{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.300156102146757,
  "eval_steps": 500,
  "global_step": 531500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.321618079091832e-06,
      "grad_norm": 3.2002530097961426,
      "learning_rate": 0.00019999832211011433,
      "loss": 9.1899,
      "step": 10
    },
    {
      "epoch": 1.8643236158183664e-05,
      "grad_norm": 2.5947623252868652,
      "learning_rate": 0.0001999964577880191,
      "loss": 8.2201,
      "step": 20
    },
    {
      "epoch": 2.7964854237275495e-05,
      "grad_norm": 2.0071773529052734,
      "learning_rate": 0.00019999459346592394,
      "loss": 7.9827,
      "step": 30
    },
    {
      "epoch": 3.728647231636733e-05,
      "grad_norm": 1.8815727233886719,
      "learning_rate": 0.00019999272914382875,
      "loss": 7.8171,
      "step": 40
    },
    {
      "epoch": 4.660809039545916e-05,
      "grad_norm": 1.8192214965820312,
      "learning_rate": 0.00019999086482173353,
      "loss": 7.5972,
      "step": 50
    },
    {
      "epoch": 5.592970847455099e-05,
      "grad_norm": 2.071150779724121,
      "learning_rate": 0.00019998900049963833,
      "loss": 7.6561,
      "step": 60
    },
    {
      "epoch": 6.525132655364283e-05,
      "grad_norm": 3.121474504470825,
      "learning_rate": 0.0001999871361775431,
      "loss": 7.0929,
      "step": 70
    },
    {
      "epoch": 7.457294463273465e-05,
      "grad_norm": 3.3654861450195312,
      "learning_rate": 0.00019998527185544792,
      "loss": 7.4497,
      "step": 80
    },
    {
      "epoch": 8.389456271182648e-05,
      "grad_norm": 2.040837526321411,
      "learning_rate": 0.00019998340753335275,
      "loss": 7.6396,
      "step": 90
    },
    {
      "epoch": 9.321618079091832e-05,
      "grad_norm": 2.3933258056640625,
      "learning_rate": 0.00019998154321125753,
      "loss": 7.1792,
      "step": 100
    },
    {
      "epoch": 0.00010253779887001015,
      "grad_norm": 2.8016998767852783,
      "learning_rate": 0.00019997967888916234,
      "loss": 7.6002,
      "step": 110
    },
    {
      "epoch": 0.00011185941694910198,
      "grad_norm": 1.8778949975967407,
      "learning_rate": 0.00019997781456706712,
      "loss": 7.4145,
      "step": 120
    },
    {
      "epoch": 0.00012118103502819381,
      "grad_norm": 1.6135320663452148,
      "learning_rate": 0.00019997595024497192,
      "loss": 7.4351,
      "step": 130
    },
    {
      "epoch": 0.00013050265310728565,
      "grad_norm": 1.3445967435836792,
      "learning_rate": 0.00019997408592287673,
      "loss": 7.4473,
      "step": 140
    },
    {
      "epoch": 0.00013982427118637748,
      "grad_norm": 4.429905414581299,
      "learning_rate": 0.00019997222160078154,
      "loss": 7.4215,
      "step": 150
    },
    {
      "epoch": 0.0001491458892654693,
      "grad_norm": 3.870464324951172,
      "learning_rate": 0.00019997035727868634,
      "loss": 7.3185,
      "step": 160
    },
    {
      "epoch": 0.00015846750734456114,
      "grad_norm": 1.592889428138733,
      "learning_rate": 0.00019996849295659112,
      "loss": 7.4256,
      "step": 170
    },
    {
      "epoch": 0.00016778912542365297,
      "grad_norm": 1.899193286895752,
      "learning_rate": 0.00019996662863449593,
      "loss": 7.2588,
      "step": 180
    },
    {
      "epoch": 0.00017711074350274482,
      "grad_norm": 2.8246569633483887,
      "learning_rate": 0.00019996476431240073,
      "loss": 7.1576,
      "step": 190
    },
    {
      "epoch": 0.00018643236158183665,
      "grad_norm": 1.6285275220870972,
      "learning_rate": 0.00019996289999030554,
      "loss": 7.4676,
      "step": 200
    },
    {
      "epoch": 0.00019575397966092848,
      "grad_norm": 1.2802704572677612,
      "learning_rate": 0.00019996103566821035,
      "loss": 7.2458,
      "step": 210
    },
    {
      "epoch": 0.0002050755977400203,
      "grad_norm": 1.3904764652252197,
      "learning_rate": 0.00019995917134611513,
      "loss": 7.3824,
      "step": 220
    },
    {
      "epoch": 0.00021439721581911213,
      "grad_norm": 1.6458265781402588,
      "learning_rate": 0.00019995730702401993,
      "loss": 7.1822,
      "step": 230
    },
    {
      "epoch": 0.00022371883389820396,
      "grad_norm": 1.5168639421463013,
      "learning_rate": 0.00019995544270192474,
      "loss": 7.0022,
      "step": 240
    },
    {
      "epoch": 0.0002330404519772958,
      "grad_norm": 1.6382197141647339,
      "learning_rate": 0.00019995357837982952,
      "loss": 7.325,
      "step": 250
    },
    {
      "epoch": 0.00024236207005638762,
      "grad_norm": 1.6937050819396973,
      "learning_rate": 0.00019995171405773435,
      "loss": 7.2266,
      "step": 260
    },
    {
      "epoch": 0.0002516836881354795,
      "grad_norm": 1.5222673416137695,
      "learning_rate": 0.00019994984973563913,
      "loss": 7.4021,
      "step": 270
    },
    {
      "epoch": 0.0002610053062145713,
      "grad_norm": 1.4677256345748901,
      "learning_rate": 0.00019994798541354394,
      "loss": 7.2444,
      "step": 280
    },
    {
      "epoch": 0.00027032692429366313,
      "grad_norm": 1.3951420783996582,
      "learning_rate": 0.00019994612109144874,
      "loss": 7.0244,
      "step": 290
    },
    {
      "epoch": 0.00027964854237275496,
      "grad_norm": 1.0766204595565796,
      "learning_rate": 0.00019994425676935352,
      "loss": 7.0672,
      "step": 300
    },
    {
      "epoch": 0.0002889701604518468,
      "grad_norm": 1.419307827949524,
      "learning_rate": 0.00019994239244725833,
      "loss": 7.1096,
      "step": 310
    },
    {
      "epoch": 0.0002982917785309386,
      "grad_norm": 1.3754544258117676,
      "learning_rate": 0.00019994052812516314,
      "loss": 7.2488,
      "step": 320
    },
    {
      "epoch": 0.00030761339661003045,
      "grad_norm": 1.6325851678848267,
      "learning_rate": 0.00019993866380306794,
      "loss": 7.0152,
      "step": 330
    },
    {
      "epoch": 0.0003169350146891223,
      "grad_norm": 1.2969365119934082,
      "learning_rate": 0.00019993679948097275,
      "loss": 7.0527,
      "step": 340
    },
    {
      "epoch": 0.0003262566327682141,
      "grad_norm": 1.1253432035446167,
      "learning_rate": 0.00019993493515887753,
      "loss": 7.0267,
      "step": 350
    },
    {
      "epoch": 0.00033557825084730593,
      "grad_norm": 1.807865023612976,
      "learning_rate": 0.00019993307083678233,
      "loss": 6.8031,
      "step": 360
    },
    {
      "epoch": 0.00034489986892639776,
      "grad_norm": 1.5728018283843994,
      "learning_rate": 0.00019993120651468714,
      "loss": 6.9796,
      "step": 370
    },
    {
      "epoch": 0.00035422148700548964,
      "grad_norm": 1.8619799613952637,
      "learning_rate": 0.00019992934219259195,
      "loss": 6.9667,
      "step": 380
    },
    {
      "epoch": 0.00036354310508458147,
      "grad_norm": 2.057664632797241,
      "learning_rate": 0.00019992747787049675,
      "loss": 7.0852,
      "step": 390
    },
    {
      "epoch": 0.0003728647231636733,
      "grad_norm": 1.520337462425232,
      "learning_rate": 0.00019992561354840153,
      "loss": 7.1992,
      "step": 400
    },
    {
      "epoch": 0.00038218634124276513,
      "grad_norm": 1.3692455291748047,
      "learning_rate": 0.00019992374922630634,
      "loss": 6.6754,
      "step": 410
    },
    {
      "epoch": 0.00039150795932185696,
      "grad_norm": 1.3424148559570312,
      "learning_rate": 0.00019992188490421114,
      "loss": 7.1068,
      "step": 420
    },
    {
      "epoch": 0.0004008295774009488,
      "grad_norm": 1.2730680704116821,
      "learning_rate": 0.00019992002058211595,
      "loss": 7.0678,
      "step": 430
    },
    {
      "epoch": 0.0004101511954800406,
      "grad_norm": 1.4230331182479858,
      "learning_rate": 0.00019991815626002076,
      "loss": 6.8743,
      "step": 440
    },
    {
      "epoch": 0.00041947281355913244,
      "grad_norm": 1.489424705505371,
      "learning_rate": 0.00019991629193792554,
      "loss": 6.8013,
      "step": 450
    },
    {
      "epoch": 0.00042879443163822427,
      "grad_norm": 1.4190672636032104,
      "learning_rate": 0.00019991442761583034,
      "loss": 6.9042,
      "step": 460
    },
    {
      "epoch": 0.0004381160497173161,
      "grad_norm": 1.275261640548706,
      "learning_rate": 0.00019991256329373512,
      "loss": 6.7855,
      "step": 470
    },
    {
      "epoch": 0.0004474376677964079,
      "grad_norm": 2.2323038578033447,
      "learning_rate": 0.00019991069897163996,
      "loss": 6.8916,
      "step": 480
    },
    {
      "epoch": 0.00045675928587549975,
      "grad_norm": 1.4984803199768066,
      "learning_rate": 0.00019990883464954476,
      "loss": 6.9483,
      "step": 490
    },
    {
      "epoch": 0.0004660809039545916,
      "grad_norm": 1.3241724967956543,
      "learning_rate": 0.00019990697032744954,
      "loss": 6.9482,
      "step": 500
    },
    {
      "epoch": 0.0004754025220336834,
      "grad_norm": 1.3920882940292358,
      "learning_rate": 0.00019990510600535435,
      "loss": 6.8723,
      "step": 510
    },
    {
      "epoch": 0.00048472414011277524,
      "grad_norm": 2.146578788757324,
      "learning_rate": 0.00019990324168325913,
      "loss": 6.9118,
      "step": 520
    },
    {
      "epoch": 0.0004940457581918671,
      "grad_norm": 1.3931065797805786,
      "learning_rate": 0.00019990137736116393,
      "loss": 6.8589,
      "step": 530
    },
    {
      "epoch": 0.000503367376270959,
      "grad_norm": 1.5255327224731445,
      "learning_rate": 0.00019989951303906877,
      "loss": 6.7164,
      "step": 540
    },
    {
      "epoch": 0.0005126889943500507,
      "grad_norm": 1.361459493637085,
      "learning_rate": 0.00019989764871697355,
      "loss": 6.6834,
      "step": 550
    },
    {
      "epoch": 0.0005220106124291426,
      "grad_norm": 1.608964443206787,
      "learning_rate": 0.00019989578439487835,
      "loss": 6.6661,
      "step": 560
    },
    {
      "epoch": 0.0005313322305082344,
      "grad_norm": 1.2415580749511719,
      "learning_rate": 0.00019989392007278313,
      "loss": 6.911,
      "step": 570
    },
    {
      "epoch": 0.0005406538485873263,
      "grad_norm": 2.1040353775024414,
      "learning_rate": 0.00019989205575068794,
      "loss": 6.79,
      "step": 580
    },
    {
      "epoch": 0.000549975466666418,
      "grad_norm": 1.3136329650878906,
      "learning_rate": 0.00019989019142859274,
      "loss": 6.7914,
      "step": 590
    },
    {
      "epoch": 0.0005592970847455099,
      "grad_norm": 1.108455777168274,
      "learning_rate": 0.00019988832710649755,
      "loss": 6.9203,
      "step": 600
    },
    {
      "epoch": 0.0005686187028246018,
      "grad_norm": 1.2124724388122559,
      "learning_rate": 0.00019988646278440236,
      "loss": 6.7748,
      "step": 610
    },
    {
      "epoch": 0.0005779403209036936,
      "grad_norm": 1.4023351669311523,
      "learning_rate": 0.00019988459846230714,
      "loss": 6.5192,
      "step": 620
    },
    {
      "epoch": 0.0005872619389827855,
      "grad_norm": 1.3030065298080444,
      "learning_rate": 0.00019988273414021194,
      "loss": 6.7158,
      "step": 630
    },
    {
      "epoch": 0.0005965835570618772,
      "grad_norm": 1.3256481885910034,
      "learning_rate": 0.00019988086981811675,
      "loss": 6.6872,
      "step": 640
    },
    {
      "epoch": 0.0006059051751409691,
      "grad_norm": 1.4297603368759155,
      "learning_rate": 0.00019987900549602155,
      "loss": 6.6444,
      "step": 650
    },
    {
      "epoch": 0.0006152267932200609,
      "grad_norm": 1.583939552307129,
      "learning_rate": 0.00019987714117392636,
      "loss": 6.5313,
      "step": 660
    },
    {
      "epoch": 0.0006245484112991528,
      "grad_norm": 1.6275238990783691,
      "learning_rate": 0.00019987527685183114,
      "loss": 6.7676,
      "step": 670
    },
    {
      "epoch": 0.0006338700293782445,
      "grad_norm": 1.3736926317214966,
      "learning_rate": 0.00019987341252973595,
      "loss": 6.6099,
      "step": 680
    },
    {
      "epoch": 0.0006431916474573364,
      "grad_norm": 1.4264907836914062,
      "learning_rate": 0.00019987154820764075,
      "loss": 6.6857,
      "step": 690
    },
    {
      "epoch": 0.0006525132655364282,
      "grad_norm": 1.1772534847259521,
      "learning_rate": 0.00019986968388554553,
      "loss": 6.9458,
      "step": 700
    },
    {
      "epoch": 0.0006618348836155201,
      "grad_norm": 2.5051887035369873,
      "learning_rate": 0.00019986781956345036,
      "loss": 6.5837,
      "step": 710
    },
    {
      "epoch": 0.0006711565016946119,
      "grad_norm": 1.3035796880722046,
      "learning_rate": 0.00019986595524135514,
      "loss": 6.8345,
      "step": 720
    },
    {
      "epoch": 0.0006804781197737037,
      "grad_norm": 1.2828218936920166,
      "learning_rate": 0.00019986409091925995,
      "loss": 6.4772,
      "step": 730
    },
    {
      "epoch": 0.0006897997378527955,
      "grad_norm": 1.4591846466064453,
      "learning_rate": 0.00019986222659716476,
      "loss": 6.6609,
      "step": 740
    },
    {
      "epoch": 0.0006991213559318874,
      "grad_norm": 1.3505735397338867,
      "learning_rate": 0.00019986036227506954,
      "loss": 6.7403,
      "step": 750
    },
    {
      "epoch": 0.0007084429740109793,
      "grad_norm": 1.3525290489196777,
      "learning_rate": 0.00019985849795297434,
      "loss": 6.6129,
      "step": 760
    },
    {
      "epoch": 0.0007177645920900711,
      "grad_norm": 1.1681386232376099,
      "learning_rate": 0.00019985663363087915,
      "loss": 6.6817,
      "step": 770
    },
    {
      "epoch": 0.0007270862101691629,
      "grad_norm": 1.4463496208190918,
      "learning_rate": 0.00019985476930878395,
      "loss": 6.4597,
      "step": 780
    },
    {
      "epoch": 0.0007364078282482547,
      "grad_norm": 1.0924383401870728,
      "learning_rate": 0.00019985290498668876,
      "loss": 6.5003,
      "step": 790
    },
    {
      "epoch": 0.0007457294463273466,
      "grad_norm": 1.252498745918274,
      "learning_rate": 0.00019985104066459354,
      "loss": 6.4871,
      "step": 800
    },
    {
      "epoch": 0.0007550510644064384,
      "grad_norm": 1.7901039123535156,
      "learning_rate": 0.00019984917634249835,
      "loss": 6.5548,
      "step": 810
    },
    {
      "epoch": 0.0007643726824855303,
      "grad_norm": 1.3178496360778809,
      "learning_rate": 0.00019984731202040315,
      "loss": 6.7518,
      "step": 820
    },
    {
      "epoch": 0.000773694300564622,
      "grad_norm": 1.3943039178848267,
      "learning_rate": 0.00019984544769830796,
      "loss": 6.6823,
      "step": 830
    },
    {
      "epoch": 0.0007830159186437139,
      "grad_norm": 1.3797706365585327,
      "learning_rate": 0.00019984358337621277,
      "loss": 6.4574,
      "step": 840
    },
    {
      "epoch": 0.0007923375367228057,
      "grad_norm": 1.2586456537246704,
      "learning_rate": 0.00019984171905411754,
      "loss": 6.6053,
      "step": 850
    },
    {
      "epoch": 0.0008016591548018976,
      "grad_norm": 1.4233512878417969,
      "learning_rate": 0.00019983985473202235,
      "loss": 6.373,
      "step": 860
    },
    {
      "epoch": 0.0008109807728809893,
      "grad_norm": 1.169721007347107,
      "learning_rate": 0.00019983799040992716,
      "loss": 6.4377,
      "step": 870
    },
    {
      "epoch": 0.0008203023909600812,
      "grad_norm": 1.2346274852752686,
      "learning_rate": 0.00019983612608783196,
      "loss": 6.6941,
      "step": 880
    },
    {
      "epoch": 0.000829624009039173,
      "grad_norm": 1.2436903715133667,
      "learning_rate": 0.00019983426176573677,
      "loss": 6.6599,
      "step": 890
    },
    {
      "epoch": 0.0008389456271182649,
      "grad_norm": 1.1152938604354858,
      "learning_rate": 0.00019983239744364155,
      "loss": 6.5135,
      "step": 900
    },
    {
      "epoch": 0.0008482672451973568,
      "grad_norm": 1.102481484413147,
      "learning_rate": 0.00019983053312154636,
      "loss": 6.4769,
      "step": 910
    },
    {
      "epoch": 0.0008575888632764485,
      "grad_norm": 1.2941477298736572,
      "learning_rate": 0.00019982866879945113,
      "loss": 6.6112,
      "step": 920
    },
    {
      "epoch": 0.0008669104813555404,
      "grad_norm": 1.3807756900787354,
      "learning_rate": 0.00019982680447735597,
      "loss": 6.2274,
      "step": 930
    },
    {
      "epoch": 0.0008762320994346322,
      "grad_norm": 1.6126453876495361,
      "learning_rate": 0.00019982494015526077,
      "loss": 6.4591,
      "step": 940
    },
    {
      "epoch": 0.0008855537175137241,
      "grad_norm": 1.204268217086792,
      "learning_rate": 0.00019982307583316555,
      "loss": 6.4005,
      "step": 950
    },
    {
      "epoch": 0.0008948753355928159,
      "grad_norm": 1.2377060651779175,
      "learning_rate": 0.00019982121151107036,
      "loss": 6.4964,
      "step": 960
    },
    {
      "epoch": 0.0009041969536719077,
      "grad_norm": 1.614266037940979,
      "learning_rate": 0.00019981934718897514,
      "loss": 6.3251,
      "step": 970
    },
    {
      "epoch": 0.0009135185717509995,
      "grad_norm": 1.041628360748291,
      "learning_rate": 0.00019981748286687995,
      "loss": 6.6314,
      "step": 980
    },
    {
      "epoch": 0.0009228401898300914,
      "grad_norm": 1.7374612092971802,
      "learning_rate": 0.00019981561854478475,
      "loss": 6.191,
      "step": 990
    },
    {
      "epoch": 0.0009321618079091832,
      "grad_norm": 1.2302523851394653,
      "learning_rate": 0.00019981375422268956,
      "loss": 6.2221,
      "step": 1000
    },
    {
      "epoch": 0.000941483425988275,
      "grad_norm": 1.307752251625061,
      "learning_rate": 0.00019981188990059436,
      "loss": 6.1177,
      "step": 1010
    },
    {
      "epoch": 0.0009508050440673668,
      "grad_norm": 1.1667742729187012,
      "learning_rate": 0.00019981002557849914,
      "loss": 6.4829,
      "step": 1020
    },
    {
      "epoch": 0.0009601266621464587,
      "grad_norm": 1.4963068962097168,
      "learning_rate": 0.00019980816125640395,
      "loss": 6.3151,
      "step": 1030
    },
    {
      "epoch": 0.0009694482802255505,
      "grad_norm": 1.3045762777328491,
      "learning_rate": 0.00019980629693430876,
      "loss": 6.555,
      "step": 1040
    },
    {
      "epoch": 0.0009787698983046425,
      "grad_norm": 1.4382593631744385,
      "learning_rate": 0.00019980443261221356,
      "loss": 6.2555,
      "step": 1050
    },
    {
      "epoch": 0.0009880915163837341,
      "grad_norm": 1.4463025331497192,
      "learning_rate": 0.00019980256829011837,
      "loss": 6.4774,
      "step": 1060
    },
    {
      "epoch": 0.000997413134462826,
      "grad_norm": 1.3769607543945312,
      "learning_rate": 0.00019980070396802315,
      "loss": 6.2783,
      "step": 1070
    },
    {
      "epoch": 0.001006734752541918,
      "grad_norm": 1.6692404747009277,
      "learning_rate": 0.00019979883964592795,
      "loss": 6.4254,
      "step": 1080
    },
    {
      "epoch": 0.0010160563706210098,
      "grad_norm": 1.1689605712890625,
      "learning_rate": 0.00019979697532383273,
      "loss": 6.4413,
      "step": 1090
    },
    {
      "epoch": 0.0010253779887001014,
      "grad_norm": 1.27359139919281,
      "learning_rate": 0.00019979511100173757,
      "loss": 6.3017,
      "step": 1100
    },
    {
      "epoch": 0.0010346996067791933,
      "grad_norm": 1.5118929147720337,
      "learning_rate": 0.00019979324667964237,
      "loss": 5.9664,
      "step": 1110
    },
    {
      "epoch": 0.0010440212248582852,
      "grad_norm": 1.4564523696899414,
      "learning_rate": 0.00019979138235754715,
      "loss": 6.3602,
      "step": 1120
    },
    {
      "epoch": 0.001053342842937377,
      "grad_norm": 1.1829404830932617,
      "learning_rate": 0.00019978951803545196,
      "loss": 5.9649,
      "step": 1130
    },
    {
      "epoch": 0.0010626644610164688,
      "grad_norm": 1.41963791847229,
      "learning_rate": 0.00019978765371335674,
      "loss": 6.2665,
      "step": 1140
    },
    {
      "epoch": 0.0010719860790955606,
      "grad_norm": 1.5920668840408325,
      "learning_rate": 0.00019978578939126154,
      "loss": 6.2583,
      "step": 1150
    },
    {
      "epoch": 0.0010813076971746525,
      "grad_norm": 1.2009505033493042,
      "learning_rate": 0.00019978392506916638,
      "loss": 6.2585,
      "step": 1160
    },
    {
      "epoch": 0.0010906293152537444,
      "grad_norm": 1.1926319599151611,
      "learning_rate": 0.00019978206074707116,
      "loss": 6.2982,
      "step": 1170
    },
    {
      "epoch": 0.001099950933332836,
      "grad_norm": 1.5066393613815308,
      "learning_rate": 0.00019978019642497596,
      "loss": 6.193,
      "step": 1180
    },
    {
      "epoch": 0.001109272551411928,
      "grad_norm": 1.331186056137085,
      "learning_rate": 0.00019977833210288074,
      "loss": 6.2763,
      "step": 1190
    },
    {
      "epoch": 0.0011185941694910198,
      "grad_norm": 1.2638782262802124,
      "learning_rate": 0.00019977646778078555,
      "loss": 6.1578,
      "step": 1200
    },
    {
      "epoch": 0.0011279157875701117,
      "grad_norm": 1.3715007305145264,
      "learning_rate": 0.00019977460345869036,
      "loss": 6.3141,
      "step": 1210
    },
    {
      "epoch": 0.0011372374056492036,
      "grad_norm": 1.441797137260437,
      "learning_rate": 0.00019977273913659516,
      "loss": 6.02,
      "step": 1220
    },
    {
      "epoch": 0.0011465590237282953,
      "grad_norm": 1.3958518505096436,
      "learning_rate": 0.00019977087481449997,
      "loss": 6.2696,
      "step": 1230
    },
    {
      "epoch": 0.0011558806418073872,
      "grad_norm": 1.293051838874817,
      "learning_rate": 0.00019976901049240475,
      "loss": 6.1604,
      "step": 1240
    },
    {
      "epoch": 0.001165202259886479,
      "grad_norm": 1.5003750324249268,
      "learning_rate": 0.00019976714617030955,
      "loss": 6.1306,
      "step": 1250
    },
    {
      "epoch": 0.001174523877965571,
      "grad_norm": 1.2592973709106445,
      "learning_rate": 0.00019976528184821436,
      "loss": 6.3005,
      "step": 1260
    },
    {
      "epoch": 0.0011838454960446626,
      "grad_norm": 1.3407094478607178,
      "learning_rate": 0.00019976341752611917,
      "loss": 6.2516,
      "step": 1270
    },
    {
      "epoch": 0.0011931671141237545,
      "grad_norm": 1.3560388088226318,
      "learning_rate": 0.00019976155320402397,
      "loss": 6.3623,
      "step": 1280
    },
    {
      "epoch": 0.0012024887322028464,
      "grad_norm": 1.0949270725250244,
      "learning_rate": 0.00019975968888192875,
      "loss": 6.1955,
      "step": 1290
    },
    {
      "epoch": 0.0012118103502819382,
      "grad_norm": 1.629128098487854,
      "learning_rate": 0.00019975782455983356,
      "loss": 6.1316,
      "step": 1300
    },
    {
      "epoch": 0.00122113196836103,
      "grad_norm": 1.3648157119750977,
      "learning_rate": 0.00019975596023773836,
      "loss": 6.1038,
      "step": 1310
    },
    {
      "epoch": 0.0012304535864401218,
      "grad_norm": 1.4169695377349854,
      "learning_rate": 0.00019975409591564317,
      "loss": 6.2731,
      "step": 1320
    },
    {
      "epoch": 0.0012397752045192137,
      "grad_norm": 1.3687059879302979,
      "learning_rate": 0.00019975223159354798,
      "loss": 6.2501,
      "step": 1330
    },
    {
      "epoch": 0.0012490968225983055,
      "grad_norm": 1.461230754852295,
      "learning_rate": 0.00019975036727145276,
      "loss": 6.1408,
      "step": 1340
    },
    {
      "epoch": 0.0012584184406773974,
      "grad_norm": 1.3677899837493896,
      "learning_rate": 0.00019974850294935756,
      "loss": 6.2774,
      "step": 1350
    },
    {
      "epoch": 0.001267740058756489,
      "grad_norm": 1.3920965194702148,
      "learning_rate": 0.00019974663862726237,
      "loss": 6.0234,
      "step": 1360
    },
    {
      "epoch": 0.001277061676835581,
      "grad_norm": 1.3761541843414307,
      "learning_rate": 0.00019974477430516715,
      "loss": 6.0593,
      "step": 1370
    },
    {
      "epoch": 0.0012863832949146729,
      "grad_norm": 1.2858134508132935,
      "learning_rate": 0.00019974290998307198,
      "loss": 6.1037,
      "step": 1380
    },
    {
      "epoch": 0.0012957049129937647,
      "grad_norm": 1.4361733198165894,
      "learning_rate": 0.00019974104566097676,
      "loss": 6.3105,
      "step": 1390
    },
    {
      "epoch": 0.0013050265310728564,
      "grad_norm": 1.4062520265579224,
      "learning_rate": 0.00019973918133888157,
      "loss": 6.2755,
      "step": 1400
    },
    {
      "epoch": 0.0013143481491519483,
      "grad_norm": 1.0689798593521118,
      "learning_rate": 0.00019973731701678637,
      "loss": 6.0287,
      "step": 1410
    },
    {
      "epoch": 0.0013236697672310402,
      "grad_norm": 1.333952784538269,
      "learning_rate": 0.00019973545269469115,
      "loss": 5.9951,
      "step": 1420
    },
    {
      "epoch": 0.001332991385310132,
      "grad_norm": 1.382594347000122,
      "learning_rate": 0.00019973358837259596,
      "loss": 6.1104,
      "step": 1430
    },
    {
      "epoch": 0.0013423130033892237,
      "grad_norm": 1.2463147640228271,
      "learning_rate": 0.00019973172405050076,
      "loss": 6.3732,
      "step": 1440
    },
    {
      "epoch": 0.0013516346214683156,
      "grad_norm": 1.1325303316116333,
      "learning_rate": 0.00019972985972840557,
      "loss": 6.0862,
      "step": 1450
    },
    {
      "epoch": 0.0013609562395474075,
      "grad_norm": 1.5340462923049927,
      "learning_rate": 0.00019972799540631038,
      "loss": 6.1943,
      "step": 1460
    },
    {
      "epoch": 0.0013702778576264994,
      "grad_norm": 1.020035743713379,
      "learning_rate": 0.00019972613108421516,
      "loss": 5.986,
      "step": 1470
    },
    {
      "epoch": 0.001379599475705591,
      "grad_norm": 1.2540322542190552,
      "learning_rate": 0.00019972426676211996,
      "loss": 6.1772,
      "step": 1480
    },
    {
      "epoch": 0.001388921093784683,
      "grad_norm": 1.27836275100708,
      "learning_rate": 0.00019972240244002477,
      "loss": 6.102,
      "step": 1490
    },
    {
      "epoch": 0.0013982427118637748,
      "grad_norm": 1.1532245874404907,
      "learning_rate": 0.00019972053811792958,
      "loss": 6.2235,
      "step": 1500
    },
    {
      "epoch": 0.0014075643299428667,
      "grad_norm": 1.227696418762207,
      "learning_rate": 0.00019971867379583438,
      "loss": 6.0564,
      "step": 1510
    },
    {
      "epoch": 0.0014168859480219586,
      "grad_norm": 1.381818413734436,
      "learning_rate": 0.00019971680947373916,
      "loss": 6.2541,
      "step": 1520
    },
    {
      "epoch": 0.0014262075661010502,
      "grad_norm": 1.3169180154800415,
      "learning_rate": 0.00019971494515164397,
      "loss": 6.1722,
      "step": 1530
    },
    {
      "epoch": 0.0014355291841801421,
      "grad_norm": 1.398491382598877,
      "learning_rate": 0.00019971308082954875,
      "loss": 6.1815,
      "step": 1540
    },
    {
      "epoch": 0.001444850802259234,
      "grad_norm": 1.3430067300796509,
      "learning_rate": 0.00019971121650745358,
      "loss": 6.0659,
      "step": 1550
    },
    {
      "epoch": 0.0014541724203383259,
      "grad_norm": 1.3426328897476196,
      "learning_rate": 0.00019970935218535839,
      "loss": 6.1898,
      "step": 1560
    },
    {
      "epoch": 0.0014634940384174175,
      "grad_norm": 1.816850185394287,
      "learning_rate": 0.00019970748786326317,
      "loss": 5.8386,
      "step": 1570
    },
    {
      "epoch": 0.0014728156564965094,
      "grad_norm": 1.354648232460022,
      "learning_rate": 0.00019970562354116797,
      "loss": 6.0253,
      "step": 1580
    },
    {
      "epoch": 0.0014821372745756013,
      "grad_norm": 1.1972473859786987,
      "learning_rate": 0.00019970375921907275,
      "loss": 5.9358,
      "step": 1590
    },
    {
      "epoch": 0.0014914588926546932,
      "grad_norm": 1.2064194679260254,
      "learning_rate": 0.00019970189489697756,
      "loss": 6.1984,
      "step": 1600
    },
    {
      "epoch": 0.0015007805107337849,
      "grad_norm": 1.4366810321807861,
      "learning_rate": 0.0001997000305748824,
      "loss": 5.99,
      "step": 1610
    },
    {
      "epoch": 0.0015101021288128767,
      "grad_norm": 1.154463291168213,
      "learning_rate": 0.00019969816625278717,
      "loss": 6.1105,
      "step": 1620
    },
    {
      "epoch": 0.0015194237468919686,
      "grad_norm": 1.238014817237854,
      "learning_rate": 0.00019969630193069198,
      "loss": 5.6891,
      "step": 1630
    },
    {
      "epoch": 0.0015287453649710605,
      "grad_norm": 1.0827627182006836,
      "learning_rate": 0.00019969443760859676,
      "loss": 6.0441,
      "step": 1640
    },
    {
      "epoch": 0.0015380669830501524,
      "grad_norm": 1.3137015104293823,
      "learning_rate": 0.00019969257328650156,
      "loss": 6.2684,
      "step": 1650
    },
    {
      "epoch": 0.001547388601129244,
      "grad_norm": 1.394424557685852,
      "learning_rate": 0.00019969070896440637,
      "loss": 6.1469,
      "step": 1660
    },
    {
      "epoch": 0.001556710219208336,
      "grad_norm": 1.1609095335006714,
      "learning_rate": 0.00019968884464231117,
      "loss": 6.01,
      "step": 1670
    },
    {
      "epoch": 0.0015660318372874278,
      "grad_norm": 1.6966915130615234,
      "learning_rate": 0.00019968698032021598,
      "loss": 5.9605,
      "step": 1680
    },
    {
      "epoch": 0.0015753534553665197,
      "grad_norm": 1.1891978979110718,
      "learning_rate": 0.00019968511599812076,
      "loss": 6.0347,
      "step": 1690
    },
    {
      "epoch": 0.0015846750734456114,
      "grad_norm": 1.3198747634887695,
      "learning_rate": 0.00019968325167602557,
      "loss": 5.8954,
      "step": 1700
    },
    {
      "epoch": 0.0015939966915247033,
      "grad_norm": 1.0945557355880737,
      "learning_rate": 0.00019968138735393037,
      "loss": 5.8878,
      "step": 1710
    },
    {
      "epoch": 0.0016033183096037951,
      "grad_norm": 1.3173582553863525,
      "learning_rate": 0.00019967952303183518,
      "loss": 6.0113,
      "step": 1720
    },
    {
      "epoch": 0.001612639927682887,
      "grad_norm": 1.2446321249008179,
      "learning_rate": 0.00019967765870973998,
      "loss": 6.0941,
      "step": 1730
    },
    {
      "epoch": 0.0016219615457619787,
      "grad_norm": 1.3846566677093506,
      "learning_rate": 0.00019967579438764476,
      "loss": 5.9155,
      "step": 1740
    },
    {
      "epoch": 0.0016312831638410706,
      "grad_norm": 1.3086119890213013,
      "learning_rate": 0.00019967393006554957,
      "loss": 5.8732,
      "step": 1750
    },
    {
      "epoch": 0.0016406047819201624,
      "grad_norm": 1.3534330129623413,
      "learning_rate": 0.00019967206574345438,
      "loss": 5.9104,
      "step": 1760
    },
    {
      "epoch": 0.0016499263999992543,
      "grad_norm": 1.3504287004470825,
      "learning_rate": 0.00019967020142135918,
      "loss": 5.9947,
      "step": 1770
    },
    {
      "epoch": 0.001659248018078346,
      "grad_norm": 1.3295890092849731,
      "learning_rate": 0.000199668337099264,
      "loss": 5.9242,
      "step": 1780
    },
    {
      "epoch": 0.0016685696361574379,
      "grad_norm": 1.148902416229248,
      "learning_rate": 0.00019966647277716877,
      "loss": 5.991,
      "step": 1790
    },
    {
      "epoch": 0.0016778912542365298,
      "grad_norm": 1.5315440893173218,
      "learning_rate": 0.00019966460845507357,
      "loss": 5.9311,
      "step": 1800
    },
    {
      "epoch": 0.0016872128723156216,
      "grad_norm": 1.2743403911590576,
      "learning_rate": 0.00019966274413297838,
      "loss": 5.9224,
      "step": 1810
    },
    {
      "epoch": 0.0016965344903947135,
      "grad_norm": 1.2901445627212524,
      "learning_rate": 0.00019966087981088316,
      "loss": 5.7913,
      "step": 1820
    },
    {
      "epoch": 0.0017058561084738052,
      "grad_norm": 1.5550379753112793,
      "learning_rate": 0.000199659015488788,
      "loss": 6.0062,
      "step": 1830
    },
    {
      "epoch": 0.001715177726552897,
      "grad_norm": 1.2582125663757324,
      "learning_rate": 0.00019965715116669277,
      "loss": 5.9277,
      "step": 1840
    },
    {
      "epoch": 0.001724499344631989,
      "grad_norm": 1.3376884460449219,
      "learning_rate": 0.00019965528684459758,
      "loss": 5.8198,
      "step": 1850
    },
    {
      "epoch": 0.0017338209627110808,
      "grad_norm": 1.200131893157959,
      "learning_rate": 0.00019965342252250239,
      "loss": 6.0146,
      "step": 1860
    },
    {
      "epoch": 0.0017431425807901725,
      "grad_norm": 1.1815059185028076,
      "learning_rate": 0.00019965155820040716,
      "loss": 5.9308,
      "step": 1870
    },
    {
      "epoch": 0.0017524641988692644,
      "grad_norm": 1.4164361953735352,
      "learning_rate": 0.00019964969387831197,
      "loss": 5.9919,
      "step": 1880
    },
    {
      "epoch": 0.0017617858169483563,
      "grad_norm": 1.4835737943649292,
      "learning_rate": 0.00019964782955621678,
      "loss": 5.9996,
      "step": 1890
    },
    {
      "epoch": 0.0017711074350274482,
      "grad_norm": 1.2593591213226318,
      "learning_rate": 0.00019964596523412158,
      "loss": 6.002,
      "step": 1900
    },
    {
      "epoch": 0.0017804290531065398,
      "grad_norm": 1.4181687831878662,
      "learning_rate": 0.0001996441009120264,
      "loss": 6.0902,
      "step": 1910
    },
    {
      "epoch": 0.0017897506711856317,
      "grad_norm": 1.2115586996078491,
      "learning_rate": 0.00019964223658993117,
      "loss": 6.0993,
      "step": 1920
    },
    {
      "epoch": 0.0017990722892647236,
      "grad_norm": 1.3498406410217285,
      "learning_rate": 0.00019964037226783598,
      "loss": 5.9847,
      "step": 1930
    },
    {
      "epoch": 0.0018083939073438155,
      "grad_norm": 1.4071052074432373,
      "learning_rate": 0.00019963850794574078,
      "loss": 5.9836,
      "step": 1940
    },
    {
      "epoch": 0.0018177155254229074,
      "grad_norm": 1.1525105237960815,
      "learning_rate": 0.0001996366436236456,
      "loss": 5.858,
      "step": 1950
    },
    {
      "epoch": 0.001827037143501999,
      "grad_norm": 1.483233094215393,
      "learning_rate": 0.0001996347793015504,
      "loss": 5.698,
      "step": 1960
    },
    {
      "epoch": 0.001836358761581091,
      "grad_norm": 1.2232613563537598,
      "learning_rate": 0.00019963291497945517,
      "loss": 5.9053,
      "step": 1970
    },
    {
      "epoch": 0.0018456803796601828,
      "grad_norm": 1.2558759450912476,
      "learning_rate": 0.00019963105065735998,
      "loss": 5.6805,
      "step": 1980
    },
    {
      "epoch": 0.0018550019977392747,
      "grad_norm": 1.382845163345337,
      "learning_rate": 0.00019962918633526479,
      "loss": 5.7904,
      "step": 1990
    },
    {
      "epoch": 0.0018643236158183663,
      "grad_norm": 1.8424104452133179,
      "learning_rate": 0.0001996273220131696,
      "loss": 5.8662,
      "step": 2000
    },
    {
      "epoch": 0.0018736452338974582,
      "grad_norm": 1.1898454427719116,
      "learning_rate": 0.0001996254576910744,
      "loss": 5.8049,
      "step": 2010
    },
    {
      "epoch": 0.00188296685197655,
      "grad_norm": 1.255414366722107,
      "learning_rate": 0.00019962359336897918,
      "loss": 5.8272,
      "step": 2020
    },
    {
      "epoch": 0.001892288470055642,
      "grad_norm": 1.2162913084030151,
      "learning_rate": 0.00019962172904688398,
      "loss": 5.8716,
      "step": 2030
    },
    {
      "epoch": 0.0019016100881347336,
      "grad_norm": 1.2049224376678467,
      "learning_rate": 0.00019961986472478876,
      "loss": 5.8946,
      "step": 2040
    },
    {
      "epoch": 0.0019109317062138255,
      "grad_norm": 1.118079662322998,
      "learning_rate": 0.00019961800040269357,
      "loss": 5.8175,
      "step": 2050
    },
    {
      "epoch": 0.0019202533242929174,
      "grad_norm": 1.8459454774856567,
      "learning_rate": 0.0001996161360805984,
      "loss": 5.7953,
      "step": 2060
    },
    {
      "epoch": 0.0019295749423720093,
      "grad_norm": 1.2114992141723633,
      "learning_rate": 0.00019961427175850318,
      "loss": 5.9572,
      "step": 2070
    },
    {
      "epoch": 0.001938896560451101,
      "grad_norm": 1.0908164978027344,
      "learning_rate": 0.000199612407436408,
      "loss": 5.7032,
      "step": 2080
    },
    {
      "epoch": 0.0019482181785301928,
      "grad_norm": 1.303984522819519,
      "learning_rate": 0.00019961054311431277,
      "loss": 5.8036,
      "step": 2090
    },
    {
      "epoch": 0.001957539796609285,
      "grad_norm": 1.3114397525787354,
      "learning_rate": 0.00019960867879221757,
      "loss": 5.8463,
      "step": 2100
    },
    {
      "epoch": 0.0019668614146883764,
      "grad_norm": 1.494646668434143,
      "learning_rate": 0.00019960681447012238,
      "loss": 5.8617,
      "step": 2110
    },
    {
      "epoch": 0.0019761830327674683,
      "grad_norm": 2.4286253452301025,
      "learning_rate": 0.0001996049501480272,
      "loss": 5.9333,
      "step": 2120
    },
    {
      "epoch": 0.00198550465084656,
      "grad_norm": 1.4013175964355469,
      "learning_rate": 0.000199603085825932,
      "loss": 5.9525,
      "step": 2130
    },
    {
      "epoch": 0.001994826268925652,
      "grad_norm": 1.3772404193878174,
      "learning_rate": 0.00019960122150383677,
      "loss": 5.9429,
      "step": 2140
    },
    {
      "epoch": 0.002004147887004744,
      "grad_norm": 1.1963543891906738,
      "learning_rate": 0.00019959935718174158,
      "loss": 5.9736,
      "step": 2150
    },
    {
      "epoch": 0.002013469505083836,
      "grad_norm": 1.3957598209381104,
      "learning_rate": 0.00019959749285964639,
      "loss": 5.7286,
      "step": 2160
    },
    {
      "epoch": 0.0020227911231629277,
      "grad_norm": 1.2272939682006836,
      "learning_rate": 0.0001995956285375512,
      "loss": 6.1031,
      "step": 2170
    },
    {
      "epoch": 0.0020321127412420196,
      "grad_norm": 0.9346178770065308,
      "learning_rate": 0.000199593764215456,
      "loss": 5.9925,
      "step": 2180
    },
    {
      "epoch": 0.002041434359321111,
      "grad_norm": 1.2703496217727661,
      "learning_rate": 0.00019959189989336078,
      "loss": 5.612,
      "step": 2190
    },
    {
      "epoch": 0.002050755977400203,
      "grad_norm": 1.2902095317840576,
      "learning_rate": 0.00019959003557126558,
      "loss": 5.865,
      "step": 2200
    },
    {
      "epoch": 0.0020600775954792948,
      "grad_norm": 1.5072613954544067,
      "learning_rate": 0.0001995881712491704,
      "loss": 5.745,
      "step": 2210
    },
    {
      "epoch": 0.0020693992135583867,
      "grad_norm": 1.2778536081314087,
      "learning_rate": 0.0001995863069270752,
      "loss": 5.913,
      "step": 2220
    },
    {
      "epoch": 0.0020787208316374785,
      "grad_norm": 1.1064364910125732,
      "learning_rate": 0.00019958444260498,
      "loss": 5.8832,
      "step": 2230
    },
    {
      "epoch": 0.0020880424497165704,
      "grad_norm": 1.5011250972747803,
      "learning_rate": 0.00019958257828288478,
      "loss": 6.0155,
      "step": 2240
    },
    {
      "epoch": 0.0020973640677956623,
      "grad_norm": 1.4736344814300537,
      "learning_rate": 0.0001995807139607896,
      "loss": 5.8554,
      "step": 2250
    },
    {
      "epoch": 0.002106685685874754,
      "grad_norm": 1.4946168661117554,
      "learning_rate": 0.0001995788496386944,
      "loss": 5.6638,
      "step": 2260
    },
    {
      "epoch": 0.002116007303953846,
      "grad_norm": 1.2639894485473633,
      "learning_rate": 0.00019957698531659917,
      "loss": 6.044,
      "step": 2270
    },
    {
      "epoch": 0.0021253289220329375,
      "grad_norm": 1.331900954246521,
      "learning_rate": 0.000199575120994504,
      "loss": 5.7338,
      "step": 2280
    },
    {
      "epoch": 0.0021346505401120294,
      "grad_norm": 1.2893610000610352,
      "learning_rate": 0.00019957325667240879,
      "loss": 5.9426,
      "step": 2290
    },
    {
      "epoch": 0.0021439721581911213,
      "grad_norm": 1.2282030582427979,
      "learning_rate": 0.0001995713923503136,
      "loss": 5.6458,
      "step": 2300
    },
    {
      "epoch": 0.002153293776270213,
      "grad_norm": 1.3002151250839233,
      "learning_rate": 0.0001995695280282184,
      "loss": 5.8605,
      "step": 2310
    },
    {
      "epoch": 0.002162615394349305,
      "grad_norm": 1.2350819110870361,
      "learning_rate": 0.00019956766370612318,
      "loss": 5.9337,
      "step": 2320
    },
    {
      "epoch": 0.002171937012428397,
      "grad_norm": 1.138680338859558,
      "learning_rate": 0.00019956579938402798,
      "loss": 5.8166,
      "step": 2330
    },
    {
      "epoch": 0.002181258630507489,
      "grad_norm": 1.160582423210144,
      "learning_rate": 0.0001995639350619328,
      "loss": 5.872,
      "step": 2340
    },
    {
      "epoch": 0.0021905802485865807,
      "grad_norm": 1.4505113363265991,
      "learning_rate": 0.0001995620707398376,
      "loss": 5.813,
      "step": 2350
    },
    {
      "epoch": 0.002199901866665672,
      "grad_norm": 1.531988263130188,
      "learning_rate": 0.00019956020641774238,
      "loss": 5.9301,
      "step": 2360
    },
    {
      "epoch": 0.002209223484744764,
      "grad_norm": 1.2693696022033691,
      "learning_rate": 0.00019955834209564718,
      "loss": 5.616,
      "step": 2370
    },
    {
      "epoch": 0.002218545102823856,
      "grad_norm": 1.250258445739746,
      "learning_rate": 0.000199556477773552,
      "loss": 5.7652,
      "step": 2380
    },
    {
      "epoch": 0.002227866720902948,
      "grad_norm": 1.2947261333465576,
      "learning_rate": 0.0001995546134514568,
      "loss": 5.8066,
      "step": 2390
    },
    {
      "epoch": 0.0022371883389820397,
      "grad_norm": 1.2564431428909302,
      "learning_rate": 0.0001995527491293616,
      "loss": 5.9769,
      "step": 2400
    },
    {
      "epoch": 0.0022465099570611316,
      "grad_norm": 1.3746691942214966,
      "learning_rate": 0.00019955088480726638,
      "loss": 5.7834,
      "step": 2410
    },
    {
      "epoch": 0.0022558315751402234,
      "grad_norm": 1.0700352191925049,
      "learning_rate": 0.0001995490204851712,
      "loss": 5.8331,
      "step": 2420
    },
    {
      "epoch": 0.0022651531932193153,
      "grad_norm": 1.4719619750976562,
      "learning_rate": 0.000199547156163076,
      "loss": 5.7878,
      "step": 2430
    },
    {
      "epoch": 0.002274474811298407,
      "grad_norm": 1.298589825630188,
      "learning_rate": 0.0001995452918409808,
      "loss": 5.4785,
      "step": 2440
    },
    {
      "epoch": 0.0022837964293774987,
      "grad_norm": 1.189931035041809,
      "learning_rate": 0.0001995434275188856,
      "loss": 5.9135,
      "step": 2450
    },
    {
      "epoch": 0.0022931180474565905,
      "grad_norm": 1.0139154195785522,
      "learning_rate": 0.00019954156319679038,
      "loss": 5.5201,
      "step": 2460
    },
    {
      "epoch": 0.0023024396655356824,
      "grad_norm": 1.186933994293213,
      "learning_rate": 0.0001995396988746952,
      "loss": 5.8116,
      "step": 2470
    },
    {
      "epoch": 0.0023117612836147743,
      "grad_norm": 1.143211007118225,
      "learning_rate": 0.0001995378345526,
      "loss": 5.6959,
      "step": 2480
    },
    {
      "epoch": 0.002321082901693866,
      "grad_norm": 1.3609297275543213,
      "learning_rate": 0.00019953597023050478,
      "loss": 5.7488,
      "step": 2490
    },
    {
      "epoch": 0.002330404519772958,
      "grad_norm": 1.3761202096939087,
      "learning_rate": 0.0001995341059084096,
      "loss": 5.664,
      "step": 2500
    },
    {
      "epoch": 0.00233972613785205,
      "grad_norm": 1.2924615144729614,
      "learning_rate": 0.0001995322415863144,
      "loss": 5.7789,
      "step": 2510
    },
    {
      "epoch": 0.002349047755931142,
      "grad_norm": 1.0873756408691406,
      "learning_rate": 0.0001995303772642192,
      "loss": 5.74,
      "step": 2520
    },
    {
      "epoch": 0.0023583693740102333,
      "grad_norm": 1.2986371517181396,
      "learning_rate": 0.000199528512942124,
      "loss": 5.5171,
      "step": 2530
    },
    {
      "epoch": 0.002367690992089325,
      "grad_norm": 1.2545480728149414,
      "learning_rate": 0.00019952664862002878,
      "loss": 5.5032,
      "step": 2540
    },
    {
      "epoch": 0.002377012610168417,
      "grad_norm": 1.243654489517212,
      "learning_rate": 0.0001995247842979336,
      "loss": 5.6625,
      "step": 2550
    },
    {
      "epoch": 0.002386334228247509,
      "grad_norm": 1.208896279335022,
      "learning_rate": 0.0001995229199758384,
      "loss": 5.7634,
      "step": 2560
    },
    {
      "epoch": 0.002395655846326601,
      "grad_norm": 1.2957215309143066,
      "learning_rate": 0.0001995210556537432,
      "loss": 5.808,
      "step": 2570
    },
    {
      "epoch": 0.0024049774644056927,
      "grad_norm": 1.363803744316101,
      "learning_rate": 0.000199519191331648,
      "loss": 5.5444,
      "step": 2580
    },
    {
      "epoch": 0.0024142990824847846,
      "grad_norm": 1.1248317956924438,
      "learning_rate": 0.00019951732700955279,
      "loss": 5.6182,
      "step": 2590
    },
    {
      "epoch": 0.0024236207005638765,
      "grad_norm": 1.3525234460830688,
      "learning_rate": 0.0001995154626874576,
      "loss": 5.7398,
      "step": 2600
    },
    {
      "epoch": 0.0024329423186429684,
      "grad_norm": 1.168745756149292,
      "learning_rate": 0.0001995135983653624,
      "loss": 5.7762,
      "step": 2610
    },
    {
      "epoch": 0.00244226393672206,
      "grad_norm": 1.2782343626022339,
      "learning_rate": 0.0001995117340432672,
      "loss": 5.7705,
      "step": 2620
    },
    {
      "epoch": 0.0024515855548011517,
      "grad_norm": 1.3210862874984741,
      "learning_rate": 0.000199509869721172,
      "loss": 5.7997,
      "step": 2630
    },
    {
      "epoch": 0.0024609071728802436,
      "grad_norm": 1.6484497785568237,
      "learning_rate": 0.0001995080053990768,
      "loss": 5.7268,
      "step": 2640
    },
    {
      "epoch": 0.0024702287909593354,
      "grad_norm": 1.3681410551071167,
      "learning_rate": 0.0001995061410769816,
      "loss": 5.7717,
      "step": 2650
    },
    {
      "epoch": 0.0024795504090384273,
      "grad_norm": 1.208825945854187,
      "learning_rate": 0.00019950427675488638,
      "loss": 5.6872,
      "step": 2660
    },
    {
      "epoch": 0.002488872027117519,
      "grad_norm": 1.2745203971862793,
      "learning_rate": 0.0001995024124327912,
      "loss": 5.4096,
      "step": 2670
    },
    {
      "epoch": 0.002498193645196611,
      "grad_norm": 1.300632357597351,
      "learning_rate": 0.00019950054811069602,
      "loss": 5.6621,
      "step": 2680
    },
    {
      "epoch": 0.002507515263275703,
      "grad_norm": 1.3528989553451538,
      "learning_rate": 0.0001994986837886008,
      "loss": 5.7386,
      "step": 2690
    },
    {
      "epoch": 0.002516836881354795,
      "grad_norm": 1.230455756187439,
      "learning_rate": 0.0001994968194665056,
      "loss": 5.6597,
      "step": 2700
    },
    {
      "epoch": 0.0025261584994338863,
      "grad_norm": 1.187113642692566,
      "learning_rate": 0.00019949495514441038,
      "loss": 5.8298,
      "step": 2710
    },
    {
      "epoch": 0.002535480117512978,
      "grad_norm": 1.225751280784607,
      "learning_rate": 0.00019949309082231519,
      "loss": 5.8685,
      "step": 2720
    },
    {
      "epoch": 0.00254480173559207,
      "grad_norm": 1.1145246028900146,
      "learning_rate": 0.00019949122650022002,
      "loss": 5.9099,
      "step": 2730
    },
    {
      "epoch": 0.002554123353671162,
      "grad_norm": 1.3539667129516602,
      "learning_rate": 0.0001994893621781248,
      "loss": 5.5627,
      "step": 2740
    },
    {
      "epoch": 0.002563444971750254,
      "grad_norm": 1.1115938425064087,
      "learning_rate": 0.0001994874978560296,
      "loss": 5.7048,
      "step": 2750
    },
    {
      "epoch": 0.0025727665898293457,
      "grad_norm": 1.3328970670700073,
      "learning_rate": 0.00019948563353393438,
      "loss": 5.7349,
      "step": 2760
    },
    {
      "epoch": 0.0025820882079084376,
      "grad_norm": 1.216753363609314,
      "learning_rate": 0.0001994837692118392,
      "loss": 5.4448,
      "step": 2770
    },
    {
      "epoch": 0.0025914098259875295,
      "grad_norm": 1.4374842643737793,
      "learning_rate": 0.000199481904889744,
      "loss": 5.6293,
      "step": 2780
    },
    {
      "epoch": 0.002600731444066621,
      "grad_norm": 1.0952367782592773,
      "learning_rate": 0.0001994800405676488,
      "loss": 5.6978,
      "step": 2790
    },
    {
      "epoch": 0.002610053062145713,
      "grad_norm": 1.4301681518554688,
      "learning_rate": 0.0001994781762455536,
      "loss": 5.7408,
      "step": 2800
    },
    {
      "epoch": 0.0026193746802248047,
      "grad_norm": 1.2883917093276978,
      "learning_rate": 0.0001994763119234584,
      "loss": 5.5578,
      "step": 2810
    },
    {
      "epoch": 0.0026286962983038966,
      "grad_norm": 1.3135972023010254,
      "learning_rate": 0.0001994744476013632,
      "loss": 5.6904,
      "step": 2820
    },
    {
      "epoch": 0.0026380179163829885,
      "grad_norm": 1.3887656927108765,
      "learning_rate": 0.000199472583279268,
      "loss": 5.9052,
      "step": 2830
    },
    {
      "epoch": 0.0026473395344620804,
      "grad_norm": 1.2662074565887451,
      "learning_rate": 0.0001994707189571728,
      "loss": 5.5887,
      "step": 2840
    },
    {
      "epoch": 0.0026566611525411722,
      "grad_norm": 1.425639033317566,
      "learning_rate": 0.00019946885463507761,
      "loss": 5.5242,
      "step": 2850
    },
    {
      "epoch": 0.002665982770620264,
      "grad_norm": 1.4637945890426636,
      "learning_rate": 0.0001994669903129824,
      "loss": 5.5908,
      "step": 2860
    },
    {
      "epoch": 0.002675304388699356,
      "grad_norm": 1.2819677591323853,
      "learning_rate": 0.0001994651259908872,
      "loss": 5.8305,
      "step": 2870
    },
    {
      "epoch": 0.0026846260067784474,
      "grad_norm": 1.0905524492263794,
      "learning_rate": 0.000199463261668792,
      "loss": 5.6117,
      "step": 2880
    },
    {
      "epoch": 0.0026939476248575393,
      "grad_norm": 1.093840479850769,
      "learning_rate": 0.0001994613973466968,
      "loss": 5.8652,
      "step": 2890
    },
    {
      "epoch": 0.002703269242936631,
      "grad_norm": 1.1936075687408447,
      "learning_rate": 0.00019945953302460162,
      "loss": 5.7183,
      "step": 2900
    },
    {
      "epoch": 0.002712590861015723,
      "grad_norm": 1.124319076538086,
      "learning_rate": 0.0001994576687025064,
      "loss": 5.7396,
      "step": 2910
    },
    {
      "epoch": 0.002721912479094815,
      "grad_norm": 1.4963290691375732,
      "learning_rate": 0.0001994558043804112,
      "loss": 5.5884,
      "step": 2920
    },
    {
      "epoch": 0.002731234097173907,
      "grad_norm": 1.247969150543213,
      "learning_rate": 0.000199453940058316,
      "loss": 5.5184,
      "step": 2930
    },
    {
      "epoch": 0.0027405557152529987,
      "grad_norm": 1.379880428314209,
      "learning_rate": 0.0001994520757362208,
      "loss": 5.539,
      "step": 2940
    },
    {
      "epoch": 0.0027498773333320906,
      "grad_norm": 1.56041419506073,
      "learning_rate": 0.00019945021141412562,
      "loss": 5.8487,
      "step": 2950
    },
    {
      "epoch": 0.002759198951411182,
      "grad_norm": 1.1487691402435303,
      "learning_rate": 0.0001994483470920304,
      "loss": 5.5802,
      "step": 2960
    },
    {
      "epoch": 0.002768520569490274,
      "grad_norm": 1.299077033996582,
      "learning_rate": 0.0001994464827699352,
      "loss": 5.756,
      "step": 2970
    },
    {
      "epoch": 0.002777842187569366,
      "grad_norm": 1.2826348543167114,
      "learning_rate": 0.00019944461844784001,
      "loss": 5.7193,
      "step": 2980
    },
    {
      "epoch": 0.0027871638056484577,
      "grad_norm": 1.452458381652832,
      "learning_rate": 0.0001994427541257448,
      "loss": 5.5401,
      "step": 2990
    },
    {
      "epoch": 0.0027964854237275496,
      "grad_norm": 1.247178316116333,
      "learning_rate": 0.0001994408898036496,
      "loss": 5.6936,
      "step": 3000
    },
    {
      "epoch": 0.0028058070418066415,
      "grad_norm": 1.3536392450332642,
      "learning_rate": 0.0001994390254815544,
      "loss": 5.6073,
      "step": 3010
    },
    {
      "epoch": 0.0028151286598857334,
      "grad_norm": 1.3040165901184082,
      "learning_rate": 0.0001994371611594592,
      "loss": 5.6739,
      "step": 3020
    },
    {
      "epoch": 0.0028244502779648253,
      "grad_norm": 1.3539677858352661,
      "learning_rate": 0.00019943529683736402,
      "loss": 5.7388,
      "step": 3030
    },
    {
      "epoch": 0.002833771896043917,
      "grad_norm": 1.2120777368545532,
      "learning_rate": 0.0001994334325152688,
      "loss": 5.6563,
      "step": 3040
    },
    {
      "epoch": 0.0028430935141230086,
      "grad_norm": 1.258194923400879,
      "learning_rate": 0.0001994315681931736,
      "loss": 5.7118,
      "step": 3050
    },
    {
      "epoch": 0.0028524151322021005,
      "grad_norm": 1.3143540620803833,
      "learning_rate": 0.0001994297038710784,
      "loss": 5.6738,
      "step": 3060
    },
    {
      "epoch": 0.0028617367502811923,
      "grad_norm": 1.3316795825958252,
      "learning_rate": 0.00019942783954898322,
      "loss": 5.6491,
      "step": 3070
    },
    {
      "epoch": 0.0028710583683602842,
      "grad_norm": 1.271754264831543,
      "learning_rate": 0.00019942597522688802,
      "loss": 5.6898,
      "step": 3080
    },
    {
      "epoch": 0.002880379986439376,
      "grad_norm": 1.39951491355896,
      "learning_rate": 0.0001994241109047928,
      "loss": 5.5332,
      "step": 3090
    },
    {
      "epoch": 0.002889701604518468,
      "grad_norm": 1.247473120689392,
      "learning_rate": 0.0001994222465826976,
      "loss": 5.5716,
      "step": 3100
    },
    {
      "epoch": 0.00289902322259756,
      "grad_norm": 1.077628254890442,
      "learning_rate": 0.0001994203822606024,
      "loss": 5.3622,
      "step": 3110
    },
    {
      "epoch": 0.0029083448406766518,
      "grad_norm": 1.4980993270874023,
      "learning_rate": 0.00019941851793850722,
      "loss": 5.6091,
      "step": 3120
    },
    {
      "epoch": 0.002917666458755743,
      "grad_norm": 1.154145359992981,
      "learning_rate": 0.00019941665361641203,
      "loss": 5.7178,
      "step": 3130
    },
    {
      "epoch": 0.002926988076834835,
      "grad_norm": 1.025425672531128,
      "learning_rate": 0.0001994147892943168,
      "loss": 5.5414,
      "step": 3140
    },
    {
      "epoch": 0.002936309694913927,
      "grad_norm": 1.3885176181793213,
      "learning_rate": 0.0001994129249722216,
      "loss": 5.6411,
      "step": 3150
    },
    {
      "epoch": 0.002945631312993019,
      "grad_norm": 1.2661043405532837,
      "learning_rate": 0.0001994110606501264,
      "loss": 5.6032,
      "step": 3160
    },
    {
      "epoch": 0.0029549529310721107,
      "grad_norm": 1.4641892910003662,
      "learning_rate": 0.0001994091963280312,
      "loss": 5.5405,
      "step": 3170
    },
    {
      "epoch": 0.0029642745491512026,
      "grad_norm": 1.1863428354263306,
      "learning_rate": 0.00019940733200593603,
      "loss": 5.3808,
      "step": 3180
    },
    {
      "epoch": 0.0029735961672302945,
      "grad_norm": 1.198477864265442,
      "learning_rate": 0.0001994054676838408,
      "loss": 5.5725,
      "step": 3190
    },
    {
      "epoch": 0.0029829177853093864,
      "grad_norm": 1.296746850013733,
      "learning_rate": 0.00019940360336174562,
      "loss": 5.7264,
      "step": 3200
    },
    {
      "epoch": 0.0029922394033884783,
      "grad_norm": 1.2926409244537354,
      "learning_rate": 0.0001994017390396504,
      "loss": 5.5574,
      "step": 3210
    },
    {
      "epoch": 0.0030015610214675697,
      "grad_norm": 1.2834935188293457,
      "learning_rate": 0.0001993998747175552,
      "loss": 5.3413,
      "step": 3220
    },
    {
      "epoch": 0.0030108826395466616,
      "grad_norm": 1.1587316989898682,
      "learning_rate": 0.00019939801039546,
      "loss": 5.6992,
      "step": 3230
    },
    {
      "epoch": 0.0030202042576257535,
      "grad_norm": 1.239972472190857,
      "learning_rate": 0.00019939614607336482,
      "loss": 5.4923,
      "step": 3240
    },
    {
      "epoch": 0.0030295258757048454,
      "grad_norm": 1.2783247232437134,
      "learning_rate": 0.00019939428175126962,
      "loss": 5.4485,
      "step": 3250
    },
    {
      "epoch": 0.0030388474937839373,
      "grad_norm": 1.225948691368103,
      "learning_rate": 0.0001993924174291744,
      "loss": 5.5722,
      "step": 3260
    },
    {
      "epoch": 0.003048169111863029,
      "grad_norm": 1.464343786239624,
      "learning_rate": 0.0001993905531070792,
      "loss": 5.492,
      "step": 3270
    },
    {
      "epoch": 0.003057490729942121,
      "grad_norm": 1.2083714008331299,
      "learning_rate": 0.00019938868878498401,
      "loss": 5.6244,
      "step": 3280
    },
    {
      "epoch": 0.003066812348021213,
      "grad_norm": 1.300954818725586,
      "learning_rate": 0.00019938682446288882,
      "loss": 5.5178,
      "step": 3290
    },
    {
      "epoch": 0.0030761339661003048,
      "grad_norm": 1.3545781373977661,
      "learning_rate": 0.00019938496014079363,
      "loss": 5.7769,
      "step": 3300
    },
    {
      "epoch": 0.0030854555841793962,
      "grad_norm": 1.2288116216659546,
      "learning_rate": 0.0001993830958186984,
      "loss": 5.7917,
      "step": 3310
    },
    {
      "epoch": 0.003094777202258488,
      "grad_norm": 1.3788201808929443,
      "learning_rate": 0.0001993812314966032,
      "loss": 5.3558,
      "step": 3320
    },
    {
      "epoch": 0.00310409882033758,
      "grad_norm": 1.299804925918579,
      "learning_rate": 0.00019937936717450802,
      "loss": 5.573,
      "step": 3330
    },
    {
      "epoch": 0.003113420438416672,
      "grad_norm": 1.2437869310379028,
      "learning_rate": 0.00019937750285241282,
      "loss": 5.503,
      "step": 3340
    },
    {
      "epoch": 0.0031227420564957638,
      "grad_norm": 1.3235256671905518,
      "learning_rate": 0.00019937563853031763,
      "loss": 5.5785,
      "step": 3350
    },
    {
      "epoch": 0.0031320636745748556,
      "grad_norm": 1.1318739652633667,
      "learning_rate": 0.0001993737742082224,
      "loss": 5.5635,
      "step": 3360
    },
    {
      "epoch": 0.0031413852926539475,
      "grad_norm": 1.3625324964523315,
      "learning_rate": 0.00019937190988612722,
      "loss": 5.4553,
      "step": 3370
    },
    {
      "epoch": 0.0031507069107330394,
      "grad_norm": 1.0852171182632446,
      "learning_rate": 0.00019937004556403202,
      "loss": 5.5201,
      "step": 3380
    },
    {
      "epoch": 0.003160028528812131,
      "grad_norm": 1.4377678632736206,
      "learning_rate": 0.0001993681812419368,
      "loss": 5.605,
      "step": 3390
    },
    {
      "epoch": 0.0031693501468912227,
      "grad_norm": 1.1983755826950073,
      "learning_rate": 0.00019936631691984164,
      "loss": 5.5059,
      "step": 3400
    },
    {
      "epoch": 0.0031786717649703146,
      "grad_norm": 1.4476436376571655,
      "learning_rate": 0.00019936445259774641,
      "loss": 5.4346,
      "step": 3410
    },
    {
      "epoch": 0.0031879933830494065,
      "grad_norm": 1.1640733480453491,
      "learning_rate": 0.00019936258827565122,
      "loss": 5.5547,
      "step": 3420
    },
    {
      "epoch": 0.0031973150011284984,
      "grad_norm": 1.5083680152893066,
      "learning_rate": 0.00019936072395355603,
      "loss": 5.4743,
      "step": 3430
    },
    {
      "epoch": 0.0032066366192075903,
      "grad_norm": 1.2621794939041138,
      "learning_rate": 0.0001993588596314608,
      "loss": 5.3268,
      "step": 3440
    },
    {
      "epoch": 0.003215958237286682,
      "grad_norm": 1.256732702255249,
      "learning_rate": 0.0001993569953093656,
      "loss": 5.4682,
      "step": 3450
    },
    {
      "epoch": 0.003225279855365774,
      "grad_norm": 1.2367398738861084,
      "learning_rate": 0.00019935513098727042,
      "loss": 5.1391,
      "step": 3460
    },
    {
      "epoch": 0.003234601473444866,
      "grad_norm": 1.2940133810043335,
      "learning_rate": 0.00019935326666517523,
      "loss": 5.7301,
      "step": 3470
    },
    {
      "epoch": 0.0032439230915239574,
      "grad_norm": 1.2407715320587158,
      "learning_rate": 0.00019935140234308003,
      "loss": 5.5051,
      "step": 3480
    },
    {
      "epoch": 0.0032532447096030493,
      "grad_norm": 1.3804930448532104,
      "learning_rate": 0.0001993495380209848,
      "loss": 5.5202,
      "step": 3490
    },
    {
      "epoch": 0.003262566327682141,
      "grad_norm": 1.3213229179382324,
      "learning_rate": 0.00019934767369888962,
      "loss": 5.5906,
      "step": 3500
    },
    {
      "epoch": 0.003271887945761233,
      "grad_norm": 1.2217659950256348,
      "learning_rate": 0.00019934580937679442,
      "loss": 5.5265,
      "step": 3510
    },
    {
      "epoch": 0.003281209563840325,
      "grad_norm": 1.2609096765518188,
      "learning_rate": 0.00019934394505469923,
      "loss": 5.5932,
      "step": 3520
    },
    {
      "epoch": 0.0032905311819194168,
      "grad_norm": 1.368153691291809,
      "learning_rate": 0.00019934208073260404,
      "loss": 5.4883,
      "step": 3530
    },
    {
      "epoch": 0.0032998527999985087,
      "grad_norm": 1.1580952405929565,
      "learning_rate": 0.00019934021641050882,
      "loss": 5.6035,
      "step": 3540
    },
    {
      "epoch": 0.0033091744180776005,
      "grad_norm": 0.9076644778251648,
      "learning_rate": 0.00019933835208841362,
      "loss": 5.8506,
      "step": 3550
    },
    {
      "epoch": 0.003318496036156692,
      "grad_norm": 1.1525126695632935,
      "learning_rate": 0.0001993364877663184,
      "loss": 5.4747,
      "step": 3560
    },
    {
      "epoch": 0.003327817654235784,
      "grad_norm": 1.1459264755249023,
      "learning_rate": 0.00019933462344422323,
      "loss": 5.4361,
      "step": 3570
    },
    {
      "epoch": 0.0033371392723148758,
      "grad_norm": 1.2631016969680786,
      "learning_rate": 0.00019933275912212804,
      "loss": 5.5771,
      "step": 3580
    },
    {
      "epoch": 0.0033464608903939676,
      "grad_norm": 1.2282088994979858,
      "learning_rate": 0.00019933089480003282,
      "loss": 5.4009,
      "step": 3590
    },
    {
      "epoch": 0.0033557825084730595,
      "grad_norm": 1.1879101991653442,
      "learning_rate": 0.00019932903047793763,
      "loss": 5.3789,
      "step": 3600
    },
    {
      "epoch": 0.0033651041265521514,
      "grad_norm": 1.3117749691009521,
      "learning_rate": 0.0001993271661558424,
      "loss": 5.533,
      "step": 3610
    },
    {
      "epoch": 0.0033744257446312433,
      "grad_norm": 1.2664544582366943,
      "learning_rate": 0.0001993253018337472,
      "loss": 5.5143,
      "step": 3620
    },
    {
      "epoch": 0.003383747362710335,
      "grad_norm": 1.4743261337280273,
      "learning_rate": 0.00019932343751165205,
      "loss": 5.5426,
      "step": 3630
    },
    {
      "epoch": 0.003393068980789427,
      "grad_norm": 1.2123234272003174,
      "learning_rate": 0.00019932157318955682,
      "loss": 5.2574,
      "step": 3640
    },
    {
      "epoch": 0.0034023905988685185,
      "grad_norm": 1.1828714609146118,
      "learning_rate": 0.00019931970886746163,
      "loss": 5.5512,
      "step": 3650
    },
    {
      "epoch": 0.0034117122169476104,
      "grad_norm": 1.1767810583114624,
      "learning_rate": 0.0001993178445453664,
      "loss": 5.3412,
      "step": 3660
    },
    {
      "epoch": 0.0034210338350267023,
      "grad_norm": 1.3130768537521362,
      "learning_rate": 0.00019931598022327122,
      "loss": 5.4696,
      "step": 3670
    },
    {
      "epoch": 0.003430355453105794,
      "grad_norm": 1.2141417264938354,
      "learning_rate": 0.00019931411590117602,
      "loss": 5.5209,
      "step": 3680
    },
    {
      "epoch": 0.003439677071184886,
      "grad_norm": 1.3696792125701904,
      "learning_rate": 0.00019931225157908083,
      "loss": 5.4467,
      "step": 3690
    },
    {
      "epoch": 0.003448998689263978,
      "grad_norm": 1.374324083328247,
      "learning_rate": 0.00019931038725698564,
      "loss": 5.5013,
      "step": 3700
    },
    {
      "epoch": 0.00345832030734307,
      "grad_norm": 1.1570488214492798,
      "learning_rate": 0.00019930852293489041,
      "loss": 5.3886,
      "step": 3710
    },
    {
      "epoch": 0.0034676419254221617,
      "grad_norm": 1.3060917854309082,
      "learning_rate": 0.00019930665861279522,
      "loss": 5.4915,
      "step": 3720
    },
    {
      "epoch": 0.003476963543501253,
      "grad_norm": 1.421693205833435,
      "learning_rate": 0.00019930479429070003,
      "loss": 5.2249,
      "step": 3730
    },
    {
      "epoch": 0.003486285161580345,
      "grad_norm": 1.1746225357055664,
      "learning_rate": 0.00019930292996860483,
      "loss": 5.7076,
      "step": 3740
    },
    {
      "epoch": 0.003495606779659437,
      "grad_norm": 1.1609456539154053,
      "learning_rate": 0.00019930106564650964,
      "loss": 5.3641,
      "step": 3750
    },
    {
      "epoch": 0.0035049283977385288,
      "grad_norm": 1.1397461891174316,
      "learning_rate": 0.00019929920132441442,
      "loss": 5.3983,
      "step": 3760
    },
    {
      "epoch": 0.0035142500158176207,
      "grad_norm": 1.1278932094573975,
      "learning_rate": 0.00019929733700231923,
      "loss": 5.4134,
      "step": 3770
    },
    {
      "epoch": 0.0035235716338967125,
      "grad_norm": 1.1916871070861816,
      "learning_rate": 0.000199295472680224,
      "loss": 5.1744,
      "step": 3780
    },
    {
      "epoch": 0.0035328932519758044,
      "grad_norm": 1.3965363502502441,
      "learning_rate": 0.00019929360835812884,
      "loss": 5.6939,
      "step": 3790
    },
    {
      "epoch": 0.0035422148700548963,
      "grad_norm": 1.3504546880722046,
      "learning_rate": 0.00019929174403603364,
      "loss": 5.6578,
      "step": 3800
    },
    {
      "epoch": 0.003551536488133988,
      "grad_norm": 1.4714974164962769,
      "learning_rate": 0.00019928987971393842,
      "loss": 5.4594,
      "step": 3810
    },
    {
      "epoch": 0.0035608581062130796,
      "grad_norm": 1.031854510307312,
      "learning_rate": 0.00019928801539184323,
      "loss": 5.5415,
      "step": 3820
    },
    {
      "epoch": 0.0035701797242921715,
      "grad_norm": 1.2571289539337158,
      "learning_rate": 0.000199286151069748,
      "loss": 5.5567,
      "step": 3830
    },
    {
      "epoch": 0.0035795013423712634,
      "grad_norm": 1.0651525259017944,
      "learning_rate": 0.00019928428674765282,
      "loss": 5.4856,
      "step": 3840
    },
    {
      "epoch": 0.0035888229604503553,
      "grad_norm": 1.1777867078781128,
      "learning_rate": 0.00019928242242555765,
      "loss": 5.1582,
      "step": 3850
    },
    {
      "epoch": 0.003598144578529447,
      "grad_norm": 1.272114872932434,
      "learning_rate": 0.00019928055810346243,
      "loss": 5.5736,
      "step": 3860
    },
    {
      "epoch": 0.003607466196608539,
      "grad_norm": 1.1237376928329468,
      "learning_rate": 0.00019927869378136723,
      "loss": 5.7088,
      "step": 3870
    },
    {
      "epoch": 0.003616787814687631,
      "grad_norm": 1.2949998378753662,
      "learning_rate": 0.000199276829459272,
      "loss": 5.1504,
      "step": 3880
    },
    {
      "epoch": 0.003626109432766723,
      "grad_norm": 1.3020434379577637,
      "learning_rate": 0.00019927496513717682,
      "loss": 5.2247,
      "step": 3890
    },
    {
      "epoch": 0.0036354310508458147,
      "grad_norm": 1.0735710859298706,
      "learning_rate": 0.00019927310081508163,
      "loss": 5.5683,
      "step": 3900
    },
    {
      "epoch": 0.003644752668924906,
      "grad_norm": 1.3284437656402588,
      "learning_rate": 0.00019927123649298643,
      "loss": 5.5571,
      "step": 3910
    },
    {
      "epoch": 0.003654074287003998,
      "grad_norm": 1.2188886404037476,
      "learning_rate": 0.00019926937217089124,
      "loss": 5.4533,
      "step": 3920
    },
    {
      "epoch": 0.00366339590508309,
      "grad_norm": 1.272570252418518,
      "learning_rate": 0.00019926750784879602,
      "loss": 5.3018,
      "step": 3930
    },
    {
      "epoch": 0.003672717523162182,
      "grad_norm": 1.0967044830322266,
      "learning_rate": 0.00019926564352670082,
      "loss": 5.5881,
      "step": 3940
    },
    {
      "epoch": 0.0036820391412412737,
      "grad_norm": 1.4160927534103394,
      "learning_rate": 0.00019926377920460563,
      "loss": 5.5331,
      "step": 3950
    },
    {
      "epoch": 0.0036913607593203656,
      "grad_norm": 1.2754175662994385,
      "learning_rate": 0.00019926191488251044,
      "loss": 5.4903,
      "step": 3960
    },
    {
      "epoch": 0.0037006823773994574,
      "grad_norm": 1.229002594947815,
      "learning_rate": 0.00019926005056041524,
      "loss": 5.3999,
      "step": 3970
    },
    {
      "epoch": 0.0037100039954785493,
      "grad_norm": 1.0431158542633057,
      "learning_rate": 0.00019925818623832002,
      "loss": 5.544,
      "step": 3980
    },
    {
      "epoch": 0.0037193256135576408,
      "grad_norm": 1.1534334421157837,
      "learning_rate": 0.00019925632191622483,
      "loss": 5.5205,
      "step": 3990
    },
    {
      "epoch": 0.0037286472316367327,
      "grad_norm": 1.301558017730713,
      "learning_rate": 0.00019925445759412963,
      "loss": 5.2903,
      "step": 4000
    },
    {
      "epoch": 0.0037379688497158245,
      "grad_norm": 1.5062683820724487,
      "learning_rate": 0.00019925259327203441,
      "loss": 5.2898,
      "step": 4010
    },
    {
      "epoch": 0.0037472904677949164,
      "grad_norm": 1.0187807083129883,
      "learning_rate": 0.00019925072894993925,
      "loss": 5.6268,
      "step": 4020
    },
    {
      "epoch": 0.0037566120858740083,
      "grad_norm": 1.282638669013977,
      "learning_rate": 0.00019924886462784403,
      "loss": 5.3641,
      "step": 4030
    },
    {
      "epoch": 0.0037659337039531,
      "grad_norm": 1.252553105354309,
      "learning_rate": 0.00019924700030574883,
      "loss": 5.4444,
      "step": 4040
    },
    {
      "epoch": 0.003775255322032192,
      "grad_norm": 1.712845802307129,
      "learning_rate": 0.00019924513598365364,
      "loss": 5.3236,
      "step": 4050
    },
    {
      "epoch": 0.003784576940111284,
      "grad_norm": 1.5193830728530884,
      "learning_rate": 0.00019924327166155842,
      "loss": 5.5538,
      "step": 4060
    },
    {
      "epoch": 0.003793898558190376,
      "grad_norm": 1.118734359741211,
      "learning_rate": 0.00019924140733946322,
      "loss": 5.4159,
      "step": 4070
    },
    {
      "epoch": 0.0038032201762694673,
      "grad_norm": 1.0855391025543213,
      "learning_rate": 0.00019923954301736803,
      "loss": 5.5071,
      "step": 4080
    },
    {
      "epoch": 0.003812541794348559,
      "grad_norm": 1.19082510471344,
      "learning_rate": 0.00019923767869527284,
      "loss": 5.3766,
      "step": 4090
    },
    {
      "epoch": 0.003821863412427651,
      "grad_norm": 1.256982684135437,
      "learning_rate": 0.00019923581437317764,
      "loss": 5.439,
      "step": 4100
    },
    {
      "epoch": 0.003831185030506743,
      "grad_norm": 1.1703884601593018,
      "learning_rate": 0.00019923395005108242,
      "loss": 5.4425,
      "step": 4110
    },
    {
      "epoch": 0.003840506648585835,
      "grad_norm": 1.477435827255249,
      "learning_rate": 0.00019923208572898723,
      "loss": 5.4855,
      "step": 4120
    },
    {
      "epoch": 0.0038498282666649267,
      "grad_norm": 1.1255111694335938,
      "learning_rate": 0.00019923022140689204,
      "loss": 5.4127,
      "step": 4130
    },
    {
      "epoch": 0.0038591498847440186,
      "grad_norm": 1.2885034084320068,
      "learning_rate": 0.00019922835708479684,
      "loss": 5.3819,
      "step": 4140
    },
    {
      "epoch": 0.0038684715028231105,
      "grad_norm": 1.031831979751587,
      "learning_rate": 0.00019922649276270165,
      "loss": 5.4945,
      "step": 4150
    },
    {
      "epoch": 0.003877793120902202,
      "grad_norm": 1.3072590827941895,
      "learning_rate": 0.00019922462844060643,
      "loss": 5.5358,
      "step": 4160
    },
    {
      "epoch": 0.003887114738981294,
      "grad_norm": 1.1616313457489014,
      "learning_rate": 0.00019922276411851123,
      "loss": 5.3367,
      "step": 4170
    },
    {
      "epoch": 0.0038964363570603857,
      "grad_norm": 1.408155083656311,
      "learning_rate": 0.00019922089979641604,
      "loss": 5.2538,
      "step": 4180
    },
    {
      "epoch": 0.0039057579751394776,
      "grad_norm": 1.2960612773895264,
      "learning_rate": 0.00019921903547432085,
      "loss": 5.4241,
      "step": 4190
    },
    {
      "epoch": 0.00391507959321857,
      "grad_norm": 1.0934237241744995,
      "learning_rate": 0.00019921717115222565,
      "loss": 5.327,
      "step": 4200
    },
    {
      "epoch": 0.003924401211297661,
      "grad_norm": 1.5017967224121094,
      "learning_rate": 0.00019921530683013043,
      "loss": 5.2333,
      "step": 4210
    },
    {
      "epoch": 0.003933722829376753,
      "grad_norm": 1.0514925718307495,
      "learning_rate": 0.00019921344250803524,
      "loss": 5.5156,
      "step": 4220
    },
    {
      "epoch": 0.003943044447455845,
      "grad_norm": 1.4204984903335571,
      "learning_rate": 0.00019921157818594002,
      "loss": 5.3267,
      "step": 4230
    },
    {
      "epoch": 0.0039523660655349365,
      "grad_norm": 1.4181143045425415,
      "learning_rate": 0.00019920971386384485,
      "loss": 5.1235,
      "step": 4240
    },
    {
      "epoch": 0.003961687683614028,
      "grad_norm": 1.4924850463867188,
      "learning_rate": 0.00019920784954174966,
      "loss": 5.3689,
      "step": 4250
    },
    {
      "epoch": 0.00397100930169312,
      "grad_norm": 1.2032670974731445,
      "learning_rate": 0.00019920598521965444,
      "loss": 5.4958,
      "step": 4260
    },
    {
      "epoch": 0.003980330919772212,
      "grad_norm": 1.1465665102005005,
      "learning_rate": 0.00019920412089755924,
      "loss": 5.309,
      "step": 4270
    },
    {
      "epoch": 0.003989652537851304,
      "grad_norm": 1.11646568775177,
      "learning_rate": 0.00019920225657546402,
      "loss": 5.3025,
      "step": 4280
    },
    {
      "epoch": 0.003998974155930396,
      "grad_norm": 1.1782361268997192,
      "learning_rate": 0.00019920039225336883,
      "loss": 5.3664,
      "step": 4290
    },
    {
      "epoch": 0.004008295774009488,
      "grad_norm": 1.1789748668670654,
      "learning_rate": 0.00019919852793127366,
      "loss": 5.3802,
      "step": 4300
    },
    {
      "epoch": 0.00401761739208858,
      "grad_norm": 1.2364659309387207,
      "learning_rate": 0.00019919666360917844,
      "loss": 5.5236,
      "step": 4310
    },
    {
      "epoch": 0.004026939010167672,
      "grad_norm": 1.1997092962265015,
      "learning_rate": 0.00019919479928708325,
      "loss": 5.596,
      "step": 4320
    },
    {
      "epoch": 0.0040362606282467635,
      "grad_norm": 1.1345164775848389,
      "learning_rate": 0.00019919293496498803,
      "loss": 5.1329,
      "step": 4330
    },
    {
      "epoch": 0.004045582246325855,
      "grad_norm": 1.4289532899856567,
      "learning_rate": 0.00019919107064289283,
      "loss": 5.2319,
      "step": 4340
    },
    {
      "epoch": 0.004054903864404947,
      "grad_norm": 1.2935742139816284,
      "learning_rate": 0.00019918920632079764,
      "loss": 5.3141,
      "step": 4350
    },
    {
      "epoch": 0.004064225482484039,
      "grad_norm": 1.1612316370010376,
      "learning_rate": 0.00019918734199870244,
      "loss": 5.253,
      "step": 4360
    },
    {
      "epoch": 0.004073547100563131,
      "grad_norm": 1.3109627962112427,
      "learning_rate": 0.00019918547767660725,
      "loss": 5.291,
      "step": 4370
    },
    {
      "epoch": 0.004082868718642222,
      "grad_norm": 1.3659842014312744,
      "learning_rate": 0.00019918361335451203,
      "loss": 5.2655,
      "step": 4380
    },
    {
      "epoch": 0.004092190336721314,
      "grad_norm": 1.3943480253219604,
      "learning_rate": 0.00019918174903241684,
      "loss": 5.4495,
      "step": 4390
    },
    {
      "epoch": 0.004101511954800406,
      "grad_norm": 1.321104884147644,
      "learning_rate": 0.00019917988471032164,
      "loss": 5.3504,
      "step": 4400
    },
    {
      "epoch": 0.004110833572879498,
      "grad_norm": 1.523200273513794,
      "learning_rate": 0.00019917802038822645,
      "loss": 5.3652,
      "step": 4410
    },
    {
      "epoch": 0.0041201551909585896,
      "grad_norm": 1.4780811071395874,
      "learning_rate": 0.00019917615606613126,
      "loss": 5.416,
      "step": 4420
    },
    {
      "epoch": 0.0041294768090376814,
      "grad_norm": 1.3386961221694946,
      "learning_rate": 0.00019917429174403603,
      "loss": 5.4294,
      "step": 4430
    },
    {
      "epoch": 0.004138798427116773,
      "grad_norm": 1.2899463176727295,
      "learning_rate": 0.00019917242742194084,
      "loss": 5.443,
      "step": 4440
    },
    {
      "epoch": 0.004148120045195865,
      "grad_norm": 1.3509970903396606,
      "learning_rate": 0.00019917056309984565,
      "loss": 5.3378,
      "step": 4450
    },
    {
      "epoch": 0.004157441663274957,
      "grad_norm": 1.3380789756774902,
      "learning_rate": 0.00019916869877775043,
      "loss": 5.5067,
      "step": 4460
    },
    {
      "epoch": 0.004166763281354049,
      "grad_norm": 1.08565354347229,
      "learning_rate": 0.00019916683445565526,
      "loss": 5.3564,
      "step": 4470
    },
    {
      "epoch": 0.004176084899433141,
      "grad_norm": 1.2319231033325195,
      "learning_rate": 0.00019916497013356004,
      "loss": 5.3349,
      "step": 4480
    },
    {
      "epoch": 0.004185406517512233,
      "grad_norm": 1.2465006113052368,
      "learning_rate": 0.00019916310581146485,
      "loss": 5.5254,
      "step": 4490
    },
    {
      "epoch": 0.004194728135591325,
      "grad_norm": 1.253631591796875,
      "learning_rate": 0.00019916124148936965,
      "loss": 5.1786,
      "step": 4500
    },
    {
      "epoch": 0.0042040497536704165,
      "grad_norm": 1.2540944814682007,
      "learning_rate": 0.00019915937716727443,
      "loss": 5.3253,
      "step": 4510
    },
    {
      "epoch": 0.004213371371749508,
      "grad_norm": 1.0001614093780518,
      "learning_rate": 0.00019915751284517924,
      "loss": 5.3676,
      "step": 4520
    },
    {
      "epoch": 0.0042226929898286,
      "grad_norm": 1.2471981048583984,
      "learning_rate": 0.00019915564852308404,
      "loss": 5.2775,
      "step": 4530
    },
    {
      "epoch": 0.004232014607907692,
      "grad_norm": 1.3270888328552246,
      "learning_rate": 0.00019915378420098885,
      "loss": 5.3276,
      "step": 4540
    },
    {
      "epoch": 0.004241336225986783,
      "grad_norm": 1.437471866607666,
      "learning_rate": 0.00019915191987889366,
      "loss": 5.1509,
      "step": 4550
    },
    {
      "epoch": 0.004250657844065875,
      "grad_norm": 1.2731117010116577,
      "learning_rate": 0.00019915005555679844,
      "loss": 5.4661,
      "step": 4560
    },
    {
      "epoch": 0.004259979462144967,
      "grad_norm": 1.3401970863342285,
      "learning_rate": 0.00019914819123470324,
      "loss": 5.4523,
      "step": 4570
    },
    {
      "epoch": 0.004269301080224059,
      "grad_norm": 1.1161812543869019,
      "learning_rate": 0.00019914632691260805,
      "loss": 5.3941,
      "step": 4580
    },
    {
      "epoch": 0.004278622698303151,
      "grad_norm": 1.3257615566253662,
      "learning_rate": 0.00019914446259051285,
      "loss": 5.1939,
      "step": 4590
    },
    {
      "epoch": 0.004287944316382243,
      "grad_norm": 1.2757081985473633,
      "learning_rate": 0.00019914259826841766,
      "loss": 5.1099,
      "step": 4600
    },
    {
      "epoch": 0.0042972659344613345,
      "grad_norm": 1.267295241355896,
      "learning_rate": 0.00019914073394632244,
      "loss": 5.1452,
      "step": 4610
    },
    {
      "epoch": 0.004306587552540426,
      "grad_norm": 1.004764437675476,
      "learning_rate": 0.00019913886962422725,
      "loss": 5.2846,
      "step": 4620
    },
    {
      "epoch": 0.004315909170619518,
      "grad_norm": 1.2154486179351807,
      "learning_rate": 0.00019913700530213205,
      "loss": 5.4634,
      "step": 4630
    },
    {
      "epoch": 0.00432523078869861,
      "grad_norm": 1.4374710321426392,
      "learning_rate": 0.00019913514098003686,
      "loss": 5.2779,
      "step": 4640
    },
    {
      "epoch": 0.004334552406777702,
      "grad_norm": 1.2544457912445068,
      "learning_rate": 0.00019913327665794167,
      "loss": 5.3621,
      "step": 4650
    },
    {
      "epoch": 0.004343874024856794,
      "grad_norm": 1.0696505308151245,
      "learning_rate": 0.00019913141233584644,
      "loss": 5.2943,
      "step": 4660
    },
    {
      "epoch": 0.004353195642935886,
      "grad_norm": 1.413886547088623,
      "learning_rate": 0.00019912954801375125,
      "loss": 5.3562,
      "step": 4670
    },
    {
      "epoch": 0.004362517261014978,
      "grad_norm": 1.100421667098999,
      "learning_rate": 0.00019912768369165603,
      "loss": 5.1926,
      "step": 4680
    },
    {
      "epoch": 0.0043718388790940695,
      "grad_norm": 1.3096504211425781,
      "learning_rate": 0.00019912581936956086,
      "loss": 5.2803,
      "step": 4690
    },
    {
      "epoch": 0.004381160497173161,
      "grad_norm": 1.0520988702774048,
      "learning_rate": 0.00019912395504746567,
      "loss": 5.3298,
      "step": 4700
    },
    {
      "epoch": 0.004390482115252253,
      "grad_norm": 1.341902732849121,
      "learning_rate": 0.00019912209072537045,
      "loss": 5.2529,
      "step": 4710
    },
    {
      "epoch": 0.004399803733331344,
      "grad_norm": 1.3497878313064575,
      "learning_rate": 0.00019912022640327526,
      "loss": 5.3267,
      "step": 4720
    },
    {
      "epoch": 0.004409125351410436,
      "grad_norm": 1.4654077291488647,
      "learning_rate": 0.00019911836208118003,
      "loss": 5.4309,
      "step": 4730
    },
    {
      "epoch": 0.004418446969489528,
      "grad_norm": 1.088441252708435,
      "learning_rate": 0.00019911649775908484,
      "loss": 5.1323,
      "step": 4740
    },
    {
      "epoch": 0.00442776858756862,
      "grad_norm": 1.0670320987701416,
      "learning_rate": 0.00019911463343698967,
      "loss": 5.2775,
      "step": 4750
    },
    {
      "epoch": 0.004437090205647712,
      "grad_norm": 1.2907274961471558,
      "learning_rate": 0.00019911276911489445,
      "loss": 5.1719,
      "step": 4760
    },
    {
      "epoch": 0.004446411823726804,
      "grad_norm": 1.250873327255249,
      "learning_rate": 0.00019911090479279926,
      "loss": 5.1324,
      "step": 4770
    },
    {
      "epoch": 0.004455733441805896,
      "grad_norm": 1.029380202293396,
      "learning_rate": 0.00019910904047070404,
      "loss": 5.3285,
      "step": 4780
    },
    {
      "epoch": 0.0044650550598849875,
      "grad_norm": 1.0265251398086548,
      "learning_rate": 0.00019910717614860885,
      "loss": 5.1574,
      "step": 4790
    },
    {
      "epoch": 0.004474376677964079,
      "grad_norm": 1.2617318630218506,
      "learning_rate": 0.00019910531182651365,
      "loss": 5.1082,
      "step": 4800
    },
    {
      "epoch": 0.004483698296043171,
      "grad_norm": 1.2759145498275757,
      "learning_rate": 0.00019910344750441846,
      "loss": 5.3644,
      "step": 4810
    },
    {
      "epoch": 0.004493019914122263,
      "grad_norm": 1.2072166204452515,
      "learning_rate": 0.00019910158318232326,
      "loss": 5.4551,
      "step": 4820
    },
    {
      "epoch": 0.004502341532201355,
      "grad_norm": 1.577711582183838,
      "learning_rate": 0.00019909971886022804,
      "loss": 5.4119,
      "step": 4830
    },
    {
      "epoch": 0.004511663150280447,
      "grad_norm": 1.1524547338485718,
      "learning_rate": 0.00019909785453813285,
      "loss": 5.0287,
      "step": 4840
    },
    {
      "epoch": 0.004520984768359539,
      "grad_norm": 1.3126821517944336,
      "learning_rate": 0.00019909599021603766,
      "loss": 5.3308,
      "step": 4850
    },
    {
      "epoch": 0.004530306386438631,
      "grad_norm": 1.1894383430480957,
      "learning_rate": 0.00019909412589394246,
      "loss": 5.1379,
      "step": 4860
    },
    {
      "epoch": 0.0045396280045177225,
      "grad_norm": 1.364798903465271,
      "learning_rate": 0.00019909226157184727,
      "loss": 5.0541,
      "step": 4870
    },
    {
      "epoch": 0.004548949622596814,
      "grad_norm": 1.4914847612380981,
      "learning_rate": 0.00019909039724975205,
      "loss": 5.2933,
      "step": 4880
    },
    {
      "epoch": 0.0045582712406759054,
      "grad_norm": 1.3886997699737549,
      "learning_rate": 0.00019908853292765685,
      "loss": 5.3607,
      "step": 4890
    },
    {
      "epoch": 0.004567592858754997,
      "grad_norm": 1.0702238082885742,
      "learning_rate": 0.00019908666860556166,
      "loss": 5.1542,
      "step": 4900
    },
    {
      "epoch": 0.004576914476834089,
      "grad_norm": 1.1650513410568237,
      "learning_rate": 0.00019908480428346644,
      "loss": 5.434,
      "step": 4910
    },
    {
      "epoch": 0.004586236094913181,
      "grad_norm": 1.2742385864257812,
      "learning_rate": 0.00019908293996137127,
      "loss": 5.2757,
      "step": 4920
    },
    {
      "epoch": 0.004595557712992273,
      "grad_norm": 1.2923774719238281,
      "learning_rate": 0.00019908107563927605,
      "loss": 5.207,
      "step": 4930
    },
    {
      "epoch": 0.004604879331071365,
      "grad_norm": 1.202679991722107,
      "learning_rate": 0.00019907921131718086,
      "loss": 5.1649,
      "step": 4940
    },
    {
      "epoch": 0.004614200949150457,
      "grad_norm": 1.4909894466400146,
      "learning_rate": 0.00019907734699508566,
      "loss": 5.4004,
      "step": 4950
    },
    {
      "epoch": 0.004623522567229549,
      "grad_norm": 1.0600632429122925,
      "learning_rate": 0.00019907548267299044,
      "loss": 5.2608,
      "step": 4960
    },
    {
      "epoch": 0.0046328441853086405,
      "grad_norm": 1.3402312994003296,
      "learning_rate": 0.00019907361835089525,
      "loss": 5.2814,
      "step": 4970
    },
    {
      "epoch": 0.004642165803387732,
      "grad_norm": 1.1433385610580444,
      "learning_rate": 0.00019907175402880006,
      "loss": 5.3429,
      "step": 4980
    },
    {
      "epoch": 0.004651487421466824,
      "grad_norm": 1.1477874517440796,
      "learning_rate": 0.00019906988970670486,
      "loss": 5.1642,
      "step": 4990
    },
    {
      "epoch": 0.004660809039545916,
      "grad_norm": 1.3452175855636597,
      "learning_rate": 0.00019906802538460967,
      "loss": 5.4284,
      "step": 5000
    },
    {
      "epoch": 0.004670130657625008,
      "grad_norm": 1.2875275611877441,
      "learning_rate": 0.00019906616106251445,
      "loss": 5.2153,
      "step": 5010
    },
    {
      "epoch": 0.0046794522757041,
      "grad_norm": 1.3351924419403076,
      "learning_rate": 0.00019906429674041925,
      "loss": 5.4499,
      "step": 5020
    },
    {
      "epoch": 0.004688773893783192,
      "grad_norm": 1.294994831085205,
      "learning_rate": 0.00019906243241832406,
      "loss": 5.3311,
      "step": 5030
    },
    {
      "epoch": 0.004698095511862284,
      "grad_norm": 1.1687626838684082,
      "learning_rate": 0.00019906056809622887,
      "loss": 5.1898,
      "step": 5040
    },
    {
      "epoch": 0.004707417129941376,
      "grad_norm": 1.243590235710144,
      "learning_rate": 0.00019905870377413365,
      "loss": 5.1779,
      "step": 5050
    },
    {
      "epoch": 0.004716738748020467,
      "grad_norm": 1.0769859552383423,
      "learning_rate": 0.00019905683945203845,
      "loss": 5.2188,
      "step": 5060
    },
    {
      "epoch": 0.0047260603660995585,
      "grad_norm": 1.101714015007019,
      "learning_rate": 0.00019905497512994326,
      "loss": 5.3516,
      "step": 5070
    },
    {
      "epoch": 0.00473538198417865,
      "grad_norm": 1.4162317514419556,
      "learning_rate": 0.00019905311080784807,
      "loss": 5.2399,
      "step": 5080
    },
    {
      "epoch": 0.004744703602257742,
      "grad_norm": 1.0552928447723389,
      "learning_rate": 0.00019905124648575287,
      "loss": 5.3459,
      "step": 5090
    },
    {
      "epoch": 0.004754025220336834,
      "grad_norm": 1.2376784086227417,
      "learning_rate": 0.00019904938216365765,
      "loss": 5.2211,
      "step": 5100
    },
    {
      "epoch": 0.004763346838415926,
      "grad_norm": 1.2079315185546875,
      "learning_rate": 0.00019904751784156246,
      "loss": 5.2095,
      "step": 5110
    },
    {
      "epoch": 0.004772668456495018,
      "grad_norm": 1.0578700304031372,
      "learning_rate": 0.00019904565351946726,
      "loss": 5.2925,
      "step": 5120
    },
    {
      "epoch": 0.00478199007457411,
      "grad_norm": 1.3376483917236328,
      "learning_rate": 0.00019904378919737204,
      "loss": 5.2089,
      "step": 5130
    },
    {
      "epoch": 0.004791311692653202,
      "grad_norm": 1.5430430173873901,
      "learning_rate": 0.00019904192487527688,
      "loss": 5.176,
      "step": 5140
    },
    {
      "epoch": 0.0048006333107322935,
      "grad_norm": 1.4371744394302368,
      "learning_rate": 0.00019904006055318166,
      "loss": 5.2182,
      "step": 5150
    },
    {
      "epoch": 0.004809954928811385,
      "grad_norm": 1.1367626190185547,
      "learning_rate": 0.00019903819623108646,
      "loss": 5.1896,
      "step": 5160
    },
    {
      "epoch": 0.004819276546890477,
      "grad_norm": 1.112873911857605,
      "learning_rate": 0.00019903633190899127,
      "loss": 5.2177,
      "step": 5170
    },
    {
      "epoch": 0.004828598164969569,
      "grad_norm": 1.36870539188385,
      "learning_rate": 0.00019903446758689605,
      "loss": 5.0179,
      "step": 5180
    },
    {
      "epoch": 0.004837919783048661,
      "grad_norm": 1.2666208744049072,
      "learning_rate": 0.00019903260326480085,
      "loss": 5.2102,
      "step": 5190
    },
    {
      "epoch": 0.004847241401127753,
      "grad_norm": 1.1836862564086914,
      "learning_rate": 0.00019903073894270566,
      "loss": 5.2429,
      "step": 5200
    },
    {
      "epoch": 0.004856563019206845,
      "grad_norm": 1.2312930822372437,
      "learning_rate": 0.00019902887462061047,
      "loss": 5.0584,
      "step": 5210
    },
    {
      "epoch": 0.004865884637285937,
      "grad_norm": 1.056840419769287,
      "learning_rate": 0.00019902701029851527,
      "loss": 5.1698,
      "step": 5220
    },
    {
      "epoch": 0.004875206255365029,
      "grad_norm": 1.2097456455230713,
      "learning_rate": 0.00019902514597642005,
      "loss": 5.4449,
      "step": 5230
    },
    {
      "epoch": 0.00488452787344412,
      "grad_norm": 1.2487527132034302,
      "learning_rate": 0.00019902328165432486,
      "loss": 5.3276,
      "step": 5240
    },
    {
      "epoch": 0.0048938494915232115,
      "grad_norm": 1.3059003353118896,
      "learning_rate": 0.00019902141733222966,
      "loss": 5.1579,
      "step": 5250
    },
    {
      "epoch": 0.004903171109602303,
      "grad_norm": 1.0184838771820068,
      "learning_rate": 0.00019901955301013447,
      "loss": 5.1944,
      "step": 5260
    },
    {
      "epoch": 0.004912492727681395,
      "grad_norm": 1.5093811750411987,
      "learning_rate": 0.00019901768868803928,
      "loss": 5.1788,
      "step": 5270
    },
    {
      "epoch": 0.004921814345760487,
      "grad_norm": 1.4566282033920288,
      "learning_rate": 0.00019901582436594406,
      "loss": 5.1845,
      "step": 5280
    },
    {
      "epoch": 0.004931135963839579,
      "grad_norm": 1.0036046504974365,
      "learning_rate": 0.00019901396004384886,
      "loss": 5.4294,
      "step": 5290
    },
    {
      "epoch": 0.004940457581918671,
      "grad_norm": 1.218375563621521,
      "learning_rate": 0.00019901209572175367,
      "loss": 5.2875,
      "step": 5300
    },
    {
      "epoch": 0.004949779199997763,
      "grad_norm": 1.0619953870773315,
      "learning_rate": 0.00019901023139965848,
      "loss": 5.3007,
      "step": 5310
    },
    {
      "epoch": 0.004959100818076855,
      "grad_norm": 1.3547497987747192,
      "learning_rate": 0.00019900836707756328,
      "loss": 5.0921,
      "step": 5320
    },
    {
      "epoch": 0.0049684224361559465,
      "grad_norm": 1.2003637552261353,
      "learning_rate": 0.00019900650275546806,
      "loss": 5.1292,
      "step": 5330
    },
    {
      "epoch": 0.004977744054235038,
      "grad_norm": 1.3883527517318726,
      "learning_rate": 0.00019900463843337287,
      "loss": 5.0688,
      "step": 5340
    },
    {
      "epoch": 0.00498706567231413,
      "grad_norm": 1.1968154907226562,
      "learning_rate": 0.00019900277411127765,
      "loss": 5.249,
      "step": 5350
    },
    {
      "epoch": 0.004996387290393222,
      "grad_norm": 1.2481530904769897,
      "learning_rate": 0.00019900090978918245,
      "loss": 5.2813,
      "step": 5360
    },
    {
      "epoch": 0.005005708908472314,
      "grad_norm": 1.1448562145233154,
      "learning_rate": 0.00019899904546708729,
      "loss": 5.0756,
      "step": 5370
    },
    {
      "epoch": 0.005015030526551406,
      "grad_norm": 1.4006481170654297,
      "learning_rate": 0.00019899718114499206,
      "loss": 5.1465,
      "step": 5380
    },
    {
      "epoch": 0.005024352144630498,
      "grad_norm": 1.4291865825653076,
      "learning_rate": 0.00019899531682289687,
      "loss": 5.1838,
      "step": 5390
    },
    {
      "epoch": 0.00503367376270959,
      "grad_norm": 1.2244784832000732,
      "learning_rate": 0.00019899345250080165,
      "loss": 5.0344,
      "step": 5400
    },
    {
      "epoch": 0.005042995380788681,
      "grad_norm": 1.1582772731781006,
      "learning_rate": 0.00019899158817870646,
      "loss": 5.1517,
      "step": 5410
    },
    {
      "epoch": 0.005052316998867773,
      "grad_norm": 1.2348134517669678,
      "learning_rate": 0.00019898972385661126,
      "loss": 5.1258,
      "step": 5420
    },
    {
      "epoch": 0.0050616386169468645,
      "grad_norm": 1.4205234050750732,
      "learning_rate": 0.00019898785953451607,
      "loss": 5.0106,
      "step": 5430
    },
    {
      "epoch": 0.005070960235025956,
      "grad_norm": 1.2600477933883667,
      "learning_rate": 0.00019898599521242088,
      "loss": 5.2971,
      "step": 5440
    },
    {
      "epoch": 0.005080281853105048,
      "grad_norm": 1.221347451210022,
      "learning_rate": 0.00019898413089032565,
      "loss": 5.2487,
      "step": 5450
    },
    {
      "epoch": 0.00508960347118414,
      "grad_norm": 1.3727588653564453,
      "learning_rate": 0.00019898226656823046,
      "loss": 4.8799,
      "step": 5460
    },
    {
      "epoch": 0.005098925089263232,
      "grad_norm": 1.3400248289108276,
      "learning_rate": 0.00019898040224613527,
      "loss": 5.1523,
      "step": 5470
    },
    {
      "epoch": 0.005108246707342324,
      "grad_norm": 1.2630524635314941,
      "learning_rate": 0.00019897853792404007,
      "loss": 5.1056,
      "step": 5480
    },
    {
      "epoch": 0.005117568325421416,
      "grad_norm": 1.4037432670593262,
      "learning_rate": 0.00019897667360194488,
      "loss": 5.0592,
      "step": 5490
    },
    {
      "epoch": 0.005126889943500508,
      "grad_norm": 1.5376112461090088,
      "learning_rate": 0.00019897480927984966,
      "loss": 5.159,
      "step": 5500
    },
    {
      "epoch": 0.0051362115615795996,
      "grad_norm": 1.1604212522506714,
      "learning_rate": 0.00019897294495775447,
      "loss": 4.9176,
      "step": 5510
    },
    {
      "epoch": 0.0051455331796586914,
      "grad_norm": 1.0964748859405518,
      "learning_rate": 0.00019897108063565927,
      "loss": 5.3084,
      "step": 5520
    },
    {
      "epoch": 0.005154854797737783,
      "grad_norm": 1.2208707332611084,
      "learning_rate": 0.00019896921631356408,
      "loss": 5.1753,
      "step": 5530
    },
    {
      "epoch": 0.005164176415816875,
      "grad_norm": 1.867410659790039,
      "learning_rate": 0.00019896735199146888,
      "loss": 5.0145,
      "step": 5540
    },
    {
      "epoch": 0.005173498033895967,
      "grad_norm": 1.2104103565216064,
      "learning_rate": 0.00019896548766937366,
      "loss": 4.9395,
      "step": 5550
    },
    {
      "epoch": 0.005182819651975059,
      "grad_norm": 1.285805583000183,
      "learning_rate": 0.00019896362334727847,
      "loss": 5.1195,
      "step": 5560
    },
    {
      "epoch": 0.005192141270054151,
      "grad_norm": 1.2046393156051636,
      "learning_rate": 0.00019896175902518328,
      "loss": 4.9267,
      "step": 5570
    },
    {
      "epoch": 0.005201462888133242,
      "grad_norm": 1.3092378377914429,
      "learning_rate": 0.00019895989470308806,
      "loss": 5.2456,
      "step": 5580
    },
    {
      "epoch": 0.005210784506212334,
      "grad_norm": 1.388725996017456,
      "learning_rate": 0.0001989580303809929,
      "loss": 5.2833,
      "step": 5590
    },
    {
      "epoch": 0.005220106124291426,
      "grad_norm": 1.210640788078308,
      "learning_rate": 0.00019895616605889767,
      "loss": 5.2182,
      "step": 5600
    },
    {
      "epoch": 0.0052294277423705175,
      "grad_norm": 1.1822795867919922,
      "learning_rate": 0.00019895430173680247,
      "loss": 5.2077,
      "step": 5610
    },
    {
      "epoch": 0.005238749360449609,
      "grad_norm": 1.219251036643982,
      "learning_rate": 0.00019895243741470728,
      "loss": 5.1898,
      "step": 5620
    },
    {
      "epoch": 0.005248070978528701,
      "grad_norm": 1.375563144683838,
      "learning_rate": 0.00019895057309261206,
      "loss": 5.0672,
      "step": 5630
    },
    {
      "epoch": 0.005257392596607793,
      "grad_norm": 1.3270974159240723,
      "learning_rate": 0.00019894870877051687,
      "loss": 5.1033,
      "step": 5640
    },
    {
      "epoch": 0.005266714214686885,
      "grad_norm": 1.1758943796157837,
      "learning_rate": 0.00019894684444842167,
      "loss": 5.0622,
      "step": 5650
    },
    {
      "epoch": 0.005276035832765977,
      "grad_norm": 1.1208724975585938,
      "learning_rate": 0.00019894498012632648,
      "loss": 5.2208,
      "step": 5660
    },
    {
      "epoch": 0.005285357450845069,
      "grad_norm": 1.337467074394226,
      "learning_rate": 0.00019894311580423129,
      "loss": 5.1301,
      "step": 5670
    },
    {
      "epoch": 0.005294679068924161,
      "grad_norm": 1.1996235847473145,
      "learning_rate": 0.00019894125148213606,
      "loss": 5.2483,
      "step": 5680
    },
    {
      "epoch": 0.005304000687003253,
      "grad_norm": 1.1906189918518066,
      "learning_rate": 0.00019893938716004087,
      "loss": 5.2066,
      "step": 5690
    },
    {
      "epoch": 0.0053133223050823445,
      "grad_norm": 1.3707860708236694,
      "learning_rate": 0.00019893752283794568,
      "loss": 5.2404,
      "step": 5700
    },
    {
      "epoch": 0.005322643923161436,
      "grad_norm": 1.417649269104004,
      "learning_rate": 0.00019893565851585048,
      "loss": 4.9153,
      "step": 5710
    },
    {
      "epoch": 0.005331965541240528,
      "grad_norm": 1.2564066648483276,
      "learning_rate": 0.0001989337941937553,
      "loss": 5.1602,
      "step": 5720
    },
    {
      "epoch": 0.00534128715931962,
      "grad_norm": 1.2228924036026,
      "learning_rate": 0.00019893192987166007,
      "loss": 5.304,
      "step": 5730
    },
    {
      "epoch": 0.005350608777398712,
      "grad_norm": 1.51832115650177,
      "learning_rate": 0.00019893006554956488,
      "loss": 5.1902,
      "step": 5740
    },
    {
      "epoch": 0.005359930395477803,
      "grad_norm": 1.300113320350647,
      "learning_rate": 0.00019892820122746968,
      "loss": 5.2532,
      "step": 5750
    },
    {
      "epoch": 0.005369252013556895,
      "grad_norm": 1.211308479309082,
      "learning_rate": 0.0001989263369053745,
      "loss": 5.0947,
      "step": 5760
    },
    {
      "epoch": 0.005378573631635987,
      "grad_norm": 1.2214182615280151,
      "learning_rate": 0.0001989244725832793,
      "loss": 5.3824,
      "step": 5770
    },
    {
      "epoch": 0.005387895249715079,
      "grad_norm": 1.2315948009490967,
      "learning_rate": 0.00019892260826118407,
      "loss": 5.172,
      "step": 5780
    },
    {
      "epoch": 0.0053972168677941705,
      "grad_norm": 1.2399879693984985,
      "learning_rate": 0.00019892074393908888,
      "loss": 5.1484,
      "step": 5790
    },
    {
      "epoch": 0.005406538485873262,
      "grad_norm": 1.2162418365478516,
      "learning_rate": 0.00019891887961699366,
      "loss": 5.1618,
      "step": 5800
    },
    {
      "epoch": 0.005415860103952354,
      "grad_norm": 1.570485234260559,
      "learning_rate": 0.0001989170152948985,
      "loss": 4.8871,
      "step": 5810
    },
    {
      "epoch": 0.005425181722031446,
      "grad_norm": 1.2628531455993652,
      "learning_rate": 0.0001989151509728033,
      "loss": 5.2982,
      "step": 5820
    },
    {
      "epoch": 0.005434503340110538,
      "grad_norm": 1.2806609869003296,
      "learning_rate": 0.00019891328665070808,
      "loss": 5.1023,
      "step": 5830
    },
    {
      "epoch": 0.00544382495818963,
      "grad_norm": 1.2390005588531494,
      "learning_rate": 0.00019891142232861288,
      "loss": 5.2572,
      "step": 5840
    },
    {
      "epoch": 0.005453146576268722,
      "grad_norm": 1.4046541452407837,
      "learning_rate": 0.00019890955800651766,
      "loss": 5.0739,
      "step": 5850
    },
    {
      "epoch": 0.005462468194347814,
      "grad_norm": 1.3272356986999512,
      "learning_rate": 0.00019890769368442247,
      "loss": 5.1326,
      "step": 5860
    },
    {
      "epoch": 0.005471789812426906,
      "grad_norm": 1.099500298500061,
      "learning_rate": 0.00019890582936232728,
      "loss": 5.1423,
      "step": 5870
    },
    {
      "epoch": 0.0054811114305059975,
      "grad_norm": 1.4753588438034058,
      "learning_rate": 0.00019890396504023208,
      "loss": 5.2778,
      "step": 5880
    },
    {
      "epoch": 0.005490433048585089,
      "grad_norm": 1.2946488857269287,
      "learning_rate": 0.0001989021007181369,
      "loss": 5.0522,
      "step": 5890
    },
    {
      "epoch": 0.005499754666664181,
      "grad_norm": 1.185023546218872,
      "learning_rate": 0.00019890023639604167,
      "loss": 5.033,
      "step": 5900
    },
    {
      "epoch": 0.005509076284743273,
      "grad_norm": 1.1120715141296387,
      "learning_rate": 0.00019889837207394647,
      "loss": 5.3046,
      "step": 5910
    },
    {
      "epoch": 0.005518397902822364,
      "grad_norm": 1.1013718843460083,
      "learning_rate": 0.00019889650775185128,
      "loss": 4.9458,
      "step": 5920
    },
    {
      "epoch": 0.005527719520901456,
      "grad_norm": 1.4224944114685059,
      "learning_rate": 0.0001988946434297561,
      "loss": 4.9638,
      "step": 5930
    },
    {
      "epoch": 0.005537041138980548,
      "grad_norm": 1.2144066095352173,
      "learning_rate": 0.0001988927791076609,
      "loss": 5.2212,
      "step": 5940
    },
    {
      "epoch": 0.00554636275705964,
      "grad_norm": 1.4543448686599731,
      "learning_rate": 0.00019889091478556567,
      "loss": 4.9461,
      "step": 5950
    },
    {
      "epoch": 0.005555684375138732,
      "grad_norm": 1.3938026428222656,
      "learning_rate": 0.00019888905046347048,
      "loss": 5.1775,
      "step": 5960
    },
    {
      "epoch": 0.0055650059932178236,
      "grad_norm": 1.1478456258773804,
      "learning_rate": 0.00019888718614137528,
      "loss": 5.3101,
      "step": 5970
    },
    {
      "epoch": 0.0055743276112969154,
      "grad_norm": 1.4637038707733154,
      "learning_rate": 0.0001988853218192801,
      "loss": 4.9345,
      "step": 5980
    },
    {
      "epoch": 0.005583649229376007,
      "grad_norm": 1.362911581993103,
      "learning_rate": 0.0001988834574971849,
      "loss": 4.8893,
      "step": 5990
    },
    {
      "epoch": 0.005592970847455099,
      "grad_norm": 1.165372610092163,
      "learning_rate": 0.00019888159317508968,
      "loss": 5.2762,
      "step": 6000
    },
    {
      "epoch": 0.005602292465534191,
      "grad_norm": 1.1097164154052734,
      "learning_rate": 0.00019887972885299448,
      "loss": 5.1003,
      "step": 6010
    },
    {
      "epoch": 0.005611614083613283,
      "grad_norm": 1.4809114933013916,
      "learning_rate": 0.0001988778645308993,
      "loss": 5.1005,
      "step": 6020
    },
    {
      "epoch": 0.005620935701692375,
      "grad_norm": 1.3912924528121948,
      "learning_rate": 0.00019887600020880407,
      "loss": 5.078,
      "step": 6030
    },
    {
      "epoch": 0.005630257319771467,
      "grad_norm": 1.1534082889556885,
      "learning_rate": 0.0001988741358867089,
      "loss": 5.104,
      "step": 6040
    },
    {
      "epoch": 0.005639578937850559,
      "grad_norm": 1.3356822729110718,
      "learning_rate": 0.00019887227156461368,
      "loss": 5.0595,
      "step": 6050
    },
    {
      "epoch": 0.0056489005559296505,
      "grad_norm": 1.2586528062820435,
      "learning_rate": 0.0001988704072425185,
      "loss": 5.0266,
      "step": 6060
    },
    {
      "epoch": 0.005658222174008742,
      "grad_norm": 1.392425298690796,
      "learning_rate": 0.0001988685429204233,
      "loss": 5.0719,
      "step": 6070
    },
    {
      "epoch": 0.005667543792087834,
      "grad_norm": 1.3261817693710327,
      "learning_rate": 0.00019886667859832807,
      "loss": 4.9859,
      "step": 6080
    },
    {
      "epoch": 0.005676865410166925,
      "grad_norm": 1.3025587797164917,
      "learning_rate": 0.00019886481427623288,
      "loss": 5.0341,
      "step": 6090
    },
    {
      "epoch": 0.005686187028246017,
      "grad_norm": 1.399071216583252,
      "learning_rate": 0.00019886294995413769,
      "loss": 5.1379,
      "step": 6100
    },
    {
      "epoch": 0.005695508646325109,
      "grad_norm": 1.3003449440002441,
      "learning_rate": 0.0001988610856320425,
      "loss": 4.8514,
      "step": 6110
    },
    {
      "epoch": 0.005704830264404201,
      "grad_norm": 1.2861438989639282,
      "learning_rate": 0.0001988592213099473,
      "loss": 5.2464,
      "step": 6120
    },
    {
      "epoch": 0.005714151882483293,
      "grad_norm": 1.346008539199829,
      "learning_rate": 0.00019885735698785208,
      "loss": 5.0573,
      "step": 6130
    },
    {
      "epoch": 0.005723473500562385,
      "grad_norm": 1.1652199029922485,
      "learning_rate": 0.00019885549266575688,
      "loss": 5.1855,
      "step": 6140
    },
    {
      "epoch": 0.005732795118641477,
      "grad_norm": 1.3582689762115479,
      "learning_rate": 0.0001988536283436617,
      "loss": 5.1143,
      "step": 6150
    },
    {
      "epoch": 0.0057421167367205685,
      "grad_norm": 1.1926723718643188,
      "learning_rate": 0.0001988517640215665,
      "loss": 5.2694,
      "step": 6160
    },
    {
      "epoch": 0.00575143835479966,
      "grad_norm": 1.525607943534851,
      "learning_rate": 0.0001988498996994713,
      "loss": 4.9664,
      "step": 6170
    },
    {
      "epoch": 0.005760759972878752,
      "grad_norm": 1.1141232252120972,
      "learning_rate": 0.00019884803537737608,
      "loss": 5.1019,
      "step": 6180
    },
    {
      "epoch": 0.005770081590957844,
      "grad_norm": 1.1829050779342651,
      "learning_rate": 0.0001988461710552809,
      "loss": 5.0421,
      "step": 6190
    },
    {
      "epoch": 0.005779403209036936,
      "grad_norm": 1.0102699995040894,
      "learning_rate": 0.0001988443067331857,
      "loss": 5.1192,
      "step": 6200
    },
    {
      "epoch": 0.005788724827116028,
      "grad_norm": 1.0801109075546265,
      "learning_rate": 0.0001988424424110905,
      "loss": 5.2317,
      "step": 6210
    },
    {
      "epoch": 0.00579804644519512,
      "grad_norm": 1.4175177812576294,
      "learning_rate": 0.0001988405780889953,
      "loss": 5.044,
      "step": 6220
    },
    {
      "epoch": 0.005807368063274212,
      "grad_norm": 1.1400365829467773,
      "learning_rate": 0.00019883871376690009,
      "loss": 5.1716,
      "step": 6230
    },
    {
      "epoch": 0.0058166896813533035,
      "grad_norm": 1.437990665435791,
      "learning_rate": 0.0001988368494448049,
      "loss": 4.9696,
      "step": 6240
    },
    {
      "epoch": 0.005826011299432395,
      "grad_norm": 1.1212600469589233,
      "learning_rate": 0.00019883498512270967,
      "loss": 5.2233,
      "step": 6250
    },
    {
      "epoch": 0.005835332917511486,
      "grad_norm": 1.2365626096725464,
      "learning_rate": 0.0001988331208006145,
      "loss": 5.094,
      "step": 6260
    },
    {
      "epoch": 0.005844654535590578,
      "grad_norm": 1.2604752779006958,
      "learning_rate": 0.0001988312564785193,
      "loss": 5.0805,
      "step": 6270
    },
    {
      "epoch": 0.00585397615366967,
      "grad_norm": 1.1505576372146606,
      "learning_rate": 0.0001988293921564241,
      "loss": 5.3072,
      "step": 6280
    },
    {
      "epoch": 0.005863297771748762,
      "grad_norm": 1.4390597343444824,
      "learning_rate": 0.0001988275278343289,
      "loss": 5.1844,
      "step": 6290
    },
    {
      "epoch": 0.005872619389827854,
      "grad_norm": 1.2144392728805542,
      "learning_rate": 0.00019882566351223368,
      "loss": 5.0047,
      "step": 6300
    },
    {
      "epoch": 0.005881941007906946,
      "grad_norm": 0.9830816388130188,
      "learning_rate": 0.00019882379919013848,
      "loss": 5.0071,
      "step": 6310
    },
    {
      "epoch": 0.005891262625986038,
      "grad_norm": 1.323015809059143,
      "learning_rate": 0.00019882193486804332,
      "loss": 4.9841,
      "step": 6320
    },
    {
      "epoch": 0.00590058424406513,
      "grad_norm": 1.1778491735458374,
      "learning_rate": 0.0001988200705459481,
      "loss": 5.2372,
      "step": 6330
    },
    {
      "epoch": 0.0059099058621442215,
      "grad_norm": 1.0651118755340576,
      "learning_rate": 0.0001988182062238529,
      "loss": 5.0832,
      "step": 6340
    },
    {
      "epoch": 0.005919227480223313,
      "grad_norm": 1.1941678524017334,
      "learning_rate": 0.00019881634190175768,
      "loss": 4.9032,
      "step": 6350
    },
    {
      "epoch": 0.005928549098302405,
      "grad_norm": 1.3345273733139038,
      "learning_rate": 0.0001988144775796625,
      "loss": 5.1932,
      "step": 6360
    },
    {
      "epoch": 0.005937870716381497,
      "grad_norm": 1.2684394121170044,
      "learning_rate": 0.0001988126132575673,
      "loss": 4.9922,
      "step": 6370
    },
    {
      "epoch": 0.005947192334460589,
      "grad_norm": 1.1524457931518555,
      "learning_rate": 0.0001988107489354721,
      "loss": 4.9626,
      "step": 6380
    },
    {
      "epoch": 0.005956513952539681,
      "grad_norm": 1.1815588474273682,
      "learning_rate": 0.0001988088846133769,
      "loss": 5.15,
      "step": 6390
    },
    {
      "epoch": 0.005965835570618773,
      "grad_norm": 1.0714011192321777,
      "learning_rate": 0.00019880702029128169,
      "loss": 5.2015,
      "step": 6400
    },
    {
      "epoch": 0.005975157188697865,
      "grad_norm": 1.1663565635681152,
      "learning_rate": 0.0001988051559691865,
      "loss": 4.965,
      "step": 6410
    },
    {
      "epoch": 0.0059844788067769565,
      "grad_norm": 1.3307820558547974,
      "learning_rate": 0.00019880329164709127,
      "loss": 5.0553,
      "step": 6420
    },
    {
      "epoch": 0.005993800424856048,
      "grad_norm": 1.2224727869033813,
      "learning_rate": 0.0001988014273249961,
      "loss": 5.1221,
      "step": 6430
    },
    {
      "epoch": 0.0060031220429351394,
      "grad_norm": 1.1968334913253784,
      "learning_rate": 0.0001987995630029009,
      "loss": 5.1913,
      "step": 6440
    },
    {
      "epoch": 0.006012443661014231,
      "grad_norm": 1.2567816972732544,
      "learning_rate": 0.0001987976986808057,
      "loss": 5.1745,
      "step": 6450
    },
    {
      "epoch": 0.006021765279093323,
      "grad_norm": 1.3199418783187866,
      "learning_rate": 0.0001987958343587105,
      "loss": 5.0484,
      "step": 6460
    },
    {
      "epoch": 0.006031086897172415,
      "grad_norm": 1.103058934211731,
      "learning_rate": 0.00019879397003661528,
      "loss": 4.9859,
      "step": 6470
    },
    {
      "epoch": 0.006040408515251507,
      "grad_norm": 1.327150821685791,
      "learning_rate": 0.00019879210571452008,
      "loss": 4.9587,
      "step": 6480
    },
    {
      "epoch": 0.006049730133330599,
      "grad_norm": 1.390529751777649,
      "learning_rate": 0.00019879024139242491,
      "loss": 4.9159,
      "step": 6490
    },
    {
      "epoch": 0.006059051751409691,
      "grad_norm": 1.0890625715255737,
      "learning_rate": 0.0001987883770703297,
      "loss": 4.99,
      "step": 6500
    },
    {
      "epoch": 0.006068373369488783,
      "grad_norm": 1.5781058073043823,
      "learning_rate": 0.0001987865127482345,
      "loss": 5.0163,
      "step": 6510
    },
    {
      "epoch": 0.0060776949875678745,
      "grad_norm": 1.2957006692886353,
      "learning_rate": 0.00019878464842613928,
      "loss": 5.1321,
      "step": 6520
    },
    {
      "epoch": 0.006087016605646966,
      "grad_norm": 1.2840747833251953,
      "learning_rate": 0.00019878278410404409,
      "loss": 4.8653,
      "step": 6530
    },
    {
      "epoch": 0.006096338223726058,
      "grad_norm": 1.137766718864441,
      "learning_rate": 0.0001987809197819489,
      "loss": 4.9539,
      "step": 6540
    },
    {
      "epoch": 0.00610565984180515,
      "grad_norm": 1.1781587600708008,
      "learning_rate": 0.0001987790554598537,
      "loss": 4.9162,
      "step": 6550
    },
    {
      "epoch": 0.006114981459884242,
      "grad_norm": 1.3972358703613281,
      "learning_rate": 0.0001987771911377585,
      "loss": 5.1333,
      "step": 6560
    },
    {
      "epoch": 0.006124303077963334,
      "grad_norm": 1.2119619846343994,
      "learning_rate": 0.00019877532681566328,
      "loss": 4.9059,
      "step": 6570
    },
    {
      "epoch": 0.006133624696042426,
      "grad_norm": 0.9277644753456116,
      "learning_rate": 0.0001987734624935681,
      "loss": 5.1498,
      "step": 6580
    },
    {
      "epoch": 0.006142946314121518,
      "grad_norm": 1.3425973653793335,
      "learning_rate": 0.0001987715981714729,
      "loss": 4.9316,
      "step": 6590
    },
    {
      "epoch": 0.0061522679322006096,
      "grad_norm": 1.3517606258392334,
      "learning_rate": 0.0001987697338493777,
      "loss": 4.9519,
      "step": 6600
    },
    {
      "epoch": 0.006161589550279701,
      "grad_norm": 1.4192169904708862,
      "learning_rate": 0.0001987678695272825,
      "loss": 5.0519,
      "step": 6610
    },
    {
      "epoch": 0.0061709111683587925,
      "grad_norm": 1.272735595703125,
      "learning_rate": 0.0001987660052051873,
      "loss": 5.186,
      "step": 6620
    },
    {
      "epoch": 0.006180232786437884,
      "grad_norm": 1.2500900030136108,
      "learning_rate": 0.0001987641408830921,
      "loss": 4.9283,
      "step": 6630
    },
    {
      "epoch": 0.006189554404516976,
      "grad_norm": 1.088495135307312,
      "learning_rate": 0.0001987622765609969,
      "loss": 5.1554,
      "step": 6640
    },
    {
      "epoch": 0.006198876022596068,
      "grad_norm": 1.3526616096496582,
      "learning_rate": 0.0001987604122389017,
      "loss": 4.935,
      "step": 6650
    },
    {
      "epoch": 0.00620819764067516,
      "grad_norm": 1.2497702836990356,
      "learning_rate": 0.00019875854791680651,
      "loss": 4.9839,
      "step": 6660
    },
    {
      "epoch": 0.006217519258754252,
      "grad_norm": 1.1069804430007935,
      "learning_rate": 0.0001987566835947113,
      "loss": 4.9872,
      "step": 6670
    },
    {
      "epoch": 0.006226840876833344,
      "grad_norm": 1.1419380903244019,
      "learning_rate": 0.0001987548192726161,
      "loss": 5.0501,
      "step": 6680
    },
    {
      "epoch": 0.006236162494912436,
      "grad_norm": 1.1816203594207764,
      "learning_rate": 0.0001987529549505209,
      "loss": 5.0927,
      "step": 6690
    },
    {
      "epoch": 0.0062454841129915275,
      "grad_norm": 1.2747328281402588,
      "learning_rate": 0.00019875109062842568,
      "loss": 4.8809,
      "step": 6700
    },
    {
      "epoch": 0.006254805731070619,
      "grad_norm": 1.4116417169570923,
      "learning_rate": 0.00019874922630633052,
      "loss": 4.8944,
      "step": 6710
    },
    {
      "epoch": 0.006264127349149711,
      "grad_norm": 1.1656869649887085,
      "learning_rate": 0.0001987473619842353,
      "loss": 4.9985,
      "step": 6720
    },
    {
      "epoch": 0.006273448967228803,
      "grad_norm": 1.2959582805633545,
      "learning_rate": 0.0001987454976621401,
      "loss": 4.9148,
      "step": 6730
    },
    {
      "epoch": 0.006282770585307895,
      "grad_norm": 1.102542519569397,
      "learning_rate": 0.0001987436333400449,
      "loss": 5.106,
      "step": 6740
    },
    {
      "epoch": 0.006292092203386987,
      "grad_norm": 1.198629379272461,
      "learning_rate": 0.0001987417690179497,
      "loss": 5.0433,
      "step": 6750
    },
    {
      "epoch": 0.006301413821466079,
      "grad_norm": 1.320414662361145,
      "learning_rate": 0.0001987399046958545,
      "loss": 5.125,
      "step": 6760
    },
    {
      "epoch": 0.006310735439545171,
      "grad_norm": 1.1532963514328003,
      "learning_rate": 0.0001987380403737593,
      "loss": 5.2629,
      "step": 6770
    },
    {
      "epoch": 0.006320057057624262,
      "grad_norm": 1.4396907091140747,
      "learning_rate": 0.0001987361760516641,
      "loss": 4.9567,
      "step": 6780
    },
    {
      "epoch": 0.006329378675703354,
      "grad_norm": 1.0336612462997437,
      "learning_rate": 0.00019873431172956891,
      "loss": 5.0743,
      "step": 6790
    },
    {
      "epoch": 0.0063387002937824455,
      "grad_norm": 1.1531814336776733,
      "learning_rate": 0.0001987324474074737,
      "loss": 4.9102,
      "step": 6800
    },
    {
      "epoch": 0.006348021911861537,
      "grad_norm": 1.267416000366211,
      "learning_rate": 0.0001987305830853785,
      "loss": 5.2677,
      "step": 6810
    },
    {
      "epoch": 0.006357343529940629,
      "grad_norm": 1.1670167446136475,
      "learning_rate": 0.0001987287187632833,
      "loss": 4.9793,
      "step": 6820
    },
    {
      "epoch": 0.006366665148019721,
      "grad_norm": 1.314016580581665,
      "learning_rate": 0.0001987268544411881,
      "loss": 4.8662,
      "step": 6830
    },
    {
      "epoch": 0.006375986766098813,
      "grad_norm": 1.173602819442749,
      "learning_rate": 0.00019872499011909292,
      "loss": 4.8278,
      "step": 6840
    },
    {
      "epoch": 0.006385308384177905,
      "grad_norm": 1.3309407234191895,
      "learning_rate": 0.0001987231257969977,
      "loss": 4.8397,
      "step": 6850
    },
    {
      "epoch": 0.006394630002256997,
      "grad_norm": 1.1805332899093628,
      "learning_rate": 0.0001987212614749025,
      "loss": 4.9054,
      "step": 6860
    },
    {
      "epoch": 0.006403951620336089,
      "grad_norm": 1.2185137271881104,
      "learning_rate": 0.00019871939715280728,
      "loss": 4.9311,
      "step": 6870
    },
    {
      "epoch": 0.0064132732384151805,
      "grad_norm": 1.2228668928146362,
      "learning_rate": 0.00019871753283071212,
      "loss": 4.9919,
      "step": 6880
    },
    {
      "epoch": 0.006422594856494272,
      "grad_norm": 1.267153263092041,
      "learning_rate": 0.00019871566850861692,
      "loss": 5.1297,
      "step": 6890
    },
    {
      "epoch": 0.006431916474573364,
      "grad_norm": 1.2345577478408813,
      "learning_rate": 0.0001987138041865217,
      "loss": 4.964,
      "step": 6900
    },
    {
      "epoch": 0.006441238092652456,
      "grad_norm": 1.0528508424758911,
      "learning_rate": 0.0001987119398644265,
      "loss": 5.1727,
      "step": 6910
    },
    {
      "epoch": 0.006450559710731548,
      "grad_norm": 1.9677157402038574,
      "learning_rate": 0.0001987100755423313,
      "loss": 4.9334,
      "step": 6920
    },
    {
      "epoch": 0.00645988132881064,
      "grad_norm": 1.1011134386062622,
      "learning_rate": 0.0001987082112202361,
      "loss": 5.2131,
      "step": 6930
    },
    {
      "epoch": 0.006469202946889732,
      "grad_norm": 1.3979896306991577,
      "learning_rate": 0.00019870634689814093,
      "loss": 5.0957,
      "step": 6940
    },
    {
      "epoch": 0.006478524564968823,
      "grad_norm": 1.1345504522323608,
      "learning_rate": 0.0001987044825760457,
      "loss": 5.0948,
      "step": 6950
    },
    {
      "epoch": 0.006487846183047915,
      "grad_norm": 1.2480262517929077,
      "learning_rate": 0.0001987026182539505,
      "loss": 4.7608,
      "step": 6960
    },
    {
      "epoch": 0.006497167801127007,
      "grad_norm": 1.170079231262207,
      "learning_rate": 0.0001987007539318553,
      "loss": 5.0351,
      "step": 6970
    },
    {
      "epoch": 0.0065064894192060985,
      "grad_norm": 0.9738718271255493,
      "learning_rate": 0.0001986988896097601,
      "loss": 5.2445,
      "step": 6980
    },
    {
      "epoch": 0.00651581103728519,
      "grad_norm": 1.3188416957855225,
      "learning_rate": 0.0001986970252876649,
      "loss": 4.8602,
      "step": 6990
    },
    {
      "epoch": 0.006525132655364282,
      "grad_norm": 1.2213165760040283,
      "learning_rate": 0.0001986951609655697,
      "loss": 5.0424,
      "step": 7000
    },
    {
      "epoch": 0.006534454273443374,
      "grad_norm": 1.4280872344970703,
      "learning_rate": 0.00019869329664347452,
      "loss": 4.9569,
      "step": 7010
    },
    {
      "epoch": 0.006543775891522466,
      "grad_norm": 1.204158902168274,
      "learning_rate": 0.0001986914323213793,
      "loss": 4.8771,
      "step": 7020
    },
    {
      "epoch": 0.006553097509601558,
      "grad_norm": 1.2270177602767944,
      "learning_rate": 0.0001986895679992841,
      "loss": 4.9454,
      "step": 7030
    },
    {
      "epoch": 0.00656241912768065,
      "grad_norm": 1.1024116277694702,
      "learning_rate": 0.0001986877036771889,
      "loss": 4.9192,
      "step": 7040
    },
    {
      "epoch": 0.006571740745759742,
      "grad_norm": 1.0751714706420898,
      "learning_rate": 0.00019868583935509372,
      "loss": 4.804,
      "step": 7050
    },
    {
      "epoch": 0.0065810623638388336,
      "grad_norm": 1.2107363939285278,
      "learning_rate": 0.00019868397503299852,
      "loss": 4.8228,
      "step": 7060
    },
    {
      "epoch": 0.0065903839819179254,
      "grad_norm": 1.2835782766342163,
      "learning_rate": 0.0001986821107109033,
      "loss": 5.1657,
      "step": 7070
    },
    {
      "epoch": 0.006599705599997017,
      "grad_norm": 1.1140700578689575,
      "learning_rate": 0.0001986802463888081,
      "loss": 5.0288,
      "step": 7080
    },
    {
      "epoch": 0.006609027218076109,
      "grad_norm": 1.0903693437576294,
      "learning_rate": 0.00019867838206671291,
      "loss": 5.0027,
      "step": 7090
    },
    {
      "epoch": 0.006618348836155201,
      "grad_norm": 1.3603298664093018,
      "learning_rate": 0.00019867651774461772,
      "loss": 4.9445,
      "step": 7100
    },
    {
      "epoch": 0.006627670454234293,
      "grad_norm": 1.186206340789795,
      "learning_rate": 0.00019867465342252253,
      "loss": 5.0772,
      "step": 7110
    },
    {
      "epoch": 0.006636992072313384,
      "grad_norm": 0.9891053438186646,
      "learning_rate": 0.0001986727891004273,
      "loss": 5.1197,
      "step": 7120
    },
    {
      "epoch": 0.006646313690392476,
      "grad_norm": 1.4031826257705688,
      "learning_rate": 0.0001986709247783321,
      "loss": 4.8764,
      "step": 7130
    },
    {
      "epoch": 0.006655635308471568,
      "grad_norm": 1.370801568031311,
      "learning_rate": 0.00019866906045623692,
      "loss": 4.7133,
      "step": 7140
    },
    {
      "epoch": 0.00666495692655066,
      "grad_norm": 1.1628410816192627,
      "learning_rate": 0.0001986671961341417,
      "loss": 5.0269,
      "step": 7150
    },
    {
      "epoch": 0.0066742785446297515,
      "grad_norm": 1.3528753519058228,
      "learning_rate": 0.00019866533181204653,
      "loss": 5.1038,
      "step": 7160
    },
    {
      "epoch": 0.006683600162708843,
      "grad_norm": 1.1814266443252563,
      "learning_rate": 0.0001986634674899513,
      "loss": 5.0099,
      "step": 7170
    },
    {
      "epoch": 0.006692921780787935,
      "grad_norm": 1.448676347732544,
      "learning_rate": 0.00019866160316785612,
      "loss": 5.094,
      "step": 7180
    },
    {
      "epoch": 0.006702243398867027,
      "grad_norm": 1.229223370552063,
      "learning_rate": 0.00019865973884576092,
      "loss": 4.8916,
      "step": 7190
    },
    {
      "epoch": 0.006711565016946119,
      "grad_norm": 1.133386492729187,
      "learning_rate": 0.0001986578745236657,
      "loss": 4.8941,
      "step": 7200
    },
    {
      "epoch": 0.006720886635025211,
      "grad_norm": 1.2828352451324463,
      "learning_rate": 0.0001986560102015705,
      "loss": 5.04,
      "step": 7210
    },
    {
      "epoch": 0.006730208253104303,
      "grad_norm": 1.2508066892623901,
      "learning_rate": 0.00019865414587947531,
      "loss": 5.1414,
      "step": 7220
    },
    {
      "epoch": 0.006739529871183395,
      "grad_norm": 1.1072447299957275,
      "learning_rate": 0.00019865228155738012,
      "loss": 4.9904,
      "step": 7230
    },
    {
      "epoch": 0.006748851489262487,
      "grad_norm": 1.245607614517212,
      "learning_rate": 0.00019865041723528493,
      "loss": 5.0367,
      "step": 7240
    },
    {
      "epoch": 0.0067581731073415785,
      "grad_norm": 1.1507384777069092,
      "learning_rate": 0.0001986485529131897,
      "loss": 5.0153,
      "step": 7250
    },
    {
      "epoch": 0.00676749472542067,
      "grad_norm": 1.2810331583023071,
      "learning_rate": 0.0001986466885910945,
      "loss": 5.031,
      "step": 7260
    },
    {
      "epoch": 0.006776816343499762,
      "grad_norm": 1.4240797758102417,
      "learning_rate": 0.00019864482426899932,
      "loss": 4.9329,
      "step": 7270
    },
    {
      "epoch": 0.006786137961578854,
      "grad_norm": 1.7313296794891357,
      "learning_rate": 0.00019864295994690413,
      "loss": 4.6968,
      "step": 7280
    },
    {
      "epoch": 0.006795459579657945,
      "grad_norm": 1.1592230796813965,
      "learning_rate": 0.00019864109562480893,
      "loss": 5.015,
      "step": 7290
    },
    {
      "epoch": 0.006804781197737037,
      "grad_norm": 1.2460203170776367,
      "learning_rate": 0.0001986392313027137,
      "loss": 4.8066,
      "step": 7300
    },
    {
      "epoch": 0.006814102815816129,
      "grad_norm": 1.226224660873413,
      "learning_rate": 0.00019863736698061852,
      "loss": 5.1517,
      "step": 7310
    },
    {
      "epoch": 0.006823424433895221,
      "grad_norm": 1.3010423183441162,
      "learning_rate": 0.0001986355026585233,
      "loss": 5.0711,
      "step": 7320
    },
    {
      "epoch": 0.006832746051974313,
      "grad_norm": 1.2032849788665771,
      "learning_rate": 0.00019863363833642813,
      "loss": 5.0585,
      "step": 7330
    },
    {
      "epoch": 0.0068420676700534045,
      "grad_norm": 1.2888182401657104,
      "learning_rate": 0.00019863177401433294,
      "loss": 5.0798,
      "step": 7340
    },
    {
      "epoch": 0.006851389288132496,
      "grad_norm": 1.046831727027893,
      "learning_rate": 0.00019862990969223772,
      "loss": 5.0522,
      "step": 7350
    },
    {
      "epoch": 0.006860710906211588,
      "grad_norm": 1.2347886562347412,
      "learning_rate": 0.00019862804537014252,
      "loss": 4.9889,
      "step": 7360
    },
    {
      "epoch": 0.00687003252429068,
      "grad_norm": 1.349624752998352,
      "learning_rate": 0.0001986261810480473,
      "loss": 4.846,
      "step": 7370
    },
    {
      "epoch": 0.006879354142369772,
      "grad_norm": 1.1261804103851318,
      "learning_rate": 0.0001986243167259521,
      "loss": 4.9107,
      "step": 7380
    },
    {
      "epoch": 0.006888675760448864,
      "grad_norm": 1.5930659770965576,
      "learning_rate": 0.00019862245240385694,
      "loss": 4.6788,
      "step": 7390
    },
    {
      "epoch": 0.006897997378527956,
      "grad_norm": 1.001201868057251,
      "learning_rate": 0.00019862058808176172,
      "loss": 5.0174,
      "step": 7400
    },
    {
      "epoch": 0.006907318996607048,
      "grad_norm": 1.1852569580078125,
      "learning_rate": 0.00019861872375966653,
      "loss": 4.7264,
      "step": 7410
    },
    {
      "epoch": 0.00691664061468614,
      "grad_norm": 1.2771883010864258,
      "learning_rate": 0.0001986168594375713,
      "loss": 4.9309,
      "step": 7420
    },
    {
      "epoch": 0.0069259622327652315,
      "grad_norm": 1.2482151985168457,
      "learning_rate": 0.0001986149951154761,
      "loss": 4.9977,
      "step": 7430
    },
    {
      "epoch": 0.006935283850844323,
      "grad_norm": 1.3599522113800049,
      "learning_rate": 0.00019861313079338092,
      "loss": 5.208,
      "step": 7440
    },
    {
      "epoch": 0.006944605468923415,
      "grad_norm": 1.4065276384353638,
      "learning_rate": 0.00019861126647128572,
      "loss": 4.9862,
      "step": 7450
    },
    {
      "epoch": 0.006953927087002506,
      "grad_norm": 1.2198668718338013,
      "learning_rate": 0.00019860940214919053,
      "loss": 5.0524,
      "step": 7460
    },
    {
      "epoch": 0.006963248705081598,
      "grad_norm": 1.0333727598190308,
      "learning_rate": 0.0001986075378270953,
      "loss": 5.1365,
      "step": 7470
    },
    {
      "epoch": 0.00697257032316069,
      "grad_norm": 1.1777441501617432,
      "learning_rate": 0.00019860567350500012,
      "loss": 4.8469,
      "step": 7480
    },
    {
      "epoch": 0.006981891941239782,
      "grad_norm": 1.0046356916427612,
      "learning_rate": 0.00019860380918290492,
      "loss": 4.8697,
      "step": 7490
    },
    {
      "epoch": 0.006991213559318874,
      "grad_norm": 1.089490294456482,
      "learning_rate": 0.00019860194486080973,
      "loss": 5.0784,
      "step": 7500
    },
    {
      "epoch": 0.007000535177397966,
      "grad_norm": 1.3107203245162964,
      "learning_rate": 0.00019860008053871453,
      "loss": 4.969,
      "step": 7510
    },
    {
      "epoch": 0.0070098567954770576,
      "grad_norm": 1.3819217681884766,
      "learning_rate": 0.00019859821621661931,
      "loss": 5.0524,
      "step": 7520
    },
    {
      "epoch": 0.0070191784135561494,
      "grad_norm": 1.4694207906723022,
      "learning_rate": 0.00019859635189452412,
      "loss": 4.8702,
      "step": 7530
    },
    {
      "epoch": 0.007028500031635241,
      "grad_norm": 1.2436174154281616,
      "learning_rate": 0.00019859448757242893,
      "loss": 4.978,
      "step": 7540
    },
    {
      "epoch": 0.007037821649714333,
      "grad_norm": 1.166760802268982,
      "learning_rate": 0.00019859262325033373,
      "loss": 4.9043,
      "step": 7550
    },
    {
      "epoch": 0.007047143267793425,
      "grad_norm": 1.129717230796814,
      "learning_rate": 0.00019859075892823854,
      "loss": 5.0673,
      "step": 7560
    },
    {
      "epoch": 0.007056464885872517,
      "grad_norm": 1.4361565113067627,
      "learning_rate": 0.00019858889460614332,
      "loss": 4.8452,
      "step": 7570
    },
    {
      "epoch": 0.007065786503951609,
      "grad_norm": 1.3157929182052612,
      "learning_rate": 0.00019858703028404812,
      "loss": 5.0559,
      "step": 7580
    },
    {
      "epoch": 0.007075108122030701,
      "grad_norm": 1.2934677600860596,
      "learning_rate": 0.00019858516596195293,
      "loss": 4.9981,
      "step": 7590
    },
    {
      "epoch": 0.007084429740109793,
      "grad_norm": 1.170439600944519,
      "learning_rate": 0.0001985833016398577,
      "loss": 4.6412,
      "step": 7600
    },
    {
      "epoch": 0.0070937513581888845,
      "grad_norm": 1.3451836109161377,
      "learning_rate": 0.00019858143731776254,
      "loss": 5.1372,
      "step": 7610
    },
    {
      "epoch": 0.007103072976267976,
      "grad_norm": 1.13083016872406,
      "learning_rate": 0.00019857957299566732,
      "loss": 4.7947,
      "step": 7620
    },
    {
      "epoch": 0.007112394594347068,
      "grad_norm": 1.2505085468292236,
      "learning_rate": 0.00019857770867357213,
      "loss": 4.9929,
      "step": 7630
    },
    {
      "epoch": 0.007121716212426159,
      "grad_norm": 1.266624927520752,
      "learning_rate": 0.00019857584435147694,
      "loss": 4.8182,
      "step": 7640
    },
    {
      "epoch": 0.007131037830505251,
      "grad_norm": 1.3006294965744019,
      "learning_rate": 0.00019857398002938171,
      "loss": 4.944,
      "step": 7650
    },
    {
      "epoch": 0.007140359448584343,
      "grad_norm": 1.0467309951782227,
      "learning_rate": 0.00019857211570728652,
      "loss": 4.9764,
      "step": 7660
    },
    {
      "epoch": 0.007149681066663435,
      "grad_norm": 1.0461772680282593,
      "learning_rate": 0.00019857025138519133,
      "loss": 4.9898,
      "step": 7670
    },
    {
      "epoch": 0.007159002684742527,
      "grad_norm": 1.0510647296905518,
      "learning_rate": 0.00019856838706309613,
      "loss": 5.0603,
      "step": 7680
    },
    {
      "epoch": 0.007168324302821619,
      "grad_norm": 1.392264723777771,
      "learning_rate": 0.00019856652274100094,
      "loss": 4.9357,
      "step": 7690
    },
    {
      "epoch": 0.007177645920900711,
      "grad_norm": 1.2270523309707642,
      "learning_rate": 0.00019856465841890572,
      "loss": 4.7602,
      "step": 7700
    },
    {
      "epoch": 0.0071869675389798025,
      "grad_norm": 1.2711174488067627,
      "learning_rate": 0.00019856279409681053,
      "loss": 4.7408,
      "step": 7710
    },
    {
      "epoch": 0.007196289157058894,
      "grad_norm": 1.1699833869934082,
      "learning_rate": 0.00019856092977471533,
      "loss": 4.9702,
      "step": 7720
    },
    {
      "epoch": 0.007205610775137986,
      "grad_norm": 1.2701038122177124,
      "learning_rate": 0.00019855906545262014,
      "loss": 5.006,
      "step": 7730
    },
    {
      "epoch": 0.007214932393217078,
      "grad_norm": 1.2821518182754517,
      "learning_rate": 0.00019855720113052494,
      "loss": 4.9692,
      "step": 7740
    },
    {
      "epoch": 0.00722425401129617,
      "grad_norm": 1.2450379133224487,
      "learning_rate": 0.00019855533680842972,
      "loss": 5.1929,
      "step": 7750
    },
    {
      "epoch": 0.007233575629375262,
      "grad_norm": 1.1742944717407227,
      "learning_rate": 0.00019855347248633453,
      "loss": 4.9997,
      "step": 7760
    },
    {
      "epoch": 0.007242897247454354,
      "grad_norm": 1.184942603111267,
      "learning_rate": 0.0001985516081642393,
      "loss": 4.9437,
      "step": 7770
    },
    {
      "epoch": 0.007252218865533446,
      "grad_norm": 1.2582730054855347,
      "learning_rate": 0.00019854974384214414,
      "loss": 4.9028,
      "step": 7780
    },
    {
      "epoch": 0.0072615404836125375,
      "grad_norm": 1.2326182126998901,
      "learning_rate": 0.00019854787952004892,
      "loss": 4.9333,
      "step": 7790
    },
    {
      "epoch": 0.007270862101691629,
      "grad_norm": 1.358544945716858,
      "learning_rate": 0.00019854601519795373,
      "loss": 4.9593,
      "step": 7800
    },
    {
      "epoch": 0.00728018371977072,
      "grad_norm": 1.300096035003662,
      "learning_rate": 0.00019854415087585853,
      "loss": 4.6981,
      "step": 7810
    },
    {
      "epoch": 0.007289505337849812,
      "grad_norm": 1.2645282745361328,
      "learning_rate": 0.0001985422865537633,
      "loss": 4.8867,
      "step": 7820
    },
    {
      "epoch": 0.007298826955928904,
      "grad_norm": 1.1361085176467896,
      "learning_rate": 0.00019854042223166812,
      "loss": 5.0435,
      "step": 7830
    },
    {
      "epoch": 0.007308148574007996,
      "grad_norm": 1.2992984056472778,
      "learning_rate": 0.00019853855790957293,
      "loss": 5.1261,
      "step": 7840
    },
    {
      "epoch": 0.007317470192087088,
      "grad_norm": 1.0319278240203857,
      "learning_rate": 0.00019853669358747773,
      "loss": 4.8696,
      "step": 7850
    },
    {
      "epoch": 0.00732679181016618,
      "grad_norm": 1.4849456548690796,
      "learning_rate": 0.00019853482926538254,
      "loss": 4.8262,
      "step": 7860
    },
    {
      "epoch": 0.007336113428245272,
      "grad_norm": 1.134456753730774,
      "learning_rate": 0.00019853296494328732,
      "loss": 4.837,
      "step": 7870
    },
    {
      "epoch": 0.007345435046324364,
      "grad_norm": 1.2650009393692017,
      "learning_rate": 0.00019853110062119212,
      "loss": 4.8709,
      "step": 7880
    },
    {
      "epoch": 0.0073547566644034555,
      "grad_norm": 1.1279746294021606,
      "learning_rate": 0.00019852923629909693,
      "loss": 4.7366,
      "step": 7890
    },
    {
      "epoch": 0.007364078282482547,
      "grad_norm": 1.1262634992599487,
      "learning_rate": 0.00019852737197700174,
      "loss": 5.1378,
      "step": 7900
    },
    {
      "epoch": 0.007373399900561639,
      "grad_norm": 1.1410820484161377,
      "learning_rate": 0.00019852550765490654,
      "loss": 4.8654,
      "step": 7910
    },
    {
      "epoch": 0.007382721518640731,
      "grad_norm": 1.239662766456604,
      "learning_rate": 0.00019852364333281132,
      "loss": 4.996,
      "step": 7920
    },
    {
      "epoch": 0.007392043136719823,
      "grad_norm": 1.1934696435928345,
      "learning_rate": 0.00019852177901071613,
      "loss": 4.7916,
      "step": 7930
    },
    {
      "epoch": 0.007401364754798915,
      "grad_norm": 1.1759147644042969,
      "learning_rate": 0.00019851991468862094,
      "loss": 5.0046,
      "step": 7940
    },
    {
      "epoch": 0.007410686372878007,
      "grad_norm": 1.1766819953918457,
      "learning_rate": 0.00019851805036652574,
      "loss": 4.8226,
      "step": 7950
    },
    {
      "epoch": 0.007420007990957099,
      "grad_norm": 1.3104219436645508,
      "learning_rate": 0.00019851618604443055,
      "loss": 5.1468,
      "step": 7960
    },
    {
      "epoch": 0.0074293296090361905,
      "grad_norm": 1.3569045066833496,
      "learning_rate": 0.00019851432172233533,
      "loss": 4.9987,
      "step": 7970
    },
    {
      "epoch": 0.0074386512271152816,
      "grad_norm": 1.3867515325546265,
      "learning_rate": 0.00019851245740024013,
      "loss": 4.8198,
      "step": 7980
    },
    {
      "epoch": 0.0074479728451943734,
      "grad_norm": 1.0835630893707275,
      "learning_rate": 0.0001985105930781449,
      "loss": 4.9173,
      "step": 7990
    },
    {
      "epoch": 0.007457294463273465,
      "grad_norm": 1.0880416631698608,
      "learning_rate": 0.00019850872875604975,
      "loss": 5.0028,
      "step": 8000
    },
    {
      "epoch": 0.007466616081352557,
      "grad_norm": 1.5901485681533813,
      "learning_rate": 0.00019850686443395455,
      "loss": 4.8541,
      "step": 8010
    },
    {
      "epoch": 0.007475937699431649,
      "grad_norm": 1.121887445449829,
      "learning_rate": 0.00019850500011185933,
      "loss": 4.9384,
      "step": 8020
    },
    {
      "epoch": 0.007485259317510741,
      "grad_norm": 1.410270094871521,
      "learning_rate": 0.00019850313578976414,
      "loss": 4.8822,
      "step": 8030
    },
    {
      "epoch": 0.007494580935589833,
      "grad_norm": 1.0813324451446533,
      "learning_rate": 0.00019850127146766892,
      "loss": 4.8157,
      "step": 8040
    },
    {
      "epoch": 0.007503902553668925,
      "grad_norm": 1.4358775615692139,
      "learning_rate": 0.00019849940714557372,
      "loss": 4.866,
      "step": 8050
    },
    {
      "epoch": 0.007513224171748017,
      "grad_norm": 1.3267444372177124,
      "learning_rate": 0.00019849754282347856,
      "loss": 4.8094,
      "step": 8060
    },
    {
      "epoch": 0.0075225457898271085,
      "grad_norm": 1.1626331806182861,
      "learning_rate": 0.00019849567850138334,
      "loss": 5.0263,
      "step": 8070
    },
    {
      "epoch": 0.0075318674079062,
      "grad_norm": 1.3010979890823364,
      "learning_rate": 0.00019849381417928814,
      "loss": 4.8239,
      "step": 8080
    },
    {
      "epoch": 0.007541189025985292,
      "grad_norm": 1.284131646156311,
      "learning_rate": 0.00019849194985719292,
      "loss": 5.027,
      "step": 8090
    },
    {
      "epoch": 0.007550510644064384,
      "grad_norm": 1.3099353313446045,
      "learning_rate": 0.00019849008553509773,
      "loss": 4.9311,
      "step": 8100
    },
    {
      "epoch": 0.007559832262143476,
      "grad_norm": 1.3480644226074219,
      "learning_rate": 0.00019848822121300253,
      "loss": 4.9997,
      "step": 8110
    },
    {
      "epoch": 0.007569153880222568,
      "grad_norm": 1.315985918045044,
      "learning_rate": 0.00019848635689090734,
      "loss": 4.7532,
      "step": 8120
    },
    {
      "epoch": 0.00757847549830166,
      "grad_norm": 1.149256944656372,
      "learning_rate": 0.00019848449256881215,
      "loss": 4.9491,
      "step": 8130
    },
    {
      "epoch": 0.007587797116380752,
      "grad_norm": 1.0530896186828613,
      "learning_rate": 0.00019848262824671693,
      "loss": 4.9384,
      "step": 8140
    },
    {
      "epoch": 0.007597118734459843,
      "grad_norm": 1.0257306098937988,
      "learning_rate": 0.00019848076392462173,
      "loss": 4.7444,
      "step": 8150
    },
    {
      "epoch": 0.007606440352538935,
      "grad_norm": 1.16636323928833,
      "learning_rate": 0.00019847889960252654,
      "loss": 5.006,
      "step": 8160
    },
    {
      "epoch": 0.0076157619706180265,
      "grad_norm": 1.4347087144851685,
      "learning_rate": 0.00019847703528043134,
      "loss": 5.0268,
      "step": 8170
    },
    {
      "epoch": 0.007625083588697118,
      "grad_norm": 1.2283785343170166,
      "learning_rate": 0.00019847517095833615,
      "loss": 4.9417,
      "step": 8180
    },
    {
      "epoch": 0.00763440520677621,
      "grad_norm": 1.1857713460922241,
      "learning_rate": 0.00019847330663624093,
      "loss": 4.9205,
      "step": 8190
    },
    {
      "epoch": 0.007643726824855302,
      "grad_norm": 1.0805878639221191,
      "learning_rate": 0.00019847144231414574,
      "loss": 4.9388,
      "step": 8200
    },
    {
      "epoch": 0.007653048442934394,
      "grad_norm": 1.1075712442398071,
      "learning_rate": 0.00019846957799205054,
      "loss": 4.7611,
      "step": 8210
    },
    {
      "epoch": 0.007662370061013486,
      "grad_norm": 1.085996150970459,
      "learning_rate": 0.00019846771366995532,
      "loss": 5.1234,
      "step": 8220
    },
    {
      "epoch": 0.007671691679092578,
      "grad_norm": 1.3047330379486084,
      "learning_rate": 0.00019846584934786016,
      "loss": 4.8907,
      "step": 8230
    },
    {
      "epoch": 0.00768101329717167,
      "grad_norm": 1.1024049520492554,
      "learning_rate": 0.00019846398502576493,
      "loss": 4.9576,
      "step": 8240
    },
    {
      "epoch": 0.0076903349152507615,
      "grad_norm": 1.0765289068222046,
      "learning_rate": 0.00019846212070366974,
      "loss": 4.7318,
      "step": 8250
    },
    {
      "epoch": 0.007699656533329853,
      "grad_norm": 1.0877125263214111,
      "learning_rate": 0.00019846025638157455,
      "loss": 4.8811,
      "step": 8260
    },
    {
      "epoch": 0.007708978151408945,
      "grad_norm": 1.410581111907959,
      "learning_rate": 0.00019845839205947933,
      "loss": 4.8602,
      "step": 8270
    },
    {
      "epoch": 0.007718299769488037,
      "grad_norm": 1.2539088726043701,
      "learning_rate": 0.00019845652773738413,
      "loss": 5.0174,
      "step": 8280
    },
    {
      "epoch": 0.007727621387567129,
      "grad_norm": 1.0088372230529785,
      "learning_rate": 0.00019845466341528894,
      "loss": 4.9453,
      "step": 8290
    },
    {
      "epoch": 0.007736943005646221,
      "grad_norm": 0.9264483451843262,
      "learning_rate": 0.00019845279909319375,
      "loss": 4.873,
      "step": 8300
    },
    {
      "epoch": 0.007746264623725313,
      "grad_norm": 1.5218459367752075,
      "learning_rate": 0.00019845093477109855,
      "loss": 5.0382,
      "step": 8310
    },
    {
      "epoch": 0.007755586241804404,
      "grad_norm": 1.2409844398498535,
      "learning_rate": 0.00019844907044900333,
      "loss": 4.9954,
      "step": 8320
    },
    {
      "epoch": 0.007764907859883496,
      "grad_norm": 1.1471346616744995,
      "learning_rate": 0.00019844720612690814,
      "loss": 4.6695,
      "step": 8330
    },
    {
      "epoch": 0.007774229477962588,
      "grad_norm": 1.3388370275497437,
      "learning_rate": 0.00019844534180481294,
      "loss": 4.6883,
      "step": 8340
    },
    {
      "epoch": 0.0077835510960416795,
      "grad_norm": 1.3641612529754639,
      "learning_rate": 0.00019844347748271775,
      "loss": 4.9652,
      "step": 8350
    },
    {
      "epoch": 0.007792872714120771,
      "grad_norm": 1.1500155925750732,
      "learning_rate": 0.00019844161316062256,
      "loss": 5.0137,
      "step": 8360
    },
    {
      "epoch": 0.007802194332199863,
      "grad_norm": 1.3136457204818726,
      "learning_rate": 0.00019843974883852734,
      "loss": 4.9993,
      "step": 8370
    },
    {
      "epoch": 0.007811515950278955,
      "grad_norm": 1.1722360849380493,
      "learning_rate": 0.00019843788451643214,
      "loss": 4.8562,
      "step": 8380
    },
    {
      "epoch": 0.007820837568358047,
      "grad_norm": 1.0925616025924683,
      "learning_rate": 0.00019843602019433695,
      "loss": 4.9766,
      "step": 8390
    },
    {
      "epoch": 0.00783015918643714,
      "grad_norm": 1.10353422164917,
      "learning_rate": 0.00019843415587224175,
      "loss": 4.9667,
      "step": 8400
    },
    {
      "epoch": 0.00783948080451623,
      "grad_norm": 1.2870936393737793,
      "learning_rate": 0.00019843229155014656,
      "loss": 4.8575,
      "step": 8410
    },
    {
      "epoch": 0.007848802422595322,
      "grad_norm": 1.4286662340164185,
      "learning_rate": 0.00019843042722805134,
      "loss": 4.8572,
      "step": 8420
    },
    {
      "epoch": 0.007858124040674415,
      "grad_norm": 1.3647680282592773,
      "learning_rate": 0.00019842856290595615,
      "loss": 4.8778,
      "step": 8430
    },
    {
      "epoch": 0.007867445658753506,
      "grad_norm": 1.2856314182281494,
      "learning_rate": 0.00019842669858386093,
      "loss": 4.8759,
      "step": 8440
    },
    {
      "epoch": 0.007876767276832598,
      "grad_norm": 1.235300898551941,
      "learning_rate": 0.00019842483426176576,
      "loss": 4.8743,
      "step": 8450
    },
    {
      "epoch": 0.00788608889491169,
      "grad_norm": 1.3286820650100708,
      "learning_rate": 0.00019842296993967056,
      "loss": 4.9838,
      "step": 8460
    },
    {
      "epoch": 0.007895410512990782,
      "grad_norm": 1.1737980842590332,
      "learning_rate": 0.00019842110561757534,
      "loss": 4.7706,
      "step": 8470
    },
    {
      "epoch": 0.007904732131069873,
      "grad_norm": 1.2003448009490967,
      "learning_rate": 0.00019841924129548015,
      "loss": 4.851,
      "step": 8480
    },
    {
      "epoch": 0.007914053749148966,
      "grad_norm": 1.2154419422149658,
      "learning_rate": 0.00019841737697338493,
      "loss": 4.8568,
      "step": 8490
    },
    {
      "epoch": 0.007923375367228057,
      "grad_norm": 1.2772656679153442,
      "learning_rate": 0.00019841551265128974,
      "loss": 4.884,
      "step": 8500
    },
    {
      "epoch": 0.00793269698530715,
      "grad_norm": 1.2344199419021606,
      "learning_rate": 0.00019841364832919457,
      "loss": 4.8544,
      "step": 8510
    },
    {
      "epoch": 0.00794201860338624,
      "grad_norm": 1.0093748569488525,
      "learning_rate": 0.00019841178400709935,
      "loss": 5.0694,
      "step": 8520
    },
    {
      "epoch": 0.007951340221465333,
      "grad_norm": 1.4661098718643188,
      "learning_rate": 0.00019840991968500415,
      "loss": 4.8269,
      "step": 8530
    },
    {
      "epoch": 0.007960661839544424,
      "grad_norm": 1.2956877946853638,
      "learning_rate": 0.00019840805536290893,
      "loss": 5.0715,
      "step": 8540
    },
    {
      "epoch": 0.007969983457623517,
      "grad_norm": 1.1796408891677856,
      "learning_rate": 0.00019840619104081374,
      "loss": 4.9344,
      "step": 8550
    },
    {
      "epoch": 0.007979305075702608,
      "grad_norm": 1.239906907081604,
      "learning_rate": 0.00019840432671871855,
      "loss": 4.7108,
      "step": 8560
    },
    {
      "epoch": 0.007988626693781701,
      "grad_norm": 1.2449662685394287,
      "learning_rate": 0.00019840246239662335,
      "loss": 4.7535,
      "step": 8570
    },
    {
      "epoch": 0.007997948311860792,
      "grad_norm": 1.3108882904052734,
      "learning_rate": 0.00019840059807452816,
      "loss": 4.9456,
      "step": 8580
    },
    {
      "epoch": 0.008007269929939883,
      "grad_norm": 1.0946863889694214,
      "learning_rate": 0.00019839873375243294,
      "loss": 4.8821,
      "step": 8590
    },
    {
      "epoch": 0.008016591548018976,
      "grad_norm": 1.3844107389450073,
      "learning_rate": 0.00019839686943033774,
      "loss": 4.826,
      "step": 8600
    },
    {
      "epoch": 0.008025913166098067,
      "grad_norm": 1.3447933197021484,
      "learning_rate": 0.00019839500510824255,
      "loss": 4.8677,
      "step": 8610
    },
    {
      "epoch": 0.00803523478417716,
      "grad_norm": 1.4394859075546265,
      "learning_rate": 0.00019839314078614736,
      "loss": 4.8709,
      "step": 8620
    },
    {
      "epoch": 0.00804455640225625,
      "grad_norm": 0.9999630451202393,
      "learning_rate": 0.00019839127646405216,
      "loss": 4.7333,
      "step": 8630
    },
    {
      "epoch": 0.008053878020335343,
      "grad_norm": 1.1669450998306274,
      "learning_rate": 0.00019838941214195694,
      "loss": 4.9496,
      "step": 8640
    },
    {
      "epoch": 0.008063199638414434,
      "grad_norm": 1.2367215156555176,
      "learning_rate": 0.00019838754781986175,
      "loss": 4.9084,
      "step": 8650
    },
    {
      "epoch": 0.008072521256493527,
      "grad_norm": 1.414663553237915,
      "learning_rate": 0.00019838568349776656,
      "loss": 4.9481,
      "step": 8660
    },
    {
      "epoch": 0.008081842874572618,
      "grad_norm": 1.1995890140533447,
      "learning_rate": 0.00019838381917567133,
      "loss": 5.0665,
      "step": 8670
    },
    {
      "epoch": 0.00809116449265171,
      "grad_norm": 0.9468734264373779,
      "learning_rate": 0.00019838195485357617,
      "loss": 4.9377,
      "step": 8680
    },
    {
      "epoch": 0.008100486110730802,
      "grad_norm": 1.1326203346252441,
      "learning_rate": 0.00019838009053148095,
      "loss": 4.9239,
      "step": 8690
    },
    {
      "epoch": 0.008109807728809895,
      "grad_norm": 1.185937523841858,
      "learning_rate": 0.00019837822620938575,
      "loss": 4.9139,
      "step": 8700
    },
    {
      "epoch": 0.008119129346888986,
      "grad_norm": 1.154168725013733,
      "learning_rate": 0.00019837636188729056,
      "loss": 4.9283,
      "step": 8710
    },
    {
      "epoch": 0.008128450964968078,
      "grad_norm": 1.4477354288101196,
      "learning_rate": 0.00019837449756519534,
      "loss": 4.9419,
      "step": 8720
    },
    {
      "epoch": 0.00813777258304717,
      "grad_norm": 1.2909587621688843,
      "learning_rate": 0.00019837263324310015,
      "loss": 4.8166,
      "step": 8730
    },
    {
      "epoch": 0.008147094201126262,
      "grad_norm": 1.159610629081726,
      "learning_rate": 0.00019837076892100495,
      "loss": 4.9629,
      "step": 8740
    },
    {
      "epoch": 0.008156415819205353,
      "grad_norm": 1.424314260482788,
      "learning_rate": 0.00019836890459890976,
      "loss": 4.9083,
      "step": 8750
    },
    {
      "epoch": 0.008165737437284444,
      "grad_norm": 1.050408124923706,
      "learning_rate": 0.00019836704027681456,
      "loss": 5.0389,
      "step": 8760
    },
    {
      "epoch": 0.008175059055363537,
      "grad_norm": 1.3990635871887207,
      "learning_rate": 0.00019836517595471934,
      "loss": 4.7955,
      "step": 8770
    },
    {
      "epoch": 0.008184380673442628,
      "grad_norm": 1.11854887008667,
      "learning_rate": 0.00019836331163262415,
      "loss": 4.9807,
      "step": 8780
    },
    {
      "epoch": 0.00819370229152172,
      "grad_norm": 1.3124966621398926,
      "learning_rate": 0.00019836144731052896,
      "loss": 4.9234,
      "step": 8790
    },
    {
      "epoch": 0.008203023909600812,
      "grad_norm": 1.2994333505630493,
      "learning_rate": 0.00019835958298843376,
      "loss": 4.7101,
      "step": 8800
    },
    {
      "epoch": 0.008212345527679904,
      "grad_norm": 1.2235041856765747,
      "learning_rate": 0.00019835771866633857,
      "loss": 4.8215,
      "step": 8810
    },
    {
      "epoch": 0.008221667145758995,
      "grad_norm": 1.1524461507797241,
      "learning_rate": 0.00019835585434424335,
      "loss": 5.0777,
      "step": 8820
    },
    {
      "epoch": 0.008230988763838088,
      "grad_norm": 1.2827242612838745,
      "learning_rate": 0.00019835399002214815,
      "loss": 4.8862,
      "step": 8830
    },
    {
      "epoch": 0.008240310381917179,
      "grad_norm": 1.3947824239730835,
      "learning_rate": 0.00019835212570005296,
      "loss": 4.6716,
      "step": 8840
    },
    {
      "epoch": 0.008249631999996272,
      "grad_norm": 1.149269700050354,
      "learning_rate": 0.00019835026137795777,
      "loss": 4.8858,
      "step": 8850
    },
    {
      "epoch": 0.008258953618075363,
      "grad_norm": 1.2473933696746826,
      "learning_rate": 0.00019834839705586257,
      "loss": 4.6884,
      "step": 8860
    },
    {
      "epoch": 0.008268275236154456,
      "grad_norm": 1.4413721561431885,
      "learning_rate": 0.00019834653273376735,
      "loss": 4.732,
      "step": 8870
    },
    {
      "epoch": 0.008277596854233547,
      "grad_norm": 1.0344067811965942,
      "learning_rate": 0.00019834466841167216,
      "loss": 4.9186,
      "step": 8880
    },
    {
      "epoch": 0.00828691847231264,
      "grad_norm": 1.1910630464553833,
      "learning_rate": 0.00019834280408957694,
      "loss": 4.5802,
      "step": 8890
    },
    {
      "epoch": 0.00829624009039173,
      "grad_norm": 1.2269830703735352,
      "learning_rate": 0.00019834093976748177,
      "loss": 4.8639,
      "step": 8900
    },
    {
      "epoch": 0.008305561708470823,
      "grad_norm": 1.0469430685043335,
      "learning_rate": 0.00019833907544538658,
      "loss": 4.7159,
      "step": 8910
    },
    {
      "epoch": 0.008314883326549914,
      "grad_norm": 1.2315315008163452,
      "learning_rate": 0.00019833721112329136,
      "loss": 4.7507,
      "step": 8920
    },
    {
      "epoch": 0.008324204944629005,
      "grad_norm": 1.1551271677017212,
      "learning_rate": 0.00019833534680119616,
      "loss": 4.7939,
      "step": 8930
    },
    {
      "epoch": 0.008333526562708098,
      "grad_norm": 1.3504985570907593,
      "learning_rate": 0.00019833348247910094,
      "loss": 4.806,
      "step": 8940
    },
    {
      "epoch": 0.008342848180787189,
      "grad_norm": 1.1749438047409058,
      "learning_rate": 0.00019833161815700575,
      "loss": 4.9503,
      "step": 8950
    },
    {
      "epoch": 0.008352169798866282,
      "grad_norm": 1.3431276082992554,
      "learning_rate": 0.00019832975383491058,
      "loss": 4.8504,
      "step": 8960
    },
    {
      "epoch": 0.008361491416945373,
      "grad_norm": 0.9832585453987122,
      "learning_rate": 0.00019832788951281536,
      "loss": 4.7243,
      "step": 8970
    },
    {
      "epoch": 0.008370813035024465,
      "grad_norm": 1.152896761894226,
      "learning_rate": 0.00019832602519072017,
      "loss": 4.8875,
      "step": 8980
    },
    {
      "epoch": 0.008380134653103556,
      "grad_norm": 1.4227979183197021,
      "learning_rate": 0.00019832416086862495,
      "loss": 4.6294,
      "step": 8990
    },
    {
      "epoch": 0.00838945627118265,
      "grad_norm": 0.9931693077087402,
      "learning_rate": 0.00019832229654652975,
      "loss": 4.7832,
      "step": 9000
    },
    {
      "epoch": 0.00839877788926174,
      "grad_norm": 1.3889684677124023,
      "learning_rate": 0.00019832043222443456,
      "loss": 4.7616,
      "step": 9010
    },
    {
      "epoch": 0.008408099507340833,
      "grad_norm": 1.4459487199783325,
      "learning_rate": 0.00019831856790233937,
      "loss": 4.8168,
      "step": 9020
    },
    {
      "epoch": 0.008417421125419924,
      "grad_norm": 1.2768502235412598,
      "learning_rate": 0.00019831670358024417,
      "loss": 4.6998,
      "step": 9030
    },
    {
      "epoch": 0.008426742743499017,
      "grad_norm": 1.1393910646438599,
      "learning_rate": 0.00019831483925814895,
      "loss": 4.653,
      "step": 9040
    },
    {
      "epoch": 0.008436064361578108,
      "grad_norm": 1.0008556842803955,
      "learning_rate": 0.00019831297493605376,
      "loss": 4.8449,
      "step": 9050
    },
    {
      "epoch": 0.0084453859796572,
      "grad_norm": 1.121551275253296,
      "learning_rate": 0.00019831111061395856,
      "loss": 4.9346,
      "step": 9060
    },
    {
      "epoch": 0.008454707597736292,
      "grad_norm": 1.1285264492034912,
      "learning_rate": 0.00019830924629186337,
      "loss": 4.842,
      "step": 9070
    },
    {
      "epoch": 0.008464029215815384,
      "grad_norm": 1.2769922018051147,
      "learning_rate": 0.00019830738196976818,
      "loss": 4.9438,
      "step": 9080
    },
    {
      "epoch": 0.008473350833894475,
      "grad_norm": 1.2200872898101807,
      "learning_rate": 0.00019830551764767296,
      "loss": 4.7913,
      "step": 9090
    },
    {
      "epoch": 0.008482672451973566,
      "grad_norm": 1.0659664869308472,
      "learning_rate": 0.00019830365332557776,
      "loss": 4.9716,
      "step": 9100
    },
    {
      "epoch": 0.008491994070052659,
      "grad_norm": 0.918128252029419,
      "learning_rate": 0.00019830178900348257,
      "loss": 4.8358,
      "step": 9110
    },
    {
      "epoch": 0.00850131568813175,
      "grad_norm": 1.352206826210022,
      "learning_rate": 0.00019829992468138737,
      "loss": 4.7573,
      "step": 9120
    },
    {
      "epoch": 0.008510637306210843,
      "grad_norm": 1.234545350074768,
      "learning_rate": 0.00019829806035929218,
      "loss": 4.7998,
      "step": 9130
    },
    {
      "epoch": 0.008519958924289934,
      "grad_norm": 1.137048363685608,
      "learning_rate": 0.00019829619603719696,
      "loss": 4.9472,
      "step": 9140
    },
    {
      "epoch": 0.008529280542369027,
      "grad_norm": 1.1441763639450073,
      "learning_rate": 0.00019829433171510177,
      "loss": 4.9086,
      "step": 9150
    },
    {
      "epoch": 0.008538602160448118,
      "grad_norm": 1.1410131454467773,
      "learning_rate": 0.00019829246739300655,
      "loss": 4.9739,
      "step": 9160
    },
    {
      "epoch": 0.00854792377852721,
      "grad_norm": 1.160415768623352,
      "learning_rate": 0.00019829060307091135,
      "loss": 4.7108,
      "step": 9170
    },
    {
      "epoch": 0.008557245396606301,
      "grad_norm": 1.2658864259719849,
      "learning_rate": 0.00019828873874881616,
      "loss": 4.6501,
      "step": 9180
    },
    {
      "epoch": 0.008566567014685394,
      "grad_norm": 1.1291899681091309,
      "learning_rate": 0.00019828687442672096,
      "loss": 4.7625,
      "step": 9190
    },
    {
      "epoch": 0.008575888632764485,
      "grad_norm": 1.1557096242904663,
      "learning_rate": 0.00019828501010462577,
      "loss": 4.8094,
      "step": 9200
    },
    {
      "epoch": 0.008585210250843578,
      "grad_norm": 1.1623601913452148,
      "learning_rate": 0.00019828314578253055,
      "loss": 4.7157,
      "step": 9210
    },
    {
      "epoch": 0.008594531868922669,
      "grad_norm": 1.1856416463851929,
      "learning_rate": 0.00019828128146043536,
      "loss": 4.7876,
      "step": 9220
    },
    {
      "epoch": 0.008603853487001762,
      "grad_norm": 1.395887017250061,
      "learning_rate": 0.00019827941713834016,
      "loss": 4.6615,
      "step": 9230
    },
    {
      "epoch": 0.008613175105080853,
      "grad_norm": 1.2434101104736328,
      "learning_rate": 0.00019827755281624497,
      "loss": 4.974,
      "step": 9240
    },
    {
      "epoch": 0.008622496723159945,
      "grad_norm": 1.2079405784606934,
      "learning_rate": 0.00019827568849414978,
      "loss": 4.9265,
      "step": 9250
    },
    {
      "epoch": 0.008631818341239036,
      "grad_norm": 1.4918084144592285,
      "learning_rate": 0.00019827382417205455,
      "loss": 4.9036,
      "step": 9260
    },
    {
      "epoch": 0.008641139959318127,
      "grad_norm": 0.9401279091835022,
      "learning_rate": 0.00019827195984995936,
      "loss": 4.8394,
      "step": 9270
    },
    {
      "epoch": 0.00865046157739722,
      "grad_norm": 1.317458987236023,
      "learning_rate": 0.00019827009552786417,
      "loss": 4.8372,
      "step": 9280
    },
    {
      "epoch": 0.008659783195476311,
      "grad_norm": 1.0999629497528076,
      "learning_rate": 0.00019826823120576897,
      "loss": 4.913,
      "step": 9290
    },
    {
      "epoch": 0.008669104813555404,
      "grad_norm": 1.1139938831329346,
      "learning_rate": 0.00019826636688367378,
      "loss": 4.7787,
      "step": 9300
    },
    {
      "epoch": 0.008678426431634495,
      "grad_norm": 1.3336213827133179,
      "learning_rate": 0.00019826450256157856,
      "loss": 4.8642,
      "step": 9310
    },
    {
      "epoch": 0.008687748049713588,
      "grad_norm": 1.2616091966629028,
      "learning_rate": 0.00019826263823948337,
      "loss": 4.7901,
      "step": 9320
    },
    {
      "epoch": 0.008697069667792679,
      "grad_norm": 1.3572897911071777,
      "learning_rate": 0.00019826077391738817,
      "loss": 4.6947,
      "step": 9330
    },
    {
      "epoch": 0.008706391285871772,
      "grad_norm": 1.1859502792358398,
      "learning_rate": 0.00019825890959529295,
      "loss": 4.6994,
      "step": 9340
    },
    {
      "epoch": 0.008715712903950863,
      "grad_norm": 1.10474693775177,
      "learning_rate": 0.00019825704527319778,
      "loss": 5.0597,
      "step": 9350
    },
    {
      "epoch": 0.008725034522029955,
      "grad_norm": 1.2649837732315063,
      "learning_rate": 0.00019825518095110256,
      "loss": 5.0487,
      "step": 9360
    },
    {
      "epoch": 0.008734356140109046,
      "grad_norm": 1.035142183303833,
      "learning_rate": 0.00019825331662900737,
      "loss": 4.747,
      "step": 9370
    },
    {
      "epoch": 0.008743677758188139,
      "grad_norm": 1.283711552619934,
      "learning_rate": 0.00019825145230691218,
      "loss": 4.717,
      "step": 9380
    },
    {
      "epoch": 0.00875299937626723,
      "grad_norm": 1.5331225395202637,
      "learning_rate": 0.00019824958798481696,
      "loss": 4.7722,
      "step": 9390
    },
    {
      "epoch": 0.008762320994346323,
      "grad_norm": 1.3223083019256592,
      "learning_rate": 0.00019824772366272176,
      "loss": 4.743,
      "step": 9400
    },
    {
      "epoch": 0.008771642612425414,
      "grad_norm": 1.3299753665924072,
      "learning_rate": 0.00019824585934062657,
      "loss": 4.8884,
      "step": 9410
    },
    {
      "epoch": 0.008780964230504507,
      "grad_norm": 1.3363990783691406,
      "learning_rate": 0.00019824399501853137,
      "loss": 4.5812,
      "step": 9420
    },
    {
      "epoch": 0.008790285848583598,
      "grad_norm": 1.1496776342391968,
      "learning_rate": 0.00019824213069643618,
      "loss": 4.9454,
      "step": 9430
    },
    {
      "epoch": 0.008799607466662689,
      "grad_norm": 1.391535997390747,
      "learning_rate": 0.00019824026637434096,
      "loss": 4.5673,
      "step": 9440
    },
    {
      "epoch": 0.008808929084741781,
      "grad_norm": 1.31840980052948,
      "learning_rate": 0.00019823840205224577,
      "loss": 4.597,
      "step": 9450
    },
    {
      "epoch": 0.008818250702820872,
      "grad_norm": 1.198955774307251,
      "learning_rate": 0.00019823653773015057,
      "loss": 4.766,
      "step": 9460
    },
    {
      "epoch": 0.008827572320899965,
      "grad_norm": 1.2395434379577637,
      "learning_rate": 0.00019823467340805538,
      "loss": 4.8353,
      "step": 9470
    },
    {
      "epoch": 0.008836893938979056,
      "grad_norm": 1.3378719091415405,
      "learning_rate": 0.00019823280908596018,
      "loss": 4.7209,
      "step": 9480
    },
    {
      "epoch": 0.008846215557058149,
      "grad_norm": 1.4461698532104492,
      "learning_rate": 0.00019823094476386496,
      "loss": 4.9664,
      "step": 9490
    },
    {
      "epoch": 0.00885553717513724,
      "grad_norm": 1.2044495344161987,
      "learning_rate": 0.00019822908044176977,
      "loss": 4.7298,
      "step": 9500
    },
    {
      "epoch": 0.008864858793216333,
      "grad_norm": 1.1503856182098389,
      "learning_rate": 0.00019822721611967458,
      "loss": 4.5852,
      "step": 9510
    },
    {
      "epoch": 0.008874180411295424,
      "grad_norm": 1.0005714893341064,
      "learning_rate": 0.00019822535179757938,
      "loss": 4.8101,
      "step": 9520
    },
    {
      "epoch": 0.008883502029374516,
      "grad_norm": 1.0689749717712402,
      "learning_rate": 0.0001982234874754842,
      "loss": 4.8804,
      "step": 9530
    },
    {
      "epoch": 0.008892823647453607,
      "grad_norm": 1.04464590549469,
      "learning_rate": 0.00019822162315338897,
      "loss": 4.75,
      "step": 9540
    },
    {
      "epoch": 0.0089021452655327,
      "grad_norm": 1.151863694190979,
      "learning_rate": 0.00019821975883129377,
      "loss": 4.8543,
      "step": 9550
    },
    {
      "epoch": 0.008911466883611791,
      "grad_norm": 1.1223912239074707,
      "learning_rate": 0.00019821789450919855,
      "loss": 4.5585,
      "step": 9560
    },
    {
      "epoch": 0.008920788501690884,
      "grad_norm": 1.2711237668991089,
      "learning_rate": 0.0001982160301871034,
      "loss": 4.6842,
      "step": 9570
    },
    {
      "epoch": 0.008930110119769975,
      "grad_norm": 1.1822965145111084,
      "learning_rate": 0.0001982141658650082,
      "loss": 4.9103,
      "step": 9580
    },
    {
      "epoch": 0.008939431737849068,
      "grad_norm": 1.2243831157684326,
      "learning_rate": 0.00019821230154291297,
      "loss": 4.5677,
      "step": 9590
    },
    {
      "epoch": 0.008948753355928159,
      "grad_norm": 1.2047131061553955,
      "learning_rate": 0.00019821043722081778,
      "loss": 4.8234,
      "step": 9600
    },
    {
      "epoch": 0.00895807497400725,
      "grad_norm": 1.2713921070098877,
      "learning_rate": 0.00019820857289872256,
      "loss": 4.9848,
      "step": 9610
    },
    {
      "epoch": 0.008967396592086343,
      "grad_norm": 1.2765344381332397,
      "learning_rate": 0.00019820670857662736,
      "loss": 4.8558,
      "step": 9620
    },
    {
      "epoch": 0.008976718210165434,
      "grad_norm": 1.3247177600860596,
      "learning_rate": 0.0001982048442545322,
      "loss": 4.4905,
      "step": 9630
    },
    {
      "epoch": 0.008986039828244526,
      "grad_norm": 1.0568859577178955,
      "learning_rate": 0.00019820297993243698,
      "loss": 4.8378,
      "step": 9640
    },
    {
      "epoch": 0.008995361446323617,
      "grad_norm": 1.2907098531723022,
      "learning_rate": 0.00019820111561034178,
      "loss": 4.96,
      "step": 9650
    },
    {
      "epoch": 0.00900468306440271,
      "grad_norm": 1.161401391029358,
      "learning_rate": 0.00019819925128824656,
      "loss": 4.806,
      "step": 9660
    },
    {
      "epoch": 0.009014004682481801,
      "grad_norm": 1.1828622817993164,
      "learning_rate": 0.00019819738696615137,
      "loss": 4.8549,
      "step": 9670
    },
    {
      "epoch": 0.009023326300560894,
      "grad_norm": 1.4585344791412354,
      "learning_rate": 0.00019819552264405618,
      "loss": 4.8369,
      "step": 9680
    },
    {
      "epoch": 0.009032647918639985,
      "grad_norm": 1.1201896667480469,
      "learning_rate": 0.00019819365832196098,
      "loss": 4.7791,
      "step": 9690
    },
    {
      "epoch": 0.009041969536719078,
      "grad_norm": 1.2823015451431274,
      "learning_rate": 0.0001981917939998658,
      "loss": 4.7583,
      "step": 9700
    },
    {
      "epoch": 0.009051291154798169,
      "grad_norm": 1.449570894241333,
      "learning_rate": 0.00019818992967777057,
      "loss": 4.8239,
      "step": 9710
    },
    {
      "epoch": 0.009060612772877261,
      "grad_norm": 1.1524322032928467,
      "learning_rate": 0.00019818806535567537,
      "loss": 4.6444,
      "step": 9720
    },
    {
      "epoch": 0.009069934390956352,
      "grad_norm": 1.1083277463912964,
      "learning_rate": 0.00019818620103358018,
      "loss": 4.8278,
      "step": 9730
    },
    {
      "epoch": 0.009079256009035445,
      "grad_norm": 1.3693807125091553,
      "learning_rate": 0.00019818433671148499,
      "loss": 4.8672,
      "step": 9740
    },
    {
      "epoch": 0.009088577627114536,
      "grad_norm": 1.428528070449829,
      "learning_rate": 0.0001981824723893898,
      "loss": 4.9482,
      "step": 9750
    },
    {
      "epoch": 0.009097899245193629,
      "grad_norm": 1.0505528450012207,
      "learning_rate": 0.00019818060806729457,
      "loss": 4.9912,
      "step": 9760
    },
    {
      "epoch": 0.00910722086327272,
      "grad_norm": 1.1825658082962036,
      "learning_rate": 0.00019817874374519938,
      "loss": 4.8327,
      "step": 9770
    },
    {
      "epoch": 0.009116542481351811,
      "grad_norm": 1.306368112564087,
      "learning_rate": 0.00019817687942310418,
      "loss": 4.7829,
      "step": 9780
    },
    {
      "epoch": 0.009125864099430904,
      "grad_norm": 1.4120190143585205,
      "learning_rate": 0.00019817501510100896,
      "loss": 4.7333,
      "step": 9790
    },
    {
      "epoch": 0.009135185717509995,
      "grad_norm": 1.1577839851379395,
      "learning_rate": 0.0001981731507789138,
      "loss": 4.3941,
      "step": 9800
    },
    {
      "epoch": 0.009144507335589087,
      "grad_norm": 2.038696527481079,
      "learning_rate": 0.00019817128645681858,
      "loss": 4.8338,
      "step": 9810
    },
    {
      "epoch": 0.009153828953668178,
      "grad_norm": 1.2206677198410034,
      "learning_rate": 0.00019816942213472338,
      "loss": 4.9216,
      "step": 9820
    },
    {
      "epoch": 0.009163150571747271,
      "grad_norm": 1.2814263105392456,
      "learning_rate": 0.0001981675578126282,
      "loss": 4.7686,
      "step": 9830
    },
    {
      "epoch": 0.009172472189826362,
      "grad_norm": 1.074680209159851,
      "learning_rate": 0.00019816569349053297,
      "loss": 4.6594,
      "step": 9840
    },
    {
      "epoch": 0.009181793807905455,
      "grad_norm": 1.206278681755066,
      "learning_rate": 0.00019816382916843777,
      "loss": 4.7214,
      "step": 9850
    },
    {
      "epoch": 0.009191115425984546,
      "grad_norm": 1.036825180053711,
      "learning_rate": 0.00019816196484634258,
      "loss": 4.6276,
      "step": 9860
    },
    {
      "epoch": 0.009200437044063639,
      "grad_norm": 1.1669695377349854,
      "learning_rate": 0.0001981601005242474,
      "loss": 4.8764,
      "step": 9870
    },
    {
      "epoch": 0.00920975866214273,
      "grad_norm": 1.2604702711105347,
      "learning_rate": 0.0001981582362021522,
      "loss": 4.6848,
      "step": 9880
    },
    {
      "epoch": 0.009219080280221822,
      "grad_norm": 0.9949540495872498,
      "learning_rate": 0.00019815637188005697,
      "loss": 4.7039,
      "step": 9890
    },
    {
      "epoch": 0.009228401898300913,
      "grad_norm": 1.3873772621154785,
      "learning_rate": 0.00019815450755796178,
      "loss": 4.7845,
      "step": 9900
    },
    {
      "epoch": 0.009237723516380006,
      "grad_norm": 1.2861835956573486,
      "learning_rate": 0.00019815264323586659,
      "loss": 4.7158,
      "step": 9910
    },
    {
      "epoch": 0.009247045134459097,
      "grad_norm": 1.3478058576583862,
      "learning_rate": 0.0001981507789137714,
      "loss": 4.8737,
      "step": 9920
    },
    {
      "epoch": 0.00925636675253819,
      "grad_norm": 1.2582911252975464,
      "learning_rate": 0.0001981489145916762,
      "loss": 4.7967,
      "step": 9930
    },
    {
      "epoch": 0.009265688370617281,
      "grad_norm": 1.1238700151443481,
      "learning_rate": 0.00019814705026958098,
      "loss": 4.7222,
      "step": 9940
    },
    {
      "epoch": 0.009275009988696372,
      "grad_norm": 1.3084579706192017,
      "learning_rate": 0.00019814518594748578,
      "loss": 4.823,
      "step": 9950
    },
    {
      "epoch": 0.009284331606775465,
      "grad_norm": 1.1611272096633911,
      "learning_rate": 0.0001981433216253906,
      "loss": 4.6355,
      "step": 9960
    },
    {
      "epoch": 0.009293653224854556,
      "grad_norm": 1.0901215076446533,
      "learning_rate": 0.0001981414573032954,
      "loss": 4.8743,
      "step": 9970
    },
    {
      "epoch": 0.009302974842933649,
      "grad_norm": 1.2823495864868164,
      "learning_rate": 0.0001981395929812002,
      "loss": 4.8621,
      "step": 9980
    },
    {
      "epoch": 0.00931229646101274,
      "grad_norm": 1.0429694652557373,
      "learning_rate": 0.00019813772865910498,
      "loss": 4.7445,
      "step": 9990
    },
    {
      "epoch": 0.009321618079091832,
      "grad_norm": 1.1023356914520264,
      "learning_rate": 0.0001981358643370098,
      "loss": 4.9804,
      "step": 10000
    },
    {
      "epoch": 0.009330939697170923,
      "grad_norm": 1.060567855834961,
      "learning_rate": 0.00019813400001491457,
      "loss": 4.7998,
      "step": 10010
    },
    {
      "epoch": 0.009340261315250016,
      "grad_norm": 1.190134882926941,
      "learning_rate": 0.0001981321356928194,
      "loss": 4.8827,
      "step": 10020
    },
    {
      "epoch": 0.009349582933329107,
      "grad_norm": 1.3546993732452393,
      "learning_rate": 0.0001981302713707242,
      "loss": 4.8279,
      "step": 10030
    },
    {
      "epoch": 0.0093589045514082,
      "grad_norm": 1.2498699426651,
      "learning_rate": 0.00019812840704862899,
      "loss": 4.8121,
      "step": 10040
    },
    {
      "epoch": 0.00936822616948729,
      "grad_norm": 1.5190991163253784,
      "learning_rate": 0.0001981265427265338,
      "loss": 4.8656,
      "step": 10050
    },
    {
      "epoch": 0.009377547787566384,
      "grad_norm": 1.1750047206878662,
      "learning_rate": 0.00019812467840443857,
      "loss": 4.7142,
      "step": 10060
    },
    {
      "epoch": 0.009386869405645475,
      "grad_norm": 1.1871801614761353,
      "learning_rate": 0.00019812281408234338,
      "loss": 4.7094,
      "step": 10070
    },
    {
      "epoch": 0.009396191023724567,
      "grad_norm": 1.2673033475875854,
      "learning_rate": 0.0001981209497602482,
      "loss": 4.6508,
      "step": 10080
    },
    {
      "epoch": 0.009405512641803658,
      "grad_norm": 1.2788951396942139,
      "learning_rate": 0.000198119085438153,
      "loss": 4.737,
      "step": 10090
    },
    {
      "epoch": 0.009414834259882751,
      "grad_norm": 1.2345712184906006,
      "learning_rate": 0.0001981172211160578,
      "loss": 4.6077,
      "step": 10100
    },
    {
      "epoch": 0.009424155877961842,
      "grad_norm": 1.0977445840835571,
      "learning_rate": 0.00019811535679396258,
      "loss": 4.8428,
      "step": 10110
    },
    {
      "epoch": 0.009433477496040933,
      "grad_norm": 1.1960400342941284,
      "learning_rate": 0.00019811349247186738,
      "loss": 4.9134,
      "step": 10120
    },
    {
      "epoch": 0.009442799114120026,
      "grad_norm": 1.0095242261886597,
      "learning_rate": 0.0001981116281497722,
      "loss": 4.8306,
      "step": 10130
    },
    {
      "epoch": 0.009452120732199117,
      "grad_norm": 1.333713412284851,
      "learning_rate": 0.000198109763827677,
      "loss": 4.9704,
      "step": 10140
    },
    {
      "epoch": 0.00946144235027821,
      "grad_norm": 1.2155758142471313,
      "learning_rate": 0.0001981078995055818,
      "loss": 4.8291,
      "step": 10150
    },
    {
      "epoch": 0.0094707639683573,
      "grad_norm": 1.255080223083496,
      "learning_rate": 0.00019810603518348658,
      "loss": 4.6917,
      "step": 10160
    },
    {
      "epoch": 0.009480085586436393,
      "grad_norm": 1.211902379989624,
      "learning_rate": 0.0001981041708613914,
      "loss": 4.8739,
      "step": 10170
    },
    {
      "epoch": 0.009489407204515484,
      "grad_norm": 1.1717884540557861,
      "learning_rate": 0.0001981023065392962,
      "loss": 4.951,
      "step": 10180
    },
    {
      "epoch": 0.009498728822594577,
      "grad_norm": 1.2073568105697632,
      "learning_rate": 0.000198100442217201,
      "loss": 4.6756,
      "step": 10190
    },
    {
      "epoch": 0.009508050440673668,
      "grad_norm": 1.1263670921325684,
      "learning_rate": 0.0001980985778951058,
      "loss": 4.8524,
      "step": 10200
    },
    {
      "epoch": 0.009517372058752761,
      "grad_norm": 1.2557705640792847,
      "learning_rate": 0.00019809671357301058,
      "loss": 4.9316,
      "step": 10210
    },
    {
      "epoch": 0.009526693676831852,
      "grad_norm": 1.3979039192199707,
      "learning_rate": 0.0001980948492509154,
      "loss": 4.8196,
      "step": 10220
    },
    {
      "epoch": 0.009536015294910945,
      "grad_norm": 1.2264798879623413,
      "learning_rate": 0.0001980929849288202,
      "loss": 4.7791,
      "step": 10230
    },
    {
      "epoch": 0.009545336912990036,
      "grad_norm": 1.2309374809265137,
      "learning_rate": 0.00019809112060672498,
      "loss": 4.8637,
      "step": 10240
    },
    {
      "epoch": 0.009554658531069129,
      "grad_norm": 1.0925109386444092,
      "learning_rate": 0.0001980892562846298,
      "loss": 4.761,
      "step": 10250
    },
    {
      "epoch": 0.00956398014914822,
      "grad_norm": 1.2449666261672974,
      "learning_rate": 0.0001980873919625346,
      "loss": 4.6833,
      "step": 10260
    },
    {
      "epoch": 0.009573301767227312,
      "grad_norm": 1.2873764038085938,
      "learning_rate": 0.0001980855276404394,
      "loss": 4.8487,
      "step": 10270
    },
    {
      "epoch": 0.009582623385306403,
      "grad_norm": 1.2902933359146118,
      "learning_rate": 0.0001980836633183442,
      "loss": 4.8435,
      "step": 10280
    },
    {
      "epoch": 0.009591945003385494,
      "grad_norm": 1.3847185373306274,
      "learning_rate": 0.00019808179899624898,
      "loss": 4.739,
      "step": 10290
    },
    {
      "epoch": 0.009601266621464587,
      "grad_norm": 1.4183646440505981,
      "learning_rate": 0.0001980799346741538,
      "loss": 4.5476,
      "step": 10300
    },
    {
      "epoch": 0.009610588239543678,
      "grad_norm": 1.1881178617477417,
      "learning_rate": 0.0001980780703520586,
      "loss": 4.7622,
      "step": 10310
    },
    {
      "epoch": 0.00961990985762277,
      "grad_norm": 1.1843262910842896,
      "learning_rate": 0.0001980762060299634,
      "loss": 4.899,
      "step": 10320
    },
    {
      "epoch": 0.009629231475701862,
      "grad_norm": 1.304220199584961,
      "learning_rate": 0.0001980743417078682,
      "loss": 4.757,
      "step": 10330
    },
    {
      "epoch": 0.009638553093780955,
      "grad_norm": 1.0121113061904907,
      "learning_rate": 0.00019807247738577299,
      "loss": 4.7178,
      "step": 10340
    },
    {
      "epoch": 0.009647874711860046,
      "grad_norm": 1.3128981590270996,
      "learning_rate": 0.0001980706130636778,
      "loss": 4.7735,
      "step": 10350
    },
    {
      "epoch": 0.009657196329939138,
      "grad_norm": 1.169118881225586,
      "learning_rate": 0.0001980687487415826,
      "loss": 4.7495,
      "step": 10360
    },
    {
      "epoch": 0.00966651794801823,
      "grad_norm": 1.1405764818191528,
      "learning_rate": 0.0001980668844194874,
      "loss": 4.894,
      "step": 10370
    },
    {
      "epoch": 0.009675839566097322,
      "grad_norm": 1.1933735609054565,
      "learning_rate": 0.0001980650200973922,
      "loss": 4.7476,
      "step": 10380
    },
    {
      "epoch": 0.009685161184176413,
      "grad_norm": 1.126145362854004,
      "learning_rate": 0.000198063155775297,
      "loss": 4.7217,
      "step": 10390
    },
    {
      "epoch": 0.009694482802255506,
      "grad_norm": 1.1216226816177368,
      "learning_rate": 0.0001980612914532018,
      "loss": 4.8073,
      "step": 10400
    },
    {
      "epoch": 0.009703804420334597,
      "grad_norm": 1.0255376100540161,
      "learning_rate": 0.0001980594271311066,
      "loss": 4.6569,
      "step": 10410
    },
    {
      "epoch": 0.00971312603841369,
      "grad_norm": 1.3872339725494385,
      "learning_rate": 0.0001980575628090114,
      "loss": 4.7453,
      "step": 10420
    },
    {
      "epoch": 0.00972244765649278,
      "grad_norm": 1.1253314018249512,
      "learning_rate": 0.00019805569848691622,
      "loss": 4.8955,
      "step": 10430
    },
    {
      "epoch": 0.009731769274571873,
      "grad_norm": 1.184105396270752,
      "learning_rate": 0.000198053834164821,
      "loss": 4.6978,
      "step": 10440
    },
    {
      "epoch": 0.009741090892650964,
      "grad_norm": 1.1383328437805176,
      "learning_rate": 0.0001980519698427258,
      "loss": 4.8576,
      "step": 10450
    },
    {
      "epoch": 0.009750412510730057,
      "grad_norm": 1.4717637300491333,
      "learning_rate": 0.00019805010552063058,
      "loss": 4.7427,
      "step": 10460
    },
    {
      "epoch": 0.009759734128809148,
      "grad_norm": 1.2791986465454102,
      "learning_rate": 0.0001980482411985354,
      "loss": 4.9562,
      "step": 10470
    },
    {
      "epoch": 0.00976905574688824,
      "grad_norm": 1.346987247467041,
      "learning_rate": 0.0001980463768764402,
      "loss": 4.6409,
      "step": 10480
    },
    {
      "epoch": 0.009778377364967332,
      "grad_norm": 1.288801670074463,
      "learning_rate": 0.000198044512554345,
      "loss": 4.6323,
      "step": 10490
    },
    {
      "epoch": 0.009787698983046423,
      "grad_norm": 1.4014240503311157,
      "learning_rate": 0.0001980426482322498,
      "loss": 4.7536,
      "step": 10500
    },
    {
      "epoch": 0.009797020601125516,
      "grad_norm": 1.4243663549423218,
      "learning_rate": 0.00019804078391015458,
      "loss": 4.7558,
      "step": 10510
    },
    {
      "epoch": 0.009806342219204607,
      "grad_norm": 1.2514128684997559,
      "learning_rate": 0.0001980389195880594,
      "loss": 4.6668,
      "step": 10520
    },
    {
      "epoch": 0.0098156638372837,
      "grad_norm": 1.4737646579742432,
      "learning_rate": 0.0001980370552659642,
      "loss": 4.712,
      "step": 10530
    },
    {
      "epoch": 0.00982498545536279,
      "grad_norm": 1.1899890899658203,
      "learning_rate": 0.000198035190943869,
      "loss": 4.8831,
      "step": 10540
    },
    {
      "epoch": 0.009834307073441883,
      "grad_norm": 1.1416115760803223,
      "learning_rate": 0.0001980333266217738,
      "loss": 4.9354,
      "step": 10550
    },
    {
      "epoch": 0.009843628691520974,
      "grad_norm": 1.1339893341064453,
      "learning_rate": 0.0001980314622996786,
      "loss": 5.0064,
      "step": 10560
    },
    {
      "epoch": 0.009852950309600067,
      "grad_norm": 1.2619154453277588,
      "learning_rate": 0.0001980295979775834,
      "loss": 4.6575,
      "step": 10570
    },
    {
      "epoch": 0.009862271927679158,
      "grad_norm": 1.1024069786071777,
      "learning_rate": 0.0001980277336554882,
      "loss": 4.7709,
      "step": 10580
    },
    {
      "epoch": 0.00987159354575825,
      "grad_norm": 1.0575047731399536,
      "learning_rate": 0.000198025869333393,
      "loss": 4.7661,
      "step": 10590
    },
    {
      "epoch": 0.009880915163837342,
      "grad_norm": 1.3332699537277222,
      "learning_rate": 0.00019802400501129781,
      "loss": 4.6711,
      "step": 10600
    },
    {
      "epoch": 0.009890236781916435,
      "grad_norm": 1.230258584022522,
      "learning_rate": 0.0001980221406892026,
      "loss": 4.6709,
      "step": 10610
    },
    {
      "epoch": 0.009899558399995526,
      "grad_norm": 1.1430306434631348,
      "learning_rate": 0.0001980202763671074,
      "loss": 4.8151,
      "step": 10620
    },
    {
      "epoch": 0.009908880018074618,
      "grad_norm": 1.190203309059143,
      "learning_rate": 0.00019801841204501218,
      "loss": 4.903,
      "step": 10630
    },
    {
      "epoch": 0.00991820163615371,
      "grad_norm": 1.3638906478881836,
      "learning_rate": 0.000198016547722917,
      "loss": 4.5662,
      "step": 10640
    },
    {
      "epoch": 0.0099275232542328,
      "grad_norm": 1.338258147239685,
      "learning_rate": 0.00019801468340082182,
      "loss": 4.7076,
      "step": 10650
    },
    {
      "epoch": 0.009936844872311893,
      "grad_norm": 1.067834496498108,
      "learning_rate": 0.0001980128190787266,
      "loss": 4.7164,
      "step": 10660
    },
    {
      "epoch": 0.009946166490390984,
      "grad_norm": 1.3805263042449951,
      "learning_rate": 0.0001980109547566314,
      "loss": 4.7633,
      "step": 10670
    },
    {
      "epoch": 0.009955488108470077,
      "grad_norm": 0.9999874830245972,
      "learning_rate": 0.00019800909043453618,
      "loss": 4.4421,
      "step": 10680
    },
    {
      "epoch": 0.009964809726549168,
      "grad_norm": 1.2799524068832397,
      "learning_rate": 0.000198007226112441,
      "loss": 4.7734,
      "step": 10690
    },
    {
      "epoch": 0.00997413134462826,
      "grad_norm": 1.1686030626296997,
      "learning_rate": 0.00019800536179034582,
      "loss": 4.966,
      "step": 10700
    },
    {
      "epoch": 0.009983452962707352,
      "grad_norm": 1.4176279306411743,
      "learning_rate": 0.0001980034974682506,
      "loss": 4.7422,
      "step": 10710
    },
    {
      "epoch": 0.009992774580786444,
      "grad_norm": 1.188894510269165,
      "learning_rate": 0.0001980016331461554,
      "loss": 4.6776,
      "step": 10720
    },
    {
      "epoch": 0.010002096198865535,
      "grad_norm": 1.2971999645233154,
      "learning_rate": 0.0001979997688240602,
      "loss": 4.878,
      "step": 10730
    },
    {
      "epoch": 0.010011417816944628,
      "grad_norm": 1.280004858970642,
      "learning_rate": 0.000197997904501965,
      "loss": 4.7386,
      "step": 10740
    },
    {
      "epoch": 0.01002073943502372,
      "grad_norm": 1.168899416923523,
      "learning_rate": 0.0001979960401798698,
      "loss": 4.7453,
      "step": 10750
    },
    {
      "epoch": 0.010030061053102812,
      "grad_norm": 1.0558325052261353,
      "learning_rate": 0.0001979941758577746,
      "loss": 4.781,
      "step": 10760
    },
    {
      "epoch": 0.010039382671181903,
      "grad_norm": 1.1092973947525024,
      "learning_rate": 0.0001979923115356794,
      "loss": 4.9564,
      "step": 10770
    },
    {
      "epoch": 0.010048704289260996,
      "grad_norm": 1.1112545728683472,
      "learning_rate": 0.0001979904472135842,
      "loss": 4.7141,
      "step": 10780
    },
    {
      "epoch": 0.010058025907340087,
      "grad_norm": 1.1877262592315674,
      "learning_rate": 0.000197988582891489,
      "loss": 4.6751,
      "step": 10790
    },
    {
      "epoch": 0.01006734752541918,
      "grad_norm": 1.2648251056671143,
      "learning_rate": 0.0001979867185693938,
      "loss": 4.6229,
      "step": 10800
    },
    {
      "epoch": 0.01007666914349827,
      "grad_norm": 1.2631586790084839,
      "learning_rate": 0.0001979848542472986,
      "loss": 4.7838,
      "step": 10810
    },
    {
      "epoch": 0.010085990761577361,
      "grad_norm": 1.1618845462799072,
      "learning_rate": 0.00019798298992520342,
      "loss": 4.6281,
      "step": 10820
    },
    {
      "epoch": 0.010095312379656454,
      "grad_norm": 1.341293215751648,
      "learning_rate": 0.0001979811256031082,
      "loss": 4.7288,
      "step": 10830
    },
    {
      "epoch": 0.010104633997735545,
      "grad_norm": 1.2613236904144287,
      "learning_rate": 0.000197979261281013,
      "loss": 4.8606,
      "step": 10840
    },
    {
      "epoch": 0.010113955615814638,
      "grad_norm": 1.2914831638336182,
      "learning_rate": 0.0001979773969589178,
      "loss": 4.7075,
      "step": 10850
    },
    {
      "epoch": 0.010123277233893729,
      "grad_norm": 1.239212989807129,
      "learning_rate": 0.00019797553263682262,
      "loss": 4.8269,
      "step": 10860
    },
    {
      "epoch": 0.010132598851972822,
      "grad_norm": 1.0564316511154175,
      "learning_rate": 0.00019797366831472742,
      "loss": 4.6882,
      "step": 10870
    },
    {
      "epoch": 0.010141920470051913,
      "grad_norm": 1.2103338241577148,
      "learning_rate": 0.0001979718039926322,
      "loss": 4.6829,
      "step": 10880
    },
    {
      "epoch": 0.010151242088131006,
      "grad_norm": 1.284735083580017,
      "learning_rate": 0.000197969939670537,
      "loss": 4.7158,
      "step": 10890
    },
    {
      "epoch": 0.010160563706210097,
      "grad_norm": 1.2965316772460938,
      "learning_rate": 0.0001979680753484418,
      "loss": 4.6658,
      "step": 10900
    },
    {
      "epoch": 0.01016988532428919,
      "grad_norm": 1.3357555866241455,
      "learning_rate": 0.0001979662110263466,
      "loss": 4.6672,
      "step": 10910
    },
    {
      "epoch": 0.01017920694236828,
      "grad_norm": 1.2624608278274536,
      "learning_rate": 0.00019796434670425143,
      "loss": 4.6723,
      "step": 10920
    },
    {
      "epoch": 0.010188528560447373,
      "grad_norm": 1.1584806442260742,
      "learning_rate": 0.0001979624823821562,
      "loss": 4.5917,
      "step": 10930
    },
    {
      "epoch": 0.010197850178526464,
      "grad_norm": 1.5195177793502808,
      "learning_rate": 0.000197960618060061,
      "loss": 4.522,
      "step": 10940
    },
    {
      "epoch": 0.010207171796605557,
      "grad_norm": 1.18136465549469,
      "learning_rate": 0.00019795875373796582,
      "loss": 4.7162,
      "step": 10950
    },
    {
      "epoch": 0.010216493414684648,
      "grad_norm": 1.2841763496398926,
      "learning_rate": 0.0001979568894158706,
      "loss": 4.7751,
      "step": 10960
    },
    {
      "epoch": 0.01022581503276374,
      "grad_norm": 1.0874789953231812,
      "learning_rate": 0.0001979550250937754,
      "loss": 4.6248,
      "step": 10970
    },
    {
      "epoch": 0.010235136650842832,
      "grad_norm": 1.1882151365280151,
      "learning_rate": 0.0001979531607716802,
      "loss": 4.8104,
      "step": 10980
    },
    {
      "epoch": 0.010244458268921923,
      "grad_norm": 1.1277858018875122,
      "learning_rate": 0.00019795129644958502,
      "loss": 4.6884,
      "step": 10990
    },
    {
      "epoch": 0.010253779887001015,
      "grad_norm": 0.9989956617355347,
      "learning_rate": 0.00019794943212748982,
      "loss": 4.8309,
      "step": 11000
    },
    {
      "epoch": 0.010263101505080106,
      "grad_norm": 0.9787694215774536,
      "learning_rate": 0.0001979475678053946,
      "loss": 4.8387,
      "step": 11010
    },
    {
      "epoch": 0.010272423123159199,
      "grad_norm": 1.723764181137085,
      "learning_rate": 0.0001979457034832994,
      "loss": 4.8773,
      "step": 11020
    },
    {
      "epoch": 0.01028174474123829,
      "grad_norm": 1.2659903764724731,
      "learning_rate": 0.00019794383916120421,
      "loss": 4.5747,
      "step": 11030
    },
    {
      "epoch": 0.010291066359317383,
      "grad_norm": 1.2373461723327637,
      "learning_rate": 0.00019794197483910902,
      "loss": 4.7724,
      "step": 11040
    },
    {
      "epoch": 0.010300387977396474,
      "grad_norm": 1.3656929731369019,
      "learning_rate": 0.00019794011051701383,
      "loss": 4.8741,
      "step": 11050
    },
    {
      "epoch": 0.010309709595475567,
      "grad_norm": 1.1666160821914673,
      "learning_rate": 0.0001979382461949186,
      "loss": 4.9831,
      "step": 11060
    },
    {
      "epoch": 0.010319031213554658,
      "grad_norm": 1.126936912536621,
      "learning_rate": 0.0001979363818728234,
      "loss": 4.5886,
      "step": 11070
    },
    {
      "epoch": 0.01032835283163375,
      "grad_norm": 1.1741900444030762,
      "learning_rate": 0.0001979345175507282,
      "loss": 4.6319,
      "step": 11080
    },
    {
      "epoch": 0.010337674449712841,
      "grad_norm": 1.3327516317367554,
      "learning_rate": 0.00019793265322863302,
      "loss": 4.7026,
      "step": 11090
    },
    {
      "epoch": 0.010346996067791934,
      "grad_norm": 1.308957815170288,
      "learning_rate": 0.00019793078890653783,
      "loss": 4.7954,
      "step": 11100
    },
    {
      "epoch": 0.010356317685871025,
      "grad_norm": 1.3498717546463013,
      "learning_rate": 0.0001979289245844426,
      "loss": 4.62,
      "step": 11110
    },
    {
      "epoch": 0.010365639303950118,
      "grad_norm": 1.2134662866592407,
      "learning_rate": 0.00019792706026234742,
      "loss": 4.6173,
      "step": 11120
    },
    {
      "epoch": 0.010374960922029209,
      "grad_norm": 1.193324327468872,
      "learning_rate": 0.0001979251959402522,
      "loss": 4.4733,
      "step": 11130
    },
    {
      "epoch": 0.010384282540108302,
      "grad_norm": 1.1285287141799927,
      "learning_rate": 0.000197923331618157,
      "loss": 4.6607,
      "step": 11140
    },
    {
      "epoch": 0.010393604158187393,
      "grad_norm": 1.1136326789855957,
      "learning_rate": 0.00019792146729606184,
      "loss": 4.7972,
      "step": 11150
    },
    {
      "epoch": 0.010402925776266484,
      "grad_norm": 1.2666780948638916,
      "learning_rate": 0.00019791960297396661,
      "loss": 4.8056,
      "step": 11160
    },
    {
      "epoch": 0.010412247394345577,
      "grad_norm": 1.2646830081939697,
      "learning_rate": 0.00019791773865187142,
      "loss": 4.7425,
      "step": 11170
    },
    {
      "epoch": 0.010421569012424668,
      "grad_norm": 1.2879892587661743,
      "learning_rate": 0.0001979158743297762,
      "loss": 4.932,
      "step": 11180
    },
    {
      "epoch": 0.01043089063050376,
      "grad_norm": 1.121221899986267,
      "learning_rate": 0.000197914010007681,
      "loss": 4.6843,
      "step": 11190
    },
    {
      "epoch": 0.010440212248582851,
      "grad_norm": 1.2783775329589844,
      "learning_rate": 0.0001979121456855858,
      "loss": 4.7301,
      "step": 11200
    },
    {
      "epoch": 0.010449533866661944,
      "grad_norm": 1.1659321784973145,
      "learning_rate": 0.00019791028136349062,
      "loss": 4.9241,
      "step": 11210
    },
    {
      "epoch": 0.010458855484741035,
      "grad_norm": 1.3595731258392334,
      "learning_rate": 0.00019790841704139543,
      "loss": 4.5638,
      "step": 11220
    },
    {
      "epoch": 0.010468177102820128,
      "grad_norm": 1.0807355642318726,
      "learning_rate": 0.0001979065527193002,
      "loss": 4.8447,
      "step": 11230
    },
    {
      "epoch": 0.010477498720899219,
      "grad_norm": 1.2740159034729004,
      "learning_rate": 0.000197904688397205,
      "loss": 4.6457,
      "step": 11240
    },
    {
      "epoch": 0.010486820338978312,
      "grad_norm": 1.182685375213623,
      "learning_rate": 0.00019790282407510982,
      "loss": 4.7466,
      "step": 11250
    },
    {
      "epoch": 0.010496141957057403,
      "grad_norm": 1.1896840333938599,
      "learning_rate": 0.00019790095975301462,
      "loss": 4.4765,
      "step": 11260
    },
    {
      "epoch": 0.010505463575136495,
      "grad_norm": 1.0717246532440186,
      "learning_rate": 0.00019789909543091943,
      "loss": 4.8576,
      "step": 11270
    },
    {
      "epoch": 0.010514785193215586,
      "grad_norm": 1.2992846965789795,
      "learning_rate": 0.0001978972311088242,
      "loss": 4.5128,
      "step": 11280
    },
    {
      "epoch": 0.010524106811294679,
      "grad_norm": 1.1548304557800293,
      "learning_rate": 0.00019789536678672902,
      "loss": 4.9362,
      "step": 11290
    },
    {
      "epoch": 0.01053342842937377,
      "grad_norm": 1.2316871881484985,
      "learning_rate": 0.00019789350246463382,
      "loss": 4.5554,
      "step": 11300
    },
    {
      "epoch": 0.010542750047452863,
      "grad_norm": 1.210260272026062,
      "learning_rate": 0.00019789163814253863,
      "loss": 4.6275,
      "step": 11310
    },
    {
      "epoch": 0.010552071665531954,
      "grad_norm": 1.114179015159607,
      "learning_rate": 0.00019788977382044343,
      "loss": 4.8509,
      "step": 11320
    },
    {
      "epoch": 0.010561393283611045,
      "grad_norm": 1.2047675848007202,
      "learning_rate": 0.00019788790949834821,
      "loss": 4.6989,
      "step": 11330
    },
    {
      "epoch": 0.010570714901690138,
      "grad_norm": 1.198034644126892,
      "learning_rate": 0.00019788604517625302,
      "loss": 4.6403,
      "step": 11340
    },
    {
      "epoch": 0.010580036519769229,
      "grad_norm": 1.2856758832931519,
      "learning_rate": 0.00019788418085415783,
      "loss": 4.7054,
      "step": 11350
    },
    {
      "epoch": 0.010589358137848321,
      "grad_norm": 1.413301944732666,
      "learning_rate": 0.0001978823165320626,
      "loss": 4.6383,
      "step": 11360
    },
    {
      "epoch": 0.010598679755927412,
      "grad_norm": 1.2697609663009644,
      "learning_rate": 0.00019788045220996744,
      "loss": 4.6142,
      "step": 11370
    },
    {
      "epoch": 0.010608001374006505,
      "grad_norm": 1.0524674654006958,
      "learning_rate": 0.00019787858788787222,
      "loss": 4.5567,
      "step": 11380
    },
    {
      "epoch": 0.010617322992085596,
      "grad_norm": 1.1526802778244019,
      "learning_rate": 0.00019787672356577702,
      "loss": 4.873,
      "step": 11390
    },
    {
      "epoch": 0.010626644610164689,
      "grad_norm": 1.3432260751724243,
      "learning_rate": 0.00019787485924368183,
      "loss": 4.5404,
      "step": 11400
    },
    {
      "epoch": 0.01063596622824378,
      "grad_norm": 1.3010451793670654,
      "learning_rate": 0.0001978729949215866,
      "loss": 4.482,
      "step": 11410
    },
    {
      "epoch": 0.010645287846322873,
      "grad_norm": 1.3385483026504517,
      "learning_rate": 0.00019787113059949142,
      "loss": 4.7251,
      "step": 11420
    },
    {
      "epoch": 0.010654609464401964,
      "grad_norm": 1.1978262662887573,
      "learning_rate": 0.00019786926627739622,
      "loss": 4.7541,
      "step": 11430
    },
    {
      "epoch": 0.010663931082481056,
      "grad_norm": 0.9666787385940552,
      "learning_rate": 0.00019786740195530103,
      "loss": 4.834,
      "step": 11440
    },
    {
      "epoch": 0.010673252700560147,
      "grad_norm": 1.158795714378357,
      "learning_rate": 0.00019786553763320584,
      "loss": 4.5876,
      "step": 11450
    },
    {
      "epoch": 0.01068257431863924,
      "grad_norm": 1.16313898563385,
      "learning_rate": 0.00019786367331111061,
      "loss": 4.678,
      "step": 11460
    },
    {
      "epoch": 0.010691895936718331,
      "grad_norm": 1.2621750831604004,
      "learning_rate": 0.00019786180898901542,
      "loss": 4.6522,
      "step": 11470
    },
    {
      "epoch": 0.010701217554797424,
      "grad_norm": 1.2329789400100708,
      "learning_rate": 0.00019785994466692023,
      "loss": 4.7996,
      "step": 11480
    },
    {
      "epoch": 0.010710539172876515,
      "grad_norm": 1.1732146739959717,
      "learning_rate": 0.00019785808034482503,
      "loss": 4.7619,
      "step": 11490
    },
    {
      "epoch": 0.010719860790955606,
      "grad_norm": 1.3189551830291748,
      "learning_rate": 0.00019785621602272984,
      "loss": 4.838,
      "step": 11500
    },
    {
      "epoch": 0.010729182409034699,
      "grad_norm": 1.574763298034668,
      "learning_rate": 0.00019785435170063462,
      "loss": 4.7824,
      "step": 11510
    },
    {
      "epoch": 0.01073850402711379,
      "grad_norm": 1.1082957983016968,
      "learning_rate": 0.00019785248737853943,
      "loss": 4.6017,
      "step": 11520
    },
    {
      "epoch": 0.010747825645192883,
      "grad_norm": 1.1895201206207275,
      "learning_rate": 0.0001978506230564442,
      "loss": 4.5556,
      "step": 11530
    },
    {
      "epoch": 0.010757147263271974,
      "grad_norm": 1.4591130018234253,
      "learning_rate": 0.00019784875873434904,
      "loss": 4.8635,
      "step": 11540
    },
    {
      "epoch": 0.010766468881351066,
      "grad_norm": 1.4023135900497437,
      "learning_rate": 0.00019784689441225384,
      "loss": 4.6071,
      "step": 11550
    },
    {
      "epoch": 0.010775790499430157,
      "grad_norm": 1.1090431213378906,
      "learning_rate": 0.00019784503009015862,
      "loss": 4.727,
      "step": 11560
    },
    {
      "epoch": 0.01078511211750925,
      "grad_norm": 1.1972178220748901,
      "learning_rate": 0.00019784316576806343,
      "loss": 4.7418,
      "step": 11570
    },
    {
      "epoch": 0.010794433735588341,
      "grad_norm": 1.2352659702301025,
      "learning_rate": 0.0001978413014459682,
      "loss": 4.7407,
      "step": 11580
    },
    {
      "epoch": 0.010803755353667434,
      "grad_norm": 1.2603613138198853,
      "learning_rate": 0.00019783943712387302,
      "loss": 4.6822,
      "step": 11590
    },
    {
      "epoch": 0.010813076971746525,
      "grad_norm": 1.327211856842041,
      "learning_rate": 0.00019783757280177785,
      "loss": 4.619,
      "step": 11600
    },
    {
      "epoch": 0.010822398589825618,
      "grad_norm": 1.4776089191436768,
      "learning_rate": 0.00019783570847968263,
      "loss": 4.5795,
      "step": 11610
    },
    {
      "epoch": 0.010831720207904709,
      "grad_norm": 0.9901697635650635,
      "learning_rate": 0.00019783384415758743,
      "loss": 4.6705,
      "step": 11620
    },
    {
      "epoch": 0.010841041825983801,
      "grad_norm": 1.232736349105835,
      "learning_rate": 0.0001978319798354922,
      "loss": 4.7825,
      "step": 11630
    },
    {
      "epoch": 0.010850363444062892,
      "grad_norm": 1.072238564491272,
      "learning_rate": 0.00019783011551339702,
      "loss": 4.6176,
      "step": 11640
    },
    {
      "epoch": 0.010859685062141985,
      "grad_norm": 1.1184239387512207,
      "learning_rate": 0.00019782825119130183,
      "loss": 4.5254,
      "step": 11650
    },
    {
      "epoch": 0.010869006680221076,
      "grad_norm": 1.267174482345581,
      "learning_rate": 0.00019782638686920663,
      "loss": 4.6721,
      "step": 11660
    },
    {
      "epoch": 0.010878328298300167,
      "grad_norm": 1.2039575576782227,
      "learning_rate": 0.00019782452254711144,
      "loss": 4.5298,
      "step": 11670
    },
    {
      "epoch": 0.01088764991637926,
      "grad_norm": 1.1937792301177979,
      "learning_rate": 0.00019782265822501622,
      "loss": 4.4862,
      "step": 11680
    },
    {
      "epoch": 0.010896971534458351,
      "grad_norm": 0.988756000995636,
      "learning_rate": 0.00019782079390292102,
      "loss": 4.9108,
      "step": 11690
    },
    {
      "epoch": 0.010906293152537444,
      "grad_norm": 1.1089746952056885,
      "learning_rate": 0.00019781892958082583,
      "loss": 4.7939,
      "step": 11700
    },
    {
      "epoch": 0.010915614770616535,
      "grad_norm": 1.2528226375579834,
      "learning_rate": 0.00019781706525873064,
      "loss": 4.7491,
      "step": 11710
    },
    {
      "epoch": 0.010924936388695627,
      "grad_norm": 1.414536476135254,
      "learning_rate": 0.00019781520093663544,
      "loss": 4.5876,
      "step": 11720
    },
    {
      "epoch": 0.010934258006774718,
      "grad_norm": 1.1733521223068237,
      "learning_rate": 0.00019781333661454022,
      "loss": 4.5764,
      "step": 11730
    },
    {
      "epoch": 0.010943579624853811,
      "grad_norm": 1.3656671047210693,
      "learning_rate": 0.00019781147229244503,
      "loss": 4.6952,
      "step": 11740
    },
    {
      "epoch": 0.010952901242932902,
      "grad_norm": 1.2032086849212646,
      "learning_rate": 0.00019780960797034983,
      "loss": 4.6902,
      "step": 11750
    },
    {
      "epoch": 0.010962222861011995,
      "grad_norm": 1.0818759202957153,
      "learning_rate": 0.00019780774364825464,
      "loss": 4.4154,
      "step": 11760
    },
    {
      "epoch": 0.010971544479091086,
      "grad_norm": 1.1229424476623535,
      "learning_rate": 0.00019780587932615945,
      "loss": 4.8044,
      "step": 11770
    },
    {
      "epoch": 0.010980866097170179,
      "grad_norm": 1.167639970779419,
      "learning_rate": 0.00019780401500406423,
      "loss": 4.5981,
      "step": 11780
    },
    {
      "epoch": 0.01099018771524927,
      "grad_norm": 1.2121373414993286,
      "learning_rate": 0.00019780215068196903,
      "loss": 4.8779,
      "step": 11790
    },
    {
      "epoch": 0.010999509333328363,
      "grad_norm": 1.0996671915054321,
      "learning_rate": 0.00019780028635987384,
      "loss": 4.7066,
      "step": 11800
    },
    {
      "epoch": 0.011008830951407454,
      "grad_norm": 1.170280933380127,
      "learning_rate": 0.00019779842203777862,
      "loss": 4.7977,
      "step": 11810
    },
    {
      "epoch": 0.011018152569486546,
      "grad_norm": 1.0869178771972656,
      "learning_rate": 0.00019779655771568345,
      "loss": 4.796,
      "step": 11820
    },
    {
      "epoch": 0.011027474187565637,
      "grad_norm": 1.0648571252822876,
      "learning_rate": 0.00019779469339358823,
      "loss": 4.5319,
      "step": 11830
    },
    {
      "epoch": 0.011036795805644728,
      "grad_norm": 1.1287767887115479,
      "learning_rate": 0.00019779282907149304,
      "loss": 4.6534,
      "step": 11840
    },
    {
      "epoch": 0.011046117423723821,
      "grad_norm": 1.2282471656799316,
      "learning_rate": 0.00019779096474939782,
      "loss": 4.7662,
      "step": 11850
    },
    {
      "epoch": 0.011055439041802912,
      "grad_norm": 1.1648250818252563,
      "learning_rate": 0.00019778910042730262,
      "loss": 4.6396,
      "step": 11860
    },
    {
      "epoch": 0.011064760659882005,
      "grad_norm": 1.2931251525878906,
      "learning_rate": 0.00019778723610520743,
      "loss": 4.6483,
      "step": 11870
    },
    {
      "epoch": 0.011074082277961096,
      "grad_norm": 1.3543866872787476,
      "learning_rate": 0.00019778537178311224,
      "loss": 4.6379,
      "step": 11880
    },
    {
      "epoch": 0.011083403896040189,
      "grad_norm": 1.289854884147644,
      "learning_rate": 0.00019778350746101704,
      "loss": 4.7323,
      "step": 11890
    },
    {
      "epoch": 0.01109272551411928,
      "grad_norm": 1.3021650314331055,
      "learning_rate": 0.00019778164313892182,
      "loss": 4.5233,
      "step": 11900
    },
    {
      "epoch": 0.011102047132198372,
      "grad_norm": 1.3946518898010254,
      "learning_rate": 0.00019777977881682663,
      "loss": 4.5092,
      "step": 11910
    },
    {
      "epoch": 0.011111368750277463,
      "grad_norm": 0.9844072461128235,
      "learning_rate": 0.00019777791449473143,
      "loss": 4.6891,
      "step": 11920
    },
    {
      "epoch": 0.011120690368356556,
      "grad_norm": 1.1855660676956177,
      "learning_rate": 0.00019777605017263624,
      "loss": 4.7639,
      "step": 11930
    },
    {
      "epoch": 0.011130011986435647,
      "grad_norm": 1.3074935674667358,
      "learning_rate": 0.00019777418585054105,
      "loss": 4.6153,
      "step": 11940
    },
    {
      "epoch": 0.01113933360451474,
      "grad_norm": 1.3982152938842773,
      "learning_rate": 0.00019777232152844583,
      "loss": 4.4365,
      "step": 11950
    },
    {
      "epoch": 0.011148655222593831,
      "grad_norm": 1.2927395105361938,
      "learning_rate": 0.00019777045720635063,
      "loss": 4.6714,
      "step": 11960
    },
    {
      "epoch": 0.011157976840672924,
      "grad_norm": 1.100450873374939,
      "learning_rate": 0.00019776859288425544,
      "loss": 4.5975,
      "step": 11970
    },
    {
      "epoch": 0.011167298458752015,
      "grad_norm": 1.2631586790084839,
      "learning_rate": 0.00019776672856216022,
      "loss": 4.4876,
      "step": 11980
    },
    {
      "epoch": 0.011176620076831107,
      "grad_norm": 1.275938868522644,
      "learning_rate": 0.00019776486424006505,
      "loss": 4.7713,
      "step": 11990
    },
    {
      "epoch": 0.011185941694910198,
      "grad_norm": 1.2341827154159546,
      "learning_rate": 0.00019776299991796983,
      "loss": 4.7848,
      "step": 12000
    },
    {
      "epoch": 0.01119526331298929,
      "grad_norm": 1.140900731086731,
      "learning_rate": 0.00019776113559587464,
      "loss": 4.8047,
      "step": 12010
    },
    {
      "epoch": 0.011204584931068382,
      "grad_norm": 1.0845210552215576,
      "learning_rate": 0.00019775927127377944,
      "loss": 4.7725,
      "step": 12020
    },
    {
      "epoch": 0.011213906549147473,
      "grad_norm": 1.0679272413253784,
      "learning_rate": 0.00019775740695168422,
      "loss": 4.6435,
      "step": 12030
    },
    {
      "epoch": 0.011223228167226566,
      "grad_norm": 1.1146315336227417,
      "learning_rate": 0.00019775554262958903,
      "loss": 4.6914,
      "step": 12040
    },
    {
      "epoch": 0.011232549785305657,
      "grad_norm": 1.2842248678207397,
      "learning_rate": 0.00019775367830749383,
      "loss": 4.62,
      "step": 12050
    },
    {
      "epoch": 0.01124187140338475,
      "grad_norm": 1.2058252096176147,
      "learning_rate": 0.00019775181398539864,
      "loss": 4.7046,
      "step": 12060
    },
    {
      "epoch": 0.01125119302146384,
      "grad_norm": 1.4024115800857544,
      "learning_rate": 0.00019774994966330345,
      "loss": 4.6583,
      "step": 12070
    },
    {
      "epoch": 0.011260514639542933,
      "grad_norm": 1.279287576675415,
      "learning_rate": 0.00019774808534120823,
      "loss": 4.6478,
      "step": 12080
    },
    {
      "epoch": 0.011269836257622024,
      "grad_norm": 0.8680769801139832,
      "learning_rate": 0.00019774622101911303,
      "loss": 4.7963,
      "step": 12090
    },
    {
      "epoch": 0.011279157875701117,
      "grad_norm": 1.3426874876022339,
      "learning_rate": 0.00019774435669701784,
      "loss": 4.7591,
      "step": 12100
    },
    {
      "epoch": 0.011288479493780208,
      "grad_norm": 1.219810128211975,
      "learning_rate": 0.00019774249237492264,
      "loss": 4.7049,
      "step": 12110
    },
    {
      "epoch": 0.011297801111859301,
      "grad_norm": 1.283286213874817,
      "learning_rate": 0.00019774062805282745,
      "loss": 4.6354,
      "step": 12120
    },
    {
      "epoch": 0.011307122729938392,
      "grad_norm": 1.0953909158706665,
      "learning_rate": 0.00019773876373073223,
      "loss": 4.7229,
      "step": 12130
    },
    {
      "epoch": 0.011316444348017485,
      "grad_norm": 1.2258269786834717,
      "learning_rate": 0.00019773689940863704,
      "loss": 4.7086,
      "step": 12140
    },
    {
      "epoch": 0.011325765966096576,
      "grad_norm": 1.1907451152801514,
      "learning_rate": 0.00019773503508654184,
      "loss": 4.6051,
      "step": 12150
    },
    {
      "epoch": 0.011335087584175669,
      "grad_norm": 0.9313935041427612,
      "learning_rate": 0.00019773317076444665,
      "loss": 4.7153,
      "step": 12160
    },
    {
      "epoch": 0.01134440920225476,
      "grad_norm": 1.1758939027786255,
      "learning_rate": 0.00019773130644235146,
      "loss": 4.4233,
      "step": 12170
    },
    {
      "epoch": 0.01135373082033385,
      "grad_norm": 1.0681064128875732,
      "learning_rate": 0.00019772944212025623,
      "loss": 4.5251,
      "step": 12180
    },
    {
      "epoch": 0.011363052438412943,
      "grad_norm": 1.1233164072036743,
      "learning_rate": 0.00019772757779816104,
      "loss": 4.7083,
      "step": 12190
    },
    {
      "epoch": 0.011372374056492034,
      "grad_norm": 1.2560549974441528,
      "learning_rate": 0.00019772571347606582,
      "loss": 4.7889,
      "step": 12200
    },
    {
      "epoch": 0.011381695674571127,
      "grad_norm": 1.0657035112380981,
      "learning_rate": 0.00019772384915397065,
      "loss": 4.6704,
      "step": 12210
    },
    {
      "epoch": 0.011391017292650218,
      "grad_norm": 1.1567423343658447,
      "learning_rate": 0.00019772198483187546,
      "loss": 4.5549,
      "step": 12220
    },
    {
      "epoch": 0.01140033891072931,
      "grad_norm": 1.3664268255233765,
      "learning_rate": 0.00019772012050978024,
      "loss": 4.5797,
      "step": 12230
    },
    {
      "epoch": 0.011409660528808402,
      "grad_norm": 1.2904472351074219,
      "learning_rate": 0.00019771825618768505,
      "loss": 4.6315,
      "step": 12240
    },
    {
      "epoch": 0.011418982146887495,
      "grad_norm": 1.3208314180374146,
      "learning_rate": 0.00019771639186558982,
      "loss": 4.7327,
      "step": 12250
    },
    {
      "epoch": 0.011428303764966586,
      "grad_norm": 1.3051873445510864,
      "learning_rate": 0.00019771452754349463,
      "loss": 4.711,
      "step": 12260
    },
    {
      "epoch": 0.011437625383045678,
      "grad_norm": 1.270342230796814,
      "learning_rate": 0.00019771266322139946,
      "loss": 4.658,
      "step": 12270
    },
    {
      "epoch": 0.01144694700112477,
      "grad_norm": 1.0336834192276,
      "learning_rate": 0.00019771079889930424,
      "loss": 4.5904,
      "step": 12280
    },
    {
      "epoch": 0.011456268619203862,
      "grad_norm": 1.2927606105804443,
      "learning_rate": 0.00019770893457720905,
      "loss": 4.5531,
      "step": 12290
    },
    {
      "epoch": 0.011465590237282953,
      "grad_norm": 1.310385823249817,
      "learning_rate": 0.00019770707025511383,
      "loss": 4.4912,
      "step": 12300
    },
    {
      "epoch": 0.011474911855362046,
      "grad_norm": 1.2712312936782837,
      "learning_rate": 0.00019770520593301864,
      "loss": 4.8065,
      "step": 12310
    },
    {
      "epoch": 0.011484233473441137,
      "grad_norm": 1.3402910232543945,
      "learning_rate": 0.00019770334161092344,
      "loss": 4.7332,
      "step": 12320
    },
    {
      "epoch": 0.01149355509152023,
      "grad_norm": 1.0866848230361938,
      "learning_rate": 0.00019770147728882825,
      "loss": 4.5417,
      "step": 12330
    },
    {
      "epoch": 0.01150287670959932,
      "grad_norm": 1.2206584215164185,
      "learning_rate": 0.00019769961296673305,
      "loss": 4.6794,
      "step": 12340
    },
    {
      "epoch": 0.011512198327678412,
      "grad_norm": 1.0400333404541016,
      "learning_rate": 0.00019769774864463783,
      "loss": 4.7074,
      "step": 12350
    },
    {
      "epoch": 0.011521519945757504,
      "grad_norm": 0.979443371295929,
      "learning_rate": 0.00019769588432254264,
      "loss": 4.8348,
      "step": 12360
    },
    {
      "epoch": 0.011530841563836595,
      "grad_norm": 1.3747246265411377,
      "learning_rate": 0.00019769402000044745,
      "loss": 4.66,
      "step": 12370
    },
    {
      "epoch": 0.011540163181915688,
      "grad_norm": 1.2172242403030396,
      "learning_rate": 0.00019769215567835225,
      "loss": 4.5097,
      "step": 12380
    },
    {
      "epoch": 0.01154948479999478,
      "grad_norm": 1.3853471279144287,
      "learning_rate": 0.00019769029135625706,
      "loss": 4.5363,
      "step": 12390
    },
    {
      "epoch": 0.011558806418073872,
      "grad_norm": 1.0263781547546387,
      "learning_rate": 0.00019768842703416184,
      "loss": 4.7845,
      "step": 12400
    },
    {
      "epoch": 0.011568128036152963,
      "grad_norm": 1.3445008993148804,
      "learning_rate": 0.00019768656271206664,
      "loss": 4.7493,
      "step": 12410
    },
    {
      "epoch": 0.011577449654232056,
      "grad_norm": 1.2730623483657837,
      "learning_rate": 0.00019768469838997145,
      "loss": 4.5258,
      "step": 12420
    },
    {
      "epoch": 0.011586771272311147,
      "grad_norm": 1.2987854480743408,
      "learning_rate": 0.00019768283406787626,
      "loss": 4.4234,
      "step": 12430
    },
    {
      "epoch": 0.01159609289039024,
      "grad_norm": 1.105466604232788,
      "learning_rate": 0.00019768096974578106,
      "loss": 4.9975,
      "step": 12440
    },
    {
      "epoch": 0.01160541450846933,
      "grad_norm": 1.1115144491195679,
      "learning_rate": 0.00019767910542368584,
      "loss": 4.6709,
      "step": 12450
    },
    {
      "epoch": 0.011614736126548423,
      "grad_norm": 1.2577496767044067,
      "learning_rate": 0.00019767724110159065,
      "loss": 4.7156,
      "step": 12460
    },
    {
      "epoch": 0.011624057744627514,
      "grad_norm": 1.196192741394043,
      "learning_rate": 0.00019767537677949546,
      "loss": 4.6191,
      "step": 12470
    },
    {
      "epoch": 0.011633379362706607,
      "grad_norm": 1.29702627658844,
      "learning_rate": 0.00019767351245740023,
      "loss": 4.6969,
      "step": 12480
    },
    {
      "epoch": 0.011642700980785698,
      "grad_norm": 1.3111809492111206,
      "learning_rate": 0.00019767164813530504,
      "loss": 4.5261,
      "step": 12490
    },
    {
      "epoch": 0.01165202259886479,
      "grad_norm": 1.1275421380996704,
      "learning_rate": 0.00019766978381320985,
      "loss": 4.7837,
      "step": 12500
    },
    {
      "epoch": 0.011661344216943882,
      "grad_norm": 1.2708603143692017,
      "learning_rate": 0.00019766791949111465,
      "loss": 4.3096,
      "step": 12510
    },
    {
      "epoch": 0.011670665835022973,
      "grad_norm": 1.2825387716293335,
      "learning_rate": 0.00019766605516901946,
      "loss": 4.8033,
      "step": 12520
    },
    {
      "epoch": 0.011679987453102066,
      "grad_norm": 1.3484225273132324,
      "learning_rate": 0.00019766419084692424,
      "loss": 4.7548,
      "step": 12530
    },
    {
      "epoch": 0.011689309071181157,
      "grad_norm": 1.0836859941482544,
      "learning_rate": 0.00019766232652482905,
      "loss": 4.916,
      "step": 12540
    },
    {
      "epoch": 0.01169863068926025,
      "grad_norm": 1.3245078325271606,
      "learning_rate": 0.00019766046220273385,
      "loss": 4.8847,
      "step": 12550
    },
    {
      "epoch": 0.01170795230733934,
      "grad_norm": 1.1478313207626343,
      "learning_rate": 0.00019765859788063866,
      "loss": 4.6515,
      "step": 12560
    },
    {
      "epoch": 0.011717273925418433,
      "grad_norm": 1.1909085512161255,
      "learning_rate": 0.00019765673355854346,
      "loss": 4.6161,
      "step": 12570
    },
    {
      "epoch": 0.011726595543497524,
      "grad_norm": 1.0271369218826294,
      "learning_rate": 0.00019765486923644824,
      "loss": 4.5848,
      "step": 12580
    },
    {
      "epoch": 0.011735917161576617,
      "grad_norm": 1.3631749153137207,
      "learning_rate": 0.00019765300491435305,
      "loss": 4.7815,
      "step": 12590
    },
    {
      "epoch": 0.011745238779655708,
      "grad_norm": 1.2389789819717407,
      "learning_rate": 0.00019765114059225786,
      "loss": 4.7508,
      "step": 12600
    },
    {
      "epoch": 0.0117545603977348,
      "grad_norm": 1.4146127700805664,
      "learning_rate": 0.00019764927627016266,
      "loss": 4.4433,
      "step": 12610
    },
    {
      "epoch": 0.011763882015813892,
      "grad_norm": 1.4589766263961792,
      "learning_rate": 0.00019764741194806747,
      "loss": 4.7614,
      "step": 12620
    },
    {
      "epoch": 0.011773203633892984,
      "grad_norm": 1.0718472003936768,
      "learning_rate": 0.00019764554762597225,
      "loss": 4.7041,
      "step": 12630
    },
    {
      "epoch": 0.011782525251972075,
      "grad_norm": 1.2327929735183716,
      "learning_rate": 0.00019764368330387705,
      "loss": 4.7072,
      "step": 12640
    },
    {
      "epoch": 0.011791846870051168,
      "grad_norm": 1.220878005027771,
      "learning_rate": 0.00019764181898178183,
      "loss": 4.512,
      "step": 12650
    },
    {
      "epoch": 0.01180116848813026,
      "grad_norm": 1.0879954099655151,
      "learning_rate": 0.00019763995465968667,
      "loss": 4.8538,
      "step": 12660
    },
    {
      "epoch": 0.011810490106209352,
      "grad_norm": 1.050572395324707,
      "learning_rate": 0.00019763809033759147,
      "loss": 4.8608,
      "step": 12670
    },
    {
      "epoch": 0.011819811724288443,
      "grad_norm": 1.2213302850723267,
      "learning_rate": 0.00019763622601549625,
      "loss": 4.5209,
      "step": 12680
    },
    {
      "epoch": 0.011829133342367536,
      "grad_norm": 1.375158667564392,
      "learning_rate": 0.00019763436169340106,
      "loss": 4.5286,
      "step": 12690
    },
    {
      "epoch": 0.011838454960446627,
      "grad_norm": 2.2986717224121094,
      "learning_rate": 0.00019763249737130584,
      "loss": 4.7763,
      "step": 12700
    },
    {
      "epoch": 0.011847776578525718,
      "grad_norm": 1.1853551864624023,
      "learning_rate": 0.00019763063304921064,
      "loss": 4.5357,
      "step": 12710
    },
    {
      "epoch": 0.01185709819660481,
      "grad_norm": 1.0633550882339478,
      "learning_rate": 0.00019762876872711548,
      "loss": 4.7219,
      "step": 12720
    },
    {
      "epoch": 0.011866419814683902,
      "grad_norm": 1.0482428073883057,
      "learning_rate": 0.00019762690440502026,
      "loss": 4.9219,
      "step": 12730
    },
    {
      "epoch": 0.011875741432762994,
      "grad_norm": 0.9563916325569153,
      "learning_rate": 0.00019762504008292506,
      "loss": 4.703,
      "step": 12740
    },
    {
      "epoch": 0.011885063050842085,
      "grad_norm": 1.1312015056610107,
      "learning_rate": 0.00019762317576082984,
      "loss": 4.5091,
      "step": 12750
    },
    {
      "epoch": 0.011894384668921178,
      "grad_norm": 1.2012325525283813,
      "learning_rate": 0.00019762131143873465,
      "loss": 4.625,
      "step": 12760
    },
    {
      "epoch": 0.011903706287000269,
      "grad_norm": 1.3088470697402954,
      "learning_rate": 0.00019761944711663945,
      "loss": 4.3811,
      "step": 12770
    },
    {
      "epoch": 0.011913027905079362,
      "grad_norm": 1.080206036567688,
      "learning_rate": 0.00019761758279454426,
      "loss": 4.6929,
      "step": 12780
    },
    {
      "epoch": 0.011922349523158453,
      "grad_norm": 1.1996368169784546,
      "learning_rate": 0.00019761571847244907,
      "loss": 4.6732,
      "step": 12790
    },
    {
      "epoch": 0.011931671141237546,
      "grad_norm": 1.39106285572052,
      "learning_rate": 0.00019761385415035385,
      "loss": 4.7267,
      "step": 12800
    },
    {
      "epoch": 0.011940992759316637,
      "grad_norm": 1.1282126903533936,
      "learning_rate": 0.00019761198982825865,
      "loss": 4.7142,
      "step": 12810
    },
    {
      "epoch": 0.01195031437739573,
      "grad_norm": 1.2700788974761963,
      "learning_rate": 0.00019761012550616346,
      "loss": 4.7675,
      "step": 12820
    },
    {
      "epoch": 0.01195963599547482,
      "grad_norm": 1.348223090171814,
      "learning_rate": 0.00019760826118406827,
      "loss": 4.7524,
      "step": 12830
    },
    {
      "epoch": 0.011968957613553913,
      "grad_norm": 1.0604645013809204,
      "learning_rate": 0.00019760639686197307,
      "loss": 4.4601,
      "step": 12840
    },
    {
      "epoch": 0.011978279231633004,
      "grad_norm": 1.1485093832015991,
      "learning_rate": 0.00019760453253987785,
      "loss": 4.6461,
      "step": 12850
    },
    {
      "epoch": 0.011987600849712097,
      "grad_norm": 1.1520196199417114,
      "learning_rate": 0.00019760266821778266,
      "loss": 4.5846,
      "step": 12860
    },
    {
      "epoch": 0.011996922467791188,
      "grad_norm": 1.2037489414215088,
      "learning_rate": 0.00019760080389568746,
      "loss": 4.6565,
      "step": 12870
    },
    {
      "epoch": 0.012006244085870279,
      "grad_norm": 1.065216064453125,
      "learning_rate": 0.00019759893957359227,
      "loss": 4.8474,
      "step": 12880
    },
    {
      "epoch": 0.012015565703949372,
      "grad_norm": 1.1594268083572388,
      "learning_rate": 0.00019759707525149708,
      "loss": 4.4472,
      "step": 12890
    },
    {
      "epoch": 0.012024887322028463,
      "grad_norm": 1.2396490573883057,
      "learning_rate": 0.00019759521092940186,
      "loss": 4.8284,
      "step": 12900
    },
    {
      "epoch": 0.012034208940107555,
      "grad_norm": 1.2758737802505493,
      "learning_rate": 0.00019759334660730666,
      "loss": 4.4162,
      "step": 12910
    },
    {
      "epoch": 0.012043530558186646,
      "grad_norm": 1.071630597114563,
      "learning_rate": 0.00019759148228521147,
      "loss": 4.6895,
      "step": 12920
    },
    {
      "epoch": 0.01205285217626574,
      "grad_norm": 1.1360751390457153,
      "learning_rate": 0.00019758961796311625,
      "loss": 4.7709,
      "step": 12930
    },
    {
      "epoch": 0.01206217379434483,
      "grad_norm": 1.7119441032409668,
      "learning_rate": 0.00019758775364102108,
      "loss": 4.482,
      "step": 12940
    },
    {
      "epoch": 0.012071495412423923,
      "grad_norm": 1.1814135313034058,
      "learning_rate": 0.00019758588931892586,
      "loss": 4.5808,
      "step": 12950
    },
    {
      "epoch": 0.012080817030503014,
      "grad_norm": 1.039850115776062,
      "learning_rate": 0.00019758402499683067,
      "loss": 4.5094,
      "step": 12960
    },
    {
      "epoch": 0.012090138648582107,
      "grad_norm": 1.4325493574142456,
      "learning_rate": 0.00019758216067473547,
      "loss": 4.576,
      "step": 12970
    },
    {
      "epoch": 0.012099460266661198,
      "grad_norm": 1.117756724357605,
      "learning_rate": 0.00019758029635264025,
      "loss": 4.6241,
      "step": 12980
    },
    {
      "epoch": 0.01210878188474029,
      "grad_norm": 1.1039752960205078,
      "learning_rate": 0.00019757843203054506,
      "loss": 4.6387,
      "step": 12990
    },
    {
      "epoch": 0.012118103502819381,
      "grad_norm": 1.3030712604522705,
      "learning_rate": 0.00019757656770844986,
      "loss": 4.687,
      "step": 13000
    },
    {
      "epoch": 0.012127425120898474,
      "grad_norm": 1.2151240110397339,
      "learning_rate": 0.00019757470338635467,
      "loss": 4.6118,
      "step": 13010
    },
    {
      "epoch": 0.012136746738977565,
      "grad_norm": 1.4980878829956055,
      "learning_rate": 0.00019757283906425948,
      "loss": 4.6359,
      "step": 13020
    },
    {
      "epoch": 0.012146068357056658,
      "grad_norm": 1.2986094951629639,
      "learning_rate": 0.00019757097474216426,
      "loss": 4.6052,
      "step": 13030
    },
    {
      "epoch": 0.012155389975135749,
      "grad_norm": 1.3786877393722534,
      "learning_rate": 0.00019756911042006906,
      "loss": 4.8498,
      "step": 13040
    },
    {
      "epoch": 0.01216471159321484,
      "grad_norm": 1.0313318967819214,
      "learning_rate": 0.00019756724609797387,
      "loss": 4.5698,
      "step": 13050
    },
    {
      "epoch": 0.012174033211293933,
      "grad_norm": 1.2049813270568848,
      "learning_rate": 0.00019756538177587868,
      "loss": 4.7521,
      "step": 13060
    },
    {
      "epoch": 0.012183354829373024,
      "grad_norm": 1.1151974201202393,
      "learning_rate": 0.00019756351745378348,
      "loss": 4.6395,
      "step": 13070
    },
    {
      "epoch": 0.012192676447452117,
      "grad_norm": 0.953339695930481,
      "learning_rate": 0.00019756165313168826,
      "loss": 4.5785,
      "step": 13080
    },
    {
      "epoch": 0.012201998065531208,
      "grad_norm": 1.311403751373291,
      "learning_rate": 0.00019755978880959307,
      "loss": 4.6028,
      "step": 13090
    },
    {
      "epoch": 0.0122113196836103,
      "grad_norm": 1.163691520690918,
      "learning_rate": 0.00019755792448749785,
      "loss": 4.6506,
      "step": 13100
    },
    {
      "epoch": 0.012220641301689391,
      "grad_norm": 1.1840766668319702,
      "learning_rate": 0.00019755606016540268,
      "loss": 4.5586,
      "step": 13110
    },
    {
      "epoch": 0.012229962919768484,
      "grad_norm": 1.150666356086731,
      "learning_rate": 0.00019755419584330749,
      "loss": 4.6729,
      "step": 13120
    },
    {
      "epoch": 0.012239284537847575,
      "grad_norm": 1.003915548324585,
      "learning_rate": 0.00019755233152121227,
      "loss": 4.8124,
      "step": 13130
    },
    {
      "epoch": 0.012248606155926668,
      "grad_norm": 1.1809171438217163,
      "learning_rate": 0.00019755046719911707,
      "loss": 4.5933,
      "step": 13140
    },
    {
      "epoch": 0.012257927774005759,
      "grad_norm": 1.260977029800415,
      "learning_rate": 0.00019754860287702185,
      "loss": 4.8976,
      "step": 13150
    },
    {
      "epoch": 0.012267249392084852,
      "grad_norm": 1.1831741333007812,
      "learning_rate": 0.00019754673855492666,
      "loss": 4.5577,
      "step": 13160
    },
    {
      "epoch": 0.012276571010163943,
      "grad_norm": 1.3199199438095093,
      "learning_rate": 0.0001975448742328315,
      "loss": 4.6118,
      "step": 13170
    },
    {
      "epoch": 0.012285892628243035,
      "grad_norm": 1.306962490081787,
      "learning_rate": 0.00019754300991073627,
      "loss": 4.5308,
      "step": 13180
    },
    {
      "epoch": 0.012295214246322126,
      "grad_norm": 1.2305041551589966,
      "learning_rate": 0.00019754114558864108,
      "loss": 4.5861,
      "step": 13190
    },
    {
      "epoch": 0.012304535864401219,
      "grad_norm": 1.2881935834884644,
      "learning_rate": 0.00019753928126654585,
      "loss": 4.6977,
      "step": 13200
    },
    {
      "epoch": 0.01231385748248031,
      "grad_norm": 1.1907974481582642,
      "learning_rate": 0.00019753741694445066,
      "loss": 4.5098,
      "step": 13210
    },
    {
      "epoch": 0.012323179100559401,
      "grad_norm": 1.181117057800293,
      "learning_rate": 0.00019753555262235547,
      "loss": 4.6534,
      "step": 13220
    },
    {
      "epoch": 0.012332500718638494,
      "grad_norm": 1.2014846801757812,
      "learning_rate": 0.00019753368830026027,
      "loss": 4.3763,
      "step": 13230
    },
    {
      "epoch": 0.012341822336717585,
      "grad_norm": 1.2450850009918213,
      "learning_rate": 0.00019753182397816508,
      "loss": 4.4741,
      "step": 13240
    },
    {
      "epoch": 0.012351143954796678,
      "grad_norm": 1.211602807044983,
      "learning_rate": 0.00019752995965606986,
      "loss": 4.5399,
      "step": 13250
    },
    {
      "epoch": 0.012360465572875769,
      "grad_norm": 1.187275767326355,
      "learning_rate": 0.00019752809533397467,
      "loss": 4.6768,
      "step": 13260
    },
    {
      "epoch": 0.012369787190954861,
      "grad_norm": 1.3823065757751465,
      "learning_rate": 0.00019752623101187947,
      "loss": 4.62,
      "step": 13270
    },
    {
      "epoch": 0.012379108809033952,
      "grad_norm": 1.2806780338287354,
      "learning_rate": 0.00019752436668978428,
      "loss": 4.7038,
      "step": 13280
    },
    {
      "epoch": 0.012388430427113045,
      "grad_norm": 1.1268961429595947,
      "learning_rate": 0.00019752250236768908,
      "loss": 4.4554,
      "step": 13290
    },
    {
      "epoch": 0.012397752045192136,
      "grad_norm": 1.2477169036865234,
      "learning_rate": 0.00019752063804559386,
      "loss": 4.6478,
      "step": 13300
    },
    {
      "epoch": 0.012407073663271229,
      "grad_norm": 1.3264281749725342,
      "learning_rate": 0.00019751877372349867,
      "loss": 4.8063,
      "step": 13310
    },
    {
      "epoch": 0.01241639528135032,
      "grad_norm": 1.3030250072479248,
      "learning_rate": 0.00019751690940140345,
      "loss": 4.4581,
      "step": 13320
    },
    {
      "epoch": 0.012425716899429413,
      "grad_norm": 1.2535392045974731,
      "learning_rate": 0.00019751504507930828,
      "loss": 4.5586,
      "step": 13330
    },
    {
      "epoch": 0.012435038517508504,
      "grad_norm": 0.9326042532920837,
      "learning_rate": 0.0001975131807572131,
      "loss": 4.6503,
      "step": 13340
    },
    {
      "epoch": 0.012444360135587597,
      "grad_norm": 1.332778811454773,
      "learning_rate": 0.00019751131643511787,
      "loss": 4.6307,
      "step": 13350
    },
    {
      "epoch": 0.012453681753666688,
      "grad_norm": 1.6392605304718018,
      "learning_rate": 0.00019750945211302267,
      "loss": 4.5735,
      "step": 13360
    },
    {
      "epoch": 0.01246300337174578,
      "grad_norm": 1.137817621231079,
      "learning_rate": 0.00019750758779092745,
      "loss": 4.5374,
      "step": 13370
    },
    {
      "epoch": 0.012472324989824871,
      "grad_norm": 0.9170074462890625,
      "learning_rate": 0.00019750572346883226,
      "loss": 4.5609,
      "step": 13380
    },
    {
      "epoch": 0.012481646607903962,
      "grad_norm": 1.3231275081634521,
      "learning_rate": 0.0001975038591467371,
      "loss": 4.5682,
      "step": 13390
    },
    {
      "epoch": 0.012490968225983055,
      "grad_norm": 1.0771247148513794,
      "learning_rate": 0.00019750199482464187,
      "loss": 4.7615,
      "step": 13400
    },
    {
      "epoch": 0.012500289844062146,
      "grad_norm": 1.196934700012207,
      "learning_rate": 0.00019750013050254668,
      "loss": 4.6208,
      "step": 13410
    },
    {
      "epoch": 0.012509611462141239,
      "grad_norm": 1.1433851718902588,
      "learning_rate": 0.00019749826618045146,
      "loss": 4.7994,
      "step": 13420
    },
    {
      "epoch": 0.01251893308022033,
      "grad_norm": 1.0329235792160034,
      "learning_rate": 0.00019749640185835626,
      "loss": 4.6803,
      "step": 13430
    },
    {
      "epoch": 0.012528254698299423,
      "grad_norm": 1.129536747932434,
      "learning_rate": 0.00019749453753626107,
      "loss": 4.594,
      "step": 13440
    },
    {
      "epoch": 0.012537576316378514,
      "grad_norm": 1.3494727611541748,
      "learning_rate": 0.00019749267321416588,
      "loss": 4.6216,
      "step": 13450
    },
    {
      "epoch": 0.012546897934457606,
      "grad_norm": 1.2231273651123047,
      "learning_rate": 0.00019749080889207068,
      "loss": 4.4608,
      "step": 13460
    },
    {
      "epoch": 0.012556219552536697,
      "grad_norm": 1.203768014907837,
      "learning_rate": 0.00019748894456997546,
      "loss": 4.6914,
      "step": 13470
    },
    {
      "epoch": 0.01256554117061579,
      "grad_norm": 1.5585161447525024,
      "learning_rate": 0.00019748708024788027,
      "loss": 4.6073,
      "step": 13480
    },
    {
      "epoch": 0.012574862788694881,
      "grad_norm": 1.0325742959976196,
      "learning_rate": 0.00019748521592578508,
      "loss": 4.6363,
      "step": 13490
    },
    {
      "epoch": 0.012584184406773974,
      "grad_norm": 1.1110585927963257,
      "learning_rate": 0.00019748335160368988,
      "loss": 4.7737,
      "step": 13500
    },
    {
      "epoch": 0.012593506024853065,
      "grad_norm": 1.2551761865615845,
      "learning_rate": 0.0001974814872815947,
      "loss": 4.4785,
      "step": 13510
    },
    {
      "epoch": 0.012602827642932158,
      "grad_norm": 1.2974624633789062,
      "learning_rate": 0.00019747962295949947,
      "loss": 4.5697,
      "step": 13520
    },
    {
      "epoch": 0.012612149261011249,
      "grad_norm": 1.4728407859802246,
      "learning_rate": 0.00019747775863740427,
      "loss": 4.6409,
      "step": 13530
    },
    {
      "epoch": 0.012621470879090341,
      "grad_norm": 1.3445494174957275,
      "learning_rate": 0.00019747589431530908,
      "loss": 4.5136,
      "step": 13540
    },
    {
      "epoch": 0.012630792497169432,
      "grad_norm": 1.171932339668274,
      "learning_rate": 0.00019747402999321386,
      "loss": 4.4562,
      "step": 13550
    },
    {
      "epoch": 0.012640114115248523,
      "grad_norm": 1.1676685810089111,
      "learning_rate": 0.0001974721656711187,
      "loss": 4.617,
      "step": 13560
    },
    {
      "epoch": 0.012649435733327616,
      "grad_norm": 0.8529885411262512,
      "learning_rate": 0.00019747030134902347,
      "loss": 4.8001,
      "step": 13570
    },
    {
      "epoch": 0.012658757351406707,
      "grad_norm": 1.0291918516159058,
      "learning_rate": 0.00019746843702692828,
      "loss": 4.7313,
      "step": 13580
    },
    {
      "epoch": 0.0126680789694858,
      "grad_norm": 1.139506459236145,
      "learning_rate": 0.00019746657270483308,
      "loss": 4.8003,
      "step": 13590
    },
    {
      "epoch": 0.012677400587564891,
      "grad_norm": 1.5068858861923218,
      "learning_rate": 0.00019746470838273786,
      "loss": 4.5084,
      "step": 13600
    },
    {
      "epoch": 0.012686722205643984,
      "grad_norm": 1.2331618070602417,
      "learning_rate": 0.00019746284406064267,
      "loss": 4.4436,
      "step": 13610
    },
    {
      "epoch": 0.012696043823723075,
      "grad_norm": 1.1810226440429688,
      "learning_rate": 0.00019746097973854748,
      "loss": 4.6442,
      "step": 13620
    },
    {
      "epoch": 0.012705365441802167,
      "grad_norm": 1.1582752466201782,
      "learning_rate": 0.00019745911541645228,
      "loss": 4.3951,
      "step": 13630
    },
    {
      "epoch": 0.012714687059881258,
      "grad_norm": 1.4209319353103638,
      "learning_rate": 0.0001974572510943571,
      "loss": 4.4466,
      "step": 13640
    },
    {
      "epoch": 0.012724008677960351,
      "grad_norm": 1.2929952144622803,
      "learning_rate": 0.00019745538677226187,
      "loss": 4.4052,
      "step": 13650
    },
    {
      "epoch": 0.012733330296039442,
      "grad_norm": 1.1203868389129639,
      "learning_rate": 0.00019745352245016667,
      "loss": 4.5592,
      "step": 13660
    },
    {
      "epoch": 0.012742651914118535,
      "grad_norm": 1.2191412448883057,
      "learning_rate": 0.00019745165812807148,
      "loss": 4.628,
      "step": 13670
    },
    {
      "epoch": 0.012751973532197626,
      "grad_norm": 1.4064608812332153,
      "learning_rate": 0.0001974497938059763,
      "loss": 4.6458,
      "step": 13680
    },
    {
      "epoch": 0.012761295150276719,
      "grad_norm": 1.0982595682144165,
      "learning_rate": 0.0001974479294838811,
      "loss": 4.6321,
      "step": 13690
    },
    {
      "epoch": 0.01277061676835581,
      "grad_norm": 1.2136070728302002,
      "learning_rate": 0.00019744606516178587,
      "loss": 4.5303,
      "step": 13700
    },
    {
      "epoch": 0.012779938386434903,
      "grad_norm": 1.3154706954956055,
      "learning_rate": 0.00019744420083969068,
      "loss": 4.5429,
      "step": 13710
    },
    {
      "epoch": 0.012789260004513994,
      "grad_norm": 0.9224554896354675,
      "learning_rate": 0.00019744233651759548,
      "loss": 4.6728,
      "step": 13720
    },
    {
      "epoch": 0.012798581622593085,
      "grad_norm": 0.9948967695236206,
      "learning_rate": 0.0001974404721955003,
      "loss": 4.5109,
      "step": 13730
    },
    {
      "epoch": 0.012807903240672177,
      "grad_norm": 1.1020865440368652,
      "learning_rate": 0.0001974386078734051,
      "loss": 4.5421,
      "step": 13740
    },
    {
      "epoch": 0.012817224858751268,
      "grad_norm": 1.318479061126709,
      "learning_rate": 0.00019743674355130988,
      "loss": 4.6748,
      "step": 13750
    },
    {
      "epoch": 0.012826546476830361,
      "grad_norm": 0.9723525047302246,
      "learning_rate": 0.00019743487922921468,
      "loss": 4.6678,
      "step": 13760
    },
    {
      "epoch": 0.012835868094909452,
      "grad_norm": 1.2034777402877808,
      "learning_rate": 0.00019743301490711946,
      "loss": 4.7253,
      "step": 13770
    },
    {
      "epoch": 0.012845189712988545,
      "grad_norm": 1.073094129562378,
      "learning_rate": 0.0001974311505850243,
      "loss": 4.5419,
      "step": 13780
    },
    {
      "epoch": 0.012854511331067636,
      "grad_norm": 1.0819003582000732,
      "learning_rate": 0.0001974292862629291,
      "loss": 4.547,
      "step": 13790
    },
    {
      "epoch": 0.012863832949146729,
      "grad_norm": 1.3740450143814087,
      "learning_rate": 0.00019742742194083388,
      "loss": 4.5193,
      "step": 13800
    },
    {
      "epoch": 0.01287315456722582,
      "grad_norm": 1.1645454168319702,
      "learning_rate": 0.0001974255576187387,
      "loss": 4.7186,
      "step": 13810
    },
    {
      "epoch": 0.012882476185304912,
      "grad_norm": 1.1539785861968994,
      "learning_rate": 0.00019742369329664347,
      "loss": 4.5478,
      "step": 13820
    },
    {
      "epoch": 0.012891797803384003,
      "grad_norm": 1.1883512735366821,
      "learning_rate": 0.00019742182897454827,
      "loss": 4.4143,
      "step": 13830
    },
    {
      "epoch": 0.012901119421463096,
      "grad_norm": 1.0664678812026978,
      "learning_rate": 0.0001974199646524531,
      "loss": 4.538,
      "step": 13840
    },
    {
      "epoch": 0.012910441039542187,
      "grad_norm": 1.3002418279647827,
      "learning_rate": 0.00019741810033035789,
      "loss": 4.4193,
      "step": 13850
    },
    {
      "epoch": 0.01291976265762128,
      "grad_norm": 1.2597792148590088,
      "learning_rate": 0.0001974162360082627,
      "loss": 4.6764,
      "step": 13860
    },
    {
      "epoch": 0.012929084275700371,
      "grad_norm": 1.1419131755828857,
      "learning_rate": 0.00019741437168616747,
      "loss": 4.7406,
      "step": 13870
    },
    {
      "epoch": 0.012938405893779464,
      "grad_norm": 1.156623363494873,
      "learning_rate": 0.00019741250736407228,
      "loss": 4.6338,
      "step": 13880
    },
    {
      "epoch": 0.012947727511858555,
      "grad_norm": 1.2742986679077148,
      "learning_rate": 0.00019741064304197708,
      "loss": 4.6494,
      "step": 13890
    },
    {
      "epoch": 0.012957049129937646,
      "grad_norm": 1.0218092203140259,
      "learning_rate": 0.0001974087787198819,
      "loss": 4.6675,
      "step": 13900
    },
    {
      "epoch": 0.012966370748016738,
      "grad_norm": 1.1143782138824463,
      "learning_rate": 0.0001974069143977867,
      "loss": 4.641,
      "step": 13910
    },
    {
      "epoch": 0.01297569236609583,
      "grad_norm": 1.0599349737167358,
      "learning_rate": 0.00019740505007569148,
      "loss": 4.5823,
      "step": 13920
    },
    {
      "epoch": 0.012985013984174922,
      "grad_norm": 1.0372275114059448,
      "learning_rate": 0.00019740318575359628,
      "loss": 4.5744,
      "step": 13930
    },
    {
      "epoch": 0.012994335602254013,
      "grad_norm": 1.2615920305252075,
      "learning_rate": 0.0001974013214315011,
      "loss": 4.5551,
      "step": 13940
    },
    {
      "epoch": 0.013003657220333106,
      "grad_norm": 1.2706372737884521,
      "learning_rate": 0.0001973994571094059,
      "loss": 4.69,
      "step": 13950
    },
    {
      "epoch": 0.013012978838412197,
      "grad_norm": 1.2567298412322998,
      "learning_rate": 0.0001973975927873107,
      "loss": 4.6023,
      "step": 13960
    },
    {
      "epoch": 0.01302230045649129,
      "grad_norm": 1.2674345970153809,
      "learning_rate": 0.00019739572846521548,
      "loss": 4.4242,
      "step": 13970
    },
    {
      "epoch": 0.01303162207457038,
      "grad_norm": 1.1064292192459106,
      "learning_rate": 0.00019739386414312029,
      "loss": 4.6011,
      "step": 13980
    },
    {
      "epoch": 0.013040943692649474,
      "grad_norm": 1.3388310670852661,
      "learning_rate": 0.0001973919998210251,
      "loss": 4.6125,
      "step": 13990
    },
    {
      "epoch": 0.013050265310728565,
      "grad_norm": 1.2205479145050049,
      "learning_rate": 0.00019739013549892987,
      "loss": 4.7592,
      "step": 14000
    },
    {
      "epoch": 0.013059586928807657,
      "grad_norm": 1.2712163925170898,
      "learning_rate": 0.0001973882711768347,
      "loss": 4.66,
      "step": 14010
    },
    {
      "epoch": 0.013068908546886748,
      "grad_norm": 1.2503936290740967,
      "learning_rate": 0.00019738640685473948,
      "loss": 4.6927,
      "step": 14020
    },
    {
      "epoch": 0.013078230164965841,
      "grad_norm": 1.3008058071136475,
      "learning_rate": 0.0001973845425326443,
      "loss": 4.4921,
      "step": 14030
    },
    {
      "epoch": 0.013087551783044932,
      "grad_norm": 1.2134071588516235,
      "learning_rate": 0.0001973826782105491,
      "loss": 4.831,
      "step": 14040
    },
    {
      "epoch": 0.013096873401124025,
      "grad_norm": 1.3433655500411987,
      "learning_rate": 0.00019738081388845388,
      "loss": 4.6331,
      "step": 14050
    },
    {
      "epoch": 0.013106195019203116,
      "grad_norm": 1.254221796989441,
      "learning_rate": 0.00019737894956635868,
      "loss": 4.6866,
      "step": 14060
    },
    {
      "epoch": 0.013115516637282207,
      "grad_norm": 1.2082555294036865,
      "learning_rate": 0.0001973770852442635,
      "loss": 4.6223,
      "step": 14070
    },
    {
      "epoch": 0.0131248382553613,
      "grad_norm": 1.3346956968307495,
      "learning_rate": 0.0001973752209221683,
      "loss": 4.5076,
      "step": 14080
    },
    {
      "epoch": 0.01313415987344039,
      "grad_norm": 1.0328843593597412,
      "learning_rate": 0.0001973733566000731,
      "loss": 4.578,
      "step": 14090
    },
    {
      "epoch": 0.013143481491519483,
      "grad_norm": 1.0859013795852661,
      "learning_rate": 0.00019737149227797788,
      "loss": 4.6691,
      "step": 14100
    },
    {
      "epoch": 0.013152803109598574,
      "grad_norm": 1.084708571434021,
      "learning_rate": 0.0001973696279558827,
      "loss": 4.6189,
      "step": 14110
    },
    {
      "epoch": 0.013162124727677667,
      "grad_norm": 1.171392560005188,
      "learning_rate": 0.0001973677636337875,
      "loss": 4.5217,
      "step": 14120
    },
    {
      "epoch": 0.013171446345756758,
      "grad_norm": 1.304352879524231,
      "learning_rate": 0.0001973658993116923,
      "loss": 4.6421,
      "step": 14130
    },
    {
      "epoch": 0.013180767963835851,
      "grad_norm": 1.225507140159607,
      "learning_rate": 0.0001973640349895971,
      "loss": 4.4624,
      "step": 14140
    },
    {
      "epoch": 0.013190089581914942,
      "grad_norm": 1.3479249477386475,
      "learning_rate": 0.00019736217066750189,
      "loss": 4.6503,
      "step": 14150
    },
    {
      "epoch": 0.013199411199994035,
      "grad_norm": 1.0172749757766724,
      "learning_rate": 0.0001973603063454067,
      "loss": 4.577,
      "step": 14160
    },
    {
      "epoch": 0.013208732818073126,
      "grad_norm": 1.0406603813171387,
      "learning_rate": 0.0001973584420233115,
      "loss": 4.5942,
      "step": 14170
    },
    {
      "epoch": 0.013218054436152218,
      "grad_norm": 1.1724786758422852,
      "learning_rate": 0.0001973565777012163,
      "loss": 4.6601,
      "step": 14180
    },
    {
      "epoch": 0.01322737605423131,
      "grad_norm": 1.3267409801483154,
      "learning_rate": 0.0001973547133791211,
      "loss": 4.4116,
      "step": 14190
    },
    {
      "epoch": 0.013236697672310402,
      "grad_norm": 1.1584765911102295,
      "learning_rate": 0.0001973528490570259,
      "loss": 4.3818,
      "step": 14200
    },
    {
      "epoch": 0.013246019290389493,
      "grad_norm": 1.1328881978988647,
      "learning_rate": 0.0001973509847349307,
      "loss": 4.5451,
      "step": 14210
    },
    {
      "epoch": 0.013255340908468586,
      "grad_norm": 1.032038688659668,
      "learning_rate": 0.00019734912041283548,
      "loss": 4.5265,
      "step": 14220
    },
    {
      "epoch": 0.013264662526547677,
      "grad_norm": 1.175805687904358,
      "learning_rate": 0.0001973472560907403,
      "loss": 4.6743,
      "step": 14230
    },
    {
      "epoch": 0.013273984144626768,
      "grad_norm": 1.118894100189209,
      "learning_rate": 0.00019734539176864511,
      "loss": 4.459,
      "step": 14240
    },
    {
      "epoch": 0.01328330576270586,
      "grad_norm": 1.176609992980957,
      "learning_rate": 0.0001973435274465499,
      "loss": 4.5926,
      "step": 14250
    },
    {
      "epoch": 0.013292627380784952,
      "grad_norm": 1.3336635828018188,
      "learning_rate": 0.0001973416631244547,
      "loss": 4.4959,
      "step": 14260
    },
    {
      "epoch": 0.013301948998864044,
      "grad_norm": 1.0857014656066895,
      "learning_rate": 0.00019733979880235948,
      "loss": 4.6487,
      "step": 14270
    },
    {
      "epoch": 0.013311270616943136,
      "grad_norm": 1.1726363897323608,
      "learning_rate": 0.00019733793448026429,
      "loss": 4.5337,
      "step": 14280
    },
    {
      "epoch": 0.013320592235022228,
      "grad_norm": 1.2783362865447998,
      "learning_rate": 0.00019733607015816912,
      "loss": 4.5319,
      "step": 14290
    },
    {
      "epoch": 0.01332991385310132,
      "grad_norm": 1.3290194272994995,
      "learning_rate": 0.0001973342058360739,
      "loss": 4.404,
      "step": 14300
    },
    {
      "epoch": 0.013339235471180412,
      "grad_norm": 1.0712889432907104,
      "learning_rate": 0.0001973323415139787,
      "loss": 4.5497,
      "step": 14310
    },
    {
      "epoch": 0.013348557089259503,
      "grad_norm": 1.162014365196228,
      "learning_rate": 0.00019733047719188348,
      "loss": 4.5493,
      "step": 14320
    },
    {
      "epoch": 0.013357878707338596,
      "grad_norm": 1.4546514749526978,
      "learning_rate": 0.0001973286128697883,
      "loss": 4.4657,
      "step": 14330
    },
    {
      "epoch": 0.013367200325417687,
      "grad_norm": 1.2480608224868774,
      "learning_rate": 0.0001973267485476931,
      "loss": 4.5999,
      "step": 14340
    },
    {
      "epoch": 0.01337652194349678,
      "grad_norm": 1.231766700744629,
      "learning_rate": 0.0001973248842255979,
      "loss": 4.62,
      "step": 14350
    },
    {
      "epoch": 0.01338584356157587,
      "grad_norm": 1.0564762353897095,
      "learning_rate": 0.0001973230199035027,
      "loss": 4.5744,
      "step": 14360
    },
    {
      "epoch": 0.013395165179654963,
      "grad_norm": 1.2368824481964111,
      "learning_rate": 0.0001973211555814075,
      "loss": 4.4338,
      "step": 14370
    },
    {
      "epoch": 0.013404486797734054,
      "grad_norm": 1.1743470430374146,
      "learning_rate": 0.0001973192912593123,
      "loss": 4.5022,
      "step": 14380
    },
    {
      "epoch": 0.013413808415813147,
      "grad_norm": 1.4512903690338135,
      "learning_rate": 0.0001973174269372171,
      "loss": 4.6209,
      "step": 14390
    },
    {
      "epoch": 0.013423130033892238,
      "grad_norm": 1.2742364406585693,
      "learning_rate": 0.0001973155626151219,
      "loss": 4.8614,
      "step": 14400
    },
    {
      "epoch": 0.013432451651971329,
      "grad_norm": 1.2468639612197876,
      "learning_rate": 0.00019731369829302671,
      "loss": 4.5443,
      "step": 14410
    },
    {
      "epoch": 0.013441773270050422,
      "grad_norm": 0.9425323605537415,
      "learning_rate": 0.0001973118339709315,
      "loss": 4.6482,
      "step": 14420
    },
    {
      "epoch": 0.013451094888129513,
      "grad_norm": 1.2870475053787231,
      "learning_rate": 0.0001973099696488363,
      "loss": 4.5742,
      "step": 14430
    },
    {
      "epoch": 0.013460416506208606,
      "grad_norm": 1.1684921979904175,
      "learning_rate": 0.0001973081053267411,
      "loss": 4.5791,
      "step": 14440
    },
    {
      "epoch": 0.013469738124287697,
      "grad_norm": 1.154624342918396,
      "learning_rate": 0.00019730624100464588,
      "loss": 4.5324,
      "step": 14450
    },
    {
      "epoch": 0.01347905974236679,
      "grad_norm": 1.0867828130722046,
      "learning_rate": 0.00019730437668255072,
      "loss": 4.6749,
      "step": 14460
    },
    {
      "epoch": 0.01348838136044588,
      "grad_norm": 1.1875011920928955,
      "learning_rate": 0.0001973025123604555,
      "loss": 4.6191,
      "step": 14470
    },
    {
      "epoch": 0.013497702978524973,
      "grad_norm": 1.0676566362380981,
      "learning_rate": 0.0001973006480383603,
      "loss": 4.5377,
      "step": 14480
    },
    {
      "epoch": 0.013507024596604064,
      "grad_norm": 1.1208975315093994,
      "learning_rate": 0.0001972987837162651,
      "loss": 4.4462,
      "step": 14490
    },
    {
      "epoch": 0.013516346214683157,
      "grad_norm": 1.2080246210098267,
      "learning_rate": 0.0001972969193941699,
      "loss": 4.7231,
      "step": 14500
    },
    {
      "epoch": 0.013525667832762248,
      "grad_norm": 1.1570172309875488,
      "learning_rate": 0.0001972950550720747,
      "loss": 4.6338,
      "step": 14510
    },
    {
      "epoch": 0.01353498945084134,
      "grad_norm": 1.131277084350586,
      "learning_rate": 0.0001972931907499795,
      "loss": 4.565,
      "step": 14520
    },
    {
      "epoch": 0.013544311068920432,
      "grad_norm": 1.2249164581298828,
      "learning_rate": 0.0001972913264278843,
      "loss": 4.4683,
      "step": 14530
    },
    {
      "epoch": 0.013553632686999524,
      "grad_norm": 1.3555735349655151,
      "learning_rate": 0.00019728946210578911,
      "loss": 4.4902,
      "step": 14540
    },
    {
      "epoch": 0.013562954305078615,
      "grad_norm": 1.1136330366134644,
      "learning_rate": 0.0001972875977836939,
      "loss": 4.6817,
      "step": 14550
    },
    {
      "epoch": 0.013572275923157708,
      "grad_norm": 1.1630266904830933,
      "learning_rate": 0.0001972857334615987,
      "loss": 4.6089,
      "step": 14560
    },
    {
      "epoch": 0.0135815975412368,
      "grad_norm": 0.9931058883666992,
      "learning_rate": 0.0001972838691395035,
      "loss": 4.6447,
      "step": 14570
    },
    {
      "epoch": 0.01359091915931589,
      "grad_norm": 1.2403388023376465,
      "learning_rate": 0.0001972820048174083,
      "loss": 4.6948,
      "step": 14580
    },
    {
      "epoch": 0.013600240777394983,
      "grad_norm": 1.2933096885681152,
      "learning_rate": 0.0001972801404953131,
      "loss": 4.5001,
      "step": 14590
    },
    {
      "epoch": 0.013609562395474074,
      "grad_norm": 1.2163969278335571,
      "learning_rate": 0.0001972782761732179,
      "loss": 4.4378,
      "step": 14600
    },
    {
      "epoch": 0.013618884013553167,
      "grad_norm": 1.057443618774414,
      "learning_rate": 0.0001972764118511227,
      "loss": 4.453,
      "step": 14610
    },
    {
      "epoch": 0.013628205631632258,
      "grad_norm": 1.1429438591003418,
      "learning_rate": 0.0001972745475290275,
      "loss": 4.5533,
      "step": 14620
    },
    {
      "epoch": 0.01363752724971135,
      "grad_norm": 1.3044370412826538,
      "learning_rate": 0.00019727268320693232,
      "loss": 4.8637,
      "step": 14630
    },
    {
      "epoch": 0.013646848867790442,
      "grad_norm": 1.2470521926879883,
      "learning_rate": 0.0001972708188848371,
      "loss": 4.4644,
      "step": 14640
    },
    {
      "epoch": 0.013656170485869534,
      "grad_norm": 1.2091494798660278,
      "learning_rate": 0.0001972689545627419,
      "loss": 4.7067,
      "step": 14650
    },
    {
      "epoch": 0.013665492103948625,
      "grad_norm": 0.9937439560890198,
      "learning_rate": 0.0001972670902406467,
      "loss": 4.7278,
      "step": 14660
    },
    {
      "epoch": 0.013674813722027718,
      "grad_norm": 1.2041183710098267,
      "learning_rate": 0.0001972652259185515,
      "loss": 4.4,
      "step": 14670
    },
    {
      "epoch": 0.013684135340106809,
      "grad_norm": 1.080781102180481,
      "learning_rate": 0.00019726336159645632,
      "loss": 4.5031,
      "step": 14680
    },
    {
      "epoch": 0.013693456958185902,
      "grad_norm": 1.1291261911392212,
      "learning_rate": 0.0001972614972743611,
      "loss": 4.4067,
      "step": 14690
    },
    {
      "epoch": 0.013702778576264993,
      "grad_norm": 0.9233282208442688,
      "learning_rate": 0.0001972596329522659,
      "loss": 4.3948,
      "step": 14700
    },
    {
      "epoch": 0.013712100194344086,
      "grad_norm": 1.2404347658157349,
      "learning_rate": 0.0001972577686301707,
      "loss": 4.4589,
      "step": 14710
    },
    {
      "epoch": 0.013721421812423177,
      "grad_norm": 1.2120004892349243,
      "learning_rate": 0.0001972559043080755,
      "loss": 4.643,
      "step": 14720
    },
    {
      "epoch": 0.01373074343050227,
      "grad_norm": 1.306998372077942,
      "learning_rate": 0.0001972540399859803,
      "loss": 4.3683,
      "step": 14730
    },
    {
      "epoch": 0.01374006504858136,
      "grad_norm": 1.2773089408874512,
      "learning_rate": 0.0001972521756638851,
      "loss": 4.5566,
      "step": 14740
    },
    {
      "epoch": 0.013749386666660451,
      "grad_norm": 1.037179946899414,
      "learning_rate": 0.0001972503113417899,
      "loss": 4.582,
      "step": 14750
    },
    {
      "epoch": 0.013758708284739544,
      "grad_norm": 1.2319579124450684,
      "learning_rate": 0.00019724844701969472,
      "loss": 4.438,
      "step": 14760
    },
    {
      "epoch": 0.013768029902818635,
      "grad_norm": 1.167433738708496,
      "learning_rate": 0.0001972465826975995,
      "loss": 4.4886,
      "step": 14770
    },
    {
      "epoch": 0.013777351520897728,
      "grad_norm": 1.3190443515777588,
      "learning_rate": 0.0001972447183755043,
      "loss": 4.5865,
      "step": 14780
    },
    {
      "epoch": 0.013786673138976819,
      "grad_norm": 1.175478458404541,
      "learning_rate": 0.0001972428540534091,
      "loss": 4.4835,
      "step": 14790
    },
    {
      "epoch": 0.013795994757055912,
      "grad_norm": 1.259690761566162,
      "learning_rate": 0.00019724098973131392,
      "loss": 4.6152,
      "step": 14800
    },
    {
      "epoch": 0.013805316375135003,
      "grad_norm": 1.1667793989181519,
      "learning_rate": 0.00019723912540921872,
      "loss": 4.5129,
      "step": 14810
    },
    {
      "epoch": 0.013814637993214095,
      "grad_norm": 1.1799503564834595,
      "learning_rate": 0.0001972372610871235,
      "loss": 4.4389,
      "step": 14820
    },
    {
      "epoch": 0.013823959611293186,
      "grad_norm": 1.2986341714859009,
      "learning_rate": 0.0001972353967650283,
      "loss": 4.6323,
      "step": 14830
    },
    {
      "epoch": 0.01383328122937228,
      "grad_norm": 1.1824814081192017,
      "learning_rate": 0.0001972335324429331,
      "loss": 4.4849,
      "step": 14840
    },
    {
      "epoch": 0.01384260284745137,
      "grad_norm": 1.460556983947754,
      "learning_rate": 0.00019723166812083792,
      "loss": 4.3904,
      "step": 14850
    },
    {
      "epoch": 0.013851924465530463,
      "grad_norm": 1.317578673362732,
      "learning_rate": 0.00019722980379874273,
      "loss": 4.6748,
      "step": 14860
    },
    {
      "epoch": 0.013861246083609554,
      "grad_norm": 1.2828431129455566,
      "learning_rate": 0.0001972279394766475,
      "loss": 4.5867,
      "step": 14870
    },
    {
      "epoch": 0.013870567701688647,
      "grad_norm": 1.24486243724823,
      "learning_rate": 0.0001972260751545523,
      "loss": 4.5907,
      "step": 14880
    },
    {
      "epoch": 0.013879889319767738,
      "grad_norm": 1.1922328472137451,
      "learning_rate": 0.0001972242108324571,
      "loss": 4.6755,
      "step": 14890
    },
    {
      "epoch": 0.01388921093784683,
      "grad_norm": 1.1863383054733276,
      "learning_rate": 0.0001972223465103619,
      "loss": 4.5777,
      "step": 14900
    },
    {
      "epoch": 0.013898532555925922,
      "grad_norm": 1.2859413623809814,
      "learning_rate": 0.00019722048218826673,
      "loss": 4.707,
      "step": 14910
    },
    {
      "epoch": 0.013907854174005013,
      "grad_norm": 1.3614192008972168,
      "learning_rate": 0.0001972186178661715,
      "loss": 4.6493,
      "step": 14920
    },
    {
      "epoch": 0.013917175792084105,
      "grad_norm": 1.4349111318588257,
      "learning_rate": 0.00019721675354407632,
      "loss": 4.4933,
      "step": 14930
    },
    {
      "epoch": 0.013926497410163196,
      "grad_norm": 1.2505768537521362,
      "learning_rate": 0.0001972148892219811,
      "loss": 4.4877,
      "step": 14940
    },
    {
      "epoch": 0.013935819028242289,
      "grad_norm": 1.2182815074920654,
      "learning_rate": 0.0001972130248998859,
      "loss": 4.8049,
      "step": 14950
    },
    {
      "epoch": 0.01394514064632138,
      "grad_norm": 1.1866166591644287,
      "learning_rate": 0.0001972111605777907,
      "loss": 4.5108,
      "step": 14960
    },
    {
      "epoch": 0.013954462264400473,
      "grad_norm": 1.1102325916290283,
      "learning_rate": 0.00019720929625569551,
      "loss": 4.5283,
      "step": 14970
    },
    {
      "epoch": 0.013963783882479564,
      "grad_norm": 1.2027043104171753,
      "learning_rate": 0.00019720743193360032,
      "loss": 4.6855,
      "step": 14980
    },
    {
      "epoch": 0.013973105500558657,
      "grad_norm": 1.0795619487762451,
      "learning_rate": 0.0001972055676115051,
      "loss": 4.544,
      "step": 14990
    },
    {
      "epoch": 0.013982427118637748,
      "grad_norm": 1.1690608263015747,
      "learning_rate": 0.0001972037032894099,
      "loss": 4.5694,
      "step": 15000
    },
    {
      "epoch": 0.01399174873671684,
      "grad_norm": 1.1501505374908447,
      "learning_rate": 0.0001972018389673147,
      "loss": 4.7352,
      "step": 15010
    },
    {
      "epoch": 0.014001070354795931,
      "grad_norm": 1.30429208278656,
      "learning_rate": 0.00019719997464521952,
      "loss": 4.4057,
      "step": 15020
    },
    {
      "epoch": 0.014010391972875024,
      "grad_norm": 1.0262491703033447,
      "learning_rate": 0.00019719811032312433,
      "loss": 4.3876,
      "step": 15030
    },
    {
      "epoch": 0.014019713590954115,
      "grad_norm": 1.6021312475204468,
      "learning_rate": 0.0001971962460010291,
      "loss": 4.5234,
      "step": 15040
    },
    {
      "epoch": 0.014029035209033208,
      "grad_norm": 1.1501836776733398,
      "learning_rate": 0.0001971943816789339,
      "loss": 4.4423,
      "step": 15050
    },
    {
      "epoch": 0.014038356827112299,
      "grad_norm": 0.986639678478241,
      "learning_rate": 0.00019719251735683872,
      "loss": 4.5828,
      "step": 15060
    },
    {
      "epoch": 0.014047678445191392,
      "grad_norm": 1.119666337966919,
      "learning_rate": 0.00019719065303474352,
      "loss": 4.5764,
      "step": 15070
    },
    {
      "epoch": 0.014057000063270483,
      "grad_norm": 1.1854208707809448,
      "learning_rate": 0.00019718878871264833,
      "loss": 4.5532,
      "step": 15080
    },
    {
      "epoch": 0.014066321681349575,
      "grad_norm": 1.2806496620178223,
      "learning_rate": 0.0001971869243905531,
      "loss": 4.615,
      "step": 15090
    },
    {
      "epoch": 0.014075643299428666,
      "grad_norm": 1.2376620769500732,
      "learning_rate": 0.00019718506006845792,
      "loss": 4.6786,
      "step": 15100
    },
    {
      "epoch": 0.014084964917507757,
      "grad_norm": 1.1436574459075928,
      "learning_rate": 0.00019718319574636272,
      "loss": 4.4705,
      "step": 15110
    },
    {
      "epoch": 0.01409428653558685,
      "grad_norm": 1.189790964126587,
      "learning_rate": 0.0001971813314242675,
      "loss": 4.7273,
      "step": 15120
    },
    {
      "epoch": 0.014103608153665941,
      "grad_norm": 1.1371002197265625,
      "learning_rate": 0.00019717946710217233,
      "loss": 4.5647,
      "step": 15130
    },
    {
      "epoch": 0.014112929771745034,
      "grad_norm": 1.3244099617004395,
      "learning_rate": 0.0001971776027800771,
      "loss": 4.3241,
      "step": 15140
    },
    {
      "epoch": 0.014122251389824125,
      "grad_norm": 1.17960786819458,
      "learning_rate": 0.00019717573845798192,
      "loss": 4.6605,
      "step": 15150
    },
    {
      "epoch": 0.014131573007903218,
      "grad_norm": 1.1162618398666382,
      "learning_rate": 0.00019717387413588673,
      "loss": 4.5696,
      "step": 15160
    },
    {
      "epoch": 0.014140894625982309,
      "grad_norm": 1.1660070419311523,
      "learning_rate": 0.0001971720098137915,
      "loss": 4.5798,
      "step": 15170
    },
    {
      "epoch": 0.014150216244061401,
      "grad_norm": 0.9883113503456116,
      "learning_rate": 0.0001971701454916963,
      "loss": 4.5564,
      "step": 15180
    },
    {
      "epoch": 0.014159537862140492,
      "grad_norm": 1.030522108078003,
      "learning_rate": 0.00019716828116960112,
      "loss": 4.7494,
      "step": 15190
    },
    {
      "epoch": 0.014168859480219585,
      "grad_norm": 1.072016954421997,
      "learning_rate": 0.00019716641684750592,
      "loss": 4.5501,
      "step": 15200
    },
    {
      "epoch": 0.014178181098298676,
      "grad_norm": 1.1946282386779785,
      "learning_rate": 0.00019716455252541073,
      "loss": 4.3455,
      "step": 15210
    },
    {
      "epoch": 0.014187502716377769,
      "grad_norm": 1.004911184310913,
      "learning_rate": 0.0001971626882033155,
      "loss": 4.6878,
      "step": 15220
    },
    {
      "epoch": 0.01419682433445686,
      "grad_norm": 1.32917058467865,
      "learning_rate": 0.00019716082388122032,
      "loss": 4.4388,
      "step": 15230
    },
    {
      "epoch": 0.014206145952535953,
      "grad_norm": 1.0663623809814453,
      "learning_rate": 0.00019715895955912512,
      "loss": 4.675,
      "step": 15240
    },
    {
      "epoch": 0.014215467570615044,
      "grad_norm": 1.128658652305603,
      "learning_rate": 0.00019715709523702993,
      "loss": 4.4117,
      "step": 15250
    },
    {
      "epoch": 0.014224789188694137,
      "grad_norm": 1.207233190536499,
      "learning_rate": 0.00019715523091493473,
      "loss": 4.6143,
      "step": 15260
    },
    {
      "epoch": 0.014234110806773228,
      "grad_norm": 1.1651039123535156,
      "learning_rate": 0.00019715336659283951,
      "loss": 4.6196,
      "step": 15270
    },
    {
      "epoch": 0.014243432424852319,
      "grad_norm": 1.1074390411376953,
      "learning_rate": 0.00019715150227074432,
      "loss": 4.6544,
      "step": 15280
    },
    {
      "epoch": 0.014252754042931411,
      "grad_norm": 1.1875381469726562,
      "learning_rate": 0.0001971496379486491,
      "loss": 4.6448,
      "step": 15290
    },
    {
      "epoch": 0.014262075661010502,
      "grad_norm": 1.2167370319366455,
      "learning_rate": 0.00019714777362655393,
      "loss": 4.5056,
      "step": 15300
    },
    {
      "epoch": 0.014271397279089595,
      "grad_norm": 1.1701884269714355,
      "learning_rate": 0.00019714590930445874,
      "loss": 4.4417,
      "step": 15310
    },
    {
      "epoch": 0.014280718897168686,
      "grad_norm": 1.4548020362854004,
      "learning_rate": 0.00019714404498236352,
      "loss": 4.3671,
      "step": 15320
    },
    {
      "epoch": 0.014290040515247779,
      "grad_norm": 1.1412996053695679,
      "learning_rate": 0.00019714218066026832,
      "loss": 4.5598,
      "step": 15330
    },
    {
      "epoch": 0.01429936213332687,
      "grad_norm": 1.1457619667053223,
      "learning_rate": 0.0001971403163381731,
      "loss": 4.6983,
      "step": 15340
    },
    {
      "epoch": 0.014308683751405963,
      "grad_norm": 1.2217504978179932,
      "learning_rate": 0.0001971384520160779,
      "loss": 4.393,
      "step": 15350
    },
    {
      "epoch": 0.014318005369485054,
      "grad_norm": 1.2563819885253906,
      "learning_rate": 0.00019713658769398274,
      "loss": 4.5156,
      "step": 15360
    },
    {
      "epoch": 0.014327326987564146,
      "grad_norm": 1.1505266427993774,
      "learning_rate": 0.00019713472337188752,
      "loss": 4.4596,
      "step": 15370
    },
    {
      "epoch": 0.014336648605643237,
      "grad_norm": 1.230899691581726,
      "learning_rate": 0.00019713285904979233,
      "loss": 4.6972,
      "step": 15380
    },
    {
      "epoch": 0.01434597022372233,
      "grad_norm": 1.3852901458740234,
      "learning_rate": 0.0001971309947276971,
      "loss": 4.6613,
      "step": 15390
    },
    {
      "epoch": 0.014355291841801421,
      "grad_norm": 1.2801262140274048,
      "learning_rate": 0.00019712913040560191,
      "loss": 4.7603,
      "step": 15400
    },
    {
      "epoch": 0.014364613459880514,
      "grad_norm": 1.0653603076934814,
      "learning_rate": 0.00019712726608350672,
      "loss": 4.4398,
      "step": 15410
    },
    {
      "epoch": 0.014373935077959605,
      "grad_norm": 1.2889834642410278,
      "learning_rate": 0.00019712540176141153,
      "loss": 4.7122,
      "step": 15420
    },
    {
      "epoch": 0.014383256696038698,
      "grad_norm": 1.3186266422271729,
      "learning_rate": 0.00019712353743931633,
      "loss": 4.4822,
      "step": 15430
    },
    {
      "epoch": 0.014392578314117789,
      "grad_norm": 1.1218372583389282,
      "learning_rate": 0.0001971216731172211,
      "loss": 4.5032,
      "step": 15440
    },
    {
      "epoch": 0.01440189993219688,
      "grad_norm": 0.9462302327156067,
      "learning_rate": 0.00019711980879512592,
      "loss": 4.6447,
      "step": 15450
    },
    {
      "epoch": 0.014411221550275972,
      "grad_norm": 1.1327317953109741,
      "learning_rate": 0.00019711794447303073,
      "loss": 4.695,
      "step": 15460
    },
    {
      "epoch": 0.014420543168355063,
      "grad_norm": 0.9211001396179199,
      "learning_rate": 0.00019711608015093553,
      "loss": 4.5235,
      "step": 15470
    },
    {
      "epoch": 0.014429864786434156,
      "grad_norm": 1.2882301807403564,
      "learning_rate": 0.00019711421582884034,
      "loss": 4.6056,
      "step": 15480
    },
    {
      "epoch": 0.014439186404513247,
      "grad_norm": 1.1264674663543701,
      "learning_rate": 0.00019711235150674512,
      "loss": 4.6856,
      "step": 15490
    },
    {
      "epoch": 0.01444850802259234,
      "grad_norm": 1.19175124168396,
      "learning_rate": 0.00019711048718464992,
      "loss": 4.327,
      "step": 15500
    },
    {
      "epoch": 0.014457829640671431,
      "grad_norm": 1.2877241373062134,
      "learning_rate": 0.00019710862286255473,
      "loss": 4.4442,
      "step": 15510
    },
    {
      "epoch": 0.014467151258750524,
      "grad_norm": 1.129870891571045,
      "learning_rate": 0.00019710675854045954,
      "loss": 4.5123,
      "step": 15520
    },
    {
      "epoch": 0.014476472876829615,
      "grad_norm": 1.2695037126541138,
      "learning_rate": 0.00019710489421836434,
      "loss": 4.5862,
      "step": 15530
    },
    {
      "epoch": 0.014485794494908708,
      "grad_norm": 1.0797818899154663,
      "learning_rate": 0.00019710302989626912,
      "loss": 4.4429,
      "step": 15540
    },
    {
      "epoch": 0.014495116112987799,
      "grad_norm": 1.1550147533416748,
      "learning_rate": 0.00019710116557417393,
      "loss": 4.7225,
      "step": 15550
    },
    {
      "epoch": 0.014504437731066891,
      "grad_norm": 1.3327094316482544,
      "learning_rate": 0.00019709930125207873,
      "loss": 4.5009,
      "step": 15560
    },
    {
      "epoch": 0.014513759349145982,
      "grad_norm": 1.3001959323883057,
      "learning_rate": 0.00019709743692998351,
      "loss": 4.5134,
      "step": 15570
    },
    {
      "epoch": 0.014523080967225075,
      "grad_norm": 1.248551607131958,
      "learning_rate": 0.00019709557260788835,
      "loss": 4.6794,
      "step": 15580
    },
    {
      "epoch": 0.014532402585304166,
      "grad_norm": 1.3199856281280518,
      "learning_rate": 0.00019709370828579313,
      "loss": 4.6641,
      "step": 15590
    },
    {
      "epoch": 0.014541724203383259,
      "grad_norm": 1.1165252923965454,
      "learning_rate": 0.00019709184396369793,
      "loss": 4.6322,
      "step": 15600
    },
    {
      "epoch": 0.01455104582146235,
      "grad_norm": 1.15933096408844,
      "learning_rate": 0.00019708997964160274,
      "loss": 4.4526,
      "step": 15610
    },
    {
      "epoch": 0.01456036743954144,
      "grad_norm": 1.0587568283081055,
      "learning_rate": 0.00019708811531950752,
      "loss": 4.5075,
      "step": 15620
    },
    {
      "epoch": 0.014569689057620534,
      "grad_norm": 1.1999831199645996,
      "learning_rate": 0.00019708625099741232,
      "loss": 4.6834,
      "step": 15630
    },
    {
      "epoch": 0.014579010675699625,
      "grad_norm": 1.4396162033081055,
      "learning_rate": 0.00019708438667531713,
      "loss": 4.5957,
      "step": 15640
    },
    {
      "epoch": 0.014588332293778717,
      "grad_norm": 1.0853149890899658,
      "learning_rate": 0.00019708252235322194,
      "loss": 4.4683,
      "step": 15650
    },
    {
      "epoch": 0.014597653911857808,
      "grad_norm": 1.126427412033081,
      "learning_rate": 0.00019708065803112674,
      "loss": 4.4424,
      "step": 15660
    },
    {
      "epoch": 0.014606975529936901,
      "grad_norm": 1.10371994972229,
      "learning_rate": 0.00019707879370903152,
      "loss": 4.7793,
      "step": 15670
    },
    {
      "epoch": 0.014616297148015992,
      "grad_norm": 1.084664225578308,
      "learning_rate": 0.00019707692938693633,
      "loss": 4.3275,
      "step": 15680
    },
    {
      "epoch": 0.014625618766095085,
      "grad_norm": 1.1763403415679932,
      "learning_rate": 0.00019707506506484114,
      "loss": 4.4605,
      "step": 15690
    },
    {
      "epoch": 0.014634940384174176,
      "grad_norm": 1.2684112787246704,
      "learning_rate": 0.00019707320074274594,
      "loss": 4.5316,
      "step": 15700
    },
    {
      "epoch": 0.014644262002253269,
      "grad_norm": 1.336646556854248,
      "learning_rate": 0.00019707133642065075,
      "loss": 4.3559,
      "step": 15710
    },
    {
      "epoch": 0.01465358362033236,
      "grad_norm": 1.129859209060669,
      "learning_rate": 0.00019706947209855553,
      "loss": 4.4544,
      "step": 15720
    },
    {
      "epoch": 0.014662905238411452,
      "grad_norm": 1.1846956014633179,
      "learning_rate": 0.00019706760777646033,
      "loss": 4.7636,
      "step": 15730
    },
    {
      "epoch": 0.014672226856490543,
      "grad_norm": 1.1434651613235474,
      "learning_rate": 0.00019706574345436514,
      "loss": 4.5762,
      "step": 15740
    },
    {
      "epoch": 0.014681548474569636,
      "grad_norm": 0.9821406602859497,
      "learning_rate": 0.00019706387913226995,
      "loss": 4.5455,
      "step": 15750
    },
    {
      "epoch": 0.014690870092648727,
      "grad_norm": 1.3292255401611328,
      "learning_rate": 0.00019706201481017475,
      "loss": 4.3813,
      "step": 15760
    },
    {
      "epoch": 0.01470019171072782,
      "grad_norm": 1.034483551979065,
      "learning_rate": 0.00019706015048807953,
      "loss": 4.5836,
      "step": 15770
    },
    {
      "epoch": 0.014709513328806911,
      "grad_norm": 1.1822052001953125,
      "learning_rate": 0.00019705828616598434,
      "loss": 4.5363,
      "step": 15780
    },
    {
      "epoch": 0.014718834946886002,
      "grad_norm": 1.0633821487426758,
      "learning_rate": 0.00019705642184388912,
      "loss": 4.4103,
      "step": 15790
    },
    {
      "epoch": 0.014728156564965095,
      "grad_norm": 1.0804098844528198,
      "learning_rate": 0.00019705455752179392,
      "loss": 4.6184,
      "step": 15800
    },
    {
      "epoch": 0.014737478183044186,
      "grad_norm": 1.2784181833267212,
      "learning_rate": 0.00019705269319969876,
      "loss": 4.5204,
      "step": 15810
    },
    {
      "epoch": 0.014746799801123278,
      "grad_norm": 1.1875311136245728,
      "learning_rate": 0.00019705082887760354,
      "loss": 4.7592,
      "step": 15820
    },
    {
      "epoch": 0.01475612141920237,
      "grad_norm": 1.1727021932601929,
      "learning_rate": 0.00019704896455550834,
      "loss": 4.7068,
      "step": 15830
    },
    {
      "epoch": 0.014765443037281462,
      "grad_norm": 1.0894187688827515,
      "learning_rate": 0.00019704710023341312,
      "loss": 4.4694,
      "step": 15840
    },
    {
      "epoch": 0.014774764655360553,
      "grad_norm": 1.183312177658081,
      "learning_rate": 0.00019704523591131793,
      "loss": 4.7549,
      "step": 15850
    },
    {
      "epoch": 0.014784086273439646,
      "grad_norm": 1.1912903785705566,
      "learning_rate": 0.00019704337158922273,
      "loss": 4.7577,
      "step": 15860
    },
    {
      "epoch": 0.014793407891518737,
      "grad_norm": 1.137644648551941,
      "learning_rate": 0.00019704150726712754,
      "loss": 4.5402,
      "step": 15870
    },
    {
      "epoch": 0.01480272950959783,
      "grad_norm": 1.1343845129013062,
      "learning_rate": 0.00019703964294503235,
      "loss": 4.3262,
      "step": 15880
    },
    {
      "epoch": 0.01481205112767692,
      "grad_norm": 1.2101548910140991,
      "learning_rate": 0.00019703777862293713,
      "loss": 4.4963,
      "step": 15890
    },
    {
      "epoch": 0.014821372745756014,
      "grad_norm": 1.195426106452942,
      "learning_rate": 0.00019703591430084193,
      "loss": 4.6749,
      "step": 15900
    },
    {
      "epoch": 0.014830694363835105,
      "grad_norm": 1.187921404838562,
      "learning_rate": 0.00019703404997874674,
      "loss": 4.5793,
      "step": 15910
    },
    {
      "epoch": 0.014840015981914197,
      "grad_norm": 1.0574133396148682,
      "learning_rate": 0.00019703218565665154,
      "loss": 4.8141,
      "step": 15920
    },
    {
      "epoch": 0.014849337599993288,
      "grad_norm": 1.1777557134628296,
      "learning_rate": 0.00019703032133455635,
      "loss": 4.7728,
      "step": 15930
    },
    {
      "epoch": 0.014858659218072381,
      "grad_norm": 1.1829129457473755,
      "learning_rate": 0.00019702845701246113,
      "loss": 4.5661,
      "step": 15940
    },
    {
      "epoch": 0.014867980836151472,
      "grad_norm": 1.068943738937378,
      "learning_rate": 0.00019702659269036594,
      "loss": 4.6588,
      "step": 15950
    },
    {
      "epoch": 0.014877302454230563,
      "grad_norm": 0.9926013350486755,
      "learning_rate": 0.00019702472836827072,
      "loss": 4.6019,
      "step": 15960
    },
    {
      "epoch": 0.014886624072309656,
      "grad_norm": 1.2847212553024292,
      "learning_rate": 0.00019702286404617555,
      "loss": 4.4969,
      "step": 15970
    },
    {
      "epoch": 0.014895945690388747,
      "grad_norm": 1.3855618238449097,
      "learning_rate": 0.00019702099972408036,
      "loss": 4.3835,
      "step": 15980
    },
    {
      "epoch": 0.01490526730846784,
      "grad_norm": 1.275460124015808,
      "learning_rate": 0.00019701913540198513,
      "loss": 4.5898,
      "step": 15990
    },
    {
      "epoch": 0.01491458892654693,
      "grad_norm": 1.112544298171997,
      "learning_rate": 0.00019701727107988994,
      "loss": 4.6266,
      "step": 16000
    },
    {
      "epoch": 0.014923910544626023,
      "grad_norm": 1.2512692213058472,
      "learning_rate": 0.00019701540675779472,
      "loss": 4.3252,
      "step": 16010
    },
    {
      "epoch": 0.014933232162705114,
      "grad_norm": 1.1742918491363525,
      "learning_rate": 0.00019701354243569953,
      "loss": 4.6177,
      "step": 16020
    },
    {
      "epoch": 0.014942553780784207,
      "grad_norm": 1.212009072303772,
      "learning_rate": 0.00019701167811360436,
      "loss": 4.4828,
      "step": 16030
    },
    {
      "epoch": 0.014951875398863298,
      "grad_norm": 1.1574987173080444,
      "learning_rate": 0.00019700981379150914,
      "loss": 4.5078,
      "step": 16040
    },
    {
      "epoch": 0.014961197016942391,
      "grad_norm": 1.1302474737167358,
      "learning_rate": 0.00019700794946941395,
      "loss": 4.5198,
      "step": 16050
    },
    {
      "epoch": 0.014970518635021482,
      "grad_norm": 1.280281662940979,
      "learning_rate": 0.00019700608514731872,
      "loss": 4.6609,
      "step": 16060
    },
    {
      "epoch": 0.014979840253100575,
      "grad_norm": 1.0537630319595337,
      "learning_rate": 0.00019700422082522353,
      "loss": 4.6987,
      "step": 16070
    },
    {
      "epoch": 0.014989161871179666,
      "grad_norm": 1.3069387674331665,
      "learning_rate": 0.00019700235650312834,
      "loss": 4.376,
      "step": 16080
    },
    {
      "epoch": 0.014998483489258758,
      "grad_norm": 1.2626792192459106,
      "learning_rate": 0.00019700049218103314,
      "loss": 4.5639,
      "step": 16090
    },
    {
      "epoch": 0.01500780510733785,
      "grad_norm": 1.0824650526046753,
      "learning_rate": 0.00019699862785893795,
      "loss": 4.5345,
      "step": 16100
    },
    {
      "epoch": 0.015017126725416942,
      "grad_norm": 1.2551994323730469,
      "learning_rate": 0.00019699676353684273,
      "loss": 4.1789,
      "step": 16110
    },
    {
      "epoch": 0.015026448343496033,
      "grad_norm": 1.0961027145385742,
      "learning_rate": 0.00019699489921474754,
      "loss": 4.4843,
      "step": 16120
    },
    {
      "epoch": 0.015035769961575124,
      "grad_norm": 1.1243242025375366,
      "learning_rate": 0.00019699303489265234,
      "loss": 4.5561,
      "step": 16130
    },
    {
      "epoch": 0.015045091579654217,
      "grad_norm": 1.0378987789154053,
      "learning_rate": 0.00019699117057055715,
      "loss": 4.4355,
      "step": 16140
    },
    {
      "epoch": 0.015054413197733308,
      "grad_norm": 1.0572618246078491,
      "learning_rate": 0.00019698930624846195,
      "loss": 4.4939,
      "step": 16150
    },
    {
      "epoch": 0.0150637348158124,
      "grad_norm": 1.16074538230896,
      "learning_rate": 0.00019698744192636673,
      "loss": 4.5162,
      "step": 16160
    },
    {
      "epoch": 0.015073056433891492,
      "grad_norm": 1.1572027206420898,
      "learning_rate": 0.00019698557760427154,
      "loss": 4.499,
      "step": 16170
    },
    {
      "epoch": 0.015082378051970585,
      "grad_norm": 1.2743443250656128,
      "learning_rate": 0.00019698371328217635,
      "loss": 4.821,
      "step": 16180
    },
    {
      "epoch": 0.015091699670049676,
      "grad_norm": 1.0955778360366821,
      "learning_rate": 0.00019698184896008115,
      "loss": 4.6301,
      "step": 16190
    },
    {
      "epoch": 0.015101021288128768,
      "grad_norm": 1.0019911527633667,
      "learning_rate": 0.00019697998463798596,
      "loss": 4.3337,
      "step": 16200
    },
    {
      "epoch": 0.01511034290620786,
      "grad_norm": 1.0956101417541504,
      "learning_rate": 0.00019697812031589074,
      "loss": 4.8183,
      "step": 16210
    },
    {
      "epoch": 0.015119664524286952,
      "grad_norm": 1.1428101062774658,
      "learning_rate": 0.00019697625599379554,
      "loss": 4.6617,
      "step": 16220
    },
    {
      "epoch": 0.015128986142366043,
      "grad_norm": 1.0649453401565552,
      "learning_rate": 0.00019697439167170035,
      "loss": 4.4787,
      "step": 16230
    },
    {
      "epoch": 0.015138307760445136,
      "grad_norm": 1.3154183626174927,
      "learning_rate": 0.00019697252734960513,
      "loss": 4.4909,
      "step": 16240
    },
    {
      "epoch": 0.015147629378524227,
      "grad_norm": 1.1835122108459473,
      "learning_rate": 0.00019697066302750996,
      "loss": 4.3425,
      "step": 16250
    },
    {
      "epoch": 0.01515695099660332,
      "grad_norm": 1.1754130125045776,
      "learning_rate": 0.00019696879870541474,
      "loss": 4.5614,
      "step": 16260
    },
    {
      "epoch": 0.01516627261468241,
      "grad_norm": 1.123792052268982,
      "learning_rate": 0.00019696693438331955,
      "loss": 4.5972,
      "step": 16270
    },
    {
      "epoch": 0.015175594232761503,
      "grad_norm": 1.1091179847717285,
      "learning_rate": 0.00019696507006122435,
      "loss": 4.4745,
      "step": 16280
    },
    {
      "epoch": 0.015184915850840594,
      "grad_norm": 1.3666237592697144,
      "learning_rate": 0.00019696320573912913,
      "loss": 4.4587,
      "step": 16290
    },
    {
      "epoch": 0.015194237468919685,
      "grad_norm": 0.9802164435386658,
      "learning_rate": 0.00019696134141703394,
      "loss": 4.68,
      "step": 16300
    },
    {
      "epoch": 0.015203559086998778,
      "grad_norm": 1.1291648149490356,
      "learning_rate": 0.00019695947709493875,
      "loss": 4.4327,
      "step": 16310
    },
    {
      "epoch": 0.01521288070507787,
      "grad_norm": 1.125247836112976,
      "learning_rate": 0.00019695761277284355,
      "loss": 4.4637,
      "step": 16320
    },
    {
      "epoch": 0.015222202323156962,
      "grad_norm": 1.2187532186508179,
      "learning_rate": 0.00019695574845074836,
      "loss": 4.5639,
      "step": 16330
    },
    {
      "epoch": 0.015231523941236053,
      "grad_norm": 1.194108247756958,
      "learning_rate": 0.00019695388412865314,
      "loss": 4.2167,
      "step": 16340
    },
    {
      "epoch": 0.015240845559315146,
      "grad_norm": 1.090761661529541,
      "learning_rate": 0.00019695201980655794,
      "loss": 4.5957,
      "step": 16350
    },
    {
      "epoch": 0.015250167177394237,
      "grad_norm": 1.206772804260254,
      "learning_rate": 0.00019695015548446275,
      "loss": 4.6344,
      "step": 16360
    },
    {
      "epoch": 0.01525948879547333,
      "grad_norm": 1.170844316482544,
      "learning_rate": 0.00019694829116236756,
      "loss": 4.5691,
      "step": 16370
    },
    {
      "epoch": 0.01526881041355242,
      "grad_norm": 1.1810002326965332,
      "learning_rate": 0.00019694642684027236,
      "loss": 4.6204,
      "step": 16380
    },
    {
      "epoch": 0.015278132031631513,
      "grad_norm": 1.1060277223587036,
      "learning_rate": 0.00019694456251817714,
      "loss": 4.6078,
      "step": 16390
    },
    {
      "epoch": 0.015287453649710604,
      "grad_norm": 1.3327161073684692,
      "learning_rate": 0.00019694269819608195,
      "loss": 4.6423,
      "step": 16400
    },
    {
      "epoch": 0.015296775267789697,
      "grad_norm": 1.8753572702407837,
      "learning_rate": 0.00019694083387398673,
      "loss": 4.4817,
      "step": 16410
    },
    {
      "epoch": 0.015306096885868788,
      "grad_norm": 1.2449177503585815,
      "learning_rate": 0.00019693896955189156,
      "loss": 4.4642,
      "step": 16420
    },
    {
      "epoch": 0.01531541850394788,
      "grad_norm": 1.1027086973190308,
      "learning_rate": 0.00019693710522979637,
      "loss": 4.631,
      "step": 16430
    },
    {
      "epoch": 0.015324740122026972,
      "grad_norm": 0.9869092106819153,
      "learning_rate": 0.00019693524090770115,
      "loss": 4.4645,
      "step": 16440
    },
    {
      "epoch": 0.015334061740106065,
      "grad_norm": 1.2935222387313843,
      "learning_rate": 0.00019693337658560595,
      "loss": 4.3961,
      "step": 16450
    },
    {
      "epoch": 0.015343383358185156,
      "grad_norm": 1.2301591634750366,
      "learning_rate": 0.00019693151226351073,
      "loss": 4.6121,
      "step": 16460
    },
    {
      "epoch": 0.015352704976264247,
      "grad_norm": 1.1695058345794678,
      "learning_rate": 0.00019692964794141554,
      "loss": 4.6858,
      "step": 16470
    },
    {
      "epoch": 0.01536202659434334,
      "grad_norm": 1.2092595100402832,
      "learning_rate": 0.00019692778361932037,
      "loss": 4.498,
      "step": 16480
    },
    {
      "epoch": 0.01537134821242243,
      "grad_norm": 1.3985393047332764,
      "learning_rate": 0.00019692591929722515,
      "loss": 4.4079,
      "step": 16490
    },
    {
      "epoch": 0.015380669830501523,
      "grad_norm": 1.0089678764343262,
      "learning_rate": 0.00019692405497512996,
      "loss": 4.4826,
      "step": 16500
    },
    {
      "epoch": 0.015389991448580614,
      "grad_norm": 1.2995035648345947,
      "learning_rate": 0.00019692219065303474,
      "loss": 4.4474,
      "step": 16510
    },
    {
      "epoch": 0.015399313066659707,
      "grad_norm": 1.0030956268310547,
      "learning_rate": 0.00019692032633093954,
      "loss": 4.631,
      "step": 16520
    },
    {
      "epoch": 0.015408634684738798,
      "grad_norm": 0.8949043154716492,
      "learning_rate": 0.00019691846200884435,
      "loss": 4.4678,
      "step": 16530
    },
    {
      "epoch": 0.01541795630281789,
      "grad_norm": 1.053755521774292,
      "learning_rate": 0.00019691659768674916,
      "loss": 4.4456,
      "step": 16540
    },
    {
      "epoch": 0.015427277920896982,
      "grad_norm": 1.2961902618408203,
      "learning_rate": 0.00019691473336465396,
      "loss": 4.6423,
      "step": 16550
    },
    {
      "epoch": 0.015436599538976074,
      "grad_norm": 1.2007018327713013,
      "learning_rate": 0.00019691286904255874,
      "loss": 4.4696,
      "step": 16560
    },
    {
      "epoch": 0.015445921157055165,
      "grad_norm": 1.2656114101409912,
      "learning_rate": 0.00019691100472046355,
      "loss": 4.4572,
      "step": 16570
    },
    {
      "epoch": 0.015455242775134258,
      "grad_norm": 1.1072591543197632,
      "learning_rate": 0.00019690914039836835,
      "loss": 4.6991,
      "step": 16580
    },
    {
      "epoch": 0.015464564393213349,
      "grad_norm": 1.0754696130752563,
      "learning_rate": 0.00019690727607627316,
      "loss": 4.4591,
      "step": 16590
    },
    {
      "epoch": 0.015473886011292442,
      "grad_norm": 1.0485687255859375,
      "learning_rate": 0.00019690541175417797,
      "loss": 4.1729,
      "step": 16600
    },
    {
      "epoch": 0.015483207629371533,
      "grad_norm": 1.2215561866760254,
      "learning_rate": 0.00019690354743208275,
      "loss": 4.674,
      "step": 16610
    },
    {
      "epoch": 0.015492529247450626,
      "grad_norm": 1.1532467603683472,
      "learning_rate": 0.00019690168310998755,
      "loss": 4.2861,
      "step": 16620
    },
    {
      "epoch": 0.015501850865529717,
      "grad_norm": 1.1850162744522095,
      "learning_rate": 0.00019689981878789236,
      "loss": 4.5646,
      "step": 16630
    },
    {
      "epoch": 0.015511172483608808,
      "grad_norm": 1.27413809299469,
      "learning_rate": 0.00019689795446579717,
      "loss": 4.4975,
      "step": 16640
    },
    {
      "epoch": 0.0155204941016879,
      "grad_norm": 1.2456251382827759,
      "learning_rate": 0.00019689609014370197,
      "loss": 4.2754,
      "step": 16650
    },
    {
      "epoch": 0.015529815719766991,
      "grad_norm": 1.087160587310791,
      "learning_rate": 0.00019689422582160675,
      "loss": 4.456,
      "step": 16660
    },
    {
      "epoch": 0.015539137337846084,
      "grad_norm": 1.0803251266479492,
      "learning_rate": 0.00019689236149951156,
      "loss": 4.6723,
      "step": 16670
    },
    {
      "epoch": 0.015548458955925175,
      "grad_norm": 1.2416343688964844,
      "learning_rate": 0.00019689049717741636,
      "loss": 4.5341,
      "step": 16680
    },
    {
      "epoch": 0.015557780574004268,
      "grad_norm": 1.1236518621444702,
      "learning_rate": 0.00019688863285532114,
      "loss": 4.5318,
      "step": 16690
    },
    {
      "epoch": 0.015567102192083359,
      "grad_norm": 1.2541742324829102,
      "learning_rate": 0.00019688676853322598,
      "loss": 4.5044,
      "step": 16700
    },
    {
      "epoch": 0.015576423810162452,
      "grad_norm": 1.0129297971725464,
      "learning_rate": 0.00019688490421113076,
      "loss": 4.6277,
      "step": 16710
    },
    {
      "epoch": 0.015585745428241543,
      "grad_norm": 1.0853214263916016,
      "learning_rate": 0.00019688303988903556,
      "loss": 4.5864,
      "step": 16720
    },
    {
      "epoch": 0.015595067046320635,
      "grad_norm": 1.1366119384765625,
      "learning_rate": 0.00019688117556694037,
      "loss": 4.4921,
      "step": 16730
    },
    {
      "epoch": 0.015604388664399726,
      "grad_norm": 1.235150933265686,
      "learning_rate": 0.00019687931124484515,
      "loss": 4.5471,
      "step": 16740
    },
    {
      "epoch": 0.01561371028247882,
      "grad_norm": 1.2786586284637451,
      "learning_rate": 0.00019687744692274995,
      "loss": 4.3611,
      "step": 16750
    },
    {
      "epoch": 0.01562303190055791,
      "grad_norm": 1.2227458953857422,
      "learning_rate": 0.00019687558260065476,
      "loss": 4.5172,
      "step": 16760
    },
    {
      "epoch": 0.015632353518637003,
      "grad_norm": 1.2546420097351074,
      "learning_rate": 0.00019687371827855957,
      "loss": 4.5621,
      "step": 16770
    },
    {
      "epoch": 0.015641675136716094,
      "grad_norm": 1.1289092302322388,
      "learning_rate": 0.00019687185395646437,
      "loss": 4.3961,
      "step": 16780
    },
    {
      "epoch": 0.015650996754795185,
      "grad_norm": 1.1408478021621704,
      "learning_rate": 0.00019686998963436915,
      "loss": 4.4333,
      "step": 16790
    },
    {
      "epoch": 0.01566031837287428,
      "grad_norm": 1.1019580364227295,
      "learning_rate": 0.00019686812531227396,
      "loss": 4.5905,
      "step": 16800
    },
    {
      "epoch": 0.01566963999095337,
      "grad_norm": 1.121171236038208,
      "learning_rate": 0.00019686626099017876,
      "loss": 4.588,
      "step": 16810
    },
    {
      "epoch": 0.01567896160903246,
      "grad_norm": 1.2111437320709229,
      "learning_rate": 0.00019686439666808357,
      "loss": 4.5766,
      "step": 16820
    },
    {
      "epoch": 0.015688283227111553,
      "grad_norm": 1.1445399522781372,
      "learning_rate": 0.00019686253234598838,
      "loss": 4.3911,
      "step": 16830
    },
    {
      "epoch": 0.015697604845190644,
      "grad_norm": 1.0726237297058105,
      "learning_rate": 0.00019686066802389316,
      "loss": 4.3842,
      "step": 16840
    },
    {
      "epoch": 0.015706926463269738,
      "grad_norm": 1.0937092304229736,
      "learning_rate": 0.00019685880370179796,
      "loss": 4.5939,
      "step": 16850
    },
    {
      "epoch": 0.01571624808134883,
      "grad_norm": 1.1100285053253174,
      "learning_rate": 0.00019685693937970274,
      "loss": 4.4305,
      "step": 16860
    },
    {
      "epoch": 0.01572556969942792,
      "grad_norm": 1.1638487577438354,
      "learning_rate": 0.00019685507505760757,
      "loss": 4.4185,
      "step": 16870
    },
    {
      "epoch": 0.01573489131750701,
      "grad_norm": 1.33119535446167,
      "learning_rate": 0.00019685321073551238,
      "loss": 4.2573,
      "step": 16880
    },
    {
      "epoch": 0.015744212935586106,
      "grad_norm": 1.398415207862854,
      "learning_rate": 0.00019685134641341716,
      "loss": 4.5618,
      "step": 16890
    },
    {
      "epoch": 0.015753534553665197,
      "grad_norm": 0.9850145578384399,
      "learning_rate": 0.00019684948209132197,
      "loss": 4.3706,
      "step": 16900
    },
    {
      "epoch": 0.015762856171744288,
      "grad_norm": 1.262485384941101,
      "learning_rate": 0.00019684761776922675,
      "loss": 4.7585,
      "step": 16910
    },
    {
      "epoch": 0.01577217778982338,
      "grad_norm": 0.9751818776130676,
      "learning_rate": 0.00019684575344713155,
      "loss": 4.379,
      "step": 16920
    },
    {
      "epoch": 0.015781499407902473,
      "grad_norm": 1.1233235597610474,
      "learning_rate": 0.00019684388912503639,
      "loss": 4.5658,
      "step": 16930
    },
    {
      "epoch": 0.015790821025981564,
      "grad_norm": 1.27347731590271,
      "learning_rate": 0.00019684202480294116,
      "loss": 4.5446,
      "step": 16940
    },
    {
      "epoch": 0.015800142644060655,
      "grad_norm": 1.1333507299423218,
      "learning_rate": 0.00019684016048084597,
      "loss": 4.4284,
      "step": 16950
    },
    {
      "epoch": 0.015809464262139746,
      "grad_norm": 1.3886981010437012,
      "learning_rate": 0.00019683829615875075,
      "loss": 4.4176,
      "step": 16960
    },
    {
      "epoch": 0.01581878588021884,
      "grad_norm": 1.116692066192627,
      "learning_rate": 0.00019683643183665556,
      "loss": 4.5047,
      "step": 16970
    },
    {
      "epoch": 0.01582810749829793,
      "grad_norm": 1.0619912147521973,
      "learning_rate": 0.00019683456751456036,
      "loss": 4.4617,
      "step": 16980
    },
    {
      "epoch": 0.015837429116377023,
      "grad_norm": 1.1627715826034546,
      "learning_rate": 0.00019683270319246517,
      "loss": 4.3765,
      "step": 16990
    },
    {
      "epoch": 0.015846750734456114,
      "grad_norm": 1.113926649093628,
      "learning_rate": 0.00019683083887036998,
      "loss": 4.6113,
      "step": 17000
    },
    {
      "epoch": 0.015856072352535205,
      "grad_norm": 1.093421459197998,
      "learning_rate": 0.00019682897454827475,
      "loss": 4.6536,
      "step": 17010
    },
    {
      "epoch": 0.0158653939706143,
      "grad_norm": 1.2781003713607788,
      "learning_rate": 0.00019682711022617956,
      "loss": 4.6614,
      "step": 17020
    },
    {
      "epoch": 0.01587471558869339,
      "grad_norm": 1.2008435726165771,
      "learning_rate": 0.00019682524590408437,
      "loss": 4.4494,
      "step": 17030
    },
    {
      "epoch": 0.01588403720677248,
      "grad_norm": 1.0976715087890625,
      "learning_rate": 0.00019682338158198917,
      "loss": 4.5129,
      "step": 17040
    },
    {
      "epoch": 0.015893358824851572,
      "grad_norm": 1.150437593460083,
      "learning_rate": 0.00019682151725989398,
      "loss": 4.3266,
      "step": 17050
    },
    {
      "epoch": 0.015902680442930667,
      "grad_norm": 1.048435091972351,
      "learning_rate": 0.00019681965293779876,
      "loss": 4.5101,
      "step": 17060
    },
    {
      "epoch": 0.015912002061009758,
      "grad_norm": 1.4805630445480347,
      "learning_rate": 0.00019681778861570357,
      "loss": 4.1838,
      "step": 17070
    },
    {
      "epoch": 0.01592132367908885,
      "grad_norm": 1.0664284229278564,
      "learning_rate": 0.00019681592429360837,
      "loss": 4.4361,
      "step": 17080
    },
    {
      "epoch": 0.01593064529716794,
      "grad_norm": 1.0228906869888306,
      "learning_rate": 0.00019681405997151318,
      "loss": 4.6335,
      "step": 17090
    },
    {
      "epoch": 0.015939966915247034,
      "grad_norm": 1.2894953489303589,
      "learning_rate": 0.00019681219564941798,
      "loss": 4.3972,
      "step": 17100
    },
    {
      "epoch": 0.015949288533326125,
      "grad_norm": 1.3029876947402954,
      "learning_rate": 0.00019681033132732276,
      "loss": 4.4576,
      "step": 17110
    },
    {
      "epoch": 0.015958610151405216,
      "grad_norm": 1.1977765560150146,
      "learning_rate": 0.00019680846700522757,
      "loss": 4.359,
      "step": 17120
    },
    {
      "epoch": 0.015967931769484307,
      "grad_norm": 1.2750426530838013,
      "learning_rate": 0.00019680660268313238,
      "loss": 4.5111,
      "step": 17130
    },
    {
      "epoch": 0.015977253387563402,
      "grad_norm": 1.0768829584121704,
      "learning_rate": 0.00019680473836103716,
      "loss": 4.424,
      "step": 17140
    },
    {
      "epoch": 0.015986575005642493,
      "grad_norm": 0.9155883193016052,
      "learning_rate": 0.000196802874038942,
      "loss": 4.4408,
      "step": 17150
    },
    {
      "epoch": 0.015995896623721584,
      "grad_norm": 1.3279091119766235,
      "learning_rate": 0.00019680100971684677,
      "loss": 4.3807,
      "step": 17160
    },
    {
      "epoch": 0.016005218241800675,
      "grad_norm": 1.0500863790512085,
      "learning_rate": 0.00019679914539475157,
      "loss": 4.489,
      "step": 17170
    },
    {
      "epoch": 0.016014539859879766,
      "grad_norm": 1.182771921157837,
      "learning_rate": 0.00019679728107265638,
      "loss": 4.5511,
      "step": 17180
    },
    {
      "epoch": 0.01602386147795886,
      "grad_norm": 1.1223487854003906,
      "learning_rate": 0.00019679541675056116,
      "loss": 4.4885,
      "step": 17190
    },
    {
      "epoch": 0.01603318309603795,
      "grad_norm": 0.9083897471427917,
      "learning_rate": 0.00019679355242846597,
      "loss": 4.5816,
      "step": 17200
    },
    {
      "epoch": 0.016042504714117042,
      "grad_norm": 1.1776821613311768,
      "learning_rate": 0.00019679168810637077,
      "loss": 4.5712,
      "step": 17210
    },
    {
      "epoch": 0.016051826332196133,
      "grad_norm": 1.0692150592803955,
      "learning_rate": 0.00019678982378427558,
      "loss": 4.6217,
      "step": 17220
    },
    {
      "epoch": 0.016061147950275228,
      "grad_norm": 1.2174395322799683,
      "learning_rate": 0.00019678795946218039,
      "loss": 4.3964,
      "step": 17230
    },
    {
      "epoch": 0.01607046956835432,
      "grad_norm": 1.0161887407302856,
      "learning_rate": 0.00019678609514008516,
      "loss": 4.4811,
      "step": 17240
    },
    {
      "epoch": 0.01607979118643341,
      "grad_norm": 1.4252760410308838,
      "learning_rate": 0.00019678423081798997,
      "loss": 4.3468,
      "step": 17250
    },
    {
      "epoch": 0.0160891128045125,
      "grad_norm": 1.1392055749893188,
      "learning_rate": 0.00019678236649589478,
      "loss": 4.3114,
      "step": 17260
    },
    {
      "epoch": 0.016098434422591595,
      "grad_norm": 1.0936288833618164,
      "learning_rate": 0.00019678050217379958,
      "loss": 4.5993,
      "step": 17270
    },
    {
      "epoch": 0.016107756040670686,
      "grad_norm": 0.9629073143005371,
      "learning_rate": 0.00019677863785170436,
      "loss": 4.523,
      "step": 17280
    },
    {
      "epoch": 0.016117077658749777,
      "grad_norm": 1.2280113697052002,
      "learning_rate": 0.00019677677352960917,
      "loss": 4.4169,
      "step": 17290
    },
    {
      "epoch": 0.01612639927682887,
      "grad_norm": 1.4447882175445557,
      "learning_rate": 0.00019677490920751397,
      "loss": 4.4569,
      "step": 17300
    },
    {
      "epoch": 0.016135720894907963,
      "grad_norm": 1.2569477558135986,
      "learning_rate": 0.00019677304488541875,
      "loss": 4.4901,
      "step": 17310
    },
    {
      "epoch": 0.016145042512987054,
      "grad_norm": 1.0999788045883179,
      "learning_rate": 0.0001967711805633236,
      "loss": 4.5784,
      "step": 17320
    },
    {
      "epoch": 0.016154364131066145,
      "grad_norm": 1.029313087463379,
      "learning_rate": 0.00019676931624122837,
      "loss": 4.3911,
      "step": 17330
    },
    {
      "epoch": 0.016163685749145236,
      "grad_norm": 1.0078456401824951,
      "learning_rate": 0.00019676745191913317,
      "loss": 4.5263,
      "step": 17340
    },
    {
      "epoch": 0.016173007367224327,
      "grad_norm": 1.1776665449142456,
      "learning_rate": 0.00019676558759703798,
      "loss": 4.3831,
      "step": 17350
    },
    {
      "epoch": 0.01618232898530342,
      "grad_norm": 1.1414107084274292,
      "learning_rate": 0.00019676372327494276,
      "loss": 4.5422,
      "step": 17360
    },
    {
      "epoch": 0.016191650603382512,
      "grad_norm": 1.196068286895752,
      "learning_rate": 0.00019676185895284756,
      "loss": 4.4734,
      "step": 17370
    },
    {
      "epoch": 0.016200972221461604,
      "grad_norm": 1.077409029006958,
      "learning_rate": 0.00019675999463075237,
      "loss": 4.597,
      "step": 17380
    },
    {
      "epoch": 0.016210293839540695,
      "grad_norm": 1.0068482160568237,
      "learning_rate": 0.00019675813030865718,
      "loss": 4.6897,
      "step": 17390
    },
    {
      "epoch": 0.01621961545761979,
      "grad_norm": 1.2509359121322632,
      "learning_rate": 0.00019675626598656198,
      "loss": 4.6688,
      "step": 17400
    },
    {
      "epoch": 0.01622893707569888,
      "grad_norm": 1.1982245445251465,
      "learning_rate": 0.00019675440166446676,
      "loss": 4.5015,
      "step": 17410
    },
    {
      "epoch": 0.01623825869377797,
      "grad_norm": 1.3006165027618408,
      "learning_rate": 0.00019675253734237157,
      "loss": 4.5816,
      "step": 17420
    },
    {
      "epoch": 0.016247580311857062,
      "grad_norm": 1.132145881652832,
      "learning_rate": 0.00019675067302027638,
      "loss": 4.5565,
      "step": 17430
    },
    {
      "epoch": 0.016256901929936157,
      "grad_norm": 1.1508245468139648,
      "learning_rate": 0.00019674880869818118,
      "loss": 4.4846,
      "step": 17440
    },
    {
      "epoch": 0.016266223548015248,
      "grad_norm": 1.2623023986816406,
      "learning_rate": 0.000196746944376086,
      "loss": 4.4505,
      "step": 17450
    },
    {
      "epoch": 0.01627554516609434,
      "grad_norm": 1.109551191329956,
      "learning_rate": 0.00019674508005399077,
      "loss": 4.4291,
      "step": 17460
    },
    {
      "epoch": 0.01628486678417343,
      "grad_norm": 1.1972815990447998,
      "learning_rate": 0.00019674321573189557,
      "loss": 4.4574,
      "step": 17470
    },
    {
      "epoch": 0.016294188402252524,
      "grad_norm": 1.097680687904358,
      "learning_rate": 0.00019674135140980038,
      "loss": 4.4958,
      "step": 17480
    },
    {
      "epoch": 0.016303510020331615,
      "grad_norm": 1.3224825859069824,
      "learning_rate": 0.00019673948708770519,
      "loss": 4.4566,
      "step": 17490
    },
    {
      "epoch": 0.016312831638410706,
      "grad_norm": 1.0231324434280396,
      "learning_rate": 0.00019673762276561,
      "loss": 4.5468,
      "step": 17500
    },
    {
      "epoch": 0.016322153256489797,
      "grad_norm": 1.2239038944244385,
      "learning_rate": 0.00019673575844351477,
      "loss": 4.6479,
      "step": 17510
    },
    {
      "epoch": 0.016331474874568888,
      "grad_norm": 1.149319052696228,
      "learning_rate": 0.00019673389412141958,
      "loss": 4.5494,
      "step": 17520
    },
    {
      "epoch": 0.016340796492647983,
      "grad_norm": 1.1128418445587158,
      "learning_rate": 0.00019673202979932436,
      "loss": 4.6072,
      "step": 17530
    },
    {
      "epoch": 0.016350118110727074,
      "grad_norm": 1.2838488817214966,
      "learning_rate": 0.0001967301654772292,
      "loss": 4.4451,
      "step": 17540
    },
    {
      "epoch": 0.016359439728806165,
      "grad_norm": 1.0720415115356445,
      "learning_rate": 0.000196728301155134,
      "loss": 4.2765,
      "step": 17550
    },
    {
      "epoch": 0.016368761346885256,
      "grad_norm": 1.0304310321807861,
      "learning_rate": 0.00019672643683303878,
      "loss": 4.4529,
      "step": 17560
    },
    {
      "epoch": 0.01637808296496435,
      "grad_norm": 1.1360918283462524,
      "learning_rate": 0.00019672457251094358,
      "loss": 4.3764,
      "step": 17570
    },
    {
      "epoch": 0.01638740458304344,
      "grad_norm": 1.0397121906280518,
      "learning_rate": 0.00019672270818884836,
      "loss": 4.7009,
      "step": 17580
    },
    {
      "epoch": 0.016396726201122532,
      "grad_norm": 1.086551547050476,
      "learning_rate": 0.00019672084386675317,
      "loss": 4.6112,
      "step": 17590
    },
    {
      "epoch": 0.016406047819201623,
      "grad_norm": 1.4094029664993286,
      "learning_rate": 0.000196718979544658,
      "loss": 4.527,
      "step": 17600
    },
    {
      "epoch": 0.016415369437280718,
      "grad_norm": 1.1831270456314087,
      "learning_rate": 0.00019671711522256278,
      "loss": 4.4886,
      "step": 17610
    },
    {
      "epoch": 0.01642469105535981,
      "grad_norm": 1.1437351703643799,
      "learning_rate": 0.0001967152509004676,
      "loss": 4.3905,
      "step": 17620
    },
    {
      "epoch": 0.0164340126734389,
      "grad_norm": 1.1590259075164795,
      "learning_rate": 0.00019671338657837237,
      "loss": 4.4686,
      "step": 17630
    },
    {
      "epoch": 0.01644333429151799,
      "grad_norm": 1.2035565376281738,
      "learning_rate": 0.00019671152225627717,
      "loss": 4.3815,
      "step": 17640
    },
    {
      "epoch": 0.016452655909597085,
      "grad_norm": 1.1693356037139893,
      "learning_rate": 0.00019670965793418198,
      "loss": 4.6372,
      "step": 17650
    },
    {
      "epoch": 0.016461977527676176,
      "grad_norm": 1.0783804655075073,
      "learning_rate": 0.00019670779361208679,
      "loss": 4.5538,
      "step": 17660
    },
    {
      "epoch": 0.016471299145755267,
      "grad_norm": 1.3145687580108643,
      "learning_rate": 0.0001967059292899916,
      "loss": 4.5529,
      "step": 17670
    },
    {
      "epoch": 0.016480620763834358,
      "grad_norm": 1.1349118947982788,
      "learning_rate": 0.00019670406496789637,
      "loss": 4.4422,
      "step": 17680
    },
    {
      "epoch": 0.01648994238191345,
      "grad_norm": 1.1428669691085815,
      "learning_rate": 0.00019670220064580118,
      "loss": 4.4615,
      "step": 17690
    },
    {
      "epoch": 0.016499263999992544,
      "grad_norm": 1.1453847885131836,
      "learning_rate": 0.00019670033632370598,
      "loss": 4.4372,
      "step": 17700
    },
    {
      "epoch": 0.016508585618071635,
      "grad_norm": 0.9910815954208374,
      "learning_rate": 0.0001966984720016108,
      "loss": 4.5543,
      "step": 17710
    },
    {
      "epoch": 0.016517907236150726,
      "grad_norm": 1.0318204164505005,
      "learning_rate": 0.0001966966076795156,
      "loss": 4.5012,
      "step": 17720
    },
    {
      "epoch": 0.016527228854229817,
      "grad_norm": 1.259966492652893,
      "learning_rate": 0.00019669474335742038,
      "loss": 4.4291,
      "step": 17730
    },
    {
      "epoch": 0.01653655047230891,
      "grad_norm": 1.1732487678527832,
      "learning_rate": 0.00019669287903532518,
      "loss": 4.5087,
      "step": 17740
    },
    {
      "epoch": 0.016545872090388002,
      "grad_norm": 0.9462860822677612,
      "learning_rate": 0.00019669101471323,
      "loss": 4.6641,
      "step": 17750
    },
    {
      "epoch": 0.016555193708467093,
      "grad_norm": 1.337151050567627,
      "learning_rate": 0.00019668915039113477,
      "loss": 4.4756,
      "step": 17760
    },
    {
      "epoch": 0.016564515326546184,
      "grad_norm": 1.0590085983276367,
      "learning_rate": 0.0001966872860690396,
      "loss": 4.3573,
      "step": 17770
    },
    {
      "epoch": 0.01657383694462528,
      "grad_norm": 1.4390627145767212,
      "learning_rate": 0.00019668542174694438,
      "loss": 4.4724,
      "step": 17780
    },
    {
      "epoch": 0.01658315856270437,
      "grad_norm": 1.1670478582382202,
      "learning_rate": 0.00019668355742484919,
      "loss": 4.3722,
      "step": 17790
    },
    {
      "epoch": 0.01659248018078346,
      "grad_norm": 1.1734064817428589,
      "learning_rate": 0.000196681693102754,
      "loss": 4.4124,
      "step": 17800
    },
    {
      "epoch": 0.016601801798862552,
      "grad_norm": 1.0035632848739624,
      "learning_rate": 0.00019667982878065877,
      "loss": 4.3401,
      "step": 17810
    },
    {
      "epoch": 0.016611123416941646,
      "grad_norm": 1.2000086307525635,
      "learning_rate": 0.00019667796445856358,
      "loss": 4.49,
      "step": 17820
    },
    {
      "epoch": 0.016620445035020737,
      "grad_norm": 1.285494089126587,
      "learning_rate": 0.00019667610013646838,
      "loss": 4.5549,
      "step": 17830
    },
    {
      "epoch": 0.01662976665309983,
      "grad_norm": 1.246864914894104,
      "learning_rate": 0.0001966742358143732,
      "loss": 4.3573,
      "step": 17840
    },
    {
      "epoch": 0.01663908827117892,
      "grad_norm": 1.1562223434448242,
      "learning_rate": 0.000196672371492278,
      "loss": 4.3576,
      "step": 17850
    },
    {
      "epoch": 0.01664840988925801,
      "grad_norm": 1.427288293838501,
      "learning_rate": 0.00019667050717018278,
      "loss": 4.3185,
      "step": 17860
    },
    {
      "epoch": 0.016657731507337105,
      "grad_norm": 1.3006247282028198,
      "learning_rate": 0.00019666864284808758,
      "loss": 4.4326,
      "step": 17870
    },
    {
      "epoch": 0.016667053125416196,
      "grad_norm": 1.1824243068695068,
      "learning_rate": 0.0001966667785259924,
      "loss": 4.646,
      "step": 17880
    },
    {
      "epoch": 0.016676374743495287,
      "grad_norm": 0.9518525004386902,
      "learning_rate": 0.0001966649142038972,
      "loss": 4.5348,
      "step": 17890
    },
    {
      "epoch": 0.016685696361574378,
      "grad_norm": 1.3271188735961914,
      "learning_rate": 0.000196663049881802,
      "loss": 4.3825,
      "step": 17900
    },
    {
      "epoch": 0.016695017979653472,
      "grad_norm": 1.2353016138076782,
      "learning_rate": 0.00019666118555970678,
      "loss": 4.7403,
      "step": 17910
    },
    {
      "epoch": 0.016704339597732563,
      "grad_norm": 1.1963491439819336,
      "learning_rate": 0.0001966593212376116,
      "loss": 4.665,
      "step": 17920
    },
    {
      "epoch": 0.016713661215811654,
      "grad_norm": 1.3270742893218994,
      "learning_rate": 0.0001966574569155164,
      "loss": 4.4051,
      "step": 17930
    },
    {
      "epoch": 0.016722982833890745,
      "grad_norm": 1.0312671661376953,
      "learning_rate": 0.0001966555925934212,
      "loss": 4.4435,
      "step": 17940
    },
    {
      "epoch": 0.01673230445196984,
      "grad_norm": 1.12940514087677,
      "learning_rate": 0.000196653728271326,
      "loss": 4.4732,
      "step": 17950
    },
    {
      "epoch": 0.01674162607004893,
      "grad_norm": 1.1567469835281372,
      "learning_rate": 0.00019665186394923078,
      "loss": 4.5952,
      "step": 17960
    },
    {
      "epoch": 0.016750947688128022,
      "grad_norm": 1.3211356401443481,
      "learning_rate": 0.0001966499996271356,
      "loss": 4.5055,
      "step": 17970
    },
    {
      "epoch": 0.016760269306207113,
      "grad_norm": 1.2657487392425537,
      "learning_rate": 0.00019664813530504037,
      "loss": 4.5417,
      "step": 17980
    },
    {
      "epoch": 0.016769590924286207,
      "grad_norm": 0.9614537954330444,
      "learning_rate": 0.0001966462709829452,
      "loss": 4.4372,
      "step": 17990
    },
    {
      "epoch": 0.0167789125423653,
      "grad_norm": 1.1549428701400757,
      "learning_rate": 0.00019664440666085,
      "loss": 4.6384,
      "step": 18000
    },
    {
      "epoch": 0.01678823416044439,
      "grad_norm": 1.3695018291473389,
      "learning_rate": 0.0001966425423387548,
      "loss": 4.2678,
      "step": 18010
    },
    {
      "epoch": 0.01679755577852348,
      "grad_norm": 1.3776980638504028,
      "learning_rate": 0.0001966406780166596,
      "loss": 4.5348,
      "step": 18020
    },
    {
      "epoch": 0.01680687739660257,
      "grad_norm": 1.1857786178588867,
      "learning_rate": 0.00019663881369456437,
      "loss": 4.3926,
      "step": 18030
    },
    {
      "epoch": 0.016816199014681666,
      "grad_norm": 1.108077049255371,
      "learning_rate": 0.00019663694937246918,
      "loss": 4.3341,
      "step": 18040
    },
    {
      "epoch": 0.016825520632760757,
      "grad_norm": 1.3041938543319702,
      "learning_rate": 0.00019663508505037401,
      "loss": 4.6475,
      "step": 18050
    },
    {
      "epoch": 0.016834842250839848,
      "grad_norm": 1.1820635795593262,
      "learning_rate": 0.0001966332207282788,
      "loss": 4.4271,
      "step": 18060
    },
    {
      "epoch": 0.01684416386891894,
      "grad_norm": 1.3178205490112305,
      "learning_rate": 0.0001966313564061836,
      "loss": 4.2831,
      "step": 18070
    },
    {
      "epoch": 0.016853485486998034,
      "grad_norm": 1.1315423250198364,
      "learning_rate": 0.00019662949208408838,
      "loss": 4.5067,
      "step": 18080
    },
    {
      "epoch": 0.016862807105077125,
      "grad_norm": 1.3123506307601929,
      "learning_rate": 0.00019662762776199319,
      "loss": 4.5111,
      "step": 18090
    },
    {
      "epoch": 0.016872128723156216,
      "grad_norm": 1.182418704032898,
      "learning_rate": 0.000196625763439898,
      "loss": 4.3307,
      "step": 18100
    },
    {
      "epoch": 0.016881450341235307,
      "grad_norm": 1.1887218952178955,
      "learning_rate": 0.0001966238991178028,
      "loss": 4.4024,
      "step": 18110
    },
    {
      "epoch": 0.0168907719593144,
      "grad_norm": 1.0443063974380493,
      "learning_rate": 0.0001966220347957076,
      "loss": 4.5582,
      "step": 18120
    },
    {
      "epoch": 0.016900093577393492,
      "grad_norm": 1.0631448030471802,
      "learning_rate": 0.00019662017047361238,
      "loss": 4.3578,
      "step": 18130
    },
    {
      "epoch": 0.016909415195472583,
      "grad_norm": 1.1584179401397705,
      "learning_rate": 0.0001966183061515172,
      "loss": 4.3276,
      "step": 18140
    },
    {
      "epoch": 0.016918736813551674,
      "grad_norm": 1.3780972957611084,
      "learning_rate": 0.000196616441829422,
      "loss": 4.6318,
      "step": 18150
    },
    {
      "epoch": 0.01692805843163077,
      "grad_norm": 1.2469985485076904,
      "learning_rate": 0.0001966145775073268,
      "loss": 4.4375,
      "step": 18160
    },
    {
      "epoch": 0.01693738004970986,
      "grad_norm": 1.2021986246109009,
      "learning_rate": 0.0001966127131852316,
      "loss": 4.2922,
      "step": 18170
    },
    {
      "epoch": 0.01694670166778895,
      "grad_norm": 1.272376298904419,
      "learning_rate": 0.0001966108488631364,
      "loss": 4.2408,
      "step": 18180
    },
    {
      "epoch": 0.01695602328586804,
      "grad_norm": 1.219962477684021,
      "learning_rate": 0.0001966089845410412,
      "loss": 4.4028,
      "step": 18190
    },
    {
      "epoch": 0.016965344903947133,
      "grad_norm": 1.0649285316467285,
      "learning_rate": 0.000196607120218946,
      "loss": 4.2616,
      "step": 18200
    },
    {
      "epoch": 0.016974666522026227,
      "grad_norm": 1.2707583904266357,
      "learning_rate": 0.00019660525589685078,
      "loss": 4.3669,
      "step": 18210
    },
    {
      "epoch": 0.016983988140105318,
      "grad_norm": 1.0662801265716553,
      "learning_rate": 0.0001966033915747556,
      "loss": 4.6813,
      "step": 18220
    },
    {
      "epoch": 0.01699330975818441,
      "grad_norm": 1.1669766902923584,
      "learning_rate": 0.0001966015272526604,
      "loss": 4.4812,
      "step": 18230
    },
    {
      "epoch": 0.0170026313762635,
      "grad_norm": 1.3150641918182373,
      "learning_rate": 0.0001965996629305652,
      "loss": 4.4164,
      "step": 18240
    },
    {
      "epoch": 0.017011952994342595,
      "grad_norm": 1.1140202283859253,
      "learning_rate": 0.00019659779860847,
      "loss": 4.3731,
      "step": 18250
    },
    {
      "epoch": 0.017021274612421686,
      "grad_norm": 1.0560606718063354,
      "learning_rate": 0.00019659593428637478,
      "loss": 4.5105,
      "step": 18260
    },
    {
      "epoch": 0.017030596230500777,
      "grad_norm": 1.1560946702957153,
      "learning_rate": 0.0001965940699642796,
      "loss": 4.4196,
      "step": 18270
    },
    {
      "epoch": 0.017039917848579868,
      "grad_norm": 0.8858310580253601,
      "learning_rate": 0.0001965922056421844,
      "loss": 4.4313,
      "step": 18280
    },
    {
      "epoch": 0.017049239466658962,
      "grad_norm": 1.2567862272262573,
      "learning_rate": 0.0001965903413200892,
      "loss": 4.3824,
      "step": 18290
    },
    {
      "epoch": 0.017058561084738053,
      "grad_norm": 1.2343858480453491,
      "learning_rate": 0.000196588476997994,
      "loss": 4.3117,
      "step": 18300
    },
    {
      "epoch": 0.017067882702817144,
      "grad_norm": 1.1311086416244507,
      "learning_rate": 0.0001965866126758988,
      "loss": 4.4706,
      "step": 18310
    },
    {
      "epoch": 0.017077204320896235,
      "grad_norm": 1.2546231746673584,
      "learning_rate": 0.0001965847483538036,
      "loss": 4.4725,
      "step": 18320
    },
    {
      "epoch": 0.01708652593897533,
      "grad_norm": 1.1378636360168457,
      "learning_rate": 0.0001965828840317084,
      "loss": 4.3941,
      "step": 18330
    },
    {
      "epoch": 0.01709584755705442,
      "grad_norm": 0.9582810401916504,
      "learning_rate": 0.0001965810197096132,
      "loss": 4.474,
      "step": 18340
    },
    {
      "epoch": 0.017105169175133512,
      "grad_norm": 1.059482455253601,
      "learning_rate": 0.00019657915538751801,
      "loss": 4.1596,
      "step": 18350
    },
    {
      "epoch": 0.017114490793212603,
      "grad_norm": 1.3457335233688354,
      "learning_rate": 0.0001965772910654228,
      "loss": 4.3252,
      "step": 18360
    },
    {
      "epoch": 0.017123812411291694,
      "grad_norm": 1.1649283170700073,
      "learning_rate": 0.0001965754267433276,
      "loss": 4.5969,
      "step": 18370
    },
    {
      "epoch": 0.01713313402937079,
      "grad_norm": 1.2191810607910156,
      "learning_rate": 0.0001965735624212324,
      "loss": 4.3903,
      "step": 18380
    },
    {
      "epoch": 0.01714245564744988,
      "grad_norm": 1.2415419816970825,
      "learning_rate": 0.0001965716980991372,
      "loss": 4.3998,
      "step": 18390
    },
    {
      "epoch": 0.01715177726552897,
      "grad_norm": 1.2433178424835205,
      "learning_rate": 0.00019656983377704202,
      "loss": 4.4417,
      "step": 18400
    },
    {
      "epoch": 0.01716109888360806,
      "grad_norm": 1.1523895263671875,
      "learning_rate": 0.0001965679694549468,
      "loss": 4.3647,
      "step": 18410
    },
    {
      "epoch": 0.017170420501687156,
      "grad_norm": 1.2098249197006226,
      "learning_rate": 0.0001965661051328516,
      "loss": 4.6005,
      "step": 18420
    },
    {
      "epoch": 0.017179742119766247,
      "grad_norm": 1.2459036111831665,
      "learning_rate": 0.00019656424081075638,
      "loss": 4.5363,
      "step": 18430
    },
    {
      "epoch": 0.017189063737845338,
      "grad_norm": 1.2123453617095947,
      "learning_rate": 0.00019656237648866122,
      "loss": 4.2542,
      "step": 18440
    },
    {
      "epoch": 0.01719838535592443,
      "grad_norm": 1.0516149997711182,
      "learning_rate": 0.00019656051216656602,
      "loss": 4.3396,
      "step": 18450
    },
    {
      "epoch": 0.017207706974003523,
      "grad_norm": 1.7774982452392578,
      "learning_rate": 0.0001965586478444708,
      "loss": 4.4647,
      "step": 18460
    },
    {
      "epoch": 0.017217028592082614,
      "grad_norm": 1.0597845315933228,
      "learning_rate": 0.0001965567835223756,
      "loss": 4.5176,
      "step": 18470
    },
    {
      "epoch": 0.017226350210161705,
      "grad_norm": 0.9691914319992065,
      "learning_rate": 0.0001965549192002804,
      "loss": 4.3885,
      "step": 18480
    },
    {
      "epoch": 0.017235671828240796,
      "grad_norm": 1.2317180633544922,
      "learning_rate": 0.0001965530548781852,
      "loss": 4.4939,
      "step": 18490
    },
    {
      "epoch": 0.01724499344631989,
      "grad_norm": 1.1248630285263062,
      "learning_rate": 0.00019655119055609003,
      "loss": 4.5623,
      "step": 18500
    },
    {
      "epoch": 0.017254315064398982,
      "grad_norm": 1.2008506059646606,
      "learning_rate": 0.0001965493262339948,
      "loss": 4.3368,
      "step": 18510
    },
    {
      "epoch": 0.017263636682478073,
      "grad_norm": 1.1992918252944946,
      "learning_rate": 0.0001965474619118996,
      "loss": 4.4742,
      "step": 18520
    },
    {
      "epoch": 0.017272958300557164,
      "grad_norm": 1.2172168493270874,
      "learning_rate": 0.0001965455975898044,
      "loss": 4.4568,
      "step": 18530
    },
    {
      "epoch": 0.017282279918636255,
      "grad_norm": 1.3720842599868774,
      "learning_rate": 0.0001965437332677092,
      "loss": 4.3584,
      "step": 18540
    },
    {
      "epoch": 0.01729160153671535,
      "grad_norm": 1.122294306755066,
      "learning_rate": 0.000196541868945614,
      "loss": 4.3468,
      "step": 18550
    },
    {
      "epoch": 0.01730092315479444,
      "grad_norm": 1.0177843570709229,
      "learning_rate": 0.0001965400046235188,
      "loss": 4.5718,
      "step": 18560
    },
    {
      "epoch": 0.01731024477287353,
      "grad_norm": 0.9936982989311218,
      "learning_rate": 0.00019653814030142362,
      "loss": 4.4696,
      "step": 18570
    },
    {
      "epoch": 0.017319566390952622,
      "grad_norm": 1.1535321474075317,
      "learning_rate": 0.0001965362759793284,
      "loss": 4.4799,
      "step": 18580
    },
    {
      "epoch": 0.017328888009031717,
      "grad_norm": 1.2542303800582886,
      "learning_rate": 0.0001965344116572332,
      "loss": 4.4423,
      "step": 18590
    },
    {
      "epoch": 0.017338209627110808,
      "grad_norm": 1.1681137084960938,
      "learning_rate": 0.000196532547335138,
      "loss": 4.3147,
      "step": 18600
    },
    {
      "epoch": 0.0173475312451899,
      "grad_norm": 1.0400207042694092,
      "learning_rate": 0.00019653068301304282,
      "loss": 4.4746,
      "step": 18610
    },
    {
      "epoch": 0.01735685286326899,
      "grad_norm": 1.297447919845581,
      "learning_rate": 0.00019652881869094762,
      "loss": 4.5589,
      "step": 18620
    },
    {
      "epoch": 0.017366174481348085,
      "grad_norm": 1.1698585748672485,
      "learning_rate": 0.0001965269543688524,
      "loss": 4.5806,
      "step": 18630
    },
    {
      "epoch": 0.017375496099427176,
      "grad_norm": 1.181631326675415,
      "learning_rate": 0.0001965250900467572,
      "loss": 4.4727,
      "step": 18640
    },
    {
      "epoch": 0.017384817717506267,
      "grad_norm": 1.0805723667144775,
      "learning_rate": 0.00019652322572466199,
      "loss": 4.2629,
      "step": 18650
    },
    {
      "epoch": 0.017394139335585358,
      "grad_norm": 1.3600494861602783,
      "learning_rate": 0.0001965213614025668,
      "loss": 4.5161,
      "step": 18660
    },
    {
      "epoch": 0.017403460953664452,
      "grad_norm": 1.3973034620285034,
      "learning_rate": 0.00019651949708047163,
      "loss": 4.4137,
      "step": 18670
    },
    {
      "epoch": 0.017412782571743543,
      "grad_norm": 1.4527422189712524,
      "learning_rate": 0.0001965176327583764,
      "loss": 4.4797,
      "step": 18680
    },
    {
      "epoch": 0.017422104189822634,
      "grad_norm": 1.070884346961975,
      "learning_rate": 0.0001965157684362812,
      "loss": 4.3097,
      "step": 18690
    },
    {
      "epoch": 0.017431425807901725,
      "grad_norm": 1.0278024673461914,
      "learning_rate": 0.000196513904114186,
      "loss": 4.1289,
      "step": 18700
    },
    {
      "epoch": 0.017440747425980816,
      "grad_norm": 1.2470118999481201,
      "learning_rate": 0.0001965120397920908,
      "loss": 4.6297,
      "step": 18710
    },
    {
      "epoch": 0.01745006904405991,
      "grad_norm": 1.3182660341262817,
      "learning_rate": 0.0001965101754699956,
      "loss": 4.2339,
      "step": 18720
    },
    {
      "epoch": 0.017459390662139,
      "grad_norm": 1.0130542516708374,
      "learning_rate": 0.0001965083111479004,
      "loss": 4.3148,
      "step": 18730
    },
    {
      "epoch": 0.017468712280218093,
      "grad_norm": 1.248651146888733,
      "learning_rate": 0.00019650644682580522,
      "loss": 4.5153,
      "step": 18740
    },
    {
      "epoch": 0.017478033898297184,
      "grad_norm": 1.1268385648727417,
      "learning_rate": 0.00019650458250371,
      "loss": 4.4495,
      "step": 18750
    },
    {
      "epoch": 0.017487355516376278,
      "grad_norm": 1.1986981630325317,
      "learning_rate": 0.0001965027181816148,
      "loss": 4.4842,
      "step": 18760
    },
    {
      "epoch": 0.01749667713445537,
      "grad_norm": 1.1590492725372314,
      "learning_rate": 0.0001965008538595196,
      "loss": 4.3717,
      "step": 18770
    },
    {
      "epoch": 0.01750599875253446,
      "grad_norm": 1.1952542066574097,
      "learning_rate": 0.00019649898953742441,
      "loss": 4.4004,
      "step": 18780
    },
    {
      "epoch": 0.01751532037061355,
      "grad_norm": 1.1124719381332397,
      "learning_rate": 0.00019649712521532922,
      "loss": 4.426,
      "step": 18790
    },
    {
      "epoch": 0.017524641988692646,
      "grad_norm": 1.135837197303772,
      "learning_rate": 0.000196495260893234,
      "loss": 4.4393,
      "step": 18800
    },
    {
      "epoch": 0.017533963606771737,
      "grad_norm": 1.3728739023208618,
      "learning_rate": 0.0001964933965711388,
      "loss": 4.5395,
      "step": 18810
    },
    {
      "epoch": 0.017543285224850828,
      "grad_norm": 1.0159664154052734,
      "learning_rate": 0.0001964915322490436,
      "loss": 4.5526,
      "step": 18820
    },
    {
      "epoch": 0.01755260684292992,
      "grad_norm": 1.075565218925476,
      "learning_rate": 0.00019648966792694842,
      "loss": 4.4279,
      "step": 18830
    },
    {
      "epoch": 0.017561928461009013,
      "grad_norm": 1.2968484163284302,
      "learning_rate": 0.00019648780360485322,
      "loss": 4.5314,
      "step": 18840
    },
    {
      "epoch": 0.017571250079088104,
      "grad_norm": 1.2768299579620361,
      "learning_rate": 0.000196485939282758,
      "loss": 4.5413,
      "step": 18850
    },
    {
      "epoch": 0.017580571697167195,
      "grad_norm": 0.9205238819122314,
      "learning_rate": 0.0001964840749606628,
      "loss": 4.4163,
      "step": 18860
    },
    {
      "epoch": 0.017589893315246286,
      "grad_norm": 1.1214430332183838,
      "learning_rate": 0.00019648221063856762,
      "loss": 4.4632,
      "step": 18870
    },
    {
      "epoch": 0.017599214933325377,
      "grad_norm": 1.1342260837554932,
      "learning_rate": 0.0001964803463164724,
      "loss": 4.5778,
      "step": 18880
    },
    {
      "epoch": 0.01760853655140447,
      "grad_norm": 0.934218168258667,
      "learning_rate": 0.00019647848199437723,
      "loss": 4.5696,
      "step": 18890
    },
    {
      "epoch": 0.017617858169483563,
      "grad_norm": 1.228265404701233,
      "learning_rate": 0.000196476617672282,
      "loss": 4.4262,
      "step": 18900
    },
    {
      "epoch": 0.017627179787562654,
      "grad_norm": 1.2346549034118652,
      "learning_rate": 0.00019647475335018681,
      "loss": 4.4007,
      "step": 18910
    },
    {
      "epoch": 0.017636501405641745,
      "grad_norm": 1.0661430358886719,
      "learning_rate": 0.00019647288902809162,
      "loss": 4.4533,
      "step": 18920
    },
    {
      "epoch": 0.01764582302372084,
      "grad_norm": 1.0338377952575684,
      "learning_rate": 0.0001964710247059964,
      "loss": 4.4851,
      "step": 18930
    },
    {
      "epoch": 0.01765514464179993,
      "grad_norm": 1.2428032159805298,
      "learning_rate": 0.0001964691603839012,
      "loss": 4.4496,
      "step": 18940
    },
    {
      "epoch": 0.01766446625987902,
      "grad_norm": 1.2769447565078735,
      "learning_rate": 0.000196467296061806,
      "loss": 4.3195,
      "step": 18950
    },
    {
      "epoch": 0.017673787877958112,
      "grad_norm": 1.106160044670105,
      "learning_rate": 0.00019646543173971082,
      "loss": 4.4371,
      "step": 18960
    },
    {
      "epoch": 0.017683109496037207,
      "grad_norm": 1.5045709609985352,
      "learning_rate": 0.00019646356741761563,
      "loss": 4.4535,
      "step": 18970
    },
    {
      "epoch": 0.017692431114116298,
      "grad_norm": 1.053049087524414,
      "learning_rate": 0.0001964617030955204,
      "loss": 4.1929,
      "step": 18980
    },
    {
      "epoch": 0.01770175273219539,
      "grad_norm": 1.2322022914886475,
      "learning_rate": 0.0001964598387734252,
      "loss": 4.3828,
      "step": 18990
    },
    {
      "epoch": 0.01771107435027448,
      "grad_norm": 1.1870911121368408,
      "learning_rate": 0.00019645797445133002,
      "loss": 4.1368,
      "step": 19000
    },
    {
      "epoch": 0.017720395968353574,
      "grad_norm": 0.9881328344345093,
      "learning_rate": 0.00019645611012923482,
      "loss": 4.5072,
      "step": 19010
    },
    {
      "epoch": 0.017729717586432665,
      "grad_norm": 1.1495641469955444,
      "learning_rate": 0.00019645424580713963,
      "loss": 4.4003,
      "step": 19020
    },
    {
      "epoch": 0.017739039204511756,
      "grad_norm": 0.9587923288345337,
      "learning_rate": 0.0001964523814850444,
      "loss": 4.4454,
      "step": 19030
    },
    {
      "epoch": 0.017748360822590847,
      "grad_norm": 1.490348219871521,
      "learning_rate": 0.00019645051716294922,
      "loss": 4.2292,
      "step": 19040
    },
    {
      "epoch": 0.01775768244066994,
      "grad_norm": 1.4241642951965332,
      "learning_rate": 0.000196448652840854,
      "loss": 4.5221,
      "step": 19050
    },
    {
      "epoch": 0.017767004058749033,
      "grad_norm": 1.0849642753601074,
      "learning_rate": 0.00019644678851875883,
      "loss": 4.414,
      "step": 19060
    },
    {
      "epoch": 0.017776325676828124,
      "grad_norm": 1.3248785734176636,
      "learning_rate": 0.00019644492419666363,
      "loss": 4.284,
      "step": 19070
    },
    {
      "epoch": 0.017785647294907215,
      "grad_norm": 1.2628592252731323,
      "learning_rate": 0.00019644305987456841,
      "loss": 4.359,
      "step": 19080
    },
    {
      "epoch": 0.017794968912986306,
      "grad_norm": 1.38740074634552,
      "learning_rate": 0.00019644119555247322,
      "loss": 4.2798,
      "step": 19090
    },
    {
      "epoch": 0.0178042905310654,
      "grad_norm": 1.234161615371704,
      "learning_rate": 0.000196439331230378,
      "loss": 4.4294,
      "step": 19100
    },
    {
      "epoch": 0.01781361214914449,
      "grad_norm": 1.3601151704788208,
      "learning_rate": 0.0001964374669082828,
      "loss": 4.707,
      "step": 19110
    },
    {
      "epoch": 0.017822933767223582,
      "grad_norm": 1.9480451345443726,
      "learning_rate": 0.00019643560258618764,
      "loss": 4.4839,
      "step": 19120
    },
    {
      "epoch": 0.017832255385302673,
      "grad_norm": 1.1767258644104004,
      "learning_rate": 0.00019643373826409242,
      "loss": 4.5533,
      "step": 19130
    },
    {
      "epoch": 0.017841577003381768,
      "grad_norm": 1.3418999910354614,
      "learning_rate": 0.00019643187394199722,
      "loss": 4.231,
      "step": 19140
    },
    {
      "epoch": 0.01785089862146086,
      "grad_norm": 1.0574172735214233,
      "learning_rate": 0.000196430009619902,
      "loss": 4.5717,
      "step": 19150
    },
    {
      "epoch": 0.01786022023953995,
      "grad_norm": 1.3653228282928467,
      "learning_rate": 0.0001964281452978068,
      "loss": 4.4263,
      "step": 19160
    },
    {
      "epoch": 0.01786954185761904,
      "grad_norm": 1.3652501106262207,
      "learning_rate": 0.00019642628097571162,
      "loss": 4.5208,
      "step": 19170
    },
    {
      "epoch": 0.017878863475698135,
      "grad_norm": 1.1187973022460938,
      "learning_rate": 0.00019642441665361642,
      "loss": 4.4413,
      "step": 19180
    },
    {
      "epoch": 0.017888185093777226,
      "grad_norm": 1.3650189638137817,
      "learning_rate": 0.00019642255233152123,
      "loss": 4.3907,
      "step": 19190
    },
    {
      "epoch": 0.017897506711856317,
      "grad_norm": 0.9688619375228882,
      "learning_rate": 0.000196420688009426,
      "loss": 4.4786,
      "step": 19200
    },
    {
      "epoch": 0.01790682832993541,
      "grad_norm": 1.298248052597046,
      "learning_rate": 0.00019641882368733081,
      "loss": 4.5751,
      "step": 19210
    },
    {
      "epoch": 0.0179161499480145,
      "grad_norm": 1.139804482460022,
      "learning_rate": 0.00019641695936523562,
      "loss": 4.5544,
      "step": 19220
    },
    {
      "epoch": 0.017925471566093594,
      "grad_norm": 1.201224684715271,
      "learning_rate": 0.00019641509504314043,
      "loss": 4.3694,
      "step": 19230
    },
    {
      "epoch": 0.017934793184172685,
      "grad_norm": 1.1764522790908813,
      "learning_rate": 0.00019641323072104523,
      "loss": 4.3512,
      "step": 19240
    },
    {
      "epoch": 0.017944114802251776,
      "grad_norm": 1.2997349500656128,
      "learning_rate": 0.00019641136639895,
      "loss": 4.4481,
      "step": 19250
    },
    {
      "epoch": 0.017953436420330867,
      "grad_norm": 1.365261197090149,
      "learning_rate": 0.00019640950207685482,
      "loss": 4.6456,
      "step": 19260
    },
    {
      "epoch": 0.01796275803840996,
      "grad_norm": 1.2987282276153564,
      "learning_rate": 0.00019640763775475963,
      "loss": 4.4443,
      "step": 19270
    },
    {
      "epoch": 0.017972079656489053,
      "grad_norm": 1.271437644958496,
      "learning_rate": 0.00019640577343266443,
      "loss": 4.2467,
      "step": 19280
    },
    {
      "epoch": 0.017981401274568144,
      "grad_norm": 1.2554585933685303,
      "learning_rate": 0.00019640390911056924,
      "loss": 4.4098,
      "step": 19290
    },
    {
      "epoch": 0.017990722892647235,
      "grad_norm": 1.128904938697815,
      "learning_rate": 0.00019640204478847402,
      "loss": 4.3351,
      "step": 19300
    },
    {
      "epoch": 0.01800004451072633,
      "grad_norm": 1.0735737085342407,
      "learning_rate": 0.00019640018046637882,
      "loss": 4.3895,
      "step": 19310
    },
    {
      "epoch": 0.01800936612880542,
      "grad_norm": 1.1193442344665527,
      "learning_rate": 0.00019639831614428363,
      "loss": 4.6629,
      "step": 19320
    },
    {
      "epoch": 0.01801868774688451,
      "grad_norm": 1.105063796043396,
      "learning_rate": 0.0001963964518221884,
      "loss": 4.4292,
      "step": 19330
    },
    {
      "epoch": 0.018028009364963602,
      "grad_norm": 1.1345516443252563,
      "learning_rate": 0.00019639458750009324,
      "loss": 4.3801,
      "step": 19340
    },
    {
      "epoch": 0.018037330983042697,
      "grad_norm": 0.9962632060050964,
      "learning_rate": 0.00019639272317799802,
      "loss": 4.3706,
      "step": 19350
    },
    {
      "epoch": 0.018046652601121788,
      "grad_norm": 1.008607268333435,
      "learning_rate": 0.00019639085885590283,
      "loss": 4.598,
      "step": 19360
    },
    {
      "epoch": 0.01805597421920088,
      "grad_norm": 1.0815128087997437,
      "learning_rate": 0.00019638899453380763,
      "loss": 4.4604,
      "step": 19370
    },
    {
      "epoch": 0.01806529583727997,
      "grad_norm": 1.1405246257781982,
      "learning_rate": 0.0001963871302117124,
      "loss": 4.4713,
      "step": 19380
    },
    {
      "epoch": 0.01807461745535906,
      "grad_norm": 1.038407325744629,
      "learning_rate": 0.00019638526588961722,
      "loss": 4.4983,
      "step": 19390
    },
    {
      "epoch": 0.018083939073438155,
      "grad_norm": 1.2526684999465942,
      "learning_rate": 0.00019638340156752203,
      "loss": 4.3759,
      "step": 19400
    },
    {
      "epoch": 0.018093260691517246,
      "grad_norm": 1.2082592248916626,
      "learning_rate": 0.00019638153724542683,
      "loss": 4.3553,
      "step": 19410
    },
    {
      "epoch": 0.018102582309596337,
      "grad_norm": 1.1537495851516724,
      "learning_rate": 0.00019637967292333164,
      "loss": 4.16,
      "step": 19420
    },
    {
      "epoch": 0.018111903927675428,
      "grad_norm": 1.1604093313217163,
      "learning_rate": 0.00019637780860123642,
      "loss": 4.6495,
      "step": 19430
    },
    {
      "epoch": 0.018121225545754523,
      "grad_norm": 1.154595971107483,
      "learning_rate": 0.00019637594427914122,
      "loss": 4.4719,
      "step": 19440
    },
    {
      "epoch": 0.018130547163833614,
      "grad_norm": 1.1945675611495972,
      "learning_rate": 0.00019637407995704603,
      "loss": 4.3308,
      "step": 19450
    },
    {
      "epoch": 0.018139868781912705,
      "grad_norm": 1.4529187679290771,
      "learning_rate": 0.00019637221563495084,
      "loss": 4.4656,
      "step": 19460
    },
    {
      "epoch": 0.018149190399991796,
      "grad_norm": 1.0295666456222534,
      "learning_rate": 0.00019637035131285564,
      "loss": 4.499,
      "step": 19470
    },
    {
      "epoch": 0.01815851201807089,
      "grad_norm": 1.1387856006622314,
      "learning_rate": 0.00019636848699076042,
      "loss": 4.4076,
      "step": 19480
    },
    {
      "epoch": 0.01816783363614998,
      "grad_norm": 1.2374390363693237,
      "learning_rate": 0.00019636662266866523,
      "loss": 4.2798,
      "step": 19490
    },
    {
      "epoch": 0.018177155254229072,
      "grad_norm": 1.1212693452835083,
      "learning_rate": 0.00019636475834657003,
      "loss": 4.4254,
      "step": 19500
    },
    {
      "epoch": 0.018186476872308163,
      "grad_norm": 1.135055422782898,
      "learning_rate": 0.00019636289402447484,
      "loss": 4.4007,
      "step": 19510
    },
    {
      "epoch": 0.018195798490387258,
      "grad_norm": 1.4091242551803589,
      "learning_rate": 0.00019636102970237965,
      "loss": 4.4974,
      "step": 19520
    },
    {
      "epoch": 0.01820512010846635,
      "grad_norm": 1.0359926223754883,
      "learning_rate": 0.00019635916538028443,
      "loss": 4.4574,
      "step": 19530
    },
    {
      "epoch": 0.01821444172654544,
      "grad_norm": 1.2142168283462524,
      "learning_rate": 0.00019635730105818923,
      "loss": 4.2786,
      "step": 19540
    },
    {
      "epoch": 0.01822376334462453,
      "grad_norm": 1.3212018013000488,
      "learning_rate": 0.000196355436736094,
      "loss": 4.2603,
      "step": 19550
    },
    {
      "epoch": 0.018233084962703622,
      "grad_norm": 0.9568943381309509,
      "learning_rate": 0.00019635357241399885,
      "loss": 4.4141,
      "step": 19560
    },
    {
      "epoch": 0.018242406580782716,
      "grad_norm": 0.91089928150177,
      "learning_rate": 0.00019635170809190365,
      "loss": 4.3969,
      "step": 19570
    },
    {
      "epoch": 0.018251728198861807,
      "grad_norm": 1.2562808990478516,
      "learning_rate": 0.00019634984376980843,
      "loss": 4.4843,
      "step": 19580
    },
    {
      "epoch": 0.0182610498169409,
      "grad_norm": 1.1273783445358276,
      "learning_rate": 0.00019634797944771324,
      "loss": 4.296,
      "step": 19590
    },
    {
      "epoch": 0.01827037143501999,
      "grad_norm": 0.9810512065887451,
      "learning_rate": 0.00019634611512561802,
      "loss": 4.4802,
      "step": 19600
    },
    {
      "epoch": 0.018279693053099084,
      "grad_norm": 1.2574806213378906,
      "learning_rate": 0.00019634425080352282,
      "loss": 4.534,
      "step": 19610
    },
    {
      "epoch": 0.018289014671178175,
      "grad_norm": 1.02178156375885,
      "learning_rate": 0.00019634238648142763,
      "loss": 4.4034,
      "step": 19620
    },
    {
      "epoch": 0.018298336289257266,
      "grad_norm": 1.2863718271255493,
      "learning_rate": 0.00019634052215933244,
      "loss": 4.2178,
      "step": 19630
    },
    {
      "epoch": 0.018307657907336357,
      "grad_norm": 1.2139842510223389,
      "learning_rate": 0.00019633865783723724,
      "loss": 4.441,
      "step": 19640
    },
    {
      "epoch": 0.01831697952541545,
      "grad_norm": 1.0023056268692017,
      "learning_rate": 0.00019633679351514202,
      "loss": 4.5606,
      "step": 19650
    },
    {
      "epoch": 0.018326301143494542,
      "grad_norm": 1.1372485160827637,
      "learning_rate": 0.00019633492919304683,
      "loss": 4.253,
      "step": 19660
    },
    {
      "epoch": 0.018335622761573633,
      "grad_norm": 0.9959949254989624,
      "learning_rate": 0.00019633306487095163,
      "loss": 4.3265,
      "step": 19670
    },
    {
      "epoch": 0.018344944379652724,
      "grad_norm": 1.2689061164855957,
      "learning_rate": 0.00019633120054885644,
      "loss": 4.3769,
      "step": 19680
    },
    {
      "epoch": 0.01835426599773182,
      "grad_norm": 1.0386719703674316,
      "learning_rate": 0.00019632933622676125,
      "loss": 4.5459,
      "step": 19690
    },
    {
      "epoch": 0.01836358761581091,
      "grad_norm": 1.3511146306991577,
      "learning_rate": 0.00019632747190466603,
      "loss": 4.3502,
      "step": 19700
    },
    {
      "epoch": 0.01837290923389,
      "grad_norm": 1.0629262924194336,
      "learning_rate": 0.00019632560758257083,
      "loss": 4.3405,
      "step": 19710
    },
    {
      "epoch": 0.018382230851969092,
      "grad_norm": 1.1152138710021973,
      "learning_rate": 0.00019632374326047564,
      "loss": 4.4349,
      "step": 19720
    },
    {
      "epoch": 0.018391552470048183,
      "grad_norm": 1.135354995727539,
      "learning_rate": 0.00019632187893838044,
      "loss": 4.4809,
      "step": 19730
    },
    {
      "epoch": 0.018400874088127277,
      "grad_norm": 1.1966078281402588,
      "learning_rate": 0.00019632001461628525,
      "loss": 4.3954,
      "step": 19740
    },
    {
      "epoch": 0.01841019570620637,
      "grad_norm": 1.2959740161895752,
      "learning_rate": 0.00019631815029419003,
      "loss": 4.5993,
      "step": 19750
    },
    {
      "epoch": 0.01841951732428546,
      "grad_norm": 1.3443161249160767,
      "learning_rate": 0.00019631628597209484,
      "loss": 4.3897,
      "step": 19760
    },
    {
      "epoch": 0.01842883894236455,
      "grad_norm": 1.0837613344192505,
      "learning_rate": 0.00019631442164999964,
      "loss": 4.6203,
      "step": 19770
    },
    {
      "epoch": 0.018438160560443645,
      "grad_norm": 1.1697224378585815,
      "learning_rate": 0.00019631255732790442,
      "loss": 4.4334,
      "step": 19780
    },
    {
      "epoch": 0.018447482178522736,
      "grad_norm": 1.128747582435608,
      "learning_rate": 0.00019631069300580926,
      "loss": 4.5457,
      "step": 19790
    },
    {
      "epoch": 0.018456803796601827,
      "grad_norm": 1.2745335102081299,
      "learning_rate": 0.00019630882868371403,
      "loss": 4.5747,
      "step": 19800
    },
    {
      "epoch": 0.018466125414680918,
      "grad_norm": 1.2107248306274414,
      "learning_rate": 0.00019630696436161884,
      "loss": 4.5185,
      "step": 19810
    },
    {
      "epoch": 0.018475447032760012,
      "grad_norm": 1.068217396736145,
      "learning_rate": 0.00019630510003952365,
      "loss": 4.2983,
      "step": 19820
    },
    {
      "epoch": 0.018484768650839103,
      "grad_norm": 1.1653037071228027,
      "learning_rate": 0.00019630323571742843,
      "loss": 4.4138,
      "step": 19830
    },
    {
      "epoch": 0.018494090268918194,
      "grad_norm": 1.1704659461975098,
      "learning_rate": 0.00019630137139533323,
      "loss": 4.2986,
      "step": 19840
    },
    {
      "epoch": 0.018503411886997286,
      "grad_norm": 1.6571992635726929,
      "learning_rate": 0.00019629950707323804,
      "loss": 4.5849,
      "step": 19850
    },
    {
      "epoch": 0.01851273350507638,
      "grad_norm": 1.1686062812805176,
      "learning_rate": 0.00019629764275114285,
      "loss": 4.4839,
      "step": 19860
    },
    {
      "epoch": 0.01852205512315547,
      "grad_norm": 1.1733182668685913,
      "learning_rate": 0.00019629577842904765,
      "loss": 4.2319,
      "step": 19870
    },
    {
      "epoch": 0.018531376741234562,
      "grad_norm": 1.1333258152008057,
      "learning_rate": 0.00019629391410695243,
      "loss": 4.4558,
      "step": 19880
    },
    {
      "epoch": 0.018540698359313653,
      "grad_norm": 1.2228325605392456,
      "learning_rate": 0.00019629204978485724,
      "loss": 4.2284,
      "step": 19890
    },
    {
      "epoch": 0.018550019977392744,
      "grad_norm": 1.272190809249878,
      "learning_rate": 0.00019629018546276204,
      "loss": 4.2343,
      "step": 19900
    },
    {
      "epoch": 0.01855934159547184,
      "grad_norm": 1.152663230895996,
      "learning_rate": 0.00019628832114066685,
      "loss": 4.3564,
      "step": 19910
    },
    {
      "epoch": 0.01856866321355093,
      "grad_norm": 1.32725989818573,
      "learning_rate": 0.00019628645681857166,
      "loss": 4.2578,
      "step": 19920
    },
    {
      "epoch": 0.01857798483163002,
      "grad_norm": 1.1691837310791016,
      "learning_rate": 0.00019628459249647643,
      "loss": 4.433,
      "step": 19930
    },
    {
      "epoch": 0.01858730644970911,
      "grad_norm": 1.1789283752441406,
      "learning_rate": 0.00019628272817438124,
      "loss": 4.4401,
      "step": 19940
    },
    {
      "epoch": 0.018596628067788206,
      "grad_norm": 0.9911158680915833,
      "learning_rate": 0.00019628086385228605,
      "loss": 4.3791,
      "step": 19950
    },
    {
      "epoch": 0.018605949685867297,
      "grad_norm": 1.2352948188781738,
      "learning_rate": 0.00019627899953019085,
      "loss": 4.5313,
      "step": 19960
    },
    {
      "epoch": 0.018615271303946388,
      "grad_norm": 1.0205719470977783,
      "learning_rate": 0.00019627713520809566,
      "loss": 4.4891,
      "step": 19970
    },
    {
      "epoch": 0.01862459292202548,
      "grad_norm": 1.3951219320297241,
      "learning_rate": 0.00019627527088600044,
      "loss": 4.248,
      "step": 19980
    },
    {
      "epoch": 0.018633914540104574,
      "grad_norm": 1.3973133563995361,
      "learning_rate": 0.00019627340656390525,
      "loss": 4.4911,
      "step": 19990
    },
    {
      "epoch": 0.018643236158183665,
      "grad_norm": 1.082179307937622,
      "learning_rate": 0.00019627154224181002,
      "loss": 4.3137,
      "step": 20000
    },
    {
      "epoch": 0.018652557776262756,
      "grad_norm": 1.1596590280532837,
      "learning_rate": 0.00019626967791971486,
      "loss": 4.5576,
      "step": 20010
    },
    {
      "epoch": 0.018661879394341847,
      "grad_norm": 1.3845595121383667,
      "learning_rate": 0.00019626781359761964,
      "loss": 4.4543,
      "step": 20020
    },
    {
      "epoch": 0.01867120101242094,
      "grad_norm": 1.1845121383666992,
      "learning_rate": 0.00019626594927552444,
      "loss": 4.3104,
      "step": 20030
    },
    {
      "epoch": 0.018680522630500032,
      "grad_norm": 1.2037529945373535,
      "learning_rate": 0.00019626408495342925,
      "loss": 4.2642,
      "step": 20040
    },
    {
      "epoch": 0.018689844248579123,
      "grad_norm": 1.0558991432189941,
      "learning_rate": 0.00019626222063133403,
      "loss": 4.3036,
      "step": 20050
    },
    {
      "epoch": 0.018699165866658214,
      "grad_norm": 1.323957085609436,
      "learning_rate": 0.00019626035630923884,
      "loss": 4.4562,
      "step": 20060
    },
    {
      "epoch": 0.018708487484737305,
      "grad_norm": 1.305452585220337,
      "learning_rate": 0.00019625849198714364,
      "loss": 4.2131,
      "step": 20070
    },
    {
      "epoch": 0.0187178091028164,
      "grad_norm": 0.8583149909973145,
      "learning_rate": 0.00019625662766504845,
      "loss": 4.3727,
      "step": 20080
    },
    {
      "epoch": 0.01872713072089549,
      "grad_norm": 1.317604660987854,
      "learning_rate": 0.00019625476334295325,
      "loss": 4.4878,
      "step": 20090
    },
    {
      "epoch": 0.01873645233897458,
      "grad_norm": 1.2243614196777344,
      "learning_rate": 0.00019625289902085803,
      "loss": 4.4556,
      "step": 20100
    },
    {
      "epoch": 0.018745773957053673,
      "grad_norm": 1.2062252759933472,
      "learning_rate": 0.00019625103469876284,
      "loss": 4.3992,
      "step": 20110
    },
    {
      "epoch": 0.018755095575132767,
      "grad_norm": 1.3009452819824219,
      "learning_rate": 0.00019624917037666765,
      "loss": 4.3974,
      "step": 20120
    },
    {
      "epoch": 0.018764417193211858,
      "grad_norm": 1.3617554903030396,
      "learning_rate": 0.00019624730605457245,
      "loss": 4.491,
      "step": 20130
    },
    {
      "epoch": 0.01877373881129095,
      "grad_norm": 1.155834674835205,
      "learning_rate": 0.00019624544173247726,
      "loss": 4.3948,
      "step": 20140
    },
    {
      "epoch": 0.01878306042937004,
      "grad_norm": 1.117685079574585,
      "learning_rate": 0.00019624357741038204,
      "loss": 4.2336,
      "step": 20150
    },
    {
      "epoch": 0.018792382047449135,
      "grad_norm": 1.089292049407959,
      "learning_rate": 0.00019624171308828684,
      "loss": 4.2923,
      "step": 20160
    },
    {
      "epoch": 0.018801703665528226,
      "grad_norm": 1.0487219095230103,
      "learning_rate": 0.00019623984876619162,
      "loss": 4.5091,
      "step": 20170
    },
    {
      "epoch": 0.018811025283607317,
      "grad_norm": 1.265276312828064,
      "learning_rate": 0.00019623798444409646,
      "loss": 4.3605,
      "step": 20180
    },
    {
      "epoch": 0.018820346901686408,
      "grad_norm": 1.204126000404358,
      "learning_rate": 0.00019623612012200126,
      "loss": 4.6098,
      "step": 20190
    },
    {
      "epoch": 0.018829668519765502,
      "grad_norm": 1.2844716310501099,
      "learning_rate": 0.00019623425579990604,
      "loss": 4.298,
      "step": 20200
    },
    {
      "epoch": 0.018838990137844593,
      "grad_norm": 1.2855230569839478,
      "learning_rate": 0.00019623239147781085,
      "loss": 4.1438,
      "step": 20210
    },
    {
      "epoch": 0.018848311755923684,
      "grad_norm": 1.3872344493865967,
      "learning_rate": 0.00019623052715571563,
      "loss": 4.3852,
      "step": 20220
    },
    {
      "epoch": 0.018857633374002775,
      "grad_norm": 1.202063798904419,
      "learning_rate": 0.00019622866283362043,
      "loss": 4.4899,
      "step": 20230
    },
    {
      "epoch": 0.018866954992081866,
      "grad_norm": 1.2650225162506104,
      "learning_rate": 0.00019622679851152527,
      "loss": 4.5143,
      "step": 20240
    },
    {
      "epoch": 0.01887627661016096,
      "grad_norm": 1.164445161819458,
      "learning_rate": 0.00019622493418943005,
      "loss": 4.5413,
      "step": 20250
    },
    {
      "epoch": 0.018885598228240052,
      "grad_norm": 1.145704746246338,
      "learning_rate": 0.00019622306986733485,
      "loss": 4.3179,
      "step": 20260
    },
    {
      "epoch": 0.018894919846319143,
      "grad_norm": 1.2930110692977905,
      "learning_rate": 0.00019622120554523963,
      "loss": 4.5036,
      "step": 20270
    },
    {
      "epoch": 0.018904241464398234,
      "grad_norm": 1.4677107334136963,
      "learning_rate": 0.00019621934122314444,
      "loss": 4.5449,
      "step": 20280
    },
    {
      "epoch": 0.01891356308247733,
      "grad_norm": 1.0137535333633423,
      "learning_rate": 0.00019621747690104925,
      "loss": 4.3998,
      "step": 20290
    },
    {
      "epoch": 0.01892288470055642,
      "grad_norm": 1.2589390277862549,
      "learning_rate": 0.00019621561257895405,
      "loss": 4.4851,
      "step": 20300
    },
    {
      "epoch": 0.01893220631863551,
      "grad_norm": 1.2181401252746582,
      "learning_rate": 0.00019621374825685886,
      "loss": 4.3876,
      "step": 20310
    },
    {
      "epoch": 0.0189415279367146,
      "grad_norm": 1.1441872119903564,
      "learning_rate": 0.00019621188393476364,
      "loss": 4.4124,
      "step": 20320
    },
    {
      "epoch": 0.018950849554793696,
      "grad_norm": 1.2047834396362305,
      "learning_rate": 0.00019621001961266844,
      "loss": 4.3817,
      "step": 20330
    },
    {
      "epoch": 0.018960171172872787,
      "grad_norm": 1.273178219795227,
      "learning_rate": 0.00019620815529057325,
      "loss": 4.0468,
      "step": 20340
    },
    {
      "epoch": 0.018969492790951878,
      "grad_norm": 1.0388067960739136,
      "learning_rate": 0.00019620629096847806,
      "loss": 4.3407,
      "step": 20350
    },
    {
      "epoch": 0.01897881440903097,
      "grad_norm": 1.04404878616333,
      "learning_rate": 0.00019620442664638286,
      "loss": 4.3713,
      "step": 20360
    },
    {
      "epoch": 0.018988136027110063,
      "grad_norm": 1.2110146284103394,
      "learning_rate": 0.00019620256232428764,
      "loss": 4.269,
      "step": 20370
    },
    {
      "epoch": 0.018997457645189154,
      "grad_norm": 1.0189298391342163,
      "learning_rate": 0.00019620069800219245,
      "loss": 4.399,
      "step": 20380
    },
    {
      "epoch": 0.019006779263268245,
      "grad_norm": 1.1470352411270142,
      "learning_rate": 0.00019619883368009725,
      "loss": 4.4328,
      "step": 20390
    },
    {
      "epoch": 0.019016100881347336,
      "grad_norm": 1.1221445798873901,
      "learning_rate": 0.00019619696935800206,
      "loss": 4.3768,
      "step": 20400
    },
    {
      "epoch": 0.019025422499426427,
      "grad_norm": 1.321506381034851,
      "learning_rate": 0.00019619510503590687,
      "loss": 4.5907,
      "step": 20410
    },
    {
      "epoch": 0.019034744117505522,
      "grad_norm": 1.0564603805541992,
      "learning_rate": 0.00019619324071381165,
      "loss": 4.5801,
      "step": 20420
    },
    {
      "epoch": 0.019044065735584613,
      "grad_norm": 1.227101445198059,
      "learning_rate": 0.00019619137639171645,
      "loss": 4.316,
      "step": 20430
    },
    {
      "epoch": 0.019053387353663704,
      "grad_norm": 1.2423146963119507,
      "learning_rate": 0.00019618951206962126,
      "loss": 4.2625,
      "step": 20440
    },
    {
      "epoch": 0.019062708971742795,
      "grad_norm": 1.0744585990905762,
      "learning_rate": 0.00019618764774752604,
      "loss": 4.5342,
      "step": 20450
    },
    {
      "epoch": 0.01907203058982189,
      "grad_norm": Infinity,
      "learning_rate": 0.00019618578342543087,
      "loss": 4.2959,
      "step": 20460
    },
    {
      "epoch": 0.01908135220790098,
      "grad_norm": 1.0980972051620483,
      "learning_rate": 0.00019618391910333565,
      "loss": 4.3438,
      "step": 20470
    },
    {
      "epoch": 0.01909067382598007,
      "grad_norm": 1.3860713243484497,
      "learning_rate": 0.00019618205478124046,
      "loss": 4.2619,
      "step": 20480
    },
    {
      "epoch": 0.019099995444059163,
      "grad_norm": 1.0166845321655273,
      "learning_rate": 0.00019618019045914526,
      "loss": 4.5034,
      "step": 20490
    },
    {
      "epoch": 0.019109317062138257,
      "grad_norm": 1.2529436349868774,
      "learning_rate": 0.00019617832613705004,
      "loss": 4.287,
      "step": 20500
    },
    {
      "epoch": 0.019118638680217348,
      "grad_norm": 0.9913250207901001,
      "learning_rate": 0.00019617646181495485,
      "loss": 4.3377,
      "step": 20510
    },
    {
      "epoch": 0.01912796029829644,
      "grad_norm": 1.158797264099121,
      "learning_rate": 0.00019617459749285965,
      "loss": 4.4818,
      "step": 20520
    },
    {
      "epoch": 0.01913728191637553,
      "grad_norm": 1.093450903892517,
      "learning_rate": 0.00019617273317076446,
      "loss": 4.3571,
      "step": 20530
    },
    {
      "epoch": 0.019146603534454625,
      "grad_norm": 1.0441466569900513,
      "learning_rate": 0.00019617086884866927,
      "loss": 4.3837,
      "step": 20540
    },
    {
      "epoch": 0.019155925152533716,
      "grad_norm": 1.356174349784851,
      "learning_rate": 0.00019616900452657405,
      "loss": 4.168,
      "step": 20550
    },
    {
      "epoch": 0.019165246770612807,
      "grad_norm": 1.1603717803955078,
      "learning_rate": 0.00019616714020447885,
      "loss": 4.561,
      "step": 20560
    },
    {
      "epoch": 0.019174568388691898,
      "grad_norm": 1.146405577659607,
      "learning_rate": 0.00019616527588238366,
      "loss": 4.3759,
      "step": 20570
    },
    {
      "epoch": 0.01918389000677099,
      "grad_norm": 1.1291903257369995,
      "learning_rate": 0.00019616341156028847,
      "loss": 4.4785,
      "step": 20580
    },
    {
      "epoch": 0.019193211624850083,
      "grad_norm": 1.001545786857605,
      "learning_rate": 0.00019616154723819327,
      "loss": 4.4736,
      "step": 20590
    },
    {
      "epoch": 0.019202533242929174,
      "grad_norm": 1.0205777883529663,
      "learning_rate": 0.00019615968291609805,
      "loss": 4.398,
      "step": 20600
    },
    {
      "epoch": 0.019211854861008265,
      "grad_norm": 1.2815208435058594,
      "learning_rate": 0.00019615781859400286,
      "loss": 4.3413,
      "step": 20610
    },
    {
      "epoch": 0.019221176479087356,
      "grad_norm": 1.1294183731079102,
      "learning_rate": 0.00019615595427190764,
      "loss": 4.3972,
      "step": 20620
    },
    {
      "epoch": 0.01923049809716645,
      "grad_norm": 1.264981746673584,
      "learning_rate": 0.00019615408994981247,
      "loss": 4.3791,
      "step": 20630
    },
    {
      "epoch": 0.01923981971524554,
      "grad_norm": 1.0754454135894775,
      "learning_rate": 0.00019615222562771728,
      "loss": 4.3661,
      "step": 20640
    },
    {
      "epoch": 0.019249141333324633,
      "grad_norm": 1.0976954698562622,
      "learning_rate": 0.00019615036130562206,
      "loss": 4.3057,
      "step": 20650
    },
    {
      "epoch": 0.019258462951403724,
      "grad_norm": 1.1836508512496948,
      "learning_rate": 0.00019614849698352686,
      "loss": 4.3794,
      "step": 20660
    },
    {
      "epoch": 0.019267784569482818,
      "grad_norm": 1.004085898399353,
      "learning_rate": 0.00019614663266143164,
      "loss": 4.2473,
      "step": 20670
    },
    {
      "epoch": 0.01927710618756191,
      "grad_norm": 1.1904170513153076,
      "learning_rate": 0.00019614476833933645,
      "loss": 4.4275,
      "step": 20680
    },
    {
      "epoch": 0.019286427805641,
      "grad_norm": 1.2462722063064575,
      "learning_rate": 0.00019614290401724128,
      "loss": 4.3749,
      "step": 20690
    },
    {
      "epoch": 0.01929574942372009,
      "grad_norm": 1.2381806373596191,
      "learning_rate": 0.00019614103969514606,
      "loss": 4.5711,
      "step": 20700
    },
    {
      "epoch": 0.019305071041799186,
      "grad_norm": 1.4246044158935547,
      "learning_rate": 0.00019613917537305087,
      "loss": 4.1033,
      "step": 20710
    },
    {
      "epoch": 0.019314392659878277,
      "grad_norm": 1.425360918045044,
      "learning_rate": 0.00019613731105095565,
      "loss": 4.3653,
      "step": 20720
    },
    {
      "epoch": 0.019323714277957368,
      "grad_norm": 1.0930945873260498,
      "learning_rate": 0.00019613544672886045,
      "loss": 4.4227,
      "step": 20730
    },
    {
      "epoch": 0.01933303589603646,
      "grad_norm": 0.9793223142623901,
      "learning_rate": 0.00019613358240676526,
      "loss": 4.4451,
      "step": 20740
    },
    {
      "epoch": 0.019342357514115553,
      "grad_norm": 1.278320074081421,
      "learning_rate": 0.00019613171808467006,
      "loss": 4.5725,
      "step": 20750
    },
    {
      "epoch": 0.019351679132194644,
      "grad_norm": 0.9707649946212769,
      "learning_rate": 0.00019612985376257487,
      "loss": 4.4364,
      "step": 20760
    },
    {
      "epoch": 0.019361000750273735,
      "grad_norm": 0.8675856590270996,
      "learning_rate": 0.00019612798944047965,
      "loss": 4.516,
      "step": 20770
    },
    {
      "epoch": 0.019370322368352826,
      "grad_norm": 1.1237068176269531,
      "learning_rate": 0.00019612612511838446,
      "loss": 4.3561,
      "step": 20780
    },
    {
      "epoch": 0.019379643986431917,
      "grad_norm": 1.0325095653533936,
      "learning_rate": 0.00019612426079628926,
      "loss": 4.3361,
      "step": 20790
    },
    {
      "epoch": 0.019388965604511012,
      "grad_norm": 1.007977843284607,
      "learning_rate": 0.00019612239647419407,
      "loss": 4.5168,
      "step": 20800
    },
    {
      "epoch": 0.019398287222590103,
      "grad_norm": 1.053512692451477,
      "learning_rate": 0.00019612053215209888,
      "loss": 4.3322,
      "step": 20810
    },
    {
      "epoch": 0.019407608840669194,
      "grad_norm": 1.0802580118179321,
      "learning_rate": 0.00019611866783000365,
      "loss": 4.2414,
      "step": 20820
    },
    {
      "epoch": 0.019416930458748285,
      "grad_norm": 1.11273992061615,
      "learning_rate": 0.00019611680350790846,
      "loss": 4.5205,
      "step": 20830
    },
    {
      "epoch": 0.01942625207682738,
      "grad_norm": 1.2483340501785278,
      "learning_rate": 0.00019611493918581327,
      "loss": 4.4662,
      "step": 20840
    },
    {
      "epoch": 0.01943557369490647,
      "grad_norm": 1.221272349357605,
      "learning_rate": 0.00019611307486371807,
      "loss": 4.5081,
      "step": 20850
    },
    {
      "epoch": 0.01944489531298556,
      "grad_norm": 1.3163392543792725,
      "learning_rate": 0.00019611121054162288,
      "loss": 4.4121,
      "step": 20860
    },
    {
      "epoch": 0.019454216931064652,
      "grad_norm": 1.2238866090774536,
      "learning_rate": 0.00019610934621952766,
      "loss": 4.343,
      "step": 20870
    },
    {
      "epoch": 0.019463538549143747,
      "grad_norm": 0.9900808334350586,
      "learning_rate": 0.00019610748189743247,
      "loss": 4.3595,
      "step": 20880
    },
    {
      "epoch": 0.019472860167222838,
      "grad_norm": 1.1244488954544067,
      "learning_rate": 0.00019610561757533727,
      "loss": 4.4142,
      "step": 20890
    },
    {
      "epoch": 0.01948218178530193,
      "grad_norm": 1.2216675281524658,
      "learning_rate": 0.00019610375325324205,
      "loss": 4.3366,
      "step": 20900
    },
    {
      "epoch": 0.01949150340338102,
      "grad_norm": 1.231162667274475,
      "learning_rate": 0.00019610188893114688,
      "loss": 4.3387,
      "step": 20910
    },
    {
      "epoch": 0.019500825021460114,
      "grad_norm": 1.213599443435669,
      "learning_rate": 0.00019610002460905166,
      "loss": 4.3334,
      "step": 20920
    },
    {
      "epoch": 0.019510146639539205,
      "grad_norm": 1.2709356546401978,
      "learning_rate": 0.00019609816028695647,
      "loss": 4.3484,
      "step": 20930
    },
    {
      "epoch": 0.019519468257618296,
      "grad_norm": 2.632432699203491,
      "learning_rate": 0.00019609629596486128,
      "loss": 4.4049,
      "step": 20940
    },
    {
      "epoch": 0.019528789875697387,
      "grad_norm": 1.3603662252426147,
      "learning_rate": 0.00019609443164276606,
      "loss": 4.0506,
      "step": 20950
    },
    {
      "epoch": 0.01953811149377648,
      "grad_norm": 1.2098608016967773,
      "learning_rate": 0.00019609256732067086,
      "loss": 4.3049,
      "step": 20960
    },
    {
      "epoch": 0.019547433111855573,
      "grad_norm": 1.065596103668213,
      "learning_rate": 0.00019609070299857567,
      "loss": 4.4763,
      "step": 20970
    },
    {
      "epoch": 0.019556754729934664,
      "grad_norm": 1.0937974452972412,
      "learning_rate": 0.00019608883867648047,
      "loss": 4.3798,
      "step": 20980
    },
    {
      "epoch": 0.019566076348013755,
      "grad_norm": 1.122941017150879,
      "learning_rate": 0.00019608697435438528,
      "loss": 4.2839,
      "step": 20990
    },
    {
      "epoch": 0.019575397966092846,
      "grad_norm": 1.1293151378631592,
      "learning_rate": 0.00019608511003229006,
      "loss": 4.2154,
      "step": 21000
    },
    {
      "epoch": 0.01958471958417194,
      "grad_norm": 1.2419328689575195,
      "learning_rate": 0.00019608324571019487,
      "loss": 4.381,
      "step": 21010
    },
    {
      "epoch": 0.01959404120225103,
      "grad_norm": 1.3447153568267822,
      "learning_rate": 0.00019608138138809967,
      "loss": 4.3638,
      "step": 21020
    },
    {
      "epoch": 0.019603362820330122,
      "grad_norm": 1.1836564540863037,
      "learning_rate": 0.00019607951706600448,
      "loss": 4.5389,
      "step": 21030
    },
    {
      "epoch": 0.019612684438409213,
      "grad_norm": 1.652296781539917,
      "learning_rate": 0.00019607765274390928,
      "loss": 4.3494,
      "step": 21040
    },
    {
      "epoch": 0.019622006056488308,
      "grad_norm": 1.255491852760315,
      "learning_rate": 0.00019607578842181406,
      "loss": 4.3775,
      "step": 21050
    },
    {
      "epoch": 0.0196313276745674,
      "grad_norm": 1.1267136335372925,
      "learning_rate": 0.00019607392409971887,
      "loss": 4.3518,
      "step": 21060
    },
    {
      "epoch": 0.01964064929264649,
      "grad_norm": 0.9447970986366272,
      "learning_rate": 0.00019607205977762365,
      "loss": 4.5664,
      "step": 21070
    },
    {
      "epoch": 0.01964997091072558,
      "grad_norm": 1.1211813688278198,
      "learning_rate": 0.00019607019545552848,
      "loss": 4.335,
      "step": 21080
    },
    {
      "epoch": 0.019659292528804675,
      "grad_norm": 0.9617564082145691,
      "learning_rate": 0.0001960683311334333,
      "loss": 4.4384,
      "step": 21090
    },
    {
      "epoch": 0.019668614146883766,
      "grad_norm": 1.064266324043274,
      "learning_rate": 0.00019606646681133807,
      "loss": 4.4923,
      "step": 21100
    },
    {
      "epoch": 0.019677935764962858,
      "grad_norm": 0.9494556188583374,
      "learning_rate": 0.00019606460248924287,
      "loss": 4.6251,
      "step": 21110
    },
    {
      "epoch": 0.01968725738304195,
      "grad_norm": 1.0776262283325195,
      "learning_rate": 0.00019606273816714765,
      "loss": 4.2602,
      "step": 21120
    },
    {
      "epoch": 0.01969657900112104,
      "grad_norm": 1.211012363433838,
      "learning_rate": 0.00019606087384505246,
      "loss": 4.1661,
      "step": 21130
    },
    {
      "epoch": 0.019705900619200134,
      "grad_norm": 1.069645881652832,
      "learning_rate": 0.0001960590095229573,
      "loss": 4.4254,
      "step": 21140
    },
    {
      "epoch": 0.019715222237279225,
      "grad_norm": 1.2595583200454712,
      "learning_rate": 0.00019605714520086207,
      "loss": 4.4824,
      "step": 21150
    },
    {
      "epoch": 0.019724543855358316,
      "grad_norm": 1.2388463020324707,
      "learning_rate": 0.00019605528087876688,
      "loss": 4.3031,
      "step": 21160
    },
    {
      "epoch": 0.019733865473437407,
      "grad_norm": 1.3554813861846924,
      "learning_rate": 0.00019605341655667166,
      "loss": 4.5508,
      "step": 21170
    },
    {
      "epoch": 0.0197431870915165,
      "grad_norm": 1.1081284284591675,
      "learning_rate": 0.00019605155223457646,
      "loss": 4.3451,
      "step": 21180
    },
    {
      "epoch": 0.019752508709595593,
      "grad_norm": 1.096108317375183,
      "learning_rate": 0.00019604968791248127,
      "loss": 4.2412,
      "step": 21190
    },
    {
      "epoch": 0.019761830327674684,
      "grad_norm": 1.2947920560836792,
      "learning_rate": 0.00019604782359038608,
      "loss": 4.3817,
      "step": 21200
    },
    {
      "epoch": 0.019771151945753775,
      "grad_norm": 1.1238563060760498,
      "learning_rate": 0.00019604595926829088,
      "loss": 4.3145,
      "step": 21210
    },
    {
      "epoch": 0.01978047356383287,
      "grad_norm": 1.1579463481903076,
      "learning_rate": 0.00019604409494619566,
      "loss": 4.4288,
      "step": 21220
    },
    {
      "epoch": 0.01978979518191196,
      "grad_norm": 1.1436575651168823,
      "learning_rate": 0.00019604223062410047,
      "loss": 4.3805,
      "step": 21230
    },
    {
      "epoch": 0.01979911679999105,
      "grad_norm": 1.1330803632736206,
      "learning_rate": 0.00019604036630200528,
      "loss": 4.2968,
      "step": 21240
    },
    {
      "epoch": 0.019808438418070142,
      "grad_norm": 1.3276994228363037,
      "learning_rate": 0.00019603850197991008,
      "loss": 4.3716,
      "step": 21250
    },
    {
      "epoch": 0.019817760036149237,
      "grad_norm": 1.2746151685714722,
      "learning_rate": 0.0001960366376578149,
      "loss": 4.3489,
      "step": 21260
    },
    {
      "epoch": 0.019827081654228328,
      "grad_norm": 1.054663896560669,
      "learning_rate": 0.00019603477333571967,
      "loss": 4.2683,
      "step": 21270
    },
    {
      "epoch": 0.01983640327230742,
      "grad_norm": 1.130021572113037,
      "learning_rate": 0.00019603290901362447,
      "loss": 4.364,
      "step": 21280
    },
    {
      "epoch": 0.01984572489038651,
      "grad_norm": 1.2551974058151245,
      "learning_rate": 0.00019603104469152928,
      "loss": 4.2897,
      "step": 21290
    },
    {
      "epoch": 0.0198550465084656,
      "grad_norm": 1.1989495754241943,
      "learning_rate": 0.00019602918036943409,
      "loss": 4.4935,
      "step": 21300
    },
    {
      "epoch": 0.019864368126544695,
      "grad_norm": 1.2480759620666504,
      "learning_rate": 0.0001960273160473389,
      "loss": 4.142,
      "step": 21310
    },
    {
      "epoch": 0.019873689744623786,
      "grad_norm": 1.147403597831726,
      "learning_rate": 0.00019602545172524367,
      "loss": 4.2805,
      "step": 21320
    },
    {
      "epoch": 0.019883011362702877,
      "grad_norm": 1.0302525758743286,
      "learning_rate": 0.00019602358740314848,
      "loss": 4.4287,
      "step": 21330
    },
    {
      "epoch": 0.019892332980781968,
      "grad_norm": 1.4924811124801636,
      "learning_rate": 0.00019602172308105328,
      "loss": 4.3534,
      "step": 21340
    },
    {
      "epoch": 0.019901654598861063,
      "grad_norm": 1.0885207653045654,
      "learning_rate": 0.00019601985875895806,
      "loss": 4.3255,
      "step": 21350
    },
    {
      "epoch": 0.019910976216940154,
      "grad_norm": 1.023325800895691,
      "learning_rate": 0.0001960179944368629,
      "loss": 4.3087,
      "step": 21360
    },
    {
      "epoch": 0.019920297835019245,
      "grad_norm": 1.2514736652374268,
      "learning_rate": 0.00019601613011476768,
      "loss": 4.4221,
      "step": 21370
    },
    {
      "epoch": 0.019929619453098336,
      "grad_norm": 1.1758372783660889,
      "learning_rate": 0.00019601426579267248,
      "loss": 4.4069,
      "step": 21380
    },
    {
      "epoch": 0.01993894107117743,
      "grad_norm": 1.1414170265197754,
      "learning_rate": 0.00019601240147057726,
      "loss": 4.2996,
      "step": 21390
    },
    {
      "epoch": 0.01994826268925652,
      "grad_norm": 1.1722439527511597,
      "learning_rate": 0.00019601053714848207,
      "loss": 4.5363,
      "step": 21400
    },
    {
      "epoch": 0.019957584307335612,
      "grad_norm": 1.1774364709854126,
      "learning_rate": 0.00019600867282638687,
      "loss": 4.5449,
      "step": 21410
    },
    {
      "epoch": 0.019966905925414703,
      "grad_norm": 1.1212202310562134,
      "learning_rate": 0.00019600680850429168,
      "loss": 4.1545,
      "step": 21420
    },
    {
      "epoch": 0.019976227543493798,
      "grad_norm": 1.0831328630447388,
      "learning_rate": 0.0001960049441821965,
      "loss": 4.5267,
      "step": 21430
    },
    {
      "epoch": 0.01998554916157289,
      "grad_norm": 1.221522331237793,
      "learning_rate": 0.00019600307986010127,
      "loss": 4.4901,
      "step": 21440
    },
    {
      "epoch": 0.01999487077965198,
      "grad_norm": 0.9726864695549011,
      "learning_rate": 0.00019600121553800607,
      "loss": 4.52,
      "step": 21450
    },
    {
      "epoch": 0.02000419239773107,
      "grad_norm": 1.2875545024871826,
      "learning_rate": 0.00019599935121591088,
      "loss": 4.3398,
      "step": 21460
    },
    {
      "epoch": 0.020013514015810162,
      "grad_norm": 1.308594822883606,
      "learning_rate": 0.00019599748689381568,
      "loss": 4.3841,
      "step": 21470
    },
    {
      "epoch": 0.020022835633889256,
      "grad_norm": 1.2532998323440552,
      "learning_rate": 0.0001959956225717205,
      "loss": 4.1561,
      "step": 21480
    },
    {
      "epoch": 0.020032157251968347,
      "grad_norm": 1.0882916450500488,
      "learning_rate": 0.00019599375824962527,
      "loss": 4.394,
      "step": 21490
    },
    {
      "epoch": 0.02004147887004744,
      "grad_norm": 1.0536552667617798,
      "learning_rate": 0.00019599189392753008,
      "loss": 4.2152,
      "step": 21500
    },
    {
      "epoch": 0.02005080048812653,
      "grad_norm": 1.5155516862869263,
      "learning_rate": 0.00019599002960543488,
      "loss": 4.4663,
      "step": 21510
    },
    {
      "epoch": 0.020060122106205624,
      "grad_norm": 1.1270482540130615,
      "learning_rate": 0.00019598816528333966,
      "loss": 4.2154,
      "step": 21520
    },
    {
      "epoch": 0.020069443724284715,
      "grad_norm": 1.2806158065795898,
      "learning_rate": 0.0001959863009612445,
      "loss": 4.3869,
      "step": 21530
    },
    {
      "epoch": 0.020078765342363806,
      "grad_norm": 1.2688546180725098,
      "learning_rate": 0.00019598443663914927,
      "loss": 4.3199,
      "step": 21540
    },
    {
      "epoch": 0.020088086960442897,
      "grad_norm": 1.1995552778244019,
      "learning_rate": 0.00019598257231705408,
      "loss": 4.3272,
      "step": 21550
    },
    {
      "epoch": 0.02009740857852199,
      "grad_norm": 1.2488222122192383,
      "learning_rate": 0.0001959807079949589,
      "loss": 4.3073,
      "step": 21560
    },
    {
      "epoch": 0.020106730196601082,
      "grad_norm": 1.2569189071655273,
      "learning_rate": 0.00019597884367286367,
      "loss": 4.1949,
      "step": 21570
    },
    {
      "epoch": 0.020116051814680173,
      "grad_norm": 1.0687663555145264,
      "learning_rate": 0.00019597697935076847,
      "loss": 4.3528,
      "step": 21580
    },
    {
      "epoch": 0.020125373432759264,
      "grad_norm": 1.1326096057891846,
      "learning_rate": 0.00019597511502867328,
      "loss": 4.3452,
      "step": 21590
    },
    {
      "epoch": 0.02013469505083836,
      "grad_norm": 1.3631855249404907,
      "learning_rate": 0.00019597325070657809,
      "loss": 3.989,
      "step": 21600
    },
    {
      "epoch": 0.02014401666891745,
      "grad_norm": 1.3452163934707642,
      "learning_rate": 0.0001959713863844829,
      "loss": 4.3201,
      "step": 21610
    },
    {
      "epoch": 0.02015333828699654,
      "grad_norm": 0.9846382737159729,
      "learning_rate": 0.00019596952206238767,
      "loss": 4.4034,
      "step": 21620
    },
    {
      "epoch": 0.020162659905075632,
      "grad_norm": 1.0474683046340942,
      "learning_rate": 0.00019596765774029248,
      "loss": 4.3683,
      "step": 21630
    },
    {
      "epoch": 0.020171981523154723,
      "grad_norm": 1.3494775295257568,
      "learning_rate": 0.00019596579341819728,
      "loss": 4.4141,
      "step": 21640
    },
    {
      "epoch": 0.020181303141233817,
      "grad_norm": 1.205356478691101,
      "learning_rate": 0.0001959639290961021,
      "loss": 4.2706,
      "step": 21650
    },
    {
      "epoch": 0.02019062475931291,
      "grad_norm": 1.2555932998657227,
      "learning_rate": 0.0001959620647740069,
      "loss": 4.3943,
      "step": 21660
    },
    {
      "epoch": 0.020199946377392,
      "grad_norm": 1.085410237312317,
      "learning_rate": 0.00019596020045191168,
      "loss": 4.4155,
      "step": 21670
    },
    {
      "epoch": 0.02020926799547109,
      "grad_norm": 1.2725627422332764,
      "learning_rate": 0.00019595833612981648,
      "loss": 4.4848,
      "step": 21680
    },
    {
      "epoch": 0.020218589613550185,
      "grad_norm": 1.1680797338485718,
      "learning_rate": 0.0001959564718077213,
      "loss": 4.2514,
      "step": 21690
    },
    {
      "epoch": 0.020227911231629276,
      "grad_norm": 1.258533239364624,
      "learning_rate": 0.0001959546074856261,
      "loss": 4.2725,
      "step": 21700
    },
    {
      "epoch": 0.020237232849708367,
      "grad_norm": 1.0629173517227173,
      "learning_rate": 0.0001959527431635309,
      "loss": 4.1452,
      "step": 21710
    },
    {
      "epoch": 0.020246554467787458,
      "grad_norm": 1.2787220478057861,
      "learning_rate": 0.00019595087884143568,
      "loss": 4.4261,
      "step": 21720
    },
    {
      "epoch": 0.020255876085866552,
      "grad_norm": 1.105391025543213,
      "learning_rate": 0.00019594901451934049,
      "loss": 4.531,
      "step": 21730
    },
    {
      "epoch": 0.020265197703945644,
      "grad_norm": 1.1917709112167358,
      "learning_rate": 0.00019594715019724527,
      "loss": 4.5368,
      "step": 21740
    },
    {
      "epoch": 0.020274519322024735,
      "grad_norm": 1.3428925275802612,
      "learning_rate": 0.0001959452858751501,
      "loss": 4.6356,
      "step": 21750
    },
    {
      "epoch": 0.020283840940103826,
      "grad_norm": 1.351650357246399,
      "learning_rate": 0.0001959434215530549,
      "loss": 4.3968,
      "step": 21760
    },
    {
      "epoch": 0.02029316255818292,
      "grad_norm": 1.2394055128097534,
      "learning_rate": 0.00019594155723095968,
      "loss": 4.4615,
      "step": 21770
    },
    {
      "epoch": 0.02030248417626201,
      "grad_norm": 1.121037244796753,
      "learning_rate": 0.0001959396929088645,
      "loss": 4.3481,
      "step": 21780
    },
    {
      "epoch": 0.020311805794341102,
      "grad_norm": 1.3296763896942139,
      "learning_rate": 0.00019593782858676927,
      "loss": 4.3788,
      "step": 21790
    },
    {
      "epoch": 0.020321127412420193,
      "grad_norm": 1.2629566192626953,
      "learning_rate": 0.00019593596426467408,
      "loss": 4.3582,
      "step": 21800
    },
    {
      "epoch": 0.020330449030499284,
      "grad_norm": 1.1138421297073364,
      "learning_rate": 0.0001959340999425789,
      "loss": 4.3736,
      "step": 21810
    },
    {
      "epoch": 0.02033977064857838,
      "grad_norm": 1.1750727891921997,
      "learning_rate": 0.0001959322356204837,
      "loss": 4.3811,
      "step": 21820
    },
    {
      "epoch": 0.02034909226665747,
      "grad_norm": 1.468910574913025,
      "learning_rate": 0.0001959303712983885,
      "loss": 4.1669,
      "step": 21830
    },
    {
      "epoch": 0.02035841388473656,
      "grad_norm": 1.198578953742981,
      "learning_rate": 0.00019592850697629327,
      "loss": 4.4037,
      "step": 21840
    },
    {
      "epoch": 0.02036773550281565,
      "grad_norm": 1.1169637441635132,
      "learning_rate": 0.00019592664265419808,
      "loss": 4.2736,
      "step": 21850
    },
    {
      "epoch": 0.020377057120894746,
      "grad_norm": 1.2395766973495483,
      "learning_rate": 0.0001959247783321029,
      "loss": 4.133,
      "step": 21860
    },
    {
      "epoch": 0.020386378738973837,
      "grad_norm": 1.220430612564087,
      "learning_rate": 0.0001959229140100077,
      "loss": 4.4267,
      "step": 21870
    },
    {
      "epoch": 0.020395700357052928,
      "grad_norm": 1.1775178909301758,
      "learning_rate": 0.0001959210496879125,
      "loss": 4.4765,
      "step": 21880
    },
    {
      "epoch": 0.02040502197513202,
      "grad_norm": 1.2147761583328247,
      "learning_rate": 0.00019591918536581728,
      "loss": 4.2999,
      "step": 21890
    },
    {
      "epoch": 0.020414343593211114,
      "grad_norm": 1.0393234491348267,
      "learning_rate": 0.00019591732104372209,
      "loss": 4.3651,
      "step": 21900
    },
    {
      "epoch": 0.020423665211290205,
      "grad_norm": 1.0905479192733765,
      "learning_rate": 0.0001959154567216269,
      "loss": 4.3058,
      "step": 21910
    },
    {
      "epoch": 0.020432986829369296,
      "grad_norm": 1.1435601711273193,
      "learning_rate": 0.0001959135923995317,
      "loss": 4.3289,
      "step": 21920
    },
    {
      "epoch": 0.020442308447448387,
      "grad_norm": 1.082533597946167,
      "learning_rate": 0.0001959117280774365,
      "loss": 4.5697,
      "step": 21930
    },
    {
      "epoch": 0.02045163006552748,
      "grad_norm": 1.1867456436157227,
      "learning_rate": 0.00019590986375534128,
      "loss": 4.1737,
      "step": 21940
    },
    {
      "epoch": 0.020460951683606572,
      "grad_norm": 1.1923692226409912,
      "learning_rate": 0.0001959079994332461,
      "loss": 4.3779,
      "step": 21950
    },
    {
      "epoch": 0.020470273301685663,
      "grad_norm": 1.093011498451233,
      "learning_rate": 0.0001959061351111509,
      "loss": 4.4719,
      "step": 21960
    },
    {
      "epoch": 0.020479594919764754,
      "grad_norm": 1.0632506608963013,
      "learning_rate": 0.00019590427078905568,
      "loss": 4.4345,
      "step": 21970
    },
    {
      "epoch": 0.020488916537843845,
      "grad_norm": 1.4903132915496826,
      "learning_rate": 0.0001959024064669605,
      "loss": 4.3393,
      "step": 21980
    },
    {
      "epoch": 0.02049823815592294,
      "grad_norm": 1.016008734703064,
      "learning_rate": 0.0001959005421448653,
      "loss": 4.3811,
      "step": 21990
    },
    {
      "epoch": 0.02050755977400203,
      "grad_norm": 1.1161080598831177,
      "learning_rate": 0.0001958986778227701,
      "loss": 4.2842,
      "step": 22000
    },
    {
      "epoch": 0.020516881392081122,
      "grad_norm": 1.3601895570755005,
      "learning_rate": 0.0001958968135006749,
      "loss": 4.3625,
      "step": 22010
    },
    {
      "epoch": 0.020526203010160213,
      "grad_norm": 1.1022660732269287,
      "learning_rate": 0.00019589494917857968,
      "loss": 4.2812,
      "step": 22020
    },
    {
      "epoch": 0.020535524628239307,
      "grad_norm": 1.231083631515503,
      "learning_rate": 0.00019589308485648449,
      "loss": 4.1614,
      "step": 22030
    },
    {
      "epoch": 0.020544846246318398,
      "grad_norm": 1.3098722696304321,
      "learning_rate": 0.0001958912205343893,
      "loss": 4.3773,
      "step": 22040
    },
    {
      "epoch": 0.02055416786439749,
      "grad_norm": 1.0540385246276855,
      "learning_rate": 0.0001958893562122941,
      "loss": 4.4868,
      "step": 22050
    },
    {
      "epoch": 0.02056348948247658,
      "grad_norm": 1.1782798767089844,
      "learning_rate": 0.0001958874918901989,
      "loss": 4.4079,
      "step": 22060
    },
    {
      "epoch": 0.020572811100555675,
      "grad_norm": 1.1417455673217773,
      "learning_rate": 0.00019588562756810368,
      "loss": 4.3763,
      "step": 22070
    },
    {
      "epoch": 0.020582132718634766,
      "grad_norm": 1.356848955154419,
      "learning_rate": 0.0001958837632460085,
      "loss": 4.4198,
      "step": 22080
    },
    {
      "epoch": 0.020591454336713857,
      "grad_norm": 1.0392814874649048,
      "learning_rate": 0.0001958818989239133,
      "loss": 4.3202,
      "step": 22090
    },
    {
      "epoch": 0.020600775954792948,
      "grad_norm": 1.3968662023544312,
      "learning_rate": 0.0001958800346018181,
      "loss": 4.2806,
      "step": 22100
    },
    {
      "epoch": 0.020610097572872042,
      "grad_norm": 1.2545632123947144,
      "learning_rate": 0.0001958781702797229,
      "loss": 4.5156,
      "step": 22110
    },
    {
      "epoch": 0.020619419190951133,
      "grad_norm": 1.1309444904327393,
      "learning_rate": 0.0001958763059576277,
      "loss": 4.2696,
      "step": 22120
    },
    {
      "epoch": 0.020628740809030224,
      "grad_norm": 1.054134488105774,
      "learning_rate": 0.0001958744416355325,
      "loss": 4.231,
      "step": 22130
    },
    {
      "epoch": 0.020638062427109315,
      "grad_norm": 1.4579472541809082,
      "learning_rate": 0.0001958725773134373,
      "loss": 4.2741,
      "step": 22140
    },
    {
      "epoch": 0.020647384045188406,
      "grad_norm": 1.281370997428894,
      "learning_rate": 0.0001958707129913421,
      "loss": 4.3361,
      "step": 22150
    },
    {
      "epoch": 0.0206567056632675,
      "grad_norm": 1.1744506359100342,
      "learning_rate": 0.00019586884866924691,
      "loss": 4.3564,
      "step": 22160
    },
    {
      "epoch": 0.020666027281346592,
      "grad_norm": 1.0106309652328491,
      "learning_rate": 0.0001958669843471517,
      "loss": 4.3045,
      "step": 22170
    },
    {
      "epoch": 0.020675348899425683,
      "grad_norm": 1.2069554328918457,
      "learning_rate": 0.0001958651200250565,
      "loss": 4.4231,
      "step": 22180
    },
    {
      "epoch": 0.020684670517504774,
      "grad_norm": 1.128846287727356,
      "learning_rate": 0.00019586325570296128,
      "loss": 4.2284,
      "step": 22190
    },
    {
      "epoch": 0.02069399213558387,
      "grad_norm": 1.2167115211486816,
      "learning_rate": 0.0001958613913808661,
      "loss": 4.2763,
      "step": 22200
    },
    {
      "epoch": 0.02070331375366296,
      "grad_norm": 1.2372170686721802,
      "learning_rate": 0.00019585952705877092,
      "loss": 4.5174,
      "step": 22210
    },
    {
      "epoch": 0.02071263537174205,
      "grad_norm": 1.0909888744354248,
      "learning_rate": 0.0001958576627366757,
      "loss": 4.1874,
      "step": 22220
    },
    {
      "epoch": 0.02072195698982114,
      "grad_norm": 1.2858998775482178,
      "learning_rate": 0.0001958557984145805,
      "loss": 4.2609,
      "step": 22230
    },
    {
      "epoch": 0.020731278607900236,
      "grad_norm": 1.1102149486541748,
      "learning_rate": 0.00019585393409248528,
      "loss": 4.4729,
      "step": 22240
    },
    {
      "epoch": 0.020740600225979327,
      "grad_norm": 1.1349430084228516,
      "learning_rate": 0.0001958520697703901,
      "loss": 4.4528,
      "step": 22250
    },
    {
      "epoch": 0.020749921844058418,
      "grad_norm": 1.0845890045166016,
      "learning_rate": 0.00019585020544829492,
      "loss": 4.4273,
      "step": 22260
    },
    {
      "epoch": 0.02075924346213751,
      "grad_norm": 1.2319930791854858,
      "learning_rate": 0.0001958483411261997,
      "loss": 4.4818,
      "step": 22270
    },
    {
      "epoch": 0.020768565080216603,
      "grad_norm": 1.229142189025879,
      "learning_rate": 0.0001958464768041045,
      "loss": 4.3634,
      "step": 22280
    },
    {
      "epoch": 0.020777886698295694,
      "grad_norm": 1.2006539106369019,
      "learning_rate": 0.0001958446124820093,
      "loss": 4.1156,
      "step": 22290
    },
    {
      "epoch": 0.020787208316374785,
      "grad_norm": 1.146466851234436,
      "learning_rate": 0.0001958427481599141,
      "loss": 4.3,
      "step": 22300
    },
    {
      "epoch": 0.020796529934453876,
      "grad_norm": 1.1792181730270386,
      "learning_rate": 0.0001958408838378189,
      "loss": 4.3164,
      "step": 22310
    },
    {
      "epoch": 0.020805851552532968,
      "grad_norm": 1.2763291597366333,
      "learning_rate": 0.0001958390195157237,
      "loss": 4.3822,
      "step": 22320
    },
    {
      "epoch": 0.020815173170612062,
      "grad_norm": 0.9975064396858215,
      "learning_rate": 0.0001958371551936285,
      "loss": 4.286,
      "step": 22330
    },
    {
      "epoch": 0.020824494788691153,
      "grad_norm": 1.282663345336914,
      "learning_rate": 0.0001958352908715333,
      "loss": 4.3912,
      "step": 22340
    },
    {
      "epoch": 0.020833816406770244,
      "grad_norm": 1.4500020742416382,
      "learning_rate": 0.0001958334265494381,
      "loss": 4.4238,
      "step": 22350
    },
    {
      "epoch": 0.020843138024849335,
      "grad_norm": 1.3177235126495361,
      "learning_rate": 0.0001958315622273429,
      "loss": 4.1981,
      "step": 22360
    },
    {
      "epoch": 0.02085245964292843,
      "grad_norm": 1.0146666765213013,
      "learning_rate": 0.0001958296979052477,
      "loss": 4.4932,
      "step": 22370
    },
    {
      "epoch": 0.02086178126100752,
      "grad_norm": 1.3934416770935059,
      "learning_rate": 0.00019582783358315252,
      "loss": 4.2317,
      "step": 22380
    },
    {
      "epoch": 0.02087110287908661,
      "grad_norm": 1.099973440170288,
      "learning_rate": 0.0001958259692610573,
      "loss": 4.4045,
      "step": 22390
    },
    {
      "epoch": 0.020880424497165703,
      "grad_norm": 1.038609504699707,
      "learning_rate": 0.0001958241049389621,
      "loss": 4.3707,
      "step": 22400
    },
    {
      "epoch": 0.020889746115244797,
      "grad_norm": 1.0610333681106567,
      "learning_rate": 0.0001958222406168669,
      "loss": 4.3803,
      "step": 22410
    },
    {
      "epoch": 0.020899067733323888,
      "grad_norm": 1.3954745531082153,
      "learning_rate": 0.0001958203762947717,
      "loss": 4.2503,
      "step": 22420
    },
    {
      "epoch": 0.02090838935140298,
      "grad_norm": 1.050191044807434,
      "learning_rate": 0.00019581851197267652,
      "loss": 4.5131,
      "step": 22430
    },
    {
      "epoch": 0.02091771096948207,
      "grad_norm": 1.0322703123092651,
      "learning_rate": 0.0001958166476505813,
      "loss": 4.3108,
      "step": 22440
    },
    {
      "epoch": 0.020927032587561165,
      "grad_norm": 1.1320111751556396,
      "learning_rate": 0.0001958147833284861,
      "loss": 4.6039,
      "step": 22450
    },
    {
      "epoch": 0.020936354205640256,
      "grad_norm": 1.1915960311889648,
      "learning_rate": 0.0001958129190063909,
      "loss": 4.313,
      "step": 22460
    },
    {
      "epoch": 0.020945675823719347,
      "grad_norm": 1.1679699420928955,
      "learning_rate": 0.0001958110546842957,
      "loss": 4.4126,
      "step": 22470
    },
    {
      "epoch": 0.020954997441798438,
      "grad_norm": 1.1587450504302979,
      "learning_rate": 0.0001958091903622005,
      "loss": 4.4677,
      "step": 22480
    },
    {
      "epoch": 0.02096431905987753,
      "grad_norm": 1.2362205982208252,
      "learning_rate": 0.0001958073260401053,
      "loss": 4.332,
      "step": 22490
    },
    {
      "epoch": 0.020973640677956623,
      "grad_norm": 1.2248927354812622,
      "learning_rate": 0.0001958054617180101,
      "loss": 4.2081,
      "step": 22500
    },
    {
      "epoch": 0.020982962296035714,
      "grad_norm": 1.209354043006897,
      "learning_rate": 0.00019580359739591492,
      "loss": 4.4512,
      "step": 22510
    },
    {
      "epoch": 0.020992283914114805,
      "grad_norm": 1.250347375869751,
      "learning_rate": 0.0001958017330738197,
      "loss": 4.3499,
      "step": 22520
    },
    {
      "epoch": 0.021001605532193896,
      "grad_norm": 1.0463215112686157,
      "learning_rate": 0.0001957998687517245,
      "loss": 4.4478,
      "step": 22530
    },
    {
      "epoch": 0.02101092715027299,
      "grad_norm": 1.1715422868728638,
      "learning_rate": 0.0001957980044296293,
      "loss": 4.4219,
      "step": 22540
    },
    {
      "epoch": 0.02102024876835208,
      "grad_norm": 1.0801199674606323,
      "learning_rate": 0.00019579614010753412,
      "loss": 4.2392,
      "step": 22550
    },
    {
      "epoch": 0.021029570386431173,
      "grad_norm": 1.160889983177185,
      "learning_rate": 0.00019579427578543892,
      "loss": 4.4334,
      "step": 22560
    },
    {
      "epoch": 0.021038892004510264,
      "grad_norm": 1.3126153945922852,
      "learning_rate": 0.0001957924114633437,
      "loss": 4.3768,
      "step": 22570
    },
    {
      "epoch": 0.021048213622589358,
      "grad_norm": 1.4406696557998657,
      "learning_rate": 0.0001957905471412485,
      "loss": 4.2523,
      "step": 22580
    },
    {
      "epoch": 0.02105753524066845,
      "grad_norm": 1.1441829204559326,
      "learning_rate": 0.00019578868281915331,
      "loss": 4.189,
      "step": 22590
    },
    {
      "epoch": 0.02106685685874754,
      "grad_norm": 1.0588065385818481,
      "learning_rate": 0.00019578681849705812,
      "loss": 4.3483,
      "step": 22600
    },
    {
      "epoch": 0.02107617847682663,
      "grad_norm": 1.1326940059661865,
      "learning_rate": 0.00019578495417496293,
      "loss": 4.3664,
      "step": 22610
    },
    {
      "epoch": 0.021085500094905726,
      "grad_norm": 1.2688612937927246,
      "learning_rate": 0.0001957830898528677,
      "loss": 4.5083,
      "step": 22620
    },
    {
      "epoch": 0.021094821712984817,
      "grad_norm": 1.1714556217193604,
      "learning_rate": 0.0001957812255307725,
      "loss": 4.4078,
      "step": 22630
    },
    {
      "epoch": 0.021104143331063908,
      "grad_norm": 1.1447511911392212,
      "learning_rate": 0.0001957793612086773,
      "loss": 4.3058,
      "step": 22640
    },
    {
      "epoch": 0.021113464949143,
      "grad_norm": 1.0545666217803955,
      "learning_rate": 0.00019577749688658212,
      "loss": 4.4685,
      "step": 22650
    },
    {
      "epoch": 0.02112278656722209,
      "grad_norm": 1.1733225584030151,
      "learning_rate": 0.00019577563256448693,
      "loss": 4.3068,
      "step": 22660
    },
    {
      "epoch": 0.021132108185301184,
      "grad_norm": 0.9371938705444336,
      "learning_rate": 0.0001957737682423917,
      "loss": 4.3523,
      "step": 22670
    },
    {
      "epoch": 0.021141429803380275,
      "grad_norm": 1.4205223321914673,
      "learning_rate": 0.00019577190392029652,
      "loss": 4.2802,
      "step": 22680
    },
    {
      "epoch": 0.021150751421459366,
      "grad_norm": 0.9921582341194153,
      "learning_rate": 0.0001957700395982013,
      "loss": 4.4632,
      "step": 22690
    },
    {
      "epoch": 0.021160073039538457,
      "grad_norm": 1.1068817377090454,
      "learning_rate": 0.0001957681752761061,
      "loss": 4.3531,
      "step": 22700
    },
    {
      "epoch": 0.021169394657617552,
      "grad_norm": 1.0341203212738037,
      "learning_rate": 0.00019576631095401094,
      "loss": 4.3005,
      "step": 22710
    },
    {
      "epoch": 0.021178716275696643,
      "grad_norm": 1.176903486251831,
      "learning_rate": 0.00019576444663191571,
      "loss": 4.4922,
      "step": 22720
    },
    {
      "epoch": 0.021188037893775734,
      "grad_norm": 1.1758729219436646,
      "learning_rate": 0.00019576258230982052,
      "loss": 4.3545,
      "step": 22730
    },
    {
      "epoch": 0.021197359511854825,
      "grad_norm": 1.1058979034423828,
      "learning_rate": 0.0001957607179877253,
      "loss": 4.3821,
      "step": 22740
    },
    {
      "epoch": 0.02120668112993392,
      "grad_norm": 1.0586563348770142,
      "learning_rate": 0.0001957588536656301,
      "loss": 4.3633,
      "step": 22750
    },
    {
      "epoch": 0.02121600274801301,
      "grad_norm": 1.177322506904602,
      "learning_rate": 0.0001957569893435349,
      "loss": 4.2011,
      "step": 22760
    },
    {
      "epoch": 0.0212253243660921,
      "grad_norm": 1.041520357131958,
      "learning_rate": 0.00019575512502143972,
      "loss": 4.4748,
      "step": 22770
    },
    {
      "epoch": 0.021234645984171192,
      "grad_norm": 1.1481729745864868,
      "learning_rate": 0.00019575326069934453,
      "loss": 4.564,
      "step": 22780
    },
    {
      "epoch": 0.021243967602250287,
      "grad_norm": 1.1243928670883179,
      "learning_rate": 0.0001957513963772493,
      "loss": 4.1662,
      "step": 22790
    },
    {
      "epoch": 0.021253289220329378,
      "grad_norm": 1.247380256652832,
      "learning_rate": 0.0001957495320551541,
      "loss": 4.4606,
      "step": 22800
    },
    {
      "epoch": 0.02126261083840847,
      "grad_norm": 1.0726087093353271,
      "learning_rate": 0.00019574766773305892,
      "loss": 4.4134,
      "step": 22810
    },
    {
      "epoch": 0.02127193245648756,
      "grad_norm": 1.1624679565429688,
      "learning_rate": 0.00019574580341096372,
      "loss": 4.124,
      "step": 22820
    },
    {
      "epoch": 0.02128125407456665,
      "grad_norm": 1.142876386642456,
      "learning_rate": 0.00019574393908886853,
      "loss": 4.3441,
      "step": 22830
    },
    {
      "epoch": 0.021290575692645745,
      "grad_norm": 1.088793158531189,
      "learning_rate": 0.0001957420747667733,
      "loss": 4.4302,
      "step": 22840
    },
    {
      "epoch": 0.021299897310724836,
      "grad_norm": 1.1302547454833984,
      "learning_rate": 0.00019574021044467812,
      "loss": 4.342,
      "step": 22850
    },
    {
      "epoch": 0.021309218928803927,
      "grad_norm": 1.3021864891052246,
      "learning_rate": 0.0001957383461225829,
      "loss": 4.2718,
      "step": 22860
    },
    {
      "epoch": 0.02131854054688302,
      "grad_norm": 1.0570056438446045,
      "learning_rate": 0.0001957364818004877,
      "loss": 4.2734,
      "step": 22870
    },
    {
      "epoch": 0.021327862164962113,
      "grad_norm": 1.3651119470596313,
      "learning_rate": 0.00019573461747839253,
      "loss": 4.3779,
      "step": 22880
    },
    {
      "epoch": 0.021337183783041204,
      "grad_norm": 1.215494990348816,
      "learning_rate": 0.0001957327531562973,
      "loss": 4.2199,
      "step": 22890
    },
    {
      "epoch": 0.021346505401120295,
      "grad_norm": 1.193767786026001,
      "learning_rate": 0.00019573088883420212,
      "loss": 4.5314,
      "step": 22900
    },
    {
      "epoch": 0.021355827019199386,
      "grad_norm": 1.5332492589950562,
      "learning_rate": 0.0001957290245121069,
      "loss": 4.3621,
      "step": 22910
    },
    {
      "epoch": 0.02136514863727848,
      "grad_norm": 1.0002565383911133,
      "learning_rate": 0.0001957271601900117,
      "loss": 4.378,
      "step": 22920
    },
    {
      "epoch": 0.02137447025535757,
      "grad_norm": 1.0564980506896973,
      "learning_rate": 0.0001957252958679165,
      "loss": 4.1586,
      "step": 22930
    },
    {
      "epoch": 0.021383791873436662,
      "grad_norm": 1.069863200187683,
      "learning_rate": 0.00019572343154582132,
      "loss": 4.3118,
      "step": 22940
    },
    {
      "epoch": 0.021393113491515754,
      "grad_norm": 1.3524945974349976,
      "learning_rate": 0.00019572156722372612,
      "loss": 4.4026,
      "step": 22950
    },
    {
      "epoch": 0.021402435109594848,
      "grad_norm": 1.0274730920791626,
      "learning_rate": 0.0001957197029016309,
      "loss": 4.3244,
      "step": 22960
    },
    {
      "epoch": 0.02141175672767394,
      "grad_norm": 1.2091405391693115,
      "learning_rate": 0.0001957178385795357,
      "loss": 4.2209,
      "step": 22970
    },
    {
      "epoch": 0.02142107834575303,
      "grad_norm": 1.244446039199829,
      "learning_rate": 0.00019571597425744052,
      "loss": 4.4118,
      "step": 22980
    },
    {
      "epoch": 0.02143039996383212,
      "grad_norm": 1.2332650423049927,
      "learning_rate": 0.00019571410993534532,
      "loss": 4.438,
      "step": 22990
    },
    {
      "epoch": 0.021439721581911212,
      "grad_norm": 1.2506284713745117,
      "learning_rate": 0.00019571224561325013,
      "loss": 4.2826,
      "step": 23000
    },
    {
      "epoch": 0.021449043199990307,
      "grad_norm": 1.0366930961608887,
      "learning_rate": 0.0001957103812911549,
      "loss": 4.4422,
      "step": 23010
    },
    {
      "epoch": 0.021458364818069398,
      "grad_norm": 1.1295262575149536,
      "learning_rate": 0.00019570851696905971,
      "loss": 4.3729,
      "step": 23020
    },
    {
      "epoch": 0.02146768643614849,
      "grad_norm": 1.1603447198867798,
      "learning_rate": 0.00019570665264696452,
      "loss": 4.1781,
      "step": 23030
    },
    {
      "epoch": 0.02147700805422758,
      "grad_norm": 1.384962797164917,
      "learning_rate": 0.00019570478832486933,
      "loss": 4.4346,
      "step": 23040
    },
    {
      "epoch": 0.021486329672306674,
      "grad_norm": 1.0860735177993774,
      "learning_rate": 0.00019570292400277413,
      "loss": 4.2271,
      "step": 23050
    },
    {
      "epoch": 0.021495651290385765,
      "grad_norm": 1.0921931266784668,
      "learning_rate": 0.0001957010596806789,
      "loss": 4.4088,
      "step": 23060
    },
    {
      "epoch": 0.021504972908464856,
      "grad_norm": 1.2638312578201294,
      "learning_rate": 0.00019569919535858372,
      "loss": 4.1269,
      "step": 23070
    },
    {
      "epoch": 0.021514294526543947,
      "grad_norm": 1.1713136434555054,
      "learning_rate": 0.00019569733103648852,
      "loss": 4.3904,
      "step": 23080
    },
    {
      "epoch": 0.02152361614462304,
      "grad_norm": 1.0495378971099854,
      "learning_rate": 0.0001956954667143933,
      "loss": 4.4027,
      "step": 23090
    },
    {
      "epoch": 0.021532937762702133,
      "grad_norm": 1.087731122970581,
      "learning_rate": 0.00019569360239229814,
      "loss": 4.2414,
      "step": 23100
    },
    {
      "epoch": 0.021542259380781224,
      "grad_norm": 1.2027287483215332,
      "learning_rate": 0.00019569173807020292,
      "loss": 4.3103,
      "step": 23110
    },
    {
      "epoch": 0.021551580998860315,
      "grad_norm": 0.9580854177474976,
      "learning_rate": 0.00019568987374810772,
      "loss": 4.5695,
      "step": 23120
    },
    {
      "epoch": 0.02156090261693941,
      "grad_norm": 1.0797076225280762,
      "learning_rate": 0.00019568800942601253,
      "loss": 4.3815,
      "step": 23130
    },
    {
      "epoch": 0.0215702242350185,
      "grad_norm": 1.2214972972869873,
      "learning_rate": 0.0001956861451039173,
      "loss": 4.3086,
      "step": 23140
    },
    {
      "epoch": 0.02157954585309759,
      "grad_norm": 1.0500385761260986,
      "learning_rate": 0.00019568428078182211,
      "loss": 4.2282,
      "step": 23150
    },
    {
      "epoch": 0.021588867471176682,
      "grad_norm": 1.1141273975372314,
      "learning_rate": 0.00019568241645972692,
      "loss": 4.4624,
      "step": 23160
    },
    {
      "epoch": 0.021598189089255773,
      "grad_norm": 1.1936089992523193,
      "learning_rate": 0.00019568055213763173,
      "loss": 4.1735,
      "step": 23170
    },
    {
      "epoch": 0.021607510707334868,
      "grad_norm": 0.9995372891426086,
      "learning_rate": 0.00019567868781553653,
      "loss": 4.4246,
      "step": 23180
    },
    {
      "epoch": 0.02161683232541396,
      "grad_norm": 0.9576570987701416,
      "learning_rate": 0.0001956768234934413,
      "loss": 4.4336,
      "step": 23190
    },
    {
      "epoch": 0.02162615394349305,
      "grad_norm": 1.122887134552002,
      "learning_rate": 0.00019567495917134612,
      "loss": 4.3382,
      "step": 23200
    },
    {
      "epoch": 0.02163547556157214,
      "grad_norm": 1.1719300746917725,
      "learning_rate": 0.00019567309484925093,
      "loss": 4.229,
      "step": 23210
    },
    {
      "epoch": 0.021644797179651235,
      "grad_norm": 1.1572595834732056,
      "learning_rate": 0.00019567123052715573,
      "loss": 4.1337,
      "step": 23220
    },
    {
      "epoch": 0.021654118797730326,
      "grad_norm": 1.3982733488082886,
      "learning_rate": 0.00019566936620506054,
      "loss": 4.362,
      "step": 23230
    },
    {
      "epoch": 0.021663440415809417,
      "grad_norm": 1.1879771947860718,
      "learning_rate": 0.00019566750188296532,
      "loss": 4.2807,
      "step": 23240
    },
    {
      "epoch": 0.021672762033888508,
      "grad_norm": 1.1324424743652344,
      "learning_rate": 0.00019566563756087012,
      "loss": 4.2227,
      "step": 23250
    },
    {
      "epoch": 0.021682083651967603,
      "grad_norm": 1.1283740997314453,
      "learning_rate": 0.00019566377323877493,
      "loss": 4.3373,
      "step": 23260
    },
    {
      "epoch": 0.021691405270046694,
      "grad_norm": 1.081175446510315,
      "learning_rate": 0.00019566190891667974,
      "loss": 4.5048,
      "step": 23270
    },
    {
      "epoch": 0.021700726888125785,
      "grad_norm": 1.031701683998108,
      "learning_rate": 0.00019566004459458454,
      "loss": 4.515,
      "step": 23280
    },
    {
      "epoch": 0.021710048506204876,
      "grad_norm": 1.1816397905349731,
      "learning_rate": 0.00019565818027248932,
      "loss": 4.1531,
      "step": 23290
    },
    {
      "epoch": 0.02171937012428397,
      "grad_norm": 1.1420246362686157,
      "learning_rate": 0.00019565631595039413,
      "loss": 4.3079,
      "step": 23300
    },
    {
      "epoch": 0.02172869174236306,
      "grad_norm": 0.994046688079834,
      "learning_rate": 0.0001956544516282989,
      "loss": 4.3041,
      "step": 23310
    },
    {
      "epoch": 0.021738013360442152,
      "grad_norm": 0.9922618865966797,
      "learning_rate": 0.00019565258730620374,
      "loss": 4.1258,
      "step": 23320
    },
    {
      "epoch": 0.021747334978521243,
      "grad_norm": 0.9681622982025146,
      "learning_rate": 0.00019565072298410855,
      "loss": 4.3658,
      "step": 23330
    },
    {
      "epoch": 0.021756656596600334,
      "grad_norm": 1.1191632747650146,
      "learning_rate": 0.00019564885866201333,
      "loss": 4.4,
      "step": 23340
    },
    {
      "epoch": 0.02176597821467943,
      "grad_norm": 1.1686795949935913,
      "learning_rate": 0.00019564699433991813,
      "loss": 4.2757,
      "step": 23350
    },
    {
      "epoch": 0.02177529983275852,
      "grad_norm": 1.3267126083374023,
      "learning_rate": 0.0001956451300178229,
      "loss": 4.5468,
      "step": 23360
    },
    {
      "epoch": 0.02178462145083761,
      "grad_norm": 1.2694510221481323,
      "learning_rate": 0.00019564326569572772,
      "loss": 4.3644,
      "step": 23370
    },
    {
      "epoch": 0.021793943068916702,
      "grad_norm": 1.2605597972869873,
      "learning_rate": 0.00019564140137363255,
      "loss": 4.1651,
      "step": 23380
    },
    {
      "epoch": 0.021803264686995796,
      "grad_norm": 1.2580034732818604,
      "learning_rate": 0.00019563953705153733,
      "loss": 4.3919,
      "step": 23390
    },
    {
      "epoch": 0.021812586305074887,
      "grad_norm": 1.1390434503555298,
      "learning_rate": 0.00019563767272944214,
      "loss": 4.3047,
      "step": 23400
    },
    {
      "epoch": 0.02182190792315398,
      "grad_norm": 0.9882580637931824,
      "learning_rate": 0.00019563580840734692,
      "loss": 4.4867,
      "step": 23410
    },
    {
      "epoch": 0.02183122954123307,
      "grad_norm": 1.0247985124588013,
      "learning_rate": 0.00019563394408525172,
      "loss": 4.2605,
      "step": 23420
    },
    {
      "epoch": 0.021840551159312164,
      "grad_norm": 0.9630834460258484,
      "learning_rate": 0.00019563207976315653,
      "loss": 4.3197,
      "step": 23430
    },
    {
      "epoch": 0.021849872777391255,
      "grad_norm": 1.1776328086853027,
      "learning_rate": 0.00019563021544106134,
      "loss": 4.4365,
      "step": 23440
    },
    {
      "epoch": 0.021859194395470346,
      "grad_norm": 1.0146737098693848,
      "learning_rate": 0.00019562835111896614,
      "loss": 4.3941,
      "step": 23450
    },
    {
      "epoch": 0.021868516013549437,
      "grad_norm": 0.8386724591255188,
      "learning_rate": 0.00019562648679687092,
      "loss": 4.4508,
      "step": 23460
    },
    {
      "epoch": 0.02187783763162853,
      "grad_norm": 1.36872398853302,
      "learning_rate": 0.00019562462247477573,
      "loss": 4.4484,
      "step": 23470
    },
    {
      "epoch": 0.021887159249707622,
      "grad_norm": 1.0963470935821533,
      "learning_rate": 0.00019562275815268053,
      "loss": 4.2349,
      "step": 23480
    },
    {
      "epoch": 0.021896480867786713,
      "grad_norm": 1.224507451057434,
      "learning_rate": 0.00019562089383058534,
      "loss": 4.1777,
      "step": 23490
    },
    {
      "epoch": 0.021905802485865804,
      "grad_norm": 1.1057335138320923,
      "learning_rate": 0.00019561902950849015,
      "loss": 4.445,
      "step": 23500
    },
    {
      "epoch": 0.021915124103944895,
      "grad_norm": 1.110528826713562,
      "learning_rate": 0.00019561716518639493,
      "loss": 4.3444,
      "step": 23510
    },
    {
      "epoch": 0.02192444572202399,
      "grad_norm": 1.0354865789413452,
      "learning_rate": 0.00019561530086429973,
      "loss": 4.2058,
      "step": 23520
    },
    {
      "epoch": 0.02193376734010308,
      "grad_norm": 1.168091893196106,
      "learning_rate": 0.00019561343654220454,
      "loss": 4.2077,
      "step": 23530
    },
    {
      "epoch": 0.021943088958182172,
      "grad_norm": 1.3559335470199585,
      "learning_rate": 0.00019561157222010932,
      "loss": 4.2089,
      "step": 23540
    },
    {
      "epoch": 0.021952410576261263,
      "grad_norm": 0.9781034588813782,
      "learning_rate": 0.00019560970789801415,
      "loss": 4.3633,
      "step": 23550
    },
    {
      "epoch": 0.021961732194340357,
      "grad_norm": 1.2421865463256836,
      "learning_rate": 0.00019560784357591893,
      "loss": 4.4835,
      "step": 23560
    },
    {
      "epoch": 0.02197105381241945,
      "grad_norm": 1.1602401733398438,
      "learning_rate": 0.00019560597925382374,
      "loss": 4.0551,
      "step": 23570
    },
    {
      "epoch": 0.02198037543049854,
      "grad_norm": 0.9430173635482788,
      "learning_rate": 0.00019560411493172854,
      "loss": 4.5029,
      "step": 23580
    },
    {
      "epoch": 0.02198969704857763,
      "grad_norm": 1.199337124824524,
      "learning_rate": 0.00019560225060963332,
      "loss": 4.5109,
      "step": 23590
    },
    {
      "epoch": 0.021999018666656725,
      "grad_norm": 1.2469041347503662,
      "learning_rate": 0.00019560038628753813,
      "loss": 4.3539,
      "step": 23600
    },
    {
      "epoch": 0.022008340284735816,
      "grad_norm": 1.1198428869247437,
      "learning_rate": 0.00019559852196544293,
      "loss": 4.5376,
      "step": 23610
    },
    {
      "epoch": 0.022017661902814907,
      "grad_norm": 0.9276149868965149,
      "learning_rate": 0.00019559665764334774,
      "loss": 4.3774,
      "step": 23620
    },
    {
      "epoch": 0.022026983520893998,
      "grad_norm": 1.3629865646362305,
      "learning_rate": 0.00019559479332125255,
      "loss": 4.3787,
      "step": 23630
    },
    {
      "epoch": 0.022036305138973093,
      "grad_norm": 1.1303437948226929,
      "learning_rate": 0.00019559292899915733,
      "loss": 4.3445,
      "step": 23640
    },
    {
      "epoch": 0.022045626757052184,
      "grad_norm": 1.174798846244812,
      "learning_rate": 0.00019559106467706213,
      "loss": 4.1758,
      "step": 23650
    },
    {
      "epoch": 0.022054948375131275,
      "grad_norm": 1.1835523843765259,
      "learning_rate": 0.00019558920035496694,
      "loss": 4.3008,
      "step": 23660
    },
    {
      "epoch": 0.022064269993210366,
      "grad_norm": 1.261952519416809,
      "learning_rate": 0.00019558733603287174,
      "loss": 4.3814,
      "step": 23670
    },
    {
      "epoch": 0.022073591611289457,
      "grad_norm": 1.1418771743774414,
      "learning_rate": 0.00019558547171077655,
      "loss": 4.2325,
      "step": 23680
    },
    {
      "epoch": 0.02208291322936855,
      "grad_norm": 1.283158540725708,
      "learning_rate": 0.00019558360738868133,
      "loss": 4.3354,
      "step": 23690
    },
    {
      "epoch": 0.022092234847447642,
      "grad_norm": 1.344997525215149,
      "learning_rate": 0.00019558174306658614,
      "loss": 4.274,
      "step": 23700
    },
    {
      "epoch": 0.022101556465526733,
      "grad_norm": 1.3954113721847534,
      "learning_rate": 0.00019557987874449094,
      "loss": 4.3119,
      "step": 23710
    },
    {
      "epoch": 0.022110878083605824,
      "grad_norm": 1.211361289024353,
      "learning_rate": 0.00019557801442239575,
      "loss": 4.1118,
      "step": 23720
    },
    {
      "epoch": 0.02212019970168492,
      "grad_norm": 1.268986701965332,
      "learning_rate": 0.00019557615010030056,
      "loss": 4.219,
      "step": 23730
    },
    {
      "epoch": 0.02212952131976401,
      "grad_norm": 1.1736904382705688,
      "learning_rate": 0.00019557428577820533,
      "loss": 4.2553,
      "step": 23740
    },
    {
      "epoch": 0.0221388429378431,
      "grad_norm": 1.192978024482727,
      "learning_rate": 0.00019557242145611014,
      "loss": 4.3537,
      "step": 23750
    },
    {
      "epoch": 0.02214816455592219,
      "grad_norm": 1.195302128791809,
      "learning_rate": 0.00019557055713401492,
      "loss": 4.3918,
      "step": 23760
    },
    {
      "epoch": 0.022157486174001286,
      "grad_norm": 1.200590968132019,
      "learning_rate": 0.00019556869281191975,
      "loss": 4.3327,
      "step": 23770
    },
    {
      "epoch": 0.022166807792080377,
      "grad_norm": 1.187246561050415,
      "learning_rate": 0.00019556682848982456,
      "loss": 4.4329,
      "step": 23780
    },
    {
      "epoch": 0.022176129410159468,
      "grad_norm": 0.9097520112991333,
      "learning_rate": 0.00019556496416772934,
      "loss": 4.4299,
      "step": 23790
    },
    {
      "epoch": 0.02218545102823856,
      "grad_norm": 1.1979793310165405,
      "learning_rate": 0.00019556309984563415,
      "loss": 4.3865,
      "step": 23800
    },
    {
      "epoch": 0.022194772646317654,
      "grad_norm": 1.094116449356079,
      "learning_rate": 0.00019556123552353892,
      "loss": 4.3767,
      "step": 23810
    },
    {
      "epoch": 0.022204094264396745,
      "grad_norm": 1.2772101163864136,
      "learning_rate": 0.00019555937120144373,
      "loss": 4.1632,
      "step": 23820
    },
    {
      "epoch": 0.022213415882475836,
      "grad_norm": 1.0829565525054932,
      "learning_rate": 0.00019555750687934856,
      "loss": 4.1045,
      "step": 23830
    },
    {
      "epoch": 0.022222737500554927,
      "grad_norm": 1.0334030389785767,
      "learning_rate": 0.00019555564255725334,
      "loss": 4.3635,
      "step": 23840
    },
    {
      "epoch": 0.022232059118634018,
      "grad_norm": 1.6389391422271729,
      "learning_rate": 0.00019555377823515815,
      "loss": 4.2591,
      "step": 23850
    },
    {
      "epoch": 0.022241380736713112,
      "grad_norm": 1.0552549362182617,
      "learning_rate": 0.00019555191391306293,
      "loss": 4.3552,
      "step": 23860
    },
    {
      "epoch": 0.022250702354792203,
      "grad_norm": 0.901793360710144,
      "learning_rate": 0.00019555004959096774,
      "loss": 4.2863,
      "step": 23870
    },
    {
      "epoch": 0.022260023972871294,
      "grad_norm": 0.9929845929145813,
      "learning_rate": 0.00019554818526887254,
      "loss": 4.3517,
      "step": 23880
    },
    {
      "epoch": 0.022269345590950385,
      "grad_norm": 1.1043962240219116,
      "learning_rate": 0.00019554632094677735,
      "loss": 4.3565,
      "step": 23890
    },
    {
      "epoch": 0.02227866720902948,
      "grad_norm": 0.8500731587409973,
      "learning_rate": 0.00019554445662468215,
      "loss": 4.4466,
      "step": 23900
    },
    {
      "epoch": 0.02228798882710857,
      "grad_norm": 0.9456198215484619,
      "learning_rate": 0.00019554259230258693,
      "loss": 4.6081,
      "step": 23910
    },
    {
      "epoch": 0.022297310445187662,
      "grad_norm": 1.4071197509765625,
      "learning_rate": 0.00019554072798049174,
      "loss": 4.3349,
      "step": 23920
    },
    {
      "epoch": 0.022306632063266753,
      "grad_norm": 1.1754246950149536,
      "learning_rate": 0.00019553886365839655,
      "loss": 3.9812,
      "step": 23930
    },
    {
      "epoch": 0.022315953681345847,
      "grad_norm": 1.094680666923523,
      "learning_rate": 0.00019553699933630135,
      "loss": 4.2958,
      "step": 23940
    },
    {
      "epoch": 0.02232527529942494,
      "grad_norm": 1.063615083694458,
      "learning_rate": 0.00019553513501420616,
      "loss": 4.2517,
      "step": 23950
    },
    {
      "epoch": 0.02233459691750403,
      "grad_norm": 1.1039029359817505,
      "learning_rate": 0.00019553327069211094,
      "loss": 4.247,
      "step": 23960
    },
    {
      "epoch": 0.02234391853558312,
      "grad_norm": 1.224463939666748,
      "learning_rate": 0.00019553140637001574,
      "loss": 4.4371,
      "step": 23970
    },
    {
      "epoch": 0.022353240153662215,
      "grad_norm": 1.2122137546539307,
      "learning_rate": 0.00019552954204792055,
      "loss": 4.4416,
      "step": 23980
    },
    {
      "epoch": 0.022362561771741306,
      "grad_norm": 1.182957410812378,
      "learning_rate": 0.00019552767772582533,
      "loss": 4.5423,
      "step": 23990
    },
    {
      "epoch": 0.022371883389820397,
      "grad_norm": 1.1071624755859375,
      "learning_rate": 0.00019552581340373016,
      "loss": 4.6173,
      "step": 24000
    },
    {
      "epoch": 0.022381205007899488,
      "grad_norm": 1.03336763381958,
      "learning_rate": 0.00019552394908163494,
      "loss": 4.3742,
      "step": 24010
    },
    {
      "epoch": 0.02239052662597858,
      "grad_norm": 0.9738283157348633,
      "learning_rate": 0.00019552208475953975,
      "loss": 4.374,
      "step": 24020
    },
    {
      "epoch": 0.022399848244057673,
      "grad_norm": 1.090299367904663,
      "learning_rate": 0.00019552022043744455,
      "loss": 4.2191,
      "step": 24030
    },
    {
      "epoch": 0.022409169862136764,
      "grad_norm": 1.025712251663208,
      "learning_rate": 0.00019551835611534933,
      "loss": 4.3048,
      "step": 24040
    },
    {
      "epoch": 0.022418491480215855,
      "grad_norm": 1.0785235166549683,
      "learning_rate": 0.00019551649179325414,
      "loss": 4.4091,
      "step": 24050
    },
    {
      "epoch": 0.022427813098294946,
      "grad_norm": 1.1313953399658203,
      "learning_rate": 0.00019551462747115895,
      "loss": 4.1306,
      "step": 24060
    },
    {
      "epoch": 0.02243713471637404,
      "grad_norm": 1.2095413208007812,
      "learning_rate": 0.00019551276314906375,
      "loss": 4.3961,
      "step": 24070
    },
    {
      "epoch": 0.022446456334453132,
      "grad_norm": 1.2434359788894653,
      "learning_rate": 0.00019551089882696853,
      "loss": 4.3244,
      "step": 24080
    },
    {
      "epoch": 0.022455777952532223,
      "grad_norm": 0.9629742503166199,
      "learning_rate": 0.00019550903450487334,
      "loss": 4.3835,
      "step": 24090
    },
    {
      "epoch": 0.022465099570611314,
      "grad_norm": 1.2211849689483643,
      "learning_rate": 0.00019550717018277814,
      "loss": 4.2251,
      "step": 24100
    },
    {
      "epoch": 0.02247442118869041,
      "grad_norm": 1.3198729753494263,
      "learning_rate": 0.00019550530586068295,
      "loss": 4.3367,
      "step": 24110
    },
    {
      "epoch": 0.0224837428067695,
      "grad_norm": 1.4003764390945435,
      "learning_rate": 0.00019550344153858776,
      "loss": 4.2068,
      "step": 24120
    },
    {
      "epoch": 0.02249306442484859,
      "grad_norm": 1.490180492401123,
      "learning_rate": 0.00019550157721649254,
      "loss": 4.0996,
      "step": 24130
    },
    {
      "epoch": 0.02250238604292768,
      "grad_norm": 1.3010557889938354,
      "learning_rate": 0.00019549971289439734,
      "loss": 4.1205,
      "step": 24140
    },
    {
      "epoch": 0.022511707661006776,
      "grad_norm": 1.342936635017395,
      "learning_rate": 0.00019549784857230215,
      "loss": 4.3631,
      "step": 24150
    },
    {
      "epoch": 0.022521029279085867,
      "grad_norm": 1.063970923423767,
      "learning_rate": 0.00019549598425020696,
      "loss": 4.1149,
      "step": 24160
    },
    {
      "epoch": 0.022530350897164958,
      "grad_norm": 1.2001233100891113,
      "learning_rate": 0.00019549411992811176,
      "loss": 4.2202,
      "step": 24170
    },
    {
      "epoch": 0.02253967251524405,
      "grad_norm": 1.203852891921997,
      "learning_rate": 0.00019549225560601654,
      "loss": 4.1891,
      "step": 24180
    },
    {
      "epoch": 0.02254899413332314,
      "grad_norm": 1.1286802291870117,
      "learning_rate": 0.00019549039128392135,
      "loss": 4.3371,
      "step": 24190
    },
    {
      "epoch": 0.022558315751402234,
      "grad_norm": 1.30306875705719,
      "learning_rate": 0.00019548852696182615,
      "loss": 4.2636,
      "step": 24200
    },
    {
      "epoch": 0.022567637369481326,
      "grad_norm": 1.3335968255996704,
      "learning_rate": 0.00019548666263973093,
      "loss": 4.3973,
      "step": 24210
    },
    {
      "epoch": 0.022576958987560417,
      "grad_norm": 1.0683847665786743,
      "learning_rate": 0.00019548479831763577,
      "loss": 4.4949,
      "step": 24220
    },
    {
      "epoch": 0.022586280605639508,
      "grad_norm": 1.1036816835403442,
      "learning_rate": 0.00019548293399554055,
      "loss": 4.3764,
      "step": 24230
    },
    {
      "epoch": 0.022595602223718602,
      "grad_norm": 1.0963846445083618,
      "learning_rate": 0.00019548106967344535,
      "loss": 4.5674,
      "step": 24240
    },
    {
      "epoch": 0.022604923841797693,
      "grad_norm": 1.483090877532959,
      "learning_rate": 0.00019547920535135016,
      "loss": 4.4586,
      "step": 24250
    },
    {
      "epoch": 0.022614245459876784,
      "grad_norm": 1.180461049079895,
      "learning_rate": 0.00019547734102925494,
      "loss": 4.2849,
      "step": 24260
    },
    {
      "epoch": 0.022623567077955875,
      "grad_norm": 1.1371605396270752,
      "learning_rate": 0.00019547547670715974,
      "loss": 4.3913,
      "step": 24270
    },
    {
      "epoch": 0.02263288869603497,
      "grad_norm": 1.0948525667190552,
      "learning_rate": 0.00019547361238506455,
      "loss": 4.3908,
      "step": 24280
    },
    {
      "epoch": 0.02264221031411406,
      "grad_norm": 1.0131280422210693,
      "learning_rate": 0.00019547174806296936,
      "loss": 4.1996,
      "step": 24290
    },
    {
      "epoch": 0.02265153193219315,
      "grad_norm": 1.044194221496582,
      "learning_rate": 0.00019546988374087416,
      "loss": 4.2432,
      "step": 24300
    },
    {
      "epoch": 0.022660853550272243,
      "grad_norm": 1.213686466217041,
      "learning_rate": 0.00019546801941877894,
      "loss": 4.2506,
      "step": 24310
    },
    {
      "epoch": 0.022670175168351337,
      "grad_norm": 0.991962194442749,
      "learning_rate": 0.00019546615509668375,
      "loss": 4.3677,
      "step": 24320
    },
    {
      "epoch": 0.022679496786430428,
      "grad_norm": 1.13752019405365,
      "learning_rate": 0.00019546429077458855,
      "loss": 4.116,
      "step": 24330
    },
    {
      "epoch": 0.02268881840450952,
      "grad_norm": 0.994701087474823,
      "learning_rate": 0.00019546242645249336,
      "loss": 4.2874,
      "step": 24340
    },
    {
      "epoch": 0.02269814002258861,
      "grad_norm": 1.097616195678711,
      "learning_rate": 0.00019546056213039817,
      "loss": 4.3238,
      "step": 24350
    },
    {
      "epoch": 0.0227074616406677,
      "grad_norm": 1.2617865800857544,
      "learning_rate": 0.00019545869780830295,
      "loss": 4.1676,
      "step": 24360
    },
    {
      "epoch": 0.022716783258746796,
      "grad_norm": 1.2567083835601807,
      "learning_rate": 0.00019545683348620775,
      "loss": 4.3038,
      "step": 24370
    },
    {
      "epoch": 0.022726104876825887,
      "grad_norm": 1.0654258728027344,
      "learning_rate": 0.00019545496916411253,
      "loss": 4.2323,
      "step": 24380
    },
    {
      "epoch": 0.022735426494904978,
      "grad_norm": 1.19419527053833,
      "learning_rate": 0.00019545310484201737,
      "loss": 4.3678,
      "step": 24390
    },
    {
      "epoch": 0.02274474811298407,
      "grad_norm": 1.0525989532470703,
      "learning_rate": 0.00019545124051992217,
      "loss": 4.3974,
      "step": 24400
    },
    {
      "epoch": 0.022754069731063163,
      "grad_norm": 1.3828620910644531,
      "learning_rate": 0.00019544937619782695,
      "loss": 4.3844,
      "step": 24410
    },
    {
      "epoch": 0.022763391349142254,
      "grad_norm": 1.305601716041565,
      "learning_rate": 0.00019544751187573176,
      "loss": 4.1593,
      "step": 24420
    },
    {
      "epoch": 0.022772712967221345,
      "grad_norm": 1.0573183298110962,
      "learning_rate": 0.00019544564755363654,
      "loss": 4.2989,
      "step": 24430
    },
    {
      "epoch": 0.022782034585300436,
      "grad_norm": 1.1054061651229858,
      "learning_rate": 0.00019544378323154134,
      "loss": 4.1618,
      "step": 24440
    },
    {
      "epoch": 0.02279135620337953,
      "grad_norm": 1.1506904363632202,
      "learning_rate": 0.00019544191890944618,
      "loss": 4.2826,
      "step": 24450
    },
    {
      "epoch": 0.02280067782145862,
      "grad_norm": 1.0768589973449707,
      "learning_rate": 0.00019544005458735096,
      "loss": 4.5679,
      "step": 24460
    },
    {
      "epoch": 0.022809999439537713,
      "grad_norm": 0.980288565158844,
      "learning_rate": 0.00019543819026525576,
      "loss": 4.4404,
      "step": 24470
    },
    {
      "epoch": 0.022819321057616804,
      "grad_norm": 0.9518328309059143,
      "learning_rate": 0.00019543632594316054,
      "loss": 4.3175,
      "step": 24480
    },
    {
      "epoch": 0.022828642675695898,
      "grad_norm": 1.193381428718567,
      "learning_rate": 0.00019543446162106535,
      "loss": 4.5597,
      "step": 24490
    },
    {
      "epoch": 0.02283796429377499,
      "grad_norm": 1.1200419664382935,
      "learning_rate": 0.00019543259729897015,
      "loss": 4.2891,
      "step": 24500
    },
    {
      "epoch": 0.02284728591185408,
      "grad_norm": 1.3487435579299927,
      "learning_rate": 0.00019543073297687496,
      "loss": 4.3139,
      "step": 24510
    },
    {
      "epoch": 0.02285660752993317,
      "grad_norm": 0.9159813523292542,
      "learning_rate": 0.00019542886865477977,
      "loss": 4.3294,
      "step": 24520
    },
    {
      "epoch": 0.022865929148012262,
      "grad_norm": 1.204926609992981,
      "learning_rate": 0.00019542700433268455,
      "loss": 4.401,
      "step": 24530
    },
    {
      "epoch": 0.022875250766091357,
      "grad_norm": 1.225124478340149,
      "learning_rate": 0.00019542514001058935,
      "loss": 4.4485,
      "step": 24540
    },
    {
      "epoch": 0.022884572384170448,
      "grad_norm": 0.9511662721633911,
      "learning_rate": 0.00019542327568849416,
      "loss": 4.3257,
      "step": 24550
    },
    {
      "epoch": 0.02289389400224954,
      "grad_norm": 1.1982462406158447,
      "learning_rate": 0.00019542141136639896,
      "loss": 4.2456,
      "step": 24560
    },
    {
      "epoch": 0.02290321562032863,
      "grad_norm": 1.101113200187683,
      "learning_rate": 0.00019541954704430377,
      "loss": 4.2005,
      "step": 24570
    },
    {
      "epoch": 0.022912537238407724,
      "grad_norm": 1.1622236967086792,
      "learning_rate": 0.00019541768272220855,
      "loss": 4.123,
      "step": 24580
    },
    {
      "epoch": 0.022921858856486815,
      "grad_norm": 1.2927500009536743,
      "learning_rate": 0.00019541581840011336,
      "loss": 4.4239,
      "step": 24590
    },
    {
      "epoch": 0.022931180474565906,
      "grad_norm": 1.0844045877456665,
      "learning_rate": 0.00019541395407801816,
      "loss": 4.1916,
      "step": 24600
    },
    {
      "epoch": 0.022940502092644997,
      "grad_norm": 1.1976271867752075,
      "learning_rate": 0.00019541208975592297,
      "loss": 4.3016,
      "step": 24610
    },
    {
      "epoch": 0.022949823710724092,
      "grad_norm": 1.2318114042282104,
      "learning_rate": 0.00019541022543382777,
      "loss": 4.2032,
      "step": 24620
    },
    {
      "epoch": 0.022959145328803183,
      "grad_norm": 1.1799439191818237,
      "learning_rate": 0.00019540836111173255,
      "loss": 4.2901,
      "step": 24630
    },
    {
      "epoch": 0.022968466946882274,
      "grad_norm": 1.0168405771255493,
      "learning_rate": 0.00019540649678963736,
      "loss": 4.5062,
      "step": 24640
    },
    {
      "epoch": 0.022977788564961365,
      "grad_norm": 1.2494419813156128,
      "learning_rate": 0.00019540463246754217,
      "loss": 4.3337,
      "step": 24650
    },
    {
      "epoch": 0.02298711018304046,
      "grad_norm": 1.258803367614746,
      "learning_rate": 0.00019540276814544695,
      "loss": 4.3573,
      "step": 24660
    },
    {
      "epoch": 0.02299643180111955,
      "grad_norm": 1.2467066049575806,
      "learning_rate": 0.00019540090382335178,
      "loss": 4.2557,
      "step": 24670
    },
    {
      "epoch": 0.02300575341919864,
      "grad_norm": 1.1284219026565552,
      "learning_rate": 0.00019539903950125656,
      "loss": 4.3272,
      "step": 24680
    },
    {
      "epoch": 0.023015075037277732,
      "grad_norm": 1.069342851638794,
      "learning_rate": 0.00019539717517916136,
      "loss": 4.3883,
      "step": 24690
    },
    {
      "epoch": 0.023024396655356823,
      "grad_norm": 1.2196285724639893,
      "learning_rate": 0.00019539531085706617,
      "loss": 4.2109,
      "step": 24700
    },
    {
      "epoch": 0.023033718273435918,
      "grad_norm": 1.1922166347503662,
      "learning_rate": 0.00019539344653497095,
      "loss": 4.2531,
      "step": 24710
    },
    {
      "epoch": 0.02304303989151501,
      "grad_norm": 1.158766269683838,
      "learning_rate": 0.00019539158221287576,
      "loss": 4.4071,
      "step": 24720
    },
    {
      "epoch": 0.0230523615095941,
      "grad_norm": 1.1905725002288818,
      "learning_rate": 0.00019538971789078056,
      "loss": 4.4486,
      "step": 24730
    },
    {
      "epoch": 0.02306168312767319,
      "grad_norm": 1.1171233654022217,
      "learning_rate": 0.00019538785356868537,
      "loss": 4.2635,
      "step": 24740
    },
    {
      "epoch": 0.023071004745752285,
      "grad_norm": 1.0142176151275635,
      "learning_rate": 0.00019538598924659018,
      "loss": 4.2853,
      "step": 24750
    },
    {
      "epoch": 0.023080326363831376,
      "grad_norm": 1.3222604990005493,
      "learning_rate": 0.00019538412492449495,
      "loss": 4.2903,
      "step": 24760
    },
    {
      "epoch": 0.023089647981910467,
      "grad_norm": 1.0093520879745483,
      "learning_rate": 0.00019538226060239976,
      "loss": 4.255,
      "step": 24770
    },
    {
      "epoch": 0.02309896959998956,
      "grad_norm": 0.9300851225852966,
      "learning_rate": 0.00019538039628030457,
      "loss": 4.4502,
      "step": 24780
    },
    {
      "epoch": 0.023108291218068653,
      "grad_norm": 1.1337575912475586,
      "learning_rate": 0.00019537853195820937,
      "loss": 4.1845,
      "step": 24790
    },
    {
      "epoch": 0.023117612836147744,
      "grad_norm": 1.1245863437652588,
      "learning_rate": 0.00019537666763611418,
      "loss": 4.3038,
      "step": 24800
    },
    {
      "epoch": 0.023126934454226835,
      "grad_norm": 1.3726731538772583,
      "learning_rate": 0.00019537480331401896,
      "loss": 4.4545,
      "step": 24810
    },
    {
      "epoch": 0.023136256072305926,
      "grad_norm": 1.3319640159606934,
      "learning_rate": 0.00019537293899192377,
      "loss": 4.3702,
      "step": 24820
    },
    {
      "epoch": 0.02314557769038502,
      "grad_norm": 1.3213223218917847,
      "learning_rate": 0.00019537107466982854,
      "loss": 4.3285,
      "step": 24830
    },
    {
      "epoch": 0.02315489930846411,
      "grad_norm": 1.4504398107528687,
      "learning_rate": 0.00019536921034773338,
      "loss": 4.2517,
      "step": 24840
    },
    {
      "epoch": 0.023164220926543203,
      "grad_norm": 1.197959542274475,
      "learning_rate": 0.00019536734602563818,
      "loss": 4.3452,
      "step": 24850
    },
    {
      "epoch": 0.023173542544622294,
      "grad_norm": 1.133874535560608,
      "learning_rate": 0.00019536548170354296,
      "loss": 4.4373,
      "step": 24860
    },
    {
      "epoch": 0.023182864162701385,
      "grad_norm": 1.2772810459136963,
      "learning_rate": 0.00019536361738144777,
      "loss": 4.3405,
      "step": 24870
    },
    {
      "epoch": 0.02319218578078048,
      "grad_norm": 1.1216360330581665,
      "learning_rate": 0.00019536175305935255,
      "loss": 4.3491,
      "step": 24880
    },
    {
      "epoch": 0.02320150739885957,
      "grad_norm": 1.1430422067642212,
      "learning_rate": 0.00019535988873725736,
      "loss": 4.3086,
      "step": 24890
    },
    {
      "epoch": 0.02321082901693866,
      "grad_norm": 0.8950048089027405,
      "learning_rate": 0.0001953580244151622,
      "loss": 4.3035,
      "step": 24900
    },
    {
      "epoch": 0.023220150635017752,
      "grad_norm": 1.271438717842102,
      "learning_rate": 0.00019535616009306697,
      "loss": 4.469,
      "step": 24910
    },
    {
      "epoch": 0.023229472253096847,
      "grad_norm": 1.1475095748901367,
      "learning_rate": 0.00019535429577097177,
      "loss": 4.3953,
      "step": 24920
    },
    {
      "epoch": 0.023238793871175938,
      "grad_norm": 1.1112210750579834,
      "learning_rate": 0.00019535243144887655,
      "loss": 4.3658,
      "step": 24930
    },
    {
      "epoch": 0.02324811548925503,
      "grad_norm": 1.031062126159668,
      "learning_rate": 0.00019535056712678136,
      "loss": 4.2962,
      "step": 24940
    },
    {
      "epoch": 0.02325743710733412,
      "grad_norm": 0.9918408989906311,
      "learning_rate": 0.00019534870280468617,
      "loss": 4.3083,
      "step": 24950
    },
    {
      "epoch": 0.023266758725413214,
      "grad_norm": 1.3816722631454468,
      "learning_rate": 0.00019534683848259097,
      "loss": 4.3821,
      "step": 24960
    },
    {
      "epoch": 0.023276080343492305,
      "grad_norm": 1.2172651290893555,
      "learning_rate": 0.00019534497416049578,
      "loss": 4.3798,
      "step": 24970
    },
    {
      "epoch": 0.023285401961571396,
      "grad_norm": 1.3566842079162598,
      "learning_rate": 0.00019534310983840056,
      "loss": 4.4028,
      "step": 24980
    },
    {
      "epoch": 0.023294723579650487,
      "grad_norm": 1.2486677169799805,
      "learning_rate": 0.00019534124551630536,
      "loss": 4.1373,
      "step": 24990
    },
    {
      "epoch": 0.02330404519772958,
      "grad_norm": 0.8885643482208252,
      "learning_rate": 0.00019533938119421017,
      "loss": 4.187,
      "step": 25000
    },
    {
      "epoch": 0.023313366815808673,
      "grad_norm": 1.3953380584716797,
      "learning_rate": 0.00019533751687211498,
      "loss": 4.1525,
      "step": 25010
    },
    {
      "epoch": 0.023322688433887764,
      "grad_norm": 1.0635651350021362,
      "learning_rate": 0.00019533565255001978,
      "loss": 4.379,
      "step": 25020
    },
    {
      "epoch": 0.023332010051966855,
      "grad_norm": 1.1968889236450195,
      "learning_rate": 0.00019533378822792456,
      "loss": 4.1016,
      "step": 25030
    },
    {
      "epoch": 0.023341331670045946,
      "grad_norm": 1.090334415435791,
      "learning_rate": 0.00019533192390582937,
      "loss": 4.1213,
      "step": 25040
    },
    {
      "epoch": 0.02335065328812504,
      "grad_norm": 1.2018760442733765,
      "learning_rate": 0.00019533005958373418,
      "loss": 4.491,
      "step": 25050
    },
    {
      "epoch": 0.02335997490620413,
      "grad_norm": 1.2252004146575928,
      "learning_rate": 0.00019532819526163898,
      "loss": 4.3795,
      "step": 25060
    },
    {
      "epoch": 0.023369296524283222,
      "grad_norm": 1.118645191192627,
      "learning_rate": 0.0001953263309395438,
      "loss": 4.1221,
      "step": 25070
    },
    {
      "epoch": 0.023378618142362313,
      "grad_norm": 1.1242575645446777,
      "learning_rate": 0.00019532446661744857,
      "loss": 4.1721,
      "step": 25080
    },
    {
      "epoch": 0.023387939760441408,
      "grad_norm": 0.9725354313850403,
      "learning_rate": 0.00019532260229535337,
      "loss": 4.2827,
      "step": 25090
    },
    {
      "epoch": 0.0233972613785205,
      "grad_norm": 0.9786919355392456,
      "learning_rate": 0.00019532073797325818,
      "loss": 4.5157,
      "step": 25100
    },
    {
      "epoch": 0.02340658299659959,
      "grad_norm": 1.2120628356933594,
      "learning_rate": 0.00019531887365116296,
      "loss": 4.2653,
      "step": 25110
    },
    {
      "epoch": 0.02341590461467868,
      "grad_norm": 1.2042920589447021,
      "learning_rate": 0.0001953170093290678,
      "loss": 4.1395,
      "step": 25120
    },
    {
      "epoch": 0.023425226232757775,
      "grad_norm": 1.1919137239456177,
      "learning_rate": 0.00019531514500697257,
      "loss": 4.4262,
      "step": 25130
    },
    {
      "epoch": 0.023434547850836866,
      "grad_norm": 1.1795544624328613,
      "learning_rate": 0.00019531328068487738,
      "loss": 4.2806,
      "step": 25140
    },
    {
      "epoch": 0.023443869468915957,
      "grad_norm": 1.0820406675338745,
      "learning_rate": 0.00019531141636278218,
      "loss": 4.4555,
      "step": 25150
    },
    {
      "epoch": 0.02345319108699505,
      "grad_norm": 1.1735633611679077,
      "learning_rate": 0.00019530955204068696,
      "loss": 4.1014,
      "step": 25160
    },
    {
      "epoch": 0.023462512705074143,
      "grad_norm": 1.1343457698822021,
      "learning_rate": 0.00019530768771859177,
      "loss": 4.3568,
      "step": 25170
    },
    {
      "epoch": 0.023471834323153234,
      "grad_norm": 1.185961365699768,
      "learning_rate": 0.00019530582339649658,
      "loss": 4.4278,
      "step": 25180
    },
    {
      "epoch": 0.023481155941232325,
      "grad_norm": 1.217828392982483,
      "learning_rate": 0.00019530395907440138,
      "loss": 4.2277,
      "step": 25190
    },
    {
      "epoch": 0.023490477559311416,
      "grad_norm": 1.2701529264450073,
      "learning_rate": 0.0001953020947523062,
      "loss": 4.5146,
      "step": 25200
    },
    {
      "epoch": 0.023499799177390507,
      "grad_norm": 1.1114646196365356,
      "learning_rate": 0.00019530023043021097,
      "loss": 4.0667,
      "step": 25210
    },
    {
      "epoch": 0.0235091207954696,
      "grad_norm": 0.9907222390174866,
      "learning_rate": 0.00019529836610811577,
      "loss": 4.2821,
      "step": 25220
    },
    {
      "epoch": 0.023518442413548692,
      "grad_norm": 1.3219910860061646,
      "learning_rate": 0.00019529650178602058,
      "loss": 4.5925,
      "step": 25230
    },
    {
      "epoch": 0.023527764031627783,
      "grad_norm": 1.2727689743041992,
      "learning_rate": 0.0001952946374639254,
      "loss": 4.2947,
      "step": 25240
    },
    {
      "epoch": 0.023537085649706874,
      "grad_norm": 1.3092213869094849,
      "learning_rate": 0.0001952927731418302,
      "loss": 4.2904,
      "step": 25250
    },
    {
      "epoch": 0.02354640726778597,
      "grad_norm": 1.1017239093780518,
      "learning_rate": 0.00019529090881973497,
      "loss": 4.4216,
      "step": 25260
    },
    {
      "epoch": 0.02355572888586506,
      "grad_norm": 2.022413730621338,
      "learning_rate": 0.00019528904449763978,
      "loss": 4.3045,
      "step": 25270
    },
    {
      "epoch": 0.02356505050394415,
      "grad_norm": 1.1321879625320435,
      "learning_rate": 0.00019528718017554456,
      "loss": 4.3955,
      "step": 25280
    },
    {
      "epoch": 0.023574372122023242,
      "grad_norm": 0.9072484374046326,
      "learning_rate": 0.0001952853158534494,
      "loss": 4.2283,
      "step": 25290
    },
    {
      "epoch": 0.023583693740102336,
      "grad_norm": 1.2103160619735718,
      "learning_rate": 0.0001952834515313542,
      "loss": 4.4656,
      "step": 25300
    },
    {
      "epoch": 0.023593015358181427,
      "grad_norm": 1.5355582237243652,
      "learning_rate": 0.00019528158720925898,
      "loss": 4.2818,
      "step": 25310
    },
    {
      "epoch": 0.02360233697626052,
      "grad_norm": 1.1358553171157837,
      "learning_rate": 0.00019527972288716378,
      "loss": 4.2948,
      "step": 25320
    },
    {
      "epoch": 0.02361165859433961,
      "grad_norm": 1.0714420080184937,
      "learning_rate": 0.00019527785856506856,
      "loss": 4.4019,
      "step": 25330
    },
    {
      "epoch": 0.023620980212418704,
      "grad_norm": 1.2087992429733276,
      "learning_rate": 0.00019527599424297337,
      "loss": 4.2834,
      "step": 25340
    },
    {
      "epoch": 0.023630301830497795,
      "grad_norm": 1.2150944471359253,
      "learning_rate": 0.0001952741299208782,
      "loss": 4.2913,
      "step": 25350
    },
    {
      "epoch": 0.023639623448576886,
      "grad_norm": 1.1687860488891602,
      "learning_rate": 0.00019527226559878298,
      "loss": 4.1426,
      "step": 25360
    },
    {
      "epoch": 0.023648945066655977,
      "grad_norm": 1.209556221961975,
      "learning_rate": 0.0001952704012766878,
      "loss": 4.1801,
      "step": 25370
    },
    {
      "epoch": 0.02365826668473507,
      "grad_norm": 0.9708238244056702,
      "learning_rate": 0.00019526853695459257,
      "loss": 4.3204,
      "step": 25380
    },
    {
      "epoch": 0.023667588302814162,
      "grad_norm": 1.0697259902954102,
      "learning_rate": 0.00019526667263249737,
      "loss": 4.3495,
      "step": 25390
    },
    {
      "epoch": 0.023676909920893253,
      "grad_norm": 1.1898391246795654,
      "learning_rate": 0.00019526480831040218,
      "loss": 4.3145,
      "step": 25400
    },
    {
      "epoch": 0.023686231538972344,
      "grad_norm": 1.0595635175704956,
      "learning_rate": 0.00019526294398830699,
      "loss": 4.2747,
      "step": 25410
    },
    {
      "epoch": 0.023695553157051435,
      "grad_norm": 1.1836415529251099,
      "learning_rate": 0.0001952610796662118,
      "loss": 4.2315,
      "step": 25420
    },
    {
      "epoch": 0.02370487477513053,
      "grad_norm": 1.435659408569336,
      "learning_rate": 0.00019525921534411657,
      "loss": 4.0833,
      "step": 25430
    },
    {
      "epoch": 0.02371419639320962,
      "grad_norm": 1.133203148841858,
      "learning_rate": 0.00019525735102202138,
      "loss": 4.4623,
      "step": 25440
    },
    {
      "epoch": 0.023723518011288712,
      "grad_norm": 1.2119722366333008,
      "learning_rate": 0.00019525548669992618,
      "loss": 4.2644,
      "step": 25450
    },
    {
      "epoch": 0.023732839629367803,
      "grad_norm": 1.2372931241989136,
      "learning_rate": 0.000195253622377831,
      "loss": 4.1689,
      "step": 25460
    },
    {
      "epoch": 0.023742161247446898,
      "grad_norm": 1.1760579347610474,
      "learning_rate": 0.0001952517580557358,
      "loss": 4.2362,
      "step": 25470
    },
    {
      "epoch": 0.02375148286552599,
      "grad_norm": 0.9510321021080017,
      "learning_rate": 0.00019524989373364058,
      "loss": 4.6018,
      "step": 25480
    },
    {
      "epoch": 0.02376080448360508,
      "grad_norm": 1.1331539154052734,
      "learning_rate": 0.00019524802941154538,
      "loss": 4.2471,
      "step": 25490
    },
    {
      "epoch": 0.02377012610168417,
      "grad_norm": 0.9939475655555725,
      "learning_rate": 0.00019524616508945016,
      "loss": 4.251,
      "step": 25500
    },
    {
      "epoch": 0.023779447719763265,
      "grad_norm": 0.9226861596107483,
      "learning_rate": 0.000195244300767355,
      "loss": 4.2114,
      "step": 25510
    },
    {
      "epoch": 0.023788769337842356,
      "grad_norm": 1.3129124641418457,
      "learning_rate": 0.0001952424364452598,
      "loss": 4.2371,
      "step": 25520
    },
    {
      "epoch": 0.023798090955921447,
      "grad_norm": 1.3315094709396362,
      "learning_rate": 0.00019524057212316458,
      "loss": 4.2591,
      "step": 25530
    },
    {
      "epoch": 0.023807412574000538,
      "grad_norm": 1.1264277696609497,
      "learning_rate": 0.00019523870780106939,
      "loss": 4.2867,
      "step": 25540
    },
    {
      "epoch": 0.023816734192079633,
      "grad_norm": 1.332444190979004,
      "learning_rate": 0.00019523684347897417,
      "loss": 4.3519,
      "step": 25550
    },
    {
      "epoch": 0.023826055810158724,
      "grad_norm": 1.2496249675750732,
      "learning_rate": 0.00019523497915687897,
      "loss": 4.3268,
      "step": 25560
    },
    {
      "epoch": 0.023835377428237815,
      "grad_norm": 1.2552169561386108,
      "learning_rate": 0.0001952331148347838,
      "loss": 4.3932,
      "step": 25570
    },
    {
      "epoch": 0.023844699046316906,
      "grad_norm": 1.0531643629074097,
      "learning_rate": 0.00019523125051268858,
      "loss": 4.1684,
      "step": 25580
    },
    {
      "epoch": 0.023854020664395997,
      "grad_norm": 1.2416619062423706,
      "learning_rate": 0.0001952293861905934,
      "loss": 4.2255,
      "step": 25590
    },
    {
      "epoch": 0.02386334228247509,
      "grad_norm": 1.138891339302063,
      "learning_rate": 0.00019522752186849817,
      "loss": 4.3715,
      "step": 25600
    },
    {
      "epoch": 0.023872663900554182,
      "grad_norm": 0.9761542081832886,
      "learning_rate": 0.00019522565754640298,
      "loss": 4.4061,
      "step": 25610
    },
    {
      "epoch": 0.023881985518633273,
      "grad_norm": 1.1026179790496826,
      "learning_rate": 0.00019522379322430778,
      "loss": 4.4388,
      "step": 25620
    },
    {
      "epoch": 0.023891307136712364,
      "grad_norm": 1.2051761150360107,
      "learning_rate": 0.0001952219289022126,
      "loss": 4.4088,
      "step": 25630
    },
    {
      "epoch": 0.02390062875479146,
      "grad_norm": 0.9162956476211548,
      "learning_rate": 0.0001952200645801174,
      "loss": 4.1773,
      "step": 25640
    },
    {
      "epoch": 0.02390995037287055,
      "grad_norm": 1.0118372440338135,
      "learning_rate": 0.00019521820025802217,
      "loss": 4.3183,
      "step": 25650
    },
    {
      "epoch": 0.02391927199094964,
      "grad_norm": 1.0276916027069092,
      "learning_rate": 0.00019521633593592698,
      "loss": 4.0406,
      "step": 25660
    },
    {
      "epoch": 0.02392859360902873,
      "grad_norm": 1.0317009687423706,
      "learning_rate": 0.0001952144716138318,
      "loss": 4.3307,
      "step": 25670
    },
    {
      "epoch": 0.023937915227107826,
      "grad_norm": 1.0358209609985352,
      "learning_rate": 0.0001952126072917366,
      "loss": 4.59,
      "step": 25680
    },
    {
      "epoch": 0.023947236845186917,
      "grad_norm": 1.180587887763977,
      "learning_rate": 0.0001952107429696414,
      "loss": 4.379,
      "step": 25690
    },
    {
      "epoch": 0.023956558463266008,
      "grad_norm": 0.9580869078636169,
      "learning_rate": 0.00019520887864754618,
      "loss": 4.3567,
      "step": 25700
    },
    {
      "epoch": 0.0239658800813451,
      "grad_norm": 1.1561044454574585,
      "learning_rate": 0.00019520701432545098,
      "loss": 4.2484,
      "step": 25710
    },
    {
      "epoch": 0.023975201699424194,
      "grad_norm": 1.3414440155029297,
      "learning_rate": 0.0001952051500033558,
      "loss": 4.1571,
      "step": 25720
    },
    {
      "epoch": 0.023984523317503285,
      "grad_norm": 1.1553242206573486,
      "learning_rate": 0.00019520328568126057,
      "loss": 4.4419,
      "step": 25730
    },
    {
      "epoch": 0.023993844935582376,
      "grad_norm": 1.2891898155212402,
      "learning_rate": 0.0001952014213591654,
      "loss": 4.5044,
      "step": 25740
    },
    {
      "epoch": 0.024003166553661467,
      "grad_norm": 0.9680271148681641,
      "learning_rate": 0.00019519955703707018,
      "loss": 4.2377,
      "step": 25750
    },
    {
      "epoch": 0.024012488171740558,
      "grad_norm": 1.1747032403945923,
      "learning_rate": 0.000195197692714975,
      "loss": 4.205,
      "step": 25760
    },
    {
      "epoch": 0.024021809789819652,
      "grad_norm": 1.1816003322601318,
      "learning_rate": 0.0001951958283928798,
      "loss": 4.356,
      "step": 25770
    },
    {
      "epoch": 0.024031131407898743,
      "grad_norm": 1.310225009918213,
      "learning_rate": 0.00019519396407078457,
      "loss": 4.4768,
      "step": 25780
    },
    {
      "epoch": 0.024040453025977834,
      "grad_norm": 0.9858599901199341,
      "learning_rate": 0.00019519209974868938,
      "loss": 4.1429,
      "step": 25790
    },
    {
      "epoch": 0.024049774644056925,
      "grad_norm": 1.0116597414016724,
      "learning_rate": 0.0001951902354265942,
      "loss": 4.3808,
      "step": 25800
    },
    {
      "epoch": 0.02405909626213602,
      "grad_norm": 1.2355382442474365,
      "learning_rate": 0.000195188371104499,
      "loss": 4.2324,
      "step": 25810
    },
    {
      "epoch": 0.02406841788021511,
      "grad_norm": 1.1173208951950073,
      "learning_rate": 0.0001951865067824038,
      "loss": 4.3871,
      "step": 25820
    },
    {
      "epoch": 0.024077739498294202,
      "grad_norm": 1.4504927396774292,
      "learning_rate": 0.00019518464246030858,
      "loss": 4.4121,
      "step": 25830
    },
    {
      "epoch": 0.024087061116373293,
      "grad_norm": 1.154530644416809,
      "learning_rate": 0.00019518277813821339,
      "loss": 4.3395,
      "step": 25840
    },
    {
      "epoch": 0.024096382734452387,
      "grad_norm": 1.134407639503479,
      "learning_rate": 0.0001951809138161182,
      "loss": 4.4013,
      "step": 25850
    },
    {
      "epoch": 0.02410570435253148,
      "grad_norm": 0.9842607378959656,
      "learning_rate": 0.000195179049494023,
      "loss": 4.1948,
      "step": 25860
    },
    {
      "epoch": 0.02411502597061057,
      "grad_norm": 1.3058041334152222,
      "learning_rate": 0.0001951771851719278,
      "loss": 4.2028,
      "step": 25870
    },
    {
      "epoch": 0.02412434758868966,
      "grad_norm": 0.9661714434623718,
      "learning_rate": 0.00019517532084983258,
      "loss": 4.3601,
      "step": 25880
    },
    {
      "epoch": 0.024133669206768755,
      "grad_norm": 1.1433345079421997,
      "learning_rate": 0.0001951734565277374,
      "loss": 4.5,
      "step": 25890
    },
    {
      "epoch": 0.024142990824847846,
      "grad_norm": 1.0052543878555298,
      "learning_rate": 0.0001951715922056422,
      "loss": 4.306,
      "step": 25900
    },
    {
      "epoch": 0.024152312442926937,
      "grad_norm": 1.4083428382873535,
      "learning_rate": 0.000195169727883547,
      "loss": 4.3123,
      "step": 25910
    },
    {
      "epoch": 0.024161634061006028,
      "grad_norm": 1.0732117891311646,
      "learning_rate": 0.0001951678635614518,
      "loss": 4.383,
      "step": 25920
    },
    {
      "epoch": 0.02417095567908512,
      "grad_norm": 1.163521409034729,
      "learning_rate": 0.0001951659992393566,
      "loss": 4.2386,
      "step": 25930
    },
    {
      "epoch": 0.024180277297164213,
      "grad_norm": 0.8232237696647644,
      "learning_rate": 0.0001951641349172614,
      "loss": 4.3398,
      "step": 25940
    },
    {
      "epoch": 0.024189598915243304,
      "grad_norm": 1.1200255155563354,
      "learning_rate": 0.00019516227059516617,
      "loss": 4.4749,
      "step": 25950
    },
    {
      "epoch": 0.024198920533322395,
      "grad_norm": 1.0774998664855957,
      "learning_rate": 0.000195160406273071,
      "loss": 4.2548,
      "step": 25960
    },
    {
      "epoch": 0.024208242151401486,
      "grad_norm": 1.175680160522461,
      "learning_rate": 0.0001951585419509758,
      "loss": 4.3912,
      "step": 25970
    },
    {
      "epoch": 0.02421756376948058,
      "grad_norm": 1.0778518915176392,
      "learning_rate": 0.0001951566776288806,
      "loss": 4.3637,
      "step": 25980
    },
    {
      "epoch": 0.024226885387559672,
      "grad_norm": 0.949350893497467,
      "learning_rate": 0.0001951548133067854,
      "loss": 4.2156,
      "step": 25990
    },
    {
      "epoch": 0.024236207005638763,
      "grad_norm": 1.155747652053833,
      "learning_rate": 0.00019515294898469018,
      "loss": 4.228,
      "step": 26000
    },
    {
      "epoch": 0.024245528623717854,
      "grad_norm": 1.0585997104644775,
      "learning_rate": 0.00019515108466259498,
      "loss": 4.1774,
      "step": 26010
    },
    {
      "epoch": 0.02425485024179695,
      "grad_norm": 1.1439288854599,
      "learning_rate": 0.00019514922034049982,
      "loss": 4.2351,
      "step": 26020
    },
    {
      "epoch": 0.02426417185987604,
      "grad_norm": 1.079811692237854,
      "learning_rate": 0.0001951473560184046,
      "loss": 4.4232,
      "step": 26030
    },
    {
      "epoch": 0.02427349347795513,
      "grad_norm": 1.2490562200546265,
      "learning_rate": 0.0001951454916963094,
      "loss": 4.1832,
      "step": 26040
    },
    {
      "epoch": 0.02428281509603422,
      "grad_norm": 1.1515179872512817,
      "learning_rate": 0.00019514362737421418,
      "loss": 4.157,
      "step": 26050
    },
    {
      "epoch": 0.024292136714113316,
      "grad_norm": 1.2545380592346191,
      "learning_rate": 0.000195141763052119,
      "loss": 4.2483,
      "step": 26060
    },
    {
      "epoch": 0.024301458332192407,
      "grad_norm": 1.0269255638122559,
      "learning_rate": 0.0001951398987300238,
      "loss": 4.3672,
      "step": 26070
    },
    {
      "epoch": 0.024310779950271498,
      "grad_norm": 0.9609196782112122,
      "learning_rate": 0.0001951380344079286,
      "loss": 4.3474,
      "step": 26080
    },
    {
      "epoch": 0.02432010156835059,
      "grad_norm": 1.2810255289077759,
      "learning_rate": 0.0001951361700858334,
      "loss": 4.2549,
      "step": 26090
    },
    {
      "epoch": 0.02432942318642968,
      "grad_norm": 1.1345463991165161,
      "learning_rate": 0.0001951343057637382,
      "loss": 4.3269,
      "step": 26100
    },
    {
      "epoch": 0.024338744804508775,
      "grad_norm": 1.2818320989608765,
      "learning_rate": 0.000195132441441643,
      "loss": 4.0483,
      "step": 26110
    },
    {
      "epoch": 0.024348066422587866,
      "grad_norm": 1.3007680177688599,
      "learning_rate": 0.0001951305771195478,
      "loss": 4.0713,
      "step": 26120
    },
    {
      "epoch": 0.024357388040666957,
      "grad_norm": 1.2240482568740845,
      "learning_rate": 0.0001951287127974526,
      "loss": 4.1875,
      "step": 26130
    },
    {
      "epoch": 0.024366709658746048,
      "grad_norm": 1.2694458961486816,
      "learning_rate": 0.0001951268484753574,
      "loss": 4.2172,
      "step": 26140
    },
    {
      "epoch": 0.024376031276825142,
      "grad_norm": 1.3325135707855225,
      "learning_rate": 0.0001951249841532622,
      "loss": 4.2186,
      "step": 26150
    },
    {
      "epoch": 0.024385352894904233,
      "grad_norm": 1.312690019607544,
      "learning_rate": 0.000195123119831167,
      "loss": 4.3795,
      "step": 26160
    },
    {
      "epoch": 0.024394674512983324,
      "grad_norm": 1.1456608772277832,
      "learning_rate": 0.0001951212555090718,
      "loss": 4.2491,
      "step": 26170
    },
    {
      "epoch": 0.024403996131062415,
      "grad_norm": 1.25475013256073,
      "learning_rate": 0.00019511939118697658,
      "loss": 4.1969,
      "step": 26180
    },
    {
      "epoch": 0.02441331774914151,
      "grad_norm": 1.2015897035598755,
      "learning_rate": 0.00019511752686488142,
      "loss": 4.3338,
      "step": 26190
    },
    {
      "epoch": 0.0244226393672206,
      "grad_norm": 1.0775506496429443,
      "learning_rate": 0.0001951156625427862,
      "loss": 4.274,
      "step": 26200
    },
    {
      "epoch": 0.02443196098529969,
      "grad_norm": 1.1862497329711914,
      "learning_rate": 0.000195113798220691,
      "loss": 4.2971,
      "step": 26210
    },
    {
      "epoch": 0.024441282603378783,
      "grad_norm": 1.227306842803955,
      "learning_rate": 0.0001951119338985958,
      "loss": 4.3902,
      "step": 26220
    },
    {
      "epoch": 0.024450604221457877,
      "grad_norm": 1.0895674228668213,
      "learning_rate": 0.0001951100695765006,
      "loss": 4.2206,
      "step": 26230
    },
    {
      "epoch": 0.024459925839536968,
      "grad_norm": 1.1198301315307617,
      "learning_rate": 0.0001951082052544054,
      "loss": 4.2971,
      "step": 26240
    },
    {
      "epoch": 0.02446924745761606,
      "grad_norm": 1.1459205150604248,
      "learning_rate": 0.0001951063409323102,
      "loss": 4.1632,
      "step": 26250
    },
    {
      "epoch": 0.02447856907569515,
      "grad_norm": 1.1950041055679321,
      "learning_rate": 0.000195104476610215,
      "loss": 4.3388,
      "step": 26260
    },
    {
      "epoch": 0.02448789069377424,
      "grad_norm": 1.3040142059326172,
      "learning_rate": 0.0001951026122881198,
      "loss": 4.2315,
      "step": 26270
    },
    {
      "epoch": 0.024497212311853336,
      "grad_norm": 1.0213428735733032,
      "learning_rate": 0.0001951007479660246,
      "loss": 4.2879,
      "step": 26280
    },
    {
      "epoch": 0.024506533929932427,
      "grad_norm": 1.1493016481399536,
      "learning_rate": 0.0001950988836439294,
      "loss": 4.4946,
      "step": 26290
    },
    {
      "epoch": 0.024515855548011518,
      "grad_norm": 1.276241421699524,
      "learning_rate": 0.0001950970193218342,
      "loss": 4.2646,
      "step": 26300
    },
    {
      "epoch": 0.02452517716609061,
      "grad_norm": 1.1096677780151367,
      "learning_rate": 0.000195095154999739,
      "loss": 4.1952,
      "step": 26310
    },
    {
      "epoch": 0.024534498784169703,
      "grad_norm": 1.2251757383346558,
      "learning_rate": 0.00019509329067764382,
      "loss": 4.3848,
      "step": 26320
    },
    {
      "epoch": 0.024543820402248794,
      "grad_norm": 1.0263757705688477,
      "learning_rate": 0.0001950914263555486,
      "loss": 4.2936,
      "step": 26330
    },
    {
      "epoch": 0.024553142020327885,
      "grad_norm": 1.175191044807434,
      "learning_rate": 0.0001950895620334534,
      "loss": 4.3406,
      "step": 26340
    },
    {
      "epoch": 0.024562463638406976,
      "grad_norm": 1.0343765020370483,
      "learning_rate": 0.0001950876977113582,
      "loss": 4.4829,
      "step": 26350
    },
    {
      "epoch": 0.02457178525648607,
      "grad_norm": 1.0078182220458984,
      "learning_rate": 0.00019508583338926302,
      "loss": 4.0843,
      "step": 26360
    },
    {
      "epoch": 0.024581106874565162,
      "grad_norm": 1.1242882013320923,
      "learning_rate": 0.00019508396906716782,
      "loss": 4.3581,
      "step": 26370
    },
    {
      "epoch": 0.024590428492644253,
      "grad_norm": 0.9683307409286499,
      "learning_rate": 0.0001950821047450726,
      "loss": 4.3089,
      "step": 26380
    },
    {
      "epoch": 0.024599750110723344,
      "grad_norm": 1.2668174505233765,
      "learning_rate": 0.0001950802404229774,
      "loss": 4.0767,
      "step": 26390
    },
    {
      "epoch": 0.024609071728802438,
      "grad_norm": 1.3483976125717163,
      "learning_rate": 0.0001950783761008822,
      "loss": 4.3309,
      "step": 26400
    },
    {
      "epoch": 0.02461839334688153,
      "grad_norm": 1.1598643064498901,
      "learning_rate": 0.00019507651177878702,
      "loss": 4.2844,
      "step": 26410
    },
    {
      "epoch": 0.02462771496496062,
      "grad_norm": 1.0849692821502686,
      "learning_rate": 0.00019507464745669183,
      "loss": 4.157,
      "step": 26420
    },
    {
      "epoch": 0.02463703658303971,
      "grad_norm": 1.072091817855835,
      "learning_rate": 0.0001950727831345966,
      "loss": 4.2372,
      "step": 26430
    },
    {
      "epoch": 0.024646358201118802,
      "grad_norm": 1.2443921566009521,
      "learning_rate": 0.0001950709188125014,
      "loss": 4.3502,
      "step": 26440
    },
    {
      "epoch": 0.024655679819197897,
      "grad_norm": 0.8131055235862732,
      "learning_rate": 0.0001950690544904062,
      "loss": 4.1696,
      "step": 26450
    },
    {
      "epoch": 0.024665001437276988,
      "grad_norm": 1.258666753768921,
      "learning_rate": 0.000195067190168311,
      "loss": 4.2789,
      "step": 26460
    },
    {
      "epoch": 0.02467432305535608,
      "grad_norm": 1.0803370475769043,
      "learning_rate": 0.00019506532584621583,
      "loss": 4.3654,
      "step": 26470
    },
    {
      "epoch": 0.02468364467343517,
      "grad_norm": 1.2808382511138916,
      "learning_rate": 0.0001950634615241206,
      "loss": 4.2448,
      "step": 26480
    },
    {
      "epoch": 0.024692966291514264,
      "grad_norm": 1.1501256227493286,
      "learning_rate": 0.00019506159720202542,
      "loss": 4.2755,
      "step": 26490
    },
    {
      "epoch": 0.024702287909593355,
      "grad_norm": 1.309529423713684,
      "learning_rate": 0.0001950597328799302,
      "loss": 4.185,
      "step": 26500
    },
    {
      "epoch": 0.024711609527672446,
      "grad_norm": 0.9050287008285522,
      "learning_rate": 0.000195057868557835,
      "loss": 4.4224,
      "step": 26510
    },
    {
      "epoch": 0.024720931145751537,
      "grad_norm": 1.145747423171997,
      "learning_rate": 0.0001950560042357398,
      "loss": 4.1551,
      "step": 26520
    },
    {
      "epoch": 0.024730252763830632,
      "grad_norm": 1.0934703350067139,
      "learning_rate": 0.00019505413991364461,
      "loss": 4.0759,
      "step": 26530
    },
    {
      "epoch": 0.024739574381909723,
      "grad_norm": 0.8947463035583496,
      "learning_rate": 0.00019505227559154942,
      "loss": 4.0686,
      "step": 26540
    },
    {
      "epoch": 0.024748895999988814,
      "grad_norm": 1.0900859832763672,
      "learning_rate": 0.0001950504112694542,
      "loss": 4.3336,
      "step": 26550
    },
    {
      "epoch": 0.024758217618067905,
      "grad_norm": 0.894111692905426,
      "learning_rate": 0.000195048546947359,
      "loss": 4.1781,
      "step": 26560
    },
    {
      "epoch": 0.024767539236147,
      "grad_norm": 1.1098744869232178,
      "learning_rate": 0.0001950466826252638,
      "loss": 4.3089,
      "step": 26570
    },
    {
      "epoch": 0.02477686085422609,
      "grad_norm": 1.1040693521499634,
      "learning_rate": 0.00019504481830316862,
      "loss": 4.3703,
      "step": 26580
    },
    {
      "epoch": 0.02478618247230518,
      "grad_norm": 1.0802429914474487,
      "learning_rate": 0.00019504295398107343,
      "loss": 4.2781,
      "step": 26590
    },
    {
      "epoch": 0.024795504090384272,
      "grad_norm": 1.036103367805481,
      "learning_rate": 0.0001950410896589782,
      "loss": 4.3728,
      "step": 26600
    },
    {
      "epoch": 0.024804825708463363,
      "grad_norm": 1.0853430032730103,
      "learning_rate": 0.000195039225336883,
      "loss": 4.3509,
      "step": 26610
    },
    {
      "epoch": 0.024814147326542458,
      "grad_norm": 1.2047678232192993,
      "learning_rate": 0.00019503736101478782,
      "loss": 4.3309,
      "step": 26620
    },
    {
      "epoch": 0.02482346894462155,
      "grad_norm": 1.2488569021224976,
      "learning_rate": 0.00019503549669269262,
      "loss": 4.3815,
      "step": 26630
    },
    {
      "epoch": 0.02483279056270064,
      "grad_norm": 1.0649772882461548,
      "learning_rate": 0.00019503363237059743,
      "loss": 4.0208,
      "step": 26640
    },
    {
      "epoch": 0.02484211218077973,
      "grad_norm": 1.1589698791503906,
      "learning_rate": 0.0001950317680485022,
      "loss": 4.251,
      "step": 26650
    },
    {
      "epoch": 0.024851433798858825,
      "grad_norm": 1.2540165185928345,
      "learning_rate": 0.00019502990372640701,
      "loss": 4.2688,
      "step": 26660
    },
    {
      "epoch": 0.024860755416937916,
      "grad_norm": 1.110235571861267,
      "learning_rate": 0.00019502803940431182,
      "loss": 4.2173,
      "step": 26670
    },
    {
      "epoch": 0.024870077035017008,
      "grad_norm": 1.200618028640747,
      "learning_rate": 0.0001950261750822166,
      "loss": 4.3415,
      "step": 26680
    },
    {
      "epoch": 0.0248793986530961,
      "grad_norm": 1.2330814599990845,
      "learning_rate": 0.00019502431076012143,
      "loss": 4.3789,
      "step": 26690
    },
    {
      "epoch": 0.024888720271175193,
      "grad_norm": 1.0663968324661255,
      "learning_rate": 0.0001950224464380262,
      "loss": 4.2421,
      "step": 26700
    },
    {
      "epoch": 0.024898041889254284,
      "grad_norm": 1.0813602209091187,
      "learning_rate": 0.00019502058211593102,
      "loss": 4.363,
      "step": 26710
    },
    {
      "epoch": 0.024907363507333375,
      "grad_norm": 1.1093820333480835,
      "learning_rate": 0.00019501871779383583,
      "loss": 4.1934,
      "step": 26720
    },
    {
      "epoch": 0.024916685125412466,
      "grad_norm": 1.1404876708984375,
      "learning_rate": 0.0001950168534717406,
      "loss": 4.23,
      "step": 26730
    },
    {
      "epoch": 0.02492600674349156,
      "grad_norm": 1.0336272716522217,
      "learning_rate": 0.0001950149891496454,
      "loss": 4.1131,
      "step": 26740
    },
    {
      "epoch": 0.02493532836157065,
      "grad_norm": 1.2391999959945679,
      "learning_rate": 0.00019501312482755022,
      "loss": 4.487,
      "step": 26750
    },
    {
      "epoch": 0.024944649979649743,
      "grad_norm": 1.11328125,
      "learning_rate": 0.00019501126050545502,
      "loss": 4.2949,
      "step": 26760
    },
    {
      "epoch": 0.024953971597728834,
      "grad_norm": 1.0190496444702148,
      "learning_rate": 0.00019500939618335983,
      "loss": 4.4316,
      "step": 26770
    },
    {
      "epoch": 0.024963293215807925,
      "grad_norm": 1.0995476245880127,
      "learning_rate": 0.0001950075318612646,
      "loss": 4.2331,
      "step": 26780
    },
    {
      "epoch": 0.02497261483388702,
      "grad_norm": 1.1929612159729004,
      "learning_rate": 0.00019500566753916942,
      "loss": 4.349,
      "step": 26790
    },
    {
      "epoch": 0.02498193645196611,
      "grad_norm": 1.1772637367248535,
      "learning_rate": 0.00019500380321707422,
      "loss": 4.2666,
      "step": 26800
    },
    {
      "epoch": 0.0249912580700452,
      "grad_norm": 1.1157186031341553,
      "learning_rate": 0.00019500193889497903,
      "loss": 4.3326,
      "step": 26810
    },
    {
      "epoch": 0.025000579688124292,
      "grad_norm": 1.0779684782028198,
      "learning_rate": 0.0001950000745728838,
      "loss": 4.2007,
      "step": 26820
    },
    {
      "epoch": 0.025009901306203387,
      "grad_norm": 1.1132296323776245,
      "learning_rate": 0.00019499821025078861,
      "loss": 4.2927,
      "step": 26830
    },
    {
      "epoch": 0.025019222924282478,
      "grad_norm": 1.0884636640548706,
      "learning_rate": 0.00019499634592869342,
      "loss": 4.2945,
      "step": 26840
    },
    {
      "epoch": 0.02502854454236157,
      "grad_norm": 1.260109305381775,
      "learning_rate": 0.0001949944816065982,
      "loss": 4.3731,
      "step": 26850
    },
    {
      "epoch": 0.02503786616044066,
      "grad_norm": 1.1866180896759033,
      "learning_rate": 0.00019499261728450303,
      "loss": 4.2327,
      "step": 26860
    },
    {
      "epoch": 0.025047187778519754,
      "grad_norm": 1.2331770658493042,
      "learning_rate": 0.0001949907529624078,
      "loss": 4.1782,
      "step": 26870
    },
    {
      "epoch": 0.025056509396598845,
      "grad_norm": 1.1961466073989868,
      "learning_rate": 0.00019498888864031262,
      "loss": 4.3988,
      "step": 26880
    },
    {
      "epoch": 0.025065831014677936,
      "grad_norm": 0.9198324084281921,
      "learning_rate": 0.00019498702431821742,
      "loss": 4.219,
      "step": 26890
    },
    {
      "epoch": 0.025075152632757027,
      "grad_norm": 1.2005844116210938,
      "learning_rate": 0.0001949851599961222,
      "loss": 4.1328,
      "step": 26900
    },
    {
      "epoch": 0.02508447425083612,
      "grad_norm": 1.1314808130264282,
      "learning_rate": 0.000194983295674027,
      "loss": 4.3626,
      "step": 26910
    },
    {
      "epoch": 0.025093795868915213,
      "grad_norm": 1.2265046834945679,
      "learning_rate": 0.00019498143135193182,
      "loss": 4.3265,
      "step": 26920
    },
    {
      "epoch": 0.025103117486994304,
      "grad_norm": 1.091578483581543,
      "learning_rate": 0.00019497956702983662,
      "loss": 4.3283,
      "step": 26930
    },
    {
      "epoch": 0.025112439105073395,
      "grad_norm": 1.0724936723709106,
      "learning_rate": 0.00019497770270774143,
      "loss": 4.23,
      "step": 26940
    },
    {
      "epoch": 0.025121760723152486,
      "grad_norm": 1.070136308670044,
      "learning_rate": 0.0001949758383856462,
      "loss": 4.1819,
      "step": 26950
    },
    {
      "epoch": 0.02513108234123158,
      "grad_norm": 1.1180752515792847,
      "learning_rate": 0.00019497397406355101,
      "loss": 4.3498,
      "step": 26960
    },
    {
      "epoch": 0.02514040395931067,
      "grad_norm": 1.2191082239151,
      "learning_rate": 0.00019497210974145582,
      "loss": 4.1884,
      "step": 26970
    },
    {
      "epoch": 0.025149725577389762,
      "grad_norm": 1.119485855102539,
      "learning_rate": 0.00019497024541936063,
      "loss": 4.2972,
      "step": 26980
    },
    {
      "epoch": 0.025159047195468853,
      "grad_norm": 1.134010910987854,
      "learning_rate": 0.00019496838109726543,
      "loss": 4.2688,
      "step": 26990
    },
    {
      "epoch": 0.025168368813547948,
      "grad_norm": 0.9564749598503113,
      "learning_rate": 0.0001949665167751702,
      "loss": 4.1645,
      "step": 27000
    },
    {
      "epoch": 0.02517769043162704,
      "grad_norm": 1.0403841733932495,
      "learning_rate": 0.00019496465245307502,
      "loss": 4.2729,
      "step": 27010
    },
    {
      "epoch": 0.02518701204970613,
      "grad_norm": 1.11350679397583,
      "learning_rate": 0.00019496278813097983,
      "loss": 4.0608,
      "step": 27020
    },
    {
      "epoch": 0.02519633366778522,
      "grad_norm": 1.1223313808441162,
      "learning_rate": 0.00019496092380888463,
      "loss": 4.3849,
      "step": 27030
    },
    {
      "epoch": 0.025205655285864315,
      "grad_norm": 1.1163564920425415,
      "learning_rate": 0.00019495905948678944,
      "loss": 4.1702,
      "step": 27040
    },
    {
      "epoch": 0.025214976903943406,
      "grad_norm": 1.1855180263519287,
      "learning_rate": 0.00019495719516469422,
      "loss": 4.3315,
      "step": 27050
    },
    {
      "epoch": 0.025224298522022497,
      "grad_norm": 1.2362064123153687,
      "learning_rate": 0.00019495533084259902,
      "loss": 4.1444,
      "step": 27060
    },
    {
      "epoch": 0.02523362014010159,
      "grad_norm": 1.20620596408844,
      "learning_rate": 0.0001949534665205038,
      "loss": 4.3334,
      "step": 27070
    },
    {
      "epoch": 0.025242941758180683,
      "grad_norm": 1.156413197517395,
      "learning_rate": 0.00019495160219840864,
      "loss": 4.4344,
      "step": 27080
    },
    {
      "epoch": 0.025252263376259774,
      "grad_norm": 1.3575183153152466,
      "learning_rate": 0.00019494973787631344,
      "loss": 4.1922,
      "step": 27090
    },
    {
      "epoch": 0.025261584994338865,
      "grad_norm": 1.153159260749817,
      "learning_rate": 0.00019494787355421822,
      "loss": 4.2356,
      "step": 27100
    },
    {
      "epoch": 0.025270906612417956,
      "grad_norm": 1.1401430368423462,
      "learning_rate": 0.00019494600923212303,
      "loss": 4.4406,
      "step": 27110
    },
    {
      "epoch": 0.025280228230497047,
      "grad_norm": 1.1915273666381836,
      "learning_rate": 0.0001949441449100278,
      "loss": 4.3126,
      "step": 27120
    },
    {
      "epoch": 0.02528954984857614,
      "grad_norm": 0.9761599898338318,
      "learning_rate": 0.0001949422805879326,
      "loss": 4.3136,
      "step": 27130
    },
    {
      "epoch": 0.025298871466655232,
      "grad_norm": 1.1661889553070068,
      "learning_rate": 0.00019494041626583745,
      "loss": 4.4377,
      "step": 27140
    },
    {
      "epoch": 0.025308193084734323,
      "grad_norm": 1.1335996389389038,
      "learning_rate": 0.00019493855194374223,
      "loss": 4.2359,
      "step": 27150
    },
    {
      "epoch": 0.025317514702813414,
      "grad_norm": 1.1962746381759644,
      "learning_rate": 0.00019493668762164703,
      "loss": 4.2355,
      "step": 27160
    },
    {
      "epoch": 0.02532683632089251,
      "grad_norm": 1.2092242240905762,
      "learning_rate": 0.0001949348232995518,
      "loss": 4.1992,
      "step": 27170
    },
    {
      "epoch": 0.0253361579389716,
      "grad_norm": 1.205977439880371,
      "learning_rate": 0.00019493295897745662,
      "loss": 4.4183,
      "step": 27180
    },
    {
      "epoch": 0.02534547955705069,
      "grad_norm": 1.1727303266525269,
      "learning_rate": 0.00019493109465536142,
      "loss": 4.293,
      "step": 27190
    },
    {
      "epoch": 0.025354801175129782,
      "grad_norm": 1.2692384719848633,
      "learning_rate": 0.00019492923033326623,
      "loss": 4.4206,
      "step": 27200
    },
    {
      "epoch": 0.025364122793208876,
      "grad_norm": 1.13283109664917,
      "learning_rate": 0.00019492736601117104,
      "loss": 4.4273,
      "step": 27210
    },
    {
      "epoch": 0.025373444411287967,
      "grad_norm": 1.059680700302124,
      "learning_rate": 0.00019492550168907582,
      "loss": 4.3351,
      "step": 27220
    },
    {
      "epoch": 0.02538276602936706,
      "grad_norm": 0.9791172742843628,
      "learning_rate": 0.00019492363736698062,
      "loss": 4.36,
      "step": 27230
    },
    {
      "epoch": 0.02539208764744615,
      "grad_norm": 1.022316336631775,
      "learning_rate": 0.00019492177304488543,
      "loss": 4.4401,
      "step": 27240
    },
    {
      "epoch": 0.025401409265525244,
      "grad_norm": 1.0657271146774292,
      "learning_rate": 0.00019491990872279023,
      "loss": 4.2419,
      "step": 27250
    },
    {
      "epoch": 0.025410730883604335,
      "grad_norm": 1.2046310901641846,
      "learning_rate": 0.00019491804440069504,
      "loss": 4.1599,
      "step": 27260
    },
    {
      "epoch": 0.025420052501683426,
      "grad_norm": 1.1195133924484253,
      "learning_rate": 0.00019491618007859982,
      "loss": 4.4975,
      "step": 27270
    },
    {
      "epoch": 0.025429374119762517,
      "grad_norm": 1.1676751375198364,
      "learning_rate": 0.00019491431575650463,
      "loss": 4.247,
      "step": 27280
    },
    {
      "epoch": 0.025438695737841608,
      "grad_norm": 1.0449965000152588,
      "learning_rate": 0.00019491245143440943,
      "loss": 4.4316,
      "step": 27290
    },
    {
      "epoch": 0.025448017355920702,
      "grad_norm": 1.218698263168335,
      "learning_rate": 0.0001949105871123142,
      "loss": 4.2482,
      "step": 27300
    },
    {
      "epoch": 0.025457338973999794,
      "grad_norm": 1.3054856061935425,
      "learning_rate": 0.00019490872279021905,
      "loss": 4.2283,
      "step": 27310
    },
    {
      "epoch": 0.025466660592078885,
      "grad_norm": 1.3127882480621338,
      "learning_rate": 0.00019490685846812382,
      "loss": 4.3748,
      "step": 27320
    },
    {
      "epoch": 0.025475982210157976,
      "grad_norm": 1.2586358785629272,
      "learning_rate": 0.00019490499414602863,
      "loss": 4.1616,
      "step": 27330
    },
    {
      "epoch": 0.02548530382823707,
      "grad_norm": 1.1028530597686768,
      "learning_rate": 0.00019490312982393344,
      "loss": 4.4898,
      "step": 27340
    },
    {
      "epoch": 0.02549462544631616,
      "grad_norm": 1.028075933456421,
      "learning_rate": 0.00019490126550183822,
      "loss": 4.2186,
      "step": 27350
    },
    {
      "epoch": 0.025503947064395252,
      "grad_norm": 1.097214698791504,
      "learning_rate": 0.00019489940117974302,
      "loss": 4.3236,
      "step": 27360
    },
    {
      "epoch": 0.025513268682474343,
      "grad_norm": 1.02894127368927,
      "learning_rate": 0.00019489753685764783,
      "loss": 4.2235,
      "step": 27370
    },
    {
      "epoch": 0.025522590300553438,
      "grad_norm": 1.360142707824707,
      "learning_rate": 0.00019489567253555264,
      "loss": 4.1145,
      "step": 27380
    },
    {
      "epoch": 0.02553191191863253,
      "grad_norm": 1.339141845703125,
      "learning_rate": 0.00019489380821345744,
      "loss": 4.2881,
      "step": 27390
    },
    {
      "epoch": 0.02554123353671162,
      "grad_norm": 1.1225017309188843,
      "learning_rate": 0.00019489194389136222,
      "loss": 4.3992,
      "step": 27400
    },
    {
      "epoch": 0.02555055515479071,
      "grad_norm": 1.1247949600219727,
      "learning_rate": 0.00019489007956926703,
      "loss": 4.2693,
      "step": 27410
    },
    {
      "epoch": 0.025559876772869805,
      "grad_norm": 1.125213384628296,
      "learning_rate": 0.00019488821524717183,
      "loss": 4.276,
      "step": 27420
    },
    {
      "epoch": 0.025569198390948896,
      "grad_norm": 0.9697474241256714,
      "learning_rate": 0.00019488635092507664,
      "loss": 4.4534,
      "step": 27430
    },
    {
      "epoch": 0.025578520009027987,
      "grad_norm": 1.1920273303985596,
      "learning_rate": 0.00019488448660298145,
      "loss": 4.1556,
      "step": 27440
    },
    {
      "epoch": 0.025587841627107078,
      "grad_norm": 1.1319034099578857,
      "learning_rate": 0.00019488262228088623,
      "loss": 4.102,
      "step": 27450
    },
    {
      "epoch": 0.02559716324518617,
      "grad_norm": 1.1034579277038574,
      "learning_rate": 0.00019488075795879103,
      "loss": 4.1413,
      "step": 27460
    },
    {
      "epoch": 0.025606484863265264,
      "grad_norm": 0.9388127326965332,
      "learning_rate": 0.00019487889363669584,
      "loss": 4.2571,
      "step": 27470
    },
    {
      "epoch": 0.025615806481344355,
      "grad_norm": 1.1541668176651,
      "learning_rate": 0.00019487702931460064,
      "loss": 4.2289,
      "step": 27480
    },
    {
      "epoch": 0.025625128099423446,
      "grad_norm": 1.1796094179153442,
      "learning_rate": 0.00019487516499250545,
      "loss": 4.179,
      "step": 27490
    },
    {
      "epoch": 0.025634449717502537,
      "grad_norm": 1.3244050741195679,
      "learning_rate": 0.00019487330067041023,
      "loss": 4.0789,
      "step": 27500
    },
    {
      "epoch": 0.02564377133558163,
      "grad_norm": 1.3902020454406738,
      "learning_rate": 0.00019487143634831504,
      "loss": 4.3607,
      "step": 27510
    },
    {
      "epoch": 0.025653092953660722,
      "grad_norm": 1.1120481491088867,
      "learning_rate": 0.00019486957202621982,
      "loss": 4.3251,
      "step": 27520
    },
    {
      "epoch": 0.025662414571739813,
      "grad_norm": 1.1542820930480957,
      "learning_rate": 0.00019486770770412465,
      "loss": 4.2395,
      "step": 27530
    },
    {
      "epoch": 0.025671736189818904,
      "grad_norm": 1.105812430381775,
      "learning_rate": 0.00019486584338202946,
      "loss": 4.2961,
      "step": 27540
    },
    {
      "epoch": 0.025681057807898,
      "grad_norm": 1.3773430585861206,
      "learning_rate": 0.00019486397905993423,
      "loss": 4.4097,
      "step": 27550
    },
    {
      "epoch": 0.02569037942597709,
      "grad_norm": 1.1232496500015259,
      "learning_rate": 0.00019486211473783904,
      "loss": 4.361,
      "step": 27560
    },
    {
      "epoch": 0.02569970104405618,
      "grad_norm": 1.2559623718261719,
      "learning_rate": 0.00019486025041574382,
      "loss": 4.3052,
      "step": 27570
    },
    {
      "epoch": 0.025709022662135272,
      "grad_norm": 0.9566503763198853,
      "learning_rate": 0.00019485838609364863,
      "loss": 4.4922,
      "step": 27580
    },
    {
      "epoch": 0.025718344280214366,
      "grad_norm": 1.2518701553344727,
      "learning_rate": 0.00019485652177155346,
      "loss": 4.2834,
      "step": 27590
    },
    {
      "epoch": 0.025727665898293457,
      "grad_norm": 1.2910116910934448,
      "learning_rate": 0.00019485465744945824,
      "loss": 4.2513,
      "step": 27600
    },
    {
      "epoch": 0.025736987516372548,
      "grad_norm": 1.0042388439178467,
      "learning_rate": 0.00019485279312736305,
      "loss": 4.4463,
      "step": 27610
    },
    {
      "epoch": 0.02574630913445164,
      "grad_norm": 1.3178033828735352,
      "learning_rate": 0.00019485092880526782,
      "loss": 4.321,
      "step": 27620
    },
    {
      "epoch": 0.02575563075253073,
      "grad_norm": 0.9830506443977356,
      "learning_rate": 0.00019484906448317263,
      "loss": 4.4695,
      "step": 27630
    },
    {
      "epoch": 0.025764952370609825,
      "grad_norm": 1.2199665307998657,
      "learning_rate": 0.00019484720016107744,
      "loss": 4.2515,
      "step": 27640
    },
    {
      "epoch": 0.025774273988688916,
      "grad_norm": 1.1124294996261597,
      "learning_rate": 0.00019484533583898224,
      "loss": 3.9199,
      "step": 27650
    },
    {
      "epoch": 0.025783595606768007,
      "grad_norm": 1.0375221967697144,
      "learning_rate": 0.00019484347151688705,
      "loss": 4.2068,
      "step": 27660
    },
    {
      "epoch": 0.025792917224847098,
      "grad_norm": 1.4398577213287354,
      "learning_rate": 0.00019484160719479183,
      "loss": 4.2631,
      "step": 27670
    },
    {
      "epoch": 0.025802238842926192,
      "grad_norm": 0.9595447778701782,
      "learning_rate": 0.00019483974287269664,
      "loss": 4.332,
      "step": 27680
    },
    {
      "epoch": 0.025811560461005283,
      "grad_norm": 0.8980116248130798,
      "learning_rate": 0.00019483787855060144,
      "loss": 4.3124,
      "step": 27690
    },
    {
      "epoch": 0.025820882079084374,
      "grad_norm": 1.3157520294189453,
      "learning_rate": 0.00019483601422850625,
      "loss": 4.1475,
      "step": 27700
    },
    {
      "epoch": 0.025830203697163465,
      "grad_norm": 1.0293772220611572,
      "learning_rate": 0.00019483414990641105,
      "loss": 4.1369,
      "step": 27710
    },
    {
      "epoch": 0.02583952531524256,
      "grad_norm": 1.2900476455688477,
      "learning_rate": 0.00019483228558431583,
      "loss": 4.3536,
      "step": 27720
    },
    {
      "epoch": 0.02584884693332165,
      "grad_norm": 1.0627479553222656,
      "learning_rate": 0.00019483042126222064,
      "loss": 4.1382,
      "step": 27730
    },
    {
      "epoch": 0.025858168551400742,
      "grad_norm": 1.013731837272644,
      "learning_rate": 0.00019482855694012545,
      "loss": 4.3093,
      "step": 27740
    },
    {
      "epoch": 0.025867490169479833,
      "grad_norm": 1.0203777551651,
      "learning_rate": 0.00019482669261803022,
      "loss": 4.1701,
      "step": 27750
    },
    {
      "epoch": 0.025876811787558927,
      "grad_norm": 1.140067219734192,
      "learning_rate": 0.00019482482829593506,
      "loss": 4.1987,
      "step": 27760
    },
    {
      "epoch": 0.02588613340563802,
      "grad_norm": 0.9398214817047119,
      "learning_rate": 0.00019482296397383984,
      "loss": 4.1444,
      "step": 27770
    },
    {
      "epoch": 0.02589545502371711,
      "grad_norm": 1.0490539073944092,
      "learning_rate": 0.00019482109965174464,
      "loss": 4.2008,
      "step": 27780
    },
    {
      "epoch": 0.0259047766417962,
      "grad_norm": 1.1988731622695923,
      "learning_rate": 0.00019481923532964945,
      "loss": 4.2229,
      "step": 27790
    },
    {
      "epoch": 0.02591409825987529,
      "grad_norm": 1.1112223863601685,
      "learning_rate": 0.00019481737100755423,
      "loss": 4.2344,
      "step": 27800
    },
    {
      "epoch": 0.025923419877954386,
      "grad_norm": 1.151888132095337,
      "learning_rate": 0.00019481550668545904,
      "loss": 4.2579,
      "step": 27810
    },
    {
      "epoch": 0.025932741496033477,
      "grad_norm": 0.9487802982330322,
      "learning_rate": 0.00019481364236336384,
      "loss": 4.3528,
      "step": 27820
    },
    {
      "epoch": 0.025942063114112568,
      "grad_norm": 1.102232575416565,
      "learning_rate": 0.00019481177804126865,
      "loss": 4.2104,
      "step": 27830
    },
    {
      "epoch": 0.02595138473219166,
      "grad_norm": 1.1896731853485107,
      "learning_rate": 0.00019480991371917345,
      "loss": 4.3368,
      "step": 27840
    },
    {
      "epoch": 0.025960706350270753,
      "grad_norm": 0.9201789498329163,
      "learning_rate": 0.00019480804939707823,
      "loss": 4.3597,
      "step": 27850
    },
    {
      "epoch": 0.025970027968349844,
      "grad_norm": 1.157020926475525,
      "learning_rate": 0.00019480618507498304,
      "loss": 4.1095,
      "step": 27860
    },
    {
      "epoch": 0.025979349586428935,
      "grad_norm": 0.9741753935813904,
      "learning_rate": 0.00019480432075288785,
      "loss": 4.2048,
      "step": 27870
    },
    {
      "epoch": 0.025988671204508026,
      "grad_norm": 1.2094454765319824,
      "learning_rate": 0.00019480245643079265,
      "loss": 4.0623,
      "step": 27880
    },
    {
      "epoch": 0.02599799282258712,
      "grad_norm": 1.1784563064575195,
      "learning_rate": 0.00019480059210869746,
      "loss": 4.2734,
      "step": 27890
    },
    {
      "epoch": 0.026007314440666212,
      "grad_norm": 1.169832706451416,
      "learning_rate": 0.00019479872778660224,
      "loss": 4.36,
      "step": 27900
    },
    {
      "epoch": 0.026016636058745303,
      "grad_norm": 1.10849928855896,
      "learning_rate": 0.00019479686346450704,
      "loss": 4.3121,
      "step": 27910
    },
    {
      "epoch": 0.026025957676824394,
      "grad_norm": 1.0116581916809082,
      "learning_rate": 0.00019479499914241185,
      "loss": 4.2837,
      "step": 27920
    },
    {
      "epoch": 0.02603527929490349,
      "grad_norm": 1.3350141048431396,
      "learning_rate": 0.00019479313482031666,
      "loss": 4.173,
      "step": 27930
    },
    {
      "epoch": 0.02604460091298258,
      "grad_norm": 1.073904037475586,
      "learning_rate": 0.00019479127049822146,
      "loss": 4.3683,
      "step": 27940
    },
    {
      "epoch": 0.02605392253106167,
      "grad_norm": 1.5044313669204712,
      "learning_rate": 0.00019478940617612624,
      "loss": 4.3531,
      "step": 27950
    },
    {
      "epoch": 0.02606324414914076,
      "grad_norm": 1.2254962921142578,
      "learning_rate": 0.00019478754185403105,
      "loss": 4.1683,
      "step": 27960
    },
    {
      "epoch": 0.026072565767219853,
      "grad_norm": 1.214392066001892,
      "learning_rate": 0.00019478567753193583,
      "loss": 4.1745,
      "step": 27970
    },
    {
      "epoch": 0.026081887385298947,
      "grad_norm": 1.0896146297454834,
      "learning_rate": 0.00019478381320984066,
      "loss": 4.5006,
      "step": 27980
    },
    {
      "epoch": 0.026091209003378038,
      "grad_norm": 1.159770131111145,
      "learning_rate": 0.00019478194888774547,
      "loss": 4.4609,
      "step": 27990
    },
    {
      "epoch": 0.02610053062145713,
      "grad_norm": 1.1267848014831543,
      "learning_rate": 0.00019478008456565025,
      "loss": 4.234,
      "step": 28000
    },
    {
      "epoch": 0.02610985223953622,
      "grad_norm": 1.2021052837371826,
      "learning_rate": 0.00019477822024355505,
      "loss": 4.2758,
      "step": 28010
    },
    {
      "epoch": 0.026119173857615315,
      "grad_norm": 1.1627283096313477,
      "learning_rate": 0.00019477635592145983,
      "loss": 4.1204,
      "step": 28020
    },
    {
      "epoch": 0.026128495475694406,
      "grad_norm": 1.282310128211975,
      "learning_rate": 0.00019477449159936464,
      "loss": 4.3247,
      "step": 28030
    },
    {
      "epoch": 0.026137817093773497,
      "grad_norm": 1.0624529123306274,
      "learning_rate": 0.00019477262727726947,
      "loss": 4.3292,
      "step": 28040
    },
    {
      "epoch": 0.026147138711852588,
      "grad_norm": 1.2989543676376343,
      "learning_rate": 0.00019477076295517425,
      "loss": 4.3845,
      "step": 28050
    },
    {
      "epoch": 0.026156460329931682,
      "grad_norm": 1.1644692420959473,
      "learning_rate": 0.00019476889863307906,
      "loss": 4.1752,
      "step": 28060
    },
    {
      "epoch": 0.026165781948010773,
      "grad_norm": 1.0323569774627686,
      "learning_rate": 0.00019476703431098384,
      "loss": 4.2585,
      "step": 28070
    },
    {
      "epoch": 0.026175103566089864,
      "grad_norm": 0.9959034323692322,
      "learning_rate": 0.00019476516998888864,
      "loss": 4.3378,
      "step": 28080
    },
    {
      "epoch": 0.026184425184168955,
      "grad_norm": 1.2495698928833008,
      "learning_rate": 0.00019476330566679345,
      "loss": 4.0509,
      "step": 28090
    },
    {
      "epoch": 0.02619374680224805,
      "grad_norm": 1.0344394445419312,
      "learning_rate": 0.00019476144134469826,
      "loss": 4.3643,
      "step": 28100
    },
    {
      "epoch": 0.02620306842032714,
      "grad_norm": 1.0963759422302246,
      "learning_rate": 0.00019475957702260306,
      "loss": 4.1813,
      "step": 28110
    },
    {
      "epoch": 0.02621239003840623,
      "grad_norm": 0.9991523623466492,
      "learning_rate": 0.00019475771270050784,
      "loss": 4.3144,
      "step": 28120
    },
    {
      "epoch": 0.026221711656485323,
      "grad_norm": 1.1508138179779053,
      "learning_rate": 0.00019475584837841265,
      "loss": 4.1226,
      "step": 28130
    },
    {
      "epoch": 0.026231033274564414,
      "grad_norm": 1.2254955768585205,
      "learning_rate": 0.00019475398405631745,
      "loss": 4.0354,
      "step": 28140
    },
    {
      "epoch": 0.026240354892643508,
      "grad_norm": 1.1771806478500366,
      "learning_rate": 0.00019475211973422226,
      "loss": 4.2897,
      "step": 28150
    },
    {
      "epoch": 0.0262496765107226,
      "grad_norm": 1.1224353313446045,
      "learning_rate": 0.00019475025541212707,
      "loss": 4.1798,
      "step": 28160
    },
    {
      "epoch": 0.02625899812880169,
      "grad_norm": 1.1317070722579956,
      "learning_rate": 0.00019474839109003185,
      "loss": 4.2297,
      "step": 28170
    },
    {
      "epoch": 0.02626831974688078,
      "grad_norm": 1.2650384902954102,
      "learning_rate": 0.00019474652676793665,
      "loss": 4.3893,
      "step": 28180
    },
    {
      "epoch": 0.026277641364959876,
      "grad_norm": 1.1106812953948975,
      "learning_rate": 0.00019474466244584143,
      "loss": 4.2143,
      "step": 28190
    },
    {
      "epoch": 0.026286962983038967,
      "grad_norm": 1.1223738193511963,
      "learning_rate": 0.00019474279812374624,
      "loss": 4.2795,
      "step": 28200
    },
    {
      "epoch": 0.026296284601118058,
      "grad_norm": 1.2385519742965698,
      "learning_rate": 0.00019474093380165107,
      "loss": 4.2809,
      "step": 28210
    },
    {
      "epoch": 0.02630560621919715,
      "grad_norm": 1.1840029954910278,
      "learning_rate": 0.00019473906947955585,
      "loss": 4.102,
      "step": 28220
    },
    {
      "epoch": 0.026314927837276243,
      "grad_norm": 1.09910249710083,
      "learning_rate": 0.00019473720515746066,
      "loss": 4.374,
      "step": 28230
    },
    {
      "epoch": 0.026324249455355334,
      "grad_norm": 1.090376853942871,
      "learning_rate": 0.00019473534083536544,
      "loss": 4.0298,
      "step": 28240
    },
    {
      "epoch": 0.026333571073434425,
      "grad_norm": 0.9261501431465149,
      "learning_rate": 0.00019473347651327024,
      "loss": 4.4591,
      "step": 28250
    },
    {
      "epoch": 0.026342892691513516,
      "grad_norm": 0.9547326564788818,
      "learning_rate": 0.00019473161219117505,
      "loss": 4.1812,
      "step": 28260
    },
    {
      "epoch": 0.02635221430959261,
      "grad_norm": 0.9868759512901306,
      "learning_rate": 0.00019472974786907985,
      "loss": 4.1775,
      "step": 28270
    },
    {
      "epoch": 0.026361535927671702,
      "grad_norm": 1.1174323558807373,
      "learning_rate": 0.00019472788354698466,
      "loss": 4.3867,
      "step": 28280
    },
    {
      "epoch": 0.026370857545750793,
      "grad_norm": 1.3655201196670532,
      "learning_rate": 0.00019472601922488944,
      "loss": 4.2346,
      "step": 28290
    },
    {
      "epoch": 0.026380179163829884,
      "grad_norm": 0.9716086983680725,
      "learning_rate": 0.00019472415490279425,
      "loss": 4.303,
      "step": 28300
    },
    {
      "epoch": 0.026389500781908975,
      "grad_norm": 1.015121340751648,
      "learning_rate": 0.00019472229058069905,
      "loss": 4.2691,
      "step": 28310
    },
    {
      "epoch": 0.02639882239998807,
      "grad_norm": 1.102526068687439,
      "learning_rate": 0.00019472042625860386,
      "loss": 4.2638,
      "step": 28320
    },
    {
      "epoch": 0.02640814401806716,
      "grad_norm": 1.1531914472579956,
      "learning_rate": 0.00019471856193650867,
      "loss": 4.1987,
      "step": 28330
    },
    {
      "epoch": 0.02641746563614625,
      "grad_norm": 1.0494741201400757,
      "learning_rate": 0.00019471669761441344,
      "loss": 4.3734,
      "step": 28340
    },
    {
      "epoch": 0.026426787254225342,
      "grad_norm": 1.1647107601165771,
      "learning_rate": 0.00019471483329231825,
      "loss": 4.3406,
      "step": 28350
    },
    {
      "epoch": 0.026436108872304437,
      "grad_norm": 1.1529090404510498,
      "learning_rate": 0.00019471296897022306,
      "loss": 4.286,
      "step": 28360
    },
    {
      "epoch": 0.026445430490383528,
      "grad_norm": 0.8858646750450134,
      "learning_rate": 0.00019471110464812786,
      "loss": 4.211,
      "step": 28370
    },
    {
      "epoch": 0.02645475210846262,
      "grad_norm": 1.1542376279830933,
      "learning_rate": 0.00019470924032603267,
      "loss": 4.2308,
      "step": 28380
    },
    {
      "epoch": 0.02646407372654171,
      "grad_norm": 1.0016499757766724,
      "learning_rate": 0.00019470737600393745,
      "loss": 4.1263,
      "step": 28390
    },
    {
      "epoch": 0.026473395344620804,
      "grad_norm": 1.1625216007232666,
      "learning_rate": 0.00019470551168184226,
      "loss": 4.2305,
      "step": 28400
    },
    {
      "epoch": 0.026482716962699895,
      "grad_norm": 1.1518560647964478,
      "learning_rate": 0.00019470364735974706,
      "loss": 4.2975,
      "step": 28410
    },
    {
      "epoch": 0.026492038580778986,
      "grad_norm": 1.2776798009872437,
      "learning_rate": 0.00019470178303765184,
      "loss": 4.1803,
      "step": 28420
    },
    {
      "epoch": 0.026501360198858077,
      "grad_norm": 1.0398823022842407,
      "learning_rate": 0.00019469991871555667,
      "loss": 4.2402,
      "step": 28430
    },
    {
      "epoch": 0.026510681816937172,
      "grad_norm": 1.2361820936203003,
      "learning_rate": 0.00019469805439346145,
      "loss": 4.2842,
      "step": 28440
    },
    {
      "epoch": 0.026520003435016263,
      "grad_norm": 1.2707715034484863,
      "learning_rate": 0.00019469619007136626,
      "loss": 4.3068,
      "step": 28450
    },
    {
      "epoch": 0.026529325053095354,
      "grad_norm": 0.9966992735862732,
      "learning_rate": 0.00019469432574927107,
      "loss": 4.1121,
      "step": 28460
    },
    {
      "epoch": 0.026538646671174445,
      "grad_norm": 1.2766801118850708,
      "learning_rate": 0.00019469246142717585,
      "loss": 4.3596,
      "step": 28470
    },
    {
      "epoch": 0.026547968289253536,
      "grad_norm": 1.0383092164993286,
      "learning_rate": 0.00019469059710508065,
      "loss": 4.2857,
      "step": 28480
    },
    {
      "epoch": 0.02655728990733263,
      "grad_norm": 1.1039258241653442,
      "learning_rate": 0.00019468873278298546,
      "loss": 4.1708,
      "step": 28490
    },
    {
      "epoch": 0.02656661152541172,
      "grad_norm": 1.1415913105010986,
      "learning_rate": 0.00019468686846089026,
      "loss": 4.3442,
      "step": 28500
    },
    {
      "epoch": 0.026575933143490812,
      "grad_norm": 0.8727560639381409,
      "learning_rate": 0.00019468500413879507,
      "loss": 4.3168,
      "step": 28510
    },
    {
      "epoch": 0.026585254761569903,
      "grad_norm": 1.0488190650939941,
      "learning_rate": 0.00019468313981669985,
      "loss": 4.136,
      "step": 28520
    },
    {
      "epoch": 0.026594576379648998,
      "grad_norm": 1.0090585947036743,
      "learning_rate": 0.00019468127549460466,
      "loss": 4.1101,
      "step": 28530
    },
    {
      "epoch": 0.02660389799772809,
      "grad_norm": 1.099616289138794,
      "learning_rate": 0.00019467941117250946,
      "loss": 4.3234,
      "step": 28540
    },
    {
      "epoch": 0.02661321961580718,
      "grad_norm": 1.1166173219680786,
      "learning_rate": 0.00019467754685041427,
      "loss": 4.3214,
      "step": 28550
    },
    {
      "epoch": 0.02662254123388627,
      "grad_norm": 1.0480883121490479,
      "learning_rate": 0.00019467568252831908,
      "loss": 4.1731,
      "step": 28560
    },
    {
      "epoch": 0.026631862851965366,
      "grad_norm": 1.1930912733078003,
      "learning_rate": 0.00019467381820622385,
      "loss": 4.192,
      "step": 28570
    },
    {
      "epoch": 0.026641184470044457,
      "grad_norm": 1.1010065078735352,
      "learning_rate": 0.00019467195388412866,
      "loss": 4.2661,
      "step": 28580
    },
    {
      "epoch": 0.026650506088123548,
      "grad_norm": 1.014986276626587,
      "learning_rate": 0.00019467008956203344,
      "loss": 4.2229,
      "step": 28590
    },
    {
      "epoch": 0.02665982770620264,
      "grad_norm": 1.2132433652877808,
      "learning_rate": 0.00019466822523993827,
      "loss": 4.2578,
      "step": 28600
    },
    {
      "epoch": 0.026669149324281733,
      "grad_norm": 1.2784295082092285,
      "learning_rate": 0.00019466636091784308,
      "loss": 4.36,
      "step": 28610
    },
    {
      "epoch": 0.026678470942360824,
      "grad_norm": 0.9309957027435303,
      "learning_rate": 0.00019466449659574786,
      "loss": 4.3587,
      "step": 28620
    },
    {
      "epoch": 0.026687792560439915,
      "grad_norm": 1.1927074193954468,
      "learning_rate": 0.00019466263227365267,
      "loss": 4.0845,
      "step": 28630
    },
    {
      "epoch": 0.026697114178519006,
      "grad_norm": 0.9887617230415344,
      "learning_rate": 0.00019466076795155744,
      "loss": 4.1896,
      "step": 28640
    },
    {
      "epoch": 0.026706435796598097,
      "grad_norm": 1.1572954654693604,
      "learning_rate": 0.00019465890362946225,
      "loss": 4.0206,
      "step": 28650
    },
    {
      "epoch": 0.02671575741467719,
      "grad_norm": 0.9519640803337097,
      "learning_rate": 0.00019465703930736708,
      "loss": 4.3151,
      "step": 28660
    },
    {
      "epoch": 0.026725079032756283,
      "grad_norm": 1.0924383401870728,
      "learning_rate": 0.00019465517498527186,
      "loss": 4.3414,
      "step": 28670
    },
    {
      "epoch": 0.026734400650835374,
      "grad_norm": 1.1783348321914673,
      "learning_rate": 0.00019465331066317667,
      "loss": 4.1015,
      "step": 28680
    },
    {
      "epoch": 0.026743722268914465,
      "grad_norm": 1.1751993894577026,
      "learning_rate": 0.00019465144634108145,
      "loss": 4.2085,
      "step": 28690
    },
    {
      "epoch": 0.02675304388699356,
      "grad_norm": 1.0536497831344604,
      "learning_rate": 0.00019464958201898626,
      "loss": 4.5019,
      "step": 28700
    },
    {
      "epoch": 0.02676236550507265,
      "grad_norm": 0.9439518451690674,
      "learning_rate": 0.00019464771769689106,
      "loss": 4.0754,
      "step": 28710
    },
    {
      "epoch": 0.02677168712315174,
      "grad_norm": 1.082790493965149,
      "learning_rate": 0.00019464585337479587,
      "loss": 4.4255,
      "step": 28720
    },
    {
      "epoch": 0.026781008741230832,
      "grad_norm": 1.3948160409927368,
      "learning_rate": 0.00019464398905270067,
      "loss": 4.2277,
      "step": 28730
    },
    {
      "epoch": 0.026790330359309927,
      "grad_norm": 0.9570707082748413,
      "learning_rate": 0.00019464212473060545,
      "loss": 4.3283,
      "step": 28740
    },
    {
      "epoch": 0.026799651977389018,
      "grad_norm": 1.1895440816879272,
      "learning_rate": 0.00019464026040851026,
      "loss": 4.1415,
      "step": 28750
    },
    {
      "epoch": 0.02680897359546811,
      "grad_norm": 1.1227682828903198,
      "learning_rate": 0.00019463839608641507,
      "loss": 4.1646,
      "step": 28760
    },
    {
      "epoch": 0.0268182952135472,
      "grad_norm": 1.0524277687072754,
      "learning_rate": 0.00019463653176431987,
      "loss": 4.1959,
      "step": 28770
    },
    {
      "epoch": 0.026827616831626294,
      "grad_norm": 1.1770564317703247,
      "learning_rate": 0.00019463466744222468,
      "loss": 4.3649,
      "step": 28780
    },
    {
      "epoch": 0.026836938449705385,
      "grad_norm": 0.9908391833305359,
      "learning_rate": 0.00019463280312012946,
      "loss": 4.1615,
      "step": 28790
    },
    {
      "epoch": 0.026846260067784476,
      "grad_norm": 1.150545358657837,
      "learning_rate": 0.00019463093879803426,
      "loss": 4.3466,
      "step": 28800
    },
    {
      "epoch": 0.026855581685863567,
      "grad_norm": 0.9483940601348877,
      "learning_rate": 0.00019462907447593907,
      "loss": 4.0479,
      "step": 28810
    },
    {
      "epoch": 0.026864903303942658,
      "grad_norm": 1.0444685220718384,
      "learning_rate": 0.00019462721015384388,
      "loss": 4.2077,
      "step": 28820
    },
    {
      "epoch": 0.026874224922021753,
      "grad_norm": 1.1011759042739868,
      "learning_rate": 0.00019462534583174868,
      "loss": 4.2643,
      "step": 28830
    },
    {
      "epoch": 0.026883546540100844,
      "grad_norm": 1.2812039852142334,
      "learning_rate": 0.00019462348150965346,
      "loss": 4.2058,
      "step": 28840
    },
    {
      "epoch": 0.026892868158179935,
      "grad_norm": 1.1920572519302368,
      "learning_rate": 0.00019462161718755827,
      "loss": 4.2223,
      "step": 28850
    },
    {
      "epoch": 0.026902189776259026,
      "grad_norm": 1.2258453369140625,
      "learning_rate": 0.00019461975286546307,
      "loss": 4.2618,
      "step": 28860
    },
    {
      "epoch": 0.02691151139433812,
      "grad_norm": 0.9299667477607727,
      "learning_rate": 0.00019461788854336785,
      "loss": 4.1606,
      "step": 28870
    },
    {
      "epoch": 0.02692083301241721,
      "grad_norm": 1.2394553422927856,
      "learning_rate": 0.0001946160242212727,
      "loss": 4.1716,
      "step": 28880
    },
    {
      "epoch": 0.026930154630496302,
      "grad_norm": 1.0781043767929077,
      "learning_rate": 0.00019461415989917747,
      "loss": 4.3625,
      "step": 28890
    },
    {
      "epoch": 0.026939476248575393,
      "grad_norm": 1.137622594833374,
      "learning_rate": 0.00019461229557708227,
      "loss": 4.2961,
      "step": 28900
    },
    {
      "epoch": 0.026948797866654488,
      "grad_norm": 1.3354579210281372,
      "learning_rate": 0.00019461043125498708,
      "loss": 4.0875,
      "step": 28910
    },
    {
      "epoch": 0.02695811948473358,
      "grad_norm": 1.0804301500320435,
      "learning_rate": 0.00019460856693289186,
      "loss": 4.3383,
      "step": 28920
    },
    {
      "epoch": 0.02696744110281267,
      "grad_norm": 1.0588536262512207,
      "learning_rate": 0.00019460670261079666,
      "loss": 4.2096,
      "step": 28930
    },
    {
      "epoch": 0.02697676272089176,
      "grad_norm": 1.0959217548370361,
      "learning_rate": 0.00019460483828870147,
      "loss": 4.1171,
      "step": 28940
    },
    {
      "epoch": 0.026986084338970855,
      "grad_norm": 1.1731961965560913,
      "learning_rate": 0.00019460297396660628,
      "loss": 4.161,
      "step": 28950
    },
    {
      "epoch": 0.026995405957049946,
      "grad_norm": 0.8345583081245422,
      "learning_rate": 0.00019460110964451108,
      "loss": 4.1373,
      "step": 28960
    },
    {
      "epoch": 0.027004727575129037,
      "grad_norm": 1.227846622467041,
      "learning_rate": 0.00019459924532241586,
      "loss": 4.0247,
      "step": 28970
    },
    {
      "epoch": 0.02701404919320813,
      "grad_norm": 1.09984290599823,
      "learning_rate": 0.00019459738100032067,
      "loss": 4.1594,
      "step": 28980
    },
    {
      "epoch": 0.02702337081128722,
      "grad_norm": 1.0612342357635498,
      "learning_rate": 0.00019459551667822548,
      "loss": 4.0493,
      "step": 28990
    },
    {
      "epoch": 0.027032692429366314,
      "grad_norm": 1.178430199623108,
      "learning_rate": 0.00019459365235613028,
      "loss": 4.1433,
      "step": 29000
    },
    {
      "epoch": 0.027042014047445405,
      "grad_norm": 1.035421371459961,
      "learning_rate": 0.0001945917880340351,
      "loss": 4.2343,
      "step": 29010
    },
    {
      "epoch": 0.027051335665524496,
      "grad_norm": 1.0976892709732056,
      "learning_rate": 0.00019458992371193987,
      "loss": 4.2715,
      "step": 29020
    },
    {
      "epoch": 0.027060657283603587,
      "grad_norm": 1.085189700126648,
      "learning_rate": 0.00019458805938984467,
      "loss": 4.2201,
      "step": 29030
    },
    {
      "epoch": 0.02706997890168268,
      "grad_norm": 1.0454773902893066,
      "learning_rate": 0.00019458619506774945,
      "loss": 4.3262,
      "step": 29040
    },
    {
      "epoch": 0.027079300519761772,
      "grad_norm": 1.033460259437561,
      "learning_rate": 0.00019458433074565429,
      "loss": 4.2414,
      "step": 29050
    },
    {
      "epoch": 0.027088622137840863,
      "grad_norm": 1.2299708127975464,
      "learning_rate": 0.0001945824664235591,
      "loss": 4.1531,
      "step": 29060
    },
    {
      "epoch": 0.027097943755919954,
      "grad_norm": 1.0889471769332886,
      "learning_rate": 0.00019458060210146387,
      "loss": 4.4035,
      "step": 29070
    },
    {
      "epoch": 0.02710726537399905,
      "grad_norm": 1.0031168460845947,
      "learning_rate": 0.00019457873777936868,
      "loss": 4.323,
      "step": 29080
    },
    {
      "epoch": 0.02711658699207814,
      "grad_norm": 0.9850956797599792,
      "learning_rate": 0.00019457687345727346,
      "loss": 4.2094,
      "step": 29090
    },
    {
      "epoch": 0.02712590861015723,
      "grad_norm": 1.158898949623108,
      "learning_rate": 0.00019457500913517826,
      "loss": 4.4195,
      "step": 29100
    },
    {
      "epoch": 0.027135230228236322,
      "grad_norm": 1.1418386697769165,
      "learning_rate": 0.0001945731448130831,
      "loss": 4.3909,
      "step": 29110
    },
    {
      "epoch": 0.027144551846315416,
      "grad_norm": 1.048790693283081,
      "learning_rate": 0.00019457128049098788,
      "loss": 4.4208,
      "step": 29120
    },
    {
      "epoch": 0.027153873464394507,
      "grad_norm": 0.9750396609306335,
      "learning_rate": 0.00019456941616889268,
      "loss": 4.3786,
      "step": 29130
    },
    {
      "epoch": 0.0271631950824736,
      "grad_norm": 1.025668740272522,
      "learning_rate": 0.00019456755184679746,
      "loss": 4.451,
      "step": 29140
    },
    {
      "epoch": 0.02717251670055269,
      "grad_norm": 1.0821212530136108,
      "learning_rate": 0.00019456568752470227,
      "loss": 4.1596,
      "step": 29150
    },
    {
      "epoch": 0.02718183831863178,
      "grad_norm": 1.1281262636184692,
      "learning_rate": 0.00019456382320260707,
      "loss": 4.3554,
      "step": 29160
    },
    {
      "epoch": 0.027191159936710875,
      "grad_norm": 1.0510941743850708,
      "learning_rate": 0.00019456195888051188,
      "loss": 4.1642,
      "step": 29170
    },
    {
      "epoch": 0.027200481554789966,
      "grad_norm": 0.9559809565544128,
      "learning_rate": 0.0001945600945584167,
      "loss": 4.449,
      "step": 29180
    },
    {
      "epoch": 0.027209803172869057,
      "grad_norm": 0.9987666010856628,
      "learning_rate": 0.00019455823023632147,
      "loss": 4.1955,
      "step": 29190
    },
    {
      "epoch": 0.027219124790948148,
      "grad_norm": 1.1287548542022705,
      "learning_rate": 0.00019455636591422627,
      "loss": 4.3434,
      "step": 29200
    },
    {
      "epoch": 0.027228446409027243,
      "grad_norm": 1.0920779705047607,
      "learning_rate": 0.00019455450159213108,
      "loss": 4.0969,
      "step": 29210
    },
    {
      "epoch": 0.027237768027106334,
      "grad_norm": 1.1443978548049927,
      "learning_rate": 0.00019455263727003588,
      "loss": 4.2832,
      "step": 29220
    },
    {
      "epoch": 0.027247089645185425,
      "grad_norm": 1.143643856048584,
      "learning_rate": 0.0001945507729479407,
      "loss": 4.1581,
      "step": 29230
    },
    {
      "epoch": 0.027256411263264516,
      "grad_norm": 1.0279771089553833,
      "learning_rate": 0.00019454890862584547,
      "loss": 4.0264,
      "step": 29240
    },
    {
      "epoch": 0.02726573288134361,
      "grad_norm": 1.184669017791748,
      "learning_rate": 0.00019454704430375028,
      "loss": 4.2995,
      "step": 29250
    },
    {
      "epoch": 0.0272750544994227,
      "grad_norm": 1.0545037984848022,
      "learning_rate": 0.00019454517998165508,
      "loss": 4.1016,
      "step": 29260
    },
    {
      "epoch": 0.027284376117501792,
      "grad_norm": 1.180772066116333,
      "learning_rate": 0.0001945433156595599,
      "loss": 4.1973,
      "step": 29270
    },
    {
      "epoch": 0.027293697735580883,
      "grad_norm": 1.3218094110488892,
      "learning_rate": 0.0001945414513374647,
      "loss": 4.3312,
      "step": 29280
    },
    {
      "epoch": 0.027303019353659978,
      "grad_norm": 1.1894714832305908,
      "learning_rate": 0.00019453958701536947,
      "loss": 4.0967,
      "step": 29290
    },
    {
      "epoch": 0.02731234097173907,
      "grad_norm": 1.0680218935012817,
      "learning_rate": 0.00019453772269327428,
      "loss": 4.2289,
      "step": 29300
    },
    {
      "epoch": 0.02732166258981816,
      "grad_norm": 1.1131433248519897,
      "learning_rate": 0.0001945358583711791,
      "loss": 4.2336,
      "step": 29310
    },
    {
      "epoch": 0.02733098420789725,
      "grad_norm": 1.078934907913208,
      "learning_rate": 0.00019453399404908387,
      "loss": 4.3221,
      "step": 29320
    },
    {
      "epoch": 0.02734030582597634,
      "grad_norm": 1.1750596761703491,
      "learning_rate": 0.0001945321297269887,
      "loss": 4.2611,
      "step": 29330
    },
    {
      "epoch": 0.027349627444055436,
      "grad_norm": 1.0845474004745483,
      "learning_rate": 0.00019453026540489348,
      "loss": 4.2088,
      "step": 29340
    },
    {
      "epoch": 0.027358949062134527,
      "grad_norm": 1.278959035873413,
      "learning_rate": 0.00019452840108279829,
      "loss": 4.1543,
      "step": 29350
    },
    {
      "epoch": 0.027368270680213618,
      "grad_norm": 1.59437096118927,
      "learning_rate": 0.0001945265367607031,
      "loss": 4.2452,
      "step": 29360
    },
    {
      "epoch": 0.02737759229829271,
      "grad_norm": 1.139586091041565,
      "learning_rate": 0.00019452467243860787,
      "loss": 4.2048,
      "step": 29370
    },
    {
      "epoch": 0.027386913916371804,
      "grad_norm": 1.0568971633911133,
      "learning_rate": 0.00019452280811651268,
      "loss": 4.1954,
      "step": 29380
    },
    {
      "epoch": 0.027396235534450895,
      "grad_norm": 1.2533947229385376,
      "learning_rate": 0.00019452094379441748,
      "loss": 4.3684,
      "step": 29390
    },
    {
      "epoch": 0.027405557152529986,
      "grad_norm": 1.0121870040893555,
      "learning_rate": 0.0001945190794723223,
      "loss": 4.2766,
      "step": 29400
    },
    {
      "epoch": 0.027414878770609077,
      "grad_norm": 1.1572916507720947,
      "learning_rate": 0.0001945172151502271,
      "loss": 3.9731,
      "step": 29410
    },
    {
      "epoch": 0.02742420038868817,
      "grad_norm": 1.0100351572036743,
      "learning_rate": 0.00019451535082813188,
      "loss": 4.2405,
      "step": 29420
    },
    {
      "epoch": 0.027433522006767262,
      "grad_norm": 1.0195560455322266,
      "learning_rate": 0.00019451348650603668,
      "loss": 4.1415,
      "step": 29430
    },
    {
      "epoch": 0.027442843624846353,
      "grad_norm": 1.1734956502914429,
      "learning_rate": 0.0001945116221839415,
      "loss": 4.124,
      "step": 29440
    },
    {
      "epoch": 0.027452165242925444,
      "grad_norm": 0.9657151103019714,
      "learning_rate": 0.0001945097578618463,
      "loss": 4.2005,
      "step": 29450
    },
    {
      "epoch": 0.02746148686100454,
      "grad_norm": 1.2245347499847412,
      "learning_rate": 0.0001945078935397511,
      "loss": 4.2618,
      "step": 29460
    },
    {
      "epoch": 0.02747080847908363,
      "grad_norm": 1.4259823560714722,
      "learning_rate": 0.00019450602921765588,
      "loss": 4.1627,
      "step": 29470
    },
    {
      "epoch": 0.02748013009716272,
      "grad_norm": 1.1593616008758545,
      "learning_rate": 0.00019450416489556069,
      "loss": 4.2613,
      "step": 29480
    },
    {
      "epoch": 0.027489451715241812,
      "grad_norm": 1.2188735008239746,
      "learning_rate": 0.00019450230057346547,
      "loss": 4.1394,
      "step": 29490
    },
    {
      "epoch": 0.027498773333320903,
      "grad_norm": 0.9380781650543213,
      "learning_rate": 0.0001945004362513703,
      "loss": 4.1846,
      "step": 29500
    },
    {
      "epoch": 0.027508094951399997,
      "grad_norm": 1.283844232559204,
      "learning_rate": 0.0001944985719292751,
      "loss": 4.3088,
      "step": 29510
    },
    {
      "epoch": 0.02751741656947909,
      "grad_norm": 1.1529028415679932,
      "learning_rate": 0.00019449670760717988,
      "loss": 4.1329,
      "step": 29520
    },
    {
      "epoch": 0.02752673818755818,
      "grad_norm": 1.1037242412567139,
      "learning_rate": 0.0001944948432850847,
      "loss": 4.1341,
      "step": 29530
    },
    {
      "epoch": 0.02753605980563727,
      "grad_norm": 0.9974387884140015,
      "learning_rate": 0.00019449297896298947,
      "loss": 4.0684,
      "step": 29540
    },
    {
      "epoch": 0.027545381423716365,
      "grad_norm": 1.0964442491531372,
      "learning_rate": 0.00019449111464089428,
      "loss": 4.1372,
      "step": 29550
    },
    {
      "epoch": 0.027554703041795456,
      "grad_norm": 0.9842365980148315,
      "learning_rate": 0.00019448925031879908,
      "loss": 4.1283,
      "step": 29560
    },
    {
      "epoch": 0.027564024659874547,
      "grad_norm": 0.9990352988243103,
      "learning_rate": 0.0001944873859967039,
      "loss": 4.3964,
      "step": 29570
    },
    {
      "epoch": 0.027573346277953638,
      "grad_norm": 1.0937678813934326,
      "learning_rate": 0.0001944855216746087,
      "loss": 4.4223,
      "step": 29580
    },
    {
      "epoch": 0.027582667896032732,
      "grad_norm": 1.1645828485488892,
      "learning_rate": 0.00019448365735251347,
      "loss": 4.0024,
      "step": 29590
    },
    {
      "epoch": 0.027591989514111823,
      "grad_norm": 1.1039122343063354,
      "learning_rate": 0.00019448179303041828,
      "loss": 4.2516,
      "step": 29600
    },
    {
      "epoch": 0.027601311132190914,
      "grad_norm": 1.2983834743499756,
      "learning_rate": 0.0001944799287083231,
      "loss": 3.9972,
      "step": 29610
    },
    {
      "epoch": 0.027610632750270005,
      "grad_norm": 1.1381375789642334,
      "learning_rate": 0.0001944780643862279,
      "loss": 4.1945,
      "step": 29620
    },
    {
      "epoch": 0.0276199543683491,
      "grad_norm": 1.1137017011642456,
      "learning_rate": 0.0001944762000641327,
      "loss": 4.1948,
      "step": 29630
    },
    {
      "epoch": 0.02762927598642819,
      "grad_norm": 0.9230836033821106,
      "learning_rate": 0.00019447433574203748,
      "loss": 4.3527,
      "step": 29640
    },
    {
      "epoch": 0.027638597604507282,
      "grad_norm": 1.05021071434021,
      "learning_rate": 0.00019447247141994229,
      "loss": 4.1774,
      "step": 29650
    },
    {
      "epoch": 0.027647919222586373,
      "grad_norm": 0.9980315566062927,
      "learning_rate": 0.0001944706070978471,
      "loss": 4.1463,
      "step": 29660
    },
    {
      "epoch": 0.027657240840665464,
      "grad_norm": 1.165527582168579,
      "learning_rate": 0.0001944687427757519,
      "loss": 4.3281,
      "step": 29670
    },
    {
      "epoch": 0.02766656245874456,
      "grad_norm": 0.9254803657531738,
      "learning_rate": 0.0001944668784536567,
      "loss": 4.464,
      "step": 29680
    },
    {
      "epoch": 0.02767588407682365,
      "grad_norm": 1.0743868350982666,
      "learning_rate": 0.00019446501413156148,
      "loss": 4.3417,
      "step": 29690
    },
    {
      "epoch": 0.02768520569490274,
      "grad_norm": 1.2196156978607178,
      "learning_rate": 0.0001944631498094663,
      "loss": 4.2004,
      "step": 29700
    },
    {
      "epoch": 0.02769452731298183,
      "grad_norm": 1.332757592201233,
      "learning_rate": 0.00019446128548737107,
      "loss": 4.3299,
      "step": 29710
    },
    {
      "epoch": 0.027703848931060926,
      "grad_norm": 1.177437663078308,
      "learning_rate": 0.0001944594211652759,
      "loss": 4.2965,
      "step": 29720
    },
    {
      "epoch": 0.027713170549140017,
      "grad_norm": 1.066603183746338,
      "learning_rate": 0.0001944575568431807,
      "loss": 4.3221,
      "step": 29730
    },
    {
      "epoch": 0.027722492167219108,
      "grad_norm": 1.1323814392089844,
      "learning_rate": 0.0001944556925210855,
      "loss": 4.1765,
      "step": 29740
    },
    {
      "epoch": 0.0277318137852982,
      "grad_norm": 1.292104721069336,
      "learning_rate": 0.0001944538281989903,
      "loss": 4.0772,
      "step": 29750
    },
    {
      "epoch": 0.027741135403377293,
      "grad_norm": 1.0282576084136963,
      "learning_rate": 0.00019445196387689507,
      "loss": 4.2939,
      "step": 29760
    },
    {
      "epoch": 0.027750457021456384,
      "grad_norm": 0.9446836113929749,
      "learning_rate": 0.00019445009955479988,
      "loss": 4.1135,
      "step": 29770
    },
    {
      "epoch": 0.027759778639535476,
      "grad_norm": 1.1790218353271484,
      "learning_rate": 0.0001944482352327047,
      "loss": 4.2941,
      "step": 29780
    },
    {
      "epoch": 0.027769100257614567,
      "grad_norm": 0.9898374080657959,
      "learning_rate": 0.0001944463709106095,
      "loss": 4.2983,
      "step": 29790
    },
    {
      "epoch": 0.02777842187569366,
      "grad_norm": 1.1479710340499878,
      "learning_rate": 0.0001944445065885143,
      "loss": 4.2331,
      "step": 29800
    },
    {
      "epoch": 0.027787743493772752,
      "grad_norm": 1.15385901927948,
      "learning_rate": 0.00019444264226641908,
      "loss": 4.1323,
      "step": 29810
    },
    {
      "epoch": 0.027797065111851843,
      "grad_norm": 1.2066665887832642,
      "learning_rate": 0.00019444077794432388,
      "loss": 4.3101,
      "step": 29820
    },
    {
      "epoch": 0.027806386729930934,
      "grad_norm": 1.2487064599990845,
      "learning_rate": 0.0001944389136222287,
      "loss": 4.2495,
      "step": 29830
    },
    {
      "epoch": 0.027815708348010025,
      "grad_norm": 1.2349793910980225,
      "learning_rate": 0.0001944370493001335,
      "loss": 4.298,
      "step": 29840
    },
    {
      "epoch": 0.02782502996608912,
      "grad_norm": 1.0675073862075806,
      "learning_rate": 0.0001944351849780383,
      "loss": 4.2703,
      "step": 29850
    },
    {
      "epoch": 0.02783435158416821,
      "grad_norm": 1.095158338546753,
      "learning_rate": 0.00019443332065594308,
      "loss": 4.2048,
      "step": 29860
    },
    {
      "epoch": 0.0278436732022473,
      "grad_norm": 1.3492717742919922,
      "learning_rate": 0.0001944314563338479,
      "loss": 4.1668,
      "step": 29870
    },
    {
      "epoch": 0.027852994820326393,
      "grad_norm": 0.9899656772613525,
      "learning_rate": 0.0001944295920117527,
      "loss": 4.3974,
      "step": 29880
    },
    {
      "epoch": 0.027862316438405487,
      "grad_norm": 0.8951921463012695,
      "learning_rate": 0.0001944277276896575,
      "loss": 4.4626,
      "step": 29890
    },
    {
      "epoch": 0.027871638056484578,
      "grad_norm": 1.100419282913208,
      "learning_rate": 0.0001944258633675623,
      "loss": 4.1757,
      "step": 29900
    },
    {
      "epoch": 0.02788095967456367,
      "grad_norm": 1.104102373123169,
      "learning_rate": 0.0001944239990454671,
      "loss": 4.1705,
      "step": 29910
    },
    {
      "epoch": 0.02789028129264276,
      "grad_norm": 1.2880393266677856,
      "learning_rate": 0.0001944221347233719,
      "loss": 4.2618,
      "step": 29920
    },
    {
      "epoch": 0.027899602910721855,
      "grad_norm": 1.0313355922698975,
      "learning_rate": 0.0001944202704012767,
      "loss": 4.067,
      "step": 29930
    },
    {
      "epoch": 0.027908924528800946,
      "grad_norm": 1.163288950920105,
      "learning_rate": 0.0001944184060791815,
      "loss": 4.0742,
      "step": 29940
    },
    {
      "epoch": 0.027918246146880037,
      "grad_norm": 1.4432181119918823,
      "learning_rate": 0.0001944165417570863,
      "loss": 4.0086,
      "step": 29950
    },
    {
      "epoch": 0.027927567764959128,
      "grad_norm": 1.1806241273880005,
      "learning_rate": 0.0001944146774349911,
      "loss": 4.3301,
      "step": 29960
    },
    {
      "epoch": 0.027936889383038222,
      "grad_norm": 0.8892717957496643,
      "learning_rate": 0.0001944128131128959,
      "loss": 4.1498,
      "step": 29970
    },
    {
      "epoch": 0.027946211001117313,
      "grad_norm": 1.1280444860458374,
      "learning_rate": 0.0001944109487908007,
      "loss": 4.4516,
      "step": 29980
    },
    {
      "epoch": 0.027955532619196404,
      "grad_norm": 1.0531505346298218,
      "learning_rate": 0.00019440908446870548,
      "loss": 4.1705,
      "step": 29990
    },
    {
      "epoch": 0.027964854237275495,
      "grad_norm": 1.0602271556854248,
      "learning_rate": 0.0001944072201466103,
      "loss": 4.242,
      "step": 30000
    },
    {
      "epoch": 0.02797417585535459,
      "grad_norm": 1.0653207302093506,
      "learning_rate": 0.0001944053558245151,
      "loss": 4.2937,
      "step": 30010
    },
    {
      "epoch": 0.02798349747343368,
      "grad_norm": 1.0337291955947876,
      "learning_rate": 0.0001944034915024199,
      "loss": 4.162,
      "step": 30020
    },
    {
      "epoch": 0.02799281909151277,
      "grad_norm": 1.0819483995437622,
      "learning_rate": 0.0001944016271803247,
      "loss": 4.1558,
      "step": 30030
    },
    {
      "epoch": 0.028002140709591863,
      "grad_norm": 1.0429236888885498,
      "learning_rate": 0.0001943997628582295,
      "loss": 4.1386,
      "step": 30040
    },
    {
      "epoch": 0.028011462327670954,
      "grad_norm": 1.15821373462677,
      "learning_rate": 0.0001943978985361343,
      "loss": 3.9929,
      "step": 30050
    },
    {
      "epoch": 0.028020783945750048,
      "grad_norm": 0.9606207013130188,
      "learning_rate": 0.0001943960342140391,
      "loss": 4.3238,
      "step": 30060
    },
    {
      "epoch": 0.02803010556382914,
      "grad_norm": 1.0747424364089966,
      "learning_rate": 0.0001943941698919439,
      "loss": 4.2483,
      "step": 30070
    },
    {
      "epoch": 0.02803942718190823,
      "grad_norm": 1.1192786693572998,
      "learning_rate": 0.0001943923055698487,
      "loss": 4.0286,
      "step": 30080
    },
    {
      "epoch": 0.02804874879998732,
      "grad_norm": 1.0124425888061523,
      "learning_rate": 0.0001943904412477535,
      "loss": 4.2082,
      "step": 30090
    },
    {
      "epoch": 0.028058070418066416,
      "grad_norm": 1.1556917428970337,
      "learning_rate": 0.0001943885769256583,
      "loss": 4.2858,
      "step": 30100
    },
    {
      "epoch": 0.028067392036145507,
      "grad_norm": 1.0085023641586304,
      "learning_rate": 0.0001943867126035631,
      "loss": 4.0395,
      "step": 30110
    },
    {
      "epoch": 0.028076713654224598,
      "grad_norm": 1.2024095058441162,
      "learning_rate": 0.0001943848482814679,
      "loss": 4.1215,
      "step": 30120
    },
    {
      "epoch": 0.02808603527230369,
      "grad_norm": 1.1313384771347046,
      "learning_rate": 0.00019438298395937272,
      "loss": 4.2428,
      "step": 30130
    },
    {
      "epoch": 0.028095356890382783,
      "grad_norm": 1.0524983406066895,
      "learning_rate": 0.0001943811196372775,
      "loss": 4.3376,
      "step": 30140
    },
    {
      "epoch": 0.028104678508461874,
      "grad_norm": 0.9869844913482666,
      "learning_rate": 0.0001943792553151823,
      "loss": 4.096,
      "step": 30150
    },
    {
      "epoch": 0.028114000126540965,
      "grad_norm": 1.1388225555419922,
      "learning_rate": 0.00019437739099308708,
      "loss": 3.9101,
      "step": 30160
    },
    {
      "epoch": 0.028123321744620056,
      "grad_norm": 1.0303579568862915,
      "learning_rate": 0.00019437552667099192,
      "loss": 4.2153,
      "step": 30170
    },
    {
      "epoch": 0.02813264336269915,
      "grad_norm": 1.2085143327713013,
      "learning_rate": 0.00019437366234889672,
      "loss": 4.269,
      "step": 30180
    },
    {
      "epoch": 0.028141964980778242,
      "grad_norm": 1.3076800107955933,
      "learning_rate": 0.0001943717980268015,
      "loss": 3.9404,
      "step": 30190
    },
    {
      "epoch": 0.028151286598857333,
      "grad_norm": 1.0717430114746094,
      "learning_rate": 0.0001943699337047063,
      "loss": 4.2475,
      "step": 30200
    },
    {
      "epoch": 0.028160608216936424,
      "grad_norm": 1.24625563621521,
      "learning_rate": 0.00019436806938261109,
      "loss": 4.2041,
      "step": 30210
    },
    {
      "epoch": 0.028169929835015515,
      "grad_norm": 0.9927229285240173,
      "learning_rate": 0.0001943662050605159,
      "loss": 4.153,
      "step": 30220
    },
    {
      "epoch": 0.02817925145309461,
      "grad_norm": 1.0605548620224,
      "learning_rate": 0.00019436434073842073,
      "loss": 4.3944,
      "step": 30230
    },
    {
      "epoch": 0.0281885730711737,
      "grad_norm": 0.9414675235748291,
      "learning_rate": 0.0001943624764163255,
      "loss": 4.2689,
      "step": 30240
    },
    {
      "epoch": 0.02819789468925279,
      "grad_norm": 1.2204649448394775,
      "learning_rate": 0.0001943606120942303,
      "loss": 4.2684,
      "step": 30250
    },
    {
      "epoch": 0.028207216307331882,
      "grad_norm": 1.161726474761963,
      "learning_rate": 0.0001943587477721351,
      "loss": 4.2133,
      "step": 30260
    },
    {
      "epoch": 0.028216537925410977,
      "grad_norm": 0.9708456993103027,
      "learning_rate": 0.0001943568834500399,
      "loss": 4.0784,
      "step": 30270
    },
    {
      "epoch": 0.028225859543490068,
      "grad_norm": 1.2743651866912842,
      "learning_rate": 0.0001943550191279447,
      "loss": 4.1713,
      "step": 30280
    },
    {
      "epoch": 0.02823518116156916,
      "grad_norm": 1.1080349683761597,
      "learning_rate": 0.0001943531548058495,
      "loss": 4.2465,
      "step": 30290
    },
    {
      "epoch": 0.02824450277964825,
      "grad_norm": 1.0082476139068604,
      "learning_rate": 0.00019435129048375432,
      "loss": 4.2253,
      "step": 30300
    },
    {
      "epoch": 0.028253824397727344,
      "grad_norm": 0.959999144077301,
      "learning_rate": 0.0001943494261616591,
      "loss": 4.1607,
      "step": 30310
    },
    {
      "epoch": 0.028263146015806435,
      "grad_norm": 1.0145454406738281,
      "learning_rate": 0.0001943475618395639,
      "loss": 4.3479,
      "step": 30320
    },
    {
      "epoch": 0.028272467633885526,
      "grad_norm": 1.0753196477890015,
      "learning_rate": 0.0001943456975174687,
      "loss": 4.1636,
      "step": 30330
    },
    {
      "epoch": 0.028281789251964617,
      "grad_norm": 1.3137712478637695,
      "learning_rate": 0.00019434383319537351,
      "loss": 4.2434,
      "step": 30340
    },
    {
      "epoch": 0.028291110870043712,
      "grad_norm": 1.3082220554351807,
      "learning_rate": 0.00019434196887327832,
      "loss": 4.2677,
      "step": 30350
    },
    {
      "epoch": 0.028300432488122803,
      "grad_norm": 1.2131245136260986,
      "learning_rate": 0.0001943401045511831,
      "loss": 4.2032,
      "step": 30360
    },
    {
      "epoch": 0.028309754106201894,
      "grad_norm": 1.0667333602905273,
      "learning_rate": 0.0001943382402290879,
      "loss": 4.1387,
      "step": 30370
    },
    {
      "epoch": 0.028319075724280985,
      "grad_norm": 1.1289845705032349,
      "learning_rate": 0.0001943363759069927,
      "loss": 4.1713,
      "step": 30380
    },
    {
      "epoch": 0.028328397342360076,
      "grad_norm": 1.0636303424835205,
      "learning_rate": 0.00019433451158489752,
      "loss": 4.1983,
      "step": 30390
    },
    {
      "epoch": 0.02833771896043917,
      "grad_norm": 1.0969558954238892,
      "learning_rate": 0.00019433264726280232,
      "loss": 4.1487,
      "step": 30400
    },
    {
      "epoch": 0.02834704057851826,
      "grad_norm": 1.1674693822860718,
      "learning_rate": 0.0001943307829407071,
      "loss": 4.2383,
      "step": 30410
    },
    {
      "epoch": 0.028356362196597353,
      "grad_norm": 1.1638989448547363,
      "learning_rate": 0.0001943289186186119,
      "loss": 4.1205,
      "step": 30420
    },
    {
      "epoch": 0.028365683814676444,
      "grad_norm": 1.0495280027389526,
      "learning_rate": 0.00019432705429651672,
      "loss": 4.2748,
      "step": 30430
    },
    {
      "epoch": 0.028375005432755538,
      "grad_norm": 1.1624623537063599,
      "learning_rate": 0.0001943251899744215,
      "loss": 4.1716,
      "step": 30440
    },
    {
      "epoch": 0.02838432705083463,
      "grad_norm": 1.2940349578857422,
      "learning_rate": 0.00019432332565232633,
      "loss": 3.9999,
      "step": 30450
    },
    {
      "epoch": 0.02839364866891372,
      "grad_norm": 1.0359212160110474,
      "learning_rate": 0.0001943214613302311,
      "loss": 4.3129,
      "step": 30460
    },
    {
      "epoch": 0.02840297028699281,
      "grad_norm": 0.9796423316001892,
      "learning_rate": 0.00019431959700813591,
      "loss": 4.07,
      "step": 30470
    },
    {
      "epoch": 0.028412291905071906,
      "grad_norm": 1.1834994554519653,
      "learning_rate": 0.00019431773268604072,
      "loss": 4.1983,
      "step": 30480
    },
    {
      "epoch": 0.028421613523150997,
      "grad_norm": 1.0716720819473267,
      "learning_rate": 0.0001943158683639455,
      "loss": 4.2448,
      "step": 30490
    },
    {
      "epoch": 0.028430935141230088,
      "grad_norm": 1.108633041381836,
      "learning_rate": 0.0001943140040418503,
      "loss": 4.1101,
      "step": 30500
    },
    {
      "epoch": 0.02844025675930918,
      "grad_norm": 1.0244076251983643,
      "learning_rate": 0.0001943121397197551,
      "loss": 4.2849,
      "step": 30510
    },
    {
      "epoch": 0.028449578377388273,
      "grad_norm": 1.1317344903945923,
      "learning_rate": 0.00019431027539765992,
      "loss": 4.1582,
      "step": 30520
    },
    {
      "epoch": 0.028458899995467364,
      "grad_norm": 1.202074408531189,
      "learning_rate": 0.00019430841107556473,
      "loss": 4.3172,
      "step": 30530
    },
    {
      "epoch": 0.028468221613546455,
      "grad_norm": 1.2606077194213867,
      "learning_rate": 0.0001943065467534695,
      "loss": 4.3392,
      "step": 30540
    },
    {
      "epoch": 0.028477543231625546,
      "grad_norm": 1.0735373497009277,
      "learning_rate": 0.0001943046824313743,
      "loss": 4.3225,
      "step": 30550
    },
    {
      "epoch": 0.028486864849704637,
      "grad_norm": 1.0153846740722656,
      "learning_rate": 0.00019430281810927912,
      "loss": 4.1706,
      "step": 30560
    },
    {
      "epoch": 0.02849618646778373,
      "grad_norm": 1.4597234725952148,
      "learning_rate": 0.00019430095378718392,
      "loss": 4.2701,
      "step": 30570
    },
    {
      "epoch": 0.028505508085862823,
      "grad_norm": 1.0922951698303223,
      "learning_rate": 0.00019429908946508873,
      "loss": 4.2464,
      "step": 30580
    },
    {
      "epoch": 0.028514829703941914,
      "grad_norm": 1.1419568061828613,
      "learning_rate": 0.0001942972251429935,
      "loss": 4.155,
      "step": 30590
    },
    {
      "epoch": 0.028524151322021005,
      "grad_norm": 1.0993181467056274,
      "learning_rate": 0.00019429536082089832,
      "loss": 4.4409,
      "step": 30600
    },
    {
      "epoch": 0.0285334729401001,
      "grad_norm": 1.04743230342865,
      "learning_rate": 0.0001942934964988031,
      "loss": 4.0645,
      "step": 30610
    },
    {
      "epoch": 0.02854279455817919,
      "grad_norm": 1.085794448852539,
      "learning_rate": 0.00019429163217670793,
      "loss": 4.3749,
      "step": 30620
    },
    {
      "epoch": 0.02855211617625828,
      "grad_norm": 1.1122981309890747,
      "learning_rate": 0.00019428976785461273,
      "loss": 4.305,
      "step": 30630
    },
    {
      "epoch": 0.028561437794337372,
      "grad_norm": 1.0709840059280396,
      "learning_rate": 0.0001942879035325175,
      "loss": 3.9346,
      "step": 30640
    },
    {
      "epoch": 0.028570759412416467,
      "grad_norm": 1.1184806823730469,
      "learning_rate": 0.00019428603921042232,
      "loss": 4.0339,
      "step": 30650
    },
    {
      "epoch": 0.028580081030495558,
      "grad_norm": 1.12869393825531,
      "learning_rate": 0.0001942841748883271,
      "loss": 4.3543,
      "step": 30660
    },
    {
      "epoch": 0.02858940264857465,
      "grad_norm": 1.1106760501861572,
      "learning_rate": 0.0001942823105662319,
      "loss": 4.2357,
      "step": 30670
    },
    {
      "epoch": 0.02859872426665374,
      "grad_norm": 1.1765520572662354,
      "learning_rate": 0.00019428044624413674,
      "loss": 4.0451,
      "step": 30680
    },
    {
      "epoch": 0.028608045884732834,
      "grad_norm": 1.0167961120605469,
      "learning_rate": 0.00019427858192204152,
      "loss": 4.171,
      "step": 30690
    },
    {
      "epoch": 0.028617367502811925,
      "grad_norm": 1.2052909135818481,
      "learning_rate": 0.00019427671759994632,
      "loss": 4.0517,
      "step": 30700
    },
    {
      "epoch": 0.028626689120891016,
      "grad_norm": 1.1739888191223145,
      "learning_rate": 0.0001942748532778511,
      "loss": 4.099,
      "step": 30710
    },
    {
      "epoch": 0.028636010738970107,
      "grad_norm": 1.1208767890930176,
      "learning_rate": 0.0001942729889557559,
      "loss": 4.2057,
      "step": 30720
    },
    {
      "epoch": 0.0286453323570492,
      "grad_norm": 1.3346012830734253,
      "learning_rate": 0.00019427112463366072,
      "loss": 4.3209,
      "step": 30730
    },
    {
      "epoch": 0.028654653975128293,
      "grad_norm": 1.248573660850525,
      "learning_rate": 0.00019426926031156552,
      "loss": 4.1957,
      "step": 30740
    },
    {
      "epoch": 0.028663975593207384,
      "grad_norm": 1.1076196432113647,
      "learning_rate": 0.00019426739598947033,
      "loss": 4.3526,
      "step": 30750
    },
    {
      "epoch": 0.028673297211286475,
      "grad_norm": 1.167251467704773,
      "learning_rate": 0.0001942655316673751,
      "loss": 4.1091,
      "step": 30760
    },
    {
      "epoch": 0.028682618829365566,
      "grad_norm": 1.2762742042541504,
      "learning_rate": 0.00019426366734527991,
      "loss": 4.1109,
      "step": 30770
    },
    {
      "epoch": 0.02869194044744466,
      "grad_norm": 1.2636421918869019,
      "learning_rate": 0.00019426180302318472,
      "loss": 4.31,
      "step": 30780
    },
    {
      "epoch": 0.02870126206552375,
      "grad_norm": 1.0301423072814941,
      "learning_rate": 0.00019425993870108953,
      "loss": 4.1196,
      "step": 30790
    },
    {
      "epoch": 0.028710583683602842,
      "grad_norm": 1.0851550102233887,
      "learning_rate": 0.00019425807437899433,
      "loss": 4.3684,
      "step": 30800
    },
    {
      "epoch": 0.028719905301681933,
      "grad_norm": 1.2577015161514282,
      "learning_rate": 0.0001942562100568991,
      "loss": 4.2259,
      "step": 30810
    },
    {
      "epoch": 0.028729226919761028,
      "grad_norm": 1.2661305665969849,
      "learning_rate": 0.00019425434573480392,
      "loss": 4.3785,
      "step": 30820
    },
    {
      "epoch": 0.02873854853784012,
      "grad_norm": 1.0486119985580444,
      "learning_rate": 0.00019425248141270872,
      "loss": 4.3778,
      "step": 30830
    },
    {
      "epoch": 0.02874787015591921,
      "grad_norm": 1.1134130954742432,
      "learning_rate": 0.00019425061709061353,
      "loss": 4.2029,
      "step": 30840
    },
    {
      "epoch": 0.0287571917739983,
      "grad_norm": 1.0229254961013794,
      "learning_rate": 0.00019424875276851834,
      "loss": 4.1935,
      "step": 30850
    },
    {
      "epoch": 0.028766513392077395,
      "grad_norm": 1.1604161262512207,
      "learning_rate": 0.00019424688844642312,
      "loss": 4.2152,
      "step": 30860
    },
    {
      "epoch": 0.028775835010156486,
      "grad_norm": 1.1942169666290283,
      "learning_rate": 0.00019424502412432792,
      "loss": 4.2018,
      "step": 30870
    },
    {
      "epoch": 0.028785156628235577,
      "grad_norm": 1.0498714447021484,
      "learning_rate": 0.0001942431598022327,
      "loss": 4.1511,
      "step": 30880
    },
    {
      "epoch": 0.02879447824631467,
      "grad_norm": 1.004705548286438,
      "learning_rate": 0.0001942412954801375,
      "loss": 4.3118,
      "step": 30890
    },
    {
      "epoch": 0.02880379986439376,
      "grad_norm": 0.9384045600891113,
      "learning_rate": 0.00019423943115804234,
      "loss": 4.3223,
      "step": 30900
    },
    {
      "epoch": 0.028813121482472854,
      "grad_norm": 1.2505364418029785,
      "learning_rate": 0.00019423756683594712,
      "loss": 4.1165,
      "step": 30910
    },
    {
      "epoch": 0.028822443100551945,
      "grad_norm": 1.1915441751480103,
      "learning_rate": 0.00019423570251385193,
      "loss": 3.9949,
      "step": 30920
    },
    {
      "epoch": 0.028831764718631036,
      "grad_norm": 1.4417967796325684,
      "learning_rate": 0.0001942338381917567,
      "loss": 4.1967,
      "step": 30930
    },
    {
      "epoch": 0.028841086336710127,
      "grad_norm": 1.1624959707260132,
      "learning_rate": 0.0001942319738696615,
      "loss": 4.2487,
      "step": 30940
    },
    {
      "epoch": 0.02885040795478922,
      "grad_norm": 1.2021456956863403,
      "learning_rate": 0.00019423010954756632,
      "loss": 4.0864,
      "step": 30950
    },
    {
      "epoch": 0.028859729572868312,
      "grad_norm": 1.3583601713180542,
      "learning_rate": 0.00019422824522547113,
      "loss": 4.228,
      "step": 30960
    },
    {
      "epoch": 0.028869051190947403,
      "grad_norm": 1.2174726724624634,
      "learning_rate": 0.00019422638090337593,
      "loss": 4.4352,
      "step": 30970
    },
    {
      "epoch": 0.028878372809026494,
      "grad_norm": 1.117678165435791,
      "learning_rate": 0.0001942245165812807,
      "loss": 4.2775,
      "step": 30980
    },
    {
      "epoch": 0.02888769442710559,
      "grad_norm": 1.155847430229187,
      "learning_rate": 0.00019422265225918552,
      "loss": 4.2097,
      "step": 30990
    },
    {
      "epoch": 0.02889701604518468,
      "grad_norm": 1.204359531402588,
      "learning_rate": 0.00019422078793709032,
      "loss": 4.1362,
      "step": 31000
    },
    {
      "epoch": 0.02890633766326377,
      "grad_norm": 1.1896674633026123,
      "learning_rate": 0.00019421892361499513,
      "loss": 4.2555,
      "step": 31010
    },
    {
      "epoch": 0.028915659281342862,
      "grad_norm": 1.188114047050476,
      "learning_rate": 0.00019421705929289994,
      "loss": 4.1158,
      "step": 31020
    },
    {
      "epoch": 0.028924980899421956,
      "grad_norm": 1.0320518016815186,
      "learning_rate": 0.00019421519497080472,
      "loss": 4.3613,
      "step": 31030
    },
    {
      "epoch": 0.028934302517501048,
      "grad_norm": 1.2484323978424072,
      "learning_rate": 0.00019421333064870952,
      "loss": 4.2283,
      "step": 31040
    },
    {
      "epoch": 0.02894362413558014,
      "grad_norm": 1.1570160388946533,
      "learning_rate": 0.00019421146632661433,
      "loss": 4.2602,
      "step": 31050
    },
    {
      "epoch": 0.02895294575365923,
      "grad_norm": 1.0835182666778564,
      "learning_rate": 0.0001942096020045191,
      "loss": 4.1604,
      "step": 31060
    },
    {
      "epoch": 0.02896226737173832,
      "grad_norm": 1.017111897468567,
      "learning_rate": 0.00019420773768242394,
      "loss": 4.2253,
      "step": 31070
    },
    {
      "epoch": 0.028971588989817415,
      "grad_norm": 1.1445168256759644,
      "learning_rate": 0.00019420587336032872,
      "loss": 4.2965,
      "step": 31080
    },
    {
      "epoch": 0.028980910607896506,
      "grad_norm": 1.1842567920684814,
      "learning_rate": 0.00019420400903823353,
      "loss": 4.3223,
      "step": 31090
    },
    {
      "epoch": 0.028990232225975597,
      "grad_norm": 1.0780984163284302,
      "learning_rate": 0.00019420214471613833,
      "loss": 4.1277,
      "step": 31100
    },
    {
      "epoch": 0.028999553844054688,
      "grad_norm": 0.9724211692810059,
      "learning_rate": 0.0001942002803940431,
      "loss": 4.2149,
      "step": 31110
    },
    {
      "epoch": 0.029008875462133783,
      "grad_norm": 1.1847158670425415,
      "learning_rate": 0.00019419841607194792,
      "loss": 4.0666,
      "step": 31120
    },
    {
      "epoch": 0.029018197080212874,
      "grad_norm": 1.1486767530441284,
      "learning_rate": 0.00019419655174985272,
      "loss": 4.135,
      "step": 31130
    },
    {
      "epoch": 0.029027518698291965,
      "grad_norm": 1.1634501218795776,
      "learning_rate": 0.00019419468742775753,
      "loss": 4.1064,
      "step": 31140
    },
    {
      "epoch": 0.029036840316371056,
      "grad_norm": 1.05243718624115,
      "learning_rate": 0.00019419282310566234,
      "loss": 4.0881,
      "step": 31150
    },
    {
      "epoch": 0.02904616193445015,
      "grad_norm": 1.3611392974853516,
      "learning_rate": 0.00019419095878356712,
      "loss": 4.1299,
      "step": 31160
    },
    {
      "epoch": 0.02905548355252924,
      "grad_norm": 1.0489792823791504,
      "learning_rate": 0.00019418909446147192,
      "loss": 4.2966,
      "step": 31170
    },
    {
      "epoch": 0.029064805170608332,
      "grad_norm": 0.9508333802223206,
      "learning_rate": 0.00019418723013937673,
      "loss": 4.2641,
      "step": 31180
    },
    {
      "epoch": 0.029074126788687423,
      "grad_norm": 1.136264443397522,
      "learning_rate": 0.00019418536581728154,
      "loss": 4.1575,
      "step": 31190
    },
    {
      "epoch": 0.029083448406766518,
      "grad_norm": 1.367606520652771,
      "learning_rate": 0.00019418350149518634,
      "loss": 4.1737,
      "step": 31200
    },
    {
      "epoch": 0.02909277002484561,
      "grad_norm": 1.0959076881408691,
      "learning_rate": 0.00019418163717309112,
      "loss": 4.1563,
      "step": 31210
    },
    {
      "epoch": 0.0291020916429247,
      "grad_norm": 1.2323631048202515,
      "learning_rate": 0.00019417977285099593,
      "loss": 4.0066,
      "step": 31220
    },
    {
      "epoch": 0.02911141326100379,
      "grad_norm": 1.3867909908294678,
      "learning_rate": 0.00019417790852890073,
      "loss": 4.0787,
      "step": 31230
    },
    {
      "epoch": 0.02912073487908288,
      "grad_norm": 1.2405859231948853,
      "learning_rate": 0.00019417604420680554,
      "loss": 4.064,
      "step": 31240
    },
    {
      "epoch": 0.029130056497161976,
      "grad_norm": 1.1321849822998047,
      "learning_rate": 0.00019417417988471035,
      "loss": 4.149,
      "step": 31250
    },
    {
      "epoch": 0.029139378115241067,
      "grad_norm": 0.9608215689659119,
      "learning_rate": 0.00019417231556261513,
      "loss": 4.1369,
      "step": 31260
    },
    {
      "epoch": 0.029148699733320158,
      "grad_norm": 1.204519510269165,
      "learning_rate": 0.00019417045124051993,
      "loss": 4.3665,
      "step": 31270
    },
    {
      "epoch": 0.02915802135139925,
      "grad_norm": 0.9490692019462585,
      "learning_rate": 0.0001941685869184247,
      "loss": 4.2159,
      "step": 31280
    },
    {
      "epoch": 0.029167342969478344,
      "grad_norm": 1.1757513284683228,
      "learning_rate": 0.00019416672259632954,
      "loss": 4.2191,
      "step": 31290
    },
    {
      "epoch": 0.029176664587557435,
      "grad_norm": 1.0680192708969116,
      "learning_rate": 0.00019416485827423435,
      "loss": 4.2521,
      "step": 31300
    },
    {
      "epoch": 0.029185986205636526,
      "grad_norm": 1.063636064529419,
      "learning_rate": 0.00019416299395213913,
      "loss": 3.9915,
      "step": 31310
    },
    {
      "epoch": 0.029195307823715617,
      "grad_norm": 1.1657469272613525,
      "learning_rate": 0.00019416112963004394,
      "loss": 4.227,
      "step": 31320
    },
    {
      "epoch": 0.02920462944179471,
      "grad_norm": 1.2537168264389038,
      "learning_rate": 0.00019415926530794872,
      "loss": 4.1102,
      "step": 31330
    },
    {
      "epoch": 0.029213951059873802,
      "grad_norm": 1.0436699390411377,
      "learning_rate": 0.00019415740098585352,
      "loss": 4.1129,
      "step": 31340
    },
    {
      "epoch": 0.029223272677952893,
      "grad_norm": 1.4360355138778687,
      "learning_rate": 0.00019415553666375835,
      "loss": 4.104,
      "step": 31350
    },
    {
      "epoch": 0.029232594296031984,
      "grad_norm": 1.3172998428344727,
      "learning_rate": 0.00019415367234166313,
      "loss": 4.2148,
      "step": 31360
    },
    {
      "epoch": 0.02924191591411108,
      "grad_norm": 1.1695877313613892,
      "learning_rate": 0.00019415180801956794,
      "loss": 4.2419,
      "step": 31370
    },
    {
      "epoch": 0.02925123753219017,
      "grad_norm": 0.9640228748321533,
      "learning_rate": 0.00019414994369747272,
      "loss": 4.1166,
      "step": 31380
    },
    {
      "epoch": 0.02926055915026926,
      "grad_norm": 1.1064560413360596,
      "learning_rate": 0.00019414807937537753,
      "loss": 4.1758,
      "step": 31390
    },
    {
      "epoch": 0.029269880768348352,
      "grad_norm": 1.2661802768707275,
      "learning_rate": 0.00019414621505328233,
      "loss": 4.2187,
      "step": 31400
    },
    {
      "epoch": 0.029279202386427443,
      "grad_norm": 1.1263340711593628,
      "learning_rate": 0.00019414435073118714,
      "loss": 4.4257,
      "step": 31410
    },
    {
      "epoch": 0.029288524004506537,
      "grad_norm": 1.114465594291687,
      "learning_rate": 0.00019414248640909194,
      "loss": 4.153,
      "step": 31420
    },
    {
      "epoch": 0.02929784562258563,
      "grad_norm": 1.2296165227890015,
      "learning_rate": 0.00019414062208699672,
      "loss": 4.265,
      "step": 31430
    },
    {
      "epoch": 0.02930716724066472,
      "grad_norm": 1.3024330139160156,
      "learning_rate": 0.00019413875776490153,
      "loss": 4.2446,
      "step": 31440
    },
    {
      "epoch": 0.02931648885874381,
      "grad_norm": 1.0702215433120728,
      "learning_rate": 0.00019413689344280634,
      "loss": 4.3904,
      "step": 31450
    },
    {
      "epoch": 0.029325810476822905,
      "grad_norm": 1.0669704675674438,
      "learning_rate": 0.00019413502912071114,
      "loss": 4.358,
      "step": 31460
    },
    {
      "epoch": 0.029335132094901996,
      "grad_norm": 1.2334884405136108,
      "learning_rate": 0.00019413316479861595,
      "loss": 4.1798,
      "step": 31470
    },
    {
      "epoch": 0.029344453712981087,
      "grad_norm": 0.9617350697517395,
      "learning_rate": 0.00019413130047652073,
      "loss": 4.2367,
      "step": 31480
    },
    {
      "epoch": 0.029353775331060178,
      "grad_norm": 1.147331953048706,
      "learning_rate": 0.00019412943615442553,
      "loss": 4.3344,
      "step": 31490
    },
    {
      "epoch": 0.029363096949139272,
      "grad_norm": 1.0954643487930298,
      "learning_rate": 0.00019412757183233034,
      "loss": 4.1375,
      "step": 31500
    },
    {
      "epoch": 0.029372418567218363,
      "grad_norm": 1.1036428213119507,
      "learning_rate": 0.00019412570751023512,
      "loss": 4.1186,
      "step": 31510
    },
    {
      "epoch": 0.029381740185297454,
      "grad_norm": 1.195631980895996,
      "learning_rate": 0.00019412384318813995,
      "loss": 4.1619,
      "step": 31520
    },
    {
      "epoch": 0.029391061803376545,
      "grad_norm": 1.191128134727478,
      "learning_rate": 0.00019412197886604473,
      "loss": 4.1943,
      "step": 31530
    },
    {
      "epoch": 0.02940038342145564,
      "grad_norm": 1.174415111541748,
      "learning_rate": 0.00019412011454394954,
      "loss": 4.4154,
      "step": 31540
    },
    {
      "epoch": 0.02940970503953473,
      "grad_norm": 1.3919439315795898,
      "learning_rate": 0.00019411825022185435,
      "loss": 4.2865,
      "step": 31550
    },
    {
      "epoch": 0.029419026657613822,
      "grad_norm": 0.9812098741531372,
      "learning_rate": 0.00019411638589975912,
      "loss": 4.0383,
      "step": 31560
    },
    {
      "epoch": 0.029428348275692913,
      "grad_norm": 1.1779046058654785,
      "learning_rate": 0.00019411452157766393,
      "loss": 3.9088,
      "step": 31570
    },
    {
      "epoch": 0.029437669893772004,
      "grad_norm": 1.151163101196289,
      "learning_rate": 0.00019411265725556874,
      "loss": 4.0641,
      "step": 31580
    },
    {
      "epoch": 0.0294469915118511,
      "grad_norm": 1.3185166120529175,
      "learning_rate": 0.00019411079293347354,
      "loss": 4.1538,
      "step": 31590
    },
    {
      "epoch": 0.02945631312993019,
      "grad_norm": 1.1031758785247803,
      "learning_rate": 0.00019410892861137835,
      "loss": 4.3333,
      "step": 31600
    },
    {
      "epoch": 0.02946563474800928,
      "grad_norm": 1.2568800449371338,
      "learning_rate": 0.00019410706428928313,
      "loss": 4.2113,
      "step": 31610
    },
    {
      "epoch": 0.02947495636608837,
      "grad_norm": 1.3528683185577393,
      "learning_rate": 0.00019410519996718794,
      "loss": 4.2897,
      "step": 31620
    },
    {
      "epoch": 0.029484277984167466,
      "grad_norm": 1.2947696447372437,
      "learning_rate": 0.00019410333564509274,
      "loss": 4.0942,
      "step": 31630
    },
    {
      "epoch": 0.029493599602246557,
      "grad_norm": 0.8445706367492676,
      "learning_rate": 0.00019410147132299755,
      "loss": 4.2048,
      "step": 31640
    },
    {
      "epoch": 0.029502921220325648,
      "grad_norm": 0.9530462622642517,
      "learning_rate": 0.00019409960700090235,
      "loss": 4.3569,
      "step": 31650
    },
    {
      "epoch": 0.02951224283840474,
      "grad_norm": 1.1190719604492188,
      "learning_rate": 0.00019409774267880713,
      "loss": 3.9597,
      "step": 31660
    },
    {
      "epoch": 0.029521564456483834,
      "grad_norm": 1.1711585521697998,
      "learning_rate": 0.00019409587835671194,
      "loss": 4.0377,
      "step": 31670
    },
    {
      "epoch": 0.029530886074562925,
      "grad_norm": 1.179612159729004,
      "learning_rate": 0.00019409401403461675,
      "loss": 4.2546,
      "step": 31680
    },
    {
      "epoch": 0.029540207692642016,
      "grad_norm": 1.215432047843933,
      "learning_rate": 0.00019409214971252155,
      "loss": 4.2353,
      "step": 31690
    },
    {
      "epoch": 0.029549529310721107,
      "grad_norm": 0.892830491065979,
      "learning_rate": 0.00019409028539042636,
      "loss": 4.3534,
      "step": 31700
    },
    {
      "epoch": 0.0295588509288002,
      "grad_norm": 1.0554001331329346,
      "learning_rate": 0.00019408842106833114,
      "loss": 4.1723,
      "step": 31710
    },
    {
      "epoch": 0.029568172546879292,
      "grad_norm": 0.89695143699646,
      "learning_rate": 0.00019408655674623594,
      "loss": 4.1878,
      "step": 31720
    },
    {
      "epoch": 0.029577494164958383,
      "grad_norm": 1.1932203769683838,
      "learning_rate": 0.00019408469242414072,
      "loss": 4.2343,
      "step": 31730
    },
    {
      "epoch": 0.029586815783037474,
      "grad_norm": 0.9796647429466248,
      "learning_rate": 0.00019408282810204556,
      "loss": 4.1453,
      "step": 31740
    },
    {
      "epoch": 0.029596137401116565,
      "grad_norm": 1.210918664932251,
      "learning_rate": 0.00019408096377995036,
      "loss": 4.2629,
      "step": 31750
    },
    {
      "epoch": 0.02960545901919566,
      "grad_norm": 1.1470444202423096,
      "learning_rate": 0.00019407909945785514,
      "loss": 4.1109,
      "step": 31760
    },
    {
      "epoch": 0.02961478063727475,
      "grad_norm": 1.014633297920227,
      "learning_rate": 0.00019407723513575995,
      "loss": 4.1871,
      "step": 31770
    },
    {
      "epoch": 0.02962410225535384,
      "grad_norm": 1.1396725177764893,
      "learning_rate": 0.00019407537081366473,
      "loss": 4.2276,
      "step": 31780
    },
    {
      "epoch": 0.029633423873432933,
      "grad_norm": 1.1600899696350098,
      "learning_rate": 0.00019407350649156953,
      "loss": 4.1908,
      "step": 31790
    },
    {
      "epoch": 0.029642745491512027,
      "grad_norm": 1.1420100927352905,
      "learning_rate": 0.00019407164216947437,
      "loss": 4.101,
      "step": 31800
    },
    {
      "epoch": 0.029652067109591118,
      "grad_norm": 0.9802015423774719,
      "learning_rate": 0.00019406977784737915,
      "loss": 4.2922,
      "step": 31810
    },
    {
      "epoch": 0.02966138872767021,
      "grad_norm": 1.1802209615707397,
      "learning_rate": 0.00019406791352528395,
      "loss": 3.9972,
      "step": 31820
    },
    {
      "epoch": 0.0296707103457493,
      "grad_norm": 1.2083922624588013,
      "learning_rate": 0.00019406604920318873,
      "loss": 4.2324,
      "step": 31830
    },
    {
      "epoch": 0.029680031963828395,
      "grad_norm": 1.14492928981781,
      "learning_rate": 0.00019406418488109354,
      "loss": 4.2206,
      "step": 31840
    },
    {
      "epoch": 0.029689353581907486,
      "grad_norm": 1.1457021236419678,
      "learning_rate": 0.00019406232055899834,
      "loss": 4.2076,
      "step": 31850
    },
    {
      "epoch": 0.029698675199986577,
      "grad_norm": 1.0566233396530151,
      "learning_rate": 0.00019406045623690315,
      "loss": 4.1471,
      "step": 31860
    },
    {
      "epoch": 0.029707996818065668,
      "grad_norm": 1.1441012620925903,
      "learning_rate": 0.00019405859191480796,
      "loss": 4.3295,
      "step": 31870
    },
    {
      "epoch": 0.029717318436144762,
      "grad_norm": 0.9331770539283752,
      "learning_rate": 0.00019405672759271274,
      "loss": 4.3818,
      "step": 31880
    },
    {
      "epoch": 0.029726640054223853,
      "grad_norm": 1.3024044036865234,
      "learning_rate": 0.00019405486327061754,
      "loss": 4.1346,
      "step": 31890
    },
    {
      "epoch": 0.029735961672302944,
      "grad_norm": 1.1556702852249146,
      "learning_rate": 0.00019405299894852235,
      "loss": 4.3213,
      "step": 31900
    },
    {
      "epoch": 0.029745283290382035,
      "grad_norm": 1.0790272951126099,
      "learning_rate": 0.00019405113462642716,
      "loss": 4.2469,
      "step": 31910
    },
    {
      "epoch": 0.029754604908461126,
      "grad_norm": 1.0064516067504883,
      "learning_rate": 0.00019404927030433196,
      "loss": 4.1735,
      "step": 31920
    },
    {
      "epoch": 0.02976392652654022,
      "grad_norm": 1.1206207275390625,
      "learning_rate": 0.00019404740598223674,
      "loss": 4.1239,
      "step": 31930
    },
    {
      "epoch": 0.029773248144619312,
      "grad_norm": 1.687496304512024,
      "learning_rate": 0.00019404554166014155,
      "loss": 4.2619,
      "step": 31940
    },
    {
      "epoch": 0.029782569762698403,
      "grad_norm": 1.2612522840499878,
      "learning_rate": 0.00019404367733804635,
      "loss": 4.3353,
      "step": 31950
    },
    {
      "epoch": 0.029791891380777494,
      "grad_norm": 0.9138712286949158,
      "learning_rate": 0.00019404181301595113,
      "loss": 4.1684,
      "step": 31960
    },
    {
      "epoch": 0.029801212998856588,
      "grad_norm": 0.9929453134536743,
      "learning_rate": 0.00019403994869385597,
      "loss": 4.0885,
      "step": 31970
    },
    {
      "epoch": 0.02981053461693568,
      "grad_norm": 0.9931709170341492,
      "learning_rate": 0.00019403808437176075,
      "loss": 4.2317,
      "step": 31980
    },
    {
      "epoch": 0.02981985623501477,
      "grad_norm": 0.8862972259521484,
      "learning_rate": 0.00019403622004966555,
      "loss": 4.272,
      "step": 31990
    },
    {
      "epoch": 0.02982917785309386,
      "grad_norm": 1.1282144784927368,
      "learning_rate": 0.00019403435572757036,
      "loss": 4.0915,
      "step": 32000
    },
    {
      "epoch": 0.029838499471172956,
      "grad_norm": 1.0891878604888916,
      "learning_rate": 0.00019403249140547514,
      "loss": 4.1957,
      "step": 32010
    },
    {
      "epoch": 0.029847821089252047,
      "grad_norm": 1.1402757167816162,
      "learning_rate": 0.00019403062708337994,
      "loss": 4.2495,
      "step": 32020
    },
    {
      "epoch": 0.029857142707331138,
      "grad_norm": 1.0377819538116455,
      "learning_rate": 0.00019402876276128475,
      "loss": 4.23,
      "step": 32030
    },
    {
      "epoch": 0.02986646432541023,
      "grad_norm": 1.2140430212020874,
      "learning_rate": 0.00019402689843918956,
      "loss": 4.0372,
      "step": 32040
    },
    {
      "epoch": 0.029875785943489323,
      "grad_norm": 1.2643791437149048,
      "learning_rate": 0.00019402503411709436,
      "loss": 4.0696,
      "step": 32050
    },
    {
      "epoch": 0.029885107561568414,
      "grad_norm": 0.9341362118721008,
      "learning_rate": 0.00019402316979499914,
      "loss": 4.2514,
      "step": 32060
    },
    {
      "epoch": 0.029894429179647505,
      "grad_norm": 1.2274776697158813,
      "learning_rate": 0.00019402130547290395,
      "loss": 4.2663,
      "step": 32070
    },
    {
      "epoch": 0.029903750797726596,
      "grad_norm": 0.9787795543670654,
      "learning_rate": 0.00019401944115080875,
      "loss": 4.3167,
      "step": 32080
    },
    {
      "epoch": 0.029913072415805687,
      "grad_norm": 1.141086459159851,
      "learning_rate": 0.00019401757682871356,
      "loss": 4.4211,
      "step": 32090
    },
    {
      "epoch": 0.029922394033884782,
      "grad_norm": 0.9898458123207092,
      "learning_rate": 0.00019401571250661837,
      "loss": 4.0699,
      "step": 32100
    },
    {
      "epoch": 0.029931715651963873,
      "grad_norm": 1.0336744785308838,
      "learning_rate": 0.00019401384818452315,
      "loss": 4.2073,
      "step": 32110
    },
    {
      "epoch": 0.029941037270042964,
      "grad_norm": 1.2898985147476196,
      "learning_rate": 0.00019401198386242795,
      "loss": 4.3292,
      "step": 32120
    },
    {
      "epoch": 0.029950358888122055,
      "grad_norm": 1.0010024309158325,
      "learning_rate": 0.00019401011954033276,
      "loss": 4.2399,
      "step": 32130
    },
    {
      "epoch": 0.02995968050620115,
      "grad_norm": 1.346633791923523,
      "learning_rate": 0.00019400825521823757,
      "loss": 4.296,
      "step": 32140
    },
    {
      "epoch": 0.02996900212428024,
      "grad_norm": 1.1327159404754639,
      "learning_rate": 0.00019400639089614237,
      "loss": 4.379,
      "step": 32150
    },
    {
      "epoch": 0.02997832374235933,
      "grad_norm": 1.0866341590881348,
      "learning_rate": 0.00019400452657404715,
      "loss": 4.1798,
      "step": 32160
    },
    {
      "epoch": 0.029987645360438422,
      "grad_norm": 1.168637752532959,
      "learning_rate": 0.00019400266225195196,
      "loss": 4.1051,
      "step": 32170
    },
    {
      "epoch": 0.029996966978517517,
      "grad_norm": 1.108092188835144,
      "learning_rate": 0.00019400079792985674,
      "loss": 4.1704,
      "step": 32180
    },
    {
      "epoch": 0.030006288596596608,
      "grad_norm": 1.1671338081359863,
      "learning_rate": 0.00019399893360776157,
      "loss": 4.2702,
      "step": 32190
    },
    {
      "epoch": 0.0300156102146757,
      "grad_norm": 1.1662986278533936,
      "learning_rate": 0.00019399706928566638,
      "loss": 4.0605,
      "step": 32200
    },
    {
      "epoch": 0.03002493183275479,
      "grad_norm": 1.2726211547851562,
      "learning_rate": 0.00019399520496357116,
      "loss": 4.0358,
      "step": 32210
    },
    {
      "epoch": 0.030034253450833884,
      "grad_norm": 1.0287293195724487,
      "learning_rate": 0.00019399334064147596,
      "loss": 4.1769,
      "step": 32220
    },
    {
      "epoch": 0.030043575068912975,
      "grad_norm": 1.0439114570617676,
      "learning_rate": 0.00019399147631938074,
      "loss": 4.3865,
      "step": 32230
    },
    {
      "epoch": 0.030052896686992066,
      "grad_norm": 1.1231848001480103,
      "learning_rate": 0.00019398961199728555,
      "loss": 4.0978,
      "step": 32240
    },
    {
      "epoch": 0.030062218305071157,
      "grad_norm": 1.1679633855819702,
      "learning_rate": 0.00019398774767519035,
      "loss": 4.2357,
      "step": 32250
    },
    {
      "epoch": 0.03007153992315025,
      "grad_norm": 1.0517910718917847,
      "learning_rate": 0.00019398588335309516,
      "loss": 4.2264,
      "step": 32260
    },
    {
      "epoch": 0.030080861541229343,
      "grad_norm": 1.0992600917816162,
      "learning_rate": 0.00019398401903099997,
      "loss": 4.2709,
      "step": 32270
    },
    {
      "epoch": 0.030090183159308434,
      "grad_norm": 1.1717044115066528,
      "learning_rate": 0.00019398215470890475,
      "loss": 4.2558,
      "step": 32280
    },
    {
      "epoch": 0.030099504777387525,
      "grad_norm": 1.1010996103286743,
      "learning_rate": 0.00019398029038680955,
      "loss": 4.2759,
      "step": 32290
    },
    {
      "epoch": 0.030108826395466616,
      "grad_norm": 1.1503701210021973,
      "learning_rate": 0.00019397842606471436,
      "loss": 4.1467,
      "step": 32300
    },
    {
      "epoch": 0.03011814801354571,
      "grad_norm": 1.0315122604370117,
      "learning_rate": 0.00019397656174261916,
      "loss": 4.1384,
      "step": 32310
    },
    {
      "epoch": 0.0301274696316248,
      "grad_norm": 1.0681538581848145,
      "learning_rate": 0.00019397469742052397,
      "loss": 4.1508,
      "step": 32320
    },
    {
      "epoch": 0.030136791249703893,
      "grad_norm": 1.2215549945831299,
      "learning_rate": 0.00019397283309842875,
      "loss": 4.1925,
      "step": 32330
    },
    {
      "epoch": 0.030146112867782984,
      "grad_norm": 0.9949830174446106,
      "learning_rate": 0.00019397096877633356,
      "loss": 4.2258,
      "step": 32340
    },
    {
      "epoch": 0.030155434485862078,
      "grad_norm": 1.0598527193069458,
      "learning_rate": 0.00019396910445423834,
      "loss": 4.1761,
      "step": 32350
    },
    {
      "epoch": 0.03016475610394117,
      "grad_norm": 1.0225567817687988,
      "learning_rate": 0.00019396724013214317,
      "loss": 4.2552,
      "step": 32360
    },
    {
      "epoch": 0.03017407772202026,
      "grad_norm": 1.1570672988891602,
      "learning_rate": 0.00019396537581004797,
      "loss": 4.1729,
      "step": 32370
    },
    {
      "epoch": 0.03018339934009935,
      "grad_norm": 1.2768572568893433,
      "learning_rate": 0.00019396351148795275,
      "loss": 4.1316,
      "step": 32380
    },
    {
      "epoch": 0.030192720958178446,
      "grad_norm": 1.2195054292678833,
      "learning_rate": 0.00019396164716585756,
      "loss": 4.2414,
      "step": 32390
    },
    {
      "epoch": 0.030202042576257537,
      "grad_norm": 1.1502571105957031,
      "learning_rate": 0.00019395978284376234,
      "loss": 4.3156,
      "step": 32400
    },
    {
      "epoch": 0.030211364194336628,
      "grad_norm": 1.1636807918548584,
      "learning_rate": 0.00019395791852166715,
      "loss": 4.117,
      "step": 32410
    },
    {
      "epoch": 0.03022068581241572,
      "grad_norm": 1.1960997581481934,
      "learning_rate": 0.00019395605419957198,
      "loss": 4.027,
      "step": 32420
    },
    {
      "epoch": 0.03023000743049481,
      "grad_norm": 0.987415611743927,
      "learning_rate": 0.00019395418987747676,
      "loss": 4.1607,
      "step": 32430
    },
    {
      "epoch": 0.030239329048573904,
      "grad_norm": 1.167460560798645,
      "learning_rate": 0.00019395232555538156,
      "loss": 4.2561,
      "step": 32440
    },
    {
      "epoch": 0.030248650666652995,
      "grad_norm": 1.0048154592514038,
      "learning_rate": 0.00019395046123328634,
      "loss": 4.1252,
      "step": 32450
    },
    {
      "epoch": 0.030257972284732086,
      "grad_norm": 1.1638702154159546,
      "learning_rate": 0.00019394859691119115,
      "loss": 4.3054,
      "step": 32460
    },
    {
      "epoch": 0.030267293902811177,
      "grad_norm": 1.1017528772354126,
      "learning_rate": 0.00019394673258909596,
      "loss": 4.2152,
      "step": 32470
    },
    {
      "epoch": 0.03027661552089027,
      "grad_norm": 1.070324182510376,
      "learning_rate": 0.00019394486826700076,
      "loss": 4.0941,
      "step": 32480
    },
    {
      "epoch": 0.030285937138969363,
      "grad_norm": 1.2833747863769531,
      "learning_rate": 0.00019394300394490557,
      "loss": 4.2217,
      "step": 32490
    },
    {
      "epoch": 0.030295258757048454,
      "grad_norm": 1.2014143466949463,
      "learning_rate": 0.00019394113962281035,
      "loss": 4.0495,
      "step": 32500
    },
    {
      "epoch": 0.030304580375127545,
      "grad_norm": 1.3043174743652344,
      "learning_rate": 0.00019393927530071515,
      "loss": 4.2039,
      "step": 32510
    },
    {
      "epoch": 0.03031390199320664,
      "grad_norm": 1.2151046991348267,
      "learning_rate": 0.00019393741097861996,
      "loss": 4.1512,
      "step": 32520
    },
    {
      "epoch": 0.03032322361128573,
      "grad_norm": 1.313153624534607,
      "learning_rate": 0.00019393554665652477,
      "loss": 4.297,
      "step": 32530
    },
    {
      "epoch": 0.03033254522936482,
      "grad_norm": 0.9782241582870483,
      "learning_rate": 0.00019393368233442957,
      "loss": 4.3288,
      "step": 32540
    },
    {
      "epoch": 0.030341866847443912,
      "grad_norm": 0.8962041735649109,
      "learning_rate": 0.00019393181801233435,
      "loss": 4.2201,
      "step": 32550
    },
    {
      "epoch": 0.030351188465523007,
      "grad_norm": 1.0422885417938232,
      "learning_rate": 0.00019392995369023916,
      "loss": 4.0758,
      "step": 32560
    },
    {
      "epoch": 0.030360510083602098,
      "grad_norm": 1.294143795967102,
      "learning_rate": 0.00019392808936814397,
      "loss": 4.1775,
      "step": 32570
    },
    {
      "epoch": 0.03036983170168119,
      "grad_norm": 1.0633379220962524,
      "learning_rate": 0.00019392622504604877,
      "loss": 4.1751,
      "step": 32580
    },
    {
      "epoch": 0.03037915331976028,
      "grad_norm": 1.1462589502334595,
      "learning_rate": 0.00019392436072395358,
      "loss": 4.343,
      "step": 32590
    },
    {
      "epoch": 0.03038847493783937,
      "grad_norm": 1.0946015119552612,
      "learning_rate": 0.00019392249640185836,
      "loss": 3.9879,
      "step": 32600
    },
    {
      "epoch": 0.030397796555918465,
      "grad_norm": 1.110753059387207,
      "learning_rate": 0.00019392063207976316,
      "loss": 4.1237,
      "step": 32610
    },
    {
      "epoch": 0.030407118173997556,
      "grad_norm": 1.3405119180679321,
      "learning_rate": 0.00019391876775766797,
      "loss": 4.1558,
      "step": 32620
    },
    {
      "epoch": 0.030416439792076647,
      "grad_norm": 1.1808550357818604,
      "learning_rate": 0.00019391690343557275,
      "loss": 4.4158,
      "step": 32630
    },
    {
      "epoch": 0.03042576141015574,
      "grad_norm": 1.0419751405715942,
      "learning_rate": 0.00019391503911347758,
      "loss": 4.1521,
      "step": 32640
    },
    {
      "epoch": 0.030435083028234833,
      "grad_norm": 1.0984809398651123,
      "learning_rate": 0.00019391317479138236,
      "loss": 4.0534,
      "step": 32650
    },
    {
      "epoch": 0.030444404646313924,
      "grad_norm": 1.303112506866455,
      "learning_rate": 0.00019391131046928717,
      "loss": 4.2113,
      "step": 32660
    },
    {
      "epoch": 0.030453726264393015,
      "grad_norm": 1.1434563398361206,
      "learning_rate": 0.00019390944614719197,
      "loss": 4.1903,
      "step": 32670
    },
    {
      "epoch": 0.030463047882472106,
      "grad_norm": 1.1680105924606323,
      "learning_rate": 0.00019390758182509675,
      "loss": 4.0452,
      "step": 32680
    },
    {
      "epoch": 0.0304723695005512,
      "grad_norm": 1.2627942562103271,
      "learning_rate": 0.00019390571750300156,
      "loss": 3.9991,
      "step": 32690
    },
    {
      "epoch": 0.03048169111863029,
      "grad_norm": 1.0457220077514648,
      "learning_rate": 0.00019390385318090637,
      "loss": 4.2862,
      "step": 32700
    },
    {
      "epoch": 0.030491012736709382,
      "grad_norm": 1.236045479774475,
      "learning_rate": 0.00019390198885881117,
      "loss": 4.2108,
      "step": 32710
    },
    {
      "epoch": 0.030500334354788473,
      "grad_norm": 1.1695642471313477,
      "learning_rate": 0.00019390012453671598,
      "loss": 4.1277,
      "step": 32720
    },
    {
      "epoch": 0.030509655972867568,
      "grad_norm": 1.1890162229537964,
      "learning_rate": 0.00019389826021462076,
      "loss": 4.2326,
      "step": 32730
    },
    {
      "epoch": 0.03051897759094666,
      "grad_norm": 1.0654661655426025,
      "learning_rate": 0.00019389639589252556,
      "loss": 4.1245,
      "step": 32740
    },
    {
      "epoch": 0.03052829920902575,
      "grad_norm": 1.3514894247055054,
      "learning_rate": 0.00019389453157043037,
      "loss": 4.2261,
      "step": 32750
    },
    {
      "epoch": 0.03053762082710484,
      "grad_norm": 1.1652426719665527,
      "learning_rate": 0.00019389266724833518,
      "loss": 4.1811,
      "step": 32760
    },
    {
      "epoch": 0.030546942445183932,
      "grad_norm": 1.0079636573791504,
      "learning_rate": 0.00019389080292623998,
      "loss": 4.2569,
      "step": 32770
    },
    {
      "epoch": 0.030556264063263026,
      "grad_norm": 1.2339545488357544,
      "learning_rate": 0.00019388893860414476,
      "loss": 4.0524,
      "step": 32780
    },
    {
      "epoch": 0.030565585681342117,
      "grad_norm": 1.1250815391540527,
      "learning_rate": 0.00019388707428204957,
      "loss": 4.3477,
      "step": 32790
    },
    {
      "epoch": 0.03057490729942121,
      "grad_norm": 1.0196667909622192,
      "learning_rate": 0.00019388520995995435,
      "loss": 4.2236,
      "step": 32800
    },
    {
      "epoch": 0.0305842289175003,
      "grad_norm": 0.949111819267273,
      "learning_rate": 0.00019388334563785918,
      "loss": 3.901,
      "step": 32810
    },
    {
      "epoch": 0.030593550535579394,
      "grad_norm": 1.0138804912567139,
      "learning_rate": 0.000193881481315764,
      "loss": 4.0771,
      "step": 32820
    },
    {
      "epoch": 0.030602872153658485,
      "grad_norm": 1.1246514320373535,
      "learning_rate": 0.00019387961699366877,
      "loss": 4.1758,
      "step": 32830
    },
    {
      "epoch": 0.030612193771737576,
      "grad_norm": 1.1863481998443604,
      "learning_rate": 0.00019387775267157357,
      "loss": 4.1521,
      "step": 32840
    },
    {
      "epoch": 0.030621515389816667,
      "grad_norm": 0.929152250289917,
      "learning_rate": 0.00019387588834947835,
      "loss": 4.0525,
      "step": 32850
    },
    {
      "epoch": 0.03063083700789576,
      "grad_norm": 1.3190611600875854,
      "learning_rate": 0.00019387402402738316,
      "loss": 4.105,
      "step": 32860
    },
    {
      "epoch": 0.030640158625974852,
      "grad_norm": 0.9306054711341858,
      "learning_rate": 0.000193872159705288,
      "loss": 4.1409,
      "step": 32870
    },
    {
      "epoch": 0.030649480244053944,
      "grad_norm": 1.2419893741607666,
      "learning_rate": 0.00019387029538319277,
      "loss": 4.1708,
      "step": 32880
    },
    {
      "epoch": 0.030658801862133035,
      "grad_norm": 1.2410808801651,
      "learning_rate": 0.00019386843106109758,
      "loss": 4.0882,
      "step": 32890
    },
    {
      "epoch": 0.03066812348021213,
      "grad_norm": 1.1140642166137695,
      "learning_rate": 0.00019386656673900236,
      "loss": 4.0428,
      "step": 32900
    },
    {
      "epoch": 0.03067744509829122,
      "grad_norm": 1.300079107284546,
      "learning_rate": 0.00019386470241690716,
      "loss": 4.1463,
      "step": 32910
    },
    {
      "epoch": 0.03068676671637031,
      "grad_norm": 1.00928795337677,
      "learning_rate": 0.00019386283809481197,
      "loss": 4.1983,
      "step": 32920
    },
    {
      "epoch": 0.030696088334449402,
      "grad_norm": 1.5210810899734497,
      "learning_rate": 0.00019386097377271678,
      "loss": 3.9721,
      "step": 32930
    },
    {
      "epoch": 0.030705409952528493,
      "grad_norm": 1.1784920692443848,
      "learning_rate": 0.00019385910945062158,
      "loss": 4.2651,
      "step": 32940
    },
    {
      "epoch": 0.030714731570607588,
      "grad_norm": 1.2530821561813354,
      "learning_rate": 0.00019385724512852636,
      "loss": 4.2973,
      "step": 32950
    },
    {
      "epoch": 0.03072405318868668,
      "grad_norm": 1.154852271080017,
      "learning_rate": 0.00019385538080643117,
      "loss": 4.2372,
      "step": 32960
    },
    {
      "epoch": 0.03073337480676577,
      "grad_norm": 1.23934805393219,
      "learning_rate": 0.00019385351648433597,
      "loss": 4.24,
      "step": 32970
    },
    {
      "epoch": 0.03074269642484486,
      "grad_norm": 1.0756442546844482,
      "learning_rate": 0.00019385165216224078,
      "loss": 4.2935,
      "step": 32980
    },
    {
      "epoch": 0.030752018042923955,
      "grad_norm": 0.8530885577201843,
      "learning_rate": 0.0001938497878401456,
      "loss": 4.2498,
      "step": 32990
    },
    {
      "epoch": 0.030761339661003046,
      "grad_norm": 1.1073813438415527,
      "learning_rate": 0.00019384792351805037,
      "loss": 4.0408,
      "step": 33000
    },
    {
      "epoch": 0.030770661279082137,
      "grad_norm": 1.1500654220581055,
      "learning_rate": 0.00019384605919595517,
      "loss": 4.3321,
      "step": 33010
    },
    {
      "epoch": 0.030779982897161228,
      "grad_norm": 1.0076868534088135,
      "learning_rate": 0.00019384419487385998,
      "loss": 4.117,
      "step": 33020
    },
    {
      "epoch": 0.030789304515240323,
      "grad_norm": 1.2810779809951782,
      "learning_rate": 0.00019384233055176478,
      "loss": 4.1931,
      "step": 33030
    },
    {
      "epoch": 0.030798626133319414,
      "grad_norm": 0.9586958289146423,
      "learning_rate": 0.0001938404662296696,
      "loss": 4.0378,
      "step": 33040
    },
    {
      "epoch": 0.030807947751398505,
      "grad_norm": 1.0542206764221191,
      "learning_rate": 0.00019383860190757437,
      "loss": 4.2382,
      "step": 33050
    },
    {
      "epoch": 0.030817269369477596,
      "grad_norm": 1.1495213508605957,
      "learning_rate": 0.00019383673758547918,
      "loss": 4.2617,
      "step": 33060
    },
    {
      "epoch": 0.03082659098755669,
      "grad_norm": 1.2933770418167114,
      "learning_rate": 0.00019383487326338398,
      "loss": 4.1098,
      "step": 33070
    },
    {
      "epoch": 0.03083591260563578,
      "grad_norm": 1.0857980251312256,
      "learning_rate": 0.00019383300894128876,
      "loss": 4.1136,
      "step": 33080
    },
    {
      "epoch": 0.030845234223714872,
      "grad_norm": 1.087378978729248,
      "learning_rate": 0.0001938311446191936,
      "loss": 4.196,
      "step": 33090
    },
    {
      "epoch": 0.030854555841793963,
      "grad_norm": 1.0370914936065674,
      "learning_rate": 0.00019382928029709837,
      "loss": 4.1033,
      "step": 33100
    },
    {
      "epoch": 0.030863877459873054,
      "grad_norm": 1.0487580299377441,
      "learning_rate": 0.00019382741597500318,
      "loss": 4.3284,
      "step": 33110
    },
    {
      "epoch": 0.03087319907795215,
      "grad_norm": 1.4315812587738037,
      "learning_rate": 0.000193825551652908,
      "loss": 4.0639,
      "step": 33120
    },
    {
      "epoch": 0.03088252069603124,
      "grad_norm": 0.9930644035339355,
      "learning_rate": 0.00019382368733081277,
      "loss": 4.2782,
      "step": 33130
    },
    {
      "epoch": 0.03089184231411033,
      "grad_norm": 1.1126712560653687,
      "learning_rate": 0.00019382182300871757,
      "loss": 4.3055,
      "step": 33140
    },
    {
      "epoch": 0.03090116393218942,
      "grad_norm": 1.0667608976364136,
      "learning_rate": 0.00019381995868662238,
      "loss": 4.3108,
      "step": 33150
    },
    {
      "epoch": 0.030910485550268516,
      "grad_norm": 1.165819525718689,
      "learning_rate": 0.00019381809436452719,
      "loss": 4.0155,
      "step": 33160
    },
    {
      "epoch": 0.030919807168347607,
      "grad_norm": 1.0682085752487183,
      "learning_rate": 0.000193816230042432,
      "loss": 3.858,
      "step": 33170
    },
    {
      "epoch": 0.030929128786426698,
      "grad_norm": 1.24373197555542,
      "learning_rate": 0.00019381436572033677,
      "loss": 4.2599,
      "step": 33180
    },
    {
      "epoch": 0.03093845040450579,
      "grad_norm": 1.0636913776397705,
      "learning_rate": 0.00019381250139824158,
      "loss": 4.0304,
      "step": 33190
    },
    {
      "epoch": 0.030947772022584884,
      "grad_norm": 0.9426274299621582,
      "learning_rate": 0.00019381063707614638,
      "loss": 4.1354,
      "step": 33200
    },
    {
      "epoch": 0.030957093640663975,
      "grad_norm": 1.088461995124817,
      "learning_rate": 0.0001938087727540512,
      "loss": 4.0797,
      "step": 33210
    },
    {
      "epoch": 0.030966415258743066,
      "grad_norm": 1.3038103580474854,
      "learning_rate": 0.000193806908431956,
      "loss": 4.1388,
      "step": 33220
    },
    {
      "epoch": 0.030975736876822157,
      "grad_norm": 1.159716248512268,
      "learning_rate": 0.00019380504410986078,
      "loss": 4.1316,
      "step": 33230
    },
    {
      "epoch": 0.03098505849490125,
      "grad_norm": 1.1869653463363647,
      "learning_rate": 0.00019380317978776558,
      "loss": 4.0832,
      "step": 33240
    },
    {
      "epoch": 0.030994380112980342,
      "grad_norm": 1.203855037689209,
      "learning_rate": 0.0001938013154656704,
      "loss": 4.0701,
      "step": 33250
    },
    {
      "epoch": 0.031003701731059433,
      "grad_norm": 1.1782304048538208,
      "learning_rate": 0.0001937994511435752,
      "loss": 4.1087,
      "step": 33260
    },
    {
      "epoch": 0.031013023349138524,
      "grad_norm": 1.0877647399902344,
      "learning_rate": 0.00019379758682148,
      "loss": 4.2157,
      "step": 33270
    },
    {
      "epoch": 0.031022344967217615,
      "grad_norm": 1.260461449623108,
      "learning_rate": 0.00019379572249938478,
      "loss": 4.2695,
      "step": 33280
    },
    {
      "epoch": 0.03103166658529671,
      "grad_norm": 1.2396223545074463,
      "learning_rate": 0.00019379385817728959,
      "loss": 4.1783,
      "step": 33290
    },
    {
      "epoch": 0.0310409882033758,
      "grad_norm": 1.160062313079834,
      "learning_rate": 0.00019379199385519437,
      "loss": 4.1884,
      "step": 33300
    },
    {
      "epoch": 0.031050309821454892,
      "grad_norm": 1.2693005800247192,
      "learning_rate": 0.00019379012953309917,
      "loss": 4.2387,
      "step": 33310
    },
    {
      "epoch": 0.031059631439533983,
      "grad_norm": 1.0129910707473755,
      "learning_rate": 0.000193788265211004,
      "loss": 4.1226,
      "step": 33320
    },
    {
      "epoch": 0.031068953057613077,
      "grad_norm": 1.1903023719787598,
      "learning_rate": 0.00019378640088890878,
      "loss": 4.0284,
      "step": 33330
    },
    {
      "epoch": 0.03107827467569217,
      "grad_norm": 1.1917833089828491,
      "learning_rate": 0.0001937845365668136,
      "loss": 4.2356,
      "step": 33340
    },
    {
      "epoch": 0.03108759629377126,
      "grad_norm": 1.1421825885772705,
      "learning_rate": 0.00019378267224471837,
      "loss": 4.1666,
      "step": 33350
    },
    {
      "epoch": 0.03109691791185035,
      "grad_norm": 0.9789568185806274,
      "learning_rate": 0.00019378080792262318,
      "loss": 4.0662,
      "step": 33360
    },
    {
      "epoch": 0.031106239529929445,
      "grad_norm": 3.5572075843811035,
      "learning_rate": 0.00019377894360052798,
      "loss": 4.2955,
      "step": 33370
    },
    {
      "epoch": 0.031115561148008536,
      "grad_norm": 0.9755647778511047,
      "learning_rate": 0.0001937770792784328,
      "loss": 4.0199,
      "step": 33380
    },
    {
      "epoch": 0.031124882766087627,
      "grad_norm": 1.1481549739837646,
      "learning_rate": 0.0001937752149563376,
      "loss": 4.2582,
      "step": 33390
    },
    {
      "epoch": 0.031134204384166718,
      "grad_norm": 1.186603307723999,
      "learning_rate": 0.00019377335063424237,
      "loss": 4.3751,
      "step": 33400
    },
    {
      "epoch": 0.031143526002245812,
      "grad_norm": 1.1959127187728882,
      "learning_rate": 0.00019377148631214718,
      "loss": 3.9717,
      "step": 33410
    },
    {
      "epoch": 0.031152847620324903,
      "grad_norm": 0.9886515140533447,
      "learning_rate": 0.000193769621990052,
      "loss": 4.1013,
      "step": 33420
    },
    {
      "epoch": 0.031162169238403994,
      "grad_norm": 1.0797505378723145,
      "learning_rate": 0.0001937677576679568,
      "loss": 4.2194,
      "step": 33430
    },
    {
      "epoch": 0.031171490856483085,
      "grad_norm": 1.218414545059204,
      "learning_rate": 0.0001937658933458616,
      "loss": 4.0605,
      "step": 33440
    },
    {
      "epoch": 0.031180812474562176,
      "grad_norm": 1.0939136743545532,
      "learning_rate": 0.00019376402902376638,
      "loss": 4.2712,
      "step": 33450
    },
    {
      "epoch": 0.03119013409264127,
      "grad_norm": 1.2458620071411133,
      "learning_rate": 0.00019376216470167118,
      "loss": 4.169,
      "step": 33460
    },
    {
      "epoch": 0.031199455710720362,
      "grad_norm": 1.2819652557373047,
      "learning_rate": 0.000193760300379576,
      "loss": 4.1438,
      "step": 33470
    },
    {
      "epoch": 0.031208777328799453,
      "grad_norm": 1.0431079864501953,
      "learning_rate": 0.0001937584360574808,
      "loss": 4.2421,
      "step": 33480
    },
    {
      "epoch": 0.031218098946878544,
      "grad_norm": 1.2537189722061157,
      "learning_rate": 0.0001937565717353856,
      "loss": 4.0145,
      "step": 33490
    },
    {
      "epoch": 0.03122742056495764,
      "grad_norm": 1.0009609460830688,
      "learning_rate": 0.00019375470741329038,
      "loss": 4.112,
      "step": 33500
    },
    {
      "epoch": 0.03123674218303673,
      "grad_norm": 1.1659718751907349,
      "learning_rate": 0.0001937528430911952,
      "loss": 4.1093,
      "step": 33510
    },
    {
      "epoch": 0.03124606380111582,
      "grad_norm": 1.2436418533325195,
      "learning_rate": 0.0001937509787691,
      "loss": 4.154,
      "step": 33520
    },
    {
      "epoch": 0.03125538541919491,
      "grad_norm": 1.1197046041488647,
      "learning_rate": 0.00019374911444700477,
      "loss": 4.449,
      "step": 33530
    },
    {
      "epoch": 0.031264707037274006,
      "grad_norm": 1.0459939241409302,
      "learning_rate": 0.0001937472501249096,
      "loss": 4.1517,
      "step": 33540
    },
    {
      "epoch": 0.031274028655353094,
      "grad_norm": 1.24307382106781,
      "learning_rate": 0.0001937453858028144,
      "loss": 4.0386,
      "step": 33550
    },
    {
      "epoch": 0.03128335027343219,
      "grad_norm": 1.3622782230377197,
      "learning_rate": 0.0001937435214807192,
      "loss": 4.1535,
      "step": 33560
    },
    {
      "epoch": 0.03129267189151128,
      "grad_norm": 1.2810947895050049,
      "learning_rate": 0.000193741657158624,
      "loss": 4.0404,
      "step": 33570
    },
    {
      "epoch": 0.03130199350959037,
      "grad_norm": 1.1635499000549316,
      "learning_rate": 0.00019373979283652878,
      "loss": 4.1705,
      "step": 33580
    },
    {
      "epoch": 0.031311315127669465,
      "grad_norm": 1.0975193977355957,
      "learning_rate": 0.00019373792851443359,
      "loss": 4.2779,
      "step": 33590
    },
    {
      "epoch": 0.03132063674574856,
      "grad_norm": 0.9475517272949219,
      "learning_rate": 0.0001937360641923384,
      "loss": 4.25,
      "step": 33600
    },
    {
      "epoch": 0.03132995836382765,
      "grad_norm": 1.0388847589492798,
      "learning_rate": 0.0001937341998702432,
      "loss": 4.2887,
      "step": 33610
    },
    {
      "epoch": 0.03133927998190674,
      "grad_norm": 1.0195189714431763,
      "learning_rate": 0.00019373233554814798,
      "loss": 4.1614,
      "step": 33620
    },
    {
      "epoch": 0.03134860159998583,
      "grad_norm": 1.333825945854187,
      "learning_rate": 0.00019373047122605278,
      "loss": 4.2199,
      "step": 33630
    },
    {
      "epoch": 0.03135792321806492,
      "grad_norm": 1.1169041395187378,
      "learning_rate": 0.0001937286069039576,
      "loss": 4.2686,
      "step": 33640
    },
    {
      "epoch": 0.03136724483614402,
      "grad_norm": 0.8800468444824219,
      "learning_rate": 0.0001937267425818624,
      "loss": 4.2141,
      "step": 33650
    },
    {
      "epoch": 0.031376566454223105,
      "grad_norm": 1.3360450267791748,
      "learning_rate": 0.0001937248782597672,
      "loss": 3.973,
      "step": 33660
    },
    {
      "epoch": 0.0313858880723022,
      "grad_norm": 1.2281818389892578,
      "learning_rate": 0.00019372301393767198,
      "loss": 4.2168,
      "step": 33670
    },
    {
      "epoch": 0.03139520969038129,
      "grad_norm": 0.9637786149978638,
      "learning_rate": 0.0001937211496155768,
      "loss": 4.0139,
      "step": 33680
    },
    {
      "epoch": 0.03140453130846038,
      "grad_norm": 1.2547768354415894,
      "learning_rate": 0.0001937192852934816,
      "loss": 4.3082,
      "step": 33690
    },
    {
      "epoch": 0.031413852926539476,
      "grad_norm": 1.0905693769454956,
      "learning_rate": 0.0001937174209713864,
      "loss": 3.9822,
      "step": 33700
    },
    {
      "epoch": 0.031423174544618564,
      "grad_norm": 1.0615050792694092,
      "learning_rate": 0.0001937155566492912,
      "loss": 4.1995,
      "step": 33710
    },
    {
      "epoch": 0.03143249616269766,
      "grad_norm": 1.0356389284133911,
      "learning_rate": 0.00019371369232719599,
      "loss": 4.112,
      "step": 33720
    },
    {
      "epoch": 0.03144181778077675,
      "grad_norm": 1.1233934164047241,
      "learning_rate": 0.0001937118280051008,
      "loss": 4.0047,
      "step": 33730
    },
    {
      "epoch": 0.03145113939885584,
      "grad_norm": 1.2266989946365356,
      "learning_rate": 0.0001937099636830056,
      "loss": 4.0138,
      "step": 33740
    },
    {
      "epoch": 0.031460461016934935,
      "grad_norm": 1.105018138885498,
      "learning_rate": 0.00019370809936091038,
      "loss": 4.0073,
      "step": 33750
    },
    {
      "epoch": 0.03146978263501402,
      "grad_norm": 1.2879492044448853,
      "learning_rate": 0.0001937062350388152,
      "loss": 4.1784,
      "step": 33760
    },
    {
      "epoch": 0.03147910425309312,
      "grad_norm": 1.1119554042816162,
      "learning_rate": 0.00019370437071672,
      "loss": 4.2229,
      "step": 33770
    },
    {
      "epoch": 0.03148842587117221,
      "grad_norm": 1.188773512840271,
      "learning_rate": 0.0001937025063946248,
      "loss": 4.1517,
      "step": 33780
    },
    {
      "epoch": 0.0314977474892513,
      "grad_norm": 1.378822922706604,
      "learning_rate": 0.0001937006420725296,
      "loss": 4.2138,
      "step": 33790
    },
    {
      "epoch": 0.03150706910733039,
      "grad_norm": 0.9857574701309204,
      "learning_rate": 0.00019369877775043438,
      "loss": 4.2705,
      "step": 33800
    },
    {
      "epoch": 0.03151639072540948,
      "grad_norm": 1.065895676612854,
      "learning_rate": 0.0001936969134283392,
      "loss": 4.1363,
      "step": 33810
    },
    {
      "epoch": 0.031525712343488575,
      "grad_norm": 1.147736668586731,
      "learning_rate": 0.000193695049106244,
      "loss": 4.0372,
      "step": 33820
    },
    {
      "epoch": 0.03153503396156767,
      "grad_norm": 1.1703882217407227,
      "learning_rate": 0.0001936931847841488,
      "loss": 4.0431,
      "step": 33830
    },
    {
      "epoch": 0.03154435557964676,
      "grad_norm": 1.0471097230911255,
      "learning_rate": 0.0001936913204620536,
      "loss": 4.1048,
      "step": 33840
    },
    {
      "epoch": 0.03155367719772585,
      "grad_norm": 1.156387448310852,
      "learning_rate": 0.0001936894561399584,
      "loss": 4.1485,
      "step": 33850
    },
    {
      "epoch": 0.031562998815804946,
      "grad_norm": 1.118208408355713,
      "learning_rate": 0.0001936875918178632,
      "loss": 4.1321,
      "step": 33860
    },
    {
      "epoch": 0.031572320433884034,
      "grad_norm": 1.1028974056243896,
      "learning_rate": 0.000193685727495768,
      "loss": 4.1805,
      "step": 33870
    },
    {
      "epoch": 0.03158164205196313,
      "grad_norm": 1.1796215772628784,
      "learning_rate": 0.0001936838631736728,
      "loss": 4.1766,
      "step": 33880
    },
    {
      "epoch": 0.031590963670042216,
      "grad_norm": 1.2137401103973389,
      "learning_rate": 0.0001936819988515776,
      "loss": 4.2923,
      "step": 33890
    },
    {
      "epoch": 0.03160028528812131,
      "grad_norm": 1.192809820175171,
      "learning_rate": 0.0001936801345294824,
      "loss": 4.3314,
      "step": 33900
    },
    {
      "epoch": 0.031609606906200405,
      "grad_norm": 1.0480808019638062,
      "learning_rate": 0.0001936782702073872,
      "loss": 4.2334,
      "step": 33910
    },
    {
      "epoch": 0.03161892852427949,
      "grad_norm": 1.1171379089355469,
      "learning_rate": 0.00019367640588529198,
      "loss": 4.2951,
      "step": 33920
    },
    {
      "epoch": 0.03162825014235859,
      "grad_norm": 0.9879500865936279,
      "learning_rate": 0.0001936745415631968,
      "loss": 4.1616,
      "step": 33930
    },
    {
      "epoch": 0.03163757176043768,
      "grad_norm": 1.1072136163711548,
      "learning_rate": 0.00019367267724110162,
      "loss": 4.1484,
      "step": 33940
    },
    {
      "epoch": 0.03164689337851677,
      "grad_norm": 1.1698812246322632,
      "learning_rate": 0.0001936708129190064,
      "loss": 4.2015,
      "step": 33950
    },
    {
      "epoch": 0.03165621499659586,
      "grad_norm": 1.1047348976135254,
      "learning_rate": 0.0001936689485969112,
      "loss": 4.1641,
      "step": 33960
    },
    {
      "epoch": 0.03166553661467495,
      "grad_norm": 1.042384147644043,
      "learning_rate": 0.00019366708427481598,
      "loss": 4.0916,
      "step": 33970
    },
    {
      "epoch": 0.031674858232754045,
      "grad_norm": 1.2426294088363647,
      "learning_rate": 0.0001936652199527208,
      "loss": 4.1778,
      "step": 33980
    },
    {
      "epoch": 0.03168417985083314,
      "grad_norm": 1.1677898168563843,
      "learning_rate": 0.00019366335563062562,
      "loss": 4.0052,
      "step": 33990
    },
    {
      "epoch": 0.03169350146891223,
      "grad_norm": 0.9215357303619385,
      "learning_rate": 0.0001936614913085304,
      "loss": 4.0803,
      "step": 34000
    },
    {
      "epoch": 0.03170282308699132,
      "grad_norm": 1.1562113761901855,
      "learning_rate": 0.0001936596269864352,
      "loss": 4.29,
      "step": 34010
    },
    {
      "epoch": 0.03171214470507041,
      "grad_norm": 0.9683641791343689,
      "learning_rate": 0.00019365776266433999,
      "loss": 4.1629,
      "step": 34020
    },
    {
      "epoch": 0.031721466323149504,
      "grad_norm": 1.1866391897201538,
      "learning_rate": 0.0001936558983422448,
      "loss": 4.0267,
      "step": 34030
    },
    {
      "epoch": 0.0317307879412286,
      "grad_norm": 0.9597491025924683,
      "learning_rate": 0.0001936540340201496,
      "loss": 4.2646,
      "step": 34040
    },
    {
      "epoch": 0.031740109559307686,
      "grad_norm": 1.1372307538986206,
      "learning_rate": 0.0001936521696980544,
      "loss": 4.0993,
      "step": 34050
    },
    {
      "epoch": 0.03174943117738678,
      "grad_norm": 1.2440176010131836,
      "learning_rate": 0.0001936503053759592,
      "loss": 4.2427,
      "step": 34060
    },
    {
      "epoch": 0.031758752795465875,
      "grad_norm": 1.3341916799545288,
      "learning_rate": 0.000193648441053864,
      "loss": 4.1361,
      "step": 34070
    },
    {
      "epoch": 0.03176807441354496,
      "grad_norm": 0.977319598197937,
      "learning_rate": 0.0001936465767317688,
      "loss": 3.9951,
      "step": 34080
    },
    {
      "epoch": 0.03177739603162406,
      "grad_norm": 1.121798038482666,
      "learning_rate": 0.0001936447124096736,
      "loss": 4.2099,
      "step": 34090
    },
    {
      "epoch": 0.031786717649703145,
      "grad_norm": 1.0670084953308105,
      "learning_rate": 0.0001936428480875784,
      "loss": 4.0989,
      "step": 34100
    },
    {
      "epoch": 0.03179603926778224,
      "grad_norm": 1.1414498090744019,
      "learning_rate": 0.00019364098376548322,
      "loss": 3.9789,
      "step": 34110
    },
    {
      "epoch": 0.031805360885861333,
      "grad_norm": 1.1092795133590698,
      "learning_rate": 0.000193639119443388,
      "loss": 4.2591,
      "step": 34120
    },
    {
      "epoch": 0.03181468250394042,
      "grad_norm": 1.166350245475769,
      "learning_rate": 0.0001936372551212928,
      "loss": 4.1724,
      "step": 34130
    },
    {
      "epoch": 0.031824004122019516,
      "grad_norm": 1.1382982730865479,
      "learning_rate": 0.0001936353907991976,
      "loss": 4.1849,
      "step": 34140
    },
    {
      "epoch": 0.0318333257400986,
      "grad_norm": 1.0480167865753174,
      "learning_rate": 0.00019363352647710241,
      "loss": 4.1555,
      "step": 34150
    },
    {
      "epoch": 0.0318426473581777,
      "grad_norm": 1.093045949935913,
      "learning_rate": 0.00019363166215500722,
      "loss": 4.191,
      "step": 34160
    },
    {
      "epoch": 0.03185196897625679,
      "grad_norm": 1.328068494796753,
      "learning_rate": 0.000193629797832912,
      "loss": 4.1134,
      "step": 34170
    },
    {
      "epoch": 0.03186129059433588,
      "grad_norm": 0.9579453468322754,
      "learning_rate": 0.0001936279335108168,
      "loss": 4.1389,
      "step": 34180
    },
    {
      "epoch": 0.031870612212414974,
      "grad_norm": 1.0709044933319092,
      "learning_rate": 0.0001936260691887216,
      "loss": 4.1876,
      "step": 34190
    },
    {
      "epoch": 0.03187993383049407,
      "grad_norm": 1.1078155040740967,
      "learning_rate": 0.0001936242048666264,
      "loss": 3.9269,
      "step": 34200
    },
    {
      "epoch": 0.031889255448573156,
      "grad_norm": 1.0197181701660156,
      "learning_rate": 0.00019362234054453122,
      "loss": 4.3219,
      "step": 34210
    },
    {
      "epoch": 0.03189857706665225,
      "grad_norm": 1.005435585975647,
      "learning_rate": 0.000193620476222436,
      "loss": 4.1999,
      "step": 34220
    },
    {
      "epoch": 0.03190789868473134,
      "grad_norm": 1.1313318014144897,
      "learning_rate": 0.0001936186119003408,
      "loss": 4.1142,
      "step": 34230
    },
    {
      "epoch": 0.03191722030281043,
      "grad_norm": 1.07282292842865,
      "learning_rate": 0.00019361674757824562,
      "loss": 4.226,
      "step": 34240
    },
    {
      "epoch": 0.03192654192088953,
      "grad_norm": 1.314040184020996,
      "learning_rate": 0.0001936148832561504,
      "loss": 4.1113,
      "step": 34250
    },
    {
      "epoch": 0.031935863538968615,
      "grad_norm": 1.206375002861023,
      "learning_rate": 0.0001936130189340552,
      "loss": 4.3894,
      "step": 34260
    },
    {
      "epoch": 0.03194518515704771,
      "grad_norm": 1.2092639207839966,
      "learning_rate": 0.00019361115461196,
      "loss": 4.2073,
      "step": 34270
    },
    {
      "epoch": 0.031954506775126804,
      "grad_norm": 1.1287683248519897,
      "learning_rate": 0.00019360929028986481,
      "loss": 4.1723,
      "step": 34280
    },
    {
      "epoch": 0.03196382839320589,
      "grad_norm": 1.2201905250549316,
      "learning_rate": 0.00019360742596776962,
      "loss": 4.2182,
      "step": 34290
    },
    {
      "epoch": 0.031973150011284986,
      "grad_norm": 1.0854533910751343,
      "learning_rate": 0.0001936055616456744,
      "loss": 4.3552,
      "step": 34300
    },
    {
      "epoch": 0.03198247162936407,
      "grad_norm": 1.4439020156860352,
      "learning_rate": 0.0001936036973235792,
      "loss": 4.2405,
      "step": 34310
    },
    {
      "epoch": 0.03199179324744317,
      "grad_norm": 1.174496054649353,
      "learning_rate": 0.000193601833001484,
      "loss": 4.0936,
      "step": 34320
    },
    {
      "epoch": 0.03200111486552226,
      "grad_norm": 1.1464020013809204,
      "learning_rate": 0.00019359996867938882,
      "loss": 4.3755,
      "step": 34330
    },
    {
      "epoch": 0.03201043648360135,
      "grad_norm": 1.2716156244277954,
      "learning_rate": 0.00019359810435729363,
      "loss": 4.2107,
      "step": 34340
    },
    {
      "epoch": 0.032019758101680444,
      "grad_norm": 1.0000500679016113,
      "learning_rate": 0.0001935962400351984,
      "loss": 4.1447,
      "step": 34350
    },
    {
      "epoch": 0.03202907971975953,
      "grad_norm": 1.119308590888977,
      "learning_rate": 0.0001935943757131032,
      "loss": 4.1283,
      "step": 34360
    },
    {
      "epoch": 0.032038401337838626,
      "grad_norm": 1.1436704397201538,
      "learning_rate": 0.000193592511391008,
      "loss": 4.1294,
      "step": 34370
    },
    {
      "epoch": 0.03204772295591772,
      "grad_norm": 1.094313621520996,
      "learning_rate": 0.00019359064706891282,
      "loss": 4.1394,
      "step": 34380
    },
    {
      "epoch": 0.03205704457399681,
      "grad_norm": 1.1081364154815674,
      "learning_rate": 0.00019358878274681763,
      "loss": 4.3358,
      "step": 34390
    },
    {
      "epoch": 0.0320663661920759,
      "grad_norm": 0.8813686370849609,
      "learning_rate": 0.0001935869184247224,
      "loss": 4.2945,
      "step": 34400
    },
    {
      "epoch": 0.032075687810155,
      "grad_norm": 1.0365159511566162,
      "learning_rate": 0.00019358505410262721,
      "loss": 4.3583,
      "step": 34410
    },
    {
      "epoch": 0.032085009428234085,
      "grad_norm": 1.0628511905670166,
      "learning_rate": 0.000193583189780532,
      "loss": 4.2082,
      "step": 34420
    },
    {
      "epoch": 0.03209433104631318,
      "grad_norm": 1.0665018558502197,
      "learning_rate": 0.0001935813254584368,
      "loss": 4.0638,
      "step": 34430
    },
    {
      "epoch": 0.03210365266439227,
      "grad_norm": 1.0635793209075928,
      "learning_rate": 0.00019357946113634163,
      "loss": 4.0721,
      "step": 34440
    },
    {
      "epoch": 0.03211297428247136,
      "grad_norm": 1.3135826587677002,
      "learning_rate": 0.0001935775968142464,
      "loss": 4.1888,
      "step": 34450
    },
    {
      "epoch": 0.032122295900550456,
      "grad_norm": 1.1003209352493286,
      "learning_rate": 0.00019357573249215122,
      "loss": 4.2455,
      "step": 34460
    },
    {
      "epoch": 0.03213161751862954,
      "grad_norm": 1.0198822021484375,
      "learning_rate": 0.000193573868170056,
      "loss": 4.1,
      "step": 34470
    },
    {
      "epoch": 0.03214093913670864,
      "grad_norm": 1.1254384517669678,
      "learning_rate": 0.0001935720038479608,
      "loss": 4.1043,
      "step": 34480
    },
    {
      "epoch": 0.032150260754787725,
      "grad_norm": 1.2586067914962769,
      "learning_rate": 0.0001935701395258656,
      "loss": 4.0226,
      "step": 34490
    },
    {
      "epoch": 0.03215958237286682,
      "grad_norm": 1.0939058065414429,
      "learning_rate": 0.00019356827520377042,
      "loss": 4.1051,
      "step": 34500
    },
    {
      "epoch": 0.032168903990945914,
      "grad_norm": 1.1908825635910034,
      "learning_rate": 0.00019356641088167522,
      "loss": 4.087,
      "step": 34510
    },
    {
      "epoch": 0.032178225609025,
      "grad_norm": 1.0478686094284058,
      "learning_rate": 0.00019356454655958,
      "loss": 4.1572,
      "step": 34520
    },
    {
      "epoch": 0.032187547227104096,
      "grad_norm": 1.246662974357605,
      "learning_rate": 0.0001935626822374848,
      "loss": 4.031,
      "step": 34530
    },
    {
      "epoch": 0.03219686884518319,
      "grad_norm": 0.941910982131958,
      "learning_rate": 0.00019356081791538962,
      "loss": 4.1037,
      "step": 34540
    },
    {
      "epoch": 0.03220619046326228,
      "grad_norm": 1.3006725311279297,
      "learning_rate": 0.00019355895359329442,
      "loss": 4.0896,
      "step": 34550
    },
    {
      "epoch": 0.03221551208134137,
      "grad_norm": 1.3279876708984375,
      "learning_rate": 0.00019355708927119923,
      "loss": 4.1611,
      "step": 34560
    },
    {
      "epoch": 0.03222483369942046,
      "grad_norm": 1.016194462776184,
      "learning_rate": 0.000193555224949104,
      "loss": 4.0477,
      "step": 34570
    },
    {
      "epoch": 0.032234155317499555,
      "grad_norm": 1.2807313203811646,
      "learning_rate": 0.00019355336062700881,
      "loss": 4.1777,
      "step": 34580
    },
    {
      "epoch": 0.03224347693557865,
      "grad_norm": 0.9206752777099609,
      "learning_rate": 0.00019355149630491362,
      "loss": 4.0123,
      "step": 34590
    },
    {
      "epoch": 0.03225279855365774,
      "grad_norm": 1.1686913967132568,
      "learning_rate": 0.00019354963198281843,
      "loss": 4.1328,
      "step": 34600
    },
    {
      "epoch": 0.03226212017173683,
      "grad_norm": 1.0351837873458862,
      "learning_rate": 0.00019354776766072323,
      "loss": 4.2025,
      "step": 34610
    },
    {
      "epoch": 0.032271441789815926,
      "grad_norm": 1.0376849174499512,
      "learning_rate": 0.000193545903338628,
      "loss": 4.0095,
      "step": 34620
    },
    {
      "epoch": 0.03228076340789501,
      "grad_norm": 1.0402971506118774,
      "learning_rate": 0.00019354403901653282,
      "loss": 3.9493,
      "step": 34630
    },
    {
      "epoch": 0.03229008502597411,
      "grad_norm": 1.282641053199768,
      "learning_rate": 0.00019354217469443762,
      "loss": 4.1612,
      "step": 34640
    },
    {
      "epoch": 0.032299406644053195,
      "grad_norm": 1.261616826057434,
      "learning_rate": 0.0001935403103723424,
      "loss": 3.8408,
      "step": 34650
    },
    {
      "epoch": 0.03230872826213229,
      "grad_norm": 1.05290687084198,
      "learning_rate": 0.00019353844605024724,
      "loss": 4.2956,
      "step": 34660
    },
    {
      "epoch": 0.032318049880211384,
      "grad_norm": 1.354961633682251,
      "learning_rate": 0.00019353658172815202,
      "loss": 4.1336,
      "step": 34670
    },
    {
      "epoch": 0.03232737149829047,
      "grad_norm": 1.213735580444336,
      "learning_rate": 0.00019353471740605682,
      "loss": 3.8343,
      "step": 34680
    },
    {
      "epoch": 0.032336693116369566,
      "grad_norm": 1.1585599184036255,
      "learning_rate": 0.00019353285308396163,
      "loss": 4.1851,
      "step": 34690
    },
    {
      "epoch": 0.032346014734448654,
      "grad_norm": 1.1205967664718628,
      "learning_rate": 0.0001935309887618664,
      "loss": 4.2056,
      "step": 34700
    },
    {
      "epoch": 0.03235533635252775,
      "grad_norm": 1.170662522315979,
      "learning_rate": 0.00019352912443977121,
      "loss": 4.143,
      "step": 34710
    },
    {
      "epoch": 0.03236465797060684,
      "grad_norm": 1.2325834035873413,
      "learning_rate": 0.00019352726011767602,
      "loss": 4.2747,
      "step": 34720
    },
    {
      "epoch": 0.03237397958868593,
      "grad_norm": 1.2575600147247314,
      "learning_rate": 0.00019352539579558083,
      "loss": 4.4065,
      "step": 34730
    },
    {
      "epoch": 0.032383301206765025,
      "grad_norm": 0.9777859449386597,
      "learning_rate": 0.00019352353147348563,
      "loss": 4.1668,
      "step": 34740
    },
    {
      "epoch": 0.03239262282484412,
      "grad_norm": 1.0599602460861206,
      "learning_rate": 0.0001935216671513904,
      "loss": 4.1713,
      "step": 34750
    },
    {
      "epoch": 0.03240194444292321,
      "grad_norm": 1.1970728635787964,
      "learning_rate": 0.00019351980282929522,
      "loss": 4.1174,
      "step": 34760
    },
    {
      "epoch": 0.0324112660610023,
      "grad_norm": 1.2268121242523193,
      "learning_rate": 0.00019351793850720003,
      "loss": 4.1105,
      "step": 34770
    },
    {
      "epoch": 0.03242058767908139,
      "grad_norm": 1.1106913089752197,
      "learning_rate": 0.00019351607418510483,
      "loss": 4.2435,
      "step": 34780
    },
    {
      "epoch": 0.032429909297160484,
      "grad_norm": 1.243887186050415,
      "learning_rate": 0.00019351420986300964,
      "loss": 4.2028,
      "step": 34790
    },
    {
      "epoch": 0.03243923091523958,
      "grad_norm": 1.2053412199020386,
      "learning_rate": 0.00019351234554091442,
      "loss": 4.2289,
      "step": 34800
    },
    {
      "epoch": 0.032448552533318666,
      "grad_norm": 0.9863215088844299,
      "learning_rate": 0.00019351048121881922,
      "loss": 4.2771,
      "step": 34810
    },
    {
      "epoch": 0.03245787415139776,
      "grad_norm": 1.2366856336593628,
      "learning_rate": 0.000193508616896724,
      "loss": 4.1658,
      "step": 34820
    },
    {
      "epoch": 0.03246719576947685,
      "grad_norm": 1.0578049421310425,
      "learning_rate": 0.00019350675257462884,
      "loss": 4.0176,
      "step": 34830
    },
    {
      "epoch": 0.03247651738755594,
      "grad_norm": 1.0739120244979858,
      "learning_rate": 0.00019350488825253364,
      "loss": 4.1656,
      "step": 34840
    },
    {
      "epoch": 0.03248583900563504,
      "grad_norm": 1.1196386814117432,
      "learning_rate": 0.00019350302393043842,
      "loss": 4.1187,
      "step": 34850
    },
    {
      "epoch": 0.032495160623714124,
      "grad_norm": 1.1284171342849731,
      "learning_rate": 0.00019350115960834323,
      "loss": 4.002,
      "step": 34860
    },
    {
      "epoch": 0.03250448224179322,
      "grad_norm": 1.1058006286621094,
      "learning_rate": 0.000193499295286248,
      "loss": 3.9843,
      "step": 34870
    },
    {
      "epoch": 0.03251380385987231,
      "grad_norm": 0.9528669714927673,
      "learning_rate": 0.0001934974309641528,
      "loss": 4.084,
      "step": 34880
    },
    {
      "epoch": 0.0325231254779514,
      "grad_norm": 1.1715654134750366,
      "learning_rate": 0.00019349556664205765,
      "loss": 4.1488,
      "step": 34890
    },
    {
      "epoch": 0.032532447096030495,
      "grad_norm": 1.1766422986984253,
      "learning_rate": 0.00019349370231996243,
      "loss": 4.0319,
      "step": 34900
    },
    {
      "epoch": 0.03254176871410958,
      "grad_norm": 1.188271164894104,
      "learning_rate": 0.00019349183799786723,
      "loss": 4.0641,
      "step": 34910
    },
    {
      "epoch": 0.03255109033218868,
      "grad_norm": 1.1889610290527344,
      "learning_rate": 0.000193489973675772,
      "loss": 4.1248,
      "step": 34920
    },
    {
      "epoch": 0.03256041195026777,
      "grad_norm": 1.0510648488998413,
      "learning_rate": 0.00019348810935367682,
      "loss": 4.2145,
      "step": 34930
    },
    {
      "epoch": 0.03256973356834686,
      "grad_norm": 0.9024167060852051,
      "learning_rate": 0.00019348624503158162,
      "loss": 4.1184,
      "step": 34940
    },
    {
      "epoch": 0.032579055186425954,
      "grad_norm": 1.1037555932998657,
      "learning_rate": 0.00019348438070948643,
      "loss": 4.1984,
      "step": 34950
    },
    {
      "epoch": 0.03258837680450505,
      "grad_norm": 1.0127732753753662,
      "learning_rate": 0.00019348251638739124,
      "loss": 4.0445,
      "step": 34960
    },
    {
      "epoch": 0.032597698422584136,
      "grad_norm": 1.0112773180007935,
      "learning_rate": 0.00019348065206529602,
      "loss": 4.2553,
      "step": 34970
    },
    {
      "epoch": 0.03260702004066323,
      "grad_norm": 1.1381269693374634,
      "learning_rate": 0.00019347878774320082,
      "loss": 4.4135,
      "step": 34980
    },
    {
      "epoch": 0.03261634165874232,
      "grad_norm": 1.3439854383468628,
      "learning_rate": 0.00019347692342110563,
      "loss": 4.1795,
      "step": 34990
    },
    {
      "epoch": 0.03262566327682141,
      "grad_norm": 0.8989523649215698,
      "learning_rate": 0.00019347505909901043,
      "loss": 4.0483,
      "step": 35000
    },
    {
      "epoch": 0.03263498489490051,
      "grad_norm": 1.1976523399353027,
      "learning_rate": 0.00019347319477691524,
      "loss": 4.0914,
      "step": 35010
    },
    {
      "epoch": 0.032644306512979594,
      "grad_norm": 1.1046137809753418,
      "learning_rate": 0.00019347133045482002,
      "loss": 4.0936,
      "step": 35020
    },
    {
      "epoch": 0.03265362813105869,
      "grad_norm": 0.9902476072311401,
      "learning_rate": 0.00019346946613272483,
      "loss": 4.1917,
      "step": 35030
    },
    {
      "epoch": 0.032662949749137776,
      "grad_norm": 1.221869945526123,
      "learning_rate": 0.0001934676018106296,
      "loss": 4.0899,
      "step": 35040
    },
    {
      "epoch": 0.03267227136721687,
      "grad_norm": 1.1407146453857422,
      "learning_rate": 0.00019346573748853444,
      "loss": 4.0348,
      "step": 35050
    },
    {
      "epoch": 0.032681592985295965,
      "grad_norm": 1.2794212102890015,
      "learning_rate": 0.00019346387316643925,
      "loss": 4.1526,
      "step": 35060
    },
    {
      "epoch": 0.03269091460337505,
      "grad_norm": 1.4210104942321777,
      "learning_rate": 0.00019346200884434402,
      "loss": 4.2075,
      "step": 35070
    },
    {
      "epoch": 0.03270023622145415,
      "grad_norm": 1.2093636989593506,
      "learning_rate": 0.00019346014452224883,
      "loss": 4.1388,
      "step": 35080
    },
    {
      "epoch": 0.03270955783953324,
      "grad_norm": 0.8809083104133606,
      "learning_rate": 0.0001934582802001536,
      "loss": 4.1955,
      "step": 35090
    },
    {
      "epoch": 0.03271887945761233,
      "grad_norm": 1.1365787982940674,
      "learning_rate": 0.00019345641587805842,
      "loss": 4.1927,
      "step": 35100
    },
    {
      "epoch": 0.032728201075691424,
      "grad_norm": 0.9295052289962769,
      "learning_rate": 0.00019345455155596325,
      "loss": 3.9926,
      "step": 35110
    },
    {
      "epoch": 0.03273752269377051,
      "grad_norm": 0.7925900816917419,
      "learning_rate": 0.00019345268723386803,
      "loss": 4.1404,
      "step": 35120
    },
    {
      "epoch": 0.032746844311849606,
      "grad_norm": 1.1571484804153442,
      "learning_rate": 0.00019345082291177284,
      "loss": 4.3182,
      "step": 35130
    },
    {
      "epoch": 0.0327561659299287,
      "grad_norm": 1.2407323122024536,
      "learning_rate": 0.00019344895858967761,
      "loss": 4.1836,
      "step": 35140
    },
    {
      "epoch": 0.03276548754800779,
      "grad_norm": 1.205796718597412,
      "learning_rate": 0.00019344709426758242,
      "loss": 4.2361,
      "step": 35150
    },
    {
      "epoch": 0.03277480916608688,
      "grad_norm": 1.0250641107559204,
      "learning_rate": 0.00019344522994548723,
      "loss": 4.181,
      "step": 35160
    },
    {
      "epoch": 0.03278413078416597,
      "grad_norm": 1.1674209833145142,
      "learning_rate": 0.00019344336562339203,
      "loss": 4.0172,
      "step": 35170
    },
    {
      "epoch": 0.032793452402245064,
      "grad_norm": 1.3377164602279663,
      "learning_rate": 0.00019344150130129684,
      "loss": 4.1992,
      "step": 35180
    },
    {
      "epoch": 0.03280277402032416,
      "grad_norm": 0.9373706579208374,
      "learning_rate": 0.00019343963697920162,
      "loss": 4.1556,
      "step": 35190
    },
    {
      "epoch": 0.032812095638403246,
      "grad_norm": 1.1434433460235596,
      "learning_rate": 0.00019343777265710643,
      "loss": 4.1253,
      "step": 35200
    },
    {
      "epoch": 0.03282141725648234,
      "grad_norm": 0.880403995513916,
      "learning_rate": 0.00019343590833501123,
      "loss": 4.2065,
      "step": 35210
    },
    {
      "epoch": 0.032830738874561435,
      "grad_norm": 1.2889735698699951,
      "learning_rate": 0.00019343404401291604,
      "loss": 4.193,
      "step": 35220
    },
    {
      "epoch": 0.03284006049264052,
      "grad_norm": 1.1926766633987427,
      "learning_rate": 0.00019343217969082084,
      "loss": 4.0727,
      "step": 35230
    },
    {
      "epoch": 0.03284938211071962,
      "grad_norm": 1.244441032409668,
      "learning_rate": 0.00019343031536872562,
      "loss": 4.2453,
      "step": 35240
    },
    {
      "epoch": 0.032858703728798705,
      "grad_norm": 1.071093201637268,
      "learning_rate": 0.00019342845104663043,
      "loss": 4.1868,
      "step": 35250
    },
    {
      "epoch": 0.0328680253468778,
      "grad_norm": 1.2601640224456787,
      "learning_rate": 0.00019342658672453524,
      "loss": 4.0965,
      "step": 35260
    },
    {
      "epoch": 0.032877346964956894,
      "grad_norm": 1.164330244064331,
      "learning_rate": 0.00019342472240244002,
      "loss": 4.0125,
      "step": 35270
    },
    {
      "epoch": 0.03288666858303598,
      "grad_norm": 0.9464453458786011,
      "learning_rate": 0.00019342285808034485,
      "loss": 4.0534,
      "step": 35280
    },
    {
      "epoch": 0.032895990201115076,
      "grad_norm": 1.0763120651245117,
      "learning_rate": 0.00019342099375824963,
      "loss": 4.0594,
      "step": 35290
    },
    {
      "epoch": 0.03290531181919417,
      "grad_norm": 1.069258451461792,
      "learning_rate": 0.00019341912943615443,
      "loss": 4.2065,
      "step": 35300
    },
    {
      "epoch": 0.03291463343727326,
      "grad_norm": 1.1749544143676758,
      "learning_rate": 0.00019341726511405924,
      "loss": 4.2275,
      "step": 35310
    },
    {
      "epoch": 0.03292395505535235,
      "grad_norm": 1.1076024770736694,
      "learning_rate": 0.00019341540079196402,
      "loss": 4.0767,
      "step": 35320
    },
    {
      "epoch": 0.03293327667343144,
      "grad_norm": 1.2721970081329346,
      "learning_rate": 0.00019341353646986883,
      "loss": 4.1992,
      "step": 35330
    },
    {
      "epoch": 0.032942598291510534,
      "grad_norm": 0.9249944090843201,
      "learning_rate": 0.00019341167214777363,
      "loss": 4.154,
      "step": 35340
    },
    {
      "epoch": 0.03295191990958963,
      "grad_norm": 1.0928775072097778,
      "learning_rate": 0.00019340980782567844,
      "loss": 4.2815,
      "step": 35350
    },
    {
      "epoch": 0.032961241527668717,
      "grad_norm": 1.0345666408538818,
      "learning_rate": 0.00019340794350358325,
      "loss": 4.1351,
      "step": 35360
    },
    {
      "epoch": 0.03297056314574781,
      "grad_norm": 1.1222738027572632,
      "learning_rate": 0.00019340607918148802,
      "loss": 4.2198,
      "step": 35370
    },
    {
      "epoch": 0.0329798847638269,
      "grad_norm": 0.9973626136779785,
      "learning_rate": 0.00019340421485939283,
      "loss": 4.1948,
      "step": 35380
    },
    {
      "epoch": 0.03298920638190599,
      "grad_norm": 1.058982014656067,
      "learning_rate": 0.00019340235053729764,
      "loss": 4.0064,
      "step": 35390
    },
    {
      "epoch": 0.03299852799998509,
      "grad_norm": 1.220003366470337,
      "learning_rate": 0.00019340048621520244,
      "loss": 4.2706,
      "step": 35400
    },
    {
      "epoch": 0.033007849618064175,
      "grad_norm": 1.2543874979019165,
      "learning_rate": 0.00019339862189310725,
      "loss": 4.0604,
      "step": 35410
    },
    {
      "epoch": 0.03301717123614327,
      "grad_norm": 1.114372730255127,
      "learning_rate": 0.00019339675757101203,
      "loss": 4.3662,
      "step": 35420
    },
    {
      "epoch": 0.033026492854222364,
      "grad_norm": 1.0303388833999634,
      "learning_rate": 0.00019339489324891684,
      "loss": 4.1948,
      "step": 35430
    },
    {
      "epoch": 0.03303581447230145,
      "grad_norm": 1.183666706085205,
      "learning_rate": 0.00019339302892682164,
      "loss": 4.1968,
      "step": 35440
    },
    {
      "epoch": 0.033045136090380546,
      "grad_norm": 1.1199066638946533,
      "learning_rate": 0.00019339116460472645,
      "loss": 4.2229,
      "step": 35450
    },
    {
      "epoch": 0.033054457708459634,
      "grad_norm": 1.0531734228134155,
      "learning_rate": 0.00019338930028263125,
      "loss": 4.2943,
      "step": 35460
    },
    {
      "epoch": 0.03306377932653873,
      "grad_norm": 0.9666541218757629,
      "learning_rate": 0.00019338743596053603,
      "loss": 4.2207,
      "step": 35470
    },
    {
      "epoch": 0.03307310094461782,
      "grad_norm": 1.01527738571167,
      "learning_rate": 0.00019338557163844084,
      "loss": 4.1003,
      "step": 35480
    },
    {
      "epoch": 0.03308242256269691,
      "grad_norm": 1.3944461345672607,
      "learning_rate": 0.00019338370731634562,
      "loss": 4.1816,
      "step": 35490
    },
    {
      "epoch": 0.033091744180776005,
      "grad_norm": 1.1572787761688232,
      "learning_rate": 0.00019338184299425045,
      "loss": 4.0305,
      "step": 35500
    },
    {
      "epoch": 0.03310106579885509,
      "grad_norm": 1.0991535186767578,
      "learning_rate": 0.00019337997867215526,
      "loss": 4.1398,
      "step": 35510
    },
    {
      "epoch": 0.03311038741693419,
      "grad_norm": 1.2000528573989868,
      "learning_rate": 0.00019337811435006004,
      "loss": 4.249,
      "step": 35520
    },
    {
      "epoch": 0.03311970903501328,
      "grad_norm": 1.157667636871338,
      "learning_rate": 0.00019337625002796484,
      "loss": 4.092,
      "step": 35530
    },
    {
      "epoch": 0.03312903065309237,
      "grad_norm": 0.9756285548210144,
      "learning_rate": 0.00019337438570586962,
      "loss": 4.2246,
      "step": 35540
    },
    {
      "epoch": 0.03313835227117146,
      "grad_norm": 1.1947349309921265,
      "learning_rate": 0.00019337252138377443,
      "loss": 4.0755,
      "step": 35550
    },
    {
      "epoch": 0.03314767388925056,
      "grad_norm": 1.157059669494629,
      "learning_rate": 0.00019337065706167926,
      "loss": 4.2217,
      "step": 35560
    },
    {
      "epoch": 0.033156995507329645,
      "grad_norm": 1.0332963466644287,
      "learning_rate": 0.00019336879273958404,
      "loss": 4.1951,
      "step": 35570
    },
    {
      "epoch": 0.03316631712540874,
      "grad_norm": 1.0227380990982056,
      "learning_rate": 0.00019336692841748885,
      "loss": 4.1868,
      "step": 35580
    },
    {
      "epoch": 0.03317563874348783,
      "grad_norm": 1.0729788541793823,
      "learning_rate": 0.00019336506409539363,
      "loss": 4.2249,
      "step": 35590
    },
    {
      "epoch": 0.03318496036156692,
      "grad_norm": 1.3036978244781494,
      "learning_rate": 0.00019336319977329843,
      "loss": 4.0979,
      "step": 35600
    },
    {
      "epoch": 0.033194281979646016,
      "grad_norm": 1.3659749031066895,
      "learning_rate": 0.00019336133545120324,
      "loss": 4.0943,
      "step": 35610
    },
    {
      "epoch": 0.033203603597725104,
      "grad_norm": 1.038750410079956,
      "learning_rate": 0.00019335947112910805,
      "loss": 4.1759,
      "step": 35620
    },
    {
      "epoch": 0.0332129252158042,
      "grad_norm": 1.1159332990646362,
      "learning_rate": 0.00019335760680701285,
      "loss": 4.1254,
      "step": 35630
    },
    {
      "epoch": 0.03322224683388329,
      "grad_norm": 1.013224482536316,
      "learning_rate": 0.00019335574248491763,
      "loss": 4.3145,
      "step": 35640
    },
    {
      "epoch": 0.03323156845196238,
      "grad_norm": 1.095300555229187,
      "learning_rate": 0.00019335387816282244,
      "loss": 4.0493,
      "step": 35650
    },
    {
      "epoch": 0.033240890070041475,
      "grad_norm": 1.1485720872879028,
      "learning_rate": 0.00019335201384072724,
      "loss": 4.0309,
      "step": 35660
    },
    {
      "epoch": 0.03325021168812056,
      "grad_norm": 1.0717219114303589,
      "learning_rate": 0.00019335014951863205,
      "loss": 3.9224,
      "step": 35670
    },
    {
      "epoch": 0.03325953330619966,
      "grad_norm": 1.1713707447052002,
      "learning_rate": 0.00019334828519653686,
      "loss": 4.1161,
      "step": 35680
    },
    {
      "epoch": 0.03326885492427875,
      "grad_norm": 0.8868288993835449,
      "learning_rate": 0.00019334642087444164,
      "loss": 4.1418,
      "step": 35690
    },
    {
      "epoch": 0.03327817654235784,
      "grad_norm": 1.1926383972167969,
      "learning_rate": 0.00019334455655234644,
      "loss": 4.3063,
      "step": 35700
    },
    {
      "epoch": 0.03328749816043693,
      "grad_norm": 1.1011382341384888,
      "learning_rate": 0.00019334269223025125,
      "loss": 4.3623,
      "step": 35710
    },
    {
      "epoch": 0.03329681977851602,
      "grad_norm": 1.2142891883850098,
      "learning_rate": 0.00019334082790815603,
      "loss": 4.2422,
      "step": 35720
    },
    {
      "epoch": 0.033306141396595115,
      "grad_norm": 1.3058940172195435,
      "learning_rate": 0.00019333896358606086,
      "loss": 4.1906,
      "step": 35730
    },
    {
      "epoch": 0.03331546301467421,
      "grad_norm": 1.121396780014038,
      "learning_rate": 0.00019333709926396564,
      "loss": 3.9651,
      "step": 35740
    },
    {
      "epoch": 0.0333247846327533,
      "grad_norm": 0.9490476846694946,
      "learning_rate": 0.00019333523494187045,
      "loss": 4.0524,
      "step": 35750
    },
    {
      "epoch": 0.03333410625083239,
      "grad_norm": 1.2674659490585327,
      "learning_rate": 0.00019333337061977525,
      "loss": 4.091,
      "step": 35760
    },
    {
      "epoch": 0.033343427868911486,
      "grad_norm": 1.0735528469085693,
      "learning_rate": 0.00019333150629768003,
      "loss": 4.2437,
      "step": 35770
    },
    {
      "epoch": 0.033352749486990574,
      "grad_norm": 1.2301735877990723,
      "learning_rate": 0.00019332964197558484,
      "loss": 4.134,
      "step": 35780
    },
    {
      "epoch": 0.03336207110506967,
      "grad_norm": 1.1450536251068115,
      "learning_rate": 0.00019332777765348965,
      "loss": 4.1119,
      "step": 35790
    },
    {
      "epoch": 0.033371392723148756,
      "grad_norm": 1.0191855430603027,
      "learning_rate": 0.00019332591333139445,
      "loss": 4.1218,
      "step": 35800
    },
    {
      "epoch": 0.03338071434122785,
      "grad_norm": 0.9961978197097778,
      "learning_rate": 0.00019332404900929926,
      "loss": 4.2778,
      "step": 35810
    },
    {
      "epoch": 0.033390035959306945,
      "grad_norm": 1.1946581602096558,
      "learning_rate": 0.00019332218468720404,
      "loss": 4.0143,
      "step": 35820
    },
    {
      "epoch": 0.03339935757738603,
      "grad_norm": 1.2114330530166626,
      "learning_rate": 0.00019332032036510884,
      "loss": 4.3724,
      "step": 35830
    },
    {
      "epoch": 0.03340867919546513,
      "grad_norm": 1.101008653640747,
      "learning_rate": 0.00019331845604301365,
      "loss": 4.2429,
      "step": 35840
    },
    {
      "epoch": 0.033418000813544214,
      "grad_norm": 1.170325517654419,
      "learning_rate": 0.00019331659172091846,
      "loss": 4.3067,
      "step": 35850
    },
    {
      "epoch": 0.03342732243162331,
      "grad_norm": 1.0897245407104492,
      "learning_rate": 0.00019331472739882326,
      "loss": 4.3481,
      "step": 35860
    },
    {
      "epoch": 0.0334366440497024,
      "grad_norm": 1.0895476341247559,
      "learning_rate": 0.00019331286307672804,
      "loss": 4.1004,
      "step": 35870
    },
    {
      "epoch": 0.03344596566778149,
      "grad_norm": 1.0363632440567017,
      "learning_rate": 0.00019331099875463285,
      "loss": 4.1222,
      "step": 35880
    },
    {
      "epoch": 0.033455287285860585,
      "grad_norm": 1.1857324838638306,
      "learning_rate": 0.00019330913443253765,
      "loss": 4.1489,
      "step": 35890
    },
    {
      "epoch": 0.03346460890393968,
      "grad_norm": 1.1100422143936157,
      "learning_rate": 0.00019330727011044246,
      "loss": 4.2295,
      "step": 35900
    },
    {
      "epoch": 0.03347393052201877,
      "grad_norm": 1.1026506423950195,
      "learning_rate": 0.00019330540578834727,
      "loss": 4.3637,
      "step": 35910
    },
    {
      "epoch": 0.03348325214009786,
      "grad_norm": 1.2043068408966064,
      "learning_rate": 0.00019330354146625205,
      "loss": 4.0626,
      "step": 35920
    },
    {
      "epoch": 0.03349257375817695,
      "grad_norm": 1.3267329931259155,
      "learning_rate": 0.00019330167714415685,
      "loss": 4.2194,
      "step": 35930
    },
    {
      "epoch": 0.033501895376256044,
      "grad_norm": 1.0405914783477783,
      "learning_rate": 0.00019329981282206163,
      "loss": 3.9382,
      "step": 35940
    },
    {
      "epoch": 0.03351121699433514,
      "grad_norm": 0.878574550151825,
      "learning_rate": 0.00019329794849996646,
      "loss": 4.2242,
      "step": 35950
    },
    {
      "epoch": 0.033520538612414226,
      "grad_norm": 1.002815842628479,
      "learning_rate": 0.00019329608417787127,
      "loss": 4.214,
      "step": 35960
    },
    {
      "epoch": 0.03352986023049332,
      "grad_norm": 1.116479754447937,
      "learning_rate": 0.00019329421985577605,
      "loss": 4.3143,
      "step": 35970
    },
    {
      "epoch": 0.033539181848572415,
      "grad_norm": 1.090591311454773,
      "learning_rate": 0.00019329235553368086,
      "loss": 4.1678,
      "step": 35980
    },
    {
      "epoch": 0.0335485034666515,
      "grad_norm": 1.0907515287399292,
      "learning_rate": 0.00019329049121158564,
      "loss": 4.1257,
      "step": 35990
    },
    {
      "epoch": 0.0335578250847306,
      "grad_norm": 1.1108161211013794,
      "learning_rate": 0.00019328862688949044,
      "loss": 4.0422,
      "step": 36000
    },
    {
      "epoch": 0.033567146702809685,
      "grad_norm": 1.3157808780670166,
      "learning_rate": 0.00019328676256739528,
      "loss": 3.9557,
      "step": 36010
    },
    {
      "epoch": 0.03357646832088878,
      "grad_norm": 1.1594616174697876,
      "learning_rate": 0.00019328489824530005,
      "loss": 3.8769,
      "step": 36020
    },
    {
      "epoch": 0.033585789938967874,
      "grad_norm": 1.177624225616455,
      "learning_rate": 0.00019328303392320486,
      "loss": 4.1424,
      "step": 36030
    },
    {
      "epoch": 0.03359511155704696,
      "grad_norm": 1.3645588159561157,
      "learning_rate": 0.00019328116960110964,
      "loss": 4.1483,
      "step": 36040
    },
    {
      "epoch": 0.033604433175126056,
      "grad_norm": 1.1082522869110107,
      "learning_rate": 0.00019327930527901445,
      "loss": 4.224,
      "step": 36050
    },
    {
      "epoch": 0.03361375479320514,
      "grad_norm": 1.0235573053359985,
      "learning_rate": 0.00019327744095691925,
      "loss": 3.8922,
      "step": 36060
    },
    {
      "epoch": 0.03362307641128424,
      "grad_norm": 1.2290693521499634,
      "learning_rate": 0.00019327557663482406,
      "loss": 3.988,
      "step": 36070
    },
    {
      "epoch": 0.03363239802936333,
      "grad_norm": 1.208329200744629,
      "learning_rate": 0.00019327371231272887,
      "loss": 4.0426,
      "step": 36080
    },
    {
      "epoch": 0.03364171964744242,
      "grad_norm": 1.0761380195617676,
      "learning_rate": 0.00019327184799063364,
      "loss": 4.1311,
      "step": 36090
    },
    {
      "epoch": 0.033651041265521514,
      "grad_norm": 1.298664927482605,
      "learning_rate": 0.00019326998366853845,
      "loss": 4.0605,
      "step": 36100
    },
    {
      "epoch": 0.03366036288360061,
      "grad_norm": 1.211341142654419,
      "learning_rate": 0.00019326811934644326,
      "loss": 4.2305,
      "step": 36110
    },
    {
      "epoch": 0.033669684501679696,
      "grad_norm": 1.0785350799560547,
      "learning_rate": 0.00019326625502434806,
      "loss": 4.034,
      "step": 36120
    },
    {
      "epoch": 0.03367900611975879,
      "grad_norm": 1.0437207221984863,
      "learning_rate": 0.00019326439070225287,
      "loss": 4.129,
      "step": 36130
    },
    {
      "epoch": 0.03368832773783788,
      "grad_norm": 1.1809451580047607,
      "learning_rate": 0.00019326252638015765,
      "loss": 4.1104,
      "step": 36140
    },
    {
      "epoch": 0.03369764935591697,
      "grad_norm": 1.0438284873962402,
      "learning_rate": 0.00019326066205806246,
      "loss": 4.1354,
      "step": 36150
    },
    {
      "epoch": 0.03370697097399607,
      "grad_norm": 1.21815824508667,
      "learning_rate": 0.00019325879773596726,
      "loss": 3.8669,
      "step": 36160
    },
    {
      "epoch": 0.033716292592075155,
      "grad_norm": 1.0735780000686646,
      "learning_rate": 0.00019325693341387204,
      "loss": 4.199,
      "step": 36170
    },
    {
      "epoch": 0.03372561421015425,
      "grad_norm": 1.1476866006851196,
      "learning_rate": 0.00019325506909177687,
      "loss": 4.0973,
      "step": 36180
    },
    {
      "epoch": 0.03373493582823334,
      "grad_norm": 1.2334471940994263,
      "learning_rate": 0.00019325320476968165,
      "loss": 4.1354,
      "step": 36190
    },
    {
      "epoch": 0.03374425744631243,
      "grad_norm": 0.9580932259559631,
      "learning_rate": 0.00019325134044758646,
      "loss": 4.1461,
      "step": 36200
    },
    {
      "epoch": 0.033753579064391526,
      "grad_norm": 0.9842838644981384,
      "learning_rate": 0.00019324947612549127,
      "loss": 3.9805,
      "step": 36210
    },
    {
      "epoch": 0.03376290068247061,
      "grad_norm": 1.1178094148635864,
      "learning_rate": 0.00019324761180339605,
      "loss": 4.0973,
      "step": 36220
    },
    {
      "epoch": 0.03377222230054971,
      "grad_norm": 1.1180914640426636,
      "learning_rate": 0.00019324574748130085,
      "loss": 4.0985,
      "step": 36230
    },
    {
      "epoch": 0.0337815439186288,
      "grad_norm": 1.1421407461166382,
      "learning_rate": 0.00019324388315920566,
      "loss": 4.0747,
      "step": 36240
    },
    {
      "epoch": 0.03379086553670789,
      "grad_norm": 1.091692328453064,
      "learning_rate": 0.00019324201883711046,
      "loss": 4.1335,
      "step": 36250
    },
    {
      "epoch": 0.033800187154786984,
      "grad_norm": 1.1831227540969849,
      "learning_rate": 0.00019324015451501527,
      "loss": 4.1504,
      "step": 36260
    },
    {
      "epoch": 0.03380950877286607,
      "grad_norm": 1.0806074142456055,
      "learning_rate": 0.00019323829019292005,
      "loss": 4.2368,
      "step": 36270
    },
    {
      "epoch": 0.033818830390945166,
      "grad_norm": 1.251516580581665,
      "learning_rate": 0.00019323642587082486,
      "loss": 4.1092,
      "step": 36280
    },
    {
      "epoch": 0.03382815200902426,
      "grad_norm": 1.065961241722107,
      "learning_rate": 0.00019323456154872966,
      "loss": 4.0867,
      "step": 36290
    },
    {
      "epoch": 0.03383747362710335,
      "grad_norm": 1.1821593046188354,
      "learning_rate": 0.00019323269722663447,
      "loss": 4.1963,
      "step": 36300
    },
    {
      "epoch": 0.03384679524518244,
      "grad_norm": 0.923466145992279,
      "learning_rate": 0.00019323083290453928,
      "loss": 3.9854,
      "step": 36310
    },
    {
      "epoch": 0.03385611686326154,
      "grad_norm": 0.9942947030067444,
      "learning_rate": 0.00019322896858244405,
      "loss": 4.1004,
      "step": 36320
    },
    {
      "epoch": 0.033865438481340625,
      "grad_norm": 1.0395580530166626,
      "learning_rate": 0.00019322710426034886,
      "loss": 4.1883,
      "step": 36330
    },
    {
      "epoch": 0.03387476009941972,
      "grad_norm": 1.0640079975128174,
      "learning_rate": 0.00019322523993825367,
      "loss": 4.0454,
      "step": 36340
    },
    {
      "epoch": 0.03388408171749881,
      "grad_norm": 1.1614371538162231,
      "learning_rate": 0.00019322337561615847,
      "loss": 4.2965,
      "step": 36350
    },
    {
      "epoch": 0.0338934033355779,
      "grad_norm": 0.9941519498825073,
      "learning_rate": 0.00019322151129406325,
      "loss": 4.1963,
      "step": 36360
    },
    {
      "epoch": 0.033902724953656996,
      "grad_norm": 1.139347791671753,
      "learning_rate": 0.00019321964697196806,
      "loss": 4.0008,
      "step": 36370
    },
    {
      "epoch": 0.03391204657173608,
      "grad_norm": 1.0413343906402588,
      "learning_rate": 0.00019321778264987287,
      "loss": 4.0181,
      "step": 36380
    },
    {
      "epoch": 0.03392136818981518,
      "grad_norm": 1.1219255924224854,
      "learning_rate": 0.00019321591832777764,
      "loss": 4.1357,
      "step": 36390
    },
    {
      "epoch": 0.033930689807894265,
      "grad_norm": 1.1728488206863403,
      "learning_rate": 0.00019321405400568248,
      "loss": 4.1326,
      "step": 36400
    },
    {
      "epoch": 0.03394001142597336,
      "grad_norm": 1.2744601964950562,
      "learning_rate": 0.00019321218968358726,
      "loss": 4.2381,
      "step": 36410
    },
    {
      "epoch": 0.033949333044052454,
      "grad_norm": 1.2352688312530518,
      "learning_rate": 0.00019321032536149206,
      "loss": 3.8836,
      "step": 36420
    },
    {
      "epoch": 0.03395865466213154,
      "grad_norm": 1.2142647504806519,
      "learning_rate": 0.00019320846103939687,
      "loss": 4.2285,
      "step": 36430
    },
    {
      "epoch": 0.033967976280210636,
      "grad_norm": 1.18814218044281,
      "learning_rate": 0.00019320659671730165,
      "loss": 4.0911,
      "step": 36440
    },
    {
      "epoch": 0.03397729789828973,
      "grad_norm": 1.0062525272369385,
      "learning_rate": 0.00019320473239520646,
      "loss": 4.4052,
      "step": 36450
    },
    {
      "epoch": 0.03398661951636882,
      "grad_norm": 1.0305320024490356,
      "learning_rate": 0.00019320286807311126,
      "loss": 4.1679,
      "step": 36460
    },
    {
      "epoch": 0.03399594113444791,
      "grad_norm": 1.2470200061798096,
      "learning_rate": 0.00019320100375101607,
      "loss": 4.2524,
      "step": 36470
    },
    {
      "epoch": 0.034005262752527,
      "grad_norm": 1.1213641166687012,
      "learning_rate": 0.00019319913942892087,
      "loss": 3.8196,
      "step": 36480
    },
    {
      "epoch": 0.034014584370606095,
      "grad_norm": 1.0253853797912598,
      "learning_rate": 0.00019319727510682565,
      "loss": 4.1343,
      "step": 36490
    },
    {
      "epoch": 0.03402390598868519,
      "grad_norm": 1.1629722118377686,
      "learning_rate": 0.00019319541078473046,
      "loss": 4.0814,
      "step": 36500
    },
    {
      "epoch": 0.03403322760676428,
      "grad_norm": 1.1139084100723267,
      "learning_rate": 0.00019319354646263527,
      "loss": 4.2745,
      "step": 36510
    },
    {
      "epoch": 0.03404254922484337,
      "grad_norm": 1.1862784624099731,
      "learning_rate": 0.00019319168214054007,
      "loss": 3.9963,
      "step": 36520
    },
    {
      "epoch": 0.03405187084292246,
      "grad_norm": 1.1578288078308105,
      "learning_rate": 0.00019318981781844488,
      "loss": 4.3136,
      "step": 36530
    },
    {
      "epoch": 0.03406119246100155,
      "grad_norm": 1.0003547668457031,
      "learning_rate": 0.00019318795349634966,
      "loss": 4.3381,
      "step": 36540
    },
    {
      "epoch": 0.03407051407908065,
      "grad_norm": 1.4256365299224854,
      "learning_rate": 0.00019318608917425446,
      "loss": 4.1882,
      "step": 36550
    },
    {
      "epoch": 0.034079835697159735,
      "grad_norm": 1.1492664813995361,
      "learning_rate": 0.00019318422485215927,
      "loss": 4.112,
      "step": 36560
    },
    {
      "epoch": 0.03408915731523883,
      "grad_norm": 0.9843607544898987,
      "learning_rate": 0.00019318236053006408,
      "loss": 4.0715,
      "step": 36570
    },
    {
      "epoch": 0.034098478933317924,
      "grad_norm": 1.179978847503662,
      "learning_rate": 0.00019318049620796888,
      "loss": 4.1024,
      "step": 36580
    },
    {
      "epoch": 0.03410780055139701,
      "grad_norm": 1.2091585397720337,
      "learning_rate": 0.00019317863188587366,
      "loss": 4.0311,
      "step": 36590
    },
    {
      "epoch": 0.034117122169476106,
      "grad_norm": 0.9912426471710205,
      "learning_rate": 0.00019317676756377847,
      "loss": 4.1866,
      "step": 36600
    },
    {
      "epoch": 0.034126443787555194,
      "grad_norm": 1.072439432144165,
      "learning_rate": 0.00019317490324168325,
      "loss": 4.3016,
      "step": 36610
    },
    {
      "epoch": 0.03413576540563429,
      "grad_norm": 0.952419102191925,
      "learning_rate": 0.00019317303891958805,
      "loss": 4.2239,
      "step": 36620
    },
    {
      "epoch": 0.03414508702371338,
      "grad_norm": 1.3493826389312744,
      "learning_rate": 0.0001931711745974929,
      "loss": 4.056,
      "step": 36630
    },
    {
      "epoch": 0.03415440864179247,
      "grad_norm": 1.2300864458084106,
      "learning_rate": 0.00019316931027539767,
      "loss": 4.2532,
      "step": 36640
    },
    {
      "epoch": 0.034163730259871565,
      "grad_norm": 1.0704797506332397,
      "learning_rate": 0.00019316744595330247,
      "loss": 4.1837,
      "step": 36650
    },
    {
      "epoch": 0.03417305187795066,
      "grad_norm": 1.2996450662612915,
      "learning_rate": 0.00019316558163120725,
      "loss": 4.0992,
      "step": 36660
    },
    {
      "epoch": 0.03418237349602975,
      "grad_norm": 0.9909504055976868,
      "learning_rate": 0.00019316371730911206,
      "loss": 4.1344,
      "step": 36670
    },
    {
      "epoch": 0.03419169511410884,
      "grad_norm": 1.1295089721679688,
      "learning_rate": 0.00019316185298701686,
      "loss": 4.1536,
      "step": 36680
    },
    {
      "epoch": 0.03420101673218793,
      "grad_norm": 1.1923083066940308,
      "learning_rate": 0.00019315998866492167,
      "loss": 4.1577,
      "step": 36690
    },
    {
      "epoch": 0.034210338350267024,
      "grad_norm": 1.1493161916732788,
      "learning_rate": 0.00019315812434282648,
      "loss": 4.1314,
      "step": 36700
    },
    {
      "epoch": 0.03421965996834612,
      "grad_norm": 1.0961583852767944,
      "learning_rate": 0.00019315626002073126,
      "loss": 4.097,
      "step": 36710
    },
    {
      "epoch": 0.034228981586425206,
      "grad_norm": 1.0564899444580078,
      "learning_rate": 0.00019315439569863606,
      "loss": 4.1622,
      "step": 36720
    },
    {
      "epoch": 0.0342383032045043,
      "grad_norm": 1.3472845554351807,
      "learning_rate": 0.00019315253137654087,
      "loss": 3.9323,
      "step": 36730
    },
    {
      "epoch": 0.03424762482258339,
      "grad_norm": 1.0874691009521484,
      "learning_rate": 0.00019315066705444568,
      "loss": 4.3297,
      "step": 36740
    },
    {
      "epoch": 0.03425694644066248,
      "grad_norm": 1.0791122913360596,
      "learning_rate": 0.00019314880273235048,
      "loss": 4.0857,
      "step": 36750
    },
    {
      "epoch": 0.03426626805874158,
      "grad_norm": 1.2574381828308105,
      "learning_rate": 0.00019314693841025526,
      "loss": 3.9959,
      "step": 36760
    },
    {
      "epoch": 0.034275589676820664,
      "grad_norm": 1.1552085876464844,
      "learning_rate": 0.00019314507408816007,
      "loss": 4.1156,
      "step": 36770
    },
    {
      "epoch": 0.03428491129489976,
      "grad_norm": 1.236830234527588,
      "learning_rate": 0.00019314320976606487,
      "loss": 4.1353,
      "step": 36780
    },
    {
      "epoch": 0.03429423291297885,
      "grad_norm": 1.315030574798584,
      "learning_rate": 0.00019314134544396968,
      "loss": 4.0036,
      "step": 36790
    },
    {
      "epoch": 0.03430355453105794,
      "grad_norm": 1.187694787979126,
      "learning_rate": 0.00019313948112187449,
      "loss": 4.2671,
      "step": 36800
    },
    {
      "epoch": 0.034312876149137035,
      "grad_norm": 1.09368896484375,
      "learning_rate": 0.00019313761679977927,
      "loss": 4.2327,
      "step": 36810
    },
    {
      "epoch": 0.03432219776721612,
      "grad_norm": 1.1997572183609009,
      "learning_rate": 0.00019313575247768407,
      "loss": 4.1669,
      "step": 36820
    },
    {
      "epoch": 0.03433151938529522,
      "grad_norm": 1.0553077459335327,
      "learning_rate": 0.00019313388815558888,
      "loss": 4.1545,
      "step": 36830
    },
    {
      "epoch": 0.03434084100337431,
      "grad_norm": 1.198721170425415,
      "learning_rate": 0.00019313202383349366,
      "loss": 4.0257,
      "step": 36840
    },
    {
      "epoch": 0.0343501626214534,
      "grad_norm": 1.4252009391784668,
      "learning_rate": 0.0001931301595113985,
      "loss": 4.0304,
      "step": 36850
    },
    {
      "epoch": 0.034359484239532494,
      "grad_norm": 1.1381385326385498,
      "learning_rate": 0.00019312829518930327,
      "loss": 4.26,
      "step": 36860
    },
    {
      "epoch": 0.03436880585761159,
      "grad_norm": 1.1939293146133423,
      "learning_rate": 0.00019312643086720808,
      "loss": 4.1445,
      "step": 36870
    },
    {
      "epoch": 0.034378127475690676,
      "grad_norm": 1.0256391763687134,
      "learning_rate": 0.00019312456654511288,
      "loss": 4.3222,
      "step": 36880
    },
    {
      "epoch": 0.03438744909376977,
      "grad_norm": 1.3088624477386475,
      "learning_rate": 0.00019312270222301766,
      "loss": 4.1603,
      "step": 36890
    },
    {
      "epoch": 0.03439677071184886,
      "grad_norm": 1.0275293588638306,
      "learning_rate": 0.00019312083790092247,
      "loss": 4.0491,
      "step": 36900
    },
    {
      "epoch": 0.03440609232992795,
      "grad_norm": 0.8876805305480957,
      "learning_rate": 0.00019311897357882727,
      "loss": 4.1309,
      "step": 36910
    },
    {
      "epoch": 0.03441541394800705,
      "grad_norm": 1.2451214790344238,
      "learning_rate": 0.00019311710925673208,
      "loss": 4.1481,
      "step": 36920
    },
    {
      "epoch": 0.034424735566086134,
      "grad_norm": 1.0675432682037354,
      "learning_rate": 0.0001931152449346369,
      "loss": 4.0584,
      "step": 36930
    },
    {
      "epoch": 0.03443405718416523,
      "grad_norm": 1.0127511024475098,
      "learning_rate": 0.00019311338061254167,
      "loss": 4.0239,
      "step": 36940
    },
    {
      "epoch": 0.034443378802244316,
      "grad_norm": 1.2265232801437378,
      "learning_rate": 0.00019311151629044647,
      "loss": 3.8945,
      "step": 36950
    },
    {
      "epoch": 0.03445270042032341,
      "grad_norm": 0.9318026900291443,
      "learning_rate": 0.00019310965196835128,
      "loss": 4.2402,
      "step": 36960
    },
    {
      "epoch": 0.034462022038402505,
      "grad_norm": 1.138334035873413,
      "learning_rate": 0.00019310778764625609,
      "loss": 4.0939,
      "step": 36970
    },
    {
      "epoch": 0.03447134365648159,
      "grad_norm": 1.0871986150741577,
      "learning_rate": 0.0001931059233241609,
      "loss": 4.1102,
      "step": 36980
    },
    {
      "epoch": 0.03448066527456069,
      "grad_norm": 1.1284918785095215,
      "learning_rate": 0.00019310405900206567,
      "loss": 4.1778,
      "step": 36990
    },
    {
      "epoch": 0.03448998689263978,
      "grad_norm": 1.3691426515579224,
      "learning_rate": 0.00019310219467997048,
      "loss": 3.9821,
      "step": 37000
    },
    {
      "epoch": 0.03449930851071887,
      "grad_norm": 1.2669047117233276,
      "learning_rate": 0.00019310033035787528,
      "loss": 4.1436,
      "step": 37010
    },
    {
      "epoch": 0.034508630128797964,
      "grad_norm": 1.4029881954193115,
      "learning_rate": 0.0001930984660357801,
      "loss": 4.0675,
      "step": 37020
    },
    {
      "epoch": 0.03451795174687705,
      "grad_norm": 0.953246533870697,
      "learning_rate": 0.0001930966017136849,
      "loss": 4.183,
      "step": 37030
    },
    {
      "epoch": 0.034527273364956146,
      "grad_norm": 1.1931242942810059,
      "learning_rate": 0.00019309473739158967,
      "loss": 4.1597,
      "step": 37040
    },
    {
      "epoch": 0.03453659498303524,
      "grad_norm": 1.086845874786377,
      "learning_rate": 0.00019309287306949448,
      "loss": 4.0815,
      "step": 37050
    },
    {
      "epoch": 0.03454591660111433,
      "grad_norm": 1.0687378644943237,
      "learning_rate": 0.00019309100874739926,
      "loss": 4.1271,
      "step": 37060
    },
    {
      "epoch": 0.03455523821919342,
      "grad_norm": 1.200475811958313,
      "learning_rate": 0.0001930891444253041,
      "loss": 4.2968,
      "step": 37070
    },
    {
      "epoch": 0.03456455983727251,
      "grad_norm": 1.2920763492584229,
      "learning_rate": 0.0001930872801032089,
      "loss": 4.1347,
      "step": 37080
    },
    {
      "epoch": 0.034573881455351604,
      "grad_norm": 1.060815453529358,
      "learning_rate": 0.00019308541578111368,
      "loss": 4.0239,
      "step": 37090
    },
    {
      "epoch": 0.0345832030734307,
      "grad_norm": 0.9542567729949951,
      "learning_rate": 0.00019308355145901849,
      "loss": 4.1661,
      "step": 37100
    },
    {
      "epoch": 0.034592524691509786,
      "grad_norm": 1.1316769123077393,
      "learning_rate": 0.00019308168713692326,
      "loss": 4.1033,
      "step": 37110
    },
    {
      "epoch": 0.03460184630958888,
      "grad_norm": 1.0580530166625977,
      "learning_rate": 0.00019307982281482807,
      "loss": 4.0931,
      "step": 37120
    },
    {
      "epoch": 0.034611167927667975,
      "grad_norm": 1.2586597204208374,
      "learning_rate": 0.00019307795849273288,
      "loss": 4.0222,
      "step": 37130
    },
    {
      "epoch": 0.03462048954574706,
      "grad_norm": 1.0800493955612183,
      "learning_rate": 0.00019307609417063768,
      "loss": 4.1476,
      "step": 37140
    },
    {
      "epoch": 0.03462981116382616,
      "grad_norm": 1.0200823545455933,
      "learning_rate": 0.0001930742298485425,
      "loss": 4.1189,
      "step": 37150
    },
    {
      "epoch": 0.034639132781905245,
      "grad_norm": 0.9602415561676025,
      "learning_rate": 0.00019307236552644727,
      "loss": 4.0825,
      "step": 37160
    },
    {
      "epoch": 0.03464845439998434,
      "grad_norm": 1.0656561851501465,
      "learning_rate": 0.00019307050120435208,
      "loss": 4.2181,
      "step": 37170
    },
    {
      "epoch": 0.034657776018063434,
      "grad_norm": 1.1389808654785156,
      "learning_rate": 0.00019306863688225688,
      "loss": 4.1881,
      "step": 37180
    },
    {
      "epoch": 0.03466709763614252,
      "grad_norm": 1.1608009338378906,
      "learning_rate": 0.0001930667725601617,
      "loss": 4.2282,
      "step": 37190
    },
    {
      "epoch": 0.034676419254221616,
      "grad_norm": 0.9503598809242249,
      "learning_rate": 0.0001930649082380665,
      "loss": 4.1213,
      "step": 37200
    },
    {
      "epoch": 0.03468574087230071,
      "grad_norm": 1.0735725164413452,
      "learning_rate": 0.00019306304391597127,
      "loss": 4.1946,
      "step": 37210
    },
    {
      "epoch": 0.0346950624903798,
      "grad_norm": 1.307177186012268,
      "learning_rate": 0.00019306117959387608,
      "loss": 4.1473,
      "step": 37220
    },
    {
      "epoch": 0.03470438410845889,
      "grad_norm": 1.0839482545852661,
      "learning_rate": 0.00019305931527178089,
      "loss": 4.1153,
      "step": 37230
    },
    {
      "epoch": 0.03471370572653798,
      "grad_norm": 1.0655816793441772,
      "learning_rate": 0.0001930574509496857,
      "loss": 4.1983,
      "step": 37240
    },
    {
      "epoch": 0.034723027344617075,
      "grad_norm": 1.2031570672988892,
      "learning_rate": 0.0001930555866275905,
      "loss": 4.2268,
      "step": 37250
    },
    {
      "epoch": 0.03473234896269617,
      "grad_norm": 1.1103659868240356,
      "learning_rate": 0.00019305372230549528,
      "loss": 4.168,
      "step": 37260
    },
    {
      "epoch": 0.03474167058077526,
      "grad_norm": 1.0467405319213867,
      "learning_rate": 0.00019305185798340008,
      "loss": 4.2669,
      "step": 37270
    },
    {
      "epoch": 0.03475099219885435,
      "grad_norm": 1.036921739578247,
      "learning_rate": 0.0001930499936613049,
      "loss": 4.0545,
      "step": 37280
    },
    {
      "epoch": 0.03476031381693344,
      "grad_norm": 1.0522373914718628,
      "learning_rate": 0.00019304812933920967,
      "loss": 4.1774,
      "step": 37290
    },
    {
      "epoch": 0.03476963543501253,
      "grad_norm": 1.236254096031189,
      "learning_rate": 0.0001930462650171145,
      "loss": 4.1376,
      "step": 37300
    },
    {
      "epoch": 0.03477895705309163,
      "grad_norm": 1.059876561164856,
      "learning_rate": 0.00019304440069501928,
      "loss": 4.0118,
      "step": 37310
    },
    {
      "epoch": 0.034788278671170715,
      "grad_norm": 1.0066449642181396,
      "learning_rate": 0.0001930425363729241,
      "loss": 4.1546,
      "step": 37320
    },
    {
      "epoch": 0.03479760028924981,
      "grad_norm": 0.9584740996360779,
      "learning_rate": 0.0001930406720508289,
      "loss": 4.028,
      "step": 37330
    },
    {
      "epoch": 0.034806921907328904,
      "grad_norm": 1.130561113357544,
      "learning_rate": 0.00019303880772873367,
      "loss": 3.974,
      "step": 37340
    },
    {
      "epoch": 0.03481624352540799,
      "grad_norm": 1.147829532623291,
      "learning_rate": 0.00019303694340663848,
      "loss": 4.2148,
      "step": 37350
    },
    {
      "epoch": 0.034825565143487086,
      "grad_norm": 1.0685862302780151,
      "learning_rate": 0.0001930350790845433,
      "loss": 3.9965,
      "step": 37360
    },
    {
      "epoch": 0.034834886761566174,
      "grad_norm": 1.1603268384933472,
      "learning_rate": 0.0001930332147624481,
      "loss": 4.1793,
      "step": 37370
    },
    {
      "epoch": 0.03484420837964527,
      "grad_norm": 1.3336553573608398,
      "learning_rate": 0.0001930313504403529,
      "loss": 3.9406,
      "step": 37380
    },
    {
      "epoch": 0.03485352999772436,
      "grad_norm": 1.2934613227844238,
      "learning_rate": 0.00019302948611825768,
      "loss": 4.1266,
      "step": 37390
    },
    {
      "epoch": 0.03486285161580345,
      "grad_norm": 0.8760841488838196,
      "learning_rate": 0.00019302762179616249,
      "loss": 3.9917,
      "step": 37400
    },
    {
      "epoch": 0.034872173233882545,
      "grad_norm": 1.1959586143493652,
      "learning_rate": 0.0001930257574740673,
      "loss": 4.19,
      "step": 37410
    },
    {
      "epoch": 0.03488149485196163,
      "grad_norm": 1.1685031652450562,
      "learning_rate": 0.0001930238931519721,
      "loss": 3.9437,
      "step": 37420
    },
    {
      "epoch": 0.03489081647004073,
      "grad_norm": 1.0413854122161865,
      "learning_rate": 0.0001930220288298769,
      "loss": 4.1721,
      "step": 37430
    },
    {
      "epoch": 0.03490013808811982,
      "grad_norm": 1.1367155313491821,
      "learning_rate": 0.00019302016450778168,
      "loss": 4.0766,
      "step": 37440
    },
    {
      "epoch": 0.03490945970619891,
      "grad_norm": 1.211472749710083,
      "learning_rate": 0.0001930183001856865,
      "loss": 3.9047,
      "step": 37450
    },
    {
      "epoch": 0.034918781324278,
      "grad_norm": 1.083451747894287,
      "learning_rate": 0.0001930164358635913,
      "loss": 3.99,
      "step": 37460
    },
    {
      "epoch": 0.0349281029423571,
      "grad_norm": 1.1669692993164062,
      "learning_rate": 0.0001930145715414961,
      "loss": 4.1988,
      "step": 37470
    },
    {
      "epoch": 0.034937424560436185,
      "grad_norm": 1.098412275314331,
      "learning_rate": 0.0001930127072194009,
      "loss": 4.124,
      "step": 37480
    },
    {
      "epoch": 0.03494674617851528,
      "grad_norm": 1.232627511024475,
      "learning_rate": 0.0001930108428973057,
      "loss": 3.8462,
      "step": 37490
    },
    {
      "epoch": 0.03495606779659437,
      "grad_norm": 1.2894049882888794,
      "learning_rate": 0.0001930089785752105,
      "loss": 4.1974,
      "step": 37500
    },
    {
      "epoch": 0.03496538941467346,
      "grad_norm": 1.2277299165725708,
      "learning_rate": 0.00019300711425311527,
      "loss": 3.9721,
      "step": 37510
    },
    {
      "epoch": 0.034974711032752556,
      "grad_norm": 0.9930118322372437,
      "learning_rate": 0.0001930052499310201,
      "loss": 4.0396,
      "step": 37520
    },
    {
      "epoch": 0.034984032650831644,
      "grad_norm": 1.0868483781814575,
      "learning_rate": 0.0001930033856089249,
      "loss": 4.2004,
      "step": 37530
    },
    {
      "epoch": 0.03499335426891074,
      "grad_norm": 1.0717490911483765,
      "learning_rate": 0.0001930015212868297,
      "loss": 4.1702,
      "step": 37540
    },
    {
      "epoch": 0.03500267588698983,
      "grad_norm": 1.0120840072631836,
      "learning_rate": 0.0001929996569647345,
      "loss": 4.1693,
      "step": 37550
    },
    {
      "epoch": 0.03501199750506892,
      "grad_norm": 1.0298526287078857,
      "learning_rate": 0.00019299779264263928,
      "loss": 3.9316,
      "step": 37560
    },
    {
      "epoch": 0.035021319123148015,
      "grad_norm": 1.0446021556854248,
      "learning_rate": 0.00019299592832054408,
      "loss": 4.2155,
      "step": 37570
    },
    {
      "epoch": 0.0350306407412271,
      "grad_norm": 0.9766624569892883,
      "learning_rate": 0.00019299406399844892,
      "loss": 4.1412,
      "step": 37580
    },
    {
      "epoch": 0.0350399623593062,
      "grad_norm": 1.1293927431106567,
      "learning_rate": 0.0001929921996763537,
      "loss": 4.0334,
      "step": 37590
    },
    {
      "epoch": 0.03504928397738529,
      "grad_norm": 1.1251236200332642,
      "learning_rate": 0.0001929903353542585,
      "loss": 4.0694,
      "step": 37600
    },
    {
      "epoch": 0.03505860559546438,
      "grad_norm": 1.1505190134048462,
      "learning_rate": 0.00019298847103216328,
      "loss": 3.9573,
      "step": 37610
    },
    {
      "epoch": 0.03506792721354347,
      "grad_norm": 1.2109756469726562,
      "learning_rate": 0.0001929866067100681,
      "loss": 4.0886,
      "step": 37620
    },
    {
      "epoch": 0.03507724883162256,
      "grad_norm": 1.030470848083496,
      "learning_rate": 0.0001929847423879729,
      "loss": 4.0496,
      "step": 37630
    },
    {
      "epoch": 0.035086570449701655,
      "grad_norm": 1.2648037672042847,
      "learning_rate": 0.0001929828780658777,
      "loss": 4.0356,
      "step": 37640
    },
    {
      "epoch": 0.03509589206778075,
      "grad_norm": 0.8847181797027588,
      "learning_rate": 0.0001929810137437825,
      "loss": 4.1396,
      "step": 37650
    },
    {
      "epoch": 0.03510521368585984,
      "grad_norm": 0.9661827087402344,
      "learning_rate": 0.0001929791494216873,
      "loss": 4.2031,
      "step": 37660
    },
    {
      "epoch": 0.03511453530393893,
      "grad_norm": 1.1864532232284546,
      "learning_rate": 0.0001929772850995921,
      "loss": 4.2077,
      "step": 37670
    },
    {
      "epoch": 0.035123856922018026,
      "grad_norm": 1.080757737159729,
      "learning_rate": 0.00019297542077749687,
      "loss": 4.312,
      "step": 37680
    },
    {
      "epoch": 0.035133178540097114,
      "grad_norm": 1.1037592887878418,
      "learning_rate": 0.0001929735564554017,
      "loss": 4.0737,
      "step": 37690
    },
    {
      "epoch": 0.03514250015817621,
      "grad_norm": 1.0806896686553955,
      "learning_rate": 0.0001929716921333065,
      "loss": 4.0457,
      "step": 37700
    },
    {
      "epoch": 0.035151821776255296,
      "grad_norm": 0.9931670427322388,
      "learning_rate": 0.0001929698278112113,
      "loss": 3.9832,
      "step": 37710
    },
    {
      "epoch": 0.03516114339433439,
      "grad_norm": 1.0792572498321533,
      "learning_rate": 0.0001929679634891161,
      "loss": 4.013,
      "step": 37720
    },
    {
      "epoch": 0.035170465012413485,
      "grad_norm": 1.1221245527267456,
      "learning_rate": 0.00019296609916702088,
      "loss": 4.1552,
      "step": 37730
    },
    {
      "epoch": 0.03517978663049257,
      "grad_norm": 1.019517421722412,
      "learning_rate": 0.00019296423484492568,
      "loss": 4.0277,
      "step": 37740
    },
    {
      "epoch": 0.03518910824857167,
      "grad_norm": 1.0643600225448608,
      "learning_rate": 0.00019296237052283052,
      "loss": 3.997,
      "step": 37750
    },
    {
      "epoch": 0.035198429866650754,
      "grad_norm": 1.2044190168380737,
      "learning_rate": 0.0001929605062007353,
      "loss": 3.9626,
      "step": 37760
    },
    {
      "epoch": 0.03520775148472985,
      "grad_norm": 1.1498409509658813,
      "learning_rate": 0.0001929586418786401,
      "loss": 4.2779,
      "step": 37770
    },
    {
      "epoch": 0.03521707310280894,
      "grad_norm": 0.9934914112091064,
      "learning_rate": 0.00019295677755654488,
      "loss": 3.9187,
      "step": 37780
    },
    {
      "epoch": 0.03522639472088803,
      "grad_norm": 1.0976499319076538,
      "learning_rate": 0.0001929549132344497,
      "loss": 4.0834,
      "step": 37790
    },
    {
      "epoch": 0.035235716338967125,
      "grad_norm": 0.9842775464057922,
      "learning_rate": 0.0001929530489123545,
      "loss": 4.2619,
      "step": 37800
    },
    {
      "epoch": 0.03524503795704622,
      "grad_norm": 1.2516199350357056,
      "learning_rate": 0.0001929511845902593,
      "loss": 4.063,
      "step": 37810
    },
    {
      "epoch": 0.03525435957512531,
      "grad_norm": 1.2339706420898438,
      "learning_rate": 0.0001929493202681641,
      "loss": 4.1045,
      "step": 37820
    },
    {
      "epoch": 0.0352636811932044,
      "grad_norm": 0.9953703284263611,
      "learning_rate": 0.00019294745594606889,
      "loss": 4.2194,
      "step": 37830
    },
    {
      "epoch": 0.03527300281128349,
      "grad_norm": 1.1261515617370605,
      "learning_rate": 0.0001929455916239737,
      "loss": 4.3766,
      "step": 37840
    },
    {
      "epoch": 0.035282324429362584,
      "grad_norm": 1.1232614517211914,
      "learning_rate": 0.0001929437273018785,
      "loss": 3.9933,
      "step": 37850
    },
    {
      "epoch": 0.03529164604744168,
      "grad_norm": 1.1376248598098755,
      "learning_rate": 0.0001929418629797833,
      "loss": 4.1745,
      "step": 37860
    },
    {
      "epoch": 0.035300967665520766,
      "grad_norm": 1.0965864658355713,
      "learning_rate": 0.0001929399986576881,
      "loss": 4.0915,
      "step": 37870
    },
    {
      "epoch": 0.03531028928359986,
      "grad_norm": 1.083086609840393,
      "learning_rate": 0.0001929381343355929,
      "loss": 4.0514,
      "step": 37880
    },
    {
      "epoch": 0.035319610901678955,
      "grad_norm": 1.0335804224014282,
      "learning_rate": 0.0001929362700134977,
      "loss": 4.1879,
      "step": 37890
    },
    {
      "epoch": 0.03532893251975804,
      "grad_norm": 1.0088423490524292,
      "learning_rate": 0.0001929344056914025,
      "loss": 4.3168,
      "step": 37900
    },
    {
      "epoch": 0.03533825413783714,
      "grad_norm": 1.0654244422912598,
      "learning_rate": 0.0001929325413693073,
      "loss": 4.0455,
      "step": 37910
    },
    {
      "epoch": 0.035347575755916225,
      "grad_norm": 1.1247202157974243,
      "learning_rate": 0.00019293067704721212,
      "loss": 4.1735,
      "step": 37920
    },
    {
      "epoch": 0.03535689737399532,
      "grad_norm": 1.1728848218917847,
      "learning_rate": 0.0001929288127251169,
      "loss": 3.9073,
      "step": 37930
    },
    {
      "epoch": 0.035366218992074414,
      "grad_norm": 1.1570117473602295,
      "learning_rate": 0.0001929269484030217,
      "loss": 4.0584,
      "step": 37940
    },
    {
      "epoch": 0.0353755406101535,
      "grad_norm": 1.2186565399169922,
      "learning_rate": 0.0001929250840809265,
      "loss": 4.0919,
      "step": 37950
    },
    {
      "epoch": 0.035384862228232596,
      "grad_norm": 1.125122308731079,
      "learning_rate": 0.00019292321975883129,
      "loss": 4.3063,
      "step": 37960
    },
    {
      "epoch": 0.03539418384631168,
      "grad_norm": 0.910286009311676,
      "learning_rate": 0.00019292135543673612,
      "loss": 4.1109,
      "step": 37970
    },
    {
      "epoch": 0.03540350546439078,
      "grad_norm": 1.1610174179077148,
      "learning_rate": 0.0001929194911146409,
      "loss": 4.13,
      "step": 37980
    },
    {
      "epoch": 0.03541282708246987,
      "grad_norm": 1.1814898252487183,
      "learning_rate": 0.0001929176267925457,
      "loss": 4.1174,
      "step": 37990
    },
    {
      "epoch": 0.03542214870054896,
      "grad_norm": 1.1213868856430054,
      "learning_rate": 0.0001929157624704505,
      "loss": 4.2011,
      "step": 38000
    },
    {
      "epoch": 0.035431470318628054,
      "grad_norm": 0.9935012459754944,
      "learning_rate": 0.0001929138981483553,
      "loss": 4.246,
      "step": 38010
    },
    {
      "epoch": 0.03544079193670715,
      "grad_norm": 1.2218350172042847,
      "learning_rate": 0.0001929120338262601,
      "loss": 4.0329,
      "step": 38020
    },
    {
      "epoch": 0.035450113554786236,
      "grad_norm": 1.1544662714004517,
      "learning_rate": 0.0001929101695041649,
      "loss": 4.2933,
      "step": 38030
    },
    {
      "epoch": 0.03545943517286533,
      "grad_norm": 1.1323243379592896,
      "learning_rate": 0.0001929083051820697,
      "loss": 4.2166,
      "step": 38040
    },
    {
      "epoch": 0.03546875679094442,
      "grad_norm": 1.1518605947494507,
      "learning_rate": 0.00019290644085997452,
      "loss": 4.1768,
      "step": 38050
    },
    {
      "epoch": 0.03547807840902351,
      "grad_norm": 1.0396889448165894,
      "learning_rate": 0.0001929045765378793,
      "loss": 4.0994,
      "step": 38060
    },
    {
      "epoch": 0.03548740002710261,
      "grad_norm": 1.106223225593567,
      "learning_rate": 0.0001929027122157841,
      "loss": 4.149,
      "step": 38070
    },
    {
      "epoch": 0.035496721645181695,
      "grad_norm": 1.1274343729019165,
      "learning_rate": 0.0001929008478936889,
      "loss": 4.2291,
      "step": 38080
    },
    {
      "epoch": 0.03550604326326079,
      "grad_norm": 0.9864222407341003,
      "learning_rate": 0.00019289898357159371,
      "loss": 3.9681,
      "step": 38090
    },
    {
      "epoch": 0.03551536488133988,
      "grad_norm": 1.151656150817871,
      "learning_rate": 0.00019289711924949852,
      "loss": 3.9877,
      "step": 38100
    },
    {
      "epoch": 0.03552468649941897,
      "grad_norm": 0.9614357352256775,
      "learning_rate": 0.0001928952549274033,
      "loss": 4.1294,
      "step": 38110
    },
    {
      "epoch": 0.035534008117498066,
      "grad_norm": 1.2208691835403442,
      "learning_rate": 0.0001928933906053081,
      "loss": 4.1737,
      "step": 38120
    },
    {
      "epoch": 0.03554332973557715,
      "grad_norm": 1.2761729955673218,
      "learning_rate": 0.00019289152628321289,
      "loss": 4.2599,
      "step": 38130
    },
    {
      "epoch": 0.03555265135365625,
      "grad_norm": 1.239065170288086,
      "learning_rate": 0.00019288966196111772,
      "loss": 4.1248,
      "step": 38140
    },
    {
      "epoch": 0.03556197297173534,
      "grad_norm": 1.058030366897583,
      "learning_rate": 0.00019288779763902252,
      "loss": 4.1793,
      "step": 38150
    },
    {
      "epoch": 0.03557129458981443,
      "grad_norm": 1.0920835733413696,
      "learning_rate": 0.0001928859333169273,
      "loss": 3.9504,
      "step": 38160
    },
    {
      "epoch": 0.035580616207893524,
      "grad_norm": 1.0125013589859009,
      "learning_rate": 0.0001928840689948321,
      "loss": 4.1622,
      "step": 38170
    },
    {
      "epoch": 0.03558993782597261,
      "grad_norm": 1.0700693130493164,
      "learning_rate": 0.0001928822046727369,
      "loss": 4.0833,
      "step": 38180
    },
    {
      "epoch": 0.035599259444051706,
      "grad_norm": 0.9074611067771912,
      "learning_rate": 0.0001928803403506417,
      "loss": 4.0804,
      "step": 38190
    },
    {
      "epoch": 0.0356085810621308,
      "grad_norm": 0.9927794337272644,
      "learning_rate": 0.00019287847602854653,
      "loss": 4.2716,
      "step": 38200
    },
    {
      "epoch": 0.03561790268020989,
      "grad_norm": 1.1263409852981567,
      "learning_rate": 0.0001928766117064513,
      "loss": 4.1746,
      "step": 38210
    },
    {
      "epoch": 0.03562722429828898,
      "grad_norm": 0.8839393258094788,
      "learning_rate": 0.00019287474738435611,
      "loss": 4.0853,
      "step": 38220
    },
    {
      "epoch": 0.03563654591636808,
      "grad_norm": 1.311875820159912,
      "learning_rate": 0.0001928728830622609,
      "loss": 4.0033,
      "step": 38230
    },
    {
      "epoch": 0.035645867534447165,
      "grad_norm": 1.1489866971969604,
      "learning_rate": 0.0001928710187401657,
      "loss": 4.0609,
      "step": 38240
    },
    {
      "epoch": 0.03565518915252626,
      "grad_norm": 1.1077262163162231,
      "learning_rate": 0.0001928691544180705,
      "loss": 4.0571,
      "step": 38250
    },
    {
      "epoch": 0.03566451077060535,
      "grad_norm": 1.1321465969085693,
      "learning_rate": 0.0001928672900959753,
      "loss": 4.127,
      "step": 38260
    },
    {
      "epoch": 0.03567383238868444,
      "grad_norm": 1.1192363500595093,
      "learning_rate": 0.00019286542577388012,
      "loss": 4.1384,
      "step": 38270
    },
    {
      "epoch": 0.035683154006763536,
      "grad_norm": 0.8083392977714539,
      "learning_rate": 0.0001928635614517849,
      "loss": 4.3295,
      "step": 38280
    },
    {
      "epoch": 0.03569247562484262,
      "grad_norm": 1.1051825284957886,
      "learning_rate": 0.0001928616971296897,
      "loss": 4.2387,
      "step": 38290
    },
    {
      "epoch": 0.03570179724292172,
      "grad_norm": 1.0385361909866333,
      "learning_rate": 0.0001928598328075945,
      "loss": 4.1072,
      "step": 38300
    },
    {
      "epoch": 0.035711118861000805,
      "grad_norm": 1.1727995872497559,
      "learning_rate": 0.00019285796848549932,
      "loss": 4.0122,
      "step": 38310
    },
    {
      "epoch": 0.0357204404790799,
      "grad_norm": 1.1293745040893555,
      "learning_rate": 0.00019285610416340412,
      "loss": 4.156,
      "step": 38320
    },
    {
      "epoch": 0.035729762097158994,
      "grad_norm": 1.1005111932754517,
      "learning_rate": 0.0001928542398413089,
      "loss": 4.072,
      "step": 38330
    },
    {
      "epoch": 0.03573908371523808,
      "grad_norm": 0.9696564078330994,
      "learning_rate": 0.0001928523755192137,
      "loss": 4.1841,
      "step": 38340
    },
    {
      "epoch": 0.035748405333317176,
      "grad_norm": 1.1355071067810059,
      "learning_rate": 0.00019285051119711852,
      "loss": 4.0455,
      "step": 38350
    },
    {
      "epoch": 0.03575772695139627,
      "grad_norm": 1.096312403678894,
      "learning_rate": 0.00019284864687502332,
      "loss": 4.0795,
      "step": 38360
    },
    {
      "epoch": 0.03576704856947536,
      "grad_norm": 1.0828289985656738,
      "learning_rate": 0.00019284678255292813,
      "loss": 3.9667,
      "step": 38370
    },
    {
      "epoch": 0.03577637018755445,
      "grad_norm": 1.1103864908218384,
      "learning_rate": 0.0001928449182308329,
      "loss": 4.0373,
      "step": 38380
    },
    {
      "epoch": 0.03578569180563354,
      "grad_norm": 1.2596715688705444,
      "learning_rate": 0.0001928430539087377,
      "loss": 4.244,
      "step": 38390
    },
    {
      "epoch": 0.035795013423712635,
      "grad_norm": 1.2255476713180542,
      "learning_rate": 0.00019284118958664252,
      "loss": 4.1719,
      "step": 38400
    },
    {
      "epoch": 0.03580433504179173,
      "grad_norm": 1.1718732118606567,
      "learning_rate": 0.0001928393252645473,
      "loss": 4.0645,
      "step": 38410
    },
    {
      "epoch": 0.03581365665987082,
      "grad_norm": 1.1468505859375,
      "learning_rate": 0.00019283746094245213,
      "loss": 3.9502,
      "step": 38420
    },
    {
      "epoch": 0.03582297827794991,
      "grad_norm": 1.115966558456421,
      "learning_rate": 0.0001928355966203569,
      "loss": 4.1416,
      "step": 38430
    },
    {
      "epoch": 0.035832299896029,
      "grad_norm": 0.9636500477790833,
      "learning_rate": 0.00019283373229826172,
      "loss": 4.0876,
      "step": 38440
    },
    {
      "epoch": 0.035841621514108093,
      "grad_norm": 1.1065759658813477,
      "learning_rate": 0.00019283186797616652,
      "loss": 4.0143,
      "step": 38450
    },
    {
      "epoch": 0.03585094313218719,
      "grad_norm": 1.0024652481079102,
      "learning_rate": 0.0001928300036540713,
      "loss": 3.9929,
      "step": 38460
    },
    {
      "epoch": 0.035860264750266276,
      "grad_norm": 1.0118972063064575,
      "learning_rate": 0.0001928281393319761,
      "loss": 4.0993,
      "step": 38470
    },
    {
      "epoch": 0.03586958636834537,
      "grad_norm": 1.3114248514175415,
      "learning_rate": 0.00019282627500988092,
      "loss": 4.2248,
      "step": 38480
    },
    {
      "epoch": 0.035878907986424464,
      "grad_norm": 1.31553053855896,
      "learning_rate": 0.00019282441068778572,
      "loss": 4.2386,
      "step": 38490
    },
    {
      "epoch": 0.03588822960450355,
      "grad_norm": 1.116809368133545,
      "learning_rate": 0.00019282254636569053,
      "loss": 4.0095,
      "step": 38500
    },
    {
      "epoch": 0.035897551222582647,
      "grad_norm": 1.1744763851165771,
      "learning_rate": 0.0001928206820435953,
      "loss": 4.0827,
      "step": 38510
    },
    {
      "epoch": 0.035906872840661734,
      "grad_norm": 1.2277768850326538,
      "learning_rate": 0.00019281881772150011,
      "loss": 3.9955,
      "step": 38520
    },
    {
      "epoch": 0.03591619445874083,
      "grad_norm": 1.0382577180862427,
      "learning_rate": 0.00019281695339940492,
      "loss": 4.244,
      "step": 38530
    },
    {
      "epoch": 0.03592551607681992,
      "grad_norm": 1.193824291229248,
      "learning_rate": 0.00019281508907730973,
      "loss": 3.9123,
      "step": 38540
    },
    {
      "epoch": 0.03593483769489901,
      "grad_norm": 1.1061679124832153,
      "learning_rate": 0.00019281322475521453,
      "loss": 4.1577,
      "step": 38550
    },
    {
      "epoch": 0.035944159312978105,
      "grad_norm": 1.2611697912216187,
      "learning_rate": 0.0001928113604331193,
      "loss": 4.3318,
      "step": 38560
    },
    {
      "epoch": 0.0359534809310572,
      "grad_norm": 1.0786017179489136,
      "learning_rate": 0.00019280949611102412,
      "loss": 4.3025,
      "step": 38570
    },
    {
      "epoch": 0.03596280254913629,
      "grad_norm": 0.9602004289627075,
      "learning_rate": 0.0001928076317889289,
      "loss": 4.1843,
      "step": 38580
    },
    {
      "epoch": 0.03597212416721538,
      "grad_norm": 1.1349087953567505,
      "learning_rate": 0.00019280576746683373,
      "loss": 4.1051,
      "step": 38590
    },
    {
      "epoch": 0.03598144578529447,
      "grad_norm": 1.0763399600982666,
      "learning_rate": 0.00019280390314473854,
      "loss": 4.2815,
      "step": 38600
    },
    {
      "epoch": 0.035990767403373564,
      "grad_norm": 1.237038493156433,
      "learning_rate": 0.00019280203882264332,
      "loss": 4.2854,
      "step": 38610
    },
    {
      "epoch": 0.03600008902145266,
      "grad_norm": 1.1187161207199097,
      "learning_rate": 0.00019280017450054812,
      "loss": 4.096,
      "step": 38620
    },
    {
      "epoch": 0.036009410639531746,
      "grad_norm": 1.0410243272781372,
      "learning_rate": 0.0001927983101784529,
      "loss": 4.0575,
      "step": 38630
    },
    {
      "epoch": 0.03601873225761084,
      "grad_norm": 1.2881778478622437,
      "learning_rate": 0.0001927964458563577,
      "loss": 4.0707,
      "step": 38640
    },
    {
      "epoch": 0.03602805387568993,
      "grad_norm": 0.9647039771080017,
      "learning_rate": 0.00019279458153426254,
      "loss": 4.0387,
      "step": 38650
    },
    {
      "epoch": 0.03603737549376902,
      "grad_norm": 1.1684190034866333,
      "learning_rate": 0.00019279271721216732,
      "loss": 3.9463,
      "step": 38660
    },
    {
      "epoch": 0.03604669711184812,
      "grad_norm": 1.1235665082931519,
      "learning_rate": 0.00019279085289007213,
      "loss": 4.2385,
      "step": 38670
    },
    {
      "epoch": 0.036056018729927204,
      "grad_norm": 1.1289571523666382,
      "learning_rate": 0.0001927889885679769,
      "loss": 4.1347,
      "step": 38680
    },
    {
      "epoch": 0.0360653403480063,
      "grad_norm": 0.9990030527114868,
      "learning_rate": 0.0001927871242458817,
      "loss": 4.0716,
      "step": 38690
    },
    {
      "epoch": 0.03607466196608539,
      "grad_norm": 1.0505616664886475,
      "learning_rate": 0.00019278525992378652,
      "loss": 4.0661,
      "step": 38700
    },
    {
      "epoch": 0.03608398358416448,
      "grad_norm": 1.0483348369598389,
      "learning_rate": 0.00019278339560169133,
      "loss": 4.2564,
      "step": 38710
    },
    {
      "epoch": 0.036093305202243575,
      "grad_norm": 1.1271648406982422,
      "learning_rate": 0.00019278153127959613,
      "loss": 4.104,
      "step": 38720
    },
    {
      "epoch": 0.03610262682032266,
      "grad_norm": 1.1400721073150635,
      "learning_rate": 0.0001927796669575009,
      "loss": 4.0892,
      "step": 38730
    },
    {
      "epoch": 0.03611194843840176,
      "grad_norm": 1.0966544151306152,
      "learning_rate": 0.00019277780263540572,
      "loss": 4.1419,
      "step": 38740
    },
    {
      "epoch": 0.03612127005648085,
      "grad_norm": 0.9571834802627563,
      "learning_rate": 0.00019277593831331052,
      "loss": 4.2092,
      "step": 38750
    },
    {
      "epoch": 0.03613059167455994,
      "grad_norm": 1.1537004709243774,
      "learning_rate": 0.00019277407399121533,
      "loss": 4.0718,
      "step": 38760
    },
    {
      "epoch": 0.036139913292639034,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019277220966912014,
      "loss": 4.1072,
      "step": 38770
    },
    {
      "epoch": 0.03614923491071812,
      "grad_norm": 1.0633338689804077,
      "learning_rate": 0.00019277034534702492,
      "loss": 4.0238,
      "step": 38780
    },
    {
      "epoch": 0.036158556528797216,
      "grad_norm": 1.1426119804382324,
      "learning_rate": 0.00019276848102492972,
      "loss": 4.1106,
      "step": 38790
    },
    {
      "epoch": 0.03616787814687631,
      "grad_norm": 1.257477045059204,
      "learning_rate": 0.00019276661670283453,
      "loss": 4.1344,
      "step": 38800
    },
    {
      "epoch": 0.0361771997649554,
      "grad_norm": 1.0791774988174438,
      "learning_rate": 0.00019276475238073933,
      "loss": 3.9236,
      "step": 38810
    },
    {
      "epoch": 0.03618652138303449,
      "grad_norm": 0.8546579480171204,
      "learning_rate": 0.00019276288805864414,
      "loss": 4.3156,
      "step": 38820
    },
    {
      "epoch": 0.03619584300111359,
      "grad_norm": 1.1986312866210938,
      "learning_rate": 0.00019276102373654892,
      "loss": 4.1482,
      "step": 38830
    },
    {
      "epoch": 0.036205164619192674,
      "grad_norm": 1.2054708003997803,
      "learning_rate": 0.00019275915941445373,
      "loss": 4.1893,
      "step": 38840
    },
    {
      "epoch": 0.03621448623727177,
      "grad_norm": 1.1487696170806885,
      "learning_rate": 0.00019275729509235853,
      "loss": 4.0968,
      "step": 38850
    },
    {
      "epoch": 0.036223807855350856,
      "grad_norm": 1.185702919960022,
      "learning_rate": 0.0001927554307702633,
      "loss": 4.0918,
      "step": 38860
    },
    {
      "epoch": 0.03623312947342995,
      "grad_norm": 0.9696216583251953,
      "learning_rate": 0.00019275356644816815,
      "loss": 4.2066,
      "step": 38870
    },
    {
      "epoch": 0.036242451091509045,
      "grad_norm": 1.1611793041229248,
      "learning_rate": 0.00019275170212607292,
      "loss": 4.0236,
      "step": 38880
    },
    {
      "epoch": 0.03625177270958813,
      "grad_norm": 1.3543829917907715,
      "learning_rate": 0.00019274983780397773,
      "loss": 4.2347,
      "step": 38890
    },
    {
      "epoch": 0.03626109432766723,
      "grad_norm": 0.9604505300521851,
      "learning_rate": 0.00019274797348188254,
      "loss": 4.2547,
      "step": 38900
    },
    {
      "epoch": 0.03627041594574632,
      "grad_norm": 1.2448031902313232,
      "learning_rate": 0.00019274610915978732,
      "loss": 4.1791,
      "step": 38910
    },
    {
      "epoch": 0.03627973756382541,
      "grad_norm": 1.2097291946411133,
      "learning_rate": 0.00019274424483769212,
      "loss": 3.8927,
      "step": 38920
    },
    {
      "epoch": 0.036289059181904504,
      "grad_norm": 1.163057565689087,
      "learning_rate": 0.00019274238051559693,
      "loss": 4.0357,
      "step": 38930
    },
    {
      "epoch": 0.03629838079998359,
      "grad_norm": 1.104348063468933,
      "learning_rate": 0.00019274051619350174,
      "loss": 4.1288,
      "step": 38940
    },
    {
      "epoch": 0.036307702418062686,
      "grad_norm": 0.955568790435791,
      "learning_rate": 0.00019273865187140654,
      "loss": 4.0984,
      "step": 38950
    },
    {
      "epoch": 0.03631702403614178,
      "grad_norm": 1.3948545455932617,
      "learning_rate": 0.00019273678754931132,
      "loss": 4.1923,
      "step": 38960
    },
    {
      "epoch": 0.03632634565422087,
      "grad_norm": 1.241774320602417,
      "learning_rate": 0.00019273492322721613,
      "loss": 4.068,
      "step": 38970
    },
    {
      "epoch": 0.03633566727229996,
      "grad_norm": 1.1264034509658813,
      "learning_rate": 0.00019273305890512093,
      "loss": 3.9806,
      "step": 38980
    },
    {
      "epoch": 0.03634498889037905,
      "grad_norm": 1.154724359512329,
      "learning_rate": 0.00019273119458302574,
      "loss": 3.9456,
      "step": 38990
    },
    {
      "epoch": 0.036354310508458144,
      "grad_norm": 0.9994842410087585,
      "learning_rate": 0.00019272933026093055,
      "loss": 3.9537,
      "step": 39000
    },
    {
      "epoch": 0.03636363212653724,
      "grad_norm": 1.1617680788040161,
      "learning_rate": 0.00019272746593883533,
      "loss": 3.9698,
      "step": 39010
    },
    {
      "epoch": 0.036372953744616326,
      "grad_norm": 1.189218521118164,
      "learning_rate": 0.00019272560161674013,
      "loss": 4.1475,
      "step": 39020
    },
    {
      "epoch": 0.03638227536269542,
      "grad_norm": 1.1119070053100586,
      "learning_rate": 0.0001927237372946449,
      "loss": 3.9629,
      "step": 39030
    },
    {
      "epoch": 0.036391596980774515,
      "grad_norm": 1.3835257291793823,
      "learning_rate": 0.00019272187297254974,
      "loss": 3.9637,
      "step": 39040
    },
    {
      "epoch": 0.0364009185988536,
      "grad_norm": 1.1404370069503784,
      "learning_rate": 0.00019272000865045452,
      "loss": 4.0598,
      "step": 39050
    },
    {
      "epoch": 0.0364102402169327,
      "grad_norm": 1.1488208770751953,
      "learning_rate": 0.00019271814432835933,
      "loss": 3.8939,
      "step": 39060
    },
    {
      "epoch": 0.036419561835011785,
      "grad_norm": 1.2045726776123047,
      "learning_rate": 0.00019271628000626414,
      "loss": 4.1271,
      "step": 39070
    },
    {
      "epoch": 0.03642888345309088,
      "grad_norm": 1.4190287590026855,
      "learning_rate": 0.00019271441568416892,
      "loss": 4.0863,
      "step": 39080
    },
    {
      "epoch": 0.036438205071169974,
      "grad_norm": 1.1023459434509277,
      "learning_rate": 0.00019271255136207372,
      "loss": 4.224,
      "step": 39090
    },
    {
      "epoch": 0.03644752668924906,
      "grad_norm": 0.9695587754249573,
      "learning_rate": 0.00019271068703997853,
      "loss": 3.958,
      "step": 39100
    },
    {
      "epoch": 0.036456848307328156,
      "grad_norm": 1.0364203453063965,
      "learning_rate": 0.00019270882271788333,
      "loss": 4.1443,
      "step": 39110
    },
    {
      "epoch": 0.036466169925407244,
      "grad_norm": 1.1127005815505981,
      "learning_rate": 0.00019270695839578814,
      "loss": 4.0868,
      "step": 39120
    },
    {
      "epoch": 0.03647549154348634,
      "grad_norm": 1.054770827293396,
      "learning_rate": 0.00019270509407369292,
      "loss": 4.0401,
      "step": 39130
    },
    {
      "epoch": 0.03648481316156543,
      "grad_norm": 1.1767961978912354,
      "learning_rate": 0.00019270322975159773,
      "loss": 4.0018,
      "step": 39140
    },
    {
      "epoch": 0.03649413477964452,
      "grad_norm": 1.0717244148254395,
      "learning_rate": 0.00019270136542950253,
      "loss": 4.1264,
      "step": 39150
    },
    {
      "epoch": 0.036503456397723615,
      "grad_norm": 1.3061492443084717,
      "learning_rate": 0.00019269950110740734,
      "loss": 4.0776,
      "step": 39160
    },
    {
      "epoch": 0.03651277801580271,
      "grad_norm": 0.9649725556373596,
      "learning_rate": 0.00019269763678531214,
      "loss": 4.0688,
      "step": 39170
    },
    {
      "epoch": 0.0365220996338818,
      "grad_norm": 1.194291591644287,
      "learning_rate": 0.00019269577246321692,
      "loss": 4.2773,
      "step": 39180
    },
    {
      "epoch": 0.03653142125196089,
      "grad_norm": 1.0688376426696777,
      "learning_rate": 0.00019269390814112173,
      "loss": 4.0999,
      "step": 39190
    },
    {
      "epoch": 0.03654074287003998,
      "grad_norm": 0.8982643485069275,
      "learning_rate": 0.00019269204381902654,
      "loss": 4.2118,
      "step": 39200
    },
    {
      "epoch": 0.03655006448811907,
      "grad_norm": 0.9316972494125366,
      "learning_rate": 0.00019269017949693134,
      "loss": 4.2897,
      "step": 39210
    },
    {
      "epoch": 0.03655938610619817,
      "grad_norm": 1.0814896821975708,
      "learning_rate": 0.00019268831517483615,
      "loss": 3.9672,
      "step": 39220
    },
    {
      "epoch": 0.036568707724277255,
      "grad_norm": 1.0388222932815552,
      "learning_rate": 0.00019268645085274093,
      "loss": 4.0041,
      "step": 39230
    },
    {
      "epoch": 0.03657802934235635,
      "grad_norm": 1.233129620552063,
      "learning_rate": 0.00019268458653064573,
      "loss": 3.8986,
      "step": 39240
    },
    {
      "epoch": 0.036587350960435444,
      "grad_norm": 1.2364407777786255,
      "learning_rate": 0.00019268272220855051,
      "loss": 4.2015,
      "step": 39250
    },
    {
      "epoch": 0.03659667257851453,
      "grad_norm": 1.185794711112976,
      "learning_rate": 0.00019268085788645535,
      "loss": 4.1237,
      "step": 39260
    },
    {
      "epoch": 0.036605994196593626,
      "grad_norm": 0.9794387817382812,
      "learning_rate": 0.00019267899356436015,
      "loss": 3.9261,
      "step": 39270
    },
    {
      "epoch": 0.036615315814672714,
      "grad_norm": 1.2820924520492554,
      "learning_rate": 0.00019267712924226493,
      "loss": 4.2621,
      "step": 39280
    },
    {
      "epoch": 0.03662463743275181,
      "grad_norm": 1.3766899108886719,
      "learning_rate": 0.00019267526492016974,
      "loss": 4.2592,
      "step": 39290
    },
    {
      "epoch": 0.0366339590508309,
      "grad_norm": 1.1560509204864502,
      "learning_rate": 0.00019267340059807452,
      "loss": 4.1825,
      "step": 39300
    },
    {
      "epoch": 0.03664328066890999,
      "grad_norm": 1.0488626956939697,
      "learning_rate": 0.00019267153627597932,
      "loss": 4.1335,
      "step": 39310
    },
    {
      "epoch": 0.036652602286989085,
      "grad_norm": 1.0607620477676392,
      "learning_rate": 0.00019266967195388416,
      "loss": 3.9644,
      "step": 39320
    },
    {
      "epoch": 0.03666192390506817,
      "grad_norm": 1.1451218128204346,
      "learning_rate": 0.00019266780763178894,
      "loss": 3.9548,
      "step": 39330
    },
    {
      "epoch": 0.03667124552314727,
      "grad_norm": 0.9877902865409851,
      "learning_rate": 0.00019266594330969374,
      "loss": 3.9771,
      "step": 39340
    },
    {
      "epoch": 0.03668056714122636,
      "grad_norm": 0.9519314169883728,
      "learning_rate": 0.00019266407898759852,
      "loss": 4.099,
      "step": 39350
    },
    {
      "epoch": 0.03668988875930545,
      "grad_norm": 1.0780771970748901,
      "learning_rate": 0.00019266221466550333,
      "loss": 4.0295,
      "step": 39360
    },
    {
      "epoch": 0.03669921037738454,
      "grad_norm": 1.0707323551177979,
      "learning_rate": 0.00019266035034340814,
      "loss": 3.9757,
      "step": 39370
    },
    {
      "epoch": 0.03670853199546364,
      "grad_norm": 1.7064807415008545,
      "learning_rate": 0.00019265848602131294,
      "loss": 4.0642,
      "step": 39380
    },
    {
      "epoch": 0.036717853613542725,
      "grad_norm": 1.04753839969635,
      "learning_rate": 0.00019265662169921775,
      "loss": 3.9464,
      "step": 39390
    },
    {
      "epoch": 0.03672717523162182,
      "grad_norm": 0.925936222076416,
      "learning_rate": 0.00019265475737712253,
      "loss": 4.2032,
      "step": 39400
    },
    {
      "epoch": 0.03673649684970091,
      "grad_norm": 1.181348443031311,
      "learning_rate": 0.00019265289305502733,
      "loss": 4.0487,
      "step": 39410
    },
    {
      "epoch": 0.03674581846778,
      "grad_norm": 1.0221325159072876,
      "learning_rate": 0.00019265102873293214,
      "loss": 4.053,
      "step": 39420
    },
    {
      "epoch": 0.036755140085859096,
      "grad_norm": 1.1047879457473755,
      "learning_rate": 0.00019264916441083695,
      "loss": 3.9664,
      "step": 39430
    },
    {
      "epoch": 0.036764461703938184,
      "grad_norm": 0.9867522716522217,
      "learning_rate": 0.00019264730008874175,
      "loss": 4.1532,
      "step": 39440
    },
    {
      "epoch": 0.03677378332201728,
      "grad_norm": 1.2413461208343506,
      "learning_rate": 0.00019264543576664653,
      "loss": 4.0548,
      "step": 39450
    },
    {
      "epoch": 0.036783104940096366,
      "grad_norm": 0.8642893433570862,
      "learning_rate": 0.00019264357144455134,
      "loss": 4.1481,
      "step": 39460
    },
    {
      "epoch": 0.03679242655817546,
      "grad_norm": 1.2273167371749878,
      "learning_rate": 0.00019264170712245614,
      "loss": 4.0836,
      "step": 39470
    },
    {
      "epoch": 0.036801748176254555,
      "grad_norm": 1.107600450515747,
      "learning_rate": 0.00019263984280036092,
      "loss": 4.0899,
      "step": 39480
    },
    {
      "epoch": 0.03681106979433364,
      "grad_norm": 1.0725640058517456,
      "learning_rate": 0.00019263797847826576,
      "loss": 4.1454,
      "step": 39490
    },
    {
      "epoch": 0.03682039141241274,
      "grad_norm": 1.159800410270691,
      "learning_rate": 0.00019263611415617054,
      "loss": 4.0964,
      "step": 39500
    },
    {
      "epoch": 0.03682971303049183,
      "grad_norm": 1.1512762308120728,
      "learning_rate": 0.00019263424983407534,
      "loss": 4.2331,
      "step": 39510
    },
    {
      "epoch": 0.03683903464857092,
      "grad_norm": 1.3117685317993164,
      "learning_rate": 0.00019263238551198015,
      "loss": 3.901,
      "step": 39520
    },
    {
      "epoch": 0.03684835626665001,
      "grad_norm": 1.3577008247375488,
      "learning_rate": 0.00019263052118988493,
      "loss": 4.1497,
      "step": 39530
    },
    {
      "epoch": 0.0368576778847291,
      "grad_norm": 1.047817587852478,
      "learning_rate": 0.00019262865686778973,
      "loss": 4.0072,
      "step": 39540
    },
    {
      "epoch": 0.036866999502808195,
      "grad_norm": 1.0553045272827148,
      "learning_rate": 0.00019262679254569454,
      "loss": 4.2596,
      "step": 39550
    },
    {
      "epoch": 0.03687632112088729,
      "grad_norm": 1.1183252334594727,
      "learning_rate": 0.00019262492822359935,
      "loss": 4.3296,
      "step": 39560
    },
    {
      "epoch": 0.03688564273896638,
      "grad_norm": 1.1432100534439087,
      "learning_rate": 0.00019262306390150415,
      "loss": 3.9839,
      "step": 39570
    },
    {
      "epoch": 0.03689496435704547,
      "grad_norm": 1.1059844493865967,
      "learning_rate": 0.00019262119957940893,
      "loss": 3.9479,
      "step": 39580
    },
    {
      "epoch": 0.036904285975124566,
      "grad_norm": 1.221907138824463,
      "learning_rate": 0.00019261933525731374,
      "loss": 4.04,
      "step": 39590
    },
    {
      "epoch": 0.036913607593203654,
      "grad_norm": 1.126650333404541,
      "learning_rate": 0.00019261747093521855,
      "loss": 4.1749,
      "step": 39600
    },
    {
      "epoch": 0.03692292921128275,
      "grad_norm": 1.2733274698257446,
      "learning_rate": 0.00019261560661312335,
      "loss": 3.9233,
      "step": 39610
    },
    {
      "epoch": 0.036932250829361836,
      "grad_norm": 1.1110692024230957,
      "learning_rate": 0.00019261374229102816,
      "loss": 4.0548,
      "step": 39620
    },
    {
      "epoch": 0.03694157244744093,
      "grad_norm": 1.2457387447357178,
      "learning_rate": 0.00019261187796893294,
      "loss": 4.1534,
      "step": 39630
    },
    {
      "epoch": 0.036950894065520025,
      "grad_norm": 1.0221339464187622,
      "learning_rate": 0.00019261001364683774,
      "loss": 4.2102,
      "step": 39640
    },
    {
      "epoch": 0.03696021568359911,
      "grad_norm": 0.9672004580497742,
      "learning_rate": 0.00019260814932474255,
      "loss": 4.2402,
      "step": 39650
    },
    {
      "epoch": 0.03696953730167821,
      "grad_norm": 0.9094820022583008,
      "learning_rate": 0.00019260628500264736,
      "loss": 4.1297,
      "step": 39660
    },
    {
      "epoch": 0.036978858919757294,
      "grad_norm": 1.3623133897781372,
      "learning_rate": 0.00019260442068055216,
      "loss": 4.0094,
      "step": 39670
    },
    {
      "epoch": 0.03698818053783639,
      "grad_norm": 0.9263396263122559,
      "learning_rate": 0.00019260255635845694,
      "loss": 4.076,
      "step": 39680
    },
    {
      "epoch": 0.03699750215591548,
      "grad_norm": 1.1157037019729614,
      "learning_rate": 0.00019260069203636175,
      "loss": 4.0345,
      "step": 39690
    },
    {
      "epoch": 0.03700682377399457,
      "grad_norm": 1.0037522315979004,
      "learning_rate": 0.00019259882771426653,
      "loss": 4.1827,
      "step": 39700
    },
    {
      "epoch": 0.037016145392073665,
      "grad_norm": 1.3378829956054688,
      "learning_rate": 0.00019259696339217136,
      "loss": 4.0781,
      "step": 39710
    },
    {
      "epoch": 0.03702546701015276,
      "grad_norm": 0.8762807250022888,
      "learning_rate": 0.00019259509907007617,
      "loss": 4.0789,
      "step": 39720
    },
    {
      "epoch": 0.03703478862823185,
      "grad_norm": 1.0886882543563843,
      "learning_rate": 0.00019259323474798095,
      "loss": 4.0842,
      "step": 39730
    },
    {
      "epoch": 0.03704411024631094,
      "grad_norm": 0.9643123149871826,
      "learning_rate": 0.00019259137042588575,
      "loss": 4.0773,
      "step": 39740
    },
    {
      "epoch": 0.03705343186439003,
      "grad_norm": 1.1102535724639893,
      "learning_rate": 0.00019258950610379053,
      "loss": 4.2452,
      "step": 39750
    },
    {
      "epoch": 0.037062753482469124,
      "grad_norm": 1.1882902383804321,
      "learning_rate": 0.00019258764178169534,
      "loss": 4.1373,
      "step": 39760
    },
    {
      "epoch": 0.03707207510054822,
      "grad_norm": 0.9426063299179077,
      "learning_rate": 0.00019258577745960017,
      "loss": 4.113,
      "step": 39770
    },
    {
      "epoch": 0.037081396718627306,
      "grad_norm": 1.1491249799728394,
      "learning_rate": 0.00019258391313750495,
      "loss": 3.9634,
      "step": 39780
    },
    {
      "epoch": 0.0370907183367064,
      "grad_norm": 1.0440428256988525,
      "learning_rate": 0.00019258204881540976,
      "loss": 3.9623,
      "step": 39790
    },
    {
      "epoch": 0.03710003995478549,
      "grad_norm": 1.1281213760375977,
      "learning_rate": 0.00019258018449331454,
      "loss": 4.1985,
      "step": 39800
    },
    {
      "epoch": 0.03710936157286458,
      "grad_norm": 1.188012719154358,
      "learning_rate": 0.00019257832017121934,
      "loss": 4.0618,
      "step": 39810
    },
    {
      "epoch": 0.03711868319094368,
      "grad_norm": 1.0811972618103027,
      "learning_rate": 0.00019257645584912415,
      "loss": 4.1465,
      "step": 39820
    },
    {
      "epoch": 0.037128004809022765,
      "grad_norm": 1.0669013261795044,
      "learning_rate": 0.00019257459152702895,
      "loss": 3.8197,
      "step": 39830
    },
    {
      "epoch": 0.03713732642710186,
      "grad_norm": 1.0709190368652344,
      "learning_rate": 0.00019257272720493376,
      "loss": 4.1995,
      "step": 39840
    },
    {
      "epoch": 0.037146648045180954,
      "grad_norm": 1.321731686592102,
      "learning_rate": 0.00019257086288283854,
      "loss": 4.1673,
      "step": 39850
    },
    {
      "epoch": 0.03715596966326004,
      "grad_norm": 1.105955719947815,
      "learning_rate": 0.00019256899856074335,
      "loss": 4.0779,
      "step": 39860
    },
    {
      "epoch": 0.037165291281339136,
      "grad_norm": 1.0385268926620483,
      "learning_rate": 0.00019256713423864815,
      "loss": 4.2004,
      "step": 39870
    },
    {
      "epoch": 0.03717461289941822,
      "grad_norm": 1.0198862552642822,
      "learning_rate": 0.00019256526991655296,
      "loss": 4.0823,
      "step": 39880
    },
    {
      "epoch": 0.03718393451749732,
      "grad_norm": 1.0722472667694092,
      "learning_rate": 0.00019256340559445777,
      "loss": 4.1019,
      "step": 39890
    },
    {
      "epoch": 0.03719325613557641,
      "grad_norm": 1.1122469902038574,
      "learning_rate": 0.00019256154127236254,
      "loss": 4.0149,
      "step": 39900
    },
    {
      "epoch": 0.0372025777536555,
      "grad_norm": 1.2172355651855469,
      "learning_rate": 0.00019255967695026735,
      "loss": 4.1104,
      "step": 39910
    },
    {
      "epoch": 0.037211899371734594,
      "grad_norm": 1.0808559656143188,
      "learning_rate": 0.00019255781262817216,
      "loss": 4.1133,
      "step": 39920
    },
    {
      "epoch": 0.03722122098981369,
      "grad_norm": 1.0808743238449097,
      "learning_rate": 0.00019255594830607694,
      "loss": 4.0914,
      "step": 39930
    },
    {
      "epoch": 0.037230542607892776,
      "grad_norm": 1.2068274021148682,
      "learning_rate": 0.00019255408398398177,
      "loss": 4.0616,
      "step": 39940
    },
    {
      "epoch": 0.03723986422597187,
      "grad_norm": 1.082275629043579,
      "learning_rate": 0.00019255221966188655,
      "loss": 4.2535,
      "step": 39950
    },
    {
      "epoch": 0.03724918584405096,
      "grad_norm": 1.120009183883667,
      "learning_rate": 0.00019255035533979136,
      "loss": 3.9919,
      "step": 39960
    },
    {
      "epoch": 0.03725850746213005,
      "grad_norm": 0.9155793190002441,
      "learning_rate": 0.00019254849101769616,
      "loss": 4.0616,
      "step": 39970
    },
    {
      "epoch": 0.03726782908020915,
      "grad_norm": 0.9312489032745361,
      "learning_rate": 0.00019254662669560094,
      "loss": 3.9894,
      "step": 39980
    },
    {
      "epoch": 0.037277150698288235,
      "grad_norm": 1.1335554122924805,
      "learning_rate": 0.00019254476237350575,
      "loss": 4.0119,
      "step": 39990
    },
    {
      "epoch": 0.03728647231636733,
      "grad_norm": 1.105095386505127,
      "learning_rate": 0.00019254289805141055,
      "loss": 4.1221,
      "step": 40000
    },
    {
      "epoch": 0.03729579393444642,
      "grad_norm": 1.1299716234207153,
      "learning_rate": 0.00019254103372931536,
      "loss": 4.0101,
      "step": 40010
    },
    {
      "epoch": 0.03730511555252551,
      "grad_norm": 1.0562375783920288,
      "learning_rate": 0.00019253916940722017,
      "loss": 4.1276,
      "step": 40020
    },
    {
      "epoch": 0.037314437170604606,
      "grad_norm": 1.2078243494033813,
      "learning_rate": 0.00019253730508512495,
      "loss": 4.1874,
      "step": 40030
    },
    {
      "epoch": 0.03732375878868369,
      "grad_norm": 1.0529720783233643,
      "learning_rate": 0.00019253544076302975,
      "loss": 4.2048,
      "step": 40040
    },
    {
      "epoch": 0.03733308040676279,
      "grad_norm": 1.056134581565857,
      "learning_rate": 0.00019253357644093456,
      "loss": 3.9749,
      "step": 40050
    },
    {
      "epoch": 0.03734240202484188,
      "grad_norm": 1.2226905822753906,
      "learning_rate": 0.00019253171211883936,
      "loss": 4.078,
      "step": 40060
    },
    {
      "epoch": 0.03735172364292097,
      "grad_norm": 1.2039107084274292,
      "learning_rate": 0.00019252984779674417,
      "loss": 4.0558,
      "step": 40070
    },
    {
      "epoch": 0.037361045261000064,
      "grad_norm": 1.0710933208465576,
      "learning_rate": 0.00019252798347464895,
      "loss": 4.0902,
      "step": 40080
    },
    {
      "epoch": 0.03737036687907915,
      "grad_norm": 1.008724331855774,
      "learning_rate": 0.00019252611915255376,
      "loss": 4.1062,
      "step": 40090
    },
    {
      "epoch": 0.037379688497158246,
      "grad_norm": 1.1888113021850586,
      "learning_rate": 0.00019252425483045856,
      "loss": 4.1096,
      "step": 40100
    },
    {
      "epoch": 0.03738901011523734,
      "grad_norm": 1.1243171691894531,
      "learning_rate": 0.00019252239050836337,
      "loss": 4.0575,
      "step": 40110
    },
    {
      "epoch": 0.03739833173331643,
      "grad_norm": 1.1862601041793823,
      "learning_rate": 0.00019252052618626817,
      "loss": 4.1645,
      "step": 40120
    },
    {
      "epoch": 0.03740765335139552,
      "grad_norm": 1.2097885608673096,
      "learning_rate": 0.00019251866186417295,
      "loss": 4.096,
      "step": 40130
    },
    {
      "epoch": 0.03741697496947461,
      "grad_norm": 1.1293683052062988,
      "learning_rate": 0.00019251679754207776,
      "loss": 3.9586,
      "step": 40140
    },
    {
      "epoch": 0.037426296587553705,
      "grad_norm": 0.9615786671638489,
      "learning_rate": 0.00019251493321998254,
      "loss": 3.9577,
      "step": 40150
    },
    {
      "epoch": 0.0374356182056328,
      "grad_norm": 1.177075982093811,
      "learning_rate": 0.00019251306889788737,
      "loss": 4.0898,
      "step": 40160
    },
    {
      "epoch": 0.03744493982371189,
      "grad_norm": 1.1683927774429321,
      "learning_rate": 0.00019251120457579218,
      "loss": 3.8716,
      "step": 40170
    },
    {
      "epoch": 0.03745426144179098,
      "grad_norm": 1.193228840827942,
      "learning_rate": 0.00019250934025369696,
      "loss": 4.1032,
      "step": 40180
    },
    {
      "epoch": 0.037463583059870076,
      "grad_norm": 0.9564664363861084,
      "learning_rate": 0.00019250747593160176,
      "loss": 4.1996,
      "step": 40190
    },
    {
      "epoch": 0.03747290467794916,
      "grad_norm": 1.1865912675857544,
      "learning_rate": 0.00019250561160950654,
      "loss": 4.3309,
      "step": 40200
    },
    {
      "epoch": 0.03748222629602826,
      "grad_norm": 1.1742639541625977,
      "learning_rate": 0.00019250374728741135,
      "loss": 4.0488,
      "step": 40210
    },
    {
      "epoch": 0.037491547914107345,
      "grad_norm": 1.2715983390808105,
      "learning_rate": 0.00019250188296531618,
      "loss": 3.8481,
      "step": 40220
    },
    {
      "epoch": 0.03750086953218644,
      "grad_norm": 1.002479910850525,
      "learning_rate": 0.00019250001864322096,
      "loss": 4.0063,
      "step": 40230
    },
    {
      "epoch": 0.037510191150265534,
      "grad_norm": 1.0000741481781006,
      "learning_rate": 0.00019249815432112577,
      "loss": 3.9949,
      "step": 40240
    },
    {
      "epoch": 0.03751951276834462,
      "grad_norm": 1.0061521530151367,
      "learning_rate": 0.00019249628999903055,
      "loss": 3.9806,
      "step": 40250
    },
    {
      "epoch": 0.037528834386423716,
      "grad_norm": 1.1144810914993286,
      "learning_rate": 0.00019249442567693535,
      "loss": 4.1026,
      "step": 40260
    },
    {
      "epoch": 0.03753815600450281,
      "grad_norm": 1.2022637128829956,
      "learning_rate": 0.00019249256135484016,
      "loss": 4.0692,
      "step": 40270
    },
    {
      "epoch": 0.0375474776225819,
      "grad_norm": 1.1990704536437988,
      "learning_rate": 0.00019249069703274497,
      "loss": 4.0565,
      "step": 40280
    },
    {
      "epoch": 0.03755679924066099,
      "grad_norm": 0.9441627264022827,
      "learning_rate": 0.00019248883271064977,
      "loss": 4.1688,
      "step": 40290
    },
    {
      "epoch": 0.03756612085874008,
      "grad_norm": 1.0954887866973877,
      "learning_rate": 0.00019248696838855455,
      "loss": 4.1233,
      "step": 40300
    },
    {
      "epoch": 0.037575442476819175,
      "grad_norm": 1.023715615272522,
      "learning_rate": 0.00019248510406645936,
      "loss": 4.184,
      "step": 40310
    },
    {
      "epoch": 0.03758476409489827,
      "grad_norm": 1.1815105676651,
      "learning_rate": 0.00019248323974436417,
      "loss": 4.1155,
      "step": 40320
    },
    {
      "epoch": 0.03759408571297736,
      "grad_norm": 0.9595500826835632,
      "learning_rate": 0.00019248137542226897,
      "loss": 4.1703,
      "step": 40330
    },
    {
      "epoch": 0.03760340733105645,
      "grad_norm": 1.0392171144485474,
      "learning_rate": 0.00019247951110017378,
      "loss": 3.984,
      "step": 40340
    },
    {
      "epoch": 0.03761272894913554,
      "grad_norm": 0.9147299528121948,
      "learning_rate": 0.00019247764677807856,
      "loss": 4.0905,
      "step": 40350
    },
    {
      "epoch": 0.037622050567214634,
      "grad_norm": 1.0878607034683228,
      "learning_rate": 0.00019247578245598336,
      "loss": 4.2877,
      "step": 40360
    },
    {
      "epoch": 0.03763137218529373,
      "grad_norm": 0.9968376159667969,
      "learning_rate": 0.00019247391813388817,
      "loss": 4.072,
      "step": 40370
    },
    {
      "epoch": 0.037640693803372816,
      "grad_norm": 0.9112571477890015,
      "learning_rate": 0.00019247205381179298,
      "loss": 4.1672,
      "step": 40380
    },
    {
      "epoch": 0.03765001542145191,
      "grad_norm": 0.8729525208473206,
      "learning_rate": 0.00019247018948969778,
      "loss": 4.2031,
      "step": 40390
    },
    {
      "epoch": 0.037659337039531005,
      "grad_norm": 1.1989470720291138,
      "learning_rate": 0.00019246832516760256,
      "loss": 3.9042,
      "step": 40400
    },
    {
      "epoch": 0.03766865865761009,
      "grad_norm": 0.9567263722419739,
      "learning_rate": 0.00019246646084550737,
      "loss": 4.1966,
      "step": 40410
    },
    {
      "epoch": 0.03767798027568919,
      "grad_norm": 0.9854222536087036,
      "learning_rate": 0.00019246459652341215,
      "loss": 4.1563,
      "step": 40420
    },
    {
      "epoch": 0.037687301893768274,
      "grad_norm": 1.1394073963165283,
      "learning_rate": 0.00019246273220131695,
      "loss": 4.1209,
      "step": 40430
    },
    {
      "epoch": 0.03769662351184737,
      "grad_norm": 1.1292216777801514,
      "learning_rate": 0.00019246086787922176,
      "loss": 4.3392,
      "step": 40440
    },
    {
      "epoch": 0.03770594512992646,
      "grad_norm": 1.0463813543319702,
      "learning_rate": 0.00019245900355712657,
      "loss": 4.033,
      "step": 40450
    },
    {
      "epoch": 0.03771526674800555,
      "grad_norm": 1.058118224143982,
      "learning_rate": 0.00019245713923503137,
      "loss": 4.1047,
      "step": 40460
    },
    {
      "epoch": 0.037724588366084645,
      "grad_norm": 1.0985584259033203,
      "learning_rate": 0.00019245527491293615,
      "loss": 4.1024,
      "step": 40470
    },
    {
      "epoch": 0.03773390998416373,
      "grad_norm": 1.3441824913024902,
      "learning_rate": 0.00019245341059084096,
      "loss": 3.942,
      "step": 40480
    },
    {
      "epoch": 0.03774323160224283,
      "grad_norm": 1.0290908813476562,
      "learning_rate": 0.00019245154626874576,
      "loss": 4.1985,
      "step": 40490
    },
    {
      "epoch": 0.03775255322032192,
      "grad_norm": 1.1242263317108154,
      "learning_rate": 0.00019244968194665057,
      "loss": 4.0728,
      "step": 40500
    },
    {
      "epoch": 0.03776187483840101,
      "grad_norm": 1.0513190031051636,
      "learning_rate": 0.00019244781762455538,
      "loss": 4.2754,
      "step": 40510
    },
    {
      "epoch": 0.037771196456480104,
      "grad_norm": 1.2158111333847046,
      "learning_rate": 0.00019244595330246016,
      "loss": 4.1854,
      "step": 40520
    },
    {
      "epoch": 0.0377805180745592,
      "grad_norm": 1.0164122581481934,
      "learning_rate": 0.00019244408898036496,
      "loss": 4.1409,
      "step": 40530
    },
    {
      "epoch": 0.037789839692638286,
      "grad_norm": 1.0760964155197144,
      "learning_rate": 0.00019244222465826977,
      "loss": 3.952,
      "step": 40540
    },
    {
      "epoch": 0.03779916131071738,
      "grad_norm": 0.9815390110015869,
      "learning_rate": 0.00019244036033617458,
      "loss": 4.2306,
      "step": 40550
    },
    {
      "epoch": 0.03780848292879647,
      "grad_norm": 1.2239798307418823,
      "learning_rate": 0.00019243849601407938,
      "loss": 4.1078,
      "step": 40560
    },
    {
      "epoch": 0.03781780454687556,
      "grad_norm": 1.0410186052322388,
      "learning_rate": 0.00019243663169198416,
      "loss": 3.9046,
      "step": 40570
    },
    {
      "epoch": 0.03782712616495466,
      "grad_norm": 0.9849880933761597,
      "learning_rate": 0.00019243476736988897,
      "loss": 4.1197,
      "step": 40580
    },
    {
      "epoch": 0.037836447783033744,
      "grad_norm": 1.0208759307861328,
      "learning_rate": 0.00019243290304779377,
      "loss": 3.9728,
      "step": 40590
    },
    {
      "epoch": 0.03784576940111284,
      "grad_norm": 1.1420024633407593,
      "learning_rate": 0.00019243103872569855,
      "loss": 4.1043,
      "step": 40600
    },
    {
      "epoch": 0.03785509101919193,
      "grad_norm": 1.2188630104064941,
      "learning_rate": 0.00019242917440360339,
      "loss": 4.0643,
      "step": 40610
    },
    {
      "epoch": 0.03786441263727102,
      "grad_norm": 1.1792287826538086,
      "learning_rate": 0.00019242731008150817,
      "loss": 4.0949,
      "step": 40620
    },
    {
      "epoch": 0.037873734255350115,
      "grad_norm": 1.1328030824661255,
      "learning_rate": 0.00019242544575941297,
      "loss": 4.0986,
      "step": 40630
    },
    {
      "epoch": 0.0378830558734292,
      "grad_norm": 1.29794442653656,
      "learning_rate": 0.00019242358143731778,
      "loss": 4.275,
      "step": 40640
    },
    {
      "epoch": 0.0378923774915083,
      "grad_norm": 1.2690117359161377,
      "learning_rate": 0.00019242171711522256,
      "loss": 4.0106,
      "step": 40650
    },
    {
      "epoch": 0.03790169910958739,
      "grad_norm": 1.0158185958862305,
      "learning_rate": 0.00019241985279312736,
      "loss": 4.0867,
      "step": 40660
    },
    {
      "epoch": 0.03791102072766648,
      "grad_norm": 1.2738795280456543,
      "learning_rate": 0.00019241798847103217,
      "loss": 4.1236,
      "step": 40670
    },
    {
      "epoch": 0.037920342345745574,
      "grad_norm": 1.0967628955841064,
      "learning_rate": 0.00019241612414893698,
      "loss": 4.0881,
      "step": 40680
    },
    {
      "epoch": 0.03792966396382466,
      "grad_norm": 1.1986576318740845,
      "learning_rate": 0.00019241425982684178,
      "loss": 4.1373,
      "step": 40690
    },
    {
      "epoch": 0.037938985581903756,
      "grad_norm": 1.2404823303222656,
      "learning_rate": 0.00019241239550474656,
      "loss": 4.0779,
      "step": 40700
    },
    {
      "epoch": 0.03794830719998285,
      "grad_norm": 1.1346545219421387,
      "learning_rate": 0.00019241053118265137,
      "loss": 3.9648,
      "step": 40710
    },
    {
      "epoch": 0.03795762881806194,
      "grad_norm": 0.9570274949073792,
      "learning_rate": 0.00019240866686055617,
      "loss": 3.9355,
      "step": 40720
    },
    {
      "epoch": 0.03796695043614103,
      "grad_norm": 0.9998208284378052,
      "learning_rate": 0.00019240680253846098,
      "loss": 4.0616,
      "step": 40730
    },
    {
      "epoch": 0.03797627205422013,
      "grad_norm": 1.1090260744094849,
      "learning_rate": 0.0001924049382163658,
      "loss": 4.1496,
      "step": 40740
    },
    {
      "epoch": 0.037985593672299214,
      "grad_norm": 1.248882532119751,
      "learning_rate": 0.00019240307389427057,
      "loss": 3.834,
      "step": 40750
    },
    {
      "epoch": 0.03799491529037831,
      "grad_norm": 1.0899525880813599,
      "learning_rate": 0.00019240120957217537,
      "loss": 4.0785,
      "step": 40760
    },
    {
      "epoch": 0.038004236908457396,
      "grad_norm": 0.9912641644477844,
      "learning_rate": 0.00019239934525008018,
      "loss": 4.2996,
      "step": 40770
    },
    {
      "epoch": 0.03801355852653649,
      "grad_norm": 1.1602658033370972,
      "learning_rate": 0.00019239748092798498,
      "loss": 4.049,
      "step": 40780
    },
    {
      "epoch": 0.038022880144615585,
      "grad_norm": 1.0054293870925903,
      "learning_rate": 0.0001923956166058898,
      "loss": 4.1096,
      "step": 40790
    },
    {
      "epoch": 0.03803220176269467,
      "grad_norm": 0.9454019069671631,
      "learning_rate": 0.00019239375228379457,
      "loss": 4.1751,
      "step": 40800
    },
    {
      "epoch": 0.03804152338077377,
      "grad_norm": 0.9638948440551758,
      "learning_rate": 0.00019239188796169938,
      "loss": 3.9751,
      "step": 40810
    },
    {
      "epoch": 0.038050844998852855,
      "grad_norm": 1.2645467519760132,
      "learning_rate": 0.00019239002363960416,
      "loss": 4.0581,
      "step": 40820
    },
    {
      "epoch": 0.03806016661693195,
      "grad_norm": 1.0194077491760254,
      "learning_rate": 0.000192388159317509,
      "loss": 4.1421,
      "step": 40830
    },
    {
      "epoch": 0.038069488235011044,
      "grad_norm": 1.212261438369751,
      "learning_rate": 0.0001923862949954138,
      "loss": 4.0528,
      "step": 40840
    },
    {
      "epoch": 0.03807880985309013,
      "grad_norm": 1.320901870727539,
      "learning_rate": 0.00019238443067331857,
      "loss": 4.088,
      "step": 40850
    },
    {
      "epoch": 0.038088131471169226,
      "grad_norm": 1.0687611103057861,
      "learning_rate": 0.00019238256635122338,
      "loss": 3.9095,
      "step": 40860
    },
    {
      "epoch": 0.03809745308924832,
      "grad_norm": 1.1938817501068115,
      "learning_rate": 0.00019238070202912816,
      "loss": 4.0009,
      "step": 40870
    },
    {
      "epoch": 0.03810677470732741,
      "grad_norm": 1.0755447149276733,
      "learning_rate": 0.00019237883770703297,
      "loss": 3.9763,
      "step": 40880
    },
    {
      "epoch": 0.0381160963254065,
      "grad_norm": 1.2436919212341309,
      "learning_rate": 0.0001923769733849378,
      "loss": 4.0807,
      "step": 40890
    },
    {
      "epoch": 0.03812541794348559,
      "grad_norm": 1.128350853919983,
      "learning_rate": 0.00019237510906284258,
      "loss": 4.2086,
      "step": 40900
    },
    {
      "epoch": 0.038134739561564684,
      "grad_norm": 1.0040346384048462,
      "learning_rate": 0.00019237324474074739,
      "loss": 4.0488,
      "step": 40910
    },
    {
      "epoch": 0.03814406117964378,
      "grad_norm": 1.130712628364563,
      "learning_rate": 0.00019237138041865216,
      "loss": 4.1555,
      "step": 40920
    },
    {
      "epoch": 0.038153382797722867,
      "grad_norm": 1.0129069089889526,
      "learning_rate": 0.00019236951609655697,
      "loss": 3.9636,
      "step": 40930
    },
    {
      "epoch": 0.03816270441580196,
      "grad_norm": 1.1103013753890991,
      "learning_rate": 0.00019236765177446178,
      "loss": 4.1489,
      "step": 40940
    },
    {
      "epoch": 0.038172026033881055,
      "grad_norm": 1.1453050374984741,
      "learning_rate": 0.00019236578745236658,
      "loss": 4.0955,
      "step": 40950
    },
    {
      "epoch": 0.03818134765196014,
      "grad_norm": 1.0954312086105347,
      "learning_rate": 0.0001923639231302714,
      "loss": 4.1032,
      "step": 40960
    },
    {
      "epoch": 0.03819066927003924,
      "grad_norm": 1.0967400074005127,
      "learning_rate": 0.00019236205880817617,
      "loss": 4.1818,
      "step": 40970
    },
    {
      "epoch": 0.038199990888118325,
      "grad_norm": 1.199920654296875,
      "learning_rate": 0.00019236019448608098,
      "loss": 4.0058,
      "step": 40980
    },
    {
      "epoch": 0.03820931250619742,
      "grad_norm": 1.1688036918640137,
      "learning_rate": 0.00019235833016398578,
      "loss": 4.0213,
      "step": 40990
    },
    {
      "epoch": 0.038218634124276514,
      "grad_norm": 1.1066886186599731,
      "learning_rate": 0.0001923564658418906,
      "loss": 4.0339,
      "step": 41000
    },
    {
      "epoch": 0.0382279557423556,
      "grad_norm": 1.1739071607589722,
      "learning_rate": 0.0001923546015197954,
      "loss": 4.0008,
      "step": 41010
    },
    {
      "epoch": 0.038237277360434696,
      "grad_norm": 0.9364327192306519,
      "learning_rate": 0.00019235273719770017,
      "loss": 4.0527,
      "step": 41020
    },
    {
      "epoch": 0.038246598978513784,
      "grad_norm": 0.9832871556282043,
      "learning_rate": 0.00019235087287560498,
      "loss": 4.0393,
      "step": 41030
    },
    {
      "epoch": 0.03825592059659288,
      "grad_norm": 1.277936339378357,
      "learning_rate": 0.00019234900855350979,
      "loss": 3.9826,
      "step": 41040
    },
    {
      "epoch": 0.03826524221467197,
      "grad_norm": 1.116632342338562,
      "learning_rate": 0.00019234714423141457,
      "loss": 4.0937,
      "step": 41050
    },
    {
      "epoch": 0.03827456383275106,
      "grad_norm": 1.097436785697937,
      "learning_rate": 0.0001923452799093194,
      "loss": 4.2081,
      "step": 41060
    },
    {
      "epoch": 0.038283885450830155,
      "grad_norm": 1.1514101028442383,
      "learning_rate": 0.00019234341558722418,
      "loss": 4.1388,
      "step": 41070
    },
    {
      "epoch": 0.03829320706890925,
      "grad_norm": 0.9343128800392151,
      "learning_rate": 0.00019234155126512898,
      "loss": 4.0572,
      "step": 41080
    },
    {
      "epoch": 0.03830252868698834,
      "grad_norm": 0.968866765499115,
      "learning_rate": 0.0001923396869430338,
      "loss": 3.9694,
      "step": 41090
    },
    {
      "epoch": 0.03831185030506743,
      "grad_norm": 1.2432528734207153,
      "learning_rate": 0.00019233782262093857,
      "loss": 4.0942,
      "step": 41100
    },
    {
      "epoch": 0.03832117192314652,
      "grad_norm": 1.0969525575637817,
      "learning_rate": 0.00019233595829884338,
      "loss": 4.094,
      "step": 41110
    },
    {
      "epoch": 0.03833049354122561,
      "grad_norm": 1.114329218864441,
      "learning_rate": 0.00019233409397674818,
      "loss": 4.0793,
      "step": 41120
    },
    {
      "epoch": 0.03833981515930471,
      "grad_norm": 1.2100253105163574,
      "learning_rate": 0.000192332229654653,
      "loss": 4.2325,
      "step": 41130
    },
    {
      "epoch": 0.038349136777383795,
      "grad_norm": 0.9319127202033997,
      "learning_rate": 0.0001923303653325578,
      "loss": 3.9833,
      "step": 41140
    },
    {
      "epoch": 0.03835845839546289,
      "grad_norm": 1.1963117122650146,
      "learning_rate": 0.00019232850101046257,
      "loss": 4.0687,
      "step": 41150
    },
    {
      "epoch": 0.03836778001354198,
      "grad_norm": 1.2115199565887451,
      "learning_rate": 0.00019232663668836738,
      "loss": 3.9312,
      "step": 41160
    },
    {
      "epoch": 0.03837710163162107,
      "grad_norm": 1.015753149986267,
      "learning_rate": 0.0001923247723662722,
      "loss": 4.1138,
      "step": 41170
    },
    {
      "epoch": 0.038386423249700166,
      "grad_norm": 0.9387322068214417,
      "learning_rate": 0.000192322908044177,
      "loss": 4.0414,
      "step": 41180
    },
    {
      "epoch": 0.038395744867779254,
      "grad_norm": 1.1709504127502441,
      "learning_rate": 0.0001923210437220818,
      "loss": 4.3251,
      "step": 41190
    },
    {
      "epoch": 0.03840506648585835,
      "grad_norm": 1.1122246980667114,
      "learning_rate": 0.00019231917939998658,
      "loss": 4.2206,
      "step": 41200
    },
    {
      "epoch": 0.03841438810393744,
      "grad_norm": 1.0870330333709717,
      "learning_rate": 0.00019231731507789138,
      "loss": 4.3527,
      "step": 41210
    },
    {
      "epoch": 0.03842370972201653,
      "grad_norm": 1.0672420263290405,
      "learning_rate": 0.0001923154507557962,
      "loss": 4.2102,
      "step": 41220
    },
    {
      "epoch": 0.038433031340095625,
      "grad_norm": 0.9543050527572632,
      "learning_rate": 0.000192313586433701,
      "loss": 4.1212,
      "step": 41230
    },
    {
      "epoch": 0.03844235295817471,
      "grad_norm": 1.207343339920044,
      "learning_rate": 0.0001923117221116058,
      "loss": 4.0301,
      "step": 41240
    },
    {
      "epoch": 0.03845167457625381,
      "grad_norm": 0.9569122195243835,
      "learning_rate": 0.00019230985778951058,
      "loss": 4.1354,
      "step": 41250
    },
    {
      "epoch": 0.0384609961943329,
      "grad_norm": 0.9593632221221924,
      "learning_rate": 0.0001923079934674154,
      "loss": 4.1039,
      "step": 41260
    },
    {
      "epoch": 0.03847031781241199,
      "grad_norm": 0.9343345165252686,
      "learning_rate": 0.00019230612914532017,
      "loss": 4.0479,
      "step": 41270
    },
    {
      "epoch": 0.03847963943049108,
      "grad_norm": 1.2653484344482422,
      "learning_rate": 0.000192304264823225,
      "loss": 3.9844,
      "step": 41280
    },
    {
      "epoch": 0.03848896104857018,
      "grad_norm": 1.1937435865402222,
      "learning_rate": 0.0001923024005011298,
      "loss": 4.1144,
      "step": 41290
    },
    {
      "epoch": 0.038498282666649265,
      "grad_norm": 0.9046143889427185,
      "learning_rate": 0.0001923005361790346,
      "loss": 4.0289,
      "step": 41300
    },
    {
      "epoch": 0.03850760428472836,
      "grad_norm": 1.159914255142212,
      "learning_rate": 0.0001922986718569394,
      "loss": 4.0216,
      "step": 41310
    },
    {
      "epoch": 0.03851692590280745,
      "grad_norm": 1.1360303163528442,
      "learning_rate": 0.00019229680753484417,
      "loss": 4.2474,
      "step": 41320
    },
    {
      "epoch": 0.03852624752088654,
      "grad_norm": 1.097258448600769,
      "learning_rate": 0.00019229494321274898,
      "loss": 4.1987,
      "step": 41330
    },
    {
      "epoch": 0.038535569138965636,
      "grad_norm": 1.020082712173462,
      "learning_rate": 0.0001922930788906538,
      "loss": 3.9782,
      "step": 41340
    },
    {
      "epoch": 0.038544890757044724,
      "grad_norm": 1.0331441164016724,
      "learning_rate": 0.0001922912145685586,
      "loss": 4.0513,
      "step": 41350
    },
    {
      "epoch": 0.03855421237512382,
      "grad_norm": 0.8815746307373047,
      "learning_rate": 0.0001922893502464634,
      "loss": 4.2768,
      "step": 41360
    },
    {
      "epoch": 0.038563533993202906,
      "grad_norm": 1.1352064609527588,
      "learning_rate": 0.00019228748592436818,
      "loss": 4.1504,
      "step": 41370
    },
    {
      "epoch": 0.038572855611282,
      "grad_norm": 1.1003061532974243,
      "learning_rate": 0.00019228562160227298,
      "loss": 4.0713,
      "step": 41380
    },
    {
      "epoch": 0.038582177229361095,
      "grad_norm": 1.098407506942749,
      "learning_rate": 0.0001922837572801778,
      "loss": 4.0915,
      "step": 41390
    },
    {
      "epoch": 0.03859149884744018,
      "grad_norm": 1.248635172843933,
      "learning_rate": 0.0001922818929580826,
      "loss": 3.9706,
      "step": 41400
    },
    {
      "epoch": 0.03860082046551928,
      "grad_norm": 0.9422776103019714,
      "learning_rate": 0.0001922800286359874,
      "loss": 4.0682,
      "step": 41410
    },
    {
      "epoch": 0.03861014208359837,
      "grad_norm": 1.1512635946273804,
      "learning_rate": 0.00019227816431389218,
      "loss": 4.1221,
      "step": 41420
    },
    {
      "epoch": 0.03861946370167746,
      "grad_norm": 1.027915358543396,
      "learning_rate": 0.000192276299991797,
      "loss": 4.1648,
      "step": 41430
    },
    {
      "epoch": 0.03862878531975655,
      "grad_norm": 0.9854348301887512,
      "learning_rate": 0.0001922744356697018,
      "loss": 3.9944,
      "step": 41440
    },
    {
      "epoch": 0.03863810693783564,
      "grad_norm": 1.2607437372207642,
      "learning_rate": 0.0001922725713476066,
      "loss": 4.0492,
      "step": 41450
    },
    {
      "epoch": 0.038647428555914735,
      "grad_norm": 1.078262209892273,
      "learning_rate": 0.0001922707070255114,
      "loss": 4.0743,
      "step": 41460
    },
    {
      "epoch": 0.03865675017399383,
      "grad_norm": 1.1828250885009766,
      "learning_rate": 0.00019226884270341619,
      "loss": 4.0456,
      "step": 41470
    },
    {
      "epoch": 0.03866607179207292,
      "grad_norm": 1.3467278480529785,
      "learning_rate": 0.000192266978381321,
      "loss": 3.8741,
      "step": 41480
    },
    {
      "epoch": 0.03867539341015201,
      "grad_norm": 1.2782721519470215,
      "learning_rate": 0.0001922651140592258,
      "loss": 4.0926,
      "step": 41490
    },
    {
      "epoch": 0.038684715028231106,
      "grad_norm": 1.365904688835144,
      "learning_rate": 0.00019226324973713058,
      "loss": 4.1916,
      "step": 41500
    },
    {
      "epoch": 0.038694036646310194,
      "grad_norm": 1.1289511919021606,
      "learning_rate": 0.0001922613854150354,
      "loss": 4.016,
      "step": 41510
    },
    {
      "epoch": 0.03870335826438929,
      "grad_norm": 1.1011450290679932,
      "learning_rate": 0.0001922595210929402,
      "loss": 4.0341,
      "step": 41520
    },
    {
      "epoch": 0.038712679882468376,
      "grad_norm": 1.1298778057098389,
      "learning_rate": 0.000192257656770845,
      "loss": 4.12,
      "step": 41530
    },
    {
      "epoch": 0.03872200150054747,
      "grad_norm": 1.3928098678588867,
      "learning_rate": 0.0001922557924487498,
      "loss": 3.943,
      "step": 41540
    },
    {
      "epoch": 0.038731323118626565,
      "grad_norm": 1.1416130065917969,
      "learning_rate": 0.00019225392812665458,
      "loss": 3.9909,
      "step": 41550
    },
    {
      "epoch": 0.03874064473670565,
      "grad_norm": 1.1745169162750244,
      "learning_rate": 0.0001922520638045594,
      "loss": 4.0404,
      "step": 41560
    },
    {
      "epoch": 0.03874996635478475,
      "grad_norm": 0.965463399887085,
      "learning_rate": 0.0001922501994824642,
      "loss": 4.269,
      "step": 41570
    },
    {
      "epoch": 0.038759287972863835,
      "grad_norm": 1.0472350120544434,
      "learning_rate": 0.000192248335160369,
      "loss": 4.0966,
      "step": 41580
    },
    {
      "epoch": 0.03876860959094293,
      "grad_norm": 1.1670883893966675,
      "learning_rate": 0.0001922464708382738,
      "loss": 3.9805,
      "step": 41590
    },
    {
      "epoch": 0.038777931209022024,
      "grad_norm": 1.0154129266738892,
      "learning_rate": 0.0001922446065161786,
      "loss": 4.2179,
      "step": 41600
    },
    {
      "epoch": 0.03878725282710111,
      "grad_norm": 1.2954118251800537,
      "learning_rate": 0.0001922427421940834,
      "loss": 4.0441,
      "step": 41610
    },
    {
      "epoch": 0.038796574445180206,
      "grad_norm": 1.016178011894226,
      "learning_rate": 0.0001922408778719882,
      "loss": 4.0607,
      "step": 41620
    },
    {
      "epoch": 0.0388058960632593,
      "grad_norm": 0.9924085736274719,
      "learning_rate": 0.000192239013549893,
      "loss": 4.0194,
      "step": 41630
    },
    {
      "epoch": 0.03881521768133839,
      "grad_norm": 1.179286241531372,
      "learning_rate": 0.0001922371492277978,
      "loss": 4.1629,
      "step": 41640
    },
    {
      "epoch": 0.03882453929941748,
      "grad_norm": 1.0534849166870117,
      "learning_rate": 0.0001922352849057026,
      "loss": 4.1322,
      "step": 41650
    },
    {
      "epoch": 0.03883386091749657,
      "grad_norm": 1.040246605873108,
      "learning_rate": 0.0001922334205836074,
      "loss": 3.9862,
      "step": 41660
    },
    {
      "epoch": 0.038843182535575664,
      "grad_norm": 1.0178368091583252,
      "learning_rate": 0.0001922315562615122,
      "loss": 4.1635,
      "step": 41670
    },
    {
      "epoch": 0.03885250415365476,
      "grad_norm": 1.0629743337631226,
      "learning_rate": 0.000192229691939417,
      "loss": 3.9297,
      "step": 41680
    },
    {
      "epoch": 0.038861825771733846,
      "grad_norm": 1.2759244441986084,
      "learning_rate": 0.00019222782761732182,
      "loss": 4.0389,
      "step": 41690
    },
    {
      "epoch": 0.03887114738981294,
      "grad_norm": 1.1328699588775635,
      "learning_rate": 0.0001922259632952266,
      "loss": 4.1638,
      "step": 41700
    },
    {
      "epoch": 0.03888046900789203,
      "grad_norm": 1.1599138975143433,
      "learning_rate": 0.0001922240989731314,
      "loss": 4.0358,
      "step": 41710
    },
    {
      "epoch": 0.03888979062597112,
      "grad_norm": 1.410398006439209,
      "learning_rate": 0.00019222223465103618,
      "loss": 4.0101,
      "step": 41720
    },
    {
      "epoch": 0.03889911224405022,
      "grad_norm": 1.2121272087097168,
      "learning_rate": 0.00019222037032894101,
      "loss": 4.1518,
      "step": 41730
    },
    {
      "epoch": 0.038908433862129305,
      "grad_norm": 0.9896315336227417,
      "learning_rate": 0.00019221850600684582,
      "loss": 4.1115,
      "step": 41740
    },
    {
      "epoch": 0.0389177554802084,
      "grad_norm": 1.0512206554412842,
      "learning_rate": 0.0001922166416847506,
      "loss": 4.1077,
      "step": 41750
    },
    {
      "epoch": 0.038927077098287494,
      "grad_norm": 1.0545541048049927,
      "learning_rate": 0.0001922147773626554,
      "loss": 4.112,
      "step": 41760
    },
    {
      "epoch": 0.03893639871636658,
      "grad_norm": 1.2304154634475708,
      "learning_rate": 0.00019221291304056019,
      "loss": 3.9905,
      "step": 41770
    },
    {
      "epoch": 0.038945720334445676,
      "grad_norm": 1.0344691276550293,
      "learning_rate": 0.000192211048718465,
      "loss": 4.0957,
      "step": 41780
    },
    {
      "epoch": 0.03895504195252476,
      "grad_norm": 1.3365747928619385,
      "learning_rate": 0.0001922091843963698,
      "loss": 4.1033,
      "step": 41790
    },
    {
      "epoch": 0.03896436357060386,
      "grad_norm": 1.0492339134216309,
      "learning_rate": 0.0001922073200742746,
      "loss": 4.0116,
      "step": 41800
    },
    {
      "epoch": 0.03897368518868295,
      "grad_norm": 0.9851827621459961,
      "learning_rate": 0.0001922054557521794,
      "loss": 4.0571,
      "step": 41810
    },
    {
      "epoch": 0.03898300680676204,
      "grad_norm": 0.8828548192977905,
      "learning_rate": 0.0001922035914300842,
      "loss": 3.9617,
      "step": 41820
    },
    {
      "epoch": 0.038992328424841134,
      "grad_norm": 1.0618507862091064,
      "learning_rate": 0.000192201727107989,
      "loss": 4.1192,
      "step": 41830
    },
    {
      "epoch": 0.03900165004292023,
      "grad_norm": 1.1128954887390137,
      "learning_rate": 0.0001921998627858938,
      "loss": 4.0814,
      "step": 41840
    },
    {
      "epoch": 0.039010971660999316,
      "grad_norm": 1.246725082397461,
      "learning_rate": 0.0001921979984637986,
      "loss": 3.9792,
      "step": 41850
    },
    {
      "epoch": 0.03902029327907841,
      "grad_norm": 1.2029876708984375,
      "learning_rate": 0.00019219613414170342,
      "loss": 4.0377,
      "step": 41860
    },
    {
      "epoch": 0.0390296148971575,
      "grad_norm": 1.0504587888717651,
      "learning_rate": 0.0001921942698196082,
      "loss": 4.2513,
      "step": 41870
    },
    {
      "epoch": 0.03903893651523659,
      "grad_norm": 1.154423475265503,
      "learning_rate": 0.000192192405497513,
      "loss": 4.0568,
      "step": 41880
    },
    {
      "epoch": 0.03904825813331569,
      "grad_norm": 1.1372793912887573,
      "learning_rate": 0.00019219054117541778,
      "loss": 4.1353,
      "step": 41890
    },
    {
      "epoch": 0.039057579751394775,
      "grad_norm": 1.061584234237671,
      "learning_rate": 0.00019218867685332261,
      "loss": 4.1542,
      "step": 41900
    },
    {
      "epoch": 0.03906690136947387,
      "grad_norm": 1.0901508331298828,
      "learning_rate": 0.00019218681253122742,
      "loss": 3.948,
      "step": 41910
    },
    {
      "epoch": 0.03907622298755296,
      "grad_norm": 1.0708706378936768,
      "learning_rate": 0.0001921849482091322,
      "loss": 4.012,
      "step": 41920
    },
    {
      "epoch": 0.03908554460563205,
      "grad_norm": 1.2266374826431274,
      "learning_rate": 0.000192183083887037,
      "loss": 4.0494,
      "step": 41930
    },
    {
      "epoch": 0.039094866223711146,
      "grad_norm": 1.1646130084991455,
      "learning_rate": 0.00019218121956494178,
      "loss": 4.0407,
      "step": 41940
    },
    {
      "epoch": 0.03910418784179023,
      "grad_norm": 1.0941312313079834,
      "learning_rate": 0.0001921793552428466,
      "loss": 3.9822,
      "step": 41950
    },
    {
      "epoch": 0.03911350945986933,
      "grad_norm": 0.9910647869110107,
      "learning_rate": 0.00019217749092075142,
      "loss": 4.2827,
      "step": 41960
    },
    {
      "epoch": 0.03912283107794842,
      "grad_norm": 1.1985737085342407,
      "learning_rate": 0.0001921756265986562,
      "loss": 4.0425,
      "step": 41970
    },
    {
      "epoch": 0.03913215269602751,
      "grad_norm": 1.0796658992767334,
      "learning_rate": 0.000192173762276561,
      "loss": 4.2928,
      "step": 41980
    },
    {
      "epoch": 0.039141474314106604,
      "grad_norm": 1.1714568138122559,
      "learning_rate": 0.0001921718979544658,
      "loss": 4.1503,
      "step": 41990
    },
    {
      "epoch": 0.03915079593218569,
      "grad_norm": 1.02399480342865,
      "learning_rate": 0.0001921700336323706,
      "loss": 3.9397,
      "step": 42000
    },
    {
      "epoch": 0.039160117550264786,
      "grad_norm": 1.2299144268035889,
      "learning_rate": 0.0001921681693102754,
      "loss": 4.149,
      "step": 42010
    },
    {
      "epoch": 0.03916943916834388,
      "grad_norm": 1.0460177659988403,
      "learning_rate": 0.0001921663049881802,
      "loss": 4.0962,
      "step": 42020
    },
    {
      "epoch": 0.03917876078642297,
      "grad_norm": 1.1986414194107056,
      "learning_rate": 0.00019216444066608501,
      "loss": 4.1007,
      "step": 42030
    },
    {
      "epoch": 0.03918808240450206,
      "grad_norm": 1.2066329717636108,
      "learning_rate": 0.0001921625763439898,
      "loss": 4.0059,
      "step": 42040
    },
    {
      "epoch": 0.03919740402258115,
      "grad_norm": 0.9798029065132141,
      "learning_rate": 0.0001921607120218946,
      "loss": 4.1928,
      "step": 42050
    },
    {
      "epoch": 0.039206725640660245,
      "grad_norm": 0.9466955065727234,
      "learning_rate": 0.0001921588476997994,
      "loss": 4.1687,
      "step": 42060
    },
    {
      "epoch": 0.03921604725873934,
      "grad_norm": 1.0013900995254517,
      "learning_rate": 0.0001921569833777042,
      "loss": 3.958,
      "step": 42070
    },
    {
      "epoch": 0.03922536887681843,
      "grad_norm": 1.2123006582260132,
      "learning_rate": 0.00019215511905560902,
      "loss": 4.005,
      "step": 42080
    },
    {
      "epoch": 0.03923469049489752,
      "grad_norm": 1.0928014516830444,
      "learning_rate": 0.0001921532547335138,
      "loss": 4.2003,
      "step": 42090
    },
    {
      "epoch": 0.039244012112976616,
      "grad_norm": 1.1101233959197998,
      "learning_rate": 0.0001921513904114186,
      "loss": 4.1562,
      "step": 42100
    },
    {
      "epoch": 0.0392533337310557,
      "grad_norm": 1.155512809753418,
      "learning_rate": 0.0001921495260893234,
      "loss": 4.0825,
      "step": 42110
    },
    {
      "epoch": 0.0392626553491348,
      "grad_norm": 1.0553812980651855,
      "learning_rate": 0.00019214766176722822,
      "loss": 3.9232,
      "step": 42120
    },
    {
      "epoch": 0.039271976967213885,
      "grad_norm": 1.0868277549743652,
      "learning_rate": 0.00019214579744513302,
      "loss": 4.0935,
      "step": 42130
    },
    {
      "epoch": 0.03928129858529298,
      "grad_norm": 1.1568758487701416,
      "learning_rate": 0.0001921439331230378,
      "loss": 3.9172,
      "step": 42140
    },
    {
      "epoch": 0.039290620203372074,
      "grad_norm": 1.0833420753479004,
      "learning_rate": 0.0001921420688009426,
      "loss": 4.1924,
      "step": 42150
    },
    {
      "epoch": 0.03929994182145116,
      "grad_norm": 1.3293167352676392,
      "learning_rate": 0.00019214020447884742,
      "loss": 4.074,
      "step": 42160
    },
    {
      "epoch": 0.039309263439530256,
      "grad_norm": 1.13907790184021,
      "learning_rate": 0.0001921383401567522,
      "loss": 4.0363,
      "step": 42170
    },
    {
      "epoch": 0.03931858505760935,
      "grad_norm": 1.0736900568008423,
      "learning_rate": 0.00019213647583465703,
      "loss": 3.8206,
      "step": 42180
    },
    {
      "epoch": 0.03932790667568844,
      "grad_norm": 1.3065459728240967,
      "learning_rate": 0.0001921346115125618,
      "loss": 4.1636,
      "step": 42190
    },
    {
      "epoch": 0.03933722829376753,
      "grad_norm": 1.0954782962799072,
      "learning_rate": 0.0001921327471904666,
      "loss": 4.1697,
      "step": 42200
    },
    {
      "epoch": 0.03934654991184662,
      "grad_norm": 1.1696150302886963,
      "learning_rate": 0.00019213088286837142,
      "loss": 4.0218,
      "step": 42210
    },
    {
      "epoch": 0.039355871529925715,
      "grad_norm": 1.143318772315979,
      "learning_rate": 0.0001921290185462762,
      "loss": 4.3448,
      "step": 42220
    },
    {
      "epoch": 0.03936519314800481,
      "grad_norm": 0.9596447348594666,
      "learning_rate": 0.000192127154224181,
      "loss": 4.1549,
      "step": 42230
    },
    {
      "epoch": 0.0393745147660839,
      "grad_norm": 1.0380823612213135,
      "learning_rate": 0.0001921252899020858,
      "loss": 4.2256,
      "step": 42240
    },
    {
      "epoch": 0.03938383638416299,
      "grad_norm": 1.0686918497085571,
      "learning_rate": 0.00019212342557999062,
      "loss": 4.1012,
      "step": 42250
    },
    {
      "epoch": 0.03939315800224208,
      "grad_norm": 1.1915467977523804,
      "learning_rate": 0.00019212156125789542,
      "loss": 4.0643,
      "step": 42260
    },
    {
      "epoch": 0.039402479620321174,
      "grad_norm": 1.1906009912490845,
      "learning_rate": 0.0001921196969358002,
      "loss": 4.0687,
      "step": 42270
    },
    {
      "epoch": 0.03941180123840027,
      "grad_norm": 0.8966283202171326,
      "learning_rate": 0.000192117832613705,
      "loss": 3.9216,
      "step": 42280
    },
    {
      "epoch": 0.039421122856479356,
      "grad_norm": 1.1797189712524414,
      "learning_rate": 0.00019211596829160982,
      "loss": 3.8217,
      "step": 42290
    },
    {
      "epoch": 0.03943044447455845,
      "grad_norm": 1.1511718034744263,
      "learning_rate": 0.00019211410396951462,
      "loss": 4.2013,
      "step": 42300
    },
    {
      "epoch": 0.039439766092637545,
      "grad_norm": 1.0186094045639038,
      "learning_rate": 0.00019211223964741943,
      "loss": 4.0447,
      "step": 42310
    },
    {
      "epoch": 0.03944908771071663,
      "grad_norm": 1.0421664714813232,
      "learning_rate": 0.0001921103753253242,
      "loss": 3.9538,
      "step": 42320
    },
    {
      "epoch": 0.03945840932879573,
      "grad_norm": 1.1282199621200562,
      "learning_rate": 0.00019210851100322901,
      "loss": 3.9109,
      "step": 42330
    },
    {
      "epoch": 0.039467730946874814,
      "grad_norm": 1.0363339185714722,
      "learning_rate": 0.0001921066466811338,
      "loss": 4.1428,
      "step": 42340
    },
    {
      "epoch": 0.03947705256495391,
      "grad_norm": 1.0955358743667603,
      "learning_rate": 0.00019210478235903863,
      "loss": 4.0277,
      "step": 42350
    },
    {
      "epoch": 0.039486374183033,
      "grad_norm": 1.0445301532745361,
      "learning_rate": 0.00019210291803694343,
      "loss": 3.9649,
      "step": 42360
    },
    {
      "epoch": 0.03949569580111209,
      "grad_norm": 1.163450002670288,
      "learning_rate": 0.0001921010537148482,
      "loss": 4.0597,
      "step": 42370
    },
    {
      "epoch": 0.039505017419191185,
      "grad_norm": 1.2793292999267578,
      "learning_rate": 0.00019209918939275302,
      "loss": 4.2579,
      "step": 42380
    },
    {
      "epoch": 0.03951433903727027,
      "grad_norm": 1.085758924484253,
      "learning_rate": 0.0001920973250706578,
      "loss": 4.0333,
      "step": 42390
    },
    {
      "epoch": 0.03952366065534937,
      "grad_norm": 1.0917407274246216,
      "learning_rate": 0.0001920954607485626,
      "loss": 3.895,
      "step": 42400
    },
    {
      "epoch": 0.03953298227342846,
      "grad_norm": 0.990318238735199,
      "learning_rate": 0.00019209359642646744,
      "loss": 4.0343,
      "step": 42410
    },
    {
      "epoch": 0.03954230389150755,
      "grad_norm": 1.0810824632644653,
      "learning_rate": 0.00019209173210437222,
      "loss": 4.0899,
      "step": 42420
    },
    {
      "epoch": 0.039551625509586644,
      "grad_norm": 1.2075937986373901,
      "learning_rate": 0.00019208986778227702,
      "loss": 4.0933,
      "step": 42430
    },
    {
      "epoch": 0.03956094712766574,
      "grad_norm": 1.0956226587295532,
      "learning_rate": 0.0001920880034601818,
      "loss": 4.1208,
      "step": 42440
    },
    {
      "epoch": 0.039570268745744826,
      "grad_norm": 1.206064224243164,
      "learning_rate": 0.0001920861391380866,
      "loss": 4.0552,
      "step": 42450
    },
    {
      "epoch": 0.03957959036382392,
      "grad_norm": 1.038028597831726,
      "learning_rate": 0.00019208427481599141,
      "loss": 4.0896,
      "step": 42460
    },
    {
      "epoch": 0.03958891198190301,
      "grad_norm": 1.2799361944198608,
      "learning_rate": 0.00019208241049389622,
      "loss": 4.0602,
      "step": 42470
    },
    {
      "epoch": 0.0395982335999821,
      "grad_norm": 1.0304887294769287,
      "learning_rate": 0.00019208054617180103,
      "loss": 4.0309,
      "step": 42480
    },
    {
      "epoch": 0.0396075552180612,
      "grad_norm": 1.1089017391204834,
      "learning_rate": 0.0001920786818497058,
      "loss": 4.1068,
      "step": 42490
    },
    {
      "epoch": 0.039616876836140284,
      "grad_norm": 1.0723966360092163,
      "learning_rate": 0.0001920768175276106,
      "loss": 4.2136,
      "step": 42500
    },
    {
      "epoch": 0.03962619845421938,
      "grad_norm": 1.153009295463562,
      "learning_rate": 0.00019207495320551542,
      "loss": 4.2015,
      "step": 42510
    },
    {
      "epoch": 0.03963552007229847,
      "grad_norm": 1.0389201641082764,
      "learning_rate": 0.00019207308888342023,
      "loss": 3.9168,
      "step": 42520
    },
    {
      "epoch": 0.03964484169037756,
      "grad_norm": 1.1957415342330933,
      "learning_rate": 0.00019207122456132503,
      "loss": 3.9987,
      "step": 42530
    },
    {
      "epoch": 0.039654163308456655,
      "grad_norm": 1.075339436531067,
      "learning_rate": 0.0001920693602392298,
      "loss": 4.1346,
      "step": 42540
    },
    {
      "epoch": 0.03966348492653574,
      "grad_norm": 1.104069709777832,
      "learning_rate": 0.00019206749591713462,
      "loss": 4.0982,
      "step": 42550
    },
    {
      "epoch": 0.03967280654461484,
      "grad_norm": 1.1499218940734863,
      "learning_rate": 0.00019206563159503942,
      "loss": 4.0189,
      "step": 42560
    },
    {
      "epoch": 0.03968212816269393,
      "grad_norm": 1.0492795705795288,
      "learning_rate": 0.00019206376727294423,
      "loss": 4.0503,
      "step": 42570
    },
    {
      "epoch": 0.03969144978077302,
      "grad_norm": 1.1545811891555786,
      "learning_rate": 0.00019206190295084904,
      "loss": 4.199,
      "step": 42580
    },
    {
      "epoch": 0.039700771398852114,
      "grad_norm": 0.9600545763969421,
      "learning_rate": 0.00019206003862875382,
      "loss": 3.95,
      "step": 42590
    },
    {
      "epoch": 0.0397100930169312,
      "grad_norm": 0.9610468745231628,
      "learning_rate": 0.00019205817430665862,
      "loss": 4.0216,
      "step": 42600
    },
    {
      "epoch": 0.039719414635010296,
      "grad_norm": 1.0158371925354004,
      "learning_rate": 0.00019205630998456343,
      "loss": 4.2085,
      "step": 42610
    },
    {
      "epoch": 0.03972873625308939,
      "grad_norm": 0.9873038530349731,
      "learning_rate": 0.0001920544456624682,
      "loss": 4.1015,
      "step": 42620
    },
    {
      "epoch": 0.03973805787116848,
      "grad_norm": 0.9817074537277222,
      "learning_rate": 0.00019205258134037304,
      "loss": 4.0115,
      "step": 42630
    },
    {
      "epoch": 0.03974737948924757,
      "grad_norm": 1.2631568908691406,
      "learning_rate": 0.00019205071701827782,
      "loss": 4.033,
      "step": 42640
    },
    {
      "epoch": 0.03975670110732667,
      "grad_norm": 1.1931650638580322,
      "learning_rate": 0.00019204885269618263,
      "loss": 4.2182,
      "step": 42650
    },
    {
      "epoch": 0.039766022725405754,
      "grad_norm": 0.9146193861961365,
      "learning_rate": 0.00019204698837408743,
      "loss": 4.0568,
      "step": 42660
    },
    {
      "epoch": 0.03977534434348485,
      "grad_norm": 1.2128678560256958,
      "learning_rate": 0.0001920451240519922,
      "loss": 3.959,
      "step": 42670
    },
    {
      "epoch": 0.039784665961563936,
      "grad_norm": 1.1438740491867065,
      "learning_rate": 0.00019204325972989702,
      "loss": 3.9526,
      "step": 42680
    },
    {
      "epoch": 0.03979398757964303,
      "grad_norm": 1.144945740699768,
      "learning_rate": 0.00019204139540780182,
      "loss": 3.9123,
      "step": 42690
    },
    {
      "epoch": 0.039803309197722125,
      "grad_norm": 0.9525318145751953,
      "learning_rate": 0.00019203953108570663,
      "loss": 4.1032,
      "step": 42700
    },
    {
      "epoch": 0.03981263081580121,
      "grad_norm": 0.9862566590309143,
      "learning_rate": 0.00019203766676361144,
      "loss": 4.1622,
      "step": 42710
    },
    {
      "epoch": 0.03982195243388031,
      "grad_norm": 1.1283583641052246,
      "learning_rate": 0.00019203580244151622,
      "loss": 4.0266,
      "step": 42720
    },
    {
      "epoch": 0.039831274051959395,
      "grad_norm": 1.278243899345398,
      "learning_rate": 0.00019203393811942102,
      "loss": 4.1518,
      "step": 42730
    },
    {
      "epoch": 0.03984059567003849,
      "grad_norm": 1.0929255485534668,
      "learning_rate": 0.00019203207379732583,
      "loss": 4.1014,
      "step": 42740
    },
    {
      "epoch": 0.039849917288117584,
      "grad_norm": 1.0354784727096558,
      "learning_rate": 0.00019203020947523063,
      "loss": 3.9305,
      "step": 42750
    },
    {
      "epoch": 0.03985923890619667,
      "grad_norm": 1.0497527122497559,
      "learning_rate": 0.00019202834515313544,
      "loss": 4.1889,
      "step": 42760
    },
    {
      "epoch": 0.039868560524275766,
      "grad_norm": 1.0971322059631348,
      "learning_rate": 0.00019202648083104022,
      "loss": 3.9646,
      "step": 42770
    },
    {
      "epoch": 0.03987788214235486,
      "grad_norm": 1.1249361038208008,
      "learning_rate": 0.00019202461650894503,
      "loss": 3.8761,
      "step": 42780
    },
    {
      "epoch": 0.03988720376043395,
      "grad_norm": 1.2208287715911865,
      "learning_rate": 0.0001920227521868498,
      "loss": 3.957,
      "step": 42790
    },
    {
      "epoch": 0.03989652537851304,
      "grad_norm": 1.222969889640808,
      "learning_rate": 0.00019202088786475464,
      "loss": 4.0306,
      "step": 42800
    },
    {
      "epoch": 0.03990584699659213,
      "grad_norm": 1.2550052404403687,
      "learning_rate": 0.00019201902354265945,
      "loss": 3.9653,
      "step": 42810
    },
    {
      "epoch": 0.039915168614671225,
      "grad_norm": 0.9823949337005615,
      "learning_rate": 0.00019201715922056422,
      "loss": 4.1961,
      "step": 42820
    },
    {
      "epoch": 0.03992449023275032,
      "grad_norm": 1.0704467296600342,
      "learning_rate": 0.00019201529489846903,
      "loss": 3.9799,
      "step": 42830
    },
    {
      "epoch": 0.03993381185082941,
      "grad_norm": 1.137158989906311,
      "learning_rate": 0.0001920134305763738,
      "loss": 3.9097,
      "step": 42840
    },
    {
      "epoch": 0.0399431334689085,
      "grad_norm": 0.9004451632499695,
      "learning_rate": 0.00019201156625427862,
      "loss": 4.3131,
      "step": 42850
    },
    {
      "epoch": 0.039952455086987596,
      "grad_norm": 1.154202938079834,
      "learning_rate": 0.00019200970193218345,
      "loss": 4.1238,
      "step": 42860
    },
    {
      "epoch": 0.03996177670506668,
      "grad_norm": 0.9608919620513916,
      "learning_rate": 0.00019200783761008823,
      "loss": 3.8764,
      "step": 42870
    },
    {
      "epoch": 0.03997109832314578,
      "grad_norm": 1.3931535482406616,
      "learning_rate": 0.00019200597328799304,
      "loss": 4.0656,
      "step": 42880
    },
    {
      "epoch": 0.039980419941224865,
      "grad_norm": 1.1796376705169678,
      "learning_rate": 0.00019200410896589781,
      "loss": 4.1001,
      "step": 42890
    },
    {
      "epoch": 0.03998974155930396,
      "grad_norm": 1.077997088432312,
      "learning_rate": 0.00019200224464380262,
      "loss": 4.1188,
      "step": 42900
    },
    {
      "epoch": 0.039999063177383054,
      "grad_norm": 1.2649714946746826,
      "learning_rate": 0.00019200038032170743,
      "loss": 3.9037,
      "step": 42910
    },
    {
      "epoch": 0.04000838479546214,
      "grad_norm": 1.1107327938079834,
      "learning_rate": 0.00019199851599961223,
      "loss": 4.158,
      "step": 42920
    },
    {
      "epoch": 0.040017706413541236,
      "grad_norm": 0.9847804307937622,
      "learning_rate": 0.00019199665167751704,
      "loss": 4.1534,
      "step": 42930
    },
    {
      "epoch": 0.040027028031620324,
      "grad_norm": 1.056971788406372,
      "learning_rate": 0.00019199478735542182,
      "loss": 4.2669,
      "step": 42940
    },
    {
      "epoch": 0.04003634964969942,
      "grad_norm": 1.1442947387695312,
      "learning_rate": 0.00019199292303332663,
      "loss": 4.094,
      "step": 42950
    },
    {
      "epoch": 0.04004567126777851,
      "grad_norm": 0.9336591362953186,
      "learning_rate": 0.00019199105871123143,
      "loss": 3.9325,
      "step": 42960
    },
    {
      "epoch": 0.0400549928858576,
      "grad_norm": 1.2682347297668457,
      "learning_rate": 0.00019198919438913624,
      "loss": 3.9837,
      "step": 42970
    },
    {
      "epoch": 0.040064314503936695,
      "grad_norm": 1.1265076398849487,
      "learning_rate": 0.00019198733006704104,
      "loss": 4.1391,
      "step": 42980
    },
    {
      "epoch": 0.04007363612201579,
      "grad_norm": 1.1109715700149536,
      "learning_rate": 0.00019198546574494582,
      "loss": 4.0581,
      "step": 42990
    },
    {
      "epoch": 0.04008295774009488,
      "grad_norm": 1.137202501296997,
      "learning_rate": 0.00019198360142285063,
      "loss": 3.9846,
      "step": 43000
    },
    {
      "epoch": 0.04009227935817397,
      "grad_norm": 1.273206353187561,
      "learning_rate": 0.00019198173710075544,
      "loss": 4.0768,
      "step": 43010
    },
    {
      "epoch": 0.04010160097625306,
      "grad_norm": 1.1512364149093628,
      "learning_rate": 0.00019197987277866024,
      "loss": 3.8837,
      "step": 43020
    },
    {
      "epoch": 0.04011092259433215,
      "grad_norm": 0.9835382103919983,
      "learning_rate": 0.00019197800845656505,
      "loss": 4.1358,
      "step": 43030
    },
    {
      "epoch": 0.04012024421241125,
      "grad_norm": 1.0918318033218384,
      "learning_rate": 0.00019197614413446983,
      "loss": 4.145,
      "step": 43040
    },
    {
      "epoch": 0.040129565830490335,
      "grad_norm": 1.1641207933425903,
      "learning_rate": 0.00019197427981237463,
      "loss": 3.9156,
      "step": 43050
    },
    {
      "epoch": 0.04013888744856943,
      "grad_norm": 0.9442983269691467,
      "learning_rate": 0.00019197241549027944,
      "loss": 3.9669,
      "step": 43060
    },
    {
      "epoch": 0.04014820906664852,
      "grad_norm": 0.9722496271133423,
      "learning_rate": 0.00019197055116818422,
      "loss": 3.9612,
      "step": 43070
    },
    {
      "epoch": 0.04015753068472761,
      "grad_norm": 1.061601996421814,
      "learning_rate": 0.00019196868684608905,
      "loss": 4.0273,
      "step": 43080
    },
    {
      "epoch": 0.040166852302806706,
      "grad_norm": 1.0513012409210205,
      "learning_rate": 0.00019196682252399383,
      "loss": 4.1261,
      "step": 43090
    },
    {
      "epoch": 0.040176173920885794,
      "grad_norm": 1.0452300310134888,
      "learning_rate": 0.00019196495820189864,
      "loss": 4.0529,
      "step": 43100
    },
    {
      "epoch": 0.04018549553896489,
      "grad_norm": 0.9608991146087646,
      "learning_rate": 0.00019196309387980345,
      "loss": 4.21,
      "step": 43110
    },
    {
      "epoch": 0.04019481715704398,
      "grad_norm": 1.1722394227981567,
      "learning_rate": 0.00019196122955770822,
      "loss": 4.1669,
      "step": 43120
    },
    {
      "epoch": 0.04020413877512307,
      "grad_norm": 1.0673527717590332,
      "learning_rate": 0.00019195936523561303,
      "loss": 3.9258,
      "step": 43130
    },
    {
      "epoch": 0.040213460393202165,
      "grad_norm": 0.9702373743057251,
      "learning_rate": 0.00019195750091351784,
      "loss": 4.147,
      "step": 43140
    },
    {
      "epoch": 0.04022278201128125,
      "grad_norm": 1.2085602283477783,
      "learning_rate": 0.00019195563659142264,
      "loss": 4.1582,
      "step": 43150
    },
    {
      "epoch": 0.04023210362936035,
      "grad_norm": 1.0937830209732056,
      "learning_rate": 0.00019195377226932742,
      "loss": 4.2095,
      "step": 43160
    },
    {
      "epoch": 0.04024142524743944,
      "grad_norm": 1.0363619327545166,
      "learning_rate": 0.00019195190794723223,
      "loss": 4.0001,
      "step": 43170
    },
    {
      "epoch": 0.04025074686551853,
      "grad_norm": 1.1828322410583496,
      "learning_rate": 0.00019195004362513704,
      "loss": 4.2313,
      "step": 43180
    },
    {
      "epoch": 0.04026006848359762,
      "grad_norm": 1.1210824251174927,
      "learning_rate": 0.00019194817930304184,
      "loss": 4.118,
      "step": 43190
    },
    {
      "epoch": 0.04026939010167672,
      "grad_norm": 1.073182225227356,
      "learning_rate": 0.00019194631498094665,
      "loss": 4.2907,
      "step": 43200
    },
    {
      "epoch": 0.040278711719755805,
      "grad_norm": 0.9939176440238953,
      "learning_rate": 0.00019194445065885143,
      "loss": 3.9797,
      "step": 43210
    },
    {
      "epoch": 0.0402880333378349,
      "grad_norm": 1.034401535987854,
      "learning_rate": 0.00019194258633675623,
      "loss": 4.1685,
      "step": 43220
    },
    {
      "epoch": 0.04029735495591399,
      "grad_norm": 1.0120575428009033,
      "learning_rate": 0.00019194072201466104,
      "loss": 4.0277,
      "step": 43230
    },
    {
      "epoch": 0.04030667657399308,
      "grad_norm": 1.0449073314666748,
      "learning_rate": 0.00019193885769256582,
      "loss": 4.0757,
      "step": 43240
    },
    {
      "epoch": 0.040315998192072176,
      "grad_norm": 1.1424229145050049,
      "learning_rate": 0.00019193699337047065,
      "loss": 4.1176,
      "step": 43250
    },
    {
      "epoch": 0.040325319810151264,
      "grad_norm": 1.1277271509170532,
      "learning_rate": 0.00019193512904837543,
      "loss": 4.1489,
      "step": 43260
    },
    {
      "epoch": 0.04033464142823036,
      "grad_norm": 1.1223136186599731,
      "learning_rate": 0.00019193326472628024,
      "loss": 4.1338,
      "step": 43270
    },
    {
      "epoch": 0.040343963046309446,
      "grad_norm": 0.923865020275116,
      "learning_rate": 0.00019193140040418504,
      "loss": 3.9443,
      "step": 43280
    },
    {
      "epoch": 0.04035328466438854,
      "grad_norm": 0.9501680731773376,
      "learning_rate": 0.00019192953608208982,
      "loss": 4.0337,
      "step": 43290
    },
    {
      "epoch": 0.040362606282467635,
      "grad_norm": 1.3166868686676025,
      "learning_rate": 0.00019192767175999463,
      "loss": 3.9779,
      "step": 43300
    },
    {
      "epoch": 0.04037192790054672,
      "grad_norm": 1.073359727859497,
      "learning_rate": 0.00019192580743789944,
      "loss": 3.9543,
      "step": 43310
    },
    {
      "epoch": 0.04038124951862582,
      "grad_norm": 1.0197432041168213,
      "learning_rate": 0.00019192394311580424,
      "loss": 3.9808,
      "step": 43320
    },
    {
      "epoch": 0.04039057113670491,
      "grad_norm": 1.209450125694275,
      "learning_rate": 0.00019192207879370905,
      "loss": 4.1762,
      "step": 43330
    },
    {
      "epoch": 0.040399892754784,
      "grad_norm": 1.2149900197982788,
      "learning_rate": 0.00019192021447161383,
      "loss": 3.9798,
      "step": 43340
    },
    {
      "epoch": 0.04040921437286309,
      "grad_norm": 0.9547286033630371,
      "learning_rate": 0.00019191835014951863,
      "loss": 4.127,
      "step": 43350
    },
    {
      "epoch": 0.04041853599094218,
      "grad_norm": 0.9419257640838623,
      "learning_rate": 0.00019191648582742344,
      "loss": 3.969,
      "step": 43360
    },
    {
      "epoch": 0.040427857609021275,
      "grad_norm": 0.9373849630355835,
      "learning_rate": 0.00019191462150532825,
      "loss": 4.0081,
      "step": 43370
    },
    {
      "epoch": 0.04043717922710037,
      "grad_norm": 1.1379274129867554,
      "learning_rate": 0.00019191275718323305,
      "loss": 3.9457,
      "step": 43380
    },
    {
      "epoch": 0.04044650084517946,
      "grad_norm": 1.3793485164642334,
      "learning_rate": 0.00019191089286113783,
      "loss": 4.1438,
      "step": 43390
    },
    {
      "epoch": 0.04045582246325855,
      "grad_norm": 0.8310623168945312,
      "learning_rate": 0.00019190902853904264,
      "loss": 4.2545,
      "step": 43400
    },
    {
      "epoch": 0.04046514408133764,
      "grad_norm": 0.9341917037963867,
      "learning_rate": 0.00019190716421694744,
      "loss": 4.07,
      "step": 43410
    },
    {
      "epoch": 0.040474465699416734,
      "grad_norm": 1.2635358572006226,
      "learning_rate": 0.00019190529989485225,
      "loss": 4.0413,
      "step": 43420
    },
    {
      "epoch": 0.04048378731749583,
      "grad_norm": 1.1788896322250366,
      "learning_rate": 0.00019190343557275706,
      "loss": 3.9992,
      "step": 43430
    },
    {
      "epoch": 0.040493108935574916,
      "grad_norm": 1.0967098474502563,
      "learning_rate": 0.00019190157125066184,
      "loss": 4.0016,
      "step": 43440
    },
    {
      "epoch": 0.04050243055365401,
      "grad_norm": 1.0004818439483643,
      "learning_rate": 0.00019189970692856664,
      "loss": 3.9199,
      "step": 43450
    },
    {
      "epoch": 0.040511752171733105,
      "grad_norm": 1.049928069114685,
      "learning_rate": 0.00019189784260647142,
      "loss": 3.967,
      "step": 43460
    },
    {
      "epoch": 0.04052107378981219,
      "grad_norm": 0.9846920967102051,
      "learning_rate": 0.00019189597828437626,
      "loss": 3.8605,
      "step": 43470
    },
    {
      "epoch": 0.04053039540789129,
      "grad_norm": 1.1640135049819946,
      "learning_rate": 0.00019189411396228106,
      "loss": 4.1721,
      "step": 43480
    },
    {
      "epoch": 0.040539717025970375,
      "grad_norm": 1.1016982793807983,
      "learning_rate": 0.00019189224964018584,
      "loss": 4.199,
      "step": 43490
    },
    {
      "epoch": 0.04054903864404947,
      "grad_norm": 1.075901746749878,
      "learning_rate": 0.00019189038531809065,
      "loss": 3.717,
      "step": 43500
    },
    {
      "epoch": 0.040558360262128564,
      "grad_norm": 1.0371310710906982,
      "learning_rate": 0.00019188852099599543,
      "loss": 4.1499,
      "step": 43510
    },
    {
      "epoch": 0.04056768188020765,
      "grad_norm": 1.0684322118759155,
      "learning_rate": 0.00019188665667390023,
      "loss": 3.8774,
      "step": 43520
    },
    {
      "epoch": 0.040577003498286746,
      "grad_norm": 1.3429325819015503,
      "learning_rate": 0.00019188479235180507,
      "loss": 4.119,
      "step": 43530
    },
    {
      "epoch": 0.04058632511636584,
      "grad_norm": 0.9828104376792908,
      "learning_rate": 0.00019188292802970985,
      "loss": 3.9684,
      "step": 43540
    },
    {
      "epoch": 0.04059564673444493,
      "grad_norm": 1.2230119705200195,
      "learning_rate": 0.00019188106370761465,
      "loss": 4.055,
      "step": 43550
    },
    {
      "epoch": 0.04060496835252402,
      "grad_norm": 1.3276348114013672,
      "learning_rate": 0.00019187919938551943,
      "loss": 3.9929,
      "step": 43560
    },
    {
      "epoch": 0.04061428997060311,
      "grad_norm": 1.0539911985397339,
      "learning_rate": 0.00019187733506342424,
      "loss": 3.9275,
      "step": 43570
    },
    {
      "epoch": 0.040623611588682204,
      "grad_norm": 1.07203209400177,
      "learning_rate": 0.00019187547074132904,
      "loss": 3.995,
      "step": 43580
    },
    {
      "epoch": 0.0406329332067613,
      "grad_norm": 1.2134802341461182,
      "learning_rate": 0.00019187360641923385,
      "loss": 4.1561,
      "step": 43590
    },
    {
      "epoch": 0.040642254824840386,
      "grad_norm": 1.1191792488098145,
      "learning_rate": 0.00019187174209713866,
      "loss": 4.2318,
      "step": 43600
    },
    {
      "epoch": 0.04065157644291948,
      "grad_norm": 1.0492390394210815,
      "learning_rate": 0.00019186987777504344,
      "loss": 4.0357,
      "step": 43610
    },
    {
      "epoch": 0.04066089806099857,
      "grad_norm": 0.9830111861228943,
      "learning_rate": 0.00019186801345294824,
      "loss": 4.0446,
      "step": 43620
    },
    {
      "epoch": 0.04067021967907766,
      "grad_norm": 1.1187617778778076,
      "learning_rate": 0.00019186614913085305,
      "loss": 3.9212,
      "step": 43630
    },
    {
      "epoch": 0.04067954129715676,
      "grad_norm": 1.2294853925704956,
      "learning_rate": 0.00019186428480875785,
      "loss": 4.1235,
      "step": 43640
    },
    {
      "epoch": 0.040688862915235845,
      "grad_norm": 0.9939768314361572,
      "learning_rate": 0.00019186242048666266,
      "loss": 4.0663,
      "step": 43650
    },
    {
      "epoch": 0.04069818453331494,
      "grad_norm": 1.1233911514282227,
      "learning_rate": 0.00019186055616456744,
      "loss": 4.0349,
      "step": 43660
    },
    {
      "epoch": 0.040707506151394034,
      "grad_norm": 1.1488655805587769,
      "learning_rate": 0.00019185869184247225,
      "loss": 3.9233,
      "step": 43670
    },
    {
      "epoch": 0.04071682776947312,
      "grad_norm": 1.1564654111862183,
      "learning_rate": 0.00019185682752037705,
      "loss": 4.1677,
      "step": 43680
    },
    {
      "epoch": 0.040726149387552216,
      "grad_norm": 1.108557105064392,
      "learning_rate": 0.00019185496319828186,
      "loss": 4.0539,
      "step": 43690
    },
    {
      "epoch": 0.0407354710056313,
      "grad_norm": 1.2820751667022705,
      "learning_rate": 0.00019185309887618667,
      "loss": 4.0342,
      "step": 43700
    },
    {
      "epoch": 0.0407447926237104,
      "grad_norm": 1.2385870218276978,
      "learning_rate": 0.00019185123455409144,
      "loss": 4.1174,
      "step": 43710
    },
    {
      "epoch": 0.04075411424178949,
      "grad_norm": 1.0474281311035156,
      "learning_rate": 0.00019184937023199625,
      "loss": 4.0151,
      "step": 43720
    },
    {
      "epoch": 0.04076343585986858,
      "grad_norm": 1.0027281045913696,
      "learning_rate": 0.00019184750590990106,
      "loss": 4.0678,
      "step": 43730
    },
    {
      "epoch": 0.040772757477947674,
      "grad_norm": 0.9671531915664673,
      "learning_rate": 0.00019184564158780584,
      "loss": 4.1976,
      "step": 43740
    },
    {
      "epoch": 0.04078207909602676,
      "grad_norm": 1.2948613166809082,
      "learning_rate": 0.00019184377726571064,
      "loss": 4.0598,
      "step": 43750
    },
    {
      "epoch": 0.040791400714105856,
      "grad_norm": 1.1558195352554321,
      "learning_rate": 0.00019184191294361545,
      "loss": 4.0151,
      "step": 43760
    },
    {
      "epoch": 0.04080072233218495,
      "grad_norm": 1.0946600437164307,
      "learning_rate": 0.00019184004862152025,
      "loss": 4.1718,
      "step": 43770
    },
    {
      "epoch": 0.04081004395026404,
      "grad_norm": 0.9462854266166687,
      "learning_rate": 0.00019183818429942506,
      "loss": 3.9153,
      "step": 43780
    },
    {
      "epoch": 0.04081936556834313,
      "grad_norm": 1.313506841659546,
      "learning_rate": 0.00019183631997732984,
      "loss": 4.1679,
      "step": 43790
    },
    {
      "epoch": 0.04082868718642223,
      "grad_norm": 1.2227962017059326,
      "learning_rate": 0.00019183445565523465,
      "loss": 3.9501,
      "step": 43800
    },
    {
      "epoch": 0.040838008804501315,
      "grad_norm": 1.0569132566452026,
      "learning_rate": 0.00019183259133313945,
      "loss": 4.0924,
      "step": 43810
    },
    {
      "epoch": 0.04084733042258041,
      "grad_norm": 1.1675416231155396,
      "learning_rate": 0.00019183072701104426,
      "loss": 4.0544,
      "step": 43820
    },
    {
      "epoch": 0.0408566520406595,
      "grad_norm": 1.0693832635879517,
      "learning_rate": 0.00019182886268894907,
      "loss": 3.956,
      "step": 43830
    },
    {
      "epoch": 0.04086597365873859,
      "grad_norm": 1.226027011871338,
      "learning_rate": 0.00019182699836685384,
      "loss": 4.0784,
      "step": 43840
    },
    {
      "epoch": 0.040875295276817686,
      "grad_norm": 1.0437766313552856,
      "learning_rate": 0.00019182513404475865,
      "loss": 4.081,
      "step": 43850
    },
    {
      "epoch": 0.04088461689489677,
      "grad_norm": 1.087611436843872,
      "learning_rate": 0.00019182326972266346,
      "loss": 4.2381,
      "step": 43860
    },
    {
      "epoch": 0.04089393851297587,
      "grad_norm": 1.1804695129394531,
      "learning_rate": 0.00019182140540056826,
      "loss": 4.0124,
      "step": 43870
    },
    {
      "epoch": 0.04090326013105496,
      "grad_norm": 1.1137412786483765,
      "learning_rate": 0.00019181954107847307,
      "loss": 4.056,
      "step": 43880
    },
    {
      "epoch": 0.04091258174913405,
      "grad_norm": 1.0946376323699951,
      "learning_rate": 0.00019181767675637785,
      "loss": 4.1774,
      "step": 43890
    },
    {
      "epoch": 0.040921903367213144,
      "grad_norm": 1.1054651737213135,
      "learning_rate": 0.00019181581243428266,
      "loss": 3.9759,
      "step": 43900
    },
    {
      "epoch": 0.04093122498529223,
      "grad_norm": 0.9183970093727112,
      "learning_rate": 0.00019181394811218743,
      "loss": 4.1394,
      "step": 43910
    },
    {
      "epoch": 0.040940546603371326,
      "grad_norm": 1.21921968460083,
      "learning_rate": 0.00019181208379009227,
      "loss": 3.935,
      "step": 43920
    },
    {
      "epoch": 0.04094986822145042,
      "grad_norm": 1.1525733470916748,
      "learning_rate": 0.00019181021946799707,
      "loss": 3.9289,
      "step": 43930
    },
    {
      "epoch": 0.04095918983952951,
      "grad_norm": 1.0907737016677856,
      "learning_rate": 0.00019180835514590185,
      "loss": 3.9611,
      "step": 43940
    },
    {
      "epoch": 0.0409685114576086,
      "grad_norm": 1.0724620819091797,
      "learning_rate": 0.00019180649082380666,
      "loss": 4.0496,
      "step": 43950
    },
    {
      "epoch": 0.04097783307568769,
      "grad_norm": 1.052040934562683,
      "learning_rate": 0.00019180462650171144,
      "loss": 4.0758,
      "step": 43960
    },
    {
      "epoch": 0.040987154693766785,
      "grad_norm": 1.0293339490890503,
      "learning_rate": 0.00019180276217961625,
      "loss": 4.1795,
      "step": 43970
    },
    {
      "epoch": 0.04099647631184588,
      "grad_norm": 1.0809273719787598,
      "learning_rate": 0.00019180089785752108,
      "loss": 4.0561,
      "step": 43980
    },
    {
      "epoch": 0.04100579792992497,
      "grad_norm": 1.0094774961471558,
      "learning_rate": 0.00019179903353542586,
      "loss": 4.1135,
      "step": 43990
    },
    {
      "epoch": 0.04101511954800406,
      "grad_norm": 1.0702857971191406,
      "learning_rate": 0.00019179716921333066,
      "loss": 4.0894,
      "step": 44000
    },
    {
      "epoch": 0.041024441166083156,
      "grad_norm": 1.2968230247497559,
      "learning_rate": 0.00019179530489123544,
      "loss": 4.0193,
      "step": 44010
    },
    {
      "epoch": 0.041033762784162243,
      "grad_norm": 0.9838600754737854,
      "learning_rate": 0.00019179344056914025,
      "loss": 4.1159,
      "step": 44020
    },
    {
      "epoch": 0.04104308440224134,
      "grad_norm": 0.932729959487915,
      "learning_rate": 0.00019179157624704506,
      "loss": 4.0181,
      "step": 44030
    },
    {
      "epoch": 0.041052406020320426,
      "grad_norm": 1.0501974821090698,
      "learning_rate": 0.00019178971192494986,
      "loss": 3.8557,
      "step": 44040
    },
    {
      "epoch": 0.04106172763839952,
      "grad_norm": 0.8952295780181885,
      "learning_rate": 0.00019178784760285467,
      "loss": 4.0296,
      "step": 44050
    },
    {
      "epoch": 0.041071049256478614,
      "grad_norm": 0.8621120452880859,
      "learning_rate": 0.00019178598328075945,
      "loss": 4.0498,
      "step": 44060
    },
    {
      "epoch": 0.0410803708745577,
      "grad_norm": 1.248132586479187,
      "learning_rate": 0.00019178411895866425,
      "loss": 4.0994,
      "step": 44070
    },
    {
      "epoch": 0.041089692492636797,
      "grad_norm": 0.8867339491844177,
      "learning_rate": 0.00019178225463656906,
      "loss": 4.0595,
      "step": 44080
    },
    {
      "epoch": 0.041099014110715884,
      "grad_norm": 1.4687039852142334,
      "learning_rate": 0.00019178039031447387,
      "loss": 4.0205,
      "step": 44090
    },
    {
      "epoch": 0.04110833572879498,
      "grad_norm": 1.0821882486343384,
      "learning_rate": 0.00019177852599237867,
      "loss": 3.971,
      "step": 44100
    },
    {
      "epoch": 0.04111765734687407,
      "grad_norm": 1.3235746622085571,
      "learning_rate": 0.00019177666167028345,
      "loss": 4.0907,
      "step": 44110
    },
    {
      "epoch": 0.04112697896495316,
      "grad_norm": 1.3849395513534546,
      "learning_rate": 0.00019177479734818826,
      "loss": 4.2746,
      "step": 44120
    },
    {
      "epoch": 0.041136300583032255,
      "grad_norm": 1.1194993257522583,
      "learning_rate": 0.00019177293302609307,
      "loss": 4.0182,
      "step": 44130
    },
    {
      "epoch": 0.04114562220111135,
      "grad_norm": 1.2416858673095703,
      "learning_rate": 0.00019177106870399787,
      "loss": 3.9803,
      "step": 44140
    },
    {
      "epoch": 0.04115494381919044,
      "grad_norm": 1.1293545961380005,
      "learning_rate": 0.00019176920438190268,
      "loss": 3.9973,
      "step": 44150
    },
    {
      "epoch": 0.04116426543726953,
      "grad_norm": 1.0404678583145142,
      "learning_rate": 0.00019176734005980746,
      "loss": 3.9648,
      "step": 44160
    },
    {
      "epoch": 0.04117358705534862,
      "grad_norm": 1.2297605276107788,
      "learning_rate": 0.00019176547573771226,
      "loss": 3.9342,
      "step": 44170
    },
    {
      "epoch": 0.041182908673427714,
      "grad_norm": 1.0772370100021362,
      "learning_rate": 0.00019176361141561707,
      "loss": 4.0409,
      "step": 44180
    },
    {
      "epoch": 0.04119223029150681,
      "grad_norm": 0.8973486423492432,
      "learning_rate": 0.00019176174709352185,
      "loss": 4.2935,
      "step": 44190
    },
    {
      "epoch": 0.041201551909585896,
      "grad_norm": 0.938811182975769,
      "learning_rate": 0.00019175988277142668,
      "loss": 4.148,
      "step": 44200
    },
    {
      "epoch": 0.04121087352766499,
      "grad_norm": 0.905759334564209,
      "learning_rate": 0.00019175801844933146,
      "loss": 3.8119,
      "step": 44210
    },
    {
      "epoch": 0.041220195145744085,
      "grad_norm": 1.1140387058258057,
      "learning_rate": 0.00019175615412723627,
      "loss": 3.8008,
      "step": 44220
    },
    {
      "epoch": 0.04122951676382317,
      "grad_norm": 1.0609649419784546,
      "learning_rate": 0.00019175428980514107,
      "loss": 4.1242,
      "step": 44230
    },
    {
      "epoch": 0.04123883838190227,
      "grad_norm": 1.10807466506958,
      "learning_rate": 0.00019175242548304585,
      "loss": 4.1701,
      "step": 44240
    },
    {
      "epoch": 0.041248159999981354,
      "grad_norm": 1.087756633758545,
      "learning_rate": 0.00019175056116095066,
      "loss": 4.1188,
      "step": 44250
    },
    {
      "epoch": 0.04125748161806045,
      "grad_norm": 0.9800270795822144,
      "learning_rate": 0.00019174869683885547,
      "loss": 3.8166,
      "step": 44260
    },
    {
      "epoch": 0.04126680323613954,
      "grad_norm": 1.0183217525482178,
      "learning_rate": 0.00019174683251676027,
      "loss": 3.9353,
      "step": 44270
    },
    {
      "epoch": 0.04127612485421863,
      "grad_norm": 1.0878762006759644,
      "learning_rate": 0.00019174496819466508,
      "loss": 4.2142,
      "step": 44280
    },
    {
      "epoch": 0.041285446472297725,
      "grad_norm": 1.26195228099823,
      "learning_rate": 0.00019174310387256986,
      "loss": 4.0346,
      "step": 44290
    },
    {
      "epoch": 0.04129476809037681,
      "grad_norm": 0.8306029438972473,
      "learning_rate": 0.00019174123955047466,
      "loss": 4.0587,
      "step": 44300
    },
    {
      "epoch": 0.04130408970845591,
      "grad_norm": 0.910949170589447,
      "learning_rate": 0.00019173937522837947,
      "loss": 4.0208,
      "step": 44310
    },
    {
      "epoch": 0.041313411326535,
      "grad_norm": 1.057174801826477,
      "learning_rate": 0.00019173751090628428,
      "loss": 3.9789,
      "step": 44320
    },
    {
      "epoch": 0.04132273294461409,
      "grad_norm": 1.157354474067688,
      "learning_rate": 0.00019173564658418908,
      "loss": 4.1423,
      "step": 44330
    },
    {
      "epoch": 0.041332054562693184,
      "grad_norm": 1.129567265510559,
      "learning_rate": 0.00019173378226209386,
      "loss": 3.9939,
      "step": 44340
    },
    {
      "epoch": 0.04134137618077228,
      "grad_norm": 1.0987212657928467,
      "learning_rate": 0.00019173191793999867,
      "loss": 4.1062,
      "step": 44350
    },
    {
      "epoch": 0.041350697798851366,
      "grad_norm": 0.9967983365058899,
      "learning_rate": 0.00019173005361790345,
      "loss": 4.2027,
      "step": 44360
    },
    {
      "epoch": 0.04136001941693046,
      "grad_norm": 1.327723503112793,
      "learning_rate": 0.00019172818929580828,
      "loss": 3.8927,
      "step": 44370
    },
    {
      "epoch": 0.04136934103500955,
      "grad_norm": 1.2333948612213135,
      "learning_rate": 0.0001917263249737131,
      "loss": 4.1817,
      "step": 44380
    },
    {
      "epoch": 0.04137866265308864,
      "grad_norm": 1.3991320133209229,
      "learning_rate": 0.00019172446065161787,
      "loss": 3.9045,
      "step": 44390
    },
    {
      "epoch": 0.04138798427116774,
      "grad_norm": 1.1256482601165771,
      "learning_rate": 0.00019172259632952267,
      "loss": 3.8486,
      "step": 44400
    },
    {
      "epoch": 0.041397305889246824,
      "grad_norm": 1.1071288585662842,
      "learning_rate": 0.00019172073200742745,
      "loss": 4.0267,
      "step": 44410
    },
    {
      "epoch": 0.04140662750732592,
      "grad_norm": 1.3289450407028198,
      "learning_rate": 0.00019171886768533226,
      "loss": 3.6911,
      "step": 44420
    },
    {
      "epoch": 0.041415949125405006,
      "grad_norm": 1.1391305923461914,
      "learning_rate": 0.0001917170033632371,
      "loss": 4.2267,
      "step": 44430
    },
    {
      "epoch": 0.0414252707434841,
      "grad_norm": 1.1508920192718506,
      "learning_rate": 0.00019171513904114187,
      "loss": 3.8847,
      "step": 44440
    },
    {
      "epoch": 0.041434592361563195,
      "grad_norm": 0.9156739711761475,
      "learning_rate": 0.00019171327471904668,
      "loss": 4.1735,
      "step": 44450
    },
    {
      "epoch": 0.04144391397964228,
      "grad_norm": 1.0270541906356812,
      "learning_rate": 0.00019171141039695146,
      "loss": 4.0544,
      "step": 44460
    },
    {
      "epoch": 0.04145323559772138,
      "grad_norm": 1.1885720491409302,
      "learning_rate": 0.00019170954607485626,
      "loss": 4.0698,
      "step": 44470
    },
    {
      "epoch": 0.04146255721580047,
      "grad_norm": 0.9972388744354248,
      "learning_rate": 0.00019170768175276107,
      "loss": 4.0177,
      "step": 44480
    },
    {
      "epoch": 0.04147187883387956,
      "grad_norm": 1.202405571937561,
      "learning_rate": 0.00019170581743066588,
      "loss": 3.9803,
      "step": 44490
    },
    {
      "epoch": 0.041481200451958654,
      "grad_norm": 0.9676540493965149,
      "learning_rate": 0.00019170395310857068,
      "loss": 4.2298,
      "step": 44500
    },
    {
      "epoch": 0.04149052207003774,
      "grad_norm": 1.1098759174346924,
      "learning_rate": 0.00019170208878647546,
      "loss": 4.0615,
      "step": 44510
    },
    {
      "epoch": 0.041499843688116836,
      "grad_norm": 1.1635162830352783,
      "learning_rate": 0.00019170022446438027,
      "loss": 3.778,
      "step": 44520
    },
    {
      "epoch": 0.04150916530619593,
      "grad_norm": 0.9262370467185974,
      "learning_rate": 0.00019169836014228507,
      "loss": 3.9517,
      "step": 44530
    },
    {
      "epoch": 0.04151848692427502,
      "grad_norm": 1.2470396757125854,
      "learning_rate": 0.00019169649582018988,
      "loss": 3.8726,
      "step": 44540
    },
    {
      "epoch": 0.04152780854235411,
      "grad_norm": 1.2725844383239746,
      "learning_rate": 0.00019169463149809469,
      "loss": 3.7898,
      "step": 44550
    },
    {
      "epoch": 0.04153713016043321,
      "grad_norm": 1.035983681678772,
      "learning_rate": 0.00019169276717599947,
      "loss": 3.996,
      "step": 44560
    },
    {
      "epoch": 0.041546451778512294,
      "grad_norm": 1.199242115020752,
      "learning_rate": 0.00019169090285390427,
      "loss": 4.0451,
      "step": 44570
    },
    {
      "epoch": 0.04155577339659139,
      "grad_norm": 1.1073036193847656,
      "learning_rate": 0.00019168903853180905,
      "loss": 4.3049,
      "step": 44580
    },
    {
      "epoch": 0.041565095014670476,
      "grad_norm": 1.251526951789856,
      "learning_rate": 0.00019168717420971388,
      "loss": 4.0301,
      "step": 44590
    },
    {
      "epoch": 0.04157441663274957,
      "grad_norm": 1.0487803220748901,
      "learning_rate": 0.0001916853098876187,
      "loss": 4.1043,
      "step": 44600
    },
    {
      "epoch": 0.041583738250828665,
      "grad_norm": 0.9388105273246765,
      "learning_rate": 0.00019168344556552347,
      "loss": 4.172,
      "step": 44610
    },
    {
      "epoch": 0.04159305986890775,
      "grad_norm": 1.1070263385772705,
      "learning_rate": 0.00019168158124342828,
      "loss": 3.9996,
      "step": 44620
    },
    {
      "epoch": 0.04160238148698685,
      "grad_norm": 1.127746820449829,
      "learning_rate": 0.00019167971692133306,
      "loss": 4.1136,
      "step": 44630
    },
    {
      "epoch": 0.041611703105065935,
      "grad_norm": 1.1898733377456665,
      "learning_rate": 0.00019167785259923786,
      "loss": 4.0633,
      "step": 44640
    },
    {
      "epoch": 0.04162102472314503,
      "grad_norm": 1.228536605834961,
      "learning_rate": 0.0001916759882771427,
      "loss": 3.957,
      "step": 44650
    },
    {
      "epoch": 0.041630346341224124,
      "grad_norm": 1.081772804260254,
      "learning_rate": 0.00019167412395504747,
      "loss": 4.1351,
      "step": 44660
    },
    {
      "epoch": 0.04163966795930321,
      "grad_norm": 0.8868312239646912,
      "learning_rate": 0.00019167225963295228,
      "loss": 3.8821,
      "step": 44670
    },
    {
      "epoch": 0.041648989577382306,
      "grad_norm": 0.7946564555168152,
      "learning_rate": 0.00019167039531085706,
      "loss": 4.0304,
      "step": 44680
    },
    {
      "epoch": 0.0416583111954614,
      "grad_norm": 0.9504060745239258,
      "learning_rate": 0.00019166853098876187,
      "loss": 4.2183,
      "step": 44690
    },
    {
      "epoch": 0.04166763281354049,
      "grad_norm": 1.1945183277130127,
      "learning_rate": 0.00019166666666666667,
      "loss": 3.882,
      "step": 44700
    },
    {
      "epoch": 0.04167695443161958,
      "grad_norm": 1.077387809753418,
      "learning_rate": 0.00019166480234457148,
      "loss": 4.1545,
      "step": 44710
    },
    {
      "epoch": 0.04168627604969867,
      "grad_norm": 1.2030160427093506,
      "learning_rate": 0.00019166293802247629,
      "loss": 3.9205,
      "step": 44720
    },
    {
      "epoch": 0.041695597667777765,
      "grad_norm": 1.1955902576446533,
      "learning_rate": 0.00019166107370038106,
      "loss": 4.3243,
      "step": 44730
    },
    {
      "epoch": 0.04170491928585686,
      "grad_norm": 1.061606764793396,
      "learning_rate": 0.00019165920937828587,
      "loss": 4.206,
      "step": 44740
    },
    {
      "epoch": 0.04171424090393595,
      "grad_norm": 1.012133002281189,
      "learning_rate": 0.00019165734505619068,
      "loss": 3.9898,
      "step": 44750
    },
    {
      "epoch": 0.04172356252201504,
      "grad_norm": 1.0716084241867065,
      "learning_rate": 0.00019165548073409548,
      "loss": 3.9694,
      "step": 44760
    },
    {
      "epoch": 0.04173288414009413,
      "grad_norm": 0.8999829888343811,
      "learning_rate": 0.0001916536164120003,
      "loss": 3.9314,
      "step": 44770
    },
    {
      "epoch": 0.04174220575817322,
      "grad_norm": 1.1583040952682495,
      "learning_rate": 0.00019165175208990507,
      "loss": 4.0408,
      "step": 44780
    },
    {
      "epoch": 0.04175152737625232,
      "grad_norm": 0.9137730002403259,
      "learning_rate": 0.00019164988776780988,
      "loss": 4.1664,
      "step": 44790
    },
    {
      "epoch": 0.041760848994331405,
      "grad_norm": 0.972925066947937,
      "learning_rate": 0.00019164802344571468,
      "loss": 4.1006,
      "step": 44800
    },
    {
      "epoch": 0.0417701706124105,
      "grad_norm": 1.1091423034667969,
      "learning_rate": 0.00019164615912361946,
      "loss": 4.1582,
      "step": 44810
    },
    {
      "epoch": 0.041779492230489594,
      "grad_norm": 1.0772532224655151,
      "learning_rate": 0.0001916442948015243,
      "loss": 3.9533,
      "step": 44820
    },
    {
      "epoch": 0.04178881384856868,
      "grad_norm": 0.9867050647735596,
      "learning_rate": 0.00019164243047942907,
      "loss": 4.1334,
      "step": 44830
    },
    {
      "epoch": 0.041798135466647776,
      "grad_norm": 1.0317695140838623,
      "learning_rate": 0.00019164056615733388,
      "loss": 4.1319,
      "step": 44840
    },
    {
      "epoch": 0.041807457084726864,
      "grad_norm": 0.9021366238594055,
      "learning_rate": 0.00019163870183523869,
      "loss": 3.9635,
      "step": 44850
    },
    {
      "epoch": 0.04181677870280596,
      "grad_norm": 1.3166422843933105,
      "learning_rate": 0.00019163683751314346,
      "loss": 4.0232,
      "step": 44860
    },
    {
      "epoch": 0.04182610032088505,
      "grad_norm": 0.9781999588012695,
      "learning_rate": 0.00019163497319104827,
      "loss": 4.1731,
      "step": 44870
    },
    {
      "epoch": 0.04183542193896414,
      "grad_norm": 1.1169246435165405,
      "learning_rate": 0.00019163310886895308,
      "loss": 4.0814,
      "step": 44880
    },
    {
      "epoch": 0.041844743557043235,
      "grad_norm": 0.9958419799804688,
      "learning_rate": 0.00019163124454685788,
      "loss": 4.0425,
      "step": 44890
    },
    {
      "epoch": 0.04185406517512233,
      "grad_norm": 1.0720497369766235,
      "learning_rate": 0.0001916293802247627,
      "loss": 3.8856,
      "step": 44900
    },
    {
      "epoch": 0.04186338679320142,
      "grad_norm": 0.986537516117096,
      "learning_rate": 0.00019162751590266747,
      "loss": 3.8919,
      "step": 44910
    },
    {
      "epoch": 0.04187270841128051,
      "grad_norm": 1.1617066860198975,
      "learning_rate": 0.00019162565158057228,
      "loss": 3.9805,
      "step": 44920
    },
    {
      "epoch": 0.0418820300293596,
      "grad_norm": 1.2911022901535034,
      "learning_rate": 0.00019162378725847708,
      "loss": 4.0566,
      "step": 44930
    },
    {
      "epoch": 0.04189135164743869,
      "grad_norm": 1.2987459897994995,
      "learning_rate": 0.0001916219229363819,
      "loss": 3.894,
      "step": 44940
    },
    {
      "epoch": 0.04190067326551779,
      "grad_norm": 1.1248220205307007,
      "learning_rate": 0.0001916200586142867,
      "loss": 4.0142,
      "step": 44950
    },
    {
      "epoch": 0.041909994883596875,
      "grad_norm": 1.3167048692703247,
      "learning_rate": 0.00019161819429219147,
      "loss": 4.0589,
      "step": 44960
    },
    {
      "epoch": 0.04191931650167597,
      "grad_norm": 1.2481271028518677,
      "learning_rate": 0.00019161632997009628,
      "loss": 4.2289,
      "step": 44970
    },
    {
      "epoch": 0.04192863811975506,
      "grad_norm": 1.1402570009231567,
      "learning_rate": 0.0001916144656480011,
      "loss": 3.966,
      "step": 44980
    },
    {
      "epoch": 0.04193795973783415,
      "grad_norm": 1.3384006023406982,
      "learning_rate": 0.0001916126013259059,
      "loss": 3.9774,
      "step": 44990
    },
    {
      "epoch": 0.041947281355913246,
      "grad_norm": 1.057352900505066,
      "learning_rate": 0.0001916107370038107,
      "loss": 4.0381,
      "step": 45000
    },
    {
      "epoch": 0.041956602973992334,
      "grad_norm": 1.3759028911590576,
      "learning_rate": 0.00019160887268171548,
      "loss": 3.8152,
      "step": 45010
    },
    {
      "epoch": 0.04196592459207143,
      "grad_norm": 1.1846784353256226,
      "learning_rate": 0.00019160700835962028,
      "loss": 3.9172,
      "step": 45020
    },
    {
      "epoch": 0.04197524621015052,
      "grad_norm": 0.964036762714386,
      "learning_rate": 0.00019160514403752506,
      "loss": 3.9748,
      "step": 45030
    },
    {
      "epoch": 0.04198456782822961,
      "grad_norm": 1.2088053226470947,
      "learning_rate": 0.0001916032797154299,
      "loss": 3.9888,
      "step": 45040
    },
    {
      "epoch": 0.041993889446308705,
      "grad_norm": 1.0547406673431396,
      "learning_rate": 0.0001916014153933347,
      "loss": 4.0124,
      "step": 45050
    },
    {
      "epoch": 0.04200321106438779,
      "grad_norm": 1.193061113357544,
      "learning_rate": 0.00019159955107123948,
      "loss": 4.1692,
      "step": 45060
    },
    {
      "epoch": 0.04201253268246689,
      "grad_norm": 1.207680344581604,
      "learning_rate": 0.0001915976867491443,
      "loss": 4.0507,
      "step": 45070
    },
    {
      "epoch": 0.04202185430054598,
      "grad_norm": 1.2424324750900269,
      "learning_rate": 0.00019159582242704907,
      "loss": 4.061,
      "step": 45080
    },
    {
      "epoch": 0.04203117591862507,
      "grad_norm": 0.9985347986221313,
      "learning_rate": 0.00019159395810495387,
      "loss": 3.9054,
      "step": 45090
    },
    {
      "epoch": 0.04204049753670416,
      "grad_norm": 1.090293526649475,
      "learning_rate": 0.0001915920937828587,
      "loss": 4.1317,
      "step": 45100
    },
    {
      "epoch": 0.04204981915478325,
      "grad_norm": 1.0869982242584229,
      "learning_rate": 0.0001915902294607635,
      "loss": 4.1022,
      "step": 45110
    },
    {
      "epoch": 0.042059140772862345,
      "grad_norm": 1.0664520263671875,
      "learning_rate": 0.0001915883651386683,
      "loss": 4.2496,
      "step": 45120
    },
    {
      "epoch": 0.04206846239094144,
      "grad_norm": 1.009251594543457,
      "learning_rate": 0.00019158650081657307,
      "loss": 3.9114,
      "step": 45130
    },
    {
      "epoch": 0.04207778400902053,
      "grad_norm": 1.0621063709259033,
      "learning_rate": 0.00019158463649447788,
      "loss": 4.0737,
      "step": 45140
    },
    {
      "epoch": 0.04208710562709962,
      "grad_norm": 1.1431525945663452,
      "learning_rate": 0.00019158277217238269,
      "loss": 3.9837,
      "step": 45150
    },
    {
      "epoch": 0.042096427245178716,
      "grad_norm": 1.1395057439804077,
      "learning_rate": 0.0001915809078502875,
      "loss": 4.2252,
      "step": 45160
    },
    {
      "epoch": 0.042105748863257804,
      "grad_norm": 0.9826952815055847,
      "learning_rate": 0.0001915790435281923,
      "loss": 4.2147,
      "step": 45170
    },
    {
      "epoch": 0.0421150704813369,
      "grad_norm": 0.9974595904350281,
      "learning_rate": 0.00019157717920609708,
      "loss": 4.0946,
      "step": 45180
    },
    {
      "epoch": 0.042124392099415986,
      "grad_norm": 1.0840483903884888,
      "learning_rate": 0.00019157531488400188,
      "loss": 4.1103,
      "step": 45190
    },
    {
      "epoch": 0.04213371371749508,
      "grad_norm": 1.3291698694229126,
      "learning_rate": 0.0001915734505619067,
      "loss": 4.2596,
      "step": 45200
    },
    {
      "epoch": 0.042143035335574175,
      "grad_norm": 1.054892659187317,
      "learning_rate": 0.0001915715862398115,
      "loss": 4.0445,
      "step": 45210
    },
    {
      "epoch": 0.04215235695365326,
      "grad_norm": 1.4400975704193115,
      "learning_rate": 0.0001915697219177163,
      "loss": 4.0238,
      "step": 45220
    },
    {
      "epoch": 0.04216167857173236,
      "grad_norm": 1.113810658454895,
      "learning_rate": 0.00019156785759562108,
      "loss": 3.996,
      "step": 45230
    },
    {
      "epoch": 0.04217100018981145,
      "grad_norm": 1.0038926601409912,
      "learning_rate": 0.0001915659932735259,
      "loss": 4.1522,
      "step": 45240
    },
    {
      "epoch": 0.04218032180789054,
      "grad_norm": 1.284421682357788,
      "learning_rate": 0.0001915641289514307,
      "loss": 3.8311,
      "step": 45250
    },
    {
      "epoch": 0.04218964342596963,
      "grad_norm": 1.1171334981918335,
      "learning_rate": 0.00019156226462933547,
      "loss": 4.0466,
      "step": 45260
    },
    {
      "epoch": 0.04219896504404872,
      "grad_norm": 1.0384961366653442,
      "learning_rate": 0.0001915604003072403,
      "loss": 3.8448,
      "step": 45270
    },
    {
      "epoch": 0.042208286662127815,
      "grad_norm": 0.9461719393730164,
      "learning_rate": 0.00019155853598514509,
      "loss": 3.9559,
      "step": 45280
    },
    {
      "epoch": 0.04221760828020691,
      "grad_norm": 1.163513422012329,
      "learning_rate": 0.0001915566716630499,
      "loss": 4.0466,
      "step": 45290
    },
    {
      "epoch": 0.042226929898286,
      "grad_norm": 1.2152289152145386,
      "learning_rate": 0.0001915548073409547,
      "loss": 4.1249,
      "step": 45300
    },
    {
      "epoch": 0.04223625151636509,
      "grad_norm": 0.9686734080314636,
      "learning_rate": 0.00019155294301885948,
      "loss": 4.0335,
      "step": 45310
    },
    {
      "epoch": 0.04224557313444418,
      "grad_norm": 1.0827611684799194,
      "learning_rate": 0.00019155107869676428,
      "loss": 4.0513,
      "step": 45320
    },
    {
      "epoch": 0.042254894752523274,
      "grad_norm": 1.0571357011795044,
      "learning_rate": 0.0001915492143746691,
      "loss": 4.1247,
      "step": 45330
    },
    {
      "epoch": 0.04226421637060237,
      "grad_norm": 0.8332387804985046,
      "learning_rate": 0.0001915473500525739,
      "loss": 4.0306,
      "step": 45340
    },
    {
      "epoch": 0.042273537988681456,
      "grad_norm": 0.9619254469871521,
      "learning_rate": 0.0001915454857304787,
      "loss": 4.1904,
      "step": 45350
    },
    {
      "epoch": 0.04228285960676055,
      "grad_norm": 1.0820086002349854,
      "learning_rate": 0.00019154362140838348,
      "loss": 3.9855,
      "step": 45360
    },
    {
      "epoch": 0.042292181224839645,
      "grad_norm": 1.1650296449661255,
      "learning_rate": 0.0001915417570862883,
      "loss": 3.8359,
      "step": 45370
    },
    {
      "epoch": 0.04230150284291873,
      "grad_norm": 1.0146751403808594,
      "learning_rate": 0.0001915398927641931,
      "loss": 4.0401,
      "step": 45380
    },
    {
      "epoch": 0.04231082446099783,
      "grad_norm": 1.0387269258499146,
      "learning_rate": 0.0001915380284420979,
      "loss": 3.9448,
      "step": 45390
    },
    {
      "epoch": 0.042320146079076915,
      "grad_norm": 1.1642860174179077,
      "learning_rate": 0.0001915361641200027,
      "loss": 4.1601,
      "step": 45400
    },
    {
      "epoch": 0.04232946769715601,
      "grad_norm": 1.0487117767333984,
      "learning_rate": 0.0001915342997979075,
      "loss": 4.0234,
      "step": 45410
    },
    {
      "epoch": 0.042338789315235104,
      "grad_norm": 1.1006386280059814,
      "learning_rate": 0.0001915324354758123,
      "loss": 4.0601,
      "step": 45420
    },
    {
      "epoch": 0.04234811093331419,
      "grad_norm": 1.0925596952438354,
      "learning_rate": 0.0001915305711537171,
      "loss": 4.0839,
      "step": 45430
    },
    {
      "epoch": 0.042357432551393286,
      "grad_norm": 1.1534074544906616,
      "learning_rate": 0.0001915287068316219,
      "loss": 3.9555,
      "step": 45440
    },
    {
      "epoch": 0.04236675416947237,
      "grad_norm": 1.096198558807373,
      "learning_rate": 0.0001915268425095267,
      "loss": 3.9548,
      "step": 45450
    },
    {
      "epoch": 0.04237607578755147,
      "grad_norm": 1.3378134965896606,
      "learning_rate": 0.0001915249781874315,
      "loss": 4.0649,
      "step": 45460
    },
    {
      "epoch": 0.04238539740563056,
      "grad_norm": 1.0337269306182861,
      "learning_rate": 0.0001915231138653363,
      "loss": 4.1315,
      "step": 45470
    },
    {
      "epoch": 0.04239471902370965,
      "grad_norm": 1.1550445556640625,
      "learning_rate": 0.00019152124954324108,
      "loss": 4.0284,
      "step": 45480
    },
    {
      "epoch": 0.042404040641788744,
      "grad_norm": 0.9327902793884277,
      "learning_rate": 0.0001915193852211459,
      "loss": 4.0456,
      "step": 45490
    },
    {
      "epoch": 0.04241336225986784,
      "grad_norm": 0.9583361148834229,
      "learning_rate": 0.00019151752089905072,
      "loss": 4.222,
      "step": 45500
    },
    {
      "epoch": 0.042422683877946926,
      "grad_norm": 0.984721839427948,
      "learning_rate": 0.0001915156565769555,
      "loss": 4.2452,
      "step": 45510
    },
    {
      "epoch": 0.04243200549602602,
      "grad_norm": 0.9431248903274536,
      "learning_rate": 0.0001915137922548603,
      "loss": 4.0742,
      "step": 45520
    },
    {
      "epoch": 0.04244132711410511,
      "grad_norm": 0.8152719140052795,
      "learning_rate": 0.00019151192793276508,
      "loss": 4.1484,
      "step": 45530
    },
    {
      "epoch": 0.0424506487321842,
      "grad_norm": 0.9871390461921692,
      "learning_rate": 0.0001915100636106699,
      "loss": 4.1216,
      "step": 45540
    },
    {
      "epoch": 0.0424599703502633,
      "grad_norm": 1.2909971475601196,
      "learning_rate": 0.00019150819928857472,
      "loss": 4.1098,
      "step": 45550
    },
    {
      "epoch": 0.042469291968342385,
      "grad_norm": 0.9385109543800354,
      "learning_rate": 0.0001915063349664795,
      "loss": 4.1751,
      "step": 45560
    },
    {
      "epoch": 0.04247861358642148,
      "grad_norm": 1.1952048540115356,
      "learning_rate": 0.0001915044706443843,
      "loss": 4.1453,
      "step": 45570
    },
    {
      "epoch": 0.042487935204500574,
      "grad_norm": 1.119652509689331,
      "learning_rate": 0.00019150260632228909,
      "loss": 4.2289,
      "step": 45580
    },
    {
      "epoch": 0.04249725682257966,
      "grad_norm": 1.0517446994781494,
      "learning_rate": 0.0001915007420001939,
      "loss": 4.0575,
      "step": 45590
    },
    {
      "epoch": 0.042506578440658756,
      "grad_norm": 0.9056764245033264,
      "learning_rate": 0.0001914988776780987,
      "loss": 4.0687,
      "step": 45600
    },
    {
      "epoch": 0.04251590005873784,
      "grad_norm": 1.3271816968917847,
      "learning_rate": 0.0001914970133560035,
      "loss": 3.9389,
      "step": 45610
    },
    {
      "epoch": 0.04252522167681694,
      "grad_norm": 1.1016104221343994,
      "learning_rate": 0.0001914951490339083,
      "loss": 3.7852,
      "step": 45620
    },
    {
      "epoch": 0.04253454329489603,
      "grad_norm": 1.0848432779312134,
      "learning_rate": 0.0001914932847118131,
      "loss": 4.1182,
      "step": 45630
    },
    {
      "epoch": 0.04254386491297512,
      "grad_norm": 0.8664813041687012,
      "learning_rate": 0.0001914914203897179,
      "loss": 4.0595,
      "step": 45640
    },
    {
      "epoch": 0.042553186531054214,
      "grad_norm": 1.0566294193267822,
      "learning_rate": 0.0001914895560676227,
      "loss": 3.9195,
      "step": 45650
    },
    {
      "epoch": 0.0425625081491333,
      "grad_norm": 0.942233145236969,
      "learning_rate": 0.0001914876917455275,
      "loss": 3.8966,
      "step": 45660
    },
    {
      "epoch": 0.042571829767212396,
      "grad_norm": 1.186556339263916,
      "learning_rate": 0.00019148582742343232,
      "loss": 4.004,
      "step": 45670
    },
    {
      "epoch": 0.04258115138529149,
      "grad_norm": 1.1208724975585938,
      "learning_rate": 0.0001914839631013371,
      "loss": 4.1052,
      "step": 45680
    },
    {
      "epoch": 0.04259047300337058,
      "grad_norm": 1.1067304611206055,
      "learning_rate": 0.0001914820987792419,
      "loss": 4.0924,
      "step": 45690
    },
    {
      "epoch": 0.04259979462144967,
      "grad_norm": 1.1276755332946777,
      "learning_rate": 0.0001914802344571467,
      "loss": 4.1955,
      "step": 45700
    },
    {
      "epoch": 0.04260911623952877,
      "grad_norm": 1.0875706672668457,
      "learning_rate": 0.00019147837013505149,
      "loss": 4.0776,
      "step": 45710
    },
    {
      "epoch": 0.042618437857607855,
      "grad_norm": 1.0052905082702637,
      "learning_rate": 0.00019147650581295632,
      "loss": 3.9886,
      "step": 45720
    },
    {
      "epoch": 0.04262775947568695,
      "grad_norm": 1.0125654935836792,
      "learning_rate": 0.0001914746414908611,
      "loss": 4.2401,
      "step": 45730
    },
    {
      "epoch": 0.04263708109376604,
      "grad_norm": 1.1230168342590332,
      "learning_rate": 0.0001914727771687659,
      "loss": 3.912,
      "step": 45740
    },
    {
      "epoch": 0.04264640271184513,
      "grad_norm": 0.9632102847099304,
      "learning_rate": 0.0001914709128466707,
      "loss": 4.1407,
      "step": 45750
    },
    {
      "epoch": 0.042655724329924226,
      "grad_norm": 1.0704128742218018,
      "learning_rate": 0.0001914690485245755,
      "loss": 4.0137,
      "step": 45760
    },
    {
      "epoch": 0.04266504594800331,
      "grad_norm": 1.1371448040008545,
      "learning_rate": 0.0001914671842024803,
      "loss": 4.0398,
      "step": 45770
    },
    {
      "epoch": 0.04267436756608241,
      "grad_norm": 1.1301398277282715,
      "learning_rate": 0.0001914653198803851,
      "loss": 4.1944,
      "step": 45780
    },
    {
      "epoch": 0.042683689184161495,
      "grad_norm": 1.1424384117126465,
      "learning_rate": 0.0001914634555582899,
      "loss": 4.14,
      "step": 45790
    },
    {
      "epoch": 0.04269301080224059,
      "grad_norm": 1.1274391412734985,
      "learning_rate": 0.00019146159123619472,
      "loss": 4.1162,
      "step": 45800
    },
    {
      "epoch": 0.042702332420319684,
      "grad_norm": 1.1795706748962402,
      "learning_rate": 0.0001914597269140995,
      "loss": 3.9284,
      "step": 45810
    },
    {
      "epoch": 0.04271165403839877,
      "grad_norm": 1.0076411962509155,
      "learning_rate": 0.0001914578625920043,
      "loss": 3.9675,
      "step": 45820
    },
    {
      "epoch": 0.042720975656477866,
      "grad_norm": 1.0093005895614624,
      "learning_rate": 0.0001914559982699091,
      "loss": 4.0153,
      "step": 45830
    },
    {
      "epoch": 0.04273029727455696,
      "grad_norm": 1.3890961408615112,
      "learning_rate": 0.00019145413394781391,
      "loss": 4.0419,
      "step": 45840
    },
    {
      "epoch": 0.04273961889263605,
      "grad_norm": 1.0422983169555664,
      "learning_rate": 0.0001914522696257187,
      "loss": 4.0092,
      "step": 45850
    },
    {
      "epoch": 0.04274894051071514,
      "grad_norm": 1.0370548963546753,
      "learning_rate": 0.0001914504053036235,
      "loss": 3.9324,
      "step": 45860
    },
    {
      "epoch": 0.04275826212879423,
      "grad_norm": 0.9664244651794434,
      "learning_rate": 0.0001914485409815283,
      "loss": 4.0302,
      "step": 45870
    },
    {
      "epoch": 0.042767583746873325,
      "grad_norm": 1.061559796333313,
      "learning_rate": 0.0001914466766594331,
      "loss": 4.0105,
      "step": 45880
    },
    {
      "epoch": 0.04277690536495242,
      "grad_norm": 1.2157294750213623,
      "learning_rate": 0.00019144481233733792,
      "loss": 4.0053,
      "step": 45890
    },
    {
      "epoch": 0.04278622698303151,
      "grad_norm": 1.1508764028549194,
      "learning_rate": 0.0001914429480152427,
      "loss": 4.0549,
      "step": 45900
    },
    {
      "epoch": 0.0427955486011106,
      "grad_norm": 1.0708227157592773,
      "learning_rate": 0.0001914410836931475,
      "loss": 4.0577,
      "step": 45910
    },
    {
      "epoch": 0.042804870219189696,
      "grad_norm": 1.124234676361084,
      "learning_rate": 0.0001914392193710523,
      "loss": 3.8204,
      "step": 45920
    },
    {
      "epoch": 0.042814191837268784,
      "grad_norm": 1.1838384866714478,
      "learning_rate": 0.0001914373550489571,
      "loss": 3.903,
      "step": 45930
    },
    {
      "epoch": 0.04282351345534788,
      "grad_norm": 1.0489598512649536,
      "learning_rate": 0.00019143549072686192,
      "loss": 3.9969,
      "step": 45940
    },
    {
      "epoch": 0.042832835073426966,
      "grad_norm": 1.0108591318130493,
      "learning_rate": 0.0001914336264047667,
      "loss": 4.0577,
      "step": 45950
    },
    {
      "epoch": 0.04284215669150606,
      "grad_norm": 0.9670005440711975,
      "learning_rate": 0.0001914317620826715,
      "loss": 4.1818,
      "step": 45960
    },
    {
      "epoch": 0.042851478309585155,
      "grad_norm": 1.144566297531128,
      "learning_rate": 0.00019142989776057631,
      "loss": 3.8572,
      "step": 45970
    },
    {
      "epoch": 0.04286079992766424,
      "grad_norm": 1.1628328561782837,
      "learning_rate": 0.0001914280334384811,
      "loss": 4.1289,
      "step": 45980
    },
    {
      "epoch": 0.04287012154574334,
      "grad_norm": 0.8468233346939087,
      "learning_rate": 0.0001914261691163859,
      "loss": 4.2848,
      "step": 45990
    },
    {
      "epoch": 0.042879443163822424,
      "grad_norm": 1.1186472177505493,
      "learning_rate": 0.0001914243047942907,
      "loss": 4.0521,
      "step": 46000
    },
    {
      "epoch": 0.04288876478190152,
      "grad_norm": 1.1572880744934082,
      "learning_rate": 0.0001914224404721955,
      "loss": 4.1117,
      "step": 46010
    },
    {
      "epoch": 0.04289808639998061,
      "grad_norm": 1.149366021156311,
      "learning_rate": 0.00019142057615010032,
      "loss": 3.8169,
      "step": 46020
    },
    {
      "epoch": 0.0429074080180597,
      "grad_norm": 1.0881938934326172,
      "learning_rate": 0.0001914187118280051,
      "loss": 4.122,
      "step": 46030
    },
    {
      "epoch": 0.042916729636138795,
      "grad_norm": 1.1438206434249878,
      "learning_rate": 0.0001914168475059099,
      "loss": 3.9234,
      "step": 46040
    },
    {
      "epoch": 0.04292605125421789,
      "grad_norm": 0.8983590006828308,
      "learning_rate": 0.0001914149831838147,
      "loss": 4.1907,
      "step": 46050
    },
    {
      "epoch": 0.04293537287229698,
      "grad_norm": 1.1610183715820312,
      "learning_rate": 0.00019141311886171952,
      "loss": 3.88,
      "step": 46060
    },
    {
      "epoch": 0.04294469449037607,
      "grad_norm": 1.071835994720459,
      "learning_rate": 0.00019141125453962432,
      "loss": 4.1656,
      "step": 46070
    },
    {
      "epoch": 0.04295401610845516,
      "grad_norm": 1.13447105884552,
      "learning_rate": 0.0001914093902175291,
      "loss": 3.9378,
      "step": 46080
    },
    {
      "epoch": 0.042963337726534254,
      "grad_norm": 0.9978758096694946,
      "learning_rate": 0.0001914075258954339,
      "loss": 3.9389,
      "step": 46090
    },
    {
      "epoch": 0.04297265934461335,
      "grad_norm": 1.1692676544189453,
      "learning_rate": 0.0001914056615733387,
      "loss": 4.0509,
      "step": 46100
    },
    {
      "epoch": 0.042981980962692436,
      "grad_norm": 1.2061854600906372,
      "learning_rate": 0.00019140379725124352,
      "loss": 4.0991,
      "step": 46110
    },
    {
      "epoch": 0.04299130258077153,
      "grad_norm": 0.8628930449485779,
      "learning_rate": 0.00019140193292914833,
      "loss": 4.0124,
      "step": 46120
    },
    {
      "epoch": 0.043000624198850625,
      "grad_norm": 1.1302820444107056,
      "learning_rate": 0.0001914000686070531,
      "loss": 4.0406,
      "step": 46130
    },
    {
      "epoch": 0.04300994581692971,
      "grad_norm": 1.1649473905563354,
      "learning_rate": 0.0001913982042849579,
      "loss": 4.0195,
      "step": 46140
    },
    {
      "epoch": 0.04301926743500881,
      "grad_norm": 0.8508267998695374,
      "learning_rate": 0.0001913963399628627,
      "loss": 4.2426,
      "step": 46150
    },
    {
      "epoch": 0.043028589053087894,
      "grad_norm": 0.9316275119781494,
      "learning_rate": 0.0001913944756407675,
      "loss": 3.9819,
      "step": 46160
    },
    {
      "epoch": 0.04303791067116699,
      "grad_norm": 1.0139033794403076,
      "learning_rate": 0.00019139261131867233,
      "loss": 4.0703,
      "step": 46170
    },
    {
      "epoch": 0.04304723228924608,
      "grad_norm": 1.1622366905212402,
      "learning_rate": 0.0001913907469965771,
      "loss": 3.8444,
      "step": 46180
    },
    {
      "epoch": 0.04305655390732517,
      "grad_norm": 1.0279303789138794,
      "learning_rate": 0.00019138888267448192,
      "loss": 4.2212,
      "step": 46190
    },
    {
      "epoch": 0.043065875525404265,
      "grad_norm": 0.9845932722091675,
      "learning_rate": 0.0001913870183523867,
      "loss": 4.0106,
      "step": 46200
    },
    {
      "epoch": 0.04307519714348335,
      "grad_norm": 1.27877938747406,
      "learning_rate": 0.0001913851540302915,
      "loss": 4.2804,
      "step": 46210
    },
    {
      "epoch": 0.04308451876156245,
      "grad_norm": 0.9878559112548828,
      "learning_rate": 0.0001913832897081963,
      "loss": 4.0498,
      "step": 46220
    },
    {
      "epoch": 0.04309384037964154,
      "grad_norm": 1.2638908624649048,
      "learning_rate": 0.00019138142538610112,
      "loss": 3.9978,
      "step": 46230
    },
    {
      "epoch": 0.04310316199772063,
      "grad_norm": 1.0615512132644653,
      "learning_rate": 0.00019137956106400592,
      "loss": 3.8582,
      "step": 46240
    },
    {
      "epoch": 0.043112483615799724,
      "grad_norm": 1.0848280191421509,
      "learning_rate": 0.0001913776967419107,
      "loss": 3.9457,
      "step": 46250
    },
    {
      "epoch": 0.04312180523387882,
      "grad_norm": 1.0811315774917603,
      "learning_rate": 0.0001913758324198155,
      "loss": 3.9706,
      "step": 46260
    },
    {
      "epoch": 0.043131126851957906,
      "grad_norm": 0.9630025029182434,
      "learning_rate": 0.00019137396809772031,
      "loss": 4.1354,
      "step": 46270
    },
    {
      "epoch": 0.043140448470037,
      "grad_norm": 0.8643697500228882,
      "learning_rate": 0.00019137210377562512,
      "loss": 4.0101,
      "step": 46280
    },
    {
      "epoch": 0.04314977008811609,
      "grad_norm": 1.203094482421875,
      "learning_rate": 0.00019137023945352993,
      "loss": 3.9769,
      "step": 46290
    },
    {
      "epoch": 0.04315909170619518,
      "grad_norm": 1.2221896648406982,
      "learning_rate": 0.0001913683751314347,
      "loss": 4.2032,
      "step": 46300
    },
    {
      "epoch": 0.04316841332427428,
      "grad_norm": 1.9472101926803589,
      "learning_rate": 0.0001913665108093395,
      "loss": 3.8222,
      "step": 46310
    },
    {
      "epoch": 0.043177734942353364,
      "grad_norm": 1.1148091554641724,
      "learning_rate": 0.00019136464648724432,
      "loss": 3.8675,
      "step": 46320
    },
    {
      "epoch": 0.04318705656043246,
      "grad_norm": 0.9763997197151184,
      "learning_rate": 0.00019136278216514912,
      "loss": 4.0378,
      "step": 46330
    },
    {
      "epoch": 0.043196378178511546,
      "grad_norm": 1.266046166419983,
      "learning_rate": 0.00019136091784305393,
      "loss": 4.1611,
      "step": 46340
    },
    {
      "epoch": 0.04320569979659064,
      "grad_norm": 1.1442521810531616,
      "learning_rate": 0.0001913590535209587,
      "loss": 4.1853,
      "step": 46350
    },
    {
      "epoch": 0.043215021414669735,
      "grad_norm": 1.0464118719100952,
      "learning_rate": 0.00019135718919886352,
      "loss": 4.0178,
      "step": 46360
    },
    {
      "epoch": 0.04322434303274882,
      "grad_norm": 1.099380612373352,
      "learning_rate": 0.00019135532487676832,
      "loss": 4.1557,
      "step": 46370
    },
    {
      "epoch": 0.04323366465082792,
      "grad_norm": 1.1001404523849487,
      "learning_rate": 0.0001913534605546731,
      "loss": 4.0732,
      "step": 46380
    },
    {
      "epoch": 0.04324298626890701,
      "grad_norm": 1.1574534177780151,
      "learning_rate": 0.00019135159623257794,
      "loss": 3.9402,
      "step": 46390
    },
    {
      "epoch": 0.0432523078869861,
      "grad_norm": 1.13152277469635,
      "learning_rate": 0.00019134973191048271,
      "loss": 4.171,
      "step": 46400
    },
    {
      "epoch": 0.043261629505065194,
      "grad_norm": 1.1092981100082397,
      "learning_rate": 0.00019134786758838752,
      "loss": 4.2283,
      "step": 46410
    },
    {
      "epoch": 0.04327095112314428,
      "grad_norm": 1.1466974020004272,
      "learning_rate": 0.00019134600326629233,
      "loss": 3.9023,
      "step": 46420
    },
    {
      "epoch": 0.043280272741223376,
      "grad_norm": 1.2158465385437012,
      "learning_rate": 0.0001913441389441971,
      "loss": 3.879,
      "step": 46430
    },
    {
      "epoch": 0.04328959435930247,
      "grad_norm": 1.0327425003051758,
      "learning_rate": 0.0001913422746221019,
      "loss": 3.9588,
      "step": 46440
    },
    {
      "epoch": 0.04329891597738156,
      "grad_norm": 1.0153011083602905,
      "learning_rate": 0.00019134041030000672,
      "loss": 4.1686,
      "step": 46450
    },
    {
      "epoch": 0.04330823759546065,
      "grad_norm": 1.154669165611267,
      "learning_rate": 0.00019133854597791153,
      "loss": 4.2206,
      "step": 46460
    },
    {
      "epoch": 0.04331755921353975,
      "grad_norm": 1.0159635543823242,
      "learning_rate": 0.00019133668165581633,
      "loss": 3.956,
      "step": 46470
    },
    {
      "epoch": 0.043326880831618834,
      "grad_norm": 1.125349521636963,
      "learning_rate": 0.0001913348173337211,
      "loss": 3.8423,
      "step": 46480
    },
    {
      "epoch": 0.04333620244969793,
      "grad_norm": 1.103493332862854,
      "learning_rate": 0.00019133295301162592,
      "loss": 3.9912,
      "step": 46490
    },
    {
      "epoch": 0.043345524067777016,
      "grad_norm": 1.2297821044921875,
      "learning_rate": 0.00019133108868953072,
      "loss": 4.1759,
      "step": 46500
    },
    {
      "epoch": 0.04335484568585611,
      "grad_norm": 1.0467761754989624,
      "learning_rate": 0.00019132922436743553,
      "loss": 4.0757,
      "step": 46510
    },
    {
      "epoch": 0.043364167303935205,
      "grad_norm": 0.9943503737449646,
      "learning_rate": 0.00019132736004534034,
      "loss": 3.933,
      "step": 46520
    },
    {
      "epoch": 0.04337348892201429,
      "grad_norm": 1.4645049571990967,
      "learning_rate": 0.00019132549572324512,
      "loss": 4.0425,
      "step": 46530
    },
    {
      "epoch": 0.04338281054009339,
      "grad_norm": 1.2078981399536133,
      "learning_rate": 0.00019132363140114992,
      "loss": 4.1359,
      "step": 46540
    },
    {
      "epoch": 0.043392132158172475,
      "grad_norm": 1.075526475906372,
      "learning_rate": 0.0001913217670790547,
      "loss": 4.0012,
      "step": 46550
    },
    {
      "epoch": 0.04340145377625157,
      "grad_norm": 1.0083037614822388,
      "learning_rate": 0.00019131990275695953,
      "loss": 4.0623,
      "step": 46560
    },
    {
      "epoch": 0.043410775394330664,
      "grad_norm": 2.201094627380371,
      "learning_rate": 0.00019131803843486434,
      "loss": 4.1605,
      "step": 46570
    },
    {
      "epoch": 0.04342009701240975,
      "grad_norm": 1.2110408544540405,
      "learning_rate": 0.00019131617411276912,
      "loss": 3.6971,
      "step": 46580
    },
    {
      "epoch": 0.043429418630488846,
      "grad_norm": 1.1424145698547363,
      "learning_rate": 0.00019131430979067393,
      "loss": 4.0028,
      "step": 46590
    },
    {
      "epoch": 0.04343874024856794,
      "grad_norm": 1.015932559967041,
      "learning_rate": 0.0001913124454685787,
      "loss": 4.0076,
      "step": 46600
    },
    {
      "epoch": 0.04344806186664703,
      "grad_norm": 1.0958811044692993,
      "learning_rate": 0.0001913105811464835,
      "loss": 3.9709,
      "step": 46610
    },
    {
      "epoch": 0.04345738348472612,
      "grad_norm": 1.1850529909133911,
      "learning_rate": 0.00019130871682438835,
      "loss": 3.9798,
      "step": 46620
    },
    {
      "epoch": 0.04346670510280521,
      "grad_norm": 1.1153439283370972,
      "learning_rate": 0.00019130685250229312,
      "loss": 3.9578,
      "step": 46630
    },
    {
      "epoch": 0.043476026720884305,
      "grad_norm": 1.0161492824554443,
      "learning_rate": 0.00019130498818019793,
      "loss": 4.1112,
      "step": 46640
    },
    {
      "epoch": 0.0434853483389634,
      "grad_norm": 1.2725374698638916,
      "learning_rate": 0.0001913031238581027,
      "loss": 4.0121,
      "step": 46650
    },
    {
      "epoch": 0.04349466995704249,
      "grad_norm": 1.1891107559204102,
      "learning_rate": 0.00019130125953600752,
      "loss": 4.0249,
      "step": 46660
    },
    {
      "epoch": 0.04350399157512158,
      "grad_norm": 1.1563851833343506,
      "learning_rate": 0.00019129939521391232,
      "loss": 4.0142,
      "step": 46670
    },
    {
      "epoch": 0.04351331319320067,
      "grad_norm": 1.063944697380066,
      "learning_rate": 0.00019129753089181713,
      "loss": 4.0515,
      "step": 46680
    },
    {
      "epoch": 0.04352263481127976,
      "grad_norm": 1.3096513748168945,
      "learning_rate": 0.00019129566656972194,
      "loss": 4.198,
      "step": 46690
    },
    {
      "epoch": 0.04353195642935886,
      "grad_norm": 1.2315565347671509,
      "learning_rate": 0.00019129380224762671,
      "loss": 4.1739,
      "step": 46700
    },
    {
      "epoch": 0.043541278047437945,
      "grad_norm": 0.956881046295166,
      "learning_rate": 0.00019129193792553152,
      "loss": 4.1269,
      "step": 46710
    },
    {
      "epoch": 0.04355059966551704,
      "grad_norm": 1.0699670314788818,
      "learning_rate": 0.00019129007360343633,
      "loss": 4.0746,
      "step": 46720
    },
    {
      "epoch": 0.043559921283596134,
      "grad_norm": 1.116010069847107,
      "learning_rate": 0.00019128820928134113,
      "loss": 4.1062,
      "step": 46730
    },
    {
      "epoch": 0.04356924290167522,
      "grad_norm": 1.0501189231872559,
      "learning_rate": 0.00019128634495924594,
      "loss": 4.0925,
      "step": 46740
    },
    {
      "epoch": 0.043578564519754316,
      "grad_norm": 0.897148847579956,
      "learning_rate": 0.00019128448063715072,
      "loss": 4.0624,
      "step": 46750
    },
    {
      "epoch": 0.043587886137833404,
      "grad_norm": 0.9580216407775879,
      "learning_rate": 0.00019128261631505553,
      "loss": 4.2519,
      "step": 46760
    },
    {
      "epoch": 0.0435972077559125,
      "grad_norm": 1.1057257652282715,
      "learning_rate": 0.00019128075199296033,
      "loss": 3.9963,
      "step": 46770
    },
    {
      "epoch": 0.04360652937399159,
      "grad_norm": 1.002915859222412,
      "learning_rate": 0.00019127888767086514,
      "loss": 3.9778,
      "step": 46780
    },
    {
      "epoch": 0.04361585099207068,
      "grad_norm": 1.254897117614746,
      "learning_rate": 0.00019127702334876994,
      "loss": 3.8878,
      "step": 46790
    },
    {
      "epoch": 0.043625172610149775,
      "grad_norm": 1.1630892753601074,
      "learning_rate": 0.00019127515902667472,
      "loss": 4.0854,
      "step": 46800
    },
    {
      "epoch": 0.04363449422822887,
      "grad_norm": 1.1037466526031494,
      "learning_rate": 0.00019127329470457953,
      "loss": 4.1415,
      "step": 46810
    },
    {
      "epoch": 0.04364381584630796,
      "grad_norm": 1.3196417093276978,
      "learning_rate": 0.00019127143038248434,
      "loss": 3.8255,
      "step": 46820
    },
    {
      "epoch": 0.04365313746438705,
      "grad_norm": 1.1014662981033325,
      "learning_rate": 0.00019126956606038912,
      "loss": 4.1521,
      "step": 46830
    },
    {
      "epoch": 0.04366245908246614,
      "grad_norm": 1.2407433986663818,
      "learning_rate": 0.00019126770173829395,
      "loss": 4.1463,
      "step": 46840
    },
    {
      "epoch": 0.04367178070054523,
      "grad_norm": 0.8676010370254517,
      "learning_rate": 0.00019126583741619873,
      "loss": 3.9704,
      "step": 46850
    },
    {
      "epoch": 0.04368110231862433,
      "grad_norm": 1.1049174070358276,
      "learning_rate": 0.00019126397309410353,
      "loss": 4.0582,
      "step": 46860
    },
    {
      "epoch": 0.043690423936703415,
      "grad_norm": 1.164529800415039,
      "learning_rate": 0.00019126210877200834,
      "loss": 4.0582,
      "step": 46870
    },
    {
      "epoch": 0.04369974555478251,
      "grad_norm": 1.183716058731079,
      "learning_rate": 0.00019126024444991312,
      "loss": 4.0939,
      "step": 46880
    },
    {
      "epoch": 0.0437090671728616,
      "grad_norm": 1.1374597549438477,
      "learning_rate": 0.00019125838012781793,
      "loss": 3.8761,
      "step": 46890
    },
    {
      "epoch": 0.04371838879094069,
      "grad_norm": 1.1274914741516113,
      "learning_rate": 0.00019125651580572273,
      "loss": 4.2172,
      "step": 46900
    },
    {
      "epoch": 0.043727710409019786,
      "grad_norm": 1.0022622346878052,
      "learning_rate": 0.00019125465148362754,
      "loss": 4.0741,
      "step": 46910
    },
    {
      "epoch": 0.043737032027098874,
      "grad_norm": 1.2090262174606323,
      "learning_rate": 0.00019125278716153234,
      "loss": 3.9777,
      "step": 46920
    },
    {
      "epoch": 0.04374635364517797,
      "grad_norm": 1.129099726676941,
      "learning_rate": 0.00019125092283943712,
      "loss": 4.0061,
      "step": 46930
    },
    {
      "epoch": 0.04375567526325706,
      "grad_norm": 1.2259598970413208,
      "learning_rate": 0.00019124905851734193,
      "loss": 3.9269,
      "step": 46940
    },
    {
      "epoch": 0.04376499688133615,
      "grad_norm": 1.003519892692566,
      "learning_rate": 0.00019124719419524674,
      "loss": 4.135,
      "step": 46950
    },
    {
      "epoch": 0.043774318499415245,
      "grad_norm": 1.103178858757019,
      "learning_rate": 0.00019124532987315154,
      "loss": 3.971,
      "step": 46960
    },
    {
      "epoch": 0.04378364011749433,
      "grad_norm": 1.1332569122314453,
      "learning_rate": 0.00019124346555105635,
      "loss": 4.1088,
      "step": 46970
    },
    {
      "epoch": 0.04379296173557343,
      "grad_norm": 0.9410074353218079,
      "learning_rate": 0.00019124160122896113,
      "loss": 3.987,
      "step": 46980
    },
    {
      "epoch": 0.04380228335365252,
      "grad_norm": 1.033058762550354,
      "learning_rate": 0.00019123973690686593,
      "loss": 4.2054,
      "step": 46990
    },
    {
      "epoch": 0.04381160497173161,
      "grad_norm": 1.1091746091842651,
      "learning_rate": 0.00019123787258477071,
      "loss": 4.1928,
      "step": 47000
    },
    {
      "epoch": 0.0438209265898107,
      "grad_norm": 1.139663577079773,
      "learning_rate": 0.00019123600826267555,
      "loss": 3.9826,
      "step": 47010
    },
    {
      "epoch": 0.04383024820788979,
      "grad_norm": 1.1202809810638428,
      "learning_rate": 0.00019123414394058035,
      "loss": 4.0727,
      "step": 47020
    },
    {
      "epoch": 0.043839569825968885,
      "grad_norm": 1.02768874168396,
      "learning_rate": 0.00019123227961848513,
      "loss": 3.9779,
      "step": 47030
    },
    {
      "epoch": 0.04384889144404798,
      "grad_norm": 1.1712260246276855,
      "learning_rate": 0.00019123041529638994,
      "loss": 4.123,
      "step": 47040
    },
    {
      "epoch": 0.04385821306212707,
      "grad_norm": 1.10421621799469,
      "learning_rate": 0.00019122855097429472,
      "loss": 4.0493,
      "step": 47050
    },
    {
      "epoch": 0.04386753468020616,
      "grad_norm": 1.0937484502792358,
      "learning_rate": 0.00019122668665219952,
      "loss": 4.1525,
      "step": 47060
    },
    {
      "epoch": 0.043876856298285256,
      "grad_norm": 1.0834124088287354,
      "learning_rate": 0.00019122482233010436,
      "loss": 3.9761,
      "step": 47070
    },
    {
      "epoch": 0.043886177916364344,
      "grad_norm": 0.9967542886734009,
      "learning_rate": 0.00019122295800800914,
      "loss": 4.1429,
      "step": 47080
    },
    {
      "epoch": 0.04389549953444344,
      "grad_norm": 1.0524089336395264,
      "learning_rate": 0.00019122109368591394,
      "loss": 3.8268,
      "step": 47090
    },
    {
      "epoch": 0.043904821152522526,
      "grad_norm": 1.0443944931030273,
      "learning_rate": 0.00019121922936381872,
      "loss": 4.0538,
      "step": 47100
    },
    {
      "epoch": 0.04391414277060162,
      "grad_norm": 0.9442196488380432,
      "learning_rate": 0.00019121736504172353,
      "loss": 4.0842,
      "step": 47110
    },
    {
      "epoch": 0.043923464388680715,
      "grad_norm": 1.1688541173934937,
      "learning_rate": 0.00019121550071962834,
      "loss": 3.8445,
      "step": 47120
    },
    {
      "epoch": 0.0439327860067598,
      "grad_norm": 0.9767686128616333,
      "learning_rate": 0.00019121363639753314,
      "loss": 4.1613,
      "step": 47130
    },
    {
      "epoch": 0.0439421076248389,
      "grad_norm": 1.0968340635299683,
      "learning_rate": 0.00019121177207543795,
      "loss": 4.0508,
      "step": 47140
    },
    {
      "epoch": 0.04395142924291799,
      "grad_norm": 1.0210418701171875,
      "learning_rate": 0.00019120990775334273,
      "loss": 4.092,
      "step": 47150
    },
    {
      "epoch": 0.04396075086099708,
      "grad_norm": 1.049296259880066,
      "learning_rate": 0.00019120804343124753,
      "loss": 4.1528,
      "step": 47160
    },
    {
      "epoch": 0.043970072479076173,
      "grad_norm": 1.0706125497817993,
      "learning_rate": 0.00019120617910915234,
      "loss": 3.9453,
      "step": 47170
    },
    {
      "epoch": 0.04397939409715526,
      "grad_norm": 0.9552757143974304,
      "learning_rate": 0.00019120431478705715,
      "loss": 3.9573,
      "step": 47180
    },
    {
      "epoch": 0.043988715715234356,
      "grad_norm": 1.219021201133728,
      "learning_rate": 0.00019120245046496195,
      "loss": 3.9468,
      "step": 47190
    },
    {
      "epoch": 0.04399803733331345,
      "grad_norm": 1.172353982925415,
      "learning_rate": 0.00019120058614286673,
      "loss": 4.0031,
      "step": 47200
    },
    {
      "epoch": 0.04400735895139254,
      "grad_norm": 1.0842781066894531,
      "learning_rate": 0.00019119872182077154,
      "loss": 4.2652,
      "step": 47210
    },
    {
      "epoch": 0.04401668056947163,
      "grad_norm": 1.201385498046875,
      "learning_rate": 0.00019119685749867632,
      "loss": 3.9775,
      "step": 47220
    },
    {
      "epoch": 0.04402600218755072,
      "grad_norm": 1.2388924360275269,
      "learning_rate": 0.00019119499317658115,
      "loss": 4.1392,
      "step": 47230
    },
    {
      "epoch": 0.044035323805629814,
      "grad_norm": 1.1955050230026245,
      "learning_rate": 0.00019119312885448596,
      "loss": 3.9076,
      "step": 47240
    },
    {
      "epoch": 0.04404464542370891,
      "grad_norm": 1.0131714344024658,
      "learning_rate": 0.00019119126453239074,
      "loss": 4.1897,
      "step": 47250
    },
    {
      "epoch": 0.044053967041787996,
      "grad_norm": 1.139032006263733,
      "learning_rate": 0.00019118940021029554,
      "loss": 3.8627,
      "step": 47260
    },
    {
      "epoch": 0.04406328865986709,
      "grad_norm": 0.9284548759460449,
      "learning_rate": 0.00019118753588820032,
      "loss": 4.0137,
      "step": 47270
    },
    {
      "epoch": 0.044072610277946185,
      "grad_norm": 1.1322872638702393,
      "learning_rate": 0.00019118567156610513,
      "loss": 4.1275,
      "step": 47280
    },
    {
      "epoch": 0.04408193189602527,
      "grad_norm": 1.0065809488296509,
      "learning_rate": 0.00019118380724400996,
      "loss": 4.1436,
      "step": 47290
    },
    {
      "epoch": 0.04409125351410437,
      "grad_norm": 1.0474355220794678,
      "learning_rate": 0.00019118194292191474,
      "loss": 4.2584,
      "step": 47300
    },
    {
      "epoch": 0.044100575132183455,
      "grad_norm": 1.1406879425048828,
      "learning_rate": 0.00019118007859981955,
      "loss": 3.8159,
      "step": 47310
    },
    {
      "epoch": 0.04410989675026255,
      "grad_norm": 1.0692840814590454,
      "learning_rate": 0.00019117821427772433,
      "loss": 4.1091,
      "step": 47320
    },
    {
      "epoch": 0.044119218368341644,
      "grad_norm": 1.3453342914581299,
      "learning_rate": 0.00019117634995562913,
      "loss": 4.0669,
      "step": 47330
    },
    {
      "epoch": 0.04412853998642073,
      "grad_norm": 1.189660668373108,
      "learning_rate": 0.00019117448563353394,
      "loss": 4.091,
      "step": 47340
    },
    {
      "epoch": 0.044137861604499826,
      "grad_norm": 1.019249677658081,
      "learning_rate": 0.00019117262131143875,
      "loss": 4.0407,
      "step": 47350
    },
    {
      "epoch": 0.04414718322257891,
      "grad_norm": 1.1304740905761719,
      "learning_rate": 0.00019117075698934355,
      "loss": 4.0421,
      "step": 47360
    },
    {
      "epoch": 0.04415650484065801,
      "grad_norm": 0.9870085120201111,
      "learning_rate": 0.00019116889266724833,
      "loss": 3.9715,
      "step": 47370
    },
    {
      "epoch": 0.0441658264587371,
      "grad_norm": 1.154498815536499,
      "learning_rate": 0.00019116702834515314,
      "loss": 3.9293,
      "step": 47380
    },
    {
      "epoch": 0.04417514807681619,
      "grad_norm": 1.043785572052002,
      "learning_rate": 0.00019116516402305794,
      "loss": 3.9987,
      "step": 47390
    },
    {
      "epoch": 0.044184469694895284,
      "grad_norm": 1.1078177690505981,
      "learning_rate": 0.00019116329970096275,
      "loss": 4.0857,
      "step": 47400
    },
    {
      "epoch": 0.04419379131297438,
      "grad_norm": 1.2080516815185547,
      "learning_rate": 0.00019116143537886756,
      "loss": 3.9977,
      "step": 47410
    },
    {
      "epoch": 0.044203112931053466,
      "grad_norm": 1.0128014087677002,
      "learning_rate": 0.00019115957105677234,
      "loss": 3.9856,
      "step": 47420
    },
    {
      "epoch": 0.04421243454913256,
      "grad_norm": 1.2393808364868164,
      "learning_rate": 0.00019115770673467714,
      "loss": 4.017,
      "step": 47430
    },
    {
      "epoch": 0.04422175616721165,
      "grad_norm": 1.0810742378234863,
      "learning_rate": 0.00019115584241258195,
      "loss": 3.991,
      "step": 47440
    },
    {
      "epoch": 0.04423107778529074,
      "grad_norm": 1.1523890495300293,
      "learning_rate": 0.00019115397809048675,
      "loss": 4.1738,
      "step": 47450
    },
    {
      "epoch": 0.04424039940336984,
      "grad_norm": 1.06068754196167,
      "learning_rate": 0.00019115211376839156,
      "loss": 4.1377,
      "step": 47460
    },
    {
      "epoch": 0.044249721021448925,
      "grad_norm": 1.204660415649414,
      "learning_rate": 0.00019115024944629634,
      "loss": 3.9801,
      "step": 47470
    },
    {
      "epoch": 0.04425904263952802,
      "grad_norm": 0.8555912971496582,
      "learning_rate": 0.00019114838512420115,
      "loss": 4.114,
      "step": 47480
    },
    {
      "epoch": 0.044268364257607114,
      "grad_norm": 1.0715936422348022,
      "learning_rate": 0.00019114652080210595,
      "loss": 3.8691,
      "step": 47490
    },
    {
      "epoch": 0.0442776858756862,
      "grad_norm": 0.9507628679275513,
      "learning_rate": 0.00019114465648001073,
      "loss": 4.1027,
      "step": 47500
    },
    {
      "epoch": 0.044287007493765296,
      "grad_norm": 1.0550121068954468,
      "learning_rate": 0.00019114279215791556,
      "loss": 3.9274,
      "step": 47510
    },
    {
      "epoch": 0.04429632911184438,
      "grad_norm": 1.0881675481796265,
      "learning_rate": 0.00019114092783582034,
      "loss": 4.0779,
      "step": 47520
    },
    {
      "epoch": 0.04430565072992348,
      "grad_norm": 1.1459206342697144,
      "learning_rate": 0.00019113906351372515,
      "loss": 4.0347,
      "step": 47530
    },
    {
      "epoch": 0.04431497234800257,
      "grad_norm": 1.0090044736862183,
      "learning_rate": 0.00019113719919162996,
      "loss": 4.1495,
      "step": 47540
    },
    {
      "epoch": 0.04432429396608166,
      "grad_norm": 1.13468599319458,
      "learning_rate": 0.00019113533486953474,
      "loss": 4.0164,
      "step": 47550
    },
    {
      "epoch": 0.044333615584160754,
      "grad_norm": 1.1540457010269165,
      "learning_rate": 0.00019113347054743954,
      "loss": 3.9663,
      "step": 47560
    },
    {
      "epoch": 0.04434293720223984,
      "grad_norm": 1.2209383249282837,
      "learning_rate": 0.00019113160622534435,
      "loss": 4.1709,
      "step": 47570
    },
    {
      "epoch": 0.044352258820318936,
      "grad_norm": 1.2767932415008545,
      "learning_rate": 0.00019112974190324915,
      "loss": 3.896,
      "step": 47580
    },
    {
      "epoch": 0.04436158043839803,
      "grad_norm": 1.1286613941192627,
      "learning_rate": 0.00019112787758115396,
      "loss": 4.1884,
      "step": 47590
    },
    {
      "epoch": 0.04437090205647712,
      "grad_norm": 1.0568464994430542,
      "learning_rate": 0.00019112601325905874,
      "loss": 4.1203,
      "step": 47600
    },
    {
      "epoch": 0.04438022367455621,
      "grad_norm": 1.2230498790740967,
      "learning_rate": 0.00019112414893696355,
      "loss": 3.9189,
      "step": 47610
    },
    {
      "epoch": 0.04438954529263531,
      "grad_norm": 1.1333074569702148,
      "learning_rate": 0.00019112228461486835,
      "loss": 3.9677,
      "step": 47620
    },
    {
      "epoch": 0.044398866910714395,
      "grad_norm": 1.0629397630691528,
      "learning_rate": 0.00019112042029277316,
      "loss": 3.9502,
      "step": 47630
    },
    {
      "epoch": 0.04440818852879349,
      "grad_norm": 1.0865617990493774,
      "learning_rate": 0.00019111855597067797,
      "loss": 3.9651,
      "step": 47640
    },
    {
      "epoch": 0.04441751014687258,
      "grad_norm": 1.0320184230804443,
      "learning_rate": 0.00019111669164858274,
      "loss": 3.9061,
      "step": 47650
    },
    {
      "epoch": 0.04442683176495167,
      "grad_norm": 1.174523115158081,
      "learning_rate": 0.00019111482732648755,
      "loss": 4.1138,
      "step": 47660
    },
    {
      "epoch": 0.044436153383030766,
      "grad_norm": 1.306697130203247,
      "learning_rate": 0.00019111296300439233,
      "loss": 4.017,
      "step": 47670
    },
    {
      "epoch": 0.04444547500110985,
      "grad_norm": 1.2173209190368652,
      "learning_rate": 0.00019111109868229716,
      "loss": 3.8183,
      "step": 47680
    },
    {
      "epoch": 0.04445479661918895,
      "grad_norm": 1.4942885637283325,
      "learning_rate": 0.00019110923436020197,
      "loss": 4.0119,
      "step": 47690
    },
    {
      "epoch": 0.044464118237268035,
      "grad_norm": 1.1709105968475342,
      "learning_rate": 0.00019110737003810675,
      "loss": 3.9597,
      "step": 47700
    },
    {
      "epoch": 0.04447343985534713,
      "grad_norm": 1.2294687032699585,
      "learning_rate": 0.00019110550571601156,
      "loss": 4.0198,
      "step": 47710
    },
    {
      "epoch": 0.044482761473426224,
      "grad_norm": 1.2606196403503418,
      "learning_rate": 0.00019110364139391633,
      "loss": 4.0402,
      "step": 47720
    },
    {
      "epoch": 0.04449208309150531,
      "grad_norm": 0.9657955765724182,
      "learning_rate": 0.00019110177707182114,
      "loss": 4.1043,
      "step": 47730
    },
    {
      "epoch": 0.044501404709584406,
      "grad_norm": 1.0346981287002563,
      "learning_rate": 0.00019109991274972597,
      "loss": 4.0472,
      "step": 47740
    },
    {
      "epoch": 0.0445107263276635,
      "grad_norm": 1.3508965969085693,
      "learning_rate": 0.00019109804842763075,
      "loss": 3.9499,
      "step": 47750
    },
    {
      "epoch": 0.04452004794574259,
      "grad_norm": 0.9837897419929504,
      "learning_rate": 0.00019109618410553556,
      "loss": 4.0213,
      "step": 47760
    },
    {
      "epoch": 0.04452936956382168,
      "grad_norm": 1.1742255687713623,
      "learning_rate": 0.00019109431978344034,
      "loss": 3.9936,
      "step": 47770
    },
    {
      "epoch": 0.04453869118190077,
      "grad_norm": 1.0953235626220703,
      "learning_rate": 0.00019109245546134515,
      "loss": 4.0511,
      "step": 47780
    },
    {
      "epoch": 0.044548012799979865,
      "grad_norm": 0.9568156003952026,
      "learning_rate": 0.00019109059113924995,
      "loss": 4.2505,
      "step": 47790
    },
    {
      "epoch": 0.04455733441805896,
      "grad_norm": 1.4872177839279175,
      "learning_rate": 0.00019108872681715476,
      "loss": 4.0775,
      "step": 47800
    },
    {
      "epoch": 0.04456665603613805,
      "grad_norm": 1.3332055807113647,
      "learning_rate": 0.00019108686249505956,
      "loss": 4.1315,
      "step": 47810
    },
    {
      "epoch": 0.04457597765421714,
      "grad_norm": 0.9385437965393066,
      "learning_rate": 0.00019108499817296434,
      "loss": 4.0049,
      "step": 47820
    },
    {
      "epoch": 0.044585299272296236,
      "grad_norm": 0.9329249858856201,
      "learning_rate": 0.00019108313385086915,
      "loss": 4.0616,
      "step": 47830
    },
    {
      "epoch": 0.044594620890375324,
      "grad_norm": 1.1292247772216797,
      "learning_rate": 0.00019108126952877396,
      "loss": 3.8317,
      "step": 47840
    },
    {
      "epoch": 0.04460394250845442,
      "grad_norm": 0.8658884763717651,
      "learning_rate": 0.00019107940520667876,
      "loss": 4.0064,
      "step": 47850
    },
    {
      "epoch": 0.044613264126533506,
      "grad_norm": 0.9039189219474792,
      "learning_rate": 0.00019107754088458357,
      "loss": 3.9462,
      "step": 47860
    },
    {
      "epoch": 0.0446225857446126,
      "grad_norm": 0.9769604206085205,
      "learning_rate": 0.00019107567656248835,
      "loss": 4.0538,
      "step": 47870
    },
    {
      "epoch": 0.044631907362691695,
      "grad_norm": 1.0351992845535278,
      "learning_rate": 0.00019107381224039315,
      "loss": 4.1642,
      "step": 47880
    },
    {
      "epoch": 0.04464122898077078,
      "grad_norm": 1.2625889778137207,
      "learning_rate": 0.00019107194791829796,
      "loss": 3.9075,
      "step": 47890
    },
    {
      "epoch": 0.04465055059884988,
      "grad_norm": 1.0803725719451904,
      "learning_rate": 0.00019107008359620277,
      "loss": 4.07,
      "step": 47900
    },
    {
      "epoch": 0.044659872216928964,
      "grad_norm": 1.23104727268219,
      "learning_rate": 0.00019106821927410757,
      "loss": 4.0527,
      "step": 47910
    },
    {
      "epoch": 0.04466919383500806,
      "grad_norm": 0.918766975402832,
      "learning_rate": 0.00019106635495201235,
      "loss": 4.1813,
      "step": 47920
    },
    {
      "epoch": 0.04467851545308715,
      "grad_norm": 1.1882961988449097,
      "learning_rate": 0.00019106449062991716,
      "loss": 4.1317,
      "step": 47930
    },
    {
      "epoch": 0.04468783707116624,
      "grad_norm": 1.1754664182662964,
      "learning_rate": 0.00019106262630782196,
      "loss": 4.0277,
      "step": 47940
    },
    {
      "epoch": 0.044697158689245335,
      "grad_norm": 1.011410117149353,
      "learning_rate": 0.00019106076198572674,
      "loss": 4.108,
      "step": 47950
    },
    {
      "epoch": 0.04470648030732443,
      "grad_norm": 1.0004020929336548,
      "learning_rate": 0.00019105889766363158,
      "loss": 4.0798,
      "step": 47960
    },
    {
      "epoch": 0.04471580192540352,
      "grad_norm": 1.021958589553833,
      "learning_rate": 0.00019105703334153636,
      "loss": 4.1374,
      "step": 47970
    },
    {
      "epoch": 0.04472512354348261,
      "grad_norm": 1.147586703300476,
      "learning_rate": 0.00019105516901944116,
      "loss": 3.9479,
      "step": 47980
    },
    {
      "epoch": 0.0447344451615617,
      "grad_norm": 1.0703593492507935,
      "learning_rate": 0.00019105330469734597,
      "loss": 4.0865,
      "step": 47990
    },
    {
      "epoch": 0.044743766779640794,
      "grad_norm": 1.44101083278656,
      "learning_rate": 0.00019105144037525075,
      "loss": 4.2429,
      "step": 48000
    },
    {
      "epoch": 0.04475308839771989,
      "grad_norm": 1.2089368104934692,
      "learning_rate": 0.00019104957605315555,
      "loss": 4.1619,
      "step": 48010
    },
    {
      "epoch": 0.044762410015798976,
      "grad_norm": 1.1085119247436523,
      "learning_rate": 0.00019104771173106036,
      "loss": 3.9008,
      "step": 48020
    },
    {
      "epoch": 0.04477173163387807,
      "grad_norm": 1.0245685577392578,
      "learning_rate": 0.00019104584740896517,
      "loss": 3.9662,
      "step": 48030
    },
    {
      "epoch": 0.04478105325195716,
      "grad_norm": 1.240609884262085,
      "learning_rate": 0.00019104398308686997,
      "loss": 4.0658,
      "step": 48040
    },
    {
      "epoch": 0.04479037487003625,
      "grad_norm": 1.0533725023269653,
      "learning_rate": 0.00019104211876477475,
      "loss": 4.0935,
      "step": 48050
    },
    {
      "epoch": 0.04479969648811535,
      "grad_norm": 1.075305700302124,
      "learning_rate": 0.00019104025444267956,
      "loss": 3.8449,
      "step": 48060
    },
    {
      "epoch": 0.044809018106194434,
      "grad_norm": 1.0421525239944458,
      "learning_rate": 0.00019103839012058437,
      "loss": 4.3222,
      "step": 48070
    },
    {
      "epoch": 0.04481833972427353,
      "grad_norm": 1.017531156539917,
      "learning_rate": 0.00019103652579848917,
      "loss": 3.9394,
      "step": 48080
    },
    {
      "epoch": 0.04482766134235262,
      "grad_norm": 0.9623063802719116,
      "learning_rate": 0.00019103466147639398,
      "loss": 4.2316,
      "step": 48090
    },
    {
      "epoch": 0.04483698296043171,
      "grad_norm": 1.2312684059143066,
      "learning_rate": 0.00019103279715429876,
      "loss": 4.1021,
      "step": 48100
    },
    {
      "epoch": 0.044846304578510805,
      "grad_norm": 1.1282175779342651,
      "learning_rate": 0.00019103093283220356,
      "loss": 4.0671,
      "step": 48110
    },
    {
      "epoch": 0.04485562619658989,
      "grad_norm": 1.0466934442520142,
      "learning_rate": 0.00019102906851010834,
      "loss": 3.8091,
      "step": 48120
    },
    {
      "epoch": 0.04486494781466899,
      "grad_norm": 0.9588691592216492,
      "learning_rate": 0.00019102720418801318,
      "loss": 3.9928,
      "step": 48130
    },
    {
      "epoch": 0.04487426943274808,
      "grad_norm": 1.019524335861206,
      "learning_rate": 0.00019102533986591798,
      "loss": 3.9535,
      "step": 48140
    },
    {
      "epoch": 0.04488359105082717,
      "grad_norm": 0.8539379239082336,
      "learning_rate": 0.00019102347554382276,
      "loss": 3.9985,
      "step": 48150
    },
    {
      "epoch": 0.044892912668906264,
      "grad_norm": 1.0060495138168335,
      "learning_rate": 0.00019102161122172757,
      "loss": 3.9163,
      "step": 48160
    },
    {
      "epoch": 0.04490223428698536,
      "grad_norm": 1.3036165237426758,
      "learning_rate": 0.00019101974689963235,
      "loss": 4.2083,
      "step": 48170
    },
    {
      "epoch": 0.044911555905064446,
      "grad_norm": 1.0218970775604248,
      "learning_rate": 0.00019101788257753715,
      "loss": 4.1404,
      "step": 48180
    },
    {
      "epoch": 0.04492087752314354,
      "grad_norm": 1.0990060567855835,
      "learning_rate": 0.000191016018255442,
      "loss": 4.1334,
      "step": 48190
    },
    {
      "epoch": 0.04493019914122263,
      "grad_norm": 1.2279356718063354,
      "learning_rate": 0.00019101415393334677,
      "loss": 4.0615,
      "step": 48200
    },
    {
      "epoch": 0.04493952075930172,
      "grad_norm": 0.9293816089630127,
      "learning_rate": 0.00019101228961125157,
      "loss": 3.9063,
      "step": 48210
    },
    {
      "epoch": 0.04494884237738082,
      "grad_norm": 1.0597482919692993,
      "learning_rate": 0.00019101042528915635,
      "loss": 4.0802,
      "step": 48220
    },
    {
      "epoch": 0.044958163995459904,
      "grad_norm": 1.1217434406280518,
      "learning_rate": 0.00019100856096706116,
      "loss": 3.9823,
      "step": 48230
    },
    {
      "epoch": 0.044967485613539,
      "grad_norm": 1.297641396522522,
      "learning_rate": 0.00019100669664496596,
      "loss": 4.1076,
      "step": 48240
    },
    {
      "epoch": 0.044976807231618086,
      "grad_norm": 1.1601338386535645,
      "learning_rate": 0.00019100483232287077,
      "loss": 4.1396,
      "step": 48250
    },
    {
      "epoch": 0.04498612884969718,
      "grad_norm": 1.0685162544250488,
      "learning_rate": 0.00019100296800077558,
      "loss": 4.155,
      "step": 48260
    },
    {
      "epoch": 0.044995450467776275,
      "grad_norm": 1.2048513889312744,
      "learning_rate": 0.00019100110367868036,
      "loss": 4.1958,
      "step": 48270
    },
    {
      "epoch": 0.04500477208585536,
      "grad_norm": 1.080828309059143,
      "learning_rate": 0.00019099923935658516,
      "loss": 4.042,
      "step": 48280
    },
    {
      "epoch": 0.04501409370393446,
      "grad_norm": 1.0769445896148682,
      "learning_rate": 0.00019099737503448997,
      "loss": 4.1956,
      "step": 48290
    },
    {
      "epoch": 0.04502341532201355,
      "grad_norm": 0.9664427042007446,
      "learning_rate": 0.00019099551071239478,
      "loss": 4.1039,
      "step": 48300
    },
    {
      "epoch": 0.04503273694009264,
      "grad_norm": 1.0444129705429077,
      "learning_rate": 0.00019099364639029958,
      "loss": 4.1236,
      "step": 48310
    },
    {
      "epoch": 0.045042058558171734,
      "grad_norm": 0.9810559153556824,
      "learning_rate": 0.00019099178206820436,
      "loss": 3.9259,
      "step": 48320
    },
    {
      "epoch": 0.04505138017625082,
      "grad_norm": 1.372706651687622,
      "learning_rate": 0.00019098991774610917,
      "loss": 4.1928,
      "step": 48330
    },
    {
      "epoch": 0.045060701794329916,
      "grad_norm": 1.2022677659988403,
      "learning_rate": 0.00019098805342401397,
      "loss": 3.9343,
      "step": 48340
    },
    {
      "epoch": 0.04507002341240901,
      "grad_norm": 2.408967971801758,
      "learning_rate": 0.00019098618910191878,
      "loss": 4.2057,
      "step": 48350
    },
    {
      "epoch": 0.0450793450304881,
      "grad_norm": 1.1608177423477173,
      "learning_rate": 0.00019098432477982359,
      "loss": 3.9087,
      "step": 48360
    },
    {
      "epoch": 0.04508866664856719,
      "grad_norm": 0.9680711627006531,
      "learning_rate": 0.00019098246045772837,
      "loss": 3.9989,
      "step": 48370
    },
    {
      "epoch": 0.04509798826664628,
      "grad_norm": 1.0590364933013916,
      "learning_rate": 0.00019098059613563317,
      "loss": 3.9868,
      "step": 48380
    },
    {
      "epoch": 0.045107309884725375,
      "grad_norm": 1.045609712600708,
      "learning_rate": 0.00019097873181353798,
      "loss": 4.1131,
      "step": 48390
    },
    {
      "epoch": 0.04511663150280447,
      "grad_norm": 1.2685140371322632,
      "learning_rate": 0.00019097686749144276,
      "loss": 3.9038,
      "step": 48400
    },
    {
      "epoch": 0.04512595312088356,
      "grad_norm": 1.2938525676727295,
      "learning_rate": 0.0001909750031693476,
      "loss": 4.1621,
      "step": 48410
    },
    {
      "epoch": 0.04513527473896265,
      "grad_norm": 0.9220318794250488,
      "learning_rate": 0.00019097313884725237,
      "loss": 4.0831,
      "step": 48420
    },
    {
      "epoch": 0.045144596357041746,
      "grad_norm": 1.1798179149627686,
      "learning_rate": 0.00019097127452515718,
      "loss": 3.9606,
      "step": 48430
    },
    {
      "epoch": 0.04515391797512083,
      "grad_norm": 1.0793646574020386,
      "learning_rate": 0.00019096941020306198,
      "loss": 3.9854,
      "step": 48440
    },
    {
      "epoch": 0.04516323959319993,
      "grad_norm": 1.2824171781539917,
      "learning_rate": 0.00019096754588096676,
      "loss": 3.9019,
      "step": 48450
    },
    {
      "epoch": 0.045172561211279015,
      "grad_norm": 0.9277963042259216,
      "learning_rate": 0.00019096568155887157,
      "loss": 3.8983,
      "step": 48460
    },
    {
      "epoch": 0.04518188282935811,
      "grad_norm": 1.1857331991195679,
      "learning_rate": 0.00019096381723677637,
      "loss": 4.0289,
      "step": 48470
    },
    {
      "epoch": 0.045191204447437204,
      "grad_norm": 1.2582961320877075,
      "learning_rate": 0.00019096195291468118,
      "loss": 4.0249,
      "step": 48480
    },
    {
      "epoch": 0.04520052606551629,
      "grad_norm": 1.1935498714447021,
      "learning_rate": 0.000190960088592586,
      "loss": 3.9369,
      "step": 48490
    },
    {
      "epoch": 0.045209847683595386,
      "grad_norm": 1.10221266746521,
      "learning_rate": 0.00019095822427049077,
      "loss": 4.1008,
      "step": 48500
    },
    {
      "epoch": 0.04521916930167448,
      "grad_norm": 1.0650482177734375,
      "learning_rate": 0.00019095635994839557,
      "loss": 3.9763,
      "step": 48510
    },
    {
      "epoch": 0.04522849091975357,
      "grad_norm": 1.0645064115524292,
      "learning_rate": 0.00019095449562630038,
      "loss": 3.9477,
      "step": 48520
    },
    {
      "epoch": 0.04523781253783266,
      "grad_norm": 1.1214148998260498,
      "learning_rate": 0.00019095263130420518,
      "loss": 4.0057,
      "step": 48530
    },
    {
      "epoch": 0.04524713415591175,
      "grad_norm": 1.1596720218658447,
      "learning_rate": 0.00019095076698211,
      "loss": 4.0388,
      "step": 48540
    },
    {
      "epoch": 0.045256455773990845,
      "grad_norm": 1.053486704826355,
      "learning_rate": 0.00019094890266001477,
      "loss": 3.6187,
      "step": 48550
    },
    {
      "epoch": 0.04526577739206994,
      "grad_norm": 1.0372201204299927,
      "learning_rate": 0.00019094703833791958,
      "loss": 4.0699,
      "step": 48560
    },
    {
      "epoch": 0.04527509901014903,
      "grad_norm": 1.133436918258667,
      "learning_rate": 0.00019094517401582436,
      "loss": 3.9747,
      "step": 48570
    },
    {
      "epoch": 0.04528442062822812,
      "grad_norm": 1.1140519380569458,
      "learning_rate": 0.0001909433096937292,
      "loss": 4.0382,
      "step": 48580
    },
    {
      "epoch": 0.04529374224630721,
      "grad_norm": 1.0306501388549805,
      "learning_rate": 0.00019094144537163397,
      "loss": 4.091,
      "step": 48590
    },
    {
      "epoch": 0.0453030638643863,
      "grad_norm": 1.1124743223190308,
      "learning_rate": 0.00019093958104953877,
      "loss": 3.9783,
      "step": 48600
    },
    {
      "epoch": 0.0453123854824654,
      "grad_norm": 1.022141933441162,
      "learning_rate": 0.00019093771672744358,
      "loss": 4.0052,
      "step": 48610
    },
    {
      "epoch": 0.045321707100544485,
      "grad_norm": 0.8295461535453796,
      "learning_rate": 0.00019093585240534836,
      "loss": 3.9221,
      "step": 48620
    },
    {
      "epoch": 0.04533102871862358,
      "grad_norm": 0.9933656454086304,
      "learning_rate": 0.00019093398808325317,
      "loss": 4.0937,
      "step": 48630
    },
    {
      "epoch": 0.045340350336702674,
      "grad_norm": 1.0240967273712158,
      "learning_rate": 0.00019093212376115797,
      "loss": 4.1703,
      "step": 48640
    },
    {
      "epoch": 0.04534967195478176,
      "grad_norm": 1.082154631614685,
      "learning_rate": 0.00019093025943906278,
      "loss": 4.21,
      "step": 48650
    },
    {
      "epoch": 0.045358993572860856,
      "grad_norm": 1.0178030729293823,
      "learning_rate": 0.00019092839511696759,
      "loss": 3.938,
      "step": 48660
    },
    {
      "epoch": 0.045368315190939944,
      "grad_norm": 1.0003572702407837,
      "learning_rate": 0.00019092653079487236,
      "loss": 4.1695,
      "step": 48670
    },
    {
      "epoch": 0.04537763680901904,
      "grad_norm": 1.1561449766159058,
      "learning_rate": 0.00019092466647277717,
      "loss": 4.1007,
      "step": 48680
    },
    {
      "epoch": 0.04538695842709813,
      "grad_norm": 1.0155256986618042,
      "learning_rate": 0.00019092280215068198,
      "loss": 3.7897,
      "step": 48690
    },
    {
      "epoch": 0.04539628004517722,
      "grad_norm": 1.1176549196243286,
      "learning_rate": 0.00019092093782858678,
      "loss": 4.1584,
      "step": 48700
    },
    {
      "epoch": 0.045405601663256315,
      "grad_norm": 1.2300630807876587,
      "learning_rate": 0.0001909190735064916,
      "loss": 4.2184,
      "step": 48710
    },
    {
      "epoch": 0.0454149232813354,
      "grad_norm": 0.8977557420730591,
      "learning_rate": 0.00019091720918439637,
      "loss": 4.131,
      "step": 48720
    },
    {
      "epoch": 0.0454242448994145,
      "grad_norm": 0.9371364712715149,
      "learning_rate": 0.00019091534486230118,
      "loss": 3.9319,
      "step": 48730
    },
    {
      "epoch": 0.04543356651749359,
      "grad_norm": 1.3954905271530151,
      "learning_rate": 0.00019091348054020598,
      "loss": 4.0016,
      "step": 48740
    },
    {
      "epoch": 0.04544288813557268,
      "grad_norm": 1.3843994140625,
      "learning_rate": 0.0001909116162181108,
      "loss": 4.066,
      "step": 48750
    },
    {
      "epoch": 0.04545220975365177,
      "grad_norm": 1.1504366397857666,
      "learning_rate": 0.0001909097518960156,
      "loss": 3.9976,
      "step": 48760
    },
    {
      "epoch": 0.04546153137173087,
      "grad_norm": 1.0226024389266968,
      "learning_rate": 0.00019090788757392037,
      "loss": 4.0825,
      "step": 48770
    },
    {
      "epoch": 0.045470852989809955,
      "grad_norm": 1.0539900064468384,
      "learning_rate": 0.00019090602325182518,
      "loss": 4.1474,
      "step": 48780
    },
    {
      "epoch": 0.04548017460788905,
      "grad_norm": 0.9256855249404907,
      "learning_rate": 0.00019090415892972996,
      "loss": 4.0418,
      "step": 48790
    },
    {
      "epoch": 0.04548949622596814,
      "grad_norm": 1.0217599868774414,
      "learning_rate": 0.0001909022946076348,
      "loss": 3.9543,
      "step": 48800
    },
    {
      "epoch": 0.04549881784404723,
      "grad_norm": 1.0485483407974243,
      "learning_rate": 0.0001909004302855396,
      "loss": 4.112,
      "step": 48810
    },
    {
      "epoch": 0.045508139462126326,
      "grad_norm": 0.9970660209655762,
      "learning_rate": 0.00019089856596344438,
      "loss": 3.9388,
      "step": 48820
    },
    {
      "epoch": 0.045517461080205414,
      "grad_norm": 1.0558634996414185,
      "learning_rate": 0.00019089670164134918,
      "loss": 4.0495,
      "step": 48830
    },
    {
      "epoch": 0.04552678269828451,
      "grad_norm": 1.1164113283157349,
      "learning_rate": 0.00019089483731925396,
      "loss": 3.9239,
      "step": 48840
    },
    {
      "epoch": 0.0455361043163636,
      "grad_norm": 0.8995474576950073,
      "learning_rate": 0.00019089297299715877,
      "loss": 4.1141,
      "step": 48850
    },
    {
      "epoch": 0.04554542593444269,
      "grad_norm": 1.1857380867004395,
      "learning_rate": 0.0001908911086750636,
      "loss": 3.993,
      "step": 48860
    },
    {
      "epoch": 0.045554747552521785,
      "grad_norm": 1.225184679031372,
      "learning_rate": 0.00019088924435296838,
      "loss": 3.8373,
      "step": 48870
    },
    {
      "epoch": 0.04556406917060087,
      "grad_norm": 1.0722023248672485,
      "learning_rate": 0.0001908873800308732,
      "loss": 3.9678,
      "step": 48880
    },
    {
      "epoch": 0.04557339078867997,
      "grad_norm": 1.208120346069336,
      "learning_rate": 0.00019088551570877797,
      "loss": 3.9712,
      "step": 48890
    },
    {
      "epoch": 0.04558271240675906,
      "grad_norm": 1.1874688863754272,
      "learning_rate": 0.00019088365138668277,
      "loss": 3.9369,
      "step": 48900
    },
    {
      "epoch": 0.04559203402483815,
      "grad_norm": 1.0628252029418945,
      "learning_rate": 0.00019088178706458758,
      "loss": 4.0351,
      "step": 48910
    },
    {
      "epoch": 0.04560135564291724,
      "grad_norm": 1.0105645656585693,
      "learning_rate": 0.0001908799227424924,
      "loss": 4.1531,
      "step": 48920
    },
    {
      "epoch": 0.04561067726099633,
      "grad_norm": 1.3097681999206543,
      "learning_rate": 0.0001908780584203972,
      "loss": 3.9027,
      "step": 48930
    },
    {
      "epoch": 0.045619998879075425,
      "grad_norm": 1.0567169189453125,
      "learning_rate": 0.00019087619409830197,
      "loss": 3.8607,
      "step": 48940
    },
    {
      "epoch": 0.04562932049715452,
      "grad_norm": 0.9905813932418823,
      "learning_rate": 0.00019087432977620678,
      "loss": 4.1508,
      "step": 48950
    },
    {
      "epoch": 0.04563864211523361,
      "grad_norm": 1.0056835412979126,
      "learning_rate": 0.00019087246545411158,
      "loss": 4.1652,
      "step": 48960
    },
    {
      "epoch": 0.0456479637333127,
      "grad_norm": 1.1679283380508423,
      "learning_rate": 0.0001908706011320164,
      "loss": 4.0216,
      "step": 48970
    },
    {
      "epoch": 0.045657285351391796,
      "grad_norm": 0.963749349117279,
      "learning_rate": 0.0001908687368099212,
      "loss": 4.1927,
      "step": 48980
    },
    {
      "epoch": 0.045666606969470884,
      "grad_norm": 1.0704704523086548,
      "learning_rate": 0.00019086687248782598,
      "loss": 3.797,
      "step": 48990
    },
    {
      "epoch": 0.04567592858754998,
      "grad_norm": 1.1272691488265991,
      "learning_rate": 0.00019086500816573078,
      "loss": 4.0543,
      "step": 49000
    },
    {
      "epoch": 0.045685250205629066,
      "grad_norm": 0.9592757821083069,
      "learning_rate": 0.0001908631438436356,
      "loss": 3.8296,
      "step": 49010
    },
    {
      "epoch": 0.04569457182370816,
      "grad_norm": 1.1458505392074585,
      "learning_rate": 0.00019086127952154037,
      "loss": 3.8689,
      "step": 49020
    },
    {
      "epoch": 0.045703893441787255,
      "grad_norm": 1.1150951385498047,
      "learning_rate": 0.0001908594151994452,
      "loss": 3.875,
      "step": 49030
    },
    {
      "epoch": 0.04571321505986634,
      "grad_norm": 0.9757285118103027,
      "learning_rate": 0.00019085755087734998,
      "loss": 4.215,
      "step": 49040
    },
    {
      "epoch": 0.04572253667794544,
      "grad_norm": 1.200183629989624,
      "learning_rate": 0.0001908556865552548,
      "loss": 4.0598,
      "step": 49050
    },
    {
      "epoch": 0.045731858296024525,
      "grad_norm": 1.2313543558120728,
      "learning_rate": 0.0001908538222331596,
      "loss": 3.8395,
      "step": 49060
    },
    {
      "epoch": 0.04574117991410362,
      "grad_norm": 1.145370364189148,
      "learning_rate": 0.00019085195791106437,
      "loss": 4.0539,
      "step": 49070
    },
    {
      "epoch": 0.045750501532182714,
      "grad_norm": 1.1394994258880615,
      "learning_rate": 0.00019085009358896918,
      "loss": 4.2232,
      "step": 49080
    },
    {
      "epoch": 0.0457598231502618,
      "grad_norm": 1.2331503629684448,
      "learning_rate": 0.00019084822926687399,
      "loss": 3.9876,
      "step": 49090
    },
    {
      "epoch": 0.045769144768340896,
      "grad_norm": 0.9716002345085144,
      "learning_rate": 0.0001908463649447788,
      "loss": 3.9544,
      "step": 49100
    },
    {
      "epoch": 0.04577846638641999,
      "grad_norm": 1.1548243761062622,
      "learning_rate": 0.0001908445006226836,
      "loss": 4.1755,
      "step": 49110
    },
    {
      "epoch": 0.04578778800449908,
      "grad_norm": 1.1178689002990723,
      "learning_rate": 0.00019084263630058838,
      "loss": 4.0209,
      "step": 49120
    },
    {
      "epoch": 0.04579710962257817,
      "grad_norm": 1.0301586389541626,
      "learning_rate": 0.00019084077197849318,
      "loss": 4.0841,
      "step": 49130
    },
    {
      "epoch": 0.04580643124065726,
      "grad_norm": 1.1156069040298462,
      "learning_rate": 0.000190838907656398,
      "loss": 4.043,
      "step": 49140
    },
    {
      "epoch": 0.045815752858736354,
      "grad_norm": 1.1244570016860962,
      "learning_rate": 0.0001908370433343028,
      "loss": 3.9201,
      "step": 49150
    },
    {
      "epoch": 0.04582507447681545,
      "grad_norm": 0.9728100895881653,
      "learning_rate": 0.0001908351790122076,
      "loss": 4.0378,
      "step": 49160
    },
    {
      "epoch": 0.045834396094894536,
      "grad_norm": 1.2266945838928223,
      "learning_rate": 0.00019083331469011238,
      "loss": 4.0341,
      "step": 49170
    },
    {
      "epoch": 0.04584371771297363,
      "grad_norm": 1.130985975265503,
      "learning_rate": 0.0001908314503680172,
      "loss": 4.1784,
      "step": 49180
    },
    {
      "epoch": 0.045853039331052725,
      "grad_norm": 0.9887365698814392,
      "learning_rate": 0.000190829586045922,
      "loss": 4.0068,
      "step": 49190
    },
    {
      "epoch": 0.04586236094913181,
      "grad_norm": 1.0146374702453613,
      "learning_rate": 0.0001908277217238268,
      "loss": 3.9226,
      "step": 49200
    },
    {
      "epoch": 0.04587168256721091,
      "grad_norm": 1.1310629844665527,
      "learning_rate": 0.0001908258574017316,
      "loss": 4.1792,
      "step": 49210
    },
    {
      "epoch": 0.045881004185289995,
      "grad_norm": 0.9981986284255981,
      "learning_rate": 0.00019082399307963639,
      "loss": 3.9153,
      "step": 49220
    },
    {
      "epoch": 0.04589032580336909,
      "grad_norm": 1.2802014350891113,
      "learning_rate": 0.0001908221287575412,
      "loss": 4.0365,
      "step": 49230
    },
    {
      "epoch": 0.045899647421448184,
      "grad_norm": 1.0922296047210693,
      "learning_rate": 0.00019082026443544597,
      "loss": 4.0892,
      "step": 49240
    },
    {
      "epoch": 0.04590896903952727,
      "grad_norm": 1.2148096561431885,
      "learning_rate": 0.0001908184001133508,
      "loss": 4.0775,
      "step": 49250
    },
    {
      "epoch": 0.045918290657606366,
      "grad_norm": 1.194090723991394,
      "learning_rate": 0.0001908165357912556,
      "loss": 3.99,
      "step": 49260
    },
    {
      "epoch": 0.04592761227568545,
      "grad_norm": 1.1852593421936035,
      "learning_rate": 0.0001908146714691604,
      "loss": 4.0048,
      "step": 49270
    },
    {
      "epoch": 0.04593693389376455,
      "grad_norm": 0.9613156914710999,
      "learning_rate": 0.0001908128071470652,
      "loss": 3.8674,
      "step": 49280
    },
    {
      "epoch": 0.04594625551184364,
      "grad_norm": 1.182906150817871,
      "learning_rate": 0.00019081094282496998,
      "loss": 3.9941,
      "step": 49290
    },
    {
      "epoch": 0.04595557712992273,
      "grad_norm": 1.0688416957855225,
      "learning_rate": 0.00019080907850287478,
      "loss": 3.9773,
      "step": 49300
    },
    {
      "epoch": 0.045964898748001824,
      "grad_norm": 1.1150480508804321,
      "learning_rate": 0.00019080721418077962,
      "loss": 4.0555,
      "step": 49310
    },
    {
      "epoch": 0.04597422036608092,
      "grad_norm": 0.8531553149223328,
      "learning_rate": 0.0001908053498586844,
      "loss": 4.0883,
      "step": 49320
    },
    {
      "epoch": 0.045983541984160006,
      "grad_norm": 0.980991005897522,
      "learning_rate": 0.0001908034855365892,
      "loss": 4.0083,
      "step": 49330
    },
    {
      "epoch": 0.0459928636022391,
      "grad_norm": 1.0465149879455566,
      "learning_rate": 0.00019080162121449398,
      "loss": 3.9067,
      "step": 49340
    },
    {
      "epoch": 0.04600218522031819,
      "grad_norm": 0.9833612442016602,
      "learning_rate": 0.0001907997568923988,
      "loss": 3.9139,
      "step": 49350
    },
    {
      "epoch": 0.04601150683839728,
      "grad_norm": 1.176278829574585,
      "learning_rate": 0.0001907978925703036,
      "loss": 4.0285,
      "step": 49360
    },
    {
      "epoch": 0.04602082845647638,
      "grad_norm": 1.091480016708374,
      "learning_rate": 0.0001907960282482084,
      "loss": 4.157,
      "step": 49370
    },
    {
      "epoch": 0.046030150074555465,
      "grad_norm": 1.0111989974975586,
      "learning_rate": 0.0001907941639261132,
      "loss": 3.8075,
      "step": 49380
    },
    {
      "epoch": 0.04603947169263456,
      "grad_norm": 1.2476634979248047,
      "learning_rate": 0.00019079229960401799,
      "loss": 3.9717,
      "step": 49390
    },
    {
      "epoch": 0.04604879331071365,
      "grad_norm": 1.0389792919158936,
      "learning_rate": 0.0001907904352819228,
      "loss": 4.1109,
      "step": 49400
    },
    {
      "epoch": 0.04605811492879274,
      "grad_norm": 1.2975809574127197,
      "learning_rate": 0.0001907885709598276,
      "loss": 4.0788,
      "step": 49410
    },
    {
      "epoch": 0.046067436546871836,
      "grad_norm": 1.144300103187561,
      "learning_rate": 0.0001907867066377324,
      "loss": 4.102,
      "step": 49420
    },
    {
      "epoch": 0.04607675816495092,
      "grad_norm": 1.01630437374115,
      "learning_rate": 0.0001907848423156372,
      "loss": 3.9335,
      "step": 49430
    },
    {
      "epoch": 0.04608607978303002,
      "grad_norm": 1.2317413091659546,
      "learning_rate": 0.000190782977993542,
      "loss": 4.0256,
      "step": 49440
    },
    {
      "epoch": 0.04609540140110911,
      "grad_norm": 1.1761225461959839,
      "learning_rate": 0.0001907811136714468,
      "loss": 3.905,
      "step": 49450
    },
    {
      "epoch": 0.0461047230191882,
      "grad_norm": 1.110052466392517,
      "learning_rate": 0.0001907792493493516,
      "loss": 4.049,
      "step": 49460
    },
    {
      "epoch": 0.046114044637267294,
      "grad_norm": 0.9977116584777832,
      "learning_rate": 0.00019077738502725638,
      "loss": 4.1524,
      "step": 49470
    },
    {
      "epoch": 0.04612336625534638,
      "grad_norm": 1.2224819660186768,
      "learning_rate": 0.00019077552070516121,
      "loss": 4.0046,
      "step": 49480
    },
    {
      "epoch": 0.046132687873425476,
      "grad_norm": 1.1212358474731445,
      "learning_rate": 0.000190773656383066,
      "loss": 4.2148,
      "step": 49490
    },
    {
      "epoch": 0.04614200949150457,
      "grad_norm": 1.2969779968261719,
      "learning_rate": 0.0001907717920609708,
      "loss": 3.9825,
      "step": 49500
    },
    {
      "epoch": 0.04615133110958366,
      "grad_norm": 1.1154277324676514,
      "learning_rate": 0.0001907699277388756,
      "loss": 4.1183,
      "step": 49510
    },
    {
      "epoch": 0.04616065272766275,
      "grad_norm": 1.1089688539505005,
      "learning_rate": 0.00019076806341678039,
      "loss": 3.7625,
      "step": 49520
    },
    {
      "epoch": 0.04616997434574185,
      "grad_norm": 0.9644889235496521,
      "learning_rate": 0.0001907661990946852,
      "loss": 3.9657,
      "step": 49530
    },
    {
      "epoch": 0.046179295963820935,
      "grad_norm": 1.05343496799469,
      "learning_rate": 0.00019076433477259,
      "loss": 3.9576,
      "step": 49540
    },
    {
      "epoch": 0.04618861758190003,
      "grad_norm": 1.14668869972229,
      "learning_rate": 0.0001907624704504948,
      "loss": 3.8003,
      "step": 49550
    },
    {
      "epoch": 0.04619793919997912,
      "grad_norm": 0.9743043780326843,
      "learning_rate": 0.0001907606061283996,
      "loss": 3.9793,
      "step": 49560
    },
    {
      "epoch": 0.04620726081805821,
      "grad_norm": 1.0457547903060913,
      "learning_rate": 0.0001907587418063044,
      "loss": 4.2142,
      "step": 49570
    },
    {
      "epoch": 0.046216582436137306,
      "grad_norm": 1.1340174674987793,
      "learning_rate": 0.0001907568774842092,
      "loss": 4.1425,
      "step": 49580
    },
    {
      "epoch": 0.046225904054216393,
      "grad_norm": 1.066458821296692,
      "learning_rate": 0.000190755013162114,
      "loss": 4.1857,
      "step": 49590
    },
    {
      "epoch": 0.04623522567229549,
      "grad_norm": 1.1168851852416992,
      "learning_rate": 0.0001907531488400188,
      "loss": 4.1749,
      "step": 49600
    },
    {
      "epoch": 0.046244547290374576,
      "grad_norm": 1.0827689170837402,
      "learning_rate": 0.00019075128451792362,
      "loss": 4.124,
      "step": 49610
    },
    {
      "epoch": 0.04625386890845367,
      "grad_norm": 0.8921254277229309,
      "learning_rate": 0.0001907494201958284,
      "loss": 4.0559,
      "step": 49620
    },
    {
      "epoch": 0.046263190526532764,
      "grad_norm": 1.6299036741256714,
      "learning_rate": 0.0001907475558737332,
      "loss": 3.8374,
      "step": 49630
    },
    {
      "epoch": 0.04627251214461185,
      "grad_norm": 1.0028979778289795,
      "learning_rate": 0.000190745691551638,
      "loss": 4.0909,
      "step": 49640
    },
    {
      "epoch": 0.046281833762690947,
      "grad_norm": 1.1552295684814453,
      "learning_rate": 0.00019074382722954281,
      "loss": 4.038,
      "step": 49650
    },
    {
      "epoch": 0.04629115538077004,
      "grad_norm": 0.992888867855072,
      "learning_rate": 0.00019074196290744762,
      "loss": 4.0484,
      "step": 49660
    },
    {
      "epoch": 0.04630047699884913,
      "grad_norm": 1.1457387208938599,
      "learning_rate": 0.0001907400985853524,
      "loss": 3.8407,
      "step": 49670
    },
    {
      "epoch": 0.04630979861692822,
      "grad_norm": 1.140293836593628,
      "learning_rate": 0.0001907382342632572,
      "loss": 3.8315,
      "step": 49680
    },
    {
      "epoch": 0.04631912023500731,
      "grad_norm": 0.8997176885604858,
      "learning_rate": 0.00019073636994116198,
      "loss": 3.8808,
      "step": 49690
    },
    {
      "epoch": 0.046328441853086405,
      "grad_norm": 1.1078063249588013,
      "learning_rate": 0.00019073450561906682,
      "loss": 4.1301,
      "step": 49700
    },
    {
      "epoch": 0.0463377634711655,
      "grad_norm": 0.8427147269248962,
      "learning_rate": 0.00019073264129697162,
      "loss": 3.9677,
      "step": 49710
    },
    {
      "epoch": 0.04634708508924459,
      "grad_norm": 1.0501736402511597,
      "learning_rate": 0.0001907307769748764,
      "loss": 4.0956,
      "step": 49720
    },
    {
      "epoch": 0.04635640670732368,
      "grad_norm": 1.2263426780700684,
      "learning_rate": 0.0001907289126527812,
      "loss": 4.0256,
      "step": 49730
    },
    {
      "epoch": 0.04636572832540277,
      "grad_norm": 1.2262300252914429,
      "learning_rate": 0.000190727048330686,
      "loss": 3.9637,
      "step": 49740
    },
    {
      "epoch": 0.046375049943481864,
      "grad_norm": 1.096021056175232,
      "learning_rate": 0.0001907251840085908,
      "loss": 4.2174,
      "step": 49750
    },
    {
      "epoch": 0.04638437156156096,
      "grad_norm": 0.9779959321022034,
      "learning_rate": 0.00019072331968649563,
      "loss": 4.1358,
      "step": 49760
    },
    {
      "epoch": 0.046393693179640046,
      "grad_norm": 1.0931410789489746,
      "learning_rate": 0.0001907214553644004,
      "loss": 4.161,
      "step": 49770
    },
    {
      "epoch": 0.04640301479771914,
      "grad_norm": 1.1807423830032349,
      "learning_rate": 0.00019071959104230521,
      "loss": 4.2084,
      "step": 49780
    },
    {
      "epoch": 0.046412336415798235,
      "grad_norm": 1.066366195678711,
      "learning_rate": 0.00019071772672021,
      "loss": 3.8758,
      "step": 49790
    },
    {
      "epoch": 0.04642165803387732,
      "grad_norm": 0.9745820164680481,
      "learning_rate": 0.0001907158623981148,
      "loss": 3.7333,
      "step": 49800
    },
    {
      "epoch": 0.04643097965195642,
      "grad_norm": 1.254355549812317,
      "learning_rate": 0.0001907139980760196,
      "loss": 4.037,
      "step": 49810
    },
    {
      "epoch": 0.046440301270035504,
      "grad_norm": 0.8815616965293884,
      "learning_rate": 0.0001907121337539244,
      "loss": 4.0997,
      "step": 49820
    },
    {
      "epoch": 0.0464496228881146,
      "grad_norm": 1.0791454315185547,
      "learning_rate": 0.00019071026943182922,
      "loss": 3.9924,
      "step": 49830
    },
    {
      "epoch": 0.04645894450619369,
      "grad_norm": 1.2352826595306396,
      "learning_rate": 0.000190708405109734,
      "loss": 3.9839,
      "step": 49840
    },
    {
      "epoch": 0.04646826612427278,
      "grad_norm": 1.2330060005187988,
      "learning_rate": 0.0001907065407876388,
      "loss": 3.9643,
      "step": 49850
    },
    {
      "epoch": 0.046477587742351875,
      "grad_norm": 1.021666169166565,
      "learning_rate": 0.0001907046764655436,
      "loss": 4.0424,
      "step": 49860
    },
    {
      "epoch": 0.04648690936043097,
      "grad_norm": 0.9924989938735962,
      "learning_rate": 0.00019070281214344842,
      "loss": 4.1096,
      "step": 49870
    },
    {
      "epoch": 0.04649623097851006,
      "grad_norm": 1.2998827695846558,
      "learning_rate": 0.00019070094782135322,
      "loss": 4.1765,
      "step": 49880
    },
    {
      "epoch": 0.04650555259658915,
      "grad_norm": 1.2589688301086426,
      "learning_rate": 0.000190699083499258,
      "loss": 3.7893,
      "step": 49890
    },
    {
      "epoch": 0.04651487421466824,
      "grad_norm": 1.1928846836090088,
      "learning_rate": 0.0001906972191771628,
      "loss": 3.7765,
      "step": 49900
    },
    {
      "epoch": 0.046524195832747334,
      "grad_norm": 1.1867181062698364,
      "learning_rate": 0.00019069535485506762,
      "loss": 3.9629,
      "step": 49910
    },
    {
      "epoch": 0.04653351745082643,
      "grad_norm": 1.1935607194900513,
      "learning_rate": 0.0001906934905329724,
      "loss": 4.1844,
      "step": 49920
    },
    {
      "epoch": 0.046542839068905516,
      "grad_norm": 0.8941218852996826,
      "learning_rate": 0.00019069162621087723,
      "loss": 4.0409,
      "step": 49930
    },
    {
      "epoch": 0.04655216068698461,
      "grad_norm": 1.1824308633804321,
      "learning_rate": 0.000190689761888782,
      "loss": 4.1239,
      "step": 49940
    },
    {
      "epoch": 0.0465614823050637,
      "grad_norm": 1.1308177709579468,
      "learning_rate": 0.0001906878975666868,
      "loss": 3.9122,
      "step": 49950
    },
    {
      "epoch": 0.04657080392314279,
      "grad_norm": 1.1684530973434448,
      "learning_rate": 0.0001906860332445916,
      "loss": 4.1661,
      "step": 49960
    },
    {
      "epoch": 0.04658012554122189,
      "grad_norm": 1.1258996725082397,
      "learning_rate": 0.0001906841689224964,
      "loss": 4.0793,
      "step": 49970
    },
    {
      "epoch": 0.046589447159300974,
      "grad_norm": 1.2383432388305664,
      "learning_rate": 0.0001906823046004012,
      "loss": 4.0368,
      "step": 49980
    },
    {
      "epoch": 0.04659876877738007,
      "grad_norm": 1.1565724611282349,
      "learning_rate": 0.000190680440278306,
      "loss": 3.9763,
      "step": 49990
    },
    {
      "epoch": 0.04660809039545916,
      "grad_norm": 0.9547459483146667,
      "learning_rate": 0.00019067857595621082,
      "loss": 4.0498,
      "step": 50000
    },
    {
      "epoch": 0.04661741201353825,
      "grad_norm": 0.9896571040153503,
      "learning_rate": 0.0001906767116341156,
      "loss": 3.9286,
      "step": 50010
    },
    {
      "epoch": 0.046626733631617345,
      "grad_norm": 1.0152113437652588,
      "learning_rate": 0.0001906748473120204,
      "loss": 3.9141,
      "step": 50020
    },
    {
      "epoch": 0.04663605524969643,
      "grad_norm": 0.9569724798202515,
      "learning_rate": 0.0001906729829899252,
      "loss": 3.986,
      "step": 50030
    },
    {
      "epoch": 0.04664537686777553,
      "grad_norm": 1.2087812423706055,
      "learning_rate": 0.00019067111866783002,
      "loss": 3.9262,
      "step": 50040
    },
    {
      "epoch": 0.04665469848585462,
      "grad_norm": 1.3434263467788696,
      "learning_rate": 0.00019066925434573482,
      "loss": 3.8357,
      "step": 50050
    },
    {
      "epoch": 0.04666402010393371,
      "grad_norm": 1.0005810260772705,
      "learning_rate": 0.0001906673900236396,
      "loss": 4.0822,
      "step": 50060
    },
    {
      "epoch": 0.046673341722012804,
      "grad_norm": 0.8842853307723999,
      "learning_rate": 0.0001906655257015444,
      "loss": 4.0513,
      "step": 50070
    },
    {
      "epoch": 0.04668266334009189,
      "grad_norm": 1.1749564409255981,
      "learning_rate": 0.00019066366137944921,
      "loss": 3.8713,
      "step": 50080
    },
    {
      "epoch": 0.046691984958170986,
      "grad_norm": 1.005937933921814,
      "learning_rate": 0.00019066179705735402,
      "loss": 3.9336,
      "step": 50090
    },
    {
      "epoch": 0.04670130657625008,
      "grad_norm": 1.0826904773712158,
      "learning_rate": 0.00019065993273525883,
      "loss": 3.8944,
      "step": 50100
    },
    {
      "epoch": 0.04671062819432917,
      "grad_norm": 1.0185068845748901,
      "learning_rate": 0.0001906580684131636,
      "loss": 3.9498,
      "step": 50110
    },
    {
      "epoch": 0.04671994981240826,
      "grad_norm": 1.0971894264221191,
      "learning_rate": 0.0001906562040910684,
      "loss": 4.0862,
      "step": 50120
    },
    {
      "epoch": 0.04672927143048736,
      "grad_norm": 1.167976975440979,
      "learning_rate": 0.00019065433976897322,
      "loss": 4.1385,
      "step": 50130
    },
    {
      "epoch": 0.046738593048566444,
      "grad_norm": 1.1671477556228638,
      "learning_rate": 0.000190652475446878,
      "loss": 3.9661,
      "step": 50140
    },
    {
      "epoch": 0.04674791466664554,
      "grad_norm": 0.9575675129890442,
      "learning_rate": 0.00019065061112478283,
      "loss": 3.7026,
      "step": 50150
    },
    {
      "epoch": 0.046757236284724626,
      "grad_norm": 1.3502731323242188,
      "learning_rate": 0.0001906487468026876,
      "loss": 3.9305,
      "step": 50160
    },
    {
      "epoch": 0.04676655790280372,
      "grad_norm": 1.1573331356048584,
      "learning_rate": 0.00019064688248059242,
      "loss": 4.0691,
      "step": 50170
    },
    {
      "epoch": 0.046775879520882815,
      "grad_norm": 1.133545160293579,
      "learning_rate": 0.00019064501815849722,
      "loss": 3.9936,
      "step": 50180
    },
    {
      "epoch": 0.0467852011389619,
      "grad_norm": 0.9067680835723877,
      "learning_rate": 0.000190643153836402,
      "loss": 3.9989,
      "step": 50190
    },
    {
      "epoch": 0.046794522757041,
      "grad_norm": 1.1888175010681152,
      "learning_rate": 0.0001906412895143068,
      "loss": 3.8875,
      "step": 50200
    },
    {
      "epoch": 0.04680384437512009,
      "grad_norm": 1.2953921556472778,
      "learning_rate": 0.00019063942519221161,
      "loss": 3.9797,
      "step": 50210
    },
    {
      "epoch": 0.04681316599319918,
      "grad_norm": 1.1069369316101074,
      "learning_rate": 0.00019063756087011642,
      "loss": 4.1169,
      "step": 50220
    },
    {
      "epoch": 0.046822487611278274,
      "grad_norm": 1.08426034450531,
      "learning_rate": 0.00019063569654802123,
      "loss": 4.0476,
      "step": 50230
    },
    {
      "epoch": 0.04683180922935736,
      "grad_norm": 0.8931918144226074,
      "learning_rate": 0.000190633832225926,
      "loss": 3.8657,
      "step": 50240
    },
    {
      "epoch": 0.046841130847436456,
      "grad_norm": 0.8348345756530762,
      "learning_rate": 0.0001906319679038308,
      "loss": 4.0827,
      "step": 50250
    },
    {
      "epoch": 0.04685045246551555,
      "grad_norm": 1.0555052757263184,
      "learning_rate": 0.00019063010358173562,
      "loss": 4.1569,
      "step": 50260
    },
    {
      "epoch": 0.04685977408359464,
      "grad_norm": 1.013444423675537,
      "learning_rate": 0.00019062823925964043,
      "loss": 4.095,
      "step": 50270
    },
    {
      "epoch": 0.04686909570167373,
      "grad_norm": 1.0526190996170044,
      "learning_rate": 0.00019062637493754523,
      "loss": 4.248,
      "step": 50280
    },
    {
      "epoch": 0.04687841731975282,
      "grad_norm": 1.0470703840255737,
      "learning_rate": 0.00019062451061545,
      "loss": 3.9868,
      "step": 50290
    },
    {
      "epoch": 0.046887738937831915,
      "grad_norm": 1.1313819885253906,
      "learning_rate": 0.00019062264629335482,
      "loss": 3.9368,
      "step": 50300
    },
    {
      "epoch": 0.04689706055591101,
      "grad_norm": 1.2340683937072754,
      "learning_rate": 0.0001906207819712596,
      "loss": 3.9396,
      "step": 50310
    },
    {
      "epoch": 0.0469063821739901,
      "grad_norm": 1.2343509197235107,
      "learning_rate": 0.00019061891764916443,
      "loss": 3.9319,
      "step": 50320
    },
    {
      "epoch": 0.04691570379206919,
      "grad_norm": 0.9660050272941589,
      "learning_rate": 0.00019061705332706924,
      "loss": 4.0185,
      "step": 50330
    },
    {
      "epoch": 0.046925025410148286,
      "grad_norm": 1.0311822891235352,
      "learning_rate": 0.00019061518900497402,
      "loss": 3.9796,
      "step": 50340
    },
    {
      "epoch": 0.04693434702822737,
      "grad_norm": 1.1220782995224,
      "learning_rate": 0.00019061332468287882,
      "loss": 4.0243,
      "step": 50350
    },
    {
      "epoch": 0.04694366864630647,
      "grad_norm": 1.1130353212356567,
      "learning_rate": 0.0001906114603607836,
      "loss": 4.0218,
      "step": 50360
    },
    {
      "epoch": 0.046952990264385555,
      "grad_norm": 1.0981487035751343,
      "learning_rate": 0.0001906095960386884,
      "loss": 4.0376,
      "step": 50370
    },
    {
      "epoch": 0.04696231188246465,
      "grad_norm": 1.2535386085510254,
      "learning_rate": 0.00019060773171659324,
      "loss": 3.9052,
      "step": 50380
    },
    {
      "epoch": 0.046971633500543744,
      "grad_norm": 0.9825420379638672,
      "learning_rate": 0.00019060586739449802,
      "loss": 3.9915,
      "step": 50390
    },
    {
      "epoch": 0.04698095511862283,
      "grad_norm": 1.1208621263504028,
      "learning_rate": 0.00019060400307240283,
      "loss": 3.7289,
      "step": 50400
    },
    {
      "epoch": 0.046990276736701926,
      "grad_norm": 1.2067627906799316,
      "learning_rate": 0.0001906021387503076,
      "loss": 3.9134,
      "step": 50410
    },
    {
      "epoch": 0.046999598354781014,
      "grad_norm": 0.9552167654037476,
      "learning_rate": 0.0001906002744282124,
      "loss": 3.9385,
      "step": 50420
    },
    {
      "epoch": 0.04700891997286011,
      "grad_norm": 1.0751625299453735,
      "learning_rate": 0.00019059841010611722,
      "loss": 4.0755,
      "step": 50430
    },
    {
      "epoch": 0.0470182415909392,
      "grad_norm": 1.145500898361206,
      "learning_rate": 0.00019059654578402202,
      "loss": 3.8682,
      "step": 50440
    },
    {
      "epoch": 0.04702756320901829,
      "grad_norm": 0.9472718834877014,
      "learning_rate": 0.00019059468146192683,
      "loss": 3.945,
      "step": 50450
    },
    {
      "epoch": 0.047036884827097385,
      "grad_norm": 0.9840209484100342,
      "learning_rate": 0.0001905928171398316,
      "loss": 4.012,
      "step": 50460
    },
    {
      "epoch": 0.04704620644517648,
      "grad_norm": 1.0221431255340576,
      "learning_rate": 0.00019059095281773642,
      "loss": 4.0021,
      "step": 50470
    },
    {
      "epoch": 0.04705552806325557,
      "grad_norm": 1.21000075340271,
      "learning_rate": 0.00019058908849564122,
      "loss": 3.926,
      "step": 50480
    },
    {
      "epoch": 0.04706484968133466,
      "grad_norm": 1.222626805305481,
      "learning_rate": 0.00019058722417354603,
      "loss": 4.0615,
      "step": 50490
    },
    {
      "epoch": 0.04707417129941375,
      "grad_norm": 1.202175259590149,
      "learning_rate": 0.00019058535985145083,
      "loss": 3.9794,
      "step": 50500
    },
    {
      "epoch": 0.04708349291749284,
      "grad_norm": 1.2254972457885742,
      "learning_rate": 0.00019058349552935561,
      "loss": 4.0129,
      "step": 50510
    },
    {
      "epoch": 0.04709281453557194,
      "grad_norm": 1.0253875255584717,
      "learning_rate": 0.00019058163120726042,
      "loss": 4.059,
      "step": 50520
    },
    {
      "epoch": 0.047102136153651025,
      "grad_norm": 1.2897530794143677,
      "learning_rate": 0.00019057976688516523,
      "loss": 3.6565,
      "step": 50530
    },
    {
      "epoch": 0.04711145777173012,
      "grad_norm": 1.1853125095367432,
      "learning_rate": 0.00019057790256307003,
      "loss": 4.0286,
      "step": 50540
    },
    {
      "epoch": 0.047120779389809214,
      "grad_norm": 1.0619306564331055,
      "learning_rate": 0.00019057603824097484,
      "loss": 3.9781,
      "step": 50550
    },
    {
      "epoch": 0.0471301010078883,
      "grad_norm": 1.1784639358520508,
      "learning_rate": 0.00019057417391887962,
      "loss": 3.8805,
      "step": 50560
    },
    {
      "epoch": 0.047139422625967396,
      "grad_norm": 1.0430952310562134,
      "learning_rate": 0.00019057230959678442,
      "loss": 3.7873,
      "step": 50570
    },
    {
      "epoch": 0.047148744244046484,
      "grad_norm": 0.8545770645141602,
      "learning_rate": 0.00019057044527468923,
      "loss": 3.8455,
      "step": 50580
    },
    {
      "epoch": 0.04715806586212558,
      "grad_norm": 1.0935766696929932,
      "learning_rate": 0.000190568580952594,
      "loss": 4.0736,
      "step": 50590
    },
    {
      "epoch": 0.04716738748020467,
      "grad_norm": 1.3848415613174438,
      "learning_rate": 0.00019056671663049884,
      "loss": 4.197,
      "step": 50600
    },
    {
      "epoch": 0.04717670909828376,
      "grad_norm": 1.2232117652893066,
      "learning_rate": 0.00019056485230840362,
      "loss": 3.9417,
      "step": 50610
    },
    {
      "epoch": 0.047186030716362855,
      "grad_norm": 1.1100850105285645,
      "learning_rate": 0.00019056298798630843,
      "loss": 3.8746,
      "step": 50620
    },
    {
      "epoch": 0.04719535233444194,
      "grad_norm": 1.1787127256393433,
      "learning_rate": 0.00019056112366421324,
      "loss": 4.0113,
      "step": 50630
    },
    {
      "epoch": 0.04720467395252104,
      "grad_norm": 1.1910654306411743,
      "learning_rate": 0.00019055925934211801,
      "loss": 3.9956,
      "step": 50640
    },
    {
      "epoch": 0.04721399557060013,
      "grad_norm": 0.9259728193283081,
      "learning_rate": 0.00019055739502002282,
      "loss": 4.1797,
      "step": 50650
    },
    {
      "epoch": 0.04722331718867922,
      "grad_norm": 1.1541355848312378,
      "learning_rate": 0.00019055553069792763,
      "loss": 4.2449,
      "step": 50660
    },
    {
      "epoch": 0.04723263880675831,
      "grad_norm": 1.1524564027786255,
      "learning_rate": 0.00019055366637583243,
      "loss": 4.0363,
      "step": 50670
    },
    {
      "epoch": 0.04724196042483741,
      "grad_norm": 0.9950040578842163,
      "learning_rate": 0.00019055180205373724,
      "loss": 3.9105,
      "step": 50680
    },
    {
      "epoch": 0.047251282042916495,
      "grad_norm": 0.9326775074005127,
      "learning_rate": 0.00019054993773164202,
      "loss": 4.0582,
      "step": 50690
    },
    {
      "epoch": 0.04726060366099559,
      "grad_norm": 1.0581929683685303,
      "learning_rate": 0.00019054807340954683,
      "loss": 3.795,
      "step": 50700
    },
    {
      "epoch": 0.04726992527907468,
      "grad_norm": 1.133413314819336,
      "learning_rate": 0.00019054620908745163,
      "loss": 4.1827,
      "step": 50710
    },
    {
      "epoch": 0.04727924689715377,
      "grad_norm": 1.0862394571304321,
      "learning_rate": 0.00019054434476535644,
      "loss": 4.0551,
      "step": 50720
    },
    {
      "epoch": 0.047288568515232866,
      "grad_norm": 1.0579445362091064,
      "learning_rate": 0.00019054248044326124,
      "loss": 4.1119,
      "step": 50730
    },
    {
      "epoch": 0.047297890133311954,
      "grad_norm": 1.1509143114089966,
      "learning_rate": 0.00019054061612116602,
      "loss": 4.1487,
      "step": 50740
    },
    {
      "epoch": 0.04730721175139105,
      "grad_norm": 1.0951472520828247,
      "learning_rate": 0.00019053875179907083,
      "loss": 3.9023,
      "step": 50750
    },
    {
      "epoch": 0.04731653336947014,
      "grad_norm": 1.0714703798294067,
      "learning_rate": 0.00019053688747697564,
      "loss": 3.9392,
      "step": 50760
    },
    {
      "epoch": 0.04732585498754923,
      "grad_norm": 1.166495680809021,
      "learning_rate": 0.00019053502315488044,
      "loss": 3.7807,
      "step": 50770
    },
    {
      "epoch": 0.047335176605628325,
      "grad_norm": 1.1145704984664917,
      "learning_rate": 0.00019053315883278525,
      "loss": 3.7435,
      "step": 50780
    },
    {
      "epoch": 0.04734449822370741,
      "grad_norm": 0.9349634647369385,
      "learning_rate": 0.00019053129451069003,
      "loss": 3.9029,
      "step": 50790
    },
    {
      "epoch": 0.04735381984178651,
      "grad_norm": 1.0552644729614258,
      "learning_rate": 0.00019052943018859483,
      "loss": 3.7435,
      "step": 50800
    },
    {
      "epoch": 0.0473631414598656,
      "grad_norm": 0.983023464679718,
      "learning_rate": 0.00019052756586649961,
      "loss": 4.0565,
      "step": 50810
    },
    {
      "epoch": 0.04737246307794469,
      "grad_norm": 1.153088092803955,
      "learning_rate": 0.00019052570154440445,
      "loss": 3.8472,
      "step": 50820
    },
    {
      "epoch": 0.04738178469602378,
      "grad_norm": 1.050127625465393,
      "learning_rate": 0.00019052383722230925,
      "loss": 3.9595,
      "step": 50830
    },
    {
      "epoch": 0.04739110631410287,
      "grad_norm": 1.0965121984481812,
      "learning_rate": 0.00019052197290021403,
      "loss": 3.8217,
      "step": 50840
    },
    {
      "epoch": 0.047400427932181965,
      "grad_norm": 0.9445101618766785,
      "learning_rate": 0.00019052010857811884,
      "loss": 4.0822,
      "step": 50850
    },
    {
      "epoch": 0.04740974955026106,
      "grad_norm": 1.2415802478790283,
      "learning_rate": 0.00019051824425602362,
      "loss": 4.0607,
      "step": 50860
    },
    {
      "epoch": 0.04741907116834015,
      "grad_norm": 1.1378780603408813,
      "learning_rate": 0.00019051637993392842,
      "loss": 3.9732,
      "step": 50870
    },
    {
      "epoch": 0.04742839278641924,
      "grad_norm": 0.9707638025283813,
      "learning_rate": 0.00019051451561183323,
      "loss": 4.1601,
      "step": 50880
    },
    {
      "epoch": 0.047437714404498336,
      "grad_norm": 1.1573679447174072,
      "learning_rate": 0.00019051265128973804,
      "loss": 3.8615,
      "step": 50890
    },
    {
      "epoch": 0.047447036022577424,
      "grad_norm": 1.276199221611023,
      "learning_rate": 0.00019051078696764284,
      "loss": 4.097,
      "step": 50900
    },
    {
      "epoch": 0.04745635764065652,
      "grad_norm": 1.0083770751953125,
      "learning_rate": 0.00019050892264554762,
      "loss": 3.9253,
      "step": 50910
    },
    {
      "epoch": 0.047465679258735606,
      "grad_norm": 0.8819169998168945,
      "learning_rate": 0.00019050705832345243,
      "loss": 4.1807,
      "step": 50920
    },
    {
      "epoch": 0.0474750008768147,
      "grad_norm": 0.9706562757492065,
      "learning_rate": 0.00019050519400135724,
      "loss": 3.9814,
      "step": 50930
    },
    {
      "epoch": 0.047484322494893795,
      "grad_norm": 0.9630593657493591,
      "learning_rate": 0.00019050332967926204,
      "loss": 3.9533,
      "step": 50940
    },
    {
      "epoch": 0.04749364411297288,
      "grad_norm": 1.1517114639282227,
      "learning_rate": 0.00019050146535716685,
      "loss": 3.9773,
      "step": 50950
    },
    {
      "epoch": 0.04750296573105198,
      "grad_norm": 1.441219687461853,
      "learning_rate": 0.00019049960103507163,
      "loss": 4.0305,
      "step": 50960
    },
    {
      "epoch": 0.047512287349131065,
      "grad_norm": 1.0695277452468872,
      "learning_rate": 0.00019049773671297643,
      "loss": 4.089,
      "step": 50970
    },
    {
      "epoch": 0.04752160896721016,
      "grad_norm": 1.1793328523635864,
      "learning_rate": 0.00019049587239088124,
      "loss": 4.0437,
      "step": 50980
    },
    {
      "epoch": 0.047530930585289254,
      "grad_norm": 1.0727170705795288,
      "learning_rate": 0.00019049400806878605,
      "loss": 3.9636,
      "step": 50990
    },
    {
      "epoch": 0.04754025220336834,
      "grad_norm": 1.1451101303100586,
      "learning_rate": 0.00019049214374669085,
      "loss": 4.1536,
      "step": 51000
    },
    {
      "epoch": 0.047549573821447436,
      "grad_norm": 1.0034621953964233,
      "learning_rate": 0.00019049027942459563,
      "loss": 3.9169,
      "step": 51010
    },
    {
      "epoch": 0.04755889543952653,
      "grad_norm": 1.1341564655303955,
      "learning_rate": 0.00019048841510250044,
      "loss": 4.0911,
      "step": 51020
    },
    {
      "epoch": 0.04756821705760562,
      "grad_norm": 0.9689690470695496,
      "learning_rate": 0.00019048655078040524,
      "loss": 4.1925,
      "step": 51030
    },
    {
      "epoch": 0.04757753867568471,
      "grad_norm": 0.933754026889801,
      "learning_rate": 0.00019048468645831002,
      "loss": 3.995,
      "step": 51040
    },
    {
      "epoch": 0.0475868602937638,
      "grad_norm": 1.084518313407898,
      "learning_rate": 0.00019048282213621486,
      "loss": 3.9474,
      "step": 51050
    },
    {
      "epoch": 0.047596181911842894,
      "grad_norm": 0.9476255178451538,
      "learning_rate": 0.00019048095781411964,
      "loss": 4.1594,
      "step": 51060
    },
    {
      "epoch": 0.04760550352992199,
      "grad_norm": 1.3127080202102661,
      "learning_rate": 0.00019047909349202444,
      "loss": 4.1183,
      "step": 51070
    },
    {
      "epoch": 0.047614825148001076,
      "grad_norm": 0.9757387042045593,
      "learning_rate": 0.00019047722916992925,
      "loss": 3.9614,
      "step": 51080
    },
    {
      "epoch": 0.04762414676608017,
      "grad_norm": 1.0055437088012695,
      "learning_rate": 0.00019047536484783403,
      "loss": 3.8664,
      "step": 51090
    },
    {
      "epoch": 0.047633468384159265,
      "grad_norm": 0.9427285194396973,
      "learning_rate": 0.00019047350052573883,
      "loss": 3.9131,
      "step": 51100
    },
    {
      "epoch": 0.04764279000223835,
      "grad_norm": 1.2023612260818481,
      "learning_rate": 0.00019047163620364364,
      "loss": 3.9825,
      "step": 51110
    },
    {
      "epoch": 0.04765211162031745,
      "grad_norm": 1.0873034000396729,
      "learning_rate": 0.00019046977188154845,
      "loss": 3.9779,
      "step": 51120
    },
    {
      "epoch": 0.047661433238396535,
      "grad_norm": 1.0903749465942383,
      "learning_rate": 0.00019046790755945325,
      "loss": 4.0034,
      "step": 51130
    },
    {
      "epoch": 0.04767075485647563,
      "grad_norm": 1.0606393814086914,
      "learning_rate": 0.00019046604323735803,
      "loss": 4.1008,
      "step": 51140
    },
    {
      "epoch": 0.047680076474554724,
      "grad_norm": 0.9371547698974609,
      "learning_rate": 0.00019046417891526284,
      "loss": 3.9095,
      "step": 51150
    },
    {
      "epoch": 0.04768939809263381,
      "grad_norm": 1.1405797004699707,
      "learning_rate": 0.00019046231459316764,
      "loss": 3.9966,
      "step": 51160
    },
    {
      "epoch": 0.047698719710712906,
      "grad_norm": 0.8405462503433228,
      "learning_rate": 0.00019046045027107245,
      "loss": 4.0359,
      "step": 51170
    },
    {
      "epoch": 0.04770804132879199,
      "grad_norm": 1.2861319780349731,
      "learning_rate": 0.00019045858594897726,
      "loss": 3.9364,
      "step": 51180
    },
    {
      "epoch": 0.04771736294687109,
      "grad_norm": 1.2619270086288452,
      "learning_rate": 0.00019045672162688204,
      "loss": 4.0187,
      "step": 51190
    },
    {
      "epoch": 0.04772668456495018,
      "grad_norm": 1.15921950340271,
      "learning_rate": 0.00019045485730478684,
      "loss": 3.9331,
      "step": 51200
    },
    {
      "epoch": 0.04773600618302927,
      "grad_norm": 1.0585511922836304,
      "learning_rate": 0.00019045299298269165,
      "loss": 4.1481,
      "step": 51210
    },
    {
      "epoch": 0.047745327801108364,
      "grad_norm": 1.1598544120788574,
      "learning_rate": 0.00019045112866059646,
      "loss": 3.8902,
      "step": 51220
    },
    {
      "epoch": 0.04775464941918746,
      "grad_norm": 1.1332167387008667,
      "learning_rate": 0.00019044926433850126,
      "loss": 3.8688,
      "step": 51230
    },
    {
      "epoch": 0.047763971037266546,
      "grad_norm": 1.1464909315109253,
      "learning_rate": 0.00019044740001640604,
      "loss": 4.1077,
      "step": 51240
    },
    {
      "epoch": 0.04777329265534564,
      "grad_norm": 1.0938739776611328,
      "learning_rate": 0.00019044553569431085,
      "loss": 4.1518,
      "step": 51250
    },
    {
      "epoch": 0.04778261427342473,
      "grad_norm": 1.122463345527649,
      "learning_rate": 0.00019044367137221563,
      "loss": 3.9953,
      "step": 51260
    },
    {
      "epoch": 0.04779193589150382,
      "grad_norm": 1.092374324798584,
      "learning_rate": 0.00019044180705012046,
      "loss": 3.9565,
      "step": 51270
    },
    {
      "epoch": 0.04780125750958292,
      "grad_norm": 1.2125144004821777,
      "learning_rate": 0.00019043994272802524,
      "loss": 4.0078,
      "step": 51280
    },
    {
      "epoch": 0.047810579127662005,
      "grad_norm": 1.2452245950698853,
      "learning_rate": 0.00019043807840593005,
      "loss": 3.7347,
      "step": 51290
    },
    {
      "epoch": 0.0478199007457411,
      "grad_norm": 1.3067923784255981,
      "learning_rate": 0.00019043621408383485,
      "loss": 4.0306,
      "step": 51300
    },
    {
      "epoch": 0.04782922236382019,
      "grad_norm": 0.9517296552658081,
      "learning_rate": 0.00019043434976173963,
      "loss": 4.0196,
      "step": 51310
    },
    {
      "epoch": 0.04783854398189928,
      "grad_norm": 1.1189574003219604,
      "learning_rate": 0.00019043248543964444,
      "loss": 3.9096,
      "step": 51320
    },
    {
      "epoch": 0.047847865599978376,
      "grad_norm": 1.0541609525680542,
      "learning_rate": 0.00019043062111754924,
      "loss": 4.1191,
      "step": 51330
    },
    {
      "epoch": 0.04785718721805746,
      "grad_norm": 1.1471840143203735,
      "learning_rate": 0.00019042875679545405,
      "loss": 3.656,
      "step": 51340
    },
    {
      "epoch": 0.04786650883613656,
      "grad_norm": 1.1840039491653442,
      "learning_rate": 0.00019042689247335886,
      "loss": 4.0642,
      "step": 51350
    },
    {
      "epoch": 0.04787583045421565,
      "grad_norm": 1.3543782234191895,
      "learning_rate": 0.00019042502815126364,
      "loss": 4.1866,
      "step": 51360
    },
    {
      "epoch": 0.04788515207229474,
      "grad_norm": 1.0135263204574585,
      "learning_rate": 0.00019042316382916844,
      "loss": 3.9629,
      "step": 51370
    },
    {
      "epoch": 0.047894473690373834,
      "grad_norm": 1.1752296686172485,
      "learning_rate": 0.00019042129950707325,
      "loss": 3.8095,
      "step": 51380
    },
    {
      "epoch": 0.04790379530845292,
      "grad_norm": 1.267585039138794,
      "learning_rate": 0.00019041943518497805,
      "loss": 4.0705,
      "step": 51390
    },
    {
      "epoch": 0.047913116926532016,
      "grad_norm": 1.2811702489852905,
      "learning_rate": 0.00019041757086288286,
      "loss": 4.0913,
      "step": 51400
    },
    {
      "epoch": 0.04792243854461111,
      "grad_norm": 1.0694555044174194,
      "learning_rate": 0.00019041570654078764,
      "loss": 3.9496,
      "step": 51410
    },
    {
      "epoch": 0.0479317601626902,
      "grad_norm": 1.0744857788085938,
      "learning_rate": 0.00019041384221869245,
      "loss": 3.8752,
      "step": 51420
    },
    {
      "epoch": 0.04794108178076929,
      "grad_norm": 1.0122047662734985,
      "learning_rate": 0.00019041197789659723,
      "loss": 4.0251,
      "step": 51430
    },
    {
      "epoch": 0.04795040339884839,
      "grad_norm": 1.1339516639709473,
      "learning_rate": 0.00019041011357450206,
      "loss": 3.9024,
      "step": 51440
    },
    {
      "epoch": 0.047959725016927475,
      "grad_norm": 1.0066001415252686,
      "learning_rate": 0.00019040824925240687,
      "loss": 4.2081,
      "step": 51450
    },
    {
      "epoch": 0.04796904663500657,
      "grad_norm": 0.8842967748641968,
      "learning_rate": 0.00019040638493031164,
      "loss": 3.9239,
      "step": 51460
    },
    {
      "epoch": 0.04797836825308566,
      "grad_norm": 1.088099718093872,
      "learning_rate": 0.00019040452060821645,
      "loss": 3.9298,
      "step": 51470
    },
    {
      "epoch": 0.04798768987116475,
      "grad_norm": 1.1985830068588257,
      "learning_rate": 0.00019040265628612123,
      "loss": 4.0215,
      "step": 51480
    },
    {
      "epoch": 0.047997011489243846,
      "grad_norm": 1.1915684938430786,
      "learning_rate": 0.00019040079196402604,
      "loss": 3.988,
      "step": 51490
    },
    {
      "epoch": 0.048006333107322934,
      "grad_norm": 1.067771315574646,
      "learning_rate": 0.00019039892764193087,
      "loss": 4.0685,
      "step": 51500
    },
    {
      "epoch": 0.04801565472540203,
      "grad_norm": 1.0823097229003906,
      "learning_rate": 0.00019039706331983565,
      "loss": 3.8199,
      "step": 51510
    },
    {
      "epoch": 0.048024976343481116,
      "grad_norm": 1.4124791622161865,
      "learning_rate": 0.00019039519899774046,
      "loss": 4.0729,
      "step": 51520
    },
    {
      "epoch": 0.04803429796156021,
      "grad_norm": 1.2278118133544922,
      "learning_rate": 0.00019039333467564523,
      "loss": 3.8751,
      "step": 51530
    },
    {
      "epoch": 0.048043619579639305,
      "grad_norm": 1.077328085899353,
      "learning_rate": 0.00019039147035355004,
      "loss": 4.1032,
      "step": 51540
    },
    {
      "epoch": 0.04805294119771839,
      "grad_norm": 1.1857000589370728,
      "learning_rate": 0.00019038960603145485,
      "loss": 4.0508,
      "step": 51550
    },
    {
      "epoch": 0.04806226281579749,
      "grad_norm": 1.041938304901123,
      "learning_rate": 0.00019038774170935965,
      "loss": 3.8829,
      "step": 51560
    },
    {
      "epoch": 0.04807158443387658,
      "grad_norm": 0.9403591156005859,
      "learning_rate": 0.00019038587738726446,
      "loss": 4.1229,
      "step": 51570
    },
    {
      "epoch": 0.04808090605195567,
      "grad_norm": 1.1640541553497314,
      "learning_rate": 0.00019038401306516924,
      "loss": 4.0308,
      "step": 51580
    },
    {
      "epoch": 0.04809022767003476,
      "grad_norm": 1.254164457321167,
      "learning_rate": 0.00019038214874307404,
      "loss": 4.086,
      "step": 51590
    },
    {
      "epoch": 0.04809954928811385,
      "grad_norm": 0.9274442791938782,
      "learning_rate": 0.00019038028442097885,
      "loss": 3.8073,
      "step": 51600
    },
    {
      "epoch": 0.048108870906192945,
      "grad_norm": 1.0021014213562012,
      "learning_rate": 0.00019037842009888366,
      "loss": 3.9733,
      "step": 51610
    },
    {
      "epoch": 0.04811819252427204,
      "grad_norm": 1.0733002424240112,
      "learning_rate": 0.00019037655577678846,
      "loss": 4.0462,
      "step": 51620
    },
    {
      "epoch": 0.04812751414235113,
      "grad_norm": 1.2907371520996094,
      "learning_rate": 0.00019037469145469324,
      "loss": 3.9818,
      "step": 51630
    },
    {
      "epoch": 0.04813683576043022,
      "grad_norm": 1.0884323120117188,
      "learning_rate": 0.00019037282713259805,
      "loss": 3.8214,
      "step": 51640
    },
    {
      "epoch": 0.04814615737850931,
      "grad_norm": 1.029462218284607,
      "learning_rate": 0.00019037096281050286,
      "loss": 3.9705,
      "step": 51650
    },
    {
      "epoch": 0.048155478996588404,
      "grad_norm": 1.1576783657073975,
      "learning_rate": 0.00019036909848840766,
      "loss": 4.1463,
      "step": 51660
    },
    {
      "epoch": 0.0481648006146675,
      "grad_norm": 1.0998468399047852,
      "learning_rate": 0.00019036723416631247,
      "loss": 4.0278,
      "step": 51670
    },
    {
      "epoch": 0.048174122232746586,
      "grad_norm": 1.114054799079895,
      "learning_rate": 0.00019036536984421725,
      "loss": 4.1066,
      "step": 51680
    },
    {
      "epoch": 0.04818344385082568,
      "grad_norm": 1.0733360052108765,
      "learning_rate": 0.00019036350552212205,
      "loss": 3.8805,
      "step": 51690
    },
    {
      "epoch": 0.048192765468904775,
      "grad_norm": 1.0103901624679565,
      "learning_rate": 0.00019036164120002686,
      "loss": 4.0941,
      "step": 51700
    },
    {
      "epoch": 0.04820208708698386,
      "grad_norm": 1.1688541173934937,
      "learning_rate": 0.00019035977687793164,
      "loss": 3.9047,
      "step": 51710
    },
    {
      "epoch": 0.04821140870506296,
      "grad_norm": 0.955794632434845,
      "learning_rate": 0.00019035791255583647,
      "loss": 4.1049,
      "step": 51720
    },
    {
      "epoch": 0.048220730323142044,
      "grad_norm": 1.1449525356292725,
      "learning_rate": 0.00019035604823374125,
      "loss": 4.0436,
      "step": 51730
    },
    {
      "epoch": 0.04823005194122114,
      "grad_norm": 0.9690750241279602,
      "learning_rate": 0.00019035418391164606,
      "loss": 3.9063,
      "step": 51740
    },
    {
      "epoch": 0.04823937355930023,
      "grad_norm": 0.9839017987251282,
      "learning_rate": 0.00019035231958955086,
      "loss": 4.178,
      "step": 51750
    },
    {
      "epoch": 0.04824869517737932,
      "grad_norm": 1.157585620880127,
      "learning_rate": 0.00019035045526745564,
      "loss": 3.9845,
      "step": 51760
    },
    {
      "epoch": 0.048258016795458415,
      "grad_norm": 1.1264690160751343,
      "learning_rate": 0.00019034859094536045,
      "loss": 4.0524,
      "step": 51770
    },
    {
      "epoch": 0.04826733841353751,
      "grad_norm": 0.8537757992744446,
      "learning_rate": 0.00019034672662326526,
      "loss": 4.0255,
      "step": 51780
    },
    {
      "epoch": 0.0482766600316166,
      "grad_norm": 0.9645856618881226,
      "learning_rate": 0.00019034486230117006,
      "loss": 3.844,
      "step": 51790
    },
    {
      "epoch": 0.04828598164969569,
      "grad_norm": 1.0810598134994507,
      "learning_rate": 0.00019034299797907487,
      "loss": 4.0062,
      "step": 51800
    },
    {
      "epoch": 0.04829530326777478,
      "grad_norm": 1.0321564674377441,
      "learning_rate": 0.00019034113365697965,
      "loss": 3.9334,
      "step": 51810
    },
    {
      "epoch": 0.048304624885853874,
      "grad_norm": 1.032718539237976,
      "learning_rate": 0.00019033926933488445,
      "loss": 4.01,
      "step": 51820
    },
    {
      "epoch": 0.04831394650393297,
      "grad_norm": 1.0359948873519897,
      "learning_rate": 0.00019033740501278926,
      "loss": 3.9936,
      "step": 51830
    },
    {
      "epoch": 0.048323268122012056,
      "grad_norm": 1.0590801239013672,
      "learning_rate": 0.00019033554069069407,
      "loss": 4.1584,
      "step": 51840
    },
    {
      "epoch": 0.04833258974009115,
      "grad_norm": 1.4014358520507812,
      "learning_rate": 0.00019033367636859887,
      "loss": 3.9558,
      "step": 51850
    },
    {
      "epoch": 0.04834191135817024,
      "grad_norm": 1.2383917570114136,
      "learning_rate": 0.00019033181204650365,
      "loss": 3.814,
      "step": 51860
    },
    {
      "epoch": 0.04835123297624933,
      "grad_norm": 1.0017070770263672,
      "learning_rate": 0.00019032994772440846,
      "loss": 4.0076,
      "step": 51870
    },
    {
      "epoch": 0.04836055459432843,
      "grad_norm": 0.9916624426841736,
      "learning_rate": 0.00019032808340231324,
      "loss": 4.0409,
      "step": 51880
    },
    {
      "epoch": 0.048369876212407514,
      "grad_norm": 1.0337773561477661,
      "learning_rate": 0.00019032621908021807,
      "loss": 3.9474,
      "step": 51890
    },
    {
      "epoch": 0.04837919783048661,
      "grad_norm": 1.023322343826294,
      "learning_rate": 0.00019032435475812288,
      "loss": 3.9431,
      "step": 51900
    },
    {
      "epoch": 0.0483885194485657,
      "grad_norm": 0.9651661515235901,
      "learning_rate": 0.00019032249043602766,
      "loss": 4.0904,
      "step": 51910
    },
    {
      "epoch": 0.04839784106664479,
      "grad_norm": 1.195092797279358,
      "learning_rate": 0.00019032062611393246,
      "loss": 4.1018,
      "step": 51920
    },
    {
      "epoch": 0.048407162684723885,
      "grad_norm": 1.2980226278305054,
      "learning_rate": 0.00019031876179183724,
      "loss": 3.8794,
      "step": 51930
    },
    {
      "epoch": 0.04841648430280297,
      "grad_norm": 1.0408236980438232,
      "learning_rate": 0.00019031689746974205,
      "loss": 4.0413,
      "step": 51940
    },
    {
      "epoch": 0.04842580592088207,
      "grad_norm": 1.1615105867385864,
      "learning_rate": 0.00019031503314764688,
      "loss": 3.9497,
      "step": 51950
    },
    {
      "epoch": 0.04843512753896116,
      "grad_norm": 1.0148766040802002,
      "learning_rate": 0.00019031316882555166,
      "loss": 3.8529,
      "step": 51960
    },
    {
      "epoch": 0.04844444915704025,
      "grad_norm": 1.1175127029418945,
      "learning_rate": 0.00019031130450345647,
      "loss": 4.0229,
      "step": 51970
    },
    {
      "epoch": 0.048453770775119344,
      "grad_norm": 1.1222851276397705,
      "learning_rate": 0.00019030944018136125,
      "loss": 3.693,
      "step": 51980
    },
    {
      "epoch": 0.04846309239319843,
      "grad_norm": 1.0603095293045044,
      "learning_rate": 0.00019030757585926605,
      "loss": 3.9226,
      "step": 51990
    },
    {
      "epoch": 0.048472414011277526,
      "grad_norm": 1.0699764490127563,
      "learning_rate": 0.00019030571153717086,
      "loss": 4.0347,
      "step": 52000
    },
    {
      "epoch": 0.04848173562935662,
      "grad_norm": 1.3677568435668945,
      "learning_rate": 0.00019030384721507567,
      "loss": 4.0199,
      "step": 52010
    },
    {
      "epoch": 0.04849105724743571,
      "grad_norm": 1.069220781326294,
      "learning_rate": 0.00019030198289298047,
      "loss": 3.9878,
      "step": 52020
    },
    {
      "epoch": 0.0485003788655148,
      "grad_norm": 0.9836592674255371,
      "learning_rate": 0.00019030011857088525,
      "loss": 3.9736,
      "step": 52030
    },
    {
      "epoch": 0.0485097004835939,
      "grad_norm": 1.0516679286956787,
      "learning_rate": 0.00019029825424879006,
      "loss": 3.9071,
      "step": 52040
    },
    {
      "epoch": 0.048519022101672984,
      "grad_norm": 1.233656883239746,
      "learning_rate": 0.00019029638992669486,
      "loss": 3.961,
      "step": 52050
    },
    {
      "epoch": 0.04852834371975208,
      "grad_norm": 0.9845261573791504,
      "learning_rate": 0.00019029452560459967,
      "loss": 4.0193,
      "step": 52060
    },
    {
      "epoch": 0.048537665337831166,
      "grad_norm": 0.9983953833580017,
      "learning_rate": 0.00019029266128250448,
      "loss": 4.0651,
      "step": 52070
    },
    {
      "epoch": 0.04854698695591026,
      "grad_norm": 1.2217745780944824,
      "learning_rate": 0.00019029079696040926,
      "loss": 4.1716,
      "step": 52080
    },
    {
      "epoch": 0.048556308573989355,
      "grad_norm": 1.1102994680404663,
      "learning_rate": 0.00019028893263831406,
      "loss": 4.0065,
      "step": 52090
    },
    {
      "epoch": 0.04856563019206844,
      "grad_norm": 1.0740454196929932,
      "learning_rate": 0.00019028706831621887,
      "loss": 4.0966,
      "step": 52100
    },
    {
      "epoch": 0.04857495181014754,
      "grad_norm": 1.3164176940917969,
      "learning_rate": 0.00019028520399412367,
      "loss": 4.0206,
      "step": 52110
    },
    {
      "epoch": 0.04858427342822663,
      "grad_norm": 0.9690202474594116,
      "learning_rate": 0.00019028333967202848,
      "loss": 3.9883,
      "step": 52120
    },
    {
      "epoch": 0.04859359504630572,
      "grad_norm": 1.0566238164901733,
      "learning_rate": 0.00019028147534993326,
      "loss": 4.0397,
      "step": 52130
    },
    {
      "epoch": 0.048602916664384814,
      "grad_norm": 0.9812890291213989,
      "learning_rate": 0.00019027961102783807,
      "loss": 3.8028,
      "step": 52140
    },
    {
      "epoch": 0.0486122382824639,
      "grad_norm": 1.133306860923767,
      "learning_rate": 0.00019027774670574287,
      "loss": 3.9434,
      "step": 52150
    },
    {
      "epoch": 0.048621559900542996,
      "grad_norm": 1.088868498802185,
      "learning_rate": 0.00019027588238364765,
      "loss": 3.9793,
      "step": 52160
    },
    {
      "epoch": 0.04863088151862209,
      "grad_norm": 1.218489646911621,
      "learning_rate": 0.00019027401806155249,
      "loss": 3.9983,
      "step": 52170
    },
    {
      "epoch": 0.04864020313670118,
      "grad_norm": 1.1420084238052368,
      "learning_rate": 0.00019027215373945726,
      "loss": 3.9683,
      "step": 52180
    },
    {
      "epoch": 0.04864952475478027,
      "grad_norm": 1.018351674079895,
      "learning_rate": 0.00019027028941736207,
      "loss": 4.0642,
      "step": 52190
    },
    {
      "epoch": 0.04865884637285936,
      "grad_norm": 1.187373161315918,
      "learning_rate": 0.00019026842509526688,
      "loss": 3.8701,
      "step": 52200
    },
    {
      "epoch": 0.048668167990938455,
      "grad_norm": 1.0723788738250732,
      "learning_rate": 0.00019026656077317166,
      "loss": 3.8407,
      "step": 52210
    },
    {
      "epoch": 0.04867748960901755,
      "grad_norm": 1.0737602710723877,
      "learning_rate": 0.00019026469645107646,
      "loss": 4.0173,
      "step": 52220
    },
    {
      "epoch": 0.04868681122709664,
      "grad_norm": 1.168038010597229,
      "learning_rate": 0.00019026283212898127,
      "loss": 3.9671,
      "step": 52230
    },
    {
      "epoch": 0.04869613284517573,
      "grad_norm": 1.157233476638794,
      "learning_rate": 0.00019026096780688608,
      "loss": 4.0713,
      "step": 52240
    },
    {
      "epoch": 0.048705454463254826,
      "grad_norm": 1.1416804790496826,
      "learning_rate": 0.00019025910348479088,
      "loss": 4.0041,
      "step": 52250
    },
    {
      "epoch": 0.04871477608133391,
      "grad_norm": 0.9044649004936218,
      "learning_rate": 0.00019025723916269566,
      "loss": 3.9099,
      "step": 52260
    },
    {
      "epoch": 0.04872409769941301,
      "grad_norm": 1.1288886070251465,
      "learning_rate": 0.00019025537484060047,
      "loss": 3.9769,
      "step": 52270
    },
    {
      "epoch": 0.048733419317492095,
      "grad_norm": 1.019107699394226,
      "learning_rate": 0.00019025351051850527,
      "loss": 3.7579,
      "step": 52280
    },
    {
      "epoch": 0.04874274093557119,
      "grad_norm": 1.4094700813293457,
      "learning_rate": 0.00019025164619641008,
      "loss": 4.1519,
      "step": 52290
    },
    {
      "epoch": 0.048752062553650284,
      "grad_norm": 0.9295184016227722,
      "learning_rate": 0.00019024978187431489,
      "loss": 3.9841,
      "step": 52300
    },
    {
      "epoch": 0.04876138417172937,
      "grad_norm": 1.1396265029907227,
      "learning_rate": 0.00019024791755221967,
      "loss": 3.7866,
      "step": 52310
    },
    {
      "epoch": 0.048770705789808466,
      "grad_norm": 1.0487240552902222,
      "learning_rate": 0.00019024605323012447,
      "loss": 3.9906,
      "step": 52320
    },
    {
      "epoch": 0.048780027407887554,
      "grad_norm": 1.0190057754516602,
      "learning_rate": 0.00019024418890802925,
      "loss": 4.0344,
      "step": 52330
    },
    {
      "epoch": 0.04878934902596665,
      "grad_norm": 1.0357906818389893,
      "learning_rate": 0.00019024232458593408,
      "loss": 4.0483,
      "step": 52340
    },
    {
      "epoch": 0.04879867064404574,
      "grad_norm": 1.032223105430603,
      "learning_rate": 0.0001902404602638389,
      "loss": 4.1659,
      "step": 52350
    },
    {
      "epoch": 0.04880799226212483,
      "grad_norm": 1.0666842460632324,
      "learning_rate": 0.00019023859594174367,
      "loss": 3.9127,
      "step": 52360
    },
    {
      "epoch": 0.048817313880203925,
      "grad_norm": 1.0624085664749146,
      "learning_rate": 0.00019023673161964848,
      "loss": 4.0133,
      "step": 52370
    },
    {
      "epoch": 0.04882663549828302,
      "grad_norm": 1.1968843936920166,
      "learning_rate": 0.00019023486729755326,
      "loss": 4.0634,
      "step": 52380
    },
    {
      "epoch": 0.04883595711636211,
      "grad_norm": 1.0159063339233398,
      "learning_rate": 0.00019023300297545806,
      "loss": 3.8738,
      "step": 52390
    },
    {
      "epoch": 0.0488452787344412,
      "grad_norm": 0.9843985438346863,
      "learning_rate": 0.0001902311386533629,
      "loss": 3.865,
      "step": 52400
    },
    {
      "epoch": 0.04885460035252029,
      "grad_norm": 1.203291893005371,
      "learning_rate": 0.00019022927433126767,
      "loss": 3.9988,
      "step": 52410
    },
    {
      "epoch": 0.04886392197059938,
      "grad_norm": 1.0978569984436035,
      "learning_rate": 0.00019022741000917248,
      "loss": 3.963,
      "step": 52420
    },
    {
      "epoch": 0.04887324358867848,
      "grad_norm": 1.132837176322937,
      "learning_rate": 0.00019022554568707726,
      "loss": 3.9874,
      "step": 52430
    },
    {
      "epoch": 0.048882565206757565,
      "grad_norm": 1.1584084033966064,
      "learning_rate": 0.00019022368136498207,
      "loss": 4.0119,
      "step": 52440
    },
    {
      "epoch": 0.04889188682483666,
      "grad_norm": 0.9946032166481018,
      "learning_rate": 0.00019022181704288687,
      "loss": 4.0329,
      "step": 52450
    },
    {
      "epoch": 0.048901208442915754,
      "grad_norm": 1.028126835823059,
      "learning_rate": 0.00019021995272079168,
      "loss": 3.9868,
      "step": 52460
    },
    {
      "epoch": 0.04891053006099484,
      "grad_norm": 1.0525895357131958,
      "learning_rate": 0.00019021808839869649,
      "loss": 3.9701,
      "step": 52470
    },
    {
      "epoch": 0.048919851679073936,
      "grad_norm": 1.0752750635147095,
      "learning_rate": 0.00019021622407660126,
      "loss": 4.0082,
      "step": 52480
    },
    {
      "epoch": 0.048929173297153024,
      "grad_norm": 1.1557427644729614,
      "learning_rate": 0.00019021435975450607,
      "loss": 3.9007,
      "step": 52490
    },
    {
      "epoch": 0.04893849491523212,
      "grad_norm": 0.9192332625389099,
      "learning_rate": 0.00019021249543241088,
      "loss": 4.0924,
      "step": 52500
    },
    {
      "epoch": 0.04894781653331121,
      "grad_norm": 1.097381353378296,
      "learning_rate": 0.00019021063111031568,
      "loss": 3.9288,
      "step": 52510
    },
    {
      "epoch": 0.0489571381513903,
      "grad_norm": 0.8682412505149841,
      "learning_rate": 0.0001902087667882205,
      "loss": 4.0346,
      "step": 52520
    },
    {
      "epoch": 0.048966459769469395,
      "grad_norm": 1.0960739850997925,
      "learning_rate": 0.00019020690246612527,
      "loss": 3.9707,
      "step": 52530
    },
    {
      "epoch": 0.04897578138754848,
      "grad_norm": 1.0120391845703125,
      "learning_rate": 0.00019020503814403008,
      "loss": 3.8762,
      "step": 52540
    },
    {
      "epoch": 0.04898510300562758,
      "grad_norm": 1.1290276050567627,
      "learning_rate": 0.00019020317382193488,
      "loss": 4.015,
      "step": 52550
    },
    {
      "epoch": 0.04899442462370667,
      "grad_norm": 0.9447847604751587,
      "learning_rate": 0.0001902013094998397,
      "loss": 4.1235,
      "step": 52560
    },
    {
      "epoch": 0.04900374624178576,
      "grad_norm": 0.8718826174736023,
      "learning_rate": 0.0001901994451777445,
      "loss": 3.8024,
      "step": 52570
    },
    {
      "epoch": 0.04901306785986485,
      "grad_norm": 1.1131418943405151,
      "learning_rate": 0.00019019758085564927,
      "loss": 4.0935,
      "step": 52580
    },
    {
      "epoch": 0.04902238947794395,
      "grad_norm": 1.2631232738494873,
      "learning_rate": 0.00019019571653355408,
      "loss": 4.0679,
      "step": 52590
    },
    {
      "epoch": 0.049031711096023035,
      "grad_norm": 1.02800714969635,
      "learning_rate": 0.00019019385221145889,
      "loss": 4.1465,
      "step": 52600
    },
    {
      "epoch": 0.04904103271410213,
      "grad_norm": 0.890275239944458,
      "learning_rate": 0.00019019198788936367,
      "loss": 3.8129,
      "step": 52610
    },
    {
      "epoch": 0.04905035433218122,
      "grad_norm": 1.060954213142395,
      "learning_rate": 0.0001901901235672685,
      "loss": 3.8293,
      "step": 52620
    },
    {
      "epoch": 0.04905967595026031,
      "grad_norm": 1.1454901695251465,
      "learning_rate": 0.00019018825924517328,
      "loss": 4.0972,
      "step": 52630
    },
    {
      "epoch": 0.049068997568339406,
      "grad_norm": 1.2567743062973022,
      "learning_rate": 0.00019018639492307808,
      "loss": 3.9775,
      "step": 52640
    },
    {
      "epoch": 0.049078319186418494,
      "grad_norm": 1.060060739517212,
      "learning_rate": 0.00019018453060098286,
      "loss": 4.0032,
      "step": 52650
    },
    {
      "epoch": 0.04908764080449759,
      "grad_norm": 1.0091736316680908,
      "learning_rate": 0.00019018266627888767,
      "loss": 4.061,
      "step": 52660
    },
    {
      "epoch": 0.049096962422576676,
      "grad_norm": 0.9816426038742065,
      "learning_rate": 0.00019018080195679248,
      "loss": 3.9758,
      "step": 52670
    },
    {
      "epoch": 0.04910628404065577,
      "grad_norm": 1.07598876953125,
      "learning_rate": 0.00019017893763469728,
      "loss": 4.1143,
      "step": 52680
    },
    {
      "epoch": 0.049115605658734865,
      "grad_norm": 1.0197527408599854,
      "learning_rate": 0.0001901770733126021,
      "loss": 3.7756,
      "step": 52690
    },
    {
      "epoch": 0.04912492727681395,
      "grad_norm": 1.0069514513015747,
      "learning_rate": 0.00019017520899050687,
      "loss": 3.9439,
      "step": 52700
    },
    {
      "epoch": 0.04913424889489305,
      "grad_norm": 1.0476287603378296,
      "learning_rate": 0.00019017334466841167,
      "loss": 4.0298,
      "step": 52710
    },
    {
      "epoch": 0.04914357051297214,
      "grad_norm": 1.0215991735458374,
      "learning_rate": 0.00019017148034631648,
      "loss": 3.8627,
      "step": 52720
    },
    {
      "epoch": 0.04915289213105123,
      "grad_norm": 1.2083076238632202,
      "learning_rate": 0.0001901696160242213,
      "loss": 3.895,
      "step": 52730
    },
    {
      "epoch": 0.049162213749130323,
      "grad_norm": 1.0498013496398926,
      "learning_rate": 0.0001901677517021261,
      "loss": 3.8756,
      "step": 52740
    },
    {
      "epoch": 0.04917153536720941,
      "grad_norm": 1.3667387962341309,
      "learning_rate": 0.00019016588738003087,
      "loss": 4.2092,
      "step": 52750
    },
    {
      "epoch": 0.049180856985288506,
      "grad_norm": 1.0388942956924438,
      "learning_rate": 0.00019016402305793568,
      "loss": 4.2216,
      "step": 52760
    },
    {
      "epoch": 0.0491901786033676,
      "grad_norm": 0.793200671672821,
      "learning_rate": 0.00019016215873584048,
      "loss": 3.8507,
      "step": 52770
    },
    {
      "epoch": 0.04919950022144669,
      "grad_norm": 0.9647352695465088,
      "learning_rate": 0.00019016029441374526,
      "loss": 3.7876,
      "step": 52780
    },
    {
      "epoch": 0.04920882183952578,
      "grad_norm": 0.9066522121429443,
      "learning_rate": 0.0001901584300916501,
      "loss": 3.7429,
      "step": 52790
    },
    {
      "epoch": 0.049218143457604877,
      "grad_norm": 0.8982858061790466,
      "learning_rate": 0.00019015656576955488,
      "loss": 3.8846,
      "step": 52800
    },
    {
      "epoch": 0.049227465075683964,
      "grad_norm": 1.1063075065612793,
      "learning_rate": 0.00019015470144745968,
      "loss": 4.1926,
      "step": 52810
    },
    {
      "epoch": 0.04923678669376306,
      "grad_norm": 0.9703342914581299,
      "learning_rate": 0.0001901528371253645,
      "loss": 3.9167,
      "step": 52820
    },
    {
      "epoch": 0.049246108311842146,
      "grad_norm": 1.1349700689315796,
      "learning_rate": 0.00019015097280326927,
      "loss": 4.0956,
      "step": 52830
    },
    {
      "epoch": 0.04925542992992124,
      "grad_norm": 0.8662824630737305,
      "learning_rate": 0.00019014910848117407,
      "loss": 3.9625,
      "step": 52840
    },
    {
      "epoch": 0.049264751548000335,
      "grad_norm": 1.1090459823608398,
      "learning_rate": 0.00019014724415907888,
      "loss": 4.1265,
      "step": 52850
    },
    {
      "epoch": 0.04927407316607942,
      "grad_norm": 1.2390105724334717,
      "learning_rate": 0.0001901453798369837,
      "loss": 4.0181,
      "step": 52860
    },
    {
      "epoch": 0.04928339478415852,
      "grad_norm": 1.2567996978759766,
      "learning_rate": 0.0001901435155148885,
      "loss": 4.0263,
      "step": 52870
    },
    {
      "epoch": 0.049292716402237605,
      "grad_norm": 1.013869285583496,
      "learning_rate": 0.00019014165119279327,
      "loss": 4.0259,
      "step": 52880
    },
    {
      "epoch": 0.0493020380203167,
      "grad_norm": 0.974222719669342,
      "learning_rate": 0.00019013978687069808,
      "loss": 3.8174,
      "step": 52890
    },
    {
      "epoch": 0.049311359638395794,
      "grad_norm": 0.9824211001396179,
      "learning_rate": 0.00019013792254860289,
      "loss": 3.8252,
      "step": 52900
    },
    {
      "epoch": 0.04932068125647488,
      "grad_norm": 1.0029547214508057,
      "learning_rate": 0.0001901360582265077,
      "loss": 3.9249,
      "step": 52910
    },
    {
      "epoch": 0.049330002874553976,
      "grad_norm": 1.0627325773239136,
      "learning_rate": 0.0001901341939044125,
      "loss": 4.0373,
      "step": 52920
    },
    {
      "epoch": 0.04933932449263307,
      "grad_norm": 1.0240182876586914,
      "learning_rate": 0.00019013232958231728,
      "loss": 4.1937,
      "step": 52930
    },
    {
      "epoch": 0.04934864611071216,
      "grad_norm": 1.0602868795394897,
      "learning_rate": 0.00019013046526022208,
      "loss": 4.096,
      "step": 52940
    },
    {
      "epoch": 0.04935796772879125,
      "grad_norm": 1.0309416055679321,
      "learning_rate": 0.0001901286009381269,
      "loss": 4.0659,
      "step": 52950
    },
    {
      "epoch": 0.04936728934687034,
      "grad_norm": 1.1004406213760376,
      "learning_rate": 0.0001901267366160317,
      "loss": 3.7543,
      "step": 52960
    },
    {
      "epoch": 0.049376610964949434,
      "grad_norm": 1.1468878984451294,
      "learning_rate": 0.0001901248722939365,
      "loss": 3.9145,
      "step": 52970
    },
    {
      "epoch": 0.04938593258302853,
      "grad_norm": 1.2198406457901,
      "learning_rate": 0.00019012300797184128,
      "loss": 3.9085,
      "step": 52980
    },
    {
      "epoch": 0.049395254201107616,
      "grad_norm": 1.1885517835617065,
      "learning_rate": 0.0001901211436497461,
      "loss": 3.9813,
      "step": 52990
    },
    {
      "epoch": 0.04940457581918671,
      "grad_norm": 0.9630497694015503,
      "learning_rate": 0.00019011927932765087,
      "loss": 3.8672,
      "step": 53000
    },
    {
      "epoch": 0.0494138974372658,
      "grad_norm": 1.0455225706100464,
      "learning_rate": 0.0001901174150055557,
      "loss": 3.9878,
      "step": 53010
    },
    {
      "epoch": 0.04942321905534489,
      "grad_norm": 1.002293348312378,
      "learning_rate": 0.0001901155506834605,
      "loss": 4.0887,
      "step": 53020
    },
    {
      "epoch": 0.04943254067342399,
      "grad_norm": 1.1717571020126343,
      "learning_rate": 0.00019011368636136529,
      "loss": 3.9558,
      "step": 53030
    },
    {
      "epoch": 0.049441862291503075,
      "grad_norm": 1.0725455284118652,
      "learning_rate": 0.0001901118220392701,
      "loss": 4.0707,
      "step": 53040
    },
    {
      "epoch": 0.04945118390958217,
      "grad_norm": 1.0287587642669678,
      "learning_rate": 0.00019010995771717487,
      "loss": 3.9231,
      "step": 53050
    },
    {
      "epoch": 0.049460505527661264,
      "grad_norm": 1.0581423044204712,
      "learning_rate": 0.00019010809339507968,
      "loss": 4.0132,
      "step": 53060
    },
    {
      "epoch": 0.04946982714574035,
      "grad_norm": 1.23546302318573,
      "learning_rate": 0.0001901062290729845,
      "loss": 4.0171,
      "step": 53070
    },
    {
      "epoch": 0.049479148763819446,
      "grad_norm": 1.1342984437942505,
      "learning_rate": 0.0001901043647508893,
      "loss": 4.0198,
      "step": 53080
    },
    {
      "epoch": 0.04948847038189853,
      "grad_norm": 1.1323219537734985,
      "learning_rate": 0.0001901025004287941,
      "loss": 3.9582,
      "step": 53090
    },
    {
      "epoch": 0.04949779199997763,
      "grad_norm": 1.1224942207336426,
      "learning_rate": 0.00019010063610669888,
      "loss": 3.8578,
      "step": 53100
    },
    {
      "epoch": 0.04950711361805672,
      "grad_norm": 1.0612050294876099,
      "learning_rate": 0.00019009877178460368,
      "loss": 4.1104,
      "step": 53110
    },
    {
      "epoch": 0.04951643523613581,
      "grad_norm": 0.9596996307373047,
      "learning_rate": 0.0001900969074625085,
      "loss": 4.1128,
      "step": 53120
    },
    {
      "epoch": 0.049525756854214904,
      "grad_norm": 1.1112821102142334,
      "learning_rate": 0.0001900950431404133,
      "loss": 4.0564,
      "step": 53130
    },
    {
      "epoch": 0.049535078472294,
      "grad_norm": 1.009475827217102,
      "learning_rate": 0.0001900931788183181,
      "loss": 3.7938,
      "step": 53140
    },
    {
      "epoch": 0.049544400090373086,
      "grad_norm": 0.985642671585083,
      "learning_rate": 0.00019009131449622288,
      "loss": 3.9746,
      "step": 53150
    },
    {
      "epoch": 0.04955372170845218,
      "grad_norm": 1.1766470670700073,
      "learning_rate": 0.0001900894501741277,
      "loss": 3.8402,
      "step": 53160
    },
    {
      "epoch": 0.04956304332653127,
      "grad_norm": 1.0497075319290161,
      "learning_rate": 0.0001900875858520325,
      "loss": 3.9996,
      "step": 53170
    },
    {
      "epoch": 0.04957236494461036,
      "grad_norm": 1.1899710893630981,
      "learning_rate": 0.0001900857215299373,
      "loss": 3.9361,
      "step": 53180
    },
    {
      "epoch": 0.04958168656268946,
      "grad_norm": 1.1460700035095215,
      "learning_rate": 0.0001900838572078421,
      "loss": 3.8134,
      "step": 53190
    },
    {
      "epoch": 0.049591008180768545,
      "grad_norm": 1.1546872854232788,
      "learning_rate": 0.00019008199288574688,
      "loss": 4.1371,
      "step": 53200
    },
    {
      "epoch": 0.04960032979884764,
      "grad_norm": 0.8299172520637512,
      "learning_rate": 0.0001900801285636517,
      "loss": 3.9959,
      "step": 53210
    },
    {
      "epoch": 0.04960965141692673,
      "grad_norm": 1.23526930809021,
      "learning_rate": 0.0001900782642415565,
      "loss": 3.9295,
      "step": 53220
    },
    {
      "epoch": 0.04961897303500582,
      "grad_norm": 0.9475097060203552,
      "learning_rate": 0.00019007639991946128,
      "loss": 4.0296,
      "step": 53230
    },
    {
      "epoch": 0.049628294653084916,
      "grad_norm": 1.120211124420166,
      "learning_rate": 0.0001900745355973661,
      "loss": 4.0829,
      "step": 53240
    },
    {
      "epoch": 0.049637616271164,
      "grad_norm": 0.9970521926879883,
      "learning_rate": 0.0001900726712752709,
      "loss": 4.2206,
      "step": 53250
    },
    {
      "epoch": 0.0496469378892431,
      "grad_norm": 1.2346781492233276,
      "learning_rate": 0.0001900708069531757,
      "loss": 4.1523,
      "step": 53260
    },
    {
      "epoch": 0.04965625950732219,
      "grad_norm": 1.223468542098999,
      "learning_rate": 0.0001900689426310805,
      "loss": 3.7545,
      "step": 53270
    },
    {
      "epoch": 0.04966558112540128,
      "grad_norm": 1.0059282779693604,
      "learning_rate": 0.00019006707830898528,
      "loss": 3.859,
      "step": 53280
    },
    {
      "epoch": 0.049674902743480374,
      "grad_norm": 1.14925217628479,
      "learning_rate": 0.0001900652139868901,
      "loss": 4.0871,
      "step": 53290
    },
    {
      "epoch": 0.04968422436155946,
      "grad_norm": 1.1807366609573364,
      "learning_rate": 0.0001900633496647949,
      "loss": 3.9277,
      "step": 53300
    },
    {
      "epoch": 0.049693545979638556,
      "grad_norm": 1.0846649408340454,
      "learning_rate": 0.0001900614853426997,
      "loss": 3.8657,
      "step": 53310
    },
    {
      "epoch": 0.04970286759771765,
      "grad_norm": 1.131718635559082,
      "learning_rate": 0.0001900596210206045,
      "loss": 4.0393,
      "step": 53320
    },
    {
      "epoch": 0.04971218921579674,
      "grad_norm": 0.9905151724815369,
      "learning_rate": 0.00019005775669850929,
      "loss": 3.8972,
      "step": 53330
    },
    {
      "epoch": 0.04972151083387583,
      "grad_norm": 1.091191053390503,
      "learning_rate": 0.0001900558923764141,
      "loss": 3.889,
      "step": 53340
    },
    {
      "epoch": 0.04973083245195492,
      "grad_norm": 1.1165807247161865,
      "learning_rate": 0.0001900540280543189,
      "loss": 3.9414,
      "step": 53350
    },
    {
      "epoch": 0.049740154070034015,
      "grad_norm": 1.1068611145019531,
      "learning_rate": 0.0001900521637322237,
      "loss": 4.0629,
      "step": 53360
    },
    {
      "epoch": 0.04974947568811311,
      "grad_norm": 0.9300864934921265,
      "learning_rate": 0.0001900502994101285,
      "loss": 3.9646,
      "step": 53370
    },
    {
      "epoch": 0.0497587973061922,
      "grad_norm": 1.1716879606246948,
      "learning_rate": 0.0001900484350880333,
      "loss": 3.9088,
      "step": 53380
    },
    {
      "epoch": 0.04976811892427129,
      "grad_norm": 1.170115351676941,
      "learning_rate": 0.0001900465707659381,
      "loss": 4.0354,
      "step": 53390
    },
    {
      "epoch": 0.049777440542350386,
      "grad_norm": 1.0930538177490234,
      "learning_rate": 0.0001900447064438429,
      "loss": 3.8476,
      "step": 53400
    },
    {
      "epoch": 0.049786762160429474,
      "grad_norm": 1.1031423807144165,
      "learning_rate": 0.0001900428421217477,
      "loss": 3.9873,
      "step": 53410
    },
    {
      "epoch": 0.04979608377850857,
      "grad_norm": 1.2593384981155396,
      "learning_rate": 0.00019004097779965252,
      "loss": 4.1057,
      "step": 53420
    },
    {
      "epoch": 0.049805405396587656,
      "grad_norm": 1.0727815628051758,
      "learning_rate": 0.0001900391134775573,
      "loss": 4.0289,
      "step": 53430
    },
    {
      "epoch": 0.04981472701466675,
      "grad_norm": 1.250939130783081,
      "learning_rate": 0.0001900372491554621,
      "loss": 3.9262,
      "step": 53440
    },
    {
      "epoch": 0.049824048632745845,
      "grad_norm": 0.9438419342041016,
      "learning_rate": 0.00019003538483336688,
      "loss": 3.9096,
      "step": 53450
    },
    {
      "epoch": 0.04983337025082493,
      "grad_norm": 1.084609031677246,
      "learning_rate": 0.0001900335205112717,
      "loss": 4.0401,
      "step": 53460
    },
    {
      "epoch": 0.04984269186890403,
      "grad_norm": 1.2394778728485107,
      "learning_rate": 0.00019003165618917652,
      "loss": 3.8765,
      "step": 53470
    },
    {
      "epoch": 0.04985201348698312,
      "grad_norm": 1.1877071857452393,
      "learning_rate": 0.0001900297918670813,
      "loss": 4.0494,
      "step": 53480
    },
    {
      "epoch": 0.04986133510506221,
      "grad_norm": 1.1064248085021973,
      "learning_rate": 0.0001900279275449861,
      "loss": 3.7949,
      "step": 53490
    },
    {
      "epoch": 0.0498706567231413,
      "grad_norm": 1.2699158191680908,
      "learning_rate": 0.00019002606322289088,
      "loss": 4.0632,
      "step": 53500
    },
    {
      "epoch": 0.04987997834122039,
      "grad_norm": 1.292856216430664,
      "learning_rate": 0.0001900241989007957,
      "loss": 4.1018,
      "step": 53510
    },
    {
      "epoch": 0.049889299959299485,
      "grad_norm": 1.0493067502975464,
      "learning_rate": 0.00019002233457870052,
      "loss": 4.1369,
      "step": 53520
    },
    {
      "epoch": 0.04989862157737858,
      "grad_norm": 0.9208949208259583,
      "learning_rate": 0.0001900204702566053,
      "loss": 4.0486,
      "step": 53530
    },
    {
      "epoch": 0.04990794319545767,
      "grad_norm": 1.047650694847107,
      "learning_rate": 0.0001900186059345101,
      "loss": 3.924,
      "step": 53540
    },
    {
      "epoch": 0.04991726481353676,
      "grad_norm": 1.0758771896362305,
      "learning_rate": 0.0001900167416124149,
      "loss": 4.133,
      "step": 53550
    },
    {
      "epoch": 0.04992658643161585,
      "grad_norm": 1.0548338890075684,
      "learning_rate": 0.0001900148772903197,
      "loss": 3.9524,
      "step": 53560
    },
    {
      "epoch": 0.049935908049694944,
      "grad_norm": 1.1946834325790405,
      "learning_rate": 0.0001900130129682245,
      "loss": 3.9741,
      "step": 53570
    },
    {
      "epoch": 0.04994522966777404,
      "grad_norm": 1.1608365774154663,
      "learning_rate": 0.0001900111486461293,
      "loss": 4.0553,
      "step": 53580
    },
    {
      "epoch": 0.049954551285853126,
      "grad_norm": 1.034985899925232,
      "learning_rate": 0.00019000928432403411,
      "loss": 4.0439,
      "step": 53590
    },
    {
      "epoch": 0.04996387290393222,
      "grad_norm": 1.265123963356018,
      "learning_rate": 0.0001900074200019389,
      "loss": 3.8613,
      "step": 53600
    },
    {
      "epoch": 0.049973194522011315,
      "grad_norm": 1.1853097677230835,
      "learning_rate": 0.0001900055556798437,
      "loss": 3.9447,
      "step": 53610
    },
    {
      "epoch": 0.0499825161400904,
      "grad_norm": 1.1016924381256104,
      "learning_rate": 0.0001900036913577485,
      "loss": 4.2946,
      "step": 53620
    },
    {
      "epoch": 0.0499918377581695,
      "grad_norm": 1.0355335474014282,
      "learning_rate": 0.0001900018270356533,
      "loss": 4.0166,
      "step": 53630
    },
    {
      "epoch": 0.050001159376248584,
      "grad_norm": 1.1224232912063599,
      "learning_rate": 0.00018999996271355812,
      "loss": 4.1158,
      "step": 53640
    },
    {
      "epoch": 0.05001048099432768,
      "grad_norm": 1.2092539072036743,
      "learning_rate": 0.0001899980983914629,
      "loss": 4.0043,
      "step": 53650
    },
    {
      "epoch": 0.05001980261240677,
      "grad_norm": 0.9483979940414429,
      "learning_rate": 0.0001899962340693677,
      "loss": 4.1426,
      "step": 53660
    },
    {
      "epoch": 0.05002912423048586,
      "grad_norm": 1.1503595113754272,
      "learning_rate": 0.0001899943697472725,
      "loss": 3.8882,
      "step": 53670
    },
    {
      "epoch": 0.050038445848564955,
      "grad_norm": 0.9981784224510193,
      "learning_rate": 0.0001899925054251773,
      "loss": 3.8037,
      "step": 53680
    },
    {
      "epoch": 0.05004776746664404,
      "grad_norm": 1.0340566635131836,
      "learning_rate": 0.00018999064110308212,
      "loss": 4.0186,
      "step": 53690
    },
    {
      "epoch": 0.05005708908472314,
      "grad_norm": 1.0315334796905518,
      "learning_rate": 0.0001899887767809869,
      "loss": 3.9266,
      "step": 53700
    },
    {
      "epoch": 0.05006641070280223,
      "grad_norm": 1.0373125076293945,
      "learning_rate": 0.0001899869124588917,
      "loss": 3.7438,
      "step": 53710
    },
    {
      "epoch": 0.05007573232088132,
      "grad_norm": 1.3293014764785767,
      "learning_rate": 0.00018998504813679651,
      "loss": 4.0291,
      "step": 53720
    },
    {
      "epoch": 0.050085053938960414,
      "grad_norm": 0.9974205493927002,
      "learning_rate": 0.0001899831838147013,
      "loss": 3.9871,
      "step": 53730
    },
    {
      "epoch": 0.05009437555703951,
      "grad_norm": 1.1111798286437988,
      "learning_rate": 0.0001899813194926061,
      "loss": 4.0187,
      "step": 53740
    },
    {
      "epoch": 0.050103697175118596,
      "grad_norm": 1.0456178188323975,
      "learning_rate": 0.0001899794551705109,
      "loss": 4.0069,
      "step": 53750
    },
    {
      "epoch": 0.05011301879319769,
      "grad_norm": 0.949039876461029,
      "learning_rate": 0.0001899775908484157,
      "loss": 4.0365,
      "step": 53760
    },
    {
      "epoch": 0.05012234041127678,
      "grad_norm": 0.9489142894744873,
      "learning_rate": 0.00018997572652632052,
      "loss": 3.9974,
      "step": 53770
    },
    {
      "epoch": 0.05013166202935587,
      "grad_norm": 1.2537903785705566,
      "learning_rate": 0.0001899738622042253,
      "loss": 3.888,
      "step": 53780
    },
    {
      "epoch": 0.05014098364743497,
      "grad_norm": 0.9995027184486389,
      "learning_rate": 0.0001899719978821301,
      "loss": 3.7991,
      "step": 53790
    },
    {
      "epoch": 0.050150305265514054,
      "grad_norm": 1.1903274059295654,
      "learning_rate": 0.0001899701335600349,
      "loss": 4.0063,
      "step": 53800
    },
    {
      "epoch": 0.05015962688359315,
      "grad_norm": 1.1222928762435913,
      "learning_rate": 0.00018996826923793972,
      "loss": 4.0106,
      "step": 53810
    },
    {
      "epoch": 0.05016894850167224,
      "grad_norm": 1.2595831155776978,
      "learning_rate": 0.00018996640491584452,
      "loss": 3.8824,
      "step": 53820
    },
    {
      "epoch": 0.05017827011975133,
      "grad_norm": 1.0167723894119263,
      "learning_rate": 0.0001899645405937493,
      "loss": 4.2878,
      "step": 53830
    },
    {
      "epoch": 0.050187591737830425,
      "grad_norm": 1.050758957862854,
      "learning_rate": 0.0001899626762716541,
      "loss": 3.8763,
      "step": 53840
    },
    {
      "epoch": 0.05019691335590951,
      "grad_norm": 0.8869226574897766,
      "learning_rate": 0.00018996081194955892,
      "loss": 4.1138,
      "step": 53850
    },
    {
      "epoch": 0.05020623497398861,
      "grad_norm": 0.9892594814300537,
      "learning_rate": 0.00018995894762746372,
      "loss": 3.8907,
      "step": 53860
    },
    {
      "epoch": 0.0502155565920677,
      "grad_norm": 1.0402002334594727,
      "learning_rate": 0.00018995708330536853,
      "loss": 4.06,
      "step": 53870
    },
    {
      "epoch": 0.05022487821014679,
      "grad_norm": 0.8666843771934509,
      "learning_rate": 0.0001899552189832733,
      "loss": 3.8608,
      "step": 53880
    },
    {
      "epoch": 0.050234199828225884,
      "grad_norm": 1.062688946723938,
      "learning_rate": 0.00018995335466117811,
      "loss": 4.1133,
      "step": 53890
    },
    {
      "epoch": 0.05024352144630497,
      "grad_norm": 1.256613850593567,
      "learning_rate": 0.0001899514903390829,
      "loss": 3.8673,
      "step": 53900
    },
    {
      "epoch": 0.050252843064384066,
      "grad_norm": 0.9581888914108276,
      "learning_rate": 0.00018994962601698773,
      "loss": 4.0597,
      "step": 53910
    },
    {
      "epoch": 0.05026216468246316,
      "grad_norm": 1.231951117515564,
      "learning_rate": 0.00018994776169489253,
      "loss": 4.0153,
      "step": 53920
    },
    {
      "epoch": 0.05027148630054225,
      "grad_norm": 1.1480845212936401,
      "learning_rate": 0.0001899458973727973,
      "loss": 3.8276,
      "step": 53930
    },
    {
      "epoch": 0.05028080791862134,
      "grad_norm": 1.0884634256362915,
      "learning_rate": 0.00018994403305070212,
      "loss": 3.772,
      "step": 53940
    },
    {
      "epoch": 0.05029012953670044,
      "grad_norm": 1.0897213220596313,
      "learning_rate": 0.0001899421687286069,
      "loss": 4.0693,
      "step": 53950
    },
    {
      "epoch": 0.050299451154779524,
      "grad_norm": 1.1261855363845825,
      "learning_rate": 0.0001899403044065117,
      "loss": 4.2126,
      "step": 53960
    },
    {
      "epoch": 0.05030877277285862,
      "grad_norm": 1.0891705751419067,
      "learning_rate": 0.00018993844008441654,
      "loss": 3.8426,
      "step": 53970
    },
    {
      "epoch": 0.050318094390937707,
      "grad_norm": 0.9789795279502869,
      "learning_rate": 0.00018993657576232132,
      "loss": 3.8285,
      "step": 53980
    },
    {
      "epoch": 0.0503274160090168,
      "grad_norm": 1.1282511949539185,
      "learning_rate": 0.00018993471144022612,
      "loss": 3.8069,
      "step": 53990
    },
    {
      "epoch": 0.050336737627095895,
      "grad_norm": 1.1246930360794067,
      "learning_rate": 0.0001899328471181309,
      "loss": 4.0806,
      "step": 54000
    },
    {
      "epoch": 0.05034605924517498,
      "grad_norm": 1.0147156715393066,
      "learning_rate": 0.0001899309827960357,
      "loss": 4.1488,
      "step": 54010
    },
    {
      "epoch": 0.05035538086325408,
      "grad_norm": 1.262258529663086,
      "learning_rate": 0.00018992911847394051,
      "loss": 3.941,
      "step": 54020
    },
    {
      "epoch": 0.050364702481333165,
      "grad_norm": 1.2037303447723389,
      "learning_rate": 0.00018992725415184532,
      "loss": 3.892,
      "step": 54030
    },
    {
      "epoch": 0.05037402409941226,
      "grad_norm": 1.1891191005706787,
      "learning_rate": 0.00018992538982975013,
      "loss": 4.0047,
      "step": 54040
    },
    {
      "epoch": 0.050383345717491354,
      "grad_norm": 1.18002188205719,
      "learning_rate": 0.0001899235255076549,
      "loss": 3.9682,
      "step": 54050
    },
    {
      "epoch": 0.05039266733557044,
      "grad_norm": 1.0445619821548462,
      "learning_rate": 0.0001899216611855597,
      "loss": 4.1319,
      "step": 54060
    },
    {
      "epoch": 0.050401988953649536,
      "grad_norm": 1.205309510231018,
      "learning_rate": 0.00018991979686346452,
      "loss": 3.9642,
      "step": 54070
    },
    {
      "epoch": 0.05041131057172863,
      "grad_norm": 1.0511980056762695,
      "learning_rate": 0.00018991793254136933,
      "loss": 3.8749,
      "step": 54080
    },
    {
      "epoch": 0.05042063218980772,
      "grad_norm": 1.1716368198394775,
      "learning_rate": 0.00018991606821927413,
      "loss": 3.8716,
      "step": 54090
    },
    {
      "epoch": 0.05042995380788681,
      "grad_norm": 1.186587929725647,
      "learning_rate": 0.0001899142038971789,
      "loss": 4.0962,
      "step": 54100
    },
    {
      "epoch": 0.0504392754259659,
      "grad_norm": 1.1793594360351562,
      "learning_rate": 0.00018991233957508372,
      "loss": 3.8378,
      "step": 54110
    },
    {
      "epoch": 0.050448597044044995,
      "grad_norm": 1.020412564277649,
      "learning_rate": 0.0001899104752529885,
      "loss": 3.9519,
      "step": 54120
    },
    {
      "epoch": 0.05045791866212409,
      "grad_norm": 0.97439044713974,
      "learning_rate": 0.0001899086109308933,
      "loss": 3.9702,
      "step": 54130
    },
    {
      "epoch": 0.05046724028020318,
      "grad_norm": 1.1034995317459106,
      "learning_rate": 0.00018990674660879814,
      "loss": 3.9884,
      "step": 54140
    },
    {
      "epoch": 0.05047656189828227,
      "grad_norm": 1.015865445137024,
      "learning_rate": 0.00018990488228670291,
      "loss": 3.9955,
      "step": 54150
    },
    {
      "epoch": 0.050485883516361366,
      "grad_norm": 1.0370631217956543,
      "learning_rate": 0.00018990301796460772,
      "loss": 3.9606,
      "step": 54160
    },
    {
      "epoch": 0.05049520513444045,
      "grad_norm": 1.0465246438980103,
      "learning_rate": 0.0001899011536425125,
      "loss": 3.9367,
      "step": 54170
    },
    {
      "epoch": 0.05050452675251955,
      "grad_norm": 1.078006625175476,
      "learning_rate": 0.0001898992893204173,
      "loss": 4.0306,
      "step": 54180
    },
    {
      "epoch": 0.050513848370598635,
      "grad_norm": 1.0303400754928589,
      "learning_rate": 0.0001898974249983221,
      "loss": 3.9521,
      "step": 54190
    },
    {
      "epoch": 0.05052316998867773,
      "grad_norm": 1.0479248762130737,
      "learning_rate": 0.00018989556067622692,
      "loss": 4.0868,
      "step": 54200
    },
    {
      "epoch": 0.050532491606756824,
      "grad_norm": 1.0889379978179932,
      "learning_rate": 0.00018989369635413173,
      "loss": 3.8489,
      "step": 54210
    },
    {
      "epoch": 0.05054181322483591,
      "grad_norm": 1.0870758295059204,
      "learning_rate": 0.0001898918320320365,
      "loss": 4.1617,
      "step": 54220
    },
    {
      "epoch": 0.050551134842915006,
      "grad_norm": 1.2487398386001587,
      "learning_rate": 0.0001898899677099413,
      "loss": 4.2416,
      "step": 54230
    },
    {
      "epoch": 0.050560456460994094,
      "grad_norm": 1.0156649351119995,
      "learning_rate": 0.00018988810338784612,
      "loss": 4.0487,
      "step": 54240
    },
    {
      "epoch": 0.05056977807907319,
      "grad_norm": 0.9794331192970276,
      "learning_rate": 0.00018988623906575092,
      "loss": 4.0417,
      "step": 54250
    },
    {
      "epoch": 0.05057909969715228,
      "grad_norm": 1.011210560798645,
      "learning_rate": 0.00018988437474365573,
      "loss": 3.7863,
      "step": 54260
    },
    {
      "epoch": 0.05058842131523137,
      "grad_norm": 1.1381922960281372,
      "learning_rate": 0.0001898825104215605,
      "loss": 3.7988,
      "step": 54270
    },
    {
      "epoch": 0.050597742933310465,
      "grad_norm": 1.2351351976394653,
      "learning_rate": 0.00018988064609946532,
      "loss": 4.0163,
      "step": 54280
    },
    {
      "epoch": 0.05060706455138956,
      "grad_norm": 1.1319317817687988,
      "learning_rate": 0.00018987878177737012,
      "loss": 3.8411,
      "step": 54290
    },
    {
      "epoch": 0.05061638616946865,
      "grad_norm": 0.9949221611022949,
      "learning_rate": 0.00018987691745527493,
      "loss": 3.9349,
      "step": 54300
    },
    {
      "epoch": 0.05062570778754774,
      "grad_norm": 0.946142315864563,
      "learning_rate": 0.00018987505313317973,
      "loss": 3.9924,
      "step": 54310
    },
    {
      "epoch": 0.05063502940562683,
      "grad_norm": 1.0172568559646606,
      "learning_rate": 0.00018987318881108451,
      "loss": 4.1213,
      "step": 54320
    },
    {
      "epoch": 0.05064435102370592,
      "grad_norm": 1.1695421934127808,
      "learning_rate": 0.00018987132448898932,
      "loss": 3.8121,
      "step": 54330
    },
    {
      "epoch": 0.05065367264178502,
      "grad_norm": 0.9489980340003967,
      "learning_rate": 0.00018986946016689413,
      "loss": 4.0546,
      "step": 54340
    },
    {
      "epoch": 0.050662994259864105,
      "grad_norm": 1.2506983280181885,
      "learning_rate": 0.0001898675958447989,
      "loss": 4.0465,
      "step": 54350
    },
    {
      "epoch": 0.0506723158779432,
      "grad_norm": Infinity,
      "learning_rate": 0.00018986573152270374,
      "loss": 4.0468,
      "step": 54360
    },
    {
      "epoch": 0.05068163749602229,
      "grad_norm": 1.1251038312911987,
      "learning_rate": 0.00018986386720060852,
      "loss": 4.0357,
      "step": 54370
    },
    {
      "epoch": 0.05069095911410138,
      "grad_norm": 1.0949782133102417,
      "learning_rate": 0.00018986200287851332,
      "loss": 3.9449,
      "step": 54380
    },
    {
      "epoch": 0.050700280732180476,
      "grad_norm": 1.2182979583740234,
      "learning_rate": 0.00018986013855641813,
      "loss": 3.7805,
      "step": 54390
    },
    {
      "epoch": 0.050709602350259564,
      "grad_norm": 1.113644003868103,
      "learning_rate": 0.0001898582742343229,
      "loss": 3.9334,
      "step": 54400
    },
    {
      "epoch": 0.05071892396833866,
      "grad_norm": 1.1298491954803467,
      "learning_rate": 0.00018985640991222772,
      "loss": 3.9697,
      "step": 54410
    },
    {
      "epoch": 0.05072824558641775,
      "grad_norm": 0.9587352275848389,
      "learning_rate": 0.00018985454559013252,
      "loss": 4.0213,
      "step": 54420
    },
    {
      "epoch": 0.05073756720449684,
      "grad_norm": 0.9741267561912537,
      "learning_rate": 0.00018985268126803733,
      "loss": 3.9903,
      "step": 54430
    },
    {
      "epoch": 0.050746888822575935,
      "grad_norm": 0.946594774723053,
      "learning_rate": 0.00018985081694594214,
      "loss": 4.0074,
      "step": 54440
    },
    {
      "epoch": 0.05075621044065502,
      "grad_norm": 1.0723285675048828,
      "learning_rate": 0.00018984895262384691,
      "loss": 4.0891,
      "step": 54450
    },
    {
      "epoch": 0.05076553205873412,
      "grad_norm": 1.1333924531936646,
      "learning_rate": 0.00018984708830175172,
      "loss": 3.847,
      "step": 54460
    },
    {
      "epoch": 0.05077485367681321,
      "grad_norm": 1.101178526878357,
      "learning_rate": 0.00018984522397965653,
      "loss": 3.9017,
      "step": 54470
    },
    {
      "epoch": 0.0507841752948923,
      "grad_norm": 1.2154256105422974,
      "learning_rate": 0.00018984335965756133,
      "loss": 3.9719,
      "step": 54480
    },
    {
      "epoch": 0.05079349691297139,
      "grad_norm": 1.0569943189620972,
      "learning_rate": 0.00018984149533546614,
      "loss": 3.9602,
      "step": 54490
    },
    {
      "epoch": 0.05080281853105049,
      "grad_norm": 1.0167988538742065,
      "learning_rate": 0.00018983963101337092,
      "loss": 3.9534,
      "step": 54500
    },
    {
      "epoch": 0.050812140149129575,
      "grad_norm": 1.103438377380371,
      "learning_rate": 0.00018983776669127573,
      "loss": 4.0198,
      "step": 54510
    },
    {
      "epoch": 0.05082146176720867,
      "grad_norm": 1.110276222229004,
      "learning_rate": 0.00018983590236918053,
      "loss": 3.8662,
      "step": 54520
    },
    {
      "epoch": 0.05083078338528776,
      "grad_norm": 1.2325329780578613,
      "learning_rate": 0.00018983403804708534,
      "loss": 4.0594,
      "step": 54530
    },
    {
      "epoch": 0.05084010500336685,
      "grad_norm": 0.9819337725639343,
      "learning_rate": 0.00018983217372499014,
      "loss": 3.9322,
      "step": 54540
    },
    {
      "epoch": 0.050849426621445946,
      "grad_norm": 1.0913150310516357,
      "learning_rate": 0.00018983030940289492,
      "loss": 4.0535,
      "step": 54550
    },
    {
      "epoch": 0.050858748239525034,
      "grad_norm": 1.0539488792419434,
      "learning_rate": 0.00018982844508079973,
      "loss": 3.9958,
      "step": 54560
    },
    {
      "epoch": 0.05086806985760413,
      "grad_norm": 1.201755404472351,
      "learning_rate": 0.0001898265807587045,
      "loss": 3.952,
      "step": 54570
    },
    {
      "epoch": 0.050877391475683216,
      "grad_norm": 1.069761037826538,
      "learning_rate": 0.00018982471643660934,
      "loss": 3.8913,
      "step": 54580
    },
    {
      "epoch": 0.05088671309376231,
      "grad_norm": 1.228421688079834,
      "learning_rate": 0.00018982285211451415,
      "loss": 3.8904,
      "step": 54590
    },
    {
      "epoch": 0.050896034711841405,
      "grad_norm": 1.0618542432785034,
      "learning_rate": 0.00018982098779241893,
      "loss": 3.9964,
      "step": 54600
    },
    {
      "epoch": 0.05090535632992049,
      "grad_norm": 1.053091287612915,
      "learning_rate": 0.00018981912347032373,
      "loss": 3.8277,
      "step": 54610
    },
    {
      "epoch": 0.05091467794799959,
      "grad_norm": 0.9875824451446533,
      "learning_rate": 0.0001898172591482285,
      "loss": 3.928,
      "step": 54620
    },
    {
      "epoch": 0.05092399956607868,
      "grad_norm": 1.0411696434020996,
      "learning_rate": 0.00018981539482613332,
      "loss": 3.9657,
      "step": 54630
    },
    {
      "epoch": 0.05093332118415777,
      "grad_norm": 1.0972607135772705,
      "learning_rate": 0.00018981353050403815,
      "loss": 3.8353,
      "step": 54640
    },
    {
      "epoch": 0.050942642802236864,
      "grad_norm": 1.0579349994659424,
      "learning_rate": 0.00018981166618194293,
      "loss": 3.9917,
      "step": 54650
    },
    {
      "epoch": 0.05095196442031595,
      "grad_norm": 1.1440762281417847,
      "learning_rate": 0.00018980980185984774,
      "loss": 4.253,
      "step": 54660
    },
    {
      "epoch": 0.050961286038395046,
      "grad_norm": 0.8934158086776733,
      "learning_rate": 0.00018980793753775252,
      "loss": 4.0154,
      "step": 54670
    },
    {
      "epoch": 0.05097060765647414,
      "grad_norm": 0.9167093634605408,
      "learning_rate": 0.00018980607321565732,
      "loss": 3.9655,
      "step": 54680
    },
    {
      "epoch": 0.05097992927455323,
      "grad_norm": 1.1283314228057861,
      "learning_rate": 0.00018980420889356213,
      "loss": 4.0915,
      "step": 54690
    },
    {
      "epoch": 0.05098925089263232,
      "grad_norm": 1.0613257884979248,
      "learning_rate": 0.00018980234457146694,
      "loss": 4.0578,
      "step": 54700
    },
    {
      "epoch": 0.05099857251071141,
      "grad_norm": 1.1052449941635132,
      "learning_rate": 0.00018980048024937174,
      "loss": 4.0925,
      "step": 54710
    },
    {
      "epoch": 0.051007894128790504,
      "grad_norm": 0.9526045918464661,
      "learning_rate": 0.00018979861592727652,
      "loss": 4.0342,
      "step": 54720
    },
    {
      "epoch": 0.0510172157468696,
      "grad_norm": 1.0092687606811523,
      "learning_rate": 0.00018979675160518133,
      "loss": 3.9687,
      "step": 54730
    },
    {
      "epoch": 0.051026537364948686,
      "grad_norm": 1.368599772453308,
      "learning_rate": 0.00018979488728308613,
      "loss": 3.961,
      "step": 54740
    },
    {
      "epoch": 0.05103585898302778,
      "grad_norm": 1.1687692403793335,
      "learning_rate": 0.00018979302296099094,
      "loss": 3.9648,
      "step": 54750
    },
    {
      "epoch": 0.051045180601106875,
      "grad_norm": 1.1918468475341797,
      "learning_rate": 0.00018979115863889575,
      "loss": 3.8141,
      "step": 54760
    },
    {
      "epoch": 0.05105450221918596,
      "grad_norm": 1.1442252397537231,
      "learning_rate": 0.00018978929431680053,
      "loss": 4.0215,
      "step": 54770
    },
    {
      "epoch": 0.05106382383726506,
      "grad_norm": 1.0858385562896729,
      "learning_rate": 0.00018978742999470533,
      "loss": 3.9267,
      "step": 54780
    },
    {
      "epoch": 0.051073145455344145,
      "grad_norm": 0.9028558731079102,
      "learning_rate": 0.00018978556567261014,
      "loss": 3.9172,
      "step": 54790
    },
    {
      "epoch": 0.05108246707342324,
      "grad_norm": 0.8852251768112183,
      "learning_rate": 0.00018978370135051492,
      "loss": 4.0348,
      "step": 54800
    },
    {
      "epoch": 0.051091788691502334,
      "grad_norm": 1.0890032052993774,
      "learning_rate": 0.00018978183702841975,
      "loss": 3.9552,
      "step": 54810
    },
    {
      "epoch": 0.05110111030958142,
      "grad_norm": 0.9055136442184448,
      "learning_rate": 0.00018977997270632453,
      "loss": 4.0467,
      "step": 54820
    },
    {
      "epoch": 0.051110431927660516,
      "grad_norm": 1.2022173404693604,
      "learning_rate": 0.00018977810838422934,
      "loss": 3.9309,
      "step": 54830
    },
    {
      "epoch": 0.05111975354573961,
      "grad_norm": 1.0120112895965576,
      "learning_rate": 0.00018977624406213414,
      "loss": 4.0431,
      "step": 54840
    },
    {
      "epoch": 0.0511290751638187,
      "grad_norm": 1.0956937074661255,
      "learning_rate": 0.00018977437974003892,
      "loss": 3.9278,
      "step": 54850
    },
    {
      "epoch": 0.05113839678189779,
      "grad_norm": 1.3976891040802002,
      "learning_rate": 0.00018977251541794373,
      "loss": 3.8169,
      "step": 54860
    },
    {
      "epoch": 0.05114771839997688,
      "grad_norm": 1.1047590970993042,
      "learning_rate": 0.00018977065109584854,
      "loss": 3.8513,
      "step": 54870
    },
    {
      "epoch": 0.051157040018055974,
      "grad_norm": 1.0813417434692383,
      "learning_rate": 0.00018976878677375334,
      "loss": 3.8697,
      "step": 54880
    },
    {
      "epoch": 0.05116636163613507,
      "grad_norm": 0.9304043054580688,
      "learning_rate": 0.00018976692245165815,
      "loss": 4.0808,
      "step": 54890
    },
    {
      "epoch": 0.051175683254214156,
      "grad_norm": 1.1221001148223877,
      "learning_rate": 0.00018976505812956293,
      "loss": 3.7505,
      "step": 54900
    },
    {
      "epoch": 0.05118500487229325,
      "grad_norm": 0.9647524952888489,
      "learning_rate": 0.00018976319380746773,
      "loss": 3.9384,
      "step": 54910
    },
    {
      "epoch": 0.05119432649037234,
      "grad_norm": 1.0898343324661255,
      "learning_rate": 0.00018976132948537254,
      "loss": 3.9514,
      "step": 54920
    },
    {
      "epoch": 0.05120364810845143,
      "grad_norm": 1.0535120964050293,
      "learning_rate": 0.00018975946516327735,
      "loss": 4.0644,
      "step": 54930
    },
    {
      "epoch": 0.05121296972653053,
      "grad_norm": 1.0267596244812012,
      "learning_rate": 0.00018975760084118215,
      "loss": 3.988,
      "step": 54940
    },
    {
      "epoch": 0.051222291344609615,
      "grad_norm": 0.9630237817764282,
      "learning_rate": 0.00018975573651908693,
      "loss": 4.168,
      "step": 54950
    },
    {
      "epoch": 0.05123161296268871,
      "grad_norm": 1.102724313735962,
      "learning_rate": 0.00018975387219699174,
      "loss": 4.1191,
      "step": 54960
    },
    {
      "epoch": 0.051240934580767804,
      "grad_norm": 1.3319966793060303,
      "learning_rate": 0.00018975200787489654,
      "loss": 4.0179,
      "step": 54970
    },
    {
      "epoch": 0.05125025619884689,
      "grad_norm": 1.1887202262878418,
      "learning_rate": 0.00018975014355280135,
      "loss": 4.0141,
      "step": 54980
    },
    {
      "epoch": 0.051259577816925986,
      "grad_norm": 1.0767474174499512,
      "learning_rate": 0.00018974827923070616,
      "loss": 4.0455,
      "step": 54990
    },
    {
      "epoch": 0.05126889943500507,
      "grad_norm": 1.0629984140396118,
      "learning_rate": 0.00018974641490861094,
      "loss": 3.928,
      "step": 55000
    },
    {
      "epoch": 0.05127822105308417,
      "grad_norm": 1.2935397624969482,
      "learning_rate": 0.00018974455058651574,
      "loss": 3.8708,
      "step": 55010
    },
    {
      "epoch": 0.05128754267116326,
      "grad_norm": 1.0130137205123901,
      "learning_rate": 0.00018974268626442052,
      "loss": 4.0835,
      "step": 55020
    },
    {
      "epoch": 0.05129686428924235,
      "grad_norm": 1.1023401021957397,
      "learning_rate": 0.00018974082194232536,
      "loss": 4.0915,
      "step": 55030
    },
    {
      "epoch": 0.051306185907321444,
      "grad_norm": 1.1682580709457397,
      "learning_rate": 0.00018973895762023016,
      "loss": 3.8545,
      "step": 55040
    },
    {
      "epoch": 0.05131550752540053,
      "grad_norm": 1.2555439472198486,
      "learning_rate": 0.00018973709329813494,
      "loss": 3.9028,
      "step": 55050
    },
    {
      "epoch": 0.051324829143479626,
      "grad_norm": 1.0197631120681763,
      "learning_rate": 0.00018973522897603975,
      "loss": 3.9915,
      "step": 55060
    },
    {
      "epoch": 0.05133415076155872,
      "grad_norm": 0.8333966135978699,
      "learning_rate": 0.00018973336465394453,
      "loss": 3.928,
      "step": 55070
    },
    {
      "epoch": 0.05134347237963781,
      "grad_norm": 0.9891183972358704,
      "learning_rate": 0.00018973150033184933,
      "loss": 4.0726,
      "step": 55080
    },
    {
      "epoch": 0.0513527939977169,
      "grad_norm": 0.8485405445098877,
      "learning_rate": 0.00018972963600975417,
      "loss": 3.9201,
      "step": 55090
    },
    {
      "epoch": 0.051362115615796,
      "grad_norm": 0.8847519755363464,
      "learning_rate": 0.00018972777168765895,
      "loss": 3.9661,
      "step": 55100
    },
    {
      "epoch": 0.051371437233875085,
      "grad_norm": 1.199973225593567,
      "learning_rate": 0.00018972590736556375,
      "loss": 3.8932,
      "step": 55110
    },
    {
      "epoch": 0.05138075885195418,
      "grad_norm": 1.0714867115020752,
      "learning_rate": 0.00018972404304346853,
      "loss": 4.0865,
      "step": 55120
    },
    {
      "epoch": 0.05139008047003327,
      "grad_norm": 1.1966164112091064,
      "learning_rate": 0.00018972217872137334,
      "loss": 3.8771,
      "step": 55130
    },
    {
      "epoch": 0.05139940208811236,
      "grad_norm": 0.9614071846008301,
      "learning_rate": 0.00018972031439927814,
      "loss": 4.0134,
      "step": 55140
    },
    {
      "epoch": 0.051408723706191456,
      "grad_norm": 1.123769760131836,
      "learning_rate": 0.00018971845007718295,
      "loss": 3.953,
      "step": 55150
    },
    {
      "epoch": 0.051418045324270543,
      "grad_norm": 0.9615602493286133,
      "learning_rate": 0.00018971658575508776,
      "loss": 3.9266,
      "step": 55160
    },
    {
      "epoch": 0.05142736694234964,
      "grad_norm": 0.9775127172470093,
      "learning_rate": 0.00018971472143299254,
      "loss": 4.1235,
      "step": 55170
    },
    {
      "epoch": 0.05143668856042873,
      "grad_norm": 1.103771686553955,
      "learning_rate": 0.00018971285711089734,
      "loss": 3.8043,
      "step": 55180
    },
    {
      "epoch": 0.05144601017850782,
      "grad_norm": 1.256375789642334,
      "learning_rate": 0.00018971099278880215,
      "loss": 3.796,
      "step": 55190
    },
    {
      "epoch": 0.051455331796586914,
      "grad_norm": 1.1532806158065796,
      "learning_rate": 0.00018970912846670695,
      "loss": 3.78,
      "step": 55200
    },
    {
      "epoch": 0.051464653414666,
      "grad_norm": 1.1189459562301636,
      "learning_rate": 0.00018970726414461176,
      "loss": 4.02,
      "step": 55210
    },
    {
      "epoch": 0.051473975032745097,
      "grad_norm": 1.0737756490707397,
      "learning_rate": 0.00018970539982251654,
      "loss": 3.9468,
      "step": 55220
    },
    {
      "epoch": 0.05148329665082419,
      "grad_norm": 1.4397850036621094,
      "learning_rate": 0.00018970353550042135,
      "loss": 3.8633,
      "step": 55230
    },
    {
      "epoch": 0.05149261826890328,
      "grad_norm": 0.9748737215995789,
      "learning_rate": 0.00018970167117832615,
      "loss": 3.9341,
      "step": 55240
    },
    {
      "epoch": 0.05150193988698237,
      "grad_norm": 1.0393997430801392,
      "learning_rate": 0.00018969980685623093,
      "loss": 4.0677,
      "step": 55250
    },
    {
      "epoch": 0.05151126150506146,
      "grad_norm": 1.2668555974960327,
      "learning_rate": 0.00018969794253413576,
      "loss": 3.9964,
      "step": 55260
    },
    {
      "epoch": 0.051520583123140555,
      "grad_norm": 1.2101713418960571,
      "learning_rate": 0.00018969607821204054,
      "loss": 4.0175,
      "step": 55270
    },
    {
      "epoch": 0.05152990474121965,
      "grad_norm": 1.0190975666046143,
      "learning_rate": 0.00018969421388994535,
      "loss": 4.0985,
      "step": 55280
    },
    {
      "epoch": 0.05153922635929874,
      "grad_norm": 1.1052603721618652,
      "learning_rate": 0.00018969234956785016,
      "loss": 3.982,
      "step": 55290
    },
    {
      "epoch": 0.05154854797737783,
      "grad_norm": 1.0226303339004517,
      "learning_rate": 0.00018969048524575494,
      "loss": 3.8428,
      "step": 55300
    },
    {
      "epoch": 0.051557869595456926,
      "grad_norm": 1.01841139793396,
      "learning_rate": 0.00018968862092365974,
      "loss": 3.8827,
      "step": 55310
    },
    {
      "epoch": 0.051567191213536014,
      "grad_norm": 1.080174207687378,
      "learning_rate": 0.00018968675660156455,
      "loss": 3.9544,
      "step": 55320
    },
    {
      "epoch": 0.05157651283161511,
      "grad_norm": 1.1758859157562256,
      "learning_rate": 0.00018968489227946935,
      "loss": 3.9303,
      "step": 55330
    },
    {
      "epoch": 0.051585834449694196,
      "grad_norm": 0.8123343586921692,
      "learning_rate": 0.00018968302795737416,
      "loss": 4.1828,
      "step": 55340
    },
    {
      "epoch": 0.05159515606777329,
      "grad_norm": 1.1176072359085083,
      "learning_rate": 0.00018968116363527894,
      "loss": 3.8173,
      "step": 55350
    },
    {
      "epoch": 0.051604477685852385,
      "grad_norm": 0.9087138772010803,
      "learning_rate": 0.00018967929931318375,
      "loss": 4.0255,
      "step": 55360
    },
    {
      "epoch": 0.05161379930393147,
      "grad_norm": 1.0913914442062378,
      "learning_rate": 0.00018967743499108855,
      "loss": 4.0672,
      "step": 55370
    },
    {
      "epoch": 0.05162312092201057,
      "grad_norm": 0.9113671779632568,
      "learning_rate": 0.00018967557066899336,
      "loss": 4.1104,
      "step": 55380
    },
    {
      "epoch": 0.05163244254008966,
      "grad_norm": 0.9521259665489197,
      "learning_rate": 0.00018967370634689814,
      "loss": 4.036,
      "step": 55390
    },
    {
      "epoch": 0.05164176415816875,
      "grad_norm": 1.006975769996643,
      "learning_rate": 0.00018967184202480294,
      "loss": 4.0844,
      "step": 55400
    },
    {
      "epoch": 0.05165108577624784,
      "grad_norm": 1.077636957168579,
      "learning_rate": 0.00018966997770270775,
      "loss": 3.9,
      "step": 55410
    },
    {
      "epoch": 0.05166040739432693,
      "grad_norm": 0.8968271613121033,
      "learning_rate": 0.00018966811338061256,
      "loss": 4.0681,
      "step": 55420
    },
    {
      "epoch": 0.051669729012406025,
      "grad_norm": 1.0158917903900146,
      "learning_rate": 0.00018966624905851736,
      "loss": 4.0786,
      "step": 55430
    },
    {
      "epoch": 0.05167905063048512,
      "grad_norm": 1.0452200174331665,
      "learning_rate": 0.00018966438473642214,
      "loss": 3.9109,
      "step": 55440
    },
    {
      "epoch": 0.05168837224856421,
      "grad_norm": 1.166783332824707,
      "learning_rate": 0.00018966252041432695,
      "loss": 4.0464,
      "step": 55450
    },
    {
      "epoch": 0.0516976938666433,
      "grad_norm": 1.0987191200256348,
      "learning_rate": 0.00018966065609223176,
      "loss": 3.8871,
      "step": 55460
    },
    {
      "epoch": 0.05170701548472239,
      "grad_norm": 1.024820327758789,
      "learning_rate": 0.00018965879177013653,
      "loss": 4.1122,
      "step": 55470
    },
    {
      "epoch": 0.051716337102801484,
      "grad_norm": 1.0297611951828003,
      "learning_rate": 0.00018965692744804137,
      "loss": 4.0371,
      "step": 55480
    },
    {
      "epoch": 0.05172565872088058,
      "grad_norm": 1.1547331809997559,
      "learning_rate": 0.00018965506312594615,
      "loss": 4.1664,
      "step": 55490
    },
    {
      "epoch": 0.051734980338959666,
      "grad_norm": 0.8860303163528442,
      "learning_rate": 0.00018965319880385095,
      "loss": 4.0014,
      "step": 55500
    },
    {
      "epoch": 0.05174430195703876,
      "grad_norm": 1.0076916217803955,
      "learning_rate": 0.00018965133448175576,
      "loss": 3.9734,
      "step": 55510
    },
    {
      "epoch": 0.051753623575117855,
      "grad_norm": 0.9687073826789856,
      "learning_rate": 0.00018964947015966054,
      "loss": 3.9306,
      "step": 55520
    },
    {
      "epoch": 0.05176294519319694,
      "grad_norm": 1.04690420627594,
      "learning_rate": 0.00018964760583756535,
      "loss": 4.1332,
      "step": 55530
    },
    {
      "epoch": 0.05177226681127604,
      "grad_norm": 1.1201962232589722,
      "learning_rate": 0.00018964574151547015,
      "loss": 3.9626,
      "step": 55540
    },
    {
      "epoch": 0.051781588429355124,
      "grad_norm": 0.9643058776855469,
      "learning_rate": 0.00018964387719337496,
      "loss": 3.9308,
      "step": 55550
    },
    {
      "epoch": 0.05179091004743422,
      "grad_norm": 1.0486053228378296,
      "learning_rate": 0.00018964201287127976,
      "loss": 4.1247,
      "step": 55560
    },
    {
      "epoch": 0.05180023166551331,
      "grad_norm": 1.1787482500076294,
      "learning_rate": 0.00018964014854918454,
      "loss": 3.927,
      "step": 55570
    },
    {
      "epoch": 0.0518095532835924,
      "grad_norm": 1.0891882181167603,
      "learning_rate": 0.00018963828422708935,
      "loss": 4.1147,
      "step": 55580
    },
    {
      "epoch": 0.051818874901671495,
      "grad_norm": 1.2688390016555786,
      "learning_rate": 0.00018963641990499416,
      "loss": 3.9724,
      "step": 55590
    },
    {
      "epoch": 0.05182819651975058,
      "grad_norm": 0.9579351544380188,
      "learning_rate": 0.00018963455558289896,
      "loss": 3.9532,
      "step": 55600
    },
    {
      "epoch": 0.05183751813782968,
      "grad_norm": 1.064541220664978,
      "learning_rate": 0.00018963269126080377,
      "loss": 3.854,
      "step": 55610
    },
    {
      "epoch": 0.05184683975590877,
      "grad_norm": 1.0476597547531128,
      "learning_rate": 0.00018963082693870855,
      "loss": 3.9965,
      "step": 55620
    },
    {
      "epoch": 0.05185616137398786,
      "grad_norm": 1.0805991888046265,
      "learning_rate": 0.00018962896261661335,
      "loss": 3.9234,
      "step": 55630
    },
    {
      "epoch": 0.051865482992066954,
      "grad_norm": 1.2042195796966553,
      "learning_rate": 0.00018962709829451813,
      "loss": 3.7973,
      "step": 55640
    },
    {
      "epoch": 0.05187480461014605,
      "grad_norm": 1.123443841934204,
      "learning_rate": 0.00018962523397242297,
      "loss": 4.1013,
      "step": 55650
    },
    {
      "epoch": 0.051884126228225136,
      "grad_norm": 0.9837304949760437,
      "learning_rate": 0.00018962336965032777,
      "loss": 3.7588,
      "step": 55660
    },
    {
      "epoch": 0.05189344784630423,
      "grad_norm": 1.2190287113189697,
      "learning_rate": 0.00018962150532823255,
      "loss": 4.0557,
      "step": 55670
    },
    {
      "epoch": 0.05190276946438332,
      "grad_norm": 1.0386936664581299,
      "learning_rate": 0.00018961964100613736,
      "loss": 4.0449,
      "step": 55680
    },
    {
      "epoch": 0.05191209108246241,
      "grad_norm": 1.0508233308792114,
      "learning_rate": 0.00018961777668404214,
      "loss": 3.9446,
      "step": 55690
    },
    {
      "epoch": 0.05192141270054151,
      "grad_norm": 1.6272276639938354,
      "learning_rate": 0.00018961591236194694,
      "loss": 3.9693,
      "step": 55700
    },
    {
      "epoch": 0.051930734318620594,
      "grad_norm": 1.1001161336898804,
      "learning_rate": 0.00018961404803985178,
      "loss": 3.9728,
      "step": 55710
    },
    {
      "epoch": 0.05194005593669969,
      "grad_norm": 1.5419069528579712,
      "learning_rate": 0.00018961218371775656,
      "loss": 3.954,
      "step": 55720
    },
    {
      "epoch": 0.05194937755477878,
      "grad_norm": 1.4434560537338257,
      "learning_rate": 0.00018961031939566136,
      "loss": 4.1373,
      "step": 55730
    },
    {
      "epoch": 0.05195869917285787,
      "grad_norm": 1.100601315498352,
      "learning_rate": 0.00018960845507356614,
      "loss": 4.097,
      "step": 55740
    },
    {
      "epoch": 0.051968020790936965,
      "grad_norm": 1.225663423538208,
      "learning_rate": 0.00018960659075147095,
      "loss": 4.0183,
      "step": 55750
    },
    {
      "epoch": 0.05197734240901605,
      "grad_norm": 1.0048164129257202,
      "learning_rate": 0.00018960472642937575,
      "loss": 3.8795,
      "step": 55760
    },
    {
      "epoch": 0.05198666402709515,
      "grad_norm": 1.0825837850570679,
      "learning_rate": 0.00018960286210728056,
      "loss": 4.0097,
      "step": 55770
    },
    {
      "epoch": 0.05199598564517424,
      "grad_norm": 0.9379742741584778,
      "learning_rate": 0.00018960099778518537,
      "loss": 3.9465,
      "step": 55780
    },
    {
      "epoch": 0.05200530726325333,
      "grad_norm": 1.012245535850525,
      "learning_rate": 0.00018959913346309015,
      "loss": 3.9716,
      "step": 55790
    },
    {
      "epoch": 0.052014628881332424,
      "grad_norm": 1.217566967010498,
      "learning_rate": 0.00018959726914099495,
      "loss": 4.0511,
      "step": 55800
    },
    {
      "epoch": 0.05202395049941151,
      "grad_norm": 0.9634447693824768,
      "learning_rate": 0.00018959540481889976,
      "loss": 4.1932,
      "step": 55810
    },
    {
      "epoch": 0.052033272117490606,
      "grad_norm": 1.073732614517212,
      "learning_rate": 0.00018959354049680457,
      "loss": 4.0142,
      "step": 55820
    },
    {
      "epoch": 0.0520425937355697,
      "grad_norm": 0.901394248008728,
      "learning_rate": 0.00018959167617470937,
      "loss": 3.9674,
      "step": 55830
    },
    {
      "epoch": 0.05205191535364879,
      "grad_norm": 1.3268890380859375,
      "learning_rate": 0.00018958981185261415,
      "loss": 4.0054,
      "step": 55840
    },
    {
      "epoch": 0.05206123697172788,
      "grad_norm": 0.9243728518486023,
      "learning_rate": 0.00018958794753051896,
      "loss": 3.9206,
      "step": 55850
    },
    {
      "epoch": 0.05207055858980698,
      "grad_norm": 0.8899453282356262,
      "learning_rate": 0.00018958608320842376,
      "loss": 4.0074,
      "step": 55860
    },
    {
      "epoch": 0.052079880207886065,
      "grad_norm": 1.0824888944625854,
      "learning_rate": 0.00018958421888632857,
      "loss": 3.8303,
      "step": 55870
    },
    {
      "epoch": 0.05208920182596516,
      "grad_norm": 1.0081292390823364,
      "learning_rate": 0.00018958235456423338,
      "loss": 3.9319,
      "step": 55880
    },
    {
      "epoch": 0.05209852344404425,
      "grad_norm": 1.0975992679595947,
      "learning_rate": 0.00018958049024213816,
      "loss": 4.0505,
      "step": 55890
    },
    {
      "epoch": 0.05210784506212334,
      "grad_norm": 1.0519739389419556,
      "learning_rate": 0.00018957862592004296,
      "loss": 3.9066,
      "step": 55900
    },
    {
      "epoch": 0.052117166680202436,
      "grad_norm": 1.1762065887451172,
      "learning_rate": 0.00018957676159794777,
      "loss": 4.0696,
      "step": 55910
    },
    {
      "epoch": 0.05212648829828152,
      "grad_norm": 1.1251134872436523,
      "learning_rate": 0.00018957489727585255,
      "loss": 3.9369,
      "step": 55920
    },
    {
      "epoch": 0.05213580991636062,
      "grad_norm": 1.2231569290161133,
      "learning_rate": 0.00018957303295375738,
      "loss": 3.9348,
      "step": 55930
    },
    {
      "epoch": 0.052145131534439705,
      "grad_norm": 1.123724341392517,
      "learning_rate": 0.00018957116863166216,
      "loss": 4.1196,
      "step": 55940
    },
    {
      "epoch": 0.0521544531525188,
      "grad_norm": 0.8773805499076843,
      "learning_rate": 0.00018956930430956697,
      "loss": 4.1575,
      "step": 55950
    },
    {
      "epoch": 0.052163774770597894,
      "grad_norm": 1.0800045728683472,
      "learning_rate": 0.00018956743998747177,
      "loss": 3.9739,
      "step": 55960
    },
    {
      "epoch": 0.05217309638867698,
      "grad_norm": 0.9064137935638428,
      "learning_rate": 0.00018956557566537655,
      "loss": 4.0818,
      "step": 55970
    },
    {
      "epoch": 0.052182418006756076,
      "grad_norm": 1.0205001831054688,
      "learning_rate": 0.00018956371134328136,
      "loss": 3.9068,
      "step": 55980
    },
    {
      "epoch": 0.05219173962483517,
      "grad_norm": 1.0986260175704956,
      "learning_rate": 0.00018956184702118616,
      "loss": 3.7722,
      "step": 55990
    },
    {
      "epoch": 0.05220106124291426,
      "grad_norm": 1.1468515396118164,
      "learning_rate": 0.00018955998269909097,
      "loss": 3.8984,
      "step": 56000
    },
    {
      "epoch": 0.05221038286099335,
      "grad_norm": 1.0893120765686035,
      "learning_rate": 0.00018955811837699578,
      "loss": 3.9951,
      "step": 56010
    },
    {
      "epoch": 0.05221970447907244,
      "grad_norm": 1.36849045753479,
      "learning_rate": 0.00018955625405490056,
      "loss": 4.0354,
      "step": 56020
    },
    {
      "epoch": 0.052229026097151535,
      "grad_norm": 0.8006148934364319,
      "learning_rate": 0.00018955438973280536,
      "loss": 4.0383,
      "step": 56030
    },
    {
      "epoch": 0.05223834771523063,
      "grad_norm": 1.0368760824203491,
      "learning_rate": 0.00018955252541071017,
      "loss": 4.0351,
      "step": 56040
    },
    {
      "epoch": 0.05224766933330972,
      "grad_norm": 0.9724010229110718,
      "learning_rate": 0.00018955066108861498,
      "loss": 4.0902,
      "step": 56050
    },
    {
      "epoch": 0.05225699095138881,
      "grad_norm": 1.0834343433380127,
      "learning_rate": 0.00018954879676651978,
      "loss": 3.8975,
      "step": 56060
    },
    {
      "epoch": 0.052266312569467906,
      "grad_norm": 1.2845942974090576,
      "learning_rate": 0.00018954693244442456,
      "loss": 4.1117,
      "step": 56070
    },
    {
      "epoch": 0.05227563418754699,
      "grad_norm": 1.130859136581421,
      "learning_rate": 0.00018954506812232937,
      "loss": 4.0049,
      "step": 56080
    },
    {
      "epoch": 0.05228495580562609,
      "grad_norm": 1.071157693862915,
      "learning_rate": 0.00018954320380023415,
      "loss": 3.8992,
      "step": 56090
    },
    {
      "epoch": 0.052294277423705175,
      "grad_norm": 1.1438372135162354,
      "learning_rate": 0.00018954133947813898,
      "loss": 4.0166,
      "step": 56100
    },
    {
      "epoch": 0.05230359904178427,
      "grad_norm": 1.0679093599319458,
      "learning_rate": 0.00018953947515604379,
      "loss": 3.8835,
      "step": 56110
    },
    {
      "epoch": 0.052312920659863364,
      "grad_norm": 1.0619947910308838,
      "learning_rate": 0.00018953761083394857,
      "loss": 3.7199,
      "step": 56120
    },
    {
      "epoch": 0.05232224227794245,
      "grad_norm": 1.1541805267333984,
      "learning_rate": 0.00018953574651185337,
      "loss": 3.8684,
      "step": 56130
    },
    {
      "epoch": 0.052331563896021546,
      "grad_norm": 1.081958532333374,
      "learning_rate": 0.00018953388218975815,
      "loss": 4.084,
      "step": 56140
    },
    {
      "epoch": 0.052340885514100634,
      "grad_norm": 1.170636534690857,
      "learning_rate": 0.00018953201786766296,
      "loss": 3.6938,
      "step": 56150
    },
    {
      "epoch": 0.05235020713217973,
      "grad_norm": 1.133048415184021,
      "learning_rate": 0.0001895301535455678,
      "loss": 3.9595,
      "step": 56160
    },
    {
      "epoch": 0.05235952875025882,
      "grad_norm": 1.0793505907058716,
      "learning_rate": 0.00018952828922347257,
      "loss": 4.05,
      "step": 56170
    },
    {
      "epoch": 0.05236885036833791,
      "grad_norm": 1.1938989162445068,
      "learning_rate": 0.00018952642490137738,
      "loss": 3.7765,
      "step": 56180
    },
    {
      "epoch": 0.052378171986417005,
      "grad_norm": 0.9472837448120117,
      "learning_rate": 0.00018952456057928216,
      "loss": 4.0045,
      "step": 56190
    },
    {
      "epoch": 0.0523874936044961,
      "grad_norm": 1.1353862285614014,
      "learning_rate": 0.00018952269625718696,
      "loss": 3.7689,
      "step": 56200
    },
    {
      "epoch": 0.05239681522257519,
      "grad_norm": 0.8400262594223022,
      "learning_rate": 0.00018952083193509177,
      "loss": 4.0135,
      "step": 56210
    },
    {
      "epoch": 0.05240613684065428,
      "grad_norm": 1.283551573753357,
      "learning_rate": 0.00018951896761299657,
      "loss": 3.9096,
      "step": 56220
    },
    {
      "epoch": 0.05241545845873337,
      "grad_norm": 1.0217962265014648,
      "learning_rate": 0.00018951710329090138,
      "loss": 3.9409,
      "step": 56230
    },
    {
      "epoch": 0.05242478007681246,
      "grad_norm": 0.926978588104248,
      "learning_rate": 0.00018951523896880616,
      "loss": 3.9463,
      "step": 56240
    },
    {
      "epoch": 0.05243410169489156,
      "grad_norm": 1.0759837627410889,
      "learning_rate": 0.00018951337464671097,
      "loss": 3.9119,
      "step": 56250
    },
    {
      "epoch": 0.052443423312970645,
      "grad_norm": 1.1623432636260986,
      "learning_rate": 0.00018951151032461577,
      "loss": 4.0509,
      "step": 56260
    },
    {
      "epoch": 0.05245274493104974,
      "grad_norm": 0.972091019153595,
      "learning_rate": 0.00018950964600252058,
      "loss": 4.0058,
      "step": 56270
    },
    {
      "epoch": 0.05246206654912883,
      "grad_norm": 0.9271437525749207,
      "learning_rate": 0.00018950778168042538,
      "loss": 4.0973,
      "step": 56280
    },
    {
      "epoch": 0.05247138816720792,
      "grad_norm": 1.0130140781402588,
      "learning_rate": 0.00018950591735833016,
      "loss": 3.9422,
      "step": 56290
    },
    {
      "epoch": 0.052480709785287016,
      "grad_norm": 1.0839310884475708,
      "learning_rate": 0.00018950405303623497,
      "loss": 4.0688,
      "step": 56300
    },
    {
      "epoch": 0.052490031403366104,
      "grad_norm": 0.9599992036819458,
      "learning_rate": 0.00018950218871413978,
      "loss": 3.8884,
      "step": 56310
    },
    {
      "epoch": 0.0524993530214452,
      "grad_norm": 1.0988277196884155,
      "learning_rate": 0.00018950032439204458,
      "loss": 4.0731,
      "step": 56320
    },
    {
      "epoch": 0.05250867463952429,
      "grad_norm": 0.9701310992240906,
      "learning_rate": 0.0001894984600699494,
      "loss": 3.8573,
      "step": 56330
    },
    {
      "epoch": 0.05251799625760338,
      "grad_norm": 1.1428641080856323,
      "learning_rate": 0.00018949659574785417,
      "loss": 4.0107,
      "step": 56340
    },
    {
      "epoch": 0.052527317875682475,
      "grad_norm": 1.301889419555664,
      "learning_rate": 0.00018949473142575897,
      "loss": 3.8176,
      "step": 56350
    },
    {
      "epoch": 0.05253663949376156,
      "grad_norm": 1.0274263620376587,
      "learning_rate": 0.00018949286710366378,
      "loss": 3.9089,
      "step": 56360
    },
    {
      "epoch": 0.05254596111184066,
      "grad_norm": 1.179519772529602,
      "learning_rate": 0.00018949100278156856,
      "loss": 3.7895,
      "step": 56370
    },
    {
      "epoch": 0.05255528272991975,
      "grad_norm": 1.0534932613372803,
      "learning_rate": 0.0001894891384594734,
      "loss": 4.115,
      "step": 56380
    },
    {
      "epoch": 0.05256460434799884,
      "grad_norm": 1.1056846380233765,
      "learning_rate": 0.00018948727413737817,
      "loss": 3.9864,
      "step": 56390
    },
    {
      "epoch": 0.05257392596607793,
      "grad_norm": 1.076305627822876,
      "learning_rate": 0.00018948540981528298,
      "loss": 3.7388,
      "step": 56400
    },
    {
      "epoch": 0.05258324758415703,
      "grad_norm": 1.2478545904159546,
      "learning_rate": 0.00018948354549318779,
      "loss": 3.9655,
      "step": 56410
    },
    {
      "epoch": 0.052592569202236115,
      "grad_norm": 1.0869758129119873,
      "learning_rate": 0.00018948168117109256,
      "loss": 3.6981,
      "step": 56420
    },
    {
      "epoch": 0.05260189082031521,
      "grad_norm": 1.127004861831665,
      "learning_rate": 0.00018947981684899737,
      "loss": 3.9463,
      "step": 56430
    },
    {
      "epoch": 0.0526112124383943,
      "grad_norm": 1.005271553993225,
      "learning_rate": 0.00018947795252690218,
      "loss": 4.0512,
      "step": 56440
    },
    {
      "epoch": 0.05262053405647339,
      "grad_norm": 1.0045711994171143,
      "learning_rate": 0.00018947608820480698,
      "loss": 4.068,
      "step": 56450
    },
    {
      "epoch": 0.052629855674552486,
      "grad_norm": 1.2300752401351929,
      "learning_rate": 0.0001894742238827118,
      "loss": 3.8629,
      "step": 56460
    },
    {
      "epoch": 0.052639177292631574,
      "grad_norm": 1.1481047868728638,
      "learning_rate": 0.00018947235956061657,
      "loss": 3.9758,
      "step": 56470
    },
    {
      "epoch": 0.05264849891071067,
      "grad_norm": 0.9325785636901855,
      "learning_rate": 0.00018947049523852138,
      "loss": 3.816,
      "step": 56480
    },
    {
      "epoch": 0.052657820528789756,
      "grad_norm": 0.9650364518165588,
      "learning_rate": 0.00018946863091642618,
      "loss": 4.0715,
      "step": 56490
    },
    {
      "epoch": 0.05266714214686885,
      "grad_norm": 1.0405371189117432,
      "learning_rate": 0.000189466766594331,
      "loss": 3.9611,
      "step": 56500
    },
    {
      "epoch": 0.052676463764947945,
      "grad_norm": 1.0047518014907837,
      "learning_rate": 0.0001894649022722358,
      "loss": 3.9151,
      "step": 56510
    },
    {
      "epoch": 0.05268578538302703,
      "grad_norm": 1.2883150577545166,
      "learning_rate": 0.00018946303795014057,
      "loss": 4.06,
      "step": 56520
    },
    {
      "epoch": 0.05269510700110613,
      "grad_norm": 1.1595332622528076,
      "learning_rate": 0.00018946117362804538,
      "loss": 4.0171,
      "step": 56530
    },
    {
      "epoch": 0.05270442861918522,
      "grad_norm": 1.0255898237228394,
      "learning_rate": 0.00018945930930595016,
      "loss": 4.0442,
      "step": 56540
    },
    {
      "epoch": 0.05271375023726431,
      "grad_norm": 1.0874170064926147,
      "learning_rate": 0.000189457444983855,
      "loss": 3.9049,
      "step": 56550
    },
    {
      "epoch": 0.052723071855343404,
      "grad_norm": 1.1549774408340454,
      "learning_rate": 0.0001894555806617598,
      "loss": 3.8957,
      "step": 56560
    },
    {
      "epoch": 0.05273239347342249,
      "grad_norm": 1.3316925764083862,
      "learning_rate": 0.00018945371633966458,
      "loss": 3.7772,
      "step": 56570
    },
    {
      "epoch": 0.052741715091501586,
      "grad_norm": 0.9972207546234131,
      "learning_rate": 0.00018945185201756938,
      "loss": 3.8989,
      "step": 56580
    },
    {
      "epoch": 0.05275103670958068,
      "grad_norm": 1.1227223873138428,
      "learning_rate": 0.00018944998769547416,
      "loss": 3.9517,
      "step": 56590
    },
    {
      "epoch": 0.05276035832765977,
      "grad_norm": 0.9834230542182922,
      "learning_rate": 0.00018944812337337897,
      "loss": 4.114,
      "step": 56600
    },
    {
      "epoch": 0.05276967994573886,
      "grad_norm": 1.2145326137542725,
      "learning_rate": 0.0001894462590512838,
      "loss": 3.9863,
      "step": 56610
    },
    {
      "epoch": 0.05277900156381795,
      "grad_norm": 1.111264944076538,
      "learning_rate": 0.00018944439472918858,
      "loss": 4.1583,
      "step": 56620
    },
    {
      "epoch": 0.052788323181897044,
      "grad_norm": 1.2165443897247314,
      "learning_rate": 0.0001894425304070934,
      "loss": 3.7719,
      "step": 56630
    },
    {
      "epoch": 0.05279764479997614,
      "grad_norm": 1.0412286520004272,
      "learning_rate": 0.00018944066608499817,
      "loss": 3.9074,
      "step": 56640
    },
    {
      "epoch": 0.052806966418055226,
      "grad_norm": 0.9434898495674133,
      "learning_rate": 0.00018943880176290297,
      "loss": 3.8519,
      "step": 56650
    },
    {
      "epoch": 0.05281628803613432,
      "grad_norm": 1.1082051992416382,
      "learning_rate": 0.00018943693744080778,
      "loss": 3.9412,
      "step": 56660
    },
    {
      "epoch": 0.052825609654213415,
      "grad_norm": 1.3576507568359375,
      "learning_rate": 0.0001894350731187126,
      "loss": 4.0857,
      "step": 56670
    },
    {
      "epoch": 0.0528349312722925,
      "grad_norm": 1.0196950435638428,
      "learning_rate": 0.0001894332087966174,
      "loss": 3.976,
      "step": 56680
    },
    {
      "epoch": 0.0528442528903716,
      "grad_norm": 0.9572871327400208,
      "learning_rate": 0.00018943134447452217,
      "loss": 3.9407,
      "step": 56690
    },
    {
      "epoch": 0.052853574508450685,
      "grad_norm": 1.09345543384552,
      "learning_rate": 0.00018942948015242698,
      "loss": 3.9256,
      "step": 56700
    },
    {
      "epoch": 0.05286289612652978,
      "grad_norm": 1.06854248046875,
      "learning_rate": 0.00018942761583033179,
      "loss": 3.7982,
      "step": 56710
    },
    {
      "epoch": 0.052872217744608874,
      "grad_norm": 1.0312362909317017,
      "learning_rate": 0.0001894257515082366,
      "loss": 4.0842,
      "step": 56720
    },
    {
      "epoch": 0.05288153936268796,
      "grad_norm": 1.2054280042648315,
      "learning_rate": 0.0001894238871861414,
      "loss": 3.7019,
      "step": 56730
    },
    {
      "epoch": 0.052890860980767056,
      "grad_norm": 1.0828039646148682,
      "learning_rate": 0.00018942202286404618,
      "loss": 3.9907,
      "step": 56740
    },
    {
      "epoch": 0.05290018259884615,
      "grad_norm": 0.836199164390564,
      "learning_rate": 0.00018942015854195098,
      "loss": 4.0119,
      "step": 56750
    },
    {
      "epoch": 0.05290950421692524,
      "grad_norm": 1.1513158082962036,
      "learning_rate": 0.00018941829421985576,
      "loss": 3.9159,
      "step": 56760
    },
    {
      "epoch": 0.05291882583500433,
      "grad_norm": 1.2025877237319946,
      "learning_rate": 0.0001894164298977606,
      "loss": 4.1213,
      "step": 56770
    },
    {
      "epoch": 0.05292814745308342,
      "grad_norm": 1.0781219005584717,
      "learning_rate": 0.0001894145655756654,
      "loss": 3.8242,
      "step": 56780
    },
    {
      "epoch": 0.052937469071162514,
      "grad_norm": 1.2209144830703735,
      "learning_rate": 0.00018941270125357018,
      "loss": 3.9251,
      "step": 56790
    },
    {
      "epoch": 0.05294679068924161,
      "grad_norm": 1.1745824813842773,
      "learning_rate": 0.000189410836931475,
      "loss": 3.9164,
      "step": 56800
    },
    {
      "epoch": 0.052956112307320696,
      "grad_norm": 1.0027670860290527,
      "learning_rate": 0.00018940897260937977,
      "loss": 3.9873,
      "step": 56810
    },
    {
      "epoch": 0.05296543392539979,
      "grad_norm": 1.1347339153289795,
      "learning_rate": 0.00018940710828728457,
      "loss": 3.7208,
      "step": 56820
    },
    {
      "epoch": 0.05297475554347888,
      "grad_norm": 1.004072904586792,
      "learning_rate": 0.0001894052439651894,
      "loss": 3.9778,
      "step": 56830
    },
    {
      "epoch": 0.05298407716155797,
      "grad_norm": 1.098321557044983,
      "learning_rate": 0.00018940337964309419,
      "loss": 3.7422,
      "step": 56840
    },
    {
      "epoch": 0.05299339877963707,
      "grad_norm": 0.9632648825645447,
      "learning_rate": 0.000189401515320999,
      "loss": 3.9808,
      "step": 56850
    },
    {
      "epoch": 0.053002720397716155,
      "grad_norm": 1.206466794013977,
      "learning_rate": 0.00018939965099890377,
      "loss": 4.0783,
      "step": 56860
    },
    {
      "epoch": 0.05301204201579525,
      "grad_norm": 1.3493552207946777,
      "learning_rate": 0.00018939778667680858,
      "loss": 4.0589,
      "step": 56870
    },
    {
      "epoch": 0.053021363633874344,
      "grad_norm": 1.063085675239563,
      "learning_rate": 0.00018939592235471338,
      "loss": 3.9173,
      "step": 56880
    },
    {
      "epoch": 0.05303068525195343,
      "grad_norm": 1.0631555318832397,
      "learning_rate": 0.0001893940580326182,
      "loss": 3.8173,
      "step": 56890
    },
    {
      "epoch": 0.053040006870032526,
      "grad_norm": 1.1657577753067017,
      "learning_rate": 0.000189392193710523,
      "loss": 4.0865,
      "step": 56900
    },
    {
      "epoch": 0.05304932848811161,
      "grad_norm": 1.174575686454773,
      "learning_rate": 0.00018939032938842778,
      "loss": 3.8616,
      "step": 56910
    },
    {
      "epoch": 0.05305865010619071,
      "grad_norm": 1.080888271331787,
      "learning_rate": 0.00018938846506633258,
      "loss": 4.1676,
      "step": 56920
    },
    {
      "epoch": 0.0530679717242698,
      "grad_norm": 1.0456974506378174,
      "learning_rate": 0.0001893866007442374,
      "loss": 3.9737,
      "step": 56930
    },
    {
      "epoch": 0.05307729334234889,
      "grad_norm": 1.0055514574050903,
      "learning_rate": 0.0001893847364221422,
      "loss": 3.9364,
      "step": 56940
    },
    {
      "epoch": 0.053086614960427984,
      "grad_norm": 1.1929343938827515,
      "learning_rate": 0.000189382872100047,
      "loss": 3.987,
      "step": 56950
    },
    {
      "epoch": 0.05309593657850707,
      "grad_norm": 1.1603786945343018,
      "learning_rate": 0.00018938100777795178,
      "loss": 3.8125,
      "step": 56960
    },
    {
      "epoch": 0.053105258196586166,
      "grad_norm": 1.0566283464431763,
      "learning_rate": 0.00018937914345585659,
      "loss": 3.6943,
      "step": 56970
    },
    {
      "epoch": 0.05311457981466526,
      "grad_norm": 1.0465292930603027,
      "learning_rate": 0.0001893772791337614,
      "loss": 3.982,
      "step": 56980
    },
    {
      "epoch": 0.05312390143274435,
      "grad_norm": 1.0484577417373657,
      "learning_rate": 0.00018937541481166617,
      "loss": 4.1095,
      "step": 56990
    },
    {
      "epoch": 0.05313322305082344,
      "grad_norm": 1.1667430400848389,
      "learning_rate": 0.000189373550489571,
      "loss": 3.9383,
      "step": 57000
    },
    {
      "epoch": 0.05314254466890254,
      "grad_norm": 0.9908518195152283,
      "learning_rate": 0.00018937168616747578,
      "loss": 3.9907,
      "step": 57010
    },
    {
      "epoch": 0.053151866286981625,
      "grad_norm": 1.2114880084991455,
      "learning_rate": 0.0001893698218453806,
      "loss": 4.1093,
      "step": 57020
    },
    {
      "epoch": 0.05316118790506072,
      "grad_norm": 1.2357362508773804,
      "learning_rate": 0.0001893679575232854,
      "loss": 4.1016,
      "step": 57030
    },
    {
      "epoch": 0.05317050952313981,
      "grad_norm": 1.0313705205917358,
      "learning_rate": 0.00018936609320119018,
      "loss": 3.8873,
      "step": 57040
    },
    {
      "epoch": 0.0531798311412189,
      "grad_norm": 1.2065749168395996,
      "learning_rate": 0.00018936422887909498,
      "loss": 3.9547,
      "step": 57050
    },
    {
      "epoch": 0.053189152759297996,
      "grad_norm": 1.0051875114440918,
      "learning_rate": 0.0001893623645569998,
      "loss": 3.8584,
      "step": 57060
    },
    {
      "epoch": 0.053198474377377084,
      "grad_norm": 1.0212798118591309,
      "learning_rate": 0.0001893605002349046,
      "loss": 4.1354,
      "step": 57070
    },
    {
      "epoch": 0.05320779599545618,
      "grad_norm": 1.1503729820251465,
      "learning_rate": 0.0001893586359128094,
      "loss": 3.9019,
      "step": 57080
    },
    {
      "epoch": 0.05321711761353527,
      "grad_norm": 1.2297279834747314,
      "learning_rate": 0.00018935677159071418,
      "loss": 3.9064,
      "step": 57090
    },
    {
      "epoch": 0.05322643923161436,
      "grad_norm": 0.9375919699668884,
      "learning_rate": 0.000189354907268619,
      "loss": 3.9277,
      "step": 57100
    },
    {
      "epoch": 0.053235760849693455,
      "grad_norm": 1.2114014625549316,
      "learning_rate": 0.0001893530429465238,
      "loss": 3.9302,
      "step": 57110
    },
    {
      "epoch": 0.05324508246777254,
      "grad_norm": 1.320792317390442,
      "learning_rate": 0.0001893511786244286,
      "loss": 4.0191,
      "step": 57120
    },
    {
      "epoch": 0.05325440408585164,
      "grad_norm": 1.2090290784835815,
      "learning_rate": 0.0001893493143023334,
      "loss": 3.796,
      "step": 57130
    },
    {
      "epoch": 0.05326372570393073,
      "grad_norm": 1.1650747060775757,
      "learning_rate": 0.00018934744998023819,
      "loss": 4.0821,
      "step": 57140
    },
    {
      "epoch": 0.05327304732200982,
      "grad_norm": 1.113503336906433,
      "learning_rate": 0.000189345585658143,
      "loss": 3.8678,
      "step": 57150
    },
    {
      "epoch": 0.05328236894008891,
      "grad_norm": 1.4728279113769531,
      "learning_rate": 0.0001893437213360478,
      "loss": 4.0189,
      "step": 57160
    },
    {
      "epoch": 0.053291690558168,
      "grad_norm": 1.1883323192596436,
      "learning_rate": 0.0001893418570139526,
      "loss": 3.9644,
      "step": 57170
    },
    {
      "epoch": 0.053301012176247095,
      "grad_norm": 1.10540771484375,
      "learning_rate": 0.0001893399926918574,
      "loss": 4.0774,
      "step": 57180
    },
    {
      "epoch": 0.05331033379432619,
      "grad_norm": 1.056796669960022,
      "learning_rate": 0.0001893381283697622,
      "loss": 3.8766,
      "step": 57190
    },
    {
      "epoch": 0.05331965541240528,
      "grad_norm": 0.9342288970947266,
      "learning_rate": 0.000189336264047667,
      "loss": 4.0401,
      "step": 57200
    },
    {
      "epoch": 0.05332897703048437,
      "grad_norm": 1.0869624614715576,
      "learning_rate": 0.00018933439972557178,
      "loss": 4.0321,
      "step": 57210
    },
    {
      "epoch": 0.053338298648563466,
      "grad_norm": 0.8290222883224487,
      "learning_rate": 0.0001893325354034766,
      "loss": 3.8947,
      "step": 57220
    },
    {
      "epoch": 0.053347620266642554,
      "grad_norm": 1.133899450302124,
      "learning_rate": 0.00018933067108138141,
      "loss": 4.023,
      "step": 57230
    },
    {
      "epoch": 0.05335694188472165,
      "grad_norm": 1.1218242645263672,
      "learning_rate": 0.0001893288067592862,
      "loss": 3.9713,
      "step": 57240
    },
    {
      "epoch": 0.053366263502800736,
      "grad_norm": 2.3854193687438965,
      "learning_rate": 0.000189326942437191,
      "loss": 3.9068,
      "step": 57250
    },
    {
      "epoch": 0.05337558512087983,
      "grad_norm": 1.2509095668792725,
      "learning_rate": 0.00018932507811509578,
      "loss": 3.9349,
      "step": 57260
    },
    {
      "epoch": 0.053384906738958925,
      "grad_norm": 1.088481068611145,
      "learning_rate": 0.00018932321379300059,
      "loss": 4.1667,
      "step": 57270
    },
    {
      "epoch": 0.05339422835703801,
      "grad_norm": 0.9394651651382446,
      "learning_rate": 0.00018932134947090542,
      "loss": 3.9757,
      "step": 57280
    },
    {
      "epoch": 0.05340354997511711,
      "grad_norm": 0.8779240846633911,
      "learning_rate": 0.0001893194851488102,
      "loss": 3.9598,
      "step": 57290
    },
    {
      "epoch": 0.053412871593196194,
      "grad_norm": 1.1024281978607178,
      "learning_rate": 0.000189317620826715,
      "loss": 3.7727,
      "step": 57300
    },
    {
      "epoch": 0.05342219321127529,
      "grad_norm": 1.1707584857940674,
      "learning_rate": 0.00018931575650461978,
      "loss": 4.0935,
      "step": 57310
    },
    {
      "epoch": 0.05343151482935438,
      "grad_norm": 0.9015759825706482,
      "learning_rate": 0.0001893138921825246,
      "loss": 3.891,
      "step": 57320
    },
    {
      "epoch": 0.05344083644743347,
      "grad_norm": 1.0795044898986816,
      "learning_rate": 0.0001893120278604294,
      "loss": 3.9564,
      "step": 57330
    },
    {
      "epoch": 0.053450158065512565,
      "grad_norm": 1.0084275007247925,
      "learning_rate": 0.0001893101635383342,
      "loss": 3.9006,
      "step": 57340
    },
    {
      "epoch": 0.05345947968359166,
      "grad_norm": 1.3166238069534302,
      "learning_rate": 0.000189308299216239,
      "loss": 4.035,
      "step": 57350
    },
    {
      "epoch": 0.05346880130167075,
      "grad_norm": 1.09453284740448,
      "learning_rate": 0.0001893064348941438,
      "loss": 3.9879,
      "step": 57360
    },
    {
      "epoch": 0.05347812291974984,
      "grad_norm": 1.0508408546447754,
      "learning_rate": 0.0001893045705720486,
      "loss": 3.9409,
      "step": 57370
    },
    {
      "epoch": 0.05348744453782893,
      "grad_norm": 1.1368176937103271,
      "learning_rate": 0.0001893027062499534,
      "loss": 3.9915,
      "step": 57380
    },
    {
      "epoch": 0.053496766155908024,
      "grad_norm": 1.0866962671279907,
      "learning_rate": 0.0001893008419278582,
      "loss": 3.7397,
      "step": 57390
    },
    {
      "epoch": 0.05350608777398712,
      "grad_norm": 1.0517877340316772,
      "learning_rate": 0.00018929897760576301,
      "loss": 3.9571,
      "step": 57400
    },
    {
      "epoch": 0.053515409392066206,
      "grad_norm": 1.039415717124939,
      "learning_rate": 0.0001892971132836678,
      "loss": 4.0152,
      "step": 57410
    },
    {
      "epoch": 0.0535247310101453,
      "grad_norm": 1.1587556600570679,
      "learning_rate": 0.0001892952489615726,
      "loss": 4.1746,
      "step": 57420
    },
    {
      "epoch": 0.053534052628224395,
      "grad_norm": 0.9320312142372131,
      "learning_rate": 0.0001892933846394774,
      "loss": 3.9946,
      "step": 57430
    },
    {
      "epoch": 0.05354337424630348,
      "grad_norm": 1.332412600517273,
      "learning_rate": 0.00018929152031738218,
      "loss": 4.1823,
      "step": 57440
    },
    {
      "epoch": 0.05355269586438258,
      "grad_norm": 0.8230718374252319,
      "learning_rate": 0.00018928965599528702,
      "loss": 3.9729,
      "step": 57450
    },
    {
      "epoch": 0.053562017482461664,
      "grad_norm": 1.2432594299316406,
      "learning_rate": 0.0001892877916731918,
      "loss": 3.5985,
      "step": 57460
    },
    {
      "epoch": 0.05357133910054076,
      "grad_norm": 1.226100206375122,
      "learning_rate": 0.0001892859273510966,
      "loss": 4.1545,
      "step": 57470
    },
    {
      "epoch": 0.05358066071861985,
      "grad_norm": 0.9827348589897156,
      "learning_rate": 0.0001892840630290014,
      "loss": 3.944,
      "step": 57480
    },
    {
      "epoch": 0.05358998233669894,
      "grad_norm": 0.9951417446136475,
      "learning_rate": 0.0001892821987069062,
      "loss": 3.9675,
      "step": 57490
    },
    {
      "epoch": 0.053599303954778035,
      "grad_norm": 1.1013599634170532,
      "learning_rate": 0.000189280334384811,
      "loss": 4.0564,
      "step": 57500
    },
    {
      "epoch": 0.05360862557285712,
      "grad_norm": 0.9789824485778809,
      "learning_rate": 0.0001892784700627158,
      "loss": 3.9492,
      "step": 57510
    },
    {
      "epoch": 0.05361794719093622,
      "grad_norm": 1.099979043006897,
      "learning_rate": 0.0001892766057406206,
      "loss": 3.9192,
      "step": 57520
    },
    {
      "epoch": 0.05362726880901531,
      "grad_norm": 1.2525718212127686,
      "learning_rate": 0.00018927474141852541,
      "loss": 3.8325,
      "step": 57530
    },
    {
      "epoch": 0.0536365904270944,
      "grad_norm": 1.0021698474884033,
      "learning_rate": 0.0001892728770964302,
      "loss": 4.0195,
      "step": 57540
    },
    {
      "epoch": 0.053645912045173494,
      "grad_norm": 1.202752709388733,
      "learning_rate": 0.000189271012774335,
      "loss": 3.9615,
      "step": 57550
    },
    {
      "epoch": 0.05365523366325259,
      "grad_norm": 1.0957101583480835,
      "learning_rate": 0.0001892691484522398,
      "loss": 3.9358,
      "step": 57560
    },
    {
      "epoch": 0.053664555281331676,
      "grad_norm": 1.097685694694519,
      "learning_rate": 0.0001892672841301446,
      "loss": 3.8617,
      "step": 57570
    },
    {
      "epoch": 0.05367387689941077,
      "grad_norm": 0.9278309941291809,
      "learning_rate": 0.00018926541980804942,
      "loss": 4.017,
      "step": 57580
    },
    {
      "epoch": 0.05368319851748986,
      "grad_norm": 1.1604267358779907,
      "learning_rate": 0.0001892635554859542,
      "loss": 3.831,
      "step": 57590
    },
    {
      "epoch": 0.05369252013556895,
      "grad_norm": 1.0702967643737793,
      "learning_rate": 0.000189261691163859,
      "loss": 3.7053,
      "step": 57600
    },
    {
      "epoch": 0.05370184175364805,
      "grad_norm": 0.9472962021827698,
      "learning_rate": 0.0001892598268417638,
      "loss": 3.876,
      "step": 57610
    },
    {
      "epoch": 0.053711163371727134,
      "grad_norm": 1.0526796579360962,
      "learning_rate": 0.00018925796251966862,
      "loss": 3.9939,
      "step": 57620
    },
    {
      "epoch": 0.05372048498980623,
      "grad_norm": 1.0962392091751099,
      "learning_rate": 0.00018925609819757342,
      "loss": 3.9793,
      "step": 57630
    },
    {
      "epoch": 0.053729806607885316,
      "grad_norm": 0.9915021657943726,
      "learning_rate": 0.0001892542338754782,
      "loss": 3.8905,
      "step": 57640
    },
    {
      "epoch": 0.05373912822596441,
      "grad_norm": 1.0625678300857544,
      "learning_rate": 0.000189252369553383,
      "loss": 3.8003,
      "step": 57650
    },
    {
      "epoch": 0.053748449844043505,
      "grad_norm": 1.016100287437439,
      "learning_rate": 0.0001892505052312878,
      "loss": 3.7814,
      "step": 57660
    },
    {
      "epoch": 0.05375777146212259,
      "grad_norm": 1.2507866621017456,
      "learning_rate": 0.00018924864090919262,
      "loss": 3.942,
      "step": 57670
    },
    {
      "epoch": 0.05376709308020169,
      "grad_norm": 1.163222074508667,
      "learning_rate": 0.00018924677658709743,
      "loss": 3.8848,
      "step": 57680
    },
    {
      "epoch": 0.05377641469828078,
      "grad_norm": 0.8886209726333618,
      "learning_rate": 0.0001892449122650022,
      "loss": 4.0654,
      "step": 57690
    },
    {
      "epoch": 0.05378573631635987,
      "grad_norm": 0.9036012887954712,
      "learning_rate": 0.000189243047942907,
      "loss": 4.1658,
      "step": 57700
    },
    {
      "epoch": 0.053795057934438964,
      "grad_norm": 1.184363603591919,
      "learning_rate": 0.0001892411836208118,
      "loss": 3.8938,
      "step": 57710
    },
    {
      "epoch": 0.05380437955251805,
      "grad_norm": 1.04966402053833,
      "learning_rate": 0.0001892393192987166,
      "loss": 3.9752,
      "step": 57720
    },
    {
      "epoch": 0.053813701170597146,
      "grad_norm": 1.1530848741531372,
      "learning_rate": 0.00018923745497662143,
      "loss": 3.8144,
      "step": 57730
    },
    {
      "epoch": 0.05382302278867624,
      "grad_norm": 1.0496588945388794,
      "learning_rate": 0.0001892355906545262,
      "loss": 3.9806,
      "step": 57740
    },
    {
      "epoch": 0.05383234440675533,
      "grad_norm": 1.0754061937332153,
      "learning_rate": 0.00018923372633243102,
      "loss": 4.1286,
      "step": 57750
    },
    {
      "epoch": 0.05384166602483442,
      "grad_norm": 0.9898808002471924,
      "learning_rate": 0.0001892318620103358,
      "loss": 4.0636,
      "step": 57760
    },
    {
      "epoch": 0.05385098764291352,
      "grad_norm": 1.0635921955108643,
      "learning_rate": 0.0001892299976882406,
      "loss": 4.0477,
      "step": 57770
    },
    {
      "epoch": 0.053860309260992605,
      "grad_norm": 0.973667562007904,
      "learning_rate": 0.0001892281333661454,
      "loss": 3.8812,
      "step": 57780
    },
    {
      "epoch": 0.0538696308790717,
      "grad_norm": 1.1676708459854126,
      "learning_rate": 0.00018922626904405022,
      "loss": 3.6957,
      "step": 57790
    },
    {
      "epoch": 0.05387895249715079,
      "grad_norm": 1.068045973777771,
      "learning_rate": 0.00018922440472195502,
      "loss": 3.9602,
      "step": 57800
    },
    {
      "epoch": 0.05388827411522988,
      "grad_norm": 1.0199830532073975,
      "learning_rate": 0.0001892225403998598,
      "loss": 4.2322,
      "step": 57810
    },
    {
      "epoch": 0.053897595733308976,
      "grad_norm": 1.1566307544708252,
      "learning_rate": 0.0001892206760777646,
      "loss": 3.8637,
      "step": 57820
    },
    {
      "epoch": 0.05390691735138806,
      "grad_norm": 0.9589557647705078,
      "learning_rate": 0.00018921881175566941,
      "loss": 3.998,
      "step": 57830
    },
    {
      "epoch": 0.05391623896946716,
      "grad_norm": 1.1440545320510864,
      "learning_rate": 0.00018921694743357422,
      "loss": 3.8323,
      "step": 57840
    },
    {
      "epoch": 0.053925560587546245,
      "grad_norm": 1.2413978576660156,
      "learning_rate": 0.00018921508311147903,
      "loss": 3.9802,
      "step": 57850
    },
    {
      "epoch": 0.05393488220562534,
      "grad_norm": 1.0527307987213135,
      "learning_rate": 0.0001892132187893838,
      "loss": 3.8296,
      "step": 57860
    },
    {
      "epoch": 0.053944203823704434,
      "grad_norm": 1.0649725198745728,
      "learning_rate": 0.0001892113544672886,
      "loss": 4.0106,
      "step": 57870
    },
    {
      "epoch": 0.05395352544178352,
      "grad_norm": 1.0263938903808594,
      "learning_rate": 0.00018920949014519342,
      "loss": 3.9802,
      "step": 57880
    },
    {
      "epoch": 0.053962847059862616,
      "grad_norm": 1.1090872287750244,
      "learning_rate": 0.00018920762582309822,
      "loss": 4.104,
      "step": 57890
    },
    {
      "epoch": 0.05397216867794171,
      "grad_norm": 1.2132422924041748,
      "learning_rate": 0.00018920576150100303,
      "loss": 3.9852,
      "step": 57900
    },
    {
      "epoch": 0.0539814902960208,
      "grad_norm": 1.12726891040802,
      "learning_rate": 0.0001892038971789078,
      "loss": 3.913,
      "step": 57910
    },
    {
      "epoch": 0.05399081191409989,
      "grad_norm": 1.1073696613311768,
      "learning_rate": 0.00018920203285681262,
      "loss": 4.0028,
      "step": 57920
    },
    {
      "epoch": 0.05400013353217898,
      "grad_norm": 1.0500679016113281,
      "learning_rate": 0.00018920016853471742,
      "loss": 3.9776,
      "step": 57930
    },
    {
      "epoch": 0.054009455150258075,
      "grad_norm": 1.0445940494537354,
      "learning_rate": 0.0001891983042126222,
      "loss": 4.02,
      "step": 57940
    },
    {
      "epoch": 0.05401877676833717,
      "grad_norm": 1.0335546731948853,
      "learning_rate": 0.000189196439890527,
      "loss": 3.8894,
      "step": 57950
    },
    {
      "epoch": 0.05402809838641626,
      "grad_norm": 1.0297021865844727,
      "learning_rate": 0.00018919457556843181,
      "loss": 4.0003,
      "step": 57960
    },
    {
      "epoch": 0.05403742000449535,
      "grad_norm": 0.9519784450531006,
      "learning_rate": 0.00018919271124633662,
      "loss": 4.0709,
      "step": 57970
    },
    {
      "epoch": 0.05404674162257444,
      "grad_norm": 0.9960577487945557,
      "learning_rate": 0.00018919084692424143,
      "loss": 4.0194,
      "step": 57980
    },
    {
      "epoch": 0.05405606324065353,
      "grad_norm": 1.0497925281524658,
      "learning_rate": 0.0001891889826021462,
      "loss": 4.1167,
      "step": 57990
    },
    {
      "epoch": 0.05406538485873263,
      "grad_norm": 1.058608055114746,
      "learning_rate": 0.000189187118280051,
      "loss": 4.0406,
      "step": 58000
    },
    {
      "epoch": 0.054074706476811715,
      "grad_norm": 1.0427907705307007,
      "learning_rate": 0.00018918525395795582,
      "loss": 4.0667,
      "step": 58010
    },
    {
      "epoch": 0.05408402809489081,
      "grad_norm": 0.9889691472053528,
      "learning_rate": 0.00018918338963586063,
      "loss": 4.1305,
      "step": 58020
    },
    {
      "epoch": 0.054093349712969904,
      "grad_norm": 1.06781804561615,
      "learning_rate": 0.00018918152531376543,
      "loss": 4.0992,
      "step": 58030
    },
    {
      "epoch": 0.05410267133104899,
      "grad_norm": 1.0734442472457886,
      "learning_rate": 0.0001891796609916702,
      "loss": 4.0842,
      "step": 58040
    },
    {
      "epoch": 0.054111992949128086,
      "grad_norm": 1.0818588733673096,
      "learning_rate": 0.00018917779666957502,
      "loss": 4.047,
      "step": 58050
    },
    {
      "epoch": 0.054121314567207174,
      "grad_norm": 1.1999236345291138,
      "learning_rate": 0.00018917593234747982,
      "loss": 3.9816,
      "step": 58060
    },
    {
      "epoch": 0.05413063618528627,
      "grad_norm": 1.1760749816894531,
      "learning_rate": 0.00018917406802538463,
      "loss": 3.9852,
      "step": 58070
    },
    {
      "epoch": 0.05413995780336536,
      "grad_norm": 1.1373846530914307,
      "learning_rate": 0.0001891722037032894,
      "loss": 4.085,
      "step": 58080
    },
    {
      "epoch": 0.05414927942144445,
      "grad_norm": 1.2057515382766724,
      "learning_rate": 0.00018917033938119422,
      "loss": 3.9106,
      "step": 58090
    },
    {
      "epoch": 0.054158601039523545,
      "grad_norm": 1.0163804292678833,
      "learning_rate": 0.00018916847505909902,
      "loss": 3.9075,
      "step": 58100
    },
    {
      "epoch": 0.05416792265760264,
      "grad_norm": 1.1735992431640625,
      "learning_rate": 0.0001891666107370038,
      "loss": 4.0154,
      "step": 58110
    },
    {
      "epoch": 0.05417724427568173,
      "grad_norm": 0.9925050139427185,
      "learning_rate": 0.00018916474641490863,
      "loss": 4.0015,
      "step": 58120
    },
    {
      "epoch": 0.05418656589376082,
      "grad_norm": 1.038506269454956,
      "learning_rate": 0.0001891628820928134,
      "loss": 4.3584,
      "step": 58130
    },
    {
      "epoch": 0.05419588751183991,
      "grad_norm": 0.9110045433044434,
      "learning_rate": 0.00018916101777071822,
      "loss": 3.9733,
      "step": 58140
    },
    {
      "epoch": 0.054205209129919,
      "grad_norm": 1.0636610984802246,
      "learning_rate": 0.00018915915344862303,
      "loss": 3.9571,
      "step": 58150
    },
    {
      "epoch": 0.0542145307479981,
      "grad_norm": 1.2302403450012207,
      "learning_rate": 0.0001891572891265278,
      "loss": 3.9364,
      "step": 58160
    },
    {
      "epoch": 0.054223852366077185,
      "grad_norm": 1.1408004760742188,
      "learning_rate": 0.0001891554248044326,
      "loss": 3.9909,
      "step": 58170
    },
    {
      "epoch": 0.05423317398415628,
      "grad_norm": 1.0673600435256958,
      "learning_rate": 0.00018915356048233742,
      "loss": 4.0336,
      "step": 58180
    },
    {
      "epoch": 0.05424249560223537,
      "grad_norm": 1.0058284997940063,
      "learning_rate": 0.00018915169616024222,
      "loss": 4.0029,
      "step": 58190
    },
    {
      "epoch": 0.05425181722031446,
      "grad_norm": 1.1287002563476562,
      "learning_rate": 0.00018914983183814703,
      "loss": 4.0495,
      "step": 58200
    },
    {
      "epoch": 0.054261138838393556,
      "grad_norm": 1.2692993879318237,
      "learning_rate": 0.0001891479675160518,
      "loss": 3.7953,
      "step": 58210
    },
    {
      "epoch": 0.054270460456472644,
      "grad_norm": 1.1233419179916382,
      "learning_rate": 0.00018914610319395662,
      "loss": 3.9278,
      "step": 58220
    },
    {
      "epoch": 0.05427978207455174,
      "grad_norm": 1.2382605075836182,
      "learning_rate": 0.00018914423887186142,
      "loss": 4.0683,
      "step": 58230
    },
    {
      "epoch": 0.05428910369263083,
      "grad_norm": 1.1628049612045288,
      "learning_rate": 0.00018914237454976623,
      "loss": 3.9475,
      "step": 58240
    },
    {
      "epoch": 0.05429842531070992,
      "grad_norm": 1.1144577264785767,
      "learning_rate": 0.00018914051022767103,
      "loss": 4.1546,
      "step": 58250
    },
    {
      "epoch": 0.054307746928789015,
      "grad_norm": 1.0832213163375854,
      "learning_rate": 0.00018913864590557581,
      "loss": 3.8679,
      "step": 58260
    },
    {
      "epoch": 0.0543170685468681,
      "grad_norm": 0.9462432861328125,
      "learning_rate": 0.00018913678158348062,
      "loss": 4.0971,
      "step": 58270
    },
    {
      "epoch": 0.0543263901649472,
      "grad_norm": 0.9938971400260925,
      "learning_rate": 0.00018913491726138543,
      "loss": 3.9607,
      "step": 58280
    },
    {
      "epoch": 0.05433571178302629,
      "grad_norm": 1.1684231758117676,
      "learning_rate": 0.00018913305293929023,
      "loss": 3.7942,
      "step": 58290
    },
    {
      "epoch": 0.05434503340110538,
      "grad_norm": 1.119873046875,
      "learning_rate": 0.00018913118861719504,
      "loss": 3.8743,
      "step": 58300
    },
    {
      "epoch": 0.054354355019184473,
      "grad_norm": 1.1904833316802979,
      "learning_rate": 0.00018912932429509982,
      "loss": 3.8744,
      "step": 58310
    },
    {
      "epoch": 0.05436367663726356,
      "grad_norm": 0.8205487728118896,
      "learning_rate": 0.00018912745997300462,
      "loss": 3.9548,
      "step": 58320
    },
    {
      "epoch": 0.054372998255342656,
      "grad_norm": 1.1937123537063599,
      "learning_rate": 0.0001891255956509094,
      "loss": 3.8977,
      "step": 58330
    },
    {
      "epoch": 0.05438231987342175,
      "grad_norm": 1.0038504600524902,
      "learning_rate": 0.00018912373132881424,
      "loss": 4.0038,
      "step": 58340
    },
    {
      "epoch": 0.05439164149150084,
      "grad_norm": 1.3459168672561646,
      "learning_rate": 0.00018912186700671904,
      "loss": 3.9183,
      "step": 58350
    },
    {
      "epoch": 0.05440096310957993,
      "grad_norm": 0.9879798293113708,
      "learning_rate": 0.00018912000268462382,
      "loss": 3.8316,
      "step": 58360
    },
    {
      "epoch": 0.054410284727659027,
      "grad_norm": 1.0304224491119385,
      "learning_rate": 0.00018911813836252863,
      "loss": 4.0505,
      "step": 58370
    },
    {
      "epoch": 0.054419606345738114,
      "grad_norm": 1.0416240692138672,
      "learning_rate": 0.0001891162740404334,
      "loss": 3.8222,
      "step": 58380
    },
    {
      "epoch": 0.05442892796381721,
      "grad_norm": 1.1709023714065552,
      "learning_rate": 0.00018911440971833821,
      "loss": 3.9034,
      "step": 58390
    },
    {
      "epoch": 0.054438249581896296,
      "grad_norm": 1.030332088470459,
      "learning_rate": 0.00018911254539624305,
      "loss": 3.6896,
      "step": 58400
    },
    {
      "epoch": 0.05444757119997539,
      "grad_norm": 1.1218899488449097,
      "learning_rate": 0.00018911068107414783,
      "loss": 4.0044,
      "step": 58410
    },
    {
      "epoch": 0.054456892818054485,
      "grad_norm": 0.89166659116745,
      "learning_rate": 0.00018910881675205263,
      "loss": 3.8284,
      "step": 58420
    },
    {
      "epoch": 0.05446621443613357,
      "grad_norm": 0.9703946709632874,
      "learning_rate": 0.0001891069524299574,
      "loss": 3.8661,
      "step": 58430
    },
    {
      "epoch": 0.05447553605421267,
      "grad_norm": 1.110171914100647,
      "learning_rate": 0.00018910508810786222,
      "loss": 4.1058,
      "step": 58440
    },
    {
      "epoch": 0.05448485767229176,
      "grad_norm": 1.1442573070526123,
      "learning_rate": 0.00018910322378576703,
      "loss": 3.9516,
      "step": 58450
    },
    {
      "epoch": 0.05449417929037085,
      "grad_norm": 1.138299822807312,
      "learning_rate": 0.00018910135946367183,
      "loss": 3.9167,
      "step": 58460
    },
    {
      "epoch": 0.054503500908449944,
      "grad_norm": 1.1223887205123901,
      "learning_rate": 0.00018909949514157664,
      "loss": 3.9842,
      "step": 58470
    },
    {
      "epoch": 0.05451282252652903,
      "grad_norm": 1.181692123413086,
      "learning_rate": 0.00018909763081948142,
      "loss": 3.8831,
      "step": 58480
    },
    {
      "epoch": 0.054522144144608126,
      "grad_norm": 1.1842516660690308,
      "learning_rate": 0.00018909576649738622,
      "loss": 3.9792,
      "step": 58490
    },
    {
      "epoch": 0.05453146576268722,
      "grad_norm": 1.2893067598342896,
      "learning_rate": 0.00018909390217529103,
      "loss": 3.9106,
      "step": 58500
    },
    {
      "epoch": 0.05454078738076631,
      "grad_norm": 1.0218502283096313,
      "learning_rate": 0.00018909203785319584,
      "loss": 4.0857,
      "step": 58510
    },
    {
      "epoch": 0.0545501089988454,
      "grad_norm": 0.9744971990585327,
      "learning_rate": 0.00018909017353110064,
      "loss": 3.8419,
      "step": 58520
    },
    {
      "epoch": 0.05455943061692449,
      "grad_norm": 1.2217622995376587,
      "learning_rate": 0.00018908830920900542,
      "loss": 3.9072,
      "step": 58530
    },
    {
      "epoch": 0.054568752235003584,
      "grad_norm": 1.0697897672653198,
      "learning_rate": 0.00018908644488691023,
      "loss": 3.8818,
      "step": 58540
    },
    {
      "epoch": 0.05457807385308268,
      "grad_norm": 1.0065933465957642,
      "learning_rate": 0.00018908458056481503,
      "loss": 3.9835,
      "step": 58550
    },
    {
      "epoch": 0.054587395471161766,
      "grad_norm": 0.9781011939048767,
      "learning_rate": 0.00018908271624271981,
      "loss": 4.09,
      "step": 58560
    },
    {
      "epoch": 0.05459671708924086,
      "grad_norm": 1.060929298400879,
      "learning_rate": 0.00018908085192062465,
      "loss": 3.8621,
      "step": 58570
    },
    {
      "epoch": 0.054606038707319955,
      "grad_norm": 1.1296871900558472,
      "learning_rate": 0.00018907898759852943,
      "loss": 3.875,
      "step": 58580
    },
    {
      "epoch": 0.05461536032539904,
      "grad_norm": 1.029658555984497,
      "learning_rate": 0.00018907712327643423,
      "loss": 3.9635,
      "step": 58590
    },
    {
      "epoch": 0.05462468194347814,
      "grad_norm": 0.8712491393089294,
      "learning_rate": 0.00018907525895433904,
      "loss": 3.9999,
      "step": 58600
    },
    {
      "epoch": 0.054634003561557225,
      "grad_norm": 1.089099645614624,
      "learning_rate": 0.00018907339463224382,
      "loss": 4.0274,
      "step": 58610
    },
    {
      "epoch": 0.05464332517963632,
      "grad_norm": 0.9435921907424927,
      "learning_rate": 0.00018907153031014862,
      "loss": 4.1123,
      "step": 58620
    },
    {
      "epoch": 0.054652646797715414,
      "grad_norm": 1.110429048538208,
      "learning_rate": 0.00018906966598805343,
      "loss": 3.9992,
      "step": 58630
    },
    {
      "epoch": 0.0546619684157945,
      "grad_norm": 0.9768145084381104,
      "learning_rate": 0.00018906780166595824,
      "loss": 3.9043,
      "step": 58640
    },
    {
      "epoch": 0.054671290033873596,
      "grad_norm": 1.3277325630187988,
      "learning_rate": 0.00018906593734386304,
      "loss": 3.9054,
      "step": 58650
    },
    {
      "epoch": 0.05468061165195268,
      "grad_norm": 0.9587363600730896,
      "learning_rate": 0.00018906407302176782,
      "loss": 3.9835,
      "step": 58660
    },
    {
      "epoch": 0.05468993327003178,
      "grad_norm": 1.4046820402145386,
      "learning_rate": 0.00018906220869967263,
      "loss": 3.9106,
      "step": 58670
    },
    {
      "epoch": 0.05469925488811087,
      "grad_norm": 0.9678030014038086,
      "learning_rate": 0.00018906034437757744,
      "loss": 3.9571,
      "step": 58680
    },
    {
      "epoch": 0.05470857650618996,
      "grad_norm": 1.0973451137542725,
      "learning_rate": 0.00018905848005548224,
      "loss": 3.9903,
      "step": 58690
    },
    {
      "epoch": 0.054717898124269054,
      "grad_norm": 1.0460411310195923,
      "learning_rate": 0.00018905661573338705,
      "loss": 4.0426,
      "step": 58700
    },
    {
      "epoch": 0.05472721974234815,
      "grad_norm": 1.0571718215942383,
      "learning_rate": 0.00018905475141129183,
      "loss": 3.7991,
      "step": 58710
    },
    {
      "epoch": 0.054736541360427236,
      "grad_norm": 0.8909826278686523,
      "learning_rate": 0.00018905288708919663,
      "loss": 3.9473,
      "step": 58720
    },
    {
      "epoch": 0.05474586297850633,
      "grad_norm": 1.1281143426895142,
      "learning_rate": 0.00018905102276710144,
      "loss": 4.1026,
      "step": 58730
    },
    {
      "epoch": 0.05475518459658542,
      "grad_norm": 1.1158814430236816,
      "learning_rate": 0.00018904915844500625,
      "loss": 3.9494,
      "step": 58740
    },
    {
      "epoch": 0.05476450621466451,
      "grad_norm": 1.213844895362854,
      "learning_rate": 0.00018904729412291105,
      "loss": 4.0034,
      "step": 58750
    },
    {
      "epoch": 0.05477382783274361,
      "grad_norm": 1.146836757659912,
      "learning_rate": 0.00018904542980081583,
      "loss": 3.9528,
      "step": 58760
    },
    {
      "epoch": 0.054783149450822695,
      "grad_norm": 0.9643229842185974,
      "learning_rate": 0.00018904356547872064,
      "loss": 3.998,
      "step": 58770
    },
    {
      "epoch": 0.05479247106890179,
      "grad_norm": 1.092533826828003,
      "learning_rate": 0.00018904170115662542,
      "loss": 3.9461,
      "step": 58780
    },
    {
      "epoch": 0.054801792686980884,
      "grad_norm": 1.11624014377594,
      "learning_rate": 0.00018903983683453025,
      "loss": 3.9091,
      "step": 58790
    },
    {
      "epoch": 0.05481111430505997,
      "grad_norm": 1.1705560684204102,
      "learning_rate": 0.00018903797251243506,
      "loss": 3.7941,
      "step": 58800
    },
    {
      "epoch": 0.054820435923139066,
      "grad_norm": 0.9780440926551819,
      "learning_rate": 0.00018903610819033984,
      "loss": 4.046,
      "step": 58810
    },
    {
      "epoch": 0.05482975754121815,
      "grad_norm": 1.2418042421340942,
      "learning_rate": 0.00018903424386824464,
      "loss": 3.8939,
      "step": 58820
    },
    {
      "epoch": 0.05483907915929725,
      "grad_norm": 1.0940011739730835,
      "learning_rate": 0.00018903237954614942,
      "loss": 3.7564,
      "step": 58830
    },
    {
      "epoch": 0.05484840077737634,
      "grad_norm": 0.8615045547485352,
      "learning_rate": 0.00018903051522405423,
      "loss": 3.9244,
      "step": 58840
    },
    {
      "epoch": 0.05485772239545543,
      "grad_norm": 1.2900489568710327,
      "learning_rate": 0.00018902865090195906,
      "loss": 4.0292,
      "step": 58850
    },
    {
      "epoch": 0.054867044013534524,
      "grad_norm": 1.1469944715499878,
      "learning_rate": 0.00018902678657986384,
      "loss": 3.9668,
      "step": 58860
    },
    {
      "epoch": 0.05487636563161361,
      "grad_norm": 1.019144058227539,
      "learning_rate": 0.00018902492225776865,
      "loss": 3.9105,
      "step": 58870
    },
    {
      "epoch": 0.054885687249692706,
      "grad_norm": 1.1821805238723755,
      "learning_rate": 0.00018902305793567343,
      "loss": 4.0477,
      "step": 58880
    },
    {
      "epoch": 0.0548950088677718,
      "grad_norm": 1.1777236461639404,
      "learning_rate": 0.00018902119361357823,
      "loss": 3.8603,
      "step": 58890
    },
    {
      "epoch": 0.05490433048585089,
      "grad_norm": 1.3118048906326294,
      "learning_rate": 0.00018901932929148304,
      "loss": 3.8787,
      "step": 58900
    },
    {
      "epoch": 0.05491365210392998,
      "grad_norm": 1.0695236921310425,
      "learning_rate": 0.00018901746496938784,
      "loss": 4.0464,
      "step": 58910
    },
    {
      "epoch": 0.05492297372200908,
      "grad_norm": 1.0137296915054321,
      "learning_rate": 0.00018901560064729265,
      "loss": 3.8151,
      "step": 58920
    },
    {
      "epoch": 0.054932295340088165,
      "grad_norm": 1.1784900426864624,
      "learning_rate": 0.00018901373632519743,
      "loss": 4.0451,
      "step": 58930
    },
    {
      "epoch": 0.05494161695816726,
      "grad_norm": 1.1377577781677246,
      "learning_rate": 0.00018901187200310224,
      "loss": 3.9347,
      "step": 58940
    },
    {
      "epoch": 0.05495093857624635,
      "grad_norm": 1.2737114429473877,
      "learning_rate": 0.00018901000768100704,
      "loss": 4.0197,
      "step": 58950
    },
    {
      "epoch": 0.05496026019432544,
      "grad_norm": 1.038218379020691,
      "learning_rate": 0.00018900814335891185,
      "loss": 4.0998,
      "step": 58960
    },
    {
      "epoch": 0.054969581812404536,
      "grad_norm": 1.4366286993026733,
      "learning_rate": 0.00018900627903681666,
      "loss": 4.0034,
      "step": 58970
    },
    {
      "epoch": 0.054978903430483624,
      "grad_norm": 1.0072746276855469,
      "learning_rate": 0.00018900441471472143,
      "loss": 3.9699,
      "step": 58980
    },
    {
      "epoch": 0.05498822504856272,
      "grad_norm": 0.9163904190063477,
      "learning_rate": 0.00018900255039262624,
      "loss": 4.0391,
      "step": 58990
    },
    {
      "epoch": 0.054997546666641806,
      "grad_norm": 1.1626819372177124,
      "learning_rate": 0.00018900068607053105,
      "loss": 3.9185,
      "step": 59000
    },
    {
      "epoch": 0.0550068682847209,
      "grad_norm": 1.2311490774154663,
      "learning_rate": 0.00018899882174843583,
      "loss": 4.1529,
      "step": 59010
    },
    {
      "epoch": 0.055016189902799995,
      "grad_norm": 1.1329009532928467,
      "learning_rate": 0.00018899695742634066,
      "loss": 3.9309,
      "step": 59020
    },
    {
      "epoch": 0.05502551152087908,
      "grad_norm": 0.9743552207946777,
      "learning_rate": 0.00018899509310424544,
      "loss": 3.9572,
      "step": 59030
    },
    {
      "epoch": 0.05503483313895818,
      "grad_norm": 1.0942891836166382,
      "learning_rate": 0.00018899322878215025,
      "loss": 3.925,
      "step": 59040
    },
    {
      "epoch": 0.05504415475703727,
      "grad_norm": 1.193483829498291,
      "learning_rate": 0.00018899136446005505,
      "loss": 3.7533,
      "step": 59050
    },
    {
      "epoch": 0.05505347637511636,
      "grad_norm": 1.1774821281433105,
      "learning_rate": 0.00018898950013795983,
      "loss": 4.0396,
      "step": 59060
    },
    {
      "epoch": 0.05506279799319545,
      "grad_norm": 1.0727295875549316,
      "learning_rate": 0.00018898763581586464,
      "loss": 4.0898,
      "step": 59070
    },
    {
      "epoch": 0.05507211961127454,
      "grad_norm": 1.236875057220459,
      "learning_rate": 0.00018898577149376944,
      "loss": 3.8089,
      "step": 59080
    },
    {
      "epoch": 0.055081441229353635,
      "grad_norm": 1.2067384719848633,
      "learning_rate": 0.00018898390717167425,
      "loss": 3.8242,
      "step": 59090
    },
    {
      "epoch": 0.05509076284743273,
      "grad_norm": 0.9077203869819641,
      "learning_rate": 0.00018898204284957906,
      "loss": 4.0078,
      "step": 59100
    },
    {
      "epoch": 0.05510008446551182,
      "grad_norm": 0.8680023550987244,
      "learning_rate": 0.00018898017852748384,
      "loss": 4.0343,
      "step": 59110
    },
    {
      "epoch": 0.05510940608359091,
      "grad_norm": 1.2063087224960327,
      "learning_rate": 0.00018897831420538864,
      "loss": 4.0239,
      "step": 59120
    },
    {
      "epoch": 0.055118727701670006,
      "grad_norm": 1.0129895210266113,
      "learning_rate": 0.00018897644988329345,
      "loss": 3.7939,
      "step": 59130
    },
    {
      "epoch": 0.055128049319749094,
      "grad_norm": 1.130779504776001,
      "learning_rate": 0.00018897458556119825,
      "loss": 4.0532,
      "step": 59140
    },
    {
      "epoch": 0.05513737093782819,
      "grad_norm": 1.0038487911224365,
      "learning_rate": 0.00018897272123910306,
      "loss": 3.8465,
      "step": 59150
    },
    {
      "epoch": 0.055146692555907276,
      "grad_norm": 1.1037139892578125,
      "learning_rate": 0.00018897085691700784,
      "loss": 4.1639,
      "step": 59160
    },
    {
      "epoch": 0.05515601417398637,
      "grad_norm": 1.066960334777832,
      "learning_rate": 0.00018896899259491265,
      "loss": 3.859,
      "step": 59170
    },
    {
      "epoch": 0.055165335792065465,
      "grad_norm": 1.0882610082626343,
      "learning_rate": 0.00018896712827281745,
      "loss": 3.8872,
      "step": 59180
    },
    {
      "epoch": 0.05517465741014455,
      "grad_norm": 1.0569089651107788,
      "learning_rate": 0.00018896526395072226,
      "loss": 3.8215,
      "step": 59190
    },
    {
      "epoch": 0.05518397902822365,
      "grad_norm": 0.9190982580184937,
      "learning_rate": 0.00018896339962862707,
      "loss": 3.8661,
      "step": 59200
    },
    {
      "epoch": 0.055193300646302734,
      "grad_norm": 1.1559709310531616,
      "learning_rate": 0.00018896153530653184,
      "loss": 3.9566,
      "step": 59210
    },
    {
      "epoch": 0.05520262226438183,
      "grad_norm": 1.1214195489883423,
      "learning_rate": 0.00018895967098443665,
      "loss": 3.9151,
      "step": 59220
    },
    {
      "epoch": 0.05521194388246092,
      "grad_norm": 1.2677782773971558,
      "learning_rate": 0.00018895780666234143,
      "loss": 3.8216,
      "step": 59230
    },
    {
      "epoch": 0.05522126550054001,
      "grad_norm": 1.0321927070617676,
      "learning_rate": 0.00018895594234024626,
      "loss": 3.9288,
      "step": 59240
    },
    {
      "epoch": 0.055230587118619105,
      "grad_norm": 1.121477484703064,
      "learning_rate": 0.00018895407801815107,
      "loss": 3.7669,
      "step": 59250
    },
    {
      "epoch": 0.0552399087366982,
      "grad_norm": 1.0483719110488892,
      "learning_rate": 0.00018895221369605585,
      "loss": 3.8838,
      "step": 59260
    },
    {
      "epoch": 0.05524923035477729,
      "grad_norm": 1.19350004196167,
      "learning_rate": 0.00018895034937396066,
      "loss": 3.9564,
      "step": 59270
    },
    {
      "epoch": 0.05525855197285638,
      "grad_norm": 1.1184533834457397,
      "learning_rate": 0.00018894848505186543,
      "loss": 3.9156,
      "step": 59280
    },
    {
      "epoch": 0.05526787359093547,
      "grad_norm": 1.0326621532440186,
      "learning_rate": 0.00018894662072977024,
      "loss": 4.057,
      "step": 59290
    },
    {
      "epoch": 0.055277195209014564,
      "grad_norm": 1.150278925895691,
      "learning_rate": 0.00018894475640767507,
      "loss": 3.9669,
      "step": 59300
    },
    {
      "epoch": 0.05528651682709366,
      "grad_norm": 1.1240441799163818,
      "learning_rate": 0.00018894289208557985,
      "loss": 3.8226,
      "step": 59310
    },
    {
      "epoch": 0.055295838445172746,
      "grad_norm": 0.9274193644523621,
      "learning_rate": 0.00018894102776348466,
      "loss": 4.0818,
      "step": 59320
    },
    {
      "epoch": 0.05530516006325184,
      "grad_norm": 1.0880051851272583,
      "learning_rate": 0.00018893916344138944,
      "loss": 3.9731,
      "step": 59330
    },
    {
      "epoch": 0.05531448168133093,
      "grad_norm": 0.946973979473114,
      "learning_rate": 0.00018893729911929425,
      "loss": 4.0782,
      "step": 59340
    },
    {
      "epoch": 0.05532380329941002,
      "grad_norm": 1.0684294700622559,
      "learning_rate": 0.00018893543479719905,
      "loss": 3.8635,
      "step": 59350
    },
    {
      "epoch": 0.05533312491748912,
      "grad_norm": 0.998764157295227,
      "learning_rate": 0.00018893357047510386,
      "loss": 3.8914,
      "step": 59360
    },
    {
      "epoch": 0.055342446535568204,
      "grad_norm": 1.0014574527740479,
      "learning_rate": 0.00018893170615300866,
      "loss": 3.8879,
      "step": 59370
    },
    {
      "epoch": 0.0553517681536473,
      "grad_norm": 0.9987020492553711,
      "learning_rate": 0.00018892984183091344,
      "loss": 3.9843,
      "step": 59380
    },
    {
      "epoch": 0.05536108977172639,
      "grad_norm": 1.1738030910491943,
      "learning_rate": 0.00018892797750881825,
      "loss": 3.8991,
      "step": 59390
    },
    {
      "epoch": 0.05537041138980548,
      "grad_norm": 1.039703369140625,
      "learning_rate": 0.00018892611318672306,
      "loss": 3.9099,
      "step": 59400
    },
    {
      "epoch": 0.055379733007884575,
      "grad_norm": 1.1993811130523682,
      "learning_rate": 0.00018892424886462786,
      "loss": 3.9093,
      "step": 59410
    },
    {
      "epoch": 0.05538905462596366,
      "grad_norm": 1.1612821817398071,
      "learning_rate": 0.00018892238454253267,
      "loss": 4.0658,
      "step": 59420
    },
    {
      "epoch": 0.05539837624404276,
      "grad_norm": 1.0214918851852417,
      "learning_rate": 0.00018892052022043745,
      "loss": 4.0554,
      "step": 59430
    },
    {
      "epoch": 0.05540769786212185,
      "grad_norm": 1.1532961130142212,
      "learning_rate": 0.00018891865589834225,
      "loss": 3.9307,
      "step": 59440
    },
    {
      "epoch": 0.05541701948020094,
      "grad_norm": 1.1140999794006348,
      "learning_rate": 0.00018891679157624703,
      "loss": 3.8773,
      "step": 59450
    },
    {
      "epoch": 0.055426341098280034,
      "grad_norm": 1.0320851802825928,
      "learning_rate": 0.00018891492725415184,
      "loss": 3.9959,
      "step": 59460
    },
    {
      "epoch": 0.05543566271635913,
      "grad_norm": 1.057146668434143,
      "learning_rate": 0.00018891306293205667,
      "loss": 3.9159,
      "step": 59470
    },
    {
      "epoch": 0.055444984334438216,
      "grad_norm": 1.1729600429534912,
      "learning_rate": 0.00018891119860996145,
      "loss": 3.8484,
      "step": 59480
    },
    {
      "epoch": 0.05545430595251731,
      "grad_norm": 1.0955878496170044,
      "learning_rate": 0.00018890933428786626,
      "loss": 3.613,
      "step": 59490
    },
    {
      "epoch": 0.0554636275705964,
      "grad_norm": 0.9989116191864014,
      "learning_rate": 0.00018890746996577104,
      "loss": 3.8524,
      "step": 59500
    },
    {
      "epoch": 0.05547294918867549,
      "grad_norm": 1.1268703937530518,
      "learning_rate": 0.00018890560564367584,
      "loss": 4.0956,
      "step": 59510
    },
    {
      "epoch": 0.05548227080675459,
      "grad_norm": 1.0046641826629639,
      "learning_rate": 0.00018890374132158065,
      "loss": 3.7822,
      "step": 59520
    },
    {
      "epoch": 0.055491592424833674,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018890187699948546,
      "loss": 4.1548,
      "step": 59530
    },
    {
      "epoch": 0.05550091404291277,
      "grad_norm": 1.1695936918258667,
      "learning_rate": 0.00018890001267739026,
      "loss": 3.8986,
      "step": 59540
    },
    {
      "epoch": 0.055510235660991857,
      "grad_norm": 0.9958563446998596,
      "learning_rate": 0.00018889814835529504,
      "loss": 3.8565,
      "step": 59550
    },
    {
      "epoch": 0.05551955727907095,
      "grad_norm": 0.9330542087554932,
      "learning_rate": 0.00018889628403319985,
      "loss": 4.0024,
      "step": 59560
    },
    {
      "epoch": 0.055528878897150045,
      "grad_norm": 1.1313014030456543,
      "learning_rate": 0.00018889441971110465,
      "loss": 3.8687,
      "step": 59570
    },
    {
      "epoch": 0.05553820051522913,
      "grad_norm": 1.0574142932891846,
      "learning_rate": 0.00018889255538900946,
      "loss": 3.9009,
      "step": 59580
    },
    {
      "epoch": 0.05554752213330823,
      "grad_norm": 1.2181931734085083,
      "learning_rate": 0.00018889069106691427,
      "loss": 3.7304,
      "step": 59590
    },
    {
      "epoch": 0.05555684375138732,
      "grad_norm": 1.1649553775787354,
      "learning_rate": 0.00018888882674481905,
      "loss": 4.219,
      "step": 59600
    },
    {
      "epoch": 0.05556616536946641,
      "grad_norm": 0.9253551959991455,
      "learning_rate": 0.00018888696242272385,
      "loss": 3.6701,
      "step": 59610
    },
    {
      "epoch": 0.055575486987545504,
      "grad_norm": 1.3116133213043213,
      "learning_rate": 0.00018888509810062866,
      "loss": 4.0886,
      "step": 59620
    },
    {
      "epoch": 0.05558480860562459,
      "grad_norm": 0.9275906682014465,
      "learning_rate": 0.00018888323377853347,
      "loss": 3.9361,
      "step": 59630
    },
    {
      "epoch": 0.055594130223703686,
      "grad_norm": 0.9087586402893066,
      "learning_rate": 0.00018888136945643827,
      "loss": 4.0028,
      "step": 59640
    },
    {
      "epoch": 0.05560345184178278,
      "grad_norm": 1.0452059507369995,
      "learning_rate": 0.00018887950513434305,
      "loss": 3.9516,
      "step": 59650
    },
    {
      "epoch": 0.05561277345986187,
      "grad_norm": 1.1402565240859985,
      "learning_rate": 0.00018887764081224786,
      "loss": 3.9606,
      "step": 59660
    },
    {
      "epoch": 0.05562209507794096,
      "grad_norm": 1.1433457136154175,
      "learning_rate": 0.00018887577649015266,
      "loss": 3.8981,
      "step": 59670
    },
    {
      "epoch": 0.05563141669602005,
      "grad_norm": 1.0557161569595337,
      "learning_rate": 0.00018887391216805744,
      "loss": 4.1579,
      "step": 59680
    },
    {
      "epoch": 0.055640738314099145,
      "grad_norm": 1.191161036491394,
      "learning_rate": 0.00018887204784596228,
      "loss": 4.0146,
      "step": 59690
    },
    {
      "epoch": 0.05565005993217824,
      "grad_norm": 1.098775029182434,
      "learning_rate": 0.00018887018352386706,
      "loss": 3.9063,
      "step": 59700
    },
    {
      "epoch": 0.05565938155025733,
      "grad_norm": 1.2544918060302734,
      "learning_rate": 0.00018886831920177186,
      "loss": 3.886,
      "step": 59710
    },
    {
      "epoch": 0.05566870316833642,
      "grad_norm": 1.2639310359954834,
      "learning_rate": 0.00018886645487967667,
      "loss": 3.9821,
      "step": 59720
    },
    {
      "epoch": 0.055678024786415516,
      "grad_norm": 1.1785985231399536,
      "learning_rate": 0.00018886459055758145,
      "loss": 4.0238,
      "step": 59730
    },
    {
      "epoch": 0.0556873464044946,
      "grad_norm": 0.9937004446983337,
      "learning_rate": 0.00018886272623548625,
      "loss": 4.1343,
      "step": 59740
    },
    {
      "epoch": 0.0556966680225737,
      "grad_norm": 1.0908716917037964,
      "learning_rate": 0.00018886086191339106,
      "loss": 3.9614,
      "step": 59750
    },
    {
      "epoch": 0.055705989640652785,
      "grad_norm": 0.9191951751708984,
      "learning_rate": 0.00018885899759129587,
      "loss": 3.9582,
      "step": 59760
    },
    {
      "epoch": 0.05571531125873188,
      "grad_norm": 1.030165433883667,
      "learning_rate": 0.00018885713326920067,
      "loss": 4.0916,
      "step": 59770
    },
    {
      "epoch": 0.055724632876810974,
      "grad_norm": 1.2110624313354492,
      "learning_rate": 0.00018885526894710545,
      "loss": 3.8941,
      "step": 59780
    },
    {
      "epoch": 0.05573395449489006,
      "grad_norm": 1.0049474239349365,
      "learning_rate": 0.00018885340462501026,
      "loss": 3.8093,
      "step": 59790
    },
    {
      "epoch": 0.055743276112969156,
      "grad_norm": 1.209747314453125,
      "learning_rate": 0.00018885154030291506,
      "loss": 3.9546,
      "step": 59800
    },
    {
      "epoch": 0.05575259773104825,
      "grad_norm": 1.1546950340270996,
      "learning_rate": 0.00018884967598081987,
      "loss": 4.021,
      "step": 59810
    },
    {
      "epoch": 0.05576191934912734,
      "grad_norm": 1.063696026802063,
      "learning_rate": 0.00018884781165872468,
      "loss": 3.9744,
      "step": 59820
    },
    {
      "epoch": 0.05577124096720643,
      "grad_norm": 0.9502264857292175,
      "learning_rate": 0.00018884594733662946,
      "loss": 3.8767,
      "step": 59830
    },
    {
      "epoch": 0.05578056258528552,
      "grad_norm": 0.8483677506446838,
      "learning_rate": 0.00018884408301453426,
      "loss": 3.8823,
      "step": 59840
    },
    {
      "epoch": 0.055789884203364615,
      "grad_norm": 1.1635220050811768,
      "learning_rate": 0.00018884221869243904,
      "loss": 3.9345,
      "step": 59850
    },
    {
      "epoch": 0.05579920582144371,
      "grad_norm": 1.0520942211151123,
      "learning_rate": 0.00018884035437034387,
      "loss": 3.9638,
      "step": 59860
    },
    {
      "epoch": 0.0558085274395228,
      "grad_norm": 1.0783486366271973,
      "learning_rate": 0.00018883849004824868,
      "loss": 4.0714,
      "step": 59870
    },
    {
      "epoch": 0.05581784905760189,
      "grad_norm": 0.9835946559906006,
      "learning_rate": 0.00018883662572615346,
      "loss": 4.0041,
      "step": 59880
    },
    {
      "epoch": 0.05582717067568098,
      "grad_norm": 0.9982288479804993,
      "learning_rate": 0.00018883476140405827,
      "loss": 3.986,
      "step": 59890
    },
    {
      "epoch": 0.05583649229376007,
      "grad_norm": 1.025846004486084,
      "learning_rate": 0.00018883289708196305,
      "loss": 3.9526,
      "step": 59900
    },
    {
      "epoch": 0.05584581391183917,
      "grad_norm": 1.1457805633544922,
      "learning_rate": 0.00018883103275986785,
      "loss": 4.0922,
      "step": 59910
    },
    {
      "epoch": 0.055855135529918255,
      "grad_norm": 1.063565731048584,
      "learning_rate": 0.00018882916843777269,
      "loss": 3.8751,
      "step": 59920
    },
    {
      "epoch": 0.05586445714799735,
      "grad_norm": 1.058162808418274,
      "learning_rate": 0.00018882730411567746,
      "loss": 4.0607,
      "step": 59930
    },
    {
      "epoch": 0.055873778766076444,
      "grad_norm": 0.9178330898284912,
      "learning_rate": 0.00018882543979358227,
      "loss": 4.0673,
      "step": 59940
    },
    {
      "epoch": 0.05588310038415553,
      "grad_norm": 1.164265751838684,
      "learning_rate": 0.00018882357547148705,
      "loss": 3.9601,
      "step": 59950
    },
    {
      "epoch": 0.055892422002234626,
      "grad_norm": 0.9350675344467163,
      "learning_rate": 0.00018882171114939186,
      "loss": 3.9236,
      "step": 59960
    },
    {
      "epoch": 0.055901743620313714,
      "grad_norm": 1.0722696781158447,
      "learning_rate": 0.00018881984682729666,
      "loss": 3.8683,
      "step": 59970
    },
    {
      "epoch": 0.05591106523839281,
      "grad_norm": 1.0685757398605347,
      "learning_rate": 0.00018881798250520147,
      "loss": 4.1133,
      "step": 59980
    },
    {
      "epoch": 0.0559203868564719,
      "grad_norm": 1.0920809507369995,
      "learning_rate": 0.00018881611818310628,
      "loss": 3.7022,
      "step": 59990
    },
    {
      "epoch": 0.05592970847455099,
      "grad_norm": 1.0560280084609985,
      "learning_rate": 0.00018881425386101105,
      "loss": 3.9421,
      "step": 60000
    },
    {
      "epoch": 0.055939030092630085,
      "grad_norm": 0.9777536392211914,
      "learning_rate": 0.00018881238953891586,
      "loss": 3.9042,
      "step": 60010
    },
    {
      "epoch": 0.05594835171070918,
      "grad_norm": 1.2016016244888306,
      "learning_rate": 0.00018881052521682067,
      "loss": 3.8157,
      "step": 60020
    },
    {
      "epoch": 0.05595767332878827,
      "grad_norm": 1.0121428966522217,
      "learning_rate": 0.00018880866089472547,
      "loss": 3.9635,
      "step": 60030
    },
    {
      "epoch": 0.05596699494686736,
      "grad_norm": 1.1533452272415161,
      "learning_rate": 0.00018880679657263028,
      "loss": 4.1127,
      "step": 60040
    },
    {
      "epoch": 0.05597631656494645,
      "grad_norm": 1.1079139709472656,
      "learning_rate": 0.00018880493225053506,
      "loss": 3.9654,
      "step": 60050
    },
    {
      "epoch": 0.05598563818302554,
      "grad_norm": 1.0638397932052612,
      "learning_rate": 0.00018880306792843987,
      "loss": 3.9301,
      "step": 60060
    },
    {
      "epoch": 0.05599495980110464,
      "grad_norm": 0.9913224577903748,
      "learning_rate": 0.00018880120360634467,
      "loss": 3.9228,
      "step": 60070
    },
    {
      "epoch": 0.056004281419183725,
      "grad_norm": 1.094778060913086,
      "learning_rate": 0.00018879933928424948,
      "loss": 3.8914,
      "step": 60080
    },
    {
      "epoch": 0.05601360303726282,
      "grad_norm": 0.9220154285430908,
      "learning_rate": 0.00018879747496215428,
      "loss": 3.8921,
      "step": 60090
    },
    {
      "epoch": 0.05602292465534191,
      "grad_norm": 1.011051058769226,
      "learning_rate": 0.00018879561064005906,
      "loss": 3.8298,
      "step": 60100
    },
    {
      "epoch": 0.056032246273421,
      "grad_norm": 1.06732177734375,
      "learning_rate": 0.00018879374631796387,
      "loss": 3.9799,
      "step": 60110
    },
    {
      "epoch": 0.056041567891500096,
      "grad_norm": 0.9875416159629822,
      "learning_rate": 0.00018879188199586868,
      "loss": 3.8985,
      "step": 60120
    },
    {
      "epoch": 0.056050889509579184,
      "grad_norm": 1.0084065198898315,
      "learning_rate": 0.00018879001767377346,
      "loss": 3.9296,
      "step": 60130
    },
    {
      "epoch": 0.05606021112765828,
      "grad_norm": 1.080754041671753,
      "learning_rate": 0.0001887881533516783,
      "loss": 3.9318,
      "step": 60140
    },
    {
      "epoch": 0.05606953274573737,
      "grad_norm": 0.8944104313850403,
      "learning_rate": 0.00018878628902958307,
      "loss": 4.1504,
      "step": 60150
    },
    {
      "epoch": 0.05607885436381646,
      "grad_norm": 1.2129710912704468,
      "learning_rate": 0.00018878442470748787,
      "loss": 4.0018,
      "step": 60160
    },
    {
      "epoch": 0.056088175981895555,
      "grad_norm": 0.9731314778327942,
      "learning_rate": 0.00018878256038539268,
      "loss": 3.8393,
      "step": 60170
    },
    {
      "epoch": 0.05609749759997464,
      "grad_norm": 1.0847197771072388,
      "learning_rate": 0.00018878069606329746,
      "loss": 3.9206,
      "step": 60180
    },
    {
      "epoch": 0.05610681921805374,
      "grad_norm": 1.1031677722930908,
      "learning_rate": 0.00018877883174120227,
      "loss": 4.0889,
      "step": 60190
    },
    {
      "epoch": 0.05611614083613283,
      "grad_norm": 1.0950045585632324,
      "learning_rate": 0.00018877696741910707,
      "loss": 3.8609,
      "step": 60200
    },
    {
      "epoch": 0.05612546245421192,
      "grad_norm": 1.076724886894226,
      "learning_rate": 0.00018877510309701188,
      "loss": 3.9916,
      "step": 60210
    },
    {
      "epoch": 0.056134784072291014,
      "grad_norm": 0.8824140429496765,
      "learning_rate": 0.00018877323877491669,
      "loss": 3.8432,
      "step": 60220
    },
    {
      "epoch": 0.0561441056903701,
      "grad_norm": 1.0062214136123657,
      "learning_rate": 0.00018877137445282146,
      "loss": 3.9548,
      "step": 60230
    },
    {
      "epoch": 0.056153427308449196,
      "grad_norm": 1.034301519393921,
      "learning_rate": 0.00018876951013072627,
      "loss": 3.8661,
      "step": 60240
    },
    {
      "epoch": 0.05616274892652829,
      "grad_norm": 0.8764976263046265,
      "learning_rate": 0.00018876764580863108,
      "loss": 4.065,
      "step": 60250
    },
    {
      "epoch": 0.05617207054460738,
      "grad_norm": 1.347791075706482,
      "learning_rate": 0.00018876578148653588,
      "loss": 3.9084,
      "step": 60260
    },
    {
      "epoch": 0.05618139216268647,
      "grad_norm": 1.264397144317627,
      "learning_rate": 0.0001887639171644407,
      "loss": 3.9602,
      "step": 60270
    },
    {
      "epoch": 0.05619071378076557,
      "grad_norm": 1.0763683319091797,
      "learning_rate": 0.00018876205284234547,
      "loss": 4.0459,
      "step": 60280
    },
    {
      "epoch": 0.056200035398844654,
      "grad_norm": 0.950629472732544,
      "learning_rate": 0.00018876018852025028,
      "loss": 4.0929,
      "step": 60290
    },
    {
      "epoch": 0.05620935701692375,
      "grad_norm": 1.081140160560608,
      "learning_rate": 0.00018875832419815505,
      "loss": 4.0167,
      "step": 60300
    },
    {
      "epoch": 0.056218678635002836,
      "grad_norm": 1.1280794143676758,
      "learning_rate": 0.0001887564598760599,
      "loss": 4.0167,
      "step": 60310
    },
    {
      "epoch": 0.05622800025308193,
      "grad_norm": 1.0212527513504028,
      "learning_rate": 0.0001887545955539647,
      "loss": 4.0262,
      "step": 60320
    },
    {
      "epoch": 0.056237321871161025,
      "grad_norm": 1.0018975734710693,
      "learning_rate": 0.00018875273123186947,
      "loss": 3.9022,
      "step": 60330
    },
    {
      "epoch": 0.05624664348924011,
      "grad_norm": 1.0796608924865723,
      "learning_rate": 0.00018875086690977428,
      "loss": 3.9289,
      "step": 60340
    },
    {
      "epoch": 0.05625596510731921,
      "grad_norm": 1.025428295135498,
      "learning_rate": 0.00018874900258767906,
      "loss": 3.9476,
      "step": 60350
    },
    {
      "epoch": 0.0562652867253983,
      "grad_norm": 0.990950882434845,
      "learning_rate": 0.00018874713826558387,
      "loss": 3.7503,
      "step": 60360
    },
    {
      "epoch": 0.05627460834347739,
      "grad_norm": 0.935235321521759,
      "learning_rate": 0.0001887452739434887,
      "loss": 3.9282,
      "step": 60370
    },
    {
      "epoch": 0.056283929961556484,
      "grad_norm": 1.0638915300369263,
      "learning_rate": 0.00018874340962139348,
      "loss": 3.8356,
      "step": 60380
    },
    {
      "epoch": 0.05629325157963557,
      "grad_norm": 1.1864280700683594,
      "learning_rate": 0.00018874154529929828,
      "loss": 3.8216,
      "step": 60390
    },
    {
      "epoch": 0.056302573197714666,
      "grad_norm": 0.9990113377571106,
      "learning_rate": 0.00018873968097720306,
      "loss": 3.7322,
      "step": 60400
    },
    {
      "epoch": 0.05631189481579376,
      "grad_norm": 1.1028528213500977,
      "learning_rate": 0.00018873781665510787,
      "loss": 4.0002,
      "step": 60410
    },
    {
      "epoch": 0.05632121643387285,
      "grad_norm": 0.9341080784797668,
      "learning_rate": 0.00018873595233301268,
      "loss": 3.8296,
      "step": 60420
    },
    {
      "epoch": 0.05633053805195194,
      "grad_norm": 1.1296148300170898,
      "learning_rate": 0.00018873408801091748,
      "loss": 3.8988,
      "step": 60430
    },
    {
      "epoch": 0.05633985967003103,
      "grad_norm": 1.2845932245254517,
      "learning_rate": 0.0001887322236888223,
      "loss": 3.7954,
      "step": 60440
    },
    {
      "epoch": 0.056349181288110124,
      "grad_norm": 1.3257935047149658,
      "learning_rate": 0.00018873035936672707,
      "loss": 3.9483,
      "step": 60450
    },
    {
      "epoch": 0.05635850290618922,
      "grad_norm": 1.05471670627594,
      "learning_rate": 0.00018872849504463187,
      "loss": 3.9405,
      "step": 60460
    },
    {
      "epoch": 0.056367824524268306,
      "grad_norm": 1.117624044418335,
      "learning_rate": 0.00018872663072253668,
      "loss": 3.9712,
      "step": 60470
    },
    {
      "epoch": 0.0563771461423474,
      "grad_norm": 0.9003488421440125,
      "learning_rate": 0.0001887247664004415,
      "loss": 4.0969,
      "step": 60480
    },
    {
      "epoch": 0.056386467760426495,
      "grad_norm": 1.039901614189148,
      "learning_rate": 0.0001887229020783463,
      "loss": 3.9892,
      "step": 60490
    },
    {
      "epoch": 0.05639578937850558,
      "grad_norm": 1.143051028251648,
      "learning_rate": 0.00018872103775625107,
      "loss": 3.9479,
      "step": 60500
    },
    {
      "epoch": 0.05640511099658468,
      "grad_norm": 1.2498266696929932,
      "learning_rate": 0.00018871917343415588,
      "loss": 3.8687,
      "step": 60510
    },
    {
      "epoch": 0.056414432614663765,
      "grad_norm": 1.214375376701355,
      "learning_rate": 0.00018871730911206068,
      "loss": 4.0108,
      "step": 60520
    },
    {
      "epoch": 0.05642375423274286,
      "grad_norm": 0.9863496422767639,
      "learning_rate": 0.0001887154447899655,
      "loss": 3.8149,
      "step": 60530
    },
    {
      "epoch": 0.056433075850821954,
      "grad_norm": 1.0645756721496582,
      "learning_rate": 0.0001887135804678703,
      "loss": 4.0328,
      "step": 60540
    },
    {
      "epoch": 0.05644239746890104,
      "grad_norm": 1.2363455295562744,
      "learning_rate": 0.00018871171614577508,
      "loss": 3.7879,
      "step": 60550
    },
    {
      "epoch": 0.056451719086980136,
      "grad_norm": 1.290746808052063,
      "learning_rate": 0.00018870985182367988,
      "loss": 3.7598,
      "step": 60560
    },
    {
      "epoch": 0.05646104070505922,
      "grad_norm": 1.0906598567962646,
      "learning_rate": 0.0001887079875015847,
      "loss": 3.8864,
      "step": 60570
    },
    {
      "epoch": 0.05647036232313832,
      "grad_norm": 1.0442672967910767,
      "learning_rate": 0.00018870612317948947,
      "loss": 3.7181,
      "step": 60580
    },
    {
      "epoch": 0.05647968394121741,
      "grad_norm": 0.9159005284309387,
      "learning_rate": 0.0001887042588573943,
      "loss": 4.0026,
      "step": 60590
    },
    {
      "epoch": 0.0564890055592965,
      "grad_norm": 1.1092381477355957,
      "learning_rate": 0.00018870239453529908,
      "loss": 3.9013,
      "step": 60600
    },
    {
      "epoch": 0.056498327177375594,
      "grad_norm": 1.0807442665100098,
      "learning_rate": 0.0001887005302132039,
      "loss": 4.0942,
      "step": 60610
    },
    {
      "epoch": 0.05650764879545469,
      "grad_norm": 0.8109168410301208,
      "learning_rate": 0.0001886986658911087,
      "loss": 3.8254,
      "step": 60620
    },
    {
      "epoch": 0.056516970413533776,
      "grad_norm": 1.0992950201034546,
      "learning_rate": 0.00018869680156901347,
      "loss": 3.8074,
      "step": 60630
    },
    {
      "epoch": 0.05652629203161287,
      "grad_norm": 1.142156720161438,
      "learning_rate": 0.00018869493724691828,
      "loss": 3.8446,
      "step": 60640
    },
    {
      "epoch": 0.05653561364969196,
      "grad_norm": 1.023116111755371,
      "learning_rate": 0.00018869307292482309,
      "loss": 4.024,
      "step": 60650
    },
    {
      "epoch": 0.05654493526777105,
      "grad_norm": 0.9769307971000671,
      "learning_rate": 0.0001886912086027279,
      "loss": 3.9548,
      "step": 60660
    },
    {
      "epoch": 0.05655425688585015,
      "grad_norm": 1.0790263414382935,
      "learning_rate": 0.0001886893442806327,
      "loss": 4.0808,
      "step": 60670
    },
    {
      "epoch": 0.056563578503929235,
      "grad_norm": 1.1627522706985474,
      "learning_rate": 0.00018868747995853748,
      "loss": 3.7869,
      "step": 60680
    },
    {
      "epoch": 0.05657290012200833,
      "grad_norm": 1.2565865516662598,
      "learning_rate": 0.00018868561563644228,
      "loss": 3.89,
      "step": 60690
    },
    {
      "epoch": 0.056582221740087424,
      "grad_norm": 1.119669795036316,
      "learning_rate": 0.0001886837513143471,
      "loss": 3.9957,
      "step": 60700
    },
    {
      "epoch": 0.05659154335816651,
      "grad_norm": 1.1159402132034302,
      "learning_rate": 0.0001886818869922519,
      "loss": 4.0533,
      "step": 60710
    },
    {
      "epoch": 0.056600864976245606,
      "grad_norm": 1.1105173826217651,
      "learning_rate": 0.0001886800226701567,
      "loss": 3.7988,
      "step": 60720
    },
    {
      "epoch": 0.05661018659432469,
      "grad_norm": 0.884002685546875,
      "learning_rate": 0.00018867815834806148,
      "loss": 4.0097,
      "step": 60730
    },
    {
      "epoch": 0.05661950821240379,
      "grad_norm": 1.1589065790176392,
      "learning_rate": 0.0001886762940259663,
      "loss": 3.9563,
      "step": 60740
    },
    {
      "epoch": 0.05662882983048288,
      "grad_norm": 1.0079007148742676,
      "learning_rate": 0.00018867442970387107,
      "loss": 3.9454,
      "step": 60750
    },
    {
      "epoch": 0.05663815144856197,
      "grad_norm": 1.0438591241836548,
      "learning_rate": 0.0001886725653817759,
      "loss": 3.7751,
      "step": 60760
    },
    {
      "epoch": 0.056647473066641064,
      "grad_norm": 1.0541142225265503,
      "learning_rate": 0.0001886707010596807,
      "loss": 4.0243,
      "step": 60770
    },
    {
      "epoch": 0.05665679468472015,
      "grad_norm": 1.091943621635437,
      "learning_rate": 0.00018866883673758549,
      "loss": 3.8922,
      "step": 60780
    },
    {
      "epoch": 0.056666116302799246,
      "grad_norm": 1.0824005603790283,
      "learning_rate": 0.0001886669724154903,
      "loss": 3.964,
      "step": 60790
    },
    {
      "epoch": 0.05667543792087834,
      "grad_norm": 0.9620575904846191,
      "learning_rate": 0.00018866510809339507,
      "loss": 3.9525,
      "step": 60800
    },
    {
      "epoch": 0.05668475953895743,
      "grad_norm": 1.1186827421188354,
      "learning_rate": 0.00018866324377129988,
      "loss": 4.005,
      "step": 60810
    },
    {
      "epoch": 0.05669408115703652,
      "grad_norm": 1.0995432138442993,
      "learning_rate": 0.00018866137944920468,
      "loss": 3.9395,
      "step": 60820
    },
    {
      "epoch": 0.05670340277511562,
      "grad_norm": 1.0693303346633911,
      "learning_rate": 0.0001886595151271095,
      "loss": 4.0123,
      "step": 60830
    },
    {
      "epoch": 0.056712724393194705,
      "grad_norm": 1.0847256183624268,
      "learning_rate": 0.0001886576508050143,
      "loss": 4.0026,
      "step": 60840
    },
    {
      "epoch": 0.0567220460112738,
      "grad_norm": 0.9889903664588928,
      "learning_rate": 0.00018865578648291908,
      "loss": 4.1098,
      "step": 60850
    },
    {
      "epoch": 0.05673136762935289,
      "grad_norm": 1.0381355285644531,
      "learning_rate": 0.00018865392216082388,
      "loss": 3.9954,
      "step": 60860
    },
    {
      "epoch": 0.05674068924743198,
      "grad_norm": 1.0741361379623413,
      "learning_rate": 0.0001886520578387287,
      "loss": 4.0479,
      "step": 60870
    },
    {
      "epoch": 0.056750010865511076,
      "grad_norm": 1.1611438989639282,
      "learning_rate": 0.0001886501935166335,
      "loss": 3.8004,
      "step": 60880
    },
    {
      "epoch": 0.056759332483590164,
      "grad_norm": 1.0828959941864014,
      "learning_rate": 0.0001886483291945383,
      "loss": 3.9946,
      "step": 60890
    },
    {
      "epoch": 0.05676865410166926,
      "grad_norm": 1.0610613822937012,
      "learning_rate": 0.00018864646487244308,
      "loss": 3.9242,
      "step": 60900
    },
    {
      "epoch": 0.056777975719748346,
      "grad_norm": 1.0504252910614014,
      "learning_rate": 0.0001886446005503479,
      "loss": 3.9694,
      "step": 60910
    },
    {
      "epoch": 0.05678729733782744,
      "grad_norm": 1.3273602724075317,
      "learning_rate": 0.0001886427362282527,
      "loss": 4.0805,
      "step": 60920
    },
    {
      "epoch": 0.056796618955906535,
      "grad_norm": 1.3014034032821655,
      "learning_rate": 0.0001886408719061575,
      "loss": 3.7998,
      "step": 60930
    },
    {
      "epoch": 0.05680594057398562,
      "grad_norm": 1.1455399990081787,
      "learning_rate": 0.0001886390075840623,
      "loss": 4.0819,
      "step": 60940
    },
    {
      "epoch": 0.05681526219206472,
      "grad_norm": 1.0094019174575806,
      "learning_rate": 0.00018863714326196708,
      "loss": 3.932,
      "step": 60950
    },
    {
      "epoch": 0.05682458381014381,
      "grad_norm": 1.147516131401062,
      "learning_rate": 0.0001886352789398719,
      "loss": 3.7883,
      "step": 60960
    },
    {
      "epoch": 0.0568339054282229,
      "grad_norm": 1.0805338621139526,
      "learning_rate": 0.00018863341461777667,
      "loss": 3.8815,
      "step": 60970
    },
    {
      "epoch": 0.05684322704630199,
      "grad_norm": 1.1531734466552734,
      "learning_rate": 0.0001886315502956815,
      "loss": 3.7818,
      "step": 60980
    },
    {
      "epoch": 0.05685254866438108,
      "grad_norm": 0.9635171294212341,
      "learning_rate": 0.0001886296859735863,
      "loss": 3.9094,
      "step": 60990
    },
    {
      "epoch": 0.056861870282460175,
      "grad_norm": 1.1432726383209229,
      "learning_rate": 0.0001886278216514911,
      "loss": 4.0496,
      "step": 61000
    },
    {
      "epoch": 0.05687119190053927,
      "grad_norm": 1.001057744026184,
      "learning_rate": 0.0001886259573293959,
      "loss": 3.7854,
      "step": 61010
    },
    {
      "epoch": 0.05688051351861836,
      "grad_norm": 1.2847130298614502,
      "learning_rate": 0.00018862409300730067,
      "loss": 3.9444,
      "step": 61020
    },
    {
      "epoch": 0.05688983513669745,
      "grad_norm": 1.181964635848999,
      "learning_rate": 0.00018862222868520548,
      "loss": 3.9791,
      "step": 61030
    },
    {
      "epoch": 0.056899156754776546,
      "grad_norm": 0.9564380049705505,
      "learning_rate": 0.00018862036436311031,
      "loss": 3.9647,
      "step": 61040
    },
    {
      "epoch": 0.056908478372855634,
      "grad_norm": 0.7805046439170837,
      "learning_rate": 0.0001886185000410151,
      "loss": 3.9632,
      "step": 61050
    },
    {
      "epoch": 0.05691779999093473,
      "grad_norm": 0.8958951830863953,
      "learning_rate": 0.0001886166357189199,
      "loss": 3.8798,
      "step": 61060
    },
    {
      "epoch": 0.056927121609013816,
      "grad_norm": 1.0127359628677368,
      "learning_rate": 0.00018861477139682468,
      "loss": 3.9191,
      "step": 61070
    },
    {
      "epoch": 0.05693644322709291,
      "grad_norm": 1.1073129177093506,
      "learning_rate": 0.00018861290707472949,
      "loss": 4.0415,
      "step": 61080
    },
    {
      "epoch": 0.056945764845172005,
      "grad_norm": 1.1520555019378662,
      "learning_rate": 0.0001886110427526343,
      "loss": 4.1682,
      "step": 61090
    },
    {
      "epoch": 0.05695508646325109,
      "grad_norm": 1.0909374952316284,
      "learning_rate": 0.0001886091784305391,
      "loss": 4.0427,
      "step": 61100
    },
    {
      "epoch": 0.05696440808133019,
      "grad_norm": 0.8848010897636414,
      "learning_rate": 0.0001886073141084439,
      "loss": 3.981,
      "step": 61110
    },
    {
      "epoch": 0.056973729699409274,
      "grad_norm": 1.1411880254745483,
      "learning_rate": 0.00018860544978634868,
      "loss": 3.8957,
      "step": 61120
    },
    {
      "epoch": 0.05698305131748837,
      "grad_norm": 1.0913642644882202,
      "learning_rate": 0.0001886035854642535,
      "loss": 3.8183,
      "step": 61130
    },
    {
      "epoch": 0.05699237293556746,
      "grad_norm": 1.015608787536621,
      "learning_rate": 0.0001886017211421583,
      "loss": 3.9811,
      "step": 61140
    },
    {
      "epoch": 0.05700169455364655,
      "grad_norm": 1.1276772022247314,
      "learning_rate": 0.0001885998568200631,
      "loss": 3.9015,
      "step": 61150
    },
    {
      "epoch": 0.057011016171725645,
      "grad_norm": 0.9656205177307129,
      "learning_rate": 0.0001885979924979679,
      "loss": 3.7583,
      "step": 61160
    },
    {
      "epoch": 0.05702033778980474,
      "grad_norm": 1.19890296459198,
      "learning_rate": 0.0001885961281758727,
      "loss": 3.828,
      "step": 61170
    },
    {
      "epoch": 0.05702965940788383,
      "grad_norm": 0.9727188348770142,
      "learning_rate": 0.0001885942638537775,
      "loss": 4.093,
      "step": 61180
    },
    {
      "epoch": 0.05703898102596292,
      "grad_norm": 0.8884298205375671,
      "learning_rate": 0.0001885923995316823,
      "loss": 3.8568,
      "step": 61190
    },
    {
      "epoch": 0.05704830264404201,
      "grad_norm": 1.0604300498962402,
      "learning_rate": 0.0001885905352095871,
      "loss": 3.8325,
      "step": 61200
    },
    {
      "epoch": 0.057057624262121104,
      "grad_norm": 1.2411597967147827,
      "learning_rate": 0.0001885886708874919,
      "loss": 4.0046,
      "step": 61210
    },
    {
      "epoch": 0.0570669458802002,
      "grad_norm": 1.023495078086853,
      "learning_rate": 0.0001885868065653967,
      "loss": 3.9944,
      "step": 61220
    },
    {
      "epoch": 0.057076267498279286,
      "grad_norm": 1.1587008237838745,
      "learning_rate": 0.0001885849422433015,
      "loss": 3.9171,
      "step": 61230
    },
    {
      "epoch": 0.05708558911635838,
      "grad_norm": 1.1676909923553467,
      "learning_rate": 0.0001885830779212063,
      "loss": 3.6241,
      "step": 61240
    },
    {
      "epoch": 0.05709491073443747,
      "grad_norm": 1.1249818801879883,
      "learning_rate": 0.00018858121359911108,
      "loss": 4.2045,
      "step": 61250
    },
    {
      "epoch": 0.05710423235251656,
      "grad_norm": 1.0747138261795044,
      "learning_rate": 0.0001885793492770159,
      "loss": 4.0438,
      "step": 61260
    },
    {
      "epoch": 0.05711355397059566,
      "grad_norm": 0.9165562391281128,
      "learning_rate": 0.0001885774849549207,
      "loss": 3.9007,
      "step": 61270
    },
    {
      "epoch": 0.057122875588674744,
      "grad_norm": 1.161377191543579,
      "learning_rate": 0.0001885756206328255,
      "loss": 4.0289,
      "step": 61280
    },
    {
      "epoch": 0.05713219720675384,
      "grad_norm": 1.1275168657302856,
      "learning_rate": 0.0001885737563107303,
      "loss": 4.0261,
      "step": 61290
    },
    {
      "epoch": 0.05714151882483293,
      "grad_norm": 1.049418568611145,
      "learning_rate": 0.0001885718919886351,
      "loss": 3.8098,
      "step": 61300
    },
    {
      "epoch": 0.05715084044291202,
      "grad_norm": 1.1360112428665161,
      "learning_rate": 0.0001885700276665399,
      "loss": 3.8553,
      "step": 61310
    },
    {
      "epoch": 0.057160162060991115,
      "grad_norm": 1.244049310684204,
      "learning_rate": 0.0001885681633444447,
      "loss": 3.8575,
      "step": 61320
    },
    {
      "epoch": 0.0571694836790702,
      "grad_norm": 0.8051106929779053,
      "learning_rate": 0.0001885662990223495,
      "loss": 3.9,
      "step": 61330
    },
    {
      "epoch": 0.0571788052971493,
      "grad_norm": 1.1673156023025513,
      "learning_rate": 0.00018856443470025431,
      "loss": 3.7774,
      "step": 61340
    },
    {
      "epoch": 0.05718812691522839,
      "grad_norm": 0.8899871706962585,
      "learning_rate": 0.0001885625703781591,
      "loss": 3.8518,
      "step": 61350
    },
    {
      "epoch": 0.05719744853330748,
      "grad_norm": 1.1372958421707153,
      "learning_rate": 0.0001885607060560639,
      "loss": 4.0417,
      "step": 61360
    },
    {
      "epoch": 0.057206770151386574,
      "grad_norm": 0.857760488986969,
      "learning_rate": 0.0001885588417339687,
      "loss": 3.8455,
      "step": 61370
    },
    {
      "epoch": 0.05721609176946567,
      "grad_norm": 1.0523592233657837,
      "learning_rate": 0.0001885569774118735,
      "loss": 4.1199,
      "step": 61380
    },
    {
      "epoch": 0.057225413387544756,
      "grad_norm": 1.1190837621688843,
      "learning_rate": 0.00018855511308977832,
      "loss": 3.8686,
      "step": 61390
    },
    {
      "epoch": 0.05723473500562385,
      "grad_norm": 1.0510231256484985,
      "learning_rate": 0.0001885532487676831,
      "loss": 3.8541,
      "step": 61400
    },
    {
      "epoch": 0.05724405662370294,
      "grad_norm": 1.076595664024353,
      "learning_rate": 0.0001885513844455879,
      "loss": 4.0496,
      "step": 61410
    },
    {
      "epoch": 0.05725337824178203,
      "grad_norm": 1.1644556522369385,
      "learning_rate": 0.00018854952012349268,
      "loss": 3.9241,
      "step": 61420
    },
    {
      "epoch": 0.05726269985986113,
      "grad_norm": 1.0757782459259033,
      "learning_rate": 0.00018854765580139752,
      "loss": 3.643,
      "step": 61430
    },
    {
      "epoch": 0.057272021477940215,
      "grad_norm": 1.107399582862854,
      "learning_rate": 0.00018854579147930232,
      "loss": 3.844,
      "step": 61440
    },
    {
      "epoch": 0.05728134309601931,
      "grad_norm": 1.3226611614227295,
      "learning_rate": 0.0001885439271572071,
      "loss": 3.7881,
      "step": 61450
    },
    {
      "epoch": 0.0572906647140984,
      "grad_norm": 1.0962920188903809,
      "learning_rate": 0.0001885420628351119,
      "loss": 3.8866,
      "step": 61460
    },
    {
      "epoch": 0.05729998633217749,
      "grad_norm": 1.4047776460647583,
      "learning_rate": 0.0001885401985130167,
      "loss": 3.7995,
      "step": 61470
    },
    {
      "epoch": 0.057309307950256586,
      "grad_norm": 1.3609732389450073,
      "learning_rate": 0.0001885383341909215,
      "loss": 3.8902,
      "step": 61480
    },
    {
      "epoch": 0.05731862956833567,
      "grad_norm": 0.9307029843330383,
      "learning_rate": 0.00018853646986882633,
      "loss": 3.8383,
      "step": 61490
    },
    {
      "epoch": 0.05732795118641477,
      "grad_norm": 1.3052703142166138,
      "learning_rate": 0.0001885346055467311,
      "loss": 3.8383,
      "step": 61500
    },
    {
      "epoch": 0.05733727280449386,
      "grad_norm": 1.0852152109146118,
      "learning_rate": 0.0001885327412246359,
      "loss": 3.9995,
      "step": 61510
    },
    {
      "epoch": 0.05734659442257295,
      "grad_norm": 0.9323749542236328,
      "learning_rate": 0.0001885308769025407,
      "loss": 4.0373,
      "step": 61520
    },
    {
      "epoch": 0.057355916040652044,
      "grad_norm": 1.1216589212417603,
      "learning_rate": 0.0001885290125804455,
      "loss": 3.8868,
      "step": 61530
    },
    {
      "epoch": 0.05736523765873113,
      "grad_norm": 1.0316311120986938,
      "learning_rate": 0.0001885271482583503,
      "loss": 3.9448,
      "step": 61540
    },
    {
      "epoch": 0.057374559276810226,
      "grad_norm": 1.0608278512954712,
      "learning_rate": 0.0001885252839362551,
      "loss": 3.8079,
      "step": 61550
    },
    {
      "epoch": 0.05738388089488932,
      "grad_norm": 1.1968145370483398,
      "learning_rate": 0.00018852341961415992,
      "loss": 3.946,
      "step": 61560
    },
    {
      "epoch": 0.05739320251296841,
      "grad_norm": 1.0830934047698975,
      "learning_rate": 0.0001885215552920647,
      "loss": 3.8365,
      "step": 61570
    },
    {
      "epoch": 0.0574025241310475,
      "grad_norm": 1.0579886436462402,
      "learning_rate": 0.0001885196909699695,
      "loss": 3.8923,
      "step": 61580
    },
    {
      "epoch": 0.05741184574912659,
      "grad_norm": 0.9528774619102478,
      "learning_rate": 0.0001885178266478743,
      "loss": 3.7786,
      "step": 61590
    },
    {
      "epoch": 0.057421167367205685,
      "grad_norm": 1.0788028240203857,
      "learning_rate": 0.00018851596232577912,
      "loss": 3.8441,
      "step": 61600
    },
    {
      "epoch": 0.05743048898528478,
      "grad_norm": 1.1216036081314087,
      "learning_rate": 0.00018851409800368392,
      "loss": 3.8654,
      "step": 61610
    },
    {
      "epoch": 0.05743981060336387,
      "grad_norm": 0.9960892200469971,
      "learning_rate": 0.0001885122336815887,
      "loss": 3.9227,
      "step": 61620
    },
    {
      "epoch": 0.05744913222144296,
      "grad_norm": 1.140289545059204,
      "learning_rate": 0.0001885103693594935,
      "loss": 3.951,
      "step": 61630
    },
    {
      "epoch": 0.057458453839522056,
      "grad_norm": 1.0844879150390625,
      "learning_rate": 0.00018850850503739831,
      "loss": 3.9024,
      "step": 61640
    },
    {
      "epoch": 0.05746777545760114,
      "grad_norm": 0.951602578163147,
      "learning_rate": 0.00018850664071530312,
      "loss": 3.8137,
      "step": 61650
    },
    {
      "epoch": 0.05747709707568024,
      "grad_norm": 1.1128751039505005,
      "learning_rate": 0.00018850477639320793,
      "loss": 3.8201,
      "step": 61660
    },
    {
      "epoch": 0.057486418693759325,
      "grad_norm": 1.061781406402588,
      "learning_rate": 0.0001885029120711127,
      "loss": 3.7068,
      "step": 61670
    },
    {
      "epoch": 0.05749574031183842,
      "grad_norm": 1.2082520723342896,
      "learning_rate": 0.0001885010477490175,
      "loss": 3.9015,
      "step": 61680
    },
    {
      "epoch": 0.057505061929917514,
      "grad_norm": 1.125262975692749,
      "learning_rate": 0.00018849918342692232,
      "loss": 4.1859,
      "step": 61690
    },
    {
      "epoch": 0.0575143835479966,
      "grad_norm": 1.0493849515914917,
      "learning_rate": 0.0001884973191048271,
      "loss": 4.0053,
      "step": 61700
    },
    {
      "epoch": 0.057523705166075696,
      "grad_norm": 1.1354519128799438,
      "learning_rate": 0.00018849545478273193,
      "loss": 3.7383,
      "step": 61710
    },
    {
      "epoch": 0.05753302678415479,
      "grad_norm": 0.9989332556724548,
      "learning_rate": 0.0001884935904606367,
      "loss": 4.0735,
      "step": 61720
    },
    {
      "epoch": 0.05754234840223388,
      "grad_norm": 1.0029746294021606,
      "learning_rate": 0.00018849172613854152,
      "loss": 4.0659,
      "step": 61730
    },
    {
      "epoch": 0.05755167002031297,
      "grad_norm": 0.9892446398735046,
      "learning_rate": 0.00018848986181644632,
      "loss": 4.111,
      "step": 61740
    },
    {
      "epoch": 0.05756099163839206,
      "grad_norm": 1.0262112617492676,
      "learning_rate": 0.0001884879974943511,
      "loss": 3.9739,
      "step": 61750
    },
    {
      "epoch": 0.057570313256471155,
      "grad_norm": 0.9880075454711914,
      "learning_rate": 0.0001884861331722559,
      "loss": 3.6526,
      "step": 61760
    },
    {
      "epoch": 0.05757963487455025,
      "grad_norm": 1.0054445266723633,
      "learning_rate": 0.00018848426885016071,
      "loss": 3.8002,
      "step": 61770
    },
    {
      "epoch": 0.05758895649262934,
      "grad_norm": 1.0293489694595337,
      "learning_rate": 0.00018848240452806552,
      "loss": 3.7886,
      "step": 61780
    },
    {
      "epoch": 0.05759827811070843,
      "grad_norm": 1.0815824270248413,
      "learning_rate": 0.00018848054020597033,
      "loss": 3.7815,
      "step": 61790
    },
    {
      "epoch": 0.05760759972878752,
      "grad_norm": 1.074552059173584,
      "learning_rate": 0.0001884786758838751,
      "loss": 4.0169,
      "step": 61800
    },
    {
      "epoch": 0.05761692134686661,
      "grad_norm": 1.111579418182373,
      "learning_rate": 0.0001884768115617799,
      "loss": 3.9433,
      "step": 61810
    },
    {
      "epoch": 0.05762624296494571,
      "grad_norm": 0.9639065265655518,
      "learning_rate": 0.00018847494723968472,
      "loss": 3.7354,
      "step": 61820
    },
    {
      "epoch": 0.057635564583024795,
      "grad_norm": 1.2207225561141968,
      "learning_rate": 0.00018847308291758953,
      "loss": 3.7768,
      "step": 61830
    },
    {
      "epoch": 0.05764488620110389,
      "grad_norm": 1.016951322555542,
      "learning_rate": 0.00018847121859549433,
      "loss": 4.0201,
      "step": 61840
    },
    {
      "epoch": 0.057654207819182984,
      "grad_norm": 1.2689944505691528,
      "learning_rate": 0.0001884693542733991,
      "loss": 3.8782,
      "step": 61850
    },
    {
      "epoch": 0.05766352943726207,
      "grad_norm": 1.193142056465149,
      "learning_rate": 0.00018846748995130392,
      "loss": 3.8881,
      "step": 61860
    },
    {
      "epoch": 0.057672851055341166,
      "grad_norm": 1.1170518398284912,
      "learning_rate": 0.0001884656256292087,
      "loss": 3.8254,
      "step": 61870
    },
    {
      "epoch": 0.057682172673420254,
      "grad_norm": 0.9378699660301208,
      "learning_rate": 0.00018846376130711353,
      "loss": 3.9605,
      "step": 61880
    },
    {
      "epoch": 0.05769149429149935,
      "grad_norm": 1.195677399635315,
      "learning_rate": 0.00018846189698501834,
      "loss": 4.0136,
      "step": 61890
    },
    {
      "epoch": 0.05770081590957844,
      "grad_norm": 0.9986094832420349,
      "learning_rate": 0.00018846003266292312,
      "loss": 3.9726,
      "step": 61900
    },
    {
      "epoch": 0.05771013752765753,
      "grad_norm": 1.0276505947113037,
      "learning_rate": 0.00018845816834082792,
      "loss": 3.9738,
      "step": 61910
    },
    {
      "epoch": 0.057719459145736625,
      "grad_norm": 1.1794102191925049,
      "learning_rate": 0.0001884563040187327,
      "loss": 3.9868,
      "step": 61920
    },
    {
      "epoch": 0.05772878076381571,
      "grad_norm": 1.0896258354187012,
      "learning_rate": 0.0001884544396966375,
      "loss": 3.9351,
      "step": 61930
    },
    {
      "epoch": 0.05773810238189481,
      "grad_norm": 1.1899709701538086,
      "learning_rate": 0.00018845257537454234,
      "loss": 3.907,
      "step": 61940
    },
    {
      "epoch": 0.0577474239999739,
      "grad_norm": 1.0860917568206787,
      "learning_rate": 0.00018845071105244712,
      "loss": 4.0669,
      "step": 61950
    },
    {
      "epoch": 0.05775674561805299,
      "grad_norm": 1.027166485786438,
      "learning_rate": 0.00018844884673035193,
      "loss": 3.7036,
      "step": 61960
    },
    {
      "epoch": 0.05776606723613208,
      "grad_norm": 0.9817315936088562,
      "learning_rate": 0.0001884469824082567,
      "loss": 3.8872,
      "step": 61970
    },
    {
      "epoch": 0.05777538885421118,
      "grad_norm": 1.1836395263671875,
      "learning_rate": 0.0001884451180861615,
      "loss": 3.607,
      "step": 61980
    },
    {
      "epoch": 0.057784710472290265,
      "grad_norm": 0.9876450896263123,
      "learning_rate": 0.00018844325376406632,
      "loss": 3.7413,
      "step": 61990
    },
    {
      "epoch": 0.05779403209036936,
      "grad_norm": 0.8835235238075256,
      "learning_rate": 0.00018844138944197112,
      "loss": 3.8859,
      "step": 62000
    },
    {
      "epoch": 0.05780335370844845,
      "grad_norm": 0.9733423590660095,
      "learning_rate": 0.00018843952511987593,
      "loss": 3.9361,
      "step": 62010
    },
    {
      "epoch": 0.05781267532652754,
      "grad_norm": 1.1358970403671265,
      "learning_rate": 0.0001884376607977807,
      "loss": 3.8885,
      "step": 62020
    },
    {
      "epoch": 0.057821996944606636,
      "grad_norm": 1.1885688304901123,
      "learning_rate": 0.00018843579647568552,
      "loss": 3.8437,
      "step": 62030
    },
    {
      "epoch": 0.057831318562685724,
      "grad_norm": 1.2119358777999878,
      "learning_rate": 0.00018843393215359032,
      "loss": 3.8072,
      "step": 62040
    },
    {
      "epoch": 0.05784064018076482,
      "grad_norm": 1.2638394832611084,
      "learning_rate": 0.00018843206783149513,
      "loss": 3.8737,
      "step": 62050
    },
    {
      "epoch": 0.05784996179884391,
      "grad_norm": 1.3089932203292847,
      "learning_rate": 0.00018843020350939993,
      "loss": 3.8373,
      "step": 62060
    },
    {
      "epoch": 0.057859283416923,
      "grad_norm": 1.1601006984710693,
      "learning_rate": 0.00018842833918730471,
      "loss": 3.9538,
      "step": 62070
    },
    {
      "epoch": 0.057868605035002095,
      "grad_norm": 0.9282958507537842,
      "learning_rate": 0.00018842647486520952,
      "loss": 3.706,
      "step": 62080
    },
    {
      "epoch": 0.05787792665308118,
      "grad_norm": 1.1623162031173706,
      "learning_rate": 0.00018842461054311433,
      "loss": 3.8922,
      "step": 62090
    },
    {
      "epoch": 0.05788724827116028,
      "grad_norm": 0.8622530698776245,
      "learning_rate": 0.00018842274622101913,
      "loss": 4.0148,
      "step": 62100
    },
    {
      "epoch": 0.05789656988923937,
      "grad_norm": 1.1007235050201416,
      "learning_rate": 0.00018842088189892394,
      "loss": 3.8671,
      "step": 62110
    },
    {
      "epoch": 0.05790589150731846,
      "grad_norm": 1.0797008275985718,
      "learning_rate": 0.00018841901757682872,
      "loss": 3.8238,
      "step": 62120
    },
    {
      "epoch": 0.057915213125397554,
      "grad_norm": 0.8930692672729492,
      "learning_rate": 0.00018841715325473352,
      "loss": 3.9822,
      "step": 62130
    },
    {
      "epoch": 0.05792453474347664,
      "grad_norm": 1.1615172624588013,
      "learning_rate": 0.00018841528893263833,
      "loss": 3.8167,
      "step": 62140
    },
    {
      "epoch": 0.057933856361555736,
      "grad_norm": 0.9655539989471436,
      "learning_rate": 0.0001884134246105431,
      "loss": 4.0369,
      "step": 62150
    },
    {
      "epoch": 0.05794317797963483,
      "grad_norm": 0.9761672616004944,
      "learning_rate": 0.00018841156028844794,
      "loss": 3.9893,
      "step": 62160
    },
    {
      "epoch": 0.05795249959771392,
      "grad_norm": 1.0660837888717651,
      "learning_rate": 0.00018840969596635272,
      "loss": 3.9155,
      "step": 62170
    },
    {
      "epoch": 0.05796182121579301,
      "grad_norm": 1.1042170524597168,
      "learning_rate": 0.00018840783164425753,
      "loss": 4.0028,
      "step": 62180
    },
    {
      "epoch": 0.05797114283387211,
      "grad_norm": 0.9217171669006348,
      "learning_rate": 0.0001884059673221623,
      "loss": 3.8086,
      "step": 62190
    },
    {
      "epoch": 0.057980464451951194,
      "grad_norm": 1.0867875814437866,
      "learning_rate": 0.00018840410300006711,
      "loss": 3.9074,
      "step": 62200
    },
    {
      "epoch": 0.05798978607003029,
      "grad_norm": 1.0432043075561523,
      "learning_rate": 0.00018840223867797192,
      "loss": 3.9493,
      "step": 62210
    },
    {
      "epoch": 0.057999107688109376,
      "grad_norm": 1.1797646284103394,
      "learning_rate": 0.00018840037435587673,
      "loss": 3.9072,
      "step": 62220
    },
    {
      "epoch": 0.05800842930618847,
      "grad_norm": 1.0946828126907349,
      "learning_rate": 0.00018839851003378153,
      "loss": 3.6695,
      "step": 62230
    },
    {
      "epoch": 0.058017750924267565,
      "grad_norm": 1.0211211442947388,
      "learning_rate": 0.0001883966457116863,
      "loss": 3.8767,
      "step": 62240
    },
    {
      "epoch": 0.05802707254234665,
      "grad_norm": 0.9970557689666748,
      "learning_rate": 0.00018839478138959112,
      "loss": 4.054,
      "step": 62250
    },
    {
      "epoch": 0.05803639416042575,
      "grad_norm": 1.0742108821868896,
      "learning_rate": 0.00018839291706749593,
      "loss": 4.0613,
      "step": 62260
    },
    {
      "epoch": 0.058045715778504835,
      "grad_norm": 1.334840178489685,
      "learning_rate": 0.00018839105274540073,
      "loss": 3.9057,
      "step": 62270
    },
    {
      "epoch": 0.05805503739658393,
      "grad_norm": 1.4603590965270996,
      "learning_rate": 0.00018838918842330554,
      "loss": 4.064,
      "step": 62280
    },
    {
      "epoch": 0.058064359014663024,
      "grad_norm": 1.104934811592102,
      "learning_rate": 0.00018838732410121032,
      "loss": 3.911,
      "step": 62290
    },
    {
      "epoch": 0.05807368063274211,
      "grad_norm": 1.0402268171310425,
      "learning_rate": 0.00018838545977911512,
      "loss": 3.8451,
      "step": 62300
    },
    {
      "epoch": 0.058083002250821206,
      "grad_norm": 1.0502946376800537,
      "learning_rate": 0.00018838359545701993,
      "loss": 4.0069,
      "step": 62310
    },
    {
      "epoch": 0.0580923238689003,
      "grad_norm": 0.9272421598434448,
      "learning_rate": 0.0001883817311349247,
      "loss": 3.7046,
      "step": 62320
    },
    {
      "epoch": 0.05810164548697939,
      "grad_norm": 1.2270028591156006,
      "learning_rate": 0.00018837986681282954,
      "loss": 3.9204,
      "step": 62330
    },
    {
      "epoch": 0.05811096710505848,
      "grad_norm": 1.1873655319213867,
      "learning_rate": 0.00018837800249073432,
      "loss": 3.6405,
      "step": 62340
    },
    {
      "epoch": 0.05812028872313757,
      "grad_norm": 1.1144887208938599,
      "learning_rate": 0.00018837613816863913,
      "loss": 3.9615,
      "step": 62350
    },
    {
      "epoch": 0.058129610341216664,
      "grad_norm": 1.4505250453948975,
      "learning_rate": 0.00018837427384654393,
      "loss": 4.0027,
      "step": 62360
    },
    {
      "epoch": 0.05813893195929576,
      "grad_norm": 1.0998868942260742,
      "learning_rate": 0.0001883724095244487,
      "loss": 3.8948,
      "step": 62370
    },
    {
      "epoch": 0.058148253577374846,
      "grad_norm": 0.9552507996559143,
      "learning_rate": 0.00018837054520235352,
      "loss": 3.9109,
      "step": 62380
    },
    {
      "epoch": 0.05815757519545394,
      "grad_norm": 1.1129313707351685,
      "learning_rate": 0.00018836868088025833,
      "loss": 4.0641,
      "step": 62390
    },
    {
      "epoch": 0.058166896813533035,
      "grad_norm": 1.1090972423553467,
      "learning_rate": 0.00018836681655816313,
      "loss": 3.859,
      "step": 62400
    },
    {
      "epoch": 0.05817621843161212,
      "grad_norm": 0.9934571385383606,
      "learning_rate": 0.00018836495223606794,
      "loss": 3.8352,
      "step": 62410
    },
    {
      "epoch": 0.05818554004969122,
      "grad_norm": 0.9442775249481201,
      "learning_rate": 0.00018836308791397272,
      "loss": 3.9156,
      "step": 62420
    },
    {
      "epoch": 0.058194861667770305,
      "grad_norm": 1.264621615409851,
      "learning_rate": 0.00018836122359187752,
      "loss": 3.8949,
      "step": 62430
    },
    {
      "epoch": 0.0582041832858494,
      "grad_norm": 1.0764919519424438,
      "learning_rate": 0.00018835935926978233,
      "loss": 3.8021,
      "step": 62440
    },
    {
      "epoch": 0.058213504903928494,
      "grad_norm": 1.1256928443908691,
      "learning_rate": 0.00018835749494768714,
      "loss": 3.8805,
      "step": 62450
    },
    {
      "epoch": 0.05822282652200758,
      "grad_norm": 1.1527992486953735,
      "learning_rate": 0.00018835563062559194,
      "loss": 4.1415,
      "step": 62460
    },
    {
      "epoch": 0.058232148140086676,
      "grad_norm": 1.037805438041687,
      "learning_rate": 0.00018835376630349672,
      "loss": 3.9409,
      "step": 62470
    },
    {
      "epoch": 0.05824146975816576,
      "grad_norm": 1.0119115114212036,
      "learning_rate": 0.00018835190198140153,
      "loss": 3.9353,
      "step": 62480
    },
    {
      "epoch": 0.05825079137624486,
      "grad_norm": 1.0199213027954102,
      "learning_rate": 0.00018835003765930633,
      "loss": 3.8306,
      "step": 62490
    },
    {
      "epoch": 0.05826011299432395,
      "grad_norm": 1.0896172523498535,
      "learning_rate": 0.00018834817333721114,
      "loss": 3.798,
      "step": 62500
    },
    {
      "epoch": 0.05826943461240304,
      "grad_norm": 0.996046245098114,
      "learning_rate": 0.00018834630901511595,
      "loss": 3.9914,
      "step": 62510
    },
    {
      "epoch": 0.058278756230482134,
      "grad_norm": 1.1976919174194336,
      "learning_rate": 0.00018834444469302073,
      "loss": 3.8109,
      "step": 62520
    },
    {
      "epoch": 0.05828807784856123,
      "grad_norm": 0.9306544065475464,
      "learning_rate": 0.00018834258037092553,
      "loss": 4.075,
      "step": 62530
    },
    {
      "epoch": 0.058297399466640316,
      "grad_norm": 1.4532818794250488,
      "learning_rate": 0.0001883407160488303,
      "loss": 3.9667,
      "step": 62540
    },
    {
      "epoch": 0.05830672108471941,
      "grad_norm": 0.8473064303398132,
      "learning_rate": 0.00018833885172673515,
      "loss": 4.0534,
      "step": 62550
    },
    {
      "epoch": 0.0583160427027985,
      "grad_norm": 1.1794421672821045,
      "learning_rate": 0.00018833698740463995,
      "loss": 3.9955,
      "step": 62560
    },
    {
      "epoch": 0.05832536432087759,
      "grad_norm": 1.0604649782180786,
      "learning_rate": 0.00018833512308254473,
      "loss": 3.9838,
      "step": 62570
    },
    {
      "epoch": 0.05833468593895669,
      "grad_norm": 0.9556775093078613,
      "learning_rate": 0.00018833325876044954,
      "loss": 3.7378,
      "step": 62580
    },
    {
      "epoch": 0.058344007557035775,
      "grad_norm": 0.9391772747039795,
      "learning_rate": 0.00018833139443835432,
      "loss": 3.8253,
      "step": 62590
    },
    {
      "epoch": 0.05835332917511487,
      "grad_norm": 1.032774567604065,
      "learning_rate": 0.00018832953011625912,
      "loss": 3.8045,
      "step": 62600
    },
    {
      "epoch": 0.05836265079319396,
      "grad_norm": 1.0866450071334839,
      "learning_rate": 0.00018832766579416396,
      "loss": 3.7914,
      "step": 62610
    },
    {
      "epoch": 0.05837197241127305,
      "grad_norm": 1.047686219215393,
      "learning_rate": 0.00018832580147206874,
      "loss": 3.9533,
      "step": 62620
    },
    {
      "epoch": 0.058381294029352146,
      "grad_norm": 1.2185959815979004,
      "learning_rate": 0.00018832393714997354,
      "loss": 3.8111,
      "step": 62630
    },
    {
      "epoch": 0.058390615647431234,
      "grad_norm": 1.0638697147369385,
      "learning_rate": 0.00018832207282787832,
      "loss": 3.9519,
      "step": 62640
    },
    {
      "epoch": 0.05839993726551033,
      "grad_norm": 1.1595772504806519,
      "learning_rate": 0.00018832020850578313,
      "loss": 3.7716,
      "step": 62650
    },
    {
      "epoch": 0.05840925888358942,
      "grad_norm": 1.1642749309539795,
      "learning_rate": 0.00018831834418368793,
      "loss": 3.7837,
      "step": 62660
    },
    {
      "epoch": 0.05841858050166851,
      "grad_norm": 1.052459955215454,
      "learning_rate": 0.00018831647986159274,
      "loss": 3.8352,
      "step": 62670
    },
    {
      "epoch": 0.058427902119747605,
      "grad_norm": 1.0770022869110107,
      "learning_rate": 0.00018831461553949755,
      "loss": 4.0039,
      "step": 62680
    },
    {
      "epoch": 0.05843722373782669,
      "grad_norm": 0.9807273745536804,
      "learning_rate": 0.00018831275121740233,
      "loss": 3.7063,
      "step": 62690
    },
    {
      "epoch": 0.05844654535590579,
      "grad_norm": 1.1584715843200684,
      "learning_rate": 0.00018831088689530713,
      "loss": 3.9931,
      "step": 62700
    },
    {
      "epoch": 0.05845586697398488,
      "grad_norm": 1.1980687379837036,
      "learning_rate": 0.00018830902257321194,
      "loss": 3.8742,
      "step": 62710
    },
    {
      "epoch": 0.05846518859206397,
      "grad_norm": 1.21001398563385,
      "learning_rate": 0.00018830715825111674,
      "loss": 3.8576,
      "step": 62720
    },
    {
      "epoch": 0.05847451021014306,
      "grad_norm": 1.1795594692230225,
      "learning_rate": 0.00018830529392902155,
      "loss": 3.8098,
      "step": 62730
    },
    {
      "epoch": 0.05848383182822216,
      "grad_norm": 1.034846305847168,
      "learning_rate": 0.00018830342960692633,
      "loss": 4.0337,
      "step": 62740
    },
    {
      "epoch": 0.058493153446301245,
      "grad_norm": 0.8432055115699768,
      "learning_rate": 0.00018830156528483114,
      "loss": 3.8169,
      "step": 62750
    },
    {
      "epoch": 0.05850247506438034,
      "grad_norm": 1.1476075649261475,
      "learning_rate": 0.00018829970096273594,
      "loss": 3.956,
      "step": 62760
    },
    {
      "epoch": 0.05851179668245943,
      "grad_norm": 1.0591439008712769,
      "learning_rate": 0.00018829783664064072,
      "loss": 3.7695,
      "step": 62770
    },
    {
      "epoch": 0.05852111830053852,
      "grad_norm": 1.1735451221466064,
      "learning_rate": 0.00018829597231854556,
      "loss": 3.8343,
      "step": 62780
    },
    {
      "epoch": 0.058530439918617616,
      "grad_norm": 1.2261366844177246,
      "learning_rate": 0.00018829410799645033,
      "loss": 3.874,
      "step": 62790
    },
    {
      "epoch": 0.058539761536696704,
      "grad_norm": 1.150250792503357,
      "learning_rate": 0.00018829224367435514,
      "loss": 3.8612,
      "step": 62800
    },
    {
      "epoch": 0.0585490831547758,
      "grad_norm": 1.312831997871399,
      "learning_rate": 0.00018829037935225995,
      "loss": 3.9617,
      "step": 62810
    },
    {
      "epoch": 0.058558404772854886,
      "grad_norm": 1.1896477937698364,
      "learning_rate": 0.00018828851503016473,
      "loss": 3.9647,
      "step": 62820
    },
    {
      "epoch": 0.05856772639093398,
      "grad_norm": 1.1132397651672363,
      "learning_rate": 0.00018828665070806953,
      "loss": 3.9671,
      "step": 62830
    },
    {
      "epoch": 0.058577048009013075,
      "grad_norm": 0.9646591544151306,
      "learning_rate": 0.00018828478638597434,
      "loss": 3.8988,
      "step": 62840
    },
    {
      "epoch": 0.05858636962709216,
      "grad_norm": 1.2006168365478516,
      "learning_rate": 0.00018828292206387915,
      "loss": 3.8831,
      "step": 62850
    },
    {
      "epoch": 0.05859569124517126,
      "grad_norm": 1.2068253755569458,
      "learning_rate": 0.00018828105774178395,
      "loss": 3.9946,
      "step": 62860
    },
    {
      "epoch": 0.05860501286325035,
      "grad_norm": 1.080079197883606,
      "learning_rate": 0.00018827919341968873,
      "loss": 3.8614,
      "step": 62870
    },
    {
      "epoch": 0.05861433448132944,
      "grad_norm": 1.0218850374221802,
      "learning_rate": 0.00018827732909759354,
      "loss": 3.7892,
      "step": 62880
    },
    {
      "epoch": 0.05862365609940853,
      "grad_norm": 1.2241642475128174,
      "learning_rate": 0.00018827546477549834,
      "loss": 3.7248,
      "step": 62890
    },
    {
      "epoch": 0.05863297771748762,
      "grad_norm": 1.0639575719833374,
      "learning_rate": 0.00018827360045340315,
      "loss": 3.9653,
      "step": 62900
    },
    {
      "epoch": 0.058642299335566715,
      "grad_norm": 1.1711089611053467,
      "learning_rate": 0.00018827173613130796,
      "loss": 3.9925,
      "step": 62910
    },
    {
      "epoch": 0.05865162095364581,
      "grad_norm": 1.2138442993164062,
      "learning_rate": 0.00018826987180921274,
      "loss": 3.9582,
      "step": 62920
    },
    {
      "epoch": 0.0586609425717249,
      "grad_norm": 1.1178416013717651,
      "learning_rate": 0.00018826800748711754,
      "loss": 3.8731,
      "step": 62930
    },
    {
      "epoch": 0.05867026418980399,
      "grad_norm": 1.2802199125289917,
      "learning_rate": 0.00018826614316502235,
      "loss": 3.8522,
      "step": 62940
    },
    {
      "epoch": 0.05867958580788308,
      "grad_norm": 1.0013686418533325,
      "learning_rate": 0.00018826427884292715,
      "loss": 4.0053,
      "step": 62950
    },
    {
      "epoch": 0.058688907425962174,
      "grad_norm": 1.0923311710357666,
      "learning_rate": 0.00018826241452083196,
      "loss": 3.8665,
      "step": 62960
    },
    {
      "epoch": 0.05869822904404127,
      "grad_norm": 1.3151280879974365,
      "learning_rate": 0.00018826055019873674,
      "loss": 3.982,
      "step": 62970
    },
    {
      "epoch": 0.058707550662120356,
      "grad_norm": 1.0511388778686523,
      "learning_rate": 0.00018825868587664155,
      "loss": 3.8813,
      "step": 62980
    },
    {
      "epoch": 0.05871687228019945,
      "grad_norm": 1.0501755475997925,
      "learning_rate": 0.00018825682155454633,
      "loss": 3.9027,
      "step": 62990
    },
    {
      "epoch": 0.058726193898278545,
      "grad_norm": 1.1243324279785156,
      "learning_rate": 0.00018825495723245116,
      "loss": 3.8811,
      "step": 63000
    },
    {
      "epoch": 0.05873551551635763,
      "grad_norm": 1.0601015090942383,
      "learning_rate": 0.00018825309291035596,
      "loss": 3.7956,
      "step": 63010
    },
    {
      "epoch": 0.05874483713443673,
      "grad_norm": 1.225864052772522,
      "learning_rate": 0.00018825122858826074,
      "loss": 3.8206,
      "step": 63020
    },
    {
      "epoch": 0.058754158752515814,
      "grad_norm": 1.0421357154846191,
      "learning_rate": 0.00018824936426616555,
      "loss": 3.857,
      "step": 63030
    },
    {
      "epoch": 0.05876348037059491,
      "grad_norm": 1.165616750717163,
      "learning_rate": 0.00018824749994407033,
      "loss": 3.8966,
      "step": 63040
    },
    {
      "epoch": 0.058772801988674,
      "grad_norm": 0.9097141027450562,
      "learning_rate": 0.00018824563562197514,
      "loss": 4.0288,
      "step": 63050
    },
    {
      "epoch": 0.05878212360675309,
      "grad_norm": 0.8036192655563354,
      "learning_rate": 0.00018824377129987997,
      "loss": 3.8646,
      "step": 63060
    },
    {
      "epoch": 0.058791445224832185,
      "grad_norm": 1.3149374723434448,
      "learning_rate": 0.00018824190697778475,
      "loss": 3.9365,
      "step": 63070
    },
    {
      "epoch": 0.05880076684291128,
      "grad_norm": 1.1293950080871582,
      "learning_rate": 0.00018824004265568955,
      "loss": 3.8098,
      "step": 63080
    },
    {
      "epoch": 0.05881008846099037,
      "grad_norm": 0.9899022579193115,
      "learning_rate": 0.00018823817833359433,
      "loss": 3.9198,
      "step": 63090
    },
    {
      "epoch": 0.05881941007906946,
      "grad_norm": 1.1317459344863892,
      "learning_rate": 0.00018823631401149914,
      "loss": 4.0388,
      "step": 63100
    },
    {
      "epoch": 0.05882873169714855,
      "grad_norm": 1.3381602764129639,
      "learning_rate": 0.00018823444968940395,
      "loss": 4.0395,
      "step": 63110
    },
    {
      "epoch": 0.058838053315227644,
      "grad_norm": 1.2660406827926636,
      "learning_rate": 0.00018823258536730875,
      "loss": 3.9639,
      "step": 63120
    },
    {
      "epoch": 0.05884737493330674,
      "grad_norm": 1.2488900423049927,
      "learning_rate": 0.00018823072104521356,
      "loss": 3.786,
      "step": 63130
    },
    {
      "epoch": 0.058856696551385826,
      "grad_norm": 1.1512982845306396,
      "learning_rate": 0.00018822885672311834,
      "loss": 3.7999,
      "step": 63140
    },
    {
      "epoch": 0.05886601816946492,
      "grad_norm": 1.234067678451538,
      "learning_rate": 0.00018822699240102314,
      "loss": 4.1489,
      "step": 63150
    },
    {
      "epoch": 0.05887533978754401,
      "grad_norm": 1.1286554336547852,
      "learning_rate": 0.00018822512807892795,
      "loss": 3.8574,
      "step": 63160
    },
    {
      "epoch": 0.0588846614056231,
      "grad_norm": 0.9536564350128174,
      "learning_rate": 0.00018822326375683276,
      "loss": 4.0307,
      "step": 63170
    },
    {
      "epoch": 0.0588939830237022,
      "grad_norm": 1.2218520641326904,
      "learning_rate": 0.00018822139943473756,
      "loss": 3.7957,
      "step": 63180
    },
    {
      "epoch": 0.058903304641781284,
      "grad_norm": 1.2117127180099487,
      "learning_rate": 0.00018821953511264234,
      "loss": 3.9501,
      "step": 63190
    },
    {
      "epoch": 0.05891262625986038,
      "grad_norm": 1.0709558725357056,
      "learning_rate": 0.00018821767079054715,
      "loss": 3.722,
      "step": 63200
    },
    {
      "epoch": 0.05892194787793947,
      "grad_norm": 1.0152441263198853,
      "learning_rate": 0.00018821580646845196,
      "loss": 3.9666,
      "step": 63210
    },
    {
      "epoch": 0.05893126949601856,
      "grad_norm": 1.0578488111495972,
      "learning_rate": 0.00018821394214635673,
      "loss": 3.8706,
      "step": 63220
    },
    {
      "epoch": 0.058940591114097655,
      "grad_norm": 0.990776777267456,
      "learning_rate": 0.00018821207782426157,
      "loss": 3.8331,
      "step": 63230
    },
    {
      "epoch": 0.05894991273217674,
      "grad_norm": 1.129975438117981,
      "learning_rate": 0.00018821021350216635,
      "loss": 3.7456,
      "step": 63240
    },
    {
      "epoch": 0.05895923435025584,
      "grad_norm": 1.1116374731063843,
      "learning_rate": 0.00018820834918007115,
      "loss": 4.0557,
      "step": 63250
    },
    {
      "epoch": 0.05896855596833493,
      "grad_norm": 1.1574655771255493,
      "learning_rate": 0.00018820648485797596,
      "loss": 3.6108,
      "step": 63260
    },
    {
      "epoch": 0.05897787758641402,
      "grad_norm": 1.1202874183654785,
      "learning_rate": 0.00018820462053588074,
      "loss": 3.953,
      "step": 63270
    },
    {
      "epoch": 0.058987199204493114,
      "grad_norm": 0.9487501382827759,
      "learning_rate": 0.00018820275621378555,
      "loss": 3.8835,
      "step": 63280
    },
    {
      "epoch": 0.0589965208225722,
      "grad_norm": 1.2062419652938843,
      "learning_rate": 0.00018820089189169035,
      "loss": 3.9794,
      "step": 63290
    },
    {
      "epoch": 0.059005842440651296,
      "grad_norm": 1.0504844188690186,
      "learning_rate": 0.00018819902756959516,
      "loss": 3.9996,
      "step": 63300
    },
    {
      "epoch": 0.05901516405873039,
      "grad_norm": 1.0797910690307617,
      "learning_rate": 0.00018819716324749996,
      "loss": 3.8772,
      "step": 63310
    },
    {
      "epoch": 0.05902448567680948,
      "grad_norm": 1.186706781387329,
      "learning_rate": 0.00018819529892540474,
      "loss": 3.709,
      "step": 63320
    },
    {
      "epoch": 0.05903380729488857,
      "grad_norm": 1.1437466144561768,
      "learning_rate": 0.00018819343460330955,
      "loss": 4.0276,
      "step": 63330
    },
    {
      "epoch": 0.05904312891296767,
      "grad_norm": 1.1152195930480957,
      "learning_rate": 0.00018819157028121436,
      "loss": 4.0064,
      "step": 63340
    },
    {
      "epoch": 0.059052450531046755,
      "grad_norm": 1.2097532749176025,
      "learning_rate": 0.00018818970595911916,
      "loss": 3.8827,
      "step": 63350
    },
    {
      "epoch": 0.05906177214912585,
      "grad_norm": 1.1766414642333984,
      "learning_rate": 0.00018818784163702397,
      "loss": 4.0902,
      "step": 63360
    },
    {
      "epoch": 0.05907109376720494,
      "grad_norm": 1.0007554292678833,
      "learning_rate": 0.00018818597731492875,
      "loss": 3.947,
      "step": 63370
    },
    {
      "epoch": 0.05908041538528403,
      "grad_norm": 1.0463061332702637,
      "learning_rate": 0.00018818411299283355,
      "loss": 3.9954,
      "step": 63380
    },
    {
      "epoch": 0.059089737003363126,
      "grad_norm": 1.1361579895019531,
      "learning_rate": 0.00018818224867073836,
      "loss": 4.0394,
      "step": 63390
    },
    {
      "epoch": 0.05909905862144221,
      "grad_norm": 1.1404438018798828,
      "learning_rate": 0.00018818038434864317,
      "loss": 3.8223,
      "step": 63400
    },
    {
      "epoch": 0.05910838023952131,
      "grad_norm": 1.0361108779907227,
      "learning_rate": 0.00018817852002654797,
      "loss": 3.9559,
      "step": 63410
    },
    {
      "epoch": 0.0591177018576004,
      "grad_norm": 1.1548340320587158,
      "learning_rate": 0.00018817665570445275,
      "loss": 3.816,
      "step": 63420
    },
    {
      "epoch": 0.05912702347567949,
      "grad_norm": 1.0509788990020752,
      "learning_rate": 0.00018817479138235756,
      "loss": 3.8438,
      "step": 63430
    },
    {
      "epoch": 0.059136345093758584,
      "grad_norm": 1.0547162294387817,
      "learning_rate": 0.00018817292706026234,
      "loss": 3.9819,
      "step": 63440
    },
    {
      "epoch": 0.05914566671183767,
      "grad_norm": 1.394352674484253,
      "learning_rate": 0.00018817106273816717,
      "loss": 3.7164,
      "step": 63450
    },
    {
      "epoch": 0.059154988329916766,
      "grad_norm": 1.125857949256897,
      "learning_rate": 0.00018816919841607198,
      "loss": 3.8519,
      "step": 63460
    },
    {
      "epoch": 0.05916430994799586,
      "grad_norm": 1.0103247165679932,
      "learning_rate": 0.00018816733409397676,
      "loss": 4.079,
      "step": 63470
    },
    {
      "epoch": 0.05917363156607495,
      "grad_norm": 1.140690803527832,
      "learning_rate": 0.00018816546977188156,
      "loss": 3.8561,
      "step": 63480
    },
    {
      "epoch": 0.05918295318415404,
      "grad_norm": 1.245308756828308,
      "learning_rate": 0.00018816360544978634,
      "loss": 3.8387,
      "step": 63490
    },
    {
      "epoch": 0.05919227480223313,
      "grad_norm": 1.1275092363357544,
      "learning_rate": 0.00018816174112769115,
      "loss": 3.7823,
      "step": 63500
    },
    {
      "epoch": 0.059201596420312225,
      "grad_norm": 1.1081327199935913,
      "learning_rate": 0.00018815987680559598,
      "loss": 4.0411,
      "step": 63510
    },
    {
      "epoch": 0.05921091803839132,
      "grad_norm": 1.1530652046203613,
      "learning_rate": 0.00018815801248350076,
      "loss": 3.9038,
      "step": 63520
    },
    {
      "epoch": 0.05922023965647041,
      "grad_norm": 0.9766855835914612,
      "learning_rate": 0.00018815614816140557,
      "loss": 4.0554,
      "step": 63530
    },
    {
      "epoch": 0.0592295612745495,
      "grad_norm": 1.1099703311920166,
      "learning_rate": 0.00018815428383931035,
      "loss": 4.0883,
      "step": 63540
    },
    {
      "epoch": 0.059238882892628596,
      "grad_norm": 1.0470476150512695,
      "learning_rate": 0.00018815241951721515,
      "loss": 3.8492,
      "step": 63550
    },
    {
      "epoch": 0.05924820451070768,
      "grad_norm": 0.957287073135376,
      "learning_rate": 0.00018815055519511996,
      "loss": 3.9102,
      "step": 63560
    },
    {
      "epoch": 0.05925752612878678,
      "grad_norm": 1.1413264274597168,
      "learning_rate": 0.00018814869087302477,
      "loss": 3.9081,
      "step": 63570
    },
    {
      "epoch": 0.059266847746865865,
      "grad_norm": 0.997186005115509,
      "learning_rate": 0.00018814682655092957,
      "loss": 4.0487,
      "step": 63580
    },
    {
      "epoch": 0.05927616936494496,
      "grad_norm": 1.1808760166168213,
      "learning_rate": 0.00018814496222883435,
      "loss": 3.7657,
      "step": 63590
    },
    {
      "epoch": 0.059285490983024054,
      "grad_norm": 1.0543168783187866,
      "learning_rate": 0.00018814309790673916,
      "loss": 4.0003,
      "step": 63600
    },
    {
      "epoch": 0.05929481260110314,
      "grad_norm": 1.1824076175689697,
      "learning_rate": 0.00018814123358464394,
      "loss": 3.8497,
      "step": 63610
    },
    {
      "epoch": 0.059304134219182236,
      "grad_norm": 1.3876545429229736,
      "learning_rate": 0.00018813936926254877,
      "loss": 4.0218,
      "step": 63620
    },
    {
      "epoch": 0.059313455837261324,
      "grad_norm": 0.9972414970397949,
      "learning_rate": 0.00018813750494045358,
      "loss": 4.1215,
      "step": 63630
    },
    {
      "epoch": 0.05932277745534042,
      "grad_norm": 1.2940113544464111,
      "learning_rate": 0.00018813564061835836,
      "loss": 3.934,
      "step": 63640
    },
    {
      "epoch": 0.05933209907341951,
      "grad_norm": 1.0514497756958008,
      "learning_rate": 0.00018813377629626316,
      "loss": 3.974,
      "step": 63650
    },
    {
      "epoch": 0.0593414206914986,
      "grad_norm": 1.133741855621338,
      "learning_rate": 0.00018813191197416794,
      "loss": 3.982,
      "step": 63660
    },
    {
      "epoch": 0.059350742309577695,
      "grad_norm": 1.0507925748825073,
      "learning_rate": 0.00018813004765207275,
      "loss": 3.7253,
      "step": 63670
    },
    {
      "epoch": 0.05936006392765679,
      "grad_norm": 1.0861009359359741,
      "learning_rate": 0.00018812818332997758,
      "loss": 3.934,
      "step": 63680
    },
    {
      "epoch": 0.05936938554573588,
      "grad_norm": 1.1518630981445312,
      "learning_rate": 0.00018812631900788236,
      "loss": 3.8761,
      "step": 63690
    },
    {
      "epoch": 0.05937870716381497,
      "grad_norm": 1.1240757703781128,
      "learning_rate": 0.00018812445468578717,
      "loss": 3.9272,
      "step": 63700
    },
    {
      "epoch": 0.05938802878189406,
      "grad_norm": 1.0427089929580688,
      "learning_rate": 0.00018812259036369195,
      "loss": 3.7779,
      "step": 63710
    },
    {
      "epoch": 0.05939735039997315,
      "grad_norm": 1.0119084119796753,
      "learning_rate": 0.00018812072604159675,
      "loss": 3.8915,
      "step": 63720
    },
    {
      "epoch": 0.05940667201805225,
      "grad_norm": 1.0342435836791992,
      "learning_rate": 0.00018811886171950156,
      "loss": 3.9411,
      "step": 63730
    },
    {
      "epoch": 0.059415993636131335,
      "grad_norm": 1.2160981893539429,
      "learning_rate": 0.00018811699739740636,
      "loss": 3.958,
      "step": 63740
    },
    {
      "epoch": 0.05942531525421043,
      "grad_norm": 1.2081561088562012,
      "learning_rate": 0.00018811513307531117,
      "loss": 3.933,
      "step": 63750
    },
    {
      "epoch": 0.059434636872289524,
      "grad_norm": 0.9848472476005554,
      "learning_rate": 0.00018811326875321595,
      "loss": 4.0674,
      "step": 63760
    },
    {
      "epoch": 0.05944395849036861,
      "grad_norm": 1.0338081121444702,
      "learning_rate": 0.00018811140443112076,
      "loss": 3.8265,
      "step": 63770
    },
    {
      "epoch": 0.059453280108447706,
      "grad_norm": 1.19691002368927,
      "learning_rate": 0.00018810954010902556,
      "loss": 3.8837,
      "step": 63780
    },
    {
      "epoch": 0.059462601726526794,
      "grad_norm": 1.0176936388015747,
      "learning_rate": 0.00018810767578693037,
      "loss": 3.8051,
      "step": 63790
    },
    {
      "epoch": 0.05947192334460589,
      "grad_norm": 1.1617166996002197,
      "learning_rate": 0.00018810581146483518,
      "loss": 3.8929,
      "step": 63800
    },
    {
      "epoch": 0.05948124496268498,
      "grad_norm": 1.0160932540893555,
      "learning_rate": 0.00018810394714273995,
      "loss": 4.0818,
      "step": 63810
    },
    {
      "epoch": 0.05949056658076407,
      "grad_norm": 1.0580905675888062,
      "learning_rate": 0.00018810208282064476,
      "loss": 3.9797,
      "step": 63820
    },
    {
      "epoch": 0.059499888198843165,
      "grad_norm": 1.1108691692352295,
      "learning_rate": 0.00018810021849854957,
      "loss": 3.6308,
      "step": 63830
    },
    {
      "epoch": 0.05950920981692225,
      "grad_norm": 1.1011738777160645,
      "learning_rate": 0.00018809835417645437,
      "loss": 3.9818,
      "step": 63840
    },
    {
      "epoch": 0.05951853143500135,
      "grad_norm": 1.098128318786621,
      "learning_rate": 0.00018809648985435918,
      "loss": 4.0317,
      "step": 63850
    },
    {
      "epoch": 0.05952785305308044,
      "grad_norm": 1.2602578401565552,
      "learning_rate": 0.00018809462553226396,
      "loss": 3.7744,
      "step": 63860
    },
    {
      "epoch": 0.05953717467115953,
      "grad_norm": 1.021629810333252,
      "learning_rate": 0.00018809276121016877,
      "loss": 4.0042,
      "step": 63870
    },
    {
      "epoch": 0.059546496289238623,
      "grad_norm": 1.1641446352005005,
      "learning_rate": 0.00018809089688807357,
      "loss": 3.9347,
      "step": 63880
    },
    {
      "epoch": 0.05955581790731772,
      "grad_norm": 1.1331961154937744,
      "learning_rate": 0.00018808903256597835,
      "loss": 4.0351,
      "step": 63890
    },
    {
      "epoch": 0.059565139525396806,
      "grad_norm": 1.0348235368728638,
      "learning_rate": 0.00018808716824388318,
      "loss": 3.8984,
      "step": 63900
    },
    {
      "epoch": 0.0595744611434759,
      "grad_norm": 1.044596791267395,
      "learning_rate": 0.00018808530392178796,
      "loss": 3.9537,
      "step": 63910
    },
    {
      "epoch": 0.05958378276155499,
      "grad_norm": 1.1176419258117676,
      "learning_rate": 0.00018808343959969277,
      "loss": 3.7542,
      "step": 63920
    },
    {
      "epoch": 0.05959310437963408,
      "grad_norm": 1.2139707803726196,
      "learning_rate": 0.00018808157527759758,
      "loss": 3.844,
      "step": 63930
    },
    {
      "epoch": 0.059602425997713177,
      "grad_norm": 1.0535190105438232,
      "learning_rate": 0.00018807971095550236,
      "loss": 3.9267,
      "step": 63940
    },
    {
      "epoch": 0.059611747615792264,
      "grad_norm": 1.0684497356414795,
      "learning_rate": 0.00018807784663340716,
      "loss": 4.0059,
      "step": 63950
    },
    {
      "epoch": 0.05962106923387136,
      "grad_norm": 1.1661566495895386,
      "learning_rate": 0.00018807598231131197,
      "loss": 3.8434,
      "step": 63960
    },
    {
      "epoch": 0.059630390851950446,
      "grad_norm": 1.184863567352295,
      "learning_rate": 0.00018807411798921677,
      "loss": 3.861,
      "step": 63970
    },
    {
      "epoch": 0.05963971247002954,
      "grad_norm": 0.9169617295265198,
      "learning_rate": 0.00018807225366712158,
      "loss": 3.8838,
      "step": 63980
    },
    {
      "epoch": 0.059649034088108635,
      "grad_norm": 0.9997320771217346,
      "learning_rate": 0.00018807038934502636,
      "loss": 3.8465,
      "step": 63990
    },
    {
      "epoch": 0.05965835570618772,
      "grad_norm": 0.9605811834335327,
      "learning_rate": 0.00018806852502293117,
      "loss": 4.0291,
      "step": 64000
    },
    {
      "epoch": 0.05966767732426682,
      "grad_norm": 1.2175803184509277,
      "learning_rate": 0.00018806666070083597,
      "loss": 3.9205,
      "step": 64010
    },
    {
      "epoch": 0.05967699894234591,
      "grad_norm": 0.973584771156311,
      "learning_rate": 0.00018806479637874078,
      "loss": 3.9556,
      "step": 64020
    },
    {
      "epoch": 0.059686320560425,
      "grad_norm": 0.9981963634490967,
      "learning_rate": 0.00018806293205664558,
      "loss": 3.9506,
      "step": 64030
    },
    {
      "epoch": 0.059695642178504094,
      "grad_norm": 1.192259669303894,
      "learning_rate": 0.00018806106773455036,
      "loss": 4.1175,
      "step": 64040
    },
    {
      "epoch": 0.05970496379658318,
      "grad_norm": 1.1798372268676758,
      "learning_rate": 0.00018805920341245517,
      "loss": 4.0002,
      "step": 64050
    },
    {
      "epoch": 0.059714285414662276,
      "grad_norm": 1.035584807395935,
      "learning_rate": 0.00018805733909035995,
      "loss": 3.9261,
      "step": 64060
    },
    {
      "epoch": 0.05972360703274137,
      "grad_norm": 1.1763451099395752,
      "learning_rate": 0.00018805547476826478,
      "loss": 3.8226,
      "step": 64070
    },
    {
      "epoch": 0.05973292865082046,
      "grad_norm": 1.2169792652130127,
      "learning_rate": 0.0001880536104461696,
      "loss": 3.9957,
      "step": 64080
    },
    {
      "epoch": 0.05974225026889955,
      "grad_norm": 1.2543030977249146,
      "learning_rate": 0.00018805174612407437,
      "loss": 4.0045,
      "step": 64090
    },
    {
      "epoch": 0.05975157188697865,
      "grad_norm": 1.0969030857086182,
      "learning_rate": 0.00018804988180197917,
      "loss": 3.8061,
      "step": 64100
    },
    {
      "epoch": 0.059760893505057734,
      "grad_norm": 0.9092556238174438,
      "learning_rate": 0.00018804801747988395,
      "loss": 3.8259,
      "step": 64110
    },
    {
      "epoch": 0.05977021512313683,
      "grad_norm": 0.9394232630729675,
      "learning_rate": 0.00018804615315778876,
      "loss": 3.9215,
      "step": 64120
    },
    {
      "epoch": 0.059779536741215916,
      "grad_norm": 1.3805817365646362,
      "learning_rate": 0.0001880442888356936,
      "loss": 3.9349,
      "step": 64130
    },
    {
      "epoch": 0.05978885835929501,
      "grad_norm": 1.4120526313781738,
      "learning_rate": 0.00018804242451359837,
      "loss": 3.716,
      "step": 64140
    },
    {
      "epoch": 0.059798179977374105,
      "grad_norm": 0.9552243947982788,
      "learning_rate": 0.00018804056019150318,
      "loss": 3.8713,
      "step": 64150
    },
    {
      "epoch": 0.05980750159545319,
      "grad_norm": 1.0525908470153809,
      "learning_rate": 0.00018803869586940796,
      "loss": 3.9904,
      "step": 64160
    },
    {
      "epoch": 0.05981682321353229,
      "grad_norm": 0.9801773428916931,
      "learning_rate": 0.00018803683154731276,
      "loss": 3.8543,
      "step": 64170
    },
    {
      "epoch": 0.059826144831611375,
      "grad_norm": 0.9516556859016418,
      "learning_rate": 0.00018803496722521757,
      "loss": 3.9019,
      "step": 64180
    },
    {
      "epoch": 0.05983546644969047,
      "grad_norm": 0.9873784184455872,
      "learning_rate": 0.00018803310290312238,
      "loss": 4.0346,
      "step": 64190
    },
    {
      "epoch": 0.059844788067769564,
      "grad_norm": 1.07377028465271,
      "learning_rate": 0.00018803123858102718,
      "loss": 4.0032,
      "step": 64200
    },
    {
      "epoch": 0.05985410968584865,
      "grad_norm": 1.0287189483642578,
      "learning_rate": 0.00018802937425893196,
      "loss": 3.8882,
      "step": 64210
    },
    {
      "epoch": 0.059863431303927746,
      "grad_norm": 1.0208022594451904,
      "learning_rate": 0.00018802750993683677,
      "loss": 4.1186,
      "step": 64220
    },
    {
      "epoch": 0.05987275292200684,
      "grad_norm": 0.8580570816993713,
      "learning_rate": 0.00018802564561474158,
      "loss": 3.9957,
      "step": 64230
    },
    {
      "epoch": 0.05988207454008593,
      "grad_norm": 1.2502729892730713,
      "learning_rate": 0.00018802378129264638,
      "loss": 3.9775,
      "step": 64240
    },
    {
      "epoch": 0.05989139615816502,
      "grad_norm": 1.0990549325942993,
      "learning_rate": 0.0001880219169705512,
      "loss": 4.107,
      "step": 64250
    },
    {
      "epoch": 0.05990071777624411,
      "grad_norm": 0.9679533243179321,
      "learning_rate": 0.00018802005264845597,
      "loss": 3.8347,
      "step": 64260
    },
    {
      "epoch": 0.059910039394323204,
      "grad_norm": 0.944072961807251,
      "learning_rate": 0.00018801818832636077,
      "loss": 3.8203,
      "step": 64270
    },
    {
      "epoch": 0.0599193610124023,
      "grad_norm": 1.2141450643539429,
      "learning_rate": 0.00018801632400426558,
      "loss": 3.8075,
      "step": 64280
    },
    {
      "epoch": 0.059928682630481386,
      "grad_norm": 0.924203634262085,
      "learning_rate": 0.00018801445968217039,
      "loss": 3.8699,
      "step": 64290
    },
    {
      "epoch": 0.05993800424856048,
      "grad_norm": 1.0174769163131714,
      "learning_rate": 0.0001880125953600752,
      "loss": 4.0889,
      "step": 64300
    },
    {
      "epoch": 0.059947325866639575,
      "grad_norm": 0.9883471131324768,
      "learning_rate": 0.00018801073103797997,
      "loss": 3.8587,
      "step": 64310
    },
    {
      "epoch": 0.05995664748471866,
      "grad_norm": 1.2071506977081299,
      "learning_rate": 0.00018800886671588478,
      "loss": 3.8715,
      "step": 64320
    },
    {
      "epoch": 0.05996596910279776,
      "grad_norm": 1.2694685459136963,
      "learning_rate": 0.00018800700239378958,
      "loss": 3.9722,
      "step": 64330
    },
    {
      "epoch": 0.059975290720876845,
      "grad_norm": 1.1609208583831787,
      "learning_rate": 0.00018800513807169436,
      "loss": 3.8355,
      "step": 64340
    },
    {
      "epoch": 0.05998461233895594,
      "grad_norm": 0.9269546866416931,
      "learning_rate": 0.0001880032737495992,
      "loss": 3.8036,
      "step": 64350
    },
    {
      "epoch": 0.059993933957035034,
      "grad_norm": 1.1179373264312744,
      "learning_rate": 0.00018800140942750398,
      "loss": 3.9166,
      "step": 64360
    },
    {
      "epoch": 0.06000325557511412,
      "grad_norm": 1.1386301517486572,
      "learning_rate": 0.00018799954510540878,
      "loss": 3.8794,
      "step": 64370
    },
    {
      "epoch": 0.060012577193193216,
      "grad_norm": 1.0436017513275146,
      "learning_rate": 0.0001879976807833136,
      "loss": 3.8453,
      "step": 64380
    },
    {
      "epoch": 0.0600218988112723,
      "grad_norm": 1.1860486268997192,
      "learning_rate": 0.00018799581646121837,
      "loss": 4.0934,
      "step": 64390
    },
    {
      "epoch": 0.0600312204293514,
      "grad_norm": 1.2047607898712158,
      "learning_rate": 0.00018799395213912317,
      "loss": 3.9369,
      "step": 64400
    },
    {
      "epoch": 0.06004054204743049,
      "grad_norm": 1.0522058010101318,
      "learning_rate": 0.00018799208781702798,
      "loss": 3.7963,
      "step": 64410
    },
    {
      "epoch": 0.06004986366550958,
      "grad_norm": 1.0410252809524536,
      "learning_rate": 0.0001879902234949328,
      "loss": 3.666,
      "step": 64420
    },
    {
      "epoch": 0.060059185283588674,
      "grad_norm": 0.9196585416793823,
      "learning_rate": 0.0001879883591728376,
      "loss": 3.8298,
      "step": 64430
    },
    {
      "epoch": 0.06006850690166777,
      "grad_norm": 0.845612108707428,
      "learning_rate": 0.00018798649485074237,
      "loss": 3.9435,
      "step": 64440
    },
    {
      "epoch": 0.060077828519746856,
      "grad_norm": 1.2057485580444336,
      "learning_rate": 0.00018798463052864718,
      "loss": 3.9441,
      "step": 64450
    },
    {
      "epoch": 0.06008715013782595,
      "grad_norm": 1.165625810623169,
      "learning_rate": 0.00018798276620655199,
      "loss": 3.8571,
      "step": 64460
    },
    {
      "epoch": 0.06009647175590504,
      "grad_norm": 1.0575535297393799,
      "learning_rate": 0.0001879809018844568,
      "loss": 3.9252,
      "step": 64470
    },
    {
      "epoch": 0.06010579337398413,
      "grad_norm": 1.01272714138031,
      "learning_rate": 0.0001879790375623616,
      "loss": 3.8347,
      "step": 64480
    },
    {
      "epoch": 0.06011511499206323,
      "grad_norm": 1.2112528085708618,
      "learning_rate": 0.00018797717324026638,
      "loss": 3.9454,
      "step": 64490
    },
    {
      "epoch": 0.060124436610142315,
      "grad_norm": 1.123066782951355,
      "learning_rate": 0.00018797530891817118,
      "loss": 3.9567,
      "step": 64500
    },
    {
      "epoch": 0.06013375822822141,
      "grad_norm": 1.0244057178497314,
      "learning_rate": 0.000187973444596076,
      "loss": 3.9417,
      "step": 64510
    },
    {
      "epoch": 0.0601430798463005,
      "grad_norm": 1.0779253244400024,
      "learning_rate": 0.0001879715802739808,
      "loss": 4.1914,
      "step": 64520
    },
    {
      "epoch": 0.06015240146437959,
      "grad_norm": 1.0331701040267944,
      "learning_rate": 0.0001879697159518856,
      "loss": 3.959,
      "step": 64530
    },
    {
      "epoch": 0.060161723082458686,
      "grad_norm": 0.9989593625068665,
      "learning_rate": 0.00018796785162979038,
      "loss": 3.8264,
      "step": 64540
    },
    {
      "epoch": 0.060171044700537774,
      "grad_norm": 1.006681203842163,
      "learning_rate": 0.0001879659873076952,
      "loss": 3.7959,
      "step": 64550
    },
    {
      "epoch": 0.06018036631861687,
      "grad_norm": 1.0439082384109497,
      "learning_rate": 0.00018796412298559997,
      "loss": 4.1206,
      "step": 64560
    },
    {
      "epoch": 0.06018968793669596,
      "grad_norm": 0.9713059663772583,
      "learning_rate": 0.00018796225866350477,
      "loss": 3.8167,
      "step": 64570
    },
    {
      "epoch": 0.06019900955477505,
      "grad_norm": 1.169514536857605,
      "learning_rate": 0.0001879603943414096,
      "loss": 3.8351,
      "step": 64580
    },
    {
      "epoch": 0.060208331172854145,
      "grad_norm": 1.1005680561065674,
      "learning_rate": 0.00018795853001931439,
      "loss": 3.9096,
      "step": 64590
    },
    {
      "epoch": 0.06021765279093323,
      "grad_norm": 1.1941906213760376,
      "learning_rate": 0.0001879566656972192,
      "loss": 3.5728,
      "step": 64600
    },
    {
      "epoch": 0.06022697440901233,
      "grad_norm": 1.1154274940490723,
      "learning_rate": 0.00018795480137512397,
      "loss": 3.8411,
      "step": 64610
    },
    {
      "epoch": 0.06023629602709142,
      "grad_norm": 0.9088528156280518,
      "learning_rate": 0.00018795293705302878,
      "loss": 3.8579,
      "step": 64620
    },
    {
      "epoch": 0.06024561764517051,
      "grad_norm": 1.07724928855896,
      "learning_rate": 0.00018795107273093358,
      "loss": 3.8658,
      "step": 64630
    },
    {
      "epoch": 0.0602549392632496,
      "grad_norm": 1.2487361431121826,
      "learning_rate": 0.0001879492084088384,
      "loss": 3.9237,
      "step": 64640
    },
    {
      "epoch": 0.0602642608813287,
      "grad_norm": 1.4439587593078613,
      "learning_rate": 0.0001879473440867432,
      "loss": 3.941,
      "step": 64650
    },
    {
      "epoch": 0.060273582499407785,
      "grad_norm": 1.2518527507781982,
      "learning_rate": 0.00018794547976464798,
      "loss": 3.7478,
      "step": 64660
    },
    {
      "epoch": 0.06028290411748688,
      "grad_norm": 1.2629590034484863,
      "learning_rate": 0.00018794361544255278,
      "loss": 4.0378,
      "step": 64670
    },
    {
      "epoch": 0.06029222573556597,
      "grad_norm": 1.0099592208862305,
      "learning_rate": 0.0001879417511204576,
      "loss": 3.706,
      "step": 64680
    },
    {
      "epoch": 0.06030154735364506,
      "grad_norm": 0.9309908151626587,
      "learning_rate": 0.0001879398867983624,
      "loss": 4.0109,
      "step": 64690
    },
    {
      "epoch": 0.060310868971724156,
      "grad_norm": 1.09247887134552,
      "learning_rate": 0.0001879380224762672,
      "loss": 3.9705,
      "step": 64700
    },
    {
      "epoch": 0.060320190589803244,
      "grad_norm": 1.208708643913269,
      "learning_rate": 0.00018793615815417198,
      "loss": 4.0274,
      "step": 64710
    },
    {
      "epoch": 0.06032951220788234,
      "grad_norm": 1.1125383377075195,
      "learning_rate": 0.0001879342938320768,
      "loss": 4.003,
      "step": 64720
    },
    {
      "epoch": 0.060338833825961426,
      "grad_norm": 0.9501692056655884,
      "learning_rate": 0.0001879324295099816,
      "loss": 3.9299,
      "step": 64730
    },
    {
      "epoch": 0.06034815544404052,
      "grad_norm": 1.0972506999969482,
      "learning_rate": 0.0001879305651878864,
      "loss": 3.8458,
      "step": 64740
    },
    {
      "epoch": 0.060357477062119615,
      "grad_norm": 0.9643729329109192,
      "learning_rate": 0.0001879287008657912,
      "loss": 3.8563,
      "step": 64750
    },
    {
      "epoch": 0.0603667986801987,
      "grad_norm": 1.3134554624557495,
      "learning_rate": 0.00018792683654369598,
      "loss": 3.906,
      "step": 64760
    },
    {
      "epoch": 0.0603761202982778,
      "grad_norm": 1.0105133056640625,
      "learning_rate": 0.0001879249722216008,
      "loss": 3.7051,
      "step": 64770
    },
    {
      "epoch": 0.06038544191635689,
      "grad_norm": 1.1126211881637573,
      "learning_rate": 0.0001879231078995056,
      "loss": 3.91,
      "step": 64780
    },
    {
      "epoch": 0.06039476353443598,
      "grad_norm": 1.0804306268692017,
      "learning_rate": 0.00018792124357741038,
      "loss": 3.9027,
      "step": 64790
    },
    {
      "epoch": 0.06040408515251507,
      "grad_norm": 1.0917010307312012,
      "learning_rate": 0.0001879193792553152,
      "loss": 3.9318,
      "step": 64800
    },
    {
      "epoch": 0.06041340677059416,
      "grad_norm": 1.016772985458374,
      "learning_rate": 0.00018791751493322,
      "loss": 3.9199,
      "step": 64810
    },
    {
      "epoch": 0.060422728388673255,
      "grad_norm": 0.9813007712364197,
      "learning_rate": 0.0001879156506111248,
      "loss": 3.8368,
      "step": 64820
    },
    {
      "epoch": 0.06043205000675235,
      "grad_norm": 1.1269391775131226,
      "learning_rate": 0.0001879137862890296,
      "loss": 3.9453,
      "step": 64830
    },
    {
      "epoch": 0.06044137162483144,
      "grad_norm": 1.0820554494857788,
      "learning_rate": 0.00018791192196693438,
      "loss": 4.0099,
      "step": 64840
    },
    {
      "epoch": 0.06045069324291053,
      "grad_norm": 0.9398381114006042,
      "learning_rate": 0.0001879100576448392,
      "loss": 4.1713,
      "step": 64850
    },
    {
      "epoch": 0.06046001486098962,
      "grad_norm": 1.1340458393096924,
      "learning_rate": 0.000187908193322744,
      "loss": 4.1324,
      "step": 64860
    },
    {
      "epoch": 0.060469336479068714,
      "grad_norm": 1.18028724193573,
      "learning_rate": 0.0001879063290006488,
      "loss": 4.0545,
      "step": 64870
    },
    {
      "epoch": 0.06047865809714781,
      "grad_norm": 1.1035226583480835,
      "learning_rate": 0.00018790446467855358,
      "loss": 3.9254,
      "step": 64880
    },
    {
      "epoch": 0.060487979715226896,
      "grad_norm": 1.0715570449829102,
      "learning_rate": 0.00018790260035645839,
      "loss": 3.7349,
      "step": 64890
    },
    {
      "epoch": 0.06049730133330599,
      "grad_norm": 1.1206382513046265,
      "learning_rate": 0.0001879007360343632,
      "loss": 3.9063,
      "step": 64900
    },
    {
      "epoch": 0.060506622951385085,
      "grad_norm": 1.0302106142044067,
      "learning_rate": 0.000187898871712268,
      "loss": 3.9624,
      "step": 64910
    },
    {
      "epoch": 0.06051594456946417,
      "grad_norm": 0.8917478322982788,
      "learning_rate": 0.0001878970073901728,
      "loss": 3.9785,
      "step": 64920
    },
    {
      "epoch": 0.06052526618754327,
      "grad_norm": 1.0306987762451172,
      "learning_rate": 0.00018789514306807758,
      "loss": 3.921,
      "step": 64930
    },
    {
      "epoch": 0.060534587805622354,
      "grad_norm": 0.9739078879356384,
      "learning_rate": 0.0001878932787459824,
      "loss": 3.9705,
      "step": 64940
    },
    {
      "epoch": 0.06054390942370145,
      "grad_norm": 1.0562341213226318,
      "learning_rate": 0.0001878914144238872,
      "loss": 3.8391,
      "step": 64950
    },
    {
      "epoch": 0.06055323104178054,
      "grad_norm": 1.1265020370483398,
      "learning_rate": 0.000187889550101792,
      "loss": 3.8424,
      "step": 64960
    },
    {
      "epoch": 0.06056255265985963,
      "grad_norm": 1.0777702331542969,
      "learning_rate": 0.0001878876857796968,
      "loss": 3.8296,
      "step": 64970
    },
    {
      "epoch": 0.060571874277938725,
      "grad_norm": 0.9297662973403931,
      "learning_rate": 0.0001878858214576016,
      "loss": 3.8225,
      "step": 64980
    },
    {
      "epoch": 0.06058119589601782,
      "grad_norm": 1.1075981855392456,
      "learning_rate": 0.0001878839571355064,
      "loss": 3.8801,
      "step": 64990
    },
    {
      "epoch": 0.06059051751409691,
      "grad_norm": 1.2085940837860107,
      "learning_rate": 0.0001878820928134112,
      "loss": 3.9394,
      "step": 65000
    },
    {
      "epoch": 0.060599839132176,
      "grad_norm": 1.2464669942855835,
      "learning_rate": 0.00018788022849131598,
      "loss": 3.8452,
      "step": 65010
    },
    {
      "epoch": 0.06060916075025509,
      "grad_norm": 1.0319335460662842,
      "learning_rate": 0.0001878783641692208,
      "loss": 4.0079,
      "step": 65020
    },
    {
      "epoch": 0.060618482368334184,
      "grad_norm": 0.9688382148742676,
      "learning_rate": 0.0001878764998471256,
      "loss": 4.0402,
      "step": 65030
    },
    {
      "epoch": 0.06062780398641328,
      "grad_norm": 1.0251909494400024,
      "learning_rate": 0.0001878746355250304,
      "loss": 3.8112,
      "step": 65040
    },
    {
      "epoch": 0.060637125604492366,
      "grad_norm": 0.9585376977920532,
      "learning_rate": 0.0001878727712029352,
      "loss": 3.7364,
      "step": 65050
    },
    {
      "epoch": 0.06064644722257146,
      "grad_norm": 1.1863813400268555,
      "learning_rate": 0.00018787090688083998,
      "loss": 3.9434,
      "step": 65060
    },
    {
      "epoch": 0.06065576884065055,
      "grad_norm": 1.094907283782959,
      "learning_rate": 0.0001878690425587448,
      "loss": 3.6583,
      "step": 65070
    },
    {
      "epoch": 0.06066509045872964,
      "grad_norm": 1.0210227966308594,
      "learning_rate": 0.0001878671782366496,
      "loss": 3.9501,
      "step": 65080
    },
    {
      "epoch": 0.06067441207680874,
      "grad_norm": 1.2662996053695679,
      "learning_rate": 0.0001878653139145544,
      "loss": 3.8935,
      "step": 65090
    },
    {
      "epoch": 0.060683733694887824,
      "grad_norm": 1.2390304803848267,
      "learning_rate": 0.0001878634495924592,
      "loss": 3.6416,
      "step": 65100
    },
    {
      "epoch": 0.06069305531296692,
      "grad_norm": 1.0468230247497559,
      "learning_rate": 0.000187861585270364,
      "loss": 4.0266,
      "step": 65110
    },
    {
      "epoch": 0.06070237693104601,
      "grad_norm": 1.123716950416565,
      "learning_rate": 0.0001878597209482688,
      "loss": 3.8495,
      "step": 65120
    },
    {
      "epoch": 0.0607116985491251,
      "grad_norm": 1.1283199787139893,
      "learning_rate": 0.0001878578566261736,
      "loss": 3.9808,
      "step": 65130
    },
    {
      "epoch": 0.060721020167204195,
      "grad_norm": 0.9939330816268921,
      "learning_rate": 0.0001878559923040784,
      "loss": 3.9669,
      "step": 65140
    },
    {
      "epoch": 0.06073034178528328,
      "grad_norm": 0.9211974740028381,
      "learning_rate": 0.00018785412798198321,
      "loss": 3.7776,
      "step": 65150
    },
    {
      "epoch": 0.06073966340336238,
      "grad_norm": 1.1875938177108765,
      "learning_rate": 0.000187852263659888,
      "loss": 4.0116,
      "step": 65160
    },
    {
      "epoch": 0.06074898502144147,
      "grad_norm": 0.9120241403579712,
      "learning_rate": 0.0001878503993377928,
      "loss": 4.0054,
      "step": 65170
    },
    {
      "epoch": 0.06075830663952056,
      "grad_norm": 1.0496785640716553,
      "learning_rate": 0.00018784853501569758,
      "loss": 3.8567,
      "step": 65180
    },
    {
      "epoch": 0.060767628257599654,
      "grad_norm": 1.0697646141052246,
      "learning_rate": 0.0001878466706936024,
      "loss": 4.013,
      "step": 65190
    },
    {
      "epoch": 0.06077694987567874,
      "grad_norm": 1.0512312650680542,
      "learning_rate": 0.00018784480637150722,
      "loss": 4.0272,
      "step": 65200
    },
    {
      "epoch": 0.060786271493757836,
      "grad_norm": 1.1065566539764404,
      "learning_rate": 0.000187842942049412,
      "loss": 3.9337,
      "step": 65210
    },
    {
      "epoch": 0.06079559311183693,
      "grad_norm": 1.0468909740447998,
      "learning_rate": 0.0001878410777273168,
      "loss": 3.9332,
      "step": 65220
    },
    {
      "epoch": 0.06080491472991602,
      "grad_norm": 1.2203806638717651,
      "learning_rate": 0.00018783921340522158,
      "loss": 4.0958,
      "step": 65230
    },
    {
      "epoch": 0.06081423634799511,
      "grad_norm": 1.1579252481460571,
      "learning_rate": 0.0001878373490831264,
      "loss": 3.6801,
      "step": 65240
    },
    {
      "epoch": 0.06082355796607421,
      "grad_norm": 0.968598484992981,
      "learning_rate": 0.00018783548476103122,
      "loss": 3.8988,
      "step": 65250
    },
    {
      "epoch": 0.060832879584153295,
      "grad_norm": 1.3055506944656372,
      "learning_rate": 0.000187833620438936,
      "loss": 3.8922,
      "step": 65260
    },
    {
      "epoch": 0.06084220120223239,
      "grad_norm": 0.9420547485351562,
      "learning_rate": 0.0001878317561168408,
      "loss": 3.8237,
      "step": 65270
    },
    {
      "epoch": 0.06085152282031148,
      "grad_norm": 1.037019968032837,
      "learning_rate": 0.0001878298917947456,
      "loss": 3.9784,
      "step": 65280
    },
    {
      "epoch": 0.06086084443839057,
      "grad_norm": 0.9882892370223999,
      "learning_rate": 0.0001878280274726504,
      "loss": 3.8729,
      "step": 65290
    },
    {
      "epoch": 0.060870166056469666,
      "grad_norm": 1.1714569330215454,
      "learning_rate": 0.0001878261631505552,
      "loss": 3.8871,
      "step": 65300
    },
    {
      "epoch": 0.06087948767454875,
      "grad_norm": 1.1460435390472412,
      "learning_rate": 0.00018782429882846,
      "loss": 4.0053,
      "step": 65310
    },
    {
      "epoch": 0.06088880929262785,
      "grad_norm": 1.0321425199508667,
      "learning_rate": 0.0001878224345063648,
      "loss": 3.8605,
      "step": 65320
    },
    {
      "epoch": 0.06089813091070694,
      "grad_norm": 1.1069554090499878,
      "learning_rate": 0.0001878205701842696,
      "loss": 3.8457,
      "step": 65330
    },
    {
      "epoch": 0.06090745252878603,
      "grad_norm": 1.1717329025268555,
      "learning_rate": 0.0001878187058621744,
      "loss": 3.6192,
      "step": 65340
    },
    {
      "epoch": 0.060916774146865124,
      "grad_norm": 1.1261985301971436,
      "learning_rate": 0.0001878168415400792,
      "loss": 4.0244,
      "step": 65350
    },
    {
      "epoch": 0.06092609576494421,
      "grad_norm": 0.9508065581321716,
      "learning_rate": 0.000187814977217984,
      "loss": 3.8757,
      "step": 65360
    },
    {
      "epoch": 0.060935417383023306,
      "grad_norm": 1.0166491270065308,
      "learning_rate": 0.00018781311289588882,
      "loss": 3.94,
      "step": 65370
    },
    {
      "epoch": 0.0609447390011024,
      "grad_norm": 0.9945535659790039,
      "learning_rate": 0.0001878112485737936,
      "loss": 3.941,
      "step": 65380
    },
    {
      "epoch": 0.06095406061918149,
      "grad_norm": 1.1650303602218628,
      "learning_rate": 0.0001878093842516984,
      "loss": 4.0154,
      "step": 65390
    },
    {
      "epoch": 0.06096338223726058,
      "grad_norm": 0.9825899004936218,
      "learning_rate": 0.0001878075199296032,
      "loss": 3.8303,
      "step": 65400
    },
    {
      "epoch": 0.06097270385533967,
      "grad_norm": 1.0800689458847046,
      "learning_rate": 0.00018780565560750802,
      "loss": 3.8158,
      "step": 65410
    },
    {
      "epoch": 0.060982025473418765,
      "grad_norm": 1.0032302141189575,
      "learning_rate": 0.00018780379128541282,
      "loss": 3.8608,
      "step": 65420
    },
    {
      "epoch": 0.06099134709149786,
      "grad_norm": 1.0982387065887451,
      "learning_rate": 0.0001878019269633176,
      "loss": 4.0549,
      "step": 65430
    },
    {
      "epoch": 0.06100066870957695,
      "grad_norm": 1.0929652452468872,
      "learning_rate": 0.0001878000626412224,
      "loss": 3.9835,
      "step": 65440
    },
    {
      "epoch": 0.06100999032765604,
      "grad_norm": 1.0581918954849243,
      "learning_rate": 0.0001877981983191272,
      "loss": 3.7724,
      "step": 65450
    },
    {
      "epoch": 0.061019311945735136,
      "grad_norm": 1.111418604850769,
      "learning_rate": 0.000187796333997032,
      "loss": 4.0294,
      "step": 65460
    },
    {
      "epoch": 0.06102863356381422,
      "grad_norm": 1.2193855047225952,
      "learning_rate": 0.00018779446967493683,
      "loss": 3.7392,
      "step": 65470
    },
    {
      "epoch": 0.06103795518189332,
      "grad_norm": 1.250640630722046,
      "learning_rate": 0.0001877926053528416,
      "loss": 3.706,
      "step": 65480
    },
    {
      "epoch": 0.061047276799972405,
      "grad_norm": 1.283843755722046,
      "learning_rate": 0.0001877907410307464,
      "loss": 4.0715,
      "step": 65490
    },
    {
      "epoch": 0.0610565984180515,
      "grad_norm": 1.2084918022155762,
      "learning_rate": 0.00018778887670865122,
      "loss": 3.8542,
      "step": 65500
    },
    {
      "epoch": 0.061065920036130594,
      "grad_norm": 1.1027501821517944,
      "learning_rate": 0.000187787012386556,
      "loss": 3.7013,
      "step": 65510
    },
    {
      "epoch": 0.06107524165420968,
      "grad_norm": 0.9270852208137512,
      "learning_rate": 0.0001877851480644608,
      "loss": 3.9552,
      "step": 65520
    },
    {
      "epoch": 0.061084563272288776,
      "grad_norm": 1.0088483095169067,
      "learning_rate": 0.0001877832837423656,
      "loss": 3.9938,
      "step": 65530
    },
    {
      "epoch": 0.061093884890367864,
      "grad_norm": 1.1117620468139648,
      "learning_rate": 0.00018778141942027042,
      "loss": 3.8826,
      "step": 65540
    },
    {
      "epoch": 0.06110320650844696,
      "grad_norm": 1.2652055025100708,
      "learning_rate": 0.00018777955509817522,
      "loss": 3.9509,
      "step": 65550
    },
    {
      "epoch": 0.06111252812652605,
      "grad_norm": 1.0904264450073242,
      "learning_rate": 0.00018777769077608,
      "loss": 3.9116,
      "step": 65560
    },
    {
      "epoch": 0.06112184974460514,
      "grad_norm": 1.1127004623413086,
      "learning_rate": 0.0001877758264539848,
      "loss": 3.8312,
      "step": 65570
    },
    {
      "epoch": 0.061131171362684235,
      "grad_norm": 1.1680891513824463,
      "learning_rate": 0.00018777396213188961,
      "loss": 3.8827,
      "step": 65580
    },
    {
      "epoch": 0.06114049298076333,
      "grad_norm": 0.9372225403785706,
      "learning_rate": 0.00018777209780979442,
      "loss": 3.8241,
      "step": 65590
    },
    {
      "epoch": 0.06114981459884242,
      "grad_norm": 1.175992727279663,
      "learning_rate": 0.00018777023348769923,
      "loss": 3.9705,
      "step": 65600
    },
    {
      "epoch": 0.06115913621692151,
      "grad_norm": 0.9638800024986267,
      "learning_rate": 0.000187768369165604,
      "loss": 3.7743,
      "step": 65610
    },
    {
      "epoch": 0.0611684578350006,
      "grad_norm": 1.1188318729400635,
      "learning_rate": 0.0001877665048435088,
      "loss": 3.9898,
      "step": 65620
    },
    {
      "epoch": 0.06117777945307969,
      "grad_norm": 0.8196282386779785,
      "learning_rate": 0.0001877646405214136,
      "loss": 4.1177,
      "step": 65630
    },
    {
      "epoch": 0.06118710107115879,
      "grad_norm": 1.0515269041061401,
      "learning_rate": 0.00018776277619931842,
      "loss": 3.8649,
      "step": 65640
    },
    {
      "epoch": 0.061196422689237875,
      "grad_norm": 1.0225671529769897,
      "learning_rate": 0.00018776091187722323,
      "loss": 3.8838,
      "step": 65650
    },
    {
      "epoch": 0.06120574430731697,
      "grad_norm": 1.4495176076889038,
      "learning_rate": 0.000187759047555128,
      "loss": 3.947,
      "step": 65660
    },
    {
      "epoch": 0.061215065925396064,
      "grad_norm": 1.0319181680679321,
      "learning_rate": 0.00018775718323303282,
      "loss": 3.8525,
      "step": 65670
    },
    {
      "epoch": 0.06122438754347515,
      "grad_norm": 1.1777392625808716,
      "learning_rate": 0.0001877553189109376,
      "loss": 3.7743,
      "step": 65680
    },
    {
      "epoch": 0.061233709161554246,
      "grad_norm": 1.08417809009552,
      "learning_rate": 0.0001877534545888424,
      "loss": 4.0121,
      "step": 65690
    },
    {
      "epoch": 0.061243030779633334,
      "grad_norm": 1.1597872972488403,
      "learning_rate": 0.00018775159026674724,
      "loss": 3.9807,
      "step": 65700
    },
    {
      "epoch": 0.06125235239771243,
      "grad_norm": 1.12163507938385,
      "learning_rate": 0.00018774972594465201,
      "loss": 3.9181,
      "step": 65710
    },
    {
      "epoch": 0.06126167401579152,
      "grad_norm": 0.9340000152587891,
      "learning_rate": 0.00018774786162255682,
      "loss": 3.9199,
      "step": 65720
    },
    {
      "epoch": 0.06127099563387061,
      "grad_norm": 1.1376882791519165,
      "learning_rate": 0.0001877459973004616,
      "loss": 3.855,
      "step": 65730
    },
    {
      "epoch": 0.061280317251949705,
      "grad_norm": 1.1718826293945312,
      "learning_rate": 0.0001877441329783664,
      "loss": 3.9055,
      "step": 65740
    },
    {
      "epoch": 0.06128963887002879,
      "grad_norm": 0.9745815992355347,
      "learning_rate": 0.0001877422686562712,
      "loss": 3.8809,
      "step": 65750
    },
    {
      "epoch": 0.06129896048810789,
      "grad_norm": 1.0059306621551514,
      "learning_rate": 0.00018774040433417602,
      "loss": 3.6881,
      "step": 65760
    },
    {
      "epoch": 0.06130828210618698,
      "grad_norm": 1.0907567739486694,
      "learning_rate": 0.00018773854001208083,
      "loss": 3.9715,
      "step": 65770
    },
    {
      "epoch": 0.06131760372426607,
      "grad_norm": 1.1413068771362305,
      "learning_rate": 0.0001877366756899856,
      "loss": 3.9049,
      "step": 65780
    },
    {
      "epoch": 0.061326925342345164,
      "grad_norm": 1.1578477621078491,
      "learning_rate": 0.0001877348113678904,
      "loss": 3.9533,
      "step": 65790
    },
    {
      "epoch": 0.06133624696042426,
      "grad_norm": 1.3245000839233398,
      "learning_rate": 0.00018773294704579522,
      "loss": 3.9901,
      "step": 65800
    },
    {
      "epoch": 0.061345568578503346,
      "grad_norm": 1.1173350811004639,
      "learning_rate": 0.00018773108272370002,
      "loss": 3.8866,
      "step": 65810
    },
    {
      "epoch": 0.06135489019658244,
      "grad_norm": 1.082315444946289,
      "learning_rate": 0.00018772921840160483,
      "loss": 3.9881,
      "step": 65820
    },
    {
      "epoch": 0.06136421181466153,
      "grad_norm": 1.1203800439834595,
      "learning_rate": 0.0001877273540795096,
      "loss": 4.0521,
      "step": 65830
    },
    {
      "epoch": 0.06137353343274062,
      "grad_norm": 1.0061825513839722,
      "learning_rate": 0.00018772548975741442,
      "loss": 4.0967,
      "step": 65840
    },
    {
      "epoch": 0.06138285505081972,
      "grad_norm": 1.1144152879714966,
      "learning_rate": 0.00018772362543531922,
      "loss": 3.9287,
      "step": 65850
    },
    {
      "epoch": 0.061392176668898804,
      "grad_norm": 1.1285017728805542,
      "learning_rate": 0.00018772176111322403,
      "loss": 3.8453,
      "step": 65860
    },
    {
      "epoch": 0.0614014982869779,
      "grad_norm": 1.200507640838623,
      "learning_rate": 0.00018771989679112883,
      "loss": 3.782,
      "step": 65870
    },
    {
      "epoch": 0.061410819905056986,
      "grad_norm": 1.060663104057312,
      "learning_rate": 0.0001877180324690336,
      "loss": 3.7346,
      "step": 65880
    },
    {
      "epoch": 0.06142014152313608,
      "grad_norm": 1.1962082386016846,
      "learning_rate": 0.00018771616814693842,
      "loss": 3.8943,
      "step": 65890
    },
    {
      "epoch": 0.061429463141215175,
      "grad_norm": 1.0443426370620728,
      "learning_rate": 0.00018771430382484323,
      "loss": 3.9344,
      "step": 65900
    },
    {
      "epoch": 0.06143878475929426,
      "grad_norm": 1.078098177909851,
      "learning_rate": 0.000187712439502748,
      "loss": 3.8749,
      "step": 65910
    },
    {
      "epoch": 0.06144810637737336,
      "grad_norm": 1.0558950901031494,
      "learning_rate": 0.00018771057518065284,
      "loss": 4.1104,
      "step": 65920
    },
    {
      "epoch": 0.06145742799545245,
      "grad_norm": 1.0212111473083496,
      "learning_rate": 0.00018770871085855762,
      "loss": 3.998,
      "step": 65930
    },
    {
      "epoch": 0.06146674961353154,
      "grad_norm": 1.2884374856948853,
      "learning_rate": 0.00018770684653646242,
      "loss": 3.9256,
      "step": 65940
    },
    {
      "epoch": 0.061476071231610634,
      "grad_norm": 1.0767021179199219,
      "learning_rate": 0.00018770498221436723,
      "loss": 3.9849,
      "step": 65950
    },
    {
      "epoch": 0.06148539284968972,
      "grad_norm": 1.024502158164978,
      "learning_rate": 0.000187703117892272,
      "loss": 3.9368,
      "step": 65960
    },
    {
      "epoch": 0.061494714467768816,
      "grad_norm": 1.1083717346191406,
      "learning_rate": 0.00018770125357017682,
      "loss": 3.8064,
      "step": 65970
    },
    {
      "epoch": 0.06150403608584791,
      "grad_norm": 1.0289098024368286,
      "learning_rate": 0.00018769938924808162,
      "loss": 4.0109,
      "step": 65980
    },
    {
      "epoch": 0.061513357703927,
      "grad_norm": 1.0022804737091064,
      "learning_rate": 0.00018769752492598643,
      "loss": 3.9764,
      "step": 65990
    },
    {
      "epoch": 0.06152267932200609,
      "grad_norm": 1.1410069465637207,
      "learning_rate": 0.00018769566060389124,
      "loss": 3.8363,
      "step": 66000
    },
    {
      "epoch": 0.06153200094008519,
      "grad_norm": 1.084214687347412,
      "learning_rate": 0.00018769379628179601,
      "loss": 4.0081,
      "step": 66010
    },
    {
      "epoch": 0.061541322558164274,
      "grad_norm": 1.0372861623764038,
      "learning_rate": 0.00018769193195970082,
      "loss": 3.9565,
      "step": 66020
    },
    {
      "epoch": 0.06155064417624337,
      "grad_norm": 1.1324331760406494,
      "learning_rate": 0.00018769006763760563,
      "loss": 3.9047,
      "step": 66030
    },
    {
      "epoch": 0.061559965794322456,
      "grad_norm": 0.990569531917572,
      "learning_rate": 0.00018768820331551043,
      "loss": 3.9588,
      "step": 66040
    },
    {
      "epoch": 0.06156928741240155,
      "grad_norm": 1.0756276845932007,
      "learning_rate": 0.00018768633899341524,
      "loss": 4.0051,
      "step": 66050
    },
    {
      "epoch": 0.061578609030480645,
      "grad_norm": 1.0506727695465088,
      "learning_rate": 0.00018768447467132002,
      "loss": 3.9342,
      "step": 66060
    },
    {
      "epoch": 0.06158793064855973,
      "grad_norm": 1.127631664276123,
      "learning_rate": 0.00018768261034922482,
      "loss": 3.924,
      "step": 66070
    },
    {
      "epoch": 0.06159725226663883,
      "grad_norm": 1.1534318923950195,
      "learning_rate": 0.0001876807460271296,
      "loss": 3.7261,
      "step": 66080
    },
    {
      "epoch": 0.061606573884717915,
      "grad_norm": 1.1030381917953491,
      "learning_rate": 0.00018767888170503444,
      "loss": 3.7187,
      "step": 66090
    },
    {
      "epoch": 0.06161589550279701,
      "grad_norm": 1.2400290966033936,
      "learning_rate": 0.00018767701738293924,
      "loss": 3.8268,
      "step": 66100
    },
    {
      "epoch": 0.061625217120876104,
      "grad_norm": 0.9376870393753052,
      "learning_rate": 0.00018767515306084402,
      "loss": 3.9558,
      "step": 66110
    },
    {
      "epoch": 0.06163453873895519,
      "grad_norm": 0.985747754573822,
      "learning_rate": 0.00018767328873874883,
      "loss": 3.9339,
      "step": 66120
    },
    {
      "epoch": 0.061643860357034286,
      "grad_norm": 1.1251182556152344,
      "learning_rate": 0.0001876714244166536,
      "loss": 3.9323,
      "step": 66130
    },
    {
      "epoch": 0.06165318197511338,
      "grad_norm": 1.0873751640319824,
      "learning_rate": 0.00018766956009455841,
      "loss": 3.8811,
      "step": 66140
    },
    {
      "epoch": 0.06166250359319247,
      "grad_norm": 1.2642227411270142,
      "learning_rate": 0.00018766769577246325,
      "loss": 3.7327,
      "step": 66150
    },
    {
      "epoch": 0.06167182521127156,
      "grad_norm": 0.9467036128044128,
      "learning_rate": 0.00018766583145036803,
      "loss": 3.971,
      "step": 66160
    },
    {
      "epoch": 0.06168114682935065,
      "grad_norm": 1.1692122220993042,
      "learning_rate": 0.00018766396712827283,
      "loss": 3.8857,
      "step": 66170
    },
    {
      "epoch": 0.061690468447429744,
      "grad_norm": 1.248228907585144,
      "learning_rate": 0.0001876621028061776,
      "loss": 3.8723,
      "step": 66180
    },
    {
      "epoch": 0.06169979006550884,
      "grad_norm": 1.1718950271606445,
      "learning_rate": 0.00018766023848408242,
      "loss": 3.8199,
      "step": 66190
    },
    {
      "epoch": 0.061709111683587926,
      "grad_norm": 1.0463827848434448,
      "learning_rate": 0.00018765837416198723,
      "loss": 3.8485,
      "step": 66200
    },
    {
      "epoch": 0.06171843330166702,
      "grad_norm": 1.0649807453155518,
      "learning_rate": 0.00018765650983989203,
      "loss": 3.7785,
      "step": 66210
    },
    {
      "epoch": 0.06172775491974611,
      "grad_norm": 0.9561249017715454,
      "learning_rate": 0.00018765464551779684,
      "loss": 4.0261,
      "step": 66220
    },
    {
      "epoch": 0.0617370765378252,
      "grad_norm": 1.0368080139160156,
      "learning_rate": 0.00018765278119570162,
      "loss": 3.8609,
      "step": 66230
    },
    {
      "epoch": 0.0617463981559043,
      "grad_norm": 1.1983909606933594,
      "learning_rate": 0.00018765091687360642,
      "loss": 3.8288,
      "step": 66240
    },
    {
      "epoch": 0.061755719773983385,
      "grad_norm": 1.134337306022644,
      "learning_rate": 0.00018764905255151123,
      "loss": 4.0481,
      "step": 66250
    },
    {
      "epoch": 0.06176504139206248,
      "grad_norm": 1.1098724603652954,
      "learning_rate": 0.00018764718822941604,
      "loss": 4.0173,
      "step": 66260
    },
    {
      "epoch": 0.061774363010141574,
      "grad_norm": 1.1411700248718262,
      "learning_rate": 0.00018764532390732084,
      "loss": 3.8998,
      "step": 66270
    },
    {
      "epoch": 0.06178368462822066,
      "grad_norm": 1.1959947347640991,
      "learning_rate": 0.00018764345958522562,
      "loss": 3.8101,
      "step": 66280
    },
    {
      "epoch": 0.061793006246299756,
      "grad_norm": 1.1363904476165771,
      "learning_rate": 0.00018764159526313043,
      "loss": 3.8712,
      "step": 66290
    },
    {
      "epoch": 0.06180232786437884,
      "grad_norm": 1.0527101755142212,
      "learning_rate": 0.0001876397309410352,
      "loss": 3.9542,
      "step": 66300
    },
    {
      "epoch": 0.06181164948245794,
      "grad_norm": 1.041581153869629,
      "learning_rate": 0.00018763786661894004,
      "loss": 4.1687,
      "step": 66310
    },
    {
      "epoch": 0.06182097110053703,
      "grad_norm": 1.0097358226776123,
      "learning_rate": 0.00018763600229684485,
      "loss": 4.0127,
      "step": 66320
    },
    {
      "epoch": 0.06183029271861612,
      "grad_norm": 1.1599644422531128,
      "learning_rate": 0.00018763413797474963,
      "loss": 4.1245,
      "step": 66330
    },
    {
      "epoch": 0.061839614336695214,
      "grad_norm": 1.1074005365371704,
      "learning_rate": 0.00018763227365265443,
      "loss": 3.7623,
      "step": 66340
    },
    {
      "epoch": 0.06184893595477431,
      "grad_norm": 1.1067888736724854,
      "learning_rate": 0.0001876304093305592,
      "loss": 4.0206,
      "step": 66350
    },
    {
      "epoch": 0.061858257572853396,
      "grad_norm": 1.0768077373504639,
      "learning_rate": 0.00018762854500846402,
      "loss": 4.0898,
      "step": 66360
    },
    {
      "epoch": 0.06186757919093249,
      "grad_norm": 1.0623459815979004,
      "learning_rate": 0.00018762668068636885,
      "loss": 3.9064,
      "step": 66370
    },
    {
      "epoch": 0.06187690080901158,
      "grad_norm": 1.1216696500778198,
      "learning_rate": 0.00018762481636427363,
      "loss": 3.9627,
      "step": 66380
    },
    {
      "epoch": 0.06188622242709067,
      "grad_norm": 1.0790938138961792,
      "learning_rate": 0.00018762295204217844,
      "loss": 3.9373,
      "step": 66390
    },
    {
      "epoch": 0.06189554404516977,
      "grad_norm": 1.1507973670959473,
      "learning_rate": 0.00018762108772008322,
      "loss": 3.7598,
      "step": 66400
    },
    {
      "epoch": 0.061904865663248855,
      "grad_norm": 0.9381277561187744,
      "learning_rate": 0.00018761922339798802,
      "loss": 3.9364,
      "step": 66410
    },
    {
      "epoch": 0.06191418728132795,
      "grad_norm": 0.9642226099967957,
      "learning_rate": 0.00018761735907589283,
      "loss": 3.8669,
      "step": 66420
    },
    {
      "epoch": 0.06192350889940704,
      "grad_norm": 1.0084564685821533,
      "learning_rate": 0.00018761549475379764,
      "loss": 4.0738,
      "step": 66430
    },
    {
      "epoch": 0.06193283051748613,
      "grad_norm": 1.191367745399475,
      "learning_rate": 0.00018761363043170244,
      "loss": 3.9275,
      "step": 66440
    },
    {
      "epoch": 0.061942152135565226,
      "grad_norm": 1.235923409461975,
      "learning_rate": 0.00018761176610960722,
      "loss": 3.9603,
      "step": 66450
    },
    {
      "epoch": 0.061951473753644314,
      "grad_norm": 1.1739673614501953,
      "learning_rate": 0.00018760990178751203,
      "loss": 3.7817,
      "step": 66460
    },
    {
      "epoch": 0.06196079537172341,
      "grad_norm": 1.0137386322021484,
      "learning_rate": 0.00018760803746541683,
      "loss": 4.0796,
      "step": 66470
    },
    {
      "epoch": 0.0619701169898025,
      "grad_norm": 1.0221152305603027,
      "learning_rate": 0.00018760617314332164,
      "loss": 4.0144,
      "step": 66480
    },
    {
      "epoch": 0.06197943860788159,
      "grad_norm": 1.169364094734192,
      "learning_rate": 0.00018760430882122645,
      "loss": 3.8792,
      "step": 66490
    },
    {
      "epoch": 0.061988760225960685,
      "grad_norm": 1.1486319303512573,
      "learning_rate": 0.00018760244449913123,
      "loss": 3.9535,
      "step": 66500
    },
    {
      "epoch": 0.06199808184403977,
      "grad_norm": 1.1033669710159302,
      "learning_rate": 0.00018760058017703603,
      "loss": 3.9343,
      "step": 66510
    },
    {
      "epoch": 0.06200740346211887,
      "grad_norm": 1.0419539213180542,
      "learning_rate": 0.00018759871585494084,
      "loss": 4.014,
      "step": 66520
    },
    {
      "epoch": 0.06201672508019796,
      "grad_norm": 1.1352756023406982,
      "learning_rate": 0.00018759685153284562,
      "loss": 3.8223,
      "step": 66530
    },
    {
      "epoch": 0.06202604669827705,
      "grad_norm": 0.8837353587150574,
      "learning_rate": 0.00018759498721075045,
      "loss": 3.7455,
      "step": 66540
    },
    {
      "epoch": 0.06203536831635614,
      "grad_norm": 0.8901716470718384,
      "learning_rate": 0.00018759312288865523,
      "loss": 3.9239,
      "step": 66550
    },
    {
      "epoch": 0.06204468993443523,
      "grad_norm": 1.2377784252166748,
      "learning_rate": 0.00018759125856656004,
      "loss": 3.9293,
      "step": 66560
    },
    {
      "epoch": 0.062054011552514325,
      "grad_norm": 1.1261968612670898,
      "learning_rate": 0.00018758939424446484,
      "loss": 3.7916,
      "step": 66570
    },
    {
      "epoch": 0.06206333317059342,
      "grad_norm": 1.0672205686569214,
      "learning_rate": 0.00018758752992236962,
      "loss": 3.8081,
      "step": 66580
    },
    {
      "epoch": 0.06207265478867251,
      "grad_norm": 0.9170063734054565,
      "learning_rate": 0.00018758566560027443,
      "loss": 4.0073,
      "step": 66590
    },
    {
      "epoch": 0.0620819764067516,
      "grad_norm": 0.8314833045005798,
      "learning_rate": 0.00018758380127817923,
      "loss": 3.9777,
      "step": 66600
    },
    {
      "epoch": 0.062091298024830696,
      "grad_norm": 1.0282893180847168,
      "learning_rate": 0.00018758193695608404,
      "loss": 3.9428,
      "step": 66610
    },
    {
      "epoch": 0.062100619642909784,
      "grad_norm": 1.1789143085479736,
      "learning_rate": 0.00018758007263398885,
      "loss": 3.8281,
      "step": 66620
    },
    {
      "epoch": 0.06210994126098888,
      "grad_norm": 1.2427901029586792,
      "learning_rate": 0.00018757820831189363,
      "loss": 3.8832,
      "step": 66630
    },
    {
      "epoch": 0.062119262879067966,
      "grad_norm": 1.1586785316467285,
      "learning_rate": 0.00018757634398979843,
      "loss": 4.0039,
      "step": 66640
    },
    {
      "epoch": 0.06212858449714706,
      "grad_norm": 1.00932776927948,
      "learning_rate": 0.00018757447966770324,
      "loss": 4.0319,
      "step": 66650
    },
    {
      "epoch": 0.062137906115226155,
      "grad_norm": 0.906520426273346,
      "learning_rate": 0.00018757261534560804,
      "loss": 3.9352,
      "step": 66660
    },
    {
      "epoch": 0.06214722773330524,
      "grad_norm": 1.1911226511001587,
      "learning_rate": 0.00018757075102351285,
      "loss": 3.8137,
      "step": 66670
    },
    {
      "epoch": 0.06215654935138434,
      "grad_norm": 1.240696668624878,
      "learning_rate": 0.00018756888670141763,
      "loss": 3.8444,
      "step": 66680
    },
    {
      "epoch": 0.06216587096946343,
      "grad_norm": 1.3002357482910156,
      "learning_rate": 0.00018756702237932244,
      "loss": 3.839,
      "step": 66690
    },
    {
      "epoch": 0.06217519258754252,
      "grad_norm": 1.0251606702804565,
      "learning_rate": 0.00018756515805722724,
      "loss": 3.738,
      "step": 66700
    },
    {
      "epoch": 0.06218451420562161,
      "grad_norm": 1.0876928567886353,
      "learning_rate": 0.00018756329373513205,
      "loss": 3.7779,
      "step": 66710
    },
    {
      "epoch": 0.0621938358237007,
      "grad_norm": 1.0446676015853882,
      "learning_rate": 0.00018756142941303686,
      "loss": 3.9957,
      "step": 66720
    },
    {
      "epoch": 0.062203157441779795,
      "grad_norm": 1.254152774810791,
      "learning_rate": 0.00018755956509094163,
      "loss": 3.9366,
      "step": 66730
    },
    {
      "epoch": 0.06221247905985889,
      "grad_norm": 1.01932954788208,
      "learning_rate": 0.00018755770076884644,
      "loss": 4.038,
      "step": 66740
    },
    {
      "epoch": 0.06222180067793798,
      "grad_norm": 0.9420297145843506,
      "learning_rate": 0.00018755583644675122,
      "loss": 3.8755,
      "step": 66750
    },
    {
      "epoch": 0.06223112229601707,
      "grad_norm": 1.1538032293319702,
      "learning_rate": 0.00018755397212465605,
      "loss": 3.9881,
      "step": 66760
    },
    {
      "epoch": 0.06224044391409616,
      "grad_norm": 1.07498300075531,
      "learning_rate": 0.00018755210780256086,
      "loss": 3.7712,
      "step": 66770
    },
    {
      "epoch": 0.062249765532175254,
      "grad_norm": 1.0905981063842773,
      "learning_rate": 0.00018755024348046564,
      "loss": 3.6466,
      "step": 66780
    },
    {
      "epoch": 0.06225908715025435,
      "grad_norm": 0.9870144128799438,
      "learning_rate": 0.00018754837915837045,
      "loss": 4.0374,
      "step": 66790
    },
    {
      "epoch": 0.062268408768333436,
      "grad_norm": 0.8476836085319519,
      "learning_rate": 0.00018754651483627522,
      "loss": 3.9158,
      "step": 66800
    },
    {
      "epoch": 0.06227773038641253,
      "grad_norm": 1.1218674182891846,
      "learning_rate": 0.00018754465051418003,
      "loss": 3.9955,
      "step": 66810
    },
    {
      "epoch": 0.062287052004491625,
      "grad_norm": 1.0781919956207275,
      "learning_rate": 0.00018754278619208486,
      "loss": 3.9886,
      "step": 66820
    },
    {
      "epoch": 0.06229637362257071,
      "grad_norm": 1.0152679681777954,
      "learning_rate": 0.00018754092186998964,
      "loss": 3.8144,
      "step": 66830
    },
    {
      "epoch": 0.06230569524064981,
      "grad_norm": 0.9620774984359741,
      "learning_rate": 0.00018753905754789445,
      "loss": 3.8614,
      "step": 66840
    },
    {
      "epoch": 0.062315016858728894,
      "grad_norm": 1.0254549980163574,
      "learning_rate": 0.00018753719322579923,
      "loss": 3.991,
      "step": 66850
    },
    {
      "epoch": 0.06232433847680799,
      "grad_norm": 0.9723402857780457,
      "learning_rate": 0.00018753532890370404,
      "loss": 4.0549,
      "step": 66860
    },
    {
      "epoch": 0.06233366009488708,
      "grad_norm": 1.071231722831726,
      "learning_rate": 0.00018753346458160884,
      "loss": 3.7626,
      "step": 66870
    },
    {
      "epoch": 0.06234298171296617,
      "grad_norm": 1.0156809091567993,
      "learning_rate": 0.00018753160025951365,
      "loss": 3.8947,
      "step": 66880
    },
    {
      "epoch": 0.062352303331045265,
      "grad_norm": 1.1244357824325562,
      "learning_rate": 0.00018752973593741845,
      "loss": 3.8193,
      "step": 66890
    },
    {
      "epoch": 0.06236162494912435,
      "grad_norm": 1.158227562904358,
      "learning_rate": 0.00018752787161532323,
      "loss": 3.6624,
      "step": 66900
    },
    {
      "epoch": 0.06237094656720345,
      "grad_norm": 0.8931680917739868,
      "learning_rate": 0.00018752600729322804,
      "loss": 4.0294,
      "step": 66910
    },
    {
      "epoch": 0.06238026818528254,
      "grad_norm": 1.074162483215332,
      "learning_rate": 0.00018752414297113285,
      "loss": 3.9243,
      "step": 66920
    },
    {
      "epoch": 0.06238958980336163,
      "grad_norm": 1.161910891532898,
      "learning_rate": 0.00018752227864903765,
      "loss": 3.7068,
      "step": 66930
    },
    {
      "epoch": 0.062398911421440724,
      "grad_norm": 1.0702064037322998,
      "learning_rate": 0.00018752041432694246,
      "loss": 3.7358,
      "step": 66940
    },
    {
      "epoch": 0.06240823303951982,
      "grad_norm": 1.0546255111694336,
      "learning_rate": 0.00018751855000484724,
      "loss": 4.0028,
      "step": 66950
    },
    {
      "epoch": 0.062417554657598906,
      "grad_norm": 1.1016870737075806,
      "learning_rate": 0.00018751668568275204,
      "loss": 3.8839,
      "step": 66960
    },
    {
      "epoch": 0.062426876275678,
      "grad_norm": 1.145249366760254,
      "learning_rate": 0.00018751482136065685,
      "loss": 3.9274,
      "step": 66970
    },
    {
      "epoch": 0.06243619789375709,
      "grad_norm": 0.9215826392173767,
      "learning_rate": 0.00018751295703856163,
      "loss": 3.773,
      "step": 66980
    },
    {
      "epoch": 0.06244551951183618,
      "grad_norm": 1.2283815145492554,
      "learning_rate": 0.00018751109271646646,
      "loss": 3.7014,
      "step": 66990
    },
    {
      "epoch": 0.06245484112991528,
      "grad_norm": 1.1979734897613525,
      "learning_rate": 0.00018750922839437124,
      "loss": 3.7778,
      "step": 67000
    },
    {
      "epoch": 0.062464162747994365,
      "grad_norm": 1.0584328174591064,
      "learning_rate": 0.00018750736407227605,
      "loss": 3.8568,
      "step": 67010
    },
    {
      "epoch": 0.06247348436607346,
      "grad_norm": 1.0970019102096558,
      "learning_rate": 0.00018750549975018086,
      "loss": 3.9443,
      "step": 67020
    },
    {
      "epoch": 0.062482805984152553,
      "grad_norm": 1.0791354179382324,
      "learning_rate": 0.00018750363542808563,
      "loss": 3.9868,
      "step": 67030
    },
    {
      "epoch": 0.06249212760223164,
      "grad_norm": 0.9677533507347107,
      "learning_rate": 0.00018750177110599044,
      "loss": 4.0323,
      "step": 67040
    },
    {
      "epoch": 0.06250144922031073,
      "grad_norm": 1.0406275987625122,
      "learning_rate": 0.00018749990678389525,
      "loss": 3.9126,
      "step": 67050
    },
    {
      "epoch": 0.06251077083838982,
      "grad_norm": 0.9457435607910156,
      "learning_rate": 0.00018749804246180005,
      "loss": 3.7831,
      "step": 67060
    },
    {
      "epoch": 0.06252009245646892,
      "grad_norm": 1.1192365884780884,
      "learning_rate": 0.00018749617813970486,
      "loss": 3.8492,
      "step": 67070
    },
    {
      "epoch": 0.06252941407454801,
      "grad_norm": 1.1498239040374756,
      "learning_rate": 0.00018749431381760964,
      "loss": 3.8687,
      "step": 67080
    },
    {
      "epoch": 0.0625387356926271,
      "grad_norm": 0.9136620163917542,
      "learning_rate": 0.00018749244949551445,
      "loss": 3.8977,
      "step": 67090
    },
    {
      "epoch": 0.06254805731070619,
      "grad_norm": 0.9530852437019348,
      "learning_rate": 0.00018749058517341925,
      "loss": 3.9954,
      "step": 67100
    },
    {
      "epoch": 0.06255737892878528,
      "grad_norm": 0.9067173600196838,
      "learning_rate": 0.00018748872085132406,
      "loss": 4.1138,
      "step": 67110
    },
    {
      "epoch": 0.06256670054686438,
      "grad_norm": 1.1119097471237183,
      "learning_rate": 0.00018748685652922886,
      "loss": 3.907,
      "step": 67120
    },
    {
      "epoch": 0.06257602216494347,
      "grad_norm": 0.9156941175460815,
      "learning_rate": 0.00018748499220713364,
      "loss": 4.0228,
      "step": 67130
    },
    {
      "epoch": 0.06258534378302257,
      "grad_norm": 0.9516434669494629,
      "learning_rate": 0.00018748312788503845,
      "loss": 4.03,
      "step": 67140
    },
    {
      "epoch": 0.06259466540110165,
      "grad_norm": 1.0067131519317627,
      "learning_rate": 0.00018748126356294326,
      "loss": 3.815,
      "step": 67150
    },
    {
      "epoch": 0.06260398701918074,
      "grad_norm": 1.2829419374465942,
      "learning_rate": 0.00018747939924084806,
      "loss": 3.8918,
      "step": 67160
    },
    {
      "epoch": 0.06261330863725983,
      "grad_norm": 1.0963943004608154,
      "learning_rate": 0.00018747753491875287,
      "loss": 4.0834,
      "step": 67170
    },
    {
      "epoch": 0.06262263025533893,
      "grad_norm": 0.995902955532074,
      "learning_rate": 0.00018747567059665765,
      "loss": 3.898,
      "step": 67180
    },
    {
      "epoch": 0.06263195187341802,
      "grad_norm": 1.1328662633895874,
      "learning_rate": 0.00018747380627456245,
      "loss": 3.8129,
      "step": 67190
    },
    {
      "epoch": 0.06264127349149712,
      "grad_norm": 0.951775074005127,
      "learning_rate": 0.00018747194195246723,
      "loss": 3.7871,
      "step": 67200
    },
    {
      "epoch": 0.0626505951095762,
      "grad_norm": 1.1809749603271484,
      "learning_rate": 0.00018747007763037207,
      "loss": 4.069,
      "step": 67210
    },
    {
      "epoch": 0.0626599167276553,
      "grad_norm": 1.026574969291687,
      "learning_rate": 0.00018746821330827687,
      "loss": 4.1002,
      "step": 67220
    },
    {
      "epoch": 0.06266923834573439,
      "grad_norm": 0.9884170889854431,
      "learning_rate": 0.00018746634898618165,
      "loss": 3.8874,
      "step": 67230
    },
    {
      "epoch": 0.06267855996381348,
      "grad_norm": 1.1149104833602905,
      "learning_rate": 0.00018746448466408646,
      "loss": 3.8027,
      "step": 67240
    },
    {
      "epoch": 0.06268788158189258,
      "grad_norm": 1.0280704498291016,
      "learning_rate": 0.00018746262034199124,
      "loss": 4.1039,
      "step": 67250
    },
    {
      "epoch": 0.06269720319997166,
      "grad_norm": 1.2269155979156494,
      "learning_rate": 0.00018746075601989604,
      "loss": 3.8102,
      "step": 67260
    },
    {
      "epoch": 0.06270652481805075,
      "grad_norm": 1.3593467473983765,
      "learning_rate": 0.00018745889169780088,
      "loss": 3.8675,
      "step": 67270
    },
    {
      "epoch": 0.06271584643612985,
      "grad_norm": 0.9402633905410767,
      "learning_rate": 0.00018745702737570566,
      "loss": 3.8962,
      "step": 67280
    },
    {
      "epoch": 0.06272516805420894,
      "grad_norm": 0.9444281458854675,
      "learning_rate": 0.00018745516305361046,
      "loss": 3.8793,
      "step": 67290
    },
    {
      "epoch": 0.06273448967228804,
      "grad_norm": 1.0783742666244507,
      "learning_rate": 0.00018745329873151524,
      "loss": 3.7716,
      "step": 67300
    },
    {
      "epoch": 0.06274381129036712,
      "grad_norm": 1.180952548980713,
      "learning_rate": 0.00018745143440942005,
      "loss": 3.8517,
      "step": 67310
    },
    {
      "epoch": 0.06275313290844621,
      "grad_norm": 1.1116952896118164,
      "learning_rate": 0.00018744957008732485,
      "loss": 3.8777,
      "step": 67320
    },
    {
      "epoch": 0.0627624545265253,
      "grad_norm": 0.930457592010498,
      "learning_rate": 0.00018744770576522966,
      "loss": 3.8438,
      "step": 67330
    },
    {
      "epoch": 0.0627717761446044,
      "grad_norm": 1.2205661535263062,
      "learning_rate": 0.00018744584144313447,
      "loss": 3.8284,
      "step": 67340
    },
    {
      "epoch": 0.0627810977626835,
      "grad_norm": 1.0616291761398315,
      "learning_rate": 0.00018744397712103925,
      "loss": 3.9587,
      "step": 67350
    },
    {
      "epoch": 0.06279041938076257,
      "grad_norm": 1.0157628059387207,
      "learning_rate": 0.00018744211279894405,
      "loss": 3.7583,
      "step": 67360
    },
    {
      "epoch": 0.06279974099884167,
      "grad_norm": 1.1203055381774902,
      "learning_rate": 0.00018744024847684886,
      "loss": 3.7676,
      "step": 67370
    },
    {
      "epoch": 0.06280906261692076,
      "grad_norm": 1.1508880853652954,
      "learning_rate": 0.00018743838415475367,
      "loss": 3.9656,
      "step": 67380
    },
    {
      "epoch": 0.06281838423499986,
      "grad_norm": 1.1220029592514038,
      "learning_rate": 0.00018743651983265847,
      "loss": 3.9285,
      "step": 67390
    },
    {
      "epoch": 0.06282770585307895,
      "grad_norm": 1.1235206127166748,
      "learning_rate": 0.00018743465551056325,
      "loss": 3.844,
      "step": 67400
    },
    {
      "epoch": 0.06283702747115803,
      "grad_norm": 1.0110057592391968,
      "learning_rate": 0.00018743279118846806,
      "loss": 3.728,
      "step": 67410
    },
    {
      "epoch": 0.06284634908923713,
      "grad_norm": 0.9719079732894897,
      "learning_rate": 0.00018743092686637286,
      "loss": 4.0524,
      "step": 67420
    },
    {
      "epoch": 0.06285567070731622,
      "grad_norm": 1.1734378337860107,
      "learning_rate": 0.00018742906254427764,
      "loss": 3.7975,
      "step": 67430
    },
    {
      "epoch": 0.06286499232539532,
      "grad_norm": 0.970500111579895,
      "learning_rate": 0.00018742719822218248,
      "loss": 3.869,
      "step": 67440
    },
    {
      "epoch": 0.06287431394347441,
      "grad_norm": 1.1658997535705566,
      "learning_rate": 0.00018742533390008726,
      "loss": 3.9906,
      "step": 67450
    },
    {
      "epoch": 0.0628836355615535,
      "grad_norm": 1.0693023204803467,
      "learning_rate": 0.00018742346957799206,
      "loss": 3.7654,
      "step": 67460
    },
    {
      "epoch": 0.06289295717963259,
      "grad_norm": 1.0276916027069092,
      "learning_rate": 0.00018742160525589687,
      "loss": 3.9949,
      "step": 67470
    },
    {
      "epoch": 0.06290227879771168,
      "grad_norm": 0.9404246211051941,
      "learning_rate": 0.00018741974093380165,
      "loss": 3.9889,
      "step": 67480
    },
    {
      "epoch": 0.06291160041579077,
      "grad_norm": 0.9527554512023926,
      "learning_rate": 0.00018741787661170645,
      "loss": 3.7995,
      "step": 67490
    },
    {
      "epoch": 0.06292092203386987,
      "grad_norm": 1.0761216878890991,
      "learning_rate": 0.00018741601228961126,
      "loss": 3.8146,
      "step": 67500
    },
    {
      "epoch": 0.06293024365194896,
      "grad_norm": 1.1218205690383911,
      "learning_rate": 0.00018741414796751607,
      "loss": 3.8945,
      "step": 67510
    },
    {
      "epoch": 0.06293956527002804,
      "grad_norm": 1.1108936071395874,
      "learning_rate": 0.00018741228364542087,
      "loss": 3.796,
      "step": 67520
    },
    {
      "epoch": 0.06294888688810714,
      "grad_norm": 1.1456272602081299,
      "learning_rate": 0.00018741041932332565,
      "loss": 3.9417,
      "step": 67530
    },
    {
      "epoch": 0.06295820850618623,
      "grad_norm": 1.0245463848114014,
      "learning_rate": 0.00018740855500123046,
      "loss": 3.8612,
      "step": 67540
    },
    {
      "epoch": 0.06296753012426533,
      "grad_norm": 0.8804943561553955,
      "learning_rate": 0.00018740669067913526,
      "loss": 3.758,
      "step": 67550
    },
    {
      "epoch": 0.06297685174234442,
      "grad_norm": 0.9710383415222168,
      "learning_rate": 0.00018740482635704007,
      "loss": 3.8258,
      "step": 67560
    },
    {
      "epoch": 0.0629861733604235,
      "grad_norm": 1.1105666160583496,
      "learning_rate": 0.00018740296203494488,
      "loss": 3.7116,
      "step": 67570
    },
    {
      "epoch": 0.0629954949785026,
      "grad_norm": 1.1308817863464355,
      "learning_rate": 0.00018740109771284966,
      "loss": 3.8864,
      "step": 67580
    },
    {
      "epoch": 0.06300481659658169,
      "grad_norm": 0.8792559504508972,
      "learning_rate": 0.00018739923339075446,
      "loss": 3.9426,
      "step": 67590
    },
    {
      "epoch": 0.06301413821466079,
      "grad_norm": 0.8660162687301636,
      "learning_rate": 0.00018739736906865927,
      "loss": 3.8367,
      "step": 67600
    },
    {
      "epoch": 0.06302345983273988,
      "grad_norm": 1.2245724201202393,
      "learning_rate": 0.00018739550474656407,
      "loss": 3.7866,
      "step": 67610
    },
    {
      "epoch": 0.06303278145081896,
      "grad_norm": 1.2911986112594604,
      "learning_rate": 0.00018739364042446885,
      "loss": 3.9206,
      "step": 67620
    },
    {
      "epoch": 0.06304210306889806,
      "grad_norm": 0.9456818103790283,
      "learning_rate": 0.00018739177610237366,
      "loss": 3.684,
      "step": 67630
    },
    {
      "epoch": 0.06305142468697715,
      "grad_norm": 1.0518876314163208,
      "learning_rate": 0.00018738991178027847,
      "loss": 4.1439,
      "step": 67640
    },
    {
      "epoch": 0.06306074630505625,
      "grad_norm": 1.024328351020813,
      "learning_rate": 0.00018738804745818325,
      "loss": 4.0062,
      "step": 67650
    },
    {
      "epoch": 0.06307006792313534,
      "grad_norm": 1.0490288734436035,
      "learning_rate": 0.00018738618313608808,
      "loss": 3.8331,
      "step": 67660
    },
    {
      "epoch": 0.06307938954121443,
      "grad_norm": 1.199557900428772,
      "learning_rate": 0.00018738431881399286,
      "loss": 3.8568,
      "step": 67670
    },
    {
      "epoch": 0.06308871115929351,
      "grad_norm": 1.3874891996383667,
      "learning_rate": 0.00018738245449189766,
      "loss": 3.9852,
      "step": 67680
    },
    {
      "epoch": 0.06309803277737261,
      "grad_norm": 1.185725450515747,
      "learning_rate": 0.00018738059016980247,
      "loss": 3.7282,
      "step": 67690
    },
    {
      "epoch": 0.0631073543954517,
      "grad_norm": 0.9379251003265381,
      "learning_rate": 0.00018737872584770725,
      "loss": 3.9206,
      "step": 67700
    },
    {
      "epoch": 0.0631166760135308,
      "grad_norm": 1.0857315063476562,
      "learning_rate": 0.00018737686152561206,
      "loss": 3.9058,
      "step": 67710
    },
    {
      "epoch": 0.06312599763160989,
      "grad_norm": 1.0316821336746216,
      "learning_rate": 0.00018737499720351686,
      "loss": 3.8145,
      "step": 67720
    },
    {
      "epoch": 0.06313531924968897,
      "grad_norm": 1.0512311458587646,
      "learning_rate": 0.00018737313288142167,
      "loss": 3.796,
      "step": 67730
    },
    {
      "epoch": 0.06314464086776807,
      "grad_norm": 1.1545113325119019,
      "learning_rate": 0.00018737126855932648,
      "loss": 4.1565,
      "step": 67740
    },
    {
      "epoch": 0.06315396248584716,
      "grad_norm": 1.0894851684570312,
      "learning_rate": 0.00018736940423723125,
      "loss": 3.7754,
      "step": 67750
    },
    {
      "epoch": 0.06316328410392626,
      "grad_norm": 1.2913717031478882,
      "learning_rate": 0.00018736753991513606,
      "loss": 3.9779,
      "step": 67760
    },
    {
      "epoch": 0.06317260572200535,
      "grad_norm": 1.0340118408203125,
      "learning_rate": 0.00018736567559304087,
      "loss": 3.8037,
      "step": 67770
    },
    {
      "epoch": 0.06318192734008443,
      "grad_norm": 1.2387787103652954,
      "learning_rate": 0.00018736381127094567,
      "loss": 3.9736,
      "step": 67780
    },
    {
      "epoch": 0.06319124895816353,
      "grad_norm": 1.0435937643051147,
      "learning_rate": 0.00018736194694885048,
      "loss": 4.0236,
      "step": 67790
    },
    {
      "epoch": 0.06320057057624262,
      "grad_norm": 1.0588651895523071,
      "learning_rate": 0.00018736008262675526,
      "loss": 3.8472,
      "step": 67800
    },
    {
      "epoch": 0.06320989219432172,
      "grad_norm": 0.9033811092376709,
      "learning_rate": 0.00018735821830466007,
      "loss": 4.0903,
      "step": 67810
    },
    {
      "epoch": 0.06321921381240081,
      "grad_norm": 0.9758737087249756,
      "learning_rate": 0.00018735635398256487,
      "loss": 3.7592,
      "step": 67820
    },
    {
      "epoch": 0.06322853543047989,
      "grad_norm": 1.1631038188934326,
      "learning_rate": 0.00018735448966046968,
      "loss": 3.9545,
      "step": 67830
    },
    {
      "epoch": 0.06323785704855898,
      "grad_norm": 1.1321356296539307,
      "learning_rate": 0.00018735262533837448,
      "loss": 3.8526,
      "step": 67840
    },
    {
      "epoch": 0.06324717866663808,
      "grad_norm": 1.031127691268921,
      "learning_rate": 0.00018735076101627926,
      "loss": 3.9127,
      "step": 67850
    },
    {
      "epoch": 0.06325650028471717,
      "grad_norm": 1.0548923015594482,
      "learning_rate": 0.00018734889669418407,
      "loss": 3.6949,
      "step": 67860
    },
    {
      "epoch": 0.06326582190279627,
      "grad_norm": 1.0753324031829834,
      "learning_rate": 0.00018734703237208885,
      "loss": 3.9508,
      "step": 67870
    },
    {
      "epoch": 0.06327514352087536,
      "grad_norm": 1.1075588464736938,
      "learning_rate": 0.00018734516804999366,
      "loss": 3.8291,
      "step": 67880
    },
    {
      "epoch": 0.06328446513895444,
      "grad_norm": 1.1075588464736938,
      "learning_rate": 0.0001873433037278985,
      "loss": 3.9359,
      "step": 67890
    },
    {
      "epoch": 0.06329378675703354,
      "grad_norm": 1.1009854078292847,
      "learning_rate": 0.00018734143940580327,
      "loss": 3.777,
      "step": 67900
    },
    {
      "epoch": 0.06330310837511263,
      "grad_norm": 1.2282772064208984,
      "learning_rate": 0.00018733957508370807,
      "loss": 3.7706,
      "step": 67910
    },
    {
      "epoch": 0.06331242999319173,
      "grad_norm": 0.9707192182540894,
      "learning_rate": 0.00018733771076161285,
      "loss": 3.7969,
      "step": 67920
    },
    {
      "epoch": 0.06332175161127082,
      "grad_norm": 1.121498703956604,
      "learning_rate": 0.00018733584643951766,
      "loss": 3.8668,
      "step": 67930
    },
    {
      "epoch": 0.0633310732293499,
      "grad_norm": 1.133091926574707,
      "learning_rate": 0.00018733398211742247,
      "loss": 3.9044,
      "step": 67940
    },
    {
      "epoch": 0.063340394847429,
      "grad_norm": 0.9863383769989014,
      "learning_rate": 0.00018733211779532727,
      "loss": 3.9421,
      "step": 67950
    },
    {
      "epoch": 0.06334971646550809,
      "grad_norm": 1.1195142269134521,
      "learning_rate": 0.00018733025347323208,
      "loss": 3.9336,
      "step": 67960
    },
    {
      "epoch": 0.06335903808358719,
      "grad_norm": 1.1541649103164673,
      "learning_rate": 0.00018732838915113686,
      "loss": 3.7128,
      "step": 67970
    },
    {
      "epoch": 0.06336835970166628,
      "grad_norm": 1.2686985731124878,
      "learning_rate": 0.00018732652482904166,
      "loss": 3.8762,
      "step": 67980
    },
    {
      "epoch": 0.06337768131974536,
      "grad_norm": 1.042608618736267,
      "learning_rate": 0.00018732466050694647,
      "loss": 3.8305,
      "step": 67990
    },
    {
      "epoch": 0.06338700293782445,
      "grad_norm": 1.0979747772216797,
      "learning_rate": 0.00018732279618485128,
      "loss": 3.9738,
      "step": 68000
    },
    {
      "epoch": 0.06339632455590355,
      "grad_norm": 1.1557835340499878,
      "learning_rate": 0.00018732093186275608,
      "loss": 3.9914,
      "step": 68010
    },
    {
      "epoch": 0.06340564617398264,
      "grad_norm": 1.251214861869812,
      "learning_rate": 0.00018731906754066086,
      "loss": 3.8789,
      "step": 68020
    },
    {
      "epoch": 0.06341496779206174,
      "grad_norm": 1.116012692451477,
      "learning_rate": 0.00018731720321856567,
      "loss": 4.0672,
      "step": 68030
    },
    {
      "epoch": 0.06342428941014082,
      "grad_norm": 0.9835241436958313,
      "learning_rate": 0.00018731533889647048,
      "loss": 3.9639,
      "step": 68040
    },
    {
      "epoch": 0.06343361102821991,
      "grad_norm": 1.1634106636047363,
      "learning_rate": 0.00018731347457437528,
      "loss": 3.8719,
      "step": 68050
    },
    {
      "epoch": 0.06344293264629901,
      "grad_norm": 1.040824294090271,
      "learning_rate": 0.0001873116102522801,
      "loss": 3.9267,
      "step": 68060
    },
    {
      "epoch": 0.0634522542643781,
      "grad_norm": 1.0972225666046143,
      "learning_rate": 0.00018730974593018487,
      "loss": 3.8155,
      "step": 68070
    },
    {
      "epoch": 0.0634615758824572,
      "grad_norm": 1.0298794507980347,
      "learning_rate": 0.00018730788160808967,
      "loss": 3.9892,
      "step": 68080
    },
    {
      "epoch": 0.06347089750053628,
      "grad_norm": 1.2180137634277344,
      "learning_rate": 0.00018730601728599448,
      "loss": 3.981,
      "step": 68090
    },
    {
      "epoch": 0.06348021911861537,
      "grad_norm": 1.0613566637039185,
      "learning_rate": 0.00018730415296389926,
      "loss": 4.0868,
      "step": 68100
    },
    {
      "epoch": 0.06348954073669447,
      "grad_norm": 1.0384711027145386,
      "learning_rate": 0.0001873022886418041,
      "loss": 3.8354,
      "step": 68110
    },
    {
      "epoch": 0.06349886235477356,
      "grad_norm": 1.156553864479065,
      "learning_rate": 0.00018730042431970887,
      "loss": 3.8237,
      "step": 68120
    },
    {
      "epoch": 0.06350818397285266,
      "grad_norm": 1.153511881828308,
      "learning_rate": 0.00018729855999761368,
      "loss": 3.791,
      "step": 68130
    },
    {
      "epoch": 0.06351750559093175,
      "grad_norm": 1.0299416780471802,
      "learning_rate": 0.00018729669567551848,
      "loss": 3.9562,
      "step": 68140
    },
    {
      "epoch": 0.06352682720901083,
      "grad_norm": 0.9765927791595459,
      "learning_rate": 0.00018729483135342326,
      "loss": 3.5951,
      "step": 68150
    },
    {
      "epoch": 0.06353614882708992,
      "grad_norm": 1.1472699642181396,
      "learning_rate": 0.00018729296703132807,
      "loss": 3.801,
      "step": 68160
    },
    {
      "epoch": 0.06354547044516902,
      "grad_norm": 1.1032836437225342,
      "learning_rate": 0.00018729110270923288,
      "loss": 3.8961,
      "step": 68170
    },
    {
      "epoch": 0.06355479206324811,
      "grad_norm": 1.0982533693313599,
      "learning_rate": 0.00018728923838713768,
      "loss": 3.7601,
      "step": 68180
    },
    {
      "epoch": 0.06356411368132721,
      "grad_norm": 1.03812575340271,
      "learning_rate": 0.0001872873740650425,
      "loss": 3.8435,
      "step": 68190
    },
    {
      "epoch": 0.06357343529940629,
      "grad_norm": 0.9885535836219788,
      "learning_rate": 0.00018728550974294727,
      "loss": 4.079,
      "step": 68200
    },
    {
      "epoch": 0.06358275691748538,
      "grad_norm": 1.1134799718856812,
      "learning_rate": 0.00018728364542085207,
      "loss": 3.7809,
      "step": 68210
    },
    {
      "epoch": 0.06359207853556448,
      "grad_norm": 0.8955547213554382,
      "learning_rate": 0.00018728178109875688,
      "loss": 3.9073,
      "step": 68220
    },
    {
      "epoch": 0.06360140015364357,
      "grad_norm": 1.006568431854248,
      "learning_rate": 0.0001872799167766617,
      "loss": 3.8978,
      "step": 68230
    },
    {
      "epoch": 0.06361072177172267,
      "grad_norm": 1.0594778060913086,
      "learning_rate": 0.0001872780524545665,
      "loss": 3.7444,
      "step": 68240
    },
    {
      "epoch": 0.06362004338980175,
      "grad_norm": 1.2313085794448853,
      "learning_rate": 0.00018727618813247127,
      "loss": 3.9476,
      "step": 68250
    },
    {
      "epoch": 0.06362936500788084,
      "grad_norm": 0.9843422174453735,
      "learning_rate": 0.00018727432381037608,
      "loss": 3.9845,
      "step": 68260
    },
    {
      "epoch": 0.06363868662595994,
      "grad_norm": 1.0489732027053833,
      "learning_rate": 0.00018727245948828088,
      "loss": 4.0173,
      "step": 68270
    },
    {
      "epoch": 0.06364800824403903,
      "grad_norm": 1.1117651462554932,
      "learning_rate": 0.0001872705951661857,
      "loss": 3.9061,
      "step": 68280
    },
    {
      "epoch": 0.06365732986211813,
      "grad_norm": 1.1475005149841309,
      "learning_rate": 0.0001872687308440905,
      "loss": 4.1531,
      "step": 68290
    },
    {
      "epoch": 0.0636666514801972,
      "grad_norm": 1.0445568561553955,
      "learning_rate": 0.00018726686652199528,
      "loss": 3.7697,
      "step": 68300
    },
    {
      "epoch": 0.0636759730982763,
      "grad_norm": 1.0717213153839111,
      "learning_rate": 0.00018726500219990008,
      "loss": 3.8634,
      "step": 68310
    },
    {
      "epoch": 0.0636852947163554,
      "grad_norm": 0.9950147867202759,
      "learning_rate": 0.00018726313787780486,
      "loss": 3.9015,
      "step": 68320
    },
    {
      "epoch": 0.06369461633443449,
      "grad_norm": 1.0589042901992798,
      "learning_rate": 0.0001872612735557097,
      "loss": 4.0307,
      "step": 68330
    },
    {
      "epoch": 0.06370393795251358,
      "grad_norm": 1.1424689292907715,
      "learning_rate": 0.0001872594092336145,
      "loss": 3.9875,
      "step": 68340
    },
    {
      "epoch": 0.06371325957059268,
      "grad_norm": 1.1051850318908691,
      "learning_rate": 0.00018725754491151928,
      "loss": 3.7485,
      "step": 68350
    },
    {
      "epoch": 0.06372258118867176,
      "grad_norm": 1.0707380771636963,
      "learning_rate": 0.0001872556805894241,
      "loss": 3.957,
      "step": 68360
    },
    {
      "epoch": 0.06373190280675085,
      "grad_norm": 1.107833743095398,
      "learning_rate": 0.00018725381626732887,
      "loss": 3.9027,
      "step": 68370
    },
    {
      "epoch": 0.06374122442482995,
      "grad_norm": 1.0089648962020874,
      "learning_rate": 0.00018725195194523367,
      "loss": 3.7183,
      "step": 68380
    },
    {
      "epoch": 0.06375054604290904,
      "grad_norm": 1.0075533390045166,
      "learning_rate": 0.00018725008762313848,
      "loss": 3.5661,
      "step": 68390
    },
    {
      "epoch": 0.06375986766098814,
      "grad_norm": 1.0693203210830688,
      "learning_rate": 0.00018724822330104329,
      "loss": 3.8244,
      "step": 68400
    },
    {
      "epoch": 0.06376918927906722,
      "grad_norm": 1.1220470666885376,
      "learning_rate": 0.0001872463589789481,
      "loss": 4.0498,
      "step": 68410
    },
    {
      "epoch": 0.06377851089714631,
      "grad_norm": 0.9362285137176514,
      "learning_rate": 0.00018724449465685287,
      "loss": 3.9241,
      "step": 68420
    },
    {
      "epoch": 0.0637878325152254,
      "grad_norm": 1.0787135362625122,
      "learning_rate": 0.00018724263033475768,
      "loss": 4.055,
      "step": 68430
    },
    {
      "epoch": 0.0637971541333045,
      "grad_norm": 0.957323431968689,
      "learning_rate": 0.00018724076601266248,
      "loss": 3.9075,
      "step": 68440
    },
    {
      "epoch": 0.0638064757513836,
      "grad_norm": 0.9574490189552307,
      "learning_rate": 0.0001872389016905673,
      "loss": 3.8809,
      "step": 68450
    },
    {
      "epoch": 0.06381579736946268,
      "grad_norm": 1.1020859479904175,
      "learning_rate": 0.0001872370373684721,
      "loss": 3.9473,
      "step": 68460
    },
    {
      "epoch": 0.06382511898754177,
      "grad_norm": 1.0827710628509521,
      "learning_rate": 0.00018723517304637688,
      "loss": 3.9047,
      "step": 68470
    },
    {
      "epoch": 0.06383444060562087,
      "grad_norm": 1.0721100568771362,
      "learning_rate": 0.00018723330872428168,
      "loss": 3.8601,
      "step": 68480
    },
    {
      "epoch": 0.06384376222369996,
      "grad_norm": 1.079822063446045,
      "learning_rate": 0.0001872314444021865,
      "loss": 3.8197,
      "step": 68490
    },
    {
      "epoch": 0.06385308384177905,
      "grad_norm": 1.087279200553894,
      "learning_rate": 0.0001872295800800913,
      "loss": 3.7774,
      "step": 68500
    },
    {
      "epoch": 0.06386240545985813,
      "grad_norm": 1.0759854316711426,
      "learning_rate": 0.0001872277157579961,
      "loss": 3.8192,
      "step": 68510
    },
    {
      "epoch": 0.06387172707793723,
      "grad_norm": 1.1836193799972534,
      "learning_rate": 0.00018722585143590088,
      "loss": 3.8691,
      "step": 68520
    },
    {
      "epoch": 0.06388104869601632,
      "grad_norm": 1.1191412210464478,
      "learning_rate": 0.00018722398711380569,
      "loss": 3.9399,
      "step": 68530
    },
    {
      "epoch": 0.06389037031409542,
      "grad_norm": 1.3291257619857788,
      "learning_rate": 0.0001872221227917105,
      "loss": 3.8719,
      "step": 68540
    },
    {
      "epoch": 0.06389969193217451,
      "grad_norm": 1.1536240577697754,
      "learning_rate": 0.00018722025846961527,
      "loss": 3.8935,
      "step": 68550
    },
    {
      "epoch": 0.06390901355025361,
      "grad_norm": 1.1286064386367798,
      "learning_rate": 0.0001872183941475201,
      "loss": 3.8213,
      "step": 68560
    },
    {
      "epoch": 0.06391833516833269,
      "grad_norm": 1.1620761156082153,
      "learning_rate": 0.00018721652982542488,
      "loss": 3.8341,
      "step": 68570
    },
    {
      "epoch": 0.06392765678641178,
      "grad_norm": 0.9555039405822754,
      "learning_rate": 0.0001872146655033297,
      "loss": 4.1175,
      "step": 68580
    },
    {
      "epoch": 0.06393697840449088,
      "grad_norm": 1.0065135955810547,
      "learning_rate": 0.0001872128011812345,
      "loss": 3.7413,
      "step": 68590
    },
    {
      "epoch": 0.06394630002256997,
      "grad_norm": 1.2035943269729614,
      "learning_rate": 0.00018721093685913928,
      "loss": 3.9688,
      "step": 68600
    },
    {
      "epoch": 0.06395562164064907,
      "grad_norm": 1.130946397781372,
      "learning_rate": 0.00018720907253704408,
      "loss": 3.8444,
      "step": 68610
    },
    {
      "epoch": 0.06396494325872815,
      "grad_norm": 0.89335697889328,
      "learning_rate": 0.0001872072082149489,
      "loss": 3.844,
      "step": 68620
    },
    {
      "epoch": 0.06397426487680724,
      "grad_norm": 1.1739921569824219,
      "learning_rate": 0.0001872053438928537,
      "loss": 3.8267,
      "step": 68630
    },
    {
      "epoch": 0.06398358649488634,
      "grad_norm": 1.0572878122329712,
      "learning_rate": 0.0001872034795707585,
      "loss": 3.8472,
      "step": 68640
    },
    {
      "epoch": 0.06399290811296543,
      "grad_norm": 1.0518288612365723,
      "learning_rate": 0.00018720161524866328,
      "loss": 4.1547,
      "step": 68650
    },
    {
      "epoch": 0.06400222973104452,
      "grad_norm": 1.0857597589492798,
      "learning_rate": 0.0001871997509265681,
      "loss": 3.8649,
      "step": 68660
    },
    {
      "epoch": 0.0640115513491236,
      "grad_norm": 0.9949666857719421,
      "learning_rate": 0.0001871978866044729,
      "loss": 4.0115,
      "step": 68670
    },
    {
      "epoch": 0.0640208729672027,
      "grad_norm": 1.001562237739563,
      "learning_rate": 0.0001871960222823777,
      "loss": 4.0121,
      "step": 68680
    },
    {
      "epoch": 0.0640301945852818,
      "grad_norm": 1.2235589027404785,
      "learning_rate": 0.0001871941579602825,
      "loss": 3.8152,
      "step": 68690
    },
    {
      "epoch": 0.06403951620336089,
      "grad_norm": 1.1883991956710815,
      "learning_rate": 0.00018719229363818728,
      "loss": 3.9022,
      "step": 68700
    },
    {
      "epoch": 0.06404883782143998,
      "grad_norm": 0.9846750497817993,
      "learning_rate": 0.0001871904293160921,
      "loss": 4.109,
      "step": 68710
    },
    {
      "epoch": 0.06405815943951906,
      "grad_norm": 1.1320477724075317,
      "learning_rate": 0.0001871885649939969,
      "loss": 3.8756,
      "step": 68720
    },
    {
      "epoch": 0.06406748105759816,
      "grad_norm": 1.1560884714126587,
      "learning_rate": 0.0001871867006719017,
      "loss": 3.8913,
      "step": 68730
    },
    {
      "epoch": 0.06407680267567725,
      "grad_norm": 1.1433327198028564,
      "learning_rate": 0.0001871848363498065,
      "loss": 3.7047,
      "step": 68740
    },
    {
      "epoch": 0.06408612429375635,
      "grad_norm": 1.1067930459976196,
      "learning_rate": 0.0001871829720277113,
      "loss": 4.0491,
      "step": 68750
    },
    {
      "epoch": 0.06409544591183544,
      "grad_norm": 1.058605670928955,
      "learning_rate": 0.0001871811077056161,
      "loss": 3.9751,
      "step": 68760
    },
    {
      "epoch": 0.06410476752991452,
      "grad_norm": 1.099094271659851,
      "learning_rate": 0.00018717924338352087,
      "loss": 4.0673,
      "step": 68770
    },
    {
      "epoch": 0.06411408914799362,
      "grad_norm": 1.0163694620132446,
      "learning_rate": 0.0001871773790614257,
      "loss": 3.76,
      "step": 68780
    },
    {
      "epoch": 0.06412341076607271,
      "grad_norm": 1.17780339717865,
      "learning_rate": 0.00018717551473933051,
      "loss": 3.8492,
      "step": 68790
    },
    {
      "epoch": 0.0641327323841518,
      "grad_norm": 1.1068075895309448,
      "learning_rate": 0.0001871736504172353,
      "loss": 3.746,
      "step": 68800
    },
    {
      "epoch": 0.0641420540022309,
      "grad_norm": 0.9602808952331543,
      "learning_rate": 0.0001871717860951401,
      "loss": 3.6723,
      "step": 68810
    },
    {
      "epoch": 0.06415137562031,
      "grad_norm": 0.984484851360321,
      "learning_rate": 0.00018716992177304488,
      "loss": 3.9925,
      "step": 68820
    },
    {
      "epoch": 0.06416069723838908,
      "grad_norm": 1.1821153163909912,
      "learning_rate": 0.00018716805745094969,
      "loss": 3.999,
      "step": 68830
    },
    {
      "epoch": 0.06417001885646817,
      "grad_norm": 1.150222659111023,
      "learning_rate": 0.00018716619312885452,
      "loss": 3.9668,
      "step": 68840
    },
    {
      "epoch": 0.06417934047454726,
      "grad_norm": 1.1956515312194824,
      "learning_rate": 0.0001871643288067593,
      "loss": 3.9788,
      "step": 68850
    },
    {
      "epoch": 0.06418866209262636,
      "grad_norm": 1.1178569793701172,
      "learning_rate": 0.0001871624644846641,
      "loss": 3.6344,
      "step": 68860
    },
    {
      "epoch": 0.06419798371070545,
      "grad_norm": 1.0489158630371094,
      "learning_rate": 0.00018716060016256888,
      "loss": 3.8286,
      "step": 68870
    },
    {
      "epoch": 0.06420730532878453,
      "grad_norm": 0.9595082998275757,
      "learning_rate": 0.0001871587358404737,
      "loss": 3.8687,
      "step": 68880
    },
    {
      "epoch": 0.06421662694686363,
      "grad_norm": 1.0643112659454346,
      "learning_rate": 0.0001871568715183785,
      "loss": 3.8243,
      "step": 68890
    },
    {
      "epoch": 0.06422594856494272,
      "grad_norm": 1.1779699325561523,
      "learning_rate": 0.0001871550071962833,
      "loss": 3.8104,
      "step": 68900
    },
    {
      "epoch": 0.06423527018302182,
      "grad_norm": 1.083402395248413,
      "learning_rate": 0.0001871531428741881,
      "loss": 3.7982,
      "step": 68910
    },
    {
      "epoch": 0.06424459180110091,
      "grad_norm": 0.9077815413475037,
      "learning_rate": 0.0001871512785520929,
      "loss": 3.8385,
      "step": 68920
    },
    {
      "epoch": 0.06425391341917999,
      "grad_norm": 1.0772298574447632,
      "learning_rate": 0.0001871494142299977,
      "loss": 3.7785,
      "step": 68930
    },
    {
      "epoch": 0.06426323503725909,
      "grad_norm": 1.099496841430664,
      "learning_rate": 0.0001871475499079025,
      "loss": 3.946,
      "step": 68940
    },
    {
      "epoch": 0.06427255665533818,
      "grad_norm": 1.1097297668457031,
      "learning_rate": 0.0001871456855858073,
      "loss": 4.0524,
      "step": 68950
    },
    {
      "epoch": 0.06428187827341728,
      "grad_norm": 1.0215481519699097,
      "learning_rate": 0.0001871438212637121,
      "loss": 3.7987,
      "step": 68960
    },
    {
      "epoch": 0.06429119989149637,
      "grad_norm": 1.2002804279327393,
      "learning_rate": 0.0001871419569416169,
      "loss": 3.7446,
      "step": 68970
    },
    {
      "epoch": 0.06430052150957545,
      "grad_norm": 1.2122728824615479,
      "learning_rate": 0.0001871400926195217,
      "loss": 3.7519,
      "step": 68980
    },
    {
      "epoch": 0.06430984312765455,
      "grad_norm": 1.043120265007019,
      "learning_rate": 0.00018713822829742648,
      "loss": 3.9396,
      "step": 68990
    },
    {
      "epoch": 0.06431916474573364,
      "grad_norm": 1.0340399742126465,
      "learning_rate": 0.00018713636397533128,
      "loss": 3.9295,
      "step": 69000
    },
    {
      "epoch": 0.06432848636381273,
      "grad_norm": 1.0534193515777588,
      "learning_rate": 0.00018713449965323612,
      "loss": 3.8161,
      "step": 69010
    },
    {
      "epoch": 0.06433780798189183,
      "grad_norm": 0.9907884001731873,
      "learning_rate": 0.0001871326353311409,
      "loss": 3.8613,
      "step": 69020
    },
    {
      "epoch": 0.06434712959997092,
      "grad_norm": 1.1394072771072388,
      "learning_rate": 0.0001871307710090457,
      "loss": 3.7268,
      "step": 69030
    },
    {
      "epoch": 0.06435645121805,
      "grad_norm": 1.2147250175476074,
      "learning_rate": 0.00018712890668695048,
      "loss": 3.8177,
      "step": 69040
    },
    {
      "epoch": 0.0643657728361291,
      "grad_norm": 1.3455044031143188,
      "learning_rate": 0.0001871270423648553,
      "loss": 3.7692,
      "step": 69050
    },
    {
      "epoch": 0.06437509445420819,
      "grad_norm": 1.0320672988891602,
      "learning_rate": 0.0001871251780427601,
      "loss": 3.7955,
      "step": 69060
    },
    {
      "epoch": 0.06438441607228729,
      "grad_norm": 1.1057367324829102,
      "learning_rate": 0.0001871233137206649,
      "loss": 3.8876,
      "step": 69070
    },
    {
      "epoch": 0.06439373769036638,
      "grad_norm": 0.9646188616752625,
      "learning_rate": 0.0001871214493985697,
      "loss": 3.8273,
      "step": 69080
    },
    {
      "epoch": 0.06440305930844546,
      "grad_norm": 1.3034390211105347,
      "learning_rate": 0.0001871195850764745,
      "loss": 3.8058,
      "step": 69090
    },
    {
      "epoch": 0.06441238092652456,
      "grad_norm": 0.8453490138053894,
      "learning_rate": 0.0001871177207543793,
      "loss": 3.9542,
      "step": 69100
    },
    {
      "epoch": 0.06442170254460365,
      "grad_norm": 0.8544497489929199,
      "learning_rate": 0.0001871158564322841,
      "loss": 4.0135,
      "step": 69110
    },
    {
      "epoch": 0.06443102416268275,
      "grad_norm": 0.9757434129714966,
      "learning_rate": 0.0001871139921101889,
      "loss": 3.8146,
      "step": 69120
    },
    {
      "epoch": 0.06444034578076184,
      "grad_norm": 0.9758973717689514,
      "learning_rate": 0.0001871121277880937,
      "loss": 3.8586,
      "step": 69130
    },
    {
      "epoch": 0.06444966739884092,
      "grad_norm": 1.2126023769378662,
      "learning_rate": 0.0001871102634659985,
      "loss": 3.999,
      "step": 69140
    },
    {
      "epoch": 0.06445898901692002,
      "grad_norm": 1.1527820825576782,
      "learning_rate": 0.0001871083991439033,
      "loss": 3.9449,
      "step": 69150
    },
    {
      "epoch": 0.06446831063499911,
      "grad_norm": 1.1616895198822021,
      "learning_rate": 0.0001871065348218081,
      "loss": 3.8239,
      "step": 69160
    },
    {
      "epoch": 0.0644776322530782,
      "grad_norm": 1.0893667936325073,
      "learning_rate": 0.0001871046704997129,
      "loss": 3.9383,
      "step": 69170
    },
    {
      "epoch": 0.0644869538711573,
      "grad_norm": 1.074220061302185,
      "learning_rate": 0.00018710280617761772,
      "loss": 3.831,
      "step": 69180
    },
    {
      "epoch": 0.06449627548923638,
      "grad_norm": 1.3098649978637695,
      "learning_rate": 0.0001871009418555225,
      "loss": 3.811,
      "step": 69190
    },
    {
      "epoch": 0.06450559710731547,
      "grad_norm": 1.0947511196136475,
      "learning_rate": 0.0001870990775334273,
      "loss": 3.8105,
      "step": 69200
    },
    {
      "epoch": 0.06451491872539457,
      "grad_norm": 1.0831280946731567,
      "learning_rate": 0.0001870972132113321,
      "loss": 4.1552,
      "step": 69210
    },
    {
      "epoch": 0.06452424034347366,
      "grad_norm": 1.0030667781829834,
      "learning_rate": 0.0001870953488892369,
      "loss": 3.8155,
      "step": 69220
    },
    {
      "epoch": 0.06453356196155276,
      "grad_norm": 1.1368155479431152,
      "learning_rate": 0.00018709348456714172,
      "loss": 3.8037,
      "step": 69230
    },
    {
      "epoch": 0.06454288357963185,
      "grad_norm": 0.8124812841415405,
      "learning_rate": 0.0001870916202450465,
      "loss": 3.7784,
      "step": 69240
    },
    {
      "epoch": 0.06455220519771093,
      "grad_norm": 1.0901297330856323,
      "learning_rate": 0.0001870897559229513,
      "loss": 4.0256,
      "step": 69250
    },
    {
      "epoch": 0.06456152681579003,
      "grad_norm": 1.1565682888031006,
      "learning_rate": 0.0001870878916008561,
      "loss": 3.8089,
      "step": 69260
    },
    {
      "epoch": 0.06457084843386912,
      "grad_norm": 1.0910532474517822,
      "learning_rate": 0.0001870860272787609,
      "loss": 3.8242,
      "step": 69270
    },
    {
      "epoch": 0.06458017005194822,
      "grad_norm": 1.197399377822876,
      "learning_rate": 0.0001870841629566657,
      "loss": 3.7516,
      "step": 69280
    },
    {
      "epoch": 0.06458949167002731,
      "grad_norm": 1.1152960062026978,
      "learning_rate": 0.0001870822986345705,
      "loss": 3.6618,
      "step": 69290
    },
    {
      "epoch": 0.06459881328810639,
      "grad_norm": 1.002990961074829,
      "learning_rate": 0.0001870804343124753,
      "loss": 3.8785,
      "step": 69300
    },
    {
      "epoch": 0.06460813490618549,
      "grad_norm": 1.0580461025238037,
      "learning_rate": 0.00018707856999038012,
      "loss": 3.904,
      "step": 69310
    },
    {
      "epoch": 0.06461745652426458,
      "grad_norm": 1.1600862741470337,
      "learning_rate": 0.0001870767056682849,
      "loss": 4.0202,
      "step": 69320
    },
    {
      "epoch": 0.06462677814234367,
      "grad_norm": 1.0150666236877441,
      "learning_rate": 0.0001870748413461897,
      "loss": 3.8229,
      "step": 69330
    },
    {
      "epoch": 0.06463609976042277,
      "grad_norm": 1.0985088348388672,
      "learning_rate": 0.0001870729770240945,
      "loss": 3.8113,
      "step": 69340
    },
    {
      "epoch": 0.06464542137850185,
      "grad_norm": 1.1152496337890625,
      "learning_rate": 0.00018707111270199932,
      "loss": 3.9931,
      "step": 69350
    },
    {
      "epoch": 0.06465474299658094,
      "grad_norm": 1.0008301734924316,
      "learning_rate": 0.00018706924837990412,
      "loss": 3.803,
      "step": 69360
    },
    {
      "epoch": 0.06466406461466004,
      "grad_norm": 1.188976526260376,
      "learning_rate": 0.0001870673840578089,
      "loss": 3.9583,
      "step": 69370
    },
    {
      "epoch": 0.06467338623273913,
      "grad_norm": 1.0282574892044067,
      "learning_rate": 0.0001870655197357137,
      "loss": 3.8785,
      "step": 69380
    },
    {
      "epoch": 0.06468270785081823,
      "grad_norm": 1.09946870803833,
      "learning_rate": 0.0001870636554136185,
      "loss": 3.7486,
      "step": 69390
    },
    {
      "epoch": 0.06469202946889731,
      "grad_norm": 1.0135091543197632,
      "learning_rate": 0.00018706179109152332,
      "loss": 3.8155,
      "step": 69400
    },
    {
      "epoch": 0.0647013510869764,
      "grad_norm": 0.8508045077323914,
      "learning_rate": 0.00018705992676942813,
      "loss": 3.9627,
      "step": 69410
    },
    {
      "epoch": 0.0647106727050555,
      "grad_norm": 1.0267682075500488,
      "learning_rate": 0.0001870580624473329,
      "loss": 4.0326,
      "step": 69420
    },
    {
      "epoch": 0.06471999432313459,
      "grad_norm": 1.0971665382385254,
      "learning_rate": 0.0001870561981252377,
      "loss": 3.8712,
      "step": 69430
    },
    {
      "epoch": 0.06472931594121369,
      "grad_norm": 1.3636173009872437,
      "learning_rate": 0.0001870543338031425,
      "loss": 3.9097,
      "step": 69440
    },
    {
      "epoch": 0.06473863755929278,
      "grad_norm": 1.053385853767395,
      "learning_rate": 0.0001870524694810473,
      "loss": 3.9259,
      "step": 69450
    },
    {
      "epoch": 0.06474795917737186,
      "grad_norm": 0.9917256832122803,
      "learning_rate": 0.00018705060515895213,
      "loss": 4.1145,
      "step": 69460
    },
    {
      "epoch": 0.06475728079545096,
      "grad_norm": 1.181381106376648,
      "learning_rate": 0.0001870487408368569,
      "loss": 3.7903,
      "step": 69470
    },
    {
      "epoch": 0.06476660241353005,
      "grad_norm": 1.0031429529190063,
      "learning_rate": 0.00018704687651476172,
      "loss": 3.8122,
      "step": 69480
    },
    {
      "epoch": 0.06477592403160914,
      "grad_norm": 1.1092466115951538,
      "learning_rate": 0.0001870450121926665,
      "loss": 3.8259,
      "step": 69490
    },
    {
      "epoch": 0.06478524564968824,
      "grad_norm": 0.9957287311553955,
      "learning_rate": 0.0001870431478705713,
      "loss": 3.951,
      "step": 69500
    },
    {
      "epoch": 0.06479456726776732,
      "grad_norm": 0.9774878025054932,
      "learning_rate": 0.0001870412835484761,
      "loss": 3.7491,
      "step": 69510
    },
    {
      "epoch": 0.06480388888584641,
      "grad_norm": 1.1593058109283447,
      "learning_rate": 0.00018703941922638091,
      "loss": 3.7372,
      "step": 69520
    },
    {
      "epoch": 0.06481321050392551,
      "grad_norm": 1.1247128248214722,
      "learning_rate": 0.00018703755490428572,
      "loss": 3.9126,
      "step": 69530
    },
    {
      "epoch": 0.0648225321220046,
      "grad_norm": 1.061985969543457,
      "learning_rate": 0.0001870356905821905,
      "loss": 3.9795,
      "step": 69540
    },
    {
      "epoch": 0.0648318537400837,
      "grad_norm": 0.9438024163246155,
      "learning_rate": 0.0001870338262600953,
      "loss": 3.8468,
      "step": 69550
    },
    {
      "epoch": 0.06484117535816278,
      "grad_norm": 1.0942230224609375,
      "learning_rate": 0.0001870319619380001,
      "loss": 4.0068,
      "step": 69560
    },
    {
      "epoch": 0.06485049697624187,
      "grad_norm": 0.9792689085006714,
      "learning_rate": 0.00018703009761590492,
      "loss": 3.9626,
      "step": 69570
    },
    {
      "epoch": 0.06485981859432097,
      "grad_norm": 0.9777472615242004,
      "learning_rate": 0.00018702823329380973,
      "loss": 3.9131,
      "step": 69580
    },
    {
      "epoch": 0.06486914021240006,
      "grad_norm": 1.0851194858551025,
      "learning_rate": 0.0001870263689717145,
      "loss": 3.9226,
      "step": 69590
    },
    {
      "epoch": 0.06487846183047916,
      "grad_norm": 1.082709789276123,
      "learning_rate": 0.0001870245046496193,
      "loss": 3.8314,
      "step": 69600
    },
    {
      "epoch": 0.06488778344855824,
      "grad_norm": 1.1286600828170776,
      "learning_rate": 0.00018702264032752412,
      "loss": 3.8512,
      "step": 69610
    },
    {
      "epoch": 0.06489710506663733,
      "grad_norm": 0.9062092900276184,
      "learning_rate": 0.00018702077600542892,
      "loss": 4.0422,
      "step": 69620
    },
    {
      "epoch": 0.06490642668471643,
      "grad_norm": 0.9989998936653137,
      "learning_rate": 0.00018701891168333373,
      "loss": 3.8918,
      "step": 69630
    },
    {
      "epoch": 0.06491574830279552,
      "grad_norm": 1.10346519947052,
      "learning_rate": 0.0001870170473612385,
      "loss": 3.7179,
      "step": 69640
    },
    {
      "epoch": 0.06492506992087461,
      "grad_norm": 1.0403332710266113,
      "learning_rate": 0.00018701518303914332,
      "loss": 3.9778,
      "step": 69650
    },
    {
      "epoch": 0.0649343915389537,
      "grad_norm": 0.9320885539054871,
      "learning_rate": 0.00018701331871704812,
      "loss": 3.8533,
      "step": 69660
    },
    {
      "epoch": 0.06494371315703279,
      "grad_norm": 1.005853533744812,
      "learning_rate": 0.0001870114543949529,
      "loss": 3.9693,
      "step": 69670
    },
    {
      "epoch": 0.06495303477511188,
      "grad_norm": 0.9099259376525879,
      "learning_rate": 0.00018700959007285773,
      "loss": 3.9393,
      "step": 69680
    },
    {
      "epoch": 0.06496235639319098,
      "grad_norm": 0.9580855965614319,
      "learning_rate": 0.0001870077257507625,
      "loss": 3.9421,
      "step": 69690
    },
    {
      "epoch": 0.06497167801127007,
      "grad_norm": 1.0119988918304443,
      "learning_rate": 0.00018700586142866732,
      "loss": 3.8975,
      "step": 69700
    },
    {
      "epoch": 0.06498099962934917,
      "grad_norm": 0.9737181663513184,
      "learning_rate": 0.00018700399710657213,
      "loss": 3.9813,
      "step": 69710
    },
    {
      "epoch": 0.06499032124742825,
      "grad_norm": 1.1910769939422607,
      "learning_rate": 0.0001870021327844769,
      "loss": 3.8264,
      "step": 69720
    },
    {
      "epoch": 0.06499964286550734,
      "grad_norm": 1.034146785736084,
      "learning_rate": 0.0001870002684623817,
      "loss": 3.9917,
      "step": 69730
    },
    {
      "epoch": 0.06500896448358644,
      "grad_norm": 1.0499281883239746,
      "learning_rate": 0.00018699840414028652,
      "loss": 4.0047,
      "step": 69740
    },
    {
      "epoch": 0.06501828610166553,
      "grad_norm": 0.961152970790863,
      "learning_rate": 0.00018699653981819132,
      "loss": 3.7743,
      "step": 69750
    },
    {
      "epoch": 0.06502760771974463,
      "grad_norm": 1.1737147569656372,
      "learning_rate": 0.00018699467549609613,
      "loss": 4.112,
      "step": 69760
    },
    {
      "epoch": 0.0650369293378237,
      "grad_norm": 1.0869160890579224,
      "learning_rate": 0.0001869928111740009,
      "loss": 3.744,
      "step": 69770
    },
    {
      "epoch": 0.0650462509559028,
      "grad_norm": 1.0648268461227417,
      "learning_rate": 0.00018699094685190572,
      "loss": 3.8722,
      "step": 69780
    },
    {
      "epoch": 0.0650555725739819,
      "grad_norm": 1.0458786487579346,
      "learning_rate": 0.00018698908252981052,
      "loss": 3.977,
      "step": 69790
    },
    {
      "epoch": 0.06506489419206099,
      "grad_norm": 1.0881026983261108,
      "learning_rate": 0.00018698721820771533,
      "loss": 3.9424,
      "step": 69800
    },
    {
      "epoch": 0.06507421581014008,
      "grad_norm": 0.9239693284034729,
      "learning_rate": 0.00018698535388562013,
      "loss": 3.816,
      "step": 69810
    },
    {
      "epoch": 0.06508353742821917,
      "grad_norm": 1.2371785640716553,
      "learning_rate": 0.00018698348956352491,
      "loss": 3.9608,
      "step": 69820
    },
    {
      "epoch": 0.06509285904629826,
      "grad_norm": 1.1048511266708374,
      "learning_rate": 0.00018698162524142972,
      "loss": 4.004,
      "step": 69830
    },
    {
      "epoch": 0.06510218066437735,
      "grad_norm": 0.9804288148880005,
      "learning_rate": 0.0001869797609193345,
      "loss": 4.0478,
      "step": 69840
    },
    {
      "epoch": 0.06511150228245645,
      "grad_norm": 1.023223876953125,
      "learning_rate": 0.00018697789659723933,
      "loss": 3.7377,
      "step": 69850
    },
    {
      "epoch": 0.06512082390053554,
      "grad_norm": 1.3043352365493774,
      "learning_rate": 0.00018697603227514414,
      "loss": 3.8678,
      "step": 69860
    },
    {
      "epoch": 0.06513014551861462,
      "grad_norm": 1.0368934869766235,
      "learning_rate": 0.00018697416795304892,
      "loss": 4.0805,
      "step": 69870
    },
    {
      "epoch": 0.06513946713669372,
      "grad_norm": 0.9784476161003113,
      "learning_rate": 0.00018697230363095372,
      "loss": 3.8026,
      "step": 69880
    },
    {
      "epoch": 0.06514878875477281,
      "grad_norm": 1.112503170967102,
      "learning_rate": 0.0001869704393088585,
      "loss": 3.8222,
      "step": 69890
    },
    {
      "epoch": 0.06515811037285191,
      "grad_norm": 0.9579881429672241,
      "learning_rate": 0.0001869685749867633,
      "loss": 3.7946,
      "step": 69900
    },
    {
      "epoch": 0.065167431990931,
      "grad_norm": 1.1819987297058105,
      "learning_rate": 0.00018696671066466814,
      "loss": 3.9781,
      "step": 69910
    },
    {
      "epoch": 0.0651767536090101,
      "grad_norm": 1.1794648170471191,
      "learning_rate": 0.00018696484634257292,
      "loss": 3.8388,
      "step": 69920
    },
    {
      "epoch": 0.06518607522708918,
      "grad_norm": 1.276341199874878,
      "learning_rate": 0.00018696298202047773,
      "loss": 3.9646,
      "step": 69930
    },
    {
      "epoch": 0.06519539684516827,
      "grad_norm": 1.1105395555496216,
      "learning_rate": 0.0001869611176983825,
      "loss": 3.7975,
      "step": 69940
    },
    {
      "epoch": 0.06520471846324737,
      "grad_norm": 0.9915034770965576,
      "learning_rate": 0.00018695925337628731,
      "loss": 4.0355,
      "step": 69950
    },
    {
      "epoch": 0.06521404008132646,
      "grad_norm": 1.1077935695648193,
      "learning_rate": 0.00018695738905419212,
      "loss": 3.8532,
      "step": 69960
    },
    {
      "epoch": 0.06522336169940555,
      "grad_norm": 1.1700782775878906,
      "learning_rate": 0.00018695552473209693,
      "loss": 3.7329,
      "step": 69970
    },
    {
      "epoch": 0.06523268331748464,
      "grad_norm": 0.9013391733169556,
      "learning_rate": 0.00018695366041000173,
      "loss": 3.8089,
      "step": 69980
    },
    {
      "epoch": 0.06524200493556373,
      "grad_norm": 1.0605616569519043,
      "learning_rate": 0.0001869517960879065,
      "loss": 3.6961,
      "step": 69990
    },
    {
      "epoch": 0.06525132655364282,
      "grad_norm": 1.196773648262024,
      "learning_rate": 0.00018694993176581132,
      "loss": 4.0077,
      "step": 70000
    },
    {
      "epoch": 0.06526064817172192,
      "grad_norm": 1.1017147302627563,
      "learning_rate": 0.00018694806744371613,
      "loss": 4.0336,
      "step": 70010
    },
    {
      "epoch": 0.06526996978980101,
      "grad_norm": 1.0573261976242065,
      "learning_rate": 0.00018694620312162093,
      "loss": 3.9082,
      "step": 70020
    },
    {
      "epoch": 0.0652792914078801,
      "grad_norm": 1.1159271001815796,
      "learning_rate": 0.00018694433879952574,
      "loss": 3.9969,
      "step": 70030
    },
    {
      "epoch": 0.06528861302595919,
      "grad_norm": 1.0249508619308472,
      "learning_rate": 0.00018694247447743052,
      "loss": 3.848,
      "step": 70040
    },
    {
      "epoch": 0.06529793464403828,
      "grad_norm": 1.2437338829040527,
      "learning_rate": 0.00018694061015533532,
      "loss": 3.8415,
      "step": 70050
    },
    {
      "epoch": 0.06530725626211738,
      "grad_norm": 1.3989238739013672,
      "learning_rate": 0.00018693874583324013,
      "loss": 3.786,
      "step": 70060
    },
    {
      "epoch": 0.06531657788019647,
      "grad_norm": 1.0564897060394287,
      "learning_rate": 0.00018693688151114494,
      "loss": 3.9341,
      "step": 70070
    },
    {
      "epoch": 0.06532589949827555,
      "grad_norm": 1.251758337020874,
      "learning_rate": 0.00018693501718904974,
      "loss": 3.9829,
      "step": 70080
    },
    {
      "epoch": 0.06533522111635465,
      "grad_norm": 1.1385719776153564,
      "learning_rate": 0.00018693315286695452,
      "loss": 3.8926,
      "step": 70090
    },
    {
      "epoch": 0.06534454273443374,
      "grad_norm": 1.1144052743911743,
      "learning_rate": 0.00018693128854485933,
      "loss": 3.8956,
      "step": 70100
    },
    {
      "epoch": 0.06535386435251284,
      "grad_norm": 0.9967808723449707,
      "learning_rate": 0.00018692942422276413,
      "loss": 3.7062,
      "step": 70110
    },
    {
      "epoch": 0.06536318597059193,
      "grad_norm": 1.0388349294662476,
      "learning_rate": 0.0001869275599006689,
      "loss": 4.0803,
      "step": 70120
    },
    {
      "epoch": 0.06537250758867102,
      "grad_norm": 1.1276410818099976,
      "learning_rate": 0.00018692569557857375,
      "loss": 3.8249,
      "step": 70130
    },
    {
      "epoch": 0.0653818292067501,
      "grad_norm": 1.048714518547058,
      "learning_rate": 0.00018692383125647853,
      "loss": 3.8576,
      "step": 70140
    },
    {
      "epoch": 0.0653911508248292,
      "grad_norm": 1.2276111841201782,
      "learning_rate": 0.00018692196693438333,
      "loss": 3.8675,
      "step": 70150
    },
    {
      "epoch": 0.0654004724429083,
      "grad_norm": 1.200774073600769,
      "learning_rate": 0.00018692010261228814,
      "loss": 4.0259,
      "step": 70160
    },
    {
      "epoch": 0.06540979406098739,
      "grad_norm": 1.2834547758102417,
      "learning_rate": 0.00018691823829019292,
      "loss": 3.8558,
      "step": 70170
    },
    {
      "epoch": 0.06541911567906648,
      "grad_norm": 1.0107951164245605,
      "learning_rate": 0.00018691637396809772,
      "loss": 3.8723,
      "step": 70180
    },
    {
      "epoch": 0.06542843729714556,
      "grad_norm": 1.1291570663452148,
      "learning_rate": 0.00018691450964600253,
      "loss": 3.9613,
      "step": 70190
    },
    {
      "epoch": 0.06543775891522466,
      "grad_norm": 0.9363600611686707,
      "learning_rate": 0.00018691264532390734,
      "loss": 3.9704,
      "step": 70200
    },
    {
      "epoch": 0.06544708053330375,
      "grad_norm": 0.9583397507667542,
      "learning_rate": 0.00018691078100181214,
      "loss": 3.828,
      "step": 70210
    },
    {
      "epoch": 0.06545640215138285,
      "grad_norm": 1.083980917930603,
      "learning_rate": 0.00018690891667971692,
      "loss": 3.911,
      "step": 70220
    },
    {
      "epoch": 0.06546572376946194,
      "grad_norm": 0.9852479100227356,
      "learning_rate": 0.00018690705235762173,
      "loss": 3.9189,
      "step": 70230
    },
    {
      "epoch": 0.06547504538754102,
      "grad_norm": 1.3355638980865479,
      "learning_rate": 0.00018690518803552653,
      "loss": 3.7914,
      "step": 70240
    },
    {
      "epoch": 0.06548436700562012,
      "grad_norm": 0.9246764183044434,
      "learning_rate": 0.00018690332371343134,
      "loss": 3.908,
      "step": 70250
    },
    {
      "epoch": 0.06549368862369921,
      "grad_norm": 1.0833697319030762,
      "learning_rate": 0.00018690145939133615,
      "loss": 3.8346,
      "step": 70260
    },
    {
      "epoch": 0.0655030102417783,
      "grad_norm": 1.1330653429031372,
      "learning_rate": 0.00018689959506924093,
      "loss": 3.8986,
      "step": 70270
    },
    {
      "epoch": 0.0655123318598574,
      "grad_norm": 1.2276819944381714,
      "learning_rate": 0.00018689773074714573,
      "loss": 3.8791,
      "step": 70280
    },
    {
      "epoch": 0.06552165347793648,
      "grad_norm": 1.0123993158340454,
      "learning_rate": 0.0001868958664250505,
      "loss": 3.8239,
      "step": 70290
    },
    {
      "epoch": 0.06553097509601558,
      "grad_norm": 0.9897657036781311,
      "learning_rate": 0.00018689400210295535,
      "loss": 4.0907,
      "step": 70300
    },
    {
      "epoch": 0.06554029671409467,
      "grad_norm": 1.2024669647216797,
      "learning_rate": 0.00018689213778086015,
      "loss": 3.784,
      "step": 70310
    },
    {
      "epoch": 0.06554961833217376,
      "grad_norm": 0.9594538807868958,
      "learning_rate": 0.00018689027345876493,
      "loss": 3.9902,
      "step": 70320
    },
    {
      "epoch": 0.06555893995025286,
      "grad_norm": 1.0460036993026733,
      "learning_rate": 0.00018688840913666974,
      "loss": 3.9045,
      "step": 70330
    },
    {
      "epoch": 0.06556826156833194,
      "grad_norm": 1.1416049003601074,
      "learning_rate": 0.00018688654481457452,
      "loss": 3.8913,
      "step": 70340
    },
    {
      "epoch": 0.06557758318641103,
      "grad_norm": 1.097662091255188,
      "learning_rate": 0.00018688468049247932,
      "loss": 3.9243,
      "step": 70350
    },
    {
      "epoch": 0.06558690480449013,
      "grad_norm": 1.0134984254837036,
      "learning_rate": 0.00018688281617038413,
      "loss": 4.1025,
      "step": 70360
    },
    {
      "epoch": 0.06559622642256922,
      "grad_norm": 1.114395022392273,
      "learning_rate": 0.00018688095184828894,
      "loss": 3.8623,
      "step": 70370
    },
    {
      "epoch": 0.06560554804064832,
      "grad_norm": 0.8942681550979614,
      "learning_rate": 0.00018687908752619374,
      "loss": 3.867,
      "step": 70380
    },
    {
      "epoch": 0.06561486965872741,
      "grad_norm": 0.9625548124313354,
      "learning_rate": 0.00018687722320409852,
      "loss": 3.8467,
      "step": 70390
    },
    {
      "epoch": 0.06562419127680649,
      "grad_norm": 1.0413542985916138,
      "learning_rate": 0.00018687535888200333,
      "loss": 3.948,
      "step": 70400
    },
    {
      "epoch": 0.06563351289488559,
      "grad_norm": 1.0860873460769653,
      "learning_rate": 0.00018687349455990813,
      "loss": 3.8922,
      "step": 70410
    },
    {
      "epoch": 0.06564283451296468,
      "grad_norm": 1.2040213346481323,
      "learning_rate": 0.00018687163023781294,
      "loss": 4.1195,
      "step": 70420
    },
    {
      "epoch": 0.06565215613104378,
      "grad_norm": 1.2421690225601196,
      "learning_rate": 0.00018686976591571775,
      "loss": 3.6631,
      "step": 70430
    },
    {
      "epoch": 0.06566147774912287,
      "grad_norm": 0.9335147738456726,
      "learning_rate": 0.00018686790159362253,
      "loss": 3.7728,
      "step": 70440
    },
    {
      "epoch": 0.06567079936720195,
      "grad_norm": 1.1067143678665161,
      "learning_rate": 0.00018686603727152733,
      "loss": 4.0345,
      "step": 70450
    },
    {
      "epoch": 0.06568012098528105,
      "grad_norm": 1.1108075380325317,
      "learning_rate": 0.00018686417294943214,
      "loss": 3.9202,
      "step": 70460
    },
    {
      "epoch": 0.06568944260336014,
      "grad_norm": 1.0951673984527588,
      "learning_rate": 0.00018686230862733694,
      "loss": 3.9492,
      "step": 70470
    },
    {
      "epoch": 0.06569876422143923,
      "grad_norm": 0.931297779083252,
      "learning_rate": 0.00018686044430524175,
      "loss": 3.8856,
      "step": 70480
    },
    {
      "epoch": 0.06570808583951833,
      "grad_norm": 0.9344638586044312,
      "learning_rate": 0.00018685857998314653,
      "loss": 4.0229,
      "step": 70490
    },
    {
      "epoch": 0.06571740745759741,
      "grad_norm": 1.2489696741104126,
      "learning_rate": 0.00018685671566105134,
      "loss": 3.6628,
      "step": 70500
    },
    {
      "epoch": 0.0657267290756765,
      "grad_norm": 1.0533032417297363,
      "learning_rate": 0.00018685485133895612,
      "loss": 3.9977,
      "step": 70510
    },
    {
      "epoch": 0.0657360506937556,
      "grad_norm": 1.083276629447937,
      "learning_rate": 0.00018685298701686095,
      "loss": 3.6681,
      "step": 70520
    },
    {
      "epoch": 0.0657453723118347,
      "grad_norm": 1.1032366752624512,
      "learning_rate": 0.00018685112269476576,
      "loss": 3.8725,
      "step": 70530
    },
    {
      "epoch": 0.06575469392991379,
      "grad_norm": 1.008278727531433,
      "learning_rate": 0.00018684925837267053,
      "loss": 3.733,
      "step": 70540
    },
    {
      "epoch": 0.06576401554799287,
      "grad_norm": 1.042938470840454,
      "learning_rate": 0.00018684739405057534,
      "loss": 3.8551,
      "step": 70550
    },
    {
      "epoch": 0.06577333716607196,
      "grad_norm": 0.8886581659317017,
      "learning_rate": 0.00018684552972848012,
      "loss": 3.9892,
      "step": 70560
    },
    {
      "epoch": 0.06578265878415106,
      "grad_norm": 1.0565869808197021,
      "learning_rate": 0.00018684366540638493,
      "loss": 3.9423,
      "step": 70570
    },
    {
      "epoch": 0.06579198040223015,
      "grad_norm": 1.0030663013458252,
      "learning_rate": 0.00018684180108428976,
      "loss": 4.0551,
      "step": 70580
    },
    {
      "epoch": 0.06580130202030925,
      "grad_norm": 0.9178146719932556,
      "learning_rate": 0.00018683993676219454,
      "loss": 3.9721,
      "step": 70590
    },
    {
      "epoch": 0.06581062363838834,
      "grad_norm": 1.1354881525039673,
      "learning_rate": 0.00018683807244009935,
      "loss": 3.7126,
      "step": 70600
    },
    {
      "epoch": 0.06581994525646742,
      "grad_norm": 1.172850251197815,
      "learning_rate": 0.00018683620811800412,
      "loss": 4.077,
      "step": 70610
    },
    {
      "epoch": 0.06582926687454652,
      "grad_norm": 1.0460474491119385,
      "learning_rate": 0.00018683434379590893,
      "loss": 3.7858,
      "step": 70620
    },
    {
      "epoch": 0.06583858849262561,
      "grad_norm": 1.1259211301803589,
      "learning_rate": 0.00018683247947381374,
      "loss": 3.8687,
      "step": 70630
    },
    {
      "epoch": 0.0658479101107047,
      "grad_norm": 1.0698590278625488,
      "learning_rate": 0.00018683061515171854,
      "loss": 3.9041,
      "step": 70640
    },
    {
      "epoch": 0.0658572317287838,
      "grad_norm": 1.139723539352417,
      "learning_rate": 0.00018682875082962335,
      "loss": 3.896,
      "step": 70650
    },
    {
      "epoch": 0.06586655334686288,
      "grad_norm": 1.1080371141433716,
      "learning_rate": 0.00018682688650752813,
      "loss": 3.5476,
      "step": 70660
    },
    {
      "epoch": 0.06587587496494197,
      "grad_norm": 0.9614083170890808,
      "learning_rate": 0.00018682502218543294,
      "loss": 3.9948,
      "step": 70670
    },
    {
      "epoch": 0.06588519658302107,
      "grad_norm": 1.0070170164108276,
      "learning_rate": 0.00018682315786333774,
      "loss": 4.0886,
      "step": 70680
    },
    {
      "epoch": 0.06589451820110016,
      "grad_norm": 1.1248301267623901,
      "learning_rate": 0.00018682129354124255,
      "loss": 3.5988,
      "step": 70690
    },
    {
      "epoch": 0.06590383981917926,
      "grad_norm": 1.0478373765945435,
      "learning_rate": 0.00018681942921914735,
      "loss": 3.7943,
      "step": 70700
    },
    {
      "epoch": 0.06591316143725834,
      "grad_norm": 1.1286410093307495,
      "learning_rate": 0.00018681756489705213,
      "loss": 3.9963,
      "step": 70710
    },
    {
      "epoch": 0.06592248305533743,
      "grad_norm": 0.9681978225708008,
      "learning_rate": 0.00018681570057495694,
      "loss": 3.9318,
      "step": 70720
    },
    {
      "epoch": 0.06593180467341653,
      "grad_norm": 1.106299638748169,
      "learning_rate": 0.00018681383625286175,
      "loss": 3.795,
      "step": 70730
    },
    {
      "epoch": 0.06594112629149562,
      "grad_norm": 1.0385140180587769,
      "learning_rate": 0.00018681197193076653,
      "loss": 3.8542,
      "step": 70740
    },
    {
      "epoch": 0.06595044790957472,
      "grad_norm": 1.1206656694412231,
      "learning_rate": 0.00018681010760867136,
      "loss": 3.7801,
      "step": 70750
    },
    {
      "epoch": 0.0659597695276538,
      "grad_norm": 0.9574277997016907,
      "learning_rate": 0.00018680824328657614,
      "loss": 3.8217,
      "step": 70760
    },
    {
      "epoch": 0.06596909114573289,
      "grad_norm": 1.0289397239685059,
      "learning_rate": 0.00018680637896448094,
      "loss": 3.7138,
      "step": 70770
    },
    {
      "epoch": 0.06597841276381199,
      "grad_norm": 1.1641209125518799,
      "learning_rate": 0.00018680451464238575,
      "loss": 3.8875,
      "step": 70780
    },
    {
      "epoch": 0.06598773438189108,
      "grad_norm": 1.042286992073059,
      "learning_rate": 0.00018680265032029053,
      "loss": 3.8897,
      "step": 70790
    },
    {
      "epoch": 0.06599705599997018,
      "grad_norm": 1.039119839668274,
      "learning_rate": 0.00018680078599819534,
      "loss": 3.8372,
      "step": 70800
    },
    {
      "epoch": 0.06600637761804927,
      "grad_norm": 1.1556150913238525,
      "learning_rate": 0.00018679892167610014,
      "loss": 4.0624,
      "step": 70810
    },
    {
      "epoch": 0.06601569923612835,
      "grad_norm": 1.2054479122161865,
      "learning_rate": 0.00018679705735400495,
      "loss": 3.9534,
      "step": 70820
    },
    {
      "epoch": 0.06602502085420744,
      "grad_norm": 0.9274373650550842,
      "learning_rate": 0.00018679519303190975,
      "loss": 3.8676,
      "step": 70830
    },
    {
      "epoch": 0.06603434247228654,
      "grad_norm": 1.1249842643737793,
      "learning_rate": 0.00018679332870981453,
      "loss": 3.9579,
      "step": 70840
    },
    {
      "epoch": 0.06604366409036563,
      "grad_norm": 1.3961122035980225,
      "learning_rate": 0.00018679146438771934,
      "loss": 3.9687,
      "step": 70850
    },
    {
      "epoch": 0.06605298570844473,
      "grad_norm": 0.915107786655426,
      "learning_rate": 0.00018678960006562415,
      "loss": 4.0026,
      "step": 70860
    },
    {
      "epoch": 0.06606230732652381,
      "grad_norm": 1.0987128019332886,
      "learning_rate": 0.00018678773574352895,
      "loss": 3.735,
      "step": 70870
    },
    {
      "epoch": 0.0660716289446029,
      "grad_norm": 1.0620630979537964,
      "learning_rate": 0.00018678587142143376,
      "loss": 4.0263,
      "step": 70880
    },
    {
      "epoch": 0.066080950562682,
      "grad_norm": 1.2225159406661987,
      "learning_rate": 0.00018678400709933854,
      "loss": 3.9616,
      "step": 70890
    },
    {
      "epoch": 0.06609027218076109,
      "grad_norm": 1.061865210533142,
      "learning_rate": 0.00018678214277724334,
      "loss": 3.9339,
      "step": 70900
    },
    {
      "epoch": 0.06609959379884019,
      "grad_norm": 1.1227729320526123,
      "learning_rate": 0.00018678027845514815,
      "loss": 3.9595,
      "step": 70910
    },
    {
      "epoch": 0.06610891541691927,
      "grad_norm": 1.1365903615951538,
      "learning_rate": 0.00018677841413305296,
      "loss": 3.7976,
      "step": 70920
    },
    {
      "epoch": 0.06611823703499836,
      "grad_norm": 0.9882156848907471,
      "learning_rate": 0.00018677654981095776,
      "loss": 3.9444,
      "step": 70930
    },
    {
      "epoch": 0.06612755865307746,
      "grad_norm": 1.1519664525985718,
      "learning_rate": 0.00018677468548886254,
      "loss": 3.7871,
      "step": 70940
    },
    {
      "epoch": 0.06613688027115655,
      "grad_norm": 1.0602521896362305,
      "learning_rate": 0.00018677282116676735,
      "loss": 3.9559,
      "step": 70950
    },
    {
      "epoch": 0.06614620188923565,
      "grad_norm": 1.1098699569702148,
      "learning_rate": 0.00018677095684467213,
      "loss": 3.7582,
      "step": 70960
    },
    {
      "epoch": 0.06615552350731473,
      "grad_norm": 1.396712303161621,
      "learning_rate": 0.00018676909252257696,
      "loss": 3.8424,
      "step": 70970
    },
    {
      "epoch": 0.06616484512539382,
      "grad_norm": 1.0412096977233887,
      "learning_rate": 0.00018676722820048177,
      "loss": 3.9616,
      "step": 70980
    },
    {
      "epoch": 0.06617416674347291,
      "grad_norm": 1.1315916776657104,
      "learning_rate": 0.00018676536387838655,
      "loss": 3.9667,
      "step": 70990
    },
    {
      "epoch": 0.06618348836155201,
      "grad_norm": 1.1741083860397339,
      "learning_rate": 0.00018676349955629135,
      "loss": 3.762,
      "step": 71000
    },
    {
      "epoch": 0.0661928099796311,
      "grad_norm": 1.0378473997116089,
      "learning_rate": 0.00018676163523419613,
      "loss": 3.865,
      "step": 71010
    },
    {
      "epoch": 0.06620213159771018,
      "grad_norm": 0.9368388652801514,
      "learning_rate": 0.00018675977091210094,
      "loss": 3.9549,
      "step": 71020
    },
    {
      "epoch": 0.06621145321578928,
      "grad_norm": 1.0032310485839844,
      "learning_rate": 0.00018675790659000577,
      "loss": 3.7765,
      "step": 71030
    },
    {
      "epoch": 0.06622077483386837,
      "grad_norm": 0.9211151599884033,
      "learning_rate": 0.00018675604226791055,
      "loss": 3.9168,
      "step": 71040
    },
    {
      "epoch": 0.06623009645194747,
      "grad_norm": 1.1556730270385742,
      "learning_rate": 0.00018675417794581536,
      "loss": 3.9328,
      "step": 71050
    },
    {
      "epoch": 0.06623941807002656,
      "grad_norm": 1.0971434116363525,
      "learning_rate": 0.00018675231362372014,
      "loss": 3.9444,
      "step": 71060
    },
    {
      "epoch": 0.06624873968810566,
      "grad_norm": 0.9978781938552856,
      "learning_rate": 0.00018675044930162494,
      "loss": 3.9355,
      "step": 71070
    },
    {
      "epoch": 0.06625806130618474,
      "grad_norm": 0.8504122495651245,
      "learning_rate": 0.00018674858497952975,
      "loss": 3.7661,
      "step": 71080
    },
    {
      "epoch": 0.06626738292426383,
      "grad_norm": 1.086637020111084,
      "learning_rate": 0.00018674672065743456,
      "loss": 3.7727,
      "step": 71090
    },
    {
      "epoch": 0.06627670454234293,
      "grad_norm": 1.1699752807617188,
      "learning_rate": 0.00018674485633533936,
      "loss": 3.9994,
      "step": 71100
    },
    {
      "epoch": 0.06628602616042202,
      "grad_norm": 1.0640227794647217,
      "learning_rate": 0.00018674299201324414,
      "loss": 3.9002,
      "step": 71110
    },
    {
      "epoch": 0.06629534777850112,
      "grad_norm": 1.1118675470352173,
      "learning_rate": 0.00018674112769114895,
      "loss": 3.8911,
      "step": 71120
    },
    {
      "epoch": 0.0663046693965802,
      "grad_norm": 1.108633279800415,
      "learning_rate": 0.00018673926336905375,
      "loss": 3.8318,
      "step": 71130
    },
    {
      "epoch": 0.06631399101465929,
      "grad_norm": 0.9931886196136475,
      "learning_rate": 0.00018673739904695856,
      "loss": 3.9151,
      "step": 71140
    },
    {
      "epoch": 0.06632331263273838,
      "grad_norm": 1.0326197147369385,
      "learning_rate": 0.00018673553472486337,
      "loss": 3.917,
      "step": 71150
    },
    {
      "epoch": 0.06633263425081748,
      "grad_norm": 0.9688319563865662,
      "learning_rate": 0.00018673367040276815,
      "loss": 3.86,
      "step": 71160
    },
    {
      "epoch": 0.06634195586889657,
      "grad_norm": 1.164057970046997,
      "learning_rate": 0.00018673180608067295,
      "loss": 3.9786,
      "step": 71170
    },
    {
      "epoch": 0.06635127748697565,
      "grad_norm": 1.1297873258590698,
      "learning_rate": 0.00018672994175857776,
      "loss": 3.8974,
      "step": 71180
    },
    {
      "epoch": 0.06636059910505475,
      "grad_norm": 0.934040904045105,
      "learning_rate": 0.00018672807743648254,
      "loss": 4.0634,
      "step": 71190
    },
    {
      "epoch": 0.06636992072313384,
      "grad_norm": 1.1793538331985474,
      "learning_rate": 0.00018672621311438737,
      "loss": 4.049,
      "step": 71200
    },
    {
      "epoch": 0.06637924234121294,
      "grad_norm": 1.088136911392212,
      "learning_rate": 0.00018672434879229215,
      "loss": 3.603,
      "step": 71210
    },
    {
      "epoch": 0.06638856395929203,
      "grad_norm": 1.1404743194580078,
      "learning_rate": 0.00018672248447019696,
      "loss": 3.8477,
      "step": 71220
    },
    {
      "epoch": 0.06639788557737111,
      "grad_norm": 1.076169729232788,
      "learning_rate": 0.00018672062014810176,
      "loss": 3.8186,
      "step": 71230
    },
    {
      "epoch": 0.06640720719545021,
      "grad_norm": 1.0949355363845825,
      "learning_rate": 0.00018671875582600654,
      "loss": 4.069,
      "step": 71240
    },
    {
      "epoch": 0.0664165288135293,
      "grad_norm": 1.144454836845398,
      "learning_rate": 0.00018671689150391135,
      "loss": 3.8638,
      "step": 71250
    },
    {
      "epoch": 0.0664258504316084,
      "grad_norm": 1.0371490716934204,
      "learning_rate": 0.00018671502718181616,
      "loss": 3.9208,
      "step": 71260
    },
    {
      "epoch": 0.06643517204968749,
      "grad_norm": 1.1952261924743652,
      "learning_rate": 0.00018671316285972096,
      "loss": 3.996,
      "step": 71270
    },
    {
      "epoch": 0.06644449366776659,
      "grad_norm": 1.205042839050293,
      "learning_rate": 0.00018671129853762577,
      "loss": 3.9053,
      "step": 71280
    },
    {
      "epoch": 0.06645381528584567,
      "grad_norm": 0.8051870465278625,
      "learning_rate": 0.00018670943421553055,
      "loss": 3.9376,
      "step": 71290
    },
    {
      "epoch": 0.06646313690392476,
      "grad_norm": 1.0783261060714722,
      "learning_rate": 0.00018670756989343535,
      "loss": 4.0071,
      "step": 71300
    },
    {
      "epoch": 0.06647245852200385,
      "grad_norm": 1.1799681186676025,
      "learning_rate": 0.00018670570557134016,
      "loss": 4.0904,
      "step": 71310
    },
    {
      "epoch": 0.06648178014008295,
      "grad_norm": 0.9999983310699463,
      "learning_rate": 0.00018670384124924497,
      "loss": 4.0922,
      "step": 71320
    },
    {
      "epoch": 0.06649110175816204,
      "grad_norm": 0.9987871646881104,
      "learning_rate": 0.00018670197692714977,
      "loss": 3.9728,
      "step": 71330
    },
    {
      "epoch": 0.06650042337624112,
      "grad_norm": 1.1303565502166748,
      "learning_rate": 0.00018670011260505455,
      "loss": 3.6579,
      "step": 71340
    },
    {
      "epoch": 0.06650974499432022,
      "grad_norm": 0.8936017751693726,
      "learning_rate": 0.00018669824828295936,
      "loss": 3.8167,
      "step": 71350
    },
    {
      "epoch": 0.06651906661239931,
      "grad_norm": 1.0183449983596802,
      "learning_rate": 0.00018669638396086416,
      "loss": 3.9311,
      "step": 71360
    },
    {
      "epoch": 0.06652838823047841,
      "grad_norm": 1.1978143453598022,
      "learning_rate": 0.00018669451963876897,
      "loss": 3.9796,
      "step": 71370
    },
    {
      "epoch": 0.0665377098485575,
      "grad_norm": 1.067436695098877,
      "learning_rate": 0.00018669265531667378,
      "loss": 3.9181,
      "step": 71380
    },
    {
      "epoch": 0.06654703146663658,
      "grad_norm": 1.0749447345733643,
      "learning_rate": 0.00018669079099457856,
      "loss": 3.6174,
      "step": 71390
    },
    {
      "epoch": 0.06655635308471568,
      "grad_norm": 0.9931052923202515,
      "learning_rate": 0.00018668892667248336,
      "loss": 3.972,
      "step": 71400
    },
    {
      "epoch": 0.06656567470279477,
      "grad_norm": 0.9314408302307129,
      "learning_rate": 0.00018668706235038814,
      "loss": 3.7217,
      "step": 71410
    },
    {
      "epoch": 0.06657499632087387,
      "grad_norm": 1.2253432273864746,
      "learning_rate": 0.00018668519802829297,
      "loss": 3.8568,
      "step": 71420
    },
    {
      "epoch": 0.06658431793895296,
      "grad_norm": 1.0408909320831299,
      "learning_rate": 0.00018668333370619778,
      "loss": 3.9762,
      "step": 71430
    },
    {
      "epoch": 0.06659363955703204,
      "grad_norm": 0.8877263069152832,
      "learning_rate": 0.00018668146938410256,
      "loss": 3.9412,
      "step": 71440
    },
    {
      "epoch": 0.06660296117511114,
      "grad_norm": 1.1839451789855957,
      "learning_rate": 0.00018667960506200737,
      "loss": 3.9946,
      "step": 71450
    },
    {
      "epoch": 0.06661228279319023,
      "grad_norm": 1.11624276638031,
      "learning_rate": 0.00018667774073991215,
      "loss": 3.9248,
      "step": 71460
    },
    {
      "epoch": 0.06662160441126933,
      "grad_norm": 1.060788631439209,
      "learning_rate": 0.00018667587641781695,
      "loss": 3.9769,
      "step": 71470
    },
    {
      "epoch": 0.06663092602934842,
      "grad_norm": 1.0673877000808716,
      "learning_rate": 0.00018667401209572179,
      "loss": 3.8988,
      "step": 71480
    },
    {
      "epoch": 0.06664024764742751,
      "grad_norm": 1.200699806213379,
      "learning_rate": 0.00018667214777362656,
      "loss": 3.7686,
      "step": 71490
    },
    {
      "epoch": 0.0666495692655066,
      "grad_norm": 1.078755497932434,
      "learning_rate": 0.00018667028345153137,
      "loss": 3.8558,
      "step": 71500
    },
    {
      "epoch": 0.06665889088358569,
      "grad_norm": 1.0346366167068481,
      "learning_rate": 0.00018666841912943615,
      "loss": 3.8848,
      "step": 71510
    },
    {
      "epoch": 0.06666821250166478,
      "grad_norm": 1.022849202156067,
      "learning_rate": 0.00018666655480734096,
      "loss": 3.838,
      "step": 71520
    },
    {
      "epoch": 0.06667753411974388,
      "grad_norm": 1.0837148427963257,
      "learning_rate": 0.00018666469048524576,
      "loss": 3.8268,
      "step": 71530
    },
    {
      "epoch": 0.06668685573782297,
      "grad_norm": 1.2043375968933105,
      "learning_rate": 0.00018666282616315057,
      "loss": 3.8631,
      "step": 71540
    },
    {
      "epoch": 0.06669617735590205,
      "grad_norm": 1.0444073677062988,
      "learning_rate": 0.00018666096184105538,
      "loss": 3.9001,
      "step": 71550
    },
    {
      "epoch": 0.06670549897398115,
      "grad_norm": 1.184926152229309,
      "learning_rate": 0.00018665909751896015,
      "loss": 3.5144,
      "step": 71560
    },
    {
      "epoch": 0.06671482059206024,
      "grad_norm": 1.0179505348205566,
      "learning_rate": 0.00018665723319686496,
      "loss": 3.9904,
      "step": 71570
    },
    {
      "epoch": 0.06672414221013934,
      "grad_norm": 1.0877659320831299,
      "learning_rate": 0.00018665536887476977,
      "loss": 3.6973,
      "step": 71580
    },
    {
      "epoch": 0.06673346382821843,
      "grad_norm": 0.9301310181617737,
      "learning_rate": 0.00018665350455267457,
      "loss": 3.972,
      "step": 71590
    },
    {
      "epoch": 0.06674278544629751,
      "grad_norm": 1.0369999408721924,
      "learning_rate": 0.00018665164023057938,
      "loss": 4.0279,
      "step": 71600
    },
    {
      "epoch": 0.0667521070643766,
      "grad_norm": 0.9432046413421631,
      "learning_rate": 0.00018664977590848416,
      "loss": 3.7462,
      "step": 71610
    },
    {
      "epoch": 0.0667614286824557,
      "grad_norm": 1.0202670097351074,
      "learning_rate": 0.00018664791158638897,
      "loss": 3.7518,
      "step": 71620
    },
    {
      "epoch": 0.0667707503005348,
      "grad_norm": 0.9467697739601135,
      "learning_rate": 0.00018664604726429377,
      "loss": 3.9818,
      "step": 71630
    },
    {
      "epoch": 0.06678007191861389,
      "grad_norm": 1.173716425895691,
      "learning_rate": 0.00018664418294219858,
      "loss": 3.8918,
      "step": 71640
    },
    {
      "epoch": 0.06678939353669297,
      "grad_norm": 1.2115784883499146,
      "learning_rate": 0.00018664231862010338,
      "loss": 3.6829,
      "step": 71650
    },
    {
      "epoch": 0.06679871515477206,
      "grad_norm": 0.9886482357978821,
      "learning_rate": 0.00018664045429800816,
      "loss": 3.8135,
      "step": 71660
    },
    {
      "epoch": 0.06680803677285116,
      "grad_norm": 1.085669994354248,
      "learning_rate": 0.00018663858997591297,
      "loss": 3.6681,
      "step": 71670
    },
    {
      "epoch": 0.06681735839093025,
      "grad_norm": 1.0655168294906616,
      "learning_rate": 0.00018663672565381775,
      "loss": 3.8519,
      "step": 71680
    },
    {
      "epoch": 0.06682668000900935,
      "grad_norm": 1.064499855041504,
      "learning_rate": 0.00018663486133172256,
      "loss": 3.8208,
      "step": 71690
    },
    {
      "epoch": 0.06683600162708843,
      "grad_norm": 0.9490384459495544,
      "learning_rate": 0.00018663299700962736,
      "loss": 4.0503,
      "step": 71700
    },
    {
      "epoch": 0.06684532324516752,
      "grad_norm": 1.2040622234344482,
      "learning_rate": 0.00018663113268753217,
      "loss": 3.8338,
      "step": 71710
    },
    {
      "epoch": 0.06685464486324662,
      "grad_norm": 1.1133613586425781,
      "learning_rate": 0.00018662926836543697,
      "loss": 3.8714,
      "step": 71720
    },
    {
      "epoch": 0.06686396648132571,
      "grad_norm": 0.8960132598876953,
      "learning_rate": 0.00018662740404334175,
      "loss": 3.8529,
      "step": 71730
    },
    {
      "epoch": 0.0668732880994048,
      "grad_norm": 1.2873107194900513,
      "learning_rate": 0.00018662553972124656,
      "loss": 3.9583,
      "step": 71740
    },
    {
      "epoch": 0.0668826097174839,
      "grad_norm": 1.042564034461975,
      "learning_rate": 0.00018662367539915137,
      "loss": 3.8356,
      "step": 71750
    },
    {
      "epoch": 0.06689193133556298,
      "grad_norm": 1.1899549961090088,
      "learning_rate": 0.00018662181107705617,
      "loss": 3.6974,
      "step": 71760
    },
    {
      "epoch": 0.06690125295364208,
      "grad_norm": 1.29695463180542,
      "learning_rate": 0.00018661994675496098,
      "loss": 3.8081,
      "step": 71770
    },
    {
      "epoch": 0.06691057457172117,
      "grad_norm": 1.1609656810760498,
      "learning_rate": 0.00018661808243286576,
      "loss": 4.0049,
      "step": 71780
    },
    {
      "epoch": 0.06691989618980027,
      "grad_norm": 0.9854831099510193,
      "learning_rate": 0.00018661621811077056,
      "loss": 3.952,
      "step": 71790
    },
    {
      "epoch": 0.06692921780787936,
      "grad_norm": 0.9729030132293701,
      "learning_rate": 0.00018661435378867537,
      "loss": 3.8639,
      "step": 71800
    },
    {
      "epoch": 0.06693853942595844,
      "grad_norm": 0.8566511273384094,
      "learning_rate": 0.00018661248946658018,
      "loss": 3.9944,
      "step": 71810
    },
    {
      "epoch": 0.06694786104403753,
      "grad_norm": 0.8499800562858582,
      "learning_rate": 0.00018661062514448498,
      "loss": 3.8815,
      "step": 71820
    },
    {
      "epoch": 0.06695718266211663,
      "grad_norm": 0.955191969871521,
      "learning_rate": 0.00018660876082238976,
      "loss": 3.8689,
      "step": 71830
    },
    {
      "epoch": 0.06696650428019572,
      "grad_norm": 0.9235876798629761,
      "learning_rate": 0.00018660689650029457,
      "loss": 3.8752,
      "step": 71840
    },
    {
      "epoch": 0.06697582589827482,
      "grad_norm": 1.0195883512496948,
      "learning_rate": 0.00018660503217819937,
      "loss": 3.8404,
      "step": 71850
    },
    {
      "epoch": 0.0669851475163539,
      "grad_norm": 0.9479868412017822,
      "learning_rate": 0.00018660316785610415,
      "loss": 4.0341,
      "step": 71860
    },
    {
      "epoch": 0.066994469134433,
      "grad_norm": 1.1972436904907227,
      "learning_rate": 0.000186601303534009,
      "loss": 4.1087,
      "step": 71870
    },
    {
      "epoch": 0.06700379075251209,
      "grad_norm": 0.9454895257949829,
      "learning_rate": 0.00018659943921191377,
      "loss": 3.7821,
      "step": 71880
    },
    {
      "epoch": 0.06701311237059118,
      "grad_norm": 1.4790810346603394,
      "learning_rate": 0.00018659757488981857,
      "loss": 4.0608,
      "step": 71890
    },
    {
      "epoch": 0.06702243398867028,
      "grad_norm": 0.9740322232246399,
      "learning_rate": 0.00018659571056772338,
      "loss": 3.9278,
      "step": 71900
    },
    {
      "epoch": 0.06703175560674936,
      "grad_norm": 1.0660532712936401,
      "learning_rate": 0.00018659384624562816,
      "loss": 3.8866,
      "step": 71910
    },
    {
      "epoch": 0.06704107722482845,
      "grad_norm": 1.0768311023712158,
      "learning_rate": 0.00018659198192353296,
      "loss": 3.6345,
      "step": 71920
    },
    {
      "epoch": 0.06705039884290755,
      "grad_norm": 0.9058472514152527,
      "learning_rate": 0.00018659011760143777,
      "loss": 3.9491,
      "step": 71930
    },
    {
      "epoch": 0.06705972046098664,
      "grad_norm": 1.2280904054641724,
      "learning_rate": 0.00018658825327934258,
      "loss": 3.6724,
      "step": 71940
    },
    {
      "epoch": 0.06706904207906574,
      "grad_norm": 0.9364687204360962,
      "learning_rate": 0.00018658638895724738,
      "loss": 3.9189,
      "step": 71950
    },
    {
      "epoch": 0.06707836369714483,
      "grad_norm": 1.1829760074615479,
      "learning_rate": 0.00018658452463515216,
      "loss": 3.9365,
      "step": 71960
    },
    {
      "epoch": 0.06708768531522391,
      "grad_norm": 0.9676421880722046,
      "learning_rate": 0.00018658266031305697,
      "loss": 3.8097,
      "step": 71970
    },
    {
      "epoch": 0.067097006933303,
      "grad_norm": 1.2254294157028198,
      "learning_rate": 0.00018658079599096178,
      "loss": 3.9096,
      "step": 71980
    },
    {
      "epoch": 0.0671063285513821,
      "grad_norm": 1.081566572189331,
      "learning_rate": 0.00018657893166886658,
      "loss": 4.0838,
      "step": 71990
    },
    {
      "epoch": 0.0671156501694612,
      "grad_norm": 1.1406786441802979,
      "learning_rate": 0.0001865770673467714,
      "loss": 4.0243,
      "step": 72000
    },
    {
      "epoch": 0.06712497178754029,
      "grad_norm": 1.0466115474700928,
      "learning_rate": 0.00018657520302467617,
      "loss": 3.8593,
      "step": 72010
    },
    {
      "epoch": 0.06713429340561937,
      "grad_norm": 0.8558056950569153,
      "learning_rate": 0.00018657333870258097,
      "loss": 3.9124,
      "step": 72020
    },
    {
      "epoch": 0.06714361502369846,
      "grad_norm": 1.0045530796051025,
      "learning_rate": 0.00018657147438048578,
      "loss": 3.8902,
      "step": 72030
    },
    {
      "epoch": 0.06715293664177756,
      "grad_norm": 1.1294972896575928,
      "learning_rate": 0.00018656961005839059,
      "loss": 3.8457,
      "step": 72040
    },
    {
      "epoch": 0.06716225825985665,
      "grad_norm": 1.088576316833496,
      "learning_rate": 0.0001865677457362954,
      "loss": 3.8391,
      "step": 72050
    },
    {
      "epoch": 0.06717157987793575,
      "grad_norm": 1.037251353263855,
      "learning_rate": 0.00018656588141420017,
      "loss": 3.8892,
      "step": 72060
    },
    {
      "epoch": 0.06718090149601483,
      "grad_norm": 1.046730637550354,
      "learning_rate": 0.00018656401709210498,
      "loss": 3.8023,
      "step": 72070
    },
    {
      "epoch": 0.06719022311409392,
      "grad_norm": 1.0955554246902466,
      "learning_rate": 0.00018656215277000976,
      "loss": 3.8346,
      "step": 72080
    },
    {
      "epoch": 0.06719954473217302,
      "grad_norm": 1.2014631032943726,
      "learning_rate": 0.0001865602884479146,
      "loss": 3.8889,
      "step": 72090
    },
    {
      "epoch": 0.06720886635025211,
      "grad_norm": 0.8736647963523865,
      "learning_rate": 0.0001865584241258194,
      "loss": 3.8906,
      "step": 72100
    },
    {
      "epoch": 0.0672181879683312,
      "grad_norm": 0.9745441675186157,
      "learning_rate": 0.00018655655980372418,
      "loss": 3.6116,
      "step": 72110
    },
    {
      "epoch": 0.06722750958641029,
      "grad_norm": 0.8527711629867554,
      "learning_rate": 0.00018655469548162898,
      "loss": 3.7577,
      "step": 72120
    },
    {
      "epoch": 0.06723683120448938,
      "grad_norm": 1.1062699556350708,
      "learning_rate": 0.00018655283115953376,
      "loss": 3.8824,
      "step": 72130
    },
    {
      "epoch": 0.06724615282256848,
      "grad_norm": 1.1235265731811523,
      "learning_rate": 0.00018655096683743857,
      "loss": 3.8936,
      "step": 72140
    },
    {
      "epoch": 0.06725547444064757,
      "grad_norm": 1.2270197868347168,
      "learning_rate": 0.0001865491025153434,
      "loss": 3.8087,
      "step": 72150
    },
    {
      "epoch": 0.06726479605872666,
      "grad_norm": 0.8910236358642578,
      "learning_rate": 0.00018654723819324818,
      "loss": 3.9653,
      "step": 72160
    },
    {
      "epoch": 0.06727411767680576,
      "grad_norm": 1.0000371932983398,
      "learning_rate": 0.000186545373871153,
      "loss": 3.7674,
      "step": 72170
    },
    {
      "epoch": 0.06728343929488484,
      "grad_norm": 1.262237548828125,
      "learning_rate": 0.00018654350954905777,
      "loss": 3.6674,
      "step": 72180
    },
    {
      "epoch": 0.06729276091296393,
      "grad_norm": 1.3048099279403687,
      "learning_rate": 0.00018654164522696257,
      "loss": 3.8598,
      "step": 72190
    },
    {
      "epoch": 0.06730208253104303,
      "grad_norm": 0.9963580369949341,
      "learning_rate": 0.00018653978090486738,
      "loss": 3.9411,
      "step": 72200
    },
    {
      "epoch": 0.06731140414912212,
      "grad_norm": 1.0528075695037842,
      "learning_rate": 0.00018653791658277219,
      "loss": 3.7849,
      "step": 72210
    },
    {
      "epoch": 0.06732072576720122,
      "grad_norm": 1.2396897077560425,
      "learning_rate": 0.000186536052260677,
      "loss": 3.7393,
      "step": 72220
    },
    {
      "epoch": 0.0673300473852803,
      "grad_norm": 0.9504073858261108,
      "learning_rate": 0.00018653418793858177,
      "loss": 3.9857,
      "step": 72230
    },
    {
      "epoch": 0.06733936900335939,
      "grad_norm": 1.133608102798462,
      "learning_rate": 0.00018653232361648658,
      "loss": 3.7419,
      "step": 72240
    },
    {
      "epoch": 0.06734869062143849,
      "grad_norm": 1.1165634393692017,
      "learning_rate": 0.00018653045929439138,
      "loss": 3.9193,
      "step": 72250
    },
    {
      "epoch": 0.06735801223951758,
      "grad_norm": 0.9645500183105469,
      "learning_rate": 0.0001865285949722962,
      "loss": 3.7739,
      "step": 72260
    },
    {
      "epoch": 0.06736733385759668,
      "grad_norm": 1.1725257635116577,
      "learning_rate": 0.000186526730650201,
      "loss": 4.0037,
      "step": 72270
    },
    {
      "epoch": 0.06737665547567576,
      "grad_norm": 1.0037704706192017,
      "learning_rate": 0.00018652486632810578,
      "loss": 3.884,
      "step": 72280
    },
    {
      "epoch": 0.06738597709375485,
      "grad_norm": 1.112116813659668,
      "learning_rate": 0.00018652300200601058,
      "loss": 3.9213,
      "step": 72290
    },
    {
      "epoch": 0.06739529871183395,
      "grad_norm": 1.0136371850967407,
      "learning_rate": 0.0001865211376839154,
      "loss": 3.8506,
      "step": 72300
    },
    {
      "epoch": 0.06740462032991304,
      "grad_norm": 1.0056384801864624,
      "learning_rate": 0.00018651927336182017,
      "loss": 3.9475,
      "step": 72310
    },
    {
      "epoch": 0.06741394194799213,
      "grad_norm": 1.4391742944717407,
      "learning_rate": 0.000186517409039725,
      "loss": 3.7998,
      "step": 72320
    },
    {
      "epoch": 0.06742326356607121,
      "grad_norm": 1.159735918045044,
      "learning_rate": 0.00018651554471762978,
      "loss": 3.8978,
      "step": 72330
    },
    {
      "epoch": 0.06743258518415031,
      "grad_norm": 1.079750657081604,
      "learning_rate": 0.00018651368039553459,
      "loss": 3.6883,
      "step": 72340
    },
    {
      "epoch": 0.0674419068022294,
      "grad_norm": 1.2652621269226074,
      "learning_rate": 0.0001865118160734394,
      "loss": 3.8561,
      "step": 72350
    },
    {
      "epoch": 0.0674512284203085,
      "grad_norm": 1.0542376041412354,
      "learning_rate": 0.00018650995175134417,
      "loss": 3.8198,
      "step": 72360
    },
    {
      "epoch": 0.06746055003838759,
      "grad_norm": 0.9228190183639526,
      "learning_rate": 0.00018650808742924898,
      "loss": 3.8722,
      "step": 72370
    },
    {
      "epoch": 0.06746987165646667,
      "grad_norm": 1.1351863145828247,
      "learning_rate": 0.00018650622310715378,
      "loss": 3.701,
      "step": 72380
    },
    {
      "epoch": 0.06747919327454577,
      "grad_norm": 1.0948870182037354,
      "learning_rate": 0.0001865043587850586,
      "loss": 4.0886,
      "step": 72390
    },
    {
      "epoch": 0.06748851489262486,
      "grad_norm": 1.3570785522460938,
      "learning_rate": 0.0001865024944629634,
      "loss": 3.7282,
      "step": 72400
    },
    {
      "epoch": 0.06749783651070396,
      "grad_norm": 0.9631621837615967,
      "learning_rate": 0.00018650063014086818,
      "loss": 4.087,
      "step": 72410
    },
    {
      "epoch": 0.06750715812878305,
      "grad_norm": 0.86967933177948,
      "learning_rate": 0.00018649876581877298,
      "loss": 4.006,
      "step": 72420
    },
    {
      "epoch": 0.06751647974686215,
      "grad_norm": 0.9387534856796265,
      "learning_rate": 0.0001864969014966778,
      "loss": 3.9379,
      "step": 72430
    },
    {
      "epoch": 0.06752580136494123,
      "grad_norm": 1.1558693647384644,
      "learning_rate": 0.0001864950371745826,
      "loss": 3.4935,
      "step": 72440
    },
    {
      "epoch": 0.06753512298302032,
      "grad_norm": 1.02225923538208,
      "learning_rate": 0.0001864931728524874,
      "loss": 3.918,
      "step": 72450
    },
    {
      "epoch": 0.06754444460109942,
      "grad_norm": 0.8635506629943848,
      "learning_rate": 0.00018649130853039218,
      "loss": 3.8249,
      "step": 72460
    },
    {
      "epoch": 0.06755376621917851,
      "grad_norm": 1.1858171224594116,
      "learning_rate": 0.000186489444208297,
      "loss": 3.7751,
      "step": 72470
    },
    {
      "epoch": 0.0675630878372576,
      "grad_norm": 1.0486010313034058,
      "learning_rate": 0.0001864875798862018,
      "loss": 3.94,
      "step": 72480
    },
    {
      "epoch": 0.06757240945533668,
      "grad_norm": 1.2156709432601929,
      "learning_rate": 0.0001864857155641066,
      "loss": 3.6043,
      "step": 72490
    },
    {
      "epoch": 0.06758173107341578,
      "grad_norm": 0.9470617175102234,
      "learning_rate": 0.0001864838512420114,
      "loss": 3.791,
      "step": 72500
    },
    {
      "epoch": 0.06759105269149487,
      "grad_norm": 1.0758092403411865,
      "learning_rate": 0.00018648198691991618,
      "loss": 4.0104,
      "step": 72510
    },
    {
      "epoch": 0.06760037430957397,
      "grad_norm": 1.1629222631454468,
      "learning_rate": 0.000186480122597821,
      "loss": 3.7323,
      "step": 72520
    },
    {
      "epoch": 0.06760969592765306,
      "grad_norm": 1.0252392292022705,
      "learning_rate": 0.00018647825827572577,
      "loss": 3.7268,
      "step": 72530
    },
    {
      "epoch": 0.06761901754573214,
      "grad_norm": 0.959296703338623,
      "learning_rate": 0.0001864763939536306,
      "loss": 4.0035,
      "step": 72540
    },
    {
      "epoch": 0.06762833916381124,
      "grad_norm": 1.0206630229949951,
      "learning_rate": 0.0001864745296315354,
      "loss": 3.8168,
      "step": 72550
    },
    {
      "epoch": 0.06763766078189033,
      "grad_norm": 0.815439760684967,
      "learning_rate": 0.0001864726653094402,
      "loss": 3.9459,
      "step": 72560
    },
    {
      "epoch": 0.06764698239996943,
      "grad_norm": 1.1128532886505127,
      "learning_rate": 0.000186470800987345,
      "loss": 3.7442,
      "step": 72570
    },
    {
      "epoch": 0.06765630401804852,
      "grad_norm": 1.132148027420044,
      "learning_rate": 0.00018646893666524977,
      "loss": 3.9161,
      "step": 72580
    },
    {
      "epoch": 0.0676656256361276,
      "grad_norm": 1.1486613750457764,
      "learning_rate": 0.00018646707234315458,
      "loss": 3.7404,
      "step": 72590
    },
    {
      "epoch": 0.0676749472542067,
      "grad_norm": 1.021363615989685,
      "learning_rate": 0.00018646520802105941,
      "loss": 3.7751,
      "step": 72600
    },
    {
      "epoch": 0.06768426887228579,
      "grad_norm": 1.057597279548645,
      "learning_rate": 0.0001864633436989642,
      "loss": 3.7745,
      "step": 72610
    },
    {
      "epoch": 0.06769359049036489,
      "grad_norm": 1.0693401098251343,
      "learning_rate": 0.000186461479376869,
      "loss": 3.87,
      "step": 72620
    },
    {
      "epoch": 0.06770291210844398,
      "grad_norm": 0.9703339338302612,
      "learning_rate": 0.00018645961505477378,
      "loss": 3.7623,
      "step": 72630
    },
    {
      "epoch": 0.06771223372652307,
      "grad_norm": 1.2641681432724,
      "learning_rate": 0.00018645775073267859,
      "loss": 3.9079,
      "step": 72640
    },
    {
      "epoch": 0.06772155534460216,
      "grad_norm": 1.1315041780471802,
      "learning_rate": 0.0001864558864105834,
      "loss": 3.7484,
      "step": 72650
    },
    {
      "epoch": 0.06773087696268125,
      "grad_norm": 1.0871647596359253,
      "learning_rate": 0.0001864540220884882,
      "loss": 3.7837,
      "step": 72660
    },
    {
      "epoch": 0.06774019858076034,
      "grad_norm": 1.1418249607086182,
      "learning_rate": 0.000186452157766393,
      "loss": 3.8151,
      "step": 72670
    },
    {
      "epoch": 0.06774952019883944,
      "grad_norm": 1.1295783519744873,
      "learning_rate": 0.00018645029344429778,
      "loss": 3.8534,
      "step": 72680
    },
    {
      "epoch": 0.06775884181691853,
      "grad_norm": 1.0945194959640503,
      "learning_rate": 0.0001864484291222026,
      "loss": 3.9333,
      "step": 72690
    },
    {
      "epoch": 0.06776816343499761,
      "grad_norm": 1.0798691511154175,
      "learning_rate": 0.0001864465648001074,
      "loss": 3.9723,
      "step": 72700
    },
    {
      "epoch": 0.06777748505307671,
      "grad_norm": 1.0897928476333618,
      "learning_rate": 0.0001864447004780122,
      "loss": 4.026,
      "step": 72710
    },
    {
      "epoch": 0.0677868066711558,
      "grad_norm": 1.066590428352356,
      "learning_rate": 0.000186442836155917,
      "loss": 3.9932,
      "step": 72720
    },
    {
      "epoch": 0.0677961282892349,
      "grad_norm": 0.8698009848594666,
      "learning_rate": 0.0001864409718338218,
      "loss": 3.8898,
      "step": 72730
    },
    {
      "epoch": 0.06780544990731399,
      "grad_norm": 0.997307538986206,
      "learning_rate": 0.0001864391075117266,
      "loss": 3.9999,
      "step": 72740
    },
    {
      "epoch": 0.06781477152539307,
      "grad_norm": 0.8941516280174255,
      "learning_rate": 0.0001864372431896314,
      "loss": 4.0145,
      "step": 72750
    },
    {
      "epoch": 0.06782409314347217,
      "grad_norm": 0.8736762404441833,
      "learning_rate": 0.00018643537886753618,
      "loss": 4.0327,
      "step": 72760
    },
    {
      "epoch": 0.06783341476155126,
      "grad_norm": 0.9999622106552124,
      "learning_rate": 0.000186433514545441,
      "loss": 3.9356,
      "step": 72770
    },
    {
      "epoch": 0.06784273637963036,
      "grad_norm": 1.1417306661605835,
      "learning_rate": 0.0001864316502233458,
      "loss": 3.9209,
      "step": 72780
    },
    {
      "epoch": 0.06785205799770945,
      "grad_norm": 1.2618321180343628,
      "learning_rate": 0.0001864297859012506,
      "loss": 3.9954,
      "step": 72790
    },
    {
      "epoch": 0.06786137961578853,
      "grad_norm": 1.150726318359375,
      "learning_rate": 0.0001864279215791554,
      "loss": 3.8086,
      "step": 72800
    },
    {
      "epoch": 0.06787070123386763,
      "grad_norm": 1.1568447351455688,
      "learning_rate": 0.00018642605725706018,
      "loss": 3.8897,
      "step": 72810
    },
    {
      "epoch": 0.06788002285194672,
      "grad_norm": 1.0838338136672974,
      "learning_rate": 0.000186424192934965,
      "loss": 3.8026,
      "step": 72820
    },
    {
      "epoch": 0.06788934447002581,
      "grad_norm": 1.098533034324646,
      "learning_rate": 0.0001864223286128698,
      "loss": 3.7792,
      "step": 72830
    },
    {
      "epoch": 0.06789866608810491,
      "grad_norm": 1.0716285705566406,
      "learning_rate": 0.0001864204642907746,
      "loss": 3.8036,
      "step": 72840
    },
    {
      "epoch": 0.067907987706184,
      "grad_norm": 1.1835206747055054,
      "learning_rate": 0.0001864185999686794,
      "loss": 3.7665,
      "step": 72850
    },
    {
      "epoch": 0.06791730932426308,
      "grad_norm": 1.0407944917678833,
      "learning_rate": 0.0001864167356465842,
      "loss": 3.7795,
      "step": 72860
    },
    {
      "epoch": 0.06792663094234218,
      "grad_norm": 1.169865369796753,
      "learning_rate": 0.000186414871324489,
      "loss": 3.7397,
      "step": 72870
    },
    {
      "epoch": 0.06793595256042127,
      "grad_norm": 1.0678359270095825,
      "learning_rate": 0.0001864130070023938,
      "loss": 3.9873,
      "step": 72880
    },
    {
      "epoch": 0.06794527417850037,
      "grad_norm": 1.0290679931640625,
      "learning_rate": 0.0001864111426802986,
      "loss": 3.9569,
      "step": 72890
    },
    {
      "epoch": 0.06795459579657946,
      "grad_norm": 1.70783531665802,
      "learning_rate": 0.00018640927835820341,
      "loss": 3.8965,
      "step": 72900
    },
    {
      "epoch": 0.06796391741465854,
      "grad_norm": 0.912217915058136,
      "learning_rate": 0.0001864074140361082,
      "loss": 3.9681,
      "step": 72910
    },
    {
      "epoch": 0.06797323903273764,
      "grad_norm": 1.2986397743225098,
      "learning_rate": 0.000186405549714013,
      "loss": 4.0012,
      "step": 72920
    },
    {
      "epoch": 0.06798256065081673,
      "grad_norm": 1.0553903579711914,
      "learning_rate": 0.0001864036853919178,
      "loss": 3.6244,
      "step": 72930
    },
    {
      "epoch": 0.06799188226889583,
      "grad_norm": 1.135485053062439,
      "learning_rate": 0.0001864018210698226,
      "loss": 3.7857,
      "step": 72940
    },
    {
      "epoch": 0.06800120388697492,
      "grad_norm": 1.0498149394989014,
      "learning_rate": 0.00018639995674772742,
      "loss": 3.8055,
      "step": 72950
    },
    {
      "epoch": 0.068010525505054,
      "grad_norm": 1.0395127534866333,
      "learning_rate": 0.0001863980924256322,
      "loss": 3.7276,
      "step": 72960
    },
    {
      "epoch": 0.0680198471231331,
      "grad_norm": 1.0005159378051758,
      "learning_rate": 0.000186396228103537,
      "loss": 3.9501,
      "step": 72970
    },
    {
      "epoch": 0.06802916874121219,
      "grad_norm": 1.1148216724395752,
      "learning_rate": 0.00018639436378144178,
      "loss": 3.7746,
      "step": 72980
    },
    {
      "epoch": 0.06803849035929128,
      "grad_norm": 1.0579040050506592,
      "learning_rate": 0.00018639249945934662,
      "loss": 4.0037,
      "step": 72990
    },
    {
      "epoch": 0.06804781197737038,
      "grad_norm": 1.0007716417312622,
      "learning_rate": 0.00018639063513725142,
      "loss": 3.9103,
      "step": 73000
    },
    {
      "epoch": 0.06805713359544946,
      "grad_norm": 0.9746633768081665,
      "learning_rate": 0.0001863887708151562,
      "loss": 3.8595,
      "step": 73010
    },
    {
      "epoch": 0.06806645521352855,
      "grad_norm": 1.0956438779830933,
      "learning_rate": 0.000186386906493061,
      "loss": 3.8831,
      "step": 73020
    },
    {
      "epoch": 0.06807577683160765,
      "grad_norm": 1.1292411088943481,
      "learning_rate": 0.0001863850421709658,
      "loss": 3.772,
      "step": 73030
    },
    {
      "epoch": 0.06808509844968674,
      "grad_norm": 1.0611923933029175,
      "learning_rate": 0.0001863831778488706,
      "loss": 3.7274,
      "step": 73040
    },
    {
      "epoch": 0.06809442006776584,
      "grad_norm": 1.1114259958267212,
      "learning_rate": 0.0001863813135267754,
      "loss": 3.977,
      "step": 73050
    },
    {
      "epoch": 0.06810374168584492,
      "grad_norm": 0.9870224595069885,
      "learning_rate": 0.0001863794492046802,
      "loss": 3.8416,
      "step": 73060
    },
    {
      "epoch": 0.06811306330392401,
      "grad_norm": 1.0208934545516968,
      "learning_rate": 0.000186377584882585,
      "loss": 3.8724,
      "step": 73070
    },
    {
      "epoch": 0.0681223849220031,
      "grad_norm": 1.1576954126358032,
      "learning_rate": 0.0001863757205604898,
      "loss": 4.05,
      "step": 73080
    },
    {
      "epoch": 0.0681317065400822,
      "grad_norm": 1.0665419101715088,
      "learning_rate": 0.0001863738562383946,
      "loss": 3.7928,
      "step": 73090
    },
    {
      "epoch": 0.0681410281581613,
      "grad_norm": 1.2151381969451904,
      "learning_rate": 0.0001863719919162994,
      "loss": 4.0287,
      "step": 73100
    },
    {
      "epoch": 0.06815034977624039,
      "grad_norm": 0.9922189116477966,
      "learning_rate": 0.0001863701275942042,
      "loss": 3.7862,
      "step": 73110
    },
    {
      "epoch": 0.06815967139431947,
      "grad_norm": 1.0403506755828857,
      "learning_rate": 0.00018636826327210902,
      "loss": 3.839,
      "step": 73120
    },
    {
      "epoch": 0.06816899301239857,
      "grad_norm": 1.055557131767273,
      "learning_rate": 0.0001863663989500138,
      "loss": 3.7819,
      "step": 73130
    },
    {
      "epoch": 0.06817831463047766,
      "grad_norm": 1.0886446237564087,
      "learning_rate": 0.0001863645346279186,
      "loss": 3.9006,
      "step": 73140
    },
    {
      "epoch": 0.06818763624855675,
      "grad_norm": 1.019480586051941,
      "learning_rate": 0.00018636267030582338,
      "loss": 3.8909,
      "step": 73150
    },
    {
      "epoch": 0.06819695786663585,
      "grad_norm": 1.0957180261611938,
      "learning_rate": 0.00018636080598372822,
      "loss": 3.8208,
      "step": 73160
    },
    {
      "epoch": 0.06820627948471493,
      "grad_norm": 0.9466241002082825,
      "learning_rate": 0.00018635894166163302,
      "loss": 3.8762,
      "step": 73170
    },
    {
      "epoch": 0.06821560110279402,
      "grad_norm": 1.2470558881759644,
      "learning_rate": 0.0001863570773395378,
      "loss": 3.909,
      "step": 73180
    },
    {
      "epoch": 0.06822492272087312,
      "grad_norm": 0.9535539746284485,
      "learning_rate": 0.0001863552130174426,
      "loss": 3.8672,
      "step": 73190
    },
    {
      "epoch": 0.06823424433895221,
      "grad_norm": 1.131238579750061,
      "learning_rate": 0.00018635334869534739,
      "loss": 3.7921,
      "step": 73200
    },
    {
      "epoch": 0.06824356595703131,
      "grad_norm": 1.1617755889892578,
      "learning_rate": 0.0001863514843732522,
      "loss": 3.7578,
      "step": 73210
    },
    {
      "epoch": 0.06825288757511039,
      "grad_norm": 1.137918472290039,
      "learning_rate": 0.00018634962005115703,
      "loss": 3.8225,
      "step": 73220
    },
    {
      "epoch": 0.06826220919318948,
      "grad_norm": 1.0361281633377075,
      "learning_rate": 0.0001863477557290618,
      "loss": 3.8205,
      "step": 73230
    },
    {
      "epoch": 0.06827153081126858,
      "grad_norm": 1.0630744695663452,
      "learning_rate": 0.0001863458914069666,
      "loss": 3.7713,
      "step": 73240
    },
    {
      "epoch": 0.06828085242934767,
      "grad_norm": 1.0348421335220337,
      "learning_rate": 0.0001863440270848714,
      "loss": 3.9485,
      "step": 73250
    },
    {
      "epoch": 0.06829017404742677,
      "grad_norm": 1.0825775861740112,
      "learning_rate": 0.0001863421627627762,
      "loss": 3.8507,
      "step": 73260
    },
    {
      "epoch": 0.06829949566550585,
      "grad_norm": 1.0291122198104858,
      "learning_rate": 0.000186340298440681,
      "loss": 3.6742,
      "step": 73270
    },
    {
      "epoch": 0.06830881728358494,
      "grad_norm": 0.8315314054489136,
      "learning_rate": 0.0001863384341185858,
      "loss": 3.6492,
      "step": 73280
    },
    {
      "epoch": 0.06831813890166404,
      "grad_norm": 1.0357081890106201,
      "learning_rate": 0.00018633656979649062,
      "loss": 3.7626,
      "step": 73290
    },
    {
      "epoch": 0.06832746051974313,
      "grad_norm": 1.2285135984420776,
      "learning_rate": 0.0001863347054743954,
      "loss": 3.7392,
      "step": 73300
    },
    {
      "epoch": 0.06833678213782222,
      "grad_norm": 1.1552594900131226,
      "learning_rate": 0.0001863328411523002,
      "loss": 3.8475,
      "step": 73310
    },
    {
      "epoch": 0.06834610375590132,
      "grad_norm": 1.1810593605041504,
      "learning_rate": 0.000186330976830205,
      "loss": 3.9208,
      "step": 73320
    },
    {
      "epoch": 0.0683554253739804,
      "grad_norm": 1.3850992918014526,
      "learning_rate": 0.00018632911250810981,
      "loss": 3.8135,
      "step": 73330
    },
    {
      "epoch": 0.0683647469920595,
      "grad_norm": 1.0243325233459473,
      "learning_rate": 0.00018632724818601462,
      "loss": 3.8823,
      "step": 73340
    },
    {
      "epoch": 0.06837406861013859,
      "grad_norm": 1.1061567068099976,
      "learning_rate": 0.0001863253838639194,
      "loss": 3.8887,
      "step": 73350
    },
    {
      "epoch": 0.06838339022821768,
      "grad_norm": 1.1052950620651245,
      "learning_rate": 0.0001863235195418242,
      "loss": 3.8325,
      "step": 73360
    },
    {
      "epoch": 0.06839271184629678,
      "grad_norm": 1.0938154458999634,
      "learning_rate": 0.000186321655219729,
      "loss": 4.082,
      "step": 73370
    },
    {
      "epoch": 0.06840203346437586,
      "grad_norm": 0.9151210784912109,
      "learning_rate": 0.00018631979089763382,
      "loss": 3.6901,
      "step": 73380
    },
    {
      "epoch": 0.06841135508245495,
      "grad_norm": 0.949916422367096,
      "learning_rate": 0.00018631792657553862,
      "loss": 3.9687,
      "step": 73390
    },
    {
      "epoch": 0.06842067670053405,
      "grad_norm": 1.1307601928710938,
      "learning_rate": 0.0001863160622534434,
      "loss": 3.933,
      "step": 73400
    },
    {
      "epoch": 0.06842999831861314,
      "grad_norm": 1.0379000902175903,
      "learning_rate": 0.0001863141979313482,
      "loss": 3.7408,
      "step": 73410
    },
    {
      "epoch": 0.06843931993669224,
      "grad_norm": 1.0044190883636475,
      "learning_rate": 0.00018631233360925302,
      "loss": 3.6725,
      "step": 73420
    },
    {
      "epoch": 0.06844864155477132,
      "grad_norm": 0.9059355854988098,
      "learning_rate": 0.0001863104692871578,
      "loss": 4.123,
      "step": 73430
    },
    {
      "epoch": 0.06845796317285041,
      "grad_norm": 0.9294950366020203,
      "learning_rate": 0.00018630860496506263,
      "loss": 3.8125,
      "step": 73440
    },
    {
      "epoch": 0.0684672847909295,
      "grad_norm": 1.0682276487350464,
      "learning_rate": 0.0001863067406429674,
      "loss": 3.9337,
      "step": 73450
    },
    {
      "epoch": 0.0684766064090086,
      "grad_norm": 1.009141445159912,
      "learning_rate": 0.00018630487632087221,
      "loss": 3.9896,
      "step": 73460
    },
    {
      "epoch": 0.0684859280270877,
      "grad_norm": 1.1643915176391602,
      "learning_rate": 0.00018630301199877702,
      "loss": 3.7338,
      "step": 73470
    },
    {
      "epoch": 0.06849524964516678,
      "grad_norm": 1.0680067539215088,
      "learning_rate": 0.0001863011476766818,
      "loss": 3.7289,
      "step": 73480
    },
    {
      "epoch": 0.06850457126324587,
      "grad_norm": 1.1027007102966309,
      "learning_rate": 0.0001862992833545866,
      "loss": 3.8539,
      "step": 73490
    },
    {
      "epoch": 0.06851389288132496,
      "grad_norm": 1.0186947584152222,
      "learning_rate": 0.0001862974190324914,
      "loss": 3.7324,
      "step": 73500
    },
    {
      "epoch": 0.06852321449940406,
      "grad_norm": 0.9843321442604065,
      "learning_rate": 0.00018629555471039622,
      "loss": 3.8111,
      "step": 73510
    },
    {
      "epoch": 0.06853253611748315,
      "grad_norm": 1.1122561693191528,
      "learning_rate": 0.00018629369038830103,
      "loss": 3.8444,
      "step": 73520
    },
    {
      "epoch": 0.06854185773556225,
      "grad_norm": 1.1260313987731934,
      "learning_rate": 0.0001862918260662058,
      "loss": 3.7839,
      "step": 73530
    },
    {
      "epoch": 0.06855117935364133,
      "grad_norm": 0.8109379410743713,
      "learning_rate": 0.0001862899617441106,
      "loss": 3.7608,
      "step": 73540
    },
    {
      "epoch": 0.06856050097172042,
      "grad_norm": 1.0537283420562744,
      "learning_rate": 0.00018628809742201542,
      "loss": 3.8745,
      "step": 73550
    },
    {
      "epoch": 0.06856982258979952,
      "grad_norm": 1.1142916679382324,
      "learning_rate": 0.00018628623309992022,
      "loss": 3.7435,
      "step": 73560
    },
    {
      "epoch": 0.06857914420787861,
      "grad_norm": 1.163662075996399,
      "learning_rate": 0.00018628436877782503,
      "loss": 4.0352,
      "step": 73570
    },
    {
      "epoch": 0.0685884658259577,
      "grad_norm": 1.1357803344726562,
      "learning_rate": 0.0001862825044557298,
      "loss": 3.9691,
      "step": 73580
    },
    {
      "epoch": 0.06859778744403679,
      "grad_norm": 1.002220630645752,
      "learning_rate": 0.00018628064013363462,
      "loss": 4.0508,
      "step": 73590
    },
    {
      "epoch": 0.06860710906211588,
      "grad_norm": 1.1542015075683594,
      "learning_rate": 0.0001862787758115394,
      "loss": 3.5242,
      "step": 73600
    },
    {
      "epoch": 0.06861643068019498,
      "grad_norm": 1.2181369066238403,
      "learning_rate": 0.00018627691148944423,
      "loss": 3.9685,
      "step": 73610
    },
    {
      "epoch": 0.06862575229827407,
      "grad_norm": 1.0628085136413574,
      "learning_rate": 0.00018627504716734903,
      "loss": 3.9177,
      "step": 73620
    },
    {
      "epoch": 0.06863507391635316,
      "grad_norm": 1.0226327180862427,
      "learning_rate": 0.00018627318284525381,
      "loss": 3.8593,
      "step": 73630
    },
    {
      "epoch": 0.06864439553443225,
      "grad_norm": 1.212085247039795,
      "learning_rate": 0.00018627131852315862,
      "loss": 3.8375,
      "step": 73640
    },
    {
      "epoch": 0.06865371715251134,
      "grad_norm": 1.0230633020401,
      "learning_rate": 0.0001862694542010634,
      "loss": 3.9068,
      "step": 73650
    },
    {
      "epoch": 0.06866303877059043,
      "grad_norm": 1.0243709087371826,
      "learning_rate": 0.0001862675898789682,
      "loss": 4.0029,
      "step": 73660
    },
    {
      "epoch": 0.06867236038866953,
      "grad_norm": 1.078334093093872,
      "learning_rate": 0.00018626572555687304,
      "loss": 3.9156,
      "step": 73670
    },
    {
      "epoch": 0.06868168200674862,
      "grad_norm": 1.1055697202682495,
      "learning_rate": 0.00018626386123477782,
      "loss": 3.8197,
      "step": 73680
    },
    {
      "epoch": 0.0686910036248277,
      "grad_norm": 1.2071399688720703,
      "learning_rate": 0.00018626199691268262,
      "loss": 3.9843,
      "step": 73690
    },
    {
      "epoch": 0.0687003252429068,
      "grad_norm": 0.9112050533294678,
      "learning_rate": 0.0001862601325905874,
      "loss": 3.7867,
      "step": 73700
    },
    {
      "epoch": 0.06870964686098589,
      "grad_norm": 1.1567519903182983,
      "learning_rate": 0.0001862582682684922,
      "loss": 3.808,
      "step": 73710
    },
    {
      "epoch": 0.06871896847906499,
      "grad_norm": 1.0282293558120728,
      "learning_rate": 0.00018625640394639702,
      "loss": 3.9282,
      "step": 73720
    },
    {
      "epoch": 0.06872829009714408,
      "grad_norm": 0.8833681344985962,
      "learning_rate": 0.00018625453962430182,
      "loss": 3.9187,
      "step": 73730
    },
    {
      "epoch": 0.06873761171522318,
      "grad_norm": 1.093584418296814,
      "learning_rate": 0.00018625267530220663,
      "loss": 3.8104,
      "step": 73740
    },
    {
      "epoch": 0.06874693333330226,
      "grad_norm": 1.1259875297546387,
      "learning_rate": 0.0001862508109801114,
      "loss": 3.7531,
      "step": 73750
    },
    {
      "epoch": 0.06875625495138135,
      "grad_norm": 0.9475046992301941,
      "learning_rate": 0.00018624894665801621,
      "loss": 3.7335,
      "step": 73760
    },
    {
      "epoch": 0.06876557656946045,
      "grad_norm": 0.8909757733345032,
      "learning_rate": 0.00018624708233592102,
      "loss": 3.8948,
      "step": 73770
    },
    {
      "epoch": 0.06877489818753954,
      "grad_norm": 1.1931651830673218,
      "learning_rate": 0.00018624521801382583,
      "loss": 3.824,
      "step": 73780
    },
    {
      "epoch": 0.06878421980561863,
      "grad_norm": 1.1681245565414429,
      "learning_rate": 0.00018624335369173063,
      "loss": 3.9353,
      "step": 73790
    },
    {
      "epoch": 0.06879354142369772,
      "grad_norm": 1.013641357421875,
      "learning_rate": 0.0001862414893696354,
      "loss": 3.8573,
      "step": 73800
    },
    {
      "epoch": 0.06880286304177681,
      "grad_norm": 1.0690491199493408,
      "learning_rate": 0.00018623962504754022,
      "loss": 3.9536,
      "step": 73810
    },
    {
      "epoch": 0.0688121846598559,
      "grad_norm": 1.0246375799179077,
      "learning_rate": 0.00018623776072544503,
      "loss": 3.899,
      "step": 73820
    },
    {
      "epoch": 0.068821506277935,
      "grad_norm": 1.1061170101165771,
      "learning_rate": 0.00018623589640334983,
      "loss": 3.9562,
      "step": 73830
    },
    {
      "epoch": 0.0688308278960141,
      "grad_norm": 1.2183371782302856,
      "learning_rate": 0.00018623403208125464,
      "loss": 3.8285,
      "step": 73840
    },
    {
      "epoch": 0.06884014951409317,
      "grad_norm": 0.8852172493934631,
      "learning_rate": 0.00018623216775915942,
      "loss": 3.7734,
      "step": 73850
    },
    {
      "epoch": 0.06884947113217227,
      "grad_norm": 1.042335033416748,
      "learning_rate": 0.00018623030343706422,
      "loss": 3.8479,
      "step": 73860
    },
    {
      "epoch": 0.06885879275025136,
      "grad_norm": 1.1711136102676392,
      "learning_rate": 0.00018622843911496903,
      "loss": 4.0089,
      "step": 73870
    },
    {
      "epoch": 0.06886811436833046,
      "grad_norm": 1.241899013519287,
      "learning_rate": 0.0001862265747928738,
      "loss": 3.9047,
      "step": 73880
    },
    {
      "epoch": 0.06887743598640955,
      "grad_norm": 1.1298152208328247,
      "learning_rate": 0.00018622471047077864,
      "loss": 3.8806,
      "step": 73890
    },
    {
      "epoch": 0.06888675760448863,
      "grad_norm": 1.1321738958358765,
      "learning_rate": 0.00018622284614868342,
      "loss": 3.9344,
      "step": 73900
    },
    {
      "epoch": 0.06889607922256773,
      "grad_norm": 0.9309511184692383,
      "learning_rate": 0.00018622098182658823,
      "loss": 3.7396,
      "step": 73910
    },
    {
      "epoch": 0.06890540084064682,
      "grad_norm": 0.9859816431999207,
      "learning_rate": 0.00018621911750449303,
      "loss": 3.8541,
      "step": 73920
    },
    {
      "epoch": 0.06891472245872592,
      "grad_norm": 1.0229365825653076,
      "learning_rate": 0.0001862172531823978,
      "loss": 3.832,
      "step": 73930
    },
    {
      "epoch": 0.06892404407680501,
      "grad_norm": 0.9407570958137512,
      "learning_rate": 0.00018621538886030262,
      "loss": 3.8819,
      "step": 73940
    },
    {
      "epoch": 0.06893336569488409,
      "grad_norm": 1.1729191541671753,
      "learning_rate": 0.00018621352453820743,
      "loss": 3.8195,
      "step": 73950
    },
    {
      "epoch": 0.06894268731296319,
      "grad_norm": 1.0060811042785645,
      "learning_rate": 0.00018621166021611223,
      "loss": 3.8481,
      "step": 73960
    },
    {
      "epoch": 0.06895200893104228,
      "grad_norm": 0.9543126225471497,
      "learning_rate": 0.00018620979589401704,
      "loss": 3.7939,
      "step": 73970
    },
    {
      "epoch": 0.06896133054912137,
      "grad_norm": 1.1719515323638916,
      "learning_rate": 0.00018620793157192182,
      "loss": 3.9484,
      "step": 73980
    },
    {
      "epoch": 0.06897065216720047,
      "grad_norm": 1.052700161933899,
      "learning_rate": 0.00018620606724982662,
      "loss": 3.8949,
      "step": 73990
    },
    {
      "epoch": 0.06897997378527956,
      "grad_norm": 1.2576818466186523,
      "learning_rate": 0.00018620420292773143,
      "loss": 3.9389,
      "step": 74000
    },
    {
      "epoch": 0.06898929540335864,
      "grad_norm": 1.1749898195266724,
      "learning_rate": 0.00018620233860563624,
      "loss": 3.8626,
      "step": 74010
    },
    {
      "epoch": 0.06899861702143774,
      "grad_norm": 0.9950922727584839,
      "learning_rate": 0.00018620047428354104,
      "loss": 3.7249,
      "step": 74020
    },
    {
      "epoch": 0.06900793863951683,
      "grad_norm": 1.0418201684951782,
      "learning_rate": 0.00018619860996144582,
      "loss": 3.8304,
      "step": 74030
    },
    {
      "epoch": 0.06901726025759593,
      "grad_norm": 1.1692564487457275,
      "learning_rate": 0.00018619674563935063,
      "loss": 3.9069,
      "step": 74040
    },
    {
      "epoch": 0.06902658187567502,
      "grad_norm": 1.1033692359924316,
      "learning_rate": 0.0001861948813172554,
      "loss": 3.894,
      "step": 74050
    },
    {
      "epoch": 0.0690359034937541,
      "grad_norm": 0.939429759979248,
      "learning_rate": 0.00018619301699516024,
      "loss": 4.0201,
      "step": 74060
    },
    {
      "epoch": 0.0690452251118332,
      "grad_norm": 0.9996210932731628,
      "learning_rate": 0.00018619115267306505,
      "loss": 3.8711,
      "step": 74070
    },
    {
      "epoch": 0.06905454672991229,
      "grad_norm": 0.8950780630111694,
      "learning_rate": 0.00018618928835096983,
      "loss": 3.9655,
      "step": 74080
    },
    {
      "epoch": 0.06906386834799139,
      "grad_norm": 1.1927027702331543,
      "learning_rate": 0.00018618742402887463,
      "loss": 3.5993,
      "step": 74090
    },
    {
      "epoch": 0.06907318996607048,
      "grad_norm": 1.1000522375106812,
      "learning_rate": 0.0001861855597067794,
      "loss": 3.8747,
      "step": 74100
    },
    {
      "epoch": 0.06908251158414956,
      "grad_norm": 1.1231689453125,
      "learning_rate": 0.00018618369538468422,
      "loss": 4.0494,
      "step": 74110
    },
    {
      "epoch": 0.06909183320222866,
      "grad_norm": 1.121559739112854,
      "learning_rate": 0.00018618183106258905,
      "loss": 3.7214,
      "step": 74120
    },
    {
      "epoch": 0.06910115482030775,
      "grad_norm": 1.0510871410369873,
      "learning_rate": 0.00018617996674049383,
      "loss": 3.9818,
      "step": 74130
    },
    {
      "epoch": 0.06911047643838684,
      "grad_norm": 1.2581771612167358,
      "learning_rate": 0.00018617810241839864,
      "loss": 3.9404,
      "step": 74140
    },
    {
      "epoch": 0.06911979805646594,
      "grad_norm": 0.9900798797607422,
      "learning_rate": 0.00018617623809630342,
      "loss": 3.7889,
      "step": 74150
    },
    {
      "epoch": 0.06912911967454502,
      "grad_norm": 1.2007054090499878,
      "learning_rate": 0.00018617437377420822,
      "loss": 3.7676,
      "step": 74160
    },
    {
      "epoch": 0.06913844129262411,
      "grad_norm": 1.0280392169952393,
      "learning_rate": 0.00018617250945211303,
      "loss": 3.8696,
      "step": 74170
    },
    {
      "epoch": 0.06914776291070321,
      "grad_norm": 1.0525439977645874,
      "learning_rate": 0.00018617064513001784,
      "loss": 3.8001,
      "step": 74180
    },
    {
      "epoch": 0.0691570845287823,
      "grad_norm": 1.1596851348876953,
      "learning_rate": 0.00018616878080792264,
      "loss": 3.7783,
      "step": 74190
    },
    {
      "epoch": 0.0691664061468614,
      "grad_norm": 1.0389493703842163,
      "learning_rate": 0.00018616691648582742,
      "loss": 3.8906,
      "step": 74200
    },
    {
      "epoch": 0.06917572776494049,
      "grad_norm": 1.003970980644226,
      "learning_rate": 0.00018616505216373223,
      "loss": 3.849,
      "step": 74210
    },
    {
      "epoch": 0.06918504938301957,
      "grad_norm": 1.1430243253707886,
      "learning_rate": 0.00018616318784163703,
      "loss": 3.9684,
      "step": 74220
    },
    {
      "epoch": 0.06919437100109867,
      "grad_norm": 1.0066595077514648,
      "learning_rate": 0.00018616132351954184,
      "loss": 3.8756,
      "step": 74230
    },
    {
      "epoch": 0.06920369261917776,
      "grad_norm": 1.037667989730835,
      "learning_rate": 0.00018615945919744665,
      "loss": 3.8086,
      "step": 74240
    },
    {
      "epoch": 0.06921301423725686,
      "grad_norm": 0.9901620745658875,
      "learning_rate": 0.00018615759487535143,
      "loss": 3.8538,
      "step": 74250
    },
    {
      "epoch": 0.06922233585533595,
      "grad_norm": 1.1089056730270386,
      "learning_rate": 0.00018615573055325623,
      "loss": 3.9028,
      "step": 74260
    },
    {
      "epoch": 0.06923165747341503,
      "grad_norm": 1.067518949508667,
      "learning_rate": 0.00018615386623116104,
      "loss": 3.8609,
      "step": 74270
    },
    {
      "epoch": 0.06924097909149413,
      "grad_norm": 1.150364637374878,
      "learning_rate": 0.00018615200190906584,
      "loss": 3.9262,
      "step": 74280
    },
    {
      "epoch": 0.06925030070957322,
      "grad_norm": 0.9974280595779419,
      "learning_rate": 0.00018615013758697065,
      "loss": 3.925,
      "step": 74290
    },
    {
      "epoch": 0.06925962232765231,
      "grad_norm": 1.311936855316162,
      "learning_rate": 0.00018614827326487543,
      "loss": 4.0601,
      "step": 74300
    },
    {
      "epoch": 0.06926894394573141,
      "grad_norm": 1.1089750528335571,
      "learning_rate": 0.00018614640894278024,
      "loss": 3.7195,
      "step": 74310
    },
    {
      "epoch": 0.06927826556381049,
      "grad_norm": 0.9485031366348267,
      "learning_rate": 0.00018614454462068504,
      "loss": 3.8185,
      "step": 74320
    },
    {
      "epoch": 0.06928758718188958,
      "grad_norm": 0.9114660620689392,
      "learning_rate": 0.00018614268029858982,
      "loss": 3.6756,
      "step": 74330
    },
    {
      "epoch": 0.06929690879996868,
      "grad_norm": 1.001649260520935,
      "learning_rate": 0.00018614081597649465,
      "loss": 3.8819,
      "step": 74340
    },
    {
      "epoch": 0.06930623041804777,
      "grad_norm": 0.9750105738639832,
      "learning_rate": 0.00018613895165439943,
      "loss": 3.9141,
      "step": 74350
    },
    {
      "epoch": 0.06931555203612687,
      "grad_norm": 0.7337448596954346,
      "learning_rate": 0.00018613708733230424,
      "loss": 3.8395,
      "step": 74360
    },
    {
      "epoch": 0.06932487365420595,
      "grad_norm": 1.1806303262710571,
      "learning_rate": 0.00018613522301020905,
      "loss": 3.8197,
      "step": 74370
    },
    {
      "epoch": 0.06933419527228504,
      "grad_norm": 1.0502721071243286,
      "learning_rate": 0.00018613335868811383,
      "loss": 3.7621,
      "step": 74380
    },
    {
      "epoch": 0.06934351689036414,
      "grad_norm": 1.0507595539093018,
      "learning_rate": 0.00018613149436601863,
      "loss": 3.9782,
      "step": 74390
    },
    {
      "epoch": 0.06935283850844323,
      "grad_norm": 1.2398895025253296,
      "learning_rate": 0.00018612963004392344,
      "loss": 3.7014,
      "step": 74400
    },
    {
      "epoch": 0.06936216012652233,
      "grad_norm": 0.935282826423645,
      "learning_rate": 0.00018612776572182824,
      "loss": 3.637,
      "step": 74410
    },
    {
      "epoch": 0.06937148174460142,
      "grad_norm": 0.920843243598938,
      "learning_rate": 0.00018612590139973302,
      "loss": 3.8892,
      "step": 74420
    },
    {
      "epoch": 0.0693808033626805,
      "grad_norm": 1.0187493562698364,
      "learning_rate": 0.00018612403707763783,
      "loss": 3.9006,
      "step": 74430
    },
    {
      "epoch": 0.0693901249807596,
      "grad_norm": 1.1472282409667969,
      "learning_rate": 0.00018612217275554264,
      "loss": 3.7792,
      "step": 74440
    },
    {
      "epoch": 0.06939944659883869,
      "grad_norm": 0.9664352536201477,
      "learning_rate": 0.00018612030843344744,
      "loss": 3.8674,
      "step": 74450
    },
    {
      "epoch": 0.06940876821691778,
      "grad_norm": 1.1754306554794312,
      "learning_rate": 0.00018611844411135225,
      "loss": 3.8684,
      "step": 74460
    },
    {
      "epoch": 0.06941808983499688,
      "grad_norm": 0.9218648672103882,
      "learning_rate": 0.00018611657978925703,
      "loss": 3.7212,
      "step": 74470
    },
    {
      "epoch": 0.06942741145307596,
      "grad_norm": 1.0869603157043457,
      "learning_rate": 0.00018611471546716183,
      "loss": 4.0656,
      "step": 74480
    },
    {
      "epoch": 0.06943673307115505,
      "grad_norm": 0.9243403077125549,
      "learning_rate": 0.00018611285114506664,
      "loss": 3.9753,
      "step": 74490
    },
    {
      "epoch": 0.06944605468923415,
      "grad_norm": 1.207317590713501,
      "learning_rate": 0.00018611098682297142,
      "loss": 3.6791,
      "step": 74500
    },
    {
      "epoch": 0.06945537630731324,
      "grad_norm": 1.3600550889968872,
      "learning_rate": 0.00018610912250087625,
      "loss": 3.9357,
      "step": 74510
    },
    {
      "epoch": 0.06946469792539234,
      "grad_norm": 1.163638710975647,
      "learning_rate": 0.00018610725817878103,
      "loss": 3.836,
      "step": 74520
    },
    {
      "epoch": 0.06947401954347142,
      "grad_norm": 1.1969044208526611,
      "learning_rate": 0.00018610539385668584,
      "loss": 3.9123,
      "step": 74530
    },
    {
      "epoch": 0.06948334116155051,
      "grad_norm": 1.188658356666565,
      "learning_rate": 0.00018610352953459065,
      "loss": 3.6937,
      "step": 74540
    },
    {
      "epoch": 0.06949266277962961,
      "grad_norm": 1.152466893196106,
      "learning_rate": 0.00018610166521249542,
      "loss": 3.922,
      "step": 74550
    },
    {
      "epoch": 0.0695019843977087,
      "grad_norm": 0.9761536717414856,
      "learning_rate": 0.00018609980089040023,
      "loss": 3.8106,
      "step": 74560
    },
    {
      "epoch": 0.0695113060157878,
      "grad_norm": 0.8856306076049805,
      "learning_rate": 0.00018609793656830504,
      "loss": 4.161,
      "step": 74570
    },
    {
      "epoch": 0.06952062763386688,
      "grad_norm": 1.0268487930297852,
      "learning_rate": 0.00018609607224620984,
      "loss": 3.8046,
      "step": 74580
    },
    {
      "epoch": 0.06952994925194597,
      "grad_norm": 1.0687386989593506,
      "learning_rate": 0.00018609420792411465,
      "loss": 3.9297,
      "step": 74590
    },
    {
      "epoch": 0.06953927087002507,
      "grad_norm": 1.0399694442749023,
      "learning_rate": 0.00018609234360201943,
      "loss": 3.9823,
      "step": 74600
    },
    {
      "epoch": 0.06954859248810416,
      "grad_norm": 1.0344343185424805,
      "learning_rate": 0.00018609047927992424,
      "loss": 3.8571,
      "step": 74610
    },
    {
      "epoch": 0.06955791410618326,
      "grad_norm": 0.9826962947845459,
      "learning_rate": 0.00018608861495782904,
      "loss": 3.9976,
      "step": 74620
    },
    {
      "epoch": 0.06956723572426234,
      "grad_norm": 1.0256730318069458,
      "learning_rate": 0.00018608675063573385,
      "loss": 3.9041,
      "step": 74630
    },
    {
      "epoch": 0.06957655734234143,
      "grad_norm": 1.1232075691223145,
      "learning_rate": 0.00018608488631363865,
      "loss": 3.8411,
      "step": 74640
    },
    {
      "epoch": 0.06958587896042052,
      "grad_norm": 1.1165199279785156,
      "learning_rate": 0.00018608302199154343,
      "loss": 3.7957,
      "step": 74650
    },
    {
      "epoch": 0.06959520057849962,
      "grad_norm": 0.9812337756156921,
      "learning_rate": 0.00018608115766944824,
      "loss": 3.6887,
      "step": 74660
    },
    {
      "epoch": 0.06960452219657871,
      "grad_norm": 0.892043948173523,
      "learning_rate": 0.00018607929334735305,
      "loss": 3.788,
      "step": 74670
    },
    {
      "epoch": 0.06961384381465781,
      "grad_norm": 0.8681343793869019,
      "learning_rate": 0.00018607742902525785,
      "loss": 3.8325,
      "step": 74680
    },
    {
      "epoch": 0.06962316543273689,
      "grad_norm": 1.0779509544372559,
      "learning_rate": 0.00018607556470316266,
      "loss": 3.7889,
      "step": 74690
    },
    {
      "epoch": 0.06963248705081598,
      "grad_norm": 1.3571935892105103,
      "learning_rate": 0.00018607370038106744,
      "loss": 3.8351,
      "step": 74700
    },
    {
      "epoch": 0.06964180866889508,
      "grad_norm": 1.1115292310714722,
      "learning_rate": 0.00018607183605897224,
      "loss": 3.9195,
      "step": 74710
    },
    {
      "epoch": 0.06965113028697417,
      "grad_norm": 1.300159215927124,
      "learning_rate": 0.00018606997173687702,
      "loss": 3.8402,
      "step": 74720
    },
    {
      "epoch": 0.06966045190505327,
      "grad_norm": 1.1506918668746948,
      "learning_rate": 0.00018606810741478186,
      "loss": 3.7907,
      "step": 74730
    },
    {
      "epoch": 0.06966977352313235,
      "grad_norm": 1.010364294052124,
      "learning_rate": 0.00018606624309268666,
      "loss": 3.9966,
      "step": 74740
    },
    {
      "epoch": 0.06967909514121144,
      "grad_norm": 1.197279930114746,
      "learning_rate": 0.00018606437877059144,
      "loss": 3.7079,
      "step": 74750
    },
    {
      "epoch": 0.06968841675929054,
      "grad_norm": 1.0575881004333496,
      "learning_rate": 0.00018606251444849625,
      "loss": 3.9252,
      "step": 74760
    },
    {
      "epoch": 0.06969773837736963,
      "grad_norm": 0.8686604499816895,
      "learning_rate": 0.00018606065012640103,
      "loss": 3.7487,
      "step": 74770
    },
    {
      "epoch": 0.06970705999544873,
      "grad_norm": 1.0863717794418335,
      "learning_rate": 0.00018605878580430583,
      "loss": 3.9131,
      "step": 74780
    },
    {
      "epoch": 0.0697163816135278,
      "grad_norm": 1.031425952911377,
      "learning_rate": 0.00018605692148221067,
      "loss": 3.8638,
      "step": 74790
    },
    {
      "epoch": 0.0697257032316069,
      "grad_norm": 1.0484769344329834,
      "learning_rate": 0.00018605505716011545,
      "loss": 3.996,
      "step": 74800
    },
    {
      "epoch": 0.069735024849686,
      "grad_norm": 0.9872491955757141,
      "learning_rate": 0.00018605319283802025,
      "loss": 3.647,
      "step": 74810
    },
    {
      "epoch": 0.06974434646776509,
      "grad_norm": 1.0154893398284912,
      "learning_rate": 0.00018605132851592503,
      "loss": 3.7701,
      "step": 74820
    },
    {
      "epoch": 0.06975366808584418,
      "grad_norm": 1.0602853298187256,
      "learning_rate": 0.00018604946419382984,
      "loss": 3.8606,
      "step": 74830
    },
    {
      "epoch": 0.06976298970392326,
      "grad_norm": 1.2524832487106323,
      "learning_rate": 0.00018604759987173465,
      "loss": 4.0698,
      "step": 74840
    },
    {
      "epoch": 0.06977231132200236,
      "grad_norm": 1.1790937185287476,
      "learning_rate": 0.00018604573554963945,
      "loss": 3.7831,
      "step": 74850
    },
    {
      "epoch": 0.06978163294008145,
      "grad_norm": 1.2661973237991333,
      "learning_rate": 0.00018604387122754426,
      "loss": 3.8571,
      "step": 74860
    },
    {
      "epoch": 0.06979095455816055,
      "grad_norm": 0.9871660470962524,
      "learning_rate": 0.00018604200690544904,
      "loss": 3.8575,
      "step": 74870
    },
    {
      "epoch": 0.06980027617623964,
      "grad_norm": 1.1179152727127075,
      "learning_rate": 0.00018604014258335384,
      "loss": 3.7054,
      "step": 74880
    },
    {
      "epoch": 0.06980959779431874,
      "grad_norm": 1.2621937990188599,
      "learning_rate": 0.00018603827826125865,
      "loss": 3.7501,
      "step": 74890
    },
    {
      "epoch": 0.06981891941239782,
      "grad_norm": 0.9689948558807373,
      "learning_rate": 0.00018603641393916346,
      "loss": 3.9924,
      "step": 74900
    },
    {
      "epoch": 0.06982824103047691,
      "grad_norm": 0.9596605896949768,
      "learning_rate": 0.00018603454961706826,
      "loss": 3.9732,
      "step": 74910
    },
    {
      "epoch": 0.069837562648556,
      "grad_norm": 1.0867834091186523,
      "learning_rate": 0.00018603268529497304,
      "loss": 3.8703,
      "step": 74920
    },
    {
      "epoch": 0.0698468842666351,
      "grad_norm": 1.3108235597610474,
      "learning_rate": 0.00018603082097287785,
      "loss": 3.7687,
      "step": 74930
    },
    {
      "epoch": 0.0698562058847142,
      "grad_norm": 1.1246867179870605,
      "learning_rate": 0.00018602895665078265,
      "loss": 3.8566,
      "step": 74940
    },
    {
      "epoch": 0.06986552750279328,
      "grad_norm": 1.0343809127807617,
      "learning_rate": 0.00018602709232868746,
      "loss": 3.7593,
      "step": 74950
    },
    {
      "epoch": 0.06987484912087237,
      "grad_norm": 1.0223634243011475,
      "learning_rate": 0.00018602522800659227,
      "loss": 3.7341,
      "step": 74960
    },
    {
      "epoch": 0.06988417073895146,
      "grad_norm": 1.3345303535461426,
      "learning_rate": 0.00018602336368449705,
      "loss": 3.6929,
      "step": 74970
    },
    {
      "epoch": 0.06989349235703056,
      "grad_norm": 1.0798686742782593,
      "learning_rate": 0.00018602149936240185,
      "loss": 3.8266,
      "step": 74980
    },
    {
      "epoch": 0.06990281397510965,
      "grad_norm": 1.083769679069519,
      "learning_rate": 0.00018601963504030666,
      "loss": 3.8313,
      "step": 74990
    },
    {
      "epoch": 0.06991213559318873,
      "grad_norm": 0.9615893959999084,
      "learning_rate": 0.00018601777071821144,
      "loss": 3.8964,
      "step": 75000
    },
    {
      "epoch": 0.06992145721126783,
      "grad_norm": 1.1958037614822388,
      "learning_rate": 0.00018601590639611624,
      "loss": 3.748,
      "step": 75010
    },
    {
      "epoch": 0.06993077882934692,
      "grad_norm": 1.0256730318069458,
      "learning_rate": 0.00018601404207402105,
      "loss": 3.8896,
      "step": 75020
    },
    {
      "epoch": 0.06994010044742602,
      "grad_norm": 1.0833489894866943,
      "learning_rate": 0.00018601217775192586,
      "loss": 4.1167,
      "step": 75030
    },
    {
      "epoch": 0.06994942206550511,
      "grad_norm": 1.2684459686279297,
      "learning_rate": 0.00018601031342983066,
      "loss": 3.9068,
      "step": 75040
    },
    {
      "epoch": 0.06995874368358419,
      "grad_norm": 0.9755167961120605,
      "learning_rate": 0.00018600844910773544,
      "loss": 3.9279,
      "step": 75050
    },
    {
      "epoch": 0.06996806530166329,
      "grad_norm": 1.0750012397766113,
      "learning_rate": 0.00018600658478564025,
      "loss": 3.7962,
      "step": 75060
    },
    {
      "epoch": 0.06997738691974238,
      "grad_norm": 1.2461652755737305,
      "learning_rate": 0.00018600472046354505,
      "loss": 3.9799,
      "step": 75070
    },
    {
      "epoch": 0.06998670853782148,
      "grad_norm": 1.1155582666397095,
      "learning_rate": 0.00018600285614144986,
      "loss": 3.8299,
      "step": 75080
    },
    {
      "epoch": 0.06999603015590057,
      "grad_norm": 1.273224949836731,
      "learning_rate": 0.00018600099181935467,
      "loss": 3.4735,
      "step": 75090
    },
    {
      "epoch": 0.07000535177397967,
      "grad_norm": 1.23464834690094,
      "learning_rate": 0.00018599912749725945,
      "loss": 3.9436,
      "step": 75100
    },
    {
      "epoch": 0.07001467339205875,
      "grad_norm": 0.8989088535308838,
      "learning_rate": 0.00018599726317516425,
      "loss": 4.0735,
      "step": 75110
    },
    {
      "epoch": 0.07002399501013784,
      "grad_norm": 1.0529736280441284,
      "learning_rate": 0.00018599539885306906,
      "loss": 3.9667,
      "step": 75120
    },
    {
      "epoch": 0.07003331662821694,
      "grad_norm": 1.0269843339920044,
      "learning_rate": 0.00018599353453097387,
      "loss": 3.7362,
      "step": 75130
    },
    {
      "epoch": 0.07004263824629603,
      "grad_norm": 0.9261894822120667,
      "learning_rate": 0.00018599167020887867,
      "loss": 3.7307,
      "step": 75140
    },
    {
      "epoch": 0.07005195986437512,
      "grad_norm": 1.2373100519180298,
      "learning_rate": 0.00018598980588678345,
      "loss": 3.891,
      "step": 75150
    },
    {
      "epoch": 0.0700612814824542,
      "grad_norm": 1.0718302726745605,
      "learning_rate": 0.00018598794156468826,
      "loss": 3.9485,
      "step": 75160
    },
    {
      "epoch": 0.0700706031005333,
      "grad_norm": 0.9708103537559509,
      "learning_rate": 0.00018598607724259304,
      "loss": 3.799,
      "step": 75170
    },
    {
      "epoch": 0.0700799247186124,
      "grad_norm": 1.1206880807876587,
      "learning_rate": 0.00018598421292049787,
      "loss": 3.7667,
      "step": 75180
    },
    {
      "epoch": 0.07008924633669149,
      "grad_norm": 1.1172244548797607,
      "learning_rate": 0.00018598234859840268,
      "loss": 3.8686,
      "step": 75190
    },
    {
      "epoch": 0.07009856795477058,
      "grad_norm": 1.0298411846160889,
      "learning_rate": 0.00018598048427630746,
      "loss": 3.82,
      "step": 75200
    },
    {
      "epoch": 0.07010788957284966,
      "grad_norm": 1.030705451965332,
      "learning_rate": 0.00018597861995421226,
      "loss": 3.834,
      "step": 75210
    },
    {
      "epoch": 0.07011721119092876,
      "grad_norm": 1.11429762840271,
      "learning_rate": 0.00018597675563211704,
      "loss": 3.822,
      "step": 75220
    },
    {
      "epoch": 0.07012653280900785,
      "grad_norm": 1.001128911972046,
      "learning_rate": 0.00018597489131002185,
      "loss": 3.8072,
      "step": 75230
    },
    {
      "epoch": 0.07013585442708695,
      "grad_norm": 0.8608624339103699,
      "learning_rate": 0.00018597302698792668,
      "loss": 3.8623,
      "step": 75240
    },
    {
      "epoch": 0.07014517604516604,
      "grad_norm": 0.8741585612297058,
      "learning_rate": 0.00018597116266583146,
      "loss": 3.759,
      "step": 75250
    },
    {
      "epoch": 0.07015449766324512,
      "grad_norm": 1.1937283277511597,
      "learning_rate": 0.00018596929834373627,
      "loss": 3.8604,
      "step": 75260
    },
    {
      "epoch": 0.07016381928132422,
      "grad_norm": 1.2035274505615234,
      "learning_rate": 0.00018596743402164105,
      "loss": 3.8717,
      "step": 75270
    },
    {
      "epoch": 0.07017314089940331,
      "grad_norm": 0.9273458123207092,
      "learning_rate": 0.00018596556969954585,
      "loss": 3.9502,
      "step": 75280
    },
    {
      "epoch": 0.0701824625174824,
      "grad_norm": 1.1426433324813843,
      "learning_rate": 0.00018596370537745066,
      "loss": 3.8015,
      "step": 75290
    },
    {
      "epoch": 0.0701917841355615,
      "grad_norm": 1.0645331144332886,
      "learning_rate": 0.00018596184105535546,
      "loss": 3.6709,
      "step": 75300
    },
    {
      "epoch": 0.07020110575364058,
      "grad_norm": 1.022886872291565,
      "learning_rate": 0.00018595997673326027,
      "loss": 3.9408,
      "step": 75310
    },
    {
      "epoch": 0.07021042737171967,
      "grad_norm": 0.931351900100708,
      "learning_rate": 0.00018595811241116505,
      "loss": 3.9102,
      "step": 75320
    },
    {
      "epoch": 0.07021974898979877,
      "grad_norm": 1.2018009424209595,
      "learning_rate": 0.00018595624808906986,
      "loss": 3.8152,
      "step": 75330
    },
    {
      "epoch": 0.07022907060787786,
      "grad_norm": 1.0935373306274414,
      "learning_rate": 0.00018595438376697466,
      "loss": 3.8991,
      "step": 75340
    },
    {
      "epoch": 0.07023839222595696,
      "grad_norm": 1.0044918060302734,
      "learning_rate": 0.00018595251944487947,
      "loss": 3.9002,
      "step": 75350
    },
    {
      "epoch": 0.07024771384403605,
      "grad_norm": 1.1348955631256104,
      "learning_rate": 0.00018595065512278428,
      "loss": 3.9759,
      "step": 75360
    },
    {
      "epoch": 0.07025703546211513,
      "grad_norm": 1.1401993036270142,
      "learning_rate": 0.00018594879080068905,
      "loss": 3.8696,
      "step": 75370
    },
    {
      "epoch": 0.07026635708019423,
      "grad_norm": 0.9962758421897888,
      "learning_rate": 0.00018594692647859386,
      "loss": 3.793,
      "step": 75380
    },
    {
      "epoch": 0.07027567869827332,
      "grad_norm": 1.1463472843170166,
      "learning_rate": 0.00018594506215649867,
      "loss": 3.8097,
      "step": 75390
    },
    {
      "epoch": 0.07028500031635242,
      "grad_norm": 1.0664265155792236,
      "learning_rate": 0.00018594319783440347,
      "loss": 3.8578,
      "step": 75400
    },
    {
      "epoch": 0.07029432193443151,
      "grad_norm": 1.0617568492889404,
      "learning_rate": 0.00018594133351230828,
      "loss": 3.6946,
      "step": 75410
    },
    {
      "epoch": 0.07030364355251059,
      "grad_norm": 0.9799560308456421,
      "learning_rate": 0.00018593946919021306,
      "loss": 3.5879,
      "step": 75420
    },
    {
      "epoch": 0.07031296517058969,
      "grad_norm": 1.1473348140716553,
      "learning_rate": 0.00018593760486811786,
      "loss": 3.8922,
      "step": 75430
    },
    {
      "epoch": 0.07032228678866878,
      "grad_norm": 1.0027554035186768,
      "learning_rate": 0.00018593574054602267,
      "loss": 3.7387,
      "step": 75440
    },
    {
      "epoch": 0.07033160840674788,
      "grad_norm": 0.8770893216133118,
      "learning_rate": 0.00018593387622392745,
      "loss": 4.1227,
      "step": 75450
    },
    {
      "epoch": 0.07034093002482697,
      "grad_norm": 0.9664875864982605,
      "learning_rate": 0.00018593201190183228,
      "loss": 3.7213,
      "step": 75460
    },
    {
      "epoch": 0.07035025164290605,
      "grad_norm": 1.1138218641281128,
      "learning_rate": 0.00018593014757973706,
      "loss": 3.683,
      "step": 75470
    },
    {
      "epoch": 0.07035957326098514,
      "grad_norm": 0.919974684715271,
      "learning_rate": 0.00018592828325764187,
      "loss": 3.9773,
      "step": 75480
    },
    {
      "epoch": 0.07036889487906424,
      "grad_norm": 1.1608843803405762,
      "learning_rate": 0.00018592641893554668,
      "loss": 3.838,
      "step": 75490
    },
    {
      "epoch": 0.07037821649714333,
      "grad_norm": 1.2195011377334595,
      "learning_rate": 0.00018592455461345145,
      "loss": 3.9882,
      "step": 75500
    },
    {
      "epoch": 0.07038753811522243,
      "grad_norm": 1.0989296436309814,
      "learning_rate": 0.00018592269029135626,
      "loss": 3.8544,
      "step": 75510
    },
    {
      "epoch": 0.07039685973330151,
      "grad_norm": 0.9844092726707458,
      "learning_rate": 0.00018592082596926107,
      "loss": 3.8803,
      "step": 75520
    },
    {
      "epoch": 0.0704061813513806,
      "grad_norm": 1.3174735307693481,
      "learning_rate": 0.00018591896164716587,
      "loss": 3.7473,
      "step": 75530
    },
    {
      "epoch": 0.0704155029694597,
      "grad_norm": 1.0416388511657715,
      "learning_rate": 0.00018591709732507068,
      "loss": 3.8089,
      "step": 75540
    },
    {
      "epoch": 0.07042482458753879,
      "grad_norm": 1.1808998584747314,
      "learning_rate": 0.00018591523300297546,
      "loss": 3.6408,
      "step": 75550
    },
    {
      "epoch": 0.07043414620561789,
      "grad_norm": 0.9892808794975281,
      "learning_rate": 0.00018591336868088027,
      "loss": 3.747,
      "step": 75560
    },
    {
      "epoch": 0.07044346782369698,
      "grad_norm": 0.9380480051040649,
      "learning_rate": 0.00018591150435878507,
      "loss": 3.8799,
      "step": 75570
    },
    {
      "epoch": 0.07045278944177606,
      "grad_norm": 0.8767538666725159,
      "learning_rate": 0.00018590964003668988,
      "loss": 4.0003,
      "step": 75580
    },
    {
      "epoch": 0.07046211105985516,
      "grad_norm": 1.0556666851043701,
      "learning_rate": 0.00018590777571459468,
      "loss": 3.7531,
      "step": 75590
    },
    {
      "epoch": 0.07047143267793425,
      "grad_norm": 1.0726581811904907,
      "learning_rate": 0.00018590591139249946,
      "loss": 3.8844,
      "step": 75600
    },
    {
      "epoch": 0.07048075429601335,
      "grad_norm": 1.2369953393936157,
      "learning_rate": 0.00018590404707040427,
      "loss": 3.6309,
      "step": 75610
    },
    {
      "epoch": 0.07049007591409244,
      "grad_norm": 1.1973047256469727,
      "learning_rate": 0.00018590218274830905,
      "loss": 3.7998,
      "step": 75620
    },
    {
      "epoch": 0.07049939753217152,
      "grad_norm": 0.9717024564743042,
      "learning_rate": 0.00018590031842621388,
      "loss": 4.0283,
      "step": 75630
    },
    {
      "epoch": 0.07050871915025061,
      "grad_norm": 1.1469064950942993,
      "learning_rate": 0.0001858984541041187,
      "loss": 3.8764,
      "step": 75640
    },
    {
      "epoch": 0.07051804076832971,
      "grad_norm": 1.0636128187179565,
      "learning_rate": 0.00018589658978202347,
      "loss": 3.676,
      "step": 75650
    },
    {
      "epoch": 0.0705273623864088,
      "grad_norm": 0.8732673525810242,
      "learning_rate": 0.00018589472545992827,
      "loss": 3.6947,
      "step": 75660
    },
    {
      "epoch": 0.0705366840044879,
      "grad_norm": 0.8830550312995911,
      "learning_rate": 0.00018589286113783305,
      "loss": 3.7368,
      "step": 75670
    },
    {
      "epoch": 0.07054600562256698,
      "grad_norm": 0.9894245266914368,
      "learning_rate": 0.00018589099681573786,
      "loss": 3.9774,
      "step": 75680
    },
    {
      "epoch": 0.07055532724064607,
      "grad_norm": 1.046812653541565,
      "learning_rate": 0.0001858891324936427,
      "loss": 3.8242,
      "step": 75690
    },
    {
      "epoch": 0.07056464885872517,
      "grad_norm": 1.0890501737594604,
      "learning_rate": 0.00018588726817154747,
      "loss": 3.6198,
      "step": 75700
    },
    {
      "epoch": 0.07057397047680426,
      "grad_norm": 1.0039286613464355,
      "learning_rate": 0.00018588540384945228,
      "loss": 3.7485,
      "step": 75710
    },
    {
      "epoch": 0.07058329209488336,
      "grad_norm": 1.0467021465301514,
      "learning_rate": 0.00018588353952735706,
      "loss": 3.6094,
      "step": 75720
    },
    {
      "epoch": 0.07059261371296244,
      "grad_norm": 1.28109610080719,
      "learning_rate": 0.00018588167520526186,
      "loss": 3.8402,
      "step": 75730
    },
    {
      "epoch": 0.07060193533104153,
      "grad_norm": 1.035730242729187,
      "learning_rate": 0.00018587981088316667,
      "loss": 4.0388,
      "step": 75740
    },
    {
      "epoch": 0.07061125694912063,
      "grad_norm": 1.0064315795898438,
      "learning_rate": 0.00018587794656107148,
      "loss": 3.7634,
      "step": 75750
    },
    {
      "epoch": 0.07062057856719972,
      "grad_norm": 1.0801936388015747,
      "learning_rate": 0.00018587608223897628,
      "loss": 3.7582,
      "step": 75760
    },
    {
      "epoch": 0.07062990018527882,
      "grad_norm": 1.0673562288284302,
      "learning_rate": 0.00018587421791688106,
      "loss": 3.9177,
      "step": 75770
    },
    {
      "epoch": 0.07063922180335791,
      "grad_norm": 1.0285909175872803,
      "learning_rate": 0.00018587235359478587,
      "loss": 3.7522,
      "step": 75780
    },
    {
      "epoch": 0.07064854342143699,
      "grad_norm": 1.0497263669967651,
      "learning_rate": 0.00018587048927269068,
      "loss": 3.9082,
      "step": 75790
    },
    {
      "epoch": 0.07065786503951609,
      "grad_norm": 1.089038610458374,
      "learning_rate": 0.00018586862495059548,
      "loss": 3.7465,
      "step": 75800
    },
    {
      "epoch": 0.07066718665759518,
      "grad_norm": 1.143201470375061,
      "learning_rate": 0.0001858667606285003,
      "loss": 3.8747,
      "step": 75810
    },
    {
      "epoch": 0.07067650827567427,
      "grad_norm": 1.0117108821868896,
      "learning_rate": 0.00018586489630640507,
      "loss": 3.8953,
      "step": 75820
    },
    {
      "epoch": 0.07068582989375337,
      "grad_norm": 1.126678705215454,
      "learning_rate": 0.00018586303198430987,
      "loss": 3.7618,
      "step": 75830
    },
    {
      "epoch": 0.07069515151183245,
      "grad_norm": 1.116170883178711,
      "learning_rate": 0.00018586116766221465,
      "loss": 3.9959,
      "step": 75840
    },
    {
      "epoch": 0.07070447312991154,
      "grad_norm": 1.2626032829284668,
      "learning_rate": 0.00018585930334011949,
      "loss": 3.7783,
      "step": 75850
    },
    {
      "epoch": 0.07071379474799064,
      "grad_norm": 1.2924535274505615,
      "learning_rate": 0.0001858574390180243,
      "loss": 3.7596,
      "step": 75860
    },
    {
      "epoch": 0.07072311636606973,
      "grad_norm": 1.2902793884277344,
      "learning_rate": 0.00018585557469592907,
      "loss": 3.9277,
      "step": 75870
    },
    {
      "epoch": 0.07073243798414883,
      "grad_norm": 1.292900562286377,
      "learning_rate": 0.00018585371037383388,
      "loss": 3.6679,
      "step": 75880
    },
    {
      "epoch": 0.07074175960222791,
      "grad_norm": 1.1686904430389404,
      "learning_rate": 0.00018585184605173866,
      "loss": 3.5718,
      "step": 75890
    },
    {
      "epoch": 0.070751081220307,
      "grad_norm": 1.2729188203811646,
      "learning_rate": 0.00018584998172964346,
      "loss": 3.84,
      "step": 75900
    },
    {
      "epoch": 0.0707604028383861,
      "grad_norm": 0.9984061121940613,
      "learning_rate": 0.0001858481174075483,
      "loss": 3.771,
      "step": 75910
    },
    {
      "epoch": 0.07076972445646519,
      "grad_norm": 1.2873964309692383,
      "learning_rate": 0.00018584625308545308,
      "loss": 3.7239,
      "step": 75920
    },
    {
      "epoch": 0.07077904607454429,
      "grad_norm": 1.1191805601119995,
      "learning_rate": 0.00018584438876335788,
      "loss": 3.7932,
      "step": 75930
    },
    {
      "epoch": 0.07078836769262337,
      "grad_norm": 1.2978702783584595,
      "learning_rate": 0.00018584252444126266,
      "loss": 3.8306,
      "step": 75940
    },
    {
      "epoch": 0.07079768931070246,
      "grad_norm": 0.9319953322410583,
      "learning_rate": 0.00018584066011916747,
      "loss": 3.9598,
      "step": 75950
    },
    {
      "epoch": 0.07080701092878156,
      "grad_norm": 1.1925963163375854,
      "learning_rate": 0.00018583879579707227,
      "loss": 3.9757,
      "step": 75960
    },
    {
      "epoch": 0.07081633254686065,
      "grad_norm": 1.03379487991333,
      "learning_rate": 0.00018583693147497708,
      "loss": 3.9188,
      "step": 75970
    },
    {
      "epoch": 0.07082565416493974,
      "grad_norm": 0.9114783406257629,
      "learning_rate": 0.0001858350671528819,
      "loss": 3.9346,
      "step": 75980
    },
    {
      "epoch": 0.07083497578301882,
      "grad_norm": 1.0329493284225464,
      "learning_rate": 0.00018583320283078667,
      "loss": 3.9813,
      "step": 75990
    },
    {
      "epoch": 0.07084429740109792,
      "grad_norm": 1.0565468072891235,
      "learning_rate": 0.00018583133850869147,
      "loss": 3.8962,
      "step": 76000
    },
    {
      "epoch": 0.07085361901917701,
      "grad_norm": 1.0819518566131592,
      "learning_rate": 0.00018582947418659628,
      "loss": 3.6768,
      "step": 76010
    },
    {
      "epoch": 0.07086294063725611,
      "grad_norm": 1.098228931427002,
      "learning_rate": 0.00018582760986450108,
      "loss": 3.9622,
      "step": 76020
    },
    {
      "epoch": 0.0708722622553352,
      "grad_norm": 0.8778983354568481,
      "learning_rate": 0.0001858257455424059,
      "loss": 3.8766,
      "step": 76030
    },
    {
      "epoch": 0.0708815838734143,
      "grad_norm": 1.1234443187713623,
      "learning_rate": 0.00018582388122031067,
      "loss": 3.7827,
      "step": 76040
    },
    {
      "epoch": 0.07089090549149338,
      "grad_norm": 1.0351436138153076,
      "learning_rate": 0.00018582201689821548,
      "loss": 3.8016,
      "step": 76050
    },
    {
      "epoch": 0.07090022710957247,
      "grad_norm": 1.1457127332687378,
      "learning_rate": 0.00018582015257612028,
      "loss": 3.9078,
      "step": 76060
    },
    {
      "epoch": 0.07090954872765157,
      "grad_norm": 1.053274154663086,
      "learning_rate": 0.00018581828825402506,
      "loss": 3.8773,
      "step": 76070
    },
    {
      "epoch": 0.07091887034573066,
      "grad_norm": 1.2078109979629517,
      "learning_rate": 0.0001858164239319299,
      "loss": 4.0194,
      "step": 76080
    },
    {
      "epoch": 0.07092819196380976,
      "grad_norm": 1.1229315996170044,
      "learning_rate": 0.00018581455960983467,
      "loss": 3.7537,
      "step": 76090
    },
    {
      "epoch": 0.07093751358188884,
      "grad_norm": 1.2436720132827759,
      "learning_rate": 0.00018581269528773948,
      "loss": 3.8924,
      "step": 76100
    },
    {
      "epoch": 0.07094683519996793,
      "grad_norm": 1.0662245750427246,
      "learning_rate": 0.0001858108309656443,
      "loss": 3.8712,
      "step": 76110
    },
    {
      "epoch": 0.07095615681804703,
      "grad_norm": 1.148516058921814,
      "learning_rate": 0.00018580896664354907,
      "loss": 3.7561,
      "step": 76120
    },
    {
      "epoch": 0.07096547843612612,
      "grad_norm": 1.143262505531311,
      "learning_rate": 0.00018580710232145387,
      "loss": 3.8067,
      "step": 76130
    },
    {
      "epoch": 0.07097480005420521,
      "grad_norm": 1.0495432615280151,
      "learning_rate": 0.00018580523799935868,
      "loss": 3.9611,
      "step": 76140
    },
    {
      "epoch": 0.0709841216722843,
      "grad_norm": 1.2873883247375488,
      "learning_rate": 0.00018580337367726349,
      "loss": 3.6247,
      "step": 76150
    },
    {
      "epoch": 0.07099344329036339,
      "grad_norm": 1.0379678010940552,
      "learning_rate": 0.0001858015093551683,
      "loss": 3.8698,
      "step": 76160
    },
    {
      "epoch": 0.07100276490844248,
      "grad_norm": 0.8533099889755249,
      "learning_rate": 0.00018579964503307307,
      "loss": 4.0004,
      "step": 76170
    },
    {
      "epoch": 0.07101208652652158,
      "grad_norm": 1.0021851062774658,
      "learning_rate": 0.00018579778071097788,
      "loss": 3.8801,
      "step": 76180
    },
    {
      "epoch": 0.07102140814460067,
      "grad_norm": 1.0728851556777954,
      "learning_rate": 0.00018579591638888268,
      "loss": 3.5923,
      "step": 76190
    },
    {
      "epoch": 0.07103072976267975,
      "grad_norm": 1.0408434867858887,
      "learning_rate": 0.0001857940520667875,
      "loss": 3.9267,
      "step": 76200
    },
    {
      "epoch": 0.07104005138075885,
      "grad_norm": 1.0206468105316162,
      "learning_rate": 0.0001857921877446923,
      "loss": 3.9753,
      "step": 76210
    },
    {
      "epoch": 0.07104937299883794,
      "grad_norm": 1.0570502281188965,
      "learning_rate": 0.00018579032342259708,
      "loss": 4.0478,
      "step": 76220
    },
    {
      "epoch": 0.07105869461691704,
      "grad_norm": 1.2406349182128906,
      "learning_rate": 0.00018578845910050188,
      "loss": 3.9587,
      "step": 76230
    },
    {
      "epoch": 0.07106801623499613,
      "grad_norm": 0.976159930229187,
      "learning_rate": 0.0001857865947784067,
      "loss": 3.616,
      "step": 76240
    },
    {
      "epoch": 0.07107733785307523,
      "grad_norm": 1.1969261169433594,
      "learning_rate": 0.0001857847304563115,
      "loss": 3.859,
      "step": 76250
    },
    {
      "epoch": 0.0710866594711543,
      "grad_norm": 1.1079967021942139,
      "learning_rate": 0.0001857828661342163,
      "loss": 3.7657,
      "step": 76260
    },
    {
      "epoch": 0.0710959810892334,
      "grad_norm": 1.2883684635162354,
      "learning_rate": 0.00018578100181212108,
      "loss": 3.8578,
      "step": 76270
    },
    {
      "epoch": 0.0711053027073125,
      "grad_norm": 1.3612744808197021,
      "learning_rate": 0.00018577913749002589,
      "loss": 3.9287,
      "step": 76280
    },
    {
      "epoch": 0.07111462432539159,
      "grad_norm": 1.0774716138839722,
      "learning_rate": 0.00018577727316793067,
      "loss": 3.8435,
      "step": 76290
    },
    {
      "epoch": 0.07112394594347068,
      "grad_norm": 1.0396082401275635,
      "learning_rate": 0.0001857754088458355,
      "loss": 4.0847,
      "step": 76300
    },
    {
      "epoch": 0.07113326756154977,
      "grad_norm": 1.0388888120651245,
      "learning_rate": 0.0001857735445237403,
      "loss": 3.9448,
      "step": 76310
    },
    {
      "epoch": 0.07114258917962886,
      "grad_norm": 1.027134895324707,
      "learning_rate": 0.00018577168020164508,
      "loss": 3.9073,
      "step": 76320
    },
    {
      "epoch": 0.07115191079770795,
      "grad_norm": 1.1312083005905151,
      "learning_rate": 0.0001857698158795499,
      "loss": 3.9075,
      "step": 76330
    },
    {
      "epoch": 0.07116123241578705,
      "grad_norm": 1.389328122138977,
      "learning_rate": 0.00018576795155745467,
      "loss": 3.7852,
      "step": 76340
    },
    {
      "epoch": 0.07117055403386614,
      "grad_norm": 1.107964038848877,
      "learning_rate": 0.00018576608723535948,
      "loss": 3.8686,
      "step": 76350
    },
    {
      "epoch": 0.07117987565194522,
      "grad_norm": 1.197181224822998,
      "learning_rate": 0.0001857642229132643,
      "loss": 3.8336,
      "step": 76360
    },
    {
      "epoch": 0.07118919727002432,
      "grad_norm": 0.8784891963005066,
      "learning_rate": 0.0001857623585911691,
      "loss": 3.7973,
      "step": 76370
    },
    {
      "epoch": 0.07119851888810341,
      "grad_norm": 1.0759315490722656,
      "learning_rate": 0.0001857604942690739,
      "loss": 3.6119,
      "step": 76380
    },
    {
      "epoch": 0.07120784050618251,
      "grad_norm": 1.0184338092803955,
      "learning_rate": 0.00018575862994697867,
      "loss": 3.8131,
      "step": 76390
    },
    {
      "epoch": 0.0712171621242616,
      "grad_norm": 1.274472713470459,
      "learning_rate": 0.00018575676562488348,
      "loss": 3.554,
      "step": 76400
    },
    {
      "epoch": 0.07122648374234068,
      "grad_norm": 1.3019208908081055,
      "learning_rate": 0.0001857549013027883,
      "loss": 3.6336,
      "step": 76410
    },
    {
      "epoch": 0.07123580536041978,
      "grad_norm": 1.1667877435684204,
      "learning_rate": 0.0001857530369806931,
      "loss": 3.8073,
      "step": 76420
    },
    {
      "epoch": 0.07124512697849887,
      "grad_norm": 1.142343282699585,
      "learning_rate": 0.0001857511726585979,
      "loss": 3.768,
      "step": 76430
    },
    {
      "epoch": 0.07125444859657797,
      "grad_norm": 0.8723321557044983,
      "learning_rate": 0.00018574930833650268,
      "loss": 3.8804,
      "step": 76440
    },
    {
      "epoch": 0.07126377021465706,
      "grad_norm": 1.0313915014266968,
      "learning_rate": 0.00018574744401440749,
      "loss": 3.8651,
      "step": 76450
    },
    {
      "epoch": 0.07127309183273615,
      "grad_norm": 1.0095351934432983,
      "learning_rate": 0.0001857455796923123,
      "loss": 3.9507,
      "step": 76460
    },
    {
      "epoch": 0.07128241345081524,
      "grad_norm": 1.1545809507369995,
      "learning_rate": 0.0001857437153702171,
      "loss": 3.8397,
      "step": 76470
    },
    {
      "epoch": 0.07129173506889433,
      "grad_norm": 1.3250758647918701,
      "learning_rate": 0.0001857418510481219,
      "loss": 3.7789,
      "step": 76480
    },
    {
      "epoch": 0.07130105668697342,
      "grad_norm": 1.0497057437896729,
      "learning_rate": 0.00018573998672602668,
      "loss": 3.7872,
      "step": 76490
    },
    {
      "epoch": 0.07131037830505252,
      "grad_norm": 0.8675214648246765,
      "learning_rate": 0.0001857381224039315,
      "loss": 3.9127,
      "step": 76500
    },
    {
      "epoch": 0.07131969992313161,
      "grad_norm": 1.1271265745162964,
      "learning_rate": 0.0001857362580818363,
      "loss": 3.7996,
      "step": 76510
    },
    {
      "epoch": 0.0713290215412107,
      "grad_norm": 1.0796054601669312,
      "learning_rate": 0.00018573439375974107,
      "loss": 3.8735,
      "step": 76520
    },
    {
      "epoch": 0.07133834315928979,
      "grad_norm": 1.0830163955688477,
      "learning_rate": 0.0001857325294376459,
      "loss": 3.8174,
      "step": 76530
    },
    {
      "epoch": 0.07134766477736888,
      "grad_norm": 0.9104672074317932,
      "learning_rate": 0.0001857306651155507,
      "loss": 3.8428,
      "step": 76540
    },
    {
      "epoch": 0.07135698639544798,
      "grad_norm": 1.2587593793869019,
      "learning_rate": 0.0001857288007934555,
      "loss": 3.9313,
      "step": 76550
    },
    {
      "epoch": 0.07136630801352707,
      "grad_norm": 0.8385816216468811,
      "learning_rate": 0.0001857269364713603,
      "loss": 3.9237,
      "step": 76560
    },
    {
      "epoch": 0.07137562963160615,
      "grad_norm": 1.1464736461639404,
      "learning_rate": 0.00018572507214926508,
      "loss": 3.7688,
      "step": 76570
    },
    {
      "epoch": 0.07138495124968525,
      "grad_norm": 1.0201045274734497,
      "learning_rate": 0.00018572320782716989,
      "loss": 3.8132,
      "step": 76580
    },
    {
      "epoch": 0.07139427286776434,
      "grad_norm": 1.1362450122833252,
      "learning_rate": 0.0001857213435050747,
      "loss": 3.9569,
      "step": 76590
    },
    {
      "epoch": 0.07140359448584344,
      "grad_norm": 0.9844101071357727,
      "learning_rate": 0.0001857194791829795,
      "loss": 3.804,
      "step": 76600
    },
    {
      "epoch": 0.07141291610392253,
      "grad_norm": 1.2685880661010742,
      "learning_rate": 0.0001857176148608843,
      "loss": 3.9218,
      "step": 76610
    },
    {
      "epoch": 0.07142223772200161,
      "grad_norm": 1.0598722696304321,
      "learning_rate": 0.00018571575053878908,
      "loss": 3.8837,
      "step": 76620
    },
    {
      "epoch": 0.0714315593400807,
      "grad_norm": 1.146013855934143,
      "learning_rate": 0.0001857138862166939,
      "loss": 3.7809,
      "step": 76630
    },
    {
      "epoch": 0.0714408809581598,
      "grad_norm": 1.0221807956695557,
      "learning_rate": 0.0001857120218945987,
      "loss": 4.0244,
      "step": 76640
    },
    {
      "epoch": 0.0714502025762389,
      "grad_norm": 0.9125640988349915,
      "learning_rate": 0.0001857101575725035,
      "loss": 3.656,
      "step": 76650
    },
    {
      "epoch": 0.07145952419431799,
      "grad_norm": 0.92231285572052,
      "learning_rate": 0.0001857082932504083,
      "loss": 3.8671,
      "step": 76660
    },
    {
      "epoch": 0.07146884581239707,
      "grad_norm": 0.9520929455757141,
      "learning_rate": 0.0001857064289283131,
      "loss": 3.7854,
      "step": 76670
    },
    {
      "epoch": 0.07147816743047616,
      "grad_norm": 1.041765809059143,
      "learning_rate": 0.0001857045646062179,
      "loss": 3.8596,
      "step": 76680
    },
    {
      "epoch": 0.07148748904855526,
      "grad_norm": 0.9902685880661011,
      "learning_rate": 0.0001857027002841227,
      "loss": 3.7669,
      "step": 76690
    },
    {
      "epoch": 0.07149681066663435,
      "grad_norm": 1.04756760597229,
      "learning_rate": 0.0001857008359620275,
      "loss": 3.6694,
      "step": 76700
    },
    {
      "epoch": 0.07150613228471345,
      "grad_norm": 1.1736359596252441,
      "learning_rate": 0.0001856989716399323,
      "loss": 3.7118,
      "step": 76710
    },
    {
      "epoch": 0.07151545390279254,
      "grad_norm": 1.079831600189209,
      "learning_rate": 0.0001856971073178371,
      "loss": 4.0133,
      "step": 76720
    },
    {
      "epoch": 0.07152477552087162,
      "grad_norm": 0.9588664174079895,
      "learning_rate": 0.0001856952429957419,
      "loss": 3.8271,
      "step": 76730
    },
    {
      "epoch": 0.07153409713895072,
      "grad_norm": 1.0272845029830933,
      "learning_rate": 0.00018569337867364668,
      "loss": 3.9585,
      "step": 76740
    },
    {
      "epoch": 0.07154341875702981,
      "grad_norm": 1.0496994256973267,
      "learning_rate": 0.0001856915143515515,
      "loss": 3.9708,
      "step": 76750
    },
    {
      "epoch": 0.0715527403751089,
      "grad_norm": 1.0075984001159668,
      "learning_rate": 0.00018568965002945632,
      "loss": 3.767,
      "step": 76760
    },
    {
      "epoch": 0.071562061993188,
      "grad_norm": 1.0348905324935913,
      "learning_rate": 0.0001856877857073611,
      "loss": 3.9294,
      "step": 76770
    },
    {
      "epoch": 0.07157138361126708,
      "grad_norm": 1.1173995733261108,
      "learning_rate": 0.0001856859213852659,
      "loss": 3.7675,
      "step": 76780
    },
    {
      "epoch": 0.07158070522934618,
      "grad_norm": 0.9723829030990601,
      "learning_rate": 0.00018568405706317068,
      "loss": 3.7432,
      "step": 76790
    },
    {
      "epoch": 0.07159002684742527,
      "grad_norm": 1.123565435409546,
      "learning_rate": 0.0001856821927410755,
      "loss": 3.9698,
      "step": 76800
    },
    {
      "epoch": 0.07159934846550436,
      "grad_norm": 1.2279634475708008,
      "learning_rate": 0.00018568032841898032,
      "loss": 3.932,
      "step": 76810
    },
    {
      "epoch": 0.07160867008358346,
      "grad_norm": 1.0413960218429565,
      "learning_rate": 0.0001856784640968851,
      "loss": 3.8092,
      "step": 76820
    },
    {
      "epoch": 0.07161799170166254,
      "grad_norm": 1.2176697254180908,
      "learning_rate": 0.0001856765997747899,
      "loss": 3.8561,
      "step": 76830
    },
    {
      "epoch": 0.07162731331974163,
      "grad_norm": 1.2448382377624512,
      "learning_rate": 0.0001856747354526947,
      "loss": 3.7888,
      "step": 76840
    },
    {
      "epoch": 0.07163663493782073,
      "grad_norm": 1.0863885879516602,
      "learning_rate": 0.0001856728711305995,
      "loss": 3.8558,
      "step": 76850
    },
    {
      "epoch": 0.07164595655589982,
      "grad_norm": 1.082230567932129,
      "learning_rate": 0.0001856710068085043,
      "loss": 3.7722,
      "step": 76860
    },
    {
      "epoch": 0.07165527817397892,
      "grad_norm": 1.1041110754013062,
      "learning_rate": 0.0001856691424864091,
      "loss": 3.9137,
      "step": 76870
    },
    {
      "epoch": 0.071664599792058,
      "grad_norm": 1.0404237508773804,
      "learning_rate": 0.0001856672781643139,
      "loss": 3.8376,
      "step": 76880
    },
    {
      "epoch": 0.07167392141013709,
      "grad_norm": 1.2560477256774902,
      "learning_rate": 0.0001856654138422187,
      "loss": 3.7864,
      "step": 76890
    },
    {
      "epoch": 0.07168324302821619,
      "grad_norm": 1.1921807527542114,
      "learning_rate": 0.0001856635495201235,
      "loss": 3.9147,
      "step": 76900
    },
    {
      "epoch": 0.07169256464629528,
      "grad_norm": 1.1548552513122559,
      "learning_rate": 0.0001856616851980283,
      "loss": 3.8592,
      "step": 76910
    },
    {
      "epoch": 0.07170188626437438,
      "grad_norm": 0.9429508447647095,
      "learning_rate": 0.0001856598208759331,
      "loss": 3.8598,
      "step": 76920
    },
    {
      "epoch": 0.07171120788245347,
      "grad_norm": 0.9704718589782715,
      "learning_rate": 0.00018565795655383792,
      "loss": 3.8396,
      "step": 76930
    },
    {
      "epoch": 0.07172052950053255,
      "grad_norm": 1.2162816524505615,
      "learning_rate": 0.0001856560922317427,
      "loss": 3.7248,
      "step": 76940
    },
    {
      "epoch": 0.07172985111861165,
      "grad_norm": 0.9843660593032837,
      "learning_rate": 0.0001856542279096475,
      "loss": 3.934,
      "step": 76950
    },
    {
      "epoch": 0.07173917273669074,
      "grad_norm": 1.245490550994873,
      "learning_rate": 0.0001856523635875523,
      "loss": 3.8056,
      "step": 76960
    },
    {
      "epoch": 0.07174849435476983,
      "grad_norm": 0.984560489654541,
      "learning_rate": 0.0001856504992654571,
      "loss": 3.8473,
      "step": 76970
    },
    {
      "epoch": 0.07175781597284893,
      "grad_norm": 1.052415132522583,
      "learning_rate": 0.00018564863494336192,
      "loss": 3.9385,
      "step": 76980
    },
    {
      "epoch": 0.07176713759092801,
      "grad_norm": 1.1984339952468872,
      "learning_rate": 0.0001856467706212667,
      "loss": 3.948,
      "step": 76990
    },
    {
      "epoch": 0.0717764592090071,
      "grad_norm": 1.228907585144043,
      "learning_rate": 0.0001856449062991715,
      "loss": 3.7206,
      "step": 77000
    },
    {
      "epoch": 0.0717857808270862,
      "grad_norm": 1.1240179538726807,
      "learning_rate": 0.0001856430419770763,
      "loss": 3.7178,
      "step": 77010
    },
    {
      "epoch": 0.07179510244516529,
      "grad_norm": 1.040103793144226,
      "learning_rate": 0.0001856411776549811,
      "loss": 4.0524,
      "step": 77020
    },
    {
      "epoch": 0.07180442406324439,
      "grad_norm": 1.2270210981369019,
      "learning_rate": 0.0001856393133328859,
      "loss": 3.7852,
      "step": 77030
    },
    {
      "epoch": 0.07181374568132347,
      "grad_norm": 1.2517287731170654,
      "learning_rate": 0.0001856374490107907,
      "loss": 3.9352,
      "step": 77040
    },
    {
      "epoch": 0.07182306729940256,
      "grad_norm": 1.076924443244934,
      "learning_rate": 0.0001856355846886955,
      "loss": 3.8333,
      "step": 77050
    },
    {
      "epoch": 0.07183238891748166,
      "grad_norm": 1.0640051364898682,
      "learning_rate": 0.00018563372036660032,
      "loss": 3.8334,
      "step": 77060
    },
    {
      "epoch": 0.07184171053556075,
      "grad_norm": 1.0474212169647217,
      "learning_rate": 0.0001856318560445051,
      "loss": 3.7312,
      "step": 77070
    },
    {
      "epoch": 0.07185103215363985,
      "grad_norm": 1.1107447147369385,
      "learning_rate": 0.0001856299917224099,
      "loss": 3.879,
      "step": 77080
    },
    {
      "epoch": 0.07186035377171893,
      "grad_norm": 1.1298840045928955,
      "learning_rate": 0.0001856281274003147,
      "loss": 3.8105,
      "step": 77090
    },
    {
      "epoch": 0.07186967538979802,
      "grad_norm": 1.008554220199585,
      "learning_rate": 0.00018562626307821952,
      "loss": 3.8832,
      "step": 77100
    },
    {
      "epoch": 0.07187899700787712,
      "grad_norm": 1.0245630741119385,
      "learning_rate": 0.00018562439875612432,
      "loss": 3.7421,
      "step": 77110
    },
    {
      "epoch": 0.07188831862595621,
      "grad_norm": 0.9907963871955872,
      "learning_rate": 0.0001856225344340291,
      "loss": 3.9389,
      "step": 77120
    },
    {
      "epoch": 0.0718976402440353,
      "grad_norm": 1.1607747077941895,
      "learning_rate": 0.0001856206701119339,
      "loss": 3.8862,
      "step": 77130
    },
    {
      "epoch": 0.0719069618621144,
      "grad_norm": 1.058404803276062,
      "learning_rate": 0.00018561880578983871,
      "loss": 3.6564,
      "step": 77140
    },
    {
      "epoch": 0.07191628348019348,
      "grad_norm": 1.1822987794876099,
      "learning_rate": 0.00018561694146774352,
      "loss": 3.7937,
      "step": 77150
    },
    {
      "epoch": 0.07192560509827257,
      "grad_norm": 1.1656818389892578,
      "learning_rate": 0.0001856150771456483,
      "loss": 3.8745,
      "step": 77160
    },
    {
      "epoch": 0.07193492671635167,
      "grad_norm": 1.0856913328170776,
      "learning_rate": 0.0001856132128235531,
      "loss": 3.9419,
      "step": 77170
    },
    {
      "epoch": 0.07194424833443076,
      "grad_norm": 1.109269380569458,
      "learning_rate": 0.0001856113485014579,
      "loss": 3.638,
      "step": 77180
    },
    {
      "epoch": 0.07195356995250986,
      "grad_norm": 1.1760554313659668,
      "learning_rate": 0.0001856094841793627,
      "loss": 3.7354,
      "step": 77190
    },
    {
      "epoch": 0.07196289157058894,
      "grad_norm": 0.9655463099479675,
      "learning_rate": 0.00018560761985726752,
      "loss": 3.8197,
      "step": 77200
    },
    {
      "epoch": 0.07197221318866803,
      "grad_norm": 1.1499770879745483,
      "learning_rate": 0.0001856057555351723,
      "loss": 3.7589,
      "step": 77210
    },
    {
      "epoch": 0.07198153480674713,
      "grad_norm": 1.2184933423995972,
      "learning_rate": 0.0001856038912130771,
      "loss": 3.5991,
      "step": 77220
    },
    {
      "epoch": 0.07199085642482622,
      "grad_norm": 1.2875512838363647,
      "learning_rate": 0.00018560202689098192,
      "loss": 3.8074,
      "step": 77230
    },
    {
      "epoch": 0.07200017804290532,
      "grad_norm": 1.233843207359314,
      "learning_rate": 0.0001856001625688867,
      "loss": 3.8133,
      "step": 77240
    },
    {
      "epoch": 0.0720094996609844,
      "grad_norm": 1.087356448173523,
      "learning_rate": 0.0001855982982467915,
      "loss": 3.6858,
      "step": 77250
    },
    {
      "epoch": 0.07201882127906349,
      "grad_norm": 0.9448338747024536,
      "learning_rate": 0.0001855964339246963,
      "loss": 3.8231,
      "step": 77260
    },
    {
      "epoch": 0.07202814289714259,
      "grad_norm": 1.079302191734314,
      "learning_rate": 0.00018559456960260111,
      "loss": 3.6768,
      "step": 77270
    },
    {
      "epoch": 0.07203746451522168,
      "grad_norm": 0.9502652287483215,
      "learning_rate": 0.00018559270528050592,
      "loss": 3.8639,
      "step": 77280
    },
    {
      "epoch": 0.07204678613330077,
      "grad_norm": 1.1358791589736938,
      "learning_rate": 0.0001855908409584107,
      "loss": 3.7091,
      "step": 77290
    },
    {
      "epoch": 0.07205610775137986,
      "grad_norm": 1.1355808973312378,
      "learning_rate": 0.0001855889766363155,
      "loss": 3.9507,
      "step": 77300
    },
    {
      "epoch": 0.07206542936945895,
      "grad_norm": 1.0794283151626587,
      "learning_rate": 0.0001855871123142203,
      "loss": 3.9717,
      "step": 77310
    },
    {
      "epoch": 0.07207475098753804,
      "grad_norm": 1.0579407215118408,
      "learning_rate": 0.00018558524799212512,
      "loss": 3.9354,
      "step": 77320
    },
    {
      "epoch": 0.07208407260561714,
      "grad_norm": 1.1358540058135986,
      "learning_rate": 0.00018558338367002993,
      "loss": 3.8078,
      "step": 77330
    },
    {
      "epoch": 0.07209339422369623,
      "grad_norm": 0.9657616019248962,
      "learning_rate": 0.0001855815193479347,
      "loss": 4.0498,
      "step": 77340
    },
    {
      "epoch": 0.07210271584177531,
      "grad_norm": 1.1862961053848267,
      "learning_rate": 0.0001855796550258395,
      "loss": 3.789,
      "step": 77350
    },
    {
      "epoch": 0.07211203745985441,
      "grad_norm": 1.00790536403656,
      "learning_rate": 0.0001855777907037443,
      "loss": 3.9452,
      "step": 77360
    },
    {
      "epoch": 0.0721213590779335,
      "grad_norm": 1.0261510610580444,
      "learning_rate": 0.00018557592638164912,
      "loss": 3.6837,
      "step": 77370
    },
    {
      "epoch": 0.0721306806960126,
      "grad_norm": 0.9098105430603027,
      "learning_rate": 0.00018557406205955393,
      "loss": 3.8964,
      "step": 77380
    },
    {
      "epoch": 0.07214000231409169,
      "grad_norm": 1.109384536743164,
      "learning_rate": 0.0001855721977374587,
      "loss": 3.7327,
      "step": 77390
    },
    {
      "epoch": 0.07214932393217079,
      "grad_norm": 0.9262708425521851,
      "learning_rate": 0.00018557033341536352,
      "loss": 3.7296,
      "step": 77400
    },
    {
      "epoch": 0.07215864555024987,
      "grad_norm": 1.0200828313827515,
      "learning_rate": 0.0001855684690932683,
      "loss": 4.0221,
      "step": 77410
    },
    {
      "epoch": 0.07216796716832896,
      "grad_norm": 0.9319530725479126,
      "learning_rate": 0.0001855666047711731,
      "loss": 4.1405,
      "step": 77420
    },
    {
      "epoch": 0.07217728878640806,
      "grad_norm": 1.0383070707321167,
      "learning_rate": 0.00018556474044907793,
      "loss": 3.8175,
      "step": 77430
    },
    {
      "epoch": 0.07218661040448715,
      "grad_norm": 1.067154884338379,
      "learning_rate": 0.0001855628761269827,
      "loss": 3.8601,
      "step": 77440
    },
    {
      "epoch": 0.07219593202256624,
      "grad_norm": 1.0838277339935303,
      "learning_rate": 0.00018556101180488752,
      "loss": 3.7637,
      "step": 77450
    },
    {
      "epoch": 0.07220525364064533,
      "grad_norm": 1.224027395248413,
      "learning_rate": 0.0001855591474827923,
      "loss": 3.8372,
      "step": 77460
    },
    {
      "epoch": 0.07221457525872442,
      "grad_norm": 1.0399123430252075,
      "learning_rate": 0.0001855572831606971,
      "loss": 3.8911,
      "step": 77470
    },
    {
      "epoch": 0.07222389687680351,
      "grad_norm": 1.1687555313110352,
      "learning_rate": 0.0001855554188386019,
      "loss": 3.8013,
      "step": 77480
    },
    {
      "epoch": 0.07223321849488261,
      "grad_norm": 1.1361477375030518,
      "learning_rate": 0.00018555355451650672,
      "loss": 4.0694,
      "step": 77490
    },
    {
      "epoch": 0.0722425401129617,
      "grad_norm": 0.9750786423683167,
      "learning_rate": 0.00018555169019441152,
      "loss": 3.8369,
      "step": 77500
    },
    {
      "epoch": 0.07225186173104078,
      "grad_norm": 1.1131843328475952,
      "learning_rate": 0.0001855498258723163,
      "loss": 3.8618,
      "step": 77510
    },
    {
      "epoch": 0.07226118334911988,
      "grad_norm": 1.156385898590088,
      "learning_rate": 0.0001855479615502211,
      "loss": 3.6304,
      "step": 77520
    },
    {
      "epoch": 0.07227050496719897,
      "grad_norm": 1.1671762466430664,
      "learning_rate": 0.00018554609722812592,
      "loss": 3.8546,
      "step": 77530
    },
    {
      "epoch": 0.07227982658527807,
      "grad_norm": 0.9951595067977905,
      "learning_rate": 0.00018554423290603072,
      "loss": 3.9708,
      "step": 77540
    },
    {
      "epoch": 0.07228914820335716,
      "grad_norm": 1.1599113941192627,
      "learning_rate": 0.00018554236858393553,
      "loss": 3.7809,
      "step": 77550
    },
    {
      "epoch": 0.07229846982143624,
      "grad_norm": 1.1937888860702515,
      "learning_rate": 0.0001855405042618403,
      "loss": 3.8232,
      "step": 77560
    },
    {
      "epoch": 0.07230779143951534,
      "grad_norm": 1.078673243522644,
      "learning_rate": 0.00018553863993974511,
      "loss": 3.8167,
      "step": 77570
    },
    {
      "epoch": 0.07231711305759443,
      "grad_norm": 1.5523794889450073,
      "learning_rate": 0.00018553677561764992,
      "loss": 3.7491,
      "step": 77580
    },
    {
      "epoch": 0.07232643467567353,
      "grad_norm": 1.1000157594680786,
      "learning_rate": 0.00018553491129555473,
      "loss": 3.5716,
      "step": 77590
    },
    {
      "epoch": 0.07233575629375262,
      "grad_norm": 1.1294492483139038,
      "learning_rate": 0.00018553304697345953,
      "loss": 3.9775,
      "step": 77600
    },
    {
      "epoch": 0.07234507791183171,
      "grad_norm": 0.9013527035713196,
      "learning_rate": 0.0001855311826513643,
      "loss": 4.0349,
      "step": 77610
    },
    {
      "epoch": 0.0723543995299108,
      "grad_norm": 1.3387014865875244,
      "learning_rate": 0.00018552931832926912,
      "loss": 3.8241,
      "step": 77620
    },
    {
      "epoch": 0.07236372114798989,
      "grad_norm": 1.1411209106445312,
      "learning_rate": 0.00018552745400717392,
      "loss": 3.7033,
      "step": 77630
    },
    {
      "epoch": 0.07237304276606898,
      "grad_norm": 1.014276146888733,
      "learning_rate": 0.0001855255896850787,
      "loss": 3.6675,
      "step": 77640
    },
    {
      "epoch": 0.07238236438414808,
      "grad_norm": 0.8564738035202026,
      "learning_rate": 0.00018552372536298354,
      "loss": 3.9052,
      "step": 77650
    },
    {
      "epoch": 0.07239168600222717,
      "grad_norm": 1.0868384838104248,
      "learning_rate": 0.00018552186104088832,
      "loss": 3.7313,
      "step": 77660
    },
    {
      "epoch": 0.07240100762030625,
      "grad_norm": 1.1175546646118164,
      "learning_rate": 0.00018551999671879312,
      "loss": 3.6308,
      "step": 77670
    },
    {
      "epoch": 0.07241032923838535,
      "grad_norm": 1.1657201051712036,
      "learning_rate": 0.00018551813239669793,
      "loss": 4.0067,
      "step": 77680
    },
    {
      "epoch": 0.07241965085646444,
      "grad_norm": 1.0188416242599487,
      "learning_rate": 0.0001855162680746027,
      "loss": 4.0388,
      "step": 77690
    },
    {
      "epoch": 0.07242897247454354,
      "grad_norm": 1.1500192880630493,
      "learning_rate": 0.00018551440375250751,
      "loss": 3.806,
      "step": 77700
    },
    {
      "epoch": 0.07243829409262263,
      "grad_norm": 1.1787482500076294,
      "learning_rate": 0.00018551253943041232,
      "loss": 4.0053,
      "step": 77710
    },
    {
      "epoch": 0.07244761571070171,
      "grad_norm": 1.26982581615448,
      "learning_rate": 0.00018551067510831713,
      "loss": 3.6681,
      "step": 77720
    },
    {
      "epoch": 0.07245693732878081,
      "grad_norm": 0.9928643107414246,
      "learning_rate": 0.00018550881078622193,
      "loss": 3.7485,
      "step": 77730
    },
    {
      "epoch": 0.0724662589468599,
      "grad_norm": 1.1550394296646118,
      "learning_rate": 0.0001855069464641267,
      "loss": 3.7899,
      "step": 77740
    },
    {
      "epoch": 0.072475580564939,
      "grad_norm": 1.2854005098342896,
      "learning_rate": 0.00018550508214203152,
      "loss": 3.8152,
      "step": 77750
    },
    {
      "epoch": 0.07248490218301809,
      "grad_norm": 1.0827692747116089,
      "learning_rate": 0.00018550321781993633,
      "loss": 3.7753,
      "step": 77760
    },
    {
      "epoch": 0.07249422380109717,
      "grad_norm": 1.1068161725997925,
      "learning_rate": 0.00018550135349784113,
      "loss": 3.9484,
      "step": 77770
    },
    {
      "epoch": 0.07250354541917627,
      "grad_norm": 0.9929633736610413,
      "learning_rate": 0.00018549948917574594,
      "loss": 3.8487,
      "step": 77780
    },
    {
      "epoch": 0.07251286703725536,
      "grad_norm": 1.1797031164169312,
      "learning_rate": 0.00018549762485365072,
      "loss": 3.8477,
      "step": 77790
    },
    {
      "epoch": 0.07252218865533445,
      "grad_norm": 1.0532326698303223,
      "learning_rate": 0.00018549576053155552,
      "loss": 4.0286,
      "step": 77800
    },
    {
      "epoch": 0.07253151027341355,
      "grad_norm": 0.8915303349494934,
      "learning_rate": 0.0001854938962094603,
      "loss": 3.8812,
      "step": 77810
    },
    {
      "epoch": 0.07254083189149264,
      "grad_norm": 1.2798585891723633,
      "learning_rate": 0.00018549203188736514,
      "loss": 3.8784,
      "step": 77820
    },
    {
      "epoch": 0.07255015350957172,
      "grad_norm": 0.9361816644668579,
      "learning_rate": 0.00018549016756526994,
      "loss": 3.9777,
      "step": 77830
    },
    {
      "epoch": 0.07255947512765082,
      "grad_norm": 1.1087063550949097,
      "learning_rate": 0.00018548830324317472,
      "loss": 3.7324,
      "step": 77840
    },
    {
      "epoch": 0.07256879674572991,
      "grad_norm": 0.9792326092720032,
      "learning_rate": 0.00018548643892107953,
      "loss": 3.8394,
      "step": 77850
    },
    {
      "epoch": 0.07257811836380901,
      "grad_norm": 1.1155027151107788,
      "learning_rate": 0.0001854845745989843,
      "loss": 3.8263,
      "step": 77860
    },
    {
      "epoch": 0.0725874399818881,
      "grad_norm": 1.0686458349227905,
      "learning_rate": 0.0001854827102768891,
      "loss": 3.9123,
      "step": 77870
    },
    {
      "epoch": 0.07259676159996718,
      "grad_norm": 1.2353484630584717,
      "learning_rate": 0.00018548084595479395,
      "loss": 3.6095,
      "step": 77880
    },
    {
      "epoch": 0.07260608321804628,
      "grad_norm": 1.094574213027954,
      "learning_rate": 0.00018547898163269873,
      "loss": 3.8562,
      "step": 77890
    },
    {
      "epoch": 0.07261540483612537,
      "grad_norm": 1.085801124572754,
      "learning_rate": 0.00018547711731060353,
      "loss": 3.8817,
      "step": 77900
    },
    {
      "epoch": 0.07262472645420447,
      "grad_norm": 1.0781190395355225,
      "learning_rate": 0.0001854752529885083,
      "loss": 3.848,
      "step": 77910
    },
    {
      "epoch": 0.07263404807228356,
      "grad_norm": 1.177941918373108,
      "learning_rate": 0.00018547338866641312,
      "loss": 3.7126,
      "step": 77920
    },
    {
      "epoch": 0.07264336969036264,
      "grad_norm": 0.974977433681488,
      "learning_rate": 0.00018547152434431792,
      "loss": 3.8253,
      "step": 77930
    },
    {
      "epoch": 0.07265269130844174,
      "grad_norm": 1.3189715147018433,
      "learning_rate": 0.00018546966002222273,
      "loss": 3.9256,
      "step": 77940
    },
    {
      "epoch": 0.07266201292652083,
      "grad_norm": 1.2934608459472656,
      "learning_rate": 0.00018546779570012754,
      "loss": 3.6446,
      "step": 77950
    },
    {
      "epoch": 0.07267133454459992,
      "grad_norm": 1.0619083642959595,
      "learning_rate": 0.00018546593137803232,
      "loss": 3.8338,
      "step": 77960
    },
    {
      "epoch": 0.07268065616267902,
      "grad_norm": 1.3278754949569702,
      "learning_rate": 0.00018546406705593712,
      "loss": 3.6722,
      "step": 77970
    },
    {
      "epoch": 0.0726899777807581,
      "grad_norm": 1.157171607017517,
      "learning_rate": 0.00018546220273384193,
      "loss": 3.8761,
      "step": 77980
    },
    {
      "epoch": 0.0726992993988372,
      "grad_norm": 1.054449200630188,
      "learning_rate": 0.00018546033841174673,
      "loss": 3.826,
      "step": 77990
    },
    {
      "epoch": 0.07270862101691629,
      "grad_norm": 1.3423575162887573,
      "learning_rate": 0.00018545847408965154,
      "loss": 3.805,
      "step": 78000
    },
    {
      "epoch": 0.07271794263499538,
      "grad_norm": 1.0577232837677002,
      "learning_rate": 0.00018545660976755632,
      "loss": 3.748,
      "step": 78010
    },
    {
      "epoch": 0.07272726425307448,
      "grad_norm": 1.0539405345916748,
      "learning_rate": 0.00018545474544546113,
      "loss": 4.0614,
      "step": 78020
    },
    {
      "epoch": 0.07273658587115356,
      "grad_norm": 1.0627669095993042,
      "learning_rate": 0.00018545288112336593,
      "loss": 3.7895,
      "step": 78030
    },
    {
      "epoch": 0.07274590748923265,
      "grad_norm": 0.9880484938621521,
      "learning_rate": 0.00018545101680127074,
      "loss": 3.8365,
      "step": 78040
    },
    {
      "epoch": 0.07275522910731175,
      "grad_norm": 1.0228451490402222,
      "learning_rate": 0.00018544915247917555,
      "loss": 4.011,
      "step": 78050
    },
    {
      "epoch": 0.07276455072539084,
      "grad_norm": 1.0602518320083618,
      "learning_rate": 0.00018544728815708032,
      "loss": 4.001,
      "step": 78060
    },
    {
      "epoch": 0.07277387234346994,
      "grad_norm": 1.0248878002166748,
      "learning_rate": 0.00018544542383498513,
      "loss": 3.9966,
      "step": 78070
    },
    {
      "epoch": 0.07278319396154903,
      "grad_norm": 0.8892022967338562,
      "learning_rate": 0.00018544355951288994,
      "loss": 3.8151,
      "step": 78080
    },
    {
      "epoch": 0.07279251557962811,
      "grad_norm": 1.050694227218628,
      "learning_rate": 0.00018544169519079472,
      "loss": 3.6154,
      "step": 78090
    },
    {
      "epoch": 0.0728018371977072,
      "grad_norm": 1.262674331665039,
      "learning_rate": 0.00018543983086869955,
      "loss": 3.8979,
      "step": 78100
    },
    {
      "epoch": 0.0728111588157863,
      "grad_norm": 1.1030241250991821,
      "learning_rate": 0.00018543796654660433,
      "loss": 3.9148,
      "step": 78110
    },
    {
      "epoch": 0.0728204804338654,
      "grad_norm": 1.265958547592163,
      "learning_rate": 0.00018543610222450914,
      "loss": 3.826,
      "step": 78120
    },
    {
      "epoch": 0.07282980205194449,
      "grad_norm": 1.3630318641662598,
      "learning_rate": 0.00018543423790241394,
      "loss": 4.0162,
      "step": 78130
    },
    {
      "epoch": 0.07283912367002357,
      "grad_norm": 1.1883052587509155,
      "learning_rate": 0.00018543237358031872,
      "loss": 3.7798,
      "step": 78140
    },
    {
      "epoch": 0.07284844528810266,
      "grad_norm": 1.1165944337844849,
      "learning_rate": 0.00018543050925822353,
      "loss": 3.8279,
      "step": 78150
    },
    {
      "epoch": 0.07285776690618176,
      "grad_norm": 1.1167912483215332,
      "learning_rate": 0.00018542864493612833,
      "loss": 4.069,
      "step": 78160
    },
    {
      "epoch": 0.07286708852426085,
      "grad_norm": 1.0464706420898438,
      "learning_rate": 0.00018542678061403314,
      "loss": 3.8284,
      "step": 78170
    },
    {
      "epoch": 0.07287641014233995,
      "grad_norm": 1.0436896085739136,
      "learning_rate": 0.00018542491629193795,
      "loss": 3.7809,
      "step": 78180
    },
    {
      "epoch": 0.07288573176041903,
      "grad_norm": 1.1300333738327026,
      "learning_rate": 0.00018542305196984273,
      "loss": 3.7939,
      "step": 78190
    },
    {
      "epoch": 0.07289505337849812,
      "grad_norm": 1.1070756912231445,
      "learning_rate": 0.00018542118764774753,
      "loss": 3.844,
      "step": 78200
    },
    {
      "epoch": 0.07290437499657722,
      "grad_norm": 1.2190059423446655,
      "learning_rate": 0.00018541932332565234,
      "loss": 3.7286,
      "step": 78210
    },
    {
      "epoch": 0.07291369661465631,
      "grad_norm": 0.9421396851539612,
      "learning_rate": 0.00018541745900355714,
      "loss": 3.8431,
      "step": 78220
    },
    {
      "epoch": 0.0729230182327354,
      "grad_norm": 1.257863998413086,
      "learning_rate": 0.00018541559468146195,
      "loss": 3.9384,
      "step": 78230
    },
    {
      "epoch": 0.07293233985081449,
      "grad_norm": 0.9447577595710754,
      "learning_rate": 0.00018541373035936673,
      "loss": 3.9728,
      "step": 78240
    },
    {
      "epoch": 0.07294166146889358,
      "grad_norm": 1.1415081024169922,
      "learning_rate": 0.00018541186603727154,
      "loss": 3.8068,
      "step": 78250
    },
    {
      "epoch": 0.07295098308697268,
      "grad_norm": 0.9552024602890015,
      "learning_rate": 0.00018541000171517632,
      "loss": 3.9031,
      "step": 78260
    },
    {
      "epoch": 0.07296030470505177,
      "grad_norm": 0.9842380285263062,
      "learning_rate": 0.00018540813739308115,
      "loss": 3.8197,
      "step": 78270
    },
    {
      "epoch": 0.07296962632313087,
      "grad_norm": 1.2560275793075562,
      "learning_rate": 0.00018540627307098596,
      "loss": 3.6245,
      "step": 78280
    },
    {
      "epoch": 0.07297894794120996,
      "grad_norm": 1.0620583295822144,
      "learning_rate": 0.00018540440874889073,
      "loss": 3.7913,
      "step": 78290
    },
    {
      "epoch": 0.07298826955928904,
      "grad_norm": 1.1870158910751343,
      "learning_rate": 0.00018540254442679554,
      "loss": 3.6993,
      "step": 78300
    },
    {
      "epoch": 0.07299759117736813,
      "grad_norm": 1.2022619247436523,
      "learning_rate": 0.00018540068010470032,
      "loss": 3.8204,
      "step": 78310
    },
    {
      "epoch": 0.07300691279544723,
      "grad_norm": 1.1837748289108276,
      "learning_rate": 0.00018539881578260513,
      "loss": 3.7434,
      "step": 78320
    },
    {
      "epoch": 0.07301623441352632,
      "grad_norm": 1.1570937633514404,
      "learning_rate": 0.00018539695146050996,
      "loss": 3.8264,
      "step": 78330
    },
    {
      "epoch": 0.07302555603160542,
      "grad_norm": 1.0159188508987427,
      "learning_rate": 0.00018539508713841474,
      "loss": 3.8838,
      "step": 78340
    },
    {
      "epoch": 0.0730348776496845,
      "grad_norm": 1.1542296409606934,
      "learning_rate": 0.00018539322281631955,
      "loss": 3.7861,
      "step": 78350
    },
    {
      "epoch": 0.0730441992677636,
      "grad_norm": 0.9909271597862244,
      "learning_rate": 0.00018539135849422432,
      "loss": 3.806,
      "step": 78360
    },
    {
      "epoch": 0.07305352088584269,
      "grad_norm": 1.0882200002670288,
      "learning_rate": 0.00018538949417212913,
      "loss": 3.8255,
      "step": 78370
    },
    {
      "epoch": 0.07306284250392178,
      "grad_norm": 1.2207289934158325,
      "learning_rate": 0.00018538762985003394,
      "loss": 4.0659,
      "step": 78380
    },
    {
      "epoch": 0.07307216412200088,
      "grad_norm": 1.0460460186004639,
      "learning_rate": 0.00018538576552793874,
      "loss": 3.7364,
      "step": 78390
    },
    {
      "epoch": 0.07308148574007996,
      "grad_norm": 1.3062236309051514,
      "learning_rate": 0.00018538390120584355,
      "loss": 3.808,
      "step": 78400
    },
    {
      "epoch": 0.07309080735815905,
      "grad_norm": 1.0045397281646729,
      "learning_rate": 0.00018538203688374833,
      "loss": 3.8457,
      "step": 78410
    },
    {
      "epoch": 0.07310012897623815,
      "grad_norm": 1.1284065246582031,
      "learning_rate": 0.00018538017256165314,
      "loss": 3.8542,
      "step": 78420
    },
    {
      "epoch": 0.07310945059431724,
      "grad_norm": 1.3530380725860596,
      "learning_rate": 0.00018537830823955794,
      "loss": 3.7568,
      "step": 78430
    },
    {
      "epoch": 0.07311877221239634,
      "grad_norm": 0.9951221942901611,
      "learning_rate": 0.00018537644391746275,
      "loss": 3.4811,
      "step": 78440
    },
    {
      "epoch": 0.07312809383047542,
      "grad_norm": 1.213641881942749,
      "learning_rate": 0.00018537457959536755,
      "loss": 3.6661,
      "step": 78450
    },
    {
      "epoch": 0.07313741544855451,
      "grad_norm": 0.9556347727775574,
      "learning_rate": 0.00018537271527327233,
      "loss": 3.8255,
      "step": 78460
    },
    {
      "epoch": 0.0731467370666336,
      "grad_norm": 1.5598087310791016,
      "learning_rate": 0.00018537085095117714,
      "loss": 3.9752,
      "step": 78470
    },
    {
      "epoch": 0.0731560586847127,
      "grad_norm": 1.1160281896591187,
      "learning_rate": 0.00018536898662908192,
      "loss": 3.8094,
      "step": 78480
    },
    {
      "epoch": 0.0731653803027918,
      "grad_norm": 0.9484207630157471,
      "learning_rate": 0.00018536712230698675,
      "loss": 3.9414,
      "step": 78490
    },
    {
      "epoch": 0.07317470192087089,
      "grad_norm": 1.3659909963607788,
      "learning_rate": 0.00018536525798489156,
      "loss": 3.7517,
      "step": 78500
    },
    {
      "epoch": 0.07318402353894997,
      "grad_norm": 0.9419592618942261,
      "learning_rate": 0.00018536339366279634,
      "loss": 3.8971,
      "step": 78510
    },
    {
      "epoch": 0.07319334515702906,
      "grad_norm": 0.9435970187187195,
      "learning_rate": 0.00018536152934070114,
      "loss": 3.7525,
      "step": 78520
    },
    {
      "epoch": 0.07320266677510816,
      "grad_norm": 1.0158039331436157,
      "learning_rate": 0.00018535966501860592,
      "loss": 3.866,
      "step": 78530
    },
    {
      "epoch": 0.07321198839318725,
      "grad_norm": 1.0003432035446167,
      "learning_rate": 0.00018535780069651073,
      "loss": 3.705,
      "step": 78540
    },
    {
      "epoch": 0.07322131001126635,
      "grad_norm": 1.1230584383010864,
      "learning_rate": 0.00018535593637441556,
      "loss": 3.9371,
      "step": 78550
    },
    {
      "epoch": 0.07323063162934543,
      "grad_norm": 1.2995814085006714,
      "learning_rate": 0.00018535407205232034,
      "loss": 3.8027,
      "step": 78560
    },
    {
      "epoch": 0.07323995324742452,
      "grad_norm": 0.9267760515213013,
      "learning_rate": 0.00018535220773022515,
      "loss": 3.8776,
      "step": 78570
    },
    {
      "epoch": 0.07324927486550362,
      "grad_norm": 1.0399836301803589,
      "learning_rate": 0.00018535034340812993,
      "loss": 4.135,
      "step": 78580
    },
    {
      "epoch": 0.07325859648358271,
      "grad_norm": 1.1114096641540527,
      "learning_rate": 0.00018534847908603473,
      "loss": 3.7659,
      "step": 78590
    },
    {
      "epoch": 0.0732679181016618,
      "grad_norm": 1.1891189813613892,
      "learning_rate": 0.00018534661476393954,
      "loss": 4.009,
      "step": 78600
    },
    {
      "epoch": 0.07327723971974089,
      "grad_norm": 0.9205805063247681,
      "learning_rate": 0.00018534475044184435,
      "loss": 3.8335,
      "step": 78610
    },
    {
      "epoch": 0.07328656133781998,
      "grad_norm": 0.993667721748352,
      "learning_rate": 0.00018534288611974915,
      "loss": 3.9991,
      "step": 78620
    },
    {
      "epoch": 0.07329588295589907,
      "grad_norm": 1.1311265230178833,
      "learning_rate": 0.00018534102179765393,
      "loss": 3.8696,
      "step": 78630
    },
    {
      "epoch": 0.07330520457397817,
      "grad_norm": 1.0626771450042725,
      "learning_rate": 0.00018533915747555874,
      "loss": 4.0054,
      "step": 78640
    },
    {
      "epoch": 0.07331452619205726,
      "grad_norm": 0.9953268766403198,
      "learning_rate": 0.00018533729315346354,
      "loss": 3.916,
      "step": 78650
    },
    {
      "epoch": 0.07332384781013634,
      "grad_norm": 1.3170369863510132,
      "learning_rate": 0.00018533542883136835,
      "loss": 3.6728,
      "step": 78660
    },
    {
      "epoch": 0.07333316942821544,
      "grad_norm": 1.1354296207427979,
      "learning_rate": 0.00018533356450927316,
      "loss": 3.7467,
      "step": 78670
    },
    {
      "epoch": 0.07334249104629453,
      "grad_norm": 1.1656030416488647,
      "learning_rate": 0.00018533170018717794,
      "loss": 3.8193,
      "step": 78680
    },
    {
      "epoch": 0.07335181266437363,
      "grad_norm": 0.9208136796951294,
      "learning_rate": 0.00018532983586508274,
      "loss": 3.8912,
      "step": 78690
    },
    {
      "epoch": 0.07336113428245272,
      "grad_norm": 1.026780366897583,
      "learning_rate": 0.00018532797154298755,
      "loss": 3.7749,
      "step": 78700
    },
    {
      "epoch": 0.07337045590053182,
      "grad_norm": 1.1891099214553833,
      "learning_rate": 0.00018532610722089236,
      "loss": 3.9741,
      "step": 78710
    },
    {
      "epoch": 0.0733797775186109,
      "grad_norm": 1.027515172958374,
      "learning_rate": 0.00018532424289879716,
      "loss": 3.924,
      "step": 78720
    },
    {
      "epoch": 0.07338909913668999,
      "grad_norm": 0.8953844308853149,
      "learning_rate": 0.00018532237857670194,
      "loss": 3.9069,
      "step": 78730
    },
    {
      "epoch": 0.07339842075476909,
      "grad_norm": 1.2197057008743286,
      "learning_rate": 0.00018532051425460675,
      "loss": 3.8226,
      "step": 78740
    },
    {
      "epoch": 0.07340774237284818,
      "grad_norm": 0.9293171167373657,
      "learning_rate": 0.00018531864993251155,
      "loss": 3.8591,
      "step": 78750
    },
    {
      "epoch": 0.07341706399092728,
      "grad_norm": 0.9070459604263306,
      "learning_rate": 0.00018531678561041633,
      "loss": 3.8945,
      "step": 78760
    },
    {
      "epoch": 0.07342638560900636,
      "grad_norm": 1.0244243144989014,
      "learning_rate": 0.00018531492128832117,
      "loss": 3.8096,
      "step": 78770
    },
    {
      "epoch": 0.07343570722708545,
      "grad_norm": 1.1040756702423096,
      "learning_rate": 0.00018531305696622595,
      "loss": 3.9747,
      "step": 78780
    },
    {
      "epoch": 0.07344502884516454,
      "grad_norm": 1.0806077718734741,
      "learning_rate": 0.00018531119264413075,
      "loss": 4.112,
      "step": 78790
    },
    {
      "epoch": 0.07345435046324364,
      "grad_norm": 1.1810085773468018,
      "learning_rate": 0.00018530932832203556,
      "loss": 3.756,
      "step": 78800
    },
    {
      "epoch": 0.07346367208132273,
      "grad_norm": 1.0795413255691528,
      "learning_rate": 0.00018530746399994034,
      "loss": 3.8914,
      "step": 78810
    },
    {
      "epoch": 0.07347299369940181,
      "grad_norm": 1.1788206100463867,
      "learning_rate": 0.00018530559967784514,
      "loss": 3.9134,
      "step": 78820
    },
    {
      "epoch": 0.07348231531748091,
      "grad_norm": 1.0479854345321655,
      "learning_rate": 0.00018530373535574995,
      "loss": 3.8911,
      "step": 78830
    },
    {
      "epoch": 0.07349163693556,
      "grad_norm": 1.2020065784454346,
      "learning_rate": 0.00018530187103365476,
      "loss": 4.0721,
      "step": 78840
    },
    {
      "epoch": 0.0735009585536391,
      "grad_norm": 1.1819288730621338,
      "learning_rate": 0.00018530000671155956,
      "loss": 3.7211,
      "step": 78850
    },
    {
      "epoch": 0.07351028017171819,
      "grad_norm": 1.1231285333633423,
      "learning_rate": 0.00018529814238946434,
      "loss": 3.7525,
      "step": 78860
    },
    {
      "epoch": 0.07351960178979727,
      "grad_norm": 0.8900389075279236,
      "learning_rate": 0.00018529627806736915,
      "loss": 3.701,
      "step": 78870
    },
    {
      "epoch": 0.07352892340787637,
      "grad_norm": 1.0231459140777588,
      "learning_rate": 0.00018529441374527395,
      "loss": 3.7768,
      "step": 78880
    },
    {
      "epoch": 0.07353824502595546,
      "grad_norm": 1.1110515594482422,
      "learning_rate": 0.00018529254942317876,
      "loss": 3.8947,
      "step": 78890
    },
    {
      "epoch": 0.07354756664403456,
      "grad_norm": 1.290038824081421,
      "learning_rate": 0.00018529068510108357,
      "loss": 3.8039,
      "step": 78900
    },
    {
      "epoch": 0.07355688826211365,
      "grad_norm": 1.01752507686615,
      "learning_rate": 0.00018528882077898835,
      "loss": 3.8601,
      "step": 78910
    },
    {
      "epoch": 0.07356620988019273,
      "grad_norm": 1.062963843345642,
      "learning_rate": 0.00018528695645689315,
      "loss": 3.636,
      "step": 78920
    },
    {
      "epoch": 0.07357553149827183,
      "grad_norm": 0.9743356108665466,
      "learning_rate": 0.00018528509213479793,
      "loss": 3.8132,
      "step": 78930
    },
    {
      "epoch": 0.07358485311635092,
      "grad_norm": 1.0480504035949707,
      "learning_rate": 0.00018528322781270277,
      "loss": 3.686,
      "step": 78940
    },
    {
      "epoch": 0.07359417473443002,
      "grad_norm": 1.0655860900878906,
      "learning_rate": 0.00018528136349060757,
      "loss": 3.6757,
      "step": 78950
    },
    {
      "epoch": 0.07360349635250911,
      "grad_norm": 1.1912130117416382,
      "learning_rate": 0.00018527949916851235,
      "loss": 3.6696,
      "step": 78960
    },
    {
      "epoch": 0.0736128179705882,
      "grad_norm": 1.2518402338027954,
      "learning_rate": 0.00018527763484641716,
      "loss": 4.062,
      "step": 78970
    },
    {
      "epoch": 0.07362213958866728,
      "grad_norm": 0.9570870399475098,
      "learning_rate": 0.00018527577052432194,
      "loss": 3.8645,
      "step": 78980
    },
    {
      "epoch": 0.07363146120674638,
      "grad_norm": 0.9797466993331909,
      "learning_rate": 0.00018527390620222674,
      "loss": 4.0889,
      "step": 78990
    },
    {
      "epoch": 0.07364078282482547,
      "grad_norm": 1.1061041355133057,
      "learning_rate": 0.00018527204188013158,
      "loss": 3.7832,
      "step": 79000
    },
    {
      "epoch": 0.07365010444290457,
      "grad_norm": 1.1529666185379028,
      "learning_rate": 0.00018527017755803636,
      "loss": 3.8096,
      "step": 79010
    },
    {
      "epoch": 0.07365942606098366,
      "grad_norm": 1.0688661336898804,
      "learning_rate": 0.00018526831323594116,
      "loss": 3.911,
      "step": 79020
    },
    {
      "epoch": 0.07366874767906274,
      "grad_norm": 1.175490379333496,
      "learning_rate": 0.00018526644891384594,
      "loss": 3.7723,
      "step": 79030
    },
    {
      "epoch": 0.07367806929714184,
      "grad_norm": 1.316123127937317,
      "learning_rate": 0.00018526458459175075,
      "loss": 3.9484,
      "step": 79040
    },
    {
      "epoch": 0.07368739091522093,
      "grad_norm": 1.0534316301345825,
      "learning_rate": 0.00018526272026965555,
      "loss": 3.7723,
      "step": 79050
    },
    {
      "epoch": 0.07369671253330003,
      "grad_norm": 1.117933988571167,
      "learning_rate": 0.00018526085594756036,
      "loss": 3.7698,
      "step": 79060
    },
    {
      "epoch": 0.07370603415137912,
      "grad_norm": 1.0817419290542603,
      "learning_rate": 0.00018525899162546517,
      "loss": 3.7862,
      "step": 79070
    },
    {
      "epoch": 0.0737153557694582,
      "grad_norm": 0.9922806620597839,
      "learning_rate": 0.00018525712730336995,
      "loss": 3.912,
      "step": 79080
    },
    {
      "epoch": 0.0737246773875373,
      "grad_norm": 1.063185453414917,
      "learning_rate": 0.00018525526298127475,
      "loss": 3.7986,
      "step": 79090
    },
    {
      "epoch": 0.07373399900561639,
      "grad_norm": 1.154409646987915,
      "learning_rate": 0.00018525339865917956,
      "loss": 3.6198,
      "step": 79100
    },
    {
      "epoch": 0.07374332062369549,
      "grad_norm": 1.1131900548934937,
      "learning_rate": 0.00018525153433708436,
      "loss": 3.6934,
      "step": 79110
    },
    {
      "epoch": 0.07375264224177458,
      "grad_norm": 0.9274615049362183,
      "learning_rate": 0.00018524967001498917,
      "loss": 3.8501,
      "step": 79120
    },
    {
      "epoch": 0.07376196385985366,
      "grad_norm": 1.1417930126190186,
      "learning_rate": 0.00018524780569289395,
      "loss": 3.7619,
      "step": 79130
    },
    {
      "epoch": 0.07377128547793275,
      "grad_norm": 1.0482580661773682,
      "learning_rate": 0.00018524594137079876,
      "loss": 3.6873,
      "step": 79140
    },
    {
      "epoch": 0.07378060709601185,
      "grad_norm": 1.1503934860229492,
      "learning_rate": 0.00018524407704870356,
      "loss": 3.8194,
      "step": 79150
    },
    {
      "epoch": 0.07378992871409094,
      "grad_norm": 0.9025560021400452,
      "learning_rate": 0.00018524221272660837,
      "loss": 3.8702,
      "step": 79160
    },
    {
      "epoch": 0.07379925033217004,
      "grad_norm": 1.1044920682907104,
      "learning_rate": 0.00018524034840451317,
      "loss": 3.7398,
      "step": 79170
    },
    {
      "epoch": 0.07380857195024913,
      "grad_norm": 2.5386290550231934,
      "learning_rate": 0.00018523848408241795,
      "loss": 3.9308,
      "step": 79180
    },
    {
      "epoch": 0.07381789356832821,
      "grad_norm": 0.9397866725921631,
      "learning_rate": 0.00018523661976032276,
      "loss": 3.6187,
      "step": 79190
    },
    {
      "epoch": 0.07382721518640731,
      "grad_norm": 0.7540329694747925,
      "learning_rate": 0.00018523475543822757,
      "loss": 3.5878,
      "step": 79200
    },
    {
      "epoch": 0.0738365368044864,
      "grad_norm": 0.8491885662078857,
      "learning_rate": 0.00018523289111613235,
      "loss": 3.8498,
      "step": 79210
    },
    {
      "epoch": 0.0738458584225655,
      "grad_norm": 1.1594064235687256,
      "learning_rate": 0.00018523102679403718,
      "loss": 3.7492,
      "step": 79220
    },
    {
      "epoch": 0.07385518004064459,
      "grad_norm": 0.9271681308746338,
      "learning_rate": 0.00018522916247194196,
      "loss": 3.9629,
      "step": 79230
    },
    {
      "epoch": 0.07386450165872367,
      "grad_norm": 1.1557846069335938,
      "learning_rate": 0.00018522729814984676,
      "loss": 3.738,
      "step": 79240
    },
    {
      "epoch": 0.07387382327680277,
      "grad_norm": 1.007734775543213,
      "learning_rate": 0.00018522543382775157,
      "loss": 3.7729,
      "step": 79250
    },
    {
      "epoch": 0.07388314489488186,
      "grad_norm": 0.9084041714668274,
      "learning_rate": 0.00018522356950565635,
      "loss": 3.9862,
      "step": 79260
    },
    {
      "epoch": 0.07389246651296096,
      "grad_norm": 1.2287400960922241,
      "learning_rate": 0.00018522170518356116,
      "loss": 3.9467,
      "step": 79270
    },
    {
      "epoch": 0.07390178813104005,
      "grad_norm": 1.0548228025436401,
      "learning_rate": 0.00018521984086146596,
      "loss": 3.7223,
      "step": 79280
    },
    {
      "epoch": 0.07391110974911913,
      "grad_norm": 1.1014212369918823,
      "learning_rate": 0.00018521797653937077,
      "loss": 3.8975,
      "step": 79290
    },
    {
      "epoch": 0.07392043136719822,
      "grad_norm": 1.0552330017089844,
      "learning_rate": 0.00018521611221727558,
      "loss": 3.914,
      "step": 79300
    },
    {
      "epoch": 0.07392975298527732,
      "grad_norm": 1.096727967262268,
      "learning_rate": 0.00018521424789518035,
      "loss": 3.8206,
      "step": 79310
    },
    {
      "epoch": 0.07393907460335641,
      "grad_norm": 0.9424831867218018,
      "learning_rate": 0.00018521238357308516,
      "loss": 3.9959,
      "step": 79320
    },
    {
      "epoch": 0.07394839622143551,
      "grad_norm": 1.2828948497772217,
      "learning_rate": 0.00018521051925098997,
      "loss": 3.7242,
      "step": 79330
    },
    {
      "epoch": 0.07395771783951459,
      "grad_norm": 0.9007053375244141,
      "learning_rate": 0.00018520865492889477,
      "loss": 3.8061,
      "step": 79340
    },
    {
      "epoch": 0.07396703945759368,
      "grad_norm": 1.1098709106445312,
      "learning_rate": 0.00018520679060679958,
      "loss": 3.9198,
      "step": 79350
    },
    {
      "epoch": 0.07397636107567278,
      "grad_norm": 0.9811954498291016,
      "learning_rate": 0.00018520492628470436,
      "loss": 3.661,
      "step": 79360
    },
    {
      "epoch": 0.07398568269375187,
      "grad_norm": 1.0589642524719238,
      "learning_rate": 0.00018520306196260917,
      "loss": 3.7138,
      "step": 79370
    },
    {
      "epoch": 0.07399500431183097,
      "grad_norm": 1.2527766227722168,
      "learning_rate": 0.00018520119764051394,
      "loss": 3.658,
      "step": 79380
    },
    {
      "epoch": 0.07400432592991006,
      "grad_norm": 1.2995247840881348,
      "learning_rate": 0.00018519933331841878,
      "loss": 3.8961,
      "step": 79390
    },
    {
      "epoch": 0.07401364754798914,
      "grad_norm": 1.0688937902450562,
      "learning_rate": 0.00018519746899632358,
      "loss": 3.787,
      "step": 79400
    },
    {
      "epoch": 0.07402296916606824,
      "grad_norm": 1.2022788524627686,
      "learning_rate": 0.00018519560467422836,
      "loss": 3.8777,
      "step": 79410
    },
    {
      "epoch": 0.07403229078414733,
      "grad_norm": 0.9362851977348328,
      "learning_rate": 0.00018519374035213317,
      "loss": 3.9495,
      "step": 79420
    },
    {
      "epoch": 0.07404161240222643,
      "grad_norm": 1.2386029958724976,
      "learning_rate": 0.00018519187603003795,
      "loss": 3.9413,
      "step": 79430
    },
    {
      "epoch": 0.07405093402030552,
      "grad_norm": 0.9747025966644287,
      "learning_rate": 0.00018519001170794276,
      "loss": 4.0347,
      "step": 79440
    },
    {
      "epoch": 0.0740602556383846,
      "grad_norm": 1.1773897409439087,
      "learning_rate": 0.0001851881473858476,
      "loss": 3.8392,
      "step": 79450
    },
    {
      "epoch": 0.0740695772564637,
      "grad_norm": 0.9605175852775574,
      "learning_rate": 0.00018518628306375237,
      "loss": 3.9929,
      "step": 79460
    },
    {
      "epoch": 0.07407889887454279,
      "grad_norm": 1.2485705614089966,
      "learning_rate": 0.00018518441874165717,
      "loss": 3.731,
      "step": 79470
    },
    {
      "epoch": 0.07408822049262188,
      "grad_norm": 1.161893367767334,
      "learning_rate": 0.00018518255441956195,
      "loss": 3.6642,
      "step": 79480
    },
    {
      "epoch": 0.07409754211070098,
      "grad_norm": 1.126102089881897,
      "learning_rate": 0.00018518069009746676,
      "loss": 3.776,
      "step": 79490
    },
    {
      "epoch": 0.07410686372878006,
      "grad_norm": 1.051212191581726,
      "learning_rate": 0.00018517882577537157,
      "loss": 3.7214,
      "step": 79500
    },
    {
      "epoch": 0.07411618534685915,
      "grad_norm": 1.132189393043518,
      "learning_rate": 0.00018517696145327637,
      "loss": 3.8494,
      "step": 79510
    },
    {
      "epoch": 0.07412550696493825,
      "grad_norm": 0.9569450616836548,
      "learning_rate": 0.00018517509713118118,
      "loss": 3.8154,
      "step": 79520
    },
    {
      "epoch": 0.07413482858301734,
      "grad_norm": 0.9987841844558716,
      "learning_rate": 0.00018517323280908596,
      "loss": 3.8783,
      "step": 79530
    },
    {
      "epoch": 0.07414415020109644,
      "grad_norm": 1.0722734928131104,
      "learning_rate": 0.00018517136848699076,
      "loss": 3.9866,
      "step": 79540
    },
    {
      "epoch": 0.07415347181917552,
      "grad_norm": 0.9905328154563904,
      "learning_rate": 0.00018516950416489557,
      "loss": 3.9549,
      "step": 79550
    },
    {
      "epoch": 0.07416279343725461,
      "grad_norm": 1.1108134984970093,
      "learning_rate": 0.00018516763984280038,
      "loss": 3.8757,
      "step": 79560
    },
    {
      "epoch": 0.0741721150553337,
      "grad_norm": 1.0047848224639893,
      "learning_rate": 0.00018516577552070518,
      "loss": 3.9405,
      "step": 79570
    },
    {
      "epoch": 0.0741814366734128,
      "grad_norm": 1.0366055965423584,
      "learning_rate": 0.00018516391119860996,
      "loss": 3.7302,
      "step": 79580
    },
    {
      "epoch": 0.0741907582914919,
      "grad_norm": 1.0598527193069458,
      "learning_rate": 0.00018516204687651477,
      "loss": 3.7584,
      "step": 79590
    },
    {
      "epoch": 0.07420007990957098,
      "grad_norm": 1.2095354795455933,
      "learning_rate": 0.00018516018255441957,
      "loss": 3.8921,
      "step": 79600
    },
    {
      "epoch": 0.07420940152765007,
      "grad_norm": 1.1372920274734497,
      "learning_rate": 0.00018515831823232438,
      "loss": 3.8659,
      "step": 79610
    },
    {
      "epoch": 0.07421872314572917,
      "grad_norm": 1.1084346771240234,
      "learning_rate": 0.0001851564539102292,
      "loss": 3.6317,
      "step": 79620
    },
    {
      "epoch": 0.07422804476380826,
      "grad_norm": 1.0829938650131226,
      "learning_rate": 0.00018515458958813397,
      "loss": 3.7853,
      "step": 79630
    },
    {
      "epoch": 0.07423736638188735,
      "grad_norm": 1.0455435514450073,
      "learning_rate": 0.00018515272526603877,
      "loss": 3.7172,
      "step": 79640
    },
    {
      "epoch": 0.07424668799996645,
      "grad_norm": 1.1194125413894653,
      "learning_rate": 0.00018515086094394358,
      "loss": 3.8077,
      "step": 79650
    },
    {
      "epoch": 0.07425600961804553,
      "grad_norm": 1.3247737884521484,
      "learning_rate": 0.00018514899662184836,
      "loss": 3.9661,
      "step": 79660
    },
    {
      "epoch": 0.07426533123612462,
      "grad_norm": 1.0234578847885132,
      "learning_rate": 0.0001851471322997532,
      "loss": 3.9135,
      "step": 79670
    },
    {
      "epoch": 0.07427465285420372,
      "grad_norm": 1.1412545442581177,
      "learning_rate": 0.00018514526797765797,
      "loss": 3.6193,
      "step": 79680
    },
    {
      "epoch": 0.07428397447228281,
      "grad_norm": 1.1402391195297241,
      "learning_rate": 0.00018514340365556278,
      "loss": 3.7819,
      "step": 79690
    },
    {
      "epoch": 0.07429329609036191,
      "grad_norm": 0.9238196015357971,
      "learning_rate": 0.00018514153933346758,
      "loss": 3.7725,
      "step": 79700
    },
    {
      "epoch": 0.07430261770844099,
      "grad_norm": 1.0697742700576782,
      "learning_rate": 0.00018513967501137236,
      "loss": 3.6477,
      "step": 79710
    },
    {
      "epoch": 0.07431193932652008,
      "grad_norm": 1.153845191001892,
      "learning_rate": 0.00018513781068927717,
      "loss": 3.8991,
      "step": 79720
    },
    {
      "epoch": 0.07432126094459918,
      "grad_norm": 0.9695149064064026,
      "learning_rate": 0.00018513594636718198,
      "loss": 3.8715,
      "step": 79730
    },
    {
      "epoch": 0.07433058256267827,
      "grad_norm": 0.953018069267273,
      "learning_rate": 0.00018513408204508678,
      "loss": 3.7132,
      "step": 79740
    },
    {
      "epoch": 0.07433990418075737,
      "grad_norm": 1.041069746017456,
      "learning_rate": 0.0001851322177229916,
      "loss": 3.7341,
      "step": 79750
    },
    {
      "epoch": 0.07434922579883645,
      "grad_norm": 1.0046310424804688,
      "learning_rate": 0.00018513035340089637,
      "loss": 3.7597,
      "step": 79760
    },
    {
      "epoch": 0.07435854741691554,
      "grad_norm": 1.0210875272750854,
      "learning_rate": 0.00018512848907880117,
      "loss": 3.7608,
      "step": 79770
    },
    {
      "epoch": 0.07436786903499464,
      "grad_norm": 1.2537965774536133,
      "learning_rate": 0.00018512662475670598,
      "loss": 3.6926,
      "step": 79780
    },
    {
      "epoch": 0.07437719065307373,
      "grad_norm": 0.9586920738220215,
      "learning_rate": 0.00018512476043461079,
      "loss": 3.975,
      "step": 79790
    },
    {
      "epoch": 0.07438651227115282,
      "grad_norm": 1.1427377462387085,
      "learning_rate": 0.0001851228961125156,
      "loss": 3.7154,
      "step": 79800
    },
    {
      "epoch": 0.0743958338892319,
      "grad_norm": 1.086852788925171,
      "learning_rate": 0.00018512103179042037,
      "loss": 3.709,
      "step": 79810
    },
    {
      "epoch": 0.074405155507311,
      "grad_norm": 1.3009737730026245,
      "learning_rate": 0.00018511916746832518,
      "loss": 3.6962,
      "step": 79820
    },
    {
      "epoch": 0.0744144771253901,
      "grad_norm": 1.1472022533416748,
      "learning_rate": 0.00018511730314622996,
      "loss": 3.7439,
      "step": 79830
    },
    {
      "epoch": 0.07442379874346919,
      "grad_norm": 0.9443302154541016,
      "learning_rate": 0.0001851154388241348,
      "loss": 3.6863,
      "step": 79840
    },
    {
      "epoch": 0.07443312036154828,
      "grad_norm": 1.14724862575531,
      "learning_rate": 0.00018511357450203957,
      "loss": 3.6678,
      "step": 79850
    },
    {
      "epoch": 0.07444244197962738,
      "grad_norm": 1.0767720937728882,
      "learning_rate": 0.00018511171017994438,
      "loss": 3.6665,
      "step": 79860
    },
    {
      "epoch": 0.07445176359770646,
      "grad_norm": 1.1365559101104736,
      "learning_rate": 0.00018510984585784918,
      "loss": 3.8199,
      "step": 79870
    },
    {
      "epoch": 0.07446108521578555,
      "grad_norm": 1.0718685388565063,
      "learning_rate": 0.00018510798153575396,
      "loss": 3.7482,
      "step": 79880
    },
    {
      "epoch": 0.07447040683386465,
      "grad_norm": 1.1507792472839355,
      "learning_rate": 0.00018510611721365877,
      "loss": 4.0545,
      "step": 79890
    },
    {
      "epoch": 0.07447972845194374,
      "grad_norm": 1.2483489513397217,
      "learning_rate": 0.00018510425289156357,
      "loss": 3.7774,
      "step": 79900
    },
    {
      "epoch": 0.07448905007002284,
      "grad_norm": 1.1744232177734375,
      "learning_rate": 0.00018510238856946838,
      "loss": 3.8315,
      "step": 79910
    },
    {
      "epoch": 0.07449837168810192,
      "grad_norm": 1.054312825202942,
      "learning_rate": 0.0001851005242473732,
      "loss": 3.8614,
      "step": 79920
    },
    {
      "epoch": 0.07450769330618101,
      "grad_norm": 1.173940658569336,
      "learning_rate": 0.00018509865992527797,
      "loss": 3.5118,
      "step": 79930
    },
    {
      "epoch": 0.0745170149242601,
      "grad_norm": 1.0723518133163452,
      "learning_rate": 0.00018509679560318277,
      "loss": 3.8641,
      "step": 79940
    },
    {
      "epoch": 0.0745263365423392,
      "grad_norm": 1.1629143953323364,
      "learning_rate": 0.00018509493128108758,
      "loss": 3.8672,
      "step": 79950
    },
    {
      "epoch": 0.0745356581604183,
      "grad_norm": 1.1506571769714355,
      "learning_rate": 0.00018509306695899239,
      "loss": 3.7606,
      "step": 79960
    },
    {
      "epoch": 0.07454497977849738,
      "grad_norm": 1.2605652809143066,
      "learning_rate": 0.0001850912026368972,
      "loss": 3.8124,
      "step": 79970
    },
    {
      "epoch": 0.07455430139657647,
      "grad_norm": 1.1004022359848022,
      "learning_rate": 0.00018508933831480197,
      "loss": 3.9018,
      "step": 79980
    },
    {
      "epoch": 0.07456362301465556,
      "grad_norm": 0.942916214466095,
      "learning_rate": 0.00018508747399270678,
      "loss": 3.7095,
      "step": 79990
    },
    {
      "epoch": 0.07457294463273466,
      "grad_norm": 1.080061674118042,
      "learning_rate": 0.00018508560967061158,
      "loss": 3.8693,
      "step": 80000
    },
    {
      "epoch": 0.07458226625081375,
      "grad_norm": 1.0025842189788818,
      "learning_rate": 0.0001850837453485164,
      "loss": 3.8233,
      "step": 80010
    },
    {
      "epoch": 0.07459158786889283,
      "grad_norm": 1.1944469213485718,
      "learning_rate": 0.0001850818810264212,
      "loss": 3.6549,
      "step": 80020
    },
    {
      "epoch": 0.07460090948697193,
      "grad_norm": 1.3277010917663574,
      "learning_rate": 0.00018508001670432598,
      "loss": 3.7108,
      "step": 80030
    },
    {
      "epoch": 0.07461023110505102,
      "grad_norm": 0.8908520340919495,
      "learning_rate": 0.00018507815238223078,
      "loss": 3.9186,
      "step": 80040
    },
    {
      "epoch": 0.07461955272313012,
      "grad_norm": 0.9831807613372803,
      "learning_rate": 0.00018507628806013556,
      "loss": 3.8606,
      "step": 80050
    },
    {
      "epoch": 0.07462887434120921,
      "grad_norm": 1.099531888961792,
      "learning_rate": 0.0001850744237380404,
      "loss": 3.7246,
      "step": 80060
    },
    {
      "epoch": 0.0746381959592883,
      "grad_norm": 1.0653398036956787,
      "learning_rate": 0.0001850725594159452,
      "loss": 3.739,
      "step": 80070
    },
    {
      "epoch": 0.07464751757736739,
      "grad_norm": 1.1138402223587036,
      "learning_rate": 0.00018507069509384998,
      "loss": 3.6927,
      "step": 80080
    },
    {
      "epoch": 0.07465683919544648,
      "grad_norm": 1.1370431184768677,
      "learning_rate": 0.00018506883077175479,
      "loss": 3.7774,
      "step": 80090
    },
    {
      "epoch": 0.07466616081352558,
      "grad_norm": 0.9090743660926819,
      "learning_rate": 0.00018506696644965957,
      "loss": 3.9217,
      "step": 80100
    },
    {
      "epoch": 0.07467548243160467,
      "grad_norm": 0.9894599914550781,
      "learning_rate": 0.00018506510212756437,
      "loss": 3.839,
      "step": 80110
    },
    {
      "epoch": 0.07468480404968376,
      "grad_norm": 1.1080138683319092,
      "learning_rate": 0.0001850632378054692,
      "loss": 3.7227,
      "step": 80120
    },
    {
      "epoch": 0.07469412566776285,
      "grad_norm": 0.9110984206199646,
      "learning_rate": 0.00018506137348337398,
      "loss": 3.8331,
      "step": 80130
    },
    {
      "epoch": 0.07470344728584194,
      "grad_norm": 1.1192741394042969,
      "learning_rate": 0.0001850595091612788,
      "loss": 3.9496,
      "step": 80140
    },
    {
      "epoch": 0.07471276890392103,
      "grad_norm": 1.079525351524353,
      "learning_rate": 0.00018505764483918357,
      "loss": 3.6871,
      "step": 80150
    },
    {
      "epoch": 0.07472209052200013,
      "grad_norm": 1.0373847484588623,
      "learning_rate": 0.00018505578051708838,
      "loss": 3.8168,
      "step": 80160
    },
    {
      "epoch": 0.07473141214007922,
      "grad_norm": 1.029648780822754,
      "learning_rate": 0.00018505391619499318,
      "loss": 3.8995,
      "step": 80170
    },
    {
      "epoch": 0.0747407337581583,
      "grad_norm": 1.0322539806365967,
      "learning_rate": 0.000185052051872898,
      "loss": 3.7938,
      "step": 80180
    },
    {
      "epoch": 0.0747500553762374,
      "grad_norm": 0.9557073712348938,
      "learning_rate": 0.0001850501875508028,
      "loss": 3.7844,
      "step": 80190
    },
    {
      "epoch": 0.07475937699431649,
      "grad_norm": 1.2273756265640259,
      "learning_rate": 0.00018504832322870757,
      "loss": 3.984,
      "step": 80200
    },
    {
      "epoch": 0.07476869861239559,
      "grad_norm": 1.1222267150878906,
      "learning_rate": 0.00018504645890661238,
      "loss": 3.9104,
      "step": 80210
    },
    {
      "epoch": 0.07477802023047468,
      "grad_norm": 1.101845145225525,
      "learning_rate": 0.0001850445945845172,
      "loss": 3.795,
      "step": 80220
    },
    {
      "epoch": 0.07478734184855376,
      "grad_norm": 1.0476614236831665,
      "learning_rate": 0.000185042730262422,
      "loss": 3.9253,
      "step": 80230
    },
    {
      "epoch": 0.07479666346663286,
      "grad_norm": 1.1890062093734741,
      "learning_rate": 0.0001850408659403268,
      "loss": 3.8271,
      "step": 80240
    },
    {
      "epoch": 0.07480598508471195,
      "grad_norm": 1.1136850118637085,
      "learning_rate": 0.00018503900161823158,
      "loss": 3.7299,
      "step": 80250
    },
    {
      "epoch": 0.07481530670279105,
      "grad_norm": 0.9716820120811462,
      "learning_rate": 0.00018503713729613638,
      "loss": 3.6792,
      "step": 80260
    },
    {
      "epoch": 0.07482462832087014,
      "grad_norm": 1.057136058807373,
      "learning_rate": 0.0001850352729740412,
      "loss": 3.7777,
      "step": 80270
    },
    {
      "epoch": 0.07483394993894922,
      "grad_norm": 1.2975666522979736,
      "learning_rate": 0.00018503340865194597,
      "loss": 3.8359,
      "step": 80280
    },
    {
      "epoch": 0.07484327155702832,
      "grad_norm": 1.0191689729690552,
      "learning_rate": 0.0001850315443298508,
      "loss": 3.7453,
      "step": 80290
    },
    {
      "epoch": 0.07485259317510741,
      "grad_norm": 1.0830340385437012,
      "learning_rate": 0.00018502968000775558,
      "loss": 3.6678,
      "step": 80300
    },
    {
      "epoch": 0.0748619147931865,
      "grad_norm": 1.1761387586593628,
      "learning_rate": 0.0001850278156856604,
      "loss": 3.6957,
      "step": 80310
    },
    {
      "epoch": 0.0748712364112656,
      "grad_norm": 1.2274283170700073,
      "learning_rate": 0.0001850259513635652,
      "loss": 3.7807,
      "step": 80320
    },
    {
      "epoch": 0.0748805580293447,
      "grad_norm": 0.9830955862998962,
      "learning_rate": 0.00018502408704146997,
      "loss": 3.7464,
      "step": 80330
    },
    {
      "epoch": 0.07488987964742377,
      "grad_norm": 1.3028446435928345,
      "learning_rate": 0.00018502222271937478,
      "loss": 3.8998,
      "step": 80340
    },
    {
      "epoch": 0.07489920126550287,
      "grad_norm": 1.1266982555389404,
      "learning_rate": 0.0001850203583972796,
      "loss": 3.8809,
      "step": 80350
    },
    {
      "epoch": 0.07490852288358196,
      "grad_norm": 1.2465615272521973,
      "learning_rate": 0.0001850184940751844,
      "loss": 3.908,
      "step": 80360
    },
    {
      "epoch": 0.07491784450166106,
      "grad_norm": 1.0517597198486328,
      "learning_rate": 0.0001850166297530892,
      "loss": 3.7315,
      "step": 80370
    },
    {
      "epoch": 0.07492716611974015,
      "grad_norm": 0.834896445274353,
      "learning_rate": 0.00018501476543099398,
      "loss": 3.7677,
      "step": 80380
    },
    {
      "epoch": 0.07493648773781923,
      "grad_norm": 0.9716144800186157,
      "learning_rate": 0.00018501290110889879,
      "loss": 3.7668,
      "step": 80390
    },
    {
      "epoch": 0.07494580935589833,
      "grad_norm": 1.0598503351211548,
      "learning_rate": 0.0001850110367868036,
      "loss": 4.0705,
      "step": 80400
    },
    {
      "epoch": 0.07495513097397742,
      "grad_norm": 1.2589986324310303,
      "learning_rate": 0.0001850091724647084,
      "loss": 3.6889,
      "step": 80410
    },
    {
      "epoch": 0.07496445259205652,
      "grad_norm": 1.0008798837661743,
      "learning_rate": 0.0001850073081426132,
      "loss": 3.8475,
      "step": 80420
    },
    {
      "epoch": 0.07497377421013561,
      "grad_norm": 1.1739660501480103,
      "learning_rate": 0.00018500544382051798,
      "loss": 3.9212,
      "step": 80430
    },
    {
      "epoch": 0.07498309582821469,
      "grad_norm": 1.124647617340088,
      "learning_rate": 0.0001850035794984228,
      "loss": 3.865,
      "step": 80440
    },
    {
      "epoch": 0.07499241744629379,
      "grad_norm": 1.0990586280822754,
      "learning_rate": 0.0001850017151763276,
      "loss": 3.7834,
      "step": 80450
    },
    {
      "epoch": 0.07500173906437288,
      "grad_norm": 0.9929418563842773,
      "learning_rate": 0.0001849998508542324,
      "loss": 3.9222,
      "step": 80460
    },
    {
      "epoch": 0.07501106068245197,
      "grad_norm": 1.1093653440475464,
      "learning_rate": 0.0001849979865321372,
      "loss": 3.7318,
      "step": 80470
    },
    {
      "epoch": 0.07502038230053107,
      "grad_norm": 1.3830503225326538,
      "learning_rate": 0.000184996122210042,
      "loss": 3.9325,
      "step": 80480
    },
    {
      "epoch": 0.07502970391861015,
      "grad_norm": 1.0423612594604492,
      "learning_rate": 0.0001849942578879468,
      "loss": 3.7273,
      "step": 80490
    },
    {
      "epoch": 0.07503902553668924,
      "grad_norm": 1.1564722061157227,
      "learning_rate": 0.00018499239356585157,
      "loss": 3.8233,
      "step": 80500
    },
    {
      "epoch": 0.07504834715476834,
      "grad_norm": 0.9875454306602478,
      "learning_rate": 0.0001849905292437564,
      "loss": 3.9071,
      "step": 80510
    },
    {
      "epoch": 0.07505766877284743,
      "grad_norm": 1.0237983465194702,
      "learning_rate": 0.0001849886649216612,
      "loss": 3.8406,
      "step": 80520
    },
    {
      "epoch": 0.07506699039092653,
      "grad_norm": 1.016145944595337,
      "learning_rate": 0.000184986800599566,
      "loss": 3.7605,
      "step": 80530
    },
    {
      "epoch": 0.07507631200900562,
      "grad_norm": 1.1865854263305664,
      "learning_rate": 0.0001849849362774708,
      "loss": 3.7549,
      "step": 80540
    },
    {
      "epoch": 0.0750856336270847,
      "grad_norm": 1.0108388662338257,
      "learning_rate": 0.00018498307195537558,
      "loss": 3.841,
      "step": 80550
    },
    {
      "epoch": 0.0750949552451638,
      "grad_norm": 1.1951607465744019,
      "learning_rate": 0.00018498120763328038,
      "loss": 3.6219,
      "step": 80560
    },
    {
      "epoch": 0.07510427686324289,
      "grad_norm": 0.9529380798339844,
      "learning_rate": 0.00018497934331118522,
      "loss": 3.9533,
      "step": 80570
    },
    {
      "epoch": 0.07511359848132199,
      "grad_norm": 1.1464850902557373,
      "learning_rate": 0.00018497747898909,
      "loss": 3.9433,
      "step": 80580
    },
    {
      "epoch": 0.07512292009940108,
      "grad_norm": 1.133994460105896,
      "learning_rate": 0.0001849756146669948,
      "loss": 3.8594,
      "step": 80590
    },
    {
      "epoch": 0.07513224171748016,
      "grad_norm": 1.1803213357925415,
      "learning_rate": 0.00018497375034489958,
      "loss": 3.8479,
      "step": 80600
    },
    {
      "epoch": 0.07514156333555926,
      "grad_norm": 1.0451217889785767,
      "learning_rate": 0.0001849718860228044,
      "loss": 3.7746,
      "step": 80610
    },
    {
      "epoch": 0.07515088495363835,
      "grad_norm": 0.9884854555130005,
      "learning_rate": 0.0001849700217007092,
      "loss": 3.5617,
      "step": 80620
    },
    {
      "epoch": 0.07516020657171744,
      "grad_norm": 1.0275049209594727,
      "learning_rate": 0.000184968157378614,
      "loss": 3.8411,
      "step": 80630
    },
    {
      "epoch": 0.07516952818979654,
      "grad_norm": 1.1418606042861938,
      "learning_rate": 0.0001849662930565188,
      "loss": 3.7062,
      "step": 80640
    },
    {
      "epoch": 0.07517884980787562,
      "grad_norm": 1.0110948085784912,
      "learning_rate": 0.0001849644287344236,
      "loss": 3.9812,
      "step": 80650
    },
    {
      "epoch": 0.07518817142595471,
      "grad_norm": 1.0479257106781006,
      "learning_rate": 0.0001849625644123284,
      "loss": 4.0277,
      "step": 80660
    },
    {
      "epoch": 0.07519749304403381,
      "grad_norm": 1.114606261253357,
      "learning_rate": 0.0001849607000902332,
      "loss": 3.9833,
      "step": 80670
    },
    {
      "epoch": 0.0752068146621129,
      "grad_norm": 1.0317565202713013,
      "learning_rate": 0.000184958835768138,
      "loss": 3.722,
      "step": 80680
    },
    {
      "epoch": 0.075216136280192,
      "grad_norm": 0.9864277243614197,
      "learning_rate": 0.0001849569714460428,
      "loss": 3.7316,
      "step": 80690
    },
    {
      "epoch": 0.07522545789827108,
      "grad_norm": 1.0446549654006958,
      "learning_rate": 0.0001849551071239476,
      "loss": 3.7886,
      "step": 80700
    },
    {
      "epoch": 0.07523477951635017,
      "grad_norm": 1.222643256187439,
      "learning_rate": 0.0001849532428018524,
      "loss": 3.7373,
      "step": 80710
    },
    {
      "epoch": 0.07524410113442927,
      "grad_norm": 0.9145188927650452,
      "learning_rate": 0.0001849513784797572,
      "loss": 3.8256,
      "step": 80720
    },
    {
      "epoch": 0.07525342275250836,
      "grad_norm": 0.9605583548545837,
      "learning_rate": 0.00018494951415766198,
      "loss": 3.723,
      "step": 80730
    },
    {
      "epoch": 0.07526274437058746,
      "grad_norm": 0.9803037047386169,
      "learning_rate": 0.00018494764983556682,
      "loss": 3.9582,
      "step": 80740
    },
    {
      "epoch": 0.07527206598866655,
      "grad_norm": 1.0578200817108154,
      "learning_rate": 0.0001849457855134716,
      "loss": 3.6708,
      "step": 80750
    },
    {
      "epoch": 0.07528138760674563,
      "grad_norm": 1.178858995437622,
      "learning_rate": 0.0001849439211913764,
      "loss": 3.8612,
      "step": 80760
    },
    {
      "epoch": 0.07529070922482473,
      "grad_norm": 1.3256014585494995,
      "learning_rate": 0.0001849420568692812,
      "loss": 3.7706,
      "step": 80770
    },
    {
      "epoch": 0.07530003084290382,
      "grad_norm": 1.150316834449768,
      "learning_rate": 0.000184940192547186,
      "loss": 3.7733,
      "step": 80780
    },
    {
      "epoch": 0.07530935246098291,
      "grad_norm": 0.9960058927536011,
      "learning_rate": 0.0001849383282250908,
      "loss": 3.7185,
      "step": 80790
    },
    {
      "epoch": 0.07531867407906201,
      "grad_norm": 1.0277824401855469,
      "learning_rate": 0.0001849364639029956,
      "loss": 3.8518,
      "step": 80800
    },
    {
      "epoch": 0.07532799569714109,
      "grad_norm": 1.1426531076431274,
      "learning_rate": 0.0001849345995809004,
      "loss": 3.9548,
      "step": 80810
    },
    {
      "epoch": 0.07533731731522018,
      "grad_norm": 1.0031025409698486,
      "learning_rate": 0.0001849327352588052,
      "loss": 3.8516,
      "step": 80820
    },
    {
      "epoch": 0.07534663893329928,
      "grad_norm": 1.1393113136291504,
      "learning_rate": 0.00018493087093671,
      "loss": 4.0157,
      "step": 80830
    },
    {
      "epoch": 0.07535596055137837,
      "grad_norm": 1.0525585412979126,
      "learning_rate": 0.0001849290066146148,
      "loss": 4.0455,
      "step": 80840
    },
    {
      "epoch": 0.07536528216945747,
      "grad_norm": 1.1889630556106567,
      "learning_rate": 0.0001849271422925196,
      "loss": 3.6247,
      "step": 80850
    },
    {
      "epoch": 0.07537460378753655,
      "grad_norm": 1.0412622690200806,
      "learning_rate": 0.0001849252779704244,
      "loss": 3.8252,
      "step": 80860
    },
    {
      "epoch": 0.07538392540561564,
      "grad_norm": 1.0427354574203491,
      "learning_rate": 0.00018492341364832922,
      "loss": 3.7058,
      "step": 80870
    },
    {
      "epoch": 0.07539324702369474,
      "grad_norm": 0.9782912731170654,
      "learning_rate": 0.000184921549326234,
      "loss": 3.8852,
      "step": 80880
    },
    {
      "epoch": 0.07540256864177383,
      "grad_norm": 1.1014715433120728,
      "learning_rate": 0.0001849196850041388,
      "loss": 3.6894,
      "step": 80890
    },
    {
      "epoch": 0.07541189025985293,
      "grad_norm": 1.1810920238494873,
      "learning_rate": 0.0001849178206820436,
      "loss": 3.7365,
      "step": 80900
    },
    {
      "epoch": 0.075421211877932,
      "grad_norm": 1.0734301805496216,
      "learning_rate": 0.00018491595635994842,
      "loss": 3.7155,
      "step": 80910
    },
    {
      "epoch": 0.0754305334960111,
      "grad_norm": 1.0183849334716797,
      "learning_rate": 0.00018491409203785322,
      "loss": 3.8154,
      "step": 80920
    },
    {
      "epoch": 0.0754398551140902,
      "grad_norm": 1.1193690299987793,
      "learning_rate": 0.000184912227715758,
      "loss": 3.6617,
      "step": 80930
    },
    {
      "epoch": 0.07544917673216929,
      "grad_norm": 1.197312831878662,
      "learning_rate": 0.0001849103633936628,
      "loss": 3.8797,
      "step": 80940
    },
    {
      "epoch": 0.07545849835024838,
      "grad_norm": 0.9572570323944092,
      "learning_rate": 0.00018490849907156759,
      "loss": 3.804,
      "step": 80950
    },
    {
      "epoch": 0.07546781996832747,
      "grad_norm": 1.1353247165679932,
      "learning_rate": 0.00018490663474947242,
      "loss": 3.7729,
      "step": 80960
    },
    {
      "epoch": 0.07547714158640656,
      "grad_norm": 1.0474032163619995,
      "learning_rate": 0.00018490477042737723,
      "loss": 3.6622,
      "step": 80970
    },
    {
      "epoch": 0.07548646320448565,
      "grad_norm": 1.1509714126586914,
      "learning_rate": 0.000184902906105282,
      "loss": 3.5922,
      "step": 80980
    },
    {
      "epoch": 0.07549578482256475,
      "grad_norm": 1.2424473762512207,
      "learning_rate": 0.0001849010417831868,
      "loss": 3.8077,
      "step": 80990
    },
    {
      "epoch": 0.07550510644064384,
      "grad_norm": 1.0256150960922241,
      "learning_rate": 0.0001848991774610916,
      "loss": 3.922,
      "step": 81000
    },
    {
      "epoch": 0.07551442805872294,
      "grad_norm": 1.0934542417526245,
      "learning_rate": 0.0001848973131389964,
      "loss": 3.7848,
      "step": 81010
    },
    {
      "epoch": 0.07552374967680202,
      "grad_norm": 1.0612624883651733,
      "learning_rate": 0.00018489544881690123,
      "loss": 3.6519,
      "step": 81020
    },
    {
      "epoch": 0.07553307129488111,
      "grad_norm": 0.9169989824295044,
      "learning_rate": 0.000184893584494806,
      "loss": 3.878,
      "step": 81030
    },
    {
      "epoch": 0.07554239291296021,
      "grad_norm": 0.9200854301452637,
      "learning_rate": 0.00018489172017271082,
      "loss": 3.7422,
      "step": 81040
    },
    {
      "epoch": 0.0755517145310393,
      "grad_norm": 1.0695017576217651,
      "learning_rate": 0.0001848898558506156,
      "loss": 3.7824,
      "step": 81050
    },
    {
      "epoch": 0.0755610361491184,
      "grad_norm": 0.8942006826400757,
      "learning_rate": 0.0001848879915285204,
      "loss": 3.7072,
      "step": 81060
    },
    {
      "epoch": 0.07557035776719748,
      "grad_norm": 1.0552952289581299,
      "learning_rate": 0.0001848861272064252,
      "loss": 3.7699,
      "step": 81070
    },
    {
      "epoch": 0.07557967938527657,
      "grad_norm": 0.9809656739234924,
      "learning_rate": 0.00018488426288433001,
      "loss": 3.842,
      "step": 81080
    },
    {
      "epoch": 0.07558900100335567,
      "grad_norm": 1.1730631589889526,
      "learning_rate": 0.00018488239856223482,
      "loss": 3.8363,
      "step": 81090
    },
    {
      "epoch": 0.07559832262143476,
      "grad_norm": 0.9259490370750427,
      "learning_rate": 0.0001848805342401396,
      "loss": 3.6616,
      "step": 81100
    },
    {
      "epoch": 0.07560764423951385,
      "grad_norm": 0.9076659679412842,
      "learning_rate": 0.0001848786699180444,
      "loss": 3.7206,
      "step": 81110
    },
    {
      "epoch": 0.07561696585759294,
      "grad_norm": 1.2718636989593506,
      "learning_rate": 0.0001848768055959492,
      "loss": 3.7594,
      "step": 81120
    },
    {
      "epoch": 0.07562628747567203,
      "grad_norm": 1.4040720462799072,
      "learning_rate": 0.00018487494127385402,
      "loss": 3.7508,
      "step": 81130
    },
    {
      "epoch": 0.07563560909375112,
      "grad_norm": 0.9901749491691589,
      "learning_rate": 0.00018487307695175882,
      "loss": 3.7617,
      "step": 81140
    },
    {
      "epoch": 0.07564493071183022,
      "grad_norm": 1.13004732131958,
      "learning_rate": 0.0001848712126296636,
      "loss": 3.9102,
      "step": 81150
    },
    {
      "epoch": 0.07565425232990931,
      "grad_norm": 0.9936280846595764,
      "learning_rate": 0.0001848693483075684,
      "loss": 3.7094,
      "step": 81160
    },
    {
      "epoch": 0.0756635739479884,
      "grad_norm": 0.9546281695365906,
      "learning_rate": 0.00018486748398547322,
      "loss": 3.7657,
      "step": 81170
    },
    {
      "epoch": 0.07567289556606749,
      "grad_norm": 1.1138981580734253,
      "learning_rate": 0.000184865619663378,
      "loss": 3.8325,
      "step": 81180
    },
    {
      "epoch": 0.07568221718414658,
      "grad_norm": 1.1848387718200684,
      "learning_rate": 0.00018486375534128283,
      "loss": 3.8686,
      "step": 81190
    },
    {
      "epoch": 0.07569153880222568,
      "grad_norm": 1.0183531045913696,
      "learning_rate": 0.0001848618910191876,
      "loss": 3.8078,
      "step": 81200
    },
    {
      "epoch": 0.07570086042030477,
      "grad_norm": 1.1595252752304077,
      "learning_rate": 0.00018486002669709241,
      "loss": 3.8267,
      "step": 81210
    },
    {
      "epoch": 0.07571018203838387,
      "grad_norm": 1.1059322357177734,
      "learning_rate": 0.0001848581623749972,
      "loss": 3.8241,
      "step": 81220
    },
    {
      "epoch": 0.07571950365646295,
      "grad_norm": 1.0431773662567139,
      "learning_rate": 0.000184856298052902,
      "loss": 3.7904,
      "step": 81230
    },
    {
      "epoch": 0.07572882527454204,
      "grad_norm": 1.1038494110107422,
      "learning_rate": 0.0001848544337308068,
      "loss": 3.6806,
      "step": 81240
    },
    {
      "epoch": 0.07573814689262114,
      "grad_norm": 1.0915961265563965,
      "learning_rate": 0.0001848525694087116,
      "loss": 3.751,
      "step": 81250
    },
    {
      "epoch": 0.07574746851070023,
      "grad_norm": 1.1874957084655762,
      "learning_rate": 0.00018485070508661642,
      "loss": 3.8605,
      "step": 81260
    },
    {
      "epoch": 0.07575679012877932,
      "grad_norm": 0.9485775828361511,
      "learning_rate": 0.0001848488407645212,
      "loss": 3.9039,
      "step": 81270
    },
    {
      "epoch": 0.0757661117468584,
      "grad_norm": 1.2642501592636108,
      "learning_rate": 0.000184846976442426,
      "loss": 3.8562,
      "step": 81280
    },
    {
      "epoch": 0.0757754333649375,
      "grad_norm": 1.0535961389541626,
      "learning_rate": 0.0001848451121203308,
      "loss": 3.6591,
      "step": 81290
    },
    {
      "epoch": 0.0757847549830166,
      "grad_norm": 1.5934374332427979,
      "learning_rate": 0.00018484324779823562,
      "loss": 3.8011,
      "step": 81300
    },
    {
      "epoch": 0.07579407660109569,
      "grad_norm": 1.1440191268920898,
      "learning_rate": 0.00018484138347614042,
      "loss": 3.6313,
      "step": 81310
    },
    {
      "epoch": 0.07580339821917478,
      "grad_norm": 1.185275673866272,
      "learning_rate": 0.0001848395191540452,
      "loss": 3.6735,
      "step": 81320
    },
    {
      "epoch": 0.07581271983725386,
      "grad_norm": 1.1383898258209229,
      "learning_rate": 0.00018483765483195,
      "loss": 3.6566,
      "step": 81330
    },
    {
      "epoch": 0.07582204145533296,
      "grad_norm": 1.0130380392074585,
      "learning_rate": 0.00018483579050985482,
      "loss": 3.9179,
      "step": 81340
    },
    {
      "epoch": 0.07583136307341205,
      "grad_norm": 1.100295066833496,
      "learning_rate": 0.00018483392618775962,
      "loss": 3.8942,
      "step": 81350
    },
    {
      "epoch": 0.07584068469149115,
      "grad_norm": 0.9987223744392395,
      "learning_rate": 0.00018483206186566443,
      "loss": 3.8334,
      "step": 81360
    },
    {
      "epoch": 0.07585000630957024,
      "grad_norm": 1.1365176439285278,
      "learning_rate": 0.0001848301975435692,
      "loss": 3.7442,
      "step": 81370
    },
    {
      "epoch": 0.07585932792764932,
      "grad_norm": 0.9800360202789307,
      "learning_rate": 0.00018482833322147401,
      "loss": 3.7899,
      "step": 81380
    },
    {
      "epoch": 0.07586864954572842,
      "grad_norm": 1.0914864540100098,
      "learning_rate": 0.00018482646889937882,
      "loss": 3.8989,
      "step": 81390
    },
    {
      "epoch": 0.07587797116380751,
      "grad_norm": 0.983963668346405,
      "learning_rate": 0.0001848246045772836,
      "loss": 3.7048,
      "step": 81400
    },
    {
      "epoch": 0.0758872927818866,
      "grad_norm": 1.0244697332382202,
      "learning_rate": 0.00018482274025518843,
      "loss": 3.8546,
      "step": 81410
    },
    {
      "epoch": 0.0758966143999657,
      "grad_norm": 1.1384459733963013,
      "learning_rate": 0.0001848208759330932,
      "loss": 3.7382,
      "step": 81420
    },
    {
      "epoch": 0.0759059360180448,
      "grad_norm": 1.157523512840271,
      "learning_rate": 0.00018481901161099802,
      "loss": 3.8188,
      "step": 81430
    },
    {
      "epoch": 0.07591525763612388,
      "grad_norm": 1.0154308080673218,
      "learning_rate": 0.00018481714728890282,
      "loss": 3.7061,
      "step": 81440
    },
    {
      "epoch": 0.07592457925420297,
      "grad_norm": 1.1504961252212524,
      "learning_rate": 0.0001848152829668076,
      "loss": 3.9431,
      "step": 81450
    },
    {
      "epoch": 0.07593390087228206,
      "grad_norm": 1.2126555442810059,
      "learning_rate": 0.0001848134186447124,
      "loss": 3.7488,
      "step": 81460
    },
    {
      "epoch": 0.07594322249036116,
      "grad_norm": 0.9156156778335571,
      "learning_rate": 0.00018481155432261722,
      "loss": 3.894,
      "step": 81470
    },
    {
      "epoch": 0.07595254410844025,
      "grad_norm": 1.0703355073928833,
      "learning_rate": 0.00018480969000052202,
      "loss": 3.7993,
      "step": 81480
    },
    {
      "epoch": 0.07596186572651933,
      "grad_norm": 1.228570580482483,
      "learning_rate": 0.00018480782567842683,
      "loss": 3.6681,
      "step": 81490
    },
    {
      "epoch": 0.07597118734459843,
      "grad_norm": 1.2088655233383179,
      "learning_rate": 0.0001848059613563316,
      "loss": 3.8039,
      "step": 81500
    },
    {
      "epoch": 0.07598050896267752,
      "grad_norm": 1.041456937789917,
      "learning_rate": 0.00018480409703423641,
      "loss": 3.99,
      "step": 81510
    },
    {
      "epoch": 0.07598983058075662,
      "grad_norm": 1.1755039691925049,
      "learning_rate": 0.00018480223271214122,
      "loss": 3.7025,
      "step": 81520
    },
    {
      "epoch": 0.07599915219883571,
      "grad_norm": 1.00324547290802,
      "learning_rate": 0.00018480036839004603,
      "loss": 3.7288,
      "step": 81530
    },
    {
      "epoch": 0.07600847381691479,
      "grad_norm": 1.058884620666504,
      "learning_rate": 0.00018479850406795083,
      "loss": 3.7864,
      "step": 81540
    },
    {
      "epoch": 0.07601779543499389,
      "grad_norm": 0.9747676849365234,
      "learning_rate": 0.0001847966397458556,
      "loss": 4.06,
      "step": 81550
    },
    {
      "epoch": 0.07602711705307298,
      "grad_norm": 1.1180719137191772,
      "learning_rate": 0.00018479477542376042,
      "loss": 3.7917,
      "step": 81560
    },
    {
      "epoch": 0.07603643867115208,
      "grad_norm": 1.214964747428894,
      "learning_rate": 0.0001847929111016652,
      "loss": 3.7826,
      "step": 81570
    },
    {
      "epoch": 0.07604576028923117,
      "grad_norm": 1.1240227222442627,
      "learning_rate": 0.00018479104677957003,
      "loss": 3.8112,
      "step": 81580
    },
    {
      "epoch": 0.07605508190731025,
      "grad_norm": 1.060699224472046,
      "learning_rate": 0.00018478918245747484,
      "loss": 3.5645,
      "step": 81590
    },
    {
      "epoch": 0.07606440352538935,
      "grad_norm": 1.1299245357513428,
      "learning_rate": 0.00018478731813537962,
      "loss": 3.9895,
      "step": 81600
    },
    {
      "epoch": 0.07607372514346844,
      "grad_norm": 1.0092805624008179,
      "learning_rate": 0.00018478545381328442,
      "loss": 3.9151,
      "step": 81610
    },
    {
      "epoch": 0.07608304676154753,
      "grad_norm": 1.2276062965393066,
      "learning_rate": 0.0001847835894911892,
      "loss": 3.8112,
      "step": 81620
    },
    {
      "epoch": 0.07609236837962663,
      "grad_norm": 1.276082158088684,
      "learning_rate": 0.000184781725169094,
      "loss": 3.7438,
      "step": 81630
    },
    {
      "epoch": 0.07610168999770571,
      "grad_norm": 1.2055764198303223,
      "learning_rate": 0.00018477986084699884,
      "loss": 4.0039,
      "step": 81640
    },
    {
      "epoch": 0.0761110116157848,
      "grad_norm": 1.1157405376434326,
      "learning_rate": 0.00018477799652490362,
      "loss": 3.7359,
      "step": 81650
    },
    {
      "epoch": 0.0761203332338639,
      "grad_norm": 1.0692774057388306,
      "learning_rate": 0.00018477613220280843,
      "loss": 3.704,
      "step": 81660
    },
    {
      "epoch": 0.076129654851943,
      "grad_norm": 0.9969778060913086,
      "learning_rate": 0.0001847742678807132,
      "loss": 3.8928,
      "step": 81670
    },
    {
      "epoch": 0.07613897647002209,
      "grad_norm": 1.1387583017349243,
      "learning_rate": 0.000184772403558618,
      "loss": 3.8776,
      "step": 81680
    },
    {
      "epoch": 0.07614829808810118,
      "grad_norm": 1.189937949180603,
      "learning_rate": 0.00018477053923652282,
      "loss": 3.837,
      "step": 81690
    },
    {
      "epoch": 0.07615761970618026,
      "grad_norm": 1.1320583820343018,
      "learning_rate": 0.00018476867491442763,
      "loss": 3.824,
      "step": 81700
    },
    {
      "epoch": 0.07616694132425936,
      "grad_norm": 1.0026909112930298,
      "learning_rate": 0.00018476681059233243,
      "loss": 3.9657,
      "step": 81710
    },
    {
      "epoch": 0.07617626294233845,
      "grad_norm": 0.9766179919242859,
      "learning_rate": 0.0001847649462702372,
      "loss": 3.8248,
      "step": 81720
    },
    {
      "epoch": 0.07618558456041755,
      "grad_norm": 1.0150718688964844,
      "learning_rate": 0.00018476308194814202,
      "loss": 3.8872,
      "step": 81730
    },
    {
      "epoch": 0.07619490617849664,
      "grad_norm": 0.9313427209854126,
      "learning_rate": 0.00018476121762604682,
      "loss": 3.7651,
      "step": 81740
    },
    {
      "epoch": 0.07620422779657572,
      "grad_norm": 1.0681241750717163,
      "learning_rate": 0.00018475935330395163,
      "loss": 3.7432,
      "step": 81750
    },
    {
      "epoch": 0.07621354941465482,
      "grad_norm": 1.0048919916152954,
      "learning_rate": 0.00018475748898185644,
      "loss": 3.6496,
      "step": 81760
    },
    {
      "epoch": 0.07622287103273391,
      "grad_norm": 1.1388778686523438,
      "learning_rate": 0.00018475562465976122,
      "loss": 3.8269,
      "step": 81770
    },
    {
      "epoch": 0.076232192650813,
      "grad_norm": 0.8937950730323792,
      "learning_rate": 0.00018475376033766602,
      "loss": 3.7908,
      "step": 81780
    },
    {
      "epoch": 0.0762415142688921,
      "grad_norm": 0.959383487701416,
      "learning_rate": 0.00018475189601557083,
      "loss": 3.6587,
      "step": 81790
    },
    {
      "epoch": 0.07625083588697118,
      "grad_norm": 0.9216596484184265,
      "learning_rate": 0.00018475003169347563,
      "loss": 3.59,
      "step": 81800
    },
    {
      "epoch": 0.07626015750505027,
      "grad_norm": 0.8696221709251404,
      "learning_rate": 0.00018474816737138044,
      "loss": 3.8083,
      "step": 81810
    },
    {
      "epoch": 0.07626947912312937,
      "grad_norm": 0.8121433854103088,
      "learning_rate": 0.00018474630304928522,
      "loss": 3.7617,
      "step": 81820
    },
    {
      "epoch": 0.07627880074120846,
      "grad_norm": 1.0324392318725586,
      "learning_rate": 0.00018474443872719003,
      "loss": 3.7238,
      "step": 81830
    },
    {
      "epoch": 0.07628812235928756,
      "grad_norm": 1.0139625072479248,
      "learning_rate": 0.00018474257440509483,
      "loss": 3.9096,
      "step": 81840
    },
    {
      "epoch": 0.07629744397736664,
      "grad_norm": 1.0568755865097046,
      "learning_rate": 0.0001847407100829996,
      "loss": 3.7859,
      "step": 81850
    },
    {
      "epoch": 0.07630676559544573,
      "grad_norm": 1.1563842296600342,
      "learning_rate": 0.00018473884576090445,
      "loss": 3.8665,
      "step": 81860
    },
    {
      "epoch": 0.07631608721352483,
      "grad_norm": 1.1449053287506104,
      "learning_rate": 0.00018473698143880922,
      "loss": 3.9066,
      "step": 81870
    },
    {
      "epoch": 0.07632540883160392,
      "grad_norm": 0.9798767566680908,
      "learning_rate": 0.00018473511711671403,
      "loss": 3.7692,
      "step": 81880
    },
    {
      "epoch": 0.07633473044968302,
      "grad_norm": 1.233120083808899,
      "learning_rate": 0.00018473325279461884,
      "loss": 3.6264,
      "step": 81890
    },
    {
      "epoch": 0.07634405206776211,
      "grad_norm": 0.9817606806755066,
      "learning_rate": 0.00018473138847252362,
      "loss": 3.877,
      "step": 81900
    },
    {
      "epoch": 0.07635337368584119,
      "grad_norm": 1.203526258468628,
      "learning_rate": 0.00018472952415042842,
      "loss": 3.645,
      "step": 81910
    },
    {
      "epoch": 0.07636269530392029,
      "grad_norm": 1.0323189496994019,
      "learning_rate": 0.00018472765982833323,
      "loss": 3.8461,
      "step": 81920
    },
    {
      "epoch": 0.07637201692199938,
      "grad_norm": 1.0935488939285278,
      "learning_rate": 0.00018472579550623804,
      "loss": 3.7338,
      "step": 81930
    },
    {
      "epoch": 0.07638133854007848,
      "grad_norm": 1.086245059967041,
      "learning_rate": 0.00018472393118414284,
      "loss": 3.9014,
      "step": 81940
    },
    {
      "epoch": 0.07639066015815757,
      "grad_norm": 1.1343306303024292,
      "learning_rate": 0.00018472206686204762,
      "loss": 3.9386,
      "step": 81950
    },
    {
      "epoch": 0.07639998177623665,
      "grad_norm": 1.137360692024231,
      "learning_rate": 0.00018472020253995243,
      "loss": 3.7211,
      "step": 81960
    },
    {
      "epoch": 0.07640930339431574,
      "grad_norm": 1.1393625736236572,
      "learning_rate": 0.00018471833821785723,
      "loss": 3.8156,
      "step": 81970
    },
    {
      "epoch": 0.07641862501239484,
      "grad_norm": 1.0153985023498535,
      "learning_rate": 0.00018471647389576204,
      "loss": 3.8093,
      "step": 81980
    },
    {
      "epoch": 0.07642794663047393,
      "grad_norm": 1.0918819904327393,
      "learning_rate": 0.00018471460957366685,
      "loss": 3.724,
      "step": 81990
    },
    {
      "epoch": 0.07643726824855303,
      "grad_norm": 1.140449047088623,
      "learning_rate": 0.00018471274525157163,
      "loss": 3.7963,
      "step": 82000
    },
    {
      "epoch": 0.07644658986663211,
      "grad_norm": 1.0606369972229004,
      "learning_rate": 0.00018471088092947643,
      "loss": 3.8419,
      "step": 82010
    },
    {
      "epoch": 0.0764559114847112,
      "grad_norm": 0.9769332408905029,
      "learning_rate": 0.00018470901660738124,
      "loss": 3.8584,
      "step": 82020
    },
    {
      "epoch": 0.0764652331027903,
      "grad_norm": 1.1222163438796997,
      "learning_rate": 0.00018470715228528604,
      "loss": 3.6573,
      "step": 82030
    },
    {
      "epoch": 0.07647455472086939,
      "grad_norm": 1.1399502754211426,
      "learning_rate": 0.00018470528796319085,
      "loss": 3.8038,
      "step": 82040
    },
    {
      "epoch": 0.07648387633894849,
      "grad_norm": 0.9271553754806519,
      "learning_rate": 0.00018470342364109563,
      "loss": 3.7956,
      "step": 82050
    },
    {
      "epoch": 0.07649319795702757,
      "grad_norm": 1.0047013759613037,
      "learning_rate": 0.00018470155931900044,
      "loss": 3.5743,
      "step": 82060
    },
    {
      "epoch": 0.07650251957510666,
      "grad_norm": 1.185818076133728,
      "learning_rate": 0.00018469969499690522,
      "loss": 3.3998,
      "step": 82070
    },
    {
      "epoch": 0.07651184119318576,
      "grad_norm": 1.1058740615844727,
      "learning_rate": 0.00018469783067481002,
      "loss": 3.8265,
      "step": 82080
    },
    {
      "epoch": 0.07652116281126485,
      "grad_norm": 1.0752984285354614,
      "learning_rate": 0.00018469596635271485,
      "loss": 3.6688,
      "step": 82090
    },
    {
      "epoch": 0.07653048442934395,
      "grad_norm": 1.2172366380691528,
      "learning_rate": 0.00018469410203061963,
      "loss": 3.8,
      "step": 82100
    },
    {
      "epoch": 0.07653980604742304,
      "grad_norm": 0.9159899353981018,
      "learning_rate": 0.00018469223770852444,
      "loss": 3.8875,
      "step": 82110
    },
    {
      "epoch": 0.07654912766550212,
      "grad_norm": 1.1713062524795532,
      "learning_rate": 0.00018469037338642922,
      "loss": 3.7389,
      "step": 82120
    },
    {
      "epoch": 0.07655844928358121,
      "grad_norm": 1.0846364498138428,
      "learning_rate": 0.00018468850906433403,
      "loss": 3.8226,
      "step": 82130
    },
    {
      "epoch": 0.07656777090166031,
      "grad_norm": 0.9705126881599426,
      "learning_rate": 0.00018468664474223883,
      "loss": 3.8375,
      "step": 82140
    },
    {
      "epoch": 0.0765770925197394,
      "grad_norm": 1.3263853788375854,
      "learning_rate": 0.00018468478042014364,
      "loss": 3.9158,
      "step": 82150
    },
    {
      "epoch": 0.0765864141378185,
      "grad_norm": 1.090998888015747,
      "learning_rate": 0.00018468291609804844,
      "loss": 3.9062,
      "step": 82160
    },
    {
      "epoch": 0.07659573575589758,
      "grad_norm": 1.1491022109985352,
      "learning_rate": 0.00018468105177595322,
      "loss": 3.7184,
      "step": 82170
    },
    {
      "epoch": 0.07660505737397667,
      "grad_norm": 1.1884344816207886,
      "learning_rate": 0.00018467918745385803,
      "loss": 3.8377,
      "step": 82180
    },
    {
      "epoch": 0.07661437899205577,
      "grad_norm": 1.0013808012008667,
      "learning_rate": 0.00018467732313176284,
      "loss": 3.9024,
      "step": 82190
    },
    {
      "epoch": 0.07662370061013486,
      "grad_norm": 1.1183732748031616,
      "learning_rate": 0.00018467545880966764,
      "loss": 3.8264,
      "step": 82200
    },
    {
      "epoch": 0.07663302222821396,
      "grad_norm": 1.0634775161743164,
      "learning_rate": 0.00018467359448757245,
      "loss": 3.8429,
      "step": 82210
    },
    {
      "epoch": 0.07664234384629304,
      "grad_norm": 0.9917502999305725,
      "learning_rate": 0.00018467173016547723,
      "loss": 3.8584,
      "step": 82220
    },
    {
      "epoch": 0.07665166546437213,
      "grad_norm": 1.0995941162109375,
      "learning_rate": 0.00018466986584338203,
      "loss": 3.6659,
      "step": 82230
    },
    {
      "epoch": 0.07666098708245123,
      "grad_norm": 1.061526894569397,
      "learning_rate": 0.00018466800152128684,
      "loss": 3.8457,
      "step": 82240
    },
    {
      "epoch": 0.07667030870053032,
      "grad_norm": 1.0357617139816284,
      "learning_rate": 0.00018466613719919165,
      "loss": 3.7252,
      "step": 82250
    },
    {
      "epoch": 0.07667963031860942,
      "grad_norm": 1.0006781816482544,
      "learning_rate": 0.00018466427287709645,
      "loss": 3.7834,
      "step": 82260
    },
    {
      "epoch": 0.0766889519366885,
      "grad_norm": 0.9462762475013733,
      "learning_rate": 0.00018466240855500123,
      "loss": 3.6459,
      "step": 82270
    },
    {
      "epoch": 0.07669827355476759,
      "grad_norm": 0.9825459718704224,
      "learning_rate": 0.00018466054423290604,
      "loss": 3.9573,
      "step": 82280
    },
    {
      "epoch": 0.07670759517284668,
      "grad_norm": 0.8484278917312622,
      "learning_rate": 0.00018465867991081085,
      "loss": 3.751,
      "step": 82290
    },
    {
      "epoch": 0.07671691679092578,
      "grad_norm": 1.2772327661514282,
      "learning_rate": 0.00018465681558871562,
      "loss": 3.8286,
      "step": 82300
    },
    {
      "epoch": 0.07672623840900487,
      "grad_norm": 0.9295668601989746,
      "learning_rate": 0.00018465495126662046,
      "loss": 3.7945,
      "step": 82310
    },
    {
      "epoch": 0.07673556002708395,
      "grad_norm": 0.8396241664886475,
      "learning_rate": 0.00018465308694452524,
      "loss": 3.7901,
      "step": 82320
    },
    {
      "epoch": 0.07674488164516305,
      "grad_norm": 0.9066523909568787,
      "learning_rate": 0.00018465122262243004,
      "loss": 3.8704,
      "step": 82330
    },
    {
      "epoch": 0.07675420326324214,
      "grad_norm": 1.2796574831008911,
      "learning_rate": 0.00018464935830033485,
      "loss": 3.6455,
      "step": 82340
    },
    {
      "epoch": 0.07676352488132124,
      "grad_norm": 1.200010061264038,
      "learning_rate": 0.00018464749397823963,
      "loss": 3.86,
      "step": 82350
    },
    {
      "epoch": 0.07677284649940033,
      "grad_norm": 1.2699658870697021,
      "learning_rate": 0.00018464562965614444,
      "loss": 3.4975,
      "step": 82360
    },
    {
      "epoch": 0.07678216811747943,
      "grad_norm": 1.0025526285171509,
      "learning_rate": 0.00018464376533404924,
      "loss": 3.9241,
      "step": 82370
    },
    {
      "epoch": 0.07679148973555851,
      "grad_norm": 1.1656056642532349,
      "learning_rate": 0.00018464190101195405,
      "loss": 3.7816,
      "step": 82380
    },
    {
      "epoch": 0.0768008113536376,
      "grad_norm": 0.8784066438674927,
      "learning_rate": 0.00018464003668985885,
      "loss": 3.7702,
      "step": 82390
    },
    {
      "epoch": 0.0768101329717167,
      "grad_norm": 1.170283317565918,
      "learning_rate": 0.00018463817236776363,
      "loss": 3.7718,
      "step": 82400
    },
    {
      "epoch": 0.07681945458979579,
      "grad_norm": 1.1899445056915283,
      "learning_rate": 0.00018463630804566844,
      "loss": 3.8015,
      "step": 82410
    },
    {
      "epoch": 0.07682877620787489,
      "grad_norm": 1.1662200689315796,
      "learning_rate": 0.00018463444372357325,
      "loss": 3.8053,
      "step": 82420
    },
    {
      "epoch": 0.07683809782595397,
      "grad_norm": 1.1026110649108887,
      "learning_rate": 0.00018463257940147805,
      "loss": 3.8317,
      "step": 82430
    },
    {
      "epoch": 0.07684741944403306,
      "grad_norm": 1.0150610208511353,
      "learning_rate": 0.00018463071507938286,
      "loss": 3.838,
      "step": 82440
    },
    {
      "epoch": 0.07685674106211215,
      "grad_norm": 1.0849363803863525,
      "learning_rate": 0.00018462885075728764,
      "loss": 4.0155,
      "step": 82450
    },
    {
      "epoch": 0.07686606268019125,
      "grad_norm": 1.0585051774978638,
      "learning_rate": 0.00018462698643519244,
      "loss": 3.8556,
      "step": 82460
    },
    {
      "epoch": 0.07687538429827034,
      "grad_norm": 1.0403252840042114,
      "learning_rate": 0.00018462512211309725,
      "loss": 3.904,
      "step": 82470
    },
    {
      "epoch": 0.07688470591634942,
      "grad_norm": 0.9398555755615234,
      "learning_rate": 0.00018462325779100206,
      "loss": 3.9799,
      "step": 82480
    },
    {
      "epoch": 0.07689402753442852,
      "grad_norm": 1.0602149963378906,
      "learning_rate": 0.00018462139346890686,
      "loss": 3.6707,
      "step": 82490
    },
    {
      "epoch": 0.07690334915250761,
      "grad_norm": 1.0258022546768188,
      "learning_rate": 0.00018461952914681164,
      "loss": 3.8596,
      "step": 82500
    },
    {
      "epoch": 0.07691267077058671,
      "grad_norm": 1.0714024305343628,
      "learning_rate": 0.00018461766482471645,
      "loss": 3.8043,
      "step": 82510
    },
    {
      "epoch": 0.0769219923886658,
      "grad_norm": 1.224311113357544,
      "learning_rate": 0.00018461580050262123,
      "loss": 3.9057,
      "step": 82520
    },
    {
      "epoch": 0.07693131400674488,
      "grad_norm": 0.848957359790802,
      "learning_rate": 0.00018461393618052606,
      "loss": 3.7587,
      "step": 82530
    },
    {
      "epoch": 0.07694063562482398,
      "grad_norm": 1.1829365491867065,
      "learning_rate": 0.00018461207185843087,
      "loss": 3.7295,
      "step": 82540
    },
    {
      "epoch": 0.07694995724290307,
      "grad_norm": 1.0562766790390015,
      "learning_rate": 0.00018461020753633565,
      "loss": 3.9588,
      "step": 82550
    },
    {
      "epoch": 0.07695927886098217,
      "grad_norm": 1.0925438404083252,
      "learning_rate": 0.00018460834321424045,
      "loss": 3.711,
      "step": 82560
    },
    {
      "epoch": 0.07696860047906126,
      "grad_norm": 0.9277344942092896,
      "learning_rate": 0.00018460647889214523,
      "loss": 3.8089,
      "step": 82570
    },
    {
      "epoch": 0.07697792209714036,
      "grad_norm": 1.1087814569473267,
      "learning_rate": 0.00018460461457005004,
      "loss": 3.9075,
      "step": 82580
    },
    {
      "epoch": 0.07698724371521944,
      "grad_norm": 1.0713386535644531,
      "learning_rate": 0.00018460275024795485,
      "loss": 3.6695,
      "step": 82590
    },
    {
      "epoch": 0.07699656533329853,
      "grad_norm": 1.3583678007125854,
      "learning_rate": 0.00018460088592585965,
      "loss": 3.6327,
      "step": 82600
    },
    {
      "epoch": 0.07700588695137763,
      "grad_norm": 0.8855668306350708,
      "learning_rate": 0.00018459902160376446,
      "loss": 3.8414,
      "step": 82610
    },
    {
      "epoch": 0.07701520856945672,
      "grad_norm": 1.173387050628662,
      "learning_rate": 0.00018459715728166924,
      "loss": 3.776,
      "step": 82620
    },
    {
      "epoch": 0.07702453018753581,
      "grad_norm": 1.0329679250717163,
      "learning_rate": 0.00018459529295957404,
      "loss": 3.7916,
      "step": 82630
    },
    {
      "epoch": 0.0770338518056149,
      "grad_norm": 0.9592461585998535,
      "learning_rate": 0.00018459342863747885,
      "loss": 3.7877,
      "step": 82640
    },
    {
      "epoch": 0.07704317342369399,
      "grad_norm": 1.1139922142028809,
      "learning_rate": 0.00018459156431538366,
      "loss": 3.8728,
      "step": 82650
    },
    {
      "epoch": 0.07705249504177308,
      "grad_norm": 1.1394973993301392,
      "learning_rate": 0.00018458969999328846,
      "loss": 3.7629,
      "step": 82660
    },
    {
      "epoch": 0.07706181665985218,
      "grad_norm": 1.0738658905029297,
      "learning_rate": 0.00018458783567119324,
      "loss": 3.8257,
      "step": 82670
    },
    {
      "epoch": 0.07707113827793127,
      "grad_norm": 1.1152980327606201,
      "learning_rate": 0.00018458597134909805,
      "loss": 3.7367,
      "step": 82680
    },
    {
      "epoch": 0.07708045989601035,
      "grad_norm": 1.075505018234253,
      "learning_rate": 0.00018458410702700283,
      "loss": 3.8896,
      "step": 82690
    },
    {
      "epoch": 0.07708978151408945,
      "grad_norm": 0.9099100828170776,
      "learning_rate": 0.00018458224270490766,
      "loss": 3.8072,
      "step": 82700
    },
    {
      "epoch": 0.07709910313216854,
      "grad_norm": 1.177032709121704,
      "learning_rate": 0.00018458037838281247,
      "loss": 3.7951,
      "step": 82710
    },
    {
      "epoch": 0.07710842475024764,
      "grad_norm": 0.990940511226654,
      "learning_rate": 0.00018457851406071725,
      "loss": 3.9336,
      "step": 82720
    },
    {
      "epoch": 0.07711774636832673,
      "grad_norm": 1.0345656871795654,
      "learning_rate": 0.00018457664973862205,
      "loss": 3.7114,
      "step": 82730
    },
    {
      "epoch": 0.07712706798640581,
      "grad_norm": 1.143316388130188,
      "learning_rate": 0.00018457478541652683,
      "loss": 3.6499,
      "step": 82740
    },
    {
      "epoch": 0.0771363896044849,
      "grad_norm": 0.9522451758384705,
      "learning_rate": 0.00018457292109443164,
      "loss": 3.5941,
      "step": 82750
    },
    {
      "epoch": 0.077145711222564,
      "grad_norm": 1.1332939863204956,
      "learning_rate": 0.00018457105677233647,
      "loss": 3.7146,
      "step": 82760
    },
    {
      "epoch": 0.0771550328406431,
      "grad_norm": 0.8539389967918396,
      "learning_rate": 0.00018456919245024125,
      "loss": 3.7938,
      "step": 82770
    },
    {
      "epoch": 0.07716435445872219,
      "grad_norm": 1.7532180547714233,
      "learning_rate": 0.00018456732812814606,
      "loss": 3.6746,
      "step": 82780
    },
    {
      "epoch": 0.07717367607680128,
      "grad_norm": 1.1678863763809204,
      "learning_rate": 0.00018456546380605084,
      "loss": 3.8249,
      "step": 82790
    },
    {
      "epoch": 0.07718299769488036,
      "grad_norm": 1.1942908763885498,
      "learning_rate": 0.00018456359948395564,
      "loss": 3.8288,
      "step": 82800
    },
    {
      "epoch": 0.07719231931295946,
      "grad_norm": 1.0804014205932617,
      "learning_rate": 0.00018456173516186045,
      "loss": 3.7731,
      "step": 82810
    },
    {
      "epoch": 0.07720164093103855,
      "grad_norm": 0.9704447388648987,
      "learning_rate": 0.00018455987083976525,
      "loss": 3.8882,
      "step": 82820
    },
    {
      "epoch": 0.07721096254911765,
      "grad_norm": 1.0693583488464355,
      "learning_rate": 0.00018455800651767006,
      "loss": 3.8386,
      "step": 82830
    },
    {
      "epoch": 0.07722028416719674,
      "grad_norm": 0.9776977896690369,
      "learning_rate": 0.00018455614219557484,
      "loss": 3.7891,
      "step": 82840
    },
    {
      "epoch": 0.07722960578527582,
      "grad_norm": 1.1605346202850342,
      "learning_rate": 0.00018455427787347965,
      "loss": 3.8774,
      "step": 82850
    },
    {
      "epoch": 0.07723892740335492,
      "grad_norm": 1.0404810905456543,
      "learning_rate": 0.00018455241355138445,
      "loss": 3.7019,
      "step": 82860
    },
    {
      "epoch": 0.07724824902143401,
      "grad_norm": 1.078220009803772,
      "learning_rate": 0.00018455054922928926,
      "loss": 3.8816,
      "step": 82870
    },
    {
      "epoch": 0.0772575706395131,
      "grad_norm": 1.3291469812393188,
      "learning_rate": 0.00018454868490719407,
      "loss": 3.7686,
      "step": 82880
    },
    {
      "epoch": 0.0772668922575922,
      "grad_norm": 0.939238965511322,
      "learning_rate": 0.00018454682058509884,
      "loss": 3.5463,
      "step": 82890
    },
    {
      "epoch": 0.07727621387567128,
      "grad_norm": 0.9460298418998718,
      "learning_rate": 0.00018454495626300365,
      "loss": 4.0134,
      "step": 82900
    },
    {
      "epoch": 0.07728553549375038,
      "grad_norm": 1.1445426940917969,
      "learning_rate": 0.00018454309194090846,
      "loss": 3.9135,
      "step": 82910
    },
    {
      "epoch": 0.07729485711182947,
      "grad_norm": 1.168283224105835,
      "learning_rate": 0.00018454122761881326,
      "loss": 3.7078,
      "step": 82920
    },
    {
      "epoch": 0.07730417872990857,
      "grad_norm": 1.209256887435913,
      "learning_rate": 0.00018453936329671807,
      "loss": 3.6808,
      "step": 82930
    },
    {
      "epoch": 0.07731350034798766,
      "grad_norm": 1.0783474445343018,
      "learning_rate": 0.00018453749897462285,
      "loss": 3.6449,
      "step": 82940
    },
    {
      "epoch": 0.07732282196606674,
      "grad_norm": 1.1262203454971313,
      "learning_rate": 0.00018453563465252766,
      "loss": 3.9775,
      "step": 82950
    },
    {
      "epoch": 0.07733214358414583,
      "grad_norm": 1.1140776872634888,
      "learning_rate": 0.00018453377033043246,
      "loss": 3.6846,
      "step": 82960
    },
    {
      "epoch": 0.07734146520222493,
      "grad_norm": 1.03538978099823,
      "learning_rate": 0.00018453190600833724,
      "loss": 3.7426,
      "step": 82970
    },
    {
      "epoch": 0.07735078682030402,
      "grad_norm": 0.9903056025505066,
      "learning_rate": 0.00018453004168624207,
      "loss": 4.0071,
      "step": 82980
    },
    {
      "epoch": 0.07736010843838312,
      "grad_norm": 1.2093877792358398,
      "learning_rate": 0.00018452817736414685,
      "loss": 3.5665,
      "step": 82990
    },
    {
      "epoch": 0.07736943005646221,
      "grad_norm": 0.9857003092765808,
      "learning_rate": 0.00018452631304205166,
      "loss": 3.9281,
      "step": 83000
    },
    {
      "epoch": 0.0773787516745413,
      "grad_norm": 0.9736860990524292,
      "learning_rate": 0.00018452444871995647,
      "loss": 3.8093,
      "step": 83010
    },
    {
      "epoch": 0.07738807329262039,
      "grad_norm": 1.2561994791030884,
      "learning_rate": 0.00018452258439786125,
      "loss": 3.7248,
      "step": 83020
    },
    {
      "epoch": 0.07739739491069948,
      "grad_norm": 0.9397568106651306,
      "learning_rate": 0.00018452072007576605,
      "loss": 3.8967,
      "step": 83030
    },
    {
      "epoch": 0.07740671652877858,
      "grad_norm": 1.4740463495254517,
      "learning_rate": 0.00018451885575367086,
      "loss": 3.8001,
      "step": 83040
    },
    {
      "epoch": 0.07741603814685767,
      "grad_norm": 0.9909406900405884,
      "learning_rate": 0.00018451699143157566,
      "loss": 3.8469,
      "step": 83050
    },
    {
      "epoch": 0.07742535976493675,
      "grad_norm": 1.0018956661224365,
      "learning_rate": 0.00018451512710948047,
      "loss": 3.8563,
      "step": 83060
    },
    {
      "epoch": 0.07743468138301585,
      "grad_norm": 1.2201755046844482,
      "learning_rate": 0.00018451326278738525,
      "loss": 3.7012,
      "step": 83070
    },
    {
      "epoch": 0.07744400300109494,
      "grad_norm": 0.9499194025993347,
      "learning_rate": 0.00018451139846529006,
      "loss": 3.7657,
      "step": 83080
    },
    {
      "epoch": 0.07745332461917404,
      "grad_norm": 1.223102331161499,
      "learning_rate": 0.00018450953414319486,
      "loss": 3.7179,
      "step": 83090
    },
    {
      "epoch": 0.07746264623725313,
      "grad_norm": 1.2275668382644653,
      "learning_rate": 0.00018450766982109967,
      "loss": 3.6998,
      "step": 83100
    },
    {
      "epoch": 0.07747196785533221,
      "grad_norm": 1.020716667175293,
      "learning_rate": 0.00018450580549900448,
      "loss": 3.9559,
      "step": 83110
    },
    {
      "epoch": 0.0774812894734113,
      "grad_norm": 0.980567455291748,
      "learning_rate": 0.00018450394117690925,
      "loss": 3.7069,
      "step": 83120
    },
    {
      "epoch": 0.0774906110914904,
      "grad_norm": 1.0390901565551758,
      "learning_rate": 0.00018450207685481406,
      "loss": 3.8703,
      "step": 83130
    },
    {
      "epoch": 0.0774999327095695,
      "grad_norm": 1.0549836158752441,
      "learning_rate": 0.00018450021253271884,
      "loss": 3.6445,
      "step": 83140
    },
    {
      "epoch": 0.07750925432764859,
      "grad_norm": 0.8757666945457458,
      "learning_rate": 0.00018449834821062367,
      "loss": 3.9184,
      "step": 83150
    },
    {
      "epoch": 0.07751857594572767,
      "grad_norm": 1.0870500802993774,
      "learning_rate": 0.00018449648388852848,
      "loss": 3.8127,
      "step": 83160
    },
    {
      "epoch": 0.07752789756380676,
      "grad_norm": 1.1250442266464233,
      "learning_rate": 0.00018449461956643326,
      "loss": 3.7685,
      "step": 83170
    },
    {
      "epoch": 0.07753721918188586,
      "grad_norm": 0.8933490514755249,
      "learning_rate": 0.00018449275524433807,
      "loss": 3.9405,
      "step": 83180
    },
    {
      "epoch": 0.07754654079996495,
      "grad_norm": 1.043131947517395,
      "learning_rate": 0.00018449089092224284,
      "loss": 3.8245,
      "step": 83190
    },
    {
      "epoch": 0.07755586241804405,
      "grad_norm": 1.2685271501541138,
      "learning_rate": 0.00018448902660014765,
      "loss": 3.8927,
      "step": 83200
    },
    {
      "epoch": 0.07756518403612313,
      "grad_norm": 1.2173686027526855,
      "learning_rate": 0.00018448716227805248,
      "loss": 3.8615,
      "step": 83210
    },
    {
      "epoch": 0.07757450565420222,
      "grad_norm": 1.1569362878799438,
      "learning_rate": 0.00018448529795595726,
      "loss": 3.9364,
      "step": 83220
    },
    {
      "epoch": 0.07758382727228132,
      "grad_norm": 1.0879191160202026,
      "learning_rate": 0.00018448343363386207,
      "loss": 3.9954,
      "step": 83230
    },
    {
      "epoch": 0.07759314889036041,
      "grad_norm": 0.9791191220283508,
      "learning_rate": 0.00018448156931176685,
      "loss": 3.5718,
      "step": 83240
    },
    {
      "epoch": 0.0776024705084395,
      "grad_norm": 1.0909606218338013,
      "learning_rate": 0.00018447970498967165,
      "loss": 3.9215,
      "step": 83250
    },
    {
      "epoch": 0.0776117921265186,
      "grad_norm": 1.0747729539871216,
      "learning_rate": 0.00018447784066757646,
      "loss": 3.5656,
      "step": 83260
    },
    {
      "epoch": 0.07762111374459768,
      "grad_norm": 1.2237269878387451,
      "learning_rate": 0.00018447597634548127,
      "loss": 3.6636,
      "step": 83270
    },
    {
      "epoch": 0.07763043536267678,
      "grad_norm": 1.102833867073059,
      "learning_rate": 0.00018447411202338607,
      "loss": 3.9343,
      "step": 83280
    },
    {
      "epoch": 0.07763975698075587,
      "grad_norm": 1.1330662965774536,
      "learning_rate": 0.00018447224770129085,
      "loss": 3.7443,
      "step": 83290
    },
    {
      "epoch": 0.07764907859883496,
      "grad_norm": 1.0868446826934814,
      "learning_rate": 0.00018447038337919566,
      "loss": 3.8113,
      "step": 83300
    },
    {
      "epoch": 0.07765840021691406,
      "grad_norm": 1.2307798862457275,
      "learning_rate": 0.00018446851905710047,
      "loss": 3.8604,
      "step": 83310
    },
    {
      "epoch": 0.07766772183499314,
      "grad_norm": 1.1191980838775635,
      "learning_rate": 0.00018446665473500527,
      "loss": 3.7087,
      "step": 83320
    },
    {
      "epoch": 0.07767704345307223,
      "grad_norm": 0.9746212363243103,
      "learning_rate": 0.00018446479041291008,
      "loss": 3.8264,
      "step": 83330
    },
    {
      "epoch": 0.07768636507115133,
      "grad_norm": 0.9947558641433716,
      "learning_rate": 0.00018446292609081486,
      "loss": 3.8742,
      "step": 83340
    },
    {
      "epoch": 0.07769568668923042,
      "grad_norm": 0.955352246761322,
      "learning_rate": 0.00018446106176871966,
      "loss": 3.8568,
      "step": 83350
    },
    {
      "epoch": 0.07770500830730952,
      "grad_norm": 0.9752621054649353,
      "learning_rate": 0.00018445919744662447,
      "loss": 4.0027,
      "step": 83360
    },
    {
      "epoch": 0.0777143299253886,
      "grad_norm": 1.0063693523406982,
      "learning_rate": 0.00018445733312452928,
      "loss": 3.828,
      "step": 83370
    },
    {
      "epoch": 0.07772365154346769,
      "grad_norm": 1.0937823057174683,
      "learning_rate": 0.00018445546880243408,
      "loss": 3.7083,
      "step": 83380
    },
    {
      "epoch": 0.07773297316154679,
      "grad_norm": 1.289968729019165,
      "learning_rate": 0.00018445360448033886,
      "loss": 3.8871,
      "step": 83390
    },
    {
      "epoch": 0.07774229477962588,
      "grad_norm": 1.117551565170288,
      "learning_rate": 0.00018445174015824367,
      "loss": 3.7086,
      "step": 83400
    },
    {
      "epoch": 0.07775161639770498,
      "grad_norm": 0.8663129210472107,
      "learning_rate": 0.00018444987583614847,
      "loss": 3.987,
      "step": 83410
    },
    {
      "epoch": 0.07776093801578406,
      "grad_norm": 0.973722517490387,
      "learning_rate": 0.00018444801151405325,
      "loss": 3.8899,
      "step": 83420
    },
    {
      "epoch": 0.07777025963386315,
      "grad_norm": 1.052068829536438,
      "learning_rate": 0.0001844461471919581,
      "loss": 3.7378,
      "step": 83430
    },
    {
      "epoch": 0.07777958125194225,
      "grad_norm": 1.1091748476028442,
      "learning_rate": 0.00018444428286986287,
      "loss": 3.7612,
      "step": 83440
    },
    {
      "epoch": 0.07778890287002134,
      "grad_norm": 0.8964629769325256,
      "learning_rate": 0.00018444241854776767,
      "loss": 3.8929,
      "step": 83450
    },
    {
      "epoch": 0.07779822448810043,
      "grad_norm": 1.17514169216156,
      "learning_rate": 0.00018444055422567248,
      "loss": 3.922,
      "step": 83460
    },
    {
      "epoch": 0.07780754610617953,
      "grad_norm": 1.1588834524154663,
      "learning_rate": 0.00018443868990357726,
      "loss": 3.9325,
      "step": 83470
    },
    {
      "epoch": 0.07781686772425861,
      "grad_norm": 0.9310903549194336,
      "learning_rate": 0.00018443682558148206,
      "loss": 3.7331,
      "step": 83480
    },
    {
      "epoch": 0.0778261893423377,
      "grad_norm": 1.114920735359192,
      "learning_rate": 0.00018443496125938687,
      "loss": 3.8743,
      "step": 83490
    },
    {
      "epoch": 0.0778355109604168,
      "grad_norm": 1.1309170722961426,
      "learning_rate": 0.00018443309693729168,
      "loss": 3.763,
      "step": 83500
    },
    {
      "epoch": 0.07784483257849589,
      "grad_norm": 1.1602767705917358,
      "learning_rate": 0.00018443123261519648,
      "loss": 3.8981,
      "step": 83510
    },
    {
      "epoch": 0.07785415419657499,
      "grad_norm": 1.0000767707824707,
      "learning_rate": 0.00018442936829310126,
      "loss": 3.8413,
      "step": 83520
    },
    {
      "epoch": 0.07786347581465407,
      "grad_norm": 1.077262282371521,
      "learning_rate": 0.00018442750397100607,
      "loss": 3.8324,
      "step": 83530
    },
    {
      "epoch": 0.07787279743273316,
      "grad_norm": 0.9876661896705627,
      "learning_rate": 0.00018442563964891088,
      "loss": 3.7575,
      "step": 83540
    },
    {
      "epoch": 0.07788211905081226,
      "grad_norm": 1.1267834901809692,
      "learning_rate": 0.00018442377532681568,
      "loss": 3.6803,
      "step": 83550
    },
    {
      "epoch": 0.07789144066889135,
      "grad_norm": 0.9661248922348022,
      "learning_rate": 0.0001844219110047205,
      "loss": 3.5887,
      "step": 83560
    },
    {
      "epoch": 0.07790076228697045,
      "grad_norm": 1.0349534749984741,
      "learning_rate": 0.00018442004668262527,
      "loss": 3.8174,
      "step": 83570
    },
    {
      "epoch": 0.07791008390504953,
      "grad_norm": 1.061428427696228,
      "learning_rate": 0.00018441818236053007,
      "loss": 3.6491,
      "step": 83580
    },
    {
      "epoch": 0.07791940552312862,
      "grad_norm": 1.2801928520202637,
      "learning_rate": 0.00018441631803843485,
      "loss": 3.7174,
      "step": 83590
    },
    {
      "epoch": 0.07792872714120772,
      "grad_norm": 1.2004503011703491,
      "learning_rate": 0.00018441445371633969,
      "loss": 3.698,
      "step": 83600
    },
    {
      "epoch": 0.07793804875928681,
      "grad_norm": 1.1670796871185303,
      "learning_rate": 0.0001844125893942445,
      "loss": 4.0375,
      "step": 83610
    },
    {
      "epoch": 0.0779473703773659,
      "grad_norm": 1.1843119859695435,
      "learning_rate": 0.00018441072507214927,
      "loss": 3.6355,
      "step": 83620
    },
    {
      "epoch": 0.07795669199544498,
      "grad_norm": 1.124712586402893,
      "learning_rate": 0.00018440886075005408,
      "loss": 3.8134,
      "step": 83630
    },
    {
      "epoch": 0.07796601361352408,
      "grad_norm": 0.9402591586112976,
      "learning_rate": 0.00018440699642795886,
      "loss": 3.7843,
      "step": 83640
    },
    {
      "epoch": 0.07797533523160317,
      "grad_norm": 0.9671542048454285,
      "learning_rate": 0.00018440513210586366,
      "loss": 3.8425,
      "step": 83650
    },
    {
      "epoch": 0.07798465684968227,
      "grad_norm": 0.9955025911331177,
      "learning_rate": 0.0001844032677837685,
      "loss": 3.8548,
      "step": 83660
    },
    {
      "epoch": 0.07799397846776136,
      "grad_norm": 1.089454174041748,
      "learning_rate": 0.00018440140346167328,
      "loss": 4.0454,
      "step": 83670
    },
    {
      "epoch": 0.07800330008584046,
      "grad_norm": 1.0798804759979248,
      "learning_rate": 0.00018439953913957808,
      "loss": 3.6904,
      "step": 83680
    },
    {
      "epoch": 0.07801262170391954,
      "grad_norm": 1.036024808883667,
      "learning_rate": 0.00018439767481748286,
      "loss": 3.7521,
      "step": 83690
    },
    {
      "epoch": 0.07802194332199863,
      "grad_norm": 1.0858783721923828,
      "learning_rate": 0.00018439581049538767,
      "loss": 3.7765,
      "step": 83700
    },
    {
      "epoch": 0.07803126494007773,
      "grad_norm": 1.2739440202713013,
      "learning_rate": 0.00018439394617329247,
      "loss": 3.7518,
      "step": 83710
    },
    {
      "epoch": 0.07804058655815682,
      "grad_norm": 0.8782214522361755,
      "learning_rate": 0.00018439208185119728,
      "loss": 3.8244,
      "step": 83720
    },
    {
      "epoch": 0.07804990817623592,
      "grad_norm": 1.1673282384872437,
      "learning_rate": 0.0001843902175291021,
      "loss": 3.8246,
      "step": 83730
    },
    {
      "epoch": 0.078059229794315,
      "grad_norm": 0.9973869323730469,
      "learning_rate": 0.00018438835320700687,
      "loss": 3.7883,
      "step": 83740
    },
    {
      "epoch": 0.07806855141239409,
      "grad_norm": 1.2040884494781494,
      "learning_rate": 0.00018438648888491167,
      "loss": 3.7088,
      "step": 83750
    },
    {
      "epoch": 0.07807787303047319,
      "grad_norm": 1.217732310295105,
      "learning_rate": 0.00018438462456281648,
      "loss": 3.7914,
      "step": 83760
    },
    {
      "epoch": 0.07808719464855228,
      "grad_norm": 1.115689992904663,
      "learning_rate": 0.00018438276024072128,
      "loss": 3.7946,
      "step": 83770
    },
    {
      "epoch": 0.07809651626663137,
      "grad_norm": 0.9594476222991943,
      "learning_rate": 0.0001843808959186261,
      "loss": 3.8298,
      "step": 83780
    },
    {
      "epoch": 0.07810583788471046,
      "grad_norm": 0.9558746814727783,
      "learning_rate": 0.00018437903159653087,
      "loss": 3.7595,
      "step": 83790
    },
    {
      "epoch": 0.07811515950278955,
      "grad_norm": 1.2413618564605713,
      "learning_rate": 0.00018437716727443568,
      "loss": 3.7228,
      "step": 83800
    },
    {
      "epoch": 0.07812448112086864,
      "grad_norm": 1.1685360670089722,
      "learning_rate": 0.00018437530295234048,
      "loss": 3.9937,
      "step": 83810
    },
    {
      "epoch": 0.07813380273894774,
      "grad_norm": 1.084624171257019,
      "learning_rate": 0.0001843734386302453,
      "loss": 3.6333,
      "step": 83820
    },
    {
      "epoch": 0.07814312435702683,
      "grad_norm": 0.876524031162262,
      "learning_rate": 0.0001843715743081501,
      "loss": 3.6948,
      "step": 83830
    },
    {
      "epoch": 0.07815244597510591,
      "grad_norm": 1.0846539735794067,
      "learning_rate": 0.00018436970998605487,
      "loss": 3.7642,
      "step": 83840
    },
    {
      "epoch": 0.07816176759318501,
      "grad_norm": 0.8976156711578369,
      "learning_rate": 0.00018436784566395968,
      "loss": 3.8905,
      "step": 83850
    },
    {
      "epoch": 0.0781710892112641,
      "grad_norm": 1.3651658296585083,
      "learning_rate": 0.0001843659813418645,
      "loss": 3.8483,
      "step": 83860
    },
    {
      "epoch": 0.0781804108293432,
      "grad_norm": 0.9962701201438904,
      "learning_rate": 0.00018436411701976927,
      "loss": 3.8149,
      "step": 83870
    },
    {
      "epoch": 0.07818973244742229,
      "grad_norm": 1.0372060537338257,
      "learning_rate": 0.0001843622526976741,
      "loss": 3.9489,
      "step": 83880
    },
    {
      "epoch": 0.07819905406550137,
      "grad_norm": 0.8550460934638977,
      "learning_rate": 0.00018436038837557888,
      "loss": 3.6806,
      "step": 83890
    },
    {
      "epoch": 0.07820837568358047,
      "grad_norm": 1.1010459661483765,
      "learning_rate": 0.00018435852405348369,
      "loss": 3.7471,
      "step": 83900
    },
    {
      "epoch": 0.07821769730165956,
      "grad_norm": 1.222216010093689,
      "learning_rate": 0.0001843566597313885,
      "loss": 3.7684,
      "step": 83910
    },
    {
      "epoch": 0.07822701891973866,
      "grad_norm": 1.0359300374984741,
      "learning_rate": 0.00018435479540929327,
      "loss": 3.8087,
      "step": 83920
    },
    {
      "epoch": 0.07823634053781775,
      "grad_norm": 1.158438801765442,
      "learning_rate": 0.00018435293108719808,
      "loss": 3.7037,
      "step": 83930
    },
    {
      "epoch": 0.07824566215589684,
      "grad_norm": 1.037607192993164,
      "learning_rate": 0.00018435106676510288,
      "loss": 3.782,
      "step": 83940
    },
    {
      "epoch": 0.07825498377397593,
      "grad_norm": 1.265540599822998,
      "learning_rate": 0.0001843492024430077,
      "loss": 3.9194,
      "step": 83950
    },
    {
      "epoch": 0.07826430539205502,
      "grad_norm": 1.198856234550476,
      "learning_rate": 0.00018434733812091247,
      "loss": 3.74,
      "step": 83960
    },
    {
      "epoch": 0.07827362701013411,
      "grad_norm": 1.1492429971694946,
      "learning_rate": 0.00018434547379881728,
      "loss": 3.8561,
      "step": 83970
    },
    {
      "epoch": 0.07828294862821321,
      "grad_norm": 1.0471323728561401,
      "learning_rate": 0.00018434360947672208,
      "loss": 3.9762,
      "step": 83980
    },
    {
      "epoch": 0.0782922702462923,
      "grad_norm": 1.134183645248413,
      "learning_rate": 0.0001843417451546269,
      "loss": 3.8519,
      "step": 83990
    },
    {
      "epoch": 0.07830159186437138,
      "grad_norm": 0.8708364367485046,
      "learning_rate": 0.0001843398808325317,
      "loss": 3.9412,
      "step": 84000
    },
    {
      "epoch": 0.07831091348245048,
      "grad_norm": 1.114454984664917,
      "learning_rate": 0.00018433801651043647,
      "loss": 3.7367,
      "step": 84010
    },
    {
      "epoch": 0.07832023510052957,
      "grad_norm": 0.9955729246139526,
      "learning_rate": 0.00018433615218834128,
      "loss": 3.6781,
      "step": 84020
    },
    {
      "epoch": 0.07832955671860867,
      "grad_norm": 1.145112156867981,
      "learning_rate": 0.00018433428786624609,
      "loss": 3.8147,
      "step": 84030
    },
    {
      "epoch": 0.07833887833668776,
      "grad_norm": 1.0517252683639526,
      "learning_rate": 0.00018433242354415087,
      "loss": 3.8061,
      "step": 84040
    },
    {
      "epoch": 0.07834819995476684,
      "grad_norm": 1.011615514755249,
      "learning_rate": 0.0001843305592220557,
      "loss": 3.8707,
      "step": 84050
    },
    {
      "epoch": 0.07835752157284594,
      "grad_norm": 1.2720999717712402,
      "learning_rate": 0.00018432869489996048,
      "loss": 3.5659,
      "step": 84060
    },
    {
      "epoch": 0.07836684319092503,
      "grad_norm": 1.0990231037139893,
      "learning_rate": 0.00018432683057786528,
      "loss": 3.7989,
      "step": 84070
    },
    {
      "epoch": 0.07837616480900413,
      "grad_norm": 1.1245120763778687,
      "learning_rate": 0.0001843249662557701,
      "loss": 3.6851,
      "step": 84080
    },
    {
      "epoch": 0.07838548642708322,
      "grad_norm": 1.1679607629776,
      "learning_rate": 0.00018432310193367487,
      "loss": 3.7562,
      "step": 84090
    },
    {
      "epoch": 0.0783948080451623,
      "grad_norm": 1.1611220836639404,
      "learning_rate": 0.00018432123761157968,
      "loss": 3.829,
      "step": 84100
    },
    {
      "epoch": 0.0784041296632414,
      "grad_norm": 0.9529655575752258,
      "learning_rate": 0.00018431937328948448,
      "loss": 3.7697,
      "step": 84110
    },
    {
      "epoch": 0.07841345128132049,
      "grad_norm": 1.2296348810195923,
      "learning_rate": 0.0001843175089673893,
      "loss": 3.5397,
      "step": 84120
    },
    {
      "epoch": 0.07842277289939958,
      "grad_norm": 1.0253795385360718,
      "learning_rate": 0.0001843156446452941,
      "loss": 3.6764,
      "step": 84130
    },
    {
      "epoch": 0.07843209451747868,
      "grad_norm": 1.0898776054382324,
      "learning_rate": 0.00018431378032319887,
      "loss": 3.6753,
      "step": 84140
    },
    {
      "epoch": 0.07844141613555777,
      "grad_norm": 1.0551025867462158,
      "learning_rate": 0.00018431191600110368,
      "loss": 3.8071,
      "step": 84150
    },
    {
      "epoch": 0.07845073775363685,
      "grad_norm": 1.0646858215332031,
      "learning_rate": 0.0001843100516790085,
      "loss": 3.6682,
      "step": 84160
    },
    {
      "epoch": 0.07846005937171595,
      "grad_norm": 1.327028751373291,
      "learning_rate": 0.0001843081873569133,
      "loss": 3.6955,
      "step": 84170
    },
    {
      "epoch": 0.07846938098979504,
      "grad_norm": 1.1067737340927124,
      "learning_rate": 0.0001843063230348181,
      "loss": 3.8273,
      "step": 84180
    },
    {
      "epoch": 0.07847870260787414,
      "grad_norm": 1.232177972793579,
      "learning_rate": 0.00018430445871272288,
      "loss": 3.7598,
      "step": 84190
    },
    {
      "epoch": 0.07848802422595323,
      "grad_norm": 1.0126584768295288,
      "learning_rate": 0.00018430259439062769,
      "loss": 3.8738,
      "step": 84200
    },
    {
      "epoch": 0.07849734584403231,
      "grad_norm": 0.983742356300354,
      "learning_rate": 0.0001843007300685325,
      "loss": 3.5861,
      "step": 84210
    },
    {
      "epoch": 0.0785066674621114,
      "grad_norm": 1.2597416639328003,
      "learning_rate": 0.0001842988657464373,
      "loss": 3.8379,
      "step": 84220
    },
    {
      "epoch": 0.0785159890801905,
      "grad_norm": 1.179137110710144,
      "learning_rate": 0.0001842970014243421,
      "loss": 3.8813,
      "step": 84230
    },
    {
      "epoch": 0.0785253106982696,
      "grad_norm": 1.04641854763031,
      "learning_rate": 0.00018429513710224688,
      "loss": 3.684,
      "step": 84240
    },
    {
      "epoch": 0.07853463231634869,
      "grad_norm": 1.1262953281402588,
      "learning_rate": 0.0001842932727801517,
      "loss": 3.8234,
      "step": 84250
    },
    {
      "epoch": 0.07854395393442777,
      "grad_norm": 1.2849595546722412,
      "learning_rate": 0.00018429140845805647,
      "loss": 3.7177,
      "step": 84260
    },
    {
      "epoch": 0.07855327555250687,
      "grad_norm": 1.0102072954177856,
      "learning_rate": 0.0001842895441359613,
      "loss": 4.0403,
      "step": 84270
    },
    {
      "epoch": 0.07856259717058596,
      "grad_norm": 1.0847790241241455,
      "learning_rate": 0.0001842876798138661,
      "loss": 3.6046,
      "step": 84280
    },
    {
      "epoch": 0.07857191878866505,
      "grad_norm": 1.239737629890442,
      "learning_rate": 0.0001842858154917709,
      "loss": 3.8402,
      "step": 84290
    },
    {
      "epoch": 0.07858124040674415,
      "grad_norm": 1.1698061227798462,
      "learning_rate": 0.0001842839511696757,
      "loss": 3.8391,
      "step": 84300
    },
    {
      "epoch": 0.07859056202482323,
      "grad_norm": 1.0555574893951416,
      "learning_rate": 0.00018428208684758047,
      "loss": 3.8124,
      "step": 84310
    },
    {
      "epoch": 0.07859988364290232,
      "grad_norm": 1.0454583168029785,
      "learning_rate": 0.00018428022252548528,
      "loss": 3.887,
      "step": 84320
    },
    {
      "epoch": 0.07860920526098142,
      "grad_norm": 1.0538222789764404,
      "learning_rate": 0.0001842783582033901,
      "loss": 3.6863,
      "step": 84330
    },
    {
      "epoch": 0.07861852687906051,
      "grad_norm": 1.1815606355667114,
      "learning_rate": 0.0001842764938812949,
      "loss": 3.8871,
      "step": 84340
    },
    {
      "epoch": 0.07862784849713961,
      "grad_norm": 1.0436738729476929,
      "learning_rate": 0.0001842746295591997,
      "loss": 3.7996,
      "step": 84350
    },
    {
      "epoch": 0.0786371701152187,
      "grad_norm": 0.9514278173446655,
      "learning_rate": 0.00018427276523710448,
      "loss": 3.7678,
      "step": 84360
    },
    {
      "epoch": 0.07864649173329778,
      "grad_norm": 1.1273123025894165,
      "learning_rate": 0.00018427090091500928,
      "loss": 3.9071,
      "step": 84370
    },
    {
      "epoch": 0.07865581335137688,
      "grad_norm": 0.9623312950134277,
      "learning_rate": 0.0001842690365929141,
      "loss": 3.7644,
      "step": 84380
    },
    {
      "epoch": 0.07866513496945597,
      "grad_norm": 1.20168936252594,
      "learning_rate": 0.0001842671722708189,
      "loss": 3.6952,
      "step": 84390
    },
    {
      "epoch": 0.07867445658753507,
      "grad_norm": 0.9949610829353333,
      "learning_rate": 0.0001842653079487237,
      "loss": 3.7994,
      "step": 84400
    },
    {
      "epoch": 0.07868377820561416,
      "grad_norm": 1.1510920524597168,
      "learning_rate": 0.00018426344362662848,
      "loss": 3.7852,
      "step": 84410
    },
    {
      "epoch": 0.07869309982369324,
      "grad_norm": 1.1216379404067993,
      "learning_rate": 0.0001842615793045333,
      "loss": 3.8337,
      "step": 84420
    },
    {
      "epoch": 0.07870242144177234,
      "grad_norm": 1.1767909526824951,
      "learning_rate": 0.0001842597149824381,
      "loss": 3.832,
      "step": 84430
    },
    {
      "epoch": 0.07871174305985143,
      "grad_norm": 0.909304678440094,
      "learning_rate": 0.0001842578506603429,
      "loss": 3.8632,
      "step": 84440
    },
    {
      "epoch": 0.07872106467793052,
      "grad_norm": 1.0011366605758667,
      "learning_rate": 0.0001842559863382477,
      "loss": 3.7499,
      "step": 84450
    },
    {
      "epoch": 0.07873038629600962,
      "grad_norm": 1.0479621887207031,
      "learning_rate": 0.0001842541220161525,
      "loss": 3.8636,
      "step": 84460
    },
    {
      "epoch": 0.0787397079140887,
      "grad_norm": 1.0732234716415405,
      "learning_rate": 0.0001842522576940573,
      "loss": 3.9293,
      "step": 84470
    },
    {
      "epoch": 0.0787490295321678,
      "grad_norm": 0.9904351830482483,
      "learning_rate": 0.0001842503933719621,
      "loss": 3.6681,
      "step": 84480
    },
    {
      "epoch": 0.07875835115024689,
      "grad_norm": 0.9443280100822449,
      "learning_rate": 0.00018424852904986688,
      "loss": 3.8094,
      "step": 84490
    },
    {
      "epoch": 0.07876767276832598,
      "grad_norm": 1.1469753980636597,
      "learning_rate": 0.0001842466647277717,
      "loss": 3.7712,
      "step": 84500
    },
    {
      "epoch": 0.07877699438640508,
      "grad_norm": 1.1131117343902588,
      "learning_rate": 0.0001842448004056765,
      "loss": 3.8465,
      "step": 84510
    },
    {
      "epoch": 0.07878631600448416,
      "grad_norm": 1.1790534257888794,
      "learning_rate": 0.0001842429360835813,
      "loss": 3.6722,
      "step": 84520
    },
    {
      "epoch": 0.07879563762256325,
      "grad_norm": 0.9649883508682251,
      "learning_rate": 0.0001842410717614861,
      "loss": 3.8123,
      "step": 84530
    },
    {
      "epoch": 0.07880495924064235,
      "grad_norm": 0.9954771399497986,
      "learning_rate": 0.00018423920743939088,
      "loss": 3.7202,
      "step": 84540
    },
    {
      "epoch": 0.07881428085872144,
      "grad_norm": 1.0212435722351074,
      "learning_rate": 0.0001842373431172957,
      "loss": 3.7701,
      "step": 84550
    },
    {
      "epoch": 0.07882360247680054,
      "grad_norm": 1.097159504890442,
      "learning_rate": 0.0001842354787952005,
      "loss": 3.6632,
      "step": 84560
    },
    {
      "epoch": 0.07883292409487962,
      "grad_norm": 1.154630422592163,
      "learning_rate": 0.0001842336144731053,
      "loss": 3.7954,
      "step": 84570
    },
    {
      "epoch": 0.07884224571295871,
      "grad_norm": 1.021910548210144,
      "learning_rate": 0.0001842317501510101,
      "loss": 3.7267,
      "step": 84580
    },
    {
      "epoch": 0.0788515673310378,
      "grad_norm": 1.2048863172531128,
      "learning_rate": 0.0001842298858289149,
      "loss": 3.8623,
      "step": 84590
    },
    {
      "epoch": 0.0788608889491169,
      "grad_norm": 1.1126295328140259,
      "learning_rate": 0.0001842280215068197,
      "loss": 3.7571,
      "step": 84600
    },
    {
      "epoch": 0.078870210567196,
      "grad_norm": 1.0838029384613037,
      "learning_rate": 0.0001842261571847245,
      "loss": 3.9039,
      "step": 84610
    },
    {
      "epoch": 0.07887953218527509,
      "grad_norm": 1.0315043926239014,
      "learning_rate": 0.0001842242928626293,
      "loss": 3.829,
      "step": 84620
    },
    {
      "epoch": 0.07888885380335417,
      "grad_norm": 0.9771035313606262,
      "learning_rate": 0.0001842224285405341,
      "loss": 3.698,
      "step": 84630
    },
    {
      "epoch": 0.07889817542143326,
      "grad_norm": 1.127875566482544,
      "learning_rate": 0.0001842205642184389,
      "loss": 3.9514,
      "step": 84640
    },
    {
      "epoch": 0.07890749703951236,
      "grad_norm": 1.121302843093872,
      "learning_rate": 0.0001842186998963437,
      "loss": 3.8343,
      "step": 84650
    },
    {
      "epoch": 0.07891681865759145,
      "grad_norm": 1.1165125370025635,
      "learning_rate": 0.0001842168355742485,
      "loss": 3.8333,
      "step": 84660
    },
    {
      "epoch": 0.07892614027567055,
      "grad_norm": 1.1416023969650269,
      "learning_rate": 0.0001842149712521533,
      "loss": 3.7029,
      "step": 84670
    },
    {
      "epoch": 0.07893546189374963,
      "grad_norm": 0.9400292634963989,
      "learning_rate": 0.00018421310693005812,
      "loss": 4.0763,
      "step": 84680
    },
    {
      "epoch": 0.07894478351182872,
      "grad_norm": 0.8694887757301331,
      "learning_rate": 0.0001842112426079629,
      "loss": 3.9657,
      "step": 84690
    },
    {
      "epoch": 0.07895410512990782,
      "grad_norm": 1.5508679151535034,
      "learning_rate": 0.0001842093782858677,
      "loss": 3.7858,
      "step": 84700
    },
    {
      "epoch": 0.07896342674798691,
      "grad_norm": 0.9177846908569336,
      "learning_rate": 0.00018420751396377248,
      "loss": 3.7105,
      "step": 84710
    },
    {
      "epoch": 0.078972748366066,
      "grad_norm": 0.9958915114402771,
      "learning_rate": 0.00018420564964167731,
      "loss": 3.9935,
      "step": 84720
    },
    {
      "epoch": 0.07898206998414509,
      "grad_norm": 1.0441303253173828,
      "learning_rate": 0.00018420378531958212,
      "loss": 3.9175,
      "step": 84730
    },
    {
      "epoch": 0.07899139160222418,
      "grad_norm": 0.9988207817077637,
      "learning_rate": 0.0001842019209974869,
      "loss": 3.8085,
      "step": 84740
    },
    {
      "epoch": 0.07900071322030328,
      "grad_norm": 1.2797143459320068,
      "learning_rate": 0.0001842000566753917,
      "loss": 3.797,
      "step": 84750
    },
    {
      "epoch": 0.07901003483838237,
      "grad_norm": 1.1548200845718384,
      "learning_rate": 0.00018419819235329649,
      "loss": 3.7997,
      "step": 84760
    },
    {
      "epoch": 0.07901935645646146,
      "grad_norm": 0.8812421560287476,
      "learning_rate": 0.0001841963280312013,
      "loss": 3.6566,
      "step": 84770
    },
    {
      "epoch": 0.07902867807454055,
      "grad_norm": 1.3446303606033325,
      "learning_rate": 0.00018419446370910613,
      "loss": 3.5789,
      "step": 84780
    },
    {
      "epoch": 0.07903799969261964,
      "grad_norm": 1.3593943119049072,
      "learning_rate": 0.0001841925993870109,
      "loss": 3.8234,
      "step": 84790
    },
    {
      "epoch": 0.07904732131069873,
      "grad_norm": 1.131202220916748,
      "learning_rate": 0.0001841907350649157,
      "loss": 3.7592,
      "step": 84800
    },
    {
      "epoch": 0.07905664292877783,
      "grad_norm": 1.0861531496047974,
      "learning_rate": 0.0001841888707428205,
      "loss": 3.8878,
      "step": 84810
    },
    {
      "epoch": 0.07906596454685692,
      "grad_norm": 0.976213276386261,
      "learning_rate": 0.0001841870064207253,
      "loss": 3.5527,
      "step": 84820
    },
    {
      "epoch": 0.07907528616493602,
      "grad_norm": 1.0512218475341797,
      "learning_rate": 0.0001841851420986301,
      "loss": 3.784,
      "step": 84830
    },
    {
      "epoch": 0.0790846077830151,
      "grad_norm": 1.0338109731674194,
      "learning_rate": 0.0001841832777765349,
      "loss": 3.8156,
      "step": 84840
    },
    {
      "epoch": 0.07909392940109419,
      "grad_norm": 1.10823392868042,
      "learning_rate": 0.00018418141345443972,
      "loss": 3.7439,
      "step": 84850
    },
    {
      "epoch": 0.07910325101917329,
      "grad_norm": 0.847698986530304,
      "learning_rate": 0.0001841795491323445,
      "loss": 3.9336,
      "step": 84860
    },
    {
      "epoch": 0.07911257263725238,
      "grad_norm": 1.2695664167404175,
      "learning_rate": 0.0001841776848102493,
      "loss": 3.6982,
      "step": 84870
    },
    {
      "epoch": 0.07912189425533148,
      "grad_norm": 1.2038811445236206,
      "learning_rate": 0.0001841758204881541,
      "loss": 3.8504,
      "step": 84880
    },
    {
      "epoch": 0.07913121587341056,
      "grad_norm": 1.0230463743209839,
      "learning_rate": 0.00018417395616605891,
      "loss": 3.7329,
      "step": 84890
    },
    {
      "epoch": 0.07914053749148965,
      "grad_norm": 0.8342705368995667,
      "learning_rate": 0.00018417209184396372,
      "loss": 3.8278,
      "step": 84900
    },
    {
      "epoch": 0.07914985910956875,
      "grad_norm": 1.163694143295288,
      "learning_rate": 0.0001841702275218685,
      "loss": 3.8028,
      "step": 84910
    },
    {
      "epoch": 0.07915918072764784,
      "grad_norm": 1.0234578847885132,
      "learning_rate": 0.0001841683631997733,
      "loss": 3.8589,
      "step": 84920
    },
    {
      "epoch": 0.07916850234572693,
      "grad_norm": 1.0066102743148804,
      "learning_rate": 0.0001841664988776781,
      "loss": 3.8175,
      "step": 84930
    },
    {
      "epoch": 0.07917782396380602,
      "grad_norm": 1.0909281969070435,
      "learning_rate": 0.0001841646345555829,
      "loss": 3.6713,
      "step": 84940
    },
    {
      "epoch": 0.07918714558188511,
      "grad_norm": 1.3643062114715576,
      "learning_rate": 0.00018416277023348772,
      "loss": 3.864,
      "step": 84950
    },
    {
      "epoch": 0.0791964671999642,
      "grad_norm": 1.1636476516723633,
      "learning_rate": 0.0001841609059113925,
      "loss": 3.7556,
      "step": 84960
    },
    {
      "epoch": 0.0792057888180433,
      "grad_norm": 1.2659858465194702,
      "learning_rate": 0.0001841590415892973,
      "loss": 3.8525,
      "step": 84970
    },
    {
      "epoch": 0.0792151104361224,
      "grad_norm": 1.0581899881362915,
      "learning_rate": 0.00018415717726720212,
      "loss": 3.8842,
      "step": 84980
    },
    {
      "epoch": 0.07922443205420147,
      "grad_norm": 1.0943363904953003,
      "learning_rate": 0.0001841553129451069,
      "loss": 3.9297,
      "step": 84990
    },
    {
      "epoch": 0.07923375367228057,
      "grad_norm": 0.9550975561141968,
      "learning_rate": 0.0001841534486230117,
      "loss": 3.8888,
      "step": 85000
    },
    {
      "epoch": 0.07924307529035966,
      "grad_norm": 1.1936850547790527,
      "learning_rate": 0.0001841515843009165,
      "loss": 3.9385,
      "step": 85010
    },
    {
      "epoch": 0.07925239690843876,
      "grad_norm": 1.0549874305725098,
      "learning_rate": 0.00018414971997882131,
      "loss": 4.0926,
      "step": 85020
    },
    {
      "epoch": 0.07926171852651785,
      "grad_norm": 1.0736641883850098,
      "learning_rate": 0.00018414785565672612,
      "loss": 3.7181,
      "step": 85030
    },
    {
      "epoch": 0.07927104014459695,
      "grad_norm": 1.0900259017944336,
      "learning_rate": 0.0001841459913346309,
      "loss": 3.7862,
      "step": 85040
    },
    {
      "epoch": 0.07928036176267603,
      "grad_norm": 1.0006166696548462,
      "learning_rate": 0.0001841441270125357,
      "loss": 3.6612,
      "step": 85050
    },
    {
      "epoch": 0.07928968338075512,
      "grad_norm": 1.0924447774887085,
      "learning_rate": 0.0001841422626904405,
      "loss": 3.5976,
      "step": 85060
    },
    {
      "epoch": 0.07929900499883422,
      "grad_norm": 1.1776107549667358,
      "learning_rate": 0.00018414039836834532,
      "loss": 3.9243,
      "step": 85070
    },
    {
      "epoch": 0.07930832661691331,
      "grad_norm": 1.0740976333618164,
      "learning_rate": 0.00018413853404625013,
      "loss": 3.9943,
      "step": 85080
    },
    {
      "epoch": 0.0793176482349924,
      "grad_norm": 1.1057941913604736,
      "learning_rate": 0.0001841366697241549,
      "loss": 3.936,
      "step": 85090
    },
    {
      "epoch": 0.07932696985307149,
      "grad_norm": 1.1965316534042358,
      "learning_rate": 0.0001841348054020597,
      "loss": 3.6351,
      "step": 85100
    },
    {
      "epoch": 0.07933629147115058,
      "grad_norm": 1.1138122081756592,
      "learning_rate": 0.00018413294107996452,
      "loss": 3.8118,
      "step": 85110
    },
    {
      "epoch": 0.07934561308922967,
      "grad_norm": 1.1469557285308838,
      "learning_rate": 0.00018413107675786932,
      "loss": 3.6665,
      "step": 85120
    },
    {
      "epoch": 0.07935493470730877,
      "grad_norm": 1.118495225906372,
      "learning_rate": 0.00018412921243577413,
      "loss": 3.9638,
      "step": 85130
    },
    {
      "epoch": 0.07936425632538786,
      "grad_norm": 1.2642390727996826,
      "learning_rate": 0.0001841273481136789,
      "loss": 3.697,
      "step": 85140
    },
    {
      "epoch": 0.07937357794346694,
      "grad_norm": 0.9739956259727478,
      "learning_rate": 0.00018412548379158372,
      "loss": 3.7185,
      "step": 85150
    },
    {
      "epoch": 0.07938289956154604,
      "grad_norm": 1.1326409578323364,
      "learning_rate": 0.0001841236194694885,
      "loss": 3.6436,
      "step": 85160
    },
    {
      "epoch": 0.07939222117962513,
      "grad_norm": 0.9808325171470642,
      "learning_rate": 0.00018412175514739333,
      "loss": 3.9744,
      "step": 85170
    },
    {
      "epoch": 0.07940154279770423,
      "grad_norm": 0.985626220703125,
      "learning_rate": 0.00018411989082529813,
      "loss": 3.7889,
      "step": 85180
    },
    {
      "epoch": 0.07941086441578332,
      "grad_norm": 1.0433235168457031,
      "learning_rate": 0.0001841180265032029,
      "loss": 3.801,
      "step": 85190
    },
    {
      "epoch": 0.0794201860338624,
      "grad_norm": 1.041653037071228,
      "learning_rate": 0.00018411616218110772,
      "loss": 3.7183,
      "step": 85200
    },
    {
      "epoch": 0.0794295076519415,
      "grad_norm": 1.2371515035629272,
      "learning_rate": 0.0001841142978590125,
      "loss": 3.8164,
      "step": 85210
    },
    {
      "epoch": 0.07943882927002059,
      "grad_norm": 1.0785897970199585,
      "learning_rate": 0.0001841124335369173,
      "loss": 3.6781,
      "step": 85220
    },
    {
      "epoch": 0.07944815088809969,
      "grad_norm": 0.9343023896217346,
      "learning_rate": 0.00018411056921482214,
      "loss": 3.8593,
      "step": 85230
    },
    {
      "epoch": 0.07945747250617878,
      "grad_norm": 1.0352178812026978,
      "learning_rate": 0.00018410870489272692,
      "loss": 3.8277,
      "step": 85240
    },
    {
      "epoch": 0.07946679412425786,
      "grad_norm": 0.9741109609603882,
      "learning_rate": 0.00018410684057063172,
      "loss": 3.819,
      "step": 85250
    },
    {
      "epoch": 0.07947611574233696,
      "grad_norm": 0.980547308921814,
      "learning_rate": 0.0001841049762485365,
      "loss": 3.8984,
      "step": 85260
    },
    {
      "epoch": 0.07948543736041605,
      "grad_norm": 1.056928277015686,
      "learning_rate": 0.0001841031119264413,
      "loss": 3.7907,
      "step": 85270
    },
    {
      "epoch": 0.07949475897849514,
      "grad_norm": 1.0453141927719116,
      "learning_rate": 0.00018410124760434612,
      "loss": 3.7565,
      "step": 85280
    },
    {
      "epoch": 0.07950408059657424,
      "grad_norm": 1.0586323738098145,
      "learning_rate": 0.00018409938328225092,
      "loss": 4.0055,
      "step": 85290
    },
    {
      "epoch": 0.07951340221465333,
      "grad_norm": 1.1289423704147339,
      "learning_rate": 0.00018409751896015573,
      "loss": 3.7942,
      "step": 85300
    },
    {
      "epoch": 0.07952272383273241,
      "grad_norm": 0.8841806054115295,
      "learning_rate": 0.0001840956546380605,
      "loss": 3.8614,
      "step": 85310
    },
    {
      "epoch": 0.07953204545081151,
      "grad_norm": 0.9278430938720703,
      "learning_rate": 0.00018409379031596531,
      "loss": 3.8544,
      "step": 85320
    },
    {
      "epoch": 0.0795413670688906,
      "grad_norm": 1.0475267171859741,
      "learning_rate": 0.00018409192599387012,
      "loss": 3.8325,
      "step": 85330
    },
    {
      "epoch": 0.0795506886869697,
      "grad_norm": 0.9517594575881958,
      "learning_rate": 0.00018409006167177493,
      "loss": 3.6956,
      "step": 85340
    },
    {
      "epoch": 0.07956001030504879,
      "grad_norm": 1.0580388307571411,
      "learning_rate": 0.00018408819734967973,
      "loss": 3.8208,
      "step": 85350
    },
    {
      "epoch": 0.07956933192312787,
      "grad_norm": 0.9327443242073059,
      "learning_rate": 0.0001840863330275845,
      "loss": 3.8884,
      "step": 85360
    },
    {
      "epoch": 0.07957865354120697,
      "grad_norm": 0.9613472819328308,
      "learning_rate": 0.00018408446870548932,
      "loss": 3.8864,
      "step": 85370
    },
    {
      "epoch": 0.07958797515928606,
      "grad_norm": 1.2380316257476807,
      "learning_rate": 0.0001840826043833941,
      "loss": 4.0024,
      "step": 85380
    },
    {
      "epoch": 0.07959729677736516,
      "grad_norm": 0.9193314909934998,
      "learning_rate": 0.0001840807400612989,
      "loss": 3.7472,
      "step": 85390
    },
    {
      "epoch": 0.07960661839544425,
      "grad_norm": 0.9377228021621704,
      "learning_rate": 0.00018407887573920374,
      "loss": 3.9261,
      "step": 85400
    },
    {
      "epoch": 0.07961594001352333,
      "grad_norm": 1.0107336044311523,
      "learning_rate": 0.00018407701141710852,
      "loss": 3.7933,
      "step": 85410
    },
    {
      "epoch": 0.07962526163160243,
      "grad_norm": 1.0399048328399658,
      "learning_rate": 0.00018407514709501332,
      "loss": 3.8144,
      "step": 85420
    },
    {
      "epoch": 0.07963458324968152,
      "grad_norm": 1.0344822406768799,
      "learning_rate": 0.0001840732827729181,
      "loss": 3.7442,
      "step": 85430
    },
    {
      "epoch": 0.07964390486776061,
      "grad_norm": 1.0051730871200562,
      "learning_rate": 0.0001840714184508229,
      "loss": 3.8527,
      "step": 85440
    },
    {
      "epoch": 0.07965322648583971,
      "grad_norm": 0.9361539483070374,
      "learning_rate": 0.00018406955412872771,
      "loss": 3.9253,
      "step": 85450
    },
    {
      "epoch": 0.07966254810391879,
      "grad_norm": 1.0752274990081787,
      "learning_rate": 0.00018406768980663252,
      "loss": 3.7602,
      "step": 85460
    },
    {
      "epoch": 0.07967186972199788,
      "grad_norm": 1.2303756475448608,
      "learning_rate": 0.00018406582548453733,
      "loss": 3.7696,
      "step": 85470
    },
    {
      "epoch": 0.07968119134007698,
      "grad_norm": 1.2040718793869019,
      "learning_rate": 0.0001840639611624421,
      "loss": 3.5875,
      "step": 85480
    },
    {
      "epoch": 0.07969051295815607,
      "grad_norm": 1.08222496509552,
      "learning_rate": 0.0001840620968403469,
      "loss": 3.632,
      "step": 85490
    },
    {
      "epoch": 0.07969983457623517,
      "grad_norm": 1.0344873666763306,
      "learning_rate": 0.00018406023251825172,
      "loss": 3.8184,
      "step": 85500
    },
    {
      "epoch": 0.07970915619431426,
      "grad_norm": 1.0591542720794678,
      "learning_rate": 0.00018405836819615653,
      "loss": 3.7231,
      "step": 85510
    },
    {
      "epoch": 0.07971847781239334,
      "grad_norm": 1.0919057130813599,
      "learning_rate": 0.00018405650387406133,
      "loss": 3.7239,
      "step": 85520
    },
    {
      "epoch": 0.07972779943047244,
      "grad_norm": 1.2973779439926147,
      "learning_rate": 0.0001840546395519661,
      "loss": 3.8624,
      "step": 85530
    },
    {
      "epoch": 0.07973712104855153,
      "grad_norm": 1.0838991403579712,
      "learning_rate": 0.00018405277522987092,
      "loss": 3.7636,
      "step": 85540
    },
    {
      "epoch": 0.07974644266663063,
      "grad_norm": 1.0984770059585571,
      "learning_rate": 0.00018405091090777572,
      "loss": 3.6863,
      "step": 85550
    },
    {
      "epoch": 0.07975576428470972,
      "grad_norm": 1.1590279340744019,
      "learning_rate": 0.00018404904658568053,
      "loss": 3.9384,
      "step": 85560
    },
    {
      "epoch": 0.0797650859027888,
      "grad_norm": 0.9886878132820129,
      "learning_rate": 0.00018404718226358534,
      "loss": 3.8282,
      "step": 85570
    },
    {
      "epoch": 0.0797744075208679,
      "grad_norm": 1.0675561428070068,
      "learning_rate": 0.00018404531794149012,
      "loss": 3.7463,
      "step": 85580
    },
    {
      "epoch": 0.07978372913894699,
      "grad_norm": 1.008531093597412,
      "learning_rate": 0.00018404345361939492,
      "loss": 3.806,
      "step": 85590
    },
    {
      "epoch": 0.07979305075702608,
      "grad_norm": 1.0168795585632324,
      "learning_rate": 0.00018404158929729973,
      "loss": 3.7237,
      "step": 85600
    },
    {
      "epoch": 0.07980237237510518,
      "grad_norm": 1.1714950799942017,
      "learning_rate": 0.0001840397249752045,
      "loss": 3.7835,
      "step": 85610
    },
    {
      "epoch": 0.07981169399318426,
      "grad_norm": 1.1439701318740845,
      "learning_rate": 0.00018403786065310934,
      "loss": 3.7361,
      "step": 85620
    },
    {
      "epoch": 0.07982101561126335,
      "grad_norm": 1.421473503112793,
      "learning_rate": 0.00018403599633101412,
      "loss": 3.9549,
      "step": 85630
    },
    {
      "epoch": 0.07983033722934245,
      "grad_norm": 0.9607113003730774,
      "learning_rate": 0.00018403413200891893,
      "loss": 3.9239,
      "step": 85640
    },
    {
      "epoch": 0.07983965884742154,
      "grad_norm": 1.2386573553085327,
      "learning_rate": 0.00018403226768682373,
      "loss": 3.7098,
      "step": 85650
    },
    {
      "epoch": 0.07984898046550064,
      "grad_norm": 0.9638813138008118,
      "learning_rate": 0.0001840304033647285,
      "loss": 3.7698,
      "step": 85660
    },
    {
      "epoch": 0.07985830208357972,
      "grad_norm": 1.1048636436462402,
      "learning_rate": 0.00018402853904263332,
      "loss": 3.851,
      "step": 85670
    },
    {
      "epoch": 0.07986762370165881,
      "grad_norm": 1.3939834833145142,
      "learning_rate": 0.00018402667472053812,
      "loss": 3.6794,
      "step": 85680
    },
    {
      "epoch": 0.07987694531973791,
      "grad_norm": 1.153687834739685,
      "learning_rate": 0.00018402481039844293,
      "loss": 3.6388,
      "step": 85690
    },
    {
      "epoch": 0.079886266937817,
      "grad_norm": 1.1596715450286865,
      "learning_rate": 0.00018402294607634774,
      "loss": 3.8487,
      "step": 85700
    },
    {
      "epoch": 0.0798955885558961,
      "grad_norm": 1.0172199010849,
      "learning_rate": 0.00018402108175425252,
      "loss": 3.8634,
      "step": 85710
    },
    {
      "epoch": 0.07990491017397519,
      "grad_norm": 1.0347466468811035,
      "learning_rate": 0.00018401921743215732,
      "loss": 3.7691,
      "step": 85720
    },
    {
      "epoch": 0.07991423179205427,
      "grad_norm": 0.8155280947685242,
      "learning_rate": 0.00018401735311006213,
      "loss": 3.6631,
      "step": 85730
    },
    {
      "epoch": 0.07992355341013337,
      "grad_norm": 0.9657805562019348,
      "learning_rate": 0.00018401548878796694,
      "loss": 3.778,
      "step": 85740
    },
    {
      "epoch": 0.07993287502821246,
      "grad_norm": 1.0087599754333496,
      "learning_rate": 0.00018401362446587174,
      "loss": 3.7298,
      "step": 85750
    },
    {
      "epoch": 0.07994219664629156,
      "grad_norm": 0.9749017357826233,
      "learning_rate": 0.00018401176014377652,
      "loss": 3.8486,
      "step": 85760
    },
    {
      "epoch": 0.07995151826437065,
      "grad_norm": 1.194705605506897,
      "learning_rate": 0.00018400989582168133,
      "loss": 3.9465,
      "step": 85770
    },
    {
      "epoch": 0.07996083988244973,
      "grad_norm": 1.0410354137420654,
      "learning_rate": 0.00018400803149958613,
      "loss": 3.6844,
      "step": 85780
    },
    {
      "epoch": 0.07997016150052882,
      "grad_norm": 1.0207102298736572,
      "learning_rate": 0.00018400616717749094,
      "loss": 3.9363,
      "step": 85790
    },
    {
      "epoch": 0.07997948311860792,
      "grad_norm": 1.0326533317565918,
      "learning_rate": 0.00018400430285539575,
      "loss": 3.781,
      "step": 85800
    },
    {
      "epoch": 0.07998880473668701,
      "grad_norm": 0.9000425338745117,
      "learning_rate": 0.00018400243853330052,
      "loss": 3.9121,
      "step": 85810
    },
    {
      "epoch": 0.07999812635476611,
      "grad_norm": 1.044683814048767,
      "learning_rate": 0.00018400057421120533,
      "loss": 3.6443,
      "step": 85820
    },
    {
      "epoch": 0.08000744797284519,
      "grad_norm": 1.1979143619537354,
      "learning_rate": 0.0001839987098891101,
      "loss": 3.8068,
      "step": 85830
    },
    {
      "epoch": 0.08001676959092428,
      "grad_norm": 1.0543488264083862,
      "learning_rate": 0.00018399684556701494,
      "loss": 3.8543,
      "step": 85840
    },
    {
      "epoch": 0.08002609120900338,
      "grad_norm": 0.9837391376495361,
      "learning_rate": 0.00018399498124491975,
      "loss": 3.6263,
      "step": 85850
    },
    {
      "epoch": 0.08003541282708247,
      "grad_norm": 1.1424713134765625,
      "learning_rate": 0.00018399311692282453,
      "loss": 3.8428,
      "step": 85860
    },
    {
      "epoch": 0.08004473444516157,
      "grad_norm": 1.1605498790740967,
      "learning_rate": 0.00018399125260072934,
      "loss": 3.7426,
      "step": 85870
    },
    {
      "epoch": 0.08005405606324065,
      "grad_norm": 1.1116801500320435,
      "learning_rate": 0.00018398938827863411,
      "loss": 3.724,
      "step": 85880
    },
    {
      "epoch": 0.08006337768131974,
      "grad_norm": 1.0646030902862549,
      "learning_rate": 0.00018398752395653892,
      "loss": 3.7617,
      "step": 85890
    },
    {
      "epoch": 0.08007269929939884,
      "grad_norm": 0.933161735534668,
      "learning_rate": 0.00018398565963444373,
      "loss": 3.8787,
      "step": 85900
    },
    {
      "epoch": 0.08008202091747793,
      "grad_norm": 1.1155601739883423,
      "learning_rate": 0.00018398379531234853,
      "loss": 3.6736,
      "step": 85910
    },
    {
      "epoch": 0.08009134253555703,
      "grad_norm": 1.1399741172790527,
      "learning_rate": 0.00018398193099025334,
      "loss": 3.7558,
      "step": 85920
    },
    {
      "epoch": 0.0801006641536361,
      "grad_norm": 1.1642236709594727,
      "learning_rate": 0.00018398006666815812,
      "loss": 3.7576,
      "step": 85930
    },
    {
      "epoch": 0.0801099857717152,
      "grad_norm": 1.1169875860214233,
      "learning_rate": 0.00018397820234606293,
      "loss": 3.694,
      "step": 85940
    },
    {
      "epoch": 0.0801193073897943,
      "grad_norm": 1.0721228122711182,
      "learning_rate": 0.00018397633802396773,
      "loss": 3.7974,
      "step": 85950
    },
    {
      "epoch": 0.08012862900787339,
      "grad_norm": 1.083777666091919,
      "learning_rate": 0.00018397447370187254,
      "loss": 3.8909,
      "step": 85960
    },
    {
      "epoch": 0.08013795062595248,
      "grad_norm": 0.9531497955322266,
      "learning_rate": 0.00018397260937977734,
      "loss": 3.7881,
      "step": 85970
    },
    {
      "epoch": 0.08014727224403158,
      "grad_norm": 1.2208172082901,
      "learning_rate": 0.00018397074505768212,
      "loss": 3.7114,
      "step": 85980
    },
    {
      "epoch": 0.08015659386211066,
      "grad_norm": 1.2289965152740479,
      "learning_rate": 0.00018396888073558693,
      "loss": 3.8726,
      "step": 85990
    },
    {
      "epoch": 0.08016591548018975,
      "grad_norm": 1.2421019077301025,
      "learning_rate": 0.00018396701641349174,
      "loss": 3.7367,
      "step": 86000
    },
    {
      "epoch": 0.08017523709826885,
      "grad_norm": 1.1848207712173462,
      "learning_rate": 0.00018396515209139654,
      "loss": 3.7776,
      "step": 86010
    },
    {
      "epoch": 0.08018455871634794,
      "grad_norm": 1.2445567846298218,
      "learning_rate": 0.00018396328776930135,
      "loss": 3.8148,
      "step": 86020
    },
    {
      "epoch": 0.08019388033442704,
      "grad_norm": 1.0715816020965576,
      "learning_rate": 0.00018396142344720613,
      "loss": 3.89,
      "step": 86030
    },
    {
      "epoch": 0.08020320195250612,
      "grad_norm": 0.959234356880188,
      "learning_rate": 0.00018395955912511093,
      "loss": 3.9502,
      "step": 86040
    },
    {
      "epoch": 0.08021252357058521,
      "grad_norm": 1.0772181749343872,
      "learning_rate": 0.00018395769480301574,
      "loss": 3.7025,
      "step": 86050
    },
    {
      "epoch": 0.0802218451886643,
      "grad_norm": 0.9631361365318298,
      "learning_rate": 0.00018395583048092052,
      "loss": 3.857,
      "step": 86060
    },
    {
      "epoch": 0.0802311668067434,
      "grad_norm": 1.1172244548797607,
      "learning_rate": 0.00018395396615882535,
      "loss": 3.7509,
      "step": 86070
    },
    {
      "epoch": 0.0802404884248225,
      "grad_norm": 0.9084822535514832,
      "learning_rate": 0.00018395210183673013,
      "loss": 3.6705,
      "step": 86080
    },
    {
      "epoch": 0.08024981004290158,
      "grad_norm": 1.1400363445281982,
      "learning_rate": 0.00018395023751463494,
      "loss": 3.9611,
      "step": 86090
    },
    {
      "epoch": 0.08025913166098067,
      "grad_norm": 1.202383279800415,
      "learning_rate": 0.00018394837319253975,
      "loss": 3.7933,
      "step": 86100
    },
    {
      "epoch": 0.08026845327905976,
      "grad_norm": 1.1467722654342651,
      "learning_rate": 0.00018394650887044452,
      "loss": 3.9456,
      "step": 86110
    },
    {
      "epoch": 0.08027777489713886,
      "grad_norm": 0.8846380710601807,
      "learning_rate": 0.00018394464454834933,
      "loss": 3.7763,
      "step": 86120
    },
    {
      "epoch": 0.08028709651521795,
      "grad_norm": 0.9401875734329224,
      "learning_rate": 0.00018394278022625414,
      "loss": 3.7926,
      "step": 86130
    },
    {
      "epoch": 0.08029641813329703,
      "grad_norm": 1.3205608129501343,
      "learning_rate": 0.00018394091590415894,
      "loss": 3.6787,
      "step": 86140
    },
    {
      "epoch": 0.08030573975137613,
      "grad_norm": 1.263063669204712,
      "learning_rate": 0.00018393905158206375,
      "loss": 3.8669,
      "step": 86150
    },
    {
      "epoch": 0.08031506136945522,
      "grad_norm": 1.107473373413086,
      "learning_rate": 0.00018393718725996853,
      "loss": 3.6793,
      "step": 86160
    },
    {
      "epoch": 0.08032438298753432,
      "grad_norm": 1.0453741550445557,
      "learning_rate": 0.00018393532293787334,
      "loss": 3.889,
      "step": 86170
    },
    {
      "epoch": 0.08033370460561341,
      "grad_norm": 1.0186234712600708,
      "learning_rate": 0.00018393345861577814,
      "loss": 3.804,
      "step": 86180
    },
    {
      "epoch": 0.0803430262236925,
      "grad_norm": 1.1373971700668335,
      "learning_rate": 0.00018393159429368295,
      "loss": 3.7556,
      "step": 86190
    },
    {
      "epoch": 0.08035234784177159,
      "grad_norm": 1.2954421043395996,
      "learning_rate": 0.00018392972997158775,
      "loss": 3.7256,
      "step": 86200
    },
    {
      "epoch": 0.08036166945985068,
      "grad_norm": 1.2928177118301392,
      "learning_rate": 0.00018392786564949253,
      "loss": 3.7952,
      "step": 86210
    },
    {
      "epoch": 0.08037099107792978,
      "grad_norm": 1.3003520965576172,
      "learning_rate": 0.00018392600132739734,
      "loss": 4.0539,
      "step": 86220
    },
    {
      "epoch": 0.08038031269600887,
      "grad_norm": 1.0587557554244995,
      "learning_rate": 0.00018392413700530215,
      "loss": 4.0069,
      "step": 86230
    },
    {
      "epoch": 0.08038963431408797,
      "grad_norm": 1.0148619413375854,
      "learning_rate": 0.00018392227268320695,
      "loss": 3.7401,
      "step": 86240
    },
    {
      "epoch": 0.08039895593216705,
      "grad_norm": 1.018943190574646,
      "learning_rate": 0.00018392040836111176,
      "loss": 3.6452,
      "step": 86250
    },
    {
      "epoch": 0.08040827755024614,
      "grad_norm": 1.1026042699813843,
      "learning_rate": 0.00018391854403901654,
      "loss": 3.674,
      "step": 86260
    },
    {
      "epoch": 0.08041759916832524,
      "grad_norm": 1.0791434049606323,
      "learning_rate": 0.00018391667971692134,
      "loss": 3.6482,
      "step": 86270
    },
    {
      "epoch": 0.08042692078640433,
      "grad_norm": 1.3303471803665161,
      "learning_rate": 0.00018391481539482612,
      "loss": 3.8006,
      "step": 86280
    },
    {
      "epoch": 0.08043624240448342,
      "grad_norm": 1.046758770942688,
      "learning_rate": 0.00018391295107273096,
      "loss": 3.8273,
      "step": 86290
    },
    {
      "epoch": 0.0804455640225625,
      "grad_norm": 1.0868022441864014,
      "learning_rate": 0.00018391108675063576,
      "loss": 3.8553,
      "step": 86300
    },
    {
      "epoch": 0.0804548856406416,
      "grad_norm": 1.0810528993606567,
      "learning_rate": 0.00018390922242854054,
      "loss": 3.8121,
      "step": 86310
    },
    {
      "epoch": 0.0804642072587207,
      "grad_norm": 1.002315878868103,
      "learning_rate": 0.00018390735810644535,
      "loss": 3.8701,
      "step": 86320
    },
    {
      "epoch": 0.08047352887679979,
      "grad_norm": 1.2069971561431885,
      "learning_rate": 0.00018390549378435013,
      "loss": 3.8172,
      "step": 86330
    },
    {
      "epoch": 0.08048285049487888,
      "grad_norm": 1.140898585319519,
      "learning_rate": 0.00018390362946225493,
      "loss": 3.6846,
      "step": 86340
    },
    {
      "epoch": 0.08049217211295796,
      "grad_norm": 0.9989913702011108,
      "learning_rate": 0.00018390176514015977,
      "loss": 3.7451,
      "step": 86350
    },
    {
      "epoch": 0.08050149373103706,
      "grad_norm": 0.9805017709732056,
      "learning_rate": 0.00018389990081806455,
      "loss": 3.774,
      "step": 86360
    },
    {
      "epoch": 0.08051081534911615,
      "grad_norm": 1.1461199522018433,
      "learning_rate": 0.00018389803649596935,
      "loss": 3.7369,
      "step": 86370
    },
    {
      "epoch": 0.08052013696719525,
      "grad_norm": 1.0761431455612183,
      "learning_rate": 0.00018389617217387413,
      "loss": 3.8779,
      "step": 86380
    },
    {
      "epoch": 0.08052945858527434,
      "grad_norm": 1.1439664363861084,
      "learning_rate": 0.00018389430785177894,
      "loss": 3.8176,
      "step": 86390
    },
    {
      "epoch": 0.08053878020335344,
      "grad_norm": 1.0807591676712036,
      "learning_rate": 0.00018389244352968374,
      "loss": 3.851,
      "step": 86400
    },
    {
      "epoch": 0.08054810182143252,
      "grad_norm": 1.0149829387664795,
      "learning_rate": 0.00018389057920758855,
      "loss": 3.8582,
      "step": 86410
    },
    {
      "epoch": 0.08055742343951161,
      "grad_norm": 1.0543861389160156,
      "learning_rate": 0.00018388871488549336,
      "loss": 3.8565,
      "step": 86420
    },
    {
      "epoch": 0.0805667450575907,
      "grad_norm": 1.107049584388733,
      "learning_rate": 0.00018388685056339814,
      "loss": 3.7408,
      "step": 86430
    },
    {
      "epoch": 0.0805760666756698,
      "grad_norm": 1.0318734645843506,
      "learning_rate": 0.00018388498624130294,
      "loss": 3.8304,
      "step": 86440
    },
    {
      "epoch": 0.0805853882937489,
      "grad_norm": 1.1896610260009766,
      "learning_rate": 0.00018388312191920775,
      "loss": 3.8366,
      "step": 86450
    },
    {
      "epoch": 0.08059470991182797,
      "grad_norm": 0.9602797627449036,
      "learning_rate": 0.00018388125759711256,
      "loss": 3.5415,
      "step": 86460
    },
    {
      "epoch": 0.08060403152990707,
      "grad_norm": 0.9590275287628174,
      "learning_rate": 0.00018387939327501736,
      "loss": 3.8329,
      "step": 86470
    },
    {
      "epoch": 0.08061335314798616,
      "grad_norm": 0.8180308938026428,
      "learning_rate": 0.00018387752895292214,
      "loss": 3.7499,
      "step": 86480
    },
    {
      "epoch": 0.08062267476606526,
      "grad_norm": 1.045941710472107,
      "learning_rate": 0.00018387566463082695,
      "loss": 3.8083,
      "step": 86490
    },
    {
      "epoch": 0.08063199638414435,
      "grad_norm": 1.2023191452026367,
      "learning_rate": 0.00018387380030873175,
      "loss": 3.7767,
      "step": 86500
    },
    {
      "epoch": 0.08064131800222343,
      "grad_norm": 1.0613890886306763,
      "learning_rate": 0.00018387193598663653,
      "loss": 3.8356,
      "step": 86510
    },
    {
      "epoch": 0.08065063962030253,
      "grad_norm": 1.0764449834823608,
      "learning_rate": 0.00018387007166454137,
      "loss": 4.0958,
      "step": 86520
    },
    {
      "epoch": 0.08065996123838162,
      "grad_norm": 1.170217514038086,
      "learning_rate": 0.00018386820734244615,
      "loss": 3.5987,
      "step": 86530
    },
    {
      "epoch": 0.08066928285646072,
      "grad_norm": 1.0368813276290894,
      "learning_rate": 0.00018386634302035095,
      "loss": 3.8775,
      "step": 86540
    },
    {
      "epoch": 0.08067860447453981,
      "grad_norm": 1.1642074584960938,
      "learning_rate": 0.00018386447869825576,
      "loss": 3.7852,
      "step": 86550
    },
    {
      "epoch": 0.08068792609261889,
      "grad_norm": 1.0248185396194458,
      "learning_rate": 0.00018386261437616054,
      "loss": 3.6533,
      "step": 86560
    },
    {
      "epoch": 0.08069724771069799,
      "grad_norm": 0.9976388216018677,
      "learning_rate": 0.00018386075005406534,
      "loss": 3.8853,
      "step": 86570
    },
    {
      "epoch": 0.08070656932877708,
      "grad_norm": 1.2950234413146973,
      "learning_rate": 0.00018385888573197015,
      "loss": 3.9087,
      "step": 86580
    },
    {
      "epoch": 0.08071589094685618,
      "grad_norm": 1.0029520988464355,
      "learning_rate": 0.00018385702140987496,
      "loss": 3.7748,
      "step": 86590
    },
    {
      "epoch": 0.08072521256493527,
      "grad_norm": 1.1471796035766602,
      "learning_rate": 0.00018385515708777976,
      "loss": 3.7454,
      "step": 86600
    },
    {
      "epoch": 0.08073453418301435,
      "grad_norm": 0.961183488368988,
      "learning_rate": 0.00018385329276568454,
      "loss": 3.7642,
      "step": 86610
    },
    {
      "epoch": 0.08074385580109344,
      "grad_norm": 1.171913743019104,
      "learning_rate": 0.00018385142844358935,
      "loss": 3.8127,
      "step": 86620
    },
    {
      "epoch": 0.08075317741917254,
      "grad_norm": 1.1132203340530396,
      "learning_rate": 0.00018384956412149415,
      "loss": 3.7799,
      "step": 86630
    },
    {
      "epoch": 0.08076249903725163,
      "grad_norm": 1.0140904188156128,
      "learning_rate": 0.00018384769979939896,
      "loss": 3.8703,
      "step": 86640
    },
    {
      "epoch": 0.08077182065533073,
      "grad_norm": 0.9471129775047302,
      "learning_rate": 0.00018384583547730374,
      "loss": 3.6219,
      "step": 86650
    },
    {
      "epoch": 0.08078114227340982,
      "grad_norm": 1.0058213472366333,
      "learning_rate": 0.00018384397115520855,
      "loss": 3.651,
      "step": 86660
    },
    {
      "epoch": 0.0807904638914889,
      "grad_norm": 1.2235145568847656,
      "learning_rate": 0.00018384210683311335,
      "loss": 3.8946,
      "step": 86670
    },
    {
      "epoch": 0.080799785509568,
      "grad_norm": 1.0207700729370117,
      "learning_rate": 0.00018384024251101816,
      "loss": 3.6566,
      "step": 86680
    },
    {
      "epoch": 0.08080910712764709,
      "grad_norm": 1.172377109527588,
      "learning_rate": 0.00018383837818892297,
      "loss": 3.8104,
      "step": 86690
    },
    {
      "epoch": 0.08081842874572619,
      "grad_norm": 0.9955160617828369,
      "learning_rate": 0.00018383651386682774,
      "loss": 3.6563,
      "step": 86700
    },
    {
      "epoch": 0.08082775036380528,
      "grad_norm": 1.0856544971466064,
      "learning_rate": 0.00018383464954473255,
      "loss": 3.846,
      "step": 86710
    },
    {
      "epoch": 0.08083707198188436,
      "grad_norm": 1.1587766408920288,
      "learning_rate": 0.00018383278522263736,
      "loss": 3.8525,
      "step": 86720
    },
    {
      "epoch": 0.08084639359996346,
      "grad_norm": 1.0754504203796387,
      "learning_rate": 0.00018383092090054214,
      "loss": 3.9893,
      "step": 86730
    },
    {
      "epoch": 0.08085571521804255,
      "grad_norm": 1.1348155736923218,
      "learning_rate": 0.00018382905657844697,
      "loss": 3.9363,
      "step": 86740
    },
    {
      "epoch": 0.08086503683612165,
      "grad_norm": 1.3068568706512451,
      "learning_rate": 0.00018382719225635175,
      "loss": 3.7354,
      "step": 86750
    },
    {
      "epoch": 0.08087435845420074,
      "grad_norm": 1.0815411806106567,
      "learning_rate": 0.00018382532793425656,
      "loss": 3.9746,
      "step": 86760
    },
    {
      "epoch": 0.08088368007227982,
      "grad_norm": 1.038948655128479,
      "learning_rate": 0.00018382346361216136,
      "loss": 3.8802,
      "step": 86770
    },
    {
      "epoch": 0.08089300169035891,
      "grad_norm": 0.9727603197097778,
      "learning_rate": 0.00018382159929006614,
      "loss": 3.7927,
      "step": 86780
    },
    {
      "epoch": 0.08090232330843801,
      "grad_norm": 1.2196565866470337,
      "learning_rate": 0.00018381973496797095,
      "loss": 3.5435,
      "step": 86790
    },
    {
      "epoch": 0.0809116449265171,
      "grad_norm": 1.1727768182754517,
      "learning_rate": 0.00018381787064587575,
      "loss": 3.7287,
      "step": 86800
    },
    {
      "epoch": 0.0809209665445962,
      "grad_norm": 1.1867502927780151,
      "learning_rate": 0.00018381600632378056,
      "loss": 3.7486,
      "step": 86810
    },
    {
      "epoch": 0.08093028816267528,
      "grad_norm": 1.1310523748397827,
      "learning_rate": 0.00018381414200168537,
      "loss": 3.8152,
      "step": 86820
    },
    {
      "epoch": 0.08093960978075437,
      "grad_norm": 1.11937415599823,
      "learning_rate": 0.00018381227767959015,
      "loss": 3.7065,
      "step": 86830
    },
    {
      "epoch": 0.08094893139883347,
      "grad_norm": 1.063103199005127,
      "learning_rate": 0.00018381041335749495,
      "loss": 3.8604,
      "step": 86840
    },
    {
      "epoch": 0.08095825301691256,
      "grad_norm": 0.997017502784729,
      "learning_rate": 0.00018380854903539976,
      "loss": 3.8214,
      "step": 86850
    },
    {
      "epoch": 0.08096757463499166,
      "grad_norm": 1.0204777717590332,
      "learning_rate": 0.00018380668471330456,
      "loss": 3.6761,
      "step": 86860
    },
    {
      "epoch": 0.08097689625307075,
      "grad_norm": 1.0650910139083862,
      "learning_rate": 0.00018380482039120937,
      "loss": 3.8837,
      "step": 86870
    },
    {
      "epoch": 0.08098621787114983,
      "grad_norm": 1.0104764699935913,
      "learning_rate": 0.00018380295606911415,
      "loss": 3.9245,
      "step": 86880
    },
    {
      "epoch": 0.08099553948922893,
      "grad_norm": 0.9447472095489502,
      "learning_rate": 0.00018380109174701896,
      "loss": 3.884,
      "step": 86890
    },
    {
      "epoch": 0.08100486110730802,
      "grad_norm": 1.0379087924957275,
      "learning_rate": 0.00018379922742492374,
      "loss": 3.7063,
      "step": 86900
    },
    {
      "epoch": 0.08101418272538712,
      "grad_norm": 1.1525237560272217,
      "learning_rate": 0.00018379736310282857,
      "loss": 3.8129,
      "step": 86910
    },
    {
      "epoch": 0.08102350434346621,
      "grad_norm": 1.2476550340652466,
      "learning_rate": 0.00018379549878073337,
      "loss": 3.7508,
      "step": 86920
    },
    {
      "epoch": 0.08103282596154529,
      "grad_norm": 0.9624122977256775,
      "learning_rate": 0.00018379363445863815,
      "loss": 3.7989,
      "step": 86930
    },
    {
      "epoch": 0.08104214757962439,
      "grad_norm": 1.1545637845993042,
      "learning_rate": 0.00018379177013654296,
      "loss": 3.5161,
      "step": 86940
    },
    {
      "epoch": 0.08105146919770348,
      "grad_norm": 0.8881451487541199,
      "learning_rate": 0.00018378990581444774,
      "loss": 3.5164,
      "step": 86950
    },
    {
      "epoch": 0.08106079081578257,
      "grad_norm": 1.070548176765442,
      "learning_rate": 0.00018378804149235255,
      "loss": 3.7209,
      "step": 86960
    },
    {
      "epoch": 0.08107011243386167,
      "grad_norm": 0.8492774963378906,
      "learning_rate": 0.00018378617717025738,
      "loss": 3.8246,
      "step": 86970
    },
    {
      "epoch": 0.08107943405194075,
      "grad_norm": 2.341379404067993,
      "learning_rate": 0.00018378431284816216,
      "loss": 3.8824,
      "step": 86980
    },
    {
      "epoch": 0.08108875567001984,
      "grad_norm": 1.0978611707687378,
      "learning_rate": 0.00018378244852606696,
      "loss": 3.7175,
      "step": 86990
    },
    {
      "epoch": 0.08109807728809894,
      "grad_norm": 1.1658436059951782,
      "learning_rate": 0.00018378058420397174,
      "loss": 3.837,
      "step": 87000
    },
    {
      "epoch": 0.08110739890617803,
      "grad_norm": 1.0778061151504517,
      "learning_rate": 0.00018377871988187655,
      "loss": 3.8537,
      "step": 87010
    },
    {
      "epoch": 0.08111672052425713,
      "grad_norm": 1.2687877416610718,
      "learning_rate": 0.00018377685555978136,
      "loss": 3.7353,
      "step": 87020
    },
    {
      "epoch": 0.08112604214233621,
      "grad_norm": 1.150300145149231,
      "learning_rate": 0.00018377499123768616,
      "loss": 3.8679,
      "step": 87030
    },
    {
      "epoch": 0.0811353637604153,
      "grad_norm": 0.848334789276123,
      "learning_rate": 0.00018377312691559097,
      "loss": 3.7532,
      "step": 87040
    },
    {
      "epoch": 0.0811446853784944,
      "grad_norm": 1.059403657913208,
      "learning_rate": 0.00018377126259349575,
      "loss": 3.7444,
      "step": 87050
    },
    {
      "epoch": 0.08115400699657349,
      "grad_norm": 0.906606912612915,
      "learning_rate": 0.00018376939827140055,
      "loss": 3.8373,
      "step": 87060
    },
    {
      "epoch": 0.08116332861465259,
      "grad_norm": 1.1472227573394775,
      "learning_rate": 0.00018376753394930536,
      "loss": 3.8033,
      "step": 87070
    },
    {
      "epoch": 0.08117265023273168,
      "grad_norm": 1.0749406814575195,
      "learning_rate": 0.00018376566962721017,
      "loss": 3.8542,
      "step": 87080
    },
    {
      "epoch": 0.08118197185081076,
      "grad_norm": 1.1105237007141113,
      "learning_rate": 0.00018376380530511497,
      "loss": 3.8686,
      "step": 87090
    },
    {
      "epoch": 0.08119129346888986,
      "grad_norm": 1.0853302478790283,
      "learning_rate": 0.00018376194098301975,
      "loss": 3.6619,
      "step": 87100
    },
    {
      "epoch": 0.08120061508696895,
      "grad_norm": 0.9563930034637451,
      "learning_rate": 0.00018376007666092456,
      "loss": 3.9629,
      "step": 87110
    },
    {
      "epoch": 0.08120993670504804,
      "grad_norm": 1.0163441896438599,
      "learning_rate": 0.00018375821233882937,
      "loss": 3.7519,
      "step": 87120
    },
    {
      "epoch": 0.08121925832312714,
      "grad_norm": 1.202728509902954,
      "learning_rate": 0.00018375634801673417,
      "loss": 3.6805,
      "step": 87130
    },
    {
      "epoch": 0.08122857994120622,
      "grad_norm": 0.9282944202423096,
      "learning_rate": 0.00018375448369463898,
      "loss": 3.5134,
      "step": 87140
    },
    {
      "epoch": 0.08123790155928531,
      "grad_norm": 1.2906594276428223,
      "learning_rate": 0.00018375261937254376,
      "loss": 3.6826,
      "step": 87150
    },
    {
      "epoch": 0.08124722317736441,
      "grad_norm": 0.9535414576530457,
      "learning_rate": 0.00018375075505044856,
      "loss": 3.7076,
      "step": 87160
    },
    {
      "epoch": 0.0812565447954435,
      "grad_norm": 1.2087180614471436,
      "learning_rate": 0.00018374889072835337,
      "loss": 3.7834,
      "step": 87170
    },
    {
      "epoch": 0.0812658664135226,
      "grad_norm": 0.9775599241256714,
      "learning_rate": 0.00018374702640625815,
      "loss": 3.8077,
      "step": 87180
    },
    {
      "epoch": 0.08127518803160168,
      "grad_norm": 1.135016679763794,
      "learning_rate": 0.00018374516208416298,
      "loss": 3.6895,
      "step": 87190
    },
    {
      "epoch": 0.08128450964968077,
      "grad_norm": 1.1685428619384766,
      "learning_rate": 0.00018374329776206776,
      "loss": 3.8787,
      "step": 87200
    },
    {
      "epoch": 0.08129383126775987,
      "grad_norm": 1.1307271718978882,
      "learning_rate": 0.00018374143343997257,
      "loss": 3.7397,
      "step": 87210
    },
    {
      "epoch": 0.08130315288583896,
      "grad_norm": 1.0731723308563232,
      "learning_rate": 0.00018373956911787737,
      "loss": 3.9423,
      "step": 87220
    },
    {
      "epoch": 0.08131247450391806,
      "grad_norm": 1.0390323400497437,
      "learning_rate": 0.00018373770479578215,
      "loss": 3.8803,
      "step": 87230
    },
    {
      "epoch": 0.08132179612199714,
      "grad_norm": 0.8863222599029541,
      "learning_rate": 0.00018373584047368696,
      "loss": 3.7478,
      "step": 87240
    },
    {
      "epoch": 0.08133111774007623,
      "grad_norm": 1.1596174240112305,
      "learning_rate": 0.00018373397615159177,
      "loss": 3.9072,
      "step": 87250
    },
    {
      "epoch": 0.08134043935815533,
      "grad_norm": 1.0059969425201416,
      "learning_rate": 0.00018373211182949657,
      "loss": 3.841,
      "step": 87260
    },
    {
      "epoch": 0.08134976097623442,
      "grad_norm": 1.2893091440200806,
      "learning_rate": 0.00018373024750740138,
      "loss": 3.9688,
      "step": 87270
    },
    {
      "epoch": 0.08135908259431351,
      "grad_norm": 1.1730643510818481,
      "learning_rate": 0.00018372838318530616,
      "loss": 3.795,
      "step": 87280
    },
    {
      "epoch": 0.08136840421239261,
      "grad_norm": 1.027681827545166,
      "learning_rate": 0.00018372651886321096,
      "loss": 3.8218,
      "step": 87290
    },
    {
      "epoch": 0.08137772583047169,
      "grad_norm": 1.141347885131836,
      "learning_rate": 0.00018372465454111577,
      "loss": 3.5413,
      "step": 87300
    },
    {
      "epoch": 0.08138704744855078,
      "grad_norm": 1.0913288593292236,
      "learning_rate": 0.00018372279021902058,
      "loss": 3.7033,
      "step": 87310
    },
    {
      "epoch": 0.08139636906662988,
      "grad_norm": 0.984118640422821,
      "learning_rate": 0.00018372092589692538,
      "loss": 3.8203,
      "step": 87320
    },
    {
      "epoch": 0.08140569068470897,
      "grad_norm": 1.160261631011963,
      "learning_rate": 0.00018371906157483016,
      "loss": 3.741,
      "step": 87330
    },
    {
      "epoch": 0.08141501230278807,
      "grad_norm": 1.1021716594696045,
      "learning_rate": 0.00018371719725273497,
      "loss": 3.829,
      "step": 87340
    },
    {
      "epoch": 0.08142433392086715,
      "grad_norm": 1.0623165369033813,
      "learning_rate": 0.00018371533293063975,
      "loss": 3.8352,
      "step": 87350
    },
    {
      "epoch": 0.08143365553894624,
      "grad_norm": 1.2760286331176758,
      "learning_rate": 0.00018371346860854458,
      "loss": 3.8656,
      "step": 87360
    },
    {
      "epoch": 0.08144297715702534,
      "grad_norm": 1.0129201412200928,
      "learning_rate": 0.0001837116042864494,
      "loss": 3.9917,
      "step": 87370
    },
    {
      "epoch": 0.08145229877510443,
      "grad_norm": 1.1343892812728882,
      "learning_rate": 0.00018370973996435417,
      "loss": 3.9051,
      "step": 87380
    },
    {
      "epoch": 0.08146162039318353,
      "grad_norm": 1.0786031484603882,
      "learning_rate": 0.00018370787564225897,
      "loss": 3.8578,
      "step": 87390
    },
    {
      "epoch": 0.0814709420112626,
      "grad_norm": 1.0248385667800903,
      "learning_rate": 0.00018370601132016375,
      "loss": 3.8259,
      "step": 87400
    },
    {
      "epoch": 0.0814802636293417,
      "grad_norm": 1.2125941514968872,
      "learning_rate": 0.00018370414699806856,
      "loss": 3.7788,
      "step": 87410
    },
    {
      "epoch": 0.0814895852474208,
      "grad_norm": 0.8906605839729309,
      "learning_rate": 0.0001837022826759734,
      "loss": 3.8559,
      "step": 87420
    },
    {
      "epoch": 0.08149890686549989,
      "grad_norm": 1.1789034605026245,
      "learning_rate": 0.00018370041835387817,
      "loss": 3.7401,
      "step": 87430
    },
    {
      "epoch": 0.08150822848357898,
      "grad_norm": 1.0645008087158203,
      "learning_rate": 0.00018369855403178298,
      "loss": 3.7498,
      "step": 87440
    },
    {
      "epoch": 0.08151755010165807,
      "grad_norm": 1.0600028038024902,
      "learning_rate": 0.00018369668970968776,
      "loss": 3.8091,
      "step": 87450
    },
    {
      "epoch": 0.08152687171973716,
      "grad_norm": 0.993247389793396,
      "learning_rate": 0.00018369482538759256,
      "loss": 3.7277,
      "step": 87460
    },
    {
      "epoch": 0.08153619333781625,
      "grad_norm": 1.3809962272644043,
      "learning_rate": 0.00018369296106549737,
      "loss": 3.5999,
      "step": 87470
    },
    {
      "epoch": 0.08154551495589535,
      "grad_norm": 1.1036032438278198,
      "learning_rate": 0.00018369109674340218,
      "loss": 3.7875,
      "step": 87480
    },
    {
      "epoch": 0.08155483657397444,
      "grad_norm": 1.138675332069397,
      "learning_rate": 0.00018368923242130698,
      "loss": 3.8071,
      "step": 87490
    },
    {
      "epoch": 0.08156415819205352,
      "grad_norm": 0.9789600968360901,
      "learning_rate": 0.00018368736809921176,
      "loss": 3.6312,
      "step": 87500
    },
    {
      "epoch": 0.08157347981013262,
      "grad_norm": 1.0860228538513184,
      "learning_rate": 0.00018368550377711657,
      "loss": 3.8464,
      "step": 87510
    },
    {
      "epoch": 0.08158280142821171,
      "grad_norm": 0.9313340783119202,
      "learning_rate": 0.00018368363945502137,
      "loss": 3.7533,
      "step": 87520
    },
    {
      "epoch": 0.08159212304629081,
      "grad_norm": 1.0795481204986572,
      "learning_rate": 0.00018368177513292618,
      "loss": 3.762,
      "step": 87530
    },
    {
      "epoch": 0.0816014446643699,
      "grad_norm": 1.0094270706176758,
      "learning_rate": 0.00018367991081083099,
      "loss": 3.8923,
      "step": 87540
    },
    {
      "epoch": 0.081610766282449,
      "grad_norm": 0.9819002151489258,
      "learning_rate": 0.00018367804648873577,
      "loss": 3.6375,
      "step": 87550
    },
    {
      "epoch": 0.08162008790052808,
      "grad_norm": 0.9652639031410217,
      "learning_rate": 0.00018367618216664057,
      "loss": 3.7721,
      "step": 87560
    },
    {
      "epoch": 0.08162940951860717,
      "grad_norm": 1.0501912832260132,
      "learning_rate": 0.00018367431784454538,
      "loss": 4.0132,
      "step": 87570
    },
    {
      "epoch": 0.08163873113668627,
      "grad_norm": 1.008823037147522,
      "learning_rate": 0.00018367245352245018,
      "loss": 3.7714,
      "step": 87580
    },
    {
      "epoch": 0.08164805275476536,
      "grad_norm": 1.0608259439468384,
      "learning_rate": 0.000183670589200355,
      "loss": 3.665,
      "step": 87590
    },
    {
      "epoch": 0.08165737437284445,
      "grad_norm": 1.2800042629241943,
      "learning_rate": 0.00018366872487825977,
      "loss": 3.9108,
      "step": 87600
    },
    {
      "epoch": 0.08166669599092354,
      "grad_norm": 1.0402580499649048,
      "learning_rate": 0.00018366686055616458,
      "loss": 3.7651,
      "step": 87610
    },
    {
      "epoch": 0.08167601760900263,
      "grad_norm": 1.0772905349731445,
      "learning_rate": 0.00018366499623406938,
      "loss": 3.8838,
      "step": 87620
    },
    {
      "epoch": 0.08168533922708172,
      "grad_norm": 1.16278874874115,
      "learning_rate": 0.00018366313191197416,
      "loss": 3.8715,
      "step": 87630
    },
    {
      "epoch": 0.08169466084516082,
      "grad_norm": 1.0169214010238647,
      "learning_rate": 0.000183661267589879,
      "loss": 3.8041,
      "step": 87640
    },
    {
      "epoch": 0.08170398246323991,
      "grad_norm": 1.280348539352417,
      "learning_rate": 0.00018365940326778377,
      "loss": 3.8241,
      "step": 87650
    },
    {
      "epoch": 0.081713304081319,
      "grad_norm": 1.0258538722991943,
      "learning_rate": 0.00018365753894568858,
      "loss": 3.8941,
      "step": 87660
    },
    {
      "epoch": 0.08172262569939809,
      "grad_norm": 0.9059931635856628,
      "learning_rate": 0.0001836556746235934,
      "loss": 3.8031,
      "step": 87670
    },
    {
      "epoch": 0.08173194731747718,
      "grad_norm": 1.1436008214950562,
      "learning_rate": 0.00018365381030149817,
      "loss": 3.8118,
      "step": 87680
    },
    {
      "epoch": 0.08174126893555628,
      "grad_norm": 1.3615731000900269,
      "learning_rate": 0.00018365194597940297,
      "loss": 3.7576,
      "step": 87690
    },
    {
      "epoch": 0.08175059055363537,
      "grad_norm": 1.1383004188537598,
      "learning_rate": 0.00018365008165730778,
      "loss": 3.7501,
      "step": 87700
    },
    {
      "epoch": 0.08175991217171445,
      "grad_norm": 0.9545424580574036,
      "learning_rate": 0.00018364821733521259,
      "loss": 3.7811,
      "step": 87710
    },
    {
      "epoch": 0.08176923378979355,
      "grad_norm": 1.2153741121292114,
      "learning_rate": 0.0001836463530131174,
      "loss": 3.6723,
      "step": 87720
    },
    {
      "epoch": 0.08177855540787264,
      "grad_norm": 1.0695600509643555,
      "learning_rate": 0.00018364448869102217,
      "loss": 3.8206,
      "step": 87730
    },
    {
      "epoch": 0.08178787702595174,
      "grad_norm": 1.0470106601715088,
      "learning_rate": 0.00018364262436892698,
      "loss": 3.9186,
      "step": 87740
    },
    {
      "epoch": 0.08179719864403083,
      "grad_norm": 0.9001947641372681,
      "learning_rate": 0.00018364076004683178,
      "loss": 3.8082,
      "step": 87750
    },
    {
      "epoch": 0.08180652026210992,
      "grad_norm": 1.0914478302001953,
      "learning_rate": 0.0001836388957247366,
      "loss": 3.724,
      "step": 87760
    },
    {
      "epoch": 0.081815841880189,
      "grad_norm": 0.9869974255561829,
      "learning_rate": 0.0001836370314026414,
      "loss": 3.9711,
      "step": 87770
    },
    {
      "epoch": 0.0818251634982681,
      "grad_norm": 1.1372376680374146,
      "learning_rate": 0.00018363516708054618,
      "loss": 3.8592,
      "step": 87780
    },
    {
      "epoch": 0.0818344851163472,
      "grad_norm": 1.1308425664901733,
      "learning_rate": 0.00018363330275845098,
      "loss": 3.8138,
      "step": 87790
    },
    {
      "epoch": 0.08184380673442629,
      "grad_norm": 0.9661073088645935,
      "learning_rate": 0.00018363143843635576,
      "loss": 3.8179,
      "step": 87800
    },
    {
      "epoch": 0.08185312835250538,
      "grad_norm": 1.0942655801773071,
      "learning_rate": 0.0001836295741142606,
      "loss": 3.7075,
      "step": 87810
    },
    {
      "epoch": 0.08186244997058446,
      "grad_norm": 1.101828932762146,
      "learning_rate": 0.0001836277097921654,
      "loss": 3.5574,
      "step": 87820
    },
    {
      "epoch": 0.08187177158866356,
      "grad_norm": 1.182621955871582,
      "learning_rate": 0.00018362584547007018,
      "loss": 3.8124,
      "step": 87830
    },
    {
      "epoch": 0.08188109320674265,
      "grad_norm": 1.201473593711853,
      "learning_rate": 0.00018362398114797499,
      "loss": 3.9161,
      "step": 87840
    },
    {
      "epoch": 0.08189041482482175,
      "grad_norm": 0.9057927131652832,
      "learning_rate": 0.00018362211682587977,
      "loss": 3.918,
      "step": 87850
    },
    {
      "epoch": 0.08189973644290084,
      "grad_norm": 1.2119102478027344,
      "learning_rate": 0.00018362025250378457,
      "loss": 3.7183,
      "step": 87860
    },
    {
      "epoch": 0.08190905806097992,
      "grad_norm": 1.2249269485473633,
      "learning_rate": 0.0001836183881816894,
      "loss": 3.7705,
      "step": 87870
    },
    {
      "epoch": 0.08191837967905902,
      "grad_norm": 1.0536214113235474,
      "learning_rate": 0.00018361652385959418,
      "loss": 3.608,
      "step": 87880
    },
    {
      "epoch": 0.08192770129713811,
      "grad_norm": 1.1568231582641602,
      "learning_rate": 0.000183614659537499,
      "loss": 3.7378,
      "step": 87890
    },
    {
      "epoch": 0.0819370229152172,
      "grad_norm": 0.9514676928520203,
      "learning_rate": 0.00018361279521540377,
      "loss": 4.0071,
      "step": 87900
    },
    {
      "epoch": 0.0819463445332963,
      "grad_norm": 0.9872128367424011,
      "learning_rate": 0.00018361093089330858,
      "loss": 3.6438,
      "step": 87910
    },
    {
      "epoch": 0.08195566615137538,
      "grad_norm": 1.0296813249588013,
      "learning_rate": 0.00018360906657121338,
      "loss": 3.7698,
      "step": 87920
    },
    {
      "epoch": 0.08196498776945448,
      "grad_norm": 0.9148175716400146,
      "learning_rate": 0.0001836072022491182,
      "loss": 4.0983,
      "step": 87930
    },
    {
      "epoch": 0.08197430938753357,
      "grad_norm": 1.111915946006775,
      "learning_rate": 0.000183605337927023,
      "loss": 3.7193,
      "step": 87940
    },
    {
      "epoch": 0.08198363100561266,
      "grad_norm": 1.0607082843780518,
      "learning_rate": 0.00018360347360492777,
      "loss": 3.8062,
      "step": 87950
    },
    {
      "epoch": 0.08199295262369176,
      "grad_norm": 1.5762559175491333,
      "learning_rate": 0.00018360160928283258,
      "loss": 3.5202,
      "step": 87960
    },
    {
      "epoch": 0.08200227424177085,
      "grad_norm": 1.460376501083374,
      "learning_rate": 0.0001835997449607374,
      "loss": 3.9201,
      "step": 87970
    },
    {
      "epoch": 0.08201159585984993,
      "grad_norm": 1.0631998777389526,
      "learning_rate": 0.0001835978806386422,
      "loss": 3.6868,
      "step": 87980
    },
    {
      "epoch": 0.08202091747792903,
      "grad_norm": 1.031119704246521,
      "learning_rate": 0.000183596016316547,
      "loss": 3.7831,
      "step": 87990
    },
    {
      "epoch": 0.08203023909600812,
      "grad_norm": 1.2752383947372437,
      "learning_rate": 0.00018359415199445178,
      "loss": 3.7485,
      "step": 88000
    },
    {
      "epoch": 0.08203956071408722,
      "grad_norm": 0.9173330068588257,
      "learning_rate": 0.00018359228767235658,
      "loss": 3.9451,
      "step": 88010
    },
    {
      "epoch": 0.08204888233216631,
      "grad_norm": 0.8877754211425781,
      "learning_rate": 0.00018359042335026136,
      "loss": 3.6935,
      "step": 88020
    },
    {
      "epoch": 0.08205820395024539,
      "grad_norm": 1.1131688356399536,
      "learning_rate": 0.0001835885590281662,
      "loss": 3.8301,
      "step": 88030
    },
    {
      "epoch": 0.08206752556832449,
      "grad_norm": 1.0680524110794067,
      "learning_rate": 0.000183586694706071,
      "loss": 3.8295,
      "step": 88040
    },
    {
      "epoch": 0.08207684718640358,
      "grad_norm": 1.1323411464691162,
      "learning_rate": 0.00018358483038397578,
      "loss": 3.6938,
      "step": 88050
    },
    {
      "epoch": 0.08208616880448268,
      "grad_norm": 1.1155723333358765,
      "learning_rate": 0.0001835829660618806,
      "loss": 3.7665,
      "step": 88060
    },
    {
      "epoch": 0.08209549042256177,
      "grad_norm": 1.286232829093933,
      "learning_rate": 0.00018358110173978537,
      "loss": 3.6636,
      "step": 88070
    },
    {
      "epoch": 0.08210481204064085,
      "grad_norm": 0.9877673387527466,
      "learning_rate": 0.00018357923741769017,
      "loss": 3.757,
      "step": 88080
    },
    {
      "epoch": 0.08211413365871995,
      "grad_norm": 1.2364414930343628,
      "learning_rate": 0.000183577373095595,
      "loss": 3.74,
      "step": 88090
    },
    {
      "epoch": 0.08212345527679904,
      "grad_norm": 1.0340639352798462,
      "learning_rate": 0.0001835755087734998,
      "loss": 3.6688,
      "step": 88100
    },
    {
      "epoch": 0.08213277689487813,
      "grad_norm": 1.0174978971481323,
      "learning_rate": 0.0001835736444514046,
      "loss": 3.7498,
      "step": 88110
    },
    {
      "epoch": 0.08214209851295723,
      "grad_norm": 1.1439777612686157,
      "learning_rate": 0.00018357178012930937,
      "loss": 3.8239,
      "step": 88120
    },
    {
      "epoch": 0.08215142013103631,
      "grad_norm": 1.1314817667007446,
      "learning_rate": 0.00018356991580721418,
      "loss": 3.7944,
      "step": 88130
    },
    {
      "epoch": 0.0821607417491154,
      "grad_norm": 0.9648788571357727,
      "learning_rate": 0.00018356805148511899,
      "loss": 3.8342,
      "step": 88140
    },
    {
      "epoch": 0.0821700633671945,
      "grad_norm": 0.9246693849563599,
      "learning_rate": 0.0001835661871630238,
      "loss": 3.7295,
      "step": 88150
    },
    {
      "epoch": 0.08217938498527359,
      "grad_norm": 1.1383540630340576,
      "learning_rate": 0.0001835643228409286,
      "loss": 3.8631,
      "step": 88160
    },
    {
      "epoch": 0.08218870660335269,
      "grad_norm": 1.0688496828079224,
      "learning_rate": 0.00018356245851883338,
      "loss": 3.7203,
      "step": 88170
    },
    {
      "epoch": 0.08219802822143177,
      "grad_norm": 1.1404973268508911,
      "learning_rate": 0.00018356059419673818,
      "loss": 3.8446,
      "step": 88180
    },
    {
      "epoch": 0.08220734983951086,
      "grad_norm": 1.2200199365615845,
      "learning_rate": 0.000183558729874643,
      "loss": 3.7464,
      "step": 88190
    },
    {
      "epoch": 0.08221667145758996,
      "grad_norm": 0.9911855459213257,
      "learning_rate": 0.0001835568655525478,
      "loss": 3.8303,
      "step": 88200
    },
    {
      "epoch": 0.08222599307566905,
      "grad_norm": 1.2118619680404663,
      "learning_rate": 0.0001835550012304526,
      "loss": 3.954,
      "step": 88210
    },
    {
      "epoch": 0.08223531469374815,
      "grad_norm": 0.9438069462776184,
      "learning_rate": 0.00018355313690835738,
      "loss": 3.7762,
      "step": 88220
    },
    {
      "epoch": 0.08224463631182724,
      "grad_norm": 1.0676623582839966,
      "learning_rate": 0.0001835512725862622,
      "loss": 3.6038,
      "step": 88230
    },
    {
      "epoch": 0.08225395792990632,
      "grad_norm": 0.9757475256919861,
      "learning_rate": 0.000183549408264167,
      "loss": 3.6268,
      "step": 88240
    },
    {
      "epoch": 0.08226327954798542,
      "grad_norm": 0.9948646426200867,
      "learning_rate": 0.00018354754394207177,
      "loss": 3.7905,
      "step": 88250
    },
    {
      "epoch": 0.08227260116606451,
      "grad_norm": 0.9417844414710999,
      "learning_rate": 0.0001835456796199766,
      "loss": 3.9521,
      "step": 88260
    },
    {
      "epoch": 0.0822819227841436,
      "grad_norm": 1.3388129472732544,
      "learning_rate": 0.00018354381529788139,
      "loss": 3.7659,
      "step": 88270
    },
    {
      "epoch": 0.0822912444022227,
      "grad_norm": 1.0382821559906006,
      "learning_rate": 0.0001835419509757862,
      "loss": 3.8808,
      "step": 88280
    },
    {
      "epoch": 0.08230056602030178,
      "grad_norm": 0.997255802154541,
      "learning_rate": 0.000183540086653691,
      "loss": 3.8416,
      "step": 88290
    },
    {
      "epoch": 0.08230988763838087,
      "grad_norm": 1.0686531066894531,
      "learning_rate": 0.00018353822233159578,
      "loss": 3.6364,
      "step": 88300
    },
    {
      "epoch": 0.08231920925645997,
      "grad_norm": 1.1059269905090332,
      "learning_rate": 0.00018353635800950058,
      "loss": 3.9126,
      "step": 88310
    },
    {
      "epoch": 0.08232853087453906,
      "grad_norm": 1.0066113471984863,
      "learning_rate": 0.0001835344936874054,
      "loss": 3.7928,
      "step": 88320
    },
    {
      "epoch": 0.08233785249261816,
      "grad_norm": 0.8655222654342651,
      "learning_rate": 0.0001835326293653102,
      "loss": 3.7258,
      "step": 88330
    },
    {
      "epoch": 0.08234717411069724,
      "grad_norm": 1.0548155307769775,
      "learning_rate": 0.000183530765043215,
      "loss": 3.8063,
      "step": 88340
    },
    {
      "epoch": 0.08235649572877633,
      "grad_norm": 1.1125612258911133,
      "learning_rate": 0.00018352890072111978,
      "loss": 3.8953,
      "step": 88350
    },
    {
      "epoch": 0.08236581734685543,
      "grad_norm": 0.9375595450401306,
      "learning_rate": 0.0001835270363990246,
      "loss": 3.8415,
      "step": 88360
    },
    {
      "epoch": 0.08237513896493452,
      "grad_norm": 1.2263931035995483,
      "learning_rate": 0.0001835251720769294,
      "loss": 3.6191,
      "step": 88370
    },
    {
      "epoch": 0.08238446058301362,
      "grad_norm": 0.8738211393356323,
      "learning_rate": 0.0001835233077548342,
      "loss": 3.9079,
      "step": 88380
    },
    {
      "epoch": 0.0823937822010927,
      "grad_norm": 1.0857397317886353,
      "learning_rate": 0.000183521443432739,
      "loss": 3.7109,
      "step": 88390
    },
    {
      "epoch": 0.08240310381917179,
      "grad_norm": 1.069955825805664,
      "learning_rate": 0.0001835195791106438,
      "loss": 3.7447,
      "step": 88400
    },
    {
      "epoch": 0.08241242543725089,
      "grad_norm": 0.9373009204864502,
      "learning_rate": 0.0001835177147885486,
      "loss": 3.6985,
      "step": 88410
    },
    {
      "epoch": 0.08242174705532998,
      "grad_norm": 0.9863550066947937,
      "learning_rate": 0.0001835158504664534,
      "loss": 3.8542,
      "step": 88420
    },
    {
      "epoch": 0.08243106867340907,
      "grad_norm": 0.869409441947937,
      "learning_rate": 0.0001835139861443582,
      "loss": 3.7345,
      "step": 88430
    },
    {
      "epoch": 0.08244039029148817,
      "grad_norm": 1.1544854640960693,
      "learning_rate": 0.000183512121822263,
      "loss": 3.9176,
      "step": 88440
    },
    {
      "epoch": 0.08244971190956725,
      "grad_norm": 1.1453007459640503,
      "learning_rate": 0.0001835102575001678,
      "loss": 3.8315,
      "step": 88450
    },
    {
      "epoch": 0.08245903352764634,
      "grad_norm": 1.0952019691467285,
      "learning_rate": 0.0001835083931780726,
      "loss": 3.6819,
      "step": 88460
    },
    {
      "epoch": 0.08246835514572544,
      "grad_norm": 1.001476526260376,
      "learning_rate": 0.00018350652885597738,
      "loss": 3.8096,
      "step": 88470
    },
    {
      "epoch": 0.08247767676380453,
      "grad_norm": 1.0210764408111572,
      "learning_rate": 0.0001835046645338822,
      "loss": 3.7683,
      "step": 88480
    },
    {
      "epoch": 0.08248699838188363,
      "grad_norm": 1.1637413501739502,
      "learning_rate": 0.00018350280021178702,
      "loss": 3.589,
      "step": 88490
    },
    {
      "epoch": 0.08249631999996271,
      "grad_norm": 1.0937833786010742,
      "learning_rate": 0.0001835009358896918,
      "loss": 3.7321,
      "step": 88500
    },
    {
      "epoch": 0.0825056416180418,
      "grad_norm": 0.8420345187187195,
      "learning_rate": 0.0001834990715675966,
      "loss": 3.8975,
      "step": 88510
    },
    {
      "epoch": 0.0825149632361209,
      "grad_norm": 1.2616578340530396,
      "learning_rate": 0.00018349720724550138,
      "loss": 3.807,
      "step": 88520
    },
    {
      "epoch": 0.08252428485419999,
      "grad_norm": 1.0722731351852417,
      "learning_rate": 0.0001834953429234062,
      "loss": 3.784,
      "step": 88530
    },
    {
      "epoch": 0.08253360647227909,
      "grad_norm": 1.0367686748504639,
      "learning_rate": 0.00018349347860131102,
      "loss": 3.7732,
      "step": 88540
    },
    {
      "epoch": 0.08254292809035817,
      "grad_norm": 0.890116810798645,
      "learning_rate": 0.0001834916142792158,
      "loss": 3.9096,
      "step": 88550
    },
    {
      "epoch": 0.08255224970843726,
      "grad_norm": 0.978173017501831,
      "learning_rate": 0.0001834897499571206,
      "loss": 3.7411,
      "step": 88560
    },
    {
      "epoch": 0.08256157132651636,
      "grad_norm": 1.134961724281311,
      "learning_rate": 0.00018348788563502539,
      "loss": 3.725,
      "step": 88570
    },
    {
      "epoch": 0.08257089294459545,
      "grad_norm": 0.8762019872665405,
      "learning_rate": 0.0001834860213129302,
      "loss": 3.7547,
      "step": 88580
    },
    {
      "epoch": 0.08258021456267454,
      "grad_norm": 1.1219862699508667,
      "learning_rate": 0.000183484156990835,
      "loss": 3.9075,
      "step": 88590
    },
    {
      "epoch": 0.08258953618075363,
      "grad_norm": 1.1431068181991577,
      "learning_rate": 0.0001834822926687398,
      "loss": 3.7931,
      "step": 88600
    },
    {
      "epoch": 0.08259885779883272,
      "grad_norm": 0.8346309065818787,
      "learning_rate": 0.0001834804283466446,
      "loss": 3.8935,
      "step": 88610
    },
    {
      "epoch": 0.08260817941691181,
      "grad_norm": 1.180863857269287,
      "learning_rate": 0.0001834785640245494,
      "loss": 3.7073,
      "step": 88620
    },
    {
      "epoch": 0.08261750103499091,
      "grad_norm": 1.0554766654968262,
      "learning_rate": 0.0001834766997024542,
      "loss": 3.6734,
      "step": 88630
    },
    {
      "epoch": 0.08262682265307,
      "grad_norm": 1.3002831935882568,
      "learning_rate": 0.000183474835380359,
      "loss": 3.6024,
      "step": 88640
    },
    {
      "epoch": 0.0826361442711491,
      "grad_norm": 1.236000657081604,
      "learning_rate": 0.0001834729710582638,
      "loss": 3.8328,
      "step": 88650
    },
    {
      "epoch": 0.08264546588922818,
      "grad_norm": 1.3297011852264404,
      "learning_rate": 0.00018347110673616862,
      "loss": 3.9487,
      "step": 88660
    },
    {
      "epoch": 0.08265478750730727,
      "grad_norm": 1.0165435075759888,
      "learning_rate": 0.0001834692424140734,
      "loss": 3.7435,
      "step": 88670
    },
    {
      "epoch": 0.08266410912538637,
      "grad_norm": 0.9814731478691101,
      "learning_rate": 0.0001834673780919782,
      "loss": 3.7761,
      "step": 88680
    },
    {
      "epoch": 0.08267343074346546,
      "grad_norm": 1.0977052450180054,
      "learning_rate": 0.000183465513769883,
      "loss": 3.9151,
      "step": 88690
    },
    {
      "epoch": 0.08268275236154456,
      "grad_norm": 1.1024372577667236,
      "learning_rate": 0.00018346364944778779,
      "loss": 3.6887,
      "step": 88700
    },
    {
      "epoch": 0.08269207397962364,
      "grad_norm": 0.8687765002250671,
      "learning_rate": 0.00018346178512569262,
      "loss": 3.9951,
      "step": 88710
    },
    {
      "epoch": 0.08270139559770273,
      "grad_norm": 1.0423399209976196,
      "learning_rate": 0.0001834599208035974,
      "loss": 3.7474,
      "step": 88720
    },
    {
      "epoch": 0.08271071721578183,
      "grad_norm": 1.1244417428970337,
      "learning_rate": 0.0001834580564815022,
      "loss": 3.728,
      "step": 88730
    },
    {
      "epoch": 0.08272003883386092,
      "grad_norm": 0.9458192586898804,
      "learning_rate": 0.000183456192159407,
      "loss": 3.983,
      "step": 88740
    },
    {
      "epoch": 0.08272936045194001,
      "grad_norm": 1.0982624292373657,
      "learning_rate": 0.0001834543278373118,
      "loss": 3.7145,
      "step": 88750
    },
    {
      "epoch": 0.0827386820700191,
      "grad_norm": 1.2170439958572388,
      "learning_rate": 0.0001834524635152166,
      "loss": 3.839,
      "step": 88760
    },
    {
      "epoch": 0.08274800368809819,
      "grad_norm": 0.7552831172943115,
      "learning_rate": 0.0001834505991931214,
      "loss": 3.7369,
      "step": 88770
    },
    {
      "epoch": 0.08275732530617728,
      "grad_norm": 1.2647005319595337,
      "learning_rate": 0.0001834487348710262,
      "loss": 3.7502,
      "step": 88780
    },
    {
      "epoch": 0.08276664692425638,
      "grad_norm": 1.1322062015533447,
      "learning_rate": 0.00018344687054893102,
      "loss": 3.9593,
      "step": 88790
    },
    {
      "epoch": 0.08277596854233547,
      "grad_norm": 1.082158088684082,
      "learning_rate": 0.0001834450062268358,
      "loss": 3.6891,
      "step": 88800
    },
    {
      "epoch": 0.08278529016041455,
      "grad_norm": 1.1518652439117432,
      "learning_rate": 0.0001834431419047406,
      "loss": 3.8907,
      "step": 88810
    },
    {
      "epoch": 0.08279461177849365,
      "grad_norm": 1.0212591886520386,
      "learning_rate": 0.0001834412775826454,
      "loss": 3.6942,
      "step": 88820
    },
    {
      "epoch": 0.08280393339657274,
      "grad_norm": 0.9406482577323914,
      "learning_rate": 0.00018343941326055021,
      "loss": 3.8922,
      "step": 88830
    },
    {
      "epoch": 0.08281325501465184,
      "grad_norm": 1.1551395654678345,
      "learning_rate": 0.00018343754893845502,
      "loss": 3.7051,
      "step": 88840
    },
    {
      "epoch": 0.08282257663273093,
      "grad_norm": 1.0697954893112183,
      "learning_rate": 0.0001834356846163598,
      "loss": 3.8934,
      "step": 88850
    },
    {
      "epoch": 0.08283189825081001,
      "grad_norm": 1.0874239206314087,
      "learning_rate": 0.0001834338202942646,
      "loss": 3.9676,
      "step": 88860
    },
    {
      "epoch": 0.08284121986888911,
      "grad_norm": 1.1732476949691772,
      "learning_rate": 0.0001834319559721694,
      "loss": 3.8683,
      "step": 88870
    },
    {
      "epoch": 0.0828505414869682,
      "grad_norm": 0.967695415019989,
      "learning_rate": 0.00018343009165007422,
      "loss": 3.584,
      "step": 88880
    },
    {
      "epoch": 0.0828598631050473,
      "grad_norm": 0.951021134853363,
      "learning_rate": 0.00018342822732797902,
      "loss": 3.5889,
      "step": 88890
    },
    {
      "epoch": 0.08286918472312639,
      "grad_norm": 1.0591014623641968,
      "learning_rate": 0.0001834263630058838,
      "loss": 3.7504,
      "step": 88900
    },
    {
      "epoch": 0.08287850634120549,
      "grad_norm": 0.9796790480613708,
      "learning_rate": 0.0001834244986837886,
      "loss": 3.7518,
      "step": 88910
    },
    {
      "epoch": 0.08288782795928457,
      "grad_norm": 0.9941726326942444,
      "learning_rate": 0.0001834226343616934,
      "loss": 3.7536,
      "step": 88920
    },
    {
      "epoch": 0.08289714957736366,
      "grad_norm": 1.150806188583374,
      "learning_rate": 0.00018342077003959822,
      "loss": 3.7023,
      "step": 88930
    },
    {
      "epoch": 0.08290647119544275,
      "grad_norm": 0.9749338030815125,
      "learning_rate": 0.00018341890571750303,
      "loss": 3.9412,
      "step": 88940
    },
    {
      "epoch": 0.08291579281352185,
      "grad_norm": 1.1354751586914062,
      "learning_rate": 0.0001834170413954078,
      "loss": 3.775,
      "step": 88950
    },
    {
      "epoch": 0.08292511443160094,
      "grad_norm": 0.991527795791626,
      "learning_rate": 0.00018341517707331261,
      "loss": 3.8435,
      "step": 88960
    },
    {
      "epoch": 0.08293443604968002,
      "grad_norm": 1.0700557231903076,
      "learning_rate": 0.0001834133127512174,
      "loss": 3.8186,
      "step": 88970
    },
    {
      "epoch": 0.08294375766775912,
      "grad_norm": 1.1595730781555176,
      "learning_rate": 0.0001834114484291222,
      "loss": 3.6796,
      "step": 88980
    },
    {
      "epoch": 0.08295307928583821,
      "grad_norm": 1.010259747505188,
      "learning_rate": 0.00018340958410702703,
      "loss": 3.6182,
      "step": 88990
    },
    {
      "epoch": 0.08296240090391731,
      "grad_norm": 0.9940686821937561,
      "learning_rate": 0.0001834077197849318,
      "loss": 3.6644,
      "step": 89000
    },
    {
      "epoch": 0.0829717225219964,
      "grad_norm": 0.9071727991104126,
      "learning_rate": 0.00018340585546283662,
      "loss": 3.9761,
      "step": 89010
    },
    {
      "epoch": 0.08298104414007548,
      "grad_norm": 1.067824363708496,
      "learning_rate": 0.0001834039911407414,
      "loss": 3.778,
      "step": 89020
    },
    {
      "epoch": 0.08299036575815458,
      "grad_norm": 1.1928831338882446,
      "learning_rate": 0.0001834021268186462,
      "loss": 3.7326,
      "step": 89030
    },
    {
      "epoch": 0.08299968737623367,
      "grad_norm": 0.979224443435669,
      "learning_rate": 0.000183400262496551,
      "loss": 3.7921,
      "step": 89040
    },
    {
      "epoch": 0.08300900899431277,
      "grad_norm": 0.8316656351089478,
      "learning_rate": 0.00018339839817445582,
      "loss": 3.8038,
      "step": 89050
    },
    {
      "epoch": 0.08301833061239186,
      "grad_norm": 1.082610011100769,
      "learning_rate": 0.00018339653385236062,
      "loss": 3.9313,
      "step": 89060
    },
    {
      "epoch": 0.08302765223047094,
      "grad_norm": 1.2927011251449585,
      "learning_rate": 0.0001833946695302654,
      "loss": 3.6388,
      "step": 89070
    },
    {
      "epoch": 0.08303697384855004,
      "grad_norm": 1.1897625923156738,
      "learning_rate": 0.0001833928052081702,
      "loss": 3.6765,
      "step": 89080
    },
    {
      "epoch": 0.08304629546662913,
      "grad_norm": 1.0410223007202148,
      "learning_rate": 0.00018339094088607502,
      "loss": 3.4404,
      "step": 89090
    },
    {
      "epoch": 0.08305561708470822,
      "grad_norm": 1.0925241708755493,
      "learning_rate": 0.00018338907656397982,
      "loss": 3.6406,
      "step": 89100
    },
    {
      "epoch": 0.08306493870278732,
      "grad_norm": 1.2766426801681519,
      "learning_rate": 0.00018338721224188463,
      "loss": 3.7139,
      "step": 89110
    },
    {
      "epoch": 0.08307426032086641,
      "grad_norm": 1.085628628730774,
      "learning_rate": 0.0001833853479197894,
      "loss": 3.708,
      "step": 89120
    },
    {
      "epoch": 0.0830835819389455,
      "grad_norm": 0.9918851852416992,
      "learning_rate": 0.00018338348359769421,
      "loss": 3.7779,
      "step": 89130
    },
    {
      "epoch": 0.08309290355702459,
      "grad_norm": 1.0304508209228516,
      "learning_rate": 0.00018338161927559902,
      "loss": 3.7812,
      "step": 89140
    },
    {
      "epoch": 0.08310222517510368,
      "grad_norm": 1.10287344455719,
      "learning_rate": 0.00018337975495350383,
      "loss": 3.7153,
      "step": 89150
    },
    {
      "epoch": 0.08311154679318278,
      "grad_norm": 0.9399433732032776,
      "learning_rate": 0.00018337789063140863,
      "loss": 3.6242,
      "step": 89160
    },
    {
      "epoch": 0.08312086841126187,
      "grad_norm": 1.3131870031356812,
      "learning_rate": 0.0001833760263093134,
      "loss": 3.8791,
      "step": 89170
    },
    {
      "epoch": 0.08313019002934095,
      "grad_norm": 0.9818447828292847,
      "learning_rate": 0.00018337416198721822,
      "loss": 3.8389,
      "step": 89180
    },
    {
      "epoch": 0.08313951164742005,
      "grad_norm": 1.208916187286377,
      "learning_rate": 0.00018337229766512302,
      "loss": 3.7961,
      "step": 89190
    },
    {
      "epoch": 0.08314883326549914,
      "grad_norm": 0.9979801774024963,
      "learning_rate": 0.0001833704333430278,
      "loss": 3.6652,
      "step": 89200
    },
    {
      "epoch": 0.08315815488357824,
      "grad_norm": 0.9683908820152283,
      "learning_rate": 0.0001833685690209326,
      "loss": 3.6903,
      "step": 89210
    },
    {
      "epoch": 0.08316747650165733,
      "grad_norm": 1.0774136781692505,
      "learning_rate": 0.00018336670469883742,
      "loss": 3.9046,
      "step": 89220
    },
    {
      "epoch": 0.08317679811973641,
      "grad_norm": 1.149869441986084,
      "learning_rate": 0.00018336484037674222,
      "loss": 3.7669,
      "step": 89230
    },
    {
      "epoch": 0.0831861197378155,
      "grad_norm": 0.8978118896484375,
      "learning_rate": 0.00018336297605464703,
      "loss": 3.8586,
      "step": 89240
    },
    {
      "epoch": 0.0831954413558946,
      "grad_norm": 0.975687563419342,
      "learning_rate": 0.0001833611117325518,
      "loss": 3.8149,
      "step": 89250
    },
    {
      "epoch": 0.0832047629739737,
      "grad_norm": 1.1780762672424316,
      "learning_rate": 0.00018335924741045661,
      "loss": 3.8646,
      "step": 89260
    },
    {
      "epoch": 0.08321408459205279,
      "grad_norm": 1.0273239612579346,
      "learning_rate": 0.00018335738308836142,
      "loss": 3.6839,
      "step": 89270
    },
    {
      "epoch": 0.08322340621013187,
      "grad_norm": 1.1586090326309204,
      "learning_rate": 0.00018335551876626623,
      "loss": 3.6703,
      "step": 89280
    },
    {
      "epoch": 0.08323272782821096,
      "grad_norm": 1.231189489364624,
      "learning_rate": 0.00018335365444417103,
      "loss": 4.0498,
      "step": 89290
    },
    {
      "epoch": 0.08324204944629006,
      "grad_norm": 1.1648848056793213,
      "learning_rate": 0.0001833517901220758,
      "loss": 3.8097,
      "step": 89300
    },
    {
      "epoch": 0.08325137106436915,
      "grad_norm": 1.2129104137420654,
      "learning_rate": 0.00018334992579998062,
      "loss": 3.826,
      "step": 89310
    },
    {
      "epoch": 0.08326069268244825,
      "grad_norm": 1.1121364831924438,
      "learning_rate": 0.00018334806147788543,
      "loss": 3.5963,
      "step": 89320
    },
    {
      "epoch": 0.08327001430052734,
      "grad_norm": 1.0576249361038208,
      "learning_rate": 0.00018334619715579023,
      "loss": 3.6736,
      "step": 89330
    },
    {
      "epoch": 0.08327933591860642,
      "grad_norm": 1.1788945198059082,
      "learning_rate": 0.00018334433283369504,
      "loss": 3.7828,
      "step": 89340
    },
    {
      "epoch": 0.08328865753668552,
      "grad_norm": 0.9107685089111328,
      "learning_rate": 0.00018334246851159982,
      "loss": 3.7978,
      "step": 89350
    },
    {
      "epoch": 0.08329797915476461,
      "grad_norm": 0.9888318181037903,
      "learning_rate": 0.00018334060418950462,
      "loss": 3.947,
      "step": 89360
    },
    {
      "epoch": 0.0833073007728437,
      "grad_norm": 1.0027799606323242,
      "learning_rate": 0.0001833387398674094,
      "loss": 3.8649,
      "step": 89370
    },
    {
      "epoch": 0.0833166223909228,
      "grad_norm": 1.237363576889038,
      "learning_rate": 0.00018333687554531424,
      "loss": 3.3992,
      "step": 89380
    },
    {
      "epoch": 0.08332594400900188,
      "grad_norm": 1.0837879180908203,
      "learning_rate": 0.00018333501122321902,
      "loss": 3.5636,
      "step": 89390
    },
    {
      "epoch": 0.08333526562708098,
      "grad_norm": 1.1015316247940063,
      "learning_rate": 0.00018333314690112382,
      "loss": 3.544,
      "step": 89400
    },
    {
      "epoch": 0.08334458724516007,
      "grad_norm": 1.1179119348526,
      "learning_rate": 0.00018333128257902863,
      "loss": 3.7044,
      "step": 89410
    },
    {
      "epoch": 0.08335390886323917,
      "grad_norm": 1.06252920627594,
      "learning_rate": 0.0001833294182569334,
      "loss": 4.0226,
      "step": 89420
    },
    {
      "epoch": 0.08336323048131826,
      "grad_norm": 1.220749020576477,
      "learning_rate": 0.0001833275539348382,
      "loss": 3.9392,
      "step": 89430
    },
    {
      "epoch": 0.08337255209939734,
      "grad_norm": 1.169368863105774,
      "learning_rate": 0.00018332568961274302,
      "loss": 3.7403,
      "step": 89440
    },
    {
      "epoch": 0.08338187371747643,
      "grad_norm": 1.0773814916610718,
      "learning_rate": 0.00018332382529064783,
      "loss": 3.8086,
      "step": 89450
    },
    {
      "epoch": 0.08339119533555553,
      "grad_norm": 1.143061876296997,
      "learning_rate": 0.00018332196096855263,
      "loss": 3.6146,
      "step": 89460
    },
    {
      "epoch": 0.08340051695363462,
      "grad_norm": 1.1379722356796265,
      "learning_rate": 0.0001833200966464574,
      "loss": 3.8028,
      "step": 89470
    },
    {
      "epoch": 0.08340983857171372,
      "grad_norm": 1.2364321947097778,
      "learning_rate": 0.00018331823232436222,
      "loss": 3.6561,
      "step": 89480
    },
    {
      "epoch": 0.0834191601897928,
      "grad_norm": 1.0680657625198364,
      "learning_rate": 0.00018331636800226702,
      "loss": 3.9059,
      "step": 89490
    },
    {
      "epoch": 0.0834284818078719,
      "grad_norm": 1.0744482278823853,
      "learning_rate": 0.00018331450368017183,
      "loss": 3.9166,
      "step": 89500
    },
    {
      "epoch": 0.08343780342595099,
      "grad_norm": 1.1839284896850586,
      "learning_rate": 0.00018331263935807664,
      "loss": 3.9625,
      "step": 89510
    },
    {
      "epoch": 0.08344712504403008,
      "grad_norm": 1.1350966691970825,
      "learning_rate": 0.00018331077503598142,
      "loss": 3.6156,
      "step": 89520
    },
    {
      "epoch": 0.08345644666210918,
      "grad_norm": 1.2867909669876099,
      "learning_rate": 0.00018330891071388622,
      "loss": 3.6837,
      "step": 89530
    },
    {
      "epoch": 0.08346576828018826,
      "grad_norm": 0.9711264967918396,
      "learning_rate": 0.00018330704639179103,
      "loss": 3.7267,
      "step": 89540
    },
    {
      "epoch": 0.08347508989826735,
      "grad_norm": 0.9662922024726868,
      "learning_rate": 0.00018330518206969583,
      "loss": 3.7905,
      "step": 89550
    },
    {
      "epoch": 0.08348441151634645,
      "grad_norm": 1.1165534257888794,
      "learning_rate": 0.00018330331774760064,
      "loss": 3.7718,
      "step": 89560
    },
    {
      "epoch": 0.08349373313442554,
      "grad_norm": 1.2157349586486816,
      "learning_rate": 0.00018330145342550542,
      "loss": 3.684,
      "step": 89570
    },
    {
      "epoch": 0.08350305475250464,
      "grad_norm": 1.0854103565216064,
      "learning_rate": 0.00018329958910341023,
      "loss": 3.8456,
      "step": 89580
    },
    {
      "epoch": 0.08351237637058373,
      "grad_norm": 0.9658782482147217,
      "learning_rate": 0.000183297724781315,
      "loss": 3.7967,
      "step": 89590
    },
    {
      "epoch": 0.08352169798866281,
      "grad_norm": 1.0613830089569092,
      "learning_rate": 0.00018329586045921984,
      "loss": 3.7642,
      "step": 89600
    },
    {
      "epoch": 0.0835310196067419,
      "grad_norm": 1.1535654067993164,
      "learning_rate": 0.00018329399613712465,
      "loss": 3.8366,
      "step": 89610
    },
    {
      "epoch": 0.083540341224821,
      "grad_norm": 1.1036841869354248,
      "learning_rate": 0.00018329213181502942,
      "loss": 3.5893,
      "step": 89620
    },
    {
      "epoch": 0.0835496628429001,
      "grad_norm": 1.0775861740112305,
      "learning_rate": 0.00018329026749293423,
      "loss": 3.73,
      "step": 89630
    },
    {
      "epoch": 0.08355898446097919,
      "grad_norm": 1.1278998851776123,
      "learning_rate": 0.000183288403170839,
      "loss": 3.8291,
      "step": 89640
    },
    {
      "epoch": 0.08356830607905827,
      "grad_norm": 1.0041297674179077,
      "learning_rate": 0.00018328653884874382,
      "loss": 3.6852,
      "step": 89650
    },
    {
      "epoch": 0.08357762769713736,
      "grad_norm": 0.9251834154129028,
      "learning_rate": 0.00018328467452664865,
      "loss": 3.9066,
      "step": 89660
    },
    {
      "epoch": 0.08358694931521646,
      "grad_norm": 0.769033670425415,
      "learning_rate": 0.00018328281020455343,
      "loss": 3.6948,
      "step": 89670
    },
    {
      "epoch": 0.08359627093329555,
      "grad_norm": 0.9831849336624146,
      "learning_rate": 0.00018328094588245824,
      "loss": 3.8821,
      "step": 89680
    },
    {
      "epoch": 0.08360559255137465,
      "grad_norm": 1.1892224550247192,
      "learning_rate": 0.00018327908156036301,
      "loss": 3.5689,
      "step": 89690
    },
    {
      "epoch": 0.08361491416945373,
      "grad_norm": 1.1267107725143433,
      "learning_rate": 0.00018327721723826782,
      "loss": 3.7245,
      "step": 89700
    },
    {
      "epoch": 0.08362423578753282,
      "grad_norm": 1.0802370309829712,
      "learning_rate": 0.00018327535291617263,
      "loss": 3.6414,
      "step": 89710
    },
    {
      "epoch": 0.08363355740561192,
      "grad_norm": 1.0349078178405762,
      "learning_rate": 0.00018327348859407743,
      "loss": 3.9159,
      "step": 89720
    },
    {
      "epoch": 0.08364287902369101,
      "grad_norm": 1.1824735403060913,
      "learning_rate": 0.00018327162427198224,
      "loss": 3.7859,
      "step": 89730
    },
    {
      "epoch": 0.0836522006417701,
      "grad_norm": 1.0639803409576416,
      "learning_rate": 0.00018326975994988702,
      "loss": 3.6769,
      "step": 89740
    },
    {
      "epoch": 0.08366152225984919,
      "grad_norm": 1.1121724843978882,
      "learning_rate": 0.00018326789562779183,
      "loss": 3.7381,
      "step": 89750
    },
    {
      "epoch": 0.08367084387792828,
      "grad_norm": 1.111352562904358,
      "learning_rate": 0.00018326603130569663,
      "loss": 3.8712,
      "step": 89760
    },
    {
      "epoch": 0.08368016549600737,
      "grad_norm": 1.2023488283157349,
      "learning_rate": 0.00018326416698360144,
      "loss": 3.7151,
      "step": 89770
    },
    {
      "epoch": 0.08368948711408647,
      "grad_norm": 1.1461617946624756,
      "learning_rate": 0.00018326230266150624,
      "loss": 3.9219,
      "step": 89780
    },
    {
      "epoch": 0.08369880873216556,
      "grad_norm": 1.2007027864456177,
      "learning_rate": 0.00018326043833941102,
      "loss": 3.8229,
      "step": 89790
    },
    {
      "epoch": 0.08370813035024466,
      "grad_norm": 1.0000767707824707,
      "learning_rate": 0.00018325857401731583,
      "loss": 3.8269,
      "step": 89800
    },
    {
      "epoch": 0.08371745196832374,
      "grad_norm": 1.0178489685058594,
      "learning_rate": 0.00018325670969522064,
      "loss": 3.7504,
      "step": 89810
    },
    {
      "epoch": 0.08372677358640283,
      "grad_norm": 1.1605408191680908,
      "learning_rate": 0.00018325484537312542,
      "loss": 3.7296,
      "step": 89820
    },
    {
      "epoch": 0.08373609520448193,
      "grad_norm": 1.1556155681610107,
      "learning_rate": 0.00018325298105103025,
      "loss": 3.4882,
      "step": 89830
    },
    {
      "epoch": 0.08374541682256102,
      "grad_norm": 1.1427212953567505,
      "learning_rate": 0.00018325111672893503,
      "loss": 3.6726,
      "step": 89840
    },
    {
      "epoch": 0.08375473844064012,
      "grad_norm": 1.1877018213272095,
      "learning_rate": 0.00018324925240683983,
      "loss": 3.6322,
      "step": 89850
    },
    {
      "epoch": 0.0837640600587192,
      "grad_norm": 1.1509456634521484,
      "learning_rate": 0.00018324738808474464,
      "loss": 3.8399,
      "step": 89860
    },
    {
      "epoch": 0.08377338167679829,
      "grad_norm": 1.0799084901809692,
      "learning_rate": 0.00018324552376264942,
      "loss": 3.8791,
      "step": 89870
    },
    {
      "epoch": 0.08378270329487739,
      "grad_norm": 1.0456904172897339,
      "learning_rate": 0.00018324365944055423,
      "loss": 3.65,
      "step": 89880
    },
    {
      "epoch": 0.08379202491295648,
      "grad_norm": 1.1394827365875244,
      "learning_rate": 0.00018324179511845903,
      "loss": 3.7543,
      "step": 89890
    },
    {
      "epoch": 0.08380134653103558,
      "grad_norm": 1.0078247785568237,
      "learning_rate": 0.00018323993079636384,
      "loss": 3.773,
      "step": 89900
    },
    {
      "epoch": 0.08381066814911466,
      "grad_norm": 1.1594544649124146,
      "learning_rate": 0.00018323806647426864,
      "loss": 3.6877,
      "step": 89910
    },
    {
      "epoch": 0.08381998976719375,
      "grad_norm": 0.9649662971496582,
      "learning_rate": 0.00018323620215217342,
      "loss": 3.7691,
      "step": 89920
    },
    {
      "epoch": 0.08382931138527284,
      "grad_norm": 1.1441607475280762,
      "learning_rate": 0.00018323433783007823,
      "loss": 3.745,
      "step": 89930
    },
    {
      "epoch": 0.08383863300335194,
      "grad_norm": 1.1886799335479736,
      "learning_rate": 0.00018323247350798304,
      "loss": 3.8344,
      "step": 89940
    },
    {
      "epoch": 0.08384795462143103,
      "grad_norm": 1.0504533052444458,
      "learning_rate": 0.00018323060918588784,
      "loss": 3.643,
      "step": 89950
    },
    {
      "epoch": 0.08385727623951011,
      "grad_norm": 1.3647034168243408,
      "learning_rate": 0.00018322874486379265,
      "loss": 3.6092,
      "step": 89960
    },
    {
      "epoch": 0.08386659785758921,
      "grad_norm": 0.9005436897277832,
      "learning_rate": 0.00018322688054169743,
      "loss": 3.8258,
      "step": 89970
    },
    {
      "epoch": 0.0838759194756683,
      "grad_norm": 1.122424602508545,
      "learning_rate": 0.00018322501621960223,
      "loss": 3.6372,
      "step": 89980
    },
    {
      "epoch": 0.0838852410937474,
      "grad_norm": 1.108848214149475,
      "learning_rate": 0.00018322315189750704,
      "loss": 3.5915,
      "step": 89990
    },
    {
      "epoch": 0.08389456271182649,
      "grad_norm": 0.935740053653717,
      "learning_rate": 0.00018322128757541185,
      "loss": 3.7464,
      "step": 90000
    },
    {
      "epoch": 0.08390388432990559,
      "grad_norm": 1.2610570192337036,
      "learning_rate": 0.00018321942325331665,
      "loss": 3.7172,
      "step": 90010
    },
    {
      "epoch": 0.08391320594798467,
      "grad_norm": 0.9961820840835571,
      "learning_rate": 0.00018321755893122143,
      "loss": 3.7618,
      "step": 90020
    },
    {
      "epoch": 0.08392252756606376,
      "grad_norm": 1.3069541454315186,
      "learning_rate": 0.00018321569460912624,
      "loss": 3.6268,
      "step": 90030
    },
    {
      "epoch": 0.08393184918414286,
      "grad_norm": 0.777191162109375,
      "learning_rate": 0.00018321383028703102,
      "loss": 3.7958,
      "step": 90040
    },
    {
      "epoch": 0.08394117080222195,
      "grad_norm": 1.1604788303375244,
      "learning_rate": 0.00018321196596493585,
      "loss": 3.79,
      "step": 90050
    },
    {
      "epoch": 0.08395049242030105,
      "grad_norm": 1.0818727016448975,
      "learning_rate": 0.00018321010164284066,
      "loss": 3.9213,
      "step": 90060
    },
    {
      "epoch": 0.08395981403838013,
      "grad_norm": 1.0360149145126343,
      "learning_rate": 0.00018320823732074544,
      "loss": 3.6943,
      "step": 90070
    },
    {
      "epoch": 0.08396913565645922,
      "grad_norm": 1.1248456239700317,
      "learning_rate": 0.00018320637299865024,
      "loss": 3.7104,
      "step": 90080
    },
    {
      "epoch": 0.08397845727453832,
      "grad_norm": 1.07644784450531,
      "learning_rate": 0.00018320450867655502,
      "loss": 3.7212,
      "step": 90090
    },
    {
      "epoch": 0.08398777889261741,
      "grad_norm": 1.1566216945648193,
      "learning_rate": 0.00018320264435445983,
      "loss": 3.8288,
      "step": 90100
    },
    {
      "epoch": 0.0839971005106965,
      "grad_norm": 0.9211863875389099,
      "learning_rate": 0.00018320078003236466,
      "loss": 3.8245,
      "step": 90110
    },
    {
      "epoch": 0.08400642212877558,
      "grad_norm": 1.0317857265472412,
      "learning_rate": 0.00018319891571026944,
      "loss": 3.7912,
      "step": 90120
    },
    {
      "epoch": 0.08401574374685468,
      "grad_norm": 1.0856891870498657,
      "learning_rate": 0.00018319705138817425,
      "loss": 3.8192,
      "step": 90130
    },
    {
      "epoch": 0.08402506536493377,
      "grad_norm": 1.0386570692062378,
      "learning_rate": 0.00018319518706607903,
      "loss": 3.7463,
      "step": 90140
    },
    {
      "epoch": 0.08403438698301287,
      "grad_norm": 1.1885182857513428,
      "learning_rate": 0.00018319332274398383,
      "loss": 3.7317,
      "step": 90150
    },
    {
      "epoch": 0.08404370860109196,
      "grad_norm": 1.2708626985549927,
      "learning_rate": 0.00018319145842188864,
      "loss": 3.6899,
      "step": 90160
    },
    {
      "epoch": 0.08405303021917104,
      "grad_norm": 0.973664402961731,
      "learning_rate": 0.00018318959409979345,
      "loss": 3.7801,
      "step": 90170
    },
    {
      "epoch": 0.08406235183725014,
      "grad_norm": 1.0522152185440063,
      "learning_rate": 0.00018318772977769825,
      "loss": 3.9389,
      "step": 90180
    },
    {
      "epoch": 0.08407167345532923,
      "grad_norm": 1.0568602085113525,
      "learning_rate": 0.00018318586545560303,
      "loss": 3.7703,
      "step": 90190
    },
    {
      "epoch": 0.08408099507340833,
      "grad_norm": 1.045660376548767,
      "learning_rate": 0.00018318400113350784,
      "loss": 3.6671,
      "step": 90200
    },
    {
      "epoch": 0.08409031669148742,
      "grad_norm": 0.9560387134552002,
      "learning_rate": 0.00018318213681141264,
      "loss": 3.665,
      "step": 90210
    },
    {
      "epoch": 0.0840996383095665,
      "grad_norm": 1.0404845476150513,
      "learning_rate": 0.00018318027248931745,
      "loss": 3.8044,
      "step": 90220
    },
    {
      "epoch": 0.0841089599276456,
      "grad_norm": 1.4802747964859009,
      "learning_rate": 0.00018317840816722226,
      "loss": 3.7299,
      "step": 90230
    },
    {
      "epoch": 0.08411828154572469,
      "grad_norm": 0.9927423000335693,
      "learning_rate": 0.00018317654384512704,
      "loss": 3.7628,
      "step": 90240
    },
    {
      "epoch": 0.08412760316380379,
      "grad_norm": 1.1458457708358765,
      "learning_rate": 0.00018317467952303184,
      "loss": 3.6636,
      "step": 90250
    },
    {
      "epoch": 0.08413692478188288,
      "grad_norm": 1.066410779953003,
      "learning_rate": 0.00018317281520093665,
      "loss": 3.8047,
      "step": 90260
    },
    {
      "epoch": 0.08414624639996197,
      "grad_norm": 1.1115853786468506,
      "learning_rate": 0.00018317095087884143,
      "loss": 3.828,
      "step": 90270
    },
    {
      "epoch": 0.08415556801804105,
      "grad_norm": 0.9976119995117188,
      "learning_rate": 0.00018316908655674626,
      "loss": 3.7371,
      "step": 90280
    },
    {
      "epoch": 0.08416488963612015,
      "grad_norm": 1.107704758644104,
      "learning_rate": 0.00018316722223465104,
      "loss": 3.9363,
      "step": 90290
    },
    {
      "epoch": 0.08417421125419924,
      "grad_norm": 1.0218737125396729,
      "learning_rate": 0.00018316535791255585,
      "loss": 3.7112,
      "step": 90300
    },
    {
      "epoch": 0.08418353287227834,
      "grad_norm": 0.945480465888977,
      "learning_rate": 0.00018316349359046065,
      "loss": 3.8707,
      "step": 90310
    },
    {
      "epoch": 0.08419285449035743,
      "grad_norm": 0.8942965865135193,
      "learning_rate": 0.00018316162926836543,
      "loss": 3.9042,
      "step": 90320
    },
    {
      "epoch": 0.08420217610843651,
      "grad_norm": 1.1018961668014526,
      "learning_rate": 0.00018315976494627024,
      "loss": 3.7661,
      "step": 90330
    },
    {
      "epoch": 0.08421149772651561,
      "grad_norm": 1.0971002578735352,
      "learning_rate": 0.00018315790062417505,
      "loss": 3.8322,
      "step": 90340
    },
    {
      "epoch": 0.0842208193445947,
      "grad_norm": 0.9910849928855896,
      "learning_rate": 0.00018315603630207985,
      "loss": 3.7469,
      "step": 90350
    },
    {
      "epoch": 0.0842301409626738,
      "grad_norm": 1.0718421936035156,
      "learning_rate": 0.00018315417197998466,
      "loss": 3.8812,
      "step": 90360
    },
    {
      "epoch": 0.08423946258075289,
      "grad_norm": 1.0482250452041626,
      "learning_rate": 0.00018315230765788944,
      "loss": 3.9865,
      "step": 90370
    },
    {
      "epoch": 0.08424878419883197,
      "grad_norm": 0.98357754945755,
      "learning_rate": 0.00018315044333579424,
      "loss": 3.9917,
      "step": 90380
    },
    {
      "epoch": 0.08425810581691107,
      "grad_norm": 1.1250815391540527,
      "learning_rate": 0.00018314857901369905,
      "loss": 3.6354,
      "step": 90390
    },
    {
      "epoch": 0.08426742743499016,
      "grad_norm": 1.0230919122695923,
      "learning_rate": 0.00018314671469160386,
      "loss": 3.7252,
      "step": 90400
    },
    {
      "epoch": 0.08427674905306926,
      "grad_norm": 1.3452774286270142,
      "learning_rate": 0.00018314485036950866,
      "loss": 3.8587,
      "step": 90410
    },
    {
      "epoch": 0.08428607067114835,
      "grad_norm": 1.221645474433899,
      "learning_rate": 0.00018314298604741344,
      "loss": 3.9327,
      "step": 90420
    },
    {
      "epoch": 0.08429539228922743,
      "grad_norm": 0.974917471408844,
      "learning_rate": 0.00018314112172531825,
      "loss": 3.8915,
      "step": 90430
    },
    {
      "epoch": 0.08430471390730652,
      "grad_norm": 1.235995888710022,
      "learning_rate": 0.00018313925740322305,
      "loss": 3.8454,
      "step": 90440
    },
    {
      "epoch": 0.08431403552538562,
      "grad_norm": 0.8691379427909851,
      "learning_rate": 0.00018313739308112786,
      "loss": 3.8335,
      "step": 90450
    },
    {
      "epoch": 0.08432335714346471,
      "grad_norm": 1.140667200088501,
      "learning_rate": 0.00018313552875903267,
      "loss": 3.7275,
      "step": 90460
    },
    {
      "epoch": 0.08433267876154381,
      "grad_norm": 1.0790637731552124,
      "learning_rate": 0.00018313366443693745,
      "loss": 3.7725,
      "step": 90470
    },
    {
      "epoch": 0.0843420003796229,
      "grad_norm": 1.0021501779556274,
      "learning_rate": 0.00018313180011484225,
      "loss": 3.8013,
      "step": 90480
    },
    {
      "epoch": 0.08435132199770198,
      "grad_norm": 1.0823966264724731,
      "learning_rate": 0.00018312993579274703,
      "loss": 3.6292,
      "step": 90490
    },
    {
      "epoch": 0.08436064361578108,
      "grad_norm": 1.112099528312683,
      "learning_rate": 0.00018312807147065186,
      "loss": 3.9491,
      "step": 90500
    },
    {
      "epoch": 0.08436996523386017,
      "grad_norm": 1.14512300491333,
      "learning_rate": 0.00018312620714855667,
      "loss": 3.7152,
      "step": 90510
    },
    {
      "epoch": 0.08437928685193927,
      "grad_norm": 1.209336757659912,
      "learning_rate": 0.00018312434282646145,
      "loss": 3.9136,
      "step": 90520
    },
    {
      "epoch": 0.08438860847001836,
      "grad_norm": 1.3037620782852173,
      "learning_rate": 0.00018312247850436626,
      "loss": 3.8955,
      "step": 90530
    },
    {
      "epoch": 0.08439793008809744,
      "grad_norm": 1.1198545694351196,
      "learning_rate": 0.00018312061418227104,
      "loss": 3.7608,
      "step": 90540
    },
    {
      "epoch": 0.08440725170617654,
      "grad_norm": 1.0128357410430908,
      "learning_rate": 0.00018311874986017584,
      "loss": 3.6585,
      "step": 90550
    },
    {
      "epoch": 0.08441657332425563,
      "grad_norm": 1.0617119073867798,
      "learning_rate": 0.00018311688553808068,
      "loss": 3.7551,
      "step": 90560
    },
    {
      "epoch": 0.08442589494233473,
      "grad_norm": 0.8415412902832031,
      "learning_rate": 0.00018311502121598545,
      "loss": 3.7601,
      "step": 90570
    },
    {
      "epoch": 0.08443521656041382,
      "grad_norm": 1.2341989278793335,
      "learning_rate": 0.00018311315689389026,
      "loss": 3.8985,
      "step": 90580
    },
    {
      "epoch": 0.0844445381784929,
      "grad_norm": 1.2975462675094604,
      "learning_rate": 0.00018311129257179504,
      "loss": 3.6439,
      "step": 90590
    },
    {
      "epoch": 0.084453859796572,
      "grad_norm": 1.1044119596481323,
      "learning_rate": 0.00018310942824969985,
      "loss": 3.6676,
      "step": 90600
    },
    {
      "epoch": 0.08446318141465109,
      "grad_norm": 0.9124386310577393,
      "learning_rate": 0.00018310756392760465,
      "loss": 3.6943,
      "step": 90610
    },
    {
      "epoch": 0.08447250303273018,
      "grad_norm": 1.2200120687484741,
      "learning_rate": 0.00018310569960550946,
      "loss": 3.9588,
      "step": 90620
    },
    {
      "epoch": 0.08448182465080928,
      "grad_norm": 0.9918093681335449,
      "learning_rate": 0.00018310383528341427,
      "loss": 3.8129,
      "step": 90630
    },
    {
      "epoch": 0.08449114626888836,
      "grad_norm": 1.0886304378509521,
      "learning_rate": 0.00018310197096131904,
      "loss": 3.7887,
      "step": 90640
    },
    {
      "epoch": 0.08450046788696745,
      "grad_norm": 1.073482632637024,
      "learning_rate": 0.00018310010663922385,
      "loss": 3.8949,
      "step": 90650
    },
    {
      "epoch": 0.08450978950504655,
      "grad_norm": 1.05215585231781,
      "learning_rate": 0.00018309824231712866,
      "loss": 3.7704,
      "step": 90660
    },
    {
      "epoch": 0.08451911112312564,
      "grad_norm": 1.0809096097946167,
      "learning_rate": 0.00018309637799503346,
      "loss": 3.7362,
      "step": 90670
    },
    {
      "epoch": 0.08452843274120474,
      "grad_norm": 1.122791051864624,
      "learning_rate": 0.00018309451367293827,
      "loss": 3.68,
      "step": 90680
    },
    {
      "epoch": 0.08453775435928383,
      "grad_norm": 1.2692192792892456,
      "learning_rate": 0.00018309264935084305,
      "loss": 3.5192,
      "step": 90690
    },
    {
      "epoch": 0.08454707597736291,
      "grad_norm": 1.0542346239089966,
      "learning_rate": 0.00018309078502874786,
      "loss": 3.5752,
      "step": 90700
    },
    {
      "epoch": 0.084556397595442,
      "grad_norm": 1.1520256996154785,
      "learning_rate": 0.00018308892070665266,
      "loss": 3.8361,
      "step": 90710
    },
    {
      "epoch": 0.0845657192135211,
      "grad_norm": 1.050302505493164,
      "learning_rate": 0.00018308705638455744,
      "loss": 3.7375,
      "step": 90720
    },
    {
      "epoch": 0.0845750408316002,
      "grad_norm": 0.9459418654441833,
      "learning_rate": 0.00018308519206246227,
      "loss": 3.6937,
      "step": 90730
    },
    {
      "epoch": 0.08458436244967929,
      "grad_norm": 0.8352731466293335,
      "learning_rate": 0.00018308332774036705,
      "loss": 3.7625,
      "step": 90740
    },
    {
      "epoch": 0.08459368406775837,
      "grad_norm": 1.2439701557159424,
      "learning_rate": 0.00018308146341827186,
      "loss": 3.7893,
      "step": 90750
    },
    {
      "epoch": 0.08460300568583747,
      "grad_norm": 1.2069499492645264,
      "learning_rate": 0.00018307959909617664,
      "loss": 3.6628,
      "step": 90760
    },
    {
      "epoch": 0.08461232730391656,
      "grad_norm": 1.1180886030197144,
      "learning_rate": 0.00018307773477408145,
      "loss": 3.7591,
      "step": 90770
    },
    {
      "epoch": 0.08462164892199565,
      "grad_norm": 1.058152675628662,
      "learning_rate": 0.00018307587045198625,
      "loss": 3.7974,
      "step": 90780
    },
    {
      "epoch": 0.08463097054007475,
      "grad_norm": 1.2063063383102417,
      "learning_rate": 0.00018307400612989106,
      "loss": 3.6574,
      "step": 90790
    },
    {
      "epoch": 0.08464029215815383,
      "grad_norm": 1.0201618671417236,
      "learning_rate": 0.00018307214180779586,
      "loss": 3.7676,
      "step": 90800
    },
    {
      "epoch": 0.08464961377623292,
      "grad_norm": 1.1455262899398804,
      "learning_rate": 0.00018307027748570064,
      "loss": 3.9056,
      "step": 90810
    },
    {
      "epoch": 0.08465893539431202,
      "grad_norm": 1.2237809896469116,
      "learning_rate": 0.00018306841316360545,
      "loss": 3.6868,
      "step": 90820
    },
    {
      "epoch": 0.08466825701239111,
      "grad_norm": 1.097634196281433,
      "learning_rate": 0.00018306654884151026,
      "loss": 3.6448,
      "step": 90830
    },
    {
      "epoch": 0.08467757863047021,
      "grad_norm": 1.1385921239852905,
      "learning_rate": 0.00018306468451941506,
      "loss": 3.754,
      "step": 90840
    },
    {
      "epoch": 0.08468690024854929,
      "grad_norm": 1.1270112991333008,
      "learning_rate": 0.00018306282019731987,
      "loss": 3.7724,
      "step": 90850
    },
    {
      "epoch": 0.08469622186662838,
      "grad_norm": 1.2633514404296875,
      "learning_rate": 0.00018306095587522465,
      "loss": 3.7134,
      "step": 90860
    },
    {
      "epoch": 0.08470554348470748,
      "grad_norm": 1.030579924583435,
      "learning_rate": 0.00018305909155312945,
      "loss": 3.7369,
      "step": 90870
    },
    {
      "epoch": 0.08471486510278657,
      "grad_norm": 1.166219711303711,
      "learning_rate": 0.00018305722723103426,
      "loss": 3.9238,
      "step": 90880
    },
    {
      "epoch": 0.08472418672086567,
      "grad_norm": 1.2537952661514282,
      "learning_rate": 0.00018305536290893907,
      "loss": 3.6765,
      "step": 90890
    },
    {
      "epoch": 0.08473350833894475,
      "grad_norm": 1.1088823080062866,
      "learning_rate": 0.00018305349858684387,
      "loss": 3.8809,
      "step": 90900
    },
    {
      "epoch": 0.08474282995702384,
      "grad_norm": 1.0465717315673828,
      "learning_rate": 0.00018305163426474865,
      "loss": 3.8979,
      "step": 90910
    },
    {
      "epoch": 0.08475215157510294,
      "grad_norm": 1.155555009841919,
      "learning_rate": 0.00018304976994265346,
      "loss": 3.6469,
      "step": 90920
    },
    {
      "epoch": 0.08476147319318203,
      "grad_norm": 0.9948186874389648,
      "learning_rate": 0.00018304790562055827,
      "loss": 3.7609,
      "step": 90930
    },
    {
      "epoch": 0.08477079481126112,
      "grad_norm": 0.918726921081543,
      "learning_rate": 0.00018304604129846304,
      "loss": 3.6763,
      "step": 90940
    },
    {
      "epoch": 0.08478011642934022,
      "grad_norm": 1.102290391921997,
      "learning_rate": 0.00018304417697636788,
      "loss": 3.8515,
      "step": 90950
    },
    {
      "epoch": 0.0847894380474193,
      "grad_norm": 1.00737726688385,
      "learning_rate": 0.00018304231265427266,
      "loss": 3.7304,
      "step": 90960
    },
    {
      "epoch": 0.0847987596654984,
      "grad_norm": 1.0778688192367554,
      "learning_rate": 0.00018304044833217746,
      "loss": 3.9536,
      "step": 90970
    },
    {
      "epoch": 0.08480808128357749,
      "grad_norm": 0.9101754426956177,
      "learning_rate": 0.00018303858401008227,
      "loss": 3.9582,
      "step": 90980
    },
    {
      "epoch": 0.08481740290165658,
      "grad_norm": 1.0170153379440308,
      "learning_rate": 0.00018303671968798705,
      "loss": 3.7503,
      "step": 90990
    },
    {
      "epoch": 0.08482672451973568,
      "grad_norm": 1.1915005445480347,
      "learning_rate": 0.00018303485536589186,
      "loss": 3.6722,
      "step": 91000
    },
    {
      "epoch": 0.08483604613781476,
      "grad_norm": 1.2564131021499634,
      "learning_rate": 0.00018303299104379666,
      "loss": 3.8901,
      "step": 91010
    },
    {
      "epoch": 0.08484536775589385,
      "grad_norm": 1.1804335117340088,
      "learning_rate": 0.00018303112672170147,
      "loss": 3.7327,
      "step": 91020
    },
    {
      "epoch": 0.08485468937397295,
      "grad_norm": 1.1029466390609741,
      "learning_rate": 0.00018302926239960627,
      "loss": 3.7015,
      "step": 91030
    },
    {
      "epoch": 0.08486401099205204,
      "grad_norm": 0.9740744233131409,
      "learning_rate": 0.00018302739807751105,
      "loss": 3.7275,
      "step": 91040
    },
    {
      "epoch": 0.08487333261013114,
      "grad_norm": 1.133829951286316,
      "learning_rate": 0.00018302553375541586,
      "loss": 3.8428,
      "step": 91050
    },
    {
      "epoch": 0.08488265422821022,
      "grad_norm": 1.0440329313278198,
      "learning_rate": 0.00018302366943332067,
      "loss": 3.8152,
      "step": 91060
    },
    {
      "epoch": 0.08489197584628931,
      "grad_norm": 1.0003288984298706,
      "learning_rate": 0.00018302180511122547,
      "loss": 3.8175,
      "step": 91070
    },
    {
      "epoch": 0.0849012974643684,
      "grad_norm": 1.2239731550216675,
      "learning_rate": 0.00018301994078913028,
      "loss": 3.7494,
      "step": 91080
    },
    {
      "epoch": 0.0849106190824475,
      "grad_norm": 1.1497533321380615,
      "learning_rate": 0.00018301807646703506,
      "loss": 3.7585,
      "step": 91090
    },
    {
      "epoch": 0.0849199407005266,
      "grad_norm": 1.1017271280288696,
      "learning_rate": 0.00018301621214493986,
      "loss": 3.8577,
      "step": 91100
    },
    {
      "epoch": 0.08492926231860567,
      "grad_norm": 1.1505180597305298,
      "learning_rate": 0.00018301434782284464,
      "loss": 3.7433,
      "step": 91110
    },
    {
      "epoch": 0.08493858393668477,
      "grad_norm": 1.0918858051300049,
      "learning_rate": 0.00018301248350074948,
      "loss": 3.9526,
      "step": 91120
    },
    {
      "epoch": 0.08494790555476386,
      "grad_norm": 1.1139600276947021,
      "learning_rate": 0.00018301061917865428,
      "loss": 3.7455,
      "step": 91130
    },
    {
      "epoch": 0.08495722717284296,
      "grad_norm": 1.1625149250030518,
      "learning_rate": 0.00018300875485655906,
      "loss": 3.6939,
      "step": 91140
    },
    {
      "epoch": 0.08496654879092205,
      "grad_norm": 1.09325110912323,
      "learning_rate": 0.00018300689053446387,
      "loss": 3.757,
      "step": 91150
    },
    {
      "epoch": 0.08497587040900115,
      "grad_norm": 1.0067170858383179,
      "learning_rate": 0.00018300502621236865,
      "loss": 3.6611,
      "step": 91160
    },
    {
      "epoch": 0.08498519202708023,
      "grad_norm": 1.0885149240493774,
      "learning_rate": 0.00018300316189027345,
      "loss": 3.8365,
      "step": 91170
    },
    {
      "epoch": 0.08499451364515932,
      "grad_norm": 1.1935430765151978,
      "learning_rate": 0.0001830012975681783,
      "loss": 3.8213,
      "step": 91180
    },
    {
      "epoch": 0.08500383526323842,
      "grad_norm": 0.84852135181427,
      "learning_rate": 0.00018299943324608307,
      "loss": 3.9547,
      "step": 91190
    },
    {
      "epoch": 0.08501315688131751,
      "grad_norm": 1.1958072185516357,
      "learning_rate": 0.00018299756892398787,
      "loss": 3.7986,
      "step": 91200
    },
    {
      "epoch": 0.0850224784993966,
      "grad_norm": 0.9203841090202332,
      "learning_rate": 0.00018299570460189265,
      "loss": 3.7679,
      "step": 91210
    },
    {
      "epoch": 0.08503180011747569,
      "grad_norm": 1.2036421298980713,
      "learning_rate": 0.00018299384027979746,
      "loss": 3.8644,
      "step": 91220
    },
    {
      "epoch": 0.08504112173555478,
      "grad_norm": 1.1162233352661133,
      "learning_rate": 0.00018299197595770226,
      "loss": 3.8104,
      "step": 91230
    },
    {
      "epoch": 0.08505044335363388,
      "grad_norm": 1.1090097427368164,
      "learning_rate": 0.00018299011163560707,
      "loss": 3.7358,
      "step": 91240
    },
    {
      "epoch": 0.08505976497171297,
      "grad_norm": 0.8866793513298035,
      "learning_rate": 0.00018298824731351188,
      "loss": 3.8207,
      "step": 91250
    },
    {
      "epoch": 0.08506908658979206,
      "grad_norm": 1.15516996383667,
      "learning_rate": 0.00018298638299141666,
      "loss": 3.8723,
      "step": 91260
    },
    {
      "epoch": 0.08507840820787115,
      "grad_norm": 1.0819809436798096,
      "learning_rate": 0.00018298451866932146,
      "loss": 3.718,
      "step": 91270
    },
    {
      "epoch": 0.08508772982595024,
      "grad_norm": 0.9199497699737549,
      "learning_rate": 0.00018298265434722627,
      "loss": 3.8054,
      "step": 91280
    },
    {
      "epoch": 0.08509705144402933,
      "grad_norm": 0.987616240978241,
      "learning_rate": 0.00018298079002513108,
      "loss": 3.7439,
      "step": 91290
    },
    {
      "epoch": 0.08510637306210843,
      "grad_norm": 1.1026971340179443,
      "learning_rate": 0.00018297892570303588,
      "loss": 3.9381,
      "step": 91300
    },
    {
      "epoch": 0.08511569468018752,
      "grad_norm": 0.9397737383842468,
      "learning_rate": 0.00018297706138094066,
      "loss": 3.7615,
      "step": 91310
    },
    {
      "epoch": 0.0851250162982666,
      "grad_norm": 1.0457862615585327,
      "learning_rate": 0.00018297519705884547,
      "loss": 3.9052,
      "step": 91320
    },
    {
      "epoch": 0.0851343379163457,
      "grad_norm": 1.145548701286316,
      "learning_rate": 0.00018297333273675027,
      "loss": 3.693,
      "step": 91330
    },
    {
      "epoch": 0.08514365953442479,
      "grad_norm": 1.00532066822052,
      "learning_rate": 0.00018297146841465508,
      "loss": 3.8779,
      "step": 91340
    },
    {
      "epoch": 0.08515298115250389,
      "grad_norm": 1.0564988851547241,
      "learning_rate": 0.00018296960409255989,
      "loss": 3.6925,
      "step": 91350
    },
    {
      "epoch": 0.08516230277058298,
      "grad_norm": 0.9107308387756348,
      "learning_rate": 0.00018296773977046467,
      "loss": 3.8035,
      "step": 91360
    },
    {
      "epoch": 0.08517162438866208,
      "grad_norm": 1.0773056745529175,
      "learning_rate": 0.00018296587544836947,
      "loss": 3.8784,
      "step": 91370
    },
    {
      "epoch": 0.08518094600674116,
      "grad_norm": 1.1865957975387573,
      "learning_rate": 0.00018296401112627428,
      "loss": 3.8163,
      "step": 91380
    },
    {
      "epoch": 0.08519026762482025,
      "grad_norm": 0.9493292570114136,
      "learning_rate": 0.00018296214680417906,
      "loss": 3.5424,
      "step": 91390
    },
    {
      "epoch": 0.08519958924289935,
      "grad_norm": 1.0764402151107788,
      "learning_rate": 0.0001829602824820839,
      "loss": 3.7797,
      "step": 91400
    },
    {
      "epoch": 0.08520891086097844,
      "grad_norm": 1.2743619680404663,
      "learning_rate": 0.00018295841815998867,
      "loss": 3.6002,
      "step": 91410
    },
    {
      "epoch": 0.08521823247905753,
      "grad_norm": 0.7772631049156189,
      "learning_rate": 0.00018295655383789348,
      "loss": 3.7515,
      "step": 91420
    },
    {
      "epoch": 0.08522755409713662,
      "grad_norm": 1.104709267616272,
      "learning_rate": 0.00018295468951579828,
      "loss": 3.9634,
      "step": 91430
    },
    {
      "epoch": 0.08523687571521571,
      "grad_norm": 1.1289148330688477,
      "learning_rate": 0.00018295282519370306,
      "loss": 3.838,
      "step": 91440
    },
    {
      "epoch": 0.0852461973332948,
      "grad_norm": 0.9707080125808716,
      "learning_rate": 0.00018295096087160787,
      "loss": 3.7911,
      "step": 91450
    },
    {
      "epoch": 0.0852555189513739,
      "grad_norm": 1.0870764255523682,
      "learning_rate": 0.00018294909654951267,
      "loss": 3.687,
      "step": 91460
    },
    {
      "epoch": 0.085264840569453,
      "grad_norm": 1.1658753156661987,
      "learning_rate": 0.00018294723222741748,
      "loss": 3.7752,
      "step": 91470
    },
    {
      "epoch": 0.08527416218753207,
      "grad_norm": 1.1620100736618042,
      "learning_rate": 0.0001829453679053223,
      "loss": 3.84,
      "step": 91480
    },
    {
      "epoch": 0.08528348380561117,
      "grad_norm": 1.1158514022827148,
      "learning_rate": 0.00018294350358322707,
      "loss": 3.7716,
      "step": 91490
    },
    {
      "epoch": 0.08529280542369026,
      "grad_norm": 1.1250081062316895,
      "learning_rate": 0.00018294163926113187,
      "loss": 3.823,
      "step": 91500
    },
    {
      "epoch": 0.08530212704176936,
      "grad_norm": 0.8562841415405273,
      "learning_rate": 0.00018293977493903668,
      "loss": 3.7299,
      "step": 91510
    },
    {
      "epoch": 0.08531144865984845,
      "grad_norm": 0.9067909121513367,
      "learning_rate": 0.00018293791061694148,
      "loss": 3.7409,
      "step": 91520
    },
    {
      "epoch": 0.08532077027792753,
      "grad_norm": 1.1084721088409424,
      "learning_rate": 0.0001829360462948463,
      "loss": 3.6228,
      "step": 91530
    },
    {
      "epoch": 0.08533009189600663,
      "grad_norm": 1.0502995252609253,
      "learning_rate": 0.00018293418197275107,
      "loss": 3.7267,
      "step": 91540
    },
    {
      "epoch": 0.08533941351408572,
      "grad_norm": 1.1430256366729736,
      "learning_rate": 0.00018293231765065588,
      "loss": 3.5794,
      "step": 91550
    },
    {
      "epoch": 0.08534873513216482,
      "grad_norm": 1.0504021644592285,
      "learning_rate": 0.00018293045332856066,
      "loss": 3.6779,
      "step": 91560
    },
    {
      "epoch": 0.08535805675024391,
      "grad_norm": 1.138213038444519,
      "learning_rate": 0.0001829285890064655,
      "loss": 3.4497,
      "step": 91570
    },
    {
      "epoch": 0.08536737836832299,
      "grad_norm": 0.957514226436615,
      "learning_rate": 0.0001829267246843703,
      "loss": 3.877,
      "step": 91580
    },
    {
      "epoch": 0.08537669998640209,
      "grad_norm": 0.9879634380340576,
      "learning_rate": 0.00018292486036227507,
      "loss": 3.8679,
      "step": 91590
    },
    {
      "epoch": 0.08538602160448118,
      "grad_norm": 0.8961361646652222,
      "learning_rate": 0.00018292299604017988,
      "loss": 3.685,
      "step": 91600
    },
    {
      "epoch": 0.08539534322256027,
      "grad_norm": 1.1841081380844116,
      "learning_rate": 0.00018292113171808466,
      "loss": 3.7503,
      "step": 91610
    },
    {
      "epoch": 0.08540466484063937,
      "grad_norm": 0.9616027474403381,
      "learning_rate": 0.00018291926739598947,
      "loss": 3.8878,
      "step": 91620
    },
    {
      "epoch": 0.08541398645871846,
      "grad_norm": 1.186718463897705,
      "learning_rate": 0.0001829174030738943,
      "loss": 3.7853,
      "step": 91630
    },
    {
      "epoch": 0.08542330807679754,
      "grad_norm": 1.010304570198059,
      "learning_rate": 0.00018291553875179908,
      "loss": 3.7579,
      "step": 91640
    },
    {
      "epoch": 0.08543262969487664,
      "grad_norm": 1.1286053657531738,
      "learning_rate": 0.00018291367442970389,
      "loss": 3.9119,
      "step": 91650
    },
    {
      "epoch": 0.08544195131295573,
      "grad_norm": 1.051630973815918,
      "learning_rate": 0.00018291181010760866,
      "loss": 3.6778,
      "step": 91660
    },
    {
      "epoch": 0.08545127293103483,
      "grad_norm": 1.0436055660247803,
      "learning_rate": 0.00018290994578551347,
      "loss": 3.6624,
      "step": 91670
    },
    {
      "epoch": 0.08546059454911392,
      "grad_norm": 1.15298330783844,
      "learning_rate": 0.00018290808146341828,
      "loss": 3.6874,
      "step": 91680
    },
    {
      "epoch": 0.085469916167193,
      "grad_norm": 1.330653429031372,
      "learning_rate": 0.00018290621714132308,
      "loss": 3.8147,
      "step": 91690
    },
    {
      "epoch": 0.0854792377852721,
      "grad_norm": 1.1219136714935303,
      "learning_rate": 0.0001829043528192279,
      "loss": 3.8344,
      "step": 91700
    },
    {
      "epoch": 0.08548855940335119,
      "grad_norm": 1.1229875087738037,
      "learning_rate": 0.00018290248849713267,
      "loss": 3.6432,
      "step": 91710
    },
    {
      "epoch": 0.08549788102143029,
      "grad_norm": 1.0853486061096191,
      "learning_rate": 0.00018290062417503748,
      "loss": 3.7339,
      "step": 91720
    },
    {
      "epoch": 0.08550720263950938,
      "grad_norm": 0.7813064455986023,
      "learning_rate": 0.00018289875985294228,
      "loss": 3.8865,
      "step": 91730
    },
    {
      "epoch": 0.08551652425758846,
      "grad_norm": 0.9296284317970276,
      "learning_rate": 0.0001828968955308471,
      "loss": 3.6774,
      "step": 91740
    },
    {
      "epoch": 0.08552584587566756,
      "grad_norm": 1.0032343864440918,
      "learning_rate": 0.0001828950312087519,
      "loss": 3.6884,
      "step": 91750
    },
    {
      "epoch": 0.08553516749374665,
      "grad_norm": 1.1853216886520386,
      "learning_rate": 0.00018289316688665667,
      "loss": 3.7448,
      "step": 91760
    },
    {
      "epoch": 0.08554448911182574,
      "grad_norm": 1.2081408500671387,
      "learning_rate": 0.00018289130256456148,
      "loss": 3.7445,
      "step": 91770
    },
    {
      "epoch": 0.08555381072990484,
      "grad_norm": 1.094301462173462,
      "learning_rate": 0.00018288943824246629,
      "loss": 3.8244,
      "step": 91780
    },
    {
      "epoch": 0.08556313234798392,
      "grad_norm": 1.0295906066894531,
      "learning_rate": 0.0001828875739203711,
      "loss": 3.7275,
      "step": 91790
    },
    {
      "epoch": 0.08557245396606301,
      "grad_norm": 0.9898737668991089,
      "learning_rate": 0.0001828857095982759,
      "loss": 3.8366,
      "step": 91800
    },
    {
      "epoch": 0.08558177558414211,
      "grad_norm": 1.2109133005142212,
      "learning_rate": 0.00018288384527618068,
      "loss": 3.8556,
      "step": 91810
    },
    {
      "epoch": 0.0855910972022212,
      "grad_norm": 0.9853612184524536,
      "learning_rate": 0.00018288198095408548,
      "loss": 3.645,
      "step": 91820
    },
    {
      "epoch": 0.0856004188203003,
      "grad_norm": 0.9801529049873352,
      "learning_rate": 0.0001828801166319903,
      "loss": 3.8935,
      "step": 91830
    },
    {
      "epoch": 0.08560974043837939,
      "grad_norm": 1.1206427812576294,
      "learning_rate": 0.00018287825230989507,
      "loss": 3.7188,
      "step": 91840
    },
    {
      "epoch": 0.08561906205645847,
      "grad_norm": 0.99134761095047,
      "learning_rate": 0.0001828763879877999,
      "loss": 3.595,
      "step": 91850
    },
    {
      "epoch": 0.08562838367453757,
      "grad_norm": 1.0674511194229126,
      "learning_rate": 0.00018287452366570468,
      "loss": 3.7281,
      "step": 91860
    },
    {
      "epoch": 0.08563770529261666,
      "grad_norm": 1.0122731924057007,
      "learning_rate": 0.0001828726593436095,
      "loss": 3.8174,
      "step": 91870
    },
    {
      "epoch": 0.08564702691069576,
      "grad_norm": 1.1549248695373535,
      "learning_rate": 0.0001828707950215143,
      "loss": 3.7613,
      "step": 91880
    },
    {
      "epoch": 0.08565634852877485,
      "grad_norm": 1.3927743434906006,
      "learning_rate": 0.00018286893069941907,
      "loss": 3.819,
      "step": 91890
    },
    {
      "epoch": 0.08566567014685393,
      "grad_norm": 0.9482816457748413,
      "learning_rate": 0.00018286706637732388,
      "loss": 3.7454,
      "step": 91900
    },
    {
      "epoch": 0.08567499176493303,
      "grad_norm": 0.9196394681930542,
      "learning_rate": 0.0001828652020552287,
      "loss": 3.7495,
      "step": 91910
    },
    {
      "epoch": 0.08568431338301212,
      "grad_norm": 1.0874590873718262,
      "learning_rate": 0.0001828633377331335,
      "loss": 3.7849,
      "step": 91920
    },
    {
      "epoch": 0.08569363500109121,
      "grad_norm": 1.0187325477600098,
      "learning_rate": 0.0001828614734110383,
      "loss": 3.9197,
      "step": 91930
    },
    {
      "epoch": 0.08570295661917031,
      "grad_norm": 0.9105011820793152,
      "learning_rate": 0.00018285960908894308,
      "loss": 3.8084,
      "step": 91940
    },
    {
      "epoch": 0.08571227823724939,
      "grad_norm": 1.147057056427002,
      "learning_rate": 0.00018285774476684789,
      "loss": 3.682,
      "step": 91950
    },
    {
      "epoch": 0.08572159985532848,
      "grad_norm": 1.067152738571167,
      "learning_rate": 0.0001828558804447527,
      "loss": 3.9321,
      "step": 91960
    },
    {
      "epoch": 0.08573092147340758,
      "grad_norm": 1.1349315643310547,
      "learning_rate": 0.0001828540161226575,
      "loss": 3.6602,
      "step": 91970
    },
    {
      "epoch": 0.08574024309148667,
      "grad_norm": 1.08203125,
      "learning_rate": 0.0001828521518005623,
      "loss": 3.804,
      "step": 91980
    },
    {
      "epoch": 0.08574956470956577,
      "grad_norm": 1.2118202447891235,
      "learning_rate": 0.00018285028747846708,
      "loss": 3.9199,
      "step": 91990
    },
    {
      "epoch": 0.08575888632764485,
      "grad_norm": 1.232101559638977,
      "learning_rate": 0.0001828484231563719,
      "loss": 3.6725,
      "step": 92000
    },
    {
      "epoch": 0.08576820794572394,
      "grad_norm": 1.1522777080535889,
      "learning_rate": 0.00018284655883427667,
      "loss": 3.8045,
      "step": 92010
    },
    {
      "epoch": 0.08577752956380304,
      "grad_norm": 1.1498953104019165,
      "learning_rate": 0.0001828446945121815,
      "loss": 3.7858,
      "step": 92020
    },
    {
      "epoch": 0.08578685118188213,
      "grad_norm": 1.1172664165496826,
      "learning_rate": 0.0001828428301900863,
      "loss": 3.767,
      "step": 92030
    },
    {
      "epoch": 0.08579617279996123,
      "grad_norm": 1.2952816486358643,
      "learning_rate": 0.0001828409658679911,
      "loss": 3.7867,
      "step": 92040
    },
    {
      "epoch": 0.08580549441804032,
      "grad_norm": 1.1205692291259766,
      "learning_rate": 0.0001828391015458959,
      "loss": 3.8787,
      "step": 92050
    },
    {
      "epoch": 0.0858148160361194,
      "grad_norm": 1.1771574020385742,
      "learning_rate": 0.00018283723722380067,
      "loss": 3.7482,
      "step": 92060
    },
    {
      "epoch": 0.0858241376541985,
      "grad_norm": 0.9389594197273254,
      "learning_rate": 0.00018283537290170548,
      "loss": 3.64,
      "step": 92070
    },
    {
      "epoch": 0.08583345927227759,
      "grad_norm": 1.1064437627792358,
      "learning_rate": 0.00018283350857961029,
      "loss": 3.6971,
      "step": 92080
    },
    {
      "epoch": 0.08584278089035668,
      "grad_norm": 1.0675454139709473,
      "learning_rate": 0.0001828316442575151,
      "loss": 3.8353,
      "step": 92090
    },
    {
      "epoch": 0.08585210250843578,
      "grad_norm": 1.049069881439209,
      "learning_rate": 0.0001828297799354199,
      "loss": 3.5809,
      "step": 92100
    },
    {
      "epoch": 0.08586142412651486,
      "grad_norm": 1.0894310474395752,
      "learning_rate": 0.00018282791561332468,
      "loss": 4.0352,
      "step": 92110
    },
    {
      "epoch": 0.08587074574459395,
      "grad_norm": 1.1056156158447266,
      "learning_rate": 0.00018282605129122948,
      "loss": 3.7105,
      "step": 92120
    },
    {
      "epoch": 0.08588006736267305,
      "grad_norm": 1.0494545698165894,
      "learning_rate": 0.0001828241869691343,
      "loss": 3.7656,
      "step": 92130
    },
    {
      "epoch": 0.08588938898075214,
      "grad_norm": 0.9302995800971985,
      "learning_rate": 0.0001828223226470391,
      "loss": 3.7667,
      "step": 92140
    },
    {
      "epoch": 0.08589871059883124,
      "grad_norm": 0.9697157144546509,
      "learning_rate": 0.0001828204583249439,
      "loss": 3.5373,
      "step": 92150
    },
    {
      "epoch": 0.08590803221691032,
      "grad_norm": 1.1644346714019775,
      "learning_rate": 0.00018281859400284868,
      "loss": 3.5915,
      "step": 92160
    },
    {
      "epoch": 0.08591735383498941,
      "grad_norm": 1.1313014030456543,
      "learning_rate": 0.0001828167296807535,
      "loss": 3.7486,
      "step": 92170
    },
    {
      "epoch": 0.08592667545306851,
      "grad_norm": 0.9707556366920471,
      "learning_rate": 0.0001828148653586583,
      "loss": 3.7341,
      "step": 92180
    },
    {
      "epoch": 0.0859359970711476,
      "grad_norm": 1.0344361066818237,
      "learning_rate": 0.0001828130010365631,
      "loss": 3.7858,
      "step": 92190
    },
    {
      "epoch": 0.0859453186892267,
      "grad_norm": 1.1275428533554077,
      "learning_rate": 0.0001828111367144679,
      "loss": 3.7337,
      "step": 92200
    },
    {
      "epoch": 0.08595464030730578,
      "grad_norm": 1.1723511219024658,
      "learning_rate": 0.0001828092723923727,
      "loss": 3.785,
      "step": 92210
    },
    {
      "epoch": 0.08596396192538487,
      "grad_norm": 0.9123407006263733,
      "learning_rate": 0.0001828074080702775,
      "loss": 3.7378,
      "step": 92220
    },
    {
      "epoch": 0.08597328354346397,
      "grad_norm": 1.1141326427459717,
      "learning_rate": 0.00018280554374818227,
      "loss": 3.7174,
      "step": 92230
    },
    {
      "epoch": 0.08598260516154306,
      "grad_norm": 1.0499588251113892,
      "learning_rate": 0.0001828036794260871,
      "loss": 3.6753,
      "step": 92240
    },
    {
      "epoch": 0.08599192677962215,
      "grad_norm": 0.9886652827262878,
      "learning_rate": 0.0001828018151039919,
      "loss": 3.962,
      "step": 92250
    },
    {
      "epoch": 0.08600124839770125,
      "grad_norm": 1.0283552408218384,
      "learning_rate": 0.0001827999507818967,
      "loss": 3.7271,
      "step": 92260
    },
    {
      "epoch": 0.08601057001578033,
      "grad_norm": 1.135788917541504,
      "learning_rate": 0.0001827980864598015,
      "loss": 3.5497,
      "step": 92270
    },
    {
      "epoch": 0.08601989163385942,
      "grad_norm": 1.1666066646575928,
      "learning_rate": 0.00018279622213770628,
      "loss": 3.7747,
      "step": 92280
    },
    {
      "epoch": 0.08602921325193852,
      "grad_norm": 0.9434643387794495,
      "learning_rate": 0.00018279435781561108,
      "loss": 3.8157,
      "step": 92290
    },
    {
      "epoch": 0.08603853487001761,
      "grad_norm": 1.0942015647888184,
      "learning_rate": 0.00018279249349351592,
      "loss": 3.7458,
      "step": 92300
    },
    {
      "epoch": 0.08604785648809671,
      "grad_norm": 0.9295677542686462,
      "learning_rate": 0.0001827906291714207,
      "loss": 3.8744,
      "step": 92310
    },
    {
      "epoch": 0.08605717810617579,
      "grad_norm": 0.7961344718933105,
      "learning_rate": 0.0001827887648493255,
      "loss": 3.7764,
      "step": 92320
    },
    {
      "epoch": 0.08606649972425488,
      "grad_norm": 0.9612470269203186,
      "learning_rate": 0.00018278690052723028,
      "loss": 3.6323,
      "step": 92330
    },
    {
      "epoch": 0.08607582134233398,
      "grad_norm": 1.02345871925354,
      "learning_rate": 0.0001827850362051351,
      "loss": 3.4557,
      "step": 92340
    },
    {
      "epoch": 0.08608514296041307,
      "grad_norm": 1.1972495317459106,
      "learning_rate": 0.0001827831718830399,
      "loss": 3.7825,
      "step": 92350
    },
    {
      "epoch": 0.08609446457849217,
      "grad_norm": 0.9293134212493896,
      "learning_rate": 0.0001827813075609447,
      "loss": 3.805,
      "step": 92360
    },
    {
      "epoch": 0.08610378619657125,
      "grad_norm": 1.2199037075042725,
      "learning_rate": 0.0001827794432388495,
      "loss": 3.847,
      "step": 92370
    },
    {
      "epoch": 0.08611310781465034,
      "grad_norm": 0.9881645441055298,
      "learning_rate": 0.00018277757891675429,
      "loss": 3.5468,
      "step": 92380
    },
    {
      "epoch": 0.08612242943272944,
      "grad_norm": 1.0352749824523926,
      "learning_rate": 0.0001827757145946591,
      "loss": 3.8798,
      "step": 92390
    },
    {
      "epoch": 0.08613175105080853,
      "grad_norm": 1.4168922901153564,
      "learning_rate": 0.0001827738502725639,
      "loss": 3.7717,
      "step": 92400
    },
    {
      "epoch": 0.08614107266888762,
      "grad_norm": 1.1408625841140747,
      "learning_rate": 0.0001827719859504687,
      "loss": 3.6308,
      "step": 92410
    },
    {
      "epoch": 0.0861503942869667,
      "grad_norm": 0.9942954182624817,
      "learning_rate": 0.0001827701216283735,
      "loss": 3.6911,
      "step": 92420
    },
    {
      "epoch": 0.0861597159050458,
      "grad_norm": 1.1280535459518433,
      "learning_rate": 0.0001827682573062783,
      "loss": 3.6809,
      "step": 92430
    },
    {
      "epoch": 0.0861690375231249,
      "grad_norm": 1.1652402877807617,
      "learning_rate": 0.0001827663929841831,
      "loss": 3.8829,
      "step": 92440
    },
    {
      "epoch": 0.08617835914120399,
      "grad_norm": 0.9848980903625488,
      "learning_rate": 0.0001827645286620879,
      "loss": 3.9533,
      "step": 92450
    },
    {
      "epoch": 0.08618768075928308,
      "grad_norm": 1.0514485836029053,
      "learning_rate": 0.0001827626643399927,
      "loss": 3.9749,
      "step": 92460
    },
    {
      "epoch": 0.08619700237736216,
      "grad_norm": 1.1599797010421753,
      "learning_rate": 0.00018276080001789752,
      "loss": 3.8111,
      "step": 92470
    },
    {
      "epoch": 0.08620632399544126,
      "grad_norm": 0.9118002653121948,
      "learning_rate": 0.0001827589356958023,
      "loss": 3.8366,
      "step": 92480
    },
    {
      "epoch": 0.08621564561352035,
      "grad_norm": 1.0661038160324097,
      "learning_rate": 0.0001827570713737071,
      "loss": 3.9118,
      "step": 92490
    },
    {
      "epoch": 0.08622496723159945,
      "grad_norm": 0.962046205997467,
      "learning_rate": 0.0001827552070516119,
      "loss": 3.9425,
      "step": 92500
    },
    {
      "epoch": 0.08623428884967854,
      "grad_norm": 1.0737711191177368,
      "learning_rate": 0.00018275334272951669,
      "loss": 3.5492,
      "step": 92510
    },
    {
      "epoch": 0.08624361046775764,
      "grad_norm": 1.0150538682937622,
      "learning_rate": 0.0001827514784074215,
      "loss": 3.7684,
      "step": 92520
    },
    {
      "epoch": 0.08625293208583672,
      "grad_norm": 1.0210139751434326,
      "learning_rate": 0.0001827496140853263,
      "loss": 3.6943,
      "step": 92530
    },
    {
      "epoch": 0.08626225370391581,
      "grad_norm": 1.0118932723999023,
      "learning_rate": 0.0001827477497632311,
      "loss": 3.6937,
      "step": 92540
    },
    {
      "epoch": 0.0862715753219949,
      "grad_norm": 0.9808450937271118,
      "learning_rate": 0.0001827458854411359,
      "loss": 3.8207,
      "step": 92550
    },
    {
      "epoch": 0.086280896940074,
      "grad_norm": 1.3859167098999023,
      "learning_rate": 0.0001827440211190407,
      "loss": 3.7011,
      "step": 92560
    },
    {
      "epoch": 0.0862902185581531,
      "grad_norm": 1.1238620281219482,
      "learning_rate": 0.0001827421567969455,
      "loss": 3.8026,
      "step": 92570
    },
    {
      "epoch": 0.08629954017623218,
      "grad_norm": 0.9747999906539917,
      "learning_rate": 0.0001827402924748503,
      "loss": 3.6871,
      "step": 92580
    },
    {
      "epoch": 0.08630886179431127,
      "grad_norm": 1.1222686767578125,
      "learning_rate": 0.0001827384281527551,
      "loss": 3.7837,
      "step": 92590
    },
    {
      "epoch": 0.08631818341239036,
      "grad_norm": 1.2889455556869507,
      "learning_rate": 0.00018273656383065992,
      "loss": 3.6458,
      "step": 92600
    },
    {
      "epoch": 0.08632750503046946,
      "grad_norm": 1.1329188346862793,
      "learning_rate": 0.0001827346995085647,
      "loss": 3.7822,
      "step": 92610
    },
    {
      "epoch": 0.08633682664854855,
      "grad_norm": 1.0484437942504883,
      "learning_rate": 0.0001827328351864695,
      "loss": 3.6197,
      "step": 92620
    },
    {
      "epoch": 0.08634614826662763,
      "grad_norm": 1.1262767314910889,
      "learning_rate": 0.0001827309708643743,
      "loss": 3.9112,
      "step": 92630
    },
    {
      "epoch": 0.08635546988470673,
      "grad_norm": 1.0254515409469604,
      "learning_rate": 0.00018272910654227911,
      "loss": 3.6705,
      "step": 92640
    },
    {
      "epoch": 0.08636479150278582,
      "grad_norm": 1.0246096849441528,
      "learning_rate": 0.00018272724222018392,
      "loss": 3.7399,
      "step": 92650
    },
    {
      "epoch": 0.08637411312086492,
      "grad_norm": 1.1553176641464233,
      "learning_rate": 0.0001827253778980887,
      "loss": 3.6929,
      "step": 92660
    },
    {
      "epoch": 0.08638343473894401,
      "grad_norm": 1.2758427858352661,
      "learning_rate": 0.0001827235135759935,
      "loss": 3.6859,
      "step": 92670
    },
    {
      "epoch": 0.08639275635702309,
      "grad_norm": 0.8455074429512024,
      "learning_rate": 0.00018272164925389828,
      "loss": 3.8036,
      "step": 92680
    },
    {
      "epoch": 0.08640207797510219,
      "grad_norm": 0.9437400698661804,
      "learning_rate": 0.00018271978493180312,
      "loss": 3.7878,
      "step": 92690
    },
    {
      "epoch": 0.08641139959318128,
      "grad_norm": 1.22421133518219,
      "learning_rate": 0.00018271792060970792,
      "loss": 3.7888,
      "step": 92700
    },
    {
      "epoch": 0.08642072121126038,
      "grad_norm": 0.9158443212509155,
      "learning_rate": 0.0001827160562876127,
      "loss": 3.8523,
      "step": 92710
    },
    {
      "epoch": 0.08643004282933947,
      "grad_norm": 1.2208123207092285,
      "learning_rate": 0.0001827141919655175,
      "loss": 3.7847,
      "step": 92720
    },
    {
      "epoch": 0.08643936444741857,
      "grad_norm": 1.1434334516525269,
      "learning_rate": 0.0001827123276434223,
      "loss": 3.6502,
      "step": 92730
    },
    {
      "epoch": 0.08644868606549765,
      "grad_norm": 0.9961088299751282,
      "learning_rate": 0.0001827104633213271,
      "loss": 3.9518,
      "step": 92740
    },
    {
      "epoch": 0.08645800768357674,
      "grad_norm": 1.2815883159637451,
      "learning_rate": 0.00018270859899923193,
      "loss": 3.6345,
      "step": 92750
    },
    {
      "epoch": 0.08646732930165583,
      "grad_norm": 1.0437499284744263,
      "learning_rate": 0.0001827067346771367,
      "loss": 3.742,
      "step": 92760
    },
    {
      "epoch": 0.08647665091973493,
      "grad_norm": 0.9877454042434692,
      "learning_rate": 0.00018270487035504151,
      "loss": 3.6843,
      "step": 92770
    },
    {
      "epoch": 0.08648597253781402,
      "grad_norm": 1.0083656311035156,
      "learning_rate": 0.0001827030060329463,
      "loss": 3.7117,
      "step": 92780
    },
    {
      "epoch": 0.0864952941558931,
      "grad_norm": 1.0016692876815796,
      "learning_rate": 0.0001827011417108511,
      "loss": 3.6984,
      "step": 92790
    },
    {
      "epoch": 0.0865046157739722,
      "grad_norm": 1.0308375358581543,
      "learning_rate": 0.0001826992773887559,
      "loss": 3.5279,
      "step": 92800
    },
    {
      "epoch": 0.0865139373920513,
      "grad_norm": 1.0741829872131348,
      "learning_rate": 0.0001826974130666607,
      "loss": 3.8777,
      "step": 92810
    },
    {
      "epoch": 0.08652325901013039,
      "grad_norm": 1.021083950996399,
      "learning_rate": 0.00018269554874456552,
      "loss": 3.697,
      "step": 92820
    },
    {
      "epoch": 0.08653258062820948,
      "grad_norm": 1.0431534051895142,
      "learning_rate": 0.0001826936844224703,
      "loss": 3.7033,
      "step": 92830
    },
    {
      "epoch": 0.08654190224628856,
      "grad_norm": 1.259042739868164,
      "learning_rate": 0.0001826918201003751,
      "loss": 3.7239,
      "step": 92840
    },
    {
      "epoch": 0.08655122386436766,
      "grad_norm": 1.1675244569778442,
      "learning_rate": 0.0001826899557782799,
      "loss": 3.9186,
      "step": 92850
    },
    {
      "epoch": 0.08656054548244675,
      "grad_norm": 1.014867901802063,
      "learning_rate": 0.00018268809145618472,
      "loss": 3.7906,
      "step": 92860
    },
    {
      "epoch": 0.08656986710052585,
      "grad_norm": 1.0781092643737793,
      "learning_rate": 0.00018268622713408952,
      "loss": 3.7082,
      "step": 92870
    },
    {
      "epoch": 0.08657918871860494,
      "grad_norm": 1.0274717807769775,
      "learning_rate": 0.0001826843628119943,
      "loss": 3.8289,
      "step": 92880
    },
    {
      "epoch": 0.08658851033668402,
      "grad_norm": 0.9954479932785034,
      "learning_rate": 0.0001826824984898991,
      "loss": 3.9024,
      "step": 92890
    },
    {
      "epoch": 0.08659783195476312,
      "grad_norm": 1.1817512512207031,
      "learning_rate": 0.00018268063416780392,
      "loss": 3.8965,
      "step": 92900
    },
    {
      "epoch": 0.08660715357284221,
      "grad_norm": 1.1876215934753418,
      "learning_rate": 0.00018267876984570872,
      "loss": 3.7234,
      "step": 92910
    },
    {
      "epoch": 0.0866164751909213,
      "grad_norm": 1.0260835886001587,
      "learning_rate": 0.00018267690552361353,
      "loss": 3.8349,
      "step": 92920
    },
    {
      "epoch": 0.0866257968090004,
      "grad_norm": 1.0840193033218384,
      "learning_rate": 0.0001826750412015183,
      "loss": 3.6391,
      "step": 92930
    },
    {
      "epoch": 0.0866351184270795,
      "grad_norm": 0.9923054575920105,
      "learning_rate": 0.0001826731768794231,
      "loss": 3.7567,
      "step": 92940
    },
    {
      "epoch": 0.08664444004515857,
      "grad_norm": 1.2081620693206787,
      "learning_rate": 0.00018267131255732792,
      "loss": 3.7296,
      "step": 92950
    },
    {
      "epoch": 0.08665376166323767,
      "grad_norm": 1.042267918586731,
      "learning_rate": 0.0001826694482352327,
      "loss": 3.8432,
      "step": 92960
    },
    {
      "epoch": 0.08666308328131676,
      "grad_norm": 1.2646610736846924,
      "learning_rate": 0.00018266758391313753,
      "loss": 3.3435,
      "step": 92970
    },
    {
      "epoch": 0.08667240489939586,
      "grad_norm": 1.1522018909454346,
      "learning_rate": 0.0001826657195910423,
      "loss": 3.785,
      "step": 92980
    },
    {
      "epoch": 0.08668172651747495,
      "grad_norm": 1.1322611570358276,
      "learning_rate": 0.00018266385526894712,
      "loss": 3.7722,
      "step": 92990
    },
    {
      "epoch": 0.08669104813555403,
      "grad_norm": 1.182770013809204,
      "learning_rate": 0.00018266199094685192,
      "loss": 3.8085,
      "step": 93000
    },
    {
      "epoch": 0.08670036975363313,
      "grad_norm": 1.0537338256835938,
      "learning_rate": 0.0001826601266247567,
      "loss": 3.8421,
      "step": 93010
    },
    {
      "epoch": 0.08670969137171222,
      "grad_norm": 1.166854739189148,
      "learning_rate": 0.0001826582623026615,
      "loss": 3.7124,
      "step": 93020
    },
    {
      "epoch": 0.08671901298979132,
      "grad_norm": 1.045494794845581,
      "learning_rate": 0.00018265639798056632,
      "loss": 3.7723,
      "step": 93030
    },
    {
      "epoch": 0.08672833460787041,
      "grad_norm": 1.1708658933639526,
      "learning_rate": 0.00018265453365847112,
      "loss": 3.7752,
      "step": 93040
    },
    {
      "epoch": 0.08673765622594949,
      "grad_norm": 1.1158781051635742,
      "learning_rate": 0.00018265266933637593,
      "loss": 3.5539,
      "step": 93050
    },
    {
      "epoch": 0.08674697784402859,
      "grad_norm": 1.0977542400360107,
      "learning_rate": 0.0001826508050142807,
      "loss": 3.7657,
      "step": 93060
    },
    {
      "epoch": 0.08675629946210768,
      "grad_norm": 1.2296364307403564,
      "learning_rate": 0.00018264894069218551,
      "loss": 3.8095,
      "step": 93070
    },
    {
      "epoch": 0.08676562108018677,
      "grad_norm": 1.1067864894866943,
      "learning_rate": 0.00018264707637009032,
      "loss": 3.7186,
      "step": 93080
    },
    {
      "epoch": 0.08677494269826587,
      "grad_norm": 1.1367707252502441,
      "learning_rate": 0.00018264521204799513,
      "loss": 3.8744,
      "step": 93090
    },
    {
      "epoch": 0.08678426431634495,
      "grad_norm": 1.1129093170166016,
      "learning_rate": 0.00018264334772589993,
      "loss": 3.9574,
      "step": 93100
    },
    {
      "epoch": 0.08679358593442404,
      "grad_norm": 1.166908621788025,
      "learning_rate": 0.0001826414834038047,
      "loss": 3.7147,
      "step": 93110
    },
    {
      "epoch": 0.08680290755250314,
      "grad_norm": 1.0785411596298218,
      "learning_rate": 0.00018263961908170952,
      "loss": 3.7902,
      "step": 93120
    },
    {
      "epoch": 0.08681222917058223,
      "grad_norm": 1.0387576818466187,
      "learning_rate": 0.0001826377547596143,
      "loss": 3.5908,
      "step": 93130
    },
    {
      "epoch": 0.08682155078866133,
      "grad_norm": 0.9810062646865845,
      "learning_rate": 0.00018263589043751913,
      "loss": 3.7515,
      "step": 93140
    },
    {
      "epoch": 0.08683087240674041,
      "grad_norm": 1.1319023370742798,
      "learning_rate": 0.00018263402611542394,
      "loss": 4.0525,
      "step": 93150
    },
    {
      "epoch": 0.0868401940248195,
      "grad_norm": 1.1542242765426636,
      "learning_rate": 0.00018263216179332872,
      "loss": 3.7253,
      "step": 93160
    },
    {
      "epoch": 0.0868495156428986,
      "grad_norm": 1.2739866971969604,
      "learning_rate": 0.00018263029747123352,
      "loss": 3.7179,
      "step": 93170
    },
    {
      "epoch": 0.08685883726097769,
      "grad_norm": 1.0725935697555542,
      "learning_rate": 0.0001826284331491383,
      "loss": 3.6599,
      "step": 93180
    },
    {
      "epoch": 0.08686815887905679,
      "grad_norm": 1.037644863128662,
      "learning_rate": 0.0001826265688270431,
      "loss": 3.6469,
      "step": 93190
    },
    {
      "epoch": 0.08687748049713588,
      "grad_norm": 1.0800327062606812,
      "learning_rate": 0.00018262470450494794,
      "loss": 3.6072,
      "step": 93200
    },
    {
      "epoch": 0.08688680211521496,
      "grad_norm": 1.0390174388885498,
      "learning_rate": 0.00018262284018285272,
      "loss": 3.8019,
      "step": 93210
    },
    {
      "epoch": 0.08689612373329406,
      "grad_norm": 1.048498272895813,
      "learning_rate": 0.00018262097586075753,
      "loss": 3.8643,
      "step": 93220
    },
    {
      "epoch": 0.08690544535137315,
      "grad_norm": 1.0794827938079834,
      "learning_rate": 0.0001826191115386623,
      "loss": 3.8858,
      "step": 93230
    },
    {
      "epoch": 0.08691476696945225,
      "grad_norm": 1.0929923057556152,
      "learning_rate": 0.0001826172472165671,
      "loss": 3.8258,
      "step": 93240
    },
    {
      "epoch": 0.08692408858753134,
      "grad_norm": 1.0211776494979858,
      "learning_rate": 0.00018261538289447192,
      "loss": 3.6533,
      "step": 93250
    },
    {
      "epoch": 0.08693341020561042,
      "grad_norm": 1.0055075883865356,
      "learning_rate": 0.00018261351857237673,
      "loss": 3.6023,
      "step": 93260
    },
    {
      "epoch": 0.08694273182368951,
      "grad_norm": 1.037573218345642,
      "learning_rate": 0.00018261165425028153,
      "loss": 3.7864,
      "step": 93270
    },
    {
      "epoch": 0.08695205344176861,
      "grad_norm": 1.041413426399231,
      "learning_rate": 0.0001826097899281863,
      "loss": 3.7981,
      "step": 93280
    },
    {
      "epoch": 0.0869613750598477,
      "grad_norm": 1.118214726448059,
      "learning_rate": 0.00018260792560609112,
      "loss": 3.7304,
      "step": 93290
    },
    {
      "epoch": 0.0869706966779268,
      "grad_norm": 1.3001654148101807,
      "learning_rate": 0.00018260606128399592,
      "loss": 3.8817,
      "step": 93300
    },
    {
      "epoch": 0.08698001829600588,
      "grad_norm": 1.1548720598220825,
      "learning_rate": 0.00018260419696190073,
      "loss": 3.6361,
      "step": 93310
    },
    {
      "epoch": 0.08698933991408497,
      "grad_norm": 1.1281657218933105,
      "learning_rate": 0.00018260233263980554,
      "loss": 3.4879,
      "step": 93320
    },
    {
      "epoch": 0.08699866153216407,
      "grad_norm": 1.060011625289917,
      "learning_rate": 0.00018260046831771032,
      "loss": 3.7981,
      "step": 93330
    },
    {
      "epoch": 0.08700798315024316,
      "grad_norm": 1.096727967262268,
      "learning_rate": 0.00018259860399561512,
      "loss": 3.7614,
      "step": 93340
    },
    {
      "epoch": 0.08701730476832226,
      "grad_norm": 1.0916783809661865,
      "learning_rate": 0.00018259673967351993,
      "loss": 3.7603,
      "step": 93350
    },
    {
      "epoch": 0.08702662638640134,
      "grad_norm": 0.9567083716392517,
      "learning_rate": 0.00018259487535142473,
      "loss": 3.9001,
      "step": 93360
    },
    {
      "epoch": 0.08703594800448043,
      "grad_norm": 1.0766583681106567,
      "learning_rate": 0.00018259301102932954,
      "loss": 3.8201,
      "step": 93370
    },
    {
      "epoch": 0.08704526962255953,
      "grad_norm": 1.3222333192825317,
      "learning_rate": 0.00018259114670723432,
      "loss": 3.7056,
      "step": 93380
    },
    {
      "epoch": 0.08705459124063862,
      "grad_norm": 0.9780269861221313,
      "learning_rate": 0.00018258928238513913,
      "loss": 3.9485,
      "step": 93390
    },
    {
      "epoch": 0.08706391285871772,
      "grad_norm": 1.0345313549041748,
      "learning_rate": 0.00018258741806304393,
      "loss": 3.7944,
      "step": 93400
    },
    {
      "epoch": 0.08707323447679681,
      "grad_norm": 1.1642400026321411,
      "learning_rate": 0.0001825855537409487,
      "loss": 3.6292,
      "step": 93410
    },
    {
      "epoch": 0.08708255609487589,
      "grad_norm": 1.143115758895874,
      "learning_rate": 0.00018258368941885355,
      "loss": 3.8167,
      "step": 93420
    },
    {
      "epoch": 0.08709187771295498,
      "grad_norm": 1.0082402229309082,
      "learning_rate": 0.00018258182509675832,
      "loss": 3.7105,
      "step": 93430
    },
    {
      "epoch": 0.08710119933103408,
      "grad_norm": 0.9888754487037659,
      "learning_rate": 0.00018257996077466313,
      "loss": 3.6978,
      "step": 93440
    },
    {
      "epoch": 0.08711052094911317,
      "grad_norm": 1.10225248336792,
      "learning_rate": 0.0001825780964525679,
      "loss": 3.8054,
      "step": 93450
    },
    {
      "epoch": 0.08711984256719227,
      "grad_norm": 1.069490909576416,
      "learning_rate": 0.00018257623213047272,
      "loss": 3.6911,
      "step": 93460
    },
    {
      "epoch": 0.08712916418527135,
      "grad_norm": 0.9425909519195557,
      "learning_rate": 0.00018257436780837752,
      "loss": 3.7532,
      "step": 93470
    },
    {
      "epoch": 0.08713848580335044,
      "grad_norm": 1.0958833694458008,
      "learning_rate": 0.00018257250348628233,
      "loss": 3.6088,
      "step": 93480
    },
    {
      "epoch": 0.08714780742142954,
      "grad_norm": 0.9080066084861755,
      "learning_rate": 0.00018257063916418714,
      "loss": 3.7813,
      "step": 93490
    },
    {
      "epoch": 0.08715712903950863,
      "grad_norm": 1.2380741834640503,
      "learning_rate": 0.00018256877484209191,
      "loss": 3.7435,
      "step": 93500
    },
    {
      "epoch": 0.08716645065758773,
      "grad_norm": 1.0448205471038818,
      "learning_rate": 0.00018256691051999672,
      "loss": 3.7777,
      "step": 93510
    },
    {
      "epoch": 0.08717577227566681,
      "grad_norm": 0.9547194242477417,
      "learning_rate": 0.00018256504619790153,
      "loss": 3.855,
      "step": 93520
    },
    {
      "epoch": 0.0871850938937459,
      "grad_norm": 0.975357711315155,
      "learning_rate": 0.00018256318187580633,
      "loss": 3.8779,
      "step": 93530
    },
    {
      "epoch": 0.087194415511825,
      "grad_norm": 0.989124059677124,
      "learning_rate": 0.00018256131755371114,
      "loss": 3.7548,
      "step": 93540
    },
    {
      "epoch": 0.08720373712990409,
      "grad_norm": 1.0170668363571167,
      "learning_rate": 0.00018255945323161592,
      "loss": 3.818,
      "step": 93550
    },
    {
      "epoch": 0.08721305874798319,
      "grad_norm": 1.0976616144180298,
      "learning_rate": 0.00018255758890952073,
      "loss": 3.9425,
      "step": 93560
    },
    {
      "epoch": 0.08722238036606227,
      "grad_norm": 1.0633999109268188,
      "learning_rate": 0.00018255572458742553,
      "loss": 3.6819,
      "step": 93570
    },
    {
      "epoch": 0.08723170198414136,
      "grad_norm": 0.8736644387245178,
      "learning_rate": 0.0001825538602653303,
      "loss": 3.8678,
      "step": 93580
    },
    {
      "epoch": 0.08724102360222045,
      "grad_norm": 1.1980876922607422,
      "learning_rate": 0.00018255199594323514,
      "loss": 3.7637,
      "step": 93590
    },
    {
      "epoch": 0.08725034522029955,
      "grad_norm": 0.9796538352966309,
      "learning_rate": 0.00018255013162113992,
      "loss": 3.5071,
      "step": 93600
    },
    {
      "epoch": 0.08725966683837864,
      "grad_norm": 1.0341980457305908,
      "learning_rate": 0.00018254826729904473,
      "loss": 3.8663,
      "step": 93610
    },
    {
      "epoch": 0.08726898845645774,
      "grad_norm": 1.1159693002700806,
      "learning_rate": 0.00018254640297694954,
      "loss": 3.752,
      "step": 93620
    },
    {
      "epoch": 0.08727831007453682,
      "grad_norm": 1.2355693578720093,
      "learning_rate": 0.00018254453865485431,
      "loss": 3.7363,
      "step": 93630
    },
    {
      "epoch": 0.08728763169261591,
      "grad_norm": 1.1588774919509888,
      "learning_rate": 0.00018254267433275912,
      "loss": 3.7471,
      "step": 93640
    },
    {
      "epoch": 0.08729695331069501,
      "grad_norm": 1.0558507442474365,
      "learning_rate": 0.00018254081001066393,
      "loss": 3.6777,
      "step": 93650
    },
    {
      "epoch": 0.0873062749287741,
      "grad_norm": 1.1953808069229126,
      "learning_rate": 0.00018253894568856873,
      "loss": 3.5743,
      "step": 93660
    },
    {
      "epoch": 0.0873155965468532,
      "grad_norm": 1.107887625694275,
      "learning_rate": 0.00018253708136647354,
      "loss": 3.8381,
      "step": 93670
    },
    {
      "epoch": 0.08732491816493228,
      "grad_norm": 1.1667391061782837,
      "learning_rate": 0.00018253521704437832,
      "loss": 3.7037,
      "step": 93680
    },
    {
      "epoch": 0.08733423978301137,
      "grad_norm": 1.1393359899520874,
      "learning_rate": 0.00018253335272228313,
      "loss": 3.7707,
      "step": 93690
    },
    {
      "epoch": 0.08734356140109047,
      "grad_norm": 1.239298701286316,
      "learning_rate": 0.00018253148840018793,
      "loss": 3.4794,
      "step": 93700
    },
    {
      "epoch": 0.08735288301916956,
      "grad_norm": 1.0128724575042725,
      "learning_rate": 0.00018252962407809274,
      "loss": 3.8226,
      "step": 93710
    },
    {
      "epoch": 0.08736220463724866,
      "grad_norm": 1.0711886882781982,
      "learning_rate": 0.00018252775975599754,
      "loss": 3.8395,
      "step": 93720
    },
    {
      "epoch": 0.08737152625532774,
      "grad_norm": 1.0626580715179443,
      "learning_rate": 0.00018252589543390232,
      "loss": 3.8784,
      "step": 93730
    },
    {
      "epoch": 0.08738084787340683,
      "grad_norm": 0.9288687109947205,
      "learning_rate": 0.00018252403111180713,
      "loss": 3.9232,
      "step": 93740
    },
    {
      "epoch": 0.08739016949148593,
      "grad_norm": 1.2648683786392212,
      "learning_rate": 0.00018252216678971194,
      "loss": 3.6054,
      "step": 93750
    },
    {
      "epoch": 0.08739949110956502,
      "grad_norm": 1.2629332542419434,
      "learning_rate": 0.00018252030246761674,
      "loss": 3.7543,
      "step": 93760
    },
    {
      "epoch": 0.08740881272764411,
      "grad_norm": 1.1963118314743042,
      "learning_rate": 0.00018251843814552155,
      "loss": 3.6827,
      "step": 93770
    },
    {
      "epoch": 0.0874181343457232,
      "grad_norm": 1.0772875547409058,
      "learning_rate": 0.00018251657382342633,
      "loss": 3.8923,
      "step": 93780
    },
    {
      "epoch": 0.08742745596380229,
      "grad_norm": 1.3927723169326782,
      "learning_rate": 0.00018251470950133113,
      "loss": 3.6576,
      "step": 93790
    },
    {
      "epoch": 0.08743677758188138,
      "grad_norm": 0.9365825653076172,
      "learning_rate": 0.00018251284517923591,
      "loss": 3.6384,
      "step": 93800
    },
    {
      "epoch": 0.08744609919996048,
      "grad_norm": 0.8377958536148071,
      "learning_rate": 0.00018251098085714075,
      "loss": 3.8246,
      "step": 93810
    },
    {
      "epoch": 0.08745542081803957,
      "grad_norm": 1.0025948286056519,
      "learning_rate": 0.00018250911653504555,
      "loss": 3.9178,
      "step": 93820
    },
    {
      "epoch": 0.08746474243611865,
      "grad_norm": 1.0904184579849243,
      "learning_rate": 0.00018250725221295033,
      "loss": 3.6921,
      "step": 93830
    },
    {
      "epoch": 0.08747406405419775,
      "grad_norm": 1.0243252515792847,
      "learning_rate": 0.00018250538789085514,
      "loss": 3.661,
      "step": 93840
    },
    {
      "epoch": 0.08748338567227684,
      "grad_norm": 1.1304911375045776,
      "learning_rate": 0.00018250352356875992,
      "loss": 3.4825,
      "step": 93850
    },
    {
      "epoch": 0.08749270729035594,
      "grad_norm": 1.009177327156067,
      "learning_rate": 0.00018250165924666472,
      "loss": 3.9549,
      "step": 93860
    },
    {
      "epoch": 0.08750202890843503,
      "grad_norm": 1.1433193683624268,
      "learning_rate": 0.00018249979492456956,
      "loss": 3.6521,
      "step": 93870
    },
    {
      "epoch": 0.08751135052651413,
      "grad_norm": 1.0642969608306885,
      "learning_rate": 0.00018249793060247434,
      "loss": 3.636,
      "step": 93880
    },
    {
      "epoch": 0.0875206721445932,
      "grad_norm": 0.9821270108222961,
      "learning_rate": 0.00018249606628037914,
      "loss": 3.6821,
      "step": 93890
    },
    {
      "epoch": 0.0875299937626723,
      "grad_norm": 1.031089186668396,
      "learning_rate": 0.00018249420195828392,
      "loss": 3.8514,
      "step": 93900
    },
    {
      "epoch": 0.0875393153807514,
      "grad_norm": 1.0199147462844849,
      "learning_rate": 0.00018249233763618873,
      "loss": 3.8577,
      "step": 93910
    },
    {
      "epoch": 0.08754863699883049,
      "grad_norm": 1.216715931892395,
      "learning_rate": 0.00018249047331409354,
      "loss": 3.745,
      "step": 93920
    },
    {
      "epoch": 0.08755795861690958,
      "grad_norm": 1.0666285753250122,
      "learning_rate": 0.00018248860899199834,
      "loss": 3.7415,
      "step": 93930
    },
    {
      "epoch": 0.08756728023498866,
      "grad_norm": 1.2199385166168213,
      "learning_rate": 0.00018248674466990315,
      "loss": 3.7278,
      "step": 93940
    },
    {
      "epoch": 0.08757660185306776,
      "grad_norm": 0.9967522025108337,
      "learning_rate": 0.00018248488034780793,
      "loss": 3.9221,
      "step": 93950
    },
    {
      "epoch": 0.08758592347114685,
      "grad_norm": 1.1867256164550781,
      "learning_rate": 0.00018248301602571273,
      "loss": 3.8027,
      "step": 93960
    },
    {
      "epoch": 0.08759524508922595,
      "grad_norm": 1.1213940382003784,
      "learning_rate": 0.00018248115170361754,
      "loss": 3.6591,
      "step": 93970
    },
    {
      "epoch": 0.08760456670730504,
      "grad_norm": 1.1576714515686035,
      "learning_rate": 0.00018247928738152235,
      "loss": 3.9124,
      "step": 93980
    },
    {
      "epoch": 0.08761388832538412,
      "grad_norm": 0.9161725044250488,
      "learning_rate": 0.00018247742305942715,
      "loss": 3.9321,
      "step": 93990
    },
    {
      "epoch": 0.08762320994346322,
      "grad_norm": 1.146060585975647,
      "learning_rate": 0.00018247555873733193,
      "loss": 3.6602,
      "step": 94000
    },
    {
      "epoch": 0.08763253156154231,
      "grad_norm": 1.0825690031051636,
      "learning_rate": 0.00018247369441523674,
      "loss": 3.689,
      "step": 94010
    },
    {
      "epoch": 0.0876418531796214,
      "grad_norm": 1.0173587799072266,
      "learning_rate": 0.00018247183009314154,
      "loss": 3.696,
      "step": 94020
    },
    {
      "epoch": 0.0876511747977005,
      "grad_norm": 1.1622083187103271,
      "learning_rate": 0.00018246996577104632,
      "loss": 3.564,
      "step": 94030
    },
    {
      "epoch": 0.08766049641577958,
      "grad_norm": 1.1053516864776611,
      "learning_rate": 0.00018246810144895116,
      "loss": 3.5584,
      "step": 94040
    },
    {
      "epoch": 0.08766981803385868,
      "grad_norm": 1.2169045209884644,
      "learning_rate": 0.00018246623712685594,
      "loss": 3.8373,
      "step": 94050
    },
    {
      "epoch": 0.08767913965193777,
      "grad_norm": 1.0819542407989502,
      "learning_rate": 0.00018246437280476074,
      "loss": 3.9655,
      "step": 94060
    },
    {
      "epoch": 0.08768846127001687,
      "grad_norm": 1.113237738609314,
      "learning_rate": 0.00018246250848266555,
      "loss": 3.8632,
      "step": 94070
    },
    {
      "epoch": 0.08769778288809596,
      "grad_norm": 0.9915596842765808,
      "learning_rate": 0.00018246064416057033,
      "loss": 3.7569,
      "step": 94080
    },
    {
      "epoch": 0.08770710450617505,
      "grad_norm": 1.0166910886764526,
      "learning_rate": 0.00018245877983847513,
      "loss": 3.4889,
      "step": 94090
    },
    {
      "epoch": 0.08771642612425413,
      "grad_norm": 1.0208711624145508,
      "learning_rate": 0.00018245691551637994,
      "loss": 3.6746,
      "step": 94100
    },
    {
      "epoch": 0.08772574774233323,
      "grad_norm": 0.9508869051933289,
      "learning_rate": 0.00018245505119428475,
      "loss": 3.646,
      "step": 94110
    },
    {
      "epoch": 0.08773506936041232,
      "grad_norm": 1.0950063467025757,
      "learning_rate": 0.00018245318687218955,
      "loss": 3.7122,
      "step": 94120
    },
    {
      "epoch": 0.08774439097849142,
      "grad_norm": 1.0152391195297241,
      "learning_rate": 0.00018245132255009433,
      "loss": 3.5626,
      "step": 94130
    },
    {
      "epoch": 0.08775371259657051,
      "grad_norm": 1.0187510251998901,
      "learning_rate": 0.00018244945822799914,
      "loss": 3.6418,
      "step": 94140
    },
    {
      "epoch": 0.0877630342146496,
      "grad_norm": 1.1696548461914062,
      "learning_rate": 0.00018244759390590394,
      "loss": 3.5416,
      "step": 94150
    },
    {
      "epoch": 0.08777235583272869,
      "grad_norm": 1.0673760175704956,
      "learning_rate": 0.00018244572958380875,
      "loss": 3.6677,
      "step": 94160
    },
    {
      "epoch": 0.08778167745080778,
      "grad_norm": 1.001146912574768,
      "learning_rate": 0.00018244386526171356,
      "loss": 3.7182,
      "step": 94170
    },
    {
      "epoch": 0.08779099906888688,
      "grad_norm": 0.9981769919395447,
      "learning_rate": 0.00018244200093961834,
      "loss": 3.7795,
      "step": 94180
    },
    {
      "epoch": 0.08780032068696597,
      "grad_norm": 1.0819969177246094,
      "learning_rate": 0.00018244013661752314,
      "loss": 3.9797,
      "step": 94190
    },
    {
      "epoch": 0.08780964230504505,
      "grad_norm": 0.9068520069122314,
      "learning_rate": 0.00018243827229542795,
      "loss": 3.9809,
      "step": 94200
    },
    {
      "epoch": 0.08781896392312415,
      "grad_norm": 0.9831337332725525,
      "learning_rate": 0.00018243640797333276,
      "loss": 3.8216,
      "step": 94210
    },
    {
      "epoch": 0.08782828554120324,
      "grad_norm": 1.8489841222763062,
      "learning_rate": 0.00018243454365123756,
      "loss": 3.6502,
      "step": 94220
    },
    {
      "epoch": 0.08783760715928234,
      "grad_norm": 0.9817131161689758,
      "learning_rate": 0.00018243267932914234,
      "loss": 3.631,
      "step": 94230
    },
    {
      "epoch": 0.08784692877736143,
      "grad_norm": 0.9849207997322083,
      "learning_rate": 0.00018243081500704715,
      "loss": 3.616,
      "step": 94240
    },
    {
      "epoch": 0.08785625039544051,
      "grad_norm": 1.2003759145736694,
      "learning_rate": 0.00018242895068495193,
      "loss": 3.7089,
      "step": 94250
    },
    {
      "epoch": 0.0878655720135196,
      "grad_norm": 1.2108666896820068,
      "learning_rate": 0.00018242708636285676,
      "loss": 3.7362,
      "step": 94260
    },
    {
      "epoch": 0.0878748936315987,
      "grad_norm": 1.0762466192245483,
      "learning_rate": 0.00018242522204076157,
      "loss": 3.6958,
      "step": 94270
    },
    {
      "epoch": 0.0878842152496778,
      "grad_norm": 1.0179500579833984,
      "learning_rate": 0.00018242335771866635,
      "loss": 3.7142,
      "step": 94280
    },
    {
      "epoch": 0.08789353686775689,
      "grad_norm": 1.0877846479415894,
      "learning_rate": 0.00018242149339657115,
      "loss": 3.7912,
      "step": 94290
    },
    {
      "epoch": 0.08790285848583598,
      "grad_norm": 1.045176386833191,
      "learning_rate": 0.00018241962907447593,
      "loss": 3.9148,
      "step": 94300
    },
    {
      "epoch": 0.08791218010391506,
      "grad_norm": 1.09493088722229,
      "learning_rate": 0.00018241776475238074,
      "loss": 3.7157,
      "step": 94310
    },
    {
      "epoch": 0.08792150172199416,
      "grad_norm": 0.9906175136566162,
      "learning_rate": 0.00018241590043028557,
      "loss": 3.8395,
      "step": 94320
    },
    {
      "epoch": 0.08793082334007325,
      "grad_norm": 1.1031864881515503,
      "learning_rate": 0.00018241403610819035,
      "loss": 3.7138,
      "step": 94330
    },
    {
      "epoch": 0.08794014495815235,
      "grad_norm": 1.1383144855499268,
      "learning_rate": 0.00018241217178609516,
      "loss": 3.5642,
      "step": 94340
    },
    {
      "epoch": 0.08794946657623144,
      "grad_norm": 1.1332347393035889,
      "learning_rate": 0.00018241030746399994,
      "loss": 3.6679,
      "step": 94350
    },
    {
      "epoch": 0.08795878819431052,
      "grad_norm": 1.1498445272445679,
      "learning_rate": 0.00018240844314190474,
      "loss": 3.9121,
      "step": 94360
    },
    {
      "epoch": 0.08796810981238962,
      "grad_norm": 1.107792854309082,
      "learning_rate": 0.00018240657881980955,
      "loss": 3.6349,
      "step": 94370
    },
    {
      "epoch": 0.08797743143046871,
      "grad_norm": 0.9927930235862732,
      "learning_rate": 0.00018240471449771435,
      "loss": 3.8839,
      "step": 94380
    },
    {
      "epoch": 0.0879867530485478,
      "grad_norm": 0.9779083728790283,
      "learning_rate": 0.00018240285017561916,
      "loss": 3.7327,
      "step": 94390
    },
    {
      "epoch": 0.0879960746666269,
      "grad_norm": 1.0007153749465942,
      "learning_rate": 0.00018240098585352394,
      "loss": 3.6803,
      "step": 94400
    },
    {
      "epoch": 0.08800539628470598,
      "grad_norm": 0.8982577323913574,
      "learning_rate": 0.00018239912153142875,
      "loss": 3.7534,
      "step": 94410
    },
    {
      "epoch": 0.08801471790278508,
      "grad_norm": 0.9902002811431885,
      "learning_rate": 0.00018239725720933355,
      "loss": 3.6388,
      "step": 94420
    },
    {
      "epoch": 0.08802403952086417,
      "grad_norm": 1.1138153076171875,
      "learning_rate": 0.00018239539288723836,
      "loss": 3.7067,
      "step": 94430
    },
    {
      "epoch": 0.08803336113894326,
      "grad_norm": 1.1534429788589478,
      "learning_rate": 0.00018239352856514317,
      "loss": 3.7322,
      "step": 94440
    },
    {
      "epoch": 0.08804268275702236,
      "grad_norm": 1.0785932540893555,
      "learning_rate": 0.00018239166424304794,
      "loss": 3.801,
      "step": 94450
    },
    {
      "epoch": 0.08805200437510144,
      "grad_norm": 1.0178853273391724,
      "learning_rate": 0.00018238979992095275,
      "loss": 3.6907,
      "step": 94460
    },
    {
      "epoch": 0.08806132599318053,
      "grad_norm": 1.0988880395889282,
      "learning_rate": 0.00018238793559885756,
      "loss": 3.7418,
      "step": 94470
    },
    {
      "epoch": 0.08807064761125963,
      "grad_norm": 1.1403851509094238,
      "learning_rate": 0.00018238607127676234,
      "loss": 3.7144,
      "step": 94480
    },
    {
      "epoch": 0.08807996922933872,
      "grad_norm": 1.029618263244629,
      "learning_rate": 0.00018238420695466717,
      "loss": 3.8389,
      "step": 94490
    },
    {
      "epoch": 0.08808929084741782,
      "grad_norm": 0.8854067325592041,
      "learning_rate": 0.00018238234263257195,
      "loss": 3.7899,
      "step": 94500
    },
    {
      "epoch": 0.0880986124654969,
      "grad_norm": 1.0004901885986328,
      "learning_rate": 0.00018238047831047676,
      "loss": 3.8811,
      "step": 94510
    },
    {
      "epoch": 0.08810793408357599,
      "grad_norm": 1.133546233177185,
      "learning_rate": 0.00018237861398838156,
      "loss": 3.8657,
      "step": 94520
    },
    {
      "epoch": 0.08811725570165509,
      "grad_norm": 1.2757034301757812,
      "learning_rate": 0.00018237674966628634,
      "loss": 3.6575,
      "step": 94530
    },
    {
      "epoch": 0.08812657731973418,
      "grad_norm": 0.9768718481063843,
      "learning_rate": 0.00018237488534419115,
      "loss": 3.6986,
      "step": 94540
    },
    {
      "epoch": 0.08813589893781328,
      "grad_norm": 1.192109227180481,
      "learning_rate": 0.00018237302102209595,
      "loss": 3.6558,
      "step": 94550
    },
    {
      "epoch": 0.08814522055589237,
      "grad_norm": 1.0481202602386475,
      "learning_rate": 0.00018237115670000076,
      "loss": 3.7247,
      "step": 94560
    },
    {
      "epoch": 0.08815454217397145,
      "grad_norm": 1.0640074014663696,
      "learning_rate": 0.00018236929237790557,
      "loss": 3.6165,
      "step": 94570
    },
    {
      "epoch": 0.08816386379205055,
      "grad_norm": 1.1772611141204834,
      "learning_rate": 0.00018236742805581035,
      "loss": 3.577,
      "step": 94580
    },
    {
      "epoch": 0.08817318541012964,
      "grad_norm": 0.8782712817192078,
      "learning_rate": 0.00018236556373371515,
      "loss": 3.7037,
      "step": 94590
    },
    {
      "epoch": 0.08818250702820873,
      "grad_norm": 1.131269097328186,
      "learning_rate": 0.00018236369941161996,
      "loss": 3.8979,
      "step": 94600
    },
    {
      "epoch": 0.08819182864628783,
      "grad_norm": 1.0553396940231323,
      "learning_rate": 0.00018236183508952476,
      "loss": 3.8152,
      "step": 94610
    },
    {
      "epoch": 0.08820115026436691,
      "grad_norm": 0.8412265181541443,
      "learning_rate": 0.00018235997076742957,
      "loss": 3.7751,
      "step": 94620
    },
    {
      "epoch": 0.088210471882446,
      "grad_norm": 1.1442219018936157,
      "learning_rate": 0.00018235810644533435,
      "loss": 3.8087,
      "step": 94630
    },
    {
      "epoch": 0.0882197935005251,
      "grad_norm": 1.1010093688964844,
      "learning_rate": 0.00018235624212323916,
      "loss": 3.7869,
      "step": 94640
    },
    {
      "epoch": 0.08822911511860419,
      "grad_norm": 1.132427453994751,
      "learning_rate": 0.00018235437780114396,
      "loss": 3.6459,
      "step": 94650
    },
    {
      "epoch": 0.08823843673668329,
      "grad_norm": 1.0957763195037842,
      "learning_rate": 0.00018235251347904877,
      "loss": 3.7875,
      "step": 94660
    },
    {
      "epoch": 0.08824775835476237,
      "grad_norm": 1.1160755157470703,
      "learning_rate": 0.00018235064915695357,
      "loss": 3.8175,
      "step": 94670
    },
    {
      "epoch": 0.08825707997284146,
      "grad_norm": 1.0898106098175049,
      "learning_rate": 0.00018234878483485835,
      "loss": 3.7844,
      "step": 94680
    },
    {
      "epoch": 0.08826640159092056,
      "grad_norm": 0.9863345623016357,
      "learning_rate": 0.00018234692051276316,
      "loss": 3.69,
      "step": 94690
    },
    {
      "epoch": 0.08827572320899965,
      "grad_norm": 0.8893749713897705,
      "learning_rate": 0.00018234505619066794,
      "loss": 3.8832,
      "step": 94700
    },
    {
      "epoch": 0.08828504482707875,
      "grad_norm": 1.0585521459579468,
      "learning_rate": 0.00018234319186857277,
      "loss": 3.7962,
      "step": 94710
    },
    {
      "epoch": 0.08829436644515783,
      "grad_norm": 1.044847011566162,
      "learning_rate": 0.00018234132754647758,
      "loss": 3.9392,
      "step": 94720
    },
    {
      "epoch": 0.08830368806323692,
      "grad_norm": 1.0755228996276855,
      "learning_rate": 0.00018233946322438236,
      "loss": 3.8741,
      "step": 94730
    },
    {
      "epoch": 0.08831300968131602,
      "grad_norm": 1.081615924835205,
      "learning_rate": 0.00018233759890228716,
      "loss": 3.8332,
      "step": 94740
    },
    {
      "epoch": 0.08832233129939511,
      "grad_norm": 1.1853135824203491,
      "learning_rate": 0.00018233573458019194,
      "loss": 3.6915,
      "step": 94750
    },
    {
      "epoch": 0.0883316529174742,
      "grad_norm": 1.4123568534851074,
      "learning_rate": 0.00018233387025809675,
      "loss": 3.755,
      "step": 94760
    },
    {
      "epoch": 0.0883409745355533,
      "grad_norm": 1.018868327140808,
      "learning_rate": 0.00018233200593600158,
      "loss": 3.6354,
      "step": 94770
    },
    {
      "epoch": 0.08835029615363238,
      "grad_norm": 1.1829547882080078,
      "learning_rate": 0.00018233014161390636,
      "loss": 3.6119,
      "step": 94780
    },
    {
      "epoch": 0.08835961777171147,
      "grad_norm": 1.044731616973877,
      "learning_rate": 0.00018232827729181117,
      "loss": 3.6162,
      "step": 94790
    },
    {
      "epoch": 0.08836893938979057,
      "grad_norm": 1.011771559715271,
      "learning_rate": 0.00018232641296971595,
      "loss": 3.761,
      "step": 94800
    },
    {
      "epoch": 0.08837826100786966,
      "grad_norm": 0.9886451363563538,
      "learning_rate": 0.00018232454864762075,
      "loss": 3.8409,
      "step": 94810
    },
    {
      "epoch": 0.08838758262594876,
      "grad_norm": 1.1704427003860474,
      "learning_rate": 0.00018232268432552556,
      "loss": 3.7938,
      "step": 94820
    },
    {
      "epoch": 0.08839690424402784,
      "grad_norm": 1.1710292100906372,
      "learning_rate": 0.00018232082000343037,
      "loss": 3.6869,
      "step": 94830
    },
    {
      "epoch": 0.08840622586210693,
      "grad_norm": 1.1234124898910522,
      "learning_rate": 0.00018231895568133517,
      "loss": 3.7761,
      "step": 94840
    },
    {
      "epoch": 0.08841554748018603,
      "grad_norm": 0.9985633492469788,
      "learning_rate": 0.00018231709135923995,
      "loss": 3.7067,
      "step": 94850
    },
    {
      "epoch": 0.08842486909826512,
      "grad_norm": 0.9954522848129272,
      "learning_rate": 0.00018231522703714476,
      "loss": 3.7256,
      "step": 94860
    },
    {
      "epoch": 0.08843419071634422,
      "grad_norm": 0.9955509305000305,
      "learning_rate": 0.00018231336271504954,
      "loss": 3.938,
      "step": 94870
    },
    {
      "epoch": 0.0884435123344233,
      "grad_norm": 1.4510160684585571,
      "learning_rate": 0.00018231149839295437,
      "loss": 3.9177,
      "step": 94880
    },
    {
      "epoch": 0.08845283395250239,
      "grad_norm": 1.1662824153900146,
      "learning_rate": 0.00018230963407085918,
      "loss": 3.6974,
      "step": 94890
    },
    {
      "epoch": 0.08846215557058149,
      "grad_norm": 1.0192780494689941,
      "learning_rate": 0.00018230776974876396,
      "loss": 3.8429,
      "step": 94900
    },
    {
      "epoch": 0.08847147718866058,
      "grad_norm": 1.0425856113433838,
      "learning_rate": 0.00018230590542666876,
      "loss": 3.7456,
      "step": 94910
    },
    {
      "epoch": 0.08848079880673967,
      "grad_norm": 0.9507821798324585,
      "learning_rate": 0.00018230404110457354,
      "loss": 3.9203,
      "step": 94920
    },
    {
      "epoch": 0.08849012042481876,
      "grad_norm": 1.03257155418396,
      "learning_rate": 0.00018230217678247835,
      "loss": 3.74,
      "step": 94930
    },
    {
      "epoch": 0.08849944204289785,
      "grad_norm": 0.9627812504768372,
      "learning_rate": 0.00018230031246038318,
      "loss": 3.8463,
      "step": 94940
    },
    {
      "epoch": 0.08850876366097694,
      "grad_norm": 1.3534293174743652,
      "learning_rate": 0.00018229844813828796,
      "loss": 3.9288,
      "step": 94950
    },
    {
      "epoch": 0.08851808527905604,
      "grad_norm": 1.204068899154663,
      "learning_rate": 0.00018229658381619277,
      "loss": 3.7497,
      "step": 94960
    },
    {
      "epoch": 0.08852740689713513,
      "grad_norm": 1.0593219995498657,
      "learning_rate": 0.00018229471949409755,
      "loss": 3.8029,
      "step": 94970
    },
    {
      "epoch": 0.08853672851521423,
      "grad_norm": 1.0513520240783691,
      "learning_rate": 0.00018229285517200235,
      "loss": 3.5973,
      "step": 94980
    },
    {
      "epoch": 0.08854605013329331,
      "grad_norm": 1.081590175628662,
      "learning_rate": 0.00018229099084990716,
      "loss": 3.7664,
      "step": 94990
    },
    {
      "epoch": 0.0885553717513724,
      "grad_norm": 1.1138639450073242,
      "learning_rate": 0.00018228912652781197,
      "loss": 3.7015,
      "step": 95000
    },
    {
      "epoch": 0.0885646933694515,
      "grad_norm": 0.9627779126167297,
      "learning_rate": 0.00018228726220571677,
      "loss": 3.8393,
      "step": 95010
    },
    {
      "epoch": 0.08857401498753059,
      "grad_norm": 1.163559079170227,
      "learning_rate": 0.00018228539788362155,
      "loss": 3.7026,
      "step": 95020
    },
    {
      "epoch": 0.08858333660560969,
      "grad_norm": 1.030408263206482,
      "learning_rate": 0.00018228353356152636,
      "loss": 3.6932,
      "step": 95030
    },
    {
      "epoch": 0.08859265822368877,
      "grad_norm": 1.0300179719924927,
      "learning_rate": 0.00018228166923943116,
      "loss": 3.789,
      "step": 95040
    },
    {
      "epoch": 0.08860197984176786,
      "grad_norm": 1.098362922668457,
      "learning_rate": 0.00018227980491733597,
      "loss": 3.6418,
      "step": 95050
    },
    {
      "epoch": 0.08861130145984696,
      "grad_norm": 1.119818925857544,
      "learning_rate": 0.00018227794059524078,
      "loss": 3.745,
      "step": 95060
    },
    {
      "epoch": 0.08862062307792605,
      "grad_norm": 1.2068992853164673,
      "learning_rate": 0.00018227607627314556,
      "loss": 3.7655,
      "step": 95070
    },
    {
      "epoch": 0.08862994469600514,
      "grad_norm": 0.9201576709747314,
      "learning_rate": 0.00018227421195105036,
      "loss": 3.7803,
      "step": 95080
    },
    {
      "epoch": 0.08863926631408423,
      "grad_norm": 0.8370517492294312,
      "learning_rate": 0.00018227234762895517,
      "loss": 3.6664,
      "step": 95090
    },
    {
      "epoch": 0.08864858793216332,
      "grad_norm": 1.0860437154769897,
      "learning_rate": 0.00018227048330685997,
      "loss": 3.6041,
      "step": 95100
    },
    {
      "epoch": 0.08865790955024241,
      "grad_norm": 1.136828064918518,
      "learning_rate": 0.00018226861898476478,
      "loss": 3.9269,
      "step": 95110
    },
    {
      "epoch": 0.08866723116832151,
      "grad_norm": 1.112484097480774,
      "learning_rate": 0.00018226675466266956,
      "loss": 3.8205,
      "step": 95120
    },
    {
      "epoch": 0.0886765527864006,
      "grad_norm": 1.208756446838379,
      "learning_rate": 0.00018226489034057437,
      "loss": 3.5251,
      "step": 95130
    },
    {
      "epoch": 0.08868587440447968,
      "grad_norm": 1.022292137145996,
      "learning_rate": 0.00018226302601847917,
      "loss": 3.5527,
      "step": 95140
    },
    {
      "epoch": 0.08869519602255878,
      "grad_norm": 1.028185248374939,
      "learning_rate": 0.00018226116169638395,
      "loss": 3.9476,
      "step": 95150
    },
    {
      "epoch": 0.08870451764063787,
      "grad_norm": 1.1589648723602295,
      "learning_rate": 0.00018225929737428879,
      "loss": 3.6645,
      "step": 95160
    },
    {
      "epoch": 0.08871383925871697,
      "grad_norm": 1.2049349546432495,
      "learning_rate": 0.00018225743305219356,
      "loss": 3.7606,
      "step": 95170
    },
    {
      "epoch": 0.08872316087679606,
      "grad_norm": 1.0444002151489258,
      "learning_rate": 0.00018225556873009837,
      "loss": 3.7583,
      "step": 95180
    },
    {
      "epoch": 0.08873248249487514,
      "grad_norm": 1.2404813766479492,
      "learning_rate": 0.00018225370440800318,
      "loss": 3.7331,
      "step": 95190
    },
    {
      "epoch": 0.08874180411295424,
      "grad_norm": 1.1407750844955444,
      "learning_rate": 0.00018225184008590796,
      "loss": 3.813,
      "step": 95200
    },
    {
      "epoch": 0.08875112573103333,
      "grad_norm": 0.9695561528205872,
      "learning_rate": 0.00018224997576381276,
      "loss": 3.7116,
      "step": 95210
    },
    {
      "epoch": 0.08876044734911243,
      "grad_norm": 0.921600878238678,
      "learning_rate": 0.00018224811144171757,
      "loss": 3.571,
      "step": 95220
    },
    {
      "epoch": 0.08876976896719152,
      "grad_norm": 0.8687877058982849,
      "learning_rate": 0.00018224624711962238,
      "loss": 3.8775,
      "step": 95230
    },
    {
      "epoch": 0.08877909058527061,
      "grad_norm": 1.282437801361084,
      "learning_rate": 0.00018224438279752718,
      "loss": 3.7186,
      "step": 95240
    },
    {
      "epoch": 0.0887884122033497,
      "grad_norm": 1.2591285705566406,
      "learning_rate": 0.00018224251847543196,
      "loss": 3.5567,
      "step": 95250
    },
    {
      "epoch": 0.08879773382142879,
      "grad_norm": 1.163934350013733,
      "learning_rate": 0.00018224065415333677,
      "loss": 3.7747,
      "step": 95260
    },
    {
      "epoch": 0.08880705543950788,
      "grad_norm": 0.9142125248908997,
      "learning_rate": 0.00018223878983124157,
      "loss": 3.5838,
      "step": 95270
    },
    {
      "epoch": 0.08881637705758698,
      "grad_norm": 1.092950701713562,
      "learning_rate": 0.00018223692550914638,
      "loss": 3.7401,
      "step": 95280
    },
    {
      "epoch": 0.08882569867566607,
      "grad_norm": 1.2490735054016113,
      "learning_rate": 0.00018223506118705119,
      "loss": 3.8059,
      "step": 95290
    },
    {
      "epoch": 0.08883502029374515,
      "grad_norm": 0.9137051105499268,
      "learning_rate": 0.00018223319686495597,
      "loss": 3.8832,
      "step": 95300
    },
    {
      "epoch": 0.08884434191182425,
      "grad_norm": 1.0453314781188965,
      "learning_rate": 0.00018223133254286077,
      "loss": 3.8137,
      "step": 95310
    },
    {
      "epoch": 0.08885366352990334,
      "grad_norm": 1.1885401010513306,
      "learning_rate": 0.00018222946822076555,
      "loss": 3.8832,
      "step": 95320
    },
    {
      "epoch": 0.08886298514798244,
      "grad_norm": 0.9556514620780945,
      "learning_rate": 0.00018222760389867038,
      "loss": 3.9309,
      "step": 95330
    },
    {
      "epoch": 0.08887230676606153,
      "grad_norm": 0.8926401734352112,
      "learning_rate": 0.0001822257395765752,
      "loss": 3.6807,
      "step": 95340
    },
    {
      "epoch": 0.08888162838414061,
      "grad_norm": 1.1589224338531494,
      "learning_rate": 0.00018222387525447997,
      "loss": 3.9573,
      "step": 95350
    },
    {
      "epoch": 0.0888909500022197,
      "grad_norm": 1.0940086841583252,
      "learning_rate": 0.00018222201093238478,
      "loss": 3.4453,
      "step": 95360
    },
    {
      "epoch": 0.0889002716202988,
      "grad_norm": 1.1364418268203735,
      "learning_rate": 0.00018222014661028956,
      "loss": 3.7848,
      "step": 95370
    },
    {
      "epoch": 0.0889095932383779,
      "grad_norm": 0.9214072227478027,
      "learning_rate": 0.00018221828228819436,
      "loss": 3.5666,
      "step": 95380
    },
    {
      "epoch": 0.08891891485645699,
      "grad_norm": 1.0102869272232056,
      "learning_rate": 0.0001822164179660992,
      "loss": 3.8538,
      "step": 95390
    },
    {
      "epoch": 0.08892823647453607,
      "grad_norm": 1.033190369606018,
      "learning_rate": 0.00018221455364400397,
      "loss": 3.7453,
      "step": 95400
    },
    {
      "epoch": 0.08893755809261517,
      "grad_norm": 1.01793372631073,
      "learning_rate": 0.00018221268932190878,
      "loss": 3.8277,
      "step": 95410
    },
    {
      "epoch": 0.08894687971069426,
      "grad_norm": 1.0034652948379517,
      "learning_rate": 0.00018221082499981356,
      "loss": 3.7166,
      "step": 95420
    },
    {
      "epoch": 0.08895620132877335,
      "grad_norm": 1.0000063180923462,
      "learning_rate": 0.00018220896067771837,
      "loss": 3.8155,
      "step": 95430
    },
    {
      "epoch": 0.08896552294685245,
      "grad_norm": 1.101114273071289,
      "learning_rate": 0.00018220709635562317,
      "loss": 3.8482,
      "step": 95440
    },
    {
      "epoch": 0.08897484456493154,
      "grad_norm": 1.1372005939483643,
      "learning_rate": 0.00018220523203352798,
      "loss": 3.6833,
      "step": 95450
    },
    {
      "epoch": 0.08898416618301062,
      "grad_norm": 1.0948591232299805,
      "learning_rate": 0.00018220336771143279,
      "loss": 3.7921,
      "step": 95460
    },
    {
      "epoch": 0.08899348780108972,
      "grad_norm": 1.0928105115890503,
      "learning_rate": 0.00018220150338933756,
      "loss": 3.686,
      "step": 95470
    },
    {
      "epoch": 0.08900280941916881,
      "grad_norm": 0.8711956739425659,
      "learning_rate": 0.00018219963906724237,
      "loss": 3.7313,
      "step": 95480
    },
    {
      "epoch": 0.08901213103724791,
      "grad_norm": 1.0670709609985352,
      "learning_rate": 0.00018219777474514718,
      "loss": 3.7174,
      "step": 95490
    },
    {
      "epoch": 0.089021452655327,
      "grad_norm": 1.0518829822540283,
      "learning_rate": 0.00018219591042305198,
      "loss": 3.7779,
      "step": 95500
    },
    {
      "epoch": 0.08903077427340608,
      "grad_norm": 1.147652268409729,
      "learning_rate": 0.0001821940461009568,
      "loss": 3.726,
      "step": 95510
    },
    {
      "epoch": 0.08904009589148518,
      "grad_norm": 1.215158462524414,
      "learning_rate": 0.00018219218177886157,
      "loss": 3.8333,
      "step": 95520
    },
    {
      "epoch": 0.08904941750956427,
      "grad_norm": 1.124511480331421,
      "learning_rate": 0.00018219031745676638,
      "loss": 3.78,
      "step": 95530
    },
    {
      "epoch": 0.08905873912764337,
      "grad_norm": 1.2555527687072754,
      "learning_rate": 0.00018218845313467118,
      "loss": 3.9094,
      "step": 95540
    },
    {
      "epoch": 0.08906806074572246,
      "grad_norm": 1.10213303565979,
      "learning_rate": 0.000182186588812576,
      "loss": 3.6232,
      "step": 95550
    },
    {
      "epoch": 0.08907738236380154,
      "grad_norm": 1.3065553903579712,
      "learning_rate": 0.0001821847244904808,
      "loss": 3.9245,
      "step": 95560
    },
    {
      "epoch": 0.08908670398188064,
      "grad_norm": 1.0205854177474976,
      "learning_rate": 0.00018218286016838557,
      "loss": 3.8244,
      "step": 95570
    },
    {
      "epoch": 0.08909602559995973,
      "grad_norm": 1.035840630531311,
      "learning_rate": 0.00018218099584629038,
      "loss": 3.9133,
      "step": 95580
    },
    {
      "epoch": 0.08910534721803882,
      "grad_norm": 1.0472699403762817,
      "learning_rate": 0.00018217913152419519,
      "loss": 3.7845,
      "step": 95590
    },
    {
      "epoch": 0.08911466883611792,
      "grad_norm": 0.9839484095573425,
      "learning_rate": 0.00018217726720209997,
      "loss": 3.747,
      "step": 95600
    },
    {
      "epoch": 0.089123990454197,
      "grad_norm": 0.9965683221817017,
      "learning_rate": 0.0001821754028800048,
      "loss": 3.8047,
      "step": 95610
    },
    {
      "epoch": 0.0891333120722761,
      "grad_norm": 1.0886679887771606,
      "learning_rate": 0.00018217353855790958,
      "loss": 3.7499,
      "step": 95620
    },
    {
      "epoch": 0.08914263369035519,
      "grad_norm": 0.8072439432144165,
      "learning_rate": 0.00018217167423581438,
      "loss": 3.7747,
      "step": 95630
    },
    {
      "epoch": 0.08915195530843428,
      "grad_norm": 0.975534200668335,
      "learning_rate": 0.0001821698099137192,
      "loss": 3.6114,
      "step": 95640
    },
    {
      "epoch": 0.08916127692651338,
      "grad_norm": 1.2545924186706543,
      "learning_rate": 0.00018216794559162397,
      "loss": 3.8062,
      "step": 95650
    },
    {
      "epoch": 0.08917059854459247,
      "grad_norm": 1.1136678457260132,
      "learning_rate": 0.00018216608126952878,
      "loss": 3.6072,
      "step": 95660
    },
    {
      "epoch": 0.08917992016267155,
      "grad_norm": 1.1722363233566284,
      "learning_rate": 0.00018216421694743358,
      "loss": 3.7638,
      "step": 95670
    },
    {
      "epoch": 0.08918924178075065,
      "grad_norm": 1.1143317222595215,
      "learning_rate": 0.0001821623526253384,
      "loss": 3.7539,
      "step": 95680
    },
    {
      "epoch": 0.08919856339882974,
      "grad_norm": 1.1052159070968628,
      "learning_rate": 0.0001821604883032432,
      "loss": 3.701,
      "step": 95690
    },
    {
      "epoch": 0.08920788501690884,
      "grad_norm": 1.1790086030960083,
      "learning_rate": 0.00018215862398114797,
      "loss": 3.8735,
      "step": 95700
    },
    {
      "epoch": 0.08921720663498793,
      "grad_norm": 1.0310972929000854,
      "learning_rate": 0.00018215675965905278,
      "loss": 3.6198,
      "step": 95710
    },
    {
      "epoch": 0.08922652825306701,
      "grad_norm": 0.9571806192398071,
      "learning_rate": 0.0001821548953369576,
      "loss": 3.7417,
      "step": 95720
    },
    {
      "epoch": 0.0892358498711461,
      "grad_norm": 1.0501275062561035,
      "learning_rate": 0.0001821530310148624,
      "loss": 3.9177,
      "step": 95730
    },
    {
      "epoch": 0.0892451714892252,
      "grad_norm": 1.2948410511016846,
      "learning_rate": 0.0001821511666927672,
      "loss": 3.6463,
      "step": 95740
    },
    {
      "epoch": 0.0892544931073043,
      "grad_norm": 0.9871208071708679,
      "learning_rate": 0.00018214930237067198,
      "loss": 3.7378,
      "step": 95750
    },
    {
      "epoch": 0.08926381472538339,
      "grad_norm": 0.9923432469367981,
      "learning_rate": 0.00018214743804857678,
      "loss": 3.8852,
      "step": 95760
    },
    {
      "epoch": 0.08927313634346247,
      "grad_norm": 1.036745309829712,
      "learning_rate": 0.0001821455737264816,
      "loss": 3.8345,
      "step": 95770
    },
    {
      "epoch": 0.08928245796154156,
      "grad_norm": 1.1357322931289673,
      "learning_rate": 0.0001821437094043864,
      "loss": 3.6959,
      "step": 95780
    },
    {
      "epoch": 0.08929177957962066,
      "grad_norm": 1.1385345458984375,
      "learning_rate": 0.0001821418450822912,
      "loss": 3.5548,
      "step": 95790
    },
    {
      "epoch": 0.08930110119769975,
      "grad_norm": 1.0236985683441162,
      "learning_rate": 0.00018213998076019598,
      "loss": 3.5617,
      "step": 95800
    },
    {
      "epoch": 0.08931042281577885,
      "grad_norm": 0.9095249772071838,
      "learning_rate": 0.0001821381164381008,
      "loss": 3.848,
      "step": 95810
    },
    {
      "epoch": 0.08931974443385793,
      "grad_norm": 1.1324381828308105,
      "learning_rate": 0.00018213625211600557,
      "loss": 3.8061,
      "step": 95820
    },
    {
      "epoch": 0.08932906605193702,
      "grad_norm": 0.8694828748703003,
      "learning_rate": 0.00018213438779391037,
      "loss": 3.8598,
      "step": 95830
    },
    {
      "epoch": 0.08933838767001612,
      "grad_norm": 1.2458330392837524,
      "learning_rate": 0.0001821325234718152,
      "loss": 3.7666,
      "step": 95840
    },
    {
      "epoch": 0.08934770928809521,
      "grad_norm": 1.0515027046203613,
      "learning_rate": 0.00018213065914972,
      "loss": 3.615,
      "step": 95850
    },
    {
      "epoch": 0.0893570309061743,
      "grad_norm": 1.000709891319275,
      "learning_rate": 0.0001821287948276248,
      "loss": 3.8137,
      "step": 95860
    },
    {
      "epoch": 0.08936635252425339,
      "grad_norm": 1.165708065032959,
      "learning_rate": 0.00018212693050552957,
      "loss": 3.5527,
      "step": 95870
    },
    {
      "epoch": 0.08937567414233248,
      "grad_norm": 1.114862322807312,
      "learning_rate": 0.00018212506618343438,
      "loss": 3.7318,
      "step": 95880
    },
    {
      "epoch": 0.08938499576041158,
      "grad_norm": 1.168062448501587,
      "learning_rate": 0.00018212320186133919,
      "loss": 3.6647,
      "step": 95890
    },
    {
      "epoch": 0.08939431737849067,
      "grad_norm": 1.0593785047531128,
      "learning_rate": 0.000182121337539244,
      "loss": 3.8346,
      "step": 95900
    },
    {
      "epoch": 0.08940363899656976,
      "grad_norm": 1.0687198638916016,
      "learning_rate": 0.0001821194732171488,
      "loss": 3.4016,
      "step": 95910
    },
    {
      "epoch": 0.08941296061464886,
      "grad_norm": 1.1161705255508423,
      "learning_rate": 0.00018211760889505358,
      "loss": 3.7178,
      "step": 95920
    },
    {
      "epoch": 0.08942228223272794,
      "grad_norm": 0.8794777989387512,
      "learning_rate": 0.00018211574457295838,
      "loss": 3.8092,
      "step": 95930
    },
    {
      "epoch": 0.08943160385080703,
      "grad_norm": 1.1467233896255493,
      "learning_rate": 0.0001821138802508632,
      "loss": 3.7942,
      "step": 95940
    },
    {
      "epoch": 0.08944092546888613,
      "grad_norm": 1.1736981868743896,
      "learning_rate": 0.000182112015928768,
      "loss": 3.6587,
      "step": 95950
    },
    {
      "epoch": 0.08945024708696522,
      "grad_norm": 1.3254305124282837,
      "learning_rate": 0.0001821101516066728,
      "loss": 3.4698,
      "step": 95960
    },
    {
      "epoch": 0.08945956870504432,
      "grad_norm": 0.9910618662834167,
      "learning_rate": 0.00018210828728457758,
      "loss": 3.8841,
      "step": 95970
    },
    {
      "epoch": 0.0894688903231234,
      "grad_norm": 1.104404330253601,
      "learning_rate": 0.0001821064229624824,
      "loss": 3.9496,
      "step": 95980
    },
    {
      "epoch": 0.08947821194120249,
      "grad_norm": 1.1189924478530884,
      "learning_rate": 0.0001821045586403872,
      "loss": 3.8363,
      "step": 95990
    },
    {
      "epoch": 0.08948753355928159,
      "grad_norm": 1.0675407648086548,
      "learning_rate": 0.000182102694318292,
      "loss": 3.7192,
      "step": 96000
    },
    {
      "epoch": 0.08949685517736068,
      "grad_norm": 0.9086053371429443,
      "learning_rate": 0.0001821008299961968,
      "loss": 3.8364,
      "step": 96010
    },
    {
      "epoch": 0.08950617679543978,
      "grad_norm": 0.9007850885391235,
      "learning_rate": 0.00018209896567410159,
      "loss": 3.7565,
      "step": 96020
    },
    {
      "epoch": 0.08951549841351886,
      "grad_norm": 0.9753954410552979,
      "learning_rate": 0.0001820971013520064,
      "loss": 3.8403,
      "step": 96030
    },
    {
      "epoch": 0.08952482003159795,
      "grad_norm": 1.0864754915237427,
      "learning_rate": 0.0001820952370299112,
      "loss": 3.7321,
      "step": 96040
    },
    {
      "epoch": 0.08953414164967705,
      "grad_norm": 1.0426992177963257,
      "learning_rate": 0.00018209337270781598,
      "loss": 3.5689,
      "step": 96050
    },
    {
      "epoch": 0.08954346326775614,
      "grad_norm": 1.1114693880081177,
      "learning_rate": 0.0001820915083857208,
      "loss": 3.7042,
      "step": 96060
    },
    {
      "epoch": 0.08955278488583523,
      "grad_norm": 1.130259394645691,
      "learning_rate": 0.0001820896440636256,
      "loss": 3.6333,
      "step": 96070
    },
    {
      "epoch": 0.08956210650391432,
      "grad_norm": 1.0480406284332275,
      "learning_rate": 0.0001820877797415304,
      "loss": 3.6124,
      "step": 96080
    },
    {
      "epoch": 0.08957142812199341,
      "grad_norm": 1.1697579622268677,
      "learning_rate": 0.0001820859154194352,
      "loss": 3.8257,
      "step": 96090
    },
    {
      "epoch": 0.0895807497400725,
      "grad_norm": 0.8057948350906372,
      "learning_rate": 0.00018208405109733998,
      "loss": 3.8447,
      "step": 96100
    },
    {
      "epoch": 0.0895900713581516,
      "grad_norm": 0.8251792192459106,
      "learning_rate": 0.0001820821867752448,
      "loss": 3.8991,
      "step": 96110
    },
    {
      "epoch": 0.0895993929762307,
      "grad_norm": 1.0776712894439697,
      "learning_rate": 0.0001820803224531496,
      "loss": 4.0038,
      "step": 96120
    },
    {
      "epoch": 0.08960871459430979,
      "grad_norm": 1.2024900913238525,
      "learning_rate": 0.0001820784581310544,
      "loss": 3.8728,
      "step": 96130
    },
    {
      "epoch": 0.08961803621238887,
      "grad_norm": 1.0502688884735107,
      "learning_rate": 0.0001820765938089592,
      "loss": 3.8212,
      "step": 96140
    },
    {
      "epoch": 0.08962735783046796,
      "grad_norm": 1.0467920303344727,
      "learning_rate": 0.000182074729486864,
      "loss": 3.8805,
      "step": 96150
    },
    {
      "epoch": 0.08963667944854706,
      "grad_norm": 1.113383412361145,
      "learning_rate": 0.0001820728651647688,
      "loss": 3.6645,
      "step": 96160
    },
    {
      "epoch": 0.08964600106662615,
      "grad_norm": 0.9607263207435608,
      "learning_rate": 0.0001820710008426736,
      "loss": 3.7937,
      "step": 96170
    },
    {
      "epoch": 0.08965532268470525,
      "grad_norm": 1.0386193990707397,
      "learning_rate": 0.0001820691365205784,
      "loss": 3.7419,
      "step": 96180
    },
    {
      "epoch": 0.08966464430278433,
      "grad_norm": 1.138810634613037,
      "learning_rate": 0.00018206727219848319,
      "loss": 3.8265,
      "step": 96190
    },
    {
      "epoch": 0.08967396592086342,
      "grad_norm": 0.8994926810264587,
      "learning_rate": 0.000182065407876388,
      "loss": 3.7895,
      "step": 96200
    },
    {
      "epoch": 0.08968328753894252,
      "grad_norm": 1.1556099653244019,
      "learning_rate": 0.0001820635435542928,
      "loss": 3.7564,
      "step": 96210
    },
    {
      "epoch": 0.08969260915702161,
      "grad_norm": 1.0433648824691772,
      "learning_rate": 0.0001820616792321976,
      "loss": 3.8047,
      "step": 96220
    },
    {
      "epoch": 0.0897019307751007,
      "grad_norm": 0.8715464472770691,
      "learning_rate": 0.0001820598149101024,
      "loss": 3.8178,
      "step": 96230
    },
    {
      "epoch": 0.08971125239317979,
      "grad_norm": 1.076880693435669,
      "learning_rate": 0.0001820579505880072,
      "loss": 3.7077,
      "step": 96240
    },
    {
      "epoch": 0.08972057401125888,
      "grad_norm": 1.056267499923706,
      "learning_rate": 0.000182056086265912,
      "loss": 3.4455,
      "step": 96250
    },
    {
      "epoch": 0.08972989562933797,
      "grad_norm": 1.0336313247680664,
      "learning_rate": 0.0001820542219438168,
      "loss": 3.7297,
      "step": 96260
    },
    {
      "epoch": 0.08973921724741707,
      "grad_norm": 0.9182226657867432,
      "learning_rate": 0.00018205235762172158,
      "loss": 3.6971,
      "step": 96270
    },
    {
      "epoch": 0.08974853886549616,
      "grad_norm": 1.2137073278427124,
      "learning_rate": 0.00018205049329962641,
      "loss": 3.8299,
      "step": 96280
    },
    {
      "epoch": 0.08975786048357524,
      "grad_norm": 1.0233862400054932,
      "learning_rate": 0.0001820486289775312,
      "loss": 3.7134,
      "step": 96290
    },
    {
      "epoch": 0.08976718210165434,
      "grad_norm": 1.0856752395629883,
      "learning_rate": 0.000182046764655436,
      "loss": 3.8292,
      "step": 96300
    },
    {
      "epoch": 0.08977650371973343,
      "grad_norm": 1.1224236488342285,
      "learning_rate": 0.0001820449003333408,
      "loss": 3.7258,
      "step": 96310
    },
    {
      "epoch": 0.08978582533781253,
      "grad_norm": 0.9203186631202698,
      "learning_rate": 0.00018204303601124559,
      "loss": 3.6513,
      "step": 96320
    },
    {
      "epoch": 0.08979514695589162,
      "grad_norm": 1.1707518100738525,
      "learning_rate": 0.0001820411716891504,
      "loss": 3.735,
      "step": 96330
    },
    {
      "epoch": 0.08980446857397072,
      "grad_norm": 0.8529047966003418,
      "learning_rate": 0.0001820393073670552,
      "loss": 3.6226,
      "step": 96340
    },
    {
      "epoch": 0.0898137901920498,
      "grad_norm": 1.0415992736816406,
      "learning_rate": 0.00018203744304496,
      "loss": 3.9428,
      "step": 96350
    },
    {
      "epoch": 0.08982311181012889,
      "grad_norm": 0.9939316511154175,
      "learning_rate": 0.0001820355787228648,
      "loss": 3.731,
      "step": 96360
    },
    {
      "epoch": 0.08983243342820799,
      "grad_norm": 1.0614707469940186,
      "learning_rate": 0.0001820337144007696,
      "loss": 3.7982,
      "step": 96370
    },
    {
      "epoch": 0.08984175504628708,
      "grad_norm": 1.0722107887268066,
      "learning_rate": 0.0001820318500786744,
      "loss": 3.7704,
      "step": 96380
    },
    {
      "epoch": 0.08985107666436618,
      "grad_norm": 1.1247183084487915,
      "learning_rate": 0.0001820299857565792,
      "loss": 3.8228,
      "step": 96390
    },
    {
      "epoch": 0.08986039828244526,
      "grad_norm": 1.0011451244354248,
      "learning_rate": 0.000182028121434484,
      "loss": 3.658,
      "step": 96400
    },
    {
      "epoch": 0.08986971990052435,
      "grad_norm": 0.9595437049865723,
      "learning_rate": 0.00018202625711238882,
      "loss": 3.7632,
      "step": 96410
    },
    {
      "epoch": 0.08987904151860344,
      "grad_norm": 1.0073682069778442,
      "learning_rate": 0.0001820243927902936,
      "loss": 3.6932,
      "step": 96420
    },
    {
      "epoch": 0.08988836313668254,
      "grad_norm": 1.1504261493682861,
      "learning_rate": 0.0001820225284681984,
      "loss": 3.6523,
      "step": 96430
    },
    {
      "epoch": 0.08989768475476163,
      "grad_norm": 1.100915551185608,
      "learning_rate": 0.00018202066414610318,
      "loss": 3.757,
      "step": 96440
    },
    {
      "epoch": 0.08990700637284071,
      "grad_norm": 1.0899481773376465,
      "learning_rate": 0.000182018799824008,
      "loss": 3.6302,
      "step": 96450
    },
    {
      "epoch": 0.08991632799091981,
      "grad_norm": 0.9828557372093201,
      "learning_rate": 0.00018201693550191282,
      "loss": 4.0152,
      "step": 96460
    },
    {
      "epoch": 0.0899256496089989,
      "grad_norm": 1.0040473937988281,
      "learning_rate": 0.0001820150711798176,
      "loss": 3.7586,
      "step": 96470
    },
    {
      "epoch": 0.089934971227078,
      "grad_norm": 1.0892332792282104,
      "learning_rate": 0.0001820132068577224,
      "loss": 3.7333,
      "step": 96480
    },
    {
      "epoch": 0.08994429284515709,
      "grad_norm": 1.0232083797454834,
      "learning_rate": 0.00018201134253562718,
      "loss": 3.6195,
      "step": 96490
    },
    {
      "epoch": 0.08995361446323617,
      "grad_norm": 1.1242674589157104,
      "learning_rate": 0.000182009478213532,
      "loss": 3.6657,
      "step": 96500
    },
    {
      "epoch": 0.08996293608131527,
      "grad_norm": 0.9409686326980591,
      "learning_rate": 0.00018200761389143682,
      "loss": 3.5591,
      "step": 96510
    },
    {
      "epoch": 0.08997225769939436,
      "grad_norm": 1.0487595796585083,
      "learning_rate": 0.0001820057495693416,
      "loss": 3.7501,
      "step": 96520
    },
    {
      "epoch": 0.08998157931747346,
      "grad_norm": 1.0925061702728271,
      "learning_rate": 0.0001820038852472464,
      "loss": 3.9234,
      "step": 96530
    },
    {
      "epoch": 0.08999090093555255,
      "grad_norm": 1.060221552848816,
      "learning_rate": 0.0001820020209251512,
      "loss": 3.8313,
      "step": 96540
    },
    {
      "epoch": 0.09000022255363165,
      "grad_norm": 1.037662148475647,
      "learning_rate": 0.000182000156603056,
      "loss": 3.6963,
      "step": 96550
    },
    {
      "epoch": 0.09000954417171073,
      "grad_norm": 1.021791934967041,
      "learning_rate": 0.0001819982922809608,
      "loss": 3.479,
      "step": 96560
    },
    {
      "epoch": 0.09001886578978982,
      "grad_norm": 1.149575114250183,
      "learning_rate": 0.0001819964279588656,
      "loss": 3.6081,
      "step": 96570
    },
    {
      "epoch": 0.09002818740786891,
      "grad_norm": 1.2030640840530396,
      "learning_rate": 0.00018199456363677041,
      "loss": 3.9529,
      "step": 96580
    },
    {
      "epoch": 0.09003750902594801,
      "grad_norm": 0.9132727980613708,
      "learning_rate": 0.0001819926993146752,
      "loss": 3.5335,
      "step": 96590
    },
    {
      "epoch": 0.0900468306440271,
      "grad_norm": 0.9586113691329956,
      "learning_rate": 0.00018199083499258,
      "loss": 3.6367,
      "step": 96600
    },
    {
      "epoch": 0.09005615226210618,
      "grad_norm": 0.9654227495193481,
      "learning_rate": 0.0001819889706704848,
      "loss": 3.9096,
      "step": 96610
    },
    {
      "epoch": 0.09006547388018528,
      "grad_norm": 1.1916311979293823,
      "learning_rate": 0.0001819871063483896,
      "loss": 3.6683,
      "step": 96620
    },
    {
      "epoch": 0.09007479549826437,
      "grad_norm": 0.9727156162261963,
      "learning_rate": 0.00018198524202629442,
      "loss": 3.7369,
      "step": 96630
    },
    {
      "epoch": 0.09008411711634347,
      "grad_norm": 1.0409913063049316,
      "learning_rate": 0.0001819833777041992,
      "loss": 3.7892,
      "step": 96640
    },
    {
      "epoch": 0.09009343873442256,
      "grad_norm": 1.1333049535751343,
      "learning_rate": 0.000181981513382104,
      "loss": 3.7605,
      "step": 96650
    },
    {
      "epoch": 0.09010276035250164,
      "grad_norm": 0.9258301854133606,
      "learning_rate": 0.0001819796490600088,
      "loss": 3.7839,
      "step": 96660
    },
    {
      "epoch": 0.09011208197058074,
      "grad_norm": 1.010486364364624,
      "learning_rate": 0.00018197778473791362,
      "loss": 3.9244,
      "step": 96670
    },
    {
      "epoch": 0.09012140358865983,
      "grad_norm": 1.1061307191848755,
      "learning_rate": 0.00018197592041581842,
      "loss": 3.6625,
      "step": 96680
    },
    {
      "epoch": 0.09013072520673893,
      "grad_norm": 1.1182944774627686,
      "learning_rate": 0.0001819740560937232,
      "loss": 3.7071,
      "step": 96690
    },
    {
      "epoch": 0.09014004682481802,
      "grad_norm": 1.084045171737671,
      "learning_rate": 0.000181972191771628,
      "loss": 3.761,
      "step": 96700
    },
    {
      "epoch": 0.0901493684428971,
      "grad_norm": 1.1279329061508179,
      "learning_rate": 0.00018197032744953281,
      "loss": 3.5972,
      "step": 96710
    },
    {
      "epoch": 0.0901586900609762,
      "grad_norm": 1.2137950658798218,
      "learning_rate": 0.0001819684631274376,
      "loss": 3.7047,
      "step": 96720
    },
    {
      "epoch": 0.09016801167905529,
      "grad_norm": 1.1225221157073975,
      "learning_rate": 0.00018196659880534243,
      "loss": 3.8602,
      "step": 96730
    },
    {
      "epoch": 0.09017733329713438,
      "grad_norm": 1.1314927339553833,
      "learning_rate": 0.0001819647344832472,
      "loss": 3.8139,
      "step": 96740
    },
    {
      "epoch": 0.09018665491521348,
      "grad_norm": 1.134422779083252,
      "learning_rate": 0.000181962870161152,
      "loss": 3.7379,
      "step": 96750
    },
    {
      "epoch": 0.09019597653329256,
      "grad_norm": 1.136172890663147,
      "learning_rate": 0.00018196100583905682,
      "loss": 3.6279,
      "step": 96760
    },
    {
      "epoch": 0.09020529815137165,
      "grad_norm": 0.9730674028396606,
      "learning_rate": 0.0001819591415169616,
      "loss": 3.6762,
      "step": 96770
    },
    {
      "epoch": 0.09021461976945075,
      "grad_norm": 1.1254628896713257,
      "learning_rate": 0.0001819572771948664,
      "loss": 3.7773,
      "step": 96780
    },
    {
      "epoch": 0.09022394138752984,
      "grad_norm": 1.1272146701812744,
      "learning_rate": 0.0001819554128727712,
      "loss": 3.6235,
      "step": 96790
    },
    {
      "epoch": 0.09023326300560894,
      "grad_norm": 0.9708561301231384,
      "learning_rate": 0.00018195354855067602,
      "loss": 3.677,
      "step": 96800
    },
    {
      "epoch": 0.09024258462368803,
      "grad_norm": 1.0407941341400146,
      "learning_rate": 0.00018195168422858082,
      "loss": 3.834,
      "step": 96810
    },
    {
      "epoch": 0.09025190624176711,
      "grad_norm": 1.0628775358200073,
      "learning_rate": 0.0001819498199064856,
      "loss": 3.8119,
      "step": 96820
    },
    {
      "epoch": 0.09026122785984621,
      "grad_norm": 0.8558712601661682,
      "learning_rate": 0.0001819479555843904,
      "loss": 3.6995,
      "step": 96830
    },
    {
      "epoch": 0.0902705494779253,
      "grad_norm": 1.1370400190353394,
      "learning_rate": 0.00018194609126229522,
      "loss": 3.6395,
      "step": 96840
    },
    {
      "epoch": 0.0902798710960044,
      "grad_norm": 1.4216971397399902,
      "learning_rate": 0.00018194422694020002,
      "loss": 3.5525,
      "step": 96850
    },
    {
      "epoch": 0.09028919271408349,
      "grad_norm": 0.9660159945487976,
      "learning_rate": 0.00018194236261810483,
      "loss": 3.7752,
      "step": 96860
    },
    {
      "epoch": 0.09029851433216257,
      "grad_norm": 1.0456576347351074,
      "learning_rate": 0.0001819404982960096,
      "loss": 3.7579,
      "step": 96870
    },
    {
      "epoch": 0.09030783595024167,
      "grad_norm": 0.9062406420707703,
      "learning_rate": 0.00018193863397391441,
      "loss": 3.7413,
      "step": 96880
    },
    {
      "epoch": 0.09031715756832076,
      "grad_norm": 1.0139306783676147,
      "learning_rate": 0.0001819367696518192,
      "loss": 3.4944,
      "step": 96890
    },
    {
      "epoch": 0.09032647918639986,
      "grad_norm": 1.073407769203186,
      "learning_rate": 0.00018193490532972403,
      "loss": 3.4255,
      "step": 96900
    },
    {
      "epoch": 0.09033580080447895,
      "grad_norm": 1.0420715808868408,
      "learning_rate": 0.00018193304100762883,
      "loss": 3.8569,
      "step": 96910
    },
    {
      "epoch": 0.09034512242255803,
      "grad_norm": 1.03793203830719,
      "learning_rate": 0.0001819311766855336,
      "loss": 3.7858,
      "step": 96920
    },
    {
      "epoch": 0.09035444404063712,
      "grad_norm": 1.1302138566970825,
      "learning_rate": 0.00018192931236343842,
      "loss": 3.7365,
      "step": 96930
    },
    {
      "epoch": 0.09036376565871622,
      "grad_norm": 1.048034906387329,
      "learning_rate": 0.0001819274480413432,
      "loss": 3.6965,
      "step": 96940
    },
    {
      "epoch": 0.09037308727679531,
      "grad_norm": 0.9091600775718689,
      "learning_rate": 0.000181925583719248,
      "loss": 3.9149,
      "step": 96950
    },
    {
      "epoch": 0.09038240889487441,
      "grad_norm": 1.042330265045166,
      "learning_rate": 0.00018192371939715284,
      "loss": 3.7628,
      "step": 96960
    },
    {
      "epoch": 0.09039173051295349,
      "grad_norm": 1.0620394945144653,
      "learning_rate": 0.00018192185507505762,
      "loss": 3.6435,
      "step": 96970
    },
    {
      "epoch": 0.09040105213103258,
      "grad_norm": 1.043370246887207,
      "learning_rate": 0.00018191999075296242,
      "loss": 3.7414,
      "step": 96980
    },
    {
      "epoch": 0.09041037374911168,
      "grad_norm": 1.2433431148529053,
      "learning_rate": 0.0001819181264308672,
      "loss": 3.6097,
      "step": 96990
    },
    {
      "epoch": 0.09041969536719077,
      "grad_norm": 1.0034077167510986,
      "learning_rate": 0.000181916262108772,
      "loss": 3.7073,
      "step": 97000
    },
    {
      "epoch": 0.09042901698526987,
      "grad_norm": 1.018541932106018,
      "learning_rate": 0.00018191439778667681,
      "loss": 3.5328,
      "step": 97010
    },
    {
      "epoch": 0.09043833860334896,
      "grad_norm": 1.2295211553573608,
      "learning_rate": 0.00018191253346458162,
      "loss": 3.8377,
      "step": 97020
    },
    {
      "epoch": 0.09044766022142804,
      "grad_norm": 0.9149842858314514,
      "learning_rate": 0.00018191066914248643,
      "loss": 3.7591,
      "step": 97030
    },
    {
      "epoch": 0.09045698183950714,
      "grad_norm": 1.1457501649856567,
      "learning_rate": 0.0001819088048203912,
      "loss": 3.7897,
      "step": 97040
    },
    {
      "epoch": 0.09046630345758623,
      "grad_norm": 1.1370770931243896,
      "learning_rate": 0.000181906940498296,
      "loss": 3.75,
      "step": 97050
    },
    {
      "epoch": 0.09047562507566533,
      "grad_norm": 0.985638439655304,
      "learning_rate": 0.00018190507617620082,
      "loss": 3.8163,
      "step": 97060
    },
    {
      "epoch": 0.09048494669374442,
      "grad_norm": 1.1109967231750488,
      "learning_rate": 0.00018190321185410563,
      "loss": 3.621,
      "step": 97070
    },
    {
      "epoch": 0.0904942683118235,
      "grad_norm": 0.9483237266540527,
      "learning_rate": 0.00018190134753201043,
      "loss": 3.8402,
      "step": 97080
    },
    {
      "epoch": 0.0905035899299026,
      "grad_norm": 1.0644365549087524,
      "learning_rate": 0.0001818994832099152,
      "loss": 3.5421,
      "step": 97090
    },
    {
      "epoch": 0.09051291154798169,
      "grad_norm": 0.9556727409362793,
      "learning_rate": 0.00018189761888782002,
      "loss": 3.819,
      "step": 97100
    },
    {
      "epoch": 0.09052223316606078,
      "grad_norm": 1.2114596366882324,
      "learning_rate": 0.00018189575456572482,
      "loss": 3.8811,
      "step": 97110
    },
    {
      "epoch": 0.09053155478413988,
      "grad_norm": 1.0943492650985718,
      "learning_rate": 0.00018189389024362963,
      "loss": 3.7023,
      "step": 97120
    },
    {
      "epoch": 0.09054087640221896,
      "grad_norm": 1.1842037439346313,
      "learning_rate": 0.00018189202592153444,
      "loss": 3.7143,
      "step": 97130
    },
    {
      "epoch": 0.09055019802029805,
      "grad_norm": 1.179864525794983,
      "learning_rate": 0.00018189016159943922,
      "loss": 3.7165,
      "step": 97140
    },
    {
      "epoch": 0.09055951963837715,
      "grad_norm": 1.0748746395111084,
      "learning_rate": 0.00018188829727734402,
      "loss": 3.8251,
      "step": 97150
    },
    {
      "epoch": 0.09056884125645624,
      "grad_norm": 1.00961434841156,
      "learning_rate": 0.00018188643295524883,
      "loss": 3.6565,
      "step": 97160
    },
    {
      "epoch": 0.09057816287453534,
      "grad_norm": 1.2001690864562988,
      "learning_rate": 0.0001818845686331536,
      "loss": 3.6972,
      "step": 97170
    },
    {
      "epoch": 0.09058748449261442,
      "grad_norm": 1.114176869392395,
      "learning_rate": 0.00018188270431105844,
      "loss": 3.7468,
      "step": 97180
    },
    {
      "epoch": 0.09059680611069351,
      "grad_norm": 0.9014995694160461,
      "learning_rate": 0.00018188083998896322,
      "loss": 3.7874,
      "step": 97190
    },
    {
      "epoch": 0.0906061277287726,
      "grad_norm": 1.0280548334121704,
      "learning_rate": 0.00018187897566686803,
      "loss": 3.5091,
      "step": 97200
    },
    {
      "epoch": 0.0906154493468517,
      "grad_norm": 1.0489293336868286,
      "learning_rate": 0.00018187711134477283,
      "loss": 3.7183,
      "step": 97210
    },
    {
      "epoch": 0.0906247709649308,
      "grad_norm": 1.1437894105911255,
      "learning_rate": 0.0001818752470226776,
      "loss": 3.9675,
      "step": 97220
    },
    {
      "epoch": 0.09063409258300989,
      "grad_norm": 1.2824063301086426,
      "learning_rate": 0.00018187338270058242,
      "loss": 3.6312,
      "step": 97230
    },
    {
      "epoch": 0.09064341420108897,
      "grad_norm": 1.1292521953582764,
      "learning_rate": 0.00018187151837848722,
      "loss": 3.6282,
      "step": 97240
    },
    {
      "epoch": 0.09065273581916806,
      "grad_norm": 1.276574730873108,
      "learning_rate": 0.00018186965405639203,
      "loss": 3.745,
      "step": 97250
    },
    {
      "epoch": 0.09066205743724716,
      "grad_norm": 0.8368966579437256,
      "learning_rate": 0.00018186778973429684,
      "loss": 3.6405,
      "step": 97260
    },
    {
      "epoch": 0.09067137905532625,
      "grad_norm": 0.9517053961753845,
      "learning_rate": 0.00018186592541220162,
      "loss": 3.7212,
      "step": 97270
    },
    {
      "epoch": 0.09068070067340535,
      "grad_norm": 1.006602168083191,
      "learning_rate": 0.00018186406109010642,
      "loss": 3.8814,
      "step": 97280
    },
    {
      "epoch": 0.09069002229148443,
      "grad_norm": 1.055483341217041,
      "learning_rate": 0.00018186219676801123,
      "loss": 3.7152,
      "step": 97290
    },
    {
      "epoch": 0.09069934390956352,
      "grad_norm": 1.1489899158477783,
      "learning_rate": 0.00018186033244591603,
      "loss": 3.5951,
      "step": 97300
    },
    {
      "epoch": 0.09070866552764262,
      "grad_norm": 1.3481189012527466,
      "learning_rate": 0.00018185846812382084,
      "loss": 3.6601,
      "step": 97310
    },
    {
      "epoch": 0.09071798714572171,
      "grad_norm": 1.186506748199463,
      "learning_rate": 0.00018185660380172562,
      "loss": 3.5591,
      "step": 97320
    },
    {
      "epoch": 0.0907273087638008,
      "grad_norm": 1.0849560499191284,
      "learning_rate": 0.00018185473947963043,
      "loss": 3.838,
      "step": 97330
    },
    {
      "epoch": 0.09073663038187989,
      "grad_norm": 1.1440247297286987,
      "learning_rate": 0.0001818528751575352,
      "loss": 3.6351,
      "step": 97340
    },
    {
      "epoch": 0.09074595199995898,
      "grad_norm": 1.1395658254623413,
      "learning_rate": 0.00018185101083544004,
      "loss": 3.5545,
      "step": 97350
    },
    {
      "epoch": 0.09075527361803808,
      "grad_norm": 1.0118591785430908,
      "learning_rate": 0.00018184914651334485,
      "loss": 3.7629,
      "step": 97360
    },
    {
      "epoch": 0.09076459523611717,
      "grad_norm": 1.0995596647262573,
      "learning_rate": 0.00018184728219124962,
      "loss": 3.7779,
      "step": 97370
    },
    {
      "epoch": 0.09077391685419627,
      "grad_norm": 1.0617872476577759,
      "learning_rate": 0.00018184541786915443,
      "loss": 3.5501,
      "step": 97380
    },
    {
      "epoch": 0.09078323847227535,
      "grad_norm": 1.065523624420166,
      "learning_rate": 0.0001818435535470592,
      "loss": 3.632,
      "step": 97390
    },
    {
      "epoch": 0.09079256009035444,
      "grad_norm": 1.025030255317688,
      "learning_rate": 0.00018184168922496402,
      "loss": 3.7695,
      "step": 97400
    },
    {
      "epoch": 0.09080188170843354,
      "grad_norm": 1.1348707675933838,
      "learning_rate": 0.00018183982490286885,
      "loss": 3.7022,
      "step": 97410
    },
    {
      "epoch": 0.09081120332651263,
      "grad_norm": 1.10777747631073,
      "learning_rate": 0.00018183796058077363,
      "loss": 3.6431,
      "step": 97420
    },
    {
      "epoch": 0.09082052494459172,
      "grad_norm": 1.0840188264846802,
      "learning_rate": 0.00018183609625867844,
      "loss": 3.8522,
      "step": 97430
    },
    {
      "epoch": 0.0908298465626708,
      "grad_norm": 1.010699987411499,
      "learning_rate": 0.00018183423193658321,
      "loss": 3.6146,
      "step": 97440
    },
    {
      "epoch": 0.0908391681807499,
      "grad_norm": 1.1955797672271729,
      "learning_rate": 0.00018183236761448802,
      "loss": 3.7653,
      "step": 97450
    },
    {
      "epoch": 0.090848489798829,
      "grad_norm": 1.2666447162628174,
      "learning_rate": 0.00018183050329239283,
      "loss": 3.7319,
      "step": 97460
    },
    {
      "epoch": 0.09085781141690809,
      "grad_norm": 1.0069345235824585,
      "learning_rate": 0.00018182863897029763,
      "loss": 3.6043,
      "step": 97470
    },
    {
      "epoch": 0.09086713303498718,
      "grad_norm": 1.0562242269515991,
      "learning_rate": 0.00018182677464820244,
      "loss": 3.8156,
      "step": 97480
    },
    {
      "epoch": 0.09087645465306628,
      "grad_norm": 1.006013035774231,
      "learning_rate": 0.00018182491032610722,
      "loss": 3.712,
      "step": 97490
    },
    {
      "epoch": 0.09088577627114536,
      "grad_norm": 0.9954500198364258,
      "learning_rate": 0.00018182304600401203,
      "loss": 3.7162,
      "step": 97500
    },
    {
      "epoch": 0.09089509788922445,
      "grad_norm": 0.9690423607826233,
      "learning_rate": 0.00018182118168191683,
      "loss": 3.6569,
      "step": 97510
    },
    {
      "epoch": 0.09090441950730355,
      "grad_norm": 1.0613842010498047,
      "learning_rate": 0.00018181931735982164,
      "loss": 3.7245,
      "step": 97520
    },
    {
      "epoch": 0.09091374112538264,
      "grad_norm": 0.8833327293395996,
      "learning_rate": 0.00018181745303772644,
      "loss": 3.5641,
      "step": 97530
    },
    {
      "epoch": 0.09092306274346174,
      "grad_norm": 1.1190400123596191,
      "learning_rate": 0.00018181558871563122,
      "loss": 3.4807,
      "step": 97540
    },
    {
      "epoch": 0.09093238436154082,
      "grad_norm": 1.1457152366638184,
      "learning_rate": 0.00018181372439353603,
      "loss": 3.7291,
      "step": 97550
    },
    {
      "epoch": 0.09094170597961991,
      "grad_norm": 0.9180312752723694,
      "learning_rate": 0.0001818118600714408,
      "loss": 3.7634,
      "step": 97560
    },
    {
      "epoch": 0.090951027597699,
      "grad_norm": 1.1369123458862305,
      "learning_rate": 0.00018180999574934564,
      "loss": 3.8354,
      "step": 97570
    },
    {
      "epoch": 0.0909603492157781,
      "grad_norm": 1.293229103088379,
      "learning_rate": 0.00018180813142725045,
      "loss": 3.7945,
      "step": 97580
    },
    {
      "epoch": 0.0909696708338572,
      "grad_norm": 1.0849710702896118,
      "learning_rate": 0.00018180626710515523,
      "loss": 3.5632,
      "step": 97590
    },
    {
      "epoch": 0.09097899245193627,
      "grad_norm": 1.2827497720718384,
      "learning_rate": 0.00018180440278306003,
      "loss": 3.7353,
      "step": 97600
    },
    {
      "epoch": 0.09098831407001537,
      "grad_norm": 1.3075796365737915,
      "learning_rate": 0.0001818025384609648,
      "loss": 3.6937,
      "step": 97610
    },
    {
      "epoch": 0.09099763568809446,
      "grad_norm": 1.1919982433319092,
      "learning_rate": 0.00018180067413886962,
      "loss": 3.6943,
      "step": 97620
    },
    {
      "epoch": 0.09100695730617356,
      "grad_norm": 1.325741171836853,
      "learning_rate": 0.00018179880981677445,
      "loss": 3.7461,
      "step": 97630
    },
    {
      "epoch": 0.09101627892425265,
      "grad_norm": 1.1168991327285767,
      "learning_rate": 0.00018179694549467923,
      "loss": 3.6342,
      "step": 97640
    },
    {
      "epoch": 0.09102560054233173,
      "grad_norm": 1.1324588060379028,
      "learning_rate": 0.00018179508117258404,
      "loss": 3.499,
      "step": 97650
    },
    {
      "epoch": 0.09103492216041083,
      "grad_norm": 1.2252880334854126,
      "learning_rate": 0.00018179321685048882,
      "loss": 3.5063,
      "step": 97660
    },
    {
      "epoch": 0.09104424377848992,
      "grad_norm": 0.9476016163825989,
      "learning_rate": 0.00018179135252839362,
      "loss": 3.782,
      "step": 97670
    },
    {
      "epoch": 0.09105356539656902,
      "grad_norm": 1.1172012090682983,
      "learning_rate": 0.00018178948820629843,
      "loss": 3.7938,
      "step": 97680
    },
    {
      "epoch": 0.09106288701464811,
      "grad_norm": 1.0395475625991821,
      "learning_rate": 0.00018178762388420324,
      "loss": 3.7126,
      "step": 97690
    },
    {
      "epoch": 0.0910722086327272,
      "grad_norm": 1.1168326139450073,
      "learning_rate": 0.00018178575956210804,
      "loss": 3.7119,
      "step": 97700
    },
    {
      "epoch": 0.09108153025080629,
      "grad_norm": 0.9747536778450012,
      "learning_rate": 0.00018178389524001282,
      "loss": 3.6365,
      "step": 97710
    },
    {
      "epoch": 0.09109085186888538,
      "grad_norm": 0.9930030703544617,
      "learning_rate": 0.00018178203091791763,
      "loss": 3.9303,
      "step": 97720
    },
    {
      "epoch": 0.09110017348696448,
      "grad_norm": 1.2797633409500122,
      "learning_rate": 0.00018178016659582243,
      "loss": 3.492,
      "step": 97730
    },
    {
      "epoch": 0.09110949510504357,
      "grad_norm": 1.0496546030044556,
      "learning_rate": 0.00018177830227372724,
      "loss": 3.7684,
      "step": 97740
    },
    {
      "epoch": 0.09111881672312266,
      "grad_norm": 1.091089129447937,
      "learning_rate": 0.00018177643795163205,
      "loss": 3.5322,
      "step": 97750
    },
    {
      "epoch": 0.09112813834120174,
      "grad_norm": 0.9372384548187256,
      "learning_rate": 0.00018177457362953683,
      "loss": 3.7423,
      "step": 97760
    },
    {
      "epoch": 0.09113745995928084,
      "grad_norm": 1.1661220788955688,
      "learning_rate": 0.00018177270930744163,
      "loss": 3.7377,
      "step": 97770
    },
    {
      "epoch": 0.09114678157735993,
      "grad_norm": 1.1012725830078125,
      "learning_rate": 0.00018177084498534644,
      "loss": 3.77,
      "step": 97780
    },
    {
      "epoch": 0.09115610319543903,
      "grad_norm": 0.9103452563285828,
      "learning_rate": 0.00018176898066325122,
      "loss": 3.7369,
      "step": 97790
    },
    {
      "epoch": 0.09116542481351812,
      "grad_norm": 1.0595115423202515,
      "learning_rate": 0.00018176711634115605,
      "loss": 3.879,
      "step": 97800
    },
    {
      "epoch": 0.0911747464315972,
      "grad_norm": 1.146592378616333,
      "learning_rate": 0.00018176525201906083,
      "loss": 3.8007,
      "step": 97810
    },
    {
      "epoch": 0.0911840680496763,
      "grad_norm": 1.0549829006195068,
      "learning_rate": 0.00018176338769696564,
      "loss": 3.7729,
      "step": 97820
    },
    {
      "epoch": 0.09119338966775539,
      "grad_norm": 1.1930350065231323,
      "learning_rate": 0.00018176152337487044,
      "loss": 3.7816,
      "step": 97830
    },
    {
      "epoch": 0.09120271128583449,
      "grad_norm": 1.0784467458724976,
      "learning_rate": 0.00018175965905277522,
      "loss": 3.6907,
      "step": 97840
    },
    {
      "epoch": 0.09121203290391358,
      "grad_norm": 1.0645114183425903,
      "learning_rate": 0.00018175779473068003,
      "loss": 3.7259,
      "step": 97850
    },
    {
      "epoch": 0.09122135452199266,
      "grad_norm": 1.1494264602661133,
      "learning_rate": 0.00018175593040858484,
      "loss": 3.7043,
      "step": 97860
    },
    {
      "epoch": 0.09123067614007176,
      "grad_norm": 1.4213964939117432,
      "learning_rate": 0.00018175406608648964,
      "loss": 3.6036,
      "step": 97870
    },
    {
      "epoch": 0.09123999775815085,
      "grad_norm": 1.1491549015045166,
      "learning_rate": 0.00018175220176439445,
      "loss": 3.8007,
      "step": 97880
    },
    {
      "epoch": 0.09124931937622995,
      "grad_norm": 1.1492321491241455,
      "learning_rate": 0.00018175033744229923,
      "loss": 3.8659,
      "step": 97890
    },
    {
      "epoch": 0.09125864099430904,
      "grad_norm": 1.0803942680358887,
      "learning_rate": 0.00018174847312020403,
      "loss": 3.5773,
      "step": 97900
    },
    {
      "epoch": 0.09126796261238813,
      "grad_norm": 1.1514997482299805,
      "learning_rate": 0.00018174660879810884,
      "loss": 3.8568,
      "step": 97910
    },
    {
      "epoch": 0.09127728423046721,
      "grad_norm": 0.9598230123519897,
      "learning_rate": 0.00018174474447601365,
      "loss": 3.7873,
      "step": 97920
    },
    {
      "epoch": 0.09128660584854631,
      "grad_norm": 1.0593504905700684,
      "learning_rate": 0.00018174288015391845,
      "loss": 3.6936,
      "step": 97930
    },
    {
      "epoch": 0.0912959274666254,
      "grad_norm": 1.4600787162780762,
      "learning_rate": 0.00018174101583182323,
      "loss": 3.871,
      "step": 97940
    },
    {
      "epoch": 0.0913052490847045,
      "grad_norm": 1.1760592460632324,
      "learning_rate": 0.00018173915150972804,
      "loss": 3.7593,
      "step": 97950
    },
    {
      "epoch": 0.09131457070278359,
      "grad_norm": 1.0404059886932373,
      "learning_rate": 0.00018173728718763284,
      "loss": 3.7428,
      "step": 97960
    },
    {
      "epoch": 0.09132389232086267,
      "grad_norm": 1.1598315238952637,
      "learning_rate": 0.00018173542286553765,
      "loss": 3.6795,
      "step": 97970
    },
    {
      "epoch": 0.09133321393894177,
      "grad_norm": 1.0912681818008423,
      "learning_rate": 0.00018173355854344246,
      "loss": 3.7159,
      "step": 97980
    },
    {
      "epoch": 0.09134253555702086,
      "grad_norm": 1.4197496175765991,
      "learning_rate": 0.00018173169422134724,
      "loss": 3.5742,
      "step": 97990
    },
    {
      "epoch": 0.09135185717509996,
      "grad_norm": 0.9649393558502197,
      "learning_rate": 0.00018172982989925204,
      "loss": 3.6057,
      "step": 98000
    },
    {
      "epoch": 0.09136117879317905,
      "grad_norm": 0.9378685355186462,
      "learning_rate": 0.00018172796557715682,
      "loss": 4.0184,
      "step": 98010
    },
    {
      "epoch": 0.09137050041125813,
      "grad_norm": 1.121505856513977,
      "learning_rate": 0.00018172610125506166,
      "loss": 3.926,
      "step": 98020
    },
    {
      "epoch": 0.09137982202933723,
      "grad_norm": 1.333176612854004,
      "learning_rate": 0.00018172423693296646,
      "loss": 3.6722,
      "step": 98030
    },
    {
      "epoch": 0.09138914364741632,
      "grad_norm": 0.891069769859314,
      "learning_rate": 0.00018172237261087124,
      "loss": 3.6301,
      "step": 98040
    },
    {
      "epoch": 0.09139846526549542,
      "grad_norm": 0.9868804812431335,
      "learning_rate": 0.00018172050828877605,
      "loss": 3.8506,
      "step": 98050
    },
    {
      "epoch": 0.09140778688357451,
      "grad_norm": 0.9849117398262024,
      "learning_rate": 0.00018171864396668083,
      "loss": 3.8102,
      "step": 98060
    },
    {
      "epoch": 0.09141710850165359,
      "grad_norm": 0.958265483379364,
      "learning_rate": 0.00018171677964458563,
      "loss": 3.702,
      "step": 98070
    },
    {
      "epoch": 0.09142643011973269,
      "grad_norm": 1.1586703062057495,
      "learning_rate": 0.00018171491532249047,
      "loss": 3.8439,
      "step": 98080
    },
    {
      "epoch": 0.09143575173781178,
      "grad_norm": 1.1937463283538818,
      "learning_rate": 0.00018171305100039525,
      "loss": 3.8355,
      "step": 98090
    },
    {
      "epoch": 0.09144507335589087,
      "grad_norm": 0.9904147982597351,
      "learning_rate": 0.00018171118667830005,
      "loss": 3.7246,
      "step": 98100
    },
    {
      "epoch": 0.09145439497396997,
      "grad_norm": 1.1695220470428467,
      "learning_rate": 0.00018170932235620483,
      "loss": 3.7629,
      "step": 98110
    },
    {
      "epoch": 0.09146371659204905,
      "grad_norm": 0.9828881621360779,
      "learning_rate": 0.00018170745803410964,
      "loss": 3.7326,
      "step": 98120
    },
    {
      "epoch": 0.09147303821012814,
      "grad_norm": 1.0696651935577393,
      "learning_rate": 0.00018170559371201444,
      "loss": 3.7258,
      "step": 98130
    },
    {
      "epoch": 0.09148235982820724,
      "grad_norm": 1.0409986972808838,
      "learning_rate": 0.00018170372938991925,
      "loss": 3.7458,
      "step": 98140
    },
    {
      "epoch": 0.09149168144628633,
      "grad_norm": 1.01959228515625,
      "learning_rate": 0.00018170186506782406,
      "loss": 3.7288,
      "step": 98150
    },
    {
      "epoch": 0.09150100306436543,
      "grad_norm": 0.952925443649292,
      "learning_rate": 0.00018170000074572884,
      "loss": 3.6749,
      "step": 98160
    },
    {
      "epoch": 0.09151032468244452,
      "grad_norm": 1.1426008939743042,
      "learning_rate": 0.00018169813642363364,
      "loss": 3.8412,
      "step": 98170
    },
    {
      "epoch": 0.0915196463005236,
      "grad_norm": 0.9401000142097473,
      "learning_rate": 0.00018169627210153845,
      "loss": 3.7385,
      "step": 98180
    },
    {
      "epoch": 0.0915289679186027,
      "grad_norm": 1.0262877941131592,
      "learning_rate": 0.00018169440777944325,
      "loss": 3.789,
      "step": 98190
    },
    {
      "epoch": 0.09153828953668179,
      "grad_norm": 1.0447685718536377,
      "learning_rate": 0.00018169254345734806,
      "loss": 3.7041,
      "step": 98200
    },
    {
      "epoch": 0.09154761115476089,
      "grad_norm": 1.1428000926971436,
      "learning_rate": 0.00018169067913525284,
      "loss": 3.6908,
      "step": 98210
    },
    {
      "epoch": 0.09155693277283998,
      "grad_norm": 1.0940741300582886,
      "learning_rate": 0.00018168881481315765,
      "loss": 3.6601,
      "step": 98220
    },
    {
      "epoch": 0.09156625439091906,
      "grad_norm": 1.144384503364563,
      "learning_rate": 0.00018168695049106245,
      "loss": 3.6865,
      "step": 98230
    },
    {
      "epoch": 0.09157557600899816,
      "grad_norm": 1.0652209520339966,
      "learning_rate": 0.00018168508616896723,
      "loss": 3.8076,
      "step": 98240
    },
    {
      "epoch": 0.09158489762707725,
      "grad_norm": 1.485579013824463,
      "learning_rate": 0.00018168322184687206,
      "loss": 3.9344,
      "step": 98250
    },
    {
      "epoch": 0.09159421924515634,
      "grad_norm": 1.7029590606689453,
      "learning_rate": 0.00018168135752477684,
      "loss": 3.6134,
      "step": 98260
    },
    {
      "epoch": 0.09160354086323544,
      "grad_norm": 0.9470342397689819,
      "learning_rate": 0.00018167949320268165,
      "loss": 3.8377,
      "step": 98270
    },
    {
      "epoch": 0.09161286248131452,
      "grad_norm": 1.1330674886703491,
      "learning_rate": 0.00018167762888058646,
      "loss": 3.7841,
      "step": 98280
    },
    {
      "epoch": 0.09162218409939361,
      "grad_norm": 1.0746930837631226,
      "learning_rate": 0.00018167576455849124,
      "loss": 3.747,
      "step": 98290
    },
    {
      "epoch": 0.09163150571747271,
      "grad_norm": 0.86655193567276,
      "learning_rate": 0.00018167390023639604,
      "loss": 3.8531,
      "step": 98300
    },
    {
      "epoch": 0.0916408273355518,
      "grad_norm": 1.1919525861740112,
      "learning_rate": 0.00018167203591430085,
      "loss": 3.7228,
      "step": 98310
    },
    {
      "epoch": 0.0916501489536309,
      "grad_norm": 1.0149197578430176,
      "learning_rate": 0.00018167017159220565,
      "loss": 3.9154,
      "step": 98320
    },
    {
      "epoch": 0.09165947057170998,
      "grad_norm": 1.0595226287841797,
      "learning_rate": 0.00018166830727011046,
      "loss": 3.6465,
      "step": 98330
    },
    {
      "epoch": 0.09166879218978907,
      "grad_norm": 1.0070611238479614,
      "learning_rate": 0.00018166644294801524,
      "loss": 3.9959,
      "step": 98340
    },
    {
      "epoch": 0.09167811380786817,
      "grad_norm": 0.8552594780921936,
      "learning_rate": 0.00018166457862592005,
      "loss": 3.735,
      "step": 98350
    },
    {
      "epoch": 0.09168743542594726,
      "grad_norm": 1.1194347143173218,
      "learning_rate": 0.00018166271430382485,
      "loss": 3.7156,
      "step": 98360
    },
    {
      "epoch": 0.09169675704402636,
      "grad_norm": 1.0654698610305786,
      "learning_rate": 0.00018166084998172966,
      "loss": 3.7029,
      "step": 98370
    },
    {
      "epoch": 0.09170607866210545,
      "grad_norm": 1.1316449642181396,
      "learning_rate": 0.00018165898565963447,
      "loss": 3.6655,
      "step": 98380
    },
    {
      "epoch": 0.09171540028018453,
      "grad_norm": 1.0493441820144653,
      "learning_rate": 0.00018165712133753924,
      "loss": 3.7527,
      "step": 98390
    },
    {
      "epoch": 0.09172472189826363,
      "grad_norm": 0.9561657309532166,
      "learning_rate": 0.00018165525701544405,
      "loss": 3.7616,
      "step": 98400
    },
    {
      "epoch": 0.09173404351634272,
      "grad_norm": 1.1015323400497437,
      "learning_rate": 0.00018165339269334886,
      "loss": 3.6201,
      "step": 98410
    },
    {
      "epoch": 0.09174336513442181,
      "grad_norm": 1.0793379545211792,
      "learning_rate": 0.00018165152837125366,
      "loss": 3.8334,
      "step": 98420
    },
    {
      "epoch": 0.09175268675250091,
      "grad_norm": 1.1218693256378174,
      "learning_rate": 0.00018164966404915847,
      "loss": 4.035,
      "step": 98430
    },
    {
      "epoch": 0.09176200837057999,
      "grad_norm": 1.0178041458129883,
      "learning_rate": 0.00018164779972706325,
      "loss": 3.9248,
      "step": 98440
    },
    {
      "epoch": 0.09177132998865908,
      "grad_norm": 1.0413610935211182,
      "learning_rate": 0.00018164593540496806,
      "loss": 3.7543,
      "step": 98450
    },
    {
      "epoch": 0.09178065160673818,
      "grad_norm": 1.1149568557739258,
      "learning_rate": 0.00018164407108287283,
      "loss": 3.6032,
      "step": 98460
    },
    {
      "epoch": 0.09178997322481727,
      "grad_norm": 1.2751553058624268,
      "learning_rate": 0.00018164220676077767,
      "loss": 3.8375,
      "step": 98470
    },
    {
      "epoch": 0.09179929484289637,
      "grad_norm": 1.1495461463928223,
      "learning_rate": 0.00018164034243868247,
      "loss": 3.5993,
      "step": 98480
    },
    {
      "epoch": 0.09180861646097545,
      "grad_norm": 1.1277596950531006,
      "learning_rate": 0.00018163847811658725,
      "loss": 3.9165,
      "step": 98490
    },
    {
      "epoch": 0.09181793807905454,
      "grad_norm": 1.0145355463027954,
      "learning_rate": 0.00018163661379449206,
      "loss": 3.5996,
      "step": 98500
    },
    {
      "epoch": 0.09182725969713364,
      "grad_norm": 1.0127670764923096,
      "learning_rate": 0.00018163474947239684,
      "loss": 3.8319,
      "step": 98510
    },
    {
      "epoch": 0.09183658131521273,
      "grad_norm": 0.9138212203979492,
      "learning_rate": 0.00018163288515030165,
      "loss": 3.5191,
      "step": 98520
    },
    {
      "epoch": 0.09184590293329183,
      "grad_norm": 0.9913418292999268,
      "learning_rate": 0.00018163102082820648,
      "loss": 3.5101,
      "step": 98530
    },
    {
      "epoch": 0.0918552245513709,
      "grad_norm": 1.1279042959213257,
      "learning_rate": 0.00018162915650611126,
      "loss": 3.6847,
      "step": 98540
    },
    {
      "epoch": 0.09186454616945,
      "grad_norm": 1.0422868728637695,
      "learning_rate": 0.00018162729218401606,
      "loss": 3.714,
      "step": 98550
    },
    {
      "epoch": 0.0918738677875291,
      "grad_norm": 1.035759687423706,
      "learning_rate": 0.00018162542786192084,
      "loss": 3.7646,
      "step": 98560
    },
    {
      "epoch": 0.09188318940560819,
      "grad_norm": 1.060461163520813,
      "learning_rate": 0.00018162356353982565,
      "loss": 3.6626,
      "step": 98570
    },
    {
      "epoch": 0.09189251102368728,
      "grad_norm": 1.111798644065857,
      "learning_rate": 0.00018162169921773046,
      "loss": 3.6368,
      "step": 98580
    },
    {
      "epoch": 0.09190183264176638,
      "grad_norm": 0.956245481967926,
      "learning_rate": 0.00018161983489563526,
      "loss": 3.7604,
      "step": 98590
    },
    {
      "epoch": 0.09191115425984546,
      "grad_norm": 1.009149432182312,
      "learning_rate": 0.00018161797057354007,
      "loss": 3.8753,
      "step": 98600
    },
    {
      "epoch": 0.09192047587792455,
      "grad_norm": 1.1448431015014648,
      "learning_rate": 0.00018161610625144485,
      "loss": 3.8287,
      "step": 98610
    },
    {
      "epoch": 0.09192979749600365,
      "grad_norm": 1.2939140796661377,
      "learning_rate": 0.00018161424192934965,
      "loss": 3.6783,
      "step": 98620
    },
    {
      "epoch": 0.09193911911408274,
      "grad_norm": 1.1236263513565063,
      "learning_rate": 0.00018161237760725446,
      "loss": 3.6135,
      "step": 98630
    },
    {
      "epoch": 0.09194844073216184,
      "grad_norm": 1.0775787830352783,
      "learning_rate": 0.00018161051328515927,
      "loss": 3.624,
      "step": 98640
    },
    {
      "epoch": 0.09195776235024092,
      "grad_norm": 1.1436105966567993,
      "learning_rate": 0.00018160864896306407,
      "loss": 3.5517,
      "step": 98650
    },
    {
      "epoch": 0.09196708396832001,
      "grad_norm": 1.134040117263794,
      "learning_rate": 0.00018160678464096885,
      "loss": 3.7348,
      "step": 98660
    },
    {
      "epoch": 0.09197640558639911,
      "grad_norm": 1.0241355895996094,
      "learning_rate": 0.00018160492031887366,
      "loss": 3.8678,
      "step": 98670
    },
    {
      "epoch": 0.0919857272044782,
      "grad_norm": 1.0400749444961548,
      "learning_rate": 0.00018160305599677847,
      "loss": 3.5529,
      "step": 98680
    },
    {
      "epoch": 0.0919950488225573,
      "grad_norm": 1.2183928489685059,
      "learning_rate": 0.00018160119167468324,
      "loss": 3.7158,
      "step": 98690
    },
    {
      "epoch": 0.09200437044063638,
      "grad_norm": 1.1184412240982056,
      "learning_rate": 0.00018159932735258808,
      "loss": 3.6213,
      "step": 98700
    },
    {
      "epoch": 0.09201369205871547,
      "grad_norm": 1.0042668581008911,
      "learning_rate": 0.00018159746303049286,
      "loss": 3.8591,
      "step": 98710
    },
    {
      "epoch": 0.09202301367679457,
      "grad_norm": 1.0609978437423706,
      "learning_rate": 0.00018159559870839766,
      "loss": 3.981,
      "step": 98720
    },
    {
      "epoch": 0.09203233529487366,
      "grad_norm": 1.0187801122665405,
      "learning_rate": 0.00018159373438630247,
      "loss": 3.6123,
      "step": 98730
    },
    {
      "epoch": 0.09204165691295275,
      "grad_norm": 1.0963318347930908,
      "learning_rate": 0.00018159187006420725,
      "loss": 3.7369,
      "step": 98740
    },
    {
      "epoch": 0.09205097853103184,
      "grad_norm": 0.8961384296417236,
      "learning_rate": 0.00018159000574211206,
      "loss": 3.7501,
      "step": 98750
    },
    {
      "epoch": 0.09206030014911093,
      "grad_norm": 1.094419002532959,
      "learning_rate": 0.00018158814142001686,
      "loss": 3.8931,
      "step": 98760
    },
    {
      "epoch": 0.09206962176719002,
      "grad_norm": 0.9126925468444824,
      "learning_rate": 0.00018158627709792167,
      "loss": 3.8648,
      "step": 98770
    },
    {
      "epoch": 0.09207894338526912,
      "grad_norm": 0.9813530445098877,
      "learning_rate": 0.00018158441277582647,
      "loss": 3.5547,
      "step": 98780
    },
    {
      "epoch": 0.09208826500334821,
      "grad_norm": 0.9544008374214172,
      "learning_rate": 0.00018158254845373125,
      "loss": 3.7855,
      "step": 98790
    },
    {
      "epoch": 0.0920975866214273,
      "grad_norm": 1.059839129447937,
      "learning_rate": 0.00018158068413163606,
      "loss": 3.8263,
      "step": 98800
    },
    {
      "epoch": 0.09210690823950639,
      "grad_norm": 1.0386300086975098,
      "learning_rate": 0.00018157881980954087,
      "loss": 3.6814,
      "step": 98810
    },
    {
      "epoch": 0.09211622985758548,
      "grad_norm": 1.096948504447937,
      "learning_rate": 0.00018157695548744567,
      "loss": 3.9014,
      "step": 98820
    },
    {
      "epoch": 0.09212555147566458,
      "grad_norm": 0.9499149322509766,
      "learning_rate": 0.00018157509116535048,
      "loss": 3.5899,
      "step": 98830
    },
    {
      "epoch": 0.09213487309374367,
      "grad_norm": 1.3622896671295166,
      "learning_rate": 0.00018157322684325526,
      "loss": 3.6696,
      "step": 98840
    },
    {
      "epoch": 0.09214419471182277,
      "grad_norm": 1.080320119857788,
      "learning_rate": 0.00018157136252116006,
      "loss": 3.7748,
      "step": 98850
    },
    {
      "epoch": 0.09215351632990185,
      "grad_norm": 1.0111645460128784,
      "learning_rate": 0.00018156949819906487,
      "loss": 3.8112,
      "step": 98860
    },
    {
      "epoch": 0.09216283794798094,
      "grad_norm": 1.056279182434082,
      "learning_rate": 0.00018156763387696968,
      "loss": 3.6739,
      "step": 98870
    },
    {
      "epoch": 0.09217215956606004,
      "grad_norm": 1.0338772535324097,
      "learning_rate": 0.00018156576955487446,
      "loss": 3.7632,
      "step": 98880
    },
    {
      "epoch": 0.09218148118413913,
      "grad_norm": 1.2518932819366455,
      "learning_rate": 0.00018156390523277926,
      "loss": 3.5791,
      "step": 98890
    },
    {
      "epoch": 0.09219080280221822,
      "grad_norm": 1.372376799583435,
      "learning_rate": 0.00018156204091068407,
      "loss": 3.7982,
      "step": 98900
    },
    {
      "epoch": 0.0922001244202973,
      "grad_norm": 1.167006015777588,
      "learning_rate": 0.00018156017658858885,
      "loss": 3.8014,
      "step": 98910
    },
    {
      "epoch": 0.0922094460383764,
      "grad_norm": 1.0605753660202026,
      "learning_rate": 0.00018155831226649368,
      "loss": 3.744,
      "step": 98920
    },
    {
      "epoch": 0.0922187676564555,
      "grad_norm": 1.2102160453796387,
      "learning_rate": 0.00018155644794439846,
      "loss": 3.6945,
      "step": 98930
    },
    {
      "epoch": 0.09222808927453459,
      "grad_norm": 0.9129966497421265,
      "learning_rate": 0.00018155458362230327,
      "loss": 3.808,
      "step": 98940
    },
    {
      "epoch": 0.09223741089261368,
      "grad_norm": 0.9761930108070374,
      "learning_rate": 0.00018155271930020807,
      "loss": 3.7562,
      "step": 98950
    },
    {
      "epoch": 0.09224673251069276,
      "grad_norm": 1.014920711517334,
      "learning_rate": 0.00018155085497811285,
      "loss": 3.6533,
      "step": 98960
    },
    {
      "epoch": 0.09225605412877186,
      "grad_norm": 1.0837379693984985,
      "learning_rate": 0.00018154899065601766,
      "loss": 3.6324,
      "step": 98970
    },
    {
      "epoch": 0.09226537574685095,
      "grad_norm": 1.0336427688598633,
      "learning_rate": 0.00018154712633392246,
      "loss": 3.701,
      "step": 98980
    },
    {
      "epoch": 0.09227469736493005,
      "grad_norm": 1.070528507232666,
      "learning_rate": 0.00018154526201182727,
      "loss": 3.5513,
      "step": 98990
    },
    {
      "epoch": 0.09228401898300914,
      "grad_norm": 1.0421063899993896,
      "learning_rate": 0.00018154339768973208,
      "loss": 3.7016,
      "step": 99000
    },
    {
      "epoch": 0.09229334060108822,
      "grad_norm": 1.0117619037628174,
      "learning_rate": 0.00018154153336763686,
      "loss": 3.84,
      "step": 99010
    },
    {
      "epoch": 0.09230266221916732,
      "grad_norm": 1.2311896085739136,
      "learning_rate": 0.00018153966904554166,
      "loss": 3.5883,
      "step": 99020
    },
    {
      "epoch": 0.09231198383724641,
      "grad_norm": 1.06157386302948,
      "learning_rate": 0.00018153780472344647,
      "loss": 3.8415,
      "step": 99030
    },
    {
      "epoch": 0.0923213054553255,
      "grad_norm": 1.0229465961456299,
      "learning_rate": 0.00018153594040135128,
      "loss": 3.6919,
      "step": 99040
    },
    {
      "epoch": 0.0923306270734046,
      "grad_norm": 1.085402488708496,
      "learning_rate": 0.00018153407607925608,
      "loss": 3.7355,
      "step": 99050
    },
    {
      "epoch": 0.0923399486914837,
      "grad_norm": 1.0498968362808228,
      "learning_rate": 0.00018153221175716086,
      "loss": 3.642,
      "step": 99060
    },
    {
      "epoch": 0.09234927030956278,
      "grad_norm": 1.0114507675170898,
      "learning_rate": 0.00018153034743506567,
      "loss": 3.6656,
      "step": 99070
    },
    {
      "epoch": 0.09235859192764187,
      "grad_norm": 0.9964722394943237,
      "learning_rate": 0.00018152848311297047,
      "loss": 3.6161,
      "step": 99080
    },
    {
      "epoch": 0.09236791354572096,
      "grad_norm": 1.0501893758773804,
      "learning_rate": 0.00018152661879087528,
      "loss": 3.4759,
      "step": 99090
    },
    {
      "epoch": 0.09237723516380006,
      "grad_norm": 1.149240255355835,
      "learning_rate": 0.00018152475446878009,
      "loss": 3.834,
      "step": 99100
    },
    {
      "epoch": 0.09238655678187915,
      "grad_norm": 0.8483846187591553,
      "learning_rate": 0.00018152289014668487,
      "loss": 3.5409,
      "step": 99110
    },
    {
      "epoch": 0.09239587839995823,
      "grad_norm": 0.948738694190979,
      "learning_rate": 0.00018152102582458967,
      "loss": 3.8807,
      "step": 99120
    },
    {
      "epoch": 0.09240520001803733,
      "grad_norm": 1.1909955739974976,
      "learning_rate": 0.00018151916150249445,
      "loss": 3.7783,
      "step": 99130
    },
    {
      "epoch": 0.09241452163611642,
      "grad_norm": 0.9830378293991089,
      "learning_rate": 0.00018151729718039926,
      "loss": 3.6627,
      "step": 99140
    },
    {
      "epoch": 0.09242384325419552,
      "grad_norm": 1.0331238508224487,
      "learning_rate": 0.0001815154328583041,
      "loss": 3.6988,
      "step": 99150
    },
    {
      "epoch": 0.09243316487227461,
      "grad_norm": 0.9275707602500916,
      "learning_rate": 0.00018151356853620887,
      "loss": 3.6891,
      "step": 99160
    },
    {
      "epoch": 0.09244248649035369,
      "grad_norm": 1.1313669681549072,
      "learning_rate": 0.00018151170421411368,
      "loss": 3.8003,
      "step": 99170
    },
    {
      "epoch": 0.09245180810843279,
      "grad_norm": 1.417948842048645,
      "learning_rate": 0.00018150983989201846,
      "loss": 3.8029,
      "step": 99180
    },
    {
      "epoch": 0.09246112972651188,
      "grad_norm": 0.9712142944335938,
      "learning_rate": 0.00018150797556992326,
      "loss": 3.6552,
      "step": 99190
    },
    {
      "epoch": 0.09247045134459098,
      "grad_norm": 0.8872452974319458,
      "learning_rate": 0.00018150611124782807,
      "loss": 3.7704,
      "step": 99200
    },
    {
      "epoch": 0.09247977296267007,
      "grad_norm": 0.9925699830055237,
      "learning_rate": 0.00018150424692573287,
      "loss": 3.7412,
      "step": 99210
    },
    {
      "epoch": 0.09248909458074915,
      "grad_norm": 1.1019749641418457,
      "learning_rate": 0.00018150238260363768,
      "loss": 3.583,
      "step": 99220
    },
    {
      "epoch": 0.09249841619882825,
      "grad_norm": 1.1926358938217163,
      "learning_rate": 0.00018150051828154246,
      "loss": 3.7864,
      "step": 99230
    },
    {
      "epoch": 0.09250773781690734,
      "grad_norm": 1.0335543155670166,
      "learning_rate": 0.00018149865395944727,
      "loss": 3.5945,
      "step": 99240
    },
    {
      "epoch": 0.09251705943498643,
      "grad_norm": 0.8326343894004822,
      "learning_rate": 0.00018149678963735207,
      "loss": 3.7486,
      "step": 99250
    },
    {
      "epoch": 0.09252638105306553,
      "grad_norm": 1.0863953828811646,
      "learning_rate": 0.00018149492531525688,
      "loss": 3.6663,
      "step": 99260
    },
    {
      "epoch": 0.09253570267114462,
      "grad_norm": 1.0370354652404785,
      "learning_rate": 0.00018149306099316168,
      "loss": 3.59,
      "step": 99270
    },
    {
      "epoch": 0.0925450242892237,
      "grad_norm": 0.9985873699188232,
      "learning_rate": 0.00018149119667106646,
      "loss": 3.7543,
      "step": 99280
    },
    {
      "epoch": 0.0925543459073028,
      "grad_norm": 1.1929833889007568,
      "learning_rate": 0.00018148933234897127,
      "loss": 3.6887,
      "step": 99290
    },
    {
      "epoch": 0.09256366752538189,
      "grad_norm": 1.1904726028442383,
      "learning_rate": 0.00018148746802687608,
      "loss": 3.8113,
      "step": 99300
    },
    {
      "epoch": 0.09257298914346099,
      "grad_norm": 1.0403600931167603,
      "learning_rate": 0.00018148560370478088,
      "loss": 3.6907,
      "step": 99310
    },
    {
      "epoch": 0.09258231076154008,
      "grad_norm": 1.2625341415405273,
      "learning_rate": 0.0001814837393826857,
      "loss": 3.6825,
      "step": 99320
    },
    {
      "epoch": 0.09259163237961916,
      "grad_norm": 1.021437168121338,
      "learning_rate": 0.00018148187506059047,
      "loss": 3.6526,
      "step": 99330
    },
    {
      "epoch": 0.09260095399769826,
      "grad_norm": 1.0518471002578735,
      "learning_rate": 0.00018148001073849527,
      "loss": 3.5072,
      "step": 99340
    },
    {
      "epoch": 0.09261027561577735,
      "grad_norm": 0.9674655199050903,
      "learning_rate": 0.00018147814641640008,
      "loss": 3.8031,
      "step": 99350
    },
    {
      "epoch": 0.09261959723385645,
      "grad_norm": 1.2242047786712646,
      "learning_rate": 0.00018147628209430486,
      "loss": 3.7244,
      "step": 99360
    },
    {
      "epoch": 0.09262891885193554,
      "grad_norm": 1.0042333602905273,
      "learning_rate": 0.0001814744177722097,
      "loss": 3.8549,
      "step": 99370
    },
    {
      "epoch": 0.09263824047001462,
      "grad_norm": 1.1132982969284058,
      "learning_rate": 0.00018147255345011447,
      "loss": 3.8522,
      "step": 99380
    },
    {
      "epoch": 0.09264756208809372,
      "grad_norm": 0.8445395231246948,
      "learning_rate": 0.00018147068912801928,
      "loss": 3.7406,
      "step": 99390
    },
    {
      "epoch": 0.09265688370617281,
      "grad_norm": 0.9157341718673706,
      "learning_rate": 0.00018146882480592409,
      "loss": 3.5146,
      "step": 99400
    },
    {
      "epoch": 0.0926662053242519,
      "grad_norm": 1.1625874042510986,
      "learning_rate": 0.00018146696048382886,
      "loss": 3.6925,
      "step": 99410
    },
    {
      "epoch": 0.092675526942331,
      "grad_norm": 1.2414031028747559,
      "learning_rate": 0.00018146509616173367,
      "loss": 3.7466,
      "step": 99420
    },
    {
      "epoch": 0.09268484856041008,
      "grad_norm": 1.0848650932312012,
      "learning_rate": 0.00018146323183963848,
      "loss": 3.6937,
      "step": 99430
    },
    {
      "epoch": 0.09269417017848917,
      "grad_norm": 1.0542137622833252,
      "learning_rate": 0.00018146136751754328,
      "loss": 3.681,
      "step": 99440
    },
    {
      "epoch": 0.09270349179656827,
      "grad_norm": 1.1985903978347778,
      "learning_rate": 0.0001814595031954481,
      "loss": 3.7797,
      "step": 99450
    },
    {
      "epoch": 0.09271281341464736,
      "grad_norm": 1.1138105392456055,
      "learning_rate": 0.00018145763887335287,
      "loss": 3.7203,
      "step": 99460
    },
    {
      "epoch": 0.09272213503272646,
      "grad_norm": 1.1192656755447388,
      "learning_rate": 0.00018145577455125768,
      "loss": 3.4267,
      "step": 99470
    },
    {
      "epoch": 0.09273145665080554,
      "grad_norm": 0.9214914441108704,
      "learning_rate": 0.00018145391022916248,
      "loss": 3.4828,
      "step": 99480
    },
    {
      "epoch": 0.09274077826888463,
      "grad_norm": 1.1308960914611816,
      "learning_rate": 0.0001814520459070673,
      "loss": 3.6566,
      "step": 99490
    },
    {
      "epoch": 0.09275009988696373,
      "grad_norm": 1.4034239053726196,
      "learning_rate": 0.0001814501815849721,
      "loss": 3.7195,
      "step": 99500
    },
    {
      "epoch": 0.09275942150504282,
      "grad_norm": 1.0370112657546997,
      "learning_rate": 0.00018144831726287687,
      "loss": 3.8123,
      "step": 99510
    },
    {
      "epoch": 0.09276874312312192,
      "grad_norm": 1.123101830482483,
      "learning_rate": 0.00018144645294078168,
      "loss": 3.5859,
      "step": 99520
    },
    {
      "epoch": 0.09277806474120101,
      "grad_norm": 1.0027554035186768,
      "learning_rate": 0.00018144458861868649,
      "loss": 3.8644,
      "step": 99530
    },
    {
      "epoch": 0.09278738635928009,
      "grad_norm": 0.99598628282547,
      "learning_rate": 0.0001814427242965913,
      "loss": 3.6657,
      "step": 99540
    },
    {
      "epoch": 0.09279670797735919,
      "grad_norm": 1.1761908531188965,
      "learning_rate": 0.0001814408599744961,
      "loss": 3.72,
      "step": 99550
    },
    {
      "epoch": 0.09280602959543828,
      "grad_norm": 1.1410590410232544,
      "learning_rate": 0.00018143899565240088,
      "loss": 3.9006,
      "step": 99560
    },
    {
      "epoch": 0.09281535121351737,
      "grad_norm": 1.0127190351486206,
      "learning_rate": 0.00018143713133030568,
      "loss": 3.8378,
      "step": 99570
    },
    {
      "epoch": 0.09282467283159647,
      "grad_norm": 1.1014403104782104,
      "learning_rate": 0.00018143526700821046,
      "loss": 3.6683,
      "step": 99580
    },
    {
      "epoch": 0.09283399444967555,
      "grad_norm": 1.2428492307662964,
      "learning_rate": 0.0001814334026861153,
      "loss": 3.6482,
      "step": 99590
    },
    {
      "epoch": 0.09284331606775464,
      "grad_norm": 1.1711925268173218,
      "learning_rate": 0.0001814315383640201,
      "loss": 3.6239,
      "step": 99600
    },
    {
      "epoch": 0.09285263768583374,
      "grad_norm": 1.0224093198776245,
      "learning_rate": 0.00018142967404192488,
      "loss": 3.6067,
      "step": 99610
    },
    {
      "epoch": 0.09286195930391283,
      "grad_norm": 1.1029492616653442,
      "learning_rate": 0.0001814278097198297,
      "loss": 3.5566,
      "step": 99620
    },
    {
      "epoch": 0.09287128092199193,
      "grad_norm": 1.1380761861801147,
      "learning_rate": 0.00018142594539773447,
      "loss": 4.0976,
      "step": 99630
    },
    {
      "epoch": 0.09288060254007101,
      "grad_norm": 0.8228969573974609,
      "learning_rate": 0.00018142408107563927,
      "loss": 3.8584,
      "step": 99640
    },
    {
      "epoch": 0.0928899241581501,
      "grad_norm": 1.1299856901168823,
      "learning_rate": 0.00018142221675354408,
      "loss": 3.7777,
      "step": 99650
    },
    {
      "epoch": 0.0928992457762292,
      "grad_norm": 1.109357237815857,
      "learning_rate": 0.0001814203524314489,
      "loss": 3.556,
      "step": 99660
    },
    {
      "epoch": 0.09290856739430829,
      "grad_norm": 1.1304885149002075,
      "learning_rate": 0.0001814184881093537,
      "loss": 3.6511,
      "step": 99670
    },
    {
      "epoch": 0.09291788901238739,
      "grad_norm": 1.0951067209243774,
      "learning_rate": 0.00018141662378725847,
      "loss": 3.7344,
      "step": 99680
    },
    {
      "epoch": 0.09292721063046647,
      "grad_norm": 1.0559638738632202,
      "learning_rate": 0.00018141475946516328,
      "loss": 3.8048,
      "step": 99690
    },
    {
      "epoch": 0.09293653224854556,
      "grad_norm": 1.0126873254776,
      "learning_rate": 0.00018141289514306809,
      "loss": 3.8482,
      "step": 99700
    },
    {
      "epoch": 0.09294585386662466,
      "grad_norm": 1.1145877838134766,
      "learning_rate": 0.0001814110308209729,
      "loss": 3.7135,
      "step": 99710
    },
    {
      "epoch": 0.09295517548470375,
      "grad_norm": 1.1906803846359253,
      "learning_rate": 0.0001814091664988777,
      "loss": 3.6085,
      "step": 99720
    },
    {
      "epoch": 0.09296449710278284,
      "grad_norm": 1.0943528413772583,
      "learning_rate": 0.00018140730217678248,
      "loss": 3.7049,
      "step": 99730
    },
    {
      "epoch": 0.09297381872086194,
      "grad_norm": 1.0763370990753174,
      "learning_rate": 0.00018140543785468728,
      "loss": 3.8114,
      "step": 99740
    },
    {
      "epoch": 0.09298314033894102,
      "grad_norm": 1.1613277196884155,
      "learning_rate": 0.0001814035735325921,
      "loss": 3.7283,
      "step": 99750
    },
    {
      "epoch": 0.09299246195702011,
      "grad_norm": 1.2238973379135132,
      "learning_rate": 0.0001814017092104969,
      "loss": 3.7961,
      "step": 99760
    },
    {
      "epoch": 0.09300178357509921,
      "grad_norm": 0.9424467086791992,
      "learning_rate": 0.0001813998448884017,
      "loss": 4.0252,
      "step": 99770
    },
    {
      "epoch": 0.0930111051931783,
      "grad_norm": 1.292129635810852,
      "learning_rate": 0.00018139798056630648,
      "loss": 3.5138,
      "step": 99780
    },
    {
      "epoch": 0.0930204268112574,
      "grad_norm": 1.0252412557601929,
      "learning_rate": 0.0001813961162442113,
      "loss": 3.6861,
      "step": 99790
    },
    {
      "epoch": 0.09302974842933648,
      "grad_norm": 0.9331527948379517,
      "learning_rate": 0.0001813942519221161,
      "loss": 3.6653,
      "step": 99800
    },
    {
      "epoch": 0.09303907004741557,
      "grad_norm": 1.081695556640625,
      "learning_rate": 0.00018139238760002087,
      "loss": 3.6509,
      "step": 99810
    },
    {
      "epoch": 0.09304839166549467,
      "grad_norm": 0.9618477821350098,
      "learning_rate": 0.0001813905232779257,
      "loss": 3.8673,
      "step": 99820
    },
    {
      "epoch": 0.09305771328357376,
      "grad_norm": 1.1654953956604004,
      "learning_rate": 0.00018138865895583049,
      "loss": 3.7102,
      "step": 99830
    },
    {
      "epoch": 0.09306703490165286,
      "grad_norm": 1.0302671194076538,
      "learning_rate": 0.0001813867946337353,
      "loss": 3.6304,
      "step": 99840
    },
    {
      "epoch": 0.09307635651973194,
      "grad_norm": 1.0255435705184937,
      "learning_rate": 0.0001813849303116401,
      "loss": 3.6543,
      "step": 99850
    },
    {
      "epoch": 0.09308567813781103,
      "grad_norm": 1.1324843168258667,
      "learning_rate": 0.00018138306598954488,
      "loss": 3.6392,
      "step": 99860
    },
    {
      "epoch": 0.09309499975589013,
      "grad_norm": 1.0183604955673218,
      "learning_rate": 0.00018138120166744968,
      "loss": 3.7455,
      "step": 99870
    },
    {
      "epoch": 0.09310432137396922,
      "grad_norm": 1.0652084350585938,
      "learning_rate": 0.0001813793373453545,
      "loss": 3.6272,
      "step": 99880
    },
    {
      "epoch": 0.09311364299204831,
      "grad_norm": 1.134305715560913,
      "learning_rate": 0.0001813774730232593,
      "loss": 3.71,
      "step": 99890
    },
    {
      "epoch": 0.0931229646101274,
      "grad_norm": 1.1194669008255005,
      "learning_rate": 0.0001813756087011641,
      "loss": 3.8807,
      "step": 99900
    },
    {
      "epoch": 0.09313228622820649,
      "grad_norm": 0.9646790623664856,
      "learning_rate": 0.00018137374437906888,
      "loss": 3.7869,
      "step": 99910
    },
    {
      "epoch": 0.09314160784628558,
      "grad_norm": 1.2594538927078247,
      "learning_rate": 0.0001813718800569737,
      "loss": 3.578,
      "step": 99920
    },
    {
      "epoch": 0.09315092946436468,
      "grad_norm": 1.0324970483779907,
      "learning_rate": 0.0001813700157348785,
      "loss": 3.6617,
      "step": 99930
    },
    {
      "epoch": 0.09316025108244377,
      "grad_norm": 1.2217997312545776,
      "learning_rate": 0.0001813681514127833,
      "loss": 3.838,
      "step": 99940
    },
    {
      "epoch": 0.09316957270052287,
      "grad_norm": 1.3238922357559204,
      "learning_rate": 0.0001813662870906881,
      "loss": 3.7593,
      "step": 99950
    },
    {
      "epoch": 0.09317889431860195,
      "grad_norm": 0.9512710571289062,
      "learning_rate": 0.0001813644227685929,
      "loss": 3.6001,
      "step": 99960
    },
    {
      "epoch": 0.09318821593668104,
      "grad_norm": 0.9294722676277161,
      "learning_rate": 0.0001813625584464977,
      "loss": 3.6439,
      "step": 99970
    },
    {
      "epoch": 0.09319753755476014,
      "grad_norm": 1.0749658346176147,
      "learning_rate": 0.0001813606941244025,
      "loss": 3.7925,
      "step": 99980
    },
    {
      "epoch": 0.09320685917283923,
      "grad_norm": 1.1725707054138184,
      "learning_rate": 0.0001813588298023073,
      "loss": 3.7176,
      "step": 99990
    },
    {
      "epoch": 0.09321618079091833,
      "grad_norm": 0.9385355710983276,
      "learning_rate": 0.0001813569654802121,
      "loss": 3.6559,
      "step": 100000
    },
    {
      "epoch": 0.09322550240899741,
      "grad_norm": 1.0985842943191528,
      "learning_rate": 0.0001813551011581169,
      "loss": 3.7498,
      "step": 100010
    },
    {
      "epoch": 0.0932348240270765,
      "grad_norm": 1.1152732372283936,
      "learning_rate": 0.0001813532368360217,
      "loss": 3.5569,
      "step": 100020
    },
    {
      "epoch": 0.0932441456451556,
      "grad_norm": 1.0111693143844604,
      "learning_rate": 0.00018135137251392648,
      "loss": 3.701,
      "step": 100030
    },
    {
      "epoch": 0.09325346726323469,
      "grad_norm": 1.0138810873031616,
      "learning_rate": 0.0001813495081918313,
      "loss": 3.6265,
      "step": 100040
    },
    {
      "epoch": 0.09326278888131379,
      "grad_norm": 0.9390239119529724,
      "learning_rate": 0.00018134764386973612,
      "loss": 3.8229,
      "step": 100050
    },
    {
      "epoch": 0.09327211049939287,
      "grad_norm": 1.0325312614440918,
      "learning_rate": 0.0001813457795476409,
      "loss": 3.786,
      "step": 100060
    },
    {
      "epoch": 0.09328143211747196,
      "grad_norm": 1.0908668041229248,
      "learning_rate": 0.0001813439152255457,
      "loss": 3.7284,
      "step": 100070
    },
    {
      "epoch": 0.09329075373555105,
      "grad_norm": 1.0087535381317139,
      "learning_rate": 0.00018134205090345048,
      "loss": 3.8202,
      "step": 100080
    },
    {
      "epoch": 0.09330007535363015,
      "grad_norm": 1.13985276222229,
      "learning_rate": 0.0001813401865813553,
      "loss": 3.8181,
      "step": 100090
    },
    {
      "epoch": 0.09330939697170924,
      "grad_norm": 1.3430569171905518,
      "learning_rate": 0.00018133832225926012,
      "loss": 3.8577,
      "step": 100100
    },
    {
      "epoch": 0.09331871858978832,
      "grad_norm": 1.2086718082427979,
      "learning_rate": 0.0001813364579371649,
      "loss": 3.3946,
      "step": 100110
    },
    {
      "epoch": 0.09332804020786742,
      "grad_norm": 1.056411862373352,
      "learning_rate": 0.0001813345936150697,
      "loss": 3.6881,
      "step": 100120
    },
    {
      "epoch": 0.09333736182594651,
      "grad_norm": 0.9950345754623413,
      "learning_rate": 0.00018133272929297449,
      "loss": 3.7204,
      "step": 100130
    },
    {
      "epoch": 0.09334668344402561,
      "grad_norm": 0.9808194637298584,
      "learning_rate": 0.0001813308649708793,
      "loss": 3.7431,
      "step": 100140
    },
    {
      "epoch": 0.0933560050621047,
      "grad_norm": 1.0146117210388184,
      "learning_rate": 0.0001813290006487841,
      "loss": 3.7904,
      "step": 100150
    },
    {
      "epoch": 0.09336532668018378,
      "grad_norm": 1.1518322229385376,
      "learning_rate": 0.0001813271363266889,
      "loss": 3.8702,
      "step": 100160
    },
    {
      "epoch": 0.09337464829826288,
      "grad_norm": 1.1876609325408936,
      "learning_rate": 0.0001813252720045937,
      "loss": 3.6773,
      "step": 100170
    },
    {
      "epoch": 0.09338396991634197,
      "grad_norm": 1.0819166898727417,
      "learning_rate": 0.0001813234076824985,
      "loss": 3.6859,
      "step": 100180
    },
    {
      "epoch": 0.09339329153442107,
      "grad_norm": 1.0423015356063843,
      "learning_rate": 0.0001813215433604033,
      "loss": 3.8278,
      "step": 100190
    },
    {
      "epoch": 0.09340261315250016,
      "grad_norm": 0.8811994194984436,
      "learning_rate": 0.0001813196790383081,
      "loss": 3.7803,
      "step": 100200
    },
    {
      "epoch": 0.09341193477057926,
      "grad_norm": 1.0833768844604492,
      "learning_rate": 0.0001813178147162129,
      "loss": 3.5784,
      "step": 100210
    },
    {
      "epoch": 0.09342125638865834,
      "grad_norm": 1.0869442224502563,
      "learning_rate": 0.00018131595039411772,
      "loss": 3.6576,
      "step": 100220
    },
    {
      "epoch": 0.09343057800673743,
      "grad_norm": 1.0031623840332031,
      "learning_rate": 0.0001813140860720225,
      "loss": 3.8163,
      "step": 100230
    },
    {
      "epoch": 0.09343989962481652,
      "grad_norm": 1.0332999229431152,
      "learning_rate": 0.0001813122217499273,
      "loss": 3.865,
      "step": 100240
    },
    {
      "epoch": 0.09344922124289562,
      "grad_norm": 0.9246392846107483,
      "learning_rate": 0.00018131035742783208,
      "loss": 3.7006,
      "step": 100250
    },
    {
      "epoch": 0.09345854286097471,
      "grad_norm": 1.1760451793670654,
      "learning_rate": 0.00018130849310573689,
      "loss": 3.6243,
      "step": 100260
    },
    {
      "epoch": 0.0934678644790538,
      "grad_norm": 0.9467703104019165,
      "learning_rate": 0.00018130662878364172,
      "loss": 3.7842,
      "step": 100270
    },
    {
      "epoch": 0.09347718609713289,
      "grad_norm": 1.117372989654541,
      "learning_rate": 0.0001813047644615465,
      "loss": 3.702,
      "step": 100280
    },
    {
      "epoch": 0.09348650771521198,
      "grad_norm": 0.9675835967063904,
      "learning_rate": 0.0001813029001394513,
      "loss": 3.6896,
      "step": 100290
    },
    {
      "epoch": 0.09349582933329108,
      "grad_norm": 1.1421147584915161,
      "learning_rate": 0.00018130103581735608,
      "loss": 3.7692,
      "step": 100300
    },
    {
      "epoch": 0.09350515095137017,
      "grad_norm": 0.9602410793304443,
      "learning_rate": 0.0001812991714952609,
      "loss": 3.6697,
      "step": 100310
    },
    {
      "epoch": 0.09351447256944925,
      "grad_norm": 1.0582351684570312,
      "learning_rate": 0.0001812973071731657,
      "loss": 3.7884,
      "step": 100320
    },
    {
      "epoch": 0.09352379418752835,
      "grad_norm": 1.0795538425445557,
      "learning_rate": 0.0001812954428510705,
      "loss": 3.7019,
      "step": 100330
    },
    {
      "epoch": 0.09353311580560744,
      "grad_norm": 1.2080409526824951,
      "learning_rate": 0.0001812935785289753,
      "loss": 3.5501,
      "step": 100340
    },
    {
      "epoch": 0.09354243742368654,
      "grad_norm": 1.0698095560073853,
      "learning_rate": 0.0001812917142068801,
      "loss": 3.689,
      "step": 100350
    },
    {
      "epoch": 0.09355175904176563,
      "grad_norm": 1.100395679473877,
      "learning_rate": 0.0001812898498847849,
      "loss": 3.7289,
      "step": 100360
    },
    {
      "epoch": 0.09356108065984471,
      "grad_norm": 1.253350019454956,
      "learning_rate": 0.0001812879855626897,
      "loss": 3.8032,
      "step": 100370
    },
    {
      "epoch": 0.0935704022779238,
      "grad_norm": 1.0446877479553223,
      "learning_rate": 0.0001812861212405945,
      "loss": 3.4248,
      "step": 100380
    },
    {
      "epoch": 0.0935797238960029,
      "grad_norm": 1.0357921123504639,
      "learning_rate": 0.00018128425691849931,
      "loss": 3.7977,
      "step": 100390
    },
    {
      "epoch": 0.093589045514082,
      "grad_norm": 1.086260199546814,
      "learning_rate": 0.0001812823925964041,
      "loss": 3.6483,
      "step": 100400
    },
    {
      "epoch": 0.09359836713216109,
      "grad_norm": 1.017174243927002,
      "learning_rate": 0.0001812805282743089,
      "loss": 3.6281,
      "step": 100410
    },
    {
      "epoch": 0.09360768875024018,
      "grad_norm": 1.1317778825759888,
      "learning_rate": 0.0001812786639522137,
      "loss": 3.6436,
      "step": 100420
    },
    {
      "epoch": 0.09361701036831926,
      "grad_norm": 1.1554248332977295,
      "learning_rate": 0.0001812767996301185,
      "loss": 3.4755,
      "step": 100430
    },
    {
      "epoch": 0.09362633198639836,
      "grad_norm": 1.1898730993270874,
      "learning_rate": 0.00018127493530802332,
      "loss": 3.8592,
      "step": 100440
    },
    {
      "epoch": 0.09363565360447745,
      "grad_norm": 1.0680294036865234,
      "learning_rate": 0.0001812730709859281,
      "loss": 3.5752,
      "step": 100450
    },
    {
      "epoch": 0.09364497522255655,
      "grad_norm": 1.10443115234375,
      "learning_rate": 0.0001812712066638329,
      "loss": 3.7235,
      "step": 100460
    },
    {
      "epoch": 0.09365429684063564,
      "grad_norm": 0.9736878871917725,
      "learning_rate": 0.0001812693423417377,
      "loss": 3.8257,
      "step": 100470
    },
    {
      "epoch": 0.09366361845871472,
      "grad_norm": 1.3818879127502441,
      "learning_rate": 0.0001812674780196425,
      "loss": 3.7419,
      "step": 100480
    },
    {
      "epoch": 0.09367294007679382,
      "grad_norm": 1.0141630172729492,
      "learning_rate": 0.00018126561369754732,
      "loss": 3.6413,
      "step": 100490
    },
    {
      "epoch": 0.09368226169487291,
      "grad_norm": 1.0945868492126465,
      "learning_rate": 0.0001812637493754521,
      "loss": 3.6347,
      "step": 100500
    },
    {
      "epoch": 0.093691583312952,
      "grad_norm": 0.9713839292526245,
      "learning_rate": 0.0001812618850533569,
      "loss": 3.7681,
      "step": 100510
    },
    {
      "epoch": 0.0937009049310311,
      "grad_norm": 1.40013587474823,
      "learning_rate": 0.00018126002073126171,
      "loss": 3.7251,
      "step": 100520
    },
    {
      "epoch": 0.09371022654911018,
      "grad_norm": 1.1832904815673828,
      "learning_rate": 0.0001812581564091665,
      "loss": 3.7592,
      "step": 100530
    },
    {
      "epoch": 0.09371954816718928,
      "grad_norm": 1.0966852903366089,
      "learning_rate": 0.0001812562920870713,
      "loss": 3.6396,
      "step": 100540
    },
    {
      "epoch": 0.09372886978526837,
      "grad_norm": 1.061509609222412,
      "learning_rate": 0.0001812544277649761,
      "loss": 3.8279,
      "step": 100550
    },
    {
      "epoch": 0.09373819140334747,
      "grad_norm": 0.9548684358596802,
      "learning_rate": 0.0001812525634428809,
      "loss": 3.884,
      "step": 100560
    },
    {
      "epoch": 0.09374751302142656,
      "grad_norm": 0.9221204519271851,
      "learning_rate": 0.00018125069912078572,
      "loss": 3.7265,
      "step": 100570
    },
    {
      "epoch": 0.09375683463950564,
      "grad_norm": 1.032770037651062,
      "learning_rate": 0.0001812488347986905,
      "loss": 3.7505,
      "step": 100580
    },
    {
      "epoch": 0.09376615625758473,
      "grad_norm": 1.0543640851974487,
      "learning_rate": 0.0001812469704765953,
      "loss": 3.6799,
      "step": 100590
    },
    {
      "epoch": 0.09377547787566383,
      "grad_norm": 1.2349722385406494,
      "learning_rate": 0.0001812451061545001,
      "loss": 3.8235,
      "step": 100600
    },
    {
      "epoch": 0.09378479949374292,
      "grad_norm": 1.1168386936187744,
      "learning_rate": 0.00018124324183240492,
      "loss": 3.7043,
      "step": 100610
    },
    {
      "epoch": 0.09379412111182202,
      "grad_norm": 1.082301378250122,
      "learning_rate": 0.00018124137751030972,
      "loss": 3.6888,
      "step": 100620
    },
    {
      "epoch": 0.09380344272990111,
      "grad_norm": 1.0980613231658936,
      "learning_rate": 0.0001812395131882145,
      "loss": 3.7631,
      "step": 100630
    },
    {
      "epoch": 0.0938127643479802,
      "grad_norm": 1.0007216930389404,
      "learning_rate": 0.0001812376488661193,
      "loss": 3.912,
      "step": 100640
    },
    {
      "epoch": 0.09382208596605929,
      "grad_norm": 1.1232976913452148,
      "learning_rate": 0.0001812357845440241,
      "loss": 3.7757,
      "step": 100650
    },
    {
      "epoch": 0.09383140758413838,
      "grad_norm": 1.035077452659607,
      "learning_rate": 0.00018123392022192892,
      "loss": 3.694,
      "step": 100660
    },
    {
      "epoch": 0.09384072920221748,
      "grad_norm": 1.0384105443954468,
      "learning_rate": 0.00018123205589983373,
      "loss": 3.6539,
      "step": 100670
    },
    {
      "epoch": 0.09385005082029657,
      "grad_norm": 0.8993772268295288,
      "learning_rate": 0.0001812301915777385,
      "loss": 3.7815,
      "step": 100680
    },
    {
      "epoch": 0.09385937243837565,
      "grad_norm": 0.9716445803642273,
      "learning_rate": 0.0001812283272556433,
      "loss": 3.5642,
      "step": 100690
    },
    {
      "epoch": 0.09386869405645475,
      "grad_norm": 0.9837935566902161,
      "learning_rate": 0.0001812264629335481,
      "loss": 3.5284,
      "step": 100700
    },
    {
      "epoch": 0.09387801567453384,
      "grad_norm": 1.0229346752166748,
      "learning_rate": 0.0001812245986114529,
      "loss": 3.9127,
      "step": 100710
    },
    {
      "epoch": 0.09388733729261294,
      "grad_norm": 1.0447509288787842,
      "learning_rate": 0.00018122273428935773,
      "loss": 3.8154,
      "step": 100720
    },
    {
      "epoch": 0.09389665891069203,
      "grad_norm": 1.0343669652938843,
      "learning_rate": 0.0001812208699672625,
      "loss": 3.6645,
      "step": 100730
    },
    {
      "epoch": 0.09390598052877111,
      "grad_norm": 0.9842157363891602,
      "learning_rate": 0.00018121900564516732,
      "loss": 3.7811,
      "step": 100740
    },
    {
      "epoch": 0.0939153021468502,
      "grad_norm": 1.1037791967391968,
      "learning_rate": 0.0001812171413230721,
      "loss": 3.6531,
      "step": 100750
    },
    {
      "epoch": 0.0939246237649293,
      "grad_norm": 1.151695966720581,
      "learning_rate": 0.0001812152770009769,
      "loss": 3.7543,
      "step": 100760
    },
    {
      "epoch": 0.0939339453830084,
      "grad_norm": 0.8551942706108093,
      "learning_rate": 0.0001812134126788817,
      "loss": 3.7503,
      "step": 100770
    },
    {
      "epoch": 0.09394326700108749,
      "grad_norm": 0.9773989319801331,
      "learning_rate": 0.00018121154835678652,
      "loss": 3.6862,
      "step": 100780
    },
    {
      "epoch": 0.09395258861916657,
      "grad_norm": 1.0859761238098145,
      "learning_rate": 0.00018120968403469132,
      "loss": 3.6853,
      "step": 100790
    },
    {
      "epoch": 0.09396191023724566,
      "grad_norm": 1.0100661516189575,
      "learning_rate": 0.0001812078197125961,
      "loss": 3.8273,
      "step": 100800
    },
    {
      "epoch": 0.09397123185532476,
      "grad_norm": 1.1006861925125122,
      "learning_rate": 0.0001812059553905009,
      "loss": 3.7096,
      "step": 100810
    },
    {
      "epoch": 0.09398055347340385,
      "grad_norm": 0.9871914386749268,
      "learning_rate": 0.00018120409106840571,
      "loss": 3.6052,
      "step": 100820
    },
    {
      "epoch": 0.09398987509148295,
      "grad_norm": 1.1581859588623047,
      "learning_rate": 0.00018120222674631052,
      "loss": 3.7557,
      "step": 100830
    },
    {
      "epoch": 0.09399919670956203,
      "grad_norm": 1.099839687347412,
      "learning_rate": 0.00018120036242421533,
      "loss": 3.6184,
      "step": 100840
    },
    {
      "epoch": 0.09400851832764112,
      "grad_norm": 1.0413182973861694,
      "learning_rate": 0.0001811984981021201,
      "loss": 3.5876,
      "step": 100850
    },
    {
      "epoch": 0.09401783994572022,
      "grad_norm": 1.2525391578674316,
      "learning_rate": 0.0001811966337800249,
      "loss": 3.7617,
      "step": 100860
    },
    {
      "epoch": 0.09402716156379931,
      "grad_norm": 1.0151983499526978,
      "learning_rate": 0.00018119476945792972,
      "loss": 3.7886,
      "step": 100870
    },
    {
      "epoch": 0.0940364831818784,
      "grad_norm": 1.299339771270752,
      "learning_rate": 0.00018119290513583452,
      "loss": 3.795,
      "step": 100880
    },
    {
      "epoch": 0.0940458047999575,
      "grad_norm": 1.1010276079177856,
      "learning_rate": 0.00018119104081373933,
      "loss": 3.8262,
      "step": 100890
    },
    {
      "epoch": 0.09405512641803658,
      "grad_norm": 1.0943019390106201,
      "learning_rate": 0.0001811891764916441,
      "loss": 3.5373,
      "step": 100900
    },
    {
      "epoch": 0.09406444803611567,
      "grad_norm": 0.9815253019332886,
      "learning_rate": 0.00018118731216954892,
      "loss": 3.7544,
      "step": 100910
    },
    {
      "epoch": 0.09407376965419477,
      "grad_norm": 1.2688872814178467,
      "learning_rate": 0.00018118544784745372,
      "loss": 3.5774,
      "step": 100920
    },
    {
      "epoch": 0.09408309127227386,
      "grad_norm": 0.9089839458465576,
      "learning_rate": 0.0001811835835253585,
      "loss": 3.8816,
      "step": 100930
    },
    {
      "epoch": 0.09409241289035296,
      "grad_norm": 1.1567142009735107,
      "learning_rate": 0.00018118171920326334,
      "loss": 3.7908,
      "step": 100940
    },
    {
      "epoch": 0.09410173450843204,
      "grad_norm": 1.1198341846466064,
      "learning_rate": 0.00018117985488116811,
      "loss": 3.6961,
      "step": 100950
    },
    {
      "epoch": 0.09411105612651113,
      "grad_norm": 1.1183339357376099,
      "learning_rate": 0.00018117799055907292,
      "loss": 3.6621,
      "step": 100960
    },
    {
      "epoch": 0.09412037774459023,
      "grad_norm": 0.9450920224189758,
      "learning_rate": 0.00018117612623697773,
      "loss": 3.4881,
      "step": 100970
    },
    {
      "epoch": 0.09412969936266932,
      "grad_norm": 1.0629671812057495,
      "learning_rate": 0.0001811742619148825,
      "loss": 3.7618,
      "step": 100980
    },
    {
      "epoch": 0.09413902098074842,
      "grad_norm": 1.1137375831604004,
      "learning_rate": 0.0001811723975927873,
      "loss": 3.7591,
      "step": 100990
    },
    {
      "epoch": 0.0941483425988275,
      "grad_norm": 1.0810186862945557,
      "learning_rate": 0.00018117053327069212,
      "loss": 3.6492,
      "step": 101000
    },
    {
      "epoch": 0.09415766421690659,
      "grad_norm": 1.14972722530365,
      "learning_rate": 0.00018116866894859693,
      "loss": 3.6559,
      "step": 101010
    },
    {
      "epoch": 0.09416698583498569,
      "grad_norm": 1.0027868747711182,
      "learning_rate": 0.00018116680462650173,
      "loss": 3.5704,
      "step": 101020
    },
    {
      "epoch": 0.09417630745306478,
      "grad_norm": 1.068345069885254,
      "learning_rate": 0.0001811649403044065,
      "loss": 3.8051,
      "step": 101030
    },
    {
      "epoch": 0.09418562907114388,
      "grad_norm": 1.2087156772613525,
      "learning_rate": 0.00018116307598231132,
      "loss": 3.6878,
      "step": 101040
    },
    {
      "epoch": 0.09419495068922296,
      "grad_norm": 0.9169450998306274,
      "learning_rate": 0.00018116121166021612,
      "loss": 3.7279,
      "step": 101050
    },
    {
      "epoch": 0.09420427230730205,
      "grad_norm": 1.2862696647644043,
      "learning_rate": 0.00018115934733812093,
      "loss": 3.6662,
      "step": 101060
    },
    {
      "epoch": 0.09421359392538114,
      "grad_norm": 1.0845117568969727,
      "learning_rate": 0.00018115748301602574,
      "loss": 3.8024,
      "step": 101070
    },
    {
      "epoch": 0.09422291554346024,
      "grad_norm": 1.082814335823059,
      "learning_rate": 0.00018115561869393052,
      "loss": 3.6183,
      "step": 101080
    },
    {
      "epoch": 0.09423223716153933,
      "grad_norm": 0.965110182762146,
      "learning_rate": 0.00018115375437183532,
      "loss": 3.8276,
      "step": 101090
    },
    {
      "epoch": 0.09424155877961843,
      "grad_norm": 1.049383521080017,
      "learning_rate": 0.0001811518900497401,
      "loss": 3.7437,
      "step": 101100
    },
    {
      "epoch": 0.09425088039769751,
      "grad_norm": 1.007385015487671,
      "learning_rate": 0.00018115002572764493,
      "loss": 3.6274,
      "step": 101110
    },
    {
      "epoch": 0.0942602020157766,
      "grad_norm": 1.136594533920288,
      "learning_rate": 0.00018114816140554974,
      "loss": 3.8815,
      "step": 101120
    },
    {
      "epoch": 0.0942695236338557,
      "grad_norm": 1.1209442615509033,
      "learning_rate": 0.00018114629708345452,
      "loss": 3.5028,
      "step": 101130
    },
    {
      "epoch": 0.09427884525193479,
      "grad_norm": 0.9759752750396729,
      "learning_rate": 0.00018114443276135933,
      "loss": 3.7777,
      "step": 101140
    },
    {
      "epoch": 0.09428816687001389,
      "grad_norm": 0.9269245862960815,
      "learning_rate": 0.0001811425684392641,
      "loss": 3.594,
      "step": 101150
    },
    {
      "epoch": 0.09429748848809297,
      "grad_norm": 1.4108097553253174,
      "learning_rate": 0.0001811407041171689,
      "loss": 3.5315,
      "step": 101160
    },
    {
      "epoch": 0.09430681010617206,
      "grad_norm": 1.199602484703064,
      "learning_rate": 0.00018113883979507375,
      "loss": 3.4281,
      "step": 101170
    },
    {
      "epoch": 0.09431613172425116,
      "grad_norm": 0.9514904618263245,
      "learning_rate": 0.00018113697547297852,
      "loss": 3.7899,
      "step": 101180
    },
    {
      "epoch": 0.09432545334233025,
      "grad_norm": 0.9456679224967957,
      "learning_rate": 0.00018113511115088333,
      "loss": 3.6393,
      "step": 101190
    },
    {
      "epoch": 0.09433477496040935,
      "grad_norm": 1.1852093935012817,
      "learning_rate": 0.0001811332468287881,
      "loss": 3.4932,
      "step": 101200
    },
    {
      "epoch": 0.09434409657848843,
      "grad_norm": 1.119893193244934,
      "learning_rate": 0.00018113138250669292,
      "loss": 3.778,
      "step": 101210
    },
    {
      "epoch": 0.09435341819656752,
      "grad_norm": 0.9489360451698303,
      "learning_rate": 0.00018112951818459772,
      "loss": 3.8309,
      "step": 101220
    },
    {
      "epoch": 0.09436273981464662,
      "grad_norm": 0.862987220287323,
      "learning_rate": 0.00018112765386250253,
      "loss": 3.7131,
      "step": 101230
    },
    {
      "epoch": 0.09437206143272571,
      "grad_norm": 0.9001147150993347,
      "learning_rate": 0.00018112578954040734,
      "loss": 3.7948,
      "step": 101240
    },
    {
      "epoch": 0.0943813830508048,
      "grad_norm": 0.9565033316612244,
      "learning_rate": 0.00018112392521831211,
      "loss": 3.7898,
      "step": 101250
    },
    {
      "epoch": 0.09439070466888388,
      "grad_norm": 1.505421757698059,
      "learning_rate": 0.00018112206089621692,
      "loss": 3.8118,
      "step": 101260
    },
    {
      "epoch": 0.09440002628696298,
      "grad_norm": 1.0799728631973267,
      "learning_rate": 0.00018112019657412173,
      "loss": 3.791,
      "step": 101270
    },
    {
      "epoch": 0.09440934790504207,
      "grad_norm": 1.1001086235046387,
      "learning_rate": 0.00018111833225202653,
      "loss": 3.5813,
      "step": 101280
    },
    {
      "epoch": 0.09441866952312117,
      "grad_norm": 1.147371768951416,
      "learning_rate": 0.00018111646792993134,
      "loss": 3.7893,
      "step": 101290
    },
    {
      "epoch": 0.09442799114120026,
      "grad_norm": 1.0571708679199219,
      "learning_rate": 0.00018111460360783612,
      "loss": 3.7215,
      "step": 101300
    },
    {
      "epoch": 0.09443731275927936,
      "grad_norm": 1.3235090970993042,
      "learning_rate": 0.00018111273928574093,
      "loss": 3.953,
      "step": 101310
    },
    {
      "epoch": 0.09444663437735844,
      "grad_norm": 1.2832828760147095,
      "learning_rate": 0.00018111087496364573,
      "loss": 3.7209,
      "step": 101320
    },
    {
      "epoch": 0.09445595599543753,
      "grad_norm": 1.1208547353744507,
      "learning_rate": 0.00018110901064155054,
      "loss": 3.5868,
      "step": 101330
    },
    {
      "epoch": 0.09446527761351663,
      "grad_norm": 1.0318108797073364,
      "learning_rate": 0.00018110714631945534,
      "loss": 3.7555,
      "step": 101340
    },
    {
      "epoch": 0.09447459923159572,
      "grad_norm": 1.0903537273406982,
      "learning_rate": 0.00018110528199736012,
      "loss": 3.7496,
      "step": 101350
    },
    {
      "epoch": 0.09448392084967482,
      "grad_norm": 1.0678017139434814,
      "learning_rate": 0.00018110341767526493,
      "loss": 3.7269,
      "step": 101360
    },
    {
      "epoch": 0.0944932424677539,
      "grad_norm": 1.1310887336730957,
      "learning_rate": 0.00018110155335316974,
      "loss": 3.7189,
      "step": 101370
    },
    {
      "epoch": 0.09450256408583299,
      "grad_norm": 1.280430555343628,
      "learning_rate": 0.00018109968903107452,
      "loss": 3.9322,
      "step": 101380
    },
    {
      "epoch": 0.09451188570391209,
      "grad_norm": 1.0263763666152954,
      "learning_rate": 0.00018109782470897935,
      "loss": 3.5942,
      "step": 101390
    },
    {
      "epoch": 0.09452120732199118,
      "grad_norm": 1.0040602684020996,
      "learning_rate": 0.00018109596038688413,
      "loss": 3.738,
      "step": 101400
    },
    {
      "epoch": 0.09453052894007027,
      "grad_norm": 0.9423338770866394,
      "learning_rate": 0.00018109409606478893,
      "loss": 3.8729,
      "step": 101410
    },
    {
      "epoch": 0.09453985055814935,
      "grad_norm": 1.0858410596847534,
      "learning_rate": 0.00018109223174269374,
      "loss": 3.5626,
      "step": 101420
    },
    {
      "epoch": 0.09454917217622845,
      "grad_norm": 1.126111626625061,
      "learning_rate": 0.00018109036742059852,
      "loss": 3.6058,
      "step": 101430
    },
    {
      "epoch": 0.09455849379430754,
      "grad_norm": 1.0947479009628296,
      "learning_rate": 0.00018108850309850333,
      "loss": 3.8184,
      "step": 101440
    },
    {
      "epoch": 0.09456781541238664,
      "grad_norm": 0.9536963105201721,
      "learning_rate": 0.00018108663877640813,
      "loss": 3.8572,
      "step": 101450
    },
    {
      "epoch": 0.09457713703046573,
      "grad_norm": 0.9488394856452942,
      "learning_rate": 0.00018108477445431294,
      "loss": 3.7688,
      "step": 101460
    },
    {
      "epoch": 0.09458645864854481,
      "grad_norm": 1.069977045059204,
      "learning_rate": 0.00018108291013221774,
      "loss": 3.5058,
      "step": 101470
    },
    {
      "epoch": 0.09459578026662391,
      "grad_norm": 1.055761456489563,
      "learning_rate": 0.00018108104581012252,
      "loss": 3.6268,
      "step": 101480
    },
    {
      "epoch": 0.094605101884703,
      "grad_norm": 1.0806218385696411,
      "learning_rate": 0.00018107918148802733,
      "loss": 3.8602,
      "step": 101490
    },
    {
      "epoch": 0.0946144235027821,
      "grad_norm": 0.8171321749687195,
      "learning_rate": 0.00018107731716593214,
      "loss": 3.7726,
      "step": 101500
    },
    {
      "epoch": 0.09462374512086119,
      "grad_norm": 1.1274913549423218,
      "learning_rate": 0.00018107545284383694,
      "loss": 3.8382,
      "step": 101510
    },
    {
      "epoch": 0.09463306673894029,
      "grad_norm": 0.9794861674308777,
      "learning_rate": 0.00018107358852174175,
      "loss": 3.8237,
      "step": 101520
    },
    {
      "epoch": 0.09464238835701937,
      "grad_norm": 1.0209879875183105,
      "learning_rate": 0.00018107172419964653,
      "loss": 3.791,
      "step": 101530
    },
    {
      "epoch": 0.09465170997509846,
      "grad_norm": 0.903498113155365,
      "learning_rate": 0.00018106985987755133,
      "loss": 3.8232,
      "step": 101540
    },
    {
      "epoch": 0.09466103159317756,
      "grad_norm": 1.1155445575714111,
      "learning_rate": 0.00018106799555545611,
      "loss": 3.646,
      "step": 101550
    },
    {
      "epoch": 0.09467035321125665,
      "grad_norm": 0.856492280960083,
      "learning_rate": 0.00018106613123336095,
      "loss": 3.7124,
      "step": 101560
    },
    {
      "epoch": 0.09467967482933574,
      "grad_norm": 0.9857640862464905,
      "learning_rate": 0.00018106426691126575,
      "loss": 3.6454,
      "step": 101570
    },
    {
      "epoch": 0.09468899644741482,
      "grad_norm": 1.2923280000686646,
      "learning_rate": 0.00018106240258917053,
      "loss": 3.8219,
      "step": 101580
    },
    {
      "epoch": 0.09469831806549392,
      "grad_norm": 0.872489333152771,
      "learning_rate": 0.00018106053826707534,
      "loss": 3.8287,
      "step": 101590
    },
    {
      "epoch": 0.09470763968357301,
      "grad_norm": 1.0007390975952148,
      "learning_rate": 0.00018105867394498012,
      "loss": 3.9057,
      "step": 101600
    },
    {
      "epoch": 0.09471696130165211,
      "grad_norm": 1.188722014427185,
      "learning_rate": 0.00018105680962288492,
      "loss": 3.6294,
      "step": 101610
    },
    {
      "epoch": 0.0947262829197312,
      "grad_norm": 1.0411831140518188,
      "learning_rate": 0.00018105494530078973,
      "loss": 3.4941,
      "step": 101620
    },
    {
      "epoch": 0.09473560453781028,
      "grad_norm": 1.0542887449264526,
      "learning_rate": 0.00018105308097869454,
      "loss": 3.654,
      "step": 101630
    },
    {
      "epoch": 0.09474492615588938,
      "grad_norm": 0.8426711559295654,
      "learning_rate": 0.00018105121665659934,
      "loss": 3.7508,
      "step": 101640
    },
    {
      "epoch": 0.09475424777396847,
      "grad_norm": 1.0384173393249512,
      "learning_rate": 0.00018104935233450412,
      "loss": 3.7188,
      "step": 101650
    },
    {
      "epoch": 0.09476356939204757,
      "grad_norm": 1.115477442741394,
      "learning_rate": 0.00018104748801240893,
      "loss": 3.6735,
      "step": 101660
    },
    {
      "epoch": 0.09477289101012666,
      "grad_norm": 1.010347843170166,
      "learning_rate": 0.00018104562369031374,
      "loss": 3.7808,
      "step": 101670
    },
    {
      "epoch": 0.09478221262820574,
      "grad_norm": 1.165307879447937,
      "learning_rate": 0.00018104375936821854,
      "loss": 3.9006,
      "step": 101680
    },
    {
      "epoch": 0.09479153424628484,
      "grad_norm": 0.996258556842804,
      "learning_rate": 0.00018104189504612335,
      "loss": 3.4379,
      "step": 101690
    },
    {
      "epoch": 0.09480085586436393,
      "grad_norm": 1.1611883640289307,
      "learning_rate": 0.00018104003072402813,
      "loss": 3.711,
      "step": 101700
    },
    {
      "epoch": 0.09481017748244303,
      "grad_norm": 1.0231870412826538,
      "learning_rate": 0.00018103816640193293,
      "loss": 3.7471,
      "step": 101710
    },
    {
      "epoch": 0.09481949910052212,
      "grad_norm": 1.1694614887237549,
      "learning_rate": 0.00018103630207983774,
      "loss": 3.8154,
      "step": 101720
    },
    {
      "epoch": 0.0948288207186012,
      "grad_norm": 1.2523738145828247,
      "learning_rate": 0.00018103443775774255,
      "loss": 3.6799,
      "step": 101730
    },
    {
      "epoch": 0.0948381423366803,
      "grad_norm": 1.2218815088272095,
      "learning_rate": 0.00018103257343564735,
      "loss": 3.9549,
      "step": 101740
    },
    {
      "epoch": 0.09484746395475939,
      "grad_norm": 1.182456612586975,
      "learning_rate": 0.00018103070911355213,
      "loss": 3.6458,
      "step": 101750
    },
    {
      "epoch": 0.09485678557283848,
      "grad_norm": 1.1126351356506348,
      "learning_rate": 0.00018102884479145694,
      "loss": 3.7553,
      "step": 101760
    },
    {
      "epoch": 0.09486610719091758,
      "grad_norm": 1.1379151344299316,
      "learning_rate": 0.00018102698046936172,
      "loss": 3.5899,
      "step": 101770
    },
    {
      "epoch": 0.09487542880899667,
      "grad_norm": 1.1189569234848022,
      "learning_rate": 0.00018102511614726655,
      "loss": 3.7832,
      "step": 101780
    },
    {
      "epoch": 0.09488475042707575,
      "grad_norm": 1.0355172157287598,
      "learning_rate": 0.00018102325182517136,
      "loss": 3.7656,
      "step": 101790
    },
    {
      "epoch": 0.09489407204515485,
      "grad_norm": 1.3148828744888306,
      "learning_rate": 0.00018102138750307614,
      "loss": 3.7479,
      "step": 101800
    },
    {
      "epoch": 0.09490339366323394,
      "grad_norm": 1.137589693069458,
      "learning_rate": 0.00018101952318098094,
      "loss": 3.7129,
      "step": 101810
    },
    {
      "epoch": 0.09491271528131304,
      "grad_norm": 1.134440302848816,
      "learning_rate": 0.00018101765885888572,
      "loss": 3.6057,
      "step": 101820
    },
    {
      "epoch": 0.09492203689939213,
      "grad_norm": 1.0162569284439087,
      "learning_rate": 0.00018101579453679053,
      "loss": 3.6868,
      "step": 101830
    },
    {
      "epoch": 0.09493135851747121,
      "grad_norm": 1.0664560794830322,
      "learning_rate": 0.00018101393021469536,
      "loss": 3.7339,
      "step": 101840
    },
    {
      "epoch": 0.0949406801355503,
      "grad_norm": 1.0164031982421875,
      "learning_rate": 0.00018101206589260014,
      "loss": 3.6138,
      "step": 101850
    },
    {
      "epoch": 0.0949500017536294,
      "grad_norm": 0.9916015267372131,
      "learning_rate": 0.00018101020157050495,
      "loss": 3.7295,
      "step": 101860
    },
    {
      "epoch": 0.0949593233717085,
      "grad_norm": 1.0188051462173462,
      "learning_rate": 0.00018100833724840973,
      "loss": 3.9557,
      "step": 101870
    },
    {
      "epoch": 0.09496864498978759,
      "grad_norm": 0.9578754305839539,
      "learning_rate": 0.00018100647292631453,
      "loss": 3.8577,
      "step": 101880
    },
    {
      "epoch": 0.09497796660786667,
      "grad_norm": 1.087977409362793,
      "learning_rate": 0.00018100460860421934,
      "loss": 3.6259,
      "step": 101890
    },
    {
      "epoch": 0.09498728822594577,
      "grad_norm": 0.9805278182029724,
      "learning_rate": 0.00018100274428212414,
      "loss": 3.7386,
      "step": 101900
    },
    {
      "epoch": 0.09499660984402486,
      "grad_norm": 1.0669893026351929,
      "learning_rate": 0.00018100087996002895,
      "loss": 3.6842,
      "step": 101910
    },
    {
      "epoch": 0.09500593146210395,
      "grad_norm": 0.9069686532020569,
      "learning_rate": 0.00018099901563793373,
      "loss": 3.5323,
      "step": 101920
    },
    {
      "epoch": 0.09501525308018305,
      "grad_norm": 1.0947707891464233,
      "learning_rate": 0.00018099715131583854,
      "loss": 3.7513,
      "step": 101930
    },
    {
      "epoch": 0.09502457469826213,
      "grad_norm": 0.9801051020622253,
      "learning_rate": 0.00018099528699374334,
      "loss": 3.9756,
      "step": 101940
    },
    {
      "epoch": 0.09503389631634122,
      "grad_norm": 1.0942362546920776,
      "learning_rate": 0.00018099342267164815,
      "loss": 3.5872,
      "step": 101950
    },
    {
      "epoch": 0.09504321793442032,
      "grad_norm": 1.1685435771942139,
      "learning_rate": 0.00018099155834955296,
      "loss": 3.7059,
      "step": 101960
    },
    {
      "epoch": 0.09505253955249941,
      "grad_norm": 0.9905388355255127,
      "learning_rate": 0.00018098969402745773,
      "loss": 3.9344,
      "step": 101970
    },
    {
      "epoch": 0.09506186117057851,
      "grad_norm": 1.0511754751205444,
      "learning_rate": 0.00018098782970536254,
      "loss": 3.6526,
      "step": 101980
    },
    {
      "epoch": 0.0950711827886576,
      "grad_norm": 1.2304891347885132,
      "learning_rate": 0.00018098596538326735,
      "loss": 3.688,
      "step": 101990
    },
    {
      "epoch": 0.09508050440673668,
      "grad_norm": 1.2017134428024292,
      "learning_rate": 0.00018098410106117213,
      "loss": 3.7558,
      "step": 102000
    },
    {
      "epoch": 0.09508982602481578,
      "grad_norm": 0.8823997974395752,
      "learning_rate": 0.00018098223673907696,
      "loss": 3.66,
      "step": 102010
    },
    {
      "epoch": 0.09509914764289487,
      "grad_norm": 0.9470906853675842,
      "learning_rate": 0.00018098037241698174,
      "loss": 3.7519,
      "step": 102020
    },
    {
      "epoch": 0.09510846926097397,
      "grad_norm": 1.1077678203582764,
      "learning_rate": 0.00018097850809488655,
      "loss": 3.7659,
      "step": 102030
    },
    {
      "epoch": 0.09511779087905306,
      "grad_norm": 1.3190867900848389,
      "learning_rate": 0.00018097664377279135,
      "loss": 3.5085,
      "step": 102040
    },
    {
      "epoch": 0.09512711249713214,
      "grad_norm": 1.0146849155426025,
      "learning_rate": 0.00018097477945069613,
      "loss": 3.4615,
      "step": 102050
    },
    {
      "epoch": 0.09513643411521124,
      "grad_norm": 1.1947722434997559,
      "learning_rate": 0.00018097291512860094,
      "loss": 3.7127,
      "step": 102060
    },
    {
      "epoch": 0.09514575573329033,
      "grad_norm": 1.092546820640564,
      "learning_rate": 0.00018097105080650574,
      "loss": 3.5658,
      "step": 102070
    },
    {
      "epoch": 0.09515507735136942,
      "grad_norm": 1.1795434951782227,
      "learning_rate": 0.00018096918648441055,
      "loss": 3.7786,
      "step": 102080
    },
    {
      "epoch": 0.09516439896944852,
      "grad_norm": 1.0411415100097656,
      "learning_rate": 0.00018096732216231536,
      "loss": 3.7225,
      "step": 102090
    },
    {
      "epoch": 0.0951737205875276,
      "grad_norm": 1.234616994857788,
      "learning_rate": 0.00018096545784022014,
      "loss": 3.7696,
      "step": 102100
    },
    {
      "epoch": 0.0951830422056067,
      "grad_norm": 1.0872442722320557,
      "learning_rate": 0.00018096359351812494,
      "loss": 3.6198,
      "step": 102110
    },
    {
      "epoch": 0.09519236382368579,
      "grad_norm": 1.1056033372879028,
      "learning_rate": 0.00018096172919602975,
      "loss": 3.7455,
      "step": 102120
    },
    {
      "epoch": 0.09520168544176488,
      "grad_norm": 1.210209608078003,
      "learning_rate": 0.00018095986487393455,
      "loss": 3.728,
      "step": 102130
    },
    {
      "epoch": 0.09521100705984398,
      "grad_norm": 1.1667577028274536,
      "learning_rate": 0.00018095800055183936,
      "loss": 3.804,
      "step": 102140
    },
    {
      "epoch": 0.09522032867792306,
      "grad_norm": 1.0600264072418213,
      "learning_rate": 0.00018095613622974414,
      "loss": 3.4649,
      "step": 102150
    },
    {
      "epoch": 0.09522965029600215,
      "grad_norm": 1.0680421590805054,
      "learning_rate": 0.00018095427190764895,
      "loss": 3.8232,
      "step": 102160
    },
    {
      "epoch": 0.09523897191408125,
      "grad_norm": 1.0149866342544556,
      "learning_rate": 0.00018095240758555375,
      "loss": 3.7687,
      "step": 102170
    },
    {
      "epoch": 0.09524829353216034,
      "grad_norm": 1.0954993963241577,
      "learning_rate": 0.00018095054326345856,
      "loss": 3.6233,
      "step": 102180
    },
    {
      "epoch": 0.09525761515023944,
      "grad_norm": 1.1119799613952637,
      "learning_rate": 0.00018094867894136337,
      "loss": 3.7276,
      "step": 102190
    },
    {
      "epoch": 0.09526693676831853,
      "grad_norm": 1.0659387111663818,
      "learning_rate": 0.00018094681461926814,
      "loss": 3.6079,
      "step": 102200
    },
    {
      "epoch": 0.09527625838639761,
      "grad_norm": 1.091385006904602,
      "learning_rate": 0.00018094495029717295,
      "loss": 3.7735,
      "step": 102210
    },
    {
      "epoch": 0.0952855800044767,
      "grad_norm": 1.180280089378357,
      "learning_rate": 0.00018094308597507773,
      "loss": 3.6773,
      "step": 102220
    },
    {
      "epoch": 0.0952949016225558,
      "grad_norm": 1.1595547199249268,
      "learning_rate": 0.00018094122165298256,
      "loss": 3.7006,
      "step": 102230
    },
    {
      "epoch": 0.0953042232406349,
      "grad_norm": 1.0639387369155884,
      "learning_rate": 0.00018093935733088737,
      "loss": 3.5944,
      "step": 102240
    },
    {
      "epoch": 0.09531354485871399,
      "grad_norm": 1.135607361793518,
      "learning_rate": 0.00018093749300879215,
      "loss": 3.5632,
      "step": 102250
    },
    {
      "epoch": 0.09532286647679307,
      "grad_norm": 0.8942543268203735,
      "learning_rate": 0.00018093562868669696,
      "loss": 3.8964,
      "step": 102260
    },
    {
      "epoch": 0.09533218809487216,
      "grad_norm": 0.8982799649238586,
      "learning_rate": 0.00018093376436460173,
      "loss": 3.7873,
      "step": 102270
    },
    {
      "epoch": 0.09534150971295126,
      "grad_norm": 1.2234348058700562,
      "learning_rate": 0.00018093190004250654,
      "loss": 3.59,
      "step": 102280
    },
    {
      "epoch": 0.09535083133103035,
      "grad_norm": 1.04619562625885,
      "learning_rate": 0.00018093003572041137,
      "loss": 3.6214,
      "step": 102290
    },
    {
      "epoch": 0.09536015294910945,
      "grad_norm": 1.032801866531372,
      "learning_rate": 0.00018092817139831615,
      "loss": 3.9109,
      "step": 102300
    },
    {
      "epoch": 0.09536947456718853,
      "grad_norm": 1.1251972913742065,
      "learning_rate": 0.00018092630707622096,
      "loss": 3.8401,
      "step": 102310
    },
    {
      "epoch": 0.09537879618526762,
      "grad_norm": 1.0467249155044556,
      "learning_rate": 0.00018092444275412574,
      "loss": 3.5469,
      "step": 102320
    },
    {
      "epoch": 0.09538811780334672,
      "grad_norm": 1.0739343166351318,
      "learning_rate": 0.00018092257843203055,
      "loss": 3.5289,
      "step": 102330
    },
    {
      "epoch": 0.09539743942142581,
      "grad_norm": 1.0207102298736572,
      "learning_rate": 0.00018092071410993535,
      "loss": 3.9171,
      "step": 102340
    },
    {
      "epoch": 0.0954067610395049,
      "grad_norm": 1.0134527683258057,
      "learning_rate": 0.00018091884978784016,
      "loss": 3.5599,
      "step": 102350
    },
    {
      "epoch": 0.09541608265758399,
      "grad_norm": 1.2242095470428467,
      "learning_rate": 0.00018091698546574496,
      "loss": 3.6795,
      "step": 102360
    },
    {
      "epoch": 0.09542540427566308,
      "grad_norm": 1.114272952079773,
      "learning_rate": 0.00018091512114364974,
      "loss": 3.6894,
      "step": 102370
    },
    {
      "epoch": 0.09543472589374218,
      "grad_norm": 1.0855746269226074,
      "learning_rate": 0.00018091325682155455,
      "loss": 3.5879,
      "step": 102380
    },
    {
      "epoch": 0.09544404751182127,
      "grad_norm": 1.0778402090072632,
      "learning_rate": 0.00018091139249945936,
      "loss": 3.6936,
      "step": 102390
    },
    {
      "epoch": 0.09545336912990036,
      "grad_norm": 1.0614439249038696,
      "learning_rate": 0.00018090952817736416,
      "loss": 3.6885,
      "step": 102400
    },
    {
      "epoch": 0.09546269074797945,
      "grad_norm": 1.0930390357971191,
      "learning_rate": 0.00018090766385526897,
      "loss": 3.78,
      "step": 102410
    },
    {
      "epoch": 0.09547201236605854,
      "grad_norm": 1.0706380605697632,
      "learning_rate": 0.00018090579953317375,
      "loss": 3.7976,
      "step": 102420
    },
    {
      "epoch": 0.09548133398413763,
      "grad_norm": 1.1528135538101196,
      "learning_rate": 0.00018090393521107855,
      "loss": 3.8943,
      "step": 102430
    },
    {
      "epoch": 0.09549065560221673,
      "grad_norm": 0.9480718970298767,
      "learning_rate": 0.00018090207088898336,
      "loss": 3.9287,
      "step": 102440
    },
    {
      "epoch": 0.09549997722029582,
      "grad_norm": 1.084897518157959,
      "learning_rate": 0.00018090020656688814,
      "loss": 3.6569,
      "step": 102450
    },
    {
      "epoch": 0.09550929883837492,
      "grad_norm": 0.9151461124420166,
      "learning_rate": 0.00018089834224479297,
      "loss": 3.8361,
      "step": 102460
    },
    {
      "epoch": 0.095518620456454,
      "grad_norm": 1.3364312648773193,
      "learning_rate": 0.00018089647792269775,
      "loss": 3.7177,
      "step": 102470
    },
    {
      "epoch": 0.09552794207453309,
      "grad_norm": 0.8767458200454712,
      "learning_rate": 0.00018089461360060256,
      "loss": 3.7779,
      "step": 102480
    },
    {
      "epoch": 0.09553726369261219,
      "grad_norm": 1.010414481163025,
      "learning_rate": 0.00018089274927850736,
      "loss": 3.7706,
      "step": 102490
    },
    {
      "epoch": 0.09554658531069128,
      "grad_norm": 1.051476240158081,
      "learning_rate": 0.00018089088495641214,
      "loss": 3.675,
      "step": 102500
    },
    {
      "epoch": 0.09555590692877038,
      "grad_norm": 1.285749077796936,
      "learning_rate": 0.00018088902063431695,
      "loss": 3.7385,
      "step": 102510
    },
    {
      "epoch": 0.09556522854684946,
      "grad_norm": 0.9602168798446655,
      "learning_rate": 0.00018088715631222176,
      "loss": 3.7924,
      "step": 102520
    },
    {
      "epoch": 0.09557455016492855,
      "grad_norm": 1.2740874290466309,
      "learning_rate": 0.00018088529199012656,
      "loss": 3.5216,
      "step": 102530
    },
    {
      "epoch": 0.09558387178300765,
      "grad_norm": 0.9348351359367371,
      "learning_rate": 0.00018088342766803137,
      "loss": 3.593,
      "step": 102540
    },
    {
      "epoch": 0.09559319340108674,
      "grad_norm": 0.9423388242721558,
      "learning_rate": 0.00018088156334593615,
      "loss": 3.8071,
      "step": 102550
    },
    {
      "epoch": 0.09560251501916583,
      "grad_norm": 1.1662547588348389,
      "learning_rate": 0.00018087969902384095,
      "loss": 3.7956,
      "step": 102560
    },
    {
      "epoch": 0.09561183663724492,
      "grad_norm": 0.9323167204856873,
      "learning_rate": 0.00018087783470174576,
      "loss": 3.7632,
      "step": 102570
    },
    {
      "epoch": 0.09562115825532401,
      "grad_norm": 0.9825250506401062,
      "learning_rate": 0.00018087597037965057,
      "loss": 3.773,
      "step": 102580
    },
    {
      "epoch": 0.0956304798734031,
      "grad_norm": 0.9233802556991577,
      "learning_rate": 0.00018087410605755537,
      "loss": 3.8641,
      "step": 102590
    },
    {
      "epoch": 0.0956398014914822,
      "grad_norm": 0.9704379439353943,
      "learning_rate": 0.00018087224173546015,
      "loss": 3.6205,
      "step": 102600
    },
    {
      "epoch": 0.0956491231095613,
      "grad_norm": 0.9100499153137207,
      "learning_rate": 0.00018087037741336496,
      "loss": 3.6443,
      "step": 102610
    },
    {
      "epoch": 0.09565844472764037,
      "grad_norm": 1.1092183589935303,
      "learning_rate": 0.00018086851309126977,
      "loss": 3.6163,
      "step": 102620
    },
    {
      "epoch": 0.09566776634571947,
      "grad_norm": 0.8074286580085754,
      "learning_rate": 0.00018086664876917457,
      "loss": 3.7972,
      "step": 102630
    },
    {
      "epoch": 0.09567708796379856,
      "grad_norm": 0.9993937015533447,
      "learning_rate": 0.00018086478444707938,
      "loss": 3.893,
      "step": 102640
    },
    {
      "epoch": 0.09568640958187766,
      "grad_norm": 1.168033480644226,
      "learning_rate": 0.00018086292012498416,
      "loss": 3.6603,
      "step": 102650
    },
    {
      "epoch": 0.09569573119995675,
      "grad_norm": 1.1024727821350098,
      "learning_rate": 0.00018086105580288896,
      "loss": 3.7382,
      "step": 102660
    },
    {
      "epoch": 0.09570505281803585,
      "grad_norm": 1.173277735710144,
      "learning_rate": 0.00018085919148079374,
      "loss": 3.8759,
      "step": 102670
    },
    {
      "epoch": 0.09571437443611493,
      "grad_norm": 1.1627455949783325,
      "learning_rate": 0.00018085732715869858,
      "loss": 3.4184,
      "step": 102680
    },
    {
      "epoch": 0.09572369605419402,
      "grad_norm": 1.1764169931411743,
      "learning_rate": 0.00018085546283660338,
      "loss": 3.7542,
      "step": 102690
    },
    {
      "epoch": 0.09573301767227312,
      "grad_norm": 1.3580563068389893,
      "learning_rate": 0.00018085359851450816,
      "loss": 3.7766,
      "step": 102700
    },
    {
      "epoch": 0.09574233929035221,
      "grad_norm": 1.0042778253555298,
      "learning_rate": 0.00018085173419241297,
      "loss": 3.6675,
      "step": 102710
    },
    {
      "epoch": 0.0957516609084313,
      "grad_norm": 1.0848504304885864,
      "learning_rate": 0.00018084986987031775,
      "loss": 3.7441,
      "step": 102720
    },
    {
      "epoch": 0.09576098252651039,
      "grad_norm": 1.214794397354126,
      "learning_rate": 0.00018084800554822255,
      "loss": 3.7254,
      "step": 102730
    },
    {
      "epoch": 0.09577030414458948,
      "grad_norm": 1.1358286142349243,
      "learning_rate": 0.0001808461412261274,
      "loss": 3.9093,
      "step": 102740
    },
    {
      "epoch": 0.09577962576266857,
      "grad_norm": 1.2553428411483765,
      "learning_rate": 0.00018084427690403217,
      "loss": 3.7403,
      "step": 102750
    },
    {
      "epoch": 0.09578894738074767,
      "grad_norm": 0.8554210066795349,
      "learning_rate": 0.00018084241258193697,
      "loss": 3.7524,
      "step": 102760
    },
    {
      "epoch": 0.09579826899882676,
      "grad_norm": 1.0263805389404297,
      "learning_rate": 0.00018084054825984175,
      "loss": 3.6361,
      "step": 102770
    },
    {
      "epoch": 0.09580759061690584,
      "grad_norm": 1.1583203077316284,
      "learning_rate": 0.00018083868393774656,
      "loss": 3.641,
      "step": 102780
    },
    {
      "epoch": 0.09581691223498494,
      "grad_norm": 1.111725926399231,
      "learning_rate": 0.00018083681961565136,
      "loss": 3.8797,
      "step": 102790
    },
    {
      "epoch": 0.09582623385306403,
      "grad_norm": 1.0218651294708252,
      "learning_rate": 0.00018083495529355617,
      "loss": 3.7061,
      "step": 102800
    },
    {
      "epoch": 0.09583555547114313,
      "grad_norm": 1.214371919631958,
      "learning_rate": 0.00018083309097146098,
      "loss": 3.6772,
      "step": 102810
    },
    {
      "epoch": 0.09584487708922222,
      "grad_norm": 1.1974419355392456,
      "learning_rate": 0.00018083122664936576,
      "loss": 3.7898,
      "step": 102820
    },
    {
      "epoch": 0.0958541987073013,
      "grad_norm": 1.0098545551300049,
      "learning_rate": 0.00018082936232727056,
      "loss": 3.8725,
      "step": 102830
    },
    {
      "epoch": 0.0958635203253804,
      "grad_norm": 1.2061556577682495,
      "learning_rate": 0.00018082749800517537,
      "loss": 3.6652,
      "step": 102840
    },
    {
      "epoch": 0.09587284194345949,
      "grad_norm": 1.1971746683120728,
      "learning_rate": 0.00018082563368308018,
      "loss": 3.833,
      "step": 102850
    },
    {
      "epoch": 0.09588216356153859,
      "grad_norm": 1.091713786125183,
      "learning_rate": 0.00018082376936098498,
      "loss": 3.6873,
      "step": 102860
    },
    {
      "epoch": 0.09589148517961768,
      "grad_norm": 1.003911018371582,
      "learning_rate": 0.00018082190503888976,
      "loss": 3.7872,
      "step": 102870
    },
    {
      "epoch": 0.09590080679769677,
      "grad_norm": 1.1617584228515625,
      "learning_rate": 0.00018082004071679457,
      "loss": 3.8703,
      "step": 102880
    },
    {
      "epoch": 0.09591012841577586,
      "grad_norm": 1.0164756774902344,
      "learning_rate": 0.00018081817639469937,
      "loss": 3.7449,
      "step": 102890
    },
    {
      "epoch": 0.09591945003385495,
      "grad_norm": 1.1151362657546997,
      "learning_rate": 0.00018081631207260418,
      "loss": 3.4256,
      "step": 102900
    },
    {
      "epoch": 0.09592877165193404,
      "grad_norm": 0.943231999874115,
      "learning_rate": 0.00018081444775050899,
      "loss": 3.749,
      "step": 102910
    },
    {
      "epoch": 0.09593809327001314,
      "grad_norm": 0.8609538674354553,
      "learning_rate": 0.00018081258342841376,
      "loss": 3.8127,
      "step": 102920
    },
    {
      "epoch": 0.09594741488809223,
      "grad_norm": 1.0590089559555054,
      "learning_rate": 0.00018081071910631857,
      "loss": 3.6751,
      "step": 102930
    },
    {
      "epoch": 0.09595673650617131,
      "grad_norm": 1.1949260234832764,
      "learning_rate": 0.00018080885478422338,
      "loss": 3.6875,
      "step": 102940
    },
    {
      "epoch": 0.09596605812425041,
      "grad_norm": 0.9960625171661377,
      "learning_rate": 0.00018080699046212816,
      "loss": 3.6868,
      "step": 102950
    },
    {
      "epoch": 0.0959753797423295,
      "grad_norm": 1.226158857345581,
      "learning_rate": 0.00018080512614003296,
      "loss": 3.4784,
      "step": 102960
    },
    {
      "epoch": 0.0959847013604086,
      "grad_norm": 1.1355928182601929,
      "learning_rate": 0.00018080326181793777,
      "loss": 3.6877,
      "step": 102970
    },
    {
      "epoch": 0.09599402297848769,
      "grad_norm": 0.8241602778434753,
      "learning_rate": 0.00018080139749584258,
      "loss": 3.7392,
      "step": 102980
    },
    {
      "epoch": 0.09600334459656677,
      "grad_norm": 1.0279706716537476,
      "learning_rate": 0.00018079953317374735,
      "loss": 3.6354,
      "step": 102990
    },
    {
      "epoch": 0.09601266621464587,
      "grad_norm": 1.0438532829284668,
      "learning_rate": 0.00018079766885165216,
      "loss": 3.5728,
      "step": 103000
    },
    {
      "epoch": 0.09602198783272496,
      "grad_norm": 1.2876126766204834,
      "learning_rate": 0.00018079580452955697,
      "loss": 3.7529,
      "step": 103010
    },
    {
      "epoch": 0.09603130945080406,
      "grad_norm": 0.9533160924911499,
      "learning_rate": 0.00018079394020746177,
      "loss": 3.9426,
      "step": 103020
    },
    {
      "epoch": 0.09604063106888315,
      "grad_norm": 0.9487892985343933,
      "learning_rate": 0.00018079207588536658,
      "loss": 3.7921,
      "step": 103030
    },
    {
      "epoch": 0.09604995268696223,
      "grad_norm": 0.896041214466095,
      "learning_rate": 0.00018079021156327136,
      "loss": 3.5562,
      "step": 103040
    },
    {
      "epoch": 0.09605927430504133,
      "grad_norm": 1.1229034662246704,
      "learning_rate": 0.00018078834724117617,
      "loss": 3.686,
      "step": 103050
    },
    {
      "epoch": 0.09606859592312042,
      "grad_norm": 1.0499640703201294,
      "learning_rate": 0.00018078648291908097,
      "loss": 3.6622,
      "step": 103060
    },
    {
      "epoch": 0.09607791754119951,
      "grad_norm": 0.989310622215271,
      "learning_rate": 0.00018078461859698578,
      "loss": 3.7789,
      "step": 103070
    },
    {
      "epoch": 0.09608723915927861,
      "grad_norm": 1.1133582592010498,
      "learning_rate": 0.00018078275427489058,
      "loss": 3.8566,
      "step": 103080
    },
    {
      "epoch": 0.09609656077735769,
      "grad_norm": 1.0382537841796875,
      "learning_rate": 0.00018078088995279536,
      "loss": 3.5197,
      "step": 103090
    },
    {
      "epoch": 0.09610588239543678,
      "grad_norm": 0.9637733697891235,
      "learning_rate": 0.00018077902563070017,
      "loss": 3.7908,
      "step": 103100
    },
    {
      "epoch": 0.09611520401351588,
      "grad_norm": 1.0506818294525146,
      "learning_rate": 0.00018077716130860498,
      "loss": 3.7775,
      "step": 103110
    },
    {
      "epoch": 0.09612452563159497,
      "grad_norm": 1.2841523885726929,
      "learning_rate": 0.00018077529698650976,
      "loss": 3.8073,
      "step": 103120
    },
    {
      "epoch": 0.09613384724967407,
      "grad_norm": 1.2412277460098267,
      "learning_rate": 0.0001807734326644146,
      "loss": 3.7874,
      "step": 103130
    },
    {
      "epoch": 0.09614316886775316,
      "grad_norm": 1.1116952896118164,
      "learning_rate": 0.00018077156834231937,
      "loss": 3.862,
      "step": 103140
    },
    {
      "epoch": 0.09615249048583224,
      "grad_norm": 0.8915697932243347,
      "learning_rate": 0.00018076970402022417,
      "loss": 3.876,
      "step": 103150
    },
    {
      "epoch": 0.09616181210391134,
      "grad_norm": 1.2747993469238281,
      "learning_rate": 0.00018076783969812898,
      "loss": 3.612,
      "step": 103160
    },
    {
      "epoch": 0.09617113372199043,
      "grad_norm": 1.0115240812301636,
      "learning_rate": 0.00018076597537603376,
      "loss": 3.6721,
      "step": 103170
    },
    {
      "epoch": 0.09618045534006953,
      "grad_norm": 1.080886960029602,
      "learning_rate": 0.00018076411105393857,
      "loss": 3.7186,
      "step": 103180
    },
    {
      "epoch": 0.09618977695814862,
      "grad_norm": 1.0933454036712646,
      "learning_rate": 0.00018076224673184337,
      "loss": 3.9849,
      "step": 103190
    },
    {
      "epoch": 0.0961990985762277,
      "grad_norm": 1.0590286254882812,
      "learning_rate": 0.00018076038240974818,
      "loss": 3.7278,
      "step": 103200
    },
    {
      "epoch": 0.0962084201943068,
      "grad_norm": 1.1855076551437378,
      "learning_rate": 0.00018075851808765299,
      "loss": 3.8491,
      "step": 103210
    },
    {
      "epoch": 0.09621774181238589,
      "grad_norm": 0.9025027751922607,
      "learning_rate": 0.00018075665376555776,
      "loss": 3.7337,
      "step": 103220
    },
    {
      "epoch": 0.09622706343046498,
      "grad_norm": 1.1770541667938232,
      "learning_rate": 0.00018075478944346257,
      "loss": 3.7513,
      "step": 103230
    },
    {
      "epoch": 0.09623638504854408,
      "grad_norm": 0.9843522310256958,
      "learning_rate": 0.00018075292512136738,
      "loss": 3.701,
      "step": 103240
    },
    {
      "epoch": 0.09624570666662316,
      "grad_norm": 1.0893287658691406,
      "learning_rate": 0.00018075106079927218,
      "loss": 3.8084,
      "step": 103250
    },
    {
      "epoch": 0.09625502828470225,
      "grad_norm": 0.9874154329299927,
      "learning_rate": 0.000180749196477177,
      "loss": 3.4542,
      "step": 103260
    },
    {
      "epoch": 0.09626434990278135,
      "grad_norm": 1.1352592706680298,
      "learning_rate": 0.00018074733215508177,
      "loss": 3.6468,
      "step": 103270
    },
    {
      "epoch": 0.09627367152086044,
      "grad_norm": 1.0050832033157349,
      "learning_rate": 0.00018074546783298658,
      "loss": 3.7439,
      "step": 103280
    },
    {
      "epoch": 0.09628299313893954,
      "grad_norm": 1.1981513500213623,
      "learning_rate": 0.00018074360351089138,
      "loss": 3.8232,
      "step": 103290
    },
    {
      "epoch": 0.09629231475701862,
      "grad_norm": 1.2175118923187256,
      "learning_rate": 0.0001807417391887962,
      "loss": 3.6777,
      "step": 103300
    },
    {
      "epoch": 0.09630163637509771,
      "grad_norm": 1.1969480514526367,
      "learning_rate": 0.000180739874866701,
      "loss": 3.7325,
      "step": 103310
    },
    {
      "epoch": 0.09631095799317681,
      "grad_norm": 0.9405122995376587,
      "learning_rate": 0.00018073801054460577,
      "loss": 3.758,
      "step": 103320
    },
    {
      "epoch": 0.0963202796112559,
      "grad_norm": 1.0854017734527588,
      "learning_rate": 0.00018073614622251058,
      "loss": 3.6134,
      "step": 103330
    },
    {
      "epoch": 0.096329601229335,
      "grad_norm": 0.8302522897720337,
      "learning_rate": 0.00018073428190041536,
      "loss": 3.6196,
      "step": 103340
    },
    {
      "epoch": 0.09633892284741409,
      "grad_norm": 0.9760847687721252,
      "learning_rate": 0.0001807324175783202,
      "loss": 3.6703,
      "step": 103350
    },
    {
      "epoch": 0.09634824446549317,
      "grad_norm": 1.1407372951507568,
      "learning_rate": 0.000180730553256225,
      "loss": 3.7795,
      "step": 103360
    },
    {
      "epoch": 0.09635756608357227,
      "grad_norm": 1.0304967164993286,
      "learning_rate": 0.00018072868893412978,
      "loss": 3.8152,
      "step": 103370
    },
    {
      "epoch": 0.09636688770165136,
      "grad_norm": 1.1049439907073975,
      "learning_rate": 0.00018072682461203458,
      "loss": 3.7899,
      "step": 103380
    },
    {
      "epoch": 0.09637620931973045,
      "grad_norm": 1.0360692739486694,
      "learning_rate": 0.00018072496028993936,
      "loss": 3.7923,
      "step": 103390
    },
    {
      "epoch": 0.09638553093780955,
      "grad_norm": 1.0604021549224854,
      "learning_rate": 0.00018072309596784417,
      "loss": 3.7951,
      "step": 103400
    },
    {
      "epoch": 0.09639485255588863,
      "grad_norm": 0.9471161365509033,
      "learning_rate": 0.000180721231645749,
      "loss": 3.743,
      "step": 103410
    },
    {
      "epoch": 0.09640417417396772,
      "grad_norm": 1.2123767137527466,
      "learning_rate": 0.00018071936732365378,
      "loss": 3.599,
      "step": 103420
    },
    {
      "epoch": 0.09641349579204682,
      "grad_norm": 0.9374601244926453,
      "learning_rate": 0.0001807175030015586,
      "loss": 3.7649,
      "step": 103430
    },
    {
      "epoch": 0.09642281741012591,
      "grad_norm": 1.0754557847976685,
      "learning_rate": 0.00018071563867946337,
      "loss": 3.5176,
      "step": 103440
    },
    {
      "epoch": 0.09643213902820501,
      "grad_norm": 0.9241998195648193,
      "learning_rate": 0.00018071377435736817,
      "loss": 3.911,
      "step": 103450
    },
    {
      "epoch": 0.09644146064628409,
      "grad_norm": 1.0887782573699951,
      "learning_rate": 0.00018071191003527298,
      "loss": 3.4897,
      "step": 103460
    },
    {
      "epoch": 0.09645078226436318,
      "grad_norm": 1.0907841920852661,
      "learning_rate": 0.0001807100457131778,
      "loss": 3.9061,
      "step": 103470
    },
    {
      "epoch": 0.09646010388244228,
      "grad_norm": 1.2453556060791016,
      "learning_rate": 0.0001807081813910826,
      "loss": 3.6237,
      "step": 103480
    },
    {
      "epoch": 0.09646942550052137,
      "grad_norm": 1.0736362934112549,
      "learning_rate": 0.00018070631706898737,
      "loss": 3.6071,
      "step": 103490
    },
    {
      "epoch": 0.09647874711860047,
      "grad_norm": 0.9650845527648926,
      "learning_rate": 0.00018070445274689218,
      "loss": 3.6274,
      "step": 103500
    },
    {
      "epoch": 0.09648806873667955,
      "grad_norm": 1.2106297016143799,
      "learning_rate": 0.00018070258842479698,
      "loss": 3.4983,
      "step": 103510
    },
    {
      "epoch": 0.09649739035475864,
      "grad_norm": 1.0035321712493896,
      "learning_rate": 0.0001807007241027018,
      "loss": 3.7459,
      "step": 103520
    },
    {
      "epoch": 0.09650671197283774,
      "grad_norm": 1.1483354568481445,
      "learning_rate": 0.0001806988597806066,
      "loss": 3.6677,
      "step": 103530
    },
    {
      "epoch": 0.09651603359091683,
      "grad_norm": 0.9624996185302734,
      "learning_rate": 0.00018069699545851138,
      "loss": 3.7419,
      "step": 103540
    },
    {
      "epoch": 0.09652535520899592,
      "grad_norm": 1.1335906982421875,
      "learning_rate": 0.00018069513113641618,
      "loss": 3.6662,
      "step": 103550
    },
    {
      "epoch": 0.09653467682707502,
      "grad_norm": 1.0340567827224731,
      "learning_rate": 0.000180693266814321,
      "loss": 3.5873,
      "step": 103560
    },
    {
      "epoch": 0.0965439984451541,
      "grad_norm": 0.9924055337905884,
      "learning_rate": 0.00018069140249222577,
      "loss": 3.7231,
      "step": 103570
    },
    {
      "epoch": 0.0965533200632332,
      "grad_norm": 1.1187551021575928,
      "learning_rate": 0.0001806895381701306,
      "loss": 3.7454,
      "step": 103580
    },
    {
      "epoch": 0.09656264168131229,
      "grad_norm": 1.081907868385315,
      "learning_rate": 0.00018068767384803538,
      "loss": 3.6239,
      "step": 103590
    },
    {
      "epoch": 0.09657196329939138,
      "grad_norm": 0.9387475848197937,
      "learning_rate": 0.0001806858095259402,
      "loss": 3.7458,
      "step": 103600
    },
    {
      "epoch": 0.09658128491747048,
      "grad_norm": 1.0217300653457642,
      "learning_rate": 0.000180683945203845,
      "loss": 3.738,
      "step": 103610
    },
    {
      "epoch": 0.09659060653554956,
      "grad_norm": 1.0564934015274048,
      "learning_rate": 0.00018068208088174977,
      "loss": 3.6558,
      "step": 103620
    },
    {
      "epoch": 0.09659992815362865,
      "grad_norm": 0.9642091393470764,
      "learning_rate": 0.00018068021655965458,
      "loss": 3.8191,
      "step": 103630
    },
    {
      "epoch": 0.09660924977170775,
      "grad_norm": 0.9722574353218079,
      "learning_rate": 0.00018067835223755939,
      "loss": 3.7631,
      "step": 103640
    },
    {
      "epoch": 0.09661857138978684,
      "grad_norm": 1.1365660429000854,
      "learning_rate": 0.0001806764879154642,
      "loss": 3.6078,
      "step": 103650
    },
    {
      "epoch": 0.09662789300786594,
      "grad_norm": 1.2646257877349854,
      "learning_rate": 0.000180674623593369,
      "loss": 3.6179,
      "step": 103660
    },
    {
      "epoch": 0.09663721462594502,
      "grad_norm": 0.9459970593452454,
      "learning_rate": 0.00018067275927127378,
      "loss": 3.7357,
      "step": 103670
    },
    {
      "epoch": 0.09664653624402411,
      "grad_norm": 1.002448558807373,
      "learning_rate": 0.00018067089494917858,
      "loss": 3.7476,
      "step": 103680
    },
    {
      "epoch": 0.0966558578621032,
      "grad_norm": 1.2426061630249023,
      "learning_rate": 0.0001806690306270834,
      "loss": 4.0219,
      "step": 103690
    },
    {
      "epoch": 0.0966651794801823,
      "grad_norm": 1.1202764511108398,
      "learning_rate": 0.0001806671663049882,
      "loss": 3.6427,
      "step": 103700
    },
    {
      "epoch": 0.0966745010982614,
      "grad_norm": 1.1006306409835815,
      "learning_rate": 0.000180665301982893,
      "loss": 3.6557,
      "step": 103710
    },
    {
      "epoch": 0.09668382271634048,
      "grad_norm": 1.1351875066757202,
      "learning_rate": 0.00018066343766079778,
      "loss": 3.8242,
      "step": 103720
    },
    {
      "epoch": 0.09669314433441957,
      "grad_norm": 1.0578761100769043,
      "learning_rate": 0.0001806615733387026,
      "loss": 3.7967,
      "step": 103730
    },
    {
      "epoch": 0.09670246595249866,
      "grad_norm": 1.0577791929244995,
      "learning_rate": 0.0001806597090166074,
      "loss": 3.7704,
      "step": 103740
    },
    {
      "epoch": 0.09671178757057776,
      "grad_norm": 0.9288763403892517,
      "learning_rate": 0.0001806578446945122,
      "loss": 3.7781,
      "step": 103750
    },
    {
      "epoch": 0.09672110918865685,
      "grad_norm": 1.2617831230163574,
      "learning_rate": 0.000180655980372417,
      "loss": 3.7649,
      "step": 103760
    },
    {
      "epoch": 0.09673043080673593,
      "grad_norm": 0.8993209004402161,
      "learning_rate": 0.00018065411605032179,
      "loss": 3.7447,
      "step": 103770
    },
    {
      "epoch": 0.09673975242481503,
      "grad_norm": 1.2462910413742065,
      "learning_rate": 0.0001806522517282266,
      "loss": 3.7232,
      "step": 103780
    },
    {
      "epoch": 0.09674907404289412,
      "grad_norm": 0.994921863079071,
      "learning_rate": 0.00018065038740613137,
      "loss": 3.5736,
      "step": 103790
    },
    {
      "epoch": 0.09675839566097322,
      "grad_norm": 1.0747909545898438,
      "learning_rate": 0.0001806485230840362,
      "loss": 3.5724,
      "step": 103800
    },
    {
      "epoch": 0.09676771727905231,
      "grad_norm": 1.0652018785476685,
      "learning_rate": 0.000180646658761941,
      "loss": 3.7035,
      "step": 103810
    },
    {
      "epoch": 0.0967770388971314,
      "grad_norm": 1.086314082145691,
      "learning_rate": 0.0001806447944398458,
      "loss": 3.8947,
      "step": 103820
    },
    {
      "epoch": 0.09678636051521049,
      "grad_norm": 1.087131142616272,
      "learning_rate": 0.0001806429301177506,
      "loss": 3.5512,
      "step": 103830
    },
    {
      "epoch": 0.09679568213328958,
      "grad_norm": 1.2556777000427246,
      "learning_rate": 0.00018064106579565538,
      "loss": 3.813,
      "step": 103840
    },
    {
      "epoch": 0.09680500375136868,
      "grad_norm": 1.0730314254760742,
      "learning_rate": 0.00018063920147356018,
      "loss": 3.8956,
      "step": 103850
    },
    {
      "epoch": 0.09681432536944777,
      "grad_norm": 0.9418400526046753,
      "learning_rate": 0.00018063733715146502,
      "loss": 3.7061,
      "step": 103860
    },
    {
      "epoch": 0.09682364698752687,
      "grad_norm": 1.1305567026138306,
      "learning_rate": 0.0001806354728293698,
      "loss": 3.6416,
      "step": 103870
    },
    {
      "epoch": 0.09683296860560595,
      "grad_norm": 1.2845218181610107,
      "learning_rate": 0.0001806336085072746,
      "loss": 3.6587,
      "step": 103880
    },
    {
      "epoch": 0.09684229022368504,
      "grad_norm": 0.880255937576294,
      "learning_rate": 0.00018063174418517938,
      "loss": 3.7725,
      "step": 103890
    },
    {
      "epoch": 0.09685161184176413,
      "grad_norm": 1.4816128015518188,
      "learning_rate": 0.0001806298798630842,
      "loss": 3.9561,
      "step": 103900
    },
    {
      "epoch": 0.09686093345984323,
      "grad_norm": 1.129249095916748,
      "learning_rate": 0.000180628015540989,
      "loss": 3.8335,
      "step": 103910
    },
    {
      "epoch": 0.09687025507792232,
      "grad_norm": 0.8826767206192017,
      "learning_rate": 0.0001806261512188938,
      "loss": 3.8089,
      "step": 103920
    },
    {
      "epoch": 0.0968795766960014,
      "grad_norm": 1.071058988571167,
      "learning_rate": 0.0001806242868967986,
      "loss": 3.5479,
      "step": 103930
    },
    {
      "epoch": 0.0968888983140805,
      "grad_norm": 1.010758876800537,
      "learning_rate": 0.00018062242257470339,
      "loss": 3.7503,
      "step": 103940
    },
    {
      "epoch": 0.0968982199321596,
      "grad_norm": 1.2752964496612549,
      "learning_rate": 0.0001806205582526082,
      "loss": 3.8164,
      "step": 103950
    },
    {
      "epoch": 0.09690754155023869,
      "grad_norm": 1.216945767402649,
      "learning_rate": 0.000180618693930513,
      "loss": 3.7654,
      "step": 103960
    },
    {
      "epoch": 0.09691686316831778,
      "grad_norm": 1.250680685043335,
      "learning_rate": 0.0001806168296084178,
      "loss": 3.746,
      "step": 103970
    },
    {
      "epoch": 0.09692618478639686,
      "grad_norm": 1.058275580406189,
      "learning_rate": 0.0001806149652863226,
      "loss": 3.6242,
      "step": 103980
    },
    {
      "epoch": 0.09693550640447596,
      "grad_norm": 1.361810564994812,
      "learning_rate": 0.0001806131009642274,
      "loss": 3.7206,
      "step": 103990
    },
    {
      "epoch": 0.09694482802255505,
      "grad_norm": 1.098504900932312,
      "learning_rate": 0.0001806112366421322,
      "loss": 3.8691,
      "step": 104000
    },
    {
      "epoch": 0.09695414964063415,
      "grad_norm": 1.1657143831253052,
      "learning_rate": 0.000180609372320037,
      "loss": 3.6322,
      "step": 104010
    },
    {
      "epoch": 0.09696347125871324,
      "grad_norm": 1.0002983808517456,
      "learning_rate": 0.00018060750799794178,
      "loss": 3.7391,
      "step": 104020
    },
    {
      "epoch": 0.09697279287679234,
      "grad_norm": 0.9441248774528503,
      "learning_rate": 0.00018060564367584661,
      "loss": 3.8551,
      "step": 104030
    },
    {
      "epoch": 0.09698211449487142,
      "grad_norm": 1.1406751871109009,
      "learning_rate": 0.0001806037793537514,
      "loss": 3.7702,
      "step": 104040
    },
    {
      "epoch": 0.09699143611295051,
      "grad_norm": 1.045914888381958,
      "learning_rate": 0.0001806019150316562,
      "loss": 3.9192,
      "step": 104050
    },
    {
      "epoch": 0.0970007577310296,
      "grad_norm": 0.8500249981880188,
      "learning_rate": 0.000180600050709561,
      "loss": 3.9081,
      "step": 104060
    },
    {
      "epoch": 0.0970100793491087,
      "grad_norm": 1.2563904523849487,
      "learning_rate": 0.00018059818638746579,
      "loss": 3.7436,
      "step": 104070
    },
    {
      "epoch": 0.0970194009671878,
      "grad_norm": 1.0123815536499023,
      "learning_rate": 0.0001805963220653706,
      "loss": 3.7875,
      "step": 104080
    },
    {
      "epoch": 0.09702872258526687,
      "grad_norm": 1.0926215648651123,
      "learning_rate": 0.0001805944577432754,
      "loss": 3.6969,
      "step": 104090
    },
    {
      "epoch": 0.09703804420334597,
      "grad_norm": 1.246465802192688,
      "learning_rate": 0.0001805925934211802,
      "loss": 3.7652,
      "step": 104100
    },
    {
      "epoch": 0.09704736582142506,
      "grad_norm": 1.1260854005813599,
      "learning_rate": 0.000180590729099085,
      "loss": 3.8139,
      "step": 104110
    },
    {
      "epoch": 0.09705668743950416,
      "grad_norm": 1.1920746564865112,
      "learning_rate": 0.0001805888647769898,
      "loss": 4.0541,
      "step": 104120
    },
    {
      "epoch": 0.09706600905758325,
      "grad_norm": 0.9844964742660522,
      "learning_rate": 0.0001805870004548946,
      "loss": 3.6236,
      "step": 104130
    },
    {
      "epoch": 0.09707533067566233,
      "grad_norm": 1.2984477281570435,
      "learning_rate": 0.0001805851361327994,
      "loss": 3.678,
      "step": 104140
    },
    {
      "epoch": 0.09708465229374143,
      "grad_norm": 1.0110043287277222,
      "learning_rate": 0.0001805832718107042,
      "loss": 3.6675,
      "step": 104150
    },
    {
      "epoch": 0.09709397391182052,
      "grad_norm": 1.0683505535125732,
      "learning_rate": 0.00018058140748860902,
      "loss": 3.58,
      "step": 104160
    },
    {
      "epoch": 0.09710329552989962,
      "grad_norm": 0.9819969534873962,
      "learning_rate": 0.0001805795431665138,
      "loss": 3.7653,
      "step": 104170
    },
    {
      "epoch": 0.09711261714797871,
      "grad_norm": 0.975877046585083,
      "learning_rate": 0.0001805776788444186,
      "loss": 3.8148,
      "step": 104180
    },
    {
      "epoch": 0.09712193876605779,
      "grad_norm": 0.9009703397750854,
      "learning_rate": 0.0001805758145223234,
      "loss": 3.7117,
      "step": 104190
    },
    {
      "epoch": 0.09713126038413689,
      "grad_norm": 1.0618224143981934,
      "learning_rate": 0.00018057395020022821,
      "loss": 3.8047,
      "step": 104200
    },
    {
      "epoch": 0.09714058200221598,
      "grad_norm": 1.1951241493225098,
      "learning_rate": 0.00018057208587813302,
      "loss": 3.6894,
      "step": 104210
    },
    {
      "epoch": 0.09714990362029507,
      "grad_norm": 0.953120768070221,
      "learning_rate": 0.0001805702215560378,
      "loss": 3.8732,
      "step": 104220
    },
    {
      "epoch": 0.09715922523837417,
      "grad_norm": 0.9790743589401245,
      "learning_rate": 0.0001805683572339426,
      "loss": 3.7055,
      "step": 104230
    },
    {
      "epoch": 0.09716854685645326,
      "grad_norm": 1.0389939546585083,
      "learning_rate": 0.00018056649291184738,
      "loss": 3.5761,
      "step": 104240
    },
    {
      "epoch": 0.09717786847453234,
      "grad_norm": 1.1138217449188232,
      "learning_rate": 0.00018056462858975222,
      "loss": 3.7689,
      "step": 104250
    },
    {
      "epoch": 0.09718719009261144,
      "grad_norm": 0.9806100726127625,
      "learning_rate": 0.00018056276426765702,
      "loss": 3.8504,
      "step": 104260
    },
    {
      "epoch": 0.09719651171069053,
      "grad_norm": 1.1847310066223145,
      "learning_rate": 0.0001805608999455618,
      "loss": 3.8155,
      "step": 104270
    },
    {
      "epoch": 0.09720583332876963,
      "grad_norm": 1.0956501960754395,
      "learning_rate": 0.0001805590356234666,
      "loss": 3.675,
      "step": 104280
    },
    {
      "epoch": 0.09721515494684872,
      "grad_norm": 0.9659402370452881,
      "learning_rate": 0.0001805571713013714,
      "loss": 3.567,
      "step": 104290
    },
    {
      "epoch": 0.0972244765649278,
      "grad_norm": 1.1485497951507568,
      "learning_rate": 0.0001805553069792762,
      "loss": 3.6881,
      "step": 104300
    },
    {
      "epoch": 0.0972337981830069,
      "grad_norm": 0.9314612746238708,
      "learning_rate": 0.00018055344265718103,
      "loss": 3.7845,
      "step": 104310
    },
    {
      "epoch": 0.09724311980108599,
      "grad_norm": 0.9483014941215515,
      "learning_rate": 0.0001805515783350858,
      "loss": 3.7439,
      "step": 104320
    },
    {
      "epoch": 0.09725244141916509,
      "grad_norm": 1.120376467704773,
      "learning_rate": 0.00018054971401299061,
      "loss": 3.6995,
      "step": 104330
    },
    {
      "epoch": 0.09726176303724418,
      "grad_norm": 1.0823677778244019,
      "learning_rate": 0.0001805478496908954,
      "loss": 3.7535,
      "step": 104340
    },
    {
      "epoch": 0.09727108465532326,
      "grad_norm": 0.9874478578567505,
      "learning_rate": 0.0001805459853688002,
      "loss": 3.7409,
      "step": 104350
    },
    {
      "epoch": 0.09728040627340236,
      "grad_norm": 1.1404588222503662,
      "learning_rate": 0.000180544121046705,
      "loss": 3.7961,
      "step": 104360
    },
    {
      "epoch": 0.09728972789148145,
      "grad_norm": 1.1689399480819702,
      "learning_rate": 0.0001805422567246098,
      "loss": 3.7592,
      "step": 104370
    },
    {
      "epoch": 0.09729904950956055,
      "grad_norm": 1.0094228982925415,
      "learning_rate": 0.00018054039240251462,
      "loss": 3.624,
      "step": 104380
    },
    {
      "epoch": 0.09730837112763964,
      "grad_norm": 1.0657371282577515,
      "learning_rate": 0.0001805385280804194,
      "loss": 3.8239,
      "step": 104390
    },
    {
      "epoch": 0.09731769274571872,
      "grad_norm": 1.3072583675384521,
      "learning_rate": 0.0001805366637583242,
      "loss": 3.946,
      "step": 104400
    },
    {
      "epoch": 0.09732701436379781,
      "grad_norm": 1.257609248161316,
      "learning_rate": 0.00018053479943622898,
      "loss": 3.8254,
      "step": 104410
    },
    {
      "epoch": 0.09733633598187691,
      "grad_norm": 1.121706247329712,
      "learning_rate": 0.00018053293511413382,
      "loss": 3.7039,
      "step": 104420
    },
    {
      "epoch": 0.097345657599956,
      "grad_norm": 1.023894190788269,
      "learning_rate": 0.00018053107079203862,
      "loss": 3.8521,
      "step": 104430
    },
    {
      "epoch": 0.0973549792180351,
      "grad_norm": 1.2927541732788086,
      "learning_rate": 0.0001805292064699434,
      "loss": 3.8167,
      "step": 104440
    },
    {
      "epoch": 0.09736430083611418,
      "grad_norm": 0.9489665627479553,
      "learning_rate": 0.0001805273421478482,
      "loss": 3.806,
      "step": 104450
    },
    {
      "epoch": 0.09737362245419327,
      "grad_norm": 1.0918493270874023,
      "learning_rate": 0.000180525477825753,
      "loss": 3.7642,
      "step": 104460
    },
    {
      "epoch": 0.09738294407227237,
      "grad_norm": 0.9511283040046692,
      "learning_rate": 0.0001805236135036578,
      "loss": 3.6319,
      "step": 104470
    },
    {
      "epoch": 0.09739226569035146,
      "grad_norm": 0.9947870969772339,
      "learning_rate": 0.00018052174918156263,
      "loss": 3.636,
      "step": 104480
    },
    {
      "epoch": 0.09740158730843056,
      "grad_norm": 1.1543853282928467,
      "learning_rate": 0.0001805198848594674,
      "loss": 3.5719,
      "step": 104490
    },
    {
      "epoch": 0.09741090892650965,
      "grad_norm": 0.99566251039505,
      "learning_rate": 0.0001805180205373722,
      "loss": 3.7784,
      "step": 104500
    },
    {
      "epoch": 0.09742023054458873,
      "grad_norm": 0.9713016748428345,
      "learning_rate": 0.000180516156215277,
      "loss": 3.6853,
      "step": 104510
    },
    {
      "epoch": 0.09742955216266783,
      "grad_norm": 1.0217782258987427,
      "learning_rate": 0.0001805142918931818,
      "loss": 3.8212,
      "step": 104520
    },
    {
      "epoch": 0.09743887378074692,
      "grad_norm": 1.0912683010101318,
      "learning_rate": 0.0001805124275710866,
      "loss": 3.9039,
      "step": 104530
    },
    {
      "epoch": 0.09744819539882602,
      "grad_norm": 1.1602989435195923,
      "learning_rate": 0.0001805105632489914,
      "loss": 3.7183,
      "step": 104540
    },
    {
      "epoch": 0.09745751701690511,
      "grad_norm": 1.0551726818084717,
      "learning_rate": 0.00018050869892689622,
      "loss": 3.7582,
      "step": 104550
    },
    {
      "epoch": 0.09746683863498419,
      "grad_norm": 0.9055781960487366,
      "learning_rate": 0.000180506834604801,
      "loss": 3.88,
      "step": 104560
    },
    {
      "epoch": 0.09747616025306328,
      "grad_norm": 1.0079420804977417,
      "learning_rate": 0.0001805049702827058,
      "loss": 3.8205,
      "step": 104570
    },
    {
      "epoch": 0.09748548187114238,
      "grad_norm": 1.1151195764541626,
      "learning_rate": 0.0001805031059606106,
      "loss": 3.7885,
      "step": 104580
    },
    {
      "epoch": 0.09749480348922147,
      "grad_norm": 1.0661001205444336,
      "learning_rate": 0.00018050124163851542,
      "loss": 3.4896,
      "step": 104590
    },
    {
      "epoch": 0.09750412510730057,
      "grad_norm": 1.003326416015625,
      "learning_rate": 0.00018049937731642022,
      "loss": 3.5795,
      "step": 104600
    },
    {
      "epoch": 0.09751344672537965,
      "grad_norm": 1.2465444803237915,
      "learning_rate": 0.000180497512994325,
      "loss": 3.4595,
      "step": 104610
    },
    {
      "epoch": 0.09752276834345874,
      "grad_norm": 0.9350901246070862,
      "learning_rate": 0.0001804956486722298,
      "loss": 3.8216,
      "step": 104620
    },
    {
      "epoch": 0.09753208996153784,
      "grad_norm": 1.1609870195388794,
      "learning_rate": 0.00018049378435013461,
      "loss": 3.5429,
      "step": 104630
    },
    {
      "epoch": 0.09754141157961693,
      "grad_norm": 1.0835095643997192,
      "learning_rate": 0.00018049192002803942,
      "loss": 3.78,
      "step": 104640
    },
    {
      "epoch": 0.09755073319769603,
      "grad_norm": 1.0005271434783936,
      "learning_rate": 0.00018049005570594423,
      "loss": 3.5814,
      "step": 104650
    },
    {
      "epoch": 0.09756005481577511,
      "grad_norm": 0.9637694358825684,
      "learning_rate": 0.000180488191383849,
      "loss": 3.7675,
      "step": 104660
    },
    {
      "epoch": 0.0975693764338542,
      "grad_norm": 0.805614173412323,
      "learning_rate": 0.0001804863270617538,
      "loss": 3.8345,
      "step": 104670
    },
    {
      "epoch": 0.0975786980519333,
      "grad_norm": 1.0945792198181152,
      "learning_rate": 0.00018048446273965862,
      "loss": 3.9126,
      "step": 104680
    },
    {
      "epoch": 0.09758801967001239,
      "grad_norm": 1.1312123537063599,
      "learning_rate": 0.0001804825984175634,
      "loss": 3.514,
      "step": 104690
    },
    {
      "epoch": 0.09759734128809149,
      "grad_norm": 1.0821669101715088,
      "learning_rate": 0.00018048073409546823,
      "loss": 3.7764,
      "step": 104700
    },
    {
      "epoch": 0.09760666290617058,
      "grad_norm": 1.1024432182312012,
      "learning_rate": 0.000180478869773373,
      "loss": 3.6962,
      "step": 104710
    },
    {
      "epoch": 0.09761598452424966,
      "grad_norm": 0.983674943447113,
      "learning_rate": 0.00018047700545127782,
      "loss": 3.6194,
      "step": 104720
    },
    {
      "epoch": 0.09762530614232875,
      "grad_norm": 0.965941846370697,
      "learning_rate": 0.00018047514112918262,
      "loss": 3.7133,
      "step": 104730
    },
    {
      "epoch": 0.09763462776040785,
      "grad_norm": 0.9927389025688171,
      "learning_rate": 0.0001804732768070874,
      "loss": 3.4589,
      "step": 104740
    },
    {
      "epoch": 0.09764394937848694,
      "grad_norm": 1.1071724891662598,
      "learning_rate": 0.0001804714124849922,
      "loss": 3.5443,
      "step": 104750
    },
    {
      "epoch": 0.09765327099656604,
      "grad_norm": 1.0694530010223389,
      "learning_rate": 0.00018046954816289701,
      "loss": 3.8056,
      "step": 104760
    },
    {
      "epoch": 0.09766259261464512,
      "grad_norm": 1.0993177890777588,
      "learning_rate": 0.00018046768384080182,
      "loss": 3.6054,
      "step": 104770
    },
    {
      "epoch": 0.09767191423272421,
      "grad_norm": 0.9072667956352234,
      "learning_rate": 0.00018046581951870663,
      "loss": 3.7783,
      "step": 104780
    },
    {
      "epoch": 0.09768123585080331,
      "grad_norm": 1.0801169872283936,
      "learning_rate": 0.0001804639551966114,
      "loss": 3.7897,
      "step": 104790
    },
    {
      "epoch": 0.0976905574688824,
      "grad_norm": 1.2150121927261353,
      "learning_rate": 0.0001804620908745162,
      "loss": 3.8915,
      "step": 104800
    },
    {
      "epoch": 0.0976998790869615,
      "grad_norm": 0.9846599698066711,
      "learning_rate": 0.00018046022655242102,
      "loss": 3.799,
      "step": 104810
    },
    {
      "epoch": 0.09770920070504058,
      "grad_norm": 0.9624829888343811,
      "learning_rate": 0.00018045836223032583,
      "loss": 3.6898,
      "step": 104820
    },
    {
      "epoch": 0.09771852232311967,
      "grad_norm": 1.1162240505218506,
      "learning_rate": 0.00018045649790823063,
      "loss": 3.7898,
      "step": 104830
    },
    {
      "epoch": 0.09772784394119877,
      "grad_norm": 1.0180253982543945,
      "learning_rate": 0.0001804546335861354,
      "loss": 3.6798,
      "step": 104840
    },
    {
      "epoch": 0.09773716555927786,
      "grad_norm": 1.0325411558151245,
      "learning_rate": 0.00018045276926404022,
      "loss": 3.8875,
      "step": 104850
    },
    {
      "epoch": 0.09774648717735696,
      "grad_norm": 1.092024803161621,
      "learning_rate": 0.000180450904941945,
      "loss": 3.8329,
      "step": 104860
    },
    {
      "epoch": 0.09775580879543604,
      "grad_norm": 1.1876264810562134,
      "learning_rate": 0.00018044904061984983,
      "loss": 3.5155,
      "step": 104870
    },
    {
      "epoch": 0.09776513041351513,
      "grad_norm": 1.1976513862609863,
      "learning_rate": 0.00018044717629775464,
      "loss": 3.5286,
      "step": 104880
    },
    {
      "epoch": 0.09777445203159423,
      "grad_norm": 1.0586787462234497,
      "learning_rate": 0.00018044531197565942,
      "loss": 3.6041,
      "step": 104890
    },
    {
      "epoch": 0.09778377364967332,
      "grad_norm": 1.344719409942627,
      "learning_rate": 0.00018044344765356422,
      "loss": 3.642,
      "step": 104900
    },
    {
      "epoch": 0.09779309526775241,
      "grad_norm": 0.9974401593208313,
      "learning_rate": 0.000180441583331469,
      "loss": 3.7781,
      "step": 104910
    },
    {
      "epoch": 0.09780241688583151,
      "grad_norm": 0.931623101234436,
      "learning_rate": 0.0001804397190093738,
      "loss": 3.6794,
      "step": 104920
    },
    {
      "epoch": 0.09781173850391059,
      "grad_norm": 1.1607099771499634,
      "learning_rate": 0.00018043785468727864,
      "loss": 3.7586,
      "step": 104930
    },
    {
      "epoch": 0.09782106012198968,
      "grad_norm": 1.1590893268585205,
      "learning_rate": 0.00018043599036518342,
      "loss": 3.7315,
      "step": 104940
    },
    {
      "epoch": 0.09783038174006878,
      "grad_norm": 0.9411575198173523,
      "learning_rate": 0.00018043412604308823,
      "loss": 3.8159,
      "step": 104950
    },
    {
      "epoch": 0.09783970335814787,
      "grad_norm": 0.9421437382698059,
      "learning_rate": 0.000180432261720993,
      "loss": 3.698,
      "step": 104960
    },
    {
      "epoch": 0.09784902497622697,
      "grad_norm": 1.0782772302627563,
      "learning_rate": 0.0001804303973988978,
      "loss": 3.5689,
      "step": 104970
    },
    {
      "epoch": 0.09785834659430605,
      "grad_norm": 1.1582210063934326,
      "learning_rate": 0.00018042853307680262,
      "loss": 3.7809,
      "step": 104980
    },
    {
      "epoch": 0.09786766821238514,
      "grad_norm": 1.0753874778747559,
      "learning_rate": 0.00018042666875470742,
      "loss": 3.8476,
      "step": 104990
    },
    {
      "epoch": 0.09787698983046424,
      "grad_norm": 0.9098877310752869,
      "learning_rate": 0.00018042480443261223,
      "loss": 3.7662,
      "step": 105000
    },
    {
      "epoch": 0.09788631144854333,
      "grad_norm": 1.0815929174423218,
      "learning_rate": 0.000180422940110517,
      "loss": 3.87,
      "step": 105010
    },
    {
      "epoch": 0.09789563306662243,
      "grad_norm": 1.0709646940231323,
      "learning_rate": 0.00018042107578842182,
      "loss": 3.564,
      "step": 105020
    },
    {
      "epoch": 0.0979049546847015,
      "grad_norm": 0.9877784848213196,
      "learning_rate": 0.00018041921146632662,
      "loss": 3.7244,
      "step": 105030
    },
    {
      "epoch": 0.0979142763027806,
      "grad_norm": 1.0887128114700317,
      "learning_rate": 0.00018041734714423143,
      "loss": 3.8572,
      "step": 105040
    },
    {
      "epoch": 0.0979235979208597,
      "grad_norm": 0.9683422446250916,
      "learning_rate": 0.00018041548282213623,
      "loss": 3.6517,
      "step": 105050
    },
    {
      "epoch": 0.09793291953893879,
      "grad_norm": 0.9916858673095703,
      "learning_rate": 0.00018041361850004101,
      "loss": 3.7936,
      "step": 105060
    },
    {
      "epoch": 0.09794224115701788,
      "grad_norm": 1.0017085075378418,
      "learning_rate": 0.00018041175417794582,
      "loss": 3.6532,
      "step": 105070
    },
    {
      "epoch": 0.09795156277509696,
      "grad_norm": 1.1503726243972778,
      "learning_rate": 0.00018040988985585063,
      "loss": 3.7061,
      "step": 105080
    },
    {
      "epoch": 0.09796088439317606,
      "grad_norm": 1.1214351654052734,
      "learning_rate": 0.00018040802553375543,
      "loss": 3.8077,
      "step": 105090
    },
    {
      "epoch": 0.09797020601125515,
      "grad_norm": 1.095819115638733,
      "learning_rate": 0.00018040616121166024,
      "loss": 3.7731,
      "step": 105100
    },
    {
      "epoch": 0.09797952762933425,
      "grad_norm": 1.1331195831298828,
      "learning_rate": 0.00018040429688956502,
      "loss": 3.8144,
      "step": 105110
    },
    {
      "epoch": 0.09798884924741334,
      "grad_norm": 1.0663831233978271,
      "learning_rate": 0.00018040243256746982,
      "loss": 3.776,
      "step": 105120
    },
    {
      "epoch": 0.09799817086549242,
      "grad_norm": 1.2604777812957764,
      "learning_rate": 0.00018040056824537463,
      "loss": 3.5137,
      "step": 105130
    },
    {
      "epoch": 0.09800749248357152,
      "grad_norm": 1.032504677772522,
      "learning_rate": 0.0001803987039232794,
      "loss": 3.699,
      "step": 105140
    },
    {
      "epoch": 0.09801681410165061,
      "grad_norm": 0.9871731400489807,
      "learning_rate": 0.00018039683960118424,
      "loss": 3.7805,
      "step": 105150
    },
    {
      "epoch": 0.0980261357197297,
      "grad_norm": 1.0334243774414062,
      "learning_rate": 0.00018039497527908902,
      "loss": 3.9816,
      "step": 105160
    },
    {
      "epoch": 0.0980354573378088,
      "grad_norm": 1.2775459289550781,
      "learning_rate": 0.00018039311095699383,
      "loss": 3.8541,
      "step": 105170
    },
    {
      "epoch": 0.0980447789558879,
      "grad_norm": 0.9414418935775757,
      "learning_rate": 0.00018039124663489864,
      "loss": 3.5975,
      "step": 105180
    },
    {
      "epoch": 0.09805410057396698,
      "grad_norm": 0.9352957606315613,
      "learning_rate": 0.00018038938231280341,
      "loss": 3.8271,
      "step": 105190
    },
    {
      "epoch": 0.09806342219204607,
      "grad_norm": 1.336381435394287,
      "learning_rate": 0.00018038751799070822,
      "loss": 3.6858,
      "step": 105200
    },
    {
      "epoch": 0.09807274381012517,
      "grad_norm": 1.0350475311279297,
      "learning_rate": 0.00018038565366861303,
      "loss": 3.5814,
      "step": 105210
    },
    {
      "epoch": 0.09808206542820426,
      "grad_norm": 0.965672492980957,
      "learning_rate": 0.00018038378934651783,
      "loss": 3.9044,
      "step": 105220
    },
    {
      "epoch": 0.09809138704628335,
      "grad_norm": 0.9401379823684692,
      "learning_rate": 0.00018038192502442264,
      "loss": 3.7992,
      "step": 105230
    },
    {
      "epoch": 0.09810070866436243,
      "grad_norm": 0.9324325919151306,
      "learning_rate": 0.00018038006070232742,
      "loss": 3.5699,
      "step": 105240
    },
    {
      "epoch": 0.09811003028244153,
      "grad_norm": 0.8713799118995667,
      "learning_rate": 0.00018037819638023223,
      "loss": 3.9474,
      "step": 105250
    },
    {
      "epoch": 0.09811935190052062,
      "grad_norm": 1.1184674501419067,
      "learning_rate": 0.00018037633205813703,
      "loss": 3.9247,
      "step": 105260
    },
    {
      "epoch": 0.09812867351859972,
      "grad_norm": 1.0968998670578003,
      "learning_rate": 0.00018037446773604184,
      "loss": 3.4454,
      "step": 105270
    },
    {
      "epoch": 0.09813799513667881,
      "grad_norm": 0.9780281186103821,
      "learning_rate": 0.00018037260341394664,
      "loss": 3.6435,
      "step": 105280
    },
    {
      "epoch": 0.0981473167547579,
      "grad_norm": 1.0141687393188477,
      "learning_rate": 0.00018037073909185142,
      "loss": 3.7263,
      "step": 105290
    },
    {
      "epoch": 0.09815663837283699,
      "grad_norm": 0.9992983341217041,
      "learning_rate": 0.00018036887476975623,
      "loss": 3.5796,
      "step": 105300
    },
    {
      "epoch": 0.09816595999091608,
      "grad_norm": 1.0883270502090454,
      "learning_rate": 0.000180367010447661,
      "loss": 3.6127,
      "step": 105310
    },
    {
      "epoch": 0.09817528160899518,
      "grad_norm": 1.049718976020813,
      "learning_rate": 0.00018036514612556584,
      "loss": 3.7438,
      "step": 105320
    },
    {
      "epoch": 0.09818460322707427,
      "grad_norm": 1.0041917562484741,
      "learning_rate": 0.00018036328180347065,
      "loss": 3.826,
      "step": 105330
    },
    {
      "epoch": 0.09819392484515335,
      "grad_norm": 1.0124168395996094,
      "learning_rate": 0.00018036141748137543,
      "loss": 3.6944,
      "step": 105340
    },
    {
      "epoch": 0.09820324646323245,
      "grad_norm": 0.936219334602356,
      "learning_rate": 0.00018035955315928023,
      "loss": 3.547,
      "step": 105350
    },
    {
      "epoch": 0.09821256808131154,
      "grad_norm": 1.1824873685836792,
      "learning_rate": 0.000180357688837185,
      "loss": 3.8256,
      "step": 105360
    },
    {
      "epoch": 0.09822188969939064,
      "grad_norm": 0.9765459299087524,
      "learning_rate": 0.00018035582451508982,
      "loss": 3.96,
      "step": 105370
    },
    {
      "epoch": 0.09823121131746973,
      "grad_norm": 1.1354502439498901,
      "learning_rate": 0.00018035396019299465,
      "loss": 3.7307,
      "step": 105380
    },
    {
      "epoch": 0.09824053293554882,
      "grad_norm": 1.1176772117614746,
      "learning_rate": 0.00018035209587089943,
      "loss": 3.5949,
      "step": 105390
    },
    {
      "epoch": 0.0982498545536279,
      "grad_norm": 1.237705945968628,
      "learning_rate": 0.00018035023154880424,
      "loss": 3.573,
      "step": 105400
    },
    {
      "epoch": 0.098259176171707,
      "grad_norm": 1.1277174949645996,
      "learning_rate": 0.00018034836722670902,
      "loss": 3.6708,
      "step": 105410
    },
    {
      "epoch": 0.0982684977897861,
      "grad_norm": 0.9561507105827332,
      "learning_rate": 0.00018034650290461382,
      "loss": 3.4807,
      "step": 105420
    },
    {
      "epoch": 0.09827781940786519,
      "grad_norm": 1.0031222105026245,
      "learning_rate": 0.00018034463858251863,
      "loss": 3.6861,
      "step": 105430
    },
    {
      "epoch": 0.09828714102594428,
      "grad_norm": 1.042980670928955,
      "learning_rate": 0.00018034277426042344,
      "loss": 3.6651,
      "step": 105440
    },
    {
      "epoch": 0.09829646264402336,
      "grad_norm": 0.9781773090362549,
      "learning_rate": 0.00018034090993832824,
      "loss": 3.7957,
      "step": 105450
    },
    {
      "epoch": 0.09830578426210246,
      "grad_norm": 1.0476962327957153,
      "learning_rate": 0.00018033904561623302,
      "loss": 3.7871,
      "step": 105460
    },
    {
      "epoch": 0.09831510588018155,
      "grad_norm": 1.0546027421951294,
      "learning_rate": 0.00018033718129413783,
      "loss": 3.6355,
      "step": 105470
    },
    {
      "epoch": 0.09832442749826065,
      "grad_norm": 0.9248684644699097,
      "learning_rate": 0.00018033531697204264,
      "loss": 3.7191,
      "step": 105480
    },
    {
      "epoch": 0.09833374911633974,
      "grad_norm": 0.9881425499916077,
      "learning_rate": 0.00018033345264994744,
      "loss": 3.6249,
      "step": 105490
    },
    {
      "epoch": 0.09834307073441882,
      "grad_norm": 1.0754468441009521,
      "learning_rate": 0.00018033158832785225,
      "loss": 3.713,
      "step": 105500
    },
    {
      "epoch": 0.09835239235249792,
      "grad_norm": 1.1049453020095825,
      "learning_rate": 0.00018032972400575703,
      "loss": 3.6456,
      "step": 105510
    },
    {
      "epoch": 0.09836171397057701,
      "grad_norm": 1.1433489322662354,
      "learning_rate": 0.00018032785968366183,
      "loss": 3.6239,
      "step": 105520
    },
    {
      "epoch": 0.0983710355886561,
      "grad_norm": 1.0241855382919312,
      "learning_rate": 0.00018032599536156664,
      "loss": 3.6635,
      "step": 105530
    },
    {
      "epoch": 0.0983803572067352,
      "grad_norm": 1.2177283763885498,
      "learning_rate": 0.00018032413103947145,
      "loss": 3.6407,
      "step": 105540
    },
    {
      "epoch": 0.09838967882481428,
      "grad_norm": 0.9144358038902283,
      "learning_rate": 0.00018032226671737625,
      "loss": 3.6848,
      "step": 105550
    },
    {
      "epoch": 0.09839900044289338,
      "grad_norm": 1.0301216840744019,
      "learning_rate": 0.00018032040239528103,
      "loss": 3.8623,
      "step": 105560
    },
    {
      "epoch": 0.09840832206097247,
      "grad_norm": 1.1048530340194702,
      "learning_rate": 0.00018031853807318584,
      "loss": 3.8069,
      "step": 105570
    },
    {
      "epoch": 0.09841764367905156,
      "grad_norm": 1.0757261514663696,
      "learning_rate": 0.00018031667375109064,
      "loss": 3.9436,
      "step": 105580
    },
    {
      "epoch": 0.09842696529713066,
      "grad_norm": 1.0112098455429077,
      "learning_rate": 0.00018031480942899542,
      "loss": 3.7479,
      "step": 105590
    },
    {
      "epoch": 0.09843628691520975,
      "grad_norm": 0.9333658814430237,
      "learning_rate": 0.00018031294510690026,
      "loss": 3.7987,
      "step": 105600
    },
    {
      "epoch": 0.09844560853328883,
      "grad_norm": 1.1522690057754517,
      "learning_rate": 0.00018031108078480504,
      "loss": 3.7472,
      "step": 105610
    },
    {
      "epoch": 0.09845493015136793,
      "grad_norm": 1.1596989631652832,
      "learning_rate": 0.00018030921646270984,
      "loss": 3.7995,
      "step": 105620
    },
    {
      "epoch": 0.09846425176944702,
      "grad_norm": 1.307211995124817,
      "learning_rate": 0.00018030735214061465,
      "loss": 3.5983,
      "step": 105630
    },
    {
      "epoch": 0.09847357338752612,
      "grad_norm": 1.0538874864578247,
      "learning_rate": 0.00018030548781851943,
      "loss": 3.8026,
      "step": 105640
    },
    {
      "epoch": 0.09848289500560521,
      "grad_norm": 1.185954213142395,
      "learning_rate": 0.00018030362349642423,
      "loss": 3.601,
      "step": 105650
    },
    {
      "epoch": 0.09849221662368429,
      "grad_norm": 1.146284580230713,
      "learning_rate": 0.00018030175917432904,
      "loss": 3.4955,
      "step": 105660
    },
    {
      "epoch": 0.09850153824176339,
      "grad_norm": 1.1101731061935425,
      "learning_rate": 0.00018029989485223385,
      "loss": 3.6758,
      "step": 105670
    },
    {
      "epoch": 0.09851085985984248,
      "grad_norm": 1.118294596672058,
      "learning_rate": 0.00018029803053013863,
      "loss": 3.6323,
      "step": 105680
    },
    {
      "epoch": 0.09852018147792158,
      "grad_norm": 1.0319796800613403,
      "learning_rate": 0.00018029616620804343,
      "loss": 3.5577,
      "step": 105690
    },
    {
      "epoch": 0.09852950309600067,
      "grad_norm": 1.1922307014465332,
      "learning_rate": 0.00018029430188594824,
      "loss": 3.6437,
      "step": 105700
    },
    {
      "epoch": 0.09853882471407975,
      "grad_norm": 0.8984537124633789,
      "learning_rate": 0.00018029243756385304,
      "loss": 3.9047,
      "step": 105710
    },
    {
      "epoch": 0.09854814633215885,
      "grad_norm": 1.1186201572418213,
      "learning_rate": 0.00018029057324175785,
      "loss": 3.7839,
      "step": 105720
    },
    {
      "epoch": 0.09855746795023794,
      "grad_norm": 1.3012237548828125,
      "learning_rate": 0.00018028870891966263,
      "loss": 3.7767,
      "step": 105730
    },
    {
      "epoch": 0.09856678956831703,
      "grad_norm": 1.078872561454773,
      "learning_rate": 0.00018028684459756744,
      "loss": 3.6956,
      "step": 105740
    },
    {
      "epoch": 0.09857611118639613,
      "grad_norm": 1.1860486268997192,
      "learning_rate": 0.00018028498027547224,
      "loss": 3.6101,
      "step": 105750
    },
    {
      "epoch": 0.09858543280447521,
      "grad_norm": 1.1512666940689087,
      "learning_rate": 0.00018028311595337702,
      "loss": 3.8271,
      "step": 105760
    },
    {
      "epoch": 0.0985947544225543,
      "grad_norm": 1.0431110858917236,
      "learning_rate": 0.00018028125163128186,
      "loss": 3.464,
      "step": 105770
    },
    {
      "epoch": 0.0986040760406334,
      "grad_norm": 0.9834201335906982,
      "learning_rate": 0.00018027938730918663,
      "loss": 3.6848,
      "step": 105780
    },
    {
      "epoch": 0.09861339765871249,
      "grad_norm": 0.9731853008270264,
      "learning_rate": 0.00018027752298709144,
      "loss": 3.6525,
      "step": 105790
    },
    {
      "epoch": 0.09862271927679159,
      "grad_norm": 1.3217378854751587,
      "learning_rate": 0.00018027565866499625,
      "loss": 3.5122,
      "step": 105800
    },
    {
      "epoch": 0.09863204089487068,
      "grad_norm": 1.0974305868148804,
      "learning_rate": 0.00018027379434290103,
      "loss": 3.5942,
      "step": 105810
    },
    {
      "epoch": 0.09864136251294976,
      "grad_norm": 1.0521605014801025,
      "learning_rate": 0.00018027193002080583,
      "loss": 3.9295,
      "step": 105820
    },
    {
      "epoch": 0.09865068413102886,
      "grad_norm": 0.8716318607330322,
      "learning_rate": 0.00018027006569871064,
      "loss": 3.8087,
      "step": 105830
    },
    {
      "epoch": 0.09866000574910795,
      "grad_norm": 1.1141932010650635,
      "learning_rate": 0.00018026820137661545,
      "loss": 3.6791,
      "step": 105840
    },
    {
      "epoch": 0.09866932736718705,
      "grad_norm": 1.1976470947265625,
      "learning_rate": 0.00018026633705452025,
      "loss": 3.6879,
      "step": 105850
    },
    {
      "epoch": 0.09867864898526614,
      "grad_norm": 1.1412166357040405,
      "learning_rate": 0.00018026447273242503,
      "loss": 3.7054,
      "step": 105860
    },
    {
      "epoch": 0.09868797060334522,
      "grad_norm": 1.1318199634552002,
      "learning_rate": 0.00018026260841032984,
      "loss": 3.5464,
      "step": 105870
    },
    {
      "epoch": 0.09869729222142432,
      "grad_norm": 1.0346331596374512,
      "learning_rate": 0.00018026074408823464,
      "loss": 3.6772,
      "step": 105880
    },
    {
      "epoch": 0.09870661383950341,
      "grad_norm": 1.1086366176605225,
      "learning_rate": 0.00018025887976613945,
      "loss": 3.7427,
      "step": 105890
    },
    {
      "epoch": 0.0987159354575825,
      "grad_norm": 0.9982828497886658,
      "learning_rate": 0.00018025701544404426,
      "loss": 3.7535,
      "step": 105900
    },
    {
      "epoch": 0.0987252570756616,
      "grad_norm": 1.0790094137191772,
      "learning_rate": 0.00018025515112194904,
      "loss": 3.7806,
      "step": 105910
    },
    {
      "epoch": 0.09873457869374068,
      "grad_norm": 1.4834554195404053,
      "learning_rate": 0.00018025328679985384,
      "loss": 3.7071,
      "step": 105920
    },
    {
      "epoch": 0.09874390031181977,
      "grad_norm": 0.9863899350166321,
      "learning_rate": 0.00018025142247775865,
      "loss": 3.9059,
      "step": 105930
    },
    {
      "epoch": 0.09875322192989887,
      "grad_norm": 1.1010812520980835,
      "learning_rate": 0.00018024955815566345,
      "loss": 3.6057,
      "step": 105940
    },
    {
      "epoch": 0.09876254354797796,
      "grad_norm": 1.1287869215011597,
      "learning_rate": 0.00018024769383356826,
      "loss": 3.7445,
      "step": 105950
    },
    {
      "epoch": 0.09877186516605706,
      "grad_norm": 1.104053258895874,
      "learning_rate": 0.00018024582951147304,
      "loss": 3.7929,
      "step": 105960
    },
    {
      "epoch": 0.09878118678413614,
      "grad_norm": 1.1141895055770874,
      "learning_rate": 0.00018024396518937785,
      "loss": 3.8554,
      "step": 105970
    },
    {
      "epoch": 0.09879050840221523,
      "grad_norm": 1.044767141342163,
      "learning_rate": 0.00018024210086728263,
      "loss": 3.7084,
      "step": 105980
    },
    {
      "epoch": 0.09879983002029433,
      "grad_norm": 1.002427339553833,
      "learning_rate": 0.00018024023654518746,
      "loss": 3.8236,
      "step": 105990
    },
    {
      "epoch": 0.09880915163837342,
      "grad_norm": 0.8539226651191711,
      "learning_rate": 0.00018023837222309226,
      "loss": 3.875,
      "step": 106000
    },
    {
      "epoch": 0.09881847325645252,
      "grad_norm": 1.1202421188354492,
      "learning_rate": 0.00018023650790099704,
      "loss": 3.5343,
      "step": 106010
    },
    {
      "epoch": 0.0988277948745316,
      "grad_norm": 1.1502856016159058,
      "learning_rate": 0.00018023464357890185,
      "loss": 3.5065,
      "step": 106020
    },
    {
      "epoch": 0.09883711649261069,
      "grad_norm": 0.888993501663208,
      "learning_rate": 0.00018023277925680663,
      "loss": 3.69,
      "step": 106030
    },
    {
      "epoch": 0.09884643811068979,
      "grad_norm": 1.1740127801895142,
      "learning_rate": 0.00018023091493471144,
      "loss": 3.5646,
      "step": 106040
    },
    {
      "epoch": 0.09885575972876888,
      "grad_norm": 1.0609571933746338,
      "learning_rate": 0.00018022905061261627,
      "loss": 3.7538,
      "step": 106050
    },
    {
      "epoch": 0.09886508134684797,
      "grad_norm": 1.0041213035583496,
      "learning_rate": 0.00018022718629052105,
      "loss": 3.7879,
      "step": 106060
    },
    {
      "epoch": 0.09887440296492707,
      "grad_norm": 0.7898191213607788,
      "learning_rate": 0.00018022532196842585,
      "loss": 3.791,
      "step": 106070
    },
    {
      "epoch": 0.09888372458300615,
      "grad_norm": 1.231170415878296,
      "learning_rate": 0.00018022345764633063,
      "loss": 3.897,
      "step": 106080
    },
    {
      "epoch": 0.09889304620108524,
      "grad_norm": 1.0847232341766357,
      "learning_rate": 0.00018022159332423544,
      "loss": 3.6594,
      "step": 106090
    },
    {
      "epoch": 0.09890236781916434,
      "grad_norm": 1.3711880445480347,
      "learning_rate": 0.00018021972900214025,
      "loss": 3.6087,
      "step": 106100
    },
    {
      "epoch": 0.09891168943724343,
      "grad_norm": 1.0044792890548706,
      "learning_rate": 0.00018021786468004505,
      "loss": 3.8415,
      "step": 106110
    },
    {
      "epoch": 0.09892101105532253,
      "grad_norm": 0.9369828701019287,
      "learning_rate": 0.00018021600035794986,
      "loss": 3.6524,
      "step": 106120
    },
    {
      "epoch": 0.09893033267340161,
      "grad_norm": 1.1579294204711914,
      "learning_rate": 0.00018021413603585464,
      "loss": 3.6847,
      "step": 106130
    },
    {
      "epoch": 0.0989396542914807,
      "grad_norm": 1.0523531436920166,
      "learning_rate": 0.00018021227171375944,
      "loss": 3.7371,
      "step": 106140
    },
    {
      "epoch": 0.0989489759095598,
      "grad_norm": 1.022750973701477,
      "learning_rate": 0.00018021040739166425,
      "loss": 3.7047,
      "step": 106150
    },
    {
      "epoch": 0.09895829752763889,
      "grad_norm": 0.994208812713623,
      "learning_rate": 0.00018020854306956906,
      "loss": 3.7721,
      "step": 106160
    },
    {
      "epoch": 0.09896761914571799,
      "grad_norm": 1.2209678888320923,
      "learning_rate": 0.00018020667874747386,
      "loss": 3.7618,
      "step": 106170
    },
    {
      "epoch": 0.09897694076379707,
      "grad_norm": 0.8159819841384888,
      "learning_rate": 0.00018020481442537864,
      "loss": 3.6874,
      "step": 106180
    },
    {
      "epoch": 0.09898626238187616,
      "grad_norm": 1.1381783485412598,
      "learning_rate": 0.00018020295010328345,
      "loss": 3.5989,
      "step": 106190
    },
    {
      "epoch": 0.09899558399995526,
      "grad_norm": 1.0233732461929321,
      "learning_rate": 0.00018020108578118826,
      "loss": 3.7958,
      "step": 106200
    },
    {
      "epoch": 0.09900490561803435,
      "grad_norm": 0.9703245162963867,
      "learning_rate": 0.00018019922145909303,
      "loss": 3.6931,
      "step": 106210
    },
    {
      "epoch": 0.09901422723611344,
      "grad_norm": 1.086772084236145,
      "learning_rate": 0.00018019735713699787,
      "loss": 3.7151,
      "step": 106220
    },
    {
      "epoch": 0.09902354885419253,
      "grad_norm": 1.1132042407989502,
      "learning_rate": 0.00018019549281490265,
      "loss": 3.6661,
      "step": 106230
    },
    {
      "epoch": 0.09903287047227162,
      "grad_norm": 0.9134249687194824,
      "learning_rate": 0.00018019362849280745,
      "loss": 3.6261,
      "step": 106240
    },
    {
      "epoch": 0.09904219209035071,
      "grad_norm": 0.9865863919258118,
      "learning_rate": 0.00018019176417071226,
      "loss": 3.6984,
      "step": 106250
    },
    {
      "epoch": 0.09905151370842981,
      "grad_norm": 1.3195925951004028,
      "learning_rate": 0.00018018989984861704,
      "loss": 3.5645,
      "step": 106260
    },
    {
      "epoch": 0.0990608353265089,
      "grad_norm": 1.020058274269104,
      "learning_rate": 0.00018018803552652185,
      "loss": 3.6363,
      "step": 106270
    },
    {
      "epoch": 0.099070156944588,
      "grad_norm": 1.2452484369277954,
      "learning_rate": 0.00018018617120442665,
      "loss": 3.783,
      "step": 106280
    },
    {
      "epoch": 0.09907947856266708,
      "grad_norm": 1.0446162223815918,
      "learning_rate": 0.00018018430688233146,
      "loss": 3.7038,
      "step": 106290
    },
    {
      "epoch": 0.09908880018074617,
      "grad_norm": 0.973305344581604,
      "learning_rate": 0.00018018244256023626,
      "loss": 3.7163,
      "step": 106300
    },
    {
      "epoch": 0.09909812179882527,
      "grad_norm": 1.0658657550811768,
      "learning_rate": 0.00018018057823814104,
      "loss": 3.5948,
      "step": 106310
    },
    {
      "epoch": 0.09910744341690436,
      "grad_norm": 0.9843757748603821,
      "learning_rate": 0.00018017871391604585,
      "loss": 3.7589,
      "step": 106320
    },
    {
      "epoch": 0.09911676503498346,
      "grad_norm": 1.3059496879577637,
      "learning_rate": 0.00018017684959395066,
      "loss": 3.606,
      "step": 106330
    },
    {
      "epoch": 0.09912608665306254,
      "grad_norm": 1.0494920015335083,
      "learning_rate": 0.00018017498527185546,
      "loss": 3.6389,
      "step": 106340
    },
    {
      "epoch": 0.09913540827114163,
      "grad_norm": 1.121013879776001,
      "learning_rate": 0.00018017312094976027,
      "loss": 3.7912,
      "step": 106350
    },
    {
      "epoch": 0.09914472988922073,
      "grad_norm": 0.8219430446624756,
      "learning_rate": 0.00018017125662766505,
      "loss": 3.6903,
      "step": 106360
    },
    {
      "epoch": 0.09915405150729982,
      "grad_norm": 1.1031193733215332,
      "learning_rate": 0.00018016939230556985,
      "loss": 3.6598,
      "step": 106370
    },
    {
      "epoch": 0.09916337312537891,
      "grad_norm": 1.0747694969177246,
      "learning_rate": 0.00018016752798347466,
      "loss": 3.8032,
      "step": 106380
    },
    {
      "epoch": 0.099172694743458,
      "grad_norm": 1.0248548984527588,
      "learning_rate": 0.00018016566366137947,
      "loss": 3.6638,
      "step": 106390
    },
    {
      "epoch": 0.09918201636153709,
      "grad_norm": 1.0712661743164062,
      "learning_rate": 0.00018016379933928427,
      "loss": 3.6817,
      "step": 106400
    },
    {
      "epoch": 0.09919133797961618,
      "grad_norm": 1.1779589653015137,
      "learning_rate": 0.00018016193501718905,
      "loss": 3.7639,
      "step": 106410
    },
    {
      "epoch": 0.09920065959769528,
      "grad_norm": 0.9919792413711548,
      "learning_rate": 0.00018016007069509386,
      "loss": 3.7246,
      "step": 106420
    },
    {
      "epoch": 0.09920998121577437,
      "grad_norm": 1.0849759578704834,
      "learning_rate": 0.00018015820637299864,
      "loss": 3.6857,
      "step": 106430
    },
    {
      "epoch": 0.09921930283385345,
      "grad_norm": 1.0266817808151245,
      "learning_rate": 0.00018015634205090347,
      "loss": 3.5555,
      "step": 106440
    },
    {
      "epoch": 0.09922862445193255,
      "grad_norm": 1.1279040575027466,
      "learning_rate": 0.00018015447772880828,
      "loss": 3.7018,
      "step": 106450
    },
    {
      "epoch": 0.09923794607001164,
      "grad_norm": 1.1641026735305786,
      "learning_rate": 0.00018015261340671306,
      "loss": 3.6011,
      "step": 106460
    },
    {
      "epoch": 0.09924726768809074,
      "grad_norm": 1.1190099716186523,
      "learning_rate": 0.00018015074908461786,
      "loss": 3.6532,
      "step": 106470
    },
    {
      "epoch": 0.09925658930616983,
      "grad_norm": 1.126613974571228,
      "learning_rate": 0.00018014888476252264,
      "loss": 3.7547,
      "step": 106480
    },
    {
      "epoch": 0.09926591092424893,
      "grad_norm": 1.221089482307434,
      "learning_rate": 0.00018014702044042745,
      "loss": 3.7917,
      "step": 106490
    },
    {
      "epoch": 0.099275232542328,
      "grad_norm": 1.1383941173553467,
      "learning_rate": 0.00018014515611833228,
      "loss": 3.8187,
      "step": 106500
    },
    {
      "epoch": 0.0992845541604071,
      "grad_norm": 1.189008116722107,
      "learning_rate": 0.00018014329179623706,
      "loss": 3.5689,
      "step": 106510
    },
    {
      "epoch": 0.0992938757784862,
      "grad_norm": 1.068647861480713,
      "learning_rate": 0.00018014142747414187,
      "loss": 3.5281,
      "step": 106520
    },
    {
      "epoch": 0.09930319739656529,
      "grad_norm": 1.029610514640808,
      "learning_rate": 0.00018013956315204665,
      "loss": 3.7135,
      "step": 106530
    },
    {
      "epoch": 0.09931251901464438,
      "grad_norm": 1.0718638896942139,
      "learning_rate": 0.00018013769882995145,
      "loss": 3.8433,
      "step": 106540
    },
    {
      "epoch": 0.09932184063272347,
      "grad_norm": 0.9680432677268982,
      "learning_rate": 0.00018013583450785626,
      "loss": 3.5733,
      "step": 106550
    },
    {
      "epoch": 0.09933116225080256,
      "grad_norm": 1.3922830820083618,
      "learning_rate": 0.00018013397018576107,
      "loss": 3.5595,
      "step": 106560
    },
    {
      "epoch": 0.09934048386888165,
      "grad_norm": 1.0050193071365356,
      "learning_rate": 0.00018013210586366587,
      "loss": 3.7769,
      "step": 106570
    },
    {
      "epoch": 0.09934980548696075,
      "grad_norm": 1.1745814085006714,
      "learning_rate": 0.00018013024154157065,
      "loss": 3.6588,
      "step": 106580
    },
    {
      "epoch": 0.09935912710503984,
      "grad_norm": 0.9548517465591431,
      "learning_rate": 0.00018012837721947546,
      "loss": 3.6787,
      "step": 106590
    },
    {
      "epoch": 0.09936844872311892,
      "grad_norm": 1.1281344890594482,
      "learning_rate": 0.00018012651289738026,
      "loss": 3.7003,
      "step": 106600
    },
    {
      "epoch": 0.09937777034119802,
      "grad_norm": 1.1598832607269287,
      "learning_rate": 0.00018012464857528507,
      "loss": 3.7149,
      "step": 106610
    },
    {
      "epoch": 0.09938709195927711,
      "grad_norm": 1.0085835456848145,
      "learning_rate": 0.00018012278425318988,
      "loss": 3.6924,
      "step": 106620
    },
    {
      "epoch": 0.09939641357735621,
      "grad_norm": 1.1990041732788086,
      "learning_rate": 0.00018012091993109466,
      "loss": 3.7928,
      "step": 106630
    },
    {
      "epoch": 0.0994057351954353,
      "grad_norm": 0.9306849241256714,
      "learning_rate": 0.00018011905560899946,
      "loss": 3.6328,
      "step": 106640
    },
    {
      "epoch": 0.09941505681351438,
      "grad_norm": 1.0121814012527466,
      "learning_rate": 0.00018011719128690427,
      "loss": 3.5758,
      "step": 106650
    },
    {
      "epoch": 0.09942437843159348,
      "grad_norm": 0.9391481280326843,
      "learning_rate": 0.00018011532696480907,
      "loss": 3.7226,
      "step": 106660
    },
    {
      "epoch": 0.09943370004967257,
      "grad_norm": 1.238592505455017,
      "learning_rate": 0.00018011346264271388,
      "loss": 3.6043,
      "step": 106670
    },
    {
      "epoch": 0.09944302166775167,
      "grad_norm": 1.002155065536499,
      "learning_rate": 0.00018011159832061866,
      "loss": 3.8265,
      "step": 106680
    },
    {
      "epoch": 0.09945234328583076,
      "grad_norm": 1.2854232788085938,
      "learning_rate": 0.00018010973399852347,
      "loss": 3.8708,
      "step": 106690
    },
    {
      "epoch": 0.09946166490390984,
      "grad_norm": 1.0453555583953857,
      "learning_rate": 0.00018010786967642827,
      "loss": 3.6218,
      "step": 106700
    },
    {
      "epoch": 0.09947098652198894,
      "grad_norm": 1.0291502475738525,
      "learning_rate": 0.00018010600535433305,
      "loss": 3.8023,
      "step": 106710
    },
    {
      "epoch": 0.09948030814006803,
      "grad_norm": 1.1512582302093506,
      "learning_rate": 0.00018010414103223789,
      "loss": 3.6877,
      "step": 106720
    },
    {
      "epoch": 0.09948962975814712,
      "grad_norm": 1.0598427057266235,
      "learning_rate": 0.00018010227671014266,
      "loss": 3.8517,
      "step": 106730
    },
    {
      "epoch": 0.09949895137622622,
      "grad_norm": 0.906507670879364,
      "learning_rate": 0.00018010041238804747,
      "loss": 3.8438,
      "step": 106740
    },
    {
      "epoch": 0.09950827299430531,
      "grad_norm": 0.9854740500450134,
      "learning_rate": 0.00018009854806595228,
      "loss": 3.8846,
      "step": 106750
    },
    {
      "epoch": 0.0995175946123844,
      "grad_norm": 1.0969552993774414,
      "learning_rate": 0.00018009668374385706,
      "loss": 3.7499,
      "step": 106760
    },
    {
      "epoch": 0.09952691623046349,
      "grad_norm": 0.9918996691703796,
      "learning_rate": 0.00018009481942176186,
      "loss": 3.6148,
      "step": 106770
    },
    {
      "epoch": 0.09953623784854258,
      "grad_norm": 1.230973482131958,
      "learning_rate": 0.00018009295509966667,
      "loss": 3.7928,
      "step": 106780
    },
    {
      "epoch": 0.09954555946662168,
      "grad_norm": 1.0464200973510742,
      "learning_rate": 0.00018009109077757148,
      "loss": 3.9009,
      "step": 106790
    },
    {
      "epoch": 0.09955488108470077,
      "grad_norm": 0.9816700220108032,
      "learning_rate": 0.00018008922645547628,
      "loss": 3.7459,
      "step": 106800
    },
    {
      "epoch": 0.09956420270277985,
      "grad_norm": 1.0193854570388794,
      "learning_rate": 0.00018008736213338106,
      "loss": 3.5686,
      "step": 106810
    },
    {
      "epoch": 0.09957352432085895,
      "grad_norm": 0.9022606015205383,
      "learning_rate": 0.00018008549781128587,
      "loss": 3.7378,
      "step": 106820
    },
    {
      "epoch": 0.09958284593893804,
      "grad_norm": 1.159156084060669,
      "learning_rate": 0.00018008363348919067,
      "loss": 3.7569,
      "step": 106830
    },
    {
      "epoch": 0.09959216755701714,
      "grad_norm": 1.1359755992889404,
      "learning_rate": 0.00018008176916709548,
      "loss": 3.6305,
      "step": 106840
    },
    {
      "epoch": 0.09960148917509623,
      "grad_norm": 1.0194653272628784,
      "learning_rate": 0.00018007990484500029,
      "loss": 3.6804,
      "step": 106850
    },
    {
      "epoch": 0.09961081079317531,
      "grad_norm": 0.9433748722076416,
      "learning_rate": 0.00018007804052290507,
      "loss": 3.6378,
      "step": 106860
    },
    {
      "epoch": 0.0996201324112544,
      "grad_norm": 1.012729525566101,
      "learning_rate": 0.00018007617620080987,
      "loss": 3.6695,
      "step": 106870
    },
    {
      "epoch": 0.0996294540293335,
      "grad_norm": 1.0138012170791626,
      "learning_rate": 0.00018007431187871465,
      "loss": 3.7479,
      "step": 106880
    },
    {
      "epoch": 0.0996387756474126,
      "grad_norm": 1.0906481742858887,
      "learning_rate": 0.00018007244755661948,
      "loss": 3.8414,
      "step": 106890
    },
    {
      "epoch": 0.09964809726549169,
      "grad_norm": 1.0845938920974731,
      "learning_rate": 0.0001800705832345243,
      "loss": 3.7443,
      "step": 106900
    },
    {
      "epoch": 0.09965741888357077,
      "grad_norm": 1.1989229917526245,
      "learning_rate": 0.00018006871891242907,
      "loss": 3.6118,
      "step": 106910
    },
    {
      "epoch": 0.09966674050164986,
      "grad_norm": 1.9637452363967896,
      "learning_rate": 0.00018006685459033388,
      "loss": 3.6394,
      "step": 106920
    },
    {
      "epoch": 0.09967606211972896,
      "grad_norm": 1.2148005962371826,
      "learning_rate": 0.00018006499026823866,
      "loss": 3.6428,
      "step": 106930
    },
    {
      "epoch": 0.09968538373780805,
      "grad_norm": 1.1533315181732178,
      "learning_rate": 0.00018006312594614346,
      "loss": 3.3042,
      "step": 106940
    },
    {
      "epoch": 0.09969470535588715,
      "grad_norm": 0.8998438119888306,
      "learning_rate": 0.0001800612616240483,
      "loss": 3.6309,
      "step": 106950
    },
    {
      "epoch": 0.09970402697396624,
      "grad_norm": 0.9662830233573914,
      "learning_rate": 0.00018005939730195307,
      "loss": 3.5775,
      "step": 106960
    },
    {
      "epoch": 0.09971334859204532,
      "grad_norm": 1.2111413478851318,
      "learning_rate": 0.00018005753297985788,
      "loss": 3.7073,
      "step": 106970
    },
    {
      "epoch": 0.09972267021012442,
      "grad_norm": 0.9689759612083435,
      "learning_rate": 0.00018005566865776266,
      "loss": 3.4761,
      "step": 106980
    },
    {
      "epoch": 0.09973199182820351,
      "grad_norm": 0.9757514595985413,
      "learning_rate": 0.00018005380433566747,
      "loss": 3.6631,
      "step": 106990
    },
    {
      "epoch": 0.0997413134462826,
      "grad_norm": 0.8579378128051758,
      "learning_rate": 0.00018005194001357227,
      "loss": 3.7373,
      "step": 107000
    },
    {
      "epoch": 0.0997506350643617,
      "grad_norm": 1.286089301109314,
      "learning_rate": 0.00018005007569147708,
      "loss": 3.5811,
      "step": 107010
    },
    {
      "epoch": 0.09975995668244078,
      "grad_norm": 1.015554428100586,
      "learning_rate": 0.00018004821136938188,
      "loss": 3.8193,
      "step": 107020
    },
    {
      "epoch": 0.09976927830051988,
      "grad_norm": 1.1219487190246582,
      "learning_rate": 0.00018004634704728666,
      "loss": 3.755,
      "step": 107030
    },
    {
      "epoch": 0.09977859991859897,
      "grad_norm": 1.2733485698699951,
      "learning_rate": 0.00018004448272519147,
      "loss": 3.7409,
      "step": 107040
    },
    {
      "epoch": 0.09978792153667806,
      "grad_norm": 0.9124148488044739,
      "learning_rate": 0.00018004261840309628,
      "loss": 3.8341,
      "step": 107050
    },
    {
      "epoch": 0.09979724315475716,
      "grad_norm": 0.9744992256164551,
      "learning_rate": 0.00018004075408100108,
      "loss": 3.7556,
      "step": 107060
    },
    {
      "epoch": 0.09980656477283624,
      "grad_norm": 1.0877479314804077,
      "learning_rate": 0.0001800388897589059,
      "loss": 3.713,
      "step": 107070
    },
    {
      "epoch": 0.09981588639091533,
      "grad_norm": 0.9365495443344116,
      "learning_rate": 0.00018003702543681067,
      "loss": 3.775,
      "step": 107080
    },
    {
      "epoch": 0.09982520800899443,
      "grad_norm": 1.101114273071289,
      "learning_rate": 0.00018003516111471547,
      "loss": 3.734,
      "step": 107090
    },
    {
      "epoch": 0.09983452962707352,
      "grad_norm": 1.0824581384658813,
      "learning_rate": 0.00018003329679262025,
      "loss": 3.7203,
      "step": 107100
    },
    {
      "epoch": 0.09984385124515262,
      "grad_norm": 1.229477882385254,
      "learning_rate": 0.0001800314324705251,
      "loss": 3.5881,
      "step": 107110
    },
    {
      "epoch": 0.0998531728632317,
      "grad_norm": 1.0856443643569946,
      "learning_rate": 0.0001800295681484299,
      "loss": 3.9148,
      "step": 107120
    },
    {
      "epoch": 0.09986249448131079,
      "grad_norm": 1.2435799837112427,
      "learning_rate": 0.00018002770382633467,
      "loss": 3.6482,
      "step": 107130
    },
    {
      "epoch": 0.09987181609938989,
      "grad_norm": 1.0931929349899292,
      "learning_rate": 0.00018002583950423948,
      "loss": 3.7836,
      "step": 107140
    },
    {
      "epoch": 0.09988113771746898,
      "grad_norm": 0.9801807999610901,
      "learning_rate": 0.00018002397518214426,
      "loss": 3.7423,
      "step": 107150
    },
    {
      "epoch": 0.09989045933554808,
      "grad_norm": 0.985741138458252,
      "learning_rate": 0.00018002211086004906,
      "loss": 3.768,
      "step": 107160
    },
    {
      "epoch": 0.09989978095362717,
      "grad_norm": 1.0901230573654175,
      "learning_rate": 0.0001800202465379539,
      "loss": 3.6891,
      "step": 107170
    },
    {
      "epoch": 0.09990910257170625,
      "grad_norm": 1.1373937129974365,
      "learning_rate": 0.00018001838221585868,
      "loss": 3.7203,
      "step": 107180
    },
    {
      "epoch": 0.09991842418978535,
      "grad_norm": 1.014169692993164,
      "learning_rate": 0.00018001651789376348,
      "loss": 3.5914,
      "step": 107190
    },
    {
      "epoch": 0.09992774580786444,
      "grad_norm": 1.0403484106063843,
      "learning_rate": 0.00018001465357166826,
      "loss": 3.5779,
      "step": 107200
    },
    {
      "epoch": 0.09993706742594353,
      "grad_norm": 1.189001441001892,
      "learning_rate": 0.00018001278924957307,
      "loss": 3.8396,
      "step": 107210
    },
    {
      "epoch": 0.09994638904402263,
      "grad_norm": 0.9387860298156738,
      "learning_rate": 0.00018001092492747788,
      "loss": 3.7092,
      "step": 107220
    },
    {
      "epoch": 0.09995571066210171,
      "grad_norm": 1.05007004737854,
      "learning_rate": 0.00018000906060538268,
      "loss": 3.8267,
      "step": 107230
    },
    {
      "epoch": 0.0999650322801808,
      "grad_norm": 1.0730382204055786,
      "learning_rate": 0.0001800071962832875,
      "loss": 3.6352,
      "step": 107240
    },
    {
      "epoch": 0.0999743538982599,
      "grad_norm": 1.0239431858062744,
      "learning_rate": 0.00018000533196119227,
      "loss": 3.8645,
      "step": 107250
    },
    {
      "epoch": 0.099983675516339,
      "grad_norm": 1.025657296180725,
      "learning_rate": 0.00018000346763909707,
      "loss": 3.7964,
      "step": 107260
    },
    {
      "epoch": 0.09999299713441809,
      "grad_norm": 1.4813494682312012,
      "learning_rate": 0.00018000160331700188,
      "loss": 3.6033,
      "step": 107270
    },
    {
      "epoch": 0.10000231875249717,
      "grad_norm": 1.1792864799499512,
      "learning_rate": 0.00017999973899490669,
      "loss": 3.7172,
      "step": 107280
    },
    {
      "epoch": 0.10001164037057626,
      "grad_norm": 0.9342910051345825,
      "learning_rate": 0.0001799978746728115,
      "loss": 3.7979,
      "step": 107290
    },
    {
      "epoch": 0.10002096198865536,
      "grad_norm": 0.8685641884803772,
      "learning_rate": 0.00017999601035071627,
      "loss": 3.8308,
      "step": 107300
    },
    {
      "epoch": 0.10003028360673445,
      "grad_norm": 1.0996421575546265,
      "learning_rate": 0.00017999414602862108,
      "loss": 3.7622,
      "step": 107310
    },
    {
      "epoch": 0.10003960522481355,
      "grad_norm": 1.2801930904388428,
      "learning_rate": 0.00017999228170652588,
      "loss": 3.7702,
      "step": 107320
    },
    {
      "epoch": 0.10004892684289263,
      "grad_norm": 1.1023914813995361,
      "learning_rate": 0.00017999041738443066,
      "loss": 3.9519,
      "step": 107330
    },
    {
      "epoch": 0.10005824846097172,
      "grad_norm": 1.0103981494903564,
      "learning_rate": 0.0001799885530623355,
      "loss": 3.58,
      "step": 107340
    },
    {
      "epoch": 0.10006757007905082,
      "grad_norm": 1.0844635963439941,
      "learning_rate": 0.00017998668874024028,
      "loss": 3.7306,
      "step": 107350
    },
    {
      "epoch": 0.10007689169712991,
      "grad_norm": 0.953117847442627,
      "learning_rate": 0.00017998482441814508,
      "loss": 3.7738,
      "step": 107360
    },
    {
      "epoch": 0.100086213315209,
      "grad_norm": 1.1063345670700073,
      "learning_rate": 0.0001799829600960499,
      "loss": 3.6357,
      "step": 107370
    },
    {
      "epoch": 0.10009553493328809,
      "grad_norm": 1.00814688205719,
      "learning_rate": 0.00017998109577395467,
      "loss": 3.6198,
      "step": 107380
    },
    {
      "epoch": 0.10010485655136718,
      "grad_norm": 1.0582672357559204,
      "learning_rate": 0.00017997923145185947,
      "loss": 3.6949,
      "step": 107390
    },
    {
      "epoch": 0.10011417816944627,
      "grad_norm": 1.0367306470870972,
      "learning_rate": 0.00017997736712976428,
      "loss": 3.7668,
      "step": 107400
    },
    {
      "epoch": 0.10012349978752537,
      "grad_norm": 0.9781925678253174,
      "learning_rate": 0.0001799755028076691,
      "loss": 3.6803,
      "step": 107410
    },
    {
      "epoch": 0.10013282140560446,
      "grad_norm": 1.1379441022872925,
      "learning_rate": 0.0001799736384855739,
      "loss": 3.6496,
      "step": 107420
    },
    {
      "epoch": 0.10014214302368356,
      "grad_norm": 1.0003819465637207,
      "learning_rate": 0.00017997177416347867,
      "loss": 3.6679,
      "step": 107430
    },
    {
      "epoch": 0.10015146464176264,
      "grad_norm": 1.0333974361419678,
      "learning_rate": 0.00017996990984138348,
      "loss": 3.7186,
      "step": 107440
    },
    {
      "epoch": 0.10016078625984173,
      "grad_norm": 1.042099952697754,
      "learning_rate": 0.00017996804551928829,
      "loss": 3.6559,
      "step": 107450
    },
    {
      "epoch": 0.10017010787792083,
      "grad_norm": 0.9886993765830994,
      "learning_rate": 0.0001799661811971931,
      "loss": 3.5745,
      "step": 107460
    },
    {
      "epoch": 0.10017942949599992,
      "grad_norm": 1.165614128112793,
      "learning_rate": 0.0001799643168750979,
      "loss": 3.5401,
      "step": 107470
    },
    {
      "epoch": 0.10018875111407902,
      "grad_norm": 1.1293797492980957,
      "learning_rate": 0.00017996245255300268,
      "loss": 3.6087,
      "step": 107480
    },
    {
      "epoch": 0.1001980727321581,
      "grad_norm": 1.0487021207809448,
      "learning_rate": 0.00017996058823090748,
      "loss": 3.5616,
      "step": 107490
    },
    {
      "epoch": 0.10020739435023719,
      "grad_norm": 1.0545425415039062,
      "learning_rate": 0.0001799587239088123,
      "loss": 3.7112,
      "step": 107500
    },
    {
      "epoch": 0.10021671596831629,
      "grad_norm": 1.198007583618164,
      "learning_rate": 0.0001799568595867171,
      "loss": 3.7247,
      "step": 107510
    },
    {
      "epoch": 0.10022603758639538,
      "grad_norm": 0.9850670099258423,
      "learning_rate": 0.0001799549952646219,
      "loss": 3.6823,
      "step": 107520
    },
    {
      "epoch": 0.10023535920447448,
      "grad_norm": 0.9616518616676331,
      "learning_rate": 0.00017995313094252668,
      "loss": 3.7349,
      "step": 107530
    },
    {
      "epoch": 0.10024468082255356,
      "grad_norm": 1.0497339963912964,
      "learning_rate": 0.0001799512666204315,
      "loss": 3.6608,
      "step": 107540
    },
    {
      "epoch": 0.10025400244063265,
      "grad_norm": 1.138270378112793,
      "learning_rate": 0.00017994940229833627,
      "loss": 3.7248,
      "step": 107550
    },
    {
      "epoch": 0.10026332405871174,
      "grad_norm": 1.1290243864059448,
      "learning_rate": 0.0001799475379762411,
      "loss": 3.6034,
      "step": 107560
    },
    {
      "epoch": 0.10027264567679084,
      "grad_norm": 1.098375678062439,
      "learning_rate": 0.0001799456736541459,
      "loss": 3.5095,
      "step": 107570
    },
    {
      "epoch": 0.10028196729486993,
      "grad_norm": 1.1036404371261597,
      "learning_rate": 0.00017994380933205069,
      "loss": 3.7464,
      "step": 107580
    },
    {
      "epoch": 0.10029128891294901,
      "grad_norm": 1.274667501449585,
      "learning_rate": 0.0001799419450099555,
      "loss": 3.6647,
      "step": 107590
    },
    {
      "epoch": 0.10030061053102811,
      "grad_norm": 0.989228367805481,
      "learning_rate": 0.00017994008068786027,
      "loss": 3.5534,
      "step": 107600
    },
    {
      "epoch": 0.1003099321491072,
      "grad_norm": 0.9573204517364502,
      "learning_rate": 0.00017993821636576508,
      "loss": 3.9633,
      "step": 107610
    },
    {
      "epoch": 0.1003192537671863,
      "grad_norm": 1.1043096780776978,
      "learning_rate": 0.0001799363520436699,
      "loss": 3.8057,
      "step": 107620
    },
    {
      "epoch": 0.10032857538526539,
      "grad_norm": 1.134360671043396,
      "learning_rate": 0.0001799344877215747,
      "loss": 3.8635,
      "step": 107630
    },
    {
      "epoch": 0.10033789700334449,
      "grad_norm": 1.1422977447509766,
      "learning_rate": 0.0001799326233994795,
      "loss": 3.7102,
      "step": 107640
    },
    {
      "epoch": 0.10034721862142357,
      "grad_norm": 1.083322286605835,
      "learning_rate": 0.00017993075907738428,
      "loss": 3.8491,
      "step": 107650
    },
    {
      "epoch": 0.10035654023950266,
      "grad_norm": 1.077358603477478,
      "learning_rate": 0.00017992889475528908,
      "loss": 3.8231,
      "step": 107660
    },
    {
      "epoch": 0.10036586185758176,
      "grad_norm": 1.1758109331130981,
      "learning_rate": 0.0001799270304331939,
      "loss": 3.4831,
      "step": 107670
    },
    {
      "epoch": 0.10037518347566085,
      "grad_norm": 1.071409821510315,
      "learning_rate": 0.0001799251661110987,
      "loss": 3.5004,
      "step": 107680
    },
    {
      "epoch": 0.10038450509373995,
      "grad_norm": 1.1180040836334229,
      "learning_rate": 0.0001799233017890035,
      "loss": 3.731,
      "step": 107690
    },
    {
      "epoch": 0.10039382671181903,
      "grad_norm": 1.2714486122131348,
      "learning_rate": 0.00017992143746690828,
      "loss": 3.6638,
      "step": 107700
    },
    {
      "epoch": 0.10040314832989812,
      "grad_norm": 1.039380669593811,
      "learning_rate": 0.0001799195731448131,
      "loss": 3.6919,
      "step": 107710
    },
    {
      "epoch": 0.10041246994797721,
      "grad_norm": 1.037070393562317,
      "learning_rate": 0.0001799177088227179,
      "loss": 3.6833,
      "step": 107720
    },
    {
      "epoch": 0.10042179156605631,
      "grad_norm": 1.168562412261963,
      "learning_rate": 0.0001799158445006227,
      "loss": 3.7215,
      "step": 107730
    },
    {
      "epoch": 0.1004311131841354,
      "grad_norm": 1.1563178300857544,
      "learning_rate": 0.0001799139801785275,
      "loss": 3.6454,
      "step": 107740
    },
    {
      "epoch": 0.10044043480221448,
      "grad_norm": 1.056588888168335,
      "learning_rate": 0.00017991211585643228,
      "loss": 3.7088,
      "step": 107750
    },
    {
      "epoch": 0.10044975642029358,
      "grad_norm": 1.0655643939971924,
      "learning_rate": 0.0001799102515343371,
      "loss": 3.7646,
      "step": 107760
    },
    {
      "epoch": 0.10045907803837267,
      "grad_norm": 1.0675628185272217,
      "learning_rate": 0.0001799083872122419,
      "loss": 3.9108,
      "step": 107770
    },
    {
      "epoch": 0.10046839965645177,
      "grad_norm": 1.0365623235702515,
      "learning_rate": 0.00017990652289014668,
      "loss": 3.7382,
      "step": 107780
    },
    {
      "epoch": 0.10047772127453086,
      "grad_norm": 1.2056467533111572,
      "learning_rate": 0.0001799046585680515,
      "loss": 3.7309,
      "step": 107790
    },
    {
      "epoch": 0.10048704289260994,
      "grad_norm": 0.9311197400093079,
      "learning_rate": 0.0001799027942459563,
      "loss": 3.7001,
      "step": 107800
    },
    {
      "epoch": 0.10049636451068904,
      "grad_norm": 1.3372389078140259,
      "learning_rate": 0.0001799009299238611,
      "loss": 3.5934,
      "step": 107810
    },
    {
      "epoch": 0.10050568612876813,
      "grad_norm": 1.00193452835083,
      "learning_rate": 0.0001798990656017659,
      "loss": 3.7019,
      "step": 107820
    },
    {
      "epoch": 0.10051500774684723,
      "grad_norm": 1.1119229793548584,
      "learning_rate": 0.00017989720127967068,
      "loss": 3.8505,
      "step": 107830
    },
    {
      "epoch": 0.10052432936492632,
      "grad_norm": 1.2607334852218628,
      "learning_rate": 0.0001798953369575755,
      "loss": 3.7346,
      "step": 107840
    },
    {
      "epoch": 0.10053365098300542,
      "grad_norm": 0.8511230945587158,
      "learning_rate": 0.0001798934726354803,
      "loss": 3.7308,
      "step": 107850
    },
    {
      "epoch": 0.1005429726010845,
      "grad_norm": 1.249578595161438,
      "learning_rate": 0.0001798916083133851,
      "loss": 3.6557,
      "step": 107860
    },
    {
      "epoch": 0.10055229421916359,
      "grad_norm": 0.891692042350769,
      "learning_rate": 0.0001798897439912899,
      "loss": 3.8437,
      "step": 107870
    },
    {
      "epoch": 0.10056161583724268,
      "grad_norm": 1.0458861589431763,
      "learning_rate": 0.00017988787966919469,
      "loss": 3.85,
      "step": 107880
    },
    {
      "epoch": 0.10057093745532178,
      "grad_norm": 1.0889415740966797,
      "learning_rate": 0.0001798860153470995,
      "loss": 3.7014,
      "step": 107890
    },
    {
      "epoch": 0.10058025907340087,
      "grad_norm": 1.2391018867492676,
      "learning_rate": 0.0001798841510250043,
      "loss": 3.7127,
      "step": 107900
    },
    {
      "epoch": 0.10058958069147995,
      "grad_norm": 0.9557119011878967,
      "learning_rate": 0.0001798822867029091,
      "loss": 3.6071,
      "step": 107910
    },
    {
      "epoch": 0.10059890230955905,
      "grad_norm": 0.9779342412948608,
      "learning_rate": 0.0001798804223808139,
      "loss": 3.5377,
      "step": 107920
    },
    {
      "epoch": 0.10060822392763814,
      "grad_norm": 1.1976547241210938,
      "learning_rate": 0.0001798785580587187,
      "loss": 3.6907,
      "step": 107930
    },
    {
      "epoch": 0.10061754554571724,
      "grad_norm": 0.9352059364318848,
      "learning_rate": 0.0001798766937366235,
      "loss": 3.6494,
      "step": 107940
    },
    {
      "epoch": 0.10062686716379633,
      "grad_norm": 1.112508773803711,
      "learning_rate": 0.0001798748294145283,
      "loss": 3.8057,
      "step": 107950
    },
    {
      "epoch": 0.10063618878187541,
      "grad_norm": 1.1806845664978027,
      "learning_rate": 0.0001798729650924331,
      "loss": 3.7495,
      "step": 107960
    },
    {
      "epoch": 0.10064551039995451,
      "grad_norm": 1.0963103771209717,
      "learning_rate": 0.00017987110077033792,
      "loss": 3.6374,
      "step": 107970
    },
    {
      "epoch": 0.1006548320180336,
      "grad_norm": 1.1786445379257202,
      "learning_rate": 0.0001798692364482427,
      "loss": 3.8114,
      "step": 107980
    },
    {
      "epoch": 0.1006641536361127,
      "grad_norm": 1.1156724691390991,
      "learning_rate": 0.0001798673721261475,
      "loss": 3.7889,
      "step": 107990
    },
    {
      "epoch": 0.10067347525419179,
      "grad_norm": 1.0324138402938843,
      "learning_rate": 0.00017986550780405228,
      "loss": 3.6153,
      "step": 108000
    },
    {
      "epoch": 0.10068279687227087,
      "grad_norm": 1.0841751098632812,
      "learning_rate": 0.0001798636434819571,
      "loss": 3.8913,
      "step": 108010
    },
    {
      "epoch": 0.10069211849034997,
      "grad_norm": 0.9374014735221863,
      "learning_rate": 0.00017986177915986192,
      "loss": 3.8629,
      "step": 108020
    },
    {
      "epoch": 0.10070144010842906,
      "grad_norm": 1.1949487924575806,
      "learning_rate": 0.0001798599148377667,
      "loss": 3.8487,
      "step": 108030
    },
    {
      "epoch": 0.10071076172650816,
      "grad_norm": 1.2141059637069702,
      "learning_rate": 0.0001798580505156715,
      "loss": 3.4698,
      "step": 108040
    },
    {
      "epoch": 0.10072008334458725,
      "grad_norm": 1.2071421146392822,
      "learning_rate": 0.00017985618619357628,
      "loss": 3.6836,
      "step": 108050
    },
    {
      "epoch": 0.10072940496266633,
      "grad_norm": 0.9356071949005127,
      "learning_rate": 0.0001798543218714811,
      "loss": 3.7482,
      "step": 108060
    },
    {
      "epoch": 0.10073872658074542,
      "grad_norm": 1.097862958908081,
      "learning_rate": 0.00017985245754938592,
      "loss": 3.5066,
      "step": 108070
    },
    {
      "epoch": 0.10074804819882452,
      "grad_norm": 1.1546932458877563,
      "learning_rate": 0.0001798505932272907,
      "loss": 3.6614,
      "step": 108080
    },
    {
      "epoch": 0.10075736981690361,
      "grad_norm": 1.0408028364181519,
      "learning_rate": 0.0001798487289051955,
      "loss": 3.6335,
      "step": 108090
    },
    {
      "epoch": 0.10076669143498271,
      "grad_norm": 1.0392898321151733,
      "learning_rate": 0.0001798468645831003,
      "loss": 3.755,
      "step": 108100
    },
    {
      "epoch": 0.1007760130530618,
      "grad_norm": 1.5802407264709473,
      "learning_rate": 0.0001798450002610051,
      "loss": 3.7402,
      "step": 108110
    },
    {
      "epoch": 0.10078533467114088,
      "grad_norm": 1.0758357048034668,
      "learning_rate": 0.0001798431359389099,
      "loss": 3.6961,
      "step": 108120
    },
    {
      "epoch": 0.10079465628921998,
      "grad_norm": 1.1538902521133423,
      "learning_rate": 0.0001798412716168147,
      "loss": 3.786,
      "step": 108130
    },
    {
      "epoch": 0.10080397790729907,
      "grad_norm": 1.01381254196167,
      "learning_rate": 0.00017983940729471951,
      "loss": 3.4008,
      "step": 108140
    },
    {
      "epoch": 0.10081329952537817,
      "grad_norm": 1.0850043296813965,
      "learning_rate": 0.0001798375429726243,
      "loss": 3.7189,
      "step": 108150
    },
    {
      "epoch": 0.10082262114345726,
      "grad_norm": 1.1709634065628052,
      "learning_rate": 0.0001798356786505291,
      "loss": 3.5754,
      "step": 108160
    },
    {
      "epoch": 0.10083194276153634,
      "grad_norm": 1.15822172164917,
      "learning_rate": 0.0001798338143284339,
      "loss": 3.6021,
      "step": 108170
    },
    {
      "epoch": 0.10084126437961544,
      "grad_norm": 1.0385963916778564,
      "learning_rate": 0.0001798319500063387,
      "loss": 3.6525,
      "step": 108180
    },
    {
      "epoch": 0.10085058599769453,
      "grad_norm": 1.0300086736679077,
      "learning_rate": 0.00017983008568424352,
      "loss": 3.5659,
      "step": 108190
    },
    {
      "epoch": 0.10085990761577363,
      "grad_norm": 1.2128963470458984,
      "learning_rate": 0.0001798282213621483,
      "loss": 3.6876,
      "step": 108200
    },
    {
      "epoch": 0.10086922923385272,
      "grad_norm": 1.16509211063385,
      "learning_rate": 0.0001798263570400531,
      "loss": 3.824,
      "step": 108210
    },
    {
      "epoch": 0.1008785508519318,
      "grad_norm": 1.1781061887741089,
      "learning_rate": 0.0001798244927179579,
      "loss": 3.7298,
      "step": 108220
    },
    {
      "epoch": 0.1008878724700109,
      "grad_norm": 0.9327958822250366,
      "learning_rate": 0.0001798226283958627,
      "loss": 3.511,
      "step": 108230
    },
    {
      "epoch": 0.10089719408808999,
      "grad_norm": 0.9143460988998413,
      "learning_rate": 0.00017982076407376752,
      "loss": 3.7571,
      "step": 108240
    },
    {
      "epoch": 0.10090651570616908,
      "grad_norm": 0.9362140893936157,
      "learning_rate": 0.0001798188997516723,
      "loss": 3.765,
      "step": 108250
    },
    {
      "epoch": 0.10091583732424818,
      "grad_norm": 0.9656478762626648,
      "learning_rate": 0.0001798170354295771,
      "loss": 3.837,
      "step": 108260
    },
    {
      "epoch": 0.10092515894232726,
      "grad_norm": 1.1104085445404053,
      "learning_rate": 0.00017981517110748191,
      "loss": 3.6224,
      "step": 108270
    },
    {
      "epoch": 0.10093448056040635,
      "grad_norm": 1.0145000219345093,
      "learning_rate": 0.0001798133067853867,
      "loss": 3.7846,
      "step": 108280
    },
    {
      "epoch": 0.10094380217848545,
      "grad_norm": 0.9785537719726562,
      "learning_rate": 0.0001798114424632915,
      "loss": 3.9011,
      "step": 108290
    },
    {
      "epoch": 0.10095312379656454,
      "grad_norm": 0.9916460514068604,
      "learning_rate": 0.0001798095781411963,
      "loss": 3.7495,
      "step": 108300
    },
    {
      "epoch": 0.10096244541464364,
      "grad_norm": 1.0783969163894653,
      "learning_rate": 0.0001798077138191011,
      "loss": 3.8407,
      "step": 108310
    },
    {
      "epoch": 0.10097176703272273,
      "grad_norm": 1.0420453548431396,
      "learning_rate": 0.00017980584949700592,
      "loss": 3.5403,
      "step": 108320
    },
    {
      "epoch": 0.10098108865080181,
      "grad_norm": 1.0982850790023804,
      "learning_rate": 0.0001798039851749107,
      "loss": 3.7023,
      "step": 108330
    },
    {
      "epoch": 0.1009904102688809,
      "grad_norm": 1.1062946319580078,
      "learning_rate": 0.0001798021208528155,
      "loss": 3.6162,
      "step": 108340
    },
    {
      "epoch": 0.10099973188696,
      "grad_norm": 1.223264455795288,
      "learning_rate": 0.0001798002565307203,
      "loss": 3.5029,
      "step": 108350
    },
    {
      "epoch": 0.1010090535050391,
      "grad_norm": 1.055923581123352,
      "learning_rate": 0.00017979839220862512,
      "loss": 3.88,
      "step": 108360
    },
    {
      "epoch": 0.10101837512311819,
      "grad_norm": 0.9675166606903076,
      "learning_rate": 0.00017979652788652992,
      "loss": 3.5046,
      "step": 108370
    },
    {
      "epoch": 0.10102769674119727,
      "grad_norm": 1.0685261487960815,
      "learning_rate": 0.0001797946635644347,
      "loss": 3.4778,
      "step": 108380
    },
    {
      "epoch": 0.10103701835927636,
      "grad_norm": 1.1035035848617554,
      "learning_rate": 0.0001797927992423395,
      "loss": 3.6954,
      "step": 108390
    },
    {
      "epoch": 0.10104633997735546,
      "grad_norm": 1.1486713886260986,
      "learning_rate": 0.00017979093492024432,
      "loss": 3.7757,
      "step": 108400
    },
    {
      "epoch": 0.10105566159543455,
      "grad_norm": 0.9933144450187683,
      "learning_rate": 0.00017978907059814912,
      "loss": 3.7375,
      "step": 108410
    },
    {
      "epoch": 0.10106498321351365,
      "grad_norm": 0.8825994729995728,
      "learning_rate": 0.0001797872062760539,
      "loss": 3.721,
      "step": 108420
    },
    {
      "epoch": 0.10107430483159273,
      "grad_norm": 1.126881718635559,
      "learning_rate": 0.0001797853419539587,
      "loss": 3.4574,
      "step": 108430
    },
    {
      "epoch": 0.10108362644967182,
      "grad_norm": 1.2778021097183228,
      "learning_rate": 0.0001797834776318635,
      "loss": 3.6319,
      "step": 108440
    },
    {
      "epoch": 0.10109294806775092,
      "grad_norm": 1.2238147258758545,
      "learning_rate": 0.0001797816133097683,
      "loss": 3.6517,
      "step": 108450
    },
    {
      "epoch": 0.10110226968583001,
      "grad_norm": 0.860493540763855,
      "learning_rate": 0.00017977974898767313,
      "loss": 3.8293,
      "step": 108460
    },
    {
      "epoch": 0.1011115913039091,
      "grad_norm": 1.1165215969085693,
      "learning_rate": 0.0001797778846655779,
      "loss": 3.7346,
      "step": 108470
    },
    {
      "epoch": 0.10112091292198819,
      "grad_norm": 1.2830119132995605,
      "learning_rate": 0.0001797760203434827,
      "loss": 3.6791,
      "step": 108480
    },
    {
      "epoch": 0.10113023454006728,
      "grad_norm": 1.076218843460083,
      "learning_rate": 0.00017977415602138752,
      "loss": 3.5858,
      "step": 108490
    },
    {
      "epoch": 0.10113955615814638,
      "grad_norm": 1.02606999874115,
      "learning_rate": 0.0001797722916992923,
      "loss": 3.5935,
      "step": 108500
    },
    {
      "epoch": 0.10114887777622547,
      "grad_norm": 0.9813755750656128,
      "learning_rate": 0.0001797704273771971,
      "loss": 3.7307,
      "step": 108510
    },
    {
      "epoch": 0.10115819939430457,
      "grad_norm": 1.0955638885498047,
      "learning_rate": 0.0001797685630551019,
      "loss": 3.7827,
      "step": 108520
    },
    {
      "epoch": 0.10116752101238366,
      "grad_norm": 0.9580429792404175,
      "learning_rate": 0.00017976669873300672,
      "loss": 3.9036,
      "step": 108530
    },
    {
      "epoch": 0.10117684263046274,
      "grad_norm": 0.9516139626502991,
      "learning_rate": 0.00017976483441091152,
      "loss": 3.5286,
      "step": 108540
    },
    {
      "epoch": 0.10118616424854183,
      "grad_norm": 1.1071386337280273,
      "learning_rate": 0.0001797629700888163,
      "loss": 3.6836,
      "step": 108550
    },
    {
      "epoch": 0.10119548586662093,
      "grad_norm": 0.9623807668685913,
      "learning_rate": 0.0001797611057667211,
      "loss": 3.7717,
      "step": 108560
    },
    {
      "epoch": 0.10120480748470002,
      "grad_norm": 1.0479047298431396,
      "learning_rate": 0.00017975924144462591,
      "loss": 3.6782,
      "step": 108570
    },
    {
      "epoch": 0.10121412910277912,
      "grad_norm": 1.3978877067565918,
      "learning_rate": 0.00017975737712253072,
      "loss": 3.6955,
      "step": 108580
    },
    {
      "epoch": 0.1012234507208582,
      "grad_norm": 1.0643798112869263,
      "learning_rate": 0.00017975551280043553,
      "loss": 3.7555,
      "step": 108590
    },
    {
      "epoch": 0.1012327723389373,
      "grad_norm": 1.163283348083496,
      "learning_rate": 0.0001797536484783403,
      "loss": 3.7412,
      "step": 108600
    },
    {
      "epoch": 0.10124209395701639,
      "grad_norm": 1.4763416051864624,
      "learning_rate": 0.0001797517841562451,
      "loss": 3.8744,
      "step": 108610
    },
    {
      "epoch": 0.10125141557509548,
      "grad_norm": 1.2687785625457764,
      "learning_rate": 0.0001797499198341499,
      "loss": 3.6096,
      "step": 108620
    },
    {
      "epoch": 0.10126073719317458,
      "grad_norm": 1.0551992654800415,
      "learning_rate": 0.00017974805551205472,
      "loss": 3.6989,
      "step": 108630
    },
    {
      "epoch": 0.10127005881125366,
      "grad_norm": 1.1453361511230469,
      "learning_rate": 0.00017974619118995953,
      "loss": 3.6612,
      "step": 108640
    },
    {
      "epoch": 0.10127938042933275,
      "grad_norm": 1.148647665977478,
      "learning_rate": 0.0001797443268678643,
      "loss": 3.6444,
      "step": 108650
    },
    {
      "epoch": 0.10128870204741185,
      "grad_norm": 0.9440999031066895,
      "learning_rate": 0.00017974246254576912,
      "loss": 3.7121,
      "step": 108660
    },
    {
      "epoch": 0.10129802366549094,
      "grad_norm": 1.078624963760376,
      "learning_rate": 0.0001797405982236739,
      "loss": 3.8008,
      "step": 108670
    },
    {
      "epoch": 0.10130734528357004,
      "grad_norm": 1.0077458620071411,
      "learning_rate": 0.0001797387339015787,
      "loss": 3.7251,
      "step": 108680
    },
    {
      "epoch": 0.10131666690164912,
      "grad_norm": 1.0682464838027954,
      "learning_rate": 0.00017973686957948354,
      "loss": 3.7473,
      "step": 108690
    },
    {
      "epoch": 0.10132598851972821,
      "grad_norm": 1.1583917140960693,
      "learning_rate": 0.00017973500525738831,
      "loss": 3.6488,
      "step": 108700
    },
    {
      "epoch": 0.1013353101378073,
      "grad_norm": 1.1142711639404297,
      "learning_rate": 0.00017973314093529312,
      "loss": 3.5794,
      "step": 108710
    },
    {
      "epoch": 0.1013446317558864,
      "grad_norm": 1.0090206861495972,
      "learning_rate": 0.0001797312766131979,
      "loss": 3.6935,
      "step": 108720
    },
    {
      "epoch": 0.1013539533739655,
      "grad_norm": 1.1027624607086182,
      "learning_rate": 0.0001797294122911027,
      "loss": 3.8253,
      "step": 108730
    },
    {
      "epoch": 0.10136327499204457,
      "grad_norm": 1.0641790628433228,
      "learning_rate": 0.0001797275479690075,
      "loss": 3.7571,
      "step": 108740
    },
    {
      "epoch": 0.10137259661012367,
      "grad_norm": 0.9848502278327942,
      "learning_rate": 0.00017972568364691232,
      "loss": 3.5112,
      "step": 108750
    },
    {
      "epoch": 0.10138191822820276,
      "grad_norm": 1.0141581296920776,
      "learning_rate": 0.00017972381932481713,
      "loss": 3.8033,
      "step": 108760
    },
    {
      "epoch": 0.10139123984628186,
      "grad_norm": 1.1021308898925781,
      "learning_rate": 0.0001797219550027219,
      "loss": 3.6198,
      "step": 108770
    },
    {
      "epoch": 0.10140056146436095,
      "grad_norm": 1.1883227825164795,
      "learning_rate": 0.0001797200906806267,
      "loss": 3.7418,
      "step": 108780
    },
    {
      "epoch": 0.10140988308244005,
      "grad_norm": 0.9841135144233704,
      "learning_rate": 0.00017971822635853152,
      "loss": 3.8095,
      "step": 108790
    },
    {
      "epoch": 0.10141920470051913,
      "grad_norm": 1.0125441551208496,
      "learning_rate": 0.00017971636203643632,
      "loss": 3.6643,
      "step": 108800
    },
    {
      "epoch": 0.10142852631859822,
      "grad_norm": 1.359955906867981,
      "learning_rate": 0.00017971449771434113,
      "loss": 3.8275,
      "step": 108810
    },
    {
      "epoch": 0.10143784793667732,
      "grad_norm": 0.9655221700668335,
      "learning_rate": 0.0001797126333922459,
      "loss": 3.5228,
      "step": 108820
    },
    {
      "epoch": 0.10144716955475641,
      "grad_norm": 1.0199780464172363,
      "learning_rate": 0.00017971076907015072,
      "loss": 3.7051,
      "step": 108830
    },
    {
      "epoch": 0.1014564911728355,
      "grad_norm": 1.2495313882827759,
      "learning_rate": 0.00017970890474805552,
      "loss": 3.6129,
      "step": 108840
    },
    {
      "epoch": 0.10146581279091459,
      "grad_norm": 0.9564783573150635,
      "learning_rate": 0.00017970704042596033,
      "loss": 3.5912,
      "step": 108850
    },
    {
      "epoch": 0.10147513440899368,
      "grad_norm": 0.9371235370635986,
      "learning_rate": 0.00017970517610386513,
      "loss": 3.63,
      "step": 108860
    },
    {
      "epoch": 0.10148445602707278,
      "grad_norm": 1.03408682346344,
      "learning_rate": 0.00017970331178176991,
      "loss": 3.8008,
      "step": 108870
    },
    {
      "epoch": 0.10149377764515187,
      "grad_norm": 1.0549116134643555,
      "learning_rate": 0.00017970144745967472,
      "loss": 3.6961,
      "step": 108880
    },
    {
      "epoch": 0.10150309926323096,
      "grad_norm": 1.2122244834899902,
      "learning_rate": 0.00017969958313757953,
      "loss": 3.7928,
      "step": 108890
    },
    {
      "epoch": 0.10151242088131004,
      "grad_norm": 1.1326091289520264,
      "learning_rate": 0.0001796977188154843,
      "loss": 3.6341,
      "step": 108900
    },
    {
      "epoch": 0.10152174249938914,
      "grad_norm": 1.1142542362213135,
      "learning_rate": 0.00017969585449338914,
      "loss": 3.9709,
      "step": 108910
    },
    {
      "epoch": 0.10153106411746823,
      "grad_norm": 0.8685027956962585,
      "learning_rate": 0.00017969399017129392,
      "loss": 3.7652,
      "step": 108920
    },
    {
      "epoch": 0.10154038573554733,
      "grad_norm": 1.0947619676589966,
      "learning_rate": 0.00017969212584919872,
      "loss": 3.7493,
      "step": 108930
    },
    {
      "epoch": 0.10154970735362642,
      "grad_norm": 1.0253565311431885,
      "learning_rate": 0.00017969026152710353,
      "loss": 3.7755,
      "step": 108940
    },
    {
      "epoch": 0.1015590289717055,
      "grad_norm": 1.0056571960449219,
      "learning_rate": 0.0001796883972050083,
      "loss": 3.6952,
      "step": 108950
    },
    {
      "epoch": 0.1015683505897846,
      "grad_norm": 1.1315269470214844,
      "learning_rate": 0.00017968653288291312,
      "loss": 3.6508,
      "step": 108960
    },
    {
      "epoch": 0.10157767220786369,
      "grad_norm": 1.2599095106124878,
      "learning_rate": 0.00017968466856081792,
      "loss": 3.6482,
      "step": 108970
    },
    {
      "epoch": 0.10158699382594279,
      "grad_norm": 1.1940944194793701,
      "learning_rate": 0.00017968280423872273,
      "loss": 3.6519,
      "step": 108980
    },
    {
      "epoch": 0.10159631544402188,
      "grad_norm": 1.2799243927001953,
      "learning_rate": 0.00017968093991662754,
      "loss": 3.683,
      "step": 108990
    },
    {
      "epoch": 0.10160563706210098,
      "grad_norm": 0.9483467936515808,
      "learning_rate": 0.00017967907559453231,
      "loss": 3.5838,
      "step": 109000
    },
    {
      "epoch": 0.10161495868018006,
      "grad_norm": 0.9871819019317627,
      "learning_rate": 0.00017967721127243712,
      "loss": 3.9837,
      "step": 109010
    },
    {
      "epoch": 0.10162428029825915,
      "grad_norm": 1.1434547901153564,
      "learning_rate": 0.00017967534695034193,
      "loss": 3.8031,
      "step": 109020
    },
    {
      "epoch": 0.10163360191633825,
      "grad_norm": 1.2690232992172241,
      "learning_rate": 0.00017967348262824673,
      "loss": 3.7857,
      "step": 109030
    },
    {
      "epoch": 0.10164292353441734,
      "grad_norm": 1.1217379570007324,
      "learning_rate": 0.00017967161830615154,
      "loss": 3.7943,
      "step": 109040
    },
    {
      "epoch": 0.10165224515249643,
      "grad_norm": 1.1069284677505493,
      "learning_rate": 0.00017966975398405632,
      "loss": 3.7092,
      "step": 109050
    },
    {
      "epoch": 0.10166156677057551,
      "grad_norm": 1.0351219177246094,
      "learning_rate": 0.00017966788966196113,
      "loss": 3.6324,
      "step": 109060
    },
    {
      "epoch": 0.10167088838865461,
      "grad_norm": 1.3332382440567017,
      "learning_rate": 0.0001796660253398659,
      "loss": 3.6968,
      "step": 109070
    },
    {
      "epoch": 0.1016802100067337,
      "grad_norm": 1.0821670293807983,
      "learning_rate": 0.00017966416101777074,
      "loss": 3.7949,
      "step": 109080
    },
    {
      "epoch": 0.1016895316248128,
      "grad_norm": 0.9463052153587341,
      "learning_rate": 0.00017966229669567554,
      "loss": 3.8392,
      "step": 109090
    },
    {
      "epoch": 0.10169885324289189,
      "grad_norm": 1.020020604133606,
      "learning_rate": 0.00017966043237358032,
      "loss": 3.7929,
      "step": 109100
    },
    {
      "epoch": 0.10170817486097097,
      "grad_norm": 1.1818541288375854,
      "learning_rate": 0.00017965856805148513,
      "loss": 3.7929,
      "step": 109110
    },
    {
      "epoch": 0.10171749647905007,
      "grad_norm": 0.9938533902168274,
      "learning_rate": 0.0001796567037293899,
      "loss": 3.6871,
      "step": 109120
    },
    {
      "epoch": 0.10172681809712916,
      "grad_norm": 1.1035147905349731,
      "learning_rate": 0.00017965483940729472,
      "loss": 3.7561,
      "step": 109130
    },
    {
      "epoch": 0.10173613971520826,
      "grad_norm": 1.0872979164123535,
      "learning_rate": 0.00017965297508519955,
      "loss": 3.7142,
      "step": 109140
    },
    {
      "epoch": 0.10174546133328735,
      "grad_norm": 1.0402041673660278,
      "learning_rate": 0.00017965111076310433,
      "loss": 3.6081,
      "step": 109150
    },
    {
      "epoch": 0.10175478295136643,
      "grad_norm": 1.2371877431869507,
      "learning_rate": 0.00017964924644100913,
      "loss": 3.6313,
      "step": 109160
    },
    {
      "epoch": 0.10176410456944553,
      "grad_norm": 1.0019047260284424,
      "learning_rate": 0.0001796473821189139,
      "loss": 3.6344,
      "step": 109170
    },
    {
      "epoch": 0.10177342618752462,
      "grad_norm": 0.9315583109855652,
      "learning_rate": 0.00017964551779681872,
      "loss": 3.7367,
      "step": 109180
    },
    {
      "epoch": 0.10178274780560372,
      "grad_norm": 1.0567628145217896,
      "learning_rate": 0.00017964365347472353,
      "loss": 3.6467,
      "step": 109190
    },
    {
      "epoch": 0.10179206942368281,
      "grad_norm": 1.1513396501541138,
      "learning_rate": 0.00017964178915262833,
      "loss": 3.6513,
      "step": 109200
    },
    {
      "epoch": 0.1018013910417619,
      "grad_norm": 1.1324893236160278,
      "learning_rate": 0.00017963992483053314,
      "loss": 3.7991,
      "step": 109210
    },
    {
      "epoch": 0.10181071265984099,
      "grad_norm": 1.1144933700561523,
      "learning_rate": 0.00017963806050843792,
      "loss": 3.8255,
      "step": 109220
    },
    {
      "epoch": 0.10182003427792008,
      "grad_norm": 0.9154757261276245,
      "learning_rate": 0.00017963619618634272,
      "loss": 3.6702,
      "step": 109230
    },
    {
      "epoch": 0.10182935589599917,
      "grad_norm": 0.9944091439247131,
      "learning_rate": 0.00017963433186424753,
      "loss": 3.6072,
      "step": 109240
    },
    {
      "epoch": 0.10183867751407827,
      "grad_norm": 0.9766130447387695,
      "learning_rate": 0.00017963246754215234,
      "loss": 3.7221,
      "step": 109250
    },
    {
      "epoch": 0.10184799913215736,
      "grad_norm": 1.2597649097442627,
      "learning_rate": 0.00017963060322005714,
      "loss": 3.7931,
      "step": 109260
    },
    {
      "epoch": 0.10185732075023644,
      "grad_norm": 0.8632124066352844,
      "learning_rate": 0.00017962873889796192,
      "loss": 3.6754,
      "step": 109270
    },
    {
      "epoch": 0.10186664236831554,
      "grad_norm": 0.9640316367149353,
      "learning_rate": 0.00017962687457586673,
      "loss": 3.7792,
      "step": 109280
    },
    {
      "epoch": 0.10187596398639463,
      "grad_norm": 1.0646696090698242,
      "learning_rate": 0.00017962501025377153,
      "loss": 3.7347,
      "step": 109290
    },
    {
      "epoch": 0.10188528560447373,
      "grad_norm": 1.137864589691162,
      "learning_rate": 0.00017962314593167634,
      "loss": 3.7,
      "step": 109300
    },
    {
      "epoch": 0.10189460722255282,
      "grad_norm": 0.9365801811218262,
      "learning_rate": 0.00017962128160958115,
      "loss": 3.6867,
      "step": 109310
    },
    {
      "epoch": 0.1019039288406319,
      "grad_norm": 1.0721896886825562,
      "learning_rate": 0.00017961941728748593,
      "loss": 3.6031,
      "step": 109320
    },
    {
      "epoch": 0.101913250458711,
      "grad_norm": 0.9788493514060974,
      "learning_rate": 0.00017961755296539073,
      "loss": 3.8559,
      "step": 109330
    },
    {
      "epoch": 0.10192257207679009,
      "grad_norm": 1.260147213935852,
      "learning_rate": 0.00017961568864329554,
      "loss": 3.8023,
      "step": 109340
    },
    {
      "epoch": 0.10193189369486919,
      "grad_norm": 0.9659048318862915,
      "learning_rate": 0.00017961382432120032,
      "loss": 3.7967,
      "step": 109350
    },
    {
      "epoch": 0.10194121531294828,
      "grad_norm": 1.1537251472473145,
      "learning_rate": 0.00017961195999910515,
      "loss": 3.7721,
      "step": 109360
    },
    {
      "epoch": 0.10195053693102736,
      "grad_norm": 1.0660498142242432,
      "learning_rate": 0.00017961009567700993,
      "loss": 3.6579,
      "step": 109370
    },
    {
      "epoch": 0.10195985854910646,
      "grad_norm": 1.0662256479263306,
      "learning_rate": 0.00017960823135491474,
      "loss": 3.606,
      "step": 109380
    },
    {
      "epoch": 0.10196918016718555,
      "grad_norm": 1.0167604684829712,
      "learning_rate": 0.00017960636703281954,
      "loss": 3.8055,
      "step": 109390
    },
    {
      "epoch": 0.10197850178526464,
      "grad_norm": 1.1001183986663818,
      "learning_rate": 0.00017960450271072432,
      "loss": 3.7992,
      "step": 109400
    },
    {
      "epoch": 0.10198782340334374,
      "grad_norm": 1.065530776977539,
      "learning_rate": 0.00017960263838862913,
      "loss": 3.8578,
      "step": 109410
    },
    {
      "epoch": 0.10199714502142282,
      "grad_norm": 0.9935506582260132,
      "learning_rate": 0.00017960077406653394,
      "loss": 3.7584,
      "step": 109420
    },
    {
      "epoch": 0.10200646663950191,
      "grad_norm": 1.2553907632827759,
      "learning_rate": 0.00017959890974443874,
      "loss": 3.6991,
      "step": 109430
    },
    {
      "epoch": 0.10201578825758101,
      "grad_norm": 1.0832465887069702,
      "learning_rate": 0.00017959704542234355,
      "loss": 3.4393,
      "step": 109440
    },
    {
      "epoch": 0.1020251098756601,
      "grad_norm": 1.3257991075515747,
      "learning_rate": 0.00017959518110024833,
      "loss": 3.7966,
      "step": 109450
    },
    {
      "epoch": 0.1020344314937392,
      "grad_norm": 1.133833646774292,
      "learning_rate": 0.00017959331677815313,
      "loss": 3.7471,
      "step": 109460
    },
    {
      "epoch": 0.10204375311181829,
      "grad_norm": 1.0865570306777954,
      "learning_rate": 0.00017959145245605794,
      "loss": 3.6799,
      "step": 109470
    },
    {
      "epoch": 0.10205307472989737,
      "grad_norm": 0.9752934575080872,
      "learning_rate": 0.00017958958813396275,
      "loss": 3.9359,
      "step": 109480
    },
    {
      "epoch": 0.10206239634797647,
      "grad_norm": 0.9792898893356323,
      "learning_rate": 0.00017958772381186755,
      "loss": 3.6956,
      "step": 109490
    },
    {
      "epoch": 0.10207171796605556,
      "grad_norm": 1.0673102140426636,
      "learning_rate": 0.00017958585948977233,
      "loss": 3.6815,
      "step": 109500
    },
    {
      "epoch": 0.10208103958413466,
      "grad_norm": 0.9928311109542847,
      "learning_rate": 0.00017958399516767714,
      "loss": 3.6377,
      "step": 109510
    },
    {
      "epoch": 0.10209036120221375,
      "grad_norm": 0.9754656553268433,
      "learning_rate": 0.00017958213084558192,
      "loss": 3.7633,
      "step": 109520
    },
    {
      "epoch": 0.10209968282029283,
      "grad_norm": 1.0397214889526367,
      "learning_rate": 0.00017958026652348675,
      "loss": 3.8892,
      "step": 109530
    },
    {
      "epoch": 0.10210900443837193,
      "grad_norm": 1.1228079795837402,
      "learning_rate": 0.00017957840220139156,
      "loss": 3.7555,
      "step": 109540
    },
    {
      "epoch": 0.10211832605645102,
      "grad_norm": 1.4533703327178955,
      "learning_rate": 0.00017957653787929634,
      "loss": 3.698,
      "step": 109550
    },
    {
      "epoch": 0.10212764767453011,
      "grad_norm": 1.017020344734192,
      "learning_rate": 0.00017957467355720114,
      "loss": 3.9427,
      "step": 109560
    },
    {
      "epoch": 0.10213696929260921,
      "grad_norm": 1.2759455442428589,
      "learning_rate": 0.00017957280923510592,
      "loss": 3.631,
      "step": 109570
    },
    {
      "epoch": 0.10214629091068829,
      "grad_norm": 0.9146705865859985,
      "learning_rate": 0.00017957094491301073,
      "loss": 3.6318,
      "step": 109580
    },
    {
      "epoch": 0.10215561252876738,
      "grad_norm": 0.8620395660400391,
      "learning_rate": 0.00017956908059091556,
      "loss": 3.6439,
      "step": 109590
    },
    {
      "epoch": 0.10216493414684648,
      "grad_norm": 0.9716647863388062,
      "learning_rate": 0.00017956721626882034,
      "loss": 3.834,
      "step": 109600
    },
    {
      "epoch": 0.10217425576492557,
      "grad_norm": 1.0866200923919678,
      "learning_rate": 0.00017956535194672515,
      "loss": 3.8284,
      "step": 109610
    },
    {
      "epoch": 0.10218357738300467,
      "grad_norm": 1.0095824003219604,
      "learning_rate": 0.00017956348762462993,
      "loss": 3.7936,
      "step": 109620
    },
    {
      "epoch": 0.10219289900108375,
      "grad_norm": 1.066497564315796,
      "learning_rate": 0.00017956162330253473,
      "loss": 3.7108,
      "step": 109630
    },
    {
      "epoch": 0.10220222061916284,
      "grad_norm": 0.9781243801116943,
      "learning_rate": 0.00017955975898043954,
      "loss": 3.5624,
      "step": 109640
    },
    {
      "epoch": 0.10221154223724194,
      "grad_norm": 1.0875552892684937,
      "learning_rate": 0.00017955789465834434,
      "loss": 3.7058,
      "step": 109650
    },
    {
      "epoch": 0.10222086385532103,
      "grad_norm": 1.0402895212173462,
      "learning_rate": 0.00017955603033624915,
      "loss": 3.7719,
      "step": 109660
    },
    {
      "epoch": 0.10223018547340013,
      "grad_norm": 1.2576427459716797,
      "learning_rate": 0.00017955416601415393,
      "loss": 3.7031,
      "step": 109670
    },
    {
      "epoch": 0.10223950709147922,
      "grad_norm": 0.965613842010498,
      "learning_rate": 0.00017955230169205874,
      "loss": 3.6389,
      "step": 109680
    },
    {
      "epoch": 0.1022488287095583,
      "grad_norm": 1.0786854028701782,
      "learning_rate": 0.00017955043736996354,
      "loss": 3.7419,
      "step": 109690
    },
    {
      "epoch": 0.1022581503276374,
      "grad_norm": 0.996769368648529,
      "learning_rate": 0.00017954857304786835,
      "loss": 3.8432,
      "step": 109700
    },
    {
      "epoch": 0.10226747194571649,
      "grad_norm": 1.1053546667099,
      "learning_rate": 0.00017954670872577316,
      "loss": 3.7349,
      "step": 109710
    },
    {
      "epoch": 0.10227679356379558,
      "grad_norm": 1.0558977127075195,
      "learning_rate": 0.00017954484440367793,
      "loss": 3.7598,
      "step": 109720
    },
    {
      "epoch": 0.10228611518187468,
      "grad_norm": 1.1385776996612549,
      "learning_rate": 0.00017954298008158274,
      "loss": 3.6224,
      "step": 109730
    },
    {
      "epoch": 0.10229543679995376,
      "grad_norm": 1.03168785572052,
      "learning_rate": 0.00017954111575948755,
      "loss": 3.6298,
      "step": 109740
    },
    {
      "epoch": 0.10230475841803285,
      "grad_norm": 0.8252755403518677,
      "learning_rate": 0.00017953925143739235,
      "loss": 3.7111,
      "step": 109750
    },
    {
      "epoch": 0.10231408003611195,
      "grad_norm": 0.9879529476165771,
      "learning_rate": 0.00017953738711529716,
      "loss": 3.5998,
      "step": 109760
    },
    {
      "epoch": 0.10232340165419104,
      "grad_norm": 1.090449571609497,
      "learning_rate": 0.00017953552279320194,
      "loss": 3.6482,
      "step": 109770
    },
    {
      "epoch": 0.10233272327227014,
      "grad_norm": 1.1541019678115845,
      "learning_rate": 0.00017953365847110675,
      "loss": 3.7055,
      "step": 109780
    },
    {
      "epoch": 0.10234204489034922,
      "grad_norm": 1.5989893674850464,
      "learning_rate": 0.00017953179414901152,
      "loss": 3.6378,
      "step": 109790
    },
    {
      "epoch": 0.10235136650842831,
      "grad_norm": 1.1265666484832764,
      "learning_rate": 0.00017952992982691633,
      "loss": 3.6191,
      "step": 109800
    },
    {
      "epoch": 0.10236068812650741,
      "grad_norm": 1.1246962547302246,
      "learning_rate": 0.00017952806550482116,
      "loss": 3.6205,
      "step": 109810
    },
    {
      "epoch": 0.1023700097445865,
      "grad_norm": 0.9702519774436951,
      "learning_rate": 0.00017952620118272594,
      "loss": 3.7001,
      "step": 109820
    },
    {
      "epoch": 0.1023793313626656,
      "grad_norm": 0.9678311347961426,
      "learning_rate": 0.00017952433686063075,
      "loss": 3.6756,
      "step": 109830
    },
    {
      "epoch": 0.10238865298074468,
      "grad_norm": 1.031138300895691,
      "learning_rate": 0.00017952247253853553,
      "loss": 3.7878,
      "step": 109840
    },
    {
      "epoch": 0.10239797459882377,
      "grad_norm": 0.9633233547210693,
      "learning_rate": 0.00017952060821644034,
      "loss": 3.7455,
      "step": 109850
    },
    {
      "epoch": 0.10240729621690287,
      "grad_norm": 0.9012268781661987,
      "learning_rate": 0.00017951874389434514,
      "loss": 3.7362,
      "step": 109860
    },
    {
      "epoch": 0.10241661783498196,
      "grad_norm": 1.1685338020324707,
      "learning_rate": 0.00017951687957224995,
      "loss": 3.5289,
      "step": 109870
    },
    {
      "epoch": 0.10242593945306105,
      "grad_norm": 1.0458906888961792,
      "learning_rate": 0.00017951501525015475,
      "loss": 3.7266,
      "step": 109880
    },
    {
      "epoch": 0.10243526107114015,
      "grad_norm": 1.2122728824615479,
      "learning_rate": 0.00017951315092805953,
      "loss": 3.599,
      "step": 109890
    },
    {
      "epoch": 0.10244458268921923,
      "grad_norm": 1.0213508605957031,
      "learning_rate": 0.00017951128660596434,
      "loss": 3.553,
      "step": 109900
    },
    {
      "epoch": 0.10245390430729832,
      "grad_norm": 0.9821968078613281,
      "learning_rate": 0.00017950942228386915,
      "loss": 3.7063,
      "step": 109910
    },
    {
      "epoch": 0.10246322592537742,
      "grad_norm": 0.8913684487342834,
      "learning_rate": 0.00017950755796177395,
      "loss": 3.6354,
      "step": 109920
    },
    {
      "epoch": 0.10247254754345651,
      "grad_norm": 1.2295031547546387,
      "learning_rate": 0.00017950569363967876,
      "loss": 3.5551,
      "step": 109930
    },
    {
      "epoch": 0.10248186916153561,
      "grad_norm": 1.0965948104858398,
      "learning_rate": 0.00017950382931758354,
      "loss": 3.4931,
      "step": 109940
    },
    {
      "epoch": 0.10249119077961469,
      "grad_norm": 0.9401003122329712,
      "learning_rate": 0.00017950196499548834,
      "loss": 3.9267,
      "step": 109950
    },
    {
      "epoch": 0.10250051239769378,
      "grad_norm": 1.123403549194336,
      "learning_rate": 0.00017950010067339315,
      "loss": 3.7296,
      "step": 109960
    },
    {
      "epoch": 0.10250983401577288,
      "grad_norm": 0.9418383240699768,
      "learning_rate": 0.00017949823635129796,
      "loss": 3.82,
      "step": 109970
    },
    {
      "epoch": 0.10251915563385197,
      "grad_norm": 1.0576351881027222,
      "learning_rate": 0.00017949637202920276,
      "loss": 3.7758,
      "step": 109980
    },
    {
      "epoch": 0.10252847725193107,
      "grad_norm": 1.5293141603469849,
      "learning_rate": 0.00017949450770710754,
      "loss": 3.6837,
      "step": 109990
    },
    {
      "epoch": 0.10253779887001015,
      "grad_norm": 1.1645101308822632,
      "learning_rate": 0.00017949264338501235,
      "loss": 3.5473,
      "step": 110000
    },
    {
      "epoch": 0.10254712048808924,
      "grad_norm": 1.017359733581543,
      "learning_rate": 0.00017949077906291716,
      "loss": 3.6669,
      "step": 110010
    },
    {
      "epoch": 0.10255644210616834,
      "grad_norm": 0.9569178819656372,
      "learning_rate": 0.00017948891474082193,
      "loss": 3.653,
      "step": 110020
    },
    {
      "epoch": 0.10256576372424743,
      "grad_norm": 1.3024696111679077,
      "learning_rate": 0.00017948705041872674,
      "loss": 3.654,
      "step": 110030
    },
    {
      "epoch": 0.10257508534232652,
      "grad_norm": 1.1916704177856445,
      "learning_rate": 0.00017948518609663155,
      "loss": 3.68,
      "step": 110040
    },
    {
      "epoch": 0.1025844069604056,
      "grad_norm": 1.0022231340408325,
      "learning_rate": 0.00017948332177453635,
      "loss": 3.6855,
      "step": 110050
    },
    {
      "epoch": 0.1025937285784847,
      "grad_norm": 1.038572072982788,
      "learning_rate": 0.00017948145745244116,
      "loss": 3.8687,
      "step": 110060
    },
    {
      "epoch": 0.1026030501965638,
      "grad_norm": 1.1360021829605103,
      "learning_rate": 0.00017947959313034594,
      "loss": 3.6094,
      "step": 110070
    },
    {
      "epoch": 0.10261237181464289,
      "grad_norm": 1.0849567651748657,
      "learning_rate": 0.00017947772880825075,
      "loss": 3.7409,
      "step": 110080
    },
    {
      "epoch": 0.10262169343272198,
      "grad_norm": 1.117535948753357,
      "learning_rate": 0.00017947586448615555,
      "loss": 3.7649,
      "step": 110090
    },
    {
      "epoch": 0.10263101505080106,
      "grad_norm": 0.9429476857185364,
      "learning_rate": 0.00017947400016406036,
      "loss": 3.6806,
      "step": 110100
    },
    {
      "epoch": 0.10264033666888016,
      "grad_norm": 0.8809558153152466,
      "learning_rate": 0.00017947213584196516,
      "loss": 3.7223,
      "step": 110110
    },
    {
      "epoch": 0.10264965828695925,
      "grad_norm": 1.243154764175415,
      "learning_rate": 0.00017947027151986994,
      "loss": 3.7885,
      "step": 110120
    },
    {
      "epoch": 0.10265897990503835,
      "grad_norm": 0.9488365054130554,
      "learning_rate": 0.00017946840719777475,
      "loss": 3.6433,
      "step": 110130
    },
    {
      "epoch": 0.10266830152311744,
      "grad_norm": 1.0828948020935059,
      "learning_rate": 0.00017946654287567956,
      "loss": 3.4967,
      "step": 110140
    },
    {
      "epoch": 0.10267762314119654,
      "grad_norm": 0.9587122201919556,
      "learning_rate": 0.00017946467855358436,
      "loss": 3.7359,
      "step": 110150
    },
    {
      "epoch": 0.10268694475927562,
      "grad_norm": 1.0378025770187378,
      "learning_rate": 0.00017946281423148917,
      "loss": 3.8089,
      "step": 110160
    },
    {
      "epoch": 0.10269626637735471,
      "grad_norm": 1.071425199508667,
      "learning_rate": 0.00017946094990939395,
      "loss": 3.8579,
      "step": 110170
    },
    {
      "epoch": 0.1027055879954338,
      "grad_norm": 1.1131452322006226,
      "learning_rate": 0.00017945908558729875,
      "loss": 3.6752,
      "step": 110180
    },
    {
      "epoch": 0.1027149096135129,
      "grad_norm": 1.212579369544983,
      "learning_rate": 0.00017945722126520353,
      "loss": 3.5391,
      "step": 110190
    },
    {
      "epoch": 0.102724231231592,
      "grad_norm": 0.8068693280220032,
      "learning_rate": 0.00017945535694310837,
      "loss": 3.8226,
      "step": 110200
    },
    {
      "epoch": 0.10273355284967108,
      "grad_norm": 1.1138640642166138,
      "learning_rate": 0.00017945349262101317,
      "loss": 3.6391,
      "step": 110210
    },
    {
      "epoch": 0.10274287446775017,
      "grad_norm": 0.9692152738571167,
      "learning_rate": 0.00017945162829891795,
      "loss": 3.7124,
      "step": 110220
    },
    {
      "epoch": 0.10275219608582926,
      "grad_norm": 1.1069800853729248,
      "learning_rate": 0.00017944976397682276,
      "loss": 3.8081,
      "step": 110230
    },
    {
      "epoch": 0.10276151770390836,
      "grad_norm": 1.1783447265625,
      "learning_rate": 0.00017944789965472754,
      "loss": 3.8833,
      "step": 110240
    },
    {
      "epoch": 0.10277083932198745,
      "grad_norm": 1.0879132747650146,
      "learning_rate": 0.00017944603533263234,
      "loss": 3.7045,
      "step": 110250
    },
    {
      "epoch": 0.10278016094006653,
      "grad_norm": 1.2700296640396118,
      "learning_rate": 0.00017944417101053718,
      "loss": 3.7362,
      "step": 110260
    },
    {
      "epoch": 0.10278948255814563,
      "grad_norm": 0.9220787286758423,
      "learning_rate": 0.00017944230668844196,
      "loss": 3.6404,
      "step": 110270
    },
    {
      "epoch": 0.10279880417622472,
      "grad_norm": 1.0467063188552856,
      "learning_rate": 0.00017944044236634676,
      "loss": 3.877,
      "step": 110280
    },
    {
      "epoch": 0.10280812579430382,
      "grad_norm": 1.189786672592163,
      "learning_rate": 0.00017943857804425154,
      "loss": 3.7426,
      "step": 110290
    },
    {
      "epoch": 0.10281744741238291,
      "grad_norm": 0.9572055339813232,
      "learning_rate": 0.00017943671372215635,
      "loss": 3.5149,
      "step": 110300
    },
    {
      "epoch": 0.10282676903046199,
      "grad_norm": 0.9407761693000793,
      "learning_rate": 0.00017943484940006115,
      "loss": 3.7998,
      "step": 110310
    },
    {
      "epoch": 0.10283609064854109,
      "grad_norm": 0.8838001489639282,
      "learning_rate": 0.00017943298507796596,
      "loss": 3.7426,
      "step": 110320
    },
    {
      "epoch": 0.10284541226662018,
      "grad_norm": 1.158008337020874,
      "learning_rate": 0.00017943112075587077,
      "loss": 3.6377,
      "step": 110330
    },
    {
      "epoch": 0.10285473388469928,
      "grad_norm": 1.0465463399887085,
      "learning_rate": 0.00017942925643377555,
      "loss": 3.3448,
      "step": 110340
    },
    {
      "epoch": 0.10286405550277837,
      "grad_norm": 1.0825225114822388,
      "learning_rate": 0.00017942739211168035,
      "loss": 3.7156,
      "step": 110350
    },
    {
      "epoch": 0.10287337712085746,
      "grad_norm": 0.9478535652160645,
      "learning_rate": 0.00017942552778958516,
      "loss": 3.6001,
      "step": 110360
    },
    {
      "epoch": 0.10288269873893655,
      "grad_norm": 1.2529464960098267,
      "learning_rate": 0.00017942366346748997,
      "loss": 3.7291,
      "step": 110370
    },
    {
      "epoch": 0.10289202035701564,
      "grad_norm": 1.1456013917922974,
      "learning_rate": 0.00017942179914539477,
      "loss": 3.6215,
      "step": 110380
    },
    {
      "epoch": 0.10290134197509473,
      "grad_norm": 1.0586464405059814,
      "learning_rate": 0.00017941993482329955,
      "loss": 3.9323,
      "step": 110390
    },
    {
      "epoch": 0.10291066359317383,
      "grad_norm": 0.9850831031799316,
      "learning_rate": 0.00017941807050120436,
      "loss": 3.8428,
      "step": 110400
    },
    {
      "epoch": 0.10291998521125292,
      "grad_norm": 1.0831204652786255,
      "learning_rate": 0.00017941620617910916,
      "loss": 3.5156,
      "step": 110410
    },
    {
      "epoch": 0.102929306829332,
      "grad_norm": 1.1359758377075195,
      "learning_rate": 0.00017941434185701397,
      "loss": 3.6814,
      "step": 110420
    },
    {
      "epoch": 0.1029386284474111,
      "grad_norm": 1.2557271718978882,
      "learning_rate": 0.00017941247753491878,
      "loss": 3.8933,
      "step": 110430
    },
    {
      "epoch": 0.10294795006549019,
      "grad_norm": 1.0016576051712036,
      "learning_rate": 0.00017941061321282356,
      "loss": 4.0179,
      "step": 110440
    },
    {
      "epoch": 0.10295727168356929,
      "grad_norm": 1.7128273248672485,
      "learning_rate": 0.00017940874889072836,
      "loss": 3.6515,
      "step": 110450
    },
    {
      "epoch": 0.10296659330164838,
      "grad_norm": 1.0727019309997559,
      "learning_rate": 0.00017940688456863317,
      "loss": 3.6583,
      "step": 110460
    },
    {
      "epoch": 0.10297591491972746,
      "grad_norm": 1.0054444074630737,
      "learning_rate": 0.00017940502024653795,
      "loss": 3.7891,
      "step": 110470
    },
    {
      "epoch": 0.10298523653780656,
      "grad_norm": 0.9798849821090698,
      "learning_rate": 0.00017940315592444278,
      "loss": 3.7433,
      "step": 110480
    },
    {
      "epoch": 0.10299455815588565,
      "grad_norm": 1.1961795091629028,
      "learning_rate": 0.00017940129160234756,
      "loss": 3.698,
      "step": 110490
    },
    {
      "epoch": 0.10300387977396475,
      "grad_norm": 1.0230634212493896,
      "learning_rate": 0.00017939942728025237,
      "loss": 3.7684,
      "step": 110500
    },
    {
      "epoch": 0.10301320139204384,
      "grad_norm": 1.0909132957458496,
      "learning_rate": 0.00017939756295815717,
      "loss": 3.6863,
      "step": 110510
    },
    {
      "epoch": 0.10302252301012292,
      "grad_norm": 1.0460419654846191,
      "learning_rate": 0.00017939569863606195,
      "loss": 3.6636,
      "step": 110520
    },
    {
      "epoch": 0.10303184462820202,
      "grad_norm": 0.9788183569908142,
      "learning_rate": 0.00017939383431396676,
      "loss": 3.6142,
      "step": 110530
    },
    {
      "epoch": 0.10304116624628111,
      "grad_norm": 1.1244193315505981,
      "learning_rate": 0.00017939196999187156,
      "loss": 3.7798,
      "step": 110540
    },
    {
      "epoch": 0.1030504878643602,
      "grad_norm": 1.2108086347579956,
      "learning_rate": 0.00017939010566977637,
      "loss": 3.7605,
      "step": 110550
    },
    {
      "epoch": 0.1030598094824393,
      "grad_norm": 1.2093243598937988,
      "learning_rate": 0.00017938824134768118,
      "loss": 3.6071,
      "step": 110560
    },
    {
      "epoch": 0.1030691311005184,
      "grad_norm": 1.082715630531311,
      "learning_rate": 0.00017938637702558596,
      "loss": 3.7512,
      "step": 110570
    },
    {
      "epoch": 0.10307845271859747,
      "grad_norm": 1.1749025583267212,
      "learning_rate": 0.00017938451270349076,
      "loss": 3.8385,
      "step": 110580
    },
    {
      "epoch": 0.10308777433667657,
      "grad_norm": 0.920302152633667,
      "learning_rate": 0.00017938264838139557,
      "loss": 3.6399,
      "step": 110590
    },
    {
      "epoch": 0.10309709595475566,
      "grad_norm": 1.0409952402114868,
      "learning_rate": 0.00017938078405930038,
      "loss": 3.7593,
      "step": 110600
    },
    {
      "epoch": 0.10310641757283476,
      "grad_norm": 1.071800708770752,
      "learning_rate": 0.00017937891973720518,
      "loss": 3.8291,
      "step": 110610
    },
    {
      "epoch": 0.10311573919091385,
      "grad_norm": 0.9828828573226929,
      "learning_rate": 0.00017937705541510996,
      "loss": 3.7071,
      "step": 110620
    },
    {
      "epoch": 0.10312506080899293,
      "grad_norm": 1.0054179430007935,
      "learning_rate": 0.00017937519109301477,
      "loss": 3.8453,
      "step": 110630
    },
    {
      "epoch": 0.10313438242707203,
      "grad_norm": 1.154179334640503,
      "learning_rate": 0.00017937332677091955,
      "loss": 3.5534,
      "step": 110640
    },
    {
      "epoch": 0.10314370404515112,
      "grad_norm": 1.14871346950531,
      "learning_rate": 0.00017937146244882438,
      "loss": 3.7938,
      "step": 110650
    },
    {
      "epoch": 0.10315302566323022,
      "grad_norm": 1.0472875833511353,
      "learning_rate": 0.00017936959812672919,
      "loss": 3.702,
      "step": 110660
    },
    {
      "epoch": 0.10316234728130931,
      "grad_norm": 1.0960618257522583,
      "learning_rate": 0.00017936773380463397,
      "loss": 3.6558,
      "step": 110670
    },
    {
      "epoch": 0.10317166889938839,
      "grad_norm": 1.0887638330459595,
      "learning_rate": 0.00017936586948253877,
      "loss": 3.6555,
      "step": 110680
    },
    {
      "epoch": 0.10318099051746749,
      "grad_norm": 1.0560210943222046,
      "learning_rate": 0.00017936400516044355,
      "loss": 3.7291,
      "step": 110690
    },
    {
      "epoch": 0.10319031213554658,
      "grad_norm": 1.3278852701187134,
      "learning_rate": 0.00017936214083834836,
      "loss": 3.7014,
      "step": 110700
    },
    {
      "epoch": 0.10319963375362567,
      "grad_norm": 1.134355902671814,
      "learning_rate": 0.0001793602765162532,
      "loss": 3.568,
      "step": 110710
    },
    {
      "epoch": 0.10320895537170477,
      "grad_norm": 1.0327059030532837,
      "learning_rate": 0.00017935841219415797,
      "loss": 3.6796,
      "step": 110720
    },
    {
      "epoch": 0.10321827698978385,
      "grad_norm": 1.0881435871124268,
      "learning_rate": 0.00017935654787206278,
      "loss": 3.6341,
      "step": 110730
    },
    {
      "epoch": 0.10322759860786294,
      "grad_norm": 1.285476803779602,
      "learning_rate": 0.00017935468354996755,
      "loss": 3.7189,
      "step": 110740
    },
    {
      "epoch": 0.10323692022594204,
      "grad_norm": 0.9802242517471313,
      "learning_rate": 0.00017935281922787236,
      "loss": 3.9557,
      "step": 110750
    },
    {
      "epoch": 0.10324624184402113,
      "grad_norm": 1.1269034147262573,
      "learning_rate": 0.00017935095490577717,
      "loss": 3.7414,
      "step": 110760
    },
    {
      "epoch": 0.10325556346210023,
      "grad_norm": 0.8273032903671265,
      "learning_rate": 0.00017934909058368197,
      "loss": 3.7829,
      "step": 110770
    },
    {
      "epoch": 0.10326488508017932,
      "grad_norm": 1.2535784244537354,
      "learning_rate": 0.00017934722626158678,
      "loss": 3.6944,
      "step": 110780
    },
    {
      "epoch": 0.1032742066982584,
      "grad_norm": 1.184275507926941,
      "learning_rate": 0.00017934536193949156,
      "loss": 3.6004,
      "step": 110790
    },
    {
      "epoch": 0.1032835283163375,
      "grad_norm": 1.1528633832931519,
      "learning_rate": 0.00017934349761739637,
      "loss": 3.7544,
      "step": 110800
    },
    {
      "epoch": 0.10329284993441659,
      "grad_norm": 1.0152466297149658,
      "learning_rate": 0.00017934163329530117,
      "loss": 3.7119,
      "step": 110810
    },
    {
      "epoch": 0.10330217155249569,
      "grad_norm": 0.9267253875732422,
      "learning_rate": 0.00017933976897320598,
      "loss": 3.6476,
      "step": 110820
    },
    {
      "epoch": 0.10331149317057478,
      "grad_norm": 0.9744769334793091,
      "learning_rate": 0.00017933790465111078,
      "loss": 3.8123,
      "step": 110830
    },
    {
      "epoch": 0.10332081478865386,
      "grad_norm": 1.0913689136505127,
      "learning_rate": 0.00017933604032901556,
      "loss": 3.6686,
      "step": 110840
    },
    {
      "epoch": 0.10333013640673296,
      "grad_norm": 1.1776951551437378,
      "learning_rate": 0.00017933417600692037,
      "loss": 3.6264,
      "step": 110850
    },
    {
      "epoch": 0.10333945802481205,
      "grad_norm": 0.9979063868522644,
      "learning_rate": 0.00017933231168482518,
      "loss": 3.5617,
      "step": 110860
    },
    {
      "epoch": 0.10334877964289114,
      "grad_norm": 1.0694999694824219,
      "learning_rate": 0.00017933044736272998,
      "loss": 3.6905,
      "step": 110870
    },
    {
      "epoch": 0.10335810126097024,
      "grad_norm": 1.0080066919326782,
      "learning_rate": 0.0001793285830406348,
      "loss": 3.7436,
      "step": 110880
    },
    {
      "epoch": 0.10336742287904932,
      "grad_norm": 1.142931342124939,
      "learning_rate": 0.00017932671871853957,
      "loss": 3.5321,
      "step": 110890
    },
    {
      "epoch": 0.10337674449712841,
      "grad_norm": 1.1500744819641113,
      "learning_rate": 0.00017932485439644437,
      "loss": 3.7252,
      "step": 110900
    },
    {
      "epoch": 0.10338606611520751,
      "grad_norm": 1.0300170183181763,
      "learning_rate": 0.00017932299007434918,
      "loss": 3.6328,
      "step": 110910
    },
    {
      "epoch": 0.1033953877332866,
      "grad_norm": 1.1913771629333496,
      "learning_rate": 0.00017932112575225396,
      "loss": 3.801,
      "step": 110920
    },
    {
      "epoch": 0.1034047093513657,
      "grad_norm": 1.1559228897094727,
      "learning_rate": 0.0001793192614301588,
      "loss": 3.8606,
      "step": 110930
    },
    {
      "epoch": 0.10341403096944478,
      "grad_norm": 1.1247929334640503,
      "learning_rate": 0.00017931739710806357,
      "loss": 3.6982,
      "step": 110940
    },
    {
      "epoch": 0.10342335258752387,
      "grad_norm": 1.0310138463974,
      "learning_rate": 0.00017931553278596838,
      "loss": 3.7356,
      "step": 110950
    },
    {
      "epoch": 0.10343267420560297,
      "grad_norm": 0.8531146049499512,
      "learning_rate": 0.00017931366846387319,
      "loss": 3.6684,
      "step": 110960
    },
    {
      "epoch": 0.10344199582368206,
      "grad_norm": 1.0520987510681152,
      "learning_rate": 0.00017931180414177796,
      "loss": 3.6439,
      "step": 110970
    },
    {
      "epoch": 0.10345131744176116,
      "grad_norm": 1.214035987854004,
      "learning_rate": 0.00017930993981968277,
      "loss": 3.6944,
      "step": 110980
    },
    {
      "epoch": 0.10346063905984024,
      "grad_norm": 0.8889604806900024,
      "learning_rate": 0.00017930807549758758,
      "loss": 3.8457,
      "step": 110990
    },
    {
      "epoch": 0.10346996067791933,
      "grad_norm": 1.0533548593521118,
      "learning_rate": 0.00017930621117549238,
      "loss": 3.8613,
      "step": 111000
    },
    {
      "epoch": 0.10347928229599843,
      "grad_norm": 1.0015431642532349,
      "learning_rate": 0.0001793043468533972,
      "loss": 3.7852,
      "step": 111010
    },
    {
      "epoch": 0.10348860391407752,
      "grad_norm": 1.1345582008361816,
      "learning_rate": 0.00017930248253130197,
      "loss": 3.6994,
      "step": 111020
    },
    {
      "epoch": 0.10349792553215661,
      "grad_norm": 1.1099085807800293,
      "learning_rate": 0.00017930061820920678,
      "loss": 3.5614,
      "step": 111030
    },
    {
      "epoch": 0.10350724715023571,
      "grad_norm": 0.8207852244377136,
      "learning_rate": 0.00017929875388711158,
      "loss": 3.6767,
      "step": 111040
    },
    {
      "epoch": 0.10351656876831479,
      "grad_norm": 1.0909111499786377,
      "learning_rate": 0.0001792968895650164,
      "loss": 3.6421,
      "step": 111050
    },
    {
      "epoch": 0.10352589038639388,
      "grad_norm": 1.1402292251586914,
      "learning_rate": 0.0001792950252429212,
      "loss": 3.5652,
      "step": 111060
    },
    {
      "epoch": 0.10353521200447298,
      "grad_norm": 0.8695898056030273,
      "learning_rate": 0.00017929316092082597,
      "loss": 3.755,
      "step": 111070
    },
    {
      "epoch": 0.10354453362255207,
      "grad_norm": 1.020575761795044,
      "learning_rate": 0.00017929129659873078,
      "loss": 3.6545,
      "step": 111080
    },
    {
      "epoch": 0.10355385524063117,
      "grad_norm": 1.0197159051895142,
      "learning_rate": 0.00017928943227663556,
      "loss": 3.62,
      "step": 111090
    },
    {
      "epoch": 0.10356317685871025,
      "grad_norm": 1.2059154510498047,
      "learning_rate": 0.0001792875679545404,
      "loss": 3.9808,
      "step": 111100
    },
    {
      "epoch": 0.10357249847678934,
      "grad_norm": 0.8243202567100525,
      "learning_rate": 0.0001792857036324452,
      "loss": 3.8636,
      "step": 111110
    },
    {
      "epoch": 0.10358182009486844,
      "grad_norm": 1.1446194648742676,
      "learning_rate": 0.00017928383931034998,
      "loss": 3.5291,
      "step": 111120
    },
    {
      "epoch": 0.10359114171294753,
      "grad_norm": 0.971986711025238,
      "learning_rate": 0.00017928197498825478,
      "loss": 3.6435,
      "step": 111130
    },
    {
      "epoch": 0.10360046333102663,
      "grad_norm": 1.0036306381225586,
      "learning_rate": 0.00017928011066615956,
      "loss": 3.8585,
      "step": 111140
    },
    {
      "epoch": 0.10360978494910571,
      "grad_norm": 1.0226842164993286,
      "learning_rate": 0.00017927824634406437,
      "loss": 3.7127,
      "step": 111150
    },
    {
      "epoch": 0.1036191065671848,
      "grad_norm": 1.0310544967651367,
      "learning_rate": 0.00017927638202196918,
      "loss": 3.6687,
      "step": 111160
    },
    {
      "epoch": 0.1036284281852639,
      "grad_norm": 1.146376609802246,
      "learning_rate": 0.00017927451769987398,
      "loss": 3.669,
      "step": 111170
    },
    {
      "epoch": 0.10363774980334299,
      "grad_norm": 1.0146328210830688,
      "learning_rate": 0.0001792726533777788,
      "loss": 3.7169,
      "step": 111180
    },
    {
      "epoch": 0.10364707142142209,
      "grad_norm": 1.0276398658752441,
      "learning_rate": 0.00017927078905568357,
      "loss": 3.8022,
      "step": 111190
    },
    {
      "epoch": 0.10365639303950117,
      "grad_norm": 1.045949935913086,
      "learning_rate": 0.00017926892473358837,
      "loss": 3.6721,
      "step": 111200
    },
    {
      "epoch": 0.10366571465758026,
      "grad_norm": 1.023749589920044,
      "learning_rate": 0.00017926706041149318,
      "loss": 3.6558,
      "step": 111210
    },
    {
      "epoch": 0.10367503627565935,
      "grad_norm": 0.9987607002258301,
      "learning_rate": 0.000179265196089398,
      "loss": 3.6882,
      "step": 111220
    },
    {
      "epoch": 0.10368435789373845,
      "grad_norm": 1.110124111175537,
      "learning_rate": 0.0001792633317673028,
      "loss": 3.7684,
      "step": 111230
    },
    {
      "epoch": 0.10369367951181754,
      "grad_norm": 1.0695356130599976,
      "learning_rate": 0.00017926146744520757,
      "loss": 3.7249,
      "step": 111240
    },
    {
      "epoch": 0.10370300112989664,
      "grad_norm": 1.057790994644165,
      "learning_rate": 0.00017925960312311238,
      "loss": 3.7027,
      "step": 111250
    },
    {
      "epoch": 0.10371232274797572,
      "grad_norm": 1.147108793258667,
      "learning_rate": 0.00017925773880101718,
      "loss": 3.6762,
      "step": 111260
    },
    {
      "epoch": 0.10372164436605481,
      "grad_norm": 1.0550681352615356,
      "learning_rate": 0.000179255874478922,
      "loss": 3.5998,
      "step": 111270
    },
    {
      "epoch": 0.10373096598413391,
      "grad_norm": 1.0488555431365967,
      "learning_rate": 0.0001792540101568268,
      "loss": 3.5799,
      "step": 111280
    },
    {
      "epoch": 0.103740287602213,
      "grad_norm": 0.8959263563156128,
      "learning_rate": 0.00017925214583473158,
      "loss": 3.7345,
      "step": 111290
    },
    {
      "epoch": 0.1037496092202921,
      "grad_norm": 0.9378392696380615,
      "learning_rate": 0.00017925028151263638,
      "loss": 3.8263,
      "step": 111300
    },
    {
      "epoch": 0.10375893083837118,
      "grad_norm": 1.0603954792022705,
      "learning_rate": 0.00017924841719054116,
      "loss": 3.6435,
      "step": 111310
    },
    {
      "epoch": 0.10376825245645027,
      "grad_norm": 1.1000539064407349,
      "learning_rate": 0.000179246552868446,
      "loss": 3.8132,
      "step": 111320
    },
    {
      "epoch": 0.10377757407452937,
      "grad_norm": 1.1018697023391724,
      "learning_rate": 0.0001792446885463508,
      "loss": 3.7124,
      "step": 111330
    },
    {
      "epoch": 0.10378689569260846,
      "grad_norm": 1.0148662328720093,
      "learning_rate": 0.00017924282422425558,
      "loss": 3.9717,
      "step": 111340
    },
    {
      "epoch": 0.10379621731068756,
      "grad_norm": 0.9609036445617676,
      "learning_rate": 0.0001792409599021604,
      "loss": 3.7203,
      "step": 111350
    },
    {
      "epoch": 0.10380553892876664,
      "grad_norm": 1.1335930824279785,
      "learning_rate": 0.00017923909558006517,
      "loss": 3.551,
      "step": 111360
    },
    {
      "epoch": 0.10381486054684573,
      "grad_norm": 1.1319109201431274,
      "learning_rate": 0.00017923723125796997,
      "loss": 3.6736,
      "step": 111370
    },
    {
      "epoch": 0.10382418216492482,
      "grad_norm": 1.0308632850646973,
      "learning_rate": 0.0001792353669358748,
      "loss": 3.4419,
      "step": 111380
    },
    {
      "epoch": 0.10383350378300392,
      "grad_norm": 1.243386149406433,
      "learning_rate": 0.00017923350261377959,
      "loss": 3.6045,
      "step": 111390
    },
    {
      "epoch": 0.10384282540108301,
      "grad_norm": 1.1568979024887085,
      "learning_rate": 0.0001792316382916844,
      "loss": 3.642,
      "step": 111400
    },
    {
      "epoch": 0.1038521470191621,
      "grad_norm": 0.9597214460372925,
      "learning_rate": 0.00017922977396958917,
      "loss": 3.7223,
      "step": 111410
    },
    {
      "epoch": 0.10386146863724119,
      "grad_norm": 1.1431736946105957,
      "learning_rate": 0.00017922790964749398,
      "loss": 3.7213,
      "step": 111420
    },
    {
      "epoch": 0.10387079025532028,
      "grad_norm": 1.1826411485671997,
      "learning_rate": 0.00017922604532539878,
      "loss": 3.6966,
      "step": 111430
    },
    {
      "epoch": 0.10388011187339938,
      "grad_norm": 1.0617265701293945,
      "learning_rate": 0.0001792241810033036,
      "loss": 3.3564,
      "step": 111440
    },
    {
      "epoch": 0.10388943349147847,
      "grad_norm": 1.0850297212600708,
      "learning_rate": 0.0001792223166812084,
      "loss": 3.7532,
      "step": 111450
    },
    {
      "epoch": 0.10389875510955757,
      "grad_norm": 1.3453037738800049,
      "learning_rate": 0.00017922045235911318,
      "loss": 3.649,
      "step": 111460
    },
    {
      "epoch": 0.10390807672763665,
      "grad_norm": 1.1846201419830322,
      "learning_rate": 0.00017921858803701798,
      "loss": 3.56,
      "step": 111470
    },
    {
      "epoch": 0.10391739834571574,
      "grad_norm": 1.1973079442977905,
      "learning_rate": 0.0001792167237149228,
      "loss": 3.6522,
      "step": 111480
    },
    {
      "epoch": 0.10392671996379484,
      "grad_norm": 0.9957706332206726,
      "learning_rate": 0.0001792148593928276,
      "loss": 3.7505,
      "step": 111490
    },
    {
      "epoch": 0.10393604158187393,
      "grad_norm": 1.1578617095947266,
      "learning_rate": 0.0001792129950707324,
      "loss": 3.7458,
      "step": 111500
    },
    {
      "epoch": 0.10394536319995303,
      "grad_norm": 1.1099426746368408,
      "learning_rate": 0.00017921113074863718,
      "loss": 3.6735,
      "step": 111510
    },
    {
      "epoch": 0.1039546848180321,
      "grad_norm": 1.1828925609588623,
      "learning_rate": 0.00017920926642654199,
      "loss": 3.7379,
      "step": 111520
    },
    {
      "epoch": 0.1039640064361112,
      "grad_norm": 1.134748101234436,
      "learning_rate": 0.0001792074021044468,
      "loss": 3.5804,
      "step": 111530
    },
    {
      "epoch": 0.1039733280541903,
      "grad_norm": 0.9696246385574341,
      "learning_rate": 0.00017920553778235157,
      "loss": 3.6795,
      "step": 111540
    },
    {
      "epoch": 0.10398264967226939,
      "grad_norm": 1.141525149345398,
      "learning_rate": 0.0001792036734602564,
      "loss": 3.7329,
      "step": 111550
    },
    {
      "epoch": 0.10399197129034848,
      "grad_norm": 1.1260874271392822,
      "learning_rate": 0.00017920180913816118,
      "loss": 3.5682,
      "step": 111560
    },
    {
      "epoch": 0.10400129290842756,
      "grad_norm": 1.0843292474746704,
      "learning_rate": 0.000179199944816066,
      "loss": 3.6958,
      "step": 111570
    },
    {
      "epoch": 0.10401061452650666,
      "grad_norm": 1.0026311874389648,
      "learning_rate": 0.0001791980804939708,
      "loss": 3.746,
      "step": 111580
    },
    {
      "epoch": 0.10401993614458575,
      "grad_norm": 1.0782256126403809,
      "learning_rate": 0.00017919621617187558,
      "loss": 3.6054,
      "step": 111590
    },
    {
      "epoch": 0.10402925776266485,
      "grad_norm": 0.9827002882957458,
      "learning_rate": 0.00017919435184978038,
      "loss": 3.8358,
      "step": 111600
    },
    {
      "epoch": 0.10403857938074394,
      "grad_norm": 1.0999642610549927,
      "learning_rate": 0.0001791924875276852,
      "loss": 3.6329,
      "step": 111610
    },
    {
      "epoch": 0.10404790099882302,
      "grad_norm": 0.9837241768836975,
      "learning_rate": 0.00017919062320559,
      "loss": 3.6862,
      "step": 111620
    },
    {
      "epoch": 0.10405722261690212,
      "grad_norm": 1.0653822422027588,
      "learning_rate": 0.0001791887588834948,
      "loss": 3.7035,
      "step": 111630
    },
    {
      "epoch": 0.10406654423498121,
      "grad_norm": 1.0362765789031982,
      "learning_rate": 0.00017918689456139958,
      "loss": 3.594,
      "step": 111640
    },
    {
      "epoch": 0.1040758658530603,
      "grad_norm": 1.072546124458313,
      "learning_rate": 0.0001791850302393044,
      "loss": 3.5189,
      "step": 111650
    },
    {
      "epoch": 0.1040851874711394,
      "grad_norm": 1.0845295190811157,
      "learning_rate": 0.0001791831659172092,
      "loss": 3.5959,
      "step": 111660
    },
    {
      "epoch": 0.10409450908921848,
      "grad_norm": 1.0524530410766602,
      "learning_rate": 0.000179181301595114,
      "loss": 3.5883,
      "step": 111670
    },
    {
      "epoch": 0.10410383070729758,
      "grad_norm": 1.10077702999115,
      "learning_rate": 0.0001791794372730188,
      "loss": 3.578,
      "step": 111680
    },
    {
      "epoch": 0.10411315232537667,
      "grad_norm": 1.2641783952713013,
      "learning_rate": 0.00017917757295092359,
      "loss": 3.6599,
      "step": 111690
    },
    {
      "epoch": 0.10412247394345577,
      "grad_norm": 1.0170564651489258,
      "learning_rate": 0.0001791757086288284,
      "loss": 3.6777,
      "step": 111700
    },
    {
      "epoch": 0.10413179556153486,
      "grad_norm": 1.0627682209014893,
      "learning_rate": 0.0001791738443067332,
      "loss": 3.6953,
      "step": 111710
    },
    {
      "epoch": 0.10414111717961395,
      "grad_norm": 1.193581461906433,
      "learning_rate": 0.000179171979984638,
      "loss": 3.7817,
      "step": 111720
    },
    {
      "epoch": 0.10415043879769303,
      "grad_norm": 1.230355978012085,
      "learning_rate": 0.0001791701156625428,
      "loss": 3.5563,
      "step": 111730
    },
    {
      "epoch": 0.10415976041577213,
      "grad_norm": 1.122516393661499,
      "learning_rate": 0.0001791682513404476,
      "loss": 3.6898,
      "step": 111740
    },
    {
      "epoch": 0.10416908203385122,
      "grad_norm": 1.1273434162139893,
      "learning_rate": 0.0001791663870183524,
      "loss": 3.7755,
      "step": 111750
    },
    {
      "epoch": 0.10417840365193032,
      "grad_norm": 1.1981199979782104,
      "learning_rate": 0.00017916452269625718,
      "loss": 3.7154,
      "step": 111760
    },
    {
      "epoch": 0.10418772527000941,
      "grad_norm": 1.032113790512085,
      "learning_rate": 0.000179162658374162,
      "loss": 3.6508,
      "step": 111770
    },
    {
      "epoch": 0.1041970468880885,
      "grad_norm": 0.9367120265960693,
      "learning_rate": 0.00017916079405206681,
      "loss": 3.5852,
      "step": 111780
    },
    {
      "epoch": 0.10420636850616759,
      "grad_norm": 1.3225126266479492,
      "learning_rate": 0.0001791589297299716,
      "loss": 3.3801,
      "step": 111790
    },
    {
      "epoch": 0.10421569012424668,
      "grad_norm": 1.0520704984664917,
      "learning_rate": 0.0001791570654078764,
      "loss": 3.6246,
      "step": 111800
    },
    {
      "epoch": 0.10422501174232578,
      "grad_norm": 1.2324186563491821,
      "learning_rate": 0.00017915520108578118,
      "loss": 3.7754,
      "step": 111810
    },
    {
      "epoch": 0.10423433336040487,
      "grad_norm": 1.194620132446289,
      "learning_rate": 0.00017915333676368599,
      "loss": 3.6721,
      "step": 111820
    },
    {
      "epoch": 0.10424365497848395,
      "grad_norm": 0.9208627343177795,
      "learning_rate": 0.00017915147244159082,
      "loss": 3.7171,
      "step": 111830
    },
    {
      "epoch": 0.10425297659656305,
      "grad_norm": 1.068359613418579,
      "learning_rate": 0.0001791496081194956,
      "loss": 3.6202,
      "step": 111840
    },
    {
      "epoch": 0.10426229821464214,
      "grad_norm": 1.2002017498016357,
      "learning_rate": 0.0001791477437974004,
      "loss": 3.6479,
      "step": 111850
    },
    {
      "epoch": 0.10427161983272124,
      "grad_norm": 0.9312076568603516,
      "learning_rate": 0.00017914587947530518,
      "loss": 3.747,
      "step": 111860
    },
    {
      "epoch": 0.10428094145080033,
      "grad_norm": 1.0297560691833496,
      "learning_rate": 0.00017914401515321,
      "loss": 3.5745,
      "step": 111870
    },
    {
      "epoch": 0.10429026306887941,
      "grad_norm": 1.2123838663101196,
      "learning_rate": 0.0001791421508311148,
      "loss": 3.8473,
      "step": 111880
    },
    {
      "epoch": 0.1042995846869585,
      "grad_norm": 1.127999186515808,
      "learning_rate": 0.0001791402865090196,
      "loss": 3.62,
      "step": 111890
    },
    {
      "epoch": 0.1043089063050376,
      "grad_norm": 1.1266703605651855,
      "learning_rate": 0.0001791384221869244,
      "loss": 3.8516,
      "step": 111900
    },
    {
      "epoch": 0.1043182279231167,
      "grad_norm": 1.1779206991195679,
      "learning_rate": 0.0001791365578648292,
      "loss": 3.7297,
      "step": 111910
    },
    {
      "epoch": 0.10432754954119579,
      "grad_norm": 1.2495696544647217,
      "learning_rate": 0.000179134693542734,
      "loss": 3.683,
      "step": 111920
    },
    {
      "epoch": 0.10433687115927488,
      "grad_norm": 1.0352888107299805,
      "learning_rate": 0.0001791328292206388,
      "loss": 3.7799,
      "step": 111930
    },
    {
      "epoch": 0.10434619277735396,
      "grad_norm": 1.1644461154937744,
      "learning_rate": 0.0001791309648985436,
      "loss": 3.4117,
      "step": 111940
    },
    {
      "epoch": 0.10435551439543306,
      "grad_norm": 1.0192216634750366,
      "learning_rate": 0.00017912910057644841,
      "loss": 3.7786,
      "step": 111950
    },
    {
      "epoch": 0.10436483601351215,
      "grad_norm": 1.1912250518798828,
      "learning_rate": 0.0001791272362543532,
      "loss": 3.6109,
      "step": 111960
    },
    {
      "epoch": 0.10437415763159125,
      "grad_norm": 1.295719027519226,
      "learning_rate": 0.000179125371932258,
      "loss": 3.7349,
      "step": 111970
    },
    {
      "epoch": 0.10438347924967034,
      "grad_norm": 1.1846896409988403,
      "learning_rate": 0.0001791235076101628,
      "loss": 3.7024,
      "step": 111980
    },
    {
      "epoch": 0.10439280086774942,
      "grad_norm": 1.142122507095337,
      "learning_rate": 0.00017912164328806758,
      "loss": 3.7473,
      "step": 111990
    },
    {
      "epoch": 0.10440212248582852,
      "grad_norm": 1.1507267951965332,
      "learning_rate": 0.00017911977896597242,
      "loss": 3.8014,
      "step": 112000
    },
    {
      "epoch": 0.10441144410390761,
      "grad_norm": 0.9337475299835205,
      "learning_rate": 0.0001791179146438772,
      "loss": 3.7665,
      "step": 112010
    },
    {
      "epoch": 0.1044207657219867,
      "grad_norm": 1.2483513355255127,
      "learning_rate": 0.000179116050321782,
      "loss": 3.9459,
      "step": 112020
    },
    {
      "epoch": 0.1044300873400658,
      "grad_norm": 0.946890115737915,
      "learning_rate": 0.0001791141859996868,
      "loss": 3.5585,
      "step": 112030
    },
    {
      "epoch": 0.10443940895814488,
      "grad_norm": 1.1335786581039429,
      "learning_rate": 0.0001791123216775916,
      "loss": 3.7579,
      "step": 112040
    },
    {
      "epoch": 0.10444873057622397,
      "grad_norm": 1.0976557731628418,
      "learning_rate": 0.0001791104573554964,
      "loss": 3.6886,
      "step": 112050
    },
    {
      "epoch": 0.10445805219430307,
      "grad_norm": 1.1310018301010132,
      "learning_rate": 0.0001791085930334012,
      "loss": 3.5331,
      "step": 112060
    },
    {
      "epoch": 0.10446737381238216,
      "grad_norm": 0.983887791633606,
      "learning_rate": 0.000179106728711306,
      "loss": 3.6048,
      "step": 112070
    },
    {
      "epoch": 0.10447669543046126,
      "grad_norm": 0.9259463548660278,
      "learning_rate": 0.00017910486438921081,
      "loss": 3.5594,
      "step": 112080
    },
    {
      "epoch": 0.10448601704854034,
      "grad_norm": 1.059718370437622,
      "learning_rate": 0.0001791030000671156,
      "loss": 3.6393,
      "step": 112090
    },
    {
      "epoch": 0.10449533866661943,
      "grad_norm": 0.8521909117698669,
      "learning_rate": 0.0001791011357450204,
      "loss": 3.7285,
      "step": 112100
    },
    {
      "epoch": 0.10450466028469853,
      "grad_norm": 0.9864389896392822,
      "learning_rate": 0.0001790992714229252,
      "loss": 3.58,
      "step": 112110
    },
    {
      "epoch": 0.10451398190277762,
      "grad_norm": 1.2755942344665527,
      "learning_rate": 0.00017909740710083,
      "loss": 3.692,
      "step": 112120
    },
    {
      "epoch": 0.10452330352085672,
      "grad_norm": 1.0077145099639893,
      "learning_rate": 0.00017909554277873482,
      "loss": 3.5504,
      "step": 112130
    },
    {
      "epoch": 0.10453262513893581,
      "grad_norm": 0.905451238155365,
      "learning_rate": 0.0001790936784566396,
      "loss": 3.606,
      "step": 112140
    },
    {
      "epoch": 0.10454194675701489,
      "grad_norm": 0.9476967453956604,
      "learning_rate": 0.0001790918141345444,
      "loss": 3.7687,
      "step": 112150
    },
    {
      "epoch": 0.10455126837509399,
      "grad_norm": 1.1762441396713257,
      "learning_rate": 0.0001790899498124492,
      "loss": 3.8202,
      "step": 112160
    },
    {
      "epoch": 0.10456058999317308,
      "grad_norm": 1.0154598951339722,
      "learning_rate": 0.00017908808549035402,
      "loss": 3.7062,
      "step": 112170
    },
    {
      "epoch": 0.10456991161125218,
      "grad_norm": 1.2987624406814575,
      "learning_rate": 0.00017908622116825882,
      "loss": 3.8129,
      "step": 112180
    },
    {
      "epoch": 0.10457923322933127,
      "grad_norm": 0.9624232649803162,
      "learning_rate": 0.0001790843568461636,
      "loss": 3.8408,
      "step": 112190
    },
    {
      "epoch": 0.10458855484741035,
      "grad_norm": 1.0065242052078247,
      "learning_rate": 0.0001790824925240684,
      "loss": 3.7497,
      "step": 112200
    },
    {
      "epoch": 0.10459787646548944,
      "grad_norm": 1.0486708879470825,
      "learning_rate": 0.0001790806282019732,
      "loss": 3.5995,
      "step": 112210
    },
    {
      "epoch": 0.10460719808356854,
      "grad_norm": 1.2252237796783447,
      "learning_rate": 0.00017907876387987802,
      "loss": 3.7739,
      "step": 112220
    },
    {
      "epoch": 0.10461651970164763,
      "grad_norm": 1.2909140586853027,
      "learning_rate": 0.00017907689955778283,
      "loss": 3.6582,
      "step": 112230
    },
    {
      "epoch": 0.10462584131972673,
      "grad_norm": 1.1853649616241455,
      "learning_rate": 0.0001790750352356876,
      "loss": 3.809,
      "step": 112240
    },
    {
      "epoch": 0.10463516293780581,
      "grad_norm": 0.9246610999107361,
      "learning_rate": 0.0001790731709135924,
      "loss": 3.6629,
      "step": 112250
    },
    {
      "epoch": 0.1046444845558849,
      "grad_norm": 1.1097757816314697,
      "learning_rate": 0.0001790713065914972,
      "loss": 3.7468,
      "step": 112260
    },
    {
      "epoch": 0.104653806173964,
      "grad_norm": 1.0601392984390259,
      "learning_rate": 0.000179069442269402,
      "loss": 3.6935,
      "step": 112270
    },
    {
      "epoch": 0.10466312779204309,
      "grad_norm": 1.1455737352371216,
      "learning_rate": 0.00017906757794730683,
      "loss": 3.6272,
      "step": 112280
    },
    {
      "epoch": 0.10467244941012219,
      "grad_norm": 1.098219394683838,
      "learning_rate": 0.0001790657136252116,
      "loss": 3.75,
      "step": 112290
    },
    {
      "epoch": 0.10468177102820127,
      "grad_norm": 1.1370301246643066,
      "learning_rate": 0.00017906384930311642,
      "loss": 3.4903,
      "step": 112300
    },
    {
      "epoch": 0.10469109264628036,
      "grad_norm": 1.0117000341415405,
      "learning_rate": 0.0001790619849810212,
      "loss": 3.541,
      "step": 112310
    },
    {
      "epoch": 0.10470041426435946,
      "grad_norm": 1.025274634361267,
      "learning_rate": 0.000179060120658926,
      "loss": 3.8121,
      "step": 112320
    },
    {
      "epoch": 0.10470973588243855,
      "grad_norm": 0.974202573299408,
      "learning_rate": 0.0001790582563368308,
      "loss": 3.6576,
      "step": 112330
    },
    {
      "epoch": 0.10471905750051765,
      "grad_norm": 1.0302646160125732,
      "learning_rate": 0.00017905639201473562,
      "loss": 3.6663,
      "step": 112340
    },
    {
      "epoch": 0.10472837911859673,
      "grad_norm": 1.1166936159133911,
      "learning_rate": 0.00017905452769264042,
      "loss": 3.7743,
      "step": 112350
    },
    {
      "epoch": 0.10473770073667582,
      "grad_norm": 1.1251282691955566,
      "learning_rate": 0.0001790526633705452,
      "loss": 3.7692,
      "step": 112360
    },
    {
      "epoch": 0.10474702235475492,
      "grad_norm": 1.077360987663269,
      "learning_rate": 0.00017905079904845,
      "loss": 3.6055,
      "step": 112370
    },
    {
      "epoch": 0.10475634397283401,
      "grad_norm": 0.9831839203834534,
      "learning_rate": 0.00017904893472635481,
      "loss": 3.6136,
      "step": 112380
    },
    {
      "epoch": 0.1047656655909131,
      "grad_norm": 1.1010282039642334,
      "learning_rate": 0.00017904707040425962,
      "loss": 3.6897,
      "step": 112390
    },
    {
      "epoch": 0.1047749872089922,
      "grad_norm": 1.2478246688842773,
      "learning_rate": 0.00017904520608216443,
      "loss": 3.6355,
      "step": 112400
    },
    {
      "epoch": 0.10478430882707128,
      "grad_norm": 1.0597375631332397,
      "learning_rate": 0.0001790433417600692,
      "loss": 3.6758,
      "step": 112410
    },
    {
      "epoch": 0.10479363044515037,
      "grad_norm": 1.0190232992172241,
      "learning_rate": 0.000179041477437974,
      "loss": 3.7325,
      "step": 112420
    },
    {
      "epoch": 0.10480295206322947,
      "grad_norm": 1.1640697717666626,
      "learning_rate": 0.00017903961311587882,
      "loss": 3.7683,
      "step": 112430
    },
    {
      "epoch": 0.10481227368130856,
      "grad_norm": 1.1363246440887451,
      "learning_rate": 0.0001790377487937836,
      "loss": 3.6519,
      "step": 112440
    },
    {
      "epoch": 0.10482159529938766,
      "grad_norm": 1.102710247039795,
      "learning_rate": 0.00017903588447168843,
      "loss": 3.7498,
      "step": 112450
    },
    {
      "epoch": 0.10483091691746674,
      "grad_norm": 0.999248206615448,
      "learning_rate": 0.0001790340201495932,
      "loss": 3.877,
      "step": 112460
    },
    {
      "epoch": 0.10484023853554583,
      "grad_norm": 1.112524151802063,
      "learning_rate": 0.00017903215582749802,
      "loss": 3.6693,
      "step": 112470
    },
    {
      "epoch": 0.10484956015362493,
      "grad_norm": 1.1797635555267334,
      "learning_rate": 0.0001790302915054028,
      "loss": 3.5701,
      "step": 112480
    },
    {
      "epoch": 0.10485888177170402,
      "grad_norm": 1.5572094917297363,
      "learning_rate": 0.0001790284271833076,
      "loss": 3.4998,
      "step": 112490
    },
    {
      "epoch": 0.10486820338978312,
      "grad_norm": 0.9137073159217834,
      "learning_rate": 0.0001790265628612124,
      "loss": 3.8059,
      "step": 112500
    },
    {
      "epoch": 0.1048775250078622,
      "grad_norm": 1.092854619026184,
      "learning_rate": 0.00017902469853911721,
      "loss": 3.6542,
      "step": 112510
    },
    {
      "epoch": 0.10488684662594129,
      "grad_norm": 1.1078518629074097,
      "learning_rate": 0.00017902283421702202,
      "loss": 3.6557,
      "step": 112520
    },
    {
      "epoch": 0.10489616824402039,
      "grad_norm": 1.0599324703216553,
      "learning_rate": 0.0001790209698949268,
      "loss": 3.709,
      "step": 112530
    },
    {
      "epoch": 0.10490548986209948,
      "grad_norm": 1.162428855895996,
      "learning_rate": 0.0001790191055728316,
      "loss": 3.7832,
      "step": 112540
    },
    {
      "epoch": 0.10491481148017857,
      "grad_norm": 0.9242422580718994,
      "learning_rate": 0.0001790172412507364,
      "loss": 3.8096,
      "step": 112550
    },
    {
      "epoch": 0.10492413309825765,
      "grad_norm": 1.0316507816314697,
      "learning_rate": 0.00017901537692864122,
      "loss": 3.6938,
      "step": 112560
    },
    {
      "epoch": 0.10493345471633675,
      "grad_norm": 1.1473709344863892,
      "learning_rate": 0.00017901351260654603,
      "loss": 3.4409,
      "step": 112570
    },
    {
      "epoch": 0.10494277633441584,
      "grad_norm": 1.0255497694015503,
      "learning_rate": 0.0001790116482844508,
      "loss": 3.9333,
      "step": 112580
    },
    {
      "epoch": 0.10495209795249494,
      "grad_norm": 1.2043077945709229,
      "learning_rate": 0.0001790097839623556,
      "loss": 3.7122,
      "step": 112590
    },
    {
      "epoch": 0.10496141957057403,
      "grad_norm": 1.0220160484313965,
      "learning_rate": 0.00017900791964026042,
      "loss": 3.9268,
      "step": 112600
    },
    {
      "epoch": 0.10497074118865313,
      "grad_norm": 1.038507103919983,
      "learning_rate": 0.00017900605531816522,
      "loss": 3.7199,
      "step": 112610
    },
    {
      "epoch": 0.10498006280673221,
      "grad_norm": 0.9654355049133301,
      "learning_rate": 0.00017900419099607003,
      "loss": 3.5544,
      "step": 112620
    },
    {
      "epoch": 0.1049893844248113,
      "grad_norm": 1.067135214805603,
      "learning_rate": 0.0001790023266739748,
      "loss": 3.5391,
      "step": 112630
    },
    {
      "epoch": 0.1049987060428904,
      "grad_norm": 0.9829971790313721,
      "learning_rate": 0.00017900046235187962,
      "loss": 3.8561,
      "step": 112640
    },
    {
      "epoch": 0.10500802766096949,
      "grad_norm": 1.065313696861267,
      "learning_rate": 0.00017899859802978442,
      "loss": 3.5772,
      "step": 112650
    },
    {
      "epoch": 0.10501734927904859,
      "grad_norm": 1.0924004316329956,
      "learning_rate": 0.0001789967337076892,
      "loss": 3.6651,
      "step": 112660
    },
    {
      "epoch": 0.10502667089712767,
      "grad_norm": 0.8937317132949829,
      "learning_rate": 0.00017899486938559403,
      "loss": 3.5482,
      "step": 112670
    },
    {
      "epoch": 0.10503599251520676,
      "grad_norm": 0.9636914134025574,
      "learning_rate": 0.0001789930050634988,
      "loss": 3.7456,
      "step": 112680
    },
    {
      "epoch": 0.10504531413328586,
      "grad_norm": 1.2691479921340942,
      "learning_rate": 0.00017899114074140362,
      "loss": 3.6425,
      "step": 112690
    },
    {
      "epoch": 0.10505463575136495,
      "grad_norm": 1.1245845556259155,
      "learning_rate": 0.00017898927641930843,
      "loss": 3.7849,
      "step": 112700
    },
    {
      "epoch": 0.10506395736944404,
      "grad_norm": 0.9302534461021423,
      "learning_rate": 0.0001789874120972132,
      "loss": 3.6923,
      "step": 112710
    },
    {
      "epoch": 0.10507327898752312,
      "grad_norm": 1.1598575115203857,
      "learning_rate": 0.000178985547775118,
      "loss": 3.7225,
      "step": 112720
    },
    {
      "epoch": 0.10508260060560222,
      "grad_norm": 1.2395483255386353,
      "learning_rate": 0.00017898368345302282,
      "loss": 3.9605,
      "step": 112730
    },
    {
      "epoch": 0.10509192222368131,
      "grad_norm": 0.9964959621429443,
      "learning_rate": 0.00017898181913092762,
      "loss": 3.621,
      "step": 112740
    },
    {
      "epoch": 0.10510124384176041,
      "grad_norm": 1.248435378074646,
      "learning_rate": 0.00017897995480883243,
      "loss": 3.6073,
      "step": 112750
    },
    {
      "epoch": 0.1051105654598395,
      "grad_norm": 1.0722613334655762,
      "learning_rate": 0.0001789780904867372,
      "loss": 3.5664,
      "step": 112760
    },
    {
      "epoch": 0.10511988707791858,
      "grad_norm": 1.0813764333724976,
      "learning_rate": 0.00017897622616464202,
      "loss": 3.6761,
      "step": 112770
    },
    {
      "epoch": 0.10512920869599768,
      "grad_norm": 1.3051869869232178,
      "learning_rate": 0.00017897436184254682,
      "loss": 3.7674,
      "step": 112780
    },
    {
      "epoch": 0.10513853031407677,
      "grad_norm": 1.227766990661621,
      "learning_rate": 0.00017897249752045163,
      "loss": 3.7317,
      "step": 112790
    },
    {
      "epoch": 0.10514785193215587,
      "grad_norm": 1.1883344650268555,
      "learning_rate": 0.00017897063319835643,
      "loss": 3.6668,
      "step": 112800
    },
    {
      "epoch": 0.10515717355023496,
      "grad_norm": 1.1153779029846191,
      "learning_rate": 0.00017896876887626121,
      "loss": 3.7649,
      "step": 112810
    },
    {
      "epoch": 0.10516649516831406,
      "grad_norm": 1.1033198833465576,
      "learning_rate": 0.00017896690455416602,
      "loss": 3.8136,
      "step": 112820
    },
    {
      "epoch": 0.10517581678639314,
      "grad_norm": 0.9082194566726685,
      "learning_rate": 0.0001789650402320708,
      "loss": 3.6732,
      "step": 112830
    },
    {
      "epoch": 0.10518513840447223,
      "grad_norm": 0.8043308854103088,
      "learning_rate": 0.00017896317590997563,
      "loss": 3.5728,
      "step": 112840
    },
    {
      "epoch": 0.10519446002255133,
      "grad_norm": 1.3073595762252808,
      "learning_rate": 0.00017896131158788044,
      "loss": 3.6428,
      "step": 112850
    },
    {
      "epoch": 0.10520378164063042,
      "grad_norm": 1.032132625579834,
      "learning_rate": 0.00017895944726578522,
      "loss": 3.6461,
      "step": 112860
    },
    {
      "epoch": 0.10521310325870951,
      "grad_norm": 1.0525671243667603,
      "learning_rate": 0.00017895758294369002,
      "loss": 3.7912,
      "step": 112870
    },
    {
      "epoch": 0.1052224248767886,
      "grad_norm": 0.9851046204566956,
      "learning_rate": 0.0001789557186215948,
      "loss": 3.5233,
      "step": 112880
    },
    {
      "epoch": 0.10523174649486769,
      "grad_norm": 1.0401948690414429,
      "learning_rate": 0.0001789538542994996,
      "loss": 3.7945,
      "step": 112890
    },
    {
      "epoch": 0.10524106811294678,
      "grad_norm": 0.9650993347167969,
      "learning_rate": 0.00017895198997740444,
      "loss": 3.95,
      "step": 112900
    },
    {
      "epoch": 0.10525038973102588,
      "grad_norm": 1.001217246055603,
      "learning_rate": 0.00017895012565530922,
      "loss": 3.4104,
      "step": 112910
    },
    {
      "epoch": 0.10525971134910497,
      "grad_norm": 1.0319730043411255,
      "learning_rate": 0.00017894826133321403,
      "loss": 3.6726,
      "step": 112920
    },
    {
      "epoch": 0.10526903296718405,
      "grad_norm": 0.9859693646430969,
      "learning_rate": 0.0001789463970111188,
      "loss": 3.8175,
      "step": 112930
    },
    {
      "epoch": 0.10527835458526315,
      "grad_norm": 1.0928699970245361,
      "learning_rate": 0.00017894453268902361,
      "loss": 3.7864,
      "step": 112940
    },
    {
      "epoch": 0.10528767620334224,
      "grad_norm": 1.176864504814148,
      "learning_rate": 0.00017894266836692842,
      "loss": 3.6867,
      "step": 112950
    },
    {
      "epoch": 0.10529699782142134,
      "grad_norm": 1.098189115524292,
      "learning_rate": 0.00017894080404483323,
      "loss": 3.7858,
      "step": 112960
    },
    {
      "epoch": 0.10530631943950043,
      "grad_norm": 1.0125923156738281,
      "learning_rate": 0.00017893893972273803,
      "loss": 3.726,
      "step": 112970
    },
    {
      "epoch": 0.10531564105757951,
      "grad_norm": 1.2058900594711304,
      "learning_rate": 0.0001789370754006428,
      "loss": 3.6778,
      "step": 112980
    },
    {
      "epoch": 0.1053249626756586,
      "grad_norm": 1.089557409286499,
      "learning_rate": 0.00017893521107854762,
      "loss": 3.6704,
      "step": 112990
    },
    {
      "epoch": 0.1053342842937377,
      "grad_norm": 1.1604357957839966,
      "learning_rate": 0.00017893334675645243,
      "loss": 3.7014,
      "step": 113000
    },
    {
      "epoch": 0.1053436059118168,
      "grad_norm": 0.9475886821746826,
      "learning_rate": 0.00017893148243435723,
      "loss": 3.5223,
      "step": 113010
    },
    {
      "epoch": 0.10535292752989589,
      "grad_norm": 1.0082238912582397,
      "learning_rate": 0.00017892961811226204,
      "loss": 3.665,
      "step": 113020
    },
    {
      "epoch": 0.10536224914797497,
      "grad_norm": 0.9850847125053406,
      "learning_rate": 0.00017892775379016682,
      "loss": 3.7747,
      "step": 113030
    },
    {
      "epoch": 0.10537157076605407,
      "grad_norm": 0.9308450818061829,
      "learning_rate": 0.00017892588946807162,
      "loss": 3.811,
      "step": 113040
    },
    {
      "epoch": 0.10538089238413316,
      "grad_norm": 1.1639649868011475,
      "learning_rate": 0.00017892402514597643,
      "loss": 3.7309,
      "step": 113050
    },
    {
      "epoch": 0.10539021400221225,
      "grad_norm": 1.0968639850616455,
      "learning_rate": 0.00017892216082388124,
      "loss": 3.683,
      "step": 113060
    },
    {
      "epoch": 0.10539953562029135,
      "grad_norm": 1.0338213443756104,
      "learning_rate": 0.00017892029650178604,
      "loss": 3.5155,
      "step": 113070
    },
    {
      "epoch": 0.10540885723837044,
      "grad_norm": 0.9494690895080566,
      "learning_rate": 0.00017891843217969082,
      "loss": 3.725,
      "step": 113080
    },
    {
      "epoch": 0.10541817885644952,
      "grad_norm": 1.256780982017517,
      "learning_rate": 0.00017891656785759563,
      "loss": 3.7074,
      "step": 113090
    },
    {
      "epoch": 0.10542750047452862,
      "grad_norm": 1.134716272354126,
      "learning_rate": 0.00017891470353550043,
      "loss": 3.6944,
      "step": 113100
    },
    {
      "epoch": 0.10543682209260771,
      "grad_norm": 0.9818904399871826,
      "learning_rate": 0.00017891283921340521,
      "loss": 3.7293,
      "step": 113110
    },
    {
      "epoch": 0.10544614371068681,
      "grad_norm": 1.2105287313461304,
      "learning_rate": 0.00017891097489131005,
      "loss": 3.8333,
      "step": 113120
    },
    {
      "epoch": 0.1054554653287659,
      "grad_norm": 1.1175533533096313,
      "learning_rate": 0.00017890911056921483,
      "loss": 3.6622,
      "step": 113130
    },
    {
      "epoch": 0.10546478694684498,
      "grad_norm": 0.9908333420753479,
      "learning_rate": 0.00017890724624711963,
      "loss": 3.5971,
      "step": 113140
    },
    {
      "epoch": 0.10547410856492408,
      "grad_norm": 1.0772634744644165,
      "learning_rate": 0.00017890538192502444,
      "loss": 3.6126,
      "step": 113150
    },
    {
      "epoch": 0.10548343018300317,
      "grad_norm": 1.0364948511123657,
      "learning_rate": 0.00017890351760292922,
      "loss": 3.621,
      "step": 113160
    },
    {
      "epoch": 0.10549275180108227,
      "grad_norm": 0.9478604197502136,
      "learning_rate": 0.00017890165328083402,
      "loss": 3.7006,
      "step": 113170
    },
    {
      "epoch": 0.10550207341916136,
      "grad_norm": 1.2059311866760254,
      "learning_rate": 0.00017889978895873883,
      "loss": 3.6911,
      "step": 113180
    },
    {
      "epoch": 0.10551139503724044,
      "grad_norm": 1.04911470413208,
      "learning_rate": 0.00017889792463664364,
      "loss": 3.6745,
      "step": 113190
    },
    {
      "epoch": 0.10552071665531954,
      "grad_norm": 1.1050198078155518,
      "learning_rate": 0.00017889606031454844,
      "loss": 3.4746,
      "step": 113200
    },
    {
      "epoch": 0.10553003827339863,
      "grad_norm": 1.3123775720596313,
      "learning_rate": 0.00017889419599245322,
      "loss": 3.6658,
      "step": 113210
    },
    {
      "epoch": 0.10553935989147772,
      "grad_norm": 1.0783218145370483,
      "learning_rate": 0.00017889233167035803,
      "loss": 3.7468,
      "step": 113220
    },
    {
      "epoch": 0.10554868150955682,
      "grad_norm": 1.0940546989440918,
      "learning_rate": 0.00017889046734826284,
      "loss": 3.8153,
      "step": 113230
    },
    {
      "epoch": 0.1055580031276359,
      "grad_norm": 1.107454776763916,
      "learning_rate": 0.00017888860302616764,
      "loss": 3.7598,
      "step": 113240
    },
    {
      "epoch": 0.105567324745715,
      "grad_norm": 1.1523189544677734,
      "learning_rate": 0.00017888673870407245,
      "loss": 3.5771,
      "step": 113250
    },
    {
      "epoch": 0.10557664636379409,
      "grad_norm": 0.8310294151306152,
      "learning_rate": 0.00017888487438197723,
      "loss": 3.7673,
      "step": 113260
    },
    {
      "epoch": 0.10558596798187318,
      "grad_norm": 1.2217066287994385,
      "learning_rate": 0.00017888301005988203,
      "loss": 3.653,
      "step": 113270
    },
    {
      "epoch": 0.10559528959995228,
      "grad_norm": 1.0931836366653442,
      "learning_rate": 0.00017888114573778684,
      "loss": 3.7085,
      "step": 113280
    },
    {
      "epoch": 0.10560461121803137,
      "grad_norm": 1.2279860973358154,
      "learning_rate": 0.00017887928141569165,
      "loss": 3.6494,
      "step": 113290
    },
    {
      "epoch": 0.10561393283611045,
      "grad_norm": 1.0033183097839355,
      "learning_rate": 0.00017887741709359645,
      "loss": 3.5887,
      "step": 113300
    },
    {
      "epoch": 0.10562325445418955,
      "grad_norm": 0.9973013997077942,
      "learning_rate": 0.00017887555277150123,
      "loss": 3.7502,
      "step": 113310
    },
    {
      "epoch": 0.10563257607226864,
      "grad_norm": 1.2295842170715332,
      "learning_rate": 0.00017887368844940604,
      "loss": 3.439,
      "step": 113320
    },
    {
      "epoch": 0.10564189769034774,
      "grad_norm": 1.1163017749786377,
      "learning_rate": 0.00017887182412731082,
      "loss": 3.4888,
      "step": 113330
    },
    {
      "epoch": 0.10565121930842683,
      "grad_norm": 0.9994301795959473,
      "learning_rate": 0.00017886995980521562,
      "loss": 3.6786,
      "step": 113340
    },
    {
      "epoch": 0.10566054092650591,
      "grad_norm": 1.0889523029327393,
      "learning_rate": 0.00017886809548312046,
      "loss": 3.6232,
      "step": 113350
    },
    {
      "epoch": 0.105669862544585,
      "grad_norm": 0.9455580711364746,
      "learning_rate": 0.00017886623116102524,
      "loss": 3.7705,
      "step": 113360
    },
    {
      "epoch": 0.1056791841626641,
      "grad_norm": 1.0054106712341309,
      "learning_rate": 0.00017886436683893004,
      "loss": 3.8467,
      "step": 113370
    },
    {
      "epoch": 0.1056885057807432,
      "grad_norm": 1.0810546875,
      "learning_rate": 0.00017886250251683482,
      "loss": 3.5861,
      "step": 113380
    },
    {
      "epoch": 0.10569782739882229,
      "grad_norm": 1.0497088432312012,
      "learning_rate": 0.00017886063819473963,
      "loss": 3.6768,
      "step": 113390
    },
    {
      "epoch": 0.10570714901690137,
      "grad_norm": 0.9532439708709717,
      "learning_rate": 0.00017885877387264443,
      "loss": 3.6927,
      "step": 113400
    },
    {
      "epoch": 0.10571647063498046,
      "grad_norm": 1.0583218336105347,
      "learning_rate": 0.00017885690955054924,
      "loss": 3.7673,
      "step": 113410
    },
    {
      "epoch": 0.10572579225305956,
      "grad_norm": 0.9219061732292175,
      "learning_rate": 0.00017885504522845405,
      "loss": 3.8366,
      "step": 113420
    },
    {
      "epoch": 0.10573511387113865,
      "grad_norm": 1.0991935729980469,
      "learning_rate": 0.00017885318090635883,
      "loss": 3.7785,
      "step": 113430
    },
    {
      "epoch": 0.10574443548921775,
      "grad_norm": 1.0129002332687378,
      "learning_rate": 0.00017885131658426363,
      "loss": 3.7062,
      "step": 113440
    },
    {
      "epoch": 0.10575375710729683,
      "grad_norm": 1.0195313692092896,
      "learning_rate": 0.00017884945226216844,
      "loss": 3.9243,
      "step": 113450
    },
    {
      "epoch": 0.10576307872537592,
      "grad_norm": 0.8792906403541565,
      "learning_rate": 0.00017884758794007324,
      "loss": 3.7271,
      "step": 113460
    },
    {
      "epoch": 0.10577240034345502,
      "grad_norm": 0.9665362238883972,
      "learning_rate": 0.00017884572361797805,
      "loss": 3.8097,
      "step": 113470
    },
    {
      "epoch": 0.10578172196153411,
      "grad_norm": 0.9644710421562195,
      "learning_rate": 0.00017884385929588283,
      "loss": 3.6955,
      "step": 113480
    },
    {
      "epoch": 0.1057910435796132,
      "grad_norm": 1.0106830596923828,
      "learning_rate": 0.00017884199497378764,
      "loss": 3.7669,
      "step": 113490
    },
    {
      "epoch": 0.1058003651976923,
      "grad_norm": 1.1690912246704102,
      "learning_rate": 0.00017884013065169244,
      "loss": 3.7976,
      "step": 113500
    },
    {
      "epoch": 0.10580968681577138,
      "grad_norm": 1.2016987800598145,
      "learning_rate": 0.00017883826632959725,
      "loss": 3.9252,
      "step": 113510
    },
    {
      "epoch": 0.10581900843385048,
      "grad_norm": 2.609215021133423,
      "learning_rate": 0.00017883640200750206,
      "loss": 3.6136,
      "step": 113520
    },
    {
      "epoch": 0.10582833005192957,
      "grad_norm": 0.9707454442977905,
      "learning_rate": 0.00017883453768540683,
      "loss": 3.6685,
      "step": 113530
    },
    {
      "epoch": 0.10583765167000866,
      "grad_norm": 1.1080636978149414,
      "learning_rate": 0.00017883267336331164,
      "loss": 3.4076,
      "step": 113540
    },
    {
      "epoch": 0.10584697328808776,
      "grad_norm": 1.1104390621185303,
      "learning_rate": 0.00017883080904121645,
      "loss": 3.6137,
      "step": 113550
    },
    {
      "epoch": 0.10585629490616684,
      "grad_norm": 1.1649106740951538,
      "learning_rate": 0.00017882894471912123,
      "loss": 3.6428,
      "step": 113560
    },
    {
      "epoch": 0.10586561652424593,
      "grad_norm": 0.9929066896438599,
      "learning_rate": 0.00017882708039702606,
      "loss": 3.7481,
      "step": 113570
    },
    {
      "epoch": 0.10587493814232503,
      "grad_norm": 0.9236127138137817,
      "learning_rate": 0.00017882521607493084,
      "loss": 3.7076,
      "step": 113580
    },
    {
      "epoch": 0.10588425976040412,
      "grad_norm": 1.331708550453186,
      "learning_rate": 0.00017882335175283565,
      "loss": 3.788,
      "step": 113590
    },
    {
      "epoch": 0.10589358137848322,
      "grad_norm": 1.151749610900879,
      "learning_rate": 0.00017882148743074045,
      "loss": 3.6828,
      "step": 113600
    },
    {
      "epoch": 0.1059029029965623,
      "grad_norm": 1.2291224002838135,
      "learning_rate": 0.00017881962310864523,
      "loss": 3.7307,
      "step": 113610
    },
    {
      "epoch": 0.10591222461464139,
      "grad_norm": 1.2673898935317993,
      "learning_rate": 0.00017881775878655004,
      "loss": 3.8415,
      "step": 113620
    },
    {
      "epoch": 0.10592154623272049,
      "grad_norm": 1.0283862352371216,
      "learning_rate": 0.00017881589446445484,
      "loss": 3.7203,
      "step": 113630
    },
    {
      "epoch": 0.10593086785079958,
      "grad_norm": 1.0066930055618286,
      "learning_rate": 0.00017881403014235965,
      "loss": 3.6858,
      "step": 113640
    },
    {
      "epoch": 0.10594018946887868,
      "grad_norm": 0.9890033602714539,
      "learning_rate": 0.00017881216582026446,
      "loss": 3.7851,
      "step": 113650
    },
    {
      "epoch": 0.10594951108695776,
      "grad_norm": 0.9413126111030579,
      "learning_rate": 0.00017881030149816924,
      "loss": 3.6636,
      "step": 113660
    },
    {
      "epoch": 0.10595883270503685,
      "grad_norm": 0.8782804012298584,
      "learning_rate": 0.00017880843717607404,
      "loss": 3.72,
      "step": 113670
    },
    {
      "epoch": 0.10596815432311595,
      "grad_norm": 0.9753142595291138,
      "learning_rate": 0.00017880657285397885,
      "loss": 3.6756,
      "step": 113680
    },
    {
      "epoch": 0.10597747594119504,
      "grad_norm": 0.9753140211105347,
      "learning_rate": 0.00017880470853188365,
      "loss": 3.539,
      "step": 113690
    },
    {
      "epoch": 0.10598679755927413,
      "grad_norm": 1.0342433452606201,
      "learning_rate": 0.00017880284420978846,
      "loss": 3.7227,
      "step": 113700
    },
    {
      "epoch": 0.10599611917735322,
      "grad_norm": 1.1145663261413574,
      "learning_rate": 0.00017880097988769324,
      "loss": 3.8284,
      "step": 113710
    },
    {
      "epoch": 0.10600544079543231,
      "grad_norm": 1.2026994228363037,
      "learning_rate": 0.00017879911556559805,
      "loss": 3.6451,
      "step": 113720
    },
    {
      "epoch": 0.1060147624135114,
      "grad_norm": 1.22223961353302,
      "learning_rate": 0.00017879725124350285,
      "loss": 3.6912,
      "step": 113730
    },
    {
      "epoch": 0.1060240840315905,
      "grad_norm": 0.9459344744682312,
      "learning_rate": 0.00017879538692140766,
      "loss": 3.8245,
      "step": 113740
    },
    {
      "epoch": 0.1060334056496696,
      "grad_norm": 1.020310878753662,
      "learning_rate": 0.00017879352259931246,
      "loss": 3.8355,
      "step": 113750
    },
    {
      "epoch": 0.10604272726774869,
      "grad_norm": 1.161301851272583,
      "learning_rate": 0.00017879165827721724,
      "loss": 3.7248,
      "step": 113760
    },
    {
      "epoch": 0.10605204888582777,
      "grad_norm": 1.1344789266586304,
      "learning_rate": 0.00017878979395512205,
      "loss": 3.7965,
      "step": 113770
    },
    {
      "epoch": 0.10606137050390686,
      "grad_norm": 0.961093008518219,
      "learning_rate": 0.00017878792963302683,
      "loss": 3.5303,
      "step": 113780
    },
    {
      "epoch": 0.10607069212198596,
      "grad_norm": 1.0496692657470703,
      "learning_rate": 0.00017878606531093166,
      "loss": 3.7932,
      "step": 113790
    },
    {
      "epoch": 0.10608001374006505,
      "grad_norm": 0.970527708530426,
      "learning_rate": 0.00017878420098883647,
      "loss": 3.7539,
      "step": 113800
    },
    {
      "epoch": 0.10608933535814415,
      "grad_norm": 2.9457263946533203,
      "learning_rate": 0.00017878233666674125,
      "loss": 3.7282,
      "step": 113810
    },
    {
      "epoch": 0.10609865697622323,
      "grad_norm": 0.8854132890701294,
      "learning_rate": 0.00017878047234464605,
      "loss": 3.873,
      "step": 113820
    },
    {
      "epoch": 0.10610797859430232,
      "grad_norm": 1.0028332471847534,
      "learning_rate": 0.00017877860802255083,
      "loss": 3.8121,
      "step": 113830
    },
    {
      "epoch": 0.10611730021238142,
      "grad_norm": 0.9226982593536377,
      "learning_rate": 0.00017877674370045564,
      "loss": 3.7966,
      "step": 113840
    },
    {
      "epoch": 0.10612662183046051,
      "grad_norm": 1.084414005279541,
      "learning_rate": 0.00017877487937836045,
      "loss": 3.5225,
      "step": 113850
    },
    {
      "epoch": 0.1061359434485396,
      "grad_norm": 1.1485264301300049,
      "learning_rate": 0.00017877301505626525,
      "loss": 3.8183,
      "step": 113860
    },
    {
      "epoch": 0.10614526506661869,
      "grad_norm": 1.1678028106689453,
      "learning_rate": 0.00017877115073417006,
      "loss": 3.7182,
      "step": 113870
    },
    {
      "epoch": 0.10615458668469778,
      "grad_norm": 1.033612847328186,
      "learning_rate": 0.00017876928641207484,
      "loss": 3.617,
      "step": 113880
    },
    {
      "epoch": 0.10616390830277687,
      "grad_norm": 1.2058887481689453,
      "learning_rate": 0.00017876742208997964,
      "loss": 3.8561,
      "step": 113890
    },
    {
      "epoch": 0.10617322992085597,
      "grad_norm": 1.0777723789215088,
      "learning_rate": 0.00017876555776788445,
      "loss": 3.5904,
      "step": 113900
    },
    {
      "epoch": 0.10618255153893506,
      "grad_norm": 1.2330684661865234,
      "learning_rate": 0.00017876369344578926,
      "loss": 3.6247,
      "step": 113910
    },
    {
      "epoch": 0.10619187315701414,
      "grad_norm": 1.1686674356460571,
      "learning_rate": 0.00017876182912369406,
      "loss": 3.638,
      "step": 113920
    },
    {
      "epoch": 0.10620119477509324,
      "grad_norm": 1.0176846981048584,
      "learning_rate": 0.00017875996480159884,
      "loss": 3.8533,
      "step": 113930
    },
    {
      "epoch": 0.10621051639317233,
      "grad_norm": 1.1243599653244019,
      "learning_rate": 0.00017875810047950365,
      "loss": 3.6013,
      "step": 113940
    },
    {
      "epoch": 0.10621983801125143,
      "grad_norm": 1.049626350402832,
      "learning_rate": 0.00017875623615740843,
      "loss": 3.6967,
      "step": 113950
    },
    {
      "epoch": 0.10622915962933052,
      "grad_norm": 0.9933354258537292,
      "learning_rate": 0.00017875437183531326,
      "loss": 3.7431,
      "step": 113960
    },
    {
      "epoch": 0.10623848124740962,
      "grad_norm": 1.025053858757019,
      "learning_rate": 0.00017875250751321807,
      "loss": 3.7074,
      "step": 113970
    },
    {
      "epoch": 0.1062478028654887,
      "grad_norm": 0.9707645773887634,
      "learning_rate": 0.00017875064319112285,
      "loss": 3.8723,
      "step": 113980
    },
    {
      "epoch": 0.10625712448356779,
      "grad_norm": 0.9749099612236023,
      "learning_rate": 0.00017874877886902765,
      "loss": 3.6969,
      "step": 113990
    },
    {
      "epoch": 0.10626644610164689,
      "grad_norm": 1.0188461542129517,
      "learning_rate": 0.00017874691454693243,
      "loss": 3.8482,
      "step": 114000
    },
    {
      "epoch": 0.10627576771972598,
      "grad_norm": 0.9671023488044739,
      "learning_rate": 0.00017874505022483724,
      "loss": 3.6424,
      "step": 114010
    },
    {
      "epoch": 0.10628508933780507,
      "grad_norm": 1.0795621871948242,
      "learning_rate": 0.00017874318590274207,
      "loss": 3.7158,
      "step": 114020
    },
    {
      "epoch": 0.10629441095588416,
      "grad_norm": 1.1826783418655396,
      "learning_rate": 0.00017874132158064685,
      "loss": 3.707,
      "step": 114030
    },
    {
      "epoch": 0.10630373257396325,
      "grad_norm": 1.067238688468933,
      "learning_rate": 0.00017873945725855166,
      "loss": 3.6122,
      "step": 114040
    },
    {
      "epoch": 0.10631305419204234,
      "grad_norm": 1.0664061307907104,
      "learning_rate": 0.00017873759293645644,
      "loss": 3.7102,
      "step": 114050
    },
    {
      "epoch": 0.10632237581012144,
      "grad_norm": 1.125530481338501,
      "learning_rate": 0.00017873572861436124,
      "loss": 3.782,
      "step": 114060
    },
    {
      "epoch": 0.10633169742820053,
      "grad_norm": 1.1172627210617065,
      "learning_rate": 0.00017873386429226605,
      "loss": 3.6888,
      "step": 114070
    },
    {
      "epoch": 0.10634101904627961,
      "grad_norm": 1.1107101440429688,
      "learning_rate": 0.00017873199997017086,
      "loss": 3.4493,
      "step": 114080
    },
    {
      "epoch": 0.10635034066435871,
      "grad_norm": 1.1764497756958008,
      "learning_rate": 0.00017873013564807566,
      "loss": 3.8347,
      "step": 114090
    },
    {
      "epoch": 0.1063596622824378,
      "grad_norm": 1.1371454000473022,
      "learning_rate": 0.00017872827132598044,
      "loss": 3.7344,
      "step": 114100
    },
    {
      "epoch": 0.1063689839005169,
      "grad_norm": 1.1362751722335815,
      "learning_rate": 0.00017872640700388525,
      "loss": 3.5379,
      "step": 114110
    },
    {
      "epoch": 0.10637830551859599,
      "grad_norm": 1.17912757396698,
      "learning_rate": 0.00017872454268179005,
      "loss": 3.602,
      "step": 114120
    },
    {
      "epoch": 0.10638762713667507,
      "grad_norm": 1.2843570709228516,
      "learning_rate": 0.00017872267835969486,
      "loss": 3.619,
      "step": 114130
    },
    {
      "epoch": 0.10639694875475417,
      "grad_norm": 0.9130699634552002,
      "learning_rate": 0.00017872081403759967,
      "loss": 3.7954,
      "step": 114140
    },
    {
      "epoch": 0.10640627037283326,
      "grad_norm": 1.3594754934310913,
      "learning_rate": 0.00017871894971550445,
      "loss": 3.7543,
      "step": 114150
    },
    {
      "epoch": 0.10641559199091236,
      "grad_norm": 1.1927688121795654,
      "learning_rate": 0.00017871708539340925,
      "loss": 3.6974,
      "step": 114160
    },
    {
      "epoch": 0.10642491360899145,
      "grad_norm": 1.115300178527832,
      "learning_rate": 0.00017871522107131406,
      "loss": 3.7409,
      "step": 114170
    },
    {
      "epoch": 0.10643423522707054,
      "grad_norm": 1.066194772720337,
      "learning_rate": 0.00017871335674921887,
      "loss": 3.7624,
      "step": 114180
    },
    {
      "epoch": 0.10644355684514963,
      "grad_norm": 1.0798203945159912,
      "learning_rate": 0.00017871149242712367,
      "loss": 3.7781,
      "step": 114190
    },
    {
      "epoch": 0.10645287846322872,
      "grad_norm": 0.885178804397583,
      "learning_rate": 0.00017870962810502845,
      "loss": 3.6106,
      "step": 114200
    },
    {
      "epoch": 0.10646220008130781,
      "grad_norm": 1.0376918315887451,
      "learning_rate": 0.00017870776378293326,
      "loss": 3.6854,
      "step": 114210
    },
    {
      "epoch": 0.10647152169938691,
      "grad_norm": 1.3033255338668823,
      "learning_rate": 0.00017870589946083806,
      "loss": 3.4809,
      "step": 114220
    },
    {
      "epoch": 0.106480843317466,
      "grad_norm": 1.043367862701416,
      "learning_rate": 0.00017870403513874284,
      "loss": 3.6943,
      "step": 114230
    },
    {
      "epoch": 0.10649016493554508,
      "grad_norm": 0.9509359002113342,
      "learning_rate": 0.00017870217081664768,
      "loss": 3.7799,
      "step": 114240
    },
    {
      "epoch": 0.10649948655362418,
      "grad_norm": 1.056605577468872,
      "learning_rate": 0.00017870030649455246,
      "loss": 3.627,
      "step": 114250
    },
    {
      "epoch": 0.10650880817170327,
      "grad_norm": 1.105416178703308,
      "learning_rate": 0.00017869844217245726,
      "loss": 3.6915,
      "step": 114260
    },
    {
      "epoch": 0.10651812978978237,
      "grad_norm": 1.099362850189209,
      "learning_rate": 0.00017869657785036207,
      "loss": 3.6172,
      "step": 114270
    },
    {
      "epoch": 0.10652745140786146,
      "grad_norm": 0.8841277956962585,
      "learning_rate": 0.00017869471352826685,
      "loss": 3.485,
      "step": 114280
    },
    {
      "epoch": 0.10653677302594054,
      "grad_norm": 1.1712085008621216,
      "learning_rate": 0.00017869284920617165,
      "loss": 3.6369,
      "step": 114290
    },
    {
      "epoch": 0.10654609464401964,
      "grad_norm": 1.0176153182983398,
      "learning_rate": 0.00017869098488407646,
      "loss": 3.6387,
      "step": 114300
    },
    {
      "epoch": 0.10655541626209873,
      "grad_norm": 1.210843801498413,
      "learning_rate": 0.00017868912056198127,
      "loss": 3.6671,
      "step": 114310
    },
    {
      "epoch": 0.10656473788017783,
      "grad_norm": 1.25995671749115,
      "learning_rate": 0.00017868725623988607,
      "loss": 3.64,
      "step": 114320
    },
    {
      "epoch": 0.10657405949825692,
      "grad_norm": 1.0050532817840576,
      "learning_rate": 0.00017868539191779085,
      "loss": 3.5306,
      "step": 114330
    },
    {
      "epoch": 0.106583381116336,
      "grad_norm": 0.778863251209259,
      "learning_rate": 0.00017868352759569566,
      "loss": 3.6089,
      "step": 114340
    },
    {
      "epoch": 0.1065927027344151,
      "grad_norm": 0.8922909498214722,
      "learning_rate": 0.00017868166327360046,
      "loss": 3.5329,
      "step": 114350
    },
    {
      "epoch": 0.10660202435249419,
      "grad_norm": 1.0555293560028076,
      "learning_rate": 0.00017867979895150527,
      "loss": 3.7319,
      "step": 114360
    },
    {
      "epoch": 0.10661134597057328,
      "grad_norm": 1.1677837371826172,
      "learning_rate": 0.00017867793462941008,
      "loss": 3.5934,
      "step": 114370
    },
    {
      "epoch": 0.10662066758865238,
      "grad_norm": 1.1300266981124878,
      "learning_rate": 0.00017867607030731486,
      "loss": 3.7557,
      "step": 114380
    },
    {
      "epoch": 0.10662998920673146,
      "grad_norm": 0.9901083111763,
      "learning_rate": 0.00017867420598521966,
      "loss": 3.6996,
      "step": 114390
    },
    {
      "epoch": 0.10663931082481055,
      "grad_norm": 1.1686803102493286,
      "learning_rate": 0.00017867234166312444,
      "loss": 3.7043,
      "step": 114400
    },
    {
      "epoch": 0.10664863244288965,
      "grad_norm": 1.0497260093688965,
      "learning_rate": 0.00017867047734102927,
      "loss": 3.6752,
      "step": 114410
    },
    {
      "epoch": 0.10665795406096874,
      "grad_norm": 1.1702455282211304,
      "learning_rate": 0.00017866861301893408,
      "loss": 3.632,
      "step": 114420
    },
    {
      "epoch": 0.10666727567904784,
      "grad_norm": 1.2526271343231201,
      "learning_rate": 0.00017866674869683886,
      "loss": 3.8236,
      "step": 114430
    },
    {
      "epoch": 0.10667659729712693,
      "grad_norm": 0.8470252156257629,
      "learning_rate": 0.00017866488437474367,
      "loss": 3.8731,
      "step": 114440
    },
    {
      "epoch": 0.10668591891520601,
      "grad_norm": 1.006264090538025,
      "learning_rate": 0.00017866302005264845,
      "loss": 3.8219,
      "step": 114450
    },
    {
      "epoch": 0.10669524053328511,
      "grad_norm": 1.0884661674499512,
      "learning_rate": 0.00017866115573055325,
      "loss": 3.5855,
      "step": 114460
    },
    {
      "epoch": 0.1067045621513642,
      "grad_norm": 1.1272578239440918,
      "learning_rate": 0.00017865929140845809,
      "loss": 3.5272,
      "step": 114470
    },
    {
      "epoch": 0.1067138837694433,
      "grad_norm": 1.0771654844284058,
      "learning_rate": 0.00017865742708636286,
      "loss": 3.8326,
      "step": 114480
    },
    {
      "epoch": 0.10672320538752239,
      "grad_norm": 1.087637186050415,
      "learning_rate": 0.00017865556276426767,
      "loss": 3.638,
      "step": 114490
    },
    {
      "epoch": 0.10673252700560147,
      "grad_norm": 1.2215272188186646,
      "learning_rate": 0.00017865369844217245,
      "loss": 3.7438,
      "step": 114500
    },
    {
      "epoch": 0.10674184862368057,
      "grad_norm": 0.9873678684234619,
      "learning_rate": 0.00017865183412007726,
      "loss": 3.809,
      "step": 114510
    },
    {
      "epoch": 0.10675117024175966,
      "grad_norm": 1.0192325115203857,
      "learning_rate": 0.00017864996979798206,
      "loss": 3.7376,
      "step": 114520
    },
    {
      "epoch": 0.10676049185983875,
      "grad_norm": 1.108919620513916,
      "learning_rate": 0.00017864810547588687,
      "loss": 3.7687,
      "step": 114530
    },
    {
      "epoch": 0.10676981347791785,
      "grad_norm": 0.9385892152786255,
      "learning_rate": 0.00017864624115379168,
      "loss": 3.8691,
      "step": 114540
    },
    {
      "epoch": 0.10677913509599693,
      "grad_norm": 0.9651651978492737,
      "learning_rate": 0.00017864437683169645,
      "loss": 3.71,
      "step": 114550
    },
    {
      "epoch": 0.10678845671407602,
      "grad_norm": 1.470248818397522,
      "learning_rate": 0.00017864251250960126,
      "loss": 3.7171,
      "step": 114560
    },
    {
      "epoch": 0.10679777833215512,
      "grad_norm": 1.3397042751312256,
      "learning_rate": 0.00017864064818750607,
      "loss": 3.4605,
      "step": 114570
    },
    {
      "epoch": 0.10680709995023421,
      "grad_norm": 1.1025155782699585,
      "learning_rate": 0.00017863878386541087,
      "loss": 3.799,
      "step": 114580
    },
    {
      "epoch": 0.10681642156831331,
      "grad_norm": 1.1336617469787598,
      "learning_rate": 0.00017863691954331568,
      "loss": 3.9044,
      "step": 114590
    },
    {
      "epoch": 0.10682574318639239,
      "grad_norm": 1.1427255868911743,
      "learning_rate": 0.00017863505522122046,
      "loss": 3.7653,
      "step": 114600
    },
    {
      "epoch": 0.10683506480447148,
      "grad_norm": 0.8711927533149719,
      "learning_rate": 0.00017863319089912527,
      "loss": 3.754,
      "step": 114610
    },
    {
      "epoch": 0.10684438642255058,
      "grad_norm": 1.2591594457626343,
      "learning_rate": 0.00017863132657703007,
      "loss": 3.768,
      "step": 114620
    },
    {
      "epoch": 0.10685370804062967,
      "grad_norm": 1.06172513961792,
      "learning_rate": 0.00017862946225493488,
      "loss": 3.4325,
      "step": 114630
    },
    {
      "epoch": 0.10686302965870877,
      "grad_norm": 1.1197774410247803,
      "learning_rate": 0.00017862759793283968,
      "loss": 3.5047,
      "step": 114640
    },
    {
      "epoch": 0.10687235127678786,
      "grad_norm": 1.012810468673706,
      "learning_rate": 0.00017862573361074446,
      "loss": 3.6608,
      "step": 114650
    },
    {
      "epoch": 0.10688167289486694,
      "grad_norm": 0.9754148721694946,
      "learning_rate": 0.00017862386928864927,
      "loss": 3.4005,
      "step": 114660
    },
    {
      "epoch": 0.10689099451294604,
      "grad_norm": 1.0916168689727783,
      "learning_rate": 0.00017862200496655408,
      "loss": 3.6424,
      "step": 114670
    },
    {
      "epoch": 0.10690031613102513,
      "grad_norm": 1.0428450107574463,
      "learning_rate": 0.00017862014064445886,
      "loss": 3.8591,
      "step": 114680
    },
    {
      "epoch": 0.10690963774910422,
      "grad_norm": 0.9769614338874817,
      "learning_rate": 0.0001786182763223637,
      "loss": 3.6696,
      "step": 114690
    },
    {
      "epoch": 0.10691895936718332,
      "grad_norm": 0.9890910387039185,
      "learning_rate": 0.00017861641200026847,
      "loss": 3.8586,
      "step": 114700
    },
    {
      "epoch": 0.1069282809852624,
      "grad_norm": 1.0749907493591309,
      "learning_rate": 0.00017861454767817327,
      "loss": 3.8025,
      "step": 114710
    },
    {
      "epoch": 0.1069376026033415,
      "grad_norm": 0.9776833653450012,
      "learning_rate": 0.00017861268335607808,
      "loss": 3.7281,
      "step": 114720
    },
    {
      "epoch": 0.10694692422142059,
      "grad_norm": 1.1760544776916504,
      "learning_rate": 0.00017861081903398286,
      "loss": 3.5798,
      "step": 114730
    },
    {
      "epoch": 0.10695624583949968,
      "grad_norm": 1.0137736797332764,
      "learning_rate": 0.00017860895471188767,
      "loss": 3.7548,
      "step": 114740
    },
    {
      "epoch": 0.10696556745757878,
      "grad_norm": 1.0421139001846313,
      "learning_rate": 0.00017860709038979247,
      "loss": 3.6185,
      "step": 114750
    },
    {
      "epoch": 0.10697488907565786,
      "grad_norm": 1.1150634288787842,
      "learning_rate": 0.00017860522606769728,
      "loss": 3.6883,
      "step": 114760
    },
    {
      "epoch": 0.10698421069373695,
      "grad_norm": 1.1723719835281372,
      "learning_rate": 0.00017860336174560209,
      "loss": 3.7577,
      "step": 114770
    },
    {
      "epoch": 0.10699353231181605,
      "grad_norm": 0.9378727674484253,
      "learning_rate": 0.00017860149742350686,
      "loss": 3.7374,
      "step": 114780
    },
    {
      "epoch": 0.10700285392989514,
      "grad_norm": 0.9144070148468018,
      "learning_rate": 0.00017859963310141167,
      "loss": 3.7252,
      "step": 114790
    },
    {
      "epoch": 0.10701217554797424,
      "grad_norm": 1.0647311210632324,
      "learning_rate": 0.00017859776877931648,
      "loss": 3.6839,
      "step": 114800
    },
    {
      "epoch": 0.10702149716605332,
      "grad_norm": 0.9588229656219482,
      "learning_rate": 0.00017859590445722128,
      "loss": 3.833,
      "step": 114810
    },
    {
      "epoch": 0.10703081878413241,
      "grad_norm": 1.0204685926437378,
      "learning_rate": 0.0001785940401351261,
      "loss": 3.5906,
      "step": 114820
    },
    {
      "epoch": 0.1070401404022115,
      "grad_norm": 1.0171087980270386,
      "learning_rate": 0.00017859217581303087,
      "loss": 3.6389,
      "step": 114830
    },
    {
      "epoch": 0.1070494620202906,
      "grad_norm": 1.2436795234680176,
      "learning_rate": 0.00017859031149093567,
      "loss": 3.6049,
      "step": 114840
    },
    {
      "epoch": 0.1070587836383697,
      "grad_norm": 1.001094937324524,
      "learning_rate": 0.00017858844716884045,
      "loss": 3.7767,
      "step": 114850
    },
    {
      "epoch": 0.10706810525644879,
      "grad_norm": 1.2960389852523804,
      "learning_rate": 0.0001785865828467453,
      "loss": 3.6868,
      "step": 114860
    },
    {
      "epoch": 0.10707742687452787,
      "grad_norm": 0.7870774269104004,
      "learning_rate": 0.0001785847185246501,
      "loss": 3.6968,
      "step": 114870
    },
    {
      "epoch": 0.10708674849260696,
      "grad_norm": 0.9989450573921204,
      "learning_rate": 0.00017858285420255487,
      "loss": 3.5534,
      "step": 114880
    },
    {
      "epoch": 0.10709607011068606,
      "grad_norm": 1.126551866531372,
      "learning_rate": 0.00017858098988045968,
      "loss": 3.8128,
      "step": 114890
    },
    {
      "epoch": 0.10710539172876515,
      "grad_norm": 0.9766791462898254,
      "learning_rate": 0.00017857912555836446,
      "loss": 3.6154,
      "step": 114900
    },
    {
      "epoch": 0.10711471334684425,
      "grad_norm": 1.478435754776001,
      "learning_rate": 0.00017857726123626926,
      "loss": 3.7045,
      "step": 114910
    },
    {
      "epoch": 0.10712403496492333,
      "grad_norm": 1.0397536754608154,
      "learning_rate": 0.0001785753969141741,
      "loss": 3.6615,
      "step": 114920
    },
    {
      "epoch": 0.10713335658300242,
      "grad_norm": 1.0147491693496704,
      "learning_rate": 0.00017857353259207888,
      "loss": 3.56,
      "step": 114930
    },
    {
      "epoch": 0.10714267820108152,
      "grad_norm": 1.256116509437561,
      "learning_rate": 0.00017857166826998368,
      "loss": 3.6098,
      "step": 114940
    },
    {
      "epoch": 0.10715199981916061,
      "grad_norm": 1.0718432664871216,
      "learning_rate": 0.00017856980394788846,
      "loss": 3.6886,
      "step": 114950
    },
    {
      "epoch": 0.1071613214372397,
      "grad_norm": 1.157861351966858,
      "learning_rate": 0.00017856793962579327,
      "loss": 3.6285,
      "step": 114960
    },
    {
      "epoch": 0.10717064305531879,
      "grad_norm": 0.9907975196838379,
      "learning_rate": 0.00017856607530369808,
      "loss": 3.6746,
      "step": 114970
    },
    {
      "epoch": 0.10717996467339788,
      "grad_norm": 1.118088722229004,
      "learning_rate": 0.00017856421098160288,
      "loss": 3.719,
      "step": 114980
    },
    {
      "epoch": 0.10718928629147698,
      "grad_norm": 1.1757739782333374,
      "learning_rate": 0.0001785623466595077,
      "loss": 3.6117,
      "step": 114990
    },
    {
      "epoch": 0.10719860790955607,
      "grad_norm": 1.1650646924972534,
      "learning_rate": 0.00017856048233741247,
      "loss": 3.5718,
      "step": 115000
    },
    {
      "epoch": 0.10720792952763517,
      "grad_norm": 0.9886050224304199,
      "learning_rate": 0.00017855861801531727,
      "loss": 3.5983,
      "step": 115010
    },
    {
      "epoch": 0.10721725114571425,
      "grad_norm": 1.0293090343475342,
      "learning_rate": 0.00017855675369322208,
      "loss": 3.9304,
      "step": 115020
    },
    {
      "epoch": 0.10722657276379334,
      "grad_norm": 1.250437617301941,
      "learning_rate": 0.00017855488937112689,
      "loss": 3.5928,
      "step": 115030
    },
    {
      "epoch": 0.10723589438187243,
      "grad_norm": 0.9346259236335754,
      "learning_rate": 0.0001785530250490317,
      "loss": 3.7047,
      "step": 115040
    },
    {
      "epoch": 0.10724521599995153,
      "grad_norm": 1.0318948030471802,
      "learning_rate": 0.00017855116072693647,
      "loss": 3.7514,
      "step": 115050
    },
    {
      "epoch": 0.10725453761803062,
      "grad_norm": 1.008011817932129,
      "learning_rate": 0.00017854929640484128,
      "loss": 3.8336,
      "step": 115060
    },
    {
      "epoch": 0.10726385923610972,
      "grad_norm": 1.1728013753890991,
      "learning_rate": 0.00017854743208274608,
      "loss": 3.6145,
      "step": 115070
    },
    {
      "epoch": 0.1072731808541888,
      "grad_norm": 1.1251529455184937,
      "learning_rate": 0.0001785455677606509,
      "loss": 3.7739,
      "step": 115080
    },
    {
      "epoch": 0.1072825024722679,
      "grad_norm": 1.157936453819275,
      "learning_rate": 0.0001785437034385557,
      "loss": 3.6014,
      "step": 115090
    },
    {
      "epoch": 0.10729182409034699,
      "grad_norm": 1.1058326959609985,
      "learning_rate": 0.00017854183911646048,
      "loss": 3.8492,
      "step": 115100
    },
    {
      "epoch": 0.10730114570842608,
      "grad_norm": 1.0055183172225952,
      "learning_rate": 0.00017853997479436528,
      "loss": 3.6951,
      "step": 115110
    },
    {
      "epoch": 0.10731046732650518,
      "grad_norm": 1.101608395576477,
      "learning_rate": 0.0001785381104722701,
      "loss": 3.773,
      "step": 115120
    },
    {
      "epoch": 0.10731978894458426,
      "grad_norm": 1.124337077140808,
      "learning_rate": 0.00017853624615017487,
      "loss": 3.8481,
      "step": 115130
    },
    {
      "epoch": 0.10732911056266335,
      "grad_norm": 1.105263352394104,
      "learning_rate": 0.0001785343818280797,
      "loss": 3.7325,
      "step": 115140
    },
    {
      "epoch": 0.10733843218074245,
      "grad_norm": 1.0021780729293823,
      "learning_rate": 0.00017853251750598448,
      "loss": 3.6678,
      "step": 115150
    },
    {
      "epoch": 0.10734775379882154,
      "grad_norm": 1.0806450843811035,
      "learning_rate": 0.0001785306531838893,
      "loss": 3.7331,
      "step": 115160
    },
    {
      "epoch": 0.10735707541690064,
      "grad_norm": 1.0398703813552856,
      "learning_rate": 0.0001785287888617941,
      "loss": 3.8022,
      "step": 115170
    },
    {
      "epoch": 0.10736639703497972,
      "grad_norm": 1.0796440839767456,
      "learning_rate": 0.00017852692453969887,
      "loss": 3.8654,
      "step": 115180
    },
    {
      "epoch": 0.10737571865305881,
      "grad_norm": 1.3319242000579834,
      "learning_rate": 0.00017852506021760368,
      "loss": 3.686,
      "step": 115190
    },
    {
      "epoch": 0.1073850402711379,
      "grad_norm": 1.035549521446228,
      "learning_rate": 0.00017852319589550849,
      "loss": 3.647,
      "step": 115200
    },
    {
      "epoch": 0.107394361889217,
      "grad_norm": 0.9877256751060486,
      "learning_rate": 0.0001785213315734133,
      "loss": 3.7398,
      "step": 115210
    },
    {
      "epoch": 0.1074036835072961,
      "grad_norm": 1.0365334749221802,
      "learning_rate": 0.00017851946725131807,
      "loss": 3.7078,
      "step": 115220
    },
    {
      "epoch": 0.10741300512537517,
      "grad_norm": 0.8555596470832825,
      "learning_rate": 0.00017851760292922288,
      "loss": 3.7785,
      "step": 115230
    },
    {
      "epoch": 0.10742232674345427,
      "grad_norm": 0.9676240682601929,
      "learning_rate": 0.00017851573860712768,
      "loss": 3.6182,
      "step": 115240
    },
    {
      "epoch": 0.10743164836153336,
      "grad_norm": 1.0563018321990967,
      "learning_rate": 0.0001785138742850325,
      "loss": 3.7358,
      "step": 115250
    },
    {
      "epoch": 0.10744096997961246,
      "grad_norm": 1.0558961629867554,
      "learning_rate": 0.0001785120099629373,
      "loss": 3.7728,
      "step": 115260
    },
    {
      "epoch": 0.10745029159769155,
      "grad_norm": 1.188346266746521,
      "learning_rate": 0.00017851014564084208,
      "loss": 3.7993,
      "step": 115270
    },
    {
      "epoch": 0.10745961321577063,
      "grad_norm": 1.0929150581359863,
      "learning_rate": 0.00017850828131874688,
      "loss": 3.6868,
      "step": 115280
    },
    {
      "epoch": 0.10746893483384973,
      "grad_norm": 0.9167892336845398,
      "learning_rate": 0.0001785064169966517,
      "loss": 3.7643,
      "step": 115290
    },
    {
      "epoch": 0.10747825645192882,
      "grad_norm": 1.0958518981933594,
      "learning_rate": 0.00017850455267455647,
      "loss": 3.8513,
      "step": 115300
    },
    {
      "epoch": 0.10748757807000792,
      "grad_norm": 0.8949140310287476,
      "learning_rate": 0.0001785026883524613,
      "loss": 3.8384,
      "step": 115310
    },
    {
      "epoch": 0.10749689968808701,
      "grad_norm": 1.039534568786621,
      "learning_rate": 0.00017850082403036608,
      "loss": 3.8025,
      "step": 115320
    },
    {
      "epoch": 0.1075062213061661,
      "grad_norm": 1.1188353300094604,
      "learning_rate": 0.00017849895970827089,
      "loss": 3.8167,
      "step": 115330
    },
    {
      "epoch": 0.10751554292424519,
      "grad_norm": 1.0722097158432007,
      "learning_rate": 0.0001784970953861757,
      "loss": 3.637,
      "step": 115340
    },
    {
      "epoch": 0.10752486454232428,
      "grad_norm": 1.1759768724441528,
      "learning_rate": 0.00017849523106408047,
      "loss": 3.6582,
      "step": 115350
    },
    {
      "epoch": 0.10753418616040337,
      "grad_norm": 0.8603085875511169,
      "learning_rate": 0.00017849336674198528,
      "loss": 3.6482,
      "step": 115360
    },
    {
      "epoch": 0.10754350777848247,
      "grad_norm": 1.0282224416732788,
      "learning_rate": 0.00017849150241989008,
      "loss": 3.5465,
      "step": 115370
    },
    {
      "epoch": 0.10755282939656156,
      "grad_norm": 1.2423778772354126,
      "learning_rate": 0.0001784896380977949,
      "loss": 3.7893,
      "step": 115380
    },
    {
      "epoch": 0.10756215101464064,
      "grad_norm": 1.091650366783142,
      "learning_rate": 0.0001784877737756997,
      "loss": 3.6678,
      "step": 115390
    },
    {
      "epoch": 0.10757147263271974,
      "grad_norm": 0.9612494111061096,
      "learning_rate": 0.00017848590945360448,
      "loss": 3.5835,
      "step": 115400
    },
    {
      "epoch": 0.10758079425079883,
      "grad_norm": 0.9275593757629395,
      "learning_rate": 0.00017848404513150928,
      "loss": 3.6982,
      "step": 115410
    },
    {
      "epoch": 0.10759011586887793,
      "grad_norm": 0.9763830900192261,
      "learning_rate": 0.0001784821808094141,
      "loss": 3.4169,
      "step": 115420
    },
    {
      "epoch": 0.10759943748695702,
      "grad_norm": 0.9206134676933289,
      "learning_rate": 0.0001784803164873189,
      "loss": 3.7899,
      "step": 115430
    },
    {
      "epoch": 0.1076087591050361,
      "grad_norm": 1.0275009870529175,
      "learning_rate": 0.0001784784521652237,
      "loss": 3.7802,
      "step": 115440
    },
    {
      "epoch": 0.1076180807231152,
      "grad_norm": 1.175580382347107,
      "learning_rate": 0.00017847658784312848,
      "loss": 3.8503,
      "step": 115450
    },
    {
      "epoch": 0.10762740234119429,
      "grad_norm": 1.0422554016113281,
      "learning_rate": 0.0001784747235210333,
      "loss": 3.5651,
      "step": 115460
    },
    {
      "epoch": 0.10763672395927339,
      "grad_norm": 0.9579250812530518,
      "learning_rate": 0.0001784728591989381,
      "loss": 3.7574,
      "step": 115470
    },
    {
      "epoch": 0.10764604557735248,
      "grad_norm": 1.0866994857788086,
      "learning_rate": 0.0001784709948768429,
      "loss": 3.6818,
      "step": 115480
    },
    {
      "epoch": 0.10765536719543156,
      "grad_norm": 1.0339000225067139,
      "learning_rate": 0.0001784691305547477,
      "loss": 3.5785,
      "step": 115490
    },
    {
      "epoch": 0.10766468881351066,
      "grad_norm": 0.8387581706047058,
      "learning_rate": 0.00017846726623265248,
      "loss": 3.6872,
      "step": 115500
    },
    {
      "epoch": 0.10767401043158975,
      "grad_norm": 0.9606189727783203,
      "learning_rate": 0.0001784654019105573,
      "loss": 3.4523,
      "step": 115510
    },
    {
      "epoch": 0.10768333204966885,
      "grad_norm": 1.0227233171463013,
      "learning_rate": 0.00017846353758846207,
      "loss": 3.6353,
      "step": 115520
    },
    {
      "epoch": 0.10769265366774794,
      "grad_norm": 1.0502146482467651,
      "learning_rate": 0.0001784616732663669,
      "loss": 3.6904,
      "step": 115530
    },
    {
      "epoch": 0.10770197528582703,
      "grad_norm": 1.0059000253677368,
      "learning_rate": 0.0001784598089442717,
      "loss": 3.7062,
      "step": 115540
    },
    {
      "epoch": 0.10771129690390611,
      "grad_norm": 1.0206087827682495,
      "learning_rate": 0.0001784579446221765,
      "loss": 3.8346,
      "step": 115550
    },
    {
      "epoch": 0.10772061852198521,
      "grad_norm": 1.0399364233016968,
      "learning_rate": 0.0001784560803000813,
      "loss": 3.6916,
      "step": 115560
    },
    {
      "epoch": 0.1077299401400643,
      "grad_norm": 0.9719926118850708,
      "learning_rate": 0.00017845421597798607,
      "loss": 3.8114,
      "step": 115570
    },
    {
      "epoch": 0.1077392617581434,
      "grad_norm": 0.8531675934791565,
      "learning_rate": 0.00017845235165589088,
      "loss": 3.7112,
      "step": 115580
    },
    {
      "epoch": 0.10774858337622249,
      "grad_norm": 0.9894998073577881,
      "learning_rate": 0.00017845048733379571,
      "loss": 3.6748,
      "step": 115590
    },
    {
      "epoch": 0.10775790499430157,
      "grad_norm": 0.8378013968467712,
      "learning_rate": 0.0001784486230117005,
      "loss": 3.8839,
      "step": 115600
    },
    {
      "epoch": 0.10776722661238067,
      "grad_norm": 1.0857831239700317,
      "learning_rate": 0.0001784467586896053,
      "loss": 3.6314,
      "step": 115610
    },
    {
      "epoch": 0.10777654823045976,
      "grad_norm": 0.9453648328781128,
      "learning_rate": 0.00017844489436751008,
      "loss": 3.6364,
      "step": 115620
    },
    {
      "epoch": 0.10778586984853886,
      "grad_norm": 0.9277169108390808,
      "learning_rate": 0.00017844303004541489,
      "loss": 3.5344,
      "step": 115630
    },
    {
      "epoch": 0.10779519146661795,
      "grad_norm": 0.9496978521347046,
      "learning_rate": 0.0001784411657233197,
      "loss": 3.7548,
      "step": 115640
    },
    {
      "epoch": 0.10780451308469703,
      "grad_norm": 0.8361845016479492,
      "learning_rate": 0.0001784393014012245,
      "loss": 3.6948,
      "step": 115650
    },
    {
      "epoch": 0.10781383470277613,
      "grad_norm": 1.1792360544204712,
      "learning_rate": 0.0001784374370791293,
      "loss": 3.637,
      "step": 115660
    },
    {
      "epoch": 0.10782315632085522,
      "grad_norm": 1.2044448852539062,
      "learning_rate": 0.00017843557275703408,
      "loss": 3.5664,
      "step": 115670
    },
    {
      "epoch": 0.10783247793893432,
      "grad_norm": 0.9540430903434753,
      "learning_rate": 0.0001784337084349389,
      "loss": 3.7146,
      "step": 115680
    },
    {
      "epoch": 0.10784179955701341,
      "grad_norm": 1.205952763557434,
      "learning_rate": 0.0001784318441128437,
      "loss": 3.6317,
      "step": 115690
    },
    {
      "epoch": 0.10785112117509249,
      "grad_norm": 0.9994767308235168,
      "learning_rate": 0.0001784299797907485,
      "loss": 3.4751,
      "step": 115700
    },
    {
      "epoch": 0.10786044279317158,
      "grad_norm": 1.153882622718811,
      "learning_rate": 0.0001784281154686533,
      "loss": 3.6511,
      "step": 115710
    },
    {
      "epoch": 0.10786976441125068,
      "grad_norm": 0.9039445519447327,
      "learning_rate": 0.0001784262511465581,
      "loss": 3.8078,
      "step": 115720
    },
    {
      "epoch": 0.10787908602932977,
      "grad_norm": 1.1494992971420288,
      "learning_rate": 0.0001784243868244629,
      "loss": 3.5762,
      "step": 115730
    },
    {
      "epoch": 0.10788840764740887,
      "grad_norm": 1.075609803199768,
      "learning_rate": 0.0001784225225023677,
      "loss": 3.5896,
      "step": 115740
    },
    {
      "epoch": 0.10789772926548796,
      "grad_norm": 1.199655532836914,
      "learning_rate": 0.00017842065818027248,
      "loss": 3.6116,
      "step": 115750
    },
    {
      "epoch": 0.10790705088356704,
      "grad_norm": 1.1456249952316284,
      "learning_rate": 0.0001784187938581773,
      "loss": 3.6735,
      "step": 115760
    },
    {
      "epoch": 0.10791637250164614,
      "grad_norm": 1.2356852293014526,
      "learning_rate": 0.0001784169295360821,
      "loss": 3.5452,
      "step": 115770
    },
    {
      "epoch": 0.10792569411972523,
      "grad_norm": 1.0569677352905273,
      "learning_rate": 0.0001784150652139869,
      "loss": 3.7856,
      "step": 115780
    },
    {
      "epoch": 0.10793501573780433,
      "grad_norm": 0.9397255778312683,
      "learning_rate": 0.0001784132008918917,
      "loss": 3.8527,
      "step": 115790
    },
    {
      "epoch": 0.10794433735588342,
      "grad_norm": 1.0703566074371338,
      "learning_rate": 0.00017841133656979648,
      "loss": 3.5077,
      "step": 115800
    },
    {
      "epoch": 0.1079536589739625,
      "grad_norm": 1.2175883054733276,
      "learning_rate": 0.0001784094722477013,
      "loss": 3.7912,
      "step": 115810
    },
    {
      "epoch": 0.1079629805920416,
      "grad_norm": 0.9768242239952087,
      "learning_rate": 0.0001784076079256061,
      "loss": 3.6218,
      "step": 115820
    },
    {
      "epoch": 0.10797230221012069,
      "grad_norm": 0.9995570182800293,
      "learning_rate": 0.0001784057436035109,
      "loss": 3.6053,
      "step": 115830
    },
    {
      "epoch": 0.10798162382819979,
      "grad_norm": 1.0751845836639404,
      "learning_rate": 0.0001784038792814157,
      "loss": 3.6151,
      "step": 115840
    },
    {
      "epoch": 0.10799094544627888,
      "grad_norm": 1.122031569480896,
      "learning_rate": 0.0001784020149593205,
      "loss": 3.5948,
      "step": 115850
    },
    {
      "epoch": 0.10800026706435796,
      "grad_norm": 1.2243244647979736,
      "learning_rate": 0.0001784001506372253,
      "loss": 3.5209,
      "step": 115860
    },
    {
      "epoch": 0.10800958868243705,
      "grad_norm": 1.057440996170044,
      "learning_rate": 0.0001783982863151301,
      "loss": 3.635,
      "step": 115870
    },
    {
      "epoch": 0.10801891030051615,
      "grad_norm": 1.1528748273849487,
      "learning_rate": 0.0001783964219930349,
      "loss": 3.7626,
      "step": 115880
    },
    {
      "epoch": 0.10802823191859524,
      "grad_norm": 1.0804557800292969,
      "learning_rate": 0.00017839455767093971,
      "loss": 3.7695,
      "step": 115890
    },
    {
      "epoch": 0.10803755353667434,
      "grad_norm": 1.0776697397232056,
      "learning_rate": 0.0001783926933488445,
      "loss": 3.6465,
      "step": 115900
    },
    {
      "epoch": 0.10804687515475342,
      "grad_norm": 1.2721128463745117,
      "learning_rate": 0.0001783908290267493,
      "loss": 3.6434,
      "step": 115910
    },
    {
      "epoch": 0.10805619677283251,
      "grad_norm": 1.039766788482666,
      "learning_rate": 0.0001783889647046541,
      "loss": 3.6453,
      "step": 115920
    },
    {
      "epoch": 0.10806551839091161,
      "grad_norm": 1.0895881652832031,
      "learning_rate": 0.0001783871003825589,
      "loss": 3.8792,
      "step": 115930
    },
    {
      "epoch": 0.1080748400089907,
      "grad_norm": 0.9111899137496948,
      "learning_rate": 0.00017838523606046372,
      "loss": 3.6535,
      "step": 115940
    },
    {
      "epoch": 0.1080841616270698,
      "grad_norm": 1.4523063898086548,
      "learning_rate": 0.0001783833717383685,
      "loss": 3.8866,
      "step": 115950
    },
    {
      "epoch": 0.10809348324514888,
      "grad_norm": 1.1633740663528442,
      "learning_rate": 0.0001783815074162733,
      "loss": 3.6421,
      "step": 115960
    },
    {
      "epoch": 0.10810280486322797,
      "grad_norm": 1.121166467666626,
      "learning_rate": 0.00017837964309417808,
      "loss": 3.6631,
      "step": 115970
    },
    {
      "epoch": 0.10811212648130707,
      "grad_norm": 0.8989716172218323,
      "learning_rate": 0.00017837777877208292,
      "loss": 3.7316,
      "step": 115980
    },
    {
      "epoch": 0.10812144809938616,
      "grad_norm": 1.2166473865509033,
      "learning_rate": 0.00017837591444998772,
      "loss": 3.8089,
      "step": 115990
    },
    {
      "epoch": 0.10813076971746526,
      "grad_norm": 1.106261968612671,
      "learning_rate": 0.0001783740501278925,
      "loss": 3.8036,
      "step": 116000
    },
    {
      "epoch": 0.10814009133554435,
      "grad_norm": 0.9797458648681641,
      "learning_rate": 0.0001783721858057973,
      "loss": 3.7582,
      "step": 116010
    },
    {
      "epoch": 0.10814941295362343,
      "grad_norm": 1.0566046237945557,
      "learning_rate": 0.0001783703214837021,
      "loss": 3.7491,
      "step": 116020
    },
    {
      "epoch": 0.10815873457170253,
      "grad_norm": 0.9760665893554688,
      "learning_rate": 0.0001783684571616069,
      "loss": 3.6315,
      "step": 116030
    },
    {
      "epoch": 0.10816805618978162,
      "grad_norm": 1.1391793489456177,
      "learning_rate": 0.00017836659283951173,
      "loss": 3.6679,
      "step": 116040
    },
    {
      "epoch": 0.10817737780786071,
      "grad_norm": 0.9534624218940735,
      "learning_rate": 0.0001783647285174165,
      "loss": 3.5401,
      "step": 116050
    },
    {
      "epoch": 0.10818669942593981,
      "grad_norm": 1.073684573173523,
      "learning_rate": 0.0001783628641953213,
      "loss": 3.6058,
      "step": 116060
    },
    {
      "epoch": 0.10819602104401889,
      "grad_norm": 1.1407967805862427,
      "learning_rate": 0.0001783609998732261,
      "loss": 3.6033,
      "step": 116070
    },
    {
      "epoch": 0.10820534266209798,
      "grad_norm": 1.1662884950637817,
      "learning_rate": 0.0001783591355511309,
      "loss": 3.6061,
      "step": 116080
    },
    {
      "epoch": 0.10821466428017708,
      "grad_norm": 1.173404335975647,
      "learning_rate": 0.0001783572712290357,
      "loss": 3.61,
      "step": 116090
    },
    {
      "epoch": 0.10822398589825617,
      "grad_norm": 0.957933783531189,
      "learning_rate": 0.0001783554069069405,
      "loss": 3.738,
      "step": 116100
    },
    {
      "epoch": 0.10823330751633527,
      "grad_norm": 1.208681344985962,
      "learning_rate": 0.00017835354258484532,
      "loss": 3.7779,
      "step": 116110
    },
    {
      "epoch": 0.10824262913441435,
      "grad_norm": 1.0298309326171875,
      "learning_rate": 0.0001783516782627501,
      "loss": 3.5759,
      "step": 116120
    },
    {
      "epoch": 0.10825195075249344,
      "grad_norm": 1.1197680234909058,
      "learning_rate": 0.0001783498139406549,
      "loss": 3.774,
      "step": 116130
    },
    {
      "epoch": 0.10826127237057254,
      "grad_norm": 1.2320141792297363,
      "learning_rate": 0.0001783479496185597,
      "loss": 3.6128,
      "step": 116140
    },
    {
      "epoch": 0.10827059398865163,
      "grad_norm": 0.9570108652114868,
      "learning_rate": 0.00017834608529646452,
      "loss": 3.7157,
      "step": 116150
    },
    {
      "epoch": 0.10827991560673073,
      "grad_norm": 1.0309315919876099,
      "learning_rate": 0.00017834422097436932,
      "loss": 3.549,
      "step": 116160
    },
    {
      "epoch": 0.1082892372248098,
      "grad_norm": 1.0062427520751953,
      "learning_rate": 0.0001783423566522741,
      "loss": 3.7402,
      "step": 116170
    },
    {
      "epoch": 0.1082985588428889,
      "grad_norm": 1.0348104238510132,
      "learning_rate": 0.0001783404923301789,
      "loss": 3.7948,
      "step": 116180
    },
    {
      "epoch": 0.108307880460968,
      "grad_norm": 0.990595817565918,
      "learning_rate": 0.0001783386280080837,
      "loss": 3.8372,
      "step": 116190
    },
    {
      "epoch": 0.10831720207904709,
      "grad_norm": 1.2164605855941772,
      "learning_rate": 0.0001783367636859885,
      "loss": 3.812,
      "step": 116200
    },
    {
      "epoch": 0.10832652369712618,
      "grad_norm": 1.172580361366272,
      "learning_rate": 0.00017833489936389333,
      "loss": 3.5691,
      "step": 116210
    },
    {
      "epoch": 0.10833584531520528,
      "grad_norm": 1.0862561464309692,
      "learning_rate": 0.0001783330350417981,
      "loss": 3.6241,
      "step": 116220
    },
    {
      "epoch": 0.10834516693328436,
      "grad_norm": 1.1113359928131104,
      "learning_rate": 0.0001783311707197029,
      "loss": 3.7052,
      "step": 116230
    },
    {
      "epoch": 0.10835448855136345,
      "grad_norm": 0.9816150069236755,
      "learning_rate": 0.00017832930639760772,
      "loss": 3.5773,
      "step": 116240
    },
    {
      "epoch": 0.10836381016944255,
      "grad_norm": 1.1849660873413086,
      "learning_rate": 0.0001783274420755125,
      "loss": 3.825,
      "step": 116250
    },
    {
      "epoch": 0.10837313178752164,
      "grad_norm": 1.0843359231948853,
      "learning_rate": 0.0001783255777534173,
      "loss": 3.8442,
      "step": 116260
    },
    {
      "epoch": 0.10838245340560074,
      "grad_norm": 1.0690045356750488,
      "learning_rate": 0.0001783237134313221,
      "loss": 3.7903,
      "step": 116270
    },
    {
      "epoch": 0.10839177502367982,
      "grad_norm": 1.074418067932129,
      "learning_rate": 0.00017832184910922692,
      "loss": 3.6803,
      "step": 116280
    },
    {
      "epoch": 0.10840109664175891,
      "grad_norm": 1.1946827173233032,
      "learning_rate": 0.00017831998478713172,
      "loss": 3.4765,
      "step": 116290
    },
    {
      "epoch": 0.108410418259838,
      "grad_norm": 1.1395447254180908,
      "learning_rate": 0.0001783181204650365,
      "loss": 3.6692,
      "step": 116300
    },
    {
      "epoch": 0.1084197398779171,
      "grad_norm": 0.9429430961608887,
      "learning_rate": 0.0001783162561429413,
      "loss": 3.7356,
      "step": 116310
    },
    {
      "epoch": 0.1084290614959962,
      "grad_norm": 1.219916582107544,
      "learning_rate": 0.00017831439182084611,
      "loss": 3.6074,
      "step": 116320
    },
    {
      "epoch": 0.10843838311407528,
      "grad_norm": 1.1317344903945923,
      "learning_rate": 0.00017831252749875092,
      "loss": 3.6898,
      "step": 116330
    },
    {
      "epoch": 0.10844770473215437,
      "grad_norm": 1.090247392654419,
      "learning_rate": 0.00017831066317665573,
      "loss": 3.6419,
      "step": 116340
    },
    {
      "epoch": 0.10845702635023347,
      "grad_norm": 1.052224040031433,
      "learning_rate": 0.0001783087988545605,
      "loss": 3.7302,
      "step": 116350
    },
    {
      "epoch": 0.10846634796831256,
      "grad_norm": 0.8791065812110901,
      "learning_rate": 0.0001783069345324653,
      "loss": 3.5773,
      "step": 116360
    },
    {
      "epoch": 0.10847566958639165,
      "grad_norm": 1.0250215530395508,
      "learning_rate": 0.00017830507021037012,
      "loss": 3.6639,
      "step": 116370
    },
    {
      "epoch": 0.10848499120447073,
      "grad_norm": 1.1348321437835693,
      "learning_rate": 0.00017830320588827492,
      "loss": 3.6756,
      "step": 116380
    },
    {
      "epoch": 0.10849431282254983,
      "grad_norm": 0.9845430850982666,
      "learning_rate": 0.00017830134156617973,
      "loss": 3.632,
      "step": 116390
    },
    {
      "epoch": 0.10850363444062892,
      "grad_norm": 1.0176514387130737,
      "learning_rate": 0.0001782994772440845,
      "loss": 3.4634,
      "step": 116400
    },
    {
      "epoch": 0.10851295605870802,
      "grad_norm": 1.1744846105575562,
      "learning_rate": 0.00017829761292198932,
      "loss": 3.6517,
      "step": 116410
    },
    {
      "epoch": 0.10852227767678711,
      "grad_norm": 1.0912933349609375,
      "learning_rate": 0.0001782957485998941,
      "loss": 3.5149,
      "step": 116420
    },
    {
      "epoch": 0.10853159929486621,
      "grad_norm": 1.0189136266708374,
      "learning_rate": 0.00017829388427779893,
      "loss": 3.6863,
      "step": 116430
    },
    {
      "epoch": 0.10854092091294529,
      "grad_norm": 0.9377653002738953,
      "learning_rate": 0.00017829201995570374,
      "loss": 3.5814,
      "step": 116440
    },
    {
      "epoch": 0.10855024253102438,
      "grad_norm": 1.2272491455078125,
      "learning_rate": 0.00017829015563360851,
      "loss": 3.6958,
      "step": 116450
    },
    {
      "epoch": 0.10855956414910348,
      "grad_norm": 0.963282585144043,
      "learning_rate": 0.00017828829131151332,
      "loss": 3.7288,
      "step": 116460
    },
    {
      "epoch": 0.10856888576718257,
      "grad_norm": 1.1921656131744385,
      "learning_rate": 0.0001782864269894181,
      "loss": 3.5819,
      "step": 116470
    },
    {
      "epoch": 0.10857820738526167,
      "grad_norm": 1.1437046527862549,
      "learning_rate": 0.0001782845626673229,
      "loss": 3.6537,
      "step": 116480
    },
    {
      "epoch": 0.10858752900334075,
      "grad_norm": 0.978589653968811,
      "learning_rate": 0.00017828269834522774,
      "loss": 3.6271,
      "step": 116490
    },
    {
      "epoch": 0.10859685062141984,
      "grad_norm": 1.1336371898651123,
      "learning_rate": 0.00017828083402313252,
      "loss": 3.7998,
      "step": 116500
    },
    {
      "epoch": 0.10860617223949894,
      "grad_norm": 0.9347367286682129,
      "learning_rate": 0.00017827896970103733,
      "loss": 3.6936,
      "step": 116510
    },
    {
      "epoch": 0.10861549385757803,
      "grad_norm": 1.0645968914031982,
      "learning_rate": 0.0001782771053789421,
      "loss": 3.7763,
      "step": 116520
    },
    {
      "epoch": 0.10862481547565712,
      "grad_norm": 1.066248893737793,
      "learning_rate": 0.0001782752410568469,
      "loss": 3.5766,
      "step": 116530
    },
    {
      "epoch": 0.1086341370937362,
      "grad_norm": 0.9871487617492676,
      "learning_rate": 0.00017827337673475172,
      "loss": 3.6381,
      "step": 116540
    },
    {
      "epoch": 0.1086434587118153,
      "grad_norm": 0.9221113324165344,
      "learning_rate": 0.00017827151241265652,
      "loss": 3.7146,
      "step": 116550
    },
    {
      "epoch": 0.1086527803298944,
      "grad_norm": 1.3839753866195679,
      "learning_rate": 0.00017826964809056133,
      "loss": 3.6945,
      "step": 116560
    },
    {
      "epoch": 0.10866210194797349,
      "grad_norm": 1.0795133113861084,
      "learning_rate": 0.0001782677837684661,
      "loss": 3.8026,
      "step": 116570
    },
    {
      "epoch": 0.10867142356605258,
      "grad_norm": 1.1000581979751587,
      "learning_rate": 0.00017826591944637092,
      "loss": 3.5215,
      "step": 116580
    },
    {
      "epoch": 0.10868074518413166,
      "grad_norm": 1.2518525123596191,
      "learning_rate": 0.00017826405512427572,
      "loss": 3.8102,
      "step": 116590
    },
    {
      "epoch": 0.10869006680221076,
      "grad_norm": 1.2160431146621704,
      "learning_rate": 0.00017826219080218053,
      "loss": 3.498,
      "step": 116600
    },
    {
      "epoch": 0.10869938842028985,
      "grad_norm": 1.2051528692245483,
      "learning_rate": 0.00017826032648008533,
      "loss": 3.7327,
      "step": 116610
    },
    {
      "epoch": 0.10870871003836895,
      "grad_norm": 1.0761317014694214,
      "learning_rate": 0.00017825846215799011,
      "loss": 3.7617,
      "step": 116620
    },
    {
      "epoch": 0.10871803165644804,
      "grad_norm": 0.9605215191841125,
      "learning_rate": 0.00017825659783589492,
      "loss": 3.8068,
      "step": 116630
    },
    {
      "epoch": 0.10872735327452712,
      "grad_norm": 1.3834513425827026,
      "learning_rate": 0.0001782547335137997,
      "loss": 3.5889,
      "step": 116640
    },
    {
      "epoch": 0.10873667489260622,
      "grad_norm": 0.9849690794944763,
      "learning_rate": 0.0001782528691917045,
      "loss": 3.4101,
      "step": 116650
    },
    {
      "epoch": 0.10874599651068531,
      "grad_norm": 1.1976560354232788,
      "learning_rate": 0.00017825100486960934,
      "loss": 3.654,
      "step": 116660
    },
    {
      "epoch": 0.1087553181287644,
      "grad_norm": 1.1973073482513428,
      "learning_rate": 0.00017824914054751412,
      "loss": 3.8749,
      "step": 116670
    },
    {
      "epoch": 0.1087646397468435,
      "grad_norm": 0.9983747601509094,
      "learning_rate": 0.00017824727622541892,
      "loss": 3.7396,
      "step": 116680
    },
    {
      "epoch": 0.1087739613649226,
      "grad_norm": 1.1940110921859741,
      "learning_rate": 0.0001782454119033237,
      "loss": 3.917,
      "step": 116690
    },
    {
      "epoch": 0.10878328298300168,
      "grad_norm": 1.1690409183502197,
      "learning_rate": 0.0001782435475812285,
      "loss": 3.7301,
      "step": 116700
    },
    {
      "epoch": 0.10879260460108077,
      "grad_norm": 1.033325433731079,
      "learning_rate": 0.00017824168325913332,
      "loss": 3.6292,
      "step": 116710
    },
    {
      "epoch": 0.10880192621915986,
      "grad_norm": 1.0899152755737305,
      "learning_rate": 0.00017823981893703812,
      "loss": 3.7328,
      "step": 116720
    },
    {
      "epoch": 0.10881124783723896,
      "grad_norm": 0.8479994535446167,
      "learning_rate": 0.00017823795461494293,
      "loss": 3.7585,
      "step": 116730
    },
    {
      "epoch": 0.10882056945531805,
      "grad_norm": 0.8115620017051697,
      "learning_rate": 0.0001782360902928477,
      "loss": 3.6571,
      "step": 116740
    },
    {
      "epoch": 0.10882989107339713,
      "grad_norm": 1.0071810483932495,
      "learning_rate": 0.00017823422597075251,
      "loss": 3.579,
      "step": 116750
    },
    {
      "epoch": 0.10883921269147623,
      "grad_norm": 1.0829603672027588,
      "learning_rate": 0.00017823236164865732,
      "loss": 3.6837,
      "step": 116760
    },
    {
      "epoch": 0.10884853430955532,
      "grad_norm": 0.8694325089454651,
      "learning_rate": 0.00017823049732656213,
      "loss": 3.6099,
      "step": 116770
    },
    {
      "epoch": 0.10885785592763442,
      "grad_norm": 1.1129872798919678,
      "learning_rate": 0.00017822863300446693,
      "loss": 3.6887,
      "step": 116780
    },
    {
      "epoch": 0.10886717754571351,
      "grad_norm": 1.1452871561050415,
      "learning_rate": 0.0001782267686823717,
      "loss": 3.7346,
      "step": 116790
    },
    {
      "epoch": 0.10887649916379259,
      "grad_norm": 1.1712921857833862,
      "learning_rate": 0.00017822490436027652,
      "loss": 3.622,
      "step": 116800
    },
    {
      "epoch": 0.10888582078187169,
      "grad_norm": 1.1321661472320557,
      "learning_rate": 0.00017822304003818133,
      "loss": 3.7798,
      "step": 116810
    },
    {
      "epoch": 0.10889514239995078,
      "grad_norm": 1.0196812152862549,
      "learning_rate": 0.00017822117571608613,
      "loss": 3.7089,
      "step": 116820
    },
    {
      "epoch": 0.10890446401802988,
      "grad_norm": 0.9853720664978027,
      "learning_rate": 0.00017821931139399094,
      "loss": 3.7489,
      "step": 116830
    },
    {
      "epoch": 0.10891378563610897,
      "grad_norm": 1.0455105304718018,
      "learning_rate": 0.00017821744707189572,
      "loss": 3.6127,
      "step": 116840
    },
    {
      "epoch": 0.10892310725418805,
      "grad_norm": 1.2851800918579102,
      "learning_rate": 0.00017821558274980052,
      "loss": 3.7661,
      "step": 116850
    },
    {
      "epoch": 0.10893242887226715,
      "grad_norm": 1.0916815996170044,
      "learning_rate": 0.00017821371842770533,
      "loss": 3.6624,
      "step": 116860
    },
    {
      "epoch": 0.10894175049034624,
      "grad_norm": 1.0482687950134277,
      "learning_rate": 0.0001782118541056101,
      "loss": 3.6695,
      "step": 116870
    },
    {
      "epoch": 0.10895107210842533,
      "grad_norm": 1.166937232017517,
      "learning_rate": 0.00017820998978351494,
      "loss": 3.5553,
      "step": 116880
    },
    {
      "epoch": 0.10896039372650443,
      "grad_norm": 1.0124502182006836,
      "learning_rate": 0.00017820812546141972,
      "loss": 3.538,
      "step": 116890
    },
    {
      "epoch": 0.10896971534458352,
      "grad_norm": 0.9795903563499451,
      "learning_rate": 0.00017820626113932453,
      "loss": 3.8104,
      "step": 116900
    },
    {
      "epoch": 0.1089790369626626,
      "grad_norm": 1.0776234865188599,
      "learning_rate": 0.00017820439681722933,
      "loss": 3.5666,
      "step": 116910
    },
    {
      "epoch": 0.1089883585807417,
      "grad_norm": 0.9880807399749756,
      "learning_rate": 0.0001782025324951341,
      "loss": 3.6411,
      "step": 116920
    },
    {
      "epoch": 0.10899768019882079,
      "grad_norm": 1.0933693647384644,
      "learning_rate": 0.00017820066817303892,
      "loss": 3.6983,
      "step": 116930
    },
    {
      "epoch": 0.10900700181689989,
      "grad_norm": 0.8670820593833923,
      "learning_rate": 0.00017819880385094373,
      "loss": 3.6311,
      "step": 116940
    },
    {
      "epoch": 0.10901632343497898,
      "grad_norm": 1.0063930749893188,
      "learning_rate": 0.00017819693952884853,
      "loss": 3.791,
      "step": 116950
    },
    {
      "epoch": 0.10902564505305806,
      "grad_norm": 0.8895095586776733,
      "learning_rate": 0.00017819507520675334,
      "loss": 3.7616,
      "step": 116960
    },
    {
      "epoch": 0.10903496667113716,
      "grad_norm": 1.2563596963882446,
      "learning_rate": 0.00017819321088465812,
      "loss": 3.919,
      "step": 116970
    },
    {
      "epoch": 0.10904428828921625,
      "grad_norm": 1.1975868940353394,
      "learning_rate": 0.00017819134656256292,
      "loss": 3.6365,
      "step": 116980
    },
    {
      "epoch": 0.10905360990729535,
      "grad_norm": 1.243299961090088,
      "learning_rate": 0.00017818948224046773,
      "loss": 3.8664,
      "step": 116990
    },
    {
      "epoch": 0.10906293152537444,
      "grad_norm": 0.9506731629371643,
      "learning_rate": 0.00017818761791837254,
      "loss": 3.6416,
      "step": 117000
    },
    {
      "epoch": 0.10907225314345352,
      "grad_norm": 0.9836694598197937,
      "learning_rate": 0.00017818575359627734,
      "loss": 3.7884,
      "step": 117010
    },
    {
      "epoch": 0.10908157476153262,
      "grad_norm": 1.1956461668014526,
      "learning_rate": 0.00017818388927418212,
      "loss": 3.5225,
      "step": 117020
    },
    {
      "epoch": 0.10909089637961171,
      "grad_norm": 1.16078782081604,
      "learning_rate": 0.00017818202495208693,
      "loss": 3.593,
      "step": 117030
    },
    {
      "epoch": 0.1091002179976908,
      "grad_norm": 0.9861001372337341,
      "learning_rate": 0.00017818016062999173,
      "loss": 3.8331,
      "step": 117040
    },
    {
      "epoch": 0.1091095396157699,
      "grad_norm": 1.0479613542556763,
      "learning_rate": 0.00017817829630789654,
      "loss": 3.7464,
      "step": 117050
    },
    {
      "epoch": 0.10911886123384898,
      "grad_norm": 1.060560941696167,
      "learning_rate": 0.00017817643198580135,
      "loss": 3.9808,
      "step": 117060
    },
    {
      "epoch": 0.10912818285192807,
      "grad_norm": 1.0735528469085693,
      "learning_rate": 0.00017817456766370613,
      "loss": 3.7226,
      "step": 117070
    },
    {
      "epoch": 0.10913750447000717,
      "grad_norm": 1.1725302934646606,
      "learning_rate": 0.00017817270334161093,
      "loss": 3.647,
      "step": 117080
    },
    {
      "epoch": 0.10914682608808626,
      "grad_norm": 1.0701054334640503,
      "learning_rate": 0.0001781708390195157,
      "loss": 3.7001,
      "step": 117090
    },
    {
      "epoch": 0.10915614770616536,
      "grad_norm": 0.9573224782943726,
      "learning_rate": 0.00017816897469742055,
      "loss": 3.4546,
      "step": 117100
    },
    {
      "epoch": 0.10916546932424445,
      "grad_norm": 1.1781500577926636,
      "learning_rate": 0.00017816711037532535,
      "loss": 3.4968,
      "step": 117110
    },
    {
      "epoch": 0.10917479094232353,
      "grad_norm": 1.1657195091247559,
      "learning_rate": 0.00017816524605323013,
      "loss": 3.5689,
      "step": 117120
    },
    {
      "epoch": 0.10918411256040263,
      "grad_norm": 1.0125936269760132,
      "learning_rate": 0.00017816338173113494,
      "loss": 3.599,
      "step": 117130
    },
    {
      "epoch": 0.10919343417848172,
      "grad_norm": 1.307905673980713,
      "learning_rate": 0.00017816151740903972,
      "loss": 3.7354,
      "step": 117140
    },
    {
      "epoch": 0.10920275579656082,
      "grad_norm": 1.088865876197815,
      "learning_rate": 0.00017815965308694452,
      "loss": 3.5985,
      "step": 117150
    },
    {
      "epoch": 0.10921207741463991,
      "grad_norm": 1.0747238397598267,
      "learning_rate": 0.00017815778876484933,
      "loss": 3.6229,
      "step": 117160
    },
    {
      "epoch": 0.10922139903271899,
      "grad_norm": 1.0427168607711792,
      "learning_rate": 0.00017815592444275414,
      "loss": 3.5535,
      "step": 117170
    },
    {
      "epoch": 0.10923072065079809,
      "grad_norm": 0.8967083692550659,
      "learning_rate": 0.00017815406012065894,
      "loss": 3.9045,
      "step": 117180
    },
    {
      "epoch": 0.10924004226887718,
      "grad_norm": 1.0024467706680298,
      "learning_rate": 0.00017815219579856372,
      "loss": 3.6784,
      "step": 117190
    },
    {
      "epoch": 0.10924936388695627,
      "grad_norm": 1.0588639974594116,
      "learning_rate": 0.00017815033147646853,
      "loss": 3.8411,
      "step": 117200
    },
    {
      "epoch": 0.10925868550503537,
      "grad_norm": 0.9717163443565369,
      "learning_rate": 0.00017814846715437333,
      "loss": 3.6187,
      "step": 117210
    },
    {
      "epoch": 0.10926800712311445,
      "grad_norm": 1.0248068571090698,
      "learning_rate": 0.00017814660283227814,
      "loss": 3.5636,
      "step": 117220
    },
    {
      "epoch": 0.10927732874119354,
      "grad_norm": 1.170605182647705,
      "learning_rate": 0.00017814473851018295,
      "loss": 3.4699,
      "step": 117230
    },
    {
      "epoch": 0.10928665035927264,
      "grad_norm": 1.0393871068954468,
      "learning_rate": 0.00017814287418808773,
      "loss": 3.7234,
      "step": 117240
    },
    {
      "epoch": 0.10929597197735173,
      "grad_norm": 1.0560634136199951,
      "learning_rate": 0.00017814100986599253,
      "loss": 3.7019,
      "step": 117250
    },
    {
      "epoch": 0.10930529359543083,
      "grad_norm": 0.787714958190918,
      "learning_rate": 0.00017813914554389734,
      "loss": 3.6439,
      "step": 117260
    },
    {
      "epoch": 0.10931461521350991,
      "grad_norm": 0.8817588686943054,
      "learning_rate": 0.00017813728122180214,
      "loss": 3.6254,
      "step": 117270
    },
    {
      "epoch": 0.109323936831589,
      "grad_norm": 1.056481957435608,
      "learning_rate": 0.00017813541689970695,
      "loss": 3.7071,
      "step": 117280
    },
    {
      "epoch": 0.1093332584496681,
      "grad_norm": 1.0901621580123901,
      "learning_rate": 0.00017813355257761173,
      "loss": 3.6574,
      "step": 117290
    },
    {
      "epoch": 0.10934258006774719,
      "grad_norm": 0.9393530488014221,
      "learning_rate": 0.00017813168825551654,
      "loss": 3.6325,
      "step": 117300
    },
    {
      "epoch": 0.10935190168582629,
      "grad_norm": 1.0901589393615723,
      "learning_rate": 0.00017812982393342134,
      "loss": 3.7118,
      "step": 117310
    },
    {
      "epoch": 0.10936122330390537,
      "grad_norm": 0.9153646230697632,
      "learning_rate": 0.00017812795961132612,
      "loss": 3.6417,
      "step": 117320
    },
    {
      "epoch": 0.10937054492198446,
      "grad_norm": 1.0894041061401367,
      "learning_rate": 0.00017812609528923096,
      "loss": 3.4497,
      "step": 117330
    },
    {
      "epoch": 0.10937986654006356,
      "grad_norm": 0.9818440079689026,
      "learning_rate": 0.00017812423096713573,
      "loss": 3.5457,
      "step": 117340
    },
    {
      "epoch": 0.10938918815814265,
      "grad_norm": 0.9672741293907166,
      "learning_rate": 0.00017812236664504054,
      "loss": 3.48,
      "step": 117350
    },
    {
      "epoch": 0.10939850977622174,
      "grad_norm": 1.1997368335723877,
      "learning_rate": 0.00017812050232294535,
      "loss": 3.5209,
      "step": 117360
    },
    {
      "epoch": 0.10940783139430084,
      "grad_norm": 0.82552570104599,
      "learning_rate": 0.00017811863800085013,
      "loss": 3.7426,
      "step": 117370
    },
    {
      "epoch": 0.10941715301237992,
      "grad_norm": 1.0766091346740723,
      "learning_rate": 0.00017811677367875493,
      "loss": 3.6008,
      "step": 117380
    },
    {
      "epoch": 0.10942647463045901,
      "grad_norm": 1.1214267015457153,
      "learning_rate": 0.00017811490935665974,
      "loss": 3.7539,
      "step": 117390
    },
    {
      "epoch": 0.10943579624853811,
      "grad_norm": 1.0096486806869507,
      "learning_rate": 0.00017811304503456455,
      "loss": 3.8266,
      "step": 117400
    },
    {
      "epoch": 0.1094451178666172,
      "grad_norm": 1.130332112312317,
      "learning_rate": 0.00017811118071246935,
      "loss": 3.536,
      "step": 117410
    },
    {
      "epoch": 0.1094544394846963,
      "grad_norm": 1.166832447052002,
      "learning_rate": 0.00017810931639037413,
      "loss": 3.6844,
      "step": 117420
    },
    {
      "epoch": 0.10946376110277538,
      "grad_norm": 1.0279130935668945,
      "learning_rate": 0.00017810745206827894,
      "loss": 3.5706,
      "step": 117430
    },
    {
      "epoch": 0.10947308272085447,
      "grad_norm": 1.1733272075653076,
      "learning_rate": 0.00017810558774618374,
      "loss": 3.6404,
      "step": 117440
    },
    {
      "epoch": 0.10948240433893357,
      "grad_norm": 1.0619083642959595,
      "learning_rate": 0.00017810372342408855,
      "loss": 3.7129,
      "step": 117450
    },
    {
      "epoch": 0.10949172595701266,
      "grad_norm": 1.1715093851089478,
      "learning_rate": 0.00017810185910199336,
      "loss": 3.5482,
      "step": 117460
    },
    {
      "epoch": 0.10950104757509176,
      "grad_norm": 0.9909091591835022,
      "learning_rate": 0.00017809999477989813,
      "loss": 3.7314,
      "step": 117470
    },
    {
      "epoch": 0.10951036919317084,
      "grad_norm": 1.052173376083374,
      "learning_rate": 0.00017809813045780294,
      "loss": 3.7413,
      "step": 117480
    },
    {
      "epoch": 0.10951969081124993,
      "grad_norm": 0.9572758674621582,
      "learning_rate": 0.00017809626613570775,
      "loss": 3.4988,
      "step": 117490
    },
    {
      "epoch": 0.10952901242932903,
      "grad_norm": 1.003812551498413,
      "learning_rate": 0.00017809440181361255,
      "loss": 3.7598,
      "step": 117500
    },
    {
      "epoch": 0.10953833404740812,
      "grad_norm": 1.0075294971466064,
      "learning_rate": 0.00017809253749151736,
      "loss": 3.5524,
      "step": 117510
    },
    {
      "epoch": 0.10954765566548721,
      "grad_norm": 1.1913353204727173,
      "learning_rate": 0.00017809067316942214,
      "loss": 3.7053,
      "step": 117520
    },
    {
      "epoch": 0.1095569772835663,
      "grad_norm": 0.9752768278121948,
      "learning_rate": 0.00017808880884732695,
      "loss": 3.8625,
      "step": 117530
    },
    {
      "epoch": 0.10956629890164539,
      "grad_norm": 1.022707223892212,
      "learning_rate": 0.00017808694452523172,
      "loss": 3.6363,
      "step": 117540
    },
    {
      "epoch": 0.10957562051972448,
      "grad_norm": 1.1879628896713257,
      "learning_rate": 0.00017808508020313656,
      "loss": 3.4965,
      "step": 117550
    },
    {
      "epoch": 0.10958494213780358,
      "grad_norm": 0.9471794366836548,
      "learning_rate": 0.00017808321588104136,
      "loss": 3.7933,
      "step": 117560
    },
    {
      "epoch": 0.10959426375588267,
      "grad_norm": 1.1416927576065063,
      "learning_rate": 0.00017808135155894614,
      "loss": 3.7322,
      "step": 117570
    },
    {
      "epoch": 0.10960358537396177,
      "grad_norm": 1.0480225086212158,
      "learning_rate": 0.00017807948723685095,
      "loss": 3.7894,
      "step": 117580
    },
    {
      "epoch": 0.10961290699204085,
      "grad_norm": 1.0633785724639893,
      "learning_rate": 0.00017807762291475573,
      "loss": 3.568,
      "step": 117590
    },
    {
      "epoch": 0.10962222861011994,
      "grad_norm": 0.9712777137756348,
      "learning_rate": 0.00017807575859266054,
      "loss": 3.7352,
      "step": 117600
    },
    {
      "epoch": 0.10963155022819904,
      "grad_norm": 1.1742663383483887,
      "learning_rate": 0.00017807389427056537,
      "loss": 3.4722,
      "step": 117610
    },
    {
      "epoch": 0.10964087184627813,
      "grad_norm": 1.152275562286377,
      "learning_rate": 0.00017807202994847015,
      "loss": 3.8802,
      "step": 117620
    },
    {
      "epoch": 0.10965019346435723,
      "grad_norm": 1.0080082416534424,
      "learning_rate": 0.00017807016562637495,
      "loss": 3.6069,
      "step": 117630
    },
    {
      "epoch": 0.1096595150824363,
      "grad_norm": 1.1483508348464966,
      "learning_rate": 0.00017806830130427973,
      "loss": 3.6209,
      "step": 117640
    },
    {
      "epoch": 0.1096688367005154,
      "grad_norm": 1.0526177883148193,
      "learning_rate": 0.00017806643698218454,
      "loss": 3.5197,
      "step": 117650
    },
    {
      "epoch": 0.1096781583185945,
      "grad_norm": 0.9739837050437927,
      "learning_rate": 0.00017806457266008935,
      "loss": 3.716,
      "step": 117660
    },
    {
      "epoch": 0.10968747993667359,
      "grad_norm": 1.0475987195968628,
      "learning_rate": 0.00017806270833799415,
      "loss": 3.7116,
      "step": 117670
    },
    {
      "epoch": 0.10969680155475268,
      "grad_norm": 1.0513758659362793,
      "learning_rate": 0.00017806084401589896,
      "loss": 3.6518,
      "step": 117680
    },
    {
      "epoch": 0.10970612317283177,
      "grad_norm": 1.0627347230911255,
      "learning_rate": 0.00017805897969380374,
      "loss": 3.729,
      "step": 117690
    },
    {
      "epoch": 0.10971544479091086,
      "grad_norm": 0.8590103983879089,
      "learning_rate": 0.00017805711537170854,
      "loss": 3.7031,
      "step": 117700
    },
    {
      "epoch": 0.10972476640898995,
      "grad_norm": 1.201129674911499,
      "learning_rate": 0.00017805525104961335,
      "loss": 3.6094,
      "step": 117710
    },
    {
      "epoch": 0.10973408802706905,
      "grad_norm": 1.0409380197525024,
      "learning_rate": 0.00017805338672751816,
      "loss": 3.8087,
      "step": 117720
    },
    {
      "epoch": 0.10974340964514814,
      "grad_norm": 1.1704837083816528,
      "learning_rate": 0.00017805152240542296,
      "loss": 3.6129,
      "step": 117730
    },
    {
      "epoch": 0.10975273126322722,
      "grad_norm": 1.1729754209518433,
      "learning_rate": 0.00017804965808332774,
      "loss": 3.7002,
      "step": 117740
    },
    {
      "epoch": 0.10976205288130632,
      "grad_norm": 1.046373963356018,
      "learning_rate": 0.00017804779376123255,
      "loss": 3.6177,
      "step": 117750
    },
    {
      "epoch": 0.10977137449938541,
      "grad_norm": 1.1118662357330322,
      "learning_rate": 0.00017804592943913736,
      "loss": 3.7686,
      "step": 117760
    },
    {
      "epoch": 0.10978069611746451,
      "grad_norm": 1.043724775314331,
      "learning_rate": 0.00017804406511704213,
      "loss": 3.6836,
      "step": 117770
    },
    {
      "epoch": 0.1097900177355436,
      "grad_norm": 1.2126703262329102,
      "learning_rate": 0.00017804220079494697,
      "loss": 3.6117,
      "step": 117780
    },
    {
      "epoch": 0.1097993393536227,
      "grad_norm": 1.2515510320663452,
      "learning_rate": 0.00017804033647285175,
      "loss": 3.8199,
      "step": 117790
    },
    {
      "epoch": 0.10980866097170178,
      "grad_norm": 0.9796364307403564,
      "learning_rate": 0.00017803847215075655,
      "loss": 3.6778,
      "step": 117800
    },
    {
      "epoch": 0.10981798258978087,
      "grad_norm": 1.0411031246185303,
      "learning_rate": 0.00017803660782866136,
      "loss": 3.45,
      "step": 117810
    },
    {
      "epoch": 0.10982730420785997,
      "grad_norm": 1.0152802467346191,
      "learning_rate": 0.00017803474350656614,
      "loss": 3.7248,
      "step": 117820
    },
    {
      "epoch": 0.10983662582593906,
      "grad_norm": 1.0937824249267578,
      "learning_rate": 0.00017803287918447095,
      "loss": 3.8004,
      "step": 117830
    },
    {
      "epoch": 0.10984594744401815,
      "grad_norm": 1.0656806230545044,
      "learning_rate": 0.00017803101486237575,
      "loss": 3.7793,
      "step": 117840
    },
    {
      "epoch": 0.10985526906209724,
      "grad_norm": 0.997896671295166,
      "learning_rate": 0.00017802915054028056,
      "loss": 3.6686,
      "step": 117850
    },
    {
      "epoch": 0.10986459068017633,
      "grad_norm": 0.9984060525894165,
      "learning_rate": 0.00017802728621818536,
      "loss": 3.5577,
      "step": 117860
    },
    {
      "epoch": 0.10987391229825542,
      "grad_norm": 1.1063913106918335,
      "learning_rate": 0.00017802542189609014,
      "loss": 3.5802,
      "step": 117870
    },
    {
      "epoch": 0.10988323391633452,
      "grad_norm": 1.1279734373092651,
      "learning_rate": 0.00017802355757399495,
      "loss": 3.6431,
      "step": 117880
    },
    {
      "epoch": 0.10989255553441361,
      "grad_norm": 0.9619153738021851,
      "learning_rate": 0.00017802169325189976,
      "loss": 3.6274,
      "step": 117890
    },
    {
      "epoch": 0.1099018771524927,
      "grad_norm": 1.159308671951294,
      "learning_rate": 0.00017801982892980456,
      "loss": 3.4399,
      "step": 117900
    },
    {
      "epoch": 0.10991119877057179,
      "grad_norm": 1.2631689310073853,
      "learning_rate": 0.00017801796460770937,
      "loss": 3.7109,
      "step": 117910
    },
    {
      "epoch": 0.10992052038865088,
      "grad_norm": 1.1727501153945923,
      "learning_rate": 0.00017801610028561415,
      "loss": 3.6243,
      "step": 117920
    },
    {
      "epoch": 0.10992984200672998,
      "grad_norm": 0.9511098861694336,
      "learning_rate": 0.00017801423596351895,
      "loss": 3.6337,
      "step": 117930
    },
    {
      "epoch": 0.10993916362480907,
      "grad_norm": 1.0795962810516357,
      "learning_rate": 0.00017801237164142376,
      "loss": 3.3975,
      "step": 117940
    },
    {
      "epoch": 0.10994848524288815,
      "grad_norm": 1.1403049230575562,
      "learning_rate": 0.00017801050731932857,
      "loss": 3.6168,
      "step": 117950
    },
    {
      "epoch": 0.10995780686096725,
      "grad_norm": 1.072850227355957,
      "learning_rate": 0.00017800864299723335,
      "loss": 3.66,
      "step": 117960
    },
    {
      "epoch": 0.10996712847904634,
      "grad_norm": 1.1603529453277588,
      "learning_rate": 0.00017800677867513815,
      "loss": 3.5902,
      "step": 117970
    },
    {
      "epoch": 0.10997645009712544,
      "grad_norm": 1.231557846069336,
      "learning_rate": 0.00017800491435304296,
      "loss": 3.6414,
      "step": 117980
    },
    {
      "epoch": 0.10998577171520453,
      "grad_norm": 1.0532724857330322,
      "learning_rate": 0.00017800305003094774,
      "loss": 3.5876,
      "step": 117990
    },
    {
      "epoch": 0.10999509333328361,
      "grad_norm": 1.1851651668548584,
      "learning_rate": 0.00017800118570885257,
      "loss": 3.8225,
      "step": 118000
    },
    {
      "epoch": 0.1100044149513627,
      "grad_norm": 1.045580506324768,
      "learning_rate": 0.00017799932138675735,
      "loss": 3.8372,
      "step": 118010
    },
    {
      "epoch": 0.1100137365694418,
      "grad_norm": 0.946937084197998,
      "learning_rate": 0.00017799745706466216,
      "loss": 3.612,
      "step": 118020
    },
    {
      "epoch": 0.1100230581875209,
      "grad_norm": 1.0883444547653198,
      "learning_rate": 0.00017799559274256696,
      "loss": 3.8454,
      "step": 118030
    },
    {
      "epoch": 0.11003237980559999,
      "grad_norm": 1.2039958238601685,
      "learning_rate": 0.00017799372842047174,
      "loss": 3.5834,
      "step": 118040
    },
    {
      "epoch": 0.11004170142367908,
      "grad_norm": 1.0712864398956299,
      "learning_rate": 0.00017799186409837655,
      "loss": 3.7798,
      "step": 118050
    },
    {
      "epoch": 0.11005102304175816,
      "grad_norm": 0.8938766717910767,
      "learning_rate": 0.00017798999977628135,
      "loss": 3.5883,
      "step": 118060
    },
    {
      "epoch": 0.11006034465983726,
      "grad_norm": 0.9831303358078003,
      "learning_rate": 0.00017798813545418616,
      "loss": 3.5969,
      "step": 118070
    },
    {
      "epoch": 0.11006966627791635,
      "grad_norm": 1.00808846950531,
      "learning_rate": 0.00017798627113209097,
      "loss": 3.6763,
      "step": 118080
    },
    {
      "epoch": 0.11007898789599545,
      "grad_norm": 1.113080620765686,
      "learning_rate": 0.00017798440680999575,
      "loss": 3.7303,
      "step": 118090
    },
    {
      "epoch": 0.11008830951407454,
      "grad_norm": 1.0820319652557373,
      "learning_rate": 0.00017798254248790055,
      "loss": 3.6848,
      "step": 118100
    },
    {
      "epoch": 0.11009763113215362,
      "grad_norm": 1.008569598197937,
      "learning_rate": 0.00017798067816580536,
      "loss": 3.4879,
      "step": 118110
    },
    {
      "epoch": 0.11010695275023272,
      "grad_norm": 1.2408761978149414,
      "learning_rate": 0.00017797881384371017,
      "loss": 3.5876,
      "step": 118120
    },
    {
      "epoch": 0.11011627436831181,
      "grad_norm": 1.2480506896972656,
      "learning_rate": 0.00017797694952161497,
      "loss": 3.5931,
      "step": 118130
    },
    {
      "epoch": 0.1101255959863909,
      "grad_norm": 1.2838771343231201,
      "learning_rate": 0.00017797508519951975,
      "loss": 3.6942,
      "step": 118140
    },
    {
      "epoch": 0.11013491760447,
      "grad_norm": 1.1985775232315063,
      "learning_rate": 0.00017797322087742456,
      "loss": 3.6302,
      "step": 118150
    },
    {
      "epoch": 0.11014423922254908,
      "grad_norm": 1.043415904045105,
      "learning_rate": 0.00017797135655532934,
      "loss": 3.805,
      "step": 118160
    },
    {
      "epoch": 0.11015356084062818,
      "grad_norm": 1.012534260749817,
      "learning_rate": 0.00017796949223323417,
      "loss": 3.616,
      "step": 118170
    },
    {
      "epoch": 0.11016288245870727,
      "grad_norm": 0.8795356154441833,
      "learning_rate": 0.00017796762791113898,
      "loss": 3.4712,
      "step": 118180
    },
    {
      "epoch": 0.11017220407678636,
      "grad_norm": 1.1134542226791382,
      "learning_rate": 0.00017796576358904376,
      "loss": 3.6686,
      "step": 118190
    },
    {
      "epoch": 0.11018152569486546,
      "grad_norm": 1.08842134475708,
      "learning_rate": 0.00017796389926694856,
      "loss": 3.4036,
      "step": 118200
    },
    {
      "epoch": 0.11019084731294454,
      "grad_norm": 1.120917797088623,
      "learning_rate": 0.00017796203494485334,
      "loss": 3.6108,
      "step": 118210
    },
    {
      "epoch": 0.11020016893102363,
      "grad_norm": 1.088823676109314,
      "learning_rate": 0.00017796017062275815,
      "loss": 3.7706,
      "step": 118220
    },
    {
      "epoch": 0.11020949054910273,
      "grad_norm": 1.0569294691085815,
      "learning_rate": 0.00017795830630066298,
      "loss": 3.6922,
      "step": 118230
    },
    {
      "epoch": 0.11021881216718182,
      "grad_norm": 1.2554551362991333,
      "learning_rate": 0.00017795644197856776,
      "loss": 3.6776,
      "step": 118240
    },
    {
      "epoch": 0.11022813378526092,
      "grad_norm": 1.097333312034607,
      "learning_rate": 0.00017795457765647257,
      "loss": 3.5834,
      "step": 118250
    },
    {
      "epoch": 0.11023745540334001,
      "grad_norm": 1.2292096614837646,
      "learning_rate": 0.00017795271333437735,
      "loss": 3.7059,
      "step": 118260
    },
    {
      "epoch": 0.11024677702141909,
      "grad_norm": 0.9614948034286499,
      "learning_rate": 0.00017795084901228215,
      "loss": 3.5671,
      "step": 118270
    },
    {
      "epoch": 0.11025609863949819,
      "grad_norm": 1.068009376525879,
      "learning_rate": 0.00017794898469018696,
      "loss": 3.6086,
      "step": 118280
    },
    {
      "epoch": 0.11026542025757728,
      "grad_norm": 1.1439688205718994,
      "learning_rate": 0.00017794712036809176,
      "loss": 3.597,
      "step": 118290
    },
    {
      "epoch": 0.11027474187565638,
      "grad_norm": 0.840604841709137,
      "learning_rate": 0.00017794525604599657,
      "loss": 3.5134,
      "step": 118300
    },
    {
      "epoch": 0.11028406349373547,
      "grad_norm": 1.1707819700241089,
      "learning_rate": 0.00017794339172390135,
      "loss": 3.6776,
      "step": 118310
    },
    {
      "epoch": 0.11029338511181455,
      "grad_norm": 1.1397219896316528,
      "learning_rate": 0.00017794152740180616,
      "loss": 3.7343,
      "step": 118320
    },
    {
      "epoch": 0.11030270672989365,
      "grad_norm": 1.1210427284240723,
      "learning_rate": 0.00017793966307971096,
      "loss": 3.4949,
      "step": 118330
    },
    {
      "epoch": 0.11031202834797274,
      "grad_norm": 1.1679253578186035,
      "learning_rate": 0.00017793779875761577,
      "loss": 3.4446,
      "step": 118340
    },
    {
      "epoch": 0.11032134996605183,
      "grad_norm": 1.066184401512146,
      "learning_rate": 0.00017793593443552058,
      "loss": 3.6216,
      "step": 118350
    },
    {
      "epoch": 0.11033067158413093,
      "grad_norm": 1.3140854835510254,
      "learning_rate": 0.00017793407011342535,
      "loss": 3.5468,
      "step": 118360
    },
    {
      "epoch": 0.11033999320221001,
      "grad_norm": 0.9711753129959106,
      "learning_rate": 0.00017793220579133016,
      "loss": 3.6561,
      "step": 118370
    },
    {
      "epoch": 0.1103493148202891,
      "grad_norm": 1.1529399156570435,
      "learning_rate": 0.00017793034146923497,
      "loss": 3.652,
      "step": 118380
    },
    {
      "epoch": 0.1103586364383682,
      "grad_norm": 1.0376012325286865,
      "learning_rate": 0.00017792847714713977,
      "loss": 3.6167,
      "step": 118390
    },
    {
      "epoch": 0.1103679580564473,
      "grad_norm": 0.8760204911231995,
      "learning_rate": 0.00017792661282504458,
      "loss": 3.6802,
      "step": 118400
    },
    {
      "epoch": 0.11037727967452639,
      "grad_norm": 1.1354899406433105,
      "learning_rate": 0.00017792474850294936,
      "loss": 3.6183,
      "step": 118410
    },
    {
      "epoch": 0.11038660129260547,
      "grad_norm": 1.1268277168273926,
      "learning_rate": 0.00017792288418085417,
      "loss": 3.5626,
      "step": 118420
    },
    {
      "epoch": 0.11039592291068456,
      "grad_norm": 1.0451151132583618,
      "learning_rate": 0.00017792101985875897,
      "loss": 3.6447,
      "step": 118430
    },
    {
      "epoch": 0.11040524452876366,
      "grad_norm": 0.8984562158584595,
      "learning_rate": 0.00017791915553666375,
      "loss": 3.9323,
      "step": 118440
    },
    {
      "epoch": 0.11041456614684275,
      "grad_norm": 0.9774864912033081,
      "learning_rate": 0.00017791729121456858,
      "loss": 3.5835,
      "step": 118450
    },
    {
      "epoch": 0.11042388776492185,
      "grad_norm": 1.0331459045410156,
      "learning_rate": 0.00017791542689247336,
      "loss": 3.7655,
      "step": 118460
    },
    {
      "epoch": 0.11043320938300094,
      "grad_norm": 1.1014314889907837,
      "learning_rate": 0.00017791356257037817,
      "loss": 3.769,
      "step": 118470
    },
    {
      "epoch": 0.11044253100108002,
      "grad_norm": 1.0296682119369507,
      "learning_rate": 0.00017791169824828298,
      "loss": 3.7115,
      "step": 118480
    },
    {
      "epoch": 0.11045185261915912,
      "grad_norm": 0.9218773245811462,
      "learning_rate": 0.00017790983392618776,
      "loss": 3.5042,
      "step": 118490
    },
    {
      "epoch": 0.11046117423723821,
      "grad_norm": 1.0858415365219116,
      "learning_rate": 0.00017790796960409256,
      "loss": 3.6732,
      "step": 118500
    },
    {
      "epoch": 0.1104704958553173,
      "grad_norm": 1.0863761901855469,
      "learning_rate": 0.00017790610528199737,
      "loss": 3.7205,
      "step": 118510
    },
    {
      "epoch": 0.1104798174733964,
      "grad_norm": 1.0757906436920166,
      "learning_rate": 0.00017790424095990217,
      "loss": 3.7167,
      "step": 118520
    },
    {
      "epoch": 0.11048913909147548,
      "grad_norm": 1.1192114353179932,
      "learning_rate": 0.00017790237663780698,
      "loss": 3.712,
      "step": 118530
    },
    {
      "epoch": 0.11049846070955457,
      "grad_norm": 1.053505539894104,
      "learning_rate": 0.00017790051231571176,
      "loss": 3.6519,
      "step": 118540
    },
    {
      "epoch": 0.11050778232763367,
      "grad_norm": 1.1580935716629028,
      "learning_rate": 0.00017789864799361657,
      "loss": 3.7025,
      "step": 118550
    },
    {
      "epoch": 0.11051710394571276,
      "grad_norm": 0.9510852098464966,
      "learning_rate": 0.00017789678367152137,
      "loss": 3.6084,
      "step": 118560
    },
    {
      "epoch": 0.11052642556379186,
      "grad_norm": 1.0615825653076172,
      "learning_rate": 0.00017789491934942618,
      "loss": 3.6613,
      "step": 118570
    },
    {
      "epoch": 0.11053574718187094,
      "grad_norm": 1.2112834453582764,
      "learning_rate": 0.00017789305502733098,
      "loss": 3.6215,
      "step": 118580
    },
    {
      "epoch": 0.11054506879995003,
      "grad_norm": 1.0309758186340332,
      "learning_rate": 0.00017789119070523576,
      "loss": 3.7659,
      "step": 118590
    },
    {
      "epoch": 0.11055439041802913,
      "grad_norm": 1.1086889505386353,
      "learning_rate": 0.00017788932638314057,
      "loss": 3.7689,
      "step": 118600
    },
    {
      "epoch": 0.11056371203610822,
      "grad_norm": 1.244404673576355,
      "learning_rate": 0.00017788746206104535,
      "loss": 3.679,
      "step": 118610
    },
    {
      "epoch": 0.11057303365418732,
      "grad_norm": 1.2236523628234863,
      "learning_rate": 0.00017788559773895018,
      "loss": 3.6197,
      "step": 118620
    },
    {
      "epoch": 0.1105823552722664,
      "grad_norm": 1.0068199634552002,
      "learning_rate": 0.000177883733416855,
      "loss": 3.6232,
      "step": 118630
    },
    {
      "epoch": 0.11059167689034549,
      "grad_norm": 1.0309960842132568,
      "learning_rate": 0.00017788186909475977,
      "loss": 3.7806,
      "step": 118640
    },
    {
      "epoch": 0.11060099850842459,
      "grad_norm": 1.083641767501831,
      "learning_rate": 0.00017788000477266457,
      "loss": 3.836,
      "step": 118650
    },
    {
      "epoch": 0.11061032012650368,
      "grad_norm": 1.0229661464691162,
      "learning_rate": 0.00017787814045056935,
      "loss": 3.5697,
      "step": 118660
    },
    {
      "epoch": 0.11061964174458278,
      "grad_norm": 0.9869531989097595,
      "learning_rate": 0.00017787627612847416,
      "loss": 3.7922,
      "step": 118670
    },
    {
      "epoch": 0.11062896336266186,
      "grad_norm": 1.1679939031600952,
      "learning_rate": 0.000177874411806379,
      "loss": 3.5159,
      "step": 118680
    },
    {
      "epoch": 0.11063828498074095,
      "grad_norm": 1.0919952392578125,
      "learning_rate": 0.00017787254748428377,
      "loss": 3.6369,
      "step": 118690
    },
    {
      "epoch": 0.11064760659882004,
      "grad_norm": 1.0254687070846558,
      "learning_rate": 0.00017787068316218858,
      "loss": 3.9344,
      "step": 118700
    },
    {
      "epoch": 0.11065692821689914,
      "grad_norm": 1.2138022184371948,
      "learning_rate": 0.00017786881884009336,
      "loss": 3.7589,
      "step": 118710
    },
    {
      "epoch": 0.11066624983497823,
      "grad_norm": 0.8948147296905518,
      "learning_rate": 0.00017786695451799816,
      "loss": 3.7321,
      "step": 118720
    },
    {
      "epoch": 0.11067557145305733,
      "grad_norm": 1.0818493366241455,
      "learning_rate": 0.00017786509019590297,
      "loss": 3.5588,
      "step": 118730
    },
    {
      "epoch": 0.11068489307113641,
      "grad_norm": 1.1326302289962769,
      "learning_rate": 0.00017786322587380778,
      "loss": 3.705,
      "step": 118740
    },
    {
      "epoch": 0.1106942146892155,
      "grad_norm": 1.004076361656189,
      "learning_rate": 0.00017786136155171258,
      "loss": 3.713,
      "step": 118750
    },
    {
      "epoch": 0.1107035363072946,
      "grad_norm": 1.2035436630249023,
      "learning_rate": 0.00017785949722961736,
      "loss": 3.8499,
      "step": 118760
    },
    {
      "epoch": 0.11071285792537369,
      "grad_norm": 0.9800660610198975,
      "learning_rate": 0.00017785763290752217,
      "loss": 3.8096,
      "step": 118770
    },
    {
      "epoch": 0.11072217954345279,
      "grad_norm": 0.9289263486862183,
      "learning_rate": 0.00017785576858542698,
      "loss": 3.6678,
      "step": 118780
    },
    {
      "epoch": 0.11073150116153187,
      "grad_norm": 1.1309504508972168,
      "learning_rate": 0.00017785390426333178,
      "loss": 3.735,
      "step": 118790
    },
    {
      "epoch": 0.11074082277961096,
      "grad_norm": 0.8842321634292603,
      "learning_rate": 0.0001778520399412366,
      "loss": 3.4764,
      "step": 118800
    },
    {
      "epoch": 0.11075014439769006,
      "grad_norm": 1.071805477142334,
      "learning_rate": 0.00017785017561914137,
      "loss": 3.6398,
      "step": 118810
    },
    {
      "epoch": 0.11075946601576915,
      "grad_norm": 1.0330795049667358,
      "learning_rate": 0.00017784831129704617,
      "loss": 3.7121,
      "step": 118820
    },
    {
      "epoch": 0.11076878763384825,
      "grad_norm": 1.053277611732483,
      "learning_rate": 0.00017784644697495098,
      "loss": 3.9988,
      "step": 118830
    },
    {
      "epoch": 0.11077810925192733,
      "grad_norm": 1.1736449003219604,
      "learning_rate": 0.00017784458265285579,
      "loss": 3.5556,
      "step": 118840
    },
    {
      "epoch": 0.11078743087000642,
      "grad_norm": 1.1659257411956787,
      "learning_rate": 0.0001778427183307606,
      "loss": 3.4629,
      "step": 118850
    },
    {
      "epoch": 0.11079675248808551,
      "grad_norm": 1.2711868286132812,
      "learning_rate": 0.00017784085400866537,
      "loss": 3.5245,
      "step": 118860
    },
    {
      "epoch": 0.11080607410616461,
      "grad_norm": 0.9310900568962097,
      "learning_rate": 0.00017783898968657018,
      "loss": 3.7,
      "step": 118870
    },
    {
      "epoch": 0.1108153957242437,
      "grad_norm": 0.9784958362579346,
      "learning_rate": 0.00017783712536447498,
      "loss": 3.6413,
      "step": 118880
    },
    {
      "epoch": 0.11082471734232278,
      "grad_norm": 0.9817454218864441,
      "learning_rate": 0.00017783526104237976,
      "loss": 3.59,
      "step": 118890
    },
    {
      "epoch": 0.11083403896040188,
      "grad_norm": 0.8801296949386597,
      "learning_rate": 0.0001778333967202846,
      "loss": 3.6332,
      "step": 118900
    },
    {
      "epoch": 0.11084336057848097,
      "grad_norm": 1.0270923376083374,
      "learning_rate": 0.00017783153239818938,
      "loss": 3.6307,
      "step": 118910
    },
    {
      "epoch": 0.11085268219656007,
      "grad_norm": 1.1716152429580688,
      "learning_rate": 0.00017782966807609418,
      "loss": 3.6582,
      "step": 118920
    },
    {
      "epoch": 0.11086200381463916,
      "grad_norm": 0.9938703179359436,
      "learning_rate": 0.000177827803753999,
      "loss": 3.6757,
      "step": 118930
    },
    {
      "epoch": 0.11087132543271826,
      "grad_norm": 1.047327995300293,
      "learning_rate": 0.00017782593943190377,
      "loss": 3.625,
      "step": 118940
    },
    {
      "epoch": 0.11088064705079734,
      "grad_norm": 1.1161284446716309,
      "learning_rate": 0.00017782407510980857,
      "loss": 3.7211,
      "step": 118950
    },
    {
      "epoch": 0.11088996866887643,
      "grad_norm": 1.0430185794830322,
      "learning_rate": 0.00017782221078771338,
      "loss": 3.5041,
      "step": 118960
    },
    {
      "epoch": 0.11089929028695553,
      "grad_norm": 1.0125178098678589,
      "learning_rate": 0.0001778203464656182,
      "loss": 3.4636,
      "step": 118970
    },
    {
      "epoch": 0.11090861190503462,
      "grad_norm": 0.9780397415161133,
      "learning_rate": 0.000177818482143523,
      "loss": 3.6358,
      "step": 118980
    },
    {
      "epoch": 0.11091793352311372,
      "grad_norm": 0.9328268766403198,
      "learning_rate": 0.00017781661782142777,
      "loss": 3.671,
      "step": 118990
    },
    {
      "epoch": 0.1109272551411928,
      "grad_norm": 1.158646583557129,
      "learning_rate": 0.00017781475349933258,
      "loss": 3.5782,
      "step": 119000
    },
    {
      "epoch": 0.11093657675927189,
      "grad_norm": 1.1309019327163696,
      "learning_rate": 0.00017781288917723738,
      "loss": 3.7733,
      "step": 119010
    },
    {
      "epoch": 0.11094589837735098,
      "grad_norm": 1.1059268712997437,
      "learning_rate": 0.0001778110248551422,
      "loss": 3.7346,
      "step": 119020
    },
    {
      "epoch": 0.11095521999543008,
      "grad_norm": 1.0519888401031494,
      "learning_rate": 0.000177809160533047,
      "loss": 3.7378,
      "step": 119030
    },
    {
      "epoch": 0.11096454161350917,
      "grad_norm": 1.089066505432129,
      "learning_rate": 0.00017780729621095178,
      "loss": 3.5737,
      "step": 119040
    },
    {
      "epoch": 0.11097386323158825,
      "grad_norm": 0.9390338659286499,
      "learning_rate": 0.00017780543188885658,
      "loss": 3.6295,
      "step": 119050
    },
    {
      "epoch": 0.11098318484966735,
      "grad_norm": 1.1180636882781982,
      "learning_rate": 0.00017780356756676136,
      "loss": 3.5866,
      "step": 119060
    },
    {
      "epoch": 0.11099250646774644,
      "grad_norm": 1.0417616367340088,
      "learning_rate": 0.0001778017032446662,
      "loss": 3.6033,
      "step": 119070
    },
    {
      "epoch": 0.11100182808582554,
      "grad_norm": 1.1661301851272583,
      "learning_rate": 0.000177799838922571,
      "loss": 3.7998,
      "step": 119080
    },
    {
      "epoch": 0.11101114970390463,
      "grad_norm": 1.0242823362350464,
      "learning_rate": 0.00017779797460047578,
      "loss": 3.6403,
      "step": 119090
    },
    {
      "epoch": 0.11102047132198371,
      "grad_norm": 1.0615131855010986,
      "learning_rate": 0.0001777961102783806,
      "loss": 3.5929,
      "step": 119100
    },
    {
      "epoch": 0.11102979294006281,
      "grad_norm": 0.9423764944076538,
      "learning_rate": 0.00017779424595628537,
      "loss": 3.7748,
      "step": 119110
    },
    {
      "epoch": 0.1110391145581419,
      "grad_norm": 1.0555205345153809,
      "learning_rate": 0.00017779238163419017,
      "loss": 3.6749,
      "step": 119120
    },
    {
      "epoch": 0.111048436176221,
      "grad_norm": 1.1116042137145996,
      "learning_rate": 0.000177790517312095,
      "loss": 3.6546,
      "step": 119130
    },
    {
      "epoch": 0.11105775779430009,
      "grad_norm": 1.224895715713501,
      "learning_rate": 0.00017778865298999979,
      "loss": 3.5494,
      "step": 119140
    },
    {
      "epoch": 0.11106707941237919,
      "grad_norm": 1.032128095626831,
      "learning_rate": 0.0001777867886679046,
      "loss": 3.7998,
      "step": 119150
    },
    {
      "epoch": 0.11107640103045827,
      "grad_norm": 1.025829553604126,
      "learning_rate": 0.00017778492434580937,
      "loss": 3.7059,
      "step": 119160
    },
    {
      "epoch": 0.11108572264853736,
      "grad_norm": 1.208142876625061,
      "learning_rate": 0.00017778306002371418,
      "loss": 3.5346,
      "step": 119170
    },
    {
      "epoch": 0.11109504426661646,
      "grad_norm": 1.257517695426941,
      "learning_rate": 0.00017778119570161898,
      "loss": 3.6097,
      "step": 119180
    },
    {
      "epoch": 0.11110436588469555,
      "grad_norm": 1.1187562942504883,
      "learning_rate": 0.0001777793313795238,
      "loss": 3.682,
      "step": 119190
    },
    {
      "epoch": 0.11111368750277464,
      "grad_norm": 1.153664231300354,
      "learning_rate": 0.0001777774670574286,
      "loss": 3.6233,
      "step": 119200
    },
    {
      "epoch": 0.11112300912085372,
      "grad_norm": 0.9733367562294006,
      "learning_rate": 0.00017777560273533338,
      "loss": 3.506,
      "step": 119210
    },
    {
      "epoch": 0.11113233073893282,
      "grad_norm": 1.1015232801437378,
      "learning_rate": 0.00017777373841323818,
      "loss": 3.6993,
      "step": 119220
    },
    {
      "epoch": 0.11114165235701191,
      "grad_norm": 1.1002182960510254,
      "learning_rate": 0.000177771874091143,
      "loss": 3.7693,
      "step": 119230
    },
    {
      "epoch": 0.11115097397509101,
      "grad_norm": 0.8947045207023621,
      "learning_rate": 0.0001777700097690478,
      "loss": 3.781,
      "step": 119240
    },
    {
      "epoch": 0.1111602955931701,
      "grad_norm": 1.1073577404022217,
      "learning_rate": 0.0001777681454469526,
      "loss": 3.6953,
      "step": 119250
    },
    {
      "epoch": 0.11116961721124918,
      "grad_norm": 1.1168100833892822,
      "learning_rate": 0.00017776628112485738,
      "loss": 3.7821,
      "step": 119260
    },
    {
      "epoch": 0.11117893882932828,
      "grad_norm": 1.2916648387908936,
      "learning_rate": 0.00017776441680276219,
      "loss": 3.9719,
      "step": 119270
    },
    {
      "epoch": 0.11118826044740737,
      "grad_norm": 1.0608808994293213,
      "learning_rate": 0.00017776255248066697,
      "loss": 3.8911,
      "step": 119280
    },
    {
      "epoch": 0.11119758206548647,
      "grad_norm": 1.079213261604309,
      "learning_rate": 0.0001777606881585718,
      "loss": 3.7786,
      "step": 119290
    },
    {
      "epoch": 0.11120690368356556,
      "grad_norm": 1.1796021461486816,
      "learning_rate": 0.0001777588238364766,
      "loss": 3.7642,
      "step": 119300
    },
    {
      "epoch": 0.11121622530164464,
      "grad_norm": 1.2960519790649414,
      "learning_rate": 0.00017775695951438138,
      "loss": 3.5895,
      "step": 119310
    },
    {
      "epoch": 0.11122554691972374,
      "grad_norm": 1.093869924545288,
      "learning_rate": 0.0001777550951922862,
      "loss": 3.5984,
      "step": 119320
    },
    {
      "epoch": 0.11123486853780283,
      "grad_norm": 0.7960858941078186,
      "learning_rate": 0.00017775323087019097,
      "loss": 3.6646,
      "step": 119330
    },
    {
      "epoch": 0.11124419015588193,
      "grad_norm": 1.0728261470794678,
      "learning_rate": 0.00017775136654809578,
      "loss": 3.779,
      "step": 119340
    },
    {
      "epoch": 0.11125351177396102,
      "grad_norm": 0.922049343585968,
      "learning_rate": 0.0001777495022260006,
      "loss": 3.7088,
      "step": 119350
    },
    {
      "epoch": 0.1112628333920401,
      "grad_norm": 0.9292984008789062,
      "learning_rate": 0.0001777476379039054,
      "loss": 3.6502,
      "step": 119360
    },
    {
      "epoch": 0.1112721550101192,
      "grad_norm": 1.0571258068084717,
      "learning_rate": 0.0001777457735818102,
      "loss": 3.6649,
      "step": 119370
    },
    {
      "epoch": 0.11128147662819829,
      "grad_norm": 1.112029790878296,
      "learning_rate": 0.00017774390925971497,
      "loss": 3.7512,
      "step": 119380
    },
    {
      "epoch": 0.11129079824627738,
      "grad_norm": 1.2551017999649048,
      "learning_rate": 0.00017774204493761978,
      "loss": 3.7182,
      "step": 119390
    },
    {
      "epoch": 0.11130011986435648,
      "grad_norm": 1.1123387813568115,
      "learning_rate": 0.0001777401806155246,
      "loss": 3.3353,
      "step": 119400
    },
    {
      "epoch": 0.11130944148243557,
      "grad_norm": 1.0211169719696045,
      "learning_rate": 0.0001777383162934294,
      "loss": 3.6385,
      "step": 119410
    },
    {
      "epoch": 0.11131876310051465,
      "grad_norm": 1.0863428115844727,
      "learning_rate": 0.0001777364519713342,
      "loss": 3.6533,
      "step": 119420
    },
    {
      "epoch": 0.11132808471859375,
      "grad_norm": 1.1036875247955322,
      "learning_rate": 0.00017773458764923898,
      "loss": 3.6378,
      "step": 119430
    },
    {
      "epoch": 0.11133740633667284,
      "grad_norm": 1.1310999393463135,
      "learning_rate": 0.00017773272332714379,
      "loss": 3.8411,
      "step": 119440
    },
    {
      "epoch": 0.11134672795475194,
      "grad_norm": 1.0717958211898804,
      "learning_rate": 0.0001777308590050486,
      "loss": 3.5423,
      "step": 119450
    },
    {
      "epoch": 0.11135604957283103,
      "grad_norm": 1.1975492238998413,
      "learning_rate": 0.0001777289946829534,
      "loss": 3.7151,
      "step": 119460
    },
    {
      "epoch": 0.11136537119091011,
      "grad_norm": 0.9342260956764221,
      "learning_rate": 0.0001777271303608582,
      "loss": 3.6963,
      "step": 119470
    },
    {
      "epoch": 0.1113746928089892,
      "grad_norm": 0.9856353998184204,
      "learning_rate": 0.00017772526603876298,
      "loss": 3.6366,
      "step": 119480
    },
    {
      "epoch": 0.1113840144270683,
      "grad_norm": 1.179819107055664,
      "learning_rate": 0.0001777234017166678,
      "loss": 3.8202,
      "step": 119490
    },
    {
      "epoch": 0.1113933360451474,
      "grad_norm": 1.0645556449890137,
      "learning_rate": 0.0001777215373945726,
      "loss": 3.5773,
      "step": 119500
    },
    {
      "epoch": 0.11140265766322649,
      "grad_norm": 0.9768077731132507,
      "learning_rate": 0.00017771967307247738,
      "loss": 3.5112,
      "step": 119510
    },
    {
      "epoch": 0.11141197928130557,
      "grad_norm": 1.1357357501983643,
      "learning_rate": 0.0001777178087503822,
      "loss": 3.6835,
      "step": 119520
    },
    {
      "epoch": 0.11142130089938466,
      "grad_norm": 1.0701326131820679,
      "learning_rate": 0.000177715944428287,
      "loss": 3.473,
      "step": 119530
    },
    {
      "epoch": 0.11143062251746376,
      "grad_norm": 1.0724900960922241,
      "learning_rate": 0.0001777140801061918,
      "loss": 3.5043,
      "step": 119540
    },
    {
      "epoch": 0.11143994413554285,
      "grad_norm": 1.038806676864624,
      "learning_rate": 0.0001777122157840966,
      "loss": 3.9709,
      "step": 119550
    },
    {
      "epoch": 0.11144926575362195,
      "grad_norm": 1.0361428260803223,
      "learning_rate": 0.00017771035146200138,
      "loss": 3.7039,
      "step": 119560
    },
    {
      "epoch": 0.11145858737170103,
      "grad_norm": 1.0045876502990723,
      "learning_rate": 0.00017770848713990619,
      "loss": 3.6901,
      "step": 119570
    },
    {
      "epoch": 0.11146790898978012,
      "grad_norm": 1.1297478675842285,
      "learning_rate": 0.000177706622817811,
      "loss": 3.6819,
      "step": 119580
    },
    {
      "epoch": 0.11147723060785922,
      "grad_norm": 1.0201948881149292,
      "learning_rate": 0.0001777047584957158,
      "loss": 3.7105,
      "step": 119590
    },
    {
      "epoch": 0.11148655222593831,
      "grad_norm": 1.0110154151916504,
      "learning_rate": 0.0001777028941736206,
      "loss": 3.6443,
      "step": 119600
    },
    {
      "epoch": 0.1114958738440174,
      "grad_norm": 1.1242170333862305,
      "learning_rate": 0.00017770102985152538,
      "loss": 3.6134,
      "step": 119610
    },
    {
      "epoch": 0.1115051954620965,
      "grad_norm": 1.1039642095565796,
      "learning_rate": 0.0001776991655294302,
      "loss": 3.7645,
      "step": 119620
    },
    {
      "epoch": 0.11151451708017558,
      "grad_norm": 0.81739741563797,
      "learning_rate": 0.000177697301207335,
      "loss": 3.725,
      "step": 119630
    },
    {
      "epoch": 0.11152383869825468,
      "grad_norm": 1.1420725584030151,
      "learning_rate": 0.0001776954368852398,
      "loss": 3.6087,
      "step": 119640
    },
    {
      "epoch": 0.11153316031633377,
      "grad_norm": 1.1947213411331177,
      "learning_rate": 0.0001776935725631446,
      "loss": 3.7162,
      "step": 119650
    },
    {
      "epoch": 0.11154248193441287,
      "grad_norm": 1.1263854503631592,
      "learning_rate": 0.0001776917082410494,
      "loss": 3.4588,
      "step": 119660
    },
    {
      "epoch": 0.11155180355249196,
      "grad_norm": 1.0622004270553589,
      "learning_rate": 0.0001776898439189542,
      "loss": 3.6357,
      "step": 119670
    },
    {
      "epoch": 0.11156112517057104,
      "grad_norm": 1.1336312294006348,
      "learning_rate": 0.000177687979596859,
      "loss": 3.5088,
      "step": 119680
    },
    {
      "epoch": 0.11157044678865013,
      "grad_norm": 1.0593080520629883,
      "learning_rate": 0.0001776861152747638,
      "loss": 3.8074,
      "step": 119690
    },
    {
      "epoch": 0.11157976840672923,
      "grad_norm": 1.0508270263671875,
      "learning_rate": 0.00017768425095266861,
      "loss": 3.6434,
      "step": 119700
    },
    {
      "epoch": 0.11158909002480832,
      "grad_norm": 1.1898643970489502,
      "learning_rate": 0.0001776823866305734,
      "loss": 3.6636,
      "step": 119710
    },
    {
      "epoch": 0.11159841164288742,
      "grad_norm": 0.9765653610229492,
      "learning_rate": 0.0001776805223084782,
      "loss": 3.6391,
      "step": 119720
    },
    {
      "epoch": 0.1116077332609665,
      "grad_norm": 0.9343181252479553,
      "learning_rate": 0.00017767865798638298,
      "loss": 3.7179,
      "step": 119730
    },
    {
      "epoch": 0.1116170548790456,
      "grad_norm": 0.9141216278076172,
      "learning_rate": 0.0001776767936642878,
      "loss": 3.8932,
      "step": 119740
    },
    {
      "epoch": 0.11162637649712469,
      "grad_norm": 1.0206869840621948,
      "learning_rate": 0.00017767492934219262,
      "loss": 3.6544,
      "step": 119750
    },
    {
      "epoch": 0.11163569811520378,
      "grad_norm": 0.922798752784729,
      "learning_rate": 0.0001776730650200974,
      "loss": 3.577,
      "step": 119760
    },
    {
      "epoch": 0.11164501973328288,
      "grad_norm": 1.1864310503005981,
      "learning_rate": 0.0001776712006980022,
      "loss": 3.8166,
      "step": 119770
    },
    {
      "epoch": 0.11165434135136196,
      "grad_norm": 1.0095925331115723,
      "learning_rate": 0.00017766933637590698,
      "loss": 3.5085,
      "step": 119780
    },
    {
      "epoch": 0.11166366296944105,
      "grad_norm": 1.1605465412139893,
      "learning_rate": 0.0001776674720538118,
      "loss": 3.4673,
      "step": 119790
    },
    {
      "epoch": 0.11167298458752015,
      "grad_norm": 1.363682508468628,
      "learning_rate": 0.00017766560773171662,
      "loss": 3.8083,
      "step": 119800
    },
    {
      "epoch": 0.11168230620559924,
      "grad_norm": 1.2346936464309692,
      "learning_rate": 0.0001776637434096214,
      "loss": 3.3971,
      "step": 119810
    },
    {
      "epoch": 0.11169162782367834,
      "grad_norm": 1.0636872053146362,
      "learning_rate": 0.0001776618790875262,
      "loss": 3.7379,
      "step": 119820
    },
    {
      "epoch": 0.11170094944175743,
      "grad_norm": 1.0979796648025513,
      "learning_rate": 0.000177660014765431,
      "loss": 3.6366,
      "step": 119830
    },
    {
      "epoch": 0.11171027105983651,
      "grad_norm": 0.8134977221488953,
      "learning_rate": 0.0001776581504433358,
      "loss": 3.762,
      "step": 119840
    },
    {
      "epoch": 0.1117195926779156,
      "grad_norm": 1.020936369895935,
      "learning_rate": 0.0001776562861212406,
      "loss": 3.6963,
      "step": 119850
    },
    {
      "epoch": 0.1117289142959947,
      "grad_norm": 0.9642991423606873,
      "learning_rate": 0.0001776544217991454,
      "loss": 3.668,
      "step": 119860
    },
    {
      "epoch": 0.1117382359140738,
      "grad_norm": 1.0217750072479248,
      "learning_rate": 0.0001776525574770502,
      "loss": 3.6065,
      "step": 119870
    },
    {
      "epoch": 0.11174755753215289,
      "grad_norm": 1.059287667274475,
      "learning_rate": 0.000177650693154955,
      "loss": 3.6599,
      "step": 119880
    },
    {
      "epoch": 0.11175687915023197,
      "grad_norm": 0.9529492855072021,
      "learning_rate": 0.0001776488288328598,
      "loss": 3.5969,
      "step": 119890
    },
    {
      "epoch": 0.11176620076831106,
      "grad_norm": 0.9154713153839111,
      "learning_rate": 0.0001776469645107646,
      "loss": 3.7765,
      "step": 119900
    },
    {
      "epoch": 0.11177552238639016,
      "grad_norm": 0.9272634983062744,
      "learning_rate": 0.0001776451001886694,
      "loss": 3.7524,
      "step": 119910
    },
    {
      "epoch": 0.11178484400446925,
      "grad_norm": 1.0301703214645386,
      "learning_rate": 0.00017764323586657422,
      "loss": 3.6109,
      "step": 119920
    },
    {
      "epoch": 0.11179416562254835,
      "grad_norm": 1.2654811143875122,
      "learning_rate": 0.000177641371544479,
      "loss": 3.6133,
      "step": 119930
    },
    {
      "epoch": 0.11180348724062743,
      "grad_norm": 1.168119192123413,
      "learning_rate": 0.0001776395072223838,
      "loss": 3.5925,
      "step": 119940
    },
    {
      "epoch": 0.11181280885870652,
      "grad_norm": 0.9224652051925659,
      "learning_rate": 0.0001776376429002886,
      "loss": 3.6871,
      "step": 119950
    },
    {
      "epoch": 0.11182213047678562,
      "grad_norm": 1.113749623298645,
      "learning_rate": 0.0001776357785781934,
      "loss": 3.6849,
      "step": 119960
    },
    {
      "epoch": 0.11183145209486471,
      "grad_norm": 0.9008247256278992,
      "learning_rate": 0.00017763391425609822,
      "loss": 3.6911,
      "step": 119970
    },
    {
      "epoch": 0.1118407737129438,
      "grad_norm": 1.1035370826721191,
      "learning_rate": 0.000177632049934003,
      "loss": 3.5069,
      "step": 119980
    },
    {
      "epoch": 0.11185009533102289,
      "grad_norm": 0.9537213444709778,
      "learning_rate": 0.0001776301856119078,
      "loss": 3.6885,
      "step": 119990
    },
    {
      "epoch": 0.11185941694910198,
      "grad_norm": 0.9565706253051758,
      "learning_rate": 0.0001776283212898126,
      "loss": 3.6377,
      "step": 120000
    },
    {
      "epoch": 0.11186873856718108,
      "grad_norm": 1.1163122653961182,
      "learning_rate": 0.0001776264569677174,
      "loss": 3.4474,
      "step": 120010
    },
    {
      "epoch": 0.11187806018526017,
      "grad_norm": 0.8565664291381836,
      "learning_rate": 0.0001776245926456222,
      "loss": 3.7466,
      "step": 120020
    },
    {
      "epoch": 0.11188738180333926,
      "grad_norm": 0.8503704071044922,
      "learning_rate": 0.000177622728323527,
      "loss": 3.6741,
      "step": 120030
    },
    {
      "epoch": 0.11189670342141836,
      "grad_norm": 1.350874423980713,
      "learning_rate": 0.0001776208640014318,
      "loss": 3.7148,
      "step": 120040
    },
    {
      "epoch": 0.11190602503949744,
      "grad_norm": 1.2434087991714478,
      "learning_rate": 0.00017761899967933662,
      "loss": 3.605,
      "step": 120050
    },
    {
      "epoch": 0.11191534665757653,
      "grad_norm": 1.1087995767593384,
      "learning_rate": 0.0001776171353572414,
      "loss": 3.732,
      "step": 120060
    },
    {
      "epoch": 0.11192466827565563,
      "grad_norm": 0.9942482113838196,
      "learning_rate": 0.0001776152710351462,
      "loss": 3.628,
      "step": 120070
    },
    {
      "epoch": 0.11193398989373472,
      "grad_norm": 0.8626347780227661,
      "learning_rate": 0.000177613406713051,
      "loss": 3.7669,
      "step": 120080
    },
    {
      "epoch": 0.11194331151181382,
      "grad_norm": 1.0411686897277832,
      "learning_rate": 0.00017761154239095582,
      "loss": 3.5283,
      "step": 120090
    },
    {
      "epoch": 0.1119526331298929,
      "grad_norm": 1.196685791015625,
      "learning_rate": 0.00017760967806886062,
      "loss": 3.6655,
      "step": 120100
    },
    {
      "epoch": 0.11196195474797199,
      "grad_norm": 1.1292014122009277,
      "learning_rate": 0.0001776078137467654,
      "loss": 3.7252,
      "step": 120110
    },
    {
      "epoch": 0.11197127636605109,
      "grad_norm": 1.0038903951644897,
      "learning_rate": 0.0001776059494246702,
      "loss": 3.8386,
      "step": 120120
    },
    {
      "epoch": 0.11198059798413018,
      "grad_norm": 1.0249866247177124,
      "learning_rate": 0.00017760408510257501,
      "loss": 3.83,
      "step": 120130
    },
    {
      "epoch": 0.11198991960220928,
      "grad_norm": 1.1554176807403564,
      "learning_rate": 0.00017760222078047982,
      "loss": 3.7336,
      "step": 120140
    },
    {
      "epoch": 0.11199924122028836,
      "grad_norm": 1.1235525608062744,
      "learning_rate": 0.00017760035645838463,
      "loss": 3.6363,
      "step": 120150
    },
    {
      "epoch": 0.11200856283836745,
      "grad_norm": 1.1354094743728638,
      "learning_rate": 0.0001775984921362894,
      "loss": 3.7094,
      "step": 120160
    },
    {
      "epoch": 0.11201788445644655,
      "grad_norm": 1.1084892749786377,
      "learning_rate": 0.0001775966278141942,
      "loss": 3.6518,
      "step": 120170
    },
    {
      "epoch": 0.11202720607452564,
      "grad_norm": 1.0024590492248535,
      "learning_rate": 0.000177594763492099,
      "loss": 3.6029,
      "step": 120180
    },
    {
      "epoch": 0.11203652769260473,
      "grad_norm": 1.1188870668411255,
      "learning_rate": 0.00017759289917000382,
      "loss": 3.4399,
      "step": 120190
    },
    {
      "epoch": 0.11204584931068381,
      "grad_norm": 0.9389472007751465,
      "learning_rate": 0.00017759103484790863,
      "loss": 3.7524,
      "step": 120200
    },
    {
      "epoch": 0.11205517092876291,
      "grad_norm": 1.1300652027130127,
      "learning_rate": 0.0001775891705258134,
      "loss": 3.5495,
      "step": 120210
    },
    {
      "epoch": 0.112064492546842,
      "grad_norm": 0.9539579749107361,
      "learning_rate": 0.00017758730620371822,
      "loss": 3.6871,
      "step": 120220
    },
    {
      "epoch": 0.1120738141649211,
      "grad_norm": 1.186675786972046,
      "learning_rate": 0.000177585441881623,
      "loss": 3.8183,
      "step": 120230
    },
    {
      "epoch": 0.11208313578300019,
      "grad_norm": 1.0407463312149048,
      "learning_rate": 0.0001775835775595278,
      "loss": 3.3257,
      "step": 120240
    },
    {
      "epoch": 0.11209245740107927,
      "grad_norm": 1.0811030864715576,
      "learning_rate": 0.00017758171323743264,
      "loss": 3.6526,
      "step": 120250
    },
    {
      "epoch": 0.11210177901915837,
      "grad_norm": 1.0731382369995117,
      "learning_rate": 0.00017757984891533741,
      "loss": 3.5392,
      "step": 120260
    },
    {
      "epoch": 0.11211110063723746,
      "grad_norm": 1.0812848806381226,
      "learning_rate": 0.00017757798459324222,
      "loss": 3.5367,
      "step": 120270
    },
    {
      "epoch": 0.11212042225531656,
      "grad_norm": 0.9697670936584473,
      "learning_rate": 0.000177576120271147,
      "loss": 3.787,
      "step": 120280
    },
    {
      "epoch": 0.11212974387339565,
      "grad_norm": 1.3385851383209229,
      "learning_rate": 0.0001775742559490518,
      "loss": 3.8865,
      "step": 120290
    },
    {
      "epoch": 0.11213906549147475,
      "grad_norm": 1.024154543876648,
      "learning_rate": 0.0001775723916269566,
      "loss": 3.7808,
      "step": 120300
    },
    {
      "epoch": 0.11214838710955383,
      "grad_norm": 1.1690517663955688,
      "learning_rate": 0.00017757052730486142,
      "loss": 3.6069,
      "step": 120310
    },
    {
      "epoch": 0.11215770872763292,
      "grad_norm": 1.1546710729599,
      "learning_rate": 0.00017756866298276623,
      "loss": 3.8356,
      "step": 120320
    },
    {
      "epoch": 0.11216703034571202,
      "grad_norm": 1.032787799835205,
      "learning_rate": 0.000177566798660671,
      "loss": 3.6646,
      "step": 120330
    },
    {
      "epoch": 0.11217635196379111,
      "grad_norm": 0.9627138376235962,
      "learning_rate": 0.0001775649343385758,
      "loss": 3.7657,
      "step": 120340
    },
    {
      "epoch": 0.1121856735818702,
      "grad_norm": 1.051624059677124,
      "learning_rate": 0.00017756307001648062,
      "loss": 3.729,
      "step": 120350
    },
    {
      "epoch": 0.11219499519994929,
      "grad_norm": 1.2613742351531982,
      "learning_rate": 0.00017756120569438542,
      "loss": 3.69,
      "step": 120360
    },
    {
      "epoch": 0.11220431681802838,
      "grad_norm": 0.9499520659446716,
      "learning_rate": 0.00017755934137229023,
      "loss": 3.7747,
      "step": 120370
    },
    {
      "epoch": 0.11221363843610747,
      "grad_norm": 1.0223366022109985,
      "learning_rate": 0.000177557477050195,
      "loss": 3.8229,
      "step": 120380
    },
    {
      "epoch": 0.11222296005418657,
      "grad_norm": 1.035463809967041,
      "learning_rate": 0.00017755561272809982,
      "loss": 3.8968,
      "step": 120390
    },
    {
      "epoch": 0.11223228167226566,
      "grad_norm": 0.88486248254776,
      "learning_rate": 0.00017755374840600462,
      "loss": 3.4945,
      "step": 120400
    },
    {
      "epoch": 0.11224160329034474,
      "grad_norm": 1.2599238157272339,
      "learning_rate": 0.00017755188408390943,
      "loss": 3.7459,
      "step": 120410
    },
    {
      "epoch": 0.11225092490842384,
      "grad_norm": 1.059855580329895,
      "learning_rate": 0.00017755001976181423,
      "loss": 3.6909,
      "step": 120420
    },
    {
      "epoch": 0.11226024652650293,
      "grad_norm": 1.3596216440200806,
      "learning_rate": 0.000177548155439719,
      "loss": 3.4212,
      "step": 120430
    },
    {
      "epoch": 0.11226956814458203,
      "grad_norm": 0.8824187517166138,
      "learning_rate": 0.00017754629111762382,
      "loss": 3.6725,
      "step": 120440
    },
    {
      "epoch": 0.11227888976266112,
      "grad_norm": 1.1483031511306763,
      "learning_rate": 0.00017754442679552863,
      "loss": 3.6405,
      "step": 120450
    },
    {
      "epoch": 0.1122882113807402,
      "grad_norm": 1.0569539070129395,
      "learning_rate": 0.0001775425624734334,
      "loss": 3.6437,
      "step": 120460
    },
    {
      "epoch": 0.1122975329988193,
      "grad_norm": 0.9504291415214539,
      "learning_rate": 0.0001775406981513382,
      "loss": 3.3676,
      "step": 120470
    },
    {
      "epoch": 0.11230685461689839,
      "grad_norm": 1.0241851806640625,
      "learning_rate": 0.00017753883382924302,
      "loss": 3.8233,
      "step": 120480
    },
    {
      "epoch": 0.11231617623497749,
      "grad_norm": 1.3599947690963745,
      "learning_rate": 0.00017753696950714782,
      "loss": 3.6368,
      "step": 120490
    },
    {
      "epoch": 0.11232549785305658,
      "grad_norm": 1.0237592458724976,
      "learning_rate": 0.00017753510518505263,
      "loss": 3.6297,
      "step": 120500
    },
    {
      "epoch": 0.11233481947113567,
      "grad_norm": 1.2506744861602783,
      "learning_rate": 0.0001775332408629574,
      "loss": 3.6426,
      "step": 120510
    },
    {
      "epoch": 0.11234414108921476,
      "grad_norm": 1.0227282047271729,
      "learning_rate": 0.00017753137654086222,
      "loss": 3.5611,
      "step": 120520
    },
    {
      "epoch": 0.11235346270729385,
      "grad_norm": 0.965911328792572,
      "learning_rate": 0.00017752951221876702,
      "loss": 3.5394,
      "step": 120530
    },
    {
      "epoch": 0.11236278432537294,
      "grad_norm": 0.9165328741073608,
      "learning_rate": 0.00017752764789667183,
      "loss": 3.7973,
      "step": 120540
    },
    {
      "epoch": 0.11237210594345204,
      "grad_norm": 0.9980548620223999,
      "learning_rate": 0.00017752578357457663,
      "loss": 3.5447,
      "step": 120550
    },
    {
      "epoch": 0.11238142756153113,
      "grad_norm": 1.032894253730774,
      "learning_rate": 0.00017752391925248141,
      "loss": 3.5701,
      "step": 120560
    },
    {
      "epoch": 0.11239074917961021,
      "grad_norm": 1.0124887228012085,
      "learning_rate": 0.00017752205493038622,
      "loss": 3.6135,
      "step": 120570
    },
    {
      "epoch": 0.11240007079768931,
      "grad_norm": 0.8613534569740295,
      "learning_rate": 0.00017752019060829103,
      "loss": 3.9945,
      "step": 120580
    },
    {
      "epoch": 0.1124093924157684,
      "grad_norm": 1.2219760417938232,
      "learning_rate": 0.00017751832628619583,
      "loss": 3.798,
      "step": 120590
    },
    {
      "epoch": 0.1124187140338475,
      "grad_norm": 1.024473786354065,
      "learning_rate": 0.00017751646196410064,
      "loss": 3.6623,
      "step": 120600
    },
    {
      "epoch": 0.11242803565192659,
      "grad_norm": 1.0251184701919556,
      "learning_rate": 0.00017751459764200542,
      "loss": 3.649,
      "step": 120610
    },
    {
      "epoch": 0.11243735727000567,
      "grad_norm": 1.1840460300445557,
      "learning_rate": 0.00017751273331991022,
      "loss": 3.594,
      "step": 120620
    },
    {
      "epoch": 0.11244667888808477,
      "grad_norm": 1.1361709833145142,
      "learning_rate": 0.000177510868997815,
      "loss": 3.5322,
      "step": 120630
    },
    {
      "epoch": 0.11245600050616386,
      "grad_norm": 1.2233083248138428,
      "learning_rate": 0.00017750900467571984,
      "loss": 3.6501,
      "step": 120640
    },
    {
      "epoch": 0.11246532212424296,
      "grad_norm": 1.15208101272583,
      "learning_rate": 0.00017750714035362462,
      "loss": 3.5708,
      "step": 120650
    },
    {
      "epoch": 0.11247464374232205,
      "grad_norm": 1.0810860395431519,
      "learning_rate": 0.00017750527603152942,
      "loss": 3.7495,
      "step": 120660
    },
    {
      "epoch": 0.11248396536040113,
      "grad_norm": 0.9859170913696289,
      "learning_rate": 0.00017750341170943423,
      "loss": 4.0007,
      "step": 120670
    },
    {
      "epoch": 0.11249328697848023,
      "grad_norm": 1.2137891054153442,
      "learning_rate": 0.000177501547387339,
      "loss": 3.709,
      "step": 120680
    },
    {
      "epoch": 0.11250260859655932,
      "grad_norm": 1.0491905212402344,
      "learning_rate": 0.00017749968306524381,
      "loss": 3.6651,
      "step": 120690
    },
    {
      "epoch": 0.11251193021463841,
      "grad_norm": 1.0486267805099487,
      "learning_rate": 0.00017749781874314862,
      "loss": 3.5609,
      "step": 120700
    },
    {
      "epoch": 0.11252125183271751,
      "grad_norm": 1.3342289924621582,
      "learning_rate": 0.00017749595442105343,
      "loss": 3.7614,
      "step": 120710
    },
    {
      "epoch": 0.1125305734507966,
      "grad_norm": 1.0911568403244019,
      "learning_rate": 0.00017749409009895823,
      "loss": 3.8246,
      "step": 120720
    },
    {
      "epoch": 0.11253989506887568,
      "grad_norm": 0.8594099879264832,
      "learning_rate": 0.000177492225776863,
      "loss": 3.7905,
      "step": 120730
    },
    {
      "epoch": 0.11254921668695478,
      "grad_norm": 1.308627724647522,
      "learning_rate": 0.00017749036145476782,
      "loss": 3.7038,
      "step": 120740
    },
    {
      "epoch": 0.11255853830503387,
      "grad_norm": 0.6785351634025574,
      "learning_rate": 0.00017748849713267263,
      "loss": 3.6901,
      "step": 120750
    },
    {
      "epoch": 0.11256785992311297,
      "grad_norm": 1.2153934240341187,
      "learning_rate": 0.00017748663281057743,
      "loss": 3.5478,
      "step": 120760
    },
    {
      "epoch": 0.11257718154119206,
      "grad_norm": 1.0107671022415161,
      "learning_rate": 0.00017748476848848224,
      "loss": 3.6888,
      "step": 120770
    },
    {
      "epoch": 0.11258650315927114,
      "grad_norm": 0.8785475492477417,
      "learning_rate": 0.00017748290416638702,
      "loss": 3.6531,
      "step": 120780
    },
    {
      "epoch": 0.11259582477735024,
      "grad_norm": 1.0509202480316162,
      "learning_rate": 0.00017748103984429182,
      "loss": 3.7137,
      "step": 120790
    },
    {
      "epoch": 0.11260514639542933,
      "grad_norm": 1.1735652685165405,
      "learning_rate": 0.00017747917552219663,
      "loss": 3.8593,
      "step": 120800
    },
    {
      "epoch": 0.11261446801350843,
      "grad_norm": 1.0196516513824463,
      "learning_rate": 0.00017747731120010144,
      "loss": 3.7699,
      "step": 120810
    },
    {
      "epoch": 0.11262378963158752,
      "grad_norm": 0.9309793710708618,
      "learning_rate": 0.00017747544687800624,
      "loss": 3.7685,
      "step": 120820
    },
    {
      "epoch": 0.1126331112496666,
      "grad_norm": 1.167647361755371,
      "learning_rate": 0.00017747358255591102,
      "loss": 3.6318,
      "step": 120830
    },
    {
      "epoch": 0.1126424328677457,
      "grad_norm": 0.912878692150116,
      "learning_rate": 0.00017747171823381583,
      "loss": 3.7911,
      "step": 120840
    },
    {
      "epoch": 0.11265175448582479,
      "grad_norm": 1.0798465013504028,
      "learning_rate": 0.0001774698539117206,
      "loss": 3.5787,
      "step": 120850
    },
    {
      "epoch": 0.11266107610390388,
      "grad_norm": 1.0158414840698242,
      "learning_rate": 0.00017746798958962544,
      "loss": 3.8249,
      "step": 120860
    },
    {
      "epoch": 0.11267039772198298,
      "grad_norm": 1.1288909912109375,
      "learning_rate": 0.00017746612526753025,
      "loss": 3.8282,
      "step": 120870
    },
    {
      "epoch": 0.11267971934006206,
      "grad_norm": 1.0262004137039185,
      "learning_rate": 0.00017746426094543503,
      "loss": 3.6736,
      "step": 120880
    },
    {
      "epoch": 0.11268904095814115,
      "grad_norm": 1.0977163314819336,
      "learning_rate": 0.00017746239662333983,
      "loss": 3.5599,
      "step": 120890
    },
    {
      "epoch": 0.11269836257622025,
      "grad_norm": 1.0368704795837402,
      "learning_rate": 0.0001774605323012446,
      "loss": 3.7035,
      "step": 120900
    },
    {
      "epoch": 0.11270768419429934,
      "grad_norm": 0.9970522522926331,
      "learning_rate": 0.00017745866797914942,
      "loss": 3.4095,
      "step": 120910
    },
    {
      "epoch": 0.11271700581237844,
      "grad_norm": 1.1240540742874146,
      "learning_rate": 0.00017745680365705425,
      "loss": 3.5987,
      "step": 120920
    },
    {
      "epoch": 0.11272632743045752,
      "grad_norm": 1.1098068952560425,
      "learning_rate": 0.00017745493933495903,
      "loss": 3.6456,
      "step": 120930
    },
    {
      "epoch": 0.11273564904853661,
      "grad_norm": 0.952595591545105,
      "learning_rate": 0.00017745307501286384,
      "loss": 3.6864,
      "step": 120940
    },
    {
      "epoch": 0.11274497066661571,
      "grad_norm": 1.0613620281219482,
      "learning_rate": 0.00017745121069076862,
      "loss": 3.4699,
      "step": 120950
    },
    {
      "epoch": 0.1127542922846948,
      "grad_norm": 0.908732533454895,
      "learning_rate": 0.00017744934636867342,
      "loss": 3.594,
      "step": 120960
    },
    {
      "epoch": 0.1127636139027739,
      "grad_norm": 0.9313386082649231,
      "learning_rate": 0.00017744748204657823,
      "loss": 3.7689,
      "step": 120970
    },
    {
      "epoch": 0.11277293552085299,
      "grad_norm": 1.2232340574264526,
      "learning_rate": 0.00017744561772448304,
      "loss": 3.5122,
      "step": 120980
    },
    {
      "epoch": 0.11278225713893207,
      "grad_norm": 0.9036661386489868,
      "learning_rate": 0.00017744375340238784,
      "loss": 3.6604,
      "step": 120990
    },
    {
      "epoch": 0.11279157875701117,
      "grad_norm": 1.0822333097457886,
      "learning_rate": 0.00017744188908029262,
      "loss": 3.9262,
      "step": 121000
    },
    {
      "epoch": 0.11280090037509026,
      "grad_norm": 1.1356158256530762,
      "learning_rate": 0.00017744002475819743,
      "loss": 3.6134,
      "step": 121010
    },
    {
      "epoch": 0.11281022199316935,
      "grad_norm": 1.1508911848068237,
      "learning_rate": 0.00017743816043610223,
      "loss": 3.7493,
      "step": 121020
    },
    {
      "epoch": 0.11281954361124845,
      "grad_norm": 1.0939065217971802,
      "learning_rate": 0.00017743629611400704,
      "loss": 3.725,
      "step": 121030
    },
    {
      "epoch": 0.11282886522932753,
      "grad_norm": 1.07114839553833,
      "learning_rate": 0.00017743443179191185,
      "loss": 3.7337,
      "step": 121040
    },
    {
      "epoch": 0.11283818684740662,
      "grad_norm": 1.1357687711715698,
      "learning_rate": 0.00017743256746981663,
      "loss": 3.7585,
      "step": 121050
    },
    {
      "epoch": 0.11284750846548572,
      "grad_norm": 0.9063073992729187,
      "learning_rate": 0.00017743070314772143,
      "loss": 3.6625,
      "step": 121060
    },
    {
      "epoch": 0.11285683008356481,
      "grad_norm": 1.2500768899917603,
      "learning_rate": 0.00017742883882562624,
      "loss": 3.6353,
      "step": 121070
    },
    {
      "epoch": 0.11286615170164391,
      "grad_norm": 1.206268072128296,
      "learning_rate": 0.00017742697450353102,
      "loss": 3.6862,
      "step": 121080
    },
    {
      "epoch": 0.11287547331972299,
      "grad_norm": 0.973924994468689,
      "learning_rate": 0.00017742511018143585,
      "loss": 3.7726,
      "step": 121090
    },
    {
      "epoch": 0.11288479493780208,
      "grad_norm": 1.077616572380066,
      "learning_rate": 0.00017742324585934063,
      "loss": 3.7423,
      "step": 121100
    },
    {
      "epoch": 0.11289411655588118,
      "grad_norm": 1.1618536710739136,
      "learning_rate": 0.00017742138153724544,
      "loss": 3.608,
      "step": 121110
    },
    {
      "epoch": 0.11290343817396027,
      "grad_norm": 1.2058883905410767,
      "learning_rate": 0.00017741951721515024,
      "loss": 3.5021,
      "step": 121120
    },
    {
      "epoch": 0.11291275979203937,
      "grad_norm": 1.0734807252883911,
      "learning_rate": 0.00017741765289305502,
      "loss": 3.7673,
      "step": 121130
    },
    {
      "epoch": 0.11292208141011845,
      "grad_norm": 0.9207904934883118,
      "learning_rate": 0.00017741578857095983,
      "loss": 3.6468,
      "step": 121140
    },
    {
      "epoch": 0.11293140302819754,
      "grad_norm": 0.8283982276916504,
      "learning_rate": 0.00017741392424886463,
      "loss": 3.341,
      "step": 121150
    },
    {
      "epoch": 0.11294072464627664,
      "grad_norm": 1.120540738105774,
      "learning_rate": 0.00017741205992676944,
      "loss": 3.7761,
      "step": 121160
    },
    {
      "epoch": 0.11295004626435573,
      "grad_norm": 1.1215052604675293,
      "learning_rate": 0.00017741019560467425,
      "loss": 3.6977,
      "step": 121170
    },
    {
      "epoch": 0.11295936788243482,
      "grad_norm": 1.1919063329696655,
      "learning_rate": 0.00017740833128257903,
      "loss": 3.8938,
      "step": 121180
    },
    {
      "epoch": 0.11296868950051392,
      "grad_norm": 1.0950422286987305,
      "learning_rate": 0.00017740646696048383,
      "loss": 3.6707,
      "step": 121190
    },
    {
      "epoch": 0.112978011118593,
      "grad_norm": 1.1913245916366577,
      "learning_rate": 0.00017740460263838864,
      "loss": 3.5788,
      "step": 121200
    },
    {
      "epoch": 0.1129873327366721,
      "grad_norm": 1.0267786979675293,
      "learning_rate": 0.00017740273831629344,
      "loss": 3.5131,
      "step": 121210
    },
    {
      "epoch": 0.11299665435475119,
      "grad_norm": 1.0008641481399536,
      "learning_rate": 0.00017740087399419825,
      "loss": 3.8104,
      "step": 121220
    },
    {
      "epoch": 0.11300597597283028,
      "grad_norm": 0.9351385831832886,
      "learning_rate": 0.00017739900967210303,
      "loss": 3.6861,
      "step": 121230
    },
    {
      "epoch": 0.11301529759090938,
      "grad_norm": 0.9898813962936401,
      "learning_rate": 0.00017739714535000784,
      "loss": 3.857,
      "step": 121240
    },
    {
      "epoch": 0.11302461920898846,
      "grad_norm": 0.9502670764923096,
      "learning_rate": 0.00017739528102791264,
      "loss": 3.7075,
      "step": 121250
    },
    {
      "epoch": 0.11303394082706755,
      "grad_norm": 1.0094281435012817,
      "learning_rate": 0.00017739341670581745,
      "loss": 3.5813,
      "step": 121260
    },
    {
      "epoch": 0.11304326244514665,
      "grad_norm": 1.1514989137649536,
      "learning_rate": 0.00017739155238372226,
      "loss": 3.5842,
      "step": 121270
    },
    {
      "epoch": 0.11305258406322574,
      "grad_norm": 0.9403769969940186,
      "learning_rate": 0.00017738968806162703,
      "loss": 3.7158,
      "step": 121280
    },
    {
      "epoch": 0.11306190568130484,
      "grad_norm": 1.1332353353500366,
      "learning_rate": 0.00017738782373953184,
      "loss": 3.5725,
      "step": 121290
    },
    {
      "epoch": 0.11307122729938392,
      "grad_norm": 1.2627179622650146,
      "learning_rate": 0.00017738595941743662,
      "loss": 3.6405,
      "step": 121300
    },
    {
      "epoch": 0.11308054891746301,
      "grad_norm": 0.958630383014679,
      "learning_rate": 0.00017738409509534145,
      "loss": 3.6389,
      "step": 121310
    },
    {
      "epoch": 0.1130898705355421,
      "grad_norm": 1.111850619316101,
      "learning_rate": 0.00017738223077324626,
      "loss": 3.6392,
      "step": 121320
    },
    {
      "epoch": 0.1130991921536212,
      "grad_norm": 1.1398133039474487,
      "learning_rate": 0.00017738036645115104,
      "loss": 3.618,
      "step": 121330
    },
    {
      "epoch": 0.1131085137717003,
      "grad_norm": 1.004584789276123,
      "learning_rate": 0.00017737850212905585,
      "loss": 3.6823,
      "step": 121340
    },
    {
      "epoch": 0.11311783538977938,
      "grad_norm": 1.1103934049606323,
      "learning_rate": 0.00017737663780696062,
      "loss": 3.4823,
      "step": 121350
    },
    {
      "epoch": 0.11312715700785847,
      "grad_norm": 1.4702765941619873,
      "learning_rate": 0.00017737477348486543,
      "loss": 3.7149,
      "step": 121360
    },
    {
      "epoch": 0.11313647862593756,
      "grad_norm": 1.1347824335098267,
      "learning_rate": 0.00017737290916277026,
      "loss": 3.5514,
      "step": 121370
    },
    {
      "epoch": 0.11314580024401666,
      "grad_norm": 0.9165499210357666,
      "learning_rate": 0.00017737104484067504,
      "loss": 3.5223,
      "step": 121380
    },
    {
      "epoch": 0.11315512186209575,
      "grad_norm": 1.0878695249557495,
      "learning_rate": 0.00017736918051857985,
      "loss": 3.7247,
      "step": 121390
    },
    {
      "epoch": 0.11316444348017485,
      "grad_norm": 0.9900115728378296,
      "learning_rate": 0.00017736731619648463,
      "loss": 3.6357,
      "step": 121400
    },
    {
      "epoch": 0.11317376509825393,
      "grad_norm": 1.0355217456817627,
      "learning_rate": 0.00017736545187438944,
      "loss": 3.6233,
      "step": 121410
    },
    {
      "epoch": 0.11318308671633302,
      "grad_norm": 1.1265920400619507,
      "learning_rate": 0.00017736358755229424,
      "loss": 3.7491,
      "step": 121420
    },
    {
      "epoch": 0.11319240833441212,
      "grad_norm": 1.0787068605422974,
      "learning_rate": 0.00017736172323019905,
      "loss": 3.5853,
      "step": 121430
    },
    {
      "epoch": 0.11320172995249121,
      "grad_norm": 1.0767614841461182,
      "learning_rate": 0.00017735985890810385,
      "loss": 3.5918,
      "step": 121440
    },
    {
      "epoch": 0.1132110515705703,
      "grad_norm": 0.9005491137504578,
      "learning_rate": 0.00017735799458600863,
      "loss": 3.4721,
      "step": 121450
    },
    {
      "epoch": 0.11322037318864939,
      "grad_norm": 0.9501983523368835,
      "learning_rate": 0.00017735613026391344,
      "loss": 3.6794,
      "step": 121460
    },
    {
      "epoch": 0.11322969480672848,
      "grad_norm": 1.1534535884857178,
      "learning_rate": 0.00017735426594181825,
      "loss": 3.5768,
      "step": 121470
    },
    {
      "epoch": 0.11323901642480758,
      "grad_norm": 0.921315610408783,
      "learning_rate": 0.00017735240161972305,
      "loss": 3.7405,
      "step": 121480
    },
    {
      "epoch": 0.11324833804288667,
      "grad_norm": 1.0435316562652588,
      "learning_rate": 0.00017735053729762786,
      "loss": 3.5762,
      "step": 121490
    },
    {
      "epoch": 0.11325765966096576,
      "grad_norm": 1.1185482740402222,
      "learning_rate": 0.00017734867297553264,
      "loss": 3.617,
      "step": 121500
    },
    {
      "epoch": 0.11326698127904485,
      "grad_norm": 1.1433087587356567,
      "learning_rate": 0.00017734680865343744,
      "loss": 3.9248,
      "step": 121510
    },
    {
      "epoch": 0.11327630289712394,
      "grad_norm": 1.0121558904647827,
      "learning_rate": 0.00017734494433134225,
      "loss": 3.7889,
      "step": 121520
    },
    {
      "epoch": 0.11328562451520303,
      "grad_norm": 1.1197290420532227,
      "learning_rate": 0.00017734308000924703,
      "loss": 3.7382,
      "step": 121530
    },
    {
      "epoch": 0.11329494613328213,
      "grad_norm": 1.2164140939712524,
      "learning_rate": 0.00017734121568715186,
      "loss": 3.7418,
      "step": 121540
    },
    {
      "epoch": 0.11330426775136122,
      "grad_norm": 1.078539252281189,
      "learning_rate": 0.00017733935136505664,
      "loss": 3.5302,
      "step": 121550
    },
    {
      "epoch": 0.1133135893694403,
      "grad_norm": 1.1219764947891235,
      "learning_rate": 0.00017733748704296145,
      "loss": 3.5795,
      "step": 121560
    },
    {
      "epoch": 0.1133229109875194,
      "grad_norm": 0.9696370363235474,
      "learning_rate": 0.00017733562272086625,
      "loss": 3.6032,
      "step": 121570
    },
    {
      "epoch": 0.11333223260559849,
      "grad_norm": 1.0781534910202026,
      "learning_rate": 0.00017733375839877103,
      "loss": 3.5365,
      "step": 121580
    },
    {
      "epoch": 0.11334155422367759,
      "grad_norm": 1.4055894613265991,
      "learning_rate": 0.00017733189407667584,
      "loss": 3.6582,
      "step": 121590
    },
    {
      "epoch": 0.11335087584175668,
      "grad_norm": 1.0306731462478638,
      "learning_rate": 0.00017733002975458065,
      "loss": 3.7636,
      "step": 121600
    },
    {
      "epoch": 0.11336019745983576,
      "grad_norm": 1.1075211763381958,
      "learning_rate": 0.00017732816543248545,
      "loss": 3.6424,
      "step": 121610
    },
    {
      "epoch": 0.11336951907791486,
      "grad_norm": 0.9478701949119568,
      "learning_rate": 0.00017732630111039026,
      "loss": 3.6137,
      "step": 121620
    },
    {
      "epoch": 0.11337884069599395,
      "grad_norm": 1.0209954977035522,
      "learning_rate": 0.00017732443678829504,
      "loss": 3.7622,
      "step": 121630
    },
    {
      "epoch": 0.11338816231407305,
      "grad_norm": 1.2958523035049438,
      "learning_rate": 0.00017732257246619984,
      "loss": 3.7603,
      "step": 121640
    },
    {
      "epoch": 0.11339748393215214,
      "grad_norm": 1.1060583591461182,
      "learning_rate": 0.00017732070814410465,
      "loss": 3.5925,
      "step": 121650
    },
    {
      "epoch": 0.11340680555023123,
      "grad_norm": 1.0772228240966797,
      "learning_rate": 0.00017731884382200946,
      "loss": 3.6365,
      "step": 121660
    },
    {
      "epoch": 0.11341612716831032,
      "grad_norm": 1.2782115936279297,
      "learning_rate": 0.00017731697949991426,
      "loss": 3.5105,
      "step": 121670
    },
    {
      "epoch": 0.11342544878638941,
      "grad_norm": 0.9809805750846863,
      "learning_rate": 0.00017731511517781904,
      "loss": 3.7452,
      "step": 121680
    },
    {
      "epoch": 0.1134347704044685,
      "grad_norm": 0.9390636086463928,
      "learning_rate": 0.00017731325085572385,
      "loss": 3.6,
      "step": 121690
    },
    {
      "epoch": 0.1134440920225476,
      "grad_norm": 1.092944860458374,
      "learning_rate": 0.00017731138653362866,
      "loss": 3.5831,
      "step": 121700
    },
    {
      "epoch": 0.1134534136406267,
      "grad_norm": 1.1697044372558594,
      "learning_rate": 0.00017730952221153346,
      "loss": 3.6172,
      "step": 121710
    },
    {
      "epoch": 0.11346273525870577,
      "grad_norm": 1.2030867338180542,
      "learning_rate": 0.00017730765788943827,
      "loss": 3.5326,
      "step": 121720
    },
    {
      "epoch": 0.11347205687678487,
      "grad_norm": 1.0650877952575684,
      "learning_rate": 0.00017730579356734305,
      "loss": 3.718,
      "step": 121730
    },
    {
      "epoch": 0.11348137849486396,
      "grad_norm": 1.1226893663406372,
      "learning_rate": 0.00017730392924524785,
      "loss": 3.4538,
      "step": 121740
    },
    {
      "epoch": 0.11349070011294306,
      "grad_norm": 0.8737329244613647,
      "learning_rate": 0.00017730206492315263,
      "loss": 3.825,
      "step": 121750
    },
    {
      "epoch": 0.11350002173102215,
      "grad_norm": 1.156095027923584,
      "learning_rate": 0.00017730020060105747,
      "loss": 3.4963,
      "step": 121760
    },
    {
      "epoch": 0.11350934334910123,
      "grad_norm": 1.0741443634033203,
      "learning_rate": 0.00017729833627896227,
      "loss": 3.7746,
      "step": 121770
    },
    {
      "epoch": 0.11351866496718033,
      "grad_norm": 1.8330098390579224,
      "learning_rate": 0.00017729647195686705,
      "loss": 3.6403,
      "step": 121780
    },
    {
      "epoch": 0.11352798658525942,
      "grad_norm": 1.096976399421692,
      "learning_rate": 0.00017729460763477186,
      "loss": 3.6923,
      "step": 121790
    },
    {
      "epoch": 0.11353730820333852,
      "grad_norm": 1.169242024421692,
      "learning_rate": 0.00017729274331267664,
      "loss": 3.5565,
      "step": 121800
    },
    {
      "epoch": 0.11354662982141761,
      "grad_norm": 0.9207853674888611,
      "learning_rate": 0.00017729087899058144,
      "loss": 3.6264,
      "step": 121810
    },
    {
      "epoch": 0.11355595143949669,
      "grad_norm": 0.8751650452613831,
      "learning_rate": 0.00017728901466848628,
      "loss": 3.6401,
      "step": 121820
    },
    {
      "epoch": 0.11356527305757579,
      "grad_norm": 1.068266749382019,
      "learning_rate": 0.00017728715034639106,
      "loss": 3.4748,
      "step": 121830
    },
    {
      "epoch": 0.11357459467565488,
      "grad_norm": 0.9996494054794312,
      "learning_rate": 0.00017728528602429586,
      "loss": 3.705,
      "step": 121840
    },
    {
      "epoch": 0.11358391629373397,
      "grad_norm": 1.1700215339660645,
      "learning_rate": 0.00017728342170220064,
      "loss": 3.7614,
      "step": 121850
    },
    {
      "epoch": 0.11359323791181307,
      "grad_norm": 1.0707975625991821,
      "learning_rate": 0.00017728155738010545,
      "loss": 3.7634,
      "step": 121860
    },
    {
      "epoch": 0.11360255952989216,
      "grad_norm": 1.1215137243270874,
      "learning_rate": 0.00017727969305801025,
      "loss": 3.7838,
      "step": 121870
    },
    {
      "epoch": 0.11361188114797124,
      "grad_norm": 1.1850627660751343,
      "learning_rate": 0.00017727782873591506,
      "loss": 3.7784,
      "step": 121880
    },
    {
      "epoch": 0.11362120276605034,
      "grad_norm": 1.1438812017440796,
      "learning_rate": 0.00017727596441381987,
      "loss": 3.6007,
      "step": 121890
    },
    {
      "epoch": 0.11363052438412943,
      "grad_norm": 1.0542958974838257,
      "learning_rate": 0.00017727410009172465,
      "loss": 3.7086,
      "step": 121900
    },
    {
      "epoch": 0.11363984600220853,
      "grad_norm": 1.2610647678375244,
      "learning_rate": 0.00017727223576962945,
      "loss": 3.5925,
      "step": 121910
    },
    {
      "epoch": 0.11364916762028762,
      "grad_norm": 0.9862150549888611,
      "learning_rate": 0.00017727037144753426,
      "loss": 3.5873,
      "step": 121920
    },
    {
      "epoch": 0.1136584892383667,
      "grad_norm": 1.0238877534866333,
      "learning_rate": 0.00017726850712543907,
      "loss": 3.7844,
      "step": 121930
    },
    {
      "epoch": 0.1136678108564458,
      "grad_norm": 1.050734281539917,
      "learning_rate": 0.00017726664280334387,
      "loss": 3.7823,
      "step": 121940
    },
    {
      "epoch": 0.11367713247452489,
      "grad_norm": 1.1188690662384033,
      "learning_rate": 0.00017726477848124865,
      "loss": 3.7304,
      "step": 121950
    },
    {
      "epoch": 0.11368645409260399,
      "grad_norm": 1.0772804021835327,
      "learning_rate": 0.00017726291415915346,
      "loss": 3.5852,
      "step": 121960
    },
    {
      "epoch": 0.11369577571068308,
      "grad_norm": 1.2501239776611328,
      "learning_rate": 0.00017726104983705826,
      "loss": 3.6737,
      "step": 121970
    },
    {
      "epoch": 0.11370509732876216,
      "grad_norm": 1.192938208580017,
      "learning_rate": 0.00017725918551496304,
      "loss": 3.6808,
      "step": 121980
    },
    {
      "epoch": 0.11371441894684126,
      "grad_norm": 1.189283013343811,
      "learning_rate": 0.00017725732119286788,
      "loss": 3.6101,
      "step": 121990
    },
    {
      "epoch": 0.11372374056492035,
      "grad_norm": 0.9211685061454773,
      "learning_rate": 0.00017725545687077266,
      "loss": 3.8833,
      "step": 122000
    },
    {
      "epoch": 0.11373306218299944,
      "grad_norm": 1.0844208002090454,
      "learning_rate": 0.00017725359254867746,
      "loss": 3.6309,
      "step": 122010
    },
    {
      "epoch": 0.11374238380107854,
      "grad_norm": 1.0987287759780884,
      "learning_rate": 0.00017725172822658224,
      "loss": 3.702,
      "step": 122020
    },
    {
      "epoch": 0.11375170541915762,
      "grad_norm": 1.0368094444274902,
      "learning_rate": 0.00017724986390448705,
      "loss": 3.6273,
      "step": 122030
    },
    {
      "epoch": 0.11376102703723671,
      "grad_norm": 0.8849214911460876,
      "learning_rate": 0.00017724799958239185,
      "loss": 3.8142,
      "step": 122040
    },
    {
      "epoch": 0.11377034865531581,
      "grad_norm": 0.9266700744628906,
      "learning_rate": 0.00017724613526029666,
      "loss": 3.7209,
      "step": 122050
    },
    {
      "epoch": 0.1137796702733949,
      "grad_norm": 1.022447109222412,
      "learning_rate": 0.00017724427093820147,
      "loss": 3.8247,
      "step": 122060
    },
    {
      "epoch": 0.113788991891474,
      "grad_norm": 1.0886508226394653,
      "learning_rate": 0.00017724240661610625,
      "loss": 3.4507,
      "step": 122070
    },
    {
      "epoch": 0.11379831350955309,
      "grad_norm": 1.200685977935791,
      "learning_rate": 0.00017724054229401105,
      "loss": 3.5192,
      "step": 122080
    },
    {
      "epoch": 0.11380763512763217,
      "grad_norm": 0.8868340253829956,
      "learning_rate": 0.00017723867797191586,
      "loss": 3.6318,
      "step": 122090
    },
    {
      "epoch": 0.11381695674571127,
      "grad_norm": 1.003303050994873,
      "learning_rate": 0.00017723681364982066,
      "loss": 3.5355,
      "step": 122100
    },
    {
      "epoch": 0.11382627836379036,
      "grad_norm": 1.0087568759918213,
      "learning_rate": 0.00017723494932772547,
      "loss": 3.6541,
      "step": 122110
    },
    {
      "epoch": 0.11383559998186946,
      "grad_norm": 1.0747694969177246,
      "learning_rate": 0.00017723308500563025,
      "loss": 3.639,
      "step": 122120
    },
    {
      "epoch": 0.11384492159994855,
      "grad_norm": 0.9431950449943542,
      "learning_rate": 0.00017723122068353506,
      "loss": 3.7597,
      "step": 122130
    },
    {
      "epoch": 0.11385424321802763,
      "grad_norm": 1.2606905698776245,
      "learning_rate": 0.00017722935636143986,
      "loss": 3.5723,
      "step": 122140
    },
    {
      "epoch": 0.11386356483610673,
      "grad_norm": 1.0548864603042603,
      "learning_rate": 0.00017722749203934467,
      "loss": 3.6055,
      "step": 122150
    },
    {
      "epoch": 0.11387288645418582,
      "grad_norm": 1.0199366807937622,
      "learning_rate": 0.00017722562771724947,
      "loss": 3.5889,
      "step": 122160
    },
    {
      "epoch": 0.11388220807226491,
      "grad_norm": 1.2058062553405762,
      "learning_rate": 0.00017722376339515425,
      "loss": 3.698,
      "step": 122170
    },
    {
      "epoch": 0.11389152969034401,
      "grad_norm": 1.0645756721496582,
      "learning_rate": 0.00017722189907305906,
      "loss": 3.3909,
      "step": 122180
    },
    {
      "epoch": 0.11390085130842309,
      "grad_norm": 1.2412028312683105,
      "learning_rate": 0.00017722003475096387,
      "loss": 3.5571,
      "step": 122190
    },
    {
      "epoch": 0.11391017292650218,
      "grad_norm": 1.1106842756271362,
      "learning_rate": 0.00017721817042886865,
      "loss": 3.6209,
      "step": 122200
    },
    {
      "epoch": 0.11391949454458128,
      "grad_norm": 1.1653803586959839,
      "learning_rate": 0.00017721630610677348,
      "loss": 3.8421,
      "step": 122210
    },
    {
      "epoch": 0.11392881616266037,
      "grad_norm": 1.240011215209961,
      "learning_rate": 0.00017721444178467826,
      "loss": 3.782,
      "step": 122220
    },
    {
      "epoch": 0.11393813778073947,
      "grad_norm": 1.0043519735336304,
      "learning_rate": 0.00017721257746258306,
      "loss": 3.7981,
      "step": 122230
    },
    {
      "epoch": 0.11394745939881855,
      "grad_norm": 1.1065853834152222,
      "learning_rate": 0.00017721071314048787,
      "loss": 3.6506,
      "step": 122240
    },
    {
      "epoch": 0.11395678101689764,
      "grad_norm": 1.172539234161377,
      "learning_rate": 0.00017720884881839265,
      "loss": 3.5266,
      "step": 122250
    },
    {
      "epoch": 0.11396610263497674,
      "grad_norm": 1.1443768739700317,
      "learning_rate": 0.00017720698449629746,
      "loss": 3.6773,
      "step": 122260
    },
    {
      "epoch": 0.11397542425305583,
      "grad_norm": 1.2080804109573364,
      "learning_rate": 0.00017720512017420226,
      "loss": 3.5021,
      "step": 122270
    },
    {
      "epoch": 0.11398474587113493,
      "grad_norm": 1.0089828968048096,
      "learning_rate": 0.00017720325585210707,
      "loss": 3.6376,
      "step": 122280
    },
    {
      "epoch": 0.11399406748921401,
      "grad_norm": 1.0635854005813599,
      "learning_rate": 0.00017720139153001188,
      "loss": 3.7246,
      "step": 122290
    },
    {
      "epoch": 0.1140033891072931,
      "grad_norm": 1.0871444940567017,
      "learning_rate": 0.00017719952720791665,
      "loss": 3.5998,
      "step": 122300
    },
    {
      "epoch": 0.1140127107253722,
      "grad_norm": 1.1248492002487183,
      "learning_rate": 0.00017719766288582146,
      "loss": 3.5675,
      "step": 122310
    },
    {
      "epoch": 0.11402203234345129,
      "grad_norm": 0.9250293374061584,
      "learning_rate": 0.00017719579856372627,
      "loss": 3.6941,
      "step": 122320
    },
    {
      "epoch": 0.11403135396153039,
      "grad_norm": 0.9215962290763855,
      "learning_rate": 0.00017719393424163107,
      "loss": 3.5502,
      "step": 122330
    },
    {
      "epoch": 0.11404067557960948,
      "grad_norm": 1.1721373796463013,
      "learning_rate": 0.00017719206991953588,
      "loss": 3.7687,
      "step": 122340
    },
    {
      "epoch": 0.11404999719768856,
      "grad_norm": 1.1555677652359009,
      "learning_rate": 0.00017719020559744066,
      "loss": 3.5991,
      "step": 122350
    },
    {
      "epoch": 0.11405931881576765,
      "grad_norm": 1.068703532218933,
      "learning_rate": 0.00017718834127534547,
      "loss": 3.6601,
      "step": 122360
    },
    {
      "epoch": 0.11406864043384675,
      "grad_norm": 1.040743350982666,
      "learning_rate": 0.00017718647695325024,
      "loss": 3.6123,
      "step": 122370
    },
    {
      "epoch": 0.11407796205192584,
      "grad_norm": 1.1303001642227173,
      "learning_rate": 0.00017718461263115508,
      "loss": 3.6301,
      "step": 122380
    },
    {
      "epoch": 0.11408728367000494,
      "grad_norm": 1.088417410850525,
      "learning_rate": 0.00017718274830905988,
      "loss": 3.6181,
      "step": 122390
    },
    {
      "epoch": 0.11409660528808402,
      "grad_norm": 1.189104676246643,
      "learning_rate": 0.00017718088398696466,
      "loss": 3.6117,
      "step": 122400
    },
    {
      "epoch": 0.11410592690616311,
      "grad_norm": 1.13886296749115,
      "learning_rate": 0.00017717901966486947,
      "loss": 3.7537,
      "step": 122410
    },
    {
      "epoch": 0.11411524852424221,
      "grad_norm": 1.117404818534851,
      "learning_rate": 0.00017717715534277425,
      "loss": 3.6601,
      "step": 122420
    },
    {
      "epoch": 0.1141245701423213,
      "grad_norm": 1.0794905424118042,
      "learning_rate": 0.00017717529102067906,
      "loss": 3.6682,
      "step": 122430
    },
    {
      "epoch": 0.1141338917604004,
      "grad_norm": 1.0466902256011963,
      "learning_rate": 0.0001771734266985839,
      "loss": 3.6705,
      "step": 122440
    },
    {
      "epoch": 0.11414321337847948,
      "grad_norm": 1.2219938039779663,
      "learning_rate": 0.00017717156237648867,
      "loss": 3.5857,
      "step": 122450
    },
    {
      "epoch": 0.11415253499655857,
      "grad_norm": 1.0646096467971802,
      "learning_rate": 0.00017716969805439347,
      "loss": 3.6399,
      "step": 122460
    },
    {
      "epoch": 0.11416185661463767,
      "grad_norm": 0.9548624157905579,
      "learning_rate": 0.00017716783373229825,
      "loss": 3.6119,
      "step": 122470
    },
    {
      "epoch": 0.11417117823271676,
      "grad_norm": 1.1166690587997437,
      "learning_rate": 0.00017716596941020306,
      "loss": 3.7073,
      "step": 122480
    },
    {
      "epoch": 0.11418049985079586,
      "grad_norm": 0.9473145604133606,
      "learning_rate": 0.00017716410508810787,
      "loss": 3.7128,
      "step": 122490
    },
    {
      "epoch": 0.11418982146887494,
      "grad_norm": 0.956308901309967,
      "learning_rate": 0.00017716224076601267,
      "loss": 3.8744,
      "step": 122500
    },
    {
      "epoch": 0.11419914308695403,
      "grad_norm": 0.8745516538619995,
      "learning_rate": 0.00017716037644391748,
      "loss": 3.6685,
      "step": 122510
    },
    {
      "epoch": 0.11420846470503312,
      "grad_norm": 1.1995073556900024,
      "learning_rate": 0.00017715851212182226,
      "loss": 3.6705,
      "step": 122520
    },
    {
      "epoch": 0.11421778632311222,
      "grad_norm": 1.0440810918807983,
      "learning_rate": 0.00017715664779972706,
      "loss": 3.8894,
      "step": 122530
    },
    {
      "epoch": 0.11422710794119131,
      "grad_norm": 1.0919564962387085,
      "learning_rate": 0.00017715478347763187,
      "loss": 3.7682,
      "step": 122540
    },
    {
      "epoch": 0.11423642955927041,
      "grad_norm": 1.075649619102478,
      "learning_rate": 0.00017715291915553668,
      "loss": 3.5342,
      "step": 122550
    },
    {
      "epoch": 0.11424575117734949,
      "grad_norm": 1.1685404777526855,
      "learning_rate": 0.00017715105483344148,
      "loss": 3.691,
      "step": 122560
    },
    {
      "epoch": 0.11425507279542858,
      "grad_norm": 1.1706732511520386,
      "learning_rate": 0.00017714919051134626,
      "loss": 3.6402,
      "step": 122570
    },
    {
      "epoch": 0.11426439441350768,
      "grad_norm": 1.1556551456451416,
      "learning_rate": 0.00017714732618925107,
      "loss": 3.7836,
      "step": 122580
    },
    {
      "epoch": 0.11427371603158677,
      "grad_norm": 0.8470343947410583,
      "learning_rate": 0.00017714546186715588,
      "loss": 3.7365,
      "step": 122590
    },
    {
      "epoch": 0.11428303764966587,
      "grad_norm": 1.0251801013946533,
      "learning_rate": 0.00017714359754506068,
      "loss": 3.72,
      "step": 122600
    },
    {
      "epoch": 0.11429235926774495,
      "grad_norm": 1.0712684392929077,
      "learning_rate": 0.0001771417332229655,
      "loss": 3.7446,
      "step": 122610
    },
    {
      "epoch": 0.11430168088582404,
      "grad_norm": 1.2997980117797852,
      "learning_rate": 0.00017713986890087027,
      "loss": 3.703,
      "step": 122620
    },
    {
      "epoch": 0.11431100250390314,
      "grad_norm": 1.042901635169983,
      "learning_rate": 0.00017713800457877507,
      "loss": 3.3654,
      "step": 122630
    },
    {
      "epoch": 0.11432032412198223,
      "grad_norm": 0.9261505007743835,
      "learning_rate": 0.00017713614025667988,
      "loss": 3.655,
      "step": 122640
    },
    {
      "epoch": 0.11432964574006133,
      "grad_norm": 0.9607062935829163,
      "learning_rate": 0.00017713427593458466,
      "loss": 3.5161,
      "step": 122650
    },
    {
      "epoch": 0.1143389673581404,
      "grad_norm": 1.045342206954956,
      "learning_rate": 0.0001771324116124895,
      "loss": 3.6184,
      "step": 122660
    },
    {
      "epoch": 0.1143482889762195,
      "grad_norm": 1.0983949899673462,
      "learning_rate": 0.00017713054729039427,
      "loss": 3.6156,
      "step": 122670
    },
    {
      "epoch": 0.1143576105942986,
      "grad_norm": 0.9301644563674927,
      "learning_rate": 0.00017712868296829908,
      "loss": 3.5064,
      "step": 122680
    },
    {
      "epoch": 0.11436693221237769,
      "grad_norm": 1.2537130117416382,
      "learning_rate": 0.00017712681864620388,
      "loss": 3.5548,
      "step": 122690
    },
    {
      "epoch": 0.11437625383045678,
      "grad_norm": 0.965533435344696,
      "learning_rate": 0.00017712495432410866,
      "loss": 3.8823,
      "step": 122700
    },
    {
      "epoch": 0.11438557544853586,
      "grad_norm": 0.9268821477890015,
      "learning_rate": 0.00017712309000201347,
      "loss": 3.5389,
      "step": 122710
    },
    {
      "epoch": 0.11439489706661496,
      "grad_norm": 1.352652907371521,
      "learning_rate": 0.00017712122567991828,
      "loss": 3.8066,
      "step": 122720
    },
    {
      "epoch": 0.11440421868469405,
      "grad_norm": 0.8958636522293091,
      "learning_rate": 0.00017711936135782308,
      "loss": 3.7827,
      "step": 122730
    },
    {
      "epoch": 0.11441354030277315,
      "grad_norm": 0.862438976764679,
      "learning_rate": 0.0001771174970357279,
      "loss": 3.6232,
      "step": 122740
    },
    {
      "epoch": 0.11442286192085224,
      "grad_norm": 1.0899770259857178,
      "learning_rate": 0.00017711563271363267,
      "loss": 3.6375,
      "step": 122750
    },
    {
      "epoch": 0.11443218353893134,
      "grad_norm": 0.7965469360351562,
      "learning_rate": 0.00017711376839153747,
      "loss": 3.6933,
      "step": 122760
    },
    {
      "epoch": 0.11444150515701042,
      "grad_norm": 0.9955074191093445,
      "learning_rate": 0.00017711190406944228,
      "loss": 3.697,
      "step": 122770
    },
    {
      "epoch": 0.11445082677508951,
      "grad_norm": 1.035031795501709,
      "learning_rate": 0.0001771100397473471,
      "loss": 3.784,
      "step": 122780
    },
    {
      "epoch": 0.1144601483931686,
      "grad_norm": 1.0415520668029785,
      "learning_rate": 0.0001771081754252519,
      "loss": 3.6187,
      "step": 122790
    },
    {
      "epoch": 0.1144694700112477,
      "grad_norm": 1.0592292547225952,
      "learning_rate": 0.00017710631110315667,
      "loss": 3.6526,
      "step": 122800
    },
    {
      "epoch": 0.1144787916293268,
      "grad_norm": 1.1080695390701294,
      "learning_rate": 0.00017710444678106148,
      "loss": 3.6503,
      "step": 122810
    },
    {
      "epoch": 0.11448811324740588,
      "grad_norm": 1.3210262060165405,
      "learning_rate": 0.00017710258245896626,
      "loss": 3.7525,
      "step": 122820
    },
    {
      "epoch": 0.11449743486548497,
      "grad_norm": 1.1816273927688599,
      "learning_rate": 0.0001771007181368711,
      "loss": 3.8266,
      "step": 122830
    },
    {
      "epoch": 0.11450675648356406,
      "grad_norm": 1.2038217782974243,
      "learning_rate": 0.0001770988538147759,
      "loss": 3.7768,
      "step": 122840
    },
    {
      "epoch": 0.11451607810164316,
      "grad_norm": 1.1628845930099487,
      "learning_rate": 0.00017709698949268068,
      "loss": 3.6669,
      "step": 122850
    },
    {
      "epoch": 0.11452539971972225,
      "grad_norm": 0.9242212772369385,
      "learning_rate": 0.00017709512517058548,
      "loss": 3.8313,
      "step": 122860
    },
    {
      "epoch": 0.11453472133780133,
      "grad_norm": 0.9726296067237854,
      "learning_rate": 0.00017709326084849026,
      "loss": 3.7752,
      "step": 122870
    },
    {
      "epoch": 0.11454404295588043,
      "grad_norm": 1.1651225090026855,
      "learning_rate": 0.00017709139652639507,
      "loss": 3.8265,
      "step": 122880
    },
    {
      "epoch": 0.11455336457395952,
      "grad_norm": 0.8563705086708069,
      "learning_rate": 0.0001770895322042999,
      "loss": 3.5791,
      "step": 122890
    },
    {
      "epoch": 0.11456268619203862,
      "grad_norm": 1.0203535556793213,
      "learning_rate": 0.00017708766788220468,
      "loss": 3.6794,
      "step": 122900
    },
    {
      "epoch": 0.11457200781011771,
      "grad_norm": 0.9257389903068542,
      "learning_rate": 0.0001770858035601095,
      "loss": 3.6534,
      "step": 122910
    },
    {
      "epoch": 0.1145813294281968,
      "grad_norm": 0.9302429556846619,
      "learning_rate": 0.00017708393923801427,
      "loss": 3.6696,
      "step": 122920
    },
    {
      "epoch": 0.11459065104627589,
      "grad_norm": 1.121915340423584,
      "learning_rate": 0.00017708207491591907,
      "loss": 3.6467,
      "step": 122930
    },
    {
      "epoch": 0.11459997266435498,
      "grad_norm": 0.9640955924987793,
      "learning_rate": 0.00017708021059382388,
      "loss": 3.5596,
      "step": 122940
    },
    {
      "epoch": 0.11460929428243408,
      "grad_norm": 1.1484683752059937,
      "learning_rate": 0.00017707834627172869,
      "loss": 3.2956,
      "step": 122950
    },
    {
      "epoch": 0.11461861590051317,
      "grad_norm": 0.9944047331809998,
      "learning_rate": 0.0001770764819496335,
      "loss": 3.6253,
      "step": 122960
    },
    {
      "epoch": 0.11462793751859225,
      "grad_norm": 1.0678496360778809,
      "learning_rate": 0.00017707461762753827,
      "loss": 3.6019,
      "step": 122970
    },
    {
      "epoch": 0.11463725913667135,
      "grad_norm": 1.0116517543792725,
      "learning_rate": 0.00017707275330544308,
      "loss": 3.8347,
      "step": 122980
    },
    {
      "epoch": 0.11464658075475044,
      "grad_norm": 1.083202838897705,
      "learning_rate": 0.00017707088898334788,
      "loss": 3.6513,
      "step": 122990
    },
    {
      "epoch": 0.11465590237282954,
      "grad_norm": 1.0526137351989746,
      "learning_rate": 0.0001770690246612527,
      "loss": 3.6198,
      "step": 123000
    },
    {
      "epoch": 0.11466522399090863,
      "grad_norm": 0.9872081279754639,
      "learning_rate": 0.0001770671603391575,
      "loss": 3.4852,
      "step": 123010
    },
    {
      "epoch": 0.11467454560898772,
      "grad_norm": 1.0116138458251953,
      "learning_rate": 0.00017706529601706228,
      "loss": 3.6808,
      "step": 123020
    },
    {
      "epoch": 0.1146838672270668,
      "grad_norm": 1.080480933189392,
      "learning_rate": 0.00017706343169496708,
      "loss": 3.6082,
      "step": 123030
    },
    {
      "epoch": 0.1146931888451459,
      "grad_norm": 1.3910987377166748,
      "learning_rate": 0.0001770615673728719,
      "loss": 3.5054,
      "step": 123040
    },
    {
      "epoch": 0.114702510463225,
      "grad_norm": 1.3679906129837036,
      "learning_rate": 0.0001770597030507767,
      "loss": 3.5924,
      "step": 123050
    },
    {
      "epoch": 0.11471183208130409,
      "grad_norm": 1.1645392179489136,
      "learning_rate": 0.0001770578387286815,
      "loss": 3.6693,
      "step": 123060
    },
    {
      "epoch": 0.11472115369938318,
      "grad_norm": 0.9869576096534729,
      "learning_rate": 0.00017705597440658628,
      "loss": 3.8274,
      "step": 123070
    },
    {
      "epoch": 0.11473047531746226,
      "grad_norm": 0.839231550693512,
      "learning_rate": 0.00017705411008449109,
      "loss": 3.8145,
      "step": 123080
    },
    {
      "epoch": 0.11473979693554136,
      "grad_norm": 0.9601564407348633,
      "learning_rate": 0.0001770522457623959,
      "loss": 3.7963,
      "step": 123090
    },
    {
      "epoch": 0.11474911855362045,
      "grad_norm": 1.0916922092437744,
      "learning_rate": 0.00017705038144030067,
      "loss": 3.7396,
      "step": 123100
    },
    {
      "epoch": 0.11475844017169955,
      "grad_norm": 0.9967746138572693,
      "learning_rate": 0.0001770485171182055,
      "loss": 3.7561,
      "step": 123110
    },
    {
      "epoch": 0.11476776178977864,
      "grad_norm": 1.127742052078247,
      "learning_rate": 0.00017704665279611028,
      "loss": 3.654,
      "step": 123120
    },
    {
      "epoch": 0.11477708340785772,
      "grad_norm": 1.1237831115722656,
      "learning_rate": 0.0001770447884740151,
      "loss": 3.6873,
      "step": 123130
    },
    {
      "epoch": 0.11478640502593682,
      "grad_norm": 1.0074865818023682,
      "learning_rate": 0.0001770429241519199,
      "loss": 3.6133,
      "step": 123140
    },
    {
      "epoch": 0.11479572664401591,
      "grad_norm": 1.0981065034866333,
      "learning_rate": 0.00017704105982982468,
      "loss": 3.6543,
      "step": 123150
    },
    {
      "epoch": 0.114805048262095,
      "grad_norm": 1.1119083166122437,
      "learning_rate": 0.00017703919550772948,
      "loss": 3.7405,
      "step": 123160
    },
    {
      "epoch": 0.1148143698801741,
      "grad_norm": 1.1551157236099243,
      "learning_rate": 0.0001770373311856343,
      "loss": 3.5431,
      "step": 123170
    },
    {
      "epoch": 0.11482369149825318,
      "grad_norm": 1.0504928827285767,
      "learning_rate": 0.0001770354668635391,
      "loss": 3.6609,
      "step": 123180
    },
    {
      "epoch": 0.11483301311633227,
      "grad_norm": 1.2004433870315552,
      "learning_rate": 0.0001770336025414439,
      "loss": 3.7346,
      "step": 123190
    },
    {
      "epoch": 0.11484233473441137,
      "grad_norm": 1.1507055759429932,
      "learning_rate": 0.00017703173821934868,
      "loss": 3.6264,
      "step": 123200
    },
    {
      "epoch": 0.11485165635249046,
      "grad_norm": 1.168506145477295,
      "learning_rate": 0.0001770298738972535,
      "loss": 3.6143,
      "step": 123210
    },
    {
      "epoch": 0.11486097797056956,
      "grad_norm": 0.9833765029907227,
      "learning_rate": 0.0001770280095751583,
      "loss": 3.5585,
      "step": 123220
    },
    {
      "epoch": 0.11487029958864865,
      "grad_norm": 1.0143961906433105,
      "learning_rate": 0.0001770261452530631,
      "loss": 3.7748,
      "step": 123230
    },
    {
      "epoch": 0.11487962120672773,
      "grad_norm": 1.0447996854782104,
      "learning_rate": 0.0001770242809309679,
      "loss": 3.733,
      "step": 123240
    },
    {
      "epoch": 0.11488894282480683,
      "grad_norm": 0.9838758111000061,
      "learning_rate": 0.00017702241660887268,
      "loss": 3.6059,
      "step": 123250
    },
    {
      "epoch": 0.11489826444288592,
      "grad_norm": 1.1228753328323364,
      "learning_rate": 0.0001770205522867775,
      "loss": 3.7908,
      "step": 123260
    },
    {
      "epoch": 0.11490758606096502,
      "grad_norm": 1.443495512008667,
      "learning_rate": 0.00017701868796468227,
      "loss": 3.5717,
      "step": 123270
    },
    {
      "epoch": 0.11491690767904411,
      "grad_norm": 1.3034354448318481,
      "learning_rate": 0.0001770168236425871,
      "loss": 3.4665,
      "step": 123280
    },
    {
      "epoch": 0.11492622929712319,
      "grad_norm": 1.0738680362701416,
      "learning_rate": 0.0001770149593204919,
      "loss": 3.6895,
      "step": 123290
    },
    {
      "epoch": 0.11493555091520229,
      "grad_norm": 1.025592565536499,
      "learning_rate": 0.0001770130949983967,
      "loss": 3.5408,
      "step": 123300
    },
    {
      "epoch": 0.11494487253328138,
      "grad_norm": 1.0832743644714355,
      "learning_rate": 0.0001770112306763015,
      "loss": 3.4796,
      "step": 123310
    },
    {
      "epoch": 0.11495419415136048,
      "grad_norm": 1.3356819152832031,
      "learning_rate": 0.00017700936635420627,
      "loss": 3.8175,
      "step": 123320
    },
    {
      "epoch": 0.11496351576943957,
      "grad_norm": 1.0573863983154297,
      "learning_rate": 0.00017700750203211108,
      "loss": 3.6629,
      "step": 123330
    },
    {
      "epoch": 0.11497283738751865,
      "grad_norm": 0.9683414101600647,
      "learning_rate": 0.00017700563771001591,
      "loss": 3.7811,
      "step": 123340
    },
    {
      "epoch": 0.11498215900559774,
      "grad_norm": 1.0752155780792236,
      "learning_rate": 0.0001770037733879207,
      "loss": 3.7028,
      "step": 123350
    },
    {
      "epoch": 0.11499148062367684,
      "grad_norm": 1.0683881044387817,
      "learning_rate": 0.0001770019090658255,
      "loss": 3.7066,
      "step": 123360
    },
    {
      "epoch": 0.11500080224175593,
      "grad_norm": 0.8657141327857971,
      "learning_rate": 0.00017700004474373028,
      "loss": 3.6779,
      "step": 123370
    },
    {
      "epoch": 0.11501012385983503,
      "grad_norm": 1.1977752447128296,
      "learning_rate": 0.00017699818042163509,
      "loss": 3.6243,
      "step": 123380
    },
    {
      "epoch": 0.11501944547791411,
      "grad_norm": 1.0898131132125854,
      "learning_rate": 0.0001769963160995399,
      "loss": 3.6109,
      "step": 123390
    },
    {
      "epoch": 0.1150287670959932,
      "grad_norm": 0.9200507402420044,
      "learning_rate": 0.0001769944517774447,
      "loss": 3.5082,
      "step": 123400
    },
    {
      "epoch": 0.1150380887140723,
      "grad_norm": 0.998116672039032,
      "learning_rate": 0.0001769925874553495,
      "loss": 3.5156,
      "step": 123410
    },
    {
      "epoch": 0.11504741033215139,
      "grad_norm": 1.146989345550537,
      "learning_rate": 0.00017699072313325428,
      "loss": 3.7201,
      "step": 123420
    },
    {
      "epoch": 0.11505673195023049,
      "grad_norm": 1.1653461456298828,
      "learning_rate": 0.0001769888588111591,
      "loss": 3.6362,
      "step": 123430
    },
    {
      "epoch": 0.11506605356830958,
      "grad_norm": 0.8708054423332214,
      "learning_rate": 0.0001769869944890639,
      "loss": 3.6559,
      "step": 123440
    },
    {
      "epoch": 0.11507537518638866,
      "grad_norm": 1.0908997058868408,
      "learning_rate": 0.0001769851301669687,
      "loss": 3.5069,
      "step": 123450
    },
    {
      "epoch": 0.11508469680446776,
      "grad_norm": 0.9306468367576599,
      "learning_rate": 0.0001769832658448735,
      "loss": 3.6124,
      "step": 123460
    },
    {
      "epoch": 0.11509401842254685,
      "grad_norm": 1.1111164093017578,
      "learning_rate": 0.0001769814015227783,
      "loss": 3.3988,
      "step": 123470
    },
    {
      "epoch": 0.11510334004062595,
      "grad_norm": 0.9354113936424255,
      "learning_rate": 0.0001769795372006831,
      "loss": 3.8746,
      "step": 123480
    },
    {
      "epoch": 0.11511266165870504,
      "grad_norm": 1.2443610429763794,
      "learning_rate": 0.00017697767287858787,
      "loss": 3.5634,
      "step": 123490
    },
    {
      "epoch": 0.11512198327678412,
      "grad_norm": 1.0050212144851685,
      "learning_rate": 0.0001769758085564927,
      "loss": 3.6109,
      "step": 123500
    },
    {
      "epoch": 0.11513130489486322,
      "grad_norm": 1.2044557332992554,
      "learning_rate": 0.0001769739442343975,
      "loss": 3.7422,
      "step": 123510
    },
    {
      "epoch": 0.11514062651294231,
      "grad_norm": 1.1970174312591553,
      "learning_rate": 0.0001769720799123023,
      "loss": 3.6236,
      "step": 123520
    },
    {
      "epoch": 0.1151499481310214,
      "grad_norm": 1.1041823625564575,
      "learning_rate": 0.0001769702155902071,
      "loss": 3.7585,
      "step": 123530
    },
    {
      "epoch": 0.1151592697491005,
      "grad_norm": 0.9362062215805054,
      "learning_rate": 0.00017696835126811188,
      "loss": 3.8786,
      "step": 123540
    },
    {
      "epoch": 0.11516859136717958,
      "grad_norm": 1.1462000608444214,
      "learning_rate": 0.00017696648694601668,
      "loss": 3.5456,
      "step": 123550
    },
    {
      "epoch": 0.11517791298525867,
      "grad_norm": 1.034037709236145,
      "learning_rate": 0.00017696462262392152,
      "loss": 3.6872,
      "step": 123560
    },
    {
      "epoch": 0.11518723460333777,
      "grad_norm": 0.8526687026023865,
      "learning_rate": 0.0001769627583018263,
      "loss": 3.7364,
      "step": 123570
    },
    {
      "epoch": 0.11519655622141686,
      "grad_norm": 1.1075325012207031,
      "learning_rate": 0.0001769608939797311,
      "loss": 3.4333,
      "step": 123580
    },
    {
      "epoch": 0.11520587783949596,
      "grad_norm": 1.002116084098816,
      "learning_rate": 0.00017695902965763588,
      "loss": 3.8249,
      "step": 123590
    },
    {
      "epoch": 0.11521519945757504,
      "grad_norm": 0.9517916440963745,
      "learning_rate": 0.0001769571653355407,
      "loss": 3.6259,
      "step": 123600
    },
    {
      "epoch": 0.11522452107565413,
      "grad_norm": 0.7786839008331299,
      "learning_rate": 0.0001769553010134455,
      "loss": 3.6144,
      "step": 123610
    },
    {
      "epoch": 0.11523384269373323,
      "grad_norm": 1.1201950311660767,
      "learning_rate": 0.0001769534366913503,
      "loss": 3.7889,
      "step": 123620
    },
    {
      "epoch": 0.11524316431181232,
      "grad_norm": 0.9272713661193848,
      "learning_rate": 0.0001769515723692551,
      "loss": 3.7246,
      "step": 123630
    },
    {
      "epoch": 0.11525248592989142,
      "grad_norm": 0.9626907110214233,
      "learning_rate": 0.0001769497080471599,
      "loss": 3.6666,
      "step": 123640
    },
    {
      "epoch": 0.1152618075479705,
      "grad_norm": 1.0732245445251465,
      "learning_rate": 0.0001769478437250647,
      "loss": 3.7139,
      "step": 123650
    },
    {
      "epoch": 0.11527112916604959,
      "grad_norm": 1.0156911611557007,
      "learning_rate": 0.0001769459794029695,
      "loss": 3.4478,
      "step": 123660
    },
    {
      "epoch": 0.11528045078412869,
      "grad_norm": 1.209210991859436,
      "learning_rate": 0.0001769441150808743,
      "loss": 3.8407,
      "step": 123670
    },
    {
      "epoch": 0.11528977240220778,
      "grad_norm": 0.9736602306365967,
      "learning_rate": 0.0001769422507587791,
      "loss": 3.7084,
      "step": 123680
    },
    {
      "epoch": 0.11529909402028687,
      "grad_norm": 1.1838068962097168,
      "learning_rate": 0.0001769403864366839,
      "loss": 3.6395,
      "step": 123690
    },
    {
      "epoch": 0.11530841563836597,
      "grad_norm": 1.0820971727371216,
      "learning_rate": 0.0001769385221145887,
      "loss": 3.3374,
      "step": 123700
    },
    {
      "epoch": 0.11531773725644505,
      "grad_norm": 0.6986299157142639,
      "learning_rate": 0.0001769366577924935,
      "loss": 3.6477,
      "step": 123710
    },
    {
      "epoch": 0.11532705887452414,
      "grad_norm": 0.9697808027267456,
      "learning_rate": 0.0001769347934703983,
      "loss": 3.6905,
      "step": 123720
    },
    {
      "epoch": 0.11533638049260324,
      "grad_norm": 1.20883047580719,
      "learning_rate": 0.00017693292914830312,
      "loss": 3.5969,
      "step": 123730
    },
    {
      "epoch": 0.11534570211068233,
      "grad_norm": 0.864802360534668,
      "learning_rate": 0.0001769310648262079,
      "loss": 3.7923,
      "step": 123740
    },
    {
      "epoch": 0.11535502372876143,
      "grad_norm": 1.1592224836349487,
      "learning_rate": 0.0001769292005041127,
      "loss": 3.6002,
      "step": 123750
    },
    {
      "epoch": 0.11536434534684051,
      "grad_norm": 1.2075163125991821,
      "learning_rate": 0.0001769273361820175,
      "loss": 3.5649,
      "step": 123760
    },
    {
      "epoch": 0.1153736669649196,
      "grad_norm": 0.9926464557647705,
      "learning_rate": 0.0001769254718599223,
      "loss": 3.5467,
      "step": 123770
    },
    {
      "epoch": 0.1153829885829987,
      "grad_norm": 1.0255579948425293,
      "learning_rate": 0.0001769236075378271,
      "loss": 3.6019,
      "step": 123780
    },
    {
      "epoch": 0.11539231020107779,
      "grad_norm": 0.9344084858894348,
      "learning_rate": 0.0001769217432157319,
      "loss": 3.7439,
      "step": 123790
    },
    {
      "epoch": 0.11540163181915689,
      "grad_norm": 1.3683109283447266,
      "learning_rate": 0.0001769198788936367,
      "loss": 3.8564,
      "step": 123800
    },
    {
      "epoch": 0.11541095343723597,
      "grad_norm": 1.2606834173202515,
      "learning_rate": 0.0001769180145715415,
      "loss": 3.623,
      "step": 123810
    },
    {
      "epoch": 0.11542027505531506,
      "grad_norm": 0.9716553688049316,
      "learning_rate": 0.0001769161502494463,
      "loss": 3.7065,
      "step": 123820
    },
    {
      "epoch": 0.11542959667339416,
      "grad_norm": 1.0542596578598022,
      "learning_rate": 0.0001769142859273511,
      "loss": 3.5978,
      "step": 123830
    },
    {
      "epoch": 0.11543891829147325,
      "grad_norm": 1.0831421613693237,
      "learning_rate": 0.0001769124216052559,
      "loss": 3.5156,
      "step": 123840
    },
    {
      "epoch": 0.11544823990955234,
      "grad_norm": 1.06266188621521,
      "learning_rate": 0.0001769105572831607,
      "loss": 3.488,
      "step": 123850
    },
    {
      "epoch": 0.11545756152763142,
      "grad_norm": 1.1333708763122559,
      "learning_rate": 0.00017690869296106552,
      "loss": 3.6968,
      "step": 123860
    },
    {
      "epoch": 0.11546688314571052,
      "grad_norm": 1.1452970504760742,
      "learning_rate": 0.0001769068286389703,
      "loss": 3.6747,
      "step": 123870
    },
    {
      "epoch": 0.11547620476378961,
      "grad_norm": 1.2238737344741821,
      "learning_rate": 0.0001769049643168751,
      "loss": 3.6505,
      "step": 123880
    },
    {
      "epoch": 0.11548552638186871,
      "grad_norm": 1.001469612121582,
      "learning_rate": 0.0001769030999947799,
      "loss": 3.6272,
      "step": 123890
    },
    {
      "epoch": 0.1154948479999478,
      "grad_norm": 1.3174359798431396,
      "learning_rate": 0.00017690123567268472,
      "loss": 3.5752,
      "step": 123900
    },
    {
      "epoch": 0.1155041696180269,
      "grad_norm": 0.9835289120674133,
      "learning_rate": 0.00017689937135058952,
      "loss": 3.6776,
      "step": 123910
    },
    {
      "epoch": 0.11551349123610598,
      "grad_norm": 0.8116448521614075,
      "learning_rate": 0.0001768975070284943,
      "loss": 3.5942,
      "step": 123920
    },
    {
      "epoch": 0.11552281285418507,
      "grad_norm": 1.1729671955108643,
      "learning_rate": 0.0001768956427063991,
      "loss": 3.5207,
      "step": 123930
    },
    {
      "epoch": 0.11553213447226417,
      "grad_norm": 1.1429563760757446,
      "learning_rate": 0.0001768937783843039,
      "loss": 3.5035,
      "step": 123940
    },
    {
      "epoch": 0.11554145609034326,
      "grad_norm": 1.0854487419128418,
      "learning_rate": 0.00017689191406220872,
      "loss": 3.5325,
      "step": 123950
    },
    {
      "epoch": 0.11555077770842236,
      "grad_norm": 1.1960309743881226,
      "learning_rate": 0.00017689004974011353,
      "loss": 3.6572,
      "step": 123960
    },
    {
      "epoch": 0.11556009932650144,
      "grad_norm": 1.1046817302703857,
      "learning_rate": 0.0001768881854180183,
      "loss": 3.6084,
      "step": 123970
    },
    {
      "epoch": 0.11556942094458053,
      "grad_norm": 1.0199917554855347,
      "learning_rate": 0.0001768863210959231,
      "loss": 3.6426,
      "step": 123980
    },
    {
      "epoch": 0.11557874256265963,
      "grad_norm": 0.9065109491348267,
      "learning_rate": 0.0001768844567738279,
      "loss": 3.4995,
      "step": 123990
    },
    {
      "epoch": 0.11558806418073872,
      "grad_norm": 0.9597660303115845,
      "learning_rate": 0.0001768825924517327,
      "loss": 3.9287,
      "step": 124000
    },
    {
      "epoch": 0.11559738579881781,
      "grad_norm": 1.138461947441101,
      "learning_rate": 0.00017688072812963753,
      "loss": 3.4921,
      "step": 124010
    },
    {
      "epoch": 0.1156067074168969,
      "grad_norm": 1.0491950511932373,
      "learning_rate": 0.0001768788638075423,
      "loss": 3.6779,
      "step": 124020
    },
    {
      "epoch": 0.11561602903497599,
      "grad_norm": 0.9639672636985779,
      "learning_rate": 0.00017687699948544712,
      "loss": 3.725,
      "step": 124030
    },
    {
      "epoch": 0.11562535065305508,
      "grad_norm": 0.9877734780311584,
      "learning_rate": 0.0001768751351633519,
      "loss": 3.7482,
      "step": 124040
    },
    {
      "epoch": 0.11563467227113418,
      "grad_norm": 1.0312529802322388,
      "learning_rate": 0.0001768732708412567,
      "loss": 3.6702,
      "step": 124050
    },
    {
      "epoch": 0.11564399388921327,
      "grad_norm": 1.038609504699707,
      "learning_rate": 0.0001768714065191615,
      "loss": 3.6233,
      "step": 124060
    },
    {
      "epoch": 0.11565331550729235,
      "grad_norm": 1.0133682489395142,
      "learning_rate": 0.00017686954219706631,
      "loss": 3.7664,
      "step": 124070
    },
    {
      "epoch": 0.11566263712537145,
      "grad_norm": 1.269438624382019,
      "learning_rate": 0.00017686767787497112,
      "loss": 3.6099,
      "step": 124080
    },
    {
      "epoch": 0.11567195874345054,
      "grad_norm": 0.9505051374435425,
      "learning_rate": 0.0001768658135528759,
      "loss": 3.7015,
      "step": 124090
    },
    {
      "epoch": 0.11568128036152964,
      "grad_norm": 0.9006125330924988,
      "learning_rate": 0.0001768639492307807,
      "loss": 3.589,
      "step": 124100
    },
    {
      "epoch": 0.11569060197960873,
      "grad_norm": 1.1197706460952759,
      "learning_rate": 0.0001768620849086855,
      "loss": 3.6942,
      "step": 124110
    },
    {
      "epoch": 0.11569992359768783,
      "grad_norm": 1.068170189857483,
      "learning_rate": 0.00017686022058659032,
      "loss": 3.6368,
      "step": 124120
    },
    {
      "epoch": 0.1157092452157669,
      "grad_norm": 1.2695746421813965,
      "learning_rate": 0.00017685835626449513,
      "loss": 3.4791,
      "step": 124130
    },
    {
      "epoch": 0.115718566833846,
      "grad_norm": 1.2232989072799683,
      "learning_rate": 0.0001768564919423999,
      "loss": 3.5345,
      "step": 124140
    },
    {
      "epoch": 0.1157278884519251,
      "grad_norm": 1.1334056854248047,
      "learning_rate": 0.0001768546276203047,
      "loss": 3.6255,
      "step": 124150
    },
    {
      "epoch": 0.11573721007000419,
      "grad_norm": 0.9791274070739746,
      "learning_rate": 0.00017685276329820952,
      "loss": 3.6124,
      "step": 124160
    },
    {
      "epoch": 0.11574653168808328,
      "grad_norm": 0.9170860648155212,
      "learning_rate": 0.00017685089897611432,
      "loss": 3.7607,
      "step": 124170
    },
    {
      "epoch": 0.11575585330616237,
      "grad_norm": 1.2111051082611084,
      "learning_rate": 0.00017684903465401913,
      "loss": 3.6979,
      "step": 124180
    },
    {
      "epoch": 0.11576517492424146,
      "grad_norm": 1.0316022634506226,
      "learning_rate": 0.0001768471703319239,
      "loss": 3.6317,
      "step": 124190
    },
    {
      "epoch": 0.11577449654232055,
      "grad_norm": 1.0509110689163208,
      "learning_rate": 0.00017684530600982871,
      "loss": 3.6936,
      "step": 124200
    },
    {
      "epoch": 0.11578381816039965,
      "grad_norm": 1.1808226108551025,
      "learning_rate": 0.00017684344168773352,
      "loss": 3.5136,
      "step": 124210
    },
    {
      "epoch": 0.11579313977847874,
      "grad_norm": 1.0699564218521118,
      "learning_rate": 0.0001768415773656383,
      "loss": 3.4815,
      "step": 124220
    },
    {
      "epoch": 0.11580246139655782,
      "grad_norm": 1.03549063205719,
      "learning_rate": 0.00017683971304354313,
      "loss": 3.6888,
      "step": 124230
    },
    {
      "epoch": 0.11581178301463692,
      "grad_norm": 1.032363772392273,
      "learning_rate": 0.0001768378487214479,
      "loss": 3.7238,
      "step": 124240
    },
    {
      "epoch": 0.11582110463271601,
      "grad_norm": 0.9955645203590393,
      "learning_rate": 0.00017683598439935272,
      "loss": 3.6099,
      "step": 124250
    },
    {
      "epoch": 0.11583042625079511,
      "grad_norm": 1.1903611421585083,
      "learning_rate": 0.00017683412007725753,
      "loss": 3.7284,
      "step": 124260
    },
    {
      "epoch": 0.1158397478688742,
      "grad_norm": 1.648152232170105,
      "learning_rate": 0.0001768322557551623,
      "loss": 3.6281,
      "step": 124270
    },
    {
      "epoch": 0.11584906948695328,
      "grad_norm": 1.2321938276290894,
      "learning_rate": 0.0001768303914330671,
      "loss": 3.6286,
      "step": 124280
    },
    {
      "epoch": 0.11585839110503238,
      "grad_norm": 1.2527289390563965,
      "learning_rate": 0.00017682852711097192,
      "loss": 3.5942,
      "step": 124290
    },
    {
      "epoch": 0.11586771272311147,
      "grad_norm": 0.8583096265792847,
      "learning_rate": 0.00017682666278887672,
      "loss": 3.8342,
      "step": 124300
    },
    {
      "epoch": 0.11587703434119057,
      "grad_norm": 0.8694546222686768,
      "learning_rate": 0.00017682479846678153,
      "loss": 3.7049,
      "step": 124310
    },
    {
      "epoch": 0.11588635595926966,
      "grad_norm": 1.1965495347976685,
      "learning_rate": 0.0001768229341446863,
      "loss": 3.5259,
      "step": 124320
    },
    {
      "epoch": 0.11589567757734875,
      "grad_norm": 1.144287347793579,
      "learning_rate": 0.00017682106982259112,
      "loss": 3.5734,
      "step": 124330
    },
    {
      "epoch": 0.11590499919542784,
      "grad_norm": 0.8927165865898132,
      "learning_rate": 0.00017681920550049592,
      "loss": 3.6407,
      "step": 124340
    },
    {
      "epoch": 0.11591432081350693,
      "grad_norm": 0.90550696849823,
      "learning_rate": 0.00017681734117840073,
      "loss": 3.6407,
      "step": 124350
    },
    {
      "epoch": 0.11592364243158602,
      "grad_norm": 1.2620244026184082,
      "learning_rate": 0.00017681547685630553,
      "loss": 3.7109,
      "step": 124360
    },
    {
      "epoch": 0.11593296404966512,
      "grad_norm": 1.1676028966903687,
      "learning_rate": 0.00017681361253421031,
      "loss": 3.8329,
      "step": 124370
    },
    {
      "epoch": 0.11594228566774421,
      "grad_norm": 1.2081276178359985,
      "learning_rate": 0.00017681174821211512,
      "loss": 3.717,
      "step": 124380
    },
    {
      "epoch": 0.1159516072858233,
      "grad_norm": 0.8987281918525696,
      "learning_rate": 0.0001768098838900199,
      "loss": 3.6806,
      "step": 124390
    },
    {
      "epoch": 0.11596092890390239,
      "grad_norm": 0.9167393445968628,
      "learning_rate": 0.00017680801956792473,
      "loss": 3.6782,
      "step": 124400
    },
    {
      "epoch": 0.11597025052198148,
      "grad_norm": 1.1303439140319824,
      "learning_rate": 0.00017680615524582954,
      "loss": 3.6156,
      "step": 124410
    },
    {
      "epoch": 0.11597957214006058,
      "grad_norm": 1.0458189249038696,
      "learning_rate": 0.00017680429092373432,
      "loss": 3.9454,
      "step": 124420
    },
    {
      "epoch": 0.11598889375813967,
      "grad_norm": 1.172226905822754,
      "learning_rate": 0.00017680242660163912,
      "loss": 3.6326,
      "step": 124430
    },
    {
      "epoch": 0.11599821537621875,
      "grad_norm": 1.271876335144043,
      "learning_rate": 0.0001768005622795439,
      "loss": 3.656,
      "step": 124440
    },
    {
      "epoch": 0.11600753699429785,
      "grad_norm": 1.1757702827453613,
      "learning_rate": 0.0001767986979574487,
      "loss": 3.7002,
      "step": 124450
    },
    {
      "epoch": 0.11601685861237694,
      "grad_norm": 1.0719199180603027,
      "learning_rate": 0.00017679683363535354,
      "loss": 3.5641,
      "step": 124460
    },
    {
      "epoch": 0.11602618023045604,
      "grad_norm": 1.104094386100769,
      "learning_rate": 0.00017679496931325832,
      "loss": 3.5812,
      "step": 124470
    },
    {
      "epoch": 0.11603550184853513,
      "grad_norm": 1.0059927701950073,
      "learning_rate": 0.00017679310499116313,
      "loss": 3.8963,
      "step": 124480
    },
    {
      "epoch": 0.11604482346661421,
      "grad_norm": 0.9404823184013367,
      "learning_rate": 0.0001767912406690679,
      "loss": 3.6672,
      "step": 124490
    },
    {
      "epoch": 0.1160541450846933,
      "grad_norm": 1.1454877853393555,
      "learning_rate": 0.00017678937634697271,
      "loss": 3.7503,
      "step": 124500
    },
    {
      "epoch": 0.1160634667027724,
      "grad_norm": 1.0977556705474854,
      "learning_rate": 0.00017678751202487752,
      "loss": 3.6072,
      "step": 124510
    },
    {
      "epoch": 0.1160727883208515,
      "grad_norm": 1.0888094902038574,
      "learning_rate": 0.00017678564770278233,
      "loss": 3.5153,
      "step": 124520
    },
    {
      "epoch": 0.11608210993893059,
      "grad_norm": 0.952049970626831,
      "learning_rate": 0.00017678378338068713,
      "loss": 3.601,
      "step": 124530
    },
    {
      "epoch": 0.11609143155700967,
      "grad_norm": 0.8738070726394653,
      "learning_rate": 0.0001767819190585919,
      "loss": 3.5739,
      "step": 124540
    },
    {
      "epoch": 0.11610075317508876,
      "grad_norm": 1.0703327655792236,
      "learning_rate": 0.00017678005473649672,
      "loss": 3.6935,
      "step": 124550
    },
    {
      "epoch": 0.11611007479316786,
      "grad_norm": 1.066652536392212,
      "learning_rate": 0.00017677819041440153,
      "loss": 3.7763,
      "step": 124560
    },
    {
      "epoch": 0.11611939641124695,
      "grad_norm": 1.1205657720565796,
      "learning_rate": 0.00017677632609230633,
      "loss": 3.7724,
      "step": 124570
    },
    {
      "epoch": 0.11612871802932605,
      "grad_norm": 1.0864795446395874,
      "learning_rate": 0.00017677446177021114,
      "loss": 3.7671,
      "step": 124580
    },
    {
      "epoch": 0.11613803964740514,
      "grad_norm": 1.093572974205017,
      "learning_rate": 0.00017677259744811592,
      "loss": 3.7119,
      "step": 124590
    },
    {
      "epoch": 0.11614736126548422,
      "grad_norm": 1.1527252197265625,
      "learning_rate": 0.00017677073312602072,
      "loss": 3.5469,
      "step": 124600
    },
    {
      "epoch": 0.11615668288356332,
      "grad_norm": 1.0300769805908203,
      "learning_rate": 0.00017676886880392553,
      "loss": 3.6273,
      "step": 124610
    },
    {
      "epoch": 0.11616600450164241,
      "grad_norm": 0.9764780402183533,
      "learning_rate": 0.00017676700448183034,
      "loss": 3.5905,
      "step": 124620
    },
    {
      "epoch": 0.1161753261197215,
      "grad_norm": 1.064547061920166,
      "learning_rate": 0.00017676514015973514,
      "loss": 3.7219,
      "step": 124630
    },
    {
      "epoch": 0.1161846477378006,
      "grad_norm": 1.2432756423950195,
      "learning_rate": 0.00017676327583763992,
      "loss": 3.7019,
      "step": 124640
    },
    {
      "epoch": 0.11619396935587968,
      "grad_norm": 1.0183554887771606,
      "learning_rate": 0.00017676141151554473,
      "loss": 3.5012,
      "step": 124650
    },
    {
      "epoch": 0.11620329097395878,
      "grad_norm": 1.224474310874939,
      "learning_rate": 0.00017675954719344953,
      "loss": 3.6459,
      "step": 124660
    },
    {
      "epoch": 0.11621261259203787,
      "grad_norm": 1.018662452697754,
      "learning_rate": 0.0001767576828713543,
      "loss": 3.727,
      "step": 124670
    },
    {
      "epoch": 0.11622193421011696,
      "grad_norm": 1.2712980508804321,
      "learning_rate": 0.00017675581854925915,
      "loss": 3.7736,
      "step": 124680
    },
    {
      "epoch": 0.11623125582819606,
      "grad_norm": 1.0419284105300903,
      "learning_rate": 0.00017675395422716393,
      "loss": 3.4852,
      "step": 124690
    },
    {
      "epoch": 0.11624057744627514,
      "grad_norm": 1.0220311880111694,
      "learning_rate": 0.00017675208990506873,
      "loss": 3.7145,
      "step": 124700
    },
    {
      "epoch": 0.11624989906435423,
      "grad_norm": 1.215589165687561,
      "learning_rate": 0.00017675022558297354,
      "loss": 3.5882,
      "step": 124710
    },
    {
      "epoch": 0.11625922068243333,
      "grad_norm": 1.0855456590652466,
      "learning_rate": 0.00017674836126087832,
      "loss": 3.5856,
      "step": 124720
    },
    {
      "epoch": 0.11626854230051242,
      "grad_norm": 1.4278144836425781,
      "learning_rate": 0.00017674649693878312,
      "loss": 3.5801,
      "step": 124730
    },
    {
      "epoch": 0.11627786391859152,
      "grad_norm": 1.0075064897537231,
      "learning_rate": 0.00017674463261668793,
      "loss": 3.3792,
      "step": 124740
    },
    {
      "epoch": 0.1162871855366706,
      "grad_norm": 0.9969713091850281,
      "learning_rate": 0.00017674276829459274,
      "loss": 3.4995,
      "step": 124750
    },
    {
      "epoch": 0.11629650715474969,
      "grad_norm": 1.046409010887146,
      "learning_rate": 0.00017674090397249752,
      "loss": 3.6683,
      "step": 124760
    },
    {
      "epoch": 0.11630582877282879,
      "grad_norm": 0.8723508715629578,
      "learning_rate": 0.00017673903965040232,
      "loss": 3.5775,
      "step": 124770
    },
    {
      "epoch": 0.11631515039090788,
      "grad_norm": 1.005792260169983,
      "learning_rate": 0.00017673717532830713,
      "loss": 3.6223,
      "step": 124780
    },
    {
      "epoch": 0.11632447200898698,
      "grad_norm": 0.9457629323005676,
      "learning_rate": 0.00017673531100621193,
      "loss": 3.7008,
      "step": 124790
    },
    {
      "epoch": 0.11633379362706607,
      "grad_norm": 1.094711422920227,
      "learning_rate": 0.00017673344668411674,
      "loss": 3.667,
      "step": 124800
    },
    {
      "epoch": 0.11634311524514515,
      "grad_norm": 1.0279301404953003,
      "learning_rate": 0.00017673158236202152,
      "loss": 3.7777,
      "step": 124810
    },
    {
      "epoch": 0.11635243686322425,
      "grad_norm": 1.0483540296554565,
      "learning_rate": 0.00017672971803992633,
      "loss": 3.692,
      "step": 124820
    },
    {
      "epoch": 0.11636175848130334,
      "grad_norm": 1.0865291357040405,
      "learning_rate": 0.00017672785371783113,
      "loss": 3.628,
      "step": 124830
    },
    {
      "epoch": 0.11637108009938243,
      "grad_norm": 0.8997886776924133,
      "learning_rate": 0.0001767259893957359,
      "loss": 3.6905,
      "step": 124840
    },
    {
      "epoch": 0.11638040171746153,
      "grad_norm": 1.389929175376892,
      "learning_rate": 0.00017672412507364075,
      "loss": 3.7192,
      "step": 124850
    },
    {
      "epoch": 0.11638972333554061,
      "grad_norm": 1.1911433935165405,
      "learning_rate": 0.00017672226075154552,
      "loss": 3.606,
      "step": 124860
    },
    {
      "epoch": 0.1163990449536197,
      "grad_norm": 1.1747453212738037,
      "learning_rate": 0.00017672039642945033,
      "loss": 3.8045,
      "step": 124870
    },
    {
      "epoch": 0.1164083665716988,
      "grad_norm": 1.019030213356018,
      "learning_rate": 0.00017671853210735514,
      "loss": 3.7865,
      "step": 124880
    },
    {
      "epoch": 0.1164176881897779,
      "grad_norm": 1.1079025268554688,
      "learning_rate": 0.00017671666778525992,
      "loss": 3.6398,
      "step": 124890
    },
    {
      "epoch": 0.11642700980785699,
      "grad_norm": 1.454966425895691,
      "learning_rate": 0.00017671480346316472,
      "loss": 3.5868,
      "step": 124900
    },
    {
      "epoch": 0.11643633142593607,
      "grad_norm": 0.9587101936340332,
      "learning_rate": 0.00017671293914106953,
      "loss": 3.7661,
      "step": 124910
    },
    {
      "epoch": 0.11644565304401516,
      "grad_norm": 1.0382572412490845,
      "learning_rate": 0.00017671107481897434,
      "loss": 3.6295,
      "step": 124920
    },
    {
      "epoch": 0.11645497466209426,
      "grad_norm": 0.961615800857544,
      "learning_rate": 0.00017670921049687914,
      "loss": 3.7106,
      "step": 124930
    },
    {
      "epoch": 0.11646429628017335,
      "grad_norm": 0.9549556374549866,
      "learning_rate": 0.00017670734617478392,
      "loss": 3.5092,
      "step": 124940
    },
    {
      "epoch": 0.11647361789825245,
      "grad_norm": 1.1873607635498047,
      "learning_rate": 0.00017670548185268873,
      "loss": 3.6721,
      "step": 124950
    },
    {
      "epoch": 0.11648293951633153,
      "grad_norm": 1.0417133569717407,
      "learning_rate": 0.00017670361753059353,
      "loss": 3.5792,
      "step": 124960
    },
    {
      "epoch": 0.11649226113441062,
      "grad_norm": 1.1006094217300415,
      "learning_rate": 0.00017670175320849834,
      "loss": 3.6217,
      "step": 124970
    },
    {
      "epoch": 0.11650158275248972,
      "grad_norm": 1.0878112316131592,
      "learning_rate": 0.00017669988888640315,
      "loss": 3.5035,
      "step": 124980
    },
    {
      "epoch": 0.11651090437056881,
      "grad_norm": 1.137436032295227,
      "learning_rate": 0.00017669802456430793,
      "loss": 3.5365,
      "step": 124990
    },
    {
      "epoch": 0.1165202259886479,
      "grad_norm": 1.074499487876892,
      "learning_rate": 0.00017669616024221273,
      "loss": 3.421,
      "step": 125000
    },
    {
      "epoch": 0.116529547606727,
      "grad_norm": 1.1050245761871338,
      "learning_rate": 0.00017669429592011754,
      "loss": 3.7606,
      "step": 125010
    },
    {
      "epoch": 0.11653886922480608,
      "grad_norm": 1.0532642602920532,
      "learning_rate": 0.00017669243159802234,
      "loss": 3.7958,
      "step": 125020
    },
    {
      "epoch": 0.11654819084288517,
      "grad_norm": 1.0546690225601196,
      "learning_rate": 0.00017669056727592715,
      "loss": 3.7,
      "step": 125030
    },
    {
      "epoch": 0.11655751246096427,
      "grad_norm": 1.221060037612915,
      "learning_rate": 0.00017668870295383193,
      "loss": 3.6366,
      "step": 125040
    },
    {
      "epoch": 0.11656683407904336,
      "grad_norm": 1.0170245170593262,
      "learning_rate": 0.00017668683863173674,
      "loss": 3.7463,
      "step": 125050
    },
    {
      "epoch": 0.11657615569712246,
      "grad_norm": 1.216467261314392,
      "learning_rate": 0.00017668497430964152,
      "loss": 3.5066,
      "step": 125060
    },
    {
      "epoch": 0.11658547731520154,
      "grad_norm": 0.9906796813011169,
      "learning_rate": 0.00017668310998754635,
      "loss": 3.7367,
      "step": 125070
    },
    {
      "epoch": 0.11659479893328063,
      "grad_norm": 0.7507331967353821,
      "learning_rate": 0.00017668124566545116,
      "loss": 3.8053,
      "step": 125080
    },
    {
      "epoch": 0.11660412055135973,
      "grad_norm": 1.3886559009552002,
      "learning_rate": 0.00017667938134335593,
      "loss": 3.6286,
      "step": 125090
    },
    {
      "epoch": 0.11661344216943882,
      "grad_norm": 1.064260482788086,
      "learning_rate": 0.00017667751702126074,
      "loss": 3.9022,
      "step": 125100
    },
    {
      "epoch": 0.11662276378751792,
      "grad_norm": 1.1245651245117188,
      "learning_rate": 0.00017667565269916552,
      "loss": 3.7676,
      "step": 125110
    },
    {
      "epoch": 0.116632085405597,
      "grad_norm": 0.8633332848548889,
      "learning_rate": 0.00017667378837707033,
      "loss": 3.5201,
      "step": 125120
    },
    {
      "epoch": 0.11664140702367609,
      "grad_norm": 1.0684080123901367,
      "learning_rate": 0.00017667192405497516,
      "loss": 3.6773,
      "step": 125130
    },
    {
      "epoch": 0.11665072864175519,
      "grad_norm": 1.0799570083618164,
      "learning_rate": 0.00017667005973287994,
      "loss": 3.7239,
      "step": 125140
    },
    {
      "epoch": 0.11666005025983428,
      "grad_norm": 0.984269380569458,
      "learning_rate": 0.00017666819541078475,
      "loss": 3.5557,
      "step": 125150
    },
    {
      "epoch": 0.11666937187791337,
      "grad_norm": 1.1814360618591309,
      "learning_rate": 0.00017666633108868952,
      "loss": 3.552,
      "step": 125160
    },
    {
      "epoch": 0.11667869349599246,
      "grad_norm": 1.0850422382354736,
      "learning_rate": 0.00017666446676659433,
      "loss": 3.689,
      "step": 125170
    },
    {
      "epoch": 0.11668801511407155,
      "grad_norm": 1.0366650819778442,
      "learning_rate": 0.00017666260244449914,
      "loss": 3.6951,
      "step": 125180
    },
    {
      "epoch": 0.11669733673215064,
      "grad_norm": 1.1300992965698242,
      "learning_rate": 0.00017666073812240394,
      "loss": 3.597,
      "step": 125190
    },
    {
      "epoch": 0.11670665835022974,
      "grad_norm": 1.0481067895889282,
      "learning_rate": 0.00017665887380030875,
      "loss": 3.7868,
      "step": 125200
    },
    {
      "epoch": 0.11671597996830883,
      "grad_norm": 1.1325534582138062,
      "learning_rate": 0.00017665700947821353,
      "loss": 3.5931,
      "step": 125210
    },
    {
      "epoch": 0.11672530158638791,
      "grad_norm": 1.0325146913528442,
      "learning_rate": 0.00017665514515611834,
      "loss": 3.6425,
      "step": 125220
    },
    {
      "epoch": 0.11673462320446701,
      "grad_norm": 1.0328388214111328,
      "learning_rate": 0.00017665328083402314,
      "loss": 3.7256,
      "step": 125230
    },
    {
      "epoch": 0.1167439448225461,
      "grad_norm": 0.8903948664665222,
      "learning_rate": 0.00017665141651192795,
      "loss": 3.7363,
      "step": 125240
    },
    {
      "epoch": 0.1167532664406252,
      "grad_norm": 1.1371797323226929,
      "learning_rate": 0.00017664955218983275,
      "loss": 3.5806,
      "step": 125250
    },
    {
      "epoch": 0.11676258805870429,
      "grad_norm": 0.9506473541259766,
      "learning_rate": 0.00017664768786773753,
      "loss": 3.6659,
      "step": 125260
    },
    {
      "epoch": 0.11677190967678339,
      "grad_norm": 1.1455180644989014,
      "learning_rate": 0.00017664582354564234,
      "loss": 3.7042,
      "step": 125270
    },
    {
      "epoch": 0.11678123129486247,
      "grad_norm": 1.081870436668396,
      "learning_rate": 0.00017664395922354715,
      "loss": 3.7316,
      "step": 125280
    },
    {
      "epoch": 0.11679055291294156,
      "grad_norm": 1.1116193532943726,
      "learning_rate": 0.00017664209490145192,
      "loss": 3.5483,
      "step": 125290
    },
    {
      "epoch": 0.11679987453102066,
      "grad_norm": 0.9781174659729004,
      "learning_rate": 0.00017664023057935676,
      "loss": 3.4984,
      "step": 125300
    },
    {
      "epoch": 0.11680919614909975,
      "grad_norm": 1.016122817993164,
      "learning_rate": 0.00017663836625726154,
      "loss": 3.6613,
      "step": 125310
    },
    {
      "epoch": 0.11681851776717884,
      "grad_norm": 0.9649555683135986,
      "learning_rate": 0.00017663650193516634,
      "loss": 3.6816,
      "step": 125320
    },
    {
      "epoch": 0.11682783938525793,
      "grad_norm": 0.9270132184028625,
      "learning_rate": 0.00017663463761307115,
      "loss": 3.7258,
      "step": 125330
    },
    {
      "epoch": 0.11683716100333702,
      "grad_norm": 0.942452073097229,
      "learning_rate": 0.00017663277329097593,
      "loss": 3.4773,
      "step": 125340
    },
    {
      "epoch": 0.11684648262141611,
      "grad_norm": 1.055534839630127,
      "learning_rate": 0.00017663090896888074,
      "loss": 3.6957,
      "step": 125350
    },
    {
      "epoch": 0.11685580423949521,
      "grad_norm": 1.289613127708435,
      "learning_rate": 0.00017662904464678554,
      "loss": 3.6972,
      "step": 125360
    },
    {
      "epoch": 0.1168651258575743,
      "grad_norm": 1.0735242366790771,
      "learning_rate": 0.00017662718032469035,
      "loss": 3.7723,
      "step": 125370
    },
    {
      "epoch": 0.11687444747565338,
      "grad_norm": 1.098652720451355,
      "learning_rate": 0.00017662531600259515,
      "loss": 3.7421,
      "step": 125380
    },
    {
      "epoch": 0.11688376909373248,
      "grad_norm": 1.002094030380249,
      "learning_rate": 0.00017662345168049993,
      "loss": 3.7234,
      "step": 125390
    },
    {
      "epoch": 0.11689309071181157,
      "grad_norm": 0.9352871179580688,
      "learning_rate": 0.00017662158735840474,
      "loss": 3.3947,
      "step": 125400
    },
    {
      "epoch": 0.11690241232989067,
      "grad_norm": 1.0062485933303833,
      "learning_rate": 0.00017661972303630955,
      "loss": 3.6003,
      "step": 125410
    },
    {
      "epoch": 0.11691173394796976,
      "grad_norm": 0.9647660851478577,
      "learning_rate": 0.00017661785871421435,
      "loss": 3.7984,
      "step": 125420
    },
    {
      "epoch": 0.11692105556604884,
      "grad_norm": 1.0359666347503662,
      "learning_rate": 0.00017661599439211916,
      "loss": 3.4621,
      "step": 125430
    },
    {
      "epoch": 0.11693037718412794,
      "grad_norm": 1.2094273567199707,
      "learning_rate": 0.00017661413007002394,
      "loss": 3.6247,
      "step": 125440
    },
    {
      "epoch": 0.11693969880220703,
      "grad_norm": 1.0839498043060303,
      "learning_rate": 0.00017661226574792874,
      "loss": 3.7107,
      "step": 125450
    },
    {
      "epoch": 0.11694902042028613,
      "grad_norm": 1.035865068435669,
      "learning_rate": 0.00017661040142583355,
      "loss": 3.646,
      "step": 125460
    },
    {
      "epoch": 0.11695834203836522,
      "grad_norm": 0.9834580421447754,
      "learning_rate": 0.00017660853710373836,
      "loss": 3.7057,
      "step": 125470
    },
    {
      "epoch": 0.11696766365644432,
      "grad_norm": 0.9425610899925232,
      "learning_rate": 0.00017660667278164316,
      "loss": 3.6534,
      "step": 125480
    },
    {
      "epoch": 0.1169769852745234,
      "grad_norm": 1.0548288822174072,
      "learning_rate": 0.00017660480845954794,
      "loss": 3.7489,
      "step": 125490
    },
    {
      "epoch": 0.11698630689260249,
      "grad_norm": 1.1892465353012085,
      "learning_rate": 0.00017660294413745275,
      "loss": 3.6836,
      "step": 125500
    },
    {
      "epoch": 0.11699562851068158,
      "grad_norm": 1.3688312768936157,
      "learning_rate": 0.00017660107981535753,
      "loss": 3.6179,
      "step": 125510
    },
    {
      "epoch": 0.11700495012876068,
      "grad_norm": 1.2443805932998657,
      "learning_rate": 0.00017659921549326236,
      "loss": 3.6134,
      "step": 125520
    },
    {
      "epoch": 0.11701427174683977,
      "grad_norm": 1.1028883457183838,
      "learning_rate": 0.00017659735117116717,
      "loss": 3.7231,
      "step": 125530
    },
    {
      "epoch": 0.11702359336491885,
      "grad_norm": 0.8539206981658936,
      "learning_rate": 0.00017659548684907195,
      "loss": 3.8341,
      "step": 125540
    },
    {
      "epoch": 0.11703291498299795,
      "grad_norm": 0.9610912203788757,
      "learning_rate": 0.00017659362252697675,
      "loss": 3.6089,
      "step": 125550
    },
    {
      "epoch": 0.11704223660107704,
      "grad_norm": 0.9757179021835327,
      "learning_rate": 0.00017659175820488153,
      "loss": 3.7425,
      "step": 125560
    },
    {
      "epoch": 0.11705155821915614,
      "grad_norm": 1.0303760766983032,
      "learning_rate": 0.00017658989388278634,
      "loss": 3.7276,
      "step": 125570
    },
    {
      "epoch": 0.11706087983723523,
      "grad_norm": 1.2174919843673706,
      "learning_rate": 0.00017658802956069117,
      "loss": 3.633,
      "step": 125580
    },
    {
      "epoch": 0.11707020145531431,
      "grad_norm": 1.088582992553711,
      "learning_rate": 0.00017658616523859595,
      "loss": 3.7881,
      "step": 125590
    },
    {
      "epoch": 0.11707952307339341,
      "grad_norm": 1.0493465662002563,
      "learning_rate": 0.00017658430091650076,
      "loss": 3.802,
      "step": 125600
    },
    {
      "epoch": 0.1170888446914725,
      "grad_norm": 1.0712343454360962,
      "learning_rate": 0.00017658243659440554,
      "loss": 3.5993,
      "step": 125610
    },
    {
      "epoch": 0.1170981663095516,
      "grad_norm": 1.0143096446990967,
      "learning_rate": 0.00017658057227231034,
      "loss": 3.5471,
      "step": 125620
    },
    {
      "epoch": 0.11710748792763069,
      "grad_norm": 1.2264106273651123,
      "learning_rate": 0.00017657870795021515,
      "loss": 3.6521,
      "step": 125630
    },
    {
      "epoch": 0.11711680954570977,
      "grad_norm": 1.0042062997817993,
      "learning_rate": 0.00017657684362811996,
      "loss": 3.5441,
      "step": 125640
    },
    {
      "epoch": 0.11712613116378887,
      "grad_norm": 1.0677356719970703,
      "learning_rate": 0.00017657497930602476,
      "loss": 3.5029,
      "step": 125650
    },
    {
      "epoch": 0.11713545278186796,
      "grad_norm": 0.9833751916885376,
      "learning_rate": 0.00017657311498392954,
      "loss": 3.6494,
      "step": 125660
    },
    {
      "epoch": 0.11714477439994705,
      "grad_norm": 1.0077593326568604,
      "learning_rate": 0.00017657125066183435,
      "loss": 3.8226,
      "step": 125670
    },
    {
      "epoch": 0.11715409601802615,
      "grad_norm": 1.1761980056762695,
      "learning_rate": 0.00017656938633973915,
      "loss": 3.5837,
      "step": 125680
    },
    {
      "epoch": 0.11716341763610524,
      "grad_norm": 1.0618205070495605,
      "learning_rate": 0.00017656752201764396,
      "loss": 3.4987,
      "step": 125690
    },
    {
      "epoch": 0.11717273925418432,
      "grad_norm": 1.0928713083267212,
      "learning_rate": 0.00017656565769554877,
      "loss": 3.6116,
      "step": 125700
    },
    {
      "epoch": 0.11718206087226342,
      "grad_norm": 1.0360387563705444,
      "learning_rate": 0.00017656379337345355,
      "loss": 3.7125,
      "step": 125710
    },
    {
      "epoch": 0.11719138249034251,
      "grad_norm": 1.1437729597091675,
      "learning_rate": 0.00017656192905135835,
      "loss": 3.5461,
      "step": 125720
    },
    {
      "epoch": 0.11720070410842161,
      "grad_norm": 1.0396952629089355,
      "learning_rate": 0.00017656006472926316,
      "loss": 3.8268,
      "step": 125730
    },
    {
      "epoch": 0.1172100257265007,
      "grad_norm": 1.051119327545166,
      "learning_rate": 0.00017655820040716794,
      "loss": 3.6407,
      "step": 125740
    },
    {
      "epoch": 0.11721934734457978,
      "grad_norm": 1.0410094261169434,
      "learning_rate": 0.00017655633608507277,
      "loss": 3.6039,
      "step": 125750
    },
    {
      "epoch": 0.11722866896265888,
      "grad_norm": 1.0950090885162354,
      "learning_rate": 0.00017655447176297755,
      "loss": 3.7773,
      "step": 125760
    },
    {
      "epoch": 0.11723799058073797,
      "grad_norm": 0.9613046050071716,
      "learning_rate": 0.00017655260744088236,
      "loss": 3.7167,
      "step": 125770
    },
    {
      "epoch": 0.11724731219881707,
      "grad_norm": 0.9854408502578735,
      "learning_rate": 0.00017655074311878716,
      "loss": 3.6229,
      "step": 125780
    },
    {
      "epoch": 0.11725663381689616,
      "grad_norm": 1.1942616701126099,
      "learning_rate": 0.00017654887879669194,
      "loss": 3.6119,
      "step": 125790
    },
    {
      "epoch": 0.11726595543497524,
      "grad_norm": 0.8411369919776917,
      "learning_rate": 0.00017654701447459675,
      "loss": 3.7539,
      "step": 125800
    },
    {
      "epoch": 0.11727527705305434,
      "grad_norm": 1.0952056646347046,
      "learning_rate": 0.00017654515015250155,
      "loss": 3.6501,
      "step": 125810
    },
    {
      "epoch": 0.11728459867113343,
      "grad_norm": 1.123618721961975,
      "learning_rate": 0.00017654328583040636,
      "loss": 3.5182,
      "step": 125820
    },
    {
      "epoch": 0.11729392028921252,
      "grad_norm": 1.11099112033844,
      "learning_rate": 0.00017654142150831117,
      "loss": 3.7269,
      "step": 125830
    },
    {
      "epoch": 0.11730324190729162,
      "grad_norm": 0.9244495034217834,
      "learning_rate": 0.00017653955718621595,
      "loss": 3.5173,
      "step": 125840
    },
    {
      "epoch": 0.1173125635253707,
      "grad_norm": 1.1579102277755737,
      "learning_rate": 0.00017653769286412075,
      "loss": 3.6446,
      "step": 125850
    },
    {
      "epoch": 0.1173218851434498,
      "grad_norm": 1.0555709600448608,
      "learning_rate": 0.00017653582854202556,
      "loss": 3.6996,
      "step": 125860
    },
    {
      "epoch": 0.11733120676152889,
      "grad_norm": 0.9301415085792542,
      "learning_rate": 0.00017653396421993037,
      "loss": 3.7515,
      "step": 125870
    },
    {
      "epoch": 0.11734052837960798,
      "grad_norm": 1.0249824523925781,
      "learning_rate": 0.00017653209989783517,
      "loss": 3.681,
      "step": 125880
    },
    {
      "epoch": 0.11734984999768708,
      "grad_norm": 1.0712294578552246,
      "learning_rate": 0.00017653023557573995,
      "loss": 3.5405,
      "step": 125890
    },
    {
      "epoch": 0.11735917161576616,
      "grad_norm": 1.1972718238830566,
      "learning_rate": 0.00017652837125364476,
      "loss": 3.581,
      "step": 125900
    },
    {
      "epoch": 0.11736849323384525,
      "grad_norm": 1.204336166381836,
      "learning_rate": 0.00017652650693154956,
      "loss": 3.5376,
      "step": 125910
    },
    {
      "epoch": 0.11737781485192435,
      "grad_norm": 1.010277271270752,
      "learning_rate": 0.00017652464260945437,
      "loss": 3.6451,
      "step": 125920
    },
    {
      "epoch": 0.11738713647000344,
      "grad_norm": 1.0887079238891602,
      "learning_rate": 0.00017652277828735918,
      "loss": 3.7452,
      "step": 125930
    },
    {
      "epoch": 0.11739645808808254,
      "grad_norm": 0.9248384833335876,
      "learning_rate": 0.00017652091396526396,
      "loss": 3.7088,
      "step": 125940
    },
    {
      "epoch": 0.11740577970616163,
      "grad_norm": 1.1795876026153564,
      "learning_rate": 0.00017651904964316876,
      "loss": 3.5116,
      "step": 125950
    },
    {
      "epoch": 0.11741510132424071,
      "grad_norm": 1.1556427478790283,
      "learning_rate": 0.00017651718532107354,
      "loss": 3.5729,
      "step": 125960
    },
    {
      "epoch": 0.1174244229423198,
      "grad_norm": 0.977462649345398,
      "learning_rate": 0.00017651532099897837,
      "loss": 3.7627,
      "step": 125970
    },
    {
      "epoch": 0.1174337445603989,
      "grad_norm": 1.049379587173462,
      "learning_rate": 0.00017651345667688318,
      "loss": 3.6246,
      "step": 125980
    },
    {
      "epoch": 0.117443066178478,
      "grad_norm": 1.072940707206726,
      "learning_rate": 0.00017651159235478796,
      "loss": 3.5208,
      "step": 125990
    },
    {
      "epoch": 0.11745238779655709,
      "grad_norm": 0.8965297937393188,
      "learning_rate": 0.00017650972803269277,
      "loss": 3.5884,
      "step": 126000
    },
    {
      "epoch": 0.11746170941463617,
      "grad_norm": 1.1101516485214233,
      "learning_rate": 0.00017650786371059755,
      "loss": 3.8478,
      "step": 126010
    },
    {
      "epoch": 0.11747103103271526,
      "grad_norm": 0.9276065826416016,
      "learning_rate": 0.00017650599938850235,
      "loss": 3.7585,
      "step": 126020
    },
    {
      "epoch": 0.11748035265079436,
      "grad_norm": 1.0270466804504395,
      "learning_rate": 0.00017650413506640719,
      "loss": 3.6306,
      "step": 126030
    },
    {
      "epoch": 0.11748967426887345,
      "grad_norm": 1.1090307235717773,
      "learning_rate": 0.00017650227074431196,
      "loss": 3.6845,
      "step": 126040
    },
    {
      "epoch": 0.11749899588695255,
      "grad_norm": 1.0524903535842896,
      "learning_rate": 0.00017650040642221677,
      "loss": 3.6119,
      "step": 126050
    },
    {
      "epoch": 0.11750831750503163,
      "grad_norm": 1.070493459701538,
      "learning_rate": 0.00017649854210012155,
      "loss": 3.6653,
      "step": 126060
    },
    {
      "epoch": 0.11751763912311072,
      "grad_norm": 1.0041857957839966,
      "learning_rate": 0.00017649667777802636,
      "loss": 3.5947,
      "step": 126070
    },
    {
      "epoch": 0.11752696074118982,
      "grad_norm": 1.1074024438858032,
      "learning_rate": 0.00017649481345593116,
      "loss": 3.8324,
      "step": 126080
    },
    {
      "epoch": 0.11753628235926891,
      "grad_norm": 1.108351707458496,
      "learning_rate": 0.00017649294913383597,
      "loss": 3.6576,
      "step": 126090
    },
    {
      "epoch": 0.117545603977348,
      "grad_norm": 0.9758886694908142,
      "learning_rate": 0.00017649108481174078,
      "loss": 3.6875,
      "step": 126100
    },
    {
      "epoch": 0.11755492559542709,
      "grad_norm": 1.1096346378326416,
      "learning_rate": 0.00017648922048964555,
      "loss": 3.6954,
      "step": 126110
    },
    {
      "epoch": 0.11756424721350618,
      "grad_norm": 0.8931911587715149,
      "learning_rate": 0.00017648735616755036,
      "loss": 3.5813,
      "step": 126120
    },
    {
      "epoch": 0.11757356883158528,
      "grad_norm": 1.138817548751831,
      "learning_rate": 0.00017648549184545514,
      "loss": 3.6599,
      "step": 126130
    },
    {
      "epoch": 0.11758289044966437,
      "grad_norm": 0.9665735363960266,
      "learning_rate": 0.00017648362752335997,
      "loss": 3.6016,
      "step": 126140
    },
    {
      "epoch": 0.11759221206774347,
      "grad_norm": 1.086459755897522,
      "learning_rate": 0.00017648176320126478,
      "loss": 3.5032,
      "step": 126150
    },
    {
      "epoch": 0.11760153368582256,
      "grad_norm": 1.0039323568344116,
      "learning_rate": 0.00017647989887916956,
      "loss": 3.4383,
      "step": 126160
    },
    {
      "epoch": 0.11761085530390164,
      "grad_norm": 1.0455098152160645,
      "learning_rate": 0.00017647803455707437,
      "loss": 3.4943,
      "step": 126170
    },
    {
      "epoch": 0.11762017692198073,
      "grad_norm": 1.0247657299041748,
      "learning_rate": 0.00017647617023497914,
      "loss": 3.6851,
      "step": 126180
    },
    {
      "epoch": 0.11762949854005983,
      "grad_norm": 1.0591920614242554,
      "learning_rate": 0.00017647430591288395,
      "loss": 3.6589,
      "step": 126190
    },
    {
      "epoch": 0.11763882015813892,
      "grad_norm": 1.0636473894119263,
      "learning_rate": 0.00017647244159078878,
      "loss": 3.6475,
      "step": 126200
    },
    {
      "epoch": 0.11764814177621802,
      "grad_norm": 1.045983910560608,
      "learning_rate": 0.00017647057726869356,
      "loss": 3.5722,
      "step": 126210
    },
    {
      "epoch": 0.1176574633942971,
      "grad_norm": 1.4478036165237427,
      "learning_rate": 0.00017646871294659837,
      "loss": 3.6393,
      "step": 126220
    },
    {
      "epoch": 0.1176667850123762,
      "grad_norm": 1.089186191558838,
      "learning_rate": 0.00017646684862450315,
      "loss": 3.6159,
      "step": 126230
    },
    {
      "epoch": 0.11767610663045529,
      "grad_norm": 1.0800795555114746,
      "learning_rate": 0.00017646498430240796,
      "loss": 3.6364,
      "step": 126240
    },
    {
      "epoch": 0.11768542824853438,
      "grad_norm": 0.8121035695075989,
      "learning_rate": 0.00017646311998031276,
      "loss": 3.6801,
      "step": 126250
    },
    {
      "epoch": 0.11769474986661348,
      "grad_norm": 0.989719808101654,
      "learning_rate": 0.00017646125565821757,
      "loss": 3.7334,
      "step": 126260
    },
    {
      "epoch": 0.11770407148469256,
      "grad_norm": 1.0277434587478638,
      "learning_rate": 0.00017645939133612237,
      "loss": 3.6073,
      "step": 126270
    },
    {
      "epoch": 0.11771339310277165,
      "grad_norm": 1.1289573907852173,
      "learning_rate": 0.00017645752701402715,
      "loss": 3.7105,
      "step": 126280
    },
    {
      "epoch": 0.11772271472085075,
      "grad_norm": 1.1565465927124023,
      "learning_rate": 0.00017645566269193196,
      "loss": 3.5422,
      "step": 126290
    },
    {
      "epoch": 0.11773203633892984,
      "grad_norm": 0.8872343897819519,
      "learning_rate": 0.00017645379836983677,
      "loss": 3.6013,
      "step": 126300
    },
    {
      "epoch": 0.11774135795700894,
      "grad_norm": 0.9920763373374939,
      "learning_rate": 0.00017645193404774157,
      "loss": 3.76,
      "step": 126310
    },
    {
      "epoch": 0.11775067957508802,
      "grad_norm": 1.1048835515975952,
      "learning_rate": 0.00017645006972564638,
      "loss": 3.6439,
      "step": 126320
    },
    {
      "epoch": 0.11776000119316711,
      "grad_norm": 1.0763030052185059,
      "learning_rate": 0.00017644820540355116,
      "loss": 3.5622,
      "step": 126330
    },
    {
      "epoch": 0.1177693228112462,
      "grad_norm": 0.9089756608009338,
      "learning_rate": 0.00017644634108145596,
      "loss": 3.7984,
      "step": 126340
    },
    {
      "epoch": 0.1177786444293253,
      "grad_norm": 1.109258770942688,
      "learning_rate": 0.00017644447675936077,
      "loss": 3.5998,
      "step": 126350
    },
    {
      "epoch": 0.1177879660474044,
      "grad_norm": 0.9663679599761963,
      "learning_rate": 0.00017644261243726558,
      "loss": 3.6627,
      "step": 126360
    },
    {
      "epoch": 0.11779728766548349,
      "grad_norm": 1.0764174461364746,
      "learning_rate": 0.00017644074811517038,
      "loss": 3.5082,
      "step": 126370
    },
    {
      "epoch": 0.11780660928356257,
      "grad_norm": 0.9700079560279846,
      "learning_rate": 0.00017643888379307516,
      "loss": 3.5103,
      "step": 126380
    },
    {
      "epoch": 0.11781593090164166,
      "grad_norm": 1.0670392513275146,
      "learning_rate": 0.00017643701947097997,
      "loss": 3.5683,
      "step": 126390
    },
    {
      "epoch": 0.11782525251972076,
      "grad_norm": 0.9538279175758362,
      "learning_rate": 0.00017643515514888477,
      "loss": 3.6996,
      "step": 126400
    },
    {
      "epoch": 0.11783457413779985,
      "grad_norm": 1.0820523500442505,
      "learning_rate": 0.00017643329082678955,
      "loss": 3.6789,
      "step": 126410
    },
    {
      "epoch": 0.11784389575587895,
      "grad_norm": 1.029651165008545,
      "learning_rate": 0.0001764314265046944,
      "loss": 3.6842,
      "step": 126420
    },
    {
      "epoch": 0.11785321737395803,
      "grad_norm": 0.9840251803398132,
      "learning_rate": 0.00017642956218259917,
      "loss": 3.5905,
      "step": 126430
    },
    {
      "epoch": 0.11786253899203712,
      "grad_norm": 0.9072517156600952,
      "learning_rate": 0.00017642769786050397,
      "loss": 3.5153,
      "step": 126440
    },
    {
      "epoch": 0.11787186061011622,
      "grad_norm": 1.1413357257843018,
      "learning_rate": 0.00017642583353840878,
      "loss": 3.4404,
      "step": 126450
    },
    {
      "epoch": 0.11788118222819531,
      "grad_norm": 1.378330111503601,
      "learning_rate": 0.00017642396921631356,
      "loss": 3.7448,
      "step": 126460
    },
    {
      "epoch": 0.1178905038462744,
      "grad_norm": 1.1246123313903809,
      "learning_rate": 0.00017642210489421836,
      "loss": 3.6209,
      "step": 126470
    },
    {
      "epoch": 0.11789982546435349,
      "grad_norm": 1.0716325044631958,
      "learning_rate": 0.00017642024057212317,
      "loss": 3.6664,
      "step": 126480
    },
    {
      "epoch": 0.11790914708243258,
      "grad_norm": 1.038562536239624,
      "learning_rate": 0.00017641837625002798,
      "loss": 3.7944,
      "step": 126490
    },
    {
      "epoch": 0.11791846870051167,
      "grad_norm": 1.047028660774231,
      "learning_rate": 0.00017641651192793278,
      "loss": 3.8063,
      "step": 126500
    },
    {
      "epoch": 0.11792779031859077,
      "grad_norm": 1.0322942733764648,
      "learning_rate": 0.00017641464760583756,
      "loss": 3.8068,
      "step": 126510
    },
    {
      "epoch": 0.11793711193666986,
      "grad_norm": 0.99982088804245,
      "learning_rate": 0.00017641278328374237,
      "loss": 3.7455,
      "step": 126520
    },
    {
      "epoch": 0.11794643355474894,
      "grad_norm": 1.2152273654937744,
      "learning_rate": 0.00017641091896164718,
      "loss": 3.6505,
      "step": 126530
    },
    {
      "epoch": 0.11795575517282804,
      "grad_norm": 0.8653807044029236,
      "learning_rate": 0.00017640905463955198,
      "loss": 3.7087,
      "step": 126540
    },
    {
      "epoch": 0.11796507679090713,
      "grad_norm": 1.2414112091064453,
      "learning_rate": 0.0001764071903174568,
      "loss": 3.6602,
      "step": 126550
    },
    {
      "epoch": 0.11797439840898623,
      "grad_norm": 1.155299186706543,
      "learning_rate": 0.00017640532599536157,
      "loss": 3.6678,
      "step": 126560
    },
    {
      "epoch": 0.11798372002706532,
      "grad_norm": 0.9124847054481506,
      "learning_rate": 0.00017640346167326637,
      "loss": 3.4451,
      "step": 126570
    },
    {
      "epoch": 0.1179930416451444,
      "grad_norm": 1.1944653987884521,
      "learning_rate": 0.00017640159735117115,
      "loss": 3.6785,
      "step": 126580
    },
    {
      "epoch": 0.1180023632632235,
      "grad_norm": 1.207639217376709,
      "learning_rate": 0.00017639973302907599,
      "loss": 3.6998,
      "step": 126590
    },
    {
      "epoch": 0.11801168488130259,
      "grad_norm": 1.0963270664215088,
      "learning_rate": 0.0001763978687069808,
      "loss": 3.7379,
      "step": 126600
    },
    {
      "epoch": 0.11802100649938169,
      "grad_norm": 1.1097557544708252,
      "learning_rate": 0.00017639600438488557,
      "loss": 3.608,
      "step": 126610
    },
    {
      "epoch": 0.11803032811746078,
      "grad_norm": 1.0951673984527588,
      "learning_rate": 0.00017639414006279038,
      "loss": 3.531,
      "step": 126620
    },
    {
      "epoch": 0.11803964973553988,
      "grad_norm": 0.9992876648902893,
      "learning_rate": 0.00017639227574069516,
      "loss": 3.6498,
      "step": 126630
    },
    {
      "epoch": 0.11804897135361896,
      "grad_norm": 0.8185586333274841,
      "learning_rate": 0.00017639041141859996,
      "loss": 3.768,
      "step": 126640
    },
    {
      "epoch": 0.11805829297169805,
      "grad_norm": 1.1636055707931519,
      "learning_rate": 0.0001763885470965048,
      "loss": 3.5806,
      "step": 126650
    },
    {
      "epoch": 0.11806761458977715,
      "grad_norm": 1.1695261001586914,
      "learning_rate": 0.00017638668277440958,
      "loss": 3.7119,
      "step": 126660
    },
    {
      "epoch": 0.11807693620785624,
      "grad_norm": 1.06326162815094,
      "learning_rate": 0.00017638481845231438,
      "loss": 3.575,
      "step": 126670
    },
    {
      "epoch": 0.11808625782593533,
      "grad_norm": 1.0126808881759644,
      "learning_rate": 0.00017638295413021916,
      "loss": 3.5903,
      "step": 126680
    },
    {
      "epoch": 0.11809557944401441,
      "grad_norm": 1.0517709255218506,
      "learning_rate": 0.00017638108980812397,
      "loss": 3.6852,
      "step": 126690
    },
    {
      "epoch": 0.11810490106209351,
      "grad_norm": 1.1574105024337769,
      "learning_rate": 0.00017637922548602877,
      "loss": 3.7711,
      "step": 126700
    },
    {
      "epoch": 0.1181142226801726,
      "grad_norm": 0.8897302150726318,
      "learning_rate": 0.00017637736116393358,
      "loss": 3.6736,
      "step": 126710
    },
    {
      "epoch": 0.1181235442982517,
      "grad_norm": 0.9068772196769714,
      "learning_rate": 0.0001763754968418384,
      "loss": 3.6218,
      "step": 126720
    },
    {
      "epoch": 0.11813286591633079,
      "grad_norm": 1.0447969436645508,
      "learning_rate": 0.00017637363251974317,
      "loss": 3.6752,
      "step": 126730
    },
    {
      "epoch": 0.11814218753440987,
      "grad_norm": 1.1656512022018433,
      "learning_rate": 0.00017637176819764797,
      "loss": 3.6277,
      "step": 126740
    },
    {
      "epoch": 0.11815150915248897,
      "grad_norm": 1.1421221494674683,
      "learning_rate": 0.00017636990387555278,
      "loss": 3.6965,
      "step": 126750
    },
    {
      "epoch": 0.11816083077056806,
      "grad_norm": 0.9863205552101135,
      "learning_rate": 0.00017636803955345758,
      "loss": 3.7136,
      "step": 126760
    },
    {
      "epoch": 0.11817015238864716,
      "grad_norm": 1.1628239154815674,
      "learning_rate": 0.0001763661752313624,
      "loss": 3.7289,
      "step": 126770
    },
    {
      "epoch": 0.11817947400672625,
      "grad_norm": 1.0478416681289673,
      "learning_rate": 0.00017636431090926717,
      "loss": 3.5382,
      "step": 126780
    },
    {
      "epoch": 0.11818879562480533,
      "grad_norm": 1.0478105545043945,
      "learning_rate": 0.00017636244658717198,
      "loss": 3.4573,
      "step": 126790
    },
    {
      "epoch": 0.11819811724288443,
      "grad_norm": 1.1094685792922974,
      "learning_rate": 0.00017636058226507678,
      "loss": 3.5654,
      "step": 126800
    },
    {
      "epoch": 0.11820743886096352,
      "grad_norm": 1.0886223316192627,
      "learning_rate": 0.0001763587179429816,
      "loss": 3.6264,
      "step": 126810
    },
    {
      "epoch": 0.11821676047904262,
      "grad_norm": 1.2822325229644775,
      "learning_rate": 0.0001763568536208864,
      "loss": 3.7247,
      "step": 126820
    },
    {
      "epoch": 0.11822608209712171,
      "grad_norm": 1.1356843709945679,
      "learning_rate": 0.00017635498929879117,
      "loss": 3.3929,
      "step": 126830
    },
    {
      "epoch": 0.1182354037152008,
      "grad_norm": 1.042893886566162,
      "learning_rate": 0.00017635312497669598,
      "loss": 3.5075,
      "step": 126840
    },
    {
      "epoch": 0.11824472533327988,
      "grad_norm": 1.3443946838378906,
      "learning_rate": 0.0001763512606546008,
      "loss": 3.5454,
      "step": 126850
    },
    {
      "epoch": 0.11825404695135898,
      "grad_norm": 1.1370410919189453,
      "learning_rate": 0.00017634939633250557,
      "loss": 3.8135,
      "step": 126860
    },
    {
      "epoch": 0.11826336856943807,
      "grad_norm": 1.0342564582824707,
      "learning_rate": 0.0001763475320104104,
      "loss": 3.5458,
      "step": 126870
    },
    {
      "epoch": 0.11827269018751717,
      "grad_norm": 1.317694067955017,
      "learning_rate": 0.00017634566768831518,
      "loss": 3.6241,
      "step": 126880
    },
    {
      "epoch": 0.11828201180559626,
      "grad_norm": 1.2076746225357056,
      "learning_rate": 0.00017634380336621999,
      "loss": 3.3849,
      "step": 126890
    },
    {
      "epoch": 0.11829133342367534,
      "grad_norm": 1.249222755432129,
      "learning_rate": 0.0001763419390441248,
      "loss": 3.4936,
      "step": 126900
    },
    {
      "epoch": 0.11830065504175444,
      "grad_norm": 0.9388324022293091,
      "learning_rate": 0.00017634007472202957,
      "loss": 3.7113,
      "step": 126910
    },
    {
      "epoch": 0.11830997665983353,
      "grad_norm": 1.1020387411117554,
      "learning_rate": 0.00017633821039993438,
      "loss": 3.8164,
      "step": 126920
    },
    {
      "epoch": 0.11831929827791263,
      "grad_norm": 1.0525225400924683,
      "learning_rate": 0.00017633634607783918,
      "loss": 3.6287,
      "step": 126930
    },
    {
      "epoch": 0.11832861989599172,
      "grad_norm": 1.0740740299224854,
      "learning_rate": 0.000176334481755744,
      "loss": 3.7295,
      "step": 126940
    },
    {
      "epoch": 0.1183379415140708,
      "grad_norm": 1.2037297487258911,
      "learning_rate": 0.0001763326174336488,
      "loss": 3.7186,
      "step": 126950
    },
    {
      "epoch": 0.1183472631321499,
      "grad_norm": 1.2562549114227295,
      "learning_rate": 0.00017633075311155358,
      "loss": 3.6813,
      "step": 126960
    },
    {
      "epoch": 0.11835658475022899,
      "grad_norm": 1.0321221351623535,
      "learning_rate": 0.00017632888878945838,
      "loss": 3.4801,
      "step": 126970
    },
    {
      "epoch": 0.11836590636830809,
      "grad_norm": 1.037137746810913,
      "learning_rate": 0.0001763270244673632,
      "loss": 3.5137,
      "step": 126980
    },
    {
      "epoch": 0.11837522798638718,
      "grad_norm": 0.9409327507019043,
      "learning_rate": 0.000176325160145268,
      "loss": 3.6863,
      "step": 126990
    },
    {
      "epoch": 0.11838454960446626,
      "grad_norm": 1.0301017761230469,
      "learning_rate": 0.0001763232958231728,
      "loss": 3.5969,
      "step": 127000
    },
    {
      "epoch": 0.11839387122254535,
      "grad_norm": 1.0464582443237305,
      "learning_rate": 0.00017632143150107758,
      "loss": 3.7227,
      "step": 127010
    },
    {
      "epoch": 0.11840319284062445,
      "grad_norm": 1.0905903577804565,
      "learning_rate": 0.00017631956717898239,
      "loss": 3.5426,
      "step": 127020
    },
    {
      "epoch": 0.11841251445870354,
      "grad_norm": 1.0180885791778564,
      "learning_rate": 0.0001763177028568872,
      "loss": 3.6307,
      "step": 127030
    },
    {
      "epoch": 0.11842183607678264,
      "grad_norm": 1.074367880821228,
      "learning_rate": 0.000176315838534792,
      "loss": 3.5753,
      "step": 127040
    },
    {
      "epoch": 0.11843115769486173,
      "grad_norm": 1.084975004196167,
      "learning_rate": 0.0001763139742126968,
      "loss": 3.7498,
      "step": 127050
    },
    {
      "epoch": 0.11844047931294081,
      "grad_norm": 1.0313175916671753,
      "learning_rate": 0.00017631210989060158,
      "loss": 3.6876,
      "step": 127060
    },
    {
      "epoch": 0.11844980093101991,
      "grad_norm": 1.0464942455291748,
      "learning_rate": 0.0001763102455685064,
      "loss": 3.6067,
      "step": 127070
    },
    {
      "epoch": 0.118459122549099,
      "grad_norm": 1.1720856428146362,
      "learning_rate": 0.00017630838124641117,
      "loss": 3.5237,
      "step": 127080
    },
    {
      "epoch": 0.1184684441671781,
      "grad_norm": 1.1900404691696167,
      "learning_rate": 0.00017630651692431598,
      "loss": 3.6772,
      "step": 127090
    },
    {
      "epoch": 0.11847776578525719,
      "grad_norm": 0.782961368560791,
      "learning_rate": 0.0001763046526022208,
      "loss": 3.5642,
      "step": 127100
    },
    {
      "epoch": 0.11848708740333627,
      "grad_norm": 1.1323039531707764,
      "learning_rate": 0.0001763027882801256,
      "loss": 3.5446,
      "step": 127110
    },
    {
      "epoch": 0.11849640902141537,
      "grad_norm": 0.9943915009498596,
      "learning_rate": 0.0001763009239580304,
      "loss": 3.49,
      "step": 127120
    },
    {
      "epoch": 0.11850573063949446,
      "grad_norm": 1.2724919319152832,
      "learning_rate": 0.00017629905963593517,
      "loss": 3.6764,
      "step": 127130
    },
    {
      "epoch": 0.11851505225757356,
      "grad_norm": 1.069848656654358,
      "learning_rate": 0.00017629719531383998,
      "loss": 3.6263,
      "step": 127140
    },
    {
      "epoch": 0.11852437387565265,
      "grad_norm": 0.9898010492324829,
      "learning_rate": 0.0001762953309917448,
      "loss": 3.8015,
      "step": 127150
    },
    {
      "epoch": 0.11853369549373173,
      "grad_norm": 1.0762070417404175,
      "learning_rate": 0.0001762934666696496,
      "loss": 3.5962,
      "step": 127160
    },
    {
      "epoch": 0.11854301711181083,
      "grad_norm": 0.8937102556228638,
      "learning_rate": 0.0001762916023475544,
      "loss": 3.6697,
      "step": 127170
    },
    {
      "epoch": 0.11855233872988992,
      "grad_norm": 1.0243841409683228,
      "learning_rate": 0.00017628973802545918,
      "loss": 3.8455,
      "step": 127180
    },
    {
      "epoch": 0.11856166034796901,
      "grad_norm": 0.8268987536430359,
      "learning_rate": 0.00017628787370336399,
      "loss": 3.6557,
      "step": 127190
    },
    {
      "epoch": 0.11857098196604811,
      "grad_norm": 1.1707855463027954,
      "learning_rate": 0.0001762860093812688,
      "loss": 3.5032,
      "step": 127200
    },
    {
      "epoch": 0.11858030358412719,
      "grad_norm": 1.1618274450302124,
      "learning_rate": 0.0001762841450591736,
      "loss": 3.7195,
      "step": 127210
    },
    {
      "epoch": 0.11858962520220628,
      "grad_norm": 0.90090012550354,
      "learning_rate": 0.0001762822807370784,
      "loss": 3.7392,
      "step": 127220
    },
    {
      "epoch": 0.11859894682028538,
      "grad_norm": 0.9404634237289429,
      "learning_rate": 0.00017628041641498318,
      "loss": 3.8017,
      "step": 127230
    },
    {
      "epoch": 0.11860826843836447,
      "grad_norm": 0.8540291786193848,
      "learning_rate": 0.000176278552092888,
      "loss": 3.6565,
      "step": 127240
    },
    {
      "epoch": 0.11861759005644357,
      "grad_norm": 1.0692410469055176,
      "learning_rate": 0.0001762766877707928,
      "loss": 3.5799,
      "step": 127250
    },
    {
      "epoch": 0.11862691167452265,
      "grad_norm": 1.0772641897201538,
      "learning_rate": 0.0001762748234486976,
      "loss": 3.5811,
      "step": 127260
    },
    {
      "epoch": 0.11863623329260174,
      "grad_norm": 0.9737386107444763,
      "learning_rate": 0.0001762729591266024,
      "loss": 3.7263,
      "step": 127270
    },
    {
      "epoch": 0.11864555491068084,
      "grad_norm": 0.8288321495056152,
      "learning_rate": 0.0001762710948045072,
      "loss": 3.8124,
      "step": 127280
    },
    {
      "epoch": 0.11865487652875993,
      "grad_norm": 0.992419958114624,
      "learning_rate": 0.000176269230482412,
      "loss": 3.7815,
      "step": 127290
    },
    {
      "epoch": 0.11866419814683903,
      "grad_norm": 1.3645923137664795,
      "learning_rate": 0.0001762673661603168,
      "loss": 3.5482,
      "step": 127300
    },
    {
      "epoch": 0.11867351976491812,
      "grad_norm": 1.0440272092819214,
      "learning_rate": 0.00017626550183822158,
      "loss": 3.7415,
      "step": 127310
    },
    {
      "epoch": 0.1186828413829972,
      "grad_norm": 0.9831867218017578,
      "learning_rate": 0.0001762636375161264,
      "loss": 3.812,
      "step": 127320
    },
    {
      "epoch": 0.1186921630010763,
      "grad_norm": 1.0516226291656494,
      "learning_rate": 0.0001762617731940312,
      "loss": 3.7376,
      "step": 127330
    },
    {
      "epoch": 0.11870148461915539,
      "grad_norm": 0.9729421734809875,
      "learning_rate": 0.000176259908871936,
      "loss": 3.6618,
      "step": 127340
    },
    {
      "epoch": 0.11871080623723448,
      "grad_norm": 1.7441717386245728,
      "learning_rate": 0.0001762580445498408,
      "loss": 3.6391,
      "step": 127350
    },
    {
      "epoch": 0.11872012785531358,
      "grad_norm": 1.2019487619400024,
      "learning_rate": 0.00017625618022774558,
      "loss": 3.5312,
      "step": 127360
    },
    {
      "epoch": 0.11872944947339266,
      "grad_norm": 1.0044326782226562,
      "learning_rate": 0.0001762543159056504,
      "loss": 3.6987,
      "step": 127370
    },
    {
      "epoch": 0.11873877109147175,
      "grad_norm": 0.9472842216491699,
      "learning_rate": 0.0001762524515835552,
      "loss": 3.6075,
      "step": 127380
    },
    {
      "epoch": 0.11874809270955085,
      "grad_norm": 1.3762685060501099,
      "learning_rate": 0.00017625058726146,
      "loss": 3.6474,
      "step": 127390
    },
    {
      "epoch": 0.11875741432762994,
      "grad_norm": 1.1950223445892334,
      "learning_rate": 0.0001762487229393648,
      "loss": 3.7438,
      "step": 127400
    },
    {
      "epoch": 0.11876673594570904,
      "grad_norm": 0.9370390772819519,
      "learning_rate": 0.0001762468586172696,
      "loss": 3.5822,
      "step": 127410
    },
    {
      "epoch": 0.11877605756378812,
      "grad_norm": 1.114024043083191,
      "learning_rate": 0.0001762449942951744,
      "loss": 3.8445,
      "step": 127420
    },
    {
      "epoch": 0.11878537918186721,
      "grad_norm": 1.0928754806518555,
      "learning_rate": 0.0001762431299730792,
      "loss": 3.6512,
      "step": 127430
    },
    {
      "epoch": 0.1187947007999463,
      "grad_norm": 1.2464308738708496,
      "learning_rate": 0.000176241265650984,
      "loss": 3.6102,
      "step": 127440
    },
    {
      "epoch": 0.1188040224180254,
      "grad_norm": 0.9225332140922546,
      "learning_rate": 0.0001762394013288888,
      "loss": 3.4752,
      "step": 127450
    },
    {
      "epoch": 0.1188133440361045,
      "grad_norm": 0.9403405785560608,
      "learning_rate": 0.0001762375370067936,
      "loss": 3.7223,
      "step": 127460
    },
    {
      "epoch": 0.11882266565418358,
      "grad_norm": 0.9923814535140991,
      "learning_rate": 0.0001762356726846984,
      "loss": 3.6456,
      "step": 127470
    },
    {
      "epoch": 0.11883198727226267,
      "grad_norm": 0.9562276005744934,
      "learning_rate": 0.0001762338083626032,
      "loss": 3.7054,
      "step": 127480
    },
    {
      "epoch": 0.11884130889034177,
      "grad_norm": 1.0468010902404785,
      "learning_rate": 0.000176231944040508,
      "loss": 3.8156,
      "step": 127490
    },
    {
      "epoch": 0.11885063050842086,
      "grad_norm": 0.829934298992157,
      "learning_rate": 0.0001762300797184128,
      "loss": 3.6598,
      "step": 127500
    },
    {
      "epoch": 0.11885995212649995,
      "grad_norm": 1.127429485321045,
      "learning_rate": 0.0001762282153963176,
      "loss": 3.6672,
      "step": 127510
    },
    {
      "epoch": 0.11886927374457905,
      "grad_norm": 0.9694768190383911,
      "learning_rate": 0.0001762263510742224,
      "loss": 3.759,
      "step": 127520
    },
    {
      "epoch": 0.11887859536265813,
      "grad_norm": 1.138008713722229,
      "learning_rate": 0.00017622448675212718,
      "loss": 3.5247,
      "step": 127530
    },
    {
      "epoch": 0.11888791698073722,
      "grad_norm": 0.953925609588623,
      "learning_rate": 0.00017622262243003202,
      "loss": 3.6062,
      "step": 127540
    },
    {
      "epoch": 0.11889723859881632,
      "grad_norm": 0.8648300170898438,
      "learning_rate": 0.0001762207581079368,
      "loss": 3.7237,
      "step": 127550
    },
    {
      "epoch": 0.11890656021689541,
      "grad_norm": 1.032384991645813,
      "learning_rate": 0.0001762188937858416,
      "loss": 3.6684,
      "step": 127560
    },
    {
      "epoch": 0.11891588183497451,
      "grad_norm": 1.0808186531066895,
      "learning_rate": 0.0001762170294637464,
      "loss": 3.7306,
      "step": 127570
    },
    {
      "epoch": 0.11892520345305359,
      "grad_norm": 0.8609361052513123,
      "learning_rate": 0.0001762151651416512,
      "loss": 3.6662,
      "step": 127580
    },
    {
      "epoch": 0.11893452507113268,
      "grad_norm": 1.0328922271728516,
      "learning_rate": 0.000176213300819556,
      "loss": 3.6435,
      "step": 127590
    },
    {
      "epoch": 0.11894384668921178,
      "grad_norm": 1.0190954208374023,
      "learning_rate": 0.0001762114364974608,
      "loss": 3.522,
      "step": 127600
    },
    {
      "epoch": 0.11895316830729087,
      "grad_norm": 1.1038146018981934,
      "learning_rate": 0.0001762095721753656,
      "loss": 3.4843,
      "step": 127610
    },
    {
      "epoch": 0.11896248992536997,
      "grad_norm": 1.1074222326278687,
      "learning_rate": 0.0001762077078532704,
      "loss": 3.5181,
      "step": 127620
    },
    {
      "epoch": 0.11897181154344905,
      "grad_norm": 1.5412298440933228,
      "learning_rate": 0.0001762058435311752,
      "loss": 3.5367,
      "step": 127630
    },
    {
      "epoch": 0.11898113316152814,
      "grad_norm": 0.9805348515510559,
      "learning_rate": 0.00017620397920908,
      "loss": 3.5367,
      "step": 127640
    },
    {
      "epoch": 0.11899045477960724,
      "grad_norm": 0.9713736176490784,
      "learning_rate": 0.0001762021148869848,
      "loss": 3.736,
      "step": 127650
    },
    {
      "epoch": 0.11899977639768633,
      "grad_norm": 0.8528377413749695,
      "learning_rate": 0.0001762002505648896,
      "loss": 3.7182,
      "step": 127660
    },
    {
      "epoch": 0.11900909801576542,
      "grad_norm": 1.1475824117660522,
      "learning_rate": 0.00017619838624279442,
      "loss": 3.7363,
      "step": 127670
    },
    {
      "epoch": 0.1190184196338445,
      "grad_norm": 1.0769643783569336,
      "learning_rate": 0.0001761965219206992,
      "loss": 3.4787,
      "step": 127680
    },
    {
      "epoch": 0.1190277412519236,
      "grad_norm": 1.1169933080673218,
      "learning_rate": 0.000176194657598604,
      "loss": 3.6292,
      "step": 127690
    },
    {
      "epoch": 0.1190370628700027,
      "grad_norm": 1.0721279382705688,
      "learning_rate": 0.00017619279327650878,
      "loss": 3.5659,
      "step": 127700
    },
    {
      "epoch": 0.11904638448808179,
      "grad_norm": 1.01203453540802,
      "learning_rate": 0.00017619092895441362,
      "loss": 3.5555,
      "step": 127710
    },
    {
      "epoch": 0.11905570610616088,
      "grad_norm": 1.2778639793395996,
      "learning_rate": 0.00017618906463231842,
      "loss": 3.6269,
      "step": 127720
    },
    {
      "epoch": 0.11906502772423998,
      "grad_norm": 1.056197166442871,
      "learning_rate": 0.0001761872003102232,
      "loss": 3.5875,
      "step": 127730
    },
    {
      "epoch": 0.11907434934231906,
      "grad_norm": 0.9366632699966431,
      "learning_rate": 0.000176185335988128,
      "loss": 3.7379,
      "step": 127740
    },
    {
      "epoch": 0.11908367096039815,
      "grad_norm": 1.0763320922851562,
      "learning_rate": 0.00017618347166603279,
      "loss": 3.6095,
      "step": 127750
    },
    {
      "epoch": 0.11909299257847725,
      "grad_norm": 0.9950545430183411,
      "learning_rate": 0.0001761816073439376,
      "loss": 3.5953,
      "step": 127760
    },
    {
      "epoch": 0.11910231419655634,
      "grad_norm": 1.0707788467407227,
      "learning_rate": 0.00017617974302184243,
      "loss": 3.6908,
      "step": 127770
    },
    {
      "epoch": 0.11911163581463544,
      "grad_norm": 1.0225856304168701,
      "learning_rate": 0.0001761778786997472,
      "loss": 3.4379,
      "step": 127780
    },
    {
      "epoch": 0.11912095743271452,
      "grad_norm": 1.1393932104110718,
      "learning_rate": 0.000176176014377652,
      "loss": 3.7636,
      "step": 127790
    },
    {
      "epoch": 0.11913027905079361,
      "grad_norm": 0.9481810927391052,
      "learning_rate": 0.0001761741500555568,
      "loss": 3.6886,
      "step": 127800
    },
    {
      "epoch": 0.1191396006688727,
      "grad_norm": 1.2220871448516846,
      "learning_rate": 0.0001761722857334616,
      "loss": 3.4817,
      "step": 127810
    },
    {
      "epoch": 0.1191489222869518,
      "grad_norm": 1.3148257732391357,
      "learning_rate": 0.0001761704214113664,
      "loss": 3.7594,
      "step": 127820
    },
    {
      "epoch": 0.1191582439050309,
      "grad_norm": 1.0177922248840332,
      "learning_rate": 0.0001761685570892712,
      "loss": 3.4783,
      "step": 127830
    },
    {
      "epoch": 0.11916756552310998,
      "grad_norm": 1.1394166946411133,
      "learning_rate": 0.00017616669276717602,
      "loss": 3.6039,
      "step": 127840
    },
    {
      "epoch": 0.11917688714118907,
      "grad_norm": 1.121200442314148,
      "learning_rate": 0.0001761648284450808,
      "loss": 3.6448,
      "step": 127850
    },
    {
      "epoch": 0.11918620875926816,
      "grad_norm": 1.1056981086730957,
      "learning_rate": 0.0001761629641229856,
      "loss": 3.5807,
      "step": 127860
    },
    {
      "epoch": 0.11919553037734726,
      "grad_norm": 0.8950501084327698,
      "learning_rate": 0.0001761610998008904,
      "loss": 3.5283,
      "step": 127870
    },
    {
      "epoch": 0.11920485199542635,
      "grad_norm": 0.9713159203529358,
      "learning_rate": 0.00017615923547879521,
      "loss": 3.6972,
      "step": 127880
    },
    {
      "epoch": 0.11921417361350543,
      "grad_norm": 1.0681921243667603,
      "learning_rate": 0.00017615737115670002,
      "loss": 3.7204,
      "step": 127890
    },
    {
      "epoch": 0.11922349523158453,
      "grad_norm": 1.1773194074630737,
      "learning_rate": 0.0001761555068346048,
      "loss": 3.7182,
      "step": 127900
    },
    {
      "epoch": 0.11923281684966362,
      "grad_norm": 0.9942113161087036,
      "learning_rate": 0.0001761536425125096,
      "loss": 3.7493,
      "step": 127910
    },
    {
      "epoch": 0.11924213846774272,
      "grad_norm": 1.0900704860687256,
      "learning_rate": 0.0001761517781904144,
      "loss": 3.5038,
      "step": 127920
    },
    {
      "epoch": 0.11925146008582181,
      "grad_norm": 1.0383093357086182,
      "learning_rate": 0.00017614991386831922,
      "loss": 3.8442,
      "step": 127930
    },
    {
      "epoch": 0.11926078170390089,
      "grad_norm": 1.1304171085357666,
      "learning_rate": 0.00017614804954622402,
      "loss": 3.631,
      "step": 127940
    },
    {
      "epoch": 0.11927010332197999,
      "grad_norm": 1.1466248035430908,
      "learning_rate": 0.0001761461852241288,
      "loss": 3.7427,
      "step": 127950
    },
    {
      "epoch": 0.11927942494005908,
      "grad_norm": 0.942615807056427,
      "learning_rate": 0.0001761443209020336,
      "loss": 3.4766,
      "step": 127960
    },
    {
      "epoch": 0.11928874655813818,
      "grad_norm": 1.2295966148376465,
      "learning_rate": 0.00017614245657993842,
      "loss": 3.7334,
      "step": 127970
    },
    {
      "epoch": 0.11929806817621727,
      "grad_norm": 1.1575027704238892,
      "learning_rate": 0.0001761405922578432,
      "loss": 3.4876,
      "step": 127980
    },
    {
      "epoch": 0.11930738979429636,
      "grad_norm": 1.1363035440444946,
      "learning_rate": 0.00017613872793574803,
      "loss": 3.7672,
      "step": 127990
    },
    {
      "epoch": 0.11931671141237545,
      "grad_norm": 0.9773939847946167,
      "learning_rate": 0.0001761368636136528,
      "loss": 3.5824,
      "step": 128000
    },
    {
      "epoch": 0.11932603303045454,
      "grad_norm": 1.1121466159820557,
      "learning_rate": 0.00017613499929155761,
      "loss": 3.6588,
      "step": 128010
    },
    {
      "epoch": 0.11933535464853363,
      "grad_norm": 1.0659654140472412,
      "learning_rate": 0.00017613313496946242,
      "loss": 3.6264,
      "step": 128020
    },
    {
      "epoch": 0.11934467626661273,
      "grad_norm": 1.1287391185760498,
      "learning_rate": 0.0001761312706473672,
      "loss": 3.5384,
      "step": 128030
    },
    {
      "epoch": 0.11935399788469182,
      "grad_norm": 1.016769528388977,
      "learning_rate": 0.000176129406325272,
      "loss": 3.5553,
      "step": 128040
    },
    {
      "epoch": 0.1193633195027709,
      "grad_norm": 1.1627018451690674,
      "learning_rate": 0.0001761275420031768,
      "loss": 3.7471,
      "step": 128050
    },
    {
      "epoch": 0.11937264112085,
      "grad_norm": 1.3477725982666016,
      "learning_rate": 0.00017612567768108162,
      "loss": 3.6505,
      "step": 128060
    },
    {
      "epoch": 0.11938196273892909,
      "grad_norm": 1.199165940284729,
      "learning_rate": 0.00017612381335898643,
      "loss": 3.545,
      "step": 128070
    },
    {
      "epoch": 0.11939128435700819,
      "grad_norm": 1.0209455490112305,
      "learning_rate": 0.0001761219490368912,
      "loss": 3.707,
      "step": 128080
    },
    {
      "epoch": 0.11940060597508728,
      "grad_norm": 0.991409957408905,
      "learning_rate": 0.000176120084714796,
      "loss": 3.6951,
      "step": 128090
    },
    {
      "epoch": 0.11940992759316636,
      "grad_norm": 1.1162314414978027,
      "learning_rate": 0.00017611822039270082,
      "loss": 3.7235,
      "step": 128100
    },
    {
      "epoch": 0.11941924921124546,
      "grad_norm": 1.0577465295791626,
      "learning_rate": 0.00017611635607060562,
      "loss": 3.548,
      "step": 128110
    },
    {
      "epoch": 0.11942857082932455,
      "grad_norm": 0.9421712756156921,
      "learning_rate": 0.00017611449174851043,
      "loss": 3.6502,
      "step": 128120
    },
    {
      "epoch": 0.11943789244740365,
      "grad_norm": 1.0651357173919678,
      "learning_rate": 0.0001761126274264152,
      "loss": 3.5185,
      "step": 128130
    },
    {
      "epoch": 0.11944721406548274,
      "grad_norm": 1.2046806812286377,
      "learning_rate": 0.00017611076310432002,
      "loss": 3.6106,
      "step": 128140
    },
    {
      "epoch": 0.11945653568356182,
      "grad_norm": 1.1250057220458984,
      "learning_rate": 0.0001761088987822248,
      "loss": 3.6328,
      "step": 128150
    },
    {
      "epoch": 0.11946585730164092,
      "grad_norm": 1.0040847063064575,
      "learning_rate": 0.00017610703446012963,
      "loss": 3.6584,
      "step": 128160
    },
    {
      "epoch": 0.11947517891972001,
      "grad_norm": 1.0214053392410278,
      "learning_rate": 0.00017610517013803443,
      "loss": 3.4972,
      "step": 128170
    },
    {
      "epoch": 0.1194845005377991,
      "grad_norm": 1.204142451286316,
      "learning_rate": 0.0001761033058159392,
      "loss": 3.6248,
      "step": 128180
    },
    {
      "epoch": 0.1194938221558782,
      "grad_norm": 1.1826860904693604,
      "learning_rate": 0.00017610144149384402,
      "loss": 3.9106,
      "step": 128190
    },
    {
      "epoch": 0.1195031437739573,
      "grad_norm": 1.07387113571167,
      "learning_rate": 0.0001760995771717488,
      "loss": 3.6531,
      "step": 128200
    },
    {
      "epoch": 0.11951246539203637,
      "grad_norm": 0.9457865357398987,
      "learning_rate": 0.0001760977128496536,
      "loss": 3.6635,
      "step": 128210
    },
    {
      "epoch": 0.11952178701011547,
      "grad_norm": 0.9788919687271118,
      "learning_rate": 0.00017609584852755844,
      "loss": 3.5532,
      "step": 128220
    },
    {
      "epoch": 0.11953110862819456,
      "grad_norm": 0.950505793094635,
      "learning_rate": 0.00017609398420546322,
      "loss": 3.5761,
      "step": 128230
    },
    {
      "epoch": 0.11954043024627366,
      "grad_norm": 0.9312366247177124,
      "learning_rate": 0.00017609211988336802,
      "loss": 3.5802,
      "step": 128240
    },
    {
      "epoch": 0.11954975186435275,
      "grad_norm": 0.9342807531356812,
      "learning_rate": 0.0001760902555612728,
      "loss": 3.6311,
      "step": 128250
    },
    {
      "epoch": 0.11955907348243183,
      "grad_norm": 1.1020166873931885,
      "learning_rate": 0.0001760883912391776,
      "loss": 3.7767,
      "step": 128260
    },
    {
      "epoch": 0.11956839510051093,
      "grad_norm": 1.3425266742706299,
      "learning_rate": 0.00017608652691708242,
      "loss": 3.5761,
      "step": 128270
    },
    {
      "epoch": 0.11957771671859002,
      "grad_norm": 1.180276870727539,
      "learning_rate": 0.00017608466259498722,
      "loss": 3.6928,
      "step": 128280
    },
    {
      "epoch": 0.11958703833666912,
      "grad_norm": 1.1069170236587524,
      "learning_rate": 0.00017608279827289203,
      "loss": 3.5782,
      "step": 128290
    },
    {
      "epoch": 0.11959635995474821,
      "grad_norm": 1.0756462812423706,
      "learning_rate": 0.0001760809339507968,
      "loss": 3.5994,
      "step": 128300
    },
    {
      "epoch": 0.11960568157282729,
      "grad_norm": 0.9894659519195557,
      "learning_rate": 0.00017607906962870161,
      "loss": 3.8054,
      "step": 128310
    },
    {
      "epoch": 0.11961500319090639,
      "grad_norm": 1.1582261323928833,
      "learning_rate": 0.00017607720530660642,
      "loss": 3.7949,
      "step": 128320
    },
    {
      "epoch": 0.11962432480898548,
      "grad_norm": 0.8898404240608215,
      "learning_rate": 0.00017607534098451123,
      "loss": 3.6861,
      "step": 128330
    },
    {
      "epoch": 0.11963364642706457,
      "grad_norm": 0.8469793796539307,
      "learning_rate": 0.00017607347666241603,
      "loss": 3.6249,
      "step": 128340
    },
    {
      "epoch": 0.11964296804514367,
      "grad_norm": 1.0233361721038818,
      "learning_rate": 0.0001760716123403208,
      "loss": 3.4884,
      "step": 128350
    },
    {
      "epoch": 0.11965228966322275,
      "grad_norm": 0.966124951839447,
      "learning_rate": 0.00017606974801822562,
      "loss": 3.757,
      "step": 128360
    },
    {
      "epoch": 0.11966161128130184,
      "grad_norm": 1.121186375617981,
      "learning_rate": 0.00017606788369613042,
      "loss": 3.6399,
      "step": 128370
    },
    {
      "epoch": 0.11967093289938094,
      "grad_norm": 1.035836935043335,
      "learning_rate": 0.00017606601937403523,
      "loss": 3.7259,
      "step": 128380
    },
    {
      "epoch": 0.11968025451746003,
      "grad_norm": 0.9569888114929199,
      "learning_rate": 0.00017606415505194004,
      "loss": 3.626,
      "step": 128390
    },
    {
      "epoch": 0.11968957613553913,
      "grad_norm": 0.9395502805709839,
      "learning_rate": 0.00017606229072984482,
      "loss": 3.5205,
      "step": 128400
    },
    {
      "epoch": 0.11969889775361822,
      "grad_norm": 1.0740033388137817,
      "learning_rate": 0.00017606042640774962,
      "loss": 3.6769,
      "step": 128410
    },
    {
      "epoch": 0.1197082193716973,
      "grad_norm": 1.1077371835708618,
      "learning_rate": 0.00017605856208565443,
      "loss": 3.7959,
      "step": 128420
    },
    {
      "epoch": 0.1197175409897764,
      "grad_norm": 1.0063492059707642,
      "learning_rate": 0.0001760566977635592,
      "loss": 3.5992,
      "step": 128430
    },
    {
      "epoch": 0.11972686260785549,
      "grad_norm": 0.9382206797599792,
      "learning_rate": 0.00017605483344146404,
      "loss": 3.8235,
      "step": 128440
    },
    {
      "epoch": 0.11973618422593459,
      "grad_norm": 0.9561017751693726,
      "learning_rate": 0.00017605296911936882,
      "loss": 3.5467,
      "step": 128450
    },
    {
      "epoch": 0.11974550584401368,
      "grad_norm": 1.1429104804992676,
      "learning_rate": 0.00017605110479727363,
      "loss": 3.4051,
      "step": 128460
    },
    {
      "epoch": 0.11975482746209276,
      "grad_norm": 1.1785694360733032,
      "learning_rate": 0.00017604924047517843,
      "loss": 3.5916,
      "step": 128470
    },
    {
      "epoch": 0.11976414908017186,
      "grad_norm": 1.007842779159546,
      "learning_rate": 0.0001760473761530832,
      "loss": 3.6329,
      "step": 128480
    },
    {
      "epoch": 0.11977347069825095,
      "grad_norm": 1.05196213722229,
      "learning_rate": 0.00017604551183098802,
      "loss": 3.6657,
      "step": 128490
    },
    {
      "epoch": 0.11978279231633004,
      "grad_norm": 1.1190959215164185,
      "learning_rate": 0.00017604364750889283,
      "loss": 3.796,
      "step": 128500
    },
    {
      "epoch": 0.11979211393440914,
      "grad_norm": 1.296353816986084,
      "learning_rate": 0.00017604178318679763,
      "loss": 3.6929,
      "step": 128510
    },
    {
      "epoch": 0.11980143555248822,
      "grad_norm": 0.9288484454154968,
      "learning_rate": 0.00017603991886470244,
      "loss": 3.6279,
      "step": 128520
    },
    {
      "epoch": 0.11981075717056731,
      "grad_norm": 1.1750396490097046,
      "learning_rate": 0.00017603805454260722,
      "loss": 3.5815,
      "step": 128530
    },
    {
      "epoch": 0.11982007878864641,
      "grad_norm": 1.0962719917297363,
      "learning_rate": 0.00017603619022051202,
      "loss": 3.6844,
      "step": 128540
    },
    {
      "epoch": 0.1198294004067255,
      "grad_norm": 1.0035099983215332,
      "learning_rate": 0.00017603432589841683,
      "loss": 3.709,
      "step": 128550
    },
    {
      "epoch": 0.1198387220248046,
      "grad_norm": 1.0437880754470825,
      "learning_rate": 0.00017603246157632164,
      "loss": 3.7564,
      "step": 128560
    },
    {
      "epoch": 0.11984804364288368,
      "grad_norm": 1.0384900569915771,
      "learning_rate": 0.00017603059725422644,
      "loss": 3.5808,
      "step": 128570
    },
    {
      "epoch": 0.11985736526096277,
      "grad_norm": 0.9479506611824036,
      "learning_rate": 0.00017602873293213122,
      "loss": 3.7279,
      "step": 128580
    },
    {
      "epoch": 0.11986668687904187,
      "grad_norm": 1.0112236738204956,
      "learning_rate": 0.00017602686861003603,
      "loss": 3.5358,
      "step": 128590
    },
    {
      "epoch": 0.11987600849712096,
      "grad_norm": 1.1404026746749878,
      "learning_rate": 0.0001760250042879408,
      "loss": 3.448,
      "step": 128600
    },
    {
      "epoch": 0.11988533011520006,
      "grad_norm": 1.117877721786499,
      "learning_rate": 0.00017602313996584564,
      "loss": 3.855,
      "step": 128610
    },
    {
      "epoch": 0.11989465173327915,
      "grad_norm": 1.025803804397583,
      "learning_rate": 0.00017602127564375045,
      "loss": 3.6442,
      "step": 128620
    },
    {
      "epoch": 0.11990397335135823,
      "grad_norm": 1.1991348266601562,
      "learning_rate": 0.00017601941132165523,
      "loss": 3.7875,
      "step": 128630
    },
    {
      "epoch": 0.11991329496943733,
      "grad_norm": 0.9524570107460022,
      "learning_rate": 0.00017601754699956003,
      "loss": 3.558,
      "step": 128640
    },
    {
      "epoch": 0.11992261658751642,
      "grad_norm": 0.9524158835411072,
      "learning_rate": 0.0001760156826774648,
      "loss": 3.5844,
      "step": 128650
    },
    {
      "epoch": 0.11993193820559551,
      "grad_norm": 1.1090574264526367,
      "learning_rate": 0.00017601381835536962,
      "loss": 3.5824,
      "step": 128660
    },
    {
      "epoch": 0.11994125982367461,
      "grad_norm": 0.9501005411148071,
      "learning_rate": 0.00017601195403327445,
      "loss": 3.6489,
      "step": 128670
    },
    {
      "epoch": 0.11995058144175369,
      "grad_norm": 1.297173261642456,
      "learning_rate": 0.00017601008971117923,
      "loss": 3.5816,
      "step": 128680
    },
    {
      "epoch": 0.11995990305983278,
      "grad_norm": 1.1004970073699951,
      "learning_rate": 0.00017600822538908404,
      "loss": 3.5765,
      "step": 128690
    },
    {
      "epoch": 0.11996922467791188,
      "grad_norm": 1.019680380821228,
      "learning_rate": 0.00017600636106698882,
      "loss": 3.6695,
      "step": 128700
    },
    {
      "epoch": 0.11997854629599097,
      "grad_norm": 1.1290262937545776,
      "learning_rate": 0.00017600449674489362,
      "loss": 3.5035,
      "step": 128710
    },
    {
      "epoch": 0.11998786791407007,
      "grad_norm": 1.035290241241455,
      "learning_rate": 0.00017600263242279843,
      "loss": 3.6313,
      "step": 128720
    },
    {
      "epoch": 0.11999718953214915,
      "grad_norm": 1.0556243658065796,
      "learning_rate": 0.00017600076810070324,
      "loss": 3.5987,
      "step": 128730
    },
    {
      "epoch": 0.12000651115022824,
      "grad_norm": 1.0365365743637085,
      "learning_rate": 0.00017599890377860804,
      "loss": 3.6937,
      "step": 128740
    },
    {
      "epoch": 0.12001583276830734,
      "grad_norm": 1.0512386560440063,
      "learning_rate": 0.00017599703945651282,
      "loss": 3.5528,
      "step": 128750
    },
    {
      "epoch": 0.12002515438638643,
      "grad_norm": 1.0443004369735718,
      "learning_rate": 0.00017599517513441763,
      "loss": 3.5799,
      "step": 128760
    },
    {
      "epoch": 0.12003447600446553,
      "grad_norm": 1.065734624862671,
      "learning_rate": 0.00017599331081232243,
      "loss": 3.7687,
      "step": 128770
    },
    {
      "epoch": 0.1200437976225446,
      "grad_norm": 1.1368542909622192,
      "learning_rate": 0.00017599144649022724,
      "loss": 3.6761,
      "step": 128780
    },
    {
      "epoch": 0.1200531192406237,
      "grad_norm": 0.9367719292640686,
      "learning_rate": 0.00017598958216813205,
      "loss": 3.7348,
      "step": 128790
    },
    {
      "epoch": 0.1200624408587028,
      "grad_norm": 1.1648091077804565,
      "learning_rate": 0.00017598771784603683,
      "loss": 3.6188,
      "step": 128800
    },
    {
      "epoch": 0.12007176247678189,
      "grad_norm": 1.2601418495178223,
      "learning_rate": 0.00017598585352394163,
      "loss": 3.6672,
      "step": 128810
    },
    {
      "epoch": 0.12008108409486098,
      "grad_norm": 1.1917920112609863,
      "learning_rate": 0.0001759839892018464,
      "loss": 3.5041,
      "step": 128820
    },
    {
      "epoch": 0.12009040571294007,
      "grad_norm": 1.2563843727111816,
      "learning_rate": 0.00017598212487975124,
      "loss": 3.6404,
      "step": 128830
    },
    {
      "epoch": 0.12009972733101916,
      "grad_norm": 1.1796783208847046,
      "learning_rate": 0.00017598026055765605,
      "loss": 3.5621,
      "step": 128840
    },
    {
      "epoch": 0.12010904894909825,
      "grad_norm": 1.0511493682861328,
      "learning_rate": 0.00017597839623556083,
      "loss": 3.7934,
      "step": 128850
    },
    {
      "epoch": 0.12011837056717735,
      "grad_norm": 1.103462815284729,
      "learning_rate": 0.00017597653191346564,
      "loss": 3.4792,
      "step": 128860
    },
    {
      "epoch": 0.12012769218525644,
      "grad_norm": 1.115983247756958,
      "learning_rate": 0.00017597466759137042,
      "loss": 3.8603,
      "step": 128870
    },
    {
      "epoch": 0.12013701380333554,
      "grad_norm": 0.9954465627670288,
      "learning_rate": 0.00017597280326927522,
      "loss": 4.0544,
      "step": 128880
    },
    {
      "epoch": 0.12014633542141462,
      "grad_norm": 0.9036950469017029,
      "learning_rate": 0.00017597093894718005,
      "loss": 3.7389,
      "step": 128890
    },
    {
      "epoch": 0.12015565703949371,
      "grad_norm": 1.0426995754241943,
      "learning_rate": 0.00017596907462508483,
      "loss": 3.6167,
      "step": 128900
    },
    {
      "epoch": 0.12016497865757281,
      "grad_norm": 1.0169861316680908,
      "learning_rate": 0.00017596721030298964,
      "loss": 3.6597,
      "step": 128910
    },
    {
      "epoch": 0.1201743002756519,
      "grad_norm": 1.0680360794067383,
      "learning_rate": 0.00017596534598089442,
      "loss": 3.5853,
      "step": 128920
    },
    {
      "epoch": 0.120183621893731,
      "grad_norm": 0.9550072550773621,
      "learning_rate": 0.00017596348165879923,
      "loss": 3.5705,
      "step": 128930
    },
    {
      "epoch": 0.12019294351181008,
      "grad_norm": 1.1453595161437988,
      "learning_rate": 0.00017596161733670403,
      "loss": 3.7276,
      "step": 128940
    },
    {
      "epoch": 0.12020226512988917,
      "grad_norm": 1.1825250387191772,
      "learning_rate": 0.00017595975301460884,
      "loss": 3.7237,
      "step": 128950
    },
    {
      "epoch": 0.12021158674796827,
      "grad_norm": 1.094770073890686,
      "learning_rate": 0.00017595788869251364,
      "loss": 3.6692,
      "step": 128960
    },
    {
      "epoch": 0.12022090836604736,
      "grad_norm": 1.074187159538269,
      "learning_rate": 0.00017595602437041842,
      "loss": 3.3858,
      "step": 128970
    },
    {
      "epoch": 0.12023022998412645,
      "grad_norm": 0.9925696849822998,
      "learning_rate": 0.00017595416004832323,
      "loss": 3.6533,
      "step": 128980
    },
    {
      "epoch": 0.12023955160220554,
      "grad_norm": 1.2053704261779785,
      "learning_rate": 0.00017595229572622804,
      "loss": 3.5733,
      "step": 128990
    },
    {
      "epoch": 0.12024887322028463,
      "grad_norm": 1.123471736907959,
      "learning_rate": 0.00017595043140413284,
      "loss": 3.7821,
      "step": 129000
    },
    {
      "epoch": 0.12025819483836372,
      "grad_norm": 1.1383732557296753,
      "learning_rate": 0.00017594856708203765,
      "loss": 3.6475,
      "step": 129010
    },
    {
      "epoch": 0.12026751645644282,
      "grad_norm": 0.943210780620575,
      "learning_rate": 0.00017594670275994243,
      "loss": 3.6731,
      "step": 129020
    },
    {
      "epoch": 0.12027683807452191,
      "grad_norm": 1.13300359249115,
      "learning_rate": 0.00017594483843784723,
      "loss": 3.816,
      "step": 129030
    },
    {
      "epoch": 0.120286159692601,
      "grad_norm": 1.0579547882080078,
      "learning_rate": 0.00017594297411575204,
      "loss": 3.5263,
      "step": 129040
    },
    {
      "epoch": 0.12029548131068009,
      "grad_norm": 1.0996118783950806,
      "learning_rate": 0.00017594110979365682,
      "loss": 3.6993,
      "step": 129050
    },
    {
      "epoch": 0.12030480292875918,
      "grad_norm": 1.118561863899231,
      "learning_rate": 0.00017593924547156165,
      "loss": 3.6909,
      "step": 129060
    },
    {
      "epoch": 0.12031412454683828,
      "grad_norm": 1.120147943496704,
      "learning_rate": 0.00017593738114946643,
      "loss": 3.6741,
      "step": 129070
    },
    {
      "epoch": 0.12032344616491737,
      "grad_norm": 0.9645351767539978,
      "learning_rate": 0.00017593551682737124,
      "loss": 3.6799,
      "step": 129080
    },
    {
      "epoch": 0.12033276778299647,
      "grad_norm": 1.2792479991912842,
      "learning_rate": 0.00017593365250527605,
      "loss": 3.8986,
      "step": 129090
    },
    {
      "epoch": 0.12034208940107555,
      "grad_norm": 0.9758214950561523,
      "learning_rate": 0.00017593178818318082,
      "loss": 3.5361,
      "step": 129100
    },
    {
      "epoch": 0.12035141101915464,
      "grad_norm": 1.246898889541626,
      "learning_rate": 0.00017592992386108563,
      "loss": 3.5425,
      "step": 129110
    },
    {
      "epoch": 0.12036073263723374,
      "grad_norm": 1.1295597553253174,
      "learning_rate": 0.00017592805953899044,
      "loss": 3.6035,
      "step": 129120
    },
    {
      "epoch": 0.12037005425531283,
      "grad_norm": 1.258718729019165,
      "learning_rate": 0.00017592619521689524,
      "loss": 3.6933,
      "step": 129130
    },
    {
      "epoch": 0.12037937587339193,
      "grad_norm": 1.1413381099700928,
      "learning_rate": 0.00017592433089480005,
      "loss": 3.7559,
      "step": 129140
    },
    {
      "epoch": 0.120388697491471,
      "grad_norm": 0.9987457990646362,
      "learning_rate": 0.00017592246657270483,
      "loss": 3.864,
      "step": 129150
    },
    {
      "epoch": 0.1203980191095501,
      "grad_norm": 1.119312047958374,
      "learning_rate": 0.00017592060225060964,
      "loss": 3.7062,
      "step": 129160
    },
    {
      "epoch": 0.1204073407276292,
      "grad_norm": 1.269895076751709,
      "learning_rate": 0.00017591873792851444,
      "loss": 3.4572,
      "step": 129170
    },
    {
      "epoch": 0.12041666234570829,
      "grad_norm": 1.2763952016830444,
      "learning_rate": 0.00017591687360641925,
      "loss": 3.4746,
      "step": 129180
    },
    {
      "epoch": 0.12042598396378738,
      "grad_norm": 1.1342511177062988,
      "learning_rate": 0.00017591500928432405,
      "loss": 3.6546,
      "step": 129190
    },
    {
      "epoch": 0.12043530558186646,
      "grad_norm": 1.106102705001831,
      "learning_rate": 0.00017591314496222883,
      "loss": 3.6766,
      "step": 129200
    },
    {
      "epoch": 0.12044462719994556,
      "grad_norm": 0.953296959400177,
      "learning_rate": 0.00017591128064013364,
      "loss": 3.6479,
      "step": 129210
    },
    {
      "epoch": 0.12045394881802465,
      "grad_norm": 0.9597297310829163,
      "learning_rate": 0.00017590941631803845,
      "loss": 3.7836,
      "step": 129220
    },
    {
      "epoch": 0.12046327043610375,
      "grad_norm": 1.134360432624817,
      "learning_rate": 0.00017590755199594325,
      "loss": 3.8377,
      "step": 129230
    },
    {
      "epoch": 0.12047259205418284,
      "grad_norm": 0.7634374499320984,
      "learning_rate": 0.00017590568767384806,
      "loss": 3.704,
      "step": 129240
    },
    {
      "epoch": 0.12048191367226192,
      "grad_norm": 1.1422725915908813,
      "learning_rate": 0.00017590382335175284,
      "loss": 3.5725,
      "step": 129250
    },
    {
      "epoch": 0.12049123529034102,
      "grad_norm": 0.9539543390274048,
      "learning_rate": 0.00017590195902965764,
      "loss": 3.6982,
      "step": 129260
    },
    {
      "epoch": 0.12050055690842011,
      "grad_norm": 1.0391850471496582,
      "learning_rate": 0.00017590009470756242,
      "loss": 3.563,
      "step": 129270
    },
    {
      "epoch": 0.1205098785264992,
      "grad_norm": 1.4800957441329956,
      "learning_rate": 0.00017589823038546726,
      "loss": 3.4627,
      "step": 129280
    },
    {
      "epoch": 0.1205192001445783,
      "grad_norm": 1.2339966297149658,
      "learning_rate": 0.00017589636606337206,
      "loss": 3.7363,
      "step": 129290
    },
    {
      "epoch": 0.1205285217626574,
      "grad_norm": 1.3220231533050537,
      "learning_rate": 0.00017589450174127684,
      "loss": 3.5506,
      "step": 129300
    },
    {
      "epoch": 0.12053784338073648,
      "grad_norm": 1.0157477855682373,
      "learning_rate": 0.00017589263741918165,
      "loss": 3.6338,
      "step": 129310
    },
    {
      "epoch": 0.12054716499881557,
      "grad_norm": 1.0510705709457397,
      "learning_rate": 0.00017589077309708643,
      "loss": 3.6704,
      "step": 129320
    },
    {
      "epoch": 0.12055648661689466,
      "grad_norm": 1.1264148950576782,
      "learning_rate": 0.00017588890877499123,
      "loss": 3.6199,
      "step": 129330
    },
    {
      "epoch": 0.12056580823497376,
      "grad_norm": 0.9781911969184875,
      "learning_rate": 0.00017588704445289607,
      "loss": 3.9055,
      "step": 129340
    },
    {
      "epoch": 0.12057512985305285,
      "grad_norm": 0.9657960534095764,
      "learning_rate": 0.00017588518013080085,
      "loss": 3.6154,
      "step": 129350
    },
    {
      "epoch": 0.12058445147113193,
      "grad_norm": 0.9718291163444519,
      "learning_rate": 0.00017588331580870565,
      "loss": 3.7082,
      "step": 129360
    },
    {
      "epoch": 0.12059377308921103,
      "grad_norm": 1.0214440822601318,
      "learning_rate": 0.00017588145148661043,
      "loss": 3.7973,
      "step": 129370
    },
    {
      "epoch": 0.12060309470729012,
      "grad_norm": 0.9695146679878235,
      "learning_rate": 0.00017587958716451524,
      "loss": 3.4327,
      "step": 129380
    },
    {
      "epoch": 0.12061241632536922,
      "grad_norm": 1.055912733078003,
      "learning_rate": 0.00017587772284242004,
      "loss": 3.6125,
      "step": 129390
    },
    {
      "epoch": 0.12062173794344831,
      "grad_norm": 0.9697468876838684,
      "learning_rate": 0.00017587585852032485,
      "loss": 3.6809,
      "step": 129400
    },
    {
      "epoch": 0.12063105956152739,
      "grad_norm": 1.010872483253479,
      "learning_rate": 0.00017587399419822966,
      "loss": 3.6495,
      "step": 129410
    },
    {
      "epoch": 0.12064038117960649,
      "grad_norm": 1.091186285018921,
      "learning_rate": 0.00017587212987613444,
      "loss": 3.5674,
      "step": 129420
    },
    {
      "epoch": 0.12064970279768558,
      "grad_norm": 0.8418741226196289,
      "learning_rate": 0.00017587026555403924,
      "loss": 3.8137,
      "step": 129430
    },
    {
      "epoch": 0.12065902441576468,
      "grad_norm": 1.1026726961135864,
      "learning_rate": 0.00017586840123194405,
      "loss": 3.7808,
      "step": 129440
    },
    {
      "epoch": 0.12066834603384377,
      "grad_norm": 1.013029932975769,
      "learning_rate": 0.00017586653690984886,
      "loss": 3.68,
      "step": 129450
    },
    {
      "epoch": 0.12067766765192285,
      "grad_norm": 0.981503963470459,
      "learning_rate": 0.00017586467258775366,
      "loss": 3.8352,
      "step": 129460
    },
    {
      "epoch": 0.12068698927000195,
      "grad_norm": 1.3851714134216309,
      "learning_rate": 0.00017586280826565844,
      "loss": 3.5443,
      "step": 129470
    },
    {
      "epoch": 0.12069631088808104,
      "grad_norm": 1.189347743988037,
      "learning_rate": 0.00017586094394356325,
      "loss": 3.6094,
      "step": 129480
    },
    {
      "epoch": 0.12070563250616013,
      "grad_norm": 1.1417903900146484,
      "learning_rate": 0.00017585907962146805,
      "loss": 3.6293,
      "step": 129490
    },
    {
      "epoch": 0.12071495412423923,
      "grad_norm": 0.9828134179115295,
      "learning_rate": 0.00017585721529937283,
      "loss": 3.5025,
      "step": 129500
    },
    {
      "epoch": 0.12072427574231831,
      "grad_norm": 1.007866382598877,
      "learning_rate": 0.00017585535097727767,
      "loss": 3.699,
      "step": 129510
    },
    {
      "epoch": 0.1207335973603974,
      "grad_norm": 0.8471770286560059,
      "learning_rate": 0.00017585348665518245,
      "loss": 3.5813,
      "step": 129520
    },
    {
      "epoch": 0.1207429189784765,
      "grad_norm": 1.0439916849136353,
      "learning_rate": 0.00017585162233308725,
      "loss": 3.706,
      "step": 129530
    },
    {
      "epoch": 0.1207522405965556,
      "grad_norm": 1.2083485126495361,
      "learning_rate": 0.00017584975801099206,
      "loss": 3.5362,
      "step": 129540
    },
    {
      "epoch": 0.12076156221463469,
      "grad_norm": 1.2020057439804077,
      "learning_rate": 0.00017584789368889684,
      "loss": 3.4753,
      "step": 129550
    },
    {
      "epoch": 0.12077088383271378,
      "grad_norm": 1.0319162607192993,
      "learning_rate": 0.00017584602936680164,
      "loss": 3.4932,
      "step": 129560
    },
    {
      "epoch": 0.12078020545079286,
      "grad_norm": 1.0571329593658447,
      "learning_rate": 0.00017584416504470645,
      "loss": 3.6343,
      "step": 129570
    },
    {
      "epoch": 0.12078952706887196,
      "grad_norm": 1.1215455532073975,
      "learning_rate": 0.00017584230072261126,
      "loss": 3.8152,
      "step": 129580
    },
    {
      "epoch": 0.12079884868695105,
      "grad_norm": 0.9959794878959656,
      "learning_rate": 0.00017584043640051606,
      "loss": 3.5458,
      "step": 129590
    },
    {
      "epoch": 0.12080817030503015,
      "grad_norm": 1.0924110412597656,
      "learning_rate": 0.00017583857207842084,
      "loss": 3.6573,
      "step": 129600
    },
    {
      "epoch": 0.12081749192310924,
      "grad_norm": 0.9967131018638611,
      "learning_rate": 0.00017583670775632565,
      "loss": 3.7995,
      "step": 129610
    },
    {
      "epoch": 0.12082681354118832,
      "grad_norm": 1.073503017425537,
      "learning_rate": 0.00017583484343423045,
      "loss": 3.5643,
      "step": 129620
    },
    {
      "epoch": 0.12083613515926742,
      "grad_norm": 1.0038316249847412,
      "learning_rate": 0.00017583297911213526,
      "loss": 3.7123,
      "step": 129630
    },
    {
      "epoch": 0.12084545677734651,
      "grad_norm": 0.9066917300224304,
      "learning_rate": 0.00017583111479004007,
      "loss": 3.6218,
      "step": 129640
    },
    {
      "epoch": 0.1208547783954256,
      "grad_norm": 0.855309784412384,
      "learning_rate": 0.00017582925046794485,
      "loss": 3.5966,
      "step": 129650
    },
    {
      "epoch": 0.1208641000135047,
      "grad_norm": 1.059466004371643,
      "learning_rate": 0.00017582738614584965,
      "loss": 3.7442,
      "step": 129660
    },
    {
      "epoch": 0.12087342163158378,
      "grad_norm": 1.1638693809509277,
      "learning_rate": 0.00017582552182375446,
      "loss": 3.8272,
      "step": 129670
    },
    {
      "epoch": 0.12088274324966287,
      "grad_norm": 1.172295093536377,
      "learning_rate": 0.00017582365750165927,
      "loss": 3.7014,
      "step": 129680
    },
    {
      "epoch": 0.12089206486774197,
      "grad_norm": 1.0712145566940308,
      "learning_rate": 0.00017582179317956407,
      "loss": 3.7216,
      "step": 129690
    },
    {
      "epoch": 0.12090138648582106,
      "grad_norm": 1.009612798690796,
      "learning_rate": 0.00017581992885746885,
      "loss": 3.6658,
      "step": 129700
    },
    {
      "epoch": 0.12091070810390016,
      "grad_norm": 0.8449264168739319,
      "learning_rate": 0.00017581806453537366,
      "loss": 3.6998,
      "step": 129710
    },
    {
      "epoch": 0.12092002972197924,
      "grad_norm": 1.2396833896636963,
      "learning_rate": 0.00017581620021327844,
      "loss": 3.7025,
      "step": 129720
    },
    {
      "epoch": 0.12092935134005833,
      "grad_norm": 1.1602931022644043,
      "learning_rate": 0.00017581433589118327,
      "loss": 3.5273,
      "step": 129730
    },
    {
      "epoch": 0.12093867295813743,
      "grad_norm": 0.915359616279602,
      "learning_rate": 0.00017581247156908808,
      "loss": 3.5995,
      "step": 129740
    },
    {
      "epoch": 0.12094799457621652,
      "grad_norm": 0.7671923041343689,
      "learning_rate": 0.00017581060724699286,
      "loss": 3.6454,
      "step": 129750
    },
    {
      "epoch": 0.12095731619429562,
      "grad_norm": 1.1770440340042114,
      "learning_rate": 0.00017580874292489766,
      "loss": 3.6558,
      "step": 129760
    },
    {
      "epoch": 0.12096663781237471,
      "grad_norm": 1.1294763088226318,
      "learning_rate": 0.00017580687860280244,
      "loss": 3.5544,
      "step": 129770
    },
    {
      "epoch": 0.12097595943045379,
      "grad_norm": 0.9258444309234619,
      "learning_rate": 0.00017580501428070725,
      "loss": 3.724,
      "step": 129780
    },
    {
      "epoch": 0.12098528104853289,
      "grad_norm": 1.1247532367706299,
      "learning_rate": 0.00017580314995861208,
      "loss": 3.5009,
      "step": 129790
    },
    {
      "epoch": 0.12099460266661198,
      "grad_norm": 1.0494980812072754,
      "learning_rate": 0.00017580128563651686,
      "loss": 3.6533,
      "step": 129800
    },
    {
      "epoch": 0.12100392428469108,
      "grad_norm": 1.1451482772827148,
      "learning_rate": 0.00017579942131442167,
      "loss": 3.5378,
      "step": 129810
    },
    {
      "epoch": 0.12101324590277017,
      "grad_norm": 1.0683764219284058,
      "learning_rate": 0.00017579755699232645,
      "loss": 3.7294,
      "step": 129820
    },
    {
      "epoch": 0.12102256752084925,
      "grad_norm": 1.1746447086334229,
      "learning_rate": 0.00017579569267023125,
      "loss": 3.5727,
      "step": 129830
    },
    {
      "epoch": 0.12103188913892834,
      "grad_norm": 0.9804233908653259,
      "learning_rate": 0.00017579382834813606,
      "loss": 3.5708,
      "step": 129840
    },
    {
      "epoch": 0.12104121075700744,
      "grad_norm": 1.2714953422546387,
      "learning_rate": 0.00017579196402604086,
      "loss": 3.552,
      "step": 129850
    },
    {
      "epoch": 0.12105053237508653,
      "grad_norm": 1.1388763189315796,
      "learning_rate": 0.00017579009970394567,
      "loss": 3.7279,
      "step": 129860
    },
    {
      "epoch": 0.12105985399316563,
      "grad_norm": 1.1552659273147583,
      "learning_rate": 0.00017578823538185045,
      "loss": 3.6181,
      "step": 129870
    },
    {
      "epoch": 0.12106917561124471,
      "grad_norm": 1.2124369144439697,
      "learning_rate": 0.00017578637105975526,
      "loss": 3.6156,
      "step": 129880
    },
    {
      "epoch": 0.1210784972293238,
      "grad_norm": 1.1616538763046265,
      "learning_rate": 0.00017578450673766006,
      "loss": 3.5571,
      "step": 129890
    },
    {
      "epoch": 0.1210878188474029,
      "grad_norm": 0.9242284297943115,
      "learning_rate": 0.00017578264241556487,
      "loss": 3.5478,
      "step": 129900
    },
    {
      "epoch": 0.12109714046548199,
      "grad_norm": 0.9904388785362244,
      "learning_rate": 0.00017578077809346967,
      "loss": 3.4016,
      "step": 129910
    },
    {
      "epoch": 0.12110646208356109,
      "grad_norm": 1.0968822240829468,
      "learning_rate": 0.00017577891377137445,
      "loss": 3.611,
      "step": 129920
    },
    {
      "epoch": 0.12111578370164017,
      "grad_norm": 1.3214269876480103,
      "learning_rate": 0.00017577704944927926,
      "loss": 3.6109,
      "step": 129930
    },
    {
      "epoch": 0.12112510531971926,
      "grad_norm": 0.9451707005500793,
      "learning_rate": 0.00017577518512718407,
      "loss": 3.7679,
      "step": 129940
    },
    {
      "epoch": 0.12113442693779836,
      "grad_norm": 1.0025278329849243,
      "learning_rate": 0.00017577332080508885,
      "loss": 3.5859,
      "step": 129950
    },
    {
      "epoch": 0.12114374855587745,
      "grad_norm": 0.9076146483421326,
      "learning_rate": 0.00017577145648299368,
      "loss": 3.4846,
      "step": 129960
    },
    {
      "epoch": 0.12115307017395655,
      "grad_norm": 0.8974772691726685,
      "learning_rate": 0.00017576959216089846,
      "loss": 3.4694,
      "step": 129970
    },
    {
      "epoch": 0.12116239179203564,
      "grad_norm": 1.0967482328414917,
      "learning_rate": 0.00017576772783880326,
      "loss": 3.5586,
      "step": 129980
    },
    {
      "epoch": 0.12117171341011472,
      "grad_norm": 1.0113012790679932,
      "learning_rate": 0.00017576586351670807,
      "loss": 3.5468,
      "step": 129990
    },
    {
      "epoch": 0.12118103502819381,
      "grad_norm": 1.0052860975265503,
      "learning_rate": 0.00017576399919461285,
      "loss": 3.7718,
      "step": 130000
    },
    {
      "epoch": 0.12119035664627291,
      "grad_norm": 1.083101511001587,
      "learning_rate": 0.00017576213487251766,
      "loss": 3.5056,
      "step": 130010
    },
    {
      "epoch": 0.121199678264352,
      "grad_norm": 0.9384697079658508,
      "learning_rate": 0.00017576027055042246,
      "loss": 3.5307,
      "step": 130020
    },
    {
      "epoch": 0.1212089998824311,
      "grad_norm": 1.113457202911377,
      "learning_rate": 0.00017575840622832727,
      "loss": 3.6011,
      "step": 130030
    },
    {
      "epoch": 0.12121832150051018,
      "grad_norm": 1.1052922010421753,
      "learning_rate": 0.00017575654190623208,
      "loss": 3.7522,
      "step": 130040
    },
    {
      "epoch": 0.12122764311858927,
      "grad_norm": 1.0229626893997192,
      "learning_rate": 0.00017575467758413685,
      "loss": 3.561,
      "step": 130050
    },
    {
      "epoch": 0.12123696473666837,
      "grad_norm": 0.8555023074150085,
      "learning_rate": 0.00017575281326204166,
      "loss": 3.6062,
      "step": 130060
    },
    {
      "epoch": 0.12124628635474746,
      "grad_norm": 1.0540715456008911,
      "learning_rate": 0.00017575094893994647,
      "loss": 3.5321,
      "step": 130070
    },
    {
      "epoch": 0.12125560797282656,
      "grad_norm": 0.9968523979187012,
      "learning_rate": 0.00017574908461785127,
      "loss": 3.6072,
      "step": 130080
    },
    {
      "epoch": 0.12126492959090564,
      "grad_norm": 0.8686766624450684,
      "learning_rate": 0.00017574722029575608,
      "loss": 3.5034,
      "step": 130090
    },
    {
      "epoch": 0.12127425120898473,
      "grad_norm": 1.0596550703048706,
      "learning_rate": 0.00017574535597366086,
      "loss": 3.5573,
      "step": 130100
    },
    {
      "epoch": 0.12128357282706383,
      "grad_norm": 1.0706164836883545,
      "learning_rate": 0.00017574349165156567,
      "loss": 3.4553,
      "step": 130110
    },
    {
      "epoch": 0.12129289444514292,
      "grad_norm": 1.0459805727005005,
      "learning_rate": 0.00017574162732947047,
      "loss": 3.5645,
      "step": 130120
    },
    {
      "epoch": 0.12130221606322202,
      "grad_norm": 1.3851487636566162,
      "learning_rate": 0.00017573976300737528,
      "loss": 3.7036,
      "step": 130130
    },
    {
      "epoch": 0.1213115376813011,
      "grad_norm": 1.0372626781463623,
      "learning_rate": 0.00017573789868528008,
      "loss": 3.4819,
      "step": 130140
    },
    {
      "epoch": 0.12132085929938019,
      "grad_norm": 0.9296603202819824,
      "learning_rate": 0.00017573603436318486,
      "loss": 3.5902,
      "step": 130150
    },
    {
      "epoch": 0.12133018091745928,
      "grad_norm": 1.0296803712844849,
      "learning_rate": 0.00017573417004108967,
      "loss": 3.5744,
      "step": 130160
    },
    {
      "epoch": 0.12133950253553838,
      "grad_norm": 1.0282992124557495,
      "learning_rate": 0.00017573230571899445,
      "loss": 3.6444,
      "step": 130170
    },
    {
      "epoch": 0.12134882415361747,
      "grad_norm": 1.0004427433013916,
      "learning_rate": 0.00017573044139689928,
      "loss": 3.5897,
      "step": 130180
    },
    {
      "epoch": 0.12135814577169655,
      "grad_norm": 0.9934120178222656,
      "learning_rate": 0.00017572857707480406,
      "loss": 3.4247,
      "step": 130190
    },
    {
      "epoch": 0.12136746738977565,
      "grad_norm": 0.9702591896057129,
      "learning_rate": 0.00017572671275270887,
      "loss": 3.8344,
      "step": 130200
    },
    {
      "epoch": 0.12137678900785474,
      "grad_norm": 1.3509103059768677,
      "learning_rate": 0.00017572484843061367,
      "loss": 3.5682,
      "step": 130210
    },
    {
      "epoch": 0.12138611062593384,
      "grad_norm": 1.0227545499801636,
      "learning_rate": 0.00017572298410851845,
      "loss": 3.8238,
      "step": 130220
    },
    {
      "epoch": 0.12139543224401293,
      "grad_norm": 1.0239888429641724,
      "learning_rate": 0.00017572111978642326,
      "loss": 3.6712,
      "step": 130230
    },
    {
      "epoch": 0.12140475386209203,
      "grad_norm": 0.8970332741737366,
      "learning_rate": 0.00017571925546432807,
      "loss": 3.5678,
      "step": 130240
    },
    {
      "epoch": 0.12141407548017111,
      "grad_norm": 1.1385531425476074,
      "learning_rate": 0.00017571739114223287,
      "loss": 3.5362,
      "step": 130250
    },
    {
      "epoch": 0.1214233970982502,
      "grad_norm": 1.0327969789505005,
      "learning_rate": 0.00017571552682013768,
      "loss": 3.6782,
      "step": 130260
    },
    {
      "epoch": 0.1214327187163293,
      "grad_norm": 0.9873246550559998,
      "learning_rate": 0.00017571366249804246,
      "loss": 3.658,
      "step": 130270
    },
    {
      "epoch": 0.12144204033440839,
      "grad_norm": 1.0068025588989258,
      "learning_rate": 0.00017571179817594726,
      "loss": 3.6665,
      "step": 130280
    },
    {
      "epoch": 0.12145136195248749,
      "grad_norm": 0.8711782097816467,
      "learning_rate": 0.00017570993385385207,
      "loss": 3.6012,
      "step": 130290
    },
    {
      "epoch": 0.12146068357056657,
      "grad_norm": 1.1401609182357788,
      "learning_rate": 0.00017570806953175688,
      "loss": 3.7266,
      "step": 130300
    },
    {
      "epoch": 0.12147000518864566,
      "grad_norm": 1.1035362482070923,
      "learning_rate": 0.00017570620520966168,
      "loss": 3.5766,
      "step": 130310
    },
    {
      "epoch": 0.12147932680672476,
      "grad_norm": 1.0156446695327759,
      "learning_rate": 0.00017570434088756646,
      "loss": 3.5698,
      "step": 130320
    },
    {
      "epoch": 0.12148864842480385,
      "grad_norm": 1.1710737943649292,
      "learning_rate": 0.00017570247656547127,
      "loss": 3.4685,
      "step": 130330
    },
    {
      "epoch": 0.12149797004288294,
      "grad_norm": 1.024937391281128,
      "learning_rate": 0.00017570061224337605,
      "loss": 3.487,
      "step": 130340
    },
    {
      "epoch": 0.12150729166096202,
      "grad_norm": 1.0015432834625244,
      "learning_rate": 0.00017569874792128088,
      "loss": 3.6578,
      "step": 130350
    },
    {
      "epoch": 0.12151661327904112,
      "grad_norm": 1.0699597597122192,
      "learning_rate": 0.0001756968835991857,
      "loss": 3.6943,
      "step": 130360
    },
    {
      "epoch": 0.12152593489712021,
      "grad_norm": 1.0252174139022827,
      "learning_rate": 0.00017569501927709047,
      "loss": 3.6779,
      "step": 130370
    },
    {
      "epoch": 0.12153525651519931,
      "grad_norm": 1.1173491477966309,
      "learning_rate": 0.00017569315495499527,
      "loss": 3.8164,
      "step": 130380
    },
    {
      "epoch": 0.1215445781332784,
      "grad_norm": 1.3401597738265991,
      "learning_rate": 0.00017569129063290005,
      "loss": 3.6072,
      "step": 130390
    },
    {
      "epoch": 0.12155389975135748,
      "grad_norm": 1.1434175968170166,
      "learning_rate": 0.00017568942631080486,
      "loss": 3.7156,
      "step": 130400
    },
    {
      "epoch": 0.12156322136943658,
      "grad_norm": 1.0223007202148438,
      "learning_rate": 0.0001756875619887097,
      "loss": 3.6137,
      "step": 130410
    },
    {
      "epoch": 0.12157254298751567,
      "grad_norm": 1.1369988918304443,
      "learning_rate": 0.00017568569766661447,
      "loss": 3.6366,
      "step": 130420
    },
    {
      "epoch": 0.12158186460559477,
      "grad_norm": 1.1682792901992798,
      "learning_rate": 0.00017568383334451928,
      "loss": 3.6059,
      "step": 130430
    },
    {
      "epoch": 0.12159118622367386,
      "grad_norm": 1.221169114112854,
      "learning_rate": 0.00017568196902242406,
      "loss": 3.4194,
      "step": 130440
    },
    {
      "epoch": 0.12160050784175296,
      "grad_norm": 1.0737935304641724,
      "learning_rate": 0.00017568010470032886,
      "loss": 3.6473,
      "step": 130450
    },
    {
      "epoch": 0.12160982945983204,
      "grad_norm": 1.3840336799621582,
      "learning_rate": 0.00017567824037823367,
      "loss": 3.5982,
      "step": 130460
    },
    {
      "epoch": 0.12161915107791113,
      "grad_norm": 1.2503461837768555,
      "learning_rate": 0.00017567637605613848,
      "loss": 3.6138,
      "step": 130470
    },
    {
      "epoch": 0.12162847269599023,
      "grad_norm": 1.1640276908874512,
      "learning_rate": 0.00017567451173404328,
      "loss": 3.5441,
      "step": 130480
    },
    {
      "epoch": 0.12163779431406932,
      "grad_norm": 0.8833040595054626,
      "learning_rate": 0.00017567264741194806,
      "loss": 3.6977,
      "step": 130490
    },
    {
      "epoch": 0.12164711593214841,
      "grad_norm": 0.8925062417984009,
      "learning_rate": 0.00017567078308985287,
      "loss": 3.6197,
      "step": 130500
    },
    {
      "epoch": 0.1216564375502275,
      "grad_norm": 1.3080840110778809,
      "learning_rate": 0.00017566891876775767,
      "loss": 3.7057,
      "step": 130510
    },
    {
      "epoch": 0.12166575916830659,
      "grad_norm": 0.9205510020256042,
      "learning_rate": 0.00017566705444566248,
      "loss": 3.6937,
      "step": 130520
    },
    {
      "epoch": 0.12167508078638568,
      "grad_norm": 1.0068162679672241,
      "learning_rate": 0.0001756651901235673,
      "loss": 3.8031,
      "step": 130530
    },
    {
      "epoch": 0.12168440240446478,
      "grad_norm": 1.1727452278137207,
      "learning_rate": 0.00017566332580147207,
      "loss": 3.4856,
      "step": 130540
    },
    {
      "epoch": 0.12169372402254387,
      "grad_norm": 1.1520496606826782,
      "learning_rate": 0.00017566146147937687,
      "loss": 3.7284,
      "step": 130550
    },
    {
      "epoch": 0.12170304564062295,
      "grad_norm": 1.1557161808013916,
      "learning_rate": 0.00017565959715728168,
      "loss": 3.6099,
      "step": 130560
    },
    {
      "epoch": 0.12171236725870205,
      "grad_norm": 0.9974095225334167,
      "learning_rate": 0.00017565773283518648,
      "loss": 3.6338,
      "step": 130570
    },
    {
      "epoch": 0.12172168887678114,
      "grad_norm": 0.8766019940376282,
      "learning_rate": 0.0001756558685130913,
      "loss": 3.7744,
      "step": 130580
    },
    {
      "epoch": 0.12173101049486024,
      "grad_norm": 0.9942687749862671,
      "learning_rate": 0.00017565400419099607,
      "loss": 3.5071,
      "step": 130590
    },
    {
      "epoch": 0.12174033211293933,
      "grad_norm": 1.1175925731658936,
      "learning_rate": 0.00017565213986890088,
      "loss": 3.5844,
      "step": 130600
    },
    {
      "epoch": 0.12174965373101841,
      "grad_norm": 1.2117035388946533,
      "learning_rate": 0.00017565027554680568,
      "loss": 3.693,
      "step": 130610
    },
    {
      "epoch": 0.1217589753490975,
      "grad_norm": 1.391219139099121,
      "learning_rate": 0.00017564841122471046,
      "loss": 3.6454,
      "step": 130620
    },
    {
      "epoch": 0.1217682969671766,
      "grad_norm": 1.1430439949035645,
      "learning_rate": 0.0001756465469026153,
      "loss": 3.8519,
      "step": 130630
    },
    {
      "epoch": 0.1217776185852557,
      "grad_norm": 0.9780706167221069,
      "learning_rate": 0.00017564468258052007,
      "loss": 3.602,
      "step": 130640
    },
    {
      "epoch": 0.12178694020333479,
      "grad_norm": 1.1124637126922607,
      "learning_rate": 0.00017564281825842488,
      "loss": 3.7943,
      "step": 130650
    },
    {
      "epoch": 0.12179626182141388,
      "grad_norm": 0.9141437411308289,
      "learning_rate": 0.0001756409539363297,
      "loss": 3.7759,
      "step": 130660
    },
    {
      "epoch": 0.12180558343949296,
      "grad_norm": 1.5386539697647095,
      "learning_rate": 0.00017563908961423447,
      "loss": 3.5431,
      "step": 130670
    },
    {
      "epoch": 0.12181490505757206,
      "grad_norm": 1.1412662267684937,
      "learning_rate": 0.00017563722529213927,
      "loss": 3.6343,
      "step": 130680
    },
    {
      "epoch": 0.12182422667565115,
      "grad_norm": 1.0850379467010498,
      "learning_rate": 0.00017563536097004408,
      "loss": 3.7309,
      "step": 130690
    },
    {
      "epoch": 0.12183354829373025,
      "grad_norm": 1.1609644889831543,
      "learning_rate": 0.00017563349664794889,
      "loss": 3.5371,
      "step": 130700
    },
    {
      "epoch": 0.12184286991180934,
      "grad_norm": 1.0386079549789429,
      "learning_rate": 0.0001756316323258537,
      "loss": 3.725,
      "step": 130710
    },
    {
      "epoch": 0.12185219152988842,
      "grad_norm": 0.9462617635726929,
      "learning_rate": 0.00017562976800375847,
      "loss": 3.7061,
      "step": 130720
    },
    {
      "epoch": 0.12186151314796752,
      "grad_norm": 1.114621877670288,
      "learning_rate": 0.00017562790368166328,
      "loss": 3.6245,
      "step": 130730
    },
    {
      "epoch": 0.12187083476604661,
      "grad_norm": 1.0626580715179443,
      "learning_rate": 0.00017562603935956808,
      "loss": 3.7238,
      "step": 130740
    },
    {
      "epoch": 0.1218801563841257,
      "grad_norm": 0.9311881065368652,
      "learning_rate": 0.0001756241750374729,
      "loss": 3.4804,
      "step": 130750
    },
    {
      "epoch": 0.1218894780022048,
      "grad_norm": 0.9979332089424133,
      "learning_rate": 0.0001756223107153777,
      "loss": 3.6232,
      "step": 130760
    },
    {
      "epoch": 0.12189879962028388,
      "grad_norm": 1.2112782001495361,
      "learning_rate": 0.00017562044639328248,
      "loss": 3.5644,
      "step": 130770
    },
    {
      "epoch": 0.12190812123836298,
      "grad_norm": 0.9375501871109009,
      "learning_rate": 0.00017561858207118728,
      "loss": 3.5246,
      "step": 130780
    },
    {
      "epoch": 0.12191744285644207,
      "grad_norm": 0.9417252540588379,
      "learning_rate": 0.0001756167177490921,
      "loss": 3.7007,
      "step": 130790
    },
    {
      "epoch": 0.12192676447452117,
      "grad_norm": 1.0752383470535278,
      "learning_rate": 0.0001756148534269969,
      "loss": 3.7588,
      "step": 130800
    },
    {
      "epoch": 0.12193608609260026,
      "grad_norm": 1.1027694940567017,
      "learning_rate": 0.0001756129891049017,
      "loss": 3.5553,
      "step": 130810
    },
    {
      "epoch": 0.12194540771067934,
      "grad_norm": 0.8822851181030273,
      "learning_rate": 0.00017561112478280648,
      "loss": 3.8074,
      "step": 130820
    },
    {
      "epoch": 0.12195472932875843,
      "grad_norm": 1.1525623798370361,
      "learning_rate": 0.00017560926046071129,
      "loss": 3.4463,
      "step": 130830
    },
    {
      "epoch": 0.12196405094683753,
      "grad_norm": 1.1348850727081299,
      "learning_rate": 0.00017560739613861607,
      "loss": 3.6657,
      "step": 130840
    },
    {
      "epoch": 0.12197337256491662,
      "grad_norm": 1.1580321788787842,
      "learning_rate": 0.0001756055318165209,
      "loss": 3.776,
      "step": 130850
    },
    {
      "epoch": 0.12198269418299572,
      "grad_norm": 1.2653661966323853,
      "learning_rate": 0.0001756036674944257,
      "loss": 3.5398,
      "step": 130860
    },
    {
      "epoch": 0.1219920158010748,
      "grad_norm": 1.0804541110992432,
      "learning_rate": 0.00017560180317233048,
      "loss": 3.7386,
      "step": 130870
    },
    {
      "epoch": 0.1220013374191539,
      "grad_norm": 1.054423213005066,
      "learning_rate": 0.0001755999388502353,
      "loss": 3.5892,
      "step": 130880
    },
    {
      "epoch": 0.12201065903723299,
      "grad_norm": 0.9601938724517822,
      "learning_rate": 0.00017559807452814007,
      "loss": 3.5706,
      "step": 130890
    },
    {
      "epoch": 0.12201998065531208,
      "grad_norm": 1.2302894592285156,
      "learning_rate": 0.00017559621020604488,
      "loss": 3.5644,
      "step": 130900
    },
    {
      "epoch": 0.12202930227339118,
      "grad_norm": 0.9925941824913025,
      "learning_rate": 0.00017559434588394968,
      "loss": 3.7529,
      "step": 130910
    },
    {
      "epoch": 0.12203862389147027,
      "grad_norm": 0.9927709698677063,
      "learning_rate": 0.0001755924815618545,
      "loss": 3.4836,
      "step": 130920
    },
    {
      "epoch": 0.12204794550954935,
      "grad_norm": 1.1119126081466675,
      "learning_rate": 0.0001755906172397593,
      "loss": 3.5019,
      "step": 130930
    },
    {
      "epoch": 0.12205726712762845,
      "grad_norm": 1.0738013982772827,
      "learning_rate": 0.00017558875291766407,
      "loss": 3.6235,
      "step": 130940
    },
    {
      "epoch": 0.12206658874570754,
      "grad_norm": 1.1419891119003296,
      "learning_rate": 0.00017558688859556888,
      "loss": 3.7631,
      "step": 130950
    },
    {
      "epoch": 0.12207591036378664,
      "grad_norm": 1.0807788372039795,
      "learning_rate": 0.0001755850242734737,
      "loss": 3.707,
      "step": 130960
    },
    {
      "epoch": 0.12208523198186573,
      "grad_norm": 1.2523932456970215,
      "learning_rate": 0.0001755831599513785,
      "loss": 3.6478,
      "step": 130970
    },
    {
      "epoch": 0.12209455359994481,
      "grad_norm": 0.9899513721466064,
      "learning_rate": 0.0001755812956292833,
      "loss": 3.4775,
      "step": 130980
    },
    {
      "epoch": 0.1221038752180239,
      "grad_norm": 0.9451441764831543,
      "learning_rate": 0.00017557943130718808,
      "loss": 3.6198,
      "step": 130990
    },
    {
      "epoch": 0.122113196836103,
      "grad_norm": 1.1855241060256958,
      "learning_rate": 0.00017557756698509288,
      "loss": 3.4851,
      "step": 131000
    },
    {
      "epoch": 0.1221225184541821,
      "grad_norm": 1.10093092918396,
      "learning_rate": 0.0001755757026629977,
      "loss": 3.6495,
      "step": 131010
    },
    {
      "epoch": 0.12213184007226119,
      "grad_norm": 1.0985432863235474,
      "learning_rate": 0.0001755738383409025,
      "loss": 3.6863,
      "step": 131020
    },
    {
      "epoch": 0.12214116169034027,
      "grad_norm": 0.9483815431594849,
      "learning_rate": 0.0001755719740188073,
      "loss": 3.7844,
      "step": 131030
    },
    {
      "epoch": 0.12215048330841936,
      "grad_norm": 0.8142539858818054,
      "learning_rate": 0.00017557010969671208,
      "loss": 3.4596,
      "step": 131040
    },
    {
      "epoch": 0.12215980492649846,
      "grad_norm": 1.0435914993286133,
      "learning_rate": 0.0001755682453746169,
      "loss": 3.6026,
      "step": 131050
    },
    {
      "epoch": 0.12216912654457755,
      "grad_norm": 1.0595223903656006,
      "learning_rate": 0.0001755663810525217,
      "loss": 3.3999,
      "step": 131060
    },
    {
      "epoch": 0.12217844816265665,
      "grad_norm": 1.0346593856811523,
      "learning_rate": 0.00017556451673042647,
      "loss": 3.4905,
      "step": 131070
    },
    {
      "epoch": 0.12218776978073573,
      "grad_norm": 1.1429044008255005,
      "learning_rate": 0.0001755626524083313,
      "loss": 3.7064,
      "step": 131080
    },
    {
      "epoch": 0.12219709139881482,
      "grad_norm": 0.911392092704773,
      "learning_rate": 0.0001755607880862361,
      "loss": 3.6579,
      "step": 131090
    },
    {
      "epoch": 0.12220641301689392,
      "grad_norm": 1.1379313468933105,
      "learning_rate": 0.0001755589237641409,
      "loss": 3.6265,
      "step": 131100
    },
    {
      "epoch": 0.12221573463497301,
      "grad_norm": 1.404171109199524,
      "learning_rate": 0.0001755570594420457,
      "loss": 3.4935,
      "step": 131110
    },
    {
      "epoch": 0.1222250562530521,
      "grad_norm": 1.1727657318115234,
      "learning_rate": 0.00017555519511995048,
      "loss": 3.6846,
      "step": 131120
    },
    {
      "epoch": 0.1222343778711312,
      "grad_norm": 1.3165645599365234,
      "learning_rate": 0.00017555333079785529,
      "loss": 3.6663,
      "step": 131130
    },
    {
      "epoch": 0.12224369948921028,
      "grad_norm": 1.067590355873108,
      "learning_rate": 0.0001755514664757601,
      "loss": 3.4977,
      "step": 131140
    },
    {
      "epoch": 0.12225302110728938,
      "grad_norm": 1.0048638582229614,
      "learning_rate": 0.0001755496021536649,
      "loss": 3.9056,
      "step": 131150
    },
    {
      "epoch": 0.12226234272536847,
      "grad_norm": 1.3175734281539917,
      "learning_rate": 0.0001755477378315697,
      "loss": 3.6104,
      "step": 131160
    },
    {
      "epoch": 0.12227166434344756,
      "grad_norm": 0.983647882938385,
      "learning_rate": 0.00017554587350947448,
      "loss": 3.8231,
      "step": 131170
    },
    {
      "epoch": 0.12228098596152666,
      "grad_norm": 1.2746639251708984,
      "learning_rate": 0.0001755440091873793,
      "loss": 3.7337,
      "step": 131180
    },
    {
      "epoch": 0.12229030757960574,
      "grad_norm": 1.0740300416946411,
      "learning_rate": 0.0001755421448652841,
      "loss": 3.6489,
      "step": 131190
    },
    {
      "epoch": 0.12229962919768483,
      "grad_norm": 1.2040247917175293,
      "learning_rate": 0.0001755402805431889,
      "loss": 3.5701,
      "step": 131200
    },
    {
      "epoch": 0.12230895081576393,
      "grad_norm": 1.3508124351501465,
      "learning_rate": 0.0001755384162210937,
      "loss": 3.81,
      "step": 131210
    },
    {
      "epoch": 0.12231827243384302,
      "grad_norm": 1.167702555656433,
      "learning_rate": 0.0001755365518989985,
      "loss": 3.6205,
      "step": 131220
    },
    {
      "epoch": 0.12232759405192212,
      "grad_norm": 0.9542905688285828,
      "learning_rate": 0.0001755346875769033,
      "loss": 3.5098,
      "step": 131230
    },
    {
      "epoch": 0.1223369156700012,
      "grad_norm": 1.0860830545425415,
      "learning_rate": 0.0001755328232548081,
      "loss": 3.4393,
      "step": 131240
    },
    {
      "epoch": 0.12234623728808029,
      "grad_norm": 1.1568734645843506,
      "learning_rate": 0.0001755309589327129,
      "loss": 3.6952,
      "step": 131250
    },
    {
      "epoch": 0.12235555890615939,
      "grad_norm": 1.0501856803894043,
      "learning_rate": 0.0001755290946106177,
      "loss": 3.7386,
      "step": 131260
    },
    {
      "epoch": 0.12236488052423848,
      "grad_norm": 1.1504783630371094,
      "learning_rate": 0.0001755272302885225,
      "loss": 3.6741,
      "step": 131270
    },
    {
      "epoch": 0.12237420214231758,
      "grad_norm": 1.0868178606033325,
      "learning_rate": 0.0001755253659664273,
      "loss": 3.6478,
      "step": 131280
    },
    {
      "epoch": 0.12238352376039666,
      "grad_norm": 1.016153335571289,
      "learning_rate": 0.00017552350164433208,
      "loss": 3.6832,
      "step": 131290
    },
    {
      "epoch": 0.12239284537847575,
      "grad_norm": 1.086226224899292,
      "learning_rate": 0.0001755216373222369,
      "loss": 3.7367,
      "step": 131300
    },
    {
      "epoch": 0.12240216699655485,
      "grad_norm": 0.8844786882400513,
      "learning_rate": 0.00017551977300014172,
      "loss": 3.6579,
      "step": 131310
    },
    {
      "epoch": 0.12241148861463394,
      "grad_norm": 1.0622913837432861,
      "learning_rate": 0.0001755179086780465,
      "loss": 3.528,
      "step": 131320
    },
    {
      "epoch": 0.12242081023271303,
      "grad_norm": 1.0709228515625,
      "learning_rate": 0.0001755160443559513,
      "loss": 3.8362,
      "step": 131330
    },
    {
      "epoch": 0.12243013185079213,
      "grad_norm": 1.2857671976089478,
      "learning_rate": 0.00017551418003385608,
      "loss": 3.6826,
      "step": 131340
    },
    {
      "epoch": 0.12243945346887121,
      "grad_norm": 0.8623893857002258,
      "learning_rate": 0.0001755123157117609,
      "loss": 3.722,
      "step": 131350
    },
    {
      "epoch": 0.1224487750869503,
      "grad_norm": 1.1167670488357544,
      "learning_rate": 0.00017551045138966572,
      "loss": 3.5838,
      "step": 131360
    },
    {
      "epoch": 0.1224580967050294,
      "grad_norm": 1.1965184211730957,
      "learning_rate": 0.0001755085870675705,
      "loss": 3.6148,
      "step": 131370
    },
    {
      "epoch": 0.12246741832310849,
      "grad_norm": 1.035229206085205,
      "learning_rate": 0.0001755067227454753,
      "loss": 3.7226,
      "step": 131380
    },
    {
      "epoch": 0.12247673994118759,
      "grad_norm": 0.9897264838218689,
      "learning_rate": 0.0001755048584233801,
      "loss": 3.6138,
      "step": 131390
    },
    {
      "epoch": 0.12248606155926667,
      "grad_norm": 1.3775668144226074,
      "learning_rate": 0.0001755029941012849,
      "loss": 3.7196,
      "step": 131400
    },
    {
      "epoch": 0.12249538317734576,
      "grad_norm": 1.015520453453064,
      "learning_rate": 0.0001755011297791897,
      "loss": 3.6797,
      "step": 131410
    },
    {
      "epoch": 0.12250470479542486,
      "grad_norm": 1.0283299684524536,
      "learning_rate": 0.0001754992654570945,
      "loss": 3.6101,
      "step": 131420
    },
    {
      "epoch": 0.12251402641350395,
      "grad_norm": 0.980323851108551,
      "learning_rate": 0.0001754974011349993,
      "loss": 3.5266,
      "step": 131430
    },
    {
      "epoch": 0.12252334803158305,
      "grad_norm": 1.0089373588562012,
      "learning_rate": 0.0001754955368129041,
      "loss": 3.4511,
      "step": 131440
    },
    {
      "epoch": 0.12253266964966213,
      "grad_norm": 1.180551528930664,
      "learning_rate": 0.0001754936724908089,
      "loss": 3.7094,
      "step": 131450
    },
    {
      "epoch": 0.12254199126774122,
      "grad_norm": 1.1817691326141357,
      "learning_rate": 0.0001754918081687137,
      "loss": 3.7068,
      "step": 131460
    },
    {
      "epoch": 0.12255131288582032,
      "grad_norm": 0.9528151154518127,
      "learning_rate": 0.0001754899438466185,
      "loss": 3.7038,
      "step": 131470
    },
    {
      "epoch": 0.12256063450389941,
      "grad_norm": 0.9950529932975769,
      "learning_rate": 0.00017548807952452332,
      "loss": 3.7131,
      "step": 131480
    },
    {
      "epoch": 0.1225699561219785,
      "grad_norm": 1.3277924060821533,
      "learning_rate": 0.0001754862152024281,
      "loss": 3.686,
      "step": 131490
    },
    {
      "epoch": 0.12257927774005759,
      "grad_norm": 0.9921277165412903,
      "learning_rate": 0.0001754843508803329,
      "loss": 3.6204,
      "step": 131500
    },
    {
      "epoch": 0.12258859935813668,
      "grad_norm": 1.0428638458251953,
      "learning_rate": 0.0001754824865582377,
      "loss": 3.6057,
      "step": 131510
    },
    {
      "epoch": 0.12259792097621577,
      "grad_norm": 1.0858396291732788,
      "learning_rate": 0.0001754806222361425,
      "loss": 3.7438,
      "step": 131520
    },
    {
      "epoch": 0.12260724259429487,
      "grad_norm": 0.8615316152572632,
      "learning_rate": 0.00017547875791404732,
      "loss": 3.545,
      "step": 131530
    },
    {
      "epoch": 0.12261656421237396,
      "grad_norm": 1.0171178579330444,
      "learning_rate": 0.0001754768935919521,
      "loss": 3.4422,
      "step": 131540
    },
    {
      "epoch": 0.12262588583045304,
      "grad_norm": 1.049585223197937,
      "learning_rate": 0.0001754750292698569,
      "loss": 3.7121,
      "step": 131550
    },
    {
      "epoch": 0.12263520744853214,
      "grad_norm": 0.9525637030601501,
      "learning_rate": 0.00017547316494776169,
      "loss": 3.5776,
      "step": 131560
    },
    {
      "epoch": 0.12264452906661123,
      "grad_norm": 1.0373530387878418,
      "learning_rate": 0.0001754713006256665,
      "loss": 3.4077,
      "step": 131570
    },
    {
      "epoch": 0.12265385068469033,
      "grad_norm": 1.0407944917678833,
      "learning_rate": 0.0001754694363035713,
      "loss": 3.5915,
      "step": 131580
    },
    {
      "epoch": 0.12266317230276942,
      "grad_norm": 1.0105819702148438,
      "learning_rate": 0.0001754675719814761,
      "loss": 3.6295,
      "step": 131590
    },
    {
      "epoch": 0.12267249392084852,
      "grad_norm": 1.1271835565567017,
      "learning_rate": 0.0001754657076593809,
      "loss": 3.6549,
      "step": 131600
    },
    {
      "epoch": 0.1226818155389276,
      "grad_norm": 1.0293514728546143,
      "learning_rate": 0.0001754638433372857,
      "loss": 3.7676,
      "step": 131610
    },
    {
      "epoch": 0.12269113715700669,
      "grad_norm": 1.0504229068756104,
      "learning_rate": 0.0001754619790151905,
      "loss": 3.5879,
      "step": 131620
    },
    {
      "epoch": 0.12270045877508579,
      "grad_norm": 0.9819664359092712,
      "learning_rate": 0.0001754601146930953,
      "loss": 3.7515,
      "step": 131630
    },
    {
      "epoch": 0.12270978039316488,
      "grad_norm": 1.148915410041809,
      "learning_rate": 0.0001754582503710001,
      "loss": 3.7229,
      "step": 131640
    },
    {
      "epoch": 0.12271910201124397,
      "grad_norm": 1.1672793626785278,
      "learning_rate": 0.00017545638604890492,
      "loss": 3.5043,
      "step": 131650
    },
    {
      "epoch": 0.12272842362932306,
      "grad_norm": 1.0830565690994263,
      "learning_rate": 0.0001754545217268097,
      "loss": 3.5066,
      "step": 131660
    },
    {
      "epoch": 0.12273774524740215,
      "grad_norm": 1.008072018623352,
      "learning_rate": 0.0001754526574047145,
      "loss": 3.4777,
      "step": 131670
    },
    {
      "epoch": 0.12274706686548124,
      "grad_norm": 1.4412652254104614,
      "learning_rate": 0.0001754507930826193,
      "loss": 3.5758,
      "step": 131680
    },
    {
      "epoch": 0.12275638848356034,
      "grad_norm": 0.916793704032898,
      "learning_rate": 0.00017544892876052411,
      "loss": 3.735,
      "step": 131690
    },
    {
      "epoch": 0.12276571010163943,
      "grad_norm": 0.887885570526123,
      "learning_rate": 0.00017544706443842892,
      "loss": 3.7208,
      "step": 131700
    },
    {
      "epoch": 0.12277503171971851,
      "grad_norm": 1.131018877029419,
      "learning_rate": 0.0001754452001163337,
      "loss": 3.4944,
      "step": 131710
    },
    {
      "epoch": 0.12278435333779761,
      "grad_norm": 1.1441673040390015,
      "learning_rate": 0.0001754433357942385,
      "loss": 3.5956,
      "step": 131720
    },
    {
      "epoch": 0.1227936749558767,
      "grad_norm": 0.9061900973320007,
      "learning_rate": 0.0001754414714721433,
      "loss": 3.611,
      "step": 131730
    },
    {
      "epoch": 0.1228029965739558,
      "grad_norm": 1.0670900344848633,
      "learning_rate": 0.0001754396071500481,
      "loss": 3.6281,
      "step": 131740
    },
    {
      "epoch": 0.12281231819203489,
      "grad_norm": 1.1914702653884888,
      "learning_rate": 0.00017543774282795292,
      "loss": 3.6695,
      "step": 131750
    },
    {
      "epoch": 0.12282163981011397,
      "grad_norm": 1.057942509651184,
      "learning_rate": 0.0001754358785058577,
      "loss": 3.7647,
      "step": 131760
    },
    {
      "epoch": 0.12283096142819307,
      "grad_norm": 0.9766467809677124,
      "learning_rate": 0.0001754340141837625,
      "loss": 3.5823,
      "step": 131770
    },
    {
      "epoch": 0.12284028304627216,
      "grad_norm": 0.9743143916130066,
      "learning_rate": 0.00017543214986166732,
      "loss": 3.546,
      "step": 131780
    },
    {
      "epoch": 0.12284960466435126,
      "grad_norm": 1.0744915008544922,
      "learning_rate": 0.0001754302855395721,
      "loss": 3.6841,
      "step": 131790
    },
    {
      "epoch": 0.12285892628243035,
      "grad_norm": 1.1866536140441895,
      "learning_rate": 0.0001754284212174769,
      "loss": 3.6562,
      "step": 131800
    },
    {
      "epoch": 0.12286824790050944,
      "grad_norm": 1.0541579723358154,
      "learning_rate": 0.0001754265568953817,
      "loss": 3.6168,
      "step": 131810
    },
    {
      "epoch": 0.12287756951858853,
      "grad_norm": 0.892989456653595,
      "learning_rate": 0.00017542469257328651,
      "loss": 3.6914,
      "step": 131820
    },
    {
      "epoch": 0.12288689113666762,
      "grad_norm": 0.9838952422142029,
      "learning_rate": 0.00017542282825119132,
      "loss": 3.6073,
      "step": 131830
    },
    {
      "epoch": 0.12289621275474671,
      "grad_norm": 1.207017421722412,
      "learning_rate": 0.0001754209639290961,
      "loss": 3.8065,
      "step": 131840
    },
    {
      "epoch": 0.12290553437282581,
      "grad_norm": 1.088984489440918,
      "learning_rate": 0.0001754190996070009,
      "loss": 3.5071,
      "step": 131850
    },
    {
      "epoch": 0.1229148559909049,
      "grad_norm": 1.1187454462051392,
      "learning_rate": 0.0001754172352849057,
      "loss": 3.6456,
      "step": 131860
    },
    {
      "epoch": 0.12292417760898398,
      "grad_norm": 0.9183536171913147,
      "learning_rate": 0.00017541537096281052,
      "loss": 3.6091,
      "step": 131870
    },
    {
      "epoch": 0.12293349922706308,
      "grad_norm": 1.076238989830017,
      "learning_rate": 0.00017541350664071533,
      "loss": 3.4266,
      "step": 131880
    },
    {
      "epoch": 0.12294282084514217,
      "grad_norm": 0.999420702457428,
      "learning_rate": 0.0001754116423186201,
      "loss": 3.695,
      "step": 131890
    },
    {
      "epoch": 0.12295214246322127,
      "grad_norm": 0.9260642528533936,
      "learning_rate": 0.0001754097779965249,
      "loss": 3.6881,
      "step": 131900
    },
    {
      "epoch": 0.12296146408130036,
      "grad_norm": 1.0907238721847534,
      "learning_rate": 0.0001754079136744297,
      "loss": 3.6161,
      "step": 131910
    },
    {
      "epoch": 0.12297078569937944,
      "grad_norm": 1.071152925491333,
      "learning_rate": 0.00017540604935233452,
      "loss": 3.6342,
      "step": 131920
    },
    {
      "epoch": 0.12298010731745854,
      "grad_norm": 0.8967026472091675,
      "learning_rate": 0.00017540418503023933,
      "loss": 3.5779,
      "step": 131930
    },
    {
      "epoch": 0.12298942893553763,
      "grad_norm": 1.184858798980713,
      "learning_rate": 0.0001754023207081441,
      "loss": 3.4574,
      "step": 131940
    },
    {
      "epoch": 0.12299875055361673,
      "grad_norm": 1.270565390586853,
      "learning_rate": 0.00017540045638604892,
      "loss": 3.7084,
      "step": 131950
    },
    {
      "epoch": 0.12300807217169582,
      "grad_norm": 0.9657249450683594,
      "learning_rate": 0.0001753985920639537,
      "loss": 3.5019,
      "step": 131960
    },
    {
      "epoch": 0.1230173937897749,
      "grad_norm": 1.0608367919921875,
      "learning_rate": 0.0001753967277418585,
      "loss": 3.6151,
      "step": 131970
    },
    {
      "epoch": 0.123026715407854,
      "grad_norm": 1.1386858224868774,
      "learning_rate": 0.00017539486341976333,
      "loss": 3.6705,
      "step": 131980
    },
    {
      "epoch": 0.12303603702593309,
      "grad_norm": 0.9812125563621521,
      "learning_rate": 0.0001753929990976681,
      "loss": 3.7265,
      "step": 131990
    },
    {
      "epoch": 0.12304535864401218,
      "grad_norm": 1.0379729270935059,
      "learning_rate": 0.00017539113477557292,
      "loss": 3.7034,
      "step": 132000
    },
    {
      "epoch": 0.12305468026209128,
      "grad_norm": 0.8093095421791077,
      "learning_rate": 0.0001753892704534777,
      "loss": 3.6376,
      "step": 132010
    },
    {
      "epoch": 0.12306400188017037,
      "grad_norm": 0.9535566568374634,
      "learning_rate": 0.0001753874061313825,
      "loss": 3.6072,
      "step": 132020
    },
    {
      "epoch": 0.12307332349824945,
      "grad_norm": 0.9876165986061096,
      "learning_rate": 0.0001753855418092873,
      "loss": 3.5234,
      "step": 132030
    },
    {
      "epoch": 0.12308264511632855,
      "grad_norm": 1.0588140487670898,
      "learning_rate": 0.00017538367748719212,
      "loss": 3.4468,
      "step": 132040
    },
    {
      "epoch": 0.12309196673440764,
      "grad_norm": 0.90489262342453,
      "learning_rate": 0.00017538181316509692,
      "loss": 3.5245,
      "step": 132050
    },
    {
      "epoch": 0.12310128835248674,
      "grad_norm": 1.052234411239624,
      "learning_rate": 0.0001753799488430017,
      "loss": 3.6356,
      "step": 132060
    },
    {
      "epoch": 0.12311060997056583,
      "grad_norm": 1.2413307428359985,
      "learning_rate": 0.0001753780845209065,
      "loss": 3.702,
      "step": 132070
    },
    {
      "epoch": 0.12311993158864491,
      "grad_norm": 1.0199605226516724,
      "learning_rate": 0.00017537622019881132,
      "loss": 3.6603,
      "step": 132080
    },
    {
      "epoch": 0.12312925320672401,
      "grad_norm": 0.9186791181564331,
      "learning_rate": 0.00017537435587671612,
      "loss": 3.7278,
      "step": 132090
    },
    {
      "epoch": 0.1231385748248031,
      "grad_norm": 1.0518609285354614,
      "learning_rate": 0.00017537249155462093,
      "loss": 3.8048,
      "step": 132100
    },
    {
      "epoch": 0.1231478964428822,
      "grad_norm": 1.2654846906661987,
      "learning_rate": 0.0001753706272325257,
      "loss": 3.5975,
      "step": 132110
    },
    {
      "epoch": 0.12315721806096129,
      "grad_norm": 1.1545381546020508,
      "learning_rate": 0.00017536876291043051,
      "loss": 3.5099,
      "step": 132120
    },
    {
      "epoch": 0.12316653967904037,
      "grad_norm": 0.8267285823822021,
      "learning_rate": 0.00017536689858833532,
      "loss": 3.8039,
      "step": 132130
    },
    {
      "epoch": 0.12317586129711947,
      "grad_norm": 1.0266146659851074,
      "learning_rate": 0.00017536503426624013,
      "loss": 3.5499,
      "step": 132140
    },
    {
      "epoch": 0.12318518291519856,
      "grad_norm": 1.2613611221313477,
      "learning_rate": 0.00017536316994414493,
      "loss": 3.7467,
      "step": 132150
    },
    {
      "epoch": 0.12319450453327765,
      "grad_norm": 1.0708348751068115,
      "learning_rate": 0.0001753613056220497,
      "loss": 3.5637,
      "step": 132160
    },
    {
      "epoch": 0.12320382615135675,
      "grad_norm": 0.8372321128845215,
      "learning_rate": 0.00017535944129995452,
      "loss": 3.5775,
      "step": 132170
    },
    {
      "epoch": 0.12321314776943583,
      "grad_norm": 1.0303703546524048,
      "learning_rate": 0.00017535757697785932,
      "loss": 3.7777,
      "step": 132180
    },
    {
      "epoch": 0.12322246938751492,
      "grad_norm": 1.1057116985321045,
      "learning_rate": 0.0001753557126557641,
      "loss": 3.4905,
      "step": 132190
    },
    {
      "epoch": 0.12323179100559402,
      "grad_norm": 1.1002861261367798,
      "learning_rate": 0.00017535384833366894,
      "loss": 3.6509,
      "step": 132200
    },
    {
      "epoch": 0.12324111262367311,
      "grad_norm": 1.012351393699646,
      "learning_rate": 0.00017535198401157372,
      "loss": 3.5239,
      "step": 132210
    },
    {
      "epoch": 0.12325043424175221,
      "grad_norm": 1.1325119733810425,
      "learning_rate": 0.00017535011968947852,
      "loss": 3.7647,
      "step": 132220
    },
    {
      "epoch": 0.12325975585983129,
      "grad_norm": 0.9934913516044617,
      "learning_rate": 0.00017534825536738333,
      "loss": 3.6128,
      "step": 132230
    },
    {
      "epoch": 0.12326907747791038,
      "grad_norm": 0.9737234711647034,
      "learning_rate": 0.0001753463910452881,
      "loss": 3.6622,
      "step": 132240
    },
    {
      "epoch": 0.12327839909598948,
      "grad_norm": 1.0258601903915405,
      "learning_rate": 0.00017534452672319291,
      "loss": 3.6395,
      "step": 132250
    },
    {
      "epoch": 0.12328772071406857,
      "grad_norm": 1.0262209177017212,
      "learning_rate": 0.00017534266240109772,
      "loss": 3.7293,
      "step": 132260
    },
    {
      "epoch": 0.12329704233214767,
      "grad_norm": 0.937189519405365,
      "learning_rate": 0.00017534079807900253,
      "loss": 3.7445,
      "step": 132270
    },
    {
      "epoch": 0.12330636395022676,
      "grad_norm": 1.200951337814331,
      "learning_rate": 0.00017533893375690733,
      "loss": 3.6609,
      "step": 132280
    },
    {
      "epoch": 0.12331568556830584,
      "grad_norm": 1.0902087688446045,
      "learning_rate": 0.0001753370694348121,
      "loss": 3.7006,
      "step": 132290
    },
    {
      "epoch": 0.12332500718638494,
      "grad_norm": 1.179603934288025,
      "learning_rate": 0.00017533520511271692,
      "loss": 3.646,
      "step": 132300
    },
    {
      "epoch": 0.12333432880446403,
      "grad_norm": 0.9269334673881531,
      "learning_rate": 0.00017533334079062173,
      "loss": 3.7209,
      "step": 132310
    },
    {
      "epoch": 0.12334365042254312,
      "grad_norm": 1.1458072662353516,
      "learning_rate": 0.00017533147646852653,
      "loss": 3.5943,
      "step": 132320
    },
    {
      "epoch": 0.12335297204062222,
      "grad_norm": 1.140795350074768,
      "learning_rate": 0.00017532961214643134,
      "loss": 3.4934,
      "step": 132330
    },
    {
      "epoch": 0.1233622936587013,
      "grad_norm": 1.2081823348999023,
      "learning_rate": 0.00017532774782433612,
      "loss": 3.704,
      "step": 132340
    },
    {
      "epoch": 0.1233716152767804,
      "grad_norm": 1.1420034170150757,
      "learning_rate": 0.00017532588350224092,
      "loss": 3.6195,
      "step": 132350
    },
    {
      "epoch": 0.12338093689485949,
      "grad_norm": 1.0118464231491089,
      "learning_rate": 0.0001753240191801457,
      "loss": 3.7327,
      "step": 132360
    },
    {
      "epoch": 0.12339025851293858,
      "grad_norm": 0.9687809348106384,
      "learning_rate": 0.00017532215485805054,
      "loss": 3.8228,
      "step": 132370
    },
    {
      "epoch": 0.12339958013101768,
      "grad_norm": 1.1763659715652466,
      "learning_rate": 0.00017532029053595534,
      "loss": 3.5157,
      "step": 132380
    },
    {
      "epoch": 0.12340890174909676,
      "grad_norm": 1.2258416414260864,
      "learning_rate": 0.00017531842621386012,
      "loss": 3.6722,
      "step": 132390
    },
    {
      "epoch": 0.12341822336717585,
      "grad_norm": 1.1193166971206665,
      "learning_rate": 0.00017531656189176493,
      "loss": 3.5298,
      "step": 132400
    },
    {
      "epoch": 0.12342754498525495,
      "grad_norm": 1.1140345335006714,
      "learning_rate": 0.0001753146975696697,
      "loss": 3.8246,
      "step": 132410
    },
    {
      "epoch": 0.12343686660333404,
      "grad_norm": 0.9521585702896118,
      "learning_rate": 0.0001753128332475745,
      "loss": 3.703,
      "step": 132420
    },
    {
      "epoch": 0.12344618822141314,
      "grad_norm": 1.090902328491211,
      "learning_rate": 0.00017531096892547935,
      "loss": 3.66,
      "step": 132430
    },
    {
      "epoch": 0.12345550983949222,
      "grad_norm": 0.9870071411132812,
      "learning_rate": 0.00017530910460338413,
      "loss": 3.8147,
      "step": 132440
    },
    {
      "epoch": 0.12346483145757131,
      "grad_norm": 1.0742088556289673,
      "learning_rate": 0.00017530724028128893,
      "loss": 3.4308,
      "step": 132450
    },
    {
      "epoch": 0.1234741530756504,
      "grad_norm": 1.4477640390396118,
      "learning_rate": 0.0001753053759591937,
      "loss": 3.6435,
      "step": 132460
    },
    {
      "epoch": 0.1234834746937295,
      "grad_norm": 0.9302557110786438,
      "learning_rate": 0.00017530351163709852,
      "loss": 3.6421,
      "step": 132470
    },
    {
      "epoch": 0.1234927963118086,
      "grad_norm": 1.170278787612915,
      "learning_rate": 0.00017530164731500332,
      "loss": 3.6946,
      "step": 132480
    },
    {
      "epoch": 0.12350211792988769,
      "grad_norm": 0.9052245616912842,
      "learning_rate": 0.00017529978299290813,
      "loss": 3.779,
      "step": 132490
    },
    {
      "epoch": 0.12351143954796677,
      "grad_norm": 1.0315465927124023,
      "learning_rate": 0.00017529791867081294,
      "loss": 3.5116,
      "step": 132500
    },
    {
      "epoch": 0.12352076116604586,
      "grad_norm": 1.2158652544021606,
      "learning_rate": 0.00017529605434871772,
      "loss": 3.5894,
      "step": 132510
    },
    {
      "epoch": 0.12353008278412496,
      "grad_norm": 1.050830602645874,
      "learning_rate": 0.00017529419002662252,
      "loss": 3.6017,
      "step": 132520
    },
    {
      "epoch": 0.12353940440220405,
      "grad_norm": 1.1529685258865356,
      "learning_rate": 0.00017529232570452733,
      "loss": 3.6475,
      "step": 132530
    },
    {
      "epoch": 0.12354872602028315,
      "grad_norm": 1.2184110879898071,
      "learning_rate": 0.00017529046138243213,
      "loss": 3.6861,
      "step": 132540
    },
    {
      "epoch": 0.12355804763836223,
      "grad_norm": 1.0519846677780151,
      "learning_rate": 0.00017528859706033694,
      "loss": 3.3325,
      "step": 132550
    },
    {
      "epoch": 0.12356736925644132,
      "grad_norm": 1.0262293815612793,
      "learning_rate": 0.00017528673273824172,
      "loss": 3.7058,
      "step": 132560
    },
    {
      "epoch": 0.12357669087452042,
      "grad_norm": 1.0262134075164795,
      "learning_rate": 0.00017528486841614653,
      "loss": 3.6121,
      "step": 132570
    },
    {
      "epoch": 0.12358601249259951,
      "grad_norm": 0.9662354588508606,
      "learning_rate": 0.00017528300409405133,
      "loss": 3.796,
      "step": 132580
    },
    {
      "epoch": 0.1235953341106786,
      "grad_norm": 1.0991923809051514,
      "learning_rate": 0.00017528113977195614,
      "loss": 3.614,
      "step": 132590
    },
    {
      "epoch": 0.12360465572875769,
      "grad_norm": 1.2438528537750244,
      "learning_rate": 0.00017527927544986095,
      "loss": 3.8419,
      "step": 132600
    },
    {
      "epoch": 0.12361397734683678,
      "grad_norm": 1.2384637594223022,
      "learning_rate": 0.00017527741112776572,
      "loss": 3.6395,
      "step": 132610
    },
    {
      "epoch": 0.12362329896491588,
      "grad_norm": 1.1122668981552124,
      "learning_rate": 0.00017527554680567053,
      "loss": 3.6226,
      "step": 132620
    },
    {
      "epoch": 0.12363262058299497,
      "grad_norm": 0.9006558060646057,
      "learning_rate": 0.00017527368248357534,
      "loss": 3.535,
      "step": 132630
    },
    {
      "epoch": 0.12364194220107406,
      "grad_norm": 0.9942691326141357,
      "learning_rate": 0.00017527181816148012,
      "loss": 3.6617,
      "step": 132640
    },
    {
      "epoch": 0.12365126381915315,
      "grad_norm": 0.9300016164779663,
      "learning_rate": 0.00017526995383938495,
      "loss": 3.6642,
      "step": 132650
    },
    {
      "epoch": 0.12366058543723224,
      "grad_norm": 0.8194965720176697,
      "learning_rate": 0.00017526808951728973,
      "loss": 3.5868,
      "step": 132660
    },
    {
      "epoch": 0.12366990705531133,
      "grad_norm": 0.8669424057006836,
      "learning_rate": 0.00017526622519519454,
      "loss": 3.7324,
      "step": 132670
    },
    {
      "epoch": 0.12367922867339043,
      "grad_norm": 0.8785707354545593,
      "learning_rate": 0.00017526436087309934,
      "loss": 3.6907,
      "step": 132680
    },
    {
      "epoch": 0.12368855029146952,
      "grad_norm": 1.0069000720977783,
      "learning_rate": 0.00017526249655100412,
      "loss": 3.7426,
      "step": 132690
    },
    {
      "epoch": 0.12369787190954862,
      "grad_norm": 1.33400297164917,
      "learning_rate": 0.00017526063222890893,
      "loss": 3.4912,
      "step": 132700
    },
    {
      "epoch": 0.1237071935276277,
      "grad_norm": 0.9030599594116211,
      "learning_rate": 0.00017525876790681373,
      "loss": 3.7206,
      "step": 132710
    },
    {
      "epoch": 0.12371651514570679,
      "grad_norm": 1.1314914226531982,
      "learning_rate": 0.00017525690358471854,
      "loss": 3.483,
      "step": 132720
    },
    {
      "epoch": 0.12372583676378589,
      "grad_norm": 1.0524625778198242,
      "learning_rate": 0.00017525503926262335,
      "loss": 3.6447,
      "step": 132730
    },
    {
      "epoch": 0.12373515838186498,
      "grad_norm": 1.0790568590164185,
      "learning_rate": 0.00017525317494052813,
      "loss": 3.5771,
      "step": 132740
    },
    {
      "epoch": 0.12374447999994408,
      "grad_norm": 0.9682123064994812,
      "learning_rate": 0.00017525131061843293,
      "loss": 3.7392,
      "step": 132750
    },
    {
      "epoch": 0.12375380161802316,
      "grad_norm": 1.0433402061462402,
      "learning_rate": 0.00017524944629633774,
      "loss": 3.5082,
      "step": 132760
    },
    {
      "epoch": 0.12376312323610225,
      "grad_norm": 1.1666251420974731,
      "learning_rate": 0.00017524758197424254,
      "loss": 3.7242,
      "step": 132770
    },
    {
      "epoch": 0.12377244485418135,
      "grad_norm": 1.1262192726135254,
      "learning_rate": 0.00017524571765214735,
      "loss": 3.5554,
      "step": 132780
    },
    {
      "epoch": 0.12378176647226044,
      "grad_norm": 1.1707583665847778,
      "learning_rate": 0.00017524385333005213,
      "loss": 3.5485,
      "step": 132790
    },
    {
      "epoch": 0.12379108809033953,
      "grad_norm": 1.0703105926513672,
      "learning_rate": 0.00017524198900795694,
      "loss": 3.5426,
      "step": 132800
    },
    {
      "epoch": 0.12380040970841862,
      "grad_norm": 1.1115763187408447,
      "learning_rate": 0.00017524012468586172,
      "loss": 3.7898,
      "step": 132810
    },
    {
      "epoch": 0.12380973132649771,
      "grad_norm": 1.215324878692627,
      "learning_rate": 0.00017523826036376655,
      "loss": 3.6579,
      "step": 132820
    },
    {
      "epoch": 0.1238190529445768,
      "grad_norm": 1.1237622499465942,
      "learning_rate": 0.00017523639604167136,
      "loss": 3.4152,
      "step": 132830
    },
    {
      "epoch": 0.1238283745626559,
      "grad_norm": 0.8318584561347961,
      "learning_rate": 0.00017523453171957613,
      "loss": 3.6595,
      "step": 132840
    },
    {
      "epoch": 0.123837696180735,
      "grad_norm": 0.9478282928466797,
      "learning_rate": 0.00017523266739748094,
      "loss": 3.6357,
      "step": 132850
    },
    {
      "epoch": 0.12384701779881407,
      "grad_norm": 1.041687250137329,
      "learning_rate": 0.00017523080307538572,
      "loss": 3.6553,
      "step": 132860
    },
    {
      "epoch": 0.12385633941689317,
      "grad_norm": 0.9247068166732788,
      "learning_rate": 0.00017522893875329053,
      "loss": 3.5594,
      "step": 132870
    },
    {
      "epoch": 0.12386566103497226,
      "grad_norm": 0.9898493885993958,
      "learning_rate": 0.00017522707443119533,
      "loss": 3.7222,
      "step": 132880
    },
    {
      "epoch": 0.12387498265305136,
      "grad_norm": 1.0841313600540161,
      "learning_rate": 0.00017522521010910014,
      "loss": 3.5233,
      "step": 132890
    },
    {
      "epoch": 0.12388430427113045,
      "grad_norm": 1.0769065618515015,
      "learning_rate": 0.00017522334578700495,
      "loss": 3.6636,
      "step": 132900
    },
    {
      "epoch": 0.12389362588920953,
      "grad_norm": 1.1615808010101318,
      "learning_rate": 0.00017522148146490972,
      "loss": 3.6729,
      "step": 132910
    },
    {
      "epoch": 0.12390294750728863,
      "grad_norm": 1.2780704498291016,
      "learning_rate": 0.00017521961714281453,
      "loss": 3.7295,
      "step": 132920
    },
    {
      "epoch": 0.12391226912536772,
      "grad_norm": 0.9601883292198181,
      "learning_rate": 0.00017521775282071934,
      "loss": 3.6891,
      "step": 132930
    },
    {
      "epoch": 0.12392159074344682,
      "grad_norm": 0.9882392883300781,
      "learning_rate": 0.00017521588849862414,
      "loss": 3.6488,
      "step": 132940
    },
    {
      "epoch": 0.12393091236152591,
      "grad_norm": 1.1209909915924072,
      "learning_rate": 0.00017521402417652895,
      "loss": 3.5863,
      "step": 132950
    },
    {
      "epoch": 0.123940233979605,
      "grad_norm": 1.1029417514801025,
      "learning_rate": 0.00017521215985443373,
      "loss": 3.5594,
      "step": 132960
    },
    {
      "epoch": 0.12394955559768409,
      "grad_norm": 1.105298399925232,
      "learning_rate": 0.00017521029553233854,
      "loss": 3.6235,
      "step": 132970
    },
    {
      "epoch": 0.12395887721576318,
      "grad_norm": 1.1415544748306274,
      "learning_rate": 0.00017520843121024334,
      "loss": 3.6796,
      "step": 132980
    },
    {
      "epoch": 0.12396819883384227,
      "grad_norm": 0.9780604243278503,
      "learning_rate": 0.00017520656688814815,
      "loss": 3.7275,
      "step": 132990
    },
    {
      "epoch": 0.12397752045192137,
      "grad_norm": 1.1663817167282104,
      "learning_rate": 0.00017520470256605295,
      "loss": 3.566,
      "step": 133000
    },
    {
      "epoch": 0.12398684207000046,
      "grad_norm": 1.0348902940750122,
      "learning_rate": 0.00017520283824395773,
      "loss": 3.6357,
      "step": 133010
    },
    {
      "epoch": 0.12399616368807954,
      "grad_norm": 1.0754092931747437,
      "learning_rate": 0.00017520097392186254,
      "loss": 3.677,
      "step": 133020
    },
    {
      "epoch": 0.12400548530615864,
      "grad_norm": 1.0249462127685547,
      "learning_rate": 0.00017519910959976732,
      "loss": 3.6924,
      "step": 133030
    },
    {
      "epoch": 0.12401480692423773,
      "grad_norm": 1.0112583637237549,
      "learning_rate": 0.00017519724527767215,
      "loss": 3.8963,
      "step": 133040
    },
    {
      "epoch": 0.12402412854231683,
      "grad_norm": 0.9502993822097778,
      "learning_rate": 0.00017519538095557696,
      "loss": 3.619,
      "step": 133050
    },
    {
      "epoch": 0.12403345016039592,
      "grad_norm": 1.1915324926376343,
      "learning_rate": 0.00017519351663348174,
      "loss": 3.5411,
      "step": 133060
    },
    {
      "epoch": 0.124042771778475,
      "grad_norm": 1.0301423072814941,
      "learning_rate": 0.00017519165231138654,
      "loss": 3.5894,
      "step": 133070
    },
    {
      "epoch": 0.1240520933965541,
      "grad_norm": 0.884244441986084,
      "learning_rate": 0.00017518978798929132,
      "loss": 3.6851,
      "step": 133080
    },
    {
      "epoch": 0.12406141501463319,
      "grad_norm": 1.155765175819397,
      "learning_rate": 0.00017518792366719613,
      "loss": 3.5494,
      "step": 133090
    },
    {
      "epoch": 0.12407073663271229,
      "grad_norm": 1.098888874053955,
      "learning_rate": 0.00017518605934510096,
      "loss": 3.7172,
      "step": 133100
    },
    {
      "epoch": 0.12408005825079138,
      "grad_norm": 0.9853249788284302,
      "learning_rate": 0.00017518419502300574,
      "loss": 3.6713,
      "step": 133110
    },
    {
      "epoch": 0.12408937986887046,
      "grad_norm": 1.1273963451385498,
      "learning_rate": 0.00017518233070091055,
      "loss": 3.5276,
      "step": 133120
    },
    {
      "epoch": 0.12409870148694956,
      "grad_norm": 1.0999846458435059,
      "learning_rate": 0.00017518046637881533,
      "loss": 3.6572,
      "step": 133130
    },
    {
      "epoch": 0.12410802310502865,
      "grad_norm": 1.2400660514831543,
      "learning_rate": 0.00017517860205672013,
      "loss": 3.7738,
      "step": 133140
    },
    {
      "epoch": 0.12411734472310774,
      "grad_norm": 1.0431703329086304,
      "learning_rate": 0.00017517673773462494,
      "loss": 3.7183,
      "step": 133150
    },
    {
      "epoch": 0.12412666634118684,
      "grad_norm": 1.2103829383850098,
      "learning_rate": 0.00017517487341252975,
      "loss": 3.6074,
      "step": 133160
    },
    {
      "epoch": 0.12413598795926593,
      "grad_norm": 1.0926673412322998,
      "learning_rate": 0.00017517300909043455,
      "loss": 3.7505,
      "step": 133170
    },
    {
      "epoch": 0.12414530957734501,
      "grad_norm": 1.1241508722305298,
      "learning_rate": 0.00017517114476833933,
      "loss": 3.7103,
      "step": 133180
    },
    {
      "epoch": 0.12415463119542411,
      "grad_norm": 1.127568006515503,
      "learning_rate": 0.00017516928044624414,
      "loss": 3.7686,
      "step": 133190
    },
    {
      "epoch": 0.1241639528135032,
      "grad_norm": 1.0132118463516235,
      "learning_rate": 0.00017516741612414894,
      "loss": 3.8248,
      "step": 133200
    },
    {
      "epoch": 0.1241732744315823,
      "grad_norm": 1.0234030485153198,
      "learning_rate": 0.00017516555180205375,
      "loss": 3.7304,
      "step": 133210
    },
    {
      "epoch": 0.12418259604966139,
      "grad_norm": 1.01263427734375,
      "learning_rate": 0.00017516368747995856,
      "loss": 3.5026,
      "step": 133220
    },
    {
      "epoch": 0.12419191766774047,
      "grad_norm": 0.9903879165649414,
      "learning_rate": 0.00017516182315786334,
      "loss": 3.6116,
      "step": 133230
    },
    {
      "epoch": 0.12420123928581957,
      "grad_norm": 0.9836918115615845,
      "learning_rate": 0.00017515995883576814,
      "loss": 3.5107,
      "step": 133240
    },
    {
      "epoch": 0.12421056090389866,
      "grad_norm": 1.1502447128295898,
      "learning_rate": 0.00017515809451367295,
      "loss": 3.6985,
      "step": 133250
    },
    {
      "epoch": 0.12421988252197776,
      "grad_norm": 1.0759875774383545,
      "learning_rate": 0.00017515623019157773,
      "loss": 3.6663,
      "step": 133260
    },
    {
      "epoch": 0.12422920414005685,
      "grad_norm": 1.0082929134368896,
      "learning_rate": 0.00017515436586948256,
      "loss": 3.5605,
      "step": 133270
    },
    {
      "epoch": 0.12423852575813593,
      "grad_norm": 0.9626066088676453,
      "learning_rate": 0.00017515250154738734,
      "loss": 3.6099,
      "step": 133280
    },
    {
      "epoch": 0.12424784737621503,
      "grad_norm": 1.1281856298446655,
      "learning_rate": 0.00017515063722529215,
      "loss": 3.5292,
      "step": 133290
    },
    {
      "epoch": 0.12425716899429412,
      "grad_norm": 1.1302330493927002,
      "learning_rate": 0.00017514877290319695,
      "loss": 3.6429,
      "step": 133300
    },
    {
      "epoch": 0.12426649061237321,
      "grad_norm": 0.884314775466919,
      "learning_rate": 0.00017514690858110173,
      "loss": 3.6223,
      "step": 133310
    },
    {
      "epoch": 0.12427581223045231,
      "grad_norm": 0.8396682143211365,
      "learning_rate": 0.00017514504425900654,
      "loss": 3.6347,
      "step": 133320
    },
    {
      "epoch": 0.12428513384853139,
      "grad_norm": 1.024278998374939,
      "learning_rate": 0.00017514317993691135,
      "loss": 3.7814,
      "step": 133330
    },
    {
      "epoch": 0.12429445546661048,
      "grad_norm": 1.1157612800598145,
      "learning_rate": 0.00017514131561481615,
      "loss": 3.6308,
      "step": 133340
    },
    {
      "epoch": 0.12430377708468958,
      "grad_norm": 1.0110288858413696,
      "learning_rate": 0.00017513945129272096,
      "loss": 3.6637,
      "step": 133350
    },
    {
      "epoch": 0.12431309870276867,
      "grad_norm": 1.077964186668396,
      "learning_rate": 0.00017513758697062574,
      "loss": 3.6915,
      "step": 133360
    },
    {
      "epoch": 0.12432242032084777,
      "grad_norm": 1.1823159456253052,
      "learning_rate": 0.00017513572264853054,
      "loss": 3.5721,
      "step": 133370
    },
    {
      "epoch": 0.12433174193892686,
      "grad_norm": 1.033267617225647,
      "learning_rate": 0.00017513385832643535,
      "loss": 3.7088,
      "step": 133380
    },
    {
      "epoch": 0.12434106355700594,
      "grad_norm": 1.0675773620605469,
      "learning_rate": 0.00017513199400434016,
      "loss": 3.5894,
      "step": 133390
    },
    {
      "epoch": 0.12435038517508504,
      "grad_norm": 0.7084217071533203,
      "learning_rate": 0.00017513012968224496,
      "loss": 3.7057,
      "step": 133400
    },
    {
      "epoch": 0.12435970679316413,
      "grad_norm": 1.1673637628555298,
      "learning_rate": 0.00017512826536014974,
      "loss": 3.807,
      "step": 133410
    },
    {
      "epoch": 0.12436902841124323,
      "grad_norm": 0.9409481883049011,
      "learning_rate": 0.00017512640103805455,
      "loss": 3.5644,
      "step": 133420
    },
    {
      "epoch": 0.12437835002932232,
      "grad_norm": 1.5236908197402954,
      "learning_rate": 0.00017512453671595935,
      "loss": 3.9389,
      "step": 133430
    },
    {
      "epoch": 0.1243876716474014,
      "grad_norm": 0.9650688767433167,
      "learning_rate": 0.00017512267239386416,
      "loss": 3.4125,
      "step": 133440
    },
    {
      "epoch": 0.1243969932654805,
      "grad_norm": 1.1725651025772095,
      "learning_rate": 0.00017512080807176897,
      "loss": 3.7492,
      "step": 133450
    },
    {
      "epoch": 0.12440631488355959,
      "grad_norm": 1.0163824558258057,
      "learning_rate": 0.00017511894374967375,
      "loss": 3.5822,
      "step": 133460
    },
    {
      "epoch": 0.12441563650163869,
      "grad_norm": 1.1729167699813843,
      "learning_rate": 0.00017511707942757855,
      "loss": 3.6469,
      "step": 133470
    },
    {
      "epoch": 0.12442495811971778,
      "grad_norm": 0.9699730277061462,
      "learning_rate": 0.00017511521510548333,
      "loss": 3.7032,
      "step": 133480
    },
    {
      "epoch": 0.12443427973779686,
      "grad_norm": 1.1358102560043335,
      "learning_rate": 0.00017511335078338816,
      "loss": 3.6778,
      "step": 133490
    },
    {
      "epoch": 0.12444360135587595,
      "grad_norm": 1.013377070426941,
      "learning_rate": 0.00017511148646129297,
      "loss": 3.7061,
      "step": 133500
    },
    {
      "epoch": 0.12445292297395505,
      "grad_norm": 0.9047257304191589,
      "learning_rate": 0.00017510962213919775,
      "loss": 3.7066,
      "step": 133510
    },
    {
      "epoch": 0.12446224459203414,
      "grad_norm": 1.1227307319641113,
      "learning_rate": 0.00017510775781710256,
      "loss": 3.6243,
      "step": 133520
    },
    {
      "epoch": 0.12447156621011324,
      "grad_norm": 1.3218516111373901,
      "learning_rate": 0.00017510589349500734,
      "loss": 3.6614,
      "step": 133530
    },
    {
      "epoch": 0.12448088782819232,
      "grad_norm": 1.0213943719863892,
      "learning_rate": 0.00017510402917291214,
      "loss": 3.693,
      "step": 133540
    },
    {
      "epoch": 0.12449020944627141,
      "grad_norm": 1.0793235301971436,
      "learning_rate": 0.00017510216485081698,
      "loss": 3.7888,
      "step": 133550
    },
    {
      "epoch": 0.12449953106435051,
      "grad_norm": 1.0566860437393188,
      "learning_rate": 0.00017510030052872175,
      "loss": 3.6823,
      "step": 133560
    },
    {
      "epoch": 0.1245088526824296,
      "grad_norm": 0.9673302173614502,
      "learning_rate": 0.00017509843620662656,
      "loss": 3.6074,
      "step": 133570
    },
    {
      "epoch": 0.1245181743005087,
      "grad_norm": 1.0008199214935303,
      "learning_rate": 0.00017509657188453134,
      "loss": 3.6922,
      "step": 133580
    },
    {
      "epoch": 0.12452749591858779,
      "grad_norm": 1.0608938932418823,
      "learning_rate": 0.00017509470756243615,
      "loss": 3.6299,
      "step": 133590
    },
    {
      "epoch": 0.12453681753666687,
      "grad_norm": 1.0873335599899292,
      "learning_rate": 0.00017509284324034095,
      "loss": 3.6474,
      "step": 133600
    },
    {
      "epoch": 0.12454613915474597,
      "grad_norm": 1.1879322528839111,
      "learning_rate": 0.00017509097891824576,
      "loss": 3.5753,
      "step": 133610
    },
    {
      "epoch": 0.12455546077282506,
      "grad_norm": 1.0146220922470093,
      "learning_rate": 0.00017508911459615057,
      "loss": 3.4812,
      "step": 133620
    },
    {
      "epoch": 0.12456478239090416,
      "grad_norm": 1.0554043054580688,
      "learning_rate": 0.00017508725027405534,
      "loss": 3.5961,
      "step": 133630
    },
    {
      "epoch": 0.12457410400898325,
      "grad_norm": 1.095160961151123,
      "learning_rate": 0.00017508538595196015,
      "loss": 3.5339,
      "step": 133640
    },
    {
      "epoch": 0.12458342562706233,
      "grad_norm": 1.0978909730911255,
      "learning_rate": 0.00017508352162986496,
      "loss": 3.6493,
      "step": 133650
    },
    {
      "epoch": 0.12459274724514142,
      "grad_norm": 1.0674854516983032,
      "learning_rate": 0.00017508165730776976,
      "loss": 3.7112,
      "step": 133660
    },
    {
      "epoch": 0.12460206886322052,
      "grad_norm": 0.9126174449920654,
      "learning_rate": 0.00017507979298567457,
      "loss": 3.7485,
      "step": 133670
    },
    {
      "epoch": 0.12461139048129961,
      "grad_norm": 1.0684447288513184,
      "learning_rate": 0.00017507792866357935,
      "loss": 3.662,
      "step": 133680
    },
    {
      "epoch": 0.12462071209937871,
      "grad_norm": 0.9257031679153442,
      "learning_rate": 0.00017507606434148416,
      "loss": 3.712,
      "step": 133690
    },
    {
      "epoch": 0.12463003371745779,
      "grad_norm": 1.1248289346694946,
      "learning_rate": 0.00017507420001938896,
      "loss": 3.6417,
      "step": 133700
    },
    {
      "epoch": 0.12463935533553688,
      "grad_norm": 1.015052080154419,
      "learning_rate": 0.00017507233569729374,
      "loss": 3.6202,
      "step": 133710
    },
    {
      "epoch": 0.12464867695361598,
      "grad_norm": 1.0388163328170776,
      "learning_rate": 0.00017507047137519857,
      "loss": 3.6302,
      "step": 133720
    },
    {
      "epoch": 0.12465799857169507,
      "grad_norm": 1.0446884632110596,
      "learning_rate": 0.00017506860705310335,
      "loss": 3.6994,
      "step": 133730
    },
    {
      "epoch": 0.12466732018977417,
      "grad_norm": 0.9631429314613342,
      "learning_rate": 0.00017506674273100816,
      "loss": 3.7204,
      "step": 133740
    },
    {
      "epoch": 0.12467664180785325,
      "grad_norm": 1.1327675580978394,
      "learning_rate": 0.00017506487840891297,
      "loss": 3.5732,
      "step": 133750
    },
    {
      "epoch": 0.12468596342593234,
      "grad_norm": 1.1122108697891235,
      "learning_rate": 0.00017506301408681775,
      "loss": 3.7085,
      "step": 133760
    },
    {
      "epoch": 0.12469528504401144,
      "grad_norm": 0.9626511335372925,
      "learning_rate": 0.00017506114976472255,
      "loss": 3.5933,
      "step": 133770
    },
    {
      "epoch": 0.12470460666209053,
      "grad_norm": 1.3228524923324585,
      "learning_rate": 0.00017505928544262736,
      "loss": 3.5258,
      "step": 133780
    },
    {
      "epoch": 0.12471392828016963,
      "grad_norm": 0.762591540813446,
      "learning_rate": 0.00017505742112053216,
      "loss": 3.4627,
      "step": 133790
    },
    {
      "epoch": 0.1247232498982487,
      "grad_norm": 1.160286545753479,
      "learning_rate": 0.00017505555679843697,
      "loss": 3.7152,
      "step": 133800
    },
    {
      "epoch": 0.1247325715163278,
      "grad_norm": 1.0743800401687622,
      "learning_rate": 0.00017505369247634175,
      "loss": 3.5277,
      "step": 133810
    },
    {
      "epoch": 0.1247418931344069,
      "grad_norm": 0.9996865391731262,
      "learning_rate": 0.00017505182815424656,
      "loss": 3.5844,
      "step": 133820
    },
    {
      "epoch": 0.12475121475248599,
      "grad_norm": 1.069851040840149,
      "learning_rate": 0.00017504996383215136,
      "loss": 3.6867,
      "step": 133830
    },
    {
      "epoch": 0.12476053637056508,
      "grad_norm": 1.008448839187622,
      "learning_rate": 0.00017504809951005617,
      "loss": 3.609,
      "step": 133840
    },
    {
      "epoch": 0.12476985798864418,
      "grad_norm": 1.133437991142273,
      "learning_rate": 0.00017504623518796098,
      "loss": 3.256,
      "step": 133850
    },
    {
      "epoch": 0.12477917960672326,
      "grad_norm": 1.1009397506713867,
      "learning_rate": 0.00017504437086586575,
      "loss": 3.6005,
      "step": 133860
    },
    {
      "epoch": 0.12478850122480235,
      "grad_norm": 1.0730618238449097,
      "learning_rate": 0.00017504250654377056,
      "loss": 3.5496,
      "step": 133870
    },
    {
      "epoch": 0.12479782284288145,
      "grad_norm": 0.9709783792495728,
      "learning_rate": 0.00017504064222167537,
      "loss": 3.6269,
      "step": 133880
    },
    {
      "epoch": 0.12480714446096054,
      "grad_norm": 1.1576677560806274,
      "learning_rate": 0.00017503877789958017,
      "loss": 3.8156,
      "step": 133890
    },
    {
      "epoch": 0.12481646607903964,
      "grad_norm": 1.1554361581802368,
      "learning_rate": 0.00017503691357748498,
      "loss": 3.7008,
      "step": 133900
    },
    {
      "epoch": 0.12482578769711872,
      "grad_norm": 0.9811626672744751,
      "learning_rate": 0.00017503504925538976,
      "loss": 3.8001,
      "step": 133910
    },
    {
      "epoch": 0.12483510931519781,
      "grad_norm": 1.0586379766464233,
      "learning_rate": 0.00017503318493329457,
      "loss": 3.7123,
      "step": 133920
    },
    {
      "epoch": 0.1248444309332769,
      "grad_norm": 1.1510834693908691,
      "learning_rate": 0.00017503132061119934,
      "loss": 3.7469,
      "step": 133930
    },
    {
      "epoch": 0.124853752551356,
      "grad_norm": 1.0901355743408203,
      "learning_rate": 0.00017502945628910418,
      "loss": 3.5774,
      "step": 133940
    },
    {
      "epoch": 0.1248630741694351,
      "grad_norm": 1.0423401594161987,
      "learning_rate": 0.00017502759196700898,
      "loss": 3.5474,
      "step": 133950
    },
    {
      "epoch": 0.12487239578751418,
      "grad_norm": 1.095463752746582,
      "learning_rate": 0.00017502572764491376,
      "loss": 3.7494,
      "step": 133960
    },
    {
      "epoch": 0.12488171740559327,
      "grad_norm": 1.138299822807312,
      "learning_rate": 0.00017502386332281857,
      "loss": 3.6626,
      "step": 133970
    },
    {
      "epoch": 0.12489103902367236,
      "grad_norm": 1.1001245975494385,
      "learning_rate": 0.00017502199900072335,
      "loss": 3.4907,
      "step": 133980
    },
    {
      "epoch": 0.12490036064175146,
      "grad_norm": 0.8905308246612549,
      "learning_rate": 0.00017502013467862816,
      "loss": 3.5634,
      "step": 133990
    },
    {
      "epoch": 0.12490968225983055,
      "grad_norm": 0.9577144384384155,
      "learning_rate": 0.000175018270356533,
      "loss": 3.6051,
      "step": 134000
    },
    {
      "epoch": 0.12491900387790963,
      "grad_norm": 0.9459570646286011,
      "learning_rate": 0.00017501640603443777,
      "loss": 3.6686,
      "step": 134010
    },
    {
      "epoch": 0.12492832549598873,
      "grad_norm": 1.0034596920013428,
      "learning_rate": 0.00017501454171234257,
      "loss": 3.7121,
      "step": 134020
    },
    {
      "epoch": 0.12493764711406782,
      "grad_norm": 1.0923861265182495,
      "learning_rate": 0.00017501267739024735,
      "loss": 3.604,
      "step": 134030
    },
    {
      "epoch": 0.12494696873214692,
      "grad_norm": 0.8141943216323853,
      "learning_rate": 0.00017501081306815216,
      "loss": 3.6066,
      "step": 134040
    },
    {
      "epoch": 0.12495629035022601,
      "grad_norm": 0.9729020595550537,
      "learning_rate": 0.00017500894874605697,
      "loss": 3.4034,
      "step": 134050
    },
    {
      "epoch": 0.12496561196830511,
      "grad_norm": 1.044270634651184,
      "learning_rate": 0.00017500708442396177,
      "loss": 3.7232,
      "step": 134060
    },
    {
      "epoch": 0.12497493358638419,
      "grad_norm": 1.008988618850708,
      "learning_rate": 0.00017500522010186658,
      "loss": 3.4336,
      "step": 134070
    },
    {
      "epoch": 0.12498425520446328,
      "grad_norm": 0.7500271201133728,
      "learning_rate": 0.00017500335577977136,
      "loss": 3.5569,
      "step": 134080
    },
    {
      "epoch": 0.12499357682254238,
      "grad_norm": 1.0234084129333496,
      "learning_rate": 0.00017500149145767616,
      "loss": 3.8568,
      "step": 134090
    },
    {
      "epoch": 0.12500289844062146,
      "grad_norm": 1.145134449005127,
      "learning_rate": 0.00017499962713558097,
      "loss": 3.6273,
      "step": 134100
    },
    {
      "epoch": 0.12501222005870055,
      "grad_norm": 1.1500186920166016,
      "learning_rate": 0.00017499776281348578,
      "loss": 3.8285,
      "step": 134110
    },
    {
      "epoch": 0.12502154167677965,
      "grad_norm": 1.0234297513961792,
      "learning_rate": 0.00017499589849139058,
      "loss": 3.762,
      "step": 134120
    },
    {
      "epoch": 0.12503086329485874,
      "grad_norm": 1.1576563119888306,
      "learning_rate": 0.00017499403416929536,
      "loss": 3.528,
      "step": 134130
    },
    {
      "epoch": 0.12504018491293784,
      "grad_norm": 1.2302656173706055,
      "learning_rate": 0.00017499216984720017,
      "loss": 3.63,
      "step": 134140
    },
    {
      "epoch": 0.12504950653101693,
      "grad_norm": 1.0790412425994873,
      "learning_rate": 0.00017499030552510497,
      "loss": 3.6512,
      "step": 134150
    },
    {
      "epoch": 0.12505882814909602,
      "grad_norm": 1.9493225812911987,
      "learning_rate": 0.00017498844120300975,
      "loss": 3.563,
      "step": 134160
    },
    {
      "epoch": 0.12506814976717512,
      "grad_norm": 1.095017433166504,
      "learning_rate": 0.0001749865768809146,
      "loss": 3.4946,
      "step": 134170
    },
    {
      "epoch": 0.1250774713852542,
      "grad_norm": 1.2141501903533936,
      "learning_rate": 0.00017498471255881937,
      "loss": 3.5324,
      "step": 134180
    },
    {
      "epoch": 0.1250867930033333,
      "grad_norm": 0.8716153502464294,
      "learning_rate": 0.00017498284823672417,
      "loss": 3.6102,
      "step": 134190
    },
    {
      "epoch": 0.12509611462141237,
      "grad_norm": 1.0922765731811523,
      "learning_rate": 0.00017498098391462898,
      "loss": 3.5155,
      "step": 134200
    },
    {
      "epoch": 0.12510543623949147,
      "grad_norm": 1.010145902633667,
      "learning_rate": 0.00017497911959253376,
      "loss": 3.5126,
      "step": 134210
    },
    {
      "epoch": 0.12511475785757056,
      "grad_norm": 1.100509524345398,
      "learning_rate": 0.00017497725527043856,
      "loss": 3.7817,
      "step": 134220
    },
    {
      "epoch": 0.12512407947564966,
      "grad_norm": 1.0535974502563477,
      "learning_rate": 0.00017497539094834337,
      "loss": 3.6377,
      "step": 134230
    },
    {
      "epoch": 0.12513340109372875,
      "grad_norm": 0.9259254932403564,
      "learning_rate": 0.00017497352662624818,
      "loss": 3.5246,
      "step": 134240
    },
    {
      "epoch": 0.12514272271180785,
      "grad_norm": 0.9726383686065674,
      "learning_rate": 0.00017497166230415296,
      "loss": 3.6774,
      "step": 134250
    },
    {
      "epoch": 0.12515204432988694,
      "grad_norm": 1.1288951635360718,
      "learning_rate": 0.00017496979798205776,
      "loss": 3.6471,
      "step": 134260
    },
    {
      "epoch": 0.12516136594796604,
      "grad_norm": 0.9499339461326599,
      "learning_rate": 0.00017496793365996257,
      "loss": 3.6104,
      "step": 134270
    },
    {
      "epoch": 0.12517068756604513,
      "grad_norm": 1.0886788368225098,
      "learning_rate": 0.00017496606933786738,
      "loss": 3.5928,
      "step": 134280
    },
    {
      "epoch": 0.12518000918412422,
      "grad_norm": 0.9997933506965637,
      "learning_rate": 0.00017496420501577218,
      "loss": 3.6465,
      "step": 134290
    },
    {
      "epoch": 0.1251893308022033,
      "grad_norm": 1.0891541242599487,
      "learning_rate": 0.00017496234069367696,
      "loss": 3.7321,
      "step": 134300
    },
    {
      "epoch": 0.12519865242028239,
      "grad_norm": 1.0756580829620361,
      "learning_rate": 0.00017496047637158177,
      "loss": 3.622,
      "step": 134310
    },
    {
      "epoch": 0.12520797403836148,
      "grad_norm": 0.9220572710037231,
      "learning_rate": 0.00017495861204948657,
      "loss": 3.6991,
      "step": 134320
    },
    {
      "epoch": 0.12521729565644057,
      "grad_norm": 0.9931114912033081,
      "learning_rate": 0.00017495674772739138,
      "loss": 3.5335,
      "step": 134330
    },
    {
      "epoch": 0.12522661727451967,
      "grad_norm": 0.9776830673217773,
      "learning_rate": 0.00017495488340529619,
      "loss": 3.6891,
      "step": 134340
    },
    {
      "epoch": 0.12523593889259876,
      "grad_norm": 1.09978187084198,
      "learning_rate": 0.00017495301908320097,
      "loss": 3.5069,
      "step": 134350
    },
    {
      "epoch": 0.12524526051067786,
      "grad_norm": 1.1889194250106812,
      "learning_rate": 0.00017495115476110577,
      "loss": 3.7041,
      "step": 134360
    },
    {
      "epoch": 0.12525458212875695,
      "grad_norm": 0.9467225670814514,
      "learning_rate": 0.00017494929043901058,
      "loss": 3.7902,
      "step": 134370
    },
    {
      "epoch": 0.12526390374683605,
      "grad_norm": 0.9482740759849548,
      "learning_rate": 0.00017494742611691536,
      "loss": 3.6319,
      "step": 134380
    },
    {
      "epoch": 0.12527322536491514,
      "grad_norm": 1.136029839515686,
      "learning_rate": 0.0001749455617948202,
      "loss": 3.5269,
      "step": 134390
    },
    {
      "epoch": 0.12528254698299424,
      "grad_norm": 1.0975388288497925,
      "learning_rate": 0.00017494369747272497,
      "loss": 3.624,
      "step": 134400
    },
    {
      "epoch": 0.1252918686010733,
      "grad_norm": 1.095794916152954,
      "learning_rate": 0.00017494183315062978,
      "loss": 3.678,
      "step": 134410
    },
    {
      "epoch": 0.1253011902191524,
      "grad_norm": 1.172442078590393,
      "learning_rate": 0.00017493996882853458,
      "loss": 3.516,
      "step": 134420
    },
    {
      "epoch": 0.1253105118372315,
      "grad_norm": 1.1310378313064575,
      "learning_rate": 0.00017493810450643936,
      "loss": 3.6839,
      "step": 134430
    },
    {
      "epoch": 0.1253198334553106,
      "grad_norm": 1.0571016073226929,
      "learning_rate": 0.00017493624018434417,
      "loss": 3.5784,
      "step": 134440
    },
    {
      "epoch": 0.12532915507338968,
      "grad_norm": 1.250441074371338,
      "learning_rate": 0.00017493437586224897,
      "loss": 3.6963,
      "step": 134450
    },
    {
      "epoch": 0.12533847669146878,
      "grad_norm": 1.0456098318099976,
      "learning_rate": 0.00017493251154015378,
      "loss": 3.6412,
      "step": 134460
    },
    {
      "epoch": 0.12534779830954787,
      "grad_norm": 1.0460636615753174,
      "learning_rate": 0.0001749306472180586,
      "loss": 3.6445,
      "step": 134470
    },
    {
      "epoch": 0.12535711992762696,
      "grad_norm": 1.0518367290496826,
      "learning_rate": 0.00017492878289596337,
      "loss": 3.5526,
      "step": 134480
    },
    {
      "epoch": 0.12536644154570606,
      "grad_norm": 0.9827843904495239,
      "learning_rate": 0.00017492691857386817,
      "loss": 3.9004,
      "step": 134490
    },
    {
      "epoch": 0.12537576316378515,
      "grad_norm": 1.146826982498169,
      "learning_rate": 0.00017492505425177298,
      "loss": 3.6869,
      "step": 134500
    },
    {
      "epoch": 0.12538508478186422,
      "grad_norm": 1.1460835933685303,
      "learning_rate": 0.00017492318992967779,
      "loss": 3.6323,
      "step": 134510
    },
    {
      "epoch": 0.12539440639994331,
      "grad_norm": 1.7241404056549072,
      "learning_rate": 0.0001749213256075826,
      "loss": 3.5643,
      "step": 134520
    },
    {
      "epoch": 0.1254037280180224,
      "grad_norm": 1.1935611963272095,
      "learning_rate": 0.00017491946128548737,
      "loss": 3.783,
      "step": 134530
    },
    {
      "epoch": 0.1254130496361015,
      "grad_norm": 1.1933835744857788,
      "learning_rate": 0.00017491759696339218,
      "loss": 3.5519,
      "step": 134540
    },
    {
      "epoch": 0.1254223712541806,
      "grad_norm": 1.1415354013442993,
      "learning_rate": 0.00017491573264129698,
      "loss": 3.5959,
      "step": 134550
    },
    {
      "epoch": 0.1254316928722597,
      "grad_norm": 1.0031521320343018,
      "learning_rate": 0.0001749138683192018,
      "loss": 3.647,
      "step": 134560
    },
    {
      "epoch": 0.1254410144903388,
      "grad_norm": 1.0577759742736816,
      "learning_rate": 0.0001749120039971066,
      "loss": 3.635,
      "step": 134570
    },
    {
      "epoch": 0.12545033610841788,
      "grad_norm": 1.106946349143982,
      "learning_rate": 0.00017491013967501137,
      "loss": 3.6346,
      "step": 134580
    },
    {
      "epoch": 0.12545965772649698,
      "grad_norm": 0.9455611109733582,
      "learning_rate": 0.00017490827535291618,
      "loss": 3.7032,
      "step": 134590
    },
    {
      "epoch": 0.12546897934457607,
      "grad_norm": 1.0651346445083618,
      "learning_rate": 0.00017490641103082096,
      "loss": 3.556,
      "step": 134600
    },
    {
      "epoch": 0.12547830096265516,
      "grad_norm": 1.1094386577606201,
      "learning_rate": 0.0001749045467087258,
      "loss": 3.6298,
      "step": 134610
    },
    {
      "epoch": 0.12548762258073423,
      "grad_norm": 1.4382061958312988,
      "learning_rate": 0.0001749026823866306,
      "loss": 3.6451,
      "step": 134620
    },
    {
      "epoch": 0.12549694419881333,
      "grad_norm": 0.937840461730957,
      "learning_rate": 0.00017490081806453538,
      "loss": 3.5784,
      "step": 134630
    },
    {
      "epoch": 0.12550626581689242,
      "grad_norm": 0.909685492515564,
      "learning_rate": 0.00017489895374244019,
      "loss": 3.5494,
      "step": 134640
    },
    {
      "epoch": 0.12551558743497152,
      "grad_norm": 1.288465976715088,
      "learning_rate": 0.00017489708942034496,
      "loss": 3.7544,
      "step": 134650
    },
    {
      "epoch": 0.1255249090530506,
      "grad_norm": 1.0936542749404907,
      "learning_rate": 0.00017489522509824977,
      "loss": 3.4285,
      "step": 134660
    },
    {
      "epoch": 0.1255342306711297,
      "grad_norm": 0.9431285262107849,
      "learning_rate": 0.0001748933607761546,
      "loss": 3.6339,
      "step": 134670
    },
    {
      "epoch": 0.1255435522892088,
      "grad_norm": 0.8742727637290955,
      "learning_rate": 0.00017489149645405938,
      "loss": 3.5053,
      "step": 134680
    },
    {
      "epoch": 0.1255528739072879,
      "grad_norm": 1.1325924396514893,
      "learning_rate": 0.0001748896321319642,
      "loss": 3.7871,
      "step": 134690
    },
    {
      "epoch": 0.125562195525367,
      "grad_norm": 0.9426765441894531,
      "learning_rate": 0.00017488776780986897,
      "loss": 3.737,
      "step": 134700
    },
    {
      "epoch": 0.12557151714344608,
      "grad_norm": 1.0099961757659912,
      "learning_rate": 0.00017488590348777378,
      "loss": 3.5684,
      "step": 134710
    },
    {
      "epoch": 0.12558083876152515,
      "grad_norm": 1.0120681524276733,
      "learning_rate": 0.00017488403916567858,
      "loss": 3.4407,
      "step": 134720
    },
    {
      "epoch": 0.12559016037960424,
      "grad_norm": 1.236869215965271,
      "learning_rate": 0.0001748821748435834,
      "loss": 3.5999,
      "step": 134730
    },
    {
      "epoch": 0.12559948199768334,
      "grad_norm": 1.1076451539993286,
      "learning_rate": 0.0001748803105214882,
      "loss": 3.5308,
      "step": 134740
    },
    {
      "epoch": 0.12560880361576243,
      "grad_norm": 1.250077724456787,
      "learning_rate": 0.00017487844619939297,
      "loss": 3.4967,
      "step": 134750
    },
    {
      "epoch": 0.12561812523384153,
      "grad_norm": 1.1100908517837524,
      "learning_rate": 0.00017487658187729778,
      "loss": 3.5606,
      "step": 134760
    },
    {
      "epoch": 0.12562744685192062,
      "grad_norm": 1.063333511352539,
      "learning_rate": 0.00017487471755520259,
      "loss": 3.606,
      "step": 134770
    },
    {
      "epoch": 0.12563676846999972,
      "grad_norm": 1.057320475578308,
      "learning_rate": 0.0001748728532331074,
      "loss": 3.6496,
      "step": 134780
    },
    {
      "epoch": 0.1256460900880788,
      "grad_norm": 1.0628803968429565,
      "learning_rate": 0.0001748709889110122,
      "loss": 3.4464,
      "step": 134790
    },
    {
      "epoch": 0.1256554117061579,
      "grad_norm": 1.04832923412323,
      "learning_rate": 0.00017486912458891698,
      "loss": 3.5259,
      "step": 134800
    },
    {
      "epoch": 0.125664733324237,
      "grad_norm": 0.9888014197349548,
      "learning_rate": 0.00017486726026682178,
      "loss": 3.4966,
      "step": 134810
    },
    {
      "epoch": 0.12567405494231607,
      "grad_norm": 0.9999973177909851,
      "learning_rate": 0.0001748653959447266,
      "loss": 3.696,
      "step": 134820
    },
    {
      "epoch": 0.12568337656039516,
      "grad_norm": 1.043678879737854,
      "learning_rate": 0.00017486353162263137,
      "loss": 3.6524,
      "step": 134830
    },
    {
      "epoch": 0.12569269817847425,
      "grad_norm": 1.0374507904052734,
      "learning_rate": 0.0001748616673005362,
      "loss": 3.5782,
      "step": 134840
    },
    {
      "epoch": 0.12570201979655335,
      "grad_norm": 1.0715649127960205,
      "learning_rate": 0.00017485980297844098,
      "loss": 3.4312,
      "step": 134850
    },
    {
      "epoch": 0.12571134141463244,
      "grad_norm": 1.140480399131775,
      "learning_rate": 0.0001748579386563458,
      "loss": 3.572,
      "step": 134860
    },
    {
      "epoch": 0.12572066303271154,
      "grad_norm": 1.1070250272750854,
      "learning_rate": 0.0001748560743342506,
      "loss": 3.5289,
      "step": 134870
    },
    {
      "epoch": 0.12572998465079063,
      "grad_norm": 0.9908485412597656,
      "learning_rate": 0.00017485421001215537,
      "loss": 3.6335,
      "step": 134880
    },
    {
      "epoch": 0.12573930626886973,
      "grad_norm": 1.2295578718185425,
      "learning_rate": 0.00017485234569006018,
      "loss": 3.5985,
      "step": 134890
    },
    {
      "epoch": 0.12574862788694882,
      "grad_norm": 1.0585098266601562,
      "learning_rate": 0.000174850481367965,
      "loss": 3.6117,
      "step": 134900
    },
    {
      "epoch": 0.12575794950502792,
      "grad_norm": 1.1566230058670044,
      "learning_rate": 0.0001748486170458698,
      "loss": 3.6536,
      "step": 134910
    },
    {
      "epoch": 0.125767271123107,
      "grad_norm": 0.9959219694137573,
      "learning_rate": 0.0001748467527237746,
      "loss": 3.5617,
      "step": 134920
    },
    {
      "epoch": 0.12577659274118608,
      "grad_norm": 1.0753259658813477,
      "learning_rate": 0.00017484488840167938,
      "loss": 3.564,
      "step": 134930
    },
    {
      "epoch": 0.12578591435926517,
      "grad_norm": 1.0192391872406006,
      "learning_rate": 0.00017484302407958419,
      "loss": 3.4767,
      "step": 134940
    },
    {
      "epoch": 0.12579523597734427,
      "grad_norm": 0.9117977023124695,
      "learning_rate": 0.000174841159757489,
      "loss": 3.6055,
      "step": 134950
    },
    {
      "epoch": 0.12580455759542336,
      "grad_norm": 0.9845012426376343,
      "learning_rate": 0.0001748392954353938,
      "loss": 3.6377,
      "step": 134960
    },
    {
      "epoch": 0.12581387921350246,
      "grad_norm": 1.1179600954055786,
      "learning_rate": 0.0001748374311132986,
      "loss": 3.7722,
      "step": 134970
    },
    {
      "epoch": 0.12582320083158155,
      "grad_norm": 0.935843825340271,
      "learning_rate": 0.00017483556679120338,
      "loss": 3.4794,
      "step": 134980
    },
    {
      "epoch": 0.12583252244966064,
      "grad_norm": 0.9548398852348328,
      "learning_rate": 0.0001748337024691082,
      "loss": 3.7351,
      "step": 134990
    },
    {
      "epoch": 0.12584184406773974,
      "grad_norm": 1.2658812999725342,
      "learning_rate": 0.000174831838147013,
      "loss": 3.4706,
      "step": 135000
    },
    {
      "epoch": 0.12585116568581883,
      "grad_norm": 0.9982867240905762,
      "learning_rate": 0.0001748299738249178,
      "loss": 3.6838,
      "step": 135010
    },
    {
      "epoch": 0.12586048730389793,
      "grad_norm": 1.0056736469268799,
      "learning_rate": 0.0001748281095028226,
      "loss": 3.6827,
      "step": 135020
    },
    {
      "epoch": 0.125869808921977,
      "grad_norm": 1.2372394800186157,
      "learning_rate": 0.0001748262451807274,
      "loss": 3.588,
      "step": 135030
    },
    {
      "epoch": 0.1258791305400561,
      "grad_norm": 1.0226905345916748,
      "learning_rate": 0.0001748243808586322,
      "loss": 3.7335,
      "step": 135040
    },
    {
      "epoch": 0.12588845215813518,
      "grad_norm": 1.0522316694259644,
      "learning_rate": 0.00017482251653653697,
      "loss": 3.6698,
      "step": 135050
    },
    {
      "epoch": 0.12589777377621428,
      "grad_norm": 1.0414899587631226,
      "learning_rate": 0.0001748206522144418,
      "loss": 3.581,
      "step": 135060
    },
    {
      "epoch": 0.12590709539429337,
      "grad_norm": 1.2440168857574463,
      "learning_rate": 0.0001748187878923466,
      "loss": 3.4701,
      "step": 135070
    },
    {
      "epoch": 0.12591641701237247,
      "grad_norm": 0.9553737640380859,
      "learning_rate": 0.0001748169235702514,
      "loss": 3.7115,
      "step": 135080
    },
    {
      "epoch": 0.12592573863045156,
      "grad_norm": 1.1013468503952026,
      "learning_rate": 0.0001748150592481562,
      "loss": 3.5716,
      "step": 135090
    },
    {
      "epoch": 0.12593506024853066,
      "grad_norm": 1.0459039211273193,
      "learning_rate": 0.00017481319492606098,
      "loss": 3.7389,
      "step": 135100
    },
    {
      "epoch": 0.12594438186660975,
      "grad_norm": 1.1339493989944458,
      "learning_rate": 0.00017481133060396578,
      "loss": 3.61,
      "step": 135110
    },
    {
      "epoch": 0.12595370348468884,
      "grad_norm": 0.9351755380630493,
      "learning_rate": 0.00017480946628187062,
      "loss": 3.7572,
      "step": 135120
    },
    {
      "epoch": 0.12596302510276794,
      "grad_norm": 0.9620692133903503,
      "learning_rate": 0.0001748076019597754,
      "loss": 3.7792,
      "step": 135130
    },
    {
      "epoch": 0.125972346720847,
      "grad_norm": 1.097632884979248,
      "learning_rate": 0.0001748057376376802,
      "loss": 3.6035,
      "step": 135140
    },
    {
      "epoch": 0.1259816683389261,
      "grad_norm": 0.9899196028709412,
      "learning_rate": 0.00017480387331558498,
      "loss": 3.5395,
      "step": 135150
    },
    {
      "epoch": 0.1259909899570052,
      "grad_norm": 0.9989465475082397,
      "learning_rate": 0.0001748020089934898,
      "loss": 3.6884,
      "step": 135160
    },
    {
      "epoch": 0.1260003115750843,
      "grad_norm": 1.0763238668441772,
      "learning_rate": 0.0001748001446713946,
      "loss": 3.6369,
      "step": 135170
    },
    {
      "epoch": 0.12600963319316338,
      "grad_norm": 0.9628206491470337,
      "learning_rate": 0.0001747982803492994,
      "loss": 3.6971,
      "step": 135180
    },
    {
      "epoch": 0.12601895481124248,
      "grad_norm": 0.9283196926116943,
      "learning_rate": 0.0001747964160272042,
      "loss": 3.8271,
      "step": 135190
    },
    {
      "epoch": 0.12602827642932157,
      "grad_norm": 1.1678162813186646,
      "learning_rate": 0.000174794551705109,
      "loss": 3.656,
      "step": 135200
    },
    {
      "epoch": 0.12603759804740067,
      "grad_norm": 1.258664846420288,
      "learning_rate": 0.0001747926873830138,
      "loss": 3.5231,
      "step": 135210
    },
    {
      "epoch": 0.12604691966547976,
      "grad_norm": 1.1600929498672485,
      "learning_rate": 0.0001747908230609186,
      "loss": 3.6798,
      "step": 135220
    },
    {
      "epoch": 0.12605624128355886,
      "grad_norm": 1.1270830631256104,
      "learning_rate": 0.0001747889587388234,
      "loss": 3.5775,
      "step": 135230
    },
    {
      "epoch": 0.12606556290163792,
      "grad_norm": 1.1634652614593506,
      "learning_rate": 0.0001747870944167282,
      "loss": 3.6081,
      "step": 135240
    },
    {
      "epoch": 0.12607488451971702,
      "grad_norm": 0.9805838465690613,
      "learning_rate": 0.000174785230094633,
      "loss": 3.7694,
      "step": 135250
    },
    {
      "epoch": 0.1260842061377961,
      "grad_norm": 1.0714987516403198,
      "learning_rate": 0.0001747833657725378,
      "loss": 3.5914,
      "step": 135260
    },
    {
      "epoch": 0.1260935277558752,
      "grad_norm": 1.027308702468872,
      "learning_rate": 0.0001747815014504426,
      "loss": 3.5579,
      "step": 135270
    },
    {
      "epoch": 0.1261028493739543,
      "grad_norm": 1.0631608963012695,
      "learning_rate": 0.00017477963712834738,
      "loss": 3.7334,
      "step": 135280
    },
    {
      "epoch": 0.1261121709920334,
      "grad_norm": 0.999114990234375,
      "learning_rate": 0.00017477777280625222,
      "loss": 3.5918,
      "step": 135290
    },
    {
      "epoch": 0.1261214926101125,
      "grad_norm": 1.1868062019348145,
      "learning_rate": 0.000174775908484157,
      "loss": 3.4687,
      "step": 135300
    },
    {
      "epoch": 0.12613081422819158,
      "grad_norm": 1.0509657859802246,
      "learning_rate": 0.0001747740441620618,
      "loss": 3.6861,
      "step": 135310
    },
    {
      "epoch": 0.12614013584627068,
      "grad_norm": 1.3020825386047363,
      "learning_rate": 0.0001747721798399666,
      "loss": 3.7206,
      "step": 135320
    },
    {
      "epoch": 0.12614945746434977,
      "grad_norm": 1.6180144548416138,
      "learning_rate": 0.0001747703155178714,
      "loss": 3.5346,
      "step": 135330
    },
    {
      "epoch": 0.12615877908242887,
      "grad_norm": Infinity,
      "learning_rate": 0.0001747684511957762,
      "loss": 3.5118,
      "step": 135340
    },
    {
      "epoch": 0.12616810070050793,
      "grad_norm": 1.0675734281539917,
      "learning_rate": 0.000174766586873681,
      "loss": 3.663,
      "step": 135350
    },
    {
      "epoch": 0.12617742231858703,
      "grad_norm": 1.1934221982955933,
      "learning_rate": 0.0001747647225515858,
      "loss": 3.678,
      "step": 135360
    },
    {
      "epoch": 0.12618674393666612,
      "grad_norm": 1.0887740850448608,
      "learning_rate": 0.0001747628582294906,
      "loss": 3.9189,
      "step": 135370
    },
    {
      "epoch": 0.12619606555474522,
      "grad_norm": 0.9094374179840088,
      "learning_rate": 0.0001747609939073954,
      "loss": 3.6537,
      "step": 135380
    },
    {
      "epoch": 0.1262053871728243,
      "grad_norm": 1.075058102607727,
      "learning_rate": 0.0001747591295853002,
      "loss": 3.6016,
      "step": 135390
    },
    {
      "epoch": 0.1262147087909034,
      "grad_norm": 1.0778430700302124,
      "learning_rate": 0.000174757265263205,
      "loss": 3.6288,
      "step": 135400
    },
    {
      "epoch": 0.1262240304089825,
      "grad_norm": 1.169247031211853,
      "learning_rate": 0.0001747554009411098,
      "loss": 3.4187,
      "step": 135410
    },
    {
      "epoch": 0.1262333520270616,
      "grad_norm": 1.1584477424621582,
      "learning_rate": 0.00017475353661901462,
      "loss": 3.7334,
      "step": 135420
    },
    {
      "epoch": 0.1262426736451407,
      "grad_norm": 0.9572957754135132,
      "learning_rate": 0.0001747516722969194,
      "loss": 3.5132,
      "step": 135430
    },
    {
      "epoch": 0.12625199526321979,
      "grad_norm": 1.0389223098754883,
      "learning_rate": 0.0001747498079748242,
      "loss": 3.7324,
      "step": 135440
    },
    {
      "epoch": 0.12626131688129885,
      "grad_norm": 1.112805724143982,
      "learning_rate": 0.000174747943652729,
      "loss": 3.3706,
      "step": 135450
    },
    {
      "epoch": 0.12627063849937795,
      "grad_norm": 1.133183479309082,
      "learning_rate": 0.00017474607933063382,
      "loss": 3.5998,
      "step": 135460
    },
    {
      "epoch": 0.12627996011745704,
      "grad_norm": 1.2500810623168945,
      "learning_rate": 0.00017474421500853862,
      "loss": 3.5686,
      "step": 135470
    },
    {
      "epoch": 0.12628928173553614,
      "grad_norm": 1.0819189548492432,
      "learning_rate": 0.0001747423506864434,
      "loss": 3.6974,
      "step": 135480
    },
    {
      "epoch": 0.12629860335361523,
      "grad_norm": 0.9560720324516296,
      "learning_rate": 0.0001747404863643482,
      "loss": 3.6166,
      "step": 135490
    },
    {
      "epoch": 0.12630792497169432,
      "grad_norm": 1.2555071115493774,
      "learning_rate": 0.00017473862204225299,
      "loss": 3.4124,
      "step": 135500
    },
    {
      "epoch": 0.12631724658977342,
      "grad_norm": 0.8301492929458618,
      "learning_rate": 0.00017473675772015782,
      "loss": 3.6474,
      "step": 135510
    },
    {
      "epoch": 0.1263265682078525,
      "grad_norm": 1.0309937000274658,
      "learning_rate": 0.00017473489339806263,
      "loss": 3.5621,
      "step": 135520
    },
    {
      "epoch": 0.1263358898259316,
      "grad_norm": 1.1604925394058228,
      "learning_rate": 0.0001747330290759674,
      "loss": 3.6103,
      "step": 135530
    },
    {
      "epoch": 0.1263452114440107,
      "grad_norm": 1.0399503707885742,
      "learning_rate": 0.0001747311647538722,
      "loss": 3.6608,
      "step": 135540
    },
    {
      "epoch": 0.1263545330620898,
      "grad_norm": 1.1070237159729004,
      "learning_rate": 0.000174729300431777,
      "loss": 3.5944,
      "step": 135550
    },
    {
      "epoch": 0.12636385468016886,
      "grad_norm": 1.1308951377868652,
      "learning_rate": 0.0001747274361096818,
      "loss": 3.4542,
      "step": 135560
    },
    {
      "epoch": 0.12637317629824796,
      "grad_norm": 0.9258748292922974,
      "learning_rate": 0.00017472557178758663,
      "loss": 3.5526,
      "step": 135570
    },
    {
      "epoch": 0.12638249791632705,
      "grad_norm": 1.113644003868103,
      "learning_rate": 0.0001747237074654914,
      "loss": 3.6803,
      "step": 135580
    },
    {
      "epoch": 0.12639181953440615,
      "grad_norm": 0.9666551351547241,
      "learning_rate": 0.00017472184314339622,
      "loss": 3.5262,
      "step": 135590
    },
    {
      "epoch": 0.12640114115248524,
      "grad_norm": 1.2312102317810059,
      "learning_rate": 0.000174719978821301,
      "loss": 3.6937,
      "step": 135600
    },
    {
      "epoch": 0.12641046277056434,
      "grad_norm": 1.1085211038589478,
      "learning_rate": 0.0001747181144992058,
      "loss": 3.5921,
      "step": 135610
    },
    {
      "epoch": 0.12641978438864343,
      "grad_norm": 0.9373235106468201,
      "learning_rate": 0.0001747162501771106,
      "loss": 3.6079,
      "step": 135620
    },
    {
      "epoch": 0.12642910600672252,
      "grad_norm": 1.2784720659255981,
      "learning_rate": 0.00017471438585501541,
      "loss": 3.6751,
      "step": 135630
    },
    {
      "epoch": 0.12643842762480162,
      "grad_norm": 1.0309377908706665,
      "learning_rate": 0.00017471252153292022,
      "loss": 3.5072,
      "step": 135640
    },
    {
      "epoch": 0.1264477492428807,
      "grad_norm": 1.0839556455612183,
      "learning_rate": 0.000174710657210825,
      "loss": 3.5879,
      "step": 135650
    },
    {
      "epoch": 0.12645707086095978,
      "grad_norm": 1.0567536354064941,
      "learning_rate": 0.0001747087928887298,
      "loss": 3.5545,
      "step": 135660
    },
    {
      "epoch": 0.12646639247903887,
      "grad_norm": 1.1836094856262207,
      "learning_rate": 0.00017470692856663459,
      "loss": 3.6444,
      "step": 135670
    },
    {
      "epoch": 0.12647571409711797,
      "grad_norm": 1.1864842176437378,
      "learning_rate": 0.00017470506424453942,
      "loss": 3.6601,
      "step": 135680
    },
    {
      "epoch": 0.12648503571519706,
      "grad_norm": 1.053471326828003,
      "learning_rate": 0.00017470319992244422,
      "loss": 3.6484,
      "step": 135690
    },
    {
      "epoch": 0.12649435733327616,
      "grad_norm": 0.9771831035614014,
      "learning_rate": 0.000174701335600349,
      "loss": 3.7175,
      "step": 135700
    },
    {
      "epoch": 0.12650367895135525,
      "grad_norm": 1.0136220455169678,
      "learning_rate": 0.0001746994712782538,
      "loss": 3.6504,
      "step": 135710
    },
    {
      "epoch": 0.12651300056943435,
      "grad_norm": 1.0165214538574219,
      "learning_rate": 0.0001746976069561586,
      "loss": 3.7676,
      "step": 135720
    },
    {
      "epoch": 0.12652232218751344,
      "grad_norm": 1.149699091911316,
      "learning_rate": 0.0001746957426340634,
      "loss": 3.5136,
      "step": 135730
    },
    {
      "epoch": 0.12653164380559254,
      "grad_norm": 1.141837239265442,
      "learning_rate": 0.00017469387831196823,
      "loss": 3.6018,
      "step": 135740
    },
    {
      "epoch": 0.12654096542367163,
      "grad_norm": 1.252392292022705,
      "learning_rate": 0.000174692013989873,
      "loss": 3.6459,
      "step": 135750
    },
    {
      "epoch": 0.12655028704175073,
      "grad_norm": 0.9684142470359802,
      "learning_rate": 0.00017469014966777781,
      "loss": 3.5415,
      "step": 135760
    },
    {
      "epoch": 0.1265596086598298,
      "grad_norm": 1.1857634782791138,
      "learning_rate": 0.0001746882853456826,
      "loss": 3.5818,
      "step": 135770
    },
    {
      "epoch": 0.1265689302779089,
      "grad_norm": 1.0905513763427734,
      "learning_rate": 0.0001746864210235874,
      "loss": 3.6271,
      "step": 135780
    },
    {
      "epoch": 0.12657825189598798,
      "grad_norm": 0.9312757849693298,
      "learning_rate": 0.0001746845567014922,
      "loss": 3.5744,
      "step": 135790
    },
    {
      "epoch": 0.12658757351406708,
      "grad_norm": 0.7945520281791687,
      "learning_rate": 0.000174682692379397,
      "loss": 3.6811,
      "step": 135800
    },
    {
      "epoch": 0.12659689513214617,
      "grad_norm": 1.0043823719024658,
      "learning_rate": 0.00017468082805730182,
      "loss": 3.5758,
      "step": 135810
    },
    {
      "epoch": 0.12660621675022526,
      "grad_norm": 1.0221952199935913,
      "learning_rate": 0.0001746789637352066,
      "loss": 3.6872,
      "step": 135820
    },
    {
      "epoch": 0.12661553836830436,
      "grad_norm": 1.0368220806121826,
      "learning_rate": 0.0001746770994131114,
      "loss": 3.5593,
      "step": 135830
    },
    {
      "epoch": 0.12662485998638345,
      "grad_norm": 1.1054986715316772,
      "learning_rate": 0.0001746752350910162,
      "loss": 3.6498,
      "step": 135840
    },
    {
      "epoch": 0.12663418160446255,
      "grad_norm": 1.074057936668396,
      "learning_rate": 0.00017467337076892102,
      "loss": 3.6141,
      "step": 135850
    },
    {
      "epoch": 0.12664350322254164,
      "grad_norm": 1.0507861375808716,
      "learning_rate": 0.00017467150644682582,
      "loss": 3.6733,
      "step": 135860
    },
    {
      "epoch": 0.1266528248406207,
      "grad_norm": 0.8835662007331848,
      "learning_rate": 0.0001746696421247306,
      "loss": 3.6881,
      "step": 135870
    },
    {
      "epoch": 0.1266621464586998,
      "grad_norm": 0.9677097797393799,
      "learning_rate": 0.0001746677778026354,
      "loss": 3.6004,
      "step": 135880
    },
    {
      "epoch": 0.1266714680767789,
      "grad_norm": 0.8422546982765198,
      "learning_rate": 0.00017466591348054022,
      "loss": 3.6845,
      "step": 135890
    },
    {
      "epoch": 0.126680789694858,
      "grad_norm": 1.0779885053634644,
      "learning_rate": 0.00017466404915844502,
      "loss": 3.6635,
      "step": 135900
    },
    {
      "epoch": 0.1266901113129371,
      "grad_norm": 1.1377500295639038,
      "learning_rate": 0.00017466218483634983,
      "loss": 3.4528,
      "step": 135910
    },
    {
      "epoch": 0.12669943293101618,
      "grad_norm": 1.1962103843688965,
      "learning_rate": 0.0001746603205142546,
      "loss": 3.7093,
      "step": 135920
    },
    {
      "epoch": 0.12670875454909528,
      "grad_norm": 1.1852680444717407,
      "learning_rate": 0.0001746584561921594,
      "loss": 3.6884,
      "step": 135930
    },
    {
      "epoch": 0.12671807616717437,
      "grad_norm": 1.042742133140564,
      "learning_rate": 0.00017465659187006422,
      "loss": 3.5047,
      "step": 135940
    },
    {
      "epoch": 0.12672739778525346,
      "grad_norm": 1.038472056388855,
      "learning_rate": 0.000174654727547969,
      "loss": 3.5776,
      "step": 135950
    },
    {
      "epoch": 0.12673671940333256,
      "grad_norm": 1.135931134223938,
      "learning_rate": 0.00017465286322587383,
      "loss": 3.6542,
      "step": 135960
    },
    {
      "epoch": 0.12674604102141165,
      "grad_norm": 1.1031031608581543,
      "learning_rate": 0.0001746509989037786,
      "loss": 3.7406,
      "step": 135970
    },
    {
      "epoch": 0.12675536263949072,
      "grad_norm": 1.0182061195373535,
      "learning_rate": 0.00017464913458168342,
      "loss": 3.7949,
      "step": 135980
    },
    {
      "epoch": 0.12676468425756982,
      "grad_norm": 1.206743836402893,
      "learning_rate": 0.00017464727025958822,
      "loss": 3.7028,
      "step": 135990
    },
    {
      "epoch": 0.1267740058756489,
      "grad_norm": 1.0275548696517944,
      "learning_rate": 0.000174645405937493,
      "loss": 3.5844,
      "step": 136000
    },
    {
      "epoch": 0.126783327493728,
      "grad_norm": 1.1192166805267334,
      "learning_rate": 0.0001746435416153978,
      "loss": 3.4804,
      "step": 136010
    },
    {
      "epoch": 0.1267926491118071,
      "grad_norm": 0.9581682085990906,
      "learning_rate": 0.00017464167729330262,
      "loss": 3.6349,
      "step": 136020
    },
    {
      "epoch": 0.1268019707298862,
      "grad_norm": 1.0157567262649536,
      "learning_rate": 0.00017463981297120742,
      "loss": 3.6981,
      "step": 136030
    },
    {
      "epoch": 0.1268112923479653,
      "grad_norm": 1.1681057214736938,
      "learning_rate": 0.00017463794864911223,
      "loss": 3.4885,
      "step": 136040
    },
    {
      "epoch": 0.12682061396604438,
      "grad_norm": 1.0172743797302246,
      "learning_rate": 0.000174636084327017,
      "loss": 3.624,
      "step": 136050
    },
    {
      "epoch": 0.12682993558412348,
      "grad_norm": 0.9691790342330933,
      "learning_rate": 0.00017463422000492181,
      "loss": 3.6727,
      "step": 136060
    },
    {
      "epoch": 0.12683925720220257,
      "grad_norm": 1.0163464546203613,
      "learning_rate": 0.00017463235568282662,
      "loss": 3.7773,
      "step": 136070
    },
    {
      "epoch": 0.12684857882028164,
      "grad_norm": 1.0530520677566528,
      "learning_rate": 0.00017463049136073143,
      "loss": 3.4642,
      "step": 136080
    },
    {
      "epoch": 0.12685790043836073,
      "grad_norm": 0.9889556765556335,
      "learning_rate": 0.00017462862703863623,
      "loss": 3.6772,
      "step": 136090
    },
    {
      "epoch": 0.12686722205643983,
      "grad_norm": 1.0163710117340088,
      "learning_rate": 0.000174626762716541,
      "loss": 3.4032,
      "step": 136100
    },
    {
      "epoch": 0.12687654367451892,
      "grad_norm": 1.1413942575454712,
      "learning_rate": 0.00017462489839444582,
      "loss": 3.5285,
      "step": 136110
    },
    {
      "epoch": 0.12688586529259802,
      "grad_norm": 1.0493357181549072,
      "learning_rate": 0.0001746230340723506,
      "loss": 3.9241,
      "step": 136120
    },
    {
      "epoch": 0.1268951869106771,
      "grad_norm": 1.2180818319320679,
      "learning_rate": 0.00017462116975025543,
      "loss": 3.6055,
      "step": 136130
    },
    {
      "epoch": 0.1269045085287562,
      "grad_norm": 1.0103191137313843,
      "learning_rate": 0.00017461930542816024,
      "loss": 3.7473,
      "step": 136140
    },
    {
      "epoch": 0.1269138301468353,
      "grad_norm": 1.058199405670166,
      "learning_rate": 0.00017461744110606502,
      "loss": 3.4725,
      "step": 136150
    },
    {
      "epoch": 0.1269231517649144,
      "grad_norm": 1.1690587997436523,
      "learning_rate": 0.00017461557678396982,
      "loss": 3.4926,
      "step": 136160
    },
    {
      "epoch": 0.1269324733829935,
      "grad_norm": 1.144351840019226,
      "learning_rate": 0.0001746137124618746,
      "loss": 3.4884,
      "step": 136170
    },
    {
      "epoch": 0.12694179500107255,
      "grad_norm": 0.9200192093849182,
      "learning_rate": 0.0001746118481397794,
      "loss": 3.3298,
      "step": 136180
    },
    {
      "epoch": 0.12695111661915165,
      "grad_norm": 0.9281372427940369,
      "learning_rate": 0.00017460998381768424,
      "loss": 3.5723,
      "step": 136190
    },
    {
      "epoch": 0.12696043823723074,
      "grad_norm": 1.1617118120193481,
      "learning_rate": 0.00017460811949558902,
      "loss": 3.544,
      "step": 136200
    },
    {
      "epoch": 0.12696975985530984,
      "grad_norm": 1.45936119556427,
      "learning_rate": 0.00017460625517349383,
      "loss": 3.3581,
      "step": 136210
    },
    {
      "epoch": 0.12697908147338893,
      "grad_norm": 1.076317310333252,
      "learning_rate": 0.0001746043908513986,
      "loss": 3.4578,
      "step": 136220
    },
    {
      "epoch": 0.12698840309146803,
      "grad_norm": 1.1611037254333496,
      "learning_rate": 0.0001746025265293034,
      "loss": 3.5631,
      "step": 136230
    },
    {
      "epoch": 0.12699772470954712,
      "grad_norm": 1.1571180820465088,
      "learning_rate": 0.00017460066220720822,
      "loss": 3.5392,
      "step": 136240
    },
    {
      "epoch": 0.12700704632762622,
      "grad_norm": 0.9535903930664062,
      "learning_rate": 0.00017459879788511303,
      "loss": 3.4462,
      "step": 136250
    },
    {
      "epoch": 0.1270163679457053,
      "grad_norm": 1.0041053295135498,
      "learning_rate": 0.00017459693356301783,
      "loss": 3.6926,
      "step": 136260
    },
    {
      "epoch": 0.1270256895637844,
      "grad_norm": 0.9856420159339905,
      "learning_rate": 0.0001745950692409226,
      "loss": 3.4764,
      "step": 136270
    },
    {
      "epoch": 0.1270350111818635,
      "grad_norm": 0.9642185568809509,
      "learning_rate": 0.00017459320491882742,
      "loss": 3.6347,
      "step": 136280
    },
    {
      "epoch": 0.12704433279994257,
      "grad_norm": 1.0332887172698975,
      "learning_rate": 0.00017459134059673222,
      "loss": 3.6563,
      "step": 136290
    },
    {
      "epoch": 0.12705365441802166,
      "grad_norm": 1.000420331954956,
      "learning_rate": 0.00017458947627463703,
      "loss": 3.6929,
      "step": 136300
    },
    {
      "epoch": 0.12706297603610076,
      "grad_norm": 1.1268601417541504,
      "learning_rate": 0.00017458761195254184,
      "loss": 3.5976,
      "step": 136310
    },
    {
      "epoch": 0.12707229765417985,
      "grad_norm": 1.177093744277954,
      "learning_rate": 0.00017458574763044662,
      "loss": 3.6328,
      "step": 136320
    },
    {
      "epoch": 0.12708161927225894,
      "grad_norm": 1.0940965414047241,
      "learning_rate": 0.00017458388330835142,
      "loss": 3.6914,
      "step": 136330
    },
    {
      "epoch": 0.12709094089033804,
      "grad_norm": 0.9829176068305969,
      "learning_rate": 0.00017458201898625623,
      "loss": 3.7328,
      "step": 136340
    },
    {
      "epoch": 0.12710026250841713,
      "grad_norm": 1.0883796215057373,
      "learning_rate": 0.00017458015466416103,
      "loss": 3.6855,
      "step": 136350
    },
    {
      "epoch": 0.12710958412649623,
      "grad_norm": 1.1487891674041748,
      "learning_rate": 0.00017457829034206584,
      "loss": 3.7279,
      "step": 136360
    },
    {
      "epoch": 0.12711890574457532,
      "grad_norm": 1.1893095970153809,
      "learning_rate": 0.00017457642601997062,
      "loss": 3.6649,
      "step": 136370
    },
    {
      "epoch": 0.12712822736265442,
      "grad_norm": 1.0077126026153564,
      "learning_rate": 0.00017457456169787543,
      "loss": 3.6431,
      "step": 136380
    },
    {
      "epoch": 0.12713754898073348,
      "grad_norm": 0.9194217324256897,
      "learning_rate": 0.00017457269737578023,
      "loss": 3.7646,
      "step": 136390
    },
    {
      "epoch": 0.12714687059881258,
      "grad_norm": 1.2209877967834473,
      "learning_rate": 0.000174570833053685,
      "loss": 3.6303,
      "step": 136400
    },
    {
      "epoch": 0.12715619221689167,
      "grad_norm": 1.106488823890686,
      "learning_rate": 0.00017456896873158985,
      "loss": 3.5194,
      "step": 136410
    },
    {
      "epoch": 0.12716551383497077,
      "grad_norm": 1.133682131767273,
      "learning_rate": 0.00017456710440949462,
      "loss": 3.4177,
      "step": 136420
    },
    {
      "epoch": 0.12717483545304986,
      "grad_norm": 0.907978355884552,
      "learning_rate": 0.00017456524008739943,
      "loss": 3.7382,
      "step": 136430
    },
    {
      "epoch": 0.12718415707112896,
      "grad_norm": 1.0129444599151611,
      "learning_rate": 0.00017456337576530424,
      "loss": 3.6519,
      "step": 136440
    },
    {
      "epoch": 0.12719347868920805,
      "grad_norm": 1.133984088897705,
      "learning_rate": 0.00017456151144320902,
      "loss": 3.8191,
      "step": 136450
    },
    {
      "epoch": 0.12720280030728714,
      "grad_norm": 1.1314470767974854,
      "learning_rate": 0.00017455964712111382,
      "loss": 3.5648,
      "step": 136460
    },
    {
      "epoch": 0.12721212192536624,
      "grad_norm": 1.1800434589385986,
      "learning_rate": 0.00017455778279901863,
      "loss": 3.5737,
      "step": 136470
    },
    {
      "epoch": 0.12722144354344533,
      "grad_norm": 1.0060151815414429,
      "learning_rate": 0.00017455591847692344,
      "loss": 3.7408,
      "step": 136480
    },
    {
      "epoch": 0.12723076516152443,
      "grad_norm": 1.001448154449463,
      "learning_rate": 0.00017455405415482824,
      "loss": 3.7615,
      "step": 136490
    },
    {
      "epoch": 0.1272400867796035,
      "grad_norm": 0.9728806018829346,
      "learning_rate": 0.00017455218983273302,
      "loss": 3.5721,
      "step": 136500
    },
    {
      "epoch": 0.1272494083976826,
      "grad_norm": 0.9253226518630981,
      "learning_rate": 0.00017455032551063783,
      "loss": 3.831,
      "step": 136510
    },
    {
      "epoch": 0.12725873001576168,
      "grad_norm": 0.8946741819381714,
      "learning_rate": 0.00017454846118854263,
      "loss": 3.7044,
      "step": 136520
    },
    {
      "epoch": 0.12726805163384078,
      "grad_norm": 1.065974473953247,
      "learning_rate": 0.00017454659686644744,
      "loss": 3.6169,
      "step": 136530
    },
    {
      "epoch": 0.12727737325191987,
      "grad_norm": 1.0538331270217896,
      "learning_rate": 0.00017454473254435225,
      "loss": 3.652,
      "step": 136540
    },
    {
      "epoch": 0.12728669486999897,
      "grad_norm": 0.7755545377731323,
      "learning_rate": 0.00017454286822225703,
      "loss": 3.7769,
      "step": 136550
    },
    {
      "epoch": 0.12729601648807806,
      "grad_norm": 1.0743272304534912,
      "learning_rate": 0.00017454100390016183,
      "loss": 3.5108,
      "step": 136560
    },
    {
      "epoch": 0.12730533810615716,
      "grad_norm": 1.0857945680618286,
      "learning_rate": 0.0001745391395780666,
      "loss": 3.5292,
      "step": 136570
    },
    {
      "epoch": 0.12731465972423625,
      "grad_norm": 0.9901275634765625,
      "learning_rate": 0.00017453727525597144,
      "loss": 3.7139,
      "step": 136580
    },
    {
      "epoch": 0.12732398134231535,
      "grad_norm": 1.0762505531311035,
      "learning_rate": 0.00017453541093387625,
      "loss": 3.6923,
      "step": 136590
    },
    {
      "epoch": 0.1273333029603944,
      "grad_norm": 0.9332774877548218,
      "learning_rate": 0.00017453354661178103,
      "loss": 3.5303,
      "step": 136600
    },
    {
      "epoch": 0.1273426245784735,
      "grad_norm": 1.0383354425430298,
      "learning_rate": 0.00017453168228968584,
      "loss": 3.6973,
      "step": 136610
    },
    {
      "epoch": 0.1273519461965526,
      "grad_norm": 0.9792401790618896,
      "learning_rate": 0.00017452981796759062,
      "loss": 3.6287,
      "step": 136620
    },
    {
      "epoch": 0.1273612678146317,
      "grad_norm": 1.1386473178863525,
      "learning_rate": 0.00017452795364549542,
      "loss": 3.6469,
      "step": 136630
    },
    {
      "epoch": 0.1273705894327108,
      "grad_norm": 1.1065402030944824,
      "learning_rate": 0.00017452608932340025,
      "loss": 3.5417,
      "step": 136640
    },
    {
      "epoch": 0.12737991105078988,
      "grad_norm": 0.8861261010169983,
      "learning_rate": 0.00017452422500130503,
      "loss": 3.6997,
      "step": 136650
    },
    {
      "epoch": 0.12738923266886898,
      "grad_norm": 1.3044469356536865,
      "learning_rate": 0.00017452236067920984,
      "loss": 3.654,
      "step": 136660
    },
    {
      "epoch": 0.12739855428694807,
      "grad_norm": 1.0695229768753052,
      "learning_rate": 0.00017452049635711462,
      "loss": 3.743,
      "step": 136670
    },
    {
      "epoch": 0.12740787590502717,
      "grad_norm": 1.0197317600250244,
      "learning_rate": 0.00017451863203501943,
      "loss": 3.6714,
      "step": 136680
    },
    {
      "epoch": 0.12741719752310626,
      "grad_norm": 1.125883936882019,
      "learning_rate": 0.00017451676771292423,
      "loss": 3.3131,
      "step": 136690
    },
    {
      "epoch": 0.12742651914118536,
      "grad_norm": 0.9703460931777954,
      "learning_rate": 0.00017451490339082904,
      "loss": 3.6518,
      "step": 136700
    },
    {
      "epoch": 0.12743584075926442,
      "grad_norm": 0.8230243921279907,
      "learning_rate": 0.00017451303906873384,
      "loss": 3.6797,
      "step": 136710
    },
    {
      "epoch": 0.12744516237734352,
      "grad_norm": 0.9087804555892944,
      "learning_rate": 0.00017451117474663862,
      "loss": 3.6707,
      "step": 136720
    },
    {
      "epoch": 0.1274544839954226,
      "grad_norm": 1.0748728513717651,
      "learning_rate": 0.00017450931042454343,
      "loss": 3.4801,
      "step": 136730
    },
    {
      "epoch": 0.1274638056135017,
      "grad_norm": 1.2325319051742554,
      "learning_rate": 0.00017450744610244824,
      "loss": 3.6631,
      "step": 136740
    },
    {
      "epoch": 0.1274731272315808,
      "grad_norm": 1.254675269126892,
      "learning_rate": 0.00017450558178035304,
      "loss": 3.649,
      "step": 136750
    },
    {
      "epoch": 0.1274824488496599,
      "grad_norm": 0.806750476360321,
      "learning_rate": 0.00017450371745825785,
      "loss": 3.7202,
      "step": 136760
    },
    {
      "epoch": 0.127491770467739,
      "grad_norm": 0.9912461042404175,
      "learning_rate": 0.00017450185313616263,
      "loss": 3.5477,
      "step": 136770
    },
    {
      "epoch": 0.12750109208581809,
      "grad_norm": 0.9440166354179382,
      "learning_rate": 0.00017449998881406743,
      "loss": 3.7102,
      "step": 136780
    },
    {
      "epoch": 0.12751041370389718,
      "grad_norm": 1.01021146774292,
      "learning_rate": 0.00017449812449197224,
      "loss": 3.5984,
      "step": 136790
    },
    {
      "epoch": 0.12751973532197627,
      "grad_norm": 1.0167258977890015,
      "learning_rate": 0.00017449626016987705,
      "loss": 3.4089,
      "step": 136800
    },
    {
      "epoch": 0.12752905694005534,
      "grad_norm": 1.0722429752349854,
      "learning_rate": 0.00017449439584778185,
      "loss": 3.6992,
      "step": 136810
    },
    {
      "epoch": 0.12753837855813444,
      "grad_norm": 0.8143043518066406,
      "learning_rate": 0.00017449253152568663,
      "loss": 3.5019,
      "step": 136820
    },
    {
      "epoch": 0.12754770017621353,
      "grad_norm": 1.0313693284988403,
      "learning_rate": 0.00017449066720359144,
      "loss": 3.637,
      "step": 136830
    },
    {
      "epoch": 0.12755702179429262,
      "grad_norm": 1.0162657499313354,
      "learning_rate": 0.00017448880288149625,
      "loss": 3.6577,
      "step": 136840
    },
    {
      "epoch": 0.12756634341237172,
      "grad_norm": 1.1540858745574951,
      "learning_rate": 0.00017448693855940102,
      "loss": 3.5903,
      "step": 136850
    },
    {
      "epoch": 0.1275756650304508,
      "grad_norm": 0.9298204183578491,
      "learning_rate": 0.00017448507423730586,
      "loss": 3.6844,
      "step": 136860
    },
    {
      "epoch": 0.1275849866485299,
      "grad_norm": 0.9139224886894226,
      "learning_rate": 0.00017448320991521064,
      "loss": 3.6318,
      "step": 136870
    },
    {
      "epoch": 0.127594308266609,
      "grad_norm": 1.0190091133117676,
      "learning_rate": 0.00017448134559311544,
      "loss": 3.7551,
      "step": 136880
    },
    {
      "epoch": 0.1276036298846881,
      "grad_norm": 1.1120611429214478,
      "learning_rate": 0.00017447948127102025,
      "loss": 3.4868,
      "step": 136890
    },
    {
      "epoch": 0.1276129515027672,
      "grad_norm": 0.9253730177879333,
      "learning_rate": 0.00017447761694892503,
      "loss": 3.6071,
      "step": 136900
    },
    {
      "epoch": 0.12762227312084629,
      "grad_norm": 1.0484797954559326,
      "learning_rate": 0.00017447575262682984,
      "loss": 3.5967,
      "step": 136910
    },
    {
      "epoch": 0.12763159473892535,
      "grad_norm": 1.2944955825805664,
      "learning_rate": 0.00017447388830473464,
      "loss": 3.4164,
      "step": 136920
    },
    {
      "epoch": 0.12764091635700445,
      "grad_norm": 1.1676102876663208,
      "learning_rate": 0.00017447202398263945,
      "loss": 3.5125,
      "step": 136930
    },
    {
      "epoch": 0.12765023797508354,
      "grad_norm": 1.1043777465820312,
      "learning_rate": 0.00017447015966054425,
      "loss": 3.5521,
      "step": 136940
    },
    {
      "epoch": 0.12765955959316264,
      "grad_norm": 0.8797561526298523,
      "learning_rate": 0.00017446829533844903,
      "loss": 3.4904,
      "step": 136950
    },
    {
      "epoch": 0.12766888121124173,
      "grad_norm": 0.9287214279174805,
      "learning_rate": 0.00017446643101635384,
      "loss": 3.4946,
      "step": 136960
    },
    {
      "epoch": 0.12767820282932082,
      "grad_norm": 1.0363620519638062,
      "learning_rate": 0.00017446456669425865,
      "loss": 3.4946,
      "step": 136970
    },
    {
      "epoch": 0.12768752444739992,
      "grad_norm": 1.2228811979293823,
      "learning_rate": 0.00017446270237216345,
      "loss": 3.7618,
      "step": 136980
    },
    {
      "epoch": 0.127696846065479,
      "grad_norm": 1.098602294921875,
      "learning_rate": 0.00017446083805006823,
      "loss": 3.8199,
      "step": 136990
    },
    {
      "epoch": 0.1277061676835581,
      "grad_norm": 0.9993955492973328,
      "learning_rate": 0.00017445897372797304,
      "loss": 3.5266,
      "step": 137000
    },
    {
      "epoch": 0.1277154893016372,
      "grad_norm": 1.1385811567306519,
      "learning_rate": 0.00017445710940587784,
      "loss": 3.6137,
      "step": 137010
    },
    {
      "epoch": 0.12772481091971627,
      "grad_norm": 0.9068946242332458,
      "learning_rate": 0.00017445524508378262,
      "loss": 3.4817,
      "step": 137020
    },
    {
      "epoch": 0.12773413253779536,
      "grad_norm": 1.169710397720337,
      "learning_rate": 0.00017445338076168746,
      "loss": 3.7314,
      "step": 137030
    },
    {
      "epoch": 0.12774345415587446,
      "grad_norm": 1.058281660079956,
      "learning_rate": 0.00017445151643959224,
      "loss": 3.5176,
      "step": 137040
    },
    {
      "epoch": 0.12775277577395355,
      "grad_norm": 1.0390291213989258,
      "learning_rate": 0.00017444965211749704,
      "loss": 3.5859,
      "step": 137050
    },
    {
      "epoch": 0.12776209739203265,
      "grad_norm": 1.1277002096176147,
      "learning_rate": 0.00017444778779540185,
      "loss": 3.4711,
      "step": 137060
    },
    {
      "epoch": 0.12777141901011174,
      "grad_norm": 1.0818487405776978,
      "learning_rate": 0.00017444592347330663,
      "loss": 3.4198,
      "step": 137070
    },
    {
      "epoch": 0.12778074062819084,
      "grad_norm": 0.9625991582870483,
      "learning_rate": 0.00017444405915121143,
      "loss": 3.5245,
      "step": 137080
    },
    {
      "epoch": 0.12779006224626993,
      "grad_norm": 1.0141077041625977,
      "learning_rate": 0.00017444219482911624,
      "loss": 3.6633,
      "step": 137090
    },
    {
      "epoch": 0.12779938386434903,
      "grad_norm": 0.7844510078430176,
      "learning_rate": 0.00017444033050702105,
      "loss": 3.772,
      "step": 137100
    },
    {
      "epoch": 0.12780870548242812,
      "grad_norm": 1.153554081916809,
      "learning_rate": 0.00017443846618492585,
      "loss": 3.6564,
      "step": 137110
    },
    {
      "epoch": 0.12781802710050721,
      "grad_norm": 0.8937996625900269,
      "learning_rate": 0.00017443660186283063,
      "loss": 3.6821,
      "step": 137120
    },
    {
      "epoch": 0.12782734871858628,
      "grad_norm": 0.9309395551681519,
      "learning_rate": 0.00017443473754073544,
      "loss": 3.6677,
      "step": 137130
    },
    {
      "epoch": 0.12783667033666538,
      "grad_norm": 1.0205435752868652,
      "learning_rate": 0.00017443287321864025,
      "loss": 3.4458,
      "step": 137140
    },
    {
      "epoch": 0.12784599195474447,
      "grad_norm": 1.1231334209442139,
      "learning_rate": 0.00017443100889654505,
      "loss": 3.6332,
      "step": 137150
    },
    {
      "epoch": 0.12785531357282356,
      "grad_norm": 0.9371398687362671,
      "learning_rate": 0.00017442914457444986,
      "loss": 3.9282,
      "step": 137160
    },
    {
      "epoch": 0.12786463519090266,
      "grad_norm": 1.0501374006271362,
      "learning_rate": 0.00017442728025235464,
      "loss": 3.5463,
      "step": 137170
    },
    {
      "epoch": 0.12787395680898175,
      "grad_norm": 1.3863555192947388,
      "learning_rate": 0.00017442541593025944,
      "loss": 3.7365,
      "step": 137180
    },
    {
      "epoch": 0.12788327842706085,
      "grad_norm": 1.1629679203033447,
      "learning_rate": 0.00017442355160816425,
      "loss": 3.58,
      "step": 137190
    },
    {
      "epoch": 0.12789260004513994,
      "grad_norm": 1.0971429347991943,
      "learning_rate": 0.00017442168728606906,
      "loss": 3.6306,
      "step": 137200
    },
    {
      "epoch": 0.12790192166321904,
      "grad_norm": 1.0604616403579712,
      "learning_rate": 0.00017441982296397386,
      "loss": 3.7122,
      "step": 137210
    },
    {
      "epoch": 0.12791124328129813,
      "grad_norm": 1.0191272497177124,
      "learning_rate": 0.00017441795864187864,
      "loss": 3.6198,
      "step": 137220
    },
    {
      "epoch": 0.1279205648993772,
      "grad_norm": 1.1557376384735107,
      "learning_rate": 0.00017441609431978345,
      "loss": 3.5638,
      "step": 137230
    },
    {
      "epoch": 0.1279298865174563,
      "grad_norm": 1.2479711771011353,
      "learning_rate": 0.00017441422999768823,
      "loss": 3.7003,
      "step": 137240
    },
    {
      "epoch": 0.1279392081355354,
      "grad_norm": 1.1400278806686401,
      "learning_rate": 0.00017441236567559306,
      "loss": 3.6851,
      "step": 137250
    },
    {
      "epoch": 0.12794852975361448,
      "grad_norm": 0.9251853823661804,
      "learning_rate": 0.00017441050135349787,
      "loss": 3.4033,
      "step": 137260
    },
    {
      "epoch": 0.12795785137169358,
      "grad_norm": 0.82173091173172,
      "learning_rate": 0.00017440863703140265,
      "loss": 3.7286,
      "step": 137270
    },
    {
      "epoch": 0.12796717298977267,
      "grad_norm": 0.8906792998313904,
      "learning_rate": 0.00017440677270930745,
      "loss": 3.7293,
      "step": 137280
    },
    {
      "epoch": 0.12797649460785177,
      "grad_norm": 1.228031873703003,
      "learning_rate": 0.00017440490838721223,
      "loss": 3.5717,
      "step": 137290
    },
    {
      "epoch": 0.12798581622593086,
      "grad_norm": 1.0177937746047974,
      "learning_rate": 0.00017440304406511704,
      "loss": 3.6849,
      "step": 137300
    },
    {
      "epoch": 0.12799513784400995,
      "grad_norm": 0.9957128763198853,
      "learning_rate": 0.00017440117974302187,
      "loss": 3.5772,
      "step": 137310
    },
    {
      "epoch": 0.12800445946208905,
      "grad_norm": 0.9880426526069641,
      "learning_rate": 0.00017439931542092665,
      "loss": 3.666,
      "step": 137320
    },
    {
      "epoch": 0.12801378108016814,
      "grad_norm": 0.9270479083061218,
      "learning_rate": 0.00017439745109883146,
      "loss": 3.4464,
      "step": 137330
    },
    {
      "epoch": 0.1280231026982472,
      "grad_norm": 1.110845923423767,
      "learning_rate": 0.00017439558677673624,
      "loss": 3.5417,
      "step": 137340
    },
    {
      "epoch": 0.1280324243163263,
      "grad_norm": 0.9786825776100159,
      "learning_rate": 0.00017439372245464104,
      "loss": 3.6826,
      "step": 137350
    },
    {
      "epoch": 0.1280417459344054,
      "grad_norm": 1.013375163078308,
      "learning_rate": 0.00017439185813254585,
      "loss": 3.3585,
      "step": 137360
    },
    {
      "epoch": 0.1280510675524845,
      "grad_norm": 0.8765349388122559,
      "learning_rate": 0.00017438999381045065,
      "loss": 3.7064,
      "step": 137370
    },
    {
      "epoch": 0.1280603891705636,
      "grad_norm": 1.1759799718856812,
      "learning_rate": 0.00017438812948835546,
      "loss": 3.4754,
      "step": 137380
    },
    {
      "epoch": 0.12806971078864268,
      "grad_norm": 1.0929181575775146,
      "learning_rate": 0.00017438626516626024,
      "loss": 3.6068,
      "step": 137390
    },
    {
      "epoch": 0.12807903240672178,
      "grad_norm": 1.408432126045227,
      "learning_rate": 0.00017438440084416505,
      "loss": 3.5774,
      "step": 137400
    },
    {
      "epoch": 0.12808835402480087,
      "grad_norm": 0.8831470012664795,
      "learning_rate": 0.00017438253652206985,
      "loss": 3.4094,
      "step": 137410
    },
    {
      "epoch": 0.12809767564287997,
      "grad_norm": 1.2951740026474,
      "learning_rate": 0.00017438067219997466,
      "loss": 3.517,
      "step": 137420
    },
    {
      "epoch": 0.12810699726095906,
      "grad_norm": 0.976463794708252,
      "learning_rate": 0.00017437880787787947,
      "loss": 3.5421,
      "step": 137430
    },
    {
      "epoch": 0.12811631887903813,
      "grad_norm": 0.9015424847602844,
      "learning_rate": 0.00017437694355578424,
      "loss": 3.6896,
      "step": 137440
    },
    {
      "epoch": 0.12812564049711722,
      "grad_norm": 1.0918426513671875,
      "learning_rate": 0.00017437507923368905,
      "loss": 3.5946,
      "step": 137450
    },
    {
      "epoch": 0.12813496211519632,
      "grad_norm": 0.9603579640388489,
      "learning_rate": 0.00017437321491159386,
      "loss": 3.5091,
      "step": 137460
    },
    {
      "epoch": 0.1281442837332754,
      "grad_norm": 1.086258053779602,
      "learning_rate": 0.00017437135058949864,
      "loss": 3.6635,
      "step": 137470
    },
    {
      "epoch": 0.1281536053513545,
      "grad_norm": 1.0275887250900269,
      "learning_rate": 0.00017436948626740347,
      "loss": 3.5611,
      "step": 137480
    },
    {
      "epoch": 0.1281629269694336,
      "grad_norm": 1.0790942907333374,
      "learning_rate": 0.00017436762194530825,
      "loss": 3.4853,
      "step": 137490
    },
    {
      "epoch": 0.1281722485875127,
      "grad_norm": 0.990916907787323,
      "learning_rate": 0.00017436575762321306,
      "loss": 3.4702,
      "step": 137500
    },
    {
      "epoch": 0.1281815702055918,
      "grad_norm": 1.0987732410430908,
      "learning_rate": 0.00017436389330111786,
      "loss": 3.5459,
      "step": 137510
    },
    {
      "epoch": 0.12819089182367088,
      "grad_norm": 1.047318458557129,
      "learning_rate": 0.00017436202897902264,
      "loss": 3.5919,
      "step": 137520
    },
    {
      "epoch": 0.12820021344174998,
      "grad_norm": 1.2145333290100098,
      "learning_rate": 0.00017436016465692745,
      "loss": 3.5069,
      "step": 137530
    },
    {
      "epoch": 0.12820953505982904,
      "grad_norm": 1.2280317544937134,
      "learning_rate": 0.00017435830033483225,
      "loss": 3.5157,
      "step": 137540
    },
    {
      "epoch": 0.12821885667790814,
      "grad_norm": 0.9827166795730591,
      "learning_rate": 0.00017435643601273706,
      "loss": 3.6219,
      "step": 137550
    },
    {
      "epoch": 0.12822817829598723,
      "grad_norm": 0.8758847117424011,
      "learning_rate": 0.00017435457169064187,
      "loss": 3.7167,
      "step": 137560
    },
    {
      "epoch": 0.12823749991406633,
      "grad_norm": 1.2073781490325928,
      "learning_rate": 0.00017435270736854665,
      "loss": 3.5845,
      "step": 137570
    },
    {
      "epoch": 0.12824682153214542,
      "grad_norm": 1.0609924793243408,
      "learning_rate": 0.00017435084304645145,
      "loss": 3.5206,
      "step": 137580
    },
    {
      "epoch": 0.12825614315022452,
      "grad_norm": 0.8895724415779114,
      "learning_rate": 0.00017434897872435626,
      "loss": 3.7057,
      "step": 137590
    },
    {
      "epoch": 0.1282654647683036,
      "grad_norm": 1.1454041004180908,
      "learning_rate": 0.00017434711440226106,
      "loss": 3.7384,
      "step": 137600
    },
    {
      "epoch": 0.1282747863863827,
      "grad_norm": 0.9834966063499451,
      "learning_rate": 0.00017434525008016587,
      "loss": 3.5985,
      "step": 137610
    },
    {
      "epoch": 0.1282841080044618,
      "grad_norm": 1.0060348510742188,
      "learning_rate": 0.00017434338575807065,
      "loss": 3.6447,
      "step": 137620
    },
    {
      "epoch": 0.1282934296225409,
      "grad_norm": 1.3666096925735474,
      "learning_rate": 0.00017434152143597546,
      "loss": 3.6246,
      "step": 137630
    },
    {
      "epoch": 0.12830275124062,
      "grad_norm": 0.9619407653808594,
      "learning_rate": 0.00017433965711388026,
      "loss": 3.7389,
      "step": 137640
    },
    {
      "epoch": 0.12831207285869906,
      "grad_norm": 1.2001750469207764,
      "learning_rate": 0.00017433779279178507,
      "loss": 3.6115,
      "step": 137650
    },
    {
      "epoch": 0.12832139447677815,
      "grad_norm": 0.7985817193984985,
      "learning_rate": 0.00017433592846968987,
      "loss": 3.7129,
      "step": 137660
    },
    {
      "epoch": 0.12833071609485724,
      "grad_norm": 1.0008156299591064,
      "learning_rate": 0.00017433406414759465,
      "loss": 3.7057,
      "step": 137670
    },
    {
      "epoch": 0.12834003771293634,
      "grad_norm": 0.8535372614860535,
      "learning_rate": 0.00017433219982549946,
      "loss": 3.6283,
      "step": 137680
    },
    {
      "epoch": 0.12834935933101543,
      "grad_norm": 1.1130671501159668,
      "learning_rate": 0.00017433033550340424,
      "loss": 3.6285,
      "step": 137690
    },
    {
      "epoch": 0.12835868094909453,
      "grad_norm": 1.3139317035675049,
      "learning_rate": 0.00017432847118130907,
      "loss": 3.5271,
      "step": 137700
    },
    {
      "epoch": 0.12836800256717362,
      "grad_norm": 1.1048270463943481,
      "learning_rate": 0.00017432660685921388,
      "loss": 3.4829,
      "step": 137710
    },
    {
      "epoch": 0.12837732418525272,
      "grad_norm": 1.1876795291900635,
      "learning_rate": 0.00017432474253711866,
      "loss": 3.8288,
      "step": 137720
    },
    {
      "epoch": 0.1283866458033318,
      "grad_norm": 0.9941903948783875,
      "learning_rate": 0.00017432287821502346,
      "loss": 3.6855,
      "step": 137730
    },
    {
      "epoch": 0.1283959674214109,
      "grad_norm": 1.0396230220794678,
      "learning_rate": 0.00017432101389292824,
      "loss": 3.755,
      "step": 137740
    },
    {
      "epoch": 0.12840528903948997,
      "grad_norm": 1.0992506742477417,
      "learning_rate": 0.00017431914957083305,
      "loss": 3.4706,
      "step": 137750
    },
    {
      "epoch": 0.12841461065756907,
      "grad_norm": 1.123295545578003,
      "learning_rate": 0.00017431728524873788,
      "loss": 3.662,
      "step": 137760
    },
    {
      "epoch": 0.12842393227564816,
      "grad_norm": 1.0754642486572266,
      "learning_rate": 0.00017431542092664266,
      "loss": 3.4805,
      "step": 137770
    },
    {
      "epoch": 0.12843325389372726,
      "grad_norm": 0.9628987312316895,
      "learning_rate": 0.00017431355660454747,
      "loss": 3.7845,
      "step": 137780
    },
    {
      "epoch": 0.12844257551180635,
      "grad_norm": 1.1658872365951538,
      "learning_rate": 0.00017431169228245225,
      "loss": 3.568,
      "step": 137790
    },
    {
      "epoch": 0.12845189712988545,
      "grad_norm": 1.0006259679794312,
      "learning_rate": 0.00017430982796035705,
      "loss": 3.558,
      "step": 137800
    },
    {
      "epoch": 0.12846121874796454,
      "grad_norm": 1.0425834655761719,
      "learning_rate": 0.00017430796363826186,
      "loss": 3.5899,
      "step": 137810
    },
    {
      "epoch": 0.12847054036604363,
      "grad_norm": 1.0634781122207642,
      "learning_rate": 0.00017430609931616667,
      "loss": 3.4806,
      "step": 137820
    },
    {
      "epoch": 0.12847986198412273,
      "grad_norm": 0.9292488098144531,
      "learning_rate": 0.00017430423499407147,
      "loss": 3.6166,
      "step": 137830
    },
    {
      "epoch": 0.12848918360220182,
      "grad_norm": 1.2108557224273682,
      "learning_rate": 0.00017430237067197625,
      "loss": 3.5179,
      "step": 137840
    },
    {
      "epoch": 0.12849850522028092,
      "grad_norm": 1.1212499141693115,
      "learning_rate": 0.00017430050634988106,
      "loss": 3.6364,
      "step": 137850
    },
    {
      "epoch": 0.12850782683835998,
      "grad_norm": 0.8467039465904236,
      "learning_rate": 0.00017429864202778587,
      "loss": 3.682,
      "step": 137860
    },
    {
      "epoch": 0.12851714845643908,
      "grad_norm": 1.0947970151901245,
      "learning_rate": 0.00017429677770569067,
      "loss": 3.6064,
      "step": 137870
    },
    {
      "epoch": 0.12852647007451817,
      "grad_norm": 0.7349905371665955,
      "learning_rate": 0.00017429491338359548,
      "loss": 3.4568,
      "step": 137880
    },
    {
      "epoch": 0.12853579169259727,
      "grad_norm": 1.2021141052246094,
      "learning_rate": 0.00017429304906150026,
      "loss": 3.5387,
      "step": 137890
    },
    {
      "epoch": 0.12854511331067636,
      "grad_norm": 1.1726970672607422,
      "learning_rate": 0.00017429118473940506,
      "loss": 3.7157,
      "step": 137900
    },
    {
      "epoch": 0.12855443492875546,
      "grad_norm": 0.9506930708885193,
      "learning_rate": 0.00017428932041730987,
      "loss": 3.7932,
      "step": 137910
    },
    {
      "epoch": 0.12856375654683455,
      "grad_norm": 1.0663115978240967,
      "learning_rate": 0.00017428745609521468,
      "loss": 3.9966,
      "step": 137920
    },
    {
      "epoch": 0.12857307816491365,
      "grad_norm": 0.9925169348716736,
      "learning_rate": 0.00017428559177311948,
      "loss": 3.534,
      "step": 137930
    },
    {
      "epoch": 0.12858239978299274,
      "grad_norm": 1.2307648658752441,
      "learning_rate": 0.00017428372745102426,
      "loss": 3.4466,
      "step": 137940
    },
    {
      "epoch": 0.12859172140107183,
      "grad_norm": 0.8628028035163879,
      "learning_rate": 0.00017428186312892907,
      "loss": 3.5158,
      "step": 137950
    },
    {
      "epoch": 0.1286010430191509,
      "grad_norm": 0.9510664343833923,
      "learning_rate": 0.00017427999880683387,
      "loss": 3.6219,
      "step": 137960
    },
    {
      "epoch": 0.12861036463723,
      "grad_norm": 0.8613534569740295,
      "learning_rate": 0.00017427813448473865,
      "loss": 3.639,
      "step": 137970
    },
    {
      "epoch": 0.1286196862553091,
      "grad_norm": 0.8771205544471741,
      "learning_rate": 0.0001742762701626435,
      "loss": 3.724,
      "step": 137980
    },
    {
      "epoch": 0.12862900787338818,
      "grad_norm": 1.2131056785583496,
      "learning_rate": 0.00017427440584054827,
      "loss": 3.6929,
      "step": 137990
    },
    {
      "epoch": 0.12863832949146728,
      "grad_norm": 1.04196298122406,
      "learning_rate": 0.00017427254151845307,
      "loss": 3.6501,
      "step": 138000
    },
    {
      "epoch": 0.12864765110954637,
      "grad_norm": 1.1456899642944336,
      "learning_rate": 0.00017427067719635788,
      "loss": 3.351,
      "step": 138010
    },
    {
      "epoch": 0.12865697272762547,
      "grad_norm": 1.0009351968765259,
      "learning_rate": 0.00017426881287426266,
      "loss": 3.688,
      "step": 138020
    },
    {
      "epoch": 0.12866629434570456,
      "grad_norm": 0.9931869506835938,
      "learning_rate": 0.00017426694855216746,
      "loss": 3.7443,
      "step": 138030
    },
    {
      "epoch": 0.12867561596378366,
      "grad_norm": 0.9766307473182678,
      "learning_rate": 0.00017426508423007227,
      "loss": 3.8447,
      "step": 138040
    },
    {
      "epoch": 0.12868493758186275,
      "grad_norm": 1.132267951965332,
      "learning_rate": 0.00017426321990797708,
      "loss": 3.528,
      "step": 138050
    },
    {
      "epoch": 0.12869425919994185,
      "grad_norm": 1.0940579175949097,
      "learning_rate": 0.00017426135558588188,
      "loss": 3.6521,
      "step": 138060
    },
    {
      "epoch": 0.1287035808180209,
      "grad_norm": 0.8464019894599915,
      "learning_rate": 0.00017425949126378666,
      "loss": 3.6371,
      "step": 138070
    },
    {
      "epoch": 0.1287129024361,
      "grad_norm": 0.862141489982605,
      "learning_rate": 0.00017425762694169147,
      "loss": 3.5351,
      "step": 138080
    },
    {
      "epoch": 0.1287222240541791,
      "grad_norm": 0.9941117763519287,
      "learning_rate": 0.00017425576261959628,
      "loss": 3.6521,
      "step": 138090
    },
    {
      "epoch": 0.1287315456722582,
      "grad_norm": 1.0312355756759644,
      "learning_rate": 0.00017425389829750108,
      "loss": 3.83,
      "step": 138100
    },
    {
      "epoch": 0.1287408672903373,
      "grad_norm": 0.8771920800209045,
      "learning_rate": 0.0001742520339754059,
      "loss": 3.553,
      "step": 138110
    },
    {
      "epoch": 0.12875018890841639,
      "grad_norm": 1.1165287494659424,
      "learning_rate": 0.00017425016965331067,
      "loss": 3.647,
      "step": 138120
    },
    {
      "epoch": 0.12875951052649548,
      "grad_norm": 1.4474451541900635,
      "learning_rate": 0.00017424830533121547,
      "loss": 3.5225,
      "step": 138130
    },
    {
      "epoch": 0.12876883214457457,
      "grad_norm": 1.0224461555480957,
      "learning_rate": 0.00017424644100912025,
      "loss": 3.6465,
      "step": 138140
    },
    {
      "epoch": 0.12877815376265367,
      "grad_norm": 0.9914897680282593,
      "learning_rate": 0.00017424457668702509,
      "loss": 3.5734,
      "step": 138150
    },
    {
      "epoch": 0.12878747538073276,
      "grad_norm": 1.08726167678833,
      "learning_rate": 0.0001742427123649299,
      "loss": 3.6607,
      "step": 138160
    },
    {
      "epoch": 0.12879679699881183,
      "grad_norm": 1.2055208683013916,
      "learning_rate": 0.00017424084804283467,
      "loss": 3.4836,
      "step": 138170
    },
    {
      "epoch": 0.12880611861689092,
      "grad_norm": 1.0343557596206665,
      "learning_rate": 0.00017423898372073948,
      "loss": 3.8242,
      "step": 138180
    },
    {
      "epoch": 0.12881544023497002,
      "grad_norm": 1.0841909646987915,
      "learning_rate": 0.00017423711939864426,
      "loss": 3.641,
      "step": 138190
    },
    {
      "epoch": 0.1288247618530491,
      "grad_norm": 0.9728406071662903,
      "learning_rate": 0.00017423525507654906,
      "loss": 3.7036,
      "step": 138200
    },
    {
      "epoch": 0.1288340834711282,
      "grad_norm": 1.122577428817749,
      "learning_rate": 0.0001742333907544539,
      "loss": 3.5929,
      "step": 138210
    },
    {
      "epoch": 0.1288434050892073,
      "grad_norm": 1.0684351921081543,
      "learning_rate": 0.00017423152643235868,
      "loss": 3.6726,
      "step": 138220
    },
    {
      "epoch": 0.1288527267072864,
      "grad_norm": 1.2023816108703613,
      "learning_rate": 0.00017422966211026348,
      "loss": 3.6388,
      "step": 138230
    },
    {
      "epoch": 0.1288620483253655,
      "grad_norm": 1.0446430444717407,
      "learning_rate": 0.00017422779778816826,
      "loss": 3.6343,
      "step": 138240
    },
    {
      "epoch": 0.12887136994344459,
      "grad_norm": 1.2505515813827515,
      "learning_rate": 0.00017422593346607307,
      "loss": 3.5273,
      "step": 138250
    },
    {
      "epoch": 0.12888069156152368,
      "grad_norm": 1.0178186893463135,
      "learning_rate": 0.00017422406914397787,
      "loss": 3.6065,
      "step": 138260
    },
    {
      "epoch": 0.12889001317960277,
      "grad_norm": 1.0533053874969482,
      "learning_rate": 0.00017422220482188268,
      "loss": 3.6711,
      "step": 138270
    },
    {
      "epoch": 0.12889933479768184,
      "grad_norm": 0.8233303427696228,
      "learning_rate": 0.0001742203404997875,
      "loss": 3.5935,
      "step": 138280
    },
    {
      "epoch": 0.12890865641576094,
      "grad_norm": 0.7837879061698914,
      "learning_rate": 0.00017421847617769227,
      "loss": 3.9067,
      "step": 138290
    },
    {
      "epoch": 0.12891797803384003,
      "grad_norm": 1.2599514722824097,
      "learning_rate": 0.00017421661185559707,
      "loss": 3.4348,
      "step": 138300
    },
    {
      "epoch": 0.12892729965191913,
      "grad_norm": 1.0868339538574219,
      "learning_rate": 0.00017421474753350188,
      "loss": 3.4583,
      "step": 138310
    },
    {
      "epoch": 0.12893662126999822,
      "grad_norm": 1.2233972549438477,
      "learning_rate": 0.00017421288321140668,
      "loss": 3.6187,
      "step": 138320
    },
    {
      "epoch": 0.12894594288807731,
      "grad_norm": 0.9506231546401978,
      "learning_rate": 0.0001742110188893115,
      "loss": 3.519,
      "step": 138330
    },
    {
      "epoch": 0.1289552645061564,
      "grad_norm": 1.13870370388031,
      "learning_rate": 0.00017420915456721627,
      "loss": 3.6895,
      "step": 138340
    },
    {
      "epoch": 0.1289645861242355,
      "grad_norm": 1.0011214017868042,
      "learning_rate": 0.00017420729024512108,
      "loss": 3.7297,
      "step": 138350
    },
    {
      "epoch": 0.1289739077423146,
      "grad_norm": 1.031735897064209,
      "learning_rate": 0.00017420542592302586,
      "loss": 3.7481,
      "step": 138360
    },
    {
      "epoch": 0.1289832293603937,
      "grad_norm": 1.2942628860473633,
      "learning_rate": 0.0001742035616009307,
      "loss": 3.4997,
      "step": 138370
    },
    {
      "epoch": 0.12899255097847276,
      "grad_norm": 0.9142616391181946,
      "learning_rate": 0.0001742016972788355,
      "loss": 3.8424,
      "step": 138380
    },
    {
      "epoch": 0.12900187259655185,
      "grad_norm": 1.1778286695480347,
      "learning_rate": 0.00017419983295674027,
      "loss": 3.6144,
      "step": 138390
    },
    {
      "epoch": 0.12901119421463095,
      "grad_norm": 0.894000768661499,
      "learning_rate": 0.00017419796863464508,
      "loss": 3.8063,
      "step": 138400
    },
    {
      "epoch": 0.12902051583271004,
      "grad_norm": 1.0274567604064941,
      "learning_rate": 0.00017419610431254986,
      "loss": 3.5672,
      "step": 138410
    },
    {
      "epoch": 0.12902983745078914,
      "grad_norm": 1.0196739435195923,
      "learning_rate": 0.00017419423999045467,
      "loss": 3.6377,
      "step": 138420
    },
    {
      "epoch": 0.12903915906886823,
      "grad_norm": 1.1837608814239502,
      "learning_rate": 0.0001741923756683595,
      "loss": 3.6222,
      "step": 138430
    },
    {
      "epoch": 0.12904848068694733,
      "grad_norm": 0.9182509183883667,
      "learning_rate": 0.00017419051134626428,
      "loss": 3.7247,
      "step": 138440
    },
    {
      "epoch": 0.12905780230502642,
      "grad_norm": 1.0743334293365479,
      "learning_rate": 0.00017418864702416909,
      "loss": 3.5176,
      "step": 138450
    },
    {
      "epoch": 0.12906712392310551,
      "grad_norm": 1.1092166900634766,
      "learning_rate": 0.00017418678270207386,
      "loss": 3.8411,
      "step": 138460
    },
    {
      "epoch": 0.1290764455411846,
      "grad_norm": 1.1682240962982178,
      "learning_rate": 0.00017418491837997867,
      "loss": 3.5976,
      "step": 138470
    },
    {
      "epoch": 0.1290857671592637,
      "grad_norm": 1.0349403619766235,
      "learning_rate": 0.00017418305405788348,
      "loss": 3.7287,
      "step": 138480
    },
    {
      "epoch": 0.12909508877734277,
      "grad_norm": 1.0500750541687012,
      "learning_rate": 0.00017418118973578828,
      "loss": 3.7368,
      "step": 138490
    },
    {
      "epoch": 0.12910441039542186,
      "grad_norm": 0.9702285528182983,
      "learning_rate": 0.0001741793254136931,
      "loss": 3.7028,
      "step": 138500
    },
    {
      "epoch": 0.12911373201350096,
      "grad_norm": 0.9516792297363281,
      "learning_rate": 0.00017417746109159787,
      "loss": 3.7388,
      "step": 138510
    },
    {
      "epoch": 0.12912305363158005,
      "grad_norm": 1.016809105873108,
      "learning_rate": 0.00017417559676950268,
      "loss": 3.6999,
      "step": 138520
    },
    {
      "epoch": 0.12913237524965915,
      "grad_norm": 1.0666102170944214,
      "learning_rate": 0.00017417373244740748,
      "loss": 3.7464,
      "step": 138530
    },
    {
      "epoch": 0.12914169686773824,
      "grad_norm": 1.0322648286819458,
      "learning_rate": 0.0001741718681253123,
      "loss": 3.6905,
      "step": 138540
    },
    {
      "epoch": 0.12915101848581734,
      "grad_norm": 1.1407155990600586,
      "learning_rate": 0.0001741700038032171,
      "loss": 3.7606,
      "step": 138550
    },
    {
      "epoch": 0.12916034010389643,
      "grad_norm": 0.9199822545051575,
      "learning_rate": 0.00017416813948112187,
      "loss": 3.5196,
      "step": 138560
    },
    {
      "epoch": 0.12916966172197553,
      "grad_norm": 1.0183939933776855,
      "learning_rate": 0.00017416627515902668,
      "loss": 3.8166,
      "step": 138570
    },
    {
      "epoch": 0.12917898334005462,
      "grad_norm": 1.0764085054397583,
      "learning_rate": 0.00017416441083693149,
      "loss": 3.5844,
      "step": 138580
    },
    {
      "epoch": 0.1291883049581337,
      "grad_norm": 1.0111336708068848,
      "learning_rate": 0.00017416254651483627,
      "loss": 3.3256,
      "step": 138590
    },
    {
      "epoch": 0.12919762657621278,
      "grad_norm": 1.2023160457611084,
      "learning_rate": 0.0001741606821927411,
      "loss": 3.5316,
      "step": 138600
    },
    {
      "epoch": 0.12920694819429188,
      "grad_norm": 1.1369601488113403,
      "learning_rate": 0.00017415881787064588,
      "loss": 3.8002,
      "step": 138610
    },
    {
      "epoch": 0.12921626981237097,
      "grad_norm": 1.0917505025863647,
      "learning_rate": 0.00017415695354855068,
      "loss": 3.6964,
      "step": 138620
    },
    {
      "epoch": 0.12922559143045007,
      "grad_norm": 1.5445951223373413,
      "learning_rate": 0.0001741550892264555,
      "loss": 3.5306,
      "step": 138630
    },
    {
      "epoch": 0.12923491304852916,
      "grad_norm": 1.2655285596847534,
      "learning_rate": 0.00017415322490436027,
      "loss": 3.663,
      "step": 138640
    },
    {
      "epoch": 0.12924423466660825,
      "grad_norm": 0.9598127007484436,
      "learning_rate": 0.00017415136058226508,
      "loss": 3.7108,
      "step": 138650
    },
    {
      "epoch": 0.12925355628468735,
      "grad_norm": 0.9525511860847473,
      "learning_rate": 0.00017414949626016988,
      "loss": 3.7178,
      "step": 138660
    },
    {
      "epoch": 0.12926287790276644,
      "grad_norm": 1.1461725234985352,
      "learning_rate": 0.0001741476319380747,
      "loss": 3.5849,
      "step": 138670
    },
    {
      "epoch": 0.12927219952084554,
      "grad_norm": 1.0271836519241333,
      "learning_rate": 0.0001741457676159795,
      "loss": 3.608,
      "step": 138680
    },
    {
      "epoch": 0.12928152113892463,
      "grad_norm": 1.1256470680236816,
      "learning_rate": 0.00017414390329388427,
      "loss": 3.3988,
      "step": 138690
    },
    {
      "epoch": 0.1292908427570037,
      "grad_norm": 1.224411129951477,
      "learning_rate": 0.00017414203897178908,
      "loss": 3.5379,
      "step": 138700
    },
    {
      "epoch": 0.1293001643750828,
      "grad_norm": 1.1593286991119385,
      "learning_rate": 0.0001741401746496939,
      "loss": 3.6344,
      "step": 138710
    },
    {
      "epoch": 0.1293094859931619,
      "grad_norm": 0.9915531873703003,
      "learning_rate": 0.0001741383103275987,
      "loss": 3.4596,
      "step": 138720
    },
    {
      "epoch": 0.12931880761124098,
      "grad_norm": 1.068571925163269,
      "learning_rate": 0.0001741364460055035,
      "loss": 3.7092,
      "step": 138730
    },
    {
      "epoch": 0.12932812922932008,
      "grad_norm": 1.0425658226013184,
      "learning_rate": 0.00017413458168340828,
      "loss": 3.736,
      "step": 138740
    },
    {
      "epoch": 0.12933745084739917,
      "grad_norm": 1.234809398651123,
      "learning_rate": 0.00017413271736131308,
      "loss": 3.6718,
      "step": 138750
    },
    {
      "epoch": 0.12934677246547827,
      "grad_norm": 1.0110670328140259,
      "learning_rate": 0.0001741308530392179,
      "loss": 3.571,
      "step": 138760
    },
    {
      "epoch": 0.12935609408355736,
      "grad_norm": 1.218756079673767,
      "learning_rate": 0.0001741289887171227,
      "loss": 3.6293,
      "step": 138770
    },
    {
      "epoch": 0.12936541570163645,
      "grad_norm": 0.9921412467956543,
      "learning_rate": 0.0001741271243950275,
      "loss": 3.7668,
      "step": 138780
    },
    {
      "epoch": 0.12937473731971555,
      "grad_norm": 1.3052124977111816,
      "learning_rate": 0.00017412526007293228,
      "loss": 3.5884,
      "step": 138790
    },
    {
      "epoch": 0.12938405893779462,
      "grad_norm": 1.1131598949432373,
      "learning_rate": 0.0001741233957508371,
      "loss": 3.7232,
      "step": 138800
    },
    {
      "epoch": 0.1293933805558737,
      "grad_norm": 0.8899785280227661,
      "learning_rate": 0.00017412153142874187,
      "loss": 3.7041,
      "step": 138810
    },
    {
      "epoch": 0.1294027021739528,
      "grad_norm": 1.1974611282348633,
      "learning_rate": 0.0001741196671066467,
      "loss": 3.8326,
      "step": 138820
    },
    {
      "epoch": 0.1294120237920319,
      "grad_norm": 1.0389786958694458,
      "learning_rate": 0.0001741178027845515,
      "loss": 3.6907,
      "step": 138830
    },
    {
      "epoch": 0.129421345410111,
      "grad_norm": 0.8399770855903625,
      "learning_rate": 0.0001741159384624563,
      "loss": 3.7205,
      "step": 138840
    },
    {
      "epoch": 0.1294306670281901,
      "grad_norm": 1.076077938079834,
      "learning_rate": 0.0001741140741403611,
      "loss": 3.5058,
      "step": 138850
    },
    {
      "epoch": 0.12943998864626918,
      "grad_norm": 0.9910942912101746,
      "learning_rate": 0.00017411220981826587,
      "loss": 3.5978,
      "step": 138860
    },
    {
      "epoch": 0.12944931026434828,
      "grad_norm": 1.0018270015716553,
      "learning_rate": 0.00017411034549617068,
      "loss": 3.5461,
      "step": 138870
    },
    {
      "epoch": 0.12945863188242737,
      "grad_norm": 1.054487705230713,
      "learning_rate": 0.0001741084811740755,
      "loss": 3.5092,
      "step": 138880
    },
    {
      "epoch": 0.12946795350050647,
      "grad_norm": 1.213483452796936,
      "learning_rate": 0.0001741066168519803,
      "loss": 3.5571,
      "step": 138890
    },
    {
      "epoch": 0.12947727511858556,
      "grad_norm": 1.0195698738098145,
      "learning_rate": 0.0001741047525298851,
      "loss": 3.7685,
      "step": 138900
    },
    {
      "epoch": 0.12948659673666463,
      "grad_norm": 1.0391411781311035,
      "learning_rate": 0.00017410288820778988,
      "loss": 3.6987,
      "step": 138910
    },
    {
      "epoch": 0.12949591835474372,
      "grad_norm": 1.1364387273788452,
      "learning_rate": 0.00017410102388569468,
      "loss": 3.5808,
      "step": 138920
    },
    {
      "epoch": 0.12950523997282282,
      "grad_norm": 1.2708367109298706,
      "learning_rate": 0.0001740991595635995,
      "loss": 3.6263,
      "step": 138930
    },
    {
      "epoch": 0.1295145615909019,
      "grad_norm": 1.0733970403671265,
      "learning_rate": 0.0001740972952415043,
      "loss": 3.5695,
      "step": 138940
    },
    {
      "epoch": 0.129523883208981,
      "grad_norm": 1.1977643966674805,
      "learning_rate": 0.0001740954309194091,
      "loss": 3.9576,
      "step": 138950
    },
    {
      "epoch": 0.1295332048270601,
      "grad_norm": 0.9999497532844543,
      "learning_rate": 0.00017409356659731388,
      "loss": 3.4553,
      "step": 138960
    },
    {
      "epoch": 0.1295425264451392,
      "grad_norm": 0.8451176285743713,
      "learning_rate": 0.0001740917022752187,
      "loss": 3.732,
      "step": 138970
    },
    {
      "epoch": 0.1295518480632183,
      "grad_norm": 0.9624674320220947,
      "learning_rate": 0.0001740898379531235,
      "loss": 3.5192,
      "step": 138980
    },
    {
      "epoch": 0.12956116968129738,
      "grad_norm": 0.9473581910133362,
      "learning_rate": 0.0001740879736310283,
      "loss": 3.5641,
      "step": 138990
    },
    {
      "epoch": 0.12957049129937648,
      "grad_norm": 0.9973599910736084,
      "learning_rate": 0.0001740861093089331,
      "loss": 3.5719,
      "step": 139000
    },
    {
      "epoch": 0.12957981291745554,
      "grad_norm": 1.0115176439285278,
      "learning_rate": 0.00017408424498683789,
      "loss": 3.4442,
      "step": 139010
    },
    {
      "epoch": 0.12958913453553464,
      "grad_norm": 0.9766863584518433,
      "learning_rate": 0.0001740823806647427,
      "loss": 3.5698,
      "step": 139020
    },
    {
      "epoch": 0.12959845615361373,
      "grad_norm": 1.1292712688446045,
      "learning_rate": 0.0001740805163426475,
      "loss": 3.5626,
      "step": 139030
    },
    {
      "epoch": 0.12960777777169283,
      "grad_norm": 1.0115509033203125,
      "learning_rate": 0.00017407865202055228,
      "loss": 3.8093,
      "step": 139040
    },
    {
      "epoch": 0.12961709938977192,
      "grad_norm": 1.2270089387893677,
      "learning_rate": 0.0001740767876984571,
      "loss": 3.6813,
      "step": 139050
    },
    {
      "epoch": 0.12962642100785102,
      "grad_norm": 1.0249024629592896,
      "learning_rate": 0.0001740749233763619,
      "loss": 3.7272,
      "step": 139060
    },
    {
      "epoch": 0.1296357426259301,
      "grad_norm": 0.9227113723754883,
      "learning_rate": 0.0001740730590542667,
      "loss": 3.7111,
      "step": 139070
    },
    {
      "epoch": 0.1296450642440092,
      "grad_norm": 0.9693912267684937,
      "learning_rate": 0.0001740711947321715,
      "loss": 3.4048,
      "step": 139080
    },
    {
      "epoch": 0.1296543858620883,
      "grad_norm": 0.9359678626060486,
      "learning_rate": 0.00017406933041007628,
      "loss": 3.805,
      "step": 139090
    },
    {
      "epoch": 0.1296637074801674,
      "grad_norm": 1.0815908908843994,
      "learning_rate": 0.0001740674660879811,
      "loss": 3.622,
      "step": 139100
    },
    {
      "epoch": 0.12967302909824646,
      "grad_norm": 1.0203911066055298,
      "learning_rate": 0.0001740656017658859,
      "loss": 3.6918,
      "step": 139110
    },
    {
      "epoch": 0.12968235071632556,
      "grad_norm": 1.181492567062378,
      "learning_rate": 0.0001740637374437907,
      "loss": 3.5372,
      "step": 139120
    },
    {
      "epoch": 0.12969167233440465,
      "grad_norm": 0.9908980131149292,
      "learning_rate": 0.0001740618731216955,
      "loss": 3.7067,
      "step": 139130
    },
    {
      "epoch": 0.12970099395248375,
      "grad_norm": 0.999086856842041,
      "learning_rate": 0.0001740600087996003,
      "loss": 3.6652,
      "step": 139140
    },
    {
      "epoch": 0.12971031557056284,
      "grad_norm": 0.9622434973716736,
      "learning_rate": 0.0001740581444775051,
      "loss": 3.6541,
      "step": 139150
    },
    {
      "epoch": 0.12971963718864193,
      "grad_norm": 1.1011337041854858,
      "learning_rate": 0.0001740562801554099,
      "loss": 3.5719,
      "step": 139160
    },
    {
      "epoch": 0.12972895880672103,
      "grad_norm": 1.4315322637557983,
      "learning_rate": 0.0001740544158333147,
      "loss": 3.6412,
      "step": 139170
    },
    {
      "epoch": 0.12973828042480012,
      "grad_norm": 0.9867069125175476,
      "learning_rate": 0.0001740525515112195,
      "loss": 3.7011,
      "step": 139180
    },
    {
      "epoch": 0.12974760204287922,
      "grad_norm": 1.086217999458313,
      "learning_rate": 0.0001740506871891243,
      "loss": 3.7523,
      "step": 139190
    },
    {
      "epoch": 0.1297569236609583,
      "grad_norm": 1.0484960079193115,
      "learning_rate": 0.0001740488228670291,
      "loss": 3.6193,
      "step": 139200
    },
    {
      "epoch": 0.1297662452790374,
      "grad_norm": 1.1103696823120117,
      "learning_rate": 0.0001740469585449339,
      "loss": 3.5881,
      "step": 139210
    },
    {
      "epoch": 0.12977556689711647,
      "grad_norm": 0.9623649716377258,
      "learning_rate": 0.0001740450942228387,
      "loss": 3.5841,
      "step": 139220
    },
    {
      "epoch": 0.12978488851519557,
      "grad_norm": 1.1875152587890625,
      "learning_rate": 0.00017404322990074352,
      "loss": 3.4116,
      "step": 139230
    },
    {
      "epoch": 0.12979421013327466,
      "grad_norm": 1.0061161518096924,
      "learning_rate": 0.0001740413655786483,
      "loss": 3.6447,
      "step": 139240
    },
    {
      "epoch": 0.12980353175135376,
      "grad_norm": 0.9484146237373352,
      "learning_rate": 0.0001740395012565531,
      "loss": 3.6123,
      "step": 139250
    },
    {
      "epoch": 0.12981285336943285,
      "grad_norm": 1.0911234617233276,
      "learning_rate": 0.00017403763693445788,
      "loss": 3.5402,
      "step": 139260
    },
    {
      "epoch": 0.12982217498751195,
      "grad_norm": 1.1343863010406494,
      "learning_rate": 0.00017403577261236271,
      "loss": 3.6885,
      "step": 139270
    },
    {
      "epoch": 0.12983149660559104,
      "grad_norm": 0.9923742413520813,
      "learning_rate": 0.00017403390829026752,
      "loss": 3.6519,
      "step": 139280
    },
    {
      "epoch": 0.12984081822367013,
      "grad_norm": 1.1040842533111572,
      "learning_rate": 0.0001740320439681723,
      "loss": 3.6209,
      "step": 139290
    },
    {
      "epoch": 0.12985013984174923,
      "grad_norm": 1.5313429832458496,
      "learning_rate": 0.0001740301796460771,
      "loss": 3.485,
      "step": 139300
    },
    {
      "epoch": 0.12985946145982832,
      "grad_norm": 0.919241189956665,
      "learning_rate": 0.00017402831532398189,
      "loss": 3.885,
      "step": 139310
    },
    {
      "epoch": 0.1298687830779074,
      "grad_norm": 1.0670942068099976,
      "learning_rate": 0.0001740264510018867,
      "loss": 3.6555,
      "step": 139320
    },
    {
      "epoch": 0.12987810469598648,
      "grad_norm": 1.0112987756729126,
      "learning_rate": 0.00017402458667979153,
      "loss": 3.7633,
      "step": 139330
    },
    {
      "epoch": 0.12988742631406558,
      "grad_norm": 1.0355018377304077,
      "learning_rate": 0.0001740227223576963,
      "loss": 3.5164,
      "step": 139340
    },
    {
      "epoch": 0.12989674793214467,
      "grad_norm": 0.8544875383377075,
      "learning_rate": 0.0001740208580356011,
      "loss": 3.6784,
      "step": 139350
    },
    {
      "epoch": 0.12990606955022377,
      "grad_norm": 0.9500458836555481,
      "learning_rate": 0.0001740189937135059,
      "loss": 3.6194,
      "step": 139360
    },
    {
      "epoch": 0.12991539116830286,
      "grad_norm": 1.1374369859695435,
      "learning_rate": 0.0001740171293914107,
      "loss": 3.8093,
      "step": 139370
    },
    {
      "epoch": 0.12992471278638196,
      "grad_norm": 0.9287722706794739,
      "learning_rate": 0.0001740152650693155,
      "loss": 3.7392,
      "step": 139380
    },
    {
      "epoch": 0.12993403440446105,
      "grad_norm": 1.0740008354187012,
      "learning_rate": 0.0001740134007472203,
      "loss": 3.702,
      "step": 139390
    },
    {
      "epoch": 0.12994335602254015,
      "grad_norm": 0.9486398696899414,
      "learning_rate": 0.00017401153642512512,
      "loss": 3.6375,
      "step": 139400
    },
    {
      "epoch": 0.12995267764061924,
      "grad_norm": 1.1699873208999634,
      "learning_rate": 0.0001740096721030299,
      "loss": 3.7026,
      "step": 139410
    },
    {
      "epoch": 0.12996199925869834,
      "grad_norm": 0.8569661378860474,
      "learning_rate": 0.0001740078077809347,
      "loss": 3.7658,
      "step": 139420
    },
    {
      "epoch": 0.1299713208767774,
      "grad_norm": 1.0196480751037598,
      "learning_rate": 0.0001740059434588395,
      "loss": 3.4736,
      "step": 139430
    },
    {
      "epoch": 0.1299806424948565,
      "grad_norm": 1.1675708293914795,
      "learning_rate": 0.00017400407913674431,
      "loss": 3.576,
      "step": 139440
    },
    {
      "epoch": 0.1299899641129356,
      "grad_norm": 1.1017647981643677,
      "learning_rate": 0.00017400221481464912,
      "loss": 3.4542,
      "step": 139450
    },
    {
      "epoch": 0.12999928573101469,
      "grad_norm": 1.0349839925765991,
      "learning_rate": 0.0001740003504925539,
      "loss": 3.5647,
      "step": 139460
    },
    {
      "epoch": 0.13000860734909378,
      "grad_norm": 0.9351635575294495,
      "learning_rate": 0.0001739984861704587,
      "loss": 3.6411,
      "step": 139470
    },
    {
      "epoch": 0.13001792896717287,
      "grad_norm": 1.0950281620025635,
      "learning_rate": 0.0001739966218483635,
      "loss": 3.6854,
      "step": 139480
    },
    {
      "epoch": 0.13002725058525197,
      "grad_norm": 1.0981038808822632,
      "learning_rate": 0.0001739947575262683,
      "loss": 3.6427,
      "step": 139490
    },
    {
      "epoch": 0.13003657220333106,
      "grad_norm": 1.0790188312530518,
      "learning_rate": 0.00017399289320417312,
      "loss": 3.6365,
      "step": 139500
    },
    {
      "epoch": 0.13004589382141016,
      "grad_norm": 1.235105037689209,
      "learning_rate": 0.0001739910288820779,
      "loss": 3.6277,
      "step": 139510
    },
    {
      "epoch": 0.13005521543948925,
      "grad_norm": 1.3750934600830078,
      "learning_rate": 0.0001739891645599827,
      "loss": 3.7898,
      "step": 139520
    },
    {
      "epoch": 0.13006453705756832,
      "grad_norm": 0.9313225150108337,
      "learning_rate": 0.00017398730023788752,
      "loss": 3.6781,
      "step": 139530
    },
    {
      "epoch": 0.1300738586756474,
      "grad_norm": 1.182573676109314,
      "learning_rate": 0.0001739854359157923,
      "loss": 3.6152,
      "step": 139540
    },
    {
      "epoch": 0.1300831802937265,
      "grad_norm": 0.7987726330757141,
      "learning_rate": 0.0001739835715936971,
      "loss": 3.7413,
      "step": 139550
    },
    {
      "epoch": 0.1300925019118056,
      "grad_norm": 1.1023656129837036,
      "learning_rate": 0.0001739817072716019,
      "loss": 3.7264,
      "step": 139560
    },
    {
      "epoch": 0.1301018235298847,
      "grad_norm": 0.9710120558738708,
      "learning_rate": 0.00017397984294950671,
      "loss": 3.6502,
      "step": 139570
    },
    {
      "epoch": 0.1301111451479638,
      "grad_norm": 1.1271870136260986,
      "learning_rate": 0.00017397797862741152,
      "loss": 3.5552,
      "step": 139580
    },
    {
      "epoch": 0.13012046676604289,
      "grad_norm": 1.1117100715637207,
      "learning_rate": 0.0001739761143053163,
      "loss": 3.4532,
      "step": 139590
    },
    {
      "epoch": 0.13012978838412198,
      "grad_norm": 0.9149348139762878,
      "learning_rate": 0.0001739742499832211,
      "loss": 3.7742,
      "step": 139600
    },
    {
      "epoch": 0.13013911000220107,
      "grad_norm": 1.088826298713684,
      "learning_rate": 0.0001739723856611259,
      "loss": 3.6108,
      "step": 139610
    },
    {
      "epoch": 0.13014843162028017,
      "grad_norm": 0.9440306425094604,
      "learning_rate": 0.00017397052133903072,
      "loss": 3.5672,
      "step": 139620
    },
    {
      "epoch": 0.13015775323835926,
      "grad_norm": 0.9262842535972595,
      "learning_rate": 0.00017396865701693553,
      "loss": 3.6888,
      "step": 139630
    },
    {
      "epoch": 0.13016707485643833,
      "grad_norm": 1.0511454343795776,
      "learning_rate": 0.0001739667926948403,
      "loss": 3.761,
      "step": 139640
    },
    {
      "epoch": 0.13017639647451743,
      "grad_norm": 0.8906614780426025,
      "learning_rate": 0.0001739649283727451,
      "loss": 3.5214,
      "step": 139650
    },
    {
      "epoch": 0.13018571809259652,
      "grad_norm": 1.0703580379486084,
      "learning_rate": 0.00017396306405064992,
      "loss": 3.6813,
      "step": 139660
    },
    {
      "epoch": 0.13019503971067561,
      "grad_norm": 1.0901762247085571,
      "learning_rate": 0.00017396119972855472,
      "loss": 3.7351,
      "step": 139670
    },
    {
      "epoch": 0.1302043613287547,
      "grad_norm": 0.9990274906158447,
      "learning_rate": 0.0001739593354064595,
      "loss": 3.6127,
      "step": 139680
    },
    {
      "epoch": 0.1302136829468338,
      "grad_norm": 1.0109539031982422,
      "learning_rate": 0.0001739574710843643,
      "loss": 3.5827,
      "step": 139690
    },
    {
      "epoch": 0.1302230045649129,
      "grad_norm": 0.85316401720047,
      "learning_rate": 0.00017395560676226912,
      "loss": 3.5715,
      "step": 139700
    },
    {
      "epoch": 0.130232326182992,
      "grad_norm": 1.2801234722137451,
      "learning_rate": 0.0001739537424401739,
      "loss": 3.5369,
      "step": 139710
    },
    {
      "epoch": 0.1302416478010711,
      "grad_norm": 1.0578385591506958,
      "learning_rate": 0.00017395187811807873,
      "loss": 3.4768,
      "step": 139720
    },
    {
      "epoch": 0.13025096941915018,
      "grad_norm": 1.0406080484390259,
      "learning_rate": 0.0001739500137959835,
      "loss": 3.5461,
      "step": 139730
    },
    {
      "epoch": 0.13026029103722925,
      "grad_norm": 1.1849963665008545,
      "learning_rate": 0.0001739481494738883,
      "loss": 3.6084,
      "step": 139740
    },
    {
      "epoch": 0.13026961265530834,
      "grad_norm": 1.143341302871704,
      "learning_rate": 0.00017394628515179312,
      "loss": 3.6334,
      "step": 139750
    },
    {
      "epoch": 0.13027893427338744,
      "grad_norm": 1.1014149188995361,
      "learning_rate": 0.0001739444208296979,
      "loss": 3.7035,
      "step": 139760
    },
    {
      "epoch": 0.13028825589146653,
      "grad_norm": 1.119918704032898,
      "learning_rate": 0.0001739425565076027,
      "loss": 3.587,
      "step": 139770
    },
    {
      "epoch": 0.13029757750954563,
      "grad_norm": 1.1237472295761108,
      "learning_rate": 0.0001739406921855075,
      "loss": 3.769,
      "step": 139780
    },
    {
      "epoch": 0.13030689912762472,
      "grad_norm": 1.2036947011947632,
      "learning_rate": 0.00017393882786341232,
      "loss": 3.5542,
      "step": 139790
    },
    {
      "epoch": 0.13031622074570381,
      "grad_norm": 1.1497013568878174,
      "learning_rate": 0.00017393696354131712,
      "loss": 3.6284,
      "step": 139800
    },
    {
      "epoch": 0.1303255423637829,
      "grad_norm": 1.148093581199646,
      "learning_rate": 0.0001739350992192219,
      "loss": 3.5278,
      "step": 139810
    },
    {
      "epoch": 0.130334863981862,
      "grad_norm": 1.2949477434158325,
      "learning_rate": 0.0001739332348971267,
      "loss": 3.3708,
      "step": 139820
    },
    {
      "epoch": 0.1303441855999411,
      "grad_norm": 1.18734610080719,
      "learning_rate": 0.00017393137057503152,
      "loss": 3.5284,
      "step": 139830
    },
    {
      "epoch": 0.1303535072180202,
      "grad_norm": 1.1244990825653076,
      "learning_rate": 0.00017392950625293632,
      "loss": 3.5384,
      "step": 139840
    },
    {
      "epoch": 0.13036282883609926,
      "grad_norm": 1.1622613668441772,
      "learning_rate": 0.00017392764193084113,
      "loss": 3.5984,
      "step": 139850
    },
    {
      "epoch": 0.13037215045417835,
      "grad_norm": 0.7882947325706482,
      "learning_rate": 0.0001739257776087459,
      "loss": 3.6842,
      "step": 139860
    },
    {
      "epoch": 0.13038147207225745,
      "grad_norm": 1.2007712125778198,
      "learning_rate": 0.00017392391328665071,
      "loss": 3.7463,
      "step": 139870
    },
    {
      "epoch": 0.13039079369033654,
      "grad_norm": 1.024770975112915,
      "learning_rate": 0.0001739220489645555,
      "loss": 3.7697,
      "step": 139880
    },
    {
      "epoch": 0.13040011530841564,
      "grad_norm": 1.152961015701294,
      "learning_rate": 0.00017392018464246033,
      "loss": 3.6183,
      "step": 139890
    },
    {
      "epoch": 0.13040943692649473,
      "grad_norm": 1.0025793313980103,
      "learning_rate": 0.00017391832032036513,
      "loss": 3.5247,
      "step": 139900
    },
    {
      "epoch": 0.13041875854457383,
      "grad_norm": 1.075016736984253,
      "learning_rate": 0.0001739164559982699,
      "loss": 3.3659,
      "step": 139910
    },
    {
      "epoch": 0.13042808016265292,
      "grad_norm": 1.05882728099823,
      "learning_rate": 0.00017391459167617472,
      "loss": 3.5528,
      "step": 139920
    },
    {
      "epoch": 0.13043740178073202,
      "grad_norm": 1.1214131116867065,
      "learning_rate": 0.0001739127273540795,
      "loss": 3.6367,
      "step": 139930
    },
    {
      "epoch": 0.1304467233988111,
      "grad_norm": 0.9922763109207153,
      "learning_rate": 0.0001739108630319843,
      "loss": 3.8602,
      "step": 139940
    },
    {
      "epoch": 0.13045604501689018,
      "grad_norm": 1.3437081575393677,
      "learning_rate": 0.00017390899870988914,
      "loss": 3.6572,
      "step": 139950
    },
    {
      "epoch": 0.13046536663496927,
      "grad_norm": 1.0689176321029663,
      "learning_rate": 0.00017390713438779392,
      "loss": 3.5892,
      "step": 139960
    },
    {
      "epoch": 0.13047468825304837,
      "grad_norm": 1.0610946416854858,
      "learning_rate": 0.00017390527006569872,
      "loss": 3.7673,
      "step": 139970
    },
    {
      "epoch": 0.13048400987112746,
      "grad_norm": 1.0417237281799316,
      "learning_rate": 0.0001739034057436035,
      "loss": 3.9099,
      "step": 139980
    },
    {
      "epoch": 0.13049333148920655,
      "grad_norm": 0.8887778520584106,
      "learning_rate": 0.0001739015414215083,
      "loss": 3.4259,
      "step": 139990
    },
    {
      "epoch": 0.13050265310728565,
      "grad_norm": 1.2045735120773315,
      "learning_rate": 0.00017389967709941311,
      "loss": 3.6165,
      "step": 140000
    },
    {
      "epoch": 0.13051197472536474,
      "grad_norm": 0.9766399264335632,
      "learning_rate": 0.00017389781277731792,
      "loss": 3.691,
      "step": 140010
    },
    {
      "epoch": 0.13052129634344384,
      "grad_norm": 0.961081326007843,
      "learning_rate": 0.00017389594845522273,
      "loss": 3.5577,
      "step": 140020
    },
    {
      "epoch": 0.13053061796152293,
      "grad_norm": 1.1226410865783691,
      "learning_rate": 0.0001738940841331275,
      "loss": 3.6106,
      "step": 140030
    },
    {
      "epoch": 0.13053993957960203,
      "grad_norm": 1.1144702434539795,
      "learning_rate": 0.0001738922198110323,
      "loss": 3.7015,
      "step": 140040
    },
    {
      "epoch": 0.13054926119768112,
      "grad_norm": 1.1037399768829346,
      "learning_rate": 0.00017389035548893712,
      "loss": 3.6009,
      "step": 140050
    },
    {
      "epoch": 0.1305585828157602,
      "grad_norm": 1.0473334789276123,
      "learning_rate": 0.00017388849116684193,
      "loss": 3.6325,
      "step": 140060
    },
    {
      "epoch": 0.13056790443383928,
      "grad_norm": 0.9970321655273438,
      "learning_rate": 0.00017388662684474673,
      "loss": 3.5871,
      "step": 140070
    },
    {
      "epoch": 0.13057722605191838,
      "grad_norm": 1.3944003582000732,
      "learning_rate": 0.0001738847625226515,
      "loss": 3.4152,
      "step": 140080
    },
    {
      "epoch": 0.13058654766999747,
      "grad_norm": 1.0289369821548462,
      "learning_rate": 0.00017388289820055632,
      "loss": 3.4317,
      "step": 140090
    },
    {
      "epoch": 0.13059586928807657,
      "grad_norm": 1.097071886062622,
      "learning_rate": 0.00017388103387846112,
      "loss": 3.5903,
      "step": 140100
    },
    {
      "epoch": 0.13060519090615566,
      "grad_norm": 1.0464481115341187,
      "learning_rate": 0.00017387916955636593,
      "loss": 3.5624,
      "step": 140110
    },
    {
      "epoch": 0.13061451252423475,
      "grad_norm": 0.8730531334877014,
      "learning_rate": 0.00017387730523427074,
      "loss": 3.4522,
      "step": 140120
    },
    {
      "epoch": 0.13062383414231385,
      "grad_norm": 1.1269654035568237,
      "learning_rate": 0.00017387544091217552,
      "loss": 3.5339,
      "step": 140130
    },
    {
      "epoch": 0.13063315576039294,
      "grad_norm": 1.0054324865341187,
      "learning_rate": 0.00017387357659008032,
      "loss": 3.6663,
      "step": 140140
    },
    {
      "epoch": 0.13064247737847204,
      "grad_norm": 1.0454670190811157,
      "learning_rate": 0.00017387171226798513,
      "loss": 3.6109,
      "step": 140150
    },
    {
      "epoch": 0.1306517989965511,
      "grad_norm": 1.1302945613861084,
      "learning_rate": 0.0001738698479458899,
      "loss": 3.6361,
      "step": 140160
    },
    {
      "epoch": 0.1306611206146302,
      "grad_norm": 1.2121580839157104,
      "learning_rate": 0.00017386798362379474,
      "loss": 3.5334,
      "step": 140170
    },
    {
      "epoch": 0.1306704422327093,
      "grad_norm": 1.0403993129730225,
      "learning_rate": 0.00017386611930169952,
      "loss": 3.4998,
      "step": 140180
    },
    {
      "epoch": 0.1306797638507884,
      "grad_norm": 1.1207317113876343,
      "learning_rate": 0.00017386425497960433,
      "loss": 3.6685,
      "step": 140190
    },
    {
      "epoch": 0.13068908546886748,
      "grad_norm": 1.1527551412582397,
      "learning_rate": 0.00017386239065750913,
      "loss": 3.6918,
      "step": 140200
    },
    {
      "epoch": 0.13069840708694658,
      "grad_norm": 1.0221290588378906,
      "learning_rate": 0.0001738605263354139,
      "loss": 3.4513,
      "step": 140210
    },
    {
      "epoch": 0.13070772870502567,
      "grad_norm": 1.214298129081726,
      "learning_rate": 0.00017385866201331872,
      "loss": 3.5078,
      "step": 140220
    },
    {
      "epoch": 0.13071705032310477,
      "grad_norm": 1.1601861715316772,
      "learning_rate": 0.00017385679769122352,
      "loss": 3.5489,
      "step": 140230
    },
    {
      "epoch": 0.13072637194118386,
      "grad_norm": 1.2301043272018433,
      "learning_rate": 0.00017385493336912833,
      "loss": 3.5786,
      "step": 140240
    },
    {
      "epoch": 0.13073569355926296,
      "grad_norm": 1.052327275276184,
      "learning_rate": 0.00017385306904703314,
      "loss": 3.6864,
      "step": 140250
    },
    {
      "epoch": 0.13074501517734205,
      "grad_norm": 1.0739463567733765,
      "learning_rate": 0.00017385120472493792,
      "loss": 3.5577,
      "step": 140260
    },
    {
      "epoch": 0.13075433679542112,
      "grad_norm": 1.0464240312576294,
      "learning_rate": 0.00017384934040284272,
      "loss": 3.6406,
      "step": 140270
    },
    {
      "epoch": 0.1307636584135002,
      "grad_norm": 1.1290260553359985,
      "learning_rate": 0.00017384747608074753,
      "loss": 3.7061,
      "step": 140280
    },
    {
      "epoch": 0.1307729800315793,
      "grad_norm": 1.058935523033142,
      "learning_rate": 0.00017384561175865233,
      "loss": 3.6243,
      "step": 140290
    },
    {
      "epoch": 0.1307823016496584,
      "grad_norm": 0.9456980228424072,
      "learning_rate": 0.00017384374743655714,
      "loss": 3.5986,
      "step": 140300
    },
    {
      "epoch": 0.1307916232677375,
      "grad_norm": 1.0614782571792603,
      "learning_rate": 0.00017384188311446192,
      "loss": 3.7534,
      "step": 140310
    },
    {
      "epoch": 0.1308009448858166,
      "grad_norm": 1.0758516788482666,
      "learning_rate": 0.00017384001879236673,
      "loss": 3.6909,
      "step": 140320
    },
    {
      "epoch": 0.13081026650389568,
      "grad_norm": 1.0578296184539795,
      "learning_rate": 0.0001738381544702715,
      "loss": 3.5921,
      "step": 140330
    },
    {
      "epoch": 0.13081958812197478,
      "grad_norm": 1.225870966911316,
      "learning_rate": 0.00017383629014817634,
      "loss": 3.4834,
      "step": 140340
    },
    {
      "epoch": 0.13082890974005387,
      "grad_norm": 1.1977938413619995,
      "learning_rate": 0.00017383442582608115,
      "loss": 3.79,
      "step": 140350
    },
    {
      "epoch": 0.13083823135813297,
      "grad_norm": 1.093934178352356,
      "learning_rate": 0.00017383256150398592,
      "loss": 3.5884,
      "step": 140360
    },
    {
      "epoch": 0.13084755297621203,
      "grad_norm": 1.1560542583465576,
      "learning_rate": 0.00017383069718189073,
      "loss": 3.6863,
      "step": 140370
    },
    {
      "epoch": 0.13085687459429113,
      "grad_norm": 1.0102026462554932,
      "learning_rate": 0.0001738288328597955,
      "loss": 3.8268,
      "step": 140380
    },
    {
      "epoch": 0.13086619621237022,
      "grad_norm": 1.308609127998352,
      "learning_rate": 0.00017382696853770032,
      "loss": 3.5179,
      "step": 140390
    },
    {
      "epoch": 0.13087551783044932,
      "grad_norm": 1.0958478450775146,
      "learning_rate": 0.00017382510421560515,
      "loss": 3.8206,
      "step": 140400
    },
    {
      "epoch": 0.1308848394485284,
      "grad_norm": 1.0192545652389526,
      "learning_rate": 0.00017382323989350993,
      "loss": 3.5919,
      "step": 140410
    },
    {
      "epoch": 0.1308941610666075,
      "grad_norm": 1.0100929737091064,
      "learning_rate": 0.00017382137557141474,
      "loss": 3.6361,
      "step": 140420
    },
    {
      "epoch": 0.1309034826846866,
      "grad_norm": 0.9730914235115051,
      "learning_rate": 0.00017381951124931951,
      "loss": 3.4278,
      "step": 140430
    },
    {
      "epoch": 0.1309128043027657,
      "grad_norm": 0.8669832348823547,
      "learning_rate": 0.00017381764692722432,
      "loss": 3.6542,
      "step": 140440
    },
    {
      "epoch": 0.1309221259208448,
      "grad_norm": 1.0188976526260376,
      "learning_rate": 0.00017381578260512913,
      "loss": 3.5319,
      "step": 140450
    },
    {
      "epoch": 0.13093144753892388,
      "grad_norm": 0.8460687398910522,
      "learning_rate": 0.00017381391828303393,
      "loss": 3.6023,
      "step": 140460
    },
    {
      "epoch": 0.13094076915700295,
      "grad_norm": 0.907326340675354,
      "learning_rate": 0.00017381205396093874,
      "loss": 3.3946,
      "step": 140470
    },
    {
      "epoch": 0.13095009077508205,
      "grad_norm": 0.9922136068344116,
      "learning_rate": 0.00017381018963884352,
      "loss": 3.5576,
      "step": 140480
    },
    {
      "epoch": 0.13095941239316114,
      "grad_norm": 1.2653526067733765,
      "learning_rate": 0.00017380832531674833,
      "loss": 3.6813,
      "step": 140490
    },
    {
      "epoch": 0.13096873401124023,
      "grad_norm": 0.8514553904533386,
      "learning_rate": 0.00017380646099465313,
      "loss": 3.6247,
      "step": 140500
    },
    {
      "epoch": 0.13097805562931933,
      "grad_norm": 1.1157382726669312,
      "learning_rate": 0.00017380459667255794,
      "loss": 3.5748,
      "step": 140510
    },
    {
      "epoch": 0.13098737724739842,
      "grad_norm": 0.9776082038879395,
      "learning_rate": 0.00017380273235046274,
      "loss": 3.6519,
      "step": 140520
    },
    {
      "epoch": 0.13099669886547752,
      "grad_norm": 1.0953508615493774,
      "learning_rate": 0.00017380086802836752,
      "loss": 3.6186,
      "step": 140530
    },
    {
      "epoch": 0.1310060204835566,
      "grad_norm": 1.0604429244995117,
      "learning_rate": 0.00017379900370627233,
      "loss": 3.719,
      "step": 140540
    },
    {
      "epoch": 0.1310153421016357,
      "grad_norm": 1.1737596988677979,
      "learning_rate": 0.00017379713938417714,
      "loss": 3.4667,
      "step": 140550
    },
    {
      "epoch": 0.1310246637197148,
      "grad_norm": 1.1606826782226562,
      "learning_rate": 0.00017379527506208194,
      "loss": 3.6321,
      "step": 140560
    },
    {
      "epoch": 0.1310339853377939,
      "grad_norm": 0.9598985314369202,
      "learning_rate": 0.00017379341073998675,
      "loss": 3.6876,
      "step": 140570
    },
    {
      "epoch": 0.13104330695587296,
      "grad_norm": 1.0782796144485474,
      "learning_rate": 0.00017379154641789153,
      "loss": 3.6524,
      "step": 140580
    },
    {
      "epoch": 0.13105262857395206,
      "grad_norm": 0.8472484350204468,
      "learning_rate": 0.00017378968209579633,
      "loss": 3.6628,
      "step": 140590
    },
    {
      "epoch": 0.13106195019203115,
      "grad_norm": 1.1985334157943726,
      "learning_rate": 0.00017378781777370114,
      "loss": 3.7265,
      "step": 140600
    },
    {
      "epoch": 0.13107127181011025,
      "grad_norm": 1.049764633178711,
      "learning_rate": 0.00017378595345160592,
      "loss": 3.576,
      "step": 140610
    },
    {
      "epoch": 0.13108059342818934,
      "grad_norm": 1.085094928741455,
      "learning_rate": 0.00017378408912951075,
      "loss": 3.5328,
      "step": 140620
    },
    {
      "epoch": 0.13108991504626843,
      "grad_norm": 0.979727566242218,
      "learning_rate": 0.00017378222480741553,
      "loss": 3.7781,
      "step": 140630
    },
    {
      "epoch": 0.13109923666434753,
      "grad_norm": 1.0328481197357178,
      "learning_rate": 0.00017378036048532034,
      "loss": 3.6033,
      "step": 140640
    },
    {
      "epoch": 0.13110855828242662,
      "grad_norm": 1.2105247974395752,
      "learning_rate": 0.00017377849616322515,
      "loss": 3.3147,
      "step": 140650
    },
    {
      "epoch": 0.13111787990050572,
      "grad_norm": 1.0974699258804321,
      "learning_rate": 0.00017377663184112992,
      "loss": 3.3031,
      "step": 140660
    },
    {
      "epoch": 0.1311272015185848,
      "grad_norm": 1.0861085653305054,
      "learning_rate": 0.00017377476751903473,
      "loss": 3.7254,
      "step": 140670
    },
    {
      "epoch": 0.13113652313666388,
      "grad_norm": 1.183046817779541,
      "learning_rate": 0.00017377290319693954,
      "loss": 3.7235,
      "step": 140680
    },
    {
      "epoch": 0.13114584475474297,
      "grad_norm": 1.0969194173812866,
      "learning_rate": 0.00017377103887484434,
      "loss": 3.6498,
      "step": 140690
    },
    {
      "epoch": 0.13115516637282207,
      "grad_norm": 0.9749898314476013,
      "learning_rate": 0.00017376917455274915,
      "loss": 3.7885,
      "step": 140700
    },
    {
      "epoch": 0.13116448799090116,
      "grad_norm": 0.9930137991905212,
      "learning_rate": 0.00017376731023065393,
      "loss": 3.7117,
      "step": 140710
    },
    {
      "epoch": 0.13117380960898026,
      "grad_norm": 0.9270446300506592,
      "learning_rate": 0.00017376544590855874,
      "loss": 3.7289,
      "step": 140720
    },
    {
      "epoch": 0.13118313122705935,
      "grad_norm": 1.1082795858383179,
      "learning_rate": 0.00017376358158646354,
      "loss": 3.8227,
      "step": 140730
    },
    {
      "epoch": 0.13119245284513845,
      "grad_norm": 1.0893665552139282,
      "learning_rate": 0.00017376171726436835,
      "loss": 3.764,
      "step": 140740
    },
    {
      "epoch": 0.13120177446321754,
      "grad_norm": 0.9890646934509277,
      "learning_rate": 0.00017375985294227315,
      "loss": 3.7928,
      "step": 140750
    },
    {
      "epoch": 0.13121109608129664,
      "grad_norm": 1.2881478071212769,
      "learning_rate": 0.00017375798862017793,
      "loss": 3.6322,
      "step": 140760
    },
    {
      "epoch": 0.13122041769937573,
      "grad_norm": 1.0076850652694702,
      "learning_rate": 0.00017375612429808274,
      "loss": 3.4694,
      "step": 140770
    },
    {
      "epoch": 0.13122973931745482,
      "grad_norm": 1.0056414604187012,
      "learning_rate": 0.00017375425997598752,
      "loss": 3.6611,
      "step": 140780
    },
    {
      "epoch": 0.1312390609355339,
      "grad_norm": 1.296365737915039,
      "learning_rate": 0.00017375239565389235,
      "loss": 3.5004,
      "step": 140790
    },
    {
      "epoch": 0.13124838255361299,
      "grad_norm": 0.9842895865440369,
      "learning_rate": 0.00017375053133179716,
      "loss": 3.7197,
      "step": 140800
    },
    {
      "epoch": 0.13125770417169208,
      "grad_norm": 1.043647289276123,
      "learning_rate": 0.00017374866700970194,
      "loss": 3.5212,
      "step": 140810
    },
    {
      "epoch": 0.13126702578977117,
      "grad_norm": 1.1735596656799316,
      "learning_rate": 0.00017374680268760674,
      "loss": 3.5899,
      "step": 140820
    },
    {
      "epoch": 0.13127634740785027,
      "grad_norm": 1.0780036449432373,
      "learning_rate": 0.00017374493836551152,
      "loss": 3.4978,
      "step": 140830
    },
    {
      "epoch": 0.13128566902592936,
      "grad_norm": 1.1562488079071045,
      "learning_rate": 0.00017374307404341633,
      "loss": 3.7172,
      "step": 140840
    },
    {
      "epoch": 0.13129499064400846,
      "grad_norm": 1.4016629457473755,
      "learning_rate": 0.00017374120972132116,
      "loss": 3.7163,
      "step": 140850
    },
    {
      "epoch": 0.13130431226208755,
      "grad_norm": 0.9476912617683411,
      "learning_rate": 0.00017373934539922594,
      "loss": 3.5144,
      "step": 140860
    },
    {
      "epoch": 0.13131363388016665,
      "grad_norm": 0.9981677532196045,
      "learning_rate": 0.00017373748107713075,
      "loss": 3.6174,
      "step": 140870
    },
    {
      "epoch": 0.13132295549824574,
      "grad_norm": 1.1253138780593872,
      "learning_rate": 0.00017373561675503553,
      "loss": 3.6711,
      "step": 140880
    },
    {
      "epoch": 0.1313322771163248,
      "grad_norm": 0.9239888191223145,
      "learning_rate": 0.00017373375243294033,
      "loss": 3.5878,
      "step": 140890
    },
    {
      "epoch": 0.1313415987344039,
      "grad_norm": 1.056989312171936,
      "learning_rate": 0.00017373188811084514,
      "loss": 3.5143,
      "step": 140900
    },
    {
      "epoch": 0.131350920352483,
      "grad_norm": 1.0732252597808838,
      "learning_rate": 0.00017373002378874995,
      "loss": 3.4994,
      "step": 140910
    },
    {
      "epoch": 0.1313602419705621,
      "grad_norm": 0.7400368452072144,
      "learning_rate": 0.00017372815946665475,
      "loss": 3.5556,
      "step": 140920
    },
    {
      "epoch": 0.13136956358864119,
      "grad_norm": 1.7020505666732788,
      "learning_rate": 0.00017372629514455953,
      "loss": 3.5328,
      "step": 140930
    },
    {
      "epoch": 0.13137888520672028,
      "grad_norm": 1.0944663286209106,
      "learning_rate": 0.00017372443082246434,
      "loss": 3.4692,
      "step": 140940
    },
    {
      "epoch": 0.13138820682479938,
      "grad_norm": 1.1551228761672974,
      "learning_rate": 0.00017372256650036914,
      "loss": 3.5519,
      "step": 140950
    },
    {
      "epoch": 0.13139752844287847,
      "grad_norm": 1.0712772607803345,
      "learning_rate": 0.00017372070217827395,
      "loss": 3.7242,
      "step": 140960
    },
    {
      "epoch": 0.13140685006095756,
      "grad_norm": 0.9994120597839355,
      "learning_rate": 0.00017371883785617876,
      "loss": 3.6892,
      "step": 140970
    },
    {
      "epoch": 0.13141617167903666,
      "grad_norm": 0.9848719835281372,
      "learning_rate": 0.00017371697353408354,
      "loss": 3.4733,
      "step": 140980
    },
    {
      "epoch": 0.13142549329711575,
      "grad_norm": 1.1086245775222778,
      "learning_rate": 0.00017371510921198834,
      "loss": 3.6578,
      "step": 140990
    },
    {
      "epoch": 0.13143481491519482,
      "grad_norm": 1.0831183195114136,
      "learning_rate": 0.00017371324488989315,
      "loss": 3.8238,
      "step": 141000
    },
    {
      "epoch": 0.13144413653327391,
      "grad_norm": 1.2069454193115234,
      "learning_rate": 0.00017371138056779796,
      "loss": 3.7739,
      "step": 141010
    },
    {
      "epoch": 0.131453458151353,
      "grad_norm": 1.0887264013290405,
      "learning_rate": 0.00017370951624570276,
      "loss": 3.8181,
      "step": 141020
    },
    {
      "epoch": 0.1314627797694321,
      "grad_norm": 0.9741010069847107,
      "learning_rate": 0.00017370765192360754,
      "loss": 3.4792,
      "step": 141030
    },
    {
      "epoch": 0.1314721013875112,
      "grad_norm": 1.038987159729004,
      "learning_rate": 0.00017370578760151235,
      "loss": 3.5942,
      "step": 141040
    },
    {
      "epoch": 0.1314814230055903,
      "grad_norm": 0.9525599479675293,
      "learning_rate": 0.00017370392327941713,
      "loss": 3.6408,
      "step": 141050
    },
    {
      "epoch": 0.1314907446236694,
      "grad_norm": 1.2068040370941162,
      "learning_rate": 0.00017370205895732193,
      "loss": 3.5853,
      "step": 141060
    },
    {
      "epoch": 0.13150006624174848,
      "grad_norm": 1.728853702545166,
      "learning_rate": 0.00017370019463522677,
      "loss": 3.6143,
      "step": 141070
    },
    {
      "epoch": 0.13150938785982758,
      "grad_norm": 0.9922076463699341,
      "learning_rate": 0.00017369833031313155,
      "loss": 3.4491,
      "step": 141080
    },
    {
      "epoch": 0.13151870947790667,
      "grad_norm": 1.0263336896896362,
      "learning_rate": 0.00017369646599103635,
      "loss": 3.5845,
      "step": 141090
    },
    {
      "epoch": 0.13152803109598574,
      "grad_norm": 0.9142574071884155,
      "learning_rate": 0.00017369460166894113,
      "loss": 3.6008,
      "step": 141100
    },
    {
      "epoch": 0.13153735271406483,
      "grad_norm": 1.1714932918548584,
      "learning_rate": 0.00017369273734684594,
      "loss": 3.5712,
      "step": 141110
    },
    {
      "epoch": 0.13154667433214393,
      "grad_norm": 1.5423892736434937,
      "learning_rate": 0.00017369087302475074,
      "loss": 3.6478,
      "step": 141120
    },
    {
      "epoch": 0.13155599595022302,
      "grad_norm": 1.231038212776184,
      "learning_rate": 0.00017368900870265555,
      "loss": 3.6559,
      "step": 141130
    },
    {
      "epoch": 0.13156531756830211,
      "grad_norm": 1.0476675033569336,
      "learning_rate": 0.00017368714438056036,
      "loss": 3.7869,
      "step": 141140
    },
    {
      "epoch": 0.1315746391863812,
      "grad_norm": 1.0560753345489502,
      "learning_rate": 0.00017368528005846514,
      "loss": 3.5237,
      "step": 141150
    },
    {
      "epoch": 0.1315839608044603,
      "grad_norm": 1.2140001058578491,
      "learning_rate": 0.00017368341573636994,
      "loss": 3.6403,
      "step": 141160
    },
    {
      "epoch": 0.1315932824225394,
      "grad_norm": 1.2333765029907227,
      "learning_rate": 0.00017368155141427475,
      "loss": 3.7658,
      "step": 141170
    },
    {
      "epoch": 0.1316026040406185,
      "grad_norm": 1.042032241821289,
      "learning_rate": 0.00017367968709217955,
      "loss": 3.6868,
      "step": 141180
    },
    {
      "epoch": 0.1316119256586976,
      "grad_norm": 1.0260076522827148,
      "learning_rate": 0.00017367782277008436,
      "loss": 3.5026,
      "step": 141190
    },
    {
      "epoch": 0.13162124727677668,
      "grad_norm": 1.1461308002471924,
      "learning_rate": 0.00017367595844798914,
      "loss": 3.6176,
      "step": 141200
    },
    {
      "epoch": 0.13163056889485575,
      "grad_norm": 1.0190231800079346,
      "learning_rate": 0.00017367409412589395,
      "loss": 3.5161,
      "step": 141210
    },
    {
      "epoch": 0.13163989051293484,
      "grad_norm": 0.9807432293891907,
      "learning_rate": 0.00017367222980379875,
      "loss": 3.6071,
      "step": 141220
    },
    {
      "epoch": 0.13164921213101394,
      "grad_norm": 1.0384992361068726,
      "learning_rate": 0.00017367036548170356,
      "loss": 3.7032,
      "step": 141230
    },
    {
      "epoch": 0.13165853374909303,
      "grad_norm": 1.1432477235794067,
      "learning_rate": 0.00017366850115960837,
      "loss": 3.4873,
      "step": 141240
    },
    {
      "epoch": 0.13166785536717213,
      "grad_norm": 1.1573340892791748,
      "learning_rate": 0.00017366663683751314,
      "loss": 3.5307,
      "step": 141250
    },
    {
      "epoch": 0.13167717698525122,
      "grad_norm": 0.9847189784049988,
      "learning_rate": 0.00017366477251541795,
      "loss": 3.5804,
      "step": 141260
    },
    {
      "epoch": 0.13168649860333032,
      "grad_norm": 1.239007592201233,
      "learning_rate": 0.00017366290819332276,
      "loss": 3.5477,
      "step": 141270
    },
    {
      "epoch": 0.1316958202214094,
      "grad_norm": 1.0642582178115845,
      "learning_rate": 0.00017366104387122754,
      "loss": 3.5125,
      "step": 141280
    },
    {
      "epoch": 0.1317051418394885,
      "grad_norm": 1.0726039409637451,
      "learning_rate": 0.00017365917954913234,
      "loss": 3.5767,
      "step": 141290
    },
    {
      "epoch": 0.1317144634575676,
      "grad_norm": 1.0112602710723877,
      "learning_rate": 0.00017365731522703715,
      "loss": 3.7742,
      "step": 141300
    },
    {
      "epoch": 0.13172378507564667,
      "grad_norm": 1.1320974826812744,
      "learning_rate": 0.00017365545090494195,
      "loss": 3.3655,
      "step": 141310
    },
    {
      "epoch": 0.13173310669372576,
      "grad_norm": 1.0218850374221802,
      "learning_rate": 0.00017365358658284676,
      "loss": 3.4284,
      "step": 141320
    },
    {
      "epoch": 0.13174242831180485,
      "grad_norm": 1.2441869974136353,
      "learning_rate": 0.00017365172226075154,
      "loss": 3.7926,
      "step": 141330
    },
    {
      "epoch": 0.13175174992988395,
      "grad_norm": 1.105584740638733,
      "learning_rate": 0.00017364985793865635,
      "loss": 3.7455,
      "step": 141340
    },
    {
      "epoch": 0.13176107154796304,
      "grad_norm": 0.9850826263427734,
      "learning_rate": 0.00017364799361656115,
      "loss": 3.7195,
      "step": 141350
    },
    {
      "epoch": 0.13177039316604214,
      "grad_norm": 1.210041880607605,
      "learning_rate": 0.00017364612929446596,
      "loss": 3.6589,
      "step": 141360
    },
    {
      "epoch": 0.13177971478412123,
      "grad_norm": 0.9024181365966797,
      "learning_rate": 0.00017364426497237077,
      "loss": 3.8297,
      "step": 141370
    },
    {
      "epoch": 0.13178903640220033,
      "grad_norm": 1.1316810846328735,
      "learning_rate": 0.00017364240065027554,
      "loss": 3.5711,
      "step": 141380
    },
    {
      "epoch": 0.13179835802027942,
      "grad_norm": 0.815209686756134,
      "learning_rate": 0.00017364053632818035,
      "loss": 3.6253,
      "step": 141390
    },
    {
      "epoch": 0.13180767963835852,
      "grad_norm": 1.117229700088501,
      "learning_rate": 0.00017363867200608516,
      "loss": 3.5831,
      "step": 141400
    },
    {
      "epoch": 0.1318170012564376,
      "grad_norm": 1.1951014995574951,
      "learning_rate": 0.00017363680768398996,
      "loss": 3.666,
      "step": 141410
    },
    {
      "epoch": 0.13182632287451668,
      "grad_norm": 1.2627999782562256,
      "learning_rate": 0.00017363494336189477,
      "loss": 3.7147,
      "step": 141420
    },
    {
      "epoch": 0.13183564449259577,
      "grad_norm": 1.4949270486831665,
      "learning_rate": 0.00017363307903979955,
      "loss": 3.6565,
      "step": 141430
    },
    {
      "epoch": 0.13184496611067487,
      "grad_norm": 1.0169702768325806,
      "learning_rate": 0.00017363121471770436,
      "loss": 3.4825,
      "step": 141440
    },
    {
      "epoch": 0.13185428772875396,
      "grad_norm": 1.1213898658752441,
      "learning_rate": 0.00017362935039560913,
      "loss": 3.485,
      "step": 141450
    },
    {
      "epoch": 0.13186360934683306,
      "grad_norm": 1.036543846130371,
      "learning_rate": 0.00017362748607351397,
      "loss": 3.6051,
      "step": 141460
    },
    {
      "epoch": 0.13187293096491215,
      "grad_norm": 0.8513811230659485,
      "learning_rate": 0.00017362562175141877,
      "loss": 3.7995,
      "step": 141470
    },
    {
      "epoch": 0.13188225258299124,
      "grad_norm": 1.0506998300552368,
      "learning_rate": 0.00017362375742932355,
      "loss": 3.4498,
      "step": 141480
    },
    {
      "epoch": 0.13189157420107034,
      "grad_norm": 1.025757908821106,
      "learning_rate": 0.00017362189310722836,
      "loss": 3.6615,
      "step": 141490
    },
    {
      "epoch": 0.13190089581914943,
      "grad_norm": 1.1028553247451782,
      "learning_rate": 0.00017362002878513314,
      "loss": 3.7444,
      "step": 141500
    },
    {
      "epoch": 0.13191021743722853,
      "grad_norm": 0.9121530652046204,
      "learning_rate": 0.00017361816446303795,
      "loss": 3.6104,
      "step": 141510
    },
    {
      "epoch": 0.1319195390553076,
      "grad_norm": 1.2583788633346558,
      "learning_rate": 0.00017361630014094278,
      "loss": 3.5902,
      "step": 141520
    },
    {
      "epoch": 0.1319288606733867,
      "grad_norm": 1.1205159425735474,
      "learning_rate": 0.00017361443581884756,
      "loss": 3.6274,
      "step": 141530
    },
    {
      "epoch": 0.13193818229146578,
      "grad_norm": 1.2459543943405151,
      "learning_rate": 0.00017361257149675236,
      "loss": 3.6325,
      "step": 141540
    },
    {
      "epoch": 0.13194750390954488,
      "grad_norm": 1.1419202089309692,
      "learning_rate": 0.00017361070717465714,
      "loss": 3.7222,
      "step": 141550
    },
    {
      "epoch": 0.13195682552762397,
      "grad_norm": 1.1592267751693726,
      "learning_rate": 0.00017360884285256195,
      "loss": 3.6505,
      "step": 141560
    },
    {
      "epoch": 0.13196614714570307,
      "grad_norm": 1.2263545989990234,
      "learning_rate": 0.00017360697853046676,
      "loss": 3.7654,
      "step": 141570
    },
    {
      "epoch": 0.13197546876378216,
      "grad_norm": 0.9917284250259399,
      "learning_rate": 0.00017360511420837156,
      "loss": 3.5963,
      "step": 141580
    },
    {
      "epoch": 0.13198479038186126,
      "grad_norm": 1.0421812534332275,
      "learning_rate": 0.00017360324988627637,
      "loss": 3.645,
      "step": 141590
    },
    {
      "epoch": 0.13199411199994035,
      "grad_norm": 1.0153907537460327,
      "learning_rate": 0.00017360138556418115,
      "loss": 3.6705,
      "step": 141600
    },
    {
      "epoch": 0.13200343361801944,
      "grad_norm": 1.045425534248352,
      "learning_rate": 0.00017359952124208595,
      "loss": 3.5991,
      "step": 141610
    },
    {
      "epoch": 0.13201275523609854,
      "grad_norm": 0.9646424055099487,
      "learning_rate": 0.00017359765691999076,
      "loss": 3.5055,
      "step": 141620
    },
    {
      "epoch": 0.1320220768541776,
      "grad_norm": 1.0350538492202759,
      "learning_rate": 0.00017359579259789557,
      "loss": 3.5096,
      "step": 141630
    },
    {
      "epoch": 0.1320313984722567,
      "grad_norm": 1.299453616142273,
      "learning_rate": 0.00017359392827580037,
      "loss": 3.7567,
      "step": 141640
    },
    {
      "epoch": 0.1320407200903358,
      "grad_norm": 1.103090524673462,
      "learning_rate": 0.00017359206395370515,
      "loss": 3.5432,
      "step": 141650
    },
    {
      "epoch": 0.1320500417084149,
      "grad_norm": 1.1456043720245361,
      "learning_rate": 0.00017359019963160996,
      "loss": 3.555,
      "step": 141660
    },
    {
      "epoch": 0.13205936332649398,
      "grad_norm": 1.0403655767440796,
      "learning_rate": 0.00017358833530951477,
      "loss": 3.3556,
      "step": 141670
    },
    {
      "epoch": 0.13206868494457308,
      "grad_norm": 1.1674830913543701,
      "learning_rate": 0.00017358647098741957,
      "loss": 3.6156,
      "step": 141680
    },
    {
      "epoch": 0.13207800656265217,
      "grad_norm": 1.0689268112182617,
      "learning_rate": 0.00017358460666532438,
      "loss": 3.5507,
      "step": 141690
    },
    {
      "epoch": 0.13208732818073127,
      "grad_norm": 1.131371259689331,
      "learning_rate": 0.00017358274234322916,
      "loss": 3.694,
      "step": 141700
    },
    {
      "epoch": 0.13209664979881036,
      "grad_norm": 0.869385302066803,
      "learning_rate": 0.00017358087802113396,
      "loss": 3.5409,
      "step": 141710
    },
    {
      "epoch": 0.13210597141688946,
      "grad_norm": 1.4031788110733032,
      "learning_rate": 0.00017357901369903877,
      "loss": 3.6172,
      "step": 141720
    },
    {
      "epoch": 0.13211529303496852,
      "grad_norm": 1.1257058382034302,
      "learning_rate": 0.00017357714937694355,
      "loss": 3.6066,
      "step": 141730
    },
    {
      "epoch": 0.13212461465304762,
      "grad_norm": 1.089459776878357,
      "learning_rate": 0.00017357528505484838,
      "loss": 3.6965,
      "step": 141740
    },
    {
      "epoch": 0.1321339362711267,
      "grad_norm": 0.9605054259300232,
      "learning_rate": 0.00017357342073275316,
      "loss": 3.526,
      "step": 141750
    },
    {
      "epoch": 0.1321432578892058,
      "grad_norm": 1.0613625049591064,
      "learning_rate": 0.00017357155641065797,
      "loss": 3.528,
      "step": 141760
    },
    {
      "epoch": 0.1321525795072849,
      "grad_norm": 1.151256799697876,
      "learning_rate": 0.00017356969208856277,
      "loss": 3.5314,
      "step": 141770
    },
    {
      "epoch": 0.132161901125364,
      "grad_norm": 1.1121647357940674,
      "learning_rate": 0.00017356782776646755,
      "loss": 3.5017,
      "step": 141780
    },
    {
      "epoch": 0.1321712227434431,
      "grad_norm": 0.832202136516571,
      "learning_rate": 0.00017356596344437236,
      "loss": 3.7585,
      "step": 141790
    },
    {
      "epoch": 0.13218054436152218,
      "grad_norm": 1.0413888692855835,
      "learning_rate": 0.00017356409912227717,
      "loss": 3.5393,
      "step": 141800
    },
    {
      "epoch": 0.13218986597960128,
      "grad_norm": 1.255972981452942,
      "learning_rate": 0.00017356223480018197,
      "loss": 3.7331,
      "step": 141810
    },
    {
      "epoch": 0.13219918759768037,
      "grad_norm": 0.9258784055709839,
      "learning_rate": 0.00017356037047808678,
      "loss": 3.5967,
      "step": 141820
    },
    {
      "epoch": 0.13220850921575944,
      "grad_norm": 0.9520998001098633,
      "learning_rate": 0.00017355850615599156,
      "loss": 3.5767,
      "step": 141830
    },
    {
      "epoch": 0.13221783083383853,
      "grad_norm": 1.001806378364563,
      "learning_rate": 0.00017355664183389636,
      "loss": 3.556,
      "step": 141840
    },
    {
      "epoch": 0.13222715245191763,
      "grad_norm": 1.057199478149414,
      "learning_rate": 0.00017355477751180117,
      "loss": 3.6835,
      "step": 141850
    },
    {
      "epoch": 0.13223647406999672,
      "grad_norm": 1.208609700202942,
      "learning_rate": 0.00017355291318970598,
      "loss": 3.5856,
      "step": 141860
    },
    {
      "epoch": 0.13224579568807582,
      "grad_norm": 0.9385389685630798,
      "learning_rate": 0.00017355104886761078,
      "loss": 3.6774,
      "step": 141870
    },
    {
      "epoch": 0.1322551173061549,
      "grad_norm": 1.0884685516357422,
      "learning_rate": 0.00017354918454551556,
      "loss": 3.523,
      "step": 141880
    },
    {
      "epoch": 0.132264438924234,
      "grad_norm": 1.0775189399719238,
      "learning_rate": 0.00017354732022342037,
      "loss": 3.5708,
      "step": 141890
    },
    {
      "epoch": 0.1322737605423131,
      "grad_norm": 1.0449870824813843,
      "learning_rate": 0.00017354545590132515,
      "loss": 3.6536,
      "step": 141900
    },
    {
      "epoch": 0.1322830821603922,
      "grad_norm": 1.0387685298919678,
      "learning_rate": 0.00017354359157922998,
      "loss": 3.7057,
      "step": 141910
    },
    {
      "epoch": 0.1322924037784713,
      "grad_norm": 1.103301763534546,
      "learning_rate": 0.0001735417272571348,
      "loss": 3.6154,
      "step": 141920
    },
    {
      "epoch": 0.13230172539655038,
      "grad_norm": 1.242194414138794,
      "learning_rate": 0.00017353986293503957,
      "loss": 3.5697,
      "step": 141930
    },
    {
      "epoch": 0.13231104701462945,
      "grad_norm": 1.2758359909057617,
      "learning_rate": 0.00017353799861294437,
      "loss": 3.6058,
      "step": 141940
    },
    {
      "epoch": 0.13232036863270855,
      "grad_norm": 1.1863385438919067,
      "learning_rate": 0.00017353613429084915,
      "loss": 3.4636,
      "step": 141950
    },
    {
      "epoch": 0.13232969025078764,
      "grad_norm": 1.0409818887710571,
      "learning_rate": 0.00017353426996875396,
      "loss": 3.7958,
      "step": 141960
    },
    {
      "epoch": 0.13233901186886673,
      "grad_norm": 0.9739046692848206,
      "learning_rate": 0.0001735324056466588,
      "loss": 3.7682,
      "step": 141970
    },
    {
      "epoch": 0.13234833348694583,
      "grad_norm": 1.055726170539856,
      "learning_rate": 0.00017353054132456357,
      "loss": 3.7021,
      "step": 141980
    },
    {
      "epoch": 0.13235765510502492,
      "grad_norm": 1.0359971523284912,
      "learning_rate": 0.00017352867700246838,
      "loss": 3.4736,
      "step": 141990
    },
    {
      "epoch": 0.13236697672310402,
      "grad_norm": 1.0226209163665771,
      "learning_rate": 0.00017352681268037316,
      "loss": 3.5366,
      "step": 142000
    },
    {
      "epoch": 0.1323762983411831,
      "grad_norm": 0.8749827742576599,
      "learning_rate": 0.00017352494835827796,
      "loss": 3.3306,
      "step": 142010
    },
    {
      "epoch": 0.1323856199592622,
      "grad_norm": 1.1870009899139404,
      "learning_rate": 0.00017352308403618277,
      "loss": 3.6172,
      "step": 142020
    },
    {
      "epoch": 0.1323949415773413,
      "grad_norm": 1.171270489692688,
      "learning_rate": 0.00017352121971408758,
      "loss": 3.6895,
      "step": 142030
    },
    {
      "epoch": 0.13240426319542037,
      "grad_norm": 1.2868225574493408,
      "learning_rate": 0.00017351935539199238,
      "loss": 3.5923,
      "step": 142040
    },
    {
      "epoch": 0.13241358481349946,
      "grad_norm": 1.0033918619155884,
      "learning_rate": 0.00017351749106989716,
      "loss": 3.4547,
      "step": 142050
    },
    {
      "epoch": 0.13242290643157856,
      "grad_norm": 1.1983593702316284,
      "learning_rate": 0.00017351562674780197,
      "loss": 3.6334,
      "step": 142060
    },
    {
      "epoch": 0.13243222804965765,
      "grad_norm": 0.9602853059768677,
      "learning_rate": 0.00017351376242570677,
      "loss": 3.7013,
      "step": 142070
    },
    {
      "epoch": 0.13244154966773675,
      "grad_norm": 1.061034917831421,
      "learning_rate": 0.00017351189810361158,
      "loss": 3.6393,
      "step": 142080
    },
    {
      "epoch": 0.13245087128581584,
      "grad_norm": 1.1534171104431152,
      "learning_rate": 0.00017351003378151639,
      "loss": 3.6048,
      "step": 142090
    },
    {
      "epoch": 0.13246019290389494,
      "grad_norm": 1.1222716569900513,
      "learning_rate": 0.00017350816945942117,
      "loss": 3.5628,
      "step": 142100
    },
    {
      "epoch": 0.13246951452197403,
      "grad_norm": 0.974503219127655,
      "learning_rate": 0.00017350630513732597,
      "loss": 3.7425,
      "step": 142110
    },
    {
      "epoch": 0.13247883614005312,
      "grad_norm": 0.9649922847747803,
      "learning_rate": 0.00017350444081523078,
      "loss": 3.6699,
      "step": 142120
    },
    {
      "epoch": 0.13248815775813222,
      "grad_norm": 1.2524654865264893,
      "learning_rate": 0.00017350257649313558,
      "loss": 3.6865,
      "step": 142130
    },
    {
      "epoch": 0.1324974793762113,
      "grad_norm": 1.1143548488616943,
      "learning_rate": 0.0001735007121710404,
      "loss": 3.4553,
      "step": 142140
    },
    {
      "epoch": 0.13250680099429038,
      "grad_norm": 1.0729470252990723,
      "learning_rate": 0.00017349884784894517,
      "loss": 3.5894,
      "step": 142150
    },
    {
      "epoch": 0.13251612261236947,
      "grad_norm": 0.9283246397972107,
      "learning_rate": 0.00017349698352684998,
      "loss": 3.6439,
      "step": 142160
    },
    {
      "epoch": 0.13252544423044857,
      "grad_norm": 1.0453267097473145,
      "learning_rate": 0.00017349511920475478,
      "loss": 3.6035,
      "step": 142170
    },
    {
      "epoch": 0.13253476584852766,
      "grad_norm": 1.0373964309692383,
      "learning_rate": 0.00017349325488265956,
      "loss": 3.4953,
      "step": 142180
    },
    {
      "epoch": 0.13254408746660676,
      "grad_norm": 0.9997280836105347,
      "learning_rate": 0.0001734913905605644,
      "loss": 3.7816,
      "step": 142190
    },
    {
      "epoch": 0.13255340908468585,
      "grad_norm": 1.058599591255188,
      "learning_rate": 0.00017348952623846917,
      "loss": 3.5349,
      "step": 142200
    },
    {
      "epoch": 0.13256273070276495,
      "grad_norm": 1.0332472324371338,
      "learning_rate": 0.00017348766191637398,
      "loss": 3.6644,
      "step": 142210
    },
    {
      "epoch": 0.13257205232084404,
      "grad_norm": 1.1989630460739136,
      "learning_rate": 0.0001734857975942788,
      "loss": 3.5872,
      "step": 142220
    },
    {
      "epoch": 0.13258137393892314,
      "grad_norm": 1.060988187789917,
      "learning_rate": 0.00017348393327218357,
      "loss": 3.5093,
      "step": 142230
    },
    {
      "epoch": 0.13259069555700223,
      "grad_norm": 0.8727290630340576,
      "learning_rate": 0.00017348206895008837,
      "loss": 3.5992,
      "step": 142240
    },
    {
      "epoch": 0.1326000171750813,
      "grad_norm": 1.040845274925232,
      "learning_rate": 0.00017348020462799318,
      "loss": 3.5431,
      "step": 142250
    },
    {
      "epoch": 0.1326093387931604,
      "grad_norm": 0.9985387325286865,
      "learning_rate": 0.00017347834030589799,
      "loss": 3.5417,
      "step": 142260
    },
    {
      "epoch": 0.1326186604112395,
      "grad_norm": 1.2391964197158813,
      "learning_rate": 0.0001734764759838028,
      "loss": 3.3264,
      "step": 142270
    },
    {
      "epoch": 0.13262798202931858,
      "grad_norm": 1.0525919198989868,
      "learning_rate": 0.00017347461166170757,
      "loss": 3.5268,
      "step": 142280
    },
    {
      "epoch": 0.13263730364739768,
      "grad_norm": 1.1082415580749512,
      "learning_rate": 0.00017347274733961238,
      "loss": 3.5104,
      "step": 142290
    },
    {
      "epoch": 0.13264662526547677,
      "grad_norm": 1.0275522470474243,
      "learning_rate": 0.00017347088301751718,
      "loss": 3.6533,
      "step": 142300
    },
    {
      "epoch": 0.13265594688355586,
      "grad_norm": 1.1035354137420654,
      "learning_rate": 0.000173469018695422,
      "loss": 3.3139,
      "step": 142310
    },
    {
      "epoch": 0.13266526850163496,
      "grad_norm": 1.042834758758545,
      "learning_rate": 0.0001734671543733268,
      "loss": 3.5278,
      "step": 142320
    },
    {
      "epoch": 0.13267459011971405,
      "grad_norm": 1.0185331106185913,
      "learning_rate": 0.00017346529005123158,
      "loss": 3.4267,
      "step": 142330
    },
    {
      "epoch": 0.13268391173779315,
      "grad_norm": 1.069031834602356,
      "learning_rate": 0.00017346342572913638,
      "loss": 3.729,
      "step": 142340
    },
    {
      "epoch": 0.13269323335587224,
      "grad_norm": 1.1731890439987183,
      "learning_rate": 0.00017346156140704116,
      "loss": 3.7217,
      "step": 142350
    },
    {
      "epoch": 0.1327025549739513,
      "grad_norm": 1.0431790351867676,
      "learning_rate": 0.000173459697084946,
      "loss": 3.7265,
      "step": 142360
    },
    {
      "epoch": 0.1327118765920304,
      "grad_norm": 1.0960304737091064,
      "learning_rate": 0.0001734578327628508,
      "loss": 3.7571,
      "step": 142370
    },
    {
      "epoch": 0.1327211982101095,
      "grad_norm": 0.9932330846786499,
      "learning_rate": 0.00017345596844075558,
      "loss": 3.6469,
      "step": 142380
    },
    {
      "epoch": 0.1327305198281886,
      "grad_norm": 1.0400649309158325,
      "learning_rate": 0.00017345410411866039,
      "loss": 3.523,
      "step": 142390
    },
    {
      "epoch": 0.1327398414462677,
      "grad_norm": 1.1622576713562012,
      "learning_rate": 0.00017345223979656516,
      "loss": 3.6418,
      "step": 142400
    },
    {
      "epoch": 0.13274916306434678,
      "grad_norm": 1.0654373168945312,
      "learning_rate": 0.00017345037547446997,
      "loss": 3.6329,
      "step": 142410
    },
    {
      "epoch": 0.13275848468242588,
      "grad_norm": 1.2913827896118164,
      "learning_rate": 0.00017344851115237478,
      "loss": 3.6356,
      "step": 142420
    },
    {
      "epoch": 0.13276780630050497,
      "grad_norm": 0.9805472493171692,
      "learning_rate": 0.00017344664683027958,
      "loss": 3.4256,
      "step": 142430
    },
    {
      "epoch": 0.13277712791858406,
      "grad_norm": 1.1378439664840698,
      "learning_rate": 0.0001734447825081844,
      "loss": 3.581,
      "step": 142440
    },
    {
      "epoch": 0.13278644953666316,
      "grad_norm": 1.0406337976455688,
      "learning_rate": 0.00017344291818608917,
      "loss": 3.6039,
      "step": 142450
    },
    {
      "epoch": 0.13279577115474223,
      "grad_norm": 1.054105281829834,
      "learning_rate": 0.00017344105386399398,
      "loss": 3.7397,
      "step": 142460
    },
    {
      "epoch": 0.13280509277282132,
      "grad_norm": 0.9645848870277405,
      "learning_rate": 0.00017343918954189878,
      "loss": 3.6373,
      "step": 142470
    },
    {
      "epoch": 0.13281441439090041,
      "grad_norm": 0.992351233959198,
      "learning_rate": 0.0001734373252198036,
      "loss": 3.5053,
      "step": 142480
    },
    {
      "epoch": 0.1328237360089795,
      "grad_norm": 1.142372727394104,
      "learning_rate": 0.0001734354608977084,
      "loss": 3.4995,
      "step": 142490
    },
    {
      "epoch": 0.1328330576270586,
      "grad_norm": 1.2667872905731201,
      "learning_rate": 0.00017343359657561317,
      "loss": 3.7047,
      "step": 142500
    },
    {
      "epoch": 0.1328423792451377,
      "grad_norm": 1.0026302337646484,
      "learning_rate": 0.00017343173225351798,
      "loss": 3.6016,
      "step": 142510
    },
    {
      "epoch": 0.1328517008632168,
      "grad_norm": 0.9520883560180664,
      "learning_rate": 0.0001734298679314228,
      "loss": 3.4661,
      "step": 142520
    },
    {
      "epoch": 0.1328610224812959,
      "grad_norm": 0.7885790467262268,
      "learning_rate": 0.0001734280036093276,
      "loss": 3.5925,
      "step": 142530
    },
    {
      "epoch": 0.13287034409937498,
      "grad_norm": 1.002431035041809,
      "learning_rate": 0.0001734261392872324,
      "loss": 3.6585,
      "step": 142540
    },
    {
      "epoch": 0.13287966571745408,
      "grad_norm": 1.0668078660964966,
      "learning_rate": 0.00017342427496513718,
      "loss": 3.6828,
      "step": 142550
    },
    {
      "epoch": 0.13288898733553317,
      "grad_norm": 1.1792443990707397,
      "learning_rate": 0.00017342241064304198,
      "loss": 3.6267,
      "step": 142560
    },
    {
      "epoch": 0.13289830895361224,
      "grad_norm": 0.9520564079284668,
      "learning_rate": 0.00017342054632094676,
      "loss": 3.5902,
      "step": 142570
    },
    {
      "epoch": 0.13290763057169133,
      "grad_norm": 1.0222845077514648,
      "learning_rate": 0.0001734186819988516,
      "loss": 3.6354,
      "step": 142580
    },
    {
      "epoch": 0.13291695218977043,
      "grad_norm": 1.2530747652053833,
      "learning_rate": 0.0001734168176767564,
      "loss": 3.6456,
      "step": 142590
    },
    {
      "epoch": 0.13292627380784952,
      "grad_norm": 1.1370378732681274,
      "learning_rate": 0.00017341495335466118,
      "loss": 3.7141,
      "step": 142600
    },
    {
      "epoch": 0.13293559542592862,
      "grad_norm": 0.9515537023544312,
      "learning_rate": 0.000173413089032566,
      "loss": 3.443,
      "step": 142610
    },
    {
      "epoch": 0.1329449170440077,
      "grad_norm": 1.1645640134811401,
      "learning_rate": 0.00017341122471047077,
      "loss": 3.6026,
      "step": 142620
    },
    {
      "epoch": 0.1329542386620868,
      "grad_norm": 1.0966320037841797,
      "learning_rate": 0.00017340936038837557,
      "loss": 3.5688,
      "step": 142630
    },
    {
      "epoch": 0.1329635602801659,
      "grad_norm": 0.9778701066970825,
      "learning_rate": 0.0001734074960662804,
      "loss": 3.5755,
      "step": 142640
    },
    {
      "epoch": 0.132972881898245,
      "grad_norm": 0.8737784624099731,
      "learning_rate": 0.0001734056317441852,
      "loss": 3.5842,
      "step": 142650
    },
    {
      "epoch": 0.1329822035163241,
      "grad_norm": 1.1284611225128174,
      "learning_rate": 0.00017340376742209,
      "loss": 3.5834,
      "step": 142660
    },
    {
      "epoch": 0.13299152513440315,
      "grad_norm": 1.1922402381896973,
      "learning_rate": 0.00017340190309999477,
      "loss": 3.6215,
      "step": 142670
    },
    {
      "epoch": 0.13300084675248225,
      "grad_norm": 1.1829324960708618,
      "learning_rate": 0.00017340003877789958,
      "loss": 3.4695,
      "step": 142680
    },
    {
      "epoch": 0.13301016837056134,
      "grad_norm": 1.1353532075881958,
      "learning_rate": 0.00017339817445580439,
      "loss": 3.3757,
      "step": 142690
    },
    {
      "epoch": 0.13301948998864044,
      "grad_norm": 1.0289020538330078,
      "learning_rate": 0.0001733963101337092,
      "loss": 3.7522,
      "step": 142700
    },
    {
      "epoch": 0.13302881160671953,
      "grad_norm": 0.9802441596984863,
      "learning_rate": 0.000173394445811614,
      "loss": 3.6793,
      "step": 142710
    },
    {
      "epoch": 0.13303813322479863,
      "grad_norm": 1.037143349647522,
      "learning_rate": 0.00017339258148951878,
      "loss": 3.648,
      "step": 142720
    },
    {
      "epoch": 0.13304745484287772,
      "grad_norm": 1.1103583574295044,
      "learning_rate": 0.00017339071716742358,
      "loss": 3.5408,
      "step": 142730
    },
    {
      "epoch": 0.13305677646095682,
      "grad_norm": 0.956678032875061,
      "learning_rate": 0.0001733888528453284,
      "loss": 3.6999,
      "step": 142740
    },
    {
      "epoch": 0.1330660980790359,
      "grad_norm": 1.075364589691162,
      "learning_rate": 0.0001733869885232332,
      "loss": 3.7139,
      "step": 142750
    },
    {
      "epoch": 0.133075419697115,
      "grad_norm": 1.3853262662887573,
      "learning_rate": 0.000173385124201138,
      "loss": 3.6952,
      "step": 142760
    },
    {
      "epoch": 0.1330847413151941,
      "grad_norm": 1.1248009204864502,
      "learning_rate": 0.00017338325987904278,
      "loss": 3.5651,
      "step": 142770
    },
    {
      "epoch": 0.13309406293327317,
      "grad_norm": 0.9850024580955505,
      "learning_rate": 0.0001733813955569476,
      "loss": 3.522,
      "step": 142780
    },
    {
      "epoch": 0.13310338455135226,
      "grad_norm": 1.0747108459472656,
      "learning_rate": 0.0001733795312348524,
      "loss": 3.7602,
      "step": 142790
    },
    {
      "epoch": 0.13311270616943136,
      "grad_norm": 1.1084247827529907,
      "learning_rate": 0.00017337766691275717,
      "loss": 3.7192,
      "step": 142800
    },
    {
      "epoch": 0.13312202778751045,
      "grad_norm": 0.9061776399612427,
      "learning_rate": 0.000173375802590662,
      "loss": 3.5809,
      "step": 142810
    },
    {
      "epoch": 0.13313134940558954,
      "grad_norm": 0.8820642232894897,
      "learning_rate": 0.00017337393826856679,
      "loss": 3.6373,
      "step": 142820
    },
    {
      "epoch": 0.13314067102366864,
      "grad_norm": 0.878497838973999,
      "learning_rate": 0.0001733720739464716,
      "loss": 3.5874,
      "step": 142830
    },
    {
      "epoch": 0.13314999264174773,
      "grad_norm": 1.1223316192626953,
      "learning_rate": 0.0001733702096243764,
      "loss": 3.8196,
      "step": 142840
    },
    {
      "epoch": 0.13315931425982683,
      "grad_norm": 1.0243375301361084,
      "learning_rate": 0.00017336834530228118,
      "loss": 3.499,
      "step": 142850
    },
    {
      "epoch": 0.13316863587790592,
      "grad_norm": 1.2386764287948608,
      "learning_rate": 0.00017336648098018598,
      "loss": 3.5722,
      "step": 142860
    },
    {
      "epoch": 0.13317795749598502,
      "grad_norm": 1.015763759613037,
      "learning_rate": 0.0001733646166580908,
      "loss": 3.5671,
      "step": 142870
    },
    {
      "epoch": 0.13318727911406408,
      "grad_norm": 0.9194005131721497,
      "learning_rate": 0.0001733627523359956,
      "loss": 3.6977,
      "step": 142880
    },
    {
      "epoch": 0.13319660073214318,
      "grad_norm": 0.9688779711723328,
      "learning_rate": 0.0001733608880139004,
      "loss": 3.6952,
      "step": 142890
    },
    {
      "epoch": 0.13320592235022227,
      "grad_norm": 1.0712558031082153,
      "learning_rate": 0.00017335902369180518,
      "loss": 3.5041,
      "step": 142900
    },
    {
      "epoch": 0.13321524396830137,
      "grad_norm": 1.0089730024337769,
      "learning_rate": 0.00017335715936971,
      "loss": 3.4339,
      "step": 142910
    },
    {
      "epoch": 0.13322456558638046,
      "grad_norm": 2.251481056213379,
      "learning_rate": 0.0001733552950476148,
      "loss": 3.5822,
      "step": 142920
    },
    {
      "epoch": 0.13323388720445956,
      "grad_norm": 0.9684910774230957,
      "learning_rate": 0.0001733534307255196,
      "loss": 3.6096,
      "step": 142930
    },
    {
      "epoch": 0.13324320882253865,
      "grad_norm": 0.9073813557624817,
      "learning_rate": 0.0001733515664034244,
      "loss": 3.6041,
      "step": 142940
    },
    {
      "epoch": 0.13325253044061774,
      "grad_norm": 0.9750484228134155,
      "learning_rate": 0.0001733497020813292,
      "loss": 3.7083,
      "step": 142950
    },
    {
      "epoch": 0.13326185205869684,
      "grad_norm": 1.1387897729873657,
      "learning_rate": 0.000173347837759234,
      "loss": 3.6031,
      "step": 142960
    },
    {
      "epoch": 0.13327117367677593,
      "grad_norm": 0.9980868101119995,
      "learning_rate": 0.0001733459734371388,
      "loss": 3.5883,
      "step": 142970
    },
    {
      "epoch": 0.13328049529485503,
      "grad_norm": 0.9558262825012207,
      "learning_rate": 0.0001733441091150436,
      "loss": 3.6582,
      "step": 142980
    },
    {
      "epoch": 0.1332898169129341,
      "grad_norm": 1.0307729244232178,
      "learning_rate": 0.0001733422447929484,
      "loss": 3.5815,
      "step": 142990
    },
    {
      "epoch": 0.1332991385310132,
      "grad_norm": 1.0884037017822266,
      "learning_rate": 0.0001733403804708532,
      "loss": 3.6342,
      "step": 143000
    },
    {
      "epoch": 0.13330846014909228,
      "grad_norm": 0.9720397591590881,
      "learning_rate": 0.000173338516148758,
      "loss": 3.602,
      "step": 143010
    },
    {
      "epoch": 0.13331778176717138,
      "grad_norm": 1.0833204984664917,
      "learning_rate": 0.00017333665182666278,
      "loss": 3.8172,
      "step": 143020
    },
    {
      "epoch": 0.13332710338525047,
      "grad_norm": 0.9677874445915222,
      "learning_rate": 0.0001733347875045676,
      "loss": 3.5703,
      "step": 143030
    },
    {
      "epoch": 0.13333642500332957,
      "grad_norm": 0.9457354545593262,
      "learning_rate": 0.00017333292318247242,
      "loss": 3.5821,
      "step": 143040
    },
    {
      "epoch": 0.13334574662140866,
      "grad_norm": 0.856688916683197,
      "learning_rate": 0.0001733310588603772,
      "loss": 3.7666,
      "step": 143050
    },
    {
      "epoch": 0.13335506823948776,
      "grad_norm": 1.0267325639724731,
      "learning_rate": 0.000173329194538282,
      "loss": 3.705,
      "step": 143060
    },
    {
      "epoch": 0.13336438985756685,
      "grad_norm": 0.8854440450668335,
      "learning_rate": 0.00017332733021618678,
      "loss": 3.584,
      "step": 143070
    },
    {
      "epoch": 0.13337371147564595,
      "grad_norm": 0.9620463848114014,
      "learning_rate": 0.0001733254658940916,
      "loss": 3.6865,
      "step": 143080
    },
    {
      "epoch": 0.133383033093725,
      "grad_norm": 1.1672909259796143,
      "learning_rate": 0.00017332360157199642,
      "loss": 3.6291,
      "step": 143090
    },
    {
      "epoch": 0.1333923547118041,
      "grad_norm": 1.117414951324463,
      "learning_rate": 0.0001733217372499012,
      "loss": 3.5879,
      "step": 143100
    },
    {
      "epoch": 0.1334016763298832,
      "grad_norm": 0.9144225716590881,
      "learning_rate": 0.000173319872927806,
      "loss": 3.5825,
      "step": 143110
    },
    {
      "epoch": 0.1334109979479623,
      "grad_norm": 1.0354607105255127,
      "learning_rate": 0.00017331800860571079,
      "loss": 3.6825,
      "step": 143120
    },
    {
      "epoch": 0.1334203195660414,
      "grad_norm": 1.0647330284118652,
      "learning_rate": 0.0001733161442836156,
      "loss": 3.7469,
      "step": 143130
    },
    {
      "epoch": 0.13342964118412048,
      "grad_norm": 1.2337175607681274,
      "learning_rate": 0.0001733142799615204,
      "loss": 3.4915,
      "step": 143140
    },
    {
      "epoch": 0.13343896280219958,
      "grad_norm": 0.8644975423812866,
      "learning_rate": 0.0001733124156394252,
      "loss": 3.6143,
      "step": 143150
    },
    {
      "epoch": 0.13344828442027867,
      "grad_norm": 1.053469181060791,
      "learning_rate": 0.00017331055131733,
      "loss": 3.5012,
      "step": 143160
    },
    {
      "epoch": 0.13345760603835777,
      "grad_norm": 0.834913432598114,
      "learning_rate": 0.0001733086869952348,
      "loss": 3.5174,
      "step": 143170
    },
    {
      "epoch": 0.13346692765643686,
      "grad_norm": 1.2208826541900635,
      "learning_rate": 0.0001733068226731396,
      "loss": 3.4391,
      "step": 143180
    },
    {
      "epoch": 0.13347624927451596,
      "grad_norm": 1.0652462244033813,
      "learning_rate": 0.0001733049583510444,
      "loss": 3.6509,
      "step": 143190
    },
    {
      "epoch": 0.13348557089259502,
      "grad_norm": 1.0261355638504028,
      "learning_rate": 0.0001733030940289492,
      "loss": 3.5318,
      "step": 143200
    },
    {
      "epoch": 0.13349489251067412,
      "grad_norm": 1.0923961400985718,
      "learning_rate": 0.00017330122970685402,
      "loss": 3.4783,
      "step": 143210
    },
    {
      "epoch": 0.1335042141287532,
      "grad_norm": 1.1484580039978027,
      "learning_rate": 0.0001732993653847588,
      "loss": 3.5852,
      "step": 143220
    },
    {
      "epoch": 0.1335135357468323,
      "grad_norm": 1.1420130729675293,
      "learning_rate": 0.0001732975010626636,
      "loss": 3.5627,
      "step": 143230
    },
    {
      "epoch": 0.1335228573649114,
      "grad_norm": 0.9962185025215149,
      "learning_rate": 0.0001732956367405684,
      "loss": 3.6943,
      "step": 143240
    },
    {
      "epoch": 0.1335321789829905,
      "grad_norm": 1.0255866050720215,
      "learning_rate": 0.00017329377241847319,
      "loss": 3.5567,
      "step": 143250
    },
    {
      "epoch": 0.1335415006010696,
      "grad_norm": 1.0679969787597656,
      "learning_rate": 0.00017329190809637802,
      "loss": 3.6979,
      "step": 143260
    },
    {
      "epoch": 0.13355082221914868,
      "grad_norm": 1.0925565958023071,
      "learning_rate": 0.0001732900437742828,
      "loss": 3.4446,
      "step": 143270
    },
    {
      "epoch": 0.13356014383722778,
      "grad_norm": 1.090665340423584,
      "learning_rate": 0.0001732881794521876,
      "loss": 3.5231,
      "step": 143280
    },
    {
      "epoch": 0.13356946545530687,
      "grad_norm": 1.0809195041656494,
      "learning_rate": 0.0001732863151300924,
      "loss": 3.6629,
      "step": 143290
    },
    {
      "epoch": 0.13357878707338594,
      "grad_norm": 0.999281108379364,
      "learning_rate": 0.0001732844508079972,
      "loss": 3.4074,
      "step": 143300
    },
    {
      "epoch": 0.13358810869146504,
      "grad_norm": 1.0433803796768188,
      "learning_rate": 0.000173282586485902,
      "loss": 3.4854,
      "step": 143310
    },
    {
      "epoch": 0.13359743030954413,
      "grad_norm": 0.9882596135139465,
      "learning_rate": 0.0001732807221638068,
      "loss": 3.5947,
      "step": 143320
    },
    {
      "epoch": 0.13360675192762322,
      "grad_norm": 1.1229534149169922,
      "learning_rate": 0.0001732788578417116,
      "loss": 3.4174,
      "step": 143330
    },
    {
      "epoch": 0.13361607354570232,
      "grad_norm": 1.0776492357254028,
      "learning_rate": 0.00017327699351961642,
      "loss": 3.6241,
      "step": 143340
    },
    {
      "epoch": 0.1336253951637814,
      "grad_norm": 0.9950577616691589,
      "learning_rate": 0.0001732751291975212,
      "loss": 3.5673,
      "step": 143350
    },
    {
      "epoch": 0.1336347167818605,
      "grad_norm": 0.9929913282394409,
      "learning_rate": 0.000173273264875426,
      "loss": 3.6659,
      "step": 143360
    },
    {
      "epoch": 0.1336440383999396,
      "grad_norm": 1.1773414611816406,
      "learning_rate": 0.0001732714005533308,
      "loss": 3.5052,
      "step": 143370
    },
    {
      "epoch": 0.1336533600180187,
      "grad_norm": 1.0543744564056396,
      "learning_rate": 0.00017326953623123561,
      "loss": 3.6403,
      "step": 143380
    },
    {
      "epoch": 0.1336626816360978,
      "grad_norm": 1.115688443183899,
      "learning_rate": 0.00017326767190914042,
      "loss": 3.5961,
      "step": 143390
    },
    {
      "epoch": 0.13367200325417686,
      "grad_norm": 0.9103766083717346,
      "learning_rate": 0.0001732658075870452,
      "loss": 3.7496,
      "step": 143400
    },
    {
      "epoch": 0.13368132487225595,
      "grad_norm": 1.337761640548706,
      "learning_rate": 0.00017326394326495,
      "loss": 3.7082,
      "step": 143410
    },
    {
      "epoch": 0.13369064649033505,
      "grad_norm": 1.261290431022644,
      "learning_rate": 0.0001732620789428548,
      "loss": 3.7055,
      "step": 143420
    },
    {
      "epoch": 0.13369996810841414,
      "grad_norm": 1.151963233947754,
      "learning_rate": 0.00017326021462075962,
      "loss": 3.4691,
      "step": 143430
    },
    {
      "epoch": 0.13370928972649324,
      "grad_norm": 1.1955766677856445,
      "learning_rate": 0.00017325835029866442,
      "loss": 3.4727,
      "step": 143440
    },
    {
      "epoch": 0.13371861134457233,
      "grad_norm": 1.107409954071045,
      "learning_rate": 0.0001732564859765692,
      "loss": 3.5942,
      "step": 143450
    },
    {
      "epoch": 0.13372793296265142,
      "grad_norm": 1.0538089275360107,
      "learning_rate": 0.000173254621654474,
      "loss": 3.7489,
      "step": 143460
    },
    {
      "epoch": 0.13373725458073052,
      "grad_norm": 2.6716911792755127,
      "learning_rate": 0.0001732527573323788,
      "loss": 3.4447,
      "step": 143470
    },
    {
      "epoch": 0.1337465761988096,
      "grad_norm": 1.2608890533447266,
      "learning_rate": 0.00017325089301028362,
      "loss": 3.4978,
      "step": 143480
    },
    {
      "epoch": 0.1337558978168887,
      "grad_norm": 0.9232627153396606,
      "learning_rate": 0.00017324902868818843,
      "loss": 3.5264,
      "step": 143490
    },
    {
      "epoch": 0.1337652194349678,
      "grad_norm": 0.9512648582458496,
      "learning_rate": 0.0001732471643660932,
      "loss": 3.5876,
      "step": 143500
    },
    {
      "epoch": 0.13377454105304687,
      "grad_norm": 1.161575436592102,
      "learning_rate": 0.00017324530004399801,
      "loss": 3.647,
      "step": 143510
    },
    {
      "epoch": 0.13378386267112596,
      "grad_norm": 1.0940988063812256,
      "learning_rate": 0.0001732434357219028,
      "loss": 3.6149,
      "step": 143520
    },
    {
      "epoch": 0.13379318428920506,
      "grad_norm": 1.06321382522583,
      "learning_rate": 0.0001732415713998076,
      "loss": 3.5069,
      "step": 143530
    },
    {
      "epoch": 0.13380250590728415,
      "grad_norm": 1.0452961921691895,
      "learning_rate": 0.00017323970707771243,
      "loss": 3.5767,
      "step": 143540
    },
    {
      "epoch": 0.13381182752536325,
      "grad_norm": 1.114756464958191,
      "learning_rate": 0.0001732378427556172,
      "loss": 3.6033,
      "step": 143550
    },
    {
      "epoch": 0.13382114914344234,
      "grad_norm": 0.9650659561157227,
      "learning_rate": 0.00017323597843352202,
      "loss": 3.896,
      "step": 143560
    },
    {
      "epoch": 0.13383047076152144,
      "grad_norm": 1.0196576118469238,
      "learning_rate": 0.0001732341141114268,
      "loss": 3.6673,
      "step": 143570
    },
    {
      "epoch": 0.13383979237960053,
      "grad_norm": 1.0234957933425903,
      "learning_rate": 0.0001732322497893316,
      "loss": 3.83,
      "step": 143580
    },
    {
      "epoch": 0.13384911399767963,
      "grad_norm": 0.9981874227523804,
      "learning_rate": 0.0001732303854672364,
      "loss": 3.7885,
      "step": 143590
    },
    {
      "epoch": 0.13385843561575872,
      "grad_norm": 1.0425575971603394,
      "learning_rate": 0.00017322852114514122,
      "loss": 3.4999,
      "step": 143600
    },
    {
      "epoch": 0.1338677572338378,
      "grad_norm": 1.1202330589294434,
      "learning_rate": 0.00017322665682304602,
      "loss": 3.6069,
      "step": 143610
    },
    {
      "epoch": 0.13387707885191688,
      "grad_norm": 1.1376935243606567,
      "learning_rate": 0.0001732247925009508,
      "loss": 3.5787,
      "step": 143620
    },
    {
      "epoch": 0.13388640046999598,
      "grad_norm": 1.1323440074920654,
      "learning_rate": 0.0001732229281788556,
      "loss": 3.4974,
      "step": 143630
    },
    {
      "epoch": 0.13389572208807507,
      "grad_norm": 1.1906358003616333,
      "learning_rate": 0.00017322106385676042,
      "loss": 3.5556,
      "step": 143640
    },
    {
      "epoch": 0.13390504370615416,
      "grad_norm": 1.1631600856781006,
      "learning_rate": 0.00017321919953466522,
      "loss": 3.5096,
      "step": 143650
    },
    {
      "epoch": 0.13391436532423326,
      "grad_norm": 0.8718444108963013,
      "learning_rate": 0.00017321733521257003,
      "loss": 3.6806,
      "step": 143660
    },
    {
      "epoch": 0.13392368694231235,
      "grad_norm": 1.0008471012115479,
      "learning_rate": 0.0001732154708904748,
      "loss": 3.4951,
      "step": 143670
    },
    {
      "epoch": 0.13393300856039145,
      "grad_norm": 0.9648892283439636,
      "learning_rate": 0.00017321360656837961,
      "loss": 3.5477,
      "step": 143680
    },
    {
      "epoch": 0.13394233017847054,
      "grad_norm": 1.176928162574768,
      "learning_rate": 0.00017321174224628442,
      "loss": 3.6574,
      "step": 143690
    },
    {
      "epoch": 0.13395165179654964,
      "grad_norm": 1.1882191896438599,
      "learning_rate": 0.0001732098779241892,
      "loss": 3.7555,
      "step": 143700
    },
    {
      "epoch": 0.13396097341462873,
      "grad_norm": 0.9316295385360718,
      "learning_rate": 0.00017320801360209403,
      "loss": 3.7278,
      "step": 143710
    },
    {
      "epoch": 0.1339702950327078,
      "grad_norm": 1.2291481494903564,
      "learning_rate": 0.0001732061492799988,
      "loss": 3.6947,
      "step": 143720
    },
    {
      "epoch": 0.1339796166507869,
      "grad_norm": 1.1492513418197632,
      "learning_rate": 0.00017320428495790362,
      "loss": 3.4626,
      "step": 143730
    },
    {
      "epoch": 0.133988938268866,
      "grad_norm": 1.1001110076904297,
      "learning_rate": 0.00017320242063580842,
      "loss": 3.6431,
      "step": 143740
    },
    {
      "epoch": 0.13399825988694508,
      "grad_norm": 1.031305193901062,
      "learning_rate": 0.0001732005563137132,
      "loss": 3.6275,
      "step": 143750
    },
    {
      "epoch": 0.13400758150502418,
      "grad_norm": 0.8322795033454895,
      "learning_rate": 0.000173198691991618,
      "loss": 3.4841,
      "step": 143760
    },
    {
      "epoch": 0.13401690312310327,
      "grad_norm": 0.9870697259902954,
      "learning_rate": 0.00017319682766952282,
      "loss": 3.3999,
      "step": 143770
    },
    {
      "epoch": 0.13402622474118236,
      "grad_norm": 0.8588347434997559,
      "learning_rate": 0.00017319496334742762,
      "loss": 3.6855,
      "step": 143780
    },
    {
      "epoch": 0.13403554635926146,
      "grad_norm": 1.1086721420288086,
      "learning_rate": 0.0001731930990253324,
      "loss": 3.6704,
      "step": 143790
    },
    {
      "epoch": 0.13404486797734055,
      "grad_norm": 1.0793911218643188,
      "learning_rate": 0.0001731912347032372,
      "loss": 3.5579,
      "step": 143800
    },
    {
      "epoch": 0.13405418959541965,
      "grad_norm": 0.9465522766113281,
      "learning_rate": 0.00017318937038114201,
      "loss": 3.7474,
      "step": 143810
    },
    {
      "epoch": 0.13406351121349872,
      "grad_norm": 1.0937775373458862,
      "learning_rate": 0.00017318750605904682,
      "loss": 3.5738,
      "step": 143820
    },
    {
      "epoch": 0.1340728328315778,
      "grad_norm": 1.0762499570846558,
      "learning_rate": 0.00017318564173695163,
      "loss": 3.5371,
      "step": 143830
    },
    {
      "epoch": 0.1340821544496569,
      "grad_norm": 1.201973795890808,
      "learning_rate": 0.0001731837774148564,
      "loss": 3.6309,
      "step": 143840
    },
    {
      "epoch": 0.134091476067736,
      "grad_norm": 1.149302363395691,
      "learning_rate": 0.0001731819130927612,
      "loss": 3.5897,
      "step": 143850
    },
    {
      "epoch": 0.1341007976858151,
      "grad_norm": 1.0858254432678223,
      "learning_rate": 0.00017318004877066602,
      "loss": 3.3966,
      "step": 143860
    },
    {
      "epoch": 0.1341101193038942,
      "grad_norm": 1.1561132669448853,
      "learning_rate": 0.00017317818444857082,
      "loss": 3.6026,
      "step": 143870
    },
    {
      "epoch": 0.13411944092197328,
      "grad_norm": 1.15341317653656,
      "learning_rate": 0.00017317632012647563,
      "loss": 3.5233,
      "step": 143880
    },
    {
      "epoch": 0.13412876254005238,
      "grad_norm": 1.0179111957550049,
      "learning_rate": 0.0001731744558043804,
      "loss": 3.6302,
      "step": 143890
    },
    {
      "epoch": 0.13413808415813147,
      "grad_norm": 1.8447751998901367,
      "learning_rate": 0.00017317259148228522,
      "loss": 3.546,
      "step": 143900
    },
    {
      "epoch": 0.13414740577621057,
      "grad_norm": 1.0792163610458374,
      "learning_rate": 0.00017317072716019002,
      "loss": 3.7961,
      "step": 143910
    },
    {
      "epoch": 0.13415672739428966,
      "grad_norm": 1.1187376976013184,
      "learning_rate": 0.0001731688628380948,
      "loss": 3.699,
      "step": 143920
    },
    {
      "epoch": 0.13416604901236873,
      "grad_norm": 1.1067891120910645,
      "learning_rate": 0.00017316699851599964,
      "loss": 3.6409,
      "step": 143930
    },
    {
      "epoch": 0.13417537063044782,
      "grad_norm": 1.2172623872756958,
      "learning_rate": 0.00017316513419390441,
      "loss": 3.6916,
      "step": 143940
    },
    {
      "epoch": 0.13418469224852692,
      "grad_norm": 1.1711597442626953,
      "learning_rate": 0.00017316326987180922,
      "loss": 3.4948,
      "step": 143950
    },
    {
      "epoch": 0.134194013866606,
      "grad_norm": 1.390174150466919,
      "learning_rate": 0.00017316140554971403,
      "loss": 3.6304,
      "step": 143960
    },
    {
      "epoch": 0.1342033354846851,
      "grad_norm": 1.046059489250183,
      "learning_rate": 0.0001731595412276188,
      "loss": 3.669,
      "step": 143970
    },
    {
      "epoch": 0.1342126571027642,
      "grad_norm": 0.8201963901519775,
      "learning_rate": 0.0001731576769055236,
      "loss": 3.5969,
      "step": 143980
    },
    {
      "epoch": 0.1342219787208433,
      "grad_norm": 1.2683024406433105,
      "learning_rate": 0.00017315581258342842,
      "loss": 3.5251,
      "step": 143990
    },
    {
      "epoch": 0.1342313003389224,
      "grad_norm": 1.0113449096679688,
      "learning_rate": 0.00017315394826133323,
      "loss": 3.3378,
      "step": 144000
    },
    {
      "epoch": 0.13424062195700148,
      "grad_norm": 1.2617781162261963,
      "learning_rate": 0.00017315208393923803,
      "loss": 3.6535,
      "step": 144010
    },
    {
      "epoch": 0.13424994357508058,
      "grad_norm": 1.104875922203064,
      "learning_rate": 0.0001731502196171428,
      "loss": 3.6197,
      "step": 144020
    },
    {
      "epoch": 0.13425926519315964,
      "grad_norm": 0.959153950214386,
      "learning_rate": 0.00017314835529504762,
      "loss": 3.5496,
      "step": 144030
    },
    {
      "epoch": 0.13426858681123874,
      "grad_norm": 1.108659267425537,
      "learning_rate": 0.00017314649097295242,
      "loss": 3.4979,
      "step": 144040
    },
    {
      "epoch": 0.13427790842931783,
      "grad_norm": 1.1252158880233765,
      "learning_rate": 0.00017314462665085723,
      "loss": 3.7222,
      "step": 144050
    },
    {
      "epoch": 0.13428723004739693,
      "grad_norm": 1.0858889818191528,
      "learning_rate": 0.00017314276232876204,
      "loss": 3.5896,
      "step": 144060
    },
    {
      "epoch": 0.13429655166547602,
      "grad_norm": 1.0233219861984253,
      "learning_rate": 0.00017314089800666682,
      "loss": 3.5897,
      "step": 144070
    },
    {
      "epoch": 0.13430587328355512,
      "grad_norm": 0.8665794134140015,
      "learning_rate": 0.00017313903368457162,
      "loss": 3.645,
      "step": 144080
    },
    {
      "epoch": 0.1343151949016342,
      "grad_norm": 1.1153357028961182,
      "learning_rate": 0.0001731371693624764,
      "loss": 3.5892,
      "step": 144090
    },
    {
      "epoch": 0.1343245165197133,
      "grad_norm": 1.0766304731369019,
      "learning_rate": 0.00017313530504038123,
      "loss": 3.6068,
      "step": 144100
    },
    {
      "epoch": 0.1343338381377924,
      "grad_norm": 1.2436283826828003,
      "learning_rate": 0.00017313344071828604,
      "loss": 3.6697,
      "step": 144110
    },
    {
      "epoch": 0.1343431597558715,
      "grad_norm": 1.0830131769180298,
      "learning_rate": 0.00017313157639619082,
      "loss": 3.6167,
      "step": 144120
    },
    {
      "epoch": 0.1343524813739506,
      "grad_norm": 0.9751098155975342,
      "learning_rate": 0.00017312971207409563,
      "loss": 3.5353,
      "step": 144130
    },
    {
      "epoch": 0.13436180299202966,
      "grad_norm": 1.045101284980774,
      "learning_rate": 0.0001731278477520004,
      "loss": 3.7323,
      "step": 144140
    },
    {
      "epoch": 0.13437112461010875,
      "grad_norm": 0.840776264667511,
      "learning_rate": 0.0001731259834299052,
      "loss": 3.6639,
      "step": 144150
    },
    {
      "epoch": 0.13438044622818784,
      "grad_norm": 0.9472745656967163,
      "learning_rate": 0.00017312411910781005,
      "loss": 3.5319,
      "step": 144160
    },
    {
      "epoch": 0.13438976784626694,
      "grad_norm": 0.7925543785095215,
      "learning_rate": 0.00017312225478571482,
      "loss": 3.6649,
      "step": 144170
    },
    {
      "epoch": 0.13439908946434603,
      "grad_norm": 1.150293231010437,
      "learning_rate": 0.00017312039046361963,
      "loss": 3.8207,
      "step": 144180
    },
    {
      "epoch": 0.13440841108242513,
      "grad_norm": 1.1004513502120972,
      "learning_rate": 0.0001731185261415244,
      "loss": 3.6315,
      "step": 144190
    },
    {
      "epoch": 0.13441773270050422,
      "grad_norm": 1.1841503381729126,
      "learning_rate": 0.00017311666181942922,
      "loss": 3.577,
      "step": 144200
    },
    {
      "epoch": 0.13442705431858332,
      "grad_norm": 1.1422585248947144,
      "learning_rate": 0.00017311479749733402,
      "loss": 3.4416,
      "step": 144210
    },
    {
      "epoch": 0.1344363759366624,
      "grad_norm": 1.001997709274292,
      "learning_rate": 0.00017311293317523883,
      "loss": 3.5235,
      "step": 144220
    },
    {
      "epoch": 0.1344456975547415,
      "grad_norm": 1.075750708580017,
      "learning_rate": 0.00017311106885314364,
      "loss": 3.5795,
      "step": 144230
    },
    {
      "epoch": 0.13445501917282057,
      "grad_norm": 1.1061866283416748,
      "learning_rate": 0.00017310920453104841,
      "loss": 3.7277,
      "step": 144240
    },
    {
      "epoch": 0.13446434079089967,
      "grad_norm": 1.0418132543563843,
      "learning_rate": 0.00017310734020895322,
      "loss": 3.4255,
      "step": 144250
    },
    {
      "epoch": 0.13447366240897876,
      "grad_norm": 1.0365899801254272,
      "learning_rate": 0.00017310547588685803,
      "loss": 3.5548,
      "step": 144260
    },
    {
      "epoch": 0.13448298402705786,
      "grad_norm": 0.8509095311164856,
      "learning_rate": 0.00017310361156476283,
      "loss": 3.5601,
      "step": 144270
    },
    {
      "epoch": 0.13449230564513695,
      "grad_norm": 1.0878934860229492,
      "learning_rate": 0.00017310174724266764,
      "loss": 3.5286,
      "step": 144280
    },
    {
      "epoch": 0.13450162726321604,
      "grad_norm": 1.046504259109497,
      "learning_rate": 0.00017309988292057242,
      "loss": 3.4621,
      "step": 144290
    },
    {
      "epoch": 0.13451094888129514,
      "grad_norm": 0.9351601600646973,
      "learning_rate": 0.00017309801859847723,
      "loss": 3.5661,
      "step": 144300
    },
    {
      "epoch": 0.13452027049937423,
      "grad_norm": 1.009705901145935,
      "learning_rate": 0.00017309615427638203,
      "loss": 3.5783,
      "step": 144310
    },
    {
      "epoch": 0.13452959211745333,
      "grad_norm": 1.143491506576538,
      "learning_rate": 0.00017309428995428684,
      "loss": 3.4507,
      "step": 144320
    },
    {
      "epoch": 0.13453891373553242,
      "grad_norm": 1.1133333444595337,
      "learning_rate": 0.00017309242563219164,
      "loss": 3.5092,
      "step": 144330
    },
    {
      "epoch": 0.13454823535361152,
      "grad_norm": 0.962555468082428,
      "learning_rate": 0.00017309056131009642,
      "loss": 3.6088,
      "step": 144340
    },
    {
      "epoch": 0.13455755697169058,
      "grad_norm": 0.9070379734039307,
      "learning_rate": 0.00017308869698800123,
      "loss": 3.5649,
      "step": 144350
    },
    {
      "epoch": 0.13456687858976968,
      "grad_norm": 1.0491176843643188,
      "learning_rate": 0.00017308683266590604,
      "loss": 3.6481,
      "step": 144360
    },
    {
      "epoch": 0.13457620020784877,
      "grad_norm": 1.315798044204712,
      "learning_rate": 0.00017308496834381082,
      "loss": 3.604,
      "step": 144370
    },
    {
      "epoch": 0.13458552182592787,
      "grad_norm": 1.0844566822052002,
      "learning_rate": 0.00017308310402171565,
      "loss": 3.7281,
      "step": 144380
    },
    {
      "epoch": 0.13459484344400696,
      "grad_norm": 0.9563794136047363,
      "learning_rate": 0.00017308123969962043,
      "loss": 3.5695,
      "step": 144390
    },
    {
      "epoch": 0.13460416506208606,
      "grad_norm": 1.1093165874481201,
      "learning_rate": 0.00017307937537752523,
      "loss": 3.7213,
      "step": 144400
    },
    {
      "epoch": 0.13461348668016515,
      "grad_norm": 0.9722132682800293,
      "learning_rate": 0.00017307751105543004,
      "loss": 3.4544,
      "step": 144410
    },
    {
      "epoch": 0.13462280829824425,
      "grad_norm": 1.3166884183883667,
      "learning_rate": 0.00017307564673333482,
      "loss": 3.5008,
      "step": 144420
    },
    {
      "epoch": 0.13463212991632334,
      "grad_norm": 1.6244454383850098,
      "learning_rate": 0.00017307378241123963,
      "loss": 3.6011,
      "step": 144430
    },
    {
      "epoch": 0.13464145153440243,
      "grad_norm": 1.1223219633102417,
      "learning_rate": 0.00017307191808914443,
      "loss": 3.5319,
      "step": 144440
    },
    {
      "epoch": 0.1346507731524815,
      "grad_norm": 1.2351188659667969,
      "learning_rate": 0.00017307005376704924,
      "loss": 3.56,
      "step": 144450
    },
    {
      "epoch": 0.1346600947705606,
      "grad_norm": 1.0961567163467407,
      "learning_rate": 0.00017306818944495404,
      "loss": 3.7193,
      "step": 144460
    },
    {
      "epoch": 0.1346694163886397,
      "grad_norm": 1.1188178062438965,
      "learning_rate": 0.00017306632512285882,
      "loss": 3.6848,
      "step": 144470
    },
    {
      "epoch": 0.13467873800671878,
      "grad_norm": 0.8019585013389587,
      "learning_rate": 0.00017306446080076363,
      "loss": 3.5994,
      "step": 144480
    },
    {
      "epoch": 0.13468805962479788,
      "grad_norm": 1.200348973274231,
      "learning_rate": 0.00017306259647866844,
      "loss": 3.6229,
      "step": 144490
    },
    {
      "epoch": 0.13469738124287697,
      "grad_norm": 1.1336361169815063,
      "learning_rate": 0.00017306073215657324,
      "loss": 3.6549,
      "step": 144500
    },
    {
      "epoch": 0.13470670286095607,
      "grad_norm": 1.0588172674179077,
      "learning_rate": 0.00017305886783447805,
      "loss": 3.7184,
      "step": 144510
    },
    {
      "epoch": 0.13471602447903516,
      "grad_norm": 1.0042952299118042,
      "learning_rate": 0.00017305700351238283,
      "loss": 3.7158,
      "step": 144520
    },
    {
      "epoch": 0.13472534609711426,
      "grad_norm": 1.0719681978225708,
      "learning_rate": 0.00017305513919028763,
      "loss": 3.6565,
      "step": 144530
    },
    {
      "epoch": 0.13473466771519335,
      "grad_norm": 1.0625343322753906,
      "learning_rate": 0.00017305327486819244,
      "loss": 3.4966,
      "step": 144540
    },
    {
      "epoch": 0.13474398933327245,
      "grad_norm": 1.0780513286590576,
      "learning_rate": 0.00017305141054609725,
      "loss": 3.7772,
      "step": 144550
    },
    {
      "epoch": 0.1347533109513515,
      "grad_norm": 1.2344698905944824,
      "learning_rate": 0.00017304954622400205,
      "loss": 3.5042,
      "step": 144560
    },
    {
      "epoch": 0.1347626325694306,
      "grad_norm": 1.1925452947616577,
      "learning_rate": 0.00017304768190190683,
      "loss": 3.4472,
      "step": 144570
    },
    {
      "epoch": 0.1347719541875097,
      "grad_norm": 1.186121940612793,
      "learning_rate": 0.00017304581757981164,
      "loss": 3.5663,
      "step": 144580
    },
    {
      "epoch": 0.1347812758055888,
      "grad_norm": 1.2644586563110352,
      "learning_rate": 0.00017304395325771642,
      "loss": 3.687,
      "step": 144590
    },
    {
      "epoch": 0.1347905974236679,
      "grad_norm": 1.2917147874832153,
      "learning_rate": 0.00017304208893562122,
      "loss": 3.5886,
      "step": 144600
    },
    {
      "epoch": 0.13479991904174699,
      "grad_norm": 1.3431143760681152,
      "learning_rate": 0.00017304022461352606,
      "loss": 3.7858,
      "step": 144610
    },
    {
      "epoch": 0.13480924065982608,
      "grad_norm": 1.0646874904632568,
      "learning_rate": 0.00017303836029143084,
      "loss": 3.5431,
      "step": 144620
    },
    {
      "epoch": 0.13481856227790517,
      "grad_norm": 1.126917839050293,
      "learning_rate": 0.00017303649596933564,
      "loss": 3.4377,
      "step": 144630
    },
    {
      "epoch": 0.13482788389598427,
      "grad_norm": 1.1277328729629517,
      "learning_rate": 0.00017303463164724042,
      "loss": 3.6516,
      "step": 144640
    },
    {
      "epoch": 0.13483720551406336,
      "grad_norm": 1.1337703466415405,
      "learning_rate": 0.00017303276732514523,
      "loss": 3.6408,
      "step": 144650
    },
    {
      "epoch": 0.13484652713214243,
      "grad_norm": 1.0524932146072388,
      "learning_rate": 0.00017303090300305004,
      "loss": 3.6115,
      "step": 144660
    },
    {
      "epoch": 0.13485584875022152,
      "grad_norm": 1.1986339092254639,
      "learning_rate": 0.00017302903868095484,
      "loss": 3.7048,
      "step": 144670
    },
    {
      "epoch": 0.13486517036830062,
      "grad_norm": 1.1033358573913574,
      "learning_rate": 0.00017302717435885965,
      "loss": 3.6486,
      "step": 144680
    },
    {
      "epoch": 0.1348744919863797,
      "grad_norm": 1.0493618249893188,
      "learning_rate": 0.00017302531003676443,
      "loss": 3.6425,
      "step": 144690
    },
    {
      "epoch": 0.1348838136044588,
      "grad_norm": 1.1264466047286987,
      "learning_rate": 0.00017302344571466923,
      "loss": 3.566,
      "step": 144700
    },
    {
      "epoch": 0.1348931352225379,
      "grad_norm": 0.998569667339325,
      "learning_rate": 0.00017302158139257404,
      "loss": 3.6131,
      "step": 144710
    },
    {
      "epoch": 0.134902456840617,
      "grad_norm": 0.9155606627464294,
      "learning_rate": 0.00017301971707047885,
      "loss": 3.6955,
      "step": 144720
    },
    {
      "epoch": 0.1349117784586961,
      "grad_norm": 0.9253014326095581,
      "learning_rate": 0.00017301785274838365,
      "loss": 3.6003,
      "step": 144730
    },
    {
      "epoch": 0.13492110007677519,
      "grad_norm": 1.2355159521102905,
      "learning_rate": 0.00017301598842628843,
      "loss": 3.6052,
      "step": 144740
    },
    {
      "epoch": 0.13493042169485428,
      "grad_norm": 1.0187609195709229,
      "learning_rate": 0.00017301412410419324,
      "loss": 3.6816,
      "step": 144750
    },
    {
      "epoch": 0.13493974331293335,
      "grad_norm": 0.8661412000656128,
      "learning_rate": 0.00017301225978209804,
      "loss": 3.6116,
      "step": 144760
    },
    {
      "epoch": 0.13494906493101244,
      "grad_norm": 0.9735150933265686,
      "learning_rate": 0.00017301039546000285,
      "loss": 3.6026,
      "step": 144770
    },
    {
      "epoch": 0.13495838654909154,
      "grad_norm": 0.9341749548912048,
      "learning_rate": 0.00017300853113790766,
      "loss": 3.6082,
      "step": 144780
    },
    {
      "epoch": 0.13496770816717063,
      "grad_norm": 1.4085427522659302,
      "learning_rate": 0.00017300666681581244,
      "loss": 3.5947,
      "step": 144790
    },
    {
      "epoch": 0.13497702978524972,
      "grad_norm": 0.9727792143821716,
      "learning_rate": 0.00017300480249371724,
      "loss": 3.5711,
      "step": 144800
    },
    {
      "epoch": 0.13498635140332882,
      "grad_norm": 0.9651365876197815,
      "learning_rate": 0.00017300293817162205,
      "loss": 3.6337,
      "step": 144810
    },
    {
      "epoch": 0.1349956730214079,
      "grad_norm": 1.1867769956588745,
      "learning_rate": 0.00017300107384952683,
      "loss": 3.4747,
      "step": 144820
    },
    {
      "epoch": 0.135004994639487,
      "grad_norm": 0.9792423248291016,
      "learning_rate": 0.00017299920952743166,
      "loss": 3.3836,
      "step": 144830
    },
    {
      "epoch": 0.1350143162575661,
      "grad_norm": 1.0508406162261963,
      "learning_rate": 0.00017299734520533644,
      "loss": 3.8039,
      "step": 144840
    },
    {
      "epoch": 0.1350236378756452,
      "grad_norm": 0.9577783346176147,
      "learning_rate": 0.00017299548088324125,
      "loss": 3.6211,
      "step": 144850
    },
    {
      "epoch": 0.1350329594937243,
      "grad_norm": 1.2310199737548828,
      "learning_rate": 0.00017299361656114605,
      "loss": 3.4746,
      "step": 144860
    },
    {
      "epoch": 0.13504228111180336,
      "grad_norm": 0.9799757599830627,
      "learning_rate": 0.00017299175223905083,
      "loss": 3.7565,
      "step": 144870
    },
    {
      "epoch": 0.13505160272988245,
      "grad_norm": 1.022999882698059,
      "learning_rate": 0.00017298988791695564,
      "loss": 3.5686,
      "step": 144880
    },
    {
      "epoch": 0.13506092434796155,
      "grad_norm": 1.1507164239883423,
      "learning_rate": 0.00017298802359486045,
      "loss": 3.6529,
      "step": 144890
    },
    {
      "epoch": 0.13507024596604064,
      "grad_norm": 1.2020059823989868,
      "learning_rate": 0.00017298615927276525,
      "loss": 3.646,
      "step": 144900
    },
    {
      "epoch": 0.13507956758411974,
      "grad_norm": 1.1940836906433105,
      "learning_rate": 0.00017298429495067006,
      "loss": 3.8366,
      "step": 144910
    },
    {
      "epoch": 0.13508888920219883,
      "grad_norm": 0.9347677230834961,
      "learning_rate": 0.00017298243062857484,
      "loss": 3.6713,
      "step": 144920
    },
    {
      "epoch": 0.13509821082027793,
      "grad_norm": 1.0577219724655151,
      "learning_rate": 0.00017298056630647964,
      "loss": 3.6785,
      "step": 144930
    },
    {
      "epoch": 0.13510753243835702,
      "grad_norm": 0.8824454545974731,
      "learning_rate": 0.00017297870198438445,
      "loss": 3.6258,
      "step": 144940
    },
    {
      "epoch": 0.13511685405643611,
      "grad_norm": 0.9695140719413757,
      "learning_rate": 0.00017297683766228926,
      "loss": 3.58,
      "step": 144950
    },
    {
      "epoch": 0.1351261756745152,
      "grad_norm": 1.0707521438598633,
      "learning_rate": 0.00017297497334019406,
      "loss": 3.6093,
      "step": 144960
    },
    {
      "epoch": 0.13513549729259428,
      "grad_norm": 1.0707335472106934,
      "learning_rate": 0.00017297310901809884,
      "loss": 3.5815,
      "step": 144970
    },
    {
      "epoch": 0.13514481891067337,
      "grad_norm": 1.1427091360092163,
      "learning_rate": 0.00017297124469600365,
      "loss": 3.6093,
      "step": 144980
    },
    {
      "epoch": 0.13515414052875246,
      "grad_norm": 1.0750941038131714,
      "learning_rate": 0.00017296938037390845,
      "loss": 3.6482,
      "step": 144990
    },
    {
      "epoch": 0.13516346214683156,
      "grad_norm": 0.8431552052497864,
      "learning_rate": 0.00017296751605181326,
      "loss": 3.5357,
      "step": 145000
    },
    {
      "epoch": 0.13517278376491065,
      "grad_norm": 1.1310092210769653,
      "learning_rate": 0.00017296565172971807,
      "loss": 3.7636,
      "step": 145010
    },
    {
      "epoch": 0.13518210538298975,
      "grad_norm": 0.9594446420669556,
      "learning_rate": 0.00017296378740762285,
      "loss": 3.5999,
      "step": 145020
    },
    {
      "epoch": 0.13519142700106884,
      "grad_norm": 1.1552965641021729,
      "learning_rate": 0.00017296192308552765,
      "loss": 3.6939,
      "step": 145030
    },
    {
      "epoch": 0.13520074861914794,
      "grad_norm": 0.9006357789039612,
      "learning_rate": 0.00017296005876343243,
      "loss": 3.4878,
      "step": 145040
    },
    {
      "epoch": 0.13521007023722703,
      "grad_norm": 0.9388425350189209,
      "learning_rate": 0.00017295819444133726,
      "loss": 3.6004,
      "step": 145050
    },
    {
      "epoch": 0.13521939185530613,
      "grad_norm": 1.44644296169281,
      "learning_rate": 0.00017295633011924207,
      "loss": 3.4241,
      "step": 145060
    },
    {
      "epoch": 0.13522871347338522,
      "grad_norm": 1.0396212339401245,
      "learning_rate": 0.00017295446579714685,
      "loss": 3.6283,
      "step": 145070
    },
    {
      "epoch": 0.1352380350914643,
      "grad_norm": 0.8879124522209167,
      "learning_rate": 0.00017295260147505166,
      "loss": 3.6905,
      "step": 145080
    },
    {
      "epoch": 0.13524735670954338,
      "grad_norm": 1.24770987033844,
      "learning_rate": 0.00017295073715295644,
      "loss": 3.5924,
      "step": 145090
    },
    {
      "epoch": 0.13525667832762248,
      "grad_norm": 1.0322043895721436,
      "learning_rate": 0.00017294887283086124,
      "loss": 3.5543,
      "step": 145100
    },
    {
      "epoch": 0.13526599994570157,
      "grad_norm": 0.9597380757331848,
      "learning_rate": 0.00017294700850876605,
      "loss": 3.6719,
      "step": 145110
    },
    {
      "epoch": 0.13527532156378066,
      "grad_norm": 1.0605207681655884,
      "learning_rate": 0.00017294514418667085,
      "loss": 3.5996,
      "step": 145120
    },
    {
      "epoch": 0.13528464318185976,
      "grad_norm": 0.9156344532966614,
      "learning_rate": 0.00017294327986457566,
      "loss": 3.5461,
      "step": 145130
    },
    {
      "epoch": 0.13529396479993885,
      "grad_norm": 0.9919852614402771,
      "learning_rate": 0.00017294141554248044,
      "loss": 3.574,
      "step": 145140
    },
    {
      "epoch": 0.13530328641801795,
      "grad_norm": 1.1131012439727783,
      "learning_rate": 0.00017293955122038525,
      "loss": 3.6673,
      "step": 145150
    },
    {
      "epoch": 0.13531260803609704,
      "grad_norm": 1.0919467210769653,
      "learning_rate": 0.00017293768689829005,
      "loss": 3.5321,
      "step": 145160
    },
    {
      "epoch": 0.13532192965417614,
      "grad_norm": 1.0806341171264648,
      "learning_rate": 0.00017293582257619486,
      "loss": 3.3928,
      "step": 145170
    },
    {
      "epoch": 0.1353312512722552,
      "grad_norm": 1.1801576614379883,
      "learning_rate": 0.00017293395825409967,
      "loss": 3.4967,
      "step": 145180
    },
    {
      "epoch": 0.1353405728903343,
      "grad_norm": 1.1323981285095215,
      "learning_rate": 0.00017293209393200444,
      "loss": 3.7854,
      "step": 145190
    },
    {
      "epoch": 0.1353498945084134,
      "grad_norm": 0.9678311347961426,
      "learning_rate": 0.00017293022960990925,
      "loss": 3.5849,
      "step": 145200
    },
    {
      "epoch": 0.1353592161264925,
      "grad_norm": 1.0731823444366455,
      "learning_rate": 0.00017292836528781403,
      "loss": 3.7035,
      "step": 145210
    },
    {
      "epoch": 0.13536853774457158,
      "grad_norm": 1.056592345237732,
      "learning_rate": 0.00017292650096571886,
      "loss": 3.5332,
      "step": 145220
    },
    {
      "epoch": 0.13537785936265068,
      "grad_norm": 1.101365566253662,
      "learning_rate": 0.00017292463664362367,
      "loss": 3.6243,
      "step": 145230
    },
    {
      "epoch": 0.13538718098072977,
      "grad_norm": 1.174312949180603,
      "learning_rate": 0.00017292277232152845,
      "loss": 3.6012,
      "step": 145240
    },
    {
      "epoch": 0.13539650259880887,
      "grad_norm": 1.181640863418579,
      "learning_rate": 0.00017292090799943326,
      "loss": 3.7465,
      "step": 145250
    },
    {
      "epoch": 0.13540582421688796,
      "grad_norm": 0.925840437412262,
      "learning_rate": 0.00017291904367733803,
      "loss": 3.8008,
      "step": 145260
    },
    {
      "epoch": 0.13541514583496705,
      "grad_norm": 1.075924277305603,
      "learning_rate": 0.00017291717935524284,
      "loss": 3.5948,
      "step": 145270
    },
    {
      "epoch": 0.13542446745304615,
      "grad_norm": 1.1735082864761353,
      "learning_rate": 0.00017291531503314767,
      "loss": 3.5952,
      "step": 145280
    },
    {
      "epoch": 0.13543378907112522,
      "grad_norm": 1.1682798862457275,
      "learning_rate": 0.00017291345071105245,
      "loss": 3.7069,
      "step": 145290
    },
    {
      "epoch": 0.1354431106892043,
      "grad_norm": 1.1559194326400757,
      "learning_rate": 0.00017291158638895726,
      "loss": 3.601,
      "step": 145300
    },
    {
      "epoch": 0.1354524323072834,
      "grad_norm": 1.042711853981018,
      "learning_rate": 0.00017290972206686204,
      "loss": 3.7006,
      "step": 145310
    },
    {
      "epoch": 0.1354617539253625,
      "grad_norm": 0.9240056872367859,
      "learning_rate": 0.00017290785774476685,
      "loss": 3.5627,
      "step": 145320
    },
    {
      "epoch": 0.1354710755434416,
      "grad_norm": 1.0172677040100098,
      "learning_rate": 0.00017290599342267165,
      "loss": 3.5659,
      "step": 145330
    },
    {
      "epoch": 0.1354803971615207,
      "grad_norm": 0.9099950194358826,
      "learning_rate": 0.00017290412910057646,
      "loss": 3.6047,
      "step": 145340
    },
    {
      "epoch": 0.13548971877959978,
      "grad_norm": 0.8888571262359619,
      "learning_rate": 0.00017290226477848126,
      "loss": 3.5899,
      "step": 145350
    },
    {
      "epoch": 0.13549904039767888,
      "grad_norm": 0.8247804045677185,
      "learning_rate": 0.00017290040045638604,
      "loss": 3.6352,
      "step": 145360
    },
    {
      "epoch": 0.13550836201575797,
      "grad_norm": 0.8765611052513123,
      "learning_rate": 0.00017289853613429085,
      "loss": 3.5598,
      "step": 145370
    },
    {
      "epoch": 0.13551768363383707,
      "grad_norm": 1.0699037313461304,
      "learning_rate": 0.00017289667181219566,
      "loss": 3.6072,
      "step": 145380
    },
    {
      "epoch": 0.13552700525191613,
      "grad_norm": 0.9582090973854065,
      "learning_rate": 0.00017289480749010046,
      "loss": 3.6403,
      "step": 145390
    },
    {
      "epoch": 0.13553632686999523,
      "grad_norm": 0.9964574575424194,
      "learning_rate": 0.00017289294316800527,
      "loss": 3.4338,
      "step": 145400
    },
    {
      "epoch": 0.13554564848807432,
      "grad_norm": 1.1098551750183105,
      "learning_rate": 0.00017289107884591005,
      "loss": 3.6454,
      "step": 145410
    },
    {
      "epoch": 0.13555497010615342,
      "grad_norm": 1.132568359375,
      "learning_rate": 0.00017288921452381485,
      "loss": 3.7529,
      "step": 145420
    },
    {
      "epoch": 0.1355642917242325,
      "grad_norm": 0.9990286827087402,
      "learning_rate": 0.00017288735020171966,
      "loss": 3.6785,
      "step": 145430
    },
    {
      "epoch": 0.1355736133423116,
      "grad_norm": 0.906966507434845,
      "learning_rate": 0.00017288548587962447,
      "loss": 3.6482,
      "step": 145440
    },
    {
      "epoch": 0.1355829349603907,
      "grad_norm": 1.2697185277938843,
      "learning_rate": 0.00017288362155752927,
      "loss": 3.6092,
      "step": 145450
    },
    {
      "epoch": 0.1355922565784698,
      "grad_norm": 1.012699842453003,
      "learning_rate": 0.00017288175723543405,
      "loss": 3.6502,
      "step": 145460
    },
    {
      "epoch": 0.1356015781965489,
      "grad_norm": 1.1635549068450928,
      "learning_rate": 0.00017287989291333886,
      "loss": 3.4389,
      "step": 145470
    },
    {
      "epoch": 0.13561089981462798,
      "grad_norm": 1.2315516471862793,
      "learning_rate": 0.00017287802859124366,
      "loss": 3.4168,
      "step": 145480
    },
    {
      "epoch": 0.13562022143270708,
      "grad_norm": 1.0938276052474976,
      "learning_rate": 0.00017287616426914844,
      "loss": 3.6687,
      "step": 145490
    },
    {
      "epoch": 0.13562954305078614,
      "grad_norm": 1.24359929561615,
      "learning_rate": 0.00017287429994705328,
      "loss": 3.707,
      "step": 145500
    },
    {
      "epoch": 0.13563886466886524,
      "grad_norm": 1.198989748954773,
      "learning_rate": 0.00017287243562495806,
      "loss": 3.6086,
      "step": 145510
    },
    {
      "epoch": 0.13564818628694433,
      "grad_norm": 0.9733096957206726,
      "learning_rate": 0.00017287057130286286,
      "loss": 3.5169,
      "step": 145520
    },
    {
      "epoch": 0.13565750790502343,
      "grad_norm": 1.2729690074920654,
      "learning_rate": 0.00017286870698076767,
      "loss": 3.6394,
      "step": 145530
    },
    {
      "epoch": 0.13566682952310252,
      "grad_norm": 1.1647378206253052,
      "learning_rate": 0.00017286684265867245,
      "loss": 3.6483,
      "step": 145540
    },
    {
      "epoch": 0.13567615114118162,
      "grad_norm": 1.1120896339416504,
      "learning_rate": 0.00017286497833657725,
      "loss": 3.5351,
      "step": 145550
    },
    {
      "epoch": 0.1356854727592607,
      "grad_norm": 1.1794849634170532,
      "learning_rate": 0.00017286311401448206,
      "loss": 3.3597,
      "step": 145560
    },
    {
      "epoch": 0.1356947943773398,
      "grad_norm": 0.986622154712677,
      "learning_rate": 0.00017286124969238687,
      "loss": 3.6168,
      "step": 145570
    },
    {
      "epoch": 0.1357041159954189,
      "grad_norm": 1.0540661811828613,
      "learning_rate": 0.00017285938537029167,
      "loss": 3.504,
      "step": 145580
    },
    {
      "epoch": 0.135713437613498,
      "grad_norm": 1.130255937576294,
      "learning_rate": 0.00017285752104819645,
      "loss": 3.5568,
      "step": 145590
    },
    {
      "epoch": 0.13572275923157706,
      "grad_norm": 1.0922402143478394,
      "learning_rate": 0.00017285565672610126,
      "loss": 3.6066,
      "step": 145600
    },
    {
      "epoch": 0.13573208084965616,
      "grad_norm": 1.0440726280212402,
      "learning_rate": 0.00017285379240400607,
      "loss": 3.7074,
      "step": 145610
    },
    {
      "epoch": 0.13574140246773525,
      "grad_norm": 1.1504302024841309,
      "learning_rate": 0.00017285192808191087,
      "loss": 3.5758,
      "step": 145620
    },
    {
      "epoch": 0.13575072408581434,
      "grad_norm": 0.9480786323547363,
      "learning_rate": 0.00017285006375981568,
      "loss": 3.7309,
      "step": 145630
    },
    {
      "epoch": 0.13576004570389344,
      "grad_norm": 1.0431106090545654,
      "learning_rate": 0.00017284819943772046,
      "loss": 3.5432,
      "step": 145640
    },
    {
      "epoch": 0.13576936732197253,
      "grad_norm": 1.103395700454712,
      "learning_rate": 0.00017284633511562526,
      "loss": 3.5929,
      "step": 145650
    },
    {
      "epoch": 0.13577868894005163,
      "grad_norm": 1.0859774351119995,
      "learning_rate": 0.00017284447079353004,
      "loss": 3.7549,
      "step": 145660
    },
    {
      "epoch": 0.13578801055813072,
      "grad_norm": 1.0780718326568604,
      "learning_rate": 0.00017284260647143488,
      "loss": 3.7597,
      "step": 145670
    },
    {
      "epoch": 0.13579733217620982,
      "grad_norm": 1.0863531827926636,
      "learning_rate": 0.00017284074214933968,
      "loss": 3.6288,
      "step": 145680
    },
    {
      "epoch": 0.1358066537942889,
      "grad_norm": 1.1562492847442627,
      "learning_rate": 0.00017283887782724446,
      "loss": 3.4961,
      "step": 145690
    },
    {
      "epoch": 0.135815975412368,
      "grad_norm": 1.0347890853881836,
      "learning_rate": 0.00017283701350514927,
      "loss": 3.6334,
      "step": 145700
    },
    {
      "epoch": 0.13582529703044707,
      "grad_norm": 0.9684320092201233,
      "learning_rate": 0.00017283514918305405,
      "loss": 3.4399,
      "step": 145710
    },
    {
      "epoch": 0.13583461864852617,
      "grad_norm": 1.0004993677139282,
      "learning_rate": 0.00017283328486095885,
      "loss": 3.6041,
      "step": 145720
    },
    {
      "epoch": 0.13584394026660526,
      "grad_norm": 1.02858567237854,
      "learning_rate": 0.0001728314205388637,
      "loss": 3.4942,
      "step": 145730
    },
    {
      "epoch": 0.13585326188468436,
      "grad_norm": 1.0219987630844116,
      "learning_rate": 0.00017282955621676847,
      "loss": 3.6022,
      "step": 145740
    },
    {
      "epoch": 0.13586258350276345,
      "grad_norm": 0.8224701881408691,
      "learning_rate": 0.00017282769189467327,
      "loss": 3.5148,
      "step": 145750
    },
    {
      "epoch": 0.13587190512084255,
      "grad_norm": 1.0340510606765747,
      "learning_rate": 0.00017282582757257805,
      "loss": 3.5578,
      "step": 145760
    },
    {
      "epoch": 0.13588122673892164,
      "grad_norm": 0.9519146084785461,
      "learning_rate": 0.00017282396325048286,
      "loss": 3.6492,
      "step": 145770
    },
    {
      "epoch": 0.13589054835700073,
      "grad_norm": 0.9832524061203003,
      "learning_rate": 0.00017282209892838766,
      "loss": 3.6384,
      "step": 145780
    },
    {
      "epoch": 0.13589986997507983,
      "grad_norm": 1.2392030954360962,
      "learning_rate": 0.00017282023460629247,
      "loss": 3.6519,
      "step": 145790
    },
    {
      "epoch": 0.13590919159315892,
      "grad_norm": 1.296831488609314,
      "learning_rate": 0.00017281837028419728,
      "loss": 3.5675,
      "step": 145800
    },
    {
      "epoch": 0.135918513211238,
      "grad_norm": 0.996372401714325,
      "learning_rate": 0.00017281650596210206,
      "loss": 3.4565,
      "step": 145810
    },
    {
      "epoch": 0.13592783482931708,
      "grad_norm": 1.0815131664276123,
      "learning_rate": 0.00017281464164000686,
      "loss": 3.725,
      "step": 145820
    },
    {
      "epoch": 0.13593715644739618,
      "grad_norm": 0.9128714799880981,
      "learning_rate": 0.00017281277731791167,
      "loss": 3.6001,
      "step": 145830
    },
    {
      "epoch": 0.13594647806547527,
      "grad_norm": 1.0175737142562866,
      "learning_rate": 0.00017281091299581648,
      "loss": 3.6131,
      "step": 145840
    },
    {
      "epoch": 0.13595579968355437,
      "grad_norm": 1.179065465927124,
      "learning_rate": 0.00017280904867372128,
      "loss": 3.4898,
      "step": 145850
    },
    {
      "epoch": 0.13596512130163346,
      "grad_norm": 1.278598427772522,
      "learning_rate": 0.00017280718435162606,
      "loss": 3.4936,
      "step": 145860
    },
    {
      "epoch": 0.13597444291971256,
      "grad_norm": 0.9636444449424744,
      "learning_rate": 0.00017280532002953087,
      "loss": 3.5788,
      "step": 145870
    },
    {
      "epoch": 0.13598376453779165,
      "grad_norm": 1.1455899477005005,
      "learning_rate": 0.00017280345570743567,
      "loss": 3.6115,
      "step": 145880
    },
    {
      "epoch": 0.13599308615587075,
      "grad_norm": 0.8677094578742981,
      "learning_rate": 0.00017280159138534048,
      "loss": 3.5728,
      "step": 145890
    },
    {
      "epoch": 0.13600240777394984,
      "grad_norm": 1.1984604597091675,
      "learning_rate": 0.00017279972706324529,
      "loss": 3.6449,
      "step": 145900
    },
    {
      "epoch": 0.13601172939202893,
      "grad_norm": 1.0096166133880615,
      "learning_rate": 0.00017279786274115007,
      "loss": 3.4859,
      "step": 145910
    },
    {
      "epoch": 0.136021051010108,
      "grad_norm": 1.1176892518997192,
      "learning_rate": 0.00017279599841905487,
      "loss": 3.5335,
      "step": 145920
    },
    {
      "epoch": 0.1360303726281871,
      "grad_norm": 1.2218505144119263,
      "learning_rate": 0.00017279413409695968,
      "loss": 3.6807,
      "step": 145930
    },
    {
      "epoch": 0.1360396942462662,
      "grad_norm": 1.3255550861358643,
      "learning_rate": 0.00017279226977486446,
      "loss": 3.5694,
      "step": 145940
    },
    {
      "epoch": 0.13604901586434529,
      "grad_norm": 1.1996318101882935,
      "learning_rate": 0.0001727904054527693,
      "loss": 3.5645,
      "step": 145950
    },
    {
      "epoch": 0.13605833748242438,
      "grad_norm": 0.9922583699226379,
      "learning_rate": 0.00017278854113067407,
      "loss": 3.5568,
      "step": 145960
    },
    {
      "epoch": 0.13606765910050347,
      "grad_norm": 1.138143539428711,
      "learning_rate": 0.00017278667680857888,
      "loss": 3.4799,
      "step": 145970
    },
    {
      "epoch": 0.13607698071858257,
      "grad_norm": 1.0491292476654053,
      "learning_rate": 0.00017278481248648368,
      "loss": 3.7073,
      "step": 145980
    },
    {
      "epoch": 0.13608630233666166,
      "grad_norm": 0.9874820709228516,
      "learning_rate": 0.00017278294816438846,
      "loss": 3.6376,
      "step": 145990
    },
    {
      "epoch": 0.13609562395474076,
      "grad_norm": 1.20046067237854,
      "learning_rate": 0.00017278108384229327,
      "loss": 3.3063,
      "step": 146000
    },
    {
      "epoch": 0.13610494557281985,
      "grad_norm": 1.0197668075561523,
      "learning_rate": 0.00017277921952019807,
      "loss": 3.5512,
      "step": 146010
    },
    {
      "epoch": 0.13611426719089892,
      "grad_norm": 1.234427571296692,
      "learning_rate": 0.00017277735519810288,
      "loss": 3.6208,
      "step": 146020
    },
    {
      "epoch": 0.136123588808978,
      "grad_norm": 1.0298857688903809,
      "learning_rate": 0.0001727754908760077,
      "loss": 3.6219,
      "step": 146030
    },
    {
      "epoch": 0.1361329104270571,
      "grad_norm": 0.9032633900642395,
      "learning_rate": 0.00017277362655391247,
      "loss": 3.5889,
      "step": 146040
    },
    {
      "epoch": 0.1361422320451362,
      "grad_norm": 0.9866229295730591,
      "learning_rate": 0.00017277176223181727,
      "loss": 3.6503,
      "step": 146050
    },
    {
      "epoch": 0.1361515536632153,
      "grad_norm": 0.9323477149009705,
      "learning_rate": 0.00017276989790972208,
      "loss": 3.6784,
      "step": 146060
    },
    {
      "epoch": 0.1361608752812944,
      "grad_norm": 0.9868946671485901,
      "learning_rate": 0.00017276803358762688,
      "loss": 3.702,
      "step": 146070
    },
    {
      "epoch": 0.13617019689937349,
      "grad_norm": 1.1795976161956787,
      "learning_rate": 0.0001727661692655317,
      "loss": 3.5786,
      "step": 146080
    },
    {
      "epoch": 0.13617951851745258,
      "grad_norm": 0.9585424065589905,
      "learning_rate": 0.00017276430494343647,
      "loss": 3.617,
      "step": 146090
    },
    {
      "epoch": 0.13618884013553167,
      "grad_norm": 0.8250843286514282,
      "learning_rate": 0.00017276244062134128,
      "loss": 3.6308,
      "step": 146100
    },
    {
      "epoch": 0.13619816175361077,
      "grad_norm": 0.984917163848877,
      "learning_rate": 0.00017276057629924606,
      "loss": 3.4248,
      "step": 146110
    },
    {
      "epoch": 0.13620748337168984,
      "grad_norm": 0.9941263198852539,
      "learning_rate": 0.0001727587119771509,
      "loss": 3.6046,
      "step": 146120
    },
    {
      "epoch": 0.13621680498976893,
      "grad_norm": 1.2463937997817993,
      "learning_rate": 0.0001727568476550557,
      "loss": 3.6906,
      "step": 146130
    },
    {
      "epoch": 0.13622612660784802,
      "grad_norm": 1.1979434490203857,
      "learning_rate": 0.00017275498333296047,
      "loss": 3.6551,
      "step": 146140
    },
    {
      "epoch": 0.13623544822592712,
      "grad_norm": 1.145933985710144,
      "learning_rate": 0.00017275311901086528,
      "loss": 3.5598,
      "step": 146150
    },
    {
      "epoch": 0.1362447698440062,
      "grad_norm": 0.906100869178772,
      "learning_rate": 0.00017275125468877006,
      "loss": 3.6248,
      "step": 146160
    },
    {
      "epoch": 0.1362540914620853,
      "grad_norm": 1.0689244270324707,
      "learning_rate": 0.00017274939036667487,
      "loss": 3.6938,
      "step": 146170
    },
    {
      "epoch": 0.1362634130801644,
      "grad_norm": 1.0085581541061401,
      "learning_rate": 0.0001727475260445797,
      "loss": 3.5946,
      "step": 146180
    },
    {
      "epoch": 0.1362727346982435,
      "grad_norm": 1.2347755432128906,
      "learning_rate": 0.00017274566172248448,
      "loss": 3.6444,
      "step": 146190
    },
    {
      "epoch": 0.1362820563163226,
      "grad_norm": 1.0805895328521729,
      "learning_rate": 0.00017274379740038929,
      "loss": 3.475,
      "step": 146200
    },
    {
      "epoch": 0.1362913779344017,
      "grad_norm": 1.115093469619751,
      "learning_rate": 0.00017274193307829406,
      "loss": 3.6415,
      "step": 146210
    },
    {
      "epoch": 0.13630069955248078,
      "grad_norm": 1.0190309286117554,
      "learning_rate": 0.00017274006875619887,
      "loss": 3.5826,
      "step": 146220
    },
    {
      "epoch": 0.13631002117055985,
      "grad_norm": 1.0398579835891724,
      "learning_rate": 0.00017273820443410368,
      "loss": 3.5896,
      "step": 146230
    },
    {
      "epoch": 0.13631934278863894,
      "grad_norm": 1.0211302042007446,
      "learning_rate": 0.00017273634011200848,
      "loss": 3.5278,
      "step": 146240
    },
    {
      "epoch": 0.13632866440671804,
      "grad_norm": 0.999500572681427,
      "learning_rate": 0.0001727344757899133,
      "loss": 3.7195,
      "step": 146250
    },
    {
      "epoch": 0.13633798602479713,
      "grad_norm": 0.9206175804138184,
      "learning_rate": 0.00017273261146781807,
      "loss": 3.5601,
      "step": 146260
    },
    {
      "epoch": 0.13634730764287623,
      "grad_norm": 1.103027105331421,
      "learning_rate": 0.00017273074714572288,
      "loss": 3.5692,
      "step": 146270
    },
    {
      "epoch": 0.13635662926095532,
      "grad_norm": 1.3333206176757812,
      "learning_rate": 0.00017272888282362768,
      "loss": 3.6196,
      "step": 146280
    },
    {
      "epoch": 0.13636595087903441,
      "grad_norm": 1.0496102571487427,
      "learning_rate": 0.0001727270185015325,
      "loss": 3.6165,
      "step": 146290
    },
    {
      "epoch": 0.1363752724971135,
      "grad_norm": 1.1457127332687378,
      "learning_rate": 0.0001727251541794373,
      "loss": 3.5086,
      "step": 146300
    },
    {
      "epoch": 0.1363845941151926,
      "grad_norm": 1.1366188526153564,
      "learning_rate": 0.00017272328985734207,
      "loss": 3.7739,
      "step": 146310
    },
    {
      "epoch": 0.1363939157332717,
      "grad_norm": 1.0969491004943848,
      "learning_rate": 0.00017272142553524688,
      "loss": 3.7097,
      "step": 146320
    },
    {
      "epoch": 0.13640323735135076,
      "grad_norm": 0.9856149554252625,
      "learning_rate": 0.00017271956121315169,
      "loss": 3.5738,
      "step": 146330
    },
    {
      "epoch": 0.13641255896942986,
      "grad_norm": 0.9706568121910095,
      "learning_rate": 0.0001727176968910565,
      "loss": 3.5696,
      "step": 146340
    },
    {
      "epoch": 0.13642188058750895,
      "grad_norm": 0.9054272174835205,
      "learning_rate": 0.0001727158325689613,
      "loss": 3.438,
      "step": 146350
    },
    {
      "epoch": 0.13643120220558805,
      "grad_norm": 1.2338683605194092,
      "learning_rate": 0.00017271396824686608,
      "loss": 3.5646,
      "step": 146360
    },
    {
      "epoch": 0.13644052382366714,
      "grad_norm": 1.0650485754013062,
      "learning_rate": 0.00017271210392477088,
      "loss": 3.5792,
      "step": 146370
    },
    {
      "epoch": 0.13644984544174624,
      "grad_norm": 1.1034940481185913,
      "learning_rate": 0.0001727102396026757,
      "loss": 3.4943,
      "step": 146380
    },
    {
      "epoch": 0.13645916705982533,
      "grad_norm": 1.1255850791931152,
      "learning_rate": 0.00017270837528058047,
      "loss": 3.6628,
      "step": 146390
    },
    {
      "epoch": 0.13646848867790443,
      "grad_norm": 1.1274875402450562,
      "learning_rate": 0.0001727065109584853,
      "loss": 3.4933,
      "step": 146400
    },
    {
      "epoch": 0.13647781029598352,
      "grad_norm": 1.8263602256774902,
      "learning_rate": 0.00017270464663639008,
      "loss": 3.4909,
      "step": 146410
    },
    {
      "epoch": 0.13648713191406261,
      "grad_norm": 0.9583276510238647,
      "learning_rate": 0.0001727027823142949,
      "loss": 3.4718,
      "step": 146420
    },
    {
      "epoch": 0.1364964535321417,
      "grad_norm": 1.073880910873413,
      "learning_rate": 0.0001727009179921997,
      "loss": 3.5172,
      "step": 146430
    },
    {
      "epoch": 0.13650577515022078,
      "grad_norm": 1.1161683797836304,
      "learning_rate": 0.00017269905367010447,
      "loss": 3.6968,
      "step": 146440
    },
    {
      "epoch": 0.13651509676829987,
      "grad_norm": 1.0338877439498901,
      "learning_rate": 0.00017269718934800928,
      "loss": 3.6421,
      "step": 146450
    },
    {
      "epoch": 0.13652441838637897,
      "grad_norm": 1.0977004766464233,
      "learning_rate": 0.0001726953250259141,
      "loss": 3.4739,
      "step": 146460
    },
    {
      "epoch": 0.13653374000445806,
      "grad_norm": 1.003587007522583,
      "learning_rate": 0.0001726934607038189,
      "loss": 3.6389,
      "step": 146470
    },
    {
      "epoch": 0.13654306162253715,
      "grad_norm": 0.9354004263877869,
      "learning_rate": 0.00017269159638172367,
      "loss": 3.6984,
      "step": 146480
    },
    {
      "epoch": 0.13655238324061625,
      "grad_norm": 1.075196623802185,
      "learning_rate": 0.00017268973205962848,
      "loss": 3.5743,
      "step": 146490
    },
    {
      "epoch": 0.13656170485869534,
      "grad_norm": 0.9744728803634644,
      "learning_rate": 0.00017268786773753328,
      "loss": 3.5162,
      "step": 146500
    },
    {
      "epoch": 0.13657102647677444,
      "grad_norm": 1.1161868572235107,
      "learning_rate": 0.0001726860034154381,
      "loss": 3.705,
      "step": 146510
    },
    {
      "epoch": 0.13658034809485353,
      "grad_norm": 1.4414429664611816,
      "learning_rate": 0.0001726841390933429,
      "loss": 3.6106,
      "step": 146520
    },
    {
      "epoch": 0.13658966971293263,
      "grad_norm": 0.9724092483520508,
      "learning_rate": 0.00017268227477124768,
      "loss": 3.5716,
      "step": 146530
    },
    {
      "epoch": 0.1365989913310117,
      "grad_norm": 0.8910855650901794,
      "learning_rate": 0.00017268041044915248,
      "loss": 3.6256,
      "step": 146540
    },
    {
      "epoch": 0.1366083129490908,
      "grad_norm": 1.0994374752044678,
      "learning_rate": 0.0001726785461270573,
      "loss": 3.6058,
      "step": 146550
    },
    {
      "epoch": 0.13661763456716988,
      "grad_norm": 1.073480486869812,
      "learning_rate": 0.00017267668180496207,
      "loss": 3.7389,
      "step": 146560
    },
    {
      "epoch": 0.13662695618524898,
      "grad_norm": 0.9445319175720215,
      "learning_rate": 0.0001726748174828669,
      "loss": 3.457,
      "step": 146570
    },
    {
      "epoch": 0.13663627780332807,
      "grad_norm": 1.276234745979309,
      "learning_rate": 0.00017267295316077168,
      "loss": 3.8455,
      "step": 146580
    },
    {
      "epoch": 0.13664559942140717,
      "grad_norm": 1.1034103631973267,
      "learning_rate": 0.0001726710888386765,
      "loss": 3.4179,
      "step": 146590
    },
    {
      "epoch": 0.13665492103948626,
      "grad_norm": 0.9936701059341431,
      "learning_rate": 0.0001726692245165813,
      "loss": 3.3088,
      "step": 146600
    },
    {
      "epoch": 0.13666424265756535,
      "grad_norm": 0.8594498038291931,
      "learning_rate": 0.00017266736019448607,
      "loss": 3.6677,
      "step": 146610
    },
    {
      "epoch": 0.13667356427564445,
      "grad_norm": 1.0523873567581177,
      "learning_rate": 0.00017266549587239088,
      "loss": 3.6115,
      "step": 146620
    },
    {
      "epoch": 0.13668288589372354,
      "grad_norm": 1.085530161857605,
      "learning_rate": 0.00017266363155029569,
      "loss": 3.7555,
      "step": 146630
    },
    {
      "epoch": 0.13669220751180264,
      "grad_norm": 1.073127031326294,
      "learning_rate": 0.0001726617672282005,
      "loss": 3.6534,
      "step": 146640
    },
    {
      "epoch": 0.1367015291298817,
      "grad_norm": 1.0418736934661865,
      "learning_rate": 0.0001726599029061053,
      "loss": 3.595,
      "step": 146650
    },
    {
      "epoch": 0.1367108507479608,
      "grad_norm": 0.923348069190979,
      "learning_rate": 0.00017265803858401008,
      "loss": 3.495,
      "step": 146660
    },
    {
      "epoch": 0.1367201723660399,
      "grad_norm": 0.9419720768928528,
      "learning_rate": 0.00017265617426191488,
      "loss": 3.6168,
      "step": 146670
    },
    {
      "epoch": 0.136729493984119,
      "grad_norm": 1.0508053302764893,
      "learning_rate": 0.0001726543099398197,
      "loss": 3.675,
      "step": 146680
    },
    {
      "epoch": 0.13673881560219808,
      "grad_norm": 0.979824960231781,
      "learning_rate": 0.0001726524456177245,
      "loss": 3.4557,
      "step": 146690
    },
    {
      "epoch": 0.13674813722027718,
      "grad_norm": 1.1061522960662842,
      "learning_rate": 0.0001726505812956293,
      "loss": 3.4522,
      "step": 146700
    },
    {
      "epoch": 0.13675745883835627,
      "grad_norm": 1.3277697563171387,
      "learning_rate": 0.00017264871697353408,
      "loss": 3.5179,
      "step": 146710
    },
    {
      "epoch": 0.13676678045643537,
      "grad_norm": 1.0922677516937256,
      "learning_rate": 0.0001726468526514389,
      "loss": 3.6528,
      "step": 146720
    },
    {
      "epoch": 0.13677610207451446,
      "grad_norm": 0.8409761190414429,
      "learning_rate": 0.0001726449883293437,
      "loss": 3.4647,
      "step": 146730
    },
    {
      "epoch": 0.13678542369259356,
      "grad_norm": 0.9671738147735596,
      "learning_rate": 0.0001726431240072485,
      "loss": 3.6699,
      "step": 146740
    },
    {
      "epoch": 0.13679474531067262,
      "grad_norm": 1.0160059928894043,
      "learning_rate": 0.0001726412596851533,
      "loss": 3.6731,
      "step": 146750
    },
    {
      "epoch": 0.13680406692875172,
      "grad_norm": 1.1770066022872925,
      "learning_rate": 0.00017263939536305809,
      "loss": 3.4844,
      "step": 146760
    },
    {
      "epoch": 0.1368133885468308,
      "grad_norm": 1.2121371030807495,
      "learning_rate": 0.0001726375310409629,
      "loss": 3.5262,
      "step": 146770
    },
    {
      "epoch": 0.1368227101649099,
      "grad_norm": 1.1058146953582764,
      "learning_rate": 0.00017263566671886767,
      "loss": 3.6026,
      "step": 146780
    },
    {
      "epoch": 0.136832031782989,
      "grad_norm": 1.4344578981399536,
      "learning_rate": 0.0001726338023967725,
      "loss": 3.7233,
      "step": 146790
    },
    {
      "epoch": 0.1368413534010681,
      "grad_norm": 1.0038748979568481,
      "learning_rate": 0.0001726319380746773,
      "loss": 3.7102,
      "step": 146800
    },
    {
      "epoch": 0.1368506750191472,
      "grad_norm": 1.0938806533813477,
      "learning_rate": 0.0001726300737525821,
      "loss": 3.7417,
      "step": 146810
    },
    {
      "epoch": 0.13685999663722628,
      "grad_norm": 1.0459229946136475,
      "learning_rate": 0.0001726282094304869,
      "loss": 3.6284,
      "step": 146820
    },
    {
      "epoch": 0.13686931825530538,
      "grad_norm": 1.0349721908569336,
      "learning_rate": 0.00017262634510839168,
      "loss": 3.5122,
      "step": 146830
    },
    {
      "epoch": 0.13687863987338447,
      "grad_norm": 0.9590821266174316,
      "learning_rate": 0.00017262448078629648,
      "loss": 3.6678,
      "step": 146840
    },
    {
      "epoch": 0.13688796149146357,
      "grad_norm": 1.20621919631958,
      "learning_rate": 0.00017262261646420132,
      "loss": 3.6595,
      "step": 146850
    },
    {
      "epoch": 0.13689728310954263,
      "grad_norm": 0.9682687520980835,
      "learning_rate": 0.0001726207521421061,
      "loss": 3.6993,
      "step": 146860
    },
    {
      "epoch": 0.13690660472762173,
      "grad_norm": 1.1005195379257202,
      "learning_rate": 0.0001726188878200109,
      "loss": 3.5419,
      "step": 146870
    },
    {
      "epoch": 0.13691592634570082,
      "grad_norm": 0.9586533904075623,
      "learning_rate": 0.00017261702349791568,
      "loss": 3.6373,
      "step": 146880
    },
    {
      "epoch": 0.13692524796377992,
      "grad_norm": 0.9189508557319641,
      "learning_rate": 0.0001726151591758205,
      "loss": 3.4562,
      "step": 146890
    },
    {
      "epoch": 0.136934569581859,
      "grad_norm": 1.0272642374038696,
      "learning_rate": 0.0001726132948537253,
      "loss": 3.5269,
      "step": 146900
    },
    {
      "epoch": 0.1369438911999381,
      "grad_norm": 0.8765383958816528,
      "learning_rate": 0.0001726114305316301,
      "loss": 3.593,
      "step": 146910
    },
    {
      "epoch": 0.1369532128180172,
      "grad_norm": 0.9582970142364502,
      "learning_rate": 0.0001726095662095349,
      "loss": 3.5908,
      "step": 146920
    },
    {
      "epoch": 0.1369625344360963,
      "grad_norm": 1.136827826499939,
      "learning_rate": 0.00017260770188743969,
      "loss": 3.54,
      "step": 146930
    },
    {
      "epoch": 0.1369718560541754,
      "grad_norm": 1.1889958381652832,
      "learning_rate": 0.0001726058375653445,
      "loss": 3.4717,
      "step": 146940
    },
    {
      "epoch": 0.13698117767225448,
      "grad_norm": 1.0339916944503784,
      "learning_rate": 0.0001726039732432493,
      "loss": 3.7072,
      "step": 146950
    },
    {
      "epoch": 0.13699049929033355,
      "grad_norm": 1.114817500114441,
      "learning_rate": 0.0001726021089211541,
      "loss": 3.5999,
      "step": 146960
    },
    {
      "epoch": 0.13699982090841265,
      "grad_norm": 1.0310617685317993,
      "learning_rate": 0.0001726002445990589,
      "loss": 3.6549,
      "step": 146970
    },
    {
      "epoch": 0.13700914252649174,
      "grad_norm": 1.1349269151687622,
      "learning_rate": 0.0001725983802769637,
      "loss": 3.7188,
      "step": 146980
    },
    {
      "epoch": 0.13701846414457083,
      "grad_norm": 1.1243655681610107,
      "learning_rate": 0.0001725965159548685,
      "loss": 3.5232,
      "step": 146990
    },
    {
      "epoch": 0.13702778576264993,
      "grad_norm": 1.0625293254852295,
      "learning_rate": 0.0001725946516327733,
      "loss": 3.5655,
      "step": 147000
    },
    {
      "epoch": 0.13703710738072902,
      "grad_norm": 0.90560382604599,
      "learning_rate": 0.00017259278731067808,
      "loss": 3.6907,
      "step": 147010
    },
    {
      "epoch": 0.13704642899880812,
      "grad_norm": 0.9396039843559265,
      "learning_rate": 0.00017259092298858291,
      "loss": 3.5089,
      "step": 147020
    },
    {
      "epoch": 0.1370557506168872,
      "grad_norm": 1.0859724283218384,
      "learning_rate": 0.0001725890586664877,
      "loss": 3.7092,
      "step": 147030
    },
    {
      "epoch": 0.1370650722349663,
      "grad_norm": 1.0914793014526367,
      "learning_rate": 0.0001725871943443925,
      "loss": 3.5828,
      "step": 147040
    },
    {
      "epoch": 0.1370743938530454,
      "grad_norm": 1.0526591539382935,
      "learning_rate": 0.0001725853300222973,
      "loss": 3.5648,
      "step": 147050
    },
    {
      "epoch": 0.1370837154711245,
      "grad_norm": 0.951218843460083,
      "learning_rate": 0.00017258346570020209,
      "loss": 3.6324,
      "step": 147060
    },
    {
      "epoch": 0.13709303708920356,
      "grad_norm": 1.1151467561721802,
      "learning_rate": 0.0001725816013781069,
      "loss": 3.6081,
      "step": 147070
    },
    {
      "epoch": 0.13710235870728266,
      "grad_norm": 0.9799051284790039,
      "learning_rate": 0.0001725797370560117,
      "loss": 3.768,
      "step": 147080
    },
    {
      "epoch": 0.13711168032536175,
      "grad_norm": 1.1616618633270264,
      "learning_rate": 0.0001725778727339165,
      "loss": 3.6518,
      "step": 147090
    },
    {
      "epoch": 0.13712100194344085,
      "grad_norm": 1.011771321296692,
      "learning_rate": 0.0001725760084118213,
      "loss": 3.4811,
      "step": 147100
    },
    {
      "epoch": 0.13713032356151994,
      "grad_norm": 0.9826416969299316,
      "learning_rate": 0.0001725741440897261,
      "loss": 3.6694,
      "step": 147110
    },
    {
      "epoch": 0.13713964517959903,
      "grad_norm": 0.8428415060043335,
      "learning_rate": 0.0001725722797676309,
      "loss": 3.5873,
      "step": 147120
    },
    {
      "epoch": 0.13714896679767813,
      "grad_norm": 1.022223949432373,
      "learning_rate": 0.0001725704154455357,
      "loss": 3.4577,
      "step": 147130
    },
    {
      "epoch": 0.13715828841575722,
      "grad_norm": 0.9176100492477417,
      "learning_rate": 0.0001725685511234405,
      "loss": 3.6644,
      "step": 147140
    },
    {
      "epoch": 0.13716761003383632,
      "grad_norm": 1.0048102140426636,
      "learning_rate": 0.00017256668680134532,
      "loss": 3.5609,
      "step": 147150
    },
    {
      "epoch": 0.1371769316519154,
      "grad_norm": 1.0199832916259766,
      "learning_rate": 0.0001725648224792501,
      "loss": 3.4646,
      "step": 147160
    },
    {
      "epoch": 0.13718625326999448,
      "grad_norm": 1.1620674133300781,
      "learning_rate": 0.0001725629581571549,
      "loss": 3.5994,
      "step": 147170
    },
    {
      "epoch": 0.13719557488807357,
      "grad_norm": 0.9806487560272217,
      "learning_rate": 0.0001725610938350597,
      "loss": 3.553,
      "step": 147180
    },
    {
      "epoch": 0.13720489650615267,
      "grad_norm": 1.2641576528549194,
      "learning_rate": 0.00017255922951296451,
      "loss": 3.5066,
      "step": 147190
    },
    {
      "epoch": 0.13721421812423176,
      "grad_norm": 1.0570704936981201,
      "learning_rate": 0.00017255736519086932,
      "loss": 3.7602,
      "step": 147200
    },
    {
      "epoch": 0.13722353974231086,
      "grad_norm": 1.002275824546814,
      "learning_rate": 0.0001725555008687741,
      "loss": 3.5373,
      "step": 147210
    },
    {
      "epoch": 0.13723286136038995,
      "grad_norm": 0.9936522841453552,
      "learning_rate": 0.0001725536365466789,
      "loss": 3.6526,
      "step": 147220
    },
    {
      "epoch": 0.13724218297846905,
      "grad_norm": 1.1370528936386108,
      "learning_rate": 0.00017255177222458368,
      "loss": 3.551,
      "step": 147230
    },
    {
      "epoch": 0.13725150459654814,
      "grad_norm": 1.1784509420394897,
      "learning_rate": 0.00017254990790248852,
      "loss": 3.816,
      "step": 147240
    },
    {
      "epoch": 0.13726082621462724,
      "grad_norm": 1.028337001800537,
      "learning_rate": 0.00017254804358039332,
      "loss": 3.6703,
      "step": 147250
    },
    {
      "epoch": 0.13727014783270633,
      "grad_norm": 1.0066019296646118,
      "learning_rate": 0.0001725461792582981,
      "loss": 3.5473,
      "step": 147260
    },
    {
      "epoch": 0.13727946945078542,
      "grad_norm": 0.9720441102981567,
      "learning_rate": 0.0001725443149362029,
      "loss": 3.5783,
      "step": 147270
    },
    {
      "epoch": 0.1372887910688645,
      "grad_norm": 1.1864949464797974,
      "learning_rate": 0.0001725424506141077,
      "loss": 3.6899,
      "step": 147280
    },
    {
      "epoch": 0.13729811268694359,
      "grad_norm": 1.048082947731018,
      "learning_rate": 0.0001725405862920125,
      "loss": 3.8128,
      "step": 147290
    },
    {
      "epoch": 0.13730743430502268,
      "grad_norm": 1.0603480339050293,
      "learning_rate": 0.00017253872196991733,
      "loss": 3.5693,
      "step": 147300
    },
    {
      "epoch": 0.13731675592310177,
      "grad_norm": 1.0813325643539429,
      "learning_rate": 0.0001725368576478221,
      "loss": 3.4997,
      "step": 147310
    },
    {
      "epoch": 0.13732607754118087,
      "grad_norm": 1.12544584274292,
      "learning_rate": 0.00017253499332572691,
      "loss": 3.7274,
      "step": 147320
    },
    {
      "epoch": 0.13733539915925996,
      "grad_norm": 1.1057053804397583,
      "learning_rate": 0.0001725331290036317,
      "loss": 3.598,
      "step": 147330
    },
    {
      "epoch": 0.13734472077733906,
      "grad_norm": 0.9703680276870728,
      "learning_rate": 0.0001725312646815365,
      "loss": 3.7295,
      "step": 147340
    },
    {
      "epoch": 0.13735404239541815,
      "grad_norm": 0.8651547431945801,
      "learning_rate": 0.0001725294003594413,
      "loss": 3.6895,
      "step": 147350
    },
    {
      "epoch": 0.13736336401349725,
      "grad_norm": 0.9066580533981323,
      "learning_rate": 0.0001725275360373461,
      "loss": 3.5615,
      "step": 147360
    },
    {
      "epoch": 0.13737268563157634,
      "grad_norm": 1.09152352809906,
      "learning_rate": 0.00017252567171525092,
      "loss": 3.6004,
      "step": 147370
    },
    {
      "epoch": 0.1373820072496554,
      "grad_norm": 1.1599133014678955,
      "learning_rate": 0.0001725238073931557,
      "loss": 3.6521,
      "step": 147380
    },
    {
      "epoch": 0.1373913288677345,
      "grad_norm": 1.025559425354004,
      "learning_rate": 0.0001725219430710605,
      "loss": 3.7367,
      "step": 147390
    },
    {
      "epoch": 0.1374006504858136,
      "grad_norm": 0.9779433608055115,
      "learning_rate": 0.0001725200787489653,
      "loss": 3.5048,
      "step": 147400
    },
    {
      "epoch": 0.1374099721038927,
      "grad_norm": 0.9913411736488342,
      "learning_rate": 0.00017251821442687012,
      "loss": 3.4766,
      "step": 147410
    },
    {
      "epoch": 0.13741929372197179,
      "grad_norm": 0.9421737790107727,
      "learning_rate": 0.00017251635010477492,
      "loss": 3.6264,
      "step": 147420
    },
    {
      "epoch": 0.13742861534005088,
      "grad_norm": 1.1606827974319458,
      "learning_rate": 0.0001725144857826797,
      "loss": 3.5421,
      "step": 147430
    },
    {
      "epoch": 0.13743793695812997,
      "grad_norm": 1.0729470252990723,
      "learning_rate": 0.0001725126214605845,
      "loss": 3.6935,
      "step": 147440
    },
    {
      "epoch": 0.13744725857620907,
      "grad_norm": 1.1006433963775635,
      "learning_rate": 0.00017251075713848932,
      "loss": 3.6278,
      "step": 147450
    },
    {
      "epoch": 0.13745658019428816,
      "grad_norm": 0.9942697286605835,
      "learning_rate": 0.0001725088928163941,
      "loss": 3.5566,
      "step": 147460
    },
    {
      "epoch": 0.13746590181236726,
      "grad_norm": 0.9987589716911316,
      "learning_rate": 0.00017250702849429893,
      "loss": 3.5915,
      "step": 147470
    },
    {
      "epoch": 0.13747522343044635,
      "grad_norm": 0.9675213098526001,
      "learning_rate": 0.0001725051641722037,
      "loss": 3.6043,
      "step": 147480
    },
    {
      "epoch": 0.13748454504852542,
      "grad_norm": 0.9725551605224609,
      "learning_rate": 0.0001725032998501085,
      "loss": 3.4388,
      "step": 147490
    },
    {
      "epoch": 0.13749386666660451,
      "grad_norm": 0.926898181438446,
      "learning_rate": 0.00017250143552801332,
      "loss": 3.6017,
      "step": 147500
    },
    {
      "epoch": 0.1375031882846836,
      "grad_norm": 1.1067382097244263,
      "learning_rate": 0.0001724995712059181,
      "loss": 3.6413,
      "step": 147510
    },
    {
      "epoch": 0.1375125099027627,
      "grad_norm": 0.8723544478416443,
      "learning_rate": 0.0001724977068838229,
      "loss": 3.5475,
      "step": 147520
    },
    {
      "epoch": 0.1375218315208418,
      "grad_norm": 1.1488865613937378,
      "learning_rate": 0.0001724958425617277,
      "loss": 3.5163,
      "step": 147530
    },
    {
      "epoch": 0.1375311531389209,
      "grad_norm": 0.7626347541809082,
      "learning_rate": 0.00017249397823963252,
      "loss": 3.7007,
      "step": 147540
    },
    {
      "epoch": 0.137540474757,
      "grad_norm": 1.0938538312911987,
      "learning_rate": 0.00017249211391753732,
      "loss": 3.5808,
      "step": 147550
    },
    {
      "epoch": 0.13754979637507908,
      "grad_norm": 1.0066038370132446,
      "learning_rate": 0.0001724902495954421,
      "loss": 3.6366,
      "step": 147560
    },
    {
      "epoch": 0.13755911799315818,
      "grad_norm": 1.385973334312439,
      "learning_rate": 0.0001724883852733469,
      "loss": 3.527,
      "step": 147570
    },
    {
      "epoch": 0.13756843961123727,
      "grad_norm": 1.1022825241088867,
      "learning_rate": 0.00017248652095125172,
      "loss": 3.5358,
      "step": 147580
    },
    {
      "epoch": 0.13757776122931634,
      "grad_norm": 1.3261289596557617,
      "learning_rate": 0.00017248465662915652,
      "loss": 3.5498,
      "step": 147590
    },
    {
      "epoch": 0.13758708284739543,
      "grad_norm": 1.3039348125457764,
      "learning_rate": 0.00017248279230706133,
      "loss": 3.4694,
      "step": 147600
    },
    {
      "epoch": 0.13759640446547453,
      "grad_norm": 1.485326886177063,
      "learning_rate": 0.0001724809279849661,
      "loss": 3.644,
      "step": 147610
    },
    {
      "epoch": 0.13760572608355362,
      "grad_norm": 1.024483323097229,
      "learning_rate": 0.00017247906366287091,
      "loss": 3.4857,
      "step": 147620
    },
    {
      "epoch": 0.13761504770163271,
      "grad_norm": 1.0893845558166504,
      "learning_rate": 0.00017247719934077572,
      "loss": 3.5466,
      "step": 147630
    },
    {
      "epoch": 0.1376243693197118,
      "grad_norm": 0.9438443779945374,
      "learning_rate": 0.00017247533501868053,
      "loss": 3.4896,
      "step": 147640
    },
    {
      "epoch": 0.1376336909377909,
      "grad_norm": 0.981330931186676,
      "learning_rate": 0.00017247347069658533,
      "loss": 3.6188,
      "step": 147650
    },
    {
      "epoch": 0.13764301255587,
      "grad_norm": 0.9481096863746643,
      "learning_rate": 0.0001724716063744901,
      "loss": 3.7327,
      "step": 147660
    },
    {
      "epoch": 0.1376523341739491,
      "grad_norm": 1.3207359313964844,
      "learning_rate": 0.00017246974205239492,
      "loss": 3.5557,
      "step": 147670
    },
    {
      "epoch": 0.1376616557920282,
      "grad_norm": 1.3957836627960205,
      "learning_rate": 0.0001724678777302997,
      "loss": 3.6119,
      "step": 147680
    },
    {
      "epoch": 0.13767097741010725,
      "grad_norm": 1.0906955003738403,
      "learning_rate": 0.00017246601340820453,
      "loss": 3.7615,
      "step": 147690
    },
    {
      "epoch": 0.13768029902818635,
      "grad_norm": 1.0697001218795776,
      "learning_rate": 0.00017246414908610934,
      "loss": 3.4847,
      "step": 147700
    },
    {
      "epoch": 0.13768962064626544,
      "grad_norm": 0.9624161720275879,
      "learning_rate": 0.00017246228476401412,
      "loss": 3.567,
      "step": 147710
    },
    {
      "epoch": 0.13769894226434454,
      "grad_norm": 1.0946813821792603,
      "learning_rate": 0.00017246042044191892,
      "loss": 3.5847,
      "step": 147720
    },
    {
      "epoch": 0.13770826388242363,
      "grad_norm": 1.0366222858428955,
      "learning_rate": 0.0001724585561198237,
      "loss": 3.6369,
      "step": 147730
    },
    {
      "epoch": 0.13771758550050273,
      "grad_norm": 1.1981475353240967,
      "learning_rate": 0.0001724566917977285,
      "loss": 3.5678,
      "step": 147740
    },
    {
      "epoch": 0.13772690711858182,
      "grad_norm": 1.196064829826355,
      "learning_rate": 0.00017245482747563334,
      "loss": 3.6858,
      "step": 147750
    },
    {
      "epoch": 0.13773622873666092,
      "grad_norm": 1.170128583908081,
      "learning_rate": 0.00017245296315353812,
      "loss": 3.6533,
      "step": 147760
    },
    {
      "epoch": 0.13774555035474,
      "grad_norm": 1.123289942741394,
      "learning_rate": 0.00017245109883144293,
      "loss": 3.6123,
      "step": 147770
    },
    {
      "epoch": 0.1377548719728191,
      "grad_norm": 1.12388014793396,
      "learning_rate": 0.0001724492345093477,
      "loss": 3.7369,
      "step": 147780
    },
    {
      "epoch": 0.1377641935908982,
      "grad_norm": 1.0346081256866455,
      "learning_rate": 0.0001724473701872525,
      "loss": 3.4265,
      "step": 147790
    },
    {
      "epoch": 0.13777351520897727,
      "grad_norm": 1.0211502313613892,
      "learning_rate": 0.00017244550586515732,
      "loss": 3.6324,
      "step": 147800
    },
    {
      "epoch": 0.13778283682705636,
      "grad_norm": 0.9460985660552979,
      "learning_rate": 0.00017244364154306213,
      "loss": 3.5263,
      "step": 147810
    },
    {
      "epoch": 0.13779215844513545,
      "grad_norm": 0.9178708791732788,
      "learning_rate": 0.00017244177722096693,
      "loss": 3.6854,
      "step": 147820
    },
    {
      "epoch": 0.13780148006321455,
      "grad_norm": 1.0770241022109985,
      "learning_rate": 0.0001724399128988717,
      "loss": 3.6541,
      "step": 147830
    },
    {
      "epoch": 0.13781080168129364,
      "grad_norm": 0.880582869052887,
      "learning_rate": 0.00017243804857677652,
      "loss": 3.4474,
      "step": 147840
    },
    {
      "epoch": 0.13782012329937274,
      "grad_norm": 1.2854584455490112,
      "learning_rate": 0.00017243618425468132,
      "loss": 3.5935,
      "step": 147850
    },
    {
      "epoch": 0.13782944491745183,
      "grad_norm": 1.0774258375167847,
      "learning_rate": 0.00017243431993258613,
      "loss": 3.7246,
      "step": 147860
    },
    {
      "epoch": 0.13783876653553093,
      "grad_norm": 1.1611998081207275,
      "learning_rate": 0.00017243245561049094,
      "loss": 3.4907,
      "step": 147870
    },
    {
      "epoch": 0.13784808815361002,
      "grad_norm": 1.0547311305999756,
      "learning_rate": 0.00017243059128839572,
      "loss": 3.6857,
      "step": 147880
    },
    {
      "epoch": 0.13785740977168912,
      "grad_norm": 1.1127214431762695,
      "learning_rate": 0.00017242872696630052,
      "loss": 3.5891,
      "step": 147890
    },
    {
      "epoch": 0.13786673138976818,
      "grad_norm": 1.1551470756530762,
      "learning_rate": 0.0001724268626442053,
      "loss": 3.8403,
      "step": 147900
    },
    {
      "epoch": 0.13787605300784728,
      "grad_norm": 1.0656911134719849,
      "learning_rate": 0.0001724249983221101,
      "loss": 3.5625,
      "step": 147910
    },
    {
      "epoch": 0.13788537462592637,
      "grad_norm": 0.8754634857177734,
      "learning_rate": 0.00017242313400001494,
      "loss": 3.6884,
      "step": 147920
    },
    {
      "epoch": 0.13789469624400547,
      "grad_norm": 0.9326872229576111,
      "learning_rate": 0.00017242126967791972,
      "loss": 3.6476,
      "step": 147930
    },
    {
      "epoch": 0.13790401786208456,
      "grad_norm": 0.9190135598182678,
      "learning_rate": 0.00017241940535582453,
      "loss": 3.5878,
      "step": 147940
    },
    {
      "epoch": 0.13791333948016365,
      "grad_norm": 1.1958801746368408,
      "learning_rate": 0.0001724175410337293,
      "loss": 3.5534,
      "step": 147950
    },
    {
      "epoch": 0.13792266109824275,
      "grad_norm": 0.9875401854515076,
      "learning_rate": 0.0001724156767116341,
      "loss": 3.6079,
      "step": 147960
    },
    {
      "epoch": 0.13793198271632184,
      "grad_norm": 1.0303606986999512,
      "learning_rate": 0.00017241381238953892,
      "loss": 3.6042,
      "step": 147970
    },
    {
      "epoch": 0.13794130433440094,
      "grad_norm": 1.0936462879180908,
      "learning_rate": 0.00017241194806744372,
      "loss": 3.6003,
      "step": 147980
    },
    {
      "epoch": 0.13795062595248003,
      "grad_norm": 1.2570513486862183,
      "learning_rate": 0.00017241008374534853,
      "loss": 3.4905,
      "step": 147990
    },
    {
      "epoch": 0.13795994757055913,
      "grad_norm": 0.8215647339820862,
      "learning_rate": 0.0001724082194232533,
      "loss": 3.7297,
      "step": 148000
    },
    {
      "epoch": 0.1379692691886382,
      "grad_norm": 0.9351413249969482,
      "learning_rate": 0.00017240635510115812,
      "loss": 3.663,
      "step": 148010
    },
    {
      "epoch": 0.1379785908067173,
      "grad_norm": 1.1077933311462402,
      "learning_rate": 0.00017240449077906292,
      "loss": 3.5545,
      "step": 148020
    },
    {
      "epoch": 0.13798791242479638,
      "grad_norm": 1.1294695138931274,
      "learning_rate": 0.00017240262645696773,
      "loss": 3.6839,
      "step": 148030
    },
    {
      "epoch": 0.13799723404287548,
      "grad_norm": 1.0304242372512817,
      "learning_rate": 0.00017240076213487253,
      "loss": 3.7222,
      "step": 148040
    },
    {
      "epoch": 0.13800655566095457,
      "grad_norm": 1.1316134929656982,
      "learning_rate": 0.00017239889781277731,
      "loss": 3.3971,
      "step": 148050
    },
    {
      "epoch": 0.13801587727903367,
      "grad_norm": 1.0966814756393433,
      "learning_rate": 0.00017239703349068212,
      "loss": 3.6004,
      "step": 148060
    },
    {
      "epoch": 0.13802519889711276,
      "grad_norm": 1.2245302200317383,
      "learning_rate": 0.00017239516916858693,
      "loss": 3.6644,
      "step": 148070
    },
    {
      "epoch": 0.13803452051519186,
      "grad_norm": 1.2425521612167358,
      "learning_rate": 0.00017239330484649173,
      "loss": 3.5705,
      "step": 148080
    },
    {
      "epoch": 0.13804384213327095,
      "grad_norm": 1.069164514541626,
      "learning_rate": 0.00017239144052439654,
      "loss": 3.6131,
      "step": 148090
    },
    {
      "epoch": 0.13805316375135004,
      "grad_norm": 2.4751009941101074,
      "learning_rate": 0.00017238957620230132,
      "loss": 3.4819,
      "step": 148100
    },
    {
      "epoch": 0.1380624853694291,
      "grad_norm": 1.0626919269561768,
      "learning_rate": 0.00017238771188020612,
      "loss": 3.5026,
      "step": 148110
    },
    {
      "epoch": 0.1380718069875082,
      "grad_norm": 1.1740086078643799,
      "learning_rate": 0.00017238584755811093,
      "loss": 3.5001,
      "step": 148120
    },
    {
      "epoch": 0.1380811286055873,
      "grad_norm": 1.0698212385177612,
      "learning_rate": 0.0001723839832360157,
      "loss": 3.4297,
      "step": 148130
    },
    {
      "epoch": 0.1380904502236664,
      "grad_norm": 0.9386858344078064,
      "learning_rate": 0.00017238211891392054,
      "loss": 3.6153,
      "step": 148140
    },
    {
      "epoch": 0.1380997718417455,
      "grad_norm": 1.1424319744110107,
      "learning_rate": 0.00017238025459182532,
      "loss": 3.5912,
      "step": 148150
    },
    {
      "epoch": 0.13810909345982458,
      "grad_norm": 1.1767220497131348,
      "learning_rate": 0.00017237839026973013,
      "loss": 3.6375,
      "step": 148160
    },
    {
      "epoch": 0.13811841507790368,
      "grad_norm": 0.8911251425743103,
      "learning_rate": 0.00017237652594763494,
      "loss": 3.7466,
      "step": 148170
    },
    {
      "epoch": 0.13812773669598277,
      "grad_norm": 0.9560375213623047,
      "learning_rate": 0.00017237466162553971,
      "loss": 3.5688,
      "step": 148180
    },
    {
      "epoch": 0.13813705831406187,
      "grad_norm": 1.0164648294448853,
      "learning_rate": 0.00017237279730344452,
      "loss": 3.5642,
      "step": 148190
    },
    {
      "epoch": 0.13814637993214096,
      "grad_norm": 1.003562092781067,
      "learning_rate": 0.00017237093298134933,
      "loss": 3.4905,
      "step": 148200
    },
    {
      "epoch": 0.13815570155022006,
      "grad_norm": 1.117732048034668,
      "learning_rate": 0.00017236906865925413,
      "loss": 3.469,
      "step": 148210
    },
    {
      "epoch": 0.13816502316829912,
      "grad_norm": 0.8804612159729004,
      "learning_rate": 0.00017236720433715894,
      "loss": 3.8333,
      "step": 148220
    },
    {
      "epoch": 0.13817434478637822,
      "grad_norm": 1.1503679752349854,
      "learning_rate": 0.00017236534001506372,
      "loss": 3.6607,
      "step": 148230
    },
    {
      "epoch": 0.1381836664044573,
      "grad_norm": 1.0828877687454224,
      "learning_rate": 0.00017236347569296853,
      "loss": 3.5842,
      "step": 148240
    },
    {
      "epoch": 0.1381929880225364,
      "grad_norm": 1.1273027658462524,
      "learning_rate": 0.00017236161137087333,
      "loss": 3.5644,
      "step": 148250
    },
    {
      "epoch": 0.1382023096406155,
      "grad_norm": 1.0351396799087524,
      "learning_rate": 0.00017235974704877814,
      "loss": 3.5819,
      "step": 148260
    },
    {
      "epoch": 0.1382116312586946,
      "grad_norm": 1.1467938423156738,
      "learning_rate": 0.00017235788272668294,
      "loss": 3.511,
      "step": 148270
    },
    {
      "epoch": 0.1382209528767737,
      "grad_norm": 0.9656757712364197,
      "learning_rate": 0.00017235601840458772,
      "loss": 3.6555,
      "step": 148280
    },
    {
      "epoch": 0.13823027449485278,
      "grad_norm": 1.047514796257019,
      "learning_rate": 0.00017235415408249253,
      "loss": 3.604,
      "step": 148290
    },
    {
      "epoch": 0.13823959611293188,
      "grad_norm": 0.9543241262435913,
      "learning_rate": 0.00017235228976039734,
      "loss": 3.3306,
      "step": 148300
    },
    {
      "epoch": 0.13824891773101097,
      "grad_norm": 1.2869212627410889,
      "learning_rate": 0.00017235042543830214,
      "loss": 3.6222,
      "step": 148310
    },
    {
      "epoch": 0.13825823934909004,
      "grad_norm": 0.8947182893753052,
      "learning_rate": 0.00017234856111620695,
      "loss": 3.6351,
      "step": 148320
    },
    {
      "epoch": 0.13826756096716913,
      "grad_norm": 0.9117727875709534,
      "learning_rate": 0.00017234669679411173,
      "loss": 3.4835,
      "step": 148330
    },
    {
      "epoch": 0.13827688258524823,
      "grad_norm": 1.0649187564849854,
      "learning_rate": 0.00017234483247201653,
      "loss": 3.4661,
      "step": 148340
    },
    {
      "epoch": 0.13828620420332732,
      "grad_norm": 1.0953569412231445,
      "learning_rate": 0.00017234296814992131,
      "loss": 3.6585,
      "step": 148350
    },
    {
      "epoch": 0.13829552582140642,
      "grad_norm": 1.0137571096420288,
      "learning_rate": 0.00017234110382782615,
      "loss": 3.6635,
      "step": 148360
    },
    {
      "epoch": 0.1383048474394855,
      "grad_norm": 1.7049164772033691,
      "learning_rate": 0.00017233923950573095,
      "loss": 3.671,
      "step": 148370
    },
    {
      "epoch": 0.1383141690575646,
      "grad_norm": 1.0921680927276611,
      "learning_rate": 0.00017233737518363573,
      "loss": 3.5289,
      "step": 148380
    },
    {
      "epoch": 0.1383234906756437,
      "grad_norm": 0.9173253774642944,
      "learning_rate": 0.00017233551086154054,
      "loss": 3.4297,
      "step": 148390
    },
    {
      "epoch": 0.1383328122937228,
      "grad_norm": 1.1270849704742432,
      "learning_rate": 0.00017233364653944532,
      "loss": 3.5798,
      "step": 148400
    },
    {
      "epoch": 0.1383421339118019,
      "grad_norm": 1.1082119941711426,
      "learning_rate": 0.00017233178221735012,
      "loss": 3.5338,
      "step": 148410
    },
    {
      "epoch": 0.13835145552988098,
      "grad_norm": 1.2873739004135132,
      "learning_rate": 0.00017232991789525493,
      "loss": 3.5821,
      "step": 148420
    },
    {
      "epoch": 0.13836077714796005,
      "grad_norm": 1.13396418094635,
      "learning_rate": 0.00017232805357315974,
      "loss": 3.8253,
      "step": 148430
    },
    {
      "epoch": 0.13837009876603915,
      "grad_norm": 0.9910464286804199,
      "learning_rate": 0.00017232618925106454,
      "loss": 3.6064,
      "step": 148440
    },
    {
      "epoch": 0.13837942038411824,
      "grad_norm": 1.1985224485397339,
      "learning_rate": 0.00017232432492896932,
      "loss": 3.6774,
      "step": 148450
    },
    {
      "epoch": 0.13838874200219733,
      "grad_norm": 1.0445325374603271,
      "learning_rate": 0.00017232246060687413,
      "loss": 3.4814,
      "step": 148460
    },
    {
      "epoch": 0.13839806362027643,
      "grad_norm": 1.1819877624511719,
      "learning_rate": 0.00017232059628477894,
      "loss": 3.6018,
      "step": 148470
    },
    {
      "epoch": 0.13840738523835552,
      "grad_norm": 1.0248377323150635,
      "learning_rate": 0.00017231873196268374,
      "loss": 3.5845,
      "step": 148480
    },
    {
      "epoch": 0.13841670685643462,
      "grad_norm": 0.8466164469718933,
      "learning_rate": 0.00017231686764058855,
      "loss": 3.6616,
      "step": 148490
    },
    {
      "epoch": 0.1384260284745137,
      "grad_norm": 1.1380256414413452,
      "learning_rate": 0.00017231500331849333,
      "loss": 3.5115,
      "step": 148500
    },
    {
      "epoch": 0.1384353500925928,
      "grad_norm": 1.3514912128448486,
      "learning_rate": 0.00017231313899639813,
      "loss": 3.5647,
      "step": 148510
    },
    {
      "epoch": 0.1384446717106719,
      "grad_norm": 1.0718600749969482,
      "learning_rate": 0.00017231127467430294,
      "loss": 3.5491,
      "step": 148520
    },
    {
      "epoch": 0.13845399332875097,
      "grad_norm": 1.3692030906677246,
      "learning_rate": 0.00017230941035220775,
      "loss": 3.467,
      "step": 148530
    },
    {
      "epoch": 0.13846331494683006,
      "grad_norm": 1.046351432800293,
      "learning_rate": 0.00017230754603011255,
      "loss": 3.4728,
      "step": 148540
    },
    {
      "epoch": 0.13847263656490916,
      "grad_norm": 1.0937635898590088,
      "learning_rate": 0.00017230568170801733,
      "loss": 3.6776,
      "step": 148550
    },
    {
      "epoch": 0.13848195818298825,
      "grad_norm": 1.021158218383789,
      "learning_rate": 0.00017230381738592214,
      "loss": 3.6594,
      "step": 148560
    },
    {
      "epoch": 0.13849127980106735,
      "grad_norm": 1.067294955253601,
      "learning_rate": 0.00017230195306382694,
      "loss": 3.6075,
      "step": 148570
    },
    {
      "epoch": 0.13850060141914644,
      "grad_norm": 1.138123631477356,
      "learning_rate": 0.00017230008874173172,
      "loss": 3.6515,
      "step": 148580
    },
    {
      "epoch": 0.13850992303722554,
      "grad_norm": 1.0360738039016724,
      "learning_rate": 0.00017229822441963656,
      "loss": 3.6278,
      "step": 148590
    },
    {
      "epoch": 0.13851924465530463,
      "grad_norm": 1.1764942407608032,
      "learning_rate": 0.00017229636009754134,
      "loss": 3.3114,
      "step": 148600
    },
    {
      "epoch": 0.13852856627338372,
      "grad_norm": 0.8939507007598877,
      "learning_rate": 0.00017229449577544614,
      "loss": 3.805,
      "step": 148610
    },
    {
      "epoch": 0.13853788789146282,
      "grad_norm": 0.9887641072273254,
      "learning_rate": 0.00017229263145335095,
      "loss": 3.6487,
      "step": 148620
    },
    {
      "epoch": 0.1385472095095419,
      "grad_norm": 1.1812617778778076,
      "learning_rate": 0.00017229076713125573,
      "loss": 3.674,
      "step": 148630
    },
    {
      "epoch": 0.13855653112762098,
      "grad_norm": 0.9774559140205383,
      "learning_rate": 0.00017228890280916053,
      "loss": 3.5143,
      "step": 148640
    },
    {
      "epoch": 0.13856585274570007,
      "grad_norm": 0.8616652488708496,
      "learning_rate": 0.00017228703848706534,
      "loss": 3.6159,
      "step": 148650
    },
    {
      "epoch": 0.13857517436377917,
      "grad_norm": 1.0724419355392456,
      "learning_rate": 0.00017228517416497015,
      "loss": 3.7621,
      "step": 148660
    },
    {
      "epoch": 0.13858449598185826,
      "grad_norm": 1.142430305480957,
      "learning_rate": 0.00017228330984287495,
      "loss": 3.5923,
      "step": 148670
    },
    {
      "epoch": 0.13859381759993736,
      "grad_norm": 1.0130743980407715,
      "learning_rate": 0.00017228144552077973,
      "loss": 3.7223,
      "step": 148680
    },
    {
      "epoch": 0.13860313921801645,
      "grad_norm": 1.1061451435089111,
      "learning_rate": 0.00017227958119868454,
      "loss": 3.4997,
      "step": 148690
    },
    {
      "epoch": 0.13861246083609555,
      "grad_norm": 1.0529462099075317,
      "learning_rate": 0.00017227771687658934,
      "loss": 3.5894,
      "step": 148700
    },
    {
      "epoch": 0.13862178245417464,
      "grad_norm": 1.0845354795455933,
      "learning_rate": 0.00017227585255449415,
      "loss": 3.6158,
      "step": 148710
    },
    {
      "epoch": 0.13863110407225374,
      "grad_norm": 1.0165389776229858,
      "learning_rate": 0.00017227398823239896,
      "loss": 3.4864,
      "step": 148720
    },
    {
      "epoch": 0.13864042569033283,
      "grad_norm": 1.1635349988937378,
      "learning_rate": 0.00017227212391030374,
      "loss": 3.5895,
      "step": 148730
    },
    {
      "epoch": 0.1386497473084119,
      "grad_norm": 0.9087709784507751,
      "learning_rate": 0.00017227025958820854,
      "loss": 3.6324,
      "step": 148740
    },
    {
      "epoch": 0.138659068926491,
      "grad_norm": 0.9682511687278748,
      "learning_rate": 0.00017226839526611335,
      "loss": 3.6868,
      "step": 148750
    },
    {
      "epoch": 0.13866839054457009,
      "grad_norm": 1.0705369710922241,
      "learning_rate": 0.00017226653094401816,
      "loss": 3.5947,
      "step": 148760
    },
    {
      "epoch": 0.13867771216264918,
      "grad_norm": 1.0329182147979736,
      "learning_rate": 0.00017226466662192296,
      "loss": 3.6045,
      "step": 148770
    },
    {
      "epoch": 0.13868703378072827,
      "grad_norm": 1.0643744468688965,
      "learning_rate": 0.00017226280229982774,
      "loss": 3.6884,
      "step": 148780
    },
    {
      "epoch": 0.13869635539880737,
      "grad_norm": 1.1416780948638916,
      "learning_rate": 0.00017226093797773255,
      "loss": 3.6879,
      "step": 148790
    },
    {
      "epoch": 0.13870567701688646,
      "grad_norm": 1.0913000106811523,
      "learning_rate": 0.00017225907365563733,
      "loss": 3.4482,
      "step": 148800
    },
    {
      "epoch": 0.13871499863496556,
      "grad_norm": 1.0127400159835815,
      "learning_rate": 0.00017225720933354216,
      "loss": 3.5524,
      "step": 148810
    },
    {
      "epoch": 0.13872432025304465,
      "grad_norm": 0.9825572967529297,
      "learning_rate": 0.00017225534501144697,
      "loss": 3.5281,
      "step": 148820
    },
    {
      "epoch": 0.13873364187112375,
      "grad_norm": 1.1428778171539307,
      "learning_rate": 0.00017225348068935175,
      "loss": 3.5946,
      "step": 148830
    },
    {
      "epoch": 0.13874296348920284,
      "grad_norm": 1.1300705671310425,
      "learning_rate": 0.00017225161636725655,
      "loss": 3.7121,
      "step": 148840
    },
    {
      "epoch": 0.1387522851072819,
      "grad_norm": 0.8354442715644836,
      "learning_rate": 0.00017224975204516133,
      "loss": 3.6403,
      "step": 148850
    },
    {
      "epoch": 0.138761606725361,
      "grad_norm": 1.059863567352295,
      "learning_rate": 0.00017224788772306614,
      "loss": 3.6551,
      "step": 148860
    },
    {
      "epoch": 0.1387709283434401,
      "grad_norm": 1.1334049701690674,
      "learning_rate": 0.00017224602340097097,
      "loss": 3.7786,
      "step": 148870
    },
    {
      "epoch": 0.1387802499615192,
      "grad_norm": 0.742498517036438,
      "learning_rate": 0.00017224415907887575,
      "loss": 3.671,
      "step": 148880
    },
    {
      "epoch": 0.1387895715795983,
      "grad_norm": 1.103466510772705,
      "learning_rate": 0.00017224229475678056,
      "loss": 3.2383,
      "step": 148890
    },
    {
      "epoch": 0.13879889319767738,
      "grad_norm": 1.0904054641723633,
      "learning_rate": 0.00017224043043468534,
      "loss": 3.4837,
      "step": 148900
    },
    {
      "epoch": 0.13880821481575648,
      "grad_norm": 1.0537915229797363,
      "learning_rate": 0.00017223856611259014,
      "loss": 3.6209,
      "step": 148910
    },
    {
      "epoch": 0.13881753643383557,
      "grad_norm": 1.1134755611419678,
      "learning_rate": 0.00017223670179049495,
      "loss": 3.5577,
      "step": 148920
    },
    {
      "epoch": 0.13882685805191466,
      "grad_norm": 0.9174754619598389,
      "learning_rate": 0.00017223483746839975,
      "loss": 3.657,
      "step": 148930
    },
    {
      "epoch": 0.13883617966999376,
      "grad_norm": 1.2333911657333374,
      "learning_rate": 0.00017223297314630456,
      "loss": 3.5676,
      "step": 148940
    },
    {
      "epoch": 0.13884550128807283,
      "grad_norm": 1.0193372964859009,
      "learning_rate": 0.00017223110882420934,
      "loss": 3.5641,
      "step": 148950
    },
    {
      "epoch": 0.13885482290615192,
      "grad_norm": 1.1404067277908325,
      "learning_rate": 0.00017222924450211415,
      "loss": 3.6901,
      "step": 148960
    },
    {
      "epoch": 0.13886414452423101,
      "grad_norm": 1.2257874011993408,
      "learning_rate": 0.00017222738018001895,
      "loss": 3.3844,
      "step": 148970
    },
    {
      "epoch": 0.1388734661423101,
      "grad_norm": 1.1039732694625854,
      "learning_rate": 0.00017222551585792376,
      "loss": 3.5993,
      "step": 148980
    },
    {
      "epoch": 0.1388827877603892,
      "grad_norm": 1.021658182144165,
      "learning_rate": 0.00017222365153582857,
      "loss": 3.7041,
      "step": 148990
    },
    {
      "epoch": 0.1388921093784683,
      "grad_norm": 1.013990044593811,
      "learning_rate": 0.00017222178721373334,
      "loss": 3.6623,
      "step": 149000
    },
    {
      "epoch": 0.1389014309965474,
      "grad_norm": 1.190229892730713,
      "learning_rate": 0.00017221992289163815,
      "loss": 3.6589,
      "step": 149010
    },
    {
      "epoch": 0.1389107526146265,
      "grad_norm": 0.9667720198631287,
      "learning_rate": 0.00017221805856954296,
      "loss": 3.6483,
      "step": 149020
    },
    {
      "epoch": 0.13892007423270558,
      "grad_norm": 1.0939226150512695,
      "learning_rate": 0.00017221619424744774,
      "loss": 3.4789,
      "step": 149030
    },
    {
      "epoch": 0.13892939585078468,
      "grad_norm": 0.8775827288627625,
      "learning_rate": 0.00017221432992535257,
      "loss": 3.5203,
      "step": 149040
    },
    {
      "epoch": 0.13893871746886374,
      "grad_norm": 1.0921088457107544,
      "learning_rate": 0.00017221246560325735,
      "loss": 3.7335,
      "step": 149050
    },
    {
      "epoch": 0.13894803908694284,
      "grad_norm": 0.8414561748504639,
      "learning_rate": 0.00017221060128116216,
      "loss": 3.6391,
      "step": 149060
    },
    {
      "epoch": 0.13895736070502193,
      "grad_norm": 0.8574928045272827,
      "learning_rate": 0.00017220873695906696,
      "loss": 3.5612,
      "step": 149070
    },
    {
      "epoch": 0.13896668232310103,
      "grad_norm": 1.1823744773864746,
      "learning_rate": 0.00017220687263697174,
      "loss": 3.5958,
      "step": 149080
    },
    {
      "epoch": 0.13897600394118012,
      "grad_norm": 0.9658820629119873,
      "learning_rate": 0.00017220500831487655,
      "loss": 3.6213,
      "step": 149090
    },
    {
      "epoch": 0.13898532555925922,
      "grad_norm": 0.8589293360710144,
      "learning_rate": 0.00017220314399278135,
      "loss": 3.6651,
      "step": 149100
    },
    {
      "epoch": 0.1389946471773383,
      "grad_norm": 1.1485543251037598,
      "learning_rate": 0.00017220127967068616,
      "loss": 3.5508,
      "step": 149110
    },
    {
      "epoch": 0.1390039687954174,
      "grad_norm": 1.2012487649917603,
      "learning_rate": 0.00017219941534859097,
      "loss": 3.4553,
      "step": 149120
    },
    {
      "epoch": 0.1390132904134965,
      "grad_norm": 1.0020724534988403,
      "learning_rate": 0.00017219755102649574,
      "loss": 3.6427,
      "step": 149130
    },
    {
      "epoch": 0.1390226120315756,
      "grad_norm": 1.0566380023956299,
      "learning_rate": 0.00017219568670440055,
      "loss": 3.6429,
      "step": 149140
    },
    {
      "epoch": 0.1390319336496547,
      "grad_norm": 0.9684475660324097,
      "learning_rate": 0.00017219382238230536,
      "loss": 3.5231,
      "step": 149150
    },
    {
      "epoch": 0.13904125526773375,
      "grad_norm": 1.0787243843078613,
      "learning_rate": 0.00017219195806021016,
      "loss": 3.5757,
      "step": 149160
    },
    {
      "epoch": 0.13905057688581285,
      "grad_norm": 1.138532280921936,
      "learning_rate": 0.00017219009373811497,
      "loss": 3.4347,
      "step": 149170
    },
    {
      "epoch": 0.13905989850389194,
      "grad_norm": 1.1728103160858154,
      "learning_rate": 0.00017218822941601975,
      "loss": 3.5555,
      "step": 149180
    },
    {
      "epoch": 0.13906922012197104,
      "grad_norm": 1.027542233467102,
      "learning_rate": 0.00017218636509392456,
      "loss": 3.6506,
      "step": 149190
    },
    {
      "epoch": 0.13907854174005013,
      "grad_norm": 1.2766826152801514,
      "learning_rate": 0.00017218450077182936,
      "loss": 3.5574,
      "step": 149200
    },
    {
      "epoch": 0.13908786335812923,
      "grad_norm": 1.113474726676941,
      "learning_rate": 0.00017218263644973417,
      "loss": 3.6417,
      "step": 149210
    },
    {
      "epoch": 0.13909718497620832,
      "grad_norm": 1.06013023853302,
      "learning_rate": 0.00017218077212763895,
      "loss": 3.5866,
      "step": 149220
    },
    {
      "epoch": 0.13910650659428742,
      "grad_norm": 1.0010827779769897,
      "learning_rate": 0.00017217890780554375,
      "loss": 3.7141,
      "step": 149230
    },
    {
      "epoch": 0.1391158282123665,
      "grad_norm": 0.9133751392364502,
      "learning_rate": 0.00017217704348344856,
      "loss": 3.536,
      "step": 149240
    },
    {
      "epoch": 0.1391251498304456,
      "grad_norm": 1.0777875185012817,
      "learning_rate": 0.00017217517916135334,
      "loss": 3.5947,
      "step": 149250
    },
    {
      "epoch": 0.13913447144852467,
      "grad_norm": 0.8998909592628479,
      "learning_rate": 0.00017217331483925817,
      "loss": 3.7516,
      "step": 149260
    },
    {
      "epoch": 0.13914379306660377,
      "grad_norm": 1.083700180053711,
      "learning_rate": 0.00017217145051716295,
      "loss": 3.6772,
      "step": 149270
    },
    {
      "epoch": 0.13915311468468286,
      "grad_norm": 1.492845058441162,
      "learning_rate": 0.00017216958619506776,
      "loss": 3.5662,
      "step": 149280
    },
    {
      "epoch": 0.13916243630276195,
      "grad_norm": 0.9988409280776978,
      "learning_rate": 0.00017216772187297256,
      "loss": 3.6875,
      "step": 149290
    },
    {
      "epoch": 0.13917175792084105,
      "grad_norm": 1.013709306716919,
      "learning_rate": 0.00017216585755087734,
      "loss": 3.4197,
      "step": 149300
    },
    {
      "epoch": 0.13918107953892014,
      "grad_norm": 1.1682193279266357,
      "learning_rate": 0.00017216399322878215,
      "loss": 3.7474,
      "step": 149310
    },
    {
      "epoch": 0.13919040115699924,
      "grad_norm": 0.8335027694702148,
      "learning_rate": 0.00017216212890668696,
      "loss": 3.4396,
      "step": 149320
    },
    {
      "epoch": 0.13919972277507833,
      "grad_norm": 1.098682165145874,
      "learning_rate": 0.00017216026458459176,
      "loss": 3.5274,
      "step": 149330
    },
    {
      "epoch": 0.13920904439315743,
      "grad_norm": 1.201509714126587,
      "learning_rate": 0.00017215840026249657,
      "loss": 3.6427,
      "step": 149340
    },
    {
      "epoch": 0.13921836601123652,
      "grad_norm": 1.157997965812683,
      "learning_rate": 0.00017215653594040135,
      "loss": 3.4797,
      "step": 149350
    },
    {
      "epoch": 0.13922768762931562,
      "grad_norm": 1.0738250017166138,
      "learning_rate": 0.00017215467161830615,
      "loss": 3.4456,
      "step": 149360
    },
    {
      "epoch": 0.13923700924739468,
      "grad_norm": 1.2510546445846558,
      "learning_rate": 0.00017215280729621096,
      "loss": 3.6122,
      "step": 149370
    },
    {
      "epoch": 0.13924633086547378,
      "grad_norm": 0.9767403602600098,
      "learning_rate": 0.00017215094297411577,
      "loss": 3.6849,
      "step": 149380
    },
    {
      "epoch": 0.13925565248355287,
      "grad_norm": 1.233514428138733,
      "learning_rate": 0.00017214907865202057,
      "loss": 3.8332,
      "step": 149390
    },
    {
      "epoch": 0.13926497410163197,
      "grad_norm": 0.9472055435180664,
      "learning_rate": 0.00017214721432992535,
      "loss": 3.5871,
      "step": 149400
    },
    {
      "epoch": 0.13927429571971106,
      "grad_norm": 1.1365759372711182,
      "learning_rate": 0.00017214535000783016,
      "loss": 3.6428,
      "step": 149410
    },
    {
      "epoch": 0.13928361733779016,
      "grad_norm": 0.9273175001144409,
      "learning_rate": 0.00017214348568573494,
      "loss": 3.5526,
      "step": 149420
    },
    {
      "epoch": 0.13929293895586925,
      "grad_norm": 1.0965079069137573,
      "learning_rate": 0.00017214162136363977,
      "loss": 3.5432,
      "step": 149430
    },
    {
      "epoch": 0.13930226057394834,
      "grad_norm": 0.877341628074646,
      "learning_rate": 0.00017213975704154458,
      "loss": 3.4122,
      "step": 149440
    },
    {
      "epoch": 0.13931158219202744,
      "grad_norm": 1.043926477432251,
      "learning_rate": 0.00017213789271944936,
      "loss": 3.314,
      "step": 149450
    },
    {
      "epoch": 0.13932090381010653,
      "grad_norm": 0.9952359199523926,
      "learning_rate": 0.00017213602839735416,
      "loss": 3.6904,
      "step": 149460
    },
    {
      "epoch": 0.1393302254281856,
      "grad_norm": 1.0355571508407593,
      "learning_rate": 0.00017213416407525894,
      "loss": 3.4894,
      "step": 149470
    },
    {
      "epoch": 0.1393395470462647,
      "grad_norm": 1.0690319538116455,
      "learning_rate": 0.00017213229975316375,
      "loss": 3.4546,
      "step": 149480
    },
    {
      "epoch": 0.1393488686643438,
      "grad_norm": 1.1340694427490234,
      "learning_rate": 0.00017213043543106858,
      "loss": 3.5949,
      "step": 149490
    },
    {
      "epoch": 0.13935819028242288,
      "grad_norm": 1.135312795639038,
      "learning_rate": 0.00017212857110897336,
      "loss": 3.3712,
      "step": 149500
    },
    {
      "epoch": 0.13936751190050198,
      "grad_norm": 1.0499786138534546,
      "learning_rate": 0.00017212670678687817,
      "loss": 3.56,
      "step": 149510
    },
    {
      "epoch": 0.13937683351858107,
      "grad_norm": 0.9331574440002441,
      "learning_rate": 0.00017212484246478295,
      "loss": 3.7322,
      "step": 149520
    },
    {
      "epoch": 0.13938615513666017,
      "grad_norm": 0.9823952913284302,
      "learning_rate": 0.00017212297814268775,
      "loss": 3.5003,
      "step": 149530
    },
    {
      "epoch": 0.13939547675473926,
      "grad_norm": 0.9085015654563904,
      "learning_rate": 0.00017212111382059256,
      "loss": 3.5996,
      "step": 149540
    },
    {
      "epoch": 0.13940479837281836,
      "grad_norm": 1.0763126611709595,
      "learning_rate": 0.00017211924949849737,
      "loss": 3.5764,
      "step": 149550
    },
    {
      "epoch": 0.13941411999089745,
      "grad_norm": 1.0052772760391235,
      "learning_rate": 0.00017211738517640217,
      "loss": 3.6549,
      "step": 149560
    },
    {
      "epoch": 0.13942344160897654,
      "grad_norm": 1.017096757888794,
      "learning_rate": 0.00017211552085430695,
      "loss": 3.3744,
      "step": 149570
    },
    {
      "epoch": 0.1394327632270556,
      "grad_norm": 1.1547714471817017,
      "learning_rate": 0.00017211365653221176,
      "loss": 3.6573,
      "step": 149580
    },
    {
      "epoch": 0.1394420848451347,
      "grad_norm": 1.0846771001815796,
      "learning_rate": 0.00017211179221011656,
      "loss": 3.5963,
      "step": 149590
    },
    {
      "epoch": 0.1394514064632138,
      "grad_norm": 1.1281037330627441,
      "learning_rate": 0.00017210992788802137,
      "loss": 3.5958,
      "step": 149600
    },
    {
      "epoch": 0.1394607280812929,
      "grad_norm": 0.9957218766212463,
      "learning_rate": 0.00017210806356592618,
      "loss": 3.4495,
      "step": 149610
    },
    {
      "epoch": 0.139470049699372,
      "grad_norm": 1.1162171363830566,
      "learning_rate": 0.00017210619924383096,
      "loss": 3.4212,
      "step": 149620
    },
    {
      "epoch": 0.13947937131745108,
      "grad_norm": 1.189754605293274,
      "learning_rate": 0.00017210433492173576,
      "loss": 3.429,
      "step": 149630
    },
    {
      "epoch": 0.13948869293553018,
      "grad_norm": 1.1801801919937134,
      "learning_rate": 0.00017210247059964057,
      "loss": 3.6215,
      "step": 149640
    },
    {
      "epoch": 0.13949801455360927,
      "grad_norm": 0.9226011037826538,
      "learning_rate": 0.00017210060627754537,
      "loss": 3.4844,
      "step": 149650
    },
    {
      "epoch": 0.13950733617168837,
      "grad_norm": 1.026723027229309,
      "learning_rate": 0.00017209874195545018,
      "loss": 3.4442,
      "step": 149660
    },
    {
      "epoch": 0.13951665778976746,
      "grad_norm": 1.246159553527832,
      "learning_rate": 0.00017209687763335496,
      "loss": 3.5651,
      "step": 149670
    },
    {
      "epoch": 0.13952597940784653,
      "grad_norm": 1.087843418121338,
      "learning_rate": 0.00017209501331125977,
      "loss": 3.5396,
      "step": 149680
    },
    {
      "epoch": 0.13953530102592562,
      "grad_norm": 1.1097980737686157,
      "learning_rate": 0.00017209314898916457,
      "loss": 3.7293,
      "step": 149690
    },
    {
      "epoch": 0.13954462264400472,
      "grad_norm": 0.8824959993362427,
      "learning_rate": 0.00017209128466706935,
      "loss": 3.5168,
      "step": 149700
    },
    {
      "epoch": 0.1395539442620838,
      "grad_norm": 0.9892568588256836,
      "learning_rate": 0.00017208942034497419,
      "loss": 3.635,
      "step": 149710
    },
    {
      "epoch": 0.1395632658801629,
      "grad_norm": 1.0398635864257812,
      "learning_rate": 0.00017208755602287896,
      "loss": 3.4859,
      "step": 149720
    },
    {
      "epoch": 0.139572587498242,
      "grad_norm": 1.248709797859192,
      "learning_rate": 0.00017208569170078377,
      "loss": 3.5725,
      "step": 149730
    },
    {
      "epoch": 0.1395819091163211,
      "grad_norm": 1.015355110168457,
      "learning_rate": 0.00017208382737868858,
      "loss": 3.6621,
      "step": 149740
    },
    {
      "epoch": 0.1395912307344002,
      "grad_norm": 1.1474074125289917,
      "learning_rate": 0.00017208196305659336,
      "loss": 3.65,
      "step": 149750
    },
    {
      "epoch": 0.13960055235247928,
      "grad_norm": 1.2040636539459229,
      "learning_rate": 0.00017208009873449816,
      "loss": 3.5398,
      "step": 149760
    },
    {
      "epoch": 0.13960987397055838,
      "grad_norm": 1.0402756929397583,
      "learning_rate": 0.00017207823441240297,
      "loss": 3.8004,
      "step": 149770
    },
    {
      "epoch": 0.13961919558863747,
      "grad_norm": 1.3829805850982666,
      "learning_rate": 0.00017207637009030778,
      "loss": 3.5265,
      "step": 149780
    },
    {
      "epoch": 0.13962851720671654,
      "grad_norm": 0.9630875587463379,
      "learning_rate": 0.00017207450576821258,
      "loss": 3.4341,
      "step": 149790
    },
    {
      "epoch": 0.13963783882479563,
      "grad_norm": 1.0048127174377441,
      "learning_rate": 0.00017207264144611736,
      "loss": 3.5653,
      "step": 149800
    },
    {
      "epoch": 0.13964716044287473,
      "grad_norm": 1.0937705039978027,
      "learning_rate": 0.00017207077712402217,
      "loss": 3.5964,
      "step": 149810
    },
    {
      "epoch": 0.13965648206095382,
      "grad_norm": 0.968631386756897,
      "learning_rate": 0.00017206891280192697,
      "loss": 3.4572,
      "step": 149820
    },
    {
      "epoch": 0.13966580367903292,
      "grad_norm": 1.053571343421936,
      "learning_rate": 0.00017206704847983178,
      "loss": 3.7472,
      "step": 149830
    },
    {
      "epoch": 0.139675125297112,
      "grad_norm": 0.950094997882843,
      "learning_rate": 0.00017206518415773659,
      "loss": 3.4412,
      "step": 149840
    },
    {
      "epoch": 0.1396844469151911,
      "grad_norm": 1.0602858066558838,
      "learning_rate": 0.00017206331983564137,
      "loss": 3.665,
      "step": 149850
    },
    {
      "epoch": 0.1396937685332702,
      "grad_norm": 1.0597643852233887,
      "learning_rate": 0.00017206145551354617,
      "loss": 3.6635,
      "step": 149860
    },
    {
      "epoch": 0.1397030901513493,
      "grad_norm": 1.028101921081543,
      "learning_rate": 0.00017205959119145095,
      "loss": 3.5399,
      "step": 149870
    },
    {
      "epoch": 0.1397124117694284,
      "grad_norm": 1.168373942375183,
      "learning_rate": 0.00017205772686935578,
      "loss": 3.6089,
      "step": 149880
    },
    {
      "epoch": 0.13972173338750746,
      "grad_norm": 1.1443262100219727,
      "learning_rate": 0.0001720558625472606,
      "loss": 3.611,
      "step": 149890
    },
    {
      "epoch": 0.13973105500558655,
      "grad_norm": 1.1638613939285278,
      "learning_rate": 0.00017205399822516537,
      "loss": 3.593,
      "step": 149900
    },
    {
      "epoch": 0.13974037662366565,
      "grad_norm": 1.1373412609100342,
      "learning_rate": 0.00017205213390307018,
      "loss": 3.7055,
      "step": 149910
    },
    {
      "epoch": 0.13974969824174474,
      "grad_norm": 1.1161960363388062,
      "learning_rate": 0.00017205026958097496,
      "loss": 3.5231,
      "step": 149920
    },
    {
      "epoch": 0.13975901985982384,
      "grad_norm": 0.945715069770813,
      "learning_rate": 0.00017204840525887976,
      "loss": 3.6353,
      "step": 149930
    },
    {
      "epoch": 0.13976834147790293,
      "grad_norm": 1.1503199338912964,
      "learning_rate": 0.0001720465409367846,
      "loss": 3.5237,
      "step": 149940
    },
    {
      "epoch": 0.13977766309598202,
      "grad_norm": 1.065725564956665,
      "learning_rate": 0.00017204467661468937,
      "loss": 3.699,
      "step": 149950
    },
    {
      "epoch": 0.13978698471406112,
      "grad_norm": 1.124087929725647,
      "learning_rate": 0.00017204281229259418,
      "loss": 3.6438,
      "step": 149960
    },
    {
      "epoch": 0.1397963063321402,
      "grad_norm": 1.0904439687728882,
      "learning_rate": 0.00017204094797049896,
      "loss": 3.5473,
      "step": 149970
    },
    {
      "epoch": 0.1398056279502193,
      "grad_norm": 0.984140932559967,
      "learning_rate": 0.00017203908364840377,
      "loss": 3.6317,
      "step": 149980
    },
    {
      "epoch": 0.1398149495682984,
      "grad_norm": 0.9975849390029907,
      "learning_rate": 0.00017203721932630857,
      "loss": 3.495,
      "step": 149990
    },
    {
      "epoch": 0.13982427118637747,
      "grad_norm": 0.9943645000457764,
      "learning_rate": 0.00017203535500421338,
      "loss": 3.5667,
      "step": 150000
    },
    {
      "epoch": 0.13983359280445656,
      "grad_norm": 1.051193118095398,
      "learning_rate": 0.00017203349068211819,
      "loss": 3.5777,
      "step": 150010
    },
    {
      "epoch": 0.13984291442253566,
      "grad_norm": 1.1788536310195923,
      "learning_rate": 0.00017203162636002296,
      "loss": 3.4418,
      "step": 150020
    },
    {
      "epoch": 0.13985223604061475,
      "grad_norm": 1.1767021417617798,
      "learning_rate": 0.00017202976203792777,
      "loss": 3.6858,
      "step": 150030
    },
    {
      "epoch": 0.13986155765869385,
      "grad_norm": 1.076693058013916,
      "learning_rate": 0.00017202789771583258,
      "loss": 3.8694,
      "step": 150040
    },
    {
      "epoch": 0.13987087927677294,
      "grad_norm": 1.1781142950057983,
      "learning_rate": 0.00017202603339373738,
      "loss": 3.3652,
      "step": 150050
    },
    {
      "epoch": 0.13988020089485204,
      "grad_norm": 0.9153635501861572,
      "learning_rate": 0.0001720241690716422,
      "loss": 3.5217,
      "step": 150060
    },
    {
      "epoch": 0.13988952251293113,
      "grad_norm": 1.073116660118103,
      "learning_rate": 0.00017202230474954697,
      "loss": 3.5882,
      "step": 150070
    },
    {
      "epoch": 0.13989884413101022,
      "grad_norm": 0.9376103281974792,
      "learning_rate": 0.00017202044042745178,
      "loss": 3.5461,
      "step": 150080
    },
    {
      "epoch": 0.13990816574908932,
      "grad_norm": 1.3077595233917236,
      "learning_rate": 0.00017201857610535658,
      "loss": 3.653,
      "step": 150090
    },
    {
      "epoch": 0.13991748736716839,
      "grad_norm": 0.9323743581771851,
      "learning_rate": 0.0001720167117832614,
      "loss": 3.5421,
      "step": 150100
    },
    {
      "epoch": 0.13992680898524748,
      "grad_norm": 1.1351069211959839,
      "learning_rate": 0.0001720148474611662,
      "loss": 3.5532,
      "step": 150110
    },
    {
      "epoch": 0.13993613060332658,
      "grad_norm": 1.0158541202545166,
      "learning_rate": 0.00017201298313907097,
      "loss": 3.6256,
      "step": 150120
    },
    {
      "epoch": 0.13994545222140567,
      "grad_norm": 0.8915566802024841,
      "learning_rate": 0.00017201111881697578,
      "loss": 3.5317,
      "step": 150130
    },
    {
      "epoch": 0.13995477383948476,
      "grad_norm": 1.0635948181152344,
      "learning_rate": 0.00017200925449488059,
      "loss": 3.6503,
      "step": 150140
    },
    {
      "epoch": 0.13996409545756386,
      "grad_norm": 1.1396212577819824,
      "learning_rate": 0.00017200739017278537,
      "loss": 3.6643,
      "step": 150150
    },
    {
      "epoch": 0.13997341707564295,
      "grad_norm": 1.13202702999115,
      "learning_rate": 0.0001720055258506902,
      "loss": 3.4996,
      "step": 150160
    },
    {
      "epoch": 0.13998273869372205,
      "grad_norm": 0.9819089770317078,
      "learning_rate": 0.00017200366152859498,
      "loss": 3.6858,
      "step": 150170
    },
    {
      "epoch": 0.13999206031180114,
      "grad_norm": 1.0640641450881958,
      "learning_rate": 0.00017200179720649978,
      "loss": 3.7176,
      "step": 150180
    },
    {
      "epoch": 0.14000138192988024,
      "grad_norm": 1.1086574792861938,
      "learning_rate": 0.0001719999328844046,
      "loss": 3.5258,
      "step": 150190
    },
    {
      "epoch": 0.14001070354795933,
      "grad_norm": 1.0475653409957886,
      "learning_rate": 0.00017199806856230937,
      "loss": 3.5132,
      "step": 150200
    },
    {
      "epoch": 0.1400200251660384,
      "grad_norm": 1.106787085533142,
      "learning_rate": 0.00017199620424021418,
      "loss": 3.6198,
      "step": 150210
    },
    {
      "epoch": 0.1400293467841175,
      "grad_norm": 0.9918833374977112,
      "learning_rate": 0.00017199433991811898,
      "loss": 3.6033,
      "step": 150220
    },
    {
      "epoch": 0.1400386684021966,
      "grad_norm": 1.1608312129974365,
      "learning_rate": 0.0001719924755960238,
      "loss": 3.5331,
      "step": 150230
    },
    {
      "epoch": 0.14004799002027568,
      "grad_norm": 1.0060430765151978,
      "learning_rate": 0.0001719906112739286,
      "loss": 3.4948,
      "step": 150240
    },
    {
      "epoch": 0.14005731163835478,
      "grad_norm": 0.9876248836517334,
      "learning_rate": 0.00017198874695183337,
      "loss": 3.667,
      "step": 150250
    },
    {
      "epoch": 0.14006663325643387,
      "grad_norm": 1.129751443862915,
      "learning_rate": 0.00017198688262973818,
      "loss": 3.5243,
      "step": 150260
    },
    {
      "epoch": 0.14007595487451296,
      "grad_norm": 1.0513386726379395,
      "learning_rate": 0.000171985018307643,
      "loss": 3.8129,
      "step": 150270
    },
    {
      "epoch": 0.14008527649259206,
      "grad_norm": 1.1788996458053589,
      "learning_rate": 0.0001719831539855478,
      "loss": 3.5757,
      "step": 150280
    },
    {
      "epoch": 0.14009459811067115,
      "grad_norm": 1.0439742803573608,
      "learning_rate": 0.0001719812896634526,
      "loss": 3.5755,
      "step": 150290
    },
    {
      "epoch": 0.14010391972875025,
      "grad_norm": 1.037935733795166,
      "learning_rate": 0.00017197942534135738,
      "loss": 3.7087,
      "step": 150300
    },
    {
      "epoch": 0.14011324134682931,
      "grad_norm": 1.0740782022476196,
      "learning_rate": 0.00017197756101926218,
      "loss": 3.5131,
      "step": 150310
    },
    {
      "epoch": 0.1401225629649084,
      "grad_norm": 1.360688328742981,
      "learning_rate": 0.00017197569669716696,
      "loss": 3.5649,
      "step": 150320
    },
    {
      "epoch": 0.1401318845829875,
      "grad_norm": 1.0022902488708496,
      "learning_rate": 0.0001719738323750718,
      "loss": 3.6051,
      "step": 150330
    },
    {
      "epoch": 0.1401412062010666,
      "grad_norm": 0.9552358984947205,
      "learning_rate": 0.0001719719680529766,
      "loss": 3.7879,
      "step": 150340
    },
    {
      "epoch": 0.1401505278191457,
      "grad_norm": 1.052632212638855,
      "learning_rate": 0.00017197010373088138,
      "loss": 3.5979,
      "step": 150350
    },
    {
      "epoch": 0.1401598494372248,
      "grad_norm": 1.1278177499771118,
      "learning_rate": 0.0001719682394087862,
      "loss": 3.6875,
      "step": 150360
    },
    {
      "epoch": 0.14016917105530388,
      "grad_norm": 1.1471278667449951,
      "learning_rate": 0.00017196637508669097,
      "loss": 3.3592,
      "step": 150370
    },
    {
      "epoch": 0.14017849267338298,
      "grad_norm": 1.1730122566223145,
      "learning_rate": 0.00017196451076459577,
      "loss": 3.6243,
      "step": 150380
    },
    {
      "epoch": 0.14018781429146207,
      "grad_norm": 0.9978194236755371,
      "learning_rate": 0.0001719626464425006,
      "loss": 3.4871,
      "step": 150390
    },
    {
      "epoch": 0.14019713590954117,
      "grad_norm": 0.9994372129440308,
      "learning_rate": 0.0001719607821204054,
      "loss": 3.614,
      "step": 150400
    },
    {
      "epoch": 0.14020645752762023,
      "grad_norm": 1.176805853843689,
      "learning_rate": 0.0001719589177983102,
      "loss": 3.7435,
      "step": 150410
    },
    {
      "epoch": 0.14021577914569933,
      "grad_norm": 1.2818363904953003,
      "learning_rate": 0.00017195705347621497,
      "loss": 3.5206,
      "step": 150420
    },
    {
      "epoch": 0.14022510076377842,
      "grad_norm": 0.9944844245910645,
      "learning_rate": 0.00017195518915411978,
      "loss": 3.5336,
      "step": 150430
    },
    {
      "epoch": 0.14023442238185752,
      "grad_norm": 0.9817957878112793,
      "learning_rate": 0.00017195332483202459,
      "loss": 3.6429,
      "step": 150440
    },
    {
      "epoch": 0.1402437439999366,
      "grad_norm": 1.0199388265609741,
      "learning_rate": 0.0001719514605099294,
      "loss": 3.5043,
      "step": 150450
    },
    {
      "epoch": 0.1402530656180157,
      "grad_norm": 0.9811981320381165,
      "learning_rate": 0.0001719495961878342,
      "loss": 3.6199,
      "step": 150460
    },
    {
      "epoch": 0.1402623872360948,
      "grad_norm": 0.9471276998519897,
      "learning_rate": 0.00017194773186573898,
      "loss": 3.5973,
      "step": 150470
    },
    {
      "epoch": 0.1402717088541739,
      "grad_norm": 1.2440382242202759,
      "learning_rate": 0.00017194586754364378,
      "loss": 3.6995,
      "step": 150480
    },
    {
      "epoch": 0.140281030472253,
      "grad_norm": 0.9718981385231018,
      "learning_rate": 0.0001719440032215486,
      "loss": 3.5849,
      "step": 150490
    },
    {
      "epoch": 0.14029035209033208,
      "grad_norm": 1.288957953453064,
      "learning_rate": 0.0001719421388994534,
      "loss": 3.4745,
      "step": 150500
    },
    {
      "epoch": 0.14029967370841118,
      "grad_norm": 1.0551308393478394,
      "learning_rate": 0.0001719402745773582,
      "loss": 3.5361,
      "step": 150510
    },
    {
      "epoch": 0.14030899532649024,
      "grad_norm": 0.9694001078605652,
      "learning_rate": 0.00017193841025526298,
      "loss": 3.4731,
      "step": 150520
    },
    {
      "epoch": 0.14031831694456934,
      "grad_norm": 1.0137829780578613,
      "learning_rate": 0.0001719365459331678,
      "loss": 3.643,
      "step": 150530
    },
    {
      "epoch": 0.14032763856264843,
      "grad_norm": 1.1221636533737183,
      "learning_rate": 0.0001719346816110726,
      "loss": 3.7181,
      "step": 150540
    },
    {
      "epoch": 0.14033696018072753,
      "grad_norm": 1.237754464149475,
      "learning_rate": 0.0001719328172889774,
      "loss": 3.4811,
      "step": 150550
    },
    {
      "epoch": 0.14034628179880662,
      "grad_norm": 1.4057585000991821,
      "learning_rate": 0.0001719309529668822,
      "loss": 3.4648,
      "step": 150560
    },
    {
      "epoch": 0.14035560341688572,
      "grad_norm": 1.0296334028244019,
      "learning_rate": 0.00017192908864478699,
      "loss": 3.6555,
      "step": 150570
    },
    {
      "epoch": 0.1403649250349648,
      "grad_norm": 1.0516870021820068,
      "learning_rate": 0.0001719272243226918,
      "loss": 3.6353,
      "step": 150580
    },
    {
      "epoch": 0.1403742466530439,
      "grad_norm": 0.9478424191474915,
      "learning_rate": 0.00017192536000059657,
      "loss": 3.5535,
      "step": 150590
    },
    {
      "epoch": 0.140383568271123,
      "grad_norm": 1.056957721710205,
      "learning_rate": 0.00017192349567850138,
      "loss": 3.7527,
      "step": 150600
    },
    {
      "epoch": 0.1403928898892021,
      "grad_norm": 0.9917262196540833,
      "learning_rate": 0.0001719216313564062,
      "loss": 3.838,
      "step": 150610
    },
    {
      "epoch": 0.14040221150728116,
      "grad_norm": 1.0113718509674072,
      "learning_rate": 0.000171919767034311,
      "loss": 3.6768,
      "step": 150620
    },
    {
      "epoch": 0.14041153312536025,
      "grad_norm": 1.2431588172912598,
      "learning_rate": 0.0001719179027122158,
      "loss": 3.6244,
      "step": 150630
    },
    {
      "epoch": 0.14042085474343935,
      "grad_norm": 1.035378336906433,
      "learning_rate": 0.00017191603839012058,
      "loss": 3.5564,
      "step": 150640
    },
    {
      "epoch": 0.14043017636151844,
      "grad_norm": 1.0280083417892456,
      "learning_rate": 0.00017191417406802538,
      "loss": 3.6532,
      "step": 150650
    },
    {
      "epoch": 0.14043949797959754,
      "grad_norm": 1.0320794582366943,
      "learning_rate": 0.0001719123097459302,
      "loss": 3.559,
      "step": 150660
    },
    {
      "epoch": 0.14044881959767663,
      "grad_norm": 1.0840777158737183,
      "learning_rate": 0.000171910445423835,
      "loss": 3.6604,
      "step": 150670
    },
    {
      "epoch": 0.14045814121575573,
      "grad_norm": 1.1020455360412598,
      "learning_rate": 0.0001719085811017398,
      "loss": 3.6895,
      "step": 150680
    },
    {
      "epoch": 0.14046746283383482,
      "grad_norm": 1.0801811218261719,
      "learning_rate": 0.00017190671677964458,
      "loss": 3.6717,
      "step": 150690
    },
    {
      "epoch": 0.14047678445191392,
      "grad_norm": 1.0547988414764404,
      "learning_rate": 0.0001719048524575494,
      "loss": 3.6104,
      "step": 150700
    },
    {
      "epoch": 0.140486106069993,
      "grad_norm": 1.0066474676132202,
      "learning_rate": 0.0001719029881354542,
      "loss": 3.7259,
      "step": 150710
    },
    {
      "epoch": 0.1404954276880721,
      "grad_norm": 1.0884374380111694,
      "learning_rate": 0.000171901123813359,
      "loss": 3.4686,
      "step": 150720
    },
    {
      "epoch": 0.14050474930615117,
      "grad_norm": 1.0499852895736694,
      "learning_rate": 0.0001718992594912638,
      "loss": 3.6909,
      "step": 150730
    },
    {
      "epoch": 0.14051407092423027,
      "grad_norm": 1.253602147102356,
      "learning_rate": 0.00017189739516916858,
      "loss": 3.5571,
      "step": 150740
    },
    {
      "epoch": 0.14052339254230936,
      "grad_norm": 1.000280499458313,
      "learning_rate": 0.0001718955308470734,
      "loss": 3.6982,
      "step": 150750
    },
    {
      "epoch": 0.14053271416038846,
      "grad_norm": 1.0823793411254883,
      "learning_rate": 0.0001718936665249782,
      "loss": 3.5055,
      "step": 150760
    },
    {
      "epoch": 0.14054203577846755,
      "grad_norm": 0.929049015045166,
      "learning_rate": 0.00017189180220288298,
      "loss": 3.5534,
      "step": 150770
    },
    {
      "epoch": 0.14055135739654664,
      "grad_norm": 1.1245802640914917,
      "learning_rate": 0.0001718899378807878,
      "loss": 3.715,
      "step": 150780
    },
    {
      "epoch": 0.14056067901462574,
      "grad_norm": 1.2652050256729126,
      "learning_rate": 0.0001718880735586926,
      "loss": 3.6847,
      "step": 150790
    },
    {
      "epoch": 0.14057000063270483,
      "grad_norm": 1.0567954778671265,
      "learning_rate": 0.0001718862092365974,
      "loss": 3.5344,
      "step": 150800
    },
    {
      "epoch": 0.14057932225078393,
      "grad_norm": 1.0829228162765503,
      "learning_rate": 0.0001718843449145022,
      "loss": 3.4773,
      "step": 150810
    },
    {
      "epoch": 0.14058864386886302,
      "grad_norm": 0.7784203886985779,
      "learning_rate": 0.00017188248059240698,
      "loss": 3.4834,
      "step": 150820
    },
    {
      "epoch": 0.1405979654869421,
      "grad_norm": 1.3278429508209229,
      "learning_rate": 0.0001718806162703118,
      "loss": 3.6332,
      "step": 150830
    },
    {
      "epoch": 0.14060728710502118,
      "grad_norm": 1.0754280090332031,
      "learning_rate": 0.0001718787519482166,
      "loss": 3.3149,
      "step": 150840
    },
    {
      "epoch": 0.14061660872310028,
      "grad_norm": 1.1041544675827026,
      "learning_rate": 0.0001718768876261214,
      "loss": 3.5313,
      "step": 150850
    },
    {
      "epoch": 0.14062593034117937,
      "grad_norm": 1.0718395709991455,
      "learning_rate": 0.0001718750233040262,
      "loss": 3.8041,
      "step": 150860
    },
    {
      "epoch": 0.14063525195925847,
      "grad_norm": 1.1956734657287598,
      "learning_rate": 0.00017187315898193099,
      "loss": 3.525,
      "step": 150870
    },
    {
      "epoch": 0.14064457357733756,
      "grad_norm": 1.0486210584640503,
      "learning_rate": 0.0001718712946598358,
      "loss": 3.5443,
      "step": 150880
    },
    {
      "epoch": 0.14065389519541666,
      "grad_norm": 0.9230801463127136,
      "learning_rate": 0.0001718694303377406,
      "loss": 3.6131,
      "step": 150890
    },
    {
      "epoch": 0.14066321681349575,
      "grad_norm": 1.3282785415649414,
      "learning_rate": 0.0001718675660156454,
      "loss": 3.6687,
      "step": 150900
    },
    {
      "epoch": 0.14067253843157485,
      "grad_norm": 0.8785030245780945,
      "learning_rate": 0.0001718657016935502,
      "loss": 3.6825,
      "step": 150910
    },
    {
      "epoch": 0.14068186004965394,
      "grad_norm": 0.9990417957305908,
      "learning_rate": 0.000171863837371455,
      "loss": 3.6865,
      "step": 150920
    },
    {
      "epoch": 0.14069118166773303,
      "grad_norm": 1.0450578927993774,
      "learning_rate": 0.0001718619730493598,
      "loss": 3.5139,
      "step": 150930
    },
    {
      "epoch": 0.1407005032858121,
      "grad_norm": 1.1176248788833618,
      "learning_rate": 0.0001718601087272646,
      "loss": 3.6683,
      "step": 150940
    },
    {
      "epoch": 0.1407098249038912,
      "grad_norm": 0.9142024517059326,
      "learning_rate": 0.0001718582444051694,
      "loss": 3.6466,
      "step": 150950
    },
    {
      "epoch": 0.1407191465219703,
      "grad_norm": 1.0598946809768677,
      "learning_rate": 0.00017185638008307422,
      "loss": 3.613,
      "step": 150960
    },
    {
      "epoch": 0.14072846814004938,
      "grad_norm": 1.1360652446746826,
      "learning_rate": 0.000171854515760979,
      "loss": 3.826,
      "step": 150970
    },
    {
      "epoch": 0.14073778975812848,
      "grad_norm": 1.1452852487564087,
      "learning_rate": 0.0001718526514388838,
      "loss": 3.4159,
      "step": 150980
    },
    {
      "epoch": 0.14074711137620757,
      "grad_norm": 1.062838077545166,
      "learning_rate": 0.00017185078711678858,
      "loss": 3.5636,
      "step": 150990
    },
    {
      "epoch": 0.14075643299428667,
      "grad_norm": 1.2802796363830566,
      "learning_rate": 0.0001718489227946934,
      "loss": 3.7331,
      "step": 151000
    },
    {
      "epoch": 0.14076575461236576,
      "grad_norm": 0.8533562421798706,
      "learning_rate": 0.00017184705847259822,
      "loss": 3.6589,
      "step": 151010
    },
    {
      "epoch": 0.14077507623044486,
      "grad_norm": 1.0972224473953247,
      "learning_rate": 0.000171845194150503,
      "loss": 3.7974,
      "step": 151020
    },
    {
      "epoch": 0.14078439784852395,
      "grad_norm": 1.0196350812911987,
      "learning_rate": 0.0001718433298284078,
      "loss": 3.752,
      "step": 151030
    },
    {
      "epoch": 0.14079371946660302,
      "grad_norm": 1.0733129978179932,
      "learning_rate": 0.00017184146550631258,
      "loss": 3.6683,
      "step": 151040
    },
    {
      "epoch": 0.1408030410846821,
      "grad_norm": 1.06462562084198,
      "learning_rate": 0.0001718396011842174,
      "loss": 3.5972,
      "step": 151050
    },
    {
      "epoch": 0.1408123627027612,
      "grad_norm": 1.0043432712554932,
      "learning_rate": 0.00017183773686212222,
      "loss": 3.5416,
      "step": 151060
    },
    {
      "epoch": 0.1408216843208403,
      "grad_norm": 0.9412878155708313,
      "learning_rate": 0.000171835872540027,
      "loss": 3.7285,
      "step": 151070
    },
    {
      "epoch": 0.1408310059389194,
      "grad_norm": 1.2343958616256714,
      "learning_rate": 0.0001718340082179318,
      "loss": 3.4647,
      "step": 151080
    },
    {
      "epoch": 0.1408403275569985,
      "grad_norm": 0.9410206079483032,
      "learning_rate": 0.0001718321438958366,
      "loss": 3.6664,
      "step": 151090
    },
    {
      "epoch": 0.14084964917507758,
      "grad_norm": 0.9825697541236877,
      "learning_rate": 0.0001718302795737414,
      "loss": 3.6333,
      "step": 151100
    },
    {
      "epoch": 0.14085897079315668,
      "grad_norm": 1.0779376029968262,
      "learning_rate": 0.0001718284152516462,
      "loss": 3.4931,
      "step": 151110
    },
    {
      "epoch": 0.14086829241123577,
      "grad_norm": 1.1612577438354492,
      "learning_rate": 0.000171826550929551,
      "loss": 3.4997,
      "step": 151120
    },
    {
      "epoch": 0.14087761402931487,
      "grad_norm": 1.065339207649231,
      "learning_rate": 0.00017182468660745581,
      "loss": 3.6877,
      "step": 151130
    },
    {
      "epoch": 0.14088693564739396,
      "grad_norm": 1.0220775604248047,
      "learning_rate": 0.0001718228222853606,
      "loss": 3.7831,
      "step": 151140
    },
    {
      "epoch": 0.14089625726547303,
      "grad_norm": 0.9899886250495911,
      "learning_rate": 0.0001718209579632654,
      "loss": 3.5018,
      "step": 151150
    },
    {
      "epoch": 0.14090557888355212,
      "grad_norm": 1.0948395729064941,
      "learning_rate": 0.0001718190936411702,
      "loss": 3.4656,
      "step": 151160
    },
    {
      "epoch": 0.14091490050163122,
      "grad_norm": 1.1314414739608765,
      "learning_rate": 0.000171817229319075,
      "loss": 3.5289,
      "step": 151170
    },
    {
      "epoch": 0.1409242221197103,
      "grad_norm": 1.0326313972473145,
      "learning_rate": 0.00017181536499697982,
      "loss": 3.6324,
      "step": 151180
    },
    {
      "epoch": 0.1409335437377894,
      "grad_norm": 1.233610987663269,
      "learning_rate": 0.0001718135006748846,
      "loss": 3.5789,
      "step": 151190
    },
    {
      "epoch": 0.1409428653558685,
      "grad_norm": 0.9269067645072937,
      "learning_rate": 0.0001718116363527894,
      "loss": 3.6218,
      "step": 151200
    },
    {
      "epoch": 0.1409521869739476,
      "grad_norm": 1.3213132619857788,
      "learning_rate": 0.0001718097720306942,
      "loss": 3.7009,
      "step": 151210
    },
    {
      "epoch": 0.1409615085920267,
      "grad_norm": 1.2927963733673096,
      "learning_rate": 0.000171807907708599,
      "loss": 3.5933,
      "step": 151220
    },
    {
      "epoch": 0.14097083021010579,
      "grad_norm": 1.0759352445602417,
      "learning_rate": 0.00017180604338650382,
      "loss": 3.6498,
      "step": 151230
    },
    {
      "epoch": 0.14098015182818488,
      "grad_norm": 1.0161383152008057,
      "learning_rate": 0.0001718041790644086,
      "loss": 3.5812,
      "step": 151240
    },
    {
      "epoch": 0.14098947344626395,
      "grad_norm": 1.1377718448638916,
      "learning_rate": 0.0001718023147423134,
      "loss": 3.6053,
      "step": 151250
    },
    {
      "epoch": 0.14099879506434304,
      "grad_norm": 1.0943703651428223,
      "learning_rate": 0.00017180045042021821,
      "loss": 3.5625,
      "step": 151260
    },
    {
      "epoch": 0.14100811668242214,
      "grad_norm": 1.0162744522094727,
      "learning_rate": 0.000171798586098123,
      "loss": 3.602,
      "step": 151270
    },
    {
      "epoch": 0.14101743830050123,
      "grad_norm": 1.1856255531311035,
      "learning_rate": 0.0001717967217760278,
      "loss": 3.6335,
      "step": 151280
    },
    {
      "epoch": 0.14102675991858032,
      "grad_norm": 1.1697006225585938,
      "learning_rate": 0.0001717948574539326,
      "loss": 3.3746,
      "step": 151290
    },
    {
      "epoch": 0.14103608153665942,
      "grad_norm": 0.9847891926765442,
      "learning_rate": 0.0001717929931318374,
      "loss": 3.6162,
      "step": 151300
    },
    {
      "epoch": 0.1410454031547385,
      "grad_norm": 0.9944294691085815,
      "learning_rate": 0.00017179112880974222,
      "loss": 3.6408,
      "step": 151310
    },
    {
      "epoch": 0.1410547247728176,
      "grad_norm": 1.1554138660430908,
      "learning_rate": 0.000171789264487647,
      "loss": 3.7096,
      "step": 151320
    },
    {
      "epoch": 0.1410640463908967,
      "grad_norm": 0.9032492637634277,
      "learning_rate": 0.0001717874001655518,
      "loss": 3.7652,
      "step": 151330
    },
    {
      "epoch": 0.1410733680089758,
      "grad_norm": 0.8800977468490601,
      "learning_rate": 0.0001717855358434566,
      "loss": 3.5545,
      "step": 151340
    },
    {
      "epoch": 0.1410826896270549,
      "grad_norm": 1.1804462671279907,
      "learning_rate": 0.00017178367152136142,
      "loss": 3.7937,
      "step": 151350
    },
    {
      "epoch": 0.14109201124513396,
      "grad_norm": 1.0784523487091064,
      "learning_rate": 0.00017178180719926622,
      "loss": 3.5601,
      "step": 151360
    },
    {
      "epoch": 0.14110133286321305,
      "grad_norm": 1.064441442489624,
      "learning_rate": 0.000171779942877171,
      "loss": 3.4318,
      "step": 151370
    },
    {
      "epoch": 0.14111065448129215,
      "grad_norm": 1.0685583353042603,
      "learning_rate": 0.0001717780785550758,
      "loss": 3.471,
      "step": 151380
    },
    {
      "epoch": 0.14111997609937124,
      "grad_norm": 1.1902403831481934,
      "learning_rate": 0.00017177621423298062,
      "loss": 3.4844,
      "step": 151390
    },
    {
      "epoch": 0.14112929771745034,
      "grad_norm": 1.0164477825164795,
      "learning_rate": 0.00017177434991088542,
      "loss": 3.5689,
      "step": 151400
    },
    {
      "epoch": 0.14113861933552943,
      "grad_norm": 0.8919214606285095,
      "learning_rate": 0.00017177248558879023,
      "loss": 3.5895,
      "step": 151410
    },
    {
      "epoch": 0.14114794095360852,
      "grad_norm": 1.1029967069625854,
      "learning_rate": 0.000171770621266695,
      "loss": 3.5306,
      "step": 151420
    },
    {
      "epoch": 0.14115726257168762,
      "grad_norm": 1.0771119594573975,
      "learning_rate": 0.00017176875694459981,
      "loss": 3.5077,
      "step": 151430
    },
    {
      "epoch": 0.14116658418976671,
      "grad_norm": 1.248900055885315,
      "learning_rate": 0.0001717668926225046,
      "loss": 3.6345,
      "step": 151440
    },
    {
      "epoch": 0.1411759058078458,
      "grad_norm": 1.0763001441955566,
      "learning_rate": 0.00017176502830040943,
      "loss": 3.54,
      "step": 151450
    },
    {
      "epoch": 0.14118522742592488,
      "grad_norm": 1.087433099746704,
      "learning_rate": 0.00017176316397831423,
      "loss": 3.6903,
      "step": 151460
    },
    {
      "epoch": 0.14119454904400397,
      "grad_norm": 1.0757092237472534,
      "learning_rate": 0.000171761299656219,
      "loss": 3.7016,
      "step": 151470
    },
    {
      "epoch": 0.14120387066208306,
      "grad_norm": 1.0614173412322998,
      "learning_rate": 0.00017175943533412382,
      "loss": 3.6793,
      "step": 151480
    },
    {
      "epoch": 0.14121319228016216,
      "grad_norm": 0.9682047963142395,
      "learning_rate": 0.0001717575710120286,
      "loss": 3.665,
      "step": 151490
    },
    {
      "epoch": 0.14122251389824125,
      "grad_norm": 1.331950306892395,
      "learning_rate": 0.0001717557066899334,
      "loss": 3.3019,
      "step": 151500
    },
    {
      "epoch": 0.14123183551632035,
      "grad_norm": 0.9804937839508057,
      "learning_rate": 0.00017175384236783824,
      "loss": 3.5089,
      "step": 151510
    },
    {
      "epoch": 0.14124115713439944,
      "grad_norm": 1.2719978094100952,
      "learning_rate": 0.00017175197804574302,
      "loss": 3.4478,
      "step": 151520
    },
    {
      "epoch": 0.14125047875247854,
      "grad_norm": 1.108605146408081,
      "learning_rate": 0.00017175011372364782,
      "loss": 3.3429,
      "step": 151530
    },
    {
      "epoch": 0.14125980037055763,
      "grad_norm": 1.4199445247650146,
      "learning_rate": 0.0001717482494015526,
      "loss": 3.479,
      "step": 151540
    },
    {
      "epoch": 0.14126912198863673,
      "grad_norm": 1.0245927572250366,
      "learning_rate": 0.0001717463850794574,
      "loss": 3.5815,
      "step": 151550
    },
    {
      "epoch": 0.14127844360671582,
      "grad_norm": 1.0207937955856323,
      "learning_rate": 0.00017174452075736221,
      "loss": 3.5752,
      "step": 151560
    },
    {
      "epoch": 0.1412877652247949,
      "grad_norm": 1.0451719760894775,
      "learning_rate": 0.00017174265643526702,
      "loss": 3.6501,
      "step": 151570
    },
    {
      "epoch": 0.14129708684287398,
      "grad_norm": 1.0603774785995483,
      "learning_rate": 0.00017174079211317183,
      "loss": 3.512,
      "step": 151580
    },
    {
      "epoch": 0.14130640846095308,
      "grad_norm": 1.0680582523345947,
      "learning_rate": 0.0001717389277910766,
      "loss": 3.7797,
      "step": 151590
    },
    {
      "epoch": 0.14131573007903217,
      "grad_norm": 1.1935174465179443,
      "learning_rate": 0.0001717370634689814,
      "loss": 3.5908,
      "step": 151600
    },
    {
      "epoch": 0.14132505169711126,
      "grad_norm": 1.1081442832946777,
      "learning_rate": 0.00017173519914688622,
      "loss": 3.4371,
      "step": 151610
    },
    {
      "epoch": 0.14133437331519036,
      "grad_norm": 1.071347951889038,
      "learning_rate": 0.00017173333482479103,
      "loss": 3.4658,
      "step": 151620
    },
    {
      "epoch": 0.14134369493326945,
      "grad_norm": 1.1837455034255981,
      "learning_rate": 0.00017173147050269583,
      "loss": 3.7585,
      "step": 151630
    },
    {
      "epoch": 0.14135301655134855,
      "grad_norm": 1.1765503883361816,
      "learning_rate": 0.0001717296061806006,
      "loss": 3.6774,
      "step": 151640
    },
    {
      "epoch": 0.14136233816942764,
      "grad_norm": 1.0323125123977661,
      "learning_rate": 0.00017172774185850542,
      "loss": 3.3347,
      "step": 151650
    },
    {
      "epoch": 0.14137165978750674,
      "grad_norm": 0.990146815776825,
      "learning_rate": 0.00017172587753641022,
      "loss": 3.5943,
      "step": 151660
    },
    {
      "epoch": 0.1413809814055858,
      "grad_norm": 0.89277583360672,
      "learning_rate": 0.00017172401321431503,
      "loss": 3.5641,
      "step": 151670
    },
    {
      "epoch": 0.1413903030236649,
      "grad_norm": 0.9830759763717651,
      "learning_rate": 0.00017172214889221984,
      "loss": 3.6761,
      "step": 151680
    },
    {
      "epoch": 0.141399624641744,
      "grad_norm": 1.0172364711761475,
      "learning_rate": 0.00017172028457012461,
      "loss": 3.4793,
      "step": 151690
    },
    {
      "epoch": 0.1414089462598231,
      "grad_norm": 0.9186538457870483,
      "learning_rate": 0.00017171842024802942,
      "loss": 3.7021,
      "step": 151700
    },
    {
      "epoch": 0.14141826787790218,
      "grad_norm": 0.9350690245628357,
      "learning_rate": 0.00017171655592593423,
      "loss": 3.5329,
      "step": 151710
    },
    {
      "epoch": 0.14142758949598128,
      "grad_norm": 1.1582064628601074,
      "learning_rate": 0.000171714691603839,
      "loss": 3.7446,
      "step": 151720
    },
    {
      "epoch": 0.14143691111406037,
      "grad_norm": 1.1066304445266724,
      "learning_rate": 0.0001717128272817438,
      "loss": 3.5922,
      "step": 151730
    },
    {
      "epoch": 0.14144623273213947,
      "grad_norm": 1.3058433532714844,
      "learning_rate": 0.00017171096295964862,
      "loss": 3.487,
      "step": 151740
    },
    {
      "epoch": 0.14145555435021856,
      "grad_norm": 1.1389663219451904,
      "learning_rate": 0.00017170909863755343,
      "loss": 3.7902,
      "step": 151750
    },
    {
      "epoch": 0.14146487596829765,
      "grad_norm": 1.0724667310714722,
      "learning_rate": 0.00017170723431545823,
      "loss": 3.5885,
      "step": 151760
    },
    {
      "epoch": 0.14147419758637675,
      "grad_norm": 1.0308279991149902,
      "learning_rate": 0.000171705369993363,
      "loss": 3.4667,
      "step": 151770
    },
    {
      "epoch": 0.14148351920445582,
      "grad_norm": 1.0551249980926514,
      "learning_rate": 0.00017170350567126782,
      "loss": 3.6208,
      "step": 151780
    },
    {
      "epoch": 0.1414928408225349,
      "grad_norm": 1.162888765335083,
      "learning_rate": 0.00017170164134917262,
      "loss": 3.3885,
      "step": 151790
    },
    {
      "epoch": 0.141502162440614,
      "grad_norm": 1.0604939460754395,
      "learning_rate": 0.00017169977702707743,
      "loss": 3.7229,
      "step": 151800
    },
    {
      "epoch": 0.1415114840586931,
      "grad_norm": 1.0905790328979492,
      "learning_rate": 0.00017169791270498224,
      "loss": 3.3144,
      "step": 151810
    },
    {
      "epoch": 0.1415208056767722,
      "grad_norm": 1.0212973356246948,
      "learning_rate": 0.00017169604838288702,
      "loss": 3.7371,
      "step": 151820
    },
    {
      "epoch": 0.1415301272948513,
      "grad_norm": 1.0246752500534058,
      "learning_rate": 0.00017169418406079182,
      "loss": 3.4112,
      "step": 151830
    },
    {
      "epoch": 0.14153944891293038,
      "grad_norm": 1.1221567392349243,
      "learning_rate": 0.00017169231973869663,
      "loss": 3.649,
      "step": 151840
    },
    {
      "epoch": 0.14154877053100948,
      "grad_norm": 1.274890422821045,
      "learning_rate": 0.00017169045541660143,
      "loss": 3.5933,
      "step": 151850
    },
    {
      "epoch": 0.14155809214908857,
      "grad_norm": 1.212722897529602,
      "learning_rate": 0.00017168859109450624,
      "loss": 3.5036,
      "step": 151860
    },
    {
      "epoch": 0.14156741376716767,
      "grad_norm": 1.0921943187713623,
      "learning_rate": 0.00017168672677241102,
      "loss": 3.5187,
      "step": 151870
    },
    {
      "epoch": 0.14157673538524673,
      "grad_norm": 1.1606316566467285,
      "learning_rate": 0.00017168486245031583,
      "loss": 3.601,
      "step": 151880
    },
    {
      "epoch": 0.14158605700332583,
      "grad_norm": 0.8461074233055115,
      "learning_rate": 0.0001716829981282206,
      "loss": 3.6906,
      "step": 151890
    },
    {
      "epoch": 0.14159537862140492,
      "grad_norm": 1.1340913772583008,
      "learning_rate": 0.00017168113380612544,
      "loss": 3.6209,
      "step": 151900
    },
    {
      "epoch": 0.14160470023948402,
      "grad_norm": 1.0415561199188232,
      "learning_rate": 0.00017167926948403025,
      "loss": 3.5424,
      "step": 151910
    },
    {
      "epoch": 0.1416140218575631,
      "grad_norm": 1.5110293626785278,
      "learning_rate": 0.00017167740516193502,
      "loss": 3.3661,
      "step": 151920
    },
    {
      "epoch": 0.1416233434756422,
      "grad_norm": 0.8944222331047058,
      "learning_rate": 0.00017167554083983983,
      "loss": 3.6281,
      "step": 151930
    },
    {
      "epoch": 0.1416326650937213,
      "grad_norm": 0.9218345880508423,
      "learning_rate": 0.0001716736765177446,
      "loss": 3.5144,
      "step": 151940
    },
    {
      "epoch": 0.1416419867118004,
      "grad_norm": 0.931260883808136,
      "learning_rate": 0.00017167181219564942,
      "loss": 3.6142,
      "step": 151950
    },
    {
      "epoch": 0.1416513083298795,
      "grad_norm": 1.1055734157562256,
      "learning_rate": 0.00017166994787355422,
      "loss": 3.5918,
      "step": 151960
    },
    {
      "epoch": 0.14166062994795858,
      "grad_norm": 1.0413482189178467,
      "learning_rate": 0.00017166808355145903,
      "loss": 3.5765,
      "step": 151970
    },
    {
      "epoch": 0.14166995156603765,
      "grad_norm": 1.0540988445281982,
      "learning_rate": 0.00017166621922936384,
      "loss": 3.4199,
      "step": 151980
    },
    {
      "epoch": 0.14167927318411674,
      "grad_norm": 0.9809122681617737,
      "learning_rate": 0.00017166435490726861,
      "loss": 3.6042,
      "step": 151990
    },
    {
      "epoch": 0.14168859480219584,
      "grad_norm": 1.139542579650879,
      "learning_rate": 0.00017166249058517342,
      "loss": 3.5409,
      "step": 152000
    },
    {
      "epoch": 0.14169791642027493,
      "grad_norm": 1.1353843212127686,
      "learning_rate": 0.00017166062626307823,
      "loss": 3.6205,
      "step": 152010
    },
    {
      "epoch": 0.14170723803835403,
      "grad_norm": 1.0567535161972046,
      "learning_rate": 0.00017165876194098303,
      "loss": 3.5858,
      "step": 152020
    },
    {
      "epoch": 0.14171655965643312,
      "grad_norm": 1.1464965343475342,
      "learning_rate": 0.00017165689761888784,
      "loss": 3.7547,
      "step": 152030
    },
    {
      "epoch": 0.14172588127451222,
      "grad_norm": 1.141146183013916,
      "learning_rate": 0.00017165503329679262,
      "loss": 3.505,
      "step": 152040
    },
    {
      "epoch": 0.1417352028925913,
      "grad_norm": 0.9731961488723755,
      "learning_rate": 0.00017165316897469743,
      "loss": 3.5063,
      "step": 152050
    },
    {
      "epoch": 0.1417445245106704,
      "grad_norm": 1.0720787048339844,
      "learning_rate": 0.00017165130465260223,
      "loss": 3.495,
      "step": 152060
    },
    {
      "epoch": 0.1417538461287495,
      "grad_norm": 0.9929549694061279,
      "learning_rate": 0.00017164944033050704,
      "loss": 3.6024,
      "step": 152070
    },
    {
      "epoch": 0.1417631677468286,
      "grad_norm": 1.3254952430725098,
      "learning_rate": 0.00017164757600841184,
      "loss": 3.6837,
      "step": 152080
    },
    {
      "epoch": 0.14177248936490766,
      "grad_norm": 0.9781723022460938,
      "learning_rate": 0.00017164571168631662,
      "loss": 3.5654,
      "step": 152090
    },
    {
      "epoch": 0.14178181098298676,
      "grad_norm": 1.3124158382415771,
      "learning_rate": 0.00017164384736422143,
      "loss": 3.4819,
      "step": 152100
    },
    {
      "epoch": 0.14179113260106585,
      "grad_norm": 1.056805968284607,
      "learning_rate": 0.0001716419830421262,
      "loss": 3.4095,
      "step": 152110
    },
    {
      "epoch": 0.14180045421914494,
      "grad_norm": 1.101967215538025,
      "learning_rate": 0.00017164011872003104,
      "loss": 3.4801,
      "step": 152120
    },
    {
      "epoch": 0.14180977583722404,
      "grad_norm": 1.0788676738739014,
      "learning_rate": 0.00017163825439793585,
      "loss": 3.6683,
      "step": 152130
    },
    {
      "epoch": 0.14181909745530313,
      "grad_norm": 1.1631507873535156,
      "learning_rate": 0.00017163639007584063,
      "loss": 3.5665,
      "step": 152140
    },
    {
      "epoch": 0.14182841907338223,
      "grad_norm": 1.0855541229248047,
      "learning_rate": 0.00017163452575374543,
      "loss": 3.5199,
      "step": 152150
    },
    {
      "epoch": 0.14183774069146132,
      "grad_norm": 0.9941251277923584,
      "learning_rate": 0.0001716326614316502,
      "loss": 3.5948,
      "step": 152160
    },
    {
      "epoch": 0.14184706230954042,
      "grad_norm": 1.3099170923233032,
      "learning_rate": 0.00017163079710955502,
      "loss": 3.6396,
      "step": 152170
    },
    {
      "epoch": 0.1418563839276195,
      "grad_norm": 1.0603601932525635,
      "learning_rate": 0.00017162893278745985,
      "loss": 3.5721,
      "step": 152180
    },
    {
      "epoch": 0.14186570554569858,
      "grad_norm": 0.9165054559707642,
      "learning_rate": 0.00017162706846536463,
      "loss": 3.2966,
      "step": 152190
    },
    {
      "epoch": 0.14187502716377767,
      "grad_norm": 1.1729772090911865,
      "learning_rate": 0.00017162520414326944,
      "loss": 3.6006,
      "step": 152200
    },
    {
      "epoch": 0.14188434878185677,
      "grad_norm": 0.975760817527771,
      "learning_rate": 0.00017162333982117422,
      "loss": 3.5945,
      "step": 152210
    },
    {
      "epoch": 0.14189367039993586,
      "grad_norm": 0.9630842804908752,
      "learning_rate": 0.00017162147549907902,
      "loss": 3.4147,
      "step": 152220
    },
    {
      "epoch": 0.14190299201801496,
      "grad_norm": 0.9766077995300293,
      "learning_rate": 0.00017161961117698383,
      "loss": 3.547,
      "step": 152230
    },
    {
      "epoch": 0.14191231363609405,
      "grad_norm": 1.057561993598938,
      "learning_rate": 0.00017161774685488864,
      "loss": 3.6671,
      "step": 152240
    },
    {
      "epoch": 0.14192163525417315,
      "grad_norm": 1.052701711654663,
      "learning_rate": 0.00017161588253279344,
      "loss": 3.5948,
      "step": 152250
    },
    {
      "epoch": 0.14193095687225224,
      "grad_norm": 1.1245943307876587,
      "learning_rate": 0.00017161401821069822,
      "loss": 3.5127,
      "step": 152260
    },
    {
      "epoch": 0.14194027849033133,
      "grad_norm": 1.3238850831985474,
      "learning_rate": 0.00017161215388860303,
      "loss": 3.6469,
      "step": 152270
    },
    {
      "epoch": 0.14194960010841043,
      "grad_norm": 1.097868800163269,
      "learning_rate": 0.00017161028956650783,
      "loss": 3.5714,
      "step": 152280
    },
    {
      "epoch": 0.14195892172648952,
      "grad_norm": 0.9571011066436768,
      "learning_rate": 0.00017160842524441264,
      "loss": 3.4537,
      "step": 152290
    },
    {
      "epoch": 0.1419682433445686,
      "grad_norm": 1.1137244701385498,
      "learning_rate": 0.00017160656092231745,
      "loss": 3.7446,
      "step": 152300
    },
    {
      "epoch": 0.14197756496264768,
      "grad_norm": 1.056086540222168,
      "learning_rate": 0.00017160469660022223,
      "loss": 3.4658,
      "step": 152310
    },
    {
      "epoch": 0.14198688658072678,
      "grad_norm": 1.1204965114593506,
      "learning_rate": 0.00017160283227812703,
      "loss": 3.6441,
      "step": 152320
    },
    {
      "epoch": 0.14199620819880587,
      "grad_norm": 0.9703114628791809,
      "learning_rate": 0.00017160096795603184,
      "loss": 3.5314,
      "step": 152330
    },
    {
      "epoch": 0.14200552981688497,
      "grad_norm": 1.1302847862243652,
      "learning_rate": 0.00017159910363393662,
      "loss": 3.5053,
      "step": 152340
    },
    {
      "epoch": 0.14201485143496406,
      "grad_norm": 1.0231887102127075,
      "learning_rate": 0.00017159723931184145,
      "loss": 3.5596,
      "step": 152350
    },
    {
      "epoch": 0.14202417305304316,
      "grad_norm": 1.0520269870758057,
      "learning_rate": 0.00017159537498974623,
      "loss": 3.6145,
      "step": 152360
    },
    {
      "epoch": 0.14203349467112225,
      "grad_norm": 1.1226046085357666,
      "learning_rate": 0.00017159351066765104,
      "loss": 3.4172,
      "step": 152370
    },
    {
      "epoch": 0.14204281628920135,
      "grad_norm": 1.0699490308761597,
      "learning_rate": 0.00017159164634555584,
      "loss": 3.6486,
      "step": 152380
    },
    {
      "epoch": 0.14205213790728044,
      "grad_norm": 1.134860634803772,
      "learning_rate": 0.00017158978202346062,
      "loss": 3.7905,
      "step": 152390
    },
    {
      "epoch": 0.1420614595253595,
      "grad_norm": 1.1408751010894775,
      "learning_rate": 0.00017158791770136543,
      "loss": 3.5114,
      "step": 152400
    },
    {
      "epoch": 0.1420707811434386,
      "grad_norm": 1.1306627988815308,
      "learning_rate": 0.00017158605337927024,
      "loss": 3.5875,
      "step": 152410
    },
    {
      "epoch": 0.1420801027615177,
      "grad_norm": 0.9642449617385864,
      "learning_rate": 0.00017158418905717504,
      "loss": 3.567,
      "step": 152420
    },
    {
      "epoch": 0.1420894243795968,
      "grad_norm": 1.1107165813446045,
      "learning_rate": 0.00017158232473507985,
      "loss": 3.6928,
      "step": 152430
    },
    {
      "epoch": 0.14209874599767588,
      "grad_norm": 0.8839775919914246,
      "learning_rate": 0.00017158046041298463,
      "loss": 3.7079,
      "step": 152440
    },
    {
      "epoch": 0.14210806761575498,
      "grad_norm": 0.9404957294464111,
      "learning_rate": 0.00017157859609088943,
      "loss": 3.5443,
      "step": 152450
    },
    {
      "epoch": 0.14211738923383407,
      "grad_norm": 1.200943112373352,
      "learning_rate": 0.00017157673176879424,
      "loss": 3.6091,
      "step": 152460
    },
    {
      "epoch": 0.14212671085191317,
      "grad_norm": 1.008620262145996,
      "learning_rate": 0.00017157486744669905,
      "loss": 3.6145,
      "step": 152470
    },
    {
      "epoch": 0.14213603246999226,
      "grad_norm": 1.0347915887832642,
      "learning_rate": 0.00017157300312460385,
      "loss": 3.684,
      "step": 152480
    },
    {
      "epoch": 0.14214535408807136,
      "grad_norm": 1.0401620864868164,
      "learning_rate": 0.00017157113880250863,
      "loss": 3.5445,
      "step": 152490
    },
    {
      "epoch": 0.14215467570615045,
      "grad_norm": 1.0332120656967163,
      "learning_rate": 0.00017156927448041344,
      "loss": 3.6238,
      "step": 152500
    },
    {
      "epoch": 0.14216399732422952,
      "grad_norm": 1.1994942426681519,
      "learning_rate": 0.00017156741015831824,
      "loss": 3.5349,
      "step": 152510
    },
    {
      "epoch": 0.1421733189423086,
      "grad_norm": 1.1372888088226318,
      "learning_rate": 0.00017156554583622305,
      "loss": 3.6527,
      "step": 152520
    },
    {
      "epoch": 0.1421826405603877,
      "grad_norm": 1.2045800685882568,
      "learning_rate": 0.00017156368151412786,
      "loss": 3.6325,
      "step": 152530
    },
    {
      "epoch": 0.1421919621784668,
      "grad_norm": 1.090290904045105,
      "learning_rate": 0.00017156181719203264,
      "loss": 3.614,
      "step": 152540
    },
    {
      "epoch": 0.1422012837965459,
      "grad_norm": 1.074967622756958,
      "learning_rate": 0.00017155995286993744,
      "loss": 3.6969,
      "step": 152550
    },
    {
      "epoch": 0.142210605414625,
      "grad_norm": 1.1833994388580322,
      "learning_rate": 0.00017155808854784222,
      "loss": 3.4695,
      "step": 152560
    },
    {
      "epoch": 0.14221992703270409,
      "grad_norm": 0.9957694411277771,
      "learning_rate": 0.00017155622422574706,
      "loss": 3.7177,
      "step": 152570
    },
    {
      "epoch": 0.14222924865078318,
      "grad_norm": 1.2781010866165161,
      "learning_rate": 0.00017155435990365186,
      "loss": 3.5485,
      "step": 152580
    },
    {
      "epoch": 0.14223857026886227,
      "grad_norm": 1.2589796781539917,
      "learning_rate": 0.00017155249558155664,
      "loss": 3.5629,
      "step": 152590
    },
    {
      "epoch": 0.14224789188694137,
      "grad_norm": 1.1124495267868042,
      "learning_rate": 0.00017155063125946145,
      "loss": 3.7096,
      "step": 152600
    },
    {
      "epoch": 0.14225721350502044,
      "grad_norm": 0.9895505309104919,
      "learning_rate": 0.00017154876693736623,
      "loss": 3.4109,
      "step": 152610
    },
    {
      "epoch": 0.14226653512309953,
      "grad_norm": 1.0142818689346313,
      "learning_rate": 0.00017154690261527103,
      "loss": 3.7775,
      "step": 152620
    },
    {
      "epoch": 0.14227585674117862,
      "grad_norm": 1.17258620262146,
      "learning_rate": 0.00017154503829317587,
      "loss": 3.5747,
      "step": 152630
    },
    {
      "epoch": 0.14228517835925772,
      "grad_norm": 1.0448200702667236,
      "learning_rate": 0.00017154317397108065,
      "loss": 3.6002,
      "step": 152640
    },
    {
      "epoch": 0.1422944999773368,
      "grad_norm": 1.1393810510635376,
      "learning_rate": 0.00017154130964898545,
      "loss": 3.5541,
      "step": 152650
    },
    {
      "epoch": 0.1423038215954159,
      "grad_norm": 1.2429590225219727,
      "learning_rate": 0.00017153944532689023,
      "loss": 3.5617,
      "step": 152660
    },
    {
      "epoch": 0.142313143213495,
      "grad_norm": 1.2342623472213745,
      "learning_rate": 0.00017153758100479504,
      "loss": 3.651,
      "step": 152670
    },
    {
      "epoch": 0.1423224648315741,
      "grad_norm": 0.9460858702659607,
      "learning_rate": 0.00017153571668269984,
      "loss": 3.5845,
      "step": 152680
    },
    {
      "epoch": 0.1423317864496532,
      "grad_norm": 1.180064082145691,
      "learning_rate": 0.00017153385236060465,
      "loss": 3.459,
      "step": 152690
    },
    {
      "epoch": 0.14234110806773229,
      "grad_norm": 0.8928459882736206,
      "learning_rate": 0.00017153198803850946,
      "loss": 3.3859,
      "step": 152700
    },
    {
      "epoch": 0.14235042968581138,
      "grad_norm": 1.0562522411346436,
      "learning_rate": 0.00017153012371641424,
      "loss": 3.5854,
      "step": 152710
    },
    {
      "epoch": 0.14235975130389045,
      "grad_norm": 1.0856349468231201,
      "learning_rate": 0.00017152825939431904,
      "loss": 3.6658,
      "step": 152720
    },
    {
      "epoch": 0.14236907292196954,
      "grad_norm": 0.9149188995361328,
      "learning_rate": 0.00017152639507222385,
      "loss": 3.69,
      "step": 152730
    },
    {
      "epoch": 0.14237839454004864,
      "grad_norm": 1.008880376815796,
      "learning_rate": 0.00017152453075012865,
      "loss": 3.5803,
      "step": 152740
    },
    {
      "epoch": 0.14238771615812773,
      "grad_norm": 1.0905548334121704,
      "learning_rate": 0.00017152266642803346,
      "loss": 3.5564,
      "step": 152750
    },
    {
      "epoch": 0.14239703777620683,
      "grad_norm": 1.0557242631912231,
      "learning_rate": 0.00017152080210593824,
      "loss": 3.4007,
      "step": 152760
    },
    {
      "epoch": 0.14240635939428592,
      "grad_norm": 0.9910137057304382,
      "learning_rate": 0.00017151893778384305,
      "loss": 3.7465,
      "step": 152770
    },
    {
      "epoch": 0.14241568101236501,
      "grad_norm": 1.1444814205169678,
      "learning_rate": 0.00017151707346174785,
      "loss": 3.7238,
      "step": 152780
    },
    {
      "epoch": 0.1424250026304441,
      "grad_norm": 1.1438616514205933,
      "learning_rate": 0.00017151520913965263,
      "loss": 3.438,
      "step": 152790
    },
    {
      "epoch": 0.1424343242485232,
      "grad_norm": 1.0761452913284302,
      "learning_rate": 0.00017151334481755746,
      "loss": 3.5616,
      "step": 152800
    },
    {
      "epoch": 0.1424436458666023,
      "grad_norm": 1.0086309909820557,
      "learning_rate": 0.00017151148049546224,
      "loss": 3.4633,
      "step": 152810
    },
    {
      "epoch": 0.14245296748468136,
      "grad_norm": 1.100712537765503,
      "learning_rate": 0.00017150961617336705,
      "loss": 3.7623,
      "step": 152820
    },
    {
      "epoch": 0.14246228910276046,
      "grad_norm": 1.027875304222107,
      "learning_rate": 0.00017150775185127186,
      "loss": 3.7484,
      "step": 152830
    },
    {
      "epoch": 0.14247161072083955,
      "grad_norm": 1.0281453132629395,
      "learning_rate": 0.00017150588752917664,
      "loss": 3.6387,
      "step": 152840
    },
    {
      "epoch": 0.14248093233891865,
      "grad_norm": 0.927763044834137,
      "learning_rate": 0.00017150402320708144,
      "loss": 3.5311,
      "step": 152850
    },
    {
      "epoch": 0.14249025395699774,
      "grad_norm": 0.9911327362060547,
      "learning_rate": 0.00017150215888498625,
      "loss": 3.6469,
      "step": 152860
    },
    {
      "epoch": 0.14249957557507684,
      "grad_norm": 1.0704671144485474,
      "learning_rate": 0.00017150029456289105,
      "loss": 3.599,
      "step": 152870
    },
    {
      "epoch": 0.14250889719315593,
      "grad_norm": 0.8737671971321106,
      "learning_rate": 0.00017149843024079586,
      "loss": 3.5399,
      "step": 152880
    },
    {
      "epoch": 0.14251821881123503,
      "grad_norm": 1.028764009475708,
      "learning_rate": 0.00017149656591870064,
      "loss": 3.8103,
      "step": 152890
    },
    {
      "epoch": 0.14252754042931412,
      "grad_norm": 0.941236138343811,
      "learning_rate": 0.00017149470159660545,
      "loss": 3.6551,
      "step": 152900
    },
    {
      "epoch": 0.14253686204739321,
      "grad_norm": 1.1398513317108154,
      "learning_rate": 0.00017149283727451025,
      "loss": 3.6218,
      "step": 152910
    },
    {
      "epoch": 0.1425461836654723,
      "grad_norm": 0.9165390729904175,
      "learning_rate": 0.00017149097295241506,
      "loss": 3.5173,
      "step": 152920
    },
    {
      "epoch": 0.14255550528355138,
      "grad_norm": 1.1385406255722046,
      "learning_rate": 0.00017148910863031987,
      "loss": 3.6547,
      "step": 152930
    },
    {
      "epoch": 0.14256482690163047,
      "grad_norm": 1.2720729112625122,
      "learning_rate": 0.00017148724430822464,
      "loss": 3.7011,
      "step": 152940
    },
    {
      "epoch": 0.14257414851970956,
      "grad_norm": 0.9273967742919922,
      "learning_rate": 0.00017148537998612945,
      "loss": 3.7557,
      "step": 152950
    },
    {
      "epoch": 0.14258347013778866,
      "grad_norm": 1.009867548942566,
      "learning_rate": 0.00017148351566403426,
      "loss": 3.4646,
      "step": 152960
    },
    {
      "epoch": 0.14259279175586775,
      "grad_norm": 1.0523558855056763,
      "learning_rate": 0.00017148165134193906,
      "loss": 3.7753,
      "step": 152970
    },
    {
      "epoch": 0.14260211337394685,
      "grad_norm": 1.0450526475906372,
      "learning_rate": 0.00017147978701984387,
      "loss": 3.6781,
      "step": 152980
    },
    {
      "epoch": 0.14261143499202594,
      "grad_norm": 1.174451470375061,
      "learning_rate": 0.00017147792269774865,
      "loss": 3.4761,
      "step": 152990
    },
    {
      "epoch": 0.14262075661010504,
      "grad_norm": 0.9414661526679993,
      "learning_rate": 0.00017147605837565346,
      "loss": 3.5344,
      "step": 153000
    },
    {
      "epoch": 0.14263007822818413,
      "grad_norm": 1.0564031600952148,
      "learning_rate": 0.00017147419405355823,
      "loss": 3.5542,
      "step": 153010
    },
    {
      "epoch": 0.14263939984626323,
      "grad_norm": 1.0135841369628906,
      "learning_rate": 0.00017147232973146307,
      "loss": 3.5249,
      "step": 153020
    },
    {
      "epoch": 0.1426487214643423,
      "grad_norm": 1.2348746061325073,
      "learning_rate": 0.00017147046540936787,
      "loss": 3.7407,
      "step": 153030
    },
    {
      "epoch": 0.1426580430824214,
      "grad_norm": 0.9765200614929199,
      "learning_rate": 0.00017146860108727265,
      "loss": 3.7978,
      "step": 153040
    },
    {
      "epoch": 0.14266736470050048,
      "grad_norm": 0.9994516968727112,
      "learning_rate": 0.00017146673676517746,
      "loss": 3.679,
      "step": 153050
    },
    {
      "epoch": 0.14267668631857958,
      "grad_norm": 1.2016370296478271,
      "learning_rate": 0.00017146487244308224,
      "loss": 3.6426,
      "step": 153060
    },
    {
      "epoch": 0.14268600793665867,
      "grad_norm": 1.1988798379898071,
      "learning_rate": 0.00017146300812098705,
      "loss": 3.6656,
      "step": 153070
    },
    {
      "epoch": 0.14269532955473777,
      "grad_norm": 1.1497993469238281,
      "learning_rate": 0.00017146114379889188,
      "loss": 3.4531,
      "step": 153080
    },
    {
      "epoch": 0.14270465117281686,
      "grad_norm": 1.3255313634872437,
      "learning_rate": 0.00017145927947679666,
      "loss": 3.5857,
      "step": 153090
    },
    {
      "epoch": 0.14271397279089595,
      "grad_norm": 1.043752908706665,
      "learning_rate": 0.00017145741515470146,
      "loss": 3.6205,
      "step": 153100
    },
    {
      "epoch": 0.14272329440897505,
      "grad_norm": 1.152408242225647,
      "learning_rate": 0.00017145555083260624,
      "loss": 3.6222,
      "step": 153110
    },
    {
      "epoch": 0.14273261602705414,
      "grad_norm": 0.8296293020248413,
      "learning_rate": 0.00017145368651051105,
      "loss": 3.516,
      "step": 153120
    },
    {
      "epoch": 0.14274193764513324,
      "grad_norm": 1.0119067430496216,
      "learning_rate": 0.00017145182218841586,
      "loss": 3.4642,
      "step": 153130
    },
    {
      "epoch": 0.1427512592632123,
      "grad_norm": 1.248132586479187,
      "learning_rate": 0.00017144995786632066,
      "loss": 3.6066,
      "step": 153140
    },
    {
      "epoch": 0.1427605808812914,
      "grad_norm": 1.0778142213821411,
      "learning_rate": 0.00017144809354422547,
      "loss": 3.4471,
      "step": 153150
    },
    {
      "epoch": 0.1427699024993705,
      "grad_norm": 1.0217572450637817,
      "learning_rate": 0.00017144622922213025,
      "loss": 3.4689,
      "step": 153160
    },
    {
      "epoch": 0.1427792241174496,
      "grad_norm": 1.0286458730697632,
      "learning_rate": 0.00017144436490003505,
      "loss": 3.6595,
      "step": 153170
    },
    {
      "epoch": 0.14278854573552868,
      "grad_norm": 1.1163667440414429,
      "learning_rate": 0.00017144250057793986,
      "loss": 3.5758,
      "step": 153180
    },
    {
      "epoch": 0.14279786735360778,
      "grad_norm": 0.8437511920928955,
      "learning_rate": 0.00017144063625584467,
      "loss": 3.7321,
      "step": 153190
    },
    {
      "epoch": 0.14280718897168687,
      "grad_norm": 0.9725773930549622,
      "learning_rate": 0.00017143877193374947,
      "loss": 3.5073,
      "step": 153200
    },
    {
      "epoch": 0.14281651058976597,
      "grad_norm": 1.0855647325515747,
      "learning_rate": 0.00017143690761165425,
      "loss": 3.7026,
      "step": 153210
    },
    {
      "epoch": 0.14282583220784506,
      "grad_norm": 1.3088511228561401,
      "learning_rate": 0.00017143504328955906,
      "loss": 3.6368,
      "step": 153220
    },
    {
      "epoch": 0.14283515382592415,
      "grad_norm": 1.018242597579956,
      "learning_rate": 0.00017143317896746386,
      "loss": 3.421,
      "step": 153230
    },
    {
      "epoch": 0.14284447544400322,
      "grad_norm": 1.1876564025878906,
      "learning_rate": 0.00017143131464536864,
      "loss": 3.5786,
      "step": 153240
    },
    {
      "epoch": 0.14285379706208232,
      "grad_norm": 1.204664707183838,
      "learning_rate": 0.00017142945032327348,
      "loss": 3.6818,
      "step": 153250
    },
    {
      "epoch": 0.1428631186801614,
      "grad_norm": 1.0197780132293701,
      "learning_rate": 0.00017142758600117826,
      "loss": 3.6191,
      "step": 153260
    },
    {
      "epoch": 0.1428724402982405,
      "grad_norm": 1.2241995334625244,
      "learning_rate": 0.00017142572167908306,
      "loss": 3.7252,
      "step": 153270
    },
    {
      "epoch": 0.1428817619163196,
      "grad_norm": 1.0267914533615112,
      "learning_rate": 0.00017142385735698784,
      "loss": 3.5173,
      "step": 153280
    },
    {
      "epoch": 0.1428910835343987,
      "grad_norm": 1.0177563428878784,
      "learning_rate": 0.00017142199303489265,
      "loss": 3.7105,
      "step": 153290
    },
    {
      "epoch": 0.1429004051524778,
      "grad_norm": 1.1174510717391968,
      "learning_rate": 0.00017142012871279745,
      "loss": 3.4828,
      "step": 153300
    },
    {
      "epoch": 0.14290972677055688,
      "grad_norm": 1.0179425477981567,
      "learning_rate": 0.00017141826439070226,
      "loss": 3.6333,
      "step": 153310
    },
    {
      "epoch": 0.14291904838863598,
      "grad_norm": 1.0202131271362305,
      "learning_rate": 0.00017141640006860707,
      "loss": 3.4638,
      "step": 153320
    },
    {
      "epoch": 0.14292837000671507,
      "grad_norm": 1.0103015899658203,
      "learning_rate": 0.00017141453574651185,
      "loss": 3.4964,
      "step": 153330
    },
    {
      "epoch": 0.14293769162479414,
      "grad_norm": 0.9153586030006409,
      "learning_rate": 0.00017141267142441665,
      "loss": 3.6358,
      "step": 153340
    },
    {
      "epoch": 0.14294701324287323,
      "grad_norm": 1.1668250560760498,
      "learning_rate": 0.00017141080710232146,
      "loss": 3.5494,
      "step": 153350
    },
    {
      "epoch": 0.14295633486095233,
      "grad_norm": 1.0667903423309326,
      "learning_rate": 0.00017140894278022627,
      "loss": 3.6032,
      "step": 153360
    },
    {
      "epoch": 0.14296565647903142,
      "grad_norm": 1.0429270267486572,
      "learning_rate": 0.00017140707845813107,
      "loss": 3.6341,
      "step": 153370
    },
    {
      "epoch": 0.14297497809711052,
      "grad_norm": 1.012746810913086,
      "learning_rate": 0.00017140521413603585,
      "loss": 3.5032,
      "step": 153380
    },
    {
      "epoch": 0.1429842997151896,
      "grad_norm": 1.01856529712677,
      "learning_rate": 0.00017140334981394066,
      "loss": 3.6267,
      "step": 153390
    },
    {
      "epoch": 0.1429936213332687,
      "grad_norm": 1.2073583602905273,
      "learning_rate": 0.00017140148549184546,
      "loss": 3.7233,
      "step": 153400
    },
    {
      "epoch": 0.1430029429513478,
      "grad_norm": 1.0476127862930298,
      "learning_rate": 0.00017139962116975027,
      "loss": 3.7418,
      "step": 153410
    },
    {
      "epoch": 0.1430122645694269,
      "grad_norm": 0.9847469329833984,
      "learning_rate": 0.00017139775684765508,
      "loss": 3.541,
      "step": 153420
    },
    {
      "epoch": 0.143021586187506,
      "grad_norm": 1.0826503038406372,
      "learning_rate": 0.00017139589252555986,
      "loss": 3.7171,
      "step": 153430
    },
    {
      "epoch": 0.14303090780558508,
      "grad_norm": 1.2092671394348145,
      "learning_rate": 0.00017139402820346466,
      "loss": 3.7331,
      "step": 153440
    },
    {
      "epoch": 0.14304022942366415,
      "grad_norm": 1.0529035329818726,
      "learning_rate": 0.00017139216388136947,
      "loss": 3.5698,
      "step": 153450
    },
    {
      "epoch": 0.14304955104174324,
      "grad_norm": 1.1252950429916382,
      "learning_rate": 0.00017139029955927425,
      "loss": 3.6058,
      "step": 153460
    },
    {
      "epoch": 0.14305887265982234,
      "grad_norm": 1.149488091468811,
      "learning_rate": 0.00017138843523717908,
      "loss": 3.543,
      "step": 153470
    },
    {
      "epoch": 0.14306819427790143,
      "grad_norm": 0.9445948004722595,
      "learning_rate": 0.00017138657091508386,
      "loss": 3.738,
      "step": 153480
    },
    {
      "epoch": 0.14307751589598053,
      "grad_norm": 1.0365134477615356,
      "learning_rate": 0.00017138470659298867,
      "loss": 3.574,
      "step": 153490
    },
    {
      "epoch": 0.14308683751405962,
      "grad_norm": 1.3899023532867432,
      "learning_rate": 0.00017138284227089347,
      "loss": 3.4076,
      "step": 153500
    },
    {
      "epoch": 0.14309615913213872,
      "grad_norm": 1.0575913190841675,
      "learning_rate": 0.00017138097794879825,
      "loss": 3.5448,
      "step": 153510
    },
    {
      "epoch": 0.1431054807502178,
      "grad_norm": 1.1404181718826294,
      "learning_rate": 0.00017137911362670306,
      "loss": 3.5839,
      "step": 153520
    },
    {
      "epoch": 0.1431148023682969,
      "grad_norm": 1.5798875093460083,
      "learning_rate": 0.00017137724930460786,
      "loss": 3.5667,
      "step": 153530
    },
    {
      "epoch": 0.143124123986376,
      "grad_norm": 0.9628497362136841,
      "learning_rate": 0.00017137538498251267,
      "loss": 3.5021,
      "step": 153540
    },
    {
      "epoch": 0.14313344560445507,
      "grad_norm": 1.098304033279419,
      "learning_rate": 0.00017137352066041748,
      "loss": 3.4794,
      "step": 153550
    },
    {
      "epoch": 0.14314276722253416,
      "grad_norm": 1.1243118047714233,
      "learning_rate": 0.00017137165633832226,
      "loss": 3.4096,
      "step": 153560
    },
    {
      "epoch": 0.14315208884061326,
      "grad_norm": 1.1787165403366089,
      "learning_rate": 0.00017136979201622706,
      "loss": 3.5436,
      "step": 153570
    },
    {
      "epoch": 0.14316141045869235,
      "grad_norm": 1.1196606159210205,
      "learning_rate": 0.00017136792769413187,
      "loss": 3.5284,
      "step": 153580
    },
    {
      "epoch": 0.14317073207677145,
      "grad_norm": 1.0168554782867432,
      "learning_rate": 0.00017136606337203668,
      "loss": 3.5723,
      "step": 153590
    },
    {
      "epoch": 0.14318005369485054,
      "grad_norm": 0.926483154296875,
      "learning_rate": 0.00017136419904994148,
      "loss": 3.531,
      "step": 153600
    },
    {
      "epoch": 0.14318937531292963,
      "grad_norm": 1.1805939674377441,
      "learning_rate": 0.00017136233472784626,
      "loss": 3.6425,
      "step": 153610
    },
    {
      "epoch": 0.14319869693100873,
      "grad_norm": 0.9590440392494202,
      "learning_rate": 0.00017136047040575107,
      "loss": 3.5719,
      "step": 153620
    },
    {
      "epoch": 0.14320801854908782,
      "grad_norm": 1.043452501296997,
      "learning_rate": 0.00017135860608365585,
      "loss": 3.7434,
      "step": 153630
    },
    {
      "epoch": 0.14321734016716692,
      "grad_norm": 0.9886943697929382,
      "learning_rate": 0.00017135674176156068,
      "loss": 3.7261,
      "step": 153640
    },
    {
      "epoch": 0.143226661785246,
      "grad_norm": 1.1323446035385132,
      "learning_rate": 0.00017135487743946549,
      "loss": 3.6292,
      "step": 153650
    },
    {
      "epoch": 0.14323598340332508,
      "grad_norm": 1.0998190641403198,
      "learning_rate": 0.00017135301311737027,
      "loss": 3.6521,
      "step": 153660
    },
    {
      "epoch": 0.14324530502140417,
      "grad_norm": 1.0065518617630005,
      "learning_rate": 0.00017135114879527507,
      "loss": 3.5256,
      "step": 153670
    },
    {
      "epoch": 0.14325462663948327,
      "grad_norm": 1.0499423742294312,
      "learning_rate": 0.00017134928447317985,
      "loss": 3.4697,
      "step": 153680
    },
    {
      "epoch": 0.14326394825756236,
      "grad_norm": 1.1695222854614258,
      "learning_rate": 0.00017134742015108466,
      "loss": 3.7664,
      "step": 153690
    },
    {
      "epoch": 0.14327326987564146,
      "grad_norm": 1.0411834716796875,
      "learning_rate": 0.0001713455558289895,
      "loss": 3.5029,
      "step": 153700
    },
    {
      "epoch": 0.14328259149372055,
      "grad_norm": 1.0620228052139282,
      "learning_rate": 0.00017134369150689427,
      "loss": 3.5734,
      "step": 153710
    },
    {
      "epoch": 0.14329191311179965,
      "grad_norm": 1.1617785692214966,
      "learning_rate": 0.00017134182718479908,
      "loss": 3.5876,
      "step": 153720
    },
    {
      "epoch": 0.14330123472987874,
      "grad_norm": 1.117738127708435,
      "learning_rate": 0.00017133996286270386,
      "loss": 3.6829,
      "step": 153730
    },
    {
      "epoch": 0.14331055634795783,
      "grad_norm": 1.0044304132461548,
      "learning_rate": 0.00017133809854060866,
      "loss": 3.6118,
      "step": 153740
    },
    {
      "epoch": 0.14331987796603693,
      "grad_norm": 0.9649195671081543,
      "learning_rate": 0.00017133623421851347,
      "loss": 3.4645,
      "step": 153750
    },
    {
      "epoch": 0.143329199584116,
      "grad_norm": 1.1061044931411743,
      "learning_rate": 0.00017133436989641827,
      "loss": 3.5325,
      "step": 153760
    },
    {
      "epoch": 0.1433385212021951,
      "grad_norm": 1.247739553451538,
      "learning_rate": 0.00017133250557432308,
      "loss": 3.6899,
      "step": 153770
    },
    {
      "epoch": 0.14334784282027419,
      "grad_norm": 0.9748040437698364,
      "learning_rate": 0.00017133064125222786,
      "loss": 3.8024,
      "step": 153780
    },
    {
      "epoch": 0.14335716443835328,
      "grad_norm": 0.8139598369598389,
      "learning_rate": 0.00017132877693013267,
      "loss": 3.8947,
      "step": 153790
    },
    {
      "epoch": 0.14336648605643237,
      "grad_norm": 1.2887179851531982,
      "learning_rate": 0.00017132691260803747,
      "loss": 3.54,
      "step": 153800
    },
    {
      "epoch": 0.14337580767451147,
      "grad_norm": 1.0559797286987305,
      "learning_rate": 0.00017132504828594228,
      "loss": 3.7217,
      "step": 153810
    },
    {
      "epoch": 0.14338512929259056,
      "grad_norm": 1.100137710571289,
      "learning_rate": 0.00017132318396384708,
      "loss": 3.4816,
      "step": 153820
    },
    {
      "epoch": 0.14339445091066966,
      "grad_norm": 1.1252919435501099,
      "learning_rate": 0.00017132131964175186,
      "loss": 3.5266,
      "step": 153830
    },
    {
      "epoch": 0.14340377252874875,
      "grad_norm": 1.1487971544265747,
      "learning_rate": 0.00017131945531965667,
      "loss": 3.7148,
      "step": 153840
    },
    {
      "epoch": 0.14341309414682785,
      "grad_norm": 1.0699281692504883,
      "learning_rate": 0.00017131759099756148,
      "loss": 3.6238,
      "step": 153850
    },
    {
      "epoch": 0.14342241576490694,
      "grad_norm": 1.0144704580307007,
      "learning_rate": 0.00017131572667546628,
      "loss": 3.542,
      "step": 153860
    },
    {
      "epoch": 0.143431737382986,
      "grad_norm": 0.989854633808136,
      "learning_rate": 0.0001713138623533711,
      "loss": 3.4492,
      "step": 153870
    },
    {
      "epoch": 0.1434410590010651,
      "grad_norm": 1.0522072315216064,
      "learning_rate": 0.00017131199803127587,
      "loss": 3.5785,
      "step": 153880
    },
    {
      "epoch": 0.1434503806191442,
      "grad_norm": 0.957607090473175,
      "learning_rate": 0.00017131013370918067,
      "loss": 3.5396,
      "step": 153890
    },
    {
      "epoch": 0.1434597022372233,
      "grad_norm": 1.0276532173156738,
      "learning_rate": 0.00017130826938708548,
      "loss": 3.6434,
      "step": 153900
    },
    {
      "epoch": 0.14346902385530239,
      "grad_norm": 1.042465090751648,
      "learning_rate": 0.00017130640506499026,
      "loss": 3.6815,
      "step": 153910
    },
    {
      "epoch": 0.14347834547338148,
      "grad_norm": 1.262345314025879,
      "learning_rate": 0.0001713045407428951,
      "loss": 3.7716,
      "step": 153920
    },
    {
      "epoch": 0.14348766709146057,
      "grad_norm": 0.9855847358703613,
      "learning_rate": 0.00017130267642079987,
      "loss": 3.6844,
      "step": 153930
    },
    {
      "epoch": 0.14349698870953967,
      "grad_norm": 0.9060105085372925,
      "learning_rate": 0.00017130081209870468,
      "loss": 3.6505,
      "step": 153940
    },
    {
      "epoch": 0.14350631032761876,
      "grad_norm": 1.03811776638031,
      "learning_rate": 0.00017129894777660949,
      "loss": 3.5046,
      "step": 153950
    },
    {
      "epoch": 0.14351563194569786,
      "grad_norm": 0.9816924929618835,
      "learning_rate": 0.00017129708345451426,
      "loss": 3.653,
      "step": 153960
    },
    {
      "epoch": 0.14352495356377692,
      "grad_norm": 1.165908932685852,
      "learning_rate": 0.00017129521913241907,
      "loss": 3.5009,
      "step": 153970
    },
    {
      "epoch": 0.14353427518185602,
      "grad_norm": 1.1006433963775635,
      "learning_rate": 0.00017129335481032388,
      "loss": 3.3672,
      "step": 153980
    },
    {
      "epoch": 0.1435435967999351,
      "grad_norm": 1.2030242681503296,
      "learning_rate": 0.00017129149048822868,
      "loss": 3.4172,
      "step": 153990
    },
    {
      "epoch": 0.1435529184180142,
      "grad_norm": 0.9918385148048401,
      "learning_rate": 0.0001712896261661335,
      "loss": 3.6037,
      "step": 154000
    },
    {
      "epoch": 0.1435622400360933,
      "grad_norm": 0.7911553382873535,
      "learning_rate": 0.00017128776184403827,
      "loss": 3.6403,
      "step": 154010
    },
    {
      "epoch": 0.1435715616541724,
      "grad_norm": 1.133812665939331,
      "learning_rate": 0.00017128589752194308,
      "loss": 3.5641,
      "step": 154020
    },
    {
      "epoch": 0.1435808832722515,
      "grad_norm": 1.049822449684143,
      "learning_rate": 0.00017128403319984788,
      "loss": 3.5436,
      "step": 154030
    },
    {
      "epoch": 0.14359020489033059,
      "grad_norm": 1.058208703994751,
      "learning_rate": 0.0001712821688777527,
      "loss": 3.6598,
      "step": 154040
    },
    {
      "epoch": 0.14359952650840968,
      "grad_norm": 1.1062344312667847,
      "learning_rate": 0.0001712803045556575,
      "loss": 3.5651,
      "step": 154050
    },
    {
      "epoch": 0.14360884812648878,
      "grad_norm": 0.8981274962425232,
      "learning_rate": 0.00017127844023356227,
      "loss": 3.6176,
      "step": 154060
    },
    {
      "epoch": 0.14361816974456787,
      "grad_norm": 1.156650185585022,
      "learning_rate": 0.00017127657591146708,
      "loss": 3.7833,
      "step": 154070
    },
    {
      "epoch": 0.14362749136264694,
      "grad_norm": 0.9687634706497192,
      "learning_rate": 0.00017127471158937186,
      "loss": 3.7244,
      "step": 154080
    },
    {
      "epoch": 0.14363681298072603,
      "grad_norm": 1.673258900642395,
      "learning_rate": 0.0001712728472672767,
      "loss": 3.4815,
      "step": 154090
    },
    {
      "epoch": 0.14364613459880513,
      "grad_norm": 1.146096110343933,
      "learning_rate": 0.0001712709829451815,
      "loss": 3.6073,
      "step": 154100
    },
    {
      "epoch": 0.14365545621688422,
      "grad_norm": 1.0877548456192017,
      "learning_rate": 0.00017126911862308628,
      "loss": 3.6612,
      "step": 154110
    },
    {
      "epoch": 0.14366477783496331,
      "grad_norm": 1.0730915069580078,
      "learning_rate": 0.00017126725430099108,
      "loss": 3.359,
      "step": 154120
    },
    {
      "epoch": 0.1436740994530424,
      "grad_norm": 1.188194751739502,
      "learning_rate": 0.00017126538997889586,
      "loss": 3.8215,
      "step": 154130
    },
    {
      "epoch": 0.1436834210711215,
      "grad_norm": 1.0260593891143799,
      "learning_rate": 0.00017126352565680067,
      "loss": 3.4974,
      "step": 154140
    },
    {
      "epoch": 0.1436927426892006,
      "grad_norm": 1.2788366079330444,
      "learning_rate": 0.0001712616613347055,
      "loss": 3.5422,
      "step": 154150
    },
    {
      "epoch": 0.1437020643072797,
      "grad_norm": 0.9669493436813354,
      "learning_rate": 0.00017125979701261028,
      "loss": 3.5656,
      "step": 154160
    },
    {
      "epoch": 0.1437113859253588,
      "grad_norm": 0.9947595000267029,
      "learning_rate": 0.0001712579326905151,
      "loss": 3.5782,
      "step": 154170
    },
    {
      "epoch": 0.14372070754343785,
      "grad_norm": 0.9862261414527893,
      "learning_rate": 0.00017125606836841987,
      "loss": 3.5401,
      "step": 154180
    },
    {
      "epoch": 0.14373002916151695,
      "grad_norm": 0.8032464385032654,
      "learning_rate": 0.00017125420404632467,
      "loss": 3.5671,
      "step": 154190
    },
    {
      "epoch": 0.14373935077959604,
      "grad_norm": 1.0265331268310547,
      "learning_rate": 0.00017125233972422948,
      "loss": 3.7109,
      "step": 154200
    },
    {
      "epoch": 0.14374867239767514,
      "grad_norm": 0.9614636301994324,
      "learning_rate": 0.0001712504754021343,
      "loss": 3.6406,
      "step": 154210
    },
    {
      "epoch": 0.14375799401575423,
      "grad_norm": 1.1203792095184326,
      "learning_rate": 0.0001712486110800391,
      "loss": 3.4339,
      "step": 154220
    },
    {
      "epoch": 0.14376731563383333,
      "grad_norm": 1.0267714262008667,
      "learning_rate": 0.00017124674675794387,
      "loss": 3.6118,
      "step": 154230
    },
    {
      "epoch": 0.14377663725191242,
      "grad_norm": 0.8440752625465393,
      "learning_rate": 0.00017124488243584868,
      "loss": 3.6037,
      "step": 154240
    },
    {
      "epoch": 0.14378595886999151,
      "grad_norm": 1.2145836353302002,
      "learning_rate": 0.00017124301811375349,
      "loss": 3.7001,
      "step": 154250
    },
    {
      "epoch": 0.1437952804880706,
      "grad_norm": 1.1536428928375244,
      "learning_rate": 0.0001712411537916583,
      "loss": 3.6666,
      "step": 154260
    },
    {
      "epoch": 0.1438046021061497,
      "grad_norm": 1.0840590000152588,
      "learning_rate": 0.0001712392894695631,
      "loss": 3.6035,
      "step": 154270
    },
    {
      "epoch": 0.1438139237242288,
      "grad_norm": 0.9748354554176331,
      "learning_rate": 0.00017123742514746788,
      "loss": 3.3654,
      "step": 154280
    },
    {
      "epoch": 0.14382324534230786,
      "grad_norm": 1.1832507848739624,
      "learning_rate": 0.00017123556082537268,
      "loss": 3.6384,
      "step": 154290
    },
    {
      "epoch": 0.14383256696038696,
      "grad_norm": 1.010113000869751,
      "learning_rate": 0.0001712336965032775,
      "loss": 3.4474,
      "step": 154300
    },
    {
      "epoch": 0.14384188857846605,
      "grad_norm": 0.9405809640884399,
      "learning_rate": 0.0001712318321811823,
      "loss": 3.6562,
      "step": 154310
    },
    {
      "epoch": 0.14385121019654515,
      "grad_norm": 1.0938429832458496,
      "learning_rate": 0.0001712299678590871,
      "loss": 3.5258,
      "step": 154320
    },
    {
      "epoch": 0.14386053181462424,
      "grad_norm": 0.9740532636642456,
      "learning_rate": 0.00017122810353699188,
      "loss": 3.5817,
      "step": 154330
    },
    {
      "epoch": 0.14386985343270334,
      "grad_norm": 1.0857306718826294,
      "learning_rate": 0.0001712262392148967,
      "loss": 3.5142,
      "step": 154340
    },
    {
      "epoch": 0.14387917505078243,
      "grad_norm": 1.2575881481170654,
      "learning_rate": 0.0001712243748928015,
      "loss": 3.5428,
      "step": 154350
    },
    {
      "epoch": 0.14388849666886153,
      "grad_norm": 1.0410816669464111,
      "learning_rate": 0.00017122251057070627,
      "loss": 3.3883,
      "step": 154360
    },
    {
      "epoch": 0.14389781828694062,
      "grad_norm": 0.9715193510055542,
      "learning_rate": 0.0001712206462486111,
      "loss": 3.6237,
      "step": 154370
    },
    {
      "epoch": 0.14390713990501972,
      "grad_norm": 0.894880473613739,
      "learning_rate": 0.00017121878192651589,
      "loss": 3.4869,
      "step": 154380
    },
    {
      "epoch": 0.14391646152309878,
      "grad_norm": 1.0667260885238647,
      "learning_rate": 0.0001712169176044207,
      "loss": 3.5324,
      "step": 154390
    },
    {
      "epoch": 0.14392578314117788,
      "grad_norm": 1.1609402894973755,
      "learning_rate": 0.0001712150532823255,
      "loss": 3.5694,
      "step": 154400
    },
    {
      "epoch": 0.14393510475925697,
      "grad_norm": 1.0254281759262085,
      "learning_rate": 0.00017121318896023028,
      "loss": 3.7294,
      "step": 154410
    },
    {
      "epoch": 0.14394442637733607,
      "grad_norm": 0.8901448249816895,
      "learning_rate": 0.00017121132463813508,
      "loss": 3.787,
      "step": 154420
    },
    {
      "epoch": 0.14395374799541516,
      "grad_norm": 0.9950999021530151,
      "learning_rate": 0.0001712094603160399,
      "loss": 3.3654,
      "step": 154430
    },
    {
      "epoch": 0.14396306961349425,
      "grad_norm": 1.0634710788726807,
      "learning_rate": 0.0001712075959939447,
      "loss": 3.4732,
      "step": 154440
    },
    {
      "epoch": 0.14397239123157335,
      "grad_norm": 1.0422698259353638,
      "learning_rate": 0.0001712057316718495,
      "loss": 3.664,
      "step": 154450
    },
    {
      "epoch": 0.14398171284965244,
      "grad_norm": 0.8609758019447327,
      "learning_rate": 0.00017120386734975428,
      "loss": 3.5415,
      "step": 154460
    },
    {
      "epoch": 0.14399103446773154,
      "grad_norm": 0.9558796286582947,
      "learning_rate": 0.0001712020030276591,
      "loss": 3.6579,
      "step": 154470
    },
    {
      "epoch": 0.14400035608581063,
      "grad_norm": 1.1053262948989868,
      "learning_rate": 0.0001712001387055639,
      "loss": 3.7138,
      "step": 154480
    },
    {
      "epoch": 0.14400967770388973,
      "grad_norm": 1.1009798049926758,
      "learning_rate": 0.0001711982743834687,
      "loss": 3.7138,
      "step": 154490
    },
    {
      "epoch": 0.1440189993219688,
      "grad_norm": 1.0504553318023682,
      "learning_rate": 0.0001711964100613735,
      "loss": 3.5525,
      "step": 154500
    },
    {
      "epoch": 0.1440283209400479,
      "grad_norm": 0.9478661417961121,
      "learning_rate": 0.00017119454573927829,
      "loss": 3.5029,
      "step": 154510
    },
    {
      "epoch": 0.14403764255812698,
      "grad_norm": 1.0469022989273071,
      "learning_rate": 0.0001711926814171831,
      "loss": 3.6408,
      "step": 154520
    },
    {
      "epoch": 0.14404696417620608,
      "grad_norm": 1.0542494058609009,
      "learning_rate": 0.00017119081709508787,
      "loss": 3.6753,
      "step": 154530
    },
    {
      "epoch": 0.14405628579428517,
      "grad_norm": 1.099100112915039,
      "learning_rate": 0.0001711889527729927,
      "loss": 3.5794,
      "step": 154540
    },
    {
      "epoch": 0.14406560741236427,
      "grad_norm": 1.1227331161499023,
      "learning_rate": 0.0001711870884508975,
      "loss": 3.6624,
      "step": 154550
    },
    {
      "epoch": 0.14407492903044336,
      "grad_norm": 1.3517693281173706,
      "learning_rate": 0.0001711852241288023,
      "loss": 3.472,
      "step": 154560
    },
    {
      "epoch": 0.14408425064852246,
      "grad_norm": 0.9884942770004272,
      "learning_rate": 0.0001711833598067071,
      "loss": 3.5797,
      "step": 154570
    },
    {
      "epoch": 0.14409357226660155,
      "grad_norm": 0.9978606104850769,
      "learning_rate": 0.00017118149548461188,
      "loss": 3.5095,
      "step": 154580
    },
    {
      "epoch": 0.14410289388468064,
      "grad_norm": 0.9167493581771851,
      "learning_rate": 0.00017117963116251668,
      "loss": 3.7169,
      "step": 154590
    },
    {
      "epoch": 0.1441122155027597,
      "grad_norm": 1.0249369144439697,
      "learning_rate": 0.00017117776684042152,
      "loss": 3.5291,
      "step": 154600
    },
    {
      "epoch": 0.1441215371208388,
      "grad_norm": 1.502007246017456,
      "learning_rate": 0.0001711759025183263,
      "loss": 3.5136,
      "step": 154610
    },
    {
      "epoch": 0.1441308587389179,
      "grad_norm": 1.0946470499038696,
      "learning_rate": 0.0001711740381962311,
      "loss": 3.4661,
      "step": 154620
    },
    {
      "epoch": 0.144140180356997,
      "grad_norm": 0.9849655628204346,
      "learning_rate": 0.00017117217387413588,
      "loss": 3.5092,
      "step": 154630
    },
    {
      "epoch": 0.1441495019750761,
      "grad_norm": 0.9713621735572815,
      "learning_rate": 0.0001711703095520407,
      "loss": 3.5543,
      "step": 154640
    },
    {
      "epoch": 0.14415882359315518,
      "grad_norm": 1.2061043977737427,
      "learning_rate": 0.0001711684452299455,
      "loss": 3.5111,
      "step": 154650
    },
    {
      "epoch": 0.14416814521123428,
      "grad_norm": 1.0030481815338135,
      "learning_rate": 0.0001711665809078503,
      "loss": 3.492,
      "step": 154660
    },
    {
      "epoch": 0.14417746682931337,
      "grad_norm": 1.0405757427215576,
      "learning_rate": 0.0001711647165857551,
      "loss": 3.3716,
      "step": 154670
    },
    {
      "epoch": 0.14418678844739247,
      "grad_norm": 0.9920593500137329,
      "learning_rate": 0.00017116285226365989,
      "loss": 3.6854,
      "step": 154680
    },
    {
      "epoch": 0.14419611006547156,
      "grad_norm": 1.0778264999389648,
      "learning_rate": 0.0001711609879415647,
      "loss": 3.627,
      "step": 154690
    },
    {
      "epoch": 0.14420543168355063,
      "grad_norm": 0.9811165928840637,
      "learning_rate": 0.0001711591236194695,
      "loss": 3.2945,
      "step": 154700
    },
    {
      "epoch": 0.14421475330162972,
      "grad_norm": 0.9127935171127319,
      "learning_rate": 0.0001711572592973743,
      "loss": 3.5388,
      "step": 154710
    },
    {
      "epoch": 0.14422407491970882,
      "grad_norm": 1.2056564092636108,
      "learning_rate": 0.0001711553949752791,
      "loss": 3.6291,
      "step": 154720
    },
    {
      "epoch": 0.1442333965377879,
      "grad_norm": 1.0967992544174194,
      "learning_rate": 0.0001711535306531839,
      "loss": 3.7551,
      "step": 154730
    },
    {
      "epoch": 0.144242718155867,
      "grad_norm": 1.143569827079773,
      "learning_rate": 0.0001711516663310887,
      "loss": 3.5298,
      "step": 154740
    },
    {
      "epoch": 0.1442520397739461,
      "grad_norm": 1.0565141439437866,
      "learning_rate": 0.00017114980200899348,
      "loss": 3.6012,
      "step": 154750
    },
    {
      "epoch": 0.1442613613920252,
      "grad_norm": 1.171521544456482,
      "learning_rate": 0.0001711479376868983,
      "loss": 3.6385,
      "step": 154760
    },
    {
      "epoch": 0.1442706830101043,
      "grad_norm": 1.0740084648132324,
      "learning_rate": 0.00017114607336480311,
      "loss": 3.62,
      "step": 154770
    },
    {
      "epoch": 0.14428000462818338,
      "grad_norm": 0.8790788054466248,
      "learning_rate": 0.0001711442090427079,
      "loss": 3.7399,
      "step": 154780
    },
    {
      "epoch": 0.14428932624626248,
      "grad_norm": 0.9078967571258545,
      "learning_rate": 0.0001711423447206127,
      "loss": 3.4725,
      "step": 154790
    },
    {
      "epoch": 0.14429864786434157,
      "grad_norm": 1.0572892427444458,
      "learning_rate": 0.00017114048039851748,
      "loss": 3.5262,
      "step": 154800
    },
    {
      "epoch": 0.14430796948242064,
      "grad_norm": 1.2241779565811157,
      "learning_rate": 0.00017113861607642229,
      "loss": 3.6357,
      "step": 154810
    },
    {
      "epoch": 0.14431729110049973,
      "grad_norm": 1.0594626665115356,
      "learning_rate": 0.00017113675175432712,
      "loss": 3.5632,
      "step": 154820
    },
    {
      "epoch": 0.14432661271857883,
      "grad_norm": 0.9793718457221985,
      "learning_rate": 0.0001711348874322319,
      "loss": 3.5184,
      "step": 154830
    },
    {
      "epoch": 0.14433593433665792,
      "grad_norm": 1.1173138618469238,
      "learning_rate": 0.0001711330231101367,
      "loss": 3.558,
      "step": 154840
    },
    {
      "epoch": 0.14434525595473702,
      "grad_norm": 1.2120680809020996,
      "learning_rate": 0.00017113115878804148,
      "loss": 3.5927,
      "step": 154850
    },
    {
      "epoch": 0.1443545775728161,
      "grad_norm": 1.0218862295150757,
      "learning_rate": 0.0001711292944659463,
      "loss": 3.6402,
      "step": 154860
    },
    {
      "epoch": 0.1443638991908952,
      "grad_norm": 0.8864419460296631,
      "learning_rate": 0.0001711274301438511,
      "loss": 3.5862,
      "step": 154870
    },
    {
      "epoch": 0.1443732208089743,
      "grad_norm": 0.9717258810997009,
      "learning_rate": 0.0001711255658217559,
      "loss": 3.6302,
      "step": 154880
    },
    {
      "epoch": 0.1443825424270534,
      "grad_norm": 0.965900182723999,
      "learning_rate": 0.0001711237014996607,
      "loss": 3.6488,
      "step": 154890
    },
    {
      "epoch": 0.1443918640451325,
      "grad_norm": 1.004298210144043,
      "learning_rate": 0.0001711218371775655,
      "loss": 3.4222,
      "step": 154900
    },
    {
      "epoch": 0.14440118566321156,
      "grad_norm": 1.1228832006454468,
      "learning_rate": 0.0001711199728554703,
      "loss": 3.6658,
      "step": 154910
    },
    {
      "epoch": 0.14441050728129065,
      "grad_norm": 0.9383174180984497,
      "learning_rate": 0.0001711181085333751,
      "loss": 3.7226,
      "step": 154920
    },
    {
      "epoch": 0.14441982889936975,
      "grad_norm": 0.9826633334159851,
      "learning_rate": 0.0001711162442112799,
      "loss": 3.6718,
      "step": 154930
    },
    {
      "epoch": 0.14442915051744884,
      "grad_norm": 1.0270881652832031,
      "learning_rate": 0.00017111437988918471,
      "loss": 3.3813,
      "step": 154940
    },
    {
      "epoch": 0.14443847213552793,
      "grad_norm": 0.9650743007659912,
      "learning_rate": 0.0001711125155670895,
      "loss": 3.639,
      "step": 154950
    },
    {
      "epoch": 0.14444779375360703,
      "grad_norm": 1.0008254051208496,
      "learning_rate": 0.0001711106512449943,
      "loss": 3.5325,
      "step": 154960
    },
    {
      "epoch": 0.14445711537168612,
      "grad_norm": 1.1811476945877075,
      "learning_rate": 0.0001711087869228991,
      "loss": 3.6358,
      "step": 154970
    },
    {
      "epoch": 0.14446643698976522,
      "grad_norm": 0.9996612668037415,
      "learning_rate": 0.0001711069226008039,
      "loss": 3.5903,
      "step": 154980
    },
    {
      "epoch": 0.1444757586078443,
      "grad_norm": 1.2509764432907104,
      "learning_rate": 0.00017110505827870872,
      "loss": 3.6654,
      "step": 154990
    },
    {
      "epoch": 0.1444850802259234,
      "grad_norm": 1.0271555185317993,
      "learning_rate": 0.0001711031939566135,
      "loss": 3.5399,
      "step": 155000
    },
    {
      "epoch": 0.1444944018440025,
      "grad_norm": 0.9835241436958313,
      "learning_rate": 0.0001711013296345183,
      "loss": 3.6598,
      "step": 155010
    },
    {
      "epoch": 0.14450372346208157,
      "grad_norm": 1.0706919431686401,
      "learning_rate": 0.0001710994653124231,
      "loss": 3.4911,
      "step": 155020
    },
    {
      "epoch": 0.14451304508016066,
      "grad_norm": 1.1103500127792358,
      "learning_rate": 0.0001710976009903279,
      "loss": 3.422,
      "step": 155030
    },
    {
      "epoch": 0.14452236669823976,
      "grad_norm": 1.1969425678253174,
      "learning_rate": 0.0001710957366682327,
      "loss": 3.8154,
      "step": 155040
    },
    {
      "epoch": 0.14453168831631885,
      "grad_norm": 1.074472427368164,
      "learning_rate": 0.0001710938723461375,
      "loss": 3.7229,
      "step": 155050
    },
    {
      "epoch": 0.14454100993439795,
      "grad_norm": 0.9452652335166931,
      "learning_rate": 0.0001710920080240423,
      "loss": 3.5535,
      "step": 155060
    },
    {
      "epoch": 0.14455033155247704,
      "grad_norm": 1.1122349500656128,
      "learning_rate": 0.00017109014370194711,
      "loss": 3.588,
      "step": 155070
    },
    {
      "epoch": 0.14455965317055613,
      "grad_norm": 0.8967034220695496,
      "learning_rate": 0.0001710882793798519,
      "loss": 3.5338,
      "step": 155080
    },
    {
      "epoch": 0.14456897478863523,
      "grad_norm": 0.897635817527771,
      "learning_rate": 0.0001710864150577567,
      "loss": 3.5994,
      "step": 155090
    },
    {
      "epoch": 0.14457829640671432,
      "grad_norm": 1.0328342914581299,
      "learning_rate": 0.0001710845507356615,
      "loss": 3.6487,
      "step": 155100
    },
    {
      "epoch": 0.14458761802479342,
      "grad_norm": 1.1884210109710693,
      "learning_rate": 0.0001710826864135663,
      "loss": 3.5038,
      "step": 155110
    },
    {
      "epoch": 0.14459693964287249,
      "grad_norm": 0.9031968116760254,
      "learning_rate": 0.00017108082209147112,
      "loss": 3.4795,
      "step": 155120
    },
    {
      "epoch": 0.14460626126095158,
      "grad_norm": 1.171974778175354,
      "learning_rate": 0.0001710789577693759,
      "loss": 3.3789,
      "step": 155130
    },
    {
      "epoch": 0.14461558287903067,
      "grad_norm": 1.0546735525131226,
      "learning_rate": 0.0001710770934472807,
      "loss": 3.4937,
      "step": 155140
    },
    {
      "epoch": 0.14462490449710977,
      "grad_norm": 1.1237800121307373,
      "learning_rate": 0.0001710752291251855,
      "loss": 3.418,
      "step": 155150
    },
    {
      "epoch": 0.14463422611518886,
      "grad_norm": 1.0276302099227905,
      "learning_rate": 0.00017107336480309032,
      "loss": 3.6508,
      "step": 155160
    },
    {
      "epoch": 0.14464354773326796,
      "grad_norm": 1.014435887336731,
      "learning_rate": 0.00017107150048099512,
      "loss": 3.725,
      "step": 155170
    },
    {
      "epoch": 0.14465286935134705,
      "grad_norm": 1.125822901725769,
      "learning_rate": 0.0001710696361588999,
      "loss": 3.5853,
      "step": 155180
    },
    {
      "epoch": 0.14466219096942615,
      "grad_norm": 1.1950304508209229,
      "learning_rate": 0.0001710677718368047,
      "loss": 3.5864,
      "step": 155190
    },
    {
      "epoch": 0.14467151258750524,
      "grad_norm": 1.329352617263794,
      "learning_rate": 0.0001710659075147095,
      "loss": 3.612,
      "step": 155200
    },
    {
      "epoch": 0.14468083420558434,
      "grad_norm": 1.0557538270950317,
      "learning_rate": 0.00017106404319261432,
      "loss": 3.8789,
      "step": 155210
    },
    {
      "epoch": 0.14469015582366343,
      "grad_norm": 0.9831245541572571,
      "learning_rate": 0.00017106217887051913,
      "loss": 3.5304,
      "step": 155220
    },
    {
      "epoch": 0.1446994774417425,
      "grad_norm": 1.1041826009750366,
      "learning_rate": 0.0001710603145484239,
      "loss": 3.4298,
      "step": 155230
    },
    {
      "epoch": 0.1447087990598216,
      "grad_norm": 1.158353328704834,
      "learning_rate": 0.0001710584502263287,
      "loss": 3.5239,
      "step": 155240
    },
    {
      "epoch": 0.14471812067790069,
      "grad_norm": 1.406095027923584,
      "learning_rate": 0.0001710565859042335,
      "loss": 3.5127,
      "step": 155250
    },
    {
      "epoch": 0.14472744229597978,
      "grad_norm": 1.0315552949905396,
      "learning_rate": 0.0001710547215821383,
      "loss": 3.643,
      "step": 155260
    },
    {
      "epoch": 0.14473676391405887,
      "grad_norm": 0.9739618897438049,
      "learning_rate": 0.00017105285726004313,
      "loss": 3.5367,
      "step": 155270
    },
    {
      "epoch": 0.14474608553213797,
      "grad_norm": 0.9850986003875732,
      "learning_rate": 0.0001710509929379479,
      "loss": 3.6214,
      "step": 155280
    },
    {
      "epoch": 0.14475540715021706,
      "grad_norm": 1.1448719501495361,
      "learning_rate": 0.00017104912861585272,
      "loss": 3.5055,
      "step": 155290
    },
    {
      "epoch": 0.14476472876829616,
      "grad_norm": 0.925722062587738,
      "learning_rate": 0.0001710472642937575,
      "loss": 3.6284,
      "step": 155300
    },
    {
      "epoch": 0.14477405038637525,
      "grad_norm": 1.1493927240371704,
      "learning_rate": 0.0001710453999716623,
      "loss": 3.6134,
      "step": 155310
    },
    {
      "epoch": 0.14478337200445435,
      "grad_norm": 1.0115233659744263,
      "learning_rate": 0.0001710435356495671,
      "loss": 3.6988,
      "step": 155320
    },
    {
      "epoch": 0.1447926936225334,
      "grad_norm": 1.010100245475769,
      "learning_rate": 0.00017104167132747192,
      "loss": 3.6993,
      "step": 155330
    },
    {
      "epoch": 0.1448020152406125,
      "grad_norm": 1.2016401290893555,
      "learning_rate": 0.00017103980700537672,
      "loss": 3.5695,
      "step": 155340
    },
    {
      "epoch": 0.1448113368586916,
      "grad_norm": 0.9830188155174255,
      "learning_rate": 0.0001710379426832815,
      "loss": 3.8568,
      "step": 155350
    },
    {
      "epoch": 0.1448206584767707,
      "grad_norm": 0.8683576583862305,
      "learning_rate": 0.0001710360783611863,
      "loss": 3.3485,
      "step": 155360
    },
    {
      "epoch": 0.1448299800948498,
      "grad_norm": 1.1652276515960693,
      "learning_rate": 0.00017103421403909111,
      "loss": 3.6293,
      "step": 155370
    },
    {
      "epoch": 0.1448393017129289,
      "grad_norm": 1.1073294878005981,
      "learning_rate": 0.00017103234971699592,
      "loss": 3.6773,
      "step": 155380
    },
    {
      "epoch": 0.14484862333100798,
      "grad_norm": 1.1399030685424805,
      "learning_rate": 0.00017103048539490073,
      "loss": 3.7579,
      "step": 155390
    },
    {
      "epoch": 0.14485794494908708,
      "grad_norm": 1.2062815427780151,
      "learning_rate": 0.0001710286210728055,
      "loss": 3.5424,
      "step": 155400
    },
    {
      "epoch": 0.14486726656716617,
      "grad_norm": 1.0127387046813965,
      "learning_rate": 0.0001710267567507103,
      "loss": 3.509,
      "step": 155410
    },
    {
      "epoch": 0.14487658818524526,
      "grad_norm": 0.9949168562889099,
      "learning_rate": 0.00017102489242861512,
      "loss": 3.7866,
      "step": 155420
    },
    {
      "epoch": 0.14488590980332436,
      "grad_norm": 0.9298756718635559,
      "learning_rate": 0.00017102302810651992,
      "loss": 3.5899,
      "step": 155430
    },
    {
      "epoch": 0.14489523142140343,
      "grad_norm": 1.088356614112854,
      "learning_rate": 0.00017102116378442473,
      "loss": 3.5773,
      "step": 155440
    },
    {
      "epoch": 0.14490455303948252,
      "grad_norm": 0.974055290222168,
      "learning_rate": 0.0001710192994623295,
      "loss": 3.5154,
      "step": 155450
    },
    {
      "epoch": 0.14491387465756161,
      "grad_norm": 1.0319206714630127,
      "learning_rate": 0.00017101743514023432,
      "loss": 3.5756,
      "step": 155460
    },
    {
      "epoch": 0.1449231962756407,
      "grad_norm": 1.1304774284362793,
      "learning_rate": 0.00017101557081813912,
      "loss": 3.3826,
      "step": 155470
    },
    {
      "epoch": 0.1449325178937198,
      "grad_norm": 1.0198512077331543,
      "learning_rate": 0.0001710137064960439,
      "loss": 3.3345,
      "step": 155480
    },
    {
      "epoch": 0.1449418395117989,
      "grad_norm": 1.1455835103988647,
      "learning_rate": 0.00017101184217394874,
      "loss": 3.6233,
      "step": 155490
    },
    {
      "epoch": 0.144951161129878,
      "grad_norm": 1.0483683347702026,
      "learning_rate": 0.00017100997785185351,
      "loss": 3.7918,
      "step": 155500
    },
    {
      "epoch": 0.1449604827479571,
      "grad_norm": 1.0771986246109009,
      "learning_rate": 0.00017100811352975832,
      "loss": 3.7649,
      "step": 155510
    },
    {
      "epoch": 0.14496980436603618,
      "grad_norm": 0.9908225536346436,
      "learning_rate": 0.00017100624920766313,
      "loss": 3.8782,
      "step": 155520
    },
    {
      "epoch": 0.14497912598411528,
      "grad_norm": 1.3050884008407593,
      "learning_rate": 0.0001710043848855679,
      "loss": 3.3932,
      "step": 155530
    },
    {
      "epoch": 0.14498844760219434,
      "grad_norm": 1.1057075262069702,
      "learning_rate": 0.0001710025205634727,
      "loss": 3.6508,
      "step": 155540
    },
    {
      "epoch": 0.14499776922027344,
      "grad_norm": 0.9873929023742676,
      "learning_rate": 0.00017100065624137752,
      "loss": 3.5178,
      "step": 155550
    },
    {
      "epoch": 0.14500709083835253,
      "grad_norm": 1.1912646293640137,
      "learning_rate": 0.00017099879191928233,
      "loss": 3.5827,
      "step": 155560
    },
    {
      "epoch": 0.14501641245643163,
      "grad_norm": 1.0701428651809692,
      "learning_rate": 0.00017099692759718713,
      "loss": 3.6605,
      "step": 155570
    },
    {
      "epoch": 0.14502573407451072,
      "grad_norm": 1.1009583473205566,
      "learning_rate": 0.0001709950632750919,
      "loss": 3.7426,
      "step": 155580
    },
    {
      "epoch": 0.14503505569258981,
      "grad_norm": 1.065053939819336,
      "learning_rate": 0.00017099319895299672,
      "loss": 3.8312,
      "step": 155590
    },
    {
      "epoch": 0.1450443773106689,
      "grad_norm": 1.0583018064498901,
      "learning_rate": 0.00017099133463090152,
      "loss": 3.6122,
      "step": 155600
    },
    {
      "epoch": 0.145053698928748,
      "grad_norm": 1.164020299911499,
      "learning_rate": 0.00017098947030880633,
      "loss": 3.5005,
      "step": 155610
    },
    {
      "epoch": 0.1450630205468271,
      "grad_norm": 1.0292507410049438,
      "learning_rate": 0.00017098760598671114,
      "loss": 3.4881,
      "step": 155620
    },
    {
      "epoch": 0.1450723421649062,
      "grad_norm": 0.9485761523246765,
      "learning_rate": 0.00017098574166461592,
      "loss": 3.6932,
      "step": 155630
    },
    {
      "epoch": 0.1450816637829853,
      "grad_norm": 0.9652391672134399,
      "learning_rate": 0.00017098387734252072,
      "loss": 3.6246,
      "step": 155640
    },
    {
      "epoch": 0.14509098540106435,
      "grad_norm": 0.9378554821014404,
      "learning_rate": 0.0001709820130204255,
      "loss": 3.622,
      "step": 155650
    },
    {
      "epoch": 0.14510030701914345,
      "grad_norm": 1.0930722951889038,
      "learning_rate": 0.00017098014869833033,
      "loss": 3.5657,
      "step": 155660
    },
    {
      "epoch": 0.14510962863722254,
      "grad_norm": 1.4249895811080933,
      "learning_rate": 0.00017097828437623514,
      "loss": 3.755,
      "step": 155670
    },
    {
      "epoch": 0.14511895025530164,
      "grad_norm": 1.1185815334320068,
      "learning_rate": 0.00017097642005413992,
      "loss": 3.5969,
      "step": 155680
    },
    {
      "epoch": 0.14512827187338073,
      "grad_norm": 1.0553245544433594,
      "learning_rate": 0.00017097455573204473,
      "loss": 3.6563,
      "step": 155690
    },
    {
      "epoch": 0.14513759349145983,
      "grad_norm": 0.9642137885093689,
      "learning_rate": 0.0001709726914099495,
      "loss": 3.3935,
      "step": 155700
    },
    {
      "epoch": 0.14514691510953892,
      "grad_norm": 1.0435887575149536,
      "learning_rate": 0.0001709708270878543,
      "loss": 3.57,
      "step": 155710
    },
    {
      "epoch": 0.14515623672761802,
      "grad_norm": 1.1062238216400146,
      "learning_rate": 0.00017096896276575915,
      "loss": 3.3157,
      "step": 155720
    },
    {
      "epoch": 0.1451655583456971,
      "grad_norm": 0.9973095655441284,
      "learning_rate": 0.00017096709844366392,
      "loss": 3.7488,
      "step": 155730
    },
    {
      "epoch": 0.1451748799637762,
      "grad_norm": 1.0482699871063232,
      "learning_rate": 0.00017096523412156873,
      "loss": 3.5202,
      "step": 155740
    },
    {
      "epoch": 0.14518420158185527,
      "grad_norm": 1.0881421566009521,
      "learning_rate": 0.0001709633697994735,
      "loss": 3.628,
      "step": 155750
    },
    {
      "epoch": 0.14519352319993437,
      "grad_norm": 0.9999487996101379,
      "learning_rate": 0.00017096150547737832,
      "loss": 3.5353,
      "step": 155760
    },
    {
      "epoch": 0.14520284481801346,
      "grad_norm": 1.0961378812789917,
      "learning_rate": 0.00017095964115528312,
      "loss": 3.4654,
      "step": 155770
    },
    {
      "epoch": 0.14521216643609255,
      "grad_norm": 0.8605664372444153,
      "learning_rate": 0.00017095777683318793,
      "loss": 3.5915,
      "step": 155780
    },
    {
      "epoch": 0.14522148805417165,
      "grad_norm": 0.9877215623855591,
      "learning_rate": 0.00017095591251109273,
      "loss": 3.548,
      "step": 155790
    },
    {
      "epoch": 0.14523080967225074,
      "grad_norm": 0.9560272097587585,
      "learning_rate": 0.00017095404818899751,
      "loss": 3.5748,
      "step": 155800
    },
    {
      "epoch": 0.14524013129032984,
      "grad_norm": 1.1618387699127197,
      "learning_rate": 0.00017095218386690232,
      "loss": 3.6555,
      "step": 155810
    },
    {
      "epoch": 0.14524945290840893,
      "grad_norm": 1.015721321105957,
      "learning_rate": 0.00017095031954480713,
      "loss": 3.4644,
      "step": 155820
    },
    {
      "epoch": 0.14525877452648803,
      "grad_norm": 1.1333097219467163,
      "learning_rate": 0.00017094845522271193,
      "loss": 3.5536,
      "step": 155830
    },
    {
      "epoch": 0.14526809614456712,
      "grad_norm": 0.8786084651947021,
      "learning_rate": 0.00017094659090061674,
      "loss": 3.515,
      "step": 155840
    },
    {
      "epoch": 0.14527741776264622,
      "grad_norm": 0.8452629446983337,
      "learning_rate": 0.00017094472657852152,
      "loss": 3.4971,
      "step": 155850
    },
    {
      "epoch": 0.14528673938072528,
      "grad_norm": 0.9725555777549744,
      "learning_rate": 0.00017094286225642632,
      "loss": 3.4511,
      "step": 155860
    },
    {
      "epoch": 0.14529606099880438,
      "grad_norm": 0.9287564158439636,
      "learning_rate": 0.00017094099793433113,
      "loss": 3.4822,
      "step": 155870
    },
    {
      "epoch": 0.14530538261688347,
      "grad_norm": 1.0329110622406006,
      "learning_rate": 0.00017093913361223594,
      "loss": 3.6645,
      "step": 155880
    },
    {
      "epoch": 0.14531470423496257,
      "grad_norm": 1.0396225452423096,
      "learning_rate": 0.00017093726929014074,
      "loss": 3.6356,
      "step": 155890
    },
    {
      "epoch": 0.14532402585304166,
      "grad_norm": 1.2422146797180176,
      "learning_rate": 0.00017093540496804552,
      "loss": 3.4918,
      "step": 155900
    },
    {
      "epoch": 0.14533334747112076,
      "grad_norm": 1.1187050342559814,
      "learning_rate": 0.00017093354064595033,
      "loss": 3.5554,
      "step": 155910
    },
    {
      "epoch": 0.14534266908919985,
      "grad_norm": 1.0632990598678589,
      "learning_rate": 0.00017093167632385514,
      "loss": 3.5501,
      "step": 155920
    },
    {
      "epoch": 0.14535199070727894,
      "grad_norm": 1.089569330215454,
      "learning_rate": 0.00017092981200175991,
      "loss": 3.5767,
      "step": 155930
    },
    {
      "epoch": 0.14536131232535804,
      "grad_norm": 0.8778188228607178,
      "learning_rate": 0.00017092794767966475,
      "loss": 3.7371,
      "step": 155940
    },
    {
      "epoch": 0.14537063394343713,
      "grad_norm": 1.1707614660263062,
      "learning_rate": 0.00017092608335756953,
      "loss": 3.7594,
      "step": 155950
    },
    {
      "epoch": 0.1453799555615162,
      "grad_norm": 1.168842077255249,
      "learning_rate": 0.00017092421903547433,
      "loss": 3.6598,
      "step": 155960
    },
    {
      "epoch": 0.1453892771795953,
      "grad_norm": 1.2234008312225342,
      "learning_rate": 0.00017092235471337914,
      "loss": 3.526,
      "step": 155970
    },
    {
      "epoch": 0.1453985987976744,
      "grad_norm": 1.016330599784851,
      "learning_rate": 0.00017092049039128392,
      "loss": 3.4477,
      "step": 155980
    },
    {
      "epoch": 0.14540792041575348,
      "grad_norm": 0.9862827658653259,
      "learning_rate": 0.00017091862606918873,
      "loss": 3.4059,
      "step": 155990
    },
    {
      "epoch": 0.14541724203383258,
      "grad_norm": 1.0010676383972168,
      "learning_rate": 0.00017091676174709353,
      "loss": 3.5921,
      "step": 156000
    },
    {
      "epoch": 0.14542656365191167,
      "grad_norm": 0.991430401802063,
      "learning_rate": 0.00017091489742499834,
      "loss": 3.5853,
      "step": 156010
    },
    {
      "epoch": 0.14543588526999077,
      "grad_norm": 1.30889892578125,
      "learning_rate": 0.00017091303310290312,
      "loss": 3.6943,
      "step": 156020
    },
    {
      "epoch": 0.14544520688806986,
      "grad_norm": 1.0736794471740723,
      "learning_rate": 0.00017091116878080792,
      "loss": 3.527,
      "step": 156030
    },
    {
      "epoch": 0.14545452850614896,
      "grad_norm": 0.9031925797462463,
      "learning_rate": 0.00017090930445871273,
      "loss": 3.4374,
      "step": 156040
    },
    {
      "epoch": 0.14546385012422805,
      "grad_norm": 1.157313346862793,
      "learning_rate": 0.00017090744013661754,
      "loss": 3.5054,
      "step": 156050
    },
    {
      "epoch": 0.14547317174230712,
      "grad_norm": 1.1220475435256958,
      "learning_rate": 0.00017090557581452234,
      "loss": 3.7429,
      "step": 156060
    },
    {
      "epoch": 0.1454824933603862,
      "grad_norm": 1.0391765832901,
      "learning_rate": 0.00017090371149242712,
      "loss": 3.4411,
      "step": 156070
    },
    {
      "epoch": 0.1454918149784653,
      "grad_norm": 1.0812103748321533,
      "learning_rate": 0.00017090184717033193,
      "loss": 3.5961,
      "step": 156080
    },
    {
      "epoch": 0.1455011365965444,
      "grad_norm": 1.0957211256027222,
      "learning_rate": 0.00017089998284823673,
      "loss": 3.5828,
      "step": 156090
    },
    {
      "epoch": 0.1455104582146235,
      "grad_norm": 0.9206916093826294,
      "learning_rate": 0.00017089811852614151,
      "loss": 3.6681,
      "step": 156100
    },
    {
      "epoch": 0.1455197798327026,
      "grad_norm": 0.881607711315155,
      "learning_rate": 0.00017089625420404635,
      "loss": 3.6729,
      "step": 156110
    },
    {
      "epoch": 0.14552910145078168,
      "grad_norm": 1.0746254920959473,
      "learning_rate": 0.00017089438988195113,
      "loss": 3.5639,
      "step": 156120
    },
    {
      "epoch": 0.14553842306886078,
      "grad_norm": 1.2327059507369995,
      "learning_rate": 0.00017089252555985593,
      "loss": 3.4305,
      "step": 156130
    },
    {
      "epoch": 0.14554774468693987,
      "grad_norm": 0.9504896998405457,
      "learning_rate": 0.00017089066123776074,
      "loss": 3.7849,
      "step": 156140
    },
    {
      "epoch": 0.14555706630501897,
      "grad_norm": 1.0733505487442017,
      "learning_rate": 0.00017088879691566552,
      "loss": 3.4762,
      "step": 156150
    },
    {
      "epoch": 0.14556638792309806,
      "grad_norm": 1.1032339334487915,
      "learning_rate": 0.00017088693259357032,
      "loss": 3.5828,
      "step": 156160
    },
    {
      "epoch": 0.14557570954117713,
      "grad_norm": 0.9108404517173767,
      "learning_rate": 0.00017088506827147513,
      "loss": 3.7092,
      "step": 156170
    },
    {
      "epoch": 0.14558503115925622,
      "grad_norm": 1.0935254096984863,
      "learning_rate": 0.00017088320394937994,
      "loss": 3.553,
      "step": 156180
    },
    {
      "epoch": 0.14559435277733532,
      "grad_norm": 1.2798879146575928,
      "learning_rate": 0.00017088133962728474,
      "loss": 3.5509,
      "step": 156190
    },
    {
      "epoch": 0.1456036743954144,
      "grad_norm": 1.208146333694458,
      "learning_rate": 0.00017087947530518952,
      "loss": 3.7013,
      "step": 156200
    },
    {
      "epoch": 0.1456129960134935,
      "grad_norm": 1.1446642875671387,
      "learning_rate": 0.00017087761098309433,
      "loss": 3.5162,
      "step": 156210
    },
    {
      "epoch": 0.1456223176315726,
      "grad_norm": 1.1278350353240967,
      "learning_rate": 0.00017087574666099914,
      "loss": 3.577,
      "step": 156220
    },
    {
      "epoch": 0.1456316392496517,
      "grad_norm": 1.4960863590240479,
      "learning_rate": 0.00017087388233890394,
      "loss": 3.6755,
      "step": 156230
    },
    {
      "epoch": 0.1456409608677308,
      "grad_norm": 1.0986278057098389,
      "learning_rate": 0.00017087201801680875,
      "loss": 3.5056,
      "step": 156240
    },
    {
      "epoch": 0.14565028248580988,
      "grad_norm": 0.9998971819877625,
      "learning_rate": 0.00017087015369471353,
      "loss": 3.5153,
      "step": 156250
    },
    {
      "epoch": 0.14565960410388898,
      "grad_norm": 0.9831448793411255,
      "learning_rate": 0.00017086828937261833,
      "loss": 3.5882,
      "step": 156260
    },
    {
      "epoch": 0.14566892572196805,
      "grad_norm": 1.0843019485473633,
      "learning_rate": 0.00017086642505052314,
      "loss": 3.5849,
      "step": 156270
    },
    {
      "epoch": 0.14567824734004714,
      "grad_norm": 1.2373254299163818,
      "learning_rate": 0.00017086456072842795,
      "loss": 3.5912,
      "step": 156280
    },
    {
      "epoch": 0.14568756895812623,
      "grad_norm": 0.983066737651825,
      "learning_rate": 0.00017086269640633275,
      "loss": 3.538,
      "step": 156290
    },
    {
      "epoch": 0.14569689057620533,
      "grad_norm": 1.0754998922348022,
      "learning_rate": 0.00017086083208423753,
      "loss": 3.5338,
      "step": 156300
    },
    {
      "epoch": 0.14570621219428442,
      "grad_norm": 1.1105148792266846,
      "learning_rate": 0.00017085896776214234,
      "loss": 3.5946,
      "step": 156310
    },
    {
      "epoch": 0.14571553381236352,
      "grad_norm": 0.9065349698066711,
      "learning_rate": 0.00017085710344004712,
      "loss": 3.611,
      "step": 156320
    },
    {
      "epoch": 0.1457248554304426,
      "grad_norm": 1.138717532157898,
      "learning_rate": 0.00017085523911795195,
      "loss": 3.553,
      "step": 156330
    },
    {
      "epoch": 0.1457341770485217,
      "grad_norm": 1.0720939636230469,
      "learning_rate": 0.00017085337479585676,
      "loss": 3.5386,
      "step": 156340
    },
    {
      "epoch": 0.1457434986666008,
      "grad_norm": 1.0477887392044067,
      "learning_rate": 0.00017085151047376154,
      "loss": 3.6495,
      "step": 156350
    },
    {
      "epoch": 0.1457528202846799,
      "grad_norm": 1.2025737762451172,
      "learning_rate": 0.00017084964615166634,
      "loss": 3.4012,
      "step": 156360
    },
    {
      "epoch": 0.145762141902759,
      "grad_norm": 1.1826233863830566,
      "learning_rate": 0.00017084778182957112,
      "loss": 3.587,
      "step": 156370
    },
    {
      "epoch": 0.14577146352083806,
      "grad_norm": 0.9666048884391785,
      "learning_rate": 0.00017084591750747593,
      "loss": 3.4775,
      "step": 156380
    },
    {
      "epoch": 0.14578078513891715,
      "grad_norm": 1.1851730346679688,
      "learning_rate": 0.00017084405318538076,
      "loss": 3.6069,
      "step": 156390
    },
    {
      "epoch": 0.14579010675699625,
      "grad_norm": 1.210972785949707,
      "learning_rate": 0.00017084218886328554,
      "loss": 3.566,
      "step": 156400
    },
    {
      "epoch": 0.14579942837507534,
      "grad_norm": 1.2964036464691162,
      "learning_rate": 0.00017084032454119035,
      "loss": 3.5424,
      "step": 156410
    },
    {
      "epoch": 0.14580874999315444,
      "grad_norm": 1.2649132013320923,
      "learning_rate": 0.00017083846021909513,
      "loss": 3.4655,
      "step": 156420
    },
    {
      "epoch": 0.14581807161123353,
      "grad_norm": 1.0491366386413574,
      "learning_rate": 0.00017083659589699993,
      "loss": 3.5752,
      "step": 156430
    },
    {
      "epoch": 0.14582739322931262,
      "grad_norm": 1.1326959133148193,
      "learning_rate": 0.00017083473157490474,
      "loss": 3.5873,
      "step": 156440
    },
    {
      "epoch": 0.14583671484739172,
      "grad_norm": 1.0295217037200928,
      "learning_rate": 0.00017083286725280954,
      "loss": 3.651,
      "step": 156450
    },
    {
      "epoch": 0.1458460364654708,
      "grad_norm": 1.0959924459457397,
      "learning_rate": 0.00017083100293071435,
      "loss": 3.5354,
      "step": 156460
    },
    {
      "epoch": 0.1458553580835499,
      "grad_norm": 1.207111120223999,
      "learning_rate": 0.00017082913860861913,
      "loss": 3.7502,
      "step": 156470
    },
    {
      "epoch": 0.14586467970162897,
      "grad_norm": 0.8226934671401978,
      "learning_rate": 0.00017082727428652394,
      "loss": 3.5404,
      "step": 156480
    },
    {
      "epoch": 0.14587400131970807,
      "grad_norm": 0.9180687069892883,
      "learning_rate": 0.00017082540996442874,
      "loss": 3.6711,
      "step": 156490
    },
    {
      "epoch": 0.14588332293778716,
      "grad_norm": 1.1485940217971802,
      "learning_rate": 0.00017082354564233355,
      "loss": 3.7264,
      "step": 156500
    },
    {
      "epoch": 0.14589264455586626,
      "grad_norm": 1.135000467300415,
      "learning_rate": 0.00017082168132023836,
      "loss": 3.4168,
      "step": 156510
    },
    {
      "epoch": 0.14590196617394535,
      "grad_norm": 1.0115901231765747,
      "learning_rate": 0.00017081981699814313,
      "loss": 3.4772,
      "step": 156520
    },
    {
      "epoch": 0.14591128779202445,
      "grad_norm": 0.9944664239883423,
      "learning_rate": 0.00017081795267604794,
      "loss": 3.4892,
      "step": 156530
    },
    {
      "epoch": 0.14592060941010354,
      "grad_norm": 0.8359456062316895,
      "learning_rate": 0.00017081608835395275,
      "loss": 3.5754,
      "step": 156540
    },
    {
      "epoch": 0.14592993102818264,
      "grad_norm": 1.128246545791626,
      "learning_rate": 0.00017081422403185753,
      "loss": 3.4811,
      "step": 156550
    },
    {
      "epoch": 0.14593925264626173,
      "grad_norm": 1.022463321685791,
      "learning_rate": 0.00017081235970976236,
      "loss": 3.4987,
      "step": 156560
    },
    {
      "epoch": 0.14594857426434082,
      "grad_norm": 1.1597108840942383,
      "learning_rate": 0.00017081049538766714,
      "loss": 3.6059,
      "step": 156570
    },
    {
      "epoch": 0.14595789588241992,
      "grad_norm": 1.24759840965271,
      "learning_rate": 0.00017080863106557195,
      "loss": 3.4441,
      "step": 156580
    },
    {
      "epoch": 0.14596721750049899,
      "grad_norm": 1.080255150794983,
      "learning_rate": 0.00017080676674347675,
      "loss": 3.3813,
      "step": 156590
    },
    {
      "epoch": 0.14597653911857808,
      "grad_norm": 0.9546015858650208,
      "learning_rate": 0.00017080490242138153,
      "loss": 3.4553,
      "step": 156600
    },
    {
      "epoch": 0.14598586073665717,
      "grad_norm": 1.1120378971099854,
      "learning_rate": 0.00017080303809928634,
      "loss": 3.5506,
      "step": 156610
    },
    {
      "epoch": 0.14599518235473627,
      "grad_norm": 1.0194828510284424,
      "learning_rate": 0.00017080117377719114,
      "loss": 3.625,
      "step": 156620
    },
    {
      "epoch": 0.14600450397281536,
      "grad_norm": 1.0166966915130615,
      "learning_rate": 0.00017079930945509595,
      "loss": 3.7436,
      "step": 156630
    },
    {
      "epoch": 0.14601382559089446,
      "grad_norm": 1.1432156562805176,
      "learning_rate": 0.00017079744513300076,
      "loss": 3.472,
      "step": 156640
    },
    {
      "epoch": 0.14602314720897355,
      "grad_norm": 1.1222739219665527,
      "learning_rate": 0.00017079558081090554,
      "loss": 3.6413,
      "step": 156650
    },
    {
      "epoch": 0.14603246882705265,
      "grad_norm": 1.3707441091537476,
      "learning_rate": 0.00017079371648881034,
      "loss": 3.6014,
      "step": 156660
    },
    {
      "epoch": 0.14604179044513174,
      "grad_norm": 0.8703301548957825,
      "learning_rate": 0.00017079185216671515,
      "loss": 3.5198,
      "step": 156670
    },
    {
      "epoch": 0.14605111206321084,
      "grad_norm": 1.063389539718628,
      "learning_rate": 0.00017078998784461995,
      "loss": 3.6313,
      "step": 156680
    },
    {
      "epoch": 0.1460604336812899,
      "grad_norm": 1.1975888013839722,
      "learning_rate": 0.00017078812352252476,
      "loss": 3.4783,
      "step": 156690
    },
    {
      "epoch": 0.146069755299369,
      "grad_norm": 1.1575291156768799,
      "learning_rate": 0.00017078625920042954,
      "loss": 3.6429,
      "step": 156700
    },
    {
      "epoch": 0.1460790769174481,
      "grad_norm": 1.0767180919647217,
      "learning_rate": 0.00017078439487833435,
      "loss": 3.4705,
      "step": 156710
    },
    {
      "epoch": 0.1460883985355272,
      "grad_norm": 0.8925917744636536,
      "learning_rate": 0.00017078253055623915,
      "loss": 3.6941,
      "step": 156720
    },
    {
      "epoch": 0.14609772015360628,
      "grad_norm": 1.086414098739624,
      "learning_rate": 0.00017078066623414396,
      "loss": 3.6588,
      "step": 156730
    },
    {
      "epoch": 0.14610704177168538,
      "grad_norm": 1.000341773033142,
      "learning_rate": 0.00017077880191204877,
      "loss": 3.4356,
      "step": 156740
    },
    {
      "epoch": 0.14611636338976447,
      "grad_norm": 0.9533630013465881,
      "learning_rate": 0.00017077693758995354,
      "loss": 3.626,
      "step": 156750
    },
    {
      "epoch": 0.14612568500784356,
      "grad_norm": 1.0266377925872803,
      "learning_rate": 0.00017077507326785835,
      "loss": 3.4762,
      "step": 156760
    },
    {
      "epoch": 0.14613500662592266,
      "grad_norm": 1.0004359483718872,
      "learning_rate": 0.00017077320894576313,
      "loss": 3.518,
      "step": 156770
    },
    {
      "epoch": 0.14614432824400175,
      "grad_norm": 0.8203158974647522,
      "learning_rate": 0.00017077134462366796,
      "loss": 3.6842,
      "step": 156780
    },
    {
      "epoch": 0.14615364986208085,
      "grad_norm": 1.1872098445892334,
      "learning_rate": 0.00017076948030157277,
      "loss": 3.6327,
      "step": 156790
    },
    {
      "epoch": 0.14616297148015991,
      "grad_norm": 1.2224482297897339,
      "learning_rate": 0.00017076761597947755,
      "loss": 3.6957,
      "step": 156800
    },
    {
      "epoch": 0.146172293098239,
      "grad_norm": 1.2171605825424194,
      "learning_rate": 0.00017076575165738236,
      "loss": 3.5753,
      "step": 156810
    },
    {
      "epoch": 0.1461816147163181,
      "grad_norm": 1.1185765266418457,
      "learning_rate": 0.00017076388733528713,
      "loss": 3.7185,
      "step": 156820
    },
    {
      "epoch": 0.1461909363343972,
      "grad_norm": 1.025970458984375,
      "learning_rate": 0.00017076202301319194,
      "loss": 3.5456,
      "step": 156830
    },
    {
      "epoch": 0.1462002579524763,
      "grad_norm": 1.092461109161377,
      "learning_rate": 0.00017076015869109677,
      "loss": 3.7482,
      "step": 156840
    },
    {
      "epoch": 0.1462095795705554,
      "grad_norm": 0.9746646881103516,
      "learning_rate": 0.00017075829436900155,
      "loss": 3.8304,
      "step": 156850
    },
    {
      "epoch": 0.14621890118863448,
      "grad_norm": 1.034100890159607,
      "learning_rate": 0.00017075643004690636,
      "loss": 3.3259,
      "step": 156860
    },
    {
      "epoch": 0.14622822280671358,
      "grad_norm": 1.038664698600769,
      "learning_rate": 0.00017075456572481114,
      "loss": 3.6373,
      "step": 156870
    },
    {
      "epoch": 0.14623754442479267,
      "grad_norm": 0.9923307299613953,
      "learning_rate": 0.00017075270140271595,
      "loss": 3.5678,
      "step": 156880
    },
    {
      "epoch": 0.14624686604287176,
      "grad_norm": 1.077027678489685,
      "learning_rate": 0.00017075083708062075,
      "loss": 3.3121,
      "step": 156890
    },
    {
      "epoch": 0.14625618766095083,
      "grad_norm": 1.0448178052902222,
      "learning_rate": 0.00017074897275852556,
      "loss": 3.6777,
      "step": 156900
    },
    {
      "epoch": 0.14626550927902993,
      "grad_norm": 1.071974754333496,
      "learning_rate": 0.00017074710843643036,
      "loss": 3.5226,
      "step": 156910
    },
    {
      "epoch": 0.14627483089710902,
      "grad_norm": 1.0502160787582397,
      "learning_rate": 0.00017074524411433514,
      "loss": 3.6727,
      "step": 156920
    },
    {
      "epoch": 0.14628415251518812,
      "grad_norm": 1.0464978218078613,
      "learning_rate": 0.00017074337979223995,
      "loss": 3.5975,
      "step": 156930
    },
    {
      "epoch": 0.1462934741332672,
      "grad_norm": 1.1766278743743896,
      "learning_rate": 0.00017074151547014476,
      "loss": 3.6502,
      "step": 156940
    },
    {
      "epoch": 0.1463027957513463,
      "grad_norm": 0.8091748356819153,
      "learning_rate": 0.00017073965114804956,
      "loss": 3.5387,
      "step": 156950
    },
    {
      "epoch": 0.1463121173694254,
      "grad_norm": 1.159175157546997,
      "learning_rate": 0.00017073778682595437,
      "loss": 3.7022,
      "step": 156960
    },
    {
      "epoch": 0.1463214389875045,
      "grad_norm": 1.1724103689193726,
      "learning_rate": 0.00017073592250385915,
      "loss": 3.4529,
      "step": 156970
    },
    {
      "epoch": 0.1463307606055836,
      "grad_norm": 0.8747779130935669,
      "learning_rate": 0.00017073405818176395,
      "loss": 3.5401,
      "step": 156980
    },
    {
      "epoch": 0.14634008222366268,
      "grad_norm": 1.100091814994812,
      "learning_rate": 0.00017073219385966876,
      "loss": 3.6308,
      "step": 156990
    },
    {
      "epoch": 0.14634940384174178,
      "grad_norm": 1.0315120220184326,
      "learning_rate": 0.00017073032953757354,
      "loss": 3.6419,
      "step": 157000
    },
    {
      "epoch": 0.14635872545982084,
      "grad_norm": 0.9615439176559448,
      "learning_rate": 0.00017072846521547837,
      "loss": 3.6247,
      "step": 157010
    },
    {
      "epoch": 0.14636804707789994,
      "grad_norm": 1.0728814601898193,
      "learning_rate": 0.00017072660089338315,
      "loss": 3.5927,
      "step": 157020
    },
    {
      "epoch": 0.14637736869597903,
      "grad_norm": 1.1860852241516113,
      "learning_rate": 0.00017072473657128796,
      "loss": 3.486,
      "step": 157030
    },
    {
      "epoch": 0.14638669031405813,
      "grad_norm": 0.9508401155471802,
      "learning_rate": 0.00017072287224919276,
      "loss": 3.559,
      "step": 157040
    },
    {
      "epoch": 0.14639601193213722,
      "grad_norm": 1.0850449800491333,
      "learning_rate": 0.00017072100792709754,
      "loss": 3.4595,
      "step": 157050
    },
    {
      "epoch": 0.14640533355021632,
      "grad_norm": 1.081631064414978,
      "learning_rate": 0.00017071914360500235,
      "loss": 3.6165,
      "step": 157060
    },
    {
      "epoch": 0.1464146551682954,
      "grad_norm": 1.0660157203674316,
      "learning_rate": 0.00017071727928290716,
      "loss": 3.5618,
      "step": 157070
    },
    {
      "epoch": 0.1464239767863745,
      "grad_norm": 1.1905694007873535,
      "learning_rate": 0.00017071541496081196,
      "loss": 3.5331,
      "step": 157080
    },
    {
      "epoch": 0.1464332984044536,
      "grad_norm": 1.0745753049850464,
      "learning_rate": 0.00017071355063871677,
      "loss": 3.6993,
      "step": 157090
    },
    {
      "epoch": 0.1464426200225327,
      "grad_norm": 1.0045796632766724,
      "learning_rate": 0.00017071168631662155,
      "loss": 3.6272,
      "step": 157100
    },
    {
      "epoch": 0.14645194164061176,
      "grad_norm": 1.0042860507965088,
      "learning_rate": 0.00017070982199452635,
      "loss": 3.6704,
      "step": 157110
    },
    {
      "epoch": 0.14646126325869085,
      "grad_norm": 1.0742460489273071,
      "learning_rate": 0.00017070795767243116,
      "loss": 3.5576,
      "step": 157120
    },
    {
      "epoch": 0.14647058487676995,
      "grad_norm": 1.1034265756607056,
      "learning_rate": 0.00017070609335033597,
      "loss": 3.5156,
      "step": 157130
    },
    {
      "epoch": 0.14647990649484904,
      "grad_norm": 1.2480723857879639,
      "learning_rate": 0.00017070422902824077,
      "loss": 3.6884,
      "step": 157140
    },
    {
      "epoch": 0.14648922811292814,
      "grad_norm": 0.91895991563797,
      "learning_rate": 0.00017070236470614555,
      "loss": 3.6571,
      "step": 157150
    },
    {
      "epoch": 0.14649854973100723,
      "grad_norm": 0.8276588916778564,
      "learning_rate": 0.00017070050038405036,
      "loss": 3.8274,
      "step": 157160
    },
    {
      "epoch": 0.14650787134908633,
      "grad_norm": 0.9359267950057983,
      "learning_rate": 0.00017069863606195517,
      "loss": 3.6568,
      "step": 157170
    },
    {
      "epoch": 0.14651719296716542,
      "grad_norm": 1.148492455482483,
      "learning_rate": 0.00017069677173985997,
      "loss": 3.5582,
      "step": 157180
    },
    {
      "epoch": 0.14652651458524452,
      "grad_norm": 0.8610503673553467,
      "learning_rate": 0.00017069490741776478,
      "loss": 3.556,
      "step": 157190
    },
    {
      "epoch": 0.1465358362033236,
      "grad_norm": 1.03269362449646,
      "learning_rate": 0.00017069304309566956,
      "loss": 3.612,
      "step": 157200
    },
    {
      "epoch": 0.1465451578214027,
      "grad_norm": 1.3188340663909912,
      "learning_rate": 0.00017069117877357436,
      "loss": 3.4191,
      "step": 157210
    },
    {
      "epoch": 0.14655447943948177,
      "grad_norm": 0.8655105233192444,
      "learning_rate": 0.00017068931445147914,
      "loss": 3.7197,
      "step": 157220
    },
    {
      "epoch": 0.14656380105756087,
      "grad_norm": 1.0423427820205688,
      "learning_rate": 0.00017068745012938398,
      "loss": 3.365,
      "step": 157230
    },
    {
      "epoch": 0.14657312267563996,
      "grad_norm": 1.1043329238891602,
      "learning_rate": 0.00017068558580728878,
      "loss": 3.4611,
      "step": 157240
    },
    {
      "epoch": 0.14658244429371906,
      "grad_norm": 1.1031240224838257,
      "learning_rate": 0.00017068372148519356,
      "loss": 3.6541,
      "step": 157250
    },
    {
      "epoch": 0.14659176591179815,
      "grad_norm": 1.0512511730194092,
      "learning_rate": 0.00017068185716309837,
      "loss": 3.4655,
      "step": 157260
    },
    {
      "epoch": 0.14660108752987724,
      "grad_norm": 0.9303786754608154,
      "learning_rate": 0.00017067999284100315,
      "loss": 3.5405,
      "step": 157270
    },
    {
      "epoch": 0.14661040914795634,
      "grad_norm": 1.1513152122497559,
      "learning_rate": 0.00017067812851890795,
      "loss": 3.5457,
      "step": 157280
    },
    {
      "epoch": 0.14661973076603543,
      "grad_norm": 1.2866888046264648,
      "learning_rate": 0.0001706762641968128,
      "loss": 3.7195,
      "step": 157290
    },
    {
      "epoch": 0.14662905238411453,
      "grad_norm": 1.0469437837600708,
      "learning_rate": 0.00017067439987471757,
      "loss": 3.4323,
      "step": 157300
    },
    {
      "epoch": 0.14663837400219362,
      "grad_norm": 1.0270922183990479,
      "learning_rate": 0.00017067253555262237,
      "loss": 3.5125,
      "step": 157310
    },
    {
      "epoch": 0.1466476956202727,
      "grad_norm": 0.9665002226829529,
      "learning_rate": 0.00017067067123052715,
      "loss": 3.4965,
      "step": 157320
    },
    {
      "epoch": 0.14665701723835178,
      "grad_norm": 0.9920130968093872,
      "learning_rate": 0.00017066880690843196,
      "loss": 3.7043,
      "step": 157330
    },
    {
      "epoch": 0.14666633885643088,
      "grad_norm": 1.085816740989685,
      "learning_rate": 0.00017066694258633676,
      "loss": 3.5537,
      "step": 157340
    },
    {
      "epoch": 0.14667566047450997,
      "grad_norm": 0.9818804860115051,
      "learning_rate": 0.00017066507826424157,
      "loss": 3.6542,
      "step": 157350
    },
    {
      "epoch": 0.14668498209258907,
      "grad_norm": 1.2168937921524048,
      "learning_rate": 0.00017066321394214638,
      "loss": 3.5908,
      "step": 157360
    },
    {
      "epoch": 0.14669430371066816,
      "grad_norm": 0.8600367903709412,
      "learning_rate": 0.00017066134962005116,
      "loss": 3.6386,
      "step": 157370
    },
    {
      "epoch": 0.14670362532874726,
      "grad_norm": 1.19365656375885,
      "learning_rate": 0.00017065948529795596,
      "loss": 3.4939,
      "step": 157380
    },
    {
      "epoch": 0.14671294694682635,
      "grad_norm": 1.04500412940979,
      "learning_rate": 0.00017065762097586074,
      "loss": 3.4267,
      "step": 157390
    },
    {
      "epoch": 0.14672226856490544,
      "grad_norm": 1.158331274986267,
      "learning_rate": 0.00017065575665376557,
      "loss": 3.4217,
      "step": 157400
    },
    {
      "epoch": 0.14673159018298454,
      "grad_norm": 0.8511186838150024,
      "learning_rate": 0.00017065389233167038,
      "loss": 3.7375,
      "step": 157410
    },
    {
      "epoch": 0.14674091180106363,
      "grad_norm": 1.0779386758804321,
      "learning_rate": 0.00017065202800957516,
      "loss": 3.5425,
      "step": 157420
    },
    {
      "epoch": 0.1467502334191427,
      "grad_norm": 1.1335598230361938,
      "learning_rate": 0.00017065016368747997,
      "loss": 3.4778,
      "step": 157430
    },
    {
      "epoch": 0.1467595550372218,
      "grad_norm": 1.131121039390564,
      "learning_rate": 0.00017064829936538475,
      "loss": 3.5064,
      "step": 157440
    },
    {
      "epoch": 0.1467688766553009,
      "grad_norm": 1.1160868406295776,
      "learning_rate": 0.00017064643504328955,
      "loss": 3.7335,
      "step": 157450
    },
    {
      "epoch": 0.14677819827337998,
      "grad_norm": 1.134381890296936,
      "learning_rate": 0.00017064457072119439,
      "loss": 3.487,
      "step": 157460
    },
    {
      "epoch": 0.14678751989145908,
      "grad_norm": 0.9474299550056458,
      "learning_rate": 0.00017064270639909916,
      "loss": 3.5762,
      "step": 157470
    },
    {
      "epoch": 0.14679684150953817,
      "grad_norm": 1.1403485536575317,
      "learning_rate": 0.00017064084207700397,
      "loss": 3.5312,
      "step": 157480
    },
    {
      "epoch": 0.14680616312761727,
      "grad_norm": 1.1513521671295166,
      "learning_rate": 0.00017063897775490875,
      "loss": 3.623,
      "step": 157490
    },
    {
      "epoch": 0.14681548474569636,
      "grad_norm": 1.0095131397247314,
      "learning_rate": 0.00017063711343281356,
      "loss": 3.5297,
      "step": 157500
    },
    {
      "epoch": 0.14682480636377546,
      "grad_norm": 1.086854338645935,
      "learning_rate": 0.00017063524911071836,
      "loss": 3.6073,
      "step": 157510
    },
    {
      "epoch": 0.14683412798185455,
      "grad_norm": 1.12337064743042,
      "learning_rate": 0.00017063338478862317,
      "loss": 3.5037,
      "step": 157520
    },
    {
      "epoch": 0.14684344959993362,
      "grad_norm": 0.9027214646339417,
      "learning_rate": 0.00017063152046652798,
      "loss": 3.3999,
      "step": 157530
    },
    {
      "epoch": 0.1468527712180127,
      "grad_norm": 1.0961817502975464,
      "learning_rate": 0.00017062965614443275,
      "loss": 3.575,
      "step": 157540
    },
    {
      "epoch": 0.1468620928360918,
      "grad_norm": 0.9911494851112366,
      "learning_rate": 0.00017062779182233756,
      "loss": 3.6122,
      "step": 157550
    },
    {
      "epoch": 0.1468714144541709,
      "grad_norm": 0.955200731754303,
      "learning_rate": 0.00017062592750024237,
      "loss": 3.7339,
      "step": 157560
    },
    {
      "epoch": 0.14688073607225,
      "grad_norm": 1.0026191473007202,
      "learning_rate": 0.00017062406317814717,
      "loss": 3.2337,
      "step": 157570
    },
    {
      "epoch": 0.1468900576903291,
      "grad_norm": 0.8582029938697815,
      "learning_rate": 0.00017062219885605198,
      "loss": 3.5671,
      "step": 157580
    },
    {
      "epoch": 0.14689937930840818,
      "grad_norm": 1.0495662689208984,
      "learning_rate": 0.00017062033453395676,
      "loss": 3.4825,
      "step": 157590
    },
    {
      "epoch": 0.14690870092648728,
      "grad_norm": 1.1785584688186646,
      "learning_rate": 0.00017061847021186157,
      "loss": 3.6402,
      "step": 157600
    },
    {
      "epoch": 0.14691802254456637,
      "grad_norm": 0.9917983412742615,
      "learning_rate": 0.00017061660588976637,
      "loss": 3.5244,
      "step": 157610
    },
    {
      "epoch": 0.14692734416264547,
      "grad_norm": 1.0535674095153809,
      "learning_rate": 0.00017061474156767118,
      "loss": 3.651,
      "step": 157620
    },
    {
      "epoch": 0.14693666578072453,
      "grad_norm": 0.9635142087936401,
      "learning_rate": 0.00017061287724557598,
      "loss": 3.5386,
      "step": 157630
    },
    {
      "epoch": 0.14694598739880363,
      "grad_norm": 0.8936102390289307,
      "learning_rate": 0.00017061101292348076,
      "loss": 3.4964,
      "step": 157640
    },
    {
      "epoch": 0.14695530901688272,
      "grad_norm": 1.1394762992858887,
      "learning_rate": 0.00017060914860138557,
      "loss": 3.5834,
      "step": 157650
    },
    {
      "epoch": 0.14696463063496182,
      "grad_norm": 0.935986340045929,
      "learning_rate": 0.00017060728427929038,
      "loss": 3.8366,
      "step": 157660
    },
    {
      "epoch": 0.1469739522530409,
      "grad_norm": 0.9937493801116943,
      "learning_rate": 0.00017060541995719516,
      "loss": 3.5185,
      "step": 157670
    },
    {
      "epoch": 0.14698327387112,
      "grad_norm": 1.099757432937622,
      "learning_rate": 0.0001706035556351,
      "loss": 3.4321,
      "step": 157680
    },
    {
      "epoch": 0.1469925954891991,
      "grad_norm": 0.8302254676818848,
      "learning_rate": 0.00017060169131300477,
      "loss": 3.5347,
      "step": 157690
    },
    {
      "epoch": 0.1470019171072782,
      "grad_norm": 1.0130687952041626,
      "learning_rate": 0.00017059982699090957,
      "loss": 3.6544,
      "step": 157700
    },
    {
      "epoch": 0.1470112387253573,
      "grad_norm": 0.9993167519569397,
      "learning_rate": 0.00017059796266881438,
      "loss": 3.4857,
      "step": 157710
    },
    {
      "epoch": 0.14702056034343639,
      "grad_norm": 0.9516943693161011,
      "learning_rate": 0.00017059609834671916,
      "loss": 3.5917,
      "step": 157720
    },
    {
      "epoch": 0.14702988196151548,
      "grad_norm": 0.9957903623580933,
      "learning_rate": 0.00017059423402462397,
      "loss": 3.8715,
      "step": 157730
    },
    {
      "epoch": 0.14703920357959455,
      "grad_norm": 1.069295883178711,
      "learning_rate": 0.00017059236970252877,
      "loss": 3.5291,
      "step": 157740
    },
    {
      "epoch": 0.14704852519767364,
      "grad_norm": 1.0761953592300415,
      "learning_rate": 0.00017059050538043358,
      "loss": 3.6334,
      "step": 157750
    },
    {
      "epoch": 0.14705784681575274,
      "grad_norm": 1.0384353399276733,
      "learning_rate": 0.00017058864105833839,
      "loss": 3.4514,
      "step": 157760
    },
    {
      "epoch": 0.14706716843383183,
      "grad_norm": 1.0322309732437134,
      "learning_rate": 0.00017058677673624316,
      "loss": 3.4681,
      "step": 157770
    },
    {
      "epoch": 0.14707649005191092,
      "grad_norm": 1.1743407249450684,
      "learning_rate": 0.00017058491241414797,
      "loss": 3.607,
      "step": 157780
    },
    {
      "epoch": 0.14708581166999002,
      "grad_norm": 1.0225551128387451,
      "learning_rate": 0.00017058304809205278,
      "loss": 3.6128,
      "step": 157790
    },
    {
      "epoch": 0.1470951332880691,
      "grad_norm": 0.9833743572235107,
      "learning_rate": 0.00017058118376995758,
      "loss": 3.664,
      "step": 157800
    },
    {
      "epoch": 0.1471044549061482,
      "grad_norm": 0.7463817000389099,
      "learning_rate": 0.0001705793194478624,
      "loss": 3.4236,
      "step": 157810
    },
    {
      "epoch": 0.1471137765242273,
      "grad_norm": 1.0333237648010254,
      "learning_rate": 0.00017057745512576717,
      "loss": 3.5284,
      "step": 157820
    },
    {
      "epoch": 0.1471230981423064,
      "grad_norm": 1.0132560729980469,
      "learning_rate": 0.00017057559080367198,
      "loss": 3.4527,
      "step": 157830
    },
    {
      "epoch": 0.14713241976038546,
      "grad_norm": 0.8140373826026917,
      "learning_rate": 0.00017057372648157675,
      "loss": 3.6562,
      "step": 157840
    },
    {
      "epoch": 0.14714174137846456,
      "grad_norm": 0.962538480758667,
      "learning_rate": 0.0001705718621594816,
      "loss": 3.581,
      "step": 157850
    },
    {
      "epoch": 0.14715106299654365,
      "grad_norm": 1.0324748754501343,
      "learning_rate": 0.0001705699978373864,
      "loss": 3.6152,
      "step": 157860
    },
    {
      "epoch": 0.14716038461462275,
      "grad_norm": 1.0286049842834473,
      "learning_rate": 0.00017056813351529117,
      "loss": 3.6719,
      "step": 157870
    },
    {
      "epoch": 0.14716970623270184,
      "grad_norm": 1.225281000137329,
      "learning_rate": 0.00017056626919319598,
      "loss": 3.5491,
      "step": 157880
    },
    {
      "epoch": 0.14717902785078094,
      "grad_norm": 1.001075029373169,
      "learning_rate": 0.00017056440487110076,
      "loss": 3.544,
      "step": 157890
    },
    {
      "epoch": 0.14718834946886003,
      "grad_norm": 1.5279185771942139,
      "learning_rate": 0.00017056254054900557,
      "loss": 3.7178,
      "step": 157900
    },
    {
      "epoch": 0.14719767108693912,
      "grad_norm": 0.9921907186508179,
      "learning_rate": 0.0001705606762269104,
      "loss": 3.7205,
      "step": 157910
    },
    {
      "epoch": 0.14720699270501822,
      "grad_norm": 1.1368789672851562,
      "learning_rate": 0.00017055881190481518,
      "loss": 3.6005,
      "step": 157920
    },
    {
      "epoch": 0.1472163143230973,
      "grad_norm": 0.9824383854866028,
      "learning_rate": 0.00017055694758271998,
      "loss": 3.609,
      "step": 157930
    },
    {
      "epoch": 0.1472256359411764,
      "grad_norm": 0.959607720375061,
      "learning_rate": 0.00017055508326062476,
      "loss": 3.5848,
      "step": 157940
    },
    {
      "epoch": 0.14723495755925547,
      "grad_norm": 0.9417222142219543,
      "learning_rate": 0.00017055321893852957,
      "loss": 3.6378,
      "step": 157950
    },
    {
      "epoch": 0.14724427917733457,
      "grad_norm": 1.0082165002822876,
      "learning_rate": 0.00017055135461643438,
      "loss": 3.7811,
      "step": 157960
    },
    {
      "epoch": 0.14725360079541366,
      "grad_norm": 1.1826285123825073,
      "learning_rate": 0.00017054949029433918,
      "loss": 3.5811,
      "step": 157970
    },
    {
      "epoch": 0.14726292241349276,
      "grad_norm": 1.069563388824463,
      "learning_rate": 0.000170547625972244,
      "loss": 3.3956,
      "step": 157980
    },
    {
      "epoch": 0.14727224403157185,
      "grad_norm": 0.9687991738319397,
      "learning_rate": 0.00017054576165014877,
      "loss": 3.4447,
      "step": 157990
    },
    {
      "epoch": 0.14728156564965095,
      "grad_norm": 1.2218987941741943,
      "learning_rate": 0.00017054389732805357,
      "loss": 3.6433,
      "step": 158000
    },
    {
      "epoch": 0.14729088726773004,
      "grad_norm": 0.9221024513244629,
      "learning_rate": 0.00017054203300595838,
      "loss": 3.4977,
      "step": 158010
    },
    {
      "epoch": 0.14730020888580914,
      "grad_norm": 1.001414179801941,
      "learning_rate": 0.0001705401686838632,
      "loss": 3.4871,
      "step": 158020
    },
    {
      "epoch": 0.14730953050388823,
      "grad_norm": 1.0821971893310547,
      "learning_rate": 0.000170538304361768,
      "loss": 3.4417,
      "step": 158030
    },
    {
      "epoch": 0.14731885212196733,
      "grad_norm": 1.0314228534698486,
      "learning_rate": 0.00017053644003967277,
      "loss": 3.6991,
      "step": 158040
    },
    {
      "epoch": 0.1473281737400464,
      "grad_norm": 0.9924076795578003,
      "learning_rate": 0.00017053457571757758,
      "loss": 3.4999,
      "step": 158050
    },
    {
      "epoch": 0.1473374953581255,
      "grad_norm": 1.29948091506958,
      "learning_rate": 0.00017053271139548238,
      "loss": 3.5451,
      "step": 158060
    },
    {
      "epoch": 0.14734681697620458,
      "grad_norm": 1.0443099737167358,
      "learning_rate": 0.0001705308470733872,
      "loss": 3.5031,
      "step": 158070
    },
    {
      "epoch": 0.14735613859428368,
      "grad_norm": 1.030447244644165,
      "learning_rate": 0.000170528982751292,
      "loss": 3.5805,
      "step": 158080
    },
    {
      "epoch": 0.14736546021236277,
      "grad_norm": 1.1286026239395142,
      "learning_rate": 0.00017052711842919678,
      "loss": 3.6162,
      "step": 158090
    },
    {
      "epoch": 0.14737478183044186,
      "grad_norm": 1.086392879486084,
      "learning_rate": 0.00017052525410710158,
      "loss": 3.5083,
      "step": 158100
    },
    {
      "epoch": 0.14738410344852096,
      "grad_norm": 1.1307718753814697,
      "learning_rate": 0.0001705233897850064,
      "loss": 3.5396,
      "step": 158110
    },
    {
      "epoch": 0.14739342506660005,
      "grad_norm": 1.304355502128601,
      "learning_rate": 0.00017052152546291117,
      "loss": 3.399,
      "step": 158120
    },
    {
      "epoch": 0.14740274668467915,
      "grad_norm": 0.8935936689376831,
      "learning_rate": 0.000170519661140816,
      "loss": 3.5454,
      "step": 158130
    },
    {
      "epoch": 0.14741206830275824,
      "grad_norm": 1.3941854238510132,
      "learning_rate": 0.00017051779681872078,
      "loss": 3.5723,
      "step": 158140
    },
    {
      "epoch": 0.14742138992083734,
      "grad_norm": 0.9650142192840576,
      "learning_rate": 0.0001705159324966256,
      "loss": 3.6552,
      "step": 158150
    },
    {
      "epoch": 0.1474307115389164,
      "grad_norm": 1.2272268533706665,
      "learning_rate": 0.0001705140681745304,
      "loss": 3.7446,
      "step": 158160
    },
    {
      "epoch": 0.1474400331569955,
      "grad_norm": 0.9880303144454956,
      "learning_rate": 0.00017051220385243517,
      "loss": 3.5856,
      "step": 158170
    },
    {
      "epoch": 0.1474493547750746,
      "grad_norm": 0.9620900750160217,
      "learning_rate": 0.00017051033953033998,
      "loss": 3.5881,
      "step": 158180
    },
    {
      "epoch": 0.1474586763931537,
      "grad_norm": 0.9056061506271362,
      "learning_rate": 0.00017050847520824479,
      "loss": 3.616,
      "step": 158190
    },
    {
      "epoch": 0.14746799801123278,
      "grad_norm": 1.1432603597640991,
      "learning_rate": 0.0001705066108861496,
      "loss": 3.3887,
      "step": 158200
    },
    {
      "epoch": 0.14747731962931188,
      "grad_norm": 1.4838320016860962,
      "learning_rate": 0.0001705047465640544,
      "loss": 3.4434,
      "step": 158210
    },
    {
      "epoch": 0.14748664124739097,
      "grad_norm": 1.0387898683547974,
      "learning_rate": 0.00017050288224195918,
      "loss": 3.7373,
      "step": 158220
    },
    {
      "epoch": 0.14749596286547006,
      "grad_norm": 1.0189706087112427,
      "learning_rate": 0.00017050101791986398,
      "loss": 3.4458,
      "step": 158230
    },
    {
      "epoch": 0.14750528448354916,
      "grad_norm": 1.1960124969482422,
      "learning_rate": 0.0001704991535977688,
      "loss": 3.5373,
      "step": 158240
    },
    {
      "epoch": 0.14751460610162825,
      "grad_norm": 1.1791350841522217,
      "learning_rate": 0.0001704972892756736,
      "loss": 3.6659,
      "step": 158250
    },
    {
      "epoch": 0.14752392771970732,
      "grad_norm": 1.1854087114334106,
      "learning_rate": 0.0001704954249535784,
      "loss": 3.4692,
      "step": 158260
    },
    {
      "epoch": 0.14753324933778642,
      "grad_norm": 1.1262915134429932,
      "learning_rate": 0.00017049356063148318,
      "loss": 3.6056,
      "step": 158270
    },
    {
      "epoch": 0.1475425709558655,
      "grad_norm": 1.284193754196167,
      "learning_rate": 0.000170491696309388,
      "loss": 3.3655,
      "step": 158280
    },
    {
      "epoch": 0.1475518925739446,
      "grad_norm": 1.0774354934692383,
      "learning_rate": 0.00017048983198729277,
      "loss": 3.5552,
      "step": 158290
    },
    {
      "epoch": 0.1475612141920237,
      "grad_norm": 1.1907057762145996,
      "learning_rate": 0.0001704879676651976,
      "loss": 3.1768,
      "step": 158300
    },
    {
      "epoch": 0.1475705358101028,
      "grad_norm": 1.073240876197815,
      "learning_rate": 0.0001704861033431024,
      "loss": 3.5296,
      "step": 158310
    },
    {
      "epoch": 0.1475798574281819,
      "grad_norm": 1.1431324481964111,
      "learning_rate": 0.00017048423902100719,
      "loss": 3.4386,
      "step": 158320
    },
    {
      "epoch": 0.14758917904626098,
      "grad_norm": 1.0327686071395874,
      "learning_rate": 0.000170482374698912,
      "loss": 3.6268,
      "step": 158330
    },
    {
      "epoch": 0.14759850066434008,
      "grad_norm": 1.1683541536331177,
      "learning_rate": 0.00017048051037681677,
      "loss": 3.6021,
      "step": 158340
    },
    {
      "epoch": 0.14760782228241917,
      "grad_norm": 1.0405675172805786,
      "learning_rate": 0.00017047864605472158,
      "loss": 3.6942,
      "step": 158350
    },
    {
      "epoch": 0.14761714390049827,
      "grad_norm": 1.1080900430679321,
      "learning_rate": 0.0001704767817326264,
      "loss": 3.5185,
      "step": 158360
    },
    {
      "epoch": 0.14762646551857733,
      "grad_norm": 0.9692206382751465,
      "learning_rate": 0.0001704749174105312,
      "loss": 3.7253,
      "step": 158370
    },
    {
      "epoch": 0.14763578713665643,
      "grad_norm": 1.2009391784667969,
      "learning_rate": 0.000170473053088436,
      "loss": 3.7144,
      "step": 158380
    },
    {
      "epoch": 0.14764510875473552,
      "grad_norm": 1.174905776977539,
      "learning_rate": 0.00017047118876634078,
      "loss": 3.5199,
      "step": 158390
    },
    {
      "epoch": 0.14765443037281462,
      "grad_norm": 0.9232954978942871,
      "learning_rate": 0.00017046932444424558,
      "loss": 3.3564,
      "step": 158400
    },
    {
      "epoch": 0.1476637519908937,
      "grad_norm": 0.9874709844589233,
      "learning_rate": 0.0001704674601221504,
      "loss": 3.6338,
      "step": 158410
    },
    {
      "epoch": 0.1476730736089728,
      "grad_norm": 0.8505376577377319,
      "learning_rate": 0.0001704655958000552,
      "loss": 3.4556,
      "step": 158420
    },
    {
      "epoch": 0.1476823952270519,
      "grad_norm": 1.0583362579345703,
      "learning_rate": 0.00017046373147796,
      "loss": 3.5738,
      "step": 158430
    },
    {
      "epoch": 0.147691716845131,
      "grad_norm": 1.0456198453903198,
      "learning_rate": 0.00017046186715586478,
      "loss": 3.6598,
      "step": 158440
    },
    {
      "epoch": 0.1477010384632101,
      "grad_norm": 0.8781313896179199,
      "learning_rate": 0.0001704600028337696,
      "loss": 3.5806,
      "step": 158450
    },
    {
      "epoch": 0.14771036008128918,
      "grad_norm": 1.141623616218567,
      "learning_rate": 0.0001704581385116744,
      "loss": 3.5694,
      "step": 158460
    },
    {
      "epoch": 0.14771968169936825,
      "grad_norm": 1.01954185962677,
      "learning_rate": 0.0001704562741895792,
      "loss": 3.5777,
      "step": 158470
    },
    {
      "epoch": 0.14772900331744734,
      "grad_norm": 1.0547353029251099,
      "learning_rate": 0.000170454409867484,
      "loss": 3.4777,
      "step": 158480
    },
    {
      "epoch": 0.14773832493552644,
      "grad_norm": 1.1539313793182373,
      "learning_rate": 0.00017045254554538878,
      "loss": 3.4817,
      "step": 158490
    },
    {
      "epoch": 0.14774764655360553,
      "grad_norm": 1.3037060499191284,
      "learning_rate": 0.0001704506812232936,
      "loss": 3.5232,
      "step": 158500
    },
    {
      "epoch": 0.14775696817168463,
      "grad_norm": 1.075512170791626,
      "learning_rate": 0.0001704488169011984,
      "loss": 3.475,
      "step": 158510
    },
    {
      "epoch": 0.14776628978976372,
      "grad_norm": 1.1165884733200073,
      "learning_rate": 0.0001704469525791032,
      "loss": 3.3652,
      "step": 158520
    },
    {
      "epoch": 0.14777561140784282,
      "grad_norm": 1.2549787759780884,
      "learning_rate": 0.000170445088257008,
      "loss": 3.5996,
      "step": 158530
    },
    {
      "epoch": 0.1477849330259219,
      "grad_norm": 0.99810391664505,
      "learning_rate": 0.0001704432239349128,
      "loss": 3.7286,
      "step": 158540
    },
    {
      "epoch": 0.147794254644001,
      "grad_norm": 0.8876933455467224,
      "learning_rate": 0.0001704413596128176,
      "loss": 3.5332,
      "step": 158550
    },
    {
      "epoch": 0.1478035762620801,
      "grad_norm": 1.1759753227233887,
      "learning_rate": 0.0001704394952907224,
      "loss": 3.6906,
      "step": 158560
    },
    {
      "epoch": 0.1478128978801592,
      "grad_norm": 1.0224330425262451,
      "learning_rate": 0.00017043763096862718,
      "loss": 3.5158,
      "step": 158570
    },
    {
      "epoch": 0.14782221949823826,
      "grad_norm": 1.0794836282730103,
      "learning_rate": 0.00017043576664653201,
      "loss": 3.4655,
      "step": 158580
    },
    {
      "epoch": 0.14783154111631736,
      "grad_norm": 1.17789888381958,
      "learning_rate": 0.0001704339023244368,
      "loss": 3.6291,
      "step": 158590
    },
    {
      "epoch": 0.14784086273439645,
      "grad_norm": 1.2822102308273315,
      "learning_rate": 0.0001704320380023416,
      "loss": 3.6507,
      "step": 158600
    },
    {
      "epoch": 0.14785018435247554,
      "grad_norm": 0.9787222743034363,
      "learning_rate": 0.0001704301736802464,
      "loss": 3.7588,
      "step": 158610
    },
    {
      "epoch": 0.14785950597055464,
      "grad_norm": 1.0904210805892944,
      "learning_rate": 0.00017042830935815119,
      "loss": 3.5706,
      "step": 158620
    },
    {
      "epoch": 0.14786882758863373,
      "grad_norm": 0.9995608329772949,
      "learning_rate": 0.000170426445036056,
      "loss": 3.5291,
      "step": 158630
    },
    {
      "epoch": 0.14787814920671283,
      "grad_norm": 1.026564359664917,
      "learning_rate": 0.0001704245807139608,
      "loss": 3.5662,
      "step": 158640
    },
    {
      "epoch": 0.14788747082479192,
      "grad_norm": 0.9851803183555603,
      "learning_rate": 0.0001704227163918656,
      "loss": 3.4281,
      "step": 158650
    },
    {
      "epoch": 0.14789679244287102,
      "grad_norm": 0.9240439534187317,
      "learning_rate": 0.0001704208520697704,
      "loss": 3.5344,
      "step": 158660
    },
    {
      "epoch": 0.1479061140609501,
      "grad_norm": 1.1077150106430054,
      "learning_rate": 0.0001704189877476752,
      "loss": 3.4219,
      "step": 158670
    },
    {
      "epoch": 0.14791543567902918,
      "grad_norm": 1.0331851243972778,
      "learning_rate": 0.00017041712342558,
      "loss": 3.5892,
      "step": 158680
    },
    {
      "epoch": 0.14792475729710827,
      "grad_norm": 1.0051921606063843,
      "learning_rate": 0.0001704152591034848,
      "loss": 3.6381,
      "step": 158690
    },
    {
      "epoch": 0.14793407891518737,
      "grad_norm": 1.0655348300933838,
      "learning_rate": 0.0001704133947813896,
      "loss": 3.4496,
      "step": 158700
    },
    {
      "epoch": 0.14794340053326646,
      "grad_norm": 1.1740765571594238,
      "learning_rate": 0.00017041153045929442,
      "loss": 3.5158,
      "step": 158710
    },
    {
      "epoch": 0.14795272215134556,
      "grad_norm": 0.9041295647621155,
      "learning_rate": 0.0001704096661371992,
      "loss": 3.4957,
      "step": 158720
    },
    {
      "epoch": 0.14796204376942465,
      "grad_norm": 1.1951860189437866,
      "learning_rate": 0.000170407801815104,
      "loss": 3.6482,
      "step": 158730
    },
    {
      "epoch": 0.14797136538750374,
      "grad_norm": 0.9910904765129089,
      "learning_rate": 0.0001704059374930088,
      "loss": 3.6335,
      "step": 158740
    },
    {
      "epoch": 0.14798068700558284,
      "grad_norm": 0.8875108957290649,
      "learning_rate": 0.0001704040731709136,
      "loss": 3.6997,
      "step": 158750
    },
    {
      "epoch": 0.14799000862366193,
      "grad_norm": 0.9150171875953674,
      "learning_rate": 0.0001704022088488184,
      "loss": 3.6998,
      "step": 158760
    },
    {
      "epoch": 0.14799933024174103,
      "grad_norm": 0.9678056240081787,
      "learning_rate": 0.0001704003445267232,
      "loss": 3.5495,
      "step": 158770
    },
    {
      "epoch": 0.14800865185982012,
      "grad_norm": 1.086863398551941,
      "learning_rate": 0.000170398480204628,
      "loss": 3.5708,
      "step": 158780
    },
    {
      "epoch": 0.1480179734778992,
      "grad_norm": 0.9939865469932556,
      "learning_rate": 0.00017039661588253278,
      "loss": 3.6589,
      "step": 158790
    },
    {
      "epoch": 0.14802729509597828,
      "grad_norm": 1.0749441385269165,
      "learning_rate": 0.00017039475156043762,
      "loss": 3.7256,
      "step": 158800
    },
    {
      "epoch": 0.14803661671405738,
      "grad_norm": 0.8282711505889893,
      "learning_rate": 0.0001703928872383424,
      "loss": 3.6758,
      "step": 158810
    },
    {
      "epoch": 0.14804593833213647,
      "grad_norm": 0.9826323986053467,
      "learning_rate": 0.0001703910229162472,
      "loss": 3.513,
      "step": 158820
    },
    {
      "epoch": 0.14805525995021557,
      "grad_norm": 1.4010827541351318,
      "learning_rate": 0.000170389158594152,
      "loss": 3.607,
      "step": 158830
    },
    {
      "epoch": 0.14806458156829466,
      "grad_norm": 0.9159415364265442,
      "learning_rate": 0.0001703872942720568,
      "loss": 3.6538,
      "step": 158840
    },
    {
      "epoch": 0.14807390318637376,
      "grad_norm": 1.091078758239746,
      "learning_rate": 0.0001703854299499616,
      "loss": 3.3694,
      "step": 158850
    },
    {
      "epoch": 0.14808322480445285,
      "grad_norm": 1.2134612798690796,
      "learning_rate": 0.0001703835656278664,
      "loss": 3.5352,
      "step": 158860
    },
    {
      "epoch": 0.14809254642253195,
      "grad_norm": 0.9743009805679321,
      "learning_rate": 0.0001703817013057712,
      "loss": 3.5613,
      "step": 158870
    },
    {
      "epoch": 0.14810186804061104,
      "grad_norm": 1.0697126388549805,
      "learning_rate": 0.00017037983698367601,
      "loss": 3.4879,
      "step": 158880
    },
    {
      "epoch": 0.1481111896586901,
      "grad_norm": 1.0895503759384155,
      "learning_rate": 0.0001703779726615808,
      "loss": 3.5865,
      "step": 158890
    },
    {
      "epoch": 0.1481205112767692,
      "grad_norm": 1.1502653360366821,
      "learning_rate": 0.0001703761083394856,
      "loss": 3.5931,
      "step": 158900
    },
    {
      "epoch": 0.1481298328948483,
      "grad_norm": 0.9448443055152893,
      "learning_rate": 0.0001703742440173904,
      "loss": 3.4938,
      "step": 158910
    },
    {
      "epoch": 0.1481391545129274,
      "grad_norm": 1.031388759613037,
      "learning_rate": 0.0001703723796952952,
      "loss": 3.3606,
      "step": 158920
    },
    {
      "epoch": 0.14814847613100648,
      "grad_norm": 1.0908894538879395,
      "learning_rate": 0.00017037051537320002,
      "loss": 3.5174,
      "step": 158930
    },
    {
      "epoch": 0.14815779774908558,
      "grad_norm": 1.2103888988494873,
      "learning_rate": 0.0001703686510511048,
      "loss": 3.6571,
      "step": 158940
    },
    {
      "epoch": 0.14816711936716467,
      "grad_norm": 1.127071499824524,
      "learning_rate": 0.0001703667867290096,
      "loss": 3.5048,
      "step": 158950
    },
    {
      "epoch": 0.14817644098524377,
      "grad_norm": 1.1029882431030273,
      "learning_rate": 0.00017036492240691438,
      "loss": 3.5108,
      "step": 158960
    },
    {
      "epoch": 0.14818576260332286,
      "grad_norm": 0.9605993032455444,
      "learning_rate": 0.00017036305808481922,
      "loss": 3.3602,
      "step": 158970
    },
    {
      "epoch": 0.14819508422140196,
      "grad_norm": 1.114558219909668,
      "learning_rate": 0.00017036119376272402,
      "loss": 3.4438,
      "step": 158980
    },
    {
      "epoch": 0.14820440583948102,
      "grad_norm": 1.1050631999969482,
      "learning_rate": 0.0001703593294406288,
      "loss": 3.3622,
      "step": 158990
    },
    {
      "epoch": 0.14821372745756012,
      "grad_norm": 1.0847584009170532,
      "learning_rate": 0.0001703574651185336,
      "loss": 3.6077,
      "step": 159000
    },
    {
      "epoch": 0.1482230490756392,
      "grad_norm": 1.1553462743759155,
      "learning_rate": 0.0001703556007964384,
      "loss": 3.497,
      "step": 159010
    },
    {
      "epoch": 0.1482323706937183,
      "grad_norm": 1.2773873805999756,
      "learning_rate": 0.0001703537364743432,
      "loss": 3.6911,
      "step": 159020
    },
    {
      "epoch": 0.1482416923117974,
      "grad_norm": 0.9396129250526428,
      "learning_rate": 0.00017035187215224803,
      "loss": 3.3938,
      "step": 159030
    },
    {
      "epoch": 0.1482510139298765,
      "grad_norm": 0.8235486149787903,
      "learning_rate": 0.0001703500078301528,
      "loss": 3.7933,
      "step": 159040
    },
    {
      "epoch": 0.1482603355479556,
      "grad_norm": 1.1064841747283936,
      "learning_rate": 0.0001703481435080576,
      "loss": 3.508,
      "step": 159050
    },
    {
      "epoch": 0.14826965716603469,
      "grad_norm": 1.129001259803772,
      "learning_rate": 0.0001703462791859624,
      "loss": 3.527,
      "step": 159060
    },
    {
      "epoch": 0.14827897878411378,
      "grad_norm": 1.0211693048477173,
      "learning_rate": 0.0001703444148638672,
      "loss": 3.5782,
      "step": 159070
    },
    {
      "epoch": 0.14828830040219287,
      "grad_norm": 0.9875473380088806,
      "learning_rate": 0.000170342550541772,
      "loss": 3.6467,
      "step": 159080
    },
    {
      "epoch": 0.14829762202027197,
      "grad_norm": 1.1554298400878906,
      "learning_rate": 0.0001703406862196768,
      "loss": 3.4748,
      "step": 159090
    },
    {
      "epoch": 0.14830694363835104,
      "grad_norm": 0.9746116399765015,
      "learning_rate": 0.00017033882189758162,
      "loss": 3.334,
      "step": 159100
    },
    {
      "epoch": 0.14831626525643013,
      "grad_norm": 1.0936334133148193,
      "learning_rate": 0.0001703369575754864,
      "loss": 3.7506,
      "step": 159110
    },
    {
      "epoch": 0.14832558687450922,
      "grad_norm": 0.9710587859153748,
      "learning_rate": 0.0001703350932533912,
      "loss": 3.5301,
      "step": 159120
    },
    {
      "epoch": 0.14833490849258832,
      "grad_norm": 1.1152311563491821,
      "learning_rate": 0.000170333228931296,
      "loss": 3.5347,
      "step": 159130
    },
    {
      "epoch": 0.1483442301106674,
      "grad_norm": 1.0672270059585571,
      "learning_rate": 0.00017033136460920082,
      "loss": 3.6278,
      "step": 159140
    },
    {
      "epoch": 0.1483535517287465,
      "grad_norm": 1.104860544204712,
      "learning_rate": 0.00017032950028710562,
      "loss": 3.6368,
      "step": 159150
    },
    {
      "epoch": 0.1483628733468256,
      "grad_norm": 1.0813692808151245,
      "learning_rate": 0.0001703276359650104,
      "loss": 3.6515,
      "step": 159160
    },
    {
      "epoch": 0.1483721949649047,
      "grad_norm": 1.0007214546203613,
      "learning_rate": 0.0001703257716429152,
      "loss": 3.6699,
      "step": 159170
    },
    {
      "epoch": 0.1483815165829838,
      "grad_norm": 1.0346230268478394,
      "learning_rate": 0.00017032390732082001,
      "loss": 3.6256,
      "step": 159180
    },
    {
      "epoch": 0.14839083820106289,
      "grad_norm": 1.1913636922836304,
      "learning_rate": 0.00017032204299872482,
      "loss": 3.5195,
      "step": 159190
    },
    {
      "epoch": 0.14840015981914195,
      "grad_norm": 1.0677821636199951,
      "learning_rate": 0.00017032017867662963,
      "loss": 3.4975,
      "step": 159200
    },
    {
      "epoch": 0.14840948143722105,
      "grad_norm": 0.9918962121009827,
      "learning_rate": 0.0001703183143545344,
      "loss": 3.5307,
      "step": 159210
    },
    {
      "epoch": 0.14841880305530014,
      "grad_norm": 1.040022850036621,
      "learning_rate": 0.0001703164500324392,
      "loss": 3.3475,
      "step": 159220
    },
    {
      "epoch": 0.14842812467337924,
      "grad_norm": 1.178839921951294,
      "learning_rate": 0.00017031458571034402,
      "loss": 3.626,
      "step": 159230
    },
    {
      "epoch": 0.14843744629145833,
      "grad_norm": 1.149971604347229,
      "learning_rate": 0.0001703127213882488,
      "loss": 3.5379,
      "step": 159240
    },
    {
      "epoch": 0.14844676790953742,
      "grad_norm": 0.9720000624656677,
      "learning_rate": 0.00017031085706615363,
      "loss": 3.6401,
      "step": 159250
    },
    {
      "epoch": 0.14845608952761652,
      "grad_norm": 0.8232358694076538,
      "learning_rate": 0.0001703089927440584,
      "loss": 3.6475,
      "step": 159260
    },
    {
      "epoch": 0.1484654111456956,
      "grad_norm": 1.0448014736175537,
      "learning_rate": 0.00017030712842196322,
      "loss": 3.55,
      "step": 159270
    },
    {
      "epoch": 0.1484747327637747,
      "grad_norm": 1.0839005708694458,
      "learning_rate": 0.00017030526409986802,
      "loss": 3.4745,
      "step": 159280
    },
    {
      "epoch": 0.1484840543818538,
      "grad_norm": 0.8858891725540161,
      "learning_rate": 0.0001703033997777728,
      "loss": 3.6444,
      "step": 159290
    },
    {
      "epoch": 0.1484933759999329,
      "grad_norm": 1.1181800365447998,
      "learning_rate": 0.0001703015354556776,
      "loss": 3.5701,
      "step": 159300
    },
    {
      "epoch": 0.14850269761801196,
      "grad_norm": 0.9553754329681396,
      "learning_rate": 0.00017029967113358241,
      "loss": 3.4785,
      "step": 159310
    },
    {
      "epoch": 0.14851201923609106,
      "grad_norm": 0.9252158403396606,
      "learning_rate": 0.00017029780681148722,
      "loss": 3.7106,
      "step": 159320
    },
    {
      "epoch": 0.14852134085417015,
      "grad_norm": 0.979966938495636,
      "learning_rate": 0.00017029594248939203,
      "loss": 3.4853,
      "step": 159330
    },
    {
      "epoch": 0.14853066247224925,
      "grad_norm": 1.2386510372161865,
      "learning_rate": 0.0001702940781672968,
      "loss": 3.6225,
      "step": 159340
    },
    {
      "epoch": 0.14853998409032834,
      "grad_norm": 1.017725944519043,
      "learning_rate": 0.0001702922138452016,
      "loss": 3.6288,
      "step": 159350
    },
    {
      "epoch": 0.14854930570840744,
      "grad_norm": 1.1890629529953003,
      "learning_rate": 0.00017029034952310642,
      "loss": 3.4626,
      "step": 159360
    },
    {
      "epoch": 0.14855862732648653,
      "grad_norm": 1.2306044101715088,
      "learning_rate": 0.00017028848520101123,
      "loss": 3.5137,
      "step": 159370
    },
    {
      "epoch": 0.14856794894456563,
      "grad_norm": 1.1208819150924683,
      "learning_rate": 0.00017028662087891603,
      "loss": 3.621,
      "step": 159380
    },
    {
      "epoch": 0.14857727056264472,
      "grad_norm": 1.1357563734054565,
      "learning_rate": 0.0001702847565568208,
      "loss": 3.5643,
      "step": 159390
    },
    {
      "epoch": 0.14858659218072381,
      "grad_norm": 1.074350357055664,
      "learning_rate": 0.00017028289223472562,
      "loss": 3.5278,
      "step": 159400
    },
    {
      "epoch": 0.14859591379880288,
      "grad_norm": 1.1611250638961792,
      "learning_rate": 0.0001702810279126304,
      "loss": 3.408,
      "step": 159410
    },
    {
      "epoch": 0.14860523541688198,
      "grad_norm": 1.1267133951187134,
      "learning_rate": 0.00017027916359053523,
      "loss": 3.4857,
      "step": 159420
    },
    {
      "epoch": 0.14861455703496107,
      "grad_norm": 1.1342583894729614,
      "learning_rate": 0.00017027729926844004,
      "loss": 3.5979,
      "step": 159430
    },
    {
      "epoch": 0.14862387865304016,
      "grad_norm": 1.1260281801223755,
      "learning_rate": 0.00017027543494634482,
      "loss": 3.5209,
      "step": 159440
    },
    {
      "epoch": 0.14863320027111926,
      "grad_norm": 1.3793370723724365,
      "learning_rate": 0.00017027357062424962,
      "loss": 3.6803,
      "step": 159450
    },
    {
      "epoch": 0.14864252188919835,
      "grad_norm": 1.1468645334243774,
      "learning_rate": 0.0001702717063021544,
      "loss": 3.516,
      "step": 159460
    },
    {
      "epoch": 0.14865184350727745,
      "grad_norm": 0.9758588671684265,
      "learning_rate": 0.0001702698419800592,
      "loss": 3.4379,
      "step": 159470
    },
    {
      "epoch": 0.14866116512535654,
      "grad_norm": 0.9619123339653015,
      "learning_rate": 0.00017026797765796404,
      "loss": 3.6487,
      "step": 159480
    },
    {
      "epoch": 0.14867048674343564,
      "grad_norm": 1.200209617614746,
      "learning_rate": 0.00017026611333586882,
      "loss": 3.5602,
      "step": 159490
    },
    {
      "epoch": 0.14867980836151473,
      "grad_norm": 1.1503713130950928,
      "learning_rate": 0.00017026424901377363,
      "loss": 3.5545,
      "step": 159500
    },
    {
      "epoch": 0.14868912997959383,
      "grad_norm": 1.7753474712371826,
      "learning_rate": 0.0001702623846916784,
      "loss": 3.5318,
      "step": 159510
    },
    {
      "epoch": 0.1486984515976729,
      "grad_norm": 1.1004234552383423,
      "learning_rate": 0.0001702605203695832,
      "loss": 3.5941,
      "step": 159520
    },
    {
      "epoch": 0.148707773215752,
      "grad_norm": 1.0216001272201538,
      "learning_rate": 0.00017025865604748802,
      "loss": 3.3833,
      "step": 159530
    },
    {
      "epoch": 0.14871709483383108,
      "grad_norm": 0.9225265383720398,
      "learning_rate": 0.00017025679172539282,
      "loss": 3.5286,
      "step": 159540
    },
    {
      "epoch": 0.14872641645191018,
      "grad_norm": 1.0562937259674072,
      "learning_rate": 0.00017025492740329763,
      "loss": 3.5954,
      "step": 159550
    },
    {
      "epoch": 0.14873573806998927,
      "grad_norm": 1.2083643674850464,
      "learning_rate": 0.0001702530630812024,
      "loss": 3.6072,
      "step": 159560
    },
    {
      "epoch": 0.14874505968806837,
      "grad_norm": 1.30120849609375,
      "learning_rate": 0.00017025119875910722,
      "loss": 3.5492,
      "step": 159570
    },
    {
      "epoch": 0.14875438130614746,
      "grad_norm": 1.0119293928146362,
      "learning_rate": 0.00017024933443701202,
      "loss": 3.7399,
      "step": 159580
    },
    {
      "epoch": 0.14876370292422655,
      "grad_norm": 1.1844240427017212,
      "learning_rate": 0.00017024747011491683,
      "loss": 3.535,
      "step": 159590
    },
    {
      "epoch": 0.14877302454230565,
      "grad_norm": 1.4573841094970703,
      "learning_rate": 0.00017024560579282163,
      "loss": 3.6763,
      "step": 159600
    },
    {
      "epoch": 0.14878234616038474,
      "grad_norm": 1.1241778135299683,
      "learning_rate": 0.00017024374147072641,
      "loss": 3.5273,
      "step": 159610
    },
    {
      "epoch": 0.1487916677784638,
      "grad_norm": 1.2764396667480469,
      "learning_rate": 0.00017024187714863122,
      "loss": 3.6597,
      "step": 159620
    },
    {
      "epoch": 0.1488009893965429,
      "grad_norm": 1.1325939893722534,
      "learning_rate": 0.00017024001282653603,
      "loss": 3.4645,
      "step": 159630
    },
    {
      "epoch": 0.148810311014622,
      "grad_norm": 1.1683320999145508,
      "learning_rate": 0.00017023814850444083,
      "loss": 3.424,
      "step": 159640
    },
    {
      "epoch": 0.1488196326327011,
      "grad_norm": 1.2031474113464355,
      "learning_rate": 0.00017023628418234564,
      "loss": 3.5722,
      "step": 159650
    },
    {
      "epoch": 0.1488289542507802,
      "grad_norm": 1.0339726209640503,
      "learning_rate": 0.00017023441986025042,
      "loss": 3.4475,
      "step": 159660
    },
    {
      "epoch": 0.14883827586885928,
      "grad_norm": 1.2529027462005615,
      "learning_rate": 0.00017023255553815522,
      "loss": 3.5862,
      "step": 159670
    },
    {
      "epoch": 0.14884759748693838,
      "grad_norm": 1.1898609399795532,
      "learning_rate": 0.00017023069121606003,
      "loss": 3.5262,
      "step": 159680
    },
    {
      "epoch": 0.14885691910501747,
      "grad_norm": 1.2746624946594238,
      "learning_rate": 0.0001702288268939648,
      "loss": 3.6632,
      "step": 159690
    },
    {
      "epoch": 0.14886624072309657,
      "grad_norm": 1.188232660293579,
      "learning_rate": 0.00017022696257186964,
      "loss": 3.5052,
      "step": 159700
    },
    {
      "epoch": 0.14887556234117566,
      "grad_norm": 0.97633957862854,
      "learning_rate": 0.00017022509824977442,
      "loss": 3.6626,
      "step": 159710
    },
    {
      "epoch": 0.14888488395925475,
      "grad_norm": 1.0002793073654175,
      "learning_rate": 0.00017022323392767923,
      "loss": 3.4721,
      "step": 159720
    },
    {
      "epoch": 0.14889420557733382,
      "grad_norm": 1.1517970561981201,
      "learning_rate": 0.00017022136960558404,
      "loss": 3.7943,
      "step": 159730
    },
    {
      "epoch": 0.14890352719541292,
      "grad_norm": 1.231310486793518,
      "learning_rate": 0.00017021950528348881,
      "loss": 3.5198,
      "step": 159740
    },
    {
      "epoch": 0.148912848813492,
      "grad_norm": 1.2063846588134766,
      "learning_rate": 0.00017021764096139362,
      "loss": 3.372,
      "step": 159750
    },
    {
      "epoch": 0.1489221704315711,
      "grad_norm": 1.0530564785003662,
      "learning_rate": 0.00017021577663929843,
      "loss": 3.7168,
      "step": 159760
    },
    {
      "epoch": 0.1489314920496502,
      "grad_norm": 1.1288726329803467,
      "learning_rate": 0.00017021391231720323,
      "loss": 3.717,
      "step": 159770
    },
    {
      "epoch": 0.1489408136677293,
      "grad_norm": 1.203505516052246,
      "learning_rate": 0.00017021204799510804,
      "loss": 3.5949,
      "step": 159780
    },
    {
      "epoch": 0.1489501352858084,
      "grad_norm": 1.011205792427063,
      "learning_rate": 0.00017021018367301282,
      "loss": 3.3741,
      "step": 159790
    },
    {
      "epoch": 0.14895945690388748,
      "grad_norm": 0.9160363078117371,
      "learning_rate": 0.00017020831935091763,
      "loss": 3.401,
      "step": 159800
    },
    {
      "epoch": 0.14896877852196658,
      "grad_norm": 0.929060697555542,
      "learning_rate": 0.00017020645502882243,
      "loss": 3.4648,
      "step": 159810
    },
    {
      "epoch": 0.14897810014004567,
      "grad_norm": 0.9589592218399048,
      "learning_rate": 0.00017020459070672724,
      "loss": 3.7799,
      "step": 159820
    },
    {
      "epoch": 0.14898742175812474,
      "grad_norm": 1.3412141799926758,
      "learning_rate": 0.00017020272638463204,
      "loss": 3.4255,
      "step": 159830
    },
    {
      "epoch": 0.14899674337620383,
      "grad_norm": 0.9489890933036804,
      "learning_rate": 0.00017020086206253682,
      "loss": 3.7121,
      "step": 159840
    },
    {
      "epoch": 0.14900606499428293,
      "grad_norm": 0.9331606030464172,
      "learning_rate": 0.00017019899774044163,
      "loss": 3.7102,
      "step": 159850
    },
    {
      "epoch": 0.14901538661236202,
      "grad_norm": 1.247061848640442,
      "learning_rate": 0.0001701971334183464,
      "loss": 3.4582,
      "step": 159860
    },
    {
      "epoch": 0.14902470823044112,
      "grad_norm": 1.1795265674591064,
      "learning_rate": 0.00017019526909625124,
      "loss": 3.6177,
      "step": 159870
    },
    {
      "epoch": 0.1490340298485202,
      "grad_norm": 0.9581311941146851,
      "learning_rate": 0.00017019340477415605,
      "loss": 3.6985,
      "step": 159880
    },
    {
      "epoch": 0.1490433514665993,
      "grad_norm": 1.0542627573013306,
      "learning_rate": 0.00017019154045206083,
      "loss": 3.4996,
      "step": 159890
    },
    {
      "epoch": 0.1490526730846784,
      "grad_norm": 0.9943599700927734,
      "learning_rate": 0.00017018967612996563,
      "loss": 3.4578,
      "step": 159900
    },
    {
      "epoch": 0.1490619947027575,
      "grad_norm": 1.0370550155639648,
      "learning_rate": 0.0001701878118078704,
      "loss": 3.4324,
      "step": 159910
    },
    {
      "epoch": 0.1490713163208366,
      "grad_norm": 1.0950193405151367,
      "learning_rate": 0.00017018594748577522,
      "loss": 3.4642,
      "step": 159920
    },
    {
      "epoch": 0.14908063793891568,
      "grad_norm": 0.8506491184234619,
      "learning_rate": 0.00017018408316368005,
      "loss": 3.6744,
      "step": 159930
    },
    {
      "epoch": 0.14908995955699475,
      "grad_norm": 1.0062934160232544,
      "learning_rate": 0.00017018221884158483,
      "loss": 3.7631,
      "step": 159940
    },
    {
      "epoch": 0.14909928117507384,
      "grad_norm": 1.0825012922286987,
      "learning_rate": 0.00017018035451948964,
      "loss": 3.5101,
      "step": 159950
    },
    {
      "epoch": 0.14910860279315294,
      "grad_norm": 1.0718317031860352,
      "learning_rate": 0.00017017849019739442,
      "loss": 3.3342,
      "step": 159960
    },
    {
      "epoch": 0.14911792441123203,
      "grad_norm": 1.1856319904327393,
      "learning_rate": 0.00017017662587529922,
      "loss": 3.363,
      "step": 159970
    },
    {
      "epoch": 0.14912724602931113,
      "grad_norm": 1.1792877912521362,
      "learning_rate": 0.00017017476155320403,
      "loss": 3.506,
      "step": 159980
    },
    {
      "epoch": 0.14913656764739022,
      "grad_norm": 0.9877211451530457,
      "learning_rate": 0.00017017289723110884,
      "loss": 3.6044,
      "step": 159990
    },
    {
      "epoch": 0.14914588926546932,
      "grad_norm": 1.2673358917236328,
      "learning_rate": 0.00017017103290901364,
      "loss": 3.7754,
      "step": 160000
    },
    {
      "epoch": 0.1491552108835484,
      "grad_norm": 1.003751277923584,
      "learning_rate": 0.00017016916858691842,
      "loss": 3.5482,
      "step": 160010
    },
    {
      "epoch": 0.1491645325016275,
      "grad_norm": 0.9569557309150696,
      "learning_rate": 0.00017016730426482323,
      "loss": 3.7476,
      "step": 160020
    },
    {
      "epoch": 0.1491738541197066,
      "grad_norm": 0.9118075370788574,
      "learning_rate": 0.00017016543994272803,
      "loss": 3.5788,
      "step": 160030
    },
    {
      "epoch": 0.14918317573778567,
      "grad_norm": 1.1106622219085693,
      "learning_rate": 0.00017016357562063284,
      "loss": 3.5163,
      "step": 160040
    },
    {
      "epoch": 0.14919249735586476,
      "grad_norm": 1.1500834226608276,
      "learning_rate": 0.00017016171129853765,
      "loss": 3.5664,
      "step": 160050
    },
    {
      "epoch": 0.14920181897394386,
      "grad_norm": 1.2677838802337646,
      "learning_rate": 0.00017015984697644243,
      "loss": 3.5604,
      "step": 160060
    },
    {
      "epoch": 0.14921114059202295,
      "grad_norm": 1.0866941213607788,
      "learning_rate": 0.00017015798265434723,
      "loss": 3.6185,
      "step": 160070
    },
    {
      "epoch": 0.14922046221010205,
      "grad_norm": 1.1275455951690674,
      "learning_rate": 0.000170156118332252,
      "loss": 3.6477,
      "step": 160080
    },
    {
      "epoch": 0.14922978382818114,
      "grad_norm": 0.942827045917511,
      "learning_rate": 0.00017015425401015685,
      "loss": 3.5271,
      "step": 160090
    },
    {
      "epoch": 0.14923910544626023,
      "grad_norm": 0.892052412033081,
      "learning_rate": 0.00017015238968806165,
      "loss": 3.453,
      "step": 160100
    },
    {
      "epoch": 0.14924842706433933,
      "grad_norm": 1.0090382099151611,
      "learning_rate": 0.00017015052536596643,
      "loss": 3.391,
      "step": 160110
    },
    {
      "epoch": 0.14925774868241842,
      "grad_norm": 1.0870615243911743,
      "learning_rate": 0.00017014866104387124,
      "loss": 3.6606,
      "step": 160120
    },
    {
      "epoch": 0.14926707030049752,
      "grad_norm": 1.0994369983673096,
      "learning_rate": 0.00017014679672177602,
      "loss": 3.5133,
      "step": 160130
    },
    {
      "epoch": 0.1492763919185766,
      "grad_norm": 1.225353479385376,
      "learning_rate": 0.00017014493239968082,
      "loss": 3.4543,
      "step": 160140
    },
    {
      "epoch": 0.14928571353665568,
      "grad_norm": 0.9260316491127014,
      "learning_rate": 0.00017014306807758566,
      "loss": 3.5212,
      "step": 160150
    },
    {
      "epoch": 0.14929503515473477,
      "grad_norm": 0.9357575178146362,
      "learning_rate": 0.00017014120375549044,
      "loss": 3.4721,
      "step": 160160
    },
    {
      "epoch": 0.14930435677281387,
      "grad_norm": 0.978412389755249,
      "learning_rate": 0.00017013933943339524,
      "loss": 3.6755,
      "step": 160170
    },
    {
      "epoch": 0.14931367839089296,
      "grad_norm": 1.1729519367218018,
      "learning_rate": 0.00017013747511130002,
      "loss": 3.4401,
      "step": 160180
    },
    {
      "epoch": 0.14932300000897206,
      "grad_norm": 1.3106648921966553,
      "learning_rate": 0.00017013561078920483,
      "loss": 3.621,
      "step": 160190
    },
    {
      "epoch": 0.14933232162705115,
      "grad_norm": 0.9303097128868103,
      "learning_rate": 0.00017013374646710963,
      "loss": 3.6555,
      "step": 160200
    },
    {
      "epoch": 0.14934164324513025,
      "grad_norm": 0.990278959274292,
      "learning_rate": 0.00017013188214501444,
      "loss": 3.7749,
      "step": 160210
    },
    {
      "epoch": 0.14935096486320934,
      "grad_norm": 0.9664717316627502,
      "learning_rate": 0.00017013001782291925,
      "loss": 3.5636,
      "step": 160220
    },
    {
      "epoch": 0.14936028648128843,
      "grad_norm": 1.0181362628936768,
      "learning_rate": 0.00017012815350082403,
      "loss": 3.5821,
      "step": 160230
    },
    {
      "epoch": 0.14936960809936753,
      "grad_norm": 1.1730103492736816,
      "learning_rate": 0.00017012628917872883,
      "loss": 3.6193,
      "step": 160240
    },
    {
      "epoch": 0.1493789297174466,
      "grad_norm": 1.210167646408081,
      "learning_rate": 0.00017012442485663364,
      "loss": 3.555,
      "step": 160250
    },
    {
      "epoch": 0.1493882513355257,
      "grad_norm": 1.1615939140319824,
      "learning_rate": 0.00017012256053453844,
      "loss": 3.6053,
      "step": 160260
    },
    {
      "epoch": 0.14939757295360478,
      "grad_norm": 1.079638957977295,
      "learning_rate": 0.00017012069621244325,
      "loss": 3.5841,
      "step": 160270
    },
    {
      "epoch": 0.14940689457168388,
      "grad_norm": 1.0543289184570312,
      "learning_rate": 0.00017011883189034803,
      "loss": 3.5913,
      "step": 160280
    },
    {
      "epoch": 0.14941621618976297,
      "grad_norm": 1.1262460947036743,
      "learning_rate": 0.00017011696756825284,
      "loss": 3.4919,
      "step": 160290
    },
    {
      "epoch": 0.14942553780784207,
      "grad_norm": 1.0850677490234375,
      "learning_rate": 0.00017011510324615764,
      "loss": 3.6344,
      "step": 160300
    },
    {
      "epoch": 0.14943485942592116,
      "grad_norm": 1.0235317945480347,
      "learning_rate": 0.00017011323892406242,
      "loss": 3.376,
      "step": 160310
    },
    {
      "epoch": 0.14944418104400026,
      "grad_norm": 1.1533020734786987,
      "learning_rate": 0.00017011137460196726,
      "loss": 3.5575,
      "step": 160320
    },
    {
      "epoch": 0.14945350266207935,
      "grad_norm": 1.0269486904144287,
      "learning_rate": 0.00017010951027987203,
      "loss": 3.4947,
      "step": 160330
    },
    {
      "epoch": 0.14946282428015845,
      "grad_norm": 1.1781367063522339,
      "learning_rate": 0.00017010764595777684,
      "loss": 3.5746,
      "step": 160340
    },
    {
      "epoch": 0.1494721458982375,
      "grad_norm": 1.3315365314483643,
      "learning_rate": 0.00017010578163568165,
      "loss": 3.4938,
      "step": 160350
    },
    {
      "epoch": 0.1494814675163166,
      "grad_norm": 1.2895150184631348,
      "learning_rate": 0.00017010391731358643,
      "loss": 3.6265,
      "step": 160360
    },
    {
      "epoch": 0.1494907891343957,
      "grad_norm": 1.216382384300232,
      "learning_rate": 0.00017010205299149123,
      "loss": 3.588,
      "step": 160370
    },
    {
      "epoch": 0.1495001107524748,
      "grad_norm": 1.000925898551941,
      "learning_rate": 0.00017010018866939604,
      "loss": 3.6258,
      "step": 160380
    },
    {
      "epoch": 0.1495094323705539,
      "grad_norm": 1.1233149766921997,
      "learning_rate": 0.00017009832434730085,
      "loss": 3.6495,
      "step": 160390
    },
    {
      "epoch": 0.14951875398863299,
      "grad_norm": 1.030566930770874,
      "learning_rate": 0.00017009646002520565,
      "loss": 3.6385,
      "step": 160400
    },
    {
      "epoch": 0.14952807560671208,
      "grad_norm": 1.1066230535507202,
      "learning_rate": 0.00017009459570311043,
      "loss": 3.6611,
      "step": 160410
    },
    {
      "epoch": 0.14953739722479117,
      "grad_norm": 1.1752015352249146,
      "learning_rate": 0.00017009273138101524,
      "loss": 3.465,
      "step": 160420
    },
    {
      "epoch": 0.14954671884287027,
      "grad_norm": 0.83444744348526,
      "learning_rate": 0.00017009086705892004,
      "loss": 3.5284,
      "step": 160430
    },
    {
      "epoch": 0.14955604046094936,
      "grad_norm": 1.1087735891342163,
      "learning_rate": 0.00017008900273682485,
      "loss": 3.6763,
      "step": 160440
    },
    {
      "epoch": 0.14956536207902846,
      "grad_norm": 0.9212445616722107,
      "learning_rate": 0.00017008713841472966,
      "loss": 3.7713,
      "step": 160450
    },
    {
      "epoch": 0.14957468369710752,
      "grad_norm": 1.0702846050262451,
      "learning_rate": 0.00017008527409263444,
      "loss": 3.5511,
      "step": 160460
    },
    {
      "epoch": 0.14958400531518662,
      "grad_norm": 1.1077653169631958,
      "learning_rate": 0.00017008340977053924,
      "loss": 3.5046,
      "step": 160470
    },
    {
      "epoch": 0.1495933269332657,
      "grad_norm": 0.9966021180152893,
      "learning_rate": 0.00017008154544844405,
      "loss": 3.543,
      "step": 160480
    },
    {
      "epoch": 0.1496026485513448,
      "grad_norm": 0.9726262092590332,
      "learning_rate": 0.00017007968112634885,
      "loss": 3.3665,
      "step": 160490
    },
    {
      "epoch": 0.1496119701694239,
      "grad_norm": 1.0212757587432861,
      "learning_rate": 0.00017007781680425366,
      "loss": 3.6351,
      "step": 160500
    },
    {
      "epoch": 0.149621291787503,
      "grad_norm": 1.1703176498413086,
      "learning_rate": 0.00017007595248215844,
      "loss": 3.646,
      "step": 160510
    },
    {
      "epoch": 0.1496306134055821,
      "grad_norm": 1.145171046257019,
      "learning_rate": 0.00017007408816006325,
      "loss": 3.4658,
      "step": 160520
    },
    {
      "epoch": 0.14963993502366119,
      "grad_norm": 1.0144453048706055,
      "learning_rate": 0.00017007222383796803,
      "loss": 3.5979,
      "step": 160530
    },
    {
      "epoch": 0.14964925664174028,
      "grad_norm": 1.1044682264328003,
      "learning_rate": 0.00017007035951587286,
      "loss": 3.5299,
      "step": 160540
    },
    {
      "epoch": 0.14965857825981937,
      "grad_norm": 0.9388201236724854,
      "learning_rate": 0.00017006849519377766,
      "loss": 3.5076,
      "step": 160550
    },
    {
      "epoch": 0.14966789987789844,
      "grad_norm": 1.1639153957366943,
      "learning_rate": 0.00017006663087168244,
      "loss": 3.6244,
      "step": 160560
    },
    {
      "epoch": 0.14967722149597754,
      "grad_norm": 1.0754073858261108,
      "learning_rate": 0.00017006476654958725,
      "loss": 3.6769,
      "step": 160570
    },
    {
      "epoch": 0.14968654311405663,
      "grad_norm": 1.139680027961731,
      "learning_rate": 0.00017006290222749203,
      "loss": 3.7132,
      "step": 160580
    },
    {
      "epoch": 0.14969586473213572,
      "grad_norm": 1.2271389961242676,
      "learning_rate": 0.00017006103790539684,
      "loss": 3.5412,
      "step": 160590
    },
    {
      "epoch": 0.14970518635021482,
      "grad_norm": 0.9597752094268799,
      "learning_rate": 0.00017005917358330167,
      "loss": 3.561,
      "step": 160600
    },
    {
      "epoch": 0.14971450796829391,
      "grad_norm": 1.0011094808578491,
      "learning_rate": 0.00017005730926120645,
      "loss": 3.695,
      "step": 160610
    },
    {
      "epoch": 0.149723829586373,
      "grad_norm": 1.0039223432540894,
      "learning_rate": 0.00017005544493911125,
      "loss": 3.5033,
      "step": 160620
    },
    {
      "epoch": 0.1497331512044521,
      "grad_norm": 0.9472789168357849,
      "learning_rate": 0.00017005358061701603,
      "loss": 3.6532,
      "step": 160630
    },
    {
      "epoch": 0.1497424728225312,
      "grad_norm": 1.0185555219650269,
      "learning_rate": 0.00017005171629492084,
      "loss": 3.4363,
      "step": 160640
    },
    {
      "epoch": 0.1497517944406103,
      "grad_norm": 1.0200773477554321,
      "learning_rate": 0.00017004985197282565,
      "loss": 3.589,
      "step": 160650
    },
    {
      "epoch": 0.1497611160586894,
      "grad_norm": 1.1340610980987549,
      "learning_rate": 0.00017004798765073045,
      "loss": 3.5896,
      "step": 160660
    },
    {
      "epoch": 0.14977043767676845,
      "grad_norm": 1.1247076988220215,
      "learning_rate": 0.00017004612332863526,
      "loss": 3.4495,
      "step": 160670
    },
    {
      "epoch": 0.14977975929484755,
      "grad_norm": 1.2144806385040283,
      "learning_rate": 0.00017004425900654004,
      "loss": 3.692,
      "step": 160680
    },
    {
      "epoch": 0.14978908091292664,
      "grad_norm": 0.9350613355636597,
      "learning_rate": 0.00017004239468444484,
      "loss": 3.6252,
      "step": 160690
    },
    {
      "epoch": 0.14979840253100574,
      "grad_norm": 0.9096329212188721,
      "learning_rate": 0.00017004053036234965,
      "loss": 3.3621,
      "step": 160700
    },
    {
      "epoch": 0.14980772414908483,
      "grad_norm": 0.9872866272926331,
      "learning_rate": 0.00017003866604025446,
      "loss": 3.6247,
      "step": 160710
    },
    {
      "epoch": 0.14981704576716393,
      "grad_norm": 1.0751127004623413,
      "learning_rate": 0.00017003680171815926,
      "loss": 3.7722,
      "step": 160720
    },
    {
      "epoch": 0.14982636738524302,
      "grad_norm": 1.1838552951812744,
      "learning_rate": 0.00017003493739606404,
      "loss": 3.3536,
      "step": 160730
    },
    {
      "epoch": 0.14983568900332211,
      "grad_norm": 0.9117421507835388,
      "learning_rate": 0.00017003307307396885,
      "loss": 3.6333,
      "step": 160740
    },
    {
      "epoch": 0.1498450106214012,
      "grad_norm": 1.0330579280853271,
      "learning_rate": 0.00017003120875187366,
      "loss": 3.4298,
      "step": 160750
    },
    {
      "epoch": 0.1498543322394803,
      "grad_norm": 1.033731460571289,
      "learning_rate": 0.00017002934442977843,
      "loss": 3.2988,
      "step": 160760
    },
    {
      "epoch": 0.14986365385755937,
      "grad_norm": 1.0639053583145142,
      "learning_rate": 0.00017002748010768327,
      "loss": 3.6182,
      "step": 160770
    },
    {
      "epoch": 0.14987297547563846,
      "grad_norm": 0.8421518802642822,
      "learning_rate": 0.00017002561578558805,
      "loss": 3.4258,
      "step": 160780
    },
    {
      "epoch": 0.14988229709371756,
      "grad_norm": 1.2175461053848267,
      "learning_rate": 0.00017002375146349285,
      "loss": 3.6085,
      "step": 160790
    },
    {
      "epoch": 0.14989161871179665,
      "grad_norm": 0.9750734567642212,
      "learning_rate": 0.00017002188714139766,
      "loss": 3.6929,
      "step": 160800
    },
    {
      "epoch": 0.14990094032987575,
      "grad_norm": 0.8801120519638062,
      "learning_rate": 0.00017002002281930244,
      "loss": 3.7025,
      "step": 160810
    },
    {
      "epoch": 0.14991026194795484,
      "grad_norm": 1.3663866519927979,
      "learning_rate": 0.00017001815849720725,
      "loss": 3.4647,
      "step": 160820
    },
    {
      "epoch": 0.14991958356603394,
      "grad_norm": 0.9272880554199219,
      "learning_rate": 0.00017001629417511205,
      "loss": 3.6373,
      "step": 160830
    },
    {
      "epoch": 0.14992890518411303,
      "grad_norm": 1.1306735277175903,
      "learning_rate": 0.00017001442985301686,
      "loss": 3.7396,
      "step": 160840
    },
    {
      "epoch": 0.14993822680219213,
      "grad_norm": 1.0083657503128052,
      "learning_rate": 0.00017001256553092166,
      "loss": 3.7202,
      "step": 160850
    },
    {
      "epoch": 0.14994754842027122,
      "grad_norm": 1.0469067096710205,
      "learning_rate": 0.00017001070120882644,
      "loss": 3.7062,
      "step": 160860
    },
    {
      "epoch": 0.14995687003835032,
      "grad_norm": 0.8811081051826477,
      "learning_rate": 0.00017000883688673125,
      "loss": 3.6108,
      "step": 160870
    },
    {
      "epoch": 0.14996619165642938,
      "grad_norm": 1.2633380889892578,
      "learning_rate": 0.00017000697256463606,
      "loss": 3.5781,
      "step": 160880
    },
    {
      "epoch": 0.14997551327450848,
      "grad_norm": 1.084689974784851,
      "learning_rate": 0.00017000510824254086,
      "loss": 3.5918,
      "step": 160890
    },
    {
      "epoch": 0.14998483489258757,
      "grad_norm": 0.9564747214317322,
      "learning_rate": 0.00017000324392044567,
      "loss": 3.5203,
      "step": 160900
    },
    {
      "epoch": 0.14999415651066667,
      "grad_norm": 1.0389511585235596,
      "learning_rate": 0.00017000137959835045,
      "loss": 3.3852,
      "step": 160910
    },
    {
      "epoch": 0.15000347812874576,
      "grad_norm": 0.9679903984069824,
      "learning_rate": 0.00016999951527625525,
      "loss": 3.6282,
      "step": 160920
    },
    {
      "epoch": 0.15001279974682485,
      "grad_norm": 0.9580904245376587,
      "learning_rate": 0.00016999765095416006,
      "loss": 3.5393,
      "step": 160930
    },
    {
      "epoch": 0.15002212136490395,
      "grad_norm": 0.902737557888031,
      "learning_rate": 0.00016999578663206487,
      "loss": 3.5777,
      "step": 160940
    },
    {
      "epoch": 0.15003144298298304,
      "grad_norm": 1.162060260772705,
      "learning_rate": 0.00016999392230996967,
      "loss": 3.5359,
      "step": 160950
    },
    {
      "epoch": 0.15004076460106214,
      "grad_norm": 0.9199435114860535,
      "learning_rate": 0.00016999205798787445,
      "loss": 3.6665,
      "step": 160960
    },
    {
      "epoch": 0.15005008621914123,
      "grad_norm": 1.1291331052780151,
      "learning_rate": 0.00016999019366577926,
      "loss": 3.5281,
      "step": 160970
    },
    {
      "epoch": 0.1500594078372203,
      "grad_norm": 0.8194773197174072,
      "learning_rate": 0.00016998832934368404,
      "loss": 3.5728,
      "step": 160980
    },
    {
      "epoch": 0.1500687294552994,
      "grad_norm": 1.0805637836456299,
      "learning_rate": 0.00016998646502158887,
      "loss": 3.529,
      "step": 160990
    },
    {
      "epoch": 0.1500780510733785,
      "grad_norm": 1.3497298955917358,
      "learning_rate": 0.00016998460069949368,
      "loss": 3.6576,
      "step": 161000
    },
    {
      "epoch": 0.15008737269145758,
      "grad_norm": 1.0528836250305176,
      "learning_rate": 0.00016998273637739846,
      "loss": 3.4561,
      "step": 161010
    },
    {
      "epoch": 0.15009669430953668,
      "grad_norm": 1.1179869174957275,
      "learning_rate": 0.00016998087205530326,
      "loss": 3.5575,
      "step": 161020
    },
    {
      "epoch": 0.15010601592761577,
      "grad_norm": 1.102738618850708,
      "learning_rate": 0.00016997900773320804,
      "loss": 3.4973,
      "step": 161030
    },
    {
      "epoch": 0.15011533754569487,
      "grad_norm": 1.1660577058792114,
      "learning_rate": 0.00016997714341111285,
      "loss": 3.6191,
      "step": 161040
    },
    {
      "epoch": 0.15012465916377396,
      "grad_norm": 1.0798497200012207,
      "learning_rate": 0.00016997527908901768,
      "loss": 3.6513,
      "step": 161050
    },
    {
      "epoch": 0.15013398078185305,
      "grad_norm": 1.0817018747329712,
      "learning_rate": 0.00016997341476692246,
      "loss": 3.5114,
      "step": 161060
    },
    {
      "epoch": 0.15014330239993215,
      "grad_norm": 1.012367606163025,
      "learning_rate": 0.00016997155044482727,
      "loss": 3.5339,
      "step": 161070
    },
    {
      "epoch": 0.15015262401801124,
      "grad_norm": 1.0050086975097656,
      "learning_rate": 0.00016996968612273205,
      "loss": 3.5062,
      "step": 161080
    },
    {
      "epoch": 0.1501619456360903,
      "grad_norm": 1.1379531621932983,
      "learning_rate": 0.00016996782180063685,
      "loss": 3.6218,
      "step": 161090
    },
    {
      "epoch": 0.1501712672541694,
      "grad_norm": 0.9892594814300537,
      "learning_rate": 0.00016996595747854166,
      "loss": 3.5817,
      "step": 161100
    },
    {
      "epoch": 0.1501805888722485,
      "grad_norm": 1.1419938802719116,
      "learning_rate": 0.00016996409315644647,
      "loss": 3.6277,
      "step": 161110
    },
    {
      "epoch": 0.1501899104903276,
      "grad_norm": 0.9308612942695618,
      "learning_rate": 0.00016996222883435127,
      "loss": 3.8712,
      "step": 161120
    },
    {
      "epoch": 0.1501992321084067,
      "grad_norm": 1.1252418756484985,
      "learning_rate": 0.00016996036451225605,
      "loss": 3.5071,
      "step": 161130
    },
    {
      "epoch": 0.15020855372648578,
      "grad_norm": 1.0376323461532593,
      "learning_rate": 0.00016995850019016086,
      "loss": 3.6206,
      "step": 161140
    },
    {
      "epoch": 0.15021787534456488,
      "grad_norm": 0.905649721622467,
      "learning_rate": 0.00016995663586806566,
      "loss": 3.4633,
      "step": 161150
    },
    {
      "epoch": 0.15022719696264397,
      "grad_norm": 1.0619523525238037,
      "learning_rate": 0.00016995477154597047,
      "loss": 3.6229,
      "step": 161160
    },
    {
      "epoch": 0.15023651858072307,
      "grad_norm": 1.0565115213394165,
      "learning_rate": 0.00016995290722387528,
      "loss": 3.5208,
      "step": 161170
    },
    {
      "epoch": 0.15024584019880216,
      "grad_norm": 1.1526014804840088,
      "learning_rate": 0.00016995104290178006,
      "loss": 3.6047,
      "step": 161180
    },
    {
      "epoch": 0.15025516181688123,
      "grad_norm": 1.0485429763793945,
      "learning_rate": 0.00016994917857968486,
      "loss": 3.5036,
      "step": 161190
    },
    {
      "epoch": 0.15026448343496032,
      "grad_norm": 1.0477460622787476,
      "learning_rate": 0.00016994731425758967,
      "loss": 3.4024,
      "step": 161200
    },
    {
      "epoch": 0.15027380505303942,
      "grad_norm": 1.1167559623718262,
      "learning_rate": 0.00016994544993549445,
      "loss": 3.7084,
      "step": 161210
    },
    {
      "epoch": 0.1502831266711185,
      "grad_norm": 1.1430023908615112,
      "learning_rate": 0.00016994358561339928,
      "loss": 3.5627,
      "step": 161220
    },
    {
      "epoch": 0.1502924482891976,
      "grad_norm": 1.1368496417999268,
      "learning_rate": 0.00016994172129130406,
      "loss": 3.4595,
      "step": 161230
    },
    {
      "epoch": 0.1503017699072767,
      "grad_norm": 1.0191097259521484,
      "learning_rate": 0.00016993985696920887,
      "loss": 3.5302,
      "step": 161240
    },
    {
      "epoch": 0.1503110915253558,
      "grad_norm": 1.14041268825531,
      "learning_rate": 0.00016993799264711367,
      "loss": 3.7909,
      "step": 161250
    },
    {
      "epoch": 0.1503204131434349,
      "grad_norm": 1.0859501361846924,
      "learning_rate": 0.00016993612832501845,
      "loss": 3.5651,
      "step": 161260
    },
    {
      "epoch": 0.15032973476151398,
      "grad_norm": 1.014524221420288,
      "learning_rate": 0.00016993426400292326,
      "loss": 3.7225,
      "step": 161270
    },
    {
      "epoch": 0.15033905637959308,
      "grad_norm": 0.7994191646575928,
      "learning_rate": 0.00016993239968082806,
      "loss": 3.6417,
      "step": 161280
    },
    {
      "epoch": 0.15034837799767217,
      "grad_norm": 0.9436491131782532,
      "learning_rate": 0.00016993053535873287,
      "loss": 3.6466,
      "step": 161290
    },
    {
      "epoch": 0.15035769961575124,
      "grad_norm": 0.9721137881278992,
      "learning_rate": 0.00016992867103663768,
      "loss": 3.3622,
      "step": 161300
    },
    {
      "epoch": 0.15036702123383033,
      "grad_norm": 0.916605532169342,
      "learning_rate": 0.00016992680671454246,
      "loss": 3.6072,
      "step": 161310
    },
    {
      "epoch": 0.15037634285190943,
      "grad_norm": 1.110803246498108,
      "learning_rate": 0.00016992494239244726,
      "loss": 3.5771,
      "step": 161320
    },
    {
      "epoch": 0.15038566446998852,
      "grad_norm": 1.0435733795166016,
      "learning_rate": 0.00016992307807035207,
      "loss": 3.4884,
      "step": 161330
    },
    {
      "epoch": 0.15039498608806762,
      "grad_norm": 1.1560862064361572,
      "learning_rate": 0.00016992121374825688,
      "loss": 3.5241,
      "step": 161340
    },
    {
      "epoch": 0.1504043077061467,
      "grad_norm": 1.0450880527496338,
      "learning_rate": 0.00016991934942616168,
      "loss": 3.6251,
      "step": 161350
    },
    {
      "epoch": 0.1504136293242258,
      "grad_norm": 1.0415740013122559,
      "learning_rate": 0.00016991748510406646,
      "loss": 3.6207,
      "step": 161360
    },
    {
      "epoch": 0.1504229509423049,
      "grad_norm": 0.9073272943496704,
      "learning_rate": 0.00016991562078197127,
      "loss": 3.5765,
      "step": 161370
    },
    {
      "epoch": 0.150432272560384,
      "grad_norm": 1.0834301710128784,
      "learning_rate": 0.00016991375645987607,
      "loss": 3.6174,
      "step": 161380
    },
    {
      "epoch": 0.1504415941784631,
      "grad_norm": 1.093972086906433,
      "learning_rate": 0.00016991189213778088,
      "loss": 3.5979,
      "step": 161390
    },
    {
      "epoch": 0.15045091579654216,
      "grad_norm": 1.0590850114822388,
      "learning_rate": 0.00016991002781568569,
      "loss": 3.6549,
      "step": 161400
    },
    {
      "epoch": 0.15046023741462125,
      "grad_norm": 1.2144583463668823,
      "learning_rate": 0.00016990816349359047,
      "loss": 3.5032,
      "step": 161410
    },
    {
      "epoch": 0.15046955903270035,
      "grad_norm": 0.9721841216087341,
      "learning_rate": 0.00016990629917149527,
      "loss": 3.6021,
      "step": 161420
    },
    {
      "epoch": 0.15047888065077944,
      "grad_norm": 1.2282416820526123,
      "learning_rate": 0.00016990443484940005,
      "loss": 3.6192,
      "step": 161430
    },
    {
      "epoch": 0.15048820226885853,
      "grad_norm": 1.079882264137268,
      "learning_rate": 0.00016990257052730488,
      "loss": 3.6871,
      "step": 161440
    },
    {
      "epoch": 0.15049752388693763,
      "grad_norm": 0.9867008924484253,
      "learning_rate": 0.00016990070620520966,
      "loss": 3.5127,
      "step": 161450
    },
    {
      "epoch": 0.15050684550501672,
      "grad_norm": 0.9659994840621948,
      "learning_rate": 0.00016989884188311447,
      "loss": 3.5875,
      "step": 161460
    },
    {
      "epoch": 0.15051616712309582,
      "grad_norm": 1.0353031158447266,
      "learning_rate": 0.00016989697756101928,
      "loss": 3.3451,
      "step": 161470
    },
    {
      "epoch": 0.1505254887411749,
      "grad_norm": 0.935965359210968,
      "learning_rate": 0.00016989511323892406,
      "loss": 3.3866,
      "step": 161480
    },
    {
      "epoch": 0.150534810359254,
      "grad_norm": 0.9397236108779907,
      "learning_rate": 0.00016989324891682886,
      "loss": 3.5726,
      "step": 161490
    },
    {
      "epoch": 0.1505441319773331,
      "grad_norm": 1.1369285583496094,
      "learning_rate": 0.00016989138459473367,
      "loss": 3.4699,
      "step": 161500
    },
    {
      "epoch": 0.15055345359541217,
      "grad_norm": 1.0530211925506592,
      "learning_rate": 0.00016988952027263847,
      "loss": 3.4453,
      "step": 161510
    },
    {
      "epoch": 0.15056277521349126,
      "grad_norm": 1.1319968700408936,
      "learning_rate": 0.00016988765595054328,
      "loss": 3.6407,
      "step": 161520
    },
    {
      "epoch": 0.15057209683157036,
      "grad_norm": 1.0168843269348145,
      "learning_rate": 0.00016988579162844806,
      "loss": 3.5606,
      "step": 161530
    },
    {
      "epoch": 0.15058141844964945,
      "grad_norm": 0.8879588842391968,
      "learning_rate": 0.00016988392730635287,
      "loss": 3.5545,
      "step": 161540
    },
    {
      "epoch": 0.15059074006772855,
      "grad_norm": 1.069268822669983,
      "learning_rate": 0.00016988206298425767,
      "loss": 3.6495,
      "step": 161550
    },
    {
      "epoch": 0.15060006168580764,
      "grad_norm": 0.9570314288139343,
      "learning_rate": 0.00016988019866216248,
      "loss": 3.6938,
      "step": 161560
    },
    {
      "epoch": 0.15060938330388673,
      "grad_norm": 1.0120877027511597,
      "learning_rate": 0.00016987833434006728,
      "loss": 3.603,
      "step": 161570
    },
    {
      "epoch": 0.15061870492196583,
      "grad_norm": 1.022396206855774,
      "learning_rate": 0.00016987647001797206,
      "loss": 3.6109,
      "step": 161580
    },
    {
      "epoch": 0.15062802654004492,
      "grad_norm": 1.11978280544281,
      "learning_rate": 0.00016987460569587687,
      "loss": 3.518,
      "step": 161590
    },
    {
      "epoch": 0.15063734815812402,
      "grad_norm": 0.7759900689125061,
      "learning_rate": 0.00016987274137378165,
      "loss": 3.7262,
      "step": 161600
    },
    {
      "epoch": 0.15064666977620308,
      "grad_norm": 0.9151729941368103,
      "learning_rate": 0.00016987087705168648,
      "loss": 3.7642,
      "step": 161610
    },
    {
      "epoch": 0.15065599139428218,
      "grad_norm": 1.189923644065857,
      "learning_rate": 0.0001698690127295913,
      "loss": 3.5828,
      "step": 161620
    },
    {
      "epoch": 0.15066531301236127,
      "grad_norm": 1.120405912399292,
      "learning_rate": 0.00016986714840749607,
      "loss": 3.4306,
      "step": 161630
    },
    {
      "epoch": 0.15067463463044037,
      "grad_norm": 1.546030044555664,
      "learning_rate": 0.00016986528408540087,
      "loss": 3.7155,
      "step": 161640
    },
    {
      "epoch": 0.15068395624851946,
      "grad_norm": 0.90840083360672,
      "learning_rate": 0.00016986341976330565,
      "loss": 3.5996,
      "step": 161650
    },
    {
      "epoch": 0.15069327786659856,
      "grad_norm": 1.1088391542434692,
      "learning_rate": 0.00016986155544121046,
      "loss": 3.6138,
      "step": 161660
    },
    {
      "epoch": 0.15070259948467765,
      "grad_norm": 1.0576465129852295,
      "learning_rate": 0.0001698596911191153,
      "loss": 3.5643,
      "step": 161670
    },
    {
      "epoch": 0.15071192110275675,
      "grad_norm": 0.8738207221031189,
      "learning_rate": 0.00016985782679702007,
      "loss": 3.6542,
      "step": 161680
    },
    {
      "epoch": 0.15072124272083584,
      "grad_norm": 0.9358232021331787,
      "learning_rate": 0.00016985596247492488,
      "loss": 3.4798,
      "step": 161690
    },
    {
      "epoch": 0.15073056433891494,
      "grad_norm": 0.8576918244361877,
      "learning_rate": 0.00016985409815282966,
      "loss": 3.7505,
      "step": 161700
    },
    {
      "epoch": 0.15073988595699403,
      "grad_norm": 1.2403686046600342,
      "learning_rate": 0.00016985223383073446,
      "loss": 3.4408,
      "step": 161710
    },
    {
      "epoch": 0.1507492075750731,
      "grad_norm": 1.122179627418518,
      "learning_rate": 0.00016985036950863927,
      "loss": 3.5375,
      "step": 161720
    },
    {
      "epoch": 0.1507585291931522,
      "grad_norm": 0.8739103078842163,
      "learning_rate": 0.00016984850518654408,
      "loss": 3.6287,
      "step": 161730
    },
    {
      "epoch": 0.15076785081123129,
      "grad_norm": 1.18622624874115,
      "learning_rate": 0.00016984664086444888,
      "loss": 3.4795,
      "step": 161740
    },
    {
      "epoch": 0.15077717242931038,
      "grad_norm": 1.163202166557312,
      "learning_rate": 0.00016984477654235366,
      "loss": 3.7331,
      "step": 161750
    },
    {
      "epoch": 0.15078649404738947,
      "grad_norm": 0.9859751462936401,
      "learning_rate": 0.00016984291222025847,
      "loss": 3.8936,
      "step": 161760
    },
    {
      "epoch": 0.15079581566546857,
      "grad_norm": 1.0658093690872192,
      "learning_rate": 0.00016984104789816328,
      "loss": 3.6863,
      "step": 161770
    },
    {
      "epoch": 0.15080513728354766,
      "grad_norm": 1.1504467725753784,
      "learning_rate": 0.00016983918357606808,
      "loss": 3.6172,
      "step": 161780
    },
    {
      "epoch": 0.15081445890162676,
      "grad_norm": 1.185095191001892,
      "learning_rate": 0.0001698373192539729,
      "loss": 3.5991,
      "step": 161790
    },
    {
      "epoch": 0.15082378051970585,
      "grad_norm": 1.0362939834594727,
      "learning_rate": 0.00016983545493187767,
      "loss": 3.6811,
      "step": 161800
    },
    {
      "epoch": 0.15083310213778495,
      "grad_norm": 1.016562819480896,
      "learning_rate": 0.00016983359060978247,
      "loss": 3.5041,
      "step": 161810
    },
    {
      "epoch": 0.150842423755864,
      "grad_norm": 0.9726581573486328,
      "learning_rate": 0.00016983172628768728,
      "loss": 3.7109,
      "step": 161820
    },
    {
      "epoch": 0.1508517453739431,
      "grad_norm": 1.0429298877716064,
      "learning_rate": 0.00016982986196559209,
      "loss": 3.6394,
      "step": 161830
    },
    {
      "epoch": 0.1508610669920222,
      "grad_norm": 1.005928874015808,
      "learning_rate": 0.0001698279976434969,
      "loss": 3.6333,
      "step": 161840
    },
    {
      "epoch": 0.1508703886101013,
      "grad_norm": 1.0627684593200684,
      "learning_rate": 0.00016982613332140167,
      "loss": 3.6611,
      "step": 161850
    },
    {
      "epoch": 0.1508797102281804,
      "grad_norm": 1.1898488998413086,
      "learning_rate": 0.00016982426899930648,
      "loss": 3.2139,
      "step": 161860
    },
    {
      "epoch": 0.15088903184625949,
      "grad_norm": 0.995201587677002,
      "learning_rate": 0.00016982240467721128,
      "loss": 3.345,
      "step": 161870
    },
    {
      "epoch": 0.15089835346433858,
      "grad_norm": 1.1440823078155518,
      "learning_rate": 0.00016982054035511606,
      "loss": 3.6496,
      "step": 161880
    },
    {
      "epoch": 0.15090767508241767,
      "grad_norm": 1.5850187540054321,
      "learning_rate": 0.0001698186760330209,
      "loss": 3.511,
      "step": 161890
    },
    {
      "epoch": 0.15091699670049677,
      "grad_norm": 1.028471827507019,
      "learning_rate": 0.00016981681171092568,
      "loss": 3.4474,
      "step": 161900
    },
    {
      "epoch": 0.15092631831857586,
      "grad_norm": 0.9207854270935059,
      "learning_rate": 0.00016981494738883048,
      "loss": 3.6106,
      "step": 161910
    },
    {
      "epoch": 0.15093563993665493,
      "grad_norm": 1.2252774238586426,
      "learning_rate": 0.0001698130830667353,
      "loss": 3.3661,
      "step": 161920
    },
    {
      "epoch": 0.15094496155473403,
      "grad_norm": 0.9145143032073975,
      "learning_rate": 0.00016981121874464007,
      "loss": 3.6644,
      "step": 161930
    },
    {
      "epoch": 0.15095428317281312,
      "grad_norm": 1.2665561437606812,
      "learning_rate": 0.00016980935442254487,
      "loss": 3.5729,
      "step": 161940
    },
    {
      "epoch": 0.15096360479089221,
      "grad_norm": 0.9316766262054443,
      "learning_rate": 0.00016980749010044968,
      "loss": 3.5863,
      "step": 161950
    },
    {
      "epoch": 0.1509729264089713,
      "grad_norm": 1.0197340250015259,
      "learning_rate": 0.0001698056257783545,
      "loss": 3.5265,
      "step": 161960
    },
    {
      "epoch": 0.1509822480270504,
      "grad_norm": 1.0325214862823486,
      "learning_rate": 0.0001698037614562593,
      "loss": 3.5957,
      "step": 161970
    },
    {
      "epoch": 0.1509915696451295,
      "grad_norm": 1.1213372945785522,
      "learning_rate": 0.00016980189713416407,
      "loss": 3.5629,
      "step": 161980
    },
    {
      "epoch": 0.1510008912632086,
      "grad_norm": 1.2069166898727417,
      "learning_rate": 0.00016980003281206888,
      "loss": 3.5079,
      "step": 161990
    },
    {
      "epoch": 0.1510102128812877,
      "grad_norm": 1.30852210521698,
      "learning_rate": 0.00016979816848997369,
      "loss": 3.5711,
      "step": 162000
    },
    {
      "epoch": 0.15101953449936678,
      "grad_norm": 1.0810370445251465,
      "learning_rate": 0.0001697963041678785,
      "loss": 3.6472,
      "step": 162010
    },
    {
      "epoch": 0.15102885611744588,
      "grad_norm": 0.8410333395004272,
      "learning_rate": 0.0001697944398457833,
      "loss": 3.6246,
      "step": 162020
    },
    {
      "epoch": 0.15103817773552494,
      "grad_norm": 0.9458770751953125,
      "learning_rate": 0.00016979257552368808,
      "loss": 3.6807,
      "step": 162030
    },
    {
      "epoch": 0.15104749935360404,
      "grad_norm": 1.0880706310272217,
      "learning_rate": 0.00016979071120159288,
      "loss": 3.539,
      "step": 162040
    },
    {
      "epoch": 0.15105682097168313,
      "grad_norm": 1.2194793224334717,
      "learning_rate": 0.0001697888468794977,
      "loss": 3.5697,
      "step": 162050
    },
    {
      "epoch": 0.15106614258976223,
      "grad_norm": 1.08863365650177,
      "learning_rate": 0.0001697869825574025,
      "loss": 3.6476,
      "step": 162060
    },
    {
      "epoch": 0.15107546420784132,
      "grad_norm": 1.0167722702026367,
      "learning_rate": 0.0001697851182353073,
      "loss": 3.4976,
      "step": 162070
    },
    {
      "epoch": 0.15108478582592041,
      "grad_norm": 0.910174548625946,
      "learning_rate": 0.00016978325391321208,
      "loss": 3.6829,
      "step": 162080
    },
    {
      "epoch": 0.1510941074439995,
      "grad_norm": 0.9548686146736145,
      "learning_rate": 0.0001697813895911169,
      "loss": 3.5284,
      "step": 162090
    },
    {
      "epoch": 0.1511034290620786,
      "grad_norm": 1.1775310039520264,
      "learning_rate": 0.00016977952526902167,
      "loss": 3.3784,
      "step": 162100
    },
    {
      "epoch": 0.1511127506801577,
      "grad_norm": 1.1371667385101318,
      "learning_rate": 0.0001697776609469265,
      "loss": 3.5675,
      "step": 162110
    },
    {
      "epoch": 0.1511220722982368,
      "grad_norm": 1.0590628385543823,
      "learning_rate": 0.0001697757966248313,
      "loss": 3.5676,
      "step": 162120
    },
    {
      "epoch": 0.15113139391631586,
      "grad_norm": 1.2231370210647583,
      "learning_rate": 0.00016977393230273609,
      "loss": 3.6179,
      "step": 162130
    },
    {
      "epoch": 0.15114071553439495,
      "grad_norm": 1.2593375444412231,
      "learning_rate": 0.0001697720679806409,
      "loss": 3.611,
      "step": 162140
    },
    {
      "epoch": 0.15115003715247405,
      "grad_norm": 1.0116301774978638,
      "learning_rate": 0.00016977020365854567,
      "loss": 3.5457,
      "step": 162150
    },
    {
      "epoch": 0.15115935877055314,
      "grad_norm": 1.1729271411895752,
      "learning_rate": 0.00016976833933645048,
      "loss": 3.6478,
      "step": 162160
    },
    {
      "epoch": 0.15116868038863224,
      "grad_norm": 1.026216983795166,
      "learning_rate": 0.00016976647501435528,
      "loss": 3.5625,
      "step": 162170
    },
    {
      "epoch": 0.15117800200671133,
      "grad_norm": 1.263694167137146,
      "learning_rate": 0.0001697646106922601,
      "loss": 3.6842,
      "step": 162180
    },
    {
      "epoch": 0.15118732362479043,
      "grad_norm": 1.0838394165039062,
      "learning_rate": 0.0001697627463701649,
      "loss": 3.56,
      "step": 162190
    },
    {
      "epoch": 0.15119664524286952,
      "grad_norm": 1.0850110054016113,
      "learning_rate": 0.00016976088204806968,
      "loss": 3.5661,
      "step": 162200
    },
    {
      "epoch": 0.15120596686094862,
      "grad_norm": 1.1017926931381226,
      "learning_rate": 0.00016975901772597448,
      "loss": 3.3173,
      "step": 162210
    },
    {
      "epoch": 0.1512152884790277,
      "grad_norm": 1.0463701486587524,
      "learning_rate": 0.0001697571534038793,
      "loss": 3.5426,
      "step": 162220
    },
    {
      "epoch": 0.1512246100971068,
      "grad_norm": 1.1988109350204468,
      "learning_rate": 0.0001697552890817841,
      "loss": 3.5478,
      "step": 162230
    },
    {
      "epoch": 0.15123393171518587,
      "grad_norm": 1.0757910013198853,
      "learning_rate": 0.0001697534247596889,
      "loss": 3.8015,
      "step": 162240
    },
    {
      "epoch": 0.15124325333326497,
      "grad_norm": 1.0328975915908813,
      "learning_rate": 0.00016975156043759368,
      "loss": 3.4286,
      "step": 162250
    },
    {
      "epoch": 0.15125257495134406,
      "grad_norm": 0.927945077419281,
      "learning_rate": 0.0001697496961154985,
      "loss": 3.6319,
      "step": 162260
    },
    {
      "epoch": 0.15126189656942315,
      "grad_norm": 1.0402036905288696,
      "learning_rate": 0.0001697478317934033,
      "loss": 3.6587,
      "step": 162270
    },
    {
      "epoch": 0.15127121818750225,
      "grad_norm": 0.8899232149124146,
      "learning_rate": 0.0001697459674713081,
      "loss": 3.4862,
      "step": 162280
    },
    {
      "epoch": 0.15128053980558134,
      "grad_norm": 1.2495402097702026,
      "learning_rate": 0.0001697441031492129,
      "loss": 3.6509,
      "step": 162290
    },
    {
      "epoch": 0.15128986142366044,
      "grad_norm": 0.9308242201805115,
      "learning_rate": 0.00016974223882711768,
      "loss": 3.5263,
      "step": 162300
    },
    {
      "epoch": 0.15129918304173953,
      "grad_norm": 0.9247366786003113,
      "learning_rate": 0.0001697403745050225,
      "loss": 3.7446,
      "step": 162310
    },
    {
      "epoch": 0.15130850465981863,
      "grad_norm": 0.8062155246734619,
      "learning_rate": 0.0001697385101829273,
      "loss": 3.4803,
      "step": 162320
    },
    {
      "epoch": 0.15131782627789772,
      "grad_norm": 1.093896746635437,
      "learning_rate": 0.00016973664586083208,
      "loss": 3.6814,
      "step": 162330
    },
    {
      "epoch": 0.1513271478959768,
      "grad_norm": 1.0843300819396973,
      "learning_rate": 0.0001697347815387369,
      "loss": 3.4123,
      "step": 162340
    },
    {
      "epoch": 0.15133646951405588,
      "grad_norm": 1.0802749395370483,
      "learning_rate": 0.0001697329172166417,
      "loss": 3.5498,
      "step": 162350
    },
    {
      "epoch": 0.15134579113213498,
      "grad_norm": 1.1738826036453247,
      "learning_rate": 0.0001697310528945465,
      "loss": 3.6896,
      "step": 162360
    },
    {
      "epoch": 0.15135511275021407,
      "grad_norm": 0.9989337921142578,
      "learning_rate": 0.0001697291885724513,
      "loss": 3.6771,
      "step": 162370
    },
    {
      "epoch": 0.15136443436829317,
      "grad_norm": 1.016026496887207,
      "learning_rate": 0.00016972732425035608,
      "loss": 3.3846,
      "step": 162380
    },
    {
      "epoch": 0.15137375598637226,
      "grad_norm": 1.0263959169387817,
      "learning_rate": 0.0001697254599282609,
      "loss": 3.4965,
      "step": 162390
    },
    {
      "epoch": 0.15138307760445135,
      "grad_norm": 1.1279083490371704,
      "learning_rate": 0.0001697235956061657,
      "loss": 3.4548,
      "step": 162400
    },
    {
      "epoch": 0.15139239922253045,
      "grad_norm": 1.0837818384170532,
      "learning_rate": 0.0001697217312840705,
      "loss": 3.4924,
      "step": 162410
    },
    {
      "epoch": 0.15140172084060954,
      "grad_norm": 1.0556931495666504,
      "learning_rate": 0.0001697198669619753,
      "loss": 3.6273,
      "step": 162420
    },
    {
      "epoch": 0.15141104245868864,
      "grad_norm": 1.2934184074401855,
      "learning_rate": 0.00016971800263988009,
      "loss": 3.509,
      "step": 162430
    },
    {
      "epoch": 0.15142036407676773,
      "grad_norm": 1.3633309602737427,
      "learning_rate": 0.0001697161383177849,
      "loss": 3.4113,
      "step": 162440
    },
    {
      "epoch": 0.1514296856948468,
      "grad_norm": 1.2150492668151855,
      "learning_rate": 0.0001697142739956897,
      "loss": 3.4539,
      "step": 162450
    },
    {
      "epoch": 0.1514390073129259,
      "grad_norm": 1.3411133289337158,
      "learning_rate": 0.0001697124096735945,
      "loss": 3.7684,
      "step": 162460
    },
    {
      "epoch": 0.151448328931005,
      "grad_norm": 1.2722549438476562,
      "learning_rate": 0.0001697105453514993,
      "loss": 3.6802,
      "step": 162470
    },
    {
      "epoch": 0.15145765054908408,
      "grad_norm": 1.0354325771331787,
      "learning_rate": 0.0001697086810294041,
      "loss": 3.3964,
      "step": 162480
    },
    {
      "epoch": 0.15146697216716318,
      "grad_norm": 1.0377546548843384,
      "learning_rate": 0.0001697068167073089,
      "loss": 3.4592,
      "step": 162490
    },
    {
      "epoch": 0.15147629378524227,
      "grad_norm": 1.4990792274475098,
      "learning_rate": 0.0001697049523852137,
      "loss": 3.5949,
      "step": 162500
    },
    {
      "epoch": 0.15148561540332137,
      "grad_norm": 0.8926501274108887,
      "learning_rate": 0.0001697030880631185,
      "loss": 3.6205,
      "step": 162510
    },
    {
      "epoch": 0.15149493702140046,
      "grad_norm": 1.0723527669906616,
      "learning_rate": 0.00016970122374102331,
      "loss": 3.4625,
      "step": 162520
    },
    {
      "epoch": 0.15150425863947956,
      "grad_norm": 0.9993528723716736,
      "learning_rate": 0.0001696993594189281,
      "loss": 3.5143,
      "step": 162530
    },
    {
      "epoch": 0.15151358025755865,
      "grad_norm": 1.1961661577224731,
      "learning_rate": 0.0001696974950968329,
      "loss": 3.4146,
      "step": 162540
    },
    {
      "epoch": 0.15152290187563772,
      "grad_norm": 1.0837770700454712,
      "learning_rate": 0.00016969563077473768,
      "loss": 3.6838,
      "step": 162550
    },
    {
      "epoch": 0.1515322234937168,
      "grad_norm": 0.9754130840301514,
      "learning_rate": 0.0001696937664526425,
      "loss": 3.5895,
      "step": 162560
    },
    {
      "epoch": 0.1515415451117959,
      "grad_norm": 1.0326743125915527,
      "learning_rate": 0.00016969190213054732,
      "loss": 3.5606,
      "step": 162570
    },
    {
      "epoch": 0.151550866729875,
      "grad_norm": 0.9660808444023132,
      "learning_rate": 0.0001696900378084521,
      "loss": 3.4036,
      "step": 162580
    },
    {
      "epoch": 0.1515601883479541,
      "grad_norm": 1.0914753675460815,
      "learning_rate": 0.0001696881734863569,
      "loss": 3.6409,
      "step": 162590
    },
    {
      "epoch": 0.1515695099660332,
      "grad_norm": 1.0737959146499634,
      "learning_rate": 0.00016968630916426168,
      "loss": 3.515,
      "step": 162600
    },
    {
      "epoch": 0.15157883158411228,
      "grad_norm": 1.0406137704849243,
      "learning_rate": 0.0001696844448421665,
      "loss": 3.651,
      "step": 162610
    },
    {
      "epoch": 0.15158815320219138,
      "grad_norm": 1.0416301488876343,
      "learning_rate": 0.00016968258052007132,
      "loss": 3.574,
      "step": 162620
    },
    {
      "epoch": 0.15159747482027047,
      "grad_norm": 1.0416675806045532,
      "learning_rate": 0.0001696807161979761,
      "loss": 3.5786,
      "step": 162630
    },
    {
      "epoch": 0.15160679643834957,
      "grad_norm": 0.9519405961036682,
      "learning_rate": 0.0001696788518758809,
      "loss": 3.4644,
      "step": 162640
    },
    {
      "epoch": 0.15161611805642866,
      "grad_norm": 1.0357434749603271,
      "learning_rate": 0.0001696769875537857,
      "loss": 3.6446,
      "step": 162650
    },
    {
      "epoch": 0.15162543967450773,
      "grad_norm": 1.0684961080551147,
      "learning_rate": 0.0001696751232316905,
      "loss": 3.6371,
      "step": 162660
    },
    {
      "epoch": 0.15163476129258682,
      "grad_norm": 2.4361228942871094,
      "learning_rate": 0.0001696732589095953,
      "loss": 3.4931,
      "step": 162670
    },
    {
      "epoch": 0.15164408291066592,
      "grad_norm": 0.8608137965202332,
      "learning_rate": 0.0001696713945875001,
      "loss": 3.5226,
      "step": 162680
    },
    {
      "epoch": 0.151653404528745,
      "grad_norm": 0.982627809047699,
      "learning_rate": 0.00016966953026540491,
      "loss": 3.5863,
      "step": 162690
    },
    {
      "epoch": 0.1516627261468241,
      "grad_norm": 1.072059988975525,
      "learning_rate": 0.0001696676659433097,
      "loss": 3.5357,
      "step": 162700
    },
    {
      "epoch": 0.1516720477649032,
      "grad_norm": 0.9948623180389404,
      "learning_rate": 0.0001696658016212145,
      "loss": 3.6386,
      "step": 162710
    },
    {
      "epoch": 0.1516813693829823,
      "grad_norm": 0.944118320941925,
      "learning_rate": 0.0001696639372991193,
      "loss": 3.5524,
      "step": 162720
    },
    {
      "epoch": 0.1516906910010614,
      "grad_norm": 1.143372654914856,
      "learning_rate": 0.0001696620729770241,
      "loss": 3.4521,
      "step": 162730
    },
    {
      "epoch": 0.15170001261914048,
      "grad_norm": 0.965133786201477,
      "learning_rate": 0.00016966020865492892,
      "loss": 3.411,
      "step": 162740
    },
    {
      "epoch": 0.15170933423721958,
      "grad_norm": 1.2544289827346802,
      "learning_rate": 0.0001696583443328337,
      "loss": 3.6461,
      "step": 162750
    },
    {
      "epoch": 0.15171865585529865,
      "grad_norm": 0.9690223932266235,
      "learning_rate": 0.0001696564800107385,
      "loss": 3.4682,
      "step": 162760
    },
    {
      "epoch": 0.15172797747337774,
      "grad_norm": 0.9912967681884766,
      "learning_rate": 0.0001696546156886433,
      "loss": 3.47,
      "step": 162770
    },
    {
      "epoch": 0.15173729909145683,
      "grad_norm": 1.2856128215789795,
      "learning_rate": 0.0001696527513665481,
      "loss": 3.3795,
      "step": 162780
    },
    {
      "epoch": 0.15174662070953593,
      "grad_norm": 1.0658901929855347,
      "learning_rate": 0.00016965088704445292,
      "loss": 3.6839,
      "step": 162790
    },
    {
      "epoch": 0.15175594232761502,
      "grad_norm": 1.1465984582901,
      "learning_rate": 0.0001696490227223577,
      "loss": 3.4567,
      "step": 162800
    },
    {
      "epoch": 0.15176526394569412,
      "grad_norm": 1.0798029899597168,
      "learning_rate": 0.0001696471584002625,
      "loss": 3.4848,
      "step": 162810
    },
    {
      "epoch": 0.1517745855637732,
      "grad_norm": 1.0705287456512451,
      "learning_rate": 0.0001696452940781673,
      "loss": 3.547,
      "step": 162820
    },
    {
      "epoch": 0.1517839071818523,
      "grad_norm": 1.2196595668792725,
      "learning_rate": 0.0001696434297560721,
      "loss": 3.6311,
      "step": 162830
    },
    {
      "epoch": 0.1517932287999314,
      "grad_norm": 1.114364743232727,
      "learning_rate": 0.0001696415654339769,
      "loss": 3.5027,
      "step": 162840
    },
    {
      "epoch": 0.1518025504180105,
      "grad_norm": 1.0899099111557007,
      "learning_rate": 0.0001696397011118817,
      "loss": 3.5471,
      "step": 162850
    },
    {
      "epoch": 0.1518118720360896,
      "grad_norm": 0.9135686755180359,
      "learning_rate": 0.0001696378367897865,
      "loss": 3.7847,
      "step": 162860
    },
    {
      "epoch": 0.15182119365416866,
      "grad_norm": 1.0847779512405396,
      "learning_rate": 0.0001696359724676913,
      "loss": 3.6386,
      "step": 162870
    },
    {
      "epoch": 0.15183051527224775,
      "grad_norm": 1.066691517829895,
      "learning_rate": 0.0001696341081455961,
      "loss": 3.5399,
      "step": 162880
    },
    {
      "epoch": 0.15183983689032685,
      "grad_norm": 1.108335256576538,
      "learning_rate": 0.0001696322438235009,
      "loss": 3.6307,
      "step": 162890
    },
    {
      "epoch": 0.15184915850840594,
      "grad_norm": 1.0637542009353638,
      "learning_rate": 0.0001696303795014057,
      "loss": 3.5595,
      "step": 162900
    },
    {
      "epoch": 0.15185848012648503,
      "grad_norm": 1.065366268157959,
      "learning_rate": 0.00016962851517931052,
      "loss": 3.5797,
      "step": 162910
    },
    {
      "epoch": 0.15186780174456413,
      "grad_norm": 1.2591632604599,
      "learning_rate": 0.0001696266508572153,
      "loss": 3.4685,
      "step": 162920
    },
    {
      "epoch": 0.15187712336264322,
      "grad_norm": 1.0215779542922974,
      "learning_rate": 0.0001696247865351201,
      "loss": 3.4147,
      "step": 162930
    },
    {
      "epoch": 0.15188644498072232,
      "grad_norm": 0.9816981554031372,
      "learning_rate": 0.0001696229222130249,
      "loss": 3.5869,
      "step": 162940
    },
    {
      "epoch": 0.1518957665988014,
      "grad_norm": 1.0246812105178833,
      "learning_rate": 0.00016962105789092972,
      "loss": 3.5321,
      "step": 162950
    },
    {
      "epoch": 0.1519050882168805,
      "grad_norm": 1.107904314994812,
      "learning_rate": 0.00016961919356883452,
      "loss": 3.6371,
      "step": 162960
    },
    {
      "epoch": 0.15191440983495957,
      "grad_norm": 0.9521296620368958,
      "learning_rate": 0.0001696173292467393,
      "loss": 3.7019,
      "step": 162970
    },
    {
      "epoch": 0.15192373145303867,
      "grad_norm": 0.9545040130615234,
      "learning_rate": 0.0001696154649246441,
      "loss": 3.6182,
      "step": 162980
    },
    {
      "epoch": 0.15193305307111776,
      "grad_norm": 1.2384546995162964,
      "learning_rate": 0.0001696136006025489,
      "loss": 3.4675,
      "step": 162990
    },
    {
      "epoch": 0.15194237468919686,
      "grad_norm": 1.0947885513305664,
      "learning_rate": 0.0001696117362804537,
      "loss": 3.7308,
      "step": 163000
    },
    {
      "epoch": 0.15195169630727595,
      "grad_norm": 1.067508578300476,
      "learning_rate": 0.00016960987195835853,
      "loss": 3.6238,
      "step": 163010
    },
    {
      "epoch": 0.15196101792535505,
      "grad_norm": 1.1404818296432495,
      "learning_rate": 0.0001696080076362633,
      "loss": 3.3962,
      "step": 163020
    },
    {
      "epoch": 0.15197033954343414,
      "grad_norm": 1.3310949802398682,
      "learning_rate": 0.0001696061433141681,
      "loss": 3.582,
      "step": 163030
    },
    {
      "epoch": 0.15197966116151324,
      "grad_norm": 0.9772700071334839,
      "learning_rate": 0.00016960427899207292,
      "loss": 3.6462,
      "step": 163040
    },
    {
      "epoch": 0.15198898277959233,
      "grad_norm": 1.3273677825927734,
      "learning_rate": 0.0001696024146699777,
      "loss": 3.5819,
      "step": 163050
    },
    {
      "epoch": 0.15199830439767142,
      "grad_norm": 0.9684823751449585,
      "learning_rate": 0.0001696005503478825,
      "loss": 3.4157,
      "step": 163060
    },
    {
      "epoch": 0.15200762601575052,
      "grad_norm": 0.9957492351531982,
      "learning_rate": 0.0001695986860257873,
      "loss": 3.627,
      "step": 163070
    },
    {
      "epoch": 0.15201694763382959,
      "grad_norm": 1.1374106407165527,
      "learning_rate": 0.00016959682170369212,
      "loss": 3.4667,
      "step": 163080
    },
    {
      "epoch": 0.15202626925190868,
      "grad_norm": 0.9518186450004578,
      "learning_rate": 0.00016959495738159692,
      "loss": 3.5956,
      "step": 163090
    },
    {
      "epoch": 0.15203559086998777,
      "grad_norm": 1.0688289403915405,
      "learning_rate": 0.0001695930930595017,
      "loss": 3.5775,
      "step": 163100
    },
    {
      "epoch": 0.15204491248806687,
      "grad_norm": 1.0547599792480469,
      "learning_rate": 0.0001695912287374065,
      "loss": 3.5697,
      "step": 163110
    },
    {
      "epoch": 0.15205423410614596,
      "grad_norm": 1.0178886651992798,
      "learning_rate": 0.00016958936441531131,
      "loss": 3.4507,
      "step": 163120
    },
    {
      "epoch": 0.15206355572422506,
      "grad_norm": 1.1993080377578735,
      "learning_rate": 0.00016958750009321612,
      "loss": 3.521,
      "step": 163130
    },
    {
      "epoch": 0.15207287734230415,
      "grad_norm": 1.1069806814193726,
      "learning_rate": 0.00016958563577112093,
      "loss": 3.5743,
      "step": 163140
    },
    {
      "epoch": 0.15208219896038325,
      "grad_norm": 1.1880466938018799,
      "learning_rate": 0.0001695837714490257,
      "loss": 3.7154,
      "step": 163150
    },
    {
      "epoch": 0.15209152057846234,
      "grad_norm": 1.1747839450836182,
      "learning_rate": 0.0001695819071269305,
      "loss": 3.5145,
      "step": 163160
    },
    {
      "epoch": 0.15210084219654144,
      "grad_norm": 1.026261568069458,
      "learning_rate": 0.0001695800428048353,
      "loss": 3.5133,
      "step": 163170
    },
    {
      "epoch": 0.1521101638146205,
      "grad_norm": 1.4932568073272705,
      "learning_rate": 0.00016957817848274012,
      "loss": 3.3926,
      "step": 163180
    },
    {
      "epoch": 0.1521194854326996,
      "grad_norm": 1.110474944114685,
      "learning_rate": 0.00016957631416064493,
      "loss": 3.6281,
      "step": 163190
    },
    {
      "epoch": 0.1521288070507787,
      "grad_norm": 0.9962716102600098,
      "learning_rate": 0.0001695744498385497,
      "loss": 3.7238,
      "step": 163200
    },
    {
      "epoch": 0.15213812866885779,
      "grad_norm": 1.0781536102294922,
      "learning_rate": 0.00016957258551645452,
      "loss": 3.5804,
      "step": 163210
    },
    {
      "epoch": 0.15214745028693688,
      "grad_norm": 1.1644619703292847,
      "learning_rate": 0.0001695707211943593,
      "loss": 3.5535,
      "step": 163220
    },
    {
      "epoch": 0.15215677190501598,
      "grad_norm": 1.1332428455352783,
      "learning_rate": 0.0001695688568722641,
      "loss": 3.6809,
      "step": 163230
    },
    {
      "epoch": 0.15216609352309507,
      "grad_norm": 1.3280004262924194,
      "learning_rate": 0.00016956699255016894,
      "loss": 3.5811,
      "step": 163240
    },
    {
      "epoch": 0.15217541514117416,
      "grad_norm": 1.4369491338729858,
      "learning_rate": 0.00016956512822807371,
      "loss": 3.7673,
      "step": 163250
    },
    {
      "epoch": 0.15218473675925326,
      "grad_norm": 1.2023943662643433,
      "learning_rate": 0.00016956326390597852,
      "loss": 3.4354,
      "step": 163260
    },
    {
      "epoch": 0.15219405837733235,
      "grad_norm": 1.0268067121505737,
      "learning_rate": 0.0001695613995838833,
      "loss": 3.4092,
      "step": 163270
    },
    {
      "epoch": 0.15220337999541142,
      "grad_norm": 1.2469233274459839,
      "learning_rate": 0.0001695595352617881,
      "loss": 3.6874,
      "step": 163280
    },
    {
      "epoch": 0.15221270161349051,
      "grad_norm": 1.117238163948059,
      "learning_rate": 0.0001695576709396929,
      "loss": 3.5931,
      "step": 163290
    },
    {
      "epoch": 0.1522220232315696,
      "grad_norm": 0.9442504048347473,
      "learning_rate": 0.00016955580661759772,
      "loss": 3.5835,
      "step": 163300
    },
    {
      "epoch": 0.1522313448496487,
      "grad_norm": 1.154772162437439,
      "learning_rate": 0.00016955394229550253,
      "loss": 3.683,
      "step": 163310
    },
    {
      "epoch": 0.1522406664677278,
      "grad_norm": 1.422876238822937,
      "learning_rate": 0.0001695520779734073,
      "loss": 3.6869,
      "step": 163320
    },
    {
      "epoch": 0.1522499880858069,
      "grad_norm": 1.064781904220581,
      "learning_rate": 0.0001695502136513121,
      "loss": 3.351,
      "step": 163330
    },
    {
      "epoch": 0.152259309703886,
      "grad_norm": 0.9799339771270752,
      "learning_rate": 0.00016954834932921692,
      "loss": 3.5031,
      "step": 163340
    },
    {
      "epoch": 0.15226863132196508,
      "grad_norm": 0.9984162449836731,
      "learning_rate": 0.00016954648500712172,
      "loss": 3.7085,
      "step": 163350
    },
    {
      "epoch": 0.15227795294004418,
      "grad_norm": 1.039784550666809,
      "learning_rate": 0.00016954462068502653,
      "loss": 3.6373,
      "step": 163360
    },
    {
      "epoch": 0.15228727455812327,
      "grad_norm": 1.0290240049362183,
      "learning_rate": 0.0001695427563629313,
      "loss": 3.3917,
      "step": 163370
    },
    {
      "epoch": 0.15229659617620236,
      "grad_norm": 1.0629733800888062,
      "learning_rate": 0.00016954089204083612,
      "loss": 3.6314,
      "step": 163380
    },
    {
      "epoch": 0.15230591779428143,
      "grad_norm": 1.0399487018585205,
      "learning_rate": 0.00016953902771874092,
      "loss": 3.6825,
      "step": 163390
    },
    {
      "epoch": 0.15231523941236053,
      "grad_norm": 1.1044726371765137,
      "learning_rate": 0.00016953716339664573,
      "loss": 3.4759,
      "step": 163400
    },
    {
      "epoch": 0.15232456103043962,
      "grad_norm": 0.9777655601501465,
      "learning_rate": 0.00016953529907455053,
      "loss": 3.6545,
      "step": 163410
    },
    {
      "epoch": 0.15233388264851871,
      "grad_norm": 1.0806714296340942,
      "learning_rate": 0.0001695334347524553,
      "loss": 3.5946,
      "step": 163420
    },
    {
      "epoch": 0.1523432042665978,
      "grad_norm": 0.904999315738678,
      "learning_rate": 0.00016953157043036012,
      "loss": 3.541,
      "step": 163430
    },
    {
      "epoch": 0.1523525258846769,
      "grad_norm": 0.9720255136489868,
      "learning_rate": 0.00016952970610826493,
      "loss": 3.402,
      "step": 163440
    },
    {
      "epoch": 0.152361847502756,
      "grad_norm": 1.0744670629501343,
      "learning_rate": 0.0001695278417861697,
      "loss": 3.6689,
      "step": 163450
    },
    {
      "epoch": 0.1523711691208351,
      "grad_norm": 1.0571439266204834,
      "learning_rate": 0.00016952597746407454,
      "loss": 3.665,
      "step": 163460
    },
    {
      "epoch": 0.1523804907389142,
      "grad_norm": 1.2398617267608643,
      "learning_rate": 0.00016952411314197932,
      "loss": 3.6062,
      "step": 163470
    },
    {
      "epoch": 0.15238981235699328,
      "grad_norm": 1.113310694694519,
      "learning_rate": 0.00016952224881988412,
      "loss": 3.6268,
      "step": 163480
    },
    {
      "epoch": 0.15239913397507235,
      "grad_norm": 1.1644489765167236,
      "learning_rate": 0.00016952038449778893,
      "loss": 3.4596,
      "step": 163490
    },
    {
      "epoch": 0.15240845559315144,
      "grad_norm": 1.3223578929901123,
      "learning_rate": 0.0001695185201756937,
      "loss": 3.689,
      "step": 163500
    },
    {
      "epoch": 0.15241777721123054,
      "grad_norm": 1.1392582654953003,
      "learning_rate": 0.00016951665585359852,
      "loss": 3.5496,
      "step": 163510
    },
    {
      "epoch": 0.15242709882930963,
      "grad_norm": 1.0321433544158936,
      "learning_rate": 0.00016951479153150332,
      "loss": 3.513,
      "step": 163520
    },
    {
      "epoch": 0.15243642044738873,
      "grad_norm": 1.1266100406646729,
      "learning_rate": 0.00016951292720940813,
      "loss": 3.5924,
      "step": 163530
    },
    {
      "epoch": 0.15244574206546782,
      "grad_norm": 1.2138229608535767,
      "learning_rate": 0.00016951106288731294,
      "loss": 3.349,
      "step": 163540
    },
    {
      "epoch": 0.15245506368354692,
      "grad_norm": 0.9865671992301941,
      "learning_rate": 0.00016950919856521771,
      "loss": 3.3853,
      "step": 163550
    },
    {
      "epoch": 0.152464385301626,
      "grad_norm": 1.162022352218628,
      "learning_rate": 0.00016950733424312252,
      "loss": 3.6084,
      "step": 163560
    },
    {
      "epoch": 0.1524737069197051,
      "grad_norm": 0.9445443153381348,
      "learning_rate": 0.00016950546992102733,
      "loss": 3.5346,
      "step": 163570
    },
    {
      "epoch": 0.1524830285377842,
      "grad_norm": 1.2414435148239136,
      "learning_rate": 0.00016950360559893213,
      "loss": 3.5155,
      "step": 163580
    },
    {
      "epoch": 0.1524923501558633,
      "grad_norm": 1.208118200302124,
      "learning_rate": 0.00016950174127683694,
      "loss": 3.3242,
      "step": 163590
    },
    {
      "epoch": 0.15250167177394236,
      "grad_norm": 1.1531468629837036,
      "learning_rate": 0.00016949987695474172,
      "loss": 3.3466,
      "step": 163600
    },
    {
      "epoch": 0.15251099339202145,
      "grad_norm": 0.8483718633651733,
      "learning_rate": 0.00016949801263264652,
      "loss": 3.6757,
      "step": 163610
    },
    {
      "epoch": 0.15252031501010055,
      "grad_norm": 1.1718777418136597,
      "learning_rate": 0.0001694961483105513,
      "loss": 3.568,
      "step": 163620
    },
    {
      "epoch": 0.15252963662817964,
      "grad_norm": 1.5133757591247559,
      "learning_rate": 0.00016949428398845614,
      "loss": 3.5476,
      "step": 163630
    },
    {
      "epoch": 0.15253895824625874,
      "grad_norm": 1.2490229606628418,
      "learning_rate": 0.00016949241966636094,
      "loss": 3.5111,
      "step": 163640
    },
    {
      "epoch": 0.15254827986433783,
      "grad_norm": 0.8065009117126465,
      "learning_rate": 0.00016949055534426572,
      "loss": 3.6903,
      "step": 163650
    },
    {
      "epoch": 0.15255760148241693,
      "grad_norm": 0.9558135271072388,
      "learning_rate": 0.00016948869102217053,
      "loss": 3.5446,
      "step": 163660
    },
    {
      "epoch": 0.15256692310049602,
      "grad_norm": 1.0638538599014282,
      "learning_rate": 0.0001694868267000753,
      "loss": 3.5367,
      "step": 163670
    },
    {
      "epoch": 0.15257624471857512,
      "grad_norm": 1.1095505952835083,
      "learning_rate": 0.00016948496237798011,
      "loss": 3.5148,
      "step": 163680
    },
    {
      "epoch": 0.1525855663366542,
      "grad_norm": 0.8700540065765381,
      "learning_rate": 0.00016948309805588495,
      "loss": 3.7594,
      "step": 163690
    },
    {
      "epoch": 0.15259488795473328,
      "grad_norm": 0.9456725716590881,
      "learning_rate": 0.00016948123373378973,
      "loss": 3.5735,
      "step": 163700
    },
    {
      "epoch": 0.15260420957281237,
      "grad_norm": 1.087522029876709,
      "learning_rate": 0.00016947936941169453,
      "loss": 3.5497,
      "step": 163710
    },
    {
      "epoch": 0.15261353119089147,
      "grad_norm": 1.004832148551941,
      "learning_rate": 0.0001694775050895993,
      "loss": 3.56,
      "step": 163720
    },
    {
      "epoch": 0.15262285280897056,
      "grad_norm": 1.2722177505493164,
      "learning_rate": 0.00016947564076750412,
      "loss": 3.544,
      "step": 163730
    },
    {
      "epoch": 0.15263217442704965,
      "grad_norm": 1.1344316005706787,
      "learning_rate": 0.00016947377644540893,
      "loss": 3.7341,
      "step": 163740
    },
    {
      "epoch": 0.15264149604512875,
      "grad_norm": 1.1341360807418823,
      "learning_rate": 0.00016947191212331373,
      "loss": 3.4643,
      "step": 163750
    },
    {
      "epoch": 0.15265081766320784,
      "grad_norm": 1.052386999130249,
      "learning_rate": 0.00016947004780121854,
      "loss": 3.6379,
      "step": 163760
    },
    {
      "epoch": 0.15266013928128694,
      "grad_norm": 0.9635305404663086,
      "learning_rate": 0.00016946818347912332,
      "loss": 3.5574,
      "step": 163770
    },
    {
      "epoch": 0.15266946089936603,
      "grad_norm": 0.8637039661407471,
      "learning_rate": 0.00016946631915702812,
      "loss": 3.5929,
      "step": 163780
    },
    {
      "epoch": 0.15267878251744513,
      "grad_norm": 0.8519114851951599,
      "learning_rate": 0.00016946445483493293,
      "loss": 3.6358,
      "step": 163790
    },
    {
      "epoch": 0.15268810413552422,
      "grad_norm": 1.0387808084487915,
      "learning_rate": 0.00016946259051283774,
      "loss": 3.5776,
      "step": 163800
    },
    {
      "epoch": 0.1526974257536033,
      "grad_norm": 1.1131799221038818,
      "learning_rate": 0.00016946072619074254,
      "loss": 3.6122,
      "step": 163810
    },
    {
      "epoch": 0.15270674737168238,
      "grad_norm": 1.020887017250061,
      "learning_rate": 0.00016945886186864732,
      "loss": 3.9024,
      "step": 163820
    },
    {
      "epoch": 0.15271606898976148,
      "grad_norm": 0.9797683358192444,
      "learning_rate": 0.00016945699754655213,
      "loss": 3.3779,
      "step": 163830
    },
    {
      "epoch": 0.15272539060784057,
      "grad_norm": 1.203139305114746,
      "learning_rate": 0.00016945513322445693,
      "loss": 3.6388,
      "step": 163840
    },
    {
      "epoch": 0.15273471222591967,
      "grad_norm": 0.8388554453849792,
      "learning_rate": 0.00016945326890236174,
      "loss": 3.5475,
      "step": 163850
    },
    {
      "epoch": 0.15274403384399876,
      "grad_norm": 1.122469425201416,
      "learning_rate": 0.00016945140458026655,
      "loss": 3.6422,
      "step": 163860
    },
    {
      "epoch": 0.15275335546207786,
      "grad_norm": 1.0588266849517822,
      "learning_rate": 0.00016944954025817133,
      "loss": 3.7395,
      "step": 163870
    },
    {
      "epoch": 0.15276267708015695,
      "grad_norm": 1.2696802616119385,
      "learning_rate": 0.00016944767593607613,
      "loss": 3.5265,
      "step": 163880
    },
    {
      "epoch": 0.15277199869823604,
      "grad_norm": 1.0558722019195557,
      "learning_rate": 0.00016944581161398094,
      "loss": 3.6112,
      "step": 163890
    },
    {
      "epoch": 0.15278132031631514,
      "grad_norm": 1.0139423608779907,
      "learning_rate": 0.00016944394729188572,
      "loss": 3.5291,
      "step": 163900
    },
    {
      "epoch": 0.1527906419343942,
      "grad_norm": 1.074783205986023,
      "learning_rate": 0.00016944208296979055,
      "loss": 3.5699,
      "step": 163910
    },
    {
      "epoch": 0.1527999635524733,
      "grad_norm": 1.0179696083068848,
      "learning_rate": 0.00016944021864769533,
      "loss": 3.7416,
      "step": 163920
    },
    {
      "epoch": 0.1528092851705524,
      "grad_norm": 1.0530341863632202,
      "learning_rate": 0.00016943835432560014,
      "loss": 3.6385,
      "step": 163930
    },
    {
      "epoch": 0.1528186067886315,
      "grad_norm": 1.0920813083648682,
      "learning_rate": 0.00016943649000350494,
      "loss": 3.5469,
      "step": 163940
    },
    {
      "epoch": 0.15282792840671058,
      "grad_norm": 1.2054592370986938,
      "learning_rate": 0.00016943462568140972,
      "loss": 3.3983,
      "step": 163950
    },
    {
      "epoch": 0.15283725002478968,
      "grad_norm": 1.0325238704681396,
      "learning_rate": 0.00016943276135931453,
      "loss": 3.6551,
      "step": 163960
    },
    {
      "epoch": 0.15284657164286877,
      "grad_norm": 1.111670970916748,
      "learning_rate": 0.00016943089703721934,
      "loss": 3.3616,
      "step": 163970
    },
    {
      "epoch": 0.15285589326094787,
      "grad_norm": 1.1169732809066772,
      "learning_rate": 0.00016942903271512414,
      "loss": 3.6327,
      "step": 163980
    },
    {
      "epoch": 0.15286521487902696,
      "grad_norm": 0.9907748103141785,
      "learning_rate": 0.00016942716839302895,
      "loss": 3.4932,
      "step": 163990
    },
    {
      "epoch": 0.15287453649710606,
      "grad_norm": 1.1148351430892944,
      "learning_rate": 0.00016942530407093373,
      "loss": 3.5697,
      "step": 164000
    },
    {
      "epoch": 0.15288385811518515,
      "grad_norm": 1.1347824335098267,
      "learning_rate": 0.00016942343974883853,
      "loss": 3.7204,
      "step": 164010
    },
    {
      "epoch": 0.15289317973326422,
      "grad_norm": 1.017690658569336,
      "learning_rate": 0.00016942157542674334,
      "loss": 3.5551,
      "step": 164020
    },
    {
      "epoch": 0.1529025013513433,
      "grad_norm": 0.9500486850738525,
      "learning_rate": 0.00016941971110464815,
      "loss": 3.397,
      "step": 164030
    },
    {
      "epoch": 0.1529118229694224,
      "grad_norm": 1.0516058206558228,
      "learning_rate": 0.00016941784678255295,
      "loss": 3.4097,
      "step": 164040
    },
    {
      "epoch": 0.1529211445875015,
      "grad_norm": 0.9835485816001892,
      "learning_rate": 0.00016941598246045773,
      "loss": 3.6088,
      "step": 164050
    },
    {
      "epoch": 0.1529304662055806,
      "grad_norm": 0.9077078104019165,
      "learning_rate": 0.00016941411813836254,
      "loss": 3.5673,
      "step": 164060
    },
    {
      "epoch": 0.1529397878236597,
      "grad_norm": 0.8204712271690369,
      "learning_rate": 0.00016941225381626732,
      "loss": 3.587,
      "step": 164070
    },
    {
      "epoch": 0.15294910944173878,
      "grad_norm": 1.1931484937667847,
      "learning_rate": 0.00016941038949417215,
      "loss": 3.8319,
      "step": 164080
    },
    {
      "epoch": 0.15295843105981788,
      "grad_norm": 1.1946784257888794,
      "learning_rate": 0.00016940852517207696,
      "loss": 3.5066,
      "step": 164090
    },
    {
      "epoch": 0.15296775267789697,
      "grad_norm": 1.041922688484192,
      "learning_rate": 0.00016940666084998174,
      "loss": 3.5626,
      "step": 164100
    },
    {
      "epoch": 0.15297707429597607,
      "grad_norm": 0.9830440282821655,
      "learning_rate": 0.00016940479652788654,
      "loss": 3.5411,
      "step": 164110
    },
    {
      "epoch": 0.15298639591405513,
      "grad_norm": 1.1841920614242554,
      "learning_rate": 0.00016940293220579132,
      "loss": 3.4669,
      "step": 164120
    },
    {
      "epoch": 0.15299571753213423,
      "grad_norm": 1.1115840673446655,
      "learning_rate": 0.00016940106788369613,
      "loss": 3.5748,
      "step": 164130
    },
    {
      "epoch": 0.15300503915021332,
      "grad_norm": 1.1193100214004517,
      "learning_rate": 0.00016939920356160096,
      "loss": 3.5861,
      "step": 164140
    },
    {
      "epoch": 0.15301436076829242,
      "grad_norm": 0.9570337533950806,
      "learning_rate": 0.00016939733923950574,
      "loss": 3.4615,
      "step": 164150
    },
    {
      "epoch": 0.1530236823863715,
      "grad_norm": 0.9643961787223816,
      "learning_rate": 0.00016939547491741055,
      "loss": 3.4949,
      "step": 164160
    },
    {
      "epoch": 0.1530330040044506,
      "grad_norm": 1.0867384672164917,
      "learning_rate": 0.00016939361059531533,
      "loss": 3.6583,
      "step": 164170
    },
    {
      "epoch": 0.1530423256225297,
      "grad_norm": 1.161342740058899,
      "learning_rate": 0.00016939174627322013,
      "loss": 3.6346,
      "step": 164180
    },
    {
      "epoch": 0.1530516472406088,
      "grad_norm": 1.1615753173828125,
      "learning_rate": 0.00016938988195112494,
      "loss": 3.4981,
      "step": 164190
    },
    {
      "epoch": 0.1530609688586879,
      "grad_norm": 0.7988343834877014,
      "learning_rate": 0.00016938801762902974,
      "loss": 3.7109,
      "step": 164200
    },
    {
      "epoch": 0.15307029047676698,
      "grad_norm": 1.3211549520492554,
      "learning_rate": 0.00016938615330693455,
      "loss": 3.6152,
      "step": 164210
    },
    {
      "epoch": 0.15307961209484608,
      "grad_norm": 1.323609709739685,
      "learning_rate": 0.00016938428898483933,
      "loss": 3.6885,
      "step": 164220
    },
    {
      "epoch": 0.15308893371292515,
      "grad_norm": 1.020005226135254,
      "learning_rate": 0.00016938242466274414,
      "loss": 3.5947,
      "step": 164230
    },
    {
      "epoch": 0.15309825533100424,
      "grad_norm": 1.0659207105636597,
      "learning_rate": 0.00016938056034064894,
      "loss": 3.5841,
      "step": 164240
    },
    {
      "epoch": 0.15310757694908333,
      "grad_norm": 1.0034948587417603,
      "learning_rate": 0.00016937869601855375,
      "loss": 3.7868,
      "step": 164250
    },
    {
      "epoch": 0.15311689856716243,
      "grad_norm": 1.0144836902618408,
      "learning_rate": 0.00016937683169645856,
      "loss": 3.6491,
      "step": 164260
    },
    {
      "epoch": 0.15312622018524152,
      "grad_norm": 1.123228669166565,
      "learning_rate": 0.00016937496737436333,
      "loss": 3.6502,
      "step": 164270
    },
    {
      "epoch": 0.15313554180332062,
      "grad_norm": 1.051798701286316,
      "learning_rate": 0.00016937310305226814,
      "loss": 3.7501,
      "step": 164280
    },
    {
      "epoch": 0.1531448634213997,
      "grad_norm": 0.9506922364234924,
      "learning_rate": 0.00016937123873017292,
      "loss": 3.4366,
      "step": 164290
    },
    {
      "epoch": 0.1531541850394788,
      "grad_norm": 1.0916208028793335,
      "learning_rate": 0.00016936937440807775,
      "loss": 3.5477,
      "step": 164300
    },
    {
      "epoch": 0.1531635066575579,
      "grad_norm": 1.0495104789733887,
      "learning_rate": 0.00016936751008598256,
      "loss": 3.6592,
      "step": 164310
    },
    {
      "epoch": 0.153172828275637,
      "grad_norm": 0.9994179010391235,
      "learning_rate": 0.00016936564576388734,
      "loss": 3.4304,
      "step": 164320
    },
    {
      "epoch": 0.15318214989371606,
      "grad_norm": 1.0084251165390015,
      "learning_rate": 0.00016936378144179215,
      "loss": 3.3755,
      "step": 164330
    },
    {
      "epoch": 0.15319147151179516,
      "grad_norm": 1.0683555603027344,
      "learning_rate": 0.00016936191711969692,
      "loss": 3.5705,
      "step": 164340
    },
    {
      "epoch": 0.15320079312987425,
      "grad_norm": 1.0120022296905518,
      "learning_rate": 0.00016936005279760173,
      "loss": 3.5815,
      "step": 164350
    },
    {
      "epoch": 0.15321011474795335,
      "grad_norm": 0.943570613861084,
      "learning_rate": 0.00016935818847550656,
      "loss": 3.7117,
      "step": 164360
    },
    {
      "epoch": 0.15321943636603244,
      "grad_norm": 1.0232412815093994,
      "learning_rate": 0.00016935632415341134,
      "loss": 3.5396,
      "step": 164370
    },
    {
      "epoch": 0.15322875798411154,
      "grad_norm": 1.0545654296875,
      "learning_rate": 0.00016935445983131615,
      "loss": 3.7486,
      "step": 164380
    },
    {
      "epoch": 0.15323807960219063,
      "grad_norm": 0.9516118764877319,
      "learning_rate": 0.00016935259550922093,
      "loss": 3.4623,
      "step": 164390
    },
    {
      "epoch": 0.15324740122026972,
      "grad_norm": 0.9458606243133545,
      "learning_rate": 0.00016935073118712574,
      "loss": 3.6064,
      "step": 164400
    },
    {
      "epoch": 0.15325672283834882,
      "grad_norm": 1.0147455930709839,
      "learning_rate": 0.00016934886686503054,
      "loss": 3.4689,
      "step": 164410
    },
    {
      "epoch": 0.1532660444564279,
      "grad_norm": 1.230864405632019,
      "learning_rate": 0.00016934700254293535,
      "loss": 3.44,
      "step": 164420
    },
    {
      "epoch": 0.153275366074507,
      "grad_norm": 1.2912039756774902,
      "learning_rate": 0.00016934513822084015,
      "loss": 3.5714,
      "step": 164430
    },
    {
      "epoch": 0.15328468769258607,
      "grad_norm": 0.9892342686653137,
      "learning_rate": 0.00016934327389874493,
      "loss": 3.5892,
      "step": 164440
    },
    {
      "epoch": 0.15329400931066517,
      "grad_norm": 0.9747449159622192,
      "learning_rate": 0.00016934140957664974,
      "loss": 3.4608,
      "step": 164450
    },
    {
      "epoch": 0.15330333092874426,
      "grad_norm": 1.1600521802902222,
      "learning_rate": 0.00016933954525455455,
      "loss": 3.6829,
      "step": 164460
    },
    {
      "epoch": 0.15331265254682336,
      "grad_norm": 0.9978689551353455,
      "learning_rate": 0.00016933768093245935,
      "loss": 3.5831,
      "step": 164470
    },
    {
      "epoch": 0.15332197416490245,
      "grad_norm": 1.1632122993469238,
      "learning_rate": 0.00016933581661036416,
      "loss": 3.5327,
      "step": 164480
    },
    {
      "epoch": 0.15333129578298155,
      "grad_norm": 1.1271965503692627,
      "learning_rate": 0.00016933395228826894,
      "loss": 3.4987,
      "step": 164490
    },
    {
      "epoch": 0.15334061740106064,
      "grad_norm": 0.9472037553787231,
      "learning_rate": 0.00016933208796617374,
      "loss": 3.5205,
      "step": 164500
    },
    {
      "epoch": 0.15334993901913974,
      "grad_norm": 1.1333045959472656,
      "learning_rate": 0.00016933022364407855,
      "loss": 3.6665,
      "step": 164510
    },
    {
      "epoch": 0.15335926063721883,
      "grad_norm": 1.3342249393463135,
      "learning_rate": 0.00016932835932198333,
      "loss": 3.7163,
      "step": 164520
    },
    {
      "epoch": 0.15336858225529792,
      "grad_norm": 1.2579971551895142,
      "learning_rate": 0.00016932649499988816,
      "loss": 3.5289,
      "step": 164530
    },
    {
      "epoch": 0.153377903873377,
      "grad_norm": 0.9780081510543823,
      "learning_rate": 0.00016932463067779294,
      "loss": 3.7638,
      "step": 164540
    },
    {
      "epoch": 0.1533872254914561,
      "grad_norm": 1.161008358001709,
      "learning_rate": 0.00016932276635569775,
      "loss": 3.6367,
      "step": 164550
    },
    {
      "epoch": 0.15339654710953518,
      "grad_norm": 1.049293875694275,
      "learning_rate": 0.00016932090203360256,
      "loss": 3.688,
      "step": 164560
    },
    {
      "epoch": 0.15340586872761428,
      "grad_norm": 1.0248833894729614,
      "learning_rate": 0.00016931903771150733,
      "loss": 3.4679,
      "step": 164570
    },
    {
      "epoch": 0.15341519034569337,
      "grad_norm": 1.0080010890960693,
      "learning_rate": 0.00016931717338941214,
      "loss": 3.6684,
      "step": 164580
    },
    {
      "epoch": 0.15342451196377246,
      "grad_norm": 1.2438863515853882,
      "learning_rate": 0.00016931530906731695,
      "loss": 3.4122,
      "step": 164590
    },
    {
      "epoch": 0.15343383358185156,
      "grad_norm": 1.0387935638427734,
      "learning_rate": 0.00016931344474522175,
      "loss": 3.6222,
      "step": 164600
    },
    {
      "epoch": 0.15344315519993065,
      "grad_norm": 1.3273874521255493,
      "learning_rate": 0.00016931158042312656,
      "loss": 3.5799,
      "step": 164610
    },
    {
      "epoch": 0.15345247681800975,
      "grad_norm": 1.2518571615219116,
      "learning_rate": 0.00016930971610103134,
      "loss": 3.6206,
      "step": 164620
    },
    {
      "epoch": 0.15346179843608884,
      "grad_norm": 1.0210148096084595,
      "learning_rate": 0.00016930785177893615,
      "loss": 3.5946,
      "step": 164630
    },
    {
      "epoch": 0.1534711200541679,
      "grad_norm": 1.3167893886566162,
      "learning_rate": 0.00016930598745684095,
      "loss": 3.6269,
      "step": 164640
    },
    {
      "epoch": 0.153480441672247,
      "grad_norm": 1.0693602561950684,
      "learning_rate": 0.00016930412313474576,
      "loss": 3.7059,
      "step": 164650
    },
    {
      "epoch": 0.1534897632903261,
      "grad_norm": 0.9184890389442444,
      "learning_rate": 0.00016930225881265056,
      "loss": 3.606,
      "step": 164660
    },
    {
      "epoch": 0.1534990849084052,
      "grad_norm": 1.2642122507095337,
      "learning_rate": 0.00016930039449055534,
      "loss": 3.632,
      "step": 164670
    },
    {
      "epoch": 0.1535084065264843,
      "grad_norm": 0.9691552519798279,
      "learning_rate": 0.00016929853016846015,
      "loss": 3.7194,
      "step": 164680
    },
    {
      "epoch": 0.15351772814456338,
      "grad_norm": 1.0893280506134033,
      "learning_rate": 0.00016929666584636496,
      "loss": 3.6262,
      "step": 164690
    },
    {
      "epoch": 0.15352704976264248,
      "grad_norm": 1.0227007865905762,
      "learning_rate": 0.00016929480152426976,
      "loss": 3.686,
      "step": 164700
    },
    {
      "epoch": 0.15353637138072157,
      "grad_norm": 1.000694990158081,
      "learning_rate": 0.00016929293720217457,
      "loss": 3.582,
      "step": 164710
    },
    {
      "epoch": 0.15354569299880066,
      "grad_norm": 1.0276196002960205,
      "learning_rate": 0.00016929107288007935,
      "loss": 3.5985,
      "step": 164720
    },
    {
      "epoch": 0.15355501461687976,
      "grad_norm": 0.9473147392272949,
      "learning_rate": 0.00016928920855798415,
      "loss": 3.4802,
      "step": 164730
    },
    {
      "epoch": 0.15356433623495885,
      "grad_norm": 1.1018095016479492,
      "learning_rate": 0.00016928734423588893,
      "loss": 3.5829,
      "step": 164740
    },
    {
      "epoch": 0.15357365785303792,
      "grad_norm": 1.0208107233047485,
      "learning_rate": 0.00016928547991379377,
      "loss": 3.6993,
      "step": 164750
    },
    {
      "epoch": 0.15358297947111701,
      "grad_norm": 1.0728960037231445,
      "learning_rate": 0.00016928361559169857,
      "loss": 3.5911,
      "step": 164760
    },
    {
      "epoch": 0.1535923010891961,
      "grad_norm": 0.9784166216850281,
      "learning_rate": 0.00016928175126960335,
      "loss": 3.526,
      "step": 164770
    },
    {
      "epoch": 0.1536016227072752,
      "grad_norm": 1.0387370586395264,
      "learning_rate": 0.00016927988694750816,
      "loss": 3.4514,
      "step": 164780
    },
    {
      "epoch": 0.1536109443253543,
      "grad_norm": 1.1136963367462158,
      "learning_rate": 0.00016927802262541294,
      "loss": 3.5082,
      "step": 164790
    },
    {
      "epoch": 0.1536202659434334,
      "grad_norm": 1.2553945779800415,
      "learning_rate": 0.00016927615830331774,
      "loss": 3.7157,
      "step": 164800
    },
    {
      "epoch": 0.1536295875615125,
      "grad_norm": 0.9416589736938477,
      "learning_rate": 0.00016927429398122258,
      "loss": 3.617,
      "step": 164810
    },
    {
      "epoch": 0.15363890917959158,
      "grad_norm": 1.0876586437225342,
      "learning_rate": 0.00016927242965912736,
      "loss": 3.6414,
      "step": 164820
    },
    {
      "epoch": 0.15364823079767068,
      "grad_norm": 0.9522705674171448,
      "learning_rate": 0.00016927056533703216,
      "loss": 3.6562,
      "step": 164830
    },
    {
      "epoch": 0.15365755241574977,
      "grad_norm": 1.2092442512512207,
      "learning_rate": 0.00016926870101493694,
      "loss": 3.5357,
      "step": 164840
    },
    {
      "epoch": 0.15366687403382884,
      "grad_norm": 1.1199679374694824,
      "learning_rate": 0.00016926683669284175,
      "loss": 3.6283,
      "step": 164850
    },
    {
      "epoch": 0.15367619565190793,
      "grad_norm": 0.9243615865707397,
      "learning_rate": 0.00016926497237074655,
      "loss": 3.613,
      "step": 164860
    },
    {
      "epoch": 0.15368551726998703,
      "grad_norm": 1.0841158628463745,
      "learning_rate": 0.00016926310804865136,
      "loss": 3.5754,
      "step": 164870
    },
    {
      "epoch": 0.15369483888806612,
      "grad_norm": 1.0733402967453003,
      "learning_rate": 0.00016926124372655617,
      "loss": 3.6297,
      "step": 164880
    },
    {
      "epoch": 0.15370416050614522,
      "grad_norm": 1.038836121559143,
      "learning_rate": 0.00016925937940446095,
      "loss": 3.5824,
      "step": 164890
    },
    {
      "epoch": 0.1537134821242243,
      "grad_norm": 0.9448086023330688,
      "learning_rate": 0.00016925751508236575,
      "loss": 3.6898,
      "step": 164900
    },
    {
      "epoch": 0.1537228037423034,
      "grad_norm": 1.0173267126083374,
      "learning_rate": 0.00016925565076027056,
      "loss": 3.5034,
      "step": 164910
    },
    {
      "epoch": 0.1537321253603825,
      "grad_norm": 1.0267633199691772,
      "learning_rate": 0.00016925378643817537,
      "loss": 3.3801,
      "step": 164920
    },
    {
      "epoch": 0.1537414469784616,
      "grad_norm": 1.031259536743164,
      "learning_rate": 0.00016925192211608017,
      "loss": 3.5437,
      "step": 164930
    },
    {
      "epoch": 0.1537507685965407,
      "grad_norm": 0.9841445684432983,
      "learning_rate": 0.00016925005779398495,
      "loss": 3.5653,
      "step": 164940
    },
    {
      "epoch": 0.15376009021461978,
      "grad_norm": 1.0512027740478516,
      "learning_rate": 0.00016924819347188976,
      "loss": 3.5415,
      "step": 164950
    },
    {
      "epoch": 0.15376941183269885,
      "grad_norm": 1.084637999534607,
      "learning_rate": 0.00016924632914979456,
      "loss": 3.3378,
      "step": 164960
    },
    {
      "epoch": 0.15377873345077794,
      "grad_norm": 0.8260011672973633,
      "learning_rate": 0.00016924446482769934,
      "loss": 3.558,
      "step": 164970
    },
    {
      "epoch": 0.15378805506885704,
      "grad_norm": 1.1594969034194946,
      "learning_rate": 0.00016924260050560418,
      "loss": 3.6917,
      "step": 164980
    },
    {
      "epoch": 0.15379737668693613,
      "grad_norm": 1.1071122884750366,
      "learning_rate": 0.00016924073618350896,
      "loss": 3.7355,
      "step": 164990
    },
    {
      "epoch": 0.15380669830501523,
      "grad_norm": 1.135420799255371,
      "learning_rate": 0.00016923887186141376,
      "loss": 3.6133,
      "step": 165000
    },
    {
      "epoch": 0.15381601992309432,
      "grad_norm": 1.15708589553833,
      "learning_rate": 0.00016923700753931857,
      "loss": 3.5712,
      "step": 165010
    },
    {
      "epoch": 0.15382534154117342,
      "grad_norm": 1.2972524166107178,
      "learning_rate": 0.00016923514321722335,
      "loss": 3.4851,
      "step": 165020
    },
    {
      "epoch": 0.1538346631592525,
      "grad_norm": 1.154036045074463,
      "learning_rate": 0.00016923327889512815,
      "loss": 3.5764,
      "step": 165030
    },
    {
      "epoch": 0.1538439847773316,
      "grad_norm": 1.193333625793457,
      "learning_rate": 0.00016923141457303296,
      "loss": 3.5854,
      "step": 165040
    },
    {
      "epoch": 0.1538533063954107,
      "grad_norm": 1.5674158334732056,
      "learning_rate": 0.00016922955025093777,
      "loss": 3.5322,
      "step": 165050
    },
    {
      "epoch": 0.15386262801348977,
      "grad_norm": 1.4601540565490723,
      "learning_rate": 0.00016922768592884257,
      "loss": 3.6065,
      "step": 165060
    },
    {
      "epoch": 0.15387194963156886,
      "grad_norm": 1.048707127571106,
      "learning_rate": 0.00016922582160674735,
      "loss": 3.499,
      "step": 165070
    },
    {
      "epoch": 0.15388127124964796,
      "grad_norm": 0.9718937873840332,
      "learning_rate": 0.00016922395728465216,
      "loss": 3.6461,
      "step": 165080
    },
    {
      "epoch": 0.15389059286772705,
      "grad_norm": 0.9229963421821594,
      "learning_rate": 0.00016922209296255696,
      "loss": 3.454,
      "step": 165090
    },
    {
      "epoch": 0.15389991448580614,
      "grad_norm": 1.131352186203003,
      "learning_rate": 0.00016922022864046177,
      "loss": 3.5255,
      "step": 165100
    },
    {
      "epoch": 0.15390923610388524,
      "grad_norm": 1.2837084531784058,
      "learning_rate": 0.00016921836431836658,
      "loss": 3.5971,
      "step": 165110
    },
    {
      "epoch": 0.15391855772196433,
      "grad_norm": 0.9375585913658142,
      "learning_rate": 0.00016921649999627136,
      "loss": 3.6329,
      "step": 165120
    },
    {
      "epoch": 0.15392787934004343,
      "grad_norm": 1.1745142936706543,
      "learning_rate": 0.00016921463567417616,
      "loss": 3.7188,
      "step": 165130
    },
    {
      "epoch": 0.15393720095812252,
      "grad_norm": 1.0054656267166138,
      "learning_rate": 0.00016921277135208097,
      "loss": 3.5332,
      "step": 165140
    },
    {
      "epoch": 0.15394652257620162,
      "grad_norm": 1.3089065551757812,
      "learning_rate": 0.00016921090702998577,
      "loss": 3.6404,
      "step": 165150
    },
    {
      "epoch": 0.1539558441942807,
      "grad_norm": 1.0575594902038574,
      "learning_rate": 0.00016920904270789058,
      "loss": 3.6157,
      "step": 165160
    },
    {
      "epoch": 0.15396516581235978,
      "grad_norm": 1.0709443092346191,
      "learning_rate": 0.00016920717838579536,
      "loss": 3.4154,
      "step": 165170
    },
    {
      "epoch": 0.15397448743043887,
      "grad_norm": 1.1495972871780396,
      "learning_rate": 0.00016920531406370017,
      "loss": 3.4835,
      "step": 165180
    },
    {
      "epoch": 0.15398380904851797,
      "grad_norm": 1.0757064819335938,
      "learning_rate": 0.00016920344974160495,
      "loss": 3.4498,
      "step": 165190
    },
    {
      "epoch": 0.15399313066659706,
      "grad_norm": 1.3630409240722656,
      "learning_rate": 0.00016920158541950978,
      "loss": 3.7134,
      "step": 165200
    },
    {
      "epoch": 0.15400245228467616,
      "grad_norm": 1.0939830541610718,
      "learning_rate": 0.00016919972109741459,
      "loss": 3.5632,
      "step": 165210
    },
    {
      "epoch": 0.15401177390275525,
      "grad_norm": 1.2081135511398315,
      "learning_rate": 0.00016919785677531936,
      "loss": 3.554,
      "step": 165220
    },
    {
      "epoch": 0.15402109552083434,
      "grad_norm": 1.126528024673462,
      "learning_rate": 0.00016919599245322417,
      "loss": 3.4883,
      "step": 165230
    },
    {
      "epoch": 0.15403041713891344,
      "grad_norm": 0.9428755640983582,
      "learning_rate": 0.00016919412813112895,
      "loss": 3.6635,
      "step": 165240
    },
    {
      "epoch": 0.15403973875699253,
      "grad_norm": 1.0213760137557983,
      "learning_rate": 0.00016919226380903376,
      "loss": 3.6887,
      "step": 165250
    },
    {
      "epoch": 0.15404906037507163,
      "grad_norm": 0.9389358758926392,
      "learning_rate": 0.0001691903994869386,
      "loss": 3.6641,
      "step": 165260
    },
    {
      "epoch": 0.1540583819931507,
      "grad_norm": 1.042423129081726,
      "learning_rate": 0.00016918853516484337,
      "loss": 3.673,
      "step": 165270
    },
    {
      "epoch": 0.1540677036112298,
      "grad_norm": 0.881078839302063,
      "learning_rate": 0.00016918667084274818,
      "loss": 3.506,
      "step": 165280
    },
    {
      "epoch": 0.15407702522930888,
      "grad_norm": 0.9689942002296448,
      "learning_rate": 0.00016918480652065295,
      "loss": 3.298,
      "step": 165290
    },
    {
      "epoch": 0.15408634684738798,
      "grad_norm": 0.9344320893287659,
      "learning_rate": 0.00016918294219855776,
      "loss": 3.6378,
      "step": 165300
    },
    {
      "epoch": 0.15409566846546707,
      "grad_norm": 0.8554613590240479,
      "learning_rate": 0.00016918107787646257,
      "loss": 3.5365,
      "step": 165310
    },
    {
      "epoch": 0.15410499008354617,
      "grad_norm": 1.0311999320983887,
      "learning_rate": 0.00016917921355436737,
      "loss": 3.316,
      "step": 165320
    },
    {
      "epoch": 0.15411431170162526,
      "grad_norm": 0.9924087524414062,
      "learning_rate": 0.00016917734923227218,
      "loss": 3.6867,
      "step": 165330
    },
    {
      "epoch": 0.15412363331970436,
      "grad_norm": 1.020005464553833,
      "learning_rate": 0.00016917548491017696,
      "loss": 3.5817,
      "step": 165340
    },
    {
      "epoch": 0.15413295493778345,
      "grad_norm": 1.0527137517929077,
      "learning_rate": 0.00016917362058808177,
      "loss": 3.5983,
      "step": 165350
    },
    {
      "epoch": 0.15414227655586255,
      "grad_norm": 0.8502556681632996,
      "learning_rate": 0.00016917175626598657,
      "loss": 3.7277,
      "step": 165360
    },
    {
      "epoch": 0.15415159817394164,
      "grad_norm": 1.1901313066482544,
      "learning_rate": 0.00016916989194389138,
      "loss": 3.7771,
      "step": 165370
    },
    {
      "epoch": 0.1541609197920207,
      "grad_norm": 0.9357127547264099,
      "learning_rate": 0.00016916802762179618,
      "loss": 3.4577,
      "step": 165380
    },
    {
      "epoch": 0.1541702414100998,
      "grad_norm": 0.9727702140808105,
      "learning_rate": 0.00016916616329970096,
      "loss": 3.4846,
      "step": 165390
    },
    {
      "epoch": 0.1541795630281789,
      "grad_norm": 0.9882925748825073,
      "learning_rate": 0.00016916429897760577,
      "loss": 3.5119,
      "step": 165400
    },
    {
      "epoch": 0.154188884646258,
      "grad_norm": 1.116114854812622,
      "learning_rate": 0.00016916243465551058,
      "loss": 3.4787,
      "step": 165410
    },
    {
      "epoch": 0.15419820626433708,
      "grad_norm": 1.100304126739502,
      "learning_rate": 0.00016916057033341536,
      "loss": 3.5191,
      "step": 165420
    },
    {
      "epoch": 0.15420752788241618,
      "grad_norm": 0.846406877040863,
      "learning_rate": 0.0001691587060113202,
      "loss": 3.5942,
      "step": 165430
    },
    {
      "epoch": 0.15421684950049527,
      "grad_norm": 1.0754395723342896,
      "learning_rate": 0.00016915684168922497,
      "loss": 3.5255,
      "step": 165440
    },
    {
      "epoch": 0.15422617111857437,
      "grad_norm": 1.0407886505126953,
      "learning_rate": 0.00016915497736712977,
      "loss": 3.5669,
      "step": 165450
    },
    {
      "epoch": 0.15423549273665346,
      "grad_norm": 1.1000332832336426,
      "learning_rate": 0.00016915311304503458,
      "loss": 3.4434,
      "step": 165460
    },
    {
      "epoch": 0.15424481435473256,
      "grad_norm": 1.059496521949768,
      "learning_rate": 0.00016915124872293936,
      "loss": 3.363,
      "step": 165470
    },
    {
      "epoch": 0.15425413597281162,
      "grad_norm": 1.0530757904052734,
      "learning_rate": 0.00016914938440084417,
      "loss": 3.4327,
      "step": 165480
    },
    {
      "epoch": 0.15426345759089072,
      "grad_norm": 1.6505805253982544,
      "learning_rate": 0.00016914752007874897,
      "loss": 3.6301,
      "step": 165490
    },
    {
      "epoch": 0.1542727792089698,
      "grad_norm": 0.7516831755638123,
      "learning_rate": 0.00016914565575665378,
      "loss": 3.592,
      "step": 165500
    },
    {
      "epoch": 0.1542821008270489,
      "grad_norm": 1.124738097190857,
      "learning_rate": 0.00016914379143455859,
      "loss": 3.6258,
      "step": 165510
    },
    {
      "epoch": 0.154291422445128,
      "grad_norm": 1.1523590087890625,
      "learning_rate": 0.00016914192711246336,
      "loss": 3.5794,
      "step": 165520
    },
    {
      "epoch": 0.1543007440632071,
      "grad_norm": 1.1694732904434204,
      "learning_rate": 0.00016914006279036817,
      "loss": 3.5024,
      "step": 165530
    },
    {
      "epoch": 0.1543100656812862,
      "grad_norm": 1.0316723585128784,
      "learning_rate": 0.00016913819846827298,
      "loss": 3.5795,
      "step": 165540
    },
    {
      "epoch": 0.15431938729936528,
      "grad_norm": 0.9658834338188171,
      "learning_rate": 0.00016913633414617778,
      "loss": 3.6012,
      "step": 165550
    },
    {
      "epoch": 0.15432870891744438,
      "grad_norm": 1.093953013420105,
      "learning_rate": 0.00016913446982408256,
      "loss": 3.6789,
      "step": 165560
    },
    {
      "epoch": 0.15433803053552347,
      "grad_norm": 0.9825448393821716,
      "learning_rate": 0.00016913260550198737,
      "loss": 3.4322,
      "step": 165570
    },
    {
      "epoch": 0.15434735215360257,
      "grad_norm": 1.1178170442581177,
      "learning_rate": 0.00016913074117989218,
      "loss": 3.6098,
      "step": 165580
    },
    {
      "epoch": 0.15435667377168164,
      "grad_norm": 1.3300706148147583,
      "learning_rate": 0.00016912887685779698,
      "loss": 3.5142,
      "step": 165590
    },
    {
      "epoch": 0.15436599538976073,
      "grad_norm": 1.0483527183532715,
      "learning_rate": 0.0001691270125357018,
      "loss": 3.4794,
      "step": 165600
    },
    {
      "epoch": 0.15437531700783982,
      "grad_norm": 0.9322287440299988,
      "learning_rate": 0.00016912514821360657,
      "loss": 3.5745,
      "step": 165610
    },
    {
      "epoch": 0.15438463862591892,
      "grad_norm": 1.0350161790847778,
      "learning_rate": 0.00016912328389151137,
      "loss": 3.5811,
      "step": 165620
    },
    {
      "epoch": 0.154393960243998,
      "grad_norm": 1.0794274806976318,
      "learning_rate": 0.00016912141956941618,
      "loss": 3.5541,
      "step": 165630
    },
    {
      "epoch": 0.1544032818620771,
      "grad_norm": 1.0636097192764282,
      "learning_rate": 0.00016911955524732096,
      "loss": 3.6854,
      "step": 165640
    },
    {
      "epoch": 0.1544126034801562,
      "grad_norm": 1.3094158172607422,
      "learning_rate": 0.0001691176909252258,
      "loss": 3.3733,
      "step": 165650
    },
    {
      "epoch": 0.1544219250982353,
      "grad_norm": 1.2188695669174194,
      "learning_rate": 0.00016911582660313057,
      "loss": 3.5832,
      "step": 165660
    },
    {
      "epoch": 0.1544312467163144,
      "grad_norm": 1.0696895122528076,
      "learning_rate": 0.00016911396228103538,
      "loss": 3.2772,
      "step": 165670
    },
    {
      "epoch": 0.15444056833439349,
      "grad_norm": 1.0243775844573975,
      "learning_rate": 0.00016911209795894018,
      "loss": 3.6253,
      "step": 165680
    },
    {
      "epoch": 0.15444988995247255,
      "grad_norm": 1.0704830884933472,
      "learning_rate": 0.00016911023363684496,
      "loss": 3.6208,
      "step": 165690
    },
    {
      "epoch": 0.15445921157055165,
      "grad_norm": 1.111835241317749,
      "learning_rate": 0.00016910836931474977,
      "loss": 3.3611,
      "step": 165700
    },
    {
      "epoch": 0.15446853318863074,
      "grad_norm": 0.8976926803588867,
      "learning_rate": 0.00016910650499265458,
      "loss": 3.5733,
      "step": 165710
    },
    {
      "epoch": 0.15447785480670984,
      "grad_norm": 0.971170961856842,
      "learning_rate": 0.00016910464067055938,
      "loss": 3.5316,
      "step": 165720
    },
    {
      "epoch": 0.15448717642478893,
      "grad_norm": 0.979515016078949,
      "learning_rate": 0.0001691027763484642,
      "loss": 3.4444,
      "step": 165730
    },
    {
      "epoch": 0.15449649804286802,
      "grad_norm": 0.8827831149101257,
      "learning_rate": 0.00016910091202636897,
      "loss": 3.701,
      "step": 165740
    },
    {
      "epoch": 0.15450581966094712,
      "grad_norm": 1.0548222064971924,
      "learning_rate": 0.00016909904770427377,
      "loss": 3.6503,
      "step": 165750
    },
    {
      "epoch": 0.1545151412790262,
      "grad_norm": 0.9844936728477478,
      "learning_rate": 0.00016909718338217858,
      "loss": 3.7242,
      "step": 165760
    },
    {
      "epoch": 0.1545244628971053,
      "grad_norm": 1.0392965078353882,
      "learning_rate": 0.0001690953190600834,
      "loss": 3.4539,
      "step": 165770
    },
    {
      "epoch": 0.1545337845151844,
      "grad_norm": 1.0766396522521973,
      "learning_rate": 0.0001690934547379882,
      "loss": 3.5107,
      "step": 165780
    },
    {
      "epoch": 0.1545431061332635,
      "grad_norm": 1.0970826148986816,
      "learning_rate": 0.00016909159041589297,
      "loss": 3.6142,
      "step": 165790
    },
    {
      "epoch": 0.15455242775134256,
      "grad_norm": 1.0840706825256348,
      "learning_rate": 0.00016908972609379778,
      "loss": 3.6538,
      "step": 165800
    },
    {
      "epoch": 0.15456174936942166,
      "grad_norm": 1.0794624090194702,
      "learning_rate": 0.00016908786177170258,
      "loss": 3.5743,
      "step": 165810
    },
    {
      "epoch": 0.15457107098750075,
      "grad_norm": 1.158182144165039,
      "learning_rate": 0.0001690859974496074,
      "loss": 3.6243,
      "step": 165820
    },
    {
      "epoch": 0.15458039260557985,
      "grad_norm": 0.9641348123550415,
      "learning_rate": 0.0001690841331275122,
      "loss": 3.7543,
      "step": 165830
    },
    {
      "epoch": 0.15458971422365894,
      "grad_norm": 1.0154342651367188,
      "learning_rate": 0.00016908226880541698,
      "loss": 3.5494,
      "step": 165840
    },
    {
      "epoch": 0.15459903584173804,
      "grad_norm": 0.9833351969718933,
      "learning_rate": 0.00016908040448332178,
      "loss": 3.5755,
      "step": 165850
    },
    {
      "epoch": 0.15460835745981713,
      "grad_norm": 0.9633689522743225,
      "learning_rate": 0.00016907854016122656,
      "loss": 3.5555,
      "step": 165860
    },
    {
      "epoch": 0.15461767907789623,
      "grad_norm": 1.1535370349884033,
      "learning_rate": 0.0001690766758391314,
      "loss": 3.4469,
      "step": 165870
    },
    {
      "epoch": 0.15462700069597532,
      "grad_norm": 1.093083143234253,
      "learning_rate": 0.0001690748115170362,
      "loss": 3.4532,
      "step": 165880
    },
    {
      "epoch": 0.15463632231405441,
      "grad_norm": 0.9951072335243225,
      "learning_rate": 0.00016907294719494098,
      "loss": 3.4696,
      "step": 165890
    },
    {
      "epoch": 0.15464564393213348,
      "grad_norm": 1.2483948469161987,
      "learning_rate": 0.0001690710828728458,
      "loss": 3.541,
      "step": 165900
    },
    {
      "epoch": 0.15465496555021258,
      "grad_norm": 1.185131549835205,
      "learning_rate": 0.00016906921855075057,
      "loss": 3.4599,
      "step": 165910
    },
    {
      "epoch": 0.15466428716829167,
      "grad_norm": 0.9807480573654175,
      "learning_rate": 0.00016906735422865537,
      "loss": 3.3786,
      "step": 165920
    },
    {
      "epoch": 0.15467360878637076,
      "grad_norm": 0.9657278656959534,
      "learning_rate": 0.0001690654899065602,
      "loss": 3.4165,
      "step": 165930
    },
    {
      "epoch": 0.15468293040444986,
      "grad_norm": 0.9534652829170227,
      "learning_rate": 0.00016906362558446499,
      "loss": 3.7254,
      "step": 165940
    },
    {
      "epoch": 0.15469225202252895,
      "grad_norm": 1.0863354206085205,
      "learning_rate": 0.0001690617612623698,
      "loss": 3.6004,
      "step": 165950
    },
    {
      "epoch": 0.15470157364060805,
      "grad_norm": 0.93277508020401,
      "learning_rate": 0.00016905989694027457,
      "loss": 3.5219,
      "step": 165960
    },
    {
      "epoch": 0.15471089525868714,
      "grad_norm": 1.0134165287017822,
      "learning_rate": 0.00016905803261817938,
      "loss": 3.4735,
      "step": 165970
    },
    {
      "epoch": 0.15472021687676624,
      "grad_norm": 0.9545138478279114,
      "learning_rate": 0.00016905616829608418,
      "loss": 3.222,
      "step": 165980
    },
    {
      "epoch": 0.15472953849484533,
      "grad_norm": 0.8253923654556274,
      "learning_rate": 0.000169054303973989,
      "loss": 3.5596,
      "step": 165990
    },
    {
      "epoch": 0.15473886011292443,
      "grad_norm": 0.9663662910461426,
      "learning_rate": 0.0001690524396518938,
      "loss": 3.4589,
      "step": 166000
    },
    {
      "epoch": 0.1547481817310035,
      "grad_norm": 1.2042481899261475,
      "learning_rate": 0.00016905057532979858,
      "loss": 3.4845,
      "step": 166010
    },
    {
      "epoch": 0.1547575033490826,
      "grad_norm": 0.9208950400352478,
      "learning_rate": 0.00016904871100770338,
      "loss": 3.5223,
      "step": 166020
    },
    {
      "epoch": 0.15476682496716168,
      "grad_norm": 1.0753068923950195,
      "learning_rate": 0.0001690468466856082,
      "loss": 3.5571,
      "step": 166030
    },
    {
      "epoch": 0.15477614658524078,
      "grad_norm": 1.1314748525619507,
      "learning_rate": 0.000169044982363513,
      "loss": 3.4596,
      "step": 166040
    },
    {
      "epoch": 0.15478546820331987,
      "grad_norm": 1.0613874197006226,
      "learning_rate": 0.0001690431180414178,
      "loss": 3.4488,
      "step": 166050
    },
    {
      "epoch": 0.15479478982139896,
      "grad_norm": 0.9099375009536743,
      "learning_rate": 0.00016904125371932258,
      "loss": 3.4447,
      "step": 166060
    },
    {
      "epoch": 0.15480411143947806,
      "grad_norm": 1.2200310230255127,
      "learning_rate": 0.00016903938939722739,
      "loss": 3.4373,
      "step": 166070
    },
    {
      "epoch": 0.15481343305755715,
      "grad_norm": 1.2460849285125732,
      "learning_rate": 0.0001690375250751322,
      "loss": 3.5452,
      "step": 166080
    },
    {
      "epoch": 0.15482275467563625,
      "grad_norm": 1.099685549736023,
      "learning_rate": 0.00016903566075303697,
      "loss": 3.5294,
      "step": 166090
    },
    {
      "epoch": 0.15483207629371534,
      "grad_norm": 0.901608407497406,
      "learning_rate": 0.0001690337964309418,
      "loss": 3.4927,
      "step": 166100
    },
    {
      "epoch": 0.1548413979117944,
      "grad_norm": 1.2977970838546753,
      "learning_rate": 0.00016903193210884658,
      "loss": 3.5034,
      "step": 166110
    },
    {
      "epoch": 0.1548507195298735,
      "grad_norm": 1.0385363101959229,
      "learning_rate": 0.0001690300677867514,
      "loss": 3.5311,
      "step": 166120
    },
    {
      "epoch": 0.1548600411479526,
      "grad_norm": 1.235582947731018,
      "learning_rate": 0.0001690282034646562,
      "loss": 3.6824,
      "step": 166130
    },
    {
      "epoch": 0.1548693627660317,
      "grad_norm": 1.0590847730636597,
      "learning_rate": 0.00016902633914256098,
      "loss": 3.6396,
      "step": 166140
    },
    {
      "epoch": 0.1548786843841108,
      "grad_norm": 1.0396394729614258,
      "learning_rate": 0.00016902447482046578,
      "loss": 3.4906,
      "step": 166150
    },
    {
      "epoch": 0.15488800600218988,
      "grad_norm": 1.1616240739822388,
      "learning_rate": 0.0001690226104983706,
      "loss": 3.5479,
      "step": 166160
    },
    {
      "epoch": 0.15489732762026898,
      "grad_norm": 1.0163296461105347,
      "learning_rate": 0.0001690207461762754,
      "loss": 3.5551,
      "step": 166170
    },
    {
      "epoch": 0.15490664923834807,
      "grad_norm": 0.9816654920578003,
      "learning_rate": 0.0001690188818541802,
      "loss": 3.5875,
      "step": 166180
    },
    {
      "epoch": 0.15491597085642717,
      "grad_norm": 1.184912919998169,
      "learning_rate": 0.00016901701753208498,
      "loss": 3.5012,
      "step": 166190
    },
    {
      "epoch": 0.15492529247450626,
      "grad_norm": 0.8462565541267395,
      "learning_rate": 0.0001690151532099898,
      "loss": 3.5932,
      "step": 166200
    },
    {
      "epoch": 0.15493461409258533,
      "grad_norm": 0.9010770916938782,
      "learning_rate": 0.0001690132888878946,
      "loss": 3.6285,
      "step": 166210
    },
    {
      "epoch": 0.15494393571066442,
      "grad_norm": 1.002359390258789,
      "learning_rate": 0.0001690114245657994,
      "loss": 3.4873,
      "step": 166220
    },
    {
      "epoch": 0.15495325732874352,
      "grad_norm": 1.1440932750701904,
      "learning_rate": 0.0001690095602437042,
      "loss": 3.4406,
      "step": 166230
    },
    {
      "epoch": 0.1549625789468226,
      "grad_norm": 1.0984374284744263,
      "learning_rate": 0.00016900769592160898,
      "loss": 3.4328,
      "step": 166240
    },
    {
      "epoch": 0.1549719005649017,
      "grad_norm": 1.0524401664733887,
      "learning_rate": 0.0001690058315995138,
      "loss": 3.7768,
      "step": 166250
    },
    {
      "epoch": 0.1549812221829808,
      "grad_norm": 0.9875895380973816,
      "learning_rate": 0.0001690039672774186,
      "loss": 3.6674,
      "step": 166260
    },
    {
      "epoch": 0.1549905438010599,
      "grad_norm": 1.1003822088241577,
      "learning_rate": 0.0001690021029553234,
      "loss": 3.5304,
      "step": 166270
    },
    {
      "epoch": 0.154999865419139,
      "grad_norm": 1.228245496749878,
      "learning_rate": 0.0001690002386332282,
      "loss": 3.6911,
      "step": 166280
    },
    {
      "epoch": 0.15500918703721808,
      "grad_norm": 1.052619218826294,
      "learning_rate": 0.000168998374311133,
      "loss": 3.4283,
      "step": 166290
    },
    {
      "epoch": 0.15501850865529718,
      "grad_norm": 1.2189091444015503,
      "learning_rate": 0.0001689965099890378,
      "loss": 3.6508,
      "step": 166300
    },
    {
      "epoch": 0.15502783027337627,
      "grad_norm": 0.9451297521591187,
      "learning_rate": 0.00016899464566694257,
      "loss": 3.5244,
      "step": 166310
    },
    {
      "epoch": 0.15503715189145534,
      "grad_norm": 1.0226436853408813,
      "learning_rate": 0.0001689927813448474,
      "loss": 3.4471,
      "step": 166320
    },
    {
      "epoch": 0.15504647350953443,
      "grad_norm": 1.1614946126937866,
      "learning_rate": 0.00016899091702275221,
      "loss": 3.4934,
      "step": 166330
    },
    {
      "epoch": 0.15505579512761353,
      "grad_norm": 1.058862566947937,
      "learning_rate": 0.000168989052700657,
      "loss": 3.6161,
      "step": 166340
    },
    {
      "epoch": 0.15506511674569262,
      "grad_norm": 1.0386860370635986,
      "learning_rate": 0.0001689871883785618,
      "loss": 3.5899,
      "step": 166350
    },
    {
      "epoch": 0.15507443836377172,
      "grad_norm": 1.0076767206192017,
      "learning_rate": 0.00016898532405646658,
      "loss": 3.6687,
      "step": 166360
    },
    {
      "epoch": 0.1550837599818508,
      "grad_norm": 1.1335031986236572,
      "learning_rate": 0.00016898345973437139,
      "loss": 3.6182,
      "step": 166370
    },
    {
      "epoch": 0.1550930815999299,
      "grad_norm": 1.212542176246643,
      "learning_rate": 0.00016898159541227622,
      "loss": 3.4722,
      "step": 166380
    },
    {
      "epoch": 0.155102403218009,
      "grad_norm": 0.9754291772842407,
      "learning_rate": 0.000168979731090181,
      "loss": 3.6091,
      "step": 166390
    },
    {
      "epoch": 0.1551117248360881,
      "grad_norm": 1.04908287525177,
      "learning_rate": 0.0001689778667680858,
      "loss": 3.6145,
      "step": 166400
    },
    {
      "epoch": 0.1551210464541672,
      "grad_norm": 1.410444736480713,
      "learning_rate": 0.00016897600244599058,
      "loss": 3.513,
      "step": 166410
    },
    {
      "epoch": 0.15513036807224626,
      "grad_norm": 0.9698530435562134,
      "learning_rate": 0.0001689741381238954,
      "loss": 3.5383,
      "step": 166420
    },
    {
      "epoch": 0.15513968969032535,
      "grad_norm": 1.189680814743042,
      "learning_rate": 0.0001689722738018002,
      "loss": 3.5124,
      "step": 166430
    },
    {
      "epoch": 0.15514901130840444,
      "grad_norm": 0.9069275259971619,
      "learning_rate": 0.000168970409479705,
      "loss": 3.5419,
      "step": 166440
    },
    {
      "epoch": 0.15515833292648354,
      "grad_norm": 1.012080430984497,
      "learning_rate": 0.0001689685451576098,
      "loss": 3.586,
      "step": 166450
    },
    {
      "epoch": 0.15516765454456263,
      "grad_norm": 1.147639274597168,
      "learning_rate": 0.0001689666808355146,
      "loss": 3.4127,
      "step": 166460
    },
    {
      "epoch": 0.15517697616264173,
      "grad_norm": 0.9583269357681274,
      "learning_rate": 0.0001689648165134194,
      "loss": 3.4005,
      "step": 166470
    },
    {
      "epoch": 0.15518629778072082,
      "grad_norm": 1.172580599784851,
      "learning_rate": 0.0001689629521913242,
      "loss": 3.4798,
      "step": 166480
    },
    {
      "epoch": 0.15519561939879992,
      "grad_norm": 1.1665916442871094,
      "learning_rate": 0.000168961087869229,
      "loss": 3.5732,
      "step": 166490
    },
    {
      "epoch": 0.155204941016879,
      "grad_norm": 1.0993471145629883,
      "learning_rate": 0.0001689592235471338,
      "loss": 3.5911,
      "step": 166500
    },
    {
      "epoch": 0.1552142626349581,
      "grad_norm": 0.908584713935852,
      "learning_rate": 0.0001689573592250386,
      "loss": 3.6219,
      "step": 166510
    },
    {
      "epoch": 0.1552235842530372,
      "grad_norm": 0.9526491761207581,
      "learning_rate": 0.0001689554949029434,
      "loss": 3.6591,
      "step": 166520
    },
    {
      "epoch": 0.15523290587111627,
      "grad_norm": 1.0634857416152954,
      "learning_rate": 0.0001689536305808482,
      "loss": 3.6839,
      "step": 166530
    },
    {
      "epoch": 0.15524222748919536,
      "grad_norm": 1.2138841152191162,
      "learning_rate": 0.00016895176625875298,
      "loss": 3.3834,
      "step": 166540
    },
    {
      "epoch": 0.15525154910727446,
      "grad_norm": 1.1216996908187866,
      "learning_rate": 0.00016894990193665782,
      "loss": 3.4488,
      "step": 166550
    },
    {
      "epoch": 0.15526087072535355,
      "grad_norm": 1.15871262550354,
      "learning_rate": 0.0001689480376145626,
      "loss": 3.6869,
      "step": 166560
    },
    {
      "epoch": 0.15527019234343264,
      "grad_norm": 0.9932476282119751,
      "learning_rate": 0.0001689461732924674,
      "loss": 3.7139,
      "step": 166570
    },
    {
      "epoch": 0.15527951396151174,
      "grad_norm": 1.095145344734192,
      "learning_rate": 0.0001689443089703722,
      "loss": 3.5155,
      "step": 166580
    },
    {
      "epoch": 0.15528883557959083,
      "grad_norm": 1.0815259218215942,
      "learning_rate": 0.000168942444648277,
      "loss": 3.5348,
      "step": 166590
    },
    {
      "epoch": 0.15529815719766993,
      "grad_norm": 1.020451307296753,
      "learning_rate": 0.0001689405803261818,
      "loss": 3.5127,
      "step": 166600
    },
    {
      "epoch": 0.15530747881574902,
      "grad_norm": 0.9295507669448853,
      "learning_rate": 0.0001689387160040866,
      "loss": 3.5453,
      "step": 166610
    },
    {
      "epoch": 0.15531680043382812,
      "grad_norm": 0.9669461250305176,
      "learning_rate": 0.0001689368516819914,
      "loss": 3.4496,
      "step": 166620
    },
    {
      "epoch": 0.15532612205190718,
      "grad_norm": 1.0948883295059204,
      "learning_rate": 0.00016893498735989621,
      "loss": 3.6924,
      "step": 166630
    },
    {
      "epoch": 0.15533544366998628,
      "grad_norm": 1.0696985721588135,
      "learning_rate": 0.000168933123037801,
      "loss": 3.5885,
      "step": 166640
    },
    {
      "epoch": 0.15534476528806537,
      "grad_norm": 0.9883484840393066,
      "learning_rate": 0.0001689312587157058,
      "loss": 3.4305,
      "step": 166650
    },
    {
      "epoch": 0.15535408690614447,
      "grad_norm": 1.0554239749908447,
      "learning_rate": 0.0001689293943936106,
      "loss": 3.6629,
      "step": 166660
    },
    {
      "epoch": 0.15536340852422356,
      "grad_norm": 1.0353482961654663,
      "learning_rate": 0.0001689275300715154,
      "loss": 3.6977,
      "step": 166670
    },
    {
      "epoch": 0.15537273014230266,
      "grad_norm": 1.046710729598999,
      "learning_rate": 0.00016892566574942022,
      "loss": 3.7566,
      "step": 166680
    },
    {
      "epoch": 0.15538205176038175,
      "grad_norm": 1.1952942609786987,
      "learning_rate": 0.000168923801427325,
      "loss": 3.6261,
      "step": 166690
    },
    {
      "epoch": 0.15539137337846085,
      "grad_norm": 1.080081820487976,
      "learning_rate": 0.0001689219371052298,
      "loss": 3.5995,
      "step": 166700
    },
    {
      "epoch": 0.15540069499653994,
      "grad_norm": 0.9173858761787415,
      "learning_rate": 0.0001689200727831346,
      "loss": 3.5567,
      "step": 166710
    },
    {
      "epoch": 0.15541001661461903,
      "grad_norm": 1.2603999376296997,
      "learning_rate": 0.00016891820846103942,
      "loss": 3.7371,
      "step": 166720
    },
    {
      "epoch": 0.15541933823269813,
      "grad_norm": 1.06190025806427,
      "learning_rate": 0.00016891634413894422,
      "loss": 3.5385,
      "step": 166730
    },
    {
      "epoch": 0.1554286598507772,
      "grad_norm": 1.0523204803466797,
      "learning_rate": 0.000168914479816849,
      "loss": 3.4968,
      "step": 166740
    },
    {
      "epoch": 0.1554379814688563,
      "grad_norm": 1.0639114379882812,
      "learning_rate": 0.0001689126154947538,
      "loss": 3.5919,
      "step": 166750
    },
    {
      "epoch": 0.15544730308693538,
      "grad_norm": 1.0633403062820435,
      "learning_rate": 0.0001689107511726586,
      "loss": 3.4785,
      "step": 166760
    },
    {
      "epoch": 0.15545662470501448,
      "grad_norm": 1.0104197263717651,
      "learning_rate": 0.00016890888685056342,
      "loss": 3.5969,
      "step": 166770
    },
    {
      "epoch": 0.15546594632309357,
      "grad_norm": 1.0758122205734253,
      "learning_rate": 0.00016890702252846823,
      "loss": 3.5955,
      "step": 166780
    },
    {
      "epoch": 0.15547526794117267,
      "grad_norm": 1.1714813709259033,
      "learning_rate": 0.000168905158206373,
      "loss": 3.5369,
      "step": 166790
    },
    {
      "epoch": 0.15548458955925176,
      "grad_norm": 1.0051909685134888,
      "learning_rate": 0.0001689032938842778,
      "loss": 3.5593,
      "step": 166800
    },
    {
      "epoch": 0.15549391117733086,
      "grad_norm": 1.2156893014907837,
      "learning_rate": 0.0001689014295621826,
      "loss": 3.5131,
      "step": 166810
    },
    {
      "epoch": 0.15550323279540995,
      "grad_norm": 1.1314343214035034,
      "learning_rate": 0.0001688995652400874,
      "loss": 3.6073,
      "step": 166820
    },
    {
      "epoch": 0.15551255441348905,
      "grad_norm": 1.0823273658752441,
      "learning_rate": 0.00016889770091799223,
      "loss": 3.5382,
      "step": 166830
    },
    {
      "epoch": 0.1555218760315681,
      "grad_norm": 1.2003529071807861,
      "learning_rate": 0.000168895836595897,
      "loss": 3.3653,
      "step": 166840
    },
    {
      "epoch": 0.1555311976496472,
      "grad_norm": 1.127121090888977,
      "learning_rate": 0.00016889397227380182,
      "loss": 3.4522,
      "step": 166850
    },
    {
      "epoch": 0.1555405192677263,
      "grad_norm": 1.1202287673950195,
      "learning_rate": 0.0001688921079517066,
      "loss": 3.509,
      "step": 166860
    },
    {
      "epoch": 0.1555498408858054,
      "grad_norm": 0.9956136345863342,
      "learning_rate": 0.0001688902436296114,
      "loss": 3.7685,
      "step": 166870
    },
    {
      "epoch": 0.1555591625038845,
      "grad_norm": 1.120324730873108,
      "learning_rate": 0.0001688883793075162,
      "loss": 3.547,
      "step": 166880
    },
    {
      "epoch": 0.15556848412196358,
      "grad_norm": 0.9497268795967102,
      "learning_rate": 0.00016888651498542102,
      "loss": 3.705,
      "step": 166890
    },
    {
      "epoch": 0.15557780574004268,
      "grad_norm": 1.0226426124572754,
      "learning_rate": 0.00016888465066332582,
      "loss": 3.5286,
      "step": 166900
    },
    {
      "epoch": 0.15558712735812177,
      "grad_norm": 1.051108479499817,
      "learning_rate": 0.0001688827863412306,
      "loss": 3.5685,
      "step": 166910
    },
    {
      "epoch": 0.15559644897620087,
      "grad_norm": 1.107204794883728,
      "learning_rate": 0.0001688809220191354,
      "loss": 3.4993,
      "step": 166920
    },
    {
      "epoch": 0.15560577059427996,
      "grad_norm": 1.0843216180801392,
      "learning_rate": 0.0001688790576970402,
      "loss": 3.4937,
      "step": 166930
    },
    {
      "epoch": 0.15561509221235906,
      "grad_norm": 1.3512359857559204,
      "learning_rate": 0.00016887719337494502,
      "loss": 3.7156,
      "step": 166940
    },
    {
      "epoch": 0.15562441383043812,
      "grad_norm": 0.9948915839195251,
      "learning_rate": 0.00016887532905284983,
      "loss": 3.5478,
      "step": 166950
    },
    {
      "epoch": 0.15563373544851722,
      "grad_norm": 1.0472779273986816,
      "learning_rate": 0.0001688734647307546,
      "loss": 3.7311,
      "step": 166960
    },
    {
      "epoch": 0.1556430570665963,
      "grad_norm": 1.091193675994873,
      "learning_rate": 0.0001688716004086594,
      "loss": 3.7464,
      "step": 166970
    },
    {
      "epoch": 0.1556523786846754,
      "grad_norm": 1.3698279857635498,
      "learning_rate": 0.0001688697360865642,
      "loss": 3.7312,
      "step": 166980
    },
    {
      "epoch": 0.1556617003027545,
      "grad_norm": 0.9614229202270508,
      "learning_rate": 0.000168867871764469,
      "loss": 3.5756,
      "step": 166990
    },
    {
      "epoch": 0.1556710219208336,
      "grad_norm": 1.0328223705291748,
      "learning_rate": 0.00016886600744237383,
      "loss": 3.4914,
      "step": 167000
    },
    {
      "epoch": 0.1556803435389127,
      "grad_norm": 1.0005080699920654,
      "learning_rate": 0.0001688641431202786,
      "loss": 3.6081,
      "step": 167010
    },
    {
      "epoch": 0.15568966515699179,
      "grad_norm": 1.117299199104309,
      "learning_rate": 0.00016886227879818342,
      "loss": 3.4095,
      "step": 167020
    },
    {
      "epoch": 0.15569898677507088,
      "grad_norm": 1.2728689908981323,
      "learning_rate": 0.0001688604144760882,
      "loss": 3.5204,
      "step": 167030
    },
    {
      "epoch": 0.15570830839314997,
      "grad_norm": 0.947207510471344,
      "learning_rate": 0.000168858550153993,
      "loss": 3.7896,
      "step": 167040
    },
    {
      "epoch": 0.15571763001122904,
      "grad_norm": 1.2145086526870728,
      "learning_rate": 0.0001688566858318978,
      "loss": 3.6919,
      "step": 167050
    },
    {
      "epoch": 0.15572695162930814,
      "grad_norm": 1.086267352104187,
      "learning_rate": 0.00016885482150980261,
      "loss": 3.5573,
      "step": 167060
    },
    {
      "epoch": 0.15573627324738723,
      "grad_norm": 0.9720004200935364,
      "learning_rate": 0.00016885295718770742,
      "loss": 3.4643,
      "step": 167070
    },
    {
      "epoch": 0.15574559486546632,
      "grad_norm": 0.98741215467453,
      "learning_rate": 0.0001688510928656122,
      "loss": 3.5919,
      "step": 167080
    },
    {
      "epoch": 0.15575491648354542,
      "grad_norm": 1.1257222890853882,
      "learning_rate": 0.000168849228543517,
      "loss": 3.6971,
      "step": 167090
    },
    {
      "epoch": 0.1557642381016245,
      "grad_norm": 1.2457163333892822,
      "learning_rate": 0.0001688473642214218,
      "loss": 3.6724,
      "step": 167100
    },
    {
      "epoch": 0.1557735597197036,
      "grad_norm": 1.0304208993911743,
      "learning_rate": 0.00016884549989932662,
      "loss": 3.7369,
      "step": 167110
    },
    {
      "epoch": 0.1557828813377827,
      "grad_norm": 0.9018560647964478,
      "learning_rate": 0.00016884363557723143,
      "loss": 3.7407,
      "step": 167120
    },
    {
      "epoch": 0.1557922029558618,
      "grad_norm": 0.813230574131012,
      "learning_rate": 0.0001688417712551362,
      "loss": 3.6158,
      "step": 167130
    },
    {
      "epoch": 0.1558015245739409,
      "grad_norm": 0.9766039848327637,
      "learning_rate": 0.000168839906933041,
      "loss": 3.4515,
      "step": 167140
    },
    {
      "epoch": 0.15581084619201999,
      "grad_norm": 1.1177974939346313,
      "learning_rate": 0.00016883804261094582,
      "loss": 3.6633,
      "step": 167150
    },
    {
      "epoch": 0.15582016781009905,
      "grad_norm": 1.1210585832595825,
      "learning_rate": 0.00016883617828885062,
      "loss": 3.5973,
      "step": 167160
    },
    {
      "epoch": 0.15582948942817815,
      "grad_norm": 1.1654597520828247,
      "learning_rate": 0.00016883431396675543,
      "loss": 3.4949,
      "step": 167170
    },
    {
      "epoch": 0.15583881104625724,
      "grad_norm": 0.9449802041053772,
      "learning_rate": 0.0001688324496446602,
      "loss": 3.5058,
      "step": 167180
    },
    {
      "epoch": 0.15584813266433634,
      "grad_norm": 1.2988642454147339,
      "learning_rate": 0.00016883058532256502,
      "loss": 3.569,
      "step": 167190
    },
    {
      "epoch": 0.15585745428241543,
      "grad_norm": 1.203901767730713,
      "learning_rate": 0.00016882872100046982,
      "loss": 3.6135,
      "step": 167200
    },
    {
      "epoch": 0.15586677590049453,
      "grad_norm": 1.218390941619873,
      "learning_rate": 0.0001688268566783746,
      "loss": 3.3248,
      "step": 167210
    },
    {
      "epoch": 0.15587609751857362,
      "grad_norm": 1.000434398651123,
      "learning_rate": 0.00016882499235627943,
      "loss": 3.6384,
      "step": 167220
    },
    {
      "epoch": 0.15588541913665271,
      "grad_norm": 0.9630733728408813,
      "learning_rate": 0.0001688231280341842,
      "loss": 3.3789,
      "step": 167230
    },
    {
      "epoch": 0.1558947407547318,
      "grad_norm": 0.8696102499961853,
      "learning_rate": 0.00016882126371208902,
      "loss": 3.6304,
      "step": 167240
    },
    {
      "epoch": 0.1559040623728109,
      "grad_norm": 0.9758385419845581,
      "learning_rate": 0.00016881939938999383,
      "loss": 3.6075,
      "step": 167250
    },
    {
      "epoch": 0.15591338399088997,
      "grad_norm": 1.2804304361343384,
      "learning_rate": 0.0001688175350678986,
      "loss": 3.54,
      "step": 167260
    },
    {
      "epoch": 0.15592270560896906,
      "grad_norm": 1.1338680982589722,
      "learning_rate": 0.0001688156707458034,
      "loss": 3.4385,
      "step": 167270
    },
    {
      "epoch": 0.15593202722704816,
      "grad_norm": 1.0149275064468384,
      "learning_rate": 0.00016881380642370822,
      "loss": 3.5642,
      "step": 167280
    },
    {
      "epoch": 0.15594134884512725,
      "grad_norm": 0.9014337062835693,
      "learning_rate": 0.00016881194210161302,
      "loss": 3.3139,
      "step": 167290
    },
    {
      "epoch": 0.15595067046320635,
      "grad_norm": 1.1547307968139648,
      "learning_rate": 0.00016881007777951783,
      "loss": 3.6538,
      "step": 167300
    },
    {
      "epoch": 0.15595999208128544,
      "grad_norm": 1.082494854927063,
      "learning_rate": 0.0001688082134574226,
      "loss": 3.5227,
      "step": 167310
    },
    {
      "epoch": 0.15596931369936454,
      "grad_norm": 0.8067773580551147,
      "learning_rate": 0.00016880634913532742,
      "loss": 3.6091,
      "step": 167320
    },
    {
      "epoch": 0.15597863531744363,
      "grad_norm": 1.3779375553131104,
      "learning_rate": 0.00016880448481323222,
      "loss": 3.5992,
      "step": 167330
    },
    {
      "epoch": 0.15598795693552273,
      "grad_norm": 1.017523169517517,
      "learning_rate": 0.00016880262049113703,
      "loss": 3.5268,
      "step": 167340
    },
    {
      "epoch": 0.15599727855360182,
      "grad_norm": 0.8955508470535278,
      "learning_rate": 0.00016880075616904183,
      "loss": 3.5841,
      "step": 167350
    },
    {
      "epoch": 0.15600660017168091,
      "grad_norm": 1.176692247390747,
      "learning_rate": 0.00016879889184694661,
      "loss": 3.5509,
      "step": 167360
    },
    {
      "epoch": 0.15601592178975998,
      "grad_norm": 0.9648507833480835,
      "learning_rate": 0.00016879702752485142,
      "loss": 3.4641,
      "step": 167370
    },
    {
      "epoch": 0.15602524340783908,
      "grad_norm": 0.9963885545730591,
      "learning_rate": 0.0001687951632027562,
      "loss": 3.6385,
      "step": 167380
    },
    {
      "epoch": 0.15603456502591817,
      "grad_norm": 1.1011197566986084,
      "learning_rate": 0.00016879329888066103,
      "loss": 3.5365,
      "step": 167390
    },
    {
      "epoch": 0.15604388664399726,
      "grad_norm": 1.0204663276672363,
      "learning_rate": 0.00016879143455856584,
      "loss": 3.5668,
      "step": 167400
    },
    {
      "epoch": 0.15605320826207636,
      "grad_norm": 0.9778481125831604,
      "learning_rate": 0.00016878957023647062,
      "loss": 3.3722,
      "step": 167410
    },
    {
      "epoch": 0.15606252988015545,
      "grad_norm": 1.1778634786605835,
      "learning_rate": 0.00016878770591437542,
      "loss": 3.4259,
      "step": 167420
    },
    {
      "epoch": 0.15607185149823455,
      "grad_norm": 0.9452992081642151,
      "learning_rate": 0.0001687858415922802,
      "loss": 3.5393,
      "step": 167430
    },
    {
      "epoch": 0.15608117311631364,
      "grad_norm": 1.2117304801940918,
      "learning_rate": 0.000168783977270185,
      "loss": 3.6179,
      "step": 167440
    },
    {
      "epoch": 0.15609049473439274,
      "grad_norm": 1.1878020763397217,
      "learning_rate": 0.00016878211294808984,
      "loss": 3.5564,
      "step": 167450
    },
    {
      "epoch": 0.15609981635247183,
      "grad_norm": 1.0123838186264038,
      "learning_rate": 0.00016878024862599462,
      "loss": 3.4739,
      "step": 167460
    },
    {
      "epoch": 0.1561091379705509,
      "grad_norm": 1.123936414718628,
      "learning_rate": 0.00016877838430389943,
      "loss": 3.4068,
      "step": 167470
    },
    {
      "epoch": 0.15611845958863,
      "grad_norm": 1.0541160106658936,
      "learning_rate": 0.0001687765199818042,
      "loss": 3.6538,
      "step": 167480
    },
    {
      "epoch": 0.1561277812067091,
      "grad_norm": 1.1963505744934082,
      "learning_rate": 0.00016877465565970901,
      "loss": 3.5564,
      "step": 167490
    },
    {
      "epoch": 0.15613710282478818,
      "grad_norm": 1.0228161811828613,
      "learning_rate": 0.00016877279133761382,
      "loss": 3.5215,
      "step": 167500
    },
    {
      "epoch": 0.15614642444286728,
      "grad_norm": 0.9912100434303284,
      "learning_rate": 0.00016877092701551863,
      "loss": 3.4884,
      "step": 167510
    },
    {
      "epoch": 0.15615574606094637,
      "grad_norm": 0.9912207722663879,
      "learning_rate": 0.00016876906269342343,
      "loss": 3.5829,
      "step": 167520
    },
    {
      "epoch": 0.15616506767902547,
      "grad_norm": 1.212174892425537,
      "learning_rate": 0.0001687671983713282,
      "loss": 3.6821,
      "step": 167530
    },
    {
      "epoch": 0.15617438929710456,
      "grad_norm": 1.2197264432907104,
      "learning_rate": 0.00016876533404923302,
      "loss": 3.6981,
      "step": 167540
    },
    {
      "epoch": 0.15618371091518365,
      "grad_norm": 1.0669991970062256,
      "learning_rate": 0.00016876346972713783,
      "loss": 3.4521,
      "step": 167550
    },
    {
      "epoch": 0.15619303253326275,
      "grad_norm": 0.8915425539016724,
      "learning_rate": 0.00016876160540504263,
      "loss": 3.4152,
      "step": 167560
    },
    {
      "epoch": 0.15620235415134182,
      "grad_norm": 0.9465056657791138,
      "learning_rate": 0.00016875974108294744,
      "loss": 3.5573,
      "step": 167570
    },
    {
      "epoch": 0.1562116757694209,
      "grad_norm": 0.9985895752906799,
      "learning_rate": 0.00016875787676085222,
      "loss": 3.6513,
      "step": 167580
    },
    {
      "epoch": 0.1562209973875,
      "grad_norm": 0.9737000465393066,
      "learning_rate": 0.00016875601243875702,
      "loss": 3.5017,
      "step": 167590
    },
    {
      "epoch": 0.1562303190055791,
      "grad_norm": 1.1870970726013184,
      "learning_rate": 0.00016875414811666183,
      "loss": 3.7328,
      "step": 167600
    },
    {
      "epoch": 0.1562396406236582,
      "grad_norm": 1.1131789684295654,
      "learning_rate": 0.00016875228379456664,
      "loss": 3.5052,
      "step": 167610
    },
    {
      "epoch": 0.1562489622417373,
      "grad_norm": 1.0718320608139038,
      "learning_rate": 0.00016875041947247144,
      "loss": 3.5655,
      "step": 167620
    },
    {
      "epoch": 0.15625828385981638,
      "grad_norm": 0.8984319567680359,
      "learning_rate": 0.00016874855515037622,
      "loss": 3.5916,
      "step": 167630
    },
    {
      "epoch": 0.15626760547789548,
      "grad_norm": 1.1703580617904663,
      "learning_rate": 0.00016874669082828103,
      "loss": 3.7061,
      "step": 167640
    },
    {
      "epoch": 0.15627692709597457,
      "grad_norm": 1.0587035417556763,
      "learning_rate": 0.00016874482650618583,
      "loss": 3.5268,
      "step": 167650
    },
    {
      "epoch": 0.15628624871405367,
      "grad_norm": 1.119568943977356,
      "learning_rate": 0.0001687429621840906,
      "loss": 3.5429,
      "step": 167660
    },
    {
      "epoch": 0.15629557033213276,
      "grad_norm": 1.1256263256072998,
      "learning_rate": 0.00016874109786199545,
      "loss": 3.5827,
      "step": 167670
    },
    {
      "epoch": 0.15630489195021183,
      "grad_norm": 0.9994404315948486,
      "learning_rate": 0.00016873923353990023,
      "loss": 3.6151,
      "step": 167680
    },
    {
      "epoch": 0.15631421356829092,
      "grad_norm": 0.9506731629371643,
      "learning_rate": 0.00016873736921780503,
      "loss": 3.5014,
      "step": 167690
    },
    {
      "epoch": 0.15632353518637002,
      "grad_norm": 1.1947400569915771,
      "learning_rate": 0.00016873550489570984,
      "loss": 3.4383,
      "step": 167700
    },
    {
      "epoch": 0.1563328568044491,
      "grad_norm": 0.8510608673095703,
      "learning_rate": 0.00016873364057361462,
      "loss": 3.5586,
      "step": 167710
    },
    {
      "epoch": 0.1563421784225282,
      "grad_norm": 0.8907521963119507,
      "learning_rate": 0.00016873177625151942,
      "loss": 3.5784,
      "step": 167720
    },
    {
      "epoch": 0.1563515000406073,
      "grad_norm": 1.0870147943496704,
      "learning_rate": 0.00016872991192942423,
      "loss": 3.484,
      "step": 167730
    },
    {
      "epoch": 0.1563608216586864,
      "grad_norm": 1.0056413412094116,
      "learning_rate": 0.00016872804760732904,
      "loss": 3.755,
      "step": 167740
    },
    {
      "epoch": 0.1563701432767655,
      "grad_norm": 0.9911902546882629,
      "learning_rate": 0.00016872618328523384,
      "loss": 3.4945,
      "step": 167750
    },
    {
      "epoch": 0.15637946489484458,
      "grad_norm": 1.4073680639266968,
      "learning_rate": 0.00016872431896313862,
      "loss": 3.6123,
      "step": 167760
    },
    {
      "epoch": 0.15638878651292368,
      "grad_norm": 1.3037173748016357,
      "learning_rate": 0.00016872245464104343,
      "loss": 3.5021,
      "step": 167770
    },
    {
      "epoch": 0.15639810813100274,
      "grad_norm": 1.2786725759506226,
      "learning_rate": 0.00016872059031894823,
      "loss": 3.5184,
      "step": 167780
    },
    {
      "epoch": 0.15640742974908184,
      "grad_norm": 1.2690519094467163,
      "learning_rate": 0.00016871872599685304,
      "loss": 3.5147,
      "step": 167790
    },
    {
      "epoch": 0.15641675136716093,
      "grad_norm": 1.032920002937317,
      "learning_rate": 0.00016871686167475785,
      "loss": 3.5633,
      "step": 167800
    },
    {
      "epoch": 0.15642607298524003,
      "grad_norm": 1.1167542934417725,
      "learning_rate": 0.00016871499735266263,
      "loss": 3.5293,
      "step": 167810
    },
    {
      "epoch": 0.15643539460331912,
      "grad_norm": 1.0594069957733154,
      "learning_rate": 0.00016871313303056743,
      "loss": 3.5355,
      "step": 167820
    },
    {
      "epoch": 0.15644471622139822,
      "grad_norm": 0.8724175095558167,
      "learning_rate": 0.0001687112687084722,
      "loss": 3.5272,
      "step": 167830
    },
    {
      "epoch": 0.1564540378394773,
      "grad_norm": 1.1945291757583618,
      "learning_rate": 0.00016870940438637705,
      "loss": 3.5252,
      "step": 167840
    },
    {
      "epoch": 0.1564633594575564,
      "grad_norm": 1.0713368654251099,
      "learning_rate": 0.00016870754006428185,
      "loss": 3.5033,
      "step": 167850
    },
    {
      "epoch": 0.1564726810756355,
      "grad_norm": 1.152013897895813,
      "learning_rate": 0.00016870567574218663,
      "loss": 3.4298,
      "step": 167860
    },
    {
      "epoch": 0.1564820026937146,
      "grad_norm": 1.2029041051864624,
      "learning_rate": 0.00016870381142009144,
      "loss": 3.5691,
      "step": 167870
    },
    {
      "epoch": 0.1564913243117937,
      "grad_norm": 1.1792081594467163,
      "learning_rate": 0.00016870194709799622,
      "loss": 3.4242,
      "step": 167880
    },
    {
      "epoch": 0.15650064592987276,
      "grad_norm": 1.1039406061172485,
      "learning_rate": 0.00016870008277590102,
      "loss": 3.4973,
      "step": 167890
    },
    {
      "epoch": 0.15650996754795185,
      "grad_norm": 1.0631757974624634,
      "learning_rate": 0.00016869821845380586,
      "loss": 3.5338,
      "step": 167900
    },
    {
      "epoch": 0.15651928916603094,
      "grad_norm": 1.119761347770691,
      "learning_rate": 0.00016869635413171064,
      "loss": 3.3988,
      "step": 167910
    },
    {
      "epoch": 0.15652861078411004,
      "grad_norm": 0.8927003145217896,
      "learning_rate": 0.00016869448980961544,
      "loss": 3.7687,
      "step": 167920
    },
    {
      "epoch": 0.15653793240218913,
      "grad_norm": 1.054305076599121,
      "learning_rate": 0.00016869262548752022,
      "loss": 3.7299,
      "step": 167930
    },
    {
      "epoch": 0.15654725402026823,
      "grad_norm": 0.6939016580581665,
      "learning_rate": 0.00016869076116542503,
      "loss": 3.609,
      "step": 167940
    },
    {
      "epoch": 0.15655657563834732,
      "grad_norm": 1.0598827600479126,
      "learning_rate": 0.00016868889684332983,
      "loss": 3.3994,
      "step": 167950
    },
    {
      "epoch": 0.15656589725642642,
      "grad_norm": 0.9588362574577332,
      "learning_rate": 0.00016868703252123464,
      "loss": 3.644,
      "step": 167960
    },
    {
      "epoch": 0.1565752188745055,
      "grad_norm": 1.2200297117233276,
      "learning_rate": 0.00016868516819913945,
      "loss": 3.6534,
      "step": 167970
    },
    {
      "epoch": 0.1565845404925846,
      "grad_norm": 1.0511387586593628,
      "learning_rate": 0.00016868330387704423,
      "loss": 3.343,
      "step": 167980
    },
    {
      "epoch": 0.15659386211066367,
      "grad_norm": 1.0570423603057861,
      "learning_rate": 0.00016868143955494903,
      "loss": 3.4072,
      "step": 167990
    },
    {
      "epoch": 0.15660318372874277,
      "grad_norm": 1.1652357578277588,
      "learning_rate": 0.00016867957523285384,
      "loss": 3.4921,
      "step": 168000
    },
    {
      "epoch": 0.15661250534682186,
      "grad_norm": 1.0267168283462524,
      "learning_rate": 0.00016867771091075864,
      "loss": 3.4154,
      "step": 168010
    },
    {
      "epoch": 0.15662182696490096,
      "grad_norm": 1.0960423946380615,
      "learning_rate": 0.00016867584658866345,
      "loss": 3.621,
      "step": 168020
    },
    {
      "epoch": 0.15663114858298005,
      "grad_norm": 0.9852246046066284,
      "learning_rate": 0.00016867398226656823,
      "loss": 3.5727,
      "step": 168030
    },
    {
      "epoch": 0.15664047020105915,
      "grad_norm": 0.9759939312934875,
      "learning_rate": 0.00016867211794447304,
      "loss": 3.5902,
      "step": 168040
    },
    {
      "epoch": 0.15664979181913824,
      "grad_norm": 1.2184194326400757,
      "learning_rate": 0.00016867025362237784,
      "loss": 3.6697,
      "step": 168050
    },
    {
      "epoch": 0.15665911343721733,
      "grad_norm": 1.144863247871399,
      "learning_rate": 0.00016866838930028265,
      "loss": 3.5815,
      "step": 168060
    },
    {
      "epoch": 0.15666843505529643,
      "grad_norm": 1.0698394775390625,
      "learning_rate": 0.00016866652497818746,
      "loss": 3.5936,
      "step": 168070
    },
    {
      "epoch": 0.15667775667337552,
      "grad_norm": 1.036157488822937,
      "learning_rate": 0.00016866466065609223,
      "loss": 3.4966,
      "step": 168080
    },
    {
      "epoch": 0.15668707829145462,
      "grad_norm": 1.106661319732666,
      "learning_rate": 0.00016866279633399704,
      "loss": 3.7118,
      "step": 168090
    },
    {
      "epoch": 0.15669639990953368,
      "grad_norm": 1.1474627256393433,
      "learning_rate": 0.00016866093201190185,
      "loss": 3.7253,
      "step": 168100
    },
    {
      "epoch": 0.15670572152761278,
      "grad_norm": 1.066663146018982,
      "learning_rate": 0.00016865906768980663,
      "loss": 3.5172,
      "step": 168110
    },
    {
      "epoch": 0.15671504314569187,
      "grad_norm": 1.018457293510437,
      "learning_rate": 0.00016865720336771146,
      "loss": 3.5394,
      "step": 168120
    },
    {
      "epoch": 0.15672436476377097,
      "grad_norm": 1.1301532983779907,
      "learning_rate": 0.00016865533904561624,
      "loss": 3.5371,
      "step": 168130
    },
    {
      "epoch": 0.15673368638185006,
      "grad_norm": 1.0586998462677002,
      "learning_rate": 0.00016865347472352105,
      "loss": 3.5887,
      "step": 168140
    },
    {
      "epoch": 0.15674300799992916,
      "grad_norm": 0.9868530631065369,
      "learning_rate": 0.00016865161040142585,
      "loss": 3.6038,
      "step": 168150
    },
    {
      "epoch": 0.15675232961800825,
      "grad_norm": 1.2408615350723267,
      "learning_rate": 0.00016864974607933063,
      "loss": 3.5693,
      "step": 168160
    },
    {
      "epoch": 0.15676165123608735,
      "grad_norm": 0.9692214727401733,
      "learning_rate": 0.00016864788175723544,
      "loss": 3.4727,
      "step": 168170
    },
    {
      "epoch": 0.15677097285416644,
      "grad_norm": 1.0247615575790405,
      "learning_rate": 0.00016864601743514024,
      "loss": 3.5208,
      "step": 168180
    },
    {
      "epoch": 0.15678029447224553,
      "grad_norm": 1.08306884765625,
      "learning_rate": 0.00016864415311304505,
      "loss": 3.5421,
      "step": 168190
    },
    {
      "epoch": 0.1567896160903246,
      "grad_norm": 0.9800275564193726,
      "learning_rate": 0.00016864228879094986,
      "loss": 3.5587,
      "step": 168200
    },
    {
      "epoch": 0.1567989377084037,
      "grad_norm": 1.0903624296188354,
      "learning_rate": 0.00016864042446885464,
      "loss": 3.66,
      "step": 168210
    },
    {
      "epoch": 0.1568082593264828,
      "grad_norm": 1.2369683980941772,
      "learning_rate": 0.00016863856014675944,
      "loss": 3.503,
      "step": 168220
    },
    {
      "epoch": 0.15681758094456189,
      "grad_norm": 1.1651920080184937,
      "learning_rate": 0.00016863669582466425,
      "loss": 3.5126,
      "step": 168230
    },
    {
      "epoch": 0.15682690256264098,
      "grad_norm": 1.1231839656829834,
      "learning_rate": 0.00016863483150256905,
      "loss": 3.5114,
      "step": 168240
    },
    {
      "epoch": 0.15683622418072007,
      "grad_norm": 1.1056983470916748,
      "learning_rate": 0.00016863296718047383,
      "loss": 3.5105,
      "step": 168250
    },
    {
      "epoch": 0.15684554579879917,
      "grad_norm": 1.1744309663772583,
      "learning_rate": 0.00016863110285837864,
      "loss": 3.5829,
      "step": 168260
    },
    {
      "epoch": 0.15685486741687826,
      "grad_norm": 1.0431818962097168,
      "learning_rate": 0.00016862923853628345,
      "loss": 3.5722,
      "step": 168270
    },
    {
      "epoch": 0.15686418903495736,
      "grad_norm": 1.1520222425460815,
      "learning_rate": 0.00016862737421418823,
      "loss": 3.5841,
      "step": 168280
    },
    {
      "epoch": 0.15687351065303645,
      "grad_norm": 1.0329238176345825,
      "learning_rate": 0.00016862550989209306,
      "loss": 3.6307,
      "step": 168290
    },
    {
      "epoch": 0.15688283227111555,
      "grad_norm": 1.050541877746582,
      "learning_rate": 0.00016862364556999784,
      "loss": 3.674,
      "step": 168300
    },
    {
      "epoch": 0.1568921538891946,
      "grad_norm": 1.0277284383773804,
      "learning_rate": 0.00016862178124790264,
      "loss": 3.5436,
      "step": 168310
    },
    {
      "epoch": 0.1569014755072737,
      "grad_norm": 1.1098649501800537,
      "learning_rate": 0.00016861991692580745,
      "loss": 3.471,
      "step": 168320
    },
    {
      "epoch": 0.1569107971253528,
      "grad_norm": 1.1108511686325073,
      "learning_rate": 0.00016861805260371223,
      "loss": 3.4073,
      "step": 168330
    },
    {
      "epoch": 0.1569201187434319,
      "grad_norm": 1.1243044137954712,
      "learning_rate": 0.00016861618828161704,
      "loss": 3.5363,
      "step": 168340
    },
    {
      "epoch": 0.156929440361511,
      "grad_norm": 1.152711272239685,
      "learning_rate": 0.00016861432395952184,
      "loss": 3.6306,
      "step": 168350
    },
    {
      "epoch": 0.15693876197959009,
      "grad_norm": 1.1980854272842407,
      "learning_rate": 0.00016861245963742665,
      "loss": 3.7201,
      "step": 168360
    },
    {
      "epoch": 0.15694808359766918,
      "grad_norm": 1.0963906049728394,
      "learning_rate": 0.00016861059531533145,
      "loss": 3.6019,
      "step": 168370
    },
    {
      "epoch": 0.15695740521574827,
      "grad_norm": 1.0281740427017212,
      "learning_rate": 0.00016860873099323623,
      "loss": 3.4272,
      "step": 168380
    },
    {
      "epoch": 0.15696672683382737,
      "grad_norm": 1.066220760345459,
      "learning_rate": 0.00016860686667114104,
      "loss": 3.6373,
      "step": 168390
    },
    {
      "epoch": 0.15697604845190646,
      "grad_norm": 1.0720444917678833,
      "learning_rate": 0.00016860500234904585,
      "loss": 3.5715,
      "step": 168400
    },
    {
      "epoch": 0.15698537006998553,
      "grad_norm": 1.0779533386230469,
      "learning_rate": 0.00016860313802695065,
      "loss": 3.6067,
      "step": 168410
    },
    {
      "epoch": 0.15699469168806462,
      "grad_norm": 0.8990948796272278,
      "learning_rate": 0.00016860127370485546,
      "loss": 3.6974,
      "step": 168420
    },
    {
      "epoch": 0.15700401330614372,
      "grad_norm": 0.9269053339958191,
      "learning_rate": 0.00016859940938276024,
      "loss": 3.4332,
      "step": 168430
    },
    {
      "epoch": 0.1570133349242228,
      "grad_norm": 0.890324592590332,
      "learning_rate": 0.00016859754506066504,
      "loss": 3.5971,
      "step": 168440
    },
    {
      "epoch": 0.1570226565423019,
      "grad_norm": 0.9665846824645996,
      "learning_rate": 0.00016859568073856985,
      "loss": 3.6742,
      "step": 168450
    },
    {
      "epoch": 0.157031978160381,
      "grad_norm": 1.0092519521713257,
      "learning_rate": 0.00016859381641647466,
      "loss": 3.5734,
      "step": 168460
    },
    {
      "epoch": 0.1570412997784601,
      "grad_norm": 1.0161865949630737,
      "learning_rate": 0.00016859195209437946,
      "loss": 3.4904,
      "step": 168470
    },
    {
      "epoch": 0.1570506213965392,
      "grad_norm": 1.1404372453689575,
      "learning_rate": 0.00016859008777228424,
      "loss": 3.406,
      "step": 168480
    },
    {
      "epoch": 0.1570599430146183,
      "grad_norm": 0.7286165952682495,
      "learning_rate": 0.00016858822345018905,
      "loss": 3.7722,
      "step": 168490
    },
    {
      "epoch": 0.15706926463269738,
      "grad_norm": 0.9858719110488892,
      "learning_rate": 0.00016858635912809383,
      "loss": 3.6102,
      "step": 168500
    },
    {
      "epoch": 0.15707858625077648,
      "grad_norm": 1.0405864715576172,
      "learning_rate": 0.00016858449480599866,
      "loss": 3.5335,
      "step": 168510
    },
    {
      "epoch": 0.15708790786885554,
      "grad_norm": 1.0109542608261108,
      "learning_rate": 0.00016858263048390347,
      "loss": 3.4921,
      "step": 168520
    },
    {
      "epoch": 0.15709722948693464,
      "grad_norm": 0.9849492311477661,
      "learning_rate": 0.00016858076616180825,
      "loss": 3.6155,
      "step": 168530
    },
    {
      "epoch": 0.15710655110501373,
      "grad_norm": 1.0043874979019165,
      "learning_rate": 0.00016857890183971305,
      "loss": 3.5631,
      "step": 168540
    },
    {
      "epoch": 0.15711587272309283,
      "grad_norm": 1.1416664123535156,
      "learning_rate": 0.00016857703751761783,
      "loss": 3.6242,
      "step": 168550
    },
    {
      "epoch": 0.15712519434117192,
      "grad_norm": 0.923664927482605,
      "learning_rate": 0.00016857517319552264,
      "loss": 3.7021,
      "step": 168560
    },
    {
      "epoch": 0.15713451595925101,
      "grad_norm": 1.043041706085205,
      "learning_rate": 0.00016857330887342747,
      "loss": 3.532,
      "step": 168570
    },
    {
      "epoch": 0.1571438375773301,
      "grad_norm": 1.1185216903686523,
      "learning_rate": 0.00016857144455133225,
      "loss": 3.7457,
      "step": 168580
    },
    {
      "epoch": 0.1571531591954092,
      "grad_norm": 1.2446284294128418,
      "learning_rate": 0.00016856958022923706,
      "loss": 3.6465,
      "step": 168590
    },
    {
      "epoch": 0.1571624808134883,
      "grad_norm": 1.100510597229004,
      "learning_rate": 0.00016856771590714184,
      "loss": 3.5888,
      "step": 168600
    },
    {
      "epoch": 0.1571718024315674,
      "grad_norm": 1.1294853687286377,
      "learning_rate": 0.00016856585158504664,
      "loss": 3.6038,
      "step": 168610
    },
    {
      "epoch": 0.15718112404964646,
      "grad_norm": 1.069553017616272,
      "learning_rate": 0.00016856398726295145,
      "loss": 3.6127,
      "step": 168620
    },
    {
      "epoch": 0.15719044566772555,
      "grad_norm": 1.0464401245117188,
      "learning_rate": 0.00016856212294085626,
      "loss": 3.6548,
      "step": 168630
    },
    {
      "epoch": 0.15719976728580465,
      "grad_norm": 1.0464298725128174,
      "learning_rate": 0.00016856025861876106,
      "loss": 3.5193,
      "step": 168640
    },
    {
      "epoch": 0.15720908890388374,
      "grad_norm": 0.9075007438659668,
      "learning_rate": 0.00016855839429666584,
      "loss": 3.4977,
      "step": 168650
    },
    {
      "epoch": 0.15721841052196284,
      "grad_norm": 1.0760070085525513,
      "learning_rate": 0.00016855652997457065,
      "loss": 3.5418,
      "step": 168660
    },
    {
      "epoch": 0.15722773214004193,
      "grad_norm": 1.1172363758087158,
      "learning_rate": 0.00016855466565247545,
      "loss": 3.6824,
      "step": 168670
    },
    {
      "epoch": 0.15723705375812103,
      "grad_norm": 0.9645384550094604,
      "learning_rate": 0.00016855280133038026,
      "loss": 3.407,
      "step": 168680
    },
    {
      "epoch": 0.15724637537620012,
      "grad_norm": 1.0190083980560303,
      "learning_rate": 0.00016855093700828507,
      "loss": 3.336,
      "step": 168690
    },
    {
      "epoch": 0.15725569699427921,
      "grad_norm": 1.0902420282363892,
      "learning_rate": 0.00016854907268618985,
      "loss": 3.6258,
      "step": 168700
    },
    {
      "epoch": 0.1572650186123583,
      "grad_norm": 1.1518645286560059,
      "learning_rate": 0.00016854720836409465,
      "loss": 3.6539,
      "step": 168710
    },
    {
      "epoch": 0.1572743402304374,
      "grad_norm": 1.2252658605575562,
      "learning_rate": 0.00016854534404199946,
      "loss": 3.4027,
      "step": 168720
    },
    {
      "epoch": 0.15728366184851647,
      "grad_norm": 1.1847801208496094,
      "learning_rate": 0.00016854347971990424,
      "loss": 3.4759,
      "step": 168730
    },
    {
      "epoch": 0.15729298346659557,
      "grad_norm": 1.1550167798995972,
      "learning_rate": 0.00016854161539780907,
      "loss": 3.509,
      "step": 168740
    },
    {
      "epoch": 0.15730230508467466,
      "grad_norm": 1.0895516872406006,
      "learning_rate": 0.00016853975107571385,
      "loss": 3.5093,
      "step": 168750
    },
    {
      "epoch": 0.15731162670275375,
      "grad_norm": 1.036055326461792,
      "learning_rate": 0.00016853788675361866,
      "loss": 3.6011,
      "step": 168760
    },
    {
      "epoch": 0.15732094832083285,
      "grad_norm": 1.0261415243148804,
      "learning_rate": 0.00016853602243152346,
      "loss": 3.4818,
      "step": 168770
    },
    {
      "epoch": 0.15733026993891194,
      "grad_norm": 0.9361350536346436,
      "learning_rate": 0.00016853415810942824,
      "loss": 3.6915,
      "step": 168780
    },
    {
      "epoch": 0.15733959155699104,
      "grad_norm": 1.1771818399429321,
      "learning_rate": 0.00016853229378733305,
      "loss": 3.5864,
      "step": 168790
    },
    {
      "epoch": 0.15734891317507013,
      "grad_norm": 1.203571081161499,
      "learning_rate": 0.00016853042946523786,
      "loss": 3.5442,
      "step": 168800
    },
    {
      "epoch": 0.15735823479314923,
      "grad_norm": 1.0854110717773438,
      "learning_rate": 0.00016852856514314266,
      "loss": 3.3861,
      "step": 168810
    },
    {
      "epoch": 0.15736755641122832,
      "grad_norm": 1.025126338005066,
      "learning_rate": 0.00016852670082104747,
      "loss": 3.6312,
      "step": 168820
    },
    {
      "epoch": 0.1573768780293074,
      "grad_norm": 1.0086551904678345,
      "learning_rate": 0.00016852483649895225,
      "loss": 3.5906,
      "step": 168830
    },
    {
      "epoch": 0.15738619964738648,
      "grad_norm": 1.0251684188842773,
      "learning_rate": 0.00016852297217685705,
      "loss": 3.5082,
      "step": 168840
    },
    {
      "epoch": 0.15739552126546558,
      "grad_norm": 1.161312222480774,
      "learning_rate": 0.00016852110785476186,
      "loss": 3.5919,
      "step": 168850
    },
    {
      "epoch": 0.15740484288354467,
      "grad_norm": 0.9056207537651062,
      "learning_rate": 0.00016851924353266667,
      "loss": 3.5114,
      "step": 168860
    },
    {
      "epoch": 0.15741416450162377,
      "grad_norm": 0.8266432285308838,
      "learning_rate": 0.00016851737921057147,
      "loss": 3.4001,
      "step": 168870
    },
    {
      "epoch": 0.15742348611970286,
      "grad_norm": 1.340043067932129,
      "learning_rate": 0.00016851551488847625,
      "loss": 3.5038,
      "step": 168880
    },
    {
      "epoch": 0.15743280773778195,
      "grad_norm": 1.0974502563476562,
      "learning_rate": 0.00016851365056638106,
      "loss": 3.5024,
      "step": 168890
    },
    {
      "epoch": 0.15744212935586105,
      "grad_norm": 1.2464815378189087,
      "learning_rate": 0.00016851178624428586,
      "loss": 3.676,
      "step": 168900
    },
    {
      "epoch": 0.15745145097394014,
      "grad_norm": 0.9912834167480469,
      "learning_rate": 0.00016850992192219067,
      "loss": 3.6544,
      "step": 168910
    },
    {
      "epoch": 0.15746077259201924,
      "grad_norm": 1.0129555463790894,
      "learning_rate": 0.00016850805760009548,
      "loss": 3.6356,
      "step": 168920
    },
    {
      "epoch": 0.1574700942100983,
      "grad_norm": 1.0501925945281982,
      "learning_rate": 0.00016850619327800026,
      "loss": 3.6396,
      "step": 168930
    },
    {
      "epoch": 0.1574794158281774,
      "grad_norm": 1.1432533264160156,
      "learning_rate": 0.00016850432895590506,
      "loss": 3.4619,
      "step": 168940
    },
    {
      "epoch": 0.1574887374462565,
      "grad_norm": 1.272882342338562,
      "learning_rate": 0.00016850246463380984,
      "loss": 3.3992,
      "step": 168950
    },
    {
      "epoch": 0.1574980590643356,
      "grad_norm": 1.0182135105133057,
      "learning_rate": 0.00016850060031171467,
      "loss": 3.6652,
      "step": 168960
    },
    {
      "epoch": 0.15750738068241468,
      "grad_norm": 0.9604377150535583,
      "learning_rate": 0.00016849873598961948,
      "loss": 3.5951,
      "step": 168970
    },
    {
      "epoch": 0.15751670230049378,
      "grad_norm": 1.0462690591812134,
      "learning_rate": 0.00016849687166752426,
      "loss": 3.6146,
      "step": 168980
    },
    {
      "epoch": 0.15752602391857287,
      "grad_norm": 0.8529119491577148,
      "learning_rate": 0.00016849500734542907,
      "loss": 3.6397,
      "step": 168990
    },
    {
      "epoch": 0.15753534553665197,
      "grad_norm": 1.1530797481536865,
      "learning_rate": 0.00016849314302333385,
      "loss": 3.4762,
      "step": 169000
    },
    {
      "epoch": 0.15754466715473106,
      "grad_norm": 1.2714109420776367,
      "learning_rate": 0.00016849127870123865,
      "loss": 3.4422,
      "step": 169010
    },
    {
      "epoch": 0.15755398877281016,
      "grad_norm": 1.0752369165420532,
      "learning_rate": 0.00016848941437914349,
      "loss": 3.49,
      "step": 169020
    },
    {
      "epoch": 0.15756331039088925,
      "grad_norm": 1.1036547422409058,
      "learning_rate": 0.00016848755005704826,
      "loss": 3.3367,
      "step": 169030
    },
    {
      "epoch": 0.15757263200896832,
      "grad_norm": 1.0364947319030762,
      "learning_rate": 0.00016848568573495307,
      "loss": 3.5983,
      "step": 169040
    },
    {
      "epoch": 0.1575819536270474,
      "grad_norm": 0.9017550945281982,
      "learning_rate": 0.00016848382141285785,
      "loss": 3.4852,
      "step": 169050
    },
    {
      "epoch": 0.1575912752451265,
      "grad_norm": 1.2042161226272583,
      "learning_rate": 0.00016848195709076266,
      "loss": 3.572,
      "step": 169060
    },
    {
      "epoch": 0.1576005968632056,
      "grad_norm": 1.1462074518203735,
      "learning_rate": 0.00016848009276866746,
      "loss": 3.5152,
      "step": 169070
    },
    {
      "epoch": 0.1576099184812847,
      "grad_norm": 1.1544300317764282,
      "learning_rate": 0.00016847822844657227,
      "loss": 3.5038,
      "step": 169080
    },
    {
      "epoch": 0.1576192400993638,
      "grad_norm": 1.1816120147705078,
      "learning_rate": 0.00016847636412447708,
      "loss": 3.492,
      "step": 169090
    },
    {
      "epoch": 0.15762856171744288,
      "grad_norm": 1.0266832113265991,
      "learning_rate": 0.00016847449980238185,
      "loss": 3.533,
      "step": 169100
    },
    {
      "epoch": 0.15763788333552198,
      "grad_norm": 1.1169637441635132,
      "learning_rate": 0.00016847263548028666,
      "loss": 3.521,
      "step": 169110
    },
    {
      "epoch": 0.15764720495360107,
      "grad_norm": 1.4573670625686646,
      "learning_rate": 0.00016847077115819147,
      "loss": 3.5628,
      "step": 169120
    },
    {
      "epoch": 0.15765652657168017,
      "grad_norm": 1.190260648727417,
      "learning_rate": 0.00016846890683609627,
      "loss": 3.5594,
      "step": 169130
    },
    {
      "epoch": 0.15766584818975923,
      "grad_norm": 1.097267508506775,
      "learning_rate": 0.00016846704251400108,
      "loss": 3.6778,
      "step": 169140
    },
    {
      "epoch": 0.15767516980783833,
      "grad_norm": 1.3606131076812744,
      "learning_rate": 0.00016846517819190586,
      "loss": 3.6679,
      "step": 169150
    },
    {
      "epoch": 0.15768449142591742,
      "grad_norm": 1.1518476009368896,
      "learning_rate": 0.00016846331386981067,
      "loss": 3.5384,
      "step": 169160
    },
    {
      "epoch": 0.15769381304399652,
      "grad_norm": 0.8475054502487183,
      "learning_rate": 0.00016846144954771547,
      "loss": 3.5192,
      "step": 169170
    },
    {
      "epoch": 0.1577031346620756,
      "grad_norm": 1.0843318700790405,
      "learning_rate": 0.00016845958522562028,
      "loss": 3.496,
      "step": 169180
    },
    {
      "epoch": 0.1577124562801547,
      "grad_norm": 0.9682395458221436,
      "learning_rate": 0.00016845772090352508,
      "loss": 3.6461,
      "step": 169190
    },
    {
      "epoch": 0.1577217778982338,
      "grad_norm": 0.9370889663696289,
      "learning_rate": 0.00016845585658142986,
      "loss": 3.5246,
      "step": 169200
    },
    {
      "epoch": 0.1577310995163129,
      "grad_norm": 0.9368862509727478,
      "learning_rate": 0.00016845399225933467,
      "loss": 3.5929,
      "step": 169210
    },
    {
      "epoch": 0.157740421134392,
      "grad_norm": 1.0608994960784912,
      "learning_rate": 0.00016845212793723948,
      "loss": 3.5154,
      "step": 169220
    },
    {
      "epoch": 0.15774974275247108,
      "grad_norm": 1.0837854146957397,
      "learning_rate": 0.00016845026361514426,
      "loss": 3.7392,
      "step": 169230
    },
    {
      "epoch": 0.15775906437055018,
      "grad_norm": 1.302369236946106,
      "learning_rate": 0.00016844839929304906,
      "loss": 3.5941,
      "step": 169240
    },
    {
      "epoch": 0.15776838598862925,
      "grad_norm": 1.0254818201065063,
      "learning_rate": 0.00016844653497095387,
      "loss": 3.5446,
      "step": 169250
    },
    {
      "epoch": 0.15777770760670834,
      "grad_norm": 1.0989291667938232,
      "learning_rate": 0.00016844467064885867,
      "loss": 3.7184,
      "step": 169260
    },
    {
      "epoch": 0.15778702922478743,
      "grad_norm": 0.9692838191986084,
      "learning_rate": 0.00016844280632676348,
      "loss": 3.5475,
      "step": 169270
    },
    {
      "epoch": 0.15779635084286653,
      "grad_norm": 1.0363025665283203,
      "learning_rate": 0.00016844094200466826,
      "loss": 3.6342,
      "step": 169280
    },
    {
      "epoch": 0.15780567246094562,
      "grad_norm": 1.1265283823013306,
      "learning_rate": 0.00016843907768257307,
      "loss": 3.614,
      "step": 169290
    },
    {
      "epoch": 0.15781499407902472,
      "grad_norm": 0.8779549598693848,
      "learning_rate": 0.00016843721336047787,
      "loss": 3.625,
      "step": 169300
    },
    {
      "epoch": 0.1578243156971038,
      "grad_norm": 1.1929082870483398,
      "learning_rate": 0.00016843534903838268,
      "loss": 3.5886,
      "step": 169310
    },
    {
      "epoch": 0.1578336373151829,
      "grad_norm": 1.2793543338775635,
      "learning_rate": 0.00016843348471628748,
      "loss": 3.59,
      "step": 169320
    },
    {
      "epoch": 0.157842958933262,
      "grad_norm": 0.8785250186920166,
      "learning_rate": 0.00016843162039419226,
      "loss": 3.578,
      "step": 169330
    },
    {
      "epoch": 0.1578522805513411,
      "grad_norm": 1.0853025913238525,
      "learning_rate": 0.00016842975607209707,
      "loss": 3.7435,
      "step": 169340
    },
    {
      "epoch": 0.15786160216942016,
      "grad_norm": 0.9279594421386719,
      "learning_rate": 0.00016842789175000188,
      "loss": 3.5824,
      "step": 169350
    },
    {
      "epoch": 0.15787092378749926,
      "grad_norm": 1.1747394800186157,
      "learning_rate": 0.00016842602742790668,
      "loss": 3.5524,
      "step": 169360
    },
    {
      "epoch": 0.15788024540557835,
      "grad_norm": 1.1109861135482788,
      "learning_rate": 0.0001684241631058115,
      "loss": 3.6014,
      "step": 169370
    },
    {
      "epoch": 0.15788956702365745,
      "grad_norm": 0.9629307985305786,
      "learning_rate": 0.00016842229878371627,
      "loss": 3.4632,
      "step": 169380
    },
    {
      "epoch": 0.15789888864173654,
      "grad_norm": 1.1010632514953613,
      "learning_rate": 0.00016842043446162107,
      "loss": 3.603,
      "step": 169390
    },
    {
      "epoch": 0.15790821025981563,
      "grad_norm": 1.1228992938995361,
      "learning_rate": 0.00016841857013952585,
      "loss": 3.5431,
      "step": 169400
    },
    {
      "epoch": 0.15791753187789473,
      "grad_norm": 1.2442147731781006,
      "learning_rate": 0.0001684167058174307,
      "loss": 3.4705,
      "step": 169410
    },
    {
      "epoch": 0.15792685349597382,
      "grad_norm": 0.9396917819976807,
      "learning_rate": 0.0001684148414953355,
      "loss": 3.5493,
      "step": 169420
    },
    {
      "epoch": 0.15793617511405292,
      "grad_norm": 1.0203660726547241,
      "learning_rate": 0.00016841297717324027,
      "loss": 3.6674,
      "step": 169430
    },
    {
      "epoch": 0.157945496732132,
      "grad_norm": 0.9181791543960571,
      "learning_rate": 0.00016841111285114508,
      "loss": 3.5466,
      "step": 169440
    },
    {
      "epoch": 0.1579548183502111,
      "grad_norm": 1.1391956806182861,
      "learning_rate": 0.00016840924852904986,
      "loss": 3.6212,
      "step": 169450
    },
    {
      "epoch": 0.15796413996829017,
      "grad_norm": 0.9750809669494629,
      "learning_rate": 0.00016840738420695466,
      "loss": 3.5999,
      "step": 169460
    },
    {
      "epoch": 0.15797346158636927,
      "grad_norm": 1.0013364553451538,
      "learning_rate": 0.0001684055198848595,
      "loss": 3.5832,
      "step": 169470
    },
    {
      "epoch": 0.15798278320444836,
      "grad_norm": 0.9132916331291199,
      "learning_rate": 0.00016840365556276428,
      "loss": 3.708,
      "step": 169480
    },
    {
      "epoch": 0.15799210482252746,
      "grad_norm": 0.9289125204086304,
      "learning_rate": 0.00016840179124066908,
      "loss": 3.8011,
      "step": 169490
    },
    {
      "epoch": 0.15800142644060655,
      "grad_norm": 1.0415385961532593,
      "learning_rate": 0.00016839992691857386,
      "loss": 3.5526,
      "step": 169500
    },
    {
      "epoch": 0.15801074805868565,
      "grad_norm": 1.0601282119750977,
      "learning_rate": 0.00016839806259647867,
      "loss": 3.4531,
      "step": 169510
    },
    {
      "epoch": 0.15802006967676474,
      "grad_norm": 1.193750262260437,
      "learning_rate": 0.00016839619827438348,
      "loss": 3.5574,
      "step": 169520
    },
    {
      "epoch": 0.15802939129484384,
      "grad_norm": 1.1287578344345093,
      "learning_rate": 0.00016839433395228828,
      "loss": 3.6137,
      "step": 169530
    },
    {
      "epoch": 0.15803871291292293,
      "grad_norm": 1.0470689535140991,
      "learning_rate": 0.0001683924696301931,
      "loss": 3.4297,
      "step": 169540
    },
    {
      "epoch": 0.15804803453100202,
      "grad_norm": 1.125144124031067,
      "learning_rate": 0.00016839060530809787,
      "loss": 3.6066,
      "step": 169550
    },
    {
      "epoch": 0.1580573561490811,
      "grad_norm": 1.0747116804122925,
      "learning_rate": 0.00016838874098600267,
      "loss": 3.7102,
      "step": 169560
    },
    {
      "epoch": 0.15806667776716019,
      "grad_norm": 1.2396787405014038,
      "learning_rate": 0.00016838687666390748,
      "loss": 3.5046,
      "step": 169570
    },
    {
      "epoch": 0.15807599938523928,
      "grad_norm": 1.1564439535140991,
      "learning_rate": 0.00016838501234181229,
      "loss": 3.5077,
      "step": 169580
    },
    {
      "epoch": 0.15808532100331837,
      "grad_norm": 1.0299102067947388,
      "learning_rate": 0.0001683831480197171,
      "loss": 3.626,
      "step": 169590
    },
    {
      "epoch": 0.15809464262139747,
      "grad_norm": 0.9021351337432861,
      "learning_rate": 0.00016838128369762187,
      "loss": 3.4539,
      "step": 169600
    },
    {
      "epoch": 0.15810396423947656,
      "grad_norm": 1.3279634714126587,
      "learning_rate": 0.00016837941937552668,
      "loss": 3.4311,
      "step": 169610
    },
    {
      "epoch": 0.15811328585755566,
      "grad_norm": 1.0563277006149292,
      "learning_rate": 0.00016837755505343146,
      "loss": 3.6032,
      "step": 169620
    },
    {
      "epoch": 0.15812260747563475,
      "grad_norm": 1.2882990837097168,
      "learning_rate": 0.0001683756907313363,
      "loss": 3.3576,
      "step": 169630
    },
    {
      "epoch": 0.15813192909371385,
      "grad_norm": 1.076720952987671,
      "learning_rate": 0.0001683738264092411,
      "loss": 3.6323,
      "step": 169640
    },
    {
      "epoch": 0.15814125071179294,
      "grad_norm": 1.1689305305480957,
      "learning_rate": 0.00016837196208714588,
      "loss": 3.6479,
      "step": 169650
    },
    {
      "epoch": 0.15815057232987204,
      "grad_norm": 1.1891001462936401,
      "learning_rate": 0.00016837009776505068,
      "loss": 3.6204,
      "step": 169660
    },
    {
      "epoch": 0.1581598939479511,
      "grad_norm": 0.9530740976333618,
      "learning_rate": 0.00016836823344295546,
      "loss": 3.4682,
      "step": 169670
    },
    {
      "epoch": 0.1581692155660302,
      "grad_norm": 1.1861475706100464,
      "learning_rate": 0.00016836636912086027,
      "loss": 3.3421,
      "step": 169680
    },
    {
      "epoch": 0.1581785371841093,
      "grad_norm": 1.2332007884979248,
      "learning_rate": 0.0001683645047987651,
      "loss": 3.5674,
      "step": 169690
    },
    {
      "epoch": 0.15818785880218839,
      "grad_norm": 1.1539981365203857,
      "learning_rate": 0.00016836264047666988,
      "loss": 3.587,
      "step": 169700
    },
    {
      "epoch": 0.15819718042026748,
      "grad_norm": 1.2389142513275146,
      "learning_rate": 0.0001683607761545747,
      "loss": 3.6359,
      "step": 169710
    },
    {
      "epoch": 0.15820650203834657,
      "grad_norm": 0.9548004865646362,
      "learning_rate": 0.00016835891183247947,
      "loss": 3.4347,
      "step": 169720
    },
    {
      "epoch": 0.15821582365642567,
      "grad_norm": 1.0777466297149658,
      "learning_rate": 0.00016835704751038427,
      "loss": 3.5135,
      "step": 169730
    },
    {
      "epoch": 0.15822514527450476,
      "grad_norm": 1.1595587730407715,
      "learning_rate": 0.00016835518318828908,
      "loss": 3.5613,
      "step": 169740
    },
    {
      "epoch": 0.15823446689258386,
      "grad_norm": 0.9870352745056152,
      "learning_rate": 0.00016835331886619389,
      "loss": 3.4119,
      "step": 169750
    },
    {
      "epoch": 0.15824378851066295,
      "grad_norm": 1.1288915872573853,
      "learning_rate": 0.0001683514545440987,
      "loss": 3.1847,
      "step": 169760
    },
    {
      "epoch": 0.15825311012874202,
      "grad_norm": 0.9805331826210022,
      "learning_rate": 0.00016834959022200347,
      "loss": 3.2844,
      "step": 169770
    },
    {
      "epoch": 0.15826243174682111,
      "grad_norm": 1.1659077405929565,
      "learning_rate": 0.00016834772589990828,
      "loss": 3.5029,
      "step": 169780
    },
    {
      "epoch": 0.1582717533649002,
      "grad_norm": 1.1051678657531738,
      "learning_rate": 0.00016834586157781308,
      "loss": 3.4785,
      "step": 169790
    },
    {
      "epoch": 0.1582810749829793,
      "grad_norm": 0.9892143607139587,
      "learning_rate": 0.0001683439972557179,
      "loss": 3.5783,
      "step": 169800
    },
    {
      "epoch": 0.1582903966010584,
      "grad_norm": 1.1795499324798584,
      "learning_rate": 0.0001683421329336227,
      "loss": 3.5452,
      "step": 169810
    },
    {
      "epoch": 0.1582997182191375,
      "grad_norm": 1.08574378490448,
      "learning_rate": 0.00016834026861152748,
      "loss": 3.5566,
      "step": 169820
    },
    {
      "epoch": 0.1583090398372166,
      "grad_norm": 0.9694989919662476,
      "learning_rate": 0.00016833840428943228,
      "loss": 3.5905,
      "step": 169830
    },
    {
      "epoch": 0.15831836145529568,
      "grad_norm": 1.088136911392212,
      "learning_rate": 0.0001683365399673371,
      "loss": 3.7902,
      "step": 169840
    },
    {
      "epoch": 0.15832768307337478,
      "grad_norm": 0.9264545440673828,
      "learning_rate": 0.00016833467564524187,
      "loss": 3.7536,
      "step": 169850
    },
    {
      "epoch": 0.15833700469145387,
      "grad_norm": 1.058925986289978,
      "learning_rate": 0.0001683328113231467,
      "loss": 3.6023,
      "step": 169860
    },
    {
      "epoch": 0.15834632630953296,
      "grad_norm": 1.169055700302124,
      "learning_rate": 0.00016833094700105148,
      "loss": 3.5057,
      "step": 169870
    },
    {
      "epoch": 0.15835564792761203,
      "grad_norm": 1.023472547531128,
      "learning_rate": 0.00016832908267895629,
      "loss": 3.6487,
      "step": 169880
    },
    {
      "epoch": 0.15836496954569113,
      "grad_norm": 1.1804293394088745,
      "learning_rate": 0.0001683272183568611,
      "loss": 3.5057,
      "step": 169890
    },
    {
      "epoch": 0.15837429116377022,
      "grad_norm": 1.0086511373519897,
      "learning_rate": 0.00016832535403476587,
      "loss": 3.6473,
      "step": 169900
    },
    {
      "epoch": 0.15838361278184931,
      "grad_norm": 1.0800986289978027,
      "learning_rate": 0.00016832348971267068,
      "loss": 3.6328,
      "step": 169910
    },
    {
      "epoch": 0.1583929343999284,
      "grad_norm": 1.211591362953186,
      "learning_rate": 0.00016832162539057548,
      "loss": 3.6301,
      "step": 169920
    },
    {
      "epoch": 0.1584022560180075,
      "grad_norm": 0.9508960843086243,
      "learning_rate": 0.0001683197610684803,
      "loss": 3.4978,
      "step": 169930
    },
    {
      "epoch": 0.1584115776360866,
      "grad_norm": 1.1355772018432617,
      "learning_rate": 0.0001683178967463851,
      "loss": 3.3887,
      "step": 169940
    },
    {
      "epoch": 0.1584208992541657,
      "grad_norm": 1.1995220184326172,
      "learning_rate": 0.00016831603242428988,
      "loss": 3.5284,
      "step": 169950
    },
    {
      "epoch": 0.1584302208722448,
      "grad_norm": 1.1559590101242065,
      "learning_rate": 0.00016831416810219468,
      "loss": 3.5655,
      "step": 169960
    },
    {
      "epoch": 0.15843954249032388,
      "grad_norm": 1.230677604675293,
      "learning_rate": 0.0001683123037800995,
      "loss": 3.448,
      "step": 169970
    },
    {
      "epoch": 0.15844886410840295,
      "grad_norm": 0.9613374471664429,
      "learning_rate": 0.0001683104394580043,
      "loss": 3.5973,
      "step": 169980
    },
    {
      "epoch": 0.15845818572648204,
      "grad_norm": 0.8700073957443237,
      "learning_rate": 0.0001683085751359091,
      "loss": 3.5796,
      "step": 169990
    },
    {
      "epoch": 0.15846750734456114,
      "grad_norm": 1.2045832872390747,
      "learning_rate": 0.00016830671081381388,
      "loss": 3.624,
      "step": 170000
    },
    {
      "epoch": 0.15847682896264023,
      "grad_norm": 1.1656681299209595,
      "learning_rate": 0.0001683048464917187,
      "loss": 3.4501,
      "step": 170010
    },
    {
      "epoch": 0.15848615058071933,
      "grad_norm": 1.0070679187774658,
      "learning_rate": 0.0001683029821696235,
      "loss": 3.7008,
      "step": 170020
    },
    {
      "epoch": 0.15849547219879842,
      "grad_norm": 1.0158113241195679,
      "learning_rate": 0.0001683011178475283,
      "loss": 3.4596,
      "step": 170030
    },
    {
      "epoch": 0.15850479381687752,
      "grad_norm": 1.0693289041519165,
      "learning_rate": 0.0001682992535254331,
      "loss": 3.6242,
      "step": 170040
    },
    {
      "epoch": 0.1585141154349566,
      "grad_norm": 0.959679901599884,
      "learning_rate": 0.00016829738920333788,
      "loss": 3.5087,
      "step": 170050
    },
    {
      "epoch": 0.1585234370530357,
      "grad_norm": 1.0011703968048096,
      "learning_rate": 0.0001682955248812427,
      "loss": 3.6504,
      "step": 170060
    },
    {
      "epoch": 0.1585327586711148,
      "grad_norm": 1.0249744653701782,
      "learning_rate": 0.00016829366055914747,
      "loss": 3.3911,
      "step": 170070
    },
    {
      "epoch": 0.1585420802891939,
      "grad_norm": 1.0698096752166748,
      "learning_rate": 0.0001682917962370523,
      "loss": 3.4733,
      "step": 170080
    },
    {
      "epoch": 0.15855140190727296,
      "grad_norm": 1.395029902458191,
      "learning_rate": 0.0001682899319149571,
      "loss": 3.6305,
      "step": 170090
    },
    {
      "epoch": 0.15856072352535205,
      "grad_norm": 0.9572962522506714,
      "learning_rate": 0.0001682880675928619,
      "loss": 3.6106,
      "step": 170100
    },
    {
      "epoch": 0.15857004514343115,
      "grad_norm": 0.9610726833343506,
      "learning_rate": 0.0001682862032707667,
      "loss": 3.4237,
      "step": 170110
    },
    {
      "epoch": 0.15857936676151024,
      "grad_norm": 1.1461433172225952,
      "learning_rate": 0.00016828433894867147,
      "loss": 3.5329,
      "step": 170120
    },
    {
      "epoch": 0.15858868837958934,
      "grad_norm": 1.0566694736480713,
      "learning_rate": 0.00016828247462657628,
      "loss": 3.572,
      "step": 170130
    },
    {
      "epoch": 0.15859800999766843,
      "grad_norm": 0.9126739501953125,
      "learning_rate": 0.00016828061030448111,
      "loss": 3.727,
      "step": 170140
    },
    {
      "epoch": 0.15860733161574753,
      "grad_norm": 1.117365837097168,
      "learning_rate": 0.0001682787459823859,
      "loss": 3.546,
      "step": 170150
    },
    {
      "epoch": 0.15861665323382662,
      "grad_norm": 1.08053457736969,
      "learning_rate": 0.0001682768816602907,
      "loss": 3.4719,
      "step": 170160
    },
    {
      "epoch": 0.15862597485190572,
      "grad_norm": 0.8324137926101685,
      "learning_rate": 0.00016827501733819548,
      "loss": 3.4778,
      "step": 170170
    },
    {
      "epoch": 0.1586352964699848,
      "grad_norm": 1.0479683876037598,
      "learning_rate": 0.00016827315301610029,
      "loss": 3.6227,
      "step": 170180
    },
    {
      "epoch": 0.15864461808806388,
      "grad_norm": 1.1348495483398438,
      "learning_rate": 0.0001682712886940051,
      "loss": 3.5928,
      "step": 170190
    },
    {
      "epoch": 0.15865393970614297,
      "grad_norm": 1.3505595922470093,
      "learning_rate": 0.0001682694243719099,
      "loss": 3.5727,
      "step": 170200
    },
    {
      "epoch": 0.15866326132422207,
      "grad_norm": 0.9256453514099121,
      "learning_rate": 0.0001682675600498147,
      "loss": 3.5819,
      "step": 170210
    },
    {
      "epoch": 0.15867258294230116,
      "grad_norm": 1.034595251083374,
      "learning_rate": 0.00016826569572771948,
      "loss": 3.504,
      "step": 170220
    },
    {
      "epoch": 0.15868190456038025,
      "grad_norm": 1.1402485370635986,
      "learning_rate": 0.0001682638314056243,
      "loss": 3.6379,
      "step": 170230
    },
    {
      "epoch": 0.15869122617845935,
      "grad_norm": 1.1635576486587524,
      "learning_rate": 0.0001682619670835291,
      "loss": 3.6586,
      "step": 170240
    },
    {
      "epoch": 0.15870054779653844,
      "grad_norm": 1.0318022966384888,
      "learning_rate": 0.0001682601027614339,
      "loss": 3.5307,
      "step": 170250
    },
    {
      "epoch": 0.15870986941461754,
      "grad_norm": 1.0402709245681763,
      "learning_rate": 0.0001682582384393387,
      "loss": 3.5615,
      "step": 170260
    },
    {
      "epoch": 0.15871919103269663,
      "grad_norm": 1.1329970359802246,
      "learning_rate": 0.0001682563741172435,
      "loss": 3.515,
      "step": 170270
    },
    {
      "epoch": 0.15872851265077573,
      "grad_norm": 1.1453660726547241,
      "learning_rate": 0.0001682545097951483,
      "loss": 3.6108,
      "step": 170280
    },
    {
      "epoch": 0.15873783426885482,
      "grad_norm": 1.027674674987793,
      "learning_rate": 0.0001682526454730531,
      "loss": 3.5079,
      "step": 170290
    },
    {
      "epoch": 0.1587471558869339,
      "grad_norm": 1.0729587078094482,
      "learning_rate": 0.00016825078115095788,
      "loss": 3.7959,
      "step": 170300
    },
    {
      "epoch": 0.15875647750501298,
      "grad_norm": 1.0171916484832764,
      "learning_rate": 0.0001682489168288627,
      "loss": 3.4248,
      "step": 170310
    },
    {
      "epoch": 0.15876579912309208,
      "grad_norm": 1.1766077280044556,
      "learning_rate": 0.0001682470525067675,
      "loss": 3.6343,
      "step": 170320
    },
    {
      "epoch": 0.15877512074117117,
      "grad_norm": 1.052202820777893,
      "learning_rate": 0.0001682451881846723,
      "loss": 3.7507,
      "step": 170330
    },
    {
      "epoch": 0.15878444235925027,
      "grad_norm": 1.0438326597213745,
      "learning_rate": 0.0001682433238625771,
      "loss": 3.4773,
      "step": 170340
    },
    {
      "epoch": 0.15879376397732936,
      "grad_norm": 0.8565565347671509,
      "learning_rate": 0.00016824145954048188,
      "loss": 3.6945,
      "step": 170350
    },
    {
      "epoch": 0.15880308559540846,
      "grad_norm": 1.2644176483154297,
      "learning_rate": 0.0001682395952183867,
      "loss": 3.5698,
      "step": 170360
    },
    {
      "epoch": 0.15881240721348755,
      "grad_norm": 1.0235543251037598,
      "learning_rate": 0.0001682377308962915,
      "loss": 3.2863,
      "step": 170370
    },
    {
      "epoch": 0.15882172883156664,
      "grad_norm": 1.1066559553146362,
      "learning_rate": 0.0001682358665741963,
      "loss": 3.6615,
      "step": 170380
    },
    {
      "epoch": 0.15883105044964574,
      "grad_norm": 0.9825319647789001,
      "learning_rate": 0.0001682340022521011,
      "loss": 3.3966,
      "step": 170390
    },
    {
      "epoch": 0.1588403720677248,
      "grad_norm": 1.050689697265625,
      "learning_rate": 0.0001682321379300059,
      "loss": 3.6193,
      "step": 170400
    },
    {
      "epoch": 0.1588496936858039,
      "grad_norm": 0.8937214612960815,
      "learning_rate": 0.0001682302736079107,
      "loss": 3.5004,
      "step": 170410
    },
    {
      "epoch": 0.158859015303883,
      "grad_norm": 1.1333101987838745,
      "learning_rate": 0.0001682284092858155,
      "loss": 3.4732,
      "step": 170420
    },
    {
      "epoch": 0.1588683369219621,
      "grad_norm": 1.048125147819519,
      "learning_rate": 0.0001682265449637203,
      "loss": 3.5921,
      "step": 170430
    },
    {
      "epoch": 0.15887765854004118,
      "grad_norm": 1.0281567573547363,
      "learning_rate": 0.00016822468064162511,
      "loss": 3.4794,
      "step": 170440
    },
    {
      "epoch": 0.15888698015812028,
      "grad_norm": 1.0793061256408691,
      "learning_rate": 0.0001682228163195299,
      "loss": 3.5159,
      "step": 170450
    },
    {
      "epoch": 0.15889630177619937,
      "grad_norm": 1.1352448463439941,
      "learning_rate": 0.0001682209519974347,
      "loss": 3.5591,
      "step": 170460
    },
    {
      "epoch": 0.15890562339427847,
      "grad_norm": 0.99876868724823,
      "learning_rate": 0.0001682190876753395,
      "loss": 3.6129,
      "step": 170470
    },
    {
      "epoch": 0.15891494501235756,
      "grad_norm": 1.2597465515136719,
      "learning_rate": 0.0001682172233532443,
      "loss": 3.5639,
      "step": 170480
    },
    {
      "epoch": 0.15892426663043666,
      "grad_norm": 1.2359555959701538,
      "learning_rate": 0.00016821535903114912,
      "loss": 3.4847,
      "step": 170490
    },
    {
      "epoch": 0.15893358824851572,
      "grad_norm": 1.0250427722930908,
      "learning_rate": 0.0001682134947090539,
      "loss": 3.5199,
      "step": 170500
    },
    {
      "epoch": 0.15894290986659482,
      "grad_norm": 1.3409531116485596,
      "learning_rate": 0.0001682116303869587,
      "loss": 3.6415,
      "step": 170510
    },
    {
      "epoch": 0.1589522314846739,
      "grad_norm": 1.1262712478637695,
      "learning_rate": 0.00016820976606486348,
      "loss": 3.3148,
      "step": 170520
    },
    {
      "epoch": 0.158961553102753,
      "grad_norm": 1.146405577659607,
      "learning_rate": 0.00016820790174276832,
      "loss": 3.576,
      "step": 170530
    },
    {
      "epoch": 0.1589708747208321,
      "grad_norm": 1.1460267305374146,
      "learning_rate": 0.00016820603742067312,
      "loss": 3.4934,
      "step": 170540
    },
    {
      "epoch": 0.1589801963389112,
      "grad_norm": 1.1410731077194214,
      "learning_rate": 0.0001682041730985779,
      "loss": 3.7318,
      "step": 170550
    },
    {
      "epoch": 0.1589895179569903,
      "grad_norm": 1.0320484638214111,
      "learning_rate": 0.0001682023087764827,
      "loss": 3.5991,
      "step": 170560
    },
    {
      "epoch": 0.15899883957506938,
      "grad_norm": 1.150269865989685,
      "learning_rate": 0.0001682004444543875,
      "loss": 3.6939,
      "step": 170570
    },
    {
      "epoch": 0.15900816119314848,
      "grad_norm": 0.9575408101081848,
      "learning_rate": 0.0001681985801322923,
      "loss": 3.6518,
      "step": 170580
    },
    {
      "epoch": 0.15901748281122757,
      "grad_norm": 1.2893210649490356,
      "learning_rate": 0.00016819671581019713,
      "loss": 3.5578,
      "step": 170590
    },
    {
      "epoch": 0.15902680442930667,
      "grad_norm": 1.0165340900421143,
      "learning_rate": 0.0001681948514881019,
      "loss": 3.6272,
      "step": 170600
    },
    {
      "epoch": 0.15903612604738573,
      "grad_norm": 1.149549126625061,
      "learning_rate": 0.0001681929871660067,
      "loss": 3.5166,
      "step": 170610
    },
    {
      "epoch": 0.15904544766546483,
      "grad_norm": 1.1320832967758179,
      "learning_rate": 0.0001681911228439115,
      "loss": 3.5736,
      "step": 170620
    },
    {
      "epoch": 0.15905476928354392,
      "grad_norm": 1.089776635169983,
      "learning_rate": 0.0001681892585218163,
      "loss": 3.6904,
      "step": 170630
    },
    {
      "epoch": 0.15906409090162302,
      "grad_norm": 1.0173215866088867,
      "learning_rate": 0.0001681873941997211,
      "loss": 3.5624,
      "step": 170640
    },
    {
      "epoch": 0.1590734125197021,
      "grad_norm": 1.074442982673645,
      "learning_rate": 0.0001681855298776259,
      "loss": 3.3585,
      "step": 170650
    },
    {
      "epoch": 0.1590827341377812,
      "grad_norm": 1.0812673568725586,
      "learning_rate": 0.00016818366555553072,
      "loss": 3.5505,
      "step": 170660
    },
    {
      "epoch": 0.1590920557558603,
      "grad_norm": 0.9601555466651917,
      "learning_rate": 0.0001681818012334355,
      "loss": 3.5101,
      "step": 170670
    },
    {
      "epoch": 0.1591013773739394,
      "grad_norm": 1.1663622856140137,
      "learning_rate": 0.0001681799369113403,
      "loss": 3.662,
      "step": 170680
    },
    {
      "epoch": 0.1591106989920185,
      "grad_norm": 1.0859016180038452,
      "learning_rate": 0.0001681780725892451,
      "loss": 3.6525,
      "step": 170690
    },
    {
      "epoch": 0.15912002061009758,
      "grad_norm": 1.1713271141052246,
      "learning_rate": 0.00016817620826714992,
      "loss": 3.7046,
      "step": 170700
    },
    {
      "epoch": 0.15912934222817665,
      "grad_norm": 1.0117686986923218,
      "learning_rate": 0.00016817434394505472,
      "loss": 3.7046,
      "step": 170710
    },
    {
      "epoch": 0.15913866384625575,
      "grad_norm": 1.142736792564392,
      "learning_rate": 0.0001681724796229595,
      "loss": 3.4694,
      "step": 170720
    },
    {
      "epoch": 0.15914798546433484,
      "grad_norm": 0.9450521469116211,
      "learning_rate": 0.0001681706153008643,
      "loss": 3.3551,
      "step": 170730
    },
    {
      "epoch": 0.15915730708241393,
      "grad_norm": 1.1738369464874268,
      "learning_rate": 0.0001681687509787691,
      "loss": 3.4132,
      "step": 170740
    },
    {
      "epoch": 0.15916662870049303,
      "grad_norm": 1.16764235496521,
      "learning_rate": 0.0001681668866566739,
      "loss": 3.5013,
      "step": 170750
    },
    {
      "epoch": 0.15917595031857212,
      "grad_norm": 1.141365647315979,
      "learning_rate": 0.00016816502233457873,
      "loss": 3.5986,
      "step": 170760
    },
    {
      "epoch": 0.15918527193665122,
      "grad_norm": 1.0488919019699097,
      "learning_rate": 0.0001681631580124835,
      "loss": 3.653,
      "step": 170770
    },
    {
      "epoch": 0.1591945935547303,
      "grad_norm": 1.028934121131897,
      "learning_rate": 0.0001681612936903883,
      "loss": 3.6611,
      "step": 170780
    },
    {
      "epoch": 0.1592039151728094,
      "grad_norm": 1.2155736684799194,
      "learning_rate": 0.00016815942936829312,
      "loss": 3.6447,
      "step": 170790
    },
    {
      "epoch": 0.1592132367908885,
      "grad_norm": 0.9160192608833313,
      "learning_rate": 0.0001681575650461979,
      "loss": 3.5213,
      "step": 170800
    },
    {
      "epoch": 0.1592225584089676,
      "grad_norm": 1.1006824970245361,
      "learning_rate": 0.0001681557007241027,
      "loss": 3.4598,
      "step": 170810
    },
    {
      "epoch": 0.15923188002704666,
      "grad_norm": 0.9829181432723999,
      "learning_rate": 0.0001681538364020075,
      "loss": 3.4156,
      "step": 170820
    },
    {
      "epoch": 0.15924120164512576,
      "grad_norm": 0.8658108711242676,
      "learning_rate": 0.00016815197207991232,
      "loss": 3.6132,
      "step": 170830
    },
    {
      "epoch": 0.15925052326320485,
      "grad_norm": 1.031430721282959,
      "learning_rate": 0.00016815010775781712,
      "loss": 3.6286,
      "step": 170840
    },
    {
      "epoch": 0.15925984488128395,
      "grad_norm": 1.0939459800720215,
      "learning_rate": 0.0001681482434357219,
      "loss": 3.4668,
      "step": 170850
    },
    {
      "epoch": 0.15926916649936304,
      "grad_norm": 1.028557300567627,
      "learning_rate": 0.0001681463791136267,
      "loss": 3.6929,
      "step": 170860
    },
    {
      "epoch": 0.15927848811744214,
      "grad_norm": 1.135581374168396,
      "learning_rate": 0.00016814451479153151,
      "loss": 3.5571,
      "step": 170870
    },
    {
      "epoch": 0.15928780973552123,
      "grad_norm": 1.1813255548477173,
      "learning_rate": 0.00016814265046943632,
      "loss": 3.4156,
      "step": 170880
    },
    {
      "epoch": 0.15929713135360032,
      "grad_norm": 1.380039095878601,
      "learning_rate": 0.00016814078614734113,
      "loss": 3.5898,
      "step": 170890
    },
    {
      "epoch": 0.15930645297167942,
      "grad_norm": 0.9877490997314453,
      "learning_rate": 0.0001681389218252459,
      "loss": 3.5673,
      "step": 170900
    },
    {
      "epoch": 0.1593157745897585,
      "grad_norm": 0.944980263710022,
      "learning_rate": 0.0001681370575031507,
      "loss": 3.3473,
      "step": 170910
    },
    {
      "epoch": 0.15932509620783758,
      "grad_norm": 1.0331168174743652,
      "learning_rate": 0.00016813519318105552,
      "loss": 3.6339,
      "step": 170920
    },
    {
      "epoch": 0.15933441782591667,
      "grad_norm": 1.1017268896102905,
      "learning_rate": 0.00016813332885896032,
      "loss": 3.5616,
      "step": 170930
    },
    {
      "epoch": 0.15934373944399577,
      "grad_norm": 0.9269038438796997,
      "learning_rate": 0.00016813146453686513,
      "loss": 3.6577,
      "step": 170940
    },
    {
      "epoch": 0.15935306106207486,
      "grad_norm": 0.9526374936103821,
      "learning_rate": 0.0001681296002147699,
      "loss": 3.4905,
      "step": 170950
    },
    {
      "epoch": 0.15936238268015396,
      "grad_norm": 0.9255598783493042,
      "learning_rate": 0.00016812773589267472,
      "loss": 3.7257,
      "step": 170960
    },
    {
      "epoch": 0.15937170429823305,
      "grad_norm": 0.9863935112953186,
      "learning_rate": 0.0001681258715705795,
      "loss": 3.4635,
      "step": 170970
    },
    {
      "epoch": 0.15938102591631215,
      "grad_norm": 1.09632408618927,
      "learning_rate": 0.00016812400724848433,
      "loss": 3.5042,
      "step": 170980
    },
    {
      "epoch": 0.15939034753439124,
      "grad_norm": 1.340378761291504,
      "learning_rate": 0.0001681221429263891,
      "loss": 3.6221,
      "step": 170990
    },
    {
      "epoch": 0.15939966915247034,
      "grad_norm": 0.9666957855224609,
      "learning_rate": 0.00016812027860429391,
      "loss": 3.5822,
      "step": 171000
    },
    {
      "epoch": 0.15940899077054943,
      "grad_norm": 1.0975581407546997,
      "learning_rate": 0.00016811841428219872,
      "loss": 3.6893,
      "step": 171010
    },
    {
      "epoch": 0.15941831238862852,
      "grad_norm": 0.921926736831665,
      "learning_rate": 0.0001681165499601035,
      "loss": 3.6036,
      "step": 171020
    },
    {
      "epoch": 0.1594276340067076,
      "grad_norm": 1.2192727327346802,
      "learning_rate": 0.0001681146856380083,
      "loss": 3.5831,
      "step": 171030
    },
    {
      "epoch": 0.15943695562478669,
      "grad_norm": 1.1568341255187988,
      "learning_rate": 0.0001681128213159131,
      "loss": 3.377,
      "step": 171040
    },
    {
      "epoch": 0.15944627724286578,
      "grad_norm": 0.9796777367591858,
      "learning_rate": 0.00016811095699381792,
      "loss": 3.4245,
      "step": 171050
    },
    {
      "epoch": 0.15945559886094487,
      "grad_norm": 1.077352523803711,
      "learning_rate": 0.00016810909267172273,
      "loss": 3.5876,
      "step": 171060
    },
    {
      "epoch": 0.15946492047902397,
      "grad_norm": 1.0732640027999878,
      "learning_rate": 0.0001681072283496275,
      "loss": 3.4012,
      "step": 171070
    },
    {
      "epoch": 0.15947424209710306,
      "grad_norm": 1.2757352590560913,
      "learning_rate": 0.0001681053640275323,
      "loss": 3.6437,
      "step": 171080
    },
    {
      "epoch": 0.15948356371518216,
      "grad_norm": 0.8680155277252197,
      "learning_rate": 0.00016810349970543712,
      "loss": 3.5061,
      "step": 171090
    },
    {
      "epoch": 0.15949288533326125,
      "grad_norm": 0.780190110206604,
      "learning_rate": 0.00016810163538334192,
      "loss": 3.6358,
      "step": 171100
    },
    {
      "epoch": 0.15950220695134035,
      "grad_norm": 1.0225727558135986,
      "learning_rate": 0.00016809977106124673,
      "loss": 3.4703,
      "step": 171110
    },
    {
      "epoch": 0.15951152856941944,
      "grad_norm": 1.2446869611740112,
      "learning_rate": 0.0001680979067391515,
      "loss": 3.5823,
      "step": 171120
    },
    {
      "epoch": 0.1595208501874985,
      "grad_norm": 1.0178172588348389,
      "learning_rate": 0.00016809604241705632,
      "loss": 3.5174,
      "step": 171130
    },
    {
      "epoch": 0.1595301718055776,
      "grad_norm": 0.988232433795929,
      "learning_rate": 0.0001680941780949611,
      "loss": 3.6908,
      "step": 171140
    },
    {
      "epoch": 0.1595394934236567,
      "grad_norm": 1.1047145128250122,
      "learning_rate": 0.00016809231377286593,
      "loss": 3.5064,
      "step": 171150
    },
    {
      "epoch": 0.1595488150417358,
      "grad_norm": 0.9645419716835022,
      "learning_rate": 0.00016809044945077073,
      "loss": 3.6247,
      "step": 171160
    },
    {
      "epoch": 0.1595581366598149,
      "grad_norm": 1.1034952402114868,
      "learning_rate": 0.00016808858512867551,
      "loss": 3.6053,
      "step": 171170
    },
    {
      "epoch": 0.15956745827789398,
      "grad_norm": 1.2058628797531128,
      "learning_rate": 0.00016808672080658032,
      "loss": 3.4778,
      "step": 171180
    },
    {
      "epoch": 0.15957677989597308,
      "grad_norm": 1.0802356004714966,
      "learning_rate": 0.0001680848564844851,
      "loss": 3.4661,
      "step": 171190
    },
    {
      "epoch": 0.15958610151405217,
      "grad_norm": 1.1744440793991089,
      "learning_rate": 0.0001680829921623899,
      "loss": 3.4484,
      "step": 171200
    },
    {
      "epoch": 0.15959542313213126,
      "grad_norm": 0.8541669845581055,
      "learning_rate": 0.00016808112784029474,
      "loss": 3.4852,
      "step": 171210
    },
    {
      "epoch": 0.15960474475021036,
      "grad_norm": 1.0153905153274536,
      "learning_rate": 0.00016807926351819952,
      "loss": 3.5113,
      "step": 171220
    },
    {
      "epoch": 0.15961406636828945,
      "grad_norm": 0.9782846570014954,
      "learning_rate": 0.00016807739919610432,
      "loss": 3.6061,
      "step": 171230
    },
    {
      "epoch": 0.15962338798636852,
      "grad_norm": 1.046229362487793,
      "learning_rate": 0.0001680755348740091,
      "loss": 3.4989,
      "step": 171240
    },
    {
      "epoch": 0.15963270960444761,
      "grad_norm": 1.0800234079360962,
      "learning_rate": 0.0001680736705519139,
      "loss": 3.4025,
      "step": 171250
    },
    {
      "epoch": 0.1596420312225267,
      "grad_norm": 0.9302319884300232,
      "learning_rate": 0.00016807180622981872,
      "loss": 3.5442,
      "step": 171260
    },
    {
      "epoch": 0.1596513528406058,
      "grad_norm": 1.1193076372146606,
      "learning_rate": 0.00016806994190772352,
      "loss": 3.5036,
      "step": 171270
    },
    {
      "epoch": 0.1596606744586849,
      "grad_norm": 1.0812958478927612,
      "learning_rate": 0.00016806807758562833,
      "loss": 3.6025,
      "step": 171280
    },
    {
      "epoch": 0.159669996076764,
      "grad_norm": 1.2554805278778076,
      "learning_rate": 0.0001680662132635331,
      "loss": 3.3692,
      "step": 171290
    },
    {
      "epoch": 0.1596793176948431,
      "grad_norm": 1.2603448629379272,
      "learning_rate": 0.00016806434894143791,
      "loss": 3.5284,
      "step": 171300
    },
    {
      "epoch": 0.15968863931292218,
      "grad_norm": 1.1971694231033325,
      "learning_rate": 0.00016806248461934272,
      "loss": 3.6408,
      "step": 171310
    },
    {
      "epoch": 0.15969796093100128,
      "grad_norm": 1.2156099081039429,
      "learning_rate": 0.00016806062029724753,
      "loss": 3.3863,
      "step": 171320
    },
    {
      "epoch": 0.15970728254908037,
      "grad_norm": 0.9963748455047607,
      "learning_rate": 0.00016805875597515233,
      "loss": 3.5853,
      "step": 171330
    },
    {
      "epoch": 0.15971660416715944,
      "grad_norm": 1.008134126663208,
      "learning_rate": 0.0001680568916530571,
      "loss": 3.5273,
      "step": 171340
    },
    {
      "epoch": 0.15972592578523853,
      "grad_norm": 1.1382478475570679,
      "learning_rate": 0.00016805502733096192,
      "loss": 3.5114,
      "step": 171350
    },
    {
      "epoch": 0.15973524740331763,
      "grad_norm": 1.1977064609527588,
      "learning_rate": 0.00016805316300886673,
      "loss": 3.6377,
      "step": 171360
    },
    {
      "epoch": 0.15974456902139672,
      "grad_norm": 1.0129843950271606,
      "learning_rate": 0.00016805129868677153,
      "loss": 3.4003,
      "step": 171370
    },
    {
      "epoch": 0.15975389063947582,
      "grad_norm": 1.0776807069778442,
      "learning_rate": 0.00016804943436467634,
      "loss": 3.4498,
      "step": 171380
    },
    {
      "epoch": 0.1597632122575549,
      "grad_norm": 0.9623083472251892,
      "learning_rate": 0.00016804757004258112,
      "loss": 3.5251,
      "step": 171390
    },
    {
      "epoch": 0.159772533875634,
      "grad_norm": 1.1325215101242065,
      "learning_rate": 0.00016804570572048592,
      "loss": 3.7449,
      "step": 171400
    },
    {
      "epoch": 0.1597818554937131,
      "grad_norm": 1.1477996110916138,
      "learning_rate": 0.00016804384139839073,
      "loss": 3.5861,
      "step": 171410
    },
    {
      "epoch": 0.1597911771117922,
      "grad_norm": 1.0875663757324219,
      "learning_rate": 0.0001680419770762955,
      "loss": 3.5978,
      "step": 171420
    },
    {
      "epoch": 0.1598004987298713,
      "grad_norm": 1.1056112051010132,
      "learning_rate": 0.00016804011275420034,
      "loss": 3.4237,
      "step": 171430
    },
    {
      "epoch": 0.15980982034795038,
      "grad_norm": 0.9452841877937317,
      "learning_rate": 0.00016803824843210512,
      "loss": 3.3901,
      "step": 171440
    },
    {
      "epoch": 0.15981914196602945,
      "grad_norm": 1.1173471212387085,
      "learning_rate": 0.00016803638411000993,
      "loss": 3.546,
      "step": 171450
    },
    {
      "epoch": 0.15982846358410854,
      "grad_norm": 0.7412100434303284,
      "learning_rate": 0.00016803451978791473,
      "loss": 3.5048,
      "step": 171460
    },
    {
      "epoch": 0.15983778520218764,
      "grad_norm": 0.927043080329895,
      "learning_rate": 0.0001680326554658195,
      "loss": 3.6197,
      "step": 171470
    },
    {
      "epoch": 0.15984710682026673,
      "grad_norm": 1.1719286441802979,
      "learning_rate": 0.00016803079114372432,
      "loss": 3.5717,
      "step": 171480
    },
    {
      "epoch": 0.15985642843834583,
      "grad_norm": 1.2103444337844849,
      "learning_rate": 0.00016802892682162913,
      "loss": 3.5919,
      "step": 171490
    },
    {
      "epoch": 0.15986575005642492,
      "grad_norm": 0.9515886306762695,
      "learning_rate": 0.00016802706249953393,
      "loss": 3.6591,
      "step": 171500
    },
    {
      "epoch": 0.15987507167450402,
      "grad_norm": 0.9016027450561523,
      "learning_rate": 0.00016802519817743874,
      "loss": 3.5861,
      "step": 171510
    },
    {
      "epoch": 0.1598843932925831,
      "grad_norm": 1.2167996168136597,
      "learning_rate": 0.00016802333385534352,
      "loss": 3.5143,
      "step": 171520
    },
    {
      "epoch": 0.1598937149106622,
      "grad_norm": 0.9850214123725891,
      "learning_rate": 0.00016802146953324832,
      "loss": 3.5171,
      "step": 171530
    },
    {
      "epoch": 0.1599030365287413,
      "grad_norm": 1.1577801704406738,
      "learning_rate": 0.00016801960521115313,
      "loss": 3.5908,
      "step": 171540
    },
    {
      "epoch": 0.15991235814682037,
      "grad_norm": 0.950685441493988,
      "learning_rate": 0.00016801774088905794,
      "loss": 3.4247,
      "step": 171550
    },
    {
      "epoch": 0.15992167976489946,
      "grad_norm": 1.1303287744522095,
      "learning_rate": 0.00016801587656696274,
      "loss": 3.4857,
      "step": 171560
    },
    {
      "epoch": 0.15993100138297855,
      "grad_norm": 1.221411943435669,
      "learning_rate": 0.00016801401224486752,
      "loss": 3.4869,
      "step": 171570
    },
    {
      "epoch": 0.15994032300105765,
      "grad_norm": 0.9802001714706421,
      "learning_rate": 0.00016801214792277233,
      "loss": 3.6109,
      "step": 171580
    },
    {
      "epoch": 0.15994964461913674,
      "grad_norm": 1.1731351613998413,
      "learning_rate": 0.0001680102836006771,
      "loss": 3.508,
      "step": 171590
    },
    {
      "epoch": 0.15995896623721584,
      "grad_norm": 1.0703705549240112,
      "learning_rate": 0.00016800841927858194,
      "loss": 3.4768,
      "step": 171600
    },
    {
      "epoch": 0.15996828785529493,
      "grad_norm": 0.8390430808067322,
      "learning_rate": 0.00016800655495648675,
      "loss": 3.5045,
      "step": 171610
    },
    {
      "epoch": 0.15997760947337403,
      "grad_norm": 1.017179250717163,
      "learning_rate": 0.00016800469063439153,
      "loss": 3.7173,
      "step": 171620
    },
    {
      "epoch": 0.15998693109145312,
      "grad_norm": 1.124981164932251,
      "learning_rate": 0.00016800282631229633,
      "loss": 3.5296,
      "step": 171630
    },
    {
      "epoch": 0.15999625270953222,
      "grad_norm": 1.2593653202056885,
      "learning_rate": 0.0001680009619902011,
      "loss": 3.7039,
      "step": 171640
    },
    {
      "epoch": 0.1600055743276113,
      "grad_norm": 0.9644660949707031,
      "learning_rate": 0.00016799909766810592,
      "loss": 3.609,
      "step": 171650
    },
    {
      "epoch": 0.16001489594569038,
      "grad_norm": 1.2216161489486694,
      "learning_rate": 0.00016799723334601075,
      "loss": 3.3015,
      "step": 171660
    },
    {
      "epoch": 0.16002421756376947,
      "grad_norm": 1.0764517784118652,
      "learning_rate": 0.00016799536902391553,
      "loss": 3.4619,
      "step": 171670
    },
    {
      "epoch": 0.16003353918184857,
      "grad_norm": 1.374338150024414,
      "learning_rate": 0.00016799350470182034,
      "loss": 3.4299,
      "step": 171680
    },
    {
      "epoch": 0.16004286079992766,
      "grad_norm": 1.2043873071670532,
      "learning_rate": 0.00016799164037972512,
      "loss": 3.6431,
      "step": 171690
    },
    {
      "epoch": 0.16005218241800676,
      "grad_norm": 0.9264429807662964,
      "learning_rate": 0.00016798977605762992,
      "loss": 3.6131,
      "step": 171700
    },
    {
      "epoch": 0.16006150403608585,
      "grad_norm": 1.2754080295562744,
      "learning_rate": 0.00016798791173553473,
      "loss": 3.5077,
      "step": 171710
    },
    {
      "epoch": 0.16007082565416494,
      "grad_norm": 1.3496394157409668,
      "learning_rate": 0.00016798604741343954,
      "loss": 3.5155,
      "step": 171720
    },
    {
      "epoch": 0.16008014727224404,
      "grad_norm": 0.9762224555015564,
      "learning_rate": 0.00016798418309134434,
      "loss": 3.5322,
      "step": 171730
    },
    {
      "epoch": 0.16008946889032313,
      "grad_norm": 1.123460292816162,
      "learning_rate": 0.00016798231876924912,
      "loss": 3.583,
      "step": 171740
    },
    {
      "epoch": 0.16009879050840223,
      "grad_norm": 1.0771199464797974,
      "learning_rate": 0.00016798045444715393,
      "loss": 3.4587,
      "step": 171750
    },
    {
      "epoch": 0.1601081121264813,
      "grad_norm": 1.0559968948364258,
      "learning_rate": 0.00016797859012505873,
      "loss": 3.4432,
      "step": 171760
    },
    {
      "epoch": 0.1601174337445604,
      "grad_norm": 1.3557630777359009,
      "learning_rate": 0.00016797672580296354,
      "loss": 3.5413,
      "step": 171770
    },
    {
      "epoch": 0.16012675536263948,
      "grad_norm": 0.8962087035179138,
      "learning_rate": 0.00016797486148086835,
      "loss": 3.3822,
      "step": 171780
    },
    {
      "epoch": 0.16013607698071858,
      "grad_norm": 1.241219162940979,
      "learning_rate": 0.00016797299715877313,
      "loss": 3.7569,
      "step": 171790
    },
    {
      "epoch": 0.16014539859879767,
      "grad_norm": 1.0830004215240479,
      "learning_rate": 0.00016797113283667793,
      "loss": 3.8225,
      "step": 171800
    },
    {
      "epoch": 0.16015472021687677,
      "grad_norm": 0.9483686089515686,
      "learning_rate": 0.00016796926851458274,
      "loss": 3.6672,
      "step": 171810
    },
    {
      "epoch": 0.16016404183495586,
      "grad_norm": 1.0507367849349976,
      "learning_rate": 0.00016796740419248754,
      "loss": 3.6277,
      "step": 171820
    },
    {
      "epoch": 0.16017336345303496,
      "grad_norm": 1.4049328565597534,
      "learning_rate": 0.00016796553987039235,
      "loss": 3.4335,
      "step": 171830
    },
    {
      "epoch": 0.16018268507111405,
      "grad_norm": 1.0762519836425781,
      "learning_rate": 0.00016796367554829713,
      "loss": 3.4703,
      "step": 171840
    },
    {
      "epoch": 0.16019200668919314,
      "grad_norm": 0.9890592694282532,
      "learning_rate": 0.00016796181122620194,
      "loss": 3.3187,
      "step": 171850
    },
    {
      "epoch": 0.1602013283072722,
      "grad_norm": 1.0835068225860596,
      "learning_rate": 0.00016795994690410674,
      "loss": 3.6319,
      "step": 171860
    },
    {
      "epoch": 0.1602106499253513,
      "grad_norm": 1.1931672096252441,
      "learning_rate": 0.00016795808258201152,
      "loss": 3.5941,
      "step": 171870
    },
    {
      "epoch": 0.1602199715434304,
      "grad_norm": 0.9062944054603577,
      "learning_rate": 0.00016795621825991635,
      "loss": 3.4648,
      "step": 171880
    },
    {
      "epoch": 0.1602292931615095,
      "grad_norm": 1.0354760885238647,
      "learning_rate": 0.00016795435393782113,
      "loss": 3.4869,
      "step": 171890
    },
    {
      "epoch": 0.1602386147795886,
      "grad_norm": 1.1350841522216797,
      "learning_rate": 0.00016795248961572594,
      "loss": 3.6157,
      "step": 171900
    },
    {
      "epoch": 0.16024793639766768,
      "grad_norm": 1.2266916036605835,
      "learning_rate": 0.00016795062529363075,
      "loss": 3.5001,
      "step": 171910
    },
    {
      "epoch": 0.16025725801574678,
      "grad_norm": 0.8674575090408325,
      "learning_rate": 0.00016794876097153553,
      "loss": 3.4566,
      "step": 171920
    },
    {
      "epoch": 0.16026657963382587,
      "grad_norm": 0.8470000624656677,
      "learning_rate": 0.00016794689664944033,
      "loss": 3.6071,
      "step": 171930
    },
    {
      "epoch": 0.16027590125190497,
      "grad_norm": 1.0536530017852783,
      "learning_rate": 0.00016794503232734514,
      "loss": 3.5651,
      "step": 171940
    },
    {
      "epoch": 0.16028522286998406,
      "grad_norm": 1.0844358205795288,
      "learning_rate": 0.00016794316800524994,
      "loss": 3.337,
      "step": 171950
    },
    {
      "epoch": 0.16029454448806316,
      "grad_norm": 0.9969109892845154,
      "learning_rate": 0.00016794130368315475,
      "loss": 3.5608,
      "step": 171960
    },
    {
      "epoch": 0.16030386610614222,
      "grad_norm": 1.095811128616333,
      "learning_rate": 0.00016793943936105953,
      "loss": 3.5842,
      "step": 171970
    },
    {
      "epoch": 0.16031318772422132,
      "grad_norm": 0.9508348703384399,
      "learning_rate": 0.00016793757503896434,
      "loss": 3.4791,
      "step": 171980
    },
    {
      "epoch": 0.1603225093423004,
      "grad_norm": 1.075769305229187,
      "learning_rate": 0.00016793571071686914,
      "loss": 3.3133,
      "step": 171990
    },
    {
      "epoch": 0.1603318309603795,
      "grad_norm": 1.0359238386154175,
      "learning_rate": 0.00016793384639477395,
      "loss": 3.3953,
      "step": 172000
    },
    {
      "epoch": 0.1603411525784586,
      "grad_norm": 0.976862907409668,
      "learning_rate": 0.00016793198207267876,
      "loss": 3.3947,
      "step": 172010
    },
    {
      "epoch": 0.1603504741965377,
      "grad_norm": 1.2162185907363892,
      "learning_rate": 0.00016793011775058353,
      "loss": 3.5609,
      "step": 172020
    },
    {
      "epoch": 0.1603597958146168,
      "grad_norm": 1.127697467803955,
      "learning_rate": 0.00016792825342848834,
      "loss": 3.391,
      "step": 172030
    },
    {
      "epoch": 0.16036911743269588,
      "grad_norm": 1.067224144935608,
      "learning_rate": 0.00016792638910639312,
      "loss": 3.5343,
      "step": 172040
    },
    {
      "epoch": 0.16037843905077498,
      "grad_norm": 0.9698837399482727,
      "learning_rate": 0.00016792452478429795,
      "loss": 3.6902,
      "step": 172050
    },
    {
      "epoch": 0.16038776066885407,
      "grad_norm": 0.9481937885284424,
      "learning_rate": 0.00016792266046220276,
      "loss": 3.5623,
      "step": 172060
    },
    {
      "epoch": 0.16039708228693314,
      "grad_norm": 0.9105319380760193,
      "learning_rate": 0.00016792079614010754,
      "loss": 3.502,
      "step": 172070
    },
    {
      "epoch": 0.16040640390501223,
      "grad_norm": 1.2712136507034302,
      "learning_rate": 0.00016791893181801235,
      "loss": 3.6339,
      "step": 172080
    },
    {
      "epoch": 0.16041572552309133,
      "grad_norm": 1.0726701021194458,
      "learning_rate": 0.00016791706749591712,
      "loss": 3.4768,
      "step": 172090
    },
    {
      "epoch": 0.16042504714117042,
      "grad_norm": 1.0181512832641602,
      "learning_rate": 0.00016791520317382193,
      "loss": 3.5273,
      "step": 172100
    },
    {
      "epoch": 0.16043436875924952,
      "grad_norm": 1.2660565376281738,
      "learning_rate": 0.00016791333885172676,
      "loss": 3.5714,
      "step": 172110
    },
    {
      "epoch": 0.1604436903773286,
      "grad_norm": 1.0902938842773438,
      "learning_rate": 0.00016791147452963154,
      "loss": 3.5863,
      "step": 172120
    },
    {
      "epoch": 0.1604530119954077,
      "grad_norm": 1.0784144401550293,
      "learning_rate": 0.00016790961020753635,
      "loss": 3.4983,
      "step": 172130
    },
    {
      "epoch": 0.1604623336134868,
      "grad_norm": 1.007941722869873,
      "learning_rate": 0.00016790774588544113,
      "loss": 3.5495,
      "step": 172140
    },
    {
      "epoch": 0.1604716552315659,
      "grad_norm": 1.070878028869629,
      "learning_rate": 0.00016790588156334594,
      "loss": 3.4172,
      "step": 172150
    },
    {
      "epoch": 0.160480976849645,
      "grad_norm": 1.107844591140747,
      "learning_rate": 0.00016790401724125074,
      "loss": 3.4837,
      "step": 172160
    },
    {
      "epoch": 0.16049029846772409,
      "grad_norm": 1.0039217472076416,
      "learning_rate": 0.00016790215291915555,
      "loss": 3.4693,
      "step": 172170
    },
    {
      "epoch": 0.16049962008580315,
      "grad_norm": 0.9736793637275696,
      "learning_rate": 0.00016790028859706035,
      "loss": 3.3918,
      "step": 172180
    },
    {
      "epoch": 0.16050894170388225,
      "grad_norm": 1.0132842063903809,
      "learning_rate": 0.00016789842427496513,
      "loss": 3.4245,
      "step": 172190
    },
    {
      "epoch": 0.16051826332196134,
      "grad_norm": 0.950014054775238,
      "learning_rate": 0.00016789655995286994,
      "loss": 3.5569,
      "step": 172200
    },
    {
      "epoch": 0.16052758494004044,
      "grad_norm": 1.019452691078186,
      "learning_rate": 0.00016789469563077475,
      "loss": 3.4836,
      "step": 172210
    },
    {
      "epoch": 0.16053690655811953,
      "grad_norm": 1.0924036502838135,
      "learning_rate": 0.00016789283130867955,
      "loss": 3.2852,
      "step": 172220
    },
    {
      "epoch": 0.16054622817619862,
      "grad_norm": 0.9758851528167725,
      "learning_rate": 0.00016789096698658436,
      "loss": 3.4537,
      "step": 172230
    },
    {
      "epoch": 0.16055554979427772,
      "grad_norm": 0.9128869771957397,
      "learning_rate": 0.00016788910266448914,
      "loss": 3.6158,
      "step": 172240
    },
    {
      "epoch": 0.1605648714123568,
      "grad_norm": 0.9989377856254578,
      "learning_rate": 0.00016788723834239394,
      "loss": 3.4838,
      "step": 172250
    },
    {
      "epoch": 0.1605741930304359,
      "grad_norm": 0.953129231929779,
      "learning_rate": 0.00016788537402029875,
      "loss": 3.5602,
      "step": 172260
    },
    {
      "epoch": 0.160583514648515,
      "grad_norm": 1.0355278253555298,
      "learning_rate": 0.00016788350969820356,
      "loss": 3.4688,
      "step": 172270
    },
    {
      "epoch": 0.16059283626659407,
      "grad_norm": 1.0104042291641235,
      "learning_rate": 0.00016788164537610836,
      "loss": 3.681,
      "step": 172280
    },
    {
      "epoch": 0.16060215788467316,
      "grad_norm": 1.1600631475448608,
      "learning_rate": 0.00016787978105401314,
      "loss": 3.4433,
      "step": 172290
    },
    {
      "epoch": 0.16061147950275226,
      "grad_norm": 0.9543317556381226,
      "learning_rate": 0.00016787791673191795,
      "loss": 3.4331,
      "step": 172300
    },
    {
      "epoch": 0.16062080112083135,
      "grad_norm": 1.1072136163711548,
      "learning_rate": 0.00016787605240982276,
      "loss": 3.6989,
      "step": 172310
    },
    {
      "epoch": 0.16063012273891045,
      "grad_norm": 1.0747160911560059,
      "learning_rate": 0.00016787418808772753,
      "loss": 3.6474,
      "step": 172320
    },
    {
      "epoch": 0.16063944435698954,
      "grad_norm": 1.1071518659591675,
      "learning_rate": 0.00016787232376563237,
      "loss": 3.478,
      "step": 172330
    },
    {
      "epoch": 0.16064876597506864,
      "grad_norm": 1.0517146587371826,
      "learning_rate": 0.00016787045944353715,
      "loss": 3.6027,
      "step": 172340
    },
    {
      "epoch": 0.16065808759314773,
      "grad_norm": 1.478809118270874,
      "learning_rate": 0.00016786859512144195,
      "loss": 3.5214,
      "step": 172350
    },
    {
      "epoch": 0.16066740921122682,
      "grad_norm": 1.304725170135498,
      "learning_rate": 0.00016786673079934673,
      "loss": 3.5194,
      "step": 172360
    },
    {
      "epoch": 0.16067673082930592,
      "grad_norm": 0.9961414337158203,
      "learning_rate": 0.00016786486647725154,
      "loss": 3.6741,
      "step": 172370
    },
    {
      "epoch": 0.160686052447385,
      "grad_norm": 1.1395765542984009,
      "learning_rate": 0.00016786300215515635,
      "loss": 3.6389,
      "step": 172380
    },
    {
      "epoch": 0.16069537406546408,
      "grad_norm": 1.0574822425842285,
      "learning_rate": 0.00016786113783306115,
      "loss": 3.6349,
      "step": 172390
    },
    {
      "epoch": 0.16070469568354318,
      "grad_norm": 1.3263883590698242,
      "learning_rate": 0.00016785927351096596,
      "loss": 3.6891,
      "step": 172400
    },
    {
      "epoch": 0.16071401730162227,
      "grad_norm": 1.2459477186203003,
      "learning_rate": 0.00016785740918887074,
      "loss": 3.3824,
      "step": 172410
    },
    {
      "epoch": 0.16072333891970136,
      "grad_norm": 1.2386705875396729,
      "learning_rate": 0.00016785554486677554,
      "loss": 3.6093,
      "step": 172420
    },
    {
      "epoch": 0.16073266053778046,
      "grad_norm": 1.0929063558578491,
      "learning_rate": 0.00016785368054468035,
      "loss": 3.6306,
      "step": 172430
    },
    {
      "epoch": 0.16074198215585955,
      "grad_norm": 1.1161116361618042,
      "learning_rate": 0.00016785181622258516,
      "loss": 3.6006,
      "step": 172440
    },
    {
      "epoch": 0.16075130377393865,
      "grad_norm": 0.9312822222709656,
      "learning_rate": 0.00016784995190048996,
      "loss": 3.508,
      "step": 172450
    },
    {
      "epoch": 0.16076062539201774,
      "grad_norm": 0.9686979651451111,
      "learning_rate": 0.00016784808757839474,
      "loss": 3.516,
      "step": 172460
    },
    {
      "epoch": 0.16076994701009684,
      "grad_norm": 1.2533886432647705,
      "learning_rate": 0.00016784622325629955,
      "loss": 3.571,
      "step": 172470
    },
    {
      "epoch": 0.16077926862817593,
      "grad_norm": 1.1131280660629272,
      "learning_rate": 0.00016784435893420435,
      "loss": 3.5283,
      "step": 172480
    },
    {
      "epoch": 0.160788590246255,
      "grad_norm": 1.1228677034378052,
      "learning_rate": 0.00016784249461210916,
      "loss": 3.6643,
      "step": 172490
    },
    {
      "epoch": 0.1607979118643341,
      "grad_norm": 1.1522626876831055,
      "learning_rate": 0.00016784063029001397,
      "loss": 3.5258,
      "step": 172500
    },
    {
      "epoch": 0.1608072334824132,
      "grad_norm": 0.9396576881408691,
      "learning_rate": 0.00016783876596791875,
      "loss": 3.624,
      "step": 172510
    },
    {
      "epoch": 0.16081655510049228,
      "grad_norm": 1.1319267749786377,
      "learning_rate": 0.00016783690164582355,
      "loss": 3.583,
      "step": 172520
    },
    {
      "epoch": 0.16082587671857138,
      "grad_norm": 1.1950379610061646,
      "learning_rate": 0.00016783503732372836,
      "loss": 3.4328,
      "step": 172530
    },
    {
      "epoch": 0.16083519833665047,
      "grad_norm": 0.9932364821434021,
      "learning_rate": 0.00016783317300163314,
      "loss": 3.5015,
      "step": 172540
    },
    {
      "epoch": 0.16084451995472956,
      "grad_norm": 1.3183777332305908,
      "learning_rate": 0.00016783130867953794,
      "loss": 3.6261,
      "step": 172550
    },
    {
      "epoch": 0.16085384157280866,
      "grad_norm": 1.2829622030258179,
      "learning_rate": 0.00016782944435744275,
      "loss": 3.5248,
      "step": 172560
    },
    {
      "epoch": 0.16086316319088775,
      "grad_norm": 1.1079102754592896,
      "learning_rate": 0.00016782758003534756,
      "loss": 3.5146,
      "step": 172570
    },
    {
      "epoch": 0.16087248480896685,
      "grad_norm": 1.3468551635742188,
      "learning_rate": 0.00016782571571325236,
      "loss": 3.6888,
      "step": 172580
    },
    {
      "epoch": 0.16088180642704594,
      "grad_norm": 1.2745347023010254,
      "learning_rate": 0.00016782385139115714,
      "loss": 3.6511,
      "step": 172590
    },
    {
      "epoch": 0.160891128045125,
      "grad_norm": 1.0688285827636719,
      "learning_rate": 0.00016782198706906195,
      "loss": 3.6083,
      "step": 172600
    },
    {
      "epoch": 0.1609004496632041,
      "grad_norm": 1.0044816732406616,
      "learning_rate": 0.00016782012274696675,
      "loss": 3.5589,
      "step": 172610
    },
    {
      "epoch": 0.1609097712812832,
      "grad_norm": 1.0040346384048462,
      "learning_rate": 0.00016781825842487156,
      "loss": 3.482,
      "step": 172620
    },
    {
      "epoch": 0.1609190928993623,
      "grad_norm": 0.9091858267784119,
      "learning_rate": 0.00016781639410277637,
      "loss": 3.4583,
      "step": 172630
    },
    {
      "epoch": 0.1609284145174414,
      "grad_norm": 1.3434038162231445,
      "learning_rate": 0.00016781452978068115,
      "loss": 3.5213,
      "step": 172640
    },
    {
      "epoch": 0.16093773613552048,
      "grad_norm": 0.9326375126838684,
      "learning_rate": 0.00016781266545858595,
      "loss": 3.5577,
      "step": 172650
    },
    {
      "epoch": 0.16094705775359958,
      "grad_norm": 1.1877950429916382,
      "learning_rate": 0.00016781080113649076,
      "loss": 3.5315,
      "step": 172660
    },
    {
      "epoch": 0.16095637937167867,
      "grad_norm": 1.0542489290237427,
      "learning_rate": 0.00016780893681439557,
      "loss": 3.43,
      "step": 172670
    },
    {
      "epoch": 0.16096570098975777,
      "grad_norm": 0.8508332967758179,
      "learning_rate": 0.00016780707249230037,
      "loss": 3.4396,
      "step": 172680
    },
    {
      "epoch": 0.16097502260783686,
      "grad_norm": 1.1220614910125732,
      "learning_rate": 0.00016780520817020515,
      "loss": 3.4361,
      "step": 172690
    },
    {
      "epoch": 0.16098434422591593,
      "grad_norm": 0.8466429114341736,
      "learning_rate": 0.00016780334384810996,
      "loss": 3.4997,
      "step": 172700
    },
    {
      "epoch": 0.16099366584399502,
      "grad_norm": 1.1572802066802979,
      "learning_rate": 0.00016780147952601474,
      "loss": 3.4076,
      "step": 172710
    },
    {
      "epoch": 0.16100298746207412,
      "grad_norm": 1.044481873512268,
      "learning_rate": 0.00016779961520391957,
      "loss": 3.6518,
      "step": 172720
    },
    {
      "epoch": 0.1610123090801532,
      "grad_norm": 1.0478276014328003,
      "learning_rate": 0.00016779775088182438,
      "loss": 3.6969,
      "step": 172730
    },
    {
      "epoch": 0.1610216306982323,
      "grad_norm": 1.003405213356018,
      "learning_rate": 0.00016779588655972916,
      "loss": 3.4846,
      "step": 172740
    },
    {
      "epoch": 0.1610309523163114,
      "grad_norm": 1.2266271114349365,
      "learning_rate": 0.00016779402223763396,
      "loss": 3.5093,
      "step": 172750
    },
    {
      "epoch": 0.1610402739343905,
      "grad_norm": 1.2996429204940796,
      "learning_rate": 0.00016779215791553874,
      "loss": 3.627,
      "step": 172760
    },
    {
      "epoch": 0.1610495955524696,
      "grad_norm": 0.8890774250030518,
      "learning_rate": 0.00016779029359344355,
      "loss": 3.7104,
      "step": 172770
    },
    {
      "epoch": 0.16105891717054868,
      "grad_norm": 1.0545800924301147,
      "learning_rate": 0.00016778842927134838,
      "loss": 3.4551,
      "step": 172780
    },
    {
      "epoch": 0.16106823878862778,
      "grad_norm": 0.8904262185096741,
      "learning_rate": 0.00016778656494925316,
      "loss": 3.7618,
      "step": 172790
    },
    {
      "epoch": 0.16107756040670687,
      "grad_norm": 1.0428071022033691,
      "learning_rate": 0.00016778470062715797,
      "loss": 3.7092,
      "step": 172800
    },
    {
      "epoch": 0.16108688202478594,
      "grad_norm": 1.065338373184204,
      "learning_rate": 0.00016778283630506275,
      "loss": 3.6183,
      "step": 172810
    },
    {
      "epoch": 0.16109620364286503,
      "grad_norm": 0.982509434223175,
      "learning_rate": 0.00016778097198296755,
      "loss": 3.3911,
      "step": 172820
    },
    {
      "epoch": 0.16110552526094413,
      "grad_norm": 0.9331887364387512,
      "learning_rate": 0.00016777910766087236,
      "loss": 3.5641,
      "step": 172830
    },
    {
      "epoch": 0.16111484687902322,
      "grad_norm": 0.9598640203475952,
      "learning_rate": 0.00016777724333877716,
      "loss": 3.6309,
      "step": 172840
    },
    {
      "epoch": 0.16112416849710232,
      "grad_norm": 1.061665415763855,
      "learning_rate": 0.00016777537901668197,
      "loss": 3.4896,
      "step": 172850
    },
    {
      "epoch": 0.1611334901151814,
      "grad_norm": 1.1129814386367798,
      "learning_rate": 0.00016777351469458675,
      "loss": 3.5246,
      "step": 172860
    },
    {
      "epoch": 0.1611428117332605,
      "grad_norm": 1.0193158388137817,
      "learning_rate": 0.00016777165037249156,
      "loss": 3.6166,
      "step": 172870
    },
    {
      "epoch": 0.1611521333513396,
      "grad_norm": 1.1551806926727295,
      "learning_rate": 0.00016776978605039636,
      "loss": 3.595,
      "step": 172880
    },
    {
      "epoch": 0.1611614549694187,
      "grad_norm": 1.0988882780075073,
      "learning_rate": 0.00016776792172830117,
      "loss": 3.5188,
      "step": 172890
    },
    {
      "epoch": 0.1611707765874978,
      "grad_norm": 1.0365636348724365,
      "learning_rate": 0.00016776605740620598,
      "loss": 3.5006,
      "step": 172900
    },
    {
      "epoch": 0.16118009820557685,
      "grad_norm": 1.124026894569397,
      "learning_rate": 0.00016776419308411075,
      "loss": 3.4672,
      "step": 172910
    },
    {
      "epoch": 0.16118941982365595,
      "grad_norm": 1.1682392358779907,
      "learning_rate": 0.00016776232876201556,
      "loss": 3.5649,
      "step": 172920
    },
    {
      "epoch": 0.16119874144173504,
      "grad_norm": 1.0487412214279175,
      "learning_rate": 0.00016776046443992037,
      "loss": 3.6405,
      "step": 172930
    },
    {
      "epoch": 0.16120806305981414,
      "grad_norm": 1.4624264240264893,
      "learning_rate": 0.00016775860011782517,
      "loss": 3.62,
      "step": 172940
    },
    {
      "epoch": 0.16121738467789323,
      "grad_norm": 1.0738117694854736,
      "learning_rate": 0.00016775673579572998,
      "loss": 3.5657,
      "step": 172950
    },
    {
      "epoch": 0.16122670629597233,
      "grad_norm": 0.9604001641273499,
      "learning_rate": 0.00016775487147363476,
      "loss": 3.4506,
      "step": 172960
    },
    {
      "epoch": 0.16123602791405142,
      "grad_norm": 0.930454671382904,
      "learning_rate": 0.00016775300715153956,
      "loss": 3.5628,
      "step": 172970
    },
    {
      "epoch": 0.16124534953213052,
      "grad_norm": 1.0037280321121216,
      "learning_rate": 0.00016775114282944437,
      "loss": 3.306,
      "step": 172980
    },
    {
      "epoch": 0.1612546711502096,
      "grad_norm": 1.014983057975769,
      "learning_rate": 0.00016774927850734915,
      "loss": 3.5848,
      "step": 172990
    },
    {
      "epoch": 0.1612639927682887,
      "grad_norm": 1.1643644571304321,
      "learning_rate": 0.00016774741418525398,
      "loss": 3.441,
      "step": 173000
    },
    {
      "epoch": 0.1612733143863678,
      "grad_norm": 1.4439085721969604,
      "learning_rate": 0.00016774554986315876,
      "loss": 3.6996,
      "step": 173010
    },
    {
      "epoch": 0.16128263600444687,
      "grad_norm": 1.1441869735717773,
      "learning_rate": 0.00016774368554106357,
      "loss": 3.3725,
      "step": 173020
    },
    {
      "epoch": 0.16129195762252596,
      "grad_norm": 1.0856355428695679,
      "learning_rate": 0.00016774182121896838,
      "loss": 3.5579,
      "step": 173030
    },
    {
      "epoch": 0.16130127924060506,
      "grad_norm": 1.2613455057144165,
      "learning_rate": 0.00016773995689687315,
      "loss": 3.7057,
      "step": 173040
    },
    {
      "epoch": 0.16131060085868415,
      "grad_norm": 1.0499718189239502,
      "learning_rate": 0.00016773809257477796,
      "loss": 3.4795,
      "step": 173050
    },
    {
      "epoch": 0.16131992247676324,
      "grad_norm": 1.0349938869476318,
      "learning_rate": 0.00016773622825268277,
      "loss": 3.6999,
      "step": 173060
    },
    {
      "epoch": 0.16132924409484234,
      "grad_norm": 0.976146399974823,
      "learning_rate": 0.00016773436393058757,
      "loss": 3.532,
      "step": 173070
    },
    {
      "epoch": 0.16133856571292143,
      "grad_norm": 1.1137430667877197,
      "learning_rate": 0.00016773249960849238,
      "loss": 3.5819,
      "step": 173080
    },
    {
      "epoch": 0.16134788733100053,
      "grad_norm": 0.9690654277801514,
      "learning_rate": 0.00016773063528639716,
      "loss": 3.5074,
      "step": 173090
    },
    {
      "epoch": 0.16135720894907962,
      "grad_norm": 1.0930125713348389,
      "learning_rate": 0.00016772877096430197,
      "loss": 3.5401,
      "step": 173100
    },
    {
      "epoch": 0.16136653056715872,
      "grad_norm": 1.0724947452545166,
      "learning_rate": 0.00016772690664220677,
      "loss": 3.5798,
      "step": 173110
    },
    {
      "epoch": 0.16137585218523778,
      "grad_norm": 0.7619490623474121,
      "learning_rate": 0.00016772504232011158,
      "loss": 3.5859,
      "step": 173120
    },
    {
      "epoch": 0.16138517380331688,
      "grad_norm": 0.9909614324569702,
      "learning_rate": 0.00016772317799801638,
      "loss": 3.4532,
      "step": 173130
    },
    {
      "epoch": 0.16139449542139597,
      "grad_norm": 0.9569826126098633,
      "learning_rate": 0.00016772131367592116,
      "loss": 3.427,
      "step": 173140
    },
    {
      "epoch": 0.16140381703947507,
      "grad_norm": 1.2527023553848267,
      "learning_rate": 0.00016771944935382597,
      "loss": 3.4446,
      "step": 173150
    },
    {
      "epoch": 0.16141313865755416,
      "grad_norm": 1.111926794052124,
      "learning_rate": 0.00016771758503173075,
      "loss": 3.3431,
      "step": 173160
    },
    {
      "epoch": 0.16142246027563326,
      "grad_norm": 1.0416297912597656,
      "learning_rate": 0.00016771572070963558,
      "loss": 3.5855,
      "step": 173170
    },
    {
      "epoch": 0.16143178189371235,
      "grad_norm": 1.2067859172821045,
      "learning_rate": 0.0001677138563875404,
      "loss": 3.5647,
      "step": 173180
    },
    {
      "epoch": 0.16144110351179145,
      "grad_norm": 1.251347541809082,
      "learning_rate": 0.00016771199206544517,
      "loss": 3.4402,
      "step": 173190
    },
    {
      "epoch": 0.16145042512987054,
      "grad_norm": 1.1524274349212646,
      "learning_rate": 0.00016771012774334997,
      "loss": 3.4129,
      "step": 173200
    },
    {
      "epoch": 0.16145974674794963,
      "grad_norm": 0.9818158745765686,
      "learning_rate": 0.00016770826342125475,
      "loss": 3.7016,
      "step": 173210
    },
    {
      "epoch": 0.1614690683660287,
      "grad_norm": 1.082371473312378,
      "learning_rate": 0.00016770639909915956,
      "loss": 3.4685,
      "step": 173220
    },
    {
      "epoch": 0.1614783899841078,
      "grad_norm": 0.8571160435676575,
      "learning_rate": 0.0001677045347770644,
      "loss": 3.6676,
      "step": 173230
    },
    {
      "epoch": 0.1614877116021869,
      "grad_norm": 0.9489167332649231,
      "learning_rate": 0.00016770267045496917,
      "loss": 3.4702,
      "step": 173240
    },
    {
      "epoch": 0.16149703322026598,
      "grad_norm": 1.0485657453536987,
      "learning_rate": 0.00016770080613287398,
      "loss": 3.4825,
      "step": 173250
    },
    {
      "epoch": 0.16150635483834508,
      "grad_norm": 1.1335452795028687,
      "learning_rate": 0.00016769894181077876,
      "loss": 3.4832,
      "step": 173260
    },
    {
      "epoch": 0.16151567645642417,
      "grad_norm": 0.9742951989173889,
      "learning_rate": 0.00016769707748868356,
      "loss": 3.4004,
      "step": 173270
    },
    {
      "epoch": 0.16152499807450327,
      "grad_norm": 1.0939730405807495,
      "learning_rate": 0.00016769521316658837,
      "loss": 3.4613,
      "step": 173280
    },
    {
      "epoch": 0.16153431969258236,
      "grad_norm": 1.0650676488876343,
      "learning_rate": 0.00016769334884449318,
      "loss": 3.5947,
      "step": 173290
    },
    {
      "epoch": 0.16154364131066146,
      "grad_norm": 0.9638115167617798,
      "learning_rate": 0.00016769148452239798,
      "loss": 3.6887,
      "step": 173300
    },
    {
      "epoch": 0.16155296292874055,
      "grad_norm": 1.0223424434661865,
      "learning_rate": 0.00016768962020030276,
      "loss": 3.5421,
      "step": 173310
    },
    {
      "epoch": 0.16156228454681965,
      "grad_norm": 1.0201255083084106,
      "learning_rate": 0.00016768775587820757,
      "loss": 3.5707,
      "step": 173320
    },
    {
      "epoch": 0.1615716061648987,
      "grad_norm": 0.9387025237083435,
      "learning_rate": 0.00016768589155611238,
      "loss": 3.6694,
      "step": 173330
    },
    {
      "epoch": 0.1615809277829778,
      "grad_norm": 1.203497290611267,
      "learning_rate": 0.00016768402723401718,
      "loss": 3.4649,
      "step": 173340
    },
    {
      "epoch": 0.1615902494010569,
      "grad_norm": 1.3401658535003662,
      "learning_rate": 0.000167682162911922,
      "loss": 3.453,
      "step": 173350
    },
    {
      "epoch": 0.161599571019136,
      "grad_norm": 1.2793638706207275,
      "learning_rate": 0.00016768029858982677,
      "loss": 3.7127,
      "step": 173360
    },
    {
      "epoch": 0.1616088926372151,
      "grad_norm": 1.076638102531433,
      "learning_rate": 0.00016767843426773157,
      "loss": 3.4129,
      "step": 173370
    },
    {
      "epoch": 0.16161821425529418,
      "grad_norm": 1.2167885303497314,
      "learning_rate": 0.00016767656994563638,
      "loss": 3.7784,
      "step": 173380
    },
    {
      "epoch": 0.16162753587337328,
      "grad_norm": 1.0553174018859863,
      "learning_rate": 0.00016767470562354119,
      "loss": 3.3549,
      "step": 173390
    },
    {
      "epoch": 0.16163685749145237,
      "grad_norm": 0.872408926486969,
      "learning_rate": 0.000167672841301446,
      "loss": 3.5605,
      "step": 173400
    },
    {
      "epoch": 0.16164617910953147,
      "grad_norm": 0.8268876671791077,
      "learning_rate": 0.00016767097697935077,
      "loss": 3.6288,
      "step": 173410
    },
    {
      "epoch": 0.16165550072761056,
      "grad_norm": 0.9847004413604736,
      "learning_rate": 0.00016766911265725558,
      "loss": 3.4855,
      "step": 173420
    },
    {
      "epoch": 0.16166482234568963,
      "grad_norm": 0.9231530427932739,
      "learning_rate": 0.00016766724833516038,
      "loss": 3.3374,
      "step": 173430
    },
    {
      "epoch": 0.16167414396376872,
      "grad_norm": 1.2394145727157593,
      "learning_rate": 0.00016766538401306516,
      "loss": 3.1975,
      "step": 173440
    },
    {
      "epoch": 0.16168346558184782,
      "grad_norm": 1.1578006744384766,
      "learning_rate": 0.00016766351969097,
      "loss": 3.5288,
      "step": 173450
    },
    {
      "epoch": 0.1616927871999269,
      "grad_norm": 1.1284760236740112,
      "learning_rate": 0.00016766165536887478,
      "loss": 3.7418,
      "step": 173460
    },
    {
      "epoch": 0.161702108818006,
      "grad_norm": 1.0624679327011108,
      "learning_rate": 0.00016765979104677958,
      "loss": 3.4973,
      "step": 173470
    },
    {
      "epoch": 0.1617114304360851,
      "grad_norm": 1.1659601926803589,
      "learning_rate": 0.0001676579267246844,
      "loss": 3.3608,
      "step": 173480
    },
    {
      "epoch": 0.1617207520541642,
      "grad_norm": 1.0243110656738281,
      "learning_rate": 0.00016765606240258917,
      "loss": 3.5245,
      "step": 173490
    },
    {
      "epoch": 0.1617300736722433,
      "grad_norm": 1.0541759729385376,
      "learning_rate": 0.00016765419808049397,
      "loss": 3.5747,
      "step": 173500
    },
    {
      "epoch": 0.16173939529032239,
      "grad_norm": 1.2446538209915161,
      "learning_rate": 0.00016765233375839878,
      "loss": 3.3836,
      "step": 173510
    },
    {
      "epoch": 0.16174871690840148,
      "grad_norm": 1.162255883216858,
      "learning_rate": 0.0001676504694363036,
      "loss": 3.7238,
      "step": 173520
    },
    {
      "epoch": 0.16175803852648057,
      "grad_norm": 1.0397151708602905,
      "learning_rate": 0.0001676486051142084,
      "loss": 3.5034,
      "step": 173530
    },
    {
      "epoch": 0.16176736014455964,
      "grad_norm": 1.0090646743774414,
      "learning_rate": 0.00016764674079211317,
      "loss": 3.5004,
      "step": 173540
    },
    {
      "epoch": 0.16177668176263874,
      "grad_norm": 1.1761834621429443,
      "learning_rate": 0.00016764487647001798,
      "loss": 3.4833,
      "step": 173550
    },
    {
      "epoch": 0.16178600338071783,
      "grad_norm": 0.9096974730491638,
      "learning_rate": 0.00016764301214792278,
      "loss": 3.5503,
      "step": 173560
    },
    {
      "epoch": 0.16179532499879692,
      "grad_norm": 1.0387412309646606,
      "learning_rate": 0.0001676411478258276,
      "loss": 3.4845,
      "step": 173570
    },
    {
      "epoch": 0.16180464661687602,
      "grad_norm": 0.9781978130340576,
      "learning_rate": 0.0001676392835037324,
      "loss": 3.4363,
      "step": 173580
    },
    {
      "epoch": 0.1618139682349551,
      "grad_norm": 0.848558247089386,
      "learning_rate": 0.00016763741918163718,
      "loss": 3.4597,
      "step": 173590
    },
    {
      "epoch": 0.1618232898530342,
      "grad_norm": 1.0730042457580566,
      "learning_rate": 0.00016763555485954198,
      "loss": 3.3445,
      "step": 173600
    },
    {
      "epoch": 0.1618326114711133,
      "grad_norm": 1.0510406494140625,
      "learning_rate": 0.00016763369053744676,
      "loss": 3.5596,
      "step": 173610
    },
    {
      "epoch": 0.1618419330891924,
      "grad_norm": 1.1489816904067993,
      "learning_rate": 0.0001676318262153516,
      "loss": 3.3265,
      "step": 173620
    },
    {
      "epoch": 0.1618512547072715,
      "grad_norm": 1.031068205833435,
      "learning_rate": 0.0001676299618932564,
      "loss": 3.5698,
      "step": 173630
    },
    {
      "epoch": 0.16186057632535056,
      "grad_norm": 1.1880924701690674,
      "learning_rate": 0.00016762809757116118,
      "loss": 3.6581,
      "step": 173640
    },
    {
      "epoch": 0.16186989794342965,
      "grad_norm": 1.1951193809509277,
      "learning_rate": 0.000167626233249066,
      "loss": 3.5736,
      "step": 173650
    },
    {
      "epoch": 0.16187921956150875,
      "grad_norm": 0.8782356977462769,
      "learning_rate": 0.00016762436892697077,
      "loss": 3.5595,
      "step": 173660
    },
    {
      "epoch": 0.16188854117958784,
      "grad_norm": 1.1198338270187378,
      "learning_rate": 0.00016762250460487557,
      "loss": 3.6059,
      "step": 173670
    },
    {
      "epoch": 0.16189786279766694,
      "grad_norm": 1.1192903518676758,
      "learning_rate": 0.00016762064028278038,
      "loss": 3.6374,
      "step": 173680
    },
    {
      "epoch": 0.16190718441574603,
      "grad_norm": 0.9748653173446655,
      "learning_rate": 0.00016761877596068519,
      "loss": 3.7486,
      "step": 173690
    },
    {
      "epoch": 0.16191650603382512,
      "grad_norm": 1.0770773887634277,
      "learning_rate": 0.00016761691163859,
      "loss": 3.4937,
      "step": 173700
    },
    {
      "epoch": 0.16192582765190422,
      "grad_norm": 1.0116194486618042,
      "learning_rate": 0.00016761504731649477,
      "loss": 3.5393,
      "step": 173710
    },
    {
      "epoch": 0.16193514926998331,
      "grad_norm": 1.0104323625564575,
      "learning_rate": 0.00016761318299439958,
      "loss": 3.5125,
      "step": 173720
    },
    {
      "epoch": 0.1619444708880624,
      "grad_norm": 1.2045960426330566,
      "learning_rate": 0.00016761131867230438,
      "loss": 3.6595,
      "step": 173730
    },
    {
      "epoch": 0.1619537925061415,
      "grad_norm": 1.1639024019241333,
      "learning_rate": 0.0001676094543502092,
      "loss": 3.5219,
      "step": 173740
    },
    {
      "epoch": 0.16196311412422057,
      "grad_norm": 1.127300500869751,
      "learning_rate": 0.000167607590028114,
      "loss": 3.5876,
      "step": 173750
    },
    {
      "epoch": 0.16197243574229966,
      "grad_norm": 1.0322517156600952,
      "learning_rate": 0.00016760572570601878,
      "loss": 3.3954,
      "step": 173760
    },
    {
      "epoch": 0.16198175736037876,
      "grad_norm": 1.1659519672393799,
      "learning_rate": 0.00016760386138392358,
      "loss": 3.4052,
      "step": 173770
    },
    {
      "epoch": 0.16199107897845785,
      "grad_norm": 1.1713060140609741,
      "learning_rate": 0.0001676019970618284,
      "loss": 3.4661,
      "step": 173780
    },
    {
      "epoch": 0.16200040059653695,
      "grad_norm": 1.0073689222335815,
      "learning_rate": 0.0001676001327397332,
      "loss": 3.5119,
      "step": 173790
    },
    {
      "epoch": 0.16200972221461604,
      "grad_norm": 1.1136492490768433,
      "learning_rate": 0.000167598268417638,
      "loss": 3.3731,
      "step": 173800
    },
    {
      "epoch": 0.16201904383269514,
      "grad_norm": 0.9198323488235474,
      "learning_rate": 0.00016759640409554278,
      "loss": 3.4918,
      "step": 173810
    },
    {
      "epoch": 0.16202836545077423,
      "grad_norm": 0.9815515279769897,
      "learning_rate": 0.00016759453977344759,
      "loss": 3.4362,
      "step": 173820
    },
    {
      "epoch": 0.16203768706885333,
      "grad_norm": 1.0708221197128296,
      "learning_rate": 0.00016759267545135237,
      "loss": 3.6091,
      "step": 173830
    },
    {
      "epoch": 0.16204700868693242,
      "grad_norm": 1.0091471672058105,
      "learning_rate": 0.0001675908111292572,
      "loss": 3.6865,
      "step": 173840
    },
    {
      "epoch": 0.1620563303050115,
      "grad_norm": 1.1459519863128662,
      "learning_rate": 0.000167588946807162,
      "loss": 3.5043,
      "step": 173850
    },
    {
      "epoch": 0.16206565192309058,
      "grad_norm": 1.1644147634506226,
      "learning_rate": 0.00016758708248506678,
      "loss": 3.6212,
      "step": 173860
    },
    {
      "epoch": 0.16207497354116968,
      "grad_norm": 1.148389458656311,
      "learning_rate": 0.0001675852181629716,
      "loss": 3.5139,
      "step": 173870
    },
    {
      "epoch": 0.16208429515924877,
      "grad_norm": 0.9616312980651855,
      "learning_rate": 0.00016758335384087637,
      "loss": 3.5077,
      "step": 173880
    },
    {
      "epoch": 0.16209361677732786,
      "grad_norm": 0.9092414379119873,
      "learning_rate": 0.00016758148951878118,
      "loss": 3.8408,
      "step": 173890
    },
    {
      "epoch": 0.16210293839540696,
      "grad_norm": 1.1186106204986572,
      "learning_rate": 0.000167579625196686,
      "loss": 3.3338,
      "step": 173900
    },
    {
      "epoch": 0.16211226001348605,
      "grad_norm": 0.923366129398346,
      "learning_rate": 0.0001675777608745908,
      "loss": 3.447,
      "step": 173910
    },
    {
      "epoch": 0.16212158163156515,
      "grad_norm": 1.163640022277832,
      "learning_rate": 0.0001675758965524956,
      "loss": 3.6185,
      "step": 173920
    },
    {
      "epoch": 0.16213090324964424,
      "grad_norm": 1.159255027770996,
      "learning_rate": 0.00016757403223040037,
      "loss": 3.5774,
      "step": 173930
    },
    {
      "epoch": 0.16214022486772334,
      "grad_norm": 0.9567604660987854,
      "learning_rate": 0.00016757216790830518,
      "loss": 3.6666,
      "step": 173940
    },
    {
      "epoch": 0.16214954648580243,
      "grad_norm": 1.1496045589447021,
      "learning_rate": 0.00016757030358621,
      "loss": 3.5325,
      "step": 173950
    },
    {
      "epoch": 0.1621588681038815,
      "grad_norm": 1.119219183921814,
      "learning_rate": 0.0001675684392641148,
      "loss": 3.6002,
      "step": 173960
    },
    {
      "epoch": 0.1621681897219606,
      "grad_norm": 0.9935429096221924,
      "learning_rate": 0.0001675665749420196,
      "loss": 3.5047,
      "step": 173970
    },
    {
      "epoch": 0.1621775113400397,
      "grad_norm": 1.0422853231430054,
      "learning_rate": 0.00016756471061992438,
      "loss": 3.4267,
      "step": 173980
    },
    {
      "epoch": 0.16218683295811878,
      "grad_norm": 1.052101731300354,
      "learning_rate": 0.00016756284629782919,
      "loss": 3.6572,
      "step": 173990
    },
    {
      "epoch": 0.16219615457619788,
      "grad_norm": 0.937035083770752,
      "learning_rate": 0.000167560981975734,
      "loss": 3.5248,
      "step": 174000
    },
    {
      "epoch": 0.16220547619427697,
      "grad_norm": 1.073370337486267,
      "learning_rate": 0.0001675591176536388,
      "loss": 3.5169,
      "step": 174010
    },
    {
      "epoch": 0.16221479781235607,
      "grad_norm": 1.121048927307129,
      "learning_rate": 0.0001675572533315436,
      "loss": 3.6656,
      "step": 174020
    },
    {
      "epoch": 0.16222411943043516,
      "grad_norm": 0.8772113919258118,
      "learning_rate": 0.00016755538900944838,
      "loss": 3.5316,
      "step": 174030
    },
    {
      "epoch": 0.16223344104851425,
      "grad_norm": 1.0006544589996338,
      "learning_rate": 0.0001675535246873532,
      "loss": 3.5058,
      "step": 174040
    },
    {
      "epoch": 0.16224276266659335,
      "grad_norm": 1.234677791595459,
      "learning_rate": 0.000167551660365258,
      "loss": 3.5785,
      "step": 174050
    },
    {
      "epoch": 0.16225208428467242,
      "grad_norm": 1.0123251676559448,
      "learning_rate": 0.00016754979604316277,
      "loss": 3.5052,
      "step": 174060
    },
    {
      "epoch": 0.1622614059027515,
      "grad_norm": 1.0569899082183838,
      "learning_rate": 0.0001675479317210676,
      "loss": 3.5851,
      "step": 174070
    },
    {
      "epoch": 0.1622707275208306,
      "grad_norm": 1.1808807849884033,
      "learning_rate": 0.0001675460673989724,
      "loss": 3.6807,
      "step": 174080
    },
    {
      "epoch": 0.1622800491389097,
      "grad_norm": 0.933255136013031,
      "learning_rate": 0.0001675442030768772,
      "loss": 3.6991,
      "step": 174090
    },
    {
      "epoch": 0.1622893707569888,
      "grad_norm": 0.9929753541946411,
      "learning_rate": 0.000167542338754782,
      "loss": 3.6056,
      "step": 174100
    },
    {
      "epoch": 0.1622986923750679,
      "grad_norm": 1.0586751699447632,
      "learning_rate": 0.00016754047443268678,
      "loss": 3.6894,
      "step": 174110
    },
    {
      "epoch": 0.16230801399314698,
      "grad_norm": 0.9542893171310425,
      "learning_rate": 0.00016753861011059159,
      "loss": 3.5907,
      "step": 174120
    },
    {
      "epoch": 0.16231733561122608,
      "grad_norm": 1.1926167011260986,
      "learning_rate": 0.0001675367457884964,
      "loss": 3.5845,
      "step": 174130
    },
    {
      "epoch": 0.16232665722930517,
      "grad_norm": 0.9439831972122192,
      "learning_rate": 0.0001675348814664012,
      "loss": 3.684,
      "step": 174140
    },
    {
      "epoch": 0.16233597884738427,
      "grad_norm": 0.9742633700370789,
      "learning_rate": 0.000167533017144306,
      "loss": 3.5504,
      "step": 174150
    },
    {
      "epoch": 0.16234530046546336,
      "grad_norm": 1.0514363050460815,
      "learning_rate": 0.00016753115282221078,
      "loss": 3.6307,
      "step": 174160
    },
    {
      "epoch": 0.16235462208354243,
      "grad_norm": 0.827884316444397,
      "learning_rate": 0.0001675292885001156,
      "loss": 3.5859,
      "step": 174170
    },
    {
      "epoch": 0.16236394370162152,
      "grad_norm": 0.9557058215141296,
      "learning_rate": 0.0001675274241780204,
      "loss": 3.4935,
      "step": 174180
    },
    {
      "epoch": 0.16237326531970062,
      "grad_norm": 1.2708028554916382,
      "learning_rate": 0.0001675255598559252,
      "loss": 3.5377,
      "step": 174190
    },
    {
      "epoch": 0.1623825869377797,
      "grad_norm": 1.0001472234725952,
      "learning_rate": 0.00016752369553383,
      "loss": 3.6097,
      "step": 174200
    },
    {
      "epoch": 0.1623919085558588,
      "grad_norm": 1.1433659791946411,
      "learning_rate": 0.0001675218312117348,
      "loss": 3.54,
      "step": 174210
    },
    {
      "epoch": 0.1624012301739379,
      "grad_norm": 0.9389582872390747,
      "learning_rate": 0.0001675199668896396,
      "loss": 3.6448,
      "step": 174220
    },
    {
      "epoch": 0.162410551792017,
      "grad_norm": 1.1928297281265259,
      "learning_rate": 0.0001675181025675444,
      "loss": 3.7651,
      "step": 174230
    },
    {
      "epoch": 0.1624198734100961,
      "grad_norm": 1.0724194049835205,
      "learning_rate": 0.0001675162382454492,
      "loss": 3.4969,
      "step": 174240
    },
    {
      "epoch": 0.16242919502817518,
      "grad_norm": 0.9767130017280579,
      "learning_rate": 0.000167514373923354,
      "loss": 3.6359,
      "step": 174250
    },
    {
      "epoch": 0.16243851664625428,
      "grad_norm": 1.058132290840149,
      "learning_rate": 0.0001675125096012588,
      "loss": 3.7222,
      "step": 174260
    },
    {
      "epoch": 0.16244783826433334,
      "grad_norm": 1.051514983177185,
      "learning_rate": 0.0001675106452791636,
      "loss": 3.4785,
      "step": 174270
    },
    {
      "epoch": 0.16245715988241244,
      "grad_norm": 1.273054838180542,
      "learning_rate": 0.00016750878095706838,
      "loss": 3.6302,
      "step": 174280
    },
    {
      "epoch": 0.16246648150049153,
      "grad_norm": 0.9671456813812256,
      "learning_rate": 0.0001675069166349732,
      "loss": 3.4711,
      "step": 174290
    },
    {
      "epoch": 0.16247580311857063,
      "grad_norm": 1.0131100416183472,
      "learning_rate": 0.00016750505231287802,
      "loss": 3.3591,
      "step": 174300
    },
    {
      "epoch": 0.16248512473664972,
      "grad_norm": 1.130884051322937,
      "learning_rate": 0.0001675031879907828,
      "loss": 3.6035,
      "step": 174310
    },
    {
      "epoch": 0.16249444635472882,
      "grad_norm": Infinity,
      "learning_rate": 0.0001675013236686876,
      "loss": 3.5263,
      "step": 174320
    },
    {
      "epoch": 0.1625037679728079,
      "grad_norm": 1.0466920137405396,
      "learning_rate": 0.00016749945934659238,
      "loss": 3.6525,
      "step": 174330
    },
    {
      "epoch": 0.162513089590887,
      "grad_norm": 1.0323399305343628,
      "learning_rate": 0.0001674975950244972,
      "loss": 3.4471,
      "step": 174340
    },
    {
      "epoch": 0.1625224112089661,
      "grad_norm": 1.2844505310058594,
      "learning_rate": 0.00016749573070240202,
      "loss": 3.555,
      "step": 174350
    },
    {
      "epoch": 0.1625317328270452,
      "grad_norm": 1.074537992477417,
      "learning_rate": 0.0001674938663803068,
      "loss": 3.4922,
      "step": 174360
    },
    {
      "epoch": 0.1625410544451243,
      "grad_norm": 0.9363279938697815,
      "learning_rate": 0.0001674920020582116,
      "loss": 3.3512,
      "step": 174370
    },
    {
      "epoch": 0.16255037606320336,
      "grad_norm": 1.1624791622161865,
      "learning_rate": 0.0001674901377361164,
      "loss": 3.3854,
      "step": 174380
    },
    {
      "epoch": 0.16255969768128245,
      "grad_norm": 0.8956302404403687,
      "learning_rate": 0.0001674882734140212,
      "loss": 3.4283,
      "step": 174390
    },
    {
      "epoch": 0.16256901929936154,
      "grad_norm": 1.0571717023849487,
      "learning_rate": 0.000167486409091926,
      "loss": 3.8054,
      "step": 174400
    },
    {
      "epoch": 0.16257834091744064,
      "grad_norm": 0.9243311285972595,
      "learning_rate": 0.0001674845447698308,
      "loss": 3.3903,
      "step": 174410
    },
    {
      "epoch": 0.16258766253551973,
      "grad_norm": 1.0770989656448364,
      "learning_rate": 0.0001674826804477356,
      "loss": 3.3569,
      "step": 174420
    },
    {
      "epoch": 0.16259698415359883,
      "grad_norm": 1.1631665229797363,
      "learning_rate": 0.0001674808161256404,
      "loss": 3.6415,
      "step": 174430
    },
    {
      "epoch": 0.16260630577167792,
      "grad_norm": 0.883770763874054,
      "learning_rate": 0.0001674789518035452,
      "loss": 3.5051,
      "step": 174440
    },
    {
      "epoch": 0.16261562738975702,
      "grad_norm": 1.0577107667922974,
      "learning_rate": 0.00016747708748145,
      "loss": 3.5647,
      "step": 174450
    },
    {
      "epoch": 0.1626249490078361,
      "grad_norm": 0.846813976764679,
      "learning_rate": 0.0001674752231593548,
      "loss": 3.7073,
      "step": 174460
    },
    {
      "epoch": 0.1626342706259152,
      "grad_norm": 1.141718864440918,
      "learning_rate": 0.00016747335883725962,
      "loss": 3.431,
      "step": 174470
    },
    {
      "epoch": 0.16264359224399427,
      "grad_norm": 1.0634691715240479,
      "learning_rate": 0.0001674714945151644,
      "loss": 3.5901,
      "step": 174480
    },
    {
      "epoch": 0.16265291386207337,
      "grad_norm": 0.9656170010566711,
      "learning_rate": 0.0001674696301930692,
      "loss": 3.5932,
      "step": 174490
    },
    {
      "epoch": 0.16266223548015246,
      "grad_norm": 0.9344741702079773,
      "learning_rate": 0.000167467765870974,
      "loss": 3.3843,
      "step": 174500
    },
    {
      "epoch": 0.16267155709823156,
      "grad_norm": 0.9417102932929993,
      "learning_rate": 0.0001674659015488788,
      "loss": 3.62,
      "step": 174510
    },
    {
      "epoch": 0.16268087871631065,
      "grad_norm": 1.0128718614578247,
      "learning_rate": 0.00016746403722678362,
      "loss": 3.7778,
      "step": 174520
    },
    {
      "epoch": 0.16269020033438975,
      "grad_norm": 1.10109281539917,
      "learning_rate": 0.0001674621729046884,
      "loss": 3.6132,
      "step": 174530
    },
    {
      "epoch": 0.16269952195246884,
      "grad_norm": 0.9708576202392578,
      "learning_rate": 0.0001674603085825932,
      "loss": 3.5908,
      "step": 174540
    },
    {
      "epoch": 0.16270884357054793,
      "grad_norm": 1.1014615297317505,
      "learning_rate": 0.000167458444260498,
      "loss": 3.6226,
      "step": 174550
    },
    {
      "epoch": 0.16271816518862703,
      "grad_norm": 1.037599802017212,
      "learning_rate": 0.0001674565799384028,
      "loss": 3.6524,
      "step": 174560
    },
    {
      "epoch": 0.16272748680670612,
      "grad_norm": 1.1944808959960938,
      "learning_rate": 0.0001674547156163076,
      "loss": 3.5519,
      "step": 174570
    },
    {
      "epoch": 0.16273680842478522,
      "grad_norm": 0.8586918711662292,
      "learning_rate": 0.0001674528512942124,
      "loss": 3.5633,
      "step": 174580
    },
    {
      "epoch": 0.16274613004286428,
      "grad_norm": 1.0372933149337769,
      "learning_rate": 0.0001674509869721172,
      "loss": 3.4714,
      "step": 174590
    },
    {
      "epoch": 0.16275545166094338,
      "grad_norm": 0.9948291778564453,
      "learning_rate": 0.00016744912265002202,
      "loss": 3.6364,
      "step": 174600
    },
    {
      "epoch": 0.16276477327902247,
      "grad_norm": 0.8962133526802063,
      "learning_rate": 0.0001674472583279268,
      "loss": 3.6913,
      "step": 174610
    },
    {
      "epoch": 0.16277409489710157,
      "grad_norm": 1.2691606283187866,
      "learning_rate": 0.0001674453940058316,
      "loss": 3.4482,
      "step": 174620
    },
    {
      "epoch": 0.16278341651518066,
      "grad_norm": 1.2490359544754028,
      "learning_rate": 0.0001674435296837364,
      "loss": 3.3872,
      "step": 174630
    },
    {
      "epoch": 0.16279273813325976,
      "grad_norm": 1.0676499605178833,
      "learning_rate": 0.00016744166536164122,
      "loss": 3.6968,
      "step": 174640
    },
    {
      "epoch": 0.16280205975133885,
      "grad_norm": 1.0077494382858276,
      "learning_rate": 0.00016743980103954602,
      "loss": 3.6619,
      "step": 174650
    },
    {
      "epoch": 0.16281138136941795,
      "grad_norm": 0.9079840779304504,
      "learning_rate": 0.0001674379367174508,
      "loss": 3.5422,
      "step": 174660
    },
    {
      "epoch": 0.16282070298749704,
      "grad_norm": 0.9390562176704407,
      "learning_rate": 0.0001674360723953556,
      "loss": 3.5423,
      "step": 174670
    },
    {
      "epoch": 0.16283002460557613,
      "grad_norm": 1.245996117591858,
      "learning_rate": 0.00016743420807326041,
      "loss": 3.4271,
      "step": 174680
    },
    {
      "epoch": 0.1628393462236552,
      "grad_norm": 1.152191400527954,
      "learning_rate": 0.00016743234375116522,
      "loss": 3.5004,
      "step": 174690
    },
    {
      "epoch": 0.1628486678417343,
      "grad_norm": 1.1525568962097168,
      "learning_rate": 0.00016743047942907003,
      "loss": 3.3922,
      "step": 174700
    },
    {
      "epoch": 0.1628579894598134,
      "grad_norm": 1.2244138717651367,
      "learning_rate": 0.0001674286151069748,
      "loss": 3.5142,
      "step": 174710
    },
    {
      "epoch": 0.16286731107789248,
      "grad_norm": 1.0292965173721313,
      "learning_rate": 0.0001674267507848796,
      "loss": 3.633,
      "step": 174720
    },
    {
      "epoch": 0.16287663269597158,
      "grad_norm": 1.2235974073410034,
      "learning_rate": 0.0001674248864627844,
      "loss": 3.6429,
      "step": 174730
    },
    {
      "epoch": 0.16288595431405067,
      "grad_norm": 1.1329911947250366,
      "learning_rate": 0.00016742302214068922,
      "loss": 3.6307,
      "step": 174740
    },
    {
      "epoch": 0.16289527593212977,
      "grad_norm": 1.0625687837600708,
      "learning_rate": 0.00016742115781859403,
      "loss": 3.5811,
      "step": 174750
    },
    {
      "epoch": 0.16290459755020886,
      "grad_norm": 1.0210615396499634,
      "learning_rate": 0.0001674192934964988,
      "loss": 3.3942,
      "step": 174760
    },
    {
      "epoch": 0.16291391916828796,
      "grad_norm": 1.1927162408828735,
      "learning_rate": 0.00016741742917440362,
      "loss": 3.7497,
      "step": 174770
    },
    {
      "epoch": 0.16292324078636705,
      "grad_norm": 1.2116332054138184,
      "learning_rate": 0.0001674155648523084,
      "loss": 3.4045,
      "step": 174780
    },
    {
      "epoch": 0.16293256240444612,
      "grad_norm": 0.9357687830924988,
      "learning_rate": 0.0001674137005302132,
      "loss": 3.2865,
      "step": 174790
    },
    {
      "epoch": 0.1629418840225252,
      "grad_norm": 0.9312347769737244,
      "learning_rate": 0.00016741183620811804,
      "loss": 3.6276,
      "step": 174800
    },
    {
      "epoch": 0.1629512056406043,
      "grad_norm": 0.9742246866226196,
      "learning_rate": 0.00016740997188602281,
      "loss": 3.6843,
      "step": 174810
    },
    {
      "epoch": 0.1629605272586834,
      "grad_norm": 0.8970563411712646,
      "learning_rate": 0.00016740810756392762,
      "loss": 3.419,
      "step": 174820
    },
    {
      "epoch": 0.1629698488767625,
      "grad_norm": 1.054429054260254,
      "learning_rate": 0.0001674062432418324,
      "loss": 3.4494,
      "step": 174830
    },
    {
      "epoch": 0.1629791704948416,
      "grad_norm": 1.0812995433807373,
      "learning_rate": 0.0001674043789197372,
      "loss": 3.3709,
      "step": 174840
    },
    {
      "epoch": 0.16298849211292069,
      "grad_norm": 0.9417543411254883,
      "learning_rate": 0.000167402514597642,
      "loss": 3.6095,
      "step": 174850
    },
    {
      "epoch": 0.16299781373099978,
      "grad_norm": 1.33489990234375,
      "learning_rate": 0.00016740065027554682,
      "loss": 3.4241,
      "step": 174860
    },
    {
      "epoch": 0.16300713534907887,
      "grad_norm": 0.9348828196525574,
      "learning_rate": 0.00016739878595345163,
      "loss": 3.7349,
      "step": 174870
    },
    {
      "epoch": 0.16301645696715797,
      "grad_norm": 1.1801562309265137,
      "learning_rate": 0.0001673969216313564,
      "loss": 3.5064,
      "step": 174880
    },
    {
      "epoch": 0.16302577858523706,
      "grad_norm": 1.0576727390289307,
      "learning_rate": 0.0001673950573092612,
      "loss": 3.6455,
      "step": 174890
    },
    {
      "epoch": 0.16303510020331613,
      "grad_norm": 1.1064038276672363,
      "learning_rate": 0.00016739319298716602,
      "loss": 3.7173,
      "step": 174900
    },
    {
      "epoch": 0.16304442182139522,
      "grad_norm": 1.2244728803634644,
      "learning_rate": 0.00016739132866507082,
      "loss": 3.5084,
      "step": 174910
    },
    {
      "epoch": 0.16305374343947432,
      "grad_norm": 1.084245204925537,
      "learning_rate": 0.00016738946434297563,
      "loss": 3.5515,
      "step": 174920
    },
    {
      "epoch": 0.1630630650575534,
      "grad_norm": 1.0035912990570068,
      "learning_rate": 0.0001673876000208804,
      "loss": 3.5432,
      "step": 174930
    },
    {
      "epoch": 0.1630723866756325,
      "grad_norm": 1.0827351808547974,
      "learning_rate": 0.00016738573569878522,
      "loss": 3.486,
      "step": 174940
    },
    {
      "epoch": 0.1630817082937116,
      "grad_norm": 0.8697677850723267,
      "learning_rate": 0.00016738387137669002,
      "loss": 3.6986,
      "step": 174950
    },
    {
      "epoch": 0.1630910299117907,
      "grad_norm": 1.1358377933502197,
      "learning_rate": 0.0001673820070545948,
      "loss": 3.5651,
      "step": 174960
    },
    {
      "epoch": 0.1631003515298698,
      "grad_norm": 1.2709870338439941,
      "learning_rate": 0.00016738014273249963,
      "loss": 3.6954,
      "step": 174970
    },
    {
      "epoch": 0.16310967314794889,
      "grad_norm": 0.9923626184463501,
      "learning_rate": 0.0001673782784104044,
      "loss": 3.5566,
      "step": 174980
    },
    {
      "epoch": 0.16311899476602798,
      "grad_norm": 0.8971962928771973,
      "learning_rate": 0.00016737641408830922,
      "loss": 3.5566,
      "step": 174990
    },
    {
      "epoch": 0.16312831638410705,
      "grad_norm": 1.0964173078536987,
      "learning_rate": 0.00016737454976621403,
      "loss": 3.5747,
      "step": 175000
    },
    {
      "epoch": 0.16313763800218614,
      "grad_norm": 0.9162952899932861,
      "learning_rate": 0.0001673726854441188,
      "loss": 3.5155,
      "step": 175010
    },
    {
      "epoch": 0.16314695962026524,
      "grad_norm": 1.0073130130767822,
      "learning_rate": 0.0001673708211220236,
      "loss": 3.5813,
      "step": 175020
    },
    {
      "epoch": 0.16315628123834433,
      "grad_norm": 0.9101647734642029,
      "learning_rate": 0.00016736895679992842,
      "loss": 3.3536,
      "step": 175030
    },
    {
      "epoch": 0.16316560285642343,
      "grad_norm": 0.9773883819580078,
      "learning_rate": 0.00016736709247783322,
      "loss": 3.5261,
      "step": 175040
    },
    {
      "epoch": 0.16317492447450252,
      "grad_norm": 1.1340218782424927,
      "learning_rate": 0.000167365228155738,
      "loss": 3.5265,
      "step": 175050
    },
    {
      "epoch": 0.16318424609258161,
      "grad_norm": 1.0151335000991821,
      "learning_rate": 0.0001673633638336428,
      "loss": 3.3737,
      "step": 175060
    },
    {
      "epoch": 0.1631935677106607,
      "grad_norm": 0.9384101033210754,
      "learning_rate": 0.00016736149951154762,
      "loss": 3.5104,
      "step": 175070
    },
    {
      "epoch": 0.1632028893287398,
      "grad_norm": 1.573607087135315,
      "learning_rate": 0.00016735963518945242,
      "loss": 3.3956,
      "step": 175080
    },
    {
      "epoch": 0.1632122109468189,
      "grad_norm": 1.0784046649932861,
      "learning_rate": 0.00016735777086735723,
      "loss": 3.5371,
      "step": 175090
    },
    {
      "epoch": 0.163221532564898,
      "grad_norm": 1.2966405153274536,
      "learning_rate": 0.000167355906545262,
      "loss": 3.4532,
      "step": 175100
    },
    {
      "epoch": 0.16323085418297706,
      "grad_norm": 1.147736668586731,
      "learning_rate": 0.00016735404222316681,
      "loss": 3.7156,
      "step": 175110
    },
    {
      "epoch": 0.16324017580105615,
      "grad_norm": 1.0864015817642212,
      "learning_rate": 0.00016735217790107162,
      "loss": 3.5138,
      "step": 175120
    },
    {
      "epoch": 0.16324949741913525,
      "grad_norm": 1.199693202972412,
      "learning_rate": 0.00016735031357897643,
      "loss": 3.624,
      "step": 175130
    },
    {
      "epoch": 0.16325881903721434,
      "grad_norm": 0.7840954661369324,
      "learning_rate": 0.00016734844925688123,
      "loss": 3.4832,
      "step": 175140
    },
    {
      "epoch": 0.16326814065529344,
      "grad_norm": 1.0421379804611206,
      "learning_rate": 0.000167346584934786,
      "loss": 3.4835,
      "step": 175150
    },
    {
      "epoch": 0.16327746227337253,
      "grad_norm": 1.0662291049957275,
      "learning_rate": 0.00016734472061269082,
      "loss": 3.5584,
      "step": 175160
    },
    {
      "epoch": 0.16328678389145163,
      "grad_norm": 1.115380048751831,
      "learning_rate": 0.00016734285629059562,
      "loss": 3.5617,
      "step": 175170
    },
    {
      "epoch": 0.16329610550953072,
      "grad_norm": 1.1992651224136353,
      "learning_rate": 0.0001673409919685004,
      "loss": 3.6752,
      "step": 175180
    },
    {
      "epoch": 0.16330542712760981,
      "grad_norm": 1.024754285812378,
      "learning_rate": 0.00016733912764640524,
      "loss": 3.7084,
      "step": 175190
    },
    {
      "epoch": 0.1633147487456889,
      "grad_norm": 1.1464102268218994,
      "learning_rate": 0.00016733726332431002,
      "loss": 3.4785,
      "step": 175200
    },
    {
      "epoch": 0.16332407036376798,
      "grad_norm": 0.9945498108863831,
      "learning_rate": 0.00016733539900221482,
      "loss": 3.4884,
      "step": 175210
    },
    {
      "epoch": 0.16333339198184707,
      "grad_norm": 0.8747284412384033,
      "learning_rate": 0.00016733353468011963,
      "loss": 3.6551,
      "step": 175220
    },
    {
      "epoch": 0.16334271359992616,
      "grad_norm": 1.0100451707839966,
      "learning_rate": 0.0001673316703580244,
      "loss": 3.7606,
      "step": 175230
    },
    {
      "epoch": 0.16335203521800526,
      "grad_norm": 0.9931305646896362,
      "learning_rate": 0.00016732980603592921,
      "loss": 3.4807,
      "step": 175240
    },
    {
      "epoch": 0.16336135683608435,
      "grad_norm": 0.9102900624275208,
      "learning_rate": 0.00016732794171383402,
      "loss": 3.4872,
      "step": 175250
    },
    {
      "epoch": 0.16337067845416345,
      "grad_norm": 1.0197241306304932,
      "learning_rate": 0.00016732607739173883,
      "loss": 3.6019,
      "step": 175260
    },
    {
      "epoch": 0.16338000007224254,
      "grad_norm": 1.3106310367584229,
      "learning_rate": 0.00016732421306964363,
      "loss": 3.5816,
      "step": 175270
    },
    {
      "epoch": 0.16338932169032164,
      "grad_norm": 1.0151516199111938,
      "learning_rate": 0.0001673223487475484,
      "loss": 3.5618,
      "step": 175280
    },
    {
      "epoch": 0.16339864330840073,
      "grad_norm": 0.9001022577285767,
      "learning_rate": 0.00016732048442545322,
      "loss": 3.6717,
      "step": 175290
    },
    {
      "epoch": 0.16340796492647983,
      "grad_norm": 0.8727662563323975,
      "learning_rate": 0.00016731862010335803,
      "loss": 3.4607,
      "step": 175300
    },
    {
      "epoch": 0.16341728654455892,
      "grad_norm": 1.040794849395752,
      "learning_rate": 0.00016731675578126283,
      "loss": 3.3951,
      "step": 175310
    },
    {
      "epoch": 0.163426608162638,
      "grad_norm": 1.0044971704483032,
      "learning_rate": 0.00016731489145916764,
      "loss": 3.5038,
      "step": 175320
    },
    {
      "epoch": 0.16343592978071708,
      "grad_norm": 0.842902660369873,
      "learning_rate": 0.00016731302713707242,
      "loss": 3.5694,
      "step": 175330
    },
    {
      "epoch": 0.16344525139879618,
      "grad_norm": 0.8672257661819458,
      "learning_rate": 0.00016731116281497722,
      "loss": 3.7052,
      "step": 175340
    },
    {
      "epoch": 0.16345457301687527,
      "grad_norm": 1.0117204189300537,
      "learning_rate": 0.000167309298492882,
      "loss": 3.4983,
      "step": 175350
    },
    {
      "epoch": 0.16346389463495437,
      "grad_norm": 0.9252756237983704,
      "learning_rate": 0.00016730743417078684,
      "loss": 3.591,
      "step": 175360
    },
    {
      "epoch": 0.16347321625303346,
      "grad_norm": 0.9745621085166931,
      "learning_rate": 0.00016730556984869164,
      "loss": 3.5486,
      "step": 175370
    },
    {
      "epoch": 0.16348253787111255,
      "grad_norm": 1.0546565055847168,
      "learning_rate": 0.00016730370552659642,
      "loss": 3.6702,
      "step": 175380
    },
    {
      "epoch": 0.16349185948919165,
      "grad_norm": 1.0433883666992188,
      "learning_rate": 0.00016730184120450123,
      "loss": 3.5847,
      "step": 175390
    },
    {
      "epoch": 0.16350118110727074,
      "grad_norm": 1.2416772842407227,
      "learning_rate": 0.000167299976882406,
      "loss": 3.6327,
      "step": 175400
    },
    {
      "epoch": 0.16351050272534984,
      "grad_norm": 1.1583266258239746,
      "learning_rate": 0.0001672981125603108,
      "loss": 3.5737,
      "step": 175410
    },
    {
      "epoch": 0.1635198243434289,
      "grad_norm": 1.1391544342041016,
      "learning_rate": 0.00016729624823821565,
      "loss": 3.6097,
      "step": 175420
    },
    {
      "epoch": 0.163529145961508,
      "grad_norm": 1.2100168466567993,
      "learning_rate": 0.00016729438391612043,
      "loss": 3.4751,
      "step": 175430
    },
    {
      "epoch": 0.1635384675795871,
      "grad_norm": 1.189939022064209,
      "learning_rate": 0.00016729251959402523,
      "loss": 3.2913,
      "step": 175440
    },
    {
      "epoch": 0.1635477891976662,
      "grad_norm": 0.986159086227417,
      "learning_rate": 0.00016729065527193,
      "loss": 3.3163,
      "step": 175450
    },
    {
      "epoch": 0.16355711081574528,
      "grad_norm": 1.0153902769088745,
      "learning_rate": 0.00016728879094983482,
      "loss": 3.6857,
      "step": 175460
    },
    {
      "epoch": 0.16356643243382438,
      "grad_norm": 1.163685917854309,
      "learning_rate": 0.00016728692662773962,
      "loss": 3.6757,
      "step": 175470
    },
    {
      "epoch": 0.16357575405190347,
      "grad_norm": 1.0675580501556396,
      "learning_rate": 0.00016728506230564443,
      "loss": 3.7662,
      "step": 175480
    },
    {
      "epoch": 0.16358507566998257,
      "grad_norm": 1.1571272611618042,
      "learning_rate": 0.00016728319798354924,
      "loss": 3.4732,
      "step": 175490
    },
    {
      "epoch": 0.16359439728806166,
      "grad_norm": 1.11605703830719,
      "learning_rate": 0.00016728133366145402,
      "loss": 3.6653,
      "step": 175500
    },
    {
      "epoch": 0.16360371890614075,
      "grad_norm": 1.2155166864395142,
      "learning_rate": 0.00016727946933935882,
      "loss": 3.573,
      "step": 175510
    },
    {
      "epoch": 0.16361304052421985,
      "grad_norm": 1.1813873052597046,
      "learning_rate": 0.00016727760501726363,
      "loss": 3.5489,
      "step": 175520
    },
    {
      "epoch": 0.16362236214229892,
      "grad_norm": 1.0067288875579834,
      "learning_rate": 0.00016727574069516843,
      "loss": 3.7052,
      "step": 175530
    },
    {
      "epoch": 0.163631683760378,
      "grad_norm": 1.1780688762664795,
      "learning_rate": 0.00016727387637307324,
      "loss": 3.6327,
      "step": 175540
    },
    {
      "epoch": 0.1636410053784571,
      "grad_norm": 0.9057259559631348,
      "learning_rate": 0.00016727201205097802,
      "loss": 3.3754,
      "step": 175550
    },
    {
      "epoch": 0.1636503269965362,
      "grad_norm": 0.999206006526947,
      "learning_rate": 0.00016727014772888283,
      "loss": 3.5099,
      "step": 175560
    },
    {
      "epoch": 0.1636596486146153,
      "grad_norm": 1.1646109819412231,
      "learning_rate": 0.00016726828340678763,
      "loss": 3.5142,
      "step": 175570
    },
    {
      "epoch": 0.1636689702326944,
      "grad_norm": 1.2954328060150146,
      "learning_rate": 0.00016726641908469244,
      "loss": 3.622,
      "step": 175580
    },
    {
      "epoch": 0.16367829185077348,
      "grad_norm": 1.2705390453338623,
      "learning_rate": 0.00016726455476259725,
      "loss": 3.621,
      "step": 175590
    },
    {
      "epoch": 0.16368761346885258,
      "grad_norm": 1.1466827392578125,
      "learning_rate": 0.00016726269044050202,
      "loss": 3.4617,
      "step": 175600
    },
    {
      "epoch": 0.16369693508693167,
      "grad_norm": 1.0987193584442139,
      "learning_rate": 0.00016726082611840683,
      "loss": 3.4508,
      "step": 175610
    },
    {
      "epoch": 0.16370625670501077,
      "grad_norm": 1.081048846244812,
      "learning_rate": 0.00016725896179631164,
      "loss": 3.5789,
      "step": 175620
    },
    {
      "epoch": 0.16371557832308983,
      "grad_norm": 1.0059059858322144,
      "learning_rate": 0.00016725709747421642,
      "loss": 3.5795,
      "step": 175630
    },
    {
      "epoch": 0.16372489994116893,
      "grad_norm": 1.161468505859375,
      "learning_rate": 0.00016725523315212125,
      "loss": 3.5157,
      "step": 175640
    },
    {
      "epoch": 0.16373422155924802,
      "grad_norm": 1.183698296546936,
      "learning_rate": 0.00016725336883002603,
      "loss": 3.3904,
      "step": 175650
    },
    {
      "epoch": 0.16374354317732712,
      "grad_norm": 1.0162338018417358,
      "learning_rate": 0.00016725150450793084,
      "loss": 3.5357,
      "step": 175660
    },
    {
      "epoch": 0.1637528647954062,
      "grad_norm": 1.1722182035446167,
      "learning_rate": 0.00016724964018583564,
      "loss": 3.563,
      "step": 175670
    },
    {
      "epoch": 0.1637621864134853,
      "grad_norm": 1.1595112085342407,
      "learning_rate": 0.00016724777586374042,
      "loss": 3.4364,
      "step": 175680
    },
    {
      "epoch": 0.1637715080315644,
      "grad_norm": 0.8559812307357788,
      "learning_rate": 0.00016724591154164523,
      "loss": 3.587,
      "step": 175690
    },
    {
      "epoch": 0.1637808296496435,
      "grad_norm": 0.9638786315917969,
      "learning_rate": 0.00016724404721955003,
      "loss": 3.3398,
      "step": 175700
    },
    {
      "epoch": 0.1637901512677226,
      "grad_norm": 0.8533244729042053,
      "learning_rate": 0.00016724218289745484,
      "loss": 3.4381,
      "step": 175710
    },
    {
      "epoch": 0.16379947288580168,
      "grad_norm": 0.9245637655258179,
      "learning_rate": 0.00016724031857535965,
      "loss": 3.593,
      "step": 175720
    },
    {
      "epoch": 0.16380879450388078,
      "grad_norm": 1.0201518535614014,
      "learning_rate": 0.00016723845425326443,
      "loss": 3.6059,
      "step": 175730
    },
    {
      "epoch": 0.16381811612195984,
      "grad_norm": 1.0344089269638062,
      "learning_rate": 0.00016723658993116923,
      "loss": 3.4726,
      "step": 175740
    },
    {
      "epoch": 0.16382743774003894,
      "grad_norm": 0.9709367752075195,
      "learning_rate": 0.00016723472560907404,
      "loss": 3.4006,
      "step": 175750
    },
    {
      "epoch": 0.16383675935811803,
      "grad_norm": 1.2142643928527832,
      "learning_rate": 0.00016723286128697884,
      "loss": 3.419,
      "step": 175760
    },
    {
      "epoch": 0.16384608097619713,
      "grad_norm": 0.9731807112693787,
      "learning_rate": 0.00016723099696488365,
      "loss": 3.6712,
      "step": 175770
    },
    {
      "epoch": 0.16385540259427622,
      "grad_norm": 1.0274248123168945,
      "learning_rate": 0.00016722913264278843,
      "loss": 3.6581,
      "step": 175780
    },
    {
      "epoch": 0.16386472421235532,
      "grad_norm": 0.944143533706665,
      "learning_rate": 0.00016722726832069324,
      "loss": 3.4033,
      "step": 175790
    },
    {
      "epoch": 0.1638740458304344,
      "grad_norm": 1.0374034643173218,
      "learning_rate": 0.00016722540399859804,
      "loss": 3.6318,
      "step": 175800
    },
    {
      "epoch": 0.1638833674485135,
      "grad_norm": 1.007778286933899,
      "learning_rate": 0.00016722353967650285,
      "loss": 3.1856,
      "step": 175810
    },
    {
      "epoch": 0.1638926890665926,
      "grad_norm": 0.9879606366157532,
      "learning_rate": 0.00016722167535440766,
      "loss": 3.7428,
      "step": 175820
    },
    {
      "epoch": 0.1639020106846717,
      "grad_norm": 0.9374687075614929,
      "learning_rate": 0.00016721981103231243,
      "loss": 3.5027,
      "step": 175830
    },
    {
      "epoch": 0.16391133230275076,
      "grad_norm": 0.926770031452179,
      "learning_rate": 0.00016721794671021724,
      "loss": 3.4776,
      "step": 175840
    },
    {
      "epoch": 0.16392065392082986,
      "grad_norm": 0.9204336404800415,
      "learning_rate": 0.00016721608238812202,
      "loss": 3.6208,
      "step": 175850
    },
    {
      "epoch": 0.16392997553890895,
      "grad_norm": 1.111167550086975,
      "learning_rate": 0.00016721421806602683,
      "loss": 3.4283,
      "step": 175860
    },
    {
      "epoch": 0.16393929715698805,
      "grad_norm": 1.2343134880065918,
      "learning_rate": 0.00016721235374393166,
      "loss": 3.4622,
      "step": 175870
    },
    {
      "epoch": 0.16394861877506714,
      "grad_norm": 1.0070257186889648,
      "learning_rate": 0.00016721048942183644,
      "loss": 3.6284,
      "step": 175880
    },
    {
      "epoch": 0.16395794039314623,
      "grad_norm": 1.378912091255188,
      "learning_rate": 0.00016720862509974125,
      "loss": 3.6433,
      "step": 175890
    },
    {
      "epoch": 0.16396726201122533,
      "grad_norm": 0.937765896320343,
      "learning_rate": 0.00016720676077764602,
      "loss": 3.5507,
      "step": 175900
    },
    {
      "epoch": 0.16397658362930442,
      "grad_norm": 1.0140464305877686,
      "learning_rate": 0.00016720489645555083,
      "loss": 3.5798,
      "step": 175910
    },
    {
      "epoch": 0.16398590524738352,
      "grad_norm": 1.061002254486084,
      "learning_rate": 0.00016720303213345564,
      "loss": 3.4188,
      "step": 175920
    },
    {
      "epoch": 0.1639952268654626,
      "grad_norm": 0.7613069415092468,
      "learning_rate": 0.00016720116781136044,
      "loss": 3.4471,
      "step": 175930
    },
    {
      "epoch": 0.1640045484835417,
      "grad_norm": 1.197857141494751,
      "learning_rate": 0.00016719930348926525,
      "loss": 3.6687,
      "step": 175940
    },
    {
      "epoch": 0.16401387010162077,
      "grad_norm": 1.1304126977920532,
      "learning_rate": 0.00016719743916717003,
      "loss": 3.3478,
      "step": 175950
    },
    {
      "epoch": 0.16402319171969987,
      "grad_norm": 0.9675527215003967,
      "learning_rate": 0.00016719557484507484,
      "loss": 3.4361,
      "step": 175960
    },
    {
      "epoch": 0.16403251333777896,
      "grad_norm": 1.052942156791687,
      "learning_rate": 0.00016719371052297964,
      "loss": 3.4828,
      "step": 175970
    },
    {
      "epoch": 0.16404183495585806,
      "grad_norm": 0.999830424785614,
      "learning_rate": 0.00016719184620088445,
      "loss": 3.5136,
      "step": 175980
    },
    {
      "epoch": 0.16405115657393715,
      "grad_norm": 1.0746318101882935,
      "learning_rate": 0.00016718998187878925,
      "loss": 3.6953,
      "step": 175990
    },
    {
      "epoch": 0.16406047819201625,
      "grad_norm": 1.0669710636138916,
      "learning_rate": 0.00016718811755669403,
      "loss": 3.5736,
      "step": 176000
    },
    {
      "epoch": 0.16406979981009534,
      "grad_norm": 1.1566972732543945,
      "learning_rate": 0.00016718625323459884,
      "loss": 3.4491,
      "step": 176010
    },
    {
      "epoch": 0.16407912142817443,
      "grad_norm": 0.9297581315040588,
      "learning_rate": 0.00016718438891250365,
      "loss": 3.3979,
      "step": 176020
    },
    {
      "epoch": 0.16408844304625353,
      "grad_norm": 1.0668072700500488,
      "learning_rate": 0.00016718252459040845,
      "loss": 3.4894,
      "step": 176030
    },
    {
      "epoch": 0.16409776466433262,
      "grad_norm": 1.041746735572815,
      "learning_rate": 0.00016718066026831326,
      "loss": 3.5975,
      "step": 176040
    },
    {
      "epoch": 0.1641070862824117,
      "grad_norm": 0.998141884803772,
      "learning_rate": 0.00016717879594621804,
      "loss": 3.6028,
      "step": 176050
    },
    {
      "epoch": 0.16411640790049078,
      "grad_norm": 0.9421073794364929,
      "learning_rate": 0.00016717693162412284,
      "loss": 3.6578,
      "step": 176060
    },
    {
      "epoch": 0.16412572951856988,
      "grad_norm": 1.0663379430770874,
      "learning_rate": 0.00016717506730202765,
      "loss": 3.6145,
      "step": 176070
    },
    {
      "epoch": 0.16413505113664897,
      "grad_norm": 1.079344391822815,
      "learning_rate": 0.00016717320297993243,
      "loss": 3.5249,
      "step": 176080
    },
    {
      "epoch": 0.16414437275472807,
      "grad_norm": 1.1887874603271484,
      "learning_rate": 0.00016717133865783726,
      "loss": 3.6574,
      "step": 176090
    },
    {
      "epoch": 0.16415369437280716,
      "grad_norm": 1.3123761415481567,
      "learning_rate": 0.00016716947433574204,
      "loss": 3.5171,
      "step": 176100
    },
    {
      "epoch": 0.16416301599088626,
      "grad_norm": 1.0941473245620728,
      "learning_rate": 0.00016716761001364685,
      "loss": 3.4428,
      "step": 176110
    },
    {
      "epoch": 0.16417233760896535,
      "grad_norm": 1.1563363075256348,
      "learning_rate": 0.00016716574569155165,
      "loss": 3.4568,
      "step": 176120
    },
    {
      "epoch": 0.16418165922704445,
      "grad_norm": 1.1450591087341309,
      "learning_rate": 0.00016716388136945643,
      "loss": 3.352,
      "step": 176130
    },
    {
      "epoch": 0.16419098084512354,
      "grad_norm": 1.0963953733444214,
      "learning_rate": 0.00016716201704736124,
      "loss": 3.7332,
      "step": 176140
    },
    {
      "epoch": 0.1642003024632026,
      "grad_norm": 0.9672756791114807,
      "learning_rate": 0.00016716015272526605,
      "loss": 3.6635,
      "step": 176150
    },
    {
      "epoch": 0.1642096240812817,
      "grad_norm": 1.0668931007385254,
      "learning_rate": 0.00016715828840317085,
      "loss": 3.4346,
      "step": 176160
    },
    {
      "epoch": 0.1642189456993608,
      "grad_norm": 1.2250295877456665,
      "learning_rate": 0.00016715642408107566,
      "loss": 3.376,
      "step": 176170
    },
    {
      "epoch": 0.1642282673174399,
      "grad_norm": 0.9429448843002319,
      "learning_rate": 0.00016715455975898044,
      "loss": 3.6075,
      "step": 176180
    },
    {
      "epoch": 0.16423758893551899,
      "grad_norm": 1.0587822198867798,
      "learning_rate": 0.00016715269543688524,
      "loss": 3.5752,
      "step": 176190
    },
    {
      "epoch": 0.16424691055359808,
      "grad_norm": 0.9688680768013,
      "learning_rate": 0.00016715083111479005,
      "loss": 3.6499,
      "step": 176200
    },
    {
      "epoch": 0.16425623217167717,
      "grad_norm": 1.0592014789581299,
      "learning_rate": 0.00016714896679269486,
      "loss": 3.4137,
      "step": 176210
    },
    {
      "epoch": 0.16426555378975627,
      "grad_norm": 0.9686341881752014,
      "learning_rate": 0.00016714710247059966,
      "loss": 3.6021,
      "step": 176220
    },
    {
      "epoch": 0.16427487540783536,
      "grad_norm": 1.0140541791915894,
      "learning_rate": 0.00016714523814850444,
      "loss": 3.4184,
      "step": 176230
    },
    {
      "epoch": 0.16428419702591446,
      "grad_norm": 1.057027816772461,
      "learning_rate": 0.00016714337382640925,
      "loss": 3.6871,
      "step": 176240
    },
    {
      "epoch": 0.16429351864399355,
      "grad_norm": 1.2448809146881104,
      "learning_rate": 0.00016714150950431406,
      "loss": 3.5785,
      "step": 176250
    },
    {
      "epoch": 0.16430284026207262,
      "grad_norm": 1.0174821615219116,
      "learning_rate": 0.00016713964518221886,
      "loss": 3.3933,
      "step": 176260
    },
    {
      "epoch": 0.1643121618801517,
      "grad_norm": 1.0529307126998901,
      "learning_rate": 0.00016713778086012367,
      "loss": 3.4703,
      "step": 176270
    },
    {
      "epoch": 0.1643214834982308,
      "grad_norm": 1.0381836891174316,
      "learning_rate": 0.00016713591653802845,
      "loss": 3.4246,
      "step": 176280
    },
    {
      "epoch": 0.1643308051163099,
      "grad_norm": 0.9950206279754639,
      "learning_rate": 0.00016713405221593325,
      "loss": 3.3414,
      "step": 176290
    },
    {
      "epoch": 0.164340126734389,
      "grad_norm": 1.0274921655654907,
      "learning_rate": 0.00016713218789383803,
      "loss": 3.753,
      "step": 176300
    },
    {
      "epoch": 0.1643494483524681,
      "grad_norm": 0.9755023717880249,
      "learning_rate": 0.00016713032357174287,
      "loss": 3.7924,
      "step": 176310
    },
    {
      "epoch": 0.16435876997054719,
      "grad_norm": 1.072566032409668,
      "learning_rate": 0.00016712845924964767,
      "loss": 3.4825,
      "step": 176320
    },
    {
      "epoch": 0.16436809158862628,
      "grad_norm": 1.27999746799469,
      "learning_rate": 0.00016712659492755245,
      "loss": 3.4945,
      "step": 176330
    },
    {
      "epoch": 0.16437741320670538,
      "grad_norm": 1.0414459705352783,
      "learning_rate": 0.00016712473060545726,
      "loss": 3.6424,
      "step": 176340
    },
    {
      "epoch": 0.16438673482478447,
      "grad_norm": 1.0983777046203613,
      "learning_rate": 0.00016712286628336204,
      "loss": 3.6554,
      "step": 176350
    },
    {
      "epoch": 0.16439605644286354,
      "grad_norm": 1.203104019165039,
      "learning_rate": 0.00016712100196126684,
      "loss": 3.6354,
      "step": 176360
    },
    {
      "epoch": 0.16440537806094263,
      "grad_norm": 0.9047368168830872,
      "learning_rate": 0.00016711913763917165,
      "loss": 3.55,
      "step": 176370
    },
    {
      "epoch": 0.16441469967902173,
      "grad_norm": 1.2208269834518433,
      "learning_rate": 0.00016711727331707646,
      "loss": 3.4891,
      "step": 176380
    },
    {
      "epoch": 0.16442402129710082,
      "grad_norm": 1.1172620058059692,
      "learning_rate": 0.00016711540899498126,
      "loss": 3.5574,
      "step": 176390
    },
    {
      "epoch": 0.16443334291517991,
      "grad_norm": 0.9568072557449341,
      "learning_rate": 0.00016711354467288604,
      "loss": 3.5603,
      "step": 176400
    },
    {
      "epoch": 0.164442664533259,
      "grad_norm": 1.0818456411361694,
      "learning_rate": 0.00016711168035079085,
      "loss": 3.3756,
      "step": 176410
    },
    {
      "epoch": 0.1644519861513381,
      "grad_norm": 1.0414304733276367,
      "learning_rate": 0.00016710981602869565,
      "loss": 3.6937,
      "step": 176420
    },
    {
      "epoch": 0.1644613077694172,
      "grad_norm": 0.9891173839569092,
      "learning_rate": 0.00016710795170660046,
      "loss": 3.4781,
      "step": 176430
    },
    {
      "epoch": 0.1644706293874963,
      "grad_norm": 1.0875334739685059,
      "learning_rate": 0.00016710608738450527,
      "loss": 3.6192,
      "step": 176440
    },
    {
      "epoch": 0.1644799510055754,
      "grad_norm": 1.0358924865722656,
      "learning_rate": 0.00016710422306241005,
      "loss": 3.538,
      "step": 176450
    },
    {
      "epoch": 0.16448927262365448,
      "grad_norm": 1.2244949340820312,
      "learning_rate": 0.00016710235874031485,
      "loss": 3.6783,
      "step": 176460
    },
    {
      "epoch": 0.16449859424173355,
      "grad_norm": 0.8096359372138977,
      "learning_rate": 0.00016710049441821963,
      "loss": 3.7357,
      "step": 176470
    },
    {
      "epoch": 0.16450791585981264,
      "grad_norm": 1.0689644813537598,
      "learning_rate": 0.00016709863009612447,
      "loss": 3.4692,
      "step": 176480
    },
    {
      "epoch": 0.16451723747789174,
      "grad_norm": 1.07223641872406,
      "learning_rate": 0.00016709676577402927,
      "loss": 3.4308,
      "step": 176490
    },
    {
      "epoch": 0.16452655909597083,
      "grad_norm": 1.0933712720870972,
      "learning_rate": 0.00016709490145193405,
      "loss": 3.4164,
      "step": 176500
    },
    {
      "epoch": 0.16453588071404993,
      "grad_norm": 1.055428147315979,
      "learning_rate": 0.00016709303712983886,
      "loss": 3.6238,
      "step": 176510
    },
    {
      "epoch": 0.16454520233212902,
      "grad_norm": 1.191725492477417,
      "learning_rate": 0.00016709117280774364,
      "loss": 3.6807,
      "step": 176520
    },
    {
      "epoch": 0.16455452395020811,
      "grad_norm": 1.160973072052002,
      "learning_rate": 0.00016708930848564844,
      "loss": 3.4951,
      "step": 176530
    },
    {
      "epoch": 0.1645638455682872,
      "grad_norm": 0.9593070149421692,
      "learning_rate": 0.00016708744416355328,
      "loss": 3.4872,
      "step": 176540
    },
    {
      "epoch": 0.1645731671863663,
      "grad_norm": 1.1112546920776367,
      "learning_rate": 0.00016708557984145806,
      "loss": 3.6167,
      "step": 176550
    },
    {
      "epoch": 0.1645824888044454,
      "grad_norm": 0.8379981517791748,
      "learning_rate": 0.00016708371551936286,
      "loss": 3.6254,
      "step": 176560
    },
    {
      "epoch": 0.16459181042252446,
      "grad_norm": 1.0892159938812256,
      "learning_rate": 0.00016708185119726764,
      "loss": 3.4676,
      "step": 176570
    },
    {
      "epoch": 0.16460113204060356,
      "grad_norm": 1.0777606964111328,
      "learning_rate": 0.00016707998687517245,
      "loss": 3.567,
      "step": 176580
    },
    {
      "epoch": 0.16461045365868265,
      "grad_norm": 0.7264725565910339,
      "learning_rate": 0.00016707812255307725,
      "loss": 3.6336,
      "step": 176590
    },
    {
      "epoch": 0.16461977527676175,
      "grad_norm": 0.9728584885597229,
      "learning_rate": 0.00016707625823098206,
      "loss": 3.5291,
      "step": 176600
    },
    {
      "epoch": 0.16462909689484084,
      "grad_norm": 1.262062668800354,
      "learning_rate": 0.00016707439390888687,
      "loss": 3.5124,
      "step": 176610
    },
    {
      "epoch": 0.16463841851291994,
      "grad_norm": 1.2045942544937134,
      "learning_rate": 0.00016707252958679165,
      "loss": 3.5805,
      "step": 176620
    },
    {
      "epoch": 0.16464774013099903,
      "grad_norm": 1.0589112043380737,
      "learning_rate": 0.00016707066526469645,
      "loss": 3.6207,
      "step": 176630
    },
    {
      "epoch": 0.16465706174907813,
      "grad_norm": 0.9885470271110535,
      "learning_rate": 0.00016706880094260126,
      "loss": 3.5591,
      "step": 176640
    },
    {
      "epoch": 0.16466638336715722,
      "grad_norm": 1.0227361917495728,
      "learning_rate": 0.00016706693662050606,
      "loss": 3.6707,
      "step": 176650
    },
    {
      "epoch": 0.16467570498523632,
      "grad_norm": 0.9303203821182251,
      "learning_rate": 0.00016706507229841087,
      "loss": 3.645,
      "step": 176660
    },
    {
      "epoch": 0.1646850266033154,
      "grad_norm": 1.0298357009887695,
      "learning_rate": 0.00016706320797631565,
      "loss": 3.567,
      "step": 176670
    },
    {
      "epoch": 0.16469434822139448,
      "grad_norm": 0.9673129916191101,
      "learning_rate": 0.00016706134365422046,
      "loss": 3.2729,
      "step": 176680
    },
    {
      "epoch": 0.16470366983947357,
      "grad_norm": 1.1483319997787476,
      "learning_rate": 0.00016705947933212526,
      "loss": 3.4069,
      "step": 176690
    },
    {
      "epoch": 0.16471299145755267,
      "grad_norm": 1.2798141241073608,
      "learning_rate": 0.00016705761501003007,
      "loss": 3.3978,
      "step": 176700
    },
    {
      "epoch": 0.16472231307563176,
      "grad_norm": 1.3909058570861816,
      "learning_rate": 0.00016705575068793487,
      "loss": 3.4734,
      "step": 176710
    },
    {
      "epoch": 0.16473163469371085,
      "grad_norm": 1.064453125,
      "learning_rate": 0.00016705388636583965,
      "loss": 3.4832,
      "step": 176720
    },
    {
      "epoch": 0.16474095631178995,
      "grad_norm": 0.981279730796814,
      "learning_rate": 0.00016705202204374446,
      "loss": 3.5608,
      "step": 176730
    },
    {
      "epoch": 0.16475027792986904,
      "grad_norm": 1.2401913404464722,
      "learning_rate": 0.00016705015772164927,
      "loss": 3.5928,
      "step": 176740
    },
    {
      "epoch": 0.16475959954794814,
      "grad_norm": 1.0383918285369873,
      "learning_rate": 0.00016704829339955405,
      "loss": 3.3363,
      "step": 176750
    },
    {
      "epoch": 0.16476892116602723,
      "grad_norm": 0.9698703289031982,
      "learning_rate": 0.00016704642907745888,
      "loss": 3.5123,
      "step": 176760
    },
    {
      "epoch": 0.16477824278410633,
      "grad_norm": 0.8786337375640869,
      "learning_rate": 0.00016704456475536366,
      "loss": 3.5618,
      "step": 176770
    },
    {
      "epoch": 0.1647875644021854,
      "grad_norm": 1.103439450263977,
      "learning_rate": 0.00016704270043326846,
      "loss": 3.5643,
      "step": 176780
    },
    {
      "epoch": 0.1647968860202645,
      "grad_norm": 1.1230179071426392,
      "learning_rate": 0.00016704083611117327,
      "loss": 3.4895,
      "step": 176790
    },
    {
      "epoch": 0.16480620763834358,
      "grad_norm": 0.9411743879318237,
      "learning_rate": 0.00016703897178907805,
      "loss": 3.4889,
      "step": 176800
    },
    {
      "epoch": 0.16481552925642268,
      "grad_norm": 0.9651448726654053,
      "learning_rate": 0.00016703710746698286,
      "loss": 3.6102,
      "step": 176810
    },
    {
      "epoch": 0.16482485087450177,
      "grad_norm": 0.9650866985321045,
      "learning_rate": 0.00016703524314488766,
      "loss": 3.4269,
      "step": 176820
    },
    {
      "epoch": 0.16483417249258087,
      "grad_norm": 1.0496591329574585,
      "learning_rate": 0.00016703337882279247,
      "loss": 3.3888,
      "step": 176830
    },
    {
      "epoch": 0.16484349411065996,
      "grad_norm": 1.11793851852417,
      "learning_rate": 0.00016703151450069728,
      "loss": 3.2511,
      "step": 176840
    },
    {
      "epoch": 0.16485281572873905,
      "grad_norm": 1.6403411626815796,
      "learning_rate": 0.00016702965017860205,
      "loss": 3.4149,
      "step": 176850
    },
    {
      "epoch": 0.16486213734681815,
      "grad_norm": 2.192204236984253,
      "learning_rate": 0.00016702778585650686,
      "loss": 3.422,
      "step": 176860
    },
    {
      "epoch": 0.16487145896489724,
      "grad_norm": 1.1585832834243774,
      "learning_rate": 0.00016702592153441167,
      "loss": 3.5391,
      "step": 176870
    },
    {
      "epoch": 0.16488078058297634,
      "grad_norm": 1.0556836128234863,
      "learning_rate": 0.00016702405721231647,
      "loss": 3.3725,
      "step": 176880
    },
    {
      "epoch": 0.1648901022010554,
      "grad_norm": 1.1515028476715088,
      "learning_rate": 0.00016702219289022128,
      "loss": 3.5801,
      "step": 176890
    },
    {
      "epoch": 0.1648994238191345,
      "grad_norm": 1.061112642288208,
      "learning_rate": 0.00016702032856812606,
      "loss": 3.6299,
      "step": 176900
    },
    {
      "epoch": 0.1649087454372136,
      "grad_norm": 1.113472580909729,
      "learning_rate": 0.00016701846424603087,
      "loss": 3.4708,
      "step": 176910
    },
    {
      "epoch": 0.1649180670552927,
      "grad_norm": 1.1856321096420288,
      "learning_rate": 0.00016701659992393564,
      "loss": 3.5865,
      "step": 176920
    },
    {
      "epoch": 0.16492738867337178,
      "grad_norm": 1.0170261859893799,
      "learning_rate": 0.00016701473560184048,
      "loss": 3.6144,
      "step": 176930
    },
    {
      "epoch": 0.16493671029145088,
      "grad_norm": 1.0735061168670654,
      "learning_rate": 0.00016701287127974528,
      "loss": 3.4542,
      "step": 176940
    },
    {
      "epoch": 0.16494603190952997,
      "grad_norm": 1.0306382179260254,
      "learning_rate": 0.00016701100695765006,
      "loss": 3.5803,
      "step": 176950
    },
    {
      "epoch": 0.16495535352760907,
      "grad_norm": 1.3795576095581055,
      "learning_rate": 0.00016700914263555487,
      "loss": 3.5858,
      "step": 176960
    },
    {
      "epoch": 0.16496467514568816,
      "grad_norm": 1.2315164804458618,
      "learning_rate": 0.00016700727831345965,
      "loss": 3.386,
      "step": 176970
    },
    {
      "epoch": 0.16497399676376726,
      "grad_norm": 0.9790834784507751,
      "learning_rate": 0.00016700541399136446,
      "loss": 3.6167,
      "step": 176980
    },
    {
      "epoch": 0.16498331838184632,
      "grad_norm": 2.44557523727417,
      "learning_rate": 0.0001670035496692693,
      "loss": 3.5538,
      "step": 176990
    },
    {
      "epoch": 0.16499263999992542,
      "grad_norm": 1.5393012762069702,
      "learning_rate": 0.00016700168534717407,
      "loss": 3.3869,
      "step": 177000
    },
    {
      "epoch": 0.1650019616180045,
      "grad_norm": 0.8876034021377563,
      "learning_rate": 0.00016699982102507887,
      "loss": 3.6756,
      "step": 177010
    },
    {
      "epoch": 0.1650112832360836,
      "grad_norm": 1.0296971797943115,
      "learning_rate": 0.00016699795670298365,
      "loss": 3.7512,
      "step": 177020
    },
    {
      "epoch": 0.1650206048541627,
      "grad_norm": 1.1200398206710815,
      "learning_rate": 0.00016699609238088846,
      "loss": 3.7112,
      "step": 177030
    },
    {
      "epoch": 0.1650299264722418,
      "grad_norm": 0.9486483335494995,
      "learning_rate": 0.00016699422805879327,
      "loss": 3.5004,
      "step": 177040
    },
    {
      "epoch": 0.1650392480903209,
      "grad_norm": 1.1442148685455322,
      "learning_rate": 0.00016699236373669807,
      "loss": 3.55,
      "step": 177050
    },
    {
      "epoch": 0.16504856970839998,
      "grad_norm": 1.081581711769104,
      "learning_rate": 0.00016699049941460288,
      "loss": 3.4474,
      "step": 177060
    },
    {
      "epoch": 0.16505789132647908,
      "grad_norm": 1.069688320159912,
      "learning_rate": 0.00016698863509250766,
      "loss": 3.5726,
      "step": 177070
    },
    {
      "epoch": 0.16506721294455817,
      "grad_norm": 0.9090253710746765,
      "learning_rate": 0.00016698677077041246,
      "loss": 3.6125,
      "step": 177080
    },
    {
      "epoch": 0.16507653456263727,
      "grad_norm": 1.2120018005371094,
      "learning_rate": 0.00016698490644831727,
      "loss": 3.6176,
      "step": 177090
    },
    {
      "epoch": 0.16508585618071633,
      "grad_norm": 1.0517622232437134,
      "learning_rate": 0.00016698304212622208,
      "loss": 3.443,
      "step": 177100
    },
    {
      "epoch": 0.16509517779879543,
      "grad_norm": 1.0133548974990845,
      "learning_rate": 0.00016698117780412688,
      "loss": 3.3854,
      "step": 177110
    },
    {
      "epoch": 0.16510449941687452,
      "grad_norm": 1.2309191226959229,
      "learning_rate": 0.00016697931348203166,
      "loss": 3.4336,
      "step": 177120
    },
    {
      "epoch": 0.16511382103495362,
      "grad_norm": 0.8883259892463684,
      "learning_rate": 0.00016697744915993647,
      "loss": 3.5291,
      "step": 177130
    },
    {
      "epoch": 0.1651231426530327,
      "grad_norm": 1.043623924255371,
      "learning_rate": 0.00016697558483784127,
      "loss": 3.5677,
      "step": 177140
    },
    {
      "epoch": 0.1651324642711118,
      "grad_norm": 0.8405873775482178,
      "learning_rate": 0.00016697372051574608,
      "loss": 3.6099,
      "step": 177150
    },
    {
      "epoch": 0.1651417858891909,
      "grad_norm": 1.0635223388671875,
      "learning_rate": 0.0001669718561936509,
      "loss": 3.353,
      "step": 177160
    },
    {
      "epoch": 0.16515110750727,
      "grad_norm": 1.0322716236114502,
      "learning_rate": 0.00016696999187155567,
      "loss": 3.5512,
      "step": 177170
    },
    {
      "epoch": 0.1651604291253491,
      "grad_norm": 1.3531596660614014,
      "learning_rate": 0.00016696812754946047,
      "loss": 3.609,
      "step": 177180
    },
    {
      "epoch": 0.16516975074342818,
      "grad_norm": 0.9480712413787842,
      "learning_rate": 0.00016696626322736528,
      "loss": 3.7073,
      "step": 177190
    },
    {
      "epoch": 0.16517907236150725,
      "grad_norm": 1.1239100694656372,
      "learning_rate": 0.00016696439890527006,
      "loss": 3.4848,
      "step": 177200
    },
    {
      "epoch": 0.16518839397958635,
      "grad_norm": 0.9238165020942688,
      "learning_rate": 0.0001669625345831749,
      "loss": 3.4718,
      "step": 177210
    },
    {
      "epoch": 0.16519771559766544,
      "grad_norm": 0.9893054962158203,
      "learning_rate": 0.00016696067026107967,
      "loss": 3.4929,
      "step": 177220
    },
    {
      "epoch": 0.16520703721574453,
      "grad_norm": 1.0561351776123047,
      "learning_rate": 0.00016695880593898448,
      "loss": 3.5155,
      "step": 177230
    },
    {
      "epoch": 0.16521635883382363,
      "grad_norm": 1.0039275884628296,
      "learning_rate": 0.00016695694161688928,
      "loss": 3.5308,
      "step": 177240
    },
    {
      "epoch": 0.16522568045190272,
      "grad_norm": 1.6951111555099487,
      "learning_rate": 0.00016695507729479406,
      "loss": 3.4278,
      "step": 177250
    },
    {
      "epoch": 0.16523500206998182,
      "grad_norm": 1.2214958667755127,
      "learning_rate": 0.00016695321297269887,
      "loss": 3.5829,
      "step": 177260
    },
    {
      "epoch": 0.1652443236880609,
      "grad_norm": 1.028956413269043,
      "learning_rate": 0.00016695134865060368,
      "loss": 3.6357,
      "step": 177270
    },
    {
      "epoch": 0.16525364530614,
      "grad_norm": 1.1612915992736816,
      "learning_rate": 0.00016694948432850848,
      "loss": 3.5086,
      "step": 177280
    },
    {
      "epoch": 0.1652629669242191,
      "grad_norm": 0.834690272808075,
      "learning_rate": 0.0001669476200064133,
      "loss": 3.3611,
      "step": 177290
    },
    {
      "epoch": 0.1652722885422982,
      "grad_norm": 1.2785699367523193,
      "learning_rate": 0.00016694575568431807,
      "loss": 3.5244,
      "step": 177300
    },
    {
      "epoch": 0.16528161016037726,
      "grad_norm": 0.9152519106864929,
      "learning_rate": 0.00016694389136222287,
      "loss": 3.5005,
      "step": 177310
    },
    {
      "epoch": 0.16529093177845636,
      "grad_norm": 1.1414921283721924,
      "learning_rate": 0.00016694202704012768,
      "loss": 3.6012,
      "step": 177320
    },
    {
      "epoch": 0.16530025339653545,
      "grad_norm": 0.962202250957489,
      "learning_rate": 0.00016694016271803249,
      "loss": 3.5484,
      "step": 177330
    },
    {
      "epoch": 0.16530957501461455,
      "grad_norm": 1.0953091382980347,
      "learning_rate": 0.0001669382983959373,
      "loss": 3.5598,
      "step": 177340
    },
    {
      "epoch": 0.16531889663269364,
      "grad_norm": 1.03326416015625,
      "learning_rate": 0.00016693643407384207,
      "loss": 3.4532,
      "step": 177350
    },
    {
      "epoch": 0.16532821825077273,
      "grad_norm": 1.598293662071228,
      "learning_rate": 0.00016693456975174688,
      "loss": 3.5704,
      "step": 177360
    },
    {
      "epoch": 0.16533753986885183,
      "grad_norm": 1.0473116636276245,
      "learning_rate": 0.00016693270542965166,
      "loss": 3.5703,
      "step": 177370
    },
    {
      "epoch": 0.16534686148693092,
      "grad_norm": 1.179372787475586,
      "learning_rate": 0.0001669308411075565,
      "loss": 3.4905,
      "step": 177380
    },
    {
      "epoch": 0.16535618310501002,
      "grad_norm": 1.016631007194519,
      "learning_rate": 0.0001669289767854613,
      "loss": 3.669,
      "step": 177390
    },
    {
      "epoch": 0.1653655047230891,
      "grad_norm": 0.8731494545936584,
      "learning_rate": 0.00016692711246336608,
      "loss": 3.6657,
      "step": 177400
    },
    {
      "epoch": 0.16537482634116818,
      "grad_norm": 1.0502030849456787,
      "learning_rate": 0.00016692524814127088,
      "loss": 3.5999,
      "step": 177410
    },
    {
      "epoch": 0.16538414795924727,
      "grad_norm": 1.3016979694366455,
      "learning_rate": 0.00016692338381917566,
      "loss": 3.6536,
      "step": 177420
    },
    {
      "epoch": 0.16539346957732637,
      "grad_norm": 1.1424449682235718,
      "learning_rate": 0.00016692151949708047,
      "loss": 3.661,
      "step": 177430
    },
    {
      "epoch": 0.16540279119540546,
      "grad_norm": 0.8861116766929626,
      "learning_rate": 0.0001669196551749853,
      "loss": 3.3675,
      "step": 177440
    },
    {
      "epoch": 0.16541211281348456,
      "grad_norm": 0.9580240249633789,
      "learning_rate": 0.00016691779085289008,
      "loss": 3.5267,
      "step": 177450
    },
    {
      "epoch": 0.16542143443156365,
      "grad_norm": 1.1423864364624023,
      "learning_rate": 0.0001669159265307949,
      "loss": 3.556,
      "step": 177460
    },
    {
      "epoch": 0.16543075604964275,
      "grad_norm": 0.8782299757003784,
      "learning_rate": 0.00016691406220869967,
      "loss": 3.656,
      "step": 177470
    },
    {
      "epoch": 0.16544007766772184,
      "grad_norm": 1.120049238204956,
      "learning_rate": 0.00016691219788660447,
      "loss": 3.3384,
      "step": 177480
    },
    {
      "epoch": 0.16544939928580094,
      "grad_norm": 1.1088042259216309,
      "learning_rate": 0.00016691033356450928,
      "loss": 3.5449,
      "step": 177490
    },
    {
      "epoch": 0.16545872090388003,
      "grad_norm": 1.1478739976882935,
      "learning_rate": 0.00016690846924241409,
      "loss": 3.5531,
      "step": 177500
    },
    {
      "epoch": 0.1654680425219591,
      "grad_norm": 1.089497447013855,
      "learning_rate": 0.0001669066049203189,
      "loss": 3.5472,
      "step": 177510
    },
    {
      "epoch": 0.1654773641400382,
      "grad_norm": 1.2194817066192627,
      "learning_rate": 0.00016690474059822367,
      "loss": 3.4662,
      "step": 177520
    },
    {
      "epoch": 0.16548668575811729,
      "grad_norm": 1.10348379611969,
      "learning_rate": 0.00016690287627612848,
      "loss": 3.6182,
      "step": 177530
    },
    {
      "epoch": 0.16549600737619638,
      "grad_norm": 0.9675232768058777,
      "learning_rate": 0.00016690101195403328,
      "loss": 3.6906,
      "step": 177540
    },
    {
      "epoch": 0.16550532899427547,
      "grad_norm": 0.831299364566803,
      "learning_rate": 0.0001668991476319381,
      "loss": 3.5555,
      "step": 177550
    },
    {
      "epoch": 0.16551465061235457,
      "grad_norm": 0.9195166826248169,
      "learning_rate": 0.0001668972833098429,
      "loss": 3.3462,
      "step": 177560
    },
    {
      "epoch": 0.16552397223043366,
      "grad_norm": 0.9762443900108337,
      "learning_rate": 0.00016689541898774768,
      "loss": 3.4877,
      "step": 177570
    },
    {
      "epoch": 0.16553329384851276,
      "grad_norm": 1.040235996246338,
      "learning_rate": 0.00016689355466565248,
      "loss": 3.4358,
      "step": 177580
    },
    {
      "epoch": 0.16554261546659185,
      "grad_norm": 0.8524937629699707,
      "learning_rate": 0.0001668916903435573,
      "loss": 3.5714,
      "step": 177590
    },
    {
      "epoch": 0.16555193708467095,
      "grad_norm": 0.9013103246688843,
      "learning_rate": 0.0001668898260214621,
      "loss": 3.4587,
      "step": 177600
    },
    {
      "epoch": 0.16556125870275004,
      "grad_norm": 1.0819839239120483,
      "learning_rate": 0.0001668879616993669,
      "loss": 3.5414,
      "step": 177610
    },
    {
      "epoch": 0.1655705803208291,
      "grad_norm": 0.9117224812507629,
      "learning_rate": 0.00016688609737727168,
      "loss": 3.6359,
      "step": 177620
    },
    {
      "epoch": 0.1655799019389082,
      "grad_norm": 1.1564717292785645,
      "learning_rate": 0.00016688423305517649,
      "loss": 3.4459,
      "step": 177630
    },
    {
      "epoch": 0.1655892235569873,
      "grad_norm": 1.09989595413208,
      "learning_rate": 0.0001668823687330813,
      "loss": 3.4452,
      "step": 177640
    },
    {
      "epoch": 0.1655985451750664,
      "grad_norm": 1.2324378490447998,
      "learning_rate": 0.00016688050441098607,
      "loss": 3.535,
      "step": 177650
    },
    {
      "epoch": 0.1656078667931455,
      "grad_norm": 1.1279910802841187,
      "learning_rate": 0.0001668786400888909,
      "loss": 3.5981,
      "step": 177660
    },
    {
      "epoch": 0.16561718841122458,
      "grad_norm": 1.1213936805725098,
      "learning_rate": 0.00016687677576679568,
      "loss": 3.521,
      "step": 177670
    },
    {
      "epoch": 0.16562651002930368,
      "grad_norm": 1.1455796957015991,
      "learning_rate": 0.0001668749114447005,
      "loss": 3.3563,
      "step": 177680
    },
    {
      "epoch": 0.16563583164738277,
      "grad_norm": 0.9958137273788452,
      "learning_rate": 0.0001668730471226053,
      "loss": 3.5815,
      "step": 177690
    },
    {
      "epoch": 0.16564515326546186,
      "grad_norm": 0.9898131489753723,
      "learning_rate": 0.00016687118280051008,
      "loss": 3.5617,
      "step": 177700
    },
    {
      "epoch": 0.16565447488354096,
      "grad_norm": 0.9588993787765503,
      "learning_rate": 0.00016686931847841488,
      "loss": 3.6384,
      "step": 177710
    },
    {
      "epoch": 0.16566379650162003,
      "grad_norm": 1.0014657974243164,
      "learning_rate": 0.0001668674541563197,
      "loss": 3.3567,
      "step": 177720
    },
    {
      "epoch": 0.16567311811969912,
      "grad_norm": 1.0744001865386963,
      "learning_rate": 0.0001668655898342245,
      "loss": 3.6182,
      "step": 177730
    },
    {
      "epoch": 0.16568243973777821,
      "grad_norm": 0.9756109714508057,
      "learning_rate": 0.0001668637255121293,
      "loss": 3.4813,
      "step": 177740
    },
    {
      "epoch": 0.1656917613558573,
      "grad_norm": 1.2592332363128662,
      "learning_rate": 0.00016686186119003408,
      "loss": 3.4616,
      "step": 177750
    },
    {
      "epoch": 0.1657010829739364,
      "grad_norm": 1.0122489929199219,
      "learning_rate": 0.0001668599968679389,
      "loss": 3.5465,
      "step": 177760
    },
    {
      "epoch": 0.1657104045920155,
      "grad_norm": 1.08082115650177,
      "learning_rate": 0.0001668581325458437,
      "loss": 3.3184,
      "step": 177770
    },
    {
      "epoch": 0.1657197262100946,
      "grad_norm": 0.9493487477302551,
      "learning_rate": 0.0001668562682237485,
      "loss": 3.4156,
      "step": 177780
    },
    {
      "epoch": 0.1657290478281737,
      "grad_norm": 1.3206536769866943,
      "learning_rate": 0.00016685440390165328,
      "loss": 3.4102,
      "step": 177790
    },
    {
      "epoch": 0.16573836944625278,
      "grad_norm": 1.171721339225769,
      "learning_rate": 0.00016685253957955808,
      "loss": 3.5404,
      "step": 177800
    },
    {
      "epoch": 0.16574769106433188,
      "grad_norm": 1.0042073726654053,
      "learning_rate": 0.0001668506752574629,
      "loss": 3.7041,
      "step": 177810
    },
    {
      "epoch": 0.16575701268241097,
      "grad_norm": 1.0156724452972412,
      "learning_rate": 0.00016684881093536767,
      "loss": 3.4986,
      "step": 177820
    },
    {
      "epoch": 0.16576633430049004,
      "grad_norm": 1.1110048294067383,
      "learning_rate": 0.0001668469466132725,
      "loss": 3.7412,
      "step": 177830
    },
    {
      "epoch": 0.16577565591856913,
      "grad_norm": 1.0596638917922974,
      "learning_rate": 0.00016684508229117728,
      "loss": 3.5179,
      "step": 177840
    },
    {
      "epoch": 0.16578497753664823,
      "grad_norm": 1.1047056913375854,
      "learning_rate": 0.0001668432179690821,
      "loss": 3.3674,
      "step": 177850
    },
    {
      "epoch": 0.16579429915472732,
      "grad_norm": 0.8350966572761536,
      "learning_rate": 0.0001668413536469869,
      "loss": 3.5728,
      "step": 177860
    },
    {
      "epoch": 0.16580362077280641,
      "grad_norm": 1.104987382888794,
      "learning_rate": 0.00016683948932489167,
      "loss": 3.5624,
      "step": 177870
    },
    {
      "epoch": 0.1658129423908855,
      "grad_norm": 0.976584255695343,
      "learning_rate": 0.00016683762500279648,
      "loss": 3.4317,
      "step": 177880
    },
    {
      "epoch": 0.1658222640089646,
      "grad_norm": 1.0379024744033813,
      "learning_rate": 0.0001668357606807013,
      "loss": 3.4702,
      "step": 177890
    },
    {
      "epoch": 0.1658315856270437,
      "grad_norm": 0.9311032295227051,
      "learning_rate": 0.0001668338963586061,
      "loss": 3.4503,
      "step": 177900
    },
    {
      "epoch": 0.1658409072451228,
      "grad_norm": 1.2495341300964355,
      "learning_rate": 0.0001668320320365109,
      "loss": 3.4654,
      "step": 177910
    },
    {
      "epoch": 0.1658502288632019,
      "grad_norm": 1.1950229406356812,
      "learning_rate": 0.00016683016771441568,
      "loss": 3.5591,
      "step": 177920
    },
    {
      "epoch": 0.16585955048128095,
      "grad_norm": 0.8685575723648071,
      "learning_rate": 0.00016682830339232049,
      "loss": 3.5057,
      "step": 177930
    },
    {
      "epoch": 0.16586887209936005,
      "grad_norm": 1.0322073698043823,
      "learning_rate": 0.0001668264390702253,
      "loss": 3.5101,
      "step": 177940
    },
    {
      "epoch": 0.16587819371743914,
      "grad_norm": 1.0471590757369995,
      "learning_rate": 0.0001668245747481301,
      "loss": 3.6853,
      "step": 177950
    },
    {
      "epoch": 0.16588751533551824,
      "grad_norm": 1.4071714878082275,
      "learning_rate": 0.0001668227104260349,
      "loss": 3.4307,
      "step": 177960
    },
    {
      "epoch": 0.16589683695359733,
      "grad_norm": 1.040490984916687,
      "learning_rate": 0.00016682084610393968,
      "loss": 3.5888,
      "step": 177970
    },
    {
      "epoch": 0.16590615857167643,
      "grad_norm": 1.0719144344329834,
      "learning_rate": 0.0001668189817818445,
      "loss": 3.7005,
      "step": 177980
    },
    {
      "epoch": 0.16591548018975552,
      "grad_norm": 1.0487257242202759,
      "learning_rate": 0.0001668171174597493,
      "loss": 3.6181,
      "step": 177990
    },
    {
      "epoch": 0.16592480180783462,
      "grad_norm": 1.0992493629455566,
      "learning_rate": 0.0001668152531376541,
      "loss": 3.7377,
      "step": 178000
    },
    {
      "epoch": 0.1659341234259137,
      "grad_norm": 0.9939232468605042,
      "learning_rate": 0.0001668133888155589,
      "loss": 3.6728,
      "step": 178010
    },
    {
      "epoch": 0.1659434450439928,
      "grad_norm": 1.0221387147903442,
      "learning_rate": 0.0001668115244934637,
      "loss": 3.6755,
      "step": 178020
    },
    {
      "epoch": 0.1659527666620719,
      "grad_norm": 0.8931737542152405,
      "learning_rate": 0.0001668096601713685,
      "loss": 3.5152,
      "step": 178030
    },
    {
      "epoch": 0.16596208828015097,
      "grad_norm": 1.1187372207641602,
      "learning_rate": 0.00016680779584927327,
      "loss": 3.4566,
      "step": 178040
    },
    {
      "epoch": 0.16597140989823006,
      "grad_norm": 1.1133027076721191,
      "learning_rate": 0.0001668059315271781,
      "loss": 3.7725,
      "step": 178050
    },
    {
      "epoch": 0.16598073151630915,
      "grad_norm": 0.9655250310897827,
      "learning_rate": 0.0001668040672050829,
      "loss": 3.5469,
      "step": 178060
    },
    {
      "epoch": 0.16599005313438825,
      "grad_norm": 1.1285699605941772,
      "learning_rate": 0.0001668022028829877,
      "loss": 3.5408,
      "step": 178070
    },
    {
      "epoch": 0.16599937475246734,
      "grad_norm": 1.0074163675308228,
      "learning_rate": 0.0001668003385608925,
      "loss": 3.5401,
      "step": 178080
    },
    {
      "epoch": 0.16600869637054644,
      "grad_norm": 1.175376057624817,
      "learning_rate": 0.00016679847423879728,
      "loss": 3.4669,
      "step": 178090
    },
    {
      "epoch": 0.16601801798862553,
      "grad_norm": 1.1131943464279175,
      "learning_rate": 0.00016679660991670208,
      "loss": 3.5577,
      "step": 178100
    },
    {
      "epoch": 0.16602733960670463,
      "grad_norm": 1.2394670248031616,
      "learning_rate": 0.00016679474559460692,
      "loss": 3.6036,
      "step": 178110
    },
    {
      "epoch": 0.16603666122478372,
      "grad_norm": 1.0537701845169067,
      "learning_rate": 0.0001667928812725117,
      "loss": 3.5767,
      "step": 178120
    },
    {
      "epoch": 0.16604598284286282,
      "grad_norm": 1.0245957374572754,
      "learning_rate": 0.0001667910169504165,
      "loss": 3.4144,
      "step": 178130
    },
    {
      "epoch": 0.16605530446094188,
      "grad_norm": 1.1557459831237793,
      "learning_rate": 0.00016678915262832128,
      "loss": 3.4885,
      "step": 178140
    },
    {
      "epoch": 0.16606462607902098,
      "grad_norm": 1.0258052349090576,
      "learning_rate": 0.0001667872883062261,
      "loss": 3.6168,
      "step": 178150
    },
    {
      "epoch": 0.16607394769710007,
      "grad_norm": 0.8781927824020386,
      "learning_rate": 0.0001667854239841309,
      "loss": 3.5955,
      "step": 178160
    },
    {
      "epoch": 0.16608326931517917,
      "grad_norm": 0.8532488346099854,
      "learning_rate": 0.0001667835596620357,
      "loss": 3.5832,
      "step": 178170
    },
    {
      "epoch": 0.16609259093325826,
      "grad_norm": 0.9314068555831909,
      "learning_rate": 0.0001667816953399405,
      "loss": 3.5199,
      "step": 178180
    },
    {
      "epoch": 0.16610191255133736,
      "grad_norm": 1.061406135559082,
      "learning_rate": 0.0001667798310178453,
      "loss": 3.6304,
      "step": 178190
    },
    {
      "epoch": 0.16611123416941645,
      "grad_norm": 1.1553843021392822,
      "learning_rate": 0.0001667779666957501,
      "loss": 3.4584,
      "step": 178200
    },
    {
      "epoch": 0.16612055578749554,
      "grad_norm": 1.0283653736114502,
      "learning_rate": 0.0001667761023736549,
      "loss": 3.8027,
      "step": 178210
    },
    {
      "epoch": 0.16612987740557464,
      "grad_norm": 1.3668311834335327,
      "learning_rate": 0.0001667742380515597,
      "loss": 3.6198,
      "step": 178220
    },
    {
      "epoch": 0.16613919902365373,
      "grad_norm": 0.977700412273407,
      "learning_rate": 0.0001667723737294645,
      "loss": 3.3868,
      "step": 178230
    },
    {
      "epoch": 0.16614852064173283,
      "grad_norm": 1.2898929119110107,
      "learning_rate": 0.0001667705094073693,
      "loss": 3.4399,
      "step": 178240
    },
    {
      "epoch": 0.1661578422598119,
      "grad_norm": 0.9894577264785767,
      "learning_rate": 0.0001667686450852741,
      "loss": 3.6116,
      "step": 178250
    },
    {
      "epoch": 0.166167163877891,
      "grad_norm": 1.0569734573364258,
      "learning_rate": 0.0001667667807631789,
      "loss": 3.5443,
      "step": 178260
    },
    {
      "epoch": 0.16617648549597008,
      "grad_norm": 0.9420716762542725,
      "learning_rate": 0.00016676491644108368,
      "loss": 3.5075,
      "step": 178270
    },
    {
      "epoch": 0.16618580711404918,
      "grad_norm": 1.193596601486206,
      "learning_rate": 0.00016676305211898852,
      "loss": 3.4326,
      "step": 178280
    },
    {
      "epoch": 0.16619512873212827,
      "grad_norm": 0.7629056572914124,
      "learning_rate": 0.0001667611877968933,
      "loss": 3.554,
      "step": 178290
    },
    {
      "epoch": 0.16620445035020737,
      "grad_norm": 1.0361157655715942,
      "learning_rate": 0.0001667593234747981,
      "loss": 3.6716,
      "step": 178300
    },
    {
      "epoch": 0.16621377196828646,
      "grad_norm": 1.162651538848877,
      "learning_rate": 0.0001667574591527029,
      "loss": 3.5112,
      "step": 178310
    },
    {
      "epoch": 0.16622309358636556,
      "grad_norm": 1.0513452291488647,
      "learning_rate": 0.0001667555948306077,
      "loss": 3.5062,
      "step": 178320
    },
    {
      "epoch": 0.16623241520444465,
      "grad_norm": 1.1472389698028564,
      "learning_rate": 0.0001667537305085125,
      "loss": 3.3841,
      "step": 178330
    },
    {
      "epoch": 0.16624173682252374,
      "grad_norm": 1.087058186531067,
      "learning_rate": 0.0001667518661864173,
      "loss": 3.5126,
      "step": 178340
    },
    {
      "epoch": 0.1662510584406028,
      "grad_norm": 0.805040717124939,
      "learning_rate": 0.0001667500018643221,
      "loss": 3.4498,
      "step": 178350
    },
    {
      "epoch": 0.1662603800586819,
      "grad_norm": 1.1732125282287598,
      "learning_rate": 0.0001667481375422269,
      "loss": 3.7292,
      "step": 178360
    },
    {
      "epoch": 0.166269701676761,
      "grad_norm": 1.142383098602295,
      "learning_rate": 0.0001667462732201317,
      "loss": 3.6939,
      "step": 178370
    },
    {
      "epoch": 0.1662790232948401,
      "grad_norm": 1.2193849086761475,
      "learning_rate": 0.0001667444088980365,
      "loss": 3.5374,
      "step": 178380
    },
    {
      "epoch": 0.1662883449129192,
      "grad_norm": 0.9552348256111145,
      "learning_rate": 0.0001667425445759413,
      "loss": 3.6067,
      "step": 178390
    },
    {
      "epoch": 0.16629766653099828,
      "grad_norm": 1.2009563446044922,
      "learning_rate": 0.0001667406802538461,
      "loss": 3.5738,
      "step": 178400
    },
    {
      "epoch": 0.16630698814907738,
      "grad_norm": 1.031578540802002,
      "learning_rate": 0.00016673881593175092,
      "loss": 3.5984,
      "step": 178410
    },
    {
      "epoch": 0.16631630976715647,
      "grad_norm": 0.860066831111908,
      "learning_rate": 0.0001667369516096557,
      "loss": 3.3891,
      "step": 178420
    },
    {
      "epoch": 0.16632563138523557,
      "grad_norm": 1.1170240640640259,
      "learning_rate": 0.0001667350872875605,
      "loss": 3.5798,
      "step": 178430
    },
    {
      "epoch": 0.16633495300331466,
      "grad_norm": 1.1509767770767212,
      "learning_rate": 0.0001667332229654653,
      "loss": 3.6117,
      "step": 178440
    },
    {
      "epoch": 0.16634427462139376,
      "grad_norm": 0.9392849206924438,
      "learning_rate": 0.00016673135864337012,
      "loss": 3.5928,
      "step": 178450
    },
    {
      "epoch": 0.16635359623947282,
      "grad_norm": 1.0496634244918823,
      "learning_rate": 0.00016672949432127492,
      "loss": 3.5298,
      "step": 178460
    },
    {
      "epoch": 0.16636291785755192,
      "grad_norm": 0.9680724143981934,
      "learning_rate": 0.0001667276299991797,
      "loss": 3.4626,
      "step": 178470
    },
    {
      "epoch": 0.166372239475631,
      "grad_norm": 1.1778568029403687,
      "learning_rate": 0.0001667257656770845,
      "loss": 3.5693,
      "step": 178480
    },
    {
      "epoch": 0.1663815610937101,
      "grad_norm": 1.2171003818511963,
      "learning_rate": 0.00016672390135498929,
      "loss": 3.5294,
      "step": 178490
    },
    {
      "epoch": 0.1663908827117892,
      "grad_norm": 1.1111141443252563,
      "learning_rate": 0.00016672203703289412,
      "loss": 3.5616,
      "step": 178500
    },
    {
      "epoch": 0.1664002043298683,
      "grad_norm": 0.99329674243927,
      "learning_rate": 0.00016672017271079893,
      "loss": 3.4251,
      "step": 178510
    },
    {
      "epoch": 0.1664095259479474,
      "grad_norm": 1.1522644758224487,
      "learning_rate": 0.0001667183083887037,
      "loss": 3.3665,
      "step": 178520
    },
    {
      "epoch": 0.16641884756602648,
      "grad_norm": 1.1668236255645752,
      "learning_rate": 0.0001667164440666085,
      "loss": 3.6568,
      "step": 178530
    },
    {
      "epoch": 0.16642816918410558,
      "grad_norm": 1.0510979890823364,
      "learning_rate": 0.0001667145797445133,
      "loss": 3.5273,
      "step": 178540
    },
    {
      "epoch": 0.16643749080218467,
      "grad_norm": 1.2905536890029907,
      "learning_rate": 0.0001667127154224181,
      "loss": 3.2863,
      "step": 178550
    },
    {
      "epoch": 0.16644681242026374,
      "grad_norm": 0.8821520805358887,
      "learning_rate": 0.00016671085110032293,
      "loss": 3.4592,
      "step": 178560
    },
    {
      "epoch": 0.16645613403834283,
      "grad_norm": 1.3061563968658447,
      "learning_rate": 0.0001667089867782277,
      "loss": 3.5065,
      "step": 178570
    },
    {
      "epoch": 0.16646545565642193,
      "grad_norm": 1.12269926071167,
      "learning_rate": 0.00016670712245613252,
      "loss": 3.7954,
      "step": 178580
    },
    {
      "epoch": 0.16647477727450102,
      "grad_norm": 1.041063904762268,
      "learning_rate": 0.0001667052581340373,
      "loss": 3.3947,
      "step": 178590
    },
    {
      "epoch": 0.16648409889258012,
      "grad_norm": 0.8590685725212097,
      "learning_rate": 0.0001667033938119421,
      "loss": 3.5515,
      "step": 178600
    },
    {
      "epoch": 0.1664934205106592,
      "grad_norm": 1.0040956735610962,
      "learning_rate": 0.0001667015294898469,
      "loss": 3.4527,
      "step": 178610
    },
    {
      "epoch": 0.1665027421287383,
      "grad_norm": 1.1408096551895142,
      "learning_rate": 0.00016669966516775171,
      "loss": 3.5296,
      "step": 178620
    },
    {
      "epoch": 0.1665120637468174,
      "grad_norm": 0.9869802594184875,
      "learning_rate": 0.00016669780084565652,
      "loss": 3.3584,
      "step": 178630
    },
    {
      "epoch": 0.1665213853648965,
      "grad_norm": 1.3981473445892334,
      "learning_rate": 0.0001666959365235613,
      "loss": 3.41,
      "step": 178640
    },
    {
      "epoch": 0.1665307069829756,
      "grad_norm": 1.1771459579467773,
      "learning_rate": 0.0001666940722014661,
      "loss": 3.4553,
      "step": 178650
    },
    {
      "epoch": 0.16654002860105468,
      "grad_norm": 0.9838051795959473,
      "learning_rate": 0.0001666922078793709,
      "loss": 3.449,
      "step": 178660
    },
    {
      "epoch": 0.16654935021913375,
      "grad_norm": 1.0490186214447021,
      "learning_rate": 0.00016669034355727572,
      "loss": 3.4162,
      "step": 178670
    },
    {
      "epoch": 0.16655867183721285,
      "grad_norm": 1.1394373178482056,
      "learning_rate": 0.00016668847923518052,
      "loss": 3.5809,
      "step": 178680
    },
    {
      "epoch": 0.16656799345529194,
      "grad_norm": 1.2266037464141846,
      "learning_rate": 0.0001666866149130853,
      "loss": 3.5334,
      "step": 178690
    },
    {
      "epoch": 0.16657731507337104,
      "grad_norm": 1.190362811088562,
      "learning_rate": 0.0001666847505909901,
      "loss": 3.4418,
      "step": 178700
    },
    {
      "epoch": 0.16658663669145013,
      "grad_norm": 0.9745271801948547,
      "learning_rate": 0.00016668288626889492,
      "loss": 3.5829,
      "step": 178710
    },
    {
      "epoch": 0.16659595830952922,
      "grad_norm": 1.0509145259857178,
      "learning_rate": 0.0001666810219467997,
      "loss": 3.5996,
      "step": 178720
    },
    {
      "epoch": 0.16660527992760832,
      "grad_norm": 0.9136136174201965,
      "learning_rate": 0.00016667915762470453,
      "loss": 3.481,
      "step": 178730
    },
    {
      "epoch": 0.1666146015456874,
      "grad_norm": 0.8757052421569824,
      "learning_rate": 0.0001666772933026093,
      "loss": 3.5787,
      "step": 178740
    },
    {
      "epoch": 0.1666239231637665,
      "grad_norm": 0.8527224063873291,
      "learning_rate": 0.00016667542898051411,
      "loss": 3.5465,
      "step": 178750
    },
    {
      "epoch": 0.1666332447818456,
      "grad_norm": 0.967746376991272,
      "learning_rate": 0.00016667356465841892,
      "loss": 3.6303,
      "step": 178760
    },
    {
      "epoch": 0.16664256639992467,
      "grad_norm": 0.91629958152771,
      "learning_rate": 0.0001666717003363237,
      "loss": 3.7441,
      "step": 178770
    },
    {
      "epoch": 0.16665188801800376,
      "grad_norm": 1.2356555461883545,
      "learning_rate": 0.0001666698360142285,
      "loss": 3.5513,
      "step": 178780
    },
    {
      "epoch": 0.16666120963608286,
      "grad_norm": 0.9610746502876282,
      "learning_rate": 0.0001666679716921333,
      "loss": 3.5213,
      "step": 178790
    },
    {
      "epoch": 0.16667053125416195,
      "grad_norm": 1.1002808809280396,
      "learning_rate": 0.00016666610737003812,
      "loss": 3.4828,
      "step": 178800
    },
    {
      "epoch": 0.16667985287224105,
      "grad_norm": 0.8481955528259277,
      "learning_rate": 0.00016666424304794293,
      "loss": 3.6652,
      "step": 178810
    },
    {
      "epoch": 0.16668917449032014,
      "grad_norm": 1.111578345298767,
      "learning_rate": 0.0001666623787258477,
      "loss": 3.5645,
      "step": 178820
    },
    {
      "epoch": 0.16669849610839924,
      "grad_norm": 1.2253086566925049,
      "learning_rate": 0.0001666605144037525,
      "loss": 3.3751,
      "step": 178830
    },
    {
      "epoch": 0.16670781772647833,
      "grad_norm": 1.0324978828430176,
      "learning_rate": 0.00016665865008165732,
      "loss": 3.5746,
      "step": 178840
    },
    {
      "epoch": 0.16671713934455742,
      "grad_norm": 0.9610598683357239,
      "learning_rate": 0.00016665678575956212,
      "loss": 3.5841,
      "step": 178850
    },
    {
      "epoch": 0.16672646096263652,
      "grad_norm": 1.0807660818099976,
      "learning_rate": 0.00016665492143746693,
      "loss": 3.426,
      "step": 178860
    },
    {
      "epoch": 0.16673578258071559,
      "grad_norm": 1.173378825187683,
      "learning_rate": 0.0001666530571153717,
      "loss": 3.645,
      "step": 178870
    },
    {
      "epoch": 0.16674510419879468,
      "grad_norm": 1.3153913021087646,
      "learning_rate": 0.00016665119279327652,
      "loss": 3.5202,
      "step": 178880
    },
    {
      "epoch": 0.16675442581687377,
      "grad_norm": 0.9511484503746033,
      "learning_rate": 0.00016664932847118132,
      "loss": 3.4706,
      "step": 178890
    },
    {
      "epoch": 0.16676374743495287,
      "grad_norm": 1.0260106325149536,
      "learning_rate": 0.00016664746414908613,
      "loss": 3.4387,
      "step": 178900
    },
    {
      "epoch": 0.16677306905303196,
      "grad_norm": 1.301238775253296,
      "learning_rate": 0.00016664559982699093,
      "loss": 3.3922,
      "step": 178910
    },
    {
      "epoch": 0.16678239067111106,
      "grad_norm": 1.1635210514068604,
      "learning_rate": 0.00016664373550489571,
      "loss": 3.4911,
      "step": 178920
    },
    {
      "epoch": 0.16679171228919015,
      "grad_norm": 1.1985400915145874,
      "learning_rate": 0.00016664187118280052,
      "loss": 3.508,
      "step": 178930
    },
    {
      "epoch": 0.16680103390726925,
      "grad_norm": 0.8955910205841064,
      "learning_rate": 0.0001666400068607053,
      "loss": 3.4825,
      "step": 178940
    },
    {
      "epoch": 0.16681035552534834,
      "grad_norm": 1.0653364658355713,
      "learning_rate": 0.00016663814253861013,
      "loss": 3.6076,
      "step": 178950
    },
    {
      "epoch": 0.16681967714342744,
      "grad_norm": 1.1177425384521484,
      "learning_rate": 0.00016663627821651494,
      "loss": 3.6509,
      "step": 178960
    },
    {
      "epoch": 0.16682899876150653,
      "grad_norm": 1.1860722303390503,
      "learning_rate": 0.00016663441389441972,
      "loss": 3.528,
      "step": 178970
    },
    {
      "epoch": 0.1668383203795856,
      "grad_norm": 1.0915340185165405,
      "learning_rate": 0.00016663254957232452,
      "loss": 3.6438,
      "step": 178980
    },
    {
      "epoch": 0.1668476419976647,
      "grad_norm": 1.1045163869857788,
      "learning_rate": 0.0001666306852502293,
      "loss": 3.5146,
      "step": 178990
    },
    {
      "epoch": 0.1668569636157438,
      "grad_norm": 1.122087001800537,
      "learning_rate": 0.0001666288209281341,
      "loss": 3.3956,
      "step": 179000
    },
    {
      "epoch": 0.16686628523382288,
      "grad_norm": 1.0950796604156494,
      "learning_rate": 0.00016662695660603894,
      "loss": 3.4324,
      "step": 179010
    },
    {
      "epoch": 0.16687560685190198,
      "grad_norm": 1.0465474128723145,
      "learning_rate": 0.00016662509228394372,
      "loss": 3.5,
      "step": 179020
    },
    {
      "epoch": 0.16688492846998107,
      "grad_norm": 1.0454895496368408,
      "learning_rate": 0.00016662322796184853,
      "loss": 3.4866,
      "step": 179030
    },
    {
      "epoch": 0.16689425008806016,
      "grad_norm": 1.0802881717681885,
      "learning_rate": 0.0001666213636397533,
      "loss": 3.436,
      "step": 179040
    },
    {
      "epoch": 0.16690357170613926,
      "grad_norm": 0.9910936951637268,
      "learning_rate": 0.00016661949931765811,
      "loss": 3.3799,
      "step": 179050
    },
    {
      "epoch": 0.16691289332421835,
      "grad_norm": 0.827364981174469,
      "learning_rate": 0.00016661763499556292,
      "loss": 3.6177,
      "step": 179060
    },
    {
      "epoch": 0.16692221494229745,
      "grad_norm": 1.2230627536773682,
      "learning_rate": 0.00016661577067346773,
      "loss": 3.5188,
      "step": 179070
    },
    {
      "epoch": 0.16693153656037651,
      "grad_norm": 1.1812880039215088,
      "learning_rate": 0.00016661390635137253,
      "loss": 3.5778,
      "step": 179080
    },
    {
      "epoch": 0.1669408581784556,
      "grad_norm": 1.1216961145401,
      "learning_rate": 0.0001666120420292773,
      "loss": 3.507,
      "step": 179090
    },
    {
      "epoch": 0.1669501797965347,
      "grad_norm": 1.1968097686767578,
      "learning_rate": 0.00016661017770718212,
      "loss": 3.5246,
      "step": 179100
    },
    {
      "epoch": 0.1669595014146138,
      "grad_norm": 1.0829716920852661,
      "learning_rate": 0.00016660831338508693,
      "loss": 3.6354,
      "step": 179110
    },
    {
      "epoch": 0.1669688230326929,
      "grad_norm": 0.9571331739425659,
      "learning_rate": 0.00016660644906299173,
      "loss": 3.416,
      "step": 179120
    },
    {
      "epoch": 0.166978144650772,
      "grad_norm": 1.1201566457748413,
      "learning_rate": 0.00016660458474089654,
      "loss": 3.6084,
      "step": 179130
    },
    {
      "epoch": 0.16698746626885108,
      "grad_norm": 0.9743264317512512,
      "learning_rate": 0.00016660272041880132,
      "loss": 3.4388,
      "step": 179140
    },
    {
      "epoch": 0.16699678788693018,
      "grad_norm": 1.1146516799926758,
      "learning_rate": 0.00016660085609670612,
      "loss": 3.6036,
      "step": 179150
    },
    {
      "epoch": 0.16700610950500927,
      "grad_norm": 1.048599123954773,
      "learning_rate": 0.0001665989917746109,
      "loss": 3.5458,
      "step": 179160
    },
    {
      "epoch": 0.16701543112308836,
      "grad_norm": 0.8323628902435303,
      "learning_rate": 0.0001665971274525157,
      "loss": 3.4442,
      "step": 179170
    },
    {
      "epoch": 0.16702475274116746,
      "grad_norm": 1.0561702251434326,
      "learning_rate": 0.00016659526313042054,
      "loss": 3.5971,
      "step": 179180
    },
    {
      "epoch": 0.16703407435924653,
      "grad_norm": 0.978247880935669,
      "learning_rate": 0.00016659339880832532,
      "loss": 3.6339,
      "step": 179190
    },
    {
      "epoch": 0.16704339597732562,
      "grad_norm": 1.1079663038253784,
      "learning_rate": 0.00016659153448623013,
      "loss": 3.3968,
      "step": 179200
    },
    {
      "epoch": 0.16705271759540471,
      "grad_norm": 1.1467300653457642,
      "learning_rate": 0.0001665896701641349,
      "loss": 3.5748,
      "step": 179210
    },
    {
      "epoch": 0.1670620392134838,
      "grad_norm": 1.301754355430603,
      "learning_rate": 0.0001665878058420397,
      "loss": 3.4145,
      "step": 179220
    },
    {
      "epoch": 0.1670713608315629,
      "grad_norm": 0.8438061475753784,
      "learning_rate": 0.00016658594151994452,
      "loss": 3.5031,
      "step": 179230
    },
    {
      "epoch": 0.167080682449642,
      "grad_norm": 1.056978464126587,
      "learning_rate": 0.00016658407719784933,
      "loss": 3.7594,
      "step": 179240
    },
    {
      "epoch": 0.1670900040677211,
      "grad_norm": 1.053452491760254,
      "learning_rate": 0.00016658221287575413,
      "loss": 3.6655,
      "step": 179250
    },
    {
      "epoch": 0.1670993256858002,
      "grad_norm": 1.1615251302719116,
      "learning_rate": 0.0001665803485536589,
      "loss": 3.5724,
      "step": 179260
    },
    {
      "epoch": 0.16710864730387928,
      "grad_norm": 1.1670507192611694,
      "learning_rate": 0.00016657848423156372,
      "loss": 3.5487,
      "step": 179270
    },
    {
      "epoch": 0.16711796892195838,
      "grad_norm": 1.0369371175765991,
      "learning_rate": 0.00016657661990946852,
      "loss": 3.583,
      "step": 179280
    },
    {
      "epoch": 0.16712729054003744,
      "grad_norm": 0.8893411755561829,
      "learning_rate": 0.00016657475558737333,
      "loss": 3.7,
      "step": 179290
    },
    {
      "epoch": 0.16713661215811654,
      "grad_norm": 0.8633598685264587,
      "learning_rate": 0.00016657289126527814,
      "loss": 3.5945,
      "step": 179300
    },
    {
      "epoch": 0.16714593377619563,
      "grad_norm": 1.1058424711227417,
      "learning_rate": 0.00016657102694318292,
      "loss": 3.5212,
      "step": 179310
    },
    {
      "epoch": 0.16715525539427473,
      "grad_norm": 1.1352168321609497,
      "learning_rate": 0.00016656916262108772,
      "loss": 3.4983,
      "step": 179320
    },
    {
      "epoch": 0.16716457701235382,
      "grad_norm": 0.9116548895835876,
      "learning_rate": 0.00016656729829899253,
      "loss": 3.4366,
      "step": 179330
    },
    {
      "epoch": 0.16717389863043292,
      "grad_norm": 1.0629829168319702,
      "learning_rate": 0.00016656543397689733,
      "loss": 3.6631,
      "step": 179340
    },
    {
      "epoch": 0.167183220248512,
      "grad_norm": 1.0156861543655396,
      "learning_rate": 0.00016656356965480214,
      "loss": 3.6036,
      "step": 179350
    },
    {
      "epoch": 0.1671925418665911,
      "grad_norm": 1.2521045207977295,
      "learning_rate": 0.00016656170533270692,
      "loss": 3.5629,
      "step": 179360
    },
    {
      "epoch": 0.1672018634846702,
      "grad_norm": 1.1118820905685425,
      "learning_rate": 0.00016655984101061173,
      "loss": 3.6049,
      "step": 179370
    },
    {
      "epoch": 0.1672111851027493,
      "grad_norm": 0.8850547671318054,
      "learning_rate": 0.00016655797668851653,
      "loss": 3.4342,
      "step": 179380
    },
    {
      "epoch": 0.1672205067208284,
      "grad_norm": 1.3850669860839844,
      "learning_rate": 0.0001665561123664213,
      "loss": 3.3641,
      "step": 179390
    },
    {
      "epoch": 0.16722982833890745,
      "grad_norm": 1.1810981035232544,
      "learning_rate": 0.00016655424804432615,
      "loss": 3.5632,
      "step": 179400
    },
    {
      "epoch": 0.16723914995698655,
      "grad_norm": 0.9508130550384521,
      "learning_rate": 0.00016655238372223092,
      "loss": 3.735,
      "step": 179410
    },
    {
      "epoch": 0.16724847157506564,
      "grad_norm": 1.0285418033599854,
      "learning_rate": 0.00016655051940013573,
      "loss": 3.7483,
      "step": 179420
    },
    {
      "epoch": 0.16725779319314474,
      "grad_norm": 1.0134379863739014,
      "learning_rate": 0.00016654865507804054,
      "loss": 3.5066,
      "step": 179430
    },
    {
      "epoch": 0.16726711481122383,
      "grad_norm": 1.0543049573898315,
      "learning_rate": 0.00016654679075594532,
      "loss": 3.4754,
      "step": 179440
    },
    {
      "epoch": 0.16727643642930293,
      "grad_norm": 1.0988004207611084,
      "learning_rate": 0.00016654492643385012,
      "loss": 3.5429,
      "step": 179450
    },
    {
      "epoch": 0.16728575804738202,
      "grad_norm": 1.199110984802246,
      "learning_rate": 0.00016654306211175493,
      "loss": 3.6387,
      "step": 179460
    },
    {
      "epoch": 0.16729507966546112,
      "grad_norm": 1.1332573890686035,
      "learning_rate": 0.00016654119778965974,
      "loss": 3.5301,
      "step": 179470
    },
    {
      "epoch": 0.1673044012835402,
      "grad_norm": 1.109920620918274,
      "learning_rate": 0.00016653933346756454,
      "loss": 3.3632,
      "step": 179480
    },
    {
      "epoch": 0.1673137229016193,
      "grad_norm": 1.1004252433776855,
      "learning_rate": 0.00016653746914546932,
      "loss": 3.5111,
      "step": 179490
    },
    {
      "epoch": 0.16732304451969837,
      "grad_norm": 1.0189266204833984,
      "learning_rate": 0.00016653560482337413,
      "loss": 3.5508,
      "step": 179500
    },
    {
      "epoch": 0.16733236613777747,
      "grad_norm": 1.0077261924743652,
      "learning_rate": 0.00016653374050127893,
      "loss": 3.5045,
      "step": 179510
    },
    {
      "epoch": 0.16734168775585656,
      "grad_norm": 0.9685155153274536,
      "learning_rate": 0.00016653187617918374,
      "loss": 3.7014,
      "step": 179520
    },
    {
      "epoch": 0.16735100937393566,
      "grad_norm": 1.0613492727279663,
      "learning_rate": 0.00016653001185708855,
      "loss": 3.5502,
      "step": 179530
    },
    {
      "epoch": 0.16736033099201475,
      "grad_norm": 1.274254322052002,
      "learning_rate": 0.00016652814753499333,
      "loss": 3.5406,
      "step": 179540
    },
    {
      "epoch": 0.16736965261009384,
      "grad_norm": 1.0433045625686646,
      "learning_rate": 0.00016652628321289813,
      "loss": 3.4506,
      "step": 179550
    },
    {
      "epoch": 0.16737897422817294,
      "grad_norm": 1.1366299390792847,
      "learning_rate": 0.00016652441889080294,
      "loss": 3.4811,
      "step": 179560
    },
    {
      "epoch": 0.16738829584625203,
      "grad_norm": 0.886374831199646,
      "learning_rate": 0.00016652255456870774,
      "loss": 3.6285,
      "step": 179570
    },
    {
      "epoch": 0.16739761746433113,
      "grad_norm": 1.0375717878341675,
      "learning_rate": 0.00016652069024661255,
      "loss": 3.4741,
      "step": 179580
    },
    {
      "epoch": 0.16740693908241022,
      "grad_norm": 1.087979793548584,
      "learning_rate": 0.00016651882592451733,
      "loss": 3.6362,
      "step": 179590
    },
    {
      "epoch": 0.16741626070048932,
      "grad_norm": 0.975075364112854,
      "learning_rate": 0.00016651696160242214,
      "loss": 3.6575,
      "step": 179600
    },
    {
      "epoch": 0.16742558231856838,
      "grad_norm": 1.0310895442962646,
      "learning_rate": 0.00016651509728032692,
      "loss": 3.4532,
      "step": 179610
    },
    {
      "epoch": 0.16743490393664748,
      "grad_norm": 1.2122609615325928,
      "learning_rate": 0.00016651323295823175,
      "loss": 3.6911,
      "step": 179620
    },
    {
      "epoch": 0.16744422555472657,
      "grad_norm": 1.0448613166809082,
      "learning_rate": 0.00016651136863613655,
      "loss": 3.4885,
      "step": 179630
    },
    {
      "epoch": 0.16745354717280567,
      "grad_norm": 0.9878147840499878,
      "learning_rate": 0.00016650950431404133,
      "loss": 3.5869,
      "step": 179640
    },
    {
      "epoch": 0.16746286879088476,
      "grad_norm": 1.2319366931915283,
      "learning_rate": 0.00016650763999194614,
      "loss": 3.7326,
      "step": 179650
    },
    {
      "epoch": 0.16747219040896386,
      "grad_norm": 1.2250179052352905,
      "learning_rate": 0.00016650577566985092,
      "loss": 3.5273,
      "step": 179660
    },
    {
      "epoch": 0.16748151202704295,
      "grad_norm": 1.0526217222213745,
      "learning_rate": 0.00016650391134775573,
      "loss": 3.4089,
      "step": 179670
    },
    {
      "epoch": 0.16749083364512204,
      "grad_norm": 1.0476984977722168,
      "learning_rate": 0.00016650204702566053,
      "loss": 3.6657,
      "step": 179680
    },
    {
      "epoch": 0.16750015526320114,
      "grad_norm": 1.014748215675354,
      "learning_rate": 0.00016650018270356534,
      "loss": 3.682,
      "step": 179690
    },
    {
      "epoch": 0.16750947688128023,
      "grad_norm": 1.149743676185608,
      "learning_rate": 0.00016649831838147014,
      "loss": 3.5718,
      "step": 179700
    },
    {
      "epoch": 0.1675187984993593,
      "grad_norm": 0.7739169001579285,
      "learning_rate": 0.00016649645405937492,
      "loss": 3.5616,
      "step": 179710
    },
    {
      "epoch": 0.1675281201174384,
      "grad_norm": 1.0725491046905518,
      "learning_rate": 0.00016649458973727973,
      "loss": 3.387,
      "step": 179720
    },
    {
      "epoch": 0.1675374417355175,
      "grad_norm": 1.0591360330581665,
      "learning_rate": 0.00016649272541518454,
      "loss": 3.4593,
      "step": 179730
    },
    {
      "epoch": 0.16754676335359658,
      "grad_norm": 1.0821025371551514,
      "learning_rate": 0.00016649086109308934,
      "loss": 3.4634,
      "step": 179740
    },
    {
      "epoch": 0.16755608497167568,
      "grad_norm": 1.0175939798355103,
      "learning_rate": 0.00016648899677099415,
      "loss": 3.5937,
      "step": 179750
    },
    {
      "epoch": 0.16756540658975477,
      "grad_norm": 0.9502071738243103,
      "learning_rate": 0.00016648713244889893,
      "loss": 3.2602,
      "step": 179760
    },
    {
      "epoch": 0.16757472820783387,
      "grad_norm": 1.2021206617355347,
      "learning_rate": 0.00016648526812680373,
      "loss": 3.6061,
      "step": 179770
    },
    {
      "epoch": 0.16758404982591296,
      "grad_norm": 1.1935138702392578,
      "learning_rate": 0.00016648340380470854,
      "loss": 3.5306,
      "step": 179780
    },
    {
      "epoch": 0.16759337144399206,
      "grad_norm": 1.0704543590545654,
      "learning_rate": 0.00016648153948261335,
      "loss": 3.6178,
      "step": 179790
    },
    {
      "epoch": 0.16760269306207115,
      "grad_norm": 1.15630304813385,
      "learning_rate": 0.00016647967516051815,
      "loss": 3.2923,
      "step": 179800
    },
    {
      "epoch": 0.16761201468015025,
      "grad_norm": 0.9043869972229004,
      "learning_rate": 0.00016647781083842293,
      "loss": 3.5508,
      "step": 179810
    },
    {
      "epoch": 0.1676213362982293,
      "grad_norm": 1.0474575757980347,
      "learning_rate": 0.00016647594651632774,
      "loss": 3.794,
      "step": 179820
    },
    {
      "epoch": 0.1676306579163084,
      "grad_norm": 1.1562050580978394,
      "learning_rate": 0.00016647408219423255,
      "loss": 3.3961,
      "step": 179830
    },
    {
      "epoch": 0.1676399795343875,
      "grad_norm": 1.2271111011505127,
      "learning_rate": 0.00016647221787213732,
      "loss": 3.5126,
      "step": 179840
    },
    {
      "epoch": 0.1676493011524666,
      "grad_norm": 1.5028656721115112,
      "learning_rate": 0.00016647035355004216,
      "loss": 3.5364,
      "step": 179850
    },
    {
      "epoch": 0.1676586227705457,
      "grad_norm": 0.7449135780334473,
      "learning_rate": 0.00016646848922794694,
      "loss": 3.4381,
      "step": 179860
    },
    {
      "epoch": 0.16766794438862478,
      "grad_norm": 1.1199055910110474,
      "learning_rate": 0.00016646662490585174,
      "loss": 3.6369,
      "step": 179870
    },
    {
      "epoch": 0.16767726600670388,
      "grad_norm": 0.8901111483573914,
      "learning_rate": 0.00016646476058375655,
      "loss": 3.7225,
      "step": 179880
    },
    {
      "epoch": 0.16768658762478297,
      "grad_norm": 1.15988028049469,
      "learning_rate": 0.00016646289626166133,
      "loss": 3.5121,
      "step": 179890
    },
    {
      "epoch": 0.16769590924286207,
      "grad_norm": 0.9537486433982849,
      "learning_rate": 0.00016646103193956614,
      "loss": 3.5687,
      "step": 179900
    },
    {
      "epoch": 0.16770523086094116,
      "grad_norm": 1.005094289779663,
      "learning_rate": 0.00016645916761747094,
      "loss": 3.5295,
      "step": 179910
    },
    {
      "epoch": 0.16771455247902023,
      "grad_norm": 0.9306091666221619,
      "learning_rate": 0.00016645730329537575,
      "loss": 3.6549,
      "step": 179920
    },
    {
      "epoch": 0.16772387409709932,
      "grad_norm": 1.1125545501708984,
      "learning_rate": 0.00016645543897328055,
      "loss": 3.4054,
      "step": 179930
    },
    {
      "epoch": 0.16773319571517842,
      "grad_norm": 0.8657830953598022,
      "learning_rate": 0.00016645357465118533,
      "loss": 3.6887,
      "step": 179940
    },
    {
      "epoch": 0.1677425173332575,
      "grad_norm": 1.2621265649795532,
      "learning_rate": 0.00016645171032909014,
      "loss": 3.4612,
      "step": 179950
    },
    {
      "epoch": 0.1677518389513366,
      "grad_norm": 1.0428041219711304,
      "learning_rate": 0.00016644984600699495,
      "loss": 3.547,
      "step": 179960
    },
    {
      "epoch": 0.1677611605694157,
      "grad_norm": 1.5313225984573364,
      "learning_rate": 0.00016644798168489975,
      "loss": 3.542,
      "step": 179970
    },
    {
      "epoch": 0.1677704821874948,
      "grad_norm": 1.0478975772857666,
      "learning_rate": 0.00016644611736280456,
      "loss": 3.5829,
      "step": 179980
    },
    {
      "epoch": 0.1677798038055739,
      "grad_norm": 1.0339785814285278,
      "learning_rate": 0.00016644425304070934,
      "loss": 3.6865,
      "step": 179990
    },
    {
      "epoch": 0.16778912542365298,
      "grad_norm": 0.7735255360603333,
      "learning_rate": 0.00016644238871861414,
      "loss": 3.6883,
      "step": 180000
    },
    {
      "epoch": 0.16779844704173208,
      "grad_norm": 0.9927285313606262,
      "learning_rate": 0.00016644052439651895,
      "loss": 3.5509,
      "step": 180010
    },
    {
      "epoch": 0.16780776865981117,
      "grad_norm": 0.8330685496330261,
      "learning_rate": 0.00016643866007442376,
      "loss": 3.7097,
      "step": 180020
    },
    {
      "epoch": 0.16781709027789024,
      "grad_norm": 1.1202483177185059,
      "learning_rate": 0.00016643679575232856,
      "loss": 3.5572,
      "step": 180030
    },
    {
      "epoch": 0.16782641189596934,
      "grad_norm": 1.102191686630249,
      "learning_rate": 0.00016643493143023334,
      "loss": 3.6597,
      "step": 180040
    },
    {
      "epoch": 0.16783573351404843,
      "grad_norm": 1.0572313070297241,
      "learning_rate": 0.00016643306710813815,
      "loss": 3.3714,
      "step": 180050
    },
    {
      "epoch": 0.16784505513212752,
      "grad_norm": 0.9209694266319275,
      "learning_rate": 0.00016643120278604293,
      "loss": 3.5022,
      "step": 180060
    },
    {
      "epoch": 0.16785437675020662,
      "grad_norm": 0.9949057698249817,
      "learning_rate": 0.00016642933846394776,
      "loss": 3.5582,
      "step": 180070
    },
    {
      "epoch": 0.1678636983682857,
      "grad_norm": 1.0304553508758545,
      "learning_rate": 0.00016642747414185257,
      "loss": 3.5583,
      "step": 180080
    },
    {
      "epoch": 0.1678730199863648,
      "grad_norm": 1.0672693252563477,
      "learning_rate": 0.00016642560981975735,
      "loss": 3.5253,
      "step": 180090
    },
    {
      "epoch": 0.1678823416044439,
      "grad_norm": 1.0847831964492798,
      "learning_rate": 0.00016642374549766215,
      "loss": 3.6417,
      "step": 180100
    },
    {
      "epoch": 0.167891663222523,
      "grad_norm": 1.0438119173049927,
      "learning_rate": 0.00016642188117556693,
      "loss": 3.3689,
      "step": 180110
    },
    {
      "epoch": 0.1679009848406021,
      "grad_norm": 1.041677474975586,
      "learning_rate": 0.00016642001685347174,
      "loss": 3.6342,
      "step": 180120
    },
    {
      "epoch": 0.16791030645868116,
      "grad_norm": 1.0642749071121216,
      "learning_rate": 0.00016641815253137657,
      "loss": 3.46,
      "step": 180130
    },
    {
      "epoch": 0.16791962807676025,
      "grad_norm": 1.0084755420684814,
      "learning_rate": 0.00016641628820928135,
      "loss": 3.5145,
      "step": 180140
    },
    {
      "epoch": 0.16792894969483935,
      "grad_norm": 0.8486252427101135,
      "learning_rate": 0.00016641442388718616,
      "loss": 3.5882,
      "step": 180150
    },
    {
      "epoch": 0.16793827131291844,
      "grad_norm": 1.3032069206237793,
      "learning_rate": 0.00016641255956509094,
      "loss": 3.3851,
      "step": 180160
    },
    {
      "epoch": 0.16794759293099754,
      "grad_norm": 1.0340659618377686,
      "learning_rate": 0.00016641069524299574,
      "loss": 3.6083,
      "step": 180170
    },
    {
      "epoch": 0.16795691454907663,
      "grad_norm": 1.0465784072875977,
      "learning_rate": 0.00016640883092090055,
      "loss": 3.7605,
      "step": 180180
    },
    {
      "epoch": 0.16796623616715572,
      "grad_norm": 0.9812043905258179,
      "learning_rate": 0.00016640696659880536,
      "loss": 3.4234,
      "step": 180190
    },
    {
      "epoch": 0.16797555778523482,
      "grad_norm": 0.8725436329841614,
      "learning_rate": 0.00016640510227671016,
      "loss": 3.6749,
      "step": 180200
    },
    {
      "epoch": 0.1679848794033139,
      "grad_norm": 0.9901911616325378,
      "learning_rate": 0.00016640323795461494,
      "loss": 3.5496,
      "step": 180210
    },
    {
      "epoch": 0.167994201021393,
      "grad_norm": 1.465112566947937,
      "learning_rate": 0.00016640137363251975,
      "loss": 3.4086,
      "step": 180220
    },
    {
      "epoch": 0.1680035226394721,
      "grad_norm": 1.2137609720230103,
      "learning_rate": 0.00016639950931042455,
      "loss": 3.5863,
      "step": 180230
    },
    {
      "epoch": 0.16801284425755117,
      "grad_norm": 1.0431053638458252,
      "learning_rate": 0.00016639764498832936,
      "loss": 3.5899,
      "step": 180240
    },
    {
      "epoch": 0.16802216587563026,
      "grad_norm": 0.9662299752235413,
      "learning_rate": 0.00016639578066623417,
      "loss": 3.5586,
      "step": 180250
    },
    {
      "epoch": 0.16803148749370936,
      "grad_norm": 1.1815738677978516,
      "learning_rate": 0.00016639391634413895,
      "loss": 3.5182,
      "step": 180260
    },
    {
      "epoch": 0.16804080911178845,
      "grad_norm": 1.0227603912353516,
      "learning_rate": 0.00016639205202204375,
      "loss": 3.5438,
      "step": 180270
    },
    {
      "epoch": 0.16805013072986755,
      "grad_norm": 0.7879714965820312,
      "learning_rate": 0.00016639018769994856,
      "loss": 3.689,
      "step": 180280
    },
    {
      "epoch": 0.16805945234794664,
      "grad_norm": 1.090765357017517,
      "learning_rate": 0.00016638832337785334,
      "loss": 3.5304,
      "step": 180290
    },
    {
      "epoch": 0.16806877396602574,
      "grad_norm": 0.914738118648529,
      "learning_rate": 0.00016638645905575817,
      "loss": 3.6129,
      "step": 180300
    },
    {
      "epoch": 0.16807809558410483,
      "grad_norm": 1.0258647203445435,
      "learning_rate": 0.00016638459473366295,
      "loss": 3.5361,
      "step": 180310
    },
    {
      "epoch": 0.16808741720218393,
      "grad_norm": 1.1658806800842285,
      "learning_rate": 0.00016638273041156776,
      "loss": 3.395,
      "step": 180320
    },
    {
      "epoch": 0.16809673882026302,
      "grad_norm": 1.1408507823944092,
      "learning_rate": 0.00016638086608947256,
      "loss": 3.5156,
      "step": 180330
    },
    {
      "epoch": 0.1681060604383421,
      "grad_norm": 0.8508766889572144,
      "learning_rate": 0.00016637900176737734,
      "loss": 3.485,
      "step": 180340
    },
    {
      "epoch": 0.16811538205642118,
      "grad_norm": 0.8550509810447693,
      "learning_rate": 0.00016637713744528215,
      "loss": 3.4865,
      "step": 180350
    },
    {
      "epoch": 0.16812470367450028,
      "grad_norm": 1.0742753744125366,
      "learning_rate": 0.00016637527312318695,
      "loss": 3.6029,
      "step": 180360
    },
    {
      "epoch": 0.16813402529257937,
      "grad_norm": 1.1949617862701416,
      "learning_rate": 0.00016637340880109176,
      "loss": 3.5722,
      "step": 180370
    },
    {
      "epoch": 0.16814334691065846,
      "grad_norm": 1.0474309921264648,
      "learning_rate": 0.00016637154447899657,
      "loss": 3.6981,
      "step": 180380
    },
    {
      "epoch": 0.16815266852873756,
      "grad_norm": 1.0363680124282837,
      "learning_rate": 0.00016636968015690135,
      "loss": 3.3448,
      "step": 180390
    },
    {
      "epoch": 0.16816199014681665,
      "grad_norm": 0.9940577149391174,
      "learning_rate": 0.00016636781583480615,
      "loss": 3.5158,
      "step": 180400
    },
    {
      "epoch": 0.16817131176489575,
      "grad_norm": 1.3604172468185425,
      "learning_rate": 0.00016636595151271096,
      "loss": 3.5487,
      "step": 180410
    },
    {
      "epoch": 0.16818063338297484,
      "grad_norm": 1.052736759185791,
      "learning_rate": 0.00016636408719061577,
      "loss": 3.4782,
      "step": 180420
    },
    {
      "epoch": 0.16818995500105394,
      "grad_norm": 1.0611096620559692,
      "learning_rate": 0.00016636222286852057,
      "loss": 3.7661,
      "step": 180430
    },
    {
      "epoch": 0.168199276619133,
      "grad_norm": 1.1173245906829834,
      "learning_rate": 0.00016636035854642535,
      "loss": 3.6758,
      "step": 180440
    },
    {
      "epoch": 0.1682085982372121,
      "grad_norm": 1.123633623123169,
      "learning_rate": 0.00016635849422433016,
      "loss": 3.4817,
      "step": 180450
    },
    {
      "epoch": 0.1682179198552912,
      "grad_norm": 1.053910493850708,
      "learning_rate": 0.00016635662990223496,
      "loss": 3.4628,
      "step": 180460
    },
    {
      "epoch": 0.1682272414733703,
      "grad_norm": 1.0379263162612915,
      "learning_rate": 0.00016635476558013977,
      "loss": 3.5696,
      "step": 180470
    },
    {
      "epoch": 0.16823656309144938,
      "grad_norm": 1.2094285488128662,
      "learning_rate": 0.00016635290125804455,
      "loss": 3.5625,
      "step": 180480
    },
    {
      "epoch": 0.16824588470952848,
      "grad_norm": 0.985787570476532,
      "learning_rate": 0.00016635103693594936,
      "loss": 3.4589,
      "step": 180490
    },
    {
      "epoch": 0.16825520632760757,
      "grad_norm": 1.1523611545562744,
      "learning_rate": 0.00016634917261385416,
      "loss": 3.4267,
      "step": 180500
    },
    {
      "epoch": 0.16826452794568666,
      "grad_norm": 1.258141279220581,
      "learning_rate": 0.00016634730829175894,
      "loss": 3.5478,
      "step": 180510
    },
    {
      "epoch": 0.16827384956376576,
      "grad_norm": 0.9686047434806824,
      "learning_rate": 0.00016634544396966377,
      "loss": 3.4926,
      "step": 180520
    },
    {
      "epoch": 0.16828317118184485,
      "grad_norm": 0.9810992479324341,
      "learning_rate": 0.00016634357964756855,
      "loss": 3.5558,
      "step": 180530
    },
    {
      "epoch": 0.16829249279992395,
      "grad_norm": 1.279492974281311,
      "learning_rate": 0.00016634171532547336,
      "loss": 3.3929,
      "step": 180540
    },
    {
      "epoch": 0.16830181441800302,
      "grad_norm": 1.228172779083252,
      "learning_rate": 0.00016633985100337817,
      "loss": 3.4345,
      "step": 180550
    },
    {
      "epoch": 0.1683111360360821,
      "grad_norm": 0.9726352691650391,
      "learning_rate": 0.00016633798668128295,
      "loss": 3.375,
      "step": 180560
    },
    {
      "epoch": 0.1683204576541612,
      "grad_norm": 1.0647765398025513,
      "learning_rate": 0.00016633612235918775,
      "loss": 3.5137,
      "step": 180570
    },
    {
      "epoch": 0.1683297792722403,
      "grad_norm": 1.185377597808838,
      "learning_rate": 0.00016633425803709256,
      "loss": 3.636,
      "step": 180580
    },
    {
      "epoch": 0.1683391008903194,
      "grad_norm": 1.0705921649932861,
      "learning_rate": 0.00016633239371499736,
      "loss": 3.6403,
      "step": 180590
    },
    {
      "epoch": 0.1683484225083985,
      "grad_norm": 1.1258792877197266,
      "learning_rate": 0.00016633052939290217,
      "loss": 3.4596,
      "step": 180600
    },
    {
      "epoch": 0.16835774412647758,
      "grad_norm": 1.0933650732040405,
      "learning_rate": 0.00016632866507080695,
      "loss": 3.5915,
      "step": 180610
    },
    {
      "epoch": 0.16836706574455668,
      "grad_norm": 0.9524616003036499,
      "learning_rate": 0.00016632680074871176,
      "loss": 3.6956,
      "step": 180620
    },
    {
      "epoch": 0.16837638736263577,
      "grad_norm": 1.0763572454452515,
      "learning_rate": 0.00016632493642661656,
      "loss": 3.47,
      "step": 180630
    },
    {
      "epoch": 0.16838570898071487,
      "grad_norm": 1.0411689281463623,
      "learning_rate": 0.00016632307210452137,
      "loss": 3.5637,
      "step": 180640
    },
    {
      "epoch": 0.16839503059879393,
      "grad_norm": 0.9907417893409729,
      "learning_rate": 0.00016632120778242618,
      "loss": 3.6847,
      "step": 180650
    },
    {
      "epoch": 0.16840435221687303,
      "grad_norm": 1.149065375328064,
      "learning_rate": 0.00016631934346033095,
      "loss": 3.4976,
      "step": 180660
    },
    {
      "epoch": 0.16841367383495212,
      "grad_norm": 0.9442956447601318,
      "learning_rate": 0.00016631747913823576,
      "loss": 3.4541,
      "step": 180670
    },
    {
      "epoch": 0.16842299545303122,
      "grad_norm": 1.0509694814682007,
      "learning_rate": 0.00016631561481614054,
      "loss": 3.6228,
      "step": 180680
    },
    {
      "epoch": 0.1684323170711103,
      "grad_norm": 1.0792075395584106,
      "learning_rate": 0.00016631375049404537,
      "loss": 3.472,
      "step": 180690
    },
    {
      "epoch": 0.1684416386891894,
      "grad_norm": 1.1112922430038452,
      "learning_rate": 0.00016631188617195018,
      "loss": 3.4685,
      "step": 180700
    },
    {
      "epoch": 0.1684509603072685,
      "grad_norm": 1.3073638677597046,
      "learning_rate": 0.00016631002184985496,
      "loss": 3.4579,
      "step": 180710
    },
    {
      "epoch": 0.1684602819253476,
      "grad_norm": 1.3221179246902466,
      "learning_rate": 0.00016630815752775977,
      "loss": 3.6583,
      "step": 180720
    },
    {
      "epoch": 0.1684696035434267,
      "grad_norm": 1.312041163444519,
      "learning_rate": 0.00016630629320566454,
      "loss": 3.3348,
      "step": 180730
    },
    {
      "epoch": 0.16847892516150578,
      "grad_norm": 1.0295053720474243,
      "learning_rate": 0.00016630442888356935,
      "loss": 3.616,
      "step": 180740
    },
    {
      "epoch": 0.16848824677958488,
      "grad_norm": 0.9491500854492188,
      "learning_rate": 0.00016630256456147418,
      "loss": 3.4777,
      "step": 180750
    },
    {
      "epoch": 0.16849756839766394,
      "grad_norm": 1.0232856273651123,
      "learning_rate": 0.00016630070023937896,
      "loss": 3.6234,
      "step": 180760
    },
    {
      "epoch": 0.16850689001574304,
      "grad_norm": 1.0558810234069824,
      "learning_rate": 0.00016629883591728377,
      "loss": 3.6444,
      "step": 180770
    },
    {
      "epoch": 0.16851621163382213,
      "grad_norm": 0.9785712957382202,
      "learning_rate": 0.00016629697159518855,
      "loss": 3.6767,
      "step": 180780
    },
    {
      "epoch": 0.16852553325190123,
      "grad_norm": 1.1313185691833496,
      "learning_rate": 0.00016629510727309335,
      "loss": 3.5914,
      "step": 180790
    },
    {
      "epoch": 0.16853485486998032,
      "grad_norm": 1.0371664762496948,
      "learning_rate": 0.00016629324295099816,
      "loss": 3.4992,
      "step": 180800
    },
    {
      "epoch": 0.16854417648805942,
      "grad_norm": 0.982977032661438,
      "learning_rate": 0.00016629137862890297,
      "loss": 3.4483,
      "step": 180810
    },
    {
      "epoch": 0.1685534981061385,
      "grad_norm": 1.289847493171692,
      "learning_rate": 0.00016628951430680777,
      "loss": 3.4543,
      "step": 180820
    },
    {
      "epoch": 0.1685628197242176,
      "grad_norm": 1.144818902015686,
      "learning_rate": 0.00016628764998471255,
      "loss": 3.4366,
      "step": 180830
    },
    {
      "epoch": 0.1685721413422967,
      "grad_norm": 0.9846621751785278,
      "learning_rate": 0.00016628578566261736,
      "loss": 3.7978,
      "step": 180840
    },
    {
      "epoch": 0.1685814629603758,
      "grad_norm": 1.2275630235671997,
      "learning_rate": 0.00016628392134052217,
      "loss": 3.5603,
      "step": 180850
    },
    {
      "epoch": 0.16859078457845486,
      "grad_norm": 1.1217012405395508,
      "learning_rate": 0.00016628205701842697,
      "loss": 3.4918,
      "step": 180860
    },
    {
      "epoch": 0.16860010619653396,
      "grad_norm": 1.0696375370025635,
      "learning_rate": 0.00016628019269633178,
      "loss": 3.7228,
      "step": 180870
    },
    {
      "epoch": 0.16860942781461305,
      "grad_norm": 1.10086989402771,
      "learning_rate": 0.00016627832837423656,
      "loss": 3.6096,
      "step": 180880
    },
    {
      "epoch": 0.16861874943269214,
      "grad_norm": 0.9533489942550659,
      "learning_rate": 0.00016627646405214136,
      "loss": 3.6989,
      "step": 180890
    },
    {
      "epoch": 0.16862807105077124,
      "grad_norm": 1.2727900743484497,
      "learning_rate": 0.00016627459973004617,
      "loss": 3.5695,
      "step": 180900
    },
    {
      "epoch": 0.16863739266885033,
      "grad_norm": 0.9843382239341736,
      "learning_rate": 0.00016627273540795098,
      "loss": 3.5767,
      "step": 180910
    },
    {
      "epoch": 0.16864671428692943,
      "grad_norm": 0.972062349319458,
      "learning_rate": 0.00016627087108585578,
      "loss": 3.6635,
      "step": 180920
    },
    {
      "epoch": 0.16865603590500852,
      "grad_norm": 0.8408653140068054,
      "learning_rate": 0.00016626900676376056,
      "loss": 3.5481,
      "step": 180930
    },
    {
      "epoch": 0.16866535752308762,
      "grad_norm": 0.9887503981590271,
      "learning_rate": 0.00016626714244166537,
      "loss": 3.6223,
      "step": 180940
    },
    {
      "epoch": 0.1686746791411667,
      "grad_norm": 0.9427910447120667,
      "learning_rate": 0.00016626527811957017,
      "loss": 3.5443,
      "step": 180950
    },
    {
      "epoch": 0.1686840007592458,
      "grad_norm": 0.9869737029075623,
      "learning_rate": 0.00016626341379747495,
      "loss": 3.5643,
      "step": 180960
    },
    {
      "epoch": 0.16869332237732487,
      "grad_norm": 1.0462290048599243,
      "learning_rate": 0.0001662615494753798,
      "loss": 3.5863,
      "step": 180970
    },
    {
      "epoch": 0.16870264399540397,
      "grad_norm": 1.0729224681854248,
      "learning_rate": 0.00016625968515328457,
      "loss": 3.3688,
      "step": 180980
    },
    {
      "epoch": 0.16871196561348306,
      "grad_norm": 0.8380435705184937,
      "learning_rate": 0.00016625782083118937,
      "loss": 3.5032,
      "step": 180990
    },
    {
      "epoch": 0.16872128723156216,
      "grad_norm": 1.2539567947387695,
      "learning_rate": 0.00016625595650909418,
      "loss": 3.4314,
      "step": 181000
    },
    {
      "epoch": 0.16873060884964125,
      "grad_norm": 0.9001010060310364,
      "learning_rate": 0.00016625409218699896,
      "loss": 3.4747,
      "step": 181010
    },
    {
      "epoch": 0.16873993046772034,
      "grad_norm": 1.2827626466751099,
      "learning_rate": 0.00016625222786490376,
      "loss": 3.5771,
      "step": 181020
    },
    {
      "epoch": 0.16874925208579944,
      "grad_norm": 1.0585389137268066,
      "learning_rate": 0.00016625036354280857,
      "loss": 3.6049,
      "step": 181030
    },
    {
      "epoch": 0.16875857370387853,
      "grad_norm": 0.8605372309684753,
      "learning_rate": 0.00016624849922071338,
      "loss": 3.3962,
      "step": 181040
    },
    {
      "epoch": 0.16876789532195763,
      "grad_norm": 1.0408357381820679,
      "learning_rate": 0.00016624663489861818,
      "loss": 3.4205,
      "step": 181050
    },
    {
      "epoch": 0.16877721694003672,
      "grad_norm": 0.911616861820221,
      "learning_rate": 0.00016624477057652296,
      "loss": 3.6637,
      "step": 181060
    },
    {
      "epoch": 0.1687865385581158,
      "grad_norm": 0.9982996582984924,
      "learning_rate": 0.00016624290625442777,
      "loss": 3.3879,
      "step": 181070
    },
    {
      "epoch": 0.16879586017619488,
      "grad_norm": 0.9979476928710938,
      "learning_rate": 0.00016624104193233258,
      "loss": 3.6168,
      "step": 181080
    },
    {
      "epoch": 0.16880518179427398,
      "grad_norm": 1.0627148151397705,
      "learning_rate": 0.00016623917761023738,
      "loss": 3.4879,
      "step": 181090
    },
    {
      "epoch": 0.16881450341235307,
      "grad_norm": 0.9110140204429626,
      "learning_rate": 0.0001662373132881422,
      "loss": 3.4945,
      "step": 181100
    },
    {
      "epoch": 0.16882382503043217,
      "grad_norm": 1.0895360708236694,
      "learning_rate": 0.00016623544896604697,
      "loss": 3.6727,
      "step": 181110
    },
    {
      "epoch": 0.16883314664851126,
      "grad_norm": 1.221374273300171,
      "learning_rate": 0.00016623358464395177,
      "loss": 3.4421,
      "step": 181120
    },
    {
      "epoch": 0.16884246826659036,
      "grad_norm": 1.0459333658218384,
      "learning_rate": 0.00016623172032185655,
      "loss": 3.607,
      "step": 181130
    },
    {
      "epoch": 0.16885178988466945,
      "grad_norm": 1.0717836618423462,
      "learning_rate": 0.00016622985599976139,
      "loss": 3.5198,
      "step": 181140
    },
    {
      "epoch": 0.16886111150274855,
      "grad_norm": 1.7967309951782227,
      "learning_rate": 0.0001662279916776662,
      "loss": 3.4634,
      "step": 181150
    },
    {
      "epoch": 0.16887043312082764,
      "grad_norm": 1.1398102045059204,
      "learning_rate": 0.00016622612735557097,
      "loss": 3.5436,
      "step": 181160
    },
    {
      "epoch": 0.16887975473890673,
      "grad_norm": 1.1408344507217407,
      "learning_rate": 0.00016622426303347578,
      "loss": 3.3906,
      "step": 181170
    },
    {
      "epoch": 0.1688890763569858,
      "grad_norm": 1.1175036430358887,
      "learning_rate": 0.00016622239871138056,
      "loss": 3.6036,
      "step": 181180
    },
    {
      "epoch": 0.1688983979750649,
      "grad_norm": 1.0417050123214722,
      "learning_rate": 0.00016622053438928536,
      "loss": 3.4962,
      "step": 181190
    },
    {
      "epoch": 0.168907719593144,
      "grad_norm": 1.0094627141952515,
      "learning_rate": 0.0001662186700671902,
      "loss": 3.5103,
      "step": 181200
    },
    {
      "epoch": 0.16891704121122308,
      "grad_norm": 1.1527999639511108,
      "learning_rate": 0.00016621680574509498,
      "loss": 3.6071,
      "step": 181210
    },
    {
      "epoch": 0.16892636282930218,
      "grad_norm": 1.0054376125335693,
      "learning_rate": 0.00016621494142299978,
      "loss": 3.6105,
      "step": 181220
    },
    {
      "epoch": 0.16893568444738127,
      "grad_norm": 0.997739315032959,
      "learning_rate": 0.00016621307710090456,
      "loss": 3.5335,
      "step": 181230
    },
    {
      "epoch": 0.16894500606546037,
      "grad_norm": 1.1473625898361206,
      "learning_rate": 0.00016621121277880937,
      "loss": 3.8149,
      "step": 181240
    },
    {
      "epoch": 0.16895432768353946,
      "grad_norm": 1.0861519575119019,
      "learning_rate": 0.00016620934845671417,
      "loss": 3.4166,
      "step": 181250
    },
    {
      "epoch": 0.16896364930161856,
      "grad_norm": 1.0722882747650146,
      "learning_rate": 0.00016620748413461898,
      "loss": 3.5892,
      "step": 181260
    },
    {
      "epoch": 0.16897297091969765,
      "grad_norm": 1.1598511934280396,
      "learning_rate": 0.0001662056198125238,
      "loss": 3.5545,
      "step": 181270
    },
    {
      "epoch": 0.16898229253777672,
      "grad_norm": 0.9915751218795776,
      "learning_rate": 0.00016620375549042857,
      "loss": 3.5464,
      "step": 181280
    },
    {
      "epoch": 0.1689916141558558,
      "grad_norm": 1.1402783393859863,
      "learning_rate": 0.00016620189116833337,
      "loss": 3.5988,
      "step": 181290
    },
    {
      "epoch": 0.1690009357739349,
      "grad_norm": 1.211227536201477,
      "learning_rate": 0.00016620002684623818,
      "loss": 3.4013,
      "step": 181300
    },
    {
      "epoch": 0.169010257392014,
      "grad_norm": 1.0570828914642334,
      "learning_rate": 0.00016619816252414298,
      "loss": 3.6388,
      "step": 181310
    },
    {
      "epoch": 0.1690195790100931,
      "grad_norm": 1.0113229751586914,
      "learning_rate": 0.0001661962982020478,
      "loss": 3.3463,
      "step": 181320
    },
    {
      "epoch": 0.1690289006281722,
      "grad_norm": 0.9250220060348511,
      "learning_rate": 0.00016619443387995257,
      "loss": 3.5075,
      "step": 181330
    },
    {
      "epoch": 0.16903822224625129,
      "grad_norm": 1.1106846332550049,
      "learning_rate": 0.00016619256955785738,
      "loss": 3.4534,
      "step": 181340
    },
    {
      "epoch": 0.16904754386433038,
      "grad_norm": 1.1108014583587646,
      "learning_rate": 0.00016619070523576218,
      "loss": 3.6032,
      "step": 181350
    },
    {
      "epoch": 0.16905686548240947,
      "grad_norm": 0.9320165514945984,
      "learning_rate": 0.000166188840913667,
      "loss": 3.4561,
      "step": 181360
    },
    {
      "epoch": 0.16906618710048857,
      "grad_norm": 1.0062000751495361,
      "learning_rate": 0.0001661869765915718,
      "loss": 3.5632,
      "step": 181370
    },
    {
      "epoch": 0.16907550871856766,
      "grad_norm": 1.0783863067626953,
      "learning_rate": 0.00016618511226947657,
      "loss": 3.4502,
      "step": 181380
    },
    {
      "epoch": 0.16908483033664673,
      "grad_norm": 1.0112075805664062,
      "learning_rate": 0.00016618324794738138,
      "loss": 3.3816,
      "step": 181390
    },
    {
      "epoch": 0.16909415195472582,
      "grad_norm": 1.1952428817749023,
      "learning_rate": 0.0001661813836252862,
      "loss": 3.4508,
      "step": 181400
    },
    {
      "epoch": 0.16910347357280492,
      "grad_norm": 0.9863359332084656,
      "learning_rate": 0.00016617951930319097,
      "loss": 3.2874,
      "step": 181410
    },
    {
      "epoch": 0.169112795190884,
      "grad_norm": 1.4068349599838257,
      "learning_rate": 0.0001661776549810958,
      "loss": 3.4422,
      "step": 181420
    },
    {
      "epoch": 0.1691221168089631,
      "grad_norm": 1.278416395187378,
      "learning_rate": 0.00016617579065900058,
      "loss": 3.4533,
      "step": 181430
    },
    {
      "epoch": 0.1691314384270422,
      "grad_norm": 1.1347339153289795,
      "learning_rate": 0.00016617392633690539,
      "loss": 3.256,
      "step": 181440
    },
    {
      "epoch": 0.1691407600451213,
      "grad_norm": 1.1033817529678345,
      "learning_rate": 0.0001661720620148102,
      "loss": 3.3902,
      "step": 181450
    },
    {
      "epoch": 0.1691500816632004,
      "grad_norm": 1.1347289085388184,
      "learning_rate": 0.00016617019769271497,
      "loss": 3.4101,
      "step": 181460
    },
    {
      "epoch": 0.16915940328127949,
      "grad_norm": 1.1311131715774536,
      "learning_rate": 0.00016616833337061978,
      "loss": 3.3938,
      "step": 181470
    },
    {
      "epoch": 0.16916872489935858,
      "grad_norm": 1.049157977104187,
      "learning_rate": 0.00016616646904852458,
      "loss": 3.4811,
      "step": 181480
    },
    {
      "epoch": 0.16917804651743765,
      "grad_norm": 1.0837208032608032,
      "learning_rate": 0.0001661646047264294,
      "loss": 3.497,
      "step": 181490
    },
    {
      "epoch": 0.16918736813551674,
      "grad_norm": 0.9911950826644897,
      "learning_rate": 0.0001661627404043342,
      "loss": 3.582,
      "step": 181500
    },
    {
      "epoch": 0.16919668975359584,
      "grad_norm": 1.024243950843811,
      "learning_rate": 0.00016616087608223898,
      "loss": 3.5165,
      "step": 181510
    },
    {
      "epoch": 0.16920601137167493,
      "grad_norm": 0.9897884726524353,
      "learning_rate": 0.00016615901176014378,
      "loss": 3.4055,
      "step": 181520
    },
    {
      "epoch": 0.16921533298975402,
      "grad_norm": 0.9427846670150757,
      "learning_rate": 0.0001661571474380486,
      "loss": 3.462,
      "step": 181530
    },
    {
      "epoch": 0.16922465460783312,
      "grad_norm": 1.1370220184326172,
      "learning_rate": 0.0001661552831159534,
      "loss": 3.5078,
      "step": 181540
    },
    {
      "epoch": 0.1692339762259122,
      "grad_norm": 2.002877712249756,
      "learning_rate": 0.0001661534187938582,
      "loss": 3.6901,
      "step": 181550
    },
    {
      "epoch": 0.1692432978439913,
      "grad_norm": 1.4082834720611572,
      "learning_rate": 0.00016615155447176298,
      "loss": 3.3333,
      "step": 181560
    },
    {
      "epoch": 0.1692526194620704,
      "grad_norm": 0.9602926969528198,
      "learning_rate": 0.00016614969014966779,
      "loss": 3.5504,
      "step": 181570
    },
    {
      "epoch": 0.1692619410801495,
      "grad_norm": 1.0627416372299194,
      "learning_rate": 0.00016614782582757257,
      "loss": 3.6424,
      "step": 181580
    },
    {
      "epoch": 0.1692712626982286,
      "grad_norm": 1.1208655834197998,
      "learning_rate": 0.0001661459615054774,
      "loss": 3.6328,
      "step": 181590
    },
    {
      "epoch": 0.16928058431630766,
      "grad_norm": 1.0361477136611938,
      "learning_rate": 0.0001661440971833822,
      "loss": 3.4574,
      "step": 181600
    },
    {
      "epoch": 0.16928990593438675,
      "grad_norm": 1.0524996519088745,
      "learning_rate": 0.00016614223286128698,
      "loss": 3.6516,
      "step": 181610
    },
    {
      "epoch": 0.16929922755246585,
      "grad_norm": 1.256291389465332,
      "learning_rate": 0.0001661403685391918,
      "loss": 3.6558,
      "step": 181620
    },
    {
      "epoch": 0.16930854917054494,
      "grad_norm": 1.0598758459091187,
      "learning_rate": 0.00016613850421709657,
      "loss": 3.6155,
      "step": 181630
    },
    {
      "epoch": 0.16931787078862404,
      "grad_norm": 1.0321825742721558,
      "learning_rate": 0.00016613663989500138,
      "loss": 3.662,
      "step": 181640
    },
    {
      "epoch": 0.16932719240670313,
      "grad_norm": 1.0776360034942627,
      "learning_rate": 0.0001661347755729062,
      "loss": 3.5844,
      "step": 181650
    },
    {
      "epoch": 0.16933651402478223,
      "grad_norm": 0.9489246010780334,
      "learning_rate": 0.000166132911250811,
      "loss": 3.5668,
      "step": 181660
    },
    {
      "epoch": 0.16934583564286132,
      "grad_norm": 1.3351250886917114,
      "learning_rate": 0.0001661310469287158,
      "loss": 3.3191,
      "step": 181670
    },
    {
      "epoch": 0.16935515726094041,
      "grad_norm": 1.0685625076293945,
      "learning_rate": 0.00016612918260662057,
      "loss": 3.3882,
      "step": 181680
    },
    {
      "epoch": 0.1693644788790195,
      "grad_norm": 0.9365777373313904,
      "learning_rate": 0.00016612731828452538,
      "loss": 3.6207,
      "step": 181690
    },
    {
      "epoch": 0.16937380049709858,
      "grad_norm": 1.1644902229309082,
      "learning_rate": 0.0001661254539624302,
      "loss": 3.6131,
      "step": 181700
    },
    {
      "epoch": 0.16938312211517767,
      "grad_norm": 1.095974087715149,
      "learning_rate": 0.000166123589640335,
      "loss": 3.5617,
      "step": 181710
    },
    {
      "epoch": 0.16939244373325676,
      "grad_norm": 0.9062914848327637,
      "learning_rate": 0.0001661217253182398,
      "loss": 3.5871,
      "step": 181720
    },
    {
      "epoch": 0.16940176535133586,
      "grad_norm": 1.1240922212600708,
      "learning_rate": 0.00016611986099614458,
      "loss": 3.3426,
      "step": 181730
    },
    {
      "epoch": 0.16941108696941495,
      "grad_norm": 1.225275993347168,
      "learning_rate": 0.00016611799667404939,
      "loss": 3.3008,
      "step": 181740
    },
    {
      "epoch": 0.16942040858749405,
      "grad_norm": 1.069132685661316,
      "learning_rate": 0.0001661161323519542,
      "loss": 3.5462,
      "step": 181750
    },
    {
      "epoch": 0.16942973020557314,
      "grad_norm": 0.9874994158744812,
      "learning_rate": 0.000166114268029859,
      "loss": 3.6367,
      "step": 181760
    },
    {
      "epoch": 0.16943905182365224,
      "grad_norm": 1.0718685388565063,
      "learning_rate": 0.0001661124037077638,
      "loss": 3.4426,
      "step": 181770
    },
    {
      "epoch": 0.16944837344173133,
      "grad_norm": 1.0490423440933228,
      "learning_rate": 0.00016611053938566858,
      "loss": 3.5035,
      "step": 181780
    },
    {
      "epoch": 0.16945769505981043,
      "grad_norm": 1.1412512063980103,
      "learning_rate": 0.0001661086750635734,
      "loss": 3.4496,
      "step": 181790
    },
    {
      "epoch": 0.1694670166778895,
      "grad_norm": 1.1816514730453491,
      "learning_rate": 0.0001661068107414782,
      "loss": 3.3982,
      "step": 181800
    },
    {
      "epoch": 0.1694763382959686,
      "grad_norm": 1.079443097114563,
      "learning_rate": 0.000166104946419383,
      "loss": 3.4347,
      "step": 181810
    },
    {
      "epoch": 0.16948565991404768,
      "grad_norm": 0.8922663927078247,
      "learning_rate": 0.0001661030820972878,
      "loss": 3.556,
      "step": 181820
    },
    {
      "epoch": 0.16949498153212678,
      "grad_norm": 1.0265246629714966,
      "learning_rate": 0.0001661012177751926,
      "loss": 3.6257,
      "step": 181830
    },
    {
      "epoch": 0.16950430315020587,
      "grad_norm": 1.0940693616867065,
      "learning_rate": 0.0001660993534530974,
      "loss": 3.4074,
      "step": 181840
    },
    {
      "epoch": 0.16951362476828497,
      "grad_norm": 1.126319169998169,
      "learning_rate": 0.00016609748913100217,
      "loss": 3.2922,
      "step": 181850
    },
    {
      "epoch": 0.16952294638636406,
      "grad_norm": 1.0255310535430908,
      "learning_rate": 0.00016609562480890698,
      "loss": 3.5491,
      "step": 181860
    },
    {
      "epoch": 0.16953226800444315,
      "grad_norm": 0.9388129115104675,
      "learning_rate": 0.0001660937604868118,
      "loss": 3.4971,
      "step": 181870
    },
    {
      "epoch": 0.16954158962252225,
      "grad_norm": 1.056456446647644,
      "learning_rate": 0.0001660918961647166,
      "loss": 3.49,
      "step": 181880
    },
    {
      "epoch": 0.16955091124060134,
      "grad_norm": 1.3727641105651855,
      "learning_rate": 0.0001660900318426214,
      "loss": 3.6555,
      "step": 181890
    },
    {
      "epoch": 0.16956023285868044,
      "grad_norm": 1.0268954038619995,
      "learning_rate": 0.00016608816752052618,
      "loss": 3.516,
      "step": 181900
    },
    {
      "epoch": 0.1695695544767595,
      "grad_norm": 1.203813076019287,
      "learning_rate": 0.00016608630319843098,
      "loss": 3.6603,
      "step": 181910
    },
    {
      "epoch": 0.1695788760948386,
      "grad_norm": 1.1246813535690308,
      "learning_rate": 0.0001660844388763358,
      "loss": 3.6815,
      "step": 181920
    },
    {
      "epoch": 0.1695881977129177,
      "grad_norm": 1.0317578315734863,
      "learning_rate": 0.0001660825745542406,
      "loss": 3.5076,
      "step": 181930
    },
    {
      "epoch": 0.1695975193309968,
      "grad_norm": 1.0420122146606445,
      "learning_rate": 0.0001660807102321454,
      "loss": 3.2356,
      "step": 181940
    },
    {
      "epoch": 0.16960684094907588,
      "grad_norm": 0.783368706703186,
      "learning_rate": 0.00016607884591005018,
      "loss": 3.4175,
      "step": 181950
    },
    {
      "epoch": 0.16961616256715498,
      "grad_norm": 1.224193811416626,
      "learning_rate": 0.000166076981587955,
      "loss": 3.5651,
      "step": 181960
    },
    {
      "epoch": 0.16962548418523407,
      "grad_norm": 1.1801140308380127,
      "learning_rate": 0.0001660751172658598,
      "loss": 3.3735,
      "step": 181970
    },
    {
      "epoch": 0.16963480580331317,
      "grad_norm": 1.0218794345855713,
      "learning_rate": 0.0001660732529437646,
      "loss": 3.4549,
      "step": 181980
    },
    {
      "epoch": 0.16964412742139226,
      "grad_norm": 1.1346278190612793,
      "learning_rate": 0.0001660713886216694,
      "loss": 3.6069,
      "step": 181990
    },
    {
      "epoch": 0.16965344903947135,
      "grad_norm": 1.2169491052627563,
      "learning_rate": 0.0001660695242995742,
      "loss": 3.5356,
      "step": 182000
    },
    {
      "epoch": 0.16966277065755042,
      "grad_norm": 1.2169060707092285,
      "learning_rate": 0.000166067659977479,
      "loss": 3.2445,
      "step": 182010
    },
    {
      "epoch": 0.16967209227562952,
      "grad_norm": 1.080383539199829,
      "learning_rate": 0.0001660657956553838,
      "loss": 3.5261,
      "step": 182020
    },
    {
      "epoch": 0.1696814138937086,
      "grad_norm": 1.1178526878356934,
      "learning_rate": 0.00016606393133328858,
      "loss": 3.5622,
      "step": 182030
    },
    {
      "epoch": 0.1696907355117877,
      "grad_norm": 1.1102854013442993,
      "learning_rate": 0.0001660620670111934,
      "loss": 3.4133,
      "step": 182040
    },
    {
      "epoch": 0.1697000571298668,
      "grad_norm": 1.038469910621643,
      "learning_rate": 0.0001660602026890982,
      "loss": 3.5408,
      "step": 182050
    },
    {
      "epoch": 0.1697093787479459,
      "grad_norm": 1.1519159078598022,
      "learning_rate": 0.000166058338367003,
      "loss": 3.4483,
      "step": 182060
    },
    {
      "epoch": 0.169718700366025,
      "grad_norm": 1.0348557233810425,
      "learning_rate": 0.0001660564740449078,
      "loss": 3.5629,
      "step": 182070
    },
    {
      "epoch": 0.16972802198410408,
      "grad_norm": 1.2013355493545532,
      "learning_rate": 0.00016605460972281258,
      "loss": 3.4859,
      "step": 182080
    },
    {
      "epoch": 0.16973734360218318,
      "grad_norm": 0.9815104007720947,
      "learning_rate": 0.0001660527454007174,
      "loss": 3.4983,
      "step": 182090
    },
    {
      "epoch": 0.16974666522026227,
      "grad_norm": 1.1386873722076416,
      "learning_rate": 0.0001660508810786222,
      "loss": 3.413,
      "step": 182100
    },
    {
      "epoch": 0.16975598683834137,
      "grad_norm": 1.2852007150650024,
      "learning_rate": 0.000166049016756527,
      "loss": 3.647,
      "step": 182110
    },
    {
      "epoch": 0.16976530845642043,
      "grad_norm": 1.040634274482727,
      "learning_rate": 0.0001660471524344318,
      "loss": 3.6158,
      "step": 182120
    },
    {
      "epoch": 0.16977463007449953,
      "grad_norm": 0.9816538095474243,
      "learning_rate": 0.0001660452881123366,
      "loss": 3.5455,
      "step": 182130
    },
    {
      "epoch": 0.16978395169257862,
      "grad_norm": 1.1700323820114136,
      "learning_rate": 0.0001660434237902414,
      "loss": 3.5881,
      "step": 182140
    },
    {
      "epoch": 0.16979327331065772,
      "grad_norm": 1.0519683361053467,
      "learning_rate": 0.0001660415594681462,
      "loss": 3.7129,
      "step": 182150
    },
    {
      "epoch": 0.1698025949287368,
      "grad_norm": 0.9899519681930542,
      "learning_rate": 0.000166039695146051,
      "loss": 3.5164,
      "step": 182160
    },
    {
      "epoch": 0.1698119165468159,
      "grad_norm": 0.9336718320846558,
      "learning_rate": 0.0001660378308239558,
      "loss": 3.3824,
      "step": 182170
    },
    {
      "epoch": 0.169821238164895,
      "grad_norm": 1.1143633127212524,
      "learning_rate": 0.0001660359665018606,
      "loss": 3.4232,
      "step": 182180
    },
    {
      "epoch": 0.1698305597829741,
      "grad_norm": 1.2837401628494263,
      "learning_rate": 0.0001660341021797654,
      "loss": 3.4558,
      "step": 182190
    },
    {
      "epoch": 0.1698398814010532,
      "grad_norm": 1.034772276878357,
      "learning_rate": 0.0001660322378576702,
      "loss": 3.5271,
      "step": 182200
    },
    {
      "epoch": 0.16984920301913228,
      "grad_norm": 0.9799492955207825,
      "learning_rate": 0.000166030373535575,
      "loss": 3.4996,
      "step": 182210
    },
    {
      "epoch": 0.16985852463721135,
      "grad_norm": 1.1009438037872314,
      "learning_rate": 0.00016602850921347982,
      "loss": 3.3647,
      "step": 182220
    },
    {
      "epoch": 0.16986784625529044,
      "grad_norm": 1.105916142463684,
      "learning_rate": 0.0001660266448913846,
      "loss": 3.4287,
      "step": 182230
    },
    {
      "epoch": 0.16987716787336954,
      "grad_norm": 1.0096626281738281,
      "learning_rate": 0.0001660247805692894,
      "loss": 3.5534,
      "step": 182240
    },
    {
      "epoch": 0.16988648949144863,
      "grad_norm": 1.1577688455581665,
      "learning_rate": 0.00016602291624719418,
      "loss": 3.5984,
      "step": 182250
    },
    {
      "epoch": 0.16989581110952773,
      "grad_norm": 1.089330792427063,
      "learning_rate": 0.00016602105192509901,
      "loss": 3.5045,
      "step": 182260
    },
    {
      "epoch": 0.16990513272760682,
      "grad_norm": 1.3046178817749023,
      "learning_rate": 0.00016601918760300382,
      "loss": 3.5469,
      "step": 182270
    },
    {
      "epoch": 0.16991445434568592,
      "grad_norm": 1.0488439798355103,
      "learning_rate": 0.0001660173232809086,
      "loss": 3.4991,
      "step": 182280
    },
    {
      "epoch": 0.169923775963765,
      "grad_norm": 1.0097957849502563,
      "learning_rate": 0.0001660154589588134,
      "loss": 3.4034,
      "step": 182290
    },
    {
      "epoch": 0.1699330975818441,
      "grad_norm": 0.964210033416748,
      "learning_rate": 0.00016601359463671819,
      "loss": 3.6761,
      "step": 182300
    },
    {
      "epoch": 0.1699424191999232,
      "grad_norm": 1.195955514907837,
      "learning_rate": 0.000166011730314623,
      "loss": 3.4813,
      "step": 182310
    },
    {
      "epoch": 0.1699517408180023,
      "grad_norm": 1.0618433952331543,
      "learning_rate": 0.00016600986599252783,
      "loss": 3.5516,
      "step": 182320
    },
    {
      "epoch": 0.16996106243608136,
      "grad_norm": 1.220900297164917,
      "learning_rate": 0.0001660080016704326,
      "loss": 3.5688,
      "step": 182330
    },
    {
      "epoch": 0.16997038405416046,
      "grad_norm": 1.0111216306686401,
      "learning_rate": 0.0001660061373483374,
      "loss": 3.6365,
      "step": 182340
    },
    {
      "epoch": 0.16997970567223955,
      "grad_norm": 1.2511365413665771,
      "learning_rate": 0.0001660042730262422,
      "loss": 3.4077,
      "step": 182350
    },
    {
      "epoch": 0.16998902729031865,
      "grad_norm": 1.0236315727233887,
      "learning_rate": 0.000166002408704147,
      "loss": 3.4516,
      "step": 182360
    },
    {
      "epoch": 0.16999834890839774,
      "grad_norm": 1.0121172666549683,
      "learning_rate": 0.0001660005443820518,
      "loss": 3.5404,
      "step": 182370
    },
    {
      "epoch": 0.17000767052647683,
      "grad_norm": 1.0076621770858765,
      "learning_rate": 0.0001659986800599566,
      "loss": 3.4538,
      "step": 182380
    },
    {
      "epoch": 0.17001699214455593,
      "grad_norm": 1.0741374492645264,
      "learning_rate": 0.00016599681573786142,
      "loss": 3.5672,
      "step": 182390
    },
    {
      "epoch": 0.17002631376263502,
      "grad_norm": 1.1782441139221191,
      "learning_rate": 0.0001659949514157662,
      "loss": 3.5141,
      "step": 182400
    },
    {
      "epoch": 0.17003563538071412,
      "grad_norm": 1.4728467464447021,
      "learning_rate": 0.000165993087093671,
      "loss": 3.5626,
      "step": 182410
    },
    {
      "epoch": 0.1700449569987932,
      "grad_norm": 1.1272681951522827,
      "learning_rate": 0.0001659912227715758,
      "loss": 3.7138,
      "step": 182420
    },
    {
      "epoch": 0.17005427861687228,
      "grad_norm": 1.1887054443359375,
      "learning_rate": 0.00016598935844948061,
      "loss": 3.6225,
      "step": 182430
    },
    {
      "epoch": 0.17006360023495137,
      "grad_norm": 1.1195452213287354,
      "learning_rate": 0.00016598749412738542,
      "loss": 3.5482,
      "step": 182440
    },
    {
      "epoch": 0.17007292185303047,
      "grad_norm": 1.0117560625076294,
      "learning_rate": 0.0001659856298052902,
      "loss": 3.3186,
      "step": 182450
    },
    {
      "epoch": 0.17008224347110956,
      "grad_norm": 0.932949960231781,
      "learning_rate": 0.000165983765483195,
      "loss": 3.5687,
      "step": 182460
    },
    {
      "epoch": 0.17009156508918866,
      "grad_norm": 1.2837475538253784,
      "learning_rate": 0.0001659819011610998,
      "loss": 3.5914,
      "step": 182470
    },
    {
      "epoch": 0.17010088670726775,
      "grad_norm": 1.1103613376617432,
      "learning_rate": 0.0001659800368390046,
      "loss": 3.5139,
      "step": 182480
    },
    {
      "epoch": 0.17011020832534685,
      "grad_norm": 1.640820860862732,
      "learning_rate": 0.00016597817251690942,
      "loss": 3.5382,
      "step": 182490
    },
    {
      "epoch": 0.17011952994342594,
      "grad_norm": 1.0279608964920044,
      "learning_rate": 0.0001659763081948142,
      "loss": 3.3758,
      "step": 182500
    },
    {
      "epoch": 0.17012885156150503,
      "grad_norm": 0.996048092842102,
      "learning_rate": 0.000165974443872719,
      "loss": 3.6374,
      "step": 182510
    },
    {
      "epoch": 0.17013817317958413,
      "grad_norm": 1.3343051671981812,
      "learning_rate": 0.00016597257955062382,
      "loss": 3.4315,
      "step": 182520
    },
    {
      "epoch": 0.17014749479766322,
      "grad_norm": 1.167927622795105,
      "learning_rate": 0.0001659707152285286,
      "loss": 3.3296,
      "step": 182530
    },
    {
      "epoch": 0.1701568164157423,
      "grad_norm": 1.0172455310821533,
      "learning_rate": 0.0001659688509064334,
      "loss": 3.4462,
      "step": 182540
    },
    {
      "epoch": 0.17016613803382138,
      "grad_norm": 1.183787226676941,
      "learning_rate": 0.0001659669865843382,
      "loss": 3.3709,
      "step": 182550
    },
    {
      "epoch": 0.17017545965190048,
      "grad_norm": 1.1694997549057007,
      "learning_rate": 0.00016596512226224301,
      "loss": 3.4439,
      "step": 182560
    },
    {
      "epoch": 0.17018478126997957,
      "grad_norm": 0.9443683624267578,
      "learning_rate": 0.00016596325794014782,
      "loss": 3.6475,
      "step": 182570
    },
    {
      "epoch": 0.17019410288805867,
      "grad_norm": 1.186086893081665,
      "learning_rate": 0.0001659613936180526,
      "loss": 3.3046,
      "step": 182580
    },
    {
      "epoch": 0.17020342450613776,
      "grad_norm": 1.080928087234497,
      "learning_rate": 0.0001659595292959574,
      "loss": 3.531,
      "step": 182590
    },
    {
      "epoch": 0.17021274612421686,
      "grad_norm": 1.0551211833953857,
      "learning_rate": 0.0001659576649738622,
      "loss": 3.4964,
      "step": 182600
    },
    {
      "epoch": 0.17022206774229595,
      "grad_norm": 1.017826795578003,
      "learning_rate": 0.00016595580065176702,
      "loss": 3.4719,
      "step": 182610
    },
    {
      "epoch": 0.17023138936037505,
      "grad_norm": 1.348719835281372,
      "learning_rate": 0.00016595393632967183,
      "loss": 3.4798,
      "step": 182620
    },
    {
      "epoch": 0.17024071097845414,
      "grad_norm": 1.19679856300354,
      "learning_rate": 0.0001659520720075766,
      "loss": 3.5361,
      "step": 182630
    },
    {
      "epoch": 0.1702500325965332,
      "grad_norm": 0.8698492050170898,
      "learning_rate": 0.0001659502076854814,
      "loss": 3.6324,
      "step": 182640
    },
    {
      "epoch": 0.1702593542146123,
      "grad_norm": 1.1196829080581665,
      "learning_rate": 0.00016594834336338622,
      "loss": 3.5159,
      "step": 182650
    },
    {
      "epoch": 0.1702686758326914,
      "grad_norm": 1.134468913078308,
      "learning_rate": 0.00016594647904129102,
      "loss": 3.6602,
      "step": 182660
    },
    {
      "epoch": 0.1702779974507705,
      "grad_norm": 1.0258097648620605,
      "learning_rate": 0.00016594461471919583,
      "loss": 3.7456,
      "step": 182670
    },
    {
      "epoch": 0.17028731906884959,
      "grad_norm": 0.9404525756835938,
      "learning_rate": 0.0001659427503971006,
      "loss": 3.4001,
      "step": 182680
    },
    {
      "epoch": 0.17029664068692868,
      "grad_norm": 1.1826574802398682,
      "learning_rate": 0.00016594088607500542,
      "loss": 3.5647,
      "step": 182690
    },
    {
      "epoch": 0.17030596230500777,
      "grad_norm": 1.0344748497009277,
      "learning_rate": 0.0001659390217529102,
      "loss": 3.5503,
      "step": 182700
    },
    {
      "epoch": 0.17031528392308687,
      "grad_norm": 1.0403399467468262,
      "learning_rate": 0.00016593715743081503,
      "loss": 3.4689,
      "step": 182710
    },
    {
      "epoch": 0.17032460554116596,
      "grad_norm": 1.1260044574737549,
      "learning_rate": 0.00016593529310871983,
      "loss": 3.6151,
      "step": 182720
    },
    {
      "epoch": 0.17033392715924506,
      "grad_norm": 0.9213557243347168,
      "learning_rate": 0.0001659334287866246,
      "loss": 3.5856,
      "step": 182730
    },
    {
      "epoch": 0.17034324877732415,
      "grad_norm": 0.9924399852752686,
      "learning_rate": 0.00016593156446452942,
      "loss": 3.4244,
      "step": 182740
    },
    {
      "epoch": 0.17035257039540322,
      "grad_norm": 1.3495943546295166,
      "learning_rate": 0.0001659297001424342,
      "loss": 3.5646,
      "step": 182750
    },
    {
      "epoch": 0.1703618920134823,
      "grad_norm": 1.0066211223602295,
      "learning_rate": 0.000165927835820339,
      "loss": 3.5058,
      "step": 182760
    },
    {
      "epoch": 0.1703712136315614,
      "grad_norm": 1.0331041812896729,
      "learning_rate": 0.00016592597149824384,
      "loss": 3.4296,
      "step": 182770
    },
    {
      "epoch": 0.1703805352496405,
      "grad_norm": 1.098710298538208,
      "learning_rate": 0.00016592410717614862,
      "loss": 3.7055,
      "step": 182780
    },
    {
      "epoch": 0.1703898568677196,
      "grad_norm": 1.0451701879501343,
      "learning_rate": 0.00016592224285405342,
      "loss": 3.5948,
      "step": 182790
    },
    {
      "epoch": 0.1703991784857987,
      "grad_norm": 1.1798765659332275,
      "learning_rate": 0.0001659203785319582,
      "loss": 3.5024,
      "step": 182800
    },
    {
      "epoch": 0.17040850010387779,
      "grad_norm": 1.0598647594451904,
      "learning_rate": 0.000165918514209863,
      "loss": 3.677,
      "step": 182810
    },
    {
      "epoch": 0.17041782172195688,
      "grad_norm": 0.8528355360031128,
      "learning_rate": 0.00016591664988776782,
      "loss": 3.503,
      "step": 182820
    },
    {
      "epoch": 0.17042714334003597,
      "grad_norm": 1.0138134956359863,
      "learning_rate": 0.00016591478556567262,
      "loss": 3.6245,
      "step": 182830
    },
    {
      "epoch": 0.17043646495811507,
      "grad_norm": 1.0066019296646118,
      "learning_rate": 0.00016591292124357743,
      "loss": 3.471,
      "step": 182840
    },
    {
      "epoch": 0.17044578657619414,
      "grad_norm": 1.347611665725708,
      "learning_rate": 0.0001659110569214822,
      "loss": 3.5023,
      "step": 182850
    },
    {
      "epoch": 0.17045510819427323,
      "grad_norm": 0.9499000906944275,
      "learning_rate": 0.00016590919259938701,
      "loss": 3.5114,
      "step": 182860
    },
    {
      "epoch": 0.17046442981235232,
      "grad_norm": 1.136216402053833,
      "learning_rate": 0.00016590732827729182,
      "loss": 3.5775,
      "step": 182870
    },
    {
      "epoch": 0.17047375143043142,
      "grad_norm": 1.049302577972412,
      "learning_rate": 0.00016590546395519663,
      "loss": 3.4502,
      "step": 182880
    },
    {
      "epoch": 0.17048307304851051,
      "grad_norm": 0.9531486630439758,
      "learning_rate": 0.00016590359963310143,
      "loss": 3.7593,
      "step": 182890
    },
    {
      "epoch": 0.1704923946665896,
      "grad_norm": 1.057679533958435,
      "learning_rate": 0.0001659017353110062,
      "loss": 3.5808,
      "step": 182900
    },
    {
      "epoch": 0.1705017162846687,
      "grad_norm": 1.2059413194656372,
      "learning_rate": 0.00016589987098891102,
      "loss": 3.3911,
      "step": 182910
    },
    {
      "epoch": 0.1705110379027478,
      "grad_norm": 1.139917016029358,
      "learning_rate": 0.00016589800666681582,
      "loss": 3.5428,
      "step": 182920
    },
    {
      "epoch": 0.1705203595208269,
      "grad_norm": 1.1145309209823608,
      "learning_rate": 0.00016589614234472063,
      "loss": 3.5829,
      "step": 182930
    },
    {
      "epoch": 0.170529681138906,
      "grad_norm": 0.9764003157615662,
      "learning_rate": 0.00016589427802262544,
      "loss": 3.5593,
      "step": 182940
    },
    {
      "epoch": 0.17053900275698508,
      "grad_norm": 1.0065388679504395,
      "learning_rate": 0.00016589241370053022,
      "loss": 3.513,
      "step": 182950
    },
    {
      "epoch": 0.17054832437506415,
      "grad_norm": 1.0264965295791626,
      "learning_rate": 0.00016589054937843502,
      "loss": 3.365,
      "step": 182960
    },
    {
      "epoch": 0.17055764599314324,
      "grad_norm": 1.0427258014678955,
      "learning_rate": 0.00016588868505633983,
      "loss": 3.4394,
      "step": 182970
    },
    {
      "epoch": 0.17056696761122234,
      "grad_norm": 0.9261929392814636,
      "learning_rate": 0.0001658868207342446,
      "loss": 3.2585,
      "step": 182980
    },
    {
      "epoch": 0.17057628922930143,
      "grad_norm": 1.1142913103103638,
      "learning_rate": 0.00016588495641214941,
      "loss": 3.4129,
      "step": 182990
    },
    {
      "epoch": 0.17058561084738053,
      "grad_norm": 1.0554957389831543,
      "learning_rate": 0.00016588309209005422,
      "loss": 3.5876,
      "step": 183000
    },
    {
      "epoch": 0.17059493246545962,
      "grad_norm": 1.013486623764038,
      "learning_rate": 0.00016588122776795903,
      "loss": 3.5049,
      "step": 183010
    },
    {
      "epoch": 0.17060425408353871,
      "grad_norm": 0.9829257726669312,
      "learning_rate": 0.00016587936344586383,
      "loss": 3.3344,
      "step": 183020
    },
    {
      "epoch": 0.1706135757016178,
      "grad_norm": 1.140071153640747,
      "learning_rate": 0.0001658774991237686,
      "loss": 3.5416,
      "step": 183030
    },
    {
      "epoch": 0.1706228973196969,
      "grad_norm": 1.0390892028808594,
      "learning_rate": 0.00016587563480167342,
      "loss": 3.5718,
      "step": 183040
    },
    {
      "epoch": 0.170632218937776,
      "grad_norm": 1.0823420286178589,
      "learning_rate": 0.00016587377047957823,
      "loss": 3.6621,
      "step": 183050
    },
    {
      "epoch": 0.17064154055585506,
      "grad_norm": 0.8150296807289124,
      "learning_rate": 0.00016587190615748303,
      "loss": 3.6866,
      "step": 183060
    },
    {
      "epoch": 0.17065086217393416,
      "grad_norm": 1.0041866302490234,
      "learning_rate": 0.00016587004183538784,
      "loss": 3.502,
      "step": 183070
    },
    {
      "epoch": 0.17066018379201325,
      "grad_norm": 1.0304317474365234,
      "learning_rate": 0.00016586817751329262,
      "loss": 3.5853,
      "step": 183080
    },
    {
      "epoch": 0.17066950541009235,
      "grad_norm": 1.048575758934021,
      "learning_rate": 0.00016586631319119742,
      "loss": 3.619,
      "step": 183090
    },
    {
      "epoch": 0.17067882702817144,
      "grad_norm": 0.8698957562446594,
      "learning_rate": 0.00016586444886910223,
      "loss": 3.5717,
      "step": 183100
    },
    {
      "epoch": 0.17068814864625054,
      "grad_norm": 0.9640294313430786,
      "learning_rate": 0.00016586258454700704,
      "loss": 3.4453,
      "step": 183110
    },
    {
      "epoch": 0.17069747026432963,
      "grad_norm": 1.2130577564239502,
      "learning_rate": 0.00016586072022491184,
      "loss": 3.5515,
      "step": 183120
    },
    {
      "epoch": 0.17070679188240873,
      "grad_norm": 1.1248884201049805,
      "learning_rate": 0.00016585885590281662,
      "loss": 3.5681,
      "step": 183130
    },
    {
      "epoch": 0.17071611350048782,
      "grad_norm": 1.0598883628845215,
      "learning_rate": 0.00016585699158072143,
      "loss": 3.5166,
      "step": 183140
    },
    {
      "epoch": 0.17072543511856692,
      "grad_norm": 1.0141595602035522,
      "learning_rate": 0.0001658551272586262,
      "loss": 3.5747,
      "step": 183150
    },
    {
      "epoch": 0.17073475673664598,
      "grad_norm": 1.0731314420700073,
      "learning_rate": 0.00016585326293653104,
      "loss": 3.6114,
      "step": 183160
    },
    {
      "epoch": 0.17074407835472508,
      "grad_norm": 1.0139645338058472,
      "learning_rate": 0.00016585139861443585,
      "loss": 3.5225,
      "step": 183170
    },
    {
      "epoch": 0.17075339997280417,
      "grad_norm": 1.0664161443710327,
      "learning_rate": 0.00016584953429234063,
      "loss": 3.4043,
      "step": 183180
    },
    {
      "epoch": 0.17076272159088327,
      "grad_norm": 1.0704472064971924,
      "learning_rate": 0.00016584766997024543,
      "loss": 3.703,
      "step": 183190
    },
    {
      "epoch": 0.17077204320896236,
      "grad_norm": 1.1619532108306885,
      "learning_rate": 0.0001658458056481502,
      "loss": 3.3032,
      "step": 183200
    },
    {
      "epoch": 0.17078136482704145,
      "grad_norm": 1.2081372737884521,
      "learning_rate": 0.00016584394132605502,
      "loss": 3.5772,
      "step": 183210
    },
    {
      "epoch": 0.17079068644512055,
      "grad_norm": 0.9857361912727356,
      "learning_rate": 0.00016584207700395982,
      "loss": 3.5817,
      "step": 183220
    },
    {
      "epoch": 0.17080000806319964,
      "grad_norm": 1.1612145900726318,
      "learning_rate": 0.00016584021268186463,
      "loss": 3.5807,
      "step": 183230
    },
    {
      "epoch": 0.17080932968127874,
      "grad_norm": 0.9207793474197388,
      "learning_rate": 0.00016583834835976944,
      "loss": 3.5498,
      "step": 183240
    },
    {
      "epoch": 0.17081865129935783,
      "grad_norm": 1.0030778646469116,
      "learning_rate": 0.00016583648403767422,
      "loss": 3.5253,
      "step": 183250
    },
    {
      "epoch": 0.17082797291743693,
      "grad_norm": 1.1067570447921753,
      "learning_rate": 0.00016583461971557902,
      "loss": 3.6608,
      "step": 183260
    },
    {
      "epoch": 0.170837294535516,
      "grad_norm": 0.9266327023506165,
      "learning_rate": 0.00016583275539348383,
      "loss": 3.7432,
      "step": 183270
    },
    {
      "epoch": 0.1708466161535951,
      "grad_norm": 0.9992384910583496,
      "learning_rate": 0.00016583089107138864,
      "loss": 3.4572,
      "step": 183280
    },
    {
      "epoch": 0.17085593777167418,
      "grad_norm": 1.0364155769348145,
      "learning_rate": 0.00016582902674929344,
      "loss": 3.4147,
      "step": 183290
    },
    {
      "epoch": 0.17086525938975328,
      "grad_norm": 1.1735048294067383,
      "learning_rate": 0.00016582716242719822,
      "loss": 3.44,
      "step": 183300
    },
    {
      "epoch": 0.17087458100783237,
      "grad_norm": 0.8458123803138733,
      "learning_rate": 0.00016582529810510303,
      "loss": 3.6249,
      "step": 183310
    },
    {
      "epoch": 0.17088390262591147,
      "grad_norm": 1.1035062074661255,
      "learning_rate": 0.00016582343378300783,
      "loss": 3.5848,
      "step": 183320
    },
    {
      "epoch": 0.17089322424399056,
      "grad_norm": 0.9288883805274963,
      "learning_rate": 0.00016582156946091264,
      "loss": 3.5291,
      "step": 183330
    },
    {
      "epoch": 0.17090254586206965,
      "grad_norm": 1.179470419883728,
      "learning_rate": 0.00016581970513881745,
      "loss": 3.6557,
      "step": 183340
    },
    {
      "epoch": 0.17091186748014875,
      "grad_norm": 0.9859893321990967,
      "learning_rate": 0.00016581784081672222,
      "loss": 3.4743,
      "step": 183350
    },
    {
      "epoch": 0.17092118909822784,
      "grad_norm": 1.1609522104263306,
      "learning_rate": 0.00016581597649462703,
      "loss": 3.5662,
      "step": 183360
    },
    {
      "epoch": 0.1709305107163069,
      "grad_norm": 0.9462592601776123,
      "learning_rate": 0.0001658141121725318,
      "loss": 3.4183,
      "step": 183370
    },
    {
      "epoch": 0.170939832334386,
      "grad_norm": 1.0516363382339478,
      "learning_rate": 0.00016581224785043664,
      "loss": 3.5846,
      "step": 183380
    },
    {
      "epoch": 0.1709491539524651,
      "grad_norm": 1.0372146368026733,
      "learning_rate": 0.00016581038352834145,
      "loss": 3.5442,
      "step": 183390
    },
    {
      "epoch": 0.1709584755705442,
      "grad_norm": 0.9492263197898865,
      "learning_rate": 0.00016580851920624623,
      "loss": 3.3616,
      "step": 183400
    },
    {
      "epoch": 0.1709677971886233,
      "grad_norm": 1.0776170492172241,
      "learning_rate": 0.00016580665488415104,
      "loss": 3.5645,
      "step": 183410
    },
    {
      "epoch": 0.17097711880670238,
      "grad_norm": 1.195189356803894,
      "learning_rate": 0.00016580479056205581,
      "loss": 3.6597,
      "step": 183420
    },
    {
      "epoch": 0.17098644042478148,
      "grad_norm": 1.043735384941101,
      "learning_rate": 0.00016580292623996062,
      "loss": 3.5293,
      "step": 183430
    },
    {
      "epoch": 0.17099576204286057,
      "grad_norm": 1.024548888206482,
      "learning_rate": 0.00016580106191786545,
      "loss": 3.5211,
      "step": 183440
    },
    {
      "epoch": 0.17100508366093967,
      "grad_norm": 0.9512718915939331,
      "learning_rate": 0.00016579919759577023,
      "loss": 3.5386,
      "step": 183450
    },
    {
      "epoch": 0.17101440527901876,
      "grad_norm": 1.0149176120758057,
      "learning_rate": 0.00016579733327367504,
      "loss": 3.667,
      "step": 183460
    },
    {
      "epoch": 0.17102372689709786,
      "grad_norm": 1.0013587474822998,
      "learning_rate": 0.00016579546895157982,
      "loss": 3.599,
      "step": 183470
    },
    {
      "epoch": 0.17103304851517692,
      "grad_norm": 1.0771268606185913,
      "learning_rate": 0.00016579360462948463,
      "loss": 3.6112,
      "step": 183480
    },
    {
      "epoch": 0.17104237013325602,
      "grad_norm": 1.0305589437484741,
      "learning_rate": 0.00016579174030738943,
      "loss": 3.4923,
      "step": 183490
    },
    {
      "epoch": 0.1710516917513351,
      "grad_norm": 1.0987815856933594,
      "learning_rate": 0.00016578987598529424,
      "loss": 3.6222,
      "step": 183500
    },
    {
      "epoch": 0.1710610133694142,
      "grad_norm": 0.9465287923812866,
      "learning_rate": 0.00016578801166319904,
      "loss": 3.4514,
      "step": 183510
    },
    {
      "epoch": 0.1710703349874933,
      "grad_norm": 1.0737594366073608,
      "learning_rate": 0.00016578614734110382,
      "loss": 3.4317,
      "step": 183520
    },
    {
      "epoch": 0.1710796566055724,
      "grad_norm": 0.8991250991821289,
      "learning_rate": 0.00016578428301900863,
      "loss": 3.4552,
      "step": 183530
    },
    {
      "epoch": 0.1710889782236515,
      "grad_norm": 1.0592695474624634,
      "learning_rate": 0.00016578241869691344,
      "loss": 3.4558,
      "step": 183540
    },
    {
      "epoch": 0.17109829984173058,
      "grad_norm": 1.3915493488311768,
      "learning_rate": 0.00016578055437481824,
      "loss": 3.6395,
      "step": 183550
    },
    {
      "epoch": 0.17110762145980968,
      "grad_norm": 1.1867859363555908,
      "learning_rate": 0.00016577869005272305,
      "loss": 3.3668,
      "step": 183560
    },
    {
      "epoch": 0.17111694307788877,
      "grad_norm": 0.9547731280326843,
      "learning_rate": 0.00016577682573062783,
      "loss": 3.556,
      "step": 183570
    },
    {
      "epoch": 0.17112626469596784,
      "grad_norm": 1.0082460641860962,
      "learning_rate": 0.00016577496140853263,
      "loss": 3.5908,
      "step": 183580
    },
    {
      "epoch": 0.17113558631404693,
      "grad_norm": 0.9408653378486633,
      "learning_rate": 0.00016577309708643744,
      "loss": 3.6897,
      "step": 183590
    },
    {
      "epoch": 0.17114490793212603,
      "grad_norm": 0.924405574798584,
      "learning_rate": 0.00016577123276434222,
      "loss": 3.5438,
      "step": 183600
    },
    {
      "epoch": 0.17115422955020512,
      "grad_norm": 0.9795662760734558,
      "learning_rate": 0.00016576936844224705,
      "loss": 3.3614,
      "step": 183610
    },
    {
      "epoch": 0.17116355116828422,
      "grad_norm": 1.076192021369934,
      "learning_rate": 0.00016576750412015183,
      "loss": 3.7851,
      "step": 183620
    },
    {
      "epoch": 0.1711728727863633,
      "grad_norm": 1.2040343284606934,
      "learning_rate": 0.00016576563979805664,
      "loss": 3.2579,
      "step": 183630
    },
    {
      "epoch": 0.1711821944044424,
      "grad_norm": 1.1535743474960327,
      "learning_rate": 0.00016576377547596145,
      "loss": 3.5554,
      "step": 183640
    },
    {
      "epoch": 0.1711915160225215,
      "grad_norm": 0.9829056859016418,
      "learning_rate": 0.00016576191115386622,
      "loss": 3.536,
      "step": 183650
    },
    {
      "epoch": 0.1712008376406006,
      "grad_norm": 1.0866459608078003,
      "learning_rate": 0.00016576004683177103,
      "loss": 3.4654,
      "step": 183660
    },
    {
      "epoch": 0.1712101592586797,
      "grad_norm": 0.952983558177948,
      "learning_rate": 0.00016575818250967584,
      "loss": 3.6753,
      "step": 183670
    },
    {
      "epoch": 0.17121948087675878,
      "grad_norm": 0.9485489130020142,
      "learning_rate": 0.00016575631818758064,
      "loss": 3.6515,
      "step": 183680
    },
    {
      "epoch": 0.17122880249483785,
      "grad_norm": 1.1694988012313843,
      "learning_rate": 0.00016575445386548545,
      "loss": 3.5623,
      "step": 183690
    },
    {
      "epoch": 0.17123812411291695,
      "grad_norm": 1.0895262956619263,
      "learning_rate": 0.00016575258954339023,
      "loss": 3.5645,
      "step": 183700
    },
    {
      "epoch": 0.17124744573099604,
      "grad_norm": 1.0662858486175537,
      "learning_rate": 0.00016575072522129504,
      "loss": 3.5919,
      "step": 183710
    },
    {
      "epoch": 0.17125676734907513,
      "grad_norm": 1.143950343132019,
      "learning_rate": 0.00016574886089919984,
      "loss": 3.5506,
      "step": 183720
    },
    {
      "epoch": 0.17126608896715423,
      "grad_norm": 1.2810724973678589,
      "learning_rate": 0.00016574699657710465,
      "loss": 3.5669,
      "step": 183730
    },
    {
      "epoch": 0.17127541058523332,
      "grad_norm": 1.110972285270691,
      "learning_rate": 0.00016574513225500945,
      "loss": 3.5778,
      "step": 183740
    },
    {
      "epoch": 0.17128473220331242,
      "grad_norm": 1.0271774530410767,
      "learning_rate": 0.00016574326793291423,
      "loss": 3.5151,
      "step": 183750
    },
    {
      "epoch": 0.1712940538213915,
      "grad_norm": 1.0896600484848022,
      "learning_rate": 0.00016574140361081904,
      "loss": 3.6724,
      "step": 183760
    },
    {
      "epoch": 0.1713033754394706,
      "grad_norm": 0.8977314233779907,
      "learning_rate": 0.00016573953928872385,
      "loss": 3.5588,
      "step": 183770
    },
    {
      "epoch": 0.1713126970575497,
      "grad_norm": 1.2501789331436157,
      "learning_rate": 0.00016573767496662865,
      "loss": 3.6481,
      "step": 183780
    },
    {
      "epoch": 0.17132201867562877,
      "grad_norm": 1.0606986284255981,
      "learning_rate": 0.00016573581064453346,
      "loss": 3.4328,
      "step": 183790
    },
    {
      "epoch": 0.17133134029370786,
      "grad_norm": 0.8601649403572083,
      "learning_rate": 0.00016573394632243824,
      "loss": 3.4781,
      "step": 183800
    },
    {
      "epoch": 0.17134066191178696,
      "grad_norm": 1.1734941005706787,
      "learning_rate": 0.00016573208200034304,
      "loss": 3.5972,
      "step": 183810
    },
    {
      "epoch": 0.17134998352986605,
      "grad_norm": 0.9998137354850769,
      "learning_rate": 0.00016573021767824782,
      "loss": 3.5966,
      "step": 183820
    },
    {
      "epoch": 0.17135930514794515,
      "grad_norm": 1.0731061697006226,
      "learning_rate": 0.00016572835335615266,
      "loss": 3.7005,
      "step": 183830
    },
    {
      "epoch": 0.17136862676602424,
      "grad_norm": 0.9912443161010742,
      "learning_rate": 0.00016572648903405746,
      "loss": 3.4744,
      "step": 183840
    },
    {
      "epoch": 0.17137794838410333,
      "grad_norm": 0.8432312607765198,
      "learning_rate": 0.00016572462471196224,
      "loss": 3.602,
      "step": 183850
    },
    {
      "epoch": 0.17138727000218243,
      "grad_norm": 1.1374919414520264,
      "learning_rate": 0.00016572276038986705,
      "loss": 3.4934,
      "step": 183860
    },
    {
      "epoch": 0.17139659162026152,
      "grad_norm": 1.0955005884170532,
      "learning_rate": 0.00016572089606777183,
      "loss": 3.6542,
      "step": 183870
    },
    {
      "epoch": 0.17140591323834062,
      "grad_norm": 1.3120393753051758,
      "learning_rate": 0.00016571903174567663,
      "loss": 3.5064,
      "step": 183880
    },
    {
      "epoch": 0.1714152348564197,
      "grad_norm": 1.2699657678604126,
      "learning_rate": 0.00016571716742358147,
      "loss": 3.4941,
      "step": 183890
    },
    {
      "epoch": 0.17142455647449878,
      "grad_norm": 1.2897863388061523,
      "learning_rate": 0.00016571530310148625,
      "loss": 3.7346,
      "step": 183900
    },
    {
      "epoch": 0.17143387809257787,
      "grad_norm": 1.08677077293396,
      "learning_rate": 0.00016571343877939105,
      "loss": 3.588,
      "step": 183910
    },
    {
      "epoch": 0.17144319971065697,
      "grad_norm": 1.061028242111206,
      "learning_rate": 0.00016571157445729583,
      "loss": 3.6154,
      "step": 183920
    },
    {
      "epoch": 0.17145252132873606,
      "grad_norm": 1.072371244430542,
      "learning_rate": 0.00016570971013520064,
      "loss": 3.5073,
      "step": 183930
    },
    {
      "epoch": 0.17146184294681516,
      "grad_norm": 0.7789543271064758,
      "learning_rate": 0.00016570784581310544,
      "loss": 3.5992,
      "step": 183940
    },
    {
      "epoch": 0.17147116456489425,
      "grad_norm": 1.1682417392730713,
      "learning_rate": 0.00016570598149101025,
      "loss": 3.4687,
      "step": 183950
    },
    {
      "epoch": 0.17148048618297335,
      "grad_norm": 1.004330039024353,
      "learning_rate": 0.00016570411716891506,
      "loss": 3.55,
      "step": 183960
    },
    {
      "epoch": 0.17148980780105244,
      "grad_norm": 0.9759143590927124,
      "learning_rate": 0.00016570225284681984,
      "loss": 3.2526,
      "step": 183970
    },
    {
      "epoch": 0.17149912941913154,
      "grad_norm": 1.1998547315597534,
      "learning_rate": 0.00016570038852472464,
      "loss": 3.5292,
      "step": 183980
    },
    {
      "epoch": 0.17150845103721063,
      "grad_norm": 1.0716884136199951,
      "learning_rate": 0.00016569852420262945,
      "loss": 3.6939,
      "step": 183990
    },
    {
      "epoch": 0.1715177726552897,
      "grad_norm": 0.9637787342071533,
      "learning_rate": 0.00016569665988053426,
      "loss": 3.455,
      "step": 184000
    },
    {
      "epoch": 0.1715270942733688,
      "grad_norm": 1.2540026903152466,
      "learning_rate": 0.00016569479555843906,
      "loss": 3.5493,
      "step": 184010
    },
    {
      "epoch": 0.17153641589144789,
      "grad_norm": 1.107737421989441,
      "learning_rate": 0.00016569293123634384,
      "loss": 3.4333,
      "step": 184020
    },
    {
      "epoch": 0.17154573750952698,
      "grad_norm": 1.0403554439544678,
      "learning_rate": 0.00016569106691424865,
      "loss": 3.4759,
      "step": 184030
    },
    {
      "epoch": 0.17155505912760607,
      "grad_norm": 1.102949857711792,
      "learning_rate": 0.00016568920259215345,
      "loss": 3.618,
      "step": 184040
    },
    {
      "epoch": 0.17156438074568517,
      "grad_norm": 1.0872527360916138,
      "learning_rate": 0.00016568733827005823,
      "loss": 3.5277,
      "step": 184050
    },
    {
      "epoch": 0.17157370236376426,
      "grad_norm": 0.8790892958641052,
      "learning_rate": 0.00016568547394796307,
      "loss": 3.4603,
      "step": 184060
    },
    {
      "epoch": 0.17158302398184336,
      "grad_norm": 1.3321728706359863,
      "learning_rate": 0.00016568360962586785,
      "loss": 3.4973,
      "step": 184070
    },
    {
      "epoch": 0.17159234559992245,
      "grad_norm": 0.9257457256317139,
      "learning_rate": 0.00016568174530377265,
      "loss": 3.5653,
      "step": 184080
    },
    {
      "epoch": 0.17160166721800155,
      "grad_norm": 1.0676817893981934,
      "learning_rate": 0.00016567988098167746,
      "loss": 3.4454,
      "step": 184090
    },
    {
      "epoch": 0.17161098883608064,
      "grad_norm": 1.061356782913208,
      "learning_rate": 0.00016567801665958224,
      "loss": 3.5886,
      "step": 184100
    },
    {
      "epoch": 0.1716203104541597,
      "grad_norm": 1.1552987098693848,
      "learning_rate": 0.00016567615233748704,
      "loss": 3.4612,
      "step": 184110
    },
    {
      "epoch": 0.1716296320722388,
      "grad_norm": 1.083443522453308,
      "learning_rate": 0.00016567428801539185,
      "loss": 3.5095,
      "step": 184120
    },
    {
      "epoch": 0.1716389536903179,
      "grad_norm": 1.0701709985733032,
      "learning_rate": 0.00016567242369329666,
      "loss": 3.5883,
      "step": 184130
    },
    {
      "epoch": 0.171648275308397,
      "grad_norm": 1.0105557441711426,
      "learning_rate": 0.00016567055937120146,
      "loss": 3.6454,
      "step": 184140
    },
    {
      "epoch": 0.17165759692647609,
      "grad_norm": 1.0502548217773438,
      "learning_rate": 0.00016566869504910624,
      "loss": 3.5592,
      "step": 184150
    },
    {
      "epoch": 0.17166691854455518,
      "grad_norm": 0.9193689823150635,
      "learning_rate": 0.00016566683072701105,
      "loss": 3.7308,
      "step": 184160
    },
    {
      "epoch": 0.17167624016263427,
      "grad_norm": 0.7957769632339478,
      "learning_rate": 0.00016566496640491585,
      "loss": 3.6539,
      "step": 184170
    },
    {
      "epoch": 0.17168556178071337,
      "grad_norm": 1.2102394104003906,
      "learning_rate": 0.00016566310208282066,
      "loss": 3.5775,
      "step": 184180
    },
    {
      "epoch": 0.17169488339879246,
      "grad_norm": 1.2659395933151245,
      "learning_rate": 0.00016566123776072547,
      "loss": 3.4905,
      "step": 184190
    },
    {
      "epoch": 0.17170420501687156,
      "grad_norm": 1.1563093662261963,
      "learning_rate": 0.00016565937343863025,
      "loss": 3.4181,
      "step": 184200
    },
    {
      "epoch": 0.17171352663495063,
      "grad_norm": 1.0740668773651123,
      "learning_rate": 0.00016565750911653505,
      "loss": 3.4743,
      "step": 184210
    },
    {
      "epoch": 0.17172284825302972,
      "grad_norm": 1.0532886981964111,
      "learning_rate": 0.00016565564479443986,
      "loss": 3.576,
      "step": 184220
    },
    {
      "epoch": 0.17173216987110881,
      "grad_norm": 1.067779541015625,
      "learning_rate": 0.00016565378047234467,
      "loss": 3.444,
      "step": 184230
    },
    {
      "epoch": 0.1717414914891879,
      "grad_norm": 1.044915795326233,
      "learning_rate": 0.00016565191615024947,
      "loss": 3.6935,
      "step": 184240
    },
    {
      "epoch": 0.171750813107267,
      "grad_norm": 1.199549674987793,
      "learning_rate": 0.00016565005182815425,
      "loss": 3.5278,
      "step": 184250
    },
    {
      "epoch": 0.1717601347253461,
      "grad_norm": 1.0606663227081299,
      "learning_rate": 0.00016564818750605906,
      "loss": 3.5494,
      "step": 184260
    },
    {
      "epoch": 0.1717694563434252,
      "grad_norm": 1.0058175325393677,
      "learning_rate": 0.00016564632318396384,
      "loss": 3.5568,
      "step": 184270
    },
    {
      "epoch": 0.1717787779615043,
      "grad_norm": 1.160150408744812,
      "learning_rate": 0.00016564445886186867,
      "loss": 3.4684,
      "step": 184280
    },
    {
      "epoch": 0.17178809957958338,
      "grad_norm": 1.0930752754211426,
      "learning_rate": 0.00016564259453977348,
      "loss": 3.5991,
      "step": 184290
    },
    {
      "epoch": 0.17179742119766248,
      "grad_norm": 0.8541244268417358,
      "learning_rate": 0.00016564073021767826,
      "loss": 3.5322,
      "step": 184300
    },
    {
      "epoch": 0.17180674281574157,
      "grad_norm": 0.7955280542373657,
      "learning_rate": 0.00016563886589558306,
      "loss": 3.5979,
      "step": 184310
    },
    {
      "epoch": 0.17181606443382064,
      "grad_norm": 1.09780752658844,
      "learning_rate": 0.00016563700157348784,
      "loss": 3.501,
      "step": 184320
    },
    {
      "epoch": 0.17182538605189973,
      "grad_norm": 1.1066867113113403,
      "learning_rate": 0.00016563513725139265,
      "loss": 3.4934,
      "step": 184330
    },
    {
      "epoch": 0.17183470766997883,
      "grad_norm": 1.0513771772384644,
      "learning_rate": 0.00016563327292929748,
      "loss": 3.4976,
      "step": 184340
    },
    {
      "epoch": 0.17184402928805792,
      "grad_norm": 1.2564456462860107,
      "learning_rate": 0.00016563140860720226,
      "loss": 3.5818,
      "step": 184350
    },
    {
      "epoch": 0.17185335090613701,
      "grad_norm": 1.103405237197876,
      "learning_rate": 0.00016562954428510707,
      "loss": 3.3463,
      "step": 184360
    },
    {
      "epoch": 0.1718626725242161,
      "grad_norm": 1.0544216632843018,
      "learning_rate": 0.00016562767996301185,
      "loss": 3.6509,
      "step": 184370
    },
    {
      "epoch": 0.1718719941422952,
      "grad_norm": 1.0222220420837402,
      "learning_rate": 0.00016562581564091665,
      "loss": 3.4271,
      "step": 184380
    },
    {
      "epoch": 0.1718813157603743,
      "grad_norm": 1.0010725259780884,
      "learning_rate": 0.00016562395131882146,
      "loss": 3.5368,
      "step": 184390
    },
    {
      "epoch": 0.1718906373784534,
      "grad_norm": 1.090559959411621,
      "learning_rate": 0.00016562208699672626,
      "loss": 3.5888,
      "step": 184400
    },
    {
      "epoch": 0.1718999589965325,
      "grad_norm": 1.219915509223938,
      "learning_rate": 0.00016562022267463107,
      "loss": 3.4216,
      "step": 184410
    },
    {
      "epoch": 0.17190928061461155,
      "grad_norm": 1.156524658203125,
      "learning_rate": 0.00016561835835253585,
      "loss": 3.5461,
      "step": 184420
    },
    {
      "epoch": 0.17191860223269065,
      "grad_norm": 1.1903578042984009,
      "learning_rate": 0.00016561649403044066,
      "loss": 3.5297,
      "step": 184430
    },
    {
      "epoch": 0.17192792385076974,
      "grad_norm": 1.0033223628997803,
      "learning_rate": 0.00016561462970834546,
      "loss": 3.6643,
      "step": 184440
    },
    {
      "epoch": 0.17193724546884884,
      "grad_norm": 0.9939575791358948,
      "learning_rate": 0.00016561276538625027,
      "loss": 3.4509,
      "step": 184450
    },
    {
      "epoch": 0.17194656708692793,
      "grad_norm": 1.0998588800430298,
      "learning_rate": 0.00016561090106415507,
      "loss": 3.4825,
      "step": 184460
    },
    {
      "epoch": 0.17195588870500703,
      "grad_norm": 1.2766144275665283,
      "learning_rate": 0.00016560903674205985,
      "loss": 3.2641,
      "step": 184470
    },
    {
      "epoch": 0.17196521032308612,
      "grad_norm": 0.9048798680305481,
      "learning_rate": 0.00016560717241996466,
      "loss": 3.4325,
      "step": 184480
    },
    {
      "epoch": 0.17197453194116522,
      "grad_norm": 1.0373936891555786,
      "learning_rate": 0.00016560530809786947,
      "loss": 3.5771,
      "step": 184490
    },
    {
      "epoch": 0.1719838535592443,
      "grad_norm": 1.1274113655090332,
      "learning_rate": 0.00016560344377577425,
      "loss": 3.4934,
      "step": 184500
    },
    {
      "epoch": 0.1719931751773234,
      "grad_norm": 0.9481971263885498,
      "learning_rate": 0.00016560157945367908,
      "loss": 3.5941,
      "step": 184510
    },
    {
      "epoch": 0.1720024967954025,
      "grad_norm": 1.1477073431015015,
      "learning_rate": 0.00016559971513158386,
      "loss": 3.5158,
      "step": 184520
    },
    {
      "epoch": 0.17201181841348157,
      "grad_norm": 1.2622556686401367,
      "learning_rate": 0.00016559785080948866,
      "loss": 3.6406,
      "step": 184530
    },
    {
      "epoch": 0.17202114003156066,
      "grad_norm": 0.8761709928512573,
      "learning_rate": 0.00016559598648739347,
      "loss": 3.5293,
      "step": 184540
    },
    {
      "epoch": 0.17203046164963975,
      "grad_norm": 1.2464137077331543,
      "learning_rate": 0.00016559412216529825,
      "loss": 3.4893,
      "step": 184550
    },
    {
      "epoch": 0.17203978326771885,
      "grad_norm": 0.9764725565910339,
      "learning_rate": 0.00016559225784320306,
      "loss": 3.4593,
      "step": 184560
    },
    {
      "epoch": 0.17204910488579794,
      "grad_norm": 1.278493881225586,
      "learning_rate": 0.00016559039352110786,
      "loss": 3.616,
      "step": 184570
    },
    {
      "epoch": 0.17205842650387704,
      "grad_norm": 0.9455174207687378,
      "learning_rate": 0.00016558852919901267,
      "loss": 3.4807,
      "step": 184580
    },
    {
      "epoch": 0.17206774812195613,
      "grad_norm": 1.159058690071106,
      "learning_rate": 0.00016558666487691745,
      "loss": 3.5628,
      "step": 184590
    },
    {
      "epoch": 0.17207706974003523,
      "grad_norm": 1.0969306230545044,
      "learning_rate": 0.00016558480055482225,
      "loss": 3.518,
      "step": 184600
    },
    {
      "epoch": 0.17208639135811432,
      "grad_norm": 1.0398081541061401,
      "learning_rate": 0.00016558293623272706,
      "loss": 3.3692,
      "step": 184610
    },
    {
      "epoch": 0.17209571297619342,
      "grad_norm": 1.1663293838500977,
      "learning_rate": 0.00016558107191063187,
      "loss": 3.559,
      "step": 184620
    },
    {
      "epoch": 0.17210503459427248,
      "grad_norm": 0.9169872999191284,
      "learning_rate": 0.00016557920758853667,
      "loss": 3.604,
      "step": 184630
    },
    {
      "epoch": 0.17211435621235158,
      "grad_norm": 1.164994716644287,
      "learning_rate": 0.00016557734326644145,
      "loss": 3.6168,
      "step": 184640
    },
    {
      "epoch": 0.17212367783043067,
      "grad_norm": 0.9042360186576843,
      "learning_rate": 0.00016557547894434626,
      "loss": 3.5874,
      "step": 184650
    },
    {
      "epoch": 0.17213299944850977,
      "grad_norm": 1.0664558410644531,
      "learning_rate": 0.00016557361462225107,
      "loss": 3.6755,
      "step": 184660
    },
    {
      "epoch": 0.17214232106658886,
      "grad_norm": 1.030246376991272,
      "learning_rate": 0.00016557175030015587,
      "loss": 3.4837,
      "step": 184670
    },
    {
      "epoch": 0.17215164268466795,
      "grad_norm": 1.210889220237732,
      "learning_rate": 0.00016556988597806068,
      "loss": 3.5653,
      "step": 184680
    },
    {
      "epoch": 0.17216096430274705,
      "grad_norm": 0.9986250400543213,
      "learning_rate": 0.00016556802165596546,
      "loss": 3.5478,
      "step": 184690
    },
    {
      "epoch": 0.17217028592082614,
      "grad_norm": 1.178429365158081,
      "learning_rate": 0.00016556615733387026,
      "loss": 3.5276,
      "step": 184700
    },
    {
      "epoch": 0.17217960753890524,
      "grad_norm": 1.1473127603530884,
      "learning_rate": 0.00016556429301177507,
      "loss": 3.431,
      "step": 184710
    },
    {
      "epoch": 0.17218892915698433,
      "grad_norm": 1.2353408336639404,
      "learning_rate": 0.00016556242868967985,
      "loss": 3.5859,
      "step": 184720
    },
    {
      "epoch": 0.1721982507750634,
      "grad_norm": 1.0258448123931885,
      "learning_rate": 0.00016556056436758468,
      "loss": 3.6328,
      "step": 184730
    },
    {
      "epoch": 0.1722075723931425,
      "grad_norm": 0.930115818977356,
      "learning_rate": 0.00016555870004548946,
      "loss": 3.4177,
      "step": 184740
    },
    {
      "epoch": 0.1722168940112216,
      "grad_norm": 1.1385738849639893,
      "learning_rate": 0.00016555683572339427,
      "loss": 3.6663,
      "step": 184750
    },
    {
      "epoch": 0.17222621562930068,
      "grad_norm": 0.9769076108932495,
      "learning_rate": 0.00016555497140129907,
      "loss": 3.6106,
      "step": 184760
    },
    {
      "epoch": 0.17223553724737978,
      "grad_norm": 1.2142695188522339,
      "learning_rate": 0.00016555310707920385,
      "loss": 3.5468,
      "step": 184770
    },
    {
      "epoch": 0.17224485886545887,
      "grad_norm": 1.0578162670135498,
      "learning_rate": 0.00016555124275710866,
      "loss": 3.6523,
      "step": 184780
    },
    {
      "epoch": 0.17225418048353797,
      "grad_norm": 1.3069381713867188,
      "learning_rate": 0.00016554937843501347,
      "loss": 3.466,
      "step": 184790
    },
    {
      "epoch": 0.17226350210161706,
      "grad_norm": 1.1493515968322754,
      "learning_rate": 0.00016554751411291827,
      "loss": 3.5395,
      "step": 184800
    },
    {
      "epoch": 0.17227282371969616,
      "grad_norm": 0.9488618969917297,
      "learning_rate": 0.00016554564979082308,
      "loss": 3.5517,
      "step": 184810
    },
    {
      "epoch": 0.17228214533777525,
      "grad_norm": 1.2977246046066284,
      "learning_rate": 0.00016554378546872786,
      "loss": 3.4019,
      "step": 184820
    },
    {
      "epoch": 0.17229146695585434,
      "grad_norm": 1.1058259010314941,
      "learning_rate": 0.00016554192114663266,
      "loss": 3.4649,
      "step": 184830
    },
    {
      "epoch": 0.1723007885739334,
      "grad_norm": 1.2857348918914795,
      "learning_rate": 0.00016554005682453747,
      "loss": 3.4266,
      "step": 184840
    },
    {
      "epoch": 0.1723101101920125,
      "grad_norm": 0.9922835230827332,
      "learning_rate": 0.00016553819250244228,
      "loss": 3.5873,
      "step": 184850
    },
    {
      "epoch": 0.1723194318100916,
      "grad_norm": 0.9662395119667053,
      "learning_rate": 0.00016553632818034708,
      "loss": 3.7226,
      "step": 184860
    },
    {
      "epoch": 0.1723287534281707,
      "grad_norm": 1.1917177438735962,
      "learning_rate": 0.00016553446385825186,
      "loss": 3.5508,
      "step": 184870
    },
    {
      "epoch": 0.1723380750462498,
      "grad_norm": 1.088181495666504,
      "learning_rate": 0.00016553259953615667,
      "loss": 3.5913,
      "step": 184880
    },
    {
      "epoch": 0.17234739666432888,
      "grad_norm": 1.1016650199890137,
      "learning_rate": 0.00016553073521406145,
      "loss": 3.407,
      "step": 184890
    },
    {
      "epoch": 0.17235671828240798,
      "grad_norm": 0.9291702508926392,
      "learning_rate": 0.00016552887089196628,
      "loss": 3.5006,
      "step": 184900
    },
    {
      "epoch": 0.17236603990048707,
      "grad_norm": 1.1037788391113281,
      "learning_rate": 0.0001655270065698711,
      "loss": 3.2717,
      "step": 184910
    },
    {
      "epoch": 0.17237536151856617,
      "grad_norm": 1.1110106706619263,
      "learning_rate": 0.00016552514224777587,
      "loss": 3.7323,
      "step": 184920
    },
    {
      "epoch": 0.17238468313664526,
      "grad_norm": 1.0285996198654175,
      "learning_rate": 0.00016552327792568067,
      "loss": 3.4772,
      "step": 184930
    },
    {
      "epoch": 0.17239400475472433,
      "grad_norm": 1.06562340259552,
      "learning_rate": 0.00016552141360358545,
      "loss": 3.587,
      "step": 184940
    },
    {
      "epoch": 0.17240332637280342,
      "grad_norm": 1.1272412538528442,
      "learning_rate": 0.00016551954928149026,
      "loss": 3.4432,
      "step": 184950
    },
    {
      "epoch": 0.17241264799088252,
      "grad_norm": 1.1100847721099854,
      "learning_rate": 0.0001655176849593951,
      "loss": 3.4118,
      "step": 184960
    },
    {
      "epoch": 0.1724219696089616,
      "grad_norm": 0.9722442626953125,
      "learning_rate": 0.00016551582063729987,
      "loss": 3.446,
      "step": 184970
    },
    {
      "epoch": 0.1724312912270407,
      "grad_norm": 1.106167197227478,
      "learning_rate": 0.00016551395631520468,
      "loss": 3.5346,
      "step": 184980
    },
    {
      "epoch": 0.1724406128451198,
      "grad_norm": 0.9724297523498535,
      "learning_rate": 0.00016551209199310946,
      "loss": 3.2492,
      "step": 184990
    },
    {
      "epoch": 0.1724499344631989,
      "grad_norm": 1.0662271976470947,
      "learning_rate": 0.00016551022767101426,
      "loss": 3.6039,
      "step": 185000
    },
    {
      "epoch": 0.172459256081278,
      "grad_norm": 1.6619737148284912,
      "learning_rate": 0.00016550836334891907,
      "loss": 3.59,
      "step": 185010
    },
    {
      "epoch": 0.17246857769935708,
      "grad_norm": 0.9976863265037537,
      "learning_rate": 0.00016550649902682388,
      "loss": 3.5096,
      "step": 185020
    },
    {
      "epoch": 0.17247789931743618,
      "grad_norm": 0.739841878414154,
      "learning_rate": 0.00016550463470472868,
      "loss": 3.5301,
      "step": 185030
    },
    {
      "epoch": 0.17248722093551527,
      "grad_norm": 0.9821991920471191,
      "learning_rate": 0.00016550277038263346,
      "loss": 3.6098,
      "step": 185040
    },
    {
      "epoch": 0.17249654255359434,
      "grad_norm": 1.1904706954956055,
      "learning_rate": 0.00016550090606053827,
      "loss": 3.4382,
      "step": 185050
    },
    {
      "epoch": 0.17250586417167343,
      "grad_norm": 1.1764808893203735,
      "learning_rate": 0.00016549904173844307,
      "loss": 3.6451,
      "step": 185060
    },
    {
      "epoch": 0.17251518578975253,
      "grad_norm": 0.946319580078125,
      "learning_rate": 0.00016549717741634788,
      "loss": 3.4063,
      "step": 185070
    },
    {
      "epoch": 0.17252450740783162,
      "grad_norm": 1.0882121324539185,
      "learning_rate": 0.00016549531309425269,
      "loss": 3.6287,
      "step": 185080
    },
    {
      "epoch": 0.17253382902591072,
      "grad_norm": 0.9911219477653503,
      "learning_rate": 0.00016549344877215747,
      "loss": 3.6197,
      "step": 185090
    },
    {
      "epoch": 0.1725431506439898,
      "grad_norm": 1.1047768592834473,
      "learning_rate": 0.00016549158445006227,
      "loss": 3.3957,
      "step": 185100
    },
    {
      "epoch": 0.1725524722620689,
      "grad_norm": 1.0144739151000977,
      "learning_rate": 0.00016548972012796708,
      "loss": 3.5874,
      "step": 185110
    },
    {
      "epoch": 0.172561793880148,
      "grad_norm": 1.1147834062576294,
      "learning_rate": 0.00016548785580587188,
      "loss": 3.6095,
      "step": 185120
    },
    {
      "epoch": 0.1725711154982271,
      "grad_norm": 1.099489688873291,
      "learning_rate": 0.0001654859914837767,
      "loss": 3.7168,
      "step": 185130
    },
    {
      "epoch": 0.1725804371163062,
      "grad_norm": 1.1097255945205688,
      "learning_rate": 0.00016548412716168147,
      "loss": 3.64,
      "step": 185140
    },
    {
      "epoch": 0.17258975873438526,
      "grad_norm": 1.1150755882263184,
      "learning_rate": 0.00016548226283958628,
      "loss": 3.5843,
      "step": 185150
    },
    {
      "epoch": 0.17259908035246435,
      "grad_norm": 1.0083450078964233,
      "learning_rate": 0.00016548039851749108,
      "loss": 3.6885,
      "step": 185160
    },
    {
      "epoch": 0.17260840197054345,
      "grad_norm": 1.1568670272827148,
      "learning_rate": 0.00016547853419539586,
      "loss": 3.5573,
      "step": 185170
    },
    {
      "epoch": 0.17261772358862254,
      "grad_norm": 1.2601629495620728,
      "learning_rate": 0.0001654766698733007,
      "loss": 3.231,
      "step": 185180
    },
    {
      "epoch": 0.17262704520670163,
      "grad_norm": 1.0483615398406982,
      "learning_rate": 0.00016547480555120547,
      "loss": 3.6087,
      "step": 185190
    },
    {
      "epoch": 0.17263636682478073,
      "grad_norm": 1.19648277759552,
      "learning_rate": 0.00016547294122911028,
      "loss": 3.477,
      "step": 185200
    },
    {
      "epoch": 0.17264568844285982,
      "grad_norm": 1.0462243556976318,
      "learning_rate": 0.0001654710769070151,
      "loss": 3.6629,
      "step": 185210
    },
    {
      "epoch": 0.17265501006093892,
      "grad_norm": 1.3784327507019043,
      "learning_rate": 0.00016546921258491987,
      "loss": 3.357,
      "step": 185220
    },
    {
      "epoch": 0.172664331679018,
      "grad_norm": 1.0480698347091675,
      "learning_rate": 0.00016546734826282467,
      "loss": 3.5856,
      "step": 185230
    },
    {
      "epoch": 0.1726736532970971,
      "grad_norm": 1.0625687837600708,
      "learning_rate": 0.00016546548394072948,
      "loss": 3.6042,
      "step": 185240
    },
    {
      "epoch": 0.1726829749151762,
      "grad_norm": 1.0962954759597778,
      "learning_rate": 0.00016546361961863429,
      "loss": 3.4541,
      "step": 185250
    },
    {
      "epoch": 0.17269229653325527,
      "grad_norm": 1.2076421976089478,
      "learning_rate": 0.0001654617552965391,
      "loss": 3.5083,
      "step": 185260
    },
    {
      "epoch": 0.17270161815133436,
      "grad_norm": 1.092888355255127,
      "learning_rate": 0.00016545989097444387,
      "loss": 3.7041,
      "step": 185270
    },
    {
      "epoch": 0.17271093976941346,
      "grad_norm": 1.1010254621505737,
      "learning_rate": 0.00016545802665234868,
      "loss": 3.5982,
      "step": 185280
    },
    {
      "epoch": 0.17272026138749255,
      "grad_norm": 1.1485793590545654,
      "learning_rate": 0.00016545616233025348,
      "loss": 3.7143,
      "step": 185290
    },
    {
      "epoch": 0.17272958300557165,
      "grad_norm": 0.9904172420501709,
      "learning_rate": 0.0001654542980081583,
      "loss": 3.7263,
      "step": 185300
    },
    {
      "epoch": 0.17273890462365074,
      "grad_norm": 0.886090874671936,
      "learning_rate": 0.0001654524336860631,
      "loss": 3.4813,
      "step": 185310
    },
    {
      "epoch": 0.17274822624172984,
      "grad_norm": 0.8199695348739624,
      "learning_rate": 0.00016545056936396788,
      "loss": 3.7277,
      "step": 185320
    },
    {
      "epoch": 0.17275754785980893,
      "grad_norm": 1.1456665992736816,
      "learning_rate": 0.00016544870504187268,
      "loss": 3.3356,
      "step": 185330
    },
    {
      "epoch": 0.17276686947788802,
      "grad_norm": 1.1698001623153687,
      "learning_rate": 0.00016544684071977746,
      "loss": 3.4687,
      "step": 185340
    },
    {
      "epoch": 0.17277619109596712,
      "grad_norm": 0.8554566502571106,
      "learning_rate": 0.0001654449763976823,
      "loss": 3.6606,
      "step": 185350
    },
    {
      "epoch": 0.17278551271404619,
      "grad_norm": 1.082099199295044,
      "learning_rate": 0.0001654431120755871,
      "loss": 3.5966,
      "step": 185360
    },
    {
      "epoch": 0.17279483433212528,
      "grad_norm": 1.1066231727600098,
      "learning_rate": 0.00016544124775349188,
      "loss": 3.642,
      "step": 185370
    },
    {
      "epoch": 0.17280415595020437,
      "grad_norm": 1.1202970743179321,
      "learning_rate": 0.00016543938343139669,
      "loss": 3.4965,
      "step": 185380
    },
    {
      "epoch": 0.17281347756828347,
      "grad_norm": 0.8925954103469849,
      "learning_rate": 0.00016543751910930147,
      "loss": 3.6057,
      "step": 185390
    },
    {
      "epoch": 0.17282279918636256,
      "grad_norm": 0.9816046357154846,
      "learning_rate": 0.00016543565478720627,
      "loss": 3.4059,
      "step": 185400
    },
    {
      "epoch": 0.17283212080444166,
      "grad_norm": 0.9290957450866699,
      "learning_rate": 0.0001654337904651111,
      "loss": 3.5584,
      "step": 185410
    },
    {
      "epoch": 0.17284144242252075,
      "grad_norm": 0.9182539582252502,
      "learning_rate": 0.00016543192614301588,
      "loss": 3.4797,
      "step": 185420
    },
    {
      "epoch": 0.17285076404059985,
      "grad_norm": 1.046804666519165,
      "learning_rate": 0.0001654300618209207,
      "loss": 3.4403,
      "step": 185430
    },
    {
      "epoch": 0.17286008565867894,
      "grad_norm": 1.0009162425994873,
      "learning_rate": 0.00016542819749882547,
      "loss": 3.6411,
      "step": 185440
    },
    {
      "epoch": 0.17286940727675804,
      "grad_norm": 0.9947389364242554,
      "learning_rate": 0.00016542633317673028,
      "loss": 3.5993,
      "step": 185450
    },
    {
      "epoch": 0.17287872889483713,
      "grad_norm": 1.1070077419281006,
      "learning_rate": 0.00016542446885463508,
      "loss": 3.3928,
      "step": 185460
    },
    {
      "epoch": 0.1728880505129162,
      "grad_norm": 1.0920336246490479,
      "learning_rate": 0.0001654226045325399,
      "loss": 3.5139,
      "step": 185470
    },
    {
      "epoch": 0.1728973721309953,
      "grad_norm": 0.9512216448783875,
      "learning_rate": 0.0001654207402104447,
      "loss": 3.4461,
      "step": 185480
    },
    {
      "epoch": 0.17290669374907439,
      "grad_norm": 1.0156538486480713,
      "learning_rate": 0.00016541887588834947,
      "loss": 3.5084,
      "step": 185490
    },
    {
      "epoch": 0.17291601536715348,
      "grad_norm": 1.2265814542770386,
      "learning_rate": 0.00016541701156625428,
      "loss": 3.5983,
      "step": 185500
    },
    {
      "epoch": 0.17292533698523258,
      "grad_norm": 1.1534826755523682,
      "learning_rate": 0.0001654151472441591,
      "loss": 3.5168,
      "step": 185510
    },
    {
      "epoch": 0.17293465860331167,
      "grad_norm": 0.847348153591156,
      "learning_rate": 0.0001654132829220639,
      "loss": 3.447,
      "step": 185520
    },
    {
      "epoch": 0.17294398022139076,
      "grad_norm": 1.0075054168701172,
      "learning_rate": 0.0001654114185999687,
      "loss": 3.6433,
      "step": 185530
    },
    {
      "epoch": 0.17295330183946986,
      "grad_norm": 0.9565919041633606,
      "learning_rate": 0.00016540955427787348,
      "loss": 3.635,
      "step": 185540
    },
    {
      "epoch": 0.17296262345754895,
      "grad_norm": 1.0326184034347534,
      "learning_rate": 0.00016540768995577828,
      "loss": 3.5533,
      "step": 185550
    },
    {
      "epoch": 0.17297194507562805,
      "grad_norm": 1.1284306049346924,
      "learning_rate": 0.0001654058256336831,
      "loss": 3.6827,
      "step": 185560
    },
    {
      "epoch": 0.17298126669370711,
      "grad_norm": 1.2196296453475952,
      "learning_rate": 0.0001654039613115879,
      "loss": 3.5821,
      "step": 185570
    },
    {
      "epoch": 0.1729905883117862,
      "grad_norm": 1.053649663925171,
      "learning_rate": 0.0001654020969894927,
      "loss": 3.292,
      "step": 185580
    },
    {
      "epoch": 0.1729999099298653,
      "grad_norm": 1.1301218271255493,
      "learning_rate": 0.00016540023266739748,
      "loss": 3.3284,
      "step": 185590
    },
    {
      "epoch": 0.1730092315479444,
      "grad_norm": 1.0146042108535767,
      "learning_rate": 0.0001653983683453023,
      "loss": 3.4349,
      "step": 185600
    },
    {
      "epoch": 0.1730185531660235,
      "grad_norm": 1.1256442070007324,
      "learning_rate": 0.0001653965040232071,
      "loss": 3.7303,
      "step": 185610
    },
    {
      "epoch": 0.1730278747841026,
      "grad_norm": 0.981357216835022,
      "learning_rate": 0.00016539463970111187,
      "loss": 3.486,
      "step": 185620
    },
    {
      "epoch": 0.17303719640218168,
      "grad_norm": 1.164154291152954,
      "learning_rate": 0.0001653927753790167,
      "loss": 3.5462,
      "step": 185630
    },
    {
      "epoch": 0.17304651802026078,
      "grad_norm": 0.9457765221595764,
      "learning_rate": 0.0001653909110569215,
      "loss": 3.5789,
      "step": 185640
    },
    {
      "epoch": 0.17305583963833987,
      "grad_norm": 1.0704594850540161,
      "learning_rate": 0.0001653890467348263,
      "loss": 3.5493,
      "step": 185650
    },
    {
      "epoch": 0.17306516125641896,
      "grad_norm": 1.0038719177246094,
      "learning_rate": 0.0001653871824127311,
      "loss": 3.5148,
      "step": 185660
    },
    {
      "epoch": 0.17307448287449806,
      "grad_norm": 0.9349478483200073,
      "learning_rate": 0.00016538531809063588,
      "loss": 3.4111,
      "step": 185670
    },
    {
      "epoch": 0.17308380449257713,
      "grad_norm": 0.9498841166496277,
      "learning_rate": 0.00016538345376854069,
      "loss": 3.5293,
      "step": 185680
    },
    {
      "epoch": 0.17309312611065622,
      "grad_norm": 1.0262712240219116,
      "learning_rate": 0.0001653815894464455,
      "loss": 3.7059,
      "step": 185690
    },
    {
      "epoch": 0.17310244772873531,
      "grad_norm": 0.9745901823043823,
      "learning_rate": 0.0001653797251243503,
      "loss": 3.5434,
      "step": 185700
    },
    {
      "epoch": 0.1731117693468144,
      "grad_norm": 1.0249992609024048,
      "learning_rate": 0.0001653778608022551,
      "loss": 3.4922,
      "step": 185710
    },
    {
      "epoch": 0.1731210909648935,
      "grad_norm": 0.8606586456298828,
      "learning_rate": 0.00016537599648015988,
      "loss": 3.5066,
      "step": 185720
    },
    {
      "epoch": 0.1731304125829726,
      "grad_norm": 1.0308418273925781,
      "learning_rate": 0.0001653741321580647,
      "loss": 3.6115,
      "step": 185730
    },
    {
      "epoch": 0.1731397342010517,
      "grad_norm": 0.9127419590950012,
      "learning_rate": 0.0001653722678359695,
      "loss": 3.6902,
      "step": 185740
    },
    {
      "epoch": 0.1731490558191308,
      "grad_norm": 1.076372742652893,
      "learning_rate": 0.0001653704035138743,
      "loss": 3.412,
      "step": 185750
    },
    {
      "epoch": 0.17315837743720988,
      "grad_norm": 1.1194581985473633,
      "learning_rate": 0.0001653685391917791,
      "loss": 3.4842,
      "step": 185760
    },
    {
      "epoch": 0.17316769905528898,
      "grad_norm": 1.0671100616455078,
      "learning_rate": 0.0001653666748696839,
      "loss": 3.614,
      "step": 185770
    },
    {
      "epoch": 0.17317702067336804,
      "grad_norm": 1.1581472158432007,
      "learning_rate": 0.0001653648105475887,
      "loss": 3.5578,
      "step": 185780
    },
    {
      "epoch": 0.17318634229144714,
      "grad_norm": 1.0232627391815186,
      "learning_rate": 0.00016536294622549347,
      "loss": 3.4708,
      "step": 185790
    },
    {
      "epoch": 0.17319566390952623,
      "grad_norm": 0.9815053939819336,
      "learning_rate": 0.0001653610819033983,
      "loss": 3.4302,
      "step": 185800
    },
    {
      "epoch": 0.17320498552760533,
      "grad_norm": 1.0355039834976196,
      "learning_rate": 0.0001653592175813031,
      "loss": 3.6071,
      "step": 185810
    },
    {
      "epoch": 0.17321430714568442,
      "grad_norm": 1.2598360776901245,
      "learning_rate": 0.0001653573532592079,
      "loss": 3.562,
      "step": 185820
    },
    {
      "epoch": 0.17322362876376352,
      "grad_norm": 1.1695724725723267,
      "learning_rate": 0.0001653554889371127,
      "loss": 3.6091,
      "step": 185830
    },
    {
      "epoch": 0.1732329503818426,
      "grad_norm": 1.083747386932373,
      "learning_rate": 0.00016535362461501748,
      "loss": 3.5788,
      "step": 185840
    },
    {
      "epoch": 0.1732422719999217,
      "grad_norm": 1.1762120723724365,
      "learning_rate": 0.00016535176029292228,
      "loss": 3.5984,
      "step": 185850
    },
    {
      "epoch": 0.1732515936180008,
      "grad_norm": 1.065969705581665,
      "learning_rate": 0.00016534989597082712,
      "loss": 3.629,
      "step": 185860
    },
    {
      "epoch": 0.1732609152360799,
      "grad_norm": 1.2713799476623535,
      "learning_rate": 0.0001653480316487319,
      "loss": 3.6496,
      "step": 185870
    },
    {
      "epoch": 0.173270236854159,
      "grad_norm": 0.9908674359321594,
      "learning_rate": 0.0001653461673266367,
      "loss": 3.4825,
      "step": 185880
    },
    {
      "epoch": 0.17327955847223805,
      "grad_norm": 0.9797390699386597,
      "learning_rate": 0.00016534430300454148,
      "loss": 3.5183,
      "step": 185890
    },
    {
      "epoch": 0.17328888009031715,
      "grad_norm": 1.1353564262390137,
      "learning_rate": 0.0001653424386824463,
      "loss": 3.6677,
      "step": 185900
    },
    {
      "epoch": 0.17329820170839624,
      "grad_norm": 1.1770589351654053,
      "learning_rate": 0.0001653405743603511,
      "loss": 3.484,
      "step": 185910
    },
    {
      "epoch": 0.17330752332647534,
      "grad_norm": 1.0799434185028076,
      "learning_rate": 0.0001653387100382559,
      "loss": 3.7071,
      "step": 185920
    },
    {
      "epoch": 0.17331684494455443,
      "grad_norm": 1.2138168811798096,
      "learning_rate": 0.0001653368457161607,
      "loss": 3.5245,
      "step": 185930
    },
    {
      "epoch": 0.17332616656263353,
      "grad_norm": 1.2389649152755737,
      "learning_rate": 0.0001653349813940655,
      "loss": 3.5446,
      "step": 185940
    },
    {
      "epoch": 0.17333548818071262,
      "grad_norm": 1.013596773147583,
      "learning_rate": 0.0001653331170719703,
      "loss": 3.4644,
      "step": 185950
    },
    {
      "epoch": 0.17334480979879172,
      "grad_norm": 1.0478781461715698,
      "learning_rate": 0.0001653312527498751,
      "loss": 3.4587,
      "step": 185960
    },
    {
      "epoch": 0.1733541314168708,
      "grad_norm": 1.1204315423965454,
      "learning_rate": 0.0001653293884277799,
      "loss": 3.6736,
      "step": 185970
    },
    {
      "epoch": 0.1733634530349499,
      "grad_norm": 1.0912854671478271,
      "learning_rate": 0.0001653275241056847,
      "loss": 3.5115,
      "step": 185980
    },
    {
      "epoch": 0.17337277465302897,
      "grad_norm": 0.9499461650848389,
      "learning_rate": 0.0001653256597835895,
      "loss": 3.5504,
      "step": 185990
    },
    {
      "epoch": 0.17338209627110807,
      "grad_norm": 1.2953455448150635,
      "learning_rate": 0.0001653237954614943,
      "loss": 3.6796,
      "step": 186000
    },
    {
      "epoch": 0.17339141788918716,
      "grad_norm": 1.0505707263946533,
      "learning_rate": 0.00016532193113939908,
      "loss": 3.4535,
      "step": 186010
    },
    {
      "epoch": 0.17340073950726625,
      "grad_norm": 1.048218846321106,
      "learning_rate": 0.0001653200668173039,
      "loss": 3.3214,
      "step": 186020
    },
    {
      "epoch": 0.17341006112534535,
      "grad_norm": 0.9934819340705872,
      "learning_rate": 0.00016531820249520872,
      "loss": 3.5747,
      "step": 186030
    },
    {
      "epoch": 0.17341938274342444,
      "grad_norm": 1.1160222291946411,
      "learning_rate": 0.0001653163381731135,
      "loss": 3.5243,
      "step": 186040
    },
    {
      "epoch": 0.17342870436150354,
      "grad_norm": 1.305309534072876,
      "learning_rate": 0.0001653144738510183,
      "loss": 3.5882,
      "step": 186050
    },
    {
      "epoch": 0.17343802597958263,
      "grad_norm": 1.0721850395202637,
      "learning_rate": 0.00016531260952892308,
      "loss": 3.3922,
      "step": 186060
    },
    {
      "epoch": 0.17344734759766173,
      "grad_norm": 1.132940649986267,
      "learning_rate": 0.0001653107452068279,
      "loss": 3.5564,
      "step": 186070
    },
    {
      "epoch": 0.17345666921574082,
      "grad_norm": 0.9169743061065674,
      "learning_rate": 0.00016530888088473272,
      "loss": 3.5588,
      "step": 186080
    },
    {
      "epoch": 0.1734659908338199,
      "grad_norm": 1.1801408529281616,
      "learning_rate": 0.0001653070165626375,
      "loss": 3.6081,
      "step": 186090
    },
    {
      "epoch": 0.17347531245189898,
      "grad_norm": 0.9934693574905396,
      "learning_rate": 0.0001653051522405423,
      "loss": 3.6167,
      "step": 186100
    },
    {
      "epoch": 0.17348463406997808,
      "grad_norm": 0.9863764047622681,
      "learning_rate": 0.00016530328791844709,
      "loss": 3.5601,
      "step": 186110
    },
    {
      "epoch": 0.17349395568805717,
      "grad_norm": 0.9085943698883057,
      "learning_rate": 0.0001653014235963519,
      "loss": 3.5353,
      "step": 186120
    },
    {
      "epoch": 0.17350327730613627,
      "grad_norm": 1.0016095638275146,
      "learning_rate": 0.0001652995592742567,
      "loss": 3.5265,
      "step": 186130
    },
    {
      "epoch": 0.17351259892421536,
      "grad_norm": 0.9777167439460754,
      "learning_rate": 0.0001652976949521615,
      "loss": 3.4624,
      "step": 186140
    },
    {
      "epoch": 0.17352192054229446,
      "grad_norm": 0.9726812839508057,
      "learning_rate": 0.0001652958306300663,
      "loss": 3.7734,
      "step": 186150
    },
    {
      "epoch": 0.17353124216037355,
      "grad_norm": 1.2504006624221802,
      "learning_rate": 0.0001652939663079711,
      "loss": 3.6368,
      "step": 186160
    },
    {
      "epoch": 0.17354056377845264,
      "grad_norm": 0.8321353197097778,
      "learning_rate": 0.0001652921019858759,
      "loss": 3.4819,
      "step": 186170
    },
    {
      "epoch": 0.17354988539653174,
      "grad_norm": 1.1106834411621094,
      "learning_rate": 0.0001652902376637807,
      "loss": 3.4054,
      "step": 186180
    },
    {
      "epoch": 0.17355920701461083,
      "grad_norm": 1.5728397369384766,
      "learning_rate": 0.0001652883733416855,
      "loss": 3.3822,
      "step": 186190
    },
    {
      "epoch": 0.1735685286326899,
      "grad_norm": 1.090702772140503,
      "learning_rate": 0.00016528650901959032,
      "loss": 3.4721,
      "step": 186200
    },
    {
      "epoch": 0.173577850250769,
      "grad_norm": 1.0229413509368896,
      "learning_rate": 0.0001652846446974951,
      "loss": 3.4912,
      "step": 186210
    },
    {
      "epoch": 0.1735871718688481,
      "grad_norm": 1.0082322359085083,
      "learning_rate": 0.0001652827803753999,
      "loss": 3.3517,
      "step": 186220
    },
    {
      "epoch": 0.17359649348692718,
      "grad_norm": 1.0714221000671387,
      "learning_rate": 0.0001652809160533047,
      "loss": 3.6311,
      "step": 186230
    },
    {
      "epoch": 0.17360581510500628,
      "grad_norm": 1.0115845203399658,
      "learning_rate": 0.0001652790517312095,
      "loss": 3.5239,
      "step": 186240
    },
    {
      "epoch": 0.17361513672308537,
      "grad_norm": 1.1838159561157227,
      "learning_rate": 0.00016527718740911432,
      "loss": 3.576,
      "step": 186250
    },
    {
      "epoch": 0.17362445834116447,
      "grad_norm": 1.145336627960205,
      "learning_rate": 0.0001652753230870191,
      "loss": 3.5834,
      "step": 186260
    },
    {
      "epoch": 0.17363377995924356,
      "grad_norm": 1.0603665113449097,
      "learning_rate": 0.0001652734587649239,
      "loss": 3.5516,
      "step": 186270
    },
    {
      "epoch": 0.17364310157732266,
      "grad_norm": 0.9301844835281372,
      "learning_rate": 0.0001652715944428287,
      "loss": 3.5117,
      "step": 186280
    },
    {
      "epoch": 0.17365242319540175,
      "grad_norm": 0.9632061123847961,
      "learning_rate": 0.0001652697301207335,
      "loss": 3.3953,
      "step": 186290
    },
    {
      "epoch": 0.17366174481348082,
      "grad_norm": 1.088657259941101,
      "learning_rate": 0.0001652678657986383,
      "loss": 3.6364,
      "step": 186300
    },
    {
      "epoch": 0.1736710664315599,
      "grad_norm": 1.1228522062301636,
      "learning_rate": 0.0001652660014765431,
      "loss": 3.5668,
      "step": 186310
    },
    {
      "epoch": 0.173680388049639,
      "grad_norm": 0.977061927318573,
      "learning_rate": 0.0001652641371544479,
      "loss": 3.7176,
      "step": 186320
    },
    {
      "epoch": 0.1736897096677181,
      "grad_norm": 0.8087542653083801,
      "learning_rate": 0.00016526227283235272,
      "loss": 3.5977,
      "step": 186330
    },
    {
      "epoch": 0.1736990312857972,
      "grad_norm": 0.9479057788848877,
      "learning_rate": 0.0001652604085102575,
      "loss": 3.4975,
      "step": 186340
    },
    {
      "epoch": 0.1737083529038763,
      "grad_norm": 1.1377003192901611,
      "learning_rate": 0.0001652585441881623,
      "loss": 3.5019,
      "step": 186350
    },
    {
      "epoch": 0.17371767452195538,
      "grad_norm": 0.9988225698471069,
      "learning_rate": 0.0001652566798660671,
      "loss": 3.5287,
      "step": 186360
    },
    {
      "epoch": 0.17372699614003448,
      "grad_norm": 1.084678053855896,
      "learning_rate": 0.00016525481554397191,
      "loss": 3.5406,
      "step": 186370
    },
    {
      "epoch": 0.17373631775811357,
      "grad_norm": 1.0764036178588867,
      "learning_rate": 0.00016525295122187672,
      "loss": 3.3826,
      "step": 186380
    },
    {
      "epoch": 0.17374563937619267,
      "grad_norm": 1.1084964275360107,
      "learning_rate": 0.0001652510868997815,
      "loss": 3.5976,
      "step": 186390
    },
    {
      "epoch": 0.17375496099427176,
      "grad_norm": 1.0351200103759766,
      "learning_rate": 0.0001652492225776863,
      "loss": 3.5677,
      "step": 186400
    },
    {
      "epoch": 0.17376428261235083,
      "grad_norm": 1.0424035787582397,
      "learning_rate": 0.0001652473582555911,
      "loss": 3.4643,
      "step": 186410
    },
    {
      "epoch": 0.17377360423042992,
      "grad_norm": 1.1373803615570068,
      "learning_rate": 0.00016524549393349592,
      "loss": 3.6009,
      "step": 186420
    },
    {
      "epoch": 0.17378292584850902,
      "grad_norm": 1.0376819372177124,
      "learning_rate": 0.00016524362961140072,
      "loss": 3.5017,
      "step": 186430
    },
    {
      "epoch": 0.1737922474665881,
      "grad_norm": 1.00558602809906,
      "learning_rate": 0.0001652417652893055,
      "loss": 3.5992,
      "step": 186440
    },
    {
      "epoch": 0.1738015690846672,
      "grad_norm": 1.3451284170150757,
      "learning_rate": 0.0001652399009672103,
      "loss": 3.6066,
      "step": 186450
    },
    {
      "epoch": 0.1738108907027463,
      "grad_norm": 1.0679776668548584,
      "learning_rate": 0.0001652380366451151,
      "loss": 3.5024,
      "step": 186460
    },
    {
      "epoch": 0.1738202123208254,
      "grad_norm": 1.155139446258545,
      "learning_rate": 0.00016523617232301992,
      "loss": 3.5506,
      "step": 186470
    },
    {
      "epoch": 0.1738295339389045,
      "grad_norm": 0.8799071907997131,
      "learning_rate": 0.00016523430800092473,
      "loss": 3.4962,
      "step": 186480
    },
    {
      "epoch": 0.17383885555698358,
      "grad_norm": 1.245072603225708,
      "learning_rate": 0.0001652324436788295,
      "loss": 3.5628,
      "step": 186490
    },
    {
      "epoch": 0.17384817717506268,
      "grad_norm": 1.0275827646255493,
      "learning_rate": 0.00016523057935673431,
      "loss": 3.5844,
      "step": 186500
    },
    {
      "epoch": 0.17385749879314175,
      "grad_norm": 0.9787331223487854,
      "learning_rate": 0.0001652287150346391,
      "loss": 3.5328,
      "step": 186510
    },
    {
      "epoch": 0.17386682041122084,
      "grad_norm": 0.9915316700935364,
      "learning_rate": 0.0001652268507125439,
      "loss": 3.7189,
      "step": 186520
    },
    {
      "epoch": 0.17387614202929993,
      "grad_norm": 0.9717321991920471,
      "learning_rate": 0.00016522498639044873,
      "loss": 3.6852,
      "step": 186530
    },
    {
      "epoch": 0.17388546364737903,
      "grad_norm": 1.3602250814437866,
      "learning_rate": 0.0001652231220683535,
      "loss": 3.6296,
      "step": 186540
    },
    {
      "epoch": 0.17389478526545812,
      "grad_norm": 0.8474898934364319,
      "learning_rate": 0.00016522125774625832,
      "loss": 3.5803,
      "step": 186550
    },
    {
      "epoch": 0.17390410688353722,
      "grad_norm": 1.1265641450881958,
      "learning_rate": 0.0001652193934241631,
      "loss": 3.5033,
      "step": 186560
    },
    {
      "epoch": 0.1739134285016163,
      "grad_norm": 1.0786539316177368,
      "learning_rate": 0.0001652175291020679,
      "loss": 3.4556,
      "step": 186570
    },
    {
      "epoch": 0.1739227501196954,
      "grad_norm": 0.9957777261734009,
      "learning_rate": 0.0001652156647799727,
      "loss": 3.4067,
      "step": 186580
    },
    {
      "epoch": 0.1739320717377745,
      "grad_norm": 1.123471975326538,
      "learning_rate": 0.00016521380045787752,
      "loss": 3.6129,
      "step": 186590
    },
    {
      "epoch": 0.1739413933558536,
      "grad_norm": 1.1041946411132812,
      "learning_rate": 0.00016521193613578232,
      "loss": 3.5399,
      "step": 186600
    },
    {
      "epoch": 0.1739507149739327,
      "grad_norm": 1.133171558380127,
      "learning_rate": 0.0001652100718136871,
      "loss": 3.3804,
      "step": 186610
    },
    {
      "epoch": 0.17396003659201176,
      "grad_norm": 1.2303887605667114,
      "learning_rate": 0.0001652082074915919,
      "loss": 3.5369,
      "step": 186620
    },
    {
      "epoch": 0.17396935821009085,
      "grad_norm": 1.09827721118927,
      "learning_rate": 0.00016520634316949672,
      "loss": 3.6663,
      "step": 186630
    },
    {
      "epoch": 0.17397867982816995,
      "grad_norm": 1.0814969539642334,
      "learning_rate": 0.00016520447884740152,
      "loss": 3.4667,
      "step": 186640
    },
    {
      "epoch": 0.17398800144624904,
      "grad_norm": 1.1496230363845825,
      "learning_rate": 0.00016520261452530633,
      "loss": 3.4869,
      "step": 186650
    },
    {
      "epoch": 0.17399732306432814,
      "grad_norm": 0.8498661518096924,
      "learning_rate": 0.0001652007502032111,
      "loss": 3.5646,
      "step": 186660
    },
    {
      "epoch": 0.17400664468240723,
      "grad_norm": 1.2892425060272217,
      "learning_rate": 0.00016519888588111591,
      "loss": 3.555,
      "step": 186670
    },
    {
      "epoch": 0.17401596630048632,
      "grad_norm": 0.9175763130187988,
      "learning_rate": 0.00016519702155902072,
      "loss": 3.6093,
      "step": 186680
    },
    {
      "epoch": 0.17402528791856542,
      "grad_norm": 0.9001764059066772,
      "learning_rate": 0.00016519515723692553,
      "loss": 3.526,
      "step": 186690
    },
    {
      "epoch": 0.1740346095366445,
      "grad_norm": 1.008259892463684,
      "learning_rate": 0.00016519329291483033,
      "loss": 3.5603,
      "step": 186700
    },
    {
      "epoch": 0.1740439311547236,
      "grad_norm": 1.1794441938400269,
      "learning_rate": 0.0001651914285927351,
      "loss": 3.5575,
      "step": 186710
    },
    {
      "epoch": 0.17405325277280267,
      "grad_norm": 1.1261159181594849,
      "learning_rate": 0.00016518956427063992,
      "loss": 3.5338,
      "step": 186720
    },
    {
      "epoch": 0.17406257439088177,
      "grad_norm": 1.20926833152771,
      "learning_rate": 0.00016518769994854472,
      "loss": 3.5998,
      "step": 186730
    },
    {
      "epoch": 0.17407189600896086,
      "grad_norm": 1.0952250957489014,
      "learning_rate": 0.0001651858356264495,
      "loss": 3.4313,
      "step": 186740
    },
    {
      "epoch": 0.17408121762703996,
      "grad_norm": 1.0682029724121094,
      "learning_rate": 0.00016518397130435434,
      "loss": 3.5377,
      "step": 186750
    },
    {
      "epoch": 0.17409053924511905,
      "grad_norm": 1.1535027027130127,
      "learning_rate": 0.00016518210698225912,
      "loss": 3.4532,
      "step": 186760
    },
    {
      "epoch": 0.17409986086319815,
      "grad_norm": 1.0561603307724,
      "learning_rate": 0.00016518024266016392,
      "loss": 3.312,
      "step": 186770
    },
    {
      "epoch": 0.17410918248127724,
      "grad_norm": 1.0174739360809326,
      "learning_rate": 0.00016517837833806873,
      "loss": 3.5669,
      "step": 186780
    },
    {
      "epoch": 0.17411850409935634,
      "grad_norm": 1.0747838020324707,
      "learning_rate": 0.0001651765140159735,
      "loss": 3.5951,
      "step": 186790
    },
    {
      "epoch": 0.17412782571743543,
      "grad_norm": 1.1808643341064453,
      "learning_rate": 0.00016517464969387831,
      "loss": 3.7364,
      "step": 186800
    },
    {
      "epoch": 0.17413714733551452,
      "grad_norm": 1.1638017892837524,
      "learning_rate": 0.00016517278537178312,
      "loss": 3.5807,
      "step": 186810
    },
    {
      "epoch": 0.17414646895359362,
      "grad_norm": 1.0104817152023315,
      "learning_rate": 0.00016517092104968793,
      "loss": 3.5137,
      "step": 186820
    },
    {
      "epoch": 0.1741557905716727,
      "grad_norm": 1.0019937753677368,
      "learning_rate": 0.00016516905672759273,
      "loss": 3.3777,
      "step": 186830
    },
    {
      "epoch": 0.17416511218975178,
      "grad_norm": 1.1460129022598267,
      "learning_rate": 0.0001651671924054975,
      "loss": 3.4667,
      "step": 186840
    },
    {
      "epoch": 0.17417443380783088,
      "grad_norm": 1.0113235712051392,
      "learning_rate": 0.00016516532808340232,
      "loss": 3.6422,
      "step": 186850
    },
    {
      "epoch": 0.17418375542590997,
      "grad_norm": 1.0672247409820557,
      "learning_rate": 0.00016516346376130713,
      "loss": 3.7324,
      "step": 186860
    },
    {
      "epoch": 0.17419307704398906,
      "grad_norm": 1.00771164894104,
      "learning_rate": 0.00016516159943921193,
      "loss": 3.4829,
      "step": 186870
    },
    {
      "epoch": 0.17420239866206816,
      "grad_norm": 1.1548302173614502,
      "learning_rate": 0.00016515973511711674,
      "loss": 3.5944,
      "step": 186880
    },
    {
      "epoch": 0.17421172028014725,
      "grad_norm": 1.0717487335205078,
      "learning_rate": 0.00016515787079502152,
      "loss": 3.5847,
      "step": 186890
    },
    {
      "epoch": 0.17422104189822635,
      "grad_norm": 1.0990381240844727,
      "learning_rate": 0.00016515600647292632,
      "loss": 3.6775,
      "step": 186900
    },
    {
      "epoch": 0.17423036351630544,
      "grad_norm": 0.8458139896392822,
      "learning_rate": 0.0001651541421508311,
      "loss": 3.5006,
      "step": 186910
    },
    {
      "epoch": 0.17423968513438454,
      "grad_norm": 1.0506718158721924,
      "learning_rate": 0.00016515227782873594,
      "loss": 3.5903,
      "step": 186920
    },
    {
      "epoch": 0.1742490067524636,
      "grad_norm": 1.1715532541275024,
      "learning_rate": 0.00016515041350664074,
      "loss": 3.6921,
      "step": 186930
    },
    {
      "epoch": 0.1742583283705427,
      "grad_norm": 1.3117711544036865,
      "learning_rate": 0.00016514854918454552,
      "loss": 3.7307,
      "step": 186940
    },
    {
      "epoch": 0.1742676499886218,
      "grad_norm": 0.899185299873352,
      "learning_rate": 0.00016514668486245033,
      "loss": 3.5621,
      "step": 186950
    },
    {
      "epoch": 0.1742769716067009,
      "grad_norm": 0.9754799008369446,
      "learning_rate": 0.0001651448205403551,
      "loss": 3.4809,
      "step": 186960
    },
    {
      "epoch": 0.17428629322477998,
      "grad_norm": 1.1224448680877686,
      "learning_rate": 0.0001651429562182599,
      "loss": 3.5094,
      "step": 186970
    },
    {
      "epoch": 0.17429561484285908,
      "grad_norm": 0.9721072316169739,
      "learning_rate": 0.00016514109189616475,
      "loss": 3.6163,
      "step": 186980
    },
    {
      "epoch": 0.17430493646093817,
      "grad_norm": 0.9158444404602051,
      "learning_rate": 0.00016513922757406953,
      "loss": 3.4405,
      "step": 186990
    },
    {
      "epoch": 0.17431425807901726,
      "grad_norm": 0.9656952619552612,
      "learning_rate": 0.00016513736325197433,
      "loss": 3.6264,
      "step": 187000
    },
    {
      "epoch": 0.17432357969709636,
      "grad_norm": 1.2933082580566406,
      "learning_rate": 0.0001651354989298791,
      "loss": 3.6985,
      "step": 187010
    },
    {
      "epoch": 0.17433290131517545,
      "grad_norm": 1.0214842557907104,
      "learning_rate": 0.00016513363460778392,
      "loss": 3.5697,
      "step": 187020
    },
    {
      "epoch": 0.17434222293325455,
      "grad_norm": 1.048687219619751,
      "learning_rate": 0.00016513177028568872,
      "loss": 3.6878,
      "step": 187030
    },
    {
      "epoch": 0.17435154455133361,
      "grad_norm": 1.6489717960357666,
      "learning_rate": 0.00016512990596359353,
      "loss": 3.3806,
      "step": 187040
    },
    {
      "epoch": 0.1743608661694127,
      "grad_norm": 1.1101655960083008,
      "learning_rate": 0.00016512804164149834,
      "loss": 3.6444,
      "step": 187050
    },
    {
      "epoch": 0.1743701877874918,
      "grad_norm": 1.0514156818389893,
      "learning_rate": 0.00016512617731940312,
      "loss": 3.4745,
      "step": 187060
    },
    {
      "epoch": 0.1743795094055709,
      "grad_norm": 1.2029988765716553,
      "learning_rate": 0.00016512431299730792,
      "loss": 3.6646,
      "step": 187070
    },
    {
      "epoch": 0.17438883102365,
      "grad_norm": 1.141459345817566,
      "learning_rate": 0.00016512244867521273,
      "loss": 3.4452,
      "step": 187080
    },
    {
      "epoch": 0.1743981526417291,
      "grad_norm": 0.9629713296890259,
      "learning_rate": 0.00016512058435311753,
      "loss": 3.3656,
      "step": 187090
    },
    {
      "epoch": 0.17440747425980818,
      "grad_norm": 1.3106037378311157,
      "learning_rate": 0.00016511872003102234,
      "loss": 3.6825,
      "step": 187100
    },
    {
      "epoch": 0.17441679587788728,
      "grad_norm": 1.0512648820877075,
      "learning_rate": 0.00016511685570892712,
      "loss": 3.5316,
      "step": 187110
    },
    {
      "epoch": 0.17442611749596637,
      "grad_norm": 1.3412976264953613,
      "learning_rate": 0.00016511499138683193,
      "loss": 3.5647,
      "step": 187120
    },
    {
      "epoch": 0.17443543911404547,
      "grad_norm": 1.2167476415634155,
      "learning_rate": 0.00016511312706473673,
      "loss": 3.5996,
      "step": 187130
    },
    {
      "epoch": 0.17444476073212453,
      "grad_norm": 1.1005486249923706,
      "learning_rate": 0.00016511126274264154,
      "loss": 3.4538,
      "step": 187140
    },
    {
      "epoch": 0.17445408235020363,
      "grad_norm": 1.1950803995132446,
      "learning_rate": 0.00016510939842054635,
      "loss": 3.539,
      "step": 187150
    },
    {
      "epoch": 0.17446340396828272,
      "grad_norm": 1.1672471761703491,
      "learning_rate": 0.00016510753409845112,
      "loss": 3.5414,
      "step": 187160
    },
    {
      "epoch": 0.17447272558636182,
      "grad_norm": 0.9576796293258667,
      "learning_rate": 0.00016510566977635593,
      "loss": 3.701,
      "step": 187170
    },
    {
      "epoch": 0.1744820472044409,
      "grad_norm": 1.0606064796447754,
      "learning_rate": 0.00016510380545426074,
      "loss": 3.4391,
      "step": 187180
    },
    {
      "epoch": 0.17449136882252,
      "grad_norm": 1.075331449508667,
      "learning_rate": 0.00016510194113216552,
      "loss": 3.4908,
      "step": 187190
    },
    {
      "epoch": 0.1745006904405991,
      "grad_norm": 0.9561894536018372,
      "learning_rate": 0.00016510007681007035,
      "loss": 3.6123,
      "step": 187200
    },
    {
      "epoch": 0.1745100120586782,
      "grad_norm": 1.1447069644927979,
      "learning_rate": 0.00016509821248797513,
      "loss": 3.5492,
      "step": 187210
    },
    {
      "epoch": 0.1745193336767573,
      "grad_norm": 0.9976171851158142,
      "learning_rate": 0.00016509634816587994,
      "loss": 3.4827,
      "step": 187220
    },
    {
      "epoch": 0.17452865529483638,
      "grad_norm": 1.034050464630127,
      "learning_rate": 0.00016509448384378474,
      "loss": 3.5398,
      "step": 187230
    },
    {
      "epoch": 0.17453797691291548,
      "grad_norm": 1.15200936794281,
      "learning_rate": 0.00016509261952168952,
      "loss": 3.4479,
      "step": 187240
    },
    {
      "epoch": 0.17454729853099454,
      "grad_norm": 1.1332073211669922,
      "learning_rate": 0.00016509075519959433,
      "loss": 3.5732,
      "step": 187250
    },
    {
      "epoch": 0.17455662014907364,
      "grad_norm": 1.3030346632003784,
      "learning_rate": 0.00016508889087749913,
      "loss": 3.5564,
      "step": 187260
    },
    {
      "epoch": 0.17456594176715273,
      "grad_norm": 1.1611484289169312,
      "learning_rate": 0.00016508702655540394,
      "loss": 3.6454,
      "step": 187270
    },
    {
      "epoch": 0.17457526338523183,
      "grad_norm": 0.9194400906562805,
      "learning_rate": 0.00016508516223330872,
      "loss": 3.6669,
      "step": 187280
    },
    {
      "epoch": 0.17458458500331092,
      "grad_norm": 0.9891896843910217,
      "learning_rate": 0.00016508329791121353,
      "loss": 3.565,
      "step": 187290
    },
    {
      "epoch": 0.17459390662139002,
      "grad_norm": 0.9805238842964172,
      "learning_rate": 0.00016508143358911833,
      "loss": 3.6969,
      "step": 187300
    },
    {
      "epoch": 0.1746032282394691,
      "grad_norm": 0.8156234622001648,
      "learning_rate": 0.00016507956926702314,
      "loss": 3.5738,
      "step": 187310
    },
    {
      "epoch": 0.1746125498575482,
      "grad_norm": 1.090161681175232,
      "learning_rate": 0.00016507770494492794,
      "loss": 3.3999,
      "step": 187320
    },
    {
      "epoch": 0.1746218714756273,
      "grad_norm": 1.1428800821304321,
      "learning_rate": 0.00016507584062283272,
      "loss": 3.569,
      "step": 187330
    },
    {
      "epoch": 0.1746311930937064,
      "grad_norm": 1.1715465784072876,
      "learning_rate": 0.00016507397630073753,
      "loss": 3.5632,
      "step": 187340
    },
    {
      "epoch": 0.17464051471178546,
      "grad_norm": 1.0680090188980103,
      "learning_rate": 0.00016507211197864234,
      "loss": 3.437,
      "step": 187350
    },
    {
      "epoch": 0.17464983632986456,
      "grad_norm": 1.2026811838150024,
      "learning_rate": 0.00016507024765654712,
      "loss": 3.4752,
      "step": 187360
    },
    {
      "epoch": 0.17465915794794365,
      "grad_norm": 1.1395201683044434,
      "learning_rate": 0.00016506838333445195,
      "loss": 3.5995,
      "step": 187370
    },
    {
      "epoch": 0.17466847956602274,
      "grad_norm": 1.135693073272705,
      "learning_rate": 0.00016506651901235673,
      "loss": 3.5899,
      "step": 187380
    },
    {
      "epoch": 0.17467780118410184,
      "grad_norm": 1.7402094602584839,
      "learning_rate": 0.00016506465469026153,
      "loss": 3.7281,
      "step": 187390
    },
    {
      "epoch": 0.17468712280218093,
      "grad_norm": 1.084130048751831,
      "learning_rate": 0.00016506279036816634,
      "loss": 3.7203,
      "step": 187400
    },
    {
      "epoch": 0.17469644442026003,
      "grad_norm": 1.179296612739563,
      "learning_rate": 0.00016506092604607112,
      "loss": 3.5058,
      "step": 187410
    },
    {
      "epoch": 0.17470576603833912,
      "grad_norm": 0.9913820028305054,
      "learning_rate": 0.00016505906172397593,
      "loss": 3.5909,
      "step": 187420
    },
    {
      "epoch": 0.17471508765641822,
      "grad_norm": 1.0266807079315186,
      "learning_rate": 0.00016505719740188073,
      "loss": 3.5133,
      "step": 187430
    },
    {
      "epoch": 0.1747244092744973,
      "grad_norm": 1.1065948009490967,
      "learning_rate": 0.00016505533307978554,
      "loss": 3.4737,
      "step": 187440
    },
    {
      "epoch": 0.17473373089257638,
      "grad_norm": 1.1165716648101807,
      "learning_rate": 0.00016505346875769034,
      "loss": 3.5131,
      "step": 187450
    },
    {
      "epoch": 0.17474305251065547,
      "grad_norm": 1.189547061920166,
      "learning_rate": 0.00016505160443559512,
      "loss": 3.5821,
      "step": 187460
    },
    {
      "epoch": 0.17475237412873457,
      "grad_norm": 1.0801513195037842,
      "learning_rate": 0.00016504974011349993,
      "loss": 3.4784,
      "step": 187470
    },
    {
      "epoch": 0.17476169574681366,
      "grad_norm": 1.0671342611312866,
      "learning_rate": 0.00016504787579140474,
      "loss": 3.5077,
      "step": 187480
    },
    {
      "epoch": 0.17477101736489276,
      "grad_norm": 0.8682488203048706,
      "learning_rate": 0.00016504601146930954,
      "loss": 3.744,
      "step": 187490
    },
    {
      "epoch": 0.17478033898297185,
      "grad_norm": 1.105470895767212,
      "learning_rate": 0.00016504414714721435,
      "loss": 3.4663,
      "step": 187500
    },
    {
      "epoch": 0.17478966060105094,
      "grad_norm": 1.0257816314697266,
      "learning_rate": 0.00016504228282511913,
      "loss": 3.3761,
      "step": 187510
    },
    {
      "epoch": 0.17479898221913004,
      "grad_norm": 1.0694749355316162,
      "learning_rate": 0.00016504041850302393,
      "loss": 3.6002,
      "step": 187520
    },
    {
      "epoch": 0.17480830383720913,
      "grad_norm": 0.9809202551841736,
      "learning_rate": 0.00016503855418092874,
      "loss": 3.4296,
      "step": 187530
    },
    {
      "epoch": 0.17481762545528823,
      "grad_norm": 1.1022199392318726,
      "learning_rate": 0.00016503668985883355,
      "loss": 3.3993,
      "step": 187540
    },
    {
      "epoch": 0.17482694707336732,
      "grad_norm": 1.1675657033920288,
      "learning_rate": 0.00016503482553673835,
      "loss": 3.4706,
      "step": 187550
    },
    {
      "epoch": 0.1748362686914464,
      "grad_norm": 1.1376861333847046,
      "learning_rate": 0.00016503296121464313,
      "loss": 3.5333,
      "step": 187560
    },
    {
      "epoch": 0.17484559030952548,
      "grad_norm": 1.235583782196045,
      "learning_rate": 0.00016503109689254794,
      "loss": 3.5908,
      "step": 187570
    },
    {
      "epoch": 0.17485491192760458,
      "grad_norm": 1.224961280822754,
      "learning_rate": 0.00016502923257045272,
      "loss": 3.6531,
      "step": 187580
    },
    {
      "epoch": 0.17486423354568367,
      "grad_norm": 1.2308359146118164,
      "learning_rate": 0.00016502736824835755,
      "loss": 3.6785,
      "step": 187590
    },
    {
      "epoch": 0.17487355516376277,
      "grad_norm": 1.0586364269256592,
      "learning_rate": 0.00016502550392626236,
      "loss": 3.54,
      "step": 187600
    },
    {
      "epoch": 0.17488287678184186,
      "grad_norm": 1.0140382051467896,
      "learning_rate": 0.00016502363960416714,
      "loss": 3.4847,
      "step": 187610
    },
    {
      "epoch": 0.17489219839992096,
      "grad_norm": 1.146638035774231,
      "learning_rate": 0.00016502177528207194,
      "loss": 3.5478,
      "step": 187620
    },
    {
      "epoch": 0.17490152001800005,
      "grad_norm": 1.1984854936599731,
      "learning_rate": 0.00016501991095997672,
      "loss": 3.5828,
      "step": 187630
    },
    {
      "epoch": 0.17491084163607915,
      "grad_norm": 0.8121299743652344,
      "learning_rate": 0.00016501804663788153,
      "loss": 3.4858,
      "step": 187640
    },
    {
      "epoch": 0.17492016325415824,
      "grad_norm": 1.0483728647232056,
      "learning_rate": 0.00016501618231578636,
      "loss": 3.6533,
      "step": 187650
    },
    {
      "epoch": 0.1749294848722373,
      "grad_norm": 1.1905057430267334,
      "learning_rate": 0.00016501431799369114,
      "loss": 3.5814,
      "step": 187660
    },
    {
      "epoch": 0.1749388064903164,
      "grad_norm": 1.0493797063827515,
      "learning_rate": 0.00016501245367159595,
      "loss": 3.395,
      "step": 187670
    },
    {
      "epoch": 0.1749481281083955,
      "grad_norm": 1.1223219633102417,
      "learning_rate": 0.00016501058934950073,
      "loss": 3.3928,
      "step": 187680
    },
    {
      "epoch": 0.1749574497264746,
      "grad_norm": 0.9399267435073853,
      "learning_rate": 0.00016500872502740553,
      "loss": 3.5706,
      "step": 187690
    },
    {
      "epoch": 0.17496677134455368,
      "grad_norm": 1.0804976224899292,
      "learning_rate": 0.00016500686070531034,
      "loss": 3.4348,
      "step": 187700
    },
    {
      "epoch": 0.17497609296263278,
      "grad_norm": 0.9688115119934082,
      "learning_rate": 0.00016500499638321515,
      "loss": 3.69,
      "step": 187710
    },
    {
      "epoch": 0.17498541458071187,
      "grad_norm": 1.0099717378616333,
      "learning_rate": 0.00016500313206111995,
      "loss": 3.4898,
      "step": 187720
    },
    {
      "epoch": 0.17499473619879097,
      "grad_norm": 1.0348906517028809,
      "learning_rate": 0.00016500126773902473,
      "loss": 3.6316,
      "step": 187730
    },
    {
      "epoch": 0.17500405781687006,
      "grad_norm": 0.95489501953125,
      "learning_rate": 0.00016499940341692954,
      "loss": 3.5055,
      "step": 187740
    },
    {
      "epoch": 0.17501337943494916,
      "grad_norm": 1.0642038583755493,
      "learning_rate": 0.00016499753909483434,
      "loss": 3.4254,
      "step": 187750
    },
    {
      "epoch": 0.17502270105302825,
      "grad_norm": 1.0201153755187988,
      "learning_rate": 0.00016499567477273915,
      "loss": 3.3675,
      "step": 187760
    },
    {
      "epoch": 0.17503202267110732,
      "grad_norm": 1.1795483827590942,
      "learning_rate": 0.00016499381045064396,
      "loss": 3.6109,
      "step": 187770
    },
    {
      "epoch": 0.1750413442891864,
      "grad_norm": 1.0749200582504272,
      "learning_rate": 0.00016499194612854874,
      "loss": 3.3992,
      "step": 187780
    },
    {
      "epoch": 0.1750506659072655,
      "grad_norm": 1.0492180585861206,
      "learning_rate": 0.00016499008180645354,
      "loss": 3.6431,
      "step": 187790
    },
    {
      "epoch": 0.1750599875253446,
      "grad_norm": 0.9315149188041687,
      "learning_rate": 0.00016498821748435835,
      "loss": 3.6394,
      "step": 187800
    },
    {
      "epoch": 0.1750693091434237,
      "grad_norm": 1.1249003410339355,
      "learning_rate": 0.00016498635316226313,
      "loss": 3.5664,
      "step": 187810
    },
    {
      "epoch": 0.1750786307615028,
      "grad_norm": 1.110705852508545,
      "learning_rate": 0.00016498448884016796,
      "loss": 3.4167,
      "step": 187820
    },
    {
      "epoch": 0.17508795237958188,
      "grad_norm": 0.9637079238891602,
      "learning_rate": 0.00016498262451807274,
      "loss": 3.438,
      "step": 187830
    },
    {
      "epoch": 0.17509727399766098,
      "grad_norm": 1.4773776531219482,
      "learning_rate": 0.00016498076019597755,
      "loss": 3.4853,
      "step": 187840
    },
    {
      "epoch": 0.17510659561574007,
      "grad_norm": 1.0474849939346313,
      "learning_rate": 0.00016497889587388235,
      "loss": 3.4143,
      "step": 187850
    },
    {
      "epoch": 0.17511591723381917,
      "grad_norm": 1.2596979141235352,
      "learning_rate": 0.00016497703155178713,
      "loss": 3.5763,
      "step": 187860
    },
    {
      "epoch": 0.17512523885189824,
      "grad_norm": 1.1668306589126587,
      "learning_rate": 0.00016497516722969194,
      "loss": 3.6338,
      "step": 187870
    },
    {
      "epoch": 0.17513456046997733,
      "grad_norm": 1.147998571395874,
      "learning_rate": 0.00016497330290759675,
      "loss": 3.5968,
      "step": 187880
    },
    {
      "epoch": 0.17514388208805642,
      "grad_norm": 1.098056435585022,
      "learning_rate": 0.00016497143858550155,
      "loss": 3.4977,
      "step": 187890
    },
    {
      "epoch": 0.17515320370613552,
      "grad_norm": 0.906793475151062,
      "learning_rate": 0.00016496957426340636,
      "loss": 3.7106,
      "step": 187900
    },
    {
      "epoch": 0.1751625253242146,
      "grad_norm": 1.2274936437606812,
      "learning_rate": 0.00016496770994131114,
      "loss": 3.4488,
      "step": 187910
    },
    {
      "epoch": 0.1751718469422937,
      "grad_norm": 1.1470918655395508,
      "learning_rate": 0.00016496584561921594,
      "loss": 3.4684,
      "step": 187920
    },
    {
      "epoch": 0.1751811685603728,
      "grad_norm": 2.4344544410705566,
      "learning_rate": 0.00016496398129712075,
      "loss": 3.4278,
      "step": 187930
    },
    {
      "epoch": 0.1751904901784519,
      "grad_norm": 1.1870509386062622,
      "learning_rate": 0.00016496211697502556,
      "loss": 3.6355,
      "step": 187940
    },
    {
      "epoch": 0.175199811796531,
      "grad_norm": 1.1532337665557861,
      "learning_rate": 0.00016496025265293036,
      "loss": 3.4641,
      "step": 187950
    },
    {
      "epoch": 0.17520913341461009,
      "grad_norm": 1.0912044048309326,
      "learning_rate": 0.00016495838833083514,
      "loss": 3.3795,
      "step": 187960
    },
    {
      "epoch": 0.17521845503268918,
      "grad_norm": 1.803109049797058,
      "learning_rate": 0.00016495652400873995,
      "loss": 3.4206,
      "step": 187970
    },
    {
      "epoch": 0.17522777665076825,
      "grad_norm": 1.0970231294631958,
      "learning_rate": 0.00016495465968664475,
      "loss": 3.5769,
      "step": 187980
    },
    {
      "epoch": 0.17523709826884734,
      "grad_norm": 1.1357706785202026,
      "learning_rate": 0.00016495279536454956,
      "loss": 3.5399,
      "step": 187990
    },
    {
      "epoch": 0.17524641988692644,
      "grad_norm": 1.0101579427719116,
      "learning_rate": 0.00016495093104245437,
      "loss": 3.5495,
      "step": 188000
    },
    {
      "epoch": 0.17525574150500553,
      "grad_norm": 1.1242343187332153,
      "learning_rate": 0.00016494906672035915,
      "loss": 3.4518,
      "step": 188010
    },
    {
      "epoch": 0.17526506312308462,
      "grad_norm": 1.267605185508728,
      "learning_rate": 0.00016494720239826395,
      "loss": 3.5027,
      "step": 188020
    },
    {
      "epoch": 0.17527438474116372,
      "grad_norm": 1.0641928911209106,
      "learning_rate": 0.00016494533807616873,
      "loss": 3.6896,
      "step": 188030
    },
    {
      "epoch": 0.1752837063592428,
      "grad_norm": 1.3150466680526733,
      "learning_rate": 0.00016494347375407356,
      "loss": 3.5135,
      "step": 188040
    },
    {
      "epoch": 0.1752930279773219,
      "grad_norm": 0.9912251234054565,
      "learning_rate": 0.00016494160943197837,
      "loss": 3.3868,
      "step": 188050
    },
    {
      "epoch": 0.175302349595401,
      "grad_norm": 1.0875006914138794,
      "learning_rate": 0.00016493974510988315,
      "loss": 3.5881,
      "step": 188060
    },
    {
      "epoch": 0.1753116712134801,
      "grad_norm": 1.0073968172073364,
      "learning_rate": 0.00016493788078778796,
      "loss": 3.6248,
      "step": 188070
    },
    {
      "epoch": 0.17532099283155916,
      "grad_norm": 1.2261329889297485,
      "learning_rate": 0.00016493601646569274,
      "loss": 3.3801,
      "step": 188080
    },
    {
      "epoch": 0.17533031444963826,
      "grad_norm": 1.0205464363098145,
      "learning_rate": 0.00016493415214359754,
      "loss": 3.4987,
      "step": 188090
    },
    {
      "epoch": 0.17533963606771735,
      "grad_norm": 1.0313169956207275,
      "learning_rate": 0.00016493228782150238,
      "loss": 3.4625,
      "step": 188100
    },
    {
      "epoch": 0.17534895768579645,
      "grad_norm": 1.1349210739135742,
      "learning_rate": 0.00016493042349940715,
      "loss": 3.4223,
      "step": 188110
    },
    {
      "epoch": 0.17535827930387554,
      "grad_norm": 1.0695812702178955,
      "learning_rate": 0.00016492855917731196,
      "loss": 3.3459,
      "step": 188120
    },
    {
      "epoch": 0.17536760092195464,
      "grad_norm": 1.0574055910110474,
      "learning_rate": 0.00016492669485521674,
      "loss": 3.6344,
      "step": 188130
    },
    {
      "epoch": 0.17537692254003373,
      "grad_norm": 1.1121087074279785,
      "learning_rate": 0.00016492483053312155,
      "loss": 3.5024,
      "step": 188140
    },
    {
      "epoch": 0.17538624415811283,
      "grad_norm": 1.1751337051391602,
      "learning_rate": 0.00016492296621102635,
      "loss": 3.3974,
      "step": 188150
    },
    {
      "epoch": 0.17539556577619192,
      "grad_norm": 1.0490479469299316,
      "learning_rate": 0.00016492110188893116,
      "loss": 3.366,
      "step": 188160
    },
    {
      "epoch": 0.17540488739427101,
      "grad_norm": 1.0512731075286865,
      "learning_rate": 0.00016491923756683597,
      "loss": 3.512,
      "step": 188170
    },
    {
      "epoch": 0.1754142090123501,
      "grad_norm": 1.2027794122695923,
      "learning_rate": 0.00016491737324474074,
      "loss": 3.5687,
      "step": 188180
    },
    {
      "epoch": 0.17542353063042918,
      "grad_norm": 1.0809890031814575,
      "learning_rate": 0.00016491550892264555,
      "loss": 3.4426,
      "step": 188190
    },
    {
      "epoch": 0.17543285224850827,
      "grad_norm": 1.1116366386413574,
      "learning_rate": 0.00016491364460055036,
      "loss": 3.4719,
      "step": 188200
    },
    {
      "epoch": 0.17544217386658736,
      "grad_norm": 1.0197579860687256,
      "learning_rate": 0.00016491178027845516,
      "loss": 3.5372,
      "step": 188210
    },
    {
      "epoch": 0.17545149548466646,
      "grad_norm": 0.966536283493042,
      "learning_rate": 0.00016490991595635997,
      "loss": 3.5138,
      "step": 188220
    },
    {
      "epoch": 0.17546081710274555,
      "grad_norm": 1.0325297117233276,
      "learning_rate": 0.00016490805163426475,
      "loss": 3.5324,
      "step": 188230
    },
    {
      "epoch": 0.17547013872082465,
      "grad_norm": 0.9711431860923767,
      "learning_rate": 0.00016490618731216956,
      "loss": 3.4632,
      "step": 188240
    },
    {
      "epoch": 0.17547946033890374,
      "grad_norm": 1.1442747116088867,
      "learning_rate": 0.00016490432299007436,
      "loss": 3.5746,
      "step": 188250
    },
    {
      "epoch": 0.17548878195698284,
      "grad_norm": 1.262807846069336,
      "learning_rate": 0.00016490245866797914,
      "loss": 3.5561,
      "step": 188260
    },
    {
      "epoch": 0.17549810357506193,
      "grad_norm": 1.0858194828033447,
      "learning_rate": 0.00016490059434588397,
      "loss": 3.4771,
      "step": 188270
    },
    {
      "epoch": 0.17550742519314103,
      "grad_norm": 1.172018051147461,
      "learning_rate": 0.00016489873002378875,
      "loss": 3.4127,
      "step": 188280
    },
    {
      "epoch": 0.1755167468112201,
      "grad_norm": 0.9605517387390137,
      "learning_rate": 0.00016489686570169356,
      "loss": 3.6015,
      "step": 188290
    },
    {
      "epoch": 0.1755260684292992,
      "grad_norm": 1.156936526298523,
      "learning_rate": 0.00016489500137959837,
      "loss": 3.4338,
      "step": 188300
    },
    {
      "epoch": 0.17553539004737828,
      "grad_norm": 1.0867555141448975,
      "learning_rate": 0.00016489313705750315,
      "loss": 3.5206,
      "step": 188310
    },
    {
      "epoch": 0.17554471166545738,
      "grad_norm": 0.9929894208908081,
      "learning_rate": 0.00016489127273540795,
      "loss": 3.5993,
      "step": 188320
    },
    {
      "epoch": 0.17555403328353647,
      "grad_norm": 1.256765365600586,
      "learning_rate": 0.00016488940841331276,
      "loss": 3.5121,
      "step": 188330
    },
    {
      "epoch": 0.17556335490161556,
      "grad_norm": 1.0511113405227661,
      "learning_rate": 0.00016488754409121756,
      "loss": 3.4403,
      "step": 188340
    },
    {
      "epoch": 0.17557267651969466,
      "grad_norm": 1.096084713935852,
      "learning_rate": 0.00016488567976912237,
      "loss": 3.5001,
      "step": 188350
    },
    {
      "epoch": 0.17558199813777375,
      "grad_norm": 0.8084904551506042,
      "learning_rate": 0.00016488381544702715,
      "loss": 3.6202,
      "step": 188360
    },
    {
      "epoch": 0.17559131975585285,
      "grad_norm": 0.8400296568870544,
      "learning_rate": 0.00016488195112493196,
      "loss": 3.4314,
      "step": 188370
    },
    {
      "epoch": 0.17560064137393194,
      "grad_norm": 1.1084166765213013,
      "learning_rate": 0.00016488008680283676,
      "loss": 3.3658,
      "step": 188380
    },
    {
      "epoch": 0.17560996299201104,
      "grad_norm": 0.9181800484657288,
      "learning_rate": 0.00016487822248074157,
      "loss": 3.4482,
      "step": 188390
    },
    {
      "epoch": 0.1756192846100901,
      "grad_norm": 1.1109272241592407,
      "learning_rate": 0.00016487635815864638,
      "loss": 3.696,
      "step": 188400
    },
    {
      "epoch": 0.1756286062281692,
      "grad_norm": 1.2865102291107178,
      "learning_rate": 0.00016487449383655115,
      "loss": 3.4977,
      "step": 188410
    },
    {
      "epoch": 0.1756379278462483,
      "grad_norm": 0.8623461723327637,
      "learning_rate": 0.00016487262951445596,
      "loss": 3.6092,
      "step": 188420
    },
    {
      "epoch": 0.1756472494643274,
      "grad_norm": 1.2401676177978516,
      "learning_rate": 0.00016487076519236077,
      "loss": 3.6603,
      "step": 188430
    },
    {
      "epoch": 0.17565657108240648,
      "grad_norm": 0.9791706800460815,
      "learning_rate": 0.00016486890087026557,
      "loss": 3.5126,
      "step": 188440
    },
    {
      "epoch": 0.17566589270048558,
      "grad_norm": 1.0854368209838867,
      "learning_rate": 0.00016486703654817038,
      "loss": 3.4674,
      "step": 188450
    },
    {
      "epoch": 0.17567521431856467,
      "grad_norm": 1.4071861505508423,
      "learning_rate": 0.00016486517222607516,
      "loss": 3.6129,
      "step": 188460
    },
    {
      "epoch": 0.17568453593664377,
      "grad_norm": 1.0383069515228271,
      "learning_rate": 0.00016486330790397997,
      "loss": 3.525,
      "step": 188470
    },
    {
      "epoch": 0.17569385755472286,
      "grad_norm": 1.2247314453125,
      "learning_rate": 0.00016486144358188474,
      "loss": 3.584,
      "step": 188480
    },
    {
      "epoch": 0.17570317917280195,
      "grad_norm": 1.1545730829238892,
      "learning_rate": 0.00016485957925978958,
      "loss": 3.5923,
      "step": 188490
    },
    {
      "epoch": 0.17571250079088102,
      "grad_norm": 0.8964565396308899,
      "learning_rate": 0.00016485771493769438,
      "loss": 3.3536,
      "step": 188500
    },
    {
      "epoch": 0.17572182240896012,
      "grad_norm": 1.054144024848938,
      "learning_rate": 0.00016485585061559916,
      "loss": 3.6663,
      "step": 188510
    },
    {
      "epoch": 0.1757311440270392,
      "grad_norm": 1.2245724201202393,
      "learning_rate": 0.00016485398629350397,
      "loss": 3.4842,
      "step": 188520
    },
    {
      "epoch": 0.1757404656451183,
      "grad_norm": 1.1560263633728027,
      "learning_rate": 0.00016485212197140875,
      "loss": 3.5591,
      "step": 188530
    },
    {
      "epoch": 0.1757497872631974,
      "grad_norm": 1.0913605690002441,
      "learning_rate": 0.00016485025764931356,
      "loss": 3.5009,
      "step": 188540
    },
    {
      "epoch": 0.1757591088812765,
      "grad_norm": 1.2311359643936157,
      "learning_rate": 0.0001648483933272184,
      "loss": 3.5482,
      "step": 188550
    },
    {
      "epoch": 0.1757684304993556,
      "grad_norm": 0.9876444339752197,
      "learning_rate": 0.00016484652900512317,
      "loss": 3.5,
      "step": 188560
    },
    {
      "epoch": 0.17577775211743468,
      "grad_norm": 0.9859775304794312,
      "learning_rate": 0.00016484466468302797,
      "loss": 3.518,
      "step": 188570
    },
    {
      "epoch": 0.17578707373551378,
      "grad_norm": 1.1851552724838257,
      "learning_rate": 0.00016484280036093275,
      "loss": 3.3939,
      "step": 188580
    },
    {
      "epoch": 0.17579639535359287,
      "grad_norm": 1.0663554668426514,
      "learning_rate": 0.00016484093603883756,
      "loss": 3.514,
      "step": 188590
    },
    {
      "epoch": 0.17580571697167197,
      "grad_norm": 1.1750737428665161,
      "learning_rate": 0.00016483907171674237,
      "loss": 3.4877,
      "step": 188600
    },
    {
      "epoch": 0.17581503858975103,
      "grad_norm": 0.97579425573349,
      "learning_rate": 0.00016483720739464717,
      "loss": 3.2606,
      "step": 188610
    },
    {
      "epoch": 0.17582436020783013,
      "grad_norm": 1.0449047088623047,
      "learning_rate": 0.00016483534307255198,
      "loss": 3.6207,
      "step": 188620
    },
    {
      "epoch": 0.17583368182590922,
      "grad_norm": 1.0486862659454346,
      "learning_rate": 0.00016483347875045676,
      "loss": 3.3843,
      "step": 188630
    },
    {
      "epoch": 0.17584300344398832,
      "grad_norm": 1.4510798454284668,
      "learning_rate": 0.00016483161442836156,
      "loss": 3.7103,
      "step": 188640
    },
    {
      "epoch": 0.1758523250620674,
      "grad_norm": 1.025452971458435,
      "learning_rate": 0.00016482975010626634,
      "loss": 3.5392,
      "step": 188650
    },
    {
      "epoch": 0.1758616466801465,
      "grad_norm": 1.418689250946045,
      "learning_rate": 0.00016482788578417118,
      "loss": 3.3784,
      "step": 188660
    },
    {
      "epoch": 0.1758709682982256,
      "grad_norm": 0.9924076199531555,
      "learning_rate": 0.00016482602146207598,
      "loss": 3.5686,
      "step": 188670
    },
    {
      "epoch": 0.1758802899163047,
      "grad_norm": 0.9900704622268677,
      "learning_rate": 0.00016482415713998076,
      "loss": 3.5155,
      "step": 188680
    },
    {
      "epoch": 0.1758896115343838,
      "grad_norm": 0.8944819569587708,
      "learning_rate": 0.00016482229281788557,
      "loss": 3.6846,
      "step": 188690
    },
    {
      "epoch": 0.17589893315246288,
      "grad_norm": 0.900047242641449,
      "learning_rate": 0.00016482042849579035,
      "loss": 3.5134,
      "step": 188700
    },
    {
      "epoch": 0.17590825477054195,
      "grad_norm": 1.0070971250534058,
      "learning_rate": 0.00016481856417369515,
      "loss": 3.4948,
      "step": 188710
    },
    {
      "epoch": 0.17591757638862104,
      "grad_norm": 1.800654649734497,
      "learning_rate": 0.0001648166998516,
      "loss": 3.3962,
      "step": 188720
    },
    {
      "epoch": 0.17592689800670014,
      "grad_norm": 1.0314462184906006,
      "learning_rate": 0.00016481483552950477,
      "loss": 3.5968,
      "step": 188730
    },
    {
      "epoch": 0.17593621962477923,
      "grad_norm": 1.0608446598052979,
      "learning_rate": 0.00016481297120740957,
      "loss": 3.6261,
      "step": 188740
    },
    {
      "epoch": 0.17594554124285833,
      "grad_norm": 1.2361375093460083,
      "learning_rate": 0.00016481110688531435,
      "loss": 3.5283,
      "step": 188750
    },
    {
      "epoch": 0.17595486286093742,
      "grad_norm": 0.9651065468788147,
      "learning_rate": 0.00016480924256321916,
      "loss": 3.6585,
      "step": 188760
    },
    {
      "epoch": 0.17596418447901652,
      "grad_norm": 1.356829047203064,
      "learning_rate": 0.00016480737824112396,
      "loss": 3.4367,
      "step": 188770
    },
    {
      "epoch": 0.1759735060970956,
      "grad_norm": 0.8638128638267517,
      "learning_rate": 0.00016480551391902877,
      "loss": 3.3987,
      "step": 188780
    },
    {
      "epoch": 0.1759828277151747,
      "grad_norm": 1.0740727186203003,
      "learning_rate": 0.00016480364959693358,
      "loss": 3.5565,
      "step": 188790
    },
    {
      "epoch": 0.1759921493332538,
      "grad_norm": 1.2744407653808594,
      "learning_rate": 0.00016480178527483836,
      "loss": 3.6387,
      "step": 188800
    },
    {
      "epoch": 0.1760014709513329,
      "grad_norm": 1.1676973104476929,
      "learning_rate": 0.00016479992095274316,
      "loss": 3.413,
      "step": 188810
    },
    {
      "epoch": 0.17601079256941196,
      "grad_norm": 0.9408824443817139,
      "learning_rate": 0.00016479805663064797,
      "loss": 3.7116,
      "step": 188820
    },
    {
      "epoch": 0.17602011418749106,
      "grad_norm": 0.8785749673843384,
      "learning_rate": 0.00016479619230855278,
      "loss": 3.5245,
      "step": 188830
    },
    {
      "epoch": 0.17602943580557015,
      "grad_norm": 0.9986494183540344,
      "learning_rate": 0.00016479432798645758,
      "loss": 3.5492,
      "step": 188840
    },
    {
      "epoch": 0.17603875742364924,
      "grad_norm": 1.2071681022644043,
      "learning_rate": 0.00016479246366436236,
      "loss": 3.4587,
      "step": 188850
    },
    {
      "epoch": 0.17604807904172834,
      "grad_norm": 1.1439862251281738,
      "learning_rate": 0.00016479059934226717,
      "loss": 3.5603,
      "step": 188860
    },
    {
      "epoch": 0.17605740065980743,
      "grad_norm": 1.2146188020706177,
      "learning_rate": 0.00016478873502017197,
      "loss": 3.2313,
      "step": 188870
    },
    {
      "epoch": 0.17606672227788653,
      "grad_norm": 1.1387282609939575,
      "learning_rate": 0.00016478687069807678,
      "loss": 3.512,
      "step": 188880
    },
    {
      "epoch": 0.17607604389596562,
      "grad_norm": 1.1620829105377197,
      "learning_rate": 0.00016478500637598159,
      "loss": 3.4764,
      "step": 188890
    },
    {
      "epoch": 0.17608536551404472,
      "grad_norm": 0.9581976532936096,
      "learning_rate": 0.00016478314205388637,
      "loss": 3.5542,
      "step": 188900
    },
    {
      "epoch": 0.1760946871321238,
      "grad_norm": 1.165504813194275,
      "learning_rate": 0.00016478127773179117,
      "loss": 3.4936,
      "step": 188910
    },
    {
      "epoch": 0.17610400875020288,
      "grad_norm": 0.996206521987915,
      "learning_rate": 0.00016477941340969598,
      "loss": 3.6867,
      "step": 188920
    },
    {
      "epoch": 0.17611333036828197,
      "grad_norm": 1.174667239189148,
      "learning_rate": 0.00016477754908760076,
      "loss": 3.3987,
      "step": 188930
    },
    {
      "epoch": 0.17612265198636107,
      "grad_norm": 0.7775231599807739,
      "learning_rate": 0.0001647756847655056,
      "loss": 3.6267,
      "step": 188940
    },
    {
      "epoch": 0.17613197360444016,
      "grad_norm": 1.0116814374923706,
      "learning_rate": 0.00016477382044341037,
      "loss": 3.4928,
      "step": 188950
    },
    {
      "epoch": 0.17614129522251926,
      "grad_norm": 1.1379408836364746,
      "learning_rate": 0.00016477195612131518,
      "loss": 3.5848,
      "step": 188960
    },
    {
      "epoch": 0.17615061684059835,
      "grad_norm": 1.0867055654525757,
      "learning_rate": 0.00016477009179921998,
      "loss": 3.6774,
      "step": 188970
    },
    {
      "epoch": 0.17615993845867745,
      "grad_norm": 1.189487338066101,
      "learning_rate": 0.00016476822747712476,
      "loss": 3.3342,
      "step": 188980
    },
    {
      "epoch": 0.17616926007675654,
      "grad_norm": 1.1104843616485596,
      "learning_rate": 0.00016476636315502957,
      "loss": 3.4102,
      "step": 188990
    },
    {
      "epoch": 0.17617858169483563,
      "grad_norm": 0.9844069480895996,
      "learning_rate": 0.00016476449883293437,
      "loss": 3.538,
      "step": 189000
    },
    {
      "epoch": 0.17618790331291473,
      "grad_norm": 1.2695682048797607,
      "learning_rate": 0.00016476263451083918,
      "loss": 3.4969,
      "step": 189010
    },
    {
      "epoch": 0.1761972249309938,
      "grad_norm": 1.2411775588989258,
      "learning_rate": 0.000164760770188744,
      "loss": 3.5886,
      "step": 189020
    },
    {
      "epoch": 0.1762065465490729,
      "grad_norm": 1.0742944478988647,
      "learning_rate": 0.00016475890586664877,
      "loss": 3.5739,
      "step": 189030
    },
    {
      "epoch": 0.17621586816715198,
      "grad_norm": 1.0885194540023804,
      "learning_rate": 0.00016475704154455357,
      "loss": 3.4533,
      "step": 189040
    },
    {
      "epoch": 0.17622518978523108,
      "grad_norm": 1.1499710083007812,
      "learning_rate": 0.00016475517722245838,
      "loss": 3.5191,
      "step": 189050
    },
    {
      "epoch": 0.17623451140331017,
      "grad_norm": 0.968062698841095,
      "learning_rate": 0.00016475331290036318,
      "loss": 3.7103,
      "step": 189060
    },
    {
      "epoch": 0.17624383302138927,
      "grad_norm": 0.9643270373344421,
      "learning_rate": 0.000164751448578268,
      "loss": 3.4928,
      "step": 189070
    },
    {
      "epoch": 0.17625315463946836,
      "grad_norm": 1.174912691116333,
      "learning_rate": 0.00016474958425617277,
      "loss": 3.5858,
      "step": 189080
    },
    {
      "epoch": 0.17626247625754746,
      "grad_norm": 0.9473608136177063,
      "learning_rate": 0.00016474771993407758,
      "loss": 3.7098,
      "step": 189090
    },
    {
      "epoch": 0.17627179787562655,
      "grad_norm": 0.9376760721206665,
      "learning_rate": 0.00016474585561198236,
      "loss": 3.5212,
      "step": 189100
    },
    {
      "epoch": 0.17628111949370565,
      "grad_norm": 0.9339778423309326,
      "learning_rate": 0.0001647439912898872,
      "loss": 3.4486,
      "step": 189110
    },
    {
      "epoch": 0.17629044111178474,
      "grad_norm": 1.3040691614151,
      "learning_rate": 0.000164742126967792,
      "loss": 3.3889,
      "step": 189120
    },
    {
      "epoch": 0.1762997627298638,
      "grad_norm": 0.9724531173706055,
      "learning_rate": 0.00016474026264569677,
      "loss": 3.5781,
      "step": 189130
    },
    {
      "epoch": 0.1763090843479429,
      "grad_norm": 1.0884010791778564,
      "learning_rate": 0.00016473839832360158,
      "loss": 3.6055,
      "step": 189140
    },
    {
      "epoch": 0.176318405966022,
      "grad_norm": 1.0510344505310059,
      "learning_rate": 0.00016473653400150636,
      "loss": 3.4394,
      "step": 189150
    },
    {
      "epoch": 0.1763277275841011,
      "grad_norm": 1.0814015865325928,
      "learning_rate": 0.00016473466967941117,
      "loss": 3.4669,
      "step": 189160
    },
    {
      "epoch": 0.17633704920218018,
      "grad_norm": 1.188492774963379,
      "learning_rate": 0.000164732805357316,
      "loss": 3.4135,
      "step": 189170
    },
    {
      "epoch": 0.17634637082025928,
      "grad_norm": 0.8937203884124756,
      "learning_rate": 0.00016473094103522078,
      "loss": 3.4987,
      "step": 189180
    },
    {
      "epoch": 0.17635569243833837,
      "grad_norm": 1.1891305446624756,
      "learning_rate": 0.00016472907671312559,
      "loss": 3.4698,
      "step": 189190
    },
    {
      "epoch": 0.17636501405641747,
      "grad_norm": 1.2019059658050537,
      "learning_rate": 0.00016472721239103036,
      "loss": 3.5597,
      "step": 189200
    },
    {
      "epoch": 0.17637433567449656,
      "grad_norm": 1.0849946737289429,
      "learning_rate": 0.00016472534806893517,
      "loss": 3.4913,
      "step": 189210
    },
    {
      "epoch": 0.17638365729257566,
      "grad_norm": 1.244524598121643,
      "learning_rate": 0.00016472348374683998,
      "loss": 3.5915,
      "step": 189220
    },
    {
      "epoch": 0.17639297891065472,
      "grad_norm": 0.9684139490127563,
      "learning_rate": 0.00016472161942474478,
      "loss": 3.5948,
      "step": 189230
    },
    {
      "epoch": 0.17640230052873382,
      "grad_norm": 1.1629703044891357,
      "learning_rate": 0.0001647197551026496,
      "loss": 3.4815,
      "step": 189240
    },
    {
      "epoch": 0.1764116221468129,
      "grad_norm": 1.280593991279602,
      "learning_rate": 0.00016471789078055437,
      "loss": 3.3892,
      "step": 189250
    },
    {
      "epoch": 0.176420943764892,
      "grad_norm": 0.8566908836364746,
      "learning_rate": 0.00016471602645845918,
      "loss": 3.4361,
      "step": 189260
    },
    {
      "epoch": 0.1764302653829711,
      "grad_norm": 0.8811302185058594,
      "learning_rate": 0.00016471416213636398,
      "loss": 3.4342,
      "step": 189270
    },
    {
      "epoch": 0.1764395870010502,
      "grad_norm": 1.1354506015777588,
      "learning_rate": 0.0001647122978142688,
      "loss": 3.4963,
      "step": 189280
    },
    {
      "epoch": 0.1764489086191293,
      "grad_norm": 0.9006388187408447,
      "learning_rate": 0.0001647104334921736,
      "loss": 3.5965,
      "step": 189290
    },
    {
      "epoch": 0.17645823023720839,
      "grad_norm": 1.2727307081222534,
      "learning_rate": 0.00016470856917007837,
      "loss": 3.533,
      "step": 189300
    },
    {
      "epoch": 0.17646755185528748,
      "grad_norm": 1.2260764837265015,
      "learning_rate": 0.00016470670484798318,
      "loss": 3.5333,
      "step": 189310
    },
    {
      "epoch": 0.17647687347336657,
      "grad_norm": 1.2037413120269775,
      "learning_rate": 0.00016470484052588799,
      "loss": 3.4967,
      "step": 189320
    },
    {
      "epoch": 0.17648619509144567,
      "grad_norm": 1.3473947048187256,
      "learning_rate": 0.0001647029762037928,
      "loss": 3.5326,
      "step": 189330
    },
    {
      "epoch": 0.17649551670952474,
      "grad_norm": 1.1466400623321533,
      "learning_rate": 0.0001647011118816976,
      "loss": 3.4688,
      "step": 189340
    },
    {
      "epoch": 0.17650483832760383,
      "grad_norm": 1.091577172279358,
      "learning_rate": 0.00016469924755960238,
      "loss": 3.4751,
      "step": 189350
    },
    {
      "epoch": 0.17651415994568292,
      "grad_norm": 0.9515365362167358,
      "learning_rate": 0.00016469738323750718,
      "loss": 3.5718,
      "step": 189360
    },
    {
      "epoch": 0.17652348156376202,
      "grad_norm": 0.8902846574783325,
      "learning_rate": 0.000164695518915412,
      "loss": 3.5462,
      "step": 189370
    },
    {
      "epoch": 0.1765328031818411,
      "grad_norm": 1.227294921875,
      "learning_rate": 0.00016469365459331677,
      "loss": 3.3217,
      "step": 189380
    },
    {
      "epoch": 0.1765421247999202,
      "grad_norm": 0.8591569662094116,
      "learning_rate": 0.0001646917902712216,
      "loss": 3.6168,
      "step": 189390
    },
    {
      "epoch": 0.1765514464179993,
      "grad_norm": 0.8767361044883728,
      "learning_rate": 0.00016468992594912638,
      "loss": 3.597,
      "step": 189400
    },
    {
      "epoch": 0.1765607680360784,
      "grad_norm": 1.1324007511138916,
      "learning_rate": 0.0001646880616270312,
      "loss": 3.4791,
      "step": 189410
    },
    {
      "epoch": 0.1765700896541575,
      "grad_norm": 1.0116338729858398,
      "learning_rate": 0.000164686197304936,
      "loss": 3.4661,
      "step": 189420
    },
    {
      "epoch": 0.17657941127223659,
      "grad_norm": 0.9598493576049805,
      "learning_rate": 0.00016468433298284077,
      "loss": 3.3902,
      "step": 189430
    },
    {
      "epoch": 0.17658873289031565,
      "grad_norm": 1.447163701057434,
      "learning_rate": 0.00016468246866074558,
      "loss": 3.5405,
      "step": 189440
    },
    {
      "epoch": 0.17659805450839475,
      "grad_norm": 0.7211007475852966,
      "learning_rate": 0.0001646806043386504,
      "loss": 3.7428,
      "step": 189450
    },
    {
      "epoch": 0.17660737612647384,
      "grad_norm": 1.0927766561508179,
      "learning_rate": 0.0001646787400165552,
      "loss": 3.4184,
      "step": 189460
    },
    {
      "epoch": 0.17661669774455294,
      "grad_norm": 0.9780182838439941,
      "learning_rate": 0.00016467687569446,
      "loss": 3.5662,
      "step": 189470
    },
    {
      "epoch": 0.17662601936263203,
      "grad_norm": 0.9656041264533997,
      "learning_rate": 0.00016467501137236478,
      "loss": 3.374,
      "step": 189480
    },
    {
      "epoch": 0.17663534098071113,
      "grad_norm": 0.867964506149292,
      "learning_rate": 0.00016467314705026959,
      "loss": 3.453,
      "step": 189490
    },
    {
      "epoch": 0.17664466259879022,
      "grad_norm": 1.3635388612747192,
      "learning_rate": 0.0001646712827281744,
      "loss": 3.5663,
      "step": 189500
    },
    {
      "epoch": 0.17665398421686931,
      "grad_norm": 1.4083096981048584,
      "learning_rate": 0.0001646694184060792,
      "loss": 3.4283,
      "step": 189510
    },
    {
      "epoch": 0.1766633058349484,
      "grad_norm": 1.1870132684707642,
      "learning_rate": 0.000164667554083984,
      "loss": 3.487,
      "step": 189520
    },
    {
      "epoch": 0.1766726274530275,
      "grad_norm": 1.1060328483581543,
      "learning_rate": 0.00016466568976188878,
      "loss": 3.6487,
      "step": 189530
    },
    {
      "epoch": 0.1766819490711066,
      "grad_norm": 1.022339105606079,
      "learning_rate": 0.0001646638254397936,
      "loss": 3.4488,
      "step": 189540
    },
    {
      "epoch": 0.17669127068918566,
      "grad_norm": 1.1704519987106323,
      "learning_rate": 0.00016466196111769837,
      "loss": 3.4816,
      "step": 189550
    },
    {
      "epoch": 0.17670059230726476,
      "grad_norm": 0.7347167730331421,
      "learning_rate": 0.0001646600967956032,
      "loss": 3.6797,
      "step": 189560
    },
    {
      "epoch": 0.17670991392534385,
      "grad_norm": 0.9868273138999939,
      "learning_rate": 0.000164658232473508,
      "loss": 3.3811,
      "step": 189570
    },
    {
      "epoch": 0.17671923554342295,
      "grad_norm": 1.1721376180648804,
      "learning_rate": 0.0001646563681514128,
      "loss": 3.5733,
      "step": 189580
    },
    {
      "epoch": 0.17672855716150204,
      "grad_norm": 0.9347448348999023,
      "learning_rate": 0.0001646545038293176,
      "loss": 3.4334,
      "step": 189590
    },
    {
      "epoch": 0.17673787877958114,
      "grad_norm": 1.2220953702926636,
      "learning_rate": 0.00016465263950722237,
      "loss": 3.4062,
      "step": 189600
    },
    {
      "epoch": 0.17674720039766023,
      "grad_norm": 1.0501925945281982,
      "learning_rate": 0.00016465077518512718,
      "loss": 3.5727,
      "step": 189610
    },
    {
      "epoch": 0.17675652201573933,
      "grad_norm": 1.0442991256713867,
      "learning_rate": 0.000164648910863032,
      "loss": 3.622,
      "step": 189620
    },
    {
      "epoch": 0.17676584363381842,
      "grad_norm": 0.9508198499679565,
      "learning_rate": 0.0001646470465409368,
      "loss": 3.3874,
      "step": 189630
    },
    {
      "epoch": 0.17677516525189751,
      "grad_norm": 1.1425234079360962,
      "learning_rate": 0.0001646451822188416,
      "loss": 3.5094,
      "step": 189640
    },
    {
      "epoch": 0.17678448686997658,
      "grad_norm": 1.1896575689315796,
      "learning_rate": 0.00016464331789674638,
      "loss": 3.3376,
      "step": 189650
    },
    {
      "epoch": 0.17679380848805568,
      "grad_norm": 0.8488199710845947,
      "learning_rate": 0.00016464145357465118,
      "loss": 3.5839,
      "step": 189660
    },
    {
      "epoch": 0.17680313010613477,
      "grad_norm": 1.144317388534546,
      "learning_rate": 0.000164639589252556,
      "loss": 3.4978,
      "step": 189670
    },
    {
      "epoch": 0.17681245172421386,
      "grad_norm": 0.9645643830299377,
      "learning_rate": 0.0001646377249304608,
      "loss": 3.5326,
      "step": 189680
    },
    {
      "epoch": 0.17682177334229296,
      "grad_norm": 1.1337672472000122,
      "learning_rate": 0.0001646358606083656,
      "loss": 3.4708,
      "step": 189690
    },
    {
      "epoch": 0.17683109496037205,
      "grad_norm": 1.0238689184188843,
      "learning_rate": 0.00016463399628627038,
      "loss": 3.7076,
      "step": 189700
    },
    {
      "epoch": 0.17684041657845115,
      "grad_norm": 1.205001950263977,
      "learning_rate": 0.0001646321319641752,
      "loss": 3.4727,
      "step": 189710
    },
    {
      "epoch": 0.17684973819653024,
      "grad_norm": 0.9631620049476624,
      "learning_rate": 0.00016463026764208,
      "loss": 3.6088,
      "step": 189720
    },
    {
      "epoch": 0.17685905981460934,
      "grad_norm": 1.1997641324996948,
      "learning_rate": 0.0001646284033199848,
      "loss": 3.5722,
      "step": 189730
    },
    {
      "epoch": 0.17686838143268843,
      "grad_norm": 1.0954434871673584,
      "learning_rate": 0.0001646265389978896,
      "loss": 3.4871,
      "step": 189740
    },
    {
      "epoch": 0.17687770305076753,
      "grad_norm": 3.735927104949951,
      "learning_rate": 0.0001646246746757944,
      "loss": 3.6042,
      "step": 189750
    },
    {
      "epoch": 0.1768870246688466,
      "grad_norm": 0.9984660148620605,
      "learning_rate": 0.0001646228103536992,
      "loss": 3.5441,
      "step": 189760
    },
    {
      "epoch": 0.1768963462869257,
      "grad_norm": 1.200547695159912,
      "learning_rate": 0.000164620946031604,
      "loss": 3.5665,
      "step": 189770
    },
    {
      "epoch": 0.17690566790500478,
      "grad_norm": 1.0756380558013916,
      "learning_rate": 0.0001646190817095088,
      "loss": 3.3922,
      "step": 189780
    },
    {
      "epoch": 0.17691498952308388,
      "grad_norm": 1.1013909578323364,
      "learning_rate": 0.0001646172173874136,
      "loss": 3.2951,
      "step": 189790
    },
    {
      "epoch": 0.17692431114116297,
      "grad_norm": 1.1320173740386963,
      "learning_rate": 0.0001646153530653184,
      "loss": 3.4008,
      "step": 189800
    },
    {
      "epoch": 0.17693363275924207,
      "grad_norm": 1.082436203956604,
      "learning_rate": 0.0001646134887432232,
      "loss": 3.7358,
      "step": 189810
    },
    {
      "epoch": 0.17694295437732116,
      "grad_norm": 1.1866772174835205,
      "learning_rate": 0.000164611624421128,
      "loss": 3.5535,
      "step": 189820
    },
    {
      "epoch": 0.17695227599540025,
      "grad_norm": 1.166903018951416,
      "learning_rate": 0.00016460976009903278,
      "loss": 3.4854,
      "step": 189830
    },
    {
      "epoch": 0.17696159761347935,
      "grad_norm": 1.0401279926300049,
      "learning_rate": 0.00016460789577693762,
      "loss": 3.444,
      "step": 189840
    },
    {
      "epoch": 0.17697091923155844,
      "grad_norm": 1.3958097696304321,
      "learning_rate": 0.0001646060314548424,
      "loss": 3.4063,
      "step": 189850
    },
    {
      "epoch": 0.1769802408496375,
      "grad_norm": 1.0581704378128052,
      "learning_rate": 0.0001646041671327472,
      "loss": 3.3905,
      "step": 189860
    },
    {
      "epoch": 0.1769895624677166,
      "grad_norm": 1.202903151512146,
      "learning_rate": 0.000164602302810652,
      "loss": 3.5215,
      "step": 189870
    },
    {
      "epoch": 0.1769988840857957,
      "grad_norm": 1.1860233545303345,
      "learning_rate": 0.0001646004384885568,
      "loss": 3.8089,
      "step": 189880
    },
    {
      "epoch": 0.1770082057038748,
      "grad_norm": 1.2474716901779175,
      "learning_rate": 0.0001645985741664616,
      "loss": 3.4159,
      "step": 189890
    },
    {
      "epoch": 0.1770175273219539,
      "grad_norm": 1.1332831382751465,
      "learning_rate": 0.0001645967098443664,
      "loss": 3.3871,
      "step": 189900
    },
    {
      "epoch": 0.17702684894003298,
      "grad_norm": 1.1242167949676514,
      "learning_rate": 0.0001645948455222712,
      "loss": 3.6921,
      "step": 189910
    },
    {
      "epoch": 0.17703617055811208,
      "grad_norm": 0.9876261353492737,
      "learning_rate": 0.000164592981200176,
      "loss": 3.3492,
      "step": 189920
    },
    {
      "epoch": 0.17704549217619117,
      "grad_norm": 1.011749505996704,
      "learning_rate": 0.0001645911168780808,
      "loss": 3.421,
      "step": 189930
    },
    {
      "epoch": 0.17705481379427027,
      "grad_norm": 1.3183048963546753,
      "learning_rate": 0.0001645892525559856,
      "loss": 3.612,
      "step": 189940
    },
    {
      "epoch": 0.17706413541234936,
      "grad_norm": 1.4091988801956177,
      "learning_rate": 0.0001645873882338904,
      "loss": 3.5778,
      "step": 189950
    },
    {
      "epoch": 0.17707345703042845,
      "grad_norm": 1.2159940004348755,
      "learning_rate": 0.0001645855239117952,
      "loss": 3.4277,
      "step": 189960
    },
    {
      "epoch": 0.17708277864850752,
      "grad_norm": 1.0430548191070557,
      "learning_rate": 0.00016458365958970002,
      "loss": 3.4938,
      "step": 189970
    },
    {
      "epoch": 0.17709210026658662,
      "grad_norm": 0.9960682988166809,
      "learning_rate": 0.0001645817952676048,
      "loss": 3.603,
      "step": 189980
    },
    {
      "epoch": 0.1771014218846657,
      "grad_norm": 1.0948207378387451,
      "learning_rate": 0.0001645799309455096,
      "loss": 3.4763,
      "step": 189990
    },
    {
      "epoch": 0.1771107435027448,
      "grad_norm": 1.1016170978546143,
      "learning_rate": 0.0001645780666234144,
      "loss": 3.5549,
      "step": 190000
    },
    {
      "epoch": 0.1771200651208239,
      "grad_norm": 1.228987693786621,
      "learning_rate": 0.00016457620230131922,
      "loss": 3.6025,
      "step": 190010
    },
    {
      "epoch": 0.177129386738903,
      "grad_norm": 1.179368019104004,
      "learning_rate": 0.000164574337979224,
      "loss": 3.6181,
      "step": 190020
    },
    {
      "epoch": 0.1771387083569821,
      "grad_norm": 1.2059975862503052,
      "learning_rate": 0.0001645724736571288,
      "loss": 3.5028,
      "step": 190030
    },
    {
      "epoch": 0.17714802997506118,
      "grad_norm": 1.1641154289245605,
      "learning_rate": 0.0001645706093350336,
      "loss": 3.6223,
      "step": 190040
    },
    {
      "epoch": 0.17715735159314028,
      "grad_norm": 1.0892833471298218,
      "learning_rate": 0.00016456874501293839,
      "loss": 3.5197,
      "step": 190050
    },
    {
      "epoch": 0.17716667321121937,
      "grad_norm": 0.9815354347229004,
      "learning_rate": 0.00016456688069084322,
      "loss": 3.4779,
      "step": 190060
    },
    {
      "epoch": 0.17717599482929844,
      "grad_norm": 1.1677404642105103,
      "learning_rate": 0.000164565016368748,
      "loss": 3.5988,
      "step": 190070
    },
    {
      "epoch": 0.17718531644737753,
      "grad_norm": 1.205916166305542,
      "learning_rate": 0.0001645631520466528,
      "loss": 3.6999,
      "step": 190080
    },
    {
      "epoch": 0.17719463806545663,
      "grad_norm": 0.8261817693710327,
      "learning_rate": 0.0001645612877245576,
      "loss": 3.5024,
      "step": 190090
    },
    {
      "epoch": 0.17720395968353572,
      "grad_norm": 1.0516517162322998,
      "learning_rate": 0.0001645594234024624,
      "loss": 3.4956,
      "step": 190100
    },
    {
      "epoch": 0.17721328130161482,
      "grad_norm": 0.8587791323661804,
      "learning_rate": 0.0001645575590803672,
      "loss": 3.5612,
      "step": 190110
    },
    {
      "epoch": 0.1772226029196939,
      "grad_norm": 1.091942310333252,
      "learning_rate": 0.000164555694758272,
      "loss": 3.5202,
      "step": 190120
    },
    {
      "epoch": 0.177231924537773,
      "grad_norm": 0.90034419298172,
      "learning_rate": 0.0001645538304361768,
      "loss": 3.355,
      "step": 190130
    },
    {
      "epoch": 0.1772412461558521,
      "grad_norm": 1.1309988498687744,
      "learning_rate": 0.00016455196611408162,
      "loss": 3.5472,
      "step": 190140
    },
    {
      "epoch": 0.1772505677739312,
      "grad_norm": 1.290824055671692,
      "learning_rate": 0.0001645501017919864,
      "loss": 3.5117,
      "step": 190150
    },
    {
      "epoch": 0.1772598893920103,
      "grad_norm": 0.872967004776001,
      "learning_rate": 0.0001645482374698912,
      "loss": 3.2658,
      "step": 190160
    },
    {
      "epoch": 0.17726921101008938,
      "grad_norm": 1.1541889905929565,
      "learning_rate": 0.000164546373147796,
      "loss": 3.3723,
      "step": 190170
    },
    {
      "epoch": 0.17727853262816845,
      "grad_norm": 1.209230899810791,
      "learning_rate": 0.00016454450882570081,
      "loss": 3.612,
      "step": 190180
    },
    {
      "epoch": 0.17728785424624754,
      "grad_norm": 1.1162632703781128,
      "learning_rate": 0.00016454264450360562,
      "loss": 3.4734,
      "step": 190190
    },
    {
      "epoch": 0.17729717586432664,
      "grad_norm": 0.7872020602226257,
      "learning_rate": 0.0001645407801815104,
      "loss": 3.7549,
      "step": 190200
    },
    {
      "epoch": 0.17730649748240573,
      "grad_norm": 0.9576411843299866,
      "learning_rate": 0.0001645389158594152,
      "loss": 3.7516,
      "step": 190210
    },
    {
      "epoch": 0.17731581910048483,
      "grad_norm": 0.9897026419639587,
      "learning_rate": 0.00016453705153731998,
      "loss": 3.5057,
      "step": 190220
    },
    {
      "epoch": 0.17732514071856392,
      "grad_norm": 0.9767795205116272,
      "learning_rate": 0.00016453518721522482,
      "loss": 3.4262,
      "step": 190230
    },
    {
      "epoch": 0.17733446233664302,
      "grad_norm": 1.0741750001907349,
      "learning_rate": 0.00016453332289312962,
      "loss": 3.471,
      "step": 190240
    },
    {
      "epoch": 0.1773437839547221,
      "grad_norm": 1.0633389949798584,
      "learning_rate": 0.0001645314585710344,
      "loss": 3.6551,
      "step": 190250
    },
    {
      "epoch": 0.1773531055728012,
      "grad_norm": 1.0783401727676392,
      "learning_rate": 0.0001645295942489392,
      "loss": 3.2736,
      "step": 190260
    },
    {
      "epoch": 0.1773624271908803,
      "grad_norm": 1.2018500566482544,
      "learning_rate": 0.000164527729926844,
      "loss": 3.5467,
      "step": 190270
    },
    {
      "epoch": 0.17737174880895937,
      "grad_norm": 1.5692428350448608,
      "learning_rate": 0.0001645258656047488,
      "loss": 3.5485,
      "step": 190280
    },
    {
      "epoch": 0.17738107042703846,
      "grad_norm": 1.0765987634658813,
      "learning_rate": 0.00016452400128265363,
      "loss": 3.5588,
      "step": 190290
    },
    {
      "epoch": 0.17739039204511756,
      "grad_norm": 0.9568713903427124,
      "learning_rate": 0.0001645221369605584,
      "loss": 3.8271,
      "step": 190300
    },
    {
      "epoch": 0.17739971366319665,
      "grad_norm": 1.2216613292694092,
      "learning_rate": 0.00016452027263846321,
      "loss": 3.5881,
      "step": 190310
    },
    {
      "epoch": 0.17740903528127575,
      "grad_norm": 1.1283332109451294,
      "learning_rate": 0.000164518408316368,
      "loss": 3.4714,
      "step": 190320
    },
    {
      "epoch": 0.17741835689935484,
      "grad_norm": 1.1033971309661865,
      "learning_rate": 0.0001645165439942728,
      "loss": 3.5114,
      "step": 190330
    },
    {
      "epoch": 0.17742767851743393,
      "grad_norm": 0.94484543800354,
      "learning_rate": 0.0001645146796721776,
      "loss": 3.5044,
      "step": 190340
    },
    {
      "epoch": 0.17743700013551303,
      "grad_norm": 1.066983699798584,
      "learning_rate": 0.0001645128153500824,
      "loss": 3.4573,
      "step": 190350
    },
    {
      "epoch": 0.17744632175359212,
      "grad_norm": 1.2116944789886475,
      "learning_rate": 0.00016451095102798722,
      "loss": 3.3054,
      "step": 190360
    },
    {
      "epoch": 0.17745564337167122,
      "grad_norm": 1.0637199878692627,
      "learning_rate": 0.000164509086705892,
      "loss": 3.6322,
      "step": 190370
    },
    {
      "epoch": 0.17746496498975028,
      "grad_norm": 1.0312674045562744,
      "learning_rate": 0.0001645072223837968,
      "loss": 3.5786,
      "step": 190380
    },
    {
      "epoch": 0.17747428660782938,
      "grad_norm": 1.0049747228622437,
      "learning_rate": 0.0001645053580617016,
      "loss": 3.4686,
      "step": 190390
    },
    {
      "epoch": 0.17748360822590847,
      "grad_norm": 1.4857614040374756,
      "learning_rate": 0.00016450349373960642,
      "loss": 3.5771,
      "step": 190400
    },
    {
      "epoch": 0.17749292984398757,
      "grad_norm": 1.3263835906982422,
      "learning_rate": 0.00016450162941751122,
      "loss": 3.3411,
      "step": 190410
    },
    {
      "epoch": 0.17750225146206666,
      "grad_norm": 0.9746300578117371,
      "learning_rate": 0.000164499765095416,
      "loss": 3.4799,
      "step": 190420
    },
    {
      "epoch": 0.17751157308014576,
      "grad_norm": 1.1680183410644531,
      "learning_rate": 0.0001644979007733208,
      "loss": 3.5246,
      "step": 190430
    },
    {
      "epoch": 0.17752089469822485,
      "grad_norm": 1.2132943868637085,
      "learning_rate": 0.00016449603645122562,
      "loss": 3.4346,
      "step": 190440
    },
    {
      "epoch": 0.17753021631630395,
      "grad_norm": 0.9990139007568359,
      "learning_rate": 0.00016449417212913042,
      "loss": 3.6271,
      "step": 190450
    },
    {
      "epoch": 0.17753953793438304,
      "grad_norm": 0.9216272830963135,
      "learning_rate": 0.00016449230780703523,
      "loss": 3.558,
      "step": 190460
    },
    {
      "epoch": 0.17754885955246213,
      "grad_norm": 1.0547075271606445,
      "learning_rate": 0.00016449044348494,
      "loss": 3.5061,
      "step": 190470
    },
    {
      "epoch": 0.17755818117054123,
      "grad_norm": 1.1499453783035278,
      "learning_rate": 0.0001644885791628448,
      "loss": 3.4968,
      "step": 190480
    },
    {
      "epoch": 0.1775675027886203,
      "grad_norm": 1.0889804363250732,
      "learning_rate": 0.00016448671484074962,
      "loss": 3.539,
      "step": 190490
    },
    {
      "epoch": 0.1775768244066994,
      "grad_norm": 1.2214367389678955,
      "learning_rate": 0.0001644848505186544,
      "loss": 3.6858,
      "step": 190500
    },
    {
      "epoch": 0.17758614602477849,
      "grad_norm": 1.1397086381912231,
      "learning_rate": 0.00016448298619655923,
      "loss": 3.5174,
      "step": 190510
    },
    {
      "epoch": 0.17759546764285758,
      "grad_norm": 0.801450788974762,
      "learning_rate": 0.000164481121874464,
      "loss": 3.5091,
      "step": 190520
    },
    {
      "epoch": 0.17760478926093667,
      "grad_norm": 0.9675693511962891,
      "learning_rate": 0.00016447925755236882,
      "loss": 3.4549,
      "step": 190530
    },
    {
      "epoch": 0.17761411087901577,
      "grad_norm": 1.0442235469818115,
      "learning_rate": 0.00016447739323027362,
      "loss": 3.364,
      "step": 190540
    },
    {
      "epoch": 0.17762343249709486,
      "grad_norm": 0.9901760816574097,
      "learning_rate": 0.0001644755289081784,
      "loss": 3.3653,
      "step": 190550
    },
    {
      "epoch": 0.17763275411517396,
      "grad_norm": 1.1602578163146973,
      "learning_rate": 0.0001644736645860832,
      "loss": 3.539,
      "step": 190560
    },
    {
      "epoch": 0.17764207573325305,
      "grad_norm": 1.2682671546936035,
      "learning_rate": 0.00016447180026398802,
      "loss": 3.2835,
      "step": 190570
    },
    {
      "epoch": 0.17765139735133215,
      "grad_norm": 0.8737831711769104,
      "learning_rate": 0.00016446993594189282,
      "loss": 3.4114,
      "step": 190580
    },
    {
      "epoch": 0.1776607189694112,
      "grad_norm": 0.8989237546920776,
      "learning_rate": 0.00016446807161979763,
      "loss": 3.4983,
      "step": 190590
    },
    {
      "epoch": 0.1776700405874903,
      "grad_norm": 2.787346601486206,
      "learning_rate": 0.0001644662072977024,
      "loss": 3.4363,
      "step": 190600
    },
    {
      "epoch": 0.1776793622055694,
      "grad_norm": 1.1150931119918823,
      "learning_rate": 0.00016446434297560721,
      "loss": 3.4463,
      "step": 190610
    },
    {
      "epoch": 0.1776886838236485,
      "grad_norm": 1.0679199695587158,
      "learning_rate": 0.00016446247865351202,
      "loss": 3.6916,
      "step": 190620
    },
    {
      "epoch": 0.1776980054417276,
      "grad_norm": 1.0622749328613281,
      "learning_rate": 0.00016446061433141683,
      "loss": 3.6592,
      "step": 190630
    },
    {
      "epoch": 0.17770732705980669,
      "grad_norm": 1.0481127500534058,
      "learning_rate": 0.00016445875000932163,
      "loss": 3.5628,
      "step": 190640
    },
    {
      "epoch": 0.17771664867788578,
      "grad_norm": 1.179555892944336,
      "learning_rate": 0.0001644568856872264,
      "loss": 3.5679,
      "step": 190650
    },
    {
      "epoch": 0.17772597029596487,
      "grad_norm": 1.4182243347167969,
      "learning_rate": 0.00016445502136513122,
      "loss": 3.5086,
      "step": 190660
    },
    {
      "epoch": 0.17773529191404397,
      "grad_norm": 1.0942155122756958,
      "learning_rate": 0.000164453157043036,
      "loss": 3.5604,
      "step": 190670
    },
    {
      "epoch": 0.17774461353212306,
      "grad_norm": 1.0562191009521484,
      "learning_rate": 0.00016445129272094083,
      "loss": 3.5784,
      "step": 190680
    },
    {
      "epoch": 0.17775393515020216,
      "grad_norm": 1.0687860250473022,
      "learning_rate": 0.00016444942839884564,
      "loss": 3.5075,
      "step": 190690
    },
    {
      "epoch": 0.17776325676828122,
      "grad_norm": 1.0700618028640747,
      "learning_rate": 0.00016444756407675042,
      "loss": 3.4397,
      "step": 190700
    },
    {
      "epoch": 0.17777257838636032,
      "grad_norm": 1.0719406604766846,
      "learning_rate": 0.00016444569975465522,
      "loss": 3.4224,
      "step": 190710
    },
    {
      "epoch": 0.1777819000044394,
      "grad_norm": 0.9400117993354797,
      "learning_rate": 0.00016444383543256,
      "loss": 3.4458,
      "step": 190720
    },
    {
      "epoch": 0.1777912216225185,
      "grad_norm": 0.8315281867980957,
      "learning_rate": 0.0001644419711104648,
      "loss": 3.5738,
      "step": 190730
    },
    {
      "epoch": 0.1778005432405976,
      "grad_norm": 1.0410797595977783,
      "learning_rate": 0.00016444010678836964,
      "loss": 3.525,
      "step": 190740
    },
    {
      "epoch": 0.1778098648586767,
      "grad_norm": 1.2029194831848145,
      "learning_rate": 0.00016443824246627442,
      "loss": 3.4767,
      "step": 190750
    },
    {
      "epoch": 0.1778191864767558,
      "grad_norm": 1.0029183626174927,
      "learning_rate": 0.00016443637814417923,
      "loss": 3.6218,
      "step": 190760
    },
    {
      "epoch": 0.1778285080948349,
      "grad_norm": 1.0936561822891235,
      "learning_rate": 0.000164434513822084,
      "loss": 3.4756,
      "step": 190770
    },
    {
      "epoch": 0.17783782971291398,
      "grad_norm": 1.0227205753326416,
      "learning_rate": 0.0001644326494999888,
      "loss": 3.6424,
      "step": 190780
    },
    {
      "epoch": 0.17784715133099308,
      "grad_norm": 1.0050402879714966,
      "learning_rate": 0.00016443078517789362,
      "loss": 3.5634,
      "step": 190790
    },
    {
      "epoch": 0.17785647294907214,
      "grad_norm": 1.0572654008865356,
      "learning_rate": 0.00016442892085579843,
      "loss": 3.2555,
      "step": 190800
    },
    {
      "epoch": 0.17786579456715124,
      "grad_norm": 1.2111432552337646,
      "learning_rate": 0.00016442705653370323,
      "loss": 3.5062,
      "step": 190810
    },
    {
      "epoch": 0.17787511618523033,
      "grad_norm": 1.0626485347747803,
      "learning_rate": 0.000164425192211608,
      "loss": 3.5803,
      "step": 190820
    },
    {
      "epoch": 0.17788443780330943,
      "grad_norm": 1.1645585298538208,
      "learning_rate": 0.00016442332788951282,
      "loss": 3.6426,
      "step": 190830
    },
    {
      "epoch": 0.17789375942138852,
      "grad_norm": 0.8421915173530579,
      "learning_rate": 0.00016442146356741762,
      "loss": 3.4644,
      "step": 190840
    },
    {
      "epoch": 0.17790308103946761,
      "grad_norm": 0.9604653716087341,
      "learning_rate": 0.00016441959924532243,
      "loss": 3.7127,
      "step": 190850
    },
    {
      "epoch": 0.1779124026575467,
      "grad_norm": 1.1451066732406616,
      "learning_rate": 0.00016441773492322724,
      "loss": 3.5171,
      "step": 190860
    },
    {
      "epoch": 0.1779217242756258,
      "grad_norm": 0.970163106918335,
      "learning_rate": 0.00016441587060113202,
      "loss": 3.3858,
      "step": 190870
    },
    {
      "epoch": 0.1779310458937049,
      "grad_norm": 0.8744580149650574,
      "learning_rate": 0.00016441400627903682,
      "loss": 3.4383,
      "step": 190880
    },
    {
      "epoch": 0.177940367511784,
      "grad_norm": 1.0958815813064575,
      "learning_rate": 0.00016441214195694163,
      "loss": 3.5031,
      "step": 190890
    },
    {
      "epoch": 0.1779496891298631,
      "grad_norm": 1.1244040727615356,
      "learning_rate": 0.00016441027763484643,
      "loss": 3.7469,
      "step": 190900
    },
    {
      "epoch": 0.17795901074794215,
      "grad_norm": 1.2854381799697876,
      "learning_rate": 0.00016440841331275124,
      "loss": 3.5691,
      "step": 190910
    },
    {
      "epoch": 0.17796833236602125,
      "grad_norm": 0.9707404971122742,
      "learning_rate": 0.00016440654899065602,
      "loss": 3.4999,
      "step": 190920
    },
    {
      "epoch": 0.17797765398410034,
      "grad_norm": 1.1199885606765747,
      "learning_rate": 0.00016440468466856083,
      "loss": 3.4813,
      "step": 190930
    },
    {
      "epoch": 0.17798697560217944,
      "grad_norm": 0.9177330136299133,
      "learning_rate": 0.00016440282034646563,
      "loss": 3.6229,
      "step": 190940
    },
    {
      "epoch": 0.17799629722025853,
      "grad_norm": 0.8993372321128845,
      "learning_rate": 0.0001644009560243704,
      "loss": 3.5672,
      "step": 190950
    },
    {
      "epoch": 0.17800561883833763,
      "grad_norm": 1.0850073099136353,
      "learning_rate": 0.00016439909170227525,
      "loss": 3.5281,
      "step": 190960
    },
    {
      "epoch": 0.17801494045641672,
      "grad_norm": 1.010055422782898,
      "learning_rate": 0.00016439722738018002,
      "loss": 3.6751,
      "step": 190970
    },
    {
      "epoch": 0.17802426207449581,
      "grad_norm": 1.0008597373962402,
      "learning_rate": 0.00016439536305808483,
      "loss": 3.4853,
      "step": 190980
    },
    {
      "epoch": 0.1780335836925749,
      "grad_norm": 1.0438154935836792,
      "learning_rate": 0.00016439349873598964,
      "loss": 3.6389,
      "step": 190990
    },
    {
      "epoch": 0.178042905310654,
      "grad_norm": 1.0476818084716797,
      "learning_rate": 0.00016439163441389442,
      "loss": 3.5174,
      "step": 191000
    },
    {
      "epoch": 0.17805222692873307,
      "grad_norm": 0.8113012313842773,
      "learning_rate": 0.00016438977009179922,
      "loss": 3.5024,
      "step": 191010
    },
    {
      "epoch": 0.17806154854681217,
      "grad_norm": 1.08038330078125,
      "learning_rate": 0.00016438790576970403,
      "loss": 3.4829,
      "step": 191020
    },
    {
      "epoch": 0.17807087016489126,
      "grad_norm": 1.378386378288269,
      "learning_rate": 0.00016438604144760884,
      "loss": 3.2914,
      "step": 191030
    },
    {
      "epoch": 0.17808019178297035,
      "grad_norm": 0.885749340057373,
      "learning_rate": 0.00016438417712551364,
      "loss": 3.6991,
      "step": 191040
    },
    {
      "epoch": 0.17808951340104945,
      "grad_norm": 1.0726573467254639,
      "learning_rate": 0.00016438231280341842,
      "loss": 3.6099,
      "step": 191050
    },
    {
      "epoch": 0.17809883501912854,
      "grad_norm": 1.1219779253005981,
      "learning_rate": 0.00016438044848132323,
      "loss": 3.597,
      "step": 191060
    },
    {
      "epoch": 0.17810815663720764,
      "grad_norm": 1.0687382221221924,
      "learning_rate": 0.00016437858415922803,
      "loss": 3.41,
      "step": 191070
    },
    {
      "epoch": 0.17811747825528673,
      "grad_norm": 1.0519509315490723,
      "learning_rate": 0.00016437671983713284,
      "loss": 3.3648,
      "step": 191080
    },
    {
      "epoch": 0.17812679987336583,
      "grad_norm": 1.2203469276428223,
      "learning_rate": 0.00016437485551503765,
      "loss": 3.4856,
      "step": 191090
    },
    {
      "epoch": 0.17813612149144492,
      "grad_norm": 1.1274312734603882,
      "learning_rate": 0.00016437299119294243,
      "loss": 3.5472,
      "step": 191100
    },
    {
      "epoch": 0.17814544310952402,
      "grad_norm": 1.1014153957366943,
      "learning_rate": 0.00016437112687084723,
      "loss": 3.5755,
      "step": 191110
    },
    {
      "epoch": 0.17815476472760308,
      "grad_norm": 1.0044251680374146,
      "learning_rate": 0.000164369262548752,
      "loss": 3.4555,
      "step": 191120
    },
    {
      "epoch": 0.17816408634568218,
      "grad_norm": 1.1897739171981812,
      "learning_rate": 0.00016436739822665684,
      "loss": 3.4917,
      "step": 191130
    },
    {
      "epoch": 0.17817340796376127,
      "grad_norm": 1.1969090700149536,
      "learning_rate": 0.00016436553390456165,
      "loss": 3.4785,
      "step": 191140
    },
    {
      "epoch": 0.17818272958184037,
      "grad_norm": 0.7955862879753113,
      "learning_rate": 0.00016436366958246643,
      "loss": 3.5449,
      "step": 191150
    },
    {
      "epoch": 0.17819205119991946,
      "grad_norm": 1.1320631504058838,
      "learning_rate": 0.00016436180526037124,
      "loss": 3.6022,
      "step": 191160
    },
    {
      "epoch": 0.17820137281799855,
      "grad_norm": 0.9760259985923767,
      "learning_rate": 0.00016435994093827601,
      "loss": 3.517,
      "step": 191170
    },
    {
      "epoch": 0.17821069443607765,
      "grad_norm": 1.0143131017684937,
      "learning_rate": 0.00016435807661618082,
      "loss": 3.4472,
      "step": 191180
    },
    {
      "epoch": 0.17822001605415674,
      "grad_norm": 0.8554642200469971,
      "learning_rate": 0.00016435621229408565,
      "loss": 3.5313,
      "step": 191190
    },
    {
      "epoch": 0.17822933767223584,
      "grad_norm": 1.130143642425537,
      "learning_rate": 0.00016435434797199043,
      "loss": 3.4112,
      "step": 191200
    },
    {
      "epoch": 0.17823865929031493,
      "grad_norm": 1.3035281896591187,
      "learning_rate": 0.00016435248364989524,
      "loss": 3.5959,
      "step": 191210
    },
    {
      "epoch": 0.178247980908394,
      "grad_norm": 1.053883671760559,
      "learning_rate": 0.00016435061932780002,
      "loss": 3.6355,
      "step": 191220
    },
    {
      "epoch": 0.1782573025264731,
      "grad_norm": 1.0456217527389526,
      "learning_rate": 0.00016434875500570483,
      "loss": 3.5605,
      "step": 191230
    },
    {
      "epoch": 0.1782666241445522,
      "grad_norm": 1.0482776165008545,
      "learning_rate": 0.00016434689068360963,
      "loss": 3.7034,
      "step": 191240
    },
    {
      "epoch": 0.17827594576263128,
      "grad_norm": 1.066895842552185,
      "learning_rate": 0.00016434502636151444,
      "loss": 3.458,
      "step": 191250
    },
    {
      "epoch": 0.17828526738071038,
      "grad_norm": 1.0950467586517334,
      "learning_rate": 0.00016434316203941924,
      "loss": 3.5899,
      "step": 191260
    },
    {
      "epoch": 0.17829458899878947,
      "grad_norm": 0.9107535481452942,
      "learning_rate": 0.00016434129771732402,
      "loss": 3.7226,
      "step": 191270
    },
    {
      "epoch": 0.17830391061686857,
      "grad_norm": 1.092189908027649,
      "learning_rate": 0.00016433943339522883,
      "loss": 3.4578,
      "step": 191280
    },
    {
      "epoch": 0.17831323223494766,
      "grad_norm": 1.0252755880355835,
      "learning_rate": 0.00016433756907313364,
      "loss": 3.6242,
      "step": 191290
    },
    {
      "epoch": 0.17832255385302676,
      "grad_norm": 1.1623834371566772,
      "learning_rate": 0.00016433570475103844,
      "loss": 3.4368,
      "step": 191300
    },
    {
      "epoch": 0.17833187547110585,
      "grad_norm": 1.0543776750564575,
      "learning_rate": 0.00016433384042894325,
      "loss": 3.5019,
      "step": 191310
    },
    {
      "epoch": 0.17834119708918494,
      "grad_norm": 0.9116623997688293,
      "learning_rate": 0.00016433197610684803,
      "loss": 3.5753,
      "step": 191320
    },
    {
      "epoch": 0.178350518707264,
      "grad_norm": 1.002020239830017,
      "learning_rate": 0.00016433011178475283,
      "loss": 3.5245,
      "step": 191330
    },
    {
      "epoch": 0.1783598403253431,
      "grad_norm": 1.0738184452056885,
      "learning_rate": 0.00016432824746265764,
      "loss": 3.5868,
      "step": 191340
    },
    {
      "epoch": 0.1783691619434222,
      "grad_norm": 1.0603009462356567,
      "learning_rate": 0.00016432638314056245,
      "loss": 3.3306,
      "step": 191350
    },
    {
      "epoch": 0.1783784835615013,
      "grad_norm": 0.9842936396598816,
      "learning_rate": 0.00016432451881846725,
      "loss": 3.4853,
      "step": 191360
    },
    {
      "epoch": 0.1783878051795804,
      "grad_norm": 1.0422612428665161,
      "learning_rate": 0.00016432265449637203,
      "loss": 3.3827,
      "step": 191370
    },
    {
      "epoch": 0.17839712679765948,
      "grad_norm": 1.0950440168380737,
      "learning_rate": 0.00016432079017427684,
      "loss": 3.4374,
      "step": 191380
    },
    {
      "epoch": 0.17840644841573858,
      "grad_norm": 1.0455799102783203,
      "learning_rate": 0.00016431892585218162,
      "loss": 3.6483,
      "step": 191390
    },
    {
      "epoch": 0.17841577003381767,
      "grad_norm": 0.899624228477478,
      "learning_rate": 0.00016431706153008642,
      "loss": 3.5225,
      "step": 191400
    },
    {
      "epoch": 0.17842509165189677,
      "grad_norm": 1.0353723764419556,
      "learning_rate": 0.00016431519720799126,
      "loss": 3.5145,
      "step": 191410
    },
    {
      "epoch": 0.17843441326997586,
      "grad_norm": 0.9517636895179749,
      "learning_rate": 0.00016431333288589604,
      "loss": 3.6145,
      "step": 191420
    },
    {
      "epoch": 0.17844373488805493,
      "grad_norm": 1.1147962808609009,
      "learning_rate": 0.00016431146856380084,
      "loss": 3.5015,
      "step": 191430
    },
    {
      "epoch": 0.17845305650613402,
      "grad_norm": 1.4045782089233398,
      "learning_rate": 0.00016430960424170562,
      "loss": 3.3949,
      "step": 191440
    },
    {
      "epoch": 0.17846237812421312,
      "grad_norm": 0.989236056804657,
      "learning_rate": 0.00016430773991961043,
      "loss": 3.4619,
      "step": 191450
    },
    {
      "epoch": 0.1784716997422922,
      "grad_norm": 0.9678468704223633,
      "learning_rate": 0.00016430587559751524,
      "loss": 3.6363,
      "step": 191460
    },
    {
      "epoch": 0.1784810213603713,
      "grad_norm": 1.1349674463272095,
      "learning_rate": 0.00016430401127542004,
      "loss": 3.6287,
      "step": 191470
    },
    {
      "epoch": 0.1784903429784504,
      "grad_norm": 0.9584408402442932,
      "learning_rate": 0.00016430214695332485,
      "loss": 3.4594,
      "step": 191480
    },
    {
      "epoch": 0.1784996645965295,
      "grad_norm": 1.0000345706939697,
      "learning_rate": 0.00016430028263122963,
      "loss": 3.5642,
      "step": 191490
    },
    {
      "epoch": 0.1785089862146086,
      "grad_norm": 1.2687877416610718,
      "learning_rate": 0.00016429841830913443,
      "loss": 3.4031,
      "step": 191500
    },
    {
      "epoch": 0.17851830783268768,
      "grad_norm": 1.0094969272613525,
      "learning_rate": 0.00016429655398703924,
      "loss": 3.5275,
      "step": 191510
    },
    {
      "epoch": 0.17852762945076678,
      "grad_norm": 1.1013092994689941,
      "learning_rate": 0.00016429468966494405,
      "loss": 3.5735,
      "step": 191520
    },
    {
      "epoch": 0.17853695106884587,
      "grad_norm": 1.1359738111495972,
      "learning_rate": 0.00016429282534284885,
      "loss": 3.4799,
      "step": 191530
    },
    {
      "epoch": 0.17854627268692494,
      "grad_norm": 1.1340910196304321,
      "learning_rate": 0.00016429096102075363,
      "loss": 3.4965,
      "step": 191540
    },
    {
      "epoch": 0.17855559430500403,
      "grad_norm": 0.9604598879814148,
      "learning_rate": 0.00016428909669865844,
      "loss": 3.4761,
      "step": 191550
    },
    {
      "epoch": 0.17856491592308313,
      "grad_norm": 0.9441665410995483,
      "learning_rate": 0.00016428723237656324,
      "loss": 3.6772,
      "step": 191560
    },
    {
      "epoch": 0.17857423754116222,
      "grad_norm": 1.2210428714752197,
      "learning_rate": 0.00016428536805446802,
      "loss": 3.4689,
      "step": 191570
    },
    {
      "epoch": 0.17858355915924132,
      "grad_norm": 1.1010392904281616,
      "learning_rate": 0.00016428350373237286,
      "loss": 3.1223,
      "step": 191580
    },
    {
      "epoch": 0.1785928807773204,
      "grad_norm": 1.2007358074188232,
      "learning_rate": 0.00016428163941027764,
      "loss": 3.4383,
      "step": 191590
    },
    {
      "epoch": 0.1786022023953995,
      "grad_norm": 0.9731091856956482,
      "learning_rate": 0.00016427977508818244,
      "loss": 3.5036,
      "step": 191600
    },
    {
      "epoch": 0.1786115240134786,
      "grad_norm": 0.962233304977417,
      "learning_rate": 0.00016427791076608725,
      "loss": 3.4952,
      "step": 191610
    },
    {
      "epoch": 0.1786208456315577,
      "grad_norm": 1.005821943283081,
      "learning_rate": 0.00016427604644399203,
      "loss": 3.3626,
      "step": 191620
    },
    {
      "epoch": 0.1786301672496368,
      "grad_norm": 1.1139909029006958,
      "learning_rate": 0.00016427418212189683,
      "loss": 3.466,
      "step": 191630
    },
    {
      "epoch": 0.17863948886771586,
      "grad_norm": 1.1199408769607544,
      "learning_rate": 0.00016427231779980164,
      "loss": 3.631,
      "step": 191640
    },
    {
      "epoch": 0.17864881048579495,
      "grad_norm": 1.1054682731628418,
      "learning_rate": 0.00016427045347770645,
      "loss": 3.4522,
      "step": 191650
    },
    {
      "epoch": 0.17865813210387405,
      "grad_norm": 0.9125702381134033,
      "learning_rate": 0.00016426858915561125,
      "loss": 3.4071,
      "step": 191660
    },
    {
      "epoch": 0.17866745372195314,
      "grad_norm": 1.1319217681884766,
      "learning_rate": 0.00016426672483351603,
      "loss": 3.5802,
      "step": 191670
    },
    {
      "epoch": 0.17867677534003223,
      "grad_norm": 0.8133047223091125,
      "learning_rate": 0.00016426486051142084,
      "loss": 3.2553,
      "step": 191680
    },
    {
      "epoch": 0.17868609695811133,
      "grad_norm": 1.0221545696258545,
      "learning_rate": 0.00016426299618932564,
      "loss": 3.6202,
      "step": 191690
    },
    {
      "epoch": 0.17869541857619042,
      "grad_norm": 1.0157276391983032,
      "learning_rate": 0.00016426113186723045,
      "loss": 3.5632,
      "step": 191700
    },
    {
      "epoch": 0.17870474019426952,
      "grad_norm": 1.0987324714660645,
      "learning_rate": 0.00016425926754513526,
      "loss": 3.5881,
      "step": 191710
    },
    {
      "epoch": 0.1787140618123486,
      "grad_norm": 0.9457873702049255,
      "learning_rate": 0.00016425740322304004,
      "loss": 3.5874,
      "step": 191720
    },
    {
      "epoch": 0.1787233834304277,
      "grad_norm": 0.9004294276237488,
      "learning_rate": 0.00016425553890094484,
      "loss": 3.6655,
      "step": 191730
    },
    {
      "epoch": 0.17873270504850677,
      "grad_norm": 0.8626555800437927,
      "learning_rate": 0.00016425367457884965,
      "loss": 3.552,
      "step": 191740
    },
    {
      "epoch": 0.17874202666658587,
      "grad_norm": 1.1219682693481445,
      "learning_rate": 0.00016425181025675446,
      "loss": 3.561,
      "step": 191750
    },
    {
      "epoch": 0.17875134828466496,
      "grad_norm": 1.1677138805389404,
      "learning_rate": 0.00016424994593465926,
      "loss": 3.3782,
      "step": 191760
    },
    {
      "epoch": 0.17876066990274406,
      "grad_norm": 1.1961759328842163,
      "learning_rate": 0.00016424808161256404,
      "loss": 3.448,
      "step": 191770
    },
    {
      "epoch": 0.17876999152082315,
      "grad_norm": 1.087111234664917,
      "learning_rate": 0.00016424621729046885,
      "loss": 3.6332,
      "step": 191780
    },
    {
      "epoch": 0.17877931313890225,
      "grad_norm": 0.9941025376319885,
      "learning_rate": 0.00016424435296837363,
      "loss": 3.662,
      "step": 191790
    },
    {
      "epoch": 0.17878863475698134,
      "grad_norm": 1.0638805627822876,
      "learning_rate": 0.00016424248864627846,
      "loss": 3.3726,
      "step": 191800
    },
    {
      "epoch": 0.17879795637506044,
      "grad_norm": 0.8552287817001343,
      "learning_rate": 0.00016424062432418327,
      "loss": 3.4552,
      "step": 191810
    },
    {
      "epoch": 0.17880727799313953,
      "grad_norm": 1.1186120510101318,
      "learning_rate": 0.00016423876000208805,
      "loss": 3.5787,
      "step": 191820
    },
    {
      "epoch": 0.17881659961121862,
      "grad_norm": 0.9890151619911194,
      "learning_rate": 0.00016423689567999285,
      "loss": 3.503,
      "step": 191830
    },
    {
      "epoch": 0.17882592122929772,
      "grad_norm": 0.9210302233695984,
      "learning_rate": 0.00016423503135789763,
      "loss": 3.411,
      "step": 191840
    },
    {
      "epoch": 0.17883524284737679,
      "grad_norm": 1.020156741142273,
      "learning_rate": 0.00016423316703580244,
      "loss": 3.4442,
      "step": 191850
    },
    {
      "epoch": 0.17884456446545588,
      "grad_norm": 1.1283053159713745,
      "learning_rate": 0.00016423130271370727,
      "loss": 3.5739,
      "step": 191860
    },
    {
      "epoch": 0.17885388608353497,
      "grad_norm": 1.048803448677063,
      "learning_rate": 0.00016422943839161205,
      "loss": 3.4963,
      "step": 191870
    },
    {
      "epoch": 0.17886320770161407,
      "grad_norm": 1.099291443824768,
      "learning_rate": 0.00016422757406951686,
      "loss": 3.5156,
      "step": 191880
    },
    {
      "epoch": 0.17887252931969316,
      "grad_norm": 1.0873723030090332,
      "learning_rate": 0.00016422570974742164,
      "loss": 3.4467,
      "step": 191890
    },
    {
      "epoch": 0.17888185093777226,
      "grad_norm": 1.0697669982910156,
      "learning_rate": 0.00016422384542532644,
      "loss": 3.4188,
      "step": 191900
    },
    {
      "epoch": 0.17889117255585135,
      "grad_norm": 1.083868384361267,
      "learning_rate": 0.00016422198110323125,
      "loss": 3.6451,
      "step": 191910
    },
    {
      "epoch": 0.17890049417393045,
      "grad_norm": 1.1153229475021362,
      "learning_rate": 0.00016422011678113605,
      "loss": 3.5645,
      "step": 191920
    },
    {
      "epoch": 0.17890981579200954,
      "grad_norm": 1.2114897966384888,
      "learning_rate": 0.00016421825245904086,
      "loss": 3.3801,
      "step": 191930
    },
    {
      "epoch": 0.17891913741008864,
      "grad_norm": 1.1588689088821411,
      "learning_rate": 0.00016421638813694564,
      "loss": 3.6347,
      "step": 191940
    },
    {
      "epoch": 0.1789284590281677,
      "grad_norm": 1.1535214185714722,
      "learning_rate": 0.00016421452381485045,
      "loss": 3.5775,
      "step": 191950
    },
    {
      "epoch": 0.1789377806462468,
      "grad_norm": 1.118955373764038,
      "learning_rate": 0.00016421265949275525,
      "loss": 3.3717,
      "step": 191960
    },
    {
      "epoch": 0.1789471022643259,
      "grad_norm": 1.0724785327911377,
      "learning_rate": 0.00016421079517066006,
      "loss": 3.3746,
      "step": 191970
    },
    {
      "epoch": 0.17895642388240499,
      "grad_norm": 1.152620792388916,
      "learning_rate": 0.00016420893084856487,
      "loss": 3.4863,
      "step": 191980
    },
    {
      "epoch": 0.17896574550048408,
      "grad_norm": 1.015843391418457,
      "learning_rate": 0.00016420706652646964,
      "loss": 3.3767,
      "step": 191990
    },
    {
      "epoch": 0.17897506711856317,
      "grad_norm": 1.0638461112976074,
      "learning_rate": 0.00016420520220437445,
      "loss": 3.4701,
      "step": 192000
    },
    {
      "epoch": 0.17898438873664227,
      "grad_norm": 1.0443059206008911,
      "learning_rate": 0.00016420333788227926,
      "loss": 3.6293,
      "step": 192010
    },
    {
      "epoch": 0.17899371035472136,
      "grad_norm": 1.202284336090088,
      "learning_rate": 0.00016420147356018404,
      "loss": 3.5106,
      "step": 192020
    },
    {
      "epoch": 0.17900303197280046,
      "grad_norm": 1.1611202955245972,
      "learning_rate": 0.00016419960923808887,
      "loss": 3.4578,
      "step": 192030
    },
    {
      "epoch": 0.17901235359087955,
      "grad_norm": 1.1212810277938843,
      "learning_rate": 0.00016419774491599365,
      "loss": 3.49,
      "step": 192040
    },
    {
      "epoch": 0.17902167520895865,
      "grad_norm": 1.1062183380126953,
      "learning_rate": 0.00016419588059389846,
      "loss": 3.4067,
      "step": 192050
    },
    {
      "epoch": 0.17903099682703771,
      "grad_norm": 1.190456748008728,
      "learning_rate": 0.00016419401627180326,
      "loss": 3.5647,
      "step": 192060
    },
    {
      "epoch": 0.1790403184451168,
      "grad_norm": 1.2040445804595947,
      "learning_rate": 0.00016419215194970804,
      "loss": 3.5517,
      "step": 192070
    },
    {
      "epoch": 0.1790496400631959,
      "grad_norm": 1.1150383949279785,
      "learning_rate": 0.00016419028762761285,
      "loss": 3.5961,
      "step": 192080
    },
    {
      "epoch": 0.179058961681275,
      "grad_norm": 0.9388807415962219,
      "learning_rate": 0.00016418842330551765,
      "loss": 3.5392,
      "step": 192090
    },
    {
      "epoch": 0.1790682832993541,
      "grad_norm": 0.9902334809303284,
      "learning_rate": 0.00016418655898342246,
      "loss": 3.3879,
      "step": 192100
    },
    {
      "epoch": 0.1790776049174332,
      "grad_norm": 1.1896946430206299,
      "learning_rate": 0.00016418469466132727,
      "loss": 3.4849,
      "step": 192110
    },
    {
      "epoch": 0.17908692653551228,
      "grad_norm": 1.08512282371521,
      "learning_rate": 0.00016418283033923205,
      "loss": 3.5559,
      "step": 192120
    },
    {
      "epoch": 0.17909624815359138,
      "grad_norm": 1.3786256313323975,
      "learning_rate": 0.00016418096601713685,
      "loss": 3.4009,
      "step": 192130
    },
    {
      "epoch": 0.17910556977167047,
      "grad_norm": 1.1152266263961792,
      "learning_rate": 0.00016417910169504166,
      "loss": 3.4896,
      "step": 192140
    },
    {
      "epoch": 0.17911489138974956,
      "grad_norm": 1.2188682556152344,
      "learning_rate": 0.00016417723737294646,
      "loss": 3.5382,
      "step": 192150
    },
    {
      "epoch": 0.17912421300782863,
      "grad_norm": 0.9470201134681702,
      "learning_rate": 0.00016417537305085127,
      "loss": 3.4846,
      "step": 192160
    },
    {
      "epoch": 0.17913353462590773,
      "grad_norm": 1.0707522630691528,
      "learning_rate": 0.00016417350872875605,
      "loss": 3.5437,
      "step": 192170
    },
    {
      "epoch": 0.17914285624398682,
      "grad_norm": 1.1444652080535889,
      "learning_rate": 0.00016417164440666086,
      "loss": 3.4847,
      "step": 192180
    },
    {
      "epoch": 0.17915217786206591,
      "grad_norm": 0.8609937429428101,
      "learning_rate": 0.00016416978008456566,
      "loss": 3.3896,
      "step": 192190
    },
    {
      "epoch": 0.179161499480145,
      "grad_norm": 0.9307360649108887,
      "learning_rate": 0.00016416791576247047,
      "loss": 3.6228,
      "step": 192200
    },
    {
      "epoch": 0.1791708210982241,
      "grad_norm": 1.2869563102722168,
      "learning_rate": 0.00016416605144037527,
      "loss": 3.2646,
      "step": 192210
    },
    {
      "epoch": 0.1791801427163032,
      "grad_norm": 1.1184667348861694,
      "learning_rate": 0.00016416418711828005,
      "loss": 3.5184,
      "step": 192220
    },
    {
      "epoch": 0.1791894643343823,
      "grad_norm": 0.8358355760574341,
      "learning_rate": 0.00016416232279618486,
      "loss": 3.5105,
      "step": 192230
    },
    {
      "epoch": 0.1791987859524614,
      "grad_norm": 1.1381237506866455,
      "learning_rate": 0.00016416045847408964,
      "loss": 3.6729,
      "step": 192240
    },
    {
      "epoch": 0.17920810757054048,
      "grad_norm": 0.9755403995513916,
      "learning_rate": 0.00016415859415199447,
      "loss": 3.4603,
      "step": 192250
    },
    {
      "epoch": 0.17921742918861958,
      "grad_norm": 1.0476057529449463,
      "learning_rate": 0.00016415672982989928,
      "loss": 3.4552,
      "step": 192260
    },
    {
      "epoch": 0.17922675080669864,
      "grad_norm": 1.243487000465393,
      "learning_rate": 0.00016415486550780406,
      "loss": 3.5168,
      "step": 192270
    },
    {
      "epoch": 0.17923607242477774,
      "grad_norm": 1.2167344093322754,
      "learning_rate": 0.00016415300118570886,
      "loss": 3.518,
      "step": 192280
    },
    {
      "epoch": 0.17924539404285683,
      "grad_norm": 1.0494226217269897,
      "learning_rate": 0.00016415113686361364,
      "loss": 3.3028,
      "step": 192290
    },
    {
      "epoch": 0.17925471566093593,
      "grad_norm": 1.1329748630523682,
      "learning_rate": 0.00016414927254151845,
      "loss": 3.4562,
      "step": 192300
    },
    {
      "epoch": 0.17926403727901502,
      "grad_norm": 0.9668267965316772,
      "learning_rate": 0.00016414740821942328,
      "loss": 3.392,
      "step": 192310
    },
    {
      "epoch": 0.17927335889709411,
      "grad_norm": 0.8612527847290039,
      "learning_rate": 0.00016414554389732806,
      "loss": 3.5491,
      "step": 192320
    },
    {
      "epoch": 0.1792826805151732,
      "grad_norm": 0.913532555103302,
      "learning_rate": 0.00016414367957523287,
      "loss": 3.5164,
      "step": 192330
    },
    {
      "epoch": 0.1792920021332523,
      "grad_norm": 1.091029167175293,
      "learning_rate": 0.00016414181525313765,
      "loss": 3.5289,
      "step": 192340
    },
    {
      "epoch": 0.1793013237513314,
      "grad_norm": 1.1246556043624878,
      "learning_rate": 0.00016413995093104245,
      "loss": 3.478,
      "step": 192350
    },
    {
      "epoch": 0.1793106453694105,
      "grad_norm": 1.0987976789474487,
      "learning_rate": 0.00016413808660894726,
      "loss": 3.4871,
      "step": 192360
    },
    {
      "epoch": 0.17931996698748956,
      "grad_norm": 1.0470962524414062,
      "learning_rate": 0.00016413622228685207,
      "loss": 3.6605,
      "step": 192370
    },
    {
      "epoch": 0.17932928860556865,
      "grad_norm": 0.945302426815033,
      "learning_rate": 0.00016413435796475687,
      "loss": 3.4921,
      "step": 192380
    },
    {
      "epoch": 0.17933861022364775,
      "grad_norm": 1.0624130964279175,
      "learning_rate": 0.00016413249364266165,
      "loss": 3.4473,
      "step": 192390
    },
    {
      "epoch": 0.17934793184172684,
      "grad_norm": 1.1812100410461426,
      "learning_rate": 0.00016413062932056646,
      "loss": 3.5035,
      "step": 192400
    },
    {
      "epoch": 0.17935725345980594,
      "grad_norm": 1.2991490364074707,
      "learning_rate": 0.00016412876499847127,
      "loss": 3.6682,
      "step": 192410
    },
    {
      "epoch": 0.17936657507788503,
      "grad_norm": 1.1144840717315674,
      "learning_rate": 0.00016412690067637607,
      "loss": 3.6885,
      "step": 192420
    },
    {
      "epoch": 0.17937589669596413,
      "grad_norm": 1.0700290203094482,
      "learning_rate": 0.00016412503635428088,
      "loss": 3.5135,
      "step": 192430
    },
    {
      "epoch": 0.17938521831404322,
      "grad_norm": 1.0423575639724731,
      "learning_rate": 0.00016412317203218566,
      "loss": 3.4631,
      "step": 192440
    },
    {
      "epoch": 0.17939453993212232,
      "grad_norm": 1.192484974861145,
      "learning_rate": 0.00016412130771009046,
      "loss": 3.4215,
      "step": 192450
    },
    {
      "epoch": 0.1794038615502014,
      "grad_norm": 1.0241835117340088,
      "learning_rate": 0.00016411944338799527,
      "loss": 3.5439,
      "step": 192460
    },
    {
      "epoch": 0.1794131831682805,
      "grad_norm": 1.0380223989486694,
      "learning_rate": 0.00016411757906590005,
      "loss": 3.5294,
      "step": 192470
    },
    {
      "epoch": 0.17942250478635957,
      "grad_norm": 1.0561221837997437,
      "learning_rate": 0.00016411571474380488,
      "loss": 3.6692,
      "step": 192480
    },
    {
      "epoch": 0.17943182640443867,
      "grad_norm": 0.9867602586746216,
      "learning_rate": 0.00016411385042170966,
      "loss": 3.4282,
      "step": 192490
    },
    {
      "epoch": 0.17944114802251776,
      "grad_norm": 1.0909408330917358,
      "learning_rate": 0.00016411198609961447,
      "loss": 3.4904,
      "step": 192500
    },
    {
      "epoch": 0.17945046964059685,
      "grad_norm": 1.1363332271575928,
      "learning_rate": 0.00016411012177751927,
      "loss": 3.5148,
      "step": 192510
    },
    {
      "epoch": 0.17945979125867595,
      "grad_norm": 1.3493534326553345,
      "learning_rate": 0.00016410825745542405,
      "loss": 3.5382,
      "step": 192520
    },
    {
      "epoch": 0.17946911287675504,
      "grad_norm": 1.1756694316864014,
      "learning_rate": 0.00016410639313332886,
      "loss": 3.591,
      "step": 192530
    },
    {
      "epoch": 0.17947843449483414,
      "grad_norm": 1.1058510541915894,
      "learning_rate": 0.00016410452881123367,
      "loss": 3.4565,
      "step": 192540
    },
    {
      "epoch": 0.17948775611291323,
      "grad_norm": 1.2107651233673096,
      "learning_rate": 0.00016410266448913847,
      "loss": 3.5667,
      "step": 192550
    },
    {
      "epoch": 0.17949707773099233,
      "grad_norm": 1.1285796165466309,
      "learning_rate": 0.00016410080016704328,
      "loss": 3.38,
      "step": 192560
    },
    {
      "epoch": 0.17950639934907142,
      "grad_norm": 0.8573475480079651,
      "learning_rate": 0.00016409893584494806,
      "loss": 3.7072,
      "step": 192570
    },
    {
      "epoch": 0.1795157209671505,
      "grad_norm": 0.9370082020759583,
      "learning_rate": 0.00016409707152285286,
      "loss": 3.4687,
      "step": 192580
    },
    {
      "epoch": 0.17952504258522958,
      "grad_norm": 1.1231155395507812,
      "learning_rate": 0.00016409520720075767,
      "loss": 3.4013,
      "step": 192590
    },
    {
      "epoch": 0.17953436420330868,
      "grad_norm": 1.1046075820922852,
      "learning_rate": 0.00016409334287866248,
      "loss": 3.652,
      "step": 192600
    },
    {
      "epoch": 0.17954368582138777,
      "grad_norm": 0.991704523563385,
      "learning_rate": 0.00016409147855656728,
      "loss": 3.5687,
      "step": 192610
    },
    {
      "epoch": 0.17955300743946687,
      "grad_norm": 1.1691490411758423,
      "learning_rate": 0.00016408961423447206,
      "loss": 3.4724,
      "step": 192620
    },
    {
      "epoch": 0.17956232905754596,
      "grad_norm": 1.2474901676177979,
      "learning_rate": 0.00016408774991237687,
      "loss": 3.5482,
      "step": 192630
    },
    {
      "epoch": 0.17957165067562506,
      "grad_norm": 0.9351772665977478,
      "learning_rate": 0.00016408588559028168,
      "loss": 3.4165,
      "step": 192640
    },
    {
      "epoch": 0.17958097229370415,
      "grad_norm": 0.7836554646492004,
      "learning_rate": 0.00016408402126818648,
      "loss": 3.5359,
      "step": 192650
    },
    {
      "epoch": 0.17959029391178324,
      "grad_norm": 1.026523232460022,
      "learning_rate": 0.0001640821569460913,
      "loss": 3.6181,
      "step": 192660
    },
    {
      "epoch": 0.17959961552986234,
      "grad_norm": 1.1901257038116455,
      "learning_rate": 0.00016408029262399607,
      "loss": 3.6693,
      "step": 192670
    },
    {
      "epoch": 0.17960893714794143,
      "grad_norm": 0.8493075966835022,
      "learning_rate": 0.00016407842830190087,
      "loss": 3.4555,
      "step": 192680
    },
    {
      "epoch": 0.1796182587660205,
      "grad_norm": 0.7391703128814697,
      "learning_rate": 0.00016407656397980565,
      "loss": 3.455,
      "step": 192690
    },
    {
      "epoch": 0.1796275803840996,
      "grad_norm": 0.9475071430206299,
      "learning_rate": 0.00016407469965771049,
      "loss": 3.4729,
      "step": 192700
    },
    {
      "epoch": 0.1796369020021787,
      "grad_norm": 0.8509281873703003,
      "learning_rate": 0.0001640728353356153,
      "loss": 3.5555,
      "step": 192710
    },
    {
      "epoch": 0.17964622362025778,
      "grad_norm": 1.1499134302139282,
      "learning_rate": 0.00016407097101352007,
      "loss": 3.564,
      "step": 192720
    },
    {
      "epoch": 0.17965554523833688,
      "grad_norm": 1.0712106227874756,
      "learning_rate": 0.00016406910669142488,
      "loss": 3.4092,
      "step": 192730
    },
    {
      "epoch": 0.17966486685641597,
      "grad_norm": 1.1992918252944946,
      "learning_rate": 0.00016406724236932966,
      "loss": 3.5273,
      "step": 192740
    },
    {
      "epoch": 0.17967418847449507,
      "grad_norm": 0.8719925284385681,
      "learning_rate": 0.00016406537804723446,
      "loss": 3.5291,
      "step": 192750
    },
    {
      "epoch": 0.17968351009257416,
      "grad_norm": 1.144896149635315,
      "learning_rate": 0.00016406351372513927,
      "loss": 3.7355,
      "step": 192760
    },
    {
      "epoch": 0.17969283171065326,
      "grad_norm": 1.0847508907318115,
      "learning_rate": 0.00016406164940304408,
      "loss": 3.6011,
      "step": 192770
    },
    {
      "epoch": 0.17970215332873235,
      "grad_norm": 1.1125667095184326,
      "learning_rate": 0.00016405978508094888,
      "loss": 3.6697,
      "step": 192780
    },
    {
      "epoch": 0.17971147494681142,
      "grad_norm": 0.8685410618782043,
      "learning_rate": 0.00016405792075885366,
      "loss": 3.7013,
      "step": 192790
    },
    {
      "epoch": 0.1797207965648905,
      "grad_norm": 1.315866470336914,
      "learning_rate": 0.00016405605643675847,
      "loss": 3.6127,
      "step": 192800
    },
    {
      "epoch": 0.1797301181829696,
      "grad_norm": 1.276174783706665,
      "learning_rate": 0.00016405419211466327,
      "loss": 3.2999,
      "step": 192810
    },
    {
      "epoch": 0.1797394398010487,
      "grad_norm": 1.1340014934539795,
      "learning_rate": 0.00016405232779256808,
      "loss": 3.4356,
      "step": 192820
    },
    {
      "epoch": 0.1797487614191278,
      "grad_norm": 0.9422038197517395,
      "learning_rate": 0.00016405046347047289,
      "loss": 3.5987,
      "step": 192830
    },
    {
      "epoch": 0.1797580830372069,
      "grad_norm": 0.9810900092124939,
      "learning_rate": 0.00016404859914837767,
      "loss": 3.3352,
      "step": 192840
    },
    {
      "epoch": 0.17976740465528598,
      "grad_norm": 1.2392741441726685,
      "learning_rate": 0.00016404673482628247,
      "loss": 3.7486,
      "step": 192850
    },
    {
      "epoch": 0.17977672627336508,
      "grad_norm": 0.95096355676651,
      "learning_rate": 0.00016404487050418725,
      "loss": 3.6395,
      "step": 192860
    },
    {
      "epoch": 0.17978604789144417,
      "grad_norm": 1.0107941627502441,
      "learning_rate": 0.00016404300618209208,
      "loss": 3.303,
      "step": 192870
    },
    {
      "epoch": 0.17979536950952327,
      "grad_norm": 1.037257194519043,
      "learning_rate": 0.0001640411418599969,
      "loss": 3.6739,
      "step": 192880
    },
    {
      "epoch": 0.17980469112760236,
      "grad_norm": 1.2803703546524048,
      "learning_rate": 0.00016403927753790167,
      "loss": 3.6992,
      "step": 192890
    },
    {
      "epoch": 0.17981401274568143,
      "grad_norm": 1.1136908531188965,
      "learning_rate": 0.00016403741321580648,
      "loss": 3.4061,
      "step": 192900
    },
    {
      "epoch": 0.17982333436376052,
      "grad_norm": 1.1458215713500977,
      "learning_rate": 0.00016403554889371126,
      "loss": 3.3121,
      "step": 192910
    },
    {
      "epoch": 0.17983265598183962,
      "grad_norm": 1.1752278804779053,
      "learning_rate": 0.00016403368457161606,
      "loss": 3.5141,
      "step": 192920
    },
    {
      "epoch": 0.1798419775999187,
      "grad_norm": 0.860593855381012,
      "learning_rate": 0.0001640318202495209,
      "loss": 3.6119,
      "step": 192930
    },
    {
      "epoch": 0.1798512992179978,
      "grad_norm": 1.0458377599716187,
      "learning_rate": 0.00016402995592742567,
      "loss": 3.3138,
      "step": 192940
    },
    {
      "epoch": 0.1798606208360769,
      "grad_norm": 1.0916318893432617,
      "learning_rate": 0.00016402809160533048,
      "loss": 3.5336,
      "step": 192950
    },
    {
      "epoch": 0.179869942454156,
      "grad_norm": 1.0472618341445923,
      "learning_rate": 0.00016402622728323526,
      "loss": 3.4848,
      "step": 192960
    },
    {
      "epoch": 0.1798792640722351,
      "grad_norm": 0.931799054145813,
      "learning_rate": 0.00016402436296114007,
      "loss": 3.6038,
      "step": 192970
    },
    {
      "epoch": 0.17988858569031418,
      "grad_norm": 1.3939602375030518,
      "learning_rate": 0.00016402249863904487,
      "loss": 3.4892,
      "step": 192980
    },
    {
      "epoch": 0.17989790730839328,
      "grad_norm": 1.025166392326355,
      "learning_rate": 0.00016402063431694968,
      "loss": 3.485,
      "step": 192990
    },
    {
      "epoch": 0.17990722892647235,
      "grad_norm": 0.9980576634407043,
      "learning_rate": 0.00016401876999485449,
      "loss": 3.4047,
      "step": 193000
    },
    {
      "epoch": 0.17991655054455144,
      "grad_norm": 0.9668683409690857,
      "learning_rate": 0.00016401690567275926,
      "loss": 3.3793,
      "step": 193010
    },
    {
      "epoch": 0.17992587216263053,
      "grad_norm": 1.2203179597854614,
      "learning_rate": 0.00016401504135066407,
      "loss": 3.4212,
      "step": 193020
    },
    {
      "epoch": 0.17993519378070963,
      "grad_norm": 1.1591209173202515,
      "learning_rate": 0.00016401317702856888,
      "loss": 3.488,
      "step": 193030
    },
    {
      "epoch": 0.17994451539878872,
      "grad_norm": 0.9791854619979858,
      "learning_rate": 0.00016401131270647368,
      "loss": 3.5983,
      "step": 193040
    },
    {
      "epoch": 0.17995383701686782,
      "grad_norm": 1.0376545190811157,
      "learning_rate": 0.0001640094483843785,
      "loss": 3.582,
      "step": 193050
    },
    {
      "epoch": 0.1799631586349469,
      "grad_norm": 1.2250910997390747,
      "learning_rate": 0.00016400758406228327,
      "loss": 3.5681,
      "step": 193060
    },
    {
      "epoch": 0.179972480253026,
      "grad_norm": 1.1330736875534058,
      "learning_rate": 0.00016400571974018808,
      "loss": 3.5528,
      "step": 193070
    },
    {
      "epoch": 0.1799818018711051,
      "grad_norm": 1.131759762763977,
      "learning_rate": 0.00016400385541809288,
      "loss": 3.5815,
      "step": 193080
    },
    {
      "epoch": 0.1799911234891842,
      "grad_norm": 0.8237096071243286,
      "learning_rate": 0.0001640019910959977,
      "loss": 3.6447,
      "step": 193090
    },
    {
      "epoch": 0.1800004451072633,
      "grad_norm": 1.210964560508728,
      "learning_rate": 0.0001640001267739025,
      "loss": 3.7378,
      "step": 193100
    },
    {
      "epoch": 0.18000976672534236,
      "grad_norm": 1.0362085103988647,
      "learning_rate": 0.00016399826245180727,
      "loss": 3.4535,
      "step": 193110
    },
    {
      "epoch": 0.18001908834342145,
      "grad_norm": 0.9856550693511963,
      "learning_rate": 0.00016399639812971208,
      "loss": 3.4562,
      "step": 193120
    },
    {
      "epoch": 0.18002840996150055,
      "grad_norm": 0.9529857039451599,
      "learning_rate": 0.00016399453380761689,
      "loss": 3.6682,
      "step": 193130
    },
    {
      "epoch": 0.18003773157957964,
      "grad_norm": 1.0757198333740234,
      "learning_rate": 0.00016399266948552167,
      "loss": 3.4817,
      "step": 193140
    },
    {
      "epoch": 0.18004705319765874,
      "grad_norm": 0.9221884608268738,
      "learning_rate": 0.0001639908051634265,
      "loss": 3.3559,
      "step": 193150
    },
    {
      "epoch": 0.18005637481573783,
      "grad_norm": 1.0744373798370361,
      "learning_rate": 0.00016398894084133128,
      "loss": 3.4828,
      "step": 193160
    },
    {
      "epoch": 0.18006569643381692,
      "grad_norm": 1.0075643062591553,
      "learning_rate": 0.00016398707651923608,
      "loss": 3.8172,
      "step": 193170
    },
    {
      "epoch": 0.18007501805189602,
      "grad_norm": 1.0740247964859009,
      "learning_rate": 0.0001639852121971409,
      "loss": 3.5209,
      "step": 193180
    },
    {
      "epoch": 0.1800843396699751,
      "grad_norm": 1.0757825374603271,
      "learning_rate": 0.00016398334787504567,
      "loss": 3.5199,
      "step": 193190
    },
    {
      "epoch": 0.1800936612880542,
      "grad_norm": 1.0656059980392456,
      "learning_rate": 0.00016398148355295048,
      "loss": 3.5692,
      "step": 193200
    },
    {
      "epoch": 0.18010298290613327,
      "grad_norm": 1.0836994647979736,
      "learning_rate": 0.00016397961923085528,
      "loss": 3.5278,
      "step": 193210
    },
    {
      "epoch": 0.18011230452421237,
      "grad_norm": 1.1731899976730347,
      "learning_rate": 0.0001639777549087601,
      "loss": 3.3587,
      "step": 193220
    },
    {
      "epoch": 0.18012162614229146,
      "grad_norm": 1.1910964250564575,
      "learning_rate": 0.0001639758905866649,
      "loss": 3.4533,
      "step": 193230
    },
    {
      "epoch": 0.18013094776037056,
      "grad_norm": 0.9575589895248413,
      "learning_rate": 0.00016397402626456967,
      "loss": 3.531,
      "step": 193240
    },
    {
      "epoch": 0.18014026937844965,
      "grad_norm": 1.0558637380599976,
      "learning_rate": 0.00016397216194247448,
      "loss": 3.5814,
      "step": 193250
    },
    {
      "epoch": 0.18014959099652875,
      "grad_norm": 1.056782603263855,
      "learning_rate": 0.0001639702976203793,
      "loss": 3.4323,
      "step": 193260
    },
    {
      "epoch": 0.18015891261460784,
      "grad_norm": 1.118984580039978,
      "learning_rate": 0.0001639684332982841,
      "loss": 3.5463,
      "step": 193270
    },
    {
      "epoch": 0.18016823423268694,
      "grad_norm": 1.1203868389129639,
      "learning_rate": 0.0001639665689761889,
      "loss": 3.6181,
      "step": 193280
    },
    {
      "epoch": 0.18017755585076603,
      "grad_norm": 0.903698742389679,
      "learning_rate": 0.00016396470465409368,
      "loss": 3.478,
      "step": 193290
    },
    {
      "epoch": 0.18018687746884512,
      "grad_norm": 0.9862079620361328,
      "learning_rate": 0.00016396284033199848,
      "loss": 3.5586,
      "step": 193300
    },
    {
      "epoch": 0.1801961990869242,
      "grad_norm": 1.2008923292160034,
      "learning_rate": 0.0001639609760099033,
      "loss": 3.4695,
      "step": 193310
    },
    {
      "epoch": 0.18020552070500329,
      "grad_norm": 1.0928353071212769,
      "learning_rate": 0.0001639591116878081,
      "loss": 3.4668,
      "step": 193320
    },
    {
      "epoch": 0.18021484232308238,
      "grad_norm": 1.2224546670913696,
      "learning_rate": 0.0001639572473657129,
      "loss": 3.356,
      "step": 193330
    },
    {
      "epoch": 0.18022416394116147,
      "grad_norm": 1.0245281457901,
      "learning_rate": 0.00016395538304361768,
      "loss": 3.5033,
      "step": 193340
    },
    {
      "epoch": 0.18023348555924057,
      "grad_norm": 1.2407665252685547,
      "learning_rate": 0.0001639535187215225,
      "loss": 3.44,
      "step": 193350
    },
    {
      "epoch": 0.18024280717731966,
      "grad_norm": 0.8816227912902832,
      "learning_rate": 0.00016395165439942727,
      "loss": 3.6021,
      "step": 193360
    },
    {
      "epoch": 0.18025212879539876,
      "grad_norm": 0.9579691886901855,
      "learning_rate": 0.00016394979007733207,
      "loss": 3.3358,
      "step": 193370
    },
    {
      "epoch": 0.18026145041347785,
      "grad_norm": 1.2545500993728638,
      "learning_rate": 0.0001639479257552369,
      "loss": 3.5792,
      "step": 193380
    },
    {
      "epoch": 0.18027077203155695,
      "grad_norm": 1.03687584400177,
      "learning_rate": 0.0001639460614331417,
      "loss": 3.4562,
      "step": 193390
    },
    {
      "epoch": 0.18028009364963604,
      "grad_norm": 1.005733847618103,
      "learning_rate": 0.0001639441971110465,
      "loss": 3.4795,
      "step": 193400
    },
    {
      "epoch": 0.18028941526771514,
      "grad_norm": 1.1991126537322998,
      "learning_rate": 0.00016394233278895127,
      "loss": 3.4902,
      "step": 193410
    },
    {
      "epoch": 0.1802987368857942,
      "grad_norm": 1.1482861042022705,
      "learning_rate": 0.00016394046846685608,
      "loss": 3.4903,
      "step": 193420
    },
    {
      "epoch": 0.1803080585038733,
      "grad_norm": 1.1213538646697998,
      "learning_rate": 0.00016393860414476089,
      "loss": 3.5487,
      "step": 193430
    },
    {
      "epoch": 0.1803173801219524,
      "grad_norm": 0.867071807384491,
      "learning_rate": 0.0001639367398226657,
      "loss": 3.5815,
      "step": 193440
    },
    {
      "epoch": 0.1803267017400315,
      "grad_norm": 1.1116948127746582,
      "learning_rate": 0.0001639348755005705,
      "loss": 3.5484,
      "step": 193450
    },
    {
      "epoch": 0.18033602335811058,
      "grad_norm": 1.2658339738845825,
      "learning_rate": 0.00016393301117847528,
      "loss": 3.4409,
      "step": 193460
    },
    {
      "epoch": 0.18034534497618968,
      "grad_norm": 0.9899161458015442,
      "learning_rate": 0.00016393114685638008,
      "loss": 3.6526,
      "step": 193470
    },
    {
      "epoch": 0.18035466659426877,
      "grad_norm": 1.0276113748550415,
      "learning_rate": 0.0001639292825342849,
      "loss": 3.6418,
      "step": 193480
    },
    {
      "epoch": 0.18036398821234786,
      "grad_norm": 0.8826069235801697,
      "learning_rate": 0.0001639274182121897,
      "loss": 3.6136,
      "step": 193490
    },
    {
      "epoch": 0.18037330983042696,
      "grad_norm": 1.0986647605895996,
      "learning_rate": 0.0001639255538900945,
      "loss": 3.6328,
      "step": 193500
    },
    {
      "epoch": 0.18038263144850605,
      "grad_norm": 0.994693398475647,
      "learning_rate": 0.00016392368956799928,
      "loss": 3.3093,
      "step": 193510
    },
    {
      "epoch": 0.18039195306658512,
      "grad_norm": 0.8932901620864868,
      "learning_rate": 0.0001639218252459041,
      "loss": 3.5693,
      "step": 193520
    },
    {
      "epoch": 0.18040127468466421,
      "grad_norm": 1.0893652439117432,
      "learning_rate": 0.0001639199609238089,
      "loss": 3.6262,
      "step": 193530
    },
    {
      "epoch": 0.1804105963027433,
      "grad_norm": 1.1235971450805664,
      "learning_rate": 0.0001639180966017137,
      "loss": 3.5117,
      "step": 193540
    },
    {
      "epoch": 0.1804199179208224,
      "grad_norm": 1.1685459613800049,
      "learning_rate": 0.0001639162322796185,
      "loss": 3.5971,
      "step": 193550
    },
    {
      "epoch": 0.1804292395389015,
      "grad_norm": 1.0288993120193481,
      "learning_rate": 0.00016391436795752329,
      "loss": 3.5052,
      "step": 193560
    },
    {
      "epoch": 0.1804385611569806,
      "grad_norm": 0.998053789138794,
      "learning_rate": 0.0001639125036354281,
      "loss": 3.3407,
      "step": 193570
    },
    {
      "epoch": 0.1804478827750597,
      "grad_norm": 0.919596791267395,
      "learning_rate": 0.0001639106393133329,
      "loss": 3.6636,
      "step": 193580
    },
    {
      "epoch": 0.18045720439313878,
      "grad_norm": 0.8084822297096252,
      "learning_rate": 0.00016390877499123768,
      "loss": 3.7214,
      "step": 193590
    },
    {
      "epoch": 0.18046652601121788,
      "grad_norm": 1.0927119255065918,
      "learning_rate": 0.0001639069106691425,
      "loss": 3.6297,
      "step": 193600
    },
    {
      "epoch": 0.18047584762929697,
      "grad_norm": 1.0232523679733276,
      "learning_rate": 0.0001639050463470473,
      "loss": 3.4539,
      "step": 193610
    },
    {
      "epoch": 0.18048516924737606,
      "grad_norm": 1.1987602710723877,
      "learning_rate": 0.0001639031820249521,
      "loss": 3.6979,
      "step": 193620
    },
    {
      "epoch": 0.18049449086545513,
      "grad_norm": 0.960829496383667,
      "learning_rate": 0.0001639013177028569,
      "loss": 3.4188,
      "step": 193630
    },
    {
      "epoch": 0.18050381248353423,
      "grad_norm": 0.9505619406700134,
      "learning_rate": 0.00016389945338076168,
      "loss": 3.6719,
      "step": 193640
    },
    {
      "epoch": 0.18051313410161332,
      "grad_norm": 1.0817551612854004,
      "learning_rate": 0.0001638975890586665,
      "loss": 3.4774,
      "step": 193650
    },
    {
      "epoch": 0.18052245571969242,
      "grad_norm": 0.9321652054786682,
      "learning_rate": 0.0001638957247365713,
      "loss": 3.4243,
      "step": 193660
    },
    {
      "epoch": 0.1805317773377715,
      "grad_norm": 1.1878381967544556,
      "learning_rate": 0.0001638938604144761,
      "loss": 3.4613,
      "step": 193670
    },
    {
      "epoch": 0.1805410989558506,
      "grad_norm": 1.2423871755599976,
      "learning_rate": 0.0001638919960923809,
      "loss": 3.65,
      "step": 193680
    },
    {
      "epoch": 0.1805504205739297,
      "grad_norm": 1.1621453762054443,
      "learning_rate": 0.0001638901317702857,
      "loss": 3.719,
      "step": 193690
    },
    {
      "epoch": 0.1805597421920088,
      "grad_norm": 1.1181793212890625,
      "learning_rate": 0.0001638882674481905,
      "loss": 3.4353,
      "step": 193700
    },
    {
      "epoch": 0.1805690638100879,
      "grad_norm": 0.995718777179718,
      "learning_rate": 0.0001638864031260953,
      "loss": 3.5134,
      "step": 193710
    },
    {
      "epoch": 0.18057838542816698,
      "grad_norm": 1.191904067993164,
      "learning_rate": 0.0001638845388040001,
      "loss": 3.4764,
      "step": 193720
    },
    {
      "epoch": 0.18058770704624605,
      "grad_norm": 0.9812692403793335,
      "learning_rate": 0.0001638826744819049,
      "loss": 3.3289,
      "step": 193730
    },
    {
      "epoch": 0.18059702866432514,
      "grad_norm": 1.1031266450881958,
      "learning_rate": 0.0001638808101598097,
      "loss": 3.4429,
      "step": 193740
    },
    {
      "epoch": 0.18060635028240424,
      "grad_norm": 0.926883339881897,
      "learning_rate": 0.0001638789458377145,
      "loss": 3.5047,
      "step": 193750
    },
    {
      "epoch": 0.18061567190048333,
      "grad_norm": 1.0088493824005127,
      "learning_rate": 0.0001638770815156193,
      "loss": 3.5068,
      "step": 193760
    },
    {
      "epoch": 0.18062499351856243,
      "grad_norm": 0.9486165642738342,
      "learning_rate": 0.0001638752171935241,
      "loss": 3.3894,
      "step": 193770
    },
    {
      "epoch": 0.18063431513664152,
      "grad_norm": 0.9440293908119202,
      "learning_rate": 0.00016387335287142892,
      "loss": 3.5272,
      "step": 193780
    },
    {
      "epoch": 0.18064363675472062,
      "grad_norm": 0.9274420738220215,
      "learning_rate": 0.0001638714885493337,
      "loss": 3.6951,
      "step": 193790
    },
    {
      "epoch": 0.1806529583727997,
      "grad_norm": 1.0775483846664429,
      "learning_rate": 0.0001638696242272385,
      "loss": 3.4563,
      "step": 193800
    },
    {
      "epoch": 0.1806622799908788,
      "grad_norm": 1.0516231060028076,
      "learning_rate": 0.00016386775990514328,
      "loss": 3.4434,
      "step": 193810
    },
    {
      "epoch": 0.1806716016089579,
      "grad_norm": 1.0796860456466675,
      "learning_rate": 0.00016386589558304811,
      "loss": 3.5374,
      "step": 193820
    },
    {
      "epoch": 0.180680923227037,
      "grad_norm": 1.1678317785263062,
      "learning_rate": 0.00016386403126095292,
      "loss": 3.4148,
      "step": 193830
    },
    {
      "epoch": 0.18069024484511606,
      "grad_norm": 0.9005653858184814,
      "learning_rate": 0.0001638621669388577,
      "loss": 3.6958,
      "step": 193840
    },
    {
      "epoch": 0.18069956646319515,
      "grad_norm": 0.8544486165046692,
      "learning_rate": 0.0001638603026167625,
      "loss": 3.4335,
      "step": 193850
    },
    {
      "epoch": 0.18070888808127425,
      "grad_norm": 1.0916295051574707,
      "learning_rate": 0.00016385843829466729,
      "loss": 3.4399,
      "step": 193860
    },
    {
      "epoch": 0.18071820969935334,
      "grad_norm": 1.261041283607483,
      "learning_rate": 0.0001638565739725721,
      "loss": 3.461,
      "step": 193870
    },
    {
      "epoch": 0.18072753131743244,
      "grad_norm": 1.4129811525344849,
      "learning_rate": 0.00016385470965047693,
      "loss": 3.4429,
      "step": 193880
    },
    {
      "epoch": 0.18073685293551153,
      "grad_norm": 1.1459614038467407,
      "learning_rate": 0.0001638528453283817,
      "loss": 3.4271,
      "step": 193890
    },
    {
      "epoch": 0.18074617455359063,
      "grad_norm": 1.0303196907043457,
      "learning_rate": 0.0001638509810062865,
      "loss": 3.5013,
      "step": 193900
    },
    {
      "epoch": 0.18075549617166972,
      "grad_norm": 1.0902442932128906,
      "learning_rate": 0.0001638491166841913,
      "loss": 3.5806,
      "step": 193910
    },
    {
      "epoch": 0.18076481778974882,
      "grad_norm": 1.10514235496521,
      "learning_rate": 0.0001638472523620961,
      "loss": 3.6608,
      "step": 193920
    },
    {
      "epoch": 0.1807741394078279,
      "grad_norm": 0.9445947408676147,
      "learning_rate": 0.0001638453880400009,
      "loss": 3.5457,
      "step": 193930
    },
    {
      "epoch": 0.18078346102590698,
      "grad_norm": 1.3388171195983887,
      "learning_rate": 0.0001638435237179057,
      "loss": 3.5984,
      "step": 193940
    },
    {
      "epoch": 0.18079278264398607,
      "grad_norm": 1.1774941682815552,
      "learning_rate": 0.00016384165939581052,
      "loss": 3.4663,
      "step": 193950
    },
    {
      "epoch": 0.18080210426206517,
      "grad_norm": 1.2336724996566772,
      "learning_rate": 0.0001638397950737153,
      "loss": 3.4811,
      "step": 193960
    },
    {
      "epoch": 0.18081142588014426,
      "grad_norm": 1.040162444114685,
      "learning_rate": 0.0001638379307516201,
      "loss": 3.3662,
      "step": 193970
    },
    {
      "epoch": 0.18082074749822336,
      "grad_norm": 1.253139615058899,
      "learning_rate": 0.0001638360664295249,
      "loss": 3.51,
      "step": 193980
    },
    {
      "epoch": 0.18083006911630245,
      "grad_norm": 1.1226284503936768,
      "learning_rate": 0.0001638342021074297,
      "loss": 3.3302,
      "step": 193990
    },
    {
      "epoch": 0.18083939073438154,
      "grad_norm": 1.0937693119049072,
      "learning_rate": 0.00016383233778533452,
      "loss": 3.5069,
      "step": 194000
    },
    {
      "epoch": 0.18084871235246064,
      "grad_norm": 0.9127707481384277,
      "learning_rate": 0.0001638304734632393,
      "loss": 3.6369,
      "step": 194010
    },
    {
      "epoch": 0.18085803397053973,
      "grad_norm": 1.0614888668060303,
      "learning_rate": 0.0001638286091411441,
      "loss": 3.5884,
      "step": 194020
    },
    {
      "epoch": 0.18086735558861883,
      "grad_norm": 0.9648026823997498,
      "learning_rate": 0.0001638267448190489,
      "loss": 3.5681,
      "step": 194030
    },
    {
      "epoch": 0.18087667720669792,
      "grad_norm": 0.8311305642127991,
      "learning_rate": 0.0001638248804969537,
      "loss": 3.6186,
      "step": 194040
    },
    {
      "epoch": 0.180885998824777,
      "grad_norm": 1.2316160202026367,
      "learning_rate": 0.00016382301617485852,
      "loss": 3.6044,
      "step": 194050
    },
    {
      "epoch": 0.18089532044285608,
      "grad_norm": 0.8779131770133972,
      "learning_rate": 0.0001638211518527633,
      "loss": 3.5098,
      "step": 194060
    },
    {
      "epoch": 0.18090464206093518,
      "grad_norm": 1.2859892845153809,
      "learning_rate": 0.0001638192875306681,
      "loss": 3.6401,
      "step": 194070
    },
    {
      "epoch": 0.18091396367901427,
      "grad_norm": 0.9849982261657715,
      "learning_rate": 0.0001638174232085729,
      "loss": 3.5286,
      "step": 194080
    },
    {
      "epoch": 0.18092328529709337,
      "grad_norm": 1.1980085372924805,
      "learning_rate": 0.0001638155588864777,
      "loss": 3.5288,
      "step": 194090
    },
    {
      "epoch": 0.18093260691517246,
      "grad_norm": 0.9717210531234741,
      "learning_rate": 0.0001638136945643825,
      "loss": 3.6254,
      "step": 194100
    },
    {
      "epoch": 0.18094192853325156,
      "grad_norm": 1.1523386240005493,
      "learning_rate": 0.0001638118302422873,
      "loss": 3.6774,
      "step": 194110
    },
    {
      "epoch": 0.18095125015133065,
      "grad_norm": 1.009248971939087,
      "learning_rate": 0.00016380996592019211,
      "loss": 3.473,
      "step": 194120
    },
    {
      "epoch": 0.18096057176940974,
      "grad_norm": 1.0613175630569458,
      "learning_rate": 0.0001638081015980969,
      "loss": 3.565,
      "step": 194130
    },
    {
      "epoch": 0.18096989338748884,
      "grad_norm": 1.0468010902404785,
      "learning_rate": 0.0001638062372760017,
      "loss": 3.5113,
      "step": 194140
    },
    {
      "epoch": 0.1809792150055679,
      "grad_norm": 1.4579991102218628,
      "learning_rate": 0.0001638043729539065,
      "loss": 3.4593,
      "step": 194150
    },
    {
      "epoch": 0.180988536623647,
      "grad_norm": 0.8339740037918091,
      "learning_rate": 0.0001638025086318113,
      "loss": 3.5157,
      "step": 194160
    },
    {
      "epoch": 0.1809978582417261,
      "grad_norm": 0.8607269525527954,
      "learning_rate": 0.00016380064430971612,
      "loss": 3.4414,
      "step": 194170
    },
    {
      "epoch": 0.1810071798598052,
      "grad_norm": 1.1286299228668213,
      "learning_rate": 0.0001637987799876209,
      "loss": 3.4315,
      "step": 194180
    },
    {
      "epoch": 0.18101650147788428,
      "grad_norm": 0.8483546376228333,
      "learning_rate": 0.0001637969156655257,
      "loss": 3.5101,
      "step": 194190
    },
    {
      "epoch": 0.18102582309596338,
      "grad_norm": 1.6278562545776367,
      "learning_rate": 0.0001637950513434305,
      "loss": 3.4172,
      "step": 194200
    },
    {
      "epoch": 0.18103514471404247,
      "grad_norm": 0.8957080841064453,
      "learning_rate": 0.00016379318702133532,
      "loss": 3.446,
      "step": 194210
    },
    {
      "epoch": 0.18104446633212157,
      "grad_norm": 1.2015866041183472,
      "learning_rate": 0.00016379132269924012,
      "loss": 3.4329,
      "step": 194220
    },
    {
      "epoch": 0.18105378795020066,
      "grad_norm": 1.3462183475494385,
      "learning_rate": 0.0001637894583771449,
      "loss": 3.4864,
      "step": 194230
    },
    {
      "epoch": 0.18106310956827976,
      "grad_norm": 1.2429311275482178,
      "learning_rate": 0.0001637875940550497,
      "loss": 3.6171,
      "step": 194240
    },
    {
      "epoch": 0.18107243118635885,
      "grad_norm": 0.961053192615509,
      "learning_rate": 0.00016378572973295451,
      "loss": 3.4456,
      "step": 194250
    },
    {
      "epoch": 0.18108175280443792,
      "grad_norm": 0.9213621020317078,
      "learning_rate": 0.0001637838654108593,
      "loss": 3.5296,
      "step": 194260
    },
    {
      "epoch": 0.181091074422517,
      "grad_norm": 0.9927917122840881,
      "learning_rate": 0.00016378200108876413,
      "loss": 3.4331,
      "step": 194270
    },
    {
      "epoch": 0.1811003960405961,
      "grad_norm": 0.9745336771011353,
      "learning_rate": 0.0001637801367666689,
      "loss": 3.5831,
      "step": 194280
    },
    {
      "epoch": 0.1811097176586752,
      "grad_norm": 1.1017276048660278,
      "learning_rate": 0.0001637782724445737,
      "loss": 3.4377,
      "step": 194290
    },
    {
      "epoch": 0.1811190392767543,
      "grad_norm": 1.1557246446609497,
      "learning_rate": 0.00016377640812247852,
      "loss": 3.7436,
      "step": 194300
    },
    {
      "epoch": 0.1811283608948334,
      "grad_norm": 1.1681463718414307,
      "learning_rate": 0.0001637745438003833,
      "loss": 3.4497,
      "step": 194310
    },
    {
      "epoch": 0.18113768251291248,
      "grad_norm": 0.8986257314682007,
      "learning_rate": 0.0001637726794782881,
      "loss": 3.4557,
      "step": 194320
    },
    {
      "epoch": 0.18114700413099158,
      "grad_norm": 0.9490654468536377,
      "learning_rate": 0.0001637708151561929,
      "loss": 3.4381,
      "step": 194330
    },
    {
      "epoch": 0.18115632574907067,
      "grad_norm": 1.2042531967163086,
      "learning_rate": 0.00016376895083409772,
      "loss": 3.3597,
      "step": 194340
    },
    {
      "epoch": 0.18116564736714977,
      "grad_norm": 1.0406237840652466,
      "learning_rate": 0.00016376708651200252,
      "loss": 3.5711,
      "step": 194350
    },
    {
      "epoch": 0.18117496898522883,
      "grad_norm": 1.005171775817871,
      "learning_rate": 0.0001637652221899073,
      "loss": 3.4042,
      "step": 194360
    },
    {
      "epoch": 0.18118429060330793,
      "grad_norm": 1.3343888521194458,
      "learning_rate": 0.0001637633578678121,
      "loss": 3.5574,
      "step": 194370
    },
    {
      "epoch": 0.18119361222138702,
      "grad_norm": 1.0586998462677002,
      "learning_rate": 0.00016376149354571692,
      "loss": 3.6341,
      "step": 194380
    },
    {
      "epoch": 0.18120293383946612,
      "grad_norm": 0.8854542970657349,
      "learning_rate": 0.00016375962922362172,
      "loss": 3.4358,
      "step": 194390
    },
    {
      "epoch": 0.1812122554575452,
      "grad_norm": 1.0708224773406982,
      "learning_rate": 0.00016375776490152653,
      "loss": 3.334,
      "step": 194400
    },
    {
      "epoch": 0.1812215770756243,
      "grad_norm": 1.0051493644714355,
      "learning_rate": 0.0001637559005794313,
      "loss": 3.5889,
      "step": 194410
    },
    {
      "epoch": 0.1812308986937034,
      "grad_norm": 1.1114075183868408,
      "learning_rate": 0.00016375403625733611,
      "loss": 3.5394,
      "step": 194420
    },
    {
      "epoch": 0.1812402203117825,
      "grad_norm": 1.1431806087493896,
      "learning_rate": 0.0001637521719352409,
      "loss": 3.5717,
      "step": 194430
    },
    {
      "epoch": 0.1812495419298616,
      "grad_norm": 1.0645500421524048,
      "learning_rate": 0.00016375030761314573,
      "loss": 3.5883,
      "step": 194440
    },
    {
      "epoch": 0.18125886354794069,
      "grad_norm": 1.0904479026794434,
      "learning_rate": 0.00016374844329105053,
      "loss": 3.5132,
      "step": 194450
    },
    {
      "epoch": 0.18126818516601978,
      "grad_norm": 1.1098768711090088,
      "learning_rate": 0.0001637465789689553,
      "loss": 3.5789,
      "step": 194460
    },
    {
      "epoch": 0.18127750678409885,
      "grad_norm": 1.4634718894958496,
      "learning_rate": 0.00016374471464686012,
      "loss": 3.5115,
      "step": 194470
    },
    {
      "epoch": 0.18128682840217794,
      "grad_norm": 0.9587938785552979,
      "learning_rate": 0.0001637428503247649,
      "loss": 3.6153,
      "step": 194480
    },
    {
      "epoch": 0.18129615002025704,
      "grad_norm": 1.1825358867645264,
      "learning_rate": 0.0001637409860026697,
      "loss": 3.6509,
      "step": 194490
    },
    {
      "epoch": 0.18130547163833613,
      "grad_norm": 1.089938998222351,
      "learning_rate": 0.00016373912168057454,
      "loss": 3.6019,
      "step": 194500
    },
    {
      "epoch": 0.18131479325641522,
      "grad_norm": 0.9804187417030334,
      "learning_rate": 0.00016373725735847932,
      "loss": 3.3899,
      "step": 194510
    },
    {
      "epoch": 0.18132411487449432,
      "grad_norm": 1.2638863325119019,
      "learning_rate": 0.00016373539303638412,
      "loss": 3.5561,
      "step": 194520
    },
    {
      "epoch": 0.1813334364925734,
      "grad_norm": 1.0449095964431763,
      "learning_rate": 0.0001637335287142889,
      "loss": 3.615,
      "step": 194530
    },
    {
      "epoch": 0.1813427581106525,
      "grad_norm": 1.0353198051452637,
      "learning_rate": 0.0001637316643921937,
      "loss": 3.5577,
      "step": 194540
    },
    {
      "epoch": 0.1813520797287316,
      "grad_norm": 1.0027838945388794,
      "learning_rate": 0.00016372980007009851,
      "loss": 3.7207,
      "step": 194550
    },
    {
      "epoch": 0.1813614013468107,
      "grad_norm": 1.2095608711242676,
      "learning_rate": 0.00016372793574800332,
      "loss": 3.4339,
      "step": 194560
    },
    {
      "epoch": 0.18137072296488976,
      "grad_norm": 1.1448719501495361,
      "learning_rate": 0.00016372607142590813,
      "loss": 3.7604,
      "step": 194570
    },
    {
      "epoch": 0.18138004458296886,
      "grad_norm": 1.1179085969924927,
      "learning_rate": 0.0001637242071038129,
      "loss": 3.4806,
      "step": 194580
    },
    {
      "epoch": 0.18138936620104795,
      "grad_norm": 1.2636759281158447,
      "learning_rate": 0.0001637223427817177,
      "loss": 3.4309,
      "step": 194590
    },
    {
      "epoch": 0.18139868781912705,
      "grad_norm": 0.9281215667724609,
      "learning_rate": 0.00016372047845962252,
      "loss": 3.5007,
      "step": 194600
    },
    {
      "epoch": 0.18140800943720614,
      "grad_norm": 1.0626624822616577,
      "learning_rate": 0.00016371861413752733,
      "loss": 3.5142,
      "step": 194610
    },
    {
      "epoch": 0.18141733105528524,
      "grad_norm": 0.9245609045028687,
      "learning_rate": 0.00016371674981543213,
      "loss": 3.4511,
      "step": 194620
    },
    {
      "epoch": 0.18142665267336433,
      "grad_norm": 0.8225257396697998,
      "learning_rate": 0.0001637148854933369,
      "loss": 3.6658,
      "step": 194630
    },
    {
      "epoch": 0.18143597429144342,
      "grad_norm": 0.9771033525466919,
      "learning_rate": 0.00016371302117124172,
      "loss": 3.4563,
      "step": 194640
    },
    {
      "epoch": 0.18144529590952252,
      "grad_norm": 1.0890297889709473,
      "learning_rate": 0.00016371115684914652,
      "loss": 3.4484,
      "step": 194650
    },
    {
      "epoch": 0.1814546175276016,
      "grad_norm": 1.2319648265838623,
      "learning_rate": 0.00016370929252705133,
      "loss": 3.6215,
      "step": 194660
    },
    {
      "epoch": 0.18146393914568068,
      "grad_norm": 0.9055860042572021,
      "learning_rate": 0.00016370742820495614,
      "loss": 3.5011,
      "step": 194670
    },
    {
      "epoch": 0.18147326076375977,
      "grad_norm": 1.0739535093307495,
      "learning_rate": 0.00016370556388286092,
      "loss": 3.5069,
      "step": 194680
    },
    {
      "epoch": 0.18148258238183887,
      "grad_norm": 1.1613517999649048,
      "learning_rate": 0.00016370369956076572,
      "loss": 3.4593,
      "step": 194690
    },
    {
      "epoch": 0.18149190399991796,
      "grad_norm": 0.9258676171302795,
      "learning_rate": 0.00016370183523867053,
      "loss": 3.5063,
      "step": 194700
    },
    {
      "epoch": 0.18150122561799706,
      "grad_norm": 1.0085172653198242,
      "learning_rate": 0.0001636999709165753,
      "loss": 3.6297,
      "step": 194710
    },
    {
      "epoch": 0.18151054723607615,
      "grad_norm": 0.7795039415359497,
      "learning_rate": 0.00016369810659448014,
      "loss": 3.6812,
      "step": 194720
    },
    {
      "epoch": 0.18151986885415525,
      "grad_norm": 1.2067346572875977,
      "learning_rate": 0.00016369624227238492,
      "loss": 3.4525,
      "step": 194730
    },
    {
      "epoch": 0.18152919047223434,
      "grad_norm": 1.039503574371338,
      "learning_rate": 0.00016369437795028973,
      "loss": 3.4575,
      "step": 194740
    },
    {
      "epoch": 0.18153851209031344,
      "grad_norm": 1.0912810564041138,
      "learning_rate": 0.00016369251362819453,
      "loss": 3.5724,
      "step": 194750
    },
    {
      "epoch": 0.18154783370839253,
      "grad_norm": 1.1493779420852661,
      "learning_rate": 0.0001636906493060993,
      "loss": 3.5239,
      "step": 194760
    },
    {
      "epoch": 0.18155715532647163,
      "grad_norm": 0.8960182070732117,
      "learning_rate": 0.00016368878498400412,
      "loss": 3.3983,
      "step": 194770
    },
    {
      "epoch": 0.1815664769445507,
      "grad_norm": 1.2410894632339478,
      "learning_rate": 0.00016368692066190892,
      "loss": 3.5314,
      "step": 194780
    },
    {
      "epoch": 0.1815757985626298,
      "grad_norm": 1.2068511247634888,
      "learning_rate": 0.00016368505633981373,
      "loss": 3.5607,
      "step": 194790
    },
    {
      "epoch": 0.18158512018070888,
      "grad_norm": 1.0645320415496826,
      "learning_rate": 0.00016368319201771854,
      "loss": 3.4318,
      "step": 194800
    },
    {
      "epoch": 0.18159444179878798,
      "grad_norm": 1.210623025894165,
      "learning_rate": 0.00016368132769562332,
      "loss": 3.5905,
      "step": 194810
    },
    {
      "epoch": 0.18160376341686707,
      "grad_norm": 1.0020495653152466,
      "learning_rate": 0.00016367946337352812,
      "loss": 3.6343,
      "step": 194820
    },
    {
      "epoch": 0.18161308503494616,
      "grad_norm": 0.9349988698959351,
      "learning_rate": 0.00016367759905143293,
      "loss": 3.5365,
      "step": 194830
    },
    {
      "epoch": 0.18162240665302526,
      "grad_norm": 1.15462327003479,
      "learning_rate": 0.00016367573472933773,
      "loss": 3.4662,
      "step": 194840
    },
    {
      "epoch": 0.18163172827110435,
      "grad_norm": 1.1112887859344482,
      "learning_rate": 0.00016367387040724254,
      "loss": 3.5613,
      "step": 194850
    },
    {
      "epoch": 0.18164104988918345,
      "grad_norm": 1.0793308019638062,
      "learning_rate": 0.00016367200608514732,
      "loss": 3.5605,
      "step": 194860
    },
    {
      "epoch": 0.18165037150726254,
      "grad_norm": 1.0437309741973877,
      "learning_rate": 0.00016367014176305213,
      "loss": 3.4471,
      "step": 194870
    },
    {
      "epoch": 0.1816596931253416,
      "grad_norm": 1.015819787979126,
      "learning_rate": 0.0001636682774409569,
      "loss": 3.5876,
      "step": 194880
    },
    {
      "epoch": 0.1816690147434207,
      "grad_norm": 1.3549680709838867,
      "learning_rate": 0.00016366641311886174,
      "loss": 3.546,
      "step": 194890
    },
    {
      "epoch": 0.1816783363614998,
      "grad_norm": 0.9662542343139648,
      "learning_rate": 0.00016366454879676655,
      "loss": 3.5355,
      "step": 194900
    },
    {
      "epoch": 0.1816876579795789,
      "grad_norm": 1.5557950735092163,
      "learning_rate": 0.00016366268447467132,
      "loss": 3.4241,
      "step": 194910
    },
    {
      "epoch": 0.181696979597658,
      "grad_norm": 1.1756783723831177,
      "learning_rate": 0.00016366082015257613,
      "loss": 3.6291,
      "step": 194920
    },
    {
      "epoch": 0.18170630121573708,
      "grad_norm": 1.0491434335708618,
      "learning_rate": 0.0001636589558304809,
      "loss": 3.5759,
      "step": 194930
    },
    {
      "epoch": 0.18171562283381618,
      "grad_norm": 0.9826769232749939,
      "learning_rate": 0.00016365709150838572,
      "loss": 3.3841,
      "step": 194940
    },
    {
      "epoch": 0.18172494445189527,
      "grad_norm": 0.9657946825027466,
      "learning_rate": 0.00016365522718629055,
      "loss": 3.6352,
      "step": 194950
    },
    {
      "epoch": 0.18173426606997437,
      "grad_norm": 1.2727752923965454,
      "learning_rate": 0.00016365336286419533,
      "loss": 3.5862,
      "step": 194960
    },
    {
      "epoch": 0.18174358768805346,
      "grad_norm": 0.8989149928092957,
      "learning_rate": 0.00016365149854210014,
      "loss": 3.427,
      "step": 194970
    },
    {
      "epoch": 0.18175290930613255,
      "grad_norm": 1.0711134672164917,
      "learning_rate": 0.00016364963422000491,
      "loss": 3.5859,
      "step": 194980
    },
    {
      "epoch": 0.18176223092421162,
      "grad_norm": 0.9009993672370911,
      "learning_rate": 0.00016364776989790972,
      "loss": 3.5207,
      "step": 194990
    },
    {
      "epoch": 0.18177155254229072,
      "grad_norm": 1.0656555891036987,
      "learning_rate": 0.00016364590557581453,
      "loss": 3.4816,
      "step": 195000
    },
    {
      "epoch": 1.0000745729446328,
      "grad_norm": 0.38841360807418823,
      "learning_rate": 0.00016364404125371933,
      "loss": 3.4632,
      "step": 195010
    },
    {
      "epoch": 1.0001491458892655,
      "grad_norm": 0.36714524030685425,
      "learning_rate": 0.00016364217693162414,
      "loss": 3.5315,
      "step": 195020
    },
    {
      "epoch": 1.0002237188338983,
      "grad_norm": 0.365043967962265,
      "learning_rate": 0.00016364031260952892,
      "loss": 3.5189,
      "step": 195030
    },
    {
      "epoch": 1.000298291778531,
      "grad_norm": 0.3782459795475006,
      "learning_rate": 0.00016363844828743373,
      "loss": 3.5677,
      "step": 195040
    },
    {
      "epoch": 1.0003728647231638,
      "grad_norm": 0.3651690185070038,
      "learning_rate": 0.00016363658396533853,
      "loss": 3.5219,
      "step": 195050
    },
    {
      "epoch": 1.0004474376677963,
      "grad_norm": 0.37616631388664246,
      "learning_rate": 0.00016363471964324334,
      "loss": 3.5386,
      "step": 195060
    },
    {
      "epoch": 1.000522010612429,
      "grad_norm": 0.3820997476577759,
      "learning_rate": 0.00016363285532114814,
      "loss": 3.5601,
      "step": 195070
    },
    {
      "epoch": 1.0005965835570618,
      "grad_norm": 0.3934790790081024,
      "learning_rate": 0.00016363099099905292,
      "loss": 3.482,
      "step": 195080
    },
    {
      "epoch": 1.0006711565016946,
      "grad_norm": 0.37657883763313293,
      "learning_rate": 0.00016362912667695773,
      "loss": 3.414,
      "step": 195090
    },
    {
      "epoch": 1.0007457294463273,
      "grad_norm": 0.3789973556995392,
      "learning_rate": 0.00016362726235486254,
      "loss": 3.5078,
      "step": 195100
    },
    {
      "epoch": 1.00082030239096,
      "grad_norm": 0.3668755888938904,
      "learning_rate": 0.00016362539803276734,
      "loss": 3.4435,
      "step": 195110
    },
    {
      "epoch": 1.0008948753355928,
      "grad_norm": 0.40987223386764526,
      "learning_rate": 0.00016362353371067215,
      "loss": 3.4946,
      "step": 195120
    },
    {
      "epoch": 1.0009694482802256,
      "grad_norm": 0.38420936465263367,
      "learning_rate": 0.00016362166938857693,
      "loss": 3.5214,
      "step": 195130
    },
    {
      "epoch": 1.0010440212248584,
      "grad_norm": 0.3597366213798523,
      "learning_rate": 0.00016361980506648173,
      "loss": 3.5988,
      "step": 195140
    },
    {
      "epoch": 1.0011185941694911,
      "grad_norm": 0.3692083954811096,
      "learning_rate": 0.00016361794074438654,
      "loss": 3.4642,
      "step": 195150
    },
    {
      "epoch": 1.0011931671141236,
      "grad_norm": 0.3522626757621765,
      "learning_rate": 0.00016361607642229132,
      "loss": 3.4561,
      "step": 195160
    },
    {
      "epoch": 1.0012677400587564,
      "grad_norm": 0.35763218998908997,
      "learning_rate": 0.00016361421210019615,
      "loss": 3.5098,
      "step": 195170
    },
    {
      "epoch": 1.0013423130033892,
      "grad_norm": 0.36342141032218933,
      "learning_rate": 0.00016361234777810093,
      "loss": 3.4978,
      "step": 195180
    },
    {
      "epoch": 1.001416885948022,
      "grad_norm": 0.3807491660118103,
      "learning_rate": 0.00016361048345600574,
      "loss": 3.5209,
      "step": 195190
    },
    {
      "epoch": 1.0014914588926547,
      "grad_norm": 0.350237101316452,
      "learning_rate": 0.00016360861913391055,
      "loss": 3.5167,
      "step": 195200
    },
    {
      "epoch": 1.0015660318372874,
      "grad_norm": 0.7932934165000916,
      "learning_rate": 0.00016360675481181532,
      "loss": 3.5006,
      "step": 195210
    },
    {
      "epoch": 1.0016406047819202,
      "grad_norm": 0.3901043236255646,
      "learning_rate": 0.00016360489048972013,
      "loss": 3.4269,
      "step": 195220
    },
    {
      "epoch": 1.001715177726553,
      "grad_norm": 0.3906938433647156,
      "learning_rate": 0.00016360302616762494,
      "loss": 3.4889,
      "step": 195230
    },
    {
      "epoch": 1.0017897506711857,
      "grad_norm": 0.3710913360118866,
      "learning_rate": 0.00016360116184552974,
      "loss": 3.4344,
      "step": 195240
    },
    {
      "epoch": 1.0018643236158185,
      "grad_norm": 0.3649033308029175,
      "learning_rate": 0.00016359929752343455,
      "loss": 3.487,
      "step": 195250
    },
    {
      "epoch": 1.0019388965604512,
      "grad_norm": 0.3474435806274414,
      "learning_rate": 0.00016359743320133933,
      "loss": 3.4765,
      "step": 195260
    },
    {
      "epoch": 1.0020134695050837,
      "grad_norm": 0.3667939007282257,
      "learning_rate": 0.00016359556887924413,
      "loss": 3.4808,
      "step": 195270
    },
    {
      "epoch": 1.0020880424497165,
      "grad_norm": 0.3677743375301361,
      "learning_rate": 0.00016359370455714894,
      "loss": 3.5041,
      "step": 195280
    },
    {
      "epoch": 1.0021626153943493,
      "grad_norm": 0.3725552558898926,
      "learning_rate": 0.00016359184023505375,
      "loss": 3.5089,
      "step": 195290
    },
    {
      "epoch": 1.002237188338982,
      "grad_norm": 0.37133359909057617,
      "learning_rate": 0.00016358997591295855,
      "loss": 3.4666,
      "step": 195300
    },
    {
      "epoch": 1.0023117612836148,
      "grad_norm": 0.39948204159736633,
      "learning_rate": 0.00016358811159086333,
      "loss": 3.5202,
      "step": 195310
    },
    {
      "epoch": 1.0023863342282475,
      "grad_norm": 0.3885781168937683,
      "learning_rate": 0.00016358624726876814,
      "loss": 3.4618,
      "step": 195320
    },
    {
      "epoch": 1.0024609071728803,
      "grad_norm": 0.412664532661438,
      "learning_rate": 0.00016358438294667292,
      "loss": 3.5281,
      "step": 195330
    },
    {
      "epoch": 1.002535480117513,
      "grad_norm": 0.3778868615627289,
      "learning_rate": 0.00016358251862457775,
      "loss": 3.4243,
      "step": 195340
    },
    {
      "epoch": 1.0026100530621458,
      "grad_norm": 0.35546931624412537,
      "learning_rate": 0.00016358065430248256,
      "loss": 3.5223,
      "step": 195350
    },
    {
      "epoch": 1.0026846260067785,
      "grad_norm": 0.36608079075813293,
      "learning_rate": 0.00016357878998038734,
      "loss": 3.4698,
      "step": 195360
    },
    {
      "epoch": 1.002759198951411,
      "grad_norm": 0.42903268337249756,
      "learning_rate": 0.00016357692565829214,
      "loss": 3.5072,
      "step": 195370
    },
    {
      "epoch": 1.0028337718960438,
      "grad_norm": 0.3416491150856018,
      "learning_rate": 0.00016357506133619692,
      "loss": 3.4629,
      "step": 195380
    },
    {
      "epoch": 1.0029083448406766,
      "grad_norm": 0.3721160590648651,
      "learning_rate": 0.00016357319701410173,
      "loss": 3.4827,
      "step": 195390
    },
    {
      "epoch": 1.0029829177853093,
      "grad_norm": 0.37401875853538513,
      "learning_rate": 0.00016357133269200656,
      "loss": 3.4542,
      "step": 195400
    },
    {
      "epoch": 1.003057490729942,
      "grad_norm": 0.356767475605011,
      "learning_rate": 0.00016356946836991134,
      "loss": 3.4287,
      "step": 195410
    },
    {
      "epoch": 1.0031320636745749,
      "grad_norm": 0.3620913624763489,
      "learning_rate": 0.00016356760404781615,
      "loss": 3.4557,
      "step": 195420
    },
    {
      "epoch": 1.0032066366192076,
      "grad_norm": 0.38758593797683716,
      "learning_rate": 0.00016356573972572093,
      "loss": 3.4778,
      "step": 195430
    },
    {
      "epoch": 1.0032812095638404,
      "grad_norm": 0.3869129717350006,
      "learning_rate": 0.00016356387540362573,
      "loss": 3.4793,
      "step": 195440
    },
    {
      "epoch": 1.0033557825084731,
      "grad_norm": 0.3849250078201294,
      "learning_rate": 0.00016356201108153054,
      "loss": 3.4529,
      "step": 195450
    },
    {
      "epoch": 1.0034303554531059,
      "grad_norm": 0.36489400267601013,
      "learning_rate": 0.00016356014675943535,
      "loss": 3.4377,
      "step": 195460
    },
    {
      "epoch": 1.0035049283977386,
      "grad_norm": 0.3591078519821167,
      "learning_rate": 0.00016355828243734015,
      "loss": 3.4722,
      "step": 195470
    },
    {
      "epoch": 1.0035795013423712,
      "grad_norm": 0.35329967737197876,
      "learning_rate": 0.00016355641811524493,
      "loss": 3.4435,
      "step": 195480
    },
    {
      "epoch": 1.003654074287004,
      "grad_norm": 0.3592555820941925,
      "learning_rate": 0.00016355455379314974,
      "loss": 3.4333,
      "step": 195490
    },
    {
      "epoch": 1.0037286472316367,
      "grad_norm": 0.3444384038448334,
      "learning_rate": 0.00016355268947105454,
      "loss": 3.547,
      "step": 195500
    },
    {
      "epoch": 1.0038032201762694,
      "grad_norm": 0.37479168176651,
      "learning_rate": 0.00016355082514895935,
      "loss": 3.4259,
      "step": 195510
    },
    {
      "epoch": 1.0038777931209022,
      "grad_norm": 0.37193289399147034,
      "learning_rate": 0.00016354896082686416,
      "loss": 3.4434,
      "step": 195520
    },
    {
      "epoch": 1.003952366065535,
      "grad_norm": 0.3725717067718506,
      "learning_rate": 0.00016354709650476894,
      "loss": 3.4791,
      "step": 195530
    },
    {
      "epoch": 1.0040269390101677,
      "grad_norm": 0.3849305212497711,
      "learning_rate": 0.00016354523218267374,
      "loss": 3.4842,
      "step": 195540
    },
    {
      "epoch": 1.0041015119548005,
      "grad_norm": 0.35401690006256104,
      "learning_rate": 0.00016354336786057852,
      "loss": 3.505,
      "step": 195550
    },
    {
      "epoch": 1.0041760848994332,
      "grad_norm": 0.4203287959098816,
      "learning_rate": 0.00016354150353848336,
      "loss": 3.4642,
      "step": 195560
    },
    {
      "epoch": 1.004250657844066,
      "grad_norm": 0.3462972044944763,
      "learning_rate": 0.00016353963921638816,
      "loss": 3.4102,
      "step": 195570
    },
    {
      "epoch": 1.0043252307886985,
      "grad_norm": 0.3425399661064148,
      "learning_rate": 0.00016353777489429294,
      "loss": 3.4172,
      "step": 195580
    },
    {
      "epoch": 1.0043998037333313,
      "grad_norm": 0.3762417733669281,
      "learning_rate": 0.00016353591057219775,
      "loss": 3.4374,
      "step": 195590
    },
    {
      "epoch": 1.004474376677964,
      "grad_norm": 0.40815508365631104,
      "learning_rate": 0.00016353404625010253,
      "loss": 3.422,
      "step": 195600
    },
    {
      "epoch": 1.0045489496225968,
      "grad_norm": 0.39035674929618835,
      "learning_rate": 0.00016353218192800733,
      "loss": 3.498,
      "step": 195610
    },
    {
      "epoch": 1.0046235225672295,
      "grad_norm": 0.3954077661037445,
      "learning_rate": 0.00016353031760591217,
      "loss": 3.3842,
      "step": 195620
    },
    {
      "epoch": 1.0046980955118623,
      "grad_norm": 0.4017237424850464,
      "learning_rate": 0.00016352845328381695,
      "loss": 3.4044,
      "step": 195630
    },
    {
      "epoch": 1.004772668456495,
      "grad_norm": 0.37432581186294556,
      "learning_rate": 0.00016352658896172175,
      "loss": 3.4129,
      "step": 195640
    },
    {
      "epoch": 1.0048472414011278,
      "grad_norm": 0.3662513792514801,
      "learning_rate": 0.00016352472463962653,
      "loss": 3.4259,
      "step": 195650
    },
    {
      "epoch": 1.0049218143457606,
      "grad_norm": 0.4087009131908417,
      "learning_rate": 0.00016352286031753134,
      "loss": 3.4527,
      "step": 195660
    },
    {
      "epoch": 1.0049963872903933,
      "grad_norm": 0.36248746514320374,
      "learning_rate": 0.00016352099599543614,
      "loss": 3.4726,
      "step": 195670
    },
    {
      "epoch": 1.005070960235026,
      "grad_norm": 0.3918258845806122,
      "learning_rate": 0.00016351913167334095,
      "loss": 3.4251,
      "step": 195680
    },
    {
      "epoch": 1.0051455331796586,
      "grad_norm": 0.389498233795166,
      "learning_rate": 0.00016351726735124576,
      "loss": 3.5081,
      "step": 195690
    },
    {
      "epoch": 1.0052201061242914,
      "grad_norm": 0.3534718155860901,
      "learning_rate": 0.00016351540302915054,
      "loss": 3.4459,
      "step": 195700
    },
    {
      "epoch": 1.0052946790689241,
      "grad_norm": 0.34866562485694885,
      "learning_rate": 0.00016351353870705534,
      "loss": 3.4856,
      "step": 195710
    },
    {
      "epoch": 1.0053692520135569,
      "grad_norm": 0.36477962136268616,
      "learning_rate": 0.00016351167438496015,
      "loss": 3.4878,
      "step": 195720
    },
    {
      "epoch": 1.0054438249581896,
      "grad_norm": 0.36393406987190247,
      "learning_rate": 0.00016350981006286495,
      "loss": 3.4474,
      "step": 195730
    },
    {
      "epoch": 1.0055183979028224,
      "grad_norm": 0.39062902331352234,
      "learning_rate": 0.00016350794574076976,
      "loss": 3.4495,
      "step": 195740
    },
    {
      "epoch": 1.0055929708474551,
      "grad_norm": 0.3929741680622101,
      "learning_rate": 0.00016350608141867454,
      "loss": 3.3773,
      "step": 195750
    },
    {
      "epoch": 1.005667543792088,
      "grad_norm": 0.3965471684932709,
      "learning_rate": 0.00016350421709657935,
      "loss": 3.4914,
      "step": 195760
    },
    {
      "epoch": 1.0057421167367206,
      "grad_norm": 0.40405768156051636,
      "learning_rate": 0.00016350235277448415,
      "loss": 3.4656,
      "step": 195770
    },
    {
      "epoch": 1.0058166896813534,
      "grad_norm": 0.37562811374664307,
      "learning_rate": 0.00016350048845238893,
      "loss": 3.405,
      "step": 195780
    },
    {
      "epoch": 1.005891262625986,
      "grad_norm": 0.3752956688404083,
      "learning_rate": 0.00016349862413029376,
      "loss": 3.4435,
      "step": 195790
    },
    {
      "epoch": 1.0059658355706187,
      "grad_norm": 0.34610220789909363,
      "learning_rate": 0.00016349675980819854,
      "loss": 3.4469,
      "step": 195800
    },
    {
      "epoch": 1.0060404085152514,
      "grad_norm": 0.3696129322052002,
      "learning_rate": 0.00016349489548610335,
      "loss": 3.434,
      "step": 195810
    },
    {
      "epoch": 1.0061149814598842,
      "grad_norm": 0.351701021194458,
      "learning_rate": 0.00016349303116400816,
      "loss": 3.4628,
      "step": 195820
    },
    {
      "epoch": 1.006189554404517,
      "grad_norm": 0.3487760126590729,
      "learning_rate": 0.00016349116684191294,
      "loss": 3.4394,
      "step": 195830
    },
    {
      "epoch": 1.0062641273491497,
      "grad_norm": 0.372453898191452,
      "learning_rate": 0.00016348930251981774,
      "loss": 3.5018,
      "step": 195840
    },
    {
      "epoch": 1.0063387002937825,
      "grad_norm": 0.3704719841480255,
      "learning_rate": 0.00016348743819772255,
      "loss": 3.4453,
      "step": 195850
    },
    {
      "epoch": 1.0064132732384152,
      "grad_norm": 0.3809497356414795,
      "learning_rate": 0.00016348557387562735,
      "loss": 3.4875,
      "step": 195860
    },
    {
      "epoch": 1.006487846183048,
      "grad_norm": 0.35781824588775635,
      "learning_rate": 0.00016348370955353216,
      "loss": 3.4352,
      "step": 195870
    },
    {
      "epoch": 1.0065624191276807,
      "grad_norm": 0.362068772315979,
      "learning_rate": 0.00016348184523143694,
      "loss": 3.451,
      "step": 195880
    },
    {
      "epoch": 1.0066369920723133,
      "grad_norm": 0.37503522634506226,
      "learning_rate": 0.00016347998090934175,
      "loss": 3.407,
      "step": 195890
    },
    {
      "epoch": 1.006711565016946,
      "grad_norm": 0.3597732186317444,
      "learning_rate": 0.00016347811658724655,
      "loss": 3.4644,
      "step": 195900
    },
    {
      "epoch": 1.0067861379615788,
      "grad_norm": 0.38214489817619324,
      "learning_rate": 0.00016347625226515136,
      "loss": 3.4349,
      "step": 195910
    },
    {
      "epoch": 1.0068607109062115,
      "grad_norm": 0.38571619987487793,
      "learning_rate": 0.00016347438794305617,
      "loss": 3.4492,
      "step": 195920
    },
    {
      "epoch": 1.0069352838508443,
      "grad_norm": 0.3924594223499298,
      "learning_rate": 0.00016347252362096094,
      "loss": 3.464,
      "step": 195930
    },
    {
      "epoch": 1.007009856795477,
      "grad_norm": 0.3892573118209839,
      "learning_rate": 0.00016347065929886575,
      "loss": 3.405,
      "step": 195940
    },
    {
      "epoch": 1.0070844297401098,
      "grad_norm": 0.3579900860786438,
      "learning_rate": 0.00016346879497677056,
      "loss": 3.4886,
      "step": 195950
    },
    {
      "epoch": 1.0071590026847426,
      "grad_norm": 0.38292354345321655,
      "learning_rate": 0.00016346693065467536,
      "loss": 3.4624,
      "step": 195960
    },
    {
      "epoch": 1.0072335756293753,
      "grad_norm": 0.3814137279987335,
      "learning_rate": 0.00016346506633258017,
      "loss": 3.3701,
      "step": 195970
    },
    {
      "epoch": 1.007308148574008,
      "grad_norm": 0.3514788746833801,
      "learning_rate": 0.00016346320201048495,
      "loss": 3.3786,
      "step": 195980
    },
    {
      "epoch": 1.0073827215186408,
      "grad_norm": 0.4094263017177582,
      "learning_rate": 0.00016346133768838976,
      "loss": 3.432,
      "step": 195990
    },
    {
      "epoch": 1.0074572944632734,
      "grad_norm": 0.35375556349754333,
      "learning_rate": 0.00016345947336629453,
      "loss": 3.4359,
      "step": 196000
    },
    {
      "epoch": 1.0075318674079061,
      "grad_norm": 0.34270891547203064,
      "learning_rate": 0.00016345760904419937,
      "loss": 3.4484,
      "step": 196010
    },
    {
      "epoch": 1.0076064403525389,
      "grad_norm": 0.40772804617881775,
      "learning_rate": 0.00016345574472210417,
      "loss": 3.4074,
      "step": 196020
    },
    {
      "epoch": 1.0076810132971716,
      "grad_norm": 0.3879689574241638,
      "learning_rate": 0.00016345388040000895,
      "loss": 3.5,
      "step": 196030
    },
    {
      "epoch": 1.0077555862418044,
      "grad_norm": 0.6462778449058533,
      "learning_rate": 0.00016345201607791376,
      "loss": 3.388,
      "step": 196040
    },
    {
      "epoch": 1.0078301591864371,
      "grad_norm": 0.4314252436161041,
      "learning_rate": 0.00016345015175581854,
      "loss": 3.4304,
      "step": 196050
    },
    {
      "epoch": 1.00790473213107,
      "grad_norm": 0.37684208154678345,
      "learning_rate": 0.00016344828743372335,
      "loss": 3.3631,
      "step": 196060
    },
    {
      "epoch": 1.0079793050757027,
      "grad_norm": 0.39245733618736267,
      "learning_rate": 0.00016344642311162818,
      "loss": 3.3917,
      "step": 196070
    },
    {
      "epoch": 1.0080538780203354,
      "grad_norm": 0.3683706223964691,
      "learning_rate": 0.00016344455878953296,
      "loss": 3.4403,
      "step": 196080
    },
    {
      "epoch": 1.0081284509649682,
      "grad_norm": 0.38466134667396545,
      "learning_rate": 0.00016344269446743776,
      "loss": 3.4004,
      "step": 196090
    },
    {
      "epoch": 1.0082030239096007,
      "grad_norm": 0.4239516854286194,
      "learning_rate": 0.00016344083014534254,
      "loss": 3.5042,
      "step": 196100
    },
    {
      "epoch": 1.0082775968542335,
      "grad_norm": 0.3772682249546051,
      "learning_rate": 0.00016343896582324735,
      "loss": 3.4273,
      "step": 196110
    },
    {
      "epoch": 1.0083521697988662,
      "grad_norm": 0.3694695234298706,
      "learning_rate": 0.00016343710150115216,
      "loss": 3.4376,
      "step": 196120
    },
    {
      "epoch": 1.008426742743499,
      "grad_norm": 0.3583752512931824,
      "learning_rate": 0.00016343523717905696,
      "loss": 3.4833,
      "step": 196130
    },
    {
      "epoch": 1.0085013156881317,
      "grad_norm": 0.38225311040878296,
      "learning_rate": 0.00016343337285696177,
      "loss": 3.391,
      "step": 196140
    },
    {
      "epoch": 1.0085758886327645,
      "grad_norm": 0.37114614248275757,
      "learning_rate": 0.00016343150853486655,
      "loss": 3.4107,
      "step": 196150
    },
    {
      "epoch": 1.0086504615773972,
      "grad_norm": 0.36642298102378845,
      "learning_rate": 0.00016342964421277135,
      "loss": 3.3924,
      "step": 196160
    },
    {
      "epoch": 1.00872503452203,
      "grad_norm": 0.384994238615036,
      "learning_rate": 0.00016342777989067616,
      "loss": 3.4215,
      "step": 196170
    },
    {
      "epoch": 1.0087996074666628,
      "grad_norm": 0.35330730676651,
      "learning_rate": 0.00016342591556858097,
      "loss": 3.405,
      "step": 196180
    },
    {
      "epoch": 1.0088741804112955,
      "grad_norm": 0.3708997070789337,
      "learning_rate": 0.00016342405124648577,
      "loss": 3.4423,
      "step": 196190
    },
    {
      "epoch": 1.0089487533559283,
      "grad_norm": 0.3891132175922394,
      "learning_rate": 0.00016342218692439055,
      "loss": 3.3931,
      "step": 196200
    },
    {
      "epoch": 1.0090233263005608,
      "grad_norm": 0.3901602327823639,
      "learning_rate": 0.00016342032260229536,
      "loss": 3.3848,
      "step": 196210
    },
    {
      "epoch": 1.0090978992451936,
      "grad_norm": 0.3550076186656952,
      "learning_rate": 0.00016341845828020017,
      "loss": 3.4103,
      "step": 196220
    },
    {
      "epoch": 1.0091724721898263,
      "grad_norm": 0.3743254542350769,
      "learning_rate": 0.00016341659395810494,
      "loss": 3.3859,
      "step": 196230
    },
    {
      "epoch": 1.009247045134459,
      "grad_norm": 0.34419041872024536,
      "learning_rate": 0.00016341472963600978,
      "loss": 3.4605,
      "step": 196240
    },
    {
      "epoch": 1.0093216180790918,
      "grad_norm": 0.34449508786201477,
      "learning_rate": 0.00016341286531391456,
      "loss": 3.4302,
      "step": 196250
    },
    {
      "epoch": 1.0093961910237246,
      "grad_norm": 0.3433244526386261,
      "learning_rate": 0.00016341100099181936,
      "loss": 3.4998,
      "step": 196260
    },
    {
      "epoch": 1.0094707639683573,
      "grad_norm": 0.3653447926044464,
      "learning_rate": 0.00016340913666972417,
      "loss": 3.4252,
      "step": 196270
    },
    {
      "epoch": 1.00954533691299,
      "grad_norm": 0.3955341577529907,
      "learning_rate": 0.00016340727234762895,
      "loss": 3.4403,
      "step": 196280
    },
    {
      "epoch": 1.0096199098576228,
      "grad_norm": 0.36569422483444214,
      "learning_rate": 0.00016340540802553376,
      "loss": 3.3724,
      "step": 196290
    },
    {
      "epoch": 1.0096944828022556,
      "grad_norm": 0.38341137766838074,
      "learning_rate": 0.00016340354370343856,
      "loss": 3.4064,
      "step": 196300
    },
    {
      "epoch": 1.0097690557468881,
      "grad_norm": 0.3861709237098694,
      "learning_rate": 0.00016340167938134337,
      "loss": 3.4387,
      "step": 196310
    },
    {
      "epoch": 1.009843628691521,
      "grad_norm": 0.38319653272628784,
      "learning_rate": 0.00016339981505924817,
      "loss": 3.3647,
      "step": 196320
    },
    {
      "epoch": 1.0099182016361536,
      "grad_norm": 0.3569517135620117,
      "learning_rate": 0.00016339795073715295,
      "loss": 3.4398,
      "step": 196330
    },
    {
      "epoch": 1.0099927745807864,
      "grad_norm": 0.35147514939308167,
      "learning_rate": 0.00016339608641505776,
      "loss": 3.3911,
      "step": 196340
    },
    {
      "epoch": 1.0100673475254192,
      "grad_norm": 0.3931194543838501,
      "learning_rate": 0.00016339422209296257,
      "loss": 3.3825,
      "step": 196350
    },
    {
      "epoch": 1.010141920470052,
      "grad_norm": 0.8105626702308655,
      "learning_rate": 0.00016339235777086737,
      "loss": 3.4085,
      "step": 196360
    },
    {
      "epoch": 1.0102164934146847,
      "grad_norm": 0.36619332432746887,
      "learning_rate": 0.00016339049344877218,
      "loss": 3.4183,
      "step": 196370
    },
    {
      "epoch": 1.0102910663593174,
      "grad_norm": 0.3729841709136963,
      "learning_rate": 0.00016338862912667696,
      "loss": 3.339,
      "step": 196380
    },
    {
      "epoch": 1.0103656393039502,
      "grad_norm": 0.3629412055015564,
      "learning_rate": 0.00016338676480458176,
      "loss": 3.4123,
      "step": 196390
    },
    {
      "epoch": 1.010440212248583,
      "grad_norm": 0.36481693387031555,
      "learning_rate": 0.00016338490048248657,
      "loss": 3.4335,
      "step": 196400
    },
    {
      "epoch": 1.0105147851932157,
      "grad_norm": 0.3736242949962616,
      "learning_rate": 0.00016338303616039138,
      "loss": 3.3368,
      "step": 196410
    },
    {
      "epoch": 1.0105893581378482,
      "grad_norm": 0.35640040040016174,
      "learning_rate": 0.00016338117183829618,
      "loss": 3.38,
      "step": 196420
    },
    {
      "epoch": 1.010663931082481,
      "grad_norm": 0.36305123567581177,
      "learning_rate": 0.00016337930751620096,
      "loss": 3.4307,
      "step": 196430
    },
    {
      "epoch": 1.0107385040271137,
      "grad_norm": 0.37641629576683044,
      "learning_rate": 0.00016337744319410577,
      "loss": 3.3915,
      "step": 196440
    },
    {
      "epoch": 1.0108130769717465,
      "grad_norm": 0.3748375177383423,
      "learning_rate": 0.00016337557887201055,
      "loss": 3.3829,
      "step": 196450
    },
    {
      "epoch": 1.0108876499163792,
      "grad_norm": 0.36433112621307373,
      "learning_rate": 0.00016337371454991538,
      "loss": 3.4051,
      "step": 196460
    },
    {
      "epoch": 1.010962222861012,
      "grad_norm": 0.37671855092048645,
      "learning_rate": 0.0001633718502278202,
      "loss": 3.4569,
      "step": 196470
    },
    {
      "epoch": 1.0110367958056448,
      "grad_norm": 0.373914897441864,
      "learning_rate": 0.00016336998590572497,
      "loss": 3.4636,
      "step": 196480
    },
    {
      "epoch": 1.0111113687502775,
      "grad_norm": 0.4113190472126007,
      "learning_rate": 0.00016336812158362977,
      "loss": 3.4403,
      "step": 196490
    },
    {
      "epoch": 1.0111859416949103,
      "grad_norm": 0.36872193217277527,
      "learning_rate": 0.00016336625726153455,
      "loss": 3.4464,
      "step": 196500
    },
    {
      "epoch": 1.011260514639543,
      "grad_norm": 0.3684009909629822,
      "learning_rate": 0.00016336439293943936,
      "loss": 3.4496,
      "step": 196510
    },
    {
      "epoch": 1.0113350875841756,
      "grad_norm": 0.39193353056907654,
      "learning_rate": 0.0001633625286173442,
      "loss": 3.4338,
      "step": 196520
    },
    {
      "epoch": 1.0114096605288083,
      "grad_norm": 0.3647139072418213,
      "learning_rate": 0.00016336066429524897,
      "loss": 3.4049,
      "step": 196530
    },
    {
      "epoch": 1.011484233473441,
      "grad_norm": 0.3802970349788666,
      "learning_rate": 0.00016335879997315378,
      "loss": 3.4883,
      "step": 196540
    },
    {
      "epoch": 1.0115588064180738,
      "grad_norm": 0.38056665658950806,
      "learning_rate": 0.00016335693565105856,
      "loss": 3.4025,
      "step": 196550
    },
    {
      "epoch": 1.0116333793627066,
      "grad_norm": 0.4069201648235321,
      "learning_rate": 0.00016335507132896336,
      "loss": 3.3716,
      "step": 196560
    },
    {
      "epoch": 1.0117079523073393,
      "grad_norm": 0.38483142852783203,
      "learning_rate": 0.00016335320700686817,
      "loss": 3.4295,
      "step": 196570
    },
    {
      "epoch": 1.011782525251972,
      "grad_norm": 0.3664044141769409,
      "learning_rate": 0.00016335134268477298,
      "loss": 3.4094,
      "step": 196580
    },
    {
      "epoch": 1.0118570981966049,
      "grad_norm": 0.38713279366493225,
      "learning_rate": 0.00016334947836267778,
      "loss": 3.3829,
      "step": 196590
    },
    {
      "epoch": 1.0119316711412376,
      "grad_norm": 0.37715256214141846,
      "learning_rate": 0.00016334761404058256,
      "loss": 3.4868,
      "step": 196600
    },
    {
      "epoch": 1.0120062440858704,
      "grad_norm": 0.5184134840965271,
      "learning_rate": 0.00016334574971848737,
      "loss": 3.4241,
      "step": 196610
    },
    {
      "epoch": 1.0120808170305031,
      "grad_norm": 0.38544124364852905,
      "learning_rate": 0.00016334388539639217,
      "loss": 3.4272,
      "step": 196620
    },
    {
      "epoch": 1.0121553899751357,
      "grad_norm": 0.3798981308937073,
      "learning_rate": 0.00016334202107429698,
      "loss": 3.4907,
      "step": 196630
    },
    {
      "epoch": 1.0122299629197684,
      "grad_norm": 0.4002630114555359,
      "learning_rate": 0.00016334015675220179,
      "loss": 3.4294,
      "step": 196640
    },
    {
      "epoch": 1.0123045358644012,
      "grad_norm": 0.36376169323921204,
      "learning_rate": 0.00016333829243010657,
      "loss": 3.477,
      "step": 196650
    },
    {
      "epoch": 1.012379108809034,
      "grad_norm": 0.3498413860797882,
      "learning_rate": 0.00016333642810801137,
      "loss": 3.3905,
      "step": 196660
    },
    {
      "epoch": 1.0124536817536667,
      "grad_norm": 0.3714005947113037,
      "learning_rate": 0.00016333456378591618,
      "loss": 3.4242,
      "step": 196670
    },
    {
      "epoch": 1.0125282546982994,
      "grad_norm": 0.3616548478603363,
      "learning_rate": 0.00016333269946382096,
      "loss": 3.4097,
      "step": 196680
    },
    {
      "epoch": 1.0126028276429322,
      "grad_norm": 0.3813785910606384,
      "learning_rate": 0.0001633308351417258,
      "loss": 3.4606,
      "step": 196690
    },
    {
      "epoch": 1.012677400587565,
      "grad_norm": 0.3627088963985443,
      "learning_rate": 0.00016332897081963057,
      "loss": 3.4316,
      "step": 196700
    },
    {
      "epoch": 1.0127519735321977,
      "grad_norm": 0.3811390995979309,
      "learning_rate": 0.00016332710649753538,
      "loss": 3.5042,
      "step": 196710
    },
    {
      "epoch": 1.0128265464768305,
      "grad_norm": 0.3732743263244629,
      "learning_rate": 0.00016332524217544018,
      "loss": 3.4124,
      "step": 196720
    },
    {
      "epoch": 1.012901119421463,
      "grad_norm": 0.4095469117164612,
      "learning_rate": 0.00016332337785334496,
      "loss": 3.3691,
      "step": 196730
    },
    {
      "epoch": 1.0129756923660957,
      "grad_norm": 0.3772827684879303,
      "learning_rate": 0.00016332151353124977,
      "loss": 3.5088,
      "step": 196740
    },
    {
      "epoch": 1.0130502653107285,
      "grad_norm": 0.36212313175201416,
      "learning_rate": 0.00016331964920915457,
      "loss": 3.463,
      "step": 196750
    },
    {
      "epoch": 1.0131248382553613,
      "grad_norm": 0.35286980867385864,
      "learning_rate": 0.00016331778488705938,
      "loss": 3.4604,
      "step": 196760
    },
    {
      "epoch": 1.013199411199994,
      "grad_norm": 0.3578338623046875,
      "learning_rate": 0.0001633159205649642,
      "loss": 3.3859,
      "step": 196770
    },
    {
      "epoch": 1.0132739841446268,
      "grad_norm": 0.38209235668182373,
      "learning_rate": 0.00016331405624286897,
      "loss": 3.4297,
      "step": 196780
    },
    {
      "epoch": 1.0133485570892595,
      "grad_norm": 0.4012892246246338,
      "learning_rate": 0.00016331219192077377,
      "loss": 3.4392,
      "step": 196790
    },
    {
      "epoch": 1.0134231300338923,
      "grad_norm": 0.3862162232398987,
      "learning_rate": 0.00016331032759867858,
      "loss": 3.3929,
      "step": 196800
    },
    {
      "epoch": 1.013497702978525,
      "grad_norm": 0.3546280562877655,
      "learning_rate": 0.00016330846327658338,
      "loss": 3.4133,
      "step": 196810
    },
    {
      "epoch": 1.0135722759231578,
      "grad_norm": 0.36144164204597473,
      "learning_rate": 0.00016330659895448816,
      "loss": 3.4606,
      "step": 196820
    },
    {
      "epoch": 1.0136468488677906,
      "grad_norm": 0.3643339276313782,
      "learning_rate": 0.00016330473463239297,
      "loss": 3.4124,
      "step": 196830
    },
    {
      "epoch": 1.013721421812423,
      "grad_norm": 0.39038214087486267,
      "learning_rate": 0.00016330287031029778,
      "loss": 3.4302,
      "step": 196840
    },
    {
      "epoch": 1.0137959947570558,
      "grad_norm": 0.35982343554496765,
      "learning_rate": 0.00016330100598820258,
      "loss": 3.4671,
      "step": 196850
    },
    {
      "epoch": 1.0138705677016886,
      "grad_norm": 0.36323586106300354,
      "learning_rate": 0.0001632991416661074,
      "loss": 3.391,
      "step": 196860
    },
    {
      "epoch": 1.0139451406463214,
      "grad_norm": 0.38563695549964905,
      "learning_rate": 0.00016329727734401217,
      "loss": 3.4092,
      "step": 196870
    },
    {
      "epoch": 1.014019713590954,
      "grad_norm": 0.3727576732635498,
      "learning_rate": 0.00016329541302191697,
      "loss": 3.396,
      "step": 196880
    },
    {
      "epoch": 1.0140942865355869,
      "grad_norm": 0.38777968287467957,
      "learning_rate": 0.00016329354869982178,
      "loss": 3.4245,
      "step": 196890
    },
    {
      "epoch": 1.0141688594802196,
      "grad_norm": 0.37046802043914795,
      "learning_rate": 0.00016329168437772656,
      "loss": 3.4347,
      "step": 196900
    },
    {
      "epoch": 1.0142434324248524,
      "grad_norm": 0.364579975605011,
      "learning_rate": 0.0001632898200556314,
      "loss": 3.4043,
      "step": 196910
    },
    {
      "epoch": 1.0143180053694851,
      "grad_norm": 0.3474312722682953,
      "learning_rate": 0.00016328795573353617,
      "loss": 3.3866,
      "step": 196920
    },
    {
      "epoch": 1.0143925783141179,
      "grad_norm": 0.37041258811950684,
      "learning_rate": 0.00016328609141144098,
      "loss": 3.4407,
      "step": 196930
    },
    {
      "epoch": 1.0144671512587504,
      "grad_norm": 0.38461440801620483,
      "learning_rate": 0.00016328422708934579,
      "loss": 3.4181,
      "step": 196940
    },
    {
      "epoch": 1.0145417242033832,
      "grad_norm": 0.36290159821510315,
      "learning_rate": 0.00016328236276725056,
      "loss": 3.4569,
      "step": 196950
    },
    {
      "epoch": 1.014616297148016,
      "grad_norm": 0.3718505799770355,
      "learning_rate": 0.00016328049844515537,
      "loss": 3.4306,
      "step": 196960
    },
    {
      "epoch": 1.0146908700926487,
      "grad_norm": 0.3738342523574829,
      "learning_rate": 0.00016327863412306018,
      "loss": 3.4179,
      "step": 196970
    },
    {
      "epoch": 1.0147654430372814,
      "grad_norm": 0.35632944107055664,
      "learning_rate": 0.00016327676980096498,
      "loss": 3.4178,
      "step": 196980
    },
    {
      "epoch": 1.0148400159819142,
      "grad_norm": 0.44073474407196045,
      "learning_rate": 0.0001632749054788698,
      "loss": 3.4164,
      "step": 196990
    },
    {
      "epoch": 1.014914588926547,
      "grad_norm": 0.383297860622406,
      "learning_rate": 0.00016327304115677457,
      "loss": 3.4706,
      "step": 197000
    },
    {
      "epoch": 1.0149891618711797,
      "grad_norm": 0.37295326590538025,
      "learning_rate": 0.00016327117683467938,
      "loss": 3.4495,
      "step": 197010
    },
    {
      "epoch": 1.0150637348158125,
      "grad_norm": 0.34487053751945496,
      "learning_rate": 0.00016326931251258418,
      "loss": 3.4188,
      "step": 197020
    },
    {
      "epoch": 1.0151383077604452,
      "grad_norm": 0.4301530122756958,
      "learning_rate": 0.000163267448190489,
      "loss": 3.4116,
      "step": 197030
    },
    {
      "epoch": 1.0152128807050778,
      "grad_norm": 0.37521588802337646,
      "learning_rate": 0.0001632655838683938,
      "loss": 3.4661,
      "step": 197040
    },
    {
      "epoch": 1.0152874536497105,
      "grad_norm": 0.4134614169597626,
      "learning_rate": 0.00016326371954629857,
      "loss": 3.37,
      "step": 197050
    },
    {
      "epoch": 1.0153620265943433,
      "grad_norm": 0.4180576801300049,
      "learning_rate": 0.00016326185522420338,
      "loss": 3.4087,
      "step": 197060
    },
    {
      "epoch": 1.015436599538976,
      "grad_norm": 0.39539602398872375,
      "learning_rate": 0.00016325999090210819,
      "loss": 3.4951,
      "step": 197070
    },
    {
      "epoch": 1.0155111724836088,
      "grad_norm": 0.36748597025871277,
      "learning_rate": 0.000163258126580013,
      "loss": 3.4198,
      "step": 197080
    },
    {
      "epoch": 1.0155857454282415,
      "grad_norm": 0.4030238389968872,
      "learning_rate": 0.0001632562622579178,
      "loss": 3.4389,
      "step": 197090
    },
    {
      "epoch": 1.0156603183728743,
      "grad_norm": 0.3923053741455078,
      "learning_rate": 0.00016325439793582258,
      "loss": 3.4797,
      "step": 197100
    },
    {
      "epoch": 1.015734891317507,
      "grad_norm": 0.3593278229236603,
      "learning_rate": 0.00016325253361372738,
      "loss": 3.4271,
      "step": 197110
    },
    {
      "epoch": 1.0158094642621398,
      "grad_norm": 0.3471822738647461,
      "learning_rate": 0.00016325066929163216,
      "loss": 3.3082,
      "step": 197120
    },
    {
      "epoch": 1.0158840372067726,
      "grad_norm": 0.3632904887199402,
      "learning_rate": 0.000163248804969537,
      "loss": 3.4259,
      "step": 197130
    },
    {
      "epoch": 1.0159586101514053,
      "grad_norm": 0.3780371844768524,
      "learning_rate": 0.0001632469406474418,
      "loss": 3.4176,
      "step": 197140
    },
    {
      "epoch": 1.0160331830960379,
      "grad_norm": 0.3851957619190216,
      "learning_rate": 0.00016324507632534658,
      "loss": 3.3741,
      "step": 197150
    },
    {
      "epoch": 1.0161077560406706,
      "grad_norm": 0.39722099900245667,
      "learning_rate": 0.0001632432120032514,
      "loss": 3.4206,
      "step": 197160
    },
    {
      "epoch": 1.0161823289853034,
      "grad_norm": 0.3897550702095032,
      "learning_rate": 0.00016324134768115617,
      "loss": 3.3932,
      "step": 197170
    },
    {
      "epoch": 1.0162569019299361,
      "grad_norm": 0.360088050365448,
      "learning_rate": 0.00016323948335906097,
      "loss": 3.446,
      "step": 197180
    },
    {
      "epoch": 1.0163314748745689,
      "grad_norm": 0.38043302297592163,
      "learning_rate": 0.00016323761903696578,
      "loss": 3.4236,
      "step": 197190
    },
    {
      "epoch": 1.0164060478192016,
      "grad_norm": 0.35140854120254517,
      "learning_rate": 0.0001632357547148706,
      "loss": 3.3983,
      "step": 197200
    },
    {
      "epoch": 1.0164806207638344,
      "grad_norm": 0.36632004380226135,
      "learning_rate": 0.0001632338903927754,
      "loss": 3.4922,
      "step": 197210
    },
    {
      "epoch": 1.0165551937084671,
      "grad_norm": 0.4021988809108734,
      "learning_rate": 0.00016323202607068017,
      "loss": 3.3874,
      "step": 197220
    },
    {
      "epoch": 1.0166297666531,
      "grad_norm": 0.36447641253471375,
      "learning_rate": 0.00016323016174858498,
      "loss": 3.4128,
      "step": 197230
    },
    {
      "epoch": 1.0167043395977327,
      "grad_norm": 0.37366965413093567,
      "learning_rate": 0.00016322829742648979,
      "loss": 3.4447,
      "step": 197240
    },
    {
      "epoch": 1.0167789125423652,
      "grad_norm": 0.3766860067844391,
      "learning_rate": 0.0001632264331043946,
      "loss": 3.4864,
      "step": 197250
    },
    {
      "epoch": 1.016853485486998,
      "grad_norm": 0.3636392652988434,
      "learning_rate": 0.0001632245687822994,
      "loss": 3.4123,
      "step": 197260
    },
    {
      "epoch": 1.0169280584316307,
      "grad_norm": 0.38728511333465576,
      "learning_rate": 0.00016322270446020418,
      "loss": 3.4266,
      "step": 197270
    },
    {
      "epoch": 1.0170026313762635,
      "grad_norm": 0.40538129210472107,
      "learning_rate": 0.00016322084013810898,
      "loss": 3.4242,
      "step": 197280
    },
    {
      "epoch": 1.0170772043208962,
      "grad_norm": 0.3713909387588501,
      "learning_rate": 0.0001632189758160138,
      "loss": 3.387,
      "step": 197290
    },
    {
      "epoch": 1.017151777265529,
      "grad_norm": 0.3641264736652374,
      "learning_rate": 0.0001632171114939186,
      "loss": 3.4267,
      "step": 197300
    },
    {
      "epoch": 1.0172263502101617,
      "grad_norm": 0.38122281432151794,
      "learning_rate": 0.0001632152471718234,
      "loss": 3.4562,
      "step": 197310
    },
    {
      "epoch": 1.0173009231547945,
      "grad_norm": 0.3727692663669586,
      "learning_rate": 0.00016321338284972818,
      "loss": 3.4001,
      "step": 197320
    },
    {
      "epoch": 1.0173754960994272,
      "grad_norm": 0.3875269889831543,
      "learning_rate": 0.000163211518527633,
      "loss": 3.3962,
      "step": 197330
    },
    {
      "epoch": 1.01745006904406,
      "grad_norm": 0.38915205001831055,
      "learning_rate": 0.0001632096542055378,
      "loss": 3.3757,
      "step": 197340
    },
    {
      "epoch": 1.0175246419886927,
      "grad_norm": 0.3930816948413849,
      "learning_rate": 0.00016320778988344257,
      "loss": 3.3918,
      "step": 197350
    },
    {
      "epoch": 1.0175992149333253,
      "grad_norm": 0.35093191266059875,
      "learning_rate": 0.0001632059255613474,
      "loss": 3.4126,
      "step": 197360
    },
    {
      "epoch": 1.017673787877958,
      "grad_norm": 0.3736390471458435,
      "learning_rate": 0.00016320406123925219,
      "loss": 3.3875,
      "step": 197370
    },
    {
      "epoch": 1.0177483608225908,
      "grad_norm": 0.38483676314353943,
      "learning_rate": 0.000163202196917157,
      "loss": 3.4509,
      "step": 197380
    },
    {
      "epoch": 1.0178229337672235,
      "grad_norm": 0.3875139057636261,
      "learning_rate": 0.0001632003325950618,
      "loss": 3.394,
      "step": 197390
    },
    {
      "epoch": 1.0178975067118563,
      "grad_norm": 0.3931399881839752,
      "learning_rate": 0.00016319846827296658,
      "loss": 3.3395,
      "step": 197400
    },
    {
      "epoch": 1.017972079656489,
      "grad_norm": 0.3661401867866516,
      "learning_rate": 0.00016319660395087138,
      "loss": 3.3837,
      "step": 197410
    },
    {
      "epoch": 1.0180466526011218,
      "grad_norm": 0.38518592715263367,
      "learning_rate": 0.0001631947396287762,
      "loss": 3.4306,
      "step": 197420
    },
    {
      "epoch": 1.0181212255457546,
      "grad_norm": 0.37755656242370605,
      "learning_rate": 0.000163192875306681,
      "loss": 3.389,
      "step": 197430
    },
    {
      "epoch": 1.0181957984903873,
      "grad_norm": 0.3658982217311859,
      "learning_rate": 0.0001631910109845858,
      "loss": 3.3041,
      "step": 197440
    },
    {
      "epoch": 1.01827037143502,
      "grad_norm": 0.3359656035900116,
      "learning_rate": 0.00016318914666249058,
      "loss": 3.4451,
      "step": 197450
    },
    {
      "epoch": 1.0183449443796526,
      "grad_norm": 0.36165040731430054,
      "learning_rate": 0.0001631872823403954,
      "loss": 3.3635,
      "step": 197460
    },
    {
      "epoch": 1.0184195173242854,
      "grad_norm": 0.4015321135520935,
      "learning_rate": 0.0001631854180183002,
      "loss": 3.4731,
      "step": 197470
    },
    {
      "epoch": 1.0184940902689181,
      "grad_norm": 0.37044018507003784,
      "learning_rate": 0.000163183553696205,
      "loss": 3.4208,
      "step": 197480
    },
    {
      "epoch": 1.0185686632135509,
      "grad_norm": 0.3893027603626251,
      "learning_rate": 0.0001631816893741098,
      "loss": 3.3749,
      "step": 197490
    },
    {
      "epoch": 1.0186432361581836,
      "grad_norm": 0.39362987875938416,
      "learning_rate": 0.0001631798250520146,
      "loss": 3.3634,
      "step": 197500
    },
    {
      "epoch": 1.0187178091028164,
      "grad_norm": 0.3944653868675232,
      "learning_rate": 0.0001631779607299194,
      "loss": 3.3608,
      "step": 197510
    },
    {
      "epoch": 1.0187923820474492,
      "grad_norm": 0.3869660496711731,
      "learning_rate": 0.0001631760964078242,
      "loss": 3.4087,
      "step": 197520
    },
    {
      "epoch": 1.018866954992082,
      "grad_norm": 0.3599407970905304,
      "learning_rate": 0.000163174232085729,
      "loss": 3.4362,
      "step": 197530
    },
    {
      "epoch": 1.0189415279367147,
      "grad_norm": 0.3477996289730072,
      "learning_rate": 0.0001631723677636338,
      "loss": 3.4329,
      "step": 197540
    },
    {
      "epoch": 1.0190161008813474,
      "grad_norm": 0.39467132091522217,
      "learning_rate": 0.0001631705034415386,
      "loss": 3.4012,
      "step": 197550
    },
    {
      "epoch": 1.0190906738259802,
      "grad_norm": 0.39672791957855225,
      "learning_rate": 0.0001631686391194434,
      "loss": 3.3945,
      "step": 197560
    },
    {
      "epoch": 1.0191652467706127,
      "grad_norm": 0.3581332862377167,
      "learning_rate": 0.00016316677479734818,
      "loss": 3.4325,
      "step": 197570
    },
    {
      "epoch": 1.0192398197152455,
      "grad_norm": 0.3584575057029724,
      "learning_rate": 0.000163164910475253,
      "loss": 3.4347,
      "step": 197580
    },
    {
      "epoch": 1.0193143926598782,
      "grad_norm": 0.3755464255809784,
      "learning_rate": 0.00016316304615315782,
      "loss": 3.3889,
      "step": 197590
    },
    {
      "epoch": 1.019388965604511,
      "grad_norm": 0.3650897741317749,
      "learning_rate": 0.0001631611818310626,
      "loss": 3.4073,
      "step": 197600
    },
    {
      "epoch": 1.0194635385491437,
      "grad_norm": 0.3603858947753906,
      "learning_rate": 0.0001631593175089674,
      "loss": 3.3717,
      "step": 197610
    },
    {
      "epoch": 1.0195381114937765,
      "grad_norm": 0.41383418440818787,
      "learning_rate": 0.00016315745318687218,
      "loss": 3.3514,
      "step": 197620
    },
    {
      "epoch": 1.0196126844384092,
      "grad_norm": 0.3536283075809479,
      "learning_rate": 0.000163155588864777,
      "loss": 3.4048,
      "step": 197630
    },
    {
      "epoch": 1.019687257383042,
      "grad_norm": 0.3759193420410156,
      "learning_rate": 0.00016315372454268182,
      "loss": 3.4309,
      "step": 197640
    },
    {
      "epoch": 1.0197618303276748,
      "grad_norm": 0.33954572677612305,
      "learning_rate": 0.0001631518602205866,
      "loss": 3.4022,
      "step": 197650
    },
    {
      "epoch": 1.0198364032723075,
      "grad_norm": 0.3966703414916992,
      "learning_rate": 0.0001631499958984914,
      "loss": 3.3924,
      "step": 197660
    },
    {
      "epoch": 1.01991097621694,
      "grad_norm": 0.3748270571231842,
      "learning_rate": 0.00016314813157639619,
      "loss": 3.4682,
      "step": 197670
    },
    {
      "epoch": 1.0199855491615728,
      "grad_norm": 0.37637266516685486,
      "learning_rate": 0.000163146267254301,
      "loss": 3.3987,
      "step": 197680
    },
    {
      "epoch": 1.0200601221062056,
      "grad_norm": 0.39798328280448914,
      "learning_rate": 0.0001631444029322058,
      "loss": 3.3823,
      "step": 197690
    },
    {
      "epoch": 1.0201346950508383,
      "grad_norm": 0.3938486874103546,
      "learning_rate": 0.0001631425386101106,
      "loss": 3.3714,
      "step": 197700
    },
    {
      "epoch": 1.020209267995471,
      "grad_norm": 0.37055835127830505,
      "learning_rate": 0.0001631406742880154,
      "loss": 3.3831,
      "step": 197710
    },
    {
      "epoch": 1.0202838409401038,
      "grad_norm": 0.3425584137439728,
      "learning_rate": 0.0001631388099659202,
      "loss": 3.4494,
      "step": 197720
    },
    {
      "epoch": 1.0203584138847366,
      "grad_norm": 0.35690605640411377,
      "learning_rate": 0.000163136945643825,
      "loss": 3.4095,
      "step": 197730
    },
    {
      "epoch": 1.0204329868293693,
      "grad_norm": 0.3823191225528717,
      "learning_rate": 0.0001631350813217298,
      "loss": 3.3862,
      "step": 197740
    },
    {
      "epoch": 1.020507559774002,
      "grad_norm": 0.364280104637146,
      "learning_rate": 0.0001631332169996346,
      "loss": 3.3871,
      "step": 197750
    },
    {
      "epoch": 1.0205821327186348,
      "grad_norm": 0.37082675099372864,
      "learning_rate": 0.00016313135267753942,
      "loss": 3.4478,
      "step": 197760
    },
    {
      "epoch": 1.0206567056632676,
      "grad_norm": 0.35443824529647827,
      "learning_rate": 0.0001631294883554442,
      "loss": 3.4096,
      "step": 197770
    },
    {
      "epoch": 1.0207312786079001,
      "grad_norm": 0.3631582260131836,
      "learning_rate": 0.000163127624033349,
      "loss": 3.4329,
      "step": 197780
    },
    {
      "epoch": 1.020805851552533,
      "grad_norm": 0.3585764765739441,
      "learning_rate": 0.0001631257597112538,
      "loss": 3.4067,
      "step": 197790
    },
    {
      "epoch": 1.0208804244971657,
      "grad_norm": 0.37136444449424744,
      "learning_rate": 0.00016312389538915859,
      "loss": 3.4054,
      "step": 197800
    },
    {
      "epoch": 1.0209549974417984,
      "grad_norm": 0.39475032687187195,
      "learning_rate": 0.00016312203106706342,
      "loss": 3.3963,
      "step": 197810
    },
    {
      "epoch": 1.0210295703864312,
      "grad_norm": 0.33708760142326355,
      "learning_rate": 0.0001631201667449682,
      "loss": 3.4206,
      "step": 197820
    },
    {
      "epoch": 1.021104143331064,
      "grad_norm": 0.3611065149307251,
      "learning_rate": 0.000163118302422873,
      "loss": 3.3656,
      "step": 197830
    },
    {
      "epoch": 1.0211787162756967,
      "grad_norm": 0.3983652889728546,
      "learning_rate": 0.0001631164381007778,
      "loss": 3.3845,
      "step": 197840
    },
    {
      "epoch": 1.0212532892203294,
      "grad_norm": 0.39630764722824097,
      "learning_rate": 0.0001631145737786826,
      "loss": 3.3409,
      "step": 197850
    },
    {
      "epoch": 1.0213278621649622,
      "grad_norm": 0.3820835053920746,
      "learning_rate": 0.0001631127094565874,
      "loss": 3.4824,
      "step": 197860
    },
    {
      "epoch": 1.021402435109595,
      "grad_norm": 0.3761790990829468,
      "learning_rate": 0.0001631108451344922,
      "loss": 3.4142,
      "step": 197870
    },
    {
      "epoch": 1.0214770080542275,
      "grad_norm": 0.37512290477752686,
      "learning_rate": 0.000163108980812397,
      "loss": 3.3439,
      "step": 197880
    },
    {
      "epoch": 1.0215515809988602,
      "grad_norm": 0.37728428840637207,
      "learning_rate": 0.00016310711649030182,
      "loss": 3.4697,
      "step": 197890
    },
    {
      "epoch": 1.021626153943493,
      "grad_norm": 0.362946480512619,
      "learning_rate": 0.0001631052521682066,
      "loss": 3.3782,
      "step": 197900
    },
    {
      "epoch": 1.0217007268881257,
      "grad_norm": 0.3682972490787506,
      "learning_rate": 0.0001631033878461114,
      "loss": 3.3971,
      "step": 197910
    },
    {
      "epoch": 1.0217752998327585,
      "grad_norm": 0.37472325563430786,
      "learning_rate": 0.0001631015235240162,
      "loss": 3.3823,
      "step": 197920
    },
    {
      "epoch": 1.0218498727773913,
      "grad_norm": 0.3518153727054596,
      "learning_rate": 0.00016309965920192101,
      "loss": 3.4165,
      "step": 197930
    },
    {
      "epoch": 1.021924445722024,
      "grad_norm": 0.3739624321460724,
      "learning_rate": 0.00016309779487982582,
      "loss": 3.4474,
      "step": 197940
    },
    {
      "epoch": 1.0219990186666568,
      "grad_norm": 0.35775116086006165,
      "learning_rate": 0.0001630959305577306,
      "loss": 3.4167,
      "step": 197950
    },
    {
      "epoch": 1.0220735916112895,
      "grad_norm": 0.3646295964717865,
      "learning_rate": 0.0001630940662356354,
      "loss": 3.3927,
      "step": 197960
    },
    {
      "epoch": 1.0221481645559223,
      "grad_norm": 0.37285205721855164,
      "learning_rate": 0.0001630922019135402,
      "loss": 3.3333,
      "step": 197970
    },
    {
      "epoch": 1.022222737500555,
      "grad_norm": 0.38209760189056396,
      "learning_rate": 0.00016309033759144502,
      "loss": 3.416,
      "step": 197980
    },
    {
      "epoch": 1.0222973104451876,
      "grad_norm": 0.3658917248249054,
      "learning_rate": 0.00016308847326934982,
      "loss": 3.3972,
      "step": 197990
    },
    {
      "epoch": 1.0223718833898203,
      "grad_norm": 0.3706904649734497,
      "learning_rate": 0.0001630866089472546,
      "loss": 3.4278,
      "step": 198000
    },
    {
      "epoch": 1.022446456334453,
      "grad_norm": 0.35729894042015076,
      "learning_rate": 0.0001630847446251594,
      "loss": 3.4336,
      "step": 198010
    },
    {
      "epoch": 1.0225210292790858,
      "grad_norm": 0.39458367228507996,
      "learning_rate": 0.0001630828803030642,
      "loss": 3.4017,
      "step": 198020
    },
    {
      "epoch": 1.0225956022237186,
      "grad_norm": 0.3849399983882904,
      "learning_rate": 0.00016308101598096902,
      "loss": 3.3957,
      "step": 198030
    },
    {
      "epoch": 1.0226701751683513,
      "grad_norm": 0.4021221101284027,
      "learning_rate": 0.00016307915165887383,
      "loss": 3.3592,
      "step": 198040
    },
    {
      "epoch": 1.022744748112984,
      "grad_norm": 0.36006027460098267,
      "learning_rate": 0.0001630772873367786,
      "loss": 3.4456,
      "step": 198050
    },
    {
      "epoch": 1.0228193210576169,
      "grad_norm": 0.36568644642829895,
      "learning_rate": 0.00016307542301468341,
      "loss": 3.3494,
      "step": 198060
    },
    {
      "epoch": 1.0228938940022496,
      "grad_norm": 0.3590013086795807,
      "learning_rate": 0.0001630735586925882,
      "loss": 3.374,
      "step": 198070
    },
    {
      "epoch": 1.0229684669468824,
      "grad_norm": 0.3720487356185913,
      "learning_rate": 0.000163071694370493,
      "loss": 3.3794,
      "step": 198080
    },
    {
      "epoch": 1.023043039891515,
      "grad_norm": 0.3977735936641693,
      "learning_rate": 0.00016306983004839783,
      "loss": 3.3951,
      "step": 198090
    },
    {
      "epoch": 1.0231176128361477,
      "grad_norm": 0.37625500559806824,
      "learning_rate": 0.0001630679657263026,
      "loss": 3.3877,
      "step": 198100
    },
    {
      "epoch": 1.0231921857807804,
      "grad_norm": 0.35419872403144836,
      "learning_rate": 0.00016306610140420742,
      "loss": 3.3817,
      "step": 198110
    },
    {
      "epoch": 1.0232667587254132,
      "grad_norm": 0.3710750937461853,
      "learning_rate": 0.0001630642370821122,
      "loss": 3.4159,
      "step": 198120
    },
    {
      "epoch": 1.023341331670046,
      "grad_norm": 0.3630494177341461,
      "learning_rate": 0.000163062372760017,
      "loss": 3.418,
      "step": 198130
    },
    {
      "epoch": 1.0234159046146787,
      "grad_norm": 0.3760637938976288,
      "learning_rate": 0.0001630605084379218,
      "loss": 3.433,
      "step": 198140
    },
    {
      "epoch": 1.0234904775593114,
      "grad_norm": 0.42847374081611633,
      "learning_rate": 0.00016305864411582662,
      "loss": 3.4646,
      "step": 198150
    },
    {
      "epoch": 1.0235650505039442,
      "grad_norm": 0.3633996844291687,
      "learning_rate": 0.00016305677979373142,
      "loss": 3.4427,
      "step": 198160
    },
    {
      "epoch": 1.023639623448577,
      "grad_norm": 0.38380759954452515,
      "learning_rate": 0.0001630549154716362,
      "loss": 3.3792,
      "step": 198170
    },
    {
      "epoch": 1.0237141963932097,
      "grad_norm": 0.37206411361694336,
      "learning_rate": 0.000163053051149541,
      "loss": 3.4408,
      "step": 198180
    },
    {
      "epoch": 1.0237887693378425,
      "grad_norm": 0.3585062623023987,
      "learning_rate": 0.0001630511868274458,
      "loss": 3.4264,
      "step": 198190
    },
    {
      "epoch": 1.023863342282475,
      "grad_norm": 0.38590481877326965,
      "learning_rate": 0.00016304932250535062,
      "loss": 3.3922,
      "step": 198200
    },
    {
      "epoch": 1.0239379152271078,
      "grad_norm": 0.38028445839881897,
      "learning_rate": 0.00016304745818325543,
      "loss": 3.3918,
      "step": 198210
    },
    {
      "epoch": 1.0240124881717405,
      "grad_norm": 0.33875781297683716,
      "learning_rate": 0.0001630455938611602,
      "loss": 3.425,
      "step": 198220
    },
    {
      "epoch": 1.0240870611163733,
      "grad_norm": 0.48771247267723083,
      "learning_rate": 0.000163043729539065,
      "loss": 3.3799,
      "step": 198230
    },
    {
      "epoch": 1.024161634061006,
      "grad_norm": 0.35352712869644165,
      "learning_rate": 0.0001630418652169698,
      "loss": 3.3569,
      "step": 198240
    },
    {
      "epoch": 1.0242362070056388,
      "grad_norm": 0.37701889872550964,
      "learning_rate": 0.0001630400008948746,
      "loss": 3.3514,
      "step": 198250
    },
    {
      "epoch": 1.0243107799502715,
      "grad_norm": 0.37969717383384705,
      "learning_rate": 0.00016303813657277943,
      "loss": 3.4297,
      "step": 198260
    },
    {
      "epoch": 1.0243853528949043,
      "grad_norm": 0.3833572268486023,
      "learning_rate": 0.0001630362722506842,
      "loss": 3.3753,
      "step": 198270
    },
    {
      "epoch": 1.024459925839537,
      "grad_norm": 0.36466073989868164,
      "learning_rate": 0.00016303440792858902,
      "loss": 3.3753,
      "step": 198280
    },
    {
      "epoch": 1.0245344987841698,
      "grad_norm": 0.3619896173477173,
      "learning_rate": 0.0001630325436064938,
      "loss": 3.3783,
      "step": 198290
    },
    {
      "epoch": 1.0246090717288023,
      "grad_norm": 0.3721392750740051,
      "learning_rate": 0.0001630306792843986,
      "loss": 3.388,
      "step": 198300
    },
    {
      "epoch": 1.024683644673435,
      "grad_norm": 0.3816111981868744,
      "learning_rate": 0.0001630288149623034,
      "loss": 3.4133,
      "step": 198310
    },
    {
      "epoch": 1.0247582176180678,
      "grad_norm": 0.367125540971756,
      "learning_rate": 0.00016302695064020822,
      "loss": 3.4388,
      "step": 198320
    },
    {
      "epoch": 1.0248327905627006,
      "grad_norm": 0.36647889018058777,
      "learning_rate": 0.00016302508631811302,
      "loss": 3.3259,
      "step": 198330
    },
    {
      "epoch": 1.0249073635073334,
      "grad_norm": 0.3794022500514984,
      "learning_rate": 0.0001630232219960178,
      "loss": 3.4248,
      "step": 198340
    },
    {
      "epoch": 1.0249819364519661,
      "grad_norm": 0.41566693782806396,
      "learning_rate": 0.0001630213576739226,
      "loss": 3.3756,
      "step": 198350
    },
    {
      "epoch": 1.0250565093965989,
      "grad_norm": 0.3652111291885376,
      "learning_rate": 0.00016301949335182741,
      "loss": 3.4156,
      "step": 198360
    },
    {
      "epoch": 1.0251310823412316,
      "grad_norm": 0.41221290826797485,
      "learning_rate": 0.00016301762902973222,
      "loss": 3.436,
      "step": 198370
    },
    {
      "epoch": 1.0252056552858644,
      "grad_norm": 0.36004477739334106,
      "learning_rate": 0.00016301576470763703,
      "loss": 3.4622,
      "step": 198380
    },
    {
      "epoch": 1.0252802282304971,
      "grad_norm": 0.3676697909832001,
      "learning_rate": 0.0001630139003855418,
      "loss": 3.3961,
      "step": 198390
    },
    {
      "epoch": 1.02535480117513,
      "grad_norm": 0.3918878436088562,
      "learning_rate": 0.0001630120360634466,
      "loss": 3.3704,
      "step": 198400
    },
    {
      "epoch": 1.0254293741197624,
      "grad_norm": 0.39748215675354004,
      "learning_rate": 0.00016301017174135142,
      "loss": 3.4261,
      "step": 198410
    },
    {
      "epoch": 1.0255039470643952,
      "grad_norm": 0.37866491079330444,
      "learning_rate": 0.00016300830741925622,
      "loss": 3.4107,
      "step": 198420
    },
    {
      "epoch": 1.025578520009028,
      "grad_norm": 0.3870278596878052,
      "learning_rate": 0.00016300644309716103,
      "loss": 3.4142,
      "step": 198430
    },
    {
      "epoch": 1.0256530929536607,
      "grad_norm": 0.38349512219429016,
      "learning_rate": 0.0001630045787750658,
      "loss": 3.3663,
      "step": 198440
    },
    {
      "epoch": 1.0257276658982935,
      "grad_norm": 0.3688141107559204,
      "learning_rate": 0.00016300271445297062,
      "loss": 3.4029,
      "step": 198450
    },
    {
      "epoch": 1.0258022388429262,
      "grad_norm": 0.39508119225502014,
      "learning_rate": 0.00016300085013087542,
      "loss": 3.4494,
      "step": 198460
    },
    {
      "epoch": 1.025876811787559,
      "grad_norm": 0.35002943873405457,
      "learning_rate": 0.0001629989858087802,
      "loss": 3.4094,
      "step": 198470
    },
    {
      "epoch": 1.0259513847321917,
      "grad_norm": 0.36832377314567566,
      "learning_rate": 0.00016299712148668504,
      "loss": 3.3629,
      "step": 198480
    },
    {
      "epoch": 1.0260259576768245,
      "grad_norm": 0.34633350372314453,
      "learning_rate": 0.00016299525716458981,
      "loss": 3.4217,
      "step": 198490
    },
    {
      "epoch": 1.0261005306214572,
      "grad_norm": 0.3438226282596588,
      "learning_rate": 0.00016299339284249462,
      "loss": 3.4148,
      "step": 198500
    },
    {
      "epoch": 1.0261751035660898,
      "grad_norm": 0.39227864146232605,
      "learning_rate": 0.00016299152852039943,
      "loss": 3.3601,
      "step": 198510
    },
    {
      "epoch": 1.0262496765107225,
      "grad_norm": 0.38229942321777344,
      "learning_rate": 0.0001629896641983042,
      "loss": 3.4089,
      "step": 198520
    },
    {
      "epoch": 1.0263242494553553,
      "grad_norm": 0.37516579031944275,
      "learning_rate": 0.000162987799876209,
      "loss": 3.412,
      "step": 198530
    },
    {
      "epoch": 1.026398822399988,
      "grad_norm": 0.3803483247756958,
      "learning_rate": 0.00016298593555411382,
      "loss": 3.3986,
      "step": 198540
    },
    {
      "epoch": 1.0264733953446208,
      "grad_norm": 0.3825448751449585,
      "learning_rate": 0.00016298407123201863,
      "loss": 3.3688,
      "step": 198550
    },
    {
      "epoch": 1.0265479682892535,
      "grad_norm": 0.3729232847690582,
      "learning_rate": 0.00016298220690992343,
      "loss": 3.4431,
      "step": 198560
    },
    {
      "epoch": 1.0266225412338863,
      "grad_norm": 0.39184367656707764,
      "learning_rate": 0.0001629803425878282,
      "loss": 3.3405,
      "step": 198570
    },
    {
      "epoch": 1.026697114178519,
      "grad_norm": 0.35920393466949463,
      "learning_rate": 0.00016297847826573302,
      "loss": 3.3778,
      "step": 198580
    },
    {
      "epoch": 1.0267716871231518,
      "grad_norm": 0.3836681544780731,
      "learning_rate": 0.00016297661394363782,
      "loss": 3.4064,
      "step": 198590
    },
    {
      "epoch": 1.0268462600677846,
      "grad_norm": 0.37298911809921265,
      "learning_rate": 0.00016297474962154263,
      "loss": 3.442,
      "step": 198600
    },
    {
      "epoch": 1.0269208330124173,
      "grad_norm": 0.3667917847633362,
      "learning_rate": 0.00016297288529944744,
      "loss": 3.3467,
      "step": 198610
    },
    {
      "epoch": 1.0269954059570499,
      "grad_norm": 0.4149191975593567,
      "learning_rate": 0.00016297102097735222,
      "loss": 3.3333,
      "step": 198620
    },
    {
      "epoch": 1.0270699789016826,
      "grad_norm": 0.4001106321811676,
      "learning_rate": 0.00016296915665525702,
      "loss": 3.3683,
      "step": 198630
    },
    {
      "epoch": 1.0271445518463154,
      "grad_norm": 0.390739381313324,
      "learning_rate": 0.0001629672923331618,
      "loss": 3.3898,
      "step": 198640
    },
    {
      "epoch": 1.0272191247909481,
      "grad_norm": 0.3733559846878052,
      "learning_rate": 0.00016296542801106663,
      "loss": 3.3956,
      "step": 198650
    },
    {
      "epoch": 1.0272936977355809,
      "grad_norm": 0.38165146112442017,
      "learning_rate": 0.00016296356368897144,
      "loss": 3.4119,
      "step": 198660
    },
    {
      "epoch": 1.0273682706802136,
      "grad_norm": 0.37501057982444763,
      "learning_rate": 0.00016296169936687622,
      "loss": 3.3847,
      "step": 198670
    },
    {
      "epoch": 1.0274428436248464,
      "grad_norm": 0.3743495047092438,
      "learning_rate": 0.00016295983504478103,
      "loss": 3.4096,
      "step": 198680
    },
    {
      "epoch": 1.0275174165694791,
      "grad_norm": 0.37098273634910583,
      "learning_rate": 0.0001629579707226858,
      "loss": 3.4304,
      "step": 198690
    },
    {
      "epoch": 1.027591989514112,
      "grad_norm": 0.3956576883792877,
      "learning_rate": 0.0001629561064005906,
      "loss": 3.4302,
      "step": 198700
    },
    {
      "epoch": 1.0276665624587447,
      "grad_norm": 0.3759501576423645,
      "learning_rate": 0.00016295424207849545,
      "loss": 3.4291,
      "step": 198710
    },
    {
      "epoch": 1.0277411354033772,
      "grad_norm": 0.3858203887939453,
      "learning_rate": 0.00016295237775640022,
      "loss": 3.3851,
      "step": 198720
    },
    {
      "epoch": 1.02781570834801,
      "grad_norm": 0.37774378061294556,
      "learning_rate": 0.00016295051343430503,
      "loss": 3.3655,
      "step": 198730
    },
    {
      "epoch": 1.0278902812926427,
      "grad_norm": 0.3721770942211151,
      "learning_rate": 0.0001629486491122098,
      "loss": 3.3303,
      "step": 198740
    },
    {
      "epoch": 1.0279648542372755,
      "grad_norm": 0.3771817982196808,
      "learning_rate": 0.00016294678479011462,
      "loss": 3.3954,
      "step": 198750
    },
    {
      "epoch": 1.0280394271819082,
      "grad_norm": 0.37736615538597107,
      "learning_rate": 0.00016294492046801942,
      "loss": 3.376,
      "step": 198760
    },
    {
      "epoch": 1.028114000126541,
      "grad_norm": 0.3793267607688904,
      "learning_rate": 0.00016294305614592423,
      "loss": 3.4018,
      "step": 198770
    },
    {
      "epoch": 1.0281885730711737,
      "grad_norm": 0.372478723526001,
      "learning_rate": 0.00016294119182382904,
      "loss": 3.3846,
      "step": 198780
    },
    {
      "epoch": 1.0282631460158065,
      "grad_norm": 0.37527433037757874,
      "learning_rate": 0.00016293932750173381,
      "loss": 3.4101,
      "step": 198790
    },
    {
      "epoch": 1.0283377189604392,
      "grad_norm": 0.3633059561252594,
      "learning_rate": 0.00016293746317963862,
      "loss": 3.4059,
      "step": 198800
    },
    {
      "epoch": 1.028412291905072,
      "grad_norm": 0.3803752064704895,
      "learning_rate": 0.00016293559885754343,
      "loss": 3.4316,
      "step": 198810
    },
    {
      "epoch": 1.0284868648497045,
      "grad_norm": 0.37184858322143555,
      "learning_rate": 0.00016293373453544823,
      "loss": 3.39,
      "step": 198820
    },
    {
      "epoch": 1.0285614377943373,
      "grad_norm": 0.3993067443370819,
      "learning_rate": 0.00016293187021335304,
      "loss": 3.4377,
      "step": 198830
    },
    {
      "epoch": 1.02863601073897,
      "grad_norm": 0.38832202553749084,
      "learning_rate": 0.00016293000589125782,
      "loss": 3.3939,
      "step": 198840
    },
    {
      "epoch": 1.0287105836836028,
      "grad_norm": 0.38440439105033875,
      "learning_rate": 0.00016292814156916263,
      "loss": 3.3674,
      "step": 198850
    },
    {
      "epoch": 1.0287851566282356,
      "grad_norm": 0.4055153429508209,
      "learning_rate": 0.00016292627724706743,
      "loss": 3.375,
      "step": 198860
    },
    {
      "epoch": 1.0288597295728683,
      "grad_norm": 0.373364120721817,
      "learning_rate": 0.00016292441292497224,
      "loss": 3.4077,
      "step": 198870
    },
    {
      "epoch": 1.028934302517501,
      "grad_norm": 0.3807014226913452,
      "learning_rate": 0.00016292254860287704,
      "loss": 3.4553,
      "step": 198880
    },
    {
      "epoch": 1.0290088754621338,
      "grad_norm": 0.38782891631126404,
      "learning_rate": 0.00016292068428078182,
      "loss": 3.3731,
      "step": 198890
    },
    {
      "epoch": 1.0290834484067666,
      "grad_norm": 0.36907634139060974,
      "learning_rate": 0.00016291881995868663,
      "loss": 3.3932,
      "step": 198900
    },
    {
      "epoch": 1.0291580213513993,
      "grad_norm": 0.3837507367134094,
      "learning_rate": 0.00016291695563659144,
      "loss": 3.397,
      "step": 198910
    },
    {
      "epoch": 1.029232594296032,
      "grad_norm": 0.3679750859737396,
      "learning_rate": 0.00016291509131449622,
      "loss": 3.4105,
      "step": 198920
    },
    {
      "epoch": 1.0293071672406646,
      "grad_norm": 0.3769056499004364,
      "learning_rate": 0.00016291322699240105,
      "loss": 3.3979,
      "step": 198930
    },
    {
      "epoch": 1.0293817401852974,
      "grad_norm": 0.39800646901130676,
      "learning_rate": 0.00016291136267030583,
      "loss": 3.4063,
      "step": 198940
    },
    {
      "epoch": 1.0294563131299301,
      "grad_norm": 0.4091331362724304,
      "learning_rate": 0.00016290949834821063,
      "loss": 3.3608,
      "step": 198950
    },
    {
      "epoch": 1.029530886074563,
      "grad_norm": 0.3729894459247589,
      "learning_rate": 0.00016290763402611544,
      "loss": 3.4289,
      "step": 198960
    },
    {
      "epoch": 1.0296054590191956,
      "grad_norm": 0.38868898153305054,
      "learning_rate": 0.00016290576970402022,
      "loss": 3.3777,
      "step": 198970
    },
    {
      "epoch": 1.0296800319638284,
      "grad_norm": 0.3874422013759613,
      "learning_rate": 0.00016290390538192503,
      "loss": 3.4271,
      "step": 198980
    },
    {
      "epoch": 1.0297546049084612,
      "grad_norm": 0.3610776364803314,
      "learning_rate": 0.00016290204105982983,
      "loss": 3.4317,
      "step": 198990
    },
    {
      "epoch": 1.029829177853094,
      "grad_norm": 0.3664528429508209,
      "learning_rate": 0.00016290017673773464,
      "loss": 3.3765,
      "step": 199000
    },
    {
      "epoch": 1.0299037507977267,
      "grad_norm": 0.37757864594459534,
      "learning_rate": 0.00016289831241563944,
      "loss": 3.3482,
      "step": 199010
    },
    {
      "epoch": 1.0299783237423594,
      "grad_norm": 0.36171698570251465,
      "learning_rate": 0.00016289644809354422,
      "loss": 3.3556,
      "step": 199020
    },
    {
      "epoch": 1.030052896686992,
      "grad_norm": 0.3567694425582886,
      "learning_rate": 0.00016289458377144903,
      "loss": 3.3742,
      "step": 199030
    },
    {
      "epoch": 1.0301274696316247,
      "grad_norm": 0.3718249797821045,
      "learning_rate": 0.00016289271944935384,
      "loss": 3.386,
      "step": 199040
    },
    {
      "epoch": 1.0302020425762575,
      "grad_norm": 0.36859890818595886,
      "learning_rate": 0.00016289085512725864,
      "loss": 3.3865,
      "step": 199050
    },
    {
      "epoch": 1.0302766155208902,
      "grad_norm": 0.3711306154727936,
      "learning_rate": 0.00016288899080516345,
      "loss": 3.3513,
      "step": 199060
    },
    {
      "epoch": 1.030351188465523,
      "grad_norm": 0.39445289969444275,
      "learning_rate": 0.00016288712648306823,
      "loss": 3.3923,
      "step": 199070
    },
    {
      "epoch": 1.0304257614101557,
      "grad_norm": 0.38835880160331726,
      "learning_rate": 0.00016288526216097303,
      "loss": 3.3653,
      "step": 199080
    },
    {
      "epoch": 1.0305003343547885,
      "grad_norm": 0.39864423871040344,
      "learning_rate": 0.00016288339783887781,
      "loss": 3.4554,
      "step": 199090
    },
    {
      "epoch": 1.0305749072994213,
      "grad_norm": 0.3929126560688019,
      "learning_rate": 0.00016288153351678265,
      "loss": 3.3921,
      "step": 199100
    },
    {
      "epoch": 1.030649480244054,
      "grad_norm": 0.3759695887565613,
      "learning_rate": 0.00016287966919468745,
      "loss": 3.3671,
      "step": 199110
    },
    {
      "epoch": 1.0307240531886868,
      "grad_norm": 0.39658740162849426,
      "learning_rate": 0.00016287780487259223,
      "loss": 3.4163,
      "step": 199120
    },
    {
      "epoch": 1.0307986261333195,
      "grad_norm": 0.35121291875839233,
      "learning_rate": 0.00016287594055049704,
      "loss": 3.3654,
      "step": 199130
    },
    {
      "epoch": 1.030873199077952,
      "grad_norm": 0.33992135524749756,
      "learning_rate": 0.00016287407622840182,
      "loss": 3.4139,
      "step": 199140
    },
    {
      "epoch": 1.0309477720225848,
      "grad_norm": 0.3541582226753235,
      "learning_rate": 0.00016287221190630662,
      "loss": 3.4484,
      "step": 199150
    },
    {
      "epoch": 1.0310223449672176,
      "grad_norm": 0.38660719990730286,
      "learning_rate": 0.00016287034758421146,
      "loss": 3.3827,
      "step": 199160
    },
    {
      "epoch": 1.0310969179118503,
      "grad_norm": 0.40047216415405273,
      "learning_rate": 0.00016286848326211624,
      "loss": 3.3439,
      "step": 199170
    },
    {
      "epoch": 1.031171490856483,
      "grad_norm": 0.36390817165374756,
      "learning_rate": 0.00016286661894002104,
      "loss": 3.3749,
      "step": 199180
    },
    {
      "epoch": 1.0312460638011158,
      "grad_norm": 0.36721184849739075,
      "learning_rate": 0.00016286475461792582,
      "loss": 3.4109,
      "step": 199190
    },
    {
      "epoch": 1.0313206367457486,
      "grad_norm": 0.38666731119155884,
      "learning_rate": 0.00016286289029583063,
      "loss": 3.3594,
      "step": 199200
    },
    {
      "epoch": 1.0313952096903813,
      "grad_norm": 0.38504815101623535,
      "learning_rate": 0.00016286102597373544,
      "loss": 3.3683,
      "step": 199210
    },
    {
      "epoch": 1.031469782635014,
      "grad_norm": 0.3502364754676819,
      "learning_rate": 0.00016285916165164024,
      "loss": 3.4307,
      "step": 199220
    },
    {
      "epoch": 1.0315443555796469,
      "grad_norm": 0.40860065817832947,
      "learning_rate": 0.00016285729732954505,
      "loss": 3.3477,
      "step": 199230
    },
    {
      "epoch": 1.0316189285242794,
      "grad_norm": 0.3940522372722626,
      "learning_rate": 0.00016285543300744983,
      "loss": 3.4229,
      "step": 199240
    },
    {
      "epoch": 1.0316935014689121,
      "grad_norm": 0.3855624794960022,
      "learning_rate": 0.00016285356868535463,
      "loss": 3.344,
      "step": 199250
    },
    {
      "epoch": 1.031768074413545,
      "grad_norm": 0.39646533131599426,
      "learning_rate": 0.00016285170436325944,
      "loss": 3.4414,
      "step": 199260
    },
    {
      "epoch": 1.0318426473581777,
      "grad_norm": 0.3616635501384735,
      "learning_rate": 0.00016284984004116425,
      "loss": 3.3777,
      "step": 199270
    },
    {
      "epoch": 1.0319172203028104,
      "grad_norm": 0.3786763846874237,
      "learning_rate": 0.00016284797571906905,
      "loss": 3.431,
      "step": 199280
    },
    {
      "epoch": 1.0319917932474432,
      "grad_norm": 0.3796563446521759,
      "learning_rate": 0.00016284611139697383,
      "loss": 3.3345,
      "step": 199290
    },
    {
      "epoch": 1.032066366192076,
      "grad_norm": 0.37480199337005615,
      "learning_rate": 0.00016284424707487864,
      "loss": 3.4058,
      "step": 199300
    },
    {
      "epoch": 1.0321409391367087,
      "grad_norm": 0.3694894313812256,
      "learning_rate": 0.00016284238275278344,
      "loss": 3.371,
      "step": 199310
    },
    {
      "epoch": 1.0322155120813414,
      "grad_norm": 0.3705400824546814,
      "learning_rate": 0.00016284051843068825,
      "loss": 3.3855,
      "step": 199320
    },
    {
      "epoch": 1.0322900850259742,
      "grad_norm": 0.370577871799469,
      "learning_rate": 0.00016283865410859306,
      "loss": 3.395,
      "step": 199330
    },
    {
      "epoch": 1.032364657970607,
      "grad_norm": 0.35779252648353577,
      "learning_rate": 0.00016283678978649784,
      "loss": 3.3989,
      "step": 199340
    },
    {
      "epoch": 1.0324392309152395,
      "grad_norm": 0.356458842754364,
      "learning_rate": 0.00016283492546440264,
      "loss": 3.4277,
      "step": 199350
    },
    {
      "epoch": 1.0325138038598722,
      "grad_norm": 0.38130250573158264,
      "learning_rate": 0.00016283306114230745,
      "loss": 3.3968,
      "step": 199360
    },
    {
      "epoch": 1.032588376804505,
      "grad_norm": 0.3739590048789978,
      "learning_rate": 0.00016283119682021223,
      "loss": 3.3754,
      "step": 199370
    },
    {
      "epoch": 1.0326629497491377,
      "grad_norm": 0.38562482595443726,
      "learning_rate": 0.00016282933249811706,
      "loss": 3.4337,
      "step": 199380
    },
    {
      "epoch": 1.0327375226937705,
      "grad_norm": 0.4045041501522064,
      "learning_rate": 0.00016282746817602184,
      "loss": 3.3146,
      "step": 199390
    },
    {
      "epoch": 1.0328120956384033,
      "grad_norm": 0.3888855278491974,
      "learning_rate": 0.00016282560385392665,
      "loss": 3.3792,
      "step": 199400
    },
    {
      "epoch": 1.032886668583036,
      "grad_norm": 0.35991403460502625,
      "learning_rate": 0.00016282373953183145,
      "loss": 3.411,
      "step": 199410
    },
    {
      "epoch": 1.0329612415276688,
      "grad_norm": 0.38491466641426086,
      "learning_rate": 0.00016282187520973623,
      "loss": 3.4125,
      "step": 199420
    },
    {
      "epoch": 1.0330358144723015,
      "grad_norm": 0.354437917470932,
      "learning_rate": 0.00016282001088764104,
      "loss": 3.4087,
      "step": 199430
    },
    {
      "epoch": 1.0331103874169343,
      "grad_norm": 0.36555761098861694,
      "learning_rate": 0.00016281814656554584,
      "loss": 3.3155,
      "step": 199440
    },
    {
      "epoch": 1.0331849603615668,
      "grad_norm": 0.39032939076423645,
      "learning_rate": 0.00016281628224345065,
      "loss": 3.2687,
      "step": 199450
    },
    {
      "epoch": 1.0332595333061996,
      "grad_norm": 0.36410459876060486,
      "learning_rate": 0.00016281441792135546,
      "loss": 3.4099,
      "step": 199460
    },
    {
      "epoch": 1.0333341062508323,
      "grad_norm": 0.38642996549606323,
      "learning_rate": 0.00016281255359926024,
      "loss": 3.4261,
      "step": 199470
    },
    {
      "epoch": 1.033408679195465,
      "grad_norm": 0.35572341084480286,
      "learning_rate": 0.00016281068927716504,
      "loss": 3.4036,
      "step": 199480
    },
    {
      "epoch": 1.0334832521400978,
      "grad_norm": 0.4032270908355713,
      "learning_rate": 0.00016280882495506985,
      "loss": 3.3842,
      "step": 199490
    },
    {
      "epoch": 1.0335578250847306,
      "grad_norm": 0.38470587134361267,
      "learning_rate": 0.00016280696063297466,
      "loss": 3.4105,
      "step": 199500
    },
    {
      "epoch": 1.0336323980293634,
      "grad_norm": 0.4018155336380005,
      "learning_rate": 0.00016280509631087946,
      "loss": 3.3699,
      "step": 199510
    },
    {
      "epoch": 1.033706970973996,
      "grad_norm": 0.37420523166656494,
      "learning_rate": 0.00016280323198878424,
      "loss": 3.3484,
      "step": 199520
    },
    {
      "epoch": 1.0337815439186289,
      "grad_norm": 0.36830052733421326,
      "learning_rate": 0.00016280136766668905,
      "loss": 3.3632,
      "step": 199530
    },
    {
      "epoch": 1.0338561168632616,
      "grad_norm": 0.4072028696537018,
      "learning_rate": 0.00016279950334459383,
      "loss": 3.3728,
      "step": 199540
    },
    {
      "epoch": 1.0339306898078942,
      "grad_norm": 0.36287838220596313,
      "learning_rate": 0.00016279763902249866,
      "loss": 3.4064,
      "step": 199550
    },
    {
      "epoch": 1.034005262752527,
      "grad_norm": 0.3760831952095032,
      "learning_rate": 0.00016279577470040344,
      "loss": 3.3741,
      "step": 199560
    },
    {
      "epoch": 1.0340798356971597,
      "grad_norm": 0.3906535804271698,
      "learning_rate": 0.00016279391037830825,
      "loss": 3.3838,
      "step": 199570
    },
    {
      "epoch": 1.0341544086417924,
      "grad_norm": 0.3703062832355499,
      "learning_rate": 0.00016279204605621305,
      "loss": 3.44,
      "step": 199580
    },
    {
      "epoch": 1.0342289815864252,
      "grad_norm": 0.35480985045433044,
      "learning_rate": 0.00016279018173411783,
      "loss": 3.4582,
      "step": 199590
    },
    {
      "epoch": 1.034303554531058,
      "grad_norm": 0.38280534744262695,
      "learning_rate": 0.00016278831741202264,
      "loss": 3.3293,
      "step": 199600
    },
    {
      "epoch": 1.0343781274756907,
      "grad_norm": 0.3637535870075226,
      "learning_rate": 0.00016278645308992744,
      "loss": 3.4361,
      "step": 199610
    },
    {
      "epoch": 1.0344527004203234,
      "grad_norm": 0.39957648515701294,
      "learning_rate": 0.00016278458876783225,
      "loss": 3.3516,
      "step": 199620
    },
    {
      "epoch": 1.0345272733649562,
      "grad_norm": 0.35985979437828064,
      "learning_rate": 0.00016278272444573706,
      "loss": 3.3681,
      "step": 199630
    },
    {
      "epoch": 1.034601846309589,
      "grad_norm": 0.37442129850387573,
      "learning_rate": 0.00016278086012364184,
      "loss": 3.3731,
      "step": 199640
    },
    {
      "epoch": 1.0346764192542217,
      "grad_norm": 0.35736119747161865,
      "learning_rate": 0.00016277899580154664,
      "loss": 3.4285,
      "step": 199650
    },
    {
      "epoch": 1.0347509921988542,
      "grad_norm": 0.38263997435569763,
      "learning_rate": 0.00016277713147945145,
      "loss": 3.4035,
      "step": 199660
    },
    {
      "epoch": 1.034825565143487,
      "grad_norm": 0.41309505701065063,
      "learning_rate": 0.00016277526715735625,
      "loss": 3.3646,
      "step": 199670
    },
    {
      "epoch": 1.0349001380881198,
      "grad_norm": 0.37347540259361267,
      "learning_rate": 0.00016277340283526106,
      "loss": 3.4325,
      "step": 199680
    },
    {
      "epoch": 1.0349747110327525,
      "grad_norm": 0.3631594777107239,
      "learning_rate": 0.00016277153851316584,
      "loss": 3.4435,
      "step": 199690
    },
    {
      "epoch": 1.0350492839773853,
      "grad_norm": 0.38668984174728394,
      "learning_rate": 0.00016276967419107065,
      "loss": 3.3798,
      "step": 199700
    },
    {
      "epoch": 1.035123856922018,
      "grad_norm": 0.4156683683395386,
      "learning_rate": 0.00016276780986897545,
      "loss": 3.4396,
      "step": 199710
    },
    {
      "epoch": 1.0351984298666508,
      "grad_norm": 0.38735395669937134,
      "learning_rate": 0.00016276594554688026,
      "loss": 3.383,
      "step": 199720
    },
    {
      "epoch": 1.0352730028112835,
      "grad_norm": 0.379219114780426,
      "learning_rate": 0.00016276408122478507,
      "loss": 3.4385,
      "step": 199730
    },
    {
      "epoch": 1.0353475757559163,
      "grad_norm": 0.37683191895484924,
      "learning_rate": 0.00016276221690268984,
      "loss": 3.3591,
      "step": 199740
    },
    {
      "epoch": 1.035422148700549,
      "grad_norm": 0.3604625463485718,
      "learning_rate": 0.00016276035258059465,
      "loss": 3.3506,
      "step": 199750
    },
    {
      "epoch": 1.0354967216451816,
      "grad_norm": 0.3950486183166504,
      "learning_rate": 0.00016275848825849943,
      "loss": 3.3845,
      "step": 199760
    },
    {
      "epoch": 1.0355712945898143,
      "grad_norm": 0.39250805974006653,
      "learning_rate": 0.00016275662393640426,
      "loss": 3.4023,
      "step": 199770
    },
    {
      "epoch": 1.035645867534447,
      "grad_norm": 0.38037732243537903,
      "learning_rate": 0.00016275475961430907,
      "loss": 3.3273,
      "step": 199780
    },
    {
      "epoch": 1.0357204404790799,
      "grad_norm": 0.4264526963233948,
      "learning_rate": 0.00016275289529221385,
      "loss": 3.3614,
      "step": 199790
    },
    {
      "epoch": 1.0357950134237126,
      "grad_norm": 0.3974258005619049,
      "learning_rate": 0.00016275103097011866,
      "loss": 3.4608,
      "step": 199800
    },
    {
      "epoch": 1.0358695863683454,
      "grad_norm": 0.37400200963020325,
      "learning_rate": 0.00016274916664802343,
      "loss": 3.4087,
      "step": 199810
    },
    {
      "epoch": 1.0359441593129781,
      "grad_norm": 0.3669130802154541,
      "learning_rate": 0.00016274730232592824,
      "loss": 3.4078,
      "step": 199820
    },
    {
      "epoch": 1.0360187322576109,
      "grad_norm": 0.3558301329612732,
      "learning_rate": 0.00016274543800383307,
      "loss": 3.3856,
      "step": 199830
    },
    {
      "epoch": 1.0360933052022436,
      "grad_norm": 0.3805789053440094,
      "learning_rate": 0.00016274357368173785,
      "loss": 3.3859,
      "step": 199840
    },
    {
      "epoch": 1.0361678781468764,
      "grad_norm": 0.39209187030792236,
      "learning_rate": 0.00016274170935964266,
      "loss": 3.3919,
      "step": 199850
    },
    {
      "epoch": 1.0362424510915091,
      "grad_norm": 0.40389350056648254,
      "learning_rate": 0.00016273984503754744,
      "loss": 3.3804,
      "step": 199860
    },
    {
      "epoch": 1.0363170240361417,
      "grad_norm": 0.393803209066391,
      "learning_rate": 0.00016273798071545225,
      "loss": 3.3749,
      "step": 199870
    },
    {
      "epoch": 1.0363915969807744,
      "grad_norm": 0.4227028489112854,
      "learning_rate": 0.00016273611639335705,
      "loss": 3.4198,
      "step": 199880
    },
    {
      "epoch": 1.0364661699254072,
      "grad_norm": 0.4150875508785248,
      "learning_rate": 0.00016273425207126186,
      "loss": 3.379,
      "step": 199890
    },
    {
      "epoch": 1.03654074287004,
      "grad_norm": 0.38430842757225037,
      "learning_rate": 0.00016273238774916666,
      "loss": 3.3938,
      "step": 199900
    },
    {
      "epoch": 1.0366153158146727,
      "grad_norm": 0.3917923867702484,
      "learning_rate": 0.00016273052342707144,
      "loss": 3.3879,
      "step": 199910
    },
    {
      "epoch": 1.0366898887593055,
      "grad_norm": 0.3875972628593445,
      "learning_rate": 0.00016272865910497625,
      "loss": 3.4169,
      "step": 199920
    },
    {
      "epoch": 1.0367644617039382,
      "grad_norm": 0.3727186918258667,
      "learning_rate": 0.00016272679478288106,
      "loss": 3.3665,
      "step": 199930
    },
    {
      "epoch": 1.036839034648571,
      "grad_norm": 0.359777569770813,
      "learning_rate": 0.00016272493046078586,
      "loss": 3.4104,
      "step": 199940
    },
    {
      "epoch": 1.0369136075932037,
      "grad_norm": 0.3773210048675537,
      "learning_rate": 0.00016272306613869067,
      "loss": 3.3473,
      "step": 199950
    },
    {
      "epoch": 1.0369881805378365,
      "grad_norm": 0.36369773745536804,
      "learning_rate": 0.00016272120181659545,
      "loss": 3.4218,
      "step": 199960
    },
    {
      "epoch": 1.037062753482469,
      "grad_norm": 0.3715546131134033,
      "learning_rate": 0.00016271933749450025,
      "loss": 3.292,
      "step": 199970
    },
    {
      "epoch": 1.0371373264271018,
      "grad_norm": 0.37573501467704773,
      "learning_rate": 0.00016271747317240506,
      "loss": 3.3888,
      "step": 199980
    },
    {
      "epoch": 1.0372118993717345,
      "grad_norm": 0.38560470938682556,
      "learning_rate": 0.00016271560885030984,
      "loss": 3.3724,
      "step": 199990
    },
    {
      "epoch": 1.0372864723163673,
      "grad_norm": 0.37599581480026245,
      "learning_rate": 0.00016271374452821467,
      "loss": 3.3874,
      "step": 200000
    },
    {
      "epoch": 1.037361045261,
      "grad_norm": 0.3648292124271393,
      "learning_rate": 0.00016271188020611945,
      "loss": 3.3787,
      "step": 200010
    },
    {
      "epoch": 1.0374356182056328,
      "grad_norm": 0.3601621091365814,
      "learning_rate": 0.00016271001588402426,
      "loss": 3.3529,
      "step": 200020
    },
    {
      "epoch": 1.0375101911502655,
      "grad_norm": 0.39431333541870117,
      "learning_rate": 0.00016270815156192906,
      "loss": 3.3595,
      "step": 200030
    },
    {
      "epoch": 1.0375847640948983,
      "grad_norm": 0.3646322190761566,
      "learning_rate": 0.00016270628723983384,
      "loss": 3.4022,
      "step": 200040
    },
    {
      "epoch": 1.037659337039531,
      "grad_norm": 0.3877968192100525,
      "learning_rate": 0.00016270442291773865,
      "loss": 3.3579,
      "step": 200050
    },
    {
      "epoch": 1.0377339099841638,
      "grad_norm": 0.37683841586112976,
      "learning_rate": 0.00016270255859564346,
      "loss": 3.4393,
      "step": 200060
    },
    {
      "epoch": 1.0378084829287966,
      "grad_norm": 0.3732657730579376,
      "learning_rate": 0.00016270069427354826,
      "loss": 3.3754,
      "step": 200070
    },
    {
      "epoch": 1.037883055873429,
      "grad_norm": 0.3830009996891022,
      "learning_rate": 0.00016269882995145307,
      "loss": 3.3812,
      "step": 200080
    },
    {
      "epoch": 1.0379576288180619,
      "grad_norm": 0.39921337366104126,
      "learning_rate": 0.00016269696562935785,
      "loss": 3.3618,
      "step": 200090
    },
    {
      "epoch": 1.0380322017626946,
      "grad_norm": 0.36591169238090515,
      "learning_rate": 0.00016269510130726265,
      "loss": 3.3743,
      "step": 200100
    },
    {
      "epoch": 1.0381067747073274,
      "grad_norm": 0.39510613679885864,
      "learning_rate": 0.00016269323698516746,
      "loss": 3.3443,
      "step": 200110
    },
    {
      "epoch": 1.0381813476519601,
      "grad_norm": 0.3859066665172577,
      "learning_rate": 0.00016269137266307227,
      "loss": 3.4199,
      "step": 200120
    },
    {
      "epoch": 1.0382559205965929,
      "grad_norm": 0.3981739282608032,
      "learning_rate": 0.00016268950834097707,
      "loss": 3.4054,
      "step": 200130
    },
    {
      "epoch": 1.0383304935412256,
      "grad_norm": 0.3734769821166992,
      "learning_rate": 0.00016268764401888185,
      "loss": 3.4017,
      "step": 200140
    },
    {
      "epoch": 1.0384050664858584,
      "grad_norm": 0.3655478060245514,
      "learning_rate": 0.00016268577969678666,
      "loss": 3.3341,
      "step": 200150
    },
    {
      "epoch": 1.0384796394304912,
      "grad_norm": 0.37752023339271545,
      "learning_rate": 0.00016268391537469147,
      "loss": 3.3441,
      "step": 200160
    },
    {
      "epoch": 1.038554212375124,
      "grad_norm": 0.3766242265701294,
      "learning_rate": 0.00016268205105259627,
      "loss": 3.3423,
      "step": 200170
    },
    {
      "epoch": 1.0386287853197564,
      "grad_norm": 0.40431755781173706,
      "learning_rate": 0.00016268018673050108,
      "loss": 3.3422,
      "step": 200180
    },
    {
      "epoch": 1.0387033582643892,
      "grad_norm": 0.4116901457309723,
      "learning_rate": 0.00016267832240840586,
      "loss": 3.4132,
      "step": 200190
    },
    {
      "epoch": 1.038777931209022,
      "grad_norm": 0.3868252635002136,
      "learning_rate": 0.00016267645808631066,
      "loss": 3.3357,
      "step": 200200
    },
    {
      "epoch": 1.0388525041536547,
      "grad_norm": 0.3824702203273773,
      "learning_rate": 0.00016267459376421544,
      "loss": 3.3505,
      "step": 200210
    },
    {
      "epoch": 1.0389270770982875,
      "grad_norm": 0.39580869674682617,
      "learning_rate": 0.00016267272944212028,
      "loss": 3.3622,
      "step": 200220
    },
    {
      "epoch": 1.0390016500429202,
      "grad_norm": 0.37046387791633606,
      "learning_rate": 0.00016267086512002508,
      "loss": 3.3846,
      "step": 200230
    },
    {
      "epoch": 1.039076222987553,
      "grad_norm": 0.35592225193977356,
      "learning_rate": 0.00016266900079792986,
      "loss": 3.3504,
      "step": 200240
    },
    {
      "epoch": 1.0391507959321857,
      "grad_norm": 0.38588011264801025,
      "learning_rate": 0.00016266713647583467,
      "loss": 3.4217,
      "step": 200250
    },
    {
      "epoch": 1.0392253688768185,
      "grad_norm": 0.36741161346435547,
      "learning_rate": 0.00016266527215373945,
      "loss": 3.3335,
      "step": 200260
    },
    {
      "epoch": 1.0392999418214512,
      "grad_norm": 0.36362192034721375,
      "learning_rate": 0.00016266340783164425,
      "loss": 3.458,
      "step": 200270
    },
    {
      "epoch": 1.039374514766084,
      "grad_norm": 0.37436676025390625,
      "learning_rate": 0.0001626615435095491,
      "loss": 3.3417,
      "step": 200280
    },
    {
      "epoch": 1.0394490877107165,
      "grad_norm": 0.3785811960697174,
      "learning_rate": 0.00016265967918745387,
      "loss": 3.326,
      "step": 200290
    },
    {
      "epoch": 1.0395236606553493,
      "grad_norm": 0.37165912985801697,
      "learning_rate": 0.00016265781486535867,
      "loss": 3.3424,
      "step": 200300
    },
    {
      "epoch": 1.039598233599982,
      "grad_norm": 0.3780234456062317,
      "learning_rate": 0.00016265595054326345,
      "loss": 3.4435,
      "step": 200310
    },
    {
      "epoch": 1.0396728065446148,
      "grad_norm": 0.3803316056728363,
      "learning_rate": 0.00016265408622116826,
      "loss": 3.4133,
      "step": 200320
    },
    {
      "epoch": 1.0397473794892476,
      "grad_norm": 0.3839457631111145,
      "learning_rate": 0.00016265222189907306,
      "loss": 3.4396,
      "step": 200330
    },
    {
      "epoch": 1.0398219524338803,
      "grad_norm": 0.36596810817718506,
      "learning_rate": 0.00016265035757697787,
      "loss": 3.3912,
      "step": 200340
    },
    {
      "epoch": 1.039896525378513,
      "grad_norm": 0.3962719440460205,
      "learning_rate": 0.00016264849325488268,
      "loss": 3.3507,
      "step": 200350
    },
    {
      "epoch": 1.0399710983231458,
      "grad_norm": 0.3662550449371338,
      "learning_rate": 0.00016264662893278746,
      "loss": 3.316,
      "step": 200360
    },
    {
      "epoch": 1.0400456712677786,
      "grad_norm": 0.3566821813583374,
      "learning_rate": 0.00016264476461069226,
      "loss": 3.4271,
      "step": 200370
    },
    {
      "epoch": 1.0401202442124113,
      "grad_norm": 0.35422319173812866,
      "learning_rate": 0.00016264290028859707,
      "loss": 3.4545,
      "step": 200380
    },
    {
      "epoch": 1.0401948171570439,
      "grad_norm": 0.3665044903755188,
      "learning_rate": 0.00016264103596650188,
      "loss": 3.3294,
      "step": 200390
    },
    {
      "epoch": 1.0402693901016766,
      "grad_norm": 0.3660798966884613,
      "learning_rate": 0.00016263917164440668,
      "loss": 3.3734,
      "step": 200400
    },
    {
      "epoch": 1.0403439630463094,
      "grad_norm": 0.3935088813304901,
      "learning_rate": 0.00016263730732231146,
      "loss": 3.4009,
      "step": 200410
    },
    {
      "epoch": 1.0404185359909421,
      "grad_norm": 0.37004554271698,
      "learning_rate": 0.00016263544300021627,
      "loss": 3.3883,
      "step": 200420
    },
    {
      "epoch": 1.040493108935575,
      "grad_norm": 0.4098469913005829,
      "learning_rate": 0.00016263357867812107,
      "loss": 3.3842,
      "step": 200430
    },
    {
      "epoch": 1.0405676818802077,
      "grad_norm": 0.3498598337173462,
      "learning_rate": 0.00016263171435602588,
      "loss": 3.4011,
      "step": 200440
    },
    {
      "epoch": 1.0406422548248404,
      "grad_norm": 0.38473862409591675,
      "learning_rate": 0.00016262985003393069,
      "loss": 3.4251,
      "step": 200450
    },
    {
      "epoch": 1.0407168277694732,
      "grad_norm": 0.36920419335365295,
      "learning_rate": 0.00016262798571183546,
      "loss": 3.3941,
      "step": 200460
    },
    {
      "epoch": 1.040791400714106,
      "grad_norm": 0.344167560338974,
      "learning_rate": 0.00016262612138974027,
      "loss": 3.4215,
      "step": 200470
    },
    {
      "epoch": 1.0408659736587387,
      "grad_norm": 0.34542593359947205,
      "learning_rate": 0.00016262425706764508,
      "loss": 3.4058,
      "step": 200480
    },
    {
      "epoch": 1.0409405466033714,
      "grad_norm": 0.37853002548217773,
      "learning_rate": 0.00016262239274554986,
      "loss": 3.4017,
      "step": 200490
    },
    {
      "epoch": 1.041015119548004,
      "grad_norm": 0.36695021390914917,
      "learning_rate": 0.00016262052842345466,
      "loss": 3.4175,
      "step": 200500
    },
    {
      "epoch": 1.0410896924926367,
      "grad_norm": 0.38112685084342957,
      "learning_rate": 0.00016261866410135947,
      "loss": 3.3367,
      "step": 200510
    },
    {
      "epoch": 1.0411642654372695,
      "grad_norm": 0.37143853306770325,
      "learning_rate": 0.00016261679977926428,
      "loss": 3.3381,
      "step": 200520
    },
    {
      "epoch": 1.0412388383819022,
      "grad_norm": 0.36953139305114746,
      "learning_rate": 0.00016261493545716908,
      "loss": 3.3919,
      "step": 200530
    },
    {
      "epoch": 1.041313411326535,
      "grad_norm": 0.3633052706718445,
      "learning_rate": 0.00016261307113507386,
      "loss": 3.319,
      "step": 200540
    },
    {
      "epoch": 1.0413879842711677,
      "grad_norm": 0.39210453629493713,
      "learning_rate": 0.00016261120681297867,
      "loss": 3.4033,
      "step": 200550
    },
    {
      "epoch": 1.0414625572158005,
      "grad_norm": 0.38161906599998474,
      "learning_rate": 0.00016260934249088347,
      "loss": 3.3936,
      "step": 200560
    },
    {
      "epoch": 1.0415371301604333,
      "grad_norm": 0.3448023498058319,
      "learning_rate": 0.00016260747816878828,
      "loss": 3.4231,
      "step": 200570
    },
    {
      "epoch": 1.041611703105066,
      "grad_norm": 0.38725560903549194,
      "learning_rate": 0.0001626056138466931,
      "loss": 3.4462,
      "step": 200580
    },
    {
      "epoch": 1.0416862760496988,
      "grad_norm": 0.38122984766960144,
      "learning_rate": 0.00016260374952459787,
      "loss": 3.425,
      "step": 200590
    },
    {
      "epoch": 1.0417608489943313,
      "grad_norm": 0.3773456811904907,
      "learning_rate": 0.00016260188520250267,
      "loss": 3.3384,
      "step": 200600
    },
    {
      "epoch": 1.041835421938964,
      "grad_norm": 0.36919060349464417,
      "learning_rate": 0.00016260002088040748,
      "loss": 3.357,
      "step": 200610
    },
    {
      "epoch": 1.0419099948835968,
      "grad_norm": 0.35671013593673706,
      "learning_rate": 0.00016259815655831228,
      "loss": 3.4057,
      "step": 200620
    },
    {
      "epoch": 1.0419845678282296,
      "grad_norm": 0.3720661699771881,
      "learning_rate": 0.0001625962922362171,
      "loss": 3.2936,
      "step": 200630
    },
    {
      "epoch": 1.0420591407728623,
      "grad_norm": 0.3749493360519409,
      "learning_rate": 0.00016259442791412187,
      "loss": 3.3817,
      "step": 200640
    },
    {
      "epoch": 1.042133713717495,
      "grad_norm": 0.3611536920070648,
      "learning_rate": 0.00016259256359202668,
      "loss": 3.3728,
      "step": 200650
    },
    {
      "epoch": 1.0422082866621278,
      "grad_norm": 0.5578046441078186,
      "learning_rate": 0.00016259069926993146,
      "loss": 3.3353,
      "step": 200660
    },
    {
      "epoch": 1.0422828596067606,
      "grad_norm": 0.38181453943252563,
      "learning_rate": 0.0001625888349478363,
      "loss": 3.3588,
      "step": 200670
    },
    {
      "epoch": 1.0423574325513933,
      "grad_norm": 0.4799559712409973,
      "learning_rate": 0.0001625869706257411,
      "loss": 3.3701,
      "step": 200680
    },
    {
      "epoch": 1.042432005496026,
      "grad_norm": 0.3911532759666443,
      "learning_rate": 0.00016258510630364587,
      "loss": 3.3371,
      "step": 200690
    },
    {
      "epoch": 1.0425065784406589,
      "grad_norm": 0.36417141556739807,
      "learning_rate": 0.00016258324198155068,
      "loss": 3.3845,
      "step": 200700
    },
    {
      "epoch": 1.0425811513852914,
      "grad_norm": 0.39378082752227783,
      "learning_rate": 0.00016258137765945546,
      "loss": 3.4045,
      "step": 200710
    },
    {
      "epoch": 1.0426557243299241,
      "grad_norm": 0.38140735030174255,
      "learning_rate": 0.00016257951333736027,
      "loss": 3.3342,
      "step": 200720
    },
    {
      "epoch": 1.042730297274557,
      "grad_norm": 0.3699084520339966,
      "learning_rate": 0.0001625776490152651,
      "loss": 3.3803,
      "step": 200730
    },
    {
      "epoch": 1.0428048702191897,
      "grad_norm": 0.39252057671546936,
      "learning_rate": 0.00016257578469316988,
      "loss": 3.3337,
      "step": 200740
    },
    {
      "epoch": 1.0428794431638224,
      "grad_norm": 0.48733487725257874,
      "learning_rate": 0.00016257392037107469,
      "loss": 3.4119,
      "step": 200750
    },
    {
      "epoch": 1.0429540161084552,
      "grad_norm": 0.39796319603919983,
      "learning_rate": 0.00016257205604897946,
      "loss": 3.3455,
      "step": 200760
    },
    {
      "epoch": 1.043028589053088,
      "grad_norm": 0.3521192669868469,
      "learning_rate": 0.00016257019172688427,
      "loss": 3.4029,
      "step": 200770
    },
    {
      "epoch": 1.0431031619977207,
      "grad_norm": 0.3466090261936188,
      "learning_rate": 0.00016256832740478908,
      "loss": 3.4121,
      "step": 200780
    },
    {
      "epoch": 1.0431777349423534,
      "grad_norm": 0.3799853026866913,
      "learning_rate": 0.00016256646308269388,
      "loss": 3.42,
      "step": 200790
    },
    {
      "epoch": 1.0432523078869862,
      "grad_norm": 0.3834293782711029,
      "learning_rate": 0.0001625645987605987,
      "loss": 3.3642,
      "step": 200800
    },
    {
      "epoch": 1.0433268808316187,
      "grad_norm": 0.365029513835907,
      "learning_rate": 0.00016256273443850347,
      "loss": 3.3824,
      "step": 200810
    },
    {
      "epoch": 1.0434014537762515,
      "grad_norm": 0.40239405632019043,
      "learning_rate": 0.00016256087011640828,
      "loss": 3.3459,
      "step": 200820
    },
    {
      "epoch": 1.0434760267208842,
      "grad_norm": 0.42213675379753113,
      "learning_rate": 0.00016255900579431308,
      "loss": 3.3375,
      "step": 200830
    },
    {
      "epoch": 1.043550599665517,
      "grad_norm": 0.39048126339912415,
      "learning_rate": 0.0001625571414722179,
      "loss": 3.4314,
      "step": 200840
    },
    {
      "epoch": 1.0436251726101498,
      "grad_norm": 0.36596760153770447,
      "learning_rate": 0.0001625552771501227,
      "loss": 3.3856,
      "step": 200850
    },
    {
      "epoch": 1.0436997455547825,
      "grad_norm": 0.37961238622665405,
      "learning_rate": 0.00016255341282802747,
      "loss": 3.3563,
      "step": 200860
    },
    {
      "epoch": 1.0437743184994153,
      "grad_norm": 0.3698575496673584,
      "learning_rate": 0.00016255154850593228,
      "loss": 3.4401,
      "step": 200870
    },
    {
      "epoch": 1.043848891444048,
      "grad_norm": 0.3657115399837494,
      "learning_rate": 0.00016254968418383706,
      "loss": 3.4013,
      "step": 200880
    },
    {
      "epoch": 1.0439234643886808,
      "grad_norm": 0.3740396797657013,
      "learning_rate": 0.0001625478198617419,
      "loss": 3.4017,
      "step": 200890
    },
    {
      "epoch": 1.0439980373333135,
      "grad_norm": 0.4048653244972229,
      "learning_rate": 0.0001625459555396467,
      "loss": 3.3111,
      "step": 200900
    },
    {
      "epoch": 1.0440726102779463,
      "grad_norm": 0.41640371084213257,
      "learning_rate": 0.00016254409121755148,
      "loss": 3.4049,
      "step": 200910
    },
    {
      "epoch": 1.0441471832225788,
      "grad_norm": 0.40346866846084595,
      "learning_rate": 0.00016254222689545628,
      "loss": 3.3511,
      "step": 200920
    },
    {
      "epoch": 1.0442217561672116,
      "grad_norm": 0.4272169768810272,
      "learning_rate": 0.00016254036257336106,
      "loss": 3.4053,
      "step": 200930
    },
    {
      "epoch": 1.0442963291118443,
      "grad_norm": 0.3623203933238983,
      "learning_rate": 0.00016253849825126587,
      "loss": 3.4398,
      "step": 200940
    },
    {
      "epoch": 1.044370902056477,
      "grad_norm": 0.3770248591899872,
      "learning_rate": 0.0001625366339291707,
      "loss": 3.3881,
      "step": 200950
    },
    {
      "epoch": 1.0444454750011098,
      "grad_norm": 0.3963110148906708,
      "learning_rate": 0.00016253476960707548,
      "loss": 3.3647,
      "step": 200960
    },
    {
      "epoch": 1.0445200479457426,
      "grad_norm": 0.3629761040210724,
      "learning_rate": 0.0001625329052849803,
      "loss": 3.387,
      "step": 200970
    },
    {
      "epoch": 1.0445946208903754,
      "grad_norm": 0.3987675905227661,
      "learning_rate": 0.00016253104096288507,
      "loss": 3.3352,
      "step": 200980
    },
    {
      "epoch": 1.0446691938350081,
      "grad_norm": 0.3813416361808777,
      "learning_rate": 0.00016252917664078987,
      "loss": 3.4079,
      "step": 200990
    },
    {
      "epoch": 1.0447437667796409,
      "grad_norm": 0.40967780351638794,
      "learning_rate": 0.00016252731231869468,
      "loss": 3.3718,
      "step": 201000
    },
    {
      "epoch": 1.0448183397242736,
      "grad_norm": 0.3919924199581146,
      "learning_rate": 0.0001625254479965995,
      "loss": 3.4008,
      "step": 201010
    },
    {
      "epoch": 1.0448929126689062,
      "grad_norm": 0.4068112373352051,
      "learning_rate": 0.0001625235836745043,
      "loss": 3.3304,
      "step": 201020
    },
    {
      "epoch": 1.044967485613539,
      "grad_norm": 0.389161616563797,
      "learning_rate": 0.00016252171935240907,
      "loss": 3.3909,
      "step": 201030
    },
    {
      "epoch": 1.0450420585581717,
      "grad_norm": 0.36363622546195984,
      "learning_rate": 0.00016251985503031388,
      "loss": 3.3412,
      "step": 201040
    },
    {
      "epoch": 1.0451166315028044,
      "grad_norm": 0.36001700162887573,
      "learning_rate": 0.00016251799070821868,
      "loss": 3.3628,
      "step": 201050
    },
    {
      "epoch": 1.0451912044474372,
      "grad_norm": 0.38783732056617737,
      "learning_rate": 0.0001625161263861235,
      "loss": 3.3159,
      "step": 201060
    },
    {
      "epoch": 1.04526577739207,
      "grad_norm": 0.3590431213378906,
      "learning_rate": 0.0001625142620640283,
      "loss": 3.2935,
      "step": 201070
    },
    {
      "epoch": 1.0453403503367027,
      "grad_norm": 0.3714081645011902,
      "learning_rate": 0.00016251239774193308,
      "loss": 3.4358,
      "step": 201080
    },
    {
      "epoch": 1.0454149232813355,
      "grad_norm": 0.3946825861930847,
      "learning_rate": 0.00016251053341983788,
      "loss": 3.3628,
      "step": 201090
    },
    {
      "epoch": 1.0454894962259682,
      "grad_norm": 0.3749505579471588,
      "learning_rate": 0.0001625086690977427,
      "loss": 3.3333,
      "step": 201100
    },
    {
      "epoch": 1.045564069170601,
      "grad_norm": 0.376199334859848,
      "learning_rate": 0.00016250680477564747,
      "loss": 3.4018,
      "step": 201110
    },
    {
      "epoch": 1.0456386421152337,
      "grad_norm": 0.37649092078208923,
      "learning_rate": 0.0001625049404535523,
      "loss": 3.2976,
      "step": 201120
    },
    {
      "epoch": 1.0457132150598663,
      "grad_norm": 0.3779640197753906,
      "learning_rate": 0.00016250307613145708,
      "loss": 3.3276,
      "step": 201130
    },
    {
      "epoch": 1.045787788004499,
      "grad_norm": 0.4071195721626282,
      "learning_rate": 0.0001625012118093619,
      "loss": 3.4229,
      "step": 201140
    },
    {
      "epoch": 1.0458623609491318,
      "grad_norm": 0.4019864499568939,
      "learning_rate": 0.0001624993474872667,
      "loss": 3.3652,
      "step": 201150
    },
    {
      "epoch": 1.0459369338937645,
      "grad_norm": 0.4003325402736664,
      "learning_rate": 0.00016249748316517147,
      "loss": 3.3524,
      "step": 201160
    },
    {
      "epoch": 1.0460115068383973,
      "grad_norm": 0.35357174277305603,
      "learning_rate": 0.00016249561884307628,
      "loss": 3.4627,
      "step": 201170
    },
    {
      "epoch": 1.04608607978303,
      "grad_norm": 0.3884296417236328,
      "learning_rate": 0.00016249375452098109,
      "loss": 3.382,
      "step": 201180
    },
    {
      "epoch": 1.0461606527276628,
      "grad_norm": 0.3668677806854248,
      "learning_rate": 0.0001624918901988859,
      "loss": 3.3472,
      "step": 201190
    },
    {
      "epoch": 1.0462352256722955,
      "grad_norm": 0.37897202372550964,
      "learning_rate": 0.0001624900258767907,
      "loss": 3.3488,
      "step": 201200
    },
    {
      "epoch": 1.0463097986169283,
      "grad_norm": 0.3991897702217102,
      "learning_rate": 0.00016248816155469548,
      "loss": 3.4105,
      "step": 201210
    },
    {
      "epoch": 1.046384371561561,
      "grad_norm": 0.35903260111808777,
      "learning_rate": 0.00016248629723260028,
      "loss": 3.4268,
      "step": 201220
    },
    {
      "epoch": 1.0464589445061936,
      "grad_norm": 0.40027913451194763,
      "learning_rate": 0.0001624844329105051,
      "loss": 3.4226,
      "step": 201230
    },
    {
      "epoch": 1.0465335174508263,
      "grad_norm": 0.38624706864356995,
      "learning_rate": 0.0001624825685884099,
      "loss": 3.2916,
      "step": 201240
    },
    {
      "epoch": 1.046608090395459,
      "grad_norm": 0.3902931213378906,
      "learning_rate": 0.0001624807042663147,
      "loss": 3.3982,
      "step": 201250
    },
    {
      "epoch": 1.0466826633400919,
      "grad_norm": 0.34722524881362915,
      "learning_rate": 0.00016247883994421948,
      "loss": 3.4428,
      "step": 201260
    },
    {
      "epoch": 1.0467572362847246,
      "grad_norm": 0.36892855167388916,
      "learning_rate": 0.0001624769756221243,
      "loss": 3.3661,
      "step": 201270
    },
    {
      "epoch": 1.0468318092293574,
      "grad_norm": 0.3697534203529358,
      "learning_rate": 0.0001624751113000291,
      "loss": 3.3223,
      "step": 201280
    },
    {
      "epoch": 1.0469063821739901,
      "grad_norm": 0.3628511130809784,
      "learning_rate": 0.0001624732469779339,
      "loss": 3.3539,
      "step": 201290
    },
    {
      "epoch": 1.0469809551186229,
      "grad_norm": 0.3571692705154419,
      "learning_rate": 0.0001624713826558387,
      "loss": 3.3832,
      "step": 201300
    },
    {
      "epoch": 1.0470555280632556,
      "grad_norm": 0.37711575627326965,
      "learning_rate": 0.00016246951833374349,
      "loss": 3.4008,
      "step": 201310
    },
    {
      "epoch": 1.0471301010078884,
      "grad_norm": 0.3753349483013153,
      "learning_rate": 0.0001624676540116483,
      "loss": 3.3731,
      "step": 201320
    },
    {
      "epoch": 1.0472046739525211,
      "grad_norm": 0.38078996539115906,
      "learning_rate": 0.00016246578968955307,
      "loss": 3.3661,
      "step": 201330
    },
    {
      "epoch": 1.0472792468971537,
      "grad_norm": 0.36709484457969666,
      "learning_rate": 0.0001624639253674579,
      "loss": 3.3952,
      "step": 201340
    },
    {
      "epoch": 1.0473538198417864,
      "grad_norm": 0.3725516200065613,
      "learning_rate": 0.0001624620610453627,
      "loss": 3.3827,
      "step": 201350
    },
    {
      "epoch": 1.0474283927864192,
      "grad_norm": 0.36423569917678833,
      "learning_rate": 0.0001624601967232675,
      "loss": 3.4042,
      "step": 201360
    },
    {
      "epoch": 1.047502965731052,
      "grad_norm": 0.3673231303691864,
      "learning_rate": 0.0001624583324011723,
      "loss": 3.3006,
      "step": 201370
    },
    {
      "epoch": 1.0475775386756847,
      "grad_norm": 0.35028737783432007,
      "learning_rate": 0.00016245646807907708,
      "loss": 3.4089,
      "step": 201380
    },
    {
      "epoch": 1.0476521116203175,
      "grad_norm": 0.36249545216560364,
      "learning_rate": 0.00016245460375698188,
      "loss": 3.3604,
      "step": 201390
    },
    {
      "epoch": 1.0477266845649502,
      "grad_norm": 0.3840806484222412,
      "learning_rate": 0.00016245273943488672,
      "loss": 3.351,
      "step": 201400
    },
    {
      "epoch": 1.047801257509583,
      "grad_norm": 0.41626766324043274,
      "learning_rate": 0.0001624508751127915,
      "loss": 3.3463,
      "step": 201410
    },
    {
      "epoch": 1.0478758304542157,
      "grad_norm": 0.4076053500175476,
      "learning_rate": 0.0001624490107906963,
      "loss": 3.3748,
      "step": 201420
    },
    {
      "epoch": 1.0479504033988485,
      "grad_norm": 0.35679012537002563,
      "learning_rate": 0.00016244714646860108,
      "loss": 3.4116,
      "step": 201430
    },
    {
      "epoch": 1.048024976343481,
      "grad_norm": 0.39297983050346375,
      "learning_rate": 0.0001624452821465059,
      "loss": 3.3596,
      "step": 201440
    },
    {
      "epoch": 1.0480995492881138,
      "grad_norm": 0.38308075070381165,
      "learning_rate": 0.0001624434178244107,
      "loss": 3.3162,
      "step": 201450
    },
    {
      "epoch": 1.0481741222327465,
      "grad_norm": 0.40734297037124634,
      "learning_rate": 0.0001624415535023155,
      "loss": 3.3377,
      "step": 201460
    },
    {
      "epoch": 1.0482486951773793,
      "grad_norm": 0.40170735120773315,
      "learning_rate": 0.0001624396891802203,
      "loss": 3.3912,
      "step": 201470
    },
    {
      "epoch": 1.048323268122012,
      "grad_norm": 0.37169349193573,
      "learning_rate": 0.00016243782485812509,
      "loss": 3.3737,
      "step": 201480
    },
    {
      "epoch": 1.0483978410666448,
      "grad_norm": 0.3819315433502197,
      "learning_rate": 0.0001624359605360299,
      "loss": 3.3846,
      "step": 201490
    },
    {
      "epoch": 1.0484724140112776,
      "grad_norm": 0.3830302357673645,
      "learning_rate": 0.0001624340962139347,
      "loss": 3.3588,
      "step": 201500
    },
    {
      "epoch": 1.0485469869559103,
      "grad_norm": 0.38354626297950745,
      "learning_rate": 0.0001624322318918395,
      "loss": 3.3944,
      "step": 201510
    },
    {
      "epoch": 1.048621559900543,
      "grad_norm": 0.3668978810310364,
      "learning_rate": 0.0001624303675697443,
      "loss": 3.3408,
      "step": 201520
    },
    {
      "epoch": 1.0486961328451758,
      "grad_norm": 0.3797697424888611,
      "learning_rate": 0.0001624285032476491,
      "loss": 3.4146,
      "step": 201530
    },
    {
      "epoch": 1.0487707057898086,
      "grad_norm": 0.3691090941429138,
      "learning_rate": 0.0001624266389255539,
      "loss": 3.3942,
      "step": 201540
    },
    {
      "epoch": 1.048845278734441,
      "grad_norm": 0.3789222836494446,
      "learning_rate": 0.0001624247746034587,
      "loss": 3.3498,
      "step": 201550
    },
    {
      "epoch": 1.0489198516790739,
      "grad_norm": 0.3755575120449066,
      "learning_rate": 0.00016242291028136348,
      "loss": 3.3866,
      "step": 201560
    },
    {
      "epoch": 1.0489944246237066,
      "grad_norm": 0.41120269894599915,
      "learning_rate": 0.00016242104595926831,
      "loss": 3.3355,
      "step": 201570
    },
    {
      "epoch": 1.0490689975683394,
      "grad_norm": 0.40797609090805054,
      "learning_rate": 0.0001624191816371731,
      "loss": 3.3415,
      "step": 201580
    },
    {
      "epoch": 1.0491435705129721,
      "grad_norm": 0.39431336522102356,
      "learning_rate": 0.0001624173173150779,
      "loss": 3.3777,
      "step": 201590
    },
    {
      "epoch": 1.049218143457605,
      "grad_norm": 0.4062853157520294,
      "learning_rate": 0.0001624154529929827,
      "loss": 3.2908,
      "step": 201600
    },
    {
      "epoch": 1.0492927164022376,
      "grad_norm": 0.4023999869823456,
      "learning_rate": 0.00016241358867088749,
      "loss": 3.3921,
      "step": 201610
    },
    {
      "epoch": 1.0493672893468704,
      "grad_norm": 0.3940250873565674,
      "learning_rate": 0.0001624117243487923,
      "loss": 3.3084,
      "step": 201620
    },
    {
      "epoch": 1.0494418622915032,
      "grad_norm": 0.3657448887825012,
      "learning_rate": 0.0001624098600266971,
      "loss": 3.3311,
      "step": 201630
    },
    {
      "epoch": 1.049516435236136,
      "grad_norm": 0.38048622012138367,
      "learning_rate": 0.0001624079957046019,
      "loss": 3.3593,
      "step": 201640
    },
    {
      "epoch": 1.0495910081807684,
      "grad_norm": 0.3703060746192932,
      "learning_rate": 0.0001624061313825067,
      "loss": 3.3753,
      "step": 201650
    },
    {
      "epoch": 1.0496655811254012,
      "grad_norm": 0.3576778769493103,
      "learning_rate": 0.0001624042670604115,
      "loss": 3.3967,
      "step": 201660
    },
    {
      "epoch": 1.049740154070034,
      "grad_norm": 0.4084343910217285,
      "learning_rate": 0.0001624024027383163,
      "loss": 3.4871,
      "step": 201670
    },
    {
      "epoch": 1.0498147270146667,
      "grad_norm": 0.3882865011692047,
      "learning_rate": 0.0001624005384162211,
      "loss": 3.3763,
      "step": 201680
    },
    {
      "epoch": 1.0498892999592995,
      "grad_norm": 0.41255098581314087,
      "learning_rate": 0.0001623986740941259,
      "loss": 3.3222,
      "step": 201690
    },
    {
      "epoch": 1.0499638729039322,
      "grad_norm": 0.37703588604927063,
      "learning_rate": 0.00016239680977203072,
      "loss": 3.423,
      "step": 201700
    },
    {
      "epoch": 1.050038445848565,
      "grad_norm": 0.36820825934410095,
      "learning_rate": 0.0001623949454499355,
      "loss": 3.3778,
      "step": 201710
    },
    {
      "epoch": 1.0501130187931977,
      "grad_norm": 0.37239837646484375,
      "learning_rate": 0.0001623930811278403,
      "loss": 3.3424,
      "step": 201720
    },
    {
      "epoch": 1.0501875917378305,
      "grad_norm": 0.3849492371082306,
      "learning_rate": 0.0001623912168057451,
      "loss": 3.3868,
      "step": 201730
    },
    {
      "epoch": 1.0502621646824633,
      "grad_norm": 0.4042765200138092,
      "learning_rate": 0.00016238935248364991,
      "loss": 3.3384,
      "step": 201740
    },
    {
      "epoch": 1.050336737627096,
      "grad_norm": 0.36430633068084717,
      "learning_rate": 0.00016238748816155472,
      "loss": 3.3567,
      "step": 201750
    },
    {
      "epoch": 1.0504113105717285,
      "grad_norm": 0.43944600224494934,
      "learning_rate": 0.0001623856238394595,
      "loss": 3.4008,
      "step": 201760
    },
    {
      "epoch": 1.0504858835163613,
      "grad_norm": 0.36331599950790405,
      "learning_rate": 0.0001623837595173643,
      "loss": 3.4035,
      "step": 201770
    },
    {
      "epoch": 1.050560456460994,
      "grad_norm": 0.3834858536720276,
      "learning_rate": 0.00016238189519526908,
      "loss": 3.3425,
      "step": 201780
    },
    {
      "epoch": 1.0506350294056268,
      "grad_norm": 0.4045805335044861,
      "learning_rate": 0.00016238003087317392,
      "loss": 3.3832,
      "step": 201790
    },
    {
      "epoch": 1.0507096023502596,
      "grad_norm": 0.3969498574733734,
      "learning_rate": 0.00016237816655107872,
      "loss": 3.3337,
      "step": 201800
    },
    {
      "epoch": 1.0507841752948923,
      "grad_norm": 0.36703866720199585,
      "learning_rate": 0.0001623763022289835,
      "loss": 3.3986,
      "step": 201810
    },
    {
      "epoch": 1.050858748239525,
      "grad_norm": 0.37666216492652893,
      "learning_rate": 0.0001623744379068883,
      "loss": 3.3969,
      "step": 201820
    },
    {
      "epoch": 1.0509333211841578,
      "grad_norm": 0.38396257162094116,
      "learning_rate": 0.0001623725735847931,
      "loss": 3.3501,
      "step": 201830
    },
    {
      "epoch": 1.0510078941287906,
      "grad_norm": 0.38639992475509644,
      "learning_rate": 0.0001623707092626979,
      "loss": 3.3128,
      "step": 201840
    },
    {
      "epoch": 1.0510824670734233,
      "grad_norm": 0.3904225528240204,
      "learning_rate": 0.00016236884494060273,
      "loss": 3.3386,
      "step": 201850
    },
    {
      "epoch": 1.0511570400180559,
      "grad_norm": 0.3846086263656616,
      "learning_rate": 0.0001623669806185075,
      "loss": 3.3397,
      "step": 201860
    },
    {
      "epoch": 1.0512316129626886,
      "grad_norm": 0.38420796394348145,
      "learning_rate": 0.00016236511629641231,
      "loss": 3.2828,
      "step": 201870
    },
    {
      "epoch": 1.0513061859073214,
      "grad_norm": 0.3726405203342438,
      "learning_rate": 0.0001623632519743171,
      "loss": 3.4466,
      "step": 201880
    },
    {
      "epoch": 1.0513807588519541,
      "grad_norm": 0.3759957551956177,
      "learning_rate": 0.0001623613876522219,
      "loss": 3.348,
      "step": 201890
    },
    {
      "epoch": 1.051455331796587,
      "grad_norm": 0.3726506531238556,
      "learning_rate": 0.0001623595233301267,
      "loss": 3.3672,
      "step": 201900
    },
    {
      "epoch": 1.0515299047412197,
      "grad_norm": 0.3565724492073059,
      "learning_rate": 0.0001623576590080315,
      "loss": 3.3991,
      "step": 201910
    },
    {
      "epoch": 1.0516044776858524,
      "grad_norm": 0.36144858598709106,
      "learning_rate": 0.00016235579468593632,
      "loss": 3.3534,
      "step": 201920
    },
    {
      "epoch": 1.0516790506304852,
      "grad_norm": 0.41320163011550903,
      "learning_rate": 0.0001623539303638411,
      "loss": 3.3611,
      "step": 201930
    },
    {
      "epoch": 1.051753623575118,
      "grad_norm": 0.3836438059806824,
      "learning_rate": 0.0001623520660417459,
      "loss": 3.3752,
      "step": 201940
    },
    {
      "epoch": 1.0518281965197507,
      "grad_norm": 0.4074147641658783,
      "learning_rate": 0.0001623502017196507,
      "loss": 3.3535,
      "step": 201950
    },
    {
      "epoch": 1.0519027694643832,
      "grad_norm": 0.3895786702632904,
      "learning_rate": 0.00016234833739755552,
      "loss": 3.4026,
      "step": 201960
    },
    {
      "epoch": 1.051977342409016,
      "grad_norm": 0.41543838381767273,
      "learning_rate": 0.00016234647307546032,
      "loss": 3.3847,
      "step": 201970
    },
    {
      "epoch": 1.0520519153536487,
      "grad_norm": 0.38099825382232666,
      "learning_rate": 0.0001623446087533651,
      "loss": 3.3324,
      "step": 201980
    },
    {
      "epoch": 1.0521264882982815,
      "grad_norm": 0.41633448004722595,
      "learning_rate": 0.0001623427444312699,
      "loss": 3.3169,
      "step": 201990
    },
    {
      "epoch": 1.0522010612429142,
      "grad_norm": 0.4263478219509125,
      "learning_rate": 0.00016234088010917471,
      "loss": 3.3097,
      "step": 202000
    },
    {
      "epoch": 1.052275634187547,
      "grad_norm": 0.3720034658908844,
      "learning_rate": 0.0001623390157870795,
      "loss": 3.341,
      "step": 202010
    },
    {
      "epoch": 1.0523502071321797,
      "grad_norm": 0.3788544237613678,
      "learning_rate": 0.00016233715146498433,
      "loss": 3.3422,
      "step": 202020
    },
    {
      "epoch": 1.0524247800768125,
      "grad_norm": 0.3870861232280731,
      "learning_rate": 0.0001623352871428891,
      "loss": 3.4072,
      "step": 202030
    },
    {
      "epoch": 1.0524993530214453,
      "grad_norm": 0.39346736669540405,
      "learning_rate": 0.0001623334228207939,
      "loss": 3.3791,
      "step": 202040
    },
    {
      "epoch": 1.052573925966078,
      "grad_norm": 0.3568359613418579,
      "learning_rate": 0.00016233155849869872,
      "loss": 3.4203,
      "step": 202050
    },
    {
      "epoch": 1.0526484989107108,
      "grad_norm": 0.3568083345890045,
      "learning_rate": 0.0001623296941766035,
      "loss": 3.4125,
      "step": 202060
    },
    {
      "epoch": 1.0527230718553433,
      "grad_norm": 0.38862067461013794,
      "learning_rate": 0.0001623278298545083,
      "loss": 3.3385,
      "step": 202070
    },
    {
      "epoch": 1.052797644799976,
      "grad_norm": 0.3797299861907959,
      "learning_rate": 0.0001623259655324131,
      "loss": 3.3903,
      "step": 202080
    },
    {
      "epoch": 1.0528722177446088,
      "grad_norm": 0.4006911516189575,
      "learning_rate": 0.00016232410121031792,
      "loss": 3.3691,
      "step": 202090
    },
    {
      "epoch": 1.0529467906892416,
      "grad_norm": 0.3813890218734741,
      "learning_rate": 0.00016232223688822272,
      "loss": 3.3652,
      "step": 202100
    },
    {
      "epoch": 1.0530213636338743,
      "grad_norm": 0.3689267039299011,
      "learning_rate": 0.0001623203725661275,
      "loss": 3.3973,
      "step": 202110
    },
    {
      "epoch": 1.053095936578507,
      "grad_norm": 0.36007627844810486,
      "learning_rate": 0.0001623185082440323,
      "loss": 3.3126,
      "step": 202120
    },
    {
      "epoch": 1.0531705095231398,
      "grad_norm": 0.36417290568351746,
      "learning_rate": 0.00016231664392193712,
      "loss": 3.3621,
      "step": 202130
    },
    {
      "epoch": 1.0532450824677726,
      "grad_norm": 0.37864792346954346,
      "learning_rate": 0.00016231477959984192,
      "loss": 3.3769,
      "step": 202140
    },
    {
      "epoch": 1.0533196554124054,
      "grad_norm": 0.38978710770606995,
      "learning_rate": 0.00016231291527774673,
      "loss": 3.4269,
      "step": 202150
    },
    {
      "epoch": 1.053394228357038,
      "grad_norm": 0.38785141706466675,
      "learning_rate": 0.0001623110509556515,
      "loss": 3.3594,
      "step": 202160
    },
    {
      "epoch": 1.0534688013016706,
      "grad_norm": 0.3810710906982422,
      "learning_rate": 0.00016230918663355631,
      "loss": 3.4216,
      "step": 202170
    },
    {
      "epoch": 1.0535433742463034,
      "grad_norm": 0.3706696927547455,
      "learning_rate": 0.00016230732231146112,
      "loss": 3.3223,
      "step": 202180
    },
    {
      "epoch": 1.0536179471909362,
      "grad_norm": 0.369245707988739,
      "learning_rate": 0.00016230545798936593,
      "loss": 3.392,
      "step": 202190
    },
    {
      "epoch": 1.053692520135569,
      "grad_norm": 0.3986053764820099,
      "learning_rate": 0.00016230359366727073,
      "loss": 3.3226,
      "step": 202200
    },
    {
      "epoch": 1.0537670930802017,
      "grad_norm": 0.3994375467300415,
      "learning_rate": 0.0001623017293451755,
      "loss": 3.384,
      "step": 202210
    },
    {
      "epoch": 1.0538416660248344,
      "grad_norm": 0.36864352226257324,
      "learning_rate": 0.00016229986502308032,
      "loss": 3.3866,
      "step": 202220
    },
    {
      "epoch": 1.0539162389694672,
      "grad_norm": 0.402646005153656,
      "learning_rate": 0.0001622980007009851,
      "loss": 3.3732,
      "step": 202230
    },
    {
      "epoch": 1.0539908119141,
      "grad_norm": 0.3987089693546295,
      "learning_rate": 0.00016229613637888993,
      "loss": 3.3602,
      "step": 202240
    },
    {
      "epoch": 1.0540653848587327,
      "grad_norm": 0.39012980461120605,
      "learning_rate": 0.0001622942720567947,
      "loss": 3.4358,
      "step": 202250
    },
    {
      "epoch": 1.0541399578033654,
      "grad_norm": 0.3893141746520996,
      "learning_rate": 0.00016229240773469952,
      "loss": 3.3815,
      "step": 202260
    },
    {
      "epoch": 1.0542145307479982,
      "grad_norm": 0.37791353464126587,
      "learning_rate": 0.00016229054341260432,
      "loss": 3.4073,
      "step": 202270
    },
    {
      "epoch": 1.0542891036926307,
      "grad_norm": 0.3781190812587738,
      "learning_rate": 0.0001622886790905091,
      "loss": 3.3592,
      "step": 202280
    },
    {
      "epoch": 1.0543636766372635,
      "grad_norm": 0.3733564019203186,
      "learning_rate": 0.0001622868147684139,
      "loss": 3.4185,
      "step": 202290
    },
    {
      "epoch": 1.0544382495818962,
      "grad_norm": 0.39259615540504456,
      "learning_rate": 0.00016228495044631871,
      "loss": 3.3784,
      "step": 202300
    },
    {
      "epoch": 1.054512822526529,
      "grad_norm": 0.3863612711429596,
      "learning_rate": 0.00016228308612422352,
      "loss": 3.4099,
      "step": 202310
    },
    {
      "epoch": 1.0545873954711618,
      "grad_norm": 0.3969041109085083,
      "learning_rate": 0.00016228122180212833,
      "loss": 3.2962,
      "step": 202320
    },
    {
      "epoch": 1.0546619684157945,
      "grad_norm": 0.3714589476585388,
      "learning_rate": 0.0001622793574800331,
      "loss": 3.3776,
      "step": 202330
    },
    {
      "epoch": 1.0547365413604273,
      "grad_norm": 0.40480056405067444,
      "learning_rate": 0.0001622774931579379,
      "loss": 3.3677,
      "step": 202340
    },
    {
      "epoch": 1.05481111430506,
      "grad_norm": 0.3771243691444397,
      "learning_rate": 0.00016227562883584272,
      "loss": 3.4013,
      "step": 202350
    },
    {
      "epoch": 1.0548856872496928,
      "grad_norm": 0.39335963129997253,
      "learning_rate": 0.00016227376451374753,
      "loss": 3.3431,
      "step": 202360
    },
    {
      "epoch": 1.0549602601943255,
      "grad_norm": 0.396899938583374,
      "learning_rate": 0.00016227190019165233,
      "loss": 3.3577,
      "step": 202370
    },
    {
      "epoch": 1.055034833138958,
      "grad_norm": 0.37651732563972473,
      "learning_rate": 0.0001622700358695571,
      "loss": 3.3803,
      "step": 202380
    },
    {
      "epoch": 1.0551094060835908,
      "grad_norm": 0.39244258403778076,
      "learning_rate": 0.00016226817154746192,
      "loss": 3.3262,
      "step": 202390
    },
    {
      "epoch": 1.0551839790282236,
      "grad_norm": 0.3593714237213135,
      "learning_rate": 0.0001622663072253667,
      "loss": 3.4239,
      "step": 202400
    },
    {
      "epoch": 1.0552585519728563,
      "grad_norm": 0.35799363255500793,
      "learning_rate": 0.00016226444290327153,
      "loss": 3.3804,
      "step": 202410
    },
    {
      "epoch": 1.055333124917489,
      "grad_norm": 0.3732289671897888,
      "learning_rate": 0.00016226257858117634,
      "loss": 3.4156,
      "step": 202420
    },
    {
      "epoch": 1.0554076978621219,
      "grad_norm": 0.37658247351646423,
      "learning_rate": 0.00016226071425908112,
      "loss": 3.3921,
      "step": 202430
    },
    {
      "epoch": 1.0554822708067546,
      "grad_norm": 0.3963947296142578,
      "learning_rate": 0.00016225884993698592,
      "loss": 3.3726,
      "step": 202440
    },
    {
      "epoch": 1.0555568437513874,
      "grad_norm": 0.40220126509666443,
      "learning_rate": 0.0001622569856148907,
      "loss": 3.3832,
      "step": 202450
    },
    {
      "epoch": 1.0556314166960201,
      "grad_norm": 0.3590339124202728,
      "learning_rate": 0.0001622551212927955,
      "loss": 3.4017,
      "step": 202460
    },
    {
      "epoch": 1.0557059896406529,
      "grad_norm": 0.3803196847438812,
      "learning_rate": 0.00016225325697070034,
      "loss": 3.4016,
      "step": 202470
    },
    {
      "epoch": 1.0557805625852854,
      "grad_norm": 0.3755674362182617,
      "learning_rate": 0.00016225139264860512,
      "loss": 3.315,
      "step": 202480
    },
    {
      "epoch": 1.0558551355299182,
      "grad_norm": 0.40980294346809387,
      "learning_rate": 0.00016224952832650993,
      "loss": 3.3637,
      "step": 202490
    },
    {
      "epoch": 1.055929708474551,
      "grad_norm": 0.3876471221446991,
      "learning_rate": 0.0001622476640044147,
      "loss": 3.4018,
      "step": 202500
    },
    {
      "epoch": 1.0560042814191837,
      "grad_norm": 0.387102872133255,
      "learning_rate": 0.0001622457996823195,
      "loss": 3.3572,
      "step": 202510
    },
    {
      "epoch": 1.0560788543638164,
      "grad_norm": 0.36982622742652893,
      "learning_rate": 0.00016224393536022432,
      "loss": 3.3463,
      "step": 202520
    },
    {
      "epoch": 1.0561534273084492,
      "grad_norm": 0.39777782559394836,
      "learning_rate": 0.00016224207103812912,
      "loss": 3.3731,
      "step": 202530
    },
    {
      "epoch": 1.056228000253082,
      "grad_norm": 0.3996470272541046,
      "learning_rate": 0.00016224020671603393,
      "loss": 3.4033,
      "step": 202540
    },
    {
      "epoch": 1.0563025731977147,
      "grad_norm": 0.40507110953330994,
      "learning_rate": 0.0001622383423939387,
      "loss": 3.3624,
      "step": 202550
    },
    {
      "epoch": 1.0563771461423475,
      "grad_norm": 0.36458417773246765,
      "learning_rate": 0.00016223647807184352,
      "loss": 3.3477,
      "step": 202560
    },
    {
      "epoch": 1.0564517190869802,
      "grad_norm": 0.37400275468826294,
      "learning_rate": 0.00016223461374974832,
      "loss": 3.3084,
      "step": 202570
    },
    {
      "epoch": 1.056526292031613,
      "grad_norm": 0.37516722083091736,
      "learning_rate": 0.00016223274942765313,
      "loss": 3.3646,
      "step": 202580
    },
    {
      "epoch": 1.0566008649762455,
      "grad_norm": 0.3742622137069702,
      "learning_rate": 0.00016223088510555793,
      "loss": 3.4174,
      "step": 202590
    },
    {
      "epoch": 1.0566754379208783,
      "grad_norm": 0.3963627815246582,
      "learning_rate": 0.00016222902078346271,
      "loss": 3.3456,
      "step": 202600
    },
    {
      "epoch": 1.056750010865511,
      "grad_norm": 0.39418989419937134,
      "learning_rate": 0.00016222715646136752,
      "loss": 3.3611,
      "step": 202610
    },
    {
      "epoch": 1.0568245838101438,
      "grad_norm": 0.38234004378318787,
      "learning_rate": 0.00016222529213927233,
      "loss": 3.2885,
      "step": 202620
    },
    {
      "epoch": 1.0568991567547765,
      "grad_norm": 0.38533133268356323,
      "learning_rate": 0.00016222342781717713,
      "loss": 3.4421,
      "step": 202630
    },
    {
      "epoch": 1.0569737296994093,
      "grad_norm": 0.3864726424217224,
      "learning_rate": 0.00016222156349508194,
      "loss": 3.3418,
      "step": 202640
    },
    {
      "epoch": 1.057048302644042,
      "grad_norm": 0.3734045922756195,
      "learning_rate": 0.00016221969917298672,
      "loss": 3.3358,
      "step": 202650
    },
    {
      "epoch": 1.0571228755886748,
      "grad_norm": 0.3626287877559662,
      "learning_rate": 0.00016221783485089152,
      "loss": 3.3739,
      "step": 202660
    },
    {
      "epoch": 1.0571974485333075,
      "grad_norm": 0.37100115418434143,
      "learning_rate": 0.00016221597052879633,
      "loss": 3.4359,
      "step": 202670
    },
    {
      "epoch": 1.0572720214779403,
      "grad_norm": 0.3453269600868225,
      "learning_rate": 0.0001622141062067011,
      "loss": 3.3354,
      "step": 202680
    },
    {
      "epoch": 1.0573465944225728,
      "grad_norm": 0.37902727723121643,
      "learning_rate": 0.00016221224188460594,
      "loss": 3.3458,
      "step": 202690
    },
    {
      "epoch": 1.0574211673672056,
      "grad_norm": 0.3780548572540283,
      "learning_rate": 0.00016221037756251072,
      "loss": 3.355,
      "step": 202700
    },
    {
      "epoch": 1.0574957403118384,
      "grad_norm": 0.3589743971824646,
      "learning_rate": 0.00016220851324041553,
      "loss": 3.2921,
      "step": 202710
    },
    {
      "epoch": 1.057570313256471,
      "grad_norm": 0.3898327350616455,
      "learning_rate": 0.00016220664891832034,
      "loss": 3.3041,
      "step": 202720
    },
    {
      "epoch": 1.0576448862011039,
      "grad_norm": 0.4010694622993469,
      "learning_rate": 0.00016220478459622511,
      "loss": 3.3593,
      "step": 202730
    },
    {
      "epoch": 1.0577194591457366,
      "grad_norm": 0.4084166884422302,
      "learning_rate": 0.00016220292027412992,
      "loss": 3.3762,
      "step": 202740
    },
    {
      "epoch": 1.0577940320903694,
      "grad_norm": 0.39388567209243774,
      "learning_rate": 0.00016220105595203473,
      "loss": 3.3628,
      "step": 202750
    },
    {
      "epoch": 1.0578686050350021,
      "grad_norm": 0.38735949993133545,
      "learning_rate": 0.00016219919162993953,
      "loss": 3.4075,
      "step": 202760
    },
    {
      "epoch": 1.0579431779796349,
      "grad_norm": 0.38689786195755005,
      "learning_rate": 0.00016219732730784434,
      "loss": 3.3356,
      "step": 202770
    },
    {
      "epoch": 1.0580177509242676,
      "grad_norm": 0.38230404257774353,
      "learning_rate": 0.00016219546298574912,
      "loss": 3.3508,
      "step": 202780
    },
    {
      "epoch": 1.0580923238689004,
      "grad_norm": 0.3835596740245819,
      "learning_rate": 0.00016219359866365393,
      "loss": 3.2892,
      "step": 202790
    },
    {
      "epoch": 1.058166896813533,
      "grad_norm": 0.3805653154850006,
      "learning_rate": 0.00016219173434155873,
      "loss": 3.3478,
      "step": 202800
    },
    {
      "epoch": 1.0582414697581657,
      "grad_norm": 0.3940104842185974,
      "learning_rate": 0.00016218987001946354,
      "loss": 3.3225,
      "step": 202810
    },
    {
      "epoch": 1.0583160427027984,
      "grad_norm": 0.3675227463245392,
      "learning_rate": 0.00016218800569736834,
      "loss": 3.3402,
      "step": 202820
    },
    {
      "epoch": 1.0583906156474312,
      "grad_norm": 0.3729732036590576,
      "learning_rate": 0.00016218614137527312,
      "loss": 3.3178,
      "step": 202830
    },
    {
      "epoch": 1.058465188592064,
      "grad_norm": 0.40192121267318726,
      "learning_rate": 0.00016218427705317793,
      "loss": 3.3538,
      "step": 202840
    },
    {
      "epoch": 1.0585397615366967,
      "grad_norm": 0.3741905987262726,
      "learning_rate": 0.0001621824127310827,
      "loss": 3.362,
      "step": 202850
    },
    {
      "epoch": 1.0586143344813295,
      "grad_norm": 0.3703516721725464,
      "learning_rate": 0.00016218054840898754,
      "loss": 3.3791,
      "step": 202860
    },
    {
      "epoch": 1.0586889074259622,
      "grad_norm": 0.35747650265693665,
      "learning_rate": 0.00016217868408689235,
      "loss": 3.3006,
      "step": 202870
    },
    {
      "epoch": 1.058763480370595,
      "grad_norm": 0.3656475841999054,
      "learning_rate": 0.00016217681976479713,
      "loss": 3.3381,
      "step": 202880
    },
    {
      "epoch": 1.0588380533152277,
      "grad_norm": 0.38162410259246826,
      "learning_rate": 0.00016217495544270193,
      "loss": 3.4318,
      "step": 202890
    },
    {
      "epoch": 1.0589126262598603,
      "grad_norm": 0.3640812635421753,
      "learning_rate": 0.0001621730911206067,
      "loss": 3.3172,
      "step": 202900
    },
    {
      "epoch": 1.058987199204493,
      "grad_norm": 0.38599276542663574,
      "learning_rate": 0.00016217122679851152,
      "loss": 3.37,
      "step": 202910
    },
    {
      "epoch": 1.0590617721491258,
      "grad_norm": 0.36262232065200806,
      "learning_rate": 0.00016216936247641635,
      "loss": 3.3701,
      "step": 202920
    },
    {
      "epoch": 1.0591363450937585,
      "grad_norm": 0.39517679810523987,
      "learning_rate": 0.00016216749815432113,
      "loss": 3.3263,
      "step": 202930
    },
    {
      "epoch": 1.0592109180383913,
      "grad_norm": 0.386079341173172,
      "learning_rate": 0.00016216563383222594,
      "loss": 3.4338,
      "step": 202940
    },
    {
      "epoch": 1.059285490983024,
      "grad_norm": 0.3879551589488983,
      "learning_rate": 0.00016216376951013072,
      "loss": 3.3468,
      "step": 202950
    },
    {
      "epoch": 1.0593600639276568,
      "grad_norm": 0.37003156542778015,
      "learning_rate": 0.00016216190518803552,
      "loss": 3.327,
      "step": 202960
    },
    {
      "epoch": 1.0594346368722896,
      "grad_norm": 0.3744369447231293,
      "learning_rate": 0.00016216004086594033,
      "loss": 3.3236,
      "step": 202970
    },
    {
      "epoch": 1.0595092098169223,
      "grad_norm": 0.3842534124851227,
      "learning_rate": 0.00016215817654384514,
      "loss": 3.3188,
      "step": 202980
    },
    {
      "epoch": 1.059583782761555,
      "grad_norm": 0.3477684557437897,
      "learning_rate": 0.00016215631222174994,
      "loss": 3.3798,
      "step": 202990
    },
    {
      "epoch": 1.0596583557061878,
      "grad_norm": 0.38065794110298157,
      "learning_rate": 0.00016215444789965472,
      "loss": 3.4099,
      "step": 203000
    },
    {
      "epoch": 1.0597329286508204,
      "grad_norm": 0.36584630608558655,
      "learning_rate": 0.00016215258357755953,
      "loss": 3.3363,
      "step": 203010
    },
    {
      "epoch": 1.0598075015954531,
      "grad_norm": 0.36164331436157227,
      "learning_rate": 0.00016215071925546434,
      "loss": 3.3864,
      "step": 203020
    },
    {
      "epoch": 1.0598820745400859,
      "grad_norm": 0.36654868721961975,
      "learning_rate": 0.00016214885493336914,
      "loss": 3.3248,
      "step": 203030
    },
    {
      "epoch": 1.0599566474847186,
      "grad_norm": 0.4002892076969147,
      "learning_rate": 0.00016214699061127395,
      "loss": 3.316,
      "step": 203040
    },
    {
      "epoch": 1.0600312204293514,
      "grad_norm": 0.3598577678203583,
      "learning_rate": 0.00016214512628917873,
      "loss": 3.3641,
      "step": 203050
    },
    {
      "epoch": 1.0601057933739841,
      "grad_norm": 0.38654279708862305,
      "learning_rate": 0.00016214326196708353,
      "loss": 3.3889,
      "step": 203060
    },
    {
      "epoch": 1.060180366318617,
      "grad_norm": 0.38883617520332336,
      "learning_rate": 0.00016214139764498834,
      "loss": 3.2853,
      "step": 203070
    },
    {
      "epoch": 1.0602549392632497,
      "grad_norm": 0.37180230021476746,
      "learning_rate": 0.00016213953332289315,
      "loss": 3.35,
      "step": 203080
    },
    {
      "epoch": 1.0603295122078824,
      "grad_norm": 0.3723490834236145,
      "learning_rate": 0.00016213766900079795,
      "loss": 3.3758,
      "step": 203090
    },
    {
      "epoch": 1.0604040851525152,
      "grad_norm": 0.39197129011154175,
      "learning_rate": 0.00016213580467870273,
      "loss": 3.3576,
      "step": 203100
    },
    {
      "epoch": 1.0604786580971477,
      "grad_norm": 0.3652418553829193,
      "learning_rate": 0.00016213394035660754,
      "loss": 3.3119,
      "step": 203110
    },
    {
      "epoch": 1.0605532310417805,
      "grad_norm": 0.38643887639045715,
      "learning_rate": 0.00016213207603451234,
      "loss": 3.3156,
      "step": 203120
    },
    {
      "epoch": 1.0606278039864132,
      "grad_norm": 0.41781285405158997,
      "learning_rate": 0.00016213021171241712,
      "loss": 3.358,
      "step": 203130
    },
    {
      "epoch": 1.060702376931046,
      "grad_norm": 0.4028626084327698,
      "learning_rate": 0.00016212834739032196,
      "loss": 3.3844,
      "step": 203140
    },
    {
      "epoch": 1.0607769498756787,
      "grad_norm": 0.3736307621002197,
      "learning_rate": 0.00016212648306822674,
      "loss": 3.3393,
      "step": 203150
    },
    {
      "epoch": 1.0608515228203115,
      "grad_norm": 0.3935365080833435,
      "learning_rate": 0.00016212461874613154,
      "loss": 3.3099,
      "step": 203160
    },
    {
      "epoch": 1.0609260957649442,
      "grad_norm": 0.39984479546546936,
      "learning_rate": 0.00016212275442403635,
      "loss": 3.3546,
      "step": 203170
    },
    {
      "epoch": 1.061000668709577,
      "grad_norm": 0.3696795105934143,
      "learning_rate": 0.00016212089010194113,
      "loss": 3.2955,
      "step": 203180
    },
    {
      "epoch": 1.0610752416542097,
      "grad_norm": 0.3796338438987732,
      "learning_rate": 0.00016211902577984593,
      "loss": 3.4086,
      "step": 203190
    },
    {
      "epoch": 1.0611498145988425,
      "grad_norm": 0.3718588054180145,
      "learning_rate": 0.00016211716145775074,
      "loss": 3.3788,
      "step": 203200
    },
    {
      "epoch": 1.0612243875434753,
      "grad_norm": 0.3939960300922394,
      "learning_rate": 0.00016211529713565555,
      "loss": 3.3512,
      "step": 203210
    },
    {
      "epoch": 1.0612989604881078,
      "grad_norm": 0.3660171329975128,
      "learning_rate": 0.00016211343281356035,
      "loss": 3.3559,
      "step": 203220
    },
    {
      "epoch": 1.0613735334327405,
      "grad_norm": 0.3746584951877594,
      "learning_rate": 0.00016211156849146513,
      "loss": 3.3354,
      "step": 203230
    },
    {
      "epoch": 1.0614481063773733,
      "grad_norm": 0.3942808508872986,
      "learning_rate": 0.00016210970416936994,
      "loss": 3.2943,
      "step": 203240
    },
    {
      "epoch": 1.061522679322006,
      "grad_norm": 0.4062563180923462,
      "learning_rate": 0.00016210783984727474,
      "loss": 3.3104,
      "step": 203250
    },
    {
      "epoch": 1.0615972522666388,
      "grad_norm": 0.36658522486686707,
      "learning_rate": 0.00016210597552517955,
      "loss": 3.3399,
      "step": 203260
    },
    {
      "epoch": 1.0616718252112716,
      "grad_norm": 0.35852041840553284,
      "learning_rate": 0.00016210411120308436,
      "loss": 3.3648,
      "step": 203270
    },
    {
      "epoch": 1.0617463981559043,
      "grad_norm": 0.3939770758152008,
      "learning_rate": 0.00016210224688098914,
      "loss": 3.3569,
      "step": 203280
    },
    {
      "epoch": 1.061820971100537,
      "grad_norm": 0.3623763620853424,
      "learning_rate": 0.00016210038255889394,
      "loss": 3.3552,
      "step": 203290
    },
    {
      "epoch": 1.0618955440451698,
      "grad_norm": 0.3650062084197998,
      "learning_rate": 0.00016209851823679872,
      "loss": 3.35,
      "step": 203300
    },
    {
      "epoch": 1.0619701169898026,
      "grad_norm": 0.3883923292160034,
      "learning_rate": 0.00016209665391470356,
      "loss": 3.3746,
      "step": 203310
    },
    {
      "epoch": 1.0620446899344351,
      "grad_norm": 0.4009672999382019,
      "learning_rate": 0.00016209478959260836,
      "loss": 3.3245,
      "step": 203320
    },
    {
      "epoch": 1.0621192628790679,
      "grad_norm": 0.3763144910335541,
      "learning_rate": 0.00016209292527051314,
      "loss": 3.3732,
      "step": 203330
    },
    {
      "epoch": 1.0621938358237006,
      "grad_norm": 0.3749147057533264,
      "learning_rate": 0.00016209106094841795,
      "loss": 3.3461,
      "step": 203340
    },
    {
      "epoch": 1.0622684087683334,
      "grad_norm": 0.3895159661769867,
      "learning_rate": 0.00016208919662632273,
      "loss": 3.3578,
      "step": 203350
    },
    {
      "epoch": 1.0623429817129662,
      "grad_norm": 0.35822540521621704,
      "learning_rate": 0.00016208733230422753,
      "loss": 3.3386,
      "step": 203360
    },
    {
      "epoch": 1.062417554657599,
      "grad_norm": 0.3516671657562256,
      "learning_rate": 0.00016208546798213237,
      "loss": 3.3387,
      "step": 203370
    },
    {
      "epoch": 1.0624921276022317,
      "grad_norm": 0.37448787689208984,
      "learning_rate": 0.00016208360366003715,
      "loss": 3.2789,
      "step": 203380
    },
    {
      "epoch": 1.0625667005468644,
      "grad_norm": 0.39328742027282715,
      "learning_rate": 0.00016208173933794195,
      "loss": 3.3764,
      "step": 203390
    },
    {
      "epoch": 1.0626412734914972,
      "grad_norm": 0.39877885580062866,
      "learning_rate": 0.00016207987501584673,
      "loss": 3.2743,
      "step": 203400
    },
    {
      "epoch": 1.06271584643613,
      "grad_norm": 0.36803868412971497,
      "learning_rate": 0.00016207801069375154,
      "loss": 3.3977,
      "step": 203410
    },
    {
      "epoch": 1.0627904193807627,
      "grad_norm": 0.40120479464530945,
      "learning_rate": 0.00016207614637165634,
      "loss": 3.2941,
      "step": 203420
    },
    {
      "epoch": 1.0628649923253952,
      "grad_norm": 0.36808741092681885,
      "learning_rate": 0.00016207428204956115,
      "loss": 3.3646,
      "step": 203430
    },
    {
      "epoch": 1.062939565270028,
      "grad_norm": 0.3666023313999176,
      "learning_rate": 0.00016207241772746596,
      "loss": 3.3345,
      "step": 203440
    },
    {
      "epoch": 1.0630141382146607,
      "grad_norm": 0.3578723967075348,
      "learning_rate": 0.00016207055340537074,
      "loss": 3.3329,
      "step": 203450
    },
    {
      "epoch": 1.0630887111592935,
      "grad_norm": 0.4294414222240448,
      "learning_rate": 0.00016206868908327554,
      "loss": 3.3683,
      "step": 203460
    },
    {
      "epoch": 1.0631632841039262,
      "grad_norm": 0.37677738070487976,
      "learning_rate": 0.00016206682476118035,
      "loss": 3.3553,
      "step": 203470
    },
    {
      "epoch": 1.063237857048559,
      "grad_norm": 0.3779554069042206,
      "learning_rate": 0.00016206496043908515,
      "loss": 3.3552,
      "step": 203480
    },
    {
      "epoch": 1.0633124299931918,
      "grad_norm": 0.3715132176876068,
      "learning_rate": 0.00016206309611698996,
      "loss": 3.3304,
      "step": 203490
    },
    {
      "epoch": 1.0633870029378245,
      "grad_norm": 0.4028547704219818,
      "learning_rate": 0.00016206123179489474,
      "loss": 3.3556,
      "step": 203500
    },
    {
      "epoch": 1.0634615758824573,
      "grad_norm": 0.3982775807380676,
      "learning_rate": 0.00016205936747279955,
      "loss": 3.4188,
      "step": 203510
    },
    {
      "epoch": 1.06353614882709,
      "grad_norm": 0.37515586614608765,
      "learning_rate": 0.00016205750315070435,
      "loss": 3.3298,
      "step": 203520
    },
    {
      "epoch": 1.0636107217717226,
      "grad_norm": 0.3861347436904907,
      "learning_rate": 0.00016205563882860916,
      "loss": 3.3894,
      "step": 203530
    },
    {
      "epoch": 1.0636852947163553,
      "grad_norm": 0.3765619993209839,
      "learning_rate": 0.00016205377450651396,
      "loss": 3.3169,
      "step": 203540
    },
    {
      "epoch": 1.063759867660988,
      "grad_norm": 0.38458383083343506,
      "learning_rate": 0.00016205191018441874,
      "loss": 3.3547,
      "step": 203550
    },
    {
      "epoch": 1.0638344406056208,
      "grad_norm": 0.35617566108703613,
      "learning_rate": 0.00016205004586232355,
      "loss": 3.3993,
      "step": 203560
    },
    {
      "epoch": 1.0639090135502536,
      "grad_norm": 0.3695771098136902,
      "learning_rate": 0.00016204818154022836,
      "loss": 3.3527,
      "step": 203570
    },
    {
      "epoch": 1.0639835864948863,
      "grad_norm": 0.3737170100212097,
      "learning_rate": 0.00016204631721813314,
      "loss": 3.2831,
      "step": 203580
    },
    {
      "epoch": 1.064058159439519,
      "grad_norm": 0.4020046293735504,
      "learning_rate": 0.00016204445289603797,
      "loss": 3.3453,
      "step": 203590
    },
    {
      "epoch": 1.0641327323841518,
      "grad_norm": 0.371156245470047,
      "learning_rate": 0.00016204258857394275,
      "loss": 3.324,
      "step": 203600
    },
    {
      "epoch": 1.0642073053287846,
      "grad_norm": 0.40634793043136597,
      "learning_rate": 0.00016204072425184755,
      "loss": 3.3778,
      "step": 203610
    },
    {
      "epoch": 1.0642818782734174,
      "grad_norm": 0.35414454340934753,
      "learning_rate": 0.00016203885992975233,
      "loss": 3.3967,
      "step": 203620
    },
    {
      "epoch": 1.0643564512180501,
      "grad_norm": 0.40053364634513855,
      "learning_rate": 0.00016203699560765714,
      "loss": 3.3277,
      "step": 203630
    },
    {
      "epoch": 1.0644310241626826,
      "grad_norm": 0.4005400240421295,
      "learning_rate": 0.00016203513128556195,
      "loss": 3.3419,
      "step": 203640
    },
    {
      "epoch": 1.0645055971073154,
      "grad_norm": 0.37940770387649536,
      "learning_rate": 0.00016203326696346675,
      "loss": 3.3251,
      "step": 203650
    },
    {
      "epoch": 1.0645801700519482,
      "grad_norm": 0.3953036069869995,
      "learning_rate": 0.00016203140264137156,
      "loss": 3.3262,
      "step": 203660
    },
    {
      "epoch": 1.064654742996581,
      "grad_norm": 0.37639209628105164,
      "learning_rate": 0.00016202953831927634,
      "loss": 3.3544,
      "step": 203670
    },
    {
      "epoch": 1.0647293159412137,
      "grad_norm": 0.39714616537094116,
      "learning_rate": 0.00016202767399718114,
      "loss": 3.3844,
      "step": 203680
    },
    {
      "epoch": 1.0648038888858464,
      "grad_norm": 0.3843604326248169,
      "learning_rate": 0.00016202580967508595,
      "loss": 3.2862,
      "step": 203690
    },
    {
      "epoch": 1.0648784618304792,
      "grad_norm": 0.3967556357383728,
      "learning_rate": 0.00016202394535299076,
      "loss": 3.3339,
      "step": 203700
    },
    {
      "epoch": 1.064953034775112,
      "grad_norm": 0.38337188959121704,
      "learning_rate": 0.00016202208103089556,
      "loss": 3.3641,
      "step": 203710
    },
    {
      "epoch": 1.0650276077197447,
      "grad_norm": 0.388139545917511,
      "learning_rate": 0.00016202021670880034,
      "loss": 3.3621,
      "step": 203720
    },
    {
      "epoch": 1.0651021806643775,
      "grad_norm": 0.386186420917511,
      "learning_rate": 0.00016201835238670515,
      "loss": 3.3156,
      "step": 203730
    },
    {
      "epoch": 1.06517675360901,
      "grad_norm": 0.4012799859046936,
      "learning_rate": 0.00016201648806460996,
      "loss": 3.3192,
      "step": 203740
    },
    {
      "epoch": 1.0652513265536427,
      "grad_norm": 0.4384860694408417,
      "learning_rate": 0.00016201462374251476,
      "loss": 3.3293,
      "step": 203750
    },
    {
      "epoch": 1.0653258994982755,
      "grad_norm": 0.3904724717140198,
      "learning_rate": 0.00016201275942041957,
      "loss": 3.3666,
      "step": 203760
    },
    {
      "epoch": 1.0654004724429083,
      "grad_norm": 0.40097519755363464,
      "learning_rate": 0.00016201089509832435,
      "loss": 3.3521,
      "step": 203770
    },
    {
      "epoch": 1.065475045387541,
      "grad_norm": 0.4048060476779938,
      "learning_rate": 0.00016200903077622915,
      "loss": 3.3468,
      "step": 203780
    },
    {
      "epoch": 1.0655496183321738,
      "grad_norm": 0.38273170590400696,
      "learning_rate": 0.00016200716645413396,
      "loss": 3.3751,
      "step": 203790
    },
    {
      "epoch": 1.0656241912768065,
      "grad_norm": 0.34334346652030945,
      "learning_rate": 0.00016200530213203874,
      "loss": 3.3723,
      "step": 203800
    },
    {
      "epoch": 1.0656987642214393,
      "grad_norm": 0.37259259819984436,
      "learning_rate": 0.00016200343780994355,
      "loss": 3.309,
      "step": 203810
    },
    {
      "epoch": 1.065773337166072,
      "grad_norm": 0.35014039278030396,
      "learning_rate": 0.00016200157348784835,
      "loss": 3.335,
      "step": 203820
    },
    {
      "epoch": 1.0658479101107048,
      "grad_norm": 0.38747039437294006,
      "learning_rate": 0.00016199970916575316,
      "loss": 3.2685,
      "step": 203830
    },
    {
      "epoch": 1.0659224830553375,
      "grad_norm": 0.36846670508384705,
      "learning_rate": 0.00016199784484365796,
      "loss": 3.3551,
      "step": 203840
    },
    {
      "epoch": 1.06599705599997,
      "grad_norm": 0.41043052077293396,
      "learning_rate": 0.00016199598052156274,
      "loss": 3.3756,
      "step": 203850
    },
    {
      "epoch": 1.0660716289446028,
      "grad_norm": 0.3673366606235504,
      "learning_rate": 0.00016199411619946755,
      "loss": 3.3666,
      "step": 203860
    },
    {
      "epoch": 1.0661462018892356,
      "grad_norm": 0.38661667704582214,
      "learning_rate": 0.00016199225187737236,
      "loss": 3.2618,
      "step": 203870
    },
    {
      "epoch": 1.0662207748338683,
      "grad_norm": 0.379199355840683,
      "learning_rate": 0.00016199038755527716,
      "loss": 3.3664,
      "step": 203880
    },
    {
      "epoch": 1.066295347778501,
      "grad_norm": 0.3705095648765564,
      "learning_rate": 0.00016198852323318197,
      "loss": 3.334,
      "step": 203890
    },
    {
      "epoch": 1.0663699207231339,
      "grad_norm": 0.3679712414741516,
      "learning_rate": 0.00016198665891108675,
      "loss": 3.3283,
      "step": 203900
    },
    {
      "epoch": 1.0664444936677666,
      "grad_norm": 0.39731690287590027,
      "learning_rate": 0.00016198479458899155,
      "loss": 3.3672,
      "step": 203910
    },
    {
      "epoch": 1.0665190666123994,
      "grad_norm": 0.37131381034851074,
      "learning_rate": 0.00016198293026689636,
      "loss": 3.3581,
      "step": 203920
    },
    {
      "epoch": 1.0665936395570321,
      "grad_norm": 0.40792393684387207,
      "learning_rate": 0.00016198106594480117,
      "loss": 3.3195,
      "step": 203930
    },
    {
      "epoch": 1.0666682125016649,
      "grad_norm": 0.3943363428115845,
      "learning_rate": 0.00016197920162270597,
      "loss": 3.3463,
      "step": 203940
    },
    {
      "epoch": 1.0667427854462974,
      "grad_norm": 0.3639625012874603,
      "learning_rate": 0.00016197733730061075,
      "loss": 3.3564,
      "step": 203950
    },
    {
      "epoch": 1.0668173583909302,
      "grad_norm": 0.38465961813926697,
      "learning_rate": 0.00016197547297851556,
      "loss": 3.3721,
      "step": 203960
    },
    {
      "epoch": 1.066891931335563,
      "grad_norm": 0.3610456585884094,
      "learning_rate": 0.00016197360865642034,
      "loss": 3.316,
      "step": 203970
    },
    {
      "epoch": 1.0669665042801957,
      "grad_norm": 0.3666291832923889,
      "learning_rate": 0.00016197174433432517,
      "loss": 3.3698,
      "step": 203980
    },
    {
      "epoch": 1.0670410772248284,
      "grad_norm": 0.3700385093688965,
      "learning_rate": 0.00016196988001222998,
      "loss": 3.3053,
      "step": 203990
    },
    {
      "epoch": 1.0671156501694612,
      "grad_norm": 0.37829598784446716,
      "learning_rate": 0.00016196801569013476,
      "loss": 3.3323,
      "step": 204000
    },
    {
      "epoch": 1.067190223114094,
      "grad_norm": 0.3694067895412445,
      "learning_rate": 0.00016196615136803956,
      "loss": 3.352,
      "step": 204010
    },
    {
      "epoch": 1.0672647960587267,
      "grad_norm": 0.39204642176628113,
      "learning_rate": 0.00016196428704594434,
      "loss": 3.4033,
      "step": 204020
    },
    {
      "epoch": 1.0673393690033595,
      "grad_norm": 0.3666112422943115,
      "learning_rate": 0.00016196242272384915,
      "loss": 3.3268,
      "step": 204030
    },
    {
      "epoch": 1.0674139419479922,
      "grad_norm": 0.40097030997276306,
      "learning_rate": 0.00016196055840175398,
      "loss": 3.407,
      "step": 204040
    },
    {
      "epoch": 1.067488514892625,
      "grad_norm": 0.38689568638801575,
      "learning_rate": 0.00016195869407965876,
      "loss": 3.3748,
      "step": 204050
    },
    {
      "epoch": 1.0675630878372575,
      "grad_norm": 0.3734901547431946,
      "learning_rate": 0.00016195682975756357,
      "loss": 3.3449,
      "step": 204060
    },
    {
      "epoch": 1.0676376607818903,
      "grad_norm": 0.3833414912223816,
      "learning_rate": 0.00016195496543546835,
      "loss": 3.3853,
      "step": 204070
    },
    {
      "epoch": 1.067712233726523,
      "grad_norm": 0.36522942781448364,
      "learning_rate": 0.00016195310111337315,
      "loss": 3.312,
      "step": 204080
    },
    {
      "epoch": 1.0677868066711558,
      "grad_norm": 0.36241617798805237,
      "learning_rate": 0.00016195123679127796,
      "loss": 3.3453,
      "step": 204090
    },
    {
      "epoch": 1.0678613796157885,
      "grad_norm": 0.39231544733047485,
      "learning_rate": 0.00016194937246918277,
      "loss": 3.3132,
      "step": 204100
    },
    {
      "epoch": 1.0679359525604213,
      "grad_norm": 0.3872148394584656,
      "learning_rate": 0.00016194750814708757,
      "loss": 3.3593,
      "step": 204110
    },
    {
      "epoch": 1.068010525505054,
      "grad_norm": 0.359695702791214,
      "learning_rate": 0.00016194564382499235,
      "loss": 3.4009,
      "step": 204120
    },
    {
      "epoch": 1.0680850984496868,
      "grad_norm": 0.3595510423183441,
      "learning_rate": 0.00016194377950289716,
      "loss": 3.3647,
      "step": 204130
    },
    {
      "epoch": 1.0681596713943196,
      "grad_norm": 0.3803676664829254,
      "learning_rate": 0.00016194191518080196,
      "loss": 3.3608,
      "step": 204140
    },
    {
      "epoch": 1.0682342443389523,
      "grad_norm": 0.35894113779067993,
      "learning_rate": 0.00016194005085870677,
      "loss": 3.3213,
      "step": 204150
    },
    {
      "epoch": 1.0683088172835848,
      "grad_norm": 0.3786028325557709,
      "learning_rate": 0.00016193818653661158,
      "loss": 3.3808,
      "step": 204160
    },
    {
      "epoch": 1.0683833902282176,
      "grad_norm": 0.363665908575058,
      "learning_rate": 0.00016193632221451636,
      "loss": 3.3414,
      "step": 204170
    },
    {
      "epoch": 1.0684579631728504,
      "grad_norm": 0.3612699806690216,
      "learning_rate": 0.00016193445789242116,
      "loss": 3.3372,
      "step": 204180
    },
    {
      "epoch": 1.0685325361174831,
      "grad_norm": 0.3608553409576416,
      "learning_rate": 0.00016193259357032597,
      "loss": 3.3515,
      "step": 204190
    },
    {
      "epoch": 1.0686071090621159,
      "grad_norm": 0.3908579647541046,
      "learning_rate": 0.00016193072924823077,
      "loss": 3.308,
      "step": 204200
    },
    {
      "epoch": 1.0686816820067486,
      "grad_norm": 0.39875316619873047,
      "learning_rate": 0.00016192886492613558,
      "loss": 3.3828,
      "step": 204210
    },
    {
      "epoch": 1.0687562549513814,
      "grad_norm": 0.4106535017490387,
      "learning_rate": 0.00016192700060404036,
      "loss": 3.3897,
      "step": 204220
    },
    {
      "epoch": 1.0688308278960141,
      "grad_norm": 0.35443201661109924,
      "learning_rate": 0.00016192513628194517,
      "loss": 3.3152,
      "step": 204230
    },
    {
      "epoch": 1.068905400840647,
      "grad_norm": 0.389021635055542,
      "learning_rate": 0.00016192327195984997,
      "loss": 3.4031,
      "step": 204240
    },
    {
      "epoch": 1.0689799737852796,
      "grad_norm": 0.3713783919811249,
      "learning_rate": 0.00016192140763775475,
      "loss": 3.3875,
      "step": 204250
    },
    {
      "epoch": 1.0690545467299124,
      "grad_norm": 0.374683141708374,
      "learning_rate": 0.00016191954331565959,
      "loss": 3.3891,
      "step": 204260
    },
    {
      "epoch": 1.069129119674545,
      "grad_norm": 0.3985580801963806,
      "learning_rate": 0.00016191767899356436,
      "loss": 3.3216,
      "step": 204270
    },
    {
      "epoch": 1.0692036926191777,
      "grad_norm": 0.3376657962799072,
      "learning_rate": 0.00016191581467146917,
      "loss": 3.3441,
      "step": 204280
    },
    {
      "epoch": 1.0692782655638104,
      "grad_norm": 0.402781218290329,
      "learning_rate": 0.00016191395034937398,
      "loss": 3.3472,
      "step": 204290
    },
    {
      "epoch": 1.0693528385084432,
      "grad_norm": 0.38909637928009033,
      "learning_rate": 0.00016191208602727876,
      "loss": 3.3253,
      "step": 204300
    },
    {
      "epoch": 1.069427411453076,
      "grad_norm": 0.40603014826774597,
      "learning_rate": 0.00016191022170518356,
      "loss": 3.3496,
      "step": 204310
    },
    {
      "epoch": 1.0695019843977087,
      "grad_norm": 0.3727039396762848,
      "learning_rate": 0.00016190835738308837,
      "loss": 3.3206,
      "step": 204320
    },
    {
      "epoch": 1.0695765573423415,
      "grad_norm": 0.3825075030326843,
      "learning_rate": 0.00016190649306099318,
      "loss": 3.3304,
      "step": 204330
    },
    {
      "epoch": 1.0696511302869742,
      "grad_norm": 0.3429698050022125,
      "learning_rate": 0.00016190462873889798,
      "loss": 3.4323,
      "step": 204340
    },
    {
      "epoch": 1.069725703231607,
      "grad_norm": 0.37041664123535156,
      "learning_rate": 0.00016190276441680276,
      "loss": 3.3531,
      "step": 204350
    },
    {
      "epoch": 1.0698002761762397,
      "grad_norm": 0.3542230725288391,
      "learning_rate": 0.00016190090009470757,
      "loss": 3.4025,
      "step": 204360
    },
    {
      "epoch": 1.0698748491208723,
      "grad_norm": 0.37691453099250793,
      "learning_rate": 0.00016189903577261237,
      "loss": 3.3642,
      "step": 204370
    },
    {
      "epoch": 1.069949422065505,
      "grad_norm": 0.3863609731197357,
      "learning_rate": 0.00016189717145051718,
      "loss": 3.3284,
      "step": 204380
    },
    {
      "epoch": 1.0700239950101378,
      "grad_norm": 0.39832738041877747,
      "learning_rate": 0.00016189530712842199,
      "loss": 3.3636,
      "step": 204390
    },
    {
      "epoch": 1.0700985679547705,
      "grad_norm": 0.4092290699481964,
      "learning_rate": 0.00016189344280632677,
      "loss": 3.2749,
      "step": 204400
    },
    {
      "epoch": 1.0701731408994033,
      "grad_norm": 0.359984427690506,
      "learning_rate": 0.00016189157848423157,
      "loss": 3.4426,
      "step": 204410
    },
    {
      "epoch": 1.070247713844036,
      "grad_norm": 0.36110320687294006,
      "learning_rate": 0.00016188971416213635,
      "loss": 3.352,
      "step": 204420
    },
    {
      "epoch": 1.0703222867886688,
      "grad_norm": 0.40354007482528687,
      "learning_rate": 0.00016188784984004118,
      "loss": 3.3968,
      "step": 204430
    },
    {
      "epoch": 1.0703968597333016,
      "grad_norm": 0.38228660821914673,
      "learning_rate": 0.000161885985517946,
      "loss": 3.3703,
      "step": 204440
    },
    {
      "epoch": 1.0704714326779343,
      "grad_norm": 0.3723279535770416,
      "learning_rate": 0.00016188412119585077,
      "loss": 3.3743,
      "step": 204450
    },
    {
      "epoch": 1.070546005622567,
      "grad_norm": 0.38322874903678894,
      "learning_rate": 0.00016188225687375558,
      "loss": 3.3339,
      "step": 204460
    },
    {
      "epoch": 1.0706205785671998,
      "grad_norm": 0.3708924353122711,
      "learning_rate": 0.00016188039255166036,
      "loss": 3.3426,
      "step": 204470
    },
    {
      "epoch": 1.0706951515118324,
      "grad_norm": 0.40042051672935486,
      "learning_rate": 0.00016187852822956516,
      "loss": 3.3722,
      "step": 204480
    },
    {
      "epoch": 1.0707697244564651,
      "grad_norm": 0.38509994745254517,
      "learning_rate": 0.00016187666390747,
      "loss": 3.3243,
      "step": 204490
    },
    {
      "epoch": 1.0708442974010979,
      "grad_norm": 0.3905913829803467,
      "learning_rate": 0.00016187479958537477,
      "loss": 3.3499,
      "step": 204500
    },
    {
      "epoch": 1.0709188703457306,
      "grad_norm": 0.3659381568431854,
      "learning_rate": 0.00016187293526327958,
      "loss": 3.3648,
      "step": 204510
    },
    {
      "epoch": 1.0709934432903634,
      "grad_norm": 0.35121169686317444,
      "learning_rate": 0.00016187107094118436,
      "loss": 3.2937,
      "step": 204520
    },
    {
      "epoch": 1.0710680162349961,
      "grad_norm": 0.3839789032936096,
      "learning_rate": 0.00016186920661908917,
      "loss": 3.3945,
      "step": 204530
    },
    {
      "epoch": 1.071142589179629,
      "grad_norm": 0.4009931683540344,
      "learning_rate": 0.00016186734229699397,
      "loss": 3.3439,
      "step": 204540
    },
    {
      "epoch": 1.0712171621242617,
      "grad_norm": 0.3659246861934662,
      "learning_rate": 0.00016186547797489878,
      "loss": 3.3677,
      "step": 204550
    },
    {
      "epoch": 1.0712917350688944,
      "grad_norm": 0.4173632264137268,
      "learning_rate": 0.00016186361365280358,
      "loss": 3.3204,
      "step": 204560
    },
    {
      "epoch": 1.0713663080135272,
      "grad_norm": 0.3815232515335083,
      "learning_rate": 0.00016186174933070836,
      "loss": 3.3478,
      "step": 204570
    },
    {
      "epoch": 1.0714408809581597,
      "grad_norm": 0.35288646817207336,
      "learning_rate": 0.00016185988500861317,
      "loss": 3.3363,
      "step": 204580
    },
    {
      "epoch": 1.0715154539027925,
      "grad_norm": 0.3805716633796692,
      "learning_rate": 0.00016185802068651798,
      "loss": 3.3465,
      "step": 204590
    },
    {
      "epoch": 1.0715900268474252,
      "grad_norm": 0.40361061692237854,
      "learning_rate": 0.00016185615636442278,
      "loss": 3.2993,
      "step": 204600
    },
    {
      "epoch": 1.071664599792058,
      "grad_norm": 0.39892005920410156,
      "learning_rate": 0.0001618542920423276,
      "loss": 3.3371,
      "step": 204610
    },
    {
      "epoch": 1.0717391727366907,
      "grad_norm": 0.38648587465286255,
      "learning_rate": 0.00016185242772023237,
      "loss": 3.3606,
      "step": 204620
    },
    {
      "epoch": 1.0718137456813235,
      "grad_norm": 0.36519762873649597,
      "learning_rate": 0.00016185056339813717,
      "loss": 3.3457,
      "step": 204630
    },
    {
      "epoch": 1.0718883186259562,
      "grad_norm": 0.4442090392112732,
      "learning_rate": 0.00016184869907604198,
      "loss": 3.332,
      "step": 204640
    },
    {
      "epoch": 1.071962891570589,
      "grad_norm": 0.36914291977882385,
      "learning_rate": 0.0001618468347539468,
      "loss": 3.3453,
      "step": 204650
    },
    {
      "epoch": 1.0720374645152217,
      "grad_norm": 0.3850769102573395,
      "learning_rate": 0.0001618449704318516,
      "loss": 3.3426,
      "step": 204660
    },
    {
      "epoch": 1.0721120374598545,
      "grad_norm": 0.3816581070423126,
      "learning_rate": 0.00016184310610975637,
      "loss": 3.3475,
      "step": 204670
    },
    {
      "epoch": 1.0721866104044873,
      "grad_norm": 0.37920254468917847,
      "learning_rate": 0.00016184124178766118,
      "loss": 3.2805,
      "step": 204680
    },
    {
      "epoch": 1.0722611833491198,
      "grad_norm": 0.3855297565460205,
      "learning_rate": 0.00016183937746556599,
      "loss": 3.3706,
      "step": 204690
    },
    {
      "epoch": 1.0723357562937526,
      "grad_norm": 0.36667829751968384,
      "learning_rate": 0.00016183751314347076,
      "loss": 3.3582,
      "step": 204700
    },
    {
      "epoch": 1.0724103292383853,
      "grad_norm": 0.4115901291370392,
      "learning_rate": 0.0001618356488213756,
      "loss": 3.3132,
      "step": 204710
    },
    {
      "epoch": 1.072484902183018,
      "grad_norm": 0.3747204840183258,
      "learning_rate": 0.00016183378449928038,
      "loss": 3.3482,
      "step": 204720
    },
    {
      "epoch": 1.0725594751276508,
      "grad_norm": 0.383635938167572,
      "learning_rate": 0.00016183192017718518,
      "loss": 3.3845,
      "step": 204730
    },
    {
      "epoch": 1.0726340480722836,
      "grad_norm": 0.383192241191864,
      "learning_rate": 0.00016183005585509,
      "loss": 3.3822,
      "step": 204740
    },
    {
      "epoch": 1.0727086210169163,
      "grad_norm": 0.39577916264533997,
      "learning_rate": 0.00016182819153299477,
      "loss": 3.3491,
      "step": 204750
    },
    {
      "epoch": 1.072783193961549,
      "grad_norm": 0.36465370655059814,
      "learning_rate": 0.00016182632721089958,
      "loss": 3.3531,
      "step": 204760
    },
    {
      "epoch": 1.0728577669061818,
      "grad_norm": 0.3724956512451172,
      "learning_rate": 0.00016182446288880438,
      "loss": 3.4117,
      "step": 204770
    },
    {
      "epoch": 1.0729323398508144,
      "grad_norm": 0.4169958829879761,
      "learning_rate": 0.0001618225985667092,
      "loss": 3.3741,
      "step": 204780
    },
    {
      "epoch": 1.0730069127954471,
      "grad_norm": 0.3601459562778473,
      "learning_rate": 0.000161820734244614,
      "loss": 3.3783,
      "step": 204790
    },
    {
      "epoch": 1.0730814857400799,
      "grad_norm": 0.3897981643676758,
      "learning_rate": 0.00016181886992251877,
      "loss": 3.3459,
      "step": 204800
    },
    {
      "epoch": 1.0731560586847126,
      "grad_norm": 0.38747674226760864,
      "learning_rate": 0.00016181700560042358,
      "loss": 3.3308,
      "step": 204810
    },
    {
      "epoch": 1.0732306316293454,
      "grad_norm": 0.3991871774196625,
      "learning_rate": 0.00016181514127832839,
      "loss": 3.3386,
      "step": 204820
    },
    {
      "epoch": 1.0733052045739782,
      "grad_norm": 0.37779611349105835,
      "learning_rate": 0.0001618132769562332,
      "loss": 3.3382,
      "step": 204830
    },
    {
      "epoch": 1.073379777518611,
      "grad_norm": 0.37594738602638245,
      "learning_rate": 0.000161811412634138,
      "loss": 3.4053,
      "step": 204840
    },
    {
      "epoch": 1.0734543504632437,
      "grad_norm": 0.3833308517932892,
      "learning_rate": 0.00016180954831204278,
      "loss": 3.3405,
      "step": 204850
    },
    {
      "epoch": 1.0735289234078764,
      "grad_norm": 0.3885756731033325,
      "learning_rate": 0.00016180768398994758,
      "loss": 3.3787,
      "step": 204860
    },
    {
      "epoch": 1.0736034963525092,
      "grad_norm": 0.3827909827232361,
      "learning_rate": 0.00016180581966785236,
      "loss": 3.3411,
      "step": 204870
    },
    {
      "epoch": 1.073678069297142,
      "grad_norm": 0.39143678545951843,
      "learning_rate": 0.0001618039553457572,
      "loss": 3.285,
      "step": 204880
    },
    {
      "epoch": 1.0737526422417747,
      "grad_norm": 0.3924223482608795,
      "learning_rate": 0.000161802091023662,
      "loss": 3.2365,
      "step": 204890
    },
    {
      "epoch": 1.0738272151864072,
      "grad_norm": 0.409170925617218,
      "learning_rate": 0.00016180022670156678,
      "loss": 3.3321,
      "step": 204900
    },
    {
      "epoch": 1.07390178813104,
      "grad_norm": 0.40525245666503906,
      "learning_rate": 0.0001617983623794716,
      "loss": 3.3101,
      "step": 204910
    },
    {
      "epoch": 1.0739763610756727,
      "grad_norm": 0.38700491189956665,
      "learning_rate": 0.00016179649805737637,
      "loss": 3.3343,
      "step": 204920
    },
    {
      "epoch": 1.0740509340203055,
      "grad_norm": 0.3989761471748352,
      "learning_rate": 0.00016179463373528117,
      "loss": 3.3877,
      "step": 204930
    },
    {
      "epoch": 1.0741255069649382,
      "grad_norm": 0.3963501453399658,
      "learning_rate": 0.000161792769413186,
      "loss": 3.3517,
      "step": 204940
    },
    {
      "epoch": 1.074200079909571,
      "grad_norm": 0.40309906005859375,
      "learning_rate": 0.0001617909050910908,
      "loss": 3.3629,
      "step": 204950
    },
    {
      "epoch": 1.0742746528542038,
      "grad_norm": 0.3596237301826477,
      "learning_rate": 0.0001617890407689956,
      "loss": 3.3631,
      "step": 204960
    },
    {
      "epoch": 1.0743492257988365,
      "grad_norm": 0.385844349861145,
      "learning_rate": 0.00016178717644690037,
      "loss": 3.3483,
      "step": 204970
    },
    {
      "epoch": 1.0744237987434693,
      "grad_norm": 0.36905401945114136,
      "learning_rate": 0.00016178531212480518,
      "loss": 3.3598,
      "step": 204980
    },
    {
      "epoch": 1.0744983716881018,
      "grad_norm": 0.39794448018074036,
      "learning_rate": 0.00016178344780270999,
      "loss": 3.3583,
      "step": 204990
    },
    {
      "epoch": 1.0745729446327346,
      "grad_norm": 0.39239302277565,
      "learning_rate": 0.0001617815834806148,
      "loss": 3.3653,
      "step": 205000
    },
    {
      "epoch": 1.0746475175773673,
      "grad_norm": 0.35660210251808167,
      "learning_rate": 0.0001617797191585196,
      "loss": 3.3022,
      "step": 205010
    },
    {
      "epoch": 1.074722090522,
      "grad_norm": 0.39115485548973083,
      "learning_rate": 0.00016177785483642438,
      "loss": 3.3695,
      "step": 205020
    },
    {
      "epoch": 1.0747966634666328,
      "grad_norm": 0.3718762993812561,
      "learning_rate": 0.00016177599051432918,
      "loss": 3.3476,
      "step": 205030
    },
    {
      "epoch": 1.0748712364112656,
      "grad_norm": 0.4064093232154846,
      "learning_rate": 0.000161774126192234,
      "loss": 3.3303,
      "step": 205040
    },
    {
      "epoch": 1.0749458093558983,
      "grad_norm": 0.3745006322860718,
      "learning_rate": 0.0001617722618701388,
      "loss": 3.351,
      "step": 205050
    },
    {
      "epoch": 1.075020382300531,
      "grad_norm": 0.38782232999801636,
      "learning_rate": 0.0001617703975480436,
      "loss": 3.3967,
      "step": 205060
    },
    {
      "epoch": 1.0750949552451639,
      "grad_norm": 0.36218807101249695,
      "learning_rate": 0.00016176853322594838,
      "loss": 3.3181,
      "step": 205070
    },
    {
      "epoch": 1.0751695281897966,
      "grad_norm": 0.37897688150405884,
      "learning_rate": 0.0001617666689038532,
      "loss": 3.3218,
      "step": 205080
    },
    {
      "epoch": 1.0752441011344294,
      "grad_norm": 0.3890945017337799,
      "learning_rate": 0.00016176480458175797,
      "loss": 3.3828,
      "step": 205090
    },
    {
      "epoch": 1.0753186740790621,
      "grad_norm": 0.3805594742298126,
      "learning_rate": 0.0001617629402596628,
      "loss": 3.29,
      "step": 205100
    },
    {
      "epoch": 1.0753932470236947,
      "grad_norm": 0.39020970463752747,
      "learning_rate": 0.0001617610759375676,
      "loss": 3.3344,
      "step": 205110
    },
    {
      "epoch": 1.0754678199683274,
      "grad_norm": 0.38299041986465454,
      "learning_rate": 0.00016175921161547239,
      "loss": 3.3503,
      "step": 205120
    },
    {
      "epoch": 1.0755423929129602,
      "grad_norm": 0.3759520649909973,
      "learning_rate": 0.0001617573472933772,
      "loss": 3.3435,
      "step": 205130
    },
    {
      "epoch": 1.075616965857593,
      "grad_norm": 0.4258597791194916,
      "learning_rate": 0.00016175548297128197,
      "loss": 3.3627,
      "step": 205140
    },
    {
      "epoch": 1.0756915388022257,
      "grad_norm": 0.3613845109939575,
      "learning_rate": 0.00016175361864918678,
      "loss": 3.376,
      "step": 205150
    },
    {
      "epoch": 1.0757661117468584,
      "grad_norm": 0.39238613843917847,
      "learning_rate": 0.0001617517543270916,
      "loss": 3.3059,
      "step": 205160
    },
    {
      "epoch": 1.0758406846914912,
      "grad_norm": 0.367120623588562,
      "learning_rate": 0.0001617498900049964,
      "loss": 3.3623,
      "step": 205170
    },
    {
      "epoch": 1.075915257636124,
      "grad_norm": 0.3824944496154785,
      "learning_rate": 0.0001617480256829012,
      "loss": 3.3081,
      "step": 205180
    },
    {
      "epoch": 1.0759898305807567,
      "grad_norm": 0.39590927958488464,
      "learning_rate": 0.00016174616136080598,
      "loss": 3.3859,
      "step": 205190
    },
    {
      "epoch": 1.0760644035253892,
      "grad_norm": 0.3934600055217743,
      "learning_rate": 0.00016174429703871078,
      "loss": 3.3711,
      "step": 205200
    },
    {
      "epoch": 1.076138976470022,
      "grad_norm": 0.38659900426864624,
      "learning_rate": 0.0001617424327166156,
      "loss": 3.2621,
      "step": 205210
    },
    {
      "epoch": 1.0762135494146547,
      "grad_norm": 0.4003968834877014,
      "learning_rate": 0.0001617405683945204,
      "loss": 3.3258,
      "step": 205220
    },
    {
      "epoch": 1.0762881223592875,
      "grad_norm": 0.37008562684059143,
      "learning_rate": 0.0001617387040724252,
      "loss": 3.3221,
      "step": 205230
    },
    {
      "epoch": 1.0763626953039203,
      "grad_norm": 0.4020996391773224,
      "learning_rate": 0.00016173683975032998,
      "loss": 3.3416,
      "step": 205240
    },
    {
      "epoch": 1.076437268248553,
      "grad_norm": 0.35605868697166443,
      "learning_rate": 0.0001617349754282348,
      "loss": 3.4007,
      "step": 205250
    },
    {
      "epoch": 1.0765118411931858,
      "grad_norm": 0.3738047182559967,
      "learning_rate": 0.0001617331111061396,
      "loss": 3.315,
      "step": 205260
    },
    {
      "epoch": 1.0765864141378185,
      "grad_norm": 0.4078294038772583,
      "learning_rate": 0.0001617312467840444,
      "loss": 3.3311,
      "step": 205270
    },
    {
      "epoch": 1.0766609870824513,
      "grad_norm": 0.383953720331192,
      "learning_rate": 0.0001617293824619492,
      "loss": 3.3137,
      "step": 205280
    },
    {
      "epoch": 1.076735560027084,
      "grad_norm": 0.36954182386398315,
      "learning_rate": 0.00016172751813985398,
      "loss": 3.3289,
      "step": 205290
    },
    {
      "epoch": 1.0768101329717168,
      "grad_norm": 0.3839687407016754,
      "learning_rate": 0.0001617256538177588,
      "loss": 3.377,
      "step": 205300
    },
    {
      "epoch": 1.0768847059163493,
      "grad_norm": 0.38434669375419617,
      "learning_rate": 0.0001617237894956636,
      "loss": 3.3866,
      "step": 205310
    },
    {
      "epoch": 1.076959278860982,
      "grad_norm": 0.377873957157135,
      "learning_rate": 0.00016172192517356838,
      "loss": 3.353,
      "step": 205320
    },
    {
      "epoch": 1.0770338518056148,
      "grad_norm": 0.39522984623908997,
      "learning_rate": 0.0001617200608514732,
      "loss": 3.3309,
      "step": 205330
    },
    {
      "epoch": 1.0771084247502476,
      "grad_norm": 0.36577391624450684,
      "learning_rate": 0.000161718196529378,
      "loss": 3.3545,
      "step": 205340
    },
    {
      "epoch": 1.0771829976948804,
      "grad_norm": 0.36083605885505676,
      "learning_rate": 0.0001617163322072828,
      "loss": 3.3358,
      "step": 205350
    },
    {
      "epoch": 1.077257570639513,
      "grad_norm": 0.3734394907951355,
      "learning_rate": 0.0001617144678851876,
      "loss": 3.3235,
      "step": 205360
    },
    {
      "epoch": 1.0773321435841459,
      "grad_norm": 0.3955950140953064,
      "learning_rate": 0.00016171260356309238,
      "loss": 3.3107,
      "step": 205370
    },
    {
      "epoch": 1.0774067165287786,
      "grad_norm": 0.35241836309432983,
      "learning_rate": 0.0001617107392409972,
      "loss": 3.3212,
      "step": 205380
    },
    {
      "epoch": 1.0774812894734114,
      "grad_norm": 0.35450050234794617,
      "learning_rate": 0.000161708874918902,
      "loss": 3.3722,
      "step": 205390
    },
    {
      "epoch": 1.0775558624180441,
      "grad_norm": 0.36742812395095825,
      "learning_rate": 0.0001617070105968068,
      "loss": 3.3434,
      "step": 205400
    },
    {
      "epoch": 1.0776304353626767,
      "grad_norm": 0.3637603521347046,
      "learning_rate": 0.0001617051462747116,
      "loss": 3.3521,
      "step": 205410
    },
    {
      "epoch": 1.0777050083073094,
      "grad_norm": 0.36236217617988586,
      "learning_rate": 0.00016170328195261639,
      "loss": 3.2644,
      "step": 205420
    },
    {
      "epoch": 1.0777795812519422,
      "grad_norm": 0.3818565905094147,
      "learning_rate": 0.0001617014176305212,
      "loss": 3.2928,
      "step": 205430
    },
    {
      "epoch": 1.077854154196575,
      "grad_norm": 0.3728593587875366,
      "learning_rate": 0.000161699553308426,
      "loss": 3.3779,
      "step": 205440
    },
    {
      "epoch": 1.0779287271412077,
      "grad_norm": 0.3845779001712799,
      "learning_rate": 0.0001616976889863308,
      "loss": 3.3532,
      "step": 205450
    },
    {
      "epoch": 1.0780033000858404,
      "grad_norm": 0.36724454164505005,
      "learning_rate": 0.0001616958246642356,
      "loss": 3.3354,
      "step": 205460
    },
    {
      "epoch": 1.0780778730304732,
      "grad_norm": 0.3687678575515747,
      "learning_rate": 0.0001616939603421404,
      "loss": 3.3642,
      "step": 205470
    },
    {
      "epoch": 1.078152445975106,
      "grad_norm": 0.38188421726226807,
      "learning_rate": 0.0001616920960200452,
      "loss": 3.3395,
      "step": 205480
    },
    {
      "epoch": 1.0782270189197387,
      "grad_norm": 0.36995241045951843,
      "learning_rate": 0.00016169023169795,
      "loss": 3.3154,
      "step": 205490
    },
    {
      "epoch": 1.0783015918643715,
      "grad_norm": 0.40561607480049133,
      "learning_rate": 0.0001616883673758548,
      "loss": 3.3254,
      "step": 205500
    },
    {
      "epoch": 1.0783761648090042,
      "grad_norm": 0.3925474286079407,
      "learning_rate": 0.00016168650305375962,
      "loss": 3.3075,
      "step": 205510
    },
    {
      "epoch": 1.0784507377536368,
      "grad_norm": 0.3780435025691986,
      "learning_rate": 0.0001616846387316644,
      "loss": 3.3649,
      "step": 205520
    },
    {
      "epoch": 1.0785253106982695,
      "grad_norm": 0.3876588046550751,
      "learning_rate": 0.0001616827744095692,
      "loss": 3.366,
      "step": 205530
    },
    {
      "epoch": 1.0785998836429023,
      "grad_norm": 0.3842916190624237,
      "learning_rate": 0.00016168091008747398,
      "loss": 3.3295,
      "step": 205540
    },
    {
      "epoch": 1.078674456587535,
      "grad_norm": 0.3638651967048645,
      "learning_rate": 0.0001616790457653788,
      "loss": 3.3231,
      "step": 205550
    },
    {
      "epoch": 1.0787490295321678,
      "grad_norm": 0.3684498965740204,
      "learning_rate": 0.00016167718144328362,
      "loss": 3.2814,
      "step": 205560
    },
    {
      "epoch": 1.0788236024768005,
      "grad_norm": 0.3730011284351349,
      "learning_rate": 0.0001616753171211884,
      "loss": 3.3373,
      "step": 205570
    },
    {
      "epoch": 1.0788981754214333,
      "grad_norm": 0.33987465500831604,
      "learning_rate": 0.0001616734527990932,
      "loss": 3.3292,
      "step": 205580
    },
    {
      "epoch": 1.078972748366066,
      "grad_norm": 0.40995246171951294,
      "learning_rate": 0.00016167158847699798,
      "loss": 3.2556,
      "step": 205590
    },
    {
      "epoch": 1.0790473213106988,
      "grad_norm": 0.3693653345108032,
      "learning_rate": 0.0001616697241549028,
      "loss": 3.4001,
      "step": 205600
    },
    {
      "epoch": 1.0791218942553316,
      "grad_norm": 0.3863571286201477,
      "learning_rate": 0.00016166785983280762,
      "loss": 3.3416,
      "step": 205610
    },
    {
      "epoch": 1.079196467199964,
      "grad_norm": 0.3494844138622284,
      "learning_rate": 0.0001616659955107124,
      "loss": 3.267,
      "step": 205620
    },
    {
      "epoch": 1.0792710401445968,
      "grad_norm": 0.39127862453460693,
      "learning_rate": 0.0001616641311886172,
      "loss": 3.3705,
      "step": 205630
    },
    {
      "epoch": 1.0793456130892296,
      "grad_norm": 0.34736260771751404,
      "learning_rate": 0.000161662266866522,
      "loss": 3.3324,
      "step": 205640
    },
    {
      "epoch": 1.0794201860338624,
      "grad_norm": 0.3999980390071869,
      "learning_rate": 0.0001616604025444268,
      "loss": 3.3669,
      "step": 205650
    },
    {
      "epoch": 1.0794947589784951,
      "grad_norm": 0.38392001390457153,
      "learning_rate": 0.0001616585382223316,
      "loss": 3.267,
      "step": 205660
    },
    {
      "epoch": 1.0795693319231279,
      "grad_norm": 0.3801661729812622,
      "learning_rate": 0.0001616566739002364,
      "loss": 3.3825,
      "step": 205670
    },
    {
      "epoch": 1.0796439048677606,
      "grad_norm": 0.3786082863807678,
      "learning_rate": 0.00016165480957814121,
      "loss": 3.3426,
      "step": 205680
    },
    {
      "epoch": 1.0797184778123934,
      "grad_norm": 0.3698558807373047,
      "learning_rate": 0.000161652945256046,
      "loss": 3.3765,
      "step": 205690
    },
    {
      "epoch": 1.0797930507570261,
      "grad_norm": 0.3812007009983063,
      "learning_rate": 0.0001616510809339508,
      "loss": 3.2997,
      "step": 205700
    },
    {
      "epoch": 1.079867623701659,
      "grad_norm": 0.3862016499042511,
      "learning_rate": 0.0001616492166118556,
      "loss": 3.3398,
      "step": 205710
    },
    {
      "epoch": 1.0799421966462917,
      "grad_norm": 0.38971707224845886,
      "learning_rate": 0.0001616473522897604,
      "loss": 3.387,
      "step": 205720
    },
    {
      "epoch": 1.0800167695909242,
      "grad_norm": 0.3794471323490143,
      "learning_rate": 0.00016164548796766522,
      "loss": 3.3777,
      "step": 205730
    },
    {
      "epoch": 1.080091342535557,
      "grad_norm": 0.37032926082611084,
      "learning_rate": 0.00016164362364557,
      "loss": 3.3518,
      "step": 205740
    },
    {
      "epoch": 1.0801659154801897,
      "grad_norm": 0.3930206894874573,
      "learning_rate": 0.0001616417593234748,
      "loss": 3.3116,
      "step": 205750
    },
    {
      "epoch": 1.0802404884248225,
      "grad_norm": 0.38849788904190063,
      "learning_rate": 0.0001616398950013796,
      "loss": 3.2982,
      "step": 205760
    },
    {
      "epoch": 1.0803150613694552,
      "grad_norm": 0.37483662366867065,
      "learning_rate": 0.0001616380306792844,
      "loss": 3.2939,
      "step": 205770
    },
    {
      "epoch": 1.080389634314088,
      "grad_norm": 0.3803492784500122,
      "learning_rate": 0.00016163616635718922,
      "loss": 3.3453,
      "step": 205780
    },
    {
      "epoch": 1.0804642072587207,
      "grad_norm": 0.36516883969306946,
      "learning_rate": 0.000161634302035094,
      "loss": 3.3112,
      "step": 205790
    },
    {
      "epoch": 1.0805387802033535,
      "grad_norm": 0.4180492162704468,
      "learning_rate": 0.0001616324377129988,
      "loss": 3.3865,
      "step": 205800
    },
    {
      "epoch": 1.0806133531479862,
      "grad_norm": 0.38213878870010376,
      "learning_rate": 0.00016163057339090361,
      "loss": 3.3375,
      "step": 205810
    },
    {
      "epoch": 1.080687926092619,
      "grad_norm": 0.3683239817619324,
      "learning_rate": 0.0001616287090688084,
      "loss": 3.3083,
      "step": 205820
    },
    {
      "epoch": 1.0807624990372515,
      "grad_norm": 0.4020049571990967,
      "learning_rate": 0.0001616268447467132,
      "loss": 3.3193,
      "step": 205830
    },
    {
      "epoch": 1.0808370719818843,
      "grad_norm": 0.36739063262939453,
      "learning_rate": 0.000161624980424618,
      "loss": 3.3306,
      "step": 205840
    },
    {
      "epoch": 1.080911644926517,
      "grad_norm": 0.3768734931945801,
      "learning_rate": 0.0001616231161025228,
      "loss": 3.3576,
      "step": 205850
    },
    {
      "epoch": 1.0809862178711498,
      "grad_norm": 0.3804568946361542,
      "learning_rate": 0.00016162125178042762,
      "loss": 3.3529,
      "step": 205860
    },
    {
      "epoch": 1.0810607908157825,
      "grad_norm": 0.38327547907829285,
      "learning_rate": 0.0001616193874583324,
      "loss": 3.367,
      "step": 205870
    },
    {
      "epoch": 1.0811353637604153,
      "grad_norm": 0.36402401328086853,
      "learning_rate": 0.0001616175231362372,
      "loss": 3.3581,
      "step": 205880
    },
    {
      "epoch": 1.081209936705048,
      "grad_norm": 0.4140135943889618,
      "learning_rate": 0.000161615658814142,
      "loss": 3.3655,
      "step": 205890
    },
    {
      "epoch": 1.0812845096496808,
      "grad_norm": 0.3662070631980896,
      "learning_rate": 0.00016161379449204682,
      "loss": 3.3492,
      "step": 205900
    },
    {
      "epoch": 1.0813590825943136,
      "grad_norm": 0.3817961812019348,
      "learning_rate": 0.00016161193016995162,
      "loss": 3.3445,
      "step": 205910
    },
    {
      "epoch": 1.0814336555389463,
      "grad_norm": 0.38499733805656433,
      "learning_rate": 0.0001616100658478564,
      "loss": 3.3362,
      "step": 205920
    },
    {
      "epoch": 1.081508228483579,
      "grad_norm": 0.4210253953933716,
      "learning_rate": 0.0001616082015257612,
      "loss": 3.3291,
      "step": 205930
    },
    {
      "epoch": 1.0815828014282116,
      "grad_norm": 0.40396785736083984,
      "learning_rate": 0.00016160633720366602,
      "loss": 3.3403,
      "step": 205940
    },
    {
      "epoch": 1.0816573743728444,
      "grad_norm": 0.3871363401412964,
      "learning_rate": 0.00016160447288157082,
      "loss": 3.3719,
      "step": 205950
    },
    {
      "epoch": 1.0817319473174771,
      "grad_norm": 0.3637710213661194,
      "learning_rate": 0.00016160260855947563,
      "loss": 3.3819,
      "step": 205960
    },
    {
      "epoch": 1.0818065202621099,
      "grad_norm": 0.36288535594940186,
      "learning_rate": 0.0001616007442373804,
      "loss": 3.394,
      "step": 205970
    },
    {
      "epoch": 1.0818810932067426,
      "grad_norm": 0.38505658507347107,
      "learning_rate": 0.0001615988799152852,
      "loss": 3.3296,
      "step": 205980
    },
    {
      "epoch": 1.0819556661513754,
      "grad_norm": 0.392228901386261,
      "learning_rate": 0.00016159701559319,
      "loss": 3.346,
      "step": 205990
    },
    {
      "epoch": 1.0820302390960082,
      "grad_norm": 0.373052179813385,
      "learning_rate": 0.00016159515127109483,
      "loss": 3.3787,
      "step": 206000
    },
    {
      "epoch": 1.082104812040641,
      "grad_norm": 0.38399550318717957,
      "learning_rate": 0.00016159328694899963,
      "loss": 3.328,
      "step": 206010
    },
    {
      "epoch": 1.0821793849852737,
      "grad_norm": 0.40285658836364746,
      "learning_rate": 0.0001615914226269044,
      "loss": 3.3574,
      "step": 206020
    },
    {
      "epoch": 1.0822539579299064,
      "grad_norm": 0.3699082136154175,
      "learning_rate": 0.00016158955830480922,
      "loss": 3.325,
      "step": 206030
    },
    {
      "epoch": 1.082328530874539,
      "grad_norm": 0.3531211316585541,
      "learning_rate": 0.000161587693982714,
      "loss": 3.343,
      "step": 206040
    },
    {
      "epoch": 1.0824031038191717,
      "grad_norm": 0.40524017810821533,
      "learning_rate": 0.0001615858296606188,
      "loss": 3.3447,
      "step": 206050
    },
    {
      "epoch": 1.0824776767638045,
      "grad_norm": 0.370296835899353,
      "learning_rate": 0.00016158396533852364,
      "loss": 3.3385,
      "step": 206060
    },
    {
      "epoch": 1.0825522497084372,
      "grad_norm": 0.3524550199508667,
      "learning_rate": 0.00016158210101642842,
      "loss": 3.3398,
      "step": 206070
    },
    {
      "epoch": 1.08262682265307,
      "grad_norm": 0.39912253618240356,
      "learning_rate": 0.00016158023669433322,
      "loss": 3.2799,
      "step": 206080
    },
    {
      "epoch": 1.0827013955977027,
      "grad_norm": 0.3806132376194,
      "learning_rate": 0.000161578372372238,
      "loss": 3.3421,
      "step": 206090
    },
    {
      "epoch": 1.0827759685423355,
      "grad_norm": 0.3822609484195709,
      "learning_rate": 0.0001615765080501428,
      "loss": 3.3896,
      "step": 206100
    },
    {
      "epoch": 1.0828505414869682,
      "grad_norm": 0.40828973054885864,
      "learning_rate": 0.00016157464372804761,
      "loss": 3.3163,
      "step": 206110
    },
    {
      "epoch": 1.082925114431601,
      "grad_norm": 0.3976960778236389,
      "learning_rate": 0.00016157277940595242,
      "loss": 3.2871,
      "step": 206120
    },
    {
      "epoch": 1.0829996873762338,
      "grad_norm": 0.3870604336261749,
      "learning_rate": 0.00016157091508385723,
      "loss": 3.3589,
      "step": 206130
    },
    {
      "epoch": 1.0830742603208665,
      "grad_norm": 0.3745461404323578,
      "learning_rate": 0.000161569050761762,
      "loss": 3.3904,
      "step": 206140
    },
    {
      "epoch": 1.083148833265499,
      "grad_norm": 0.38528549671173096,
      "learning_rate": 0.0001615671864396668,
      "loss": 3.3014,
      "step": 206150
    },
    {
      "epoch": 1.0832234062101318,
      "grad_norm": 0.377834290266037,
      "learning_rate": 0.00016156532211757162,
      "loss": 3.3907,
      "step": 206160
    },
    {
      "epoch": 1.0832979791547646,
      "grad_norm": 0.3686562776565552,
      "learning_rate": 0.00016156345779547642,
      "loss": 3.3126,
      "step": 206170
    },
    {
      "epoch": 1.0833725520993973,
      "grad_norm": 0.35525718331336975,
      "learning_rate": 0.00016156159347338123,
      "loss": 3.3406,
      "step": 206180
    },
    {
      "epoch": 1.08344712504403,
      "grad_norm": 0.3952415883541107,
      "learning_rate": 0.000161559729151286,
      "loss": 3.3415,
      "step": 206190
    },
    {
      "epoch": 1.0835216979886628,
      "grad_norm": 0.36412015557289124,
      "learning_rate": 0.00016155786482919082,
      "loss": 3.3679,
      "step": 206200
    },
    {
      "epoch": 1.0835962709332956,
      "grad_norm": 0.3943262994289398,
      "learning_rate": 0.00016155600050709562,
      "loss": 3.3499,
      "step": 206210
    },
    {
      "epoch": 1.0836708438779283,
      "grad_norm": 0.3650711178779602,
      "learning_rate": 0.0001615541361850004,
      "loss": 3.3363,
      "step": 206220
    },
    {
      "epoch": 1.083745416822561,
      "grad_norm": 0.38578781485557556,
      "learning_rate": 0.00016155227186290524,
      "loss": 3.269,
      "step": 206230
    },
    {
      "epoch": 1.0838199897671938,
      "grad_norm": 0.3700767755508423,
      "learning_rate": 0.00016155040754081001,
      "loss": 3.3157,
      "step": 206240
    },
    {
      "epoch": 1.0838945627118264,
      "grad_norm": 0.3758215606212616,
      "learning_rate": 0.00016154854321871482,
      "loss": 3.4294,
      "step": 206250
    },
    {
      "epoch": 1.0839691356564591,
      "grad_norm": 0.36088496446609497,
      "learning_rate": 0.00016154667889661963,
      "loss": 3.3103,
      "step": 206260
    },
    {
      "epoch": 1.084043708601092,
      "grad_norm": 0.3895786702632904,
      "learning_rate": 0.0001615448145745244,
      "loss": 3.3569,
      "step": 206270
    },
    {
      "epoch": 1.0841182815457246,
      "grad_norm": 0.3715839684009552,
      "learning_rate": 0.0001615429502524292,
      "loss": 3.3175,
      "step": 206280
    },
    {
      "epoch": 1.0841928544903574,
      "grad_norm": 0.37120452523231506,
      "learning_rate": 0.00016154108593033402,
      "loss": 3.4163,
      "step": 206290
    },
    {
      "epoch": 1.0842674274349902,
      "grad_norm": 0.3677530288696289,
      "learning_rate": 0.00016153922160823883,
      "loss": 3.3398,
      "step": 206300
    },
    {
      "epoch": 1.084342000379623,
      "grad_norm": 0.34397947788238525,
      "learning_rate": 0.00016153735728614363,
      "loss": 3.3368,
      "step": 206310
    },
    {
      "epoch": 1.0844165733242557,
      "grad_norm": 0.3865191638469696,
      "learning_rate": 0.0001615354929640484,
      "loss": 3.3461,
      "step": 206320
    },
    {
      "epoch": 1.0844911462688884,
      "grad_norm": 0.38098010420799255,
      "learning_rate": 0.00016153362864195322,
      "loss": 3.3337,
      "step": 206330
    },
    {
      "epoch": 1.0845657192135212,
      "grad_norm": 0.38198941946029663,
      "learning_rate": 0.00016153176431985802,
      "loss": 3.313,
      "step": 206340
    },
    {
      "epoch": 1.084640292158154,
      "grad_norm": 0.4301677346229553,
      "learning_rate": 0.00016152989999776283,
      "loss": 3.343,
      "step": 206350
    },
    {
      "epoch": 1.0847148651027865,
      "grad_norm": 0.3873157799243927,
      "learning_rate": 0.0001615280356756676,
      "loss": 3.3495,
      "step": 206360
    },
    {
      "epoch": 1.0847894380474192,
      "grad_norm": 0.3789125382900238,
      "learning_rate": 0.00016152617135357242,
      "loss": 3.3532,
      "step": 206370
    },
    {
      "epoch": 1.084864010992052,
      "grad_norm": 0.3801932632923126,
      "learning_rate": 0.00016152430703147722,
      "loss": 3.4149,
      "step": 206380
    },
    {
      "epoch": 1.0849385839366847,
      "grad_norm": 0.3719324767589569,
      "learning_rate": 0.00016152244270938203,
      "loss": 3.2952,
      "step": 206390
    },
    {
      "epoch": 1.0850131568813175,
      "grad_norm": 0.3745438754558563,
      "learning_rate": 0.00016152057838728683,
      "loss": 3.3734,
      "step": 206400
    },
    {
      "epoch": 1.0850877298259503,
      "grad_norm": 0.3924497663974762,
      "learning_rate": 0.00016151871406519161,
      "loss": 3.3159,
      "step": 206410
    },
    {
      "epoch": 1.085162302770583,
      "grad_norm": 0.397261381149292,
      "learning_rate": 0.00016151684974309642,
      "loss": 3.2447,
      "step": 206420
    },
    {
      "epoch": 1.0852368757152158,
      "grad_norm": 0.37048816680908203,
      "learning_rate": 0.00016151498542100123,
      "loss": 3.3249,
      "step": 206430
    },
    {
      "epoch": 1.0853114486598485,
      "grad_norm": 0.3543333113193512,
      "learning_rate": 0.000161513121098906,
      "loss": 3.2754,
      "step": 206440
    },
    {
      "epoch": 1.0853860216044813,
      "grad_norm": 0.36053958535194397,
      "learning_rate": 0.00016151125677681084,
      "loss": 3.3898,
      "step": 206450
    },
    {
      "epoch": 1.0854605945491138,
      "grad_norm": 0.3681241571903229,
      "learning_rate": 0.00016150939245471562,
      "loss": 3.3343,
      "step": 206460
    },
    {
      "epoch": 1.0855351674937466,
      "grad_norm": 0.4213724732398987,
      "learning_rate": 0.00016150752813262042,
      "loss": 3.3434,
      "step": 206470
    },
    {
      "epoch": 1.0856097404383793,
      "grad_norm": 0.36572661995887756,
      "learning_rate": 0.00016150566381052523,
      "loss": 3.2859,
      "step": 206480
    },
    {
      "epoch": 1.085684313383012,
      "grad_norm": 0.3840148150920868,
      "learning_rate": 0.00016150379948843,
      "loss": 3.3652,
      "step": 206490
    },
    {
      "epoch": 1.0857588863276448,
      "grad_norm": 0.3839442729949951,
      "learning_rate": 0.00016150193516633482,
      "loss": 3.3898,
      "step": 206500
    },
    {
      "epoch": 1.0858334592722776,
      "grad_norm": 0.37928247451782227,
      "learning_rate": 0.00016150007084423962,
      "loss": 3.4335,
      "step": 206510
    },
    {
      "epoch": 1.0859080322169103,
      "grad_norm": 0.35988011956214905,
      "learning_rate": 0.00016149820652214443,
      "loss": 3.3156,
      "step": 206520
    },
    {
      "epoch": 1.085982605161543,
      "grad_norm": 0.37070319056510925,
      "learning_rate": 0.00016149634220004924,
      "loss": 3.3659,
      "step": 206530
    },
    {
      "epoch": 1.0860571781061759,
      "grad_norm": 0.40139302611351013,
      "learning_rate": 0.00016149447787795401,
      "loss": 3.3316,
      "step": 206540
    },
    {
      "epoch": 1.0861317510508086,
      "grad_norm": 0.390514999628067,
      "learning_rate": 0.00016149261355585882,
      "loss": 3.316,
      "step": 206550
    },
    {
      "epoch": 1.0862063239954414,
      "grad_norm": 0.3572389483451843,
      "learning_rate": 0.00016149074923376363,
      "loss": 3.3099,
      "step": 206560
    },
    {
      "epoch": 1.086280896940074,
      "grad_norm": 0.35080960392951965,
      "learning_rate": 0.00016148888491166843,
      "loss": 3.3229,
      "step": 206570
    },
    {
      "epoch": 1.0863554698847067,
      "grad_norm": 0.3889574408531189,
      "learning_rate": 0.00016148702058957324,
      "loss": 3.3491,
      "step": 206580
    },
    {
      "epoch": 1.0864300428293394,
      "grad_norm": 0.397882342338562,
      "learning_rate": 0.00016148515626747802,
      "loss": 3.3348,
      "step": 206590
    },
    {
      "epoch": 1.0865046157739722,
      "grad_norm": 0.3903547525405884,
      "learning_rate": 0.00016148329194538283,
      "loss": 3.3222,
      "step": 206600
    },
    {
      "epoch": 1.086579188718605,
      "grad_norm": 0.3694143295288086,
      "learning_rate": 0.0001614814276232876,
      "loss": 3.2555,
      "step": 206610
    },
    {
      "epoch": 1.0866537616632377,
      "grad_norm": 0.35756853222846985,
      "learning_rate": 0.00016147956330119244,
      "loss": 3.2965,
      "step": 206620
    },
    {
      "epoch": 1.0867283346078704,
      "grad_norm": 0.3777475655078888,
      "learning_rate": 0.00016147769897909724,
      "loss": 3.3308,
      "step": 206630
    },
    {
      "epoch": 1.0868029075525032,
      "grad_norm": 0.3690352439880371,
      "learning_rate": 0.00016147583465700202,
      "loss": 3.2898,
      "step": 206640
    },
    {
      "epoch": 1.086877480497136,
      "grad_norm": 0.3974076509475708,
      "learning_rate": 0.00016147397033490683,
      "loss": 3.295,
      "step": 206650
    },
    {
      "epoch": 1.0869520534417687,
      "grad_norm": 0.401691734790802,
      "learning_rate": 0.0001614721060128116,
      "loss": 3.3219,
      "step": 206660
    },
    {
      "epoch": 1.0870266263864012,
      "grad_norm": 0.3718796968460083,
      "learning_rate": 0.00016147024169071642,
      "loss": 3.3505,
      "step": 206670
    },
    {
      "epoch": 1.087101199331034,
      "grad_norm": 0.38613083958625793,
      "learning_rate": 0.00016146837736862125,
      "loss": 3.3423,
      "step": 206680
    },
    {
      "epoch": 1.0871757722756668,
      "grad_norm": 0.3533317446708679,
      "learning_rate": 0.00016146651304652603,
      "loss": 3.3301,
      "step": 206690
    },
    {
      "epoch": 1.0872503452202995,
      "grad_norm": 0.38343310356140137,
      "learning_rate": 0.00016146464872443083,
      "loss": 3.3506,
      "step": 206700
    },
    {
      "epoch": 1.0873249181649323,
      "grad_norm": 0.37231186032295227,
      "learning_rate": 0.0001614627844023356,
      "loss": 3.2788,
      "step": 206710
    },
    {
      "epoch": 1.087399491109565,
      "grad_norm": 0.38150742650032043,
      "learning_rate": 0.00016146092008024042,
      "loss": 3.3894,
      "step": 206720
    },
    {
      "epoch": 1.0874740640541978,
      "grad_norm": 0.3893791139125824,
      "learning_rate": 0.00016145905575814523,
      "loss": 3.3545,
      "step": 206730
    },
    {
      "epoch": 1.0875486369988305,
      "grad_norm": 0.3499162495136261,
      "learning_rate": 0.00016145719143605003,
      "loss": 3.277,
      "step": 206740
    },
    {
      "epoch": 1.0876232099434633,
      "grad_norm": 0.3625462055206299,
      "learning_rate": 0.00016145532711395484,
      "loss": 3.3614,
      "step": 206750
    },
    {
      "epoch": 1.087697782888096,
      "grad_norm": 0.41259607672691345,
      "learning_rate": 0.00016145346279185962,
      "loss": 3.3067,
      "step": 206760
    },
    {
      "epoch": 1.0877723558327288,
      "grad_norm": 0.3854861557483673,
      "learning_rate": 0.00016145159846976442,
      "loss": 3.318,
      "step": 206770
    },
    {
      "epoch": 1.0878469287773613,
      "grad_norm": 0.39616093039512634,
      "learning_rate": 0.00016144973414766923,
      "loss": 3.273,
      "step": 206780
    },
    {
      "epoch": 1.087921501721994,
      "grad_norm": 0.37960657477378845,
      "learning_rate": 0.00016144786982557404,
      "loss": 3.3539,
      "step": 206790
    },
    {
      "epoch": 1.0879960746666268,
      "grad_norm": 0.3749178946018219,
      "learning_rate": 0.00016144600550347884,
      "loss": 3.3737,
      "step": 206800
    },
    {
      "epoch": 1.0880706476112596,
      "grad_norm": 0.36955857276916504,
      "learning_rate": 0.00016144414118138362,
      "loss": 3.3756,
      "step": 206810
    },
    {
      "epoch": 1.0881452205558924,
      "grad_norm": 0.3761894702911377,
      "learning_rate": 0.00016144227685928843,
      "loss": 3.3782,
      "step": 206820
    },
    {
      "epoch": 1.0882197935005251,
      "grad_norm": 0.36727625131607056,
      "learning_rate": 0.00016144041253719323,
      "loss": 3.3606,
      "step": 206830
    },
    {
      "epoch": 1.0882943664451579,
      "grad_norm": 0.3441997468471527,
      "learning_rate": 0.00016143854821509804,
      "loss": 3.2999,
      "step": 206840
    },
    {
      "epoch": 1.0883689393897906,
      "grad_norm": 0.4093201160430908,
      "learning_rate": 0.00016143668389300285,
      "loss": 3.3496,
      "step": 206850
    },
    {
      "epoch": 1.0884435123344234,
      "grad_norm": 0.3865653872489929,
      "learning_rate": 0.00016143481957090763,
      "loss": 3.3657,
      "step": 206860
    },
    {
      "epoch": 1.0885180852790561,
      "grad_norm": 0.3929993212223053,
      "learning_rate": 0.00016143295524881243,
      "loss": 3.281,
      "step": 206870
    },
    {
      "epoch": 1.0885926582236887,
      "grad_norm": 0.3810722827911377,
      "learning_rate": 0.00016143109092671724,
      "loss": 3.3289,
      "step": 206880
    },
    {
      "epoch": 1.0886672311683214,
      "grad_norm": 0.3571613132953644,
      "learning_rate": 0.00016142922660462202,
      "loss": 3.3509,
      "step": 206890
    },
    {
      "epoch": 1.0887418041129542,
      "grad_norm": 0.39506691694259644,
      "learning_rate": 0.00016142736228252685,
      "loss": 3.3218,
      "step": 206900
    },
    {
      "epoch": 1.088816377057587,
      "grad_norm": 0.44069257378578186,
      "learning_rate": 0.00016142549796043163,
      "loss": 3.3057,
      "step": 206910
    },
    {
      "epoch": 1.0888909500022197,
      "grad_norm": 0.38632360100746155,
      "learning_rate": 0.00016142363363833644,
      "loss": 3.3142,
      "step": 206920
    },
    {
      "epoch": 1.0889655229468524,
      "grad_norm": 0.39079684019088745,
      "learning_rate": 0.00016142176931624124,
      "loss": 3.3025,
      "step": 206930
    },
    {
      "epoch": 1.0890400958914852,
      "grad_norm": 0.3844844102859497,
      "learning_rate": 0.00016141990499414602,
      "loss": 3.3469,
      "step": 206940
    },
    {
      "epoch": 1.089114668836118,
      "grad_norm": 0.3707209527492523,
      "learning_rate": 0.00016141804067205083,
      "loss": 3.2801,
      "step": 206950
    },
    {
      "epoch": 1.0891892417807507,
      "grad_norm": 0.3840257227420807,
      "learning_rate": 0.00016141617634995564,
      "loss": 3.3348,
      "step": 206960
    },
    {
      "epoch": 1.0892638147253835,
      "grad_norm": 0.38936948776245117,
      "learning_rate": 0.00016141431202786044,
      "loss": 3.3902,
      "step": 206970
    },
    {
      "epoch": 1.0893383876700162,
      "grad_norm": 0.38307780027389526,
      "learning_rate": 0.00016141244770576525,
      "loss": 3.2477,
      "step": 206980
    },
    {
      "epoch": 1.0894129606146488,
      "grad_norm": 0.38378727436065674,
      "learning_rate": 0.00016141058338367003,
      "loss": 3.3359,
      "step": 206990
    },
    {
      "epoch": 1.0894875335592815,
      "grad_norm": 0.3996177613735199,
      "learning_rate": 0.00016140871906157483,
      "loss": 3.3381,
      "step": 207000
    },
    {
      "epoch": 1.0895621065039143,
      "grad_norm": 0.37423408031463623,
      "learning_rate": 0.00016140685473947964,
      "loss": 3.2904,
      "step": 207010
    },
    {
      "epoch": 1.089636679448547,
      "grad_norm": 0.41232964396476746,
      "learning_rate": 0.00016140499041738445,
      "loss": 3.3336,
      "step": 207020
    },
    {
      "epoch": 1.0897112523931798,
      "grad_norm": 0.35198652744293213,
      "learning_rate": 0.00016140312609528925,
      "loss": 3.3397,
      "step": 207030
    },
    {
      "epoch": 1.0897858253378125,
      "grad_norm": 0.3755631148815155,
      "learning_rate": 0.00016140126177319403,
      "loss": 3.4298,
      "step": 207040
    },
    {
      "epoch": 1.0898603982824453,
      "grad_norm": 0.3865205943584442,
      "learning_rate": 0.00016139939745109884,
      "loss": 3.3354,
      "step": 207050
    },
    {
      "epoch": 1.089934971227078,
      "grad_norm": 0.3790619671344757,
      "learning_rate": 0.00016139753312900364,
      "loss": 3.3203,
      "step": 207060
    },
    {
      "epoch": 1.0900095441717108,
      "grad_norm": 0.39718878269195557,
      "learning_rate": 0.00016139566880690845,
      "loss": 3.3329,
      "step": 207070
    },
    {
      "epoch": 1.0900841171163436,
      "grad_norm": 0.37399280071258545,
      "learning_rate": 0.00016139380448481326,
      "loss": 3.2815,
      "step": 207080
    },
    {
      "epoch": 1.090158690060976,
      "grad_norm": 0.39591845870018005,
      "learning_rate": 0.00016139194016271804,
      "loss": 3.3578,
      "step": 207090
    },
    {
      "epoch": 1.0902332630056089,
      "grad_norm": 0.3573071360588074,
      "learning_rate": 0.00016139007584062284,
      "loss": 3.3652,
      "step": 207100
    },
    {
      "epoch": 1.0903078359502416,
      "grad_norm": 0.4074814021587372,
      "learning_rate": 0.00016138821151852762,
      "loss": 3.3787,
      "step": 207110
    },
    {
      "epoch": 1.0903824088948744,
      "grad_norm": 0.36231786012649536,
      "learning_rate": 0.00016138634719643243,
      "loss": 3.3328,
      "step": 207120
    },
    {
      "epoch": 1.0904569818395071,
      "grad_norm": 0.36825069785118103,
      "learning_rate": 0.00016138448287433726,
      "loss": 3.4253,
      "step": 207130
    },
    {
      "epoch": 1.0905315547841399,
      "grad_norm": 0.4027332067489624,
      "learning_rate": 0.00016138261855224204,
      "loss": 3.3844,
      "step": 207140
    },
    {
      "epoch": 1.0906061277287726,
      "grad_norm": 0.37217703461647034,
      "learning_rate": 0.00016138075423014685,
      "loss": 3.3075,
      "step": 207150
    },
    {
      "epoch": 1.0906807006734054,
      "grad_norm": 0.38428670167922974,
      "learning_rate": 0.00016137888990805163,
      "loss": 3.3328,
      "step": 207160
    },
    {
      "epoch": 1.0907552736180381,
      "grad_norm": 0.3767670691013336,
      "learning_rate": 0.00016137702558595643,
      "loss": 3.3343,
      "step": 207170
    },
    {
      "epoch": 1.090829846562671,
      "grad_norm": 0.3756343424320221,
      "learning_rate": 0.00016137516126386124,
      "loss": 3.3248,
      "step": 207180
    },
    {
      "epoch": 1.0909044195073037,
      "grad_norm": 0.3980812132358551,
      "learning_rate": 0.00016137329694176604,
      "loss": 3.3668,
      "step": 207190
    },
    {
      "epoch": 1.0909789924519362,
      "grad_norm": 0.3813735246658325,
      "learning_rate": 0.00016137143261967085,
      "loss": 3.3203,
      "step": 207200
    },
    {
      "epoch": 1.091053565396569,
      "grad_norm": 0.37032902240753174,
      "learning_rate": 0.00016136956829757563,
      "loss": 3.2801,
      "step": 207210
    },
    {
      "epoch": 1.0911281383412017,
      "grad_norm": 0.3710041344165802,
      "learning_rate": 0.00016136770397548044,
      "loss": 3.2977,
      "step": 207220
    },
    {
      "epoch": 1.0912027112858345,
      "grad_norm": 0.3974548876285553,
      "learning_rate": 0.00016136583965338524,
      "loss": 3.3648,
      "step": 207230
    },
    {
      "epoch": 1.0912772842304672,
      "grad_norm": 0.38589656352996826,
      "learning_rate": 0.00016136397533129005,
      "loss": 3.2726,
      "step": 207240
    },
    {
      "epoch": 1.0913518571751,
      "grad_norm": 0.36404165625572205,
      "learning_rate": 0.00016136211100919486,
      "loss": 3.3896,
      "step": 207250
    },
    {
      "epoch": 1.0914264301197327,
      "grad_norm": 0.348198801279068,
      "learning_rate": 0.00016136024668709963,
      "loss": 3.3479,
      "step": 207260
    },
    {
      "epoch": 1.0915010030643655,
      "grad_norm": 0.3870996832847595,
      "learning_rate": 0.00016135838236500444,
      "loss": 3.3435,
      "step": 207270
    },
    {
      "epoch": 1.0915755760089982,
      "grad_norm": 0.37778544425964355,
      "learning_rate": 0.00016135651804290925,
      "loss": 3.3642,
      "step": 207280
    },
    {
      "epoch": 1.091650148953631,
      "grad_norm": 0.404936820268631,
      "learning_rate": 0.00016135465372081405,
      "loss": 3.3431,
      "step": 207290
    },
    {
      "epoch": 1.0917247218982635,
      "grad_norm": 0.37995919585227966,
      "learning_rate": 0.00016135278939871886,
      "loss": 3.3544,
      "step": 207300
    },
    {
      "epoch": 1.0917992948428963,
      "grad_norm": 0.39589226245880127,
      "learning_rate": 0.00016135092507662364,
      "loss": 3.3279,
      "step": 207310
    },
    {
      "epoch": 1.091873867787529,
      "grad_norm": 0.4089115560054779,
      "learning_rate": 0.00016134906075452845,
      "loss": 3.2953,
      "step": 207320
    },
    {
      "epoch": 1.0919484407321618,
      "grad_norm": 0.3797878623008728,
      "learning_rate": 0.00016134719643243325,
      "loss": 3.3228,
      "step": 207330
    },
    {
      "epoch": 1.0920230136767946,
      "grad_norm": 0.3741380274295807,
      "learning_rate": 0.00016134533211033803,
      "loss": 3.2765,
      "step": 207340
    },
    {
      "epoch": 1.0920975866214273,
      "grad_norm": 0.372820109128952,
      "learning_rate": 0.00016134346778824286,
      "loss": 3.3575,
      "step": 207350
    },
    {
      "epoch": 1.09217215956606,
      "grad_norm": 0.3864259123802185,
      "learning_rate": 0.00016134160346614764,
      "loss": 3.3224,
      "step": 207360
    },
    {
      "epoch": 1.0922467325106928,
      "grad_norm": 0.3960728049278259,
      "learning_rate": 0.00016133973914405245,
      "loss": 3.3285,
      "step": 207370
    },
    {
      "epoch": 1.0923213054553256,
      "grad_norm": 0.3731151223182678,
      "learning_rate": 0.00016133787482195726,
      "loss": 3.2458,
      "step": 207380
    },
    {
      "epoch": 1.0923958783999583,
      "grad_norm": 0.34949344396591187,
      "learning_rate": 0.00016133601049986204,
      "loss": 3.3412,
      "step": 207390
    },
    {
      "epoch": 1.092470451344591,
      "grad_norm": 0.38243624567985535,
      "learning_rate": 0.00016133414617776684,
      "loss": 3.3102,
      "step": 207400
    },
    {
      "epoch": 1.0925450242892236,
      "grad_norm": 0.40058761835098267,
      "learning_rate": 0.00016133228185567165,
      "loss": 3.3766,
      "step": 207410
    },
    {
      "epoch": 1.0926195972338564,
      "grad_norm": 0.36604952812194824,
      "learning_rate": 0.00016133041753357645,
      "loss": 3.3431,
      "step": 207420
    },
    {
      "epoch": 1.0926941701784891,
      "grad_norm": 0.384255051612854,
      "learning_rate": 0.00016132855321148126,
      "loss": 3.3411,
      "step": 207430
    },
    {
      "epoch": 1.0927687431231219,
      "grad_norm": 0.3920537829399109,
      "learning_rate": 0.00016132668888938604,
      "loss": 3.3305,
      "step": 207440
    },
    {
      "epoch": 1.0928433160677546,
      "grad_norm": 0.33902356028556824,
      "learning_rate": 0.00016132482456729085,
      "loss": 3.2555,
      "step": 207450
    },
    {
      "epoch": 1.0929178890123874,
      "grad_norm": 0.3637230694293976,
      "learning_rate": 0.00016132296024519565,
      "loss": 3.3091,
      "step": 207460
    },
    {
      "epoch": 1.0929924619570202,
      "grad_norm": 0.3654659688472748,
      "learning_rate": 0.00016132109592310046,
      "loss": 3.3304,
      "step": 207470
    },
    {
      "epoch": 1.093067034901653,
      "grad_norm": 0.3888222575187683,
      "learning_rate": 0.00016131923160100527,
      "loss": 3.2938,
      "step": 207480
    },
    {
      "epoch": 1.0931416078462857,
      "grad_norm": 0.3699641525745392,
      "learning_rate": 0.00016131736727891004,
      "loss": 3.3501,
      "step": 207490
    },
    {
      "epoch": 1.0932161807909184,
      "grad_norm": 0.3906410038471222,
      "learning_rate": 0.00016131550295681485,
      "loss": 3.316,
      "step": 207500
    },
    {
      "epoch": 1.093290753735551,
      "grad_norm": 0.3847300410270691,
      "learning_rate": 0.00016131363863471966,
      "loss": 3.2825,
      "step": 207510
    },
    {
      "epoch": 1.0933653266801837,
      "grad_norm": 0.3634331226348877,
      "learning_rate": 0.00016131177431262446,
      "loss": 3.295,
      "step": 207520
    },
    {
      "epoch": 1.0934398996248165,
      "grad_norm": 0.3907432556152344,
      "learning_rate": 0.00016130990999052927,
      "loss": 3.2896,
      "step": 207530
    },
    {
      "epoch": 1.0935144725694492,
      "grad_norm": 0.40356385707855225,
      "learning_rate": 0.00016130804566843405,
      "loss": 3.3229,
      "step": 207540
    },
    {
      "epoch": 1.093589045514082,
      "grad_norm": 0.40974971652030945,
      "learning_rate": 0.00016130618134633886,
      "loss": 3.3232,
      "step": 207550
    },
    {
      "epoch": 1.0936636184587147,
      "grad_norm": 0.4021227955818176,
      "learning_rate": 0.00016130431702424363,
      "loss": 3.3268,
      "step": 207560
    },
    {
      "epoch": 1.0937381914033475,
      "grad_norm": 0.3777197003364563,
      "learning_rate": 0.00016130245270214847,
      "loss": 3.3642,
      "step": 207570
    },
    {
      "epoch": 1.0938127643479802,
      "grad_norm": 0.372499018907547,
      "learning_rate": 0.00016130058838005327,
      "loss": 3.3103,
      "step": 207580
    },
    {
      "epoch": 1.093887337292613,
      "grad_norm": 0.4022258520126343,
      "learning_rate": 0.00016129872405795805,
      "loss": 3.2594,
      "step": 207590
    },
    {
      "epoch": 1.0939619102372458,
      "grad_norm": 0.4048573076725006,
      "learning_rate": 0.00016129685973586286,
      "loss": 3.2948,
      "step": 207600
    },
    {
      "epoch": 1.0940364831818785,
      "grad_norm": 0.3645334541797638,
      "learning_rate": 0.00016129499541376764,
      "loss": 3.3783,
      "step": 207610
    },
    {
      "epoch": 1.094111056126511,
      "grad_norm": 0.3766881823539734,
      "learning_rate": 0.00016129313109167245,
      "loss": 3.2765,
      "step": 207620
    },
    {
      "epoch": 1.0941856290711438,
      "grad_norm": 0.3793986737728119,
      "learning_rate": 0.00016129126676957725,
      "loss": 3.323,
      "step": 207630
    },
    {
      "epoch": 1.0942602020157766,
      "grad_norm": 0.3898875415325165,
      "learning_rate": 0.00016128940244748206,
      "loss": 3.3097,
      "step": 207640
    },
    {
      "epoch": 1.0943347749604093,
      "grad_norm": 0.3642154633998871,
      "learning_rate": 0.00016128753812538686,
      "loss": 3.3046,
      "step": 207650
    },
    {
      "epoch": 1.094409347905042,
      "grad_norm": 0.39841535687446594,
      "learning_rate": 0.00016128567380329164,
      "loss": 3.3392,
      "step": 207660
    },
    {
      "epoch": 1.0944839208496748,
      "grad_norm": 0.3781241476535797,
      "learning_rate": 0.00016128380948119645,
      "loss": 3.3991,
      "step": 207670
    },
    {
      "epoch": 1.0945584937943076,
      "grad_norm": 0.367697149515152,
      "learning_rate": 0.00016128194515910126,
      "loss": 3.3027,
      "step": 207680
    },
    {
      "epoch": 1.0946330667389403,
      "grad_norm": 0.3772820830345154,
      "learning_rate": 0.00016128008083700606,
      "loss": 3.3391,
      "step": 207690
    },
    {
      "epoch": 1.094707639683573,
      "grad_norm": 0.40583497285842896,
      "learning_rate": 0.00016127821651491087,
      "loss": 3.3104,
      "step": 207700
    },
    {
      "epoch": 1.0947822126282056,
      "grad_norm": 0.38297098875045776,
      "learning_rate": 0.00016127635219281565,
      "loss": 3.2245,
      "step": 207710
    },
    {
      "epoch": 1.0948567855728384,
      "grad_norm": 0.3696284294128418,
      "learning_rate": 0.00016127448787072045,
      "loss": 3.4067,
      "step": 207720
    },
    {
      "epoch": 1.0949313585174711,
      "grad_norm": 0.3959190249443054,
      "learning_rate": 0.00016127262354862523,
      "loss": 3.3153,
      "step": 207730
    },
    {
      "epoch": 1.095005931462104,
      "grad_norm": 0.381076842546463,
      "learning_rate": 0.00016127075922653007,
      "loss": 3.3202,
      "step": 207740
    },
    {
      "epoch": 1.0950805044067367,
      "grad_norm": 0.38109567761421204,
      "learning_rate": 0.00016126889490443487,
      "loss": 3.2628,
      "step": 207750
    },
    {
      "epoch": 1.0951550773513694,
      "grad_norm": 0.39637964963912964,
      "learning_rate": 0.00016126703058233965,
      "loss": 3.2879,
      "step": 207760
    },
    {
      "epoch": 1.0952296502960022,
      "grad_norm": 0.3815820515155792,
      "learning_rate": 0.00016126516626024446,
      "loss": 3.3648,
      "step": 207770
    },
    {
      "epoch": 1.095304223240635,
      "grad_norm": 0.3500802218914032,
      "learning_rate": 0.00016126330193814924,
      "loss": 3.2659,
      "step": 207780
    },
    {
      "epoch": 1.0953787961852677,
      "grad_norm": 0.38190925121307373,
      "learning_rate": 0.00016126143761605404,
      "loss": 3.4321,
      "step": 207790
    },
    {
      "epoch": 1.0954533691299004,
      "grad_norm": 0.38458549976348877,
      "learning_rate": 0.00016125957329395888,
      "loss": 3.317,
      "step": 207800
    },
    {
      "epoch": 1.0955279420745332,
      "grad_norm": 0.4036242365837097,
      "learning_rate": 0.00016125770897186366,
      "loss": 3.3178,
      "step": 207810
    },
    {
      "epoch": 1.095602515019166,
      "grad_norm": 0.37597107887268066,
      "learning_rate": 0.00016125584464976846,
      "loss": 3.3694,
      "step": 207820
    },
    {
      "epoch": 1.0956770879637985,
      "grad_norm": 0.36421114206314087,
      "learning_rate": 0.00016125398032767324,
      "loss": 3.3013,
      "step": 207830
    },
    {
      "epoch": 1.0957516609084312,
      "grad_norm": 0.41624921560287476,
      "learning_rate": 0.00016125211600557805,
      "loss": 3.3779,
      "step": 207840
    },
    {
      "epoch": 1.095826233853064,
      "grad_norm": 0.37803971767425537,
      "learning_rate": 0.00016125025168348285,
      "loss": 3.2451,
      "step": 207850
    },
    {
      "epoch": 1.0959008067976967,
      "grad_norm": 0.3757801353931427,
      "learning_rate": 0.00016124838736138766,
      "loss": 3.3046,
      "step": 207860
    },
    {
      "epoch": 1.0959753797423295,
      "grad_norm": 0.3596298396587372,
      "learning_rate": 0.00016124652303929247,
      "loss": 3.3212,
      "step": 207870
    },
    {
      "epoch": 1.0960499526869623,
      "grad_norm": 0.36749836802482605,
      "learning_rate": 0.00016124465871719725,
      "loss": 3.3382,
      "step": 207880
    },
    {
      "epoch": 1.096124525631595,
      "grad_norm": 0.3903803825378418,
      "learning_rate": 0.00016124279439510205,
      "loss": 3.2612,
      "step": 207890
    },
    {
      "epoch": 1.0961990985762278,
      "grad_norm": 0.38986703753471375,
      "learning_rate": 0.00016124093007300686,
      "loss": 3.2901,
      "step": 207900
    },
    {
      "epoch": 1.0962736715208605,
      "grad_norm": 0.36154043674468994,
      "learning_rate": 0.00016123906575091167,
      "loss": 3.3421,
      "step": 207910
    },
    {
      "epoch": 1.096348244465493,
      "grad_norm": 0.3629026710987091,
      "learning_rate": 0.00016123720142881647,
      "loss": 3.331,
      "step": 207920
    },
    {
      "epoch": 1.0964228174101258,
      "grad_norm": 0.38889095187187195,
      "learning_rate": 0.00016123533710672125,
      "loss": 3.2691,
      "step": 207930
    },
    {
      "epoch": 1.0964973903547586,
      "grad_norm": 0.3939788341522217,
      "learning_rate": 0.00016123347278462606,
      "loss": 3.3284,
      "step": 207940
    },
    {
      "epoch": 1.0965719632993913,
      "grad_norm": 0.36562612652778625,
      "learning_rate": 0.00016123160846253086,
      "loss": 3.366,
      "step": 207950
    },
    {
      "epoch": 1.096646536244024,
      "grad_norm": 0.38295069336891174,
      "learning_rate": 0.00016122974414043567,
      "loss": 3.3129,
      "step": 207960
    },
    {
      "epoch": 1.0967211091886568,
      "grad_norm": 0.3977161943912506,
      "learning_rate": 0.00016122787981834048,
      "loss": 3.339,
      "step": 207970
    },
    {
      "epoch": 1.0967956821332896,
      "grad_norm": 0.4045949876308441,
      "learning_rate": 0.00016122601549624526,
      "loss": 3.4141,
      "step": 207980
    },
    {
      "epoch": 1.0968702550779224,
      "grad_norm": 0.36769258975982666,
      "learning_rate": 0.00016122415117415006,
      "loss": 3.2859,
      "step": 207990
    },
    {
      "epoch": 1.096944828022555,
      "grad_norm": 0.3960282802581787,
      "learning_rate": 0.00016122228685205487,
      "loss": 3.333,
      "step": 208000
    },
    {
      "epoch": 1.0970194009671879,
      "grad_norm": 0.3660840094089508,
      "learning_rate": 0.00016122042252995965,
      "loss": 3.3128,
      "step": 208010
    },
    {
      "epoch": 1.0970939739118206,
      "grad_norm": 0.34320196509361267,
      "learning_rate": 0.00016121855820786448,
      "loss": 3.3072,
      "step": 208020
    },
    {
      "epoch": 1.0971685468564534,
      "grad_norm": 0.38615643978118896,
      "learning_rate": 0.00016121669388576926,
      "loss": 3.3789,
      "step": 208030
    },
    {
      "epoch": 1.097243119801086,
      "grad_norm": 0.4020233452320099,
      "learning_rate": 0.00016121482956367407,
      "loss": 3.3489,
      "step": 208040
    },
    {
      "epoch": 1.0973176927457187,
      "grad_norm": 0.38467729091644287,
      "learning_rate": 0.00016121296524157887,
      "loss": 3.3422,
      "step": 208050
    },
    {
      "epoch": 1.0973922656903514,
      "grad_norm": 0.3624556064605713,
      "learning_rate": 0.00016121110091948365,
      "loss": 3.2805,
      "step": 208060
    },
    {
      "epoch": 1.0974668386349842,
      "grad_norm": 0.37623417377471924,
      "learning_rate": 0.00016120923659738846,
      "loss": 3.2821,
      "step": 208070
    },
    {
      "epoch": 1.097541411579617,
      "grad_norm": 0.3956070840358734,
      "learning_rate": 0.00016120737227529326,
      "loss": 3.3788,
      "step": 208080
    },
    {
      "epoch": 1.0976159845242497,
      "grad_norm": 0.3562900424003601,
      "learning_rate": 0.00016120550795319807,
      "loss": 3.2831,
      "step": 208090
    },
    {
      "epoch": 1.0976905574688824,
      "grad_norm": 0.3697111904621124,
      "learning_rate": 0.00016120364363110288,
      "loss": 3.3899,
      "step": 208100
    },
    {
      "epoch": 1.0977651304135152,
      "grad_norm": 0.3912684917449951,
      "learning_rate": 0.00016120177930900766,
      "loss": 3.3218,
      "step": 208110
    },
    {
      "epoch": 1.097839703358148,
      "grad_norm": 0.38361501693725586,
      "learning_rate": 0.00016119991498691246,
      "loss": 3.335,
      "step": 208120
    },
    {
      "epoch": 1.0979142763027805,
      "grad_norm": 0.3536182940006256,
      "learning_rate": 0.00016119805066481727,
      "loss": 3.3738,
      "step": 208130
    },
    {
      "epoch": 1.0979888492474132,
      "grad_norm": 0.3591531813144684,
      "learning_rate": 0.00016119618634272208,
      "loss": 3.3571,
      "step": 208140
    },
    {
      "epoch": 1.098063422192046,
      "grad_norm": 0.38181155920028687,
      "learning_rate": 0.00016119432202062688,
      "loss": 3.3084,
      "step": 208150
    },
    {
      "epoch": 1.0981379951366788,
      "grad_norm": 0.40130263566970825,
      "learning_rate": 0.00016119245769853166,
      "loss": 3.3054,
      "step": 208160
    },
    {
      "epoch": 1.0982125680813115,
      "grad_norm": 0.42255327105522156,
      "learning_rate": 0.00016119059337643647,
      "loss": 3.3093,
      "step": 208170
    },
    {
      "epoch": 1.0982871410259443,
      "grad_norm": 0.36495575308799744,
      "learning_rate": 0.00016118872905434125,
      "loss": 3.2852,
      "step": 208180
    },
    {
      "epoch": 1.098361713970577,
      "grad_norm": 0.3798651695251465,
      "learning_rate": 0.00016118686473224608,
      "loss": 3.3358,
      "step": 208190
    },
    {
      "epoch": 1.0984362869152098,
      "grad_norm": 0.3604116141796112,
      "learning_rate": 0.00016118500041015089,
      "loss": 3.3449,
      "step": 208200
    },
    {
      "epoch": 1.0985108598598425,
      "grad_norm": 0.3750329315662384,
      "learning_rate": 0.00016118313608805567,
      "loss": 3.253,
      "step": 208210
    },
    {
      "epoch": 1.0985854328044753,
      "grad_norm": 0.3669683635234833,
      "learning_rate": 0.00016118127176596047,
      "loss": 3.3289,
      "step": 208220
    },
    {
      "epoch": 1.098660005749108,
      "grad_norm": 0.38085490465164185,
      "learning_rate": 0.00016117940744386525,
      "loss": 3.3403,
      "step": 208230
    },
    {
      "epoch": 1.0987345786937406,
      "grad_norm": 0.4182053804397583,
      "learning_rate": 0.00016117754312177006,
      "loss": 3.2417,
      "step": 208240
    },
    {
      "epoch": 1.0988091516383733,
      "grad_norm": 0.3760565221309662,
      "learning_rate": 0.0001611756787996749,
      "loss": 3.2578,
      "step": 208250
    },
    {
      "epoch": 1.098883724583006,
      "grad_norm": 0.39316868782043457,
      "learning_rate": 0.00016117381447757967,
      "loss": 3.29,
      "step": 208260
    },
    {
      "epoch": 1.0989582975276388,
      "grad_norm": 0.38328349590301514,
      "learning_rate": 0.00016117195015548448,
      "loss": 3.3255,
      "step": 208270
    },
    {
      "epoch": 1.0990328704722716,
      "grad_norm": 0.3853141963481903,
      "learning_rate": 0.00016117008583338925,
      "loss": 3.3529,
      "step": 208280
    },
    {
      "epoch": 1.0991074434169044,
      "grad_norm": 0.37934568524360657,
      "learning_rate": 0.00016116822151129406,
      "loss": 3.3654,
      "step": 208290
    },
    {
      "epoch": 1.0991820163615371,
      "grad_norm": 0.41124051809310913,
      "learning_rate": 0.00016116635718919887,
      "loss": 3.2667,
      "step": 208300
    },
    {
      "epoch": 1.0992565893061699,
      "grad_norm": 0.4115390479564667,
      "learning_rate": 0.00016116449286710367,
      "loss": 3.3231,
      "step": 208310
    },
    {
      "epoch": 1.0993311622508026,
      "grad_norm": 0.37728554010391235,
      "learning_rate": 0.00016116262854500848,
      "loss": 3.3284,
      "step": 208320
    },
    {
      "epoch": 1.0994057351954354,
      "grad_norm": 0.37026333808898926,
      "learning_rate": 0.00016116076422291326,
      "loss": 3.2544,
      "step": 208330
    },
    {
      "epoch": 1.099480308140068,
      "grad_norm": 0.35394102334976196,
      "learning_rate": 0.00016115889990081807,
      "loss": 3.3056,
      "step": 208340
    },
    {
      "epoch": 1.0995548810847007,
      "grad_norm": 0.38491010665893555,
      "learning_rate": 0.00016115703557872287,
      "loss": 3.3741,
      "step": 208350
    },
    {
      "epoch": 1.0996294540293334,
      "grad_norm": 0.3907153606414795,
      "learning_rate": 0.00016115517125662768,
      "loss": 3.2692,
      "step": 208360
    },
    {
      "epoch": 1.0997040269739662,
      "grad_norm": 0.3819847106933594,
      "learning_rate": 0.00016115330693453248,
      "loss": 3.34,
      "step": 208370
    },
    {
      "epoch": 1.099778599918599,
      "grad_norm": 0.3653874695301056,
      "learning_rate": 0.00016115144261243726,
      "loss": 3.2821,
      "step": 208380
    },
    {
      "epoch": 1.0998531728632317,
      "grad_norm": 0.3678254187107086,
      "learning_rate": 0.00016114957829034207,
      "loss": 3.284,
      "step": 208390
    },
    {
      "epoch": 1.0999277458078645,
      "grad_norm": 0.38791781663894653,
      "learning_rate": 0.00016114771396824688,
      "loss": 3.2961,
      "step": 208400
    },
    {
      "epoch": 1.1000023187524972,
      "grad_norm": 0.3676512837409973,
      "learning_rate": 0.00016114584964615168,
      "loss": 3.297,
      "step": 208410
    },
    {
      "epoch": 1.10007689169713,
      "grad_norm": 0.38970720767974854,
      "learning_rate": 0.0001611439853240565,
      "loss": 3.3447,
      "step": 208420
    },
    {
      "epoch": 1.1001514646417627,
      "grad_norm": 0.38653817772865295,
      "learning_rate": 0.00016114212100196127,
      "loss": 3.3735,
      "step": 208430
    },
    {
      "epoch": 1.1002260375863955,
      "grad_norm": 0.39038437604904175,
      "learning_rate": 0.00016114025667986607,
      "loss": 3.3389,
      "step": 208440
    },
    {
      "epoch": 1.100300610531028,
      "grad_norm": 0.38671383261680603,
      "learning_rate": 0.00016113839235777088,
      "loss": 3.3255,
      "step": 208450
    },
    {
      "epoch": 1.1003751834756608,
      "grad_norm": 0.36502355337142944,
      "learning_rate": 0.00016113652803567566,
      "loss": 3.37,
      "step": 208460
    },
    {
      "epoch": 1.1004497564202935,
      "grad_norm": 0.3831137716770172,
      "learning_rate": 0.0001611346637135805,
      "loss": 3.3108,
      "step": 208470
    },
    {
      "epoch": 1.1005243293649263,
      "grad_norm": 0.3721863031387329,
      "learning_rate": 0.00016113279939148527,
      "loss": 3.2828,
      "step": 208480
    },
    {
      "epoch": 1.100598902309559,
      "grad_norm": 0.36352062225341797,
      "learning_rate": 0.00016113093506939008,
      "loss": 3.3104,
      "step": 208490
    },
    {
      "epoch": 1.1006734752541918,
      "grad_norm": 0.3850226402282715,
      "learning_rate": 0.00016112907074729489,
      "loss": 3.2944,
      "step": 208500
    },
    {
      "epoch": 1.1007480481988245,
      "grad_norm": 0.3811224699020386,
      "learning_rate": 0.00016112720642519966,
      "loss": 3.2774,
      "step": 208510
    },
    {
      "epoch": 1.1008226211434573,
      "grad_norm": 0.36111685633659363,
      "learning_rate": 0.00016112534210310447,
      "loss": 3.3377,
      "step": 208520
    },
    {
      "epoch": 1.10089719408809,
      "grad_norm": 0.40150439739227295,
      "learning_rate": 0.00016112347778100928,
      "loss": 3.2911,
      "step": 208530
    },
    {
      "epoch": 1.1009717670327228,
      "grad_norm": 0.38757798075675964,
      "learning_rate": 0.00016112161345891408,
      "loss": 3.3241,
      "step": 208540
    },
    {
      "epoch": 1.1010463399773553,
      "grad_norm": 0.4030080735683441,
      "learning_rate": 0.0001611197491368189,
      "loss": 3.3258,
      "step": 208550
    },
    {
      "epoch": 1.101120912921988,
      "grad_norm": 0.41431868076324463,
      "learning_rate": 0.00016111788481472367,
      "loss": 3.3107,
      "step": 208560
    },
    {
      "epoch": 1.1011954858666209,
      "grad_norm": 0.3782653212547302,
      "learning_rate": 0.00016111602049262848,
      "loss": 3.2478,
      "step": 208570
    },
    {
      "epoch": 1.1012700588112536,
      "grad_norm": 0.35770514607429504,
      "learning_rate": 0.00016111415617053328,
      "loss": 3.3203,
      "step": 208580
    },
    {
      "epoch": 1.1013446317558864,
      "grad_norm": 0.435045063495636,
      "learning_rate": 0.0001611122918484381,
      "loss": 3.3521,
      "step": 208590
    },
    {
      "epoch": 1.1014192047005191,
      "grad_norm": 0.4143163859844208,
      "learning_rate": 0.0001611104275263429,
      "loss": 3.3104,
      "step": 208600
    },
    {
      "epoch": 1.1014937776451519,
      "grad_norm": 0.40574508905410767,
      "learning_rate": 0.00016110856320424767,
      "loss": 3.2619,
      "step": 208610
    },
    {
      "epoch": 1.1015683505897846,
      "grad_norm": 0.3780346214771271,
      "learning_rate": 0.00016110669888215248,
      "loss": 3.3189,
      "step": 208620
    },
    {
      "epoch": 1.1016429235344174,
      "grad_norm": 0.4199724793434143,
      "learning_rate": 0.00016110483456005726,
      "loss": 3.3884,
      "step": 208630
    },
    {
      "epoch": 1.1017174964790502,
      "grad_norm": 0.3736705183982849,
      "learning_rate": 0.0001611029702379621,
      "loss": 3.2956,
      "step": 208640
    },
    {
      "epoch": 1.101792069423683,
      "grad_norm": 0.37731510400772095,
      "learning_rate": 0.0001611011059158669,
      "loss": 3.312,
      "step": 208650
    },
    {
      "epoch": 1.1018666423683154,
      "grad_norm": 0.3695484697818756,
      "learning_rate": 0.00016109924159377168,
      "loss": 3.3055,
      "step": 208660
    },
    {
      "epoch": 1.1019412153129482,
      "grad_norm": 0.3708946704864502,
      "learning_rate": 0.00016109737727167648,
      "loss": 3.375,
      "step": 208670
    },
    {
      "epoch": 1.102015788257581,
      "grad_norm": 0.3991664946079254,
      "learning_rate": 0.00016109551294958126,
      "loss": 3.341,
      "step": 208680
    },
    {
      "epoch": 1.1020903612022137,
      "grad_norm": 0.4003187119960785,
      "learning_rate": 0.00016109364862748607,
      "loss": 3.3134,
      "step": 208690
    },
    {
      "epoch": 1.1021649341468465,
      "grad_norm": 0.3764178454875946,
      "learning_rate": 0.0001610917843053909,
      "loss": 3.3395,
      "step": 208700
    },
    {
      "epoch": 1.1022395070914792,
      "grad_norm": 0.3893800973892212,
      "learning_rate": 0.00016108991998329568,
      "loss": 3.3307,
      "step": 208710
    },
    {
      "epoch": 1.102314080036112,
      "grad_norm": 0.36276498436927795,
      "learning_rate": 0.0001610880556612005,
      "loss": 3.3146,
      "step": 208720
    },
    {
      "epoch": 1.1023886529807447,
      "grad_norm": 0.3706017732620239,
      "learning_rate": 0.00016108619133910527,
      "loss": 3.3048,
      "step": 208730
    },
    {
      "epoch": 1.1024632259253775,
      "grad_norm": 0.37598279118537903,
      "learning_rate": 0.00016108432701701007,
      "loss": 3.3152,
      "step": 208740
    },
    {
      "epoch": 1.1025377988700102,
      "grad_norm": 0.41930946707725525,
      "learning_rate": 0.00016108246269491488,
      "loss": 3.3414,
      "step": 208750
    },
    {
      "epoch": 1.1026123718146428,
      "grad_norm": 0.38007834553718567,
      "learning_rate": 0.0001610805983728197,
      "loss": 3.32,
      "step": 208760
    },
    {
      "epoch": 1.1026869447592755,
      "grad_norm": 0.37149778008461,
      "learning_rate": 0.0001610787340507245,
      "loss": 3.3165,
      "step": 208770
    },
    {
      "epoch": 1.1027615177039083,
      "grad_norm": 0.37386608123779297,
      "learning_rate": 0.00016107686972862927,
      "loss": 3.3377,
      "step": 208780
    },
    {
      "epoch": 1.102836090648541,
      "grad_norm": 0.40018218755722046,
      "learning_rate": 0.00016107500540653408,
      "loss": 3.3438,
      "step": 208790
    },
    {
      "epoch": 1.1029106635931738,
      "grad_norm": 0.34054362773895264,
      "learning_rate": 0.00016107314108443888,
      "loss": 3.3014,
      "step": 208800
    },
    {
      "epoch": 1.1029852365378066,
      "grad_norm": 0.3758009374141693,
      "learning_rate": 0.0001610712767623437,
      "loss": 3.2947,
      "step": 208810
    },
    {
      "epoch": 1.1030598094824393,
      "grad_norm": 0.3941083252429962,
      "learning_rate": 0.0001610694124402485,
      "loss": 3.34,
      "step": 208820
    },
    {
      "epoch": 1.103134382427072,
      "grad_norm": 0.3880746066570282,
      "learning_rate": 0.00016106754811815328,
      "loss": 3.3002,
      "step": 208830
    },
    {
      "epoch": 1.1032089553717048,
      "grad_norm": 0.3679834306240082,
      "learning_rate": 0.00016106568379605808,
      "loss": 3.2756,
      "step": 208840
    },
    {
      "epoch": 1.1032835283163376,
      "grad_norm": 0.38182052969932556,
      "learning_rate": 0.0001610638194739629,
      "loss": 3.271,
      "step": 208850
    },
    {
      "epoch": 1.1033581012609703,
      "grad_norm": 0.4120330512523651,
      "learning_rate": 0.0001610619551518677,
      "loss": 3.298,
      "step": 208860
    },
    {
      "epoch": 1.1034326742056029,
      "grad_norm": 0.40872493386268616,
      "learning_rate": 0.0001610600908297725,
      "loss": 3.3098,
      "step": 208870
    },
    {
      "epoch": 1.1035072471502356,
      "grad_norm": 0.38198184967041016,
      "learning_rate": 0.00016105822650767728,
      "loss": 3.3547,
      "step": 208880
    },
    {
      "epoch": 1.1035818200948684,
      "grad_norm": 0.3688131868839264,
      "learning_rate": 0.0001610563621855821,
      "loss": 3.3758,
      "step": 208890
    },
    {
      "epoch": 1.1036563930395011,
      "grad_norm": 0.3707377016544342,
      "learning_rate": 0.0001610544978634869,
      "loss": 3.2987,
      "step": 208900
    },
    {
      "epoch": 1.103730965984134,
      "grad_norm": 0.35932260751724243,
      "learning_rate": 0.00016105263354139167,
      "loss": 3.3028,
      "step": 208910
    },
    {
      "epoch": 1.1038055389287666,
      "grad_norm": 0.392428457736969,
      "learning_rate": 0.0001610507692192965,
      "loss": 3.323,
      "step": 208920
    },
    {
      "epoch": 1.1038801118733994,
      "grad_norm": 0.39159947633743286,
      "learning_rate": 0.00016104890489720129,
      "loss": 3.3161,
      "step": 208930
    },
    {
      "epoch": 1.1039546848180322,
      "grad_norm": 0.3803968131542206,
      "learning_rate": 0.0001610470405751061,
      "loss": 3.3225,
      "step": 208940
    },
    {
      "epoch": 1.104029257762665,
      "grad_norm": 0.3759269118309021,
      "learning_rate": 0.0001610451762530109,
      "loss": 3.3007,
      "step": 208950
    },
    {
      "epoch": 1.1041038307072977,
      "grad_norm": 0.37918519973754883,
      "learning_rate": 0.00016104331193091568,
      "loss": 3.3268,
      "step": 208960
    },
    {
      "epoch": 1.1041784036519302,
      "grad_norm": 0.37274861335754395,
      "learning_rate": 0.00016104144760882048,
      "loss": 3.2964,
      "step": 208970
    },
    {
      "epoch": 1.104252976596563,
      "grad_norm": 0.3811642825603485,
      "learning_rate": 0.0001610395832867253,
      "loss": 3.3215,
      "step": 208980
    },
    {
      "epoch": 1.1043275495411957,
      "grad_norm": 0.3818577826023102,
      "learning_rate": 0.0001610377189646301,
      "loss": 3.224,
      "step": 208990
    },
    {
      "epoch": 1.1044021224858285,
      "grad_norm": 0.37355682253837585,
      "learning_rate": 0.0001610358546425349,
      "loss": 3.3772,
      "step": 209000
    },
    {
      "epoch": 1.1044766954304612,
      "grad_norm": 0.36059534549713135,
      "learning_rate": 0.00016103399032043968,
      "loss": 3.3707,
      "step": 209010
    },
    {
      "epoch": 1.104551268375094,
      "grad_norm": 0.3695971965789795,
      "learning_rate": 0.0001610321259983445,
      "loss": 3.3883,
      "step": 209020
    },
    {
      "epoch": 1.1046258413197267,
      "grad_norm": 0.38991445302963257,
      "learning_rate": 0.0001610302616762493,
      "loss": 3.3174,
      "step": 209030
    },
    {
      "epoch": 1.1047004142643595,
      "grad_norm": 0.38114336133003235,
      "learning_rate": 0.0001610283973541541,
      "loss": 3.3401,
      "step": 209040
    },
    {
      "epoch": 1.1047749872089923,
      "grad_norm": 0.4161517918109894,
      "learning_rate": 0.00016102653303205888,
      "loss": 3.3371,
      "step": 209050
    },
    {
      "epoch": 1.104849560153625,
      "grad_norm": 0.4032602906227112,
      "learning_rate": 0.00016102466870996369,
      "loss": 3.3135,
      "step": 209060
    },
    {
      "epoch": 1.1049241330982578,
      "grad_norm": 0.3686487078666687,
      "learning_rate": 0.0001610228043878685,
      "loss": 3.3704,
      "step": 209070
    },
    {
      "epoch": 1.1049987060428903,
      "grad_norm": 0.37871429324150085,
      "learning_rate": 0.00016102094006577327,
      "loss": 3.3203,
      "step": 209080
    },
    {
      "epoch": 1.105073278987523,
      "grad_norm": 0.38934412598609924,
      "learning_rate": 0.0001610190757436781,
      "loss": 3.3376,
      "step": 209090
    },
    {
      "epoch": 1.1051478519321558,
      "grad_norm": 0.37457767128944397,
      "learning_rate": 0.00016101721142158288,
      "loss": 3.2944,
      "step": 209100
    },
    {
      "epoch": 1.1052224248767886,
      "grad_norm": 0.38693419098854065,
      "learning_rate": 0.0001610153470994877,
      "loss": 3.3403,
      "step": 209110
    },
    {
      "epoch": 1.1052969978214213,
      "grad_norm": 0.3921462297439575,
      "learning_rate": 0.0001610134827773925,
      "loss": 3.3557,
      "step": 209120
    },
    {
      "epoch": 1.105371570766054,
      "grad_norm": 0.40796422958374023,
      "learning_rate": 0.00016101161845529728,
      "loss": 3.3609,
      "step": 209130
    },
    {
      "epoch": 1.1054461437106868,
      "grad_norm": 0.3973848819732666,
      "learning_rate": 0.00016100975413320208,
      "loss": 3.2974,
      "step": 209140
    },
    {
      "epoch": 1.1055207166553196,
      "grad_norm": 0.40016061067581177,
      "learning_rate": 0.0001610078898111069,
      "loss": 3.3294,
      "step": 209150
    },
    {
      "epoch": 1.1055952895999523,
      "grad_norm": 0.35634663701057434,
      "learning_rate": 0.0001610060254890117,
      "loss": 3.3226,
      "step": 209160
    },
    {
      "epoch": 1.105669862544585,
      "grad_norm": 0.3960696756839752,
      "learning_rate": 0.0001610041611669165,
      "loss": 3.3543,
      "step": 209170
    },
    {
      "epoch": 1.1057444354892176,
      "grad_norm": 0.35936111211776733,
      "learning_rate": 0.00016100229684482128,
      "loss": 3.2904,
      "step": 209180
    },
    {
      "epoch": 1.1058190084338504,
      "grad_norm": 0.38292551040649414,
      "learning_rate": 0.0001610004325227261,
      "loss": 3.3194,
      "step": 209190
    },
    {
      "epoch": 1.1058935813784831,
      "grad_norm": 0.4140518605709076,
      "learning_rate": 0.0001609985682006309,
      "loss": 3.3386,
      "step": 209200
    },
    {
      "epoch": 1.105968154323116,
      "grad_norm": 0.3500964641571045,
      "learning_rate": 0.0001609967038785357,
      "loss": 3.3735,
      "step": 209210
    },
    {
      "epoch": 1.1060427272677487,
      "grad_norm": 0.3809668719768524,
      "learning_rate": 0.0001609948395564405,
      "loss": 3.3713,
      "step": 209220
    },
    {
      "epoch": 1.1061173002123814,
      "grad_norm": 0.3520187735557556,
      "learning_rate": 0.00016099297523434529,
      "loss": 3.3037,
      "step": 209230
    },
    {
      "epoch": 1.1061918731570142,
      "grad_norm": 0.3510408103466034,
      "learning_rate": 0.0001609911109122501,
      "loss": 3.3306,
      "step": 209240
    },
    {
      "epoch": 1.106266446101647,
      "grad_norm": 0.37493693828582764,
      "learning_rate": 0.0001609892465901549,
      "loss": 3.354,
      "step": 209250
    },
    {
      "epoch": 1.1063410190462797,
      "grad_norm": 0.4042972922325134,
      "learning_rate": 0.0001609873822680597,
      "loss": 3.3318,
      "step": 209260
    },
    {
      "epoch": 1.1064155919909124,
      "grad_norm": 0.35391563177108765,
      "learning_rate": 0.0001609855179459645,
      "loss": 3.334,
      "step": 209270
    },
    {
      "epoch": 1.1064901649355452,
      "grad_norm": 0.3941592872142792,
      "learning_rate": 0.0001609836536238693,
      "loss": 3.3565,
      "step": 209280
    },
    {
      "epoch": 1.1065647378801777,
      "grad_norm": 0.37821951508522034,
      "learning_rate": 0.0001609817893017741,
      "loss": 3.2837,
      "step": 209290
    },
    {
      "epoch": 1.1066393108248105,
      "grad_norm": 0.38145673274993896,
      "learning_rate": 0.00016097992497967888,
      "loss": 3.3344,
      "step": 209300
    },
    {
      "epoch": 1.1067138837694432,
      "grad_norm": 0.3708159327507019,
      "learning_rate": 0.0001609780606575837,
      "loss": 3.3527,
      "step": 209310
    },
    {
      "epoch": 1.106788456714076,
      "grad_norm": 0.37736719846725464,
      "learning_rate": 0.00016097619633548851,
      "loss": 3.3056,
      "step": 209320
    },
    {
      "epoch": 1.1068630296587088,
      "grad_norm": 0.3795693814754486,
      "learning_rate": 0.0001609743320133933,
      "loss": 3.2173,
      "step": 209330
    },
    {
      "epoch": 1.1069376026033415,
      "grad_norm": 0.37921497225761414,
      "learning_rate": 0.0001609724676912981,
      "loss": 3.3297,
      "step": 209340
    },
    {
      "epoch": 1.1070121755479743,
      "grad_norm": 0.37439873814582825,
      "learning_rate": 0.00016097060336920288,
      "loss": 3.3415,
      "step": 209350
    },
    {
      "epoch": 1.107086748492607,
      "grad_norm": 0.37390097975730896,
      "learning_rate": 0.00016096873904710769,
      "loss": 3.3655,
      "step": 209360
    },
    {
      "epoch": 1.1071613214372398,
      "grad_norm": 0.38186317682266235,
      "learning_rate": 0.00016096687472501252,
      "loss": 3.3129,
      "step": 209370
    },
    {
      "epoch": 1.1072358943818725,
      "grad_norm": 0.3570472002029419,
      "learning_rate": 0.0001609650104029173,
      "loss": 3.3747,
      "step": 209380
    },
    {
      "epoch": 1.107310467326505,
      "grad_norm": 0.41569674015045166,
      "learning_rate": 0.0001609631460808221,
      "loss": 3.3387,
      "step": 209390
    },
    {
      "epoch": 1.1073850402711378,
      "grad_norm": 0.36735406517982483,
      "learning_rate": 0.00016096128175872688,
      "loss": 3.3317,
      "step": 209400
    },
    {
      "epoch": 1.1074596132157706,
      "grad_norm": 0.4008025825023651,
      "learning_rate": 0.0001609594174366317,
      "loss": 3.3514,
      "step": 209410
    },
    {
      "epoch": 1.1075341861604033,
      "grad_norm": 0.38884082436561584,
      "learning_rate": 0.0001609575531145365,
      "loss": 3.3299,
      "step": 209420
    },
    {
      "epoch": 1.107608759105036,
      "grad_norm": 0.3655414283275604,
      "learning_rate": 0.0001609556887924413,
      "loss": 3.3116,
      "step": 209430
    },
    {
      "epoch": 1.1076833320496688,
      "grad_norm": 0.3726980686187744,
      "learning_rate": 0.0001609538244703461,
      "loss": 3.3133,
      "step": 209440
    },
    {
      "epoch": 1.1077579049943016,
      "grad_norm": 0.3822226822376251,
      "learning_rate": 0.0001609519601482509,
      "loss": 3.3091,
      "step": 209450
    },
    {
      "epoch": 1.1078324779389344,
      "grad_norm": 0.3786298930644989,
      "learning_rate": 0.0001609500958261557,
      "loss": 3.2752,
      "step": 209460
    },
    {
      "epoch": 1.1079070508835671,
      "grad_norm": 0.38947758078575134,
      "learning_rate": 0.0001609482315040605,
      "loss": 3.3431,
      "step": 209470
    },
    {
      "epoch": 1.1079816238281999,
      "grad_norm": 0.39473608136177063,
      "learning_rate": 0.0001609463671819653,
      "loss": 3.336,
      "step": 209480
    },
    {
      "epoch": 1.1080561967728326,
      "grad_norm": 0.37454333901405334,
      "learning_rate": 0.00016094450285987011,
      "loss": 3.3681,
      "step": 209490
    },
    {
      "epoch": 1.1081307697174652,
      "grad_norm": 0.39757025241851807,
      "learning_rate": 0.0001609426385377749,
      "loss": 3.3761,
      "step": 209500
    },
    {
      "epoch": 1.108205342662098,
      "grad_norm": 0.39325150847435,
      "learning_rate": 0.0001609407742156797,
      "loss": 3.2854,
      "step": 209510
    },
    {
      "epoch": 1.1082799156067307,
      "grad_norm": 0.6239866614341736,
      "learning_rate": 0.0001609389098935845,
      "loss": 3.3243,
      "step": 209520
    },
    {
      "epoch": 1.1083544885513634,
      "grad_norm": 0.38877928256988525,
      "learning_rate": 0.00016093704557148928,
      "loss": 3.2697,
      "step": 209530
    },
    {
      "epoch": 1.1084290614959962,
      "grad_norm": 0.38107553124427795,
      "learning_rate": 0.00016093518124939412,
      "loss": 3.3043,
      "step": 209540
    },
    {
      "epoch": 1.108503634440629,
      "grad_norm": 0.35960090160369873,
      "learning_rate": 0.0001609333169272989,
      "loss": 3.2898,
      "step": 209550
    },
    {
      "epoch": 1.1085782073852617,
      "grad_norm": 0.38654637336730957,
      "learning_rate": 0.0001609314526052037,
      "loss": 3.365,
      "step": 209560
    },
    {
      "epoch": 1.1086527803298944,
      "grad_norm": 0.3777846395969391,
      "learning_rate": 0.0001609295882831085,
      "loss": 3.4155,
      "step": 209570
    },
    {
      "epoch": 1.1087273532745272,
      "grad_norm": 0.3833538293838501,
      "learning_rate": 0.0001609277239610133,
      "loss": 3.3183,
      "step": 209580
    },
    {
      "epoch": 1.10880192621916,
      "grad_norm": 0.3920147120952606,
      "learning_rate": 0.0001609258596389181,
      "loss": 3.2719,
      "step": 209590
    },
    {
      "epoch": 1.1088764991637925,
      "grad_norm": 0.38993585109710693,
      "learning_rate": 0.0001609239953168229,
      "loss": 3.2559,
      "step": 209600
    },
    {
      "epoch": 1.1089510721084253,
      "grad_norm": 0.3872157037258148,
      "learning_rate": 0.0001609221309947277,
      "loss": 3.3558,
      "step": 209610
    },
    {
      "epoch": 1.109025645053058,
      "grad_norm": 0.3810243606567383,
      "learning_rate": 0.00016092026667263251,
      "loss": 3.336,
      "step": 209620
    },
    {
      "epoch": 1.1091002179976908,
      "grad_norm": 0.35689952969551086,
      "learning_rate": 0.0001609184023505373,
      "loss": 3.3557,
      "step": 209630
    },
    {
      "epoch": 1.1091747909423235,
      "grad_norm": 0.3848147392272949,
      "learning_rate": 0.0001609165380284421,
      "loss": 3.4031,
      "step": 209640
    },
    {
      "epoch": 1.1092493638869563,
      "grad_norm": 0.36117634177207947,
      "learning_rate": 0.0001609146737063469,
      "loss": 3.2512,
      "step": 209650
    },
    {
      "epoch": 1.109323936831589,
      "grad_norm": 0.39045169949531555,
      "learning_rate": 0.0001609128093842517,
      "loss": 3.3459,
      "step": 209660
    },
    {
      "epoch": 1.1093985097762218,
      "grad_norm": 0.3712258040904999,
      "learning_rate": 0.00016091094506215652,
      "loss": 3.3353,
      "step": 209670
    },
    {
      "epoch": 1.1094730827208545,
      "grad_norm": 0.4125850200653076,
      "learning_rate": 0.0001609090807400613,
      "loss": 3.3677,
      "step": 209680
    },
    {
      "epoch": 1.1095476556654873,
      "grad_norm": 0.37353014945983887,
      "learning_rate": 0.0001609072164179661,
      "loss": 3.3326,
      "step": 209690
    },
    {
      "epoch": 1.10962222861012,
      "grad_norm": 0.36849507689476013,
      "learning_rate": 0.0001609053520958709,
      "loss": 3.2522,
      "step": 209700
    },
    {
      "epoch": 1.1096968015547526,
      "grad_norm": 0.37686076760292053,
      "learning_rate": 0.00016090348777377572,
      "loss": 3.2879,
      "step": 209710
    },
    {
      "epoch": 1.1097713744993853,
      "grad_norm": 0.38165053725242615,
      "learning_rate": 0.00016090162345168052,
      "loss": 3.3494,
      "step": 209720
    },
    {
      "epoch": 1.109845947444018,
      "grad_norm": 0.3664742410182953,
      "learning_rate": 0.0001608997591295853,
      "loss": 3.2814,
      "step": 209730
    },
    {
      "epoch": 1.1099205203886509,
      "grad_norm": 0.4516686499118805,
      "learning_rate": 0.0001608978948074901,
      "loss": 3.2475,
      "step": 209740
    },
    {
      "epoch": 1.1099950933332836,
      "grad_norm": 0.3782787322998047,
      "learning_rate": 0.0001608960304853949,
      "loss": 3.326,
      "step": 209750
    },
    {
      "epoch": 1.1100696662779164,
      "grad_norm": 0.4073173999786377,
      "learning_rate": 0.00016089416616329972,
      "loss": 3.3159,
      "step": 209760
    },
    {
      "epoch": 1.1101442392225491,
      "grad_norm": 0.38207995891571045,
      "learning_rate": 0.00016089230184120453,
      "loss": 3.2707,
      "step": 209770
    },
    {
      "epoch": 1.1102188121671819,
      "grad_norm": 0.40363121032714844,
      "learning_rate": 0.0001608904375191093,
      "loss": 3.3453,
      "step": 209780
    },
    {
      "epoch": 1.1102933851118146,
      "grad_norm": 0.346671462059021,
      "learning_rate": 0.0001608885731970141,
      "loss": 3.3394,
      "step": 209790
    },
    {
      "epoch": 1.1103679580564474,
      "grad_norm": 0.37420403957366943,
      "learning_rate": 0.0001608867088749189,
      "loss": 3.369,
      "step": 209800
    },
    {
      "epoch": 1.11044253100108,
      "grad_norm": 0.3805752694606781,
      "learning_rate": 0.0001608848445528237,
      "loss": 3.3673,
      "step": 209810
    },
    {
      "epoch": 1.1105171039457127,
      "grad_norm": 0.38285213708877563,
      "learning_rate": 0.00016088298023072853,
      "loss": 3.3026,
      "step": 209820
    },
    {
      "epoch": 1.1105916768903454,
      "grad_norm": 0.3989125192165375,
      "learning_rate": 0.0001608811159086333,
      "loss": 3.2831,
      "step": 209830
    },
    {
      "epoch": 1.1106662498349782,
      "grad_norm": 0.3671664893627167,
      "learning_rate": 0.00016087925158653812,
      "loss": 3.3611,
      "step": 209840
    },
    {
      "epoch": 1.110740822779611,
      "grad_norm": 0.40089964866638184,
      "learning_rate": 0.0001608773872644429,
      "loss": 3.3105,
      "step": 209850
    },
    {
      "epoch": 1.1108153957242437,
      "grad_norm": 0.3751146197319031,
      "learning_rate": 0.0001608755229423477,
      "loss": 3.3769,
      "step": 209860
    },
    {
      "epoch": 1.1108899686688765,
      "grad_norm": 0.37974709272384644,
      "learning_rate": 0.0001608736586202525,
      "loss": 3.3467,
      "step": 209870
    },
    {
      "epoch": 1.1109645416135092,
      "grad_norm": 0.34646281599998474,
      "learning_rate": 0.00016087179429815732,
      "loss": 3.3155,
      "step": 209880
    },
    {
      "epoch": 1.111039114558142,
      "grad_norm": 0.37608519196510315,
      "learning_rate": 0.00016086992997606212,
      "loss": 3.3362,
      "step": 209890
    },
    {
      "epoch": 1.1111136875027747,
      "grad_norm": 0.3784770369529724,
      "learning_rate": 0.0001608680656539669,
      "loss": 3.2709,
      "step": 209900
    },
    {
      "epoch": 1.1111882604474075,
      "grad_norm": 0.3647305965423584,
      "learning_rate": 0.0001608662013318717,
      "loss": 3.2841,
      "step": 209910
    },
    {
      "epoch": 1.11126283339204,
      "grad_norm": 0.3966321647167206,
      "learning_rate": 0.00016086433700977651,
      "loss": 3.2595,
      "step": 209920
    },
    {
      "epoch": 1.1113374063366728,
      "grad_norm": 0.3561705946922302,
      "learning_rate": 0.00016086247268768132,
      "loss": 3.2516,
      "step": 209930
    },
    {
      "epoch": 1.1114119792813055,
      "grad_norm": 0.36603695154190063,
      "learning_rate": 0.00016086060836558613,
      "loss": 3.3168,
      "step": 209940
    },
    {
      "epoch": 1.1114865522259383,
      "grad_norm": 0.35929539799690247,
      "learning_rate": 0.0001608587440434909,
      "loss": 3.3041,
      "step": 209950
    },
    {
      "epoch": 1.111561125170571,
      "grad_norm": 0.36134761571884155,
      "learning_rate": 0.0001608568797213957,
      "loss": 3.2666,
      "step": 209960
    },
    {
      "epoch": 1.1116356981152038,
      "grad_norm": 0.3620706796646118,
      "learning_rate": 0.00016085501539930052,
      "loss": 3.3117,
      "step": 209970
    },
    {
      "epoch": 1.1117102710598366,
      "grad_norm": 0.3686923384666443,
      "learning_rate": 0.0001608531510772053,
      "loss": 3.3331,
      "step": 209980
    },
    {
      "epoch": 1.1117848440044693,
      "grad_norm": 0.3716854155063629,
      "learning_rate": 0.00016085128675511013,
      "loss": 3.292,
      "step": 209990
    },
    {
      "epoch": 1.111859416949102,
      "grad_norm": 0.3687630891799927,
      "learning_rate": 0.0001608494224330149,
      "loss": 3.3074,
      "step": 210000
    },
    {
      "epoch": 1.1119339898937348,
      "grad_norm": 0.3858794569969177,
      "learning_rate": 0.00016084755811091972,
      "loss": 3.3025,
      "step": 210010
    },
    {
      "epoch": 1.1120085628383674,
      "grad_norm": 0.3556927740573883,
      "learning_rate": 0.00016084569378882452,
      "loss": 3.2547,
      "step": 210020
    },
    {
      "epoch": 1.112083135783,
      "grad_norm": 0.36691713333129883,
      "learning_rate": 0.0001608438294667293,
      "loss": 3.2846,
      "step": 210030
    },
    {
      "epoch": 1.1121577087276329,
      "grad_norm": 0.3930398225784302,
      "learning_rate": 0.0001608419651446341,
      "loss": 3.3298,
      "step": 210040
    },
    {
      "epoch": 1.1122322816722656,
      "grad_norm": 0.3917334973812103,
      "learning_rate": 0.00016084010082253891,
      "loss": 3.3604,
      "step": 210050
    },
    {
      "epoch": 1.1123068546168984,
      "grad_norm": 0.3726552724838257,
      "learning_rate": 0.00016083823650044372,
      "loss": 3.3751,
      "step": 210060
    },
    {
      "epoch": 1.1123814275615311,
      "grad_norm": 0.38152557611465454,
      "learning_rate": 0.00016083637217834853,
      "loss": 3.3273,
      "step": 210070
    },
    {
      "epoch": 1.112456000506164,
      "grad_norm": 0.38524651527404785,
      "learning_rate": 0.0001608345078562533,
      "loss": 3.2901,
      "step": 210080
    },
    {
      "epoch": 1.1125305734507966,
      "grad_norm": 0.38159704208374023,
      "learning_rate": 0.0001608326435341581,
      "loss": 3.3755,
      "step": 210090
    },
    {
      "epoch": 1.1126051463954294,
      "grad_norm": 0.3813578486442566,
      "learning_rate": 0.00016083077921206292,
      "loss": 3.2986,
      "step": 210100
    },
    {
      "epoch": 1.1126797193400622,
      "grad_norm": 0.39865103363990784,
      "learning_rate": 0.00016082891488996773,
      "loss": 3.2731,
      "step": 210110
    },
    {
      "epoch": 1.112754292284695,
      "grad_norm": 0.4043910801410675,
      "learning_rate": 0.00016082705056787253,
      "loss": 3.286,
      "step": 210120
    },
    {
      "epoch": 1.1128288652293274,
      "grad_norm": 0.3986341059207916,
      "learning_rate": 0.0001608251862457773,
      "loss": 3.2621,
      "step": 210130
    },
    {
      "epoch": 1.1129034381739602,
      "grad_norm": 0.3803565502166748,
      "learning_rate": 0.00016082332192368212,
      "loss": 3.3093,
      "step": 210140
    },
    {
      "epoch": 1.112978011118593,
      "grad_norm": 0.41276228427886963,
      "learning_rate": 0.00016082145760158692,
      "loss": 3.2933,
      "step": 210150
    },
    {
      "epoch": 1.1130525840632257,
      "grad_norm": 0.3910941481590271,
      "learning_rate": 0.00016081959327949173,
      "loss": 3.3304,
      "step": 210160
    },
    {
      "epoch": 1.1131271570078585,
      "grad_norm": 0.3814839720726013,
      "learning_rate": 0.00016081772895739654,
      "loss": 3.3035,
      "step": 210170
    },
    {
      "epoch": 1.1132017299524912,
      "grad_norm": 0.38814499974250793,
      "learning_rate": 0.00016081586463530132,
      "loss": 3.315,
      "step": 210180
    },
    {
      "epoch": 1.113276302897124,
      "grad_norm": 0.38217198848724365,
      "learning_rate": 0.00016081400031320612,
      "loss": 3.3752,
      "step": 210190
    },
    {
      "epoch": 1.1133508758417567,
      "grad_norm": 0.3861702084541321,
      "learning_rate": 0.0001608121359911109,
      "loss": 3.2826,
      "step": 210200
    },
    {
      "epoch": 1.1134254487863895,
      "grad_norm": 0.380752831697464,
      "learning_rate": 0.00016081027166901573,
      "loss": 3.3016,
      "step": 210210
    },
    {
      "epoch": 1.1135000217310222,
      "grad_norm": 0.4025547504425049,
      "learning_rate": 0.00016080840734692054,
      "loss": 3.2945,
      "step": 210220
    },
    {
      "epoch": 1.1135745946756548,
      "grad_norm": 0.38080862164497375,
      "learning_rate": 0.00016080654302482532,
      "loss": 3.3037,
      "step": 210230
    },
    {
      "epoch": 1.1136491676202875,
      "grad_norm": 0.36631742119789124,
      "learning_rate": 0.00016080467870273013,
      "loss": 3.3335,
      "step": 210240
    },
    {
      "epoch": 1.1137237405649203,
      "grad_norm": 0.35106340050697327,
      "learning_rate": 0.0001608028143806349,
      "loss": 3.2653,
      "step": 210250
    },
    {
      "epoch": 1.113798313509553,
      "grad_norm": 0.3817857801914215,
      "learning_rate": 0.0001608009500585397,
      "loss": 3.3132,
      "step": 210260
    },
    {
      "epoch": 1.1138728864541858,
      "grad_norm": 0.36679548025131226,
      "learning_rate": 0.00016079908573644454,
      "loss": 3.3049,
      "step": 210270
    },
    {
      "epoch": 1.1139474593988186,
      "grad_norm": 0.38241150975227356,
      "learning_rate": 0.00016079722141434932,
      "loss": 3.346,
      "step": 210280
    },
    {
      "epoch": 1.1140220323434513,
      "grad_norm": 0.38677164912223816,
      "learning_rate": 0.00016079535709225413,
      "loss": 3.3069,
      "step": 210290
    },
    {
      "epoch": 1.114096605288084,
      "grad_norm": 0.38278597593307495,
      "learning_rate": 0.0001607934927701589,
      "loss": 3.2921,
      "step": 210300
    },
    {
      "epoch": 1.1141711782327168,
      "grad_norm": 0.3883728086948395,
      "learning_rate": 0.00016079162844806372,
      "loss": 3.3218,
      "step": 210310
    },
    {
      "epoch": 1.1142457511773496,
      "grad_norm": 0.4588638246059418,
      "learning_rate": 0.00016078976412596852,
      "loss": 3.2743,
      "step": 210320
    },
    {
      "epoch": 1.1143203241219823,
      "grad_norm": 0.39326211810112,
      "learning_rate": 0.00016078789980387333,
      "loss": 3.3171,
      "step": 210330
    },
    {
      "epoch": 1.1143948970666149,
      "grad_norm": 0.35806727409362793,
      "learning_rate": 0.00016078603548177813,
      "loss": 3.2839,
      "step": 210340
    },
    {
      "epoch": 1.1144694700112476,
      "grad_norm": 0.3671194911003113,
      "learning_rate": 0.00016078417115968291,
      "loss": 3.34,
      "step": 210350
    },
    {
      "epoch": 1.1145440429558804,
      "grad_norm": 0.3982205092906952,
      "learning_rate": 0.00016078230683758772,
      "loss": 3.2938,
      "step": 210360
    },
    {
      "epoch": 1.1146186159005131,
      "grad_norm": 0.37766727805137634,
      "learning_rate": 0.00016078044251549253,
      "loss": 3.2983,
      "step": 210370
    },
    {
      "epoch": 1.114693188845146,
      "grad_norm": 0.37777623534202576,
      "learning_rate": 0.00016077857819339733,
      "loss": 3.3457,
      "step": 210380
    },
    {
      "epoch": 1.1147677617897787,
      "grad_norm": 0.37565305829048157,
      "learning_rate": 0.00016077671387130214,
      "loss": 3.2776,
      "step": 210390
    },
    {
      "epoch": 1.1148423347344114,
      "grad_norm": 0.3908539414405823,
      "learning_rate": 0.00016077484954920692,
      "loss": 3.3799,
      "step": 210400
    },
    {
      "epoch": 1.1149169076790442,
      "grad_norm": 0.4068916141986847,
      "learning_rate": 0.00016077298522711172,
      "loss": 3.3474,
      "step": 210410
    },
    {
      "epoch": 1.114991480623677,
      "grad_norm": 0.38571009039878845,
      "learning_rate": 0.0001607711209050165,
      "loss": 3.3073,
      "step": 210420
    },
    {
      "epoch": 1.1150660535683097,
      "grad_norm": 0.3909994661808014,
      "learning_rate": 0.0001607692565829213,
      "loss": 3.3006,
      "step": 210430
    },
    {
      "epoch": 1.1151406265129422,
      "grad_norm": 0.38482865691185,
      "learning_rate": 0.00016076739226082614,
      "loss": 3.3271,
      "step": 210440
    },
    {
      "epoch": 1.115215199457575,
      "grad_norm": 0.4102254807949066,
      "learning_rate": 0.00016076552793873092,
      "loss": 3.2717,
      "step": 210450
    },
    {
      "epoch": 1.1152897724022077,
      "grad_norm": 0.37488260865211487,
      "learning_rate": 0.00016076366361663573,
      "loss": 3.3184,
      "step": 210460
    },
    {
      "epoch": 1.1153643453468405,
      "grad_norm": 0.40094026923179626,
      "learning_rate": 0.0001607617992945405,
      "loss": 3.2876,
      "step": 210470
    },
    {
      "epoch": 1.1154389182914732,
      "grad_norm": 0.40857404470443726,
      "learning_rate": 0.00016075993497244531,
      "loss": 3.33,
      "step": 210480
    },
    {
      "epoch": 1.115513491236106,
      "grad_norm": 0.3813484013080597,
      "learning_rate": 0.00016075807065035012,
      "loss": 3.2956,
      "step": 210490
    },
    {
      "epoch": 1.1155880641807387,
      "grad_norm": 0.3890865445137024,
      "learning_rate": 0.00016075620632825493,
      "loss": 3.3192,
      "step": 210500
    },
    {
      "epoch": 1.1156626371253715,
      "grad_norm": 0.38840678334236145,
      "learning_rate": 0.00016075434200615973,
      "loss": 3.3461,
      "step": 210510
    },
    {
      "epoch": 1.1157372100700043,
      "grad_norm": 0.3429664969444275,
      "learning_rate": 0.0001607524776840645,
      "loss": 3.3147,
      "step": 210520
    },
    {
      "epoch": 1.115811783014637,
      "grad_norm": 0.3806402385234833,
      "learning_rate": 0.00016075061336196932,
      "loss": 3.2666,
      "step": 210530
    },
    {
      "epoch": 1.1158863559592698,
      "grad_norm": 0.366720587015152,
      "learning_rate": 0.00016074874903987413,
      "loss": 3.3494,
      "step": 210540
    },
    {
      "epoch": 1.1159609289039023,
      "grad_norm": 0.3685673177242279,
      "learning_rate": 0.00016074688471777893,
      "loss": 3.2575,
      "step": 210550
    },
    {
      "epoch": 1.116035501848535,
      "grad_norm": 0.3616406321525574,
      "learning_rate": 0.00016074502039568374,
      "loss": 3.2886,
      "step": 210560
    },
    {
      "epoch": 1.1161100747931678,
      "grad_norm": 0.38098394870758057,
      "learning_rate": 0.00016074315607358852,
      "loss": 3.3064,
      "step": 210570
    },
    {
      "epoch": 1.1161846477378006,
      "grad_norm": 0.3825506567955017,
      "learning_rate": 0.00016074129175149332,
      "loss": 3.306,
      "step": 210580
    },
    {
      "epoch": 1.1162592206824333,
      "grad_norm": 0.3746868371963501,
      "learning_rate": 0.00016073942742939813,
      "loss": 3.3358,
      "step": 210590
    },
    {
      "epoch": 1.116333793627066,
      "grad_norm": 0.4054054915904999,
      "learning_rate": 0.00016073756310730294,
      "loss": 3.291,
      "step": 210600
    },
    {
      "epoch": 1.1164083665716988,
      "grad_norm": 0.3706802427768707,
      "learning_rate": 0.00016073569878520774,
      "loss": 3.4,
      "step": 210610
    },
    {
      "epoch": 1.1164829395163316,
      "grad_norm": 0.38734927773475647,
      "learning_rate": 0.00016073383446311252,
      "loss": 3.2946,
      "step": 210620
    },
    {
      "epoch": 1.1165575124609644,
      "grad_norm": 0.3613775372505188,
      "learning_rate": 0.00016073197014101733,
      "loss": 3.2653,
      "step": 210630
    },
    {
      "epoch": 1.1166320854055969,
      "grad_norm": 0.3941788673400879,
      "learning_rate": 0.00016073010581892213,
      "loss": 3.2676,
      "step": 210640
    },
    {
      "epoch": 1.1167066583502296,
      "grad_norm": 0.3549763262271881,
      "learning_rate": 0.00016072824149682691,
      "loss": 3.3415,
      "step": 210650
    },
    {
      "epoch": 1.1167812312948624,
      "grad_norm": 0.3624010980129242,
      "learning_rate": 0.00016072637717473175,
      "loss": 3.2914,
      "step": 210660
    },
    {
      "epoch": 1.1168558042394952,
      "grad_norm": 0.3971410393714905,
      "learning_rate": 0.00016072451285263653,
      "loss": 3.3013,
      "step": 210670
    },
    {
      "epoch": 1.116930377184128,
      "grad_norm": 0.38170522451400757,
      "learning_rate": 0.00016072264853054133,
      "loss": 3.341,
      "step": 210680
    },
    {
      "epoch": 1.1170049501287607,
      "grad_norm": 0.37528011202812195,
      "learning_rate": 0.00016072078420844614,
      "loss": 3.3428,
      "step": 210690
    },
    {
      "epoch": 1.1170795230733934,
      "grad_norm": 0.3814779222011566,
      "learning_rate": 0.00016071891988635092,
      "loss": 3.3012,
      "step": 210700
    },
    {
      "epoch": 1.1171540960180262,
      "grad_norm": 0.3594461977481842,
      "learning_rate": 0.00016071705556425572,
      "loss": 3.3022,
      "step": 210710
    },
    {
      "epoch": 1.117228668962659,
      "grad_norm": 0.3897134065628052,
      "learning_rate": 0.00016071519124216053,
      "loss": 3.3112,
      "step": 210720
    },
    {
      "epoch": 1.1173032419072917,
      "grad_norm": 0.40142562985420227,
      "learning_rate": 0.00016071332692006534,
      "loss": 3.3169,
      "step": 210730
    },
    {
      "epoch": 1.1173778148519244,
      "grad_norm": 0.36690035462379456,
      "learning_rate": 0.00016071146259797014,
      "loss": 3.341,
      "step": 210740
    },
    {
      "epoch": 1.1174523877965572,
      "grad_norm": 0.3918124735355377,
      "learning_rate": 0.00016070959827587492,
      "loss": 3.3017,
      "step": 210750
    },
    {
      "epoch": 1.1175269607411897,
      "grad_norm": 0.37275174260139465,
      "learning_rate": 0.00016070773395377973,
      "loss": 3.2671,
      "step": 210760
    },
    {
      "epoch": 1.1176015336858225,
      "grad_norm": 0.37872645258903503,
      "learning_rate": 0.00016070586963168454,
      "loss": 3.3346,
      "step": 210770
    },
    {
      "epoch": 1.1176761066304552,
      "grad_norm": 0.36095550656318665,
      "learning_rate": 0.00016070400530958934,
      "loss": 3.3462,
      "step": 210780
    },
    {
      "epoch": 1.117750679575088,
      "grad_norm": 0.3503425717353821,
      "learning_rate": 0.00016070214098749415,
      "loss": 3.3507,
      "step": 210790
    },
    {
      "epoch": 1.1178252525197208,
      "grad_norm": 0.3729821741580963,
      "learning_rate": 0.00016070027666539893,
      "loss": 3.2733,
      "step": 210800
    },
    {
      "epoch": 1.1178998254643535,
      "grad_norm": 0.3775716722011566,
      "learning_rate": 0.00016069841234330373,
      "loss": 3.2964,
      "step": 210810
    },
    {
      "epoch": 1.1179743984089863,
      "grad_norm": 0.378314346075058,
      "learning_rate": 0.00016069654802120854,
      "loss": 3.362,
      "step": 210820
    },
    {
      "epoch": 1.118048971353619,
      "grad_norm": 0.3785650432109833,
      "learning_rate": 0.00016069468369911335,
      "loss": 3.306,
      "step": 210830
    },
    {
      "epoch": 1.1181235442982518,
      "grad_norm": 0.4089205265045166,
      "learning_rate": 0.00016069281937701815,
      "loss": 3.3706,
      "step": 210840
    },
    {
      "epoch": 1.1181981172428843,
      "grad_norm": 0.3483867943286896,
      "learning_rate": 0.00016069095505492293,
      "loss": 3.3562,
      "step": 210850
    },
    {
      "epoch": 1.118272690187517,
      "grad_norm": 0.38268348574638367,
      "learning_rate": 0.00016068909073282774,
      "loss": 3.3533,
      "step": 210860
    },
    {
      "epoch": 1.1183472631321498,
      "grad_norm": 0.40082594752311707,
      "learning_rate": 0.00016068722641073252,
      "loss": 3.3368,
      "step": 210870
    },
    {
      "epoch": 1.1184218360767826,
      "grad_norm": 0.36619800329208374,
      "learning_rate": 0.00016068536208863735,
      "loss": 3.2451,
      "step": 210880
    },
    {
      "epoch": 1.1184964090214153,
      "grad_norm": 0.3805232346057892,
      "learning_rate": 0.00016068349776654216,
      "loss": 3.282,
      "step": 210890
    },
    {
      "epoch": 1.118570981966048,
      "grad_norm": 0.3722974359989166,
      "learning_rate": 0.00016068163344444694,
      "loss": 3.358,
      "step": 210900
    },
    {
      "epoch": 1.1186455549106809,
      "grad_norm": 0.3776610791683197,
      "learning_rate": 0.00016067976912235174,
      "loss": 3.4015,
      "step": 210910
    },
    {
      "epoch": 1.1187201278553136,
      "grad_norm": 0.358610600233078,
      "learning_rate": 0.00016067790480025652,
      "loss": 3.2855,
      "step": 210920
    },
    {
      "epoch": 1.1187947007999464,
      "grad_norm": 0.38626959919929504,
      "learning_rate": 0.00016067604047816133,
      "loss": 3.3109,
      "step": 210930
    },
    {
      "epoch": 1.1188692737445791,
      "grad_norm": 0.3760049343109131,
      "learning_rate": 0.00016067417615606613,
      "loss": 3.3289,
      "step": 210940
    },
    {
      "epoch": 1.1189438466892119,
      "grad_norm": 0.3719380497932434,
      "learning_rate": 0.00016067231183397094,
      "loss": 3.3127,
      "step": 210950
    },
    {
      "epoch": 1.1190184196338444,
      "grad_norm": 0.36201342940330505,
      "learning_rate": 0.00016067044751187575,
      "loss": 3.2658,
      "step": 210960
    },
    {
      "epoch": 1.1190929925784772,
      "grad_norm": 0.3744107484817505,
      "learning_rate": 0.00016066858318978053,
      "loss": 3.2535,
      "step": 210970
    },
    {
      "epoch": 1.11916756552311,
      "grad_norm": 0.37203073501586914,
      "learning_rate": 0.00016066671886768533,
      "loss": 3.3686,
      "step": 210980
    },
    {
      "epoch": 1.1192421384677427,
      "grad_norm": 0.39743301272392273,
      "learning_rate": 0.00016066485454559014,
      "loss": 3.3332,
      "step": 210990
    },
    {
      "epoch": 1.1193167114123754,
      "grad_norm": 0.37646159529685974,
      "learning_rate": 0.00016066299022349494,
      "loss": 3.3525,
      "step": 211000
    },
    {
      "epoch": 1.1193912843570082,
      "grad_norm": 0.38553303480148315,
      "learning_rate": 0.00016066112590139975,
      "loss": 3.334,
      "step": 211010
    },
    {
      "epoch": 1.119465857301641,
      "grad_norm": 0.41039398312568665,
      "learning_rate": 0.00016065926157930453,
      "loss": 3.301,
      "step": 211020
    },
    {
      "epoch": 1.1195404302462737,
      "grad_norm": 0.3806004822254181,
      "learning_rate": 0.00016065739725720934,
      "loss": 3.2846,
      "step": 211030
    },
    {
      "epoch": 1.1196150031909065,
      "grad_norm": 0.3819076120853424,
      "learning_rate": 0.00016065553293511414,
      "loss": 3.3058,
      "step": 211040
    },
    {
      "epoch": 1.1196895761355392,
      "grad_norm": 0.3703717887401581,
      "learning_rate": 0.00016065366861301895,
      "loss": 3.312,
      "step": 211050
    },
    {
      "epoch": 1.1197641490801717,
      "grad_norm": 0.3859030604362488,
      "learning_rate": 0.00016065180429092376,
      "loss": 3.2931,
      "step": 211060
    },
    {
      "epoch": 1.1198387220248045,
      "grad_norm": 0.3654191792011261,
      "learning_rate": 0.00016064993996882853,
      "loss": 3.2688,
      "step": 211070
    },
    {
      "epoch": 1.1199132949694373,
      "grad_norm": 0.39000919461250305,
      "learning_rate": 0.00016064807564673334,
      "loss": 3.3671,
      "step": 211080
    },
    {
      "epoch": 1.11998786791407,
      "grad_norm": 0.3617355227470398,
      "learning_rate": 0.00016064621132463815,
      "loss": 3.2957,
      "step": 211090
    },
    {
      "epoch": 1.1200624408587028,
      "grad_norm": 0.34127742052078247,
      "learning_rate": 0.00016064434700254293,
      "loss": 3.3001,
      "step": 211100
    },
    {
      "epoch": 1.1201370138033355,
      "grad_norm": 0.3844374120235443,
      "learning_rate": 0.00016064248268044776,
      "loss": 3.3048,
      "step": 211110
    },
    {
      "epoch": 1.1202115867479683,
      "grad_norm": 0.38138145208358765,
      "learning_rate": 0.00016064061835835254,
      "loss": 3.311,
      "step": 211120
    },
    {
      "epoch": 1.120286159692601,
      "grad_norm": 0.39409124851226807,
      "learning_rate": 0.00016063875403625735,
      "loss": 3.2733,
      "step": 211130
    },
    {
      "epoch": 1.1203607326372338,
      "grad_norm": 0.3781566023826599,
      "learning_rate": 0.00016063688971416215,
      "loss": 3.2916,
      "step": 211140
    },
    {
      "epoch": 1.1204353055818665,
      "grad_norm": 0.3742274343967438,
      "learning_rate": 0.00016063502539206693,
      "loss": 3.342,
      "step": 211150
    },
    {
      "epoch": 1.1205098785264993,
      "grad_norm": 0.376830130815506,
      "learning_rate": 0.00016063316106997174,
      "loss": 3.2976,
      "step": 211160
    },
    {
      "epoch": 1.1205844514711318,
      "grad_norm": 0.3744322657585144,
      "learning_rate": 0.00016063129674787654,
      "loss": 3.2616,
      "step": 211170
    },
    {
      "epoch": 1.1206590244157646,
      "grad_norm": 0.4109537601470947,
      "learning_rate": 0.00016062943242578135,
      "loss": 3.3097,
      "step": 211180
    },
    {
      "epoch": 1.1207335973603973,
      "grad_norm": 0.38695085048675537,
      "learning_rate": 0.00016062756810368616,
      "loss": 3.2766,
      "step": 211190
    },
    {
      "epoch": 1.12080817030503,
      "grad_norm": 0.38119617104530334,
      "learning_rate": 0.00016062570378159094,
      "loss": 3.3548,
      "step": 211200
    },
    {
      "epoch": 1.1208827432496629,
      "grad_norm": 0.38341984152793884,
      "learning_rate": 0.00016062383945949574,
      "loss": 3.3296,
      "step": 211210
    },
    {
      "epoch": 1.1209573161942956,
      "grad_norm": 0.37625551223754883,
      "learning_rate": 0.00016062197513740055,
      "loss": 3.2437,
      "step": 211220
    },
    {
      "epoch": 1.1210318891389284,
      "grad_norm": 0.3804115355014801,
      "learning_rate": 0.00016062011081530535,
      "loss": 3.2949,
      "step": 211230
    },
    {
      "epoch": 1.1211064620835611,
      "grad_norm": 0.37368202209472656,
      "learning_rate": 0.00016061824649321016,
      "loss": 3.2627,
      "step": 211240
    },
    {
      "epoch": 1.1211810350281939,
      "grad_norm": 0.38700005412101746,
      "learning_rate": 0.00016061638217111494,
      "loss": 3.2357,
      "step": 211250
    },
    {
      "epoch": 1.1212556079728266,
      "grad_norm": 0.36195144057273865,
      "learning_rate": 0.00016061451784901975,
      "loss": 3.2657,
      "step": 211260
    },
    {
      "epoch": 1.1213301809174592,
      "grad_norm": 0.37532684206962585,
      "learning_rate": 0.00016061265352692455,
      "loss": 3.3381,
      "step": 211270
    },
    {
      "epoch": 1.121404753862092,
      "grad_norm": 0.3774757981300354,
      "learning_rate": 0.00016061078920482936,
      "loss": 3.2909,
      "step": 211280
    },
    {
      "epoch": 1.1214793268067247,
      "grad_norm": 0.3886171877384186,
      "learning_rate": 0.00016060892488273416,
      "loss": 3.3071,
      "step": 211290
    },
    {
      "epoch": 1.1215538997513574,
      "grad_norm": 0.4078722894191742,
      "learning_rate": 0.00016060706056063894,
      "loss": 3.28,
      "step": 211300
    },
    {
      "epoch": 1.1216284726959902,
      "grad_norm": 0.3719671666622162,
      "learning_rate": 0.00016060519623854375,
      "loss": 3.3448,
      "step": 211310
    },
    {
      "epoch": 1.121703045640623,
      "grad_norm": 0.3636742830276489,
      "learning_rate": 0.00016060333191644853,
      "loss": 3.2381,
      "step": 211320
    },
    {
      "epoch": 1.1217776185852557,
      "grad_norm": 0.41404980421066284,
      "learning_rate": 0.00016060146759435336,
      "loss": 3.2614,
      "step": 211330
    },
    {
      "epoch": 1.1218521915298885,
      "grad_norm": 0.3574541509151459,
      "learning_rate": 0.00016059960327225817,
      "loss": 3.2664,
      "step": 211340
    },
    {
      "epoch": 1.1219267644745212,
      "grad_norm": 0.3994460999965668,
      "learning_rate": 0.00016059773895016295,
      "loss": 3.3088,
      "step": 211350
    },
    {
      "epoch": 1.122001337419154,
      "grad_norm": 0.39273035526275635,
      "learning_rate": 0.00016059587462806775,
      "loss": 3.2514,
      "step": 211360
    },
    {
      "epoch": 1.1220759103637867,
      "grad_norm": 0.37241044640541077,
      "learning_rate": 0.00016059401030597253,
      "loss": 3.2841,
      "step": 211370
    },
    {
      "epoch": 1.1221504833084193,
      "grad_norm": 0.361427903175354,
      "learning_rate": 0.00016059214598387734,
      "loss": 3.305,
      "step": 211380
    },
    {
      "epoch": 1.122225056253052,
      "grad_norm": 0.39175546169281006,
      "learning_rate": 0.00016059028166178217,
      "loss": 3.3021,
      "step": 211390
    },
    {
      "epoch": 1.1222996291976848,
      "grad_norm": 0.4079131484031677,
      "learning_rate": 0.00016058841733968695,
      "loss": 3.3398,
      "step": 211400
    },
    {
      "epoch": 1.1223742021423175,
      "grad_norm": 0.3690301775932312,
      "learning_rate": 0.00016058655301759176,
      "loss": 3.2471,
      "step": 211410
    },
    {
      "epoch": 1.1224487750869503,
      "grad_norm": 0.410428524017334,
      "learning_rate": 0.00016058468869549654,
      "loss": 3.2876,
      "step": 211420
    },
    {
      "epoch": 1.122523348031583,
      "grad_norm": 0.38095057010650635,
      "learning_rate": 0.00016058282437340134,
      "loss": 3.2936,
      "step": 211430
    },
    {
      "epoch": 1.1225979209762158,
      "grad_norm": 0.3617936372756958,
      "learning_rate": 0.00016058096005130615,
      "loss": 3.2922,
      "step": 211440
    },
    {
      "epoch": 1.1226724939208486,
      "grad_norm": 0.37447798252105713,
      "learning_rate": 0.00016057909572921096,
      "loss": 3.3236,
      "step": 211450
    },
    {
      "epoch": 1.1227470668654813,
      "grad_norm": 0.37515386939048767,
      "learning_rate": 0.00016057723140711576,
      "loss": 3.3292,
      "step": 211460
    },
    {
      "epoch": 1.122821639810114,
      "grad_norm": 0.36762362718582153,
      "learning_rate": 0.00016057536708502054,
      "loss": 3.3252,
      "step": 211470
    },
    {
      "epoch": 1.1228962127547466,
      "grad_norm": 0.37521594762802124,
      "learning_rate": 0.00016057350276292535,
      "loss": 3.298,
      "step": 211480
    },
    {
      "epoch": 1.1229707856993794,
      "grad_norm": 0.3879474103450775,
      "learning_rate": 0.00016057163844083016,
      "loss": 3.3563,
      "step": 211490
    },
    {
      "epoch": 1.1230453586440121,
      "grad_norm": 0.406729131937027,
      "learning_rate": 0.00016056977411873496,
      "loss": 3.2862,
      "step": 211500
    },
    {
      "epoch": 1.1231199315886449,
      "grad_norm": 0.3733818531036377,
      "learning_rate": 0.00016056790979663977,
      "loss": 3.2712,
      "step": 211510
    },
    {
      "epoch": 1.1231945045332776,
      "grad_norm": 0.3576943576335907,
      "learning_rate": 0.00016056604547454455,
      "loss": 3.3864,
      "step": 211520
    },
    {
      "epoch": 1.1232690774779104,
      "grad_norm": 0.42644503712654114,
      "learning_rate": 0.00016056418115244935,
      "loss": 3.2414,
      "step": 211530
    },
    {
      "epoch": 1.1233436504225431,
      "grad_norm": 0.38180452585220337,
      "learning_rate": 0.00016056231683035416,
      "loss": 3.2927,
      "step": 211540
    },
    {
      "epoch": 1.123418223367176,
      "grad_norm": 0.36700180172920227,
      "learning_rate": 0.00016056045250825894,
      "loss": 3.3037,
      "step": 211550
    },
    {
      "epoch": 1.1234927963118087,
      "grad_norm": 0.38162970542907715,
      "learning_rate": 0.00016055858818616377,
      "loss": 3.2639,
      "step": 211560
    },
    {
      "epoch": 1.1235673692564414,
      "grad_norm": 0.37514519691467285,
      "learning_rate": 0.00016055672386406855,
      "loss": 3.3018,
      "step": 211570
    },
    {
      "epoch": 1.1236419422010742,
      "grad_norm": 0.4093427062034607,
      "learning_rate": 0.00016055485954197336,
      "loss": 3.2663,
      "step": 211580
    },
    {
      "epoch": 1.1237165151457067,
      "grad_norm": 0.37856268882751465,
      "learning_rate": 0.00016055299521987816,
      "loss": 3.3473,
      "step": 211590
    },
    {
      "epoch": 1.1237910880903395,
      "grad_norm": 0.3845929503440857,
      "learning_rate": 0.00016055113089778294,
      "loss": 3.3341,
      "step": 211600
    },
    {
      "epoch": 1.1238656610349722,
      "grad_norm": 0.3840566873550415,
      "learning_rate": 0.00016054926657568775,
      "loss": 3.3331,
      "step": 211610
    },
    {
      "epoch": 1.123940233979605,
      "grad_norm": 0.38024038076400757,
      "learning_rate": 0.00016054740225359256,
      "loss": 3.2879,
      "step": 211620
    },
    {
      "epoch": 1.1240148069242377,
      "grad_norm": 0.3805463910102844,
      "learning_rate": 0.00016054553793149736,
      "loss": 3.3293,
      "step": 211630
    },
    {
      "epoch": 1.1240893798688705,
      "grad_norm": 0.3809668719768524,
      "learning_rate": 0.00016054367360940217,
      "loss": 3.3064,
      "step": 211640
    },
    {
      "epoch": 1.1241639528135032,
      "grad_norm": 0.3945772647857666,
      "learning_rate": 0.00016054180928730695,
      "loss": 3.2921,
      "step": 211650
    },
    {
      "epoch": 1.124238525758136,
      "grad_norm": 0.42195233702659607,
      "learning_rate": 0.00016053994496521175,
      "loss": 3.2771,
      "step": 211660
    },
    {
      "epoch": 1.1243130987027687,
      "grad_norm": 0.3987444043159485,
      "learning_rate": 0.00016053808064311656,
      "loss": 3.2568,
      "step": 211670
    },
    {
      "epoch": 1.1243876716474015,
      "grad_norm": 0.38801443576812744,
      "learning_rate": 0.00016053621632102137,
      "loss": 3.2368,
      "step": 211680
    },
    {
      "epoch": 1.124462244592034,
      "grad_norm": 0.3997238874435425,
      "learning_rate": 0.00016053435199892617,
      "loss": 3.3506,
      "step": 211690
    },
    {
      "epoch": 1.1245368175366668,
      "grad_norm": 0.3919489085674286,
      "learning_rate": 0.00016053248767683095,
      "loss": 3.3244,
      "step": 211700
    },
    {
      "epoch": 1.1246113904812995,
      "grad_norm": 0.3806440532207489,
      "learning_rate": 0.00016053062335473576,
      "loss": 3.3606,
      "step": 211710
    },
    {
      "epoch": 1.1246859634259323,
      "grad_norm": 0.36929914355278015,
      "learning_rate": 0.00016052875903264057,
      "loss": 3.3433,
      "step": 211720
    },
    {
      "epoch": 1.124760536370565,
      "grad_norm": 0.3835015296936035,
      "learning_rate": 0.00016052689471054537,
      "loss": 3.329,
      "step": 211730
    },
    {
      "epoch": 1.1248351093151978,
      "grad_norm": 0.329230397939682,
      "learning_rate": 0.00016052503038845018,
      "loss": 3.3449,
      "step": 211740
    },
    {
      "epoch": 1.1249096822598306,
      "grad_norm": 0.3589043617248535,
      "learning_rate": 0.00016052316606635496,
      "loss": 3.2736,
      "step": 211750
    },
    {
      "epoch": 1.1249842552044633,
      "grad_norm": 0.3789916932582855,
      "learning_rate": 0.00016052130174425976,
      "loss": 3.3919,
      "step": 211760
    },
    {
      "epoch": 1.125058828149096,
      "grad_norm": 0.3756532669067383,
      "learning_rate": 0.00016051943742216454,
      "loss": 3.2561,
      "step": 211770
    },
    {
      "epoch": 1.1251334010937288,
      "grad_norm": 0.3527578115463257,
      "learning_rate": 0.00016051757310006938,
      "loss": 3.2235,
      "step": 211780
    },
    {
      "epoch": 1.1252079740383616,
      "grad_norm": 0.3880227506160736,
      "learning_rate": 0.00016051570877797416,
      "loss": 3.3215,
      "step": 211790
    },
    {
      "epoch": 1.1252825469829943,
      "grad_norm": 0.3604639172554016,
      "learning_rate": 0.00016051384445587896,
      "loss": 3.3229,
      "step": 211800
    },
    {
      "epoch": 1.1253571199276269,
      "grad_norm": 0.3754532039165497,
      "learning_rate": 0.00016051198013378377,
      "loss": 3.2996,
      "step": 211810
    },
    {
      "epoch": 1.1254316928722596,
      "grad_norm": 0.3783421516418457,
      "learning_rate": 0.00016051011581168855,
      "loss": 3.3134,
      "step": 211820
    },
    {
      "epoch": 1.1255062658168924,
      "grad_norm": 0.41561374068260193,
      "learning_rate": 0.00016050825148959335,
      "loss": 3.2792,
      "step": 211830
    },
    {
      "epoch": 1.1255808387615251,
      "grad_norm": 0.4509342312812805,
      "learning_rate": 0.00016050638716749816,
      "loss": 3.3094,
      "step": 211840
    },
    {
      "epoch": 1.125655411706158,
      "grad_norm": 0.39399078488349915,
      "learning_rate": 0.00016050452284540297,
      "loss": 3.3025,
      "step": 211850
    },
    {
      "epoch": 1.1257299846507907,
      "grad_norm": 0.3720542788505554,
      "learning_rate": 0.00016050265852330777,
      "loss": 3.2795,
      "step": 211860
    },
    {
      "epoch": 1.1258045575954234,
      "grad_norm": 0.3811119496822357,
      "learning_rate": 0.00016050079420121255,
      "loss": 3.3411,
      "step": 211870
    },
    {
      "epoch": 1.1258791305400562,
      "grad_norm": 0.37685921788215637,
      "learning_rate": 0.00016049892987911736,
      "loss": 3.3558,
      "step": 211880
    },
    {
      "epoch": 1.125953703484689,
      "grad_norm": 0.37019771337509155,
      "learning_rate": 0.00016049706555702216,
      "loss": 3.3065,
      "step": 211890
    },
    {
      "epoch": 1.1260282764293215,
      "grad_norm": 0.3743123412132263,
      "learning_rate": 0.00016049520123492697,
      "loss": 3.3847,
      "step": 211900
    },
    {
      "epoch": 1.1261028493739542,
      "grad_norm": 0.37972167134284973,
      "learning_rate": 0.00016049333691283178,
      "loss": 3.2542,
      "step": 211910
    },
    {
      "epoch": 1.126177422318587,
      "grad_norm": 0.3850686550140381,
      "learning_rate": 0.00016049147259073656,
      "loss": 3.281,
      "step": 211920
    },
    {
      "epoch": 1.1262519952632197,
      "grad_norm": 0.3684253394603729,
      "learning_rate": 0.00016048960826864136,
      "loss": 3.3251,
      "step": 211930
    },
    {
      "epoch": 1.1263265682078525,
      "grad_norm": 0.38655251264572144,
      "learning_rate": 0.00016048774394654614,
      "loss": 3.3616,
      "step": 211940
    },
    {
      "epoch": 1.1264011411524852,
      "grad_norm": 0.3495771884918213,
      "learning_rate": 0.00016048587962445097,
      "loss": 3.2888,
      "step": 211950
    },
    {
      "epoch": 1.126475714097118,
      "grad_norm": 0.385503351688385,
      "learning_rate": 0.00016048401530235578,
      "loss": 3.3426,
      "step": 211960
    },
    {
      "epoch": 1.1265502870417508,
      "grad_norm": 0.38100624084472656,
      "learning_rate": 0.00016048215098026056,
      "loss": 3.3263,
      "step": 211970
    },
    {
      "epoch": 1.1266248599863835,
      "grad_norm": 0.3786214590072632,
      "learning_rate": 0.00016048028665816537,
      "loss": 3.3158,
      "step": 211980
    },
    {
      "epoch": 1.1266994329310163,
      "grad_norm": 0.39539673924446106,
      "learning_rate": 0.00016047842233607015,
      "loss": 3.3231,
      "step": 211990
    },
    {
      "epoch": 1.126774005875649,
      "grad_norm": 0.4011882245540619,
      "learning_rate": 0.00016047655801397495,
      "loss": 3.2773,
      "step": 212000
    },
    {
      "epoch": 1.1268485788202816,
      "grad_norm": 0.38649681210517883,
      "learning_rate": 0.00016047469369187979,
      "loss": 3.2845,
      "step": 212010
    },
    {
      "epoch": 1.1269231517649143,
      "grad_norm": 0.3596334755420685,
      "learning_rate": 0.00016047282936978456,
      "loss": 3.2642,
      "step": 212020
    },
    {
      "epoch": 1.126997724709547,
      "grad_norm": 0.37214940786361694,
      "learning_rate": 0.00016047096504768937,
      "loss": 3.2803,
      "step": 212030
    },
    {
      "epoch": 1.1270722976541798,
      "grad_norm": 0.3748602271080017,
      "learning_rate": 0.00016046910072559415,
      "loss": 3.3259,
      "step": 212040
    },
    {
      "epoch": 1.1271468705988126,
      "grad_norm": 0.39249199628829956,
      "learning_rate": 0.00016046723640349896,
      "loss": 3.3124,
      "step": 212050
    },
    {
      "epoch": 1.1272214435434453,
      "grad_norm": 0.3661969304084778,
      "learning_rate": 0.00016046537208140376,
      "loss": 3.33,
      "step": 212060
    },
    {
      "epoch": 1.127296016488078,
      "grad_norm": 0.3810150921344757,
      "learning_rate": 0.00016046350775930857,
      "loss": 3.3764,
      "step": 212070
    },
    {
      "epoch": 1.1273705894327108,
      "grad_norm": 0.4018070101737976,
      "learning_rate": 0.00016046164343721338,
      "loss": 3.3683,
      "step": 212080
    },
    {
      "epoch": 1.1274451623773436,
      "grad_norm": 0.3793647289276123,
      "learning_rate": 0.00016045977911511815,
      "loss": 3.2595,
      "step": 212090
    },
    {
      "epoch": 1.1275197353219764,
      "grad_norm": 0.3946397006511688,
      "learning_rate": 0.00016045791479302296,
      "loss": 3.265,
      "step": 212100
    },
    {
      "epoch": 1.127594308266609,
      "grad_norm": 0.40007129311561584,
      "learning_rate": 0.00016045605047092777,
      "loss": 3.3609,
      "step": 212110
    },
    {
      "epoch": 1.1276688812112416,
      "grad_norm": 0.3846946060657501,
      "learning_rate": 0.00016045418614883257,
      "loss": 3.2473,
      "step": 212120
    },
    {
      "epoch": 1.1277434541558744,
      "grad_norm": 0.3763236701488495,
      "learning_rate": 0.00016045232182673738,
      "loss": 3.2639,
      "step": 212130
    },
    {
      "epoch": 1.1278180271005072,
      "grad_norm": 0.37458252906799316,
      "learning_rate": 0.00016045045750464216,
      "loss": 3.3195,
      "step": 212140
    },
    {
      "epoch": 1.12789260004514,
      "grad_norm": 0.3844352960586548,
      "learning_rate": 0.00016044859318254697,
      "loss": 3.2495,
      "step": 212150
    },
    {
      "epoch": 1.1279671729897727,
      "grad_norm": 0.3741457760334015,
      "learning_rate": 0.00016044672886045177,
      "loss": 3.3028,
      "step": 212160
    },
    {
      "epoch": 1.1280417459344054,
      "grad_norm": 0.40073302388191223,
      "learning_rate": 0.00016044486453835658,
      "loss": 3.2404,
      "step": 212170
    },
    {
      "epoch": 1.1281163188790382,
      "grad_norm": 0.40169358253479004,
      "learning_rate": 0.00016044300021626138,
      "loss": 3.294,
      "step": 212180
    },
    {
      "epoch": 1.128190891823671,
      "grad_norm": 0.39229053258895874,
      "learning_rate": 0.00016044113589416616,
      "loss": 3.3636,
      "step": 212190
    },
    {
      "epoch": 1.1282654647683037,
      "grad_norm": 0.36399221420288086,
      "learning_rate": 0.00016043927157207097,
      "loss": 3.295,
      "step": 212200
    },
    {
      "epoch": 1.1283400377129365,
      "grad_norm": 0.365318238735199,
      "learning_rate": 0.00016043740724997578,
      "loss": 3.2902,
      "step": 212210
    },
    {
      "epoch": 1.128414610657569,
      "grad_norm": 0.41508322954177856,
      "learning_rate": 0.00016043554292788056,
      "loss": 3.2836,
      "step": 212220
    },
    {
      "epoch": 1.1284891836022017,
      "grad_norm": 0.5086601376533508,
      "learning_rate": 0.0001604336786057854,
      "loss": 3.2961,
      "step": 212230
    },
    {
      "epoch": 1.1285637565468345,
      "grad_norm": 0.38650140166282654,
      "learning_rate": 0.00016043181428369017,
      "loss": 3.2985,
      "step": 212240
    },
    {
      "epoch": 1.1286383294914673,
      "grad_norm": 0.4195167124271393,
      "learning_rate": 0.00016042994996159497,
      "loss": 3.3365,
      "step": 212250
    },
    {
      "epoch": 1.1287129024361,
      "grad_norm": 0.36093735694885254,
      "learning_rate": 0.00016042808563949978,
      "loss": 3.2743,
      "step": 212260
    },
    {
      "epoch": 1.1287874753807328,
      "grad_norm": 0.3801356554031372,
      "learning_rate": 0.00016042622131740456,
      "loss": 3.3606,
      "step": 212270
    },
    {
      "epoch": 1.1288620483253655,
      "grad_norm": 0.3544180989265442,
      "learning_rate": 0.00016042435699530937,
      "loss": 3.3268,
      "step": 212280
    },
    {
      "epoch": 1.1289366212699983,
      "grad_norm": 0.36711061000823975,
      "learning_rate": 0.00016042249267321417,
      "loss": 3.2879,
      "step": 212290
    },
    {
      "epoch": 1.129011194214631,
      "grad_norm": 0.36763080954551697,
      "learning_rate": 0.00016042062835111898,
      "loss": 3.3093,
      "step": 212300
    },
    {
      "epoch": 1.1290857671592638,
      "grad_norm": 0.3758286237716675,
      "learning_rate": 0.00016041876402902379,
      "loss": 3.3281,
      "step": 212310
    },
    {
      "epoch": 1.1291603401038963,
      "grad_norm": 0.38582465052604675,
      "learning_rate": 0.00016041689970692856,
      "loss": 3.2699,
      "step": 212320
    },
    {
      "epoch": 1.129234913048529,
      "grad_norm": 0.36518439650535583,
      "learning_rate": 0.00016041503538483337,
      "loss": 3.2929,
      "step": 212330
    },
    {
      "epoch": 1.1293094859931618,
      "grad_norm": 0.37727752327919006,
      "learning_rate": 0.00016041317106273818,
      "loss": 3.2645,
      "step": 212340
    },
    {
      "epoch": 1.1293840589377946,
      "grad_norm": 0.3942629396915436,
      "learning_rate": 0.00016041130674064298,
      "loss": 3.304,
      "step": 212350
    },
    {
      "epoch": 1.1294586318824273,
      "grad_norm": 0.3705013692378998,
      "learning_rate": 0.0001604094424185478,
      "loss": 3.2408,
      "step": 212360
    },
    {
      "epoch": 1.12953320482706,
      "grad_norm": 0.3884001672267914,
      "learning_rate": 0.00016040757809645257,
      "loss": 3.2987,
      "step": 212370
    },
    {
      "epoch": 1.1296077777716929,
      "grad_norm": 0.36691412329673767,
      "learning_rate": 0.00016040571377435737,
      "loss": 3.2802,
      "step": 212380
    },
    {
      "epoch": 1.1296823507163256,
      "grad_norm": 0.37581002712249756,
      "learning_rate": 0.00016040384945226215,
      "loss": 3.2684,
      "step": 212390
    },
    {
      "epoch": 1.1297569236609584,
      "grad_norm": 0.3775714635848999,
      "learning_rate": 0.000160401985130167,
      "loss": 3.3532,
      "step": 212400
    },
    {
      "epoch": 1.1298314966055911,
      "grad_norm": 0.3787691593170166,
      "learning_rate": 0.0001604001208080718,
      "loss": 3.2974,
      "step": 212410
    },
    {
      "epoch": 1.1299060695502239,
      "grad_norm": 0.3766005337238312,
      "learning_rate": 0.00016039825648597657,
      "loss": 3.3476,
      "step": 212420
    },
    {
      "epoch": 1.1299806424948564,
      "grad_norm": 0.3633253872394562,
      "learning_rate": 0.00016039639216388138,
      "loss": 3.3067,
      "step": 212430
    },
    {
      "epoch": 1.1300552154394892,
      "grad_norm": 0.3794480264186859,
      "learning_rate": 0.00016039452784178616,
      "loss": 3.2673,
      "step": 212440
    },
    {
      "epoch": 1.130129788384122,
      "grad_norm": 0.381497859954834,
      "learning_rate": 0.00016039266351969096,
      "loss": 3.2583,
      "step": 212450
    },
    {
      "epoch": 1.1302043613287547,
      "grad_norm": 0.3881550431251526,
      "learning_rate": 0.0001603907991975958,
      "loss": 3.2542,
      "step": 212460
    },
    {
      "epoch": 1.1302789342733874,
      "grad_norm": 0.3652009665966034,
      "learning_rate": 0.00016038893487550058,
      "loss": 3.2858,
      "step": 212470
    },
    {
      "epoch": 1.1303535072180202,
      "grad_norm": 0.3570718467235565,
      "learning_rate": 0.00016038707055340538,
      "loss": 3.3082,
      "step": 212480
    },
    {
      "epoch": 1.130428080162653,
      "grad_norm": 0.3551473617553711,
      "learning_rate": 0.00016038520623131016,
      "loss": 3.3261,
      "step": 212490
    },
    {
      "epoch": 1.1305026531072857,
      "grad_norm": 0.3770352303981781,
      "learning_rate": 0.00016038334190921497,
      "loss": 3.321,
      "step": 212500
    },
    {
      "epoch": 1.1305772260519185,
      "grad_norm": 0.3739619851112366,
      "learning_rate": 0.00016038147758711978,
      "loss": 3.317,
      "step": 212510
    },
    {
      "epoch": 1.130651798996551,
      "grad_norm": 0.3837162256240845,
      "learning_rate": 0.00016037961326502458,
      "loss": 3.286,
      "step": 212520
    },
    {
      "epoch": 1.1307263719411837,
      "grad_norm": 0.35790666937828064,
      "learning_rate": 0.0001603777489429294,
      "loss": 3.3185,
      "step": 212530
    },
    {
      "epoch": 1.1308009448858165,
      "grad_norm": 0.3948546051979065,
      "learning_rate": 0.00016037588462083417,
      "loss": 3.2931,
      "step": 212540
    },
    {
      "epoch": 1.1308755178304493,
      "grad_norm": 0.40155622363090515,
      "learning_rate": 0.00016037402029873897,
      "loss": 3.2403,
      "step": 212550
    },
    {
      "epoch": 1.130950090775082,
      "grad_norm": 0.4119776785373688,
      "learning_rate": 0.00016037215597664378,
      "loss": 3.2449,
      "step": 212560
    },
    {
      "epoch": 1.1310246637197148,
      "grad_norm": 0.3825076222419739,
      "learning_rate": 0.00016037029165454859,
      "loss": 3.3389,
      "step": 212570
    },
    {
      "epoch": 1.1310992366643475,
      "grad_norm": 0.3788303732872009,
      "learning_rate": 0.0001603684273324534,
      "loss": 3.3345,
      "step": 212580
    },
    {
      "epoch": 1.1311738096089803,
      "grad_norm": 0.3608701229095459,
      "learning_rate": 0.00016036656301035817,
      "loss": 3.2957,
      "step": 212590
    },
    {
      "epoch": 1.131248382553613,
      "grad_norm": 0.3796294927597046,
      "learning_rate": 0.00016036469868826298,
      "loss": 3.3382,
      "step": 212600
    },
    {
      "epoch": 1.1313229554982458,
      "grad_norm": 0.41076332330703735,
      "learning_rate": 0.00016036283436616778,
      "loss": 3.314,
      "step": 212610
    },
    {
      "epoch": 1.1313975284428786,
      "grad_norm": 0.3840809762477875,
      "learning_rate": 0.0001603609700440726,
      "loss": 3.2554,
      "step": 212620
    },
    {
      "epoch": 1.1314721013875113,
      "grad_norm": 0.38322243094444275,
      "learning_rate": 0.0001603591057219774,
      "loss": 3.3506,
      "step": 212630
    },
    {
      "epoch": 1.1315466743321438,
      "grad_norm": 0.3826448619365692,
      "learning_rate": 0.00016035724139988218,
      "loss": 3.3443,
      "step": 212640
    },
    {
      "epoch": 1.1316212472767766,
      "grad_norm": 0.40047726035118103,
      "learning_rate": 0.00016035537707778698,
      "loss": 3.2913,
      "step": 212650
    },
    {
      "epoch": 1.1316958202214094,
      "grad_norm": 0.35983216762542725,
      "learning_rate": 0.0001603535127556918,
      "loss": 3.3504,
      "step": 212660
    },
    {
      "epoch": 1.131770393166042,
      "grad_norm": 0.411259263753891,
      "learning_rate": 0.00016035164843359657,
      "loss": 3.3349,
      "step": 212670
    },
    {
      "epoch": 1.1318449661106749,
      "grad_norm": 0.39135342836380005,
      "learning_rate": 0.0001603497841115014,
      "loss": 3.3362,
      "step": 212680
    },
    {
      "epoch": 1.1319195390553076,
      "grad_norm": 0.3635160028934479,
      "learning_rate": 0.00016034791978940618,
      "loss": 3.2867,
      "step": 212690
    },
    {
      "epoch": 1.1319941119999404,
      "grad_norm": 0.3878104090690613,
      "learning_rate": 0.000160346055467311,
      "loss": 3.2384,
      "step": 212700
    },
    {
      "epoch": 1.1320686849445731,
      "grad_norm": 0.36755380034446716,
      "learning_rate": 0.0001603441911452158,
      "loss": 3.2974,
      "step": 212710
    },
    {
      "epoch": 1.132143257889206,
      "grad_norm": 0.38483381271362305,
      "learning_rate": 0.00016034232682312057,
      "loss": 3.2617,
      "step": 212720
    },
    {
      "epoch": 1.1322178308338384,
      "grad_norm": 0.3897824287414551,
      "learning_rate": 0.00016034046250102538,
      "loss": 3.2891,
      "step": 212730
    },
    {
      "epoch": 1.1322924037784712,
      "grad_norm": 0.4027535915374756,
      "learning_rate": 0.00016033859817893019,
      "loss": 3.311,
      "step": 212740
    },
    {
      "epoch": 1.132366976723104,
      "grad_norm": 0.3706211447715759,
      "learning_rate": 0.000160336733856835,
      "loss": 3.2835,
      "step": 212750
    },
    {
      "epoch": 1.1324415496677367,
      "grad_norm": 0.36997005343437195,
      "learning_rate": 0.0001603348695347398,
      "loss": 3.2733,
      "step": 212760
    },
    {
      "epoch": 1.1325161226123694,
      "grad_norm": 0.366822212934494,
      "learning_rate": 0.00016033300521264458,
      "loss": 3.2417,
      "step": 212770
    },
    {
      "epoch": 1.1325906955570022,
      "grad_norm": 0.4049760103225708,
      "learning_rate": 0.00016033114089054938,
      "loss": 3.2131,
      "step": 212780
    },
    {
      "epoch": 1.132665268501635,
      "grad_norm": 0.3657650351524353,
      "learning_rate": 0.0001603292765684542,
      "loss": 3.3308,
      "step": 212790
    },
    {
      "epoch": 1.1327398414462677,
      "grad_norm": 0.39916715025901794,
      "learning_rate": 0.000160327412246359,
      "loss": 3.2811,
      "step": 212800
    },
    {
      "epoch": 1.1328144143909005,
      "grad_norm": 0.39722931385040283,
      "learning_rate": 0.0001603255479242638,
      "loss": 3.2662,
      "step": 212810
    },
    {
      "epoch": 1.1328889873355332,
      "grad_norm": 0.3996585011482239,
      "learning_rate": 0.00016032368360216858,
      "loss": 3.3016,
      "step": 212820
    },
    {
      "epoch": 1.132963560280166,
      "grad_norm": 0.3539544939994812,
      "learning_rate": 0.0001603218192800734,
      "loss": 3.3419,
      "step": 212830
    },
    {
      "epoch": 1.1330381332247987,
      "grad_norm": 0.3899043798446655,
      "learning_rate": 0.00016031995495797817,
      "loss": 3.2435,
      "step": 212840
    },
    {
      "epoch": 1.1331127061694313,
      "grad_norm": 0.400189071893692,
      "learning_rate": 0.000160318090635883,
      "loss": 3.3356,
      "step": 212850
    },
    {
      "epoch": 1.133187279114064,
      "grad_norm": 0.38665467500686646,
      "learning_rate": 0.0001603162263137878,
      "loss": 3.3621,
      "step": 212860
    },
    {
      "epoch": 1.1332618520586968,
      "grad_norm": 0.389278769493103,
      "learning_rate": 0.00016031436199169259,
      "loss": 3.2195,
      "step": 212870
    },
    {
      "epoch": 1.1333364250033295,
      "grad_norm": 0.3846675753593445,
      "learning_rate": 0.0001603124976695974,
      "loss": 3.2773,
      "step": 212880
    },
    {
      "epoch": 1.1334109979479623,
      "grad_norm": 0.3937523663043976,
      "learning_rate": 0.00016031063334750217,
      "loss": 3.3242,
      "step": 212890
    },
    {
      "epoch": 1.133485570892595,
      "grad_norm": 0.3865587115287781,
      "learning_rate": 0.00016030876902540698,
      "loss": 3.3566,
      "step": 212900
    },
    {
      "epoch": 1.1335601438372278,
      "grad_norm": 0.39912763237953186,
      "learning_rate": 0.0001603069047033118,
      "loss": 3.3368,
      "step": 212910
    },
    {
      "epoch": 1.1336347167818606,
      "grad_norm": 0.3954735994338989,
      "learning_rate": 0.0001603050403812166,
      "loss": 3.3332,
      "step": 212920
    },
    {
      "epoch": 1.1337092897264933,
      "grad_norm": 0.3708229959011078,
      "learning_rate": 0.0001603031760591214,
      "loss": 3.2721,
      "step": 212930
    },
    {
      "epoch": 1.1337838626711259,
      "grad_norm": 0.3891301453113556,
      "learning_rate": 0.00016030131173702618,
      "loss": 3.3176,
      "step": 212940
    },
    {
      "epoch": 1.1338584356157586,
      "grad_norm": 0.3648562729358673,
      "learning_rate": 0.00016029944741493098,
      "loss": 3.2194,
      "step": 212950
    },
    {
      "epoch": 1.1339330085603914,
      "grad_norm": 0.3677641451358795,
      "learning_rate": 0.0001602975830928358,
      "loss": 3.2999,
      "step": 212960
    },
    {
      "epoch": 1.1340075815050241,
      "grad_norm": 0.39291703701019287,
      "learning_rate": 0.0001602957187707406,
      "loss": 3.3433,
      "step": 212970
    },
    {
      "epoch": 1.1340821544496569,
      "grad_norm": 0.3814753293991089,
      "learning_rate": 0.0001602938544486454,
      "loss": 3.3012,
      "step": 212980
    },
    {
      "epoch": 1.1341567273942896,
      "grad_norm": 0.376116544008255,
      "learning_rate": 0.00016029199012655018,
      "loss": 3.2462,
      "step": 212990
    },
    {
      "epoch": 1.1342313003389224,
      "grad_norm": 0.4002167880535126,
      "learning_rate": 0.000160290125804455,
      "loss": 3.3098,
      "step": 213000
    },
    {
      "epoch": 1.1343058732835551,
      "grad_norm": 0.36146295070648193,
      "learning_rate": 0.0001602882614823598,
      "loss": 3.2941,
      "step": 213010
    },
    {
      "epoch": 1.134380446228188,
      "grad_norm": 0.36297041177749634,
      "learning_rate": 0.0001602863971602646,
      "loss": 3.3507,
      "step": 213020
    },
    {
      "epoch": 1.1344550191728207,
      "grad_norm": 0.3811107873916626,
      "learning_rate": 0.0001602845328381694,
      "loss": 3.2644,
      "step": 213030
    },
    {
      "epoch": 1.1345295921174534,
      "grad_norm": 0.37343165278434753,
      "learning_rate": 0.00016028266851607418,
      "loss": 3.2568,
      "step": 213040
    },
    {
      "epoch": 1.1346041650620862,
      "grad_norm": 0.3599069118499756,
      "learning_rate": 0.000160280804193979,
      "loss": 3.2285,
      "step": 213050
    },
    {
      "epoch": 1.1346787380067187,
      "grad_norm": 0.37770313024520874,
      "learning_rate": 0.0001602789398718838,
      "loss": 3.2562,
      "step": 213060
    },
    {
      "epoch": 1.1347533109513515,
      "grad_norm": 0.3810329735279083,
      "learning_rate": 0.0001602770755497886,
      "loss": 3.3302,
      "step": 213070
    },
    {
      "epoch": 1.1348278838959842,
      "grad_norm": 0.3617894649505615,
      "learning_rate": 0.0001602752112276934,
      "loss": 3.3433,
      "step": 213080
    },
    {
      "epoch": 1.134902456840617,
      "grad_norm": 0.35434287786483765,
      "learning_rate": 0.0001602733469055982,
      "loss": 3.313,
      "step": 213090
    },
    {
      "epoch": 1.1349770297852497,
      "grad_norm": 0.396220862865448,
      "learning_rate": 0.000160271482583503,
      "loss": 3.2369,
      "step": 213100
    },
    {
      "epoch": 1.1350516027298825,
      "grad_norm": 0.3770996332168579,
      "learning_rate": 0.0001602696182614078,
      "loss": 3.302,
      "step": 213110
    },
    {
      "epoch": 1.1351261756745152,
      "grad_norm": 0.3537047803401947,
      "learning_rate": 0.00016026775393931258,
      "loss": 3.3361,
      "step": 213120
    },
    {
      "epoch": 1.135200748619148,
      "grad_norm": 0.423214852809906,
      "learning_rate": 0.00016026588961721741,
      "loss": 3.2933,
      "step": 213130
    },
    {
      "epoch": 1.1352753215637807,
      "grad_norm": 0.3765217065811157,
      "learning_rate": 0.0001602640252951222,
      "loss": 3.2504,
      "step": 213140
    },
    {
      "epoch": 1.1353498945084133,
      "grad_norm": 0.3772292733192444,
      "learning_rate": 0.000160262160973027,
      "loss": 3.2945,
      "step": 213150
    },
    {
      "epoch": 1.135424467453046,
      "grad_norm": 0.38664382696151733,
      "learning_rate": 0.00016026029665093178,
      "loss": 3.3006,
      "step": 213160
    },
    {
      "epoch": 1.1354990403976788,
      "grad_norm": 0.3711012303829193,
      "learning_rate": 0.00016025843232883659,
      "loss": 3.3203,
      "step": 213170
    },
    {
      "epoch": 1.1355736133423115,
      "grad_norm": 0.37937766313552856,
      "learning_rate": 0.0001602565680067414,
      "loss": 3.3084,
      "step": 213180
    },
    {
      "epoch": 1.1356481862869443,
      "grad_norm": 0.38771021366119385,
      "learning_rate": 0.0001602547036846462,
      "loss": 3.3147,
      "step": 213190
    },
    {
      "epoch": 1.135722759231577,
      "grad_norm": 0.4025737941265106,
      "learning_rate": 0.000160252839362551,
      "loss": 3.3229,
      "step": 213200
    },
    {
      "epoch": 1.1357973321762098,
      "grad_norm": 0.3832094669342041,
      "learning_rate": 0.00016025097504045578,
      "loss": 3.3197,
      "step": 213210
    },
    {
      "epoch": 1.1358719051208426,
      "grad_norm": 0.3792611360549927,
      "learning_rate": 0.0001602491107183606,
      "loss": 3.2742,
      "step": 213220
    },
    {
      "epoch": 1.1359464780654753,
      "grad_norm": 0.37642353773117065,
      "learning_rate": 0.0001602472463962654,
      "loss": 3.3415,
      "step": 213230
    },
    {
      "epoch": 1.136021051010108,
      "grad_norm": 0.3736172318458557,
      "learning_rate": 0.0001602453820741702,
      "loss": 3.2588,
      "step": 213240
    },
    {
      "epoch": 1.1360956239547408,
      "grad_norm": 0.3947613835334778,
      "learning_rate": 0.000160243517752075,
      "loss": 3.2765,
      "step": 213250
    },
    {
      "epoch": 1.1361701968993736,
      "grad_norm": 0.3769071102142334,
      "learning_rate": 0.0001602416534299798,
      "loss": 3.3163,
      "step": 213260
    },
    {
      "epoch": 1.1362447698440061,
      "grad_norm": 0.3919846713542938,
      "learning_rate": 0.0001602397891078846,
      "loss": 3.2896,
      "step": 213270
    },
    {
      "epoch": 1.1363193427886389,
      "grad_norm": 0.40625059604644775,
      "learning_rate": 0.0001602379247857894,
      "loss": 3.34,
      "step": 213280
    },
    {
      "epoch": 1.1363939157332716,
      "grad_norm": 0.3824109137058258,
      "learning_rate": 0.00016023606046369418,
      "loss": 3.2713,
      "step": 213290
    },
    {
      "epoch": 1.1364684886779044,
      "grad_norm": 0.36154651641845703,
      "learning_rate": 0.000160234196141599,
      "loss": 3.2879,
      "step": 213300
    },
    {
      "epoch": 1.1365430616225372,
      "grad_norm": 0.3693339228630066,
      "learning_rate": 0.0001602323318195038,
      "loss": 3.2239,
      "step": 213310
    },
    {
      "epoch": 1.13661763456717,
      "grad_norm": 0.3507307767868042,
      "learning_rate": 0.0001602304674974086,
      "loss": 3.3493,
      "step": 213320
    },
    {
      "epoch": 1.1366922075118027,
      "grad_norm": 0.4161408841609955,
      "learning_rate": 0.0001602286031753134,
      "loss": 3.3172,
      "step": 213330
    },
    {
      "epoch": 1.1367667804564354,
      "grad_norm": 0.360556036233902,
      "learning_rate": 0.00016022673885321818,
      "loss": 3.3275,
      "step": 213340
    },
    {
      "epoch": 1.1368413534010682,
      "grad_norm": 0.3780098259449005,
      "learning_rate": 0.000160224874531123,
      "loss": 3.3297,
      "step": 213350
    },
    {
      "epoch": 1.1369159263457007,
      "grad_norm": 0.3397817313671112,
      "learning_rate": 0.0001602230102090278,
      "loss": 3.2991,
      "step": 213360
    },
    {
      "epoch": 1.1369904992903335,
      "grad_norm": 0.37189972400665283,
      "learning_rate": 0.0001602211458869326,
      "loss": 3.2953,
      "step": 213370
    },
    {
      "epoch": 1.1370650722349662,
      "grad_norm": 0.36227983236312866,
      "learning_rate": 0.0001602192815648374,
      "loss": 3.3949,
      "step": 213380
    },
    {
      "epoch": 1.137139645179599,
      "grad_norm": 0.3918282985687256,
      "learning_rate": 0.0001602174172427422,
      "loss": 3.2709,
      "step": 213390
    },
    {
      "epoch": 1.1372142181242317,
      "grad_norm": 0.38142529129981995,
      "learning_rate": 0.000160215552920647,
      "loss": 3.2631,
      "step": 213400
    },
    {
      "epoch": 1.1372887910688645,
      "grad_norm": 0.3896181285381317,
      "learning_rate": 0.0001602136885985518,
      "loss": 3.2908,
      "step": 213410
    },
    {
      "epoch": 1.1373633640134972,
      "grad_norm": 0.36298301815986633,
      "learning_rate": 0.0001602118242764566,
      "loss": 3.3177,
      "step": 213420
    },
    {
      "epoch": 1.13743793695813,
      "grad_norm": 0.369152307510376,
      "learning_rate": 0.00016020995995436141,
      "loss": 3.3014,
      "step": 213430
    },
    {
      "epoch": 1.1375125099027628,
      "grad_norm": 0.39280709624290466,
      "learning_rate": 0.0001602080956322662,
      "loss": 3.2244,
      "step": 213440
    },
    {
      "epoch": 1.1375870828473955,
      "grad_norm": 0.38340815901756287,
      "learning_rate": 0.000160206231310171,
      "loss": 3.2811,
      "step": 213450
    },
    {
      "epoch": 1.1376616557920283,
      "grad_norm": 0.38103970885276794,
      "learning_rate": 0.0001602043669880758,
      "loss": 3.2873,
      "step": 213460
    },
    {
      "epoch": 1.137736228736661,
      "grad_norm": 0.3818908631801605,
      "learning_rate": 0.0001602025026659806,
      "loss": 3.286,
      "step": 213470
    },
    {
      "epoch": 1.1378108016812936,
      "grad_norm": 0.3665548264980316,
      "learning_rate": 0.00016020063834388542,
      "loss": 3.3105,
      "step": 213480
    },
    {
      "epoch": 1.1378853746259263,
      "grad_norm": 0.35391905903816223,
      "learning_rate": 0.0001601987740217902,
      "loss": 3.3468,
      "step": 213490
    },
    {
      "epoch": 1.137959947570559,
      "grad_norm": 0.34399405121803284,
      "learning_rate": 0.000160196909699695,
      "loss": 3.2517,
      "step": 213500
    },
    {
      "epoch": 1.1380345205151918,
      "grad_norm": 0.37382838129997253,
      "learning_rate": 0.00016019504537759978,
      "loss": 3.3179,
      "step": 213510
    },
    {
      "epoch": 1.1381090934598246,
      "grad_norm": 0.38883712887763977,
      "learning_rate": 0.00016019318105550462,
      "loss": 3.268,
      "step": 213520
    },
    {
      "epoch": 1.1381836664044573,
      "grad_norm": 0.3651927709579468,
      "learning_rate": 0.00016019131673340942,
      "loss": 3.3161,
      "step": 213530
    },
    {
      "epoch": 1.13825823934909,
      "grad_norm": 0.36743319034576416,
      "learning_rate": 0.0001601894524113142,
      "loss": 3.3062,
      "step": 213540
    },
    {
      "epoch": 1.1383328122937229,
      "grad_norm": 0.3700407147407532,
      "learning_rate": 0.000160187588089219,
      "loss": 3.2896,
      "step": 213550
    },
    {
      "epoch": 1.1384073852383556,
      "grad_norm": 0.37709468603134155,
      "learning_rate": 0.0001601857237671238,
      "loss": 3.2994,
      "step": 213560
    },
    {
      "epoch": 1.1384819581829881,
      "grad_norm": 0.34138235449790955,
      "learning_rate": 0.0001601838594450286,
      "loss": 3.3277,
      "step": 213570
    },
    {
      "epoch": 1.138556531127621,
      "grad_norm": 0.37288203835487366,
      "learning_rate": 0.00016018199512293343,
      "loss": 3.343,
      "step": 213580
    },
    {
      "epoch": 1.1386311040722537,
      "grad_norm": 0.3845250904560089,
      "learning_rate": 0.0001601801308008382,
      "loss": 3.2389,
      "step": 213590
    },
    {
      "epoch": 1.1387056770168864,
      "grad_norm": 0.39195534586906433,
      "learning_rate": 0.000160178266478743,
      "loss": 3.1978,
      "step": 213600
    },
    {
      "epoch": 1.1387802499615192,
      "grad_norm": 0.35520607233047485,
      "learning_rate": 0.0001601764021566478,
      "loss": 3.2846,
      "step": 213610
    },
    {
      "epoch": 1.138854822906152,
      "grad_norm": 0.3699495196342468,
      "learning_rate": 0.0001601745378345526,
      "loss": 3.321,
      "step": 213620
    },
    {
      "epoch": 1.1389293958507847,
      "grad_norm": 0.3764686584472656,
      "learning_rate": 0.0001601726735124574,
      "loss": 3.3247,
      "step": 213630
    },
    {
      "epoch": 1.1390039687954174,
      "grad_norm": 0.6568909883499146,
      "learning_rate": 0.0001601708091903622,
      "loss": 3.2704,
      "step": 213640
    },
    {
      "epoch": 1.1390785417400502,
      "grad_norm": 0.35664165019989014,
      "learning_rate": 0.00016016894486826702,
      "loss": 3.254,
      "step": 213650
    },
    {
      "epoch": 1.139153114684683,
      "grad_norm": 0.36895251274108887,
      "learning_rate": 0.0001601670805461718,
      "loss": 3.3308,
      "step": 213660
    },
    {
      "epoch": 1.1392276876293157,
      "grad_norm": 0.35411813855171204,
      "learning_rate": 0.0001601652162240766,
      "loss": 3.3151,
      "step": 213670
    },
    {
      "epoch": 1.1393022605739485,
      "grad_norm": 0.3786846995353699,
      "learning_rate": 0.0001601633519019814,
      "loss": 3.2732,
      "step": 213680
    },
    {
      "epoch": 1.139376833518581,
      "grad_norm": 0.36534878611564636,
      "learning_rate": 0.00016016148757988622,
      "loss": 3.2978,
      "step": 213690
    },
    {
      "epoch": 1.1394514064632137,
      "grad_norm": 0.37328076362609863,
      "learning_rate": 0.00016015962325779102,
      "loss": 3.3043,
      "step": 213700
    },
    {
      "epoch": 1.1395259794078465,
      "grad_norm": 0.4181976914405823,
      "learning_rate": 0.0001601577589356958,
      "loss": 3.2584,
      "step": 213710
    },
    {
      "epoch": 1.1396005523524793,
      "grad_norm": 0.377590537071228,
      "learning_rate": 0.0001601558946136006,
      "loss": 3.255,
      "step": 213720
    },
    {
      "epoch": 1.139675125297112,
      "grad_norm": 0.35301682353019714,
      "learning_rate": 0.0001601540302915054,
      "loss": 3.2363,
      "step": 213730
    },
    {
      "epoch": 1.1397496982417448,
      "grad_norm": 0.39105087518692017,
      "learning_rate": 0.0001601521659694102,
      "loss": 3.3324,
      "step": 213740
    },
    {
      "epoch": 1.1398242711863775,
      "grad_norm": 0.3758735954761505,
      "learning_rate": 0.00016015030164731503,
      "loss": 3.2866,
      "step": 213750
    },
    {
      "epoch": 1.1398988441310103,
      "grad_norm": 0.40227681398391724,
      "learning_rate": 0.0001601484373252198,
      "loss": 3.2608,
      "step": 213760
    },
    {
      "epoch": 1.139973417075643,
      "grad_norm": 0.3882773518562317,
      "learning_rate": 0.0001601465730031246,
      "loss": 3.2965,
      "step": 213770
    },
    {
      "epoch": 1.1400479900202756,
      "grad_norm": 0.3785251975059509,
      "learning_rate": 0.00016014470868102942,
      "loss": 3.3166,
      "step": 213780
    },
    {
      "epoch": 1.1401225629649083,
      "grad_norm": 0.3797489404678345,
      "learning_rate": 0.0001601428443589342,
      "loss": 3.26,
      "step": 213790
    },
    {
      "epoch": 1.140197135909541,
      "grad_norm": 0.363781213760376,
      "learning_rate": 0.000160140980036839,
      "loss": 3.2517,
      "step": 213800
    },
    {
      "epoch": 1.1402717088541738,
      "grad_norm": 0.38426724076271057,
      "learning_rate": 0.0001601391157147438,
      "loss": 3.2246,
      "step": 213810
    },
    {
      "epoch": 1.1403462817988066,
      "grad_norm": 0.40827226638793945,
      "learning_rate": 0.00016013725139264862,
      "loss": 3.2224,
      "step": 213820
    },
    {
      "epoch": 1.1404208547434393,
      "grad_norm": 0.3795827031135559,
      "learning_rate": 0.00016013538707055342,
      "loss": 3.287,
      "step": 213830
    },
    {
      "epoch": 1.140495427688072,
      "grad_norm": 0.36654216051101685,
      "learning_rate": 0.0001601335227484582,
      "loss": 3.2856,
      "step": 213840
    },
    {
      "epoch": 1.1405700006327049,
      "grad_norm": 0.3805990219116211,
      "learning_rate": 0.000160131658426363,
      "loss": 3.2994,
      "step": 213850
    },
    {
      "epoch": 1.1406445735773376,
      "grad_norm": 0.38496866822242737,
      "learning_rate": 0.00016012979410426781,
      "loss": 3.3122,
      "step": 213860
    },
    {
      "epoch": 1.1407191465219704,
      "grad_norm": 0.35565486550331116,
      "learning_rate": 0.00016012792978217262,
      "loss": 3.3124,
      "step": 213870
    },
    {
      "epoch": 1.1407937194666031,
      "grad_norm": 0.3619648814201355,
      "learning_rate": 0.00016012606546007743,
      "loss": 3.2588,
      "step": 213880
    },
    {
      "epoch": 1.1408682924112359,
      "grad_norm": 0.3711971044540405,
      "learning_rate": 0.0001601242011379822,
      "loss": 3.3149,
      "step": 213890
    },
    {
      "epoch": 1.1409428653558684,
      "grad_norm": 0.3774645924568176,
      "learning_rate": 0.000160122336815887,
      "loss": 3.3453,
      "step": 213900
    },
    {
      "epoch": 1.1410174383005012,
      "grad_norm": 0.3870936930179596,
      "learning_rate": 0.00016012047249379182,
      "loss": 3.2548,
      "step": 213910
    },
    {
      "epoch": 1.141092011245134,
      "grad_norm": 0.36800721287727356,
      "learning_rate": 0.00016011860817169662,
      "loss": 3.3522,
      "step": 213920
    },
    {
      "epoch": 1.1411665841897667,
      "grad_norm": 0.36848339438438416,
      "learning_rate": 0.00016011674384960143,
      "loss": 3.3066,
      "step": 213930
    },
    {
      "epoch": 1.1412411571343994,
      "grad_norm": 0.3726692199707031,
      "learning_rate": 0.0001601148795275062,
      "loss": 3.2346,
      "step": 213940
    },
    {
      "epoch": 1.1413157300790322,
      "grad_norm": 0.385658323764801,
      "learning_rate": 0.00016011301520541102,
      "loss": 3.3279,
      "step": 213950
    },
    {
      "epoch": 1.141390303023665,
      "grad_norm": 0.4010484516620636,
      "learning_rate": 0.0001601111508833158,
      "loss": 3.3048,
      "step": 213960
    },
    {
      "epoch": 1.1414648759682977,
      "grad_norm": 0.37134841084480286,
      "learning_rate": 0.00016010928656122063,
      "loss": 3.2177,
      "step": 213970
    },
    {
      "epoch": 1.1415394489129305,
      "grad_norm": 0.39616429805755615,
      "learning_rate": 0.00016010742223912544,
      "loss": 3.2874,
      "step": 213980
    },
    {
      "epoch": 1.141614021857563,
      "grad_norm": 0.35848382115364075,
      "learning_rate": 0.00016010555791703021,
      "loss": 3.3379,
      "step": 213990
    },
    {
      "epoch": 1.1416885948021958,
      "grad_norm": 0.3978700637817383,
      "learning_rate": 0.00016010369359493502,
      "loss": 3.2448,
      "step": 214000
    },
    {
      "epoch": 1.1417631677468285,
      "grad_norm": 0.3722098171710968,
      "learning_rate": 0.0001601018292728398,
      "loss": 3.3271,
      "step": 214010
    },
    {
      "epoch": 1.1418377406914613,
      "grad_norm": 0.36351650953292847,
      "learning_rate": 0.0001600999649507446,
      "loss": 3.3448,
      "step": 214020
    },
    {
      "epoch": 1.141912313636094,
      "grad_norm": 0.38147109746932983,
      "learning_rate": 0.00016009810062864944,
      "loss": 3.3393,
      "step": 214030
    },
    {
      "epoch": 1.1419868865807268,
      "grad_norm": 0.3808027505874634,
      "learning_rate": 0.00016009623630655422,
      "loss": 3.329,
      "step": 214040
    },
    {
      "epoch": 1.1420614595253595,
      "grad_norm": 0.37981554865837097,
      "learning_rate": 0.00016009437198445903,
      "loss": 3.2695,
      "step": 214050
    },
    {
      "epoch": 1.1421360324699923,
      "grad_norm": 0.40200355648994446,
      "learning_rate": 0.0001600925076623638,
      "loss": 3.2763,
      "step": 214060
    },
    {
      "epoch": 1.142210605414625,
      "grad_norm": 0.35942724347114563,
      "learning_rate": 0.0001600906433402686,
      "loss": 3.2686,
      "step": 214070
    },
    {
      "epoch": 1.1422851783592578,
      "grad_norm": 0.3941386044025421,
      "learning_rate": 0.00016008877901817342,
      "loss": 3.3273,
      "step": 214080
    },
    {
      "epoch": 1.1423597513038906,
      "grad_norm": 0.3707486391067505,
      "learning_rate": 0.00016008691469607822,
      "loss": 3.2967,
      "step": 214090
    },
    {
      "epoch": 1.1424343242485233,
      "grad_norm": 0.38966867327690125,
      "learning_rate": 0.00016008505037398303,
      "loss": 3.2731,
      "step": 214100
    },
    {
      "epoch": 1.1425088971931558,
      "grad_norm": 0.36806270480155945,
      "learning_rate": 0.0001600831860518878,
      "loss": 3.3109,
      "step": 214110
    },
    {
      "epoch": 1.1425834701377886,
      "grad_norm": 0.3430556058883667,
      "learning_rate": 0.00016008132172979262,
      "loss": 3.2625,
      "step": 214120
    },
    {
      "epoch": 1.1426580430824214,
      "grad_norm": 0.3421136736869812,
      "learning_rate": 0.00016007945740769742,
      "loss": 3.3161,
      "step": 214130
    },
    {
      "epoch": 1.1427326160270541,
      "grad_norm": 0.3868832290172577,
      "learning_rate": 0.00016007759308560223,
      "loss": 3.3238,
      "step": 214140
    },
    {
      "epoch": 1.1428071889716869,
      "grad_norm": 0.37501880526542664,
      "learning_rate": 0.00016007572876350703,
      "loss": 3.2927,
      "step": 214150
    },
    {
      "epoch": 1.1428817619163196,
      "grad_norm": 0.37555986642837524,
      "learning_rate": 0.00016007386444141181,
      "loss": 3.242,
      "step": 214160
    },
    {
      "epoch": 1.1429563348609524,
      "grad_norm": 0.3482813239097595,
      "learning_rate": 0.00016007200011931662,
      "loss": 3.3504,
      "step": 214170
    },
    {
      "epoch": 1.1430309078055851,
      "grad_norm": 0.35903874039649963,
      "learning_rate": 0.00016007013579722143,
      "loss": 3.2569,
      "step": 214180
    },
    {
      "epoch": 1.143105480750218,
      "grad_norm": 0.39469388127326965,
      "learning_rate": 0.00016006827147512623,
      "loss": 3.3044,
      "step": 214190
    },
    {
      "epoch": 1.1431800536948504,
      "grad_norm": 0.3994810879230499,
      "learning_rate": 0.00016006640715303104,
      "loss": 3.29,
      "step": 214200
    },
    {
      "epoch": 1.1432546266394832,
      "grad_norm": 0.36433321237564087,
      "learning_rate": 0.00016006454283093582,
      "loss": 3.3346,
      "step": 214210
    },
    {
      "epoch": 1.143329199584116,
      "grad_norm": 0.3863792419433594,
      "learning_rate": 0.00016006267850884062,
      "loss": 3.3155,
      "step": 214220
    },
    {
      "epoch": 1.1434037725287487,
      "grad_norm": 0.37037622928619385,
      "learning_rate": 0.00016006081418674543,
      "loss": 3.3124,
      "step": 214230
    },
    {
      "epoch": 1.1434783454733815,
      "grad_norm": 0.39590778946876526,
      "learning_rate": 0.0001600589498646502,
      "loss": 3.2895,
      "step": 214240
    },
    {
      "epoch": 1.1435529184180142,
      "grad_norm": 0.3732934296131134,
      "learning_rate": 0.00016005708554255502,
      "loss": 3.3448,
      "step": 214250
    },
    {
      "epoch": 1.143627491362647,
      "grad_norm": 0.35538050532341003,
      "learning_rate": 0.00016005522122045982,
      "loss": 3.3122,
      "step": 214260
    },
    {
      "epoch": 1.1437020643072797,
      "grad_norm": 0.3805124759674072,
      "learning_rate": 0.00016005335689836463,
      "loss": 3.2746,
      "step": 214270
    },
    {
      "epoch": 1.1437766372519125,
      "grad_norm": 0.3763389587402344,
      "learning_rate": 0.00016005149257626944,
      "loss": 3.3206,
      "step": 214280
    },
    {
      "epoch": 1.1438512101965452,
      "grad_norm": 0.4069874882698059,
      "learning_rate": 0.00016004962825417421,
      "loss": 3.3099,
      "step": 214290
    },
    {
      "epoch": 1.143925783141178,
      "grad_norm": 0.362858384847641,
      "learning_rate": 0.00016004776393207902,
      "loss": 3.2255,
      "step": 214300
    },
    {
      "epoch": 1.1440003560858107,
      "grad_norm": 0.3547375798225403,
      "learning_rate": 0.00016004589960998383,
      "loss": 3.1962,
      "step": 214310
    },
    {
      "epoch": 1.1440749290304433,
      "grad_norm": 0.35660621523857117,
      "learning_rate": 0.00016004403528788863,
      "loss": 3.3641,
      "step": 214320
    },
    {
      "epoch": 1.144149501975076,
      "grad_norm": 0.3806912899017334,
      "learning_rate": 0.00016004217096579344,
      "loss": 3.3127,
      "step": 214330
    },
    {
      "epoch": 1.1442240749197088,
      "grad_norm": 0.36586353182792664,
      "learning_rate": 0.00016004030664369822,
      "loss": 3.3378,
      "step": 214340
    },
    {
      "epoch": 1.1442986478643415,
      "grad_norm": 0.3838302791118622,
      "learning_rate": 0.00016003844232160303,
      "loss": 3.2813,
      "step": 214350
    },
    {
      "epoch": 1.1443732208089743,
      "grad_norm": 0.37085846066474915,
      "learning_rate": 0.00016003657799950783,
      "loss": 3.2928,
      "step": 214360
    },
    {
      "epoch": 1.144447793753607,
      "grad_norm": 0.3892643451690674,
      "learning_rate": 0.00016003471367741264,
      "loss": 3.286,
      "step": 214370
    },
    {
      "epoch": 1.1445223666982398,
      "grad_norm": 0.3752046227455139,
      "learning_rate": 0.00016003284935531744,
      "loss": 3.3371,
      "step": 214380
    },
    {
      "epoch": 1.1445969396428726,
      "grad_norm": 0.3739266097545624,
      "learning_rate": 0.00016003098503322222,
      "loss": 3.2929,
      "step": 214390
    },
    {
      "epoch": 1.1446715125875053,
      "grad_norm": 0.3667449653148651,
      "learning_rate": 0.00016002912071112703,
      "loss": 3.2121,
      "step": 214400
    },
    {
      "epoch": 1.1447460855321379,
      "grad_norm": 0.3999760150909424,
      "learning_rate": 0.0001600272563890318,
      "loss": 3.3,
      "step": 214410
    },
    {
      "epoch": 1.1448206584767706,
      "grad_norm": 0.36935120820999146,
      "learning_rate": 0.00016002539206693664,
      "loss": 3.3278,
      "step": 214420
    },
    {
      "epoch": 1.1448952314214034,
      "grad_norm": 0.3810690939426422,
      "learning_rate": 0.00016002352774484145,
      "loss": 3.2918,
      "step": 214430
    },
    {
      "epoch": 1.1449698043660361,
      "grad_norm": 0.36215126514434814,
      "learning_rate": 0.00016002166342274623,
      "loss": 3.3307,
      "step": 214440
    },
    {
      "epoch": 1.1450443773106689,
      "grad_norm": 0.37438827753067017,
      "learning_rate": 0.00016001979910065103,
      "loss": 3.2646,
      "step": 214450
    },
    {
      "epoch": 1.1451189502553016,
      "grad_norm": 0.3466658890247345,
      "learning_rate": 0.0001600179347785558,
      "loss": 3.3016,
      "step": 214460
    },
    {
      "epoch": 1.1451935231999344,
      "grad_norm": 0.36939796805381775,
      "learning_rate": 0.00016001607045646062,
      "loss": 3.3185,
      "step": 214470
    },
    {
      "epoch": 1.1452680961445671,
      "grad_norm": 0.3953664004802704,
      "learning_rate": 0.00016001420613436543,
      "loss": 3.3279,
      "step": 214480
    },
    {
      "epoch": 1.1453426690892,
      "grad_norm": 0.39360037446022034,
      "learning_rate": 0.00016001234181227023,
      "loss": 3.3733,
      "step": 214490
    },
    {
      "epoch": 1.1454172420338327,
      "grad_norm": 0.36009085178375244,
      "learning_rate": 0.00016001047749017504,
      "loss": 3.3293,
      "step": 214500
    },
    {
      "epoch": 1.1454918149784654,
      "grad_norm": 0.39318057894706726,
      "learning_rate": 0.00016000861316807982,
      "loss": 3.2932,
      "step": 214510
    },
    {
      "epoch": 1.1455663879230982,
      "grad_norm": 0.3724280893802643,
      "learning_rate": 0.00016000674884598462,
      "loss": 3.3237,
      "step": 214520
    },
    {
      "epoch": 1.1456409608677307,
      "grad_norm": 0.3952750265598297,
      "learning_rate": 0.00016000488452388943,
      "loss": 3.2564,
      "step": 214530
    },
    {
      "epoch": 1.1457155338123635,
      "grad_norm": 0.3907167315483093,
      "learning_rate": 0.00016000302020179424,
      "loss": 3.3142,
      "step": 214540
    },
    {
      "epoch": 1.1457901067569962,
      "grad_norm": 0.4139921963214874,
      "learning_rate": 0.00016000115587969904,
      "loss": 3.231,
      "step": 214550
    },
    {
      "epoch": 1.145864679701629,
      "grad_norm": 0.3759714961051941,
      "learning_rate": 0.00015999929155760382,
      "loss": 3.2524,
      "step": 214560
    },
    {
      "epoch": 1.1459392526462617,
      "grad_norm": 0.37333592772483826,
      "learning_rate": 0.00015999742723550863,
      "loss": 3.2971,
      "step": 214570
    },
    {
      "epoch": 1.1460138255908945,
      "grad_norm": 0.39286381006240845,
      "learning_rate": 0.00015999556291341343,
      "loss": 3.3245,
      "step": 214580
    },
    {
      "epoch": 1.1460883985355272,
      "grad_norm": 0.35714083909988403,
      "learning_rate": 0.00015999369859131824,
      "loss": 3.2104,
      "step": 214590
    },
    {
      "epoch": 1.14616297148016,
      "grad_norm": 0.38610541820526123,
      "learning_rate": 0.00015999183426922305,
      "loss": 3.2751,
      "step": 214600
    },
    {
      "epoch": 1.1462375444247928,
      "grad_norm": 0.3738715648651123,
      "learning_rate": 0.00015998996994712783,
      "loss": 3.2939,
      "step": 214610
    },
    {
      "epoch": 1.1463121173694253,
      "grad_norm": 0.3890984058380127,
      "learning_rate": 0.00015998810562503263,
      "loss": 3.2501,
      "step": 214620
    },
    {
      "epoch": 1.146386690314058,
      "grad_norm": 0.37882596254348755,
      "learning_rate": 0.0001599862413029374,
      "loss": 3.2752,
      "step": 214630
    },
    {
      "epoch": 1.1464612632586908,
      "grad_norm": 0.3833908140659332,
      "learning_rate": 0.00015998437698084225,
      "loss": 3.2947,
      "step": 214640
    },
    {
      "epoch": 1.1465358362033236,
      "grad_norm": 0.3987097144126892,
      "learning_rate": 0.00015998251265874705,
      "loss": 3.2649,
      "step": 214650
    },
    {
      "epoch": 1.1466104091479563,
      "grad_norm": 0.3761288523674011,
      "learning_rate": 0.00015998064833665183,
      "loss": 3.2761,
      "step": 214660
    },
    {
      "epoch": 1.146684982092589,
      "grad_norm": 0.37129032611846924,
      "learning_rate": 0.00015997878401455664,
      "loss": 3.2477,
      "step": 214670
    },
    {
      "epoch": 1.1467595550372218,
      "grad_norm": 0.3756607174873352,
      "learning_rate": 0.00015997691969246142,
      "loss": 3.2862,
      "step": 214680
    },
    {
      "epoch": 1.1468341279818546,
      "grad_norm": 0.3512062728404999,
      "learning_rate": 0.00015997505537036622,
      "loss": 3.2617,
      "step": 214690
    },
    {
      "epoch": 1.1469087009264873,
      "grad_norm": 0.38412559032440186,
      "learning_rate": 0.00015997319104827106,
      "loss": 3.2782,
      "step": 214700
    },
    {
      "epoch": 1.14698327387112,
      "grad_norm": 0.3653768002986908,
      "learning_rate": 0.00015997132672617584,
      "loss": 3.3241,
      "step": 214710
    },
    {
      "epoch": 1.1470578468157528,
      "grad_norm": 0.3955306112766266,
      "learning_rate": 0.00015996946240408064,
      "loss": 3.2637,
      "step": 214720
    },
    {
      "epoch": 1.1471324197603854,
      "grad_norm": 0.36197569966316223,
      "learning_rate": 0.00015996759808198542,
      "loss": 3.2866,
      "step": 214730
    },
    {
      "epoch": 1.1472069927050181,
      "grad_norm": 0.3890834152698517,
      "learning_rate": 0.00015996573375989023,
      "loss": 3.2939,
      "step": 214740
    },
    {
      "epoch": 1.147281565649651,
      "grad_norm": 0.3878423571586609,
      "learning_rate": 0.00015996386943779503,
      "loss": 3.2608,
      "step": 214750
    },
    {
      "epoch": 1.1473561385942836,
      "grad_norm": 0.36793121695518494,
      "learning_rate": 0.00015996200511569984,
      "loss": 3.3019,
      "step": 214760
    },
    {
      "epoch": 1.1474307115389164,
      "grad_norm": 0.39133983850479126,
      "learning_rate": 0.00015996014079360465,
      "loss": 3.2771,
      "step": 214770
    },
    {
      "epoch": 1.1475052844835492,
      "grad_norm": 0.37919262051582336,
      "learning_rate": 0.00015995827647150943,
      "loss": 3.2713,
      "step": 214780
    },
    {
      "epoch": 1.147579857428182,
      "grad_norm": 0.3617783486843109,
      "learning_rate": 0.00015995641214941423,
      "loss": 3.3037,
      "step": 214790
    },
    {
      "epoch": 1.1476544303728147,
      "grad_norm": 0.376777708530426,
      "learning_rate": 0.00015995454782731904,
      "loss": 3.2492,
      "step": 214800
    },
    {
      "epoch": 1.1477290033174474,
      "grad_norm": 0.3961212933063507,
      "learning_rate": 0.00015995268350522384,
      "loss": 3.2997,
      "step": 214810
    },
    {
      "epoch": 1.1478035762620802,
      "grad_norm": 0.38455501198768616,
      "learning_rate": 0.00015995081918312865,
      "loss": 3.2996,
      "step": 214820
    },
    {
      "epoch": 1.1478781492067127,
      "grad_norm": 0.39590078592300415,
      "learning_rate": 0.00015994895486103343,
      "loss": 3.2474,
      "step": 214830
    },
    {
      "epoch": 1.1479527221513455,
      "grad_norm": 0.41040074825286865,
      "learning_rate": 0.00015994709053893824,
      "loss": 3.3312,
      "step": 214840
    },
    {
      "epoch": 1.1480272950959782,
      "grad_norm": 0.4146062731742859,
      "learning_rate": 0.00015994522621684304,
      "loss": 3.2717,
      "step": 214850
    },
    {
      "epoch": 1.148101868040611,
      "grad_norm": 0.3982124328613281,
      "learning_rate": 0.00015994336189474782,
      "loss": 3.2162,
      "step": 214860
    },
    {
      "epoch": 1.1481764409852437,
      "grad_norm": 0.3829106092453003,
      "learning_rate": 0.00015994149757265266,
      "loss": 3.2574,
      "step": 214870
    },
    {
      "epoch": 1.1482510139298765,
      "grad_norm": 0.3454431891441345,
      "learning_rate": 0.00015993963325055743,
      "loss": 3.2888,
      "step": 214880
    },
    {
      "epoch": 1.1483255868745093,
      "grad_norm": 0.3986731767654419,
      "learning_rate": 0.00015993776892846224,
      "loss": 3.2531,
      "step": 214890
    },
    {
      "epoch": 1.148400159819142,
      "grad_norm": 0.3818153738975525,
      "learning_rate": 0.00015993590460636705,
      "loss": 3.2435,
      "step": 214900
    },
    {
      "epoch": 1.1484747327637748,
      "grad_norm": 0.3920607268810272,
      "learning_rate": 0.00015993404028427183,
      "loss": 3.2871,
      "step": 214910
    },
    {
      "epoch": 1.1485493057084075,
      "grad_norm": 0.36182618141174316,
      "learning_rate": 0.00015993217596217663,
      "loss": 3.3082,
      "step": 214920
    },
    {
      "epoch": 1.1486238786530403,
      "grad_norm": 0.4081985056400299,
      "learning_rate": 0.00015993031164008144,
      "loss": 3.3155,
      "step": 214930
    },
    {
      "epoch": 1.1486984515976728,
      "grad_norm": 0.3958371579647064,
      "learning_rate": 0.00015992844731798625,
      "loss": 3.288,
      "step": 214940
    },
    {
      "epoch": 1.1487730245423056,
      "grad_norm": 0.34944480657577515,
      "learning_rate": 0.00015992658299589105,
      "loss": 3.2711,
      "step": 214950
    },
    {
      "epoch": 1.1488475974869383,
      "grad_norm": 0.3927895724773407,
      "learning_rate": 0.00015992471867379583,
      "loss": 3.3133,
      "step": 214960
    },
    {
      "epoch": 1.148922170431571,
      "grad_norm": 0.3806324303150177,
      "learning_rate": 0.00015992285435170064,
      "loss": 3.2453,
      "step": 214970
    },
    {
      "epoch": 1.1489967433762038,
      "grad_norm": 0.38537171483039856,
      "learning_rate": 0.00015992099002960544,
      "loss": 3.2371,
      "step": 214980
    },
    {
      "epoch": 1.1490713163208366,
      "grad_norm": 0.36155053973197937,
      "learning_rate": 0.00015991912570751025,
      "loss": 3.3174,
      "step": 214990
    },
    {
      "epoch": 1.1491458892654693,
      "grad_norm": 0.37976914644241333,
      "learning_rate": 0.00015991726138541506,
      "loss": 3.2468,
      "step": 215000
    },
    {
      "epoch": 1.149220462210102,
      "grad_norm": 0.3418916165828705,
      "learning_rate": 0.00015991539706331983,
      "loss": 3.2998,
      "step": 215010
    },
    {
      "epoch": 1.1492950351547349,
      "grad_norm": 0.3990996479988098,
      "learning_rate": 0.00015991353274122464,
      "loss": 3.312,
      "step": 215020
    },
    {
      "epoch": 1.1493696080993676,
      "grad_norm": 0.3768617808818817,
      "learning_rate": 0.00015991166841912945,
      "loss": 3.269,
      "step": 215030
    },
    {
      "epoch": 1.1494441810440001,
      "grad_norm": 0.4061301350593567,
      "learning_rate": 0.00015990980409703425,
      "loss": 3.2824,
      "step": 215040
    },
    {
      "epoch": 1.149518753988633,
      "grad_norm": 0.3739570379257202,
      "learning_rate": 0.00015990793977493906,
      "loss": 3.2661,
      "step": 215050
    },
    {
      "epoch": 1.1495933269332657,
      "grad_norm": 0.38258981704711914,
      "learning_rate": 0.00015990607545284384,
      "loss": 3.3328,
      "step": 215060
    },
    {
      "epoch": 1.1496678998778984,
      "grad_norm": 0.38168612122535706,
      "learning_rate": 0.00015990421113074865,
      "loss": 3.3212,
      "step": 215070
    },
    {
      "epoch": 1.1497424728225312,
      "grad_norm": 0.38615041971206665,
      "learning_rate": 0.00015990234680865342,
      "loss": 3.2659,
      "step": 215080
    },
    {
      "epoch": 1.149817045767164,
      "grad_norm": 0.3861806094646454,
      "learning_rate": 0.00015990048248655826,
      "loss": 3.3732,
      "step": 215090
    },
    {
      "epoch": 1.1498916187117967,
      "grad_norm": 0.3856545090675354,
      "learning_rate": 0.00015989861816446306,
      "loss": 3.2806,
      "step": 215100
    },
    {
      "epoch": 1.1499661916564294,
      "grad_norm": 0.4019118547439575,
      "learning_rate": 0.00015989675384236784,
      "loss": 3.24,
      "step": 215110
    },
    {
      "epoch": 1.1500407646010622,
      "grad_norm": 0.40925490856170654,
      "learning_rate": 0.00015989488952027265,
      "loss": 3.309,
      "step": 215120
    },
    {
      "epoch": 1.150115337545695,
      "grad_norm": 0.38617607951164246,
      "learning_rate": 0.00015989302519817743,
      "loss": 3.2735,
      "step": 215130
    },
    {
      "epoch": 1.1501899104903277,
      "grad_norm": 0.4057895243167877,
      "learning_rate": 0.00015989116087608224,
      "loss": 3.3403,
      "step": 215140
    },
    {
      "epoch": 1.1502644834349602,
      "grad_norm": 0.38036268949508667,
      "learning_rate": 0.00015988929655398707,
      "loss": 3.2979,
      "step": 215150
    },
    {
      "epoch": 1.150339056379593,
      "grad_norm": 0.3676172196865082,
      "learning_rate": 0.00015988743223189185,
      "loss": 3.2832,
      "step": 215160
    },
    {
      "epoch": 1.1504136293242258,
      "grad_norm": 0.35908135771751404,
      "learning_rate": 0.00015988556790979665,
      "loss": 3.3133,
      "step": 215170
    },
    {
      "epoch": 1.1504882022688585,
      "grad_norm": 0.3594930171966553,
      "learning_rate": 0.00015988370358770143,
      "loss": 3.3429,
      "step": 215180
    },
    {
      "epoch": 1.1505627752134913,
      "grad_norm": 0.3733145594596863,
      "learning_rate": 0.00015988183926560624,
      "loss": 3.3242,
      "step": 215190
    },
    {
      "epoch": 1.150637348158124,
      "grad_norm": 0.37925636768341064,
      "learning_rate": 0.00015987997494351105,
      "loss": 3.3206,
      "step": 215200
    },
    {
      "epoch": 1.1507119211027568,
      "grad_norm": 0.3940190374851227,
      "learning_rate": 0.00015987811062141585,
      "loss": 3.2368,
      "step": 215210
    },
    {
      "epoch": 1.1507864940473895,
      "grad_norm": 0.37965670228004456,
      "learning_rate": 0.00015987624629932066,
      "loss": 3.2917,
      "step": 215220
    },
    {
      "epoch": 1.1508610669920223,
      "grad_norm": 0.37174201011657715,
      "learning_rate": 0.00015987438197722544,
      "loss": 3.319,
      "step": 215230
    },
    {
      "epoch": 1.1509356399366548,
      "grad_norm": 0.3569222092628479,
      "learning_rate": 0.00015987251765513024,
      "loss": 3.3155,
      "step": 215240
    },
    {
      "epoch": 1.1510102128812876,
      "grad_norm": 0.3716641962528229,
      "learning_rate": 0.00015987065333303505,
      "loss": 3.3357,
      "step": 215250
    },
    {
      "epoch": 1.1510847858259203,
      "grad_norm": 0.3567156195640564,
      "learning_rate": 0.00015986878901093986,
      "loss": 3.3253,
      "step": 215260
    },
    {
      "epoch": 1.151159358770553,
      "grad_norm": 0.35434192419052124,
      "learning_rate": 0.00015986692468884466,
      "loss": 3.2849,
      "step": 215270
    },
    {
      "epoch": 1.1512339317151858,
      "grad_norm": 0.38773471117019653,
      "learning_rate": 0.00015986506036674944,
      "loss": 3.2911,
      "step": 215280
    },
    {
      "epoch": 1.1513085046598186,
      "grad_norm": 0.416944295167923,
      "learning_rate": 0.00015986319604465425,
      "loss": 3.283,
      "step": 215290
    },
    {
      "epoch": 1.1513830776044514,
      "grad_norm": 0.38717278838157654,
      "learning_rate": 0.00015986133172255906,
      "loss": 3.3035,
      "step": 215300
    },
    {
      "epoch": 1.151457650549084,
      "grad_norm": 0.3617466390132904,
      "learning_rate": 0.00015985946740046383,
      "loss": 3.2805,
      "step": 215310
    },
    {
      "epoch": 1.1515322234937169,
      "grad_norm": 0.37861523032188416,
      "learning_rate": 0.00015985760307836867,
      "loss": 3.2676,
      "step": 215320
    },
    {
      "epoch": 1.1516067964383496,
      "grad_norm": 0.39471200108528137,
      "learning_rate": 0.00015985573875627345,
      "loss": 3.2295,
      "step": 215330
    },
    {
      "epoch": 1.1516813693829824,
      "grad_norm": 0.39419978857040405,
      "learning_rate": 0.00015985387443417825,
      "loss": 3.3206,
      "step": 215340
    },
    {
      "epoch": 1.1517559423276151,
      "grad_norm": 0.37104010581970215,
      "learning_rate": 0.00015985201011208306,
      "loss": 3.3193,
      "step": 215350
    },
    {
      "epoch": 1.1518305152722477,
      "grad_norm": 0.3691539168357849,
      "learning_rate": 0.00015985014578998784,
      "loss": 3.2566,
      "step": 215360
    },
    {
      "epoch": 1.1519050882168804,
      "grad_norm": 0.3783223628997803,
      "learning_rate": 0.00015984828146789265,
      "loss": 3.3687,
      "step": 215370
    },
    {
      "epoch": 1.1519796611615132,
      "grad_norm": 0.3794434368610382,
      "learning_rate": 0.00015984641714579745,
      "loss": 3.3226,
      "step": 215380
    },
    {
      "epoch": 1.152054234106146,
      "grad_norm": 0.3761386573314667,
      "learning_rate": 0.00015984455282370226,
      "loss": 3.3415,
      "step": 215390
    },
    {
      "epoch": 1.1521288070507787,
      "grad_norm": 0.368582159280777,
      "learning_rate": 0.00015984268850160706,
      "loss": 3.3157,
      "step": 215400
    },
    {
      "epoch": 1.1522033799954114,
      "grad_norm": 0.3662542998790741,
      "learning_rate": 0.00015984082417951184,
      "loss": 3.296,
      "step": 215410
    },
    {
      "epoch": 1.1522779529400442,
      "grad_norm": 0.39352506399154663,
      "learning_rate": 0.00015983895985741665,
      "loss": 3.2665,
      "step": 215420
    },
    {
      "epoch": 1.152352525884677,
      "grad_norm": 0.361630916595459,
      "learning_rate": 0.00015983709553532146,
      "loss": 3.3283,
      "step": 215430
    },
    {
      "epoch": 1.1524270988293097,
      "grad_norm": 0.36641091108322144,
      "learning_rate": 0.00015983523121322626,
      "loss": 3.335,
      "step": 215440
    },
    {
      "epoch": 1.1525016717739422,
      "grad_norm": 0.41733354330062866,
      "learning_rate": 0.00015983336689113107,
      "loss": 3.3005,
      "step": 215450
    },
    {
      "epoch": 1.152576244718575,
      "grad_norm": 0.37736231088638306,
      "learning_rate": 0.00015983150256903585,
      "loss": 3.2757,
      "step": 215460
    },
    {
      "epoch": 1.1526508176632078,
      "grad_norm": 0.3687831163406372,
      "learning_rate": 0.00015982963824694065,
      "loss": 3.3655,
      "step": 215470
    },
    {
      "epoch": 1.1527253906078405,
      "grad_norm": 0.3626371920108795,
      "learning_rate": 0.00015982777392484546,
      "loss": 3.3058,
      "step": 215480
    },
    {
      "epoch": 1.1527999635524733,
      "grad_norm": 0.39931049942970276,
      "learning_rate": 0.00015982590960275027,
      "loss": 3.2369,
      "step": 215490
    },
    {
      "epoch": 1.152874536497106,
      "grad_norm": 0.377530574798584,
      "learning_rate": 0.00015982404528065507,
      "loss": 3.2976,
      "step": 215500
    },
    {
      "epoch": 1.1529491094417388,
      "grad_norm": 0.4062236547470093,
      "learning_rate": 0.00015982218095855985,
      "loss": 3.2852,
      "step": 215510
    },
    {
      "epoch": 1.1530236823863715,
      "grad_norm": 0.366414338350296,
      "learning_rate": 0.00015982031663646466,
      "loss": 3.3222,
      "step": 215520
    },
    {
      "epoch": 1.1530982553310043,
      "grad_norm": 0.3599797785282135,
      "learning_rate": 0.00015981845231436944,
      "loss": 3.265,
      "step": 215530
    },
    {
      "epoch": 1.153172828275637,
      "grad_norm": 0.3628811538219452,
      "learning_rate": 0.00015981658799227427,
      "loss": 3.2659,
      "step": 215540
    },
    {
      "epoch": 1.1532474012202698,
      "grad_norm": 0.37438058853149414,
      "learning_rate": 0.00015981472367017908,
      "loss": 3.2892,
      "step": 215550
    },
    {
      "epoch": 1.1533219741649026,
      "grad_norm": 0.3675566613674164,
      "learning_rate": 0.00015981285934808386,
      "loss": 3.2538,
      "step": 215560
    },
    {
      "epoch": 1.153396547109535,
      "grad_norm": 0.38150590658187866,
      "learning_rate": 0.00015981099502598866,
      "loss": 3.3145,
      "step": 215570
    },
    {
      "epoch": 1.1534711200541679,
      "grad_norm": 0.374281108379364,
      "learning_rate": 0.00015980913070389344,
      "loss": 3.3011,
      "step": 215580
    },
    {
      "epoch": 1.1535456929988006,
      "grad_norm": 0.3904232382774353,
      "learning_rate": 0.00015980726638179825,
      "loss": 3.3402,
      "step": 215590
    },
    {
      "epoch": 1.1536202659434334,
      "grad_norm": 0.39935964345932007,
      "learning_rate": 0.00015980540205970308,
      "loss": 3.2531,
      "step": 215600
    },
    {
      "epoch": 1.1536948388880661,
      "grad_norm": 0.37424176931381226,
      "learning_rate": 0.00015980353773760786,
      "loss": 3.3092,
      "step": 215610
    },
    {
      "epoch": 1.1537694118326989,
      "grad_norm": 0.3898569941520691,
      "learning_rate": 0.00015980167341551267,
      "loss": 3.2568,
      "step": 215620
    },
    {
      "epoch": 1.1538439847773316,
      "grad_norm": 0.3782808780670166,
      "learning_rate": 0.00015979980909341745,
      "loss": 3.3261,
      "step": 215630
    },
    {
      "epoch": 1.1539185577219644,
      "grad_norm": 0.5519580841064453,
      "learning_rate": 0.00015979794477132225,
      "loss": 3.3112,
      "step": 215640
    },
    {
      "epoch": 1.1539931306665971,
      "grad_norm": 0.3848615288734436,
      "learning_rate": 0.00015979608044922706,
      "loss": 3.2934,
      "step": 215650
    },
    {
      "epoch": 1.1540677036112297,
      "grad_norm": 0.3524031341075897,
      "learning_rate": 0.00015979421612713187,
      "loss": 3.328,
      "step": 215660
    },
    {
      "epoch": 1.1541422765558624,
      "grad_norm": 0.3820040822029114,
      "learning_rate": 0.00015979235180503667,
      "loss": 3.2557,
      "step": 215670
    },
    {
      "epoch": 1.1542168495004952,
      "grad_norm": 0.3654330372810364,
      "learning_rate": 0.00015979048748294145,
      "loss": 3.3533,
      "step": 215680
    },
    {
      "epoch": 1.154291422445128,
      "grad_norm": 0.3576078712940216,
      "learning_rate": 0.00015978862316084626,
      "loss": 3.2625,
      "step": 215690
    },
    {
      "epoch": 1.1543659953897607,
      "grad_norm": 0.3871668875217438,
      "learning_rate": 0.00015978675883875106,
      "loss": 3.2997,
      "step": 215700
    },
    {
      "epoch": 1.1544405683343935,
      "grad_norm": 0.3881453275680542,
      "learning_rate": 0.00015978489451665587,
      "loss": 3.3708,
      "step": 215710
    },
    {
      "epoch": 1.1545151412790262,
      "grad_norm": 0.3828691244125366,
      "learning_rate": 0.00015978303019456068,
      "loss": 3.2742,
      "step": 215720
    },
    {
      "epoch": 1.154589714223659,
      "grad_norm": 0.4025832414627075,
      "learning_rate": 0.00015978116587246546,
      "loss": 3.3192,
      "step": 215730
    },
    {
      "epoch": 1.1546642871682917,
      "grad_norm": 0.34329912066459656,
      "learning_rate": 0.00015977930155037026,
      "loss": 3.2806,
      "step": 215740
    },
    {
      "epoch": 1.1547388601129245,
      "grad_norm": 0.35422199964523315,
      "learning_rate": 0.00015977743722827507,
      "loss": 3.3241,
      "step": 215750
    },
    {
      "epoch": 1.1548134330575572,
      "grad_norm": 0.36820685863494873,
      "learning_rate": 0.00015977557290617985,
      "loss": 3.2131,
      "step": 215760
    },
    {
      "epoch": 1.15488800600219,
      "grad_norm": 0.39247092604637146,
      "learning_rate": 0.00015977370858408468,
      "loss": 3.2693,
      "step": 215770
    },
    {
      "epoch": 1.1549625789468225,
      "grad_norm": 0.35834038257598877,
      "learning_rate": 0.00015977184426198946,
      "loss": 3.3019,
      "step": 215780
    },
    {
      "epoch": 1.1550371518914553,
      "grad_norm": 0.3603430986404419,
      "learning_rate": 0.00015976997993989427,
      "loss": 3.2612,
      "step": 215790
    },
    {
      "epoch": 1.155111724836088,
      "grad_norm": 0.3781728446483612,
      "learning_rate": 0.00015976811561779907,
      "loss": 3.3078,
      "step": 215800
    },
    {
      "epoch": 1.1551862977807208,
      "grad_norm": 0.37095582485198975,
      "learning_rate": 0.00015976625129570385,
      "loss": 3.2643,
      "step": 215810
    },
    {
      "epoch": 1.1552608707253536,
      "grad_norm": 0.3919539153575897,
      "learning_rate": 0.00015976438697360866,
      "loss": 3.317,
      "step": 215820
    },
    {
      "epoch": 1.1553354436699863,
      "grad_norm": 0.4104246199131012,
      "learning_rate": 0.00015976252265151346,
      "loss": 3.2512,
      "step": 215830
    },
    {
      "epoch": 1.155410016614619,
      "grad_norm": 0.37791743874549866,
      "learning_rate": 0.00015976065832941827,
      "loss": 3.2954,
      "step": 215840
    },
    {
      "epoch": 1.1554845895592518,
      "grad_norm": 0.37136489152908325,
      "learning_rate": 0.00015975879400732305,
      "loss": 3.2518,
      "step": 215850
    },
    {
      "epoch": 1.1555591625038846,
      "grad_norm": 0.39434775710105896,
      "learning_rate": 0.00015975692968522786,
      "loss": 3.2664,
      "step": 215860
    },
    {
      "epoch": 1.155633735448517,
      "grad_norm": 0.3799855411052704,
      "learning_rate": 0.00015975506536313266,
      "loss": 3.2624,
      "step": 215870
    },
    {
      "epoch": 1.1557083083931499,
      "grad_norm": 0.3897678852081299,
      "learning_rate": 0.00015975320104103747,
      "loss": 3.3458,
      "step": 215880
    },
    {
      "epoch": 1.1557828813377826,
      "grad_norm": 0.3619040250778198,
      "learning_rate": 0.00015975133671894228,
      "loss": 3.294,
      "step": 215890
    },
    {
      "epoch": 1.1558574542824154,
      "grad_norm": 0.3677838444709778,
      "learning_rate": 0.00015974947239684705,
      "loss": 3.2593,
      "step": 215900
    },
    {
      "epoch": 1.1559320272270481,
      "grad_norm": 0.3768633306026459,
      "learning_rate": 0.00015974760807475186,
      "loss": 3.3032,
      "step": 215910
    },
    {
      "epoch": 1.1560066001716809,
      "grad_norm": 0.36480286717414856,
      "learning_rate": 0.00015974574375265667,
      "loss": 3.2911,
      "step": 215920
    },
    {
      "epoch": 1.1560811731163136,
      "grad_norm": 0.35989445447921753,
      "learning_rate": 0.00015974387943056147,
      "loss": 3.3254,
      "step": 215930
    },
    {
      "epoch": 1.1561557460609464,
      "grad_norm": 0.3743935227394104,
      "learning_rate": 0.00015974201510846628,
      "loss": 3.2639,
      "step": 215940
    },
    {
      "epoch": 1.1562303190055792,
      "grad_norm": 0.38597115874290466,
      "learning_rate": 0.00015974015078637106,
      "loss": 3.3163,
      "step": 215950
    },
    {
      "epoch": 1.156304891950212,
      "grad_norm": 0.3560324013233185,
      "learning_rate": 0.00015973828646427587,
      "loss": 3.2667,
      "step": 215960
    },
    {
      "epoch": 1.1563794648948447,
      "grad_norm": 0.40448230504989624,
      "learning_rate": 0.00015973642214218067,
      "loss": 3.332,
      "step": 215970
    },
    {
      "epoch": 1.1564540378394774,
      "grad_norm": 0.3711065649986267,
      "learning_rate": 0.00015973455782008545,
      "loss": 3.2508,
      "step": 215980
    },
    {
      "epoch": 1.15652861078411,
      "grad_norm": 0.3442196249961853,
      "learning_rate": 0.00015973269349799028,
      "loss": 3.3101,
      "step": 215990
    },
    {
      "epoch": 1.1566031837287427,
      "grad_norm": 0.38378992676734924,
      "learning_rate": 0.00015973082917589506,
      "loss": 3.2935,
      "step": 216000
    },
    {
      "epoch": 1.1566777566733755,
      "grad_norm": 0.3791636526584625,
      "learning_rate": 0.00015972896485379987,
      "loss": 3.3058,
      "step": 216010
    },
    {
      "epoch": 1.1567523296180082,
      "grad_norm": 0.3653338551521301,
      "learning_rate": 0.00015972710053170468,
      "loss": 3.2467,
      "step": 216020
    },
    {
      "epoch": 1.156826902562641,
      "grad_norm": 0.36529025435447693,
      "learning_rate": 0.00015972523620960946,
      "loss": 3.3104,
      "step": 216030
    },
    {
      "epoch": 1.1569014755072737,
      "grad_norm": 0.38300737738609314,
      "learning_rate": 0.00015972337188751426,
      "loss": 3.3082,
      "step": 216040
    },
    {
      "epoch": 1.1569760484519065,
      "grad_norm": 0.36986443400382996,
      "learning_rate": 0.00015972150756541907,
      "loss": 3.2815,
      "step": 216050
    },
    {
      "epoch": 1.1570506213965392,
      "grad_norm": 0.3877127170562744,
      "learning_rate": 0.00015971964324332387,
      "loss": 3.23,
      "step": 216060
    },
    {
      "epoch": 1.157125194341172,
      "grad_norm": 0.3571929335594177,
      "learning_rate": 0.00015971777892122868,
      "loss": 3.2842,
      "step": 216070
    },
    {
      "epoch": 1.1571997672858045,
      "grad_norm": 0.3645228147506714,
      "learning_rate": 0.00015971591459913346,
      "loss": 3.2697,
      "step": 216080
    },
    {
      "epoch": 1.1572743402304373,
      "grad_norm": 0.3729952573776245,
      "learning_rate": 0.00015971405027703827,
      "loss": 3.2269,
      "step": 216090
    },
    {
      "epoch": 1.15734891317507,
      "grad_norm": 0.36642393469810486,
      "learning_rate": 0.00015971218595494307,
      "loss": 3.2319,
      "step": 216100
    },
    {
      "epoch": 1.1574234861197028,
      "grad_norm": 0.3792392313480377,
      "learning_rate": 0.00015971032163284788,
      "loss": 3.294,
      "step": 216110
    },
    {
      "epoch": 1.1574980590643356,
      "grad_norm": 0.3704824447631836,
      "learning_rate": 0.00015970845731075268,
      "loss": 3.2416,
      "step": 216120
    },
    {
      "epoch": 1.1575726320089683,
      "grad_norm": 0.3528258502483368,
      "learning_rate": 0.00015970659298865746,
      "loss": 3.2651,
      "step": 216130
    },
    {
      "epoch": 1.157647204953601,
      "grad_norm": 0.41477474570274353,
      "learning_rate": 0.00015970472866656227,
      "loss": 3.2946,
      "step": 216140
    },
    {
      "epoch": 1.1577217778982338,
      "grad_norm": 0.40655601024627686,
      "learning_rate": 0.00015970286434446705,
      "loss": 3.2858,
      "step": 216150
    },
    {
      "epoch": 1.1577963508428666,
      "grad_norm": 0.3929535150527954,
      "learning_rate": 0.00015970100002237188,
      "loss": 3.3332,
      "step": 216160
    },
    {
      "epoch": 1.1578709237874993,
      "grad_norm": 0.36304736137390137,
      "learning_rate": 0.0001596991357002767,
      "loss": 3.2805,
      "step": 216170
    },
    {
      "epoch": 1.157945496732132,
      "grad_norm": 0.36715424060821533,
      "learning_rate": 0.00015969727137818147,
      "loss": 3.1892,
      "step": 216180
    },
    {
      "epoch": 1.1580200696767649,
      "grad_norm": 0.3891381323337555,
      "learning_rate": 0.00015969540705608627,
      "loss": 3.3697,
      "step": 216190
    },
    {
      "epoch": 1.1580946426213974,
      "grad_norm": 0.3871602714061737,
      "learning_rate": 0.00015969354273399105,
      "loss": 3.2838,
      "step": 216200
    },
    {
      "epoch": 1.1581692155660301,
      "grad_norm": 0.42667078971862793,
      "learning_rate": 0.00015969167841189586,
      "loss": 3.2693,
      "step": 216210
    },
    {
      "epoch": 1.158243788510663,
      "grad_norm": 0.38843834400177,
      "learning_rate": 0.0001596898140898007,
      "loss": 3.3071,
      "step": 216220
    },
    {
      "epoch": 1.1583183614552957,
      "grad_norm": 0.41630977392196655,
      "learning_rate": 0.00015968794976770547,
      "loss": 3.3472,
      "step": 216230
    },
    {
      "epoch": 1.1583929343999284,
      "grad_norm": 0.36610278487205505,
      "learning_rate": 0.00015968608544561028,
      "loss": 3.2653,
      "step": 216240
    },
    {
      "epoch": 1.1584675073445612,
      "grad_norm": 0.41582825779914856,
      "learning_rate": 0.00015968422112351506,
      "loss": 3.2702,
      "step": 216250
    },
    {
      "epoch": 1.158542080289194,
      "grad_norm": 0.3737486004829407,
      "learning_rate": 0.00015968235680141986,
      "loss": 3.28,
      "step": 216260
    },
    {
      "epoch": 1.1586166532338267,
      "grad_norm": 0.3965378403663635,
      "learning_rate": 0.00015968049247932467,
      "loss": 3.1856,
      "step": 216270
    },
    {
      "epoch": 1.1586912261784594,
      "grad_norm": 0.3733866214752197,
      "learning_rate": 0.00015967862815722948,
      "loss": 3.2909,
      "step": 216280
    },
    {
      "epoch": 1.158765799123092,
      "grad_norm": 0.36327823996543884,
      "learning_rate": 0.00015967676383513428,
      "loss": 3.2517,
      "step": 216290
    },
    {
      "epoch": 1.1588403720677247,
      "grad_norm": 0.3872983753681183,
      "learning_rate": 0.00015967489951303906,
      "loss": 3.3796,
      "step": 216300
    },
    {
      "epoch": 1.1589149450123575,
      "grad_norm": 0.372650146484375,
      "learning_rate": 0.00015967303519094387,
      "loss": 3.1961,
      "step": 216310
    },
    {
      "epoch": 1.1589895179569902,
      "grad_norm": 0.36597079038619995,
      "learning_rate": 0.00015967117086884868,
      "loss": 3.3015,
      "step": 216320
    },
    {
      "epoch": 1.159064090901623,
      "grad_norm": 0.3933015763759613,
      "learning_rate": 0.00015966930654675348,
      "loss": 3.2607,
      "step": 216330
    },
    {
      "epoch": 1.1591386638462557,
      "grad_norm": 0.37859076261520386,
      "learning_rate": 0.0001596674422246583,
      "loss": 3.265,
      "step": 216340
    },
    {
      "epoch": 1.1592132367908885,
      "grad_norm": 0.3816450238227844,
      "learning_rate": 0.00015966557790256307,
      "loss": 3.2307,
      "step": 216350
    },
    {
      "epoch": 1.1592878097355213,
      "grad_norm": 0.38465848565101624,
      "learning_rate": 0.00015966371358046787,
      "loss": 3.2641,
      "step": 216360
    },
    {
      "epoch": 1.159362382680154,
      "grad_norm": 0.40000486373901367,
      "learning_rate": 0.00015966184925837268,
      "loss": 3.2751,
      "step": 216370
    },
    {
      "epoch": 1.1594369556247868,
      "grad_norm": 0.38802075386047363,
      "learning_rate": 0.00015965998493627749,
      "loss": 3.2238,
      "step": 216380
    },
    {
      "epoch": 1.1595115285694195,
      "grad_norm": 0.37421274185180664,
      "learning_rate": 0.0001596581206141823,
      "loss": 3.2756,
      "step": 216390
    },
    {
      "epoch": 1.1595861015140523,
      "grad_norm": 0.4096425473690033,
      "learning_rate": 0.00015965625629208707,
      "loss": 3.2724,
      "step": 216400
    },
    {
      "epoch": 1.1596606744586848,
      "grad_norm": 0.4026188552379608,
      "learning_rate": 0.00015965439196999188,
      "loss": 3.2924,
      "step": 216410
    },
    {
      "epoch": 1.1597352474033176,
      "grad_norm": 0.3663683235645294,
      "learning_rate": 0.00015965252764789668,
      "loss": 3.3493,
      "step": 216420
    },
    {
      "epoch": 1.1598098203479503,
      "grad_norm": 0.37758970260620117,
      "learning_rate": 0.00015965066332580146,
      "loss": 3.3244,
      "step": 216430
    },
    {
      "epoch": 1.159884393292583,
      "grad_norm": 0.3993645906448364,
      "learning_rate": 0.0001596487990037063,
      "loss": 3.315,
      "step": 216440
    },
    {
      "epoch": 1.1599589662372158,
      "grad_norm": 0.4085935056209564,
      "learning_rate": 0.00015964693468161108,
      "loss": 3.2211,
      "step": 216450
    },
    {
      "epoch": 1.1600335391818486,
      "grad_norm": 0.3776691257953644,
      "learning_rate": 0.00015964507035951588,
      "loss": 3.3513,
      "step": 216460
    },
    {
      "epoch": 1.1601081121264814,
      "grad_norm": 0.3530852794647217,
      "learning_rate": 0.0001596432060374207,
      "loss": 3.2448,
      "step": 216470
    },
    {
      "epoch": 1.160182685071114,
      "grad_norm": 0.3914662003517151,
      "learning_rate": 0.00015964134171532547,
      "loss": 3.2659,
      "step": 216480
    },
    {
      "epoch": 1.1602572580157469,
      "grad_norm": 0.38485240936279297,
      "learning_rate": 0.00015963947739323027,
      "loss": 3.2991,
      "step": 216490
    },
    {
      "epoch": 1.1603318309603794,
      "grad_norm": 0.38754916191101074,
      "learning_rate": 0.00015963761307113508,
      "loss": 3.2125,
      "step": 216500
    },
    {
      "epoch": 1.1604064039050122,
      "grad_norm": 0.3624873459339142,
      "learning_rate": 0.0001596357487490399,
      "loss": 3.2493,
      "step": 216510
    },
    {
      "epoch": 1.160480976849645,
      "grad_norm": 0.37378406524658203,
      "learning_rate": 0.0001596338844269447,
      "loss": 3.3026,
      "step": 216520
    },
    {
      "epoch": 1.1605555497942777,
      "grad_norm": 0.3828743100166321,
      "learning_rate": 0.00015963202010484947,
      "loss": 3.2424,
      "step": 216530
    },
    {
      "epoch": 1.1606301227389104,
      "grad_norm": 0.36485859751701355,
      "learning_rate": 0.00015963015578275428,
      "loss": 3.3193,
      "step": 216540
    },
    {
      "epoch": 1.1607046956835432,
      "grad_norm": 0.4091915190219879,
      "learning_rate": 0.00015962829146065908,
      "loss": 3.2779,
      "step": 216550
    },
    {
      "epoch": 1.160779268628176,
      "grad_norm": 0.37104859948158264,
      "learning_rate": 0.0001596264271385639,
      "loss": 3.2551,
      "step": 216560
    },
    {
      "epoch": 1.1608538415728087,
      "grad_norm": 0.39210131764411926,
      "learning_rate": 0.0001596245628164687,
      "loss": 3.2354,
      "step": 216570
    },
    {
      "epoch": 1.1609284145174414,
      "grad_norm": 0.34467506408691406,
      "learning_rate": 0.00015962269849437348,
      "loss": 3.2794,
      "step": 216580
    },
    {
      "epoch": 1.1610029874620742,
      "grad_norm": 0.3979927897453308,
      "learning_rate": 0.00015962083417227828,
      "loss": 3.2727,
      "step": 216590
    },
    {
      "epoch": 1.161077560406707,
      "grad_norm": 0.3603101968765259,
      "learning_rate": 0.00015961896985018306,
      "loss": 3.2948,
      "step": 216600
    },
    {
      "epoch": 1.1611521333513397,
      "grad_norm": 0.38563278317451477,
      "learning_rate": 0.0001596171055280879,
      "loss": 3.2644,
      "step": 216610
    },
    {
      "epoch": 1.1612267062959722,
      "grad_norm": 0.38455891609191895,
      "learning_rate": 0.0001596152412059927,
      "loss": 3.2552,
      "step": 216620
    },
    {
      "epoch": 1.161301279240605,
      "grad_norm": 0.3538244664669037,
      "learning_rate": 0.00015961337688389748,
      "loss": 3.2679,
      "step": 216630
    },
    {
      "epoch": 1.1613758521852378,
      "grad_norm": 0.38571232557296753,
      "learning_rate": 0.0001596115125618023,
      "loss": 3.2515,
      "step": 216640
    },
    {
      "epoch": 1.1614504251298705,
      "grad_norm": 0.3649024963378906,
      "learning_rate": 0.00015960964823970707,
      "loss": 3.2685,
      "step": 216650
    },
    {
      "epoch": 1.1615249980745033,
      "grad_norm": 0.3547821342945099,
      "learning_rate": 0.00015960778391761187,
      "loss": 3.2965,
      "step": 216660
    },
    {
      "epoch": 1.161599571019136,
      "grad_norm": 0.34854018688201904,
      "learning_rate": 0.0001596059195955167,
      "loss": 3.2769,
      "step": 216670
    },
    {
      "epoch": 1.1616741439637688,
      "grad_norm": 0.3572191894054413,
      "learning_rate": 0.00015960405527342149,
      "loss": 3.2743,
      "step": 216680
    },
    {
      "epoch": 1.1617487169084015,
      "grad_norm": 0.3564607501029968,
      "learning_rate": 0.0001596021909513263,
      "loss": 3.2891,
      "step": 216690
    },
    {
      "epoch": 1.1618232898530343,
      "grad_norm": 0.3613826334476471,
      "learning_rate": 0.00015960032662923107,
      "loss": 3.2643,
      "step": 216700
    },
    {
      "epoch": 1.1618978627976668,
      "grad_norm": 0.3659050464630127,
      "learning_rate": 0.00015959846230713588,
      "loss": 3.3194,
      "step": 216710
    },
    {
      "epoch": 1.1619724357422996,
      "grad_norm": 0.3722460865974426,
      "learning_rate": 0.00015959659798504068,
      "loss": 3.3052,
      "step": 216720
    },
    {
      "epoch": 1.1620470086869323,
      "grad_norm": 0.37007778882980347,
      "learning_rate": 0.0001595947336629455,
      "loss": 3.2626,
      "step": 216730
    },
    {
      "epoch": 1.162121581631565,
      "grad_norm": 0.3749721944332123,
      "learning_rate": 0.0001595928693408503,
      "loss": 3.3248,
      "step": 216740
    },
    {
      "epoch": 1.1621961545761978,
      "grad_norm": 0.38550010323524475,
      "learning_rate": 0.00015959100501875508,
      "loss": 3.3014,
      "step": 216750
    },
    {
      "epoch": 1.1622707275208306,
      "grad_norm": 0.4009402096271515,
      "learning_rate": 0.00015958914069665988,
      "loss": 3.3085,
      "step": 216760
    },
    {
      "epoch": 1.1623453004654634,
      "grad_norm": 0.3974381685256958,
      "learning_rate": 0.0001595872763745647,
      "loss": 3.3002,
      "step": 216770
    },
    {
      "epoch": 1.1624198734100961,
      "grad_norm": 0.37696847319602966,
      "learning_rate": 0.0001595854120524695,
      "loss": 3.2313,
      "step": 216780
    },
    {
      "epoch": 1.1624944463547289,
      "grad_norm": 0.43271028995513916,
      "learning_rate": 0.0001595835477303743,
      "loss": 3.331,
      "step": 216790
    },
    {
      "epoch": 1.1625690192993616,
      "grad_norm": 0.36293116211891174,
      "learning_rate": 0.00015958168340827908,
      "loss": 3.312,
      "step": 216800
    },
    {
      "epoch": 1.1626435922439944,
      "grad_norm": 0.3780125379562378,
      "learning_rate": 0.00015957981908618389,
      "loss": 3.2982,
      "step": 216810
    },
    {
      "epoch": 1.1627181651886271,
      "grad_norm": 0.38156160712242126,
      "learning_rate": 0.0001595779547640887,
      "loss": 3.3014,
      "step": 216820
    },
    {
      "epoch": 1.1627927381332597,
      "grad_norm": 0.3766763210296631,
      "learning_rate": 0.0001595760904419935,
      "loss": 3.3039,
      "step": 216830
    },
    {
      "epoch": 1.1628673110778924,
      "grad_norm": 0.38389816880226135,
      "learning_rate": 0.0001595742261198983,
      "loss": 3.3,
      "step": 216840
    },
    {
      "epoch": 1.1629418840225252,
      "grad_norm": 0.40560951828956604,
      "learning_rate": 0.00015957236179780308,
      "loss": 3.2432,
      "step": 216850
    },
    {
      "epoch": 1.163016456967158,
      "grad_norm": 0.3910263180732727,
      "learning_rate": 0.0001595704974757079,
      "loss": 3.2712,
      "step": 216860
    },
    {
      "epoch": 1.1630910299117907,
      "grad_norm": 0.39927244186401367,
      "learning_rate": 0.0001595686331536127,
      "loss": 3.245,
      "step": 216870
    },
    {
      "epoch": 1.1631656028564235,
      "grad_norm": 0.385522723197937,
      "learning_rate": 0.00015956676883151748,
      "loss": 3.2603,
      "step": 216880
    },
    {
      "epoch": 1.1632401758010562,
      "grad_norm": 0.38268032670021057,
      "learning_rate": 0.0001595649045094223,
      "loss": 3.2573,
      "step": 216890
    },
    {
      "epoch": 1.163314748745689,
      "grad_norm": 0.3909124732017517,
      "learning_rate": 0.0001595630401873271,
      "loss": 3.2736,
      "step": 216900
    },
    {
      "epoch": 1.1633893216903217,
      "grad_norm": 0.3663279414176941,
      "learning_rate": 0.0001595611758652319,
      "loss": 3.1914,
      "step": 216910
    },
    {
      "epoch": 1.1634638946349543,
      "grad_norm": 0.37704843282699585,
      "learning_rate": 0.0001595593115431367,
      "loss": 3.3106,
      "step": 216920
    },
    {
      "epoch": 1.163538467579587,
      "grad_norm": 0.3896007537841797,
      "learning_rate": 0.00015955744722104148,
      "loss": 3.2946,
      "step": 216930
    },
    {
      "epoch": 1.1636130405242198,
      "grad_norm": 0.3960455656051636,
      "learning_rate": 0.0001595555828989463,
      "loss": 3.2205,
      "step": 216940
    },
    {
      "epoch": 1.1636876134688525,
      "grad_norm": 0.3768599331378937,
      "learning_rate": 0.0001595537185768511,
      "loss": 3.2907,
      "step": 216950
    },
    {
      "epoch": 1.1637621864134853,
      "grad_norm": 0.37588241696357727,
      "learning_rate": 0.0001595518542547559,
      "loss": 3.1899,
      "step": 216960
    },
    {
      "epoch": 1.163836759358118,
      "grad_norm": 0.3810625672340393,
      "learning_rate": 0.0001595499899326607,
      "loss": 3.2606,
      "step": 216970
    },
    {
      "epoch": 1.1639113323027508,
      "grad_norm": 0.37495166063308716,
      "learning_rate": 0.00015954812561056549,
      "loss": 3.2594,
      "step": 216980
    },
    {
      "epoch": 1.1639859052473835,
      "grad_norm": 0.3629707992076874,
      "learning_rate": 0.0001595462612884703,
      "loss": 3.2396,
      "step": 216990
    },
    {
      "epoch": 1.1640604781920163,
      "grad_norm": 0.3611636161804199,
      "learning_rate": 0.0001595443969663751,
      "loss": 3.2141,
      "step": 217000
    },
    {
      "epoch": 1e-05,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001565982,
      "loss": 3.4226,
      "step": 217010
    },
    {
      "epoch": 2e-05,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00015659620000000002,
      "loss": 2.9011,
      "step": 217020
    },
    {
      "epoch": 3e-05,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001565942,
      "loss": 3.1424,
      "step": 217030
    },
    {
      "epoch": 4e-05,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001565922,
      "loss": 3.5469,
      "step": 217040
    },
    {
      "epoch": 5e-05,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001565902,
      "loss": 2.8658,
      "step": 217050
    },
    {
      "epoch": 6e-05,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001565882,
      "loss": 3.4458,
      "step": 217060
    },
    {
      "epoch": 7e-05,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015658620000000002,
      "loss": 2.7824,
      "step": 217070
    },
    {
      "epoch": 8e-05,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001565842,
      "loss": 3.5382,
      "step": 217080
    },
    {
      "epoch": 9e-05,
      "grad_norm": 2.046875,
      "learning_rate": 0.0001565822,
      "loss": 3.0843,
      "step": 217090
    },
    {
      "epoch": 0.0001,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001565802,
      "loss": 2.3826,
      "step": 217100
    },
    {
      "epoch": 0.00011,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001565782,
      "loss": 2.9621,
      "step": 217110
    },
    {
      "epoch": 0.00012,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015657620000000003,
      "loss": 3.2387,
      "step": 217120
    },
    {
      "epoch": 0.00013,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001565742,
      "loss": 3.397,
      "step": 217130
    },
    {
      "epoch": 0.00014,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015657220000000002,
      "loss": 3.3995,
      "step": 217140
    },
    {
      "epoch": 0.00015,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001565702,
      "loss": 3.0792,
      "step": 217150
    },
    {
      "epoch": 0.00016,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001565682,
      "loss": 3.2845,
      "step": 217160
    },
    {
      "epoch": 0.00017,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001565662,
      "loss": 2.988,
      "step": 217170
    },
    {
      "epoch": 0.00018,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001565642,
      "loss": 2.9677,
      "step": 217180
    },
    {
      "epoch": 0.00019,
      "grad_norm": 1.109375,
      "learning_rate": 0.00015656220000000002,
      "loss": 3.0822,
      "step": 217190
    },
    {
      "epoch": 0.0002,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001565602,
      "loss": 3.0493,
      "step": 217200
    },
    {
      "epoch": 0.00021,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00015655820000000001,
      "loss": 3.3909,
      "step": 217210
    },
    {
      "epoch": 0.00022,
      "grad_norm": 1.78125,
      "learning_rate": 0.0001565562,
      "loss": 3.2493,
      "step": 217220
    },
    {
      "epoch": 0.00023,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001565542,
      "loss": 3.0252,
      "step": 217230
    },
    {
      "epoch": 0.00024,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00015655220000000003,
      "loss": 3.0482,
      "step": 217240
    },
    {
      "epoch": 0.00025,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001565502,
      "loss": 3.388,
      "step": 217250
    },
    {
      "epoch": 0.00026,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015654820000000002,
      "loss": 3.1919,
      "step": 217260
    },
    {
      "epoch": 0.00027,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001565462,
      "loss": 1.8602,
      "step": 217270
    },
    {
      "epoch": 0.00028,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001565442,
      "loss": 2.6946,
      "step": 217280
    },
    {
      "epoch": 0.00029,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001565422,
      "loss": 2.708,
      "step": 217290
    },
    {
      "epoch": 0.0003,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001565402,
      "loss": 3.7701,
      "step": 217300
    },
    {
      "epoch": 0.00031,
      "grad_norm": 1.484375,
      "learning_rate": 0.00015653820000000003,
      "loss": 3.1889,
      "step": 217310
    },
    {
      "epoch": 0.00032,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001565362,
      "loss": 3.2233,
      "step": 217320
    },
    {
      "epoch": 0.00033,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015653420000000002,
      "loss": 3.4142,
      "step": 217330
    },
    {
      "epoch": 0.00034,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001565322,
      "loss": 3.1823,
      "step": 217340
    },
    {
      "epoch": 0.00035,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001565302,
      "loss": 3.1605,
      "step": 217350
    },
    {
      "epoch": 0.00036,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001565282,
      "loss": 3.1627,
      "step": 217360
    },
    {
      "epoch": 0.00037,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001565262,
      "loss": 2.8767,
      "step": 217370
    },
    {
      "epoch": 0.00038,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015652420000000002,
      "loss": 3.2323,
      "step": 217380
    },
    {
      "epoch": 0.00039,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001565222,
      "loss": 3.2541,
      "step": 217390
    },
    {
      "epoch": 0.0004,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001565202,
      "loss": 2.668,
      "step": 217400
    },
    {
      "epoch": 0.00041,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001565182,
      "loss": 2.9357,
      "step": 217410
    },
    {
      "epoch": 0.00042,
      "grad_norm": 0.75,
      "learning_rate": 0.0001565162,
      "loss": 2.7239,
      "step": 217420
    },
    {
      "epoch": 0.00043,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001565142,
      "loss": 3.1095,
      "step": 217430
    },
    {
      "epoch": 0.00044,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001565122,
      "loss": 3.3871,
      "step": 217440
    },
    {
      "epoch": 0.00045,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015651020000000002,
      "loss": 2.9161,
      "step": 217450
    },
    {
      "epoch": 0.00046,
      "grad_norm": 1.078125,
      "learning_rate": 0.00015650820000000001,
      "loss": 2.6243,
      "step": 217460
    },
    {
      "epoch": 0.00047,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001565062,
      "loss": 2.8778,
      "step": 217470
    },
    {
      "epoch": 0.00048,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001565042,
      "loss": 3.3637,
      "step": 217480
    },
    {
      "epoch": 0.00049,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001565022,
      "loss": 2.6579,
      "step": 217490
    },
    {
      "epoch": 0.0005,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001565002,
      "loss": 2.9816,
      "step": 217500
    },
    {
      "epoch": 1.8643236158183664e-05,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015649820000000002,
      "loss": 3.1213,
      "step": 217510
    },
    {
      "epoch": 3.728647231636733e-05,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015649620000000001,
      "loss": 3.1743,
      "step": 217520
    },
    {
      "epoch": 5.592970847455099e-05,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001564942,
      "loss": 3.2243,
      "step": 217530
    },
    {
      "epoch": 7.457294463273465e-05,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001564922,
      "loss": 3.2261,
      "step": 217540
    },
    {
      "epoch": 9.321618079091832e-05,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001564902,
      "loss": 3.2451,
      "step": 217550
    },
    {
      "epoch": 0.00011185941694910198,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015648820000000002,
      "loss": 3.1793,
      "step": 217560
    },
    {
      "epoch": 0.00013050265310728565,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001564862,
      "loss": 3.164,
      "step": 217570
    },
    {
      "epoch": 0.0001491458892654693,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015648420000000002,
      "loss": 3.1636,
      "step": 217580
    },
    {
      "epoch": 0.00016778912542365297,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001564822,
      "loss": 3.1867,
      "step": 217590
    },
    {
      "epoch": 0.00018643236158183665,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001564802,
      "loss": 3.2631,
      "step": 217600
    },
    {
      "epoch": 0.0002050755977400203,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001564782,
      "loss": 3.2064,
      "step": 217610
    },
    {
      "epoch": 0.00022371883389820396,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001564762,
      "loss": 3.1391,
      "step": 217620
    },
    {
      "epoch": 0.00024236207005638762,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015647420000000002,
      "loss": 3.1729,
      "step": 217630
    },
    {
      "epoch": 0.0002610053062145713,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001564722,
      "loss": 3.1674,
      "step": 217640
    },
    {
      "epoch": 0.00027964854237275496,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001564702,
      "loss": 3.2007,
      "step": 217650
    },
    {
      "epoch": 0.0002982917785309386,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001564682,
      "loss": 3.3106,
      "step": 217660
    },
    {
      "epoch": 0.0003169350146891223,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001564662,
      "loss": 3.1071,
      "step": 217670
    },
    {
      "epoch": 0.00033557825084730593,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015646420000000003,
      "loss": 3.2253,
      "step": 217680
    },
    {
      "epoch": 0.00035422148700548964,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001564622,
      "loss": 3.1753,
      "step": 217690
    },
    {
      "epoch": 0.0003728647231636733,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015646020000000002,
      "loss": 3.2087,
      "step": 217700
    },
    {
      "epoch": 0.00039150795932185696,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001564582,
      "loss": 3.183,
      "step": 217710
    },
    {
      "epoch": 0.0004101511954800406,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001564562,
      "loss": 3.065,
      "step": 217720
    },
    {
      "epoch": 0.00042879443163822427,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015645420000000003,
      "loss": 3.2313,
      "step": 217730
    },
    {
      "epoch": 0.0004474376677964079,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001564522,
      "loss": 3.1268,
      "step": 217740
    },
    {
      "epoch": 0.0004660809039545916,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015645020000000002,
      "loss": 3.2973,
      "step": 217750
    },
    {
      "epoch": 0.00048472414011277524,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001564482,
      "loss": 3.1162,
      "step": 217760
    },
    {
      "epoch": 0.000503367376270959,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001564462,
      "loss": 3.1388,
      "step": 217770
    },
    {
      "epoch": 0.0005220106124291426,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001564442,
      "loss": 3.1618,
      "step": 217780
    },
    {
      "epoch": 0.0005406538485873263,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001564422,
      "loss": 3.0818,
      "step": 217790
    },
    {
      "epoch": 0.0005592970847455099,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015644020000000003,
      "loss": 3.194,
      "step": 217800
    },
    {
      "epoch": 0.0005779403209036936,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001564382,
      "loss": 3.1221,
      "step": 217810
    },
    {
      "epoch": 0.0005965835570618772,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015643620000000002,
      "loss": 3.0563,
      "step": 217820
    },
    {
      "epoch": 0.0006152267932200609,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001564342,
      "loss": 3.2164,
      "step": 217830
    },
    {
      "epoch": 0.0006338700293782445,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001564322,
      "loss": 3.2609,
      "step": 217840
    },
    {
      "epoch": 0.0006525132655364282,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001564302,
      "loss": 3.1828,
      "step": 217850
    },
    {
      "epoch": 0.0006711565016946119,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001564282,
      "loss": 3.2498,
      "step": 217860
    },
    {
      "epoch": 0.0006897997378527955,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015642620000000002,
      "loss": 3.11,
      "step": 217870
    },
    {
      "epoch": 0.0007084429740109793,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001564242,
      "loss": 3.1502,
      "step": 217880
    },
    {
      "epoch": 0.0007270862101691629,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015642220000000001,
      "loss": 3.2752,
      "step": 217890
    },
    {
      "epoch": 0.0007457294463273466,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001564202,
      "loss": 3.1513,
      "step": 217900
    },
    {
      "epoch": 0.0007643726824855303,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001564182,
      "loss": 3.1744,
      "step": 217910
    },
    {
      "epoch": 0.0007830159186437139,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001564162,
      "loss": 3.2195,
      "step": 217920
    },
    {
      "epoch": 0.0008016591548018976,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001564142,
      "loss": 3.0627,
      "step": 217930
    },
    {
      "epoch": 0.0008203023909600812,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015641220000000002,
      "loss": 3.171,
      "step": 217940
    },
    {
      "epoch": 0.0008389456271182649,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015641020000000002,
      "loss": 3.1295,
      "step": 217950
    },
    {
      "epoch": 0.0008575888632764485,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001564082,
      "loss": 3.1571,
      "step": 217960
    },
    {
      "epoch": 0.0008762320994346322,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001564062,
      "loss": 3.2202,
      "step": 217970
    },
    {
      "epoch": 0.0008948753355928159,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001564042,
      "loss": 3.2075,
      "step": 217980
    },
    {
      "epoch": 0.0009135185717509995,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001564022,
      "loss": 3.017,
      "step": 217990
    },
    {
      "epoch": 0.0009321618079091832,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001564002,
      "loss": 3.2945,
      "step": 218000
    },
    {
      "epoch": 0.0009508050440673668,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015639820000000002,
      "loss": 3.1543,
      "step": 218010
    },
    {
      "epoch": 0.0009694482802255505,
      "grad_norm": 0.75,
      "learning_rate": 0.0001563962,
      "loss": 3.3418,
      "step": 218020
    },
    {
      "epoch": 0.0009880915163837341,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001563942,
      "loss": 3.2928,
      "step": 218030
    },
    {
      "epoch": 0.001006734752541918,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001563922,
      "loss": 3.0912,
      "step": 218040
    },
    {
      "epoch": 0.0010253779887001014,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001563902,
      "loss": 3.1381,
      "step": 218050
    },
    {
      "epoch": 0.0010440212248582852,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001563882,
      "loss": 3.0858,
      "step": 218060
    },
    {
      "epoch": 0.0010626644610164688,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015638620000000002,
      "loss": 3.0539,
      "step": 218070
    },
    {
      "epoch": 0.0010813076971746525,
      "grad_norm": 0.75,
      "learning_rate": 0.0001563842,
      "loss": 3.1225,
      "step": 218080
    },
    {
      "epoch": 0.001099950933332836,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001563822,
      "loss": 3.1952,
      "step": 218090
    },
    {
      "epoch": 0.0011185941694910198,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001563802,
      "loss": 3.1848,
      "step": 218100
    },
    {
      "epoch": 0.0011372374056492036,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001563782,
      "loss": 3.0716,
      "step": 218110
    },
    {
      "epoch": 0.0011558806418073872,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015637620000000002,
      "loss": 3.1712,
      "step": 218120
    },
    {
      "epoch": 0.001174523877965571,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001563742,
      "loss": 3.083,
      "step": 218130
    },
    {
      "epoch": 0.0011931671141237545,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015637220000000001,
      "loss": 3.2466,
      "step": 218140
    },
    {
      "epoch": 0.0012118103502819382,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001563702,
      "loss": 3.2464,
      "step": 218150
    },
    {
      "epoch": 0.0012304535864401218,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001563682,
      "loss": 3.171,
      "step": 218160
    },
    {
      "epoch": 0.0012490968225983055,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015636620000000003,
      "loss": 3.2805,
      "step": 218170
    },
    {
      "epoch": 0.001267740058756489,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001563642,
      "loss": 3.173,
      "step": 218180
    },
    {
      "epoch": 0.0012863832949146729,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015636220000000002,
      "loss": 3.1231,
      "step": 218190
    },
    {
      "epoch": 0.0013050265310728564,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015636020000000001,
      "loss": 3.1545,
      "step": 218200
    },
    {
      "epoch": 0.0013236697672310402,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001563582,
      "loss": 3.2185,
      "step": 218210
    },
    {
      "epoch": 0.0013423130033892237,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001563562,
      "loss": 3.2478,
      "step": 218220
    },
    {
      "epoch": 0.0013609562395474075,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001563542,
      "loss": 3.0021,
      "step": 218230
    },
    {
      "epoch": 0.001379599475705591,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015635220000000002,
      "loss": 3.2357,
      "step": 218240
    },
    {
      "epoch": 0.0013982427118637748,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001563502,
      "loss": 3.2237,
      "step": 218250
    },
    {
      "epoch": 0.0014168859480219586,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015634820000000002,
      "loss": 3.3615,
      "step": 218260
    },
    {
      "epoch": 0.0014355291841801421,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001563462,
      "loss": 3.2233,
      "step": 218270
    },
    {
      "epoch": 0.0014541724203383259,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001563442,
      "loss": 3.1767,
      "step": 218280
    },
    {
      "epoch": 0.0014728156564965094,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015634220000000003,
      "loss": 3.1242,
      "step": 218290
    },
    {
      "epoch": 0.0014914588926546932,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001563402,
      "loss": 3.1963,
      "step": 218300
    },
    {
      "epoch": 0.0015101021288128767,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015633820000000002,
      "loss": 3.3268,
      "step": 218310
    },
    {
      "epoch": 0.0015287453649710605,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001563362,
      "loss": 3.1486,
      "step": 218320
    },
    {
      "epoch": 0.001547388601129244,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001563342,
      "loss": 3.1682,
      "step": 218330
    },
    {
      "epoch": 0.0015660318372874278,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001563322,
      "loss": 3.0878,
      "step": 218340
    },
    {
      "epoch": 0.0015846750734456114,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001563302,
      "loss": 3.0109,
      "step": 218350
    },
    {
      "epoch": 0.0016033183096037951,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015632820000000003,
      "loss": 3.1411,
      "step": 218360
    },
    {
      "epoch": 0.0016219615457619787,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001563262,
      "loss": 3.1244,
      "step": 218370
    },
    {
      "epoch": 0.0016406047819201624,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015632420000000002,
      "loss": 3.1713,
      "step": 218380
    },
    {
      "epoch": 0.001659248018078346,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015632219999999999,
      "loss": 3.2125,
      "step": 218390
    },
    {
      "epoch": 0.0016778912542365298,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001563202,
      "loss": 3.1604,
      "step": 218400
    },
    {
      "epoch": 0.0016965344903947135,
      "grad_norm": 1.75,
      "learning_rate": 0.0001563182,
      "loss": 3.1855,
      "step": 218410
    },
    {
      "epoch": 0.001715177726552897,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001563162,
      "loss": 3.1688,
      "step": 218420
    },
    {
      "epoch": 0.0017338209627110808,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015631420000000002,
      "loss": 3.2898,
      "step": 218430
    },
    {
      "epoch": 0.0017524641988692644,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001563122,
      "loss": 3.1535,
      "step": 218440
    },
    {
      "epoch": 0.0017711074350274482,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001563102,
      "loss": 3.2824,
      "step": 218450
    },
    {
      "epoch": 0.0017897506711856317,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001563082,
      "loss": 3.1075,
      "step": 218460
    },
    {
      "epoch": 0.0018083939073438155,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001563062,
      "loss": 3.2357,
      "step": 218470
    },
    {
      "epoch": 0.001827037143501999,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001563042,
      "loss": 3.1239,
      "step": 218480
    },
    {
      "epoch": 0.0018456803796601828,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001563022,
      "loss": 3.1921,
      "step": 218490
    },
    {
      "epoch": 0.0018643236158183663,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015630020000000002,
      "loss": 3.2015,
      "step": 218500
    },
    {
      "epoch": 0.00188296685197655,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00015629820000000001,
      "loss": 3.1303,
      "step": 218510
    },
    {
      "epoch": 0.0019016100881347336,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001562962,
      "loss": 3.1601,
      "step": 218520
    },
    {
      "epoch": 0.0019202533242929174,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001562942,
      "loss": 3.1924,
      "step": 218530
    },
    {
      "epoch": 0.001938896560451101,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001562922,
      "loss": 3.2374,
      "step": 218540
    },
    {
      "epoch": 0.001957539796609285,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001562902,
      "loss": 3.1592,
      "step": 218550
    },
    {
      "epoch": 0.0019761830327674683,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015628820000000002,
      "loss": 3.2692,
      "step": 218560
    },
    {
      "epoch": 0.001994826268925652,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015628620000000001,
      "loss": 3.1443,
      "step": 218570
    },
    {
      "epoch": 0.002013469505083836,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001562842,
      "loss": 3.147,
      "step": 218580
    },
    {
      "epoch": 0.0020321127412420196,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001562822,
      "loss": 3.1767,
      "step": 218590
    },
    {
      "epoch": 0.002050755977400203,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001562802,
      "loss": 3.165,
      "step": 218600
    },
    {
      "epoch": 0.0020693992135583867,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001562782,
      "loss": 3.1949,
      "step": 218610
    },
    {
      "epoch": 0.0020880424497165704,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001562762,
      "loss": 3.1782,
      "step": 218620
    },
    {
      "epoch": 0.002106685685874754,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015627420000000002,
      "loss": 3.2678,
      "step": 218630
    },
    {
      "epoch": 0.0021253289220329375,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001562722,
      "loss": 3.2287,
      "step": 218640
    },
    {
      "epoch": 0.0021439721581911213,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001562702,
      "loss": 3.1203,
      "step": 218650
    },
    {
      "epoch": 0.002162615394349305,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001562682,
      "loss": 3.1869,
      "step": 218660
    },
    {
      "epoch": 0.002181258630507489,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001562662,
      "loss": 3.0919,
      "step": 218670
    },
    {
      "epoch": 0.002199901866665672,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015626420000000002,
      "loss": 3.149,
      "step": 218680
    },
    {
      "epoch": 0.002218545102823856,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015626220000000002,
      "loss": 3.2064,
      "step": 218690
    },
    {
      "epoch": 0.0022371883389820397,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001562602,
      "loss": 3.1707,
      "step": 218700
    },
    {
      "epoch": 0.0022558315751402234,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001562582,
      "loss": 3.1673,
      "step": 218710
    },
    {
      "epoch": 0.002274474811298407,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001562562,
      "loss": 3.1548,
      "step": 218720
    },
    {
      "epoch": 0.0022931180474565905,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015625420000000003,
      "loss": 3.2118,
      "step": 218730
    },
    {
      "epoch": 0.0023117612836147743,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001562522,
      "loss": 3.1303,
      "step": 218740
    },
    {
      "epoch": 0.002330404519772958,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015625020000000002,
      "loss": 3.2335,
      "step": 218750
    },
    {
      "epoch": 0.002349047755931142,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001562482,
      "loss": 3.191,
      "step": 218760
    },
    {
      "epoch": 0.002367690992089325,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001562462,
      "loss": 3.0736,
      "step": 218770
    },
    {
      "epoch": 0.002386334228247509,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015624420000000003,
      "loss": 3.2155,
      "step": 218780
    },
    {
      "epoch": 0.0024049774644056927,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001562422,
      "loss": 3.1597,
      "step": 218790
    },
    {
      "epoch": 0.0024236207005638765,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015624020000000002,
      "loss": 3.1325,
      "step": 218800
    },
    {
      "epoch": 0.00244226393672206,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001562382,
      "loss": 3.2581,
      "step": 218810
    },
    {
      "epoch": 0.0024609071728802436,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015623620000000001,
      "loss": 3.2195,
      "step": 218820
    },
    {
      "epoch": 0.0024795504090384273,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001562342,
      "loss": 3.1785,
      "step": 218830
    },
    {
      "epoch": 0.002498193645196611,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001562322,
      "loss": 3.0422,
      "step": 218840
    },
    {
      "epoch": 0.002516836881354795,
      "grad_norm": 0.75,
      "learning_rate": 0.00015623020000000003,
      "loss": 3.1544,
      "step": 218850
    },
    {
      "epoch": 0.002535480117512978,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001562282,
      "loss": 3.0149,
      "step": 218860
    },
    {
      "epoch": 0.002554123353671162,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015622620000000002,
      "loss": 3.1015,
      "step": 218870
    },
    {
      "epoch": 0.0025727665898293457,
      "grad_norm": 0.75,
      "learning_rate": 0.0001562242,
      "loss": 3.2013,
      "step": 218880
    },
    {
      "epoch": 0.0025914098259875295,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001562222,
      "loss": 3.268,
      "step": 218890
    },
    {
      "epoch": 0.002610053062145713,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001562202,
      "loss": 3.0656,
      "step": 218900
    },
    {
      "epoch": 0.0026286962983038966,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001562182,
      "loss": 3.2027,
      "step": 218910
    },
    {
      "epoch": 0.0026473395344620804,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015621620000000002,
      "loss": 3.1543,
      "step": 218920
    },
    {
      "epoch": 0.002665982770620264,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001562142,
      "loss": 3.1494,
      "step": 218930
    },
    {
      "epoch": 0.0026846260067784474,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015621220000000002,
      "loss": 3.1269,
      "step": 218940
    },
    {
      "epoch": 0.002703269242936631,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001562102,
      "loss": 3.2299,
      "step": 218950
    },
    {
      "epoch": 0.002721912479094815,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001562082,
      "loss": 3.0885,
      "step": 218960
    },
    {
      "epoch": 0.0027405557152529987,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001562062,
      "loss": 3.2048,
      "step": 218970
    },
    {
      "epoch": 0.002759198951411182,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001562042,
      "loss": 3.17,
      "step": 218980
    },
    {
      "epoch": 0.002777842187569366,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015620220000000002,
      "loss": 3.1317,
      "step": 218990
    },
    {
      "epoch": 0.0027964854237275496,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001562002,
      "loss": 3.1465,
      "step": 219000
    },
    {
      "epoch": 0.0028151286598857334,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001561982,
      "loss": 3.1374,
      "step": 219010
    },
    {
      "epoch": 0.002833771896043917,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001561962,
      "loss": 3.0707,
      "step": 219020
    },
    {
      "epoch": 0.0028524151322021005,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001561942,
      "loss": 3.0878,
      "step": 219030
    },
    {
      "epoch": 0.0028710583683602842,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001561922,
      "loss": 3.0461,
      "step": 219040
    },
    {
      "epoch": 0.002889701604518468,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001561902,
      "loss": 3.328,
      "step": 219050
    },
    {
      "epoch": 0.0029083448406766518,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015618820000000002,
      "loss": 3.1394,
      "step": 219060
    },
    {
      "epoch": 0.002926988076834835,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001561862,
      "loss": 3.1302,
      "step": 219070
    },
    {
      "epoch": 0.002945631312993019,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001561842,
      "loss": 3.1088,
      "step": 219080
    },
    {
      "epoch": 0.0029642745491512026,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001561822,
      "loss": 3.1697,
      "step": 219090
    },
    {
      "epoch": 0.0029829177853093864,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001561802,
      "loss": 3.2344,
      "step": 219100
    },
    {
      "epoch": 0.0030015610214675697,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001561782,
      "loss": 2.9618,
      "step": 219110
    },
    {
      "epoch": 0.0030202042576257535,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015617620000000002,
      "loss": 3.2274,
      "step": 219120
    },
    {
      "epoch": 0.0030388474937839373,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001561742,
      "loss": 3.2077,
      "step": 219130
    },
    {
      "epoch": 0.003057490729942121,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001561722,
      "loss": 3.1006,
      "step": 219140
    },
    {
      "epoch": 0.0030761339661003048,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001561702,
      "loss": 3.2034,
      "step": 219150
    },
    {
      "epoch": 0.003094777202258488,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001561682,
      "loss": 3.2723,
      "step": 219160
    },
    {
      "epoch": 0.003113420438416672,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015616620000000002,
      "loss": 3.147,
      "step": 219170
    },
    {
      "epoch": 0.0031320636745748556,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015616420000000002,
      "loss": 3.1696,
      "step": 219180
    },
    {
      "epoch": 0.0031507069107330394,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00015616220000000001,
      "loss": 3.1999,
      "step": 219190
    },
    {
      "epoch": 0.0031693501468912227,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001561602,
      "loss": 3.2951,
      "step": 219200
    },
    {
      "epoch": 0.0031879933830494065,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001561582,
      "loss": 3.2152,
      "step": 219210
    },
    {
      "epoch": 0.0032066366192075903,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001561562,
      "loss": 3.0499,
      "step": 219220
    },
    {
      "epoch": 0.003225279855365774,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001561542,
      "loss": 3.323,
      "step": 219230
    },
    {
      "epoch": 0.0032439230915239574,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015615220000000002,
      "loss": 3.2816,
      "step": 219240
    },
    {
      "epoch": 0.003262566327682141,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015615020000000001,
      "loss": 3.1913,
      "step": 219250
    },
    {
      "epoch": 0.003281209563840325,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001561482,
      "loss": 3.2484,
      "step": 219260
    },
    {
      "epoch": 0.0032998527999985087,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001561462,
      "loss": 3.2281,
      "step": 219270
    },
    {
      "epoch": 0.003318496036156692,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001561442,
      "loss": 3.0379,
      "step": 219280
    },
    {
      "epoch": 0.0033371392723148758,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015614220000000002,
      "loss": 3.1854,
      "step": 219290
    },
    {
      "epoch": 0.0033557825084730595,
      "grad_norm": 0.75,
      "learning_rate": 0.0001561402,
      "loss": 3.2471,
      "step": 219300
    },
    {
      "epoch": 0.0033744257446312433,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015613820000000002,
      "loss": 3.2126,
      "step": 219310
    },
    {
      "epoch": 0.003393068980789427,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001561362,
      "loss": 3.0806,
      "step": 219320
    },
    {
      "epoch": 0.0034117122169476104,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001561342,
      "loss": 3.2207,
      "step": 219330
    },
    {
      "epoch": 0.003430355453105794,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015613220000000003,
      "loss": 3.1933,
      "step": 219340
    },
    {
      "epoch": 0.003448998689263978,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001561302,
      "loss": 3.2437,
      "step": 219350
    },
    {
      "epoch": 0.0034676419254221617,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015612820000000002,
      "loss": 3.1664,
      "step": 219360
    },
    {
      "epoch": 0.003486285161580345,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001561262,
      "loss": 3.2325,
      "step": 219370
    },
    {
      "epoch": 0.0035049283977385288,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001561242,
      "loss": 3.1932,
      "step": 219380
    },
    {
      "epoch": 0.0035235716338967125,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001561222,
      "loss": 3.1506,
      "step": 219390
    },
    {
      "epoch": 0.0035422148700548963,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001561202,
      "loss": 3.0993,
      "step": 219400
    },
    {
      "epoch": 0.0035608581062130796,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015611820000000003,
      "loss": 3.1105,
      "step": 219410
    },
    {
      "epoch": 0.0035795013423712634,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001561162,
      "loss": 3.1483,
      "step": 219420
    },
    {
      "epoch": 0.003598144578529447,
      "grad_norm": 0.625,
      "learning_rate": 0.00015611420000000002,
      "loss": 3.1011,
      "step": 219430
    },
    {
      "epoch": 0.003616787814687631,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015611219999999999,
      "loss": 3.2298,
      "step": 219440
    },
    {
      "epoch": 0.0036354310508458147,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001561102,
      "loss": 3.1792,
      "step": 219450
    },
    {
      "epoch": 0.003654074287003998,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001561082,
      "loss": 3.1936,
      "step": 219460
    },
    {
      "epoch": 0.003672717523162182,
      "grad_norm": 0.625,
      "learning_rate": 0.0001561062,
      "loss": 3.243,
      "step": 219470
    },
    {
      "epoch": 0.0036913607593203656,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015610420000000002,
      "loss": 3.0836,
      "step": 219480
    },
    {
      "epoch": 0.0037100039954785493,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001561022,
      "loss": 3.0788,
      "step": 219490
    },
    {
      "epoch": 0.0037286472316367327,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015610020000000001,
      "loss": 3.1143,
      "step": 219500
    },
    {
      "epoch": 0.0037472904677949164,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001560982,
      "loss": 3.1828,
      "step": 219510
    },
    {
      "epoch": 0.0037659337039531,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001560962,
      "loss": 3.277,
      "step": 219520
    },
    {
      "epoch": 0.003784576940111284,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001560942,
      "loss": 3.3055,
      "step": 219530
    },
    {
      "epoch": 0.0038032201762694673,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001560922,
      "loss": 3.0549,
      "step": 219540
    },
    {
      "epoch": 0.003821863412427651,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015609020000000002,
      "loss": 3.1857,
      "step": 219550
    },
    {
      "epoch": 0.003840506648585835,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015608820000000001,
      "loss": 3.2028,
      "step": 219560
    },
    {
      "epoch": 0.0038591498847440186,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001560862,
      "loss": 3.1085,
      "step": 219570
    },
    {
      "epoch": 0.003877793120902202,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001560842,
      "loss": 3.3137,
      "step": 219580
    },
    {
      "epoch": 0.0038964363570603857,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001560822,
      "loss": 3.1724,
      "step": 219590
    },
    {
      "epoch": 0.00391507959321857,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001560802,
      "loss": 3.2347,
      "step": 219600
    },
    {
      "epoch": 0.003933722829376753,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001560782,
      "loss": 3.1991,
      "step": 219610
    },
    {
      "epoch": 0.0039523660655349365,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015607620000000002,
      "loss": 3.2151,
      "step": 219620
    },
    {
      "epoch": 0.00397100930169312,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001560742,
      "loss": 3.2389,
      "step": 219630
    },
    {
      "epoch": 0.003989652537851304,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001560722,
      "loss": 3.19,
      "step": 219640
    },
    {
      "epoch": 0.004008295774009488,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001560702,
      "loss": 3.1906,
      "step": 219650
    },
    {
      "epoch": 0.004026939010167672,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001560682,
      "loss": 3.0947,
      "step": 219660
    },
    {
      "epoch": 0.004045582246325855,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015606620000000002,
      "loss": 3.0743,
      "step": 219670
    },
    {
      "epoch": 0.004064225482484039,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015606420000000002,
      "loss": 3.1976,
      "step": 219680
    },
    {
      "epoch": 0.004082868718642222,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001560622,
      "loss": 3.2345,
      "step": 219690
    },
    {
      "epoch": 0.004101511954800406,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001560602,
      "loss": 3.1786,
      "step": 219700
    },
    {
      "epoch": 0.0041201551909585896,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001560582,
      "loss": 3.0247,
      "step": 219710
    },
    {
      "epoch": 0.004138798427116773,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001560562,
      "loss": 3.2157,
      "step": 219720
    },
    {
      "epoch": 0.004157441663274957,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015605420000000002,
      "loss": 3.0518,
      "step": 219730
    },
    {
      "epoch": 0.004176084899433141,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015605220000000002,
      "loss": 3.2051,
      "step": 219740
    },
    {
      "epoch": 0.004194728135591325,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001560502,
      "loss": 3.1331,
      "step": 219750
    },
    {
      "epoch": 0.004213371371749508,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001560482,
      "loss": 3.0942,
      "step": 219760
    },
    {
      "epoch": 0.004232014607907692,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001560462,
      "loss": 3.1671,
      "step": 219770
    },
    {
      "epoch": 0.004250657844065875,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015604420000000003,
      "loss": 3.1989,
      "step": 219780
    },
    {
      "epoch": 0.004269301080224059,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001560422,
      "loss": 3.1055,
      "step": 219790
    },
    {
      "epoch": 0.004287944316382243,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015604020000000002,
      "loss": 3.2084,
      "step": 219800
    },
    {
      "epoch": 0.004306587552540426,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001560382,
      "loss": 3.1696,
      "step": 219810
    },
    {
      "epoch": 0.00432523078869861,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001560362,
      "loss": 3.0872,
      "step": 219820
    },
    {
      "epoch": 0.004343874024856794,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001560342,
      "loss": 3.1764,
      "step": 219830
    },
    {
      "epoch": 0.004362517261014978,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001560322,
      "loss": 3.2748,
      "step": 219840
    },
    {
      "epoch": 0.004381160497173161,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015603020000000002,
      "loss": 3.2139,
      "step": 219850
    },
    {
      "epoch": 0.004399803733331344,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001560282,
      "loss": 3.192,
      "step": 219860
    },
    {
      "epoch": 0.004418446969489528,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015602620000000001,
      "loss": 3.4071,
      "step": 219870
    },
    {
      "epoch": 0.004437090205647712,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001560242,
      "loss": 3.1773,
      "step": 219880
    },
    {
      "epoch": 0.004455733441805896,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001560222,
      "loss": 3.137,
      "step": 219890
    },
    {
      "epoch": 0.004474376677964079,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015602020000000003,
      "loss": 3.1852,
      "step": 219900
    },
    {
      "epoch": 0.004493019914122263,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001560182,
      "loss": 3.1612,
      "step": 219910
    },
    {
      "epoch": 0.004511663150280447,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015601620000000002,
      "loss": 3.1361,
      "step": 219920
    },
    {
      "epoch": 0.004530306386438631,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001560142,
      "loss": 3.0585,
      "step": 219930
    },
    {
      "epoch": 0.004548949622596814,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001560122,
      "loss": 3.0998,
      "step": 219940
    },
    {
      "epoch": 0.004567592858754997,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001560102,
      "loss": 3.2037,
      "step": 219950
    },
    {
      "epoch": 0.004586236094913181,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001560082,
      "loss": 3.0325,
      "step": 219960
    },
    {
      "epoch": 0.004604879331071365,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015600620000000002,
      "loss": 3.149,
      "step": 219970
    },
    {
      "epoch": 0.004623522567229549,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001560042,
      "loss": 3.363,
      "step": 219980
    },
    {
      "epoch": 0.004642165803387732,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015600220000000002,
      "loss": 3.1822,
      "step": 219990
    },
    {
      "epoch": 0.004660809039545916,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001560002,
      "loss": 3.0914,
      "step": 220000
    },
    {
      "epoch": 0.0046794522757041,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001559982,
      "loss": 3.212,
      "step": 220010
    },
    {
      "epoch": 0.004698095511862284,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001559962,
      "loss": 3.2451,
      "step": 220020
    },
    {
      "epoch": 0.004716738748020467,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001559942,
      "loss": 3.1324,
      "step": 220030
    },
    {
      "epoch": 0.00473538198417865,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015599220000000002,
      "loss": 3.191,
      "step": 220040
    },
    {
      "epoch": 0.004754025220336834,
      "grad_norm": 0.75,
      "learning_rate": 0.0001559902,
      "loss": 3.2387,
      "step": 220050
    },
    {
      "epoch": 0.004772668456495018,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001559882,
      "loss": 3.1505,
      "step": 220060
    },
    {
      "epoch": 0.004791311692653202,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001559862,
      "loss": 3.2401,
      "step": 220070
    },
    {
      "epoch": 0.004809954928811385,
      "grad_norm": 0.75,
      "learning_rate": 0.0001559842,
      "loss": 3.2593,
      "step": 220080
    },
    {
      "epoch": 0.004828598164969569,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001559822,
      "loss": 3.1412,
      "step": 220090
    },
    {
      "epoch": 0.004847241401127753,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001559802,
      "loss": 3.079,
      "step": 220100
    },
    {
      "epoch": 0.004865884637285937,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015597820000000002,
      "loss": 3.2688,
      "step": 220110
    },
    {
      "epoch": 0.00488452787344412,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001559762,
      "loss": 3.1976,
      "step": 220120
    },
    {
      "epoch": 0.004903171109602303,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001559742,
      "loss": 3.1757,
      "step": 220130
    },
    {
      "epoch": 0.004921814345760487,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001559722,
      "loss": 3.1711,
      "step": 220140
    },
    {
      "epoch": 0.004940457581918671,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001559702,
      "loss": 3.2836,
      "step": 220150
    },
    {
      "epoch": 0.004959100818076855,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001559682,
      "loss": 3.07,
      "step": 220160
    },
    {
      "epoch": 0.004977744054235038,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015596620000000002,
      "loss": 3.1676,
      "step": 220170
    },
    {
      "epoch": 0.004996387290393222,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001559642,
      "loss": 3.189,
      "step": 220180
    },
    {
      "epoch": 0.005015030526551406,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001559622,
      "loss": 3.1422,
      "step": 220190
    },
    {
      "epoch": 0.00503367376270959,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001559602,
      "loss": 3.1444,
      "step": 220200
    },
    {
      "epoch": 0.005052316998867773,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001559582,
      "loss": 3.2739,
      "step": 220210
    },
    {
      "epoch": 0.005070960235025956,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001559562,
      "loss": 3.0126,
      "step": 220220
    },
    {
      "epoch": 0.00508960347118414,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015595420000000002,
      "loss": 3.2192,
      "step": 220230
    },
    {
      "epoch": 0.005108246707342324,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015595220000000001,
      "loss": 3.0638,
      "step": 220240
    },
    {
      "epoch": 0.005126889943500508,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001559502,
      "loss": 3.1992,
      "step": 220250
    },
    {
      "epoch": 0.0051455331796586914,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001559482,
      "loss": 3.1737,
      "step": 220260
    },
    {
      "epoch": 0.005164176415816875,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001559462,
      "loss": 3.1514,
      "step": 220270
    },
    {
      "epoch": 0.005182819651975059,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001559442,
      "loss": 2.9076,
      "step": 220280
    },
    {
      "epoch": 0.005201462888133242,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015594220000000002,
      "loss": 3.1323,
      "step": 220290
    },
    {
      "epoch": 0.005220106124291426,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015594020000000002,
      "loss": 3.1629,
      "step": 220300
    },
    {
      "epoch": 0.005238749360449609,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001559382,
      "loss": 3.1423,
      "step": 220310
    },
    {
      "epoch": 0.005257392596607793,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001559362,
      "loss": 3.1032,
      "step": 220320
    },
    {
      "epoch": 0.005276035832765977,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001559342,
      "loss": 3.2723,
      "step": 220330
    },
    {
      "epoch": 0.005294679068924161,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015593220000000002,
      "loss": 3.2287,
      "step": 220340
    },
    {
      "epoch": 0.0053133223050823445,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001559302,
      "loss": 3.2922,
      "step": 220350
    },
    {
      "epoch": 0.005331965541240528,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015592820000000002,
      "loss": 3.0984,
      "step": 220360
    },
    {
      "epoch": 0.005350608777398712,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001559262,
      "loss": 3.0946,
      "step": 220370
    },
    {
      "epoch": 0.005369252013556895,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001559242,
      "loss": 3.1224,
      "step": 220380
    },
    {
      "epoch": 0.005387895249715079,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015592220000000003,
      "loss": 3.1425,
      "step": 220390
    },
    {
      "epoch": 0.005406538485873262,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001559202,
      "loss": 3.0984,
      "step": 220400
    },
    {
      "epoch": 0.005425181722031446,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015591820000000002,
      "loss": 3.1339,
      "step": 220410
    },
    {
      "epoch": 0.00544382495818963,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001559162,
      "loss": 3.0912,
      "step": 220420
    },
    {
      "epoch": 0.005462468194347814,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001559142,
      "loss": 3.1078,
      "step": 220430
    },
    {
      "epoch": 0.0054811114305059975,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001559122,
      "loss": 3.0728,
      "step": 220440
    },
    {
      "epoch": 0.005499754666664181,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001559102,
      "loss": 3.2131,
      "step": 220450
    },
    {
      "epoch": 0.005518397902822364,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00015590820000000003,
      "loss": 3.0562,
      "step": 220460
    },
    {
      "epoch": 0.005537041138980548,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001559062,
      "loss": 3.1136,
      "step": 220470
    },
    {
      "epoch": 0.005555684375138732,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015590420000000002,
      "loss": 3.1593,
      "step": 220480
    },
    {
      "epoch": 0.0055743276112969154,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015590219999999999,
      "loss": 3.0963,
      "step": 220490
    },
    {
      "epoch": 0.005592970847455099,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001559002,
      "loss": 3.2237,
      "step": 220500
    },
    {
      "epoch": 0.005611614083613283,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001558982,
      "loss": 3.1904,
      "step": 220510
    },
    {
      "epoch": 0.005630257319771467,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001558962,
      "loss": 3.0032,
      "step": 220520
    },
    {
      "epoch": 0.0056489005559296505,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015589420000000002,
      "loss": 3.1614,
      "step": 220530
    },
    {
      "epoch": 0.005667543792087834,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001558922,
      "loss": 3.0742,
      "step": 220540
    },
    {
      "epoch": 0.005686187028246017,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015589020000000001,
      "loss": 2.9686,
      "step": 220550
    },
    {
      "epoch": 0.005704830264404201,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001558882,
      "loss": 3.1681,
      "step": 220560
    },
    {
      "epoch": 0.005723473500562385,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001558862,
      "loss": 3.232,
      "step": 220570
    },
    {
      "epoch": 0.0057421167367205685,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001558842,
      "loss": 3.1151,
      "step": 220580
    },
    {
      "epoch": 0.005760759972878752,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001558822,
      "loss": 3.2148,
      "step": 220590
    },
    {
      "epoch": 0.005779403209036936,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015588020000000002,
      "loss": 3.1659,
      "step": 220600
    },
    {
      "epoch": 0.00579804644519512,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001558782,
      "loss": 3.2747,
      "step": 220610
    },
    {
      "epoch": 0.0058166896813533035,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001558762,
      "loss": 3.1436,
      "step": 220620
    },
    {
      "epoch": 0.005835332917511486,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001558742,
      "loss": 3.3238,
      "step": 220630
    },
    {
      "epoch": 0.00585397615366967,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001558722,
      "loss": 3.1713,
      "step": 220640
    },
    {
      "epoch": 0.005872619389827854,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001558702,
      "loss": 3.1054,
      "step": 220650
    },
    {
      "epoch": 0.005891262625986038,
      "grad_norm": 0.75,
      "learning_rate": 0.0001558682,
      "loss": 3.1045,
      "step": 220660
    },
    {
      "epoch": 0.0059099058621442215,
      "grad_norm": 0.75,
      "learning_rate": 0.00015586620000000002,
      "loss": 3.0944,
      "step": 220670
    },
    {
      "epoch": 0.005928549098302405,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001558642,
      "loss": 3.1539,
      "step": 220680
    },
    {
      "epoch": 0.005947192334460589,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001558622,
      "loss": 3.2027,
      "step": 220690
    },
    {
      "epoch": 0.005965835570618773,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001558602,
      "loss": 3.1758,
      "step": 220700
    },
    {
      "epoch": 0.0059844788067769565,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001558582,
      "loss": 3.1008,
      "step": 220710
    },
    {
      "epoch": 0.0060031220429351394,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015585620000000002,
      "loss": 3.2946,
      "step": 220720
    },
    {
      "epoch": 0.006021765279093323,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015585420000000002,
      "loss": 3.1668,
      "step": 220730
    },
    {
      "epoch": 0.006040408515251507,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001558522,
      "loss": 3.1736,
      "step": 220740
    },
    {
      "epoch": 0.006059051751409691,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001558502,
      "loss": 3.0955,
      "step": 220750
    },
    {
      "epoch": 0.0060776949875678745,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001558482,
      "loss": 3.0663,
      "step": 220760
    },
    {
      "epoch": 0.006096338223726058,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001558462,
      "loss": 3.2817,
      "step": 220770
    },
    {
      "epoch": 0.006114981459884242,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015584420000000002,
      "loss": 3.0997,
      "step": 220780
    },
    {
      "epoch": 0.006133624696042426,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015584220000000002,
      "loss": 2.99,
      "step": 220790
    },
    {
      "epoch": 0.0061522679322006096,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001558402,
      "loss": 2.9988,
      "step": 220800
    },
    {
      "epoch": 0.0061709111683587925,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001558382,
      "loss": 3.0819,
      "step": 220810
    },
    {
      "epoch": 0.006189554404516976,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001558362,
      "loss": 3.1018,
      "step": 220820
    },
    {
      "epoch": 0.00620819764067516,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001558342,
      "loss": 3.1227,
      "step": 220830
    },
    {
      "epoch": 0.006226840876833344,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001558322,
      "loss": 3.1776,
      "step": 220840
    },
    {
      "epoch": 0.0062454841129915275,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015583020000000002,
      "loss": 3.1832,
      "step": 220850
    },
    {
      "epoch": 0.006264127349149711,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001558282,
      "loss": 3.1003,
      "step": 220860
    },
    {
      "epoch": 0.006282770585307895,
      "grad_norm": 0.75,
      "learning_rate": 0.0001558262,
      "loss": 3.118,
      "step": 220870
    },
    {
      "epoch": 0.006301413821466079,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001558242,
      "loss": 3.1495,
      "step": 220880
    },
    {
      "epoch": 0.006320057057624262,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001558222,
      "loss": 3.0946,
      "step": 220890
    },
    {
      "epoch": 0.0063387002937824455,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015582020000000002,
      "loss": 3.1126,
      "step": 220900
    },
    {
      "epoch": 0.006357343529940629,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001558182,
      "loss": 3.1297,
      "step": 220910
    },
    {
      "epoch": 0.006375986766098813,
      "grad_norm": 0.75,
      "learning_rate": 0.00015581620000000001,
      "loss": 3.1641,
      "step": 220920
    },
    {
      "epoch": 0.006394630002256997,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001558142,
      "loss": 3.1761,
      "step": 220930
    },
    {
      "epoch": 0.0064132732384151805,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001558122,
      "loss": 3.1216,
      "step": 220940
    },
    {
      "epoch": 0.006431916474573364,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015581020000000003,
      "loss": 3.1273,
      "step": 220950
    },
    {
      "epoch": 0.006450559710731548,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001558082,
      "loss": 3.1458,
      "step": 220960
    },
    {
      "epoch": 0.006469202946889732,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015580620000000002,
      "loss": 3.1885,
      "step": 220970
    },
    {
      "epoch": 0.006487846183047915,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001558042,
      "loss": 3.1297,
      "step": 220980
    },
    {
      "epoch": 0.0065064894192060985,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001558022,
      "loss": 3.2001,
      "step": 220990
    },
    {
      "epoch": 0.006525132655364282,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001558002,
      "loss": 3.1525,
      "step": 221000
    },
    {
      "epoch": 0.006543775891522466,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001557982,
      "loss": 3.055,
      "step": 221010
    },
    {
      "epoch": 0.00656241912768065,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015579620000000002,
      "loss": 3.0683,
      "step": 221020
    },
    {
      "epoch": 0.0065810623638388336,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001557942,
      "loss": 3.1201,
      "step": 221030
    },
    {
      "epoch": 0.006599705599997017,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015579220000000002,
      "loss": 3.1657,
      "step": 221040
    },
    {
      "epoch": 0.006618348836155201,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015579019999999998,
      "loss": 3.1492,
      "step": 221050
    },
    {
      "epoch": 0.006636992072313384,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001557882,
      "loss": 3.2268,
      "step": 221060
    },
    {
      "epoch": 0.006655635308471568,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001557862,
      "loss": 3.1233,
      "step": 221070
    },
    {
      "epoch": 0.0066742785446297515,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001557842,
      "loss": 3.2083,
      "step": 221080
    },
    {
      "epoch": 0.006692921780787935,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015578220000000002,
      "loss": 3.1683,
      "step": 221090
    },
    {
      "epoch": 0.006711565016946119,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001557802,
      "loss": 3.0992,
      "step": 221100
    },
    {
      "epoch": 0.006730208253104303,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001557782,
      "loss": 3.1465,
      "step": 221110
    },
    {
      "epoch": 0.006748851489262487,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001557762,
      "loss": 3.1428,
      "step": 221120
    },
    {
      "epoch": 0.00676749472542067,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001557742,
      "loss": 3.1668,
      "step": 221130
    },
    {
      "epoch": 0.006786137961578854,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001557722,
      "loss": 3.1132,
      "step": 221140
    },
    {
      "epoch": 0.006804781197737037,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001557702,
      "loss": 3.2215,
      "step": 221150
    },
    {
      "epoch": 0.006823424433895221,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015576820000000002,
      "loss": 3.1205,
      "step": 221160
    },
    {
      "epoch": 0.0068420676700534045,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001557662,
      "loss": 3.104,
      "step": 221170
    },
    {
      "epoch": 0.006860710906211588,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001557642,
      "loss": 3.138,
      "step": 221180
    },
    {
      "epoch": 0.006879354142369772,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001557622,
      "loss": 3.1269,
      "step": 221190
    },
    {
      "epoch": 0.006897997378527956,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001557602,
      "loss": 3.1319,
      "step": 221200
    },
    {
      "epoch": 0.00691664061468614,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015575820000000002,
      "loss": 3.253,
      "step": 221210
    },
    {
      "epoch": 0.006935283850844323,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001557562,
      "loss": 3.2601,
      "step": 221220
    },
    {
      "epoch": 0.006953927087002506,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015575420000000001,
      "loss": 3.0504,
      "step": 221230
    },
    {
      "epoch": 0.00697257032316069,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001557522,
      "loss": 3.1925,
      "step": 221240
    },
    {
      "epoch": 0.006991213559318874,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001557502,
      "loss": 3.096,
      "step": 221250
    },
    {
      "epoch": 0.0070098567954770576,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001557482,
      "loss": 3.1191,
      "step": 221260
    },
    {
      "epoch": 0.007028500031635241,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001557462,
      "loss": 3.1037,
      "step": 221270
    },
    {
      "epoch": 0.007047143267793425,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015574420000000002,
      "loss": 3.1438,
      "step": 221280
    },
    {
      "epoch": 0.007065786503951609,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015574220000000001,
      "loss": 3.2143,
      "step": 221290
    },
    {
      "epoch": 0.007084429740109793,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001557402,
      "loss": 3.1885,
      "step": 221300
    },
    {
      "epoch": 0.007103072976267976,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001557382,
      "loss": 3.1972,
      "step": 221310
    },
    {
      "epoch": 0.007121716212426159,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001557362,
      "loss": 3.1851,
      "step": 221320
    },
    {
      "epoch": 0.007140359448584343,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001557342,
      "loss": 3.0752,
      "step": 221330
    },
    {
      "epoch": 0.007159002684742527,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015573220000000002,
      "loss": 3.2054,
      "step": 221340
    },
    {
      "epoch": 0.007177645920900711,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015573020000000002,
      "loss": 3.1583,
      "step": 221350
    },
    {
      "epoch": 0.007196289157058894,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001557282,
      "loss": 3.2334,
      "step": 221360
    },
    {
      "epoch": 0.007214932393217078,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001557262,
      "loss": 3.0922,
      "step": 221370
    },
    {
      "epoch": 0.007233575629375262,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001557242,
      "loss": 3.1256,
      "step": 221380
    },
    {
      "epoch": 0.007252218865533446,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015572220000000003,
      "loss": 3.2324,
      "step": 221390
    },
    {
      "epoch": 0.007270862101691629,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001557202,
      "loss": 3.1999,
      "step": 221400
    },
    {
      "epoch": 0.007289505337849812,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015571820000000002,
      "loss": 3.2516,
      "step": 221410
    },
    {
      "epoch": 0.007308148574007996,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001557162,
      "loss": 3.2338,
      "step": 221420
    },
    {
      "epoch": 0.00732679181016618,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001557142,
      "loss": 3.2136,
      "step": 221430
    },
    {
      "epoch": 0.007345435046324364,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001557122,
      "loss": 3.1024,
      "step": 221440
    },
    {
      "epoch": 0.007364078282482547,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001557102,
      "loss": 3.1348,
      "step": 221450
    },
    {
      "epoch": 0.007382721518640731,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015570820000000002,
      "loss": 3.1818,
      "step": 221460
    },
    {
      "epoch": 0.007401364754798915,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001557062,
      "loss": 3.0537,
      "step": 221470
    },
    {
      "epoch": 0.007420007990957099,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001557042,
      "loss": 3.1689,
      "step": 221480
    },
    {
      "epoch": 0.0074386512271152816,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001557022,
      "loss": 3.2524,
      "step": 221490
    },
    {
      "epoch": 0.007457294463273465,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001557002,
      "loss": 3.2823,
      "step": 221500
    },
    {
      "epoch": 0.007475937699431649,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015569820000000003,
      "loss": 3.1927,
      "step": 221510
    },
    {
      "epoch": 0.007494580935589833,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001556962,
      "loss": 3.0717,
      "step": 221520
    },
    {
      "epoch": 0.007513224171748017,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015569420000000002,
      "loss": 3.1129,
      "step": 221530
    },
    {
      "epoch": 0.0075318674079062,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015569219999999999,
      "loss": 3.0717,
      "step": 221540
    },
    {
      "epoch": 0.007550510644064384,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001556902,
      "loss": 3.1845,
      "step": 221550
    },
    {
      "epoch": 0.007569153880222568,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001556882,
      "loss": 3.0657,
      "step": 221560
    },
    {
      "epoch": 0.007587797116380752,
      "grad_norm": 0.75,
      "learning_rate": 0.0001556862,
      "loss": 3.0929,
      "step": 221570
    },
    {
      "epoch": 0.007606440352538935,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015568420000000002,
      "loss": 3.2565,
      "step": 221580
    },
    {
      "epoch": 0.007625083588697118,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001556822,
      "loss": 3.1745,
      "step": 221590
    },
    {
      "epoch": 0.007643726824855302,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015568020000000001,
      "loss": 3.3294,
      "step": 221600
    },
    {
      "epoch": 0.007662370061013486,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001556782,
      "loss": 3.1242,
      "step": 221610
    },
    {
      "epoch": 0.00768101329717167,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001556762,
      "loss": 2.9971,
      "step": 221620
    },
    {
      "epoch": 0.007699656533329853,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001556742,
      "loss": 3.1214,
      "step": 221630
    },
    {
      "epoch": 0.007718299769488037,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001556722,
      "loss": 3.2291,
      "step": 221640
    },
    {
      "epoch": 0.007736943005646221,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015567020000000002,
      "loss": 3.1387,
      "step": 221650
    },
    {
      "epoch": 0.007755586241804404,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001556682,
      "loss": 3.1572,
      "step": 221660
    },
    {
      "epoch": 0.007774229477962588,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001556662,
      "loss": 3.0162,
      "step": 221670
    },
    {
      "epoch": 0.007792872714120771,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001556642,
      "loss": 3.2005,
      "step": 221680
    },
    {
      "epoch": 0.007811515950278955,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001556622,
      "loss": 3.0812,
      "step": 221690
    },
    {
      "epoch": 0.00783015918643714,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015566020000000002,
      "loss": 3.1482,
      "step": 221700
    },
    {
      "epoch": 0.007848802422595322,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001556582,
      "loss": 3.0664,
      "step": 221710
    },
    {
      "epoch": 0.007867445658753506,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015565620000000002,
      "loss": 3.0958,
      "step": 221720
    },
    {
      "epoch": 0.00788608889491169,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001556542,
      "loss": 3.2175,
      "step": 221730
    },
    {
      "epoch": 0.007904732131069873,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001556522,
      "loss": 3.0996,
      "step": 221740
    },
    {
      "epoch": 0.007923375367228057,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001556502,
      "loss": 3.2204,
      "step": 221750
    },
    {
      "epoch": 0.00794201860338624,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001556482,
      "loss": 3.2111,
      "step": 221760
    },
    {
      "epoch": 0.007960661839544424,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015564620000000002,
      "loss": 3.2156,
      "step": 221770
    },
    {
      "epoch": 0.007979305075702608,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015564420000000002,
      "loss": 3.0981,
      "step": 221780
    },
    {
      "epoch": 0.007997948311860792,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001556422,
      "loss": 3.1077,
      "step": 221790
    },
    {
      "epoch": 0.008016591548018976,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001556402,
      "loss": 3.1657,
      "step": 221800
    },
    {
      "epoch": 0.00803523478417716,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001556382,
      "loss": 3.1316,
      "step": 221810
    },
    {
      "epoch": 0.008053878020335343,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001556362,
      "loss": 3.1586,
      "step": 221820
    },
    {
      "epoch": 0.008072521256493527,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001556342,
      "loss": 3.1932,
      "step": 221830
    },
    {
      "epoch": 0.00809116449265171,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015563220000000002,
      "loss": 3.1319,
      "step": 221840
    },
    {
      "epoch": 0.008109807728809895,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001556302,
      "loss": 3.181,
      "step": 221850
    },
    {
      "epoch": 0.008128450964968078,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001556282,
      "loss": 3.0894,
      "step": 221860
    },
    {
      "epoch": 0.008147094201126262,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001556262,
      "loss": 3.153,
      "step": 221870
    },
    {
      "epoch": 0.008165737437284444,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001556242,
      "loss": 3.263,
      "step": 221880
    },
    {
      "epoch": 0.008184380673442628,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001556222,
      "loss": 3.0443,
      "step": 221890
    },
    {
      "epoch": 0.008203023909600812,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015562020000000002,
      "loss": 3.0851,
      "step": 221900
    },
    {
      "epoch": 0.008221667145758995,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015561820000000001,
      "loss": 3.0582,
      "step": 221910
    },
    {
      "epoch": 0.008240310381917179,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001556162,
      "loss": 2.9943,
      "step": 221920
    },
    {
      "epoch": 0.008258953618075363,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001556142,
      "loss": 3.1268,
      "step": 221930
    },
    {
      "epoch": 0.008277596854233547,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001556122,
      "loss": 3.0967,
      "step": 221940
    },
    {
      "epoch": 0.00829624009039173,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015561020000000002,
      "loss": 3.0817,
      "step": 221950
    },
    {
      "epoch": 0.008314883326549914,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001556082,
      "loss": 3.1243,
      "step": 221960
    },
    {
      "epoch": 0.008333526562708098,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015560620000000001,
      "loss": 3.0161,
      "step": 221970
    },
    {
      "epoch": 0.008352169798866282,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001556042,
      "loss": 3.1169,
      "step": 221980
    },
    {
      "epoch": 0.008370813035024465,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001556022,
      "loss": 3.2578,
      "step": 221990
    },
    {
      "epoch": 0.00838945627118265,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015560020000000003,
      "loss": 3.1282,
      "step": 222000
    },
    {
      "epoch": 0.008408099507340833,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001555982,
      "loss": 3.155,
      "step": 222010
    },
    {
      "epoch": 0.008426742743499017,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015559620000000002,
      "loss": 3.0704,
      "step": 222020
    },
    {
      "epoch": 0.0084453859796572,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001555942,
      "loss": 3.19,
      "step": 222030
    },
    {
      "epoch": 0.008464029215815384,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001555922,
      "loss": 3.1074,
      "step": 222040
    },
    {
      "epoch": 0.008482672451973566,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001555902,
      "loss": 3.0979,
      "step": 222050
    },
    {
      "epoch": 0.00850131568813175,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001555882,
      "loss": 3.1279,
      "step": 222060
    },
    {
      "epoch": 0.008519958924289934,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015558620000000002,
      "loss": 3.2482,
      "step": 222070
    },
    {
      "epoch": 0.008538602160448118,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001555842,
      "loss": 3.2131,
      "step": 222080
    },
    {
      "epoch": 0.008557245396606301,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015558220000000002,
      "loss": 3.096,
      "step": 222090
    },
    {
      "epoch": 0.008575888632764485,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015558019999999998,
      "loss": 3.1957,
      "step": 222100
    },
    {
      "epoch": 0.008594531868922669,
      "grad_norm": 0.875,
      "learning_rate": 0.0001555782,
      "loss": 3.1785,
      "step": 222110
    },
    {
      "epoch": 0.008613175105080853,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001555762,
      "loss": 3.0902,
      "step": 222120
    },
    {
      "epoch": 0.008631818341239036,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001555742,
      "loss": 3.1191,
      "step": 222130
    },
    {
      "epoch": 0.00865046157739722,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015557220000000002,
      "loss": 3.253,
      "step": 222140
    },
    {
      "epoch": 0.008669104813555404,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001555702,
      "loss": 3.1338,
      "step": 222150
    },
    {
      "epoch": 0.008687748049713588,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001555682,
      "loss": 3.1889,
      "step": 222160
    },
    {
      "epoch": 0.008706391285871772,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001555662,
      "loss": 3.1317,
      "step": 222170
    },
    {
      "epoch": 0.008725034522029955,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001555642,
      "loss": 3.2292,
      "step": 222180
    },
    {
      "epoch": 0.008743677758188139,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015556220000000003,
      "loss": 3.084,
      "step": 222190
    },
    {
      "epoch": 0.008762320994346323,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001555602,
      "loss": 3.1348,
      "step": 222200
    },
    {
      "epoch": 0.008780964230504507,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015555820000000002,
      "loss": 3.0338,
      "step": 222210
    },
    {
      "epoch": 0.008799607466662689,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015555619999999999,
      "loss": 3.0648,
      "step": 222220
    },
    {
      "epoch": 0.008818250702820872,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001555542,
      "loss": 3.1091,
      "step": 222230
    },
    {
      "epoch": 0.008836893938979056,
      "grad_norm": 0.75,
      "learning_rate": 0.0001555522,
      "loss": 3.0367,
      "step": 222240
    },
    {
      "epoch": 0.00885553717513724,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001555502,
      "loss": 3.129,
      "step": 222250
    },
    {
      "epoch": 0.008874180411295424,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015554820000000002,
      "loss": 3.2505,
      "step": 222260
    },
    {
      "epoch": 0.008892823647453607,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001555462,
      "loss": 3.1322,
      "step": 222270
    },
    {
      "epoch": 0.008911466883611791,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015554420000000001,
      "loss": 3.2123,
      "step": 222280
    },
    {
      "epoch": 0.008930110119769975,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001555422,
      "loss": 3.1407,
      "step": 222290
    },
    {
      "epoch": 0.008948753355928159,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001555402,
      "loss": 3.1126,
      "step": 222300
    },
    {
      "epoch": 0.008967396592086343,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001555382,
      "loss": 3.1438,
      "step": 222310
    },
    {
      "epoch": 0.008986039828244526,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001555362,
      "loss": 3.222,
      "step": 222320
    },
    {
      "epoch": 0.00900468306440271,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015553420000000002,
      "loss": 3.2253,
      "step": 222330
    },
    {
      "epoch": 0.009023326300560894,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015553220000000001,
      "loss": 3.1656,
      "step": 222340
    },
    {
      "epoch": 0.009041969536719078,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001555302,
      "loss": 3.1625,
      "step": 222350
    },
    {
      "epoch": 0.009060612772877261,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001555282,
      "loss": 3.1367,
      "step": 222360
    },
    {
      "epoch": 0.009079256009035445,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001555262,
      "loss": 3.224,
      "step": 222370
    },
    {
      "epoch": 0.009097899245193629,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001555242,
      "loss": 3.0592,
      "step": 222380
    },
    {
      "epoch": 0.009116542481351811,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015552220000000002,
      "loss": 3.1079,
      "step": 222390
    },
    {
      "epoch": 0.009135185717509995,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015552020000000002,
      "loss": 3.116,
      "step": 222400
    },
    {
      "epoch": 0.009153828953668178,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001555182,
      "loss": 3.1846,
      "step": 222410
    },
    {
      "epoch": 0.009172472189826362,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001555162,
      "loss": 3.0946,
      "step": 222420
    },
    {
      "epoch": 0.009191115425984546,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001555142,
      "loss": 3.0581,
      "step": 222430
    },
    {
      "epoch": 0.00920975866214273,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001555122,
      "loss": 2.9705,
      "step": 222440
    },
    {
      "epoch": 0.009228401898300913,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001555102,
      "loss": 3.1351,
      "step": 222450
    },
    {
      "epoch": 0.009247045134459097,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015550820000000002,
      "loss": 3.1663,
      "step": 222460
    },
    {
      "epoch": 0.009265688370617281,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001555062,
      "loss": 3.16,
      "step": 222470
    },
    {
      "epoch": 0.009284331606775465,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001555042,
      "loss": 3.2145,
      "step": 222480
    },
    {
      "epoch": 0.009302974842933649,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001555022,
      "loss": 3.2129,
      "step": 222490
    },
    {
      "epoch": 0.009321618079091832,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001555002,
      "loss": 3.1415,
      "step": 222500
    },
    {
      "epoch": 0.009340261315250016,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015549820000000002,
      "loss": 3.0951,
      "step": 222510
    },
    {
      "epoch": 0.0093589045514082,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001554962,
      "loss": 3.0822,
      "step": 222520
    },
    {
      "epoch": 0.009377547787566384,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001554942,
      "loss": 3.274,
      "step": 222530
    },
    {
      "epoch": 0.009396191023724567,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001554922,
      "loss": 3.1293,
      "step": 222540
    },
    {
      "epoch": 0.009414834259882751,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001554902,
      "loss": 3.2863,
      "step": 222550
    },
    {
      "epoch": 0.009433477496040933,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015548820000000003,
      "loss": 2.9907,
      "step": 222560
    },
    {
      "epoch": 0.009452120732199117,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001554862,
      "loss": 3.2563,
      "step": 222570
    },
    {
      "epoch": 0.0094707639683573,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015548420000000002,
      "loss": 3.2543,
      "step": 222580
    },
    {
      "epoch": 0.009489407204515484,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001554822,
      "loss": 3.1358,
      "step": 222590
    },
    {
      "epoch": 0.009508050440673668,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001554802,
      "loss": 3.1499,
      "step": 222600
    },
    {
      "epoch": 0.009526693676831852,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001554782,
      "loss": 3.1953,
      "step": 222610
    },
    {
      "epoch": 0.009545336912990036,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001554762,
      "loss": 3.0453,
      "step": 222620
    },
    {
      "epoch": 0.00956398014914822,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015547420000000002,
      "loss": 2.9821,
      "step": 222630
    },
    {
      "epoch": 0.009582623385306403,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001554722,
      "loss": 3.2036,
      "step": 222640
    },
    {
      "epoch": 0.009601266621464587,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015547020000000001,
      "loss": 3.117,
      "step": 222650
    },
    {
      "epoch": 0.00961990985762277,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001554682,
      "loss": 3.09,
      "step": 222660
    },
    {
      "epoch": 0.009638553093780955,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001554662,
      "loss": 3.0989,
      "step": 222670
    },
    {
      "epoch": 0.009657196329939138,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015546420000000003,
      "loss": 3.0865,
      "step": 222680
    },
    {
      "epoch": 0.009675839566097322,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001554622,
      "loss": 3.1053,
      "step": 222690
    },
    {
      "epoch": 0.009694482802255506,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015546020000000002,
      "loss": 3.0862,
      "step": 222700
    },
    {
      "epoch": 0.00971312603841369,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001554582,
      "loss": 3.1883,
      "step": 222710
    },
    {
      "epoch": 0.009731769274571873,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001554562,
      "loss": 3.1156,
      "step": 222720
    },
    {
      "epoch": 0.009750412510730057,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001554542,
      "loss": 3.079,
      "step": 222730
    },
    {
      "epoch": 0.00976905574688824,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001554522,
      "loss": 3.191,
      "step": 222740
    },
    {
      "epoch": 0.009787698983046423,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015545020000000002,
      "loss": 3.1683,
      "step": 222750
    },
    {
      "epoch": 0.009806342219204607,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001554482,
      "loss": 3.1985,
      "step": 222760
    },
    {
      "epoch": 0.00982498545536279,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015544620000000002,
      "loss": 3.0635,
      "step": 222770
    },
    {
      "epoch": 0.009843628691520974,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001554442,
      "loss": 3.1327,
      "step": 222780
    },
    {
      "epoch": 0.009862271927679158,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001554422,
      "loss": 3.0819,
      "step": 222790
    },
    {
      "epoch": 0.009880915163837342,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001554402,
      "loss": 3.2369,
      "step": 222800
    },
    {
      "epoch": 0.009899558399995526,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001554382,
      "loss": 3.0902,
      "step": 222810
    },
    {
      "epoch": 0.00991820163615371,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015543620000000002,
      "loss": 3.1748,
      "step": 222820
    },
    {
      "epoch": 0.009936844872311893,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001554342,
      "loss": 3.2941,
      "step": 222830
    },
    {
      "epoch": 0.009955488108470077,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001554322,
      "loss": 3.2155,
      "step": 222840
    },
    {
      "epoch": 0.00997413134462826,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001554302,
      "loss": 3.0178,
      "step": 222850
    },
    {
      "epoch": 0.009992774580786444,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001554282,
      "loss": 3.03,
      "step": 222860
    },
    {
      "epoch": 0.010011417816944628,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001554262,
      "loss": 3.1208,
      "step": 222870
    },
    {
      "epoch": 0.010030061053102812,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001554242,
      "loss": 3.1084,
      "step": 222880
    },
    {
      "epoch": 0.010048704289260996,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015542220000000002,
      "loss": 3.2124,
      "step": 222890
    },
    {
      "epoch": 0.01006734752541918,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001554202,
      "loss": 3.1017,
      "step": 222900
    },
    {
      "epoch": 0.010085990761577361,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001554182,
      "loss": 3.1478,
      "step": 222910
    },
    {
      "epoch": 0.010104633997735545,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001554162,
      "loss": 3.2454,
      "step": 222920
    },
    {
      "epoch": 0.010123277233893729,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001554142,
      "loss": 3.2596,
      "step": 222930
    },
    {
      "epoch": 0.010141920470051913,
      "grad_norm": 0.75,
      "learning_rate": 0.0001554122,
      "loss": 3.2368,
      "step": 222940
    },
    {
      "epoch": 0.010160563706210097,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015541020000000002,
      "loss": 3.2028,
      "step": 222950
    },
    {
      "epoch": 0.01017920694236828,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015540820000000001,
      "loss": 3.095,
      "step": 222960
    },
    {
      "epoch": 0.010197850178526464,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001554062,
      "loss": 3.0841,
      "step": 222970
    },
    {
      "epoch": 0.010216493414684648,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001554042,
      "loss": 3.1156,
      "step": 222980
    },
    {
      "epoch": 0.010235136650842832,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001554022,
      "loss": 3.3035,
      "step": 222990
    },
    {
      "epoch": 0.010253779887001015,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015540020000000002,
      "loss": 3.1037,
      "step": 223000
    },
    {
      "epoch": 0.010272423123159199,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001553982,
      "loss": 3.1218,
      "step": 223010
    },
    {
      "epoch": 0.010291066359317383,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015539620000000001,
      "loss": 3.1403,
      "step": 223020
    },
    {
      "epoch": 0.010309709595475567,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001553942,
      "loss": 3.2474,
      "step": 223030
    },
    {
      "epoch": 0.01032835283163375,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001553922,
      "loss": 3.2273,
      "step": 223040
    },
    {
      "epoch": 0.010346996067791934,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001553902,
      "loss": 3.1212,
      "step": 223050
    },
    {
      "epoch": 0.010365639303950118,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001553882,
      "loss": 3.0628,
      "step": 223060
    },
    {
      "epoch": 0.010384282540108302,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015538620000000002,
      "loss": 3.1541,
      "step": 223070
    },
    {
      "epoch": 0.010402925776266484,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001553842,
      "loss": 3.0544,
      "step": 223080
    },
    {
      "epoch": 0.010421569012424668,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001553822,
      "loss": 3.1902,
      "step": 223090
    },
    {
      "epoch": 0.010440212248582851,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001553802,
      "loss": 3.084,
      "step": 223100
    },
    {
      "epoch": 0.010458855484741035,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001553782,
      "loss": 3.1326,
      "step": 223110
    },
    {
      "epoch": 0.010477498720899219,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015537620000000003,
      "loss": 3.1266,
      "step": 223120
    },
    {
      "epoch": 0.010496141957057403,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001553742,
      "loss": 3.2059,
      "step": 223130
    },
    {
      "epoch": 0.010514785193215586,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015537220000000002,
      "loss": 3.1573,
      "step": 223140
    },
    {
      "epoch": 0.01053342842937377,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001553702,
      "loss": 3.249,
      "step": 223150
    },
    {
      "epoch": 0.010552071665531954,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001553682,
      "loss": 3.1784,
      "step": 223160
    },
    {
      "epoch": 0.010570714901690138,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015536620000000003,
      "loss": 3.2965,
      "step": 223170
    },
    {
      "epoch": 0.010589358137848321,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001553642,
      "loss": 3.1019,
      "step": 223180
    },
    {
      "epoch": 0.010608001374006505,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015536220000000002,
      "loss": 3.2695,
      "step": 223190
    },
    {
      "epoch": 0.010626644610164689,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001553602,
      "loss": 3.1547,
      "step": 223200
    },
    {
      "epoch": 0.010645287846322873,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001553582,
      "loss": 3.2497,
      "step": 223210
    },
    {
      "epoch": 0.010663931082481056,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001553562,
      "loss": 3.1414,
      "step": 223220
    },
    {
      "epoch": 0.01068257431863924,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001553542,
      "loss": 3.2742,
      "step": 223230
    },
    {
      "epoch": 0.010701217554797424,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015535220000000003,
      "loss": 2.9949,
      "step": 223240
    },
    {
      "epoch": 0.010719860790955606,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001553502,
      "loss": 3.1741,
      "step": 223250
    },
    {
      "epoch": 0.01073850402711379,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015534820000000002,
      "loss": 3.1543,
      "step": 223260
    },
    {
      "epoch": 0.010757147263271974,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001553462,
      "loss": 3.1299,
      "step": 223270
    },
    {
      "epoch": 0.010775790499430157,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001553442,
      "loss": 3.0993,
      "step": 223280
    },
    {
      "epoch": 0.010794433735588341,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001553422,
      "loss": 3.1383,
      "step": 223290
    },
    {
      "epoch": 0.010813076971746525,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001553402,
      "loss": 3.3105,
      "step": 223300
    },
    {
      "epoch": 0.010831720207904709,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015533820000000002,
      "loss": 3.2573,
      "step": 223310
    },
    {
      "epoch": 0.010850363444062892,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001553362,
      "loss": 3.0444,
      "step": 223320
    },
    {
      "epoch": 0.010869006680221076,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015533420000000001,
      "loss": 3.2036,
      "step": 223330
    },
    {
      "epoch": 0.01088764991637926,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001553322,
      "loss": 3.1133,
      "step": 223340
    },
    {
      "epoch": 0.010906293152537444,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001553302,
      "loss": 3.138,
      "step": 223350
    },
    {
      "epoch": 0.010924936388695627,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001553282,
      "loss": 3.0245,
      "step": 223360
    },
    {
      "epoch": 0.010943579624853811,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001553262,
      "loss": 3.278,
      "step": 223370
    },
    {
      "epoch": 0.010962222861011995,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015532420000000002,
      "loss": 3.0342,
      "step": 223380
    },
    {
      "epoch": 0.010980866097170179,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015532220000000002,
      "loss": 3.077,
      "step": 223390
    },
    {
      "epoch": 0.010999509333328363,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001553202,
      "loss": 3.0524,
      "step": 223400
    },
    {
      "epoch": 0.011018152569486546,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001553182,
      "loss": 3.0809,
      "step": 223410
    },
    {
      "epoch": 0.011036795805644728,
      "grad_norm": 0.75,
      "learning_rate": 0.0001553162,
      "loss": 3.1328,
      "step": 223420
    },
    {
      "epoch": 0.011055439041802912,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001553142,
      "loss": 3.131,
      "step": 223430
    },
    {
      "epoch": 0.011074082277961096,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001553122,
      "loss": 3.0814,
      "step": 223440
    },
    {
      "epoch": 0.01109272551411928,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015531020000000002,
      "loss": 3.0607,
      "step": 223450
    },
    {
      "epoch": 0.011111368750277463,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001553082,
      "loss": 3.1137,
      "step": 223460
    },
    {
      "epoch": 0.011130011986435647,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001553062,
      "loss": 3.1003,
      "step": 223470
    },
    {
      "epoch": 0.011148655222593831,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001553042,
      "loss": 2.9926,
      "step": 223480
    },
    {
      "epoch": 0.011167298458752015,
      "grad_norm": 0.75,
      "learning_rate": 0.0001553022,
      "loss": 3.0709,
      "step": 223490
    },
    {
      "epoch": 0.011185941694910198,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001553002,
      "loss": 2.9937,
      "step": 223500
    },
    {
      "epoch": 0.011204584931068382,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015529820000000002,
      "loss": 3.1733,
      "step": 223510
    },
    {
      "epoch": 0.011223228167226566,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001552962,
      "loss": 3.1282,
      "step": 223520
    },
    {
      "epoch": 0.01124187140338475,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001552942,
      "loss": 3.2019,
      "step": 223530
    },
    {
      "epoch": 0.011260514639542933,
      "grad_norm": 0.75,
      "learning_rate": 0.0001552922,
      "loss": 3.2099,
      "step": 223540
    },
    {
      "epoch": 0.011279157875701117,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001552902,
      "loss": 3.1615,
      "step": 223550
    },
    {
      "epoch": 0.011297801111859301,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015528820000000002,
      "loss": 3.1172,
      "step": 223560
    },
    {
      "epoch": 0.011316444348017485,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001552862,
      "loss": 3.1711,
      "step": 223570
    },
    {
      "epoch": 0.011335087584175669,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001552842,
      "loss": 3.0827,
      "step": 223580
    },
    {
      "epoch": 0.01135373082033385,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001552822,
      "loss": 3.1195,
      "step": 223590
    },
    {
      "epoch": 0.011372374056492034,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001552802,
      "loss": 3.1358,
      "step": 223600
    },
    {
      "epoch": 0.011391017292650218,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015527820000000003,
      "loss": 3.1444,
      "step": 223610
    },
    {
      "epoch": 0.011409660528808402,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001552762,
      "loss": 3.1481,
      "step": 223620
    },
    {
      "epoch": 0.011428303764966586,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015527420000000002,
      "loss": 3.0994,
      "step": 223630
    },
    {
      "epoch": 0.01144694700112477,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015527220000000001,
      "loss": 3.0596,
      "step": 223640
    },
    {
      "epoch": 0.011465590237282953,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001552702,
      "loss": 3.3068,
      "step": 223650
    },
    {
      "epoch": 0.011484233473441137,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001552682,
      "loss": 3.1816,
      "step": 223660
    },
    {
      "epoch": 0.01150287670959932,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001552662,
      "loss": 3.1396,
      "step": 223670
    },
    {
      "epoch": 0.011521519945757504,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015526420000000002,
      "loss": 3.2191,
      "step": 223680
    },
    {
      "epoch": 0.011540163181915688,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001552622,
      "loss": 3.0783,
      "step": 223690
    },
    {
      "epoch": 0.011558806418073872,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015526020000000001,
      "loss": 3.152,
      "step": 223700
    },
    {
      "epoch": 0.011577449654232056,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001552582,
      "loss": 2.8889,
      "step": 223710
    },
    {
      "epoch": 0.01159609289039024,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001552562,
      "loss": 3.039,
      "step": 223720
    },
    {
      "epoch": 0.011614736126548423,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015525420000000003,
      "loss": 3.2063,
      "step": 223730
    },
    {
      "epoch": 0.011633379362706607,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001552522,
      "loss": 3.2391,
      "step": 223740
    },
    {
      "epoch": 0.01165202259886479,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015525020000000002,
      "loss": 3.1995,
      "step": 223750
    },
    {
      "epoch": 0.011670665835022973,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001552482,
      "loss": 3.1183,
      "step": 223760
    },
    {
      "epoch": 0.011689309071181157,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001552462,
      "loss": 3.125,
      "step": 223770
    },
    {
      "epoch": 0.01170795230733934,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001552442,
      "loss": 3.148,
      "step": 223780
    },
    {
      "epoch": 0.011726595543497524,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001552422,
      "loss": 3.2412,
      "step": 223790
    },
    {
      "epoch": 0.011745238779655708,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015524020000000003,
      "loss": 3.2227,
      "step": 223800
    },
    {
      "epoch": 0.011763882015813892,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001552382,
      "loss": 3.1255,
      "step": 223810
    },
    {
      "epoch": 0.011782525251972075,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015523620000000002,
      "loss": 3.1022,
      "step": 223820
    },
    {
      "epoch": 0.01180116848813026,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001552342,
      "loss": 3.1686,
      "step": 223830
    },
    {
      "epoch": 0.011819811724288443,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001552322,
      "loss": 3.1885,
      "step": 223840
    },
    {
      "epoch": 0.011838454960446627,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001552302,
      "loss": 3.0964,
      "step": 223850
    },
    {
      "epoch": 0.01185709819660481,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001552282,
      "loss": 3.1637,
      "step": 223860
    },
    {
      "epoch": 0.011875741432762994,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015522620000000002,
      "loss": 3.0563,
      "step": 223870
    },
    {
      "epoch": 0.011894384668921178,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001552242,
      "loss": 3.1385,
      "step": 223880
    },
    {
      "epoch": 0.011913027905079362,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001552222,
      "loss": 3.1962,
      "step": 223890
    },
    {
      "epoch": 0.011931671141237546,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001552202,
      "loss": 3.1429,
      "step": 223900
    },
    {
      "epoch": 0.01195031437739573,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001552182,
      "loss": 3.0825,
      "step": 223910
    },
    {
      "epoch": 0.011968957613553913,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001552162,
      "loss": 3.2386,
      "step": 223920
    },
    {
      "epoch": 0.011987600849712097,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001552142,
      "loss": 3.1687,
      "step": 223930
    },
    {
      "epoch": 0.012006244085870279,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015521220000000002,
      "loss": 3.0748,
      "step": 223940
    },
    {
      "epoch": 0.012024887322028463,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015521020000000001,
      "loss": 3.1126,
      "step": 223950
    },
    {
      "epoch": 0.012043530558186646,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001552082,
      "loss": 3.1569,
      "step": 223960
    },
    {
      "epoch": 0.01206217379434483,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001552062,
      "loss": 3.1921,
      "step": 223970
    },
    {
      "epoch": 0.012080817030503014,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001552042,
      "loss": 3.1296,
      "step": 223980
    },
    {
      "epoch": 0.012099460266661198,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001552022,
      "loss": 2.9738,
      "step": 223990
    },
    {
      "epoch": 0.012118103502819381,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015520020000000002,
      "loss": 3.0526,
      "step": 224000
    },
    {
      "epoch": 0.012136746738977565,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015519820000000001,
      "loss": 3.2319,
      "step": 224010
    },
    {
      "epoch": 0.012155389975135749,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001551962,
      "loss": 3.1211,
      "step": 224020
    },
    {
      "epoch": 0.012174033211293933,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001551942,
      "loss": 3.0978,
      "step": 224030
    },
    {
      "epoch": 0.012192676447452117,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001551922,
      "loss": 3.1917,
      "step": 224040
    },
    {
      "epoch": 0.0122113196836103,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001551902,
      "loss": 3.2755,
      "step": 224050
    },
    {
      "epoch": 0.012229962919768484,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001551882,
      "loss": 3.1256,
      "step": 224060
    },
    {
      "epoch": 0.012248606155926668,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015518620000000002,
      "loss": 3.0486,
      "step": 224070
    },
    {
      "epoch": 0.012267249392084852,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001551842,
      "loss": 3.168,
      "step": 224080
    },
    {
      "epoch": 0.012285892628243035,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001551822,
      "loss": 3.1908,
      "step": 224090
    },
    {
      "epoch": 0.012304535864401219,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001551802,
      "loss": 3.2036,
      "step": 224100
    },
    {
      "epoch": 0.012323179100559401,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001551782,
      "loss": 3.2326,
      "step": 224110
    },
    {
      "epoch": 0.012341822336717585,
      "grad_norm": 0.59375,
      "learning_rate": 0.00015517620000000002,
      "loss": 3.0472,
      "step": 224120
    },
    {
      "epoch": 0.012360465572875769,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015517420000000002,
      "loss": 3.1549,
      "step": 224130
    },
    {
      "epoch": 0.012379108809033952,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001551722,
      "loss": 3.1351,
      "step": 224140
    },
    {
      "epoch": 0.012397752045192136,
      "grad_norm": 0.75,
      "learning_rate": 0.0001551702,
      "loss": 3.1254,
      "step": 224150
    },
    {
      "epoch": 0.01241639528135032,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001551682,
      "loss": 3.1849,
      "step": 224160
    },
    {
      "epoch": 0.012435038517508504,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015516620000000003,
      "loss": 3.2397,
      "step": 224170
    },
    {
      "epoch": 0.012453681753666688,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001551642,
      "loss": 3.0774,
      "step": 224180
    },
    {
      "epoch": 0.012472324989824871,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015516220000000002,
      "loss": 3.2419,
      "step": 224190
    },
    {
      "epoch": 0.012490968225983055,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001551602,
      "loss": 3.1888,
      "step": 224200
    },
    {
      "epoch": 0.012509611462141239,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001551582,
      "loss": 2.995,
      "step": 224210
    },
    {
      "epoch": 0.012528254698299423,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015515620000000003,
      "loss": 3.0243,
      "step": 224220
    },
    {
      "epoch": 0.012546897934457606,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001551542,
      "loss": 3.1971,
      "step": 224230
    },
    {
      "epoch": 0.01256554117061579,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015515220000000002,
      "loss": 3.1405,
      "step": 224240
    },
    {
      "epoch": 0.012584184406773974,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001551502,
      "loss": 3.2816,
      "step": 224250
    },
    {
      "epoch": 0.012602827642932158,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001551482,
      "loss": 3.1173,
      "step": 224260
    },
    {
      "epoch": 0.012621470879090341,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001551462,
      "loss": 3.152,
      "step": 224270
    },
    {
      "epoch": 0.012640114115248523,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001551442,
      "loss": 3.2507,
      "step": 224280
    },
    {
      "epoch": 0.012658757351406707,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015514220000000003,
      "loss": 3.1526,
      "step": 224290
    },
    {
      "epoch": 0.012677400587564891,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001551402,
      "loss": 3.1908,
      "step": 224300
    },
    {
      "epoch": 0.012696043823723075,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015513820000000002,
      "loss": 3.2143,
      "step": 224310
    },
    {
      "epoch": 0.012714687059881258,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001551362,
      "loss": 3.132,
      "step": 224320
    },
    {
      "epoch": 0.012733330296039442,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001551342,
      "loss": 3.1037,
      "step": 224330
    },
    {
      "epoch": 0.012751973532197626,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001551322,
      "loss": 3.1842,
      "step": 224340
    },
    {
      "epoch": 0.01277061676835581,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001551302,
      "loss": 3.1762,
      "step": 224350
    },
    {
      "epoch": 0.012789260004513994,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015512820000000002,
      "loss": 3.1388,
      "step": 224360
    },
    {
      "epoch": 0.012807903240672177,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001551262,
      "loss": 3.2115,
      "step": 224370
    },
    {
      "epoch": 0.012826546476830361,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015512420000000001,
      "loss": 3.214,
      "step": 224380
    },
    {
      "epoch": 0.012845189712988545,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001551222,
      "loss": 3.206,
      "step": 224390
    },
    {
      "epoch": 0.012863832949146729,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001551202,
      "loss": 3.11,
      "step": 224400
    },
    {
      "epoch": 0.012882476185304912,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001551182,
      "loss": 3.1103,
      "step": 224410
    },
    {
      "epoch": 0.012901119421463096,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001551162,
      "loss": 3.1292,
      "step": 224420
    },
    {
      "epoch": 0.01291976265762128,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015511420000000002,
      "loss": 3.1947,
      "step": 224430
    },
    {
      "epoch": 0.012938405893779464,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001551122,
      "loss": 3.1805,
      "step": 224440
    },
    {
      "epoch": 0.012957049129937646,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001551102,
      "loss": 3.1637,
      "step": 224450
    },
    {
      "epoch": 0.01297569236609583,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001551082,
      "loss": 3.1108,
      "step": 224460
    },
    {
      "epoch": 0.012994335602254013,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001551062,
      "loss": 3.2119,
      "step": 224470
    },
    {
      "epoch": 0.013012978838412197,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001551042,
      "loss": 3.0838,
      "step": 224480
    },
    {
      "epoch": 0.01303162207457038,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001551022,
      "loss": 3.109,
      "step": 224490
    },
    {
      "epoch": 0.013050265310728565,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015510020000000002,
      "loss": 3.0367,
      "step": 224500
    },
    {
      "epoch": 0.013068908546886748,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001550982,
      "loss": 3.1714,
      "step": 224510
    },
    {
      "epoch": 0.013087551783044932,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001550962,
      "loss": 3.1491,
      "step": 224520
    },
    {
      "epoch": 0.013106195019203116,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001550942,
      "loss": 3.1658,
      "step": 224530
    },
    {
      "epoch": 0.0131248382553613,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001550922,
      "loss": 3.2489,
      "step": 224540
    },
    {
      "epoch": 0.013143481491519483,
      "grad_norm": 0.75,
      "learning_rate": 0.0001550902,
      "loss": 3.1405,
      "step": 224550
    },
    {
      "epoch": 0.013162124727677667,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015508820000000002,
      "loss": 3.1682,
      "step": 224560
    },
    {
      "epoch": 0.013180767963835851,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001550862,
      "loss": 3.1634,
      "step": 224570
    },
    {
      "epoch": 0.013199411199994035,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001550842,
      "loss": 3.1802,
      "step": 224580
    },
    {
      "epoch": 0.013218054436152218,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001550822,
      "loss": 3.2336,
      "step": 224590
    },
    {
      "epoch": 0.013236697672310402,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001550802,
      "loss": 3.0922,
      "step": 224600
    },
    {
      "epoch": 0.013255340908468586,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015507820000000002,
      "loss": 3.2509,
      "step": 224610
    },
    {
      "epoch": 0.013273984144626768,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015507620000000002,
      "loss": 3.0917,
      "step": 224620
    },
    {
      "epoch": 0.013292627380784952,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015507420000000001,
      "loss": 3.1951,
      "step": 224630
    },
    {
      "epoch": 0.013311270616943136,
      "grad_norm": 0.75,
      "learning_rate": 0.0001550722,
      "loss": 3.1587,
      "step": 224640
    },
    {
      "epoch": 0.01332991385310132,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001550702,
      "loss": 3.026,
      "step": 224650
    },
    {
      "epoch": 0.013348557089259503,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001550682,
      "loss": 3.1712,
      "step": 224660
    },
    {
      "epoch": 0.013367200325417687,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001550662,
      "loss": 3.2761,
      "step": 224670
    },
    {
      "epoch": 0.01338584356157587,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015506420000000002,
      "loss": 3.0982,
      "step": 224680
    },
    {
      "epoch": 0.013404486797734054,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015506220000000001,
      "loss": 3.2258,
      "step": 224690
    },
    {
      "epoch": 0.013423130033892238,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001550602,
      "loss": 3.1272,
      "step": 224700
    },
    {
      "epoch": 0.013441773270050422,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001550582,
      "loss": 3.1577,
      "step": 224710
    },
    {
      "epoch": 0.013460416506208606,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001550562,
      "loss": 3.2198,
      "step": 224720
    },
    {
      "epoch": 0.01347905974236679,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015505420000000002,
      "loss": 3.0931,
      "step": 224730
    },
    {
      "epoch": 0.013497702978524973,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001550522,
      "loss": 3.2129,
      "step": 224740
    },
    {
      "epoch": 0.013516346214683157,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015505020000000002,
      "loss": 3.0081,
      "step": 224750
    },
    {
      "epoch": 0.01353498945084134,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001550482,
      "loss": 3.0958,
      "step": 224760
    },
    {
      "epoch": 0.013553632686999524,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001550462,
      "loss": 3.2294,
      "step": 224770
    },
    {
      "epoch": 0.013572275923157708,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015504420000000003,
      "loss": 3.1471,
      "step": 224780
    },
    {
      "epoch": 0.01359091915931589,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001550422,
      "loss": 3.1456,
      "step": 224790
    },
    {
      "epoch": 0.013609562395474074,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015504020000000002,
      "loss": 3.1005,
      "step": 224800
    },
    {
      "epoch": 0.013628205631632258,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001550382,
      "loss": 3.2086,
      "step": 224810
    },
    {
      "epoch": 0.013646848867790442,
      "grad_norm": 0.75,
      "learning_rate": 0.0001550362,
      "loss": 3.1827,
      "step": 224820
    },
    {
      "epoch": 0.013665492103948625,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001550342,
      "loss": 3.1263,
      "step": 224830
    },
    {
      "epoch": 0.013684135340106809,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001550322,
      "loss": 3.0935,
      "step": 224840
    },
    {
      "epoch": 0.013702778576264993,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015503020000000003,
      "loss": 3.187,
      "step": 224850
    },
    {
      "epoch": 0.013721421812423177,
      "grad_norm": 0.75,
      "learning_rate": 0.0001550282,
      "loss": 3.1266,
      "step": 224860
    },
    {
      "epoch": 0.01374006504858136,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015502620000000002,
      "loss": 3.0768,
      "step": 224870
    },
    {
      "epoch": 0.013758708284739544,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015502419999999999,
      "loss": 3.1814,
      "step": 224880
    },
    {
      "epoch": 0.013777351520897728,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001550222,
      "loss": 3.1877,
      "step": 224890
    },
    {
      "epoch": 0.013795994757055912,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001550202,
      "loss": 3.1601,
      "step": 224900
    },
    {
      "epoch": 0.013814637993214095,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001550182,
      "loss": 3.2095,
      "step": 224910
    },
    {
      "epoch": 0.01383328122937228,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015501620000000002,
      "loss": 3.2207,
      "step": 224920
    },
    {
      "epoch": 0.013851924465530463,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001550142,
      "loss": 3.2159,
      "step": 224930
    },
    {
      "epoch": 0.013870567701688647,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001550122,
      "loss": 3.0651,
      "step": 224940
    },
    {
      "epoch": 0.01388921093784683,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001550102,
      "loss": 3.083,
      "step": 224950
    },
    {
      "epoch": 0.013907854174005013,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001550082,
      "loss": 3.2187,
      "step": 224960
    },
    {
      "epoch": 0.013926497410163196,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001550062,
      "loss": 3.1434,
      "step": 224970
    },
    {
      "epoch": 0.01394514064632138,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001550042,
      "loss": 3.1887,
      "step": 224980
    },
    {
      "epoch": 0.013963783882479564,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015500220000000002,
      "loss": 3.1307,
      "step": 224990
    },
    {
      "epoch": 0.013982427118637748,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015500020000000001,
      "loss": 3.0957,
      "step": 225000
    },
    {
      "epoch": 0.014001070354795931,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001549982,
      "loss": 3.2633,
      "step": 225010
    },
    {
      "epoch": 0.014019713590954115,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001549962,
      "loss": 3.1777,
      "step": 225020
    },
    {
      "epoch": 0.014038356827112299,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001549942,
      "loss": 3.1158,
      "step": 225030
    },
    {
      "epoch": 0.014057000063270483,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001549922,
      "loss": 3.1681,
      "step": 225040
    },
    {
      "epoch": 0.014075643299428666,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001549902,
      "loss": 3.2422,
      "step": 225050
    },
    {
      "epoch": 0.01409428653558685,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015498820000000001,
      "loss": 3.1289,
      "step": 225060
    },
    {
      "epoch": 0.014112929771745034,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001549862,
      "loss": 3.0672,
      "step": 225070
    },
    {
      "epoch": 0.014131573007903218,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001549842,
      "loss": 3.1546,
      "step": 225080
    },
    {
      "epoch": 0.014150216244061401,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001549822,
      "loss": 3.2226,
      "step": 225090
    },
    {
      "epoch": 0.014168859480219585,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001549802,
      "loss": 3.1027,
      "step": 225100
    },
    {
      "epoch": 0.014187502716377769,
      "grad_norm": 0.75,
      "learning_rate": 0.00015497820000000002,
      "loss": 3.0709,
      "step": 225110
    },
    {
      "epoch": 0.014206145952535953,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015497620000000002,
      "loss": 3.0909,
      "step": 225120
    },
    {
      "epoch": 0.014224789188694137,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001549742,
      "loss": 3.096,
      "step": 225130
    },
    {
      "epoch": 0.014243432424852319,
      "grad_norm": 0.75,
      "learning_rate": 0.0001549722,
      "loss": 3.2062,
      "step": 225140
    },
    {
      "epoch": 0.014262075661010502,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001549702,
      "loss": 3.1851,
      "step": 225150
    },
    {
      "epoch": 0.014280718897168686,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001549682,
      "loss": 3.1618,
      "step": 225160
    },
    {
      "epoch": 0.01429936213332687,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015496620000000002,
      "loss": 3.2883,
      "step": 225170
    },
    {
      "epoch": 0.014318005369485054,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015496420000000002,
      "loss": 3.0798,
      "step": 225180
    },
    {
      "epoch": 0.014336648605643237,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001549622,
      "loss": 2.958,
      "step": 225190
    },
    {
      "epoch": 0.014355291841801421,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001549602,
      "loss": 3.2147,
      "step": 225200
    },
    {
      "epoch": 0.014373935077959605,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001549582,
      "loss": 3.1304,
      "step": 225210
    },
    {
      "epoch": 0.014392578314117789,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015495620000000003,
      "loss": 2.9798,
      "step": 225220
    },
    {
      "epoch": 0.014411221550275972,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001549542,
      "loss": 3.1745,
      "step": 225230
    },
    {
      "epoch": 0.014429864786434156,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015495220000000002,
      "loss": 3.1604,
      "step": 225240
    },
    {
      "epoch": 0.01444850802259234,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001549502,
      "loss": 3.0909,
      "step": 225250
    },
    {
      "epoch": 0.014467151258750524,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001549482,
      "loss": 3.1947,
      "step": 225260
    },
    {
      "epoch": 0.014485794494908708,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001549462,
      "loss": 3.0541,
      "step": 225270
    },
    {
      "epoch": 0.014504437731066891,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001549442,
      "loss": 3.144,
      "step": 225280
    },
    {
      "epoch": 0.014523080967225075,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00015494220000000002,
      "loss": 3.136,
      "step": 225290
    },
    {
      "epoch": 0.014541724203383259,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001549402,
      "loss": 3.1479,
      "step": 225300
    },
    {
      "epoch": 0.01456036743954144,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001549382,
      "loss": 3.1995,
      "step": 225310
    },
    {
      "epoch": 0.014579010675699625,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001549362,
      "loss": 3.2017,
      "step": 225320
    },
    {
      "epoch": 0.014597653911857808,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001549342,
      "loss": 3.1457,
      "step": 225330
    },
    {
      "epoch": 0.014616297148015992,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015493220000000003,
      "loss": 3.0676,
      "step": 225340
    },
    {
      "epoch": 0.014634940384174176,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001549302,
      "loss": 3.1165,
      "step": 225350
    },
    {
      "epoch": 0.01465358362033236,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015492820000000002,
      "loss": 3.1972,
      "step": 225360
    },
    {
      "epoch": 0.014672226856490543,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001549262,
      "loss": 3.0758,
      "step": 225370
    },
    {
      "epoch": 0.014690870092648727,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001549242,
      "loss": 3.1069,
      "step": 225380
    },
    {
      "epoch": 0.014709513328806911,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001549222,
      "loss": 3.1423,
      "step": 225390
    },
    {
      "epoch": 0.014728156564965095,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001549202,
      "loss": 3.1414,
      "step": 225400
    },
    {
      "epoch": 0.014746799801123278,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00015491820000000002,
      "loss": 3.1622,
      "step": 225410
    },
    {
      "epoch": 0.014765443037281462,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001549162,
      "loss": 3.2764,
      "step": 225420
    },
    {
      "epoch": 0.014784086273439646,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015491420000000002,
      "loss": 3.0955,
      "step": 225430
    },
    {
      "epoch": 0.01480272950959783,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001549122,
      "loss": 3.073,
      "step": 225440
    },
    {
      "epoch": 0.014821372745756014,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001549102,
      "loss": 3.1647,
      "step": 225450
    },
    {
      "epoch": 0.014840015981914197,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001549082,
      "loss": 3.2229,
      "step": 225460
    },
    {
      "epoch": 0.014858659218072381,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001549062,
      "loss": 3.0345,
      "step": 225470
    },
    {
      "epoch": 0.014877302454230563,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015490420000000002,
      "loss": 3.0972,
      "step": 225480
    },
    {
      "epoch": 0.014895945690388747,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001549022,
      "loss": 3.0957,
      "step": 225490
    },
    {
      "epoch": 0.01491458892654693,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001549002,
      "loss": 3.1034,
      "step": 225500
    },
    {
      "epoch": 0.014933232162705114,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001548982,
      "loss": 3.174,
      "step": 225510
    },
    {
      "epoch": 0.014951875398863298,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001548962,
      "loss": 3.0491,
      "step": 225520
    },
    {
      "epoch": 0.014970518635021482,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001548942,
      "loss": 3.1326,
      "step": 225530
    },
    {
      "epoch": 0.014989161871179666,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001548922,
      "loss": 3.0883,
      "step": 225540
    },
    {
      "epoch": 0.01500780510733785,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015489020000000002,
      "loss": 3.0609,
      "step": 225550
    },
    {
      "epoch": 0.015026448343496033,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001548882,
      "loss": 3.06,
      "step": 225560
    },
    {
      "epoch": 0.015045091579654217,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001548862,
      "loss": 3.2634,
      "step": 225570
    },
    {
      "epoch": 0.0150637348158124,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001548842,
      "loss": 3.1453,
      "step": 225580
    },
    {
      "epoch": 0.015082378051970585,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001548822,
      "loss": 3.1595,
      "step": 225590
    },
    {
      "epoch": 0.015101021288128768,
      "grad_norm": 0.75,
      "learning_rate": 0.00015488020000000002,
      "loss": 3.1057,
      "step": 225600
    },
    {
      "epoch": 0.015119664524286952,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015487820000000002,
      "loss": 3.234,
      "step": 225610
    },
    {
      "epoch": 0.015138307760445136,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001548762,
      "loss": 3.0067,
      "step": 225620
    },
    {
      "epoch": 0.01515695099660332,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001548742,
      "loss": 3.1878,
      "step": 225630
    },
    {
      "epoch": 0.015175594232761503,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001548722,
      "loss": 3.1752,
      "step": 225640
    },
    {
      "epoch": 0.015194237468919685,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001548702,
      "loss": 3.1085,
      "step": 225650
    },
    {
      "epoch": 0.01521288070507787,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001548682,
      "loss": 3.0033,
      "step": 225660
    },
    {
      "epoch": 0.015231523941236053,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015486620000000002,
      "loss": 3.172,
      "step": 225670
    },
    {
      "epoch": 0.015250167177394237,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015486420000000001,
      "loss": 3.0546,
      "step": 225680
    },
    {
      "epoch": 0.01526881041355242,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001548622,
      "loss": 3.1508,
      "step": 225690
    },
    {
      "epoch": 0.015287453649710604,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001548602,
      "loss": 3.067,
      "step": 225700
    },
    {
      "epoch": 0.015306096885868788,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001548582,
      "loss": 3.0995,
      "step": 225710
    },
    {
      "epoch": 0.015324740122026972,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001548562,
      "loss": 3.1743,
      "step": 225720
    },
    {
      "epoch": 0.015343383358185156,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015485420000000002,
      "loss": 3.2235,
      "step": 225730
    },
    {
      "epoch": 0.01536202659434334,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015485220000000001,
      "loss": 3.1215,
      "step": 225740
    },
    {
      "epoch": 0.015380669830501523,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001548502,
      "loss": 3.2892,
      "step": 225750
    },
    {
      "epoch": 0.015399313066659707,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001548482,
      "loss": 3.1673,
      "step": 225760
    },
    {
      "epoch": 0.01541795630281789,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001548462,
      "loss": 3.2269,
      "step": 225770
    },
    {
      "epoch": 0.015436599538976074,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015484420000000002,
      "loss": 3.0657,
      "step": 225780
    },
    {
      "epoch": 0.015455242775134258,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001548422,
      "loss": 3.1624,
      "step": 225790
    },
    {
      "epoch": 0.015473886011292442,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015484020000000002,
      "loss": 3.0126,
      "step": 225800
    },
    {
      "epoch": 0.015492529247450626,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001548382,
      "loss": 3.1728,
      "step": 225810
    },
    {
      "epoch": 0.015511172483608808,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001548362,
      "loss": 3.0777,
      "step": 225820
    },
    {
      "epoch": 0.015529815719766991,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015483420000000003,
      "loss": 3.1833,
      "step": 225830
    },
    {
      "epoch": 0.015548458955925175,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001548322,
      "loss": 3.0918,
      "step": 225840
    },
    {
      "epoch": 0.015567102192083359,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015483020000000002,
      "loss": 3.1391,
      "step": 225850
    },
    {
      "epoch": 0.015585745428241543,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001548282,
      "loss": 3.1141,
      "step": 225860
    },
    {
      "epoch": 0.015604388664399726,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001548262,
      "loss": 3.1681,
      "step": 225870
    },
    {
      "epoch": 0.01562303190055791,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001548242,
      "loss": 3.0389,
      "step": 225880
    },
    {
      "epoch": 0.015641675136716094,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001548222,
      "loss": 3.083,
      "step": 225890
    },
    {
      "epoch": 0.01566031837287428,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015482020000000003,
      "loss": 3.2214,
      "step": 225900
    },
    {
      "epoch": 0.01567896160903246,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001548182,
      "loss": 3.0575,
      "step": 225910
    },
    {
      "epoch": 0.015697604845190644,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015481620000000002,
      "loss": 3.1073,
      "step": 225920
    },
    {
      "epoch": 0.01571624808134883,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015481419999999999,
      "loss": 3.0182,
      "step": 225930
    },
    {
      "epoch": 0.01573489131750701,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001548122,
      "loss": 3.0523,
      "step": 225940
    },
    {
      "epoch": 0.015753534553665197,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001548102,
      "loss": 3.1565,
      "step": 225950
    },
    {
      "epoch": 0.01577217778982338,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001548082,
      "loss": 3.002,
      "step": 225960
    },
    {
      "epoch": 0.015790821025981564,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015480620000000002,
      "loss": 3.0324,
      "step": 225970
    },
    {
      "epoch": 0.015809464262139746,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001548042,
      "loss": 3.0752,
      "step": 225980
    },
    {
      "epoch": 0.01582810749829793,
      "grad_norm": 0.75,
      "learning_rate": 0.0001548022,
      "loss": 3.1153,
      "step": 225990
    },
    {
      "epoch": 0.015846750734456114,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001548002,
      "loss": 3.0599,
      "step": 226000
    },
    {
      "epoch": 0.0158653939706143,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001547982,
      "loss": 3.0118,
      "step": 226010
    },
    {
      "epoch": 0.01588403720677248,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001547962,
      "loss": 3.199,
      "step": 226020
    },
    {
      "epoch": 0.015902680442930667,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001547942,
      "loss": 3.1964,
      "step": 226030
    },
    {
      "epoch": 0.01592132367908885,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015479220000000002,
      "loss": 3.1182,
      "step": 226040
    },
    {
      "epoch": 0.015939966915247034,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001547902,
      "loss": 3.0739,
      "step": 226050
    },
    {
      "epoch": 0.015958610151405216,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001547882,
      "loss": 2.9822,
      "step": 226060
    },
    {
      "epoch": 0.015977253387563402,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001547862,
      "loss": 3.0901,
      "step": 226070
    },
    {
      "epoch": 0.015995896623721584,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001547842,
      "loss": 3.086,
      "step": 226080
    },
    {
      "epoch": 0.016014539859879766,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015478220000000002,
      "loss": 3.1715,
      "step": 226090
    },
    {
      "epoch": 0.01603318309603795,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001547802,
      "loss": 3.1162,
      "step": 226100
    },
    {
      "epoch": 0.016051826332196133,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015477820000000002,
      "loss": 3.0931,
      "step": 226110
    },
    {
      "epoch": 0.01607046956835432,
      "grad_norm": 0.75,
      "learning_rate": 0.0001547762,
      "loss": 3.0431,
      "step": 226120
    },
    {
      "epoch": 0.0160891128045125,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001547742,
      "loss": 3.1742,
      "step": 226130
    },
    {
      "epoch": 0.016107756040670686,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001547722,
      "loss": 3.2176,
      "step": 226140
    },
    {
      "epoch": 0.01612639927682887,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001547702,
      "loss": 2.9967,
      "step": 226150
    },
    {
      "epoch": 0.016145042512987054,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015476820000000002,
      "loss": 3.2105,
      "step": 226160
    },
    {
      "epoch": 0.016163685749145236,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015476620000000002,
      "loss": 3.0545,
      "step": 226170
    },
    {
      "epoch": 0.01618232898530342,
      "grad_norm": 0.75,
      "learning_rate": 0.0001547642,
      "loss": 2.9999,
      "step": 226180
    },
    {
      "epoch": 0.016200972221461604,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001547622,
      "loss": 3.0245,
      "step": 226190
    },
    {
      "epoch": 0.01621961545761979,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001547602,
      "loss": 3.0489,
      "step": 226200
    },
    {
      "epoch": 0.01623825869377797,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001547582,
      "loss": 3.1001,
      "step": 226210
    },
    {
      "epoch": 0.016256901929936157,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015475620000000002,
      "loss": 2.9757,
      "step": 226220
    },
    {
      "epoch": 0.01627554516609434,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015475420000000002,
      "loss": 3.212,
      "step": 226230
    },
    {
      "epoch": 0.016294188402252524,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001547522,
      "loss": 3.177,
      "step": 226240
    },
    {
      "epoch": 0.016312831638410706,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001547502,
      "loss": 3.1568,
      "step": 226250
    },
    {
      "epoch": 0.016331474874568888,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001547482,
      "loss": 3.1109,
      "step": 226260
    },
    {
      "epoch": 0.016350118110727074,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001547462,
      "loss": 3.2093,
      "step": 226270
    },
    {
      "epoch": 0.016368761346885256,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001547442,
      "loss": 3.0834,
      "step": 226280
    },
    {
      "epoch": 0.01638740458304344,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015474220000000002,
      "loss": 3.2063,
      "step": 226290
    },
    {
      "epoch": 0.016406047819201623,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001547402,
      "loss": 3.293,
      "step": 226300
    },
    {
      "epoch": 0.01642469105535981,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001547382,
      "loss": 3.0897,
      "step": 226310
    },
    {
      "epoch": 0.01644333429151799,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001547362,
      "loss": 3.0348,
      "step": 226320
    },
    {
      "epoch": 0.016461977527676176,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001547342,
      "loss": 3.1842,
      "step": 226330
    },
    {
      "epoch": 0.016480620763834358,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015473220000000002,
      "loss": 3.0933,
      "step": 226340
    },
    {
      "epoch": 0.016499263999992544,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001547302,
      "loss": 2.9944,
      "step": 226350
    },
    {
      "epoch": 0.016517907236150726,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015472820000000001,
      "loss": 3.1309,
      "step": 226360
    },
    {
      "epoch": 0.01653655047230891,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001547262,
      "loss": 3.1208,
      "step": 226370
    },
    {
      "epoch": 0.016555193708467093,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001547242,
      "loss": 3.1438,
      "step": 226380
    },
    {
      "epoch": 0.01657383694462528,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015472220000000003,
      "loss": 3.1722,
      "step": 226390
    },
    {
      "epoch": 0.01659248018078346,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001547202,
      "loss": 3.0155,
      "step": 226400
    },
    {
      "epoch": 0.016611123416941646,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015471820000000002,
      "loss": 3.0586,
      "step": 226410
    },
    {
      "epoch": 0.01662976665309983,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001547162,
      "loss": 3.0847,
      "step": 226420
    },
    {
      "epoch": 0.01664840988925801,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001547142,
      "loss": 3.1353,
      "step": 226430
    },
    {
      "epoch": 0.016667053125416196,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001547122,
      "loss": 3.231,
      "step": 226440
    },
    {
      "epoch": 0.016685696361574378,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001547102,
      "loss": 3.1492,
      "step": 226450
    },
    {
      "epoch": 0.016704339597732563,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015470820000000002,
      "loss": 3.1375,
      "step": 226460
    },
    {
      "epoch": 0.016722982833890745,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001547062,
      "loss": 3.0215,
      "step": 226470
    },
    {
      "epoch": 0.01674162607004893,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015470420000000002,
      "loss": 3.1513,
      "step": 226480
    },
    {
      "epoch": 0.016760269306207113,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015470219999999998,
      "loss": 3.1552,
      "step": 226490
    },
    {
      "epoch": 0.0167789125423653,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001547002,
      "loss": 3.2354,
      "step": 226500
    },
    {
      "epoch": 0.01679755577852348,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001546982,
      "loss": 3.159,
      "step": 226510
    },
    {
      "epoch": 0.016816199014681666,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001546962,
      "loss": 3.1656,
      "step": 226520
    },
    {
      "epoch": 0.016834842250839848,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015469420000000002,
      "loss": 3.0994,
      "step": 226530
    },
    {
      "epoch": 0.016853485486998034,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001546922,
      "loss": 3.1287,
      "step": 226540
    },
    {
      "epoch": 0.016872128723156216,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001546902,
      "loss": 3.1055,
      "step": 226550
    },
    {
      "epoch": 0.0168907719593144,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001546882,
      "loss": 3.2046,
      "step": 226560
    },
    {
      "epoch": 0.016909415195472583,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001546862,
      "loss": 3.226,
      "step": 226570
    },
    {
      "epoch": 0.01692805843163077,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015468420000000003,
      "loss": 3.175,
      "step": 226580
    },
    {
      "epoch": 0.01694670166778895,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001546822,
      "loss": 3.1787,
      "step": 226590
    },
    {
      "epoch": 0.016965344903947133,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015468020000000002,
      "loss": 3.1212,
      "step": 226600
    },
    {
      "epoch": 0.016983988140105318,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001546782,
      "loss": 3.2517,
      "step": 226610
    },
    {
      "epoch": 0.0170026313762635,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001546762,
      "loss": 3.0366,
      "step": 226620
    },
    {
      "epoch": 0.017021274612421686,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001546742,
      "loss": 3.2705,
      "step": 226630
    },
    {
      "epoch": 0.017039917848579868,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001546722,
      "loss": 3.182,
      "step": 226640
    },
    {
      "epoch": 0.017058561084738053,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015467020000000002,
      "loss": 2.9748,
      "step": 226650
    },
    {
      "epoch": 0.017077204320896235,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001546682,
      "loss": 3.1415,
      "step": 226660
    },
    {
      "epoch": 0.01709584755705442,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001546662,
      "loss": 3.1011,
      "step": 226670
    },
    {
      "epoch": 0.017114490793212603,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001546642,
      "loss": 3.1412,
      "step": 226680
    },
    {
      "epoch": 0.01713313402937079,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001546622,
      "loss": 3.2282,
      "step": 226690
    },
    {
      "epoch": 0.01715177726552897,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001546602,
      "loss": 3.0794,
      "step": 226700
    },
    {
      "epoch": 0.017170420501687156,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001546582,
      "loss": 3.1533,
      "step": 226710
    },
    {
      "epoch": 0.017189063737845338,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015465620000000002,
      "loss": 3.1438,
      "step": 226720
    },
    {
      "epoch": 0.017207706974003523,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015465420000000001,
      "loss": 3.0574,
      "step": 226730
    },
    {
      "epoch": 0.017226350210161705,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001546522,
      "loss": 3.229,
      "step": 226740
    },
    {
      "epoch": 0.01724499344631989,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001546502,
      "loss": 3.218,
      "step": 226750
    },
    {
      "epoch": 0.017263636682478073,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001546482,
      "loss": 3.3038,
      "step": 226760
    },
    {
      "epoch": 0.017282279918636255,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001546462,
      "loss": 3.1726,
      "step": 226770
    },
    {
      "epoch": 0.01730092315479444,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015464420000000002,
      "loss": 3.0923,
      "step": 226780
    },
    {
      "epoch": 0.017319566390952622,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015464220000000002,
      "loss": 3.1105,
      "step": 226790
    },
    {
      "epoch": 0.017338209627110808,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001546402,
      "loss": 3.2051,
      "step": 226800
    },
    {
      "epoch": 0.01735685286326899,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001546382,
      "loss": 3.1802,
      "step": 226810
    },
    {
      "epoch": 0.017375496099427176,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001546362,
      "loss": 2.9969,
      "step": 226820
    },
    {
      "epoch": 0.017394139335585358,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015463420000000002,
      "loss": 3.2269,
      "step": 226830
    },
    {
      "epoch": 0.017412782571743543,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001546322,
      "loss": 3.2199,
      "step": 226840
    },
    {
      "epoch": 0.017431425807901725,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015463020000000002,
      "loss": 3.1692,
      "step": 226850
    },
    {
      "epoch": 0.01745006904405991,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001546282,
      "loss": 3.0736,
      "step": 226860
    },
    {
      "epoch": 0.017468712280218093,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001546262,
      "loss": 3.148,
      "step": 226870
    },
    {
      "epoch": 0.017487355516376278,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001546242,
      "loss": 3.104,
      "step": 226880
    },
    {
      "epoch": 0.01750599875253446,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001546222,
      "loss": 3.1324,
      "step": 226890
    },
    {
      "epoch": 0.017524641988692646,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015462020000000002,
      "loss": 3.0922,
      "step": 226900
    },
    {
      "epoch": 0.017543285224850828,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001546182,
      "loss": 3.1571,
      "step": 226910
    },
    {
      "epoch": 0.017561928461009013,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001546162,
      "loss": 3.1721,
      "step": 226920
    },
    {
      "epoch": 0.017580571697167195,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001546142,
      "loss": 3.0357,
      "step": 226930
    },
    {
      "epoch": 0.017599214933325377,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001546122,
      "loss": 3.222,
      "step": 226940
    },
    {
      "epoch": 0.017617858169483563,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015461020000000003,
      "loss": 3.1106,
      "step": 226950
    },
    {
      "epoch": 0.017636501405641745,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001546082,
      "loss": 3.1481,
      "step": 226960
    },
    {
      "epoch": 0.01765514464179993,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015460620000000002,
      "loss": 3.158,
      "step": 226970
    },
    {
      "epoch": 0.017673787877958112,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00015460419999999999,
      "loss": 3.0792,
      "step": 226980
    },
    {
      "epoch": 0.017692431114116298,
      "grad_norm": 0.75,
      "learning_rate": 0.0001546022,
      "loss": 3.2419,
      "step": 226990
    },
    {
      "epoch": 0.01771107435027448,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001546002,
      "loss": 3.2308,
      "step": 227000
    },
    {
      "epoch": 0.017729717586432665,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001545982,
      "loss": 3.1585,
      "step": 227010
    },
    {
      "epoch": 0.017748360822590847,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015459620000000002,
      "loss": 3.0837,
      "step": 227020
    },
    {
      "epoch": 0.017767004058749033,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001545942,
      "loss": 3.0455,
      "step": 227030
    },
    {
      "epoch": 0.017785647294907215,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015459220000000001,
      "loss": 3.0196,
      "step": 227040
    },
    {
      "epoch": 0.0178042905310654,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001545902,
      "loss": 3.2734,
      "step": 227050
    },
    {
      "epoch": 0.017822933767223582,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001545882,
      "loss": 3.1481,
      "step": 227060
    },
    {
      "epoch": 0.017841577003381768,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015458620000000003,
      "loss": 3.1577,
      "step": 227070
    },
    {
      "epoch": 0.01786022023953995,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001545842,
      "loss": 3.1607,
      "step": 227080
    },
    {
      "epoch": 0.017878863475698135,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015458220000000002,
      "loss": 3.2891,
      "step": 227090
    },
    {
      "epoch": 0.017897506711856317,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001545802,
      "loss": 3.1548,
      "step": 227100
    },
    {
      "epoch": 0.0179161499480145,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001545782,
      "loss": 2.9458,
      "step": 227110
    },
    {
      "epoch": 0.017934793184172685,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001545762,
      "loss": 3.175,
      "step": 227120
    },
    {
      "epoch": 0.017953436420330867,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001545742,
      "loss": 3.0335,
      "step": 227130
    },
    {
      "epoch": 0.017972079656489053,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015457220000000002,
      "loss": 3.0983,
      "step": 227140
    },
    {
      "epoch": 0.017990722892647235,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001545702,
      "loss": 3.1215,
      "step": 227150
    },
    {
      "epoch": 0.01800936612880542,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015456820000000002,
      "loss": 2.989,
      "step": 227160
    },
    {
      "epoch": 0.018028009364963602,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001545662,
      "loss": 3.1146,
      "step": 227170
    },
    {
      "epoch": 0.018046652601121788,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001545642,
      "loss": 3.1823,
      "step": 227180
    },
    {
      "epoch": 0.01806529583727997,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001545622,
      "loss": 3.1219,
      "step": 227190
    },
    {
      "epoch": 0.018083939073438155,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001545602,
      "loss": 3.0149,
      "step": 227200
    },
    {
      "epoch": 0.018102582309596337,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00015455820000000002,
      "loss": 3.3124,
      "step": 227210
    },
    {
      "epoch": 0.018121225545754523,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015455620000000002,
      "loss": 3.1468,
      "step": 227220
    },
    {
      "epoch": 0.018139868781912705,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001545542,
      "loss": 3.2496,
      "step": 227230
    },
    {
      "epoch": 0.01815851201807089,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001545522,
      "loss": 3.1947,
      "step": 227240
    },
    {
      "epoch": 0.018177155254229072,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001545502,
      "loss": 3.0314,
      "step": 227250
    },
    {
      "epoch": 0.018195798490387258,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001545482,
      "loss": 3.1522,
      "step": 227260
    },
    {
      "epoch": 0.01821444172654544,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001545462,
      "loss": 3.1348,
      "step": 227270
    },
    {
      "epoch": 0.018233084962703622,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015454420000000002,
      "loss": 3.215,
      "step": 227280
    },
    {
      "epoch": 0.018251728198861807,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001545422,
      "loss": 3.2357,
      "step": 227290
    },
    {
      "epoch": 0.01827037143501999,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001545402,
      "loss": 3.2024,
      "step": 227300
    },
    {
      "epoch": 0.018289014671178175,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001545382,
      "loss": 3.1299,
      "step": 227310
    },
    {
      "epoch": 0.018307657907336357,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001545362,
      "loss": 3.1359,
      "step": 227320
    },
    {
      "epoch": 0.018326301143494542,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001545342,
      "loss": 3.1304,
      "step": 227330
    },
    {
      "epoch": 0.018344944379652724,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015453220000000002,
      "loss": 3.0843,
      "step": 227340
    },
    {
      "epoch": 0.01836358761581091,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001545302,
      "loss": 2.9825,
      "step": 227350
    },
    {
      "epoch": 0.018382230851969092,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001545282,
      "loss": 3.257,
      "step": 227360
    },
    {
      "epoch": 0.018400874088127277,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001545262,
      "loss": 3.1208,
      "step": 227370
    },
    {
      "epoch": 0.01841951732428546,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001545242,
      "loss": 3.1564,
      "step": 227380
    },
    {
      "epoch": 0.018438160560443645,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015452220000000002,
      "loss": 3.1727,
      "step": 227390
    },
    {
      "epoch": 0.018456803796601827,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001545202,
      "loss": 3.1846,
      "step": 227400
    },
    {
      "epoch": 0.018475447032760012,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015451820000000001,
      "loss": 3.2417,
      "step": 227410
    },
    {
      "epoch": 0.018494090268918194,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001545162,
      "loss": 3.2519,
      "step": 227420
    },
    {
      "epoch": 0.01851273350507638,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001545142,
      "loss": 3.1167,
      "step": 227430
    },
    {
      "epoch": 0.018531376741234562,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015451220000000003,
      "loss": 3.1129,
      "step": 227440
    },
    {
      "epoch": 0.018550019977392744,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001545102,
      "loss": 3.17,
      "step": 227450
    },
    {
      "epoch": 0.01856866321355093,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015450820000000002,
      "loss": 3.2861,
      "step": 227460
    },
    {
      "epoch": 0.01858730644970911,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001545062,
      "loss": 3.0948,
      "step": 227470
    },
    {
      "epoch": 0.018605949685867297,
      "grad_norm": 0.75,
      "learning_rate": 0.0001545042,
      "loss": 3.0554,
      "step": 227480
    },
    {
      "epoch": 0.01862459292202548,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001545022,
      "loss": 3.1959,
      "step": 227490
    },
    {
      "epoch": 0.018643236158183665,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001545002,
      "loss": 3.1274,
      "step": 227500
    },
    {
      "epoch": 0.018661879394341847,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015449820000000002,
      "loss": 3.2175,
      "step": 227510
    },
    {
      "epoch": 0.018680522630500032,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001544962,
      "loss": 3.1619,
      "step": 227520
    },
    {
      "epoch": 0.018699165866658214,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015449420000000002,
      "loss": 3.1603,
      "step": 227530
    },
    {
      "epoch": 0.0187178091028164,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001544922,
      "loss": 3.1966,
      "step": 227540
    },
    {
      "epoch": 0.01873645233897458,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001544902,
      "loss": 3.099,
      "step": 227550
    },
    {
      "epoch": 0.018755095575132767,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015448820000000003,
      "loss": 3.1253,
      "step": 227560
    },
    {
      "epoch": 0.01877373881129095,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001544862,
      "loss": 3.2697,
      "step": 227570
    },
    {
      "epoch": 0.018792382047449135,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015448420000000002,
      "loss": 2.9813,
      "step": 227580
    },
    {
      "epoch": 0.018811025283607317,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001544822,
      "loss": 3.1932,
      "step": 227590
    },
    {
      "epoch": 0.018829668519765502,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001544802,
      "loss": 3.1573,
      "step": 227600
    },
    {
      "epoch": 0.018848311755923684,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001544782,
      "loss": 3.2069,
      "step": 227610
    },
    {
      "epoch": 0.018866954992081866,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001544762,
      "loss": 3.1508,
      "step": 227620
    },
    {
      "epoch": 0.018885598228240052,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015447420000000003,
      "loss": 2.9947,
      "step": 227630
    },
    {
      "epoch": 0.018904241464398234,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001544722,
      "loss": 2.943,
      "step": 227640
    },
    {
      "epoch": 0.01892288470055642,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015447020000000002,
      "loss": 3.0075,
      "step": 227650
    },
    {
      "epoch": 0.0189415279367146,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001544682,
      "loss": 3.3196,
      "step": 227660
    },
    {
      "epoch": 0.018960171172872787,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001544662,
      "loss": 3.2442,
      "step": 227670
    },
    {
      "epoch": 0.01897881440903097,
      "grad_norm": 0.75,
      "learning_rate": 0.0001544642,
      "loss": 3.0928,
      "step": 227680
    },
    {
      "epoch": 0.018997457645189154,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001544622,
      "loss": 3.2017,
      "step": 227690
    },
    {
      "epoch": 0.019016100881347336,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015446020000000002,
      "loss": 3.1821,
      "step": 227700
    },
    {
      "epoch": 0.019034744117505522,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001544582,
      "loss": 3.2581,
      "step": 227710
    },
    {
      "epoch": 0.019053387353663704,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015445620000000001,
      "loss": 3.1116,
      "step": 227720
    },
    {
      "epoch": 0.01907203058982189,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001544542,
      "loss": 3.169,
      "step": 227730
    },
    {
      "epoch": 0.01909067382598007,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001544522,
      "loss": 3.2537,
      "step": 227740
    },
    {
      "epoch": 0.019109317062138257,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001544502,
      "loss": 3.0962,
      "step": 227750
    },
    {
      "epoch": 0.01912796029829644,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001544482,
      "loss": 3.2413,
      "step": 227760
    },
    {
      "epoch": 0.019146603534454625,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015444620000000002,
      "loss": 3.1317,
      "step": 227770
    },
    {
      "epoch": 0.019165246770612807,
      "grad_norm": 0.625,
      "learning_rate": 0.00015444420000000001,
      "loss": 3.0863,
      "step": 227780
    },
    {
      "epoch": 0.01918389000677099,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001544422,
      "loss": 3.0427,
      "step": 227790
    },
    {
      "epoch": 0.019202533242929174,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001544402,
      "loss": 3.1314,
      "step": 227800
    },
    {
      "epoch": 0.019221176479087356,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001544382,
      "loss": 3.1424,
      "step": 227810
    },
    {
      "epoch": 0.01923981971524554,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001544362,
      "loss": 3.188,
      "step": 227820
    },
    {
      "epoch": 0.019258462951403724,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015443420000000002,
      "loss": 3.2318,
      "step": 227830
    },
    {
      "epoch": 0.01927710618756191,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015443220000000002,
      "loss": 3.0235,
      "step": 227840
    },
    {
      "epoch": 0.01929574942372009,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001544302,
      "loss": 3.2198,
      "step": 227850
    },
    {
      "epoch": 0.019314392659878277,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001544282,
      "loss": 3.152,
      "step": 227860
    },
    {
      "epoch": 0.01933303589603646,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001544262,
      "loss": 3.0918,
      "step": 227870
    },
    {
      "epoch": 0.019351679132194644,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001544242,
      "loss": 3.0736,
      "step": 227880
    },
    {
      "epoch": 0.019370322368352826,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001544222,
      "loss": 3.1223,
      "step": 227890
    },
    {
      "epoch": 0.019388965604511012,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015442020000000002,
      "loss": 3.1918,
      "step": 227900
    },
    {
      "epoch": 0.019407608840669194,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001544182,
      "loss": 3.1816,
      "step": 227910
    },
    {
      "epoch": 0.01942625207682738,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001544162,
      "loss": 3.1788,
      "step": 227920
    },
    {
      "epoch": 0.01944489531298556,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001544142,
      "loss": 3.1419,
      "step": 227930
    },
    {
      "epoch": 0.019463538549143747,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001544122,
      "loss": 3.0955,
      "step": 227940
    },
    {
      "epoch": 0.01948218178530193,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015441020000000002,
      "loss": 3.0589,
      "step": 227950
    },
    {
      "epoch": 0.019500825021460114,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001544082,
      "loss": 3.2057,
      "step": 227960
    },
    {
      "epoch": 0.019519468257618296,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001544062,
      "loss": 3.1813,
      "step": 227970
    },
    {
      "epoch": 0.01953811149377648,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001544042,
      "loss": 3.2586,
      "step": 227980
    },
    {
      "epoch": 0.019556754729934664,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001544022,
      "loss": 3.2142,
      "step": 227990
    },
    {
      "epoch": 0.019575397966092846,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015440020000000003,
      "loss": 3.0905,
      "step": 228000
    },
    {
      "epoch": 0.01959404120225103,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001543982,
      "loss": 3.0712,
      "step": 228010
    },
    {
      "epoch": 0.019612684438409213,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015439620000000002,
      "loss": 3.1275,
      "step": 228020
    },
    {
      "epoch": 0.0196313276745674,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001543942,
      "loss": 3.1444,
      "step": 228030
    },
    {
      "epoch": 0.01964997091072558,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001543922,
      "loss": 3.1063,
      "step": 228040
    },
    {
      "epoch": 0.019668614146883766,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015439020000000003,
      "loss": 3.0766,
      "step": 228050
    },
    {
      "epoch": 0.01968725738304195,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001543882,
      "loss": 3.2071,
      "step": 228060
    },
    {
      "epoch": 0.019705900619200134,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015438620000000002,
      "loss": 3.1233,
      "step": 228070
    },
    {
      "epoch": 0.019724543855358316,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001543842,
      "loss": 3.0889,
      "step": 228080
    },
    {
      "epoch": 0.0197431870915165,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015438220000000001,
      "loss": 3.0327,
      "step": 228090
    },
    {
      "epoch": 0.019761830327674684,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001543802,
      "loss": 3.1872,
      "step": 228100
    },
    {
      "epoch": 0.01978047356383287,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001543782,
      "loss": 3.2153,
      "step": 228110
    },
    {
      "epoch": 0.01979911679999105,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015437620000000003,
      "loss": 3.0783,
      "step": 228120
    },
    {
      "epoch": 0.019817760036149237,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001543742,
      "loss": 3.1704,
      "step": 228130
    },
    {
      "epoch": 0.01983640327230742,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015437220000000002,
      "loss": 3.1001,
      "step": 228140
    },
    {
      "epoch": 0.0198550465084656,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001543702,
      "loss": 3.1701,
      "step": 228150
    },
    {
      "epoch": 0.019873689744623786,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001543682,
      "loss": 3.2358,
      "step": 228160
    },
    {
      "epoch": 0.019892332980781968,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001543662,
      "loss": 3.2305,
      "step": 228170
    },
    {
      "epoch": 0.019910976216940154,
      "grad_norm": 0.75,
      "learning_rate": 0.0001543642,
      "loss": 3.1666,
      "step": 228180
    },
    {
      "epoch": 0.019929619453098336,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015436220000000002,
      "loss": 3.1831,
      "step": 228190
    },
    {
      "epoch": 0.01994826268925652,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001543602,
      "loss": 3.2243,
      "step": 228200
    },
    {
      "epoch": 0.019966905925414703,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015435820000000002,
      "loss": 3.1174,
      "step": 228210
    },
    {
      "epoch": 0.01998554916157289,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001543562,
      "loss": 3.1412,
      "step": 228220
    },
    {
      "epoch": 0.02000419239773107,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001543542,
      "loss": 3.0843,
      "step": 228230
    },
    {
      "epoch": 0.020022835633889256,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001543522,
      "loss": 3.1187,
      "step": 228240
    },
    {
      "epoch": 0.02004147887004744,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001543502,
      "loss": 3.1032,
      "step": 228250
    },
    {
      "epoch": 0.020060122106205624,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015434820000000002,
      "loss": 3.1354,
      "step": 228260
    },
    {
      "epoch": 0.020078765342363806,
      "grad_norm": 0.875,
      "learning_rate": 0.0001543462,
      "loss": 3.0721,
      "step": 228270
    },
    {
      "epoch": 0.02009740857852199,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001543442,
      "loss": 3.198,
      "step": 228280
    },
    {
      "epoch": 0.020116051814680173,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001543422,
      "loss": 3.1347,
      "step": 228290
    },
    {
      "epoch": 0.02013469505083836,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001543402,
      "loss": 3.1329,
      "step": 228300
    },
    {
      "epoch": 0.02015333828699654,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001543382,
      "loss": 3.0342,
      "step": 228310
    },
    {
      "epoch": 0.020171981523154723,
      "grad_norm": 2.328125,
      "learning_rate": 0.0001543362,
      "loss": 3.1734,
      "step": 228320
    },
    {
      "epoch": 0.02019062475931291,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015433420000000002,
      "loss": 3.0704,
      "step": 228330
    },
    {
      "epoch": 0.02020926799547109,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001543322,
      "loss": 3.0982,
      "step": 228340
    },
    {
      "epoch": 0.020227911231629276,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001543302,
      "loss": 2.9317,
      "step": 228350
    },
    {
      "epoch": 0.020246554467787458,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001543282,
      "loss": 3.0039,
      "step": 228360
    },
    {
      "epoch": 0.020265197703945644,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001543262,
      "loss": 3.0999,
      "step": 228370
    },
    {
      "epoch": 0.020283840940103826,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001543242,
      "loss": 3.0284,
      "step": 228380
    },
    {
      "epoch": 0.02030248417626201,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015432220000000002,
      "loss": 3.0463,
      "step": 228390
    },
    {
      "epoch": 0.020321127412420193,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015432020000000001,
      "loss": 3.1226,
      "step": 228400
    },
    {
      "epoch": 0.02033977064857838,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001543182,
      "loss": 3.0336,
      "step": 228410
    },
    {
      "epoch": 0.02035841388473656,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001543162,
      "loss": 3.2509,
      "step": 228420
    },
    {
      "epoch": 0.020377057120894746,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001543142,
      "loss": 3.1568,
      "step": 228430
    },
    {
      "epoch": 0.020395700357052928,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015431220000000002,
      "loss": 3.0325,
      "step": 228440
    },
    {
      "epoch": 0.020414343593211114,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001543102,
      "loss": 3.1803,
      "step": 228450
    },
    {
      "epoch": 0.020432986829369296,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015430820000000001,
      "loss": 3.1567,
      "step": 228460
    },
    {
      "epoch": 0.02045163006552748,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001543062,
      "loss": 3.0097,
      "step": 228470
    },
    {
      "epoch": 0.020470273301685663,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001543042,
      "loss": 3.1795,
      "step": 228480
    },
    {
      "epoch": 0.020488916537843845,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001543022,
      "loss": 3.1621,
      "step": 228490
    },
    {
      "epoch": 0.02050755977400203,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001543002,
      "loss": 3.0953,
      "step": 228500
    },
    {
      "epoch": 0.020526203010160213,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015429820000000002,
      "loss": 3.0435,
      "step": 228510
    },
    {
      "epoch": 0.020544846246318398,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015429620000000002,
      "loss": 3.0763,
      "step": 228520
    },
    {
      "epoch": 0.02056348948247658,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001542942,
      "loss": 3.2112,
      "step": 228530
    },
    {
      "epoch": 0.020582132718634766,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001542922,
      "loss": 3.2517,
      "step": 228540
    },
    {
      "epoch": 0.020600775954792948,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001542902,
      "loss": 3.1565,
      "step": 228550
    },
    {
      "epoch": 0.020619419190951133,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015428820000000002,
      "loss": 3.1188,
      "step": 228560
    },
    {
      "epoch": 0.020638062427109315,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001542862,
      "loss": 3.2429,
      "step": 228570
    },
    {
      "epoch": 0.0206567056632675,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015428420000000002,
      "loss": 3.1453,
      "step": 228580
    },
    {
      "epoch": 0.020675348899425683,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001542822,
      "loss": 3.1121,
      "step": 228590
    },
    {
      "epoch": 0.02069399213558387,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001542802,
      "loss": 3.0607,
      "step": 228600
    },
    {
      "epoch": 0.02071263537174205,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015427820000000003,
      "loss": 3.0821,
      "step": 228610
    },
    {
      "epoch": 0.020731278607900236,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001542762,
      "loss": 3.1212,
      "step": 228620
    },
    {
      "epoch": 0.020749921844058418,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015427420000000002,
      "loss": 3.0853,
      "step": 228630
    },
    {
      "epoch": 0.020768565080216603,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001542722,
      "loss": 3.2346,
      "step": 228640
    },
    {
      "epoch": 0.020787208316374785,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001542702,
      "loss": 3.276,
      "step": 228650
    },
    {
      "epoch": 0.020805851552532968,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001542682,
      "loss": 3.1246,
      "step": 228660
    },
    {
      "epoch": 0.020824494788691153,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001542662,
      "loss": 3.1552,
      "step": 228670
    },
    {
      "epoch": 0.020843138024849335,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015426420000000003,
      "loss": 3.1462,
      "step": 228680
    },
    {
      "epoch": 0.02086178126100752,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001542622,
      "loss": 3.0543,
      "step": 228690
    },
    {
      "epoch": 0.020880424497165703,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015426020000000002,
      "loss": 3.1498,
      "step": 228700
    },
    {
      "epoch": 0.020899067733323888,
      "grad_norm": 0.75,
      "learning_rate": 0.00015425819999999999,
      "loss": 3.0834,
      "step": 228710
    },
    {
      "epoch": 0.02091771096948207,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001542562,
      "loss": 3.0785,
      "step": 228720
    },
    {
      "epoch": 0.020936354205640256,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001542542,
      "loss": 3.2872,
      "step": 228730
    },
    {
      "epoch": 0.020954997441798438,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001542522,
      "loss": 3.0721,
      "step": 228740
    },
    {
      "epoch": 0.020973640677956623,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015425020000000002,
      "loss": 3.0805,
      "step": 228750
    },
    {
      "epoch": 0.020992283914114805,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001542482,
      "loss": 3.146,
      "step": 228760
    },
    {
      "epoch": 0.02101092715027299,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015424620000000001,
      "loss": 3.1455,
      "step": 228770
    },
    {
      "epoch": 0.021029570386431173,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001542442,
      "loss": 3.13,
      "step": 228780
    },
    {
      "epoch": 0.021048213622589358,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001542422,
      "loss": 3.173,
      "step": 228790
    },
    {
      "epoch": 0.02106685685874754,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001542402,
      "loss": 3.1205,
      "step": 228800
    },
    {
      "epoch": 0.021085500094905726,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001542382,
      "loss": 3.1455,
      "step": 228810
    },
    {
      "epoch": 0.021104143331063908,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015423620000000002,
      "loss": 3.1173,
      "step": 228820
    },
    {
      "epoch": 0.02112278656722209,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015423420000000001,
      "loss": 3.2191,
      "step": 228830
    },
    {
      "epoch": 0.021141429803380275,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001542322,
      "loss": 3.1706,
      "step": 228840
    },
    {
      "epoch": 0.021160073039538457,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001542302,
      "loss": 3.1517,
      "step": 228850
    },
    {
      "epoch": 0.021178716275696643,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001542282,
      "loss": 3.2257,
      "step": 228860
    },
    {
      "epoch": 0.021197359511854825,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001542262,
      "loss": 3.3287,
      "step": 228870
    },
    {
      "epoch": 0.02121600274801301,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001542242,
      "loss": 3.0764,
      "step": 228880
    },
    {
      "epoch": 0.021234645984171192,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015422220000000002,
      "loss": 3.1885,
      "step": 228890
    },
    {
      "epoch": 0.021253289220329378,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001542202,
      "loss": 3.0906,
      "step": 228900
    },
    {
      "epoch": 0.02127193245648756,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001542182,
      "loss": 3.14,
      "step": 228910
    },
    {
      "epoch": 0.021290575692645745,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001542162,
      "loss": 3.2427,
      "step": 228920
    },
    {
      "epoch": 0.021309218928803927,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001542142,
      "loss": 3.1902,
      "step": 228930
    },
    {
      "epoch": 0.021327862164962113,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001542122,
      "loss": 3.0566,
      "step": 228940
    },
    {
      "epoch": 0.021346505401120295,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015421020000000002,
      "loss": 3.059,
      "step": 228950
    },
    {
      "epoch": 0.02136514863727848,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001542082,
      "loss": 3.2,
      "step": 228960
    },
    {
      "epoch": 0.021383791873436662,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001542062,
      "loss": 3.1838,
      "step": 228970
    },
    {
      "epoch": 0.021402435109594848,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001542042,
      "loss": 3.1657,
      "step": 228980
    },
    {
      "epoch": 0.02142107834575303,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001542022,
      "loss": 3.0792,
      "step": 228990
    },
    {
      "epoch": 0.021439721581911212,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015420020000000002,
      "loss": 3.2022,
      "step": 229000
    },
    {
      "epoch": 0.021458364818069398,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015419820000000002,
      "loss": 3.1287,
      "step": 229010
    },
    {
      "epoch": 0.02147700805422758,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001541962,
      "loss": 3.1034,
      "step": 229020
    },
    {
      "epoch": 0.021495651290385765,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001541942,
      "loss": 3.2552,
      "step": 229030
    },
    {
      "epoch": 0.021514294526543947,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001541922,
      "loss": 3.0063,
      "step": 229040
    },
    {
      "epoch": 0.021532937762702133,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015419020000000003,
      "loss": 3.1644,
      "step": 229050
    },
    {
      "epoch": 0.021551580998860315,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001541882,
      "loss": 3.3348,
      "step": 229060
    },
    {
      "epoch": 0.0215702242350185,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015418620000000002,
      "loss": 3.1579,
      "step": 229070
    },
    {
      "epoch": 0.021588867471176682,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015418420000000001,
      "loss": 3.1473,
      "step": 229080
    },
    {
      "epoch": 0.021607510707334868,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001541822,
      "loss": 3.042,
      "step": 229090
    },
    {
      "epoch": 0.02162615394349305,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001541802,
      "loss": 3.1422,
      "step": 229100
    },
    {
      "epoch": 0.021644797179651235,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001541782,
      "loss": 3.1063,
      "step": 229110
    },
    {
      "epoch": 0.021663440415809417,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015417620000000002,
      "loss": 3.0525,
      "step": 229120
    },
    {
      "epoch": 0.021682083651967603,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001541742,
      "loss": 3.0958,
      "step": 229130
    },
    {
      "epoch": 0.021700726888125785,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015417220000000001,
      "loss": 3.0685,
      "step": 229140
    },
    {
      "epoch": 0.02171937012428397,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001541702,
      "loss": 3.1849,
      "step": 229150
    },
    {
      "epoch": 0.021738013360442152,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001541682,
      "loss": 3.1851,
      "step": 229160
    },
    {
      "epoch": 0.021756656596600334,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015416620000000003,
      "loss": 2.912,
      "step": 229170
    },
    {
      "epoch": 0.02177529983275852,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001541642,
      "loss": 3.1475,
      "step": 229180
    },
    {
      "epoch": 0.021793943068916702,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015416220000000002,
      "loss": 3.0977,
      "step": 229190
    },
    {
      "epoch": 0.021812586305074887,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001541602,
      "loss": 3.1798,
      "step": 229200
    },
    {
      "epoch": 0.02183122954123307,
      "grad_norm": 0.75,
      "learning_rate": 0.0001541582,
      "loss": 3.1142,
      "step": 229210
    },
    {
      "epoch": 0.021849872777391255,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001541562,
      "loss": 3.1227,
      "step": 229220
    },
    {
      "epoch": 0.021868516013549437,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001541542,
      "loss": 3.101,
      "step": 229230
    },
    {
      "epoch": 0.021887159249707622,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015415220000000002,
      "loss": 3.1028,
      "step": 229240
    },
    {
      "epoch": 0.021905802485865804,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001541502,
      "loss": 3.0725,
      "step": 229250
    },
    {
      "epoch": 0.02192444572202399,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015414820000000002,
      "loss": 3.0046,
      "step": 229260
    },
    {
      "epoch": 0.021943088958182172,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001541462,
      "loss": 3.1159,
      "step": 229270
    },
    {
      "epoch": 0.021961732194340357,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001541442,
      "loss": 3.078,
      "step": 229280
    },
    {
      "epoch": 0.02198037543049854,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001541422,
      "loss": 3.2289,
      "step": 229290
    },
    {
      "epoch": 0.021999018666656725,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001541402,
      "loss": 3.1915,
      "step": 229300
    },
    {
      "epoch": 0.022017661902814907,
      "grad_norm": 0.75,
      "learning_rate": 0.00015413820000000002,
      "loss": 3.0689,
      "step": 229310
    },
    {
      "epoch": 0.022036305138973093,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001541362,
      "loss": 3.1149,
      "step": 229320
    },
    {
      "epoch": 0.022054948375131275,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001541342,
      "loss": 3.2066,
      "step": 229330
    },
    {
      "epoch": 0.022073591611289457,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001541322,
      "loss": 3.0455,
      "step": 229340
    },
    {
      "epoch": 0.022092234847447642,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001541302,
      "loss": 3.0298,
      "step": 229350
    },
    {
      "epoch": 0.022110878083605824,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001541282,
      "loss": 3.1543,
      "step": 229360
    },
    {
      "epoch": 0.02212952131976401,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001541262,
      "loss": 3.1437,
      "step": 229370
    },
    {
      "epoch": 0.02214816455592219,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015412420000000002,
      "loss": 3.0729,
      "step": 229380
    },
    {
      "epoch": 0.022166807792080377,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001541222,
      "loss": 3.1145,
      "step": 229390
    },
    {
      "epoch": 0.02218545102823856,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001541202,
      "loss": 2.923,
      "step": 229400
    },
    {
      "epoch": 0.022204094264396745,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001541182,
      "loss": 3.1873,
      "step": 229410
    },
    {
      "epoch": 0.022222737500554927,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001541162,
      "loss": 3.3026,
      "step": 229420
    },
    {
      "epoch": 0.022241380736713112,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001541142,
      "loss": 3.1317,
      "step": 229430
    },
    {
      "epoch": 0.022260023972871294,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015411220000000002,
      "loss": 3.1841,
      "step": 229440
    },
    {
      "epoch": 0.02227866720902948,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015411020000000001,
      "loss": 3.1365,
      "step": 229450
    },
    {
      "epoch": 0.022297310445187662,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001541082,
      "loss": 3.1885,
      "step": 229460
    },
    {
      "epoch": 0.022315953681345847,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001541062,
      "loss": 3.1686,
      "step": 229470
    },
    {
      "epoch": 0.02233459691750403,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001541042,
      "loss": 3.2353,
      "step": 229480
    },
    {
      "epoch": 0.022353240153662215,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001541022,
      "loss": 3.0459,
      "step": 229490
    },
    {
      "epoch": 0.022371883389820397,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015410020000000002,
      "loss": 3.0911,
      "step": 229500
    },
    {
      "epoch": 0.02239052662597858,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015409820000000001,
      "loss": 3.2048,
      "step": 229510
    },
    {
      "epoch": 0.022409169862136764,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001540962,
      "loss": 3.1198,
      "step": 229520
    },
    {
      "epoch": 0.022427813098294946,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001540942,
      "loss": 3.1864,
      "step": 229530
    },
    {
      "epoch": 0.022446456334453132,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001540922,
      "loss": 3.1564,
      "step": 229540
    },
    {
      "epoch": 0.022465099570611314,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001540902,
      "loss": 3.0722,
      "step": 229550
    },
    {
      "epoch": 0.0224837428067695,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015408820000000002,
      "loss": 3.0811,
      "step": 229560
    },
    {
      "epoch": 0.02250238604292768,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00015408620000000002,
      "loss": 3.0428,
      "step": 229570
    },
    {
      "epoch": 0.022521029279085867,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001540842,
      "loss": 3.1913,
      "step": 229580
    },
    {
      "epoch": 0.02253967251524405,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001540822,
      "loss": 3.1342,
      "step": 229590
    },
    {
      "epoch": 0.022558315751402234,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001540802,
      "loss": 3.1755,
      "step": 229600
    },
    {
      "epoch": 0.022576958987560417,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015407820000000003,
      "loss": 3.2386,
      "step": 229610
    },
    {
      "epoch": 0.022595602223718602,
      "grad_norm": 0.75,
      "learning_rate": 0.0001540762,
      "loss": 3.2245,
      "step": 229620
    },
    {
      "epoch": 0.022614245459876784,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015407420000000002,
      "loss": 3.1855,
      "step": 229630
    },
    {
      "epoch": 0.02263288869603497,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001540722,
      "loss": 3.1813,
      "step": 229640
    },
    {
      "epoch": 0.02265153193219315,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001540702,
      "loss": 3.2662,
      "step": 229650
    },
    {
      "epoch": 0.022670175168351337,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015406820000000003,
      "loss": 3.1753,
      "step": 229660
    },
    {
      "epoch": 0.02268881840450952,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001540662,
      "loss": 3.1171,
      "step": 229670
    },
    {
      "epoch": 0.0227074616406677,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015406420000000002,
      "loss": 3.1099,
      "step": 229680
    },
    {
      "epoch": 0.022726104876825887,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001540622,
      "loss": 3.2919,
      "step": 229690
    },
    {
      "epoch": 0.02274474811298407,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001540602,
      "loss": 3.1407,
      "step": 229700
    },
    {
      "epoch": 0.022763391349142254,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001540582,
      "loss": 3.2135,
      "step": 229710
    },
    {
      "epoch": 0.022782034585300436,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001540562,
      "loss": 3.2345,
      "step": 229720
    },
    {
      "epoch": 0.02280067782145862,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015405420000000003,
      "loss": 3.1031,
      "step": 229730
    },
    {
      "epoch": 0.022819321057616804,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001540522,
      "loss": 2.9518,
      "step": 229740
    },
    {
      "epoch": 0.02283796429377499,
      "grad_norm": 0.625,
      "learning_rate": 0.00015405020000000002,
      "loss": 3.0776,
      "step": 229750
    },
    {
      "epoch": 0.02285660752993317,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001540482,
      "loss": 3.1981,
      "step": 229760
    },
    {
      "epoch": 0.022875250766091357,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001540462,
      "loss": 3.1953,
      "step": 229770
    },
    {
      "epoch": 0.02289389400224954,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001540442,
      "loss": 3.1822,
      "step": 229780
    },
    {
      "epoch": 0.022912537238407724,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001540422,
      "loss": 3.1917,
      "step": 229790
    },
    {
      "epoch": 0.022931180474565906,
      "grad_norm": 0.75,
      "learning_rate": 0.00015404020000000002,
      "loss": 3.2362,
      "step": 229800
    },
    {
      "epoch": 0.022949823710724092,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001540382,
      "loss": 3.0441,
      "step": 229810
    },
    {
      "epoch": 0.022968466946882274,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015403620000000001,
      "loss": 3.1942,
      "step": 229820
    },
    {
      "epoch": 0.02298711018304046,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001540342,
      "loss": 3.146,
      "step": 229830
    },
    {
      "epoch": 0.02300575341919864,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001540322,
      "loss": 3.2263,
      "step": 229840
    },
    {
      "epoch": 0.023024396655356823,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001540302,
      "loss": 3.1941,
      "step": 229850
    },
    {
      "epoch": 0.02304303989151501,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001540282,
      "loss": 2.9814,
      "step": 229860
    },
    {
      "epoch": 0.02306168312767319,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015402620000000002,
      "loss": 3.1851,
      "step": 229870
    },
    {
      "epoch": 0.023080326363831376,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015402420000000002,
      "loss": 3.0927,
      "step": 229880
    },
    {
      "epoch": 0.02309896959998956,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001540222,
      "loss": 3.1071,
      "step": 229890
    },
    {
      "epoch": 0.023117612836147744,
      "grad_norm": 0.75,
      "learning_rate": 0.0001540202,
      "loss": 3.1993,
      "step": 229900
    },
    {
      "epoch": 0.023136256072305926,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001540182,
      "loss": 3.218,
      "step": 229910
    },
    {
      "epoch": 0.02315489930846411,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001540162,
      "loss": 3.1616,
      "step": 229920
    },
    {
      "epoch": 0.023173542544622294,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001540142,
      "loss": 3.1366,
      "step": 229930
    },
    {
      "epoch": 0.02319218578078048,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015401220000000002,
      "loss": 3.2436,
      "step": 229940
    },
    {
      "epoch": 0.02321082901693866,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001540102,
      "loss": 3.1718,
      "step": 229950
    },
    {
      "epoch": 0.023229472253096847,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001540082,
      "loss": 3.0665,
      "step": 229960
    },
    {
      "epoch": 0.02324811548925503,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001540062,
      "loss": 2.9664,
      "step": 229970
    },
    {
      "epoch": 0.023266758725413214,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001540042,
      "loss": 2.9953,
      "step": 229980
    },
    {
      "epoch": 0.023285401961571396,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001540022,
      "loss": 3.1317,
      "step": 229990
    },
    {
      "epoch": 0.02330404519772958,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015400020000000002,
      "loss": 3.107,
      "step": 230000
    },
    {
      "epoch": 0.023322688433887764,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001539982,
      "loss": 3.193,
      "step": 230010
    },
    {
      "epoch": 0.023341331670045946,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001539962,
      "loss": 3.1935,
      "step": 230020
    },
    {
      "epoch": 0.02335997490620413,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001539942,
      "loss": 3.1806,
      "step": 230030
    },
    {
      "epoch": 0.023378618142362313,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001539922,
      "loss": 3.0962,
      "step": 230040
    },
    {
      "epoch": 0.0233972613785205,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015399020000000002,
      "loss": 3.1518,
      "step": 230050
    },
    {
      "epoch": 0.02341590461467868,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015398820000000002,
      "loss": 3.2247,
      "step": 230060
    },
    {
      "epoch": 0.023434547850836866,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001539862,
      "loss": 3.1858,
      "step": 230070
    },
    {
      "epoch": 0.02345319108699505,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001539842,
      "loss": 3.1091,
      "step": 230080
    },
    {
      "epoch": 0.023471834323153234,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001539822,
      "loss": 3.2649,
      "step": 230090
    },
    {
      "epoch": 0.023490477559311416,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001539802,
      "loss": 3.1523,
      "step": 230100
    },
    {
      "epoch": 0.0235091207954696,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001539782,
      "loss": 3.2327,
      "step": 230110
    },
    {
      "epoch": 0.023527764031627783,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015397620000000002,
      "loss": 3.1155,
      "step": 230120
    },
    {
      "epoch": 0.02354640726778597,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015397420000000001,
      "loss": 3.2161,
      "step": 230130
    },
    {
      "epoch": 0.02356505050394415,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001539722,
      "loss": 3.1339,
      "step": 230140
    },
    {
      "epoch": 0.023583693740102336,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001539702,
      "loss": 3.1273,
      "step": 230150
    },
    {
      "epoch": 0.02360233697626052,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001539682,
      "loss": 3.1431,
      "step": 230160
    },
    {
      "epoch": 0.023620980212418704,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015396620000000002,
      "loss": 3.1757,
      "step": 230170
    },
    {
      "epoch": 0.023639623448576886,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001539642,
      "loss": 3.0946,
      "step": 230180
    },
    {
      "epoch": 0.02365826668473507,
      "grad_norm": 0.625,
      "learning_rate": 0.00015396220000000001,
      "loss": 3.0502,
      "step": 230190
    },
    {
      "epoch": 0.023676909920893253,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001539602,
      "loss": 3.275,
      "step": 230200
    },
    {
      "epoch": 0.023695553157051435,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001539582,
      "loss": 3.1746,
      "step": 230210
    },
    {
      "epoch": 0.02371419639320962,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015395620000000003,
      "loss": 3.0327,
      "step": 230220
    },
    {
      "epoch": 0.023732839629367803,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001539542,
      "loss": 3.126,
      "step": 230230
    },
    {
      "epoch": 0.02375148286552599,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015395220000000002,
      "loss": 3.0853,
      "step": 230240
    },
    {
      "epoch": 0.02377012610168417,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001539502,
      "loss": 3.1886,
      "step": 230250
    },
    {
      "epoch": 0.023788769337842356,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001539482,
      "loss": 3.0736,
      "step": 230260
    },
    {
      "epoch": 0.023807412574000538,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001539462,
      "loss": 3.1365,
      "step": 230270
    },
    {
      "epoch": 0.023826055810158724,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001539442,
      "loss": 3.1234,
      "step": 230280
    },
    {
      "epoch": 0.023844699046316906,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015394220000000003,
      "loss": 3.1278,
      "step": 230290
    },
    {
      "epoch": 0.02386334228247509,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001539402,
      "loss": 3.0253,
      "step": 230300
    },
    {
      "epoch": 0.023881985518633273,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015393820000000002,
      "loss": 3.0967,
      "step": 230310
    },
    {
      "epoch": 0.02390062875479146,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015393619999999998,
      "loss": 3.1345,
      "step": 230320
    },
    {
      "epoch": 0.02391927199094964,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001539342,
      "loss": 3.1312,
      "step": 230330
    },
    {
      "epoch": 0.023937915227107826,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001539322,
      "loss": 3.0837,
      "step": 230340
    },
    {
      "epoch": 0.023956558463266008,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001539302,
      "loss": 3.0574,
      "step": 230350
    },
    {
      "epoch": 0.023975201699424194,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015392820000000002,
      "loss": 3.1559,
      "step": 230360
    },
    {
      "epoch": 0.023993844935582376,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001539262,
      "loss": 3.1441,
      "step": 230370
    },
    {
      "epoch": 0.024012488171740558,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001539242,
      "loss": 3.1273,
      "step": 230380
    },
    {
      "epoch": 0.024031131407898743,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001539222,
      "loss": 3.1102,
      "step": 230390
    },
    {
      "epoch": 0.024049774644056925,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001539202,
      "loss": 3.1622,
      "step": 230400
    },
    {
      "epoch": 0.02406841788021511,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001539182,
      "loss": 3.0819,
      "step": 230410
    },
    {
      "epoch": 0.024087061116373293,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001539162,
      "loss": 3.2227,
      "step": 230420
    },
    {
      "epoch": 0.02410570435253148,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015391420000000002,
      "loss": 3.264,
      "step": 230430
    },
    {
      "epoch": 0.02412434758868966,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001539122,
      "loss": 3.0423,
      "step": 230440
    },
    {
      "epoch": 0.024142990824847846,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001539102,
      "loss": 3.0536,
      "step": 230450
    },
    {
      "epoch": 0.024161634061006028,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001539082,
      "loss": 3.1715,
      "step": 230460
    },
    {
      "epoch": 0.024180277297164213,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001539062,
      "loss": 3.1328,
      "step": 230470
    },
    {
      "epoch": 0.024198920533322395,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001539042,
      "loss": 3.167,
      "step": 230480
    },
    {
      "epoch": 0.02421756376948058,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001539022,
      "loss": 3.237,
      "step": 230490
    },
    {
      "epoch": 0.024236207005638763,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015390020000000001,
      "loss": 3.1321,
      "step": 230500
    },
    {
      "epoch": 0.02425485024179695,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001538982,
      "loss": 3.0871,
      "step": 230510
    },
    {
      "epoch": 0.02427349347795513,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001538962,
      "loss": 3.0856,
      "step": 230520
    },
    {
      "epoch": 0.024292136714113316,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001538942,
      "loss": 3.2064,
      "step": 230530
    },
    {
      "epoch": 0.024310779950271498,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001538922,
      "loss": 3.1966,
      "step": 230540
    },
    {
      "epoch": 0.02432942318642968,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015389020000000002,
      "loss": 3.1739,
      "step": 230550
    },
    {
      "epoch": 0.024348066422587866,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015388820000000002,
      "loss": 3.0916,
      "step": 230560
    },
    {
      "epoch": 0.024366709658746048,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001538862,
      "loss": 3.1569,
      "step": 230570
    },
    {
      "epoch": 0.024385352894904233,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001538842,
      "loss": 3.0881,
      "step": 230580
    },
    {
      "epoch": 0.024403996131062415,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001538822,
      "loss": 3.0572,
      "step": 230590
    },
    {
      "epoch": 0.0244226393672206,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001538802,
      "loss": 3.0427,
      "step": 230600
    },
    {
      "epoch": 0.024441282603378783,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015387820000000002,
      "loss": 3.1694,
      "step": 230610
    },
    {
      "epoch": 0.024459925839536968,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015387620000000002,
      "loss": 2.9307,
      "step": 230620
    },
    {
      "epoch": 0.02447856907569515,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001538742,
      "loss": 3.1815,
      "step": 230630
    },
    {
      "epoch": 0.024497212311853336,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001538722,
      "loss": 3.2117,
      "step": 230640
    },
    {
      "epoch": 0.024515855548011518,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001538702,
      "loss": 3.2801,
      "step": 230650
    },
    {
      "epoch": 0.024534498784169703,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015386820000000003,
      "loss": 3.0169,
      "step": 230660
    },
    {
      "epoch": 0.024553142020327885,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001538662,
      "loss": 3.1814,
      "step": 230670
    },
    {
      "epoch": 0.02457178525648607,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015386420000000002,
      "loss": 3.0396,
      "step": 230680
    },
    {
      "epoch": 0.024590428492644253,
      "grad_norm": 0.75,
      "learning_rate": 0.0001538622,
      "loss": 3.2718,
      "step": 230690
    },
    {
      "epoch": 0.024609071728802438,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001538602,
      "loss": 3.1431,
      "step": 230700
    },
    {
      "epoch": 0.02462771496496062,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001538582,
      "loss": 3.2875,
      "step": 230710
    },
    {
      "epoch": 0.024646358201118802,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001538562,
      "loss": 3.0882,
      "step": 230720
    },
    {
      "epoch": 0.024665001437276988,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015385420000000002,
      "loss": 3.2984,
      "step": 230730
    },
    {
      "epoch": 0.02468364467343517,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001538522,
      "loss": 3.1705,
      "step": 230740
    },
    {
      "epoch": 0.024702287909593355,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001538502,
      "loss": 3.054,
      "step": 230750
    },
    {
      "epoch": 0.024720931145751537,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001538482,
      "loss": 3.155,
      "step": 230760
    },
    {
      "epoch": 0.024739574381909723,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001538462,
      "loss": 3.1377,
      "step": 230770
    },
    {
      "epoch": 0.024758217618067905,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015384420000000003,
      "loss": 3.0451,
      "step": 230780
    },
    {
      "epoch": 0.02477686085422609,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001538422,
      "loss": 3.2591,
      "step": 230790
    },
    {
      "epoch": 0.024795504090384272,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015384020000000002,
      "loss": 3.2216,
      "step": 230800
    },
    {
      "epoch": 0.024814147326542458,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001538382,
      "loss": 3.0382,
      "step": 230810
    },
    {
      "epoch": 0.02483279056270064,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001538362,
      "loss": 3.1974,
      "step": 230820
    },
    {
      "epoch": 0.024851433798858825,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001538342,
      "loss": 3.1092,
      "step": 230830
    },
    {
      "epoch": 0.024870077035017008,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001538322,
      "loss": 3.1342,
      "step": 230840
    },
    {
      "epoch": 0.024888720271175193,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015383020000000002,
      "loss": 3.2403,
      "step": 230850
    },
    {
      "epoch": 0.024907363507333375,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001538282,
      "loss": 3.2181,
      "step": 230860
    },
    {
      "epoch": 0.02492600674349156,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015382620000000001,
      "loss": 3.2568,
      "step": 230870
    },
    {
      "epoch": 0.024944649979649743,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001538242,
      "loss": 3.0823,
      "step": 230880
    },
    {
      "epoch": 0.024963293215807925,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001538222,
      "loss": 3.3206,
      "step": 230890
    },
    {
      "epoch": 0.02498193645196611,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001538202,
      "loss": 3.2407,
      "step": 230900
    },
    {
      "epoch": 0.025000579688124292,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001538182,
      "loss": 3.0233,
      "step": 230910
    },
    {
      "epoch": 0.025019222924282478,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015381620000000002,
      "loss": 3.1033,
      "step": 230920
    },
    {
      "epoch": 0.02503786616044066,
      "grad_norm": 0.75,
      "learning_rate": 0.0001538142,
      "loss": 3.0926,
      "step": 230930
    },
    {
      "epoch": 0.025056509396598845,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001538122,
      "loss": 3.0784,
      "step": 230940
    },
    {
      "epoch": 0.025075152632757027,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001538102,
      "loss": 3.1979,
      "step": 230950
    },
    {
      "epoch": 0.025093795868915213,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001538082,
      "loss": 3.1635,
      "step": 230960
    },
    {
      "epoch": 0.025112439105073395,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001538062,
      "loss": 3.2044,
      "step": 230970
    },
    {
      "epoch": 0.02513108234123158,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001538042,
      "loss": 3.1481,
      "step": 230980
    },
    {
      "epoch": 0.025149725577389762,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015380220000000002,
      "loss": 3.2313,
      "step": 230990
    },
    {
      "epoch": 0.025168368813547948,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001538002,
      "loss": 3.1246,
      "step": 231000
    },
    {
      "epoch": 0.02518701204970613,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001537982,
      "loss": 3.0985,
      "step": 231010
    },
    {
      "epoch": 0.025205655285864315,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001537962,
      "loss": 3.1565,
      "step": 231020
    },
    {
      "epoch": 0.025224298522022497,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001537942,
      "loss": 3.1111,
      "step": 231030
    },
    {
      "epoch": 0.025242941758180683,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015379220000000002,
      "loss": 3.2172,
      "step": 231040
    },
    {
      "epoch": 0.025261584994338865,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015379020000000002,
      "loss": 3.161,
      "step": 231050
    },
    {
      "epoch": 0.025280228230497047,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001537882,
      "loss": 3.0089,
      "step": 231060
    },
    {
      "epoch": 0.025298871466655232,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001537862,
      "loss": 3.1243,
      "step": 231070
    },
    {
      "epoch": 0.025317514702813414,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001537842,
      "loss": 3.1487,
      "step": 231080
    },
    {
      "epoch": 0.0253361579389716,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001537822,
      "loss": 3.1448,
      "step": 231090
    },
    {
      "epoch": 0.025354801175129782,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001537802,
      "loss": 3.0973,
      "step": 231100
    },
    {
      "epoch": 0.025373444411287967,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015377820000000002,
      "loss": 2.9716,
      "step": 231110
    },
    {
      "epoch": 0.02539208764744615,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001537762,
      "loss": 3.1634,
      "step": 231120
    },
    {
      "epoch": 0.025410730883604335,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001537742,
      "loss": 3.1881,
      "step": 231130
    },
    {
      "epoch": 0.025429374119762517,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001537722,
      "loss": 3.2091,
      "step": 231140
    },
    {
      "epoch": 0.025448017355920702,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001537702,
      "loss": 3.0258,
      "step": 231150
    },
    {
      "epoch": 0.025466660592078885,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001537682,
      "loss": 3.1031,
      "step": 231160
    },
    {
      "epoch": 0.02548530382823707,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015376620000000002,
      "loss": 3.0734,
      "step": 231170
    },
    {
      "epoch": 0.025503947064395252,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015376420000000001,
      "loss": 3.0974,
      "step": 231180
    },
    {
      "epoch": 0.025522590300553438,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001537622,
      "loss": 3.1522,
      "step": 231190
    },
    {
      "epoch": 0.02554123353671162,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001537602,
      "loss": 3.0726,
      "step": 231200
    },
    {
      "epoch": 0.025559876772869805,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001537582,
      "loss": 3.1619,
      "step": 231210
    },
    {
      "epoch": 0.025578520009027987,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015375620000000002,
      "loss": 3.0643,
      "step": 231220
    },
    {
      "epoch": 0.02559716324518617,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001537542,
      "loss": 3.1254,
      "step": 231230
    },
    {
      "epoch": 0.025615806481344355,
      "grad_norm": 0.75,
      "learning_rate": 0.00015375220000000002,
      "loss": 3.2631,
      "step": 231240
    },
    {
      "epoch": 0.025634449717502537,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001537502,
      "loss": 3.1441,
      "step": 231250
    },
    {
      "epoch": 0.025653092953660722,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001537482,
      "loss": 3.1242,
      "step": 231260
    },
    {
      "epoch": 0.025671736189818904,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015374620000000003,
      "loss": 3.1024,
      "step": 231270
    },
    {
      "epoch": 0.02569037942597709,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001537442,
      "loss": 3.1554,
      "step": 231280
    },
    {
      "epoch": 0.025709022662135272,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015374220000000002,
      "loss": 3.0915,
      "step": 231290
    },
    {
      "epoch": 0.025727665898293457,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001537402,
      "loss": 3.2676,
      "step": 231300
    },
    {
      "epoch": 0.02574630913445164,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001537382,
      "loss": 3.1805,
      "step": 231310
    },
    {
      "epoch": 0.025764952370609825,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001537362,
      "loss": 3.0789,
      "step": 231320
    },
    {
      "epoch": 0.025783595606768007,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001537342,
      "loss": 3.1139,
      "step": 231330
    },
    {
      "epoch": 0.025802238842926192,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015373220000000003,
      "loss": 3.15,
      "step": 231340
    },
    {
      "epoch": 0.025820882079084374,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001537302,
      "loss": 3.2129,
      "step": 231350
    },
    {
      "epoch": 0.02583952531524256,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015372820000000002,
      "loss": 3.1636,
      "step": 231360
    },
    {
      "epoch": 0.025858168551400742,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015372619999999999,
      "loss": 3.1814,
      "step": 231370
    },
    {
      "epoch": 0.025876811787558927,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001537242,
      "loss": 3.1322,
      "step": 231380
    },
    {
      "epoch": 0.02589545502371711,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001537222,
      "loss": 3.1382,
      "step": 231390
    },
    {
      "epoch": 0.02591409825987529,
      "grad_norm": 0.75,
      "learning_rate": 0.0001537202,
      "loss": 3.2458,
      "step": 231400
    },
    {
      "epoch": 0.025932741496033477,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015371820000000002,
      "loss": 3.1154,
      "step": 231410
    },
    {
      "epoch": 0.02595138473219166,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001537162,
      "loss": 3.1645,
      "step": 231420
    },
    {
      "epoch": 0.025970027968349844,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001537142,
      "loss": 3.033,
      "step": 231430
    },
    {
      "epoch": 0.025988671204508026,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001537122,
      "loss": 3.1428,
      "step": 231440
    },
    {
      "epoch": 0.026007314440666212,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001537102,
      "loss": 3.1641,
      "step": 231450
    },
    {
      "epoch": 0.026025957676824394,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001537082,
      "loss": 3.2407,
      "step": 231460
    },
    {
      "epoch": 0.02604460091298258,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001537062,
      "loss": 3.2164,
      "step": 231470
    },
    {
      "epoch": 0.02606324414914076,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015370420000000002,
      "loss": 3.2267,
      "step": 231480
    },
    {
      "epoch": 0.026081887385298947,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015370220000000001,
      "loss": 3.1149,
      "step": 231490
    },
    {
      "epoch": 0.02610053062145713,
      "grad_norm": 0.625,
      "learning_rate": 0.0001537002,
      "loss": 3.1855,
      "step": 231500
    },
    {
      "epoch": 0.026119173857615315,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001536982,
      "loss": 3.1044,
      "step": 231510
    },
    {
      "epoch": 0.026137817093773497,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001536962,
      "loss": 3.1499,
      "step": 231520
    },
    {
      "epoch": 0.026156460329931682,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015369420000000002,
      "loss": 3.1947,
      "step": 231530
    },
    {
      "epoch": 0.026175103566089864,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001536922,
      "loss": 3.165,
      "step": 231540
    },
    {
      "epoch": 0.02619374680224805,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015369020000000001,
      "loss": 3.2087,
      "step": 231550
    },
    {
      "epoch": 0.02621239003840623,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001536882,
      "loss": 3.1227,
      "step": 231560
    },
    {
      "epoch": 0.026231033274564414,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001536862,
      "loss": 3.2334,
      "step": 231570
    },
    {
      "epoch": 0.0262496765107226,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001536842,
      "loss": 3.0547,
      "step": 231580
    },
    {
      "epoch": 0.02626831974688078,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001536822,
      "loss": 3.2006,
      "step": 231590
    },
    {
      "epoch": 0.026286962983038967,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015368020000000002,
      "loss": 3.0124,
      "step": 231600
    },
    {
      "epoch": 0.02630560621919715,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015367820000000002,
      "loss": 3.042,
      "step": 231610
    },
    {
      "epoch": 0.026324249455355334,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001536762,
      "loss": 3.11,
      "step": 231620
    },
    {
      "epoch": 0.026342892691513516,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001536742,
      "loss": 3.0974,
      "step": 231630
    },
    {
      "epoch": 0.026361535927671702,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001536722,
      "loss": 3.1608,
      "step": 231640
    },
    {
      "epoch": 0.026380179163829884,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001536702,
      "loss": 2.9736,
      "step": 231650
    },
    {
      "epoch": 0.02639882239998807,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015366820000000002,
      "loss": 3.0899,
      "step": 231660
    },
    {
      "epoch": 0.02641746563614625,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015366620000000002,
      "loss": 3.0602,
      "step": 231670
    },
    {
      "epoch": 0.026436108872304437,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001536642,
      "loss": 3.1968,
      "step": 231680
    },
    {
      "epoch": 0.02645475210846262,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001536622,
      "loss": 3.1384,
      "step": 231690
    },
    {
      "epoch": 0.026473395344620804,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001536602,
      "loss": 3.062,
      "step": 231700
    },
    {
      "epoch": 0.026492038580778986,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001536582,
      "loss": 3.2331,
      "step": 231710
    },
    {
      "epoch": 0.026510681816937172,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001536562,
      "loss": 3.0391,
      "step": 231720
    },
    {
      "epoch": 0.026529325053095354,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015365420000000002,
      "loss": 3.0565,
      "step": 231730
    },
    {
      "epoch": 0.026547968289253536,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001536522,
      "loss": 3.0626,
      "step": 231740
    },
    {
      "epoch": 0.02656661152541172,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001536502,
      "loss": 3.2049,
      "step": 231750
    },
    {
      "epoch": 0.026585254761569903,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001536482,
      "loss": 2.9882,
      "step": 231760
    },
    {
      "epoch": 0.02660389799772809,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001536462,
      "loss": 3.0475,
      "step": 231770
    },
    {
      "epoch": 0.02662254123388627,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015364420000000002,
      "loss": 3.2054,
      "step": 231780
    },
    {
      "epoch": 0.026641184470044457,
      "grad_norm": 0.75,
      "learning_rate": 0.0001536422,
      "loss": 3.0965,
      "step": 231790
    },
    {
      "epoch": 0.02665982770620264,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001536402,
      "loss": 3.1593,
      "step": 231800
    },
    {
      "epoch": 0.026678470942360824,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001536382,
      "loss": 3.0288,
      "step": 231810
    },
    {
      "epoch": 0.026697114178519006,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001536362,
      "loss": 3.0913,
      "step": 231820
    },
    {
      "epoch": 0.02671575741467719,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015363420000000003,
      "loss": 3.1736,
      "step": 231830
    },
    {
      "epoch": 0.026734400650835374,
      "grad_norm": 0.75,
      "learning_rate": 0.0001536322,
      "loss": 3.1604,
      "step": 231840
    },
    {
      "epoch": 0.02675304388699356,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015363020000000002,
      "loss": 3.1388,
      "step": 231850
    },
    {
      "epoch": 0.02677168712315174,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001536282,
      "loss": 3.1102,
      "step": 231860
    },
    {
      "epoch": 0.026790330359309927,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001536262,
      "loss": 3.1736,
      "step": 231870
    },
    {
      "epoch": 0.02680897359546811,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001536242,
      "loss": 3.0739,
      "step": 231880
    },
    {
      "epoch": 0.026827616831626294,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001536222,
      "loss": 3.1758,
      "step": 231890
    },
    {
      "epoch": 0.026846260067784476,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015362020000000002,
      "loss": 3.2451,
      "step": 231900
    },
    {
      "epoch": 0.026864903303942658,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001536182,
      "loss": 3.235,
      "step": 231910
    },
    {
      "epoch": 0.026883546540100844,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015361620000000002,
      "loss": 3.1223,
      "step": 231920
    },
    {
      "epoch": 0.026902189776259026,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015361419999999998,
      "loss": 3.2599,
      "step": 231930
    },
    {
      "epoch": 0.02692083301241721,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001536122,
      "loss": 3.1143,
      "step": 231940
    },
    {
      "epoch": 0.026939476248575393,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001536102,
      "loss": 3.1497,
      "step": 231950
    },
    {
      "epoch": 0.02695811948473358,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001536082,
      "loss": 3.2731,
      "step": 231960
    },
    {
      "epoch": 0.02697676272089176,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015360620000000002,
      "loss": 3.182,
      "step": 231970
    },
    {
      "epoch": 0.026995405957049946,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001536042,
      "loss": 3.0652,
      "step": 231980
    },
    {
      "epoch": 0.02701404919320813,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001536022,
      "loss": 3.1191,
      "step": 231990
    },
    {
      "epoch": 0.027032692429366314,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001536002,
      "loss": 3.0933,
      "step": 232000
    },
    {
      "epoch": 0.027051335665524496,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001535982,
      "loss": 3.1228,
      "step": 232010
    },
    {
      "epoch": 0.02706997890168268,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015359620000000003,
      "loss": 3.066,
      "step": 232020
    },
    {
      "epoch": 0.027088622137840863,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001535942,
      "loss": 3.2414,
      "step": 232030
    },
    {
      "epoch": 0.02710726537399905,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015359220000000002,
      "loss": 3.1172,
      "step": 232040
    },
    {
      "epoch": 0.02712590861015723,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001535902,
      "loss": 3.1166,
      "step": 232050
    },
    {
      "epoch": 0.027144551846315416,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001535882,
      "loss": 3.141,
      "step": 232060
    },
    {
      "epoch": 0.0271631950824736,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001535862,
      "loss": 3.1364,
      "step": 232070
    },
    {
      "epoch": 0.02718183831863178,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001535842,
      "loss": 3.1683,
      "step": 232080
    },
    {
      "epoch": 0.027200481554789966,
      "grad_norm": 0.75,
      "learning_rate": 0.00015358220000000002,
      "loss": 3.0581,
      "step": 232090
    },
    {
      "epoch": 0.027219124790948148,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001535802,
      "loss": 3.0337,
      "step": 232100
    },
    {
      "epoch": 0.027237768027106334,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001535782,
      "loss": 3.1142,
      "step": 232110
    },
    {
      "epoch": 0.027256411263264516,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001535762,
      "loss": 3.1789,
      "step": 232120
    },
    {
      "epoch": 0.0272750544994227,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001535742,
      "loss": 3.0007,
      "step": 232130
    },
    {
      "epoch": 0.027293697735580883,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001535722,
      "loss": 3.0935,
      "step": 232140
    },
    {
      "epoch": 0.02731234097173907,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001535702,
      "loss": 3.1825,
      "step": 232150
    },
    {
      "epoch": 0.02733098420789725,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015356820000000002,
      "loss": 3.1078,
      "step": 232160
    },
    {
      "epoch": 0.027349627444055436,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015356620000000001,
      "loss": 3.2981,
      "step": 232170
    },
    {
      "epoch": 0.027368270680213618,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001535642,
      "loss": 3.2736,
      "step": 232180
    },
    {
      "epoch": 0.027386913916371804,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001535622,
      "loss": 3.2358,
      "step": 232190
    },
    {
      "epoch": 0.027405557152529986,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001535602,
      "loss": 3.0275,
      "step": 232200
    },
    {
      "epoch": 0.02742420038868817,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001535582,
      "loss": 3.046,
      "step": 232210
    },
    {
      "epoch": 0.027442843624846353,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015355620000000002,
      "loss": 3.0854,
      "step": 232220
    },
    {
      "epoch": 0.02746148686100454,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015355420000000001,
      "loss": 3.1388,
      "step": 232230
    },
    {
      "epoch": 0.02748013009716272,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001535522,
      "loss": 3.1272,
      "step": 232240
    },
    {
      "epoch": 0.027498773333320903,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001535502,
      "loss": 3.051,
      "step": 232250
    },
    {
      "epoch": 0.02751741656947909,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001535482,
      "loss": 3.034,
      "step": 232260
    },
    {
      "epoch": 0.02753605980563727,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015354620000000002,
      "loss": 3.1939,
      "step": 232270
    },
    {
      "epoch": 0.027554703041795456,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001535442,
      "loss": 3.1974,
      "step": 232280
    },
    {
      "epoch": 0.027573346277953638,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015354220000000002,
      "loss": 3.1464,
      "step": 232290
    },
    {
      "epoch": 0.027591989514111823,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001535402,
      "loss": 3.0866,
      "step": 232300
    },
    {
      "epoch": 0.027610632750270005,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001535382,
      "loss": 3.0628,
      "step": 232310
    },
    {
      "epoch": 0.02762927598642819,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001535362,
      "loss": 3.0474,
      "step": 232320
    },
    {
      "epoch": 0.027647919222586373,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001535342,
      "loss": 3.2122,
      "step": 232330
    },
    {
      "epoch": 0.02766656245874456,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00015353220000000002,
      "loss": 3.1123,
      "step": 232340
    },
    {
      "epoch": 0.02768520569490274,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001535302,
      "loss": 3.1096,
      "step": 232350
    },
    {
      "epoch": 0.027703848931060926,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001535282,
      "loss": 3.0168,
      "step": 232360
    },
    {
      "epoch": 0.027722492167219108,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001535262,
      "loss": 3.1402,
      "step": 232370
    },
    {
      "epoch": 0.027741135403377293,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001535242,
      "loss": 3.1789,
      "step": 232380
    },
    {
      "epoch": 0.027759778639535476,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015352220000000003,
      "loss": 3.2309,
      "step": 232390
    },
    {
      "epoch": 0.02777842187569366,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001535202,
      "loss": 3.2011,
      "step": 232400
    },
    {
      "epoch": 0.027797065111851843,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015351820000000002,
      "loss": 3.0945,
      "step": 232410
    },
    {
      "epoch": 0.027815708348010025,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015351619999999999,
      "loss": 3.1553,
      "step": 232420
    },
    {
      "epoch": 0.02783435158416821,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001535142,
      "loss": 3.1846,
      "step": 232430
    },
    {
      "epoch": 0.027852994820326393,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001535122,
      "loss": 3.1463,
      "step": 232440
    },
    {
      "epoch": 0.027871638056484578,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001535102,
      "loss": 3.1047,
      "step": 232450
    },
    {
      "epoch": 0.02789028129264276,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015350820000000002,
      "loss": 3.1783,
      "step": 232460
    },
    {
      "epoch": 0.027908924528800946,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001535062,
      "loss": 3.2248,
      "step": 232470
    },
    {
      "epoch": 0.027927567764959128,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001535042,
      "loss": 3.1153,
      "step": 232480
    },
    {
      "epoch": 0.027946211001117313,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001535022,
      "loss": 3.048,
      "step": 232490
    },
    {
      "epoch": 0.027964854237275495,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001535002,
      "loss": 3.0815,
      "step": 232500
    },
    {
      "epoch": 0.02798349747343368,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015349820000000003,
      "loss": 3.2036,
      "step": 232510
    },
    {
      "epoch": 0.028002140709591863,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001534962,
      "loss": 3.15,
      "step": 232520
    },
    {
      "epoch": 0.028020783945750048,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015349420000000002,
      "loss": 3.0833,
      "step": 232530
    },
    {
      "epoch": 0.02803942718190823,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001534922,
      "loss": 3.1751,
      "step": 232540
    },
    {
      "epoch": 0.028058070418066416,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001534902,
      "loss": 3.1381,
      "step": 232550
    },
    {
      "epoch": 0.028076713654224598,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001534882,
      "loss": 3.2235,
      "step": 232560
    },
    {
      "epoch": 0.028095356890382783,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001534862,
      "loss": 3.0159,
      "step": 232570
    },
    {
      "epoch": 0.028114000126540965,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015348420000000002,
      "loss": 3.1849,
      "step": 232580
    },
    {
      "epoch": 0.02813264336269915,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001534822,
      "loss": 3.206,
      "step": 232590
    },
    {
      "epoch": 0.028151286598857333,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015348020000000001,
      "loss": 3.0489,
      "step": 232600
    },
    {
      "epoch": 0.028169929835015515,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001534782,
      "loss": 3.2124,
      "step": 232610
    },
    {
      "epoch": 0.0281885730711737,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001534762,
      "loss": 3.1416,
      "step": 232620
    },
    {
      "epoch": 0.028207216307331882,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001534742,
      "loss": 2.9797,
      "step": 232630
    },
    {
      "epoch": 0.028225859543490068,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001534722,
      "loss": 3.0714,
      "step": 232640
    },
    {
      "epoch": 0.02824450277964825,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015347020000000002,
      "loss": 3.1273,
      "step": 232650
    },
    {
      "epoch": 0.028263146015806435,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015346820000000002,
      "loss": 3.1838,
      "step": 232660
    },
    {
      "epoch": 0.028281789251964617,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001534662,
      "loss": 3.1757,
      "step": 232670
    },
    {
      "epoch": 0.028300432488122803,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001534642,
      "loss": 3.0909,
      "step": 232680
    },
    {
      "epoch": 0.028319075724280985,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001534622,
      "loss": 3.1288,
      "step": 232690
    },
    {
      "epoch": 0.02833771896043917,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001534602,
      "loss": 3.2304,
      "step": 232700
    },
    {
      "epoch": 0.028356362196597353,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001534582,
      "loss": 3.1077,
      "step": 232710
    },
    {
      "epoch": 0.028375005432755538,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015345620000000002,
      "loss": 3.1207,
      "step": 232720
    },
    {
      "epoch": 0.02839364866891372,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001534542,
      "loss": 3.07,
      "step": 232730
    },
    {
      "epoch": 0.028412291905071906,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001534522,
      "loss": 3.1225,
      "step": 232740
    },
    {
      "epoch": 0.028430935141230088,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001534502,
      "loss": 3.0552,
      "step": 232750
    },
    {
      "epoch": 0.028449578377388273,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001534482,
      "loss": 3.1614,
      "step": 232760
    },
    {
      "epoch": 0.028468221613546455,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001534462,
      "loss": 3.1412,
      "step": 232770
    },
    {
      "epoch": 0.028486864849704637,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015344420000000002,
      "loss": 3.0903,
      "step": 232780
    },
    {
      "epoch": 0.028505508085862823,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001534422,
      "loss": 3.2785,
      "step": 232790
    },
    {
      "epoch": 0.028524151322021005,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001534402,
      "loss": 3.2624,
      "step": 232800
    },
    {
      "epoch": 0.02854279455817919,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001534382,
      "loss": 3.0817,
      "step": 232810
    },
    {
      "epoch": 0.028561437794337372,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001534362,
      "loss": 3.1092,
      "step": 232820
    },
    {
      "epoch": 0.028580081030495558,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015343420000000002,
      "loss": 3.2612,
      "step": 232830
    },
    {
      "epoch": 0.02859872426665374,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001534322,
      "loss": 3.1851,
      "step": 232840
    },
    {
      "epoch": 0.028617367502811925,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015343020000000001,
      "loss": 3.1196,
      "step": 232850
    },
    {
      "epoch": 0.028636010738970107,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001534282,
      "loss": 3.1592,
      "step": 232860
    },
    {
      "epoch": 0.028654653975128293,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001534262,
      "loss": 3.1848,
      "step": 232870
    },
    {
      "epoch": 0.028673297211286475,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015342420000000003,
      "loss": 3.0637,
      "step": 232880
    },
    {
      "epoch": 0.02869194044744466,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001534222,
      "loss": 3.066,
      "step": 232890
    },
    {
      "epoch": 0.028710583683602842,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015342020000000002,
      "loss": 3.1085,
      "step": 232900
    },
    {
      "epoch": 0.028729226919761028,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001534182,
      "loss": 3.1345,
      "step": 232910
    },
    {
      "epoch": 0.02874787015591921,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001534162,
      "loss": 2.9109,
      "step": 232920
    },
    {
      "epoch": 0.028766513392077395,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001534142,
      "loss": 3.1921,
      "step": 232930
    },
    {
      "epoch": 0.028785156628235577,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001534122,
      "loss": 3.078,
      "step": 232940
    },
    {
      "epoch": 0.02880379986439376,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015341020000000002,
      "loss": 3.188,
      "step": 232950
    },
    {
      "epoch": 0.028822443100551945,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001534082,
      "loss": 3.1798,
      "step": 232960
    },
    {
      "epoch": 0.028841086336710127,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015340620000000002,
      "loss": 3.1716,
      "step": 232970
    },
    {
      "epoch": 0.028859729572868312,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001534042,
      "loss": 3.0893,
      "step": 232980
    },
    {
      "epoch": 0.028878372809026494,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001534022,
      "loss": 3.0948,
      "step": 232990
    },
    {
      "epoch": 0.02889701604518468,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015340020000000003,
      "loss": 3.0866,
      "step": 233000
    },
    {
      "epoch": 0.028915659281342862,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001533982,
      "loss": 3.1273,
      "step": 233010
    },
    {
      "epoch": 0.028934302517501048,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015339620000000002,
      "loss": 3.1178,
      "step": 233020
    },
    {
      "epoch": 0.02895294575365923,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001533942,
      "loss": 3.2521,
      "step": 233030
    },
    {
      "epoch": 0.028971588989817415,
      "grad_norm": 0.625,
      "learning_rate": 0.0001533922,
      "loss": 3.0703,
      "step": 233040
    },
    {
      "epoch": 0.028990232225975597,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001533902,
      "loss": 3.1982,
      "step": 233050
    },
    {
      "epoch": 0.029008875462133783,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001533882,
      "loss": 3.1122,
      "step": 233060
    },
    {
      "epoch": 0.029027518698291965,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015338620000000003,
      "loss": 3.1578,
      "step": 233070
    },
    {
      "epoch": 0.02904616193445015,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001533842,
      "loss": 2.9797,
      "step": 233080
    },
    {
      "epoch": 0.029064805170608332,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015338220000000002,
      "loss": 3.1978,
      "step": 233090
    },
    {
      "epoch": 0.029083448406766518,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001533802,
      "loss": 3.1039,
      "step": 233100
    },
    {
      "epoch": 0.0291020916429247,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001533782,
      "loss": 3.2489,
      "step": 233110
    },
    {
      "epoch": 0.02912073487908288,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001533762,
      "loss": 3.13,
      "step": 233120
    },
    {
      "epoch": 0.029139378115241067,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001533742,
      "loss": 3.155,
      "step": 233130
    },
    {
      "epoch": 0.02915802135139925,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015337220000000002,
      "loss": 3.192,
      "step": 233140
    },
    {
      "epoch": 0.029176664587557435,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001533702,
      "loss": 3.0818,
      "step": 233150
    },
    {
      "epoch": 0.029195307823715617,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001533682,
      "loss": 3.0859,
      "step": 233160
    },
    {
      "epoch": 0.029213951059873802,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001533662,
      "loss": 3.2043,
      "step": 233170
    },
    {
      "epoch": 0.029232594296031984,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001533642,
      "loss": 3.0937,
      "step": 233180
    },
    {
      "epoch": 0.02925123753219017,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001533622,
      "loss": 3.1076,
      "step": 233190
    },
    {
      "epoch": 0.029269880768348352,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001533602,
      "loss": 3.1417,
      "step": 233200
    },
    {
      "epoch": 0.029288524004506537,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015335820000000002,
      "loss": 3.1049,
      "step": 233210
    },
    {
      "epoch": 0.02930716724066472,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015335620000000001,
      "loss": 3.0976,
      "step": 233220
    },
    {
      "epoch": 0.029325810476822905,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001533542,
      "loss": 3.1994,
      "step": 233230
    },
    {
      "epoch": 0.029344453712981087,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001533522,
      "loss": 3.1839,
      "step": 233240
    },
    {
      "epoch": 0.029363096949139272,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001533502,
      "loss": 3.2315,
      "step": 233250
    },
    {
      "epoch": 0.029381740185297454,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001533482,
      "loss": 3.1719,
      "step": 233260
    },
    {
      "epoch": 0.02940038342145564,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015334620000000002,
      "loss": 3.1532,
      "step": 233270
    },
    {
      "epoch": 0.029419026657613822,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015334420000000001,
      "loss": 3.0738,
      "step": 233280
    },
    {
      "epoch": 0.029437669893772004,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001533422,
      "loss": 3.1637,
      "step": 233290
    },
    {
      "epoch": 0.02945631312993019,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001533402,
      "loss": 3.3058,
      "step": 233300
    },
    {
      "epoch": 0.02947495636608837,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001533382,
      "loss": 3.0694,
      "step": 233310
    },
    {
      "epoch": 0.029493599602246557,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001533362,
      "loss": 3.1094,
      "step": 233320
    },
    {
      "epoch": 0.02951224283840474,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001533342,
      "loss": 3.0576,
      "step": 233330
    },
    {
      "epoch": 0.029530886074562925,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015333220000000002,
      "loss": 3.1324,
      "step": 233340
    },
    {
      "epoch": 0.029549529310721107,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001533302,
      "loss": 3.0154,
      "step": 233350
    },
    {
      "epoch": 0.029568172546879292,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001533282,
      "loss": 3.112,
      "step": 233360
    },
    {
      "epoch": 0.029586815783037474,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001533262,
      "loss": 3.1141,
      "step": 233370
    },
    {
      "epoch": 0.02960545901919566,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001533242,
      "loss": 3.1534,
      "step": 233380
    },
    {
      "epoch": 0.02962410225535384,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015332220000000002,
      "loss": 3.1655,
      "step": 233390
    },
    {
      "epoch": 0.029642745491512027,
      "grad_norm": 0.75,
      "learning_rate": 0.0001533202,
      "loss": 3.1687,
      "step": 233400
    },
    {
      "epoch": 0.02966138872767021,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001533182,
      "loss": 3.2331,
      "step": 233410
    },
    {
      "epoch": 0.029680031963828395,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001533162,
      "loss": 3.1327,
      "step": 233420
    },
    {
      "epoch": 0.029698675199986577,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001533142,
      "loss": 3.1941,
      "step": 233430
    },
    {
      "epoch": 0.029717318436144762,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015331220000000003,
      "loss": 3.1077,
      "step": 233440
    },
    {
      "epoch": 0.029735961672302944,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001533102,
      "loss": 3.0517,
      "step": 233450
    },
    {
      "epoch": 0.029754604908461126,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015330820000000002,
      "loss": 3.163,
      "step": 233460
    },
    {
      "epoch": 0.029773248144619312,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001533062,
      "loss": 3.1058,
      "step": 233470
    },
    {
      "epoch": 0.029791891380777494,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001533042,
      "loss": 3.1578,
      "step": 233480
    },
    {
      "epoch": 0.02981053461693568,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015330220000000003,
      "loss": 3.1907,
      "step": 233490
    },
    {
      "epoch": 0.02982917785309386,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001533002,
      "loss": 3.0919,
      "step": 233500
    },
    {
      "epoch": 0.029847821089252047,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015329820000000002,
      "loss": 3.1321,
      "step": 233510
    },
    {
      "epoch": 0.02986646432541023,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001532962,
      "loss": 3.0665,
      "step": 233520
    },
    {
      "epoch": 0.029885107561568414,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015329420000000001,
      "loss": 3.1324,
      "step": 233530
    },
    {
      "epoch": 0.029903750797726596,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001532922,
      "loss": 3.0898,
      "step": 233540
    },
    {
      "epoch": 0.029922394033884782,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001532902,
      "loss": 3.2226,
      "step": 233550
    },
    {
      "epoch": 0.029941037270042964,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015328820000000003,
      "loss": 3.1559,
      "step": 233560
    },
    {
      "epoch": 0.02995968050620115,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001532862,
      "loss": 3.2782,
      "step": 233570
    },
    {
      "epoch": 0.02997832374235933,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015328420000000002,
      "loss": 3.1817,
      "step": 233580
    },
    {
      "epoch": 0.029996966978517517,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001532822,
      "loss": 3.1748,
      "step": 233590
    },
    {
      "epoch": 0.0300156102146757,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001532802,
      "loss": 3.1848,
      "step": 233600
    },
    {
      "epoch": 0.030034253450833884,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001532782,
      "loss": 3.1903,
      "step": 233610
    },
    {
      "epoch": 0.030052896686992066,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001532762,
      "loss": 3.1624,
      "step": 233620
    },
    {
      "epoch": 0.03007153992315025,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015327420000000002,
      "loss": 3.1904,
      "step": 233630
    },
    {
      "epoch": 0.030090183159308434,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001532722,
      "loss": 3.0897,
      "step": 233640
    },
    {
      "epoch": 0.030108826395466616,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015327020000000002,
      "loss": 3.1759,
      "step": 233650
    },
    {
      "epoch": 0.0301274696316248,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001532682,
      "loss": 3.1966,
      "step": 233660
    },
    {
      "epoch": 0.030146112867782984,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001532662,
      "loss": 3.0265,
      "step": 233670
    },
    {
      "epoch": 0.03016475610394117,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001532642,
      "loss": 3.2804,
      "step": 233680
    },
    {
      "epoch": 0.03018339934009935,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001532622,
      "loss": 3.2189,
      "step": 233690
    },
    {
      "epoch": 0.030202042576257537,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015326020000000002,
      "loss": 3.1806,
      "step": 233700
    },
    {
      "epoch": 0.03022068581241572,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015325820000000002,
      "loss": 3.1775,
      "step": 233710
    },
    {
      "epoch": 0.030239329048573904,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001532562,
      "loss": 3.1763,
      "step": 233720
    },
    {
      "epoch": 0.030257972284732086,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001532542,
      "loss": 3.0834,
      "step": 233730
    },
    {
      "epoch": 0.03027661552089027,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001532522,
      "loss": 3.0813,
      "step": 233740
    },
    {
      "epoch": 0.030295258757048454,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001532502,
      "loss": 3.1147,
      "step": 233750
    },
    {
      "epoch": 0.03031390199320664,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001532482,
      "loss": 3.0444,
      "step": 233760
    },
    {
      "epoch": 0.03033254522936482,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015324620000000002,
      "loss": 3.0934,
      "step": 233770
    },
    {
      "epoch": 0.030351188465523007,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001532442,
      "loss": 3.0042,
      "step": 233780
    },
    {
      "epoch": 0.03036983170168119,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001532422,
      "loss": 3.0707,
      "step": 233790
    },
    {
      "epoch": 0.03038847493783937,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001532402,
      "loss": 3.1239,
      "step": 233800
    },
    {
      "epoch": 0.030407118173997556,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001532382,
      "loss": 3.2842,
      "step": 233810
    },
    {
      "epoch": 0.03042576141015574,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001532362,
      "loss": 2.9944,
      "step": 233820
    },
    {
      "epoch": 0.030444404646313924,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00015323420000000002,
      "loss": 3.2249,
      "step": 233830
    },
    {
      "epoch": 0.030463047882472106,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001532322,
      "loss": 3.1086,
      "step": 233840
    },
    {
      "epoch": 0.03048169111863029,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001532302,
      "loss": 3.1477,
      "step": 233850
    },
    {
      "epoch": 0.030500334354788473,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001532282,
      "loss": 3.0804,
      "step": 233860
    },
    {
      "epoch": 0.03051897759094666,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001532262,
      "loss": 3.0327,
      "step": 233870
    },
    {
      "epoch": 0.03053762082710484,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015322420000000002,
      "loss": 3.162,
      "step": 233880
    },
    {
      "epoch": 0.030556264063263026,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001532222,
      "loss": 3.317,
      "step": 233890
    },
    {
      "epoch": 0.03057490729942121,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015322020000000001,
      "loss": 3.2676,
      "step": 233900
    },
    {
      "epoch": 0.030593550535579394,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001532182,
      "loss": 3.1181,
      "step": 233910
    },
    {
      "epoch": 0.030612193771737576,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001532162,
      "loss": 3.1192,
      "step": 233920
    },
    {
      "epoch": 0.03063083700789576,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001532142,
      "loss": 3.1854,
      "step": 233930
    },
    {
      "epoch": 0.030649480244053944,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001532122,
      "loss": 3.0777,
      "step": 233940
    },
    {
      "epoch": 0.03066812348021213,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015321020000000002,
      "loss": 2.9939,
      "step": 233950
    },
    {
      "epoch": 0.03068676671637031,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015320820000000001,
      "loss": 3.134,
      "step": 233960
    },
    {
      "epoch": 0.030705409952528493,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001532062,
      "loss": 3.2411,
      "step": 233970
    },
    {
      "epoch": 0.03072405318868668,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001532042,
      "loss": 2.9786,
      "step": 233980
    },
    {
      "epoch": 0.03074269642484486,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001532022,
      "loss": 3.155,
      "step": 233990
    },
    {
      "epoch": 0.030761339661003046,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015320020000000002,
      "loss": 3.1544,
      "step": 234000
    },
    {
      "epoch": 0.030779982897161228,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001531982,
      "loss": 3.0691,
      "step": 234010
    },
    {
      "epoch": 0.030798626133319414,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015319620000000002,
      "loss": 3.2177,
      "step": 234020
    },
    {
      "epoch": 0.030817269369477596,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001531942,
      "loss": 2.9948,
      "step": 234030
    },
    {
      "epoch": 0.03083591260563578,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001531922,
      "loss": 3.1785,
      "step": 234040
    },
    {
      "epoch": 0.030854555841793963,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015319020000000003,
      "loss": 3.0399,
      "step": 234050
    },
    {
      "epoch": 0.03087319907795215,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001531882,
      "loss": 3.1649,
      "step": 234060
    },
    {
      "epoch": 0.03089184231411033,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015318620000000002,
      "loss": 3.354,
      "step": 234070
    },
    {
      "epoch": 0.030910485550268516,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001531842,
      "loss": 3.2271,
      "step": 234080
    },
    {
      "epoch": 0.030929128786426698,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001531822,
      "loss": 3.0994,
      "step": 234090
    },
    {
      "epoch": 0.030947772022584884,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001531802,
      "loss": 3.1846,
      "step": 234100
    },
    {
      "epoch": 0.030966415258743066,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001531782,
      "loss": 3.1783,
      "step": 234110
    },
    {
      "epoch": 0.03098505849490125,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015317620000000003,
      "loss": 3.1698,
      "step": 234120
    },
    {
      "epoch": 0.031003701731059433,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001531742,
      "loss": 3.1207,
      "step": 234130
    },
    {
      "epoch": 0.031022344967217615,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015317220000000002,
      "loss": 3.2016,
      "step": 234140
    },
    {
      "epoch": 0.0310409882033758,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015317019999999999,
      "loss": 3.0116,
      "step": 234150
    },
    {
      "epoch": 0.031059631439533983,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001531682,
      "loss": 3.1666,
      "step": 234160
    },
    {
      "epoch": 0.03107827467569217,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001531662,
      "loss": 3.1947,
      "step": 234170
    },
    {
      "epoch": 0.03109691791185035,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001531642,
      "loss": 3.0945,
      "step": 234180
    },
    {
      "epoch": 0.031115561148008536,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015316220000000002,
      "loss": 3.0293,
      "step": 234190
    },
    {
      "epoch": 0.031134204384166718,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001531602,
      "loss": 3.1831,
      "step": 234200
    },
    {
      "epoch": 0.031152847620324903,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015315820000000001,
      "loss": 3.1098,
      "step": 234210
    },
    {
      "epoch": 0.031171490856483085,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001531562,
      "loss": 3.2174,
      "step": 234220
    },
    {
      "epoch": 0.03119013409264127,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001531542,
      "loss": 3.1899,
      "step": 234230
    },
    {
      "epoch": 0.031208777328799453,
      "grad_norm": 0.75,
      "learning_rate": 0.0001531522,
      "loss": 3.0258,
      "step": 234240
    },
    {
      "epoch": 0.03122742056495764,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001531502,
      "loss": 3.1171,
      "step": 234250
    },
    {
      "epoch": 0.03124606380111582,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015314820000000002,
      "loss": 3.0306,
      "step": 234260
    },
    {
      "epoch": 0.031264707037274006,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015314620000000001,
      "loss": 3.1012,
      "step": 234270
    },
    {
      "epoch": 0.03128335027343219,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001531442,
      "loss": 3.1393,
      "step": 234280
    },
    {
      "epoch": 0.03130199350959037,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001531422,
      "loss": 3.0577,
      "step": 234290
    },
    {
      "epoch": 0.03132063674574856,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001531402,
      "loss": 3.1953,
      "step": 234300
    },
    {
      "epoch": 0.03133927998190674,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001531382,
      "loss": 3.0673,
      "step": 234310
    },
    {
      "epoch": 0.03135792321806492,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001531362,
      "loss": 3.2294,
      "step": 234320
    },
    {
      "epoch": 0.031376566454223105,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015313420000000002,
      "loss": 3.0459,
      "step": 234330
    },
    {
      "epoch": 0.03139520969038129,
      "grad_norm": 0.75,
      "learning_rate": 0.0001531322,
      "loss": 3.2029,
      "step": 234340
    },
    {
      "epoch": 0.031413852926539476,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001531302,
      "loss": 3.0806,
      "step": 234350
    },
    {
      "epoch": 0.03143249616269766,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001531282,
      "loss": 3.1408,
      "step": 234360
    },
    {
      "epoch": 0.03145113939885584,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001531262,
      "loss": 3.1551,
      "step": 234370
    },
    {
      "epoch": 0.03146978263501402,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001531242,
      "loss": 3.126,
      "step": 234380
    },
    {
      "epoch": 0.03148842587117221,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015312220000000002,
      "loss": 3.018,
      "step": 234390
    },
    {
      "epoch": 0.03150706910733039,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001531202,
      "loss": 3.0974,
      "step": 234400
    },
    {
      "epoch": 0.031525712343488575,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001531182,
      "loss": 3.1261,
      "step": 234410
    },
    {
      "epoch": 0.03154435557964676,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001531162,
      "loss": 3.2307,
      "step": 234420
    },
    {
      "epoch": 0.031562998815804946,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001531142,
      "loss": 3.0645,
      "step": 234430
    },
    {
      "epoch": 0.03158164205196313,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015311220000000002,
      "loss": 3.2942,
      "step": 234440
    },
    {
      "epoch": 0.03160028528812131,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015311020000000002,
      "loss": 3.1407,
      "step": 234450
    },
    {
      "epoch": 0.03161892852427949,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001531082,
      "loss": 3.085,
      "step": 234460
    },
    {
      "epoch": 0.03163757176043768,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001531062,
      "loss": 3.163,
      "step": 234470
    },
    {
      "epoch": 0.03165621499659586,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001531042,
      "loss": 3.0845,
      "step": 234480
    },
    {
      "epoch": 0.031674858232754045,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015310220000000003,
      "loss": 3.2679,
      "step": 234490
    },
    {
      "epoch": 0.03169350146891223,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001531002,
      "loss": 3.165,
      "step": 234500
    },
    {
      "epoch": 0.03171214470507041,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015309820000000002,
      "loss": 3.0927,
      "step": 234510
    },
    {
      "epoch": 0.0317307879412286,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001530962,
      "loss": 3.1812,
      "step": 234520
    },
    {
      "epoch": 0.03174943117738678,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001530942,
      "loss": 3.1476,
      "step": 234530
    },
    {
      "epoch": 0.03176807441354496,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001530922,
      "loss": 3.209,
      "step": 234540
    },
    {
      "epoch": 0.031786717649703145,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001530902,
      "loss": 3.2431,
      "step": 234550
    },
    {
      "epoch": 0.031805360885861333,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015308820000000002,
      "loss": 3.0298,
      "step": 234560
    },
    {
      "epoch": 0.031824004122019516,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001530862,
      "loss": 3.1105,
      "step": 234570
    },
    {
      "epoch": 0.0318426473581777,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015308420000000001,
      "loss": 3.1264,
      "step": 234580
    },
    {
      "epoch": 0.03186129059433588,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001530822,
      "loss": 3.2374,
      "step": 234590
    },
    {
      "epoch": 0.03187993383049407,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001530802,
      "loss": 3.1246,
      "step": 234600
    },
    {
      "epoch": 0.03189857706665225,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015307820000000003,
      "loss": 3.2286,
      "step": 234610
    },
    {
      "epoch": 0.03191722030281043,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001530762,
      "loss": 3.2184,
      "step": 234620
    },
    {
      "epoch": 0.031935863538968615,
      "grad_norm": 0.625,
      "learning_rate": 0.00015307420000000002,
      "loss": 3.1289,
      "step": 234630
    },
    {
      "epoch": 0.031954506775126804,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001530722,
      "loss": 3.1078,
      "step": 234640
    },
    {
      "epoch": 0.031973150011284986,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001530702,
      "loss": 3.1722,
      "step": 234650
    },
    {
      "epoch": 0.03199179324744317,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001530682,
      "loss": 3.1061,
      "step": 234660
    },
    {
      "epoch": 0.03201043648360135,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001530662,
      "loss": 3.2851,
      "step": 234670
    },
    {
      "epoch": 0.03202907971975953,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015306420000000002,
      "loss": 3.2166,
      "step": 234680
    },
    {
      "epoch": 0.03204772295591772,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001530622,
      "loss": 2.9922,
      "step": 234690
    },
    {
      "epoch": 0.0320663661920759,
      "grad_norm": 0.75,
      "learning_rate": 0.00015306020000000002,
      "loss": 3.0968,
      "step": 234700
    },
    {
      "epoch": 0.032085009428234085,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001530582,
      "loss": 3.1885,
      "step": 234710
    },
    {
      "epoch": 0.03210365266439227,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001530562,
      "loss": 2.9718,
      "step": 234720
    },
    {
      "epoch": 0.032122295900550456,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001530542,
      "loss": 3.2988,
      "step": 234730
    },
    {
      "epoch": 0.03214093913670864,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001530522,
      "loss": 3.1474,
      "step": 234740
    },
    {
      "epoch": 0.03215958237286682,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015305020000000002,
      "loss": 3.1246,
      "step": 234750
    },
    {
      "epoch": 0.032178225609025,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001530482,
      "loss": 3.2283,
      "step": 234760
    },
    {
      "epoch": 0.03219686884518319,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001530462,
      "loss": 3.1702,
      "step": 234770
    },
    {
      "epoch": 0.03221551208134137,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001530442,
      "loss": 3.1707,
      "step": 234780
    },
    {
      "epoch": 0.032234155317499555,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001530422,
      "loss": 3.149,
      "step": 234790
    },
    {
      "epoch": 0.03225279855365774,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001530402,
      "loss": 3.1455,
      "step": 234800
    },
    {
      "epoch": 0.032271441789815926,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001530382,
      "loss": 3.1048,
      "step": 234810
    },
    {
      "epoch": 0.03229008502597411,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015303620000000002,
      "loss": 3.0554,
      "step": 234820
    },
    {
      "epoch": 0.03230872826213229,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001530342,
      "loss": 3.1698,
      "step": 234830
    },
    {
      "epoch": 0.03232737149829047,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001530322,
      "loss": 3.1672,
      "step": 234840
    },
    {
      "epoch": 0.032346014734448654,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001530302,
      "loss": 3.197,
      "step": 234850
    },
    {
      "epoch": 0.03236465797060684,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001530282,
      "loss": 3.1975,
      "step": 234860
    },
    {
      "epoch": 0.032383301206765025,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001530262,
      "loss": 3.1831,
      "step": 234870
    },
    {
      "epoch": 0.03240194444292321,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015302420000000002,
      "loss": 3.0377,
      "step": 234880
    },
    {
      "epoch": 0.03242058767908139,
      "grad_norm": 0.59375,
      "learning_rate": 0.00015302220000000001,
      "loss": 3.2315,
      "step": 234890
    },
    {
      "epoch": 0.03243923091523958,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001530202,
      "loss": 3.1069,
      "step": 234900
    },
    {
      "epoch": 0.03245787415139776,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001530182,
      "loss": 3.0955,
      "step": 234910
    },
    {
      "epoch": 0.03247651738755594,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001530162,
      "loss": 3.0988,
      "step": 234920
    },
    {
      "epoch": 0.032495160623714124,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001530142,
      "loss": 3.1986,
      "step": 234930
    },
    {
      "epoch": 0.03251380385987231,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015301220000000002,
      "loss": 3.2306,
      "step": 234940
    },
    {
      "epoch": 0.032532447096030495,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015301020000000001,
      "loss": 3.299,
      "step": 234950
    },
    {
      "epoch": 0.03255109033218868,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001530082,
      "loss": 3.175,
      "step": 234960
    },
    {
      "epoch": 0.03256973356834686,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001530062,
      "loss": 3.0919,
      "step": 234970
    },
    {
      "epoch": 0.03258837680450505,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001530042,
      "loss": 3.211,
      "step": 234980
    },
    {
      "epoch": 0.03260702004066323,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001530022,
      "loss": 3.2108,
      "step": 234990
    },
    {
      "epoch": 0.03262566327682141,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015300020000000002,
      "loss": 3.0791,
      "step": 235000
    },
    {
      "epoch": 0.032644306512979594,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015299820000000002,
      "loss": 3.1192,
      "step": 235010
    },
    {
      "epoch": 0.032662949749137776,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001529962,
      "loss": 3.1569,
      "step": 235020
    },
    {
      "epoch": 0.032681592985295965,
      "grad_norm": 0.75,
      "learning_rate": 0.0001529942,
      "loss": 3.0675,
      "step": 235030
    },
    {
      "epoch": 0.03270023622145415,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001529922,
      "loss": 3.1577,
      "step": 235040
    },
    {
      "epoch": 0.03271887945761233,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015299020000000002,
      "loss": 3.0079,
      "step": 235050
    },
    {
      "epoch": 0.03273752269377051,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001529882,
      "loss": 3.1657,
      "step": 235060
    },
    {
      "epoch": 0.0327561659299287,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015298620000000002,
      "loss": 3.1386,
      "step": 235070
    },
    {
      "epoch": 0.03277480916608688,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001529842,
      "loss": 3.298,
      "step": 235080
    },
    {
      "epoch": 0.032793452402245064,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001529822,
      "loss": 3.2287,
      "step": 235090
    },
    {
      "epoch": 0.032812095638403246,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015298020000000003,
      "loss": 3.113,
      "step": 235100
    },
    {
      "epoch": 0.032830738874561435,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001529782,
      "loss": 3.0713,
      "step": 235110
    },
    {
      "epoch": 0.03284938211071962,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015297620000000002,
      "loss": 3.1188,
      "step": 235120
    },
    {
      "epoch": 0.0328680253468778,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001529742,
      "loss": 3.1437,
      "step": 235130
    },
    {
      "epoch": 0.03288666858303598,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001529722,
      "loss": 3.1487,
      "step": 235140
    },
    {
      "epoch": 0.03290531181919417,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001529702,
      "loss": 3.1258,
      "step": 235150
    },
    {
      "epoch": 0.03292395505535235,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001529682,
      "loss": 3.2886,
      "step": 235160
    },
    {
      "epoch": 0.032942598291510534,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015296620000000003,
      "loss": 3.0916,
      "step": 235170
    },
    {
      "epoch": 0.032961241527668717,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001529642,
      "loss": 3.1795,
      "step": 235180
    },
    {
      "epoch": 0.0329798847638269,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015296220000000002,
      "loss": 3.0662,
      "step": 235190
    },
    {
      "epoch": 0.03299852799998509,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015296019999999999,
      "loss": 2.9885,
      "step": 235200
    },
    {
      "epoch": 0.03301717123614327,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001529582,
      "loss": 3.0444,
      "step": 235210
    },
    {
      "epoch": 0.03303581447230145,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001529562,
      "loss": 3.1659,
      "step": 235220
    },
    {
      "epoch": 0.033054457708459634,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001529542,
      "loss": 3.164,
      "step": 235230
    },
    {
      "epoch": 0.03307310094461782,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015295220000000002,
      "loss": 3.1501,
      "step": 235240
    },
    {
      "epoch": 0.033091744180776005,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001529502,
      "loss": 3.1892,
      "step": 235250
    },
    {
      "epoch": 0.03311038741693419,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015294820000000001,
      "loss": 3.0975,
      "step": 235260
    },
    {
      "epoch": 0.03312903065309237,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001529462,
      "loss": 3.1097,
      "step": 235270
    },
    {
      "epoch": 0.03314767388925056,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001529442,
      "loss": 3.1297,
      "step": 235280
    },
    {
      "epoch": 0.03316631712540874,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001529422,
      "loss": 3.128,
      "step": 235290
    },
    {
      "epoch": 0.03318496036156692,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001529402,
      "loss": 3.1615,
      "step": 235300
    },
    {
      "epoch": 0.033203603597725104,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015293820000000002,
      "loss": 3.1194,
      "step": 235310
    },
    {
      "epoch": 0.03322224683388329,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015293620000000001,
      "loss": 3.0807,
      "step": 235320
    },
    {
      "epoch": 0.033240890070041475,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001529342,
      "loss": 3.0989,
      "step": 235330
    },
    {
      "epoch": 0.03325953330619966,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001529322,
      "loss": 3.0173,
      "step": 235340
    },
    {
      "epoch": 0.03327817654235784,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001529302,
      "loss": 3.1172,
      "step": 235350
    },
    {
      "epoch": 0.03329681977851602,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001529282,
      "loss": 3.2473,
      "step": 235360
    },
    {
      "epoch": 0.03331546301467421,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001529262,
      "loss": 3.0621,
      "step": 235370
    },
    {
      "epoch": 0.03333410625083239,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015292420000000002,
      "loss": 3.0129,
      "step": 235380
    },
    {
      "epoch": 0.033352749486990574,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001529222,
      "loss": 2.9948,
      "step": 235390
    },
    {
      "epoch": 0.033371392723148756,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001529202,
      "loss": 3.0985,
      "step": 235400
    },
    {
      "epoch": 0.033390035959306945,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001529182,
      "loss": 3.1843,
      "step": 235410
    },
    {
      "epoch": 0.03340867919546513,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001529162,
      "loss": 3.1405,
      "step": 235420
    },
    {
      "epoch": 0.03342732243162331,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015291420000000002,
      "loss": 3.2357,
      "step": 235430
    },
    {
      "epoch": 0.03344596566778149,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015291220000000002,
      "loss": 3.0904,
      "step": 235440
    },
    {
      "epoch": 0.03346460890393968,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001529102,
      "loss": 3.1708,
      "step": 235450
    },
    {
      "epoch": 0.03348325214009786,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001529082,
      "loss": 3.2559,
      "step": 235460
    },
    {
      "epoch": 0.033501895376256044,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001529062,
      "loss": 3.3387,
      "step": 235470
    },
    {
      "epoch": 0.033520538612414226,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001529042,
      "loss": 3.1229,
      "step": 235480
    },
    {
      "epoch": 0.033539181848572415,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015290220000000002,
      "loss": 3.0956,
      "step": 235490
    },
    {
      "epoch": 0.0335578250847306,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015290020000000002,
      "loss": 3.1293,
      "step": 235500
    },
    {
      "epoch": 0.03357646832088878,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001528982,
      "loss": 3.0809,
      "step": 235510
    },
    {
      "epoch": 0.03359511155704696,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001528962,
      "loss": 3.2893,
      "step": 235520
    },
    {
      "epoch": 0.03361375479320514,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001528942,
      "loss": 3.1847,
      "step": 235530
    },
    {
      "epoch": 0.03363239802936333,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001528922,
      "loss": 3.1784,
      "step": 235540
    },
    {
      "epoch": 0.033651041265521514,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001528902,
      "loss": 3.2312,
      "step": 235550
    },
    {
      "epoch": 0.033669684501679696,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015288820000000002,
      "loss": 2.9432,
      "step": 235560
    },
    {
      "epoch": 0.03368832773783788,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001528862,
      "loss": 3.1914,
      "step": 235570
    },
    {
      "epoch": 0.03370697097399607,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001528842,
      "loss": 3.1758,
      "step": 235580
    },
    {
      "epoch": 0.03372561421015425,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001528822,
      "loss": 3.2695,
      "step": 235590
    },
    {
      "epoch": 0.03374425744631243,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001528802,
      "loss": 3.1893,
      "step": 235600
    },
    {
      "epoch": 0.03376290068247061,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015287820000000002,
      "loss": 3.1345,
      "step": 235610
    },
    {
      "epoch": 0.0337815439186288,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001528762,
      "loss": 3.1975,
      "step": 235620
    },
    {
      "epoch": 0.033800187154786984,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015287420000000001,
      "loss": 3.056,
      "step": 235630
    },
    {
      "epoch": 0.033818830390945166,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001528722,
      "loss": 3.0566,
      "step": 235640
    },
    {
      "epoch": 0.03383747362710335,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001528702,
      "loss": 3.0146,
      "step": 235650
    },
    {
      "epoch": 0.03385611686326154,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015286820000000003,
      "loss": 3.0292,
      "step": 235660
    },
    {
      "epoch": 0.03387476009941972,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001528662,
      "loss": 3.1086,
      "step": 235670
    },
    {
      "epoch": 0.0338934033355779,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015286420000000002,
      "loss": 3.1656,
      "step": 235680
    },
    {
      "epoch": 0.03391204657173608,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001528622,
      "loss": 3.1829,
      "step": 235690
    },
    {
      "epoch": 0.033930689807894265,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001528602,
      "loss": 3.1648,
      "step": 235700
    },
    {
      "epoch": 0.033949333044052454,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001528582,
      "loss": 3.2211,
      "step": 235710
    },
    {
      "epoch": 0.033967976280210636,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001528562,
      "loss": 3.1462,
      "step": 235720
    },
    {
      "epoch": 0.03398661951636882,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015285420000000002,
      "loss": 3.0561,
      "step": 235730
    },
    {
      "epoch": 0.034005262752527,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001528522,
      "loss": 3.2145,
      "step": 235740
    },
    {
      "epoch": 0.03402390598868519,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015285020000000002,
      "loss": 3.1682,
      "step": 235750
    },
    {
      "epoch": 0.03404254922484337,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015284819999999998,
      "loss": 3.1775,
      "step": 235760
    },
    {
      "epoch": 0.03406119246100155,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001528462,
      "loss": 3.1553,
      "step": 235770
    },
    {
      "epoch": 0.034079835697159735,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001528442,
      "loss": 3.1446,
      "step": 235780
    },
    {
      "epoch": 0.034098478933317924,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001528422,
      "loss": 3.1299,
      "step": 235790
    },
    {
      "epoch": 0.034117122169476106,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015284020000000002,
      "loss": 3.2003,
      "step": 235800
    },
    {
      "epoch": 0.03413576540563429,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001528382,
      "loss": 3.0511,
      "step": 235810
    },
    {
      "epoch": 0.03415440864179247,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001528362,
      "loss": 3.3401,
      "step": 235820
    },
    {
      "epoch": 0.03417305187795066,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001528342,
      "loss": 3.0368,
      "step": 235830
    },
    {
      "epoch": 0.03419169511410884,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001528322,
      "loss": 3.2231,
      "step": 235840
    },
    {
      "epoch": 0.034210338350267024,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001528302,
      "loss": 3.1895,
      "step": 235850
    },
    {
      "epoch": 0.034228981586425206,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001528282,
      "loss": 3.0368,
      "step": 235860
    },
    {
      "epoch": 0.03424762482258339,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015282620000000002,
      "loss": 3.1314,
      "step": 235870
    },
    {
      "epoch": 0.03426626805874158,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001528242,
      "loss": 3.1602,
      "step": 235880
    },
    {
      "epoch": 0.03428491129489976,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001528222,
      "loss": 3.2579,
      "step": 235890
    },
    {
      "epoch": 0.03430355453105794,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001528202,
      "loss": 3.1506,
      "step": 235900
    },
    {
      "epoch": 0.03432219776721612,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001528182,
      "loss": 3.0841,
      "step": 235910
    },
    {
      "epoch": 0.03434084100337431,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015281620000000002,
      "loss": 3.1108,
      "step": 235920
    },
    {
      "epoch": 0.034359484239532494,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001528142,
      "loss": 3.1437,
      "step": 235930
    },
    {
      "epoch": 0.034378127475690676,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015281220000000001,
      "loss": 3.0792,
      "step": 235940
    },
    {
      "epoch": 0.03439677071184886,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001528102,
      "loss": 3.027,
      "step": 235950
    },
    {
      "epoch": 0.03441541394800705,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001528082,
      "loss": 3.0725,
      "step": 235960
    },
    {
      "epoch": 0.03443405718416523,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001528062,
      "loss": 3.1897,
      "step": 235970
    },
    {
      "epoch": 0.03445270042032341,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001528042,
      "loss": 3.1343,
      "step": 235980
    },
    {
      "epoch": 0.03447134365648159,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015280220000000002,
      "loss": 3.1927,
      "step": 235990
    },
    {
      "epoch": 0.03448998689263978,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015280020000000001,
      "loss": 3.1922,
      "step": 236000
    },
    {
      "epoch": 0.034508630128797964,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001527982,
      "loss": 3.1579,
      "step": 236010
    },
    {
      "epoch": 0.034527273364956146,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001527962,
      "loss": 3.2399,
      "step": 236020
    },
    {
      "epoch": 0.03454591660111433,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001527942,
      "loss": 3.0894,
      "step": 236030
    },
    {
      "epoch": 0.03456455983727251,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001527922,
      "loss": 2.9945,
      "step": 236040
    },
    {
      "epoch": 0.0345832030734307,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015279020000000002,
      "loss": 3.1122,
      "step": 236050
    },
    {
      "epoch": 0.03460184630958888,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015278820000000002,
      "loss": 3.0875,
      "step": 236060
    },
    {
      "epoch": 0.03462048954574706,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001527862,
      "loss": 3.1757,
      "step": 236070
    },
    {
      "epoch": 0.034639132781905245,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001527842,
      "loss": 3.2005,
      "step": 236080
    },
    {
      "epoch": 0.034657776018063434,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001527822,
      "loss": 3.1589,
      "step": 236090
    },
    {
      "epoch": 0.034676419254221616,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015278020000000003,
      "loss": 3.0437,
      "step": 236100
    },
    {
      "epoch": 0.0346950624903798,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001527782,
      "loss": 3.2227,
      "step": 236110
    },
    {
      "epoch": 0.03471370572653798,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015277620000000002,
      "loss": 3.1106,
      "step": 236120
    },
    {
      "epoch": 0.03473234896269617,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001527742,
      "loss": 3.25,
      "step": 236130
    },
    {
      "epoch": 0.03475099219885435,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001527722,
      "loss": 3.1896,
      "step": 236140
    },
    {
      "epoch": 0.03476963543501253,
      "grad_norm": 0.75,
      "learning_rate": 0.0001527702,
      "loss": 3.1939,
      "step": 236150
    },
    {
      "epoch": 0.034788278671170715,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001527682,
      "loss": 3.0412,
      "step": 236160
    },
    {
      "epoch": 0.034806921907328904,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015276620000000002,
      "loss": 3.2271,
      "step": 236170
    },
    {
      "epoch": 0.034825565143487086,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001527642,
      "loss": 3.188,
      "step": 236180
    },
    {
      "epoch": 0.03484420837964527,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001527622,
      "loss": 3.1264,
      "step": 236190
    },
    {
      "epoch": 0.03486285161580345,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001527602,
      "loss": 3.1367,
      "step": 236200
    },
    {
      "epoch": 0.03488149485196163,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001527582,
      "loss": 3.1663,
      "step": 236210
    },
    {
      "epoch": 0.03490013808811982,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015275620000000003,
      "loss": 3.1204,
      "step": 236220
    },
    {
      "epoch": 0.034918781324278,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001527542,
      "loss": 3.0612,
      "step": 236230
    },
    {
      "epoch": 0.034937424560436185,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015275220000000002,
      "loss": 3.0989,
      "step": 236240
    },
    {
      "epoch": 0.03495606779659437,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015275019999999999,
      "loss": 3.1062,
      "step": 236250
    },
    {
      "epoch": 0.034974711032752556,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001527482,
      "loss": 3.1217,
      "step": 236260
    },
    {
      "epoch": 0.03499335426891074,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001527462,
      "loss": 3.2045,
      "step": 236270
    },
    {
      "epoch": 0.03501199750506892,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001527442,
      "loss": 3.2447,
      "step": 236280
    },
    {
      "epoch": 0.0350306407412271,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00015274220000000002,
      "loss": 3.0388,
      "step": 236290
    },
    {
      "epoch": 0.03504928397738529,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001527402,
      "loss": 3.2436,
      "step": 236300
    },
    {
      "epoch": 0.03506792721354347,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015273820000000001,
      "loss": 3.0174,
      "step": 236310
    },
    {
      "epoch": 0.035086570449701655,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001527362,
      "loss": 3.2303,
      "step": 236320
    },
    {
      "epoch": 0.03510521368585984,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001527342,
      "loss": 3.1719,
      "step": 236330
    },
    {
      "epoch": 0.035123856922018026,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001527322,
      "loss": 3.0626,
      "step": 236340
    },
    {
      "epoch": 0.03514250015817621,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001527302,
      "loss": 3.0041,
      "step": 236350
    },
    {
      "epoch": 0.03516114339433439,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015272820000000002,
      "loss": 3.0932,
      "step": 236360
    },
    {
      "epoch": 0.03517978663049257,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001527262,
      "loss": 3.1881,
      "step": 236370
    },
    {
      "epoch": 0.035198429866650754,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001527242,
      "loss": 2.9976,
      "step": 236380
    },
    {
      "epoch": 0.03521707310280894,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001527222,
      "loss": 2.9404,
      "step": 236390
    },
    {
      "epoch": 0.035235716338967125,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001527202,
      "loss": 3.0977,
      "step": 236400
    },
    {
      "epoch": 0.03525435957512531,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015271820000000002,
      "loss": 3.1603,
      "step": 236410
    },
    {
      "epoch": 0.03527300281128349,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001527162,
      "loss": 3.205,
      "step": 236420
    },
    {
      "epoch": 0.03529164604744168,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015271420000000002,
      "loss": 3.1263,
      "step": 236430
    },
    {
      "epoch": 0.03531028928359986,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001527122,
      "loss": 2.994,
      "step": 236440
    },
    {
      "epoch": 0.03532893251975804,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001527102,
      "loss": 3.1558,
      "step": 236450
    },
    {
      "epoch": 0.035347575755916225,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001527082,
      "loss": 3.214,
      "step": 236460
    },
    {
      "epoch": 0.035366218992074414,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001527062,
      "loss": 3.1832,
      "step": 236470
    },
    {
      "epoch": 0.035384862228232596,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015270420000000002,
      "loss": 3.1841,
      "step": 236480
    },
    {
      "epoch": 0.03540350546439078,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015270220000000002,
      "loss": 3.1111,
      "step": 236490
    },
    {
      "epoch": 0.03542214870054896,
      "grad_norm": 0.625,
      "learning_rate": 0.0001527002,
      "loss": 3.0305,
      "step": 236500
    },
    {
      "epoch": 0.03544079193670715,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001526982,
      "loss": 3.0354,
      "step": 236510
    },
    {
      "epoch": 0.03545943517286533,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001526962,
      "loss": 3.1634,
      "step": 236520
    },
    {
      "epoch": 0.03547807840902351,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001526942,
      "loss": 3.0947,
      "step": 236530
    },
    {
      "epoch": 0.035496721645181695,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001526922,
      "loss": 3.1387,
      "step": 236540
    },
    {
      "epoch": 0.03551536488133988,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015269020000000002,
      "loss": 3.0043,
      "step": 236550
    },
    {
      "epoch": 0.035534008117498066,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001526882,
      "loss": 3.1146,
      "step": 236560
    },
    {
      "epoch": 0.03555265135365625,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001526862,
      "loss": 3.1011,
      "step": 236570
    },
    {
      "epoch": 0.03557129458981443,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001526842,
      "loss": 3.227,
      "step": 236580
    },
    {
      "epoch": 0.03558993782597261,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001526822,
      "loss": 3.0164,
      "step": 236590
    },
    {
      "epoch": 0.0356085810621308,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001526802,
      "loss": 3.1165,
      "step": 236600
    },
    {
      "epoch": 0.03562722429828898,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015267820000000002,
      "loss": 3.061,
      "step": 236610
    },
    {
      "epoch": 0.035645867534447165,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015267620000000001,
      "loss": 3.1097,
      "step": 236620
    },
    {
      "epoch": 0.03566451077060535,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001526742,
      "loss": 3.0316,
      "step": 236630
    },
    {
      "epoch": 0.035683154006763536,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001526722,
      "loss": 3.0948,
      "step": 236640
    },
    {
      "epoch": 0.03570179724292172,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001526702,
      "loss": 3.1225,
      "step": 236650
    },
    {
      "epoch": 0.0357204404790799,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015266820000000002,
      "loss": 3.1092,
      "step": 236660
    },
    {
      "epoch": 0.03573908371523808,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001526662,
      "loss": 3.2107,
      "step": 236670
    },
    {
      "epoch": 0.03575772695139627,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015266420000000001,
      "loss": 3.1212,
      "step": 236680
    },
    {
      "epoch": 0.03577637018755445,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001526622,
      "loss": 3.1588,
      "step": 236690
    },
    {
      "epoch": 0.035795013423712635,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001526602,
      "loss": 2.9807,
      "step": 236700
    },
    {
      "epoch": 0.03581365665987082,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015265820000000003,
      "loss": 3.0253,
      "step": 236710
    },
    {
      "epoch": 0.035832299896029,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001526562,
      "loss": 3.1241,
      "step": 236720
    },
    {
      "epoch": 0.03585094313218719,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015265420000000002,
      "loss": 3.1574,
      "step": 236730
    },
    {
      "epoch": 0.03586958636834537,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001526522,
      "loss": 3.1462,
      "step": 236740
    },
    {
      "epoch": 0.03588822960450355,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001526502,
      "loss": 3.0696,
      "step": 236750
    },
    {
      "epoch": 0.035906872840661734,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001526482,
      "loss": 3.122,
      "step": 236760
    },
    {
      "epoch": 0.03592551607681992,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001526462,
      "loss": 3.131,
      "step": 236770
    },
    {
      "epoch": 0.035944159312978105,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015264420000000003,
      "loss": 3.0497,
      "step": 236780
    },
    {
      "epoch": 0.03596280254913629,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001526422,
      "loss": 3.1164,
      "step": 236790
    },
    {
      "epoch": 0.03598144578529447,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015264020000000002,
      "loss": 3.0097,
      "step": 236800
    },
    {
      "epoch": 0.03600008902145266,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015263819999999998,
      "loss": 2.9231,
      "step": 236810
    },
    {
      "epoch": 0.03601873225761084,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001526362,
      "loss": 3.121,
      "step": 236820
    },
    {
      "epoch": 0.03603737549376902,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001526342,
      "loss": 3.0777,
      "step": 236830
    },
    {
      "epoch": 0.036056018729927204,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001526322,
      "loss": 3.1826,
      "step": 236840
    },
    {
      "epoch": 0.03607466196608539,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015263020000000002,
      "loss": 3.1431,
      "step": 236850
    },
    {
      "epoch": 0.036093305202243575,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001526282,
      "loss": 3.2625,
      "step": 236860
    },
    {
      "epoch": 0.03611194843840176,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001526262,
      "loss": 3.2685,
      "step": 236870
    },
    {
      "epoch": 0.03613059167455994,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001526242,
      "loss": 3.23,
      "step": 236880
    },
    {
      "epoch": 0.03614923491071812,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001526222,
      "loss": 3.1792,
      "step": 236890
    },
    {
      "epoch": 0.03616787814687631,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015262020000000003,
      "loss": 3.0737,
      "step": 236900
    },
    {
      "epoch": 0.03618652138303449,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001526182,
      "loss": 3.1486,
      "step": 236910
    },
    {
      "epoch": 0.036205164619192674,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015261620000000002,
      "loss": 3.1208,
      "step": 236920
    },
    {
      "epoch": 0.036223807855350856,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001526142,
      "loss": 3.0946,
      "step": 236930
    },
    {
      "epoch": 0.036242451091509045,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001526122,
      "loss": 3.1736,
      "step": 236940
    },
    {
      "epoch": 0.03626109432766723,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001526102,
      "loss": 3.2081,
      "step": 236950
    },
    {
      "epoch": 0.03627973756382541,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001526082,
      "loss": 3.1413,
      "step": 236960
    },
    {
      "epoch": 0.03629838079998359,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015260620000000002,
      "loss": 3.2168,
      "step": 236970
    },
    {
      "epoch": 0.03631702403614178,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001526042,
      "loss": 3.1872,
      "step": 236980
    },
    {
      "epoch": 0.03633566727229996,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015260220000000001,
      "loss": 3.2695,
      "step": 236990
    },
    {
      "epoch": 0.036354310508458144,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001526002,
      "loss": 3.1042,
      "step": 237000
    },
    {
      "epoch": 0.036372953744616326,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001525982,
      "loss": 3.2603,
      "step": 237010
    },
    {
      "epoch": 0.036391596980774515,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001525962,
      "loss": 3.118,
      "step": 237020
    },
    {
      "epoch": 0.0364102402169327,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001525942,
      "loss": 3.109,
      "step": 237030
    },
    {
      "epoch": 0.03642888345309088,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015259220000000002,
      "loss": 3.1957,
      "step": 237040
    },
    {
      "epoch": 0.03644752668924906,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015259020000000002,
      "loss": 3.3178,
      "step": 237050
    },
    {
      "epoch": 0.036466169925407244,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001525882,
      "loss": 3.1761,
      "step": 237060
    },
    {
      "epoch": 0.03648481316156543,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001525862,
      "loss": 3.1781,
      "step": 237070
    },
    {
      "epoch": 0.036503456397723615,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001525842,
      "loss": 3.1622,
      "step": 237080
    },
    {
      "epoch": 0.0365220996338818,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001525822,
      "loss": 3.0482,
      "step": 237090
    },
    {
      "epoch": 0.03654074287003998,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015258020000000002,
      "loss": 3.2547,
      "step": 237100
    },
    {
      "epoch": 0.03655938610619817,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015257820000000002,
      "loss": 3.0947,
      "step": 237110
    },
    {
      "epoch": 0.03657802934235635,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001525762,
      "loss": 3.1832,
      "step": 237120
    },
    {
      "epoch": 0.03659667257851453,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001525742,
      "loss": 3.0913,
      "step": 237130
    },
    {
      "epoch": 0.036615315814672714,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001525722,
      "loss": 3.222,
      "step": 237140
    },
    {
      "epoch": 0.0366339590508309,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001525702,
      "loss": 3.0911,
      "step": 237150
    },
    {
      "epoch": 0.036652602286989085,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001525682,
      "loss": 3.1591,
      "step": 237160
    },
    {
      "epoch": 0.03667124552314727,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015256620000000002,
      "loss": 3.1063,
      "step": 237170
    },
    {
      "epoch": 0.03668988875930545,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001525642,
      "loss": 3.1255,
      "step": 237180
    },
    {
      "epoch": 0.03670853199546364,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001525622,
      "loss": 3.1506,
      "step": 237190
    },
    {
      "epoch": 0.03672717523162182,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001525602,
      "loss": 3.0592,
      "step": 237200
    },
    {
      "epoch": 0.03674581846778,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001525582,
      "loss": 3.0971,
      "step": 237210
    },
    {
      "epoch": 0.036764461703938184,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015255620000000002,
      "loss": 3.1341,
      "step": 237220
    },
    {
      "epoch": 0.036783104940096366,
      "grad_norm": 0.75,
      "learning_rate": 0.0001525542,
      "loss": 3.1894,
      "step": 237230
    },
    {
      "epoch": 0.036801748176254555,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001525522,
      "loss": 3.0664,
      "step": 237240
    },
    {
      "epoch": 0.03682039141241274,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001525502,
      "loss": 3.1987,
      "step": 237250
    },
    {
      "epoch": 0.03683903464857092,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001525482,
      "loss": 3.1133,
      "step": 237260
    },
    {
      "epoch": 0.0368576778847291,
      "grad_norm": 0.75,
      "learning_rate": 0.00015254620000000003,
      "loss": 3.0465,
      "step": 237270
    },
    {
      "epoch": 0.03687632112088729,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001525442,
      "loss": 3.1421,
      "step": 237280
    },
    {
      "epoch": 0.03689496435704547,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015254220000000002,
      "loss": 3.2062,
      "step": 237290
    },
    {
      "epoch": 0.036913607593203654,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001525402,
      "loss": 3.1349,
      "step": 237300
    },
    {
      "epoch": 0.036932250829361836,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001525382,
      "loss": 3.1387,
      "step": 237310
    },
    {
      "epoch": 0.036950894065520025,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001525362,
      "loss": 2.9315,
      "step": 237320
    },
    {
      "epoch": 0.03696953730167821,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001525342,
      "loss": 3.0719,
      "step": 237330
    },
    {
      "epoch": 0.03698818053783639,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015253220000000002,
      "loss": 3.1358,
      "step": 237340
    },
    {
      "epoch": 0.03700682377399457,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001525302,
      "loss": 3.1825,
      "step": 237350
    },
    {
      "epoch": 0.03702546701015276,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015252820000000001,
      "loss": 3.1839,
      "step": 237360
    },
    {
      "epoch": 0.03704411024631094,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001525262,
      "loss": 3.1374,
      "step": 237370
    },
    {
      "epoch": 0.037062753482469124,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001525242,
      "loss": 3.2171,
      "step": 237380
    },
    {
      "epoch": 0.037081396718627306,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015252220000000003,
      "loss": 3.1765,
      "step": 237390
    },
    {
      "epoch": 0.03710003995478549,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001525202,
      "loss": 3.0426,
      "step": 237400
    },
    {
      "epoch": 0.03711868319094368,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015251820000000002,
      "loss": 3.2211,
      "step": 237410
    },
    {
      "epoch": 0.03713732642710186,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001525162,
      "loss": 3.1124,
      "step": 237420
    },
    {
      "epoch": 0.03715596966326004,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001525142,
      "loss": 3.1777,
      "step": 237430
    },
    {
      "epoch": 0.03717461289941822,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001525122,
      "loss": 3.1597,
      "step": 237440
    },
    {
      "epoch": 0.03719325613557641,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001525102,
      "loss": 3.0653,
      "step": 237450
    },
    {
      "epoch": 0.037211899371734594,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015250820000000002,
      "loss": 3.1226,
      "step": 237460
    },
    {
      "epoch": 0.037230542607892776,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001525062,
      "loss": 3.1589,
      "step": 237470
    },
    {
      "epoch": 0.03724918584405096,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015250420000000002,
      "loss": 3.2336,
      "step": 237480
    },
    {
      "epoch": 0.03726782908020915,
      "grad_norm": 0.75,
      "learning_rate": 0.0001525022,
      "loss": 3.186,
      "step": 237490
    },
    {
      "epoch": 0.03728647231636733,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001525002,
      "loss": 3.0354,
      "step": 237500
    },
    {
      "epoch": 0.03730511555252551,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001524982,
      "loss": 3.0667,
      "step": 237510
    },
    {
      "epoch": 0.03732375878868369,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001524962,
      "loss": 3.1812,
      "step": 237520
    },
    {
      "epoch": 0.03734240202484188,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015249420000000002,
      "loss": 3.1812,
      "step": 237530
    },
    {
      "epoch": 0.037361045261000064,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015249220000000002,
      "loss": 3.1787,
      "step": 237540
    },
    {
      "epoch": 0.037379688497158246,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001524902,
      "loss": 2.9997,
      "step": 237550
    },
    {
      "epoch": 0.03739833173331643,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001524882,
      "loss": 3.0545,
      "step": 237560
    },
    {
      "epoch": 0.03741697496947461,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001524862,
      "loss": 3.0255,
      "step": 237570
    },
    {
      "epoch": 0.0374356182056328,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001524842,
      "loss": 3.1861,
      "step": 237580
    },
    {
      "epoch": 0.03745426144179098,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001524822,
      "loss": 3.1835,
      "step": 237590
    },
    {
      "epoch": 0.03747290467794916,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015248020000000002,
      "loss": 3.1795,
      "step": 237600
    },
    {
      "epoch": 0.037491547914107345,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001524782,
      "loss": 3.2577,
      "step": 237610
    },
    {
      "epoch": 0.037510191150265534,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001524762,
      "loss": 3.0112,
      "step": 237620
    },
    {
      "epoch": 0.037528834386423716,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001524742,
      "loss": 3.1444,
      "step": 237630
    },
    {
      "epoch": 0.0375474776225819,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001524722,
      "loss": 2.9541,
      "step": 237640
    },
    {
      "epoch": 0.03756612085874008,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001524702,
      "loss": 3.1808,
      "step": 237650
    },
    {
      "epoch": 0.03758476409489827,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015246820000000002,
      "loss": 3.0619,
      "step": 237660
    },
    {
      "epoch": 0.03760340733105645,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015246620000000001,
      "loss": 3.2476,
      "step": 237670
    },
    {
      "epoch": 0.037622050567214634,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001524642,
      "loss": 3.2546,
      "step": 237680
    },
    {
      "epoch": 0.037640693803372816,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001524622,
      "loss": 3.0653,
      "step": 237690
    },
    {
      "epoch": 0.037659337039531005,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001524602,
      "loss": 3.0661,
      "step": 237700
    },
    {
      "epoch": 0.03767798027568919,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015245820000000002,
      "loss": 3.1404,
      "step": 237710
    },
    {
      "epoch": 0.03769662351184737,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001524562,
      "loss": 3.1401,
      "step": 237720
    },
    {
      "epoch": 0.03771526674800555,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015245420000000001,
      "loss": 3.1551,
      "step": 237730
    },
    {
      "epoch": 0.03773390998416373,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001524522,
      "loss": 3.0898,
      "step": 237740
    },
    {
      "epoch": 0.03775255322032192,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001524502,
      "loss": 3.1026,
      "step": 237750
    },
    {
      "epoch": 0.037771196456480104,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001524482,
      "loss": 3.103,
      "step": 237760
    },
    {
      "epoch": 0.037789839692638286,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001524462,
      "loss": 3.2957,
      "step": 237770
    },
    {
      "epoch": 0.03780848292879647,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015244420000000002,
      "loss": 3.0675,
      "step": 237780
    },
    {
      "epoch": 0.03782712616495466,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001524422,
      "loss": 3.1283,
      "step": 237790
    },
    {
      "epoch": 0.03784576940111284,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001524402,
      "loss": 3.1824,
      "step": 237800
    },
    {
      "epoch": 0.03786441263727102,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001524382,
      "loss": 3.1943,
      "step": 237810
    },
    {
      "epoch": 0.0378830558734292,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001524362,
      "loss": 3.0557,
      "step": 237820
    },
    {
      "epoch": 0.03790169910958739,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015243420000000003,
      "loss": 3.0944,
      "step": 237830
    },
    {
      "epoch": 0.037920342345745574,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001524322,
      "loss": 3.1741,
      "step": 237840
    },
    {
      "epoch": 0.037938985581903756,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015243020000000002,
      "loss": 3.2271,
      "step": 237850
    },
    {
      "epoch": 0.03795762881806194,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001524282,
      "loss": 3.1165,
      "step": 237860
    },
    {
      "epoch": 0.03797627205422013,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001524262,
      "loss": 3.228,
      "step": 237870
    },
    {
      "epoch": 0.03799491529037831,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015242420000000003,
      "loss": 3.0396,
      "step": 237880
    },
    {
      "epoch": 0.03801355852653649,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001524222,
      "loss": 3.2507,
      "step": 237890
    },
    {
      "epoch": 0.03803220176269467,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015242020000000002,
      "loss": 3.0474,
      "step": 237900
    },
    {
      "epoch": 0.038050844998852855,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001524182,
      "loss": 3.1452,
      "step": 237910
    },
    {
      "epoch": 0.038069488235011044,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001524162,
      "loss": 3.0693,
      "step": 237920
    },
    {
      "epoch": 0.038088131471169226,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001524142,
      "loss": 3.0726,
      "step": 237930
    },
    {
      "epoch": 0.03810677470732741,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001524122,
      "loss": 3.1093,
      "step": 237940
    },
    {
      "epoch": 0.03812541794348559,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015241020000000003,
      "loss": 3.2009,
      "step": 237950
    },
    {
      "epoch": 0.03814406117964378,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001524082,
      "loss": 3.1108,
      "step": 237960
    },
    {
      "epoch": 0.03816270441580196,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015240620000000002,
      "loss": 3.1695,
      "step": 237970
    },
    {
      "epoch": 0.03818134765196014,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001524042,
      "loss": 3.2353,
      "step": 237980
    },
    {
      "epoch": 0.038199990888118325,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001524022,
      "loss": 3.1635,
      "step": 237990
    },
    {
      "epoch": 0.038218634124276514,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001524002,
      "loss": 3.1114,
      "step": 238000
    },
    {
      "epoch": 0.038237277360434696,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001523982,
      "loss": 3.1364,
      "step": 238010
    },
    {
      "epoch": 0.03825592059659288,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015239620000000002,
      "loss": 3.1527,
      "step": 238020
    },
    {
      "epoch": 0.03827456383275106,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001523942,
      "loss": 3.102,
      "step": 238030
    },
    {
      "epoch": 0.03829320706890925,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015239220000000001,
      "loss": 3.1076,
      "step": 238040
    },
    {
      "epoch": 0.03831185030506743,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001523902,
      "loss": 3.206,
      "step": 238050
    },
    {
      "epoch": 0.03833049354122561,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001523882,
      "loss": 3.1338,
      "step": 238060
    },
    {
      "epoch": 0.038349136777383795,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001523862,
      "loss": 3.145,
      "step": 238070
    },
    {
      "epoch": 0.03836778001354198,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001523842,
      "loss": 3.0371,
      "step": 238080
    },
    {
      "epoch": 0.038386423249700166,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015238220000000002,
      "loss": 3.2668,
      "step": 238090
    },
    {
      "epoch": 0.03840506648585835,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015238020000000002,
      "loss": 3.157,
      "step": 238100
    },
    {
      "epoch": 0.03842370972201653,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001523782,
      "loss": 3.1597,
      "step": 238110
    },
    {
      "epoch": 0.03844235295817471,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001523762,
      "loss": 3.0843,
      "step": 238120
    },
    {
      "epoch": 0.0384609961943329,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001523742,
      "loss": 3.1023,
      "step": 238130
    },
    {
      "epoch": 0.03847963943049108,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001523722,
      "loss": 3.343,
      "step": 238140
    },
    {
      "epoch": 0.038498282666649265,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001523702,
      "loss": 3.2029,
      "step": 238150
    },
    {
      "epoch": 0.03851692590280745,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015236820000000002,
      "loss": 3.0292,
      "step": 238160
    },
    {
      "epoch": 0.038535569138965636,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001523662,
      "loss": 3.05,
      "step": 238170
    },
    {
      "epoch": 0.03855421237512382,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001523642,
      "loss": 2.9306,
      "step": 238180
    },
    {
      "epoch": 0.038572855611282,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001523622,
      "loss": 3.2199,
      "step": 238190
    },
    {
      "epoch": 0.03859149884744018,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001523602,
      "loss": 3.1867,
      "step": 238200
    },
    {
      "epoch": 0.03861014208359837,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001523582,
      "loss": 2.9557,
      "step": 238210
    },
    {
      "epoch": 0.03862878531975655,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015235620000000002,
      "loss": 3.2227,
      "step": 238220
    },
    {
      "epoch": 0.038647428555914735,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001523542,
      "loss": 3.1749,
      "step": 238230
    },
    {
      "epoch": 0.03866607179207292,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001523522,
      "loss": 3.1397,
      "step": 238240
    },
    {
      "epoch": 0.038684715028231106,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001523502,
      "loss": 3.0754,
      "step": 238250
    },
    {
      "epoch": 0.03870335826438929,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001523482,
      "loss": 3.1176,
      "step": 238260
    },
    {
      "epoch": 0.03872200150054747,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015234620000000002,
      "loss": 3.174,
      "step": 238270
    },
    {
      "epoch": 0.03874064473670565,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001523442,
      "loss": 3.1758,
      "step": 238280
    },
    {
      "epoch": 0.038759287972863835,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001523422,
      "loss": 3.1904,
      "step": 238290
    },
    {
      "epoch": 0.038777931209022024,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001523402,
      "loss": 3.1988,
      "step": 238300
    },
    {
      "epoch": 0.038796574445180206,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001523382,
      "loss": 3.1708,
      "step": 238310
    },
    {
      "epoch": 0.03881521768133839,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015233620000000003,
      "loss": 3.1541,
      "step": 238320
    },
    {
      "epoch": 0.03883386091749657,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001523342,
      "loss": 3.1158,
      "step": 238330
    },
    {
      "epoch": 0.03885250415365476,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015233220000000002,
      "loss": 3.1136,
      "step": 238340
    },
    {
      "epoch": 0.03887114738981294,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015233020000000001,
      "loss": 3.2858,
      "step": 238350
    },
    {
      "epoch": 0.03888979062597112,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001523282,
      "loss": 3.1953,
      "step": 238360
    },
    {
      "epoch": 0.038908433862129305,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001523262,
      "loss": 3.0394,
      "step": 238370
    },
    {
      "epoch": 0.038927077098287494,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001523242,
      "loss": 3.0605,
      "step": 238380
    },
    {
      "epoch": 0.038945720334445676,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015232220000000002,
      "loss": 2.9904,
      "step": 238390
    },
    {
      "epoch": 0.03896436357060386,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001523202,
      "loss": 3.1078,
      "step": 238400
    },
    {
      "epoch": 0.03898300680676204,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015231820000000001,
      "loss": 3.1294,
      "step": 238410
    },
    {
      "epoch": 0.03900165004292023,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001523162,
      "loss": 3.1287,
      "step": 238420
    },
    {
      "epoch": 0.03902029327907841,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001523142,
      "loss": 3.1014,
      "step": 238430
    },
    {
      "epoch": 0.03903893651523659,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015231220000000003,
      "loss": 3.1256,
      "step": 238440
    },
    {
      "epoch": 0.039057579751394775,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001523102,
      "loss": 3.2711,
      "step": 238450
    },
    {
      "epoch": 0.03907622298755296,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015230820000000002,
      "loss": 3.1288,
      "step": 238460
    },
    {
      "epoch": 0.039094866223711146,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001523062,
      "loss": 3.2593,
      "step": 238470
    },
    {
      "epoch": 0.03911350945986933,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001523042,
      "loss": 3.1623,
      "step": 238480
    },
    {
      "epoch": 0.03913215269602751,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001523022,
      "loss": 3.2753,
      "step": 238490
    },
    {
      "epoch": 0.03915079593218569,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001523002,
      "loss": 3.1284,
      "step": 238500
    },
    {
      "epoch": 0.03916943916834388,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015229820000000003,
      "loss": 3.1708,
      "step": 238510
    },
    {
      "epoch": 0.03918808240450206,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001522962,
      "loss": 3.0417,
      "step": 238520
    },
    {
      "epoch": 0.039206725640660245,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015229420000000002,
      "loss": 3.1679,
      "step": 238530
    },
    {
      "epoch": 0.03922536887681843,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001522922,
      "loss": 3.1346,
      "step": 238540
    },
    {
      "epoch": 0.039244012112976616,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001522902,
      "loss": 3.0926,
      "step": 238550
    },
    {
      "epoch": 0.0392626553491348,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001522882,
      "loss": 3.1369,
      "step": 238560
    },
    {
      "epoch": 0.03928129858529298,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001522862,
      "loss": 3.1572,
      "step": 238570
    },
    {
      "epoch": 0.03929994182145116,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015228420000000002,
      "loss": 3.1245,
      "step": 238580
    },
    {
      "epoch": 0.03931858505760935,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001522822,
      "loss": 3.1796,
      "step": 238590
    },
    {
      "epoch": 0.03933722829376753,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001522802,
      "loss": 2.9998,
      "step": 238600
    },
    {
      "epoch": 0.039355871529925715,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001522782,
      "loss": 3.1253,
      "step": 238610
    },
    {
      "epoch": 0.0393745147660839,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001522762,
      "loss": 3.0738,
      "step": 238620
    },
    {
      "epoch": 0.03939315800224208,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001522742,
      "loss": 3.1967,
      "step": 238630
    },
    {
      "epoch": 0.03941180123840027,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001522722,
      "loss": 3.0942,
      "step": 238640
    },
    {
      "epoch": 0.03943044447455845,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015227020000000002,
      "loss": 3.1863,
      "step": 238650
    },
    {
      "epoch": 0.03944908771071663,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015226820000000001,
      "loss": 3.1004,
      "step": 238660
    },
    {
      "epoch": 0.039467730946874814,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001522662,
      "loss": 3.1223,
      "step": 238670
    },
    {
      "epoch": 0.039486374183033,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001522642,
      "loss": 3.236,
      "step": 238680
    },
    {
      "epoch": 0.039505017419191185,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001522622,
      "loss": 3.1562,
      "step": 238690
    },
    {
      "epoch": 0.03952366065534937,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001522602,
      "loss": 3.2003,
      "step": 238700
    },
    {
      "epoch": 0.03954230389150755,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015225820000000002,
      "loss": 3.1106,
      "step": 238710
    },
    {
      "epoch": 0.03956094712766574,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015225620000000001,
      "loss": 3.2126,
      "step": 238720
    },
    {
      "epoch": 0.03957959036382392,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001522542,
      "loss": 3.23,
      "step": 238730
    },
    {
      "epoch": 0.0395982335999821,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001522522,
      "loss": 3.0944,
      "step": 238740
    },
    {
      "epoch": 0.039616876836140284,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001522502,
      "loss": 3.0211,
      "step": 238750
    },
    {
      "epoch": 0.03963552007229847,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001522482,
      "loss": 3.1977,
      "step": 238760
    },
    {
      "epoch": 0.039654163308456655,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001522462,
      "loss": 3.0179,
      "step": 238770
    },
    {
      "epoch": 0.03967280654461484,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015224420000000002,
      "loss": 3.1107,
      "step": 238780
    },
    {
      "epoch": 0.03969144978077302,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001522422,
      "loss": 3.2239,
      "step": 238790
    },
    {
      "epoch": 0.0397100930169312,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001522402,
      "loss": 3.1891,
      "step": 238800
    },
    {
      "epoch": 0.03972873625308939,
      "grad_norm": 0.75,
      "learning_rate": 0.0001522382,
      "loss": 3.1273,
      "step": 238810
    },
    {
      "epoch": 0.03974737948924757,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001522362,
      "loss": 2.8796,
      "step": 238820
    },
    {
      "epoch": 0.039766022725405754,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015223420000000002,
      "loss": 3.0189,
      "step": 238830
    },
    {
      "epoch": 0.039784665961563936,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015223220000000002,
      "loss": 3.0464,
      "step": 238840
    },
    {
      "epoch": 0.039803309197722125,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001522302,
      "loss": 3.2493,
      "step": 238850
    },
    {
      "epoch": 0.03982195243388031,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001522282,
      "loss": 3.1872,
      "step": 238860
    },
    {
      "epoch": 0.03984059567003849,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001522262,
      "loss": 3.0365,
      "step": 238870
    },
    {
      "epoch": 0.03985923890619667,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015222420000000003,
      "loss": 3.0837,
      "step": 238880
    },
    {
      "epoch": 0.03987788214235486,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001522222,
      "loss": 3.0132,
      "step": 238890
    },
    {
      "epoch": 0.03989652537851304,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015222020000000002,
      "loss": 3.0585,
      "step": 238900
    },
    {
      "epoch": 0.039915168614671225,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001522182,
      "loss": 3.0496,
      "step": 238910
    },
    {
      "epoch": 0.03993381185082941,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001522162,
      "loss": 3.1253,
      "step": 238920
    },
    {
      "epoch": 0.039952455086987596,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015221420000000003,
      "loss": 3.1364,
      "step": 238930
    },
    {
      "epoch": 0.03997109832314578,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001522122,
      "loss": 3.1593,
      "step": 238940
    },
    {
      "epoch": 0.03998974155930396,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015221020000000002,
      "loss": 3.1726,
      "step": 238950
    },
    {
      "epoch": 0.04000838479546214,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001522082,
      "loss": 3.2063,
      "step": 238960
    },
    {
      "epoch": 0.040027028031620324,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001522062,
      "loss": 3.0651,
      "step": 238970
    },
    {
      "epoch": 0.04004567126777851,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001522042,
      "loss": 3.0969,
      "step": 238980
    },
    {
      "epoch": 0.040064314503936695,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001522022,
      "loss": 3.0268,
      "step": 238990
    },
    {
      "epoch": 0.04008295774009488,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015220020000000003,
      "loss": 3.124,
      "step": 239000
    },
    {
      "epoch": 0.04010160097625306,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001521982,
      "loss": 3.1285,
      "step": 239010
    },
    {
      "epoch": 0.04012024421241125,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00015219620000000002,
      "loss": 3.1115,
      "step": 239020
    },
    {
      "epoch": 0.04013888744856943,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001521942,
      "loss": 3.17,
      "step": 239030
    },
    {
      "epoch": 0.04015753068472761,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001521922,
      "loss": 3.1464,
      "step": 239040
    },
    {
      "epoch": 0.040176173920885794,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001521902,
      "loss": 3.1708,
      "step": 239050
    },
    {
      "epoch": 0.04019481715704398,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001521882,
      "loss": 3.0882,
      "step": 239060
    },
    {
      "epoch": 0.040213460393202165,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015218620000000002,
      "loss": 3.2281,
      "step": 239070
    },
    {
      "epoch": 0.04023210362936035,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001521842,
      "loss": 3.1067,
      "step": 239080
    },
    {
      "epoch": 0.04025074686551853,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015218220000000001,
      "loss": 3.1569,
      "step": 239090
    },
    {
      "epoch": 0.04026939010167672,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001521802,
      "loss": 3.1721,
      "step": 239100
    },
    {
      "epoch": 0.0402880333378349,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001521782,
      "loss": 3.2789,
      "step": 239110
    },
    {
      "epoch": 0.04030667657399308,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001521762,
      "loss": 3.1181,
      "step": 239120
    },
    {
      "epoch": 0.040325319810151264,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001521742,
      "loss": 3.0994,
      "step": 239130
    },
    {
      "epoch": 0.040343963046309446,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015217220000000002,
      "loss": 3.0831,
      "step": 239140
    },
    {
      "epoch": 0.040362606282467635,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015217020000000002,
      "loss": 3.1644,
      "step": 239150
    },
    {
      "epoch": 0.04038124951862582,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001521682,
      "loss": 3.1411,
      "step": 239160
    },
    {
      "epoch": 0.040399892754784,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001521662,
      "loss": 3.1066,
      "step": 239170
    },
    {
      "epoch": 0.04041853599094218,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001521642,
      "loss": 3.1068,
      "step": 239180
    },
    {
      "epoch": 0.04043717922710037,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001521622,
      "loss": 3.2455,
      "step": 239190
    },
    {
      "epoch": 0.04045582246325855,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001521602,
      "loss": 3.2234,
      "step": 239200
    },
    {
      "epoch": 0.040474465699416734,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015215820000000002,
      "loss": 3.1856,
      "step": 239210
    },
    {
      "epoch": 0.040493108935574916,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001521562,
      "loss": 3.2398,
      "step": 239220
    },
    {
      "epoch": 0.040511752171733105,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001521542,
      "loss": 2.9528,
      "step": 239230
    },
    {
      "epoch": 0.04053039540789129,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001521522,
      "loss": 3.0732,
      "step": 239240
    },
    {
      "epoch": 0.04054903864404947,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001521502,
      "loss": 3.124,
      "step": 239250
    },
    {
      "epoch": 0.04056768188020765,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001521482,
      "loss": 3.1875,
      "step": 239260
    },
    {
      "epoch": 0.04058632511636584,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015214620000000002,
      "loss": 3.2876,
      "step": 239270
    },
    {
      "epoch": 0.04060496835252402,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001521442,
      "loss": 3.1742,
      "step": 239280
    },
    {
      "epoch": 0.040623611588682204,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001521422,
      "loss": 3.0844,
      "step": 239290
    },
    {
      "epoch": 0.040642254824840386,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001521402,
      "loss": 3.2198,
      "step": 239300
    },
    {
      "epoch": 0.04066089806099857,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001521382,
      "loss": 3.1947,
      "step": 239310
    },
    {
      "epoch": 0.04067954129715676,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015213620000000002,
      "loss": 2.9739,
      "step": 239320
    },
    {
      "epoch": 0.04069818453331494,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015213420000000002,
      "loss": 3.1409,
      "step": 239330
    },
    {
      "epoch": 0.04071682776947312,
      "grad_norm": 0.75,
      "learning_rate": 0.00015213220000000001,
      "loss": 3.0937,
      "step": 239340
    },
    {
      "epoch": 0.0407354710056313,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001521302,
      "loss": 3.1211,
      "step": 239350
    },
    {
      "epoch": 0.04075411424178949,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001521282,
      "loss": 3.1777,
      "step": 239360
    },
    {
      "epoch": 0.040772757477947674,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001521262,
      "loss": 3.1085,
      "step": 239370
    },
    {
      "epoch": 0.040791400714105856,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001521242,
      "loss": 3.1548,
      "step": 239380
    },
    {
      "epoch": 0.04081004395026404,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015212220000000002,
      "loss": 3.1202,
      "step": 239390
    },
    {
      "epoch": 0.04082868718642223,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015212020000000001,
      "loss": 3.2251,
      "step": 239400
    },
    {
      "epoch": 0.04084733042258041,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001521182,
      "loss": 3.0325,
      "step": 239410
    },
    {
      "epoch": 0.04086597365873859,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001521162,
      "loss": 3.1219,
      "step": 239420
    },
    {
      "epoch": 0.04088461689489677,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001521142,
      "loss": 3.1513,
      "step": 239430
    },
    {
      "epoch": 0.04090326013105496,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015211220000000002,
      "loss": 3.0949,
      "step": 239440
    },
    {
      "epoch": 0.040921903367213144,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001521102,
      "loss": 3.126,
      "step": 239450
    },
    {
      "epoch": 0.040940546603371326,
      "grad_norm": 0.625,
      "learning_rate": 0.00015210820000000002,
      "loss": 3.1355,
      "step": 239460
    },
    {
      "epoch": 0.04095918983952951,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001521062,
      "loss": 3.1588,
      "step": 239470
    },
    {
      "epoch": 0.04097783307568769,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001521042,
      "loss": 3.2908,
      "step": 239480
    },
    {
      "epoch": 0.04099647631184588,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015210220000000003,
      "loss": 3.1872,
      "step": 239490
    },
    {
      "epoch": 0.04101511954800406,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001521002,
      "loss": 3.0762,
      "step": 239500
    },
    {
      "epoch": 0.041033762784162243,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015209820000000002,
      "loss": 3.226,
      "step": 239510
    },
    {
      "epoch": 0.041052406020320426,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001520962,
      "loss": 3.0469,
      "step": 239520
    },
    {
      "epoch": 0.041071049256478614,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001520942,
      "loss": 3.0302,
      "step": 239530
    },
    {
      "epoch": 0.041089692492636797,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001520922,
      "loss": 3.1467,
      "step": 239540
    },
    {
      "epoch": 0.04110833572879498,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001520902,
      "loss": 3.2803,
      "step": 239550
    },
    {
      "epoch": 0.04112697896495316,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015208820000000003,
      "loss": 3.1081,
      "step": 239560
    },
    {
      "epoch": 0.04114562220111135,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001520862,
      "loss": 3.2361,
      "step": 239570
    },
    {
      "epoch": 0.04116426543726953,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015208420000000002,
      "loss": 3.0425,
      "step": 239580
    },
    {
      "epoch": 0.041182908673427714,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015208219999999999,
      "loss": 3.1479,
      "step": 239590
    },
    {
      "epoch": 0.041201551909585896,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001520802,
      "loss": 3.1606,
      "step": 239600
    },
    {
      "epoch": 0.041220195145744085,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001520782,
      "loss": 3.3058,
      "step": 239610
    },
    {
      "epoch": 0.04123883838190227,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001520762,
      "loss": 3.068,
      "step": 239620
    },
    {
      "epoch": 0.04125748161806045,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015207420000000002,
      "loss": 3.112,
      "step": 239630
    },
    {
      "epoch": 0.04127612485421863,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001520722,
      "loss": 3.0499,
      "step": 239640
    },
    {
      "epoch": 0.04129476809037681,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001520702,
      "loss": 3.1776,
      "step": 239650
    },
    {
      "epoch": 0.041313411326535,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001520682,
      "loss": 3.2255,
      "step": 239660
    },
    {
      "epoch": 0.041332054562693184,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001520662,
      "loss": 3.006,
      "step": 239670
    },
    {
      "epoch": 0.041350697798851366,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001520642,
      "loss": 3.1802,
      "step": 239680
    },
    {
      "epoch": 0.04136934103500955,
      "grad_norm": 0.625,
      "learning_rate": 0.0001520622,
      "loss": 3.2238,
      "step": 239690
    },
    {
      "epoch": 0.04138798427116774,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015206020000000002,
      "loss": 3.2017,
      "step": 239700
    },
    {
      "epoch": 0.04140662750732592,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015205820000000001,
      "loss": 3.1443,
      "step": 239710
    },
    {
      "epoch": 0.0414252707434841,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001520562,
      "loss": 3.0403,
      "step": 239720
    },
    {
      "epoch": 0.04144391397964228,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001520542,
      "loss": 3.2523,
      "step": 239730
    },
    {
      "epoch": 0.04146255721580047,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001520522,
      "loss": 3.1632,
      "step": 239740
    },
    {
      "epoch": 0.041481200451958654,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001520502,
      "loss": 3.241,
      "step": 239750
    },
    {
      "epoch": 0.041499843688116836,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001520482,
      "loss": 3.0881,
      "step": 239760
    },
    {
      "epoch": 0.04151848692427502,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015204620000000001,
      "loss": 3.1931,
      "step": 239770
    },
    {
      "epoch": 0.04153713016043321,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001520442,
      "loss": 3.0111,
      "step": 239780
    },
    {
      "epoch": 0.04155577339659139,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001520422,
      "loss": 3.188,
      "step": 239790
    },
    {
      "epoch": 0.04157441663274957,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001520402,
      "loss": 3.1115,
      "step": 239800
    },
    {
      "epoch": 0.04159305986890775,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001520382,
      "loss": 3.1461,
      "step": 239810
    },
    {
      "epoch": 0.041611703105065935,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001520362,
      "loss": 3.1169,
      "step": 239820
    },
    {
      "epoch": 0.041630346341224124,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015203420000000002,
      "loss": 3.1438,
      "step": 239830
    },
    {
      "epoch": 0.041648989577382306,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001520322,
      "loss": 3.1343,
      "step": 239840
    },
    {
      "epoch": 0.04166763281354049,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001520302,
      "loss": 3.174,
      "step": 239850
    },
    {
      "epoch": 0.04168627604969867,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001520282,
      "loss": 3.0598,
      "step": 239860
    },
    {
      "epoch": 0.04170491928585686,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001520262,
      "loss": 3.1541,
      "step": 239870
    },
    {
      "epoch": 0.04172356252201504,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015202420000000002,
      "loss": 3.0728,
      "step": 239880
    },
    {
      "epoch": 0.04174220575817322,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015202220000000002,
      "loss": 3.1178,
      "step": 239890
    },
    {
      "epoch": 0.041760848994331405,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001520202,
      "loss": 3.1821,
      "step": 239900
    },
    {
      "epoch": 0.041779492230489594,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001520182,
      "loss": 3.1261,
      "step": 239910
    },
    {
      "epoch": 0.041798135466647776,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001520162,
      "loss": 3.1487,
      "step": 239920
    },
    {
      "epoch": 0.04181677870280596,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015201420000000003,
      "loss": 3.2004,
      "step": 239930
    },
    {
      "epoch": 0.04183542193896414,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001520122,
      "loss": 3.1373,
      "step": 239940
    },
    {
      "epoch": 0.04185406517512233,
      "grad_norm": 0.75,
      "learning_rate": 0.00015201020000000002,
      "loss": 3.1036,
      "step": 239950
    },
    {
      "epoch": 0.04187270841128051,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001520082,
      "loss": 3.1838,
      "step": 239960
    },
    {
      "epoch": 0.04189135164743869,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001520062,
      "loss": 3.0342,
      "step": 239970
    },
    {
      "epoch": 0.041909994883596875,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001520042,
      "loss": 3.1355,
      "step": 239980
    },
    {
      "epoch": 0.04192863811975506,
      "grad_norm": 0.75,
      "learning_rate": 0.0001520022,
      "loss": 3.0764,
      "step": 239990
    },
    {
      "epoch": 0.041947281355913246,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015200020000000002,
      "loss": 3.1274,
      "step": 240000
    },
    {
      "epoch": 0.04196592459207143,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001519982,
      "loss": 3.1324,
      "step": 240010
    },
    {
      "epoch": 0.04198456782822961,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015199620000000001,
      "loss": 3.1489,
      "step": 240020
    },
    {
      "epoch": 0.04200321106438779,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001519942,
      "loss": 3.151,
      "step": 240030
    },
    {
      "epoch": 0.04202185430054598,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001519922,
      "loss": 3.0758,
      "step": 240040
    },
    {
      "epoch": 0.04204049753670416,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00015199020000000003,
      "loss": 3.168,
      "step": 240050
    },
    {
      "epoch": 0.042059140772862345,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001519882,
      "loss": 3.0949,
      "step": 240060
    },
    {
      "epoch": 0.04207778400902053,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015198620000000002,
      "loss": 3.2201,
      "step": 240070
    },
    {
      "epoch": 0.042096427245178716,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001519842,
      "loss": 3.2462,
      "step": 240080
    },
    {
      "epoch": 0.0421150704813369,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001519822,
      "loss": 3.0813,
      "step": 240090
    },
    {
      "epoch": 0.04213371371749508,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001519802,
      "loss": 3.3151,
      "step": 240100
    },
    {
      "epoch": 0.04215235695365326,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001519782,
      "loss": 3.2461,
      "step": 240110
    },
    {
      "epoch": 0.04217100018981145,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015197620000000002,
      "loss": 3.0577,
      "step": 240120
    },
    {
      "epoch": 0.04218964342596963,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001519742,
      "loss": 3.2654,
      "step": 240130
    },
    {
      "epoch": 0.042208286662127815,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015197220000000002,
      "loss": 3.1667,
      "step": 240140
    },
    {
      "epoch": 0.042226929898286,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001519702,
      "loss": 3.099,
      "step": 240150
    },
    {
      "epoch": 0.04224557313444418,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001519682,
      "loss": 3.1729,
      "step": 240160
    },
    {
      "epoch": 0.04226421637060237,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001519662,
      "loss": 3.0839,
      "step": 240170
    },
    {
      "epoch": 0.04228285960676055,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001519642,
      "loss": 3.1457,
      "step": 240180
    },
    {
      "epoch": 0.04230150284291873,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015196220000000002,
      "loss": 3.0403,
      "step": 240190
    },
    {
      "epoch": 0.042320146079076915,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001519602,
      "loss": 3.2114,
      "step": 240200
    },
    {
      "epoch": 0.042338789315235104,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001519582,
      "loss": 3.171,
      "step": 240210
    },
    {
      "epoch": 0.042357432551393286,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001519562,
      "loss": 3.0794,
      "step": 240220
    },
    {
      "epoch": 0.04237607578755147,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001519542,
      "loss": 3.1441,
      "step": 240230
    },
    {
      "epoch": 0.04239471902370965,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001519522,
      "loss": 3.1316,
      "step": 240240
    },
    {
      "epoch": 0.04241336225986784,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001519502,
      "loss": 3.1079,
      "step": 240250
    },
    {
      "epoch": 0.04243200549602602,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015194820000000002,
      "loss": 3.2752,
      "step": 240260
    },
    {
      "epoch": 0.0424506487321842,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001519462,
      "loss": 3.0959,
      "step": 240270
    },
    {
      "epoch": 0.042469291968342385,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001519442,
      "loss": 3.0782,
      "step": 240280
    },
    {
      "epoch": 0.042487935204500574,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001519422,
      "loss": 3.0651,
      "step": 240290
    },
    {
      "epoch": 0.042506578440658756,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001519402,
      "loss": 3.0878,
      "step": 240300
    },
    {
      "epoch": 0.04252522167681694,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001519382,
      "loss": 3.1645,
      "step": 240310
    },
    {
      "epoch": 0.04254386491297512,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015193620000000002,
      "loss": 3.1114,
      "step": 240320
    },
    {
      "epoch": 0.0425625081491333,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001519342,
      "loss": 3.2271,
      "step": 240330
    },
    {
      "epoch": 0.04258115138529149,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001519322,
      "loss": 3.0563,
      "step": 240340
    },
    {
      "epoch": 0.04259979462144967,
      "grad_norm": 0.75,
      "learning_rate": 0.0001519302,
      "loss": 3.2033,
      "step": 240350
    },
    {
      "epoch": 0.042618437857607855,
      "grad_norm": 0.75,
      "learning_rate": 0.0001519282,
      "loss": 3.256,
      "step": 240360
    },
    {
      "epoch": 0.04263708109376604,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001519262,
      "loss": 3.0397,
      "step": 240370
    },
    {
      "epoch": 0.042655724329924226,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015192420000000002,
      "loss": 3.1371,
      "step": 240380
    },
    {
      "epoch": 0.04267436756608241,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015192220000000001,
      "loss": 3.1814,
      "step": 240390
    },
    {
      "epoch": 0.04269301080224059,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001519202,
      "loss": 3.0476,
      "step": 240400
    },
    {
      "epoch": 0.04271165403839877,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001519182,
      "loss": 3.1408,
      "step": 240410
    },
    {
      "epoch": 0.04273029727455696,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001519162,
      "loss": 3.0469,
      "step": 240420
    },
    {
      "epoch": 0.04274894051071514,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001519142,
      "loss": 3.1489,
      "step": 240430
    },
    {
      "epoch": 0.042767583746873325,
      "grad_norm": 0.75,
      "learning_rate": 0.00015191220000000002,
      "loss": 3.1375,
      "step": 240440
    },
    {
      "epoch": 0.04278622698303151,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015191020000000001,
      "loss": 3.202,
      "step": 240450
    },
    {
      "epoch": 0.042804870219189696,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001519082,
      "loss": 3.0574,
      "step": 240460
    },
    {
      "epoch": 0.04282351345534788,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001519062,
      "loss": 3.1139,
      "step": 240470
    },
    {
      "epoch": 0.04284215669150606,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001519042,
      "loss": 3.1754,
      "step": 240480
    },
    {
      "epoch": 0.04286079992766424,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015190220000000002,
      "loss": 3.1323,
      "step": 240490
    },
    {
      "epoch": 0.042879443163822424,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001519002,
      "loss": 3.0924,
      "step": 240500
    },
    {
      "epoch": 0.04289808639998061,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015189820000000002,
      "loss": 3.1334,
      "step": 240510
    },
    {
      "epoch": 0.042916729636138795,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001518962,
      "loss": 3.1434,
      "step": 240520
    },
    {
      "epoch": 0.04293537287229698,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001518942,
      "loss": 3.0711,
      "step": 240530
    },
    {
      "epoch": 0.04295401610845516,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015189220000000003,
      "loss": 3.1298,
      "step": 240540
    },
    {
      "epoch": 0.04297265934461335,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001518902,
      "loss": 3.1738,
      "step": 240550
    },
    {
      "epoch": 0.04299130258077153,
      "grad_norm": 0.75,
      "learning_rate": 0.00015188820000000002,
      "loss": 3.0371,
      "step": 240560
    },
    {
      "epoch": 0.04300994581692971,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001518862,
      "loss": 3.1547,
      "step": 240570
    },
    {
      "epoch": 0.043028589053087894,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001518842,
      "loss": 3.1563,
      "step": 240580
    },
    {
      "epoch": 0.04304723228924608,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001518822,
      "loss": 3.1071,
      "step": 240590
    },
    {
      "epoch": 0.043065875525404265,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001518802,
      "loss": 3.1158,
      "step": 240600
    },
    {
      "epoch": 0.04308451876156245,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015187820000000003,
      "loss": 3.0867,
      "step": 240610
    },
    {
      "epoch": 0.04310316199772063,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001518762,
      "loss": 3.1355,
      "step": 240620
    },
    {
      "epoch": 0.04312180523387882,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015187420000000002,
      "loss": 3.0567,
      "step": 240630
    },
    {
      "epoch": 0.043140448470037,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015187219999999999,
      "loss": 3.1841,
      "step": 240640
    },
    {
      "epoch": 0.04315909170619518,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001518702,
      "loss": 3.0883,
      "step": 240650
    },
    {
      "epoch": 0.043177734942353364,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001518682,
      "loss": 3.1557,
      "step": 240660
    },
    {
      "epoch": 0.043196378178511546,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001518662,
      "loss": 3.1628,
      "step": 240670
    },
    {
      "epoch": 0.043215021414669735,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015186420000000002,
      "loss": 3.1177,
      "step": 240680
    },
    {
      "epoch": 0.04323366465082792,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001518622,
      "loss": 2.9686,
      "step": 240690
    },
    {
      "epoch": 0.0432523078869861,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001518602,
      "loss": 2.9941,
      "step": 240700
    },
    {
      "epoch": 0.04327095112314428,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001518582,
      "loss": 3.1735,
      "step": 240710
    },
    {
      "epoch": 0.04328959435930247,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001518562,
      "loss": 3.0773,
      "step": 240720
    },
    {
      "epoch": 0.04330823759546065,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001518542,
      "loss": 3.0493,
      "step": 240730
    },
    {
      "epoch": 0.043326880831618834,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001518522,
      "loss": 3.1051,
      "step": 240740
    },
    {
      "epoch": 0.043345524067777016,
      "grad_norm": 0.75,
      "learning_rate": 0.00015185020000000002,
      "loss": 2.9595,
      "step": 240750
    },
    {
      "epoch": 0.043364167303935205,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015184820000000001,
      "loss": 3.0267,
      "step": 240760
    },
    {
      "epoch": 0.04338281054009339,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001518462,
      "loss": 3.0875,
      "step": 240770
    },
    {
      "epoch": 0.04340145377625157,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001518442,
      "loss": 3.13,
      "step": 240780
    },
    {
      "epoch": 0.04342009701240975,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001518422,
      "loss": 3.0684,
      "step": 240790
    },
    {
      "epoch": 0.04343874024856794,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001518402,
      "loss": 3.1159,
      "step": 240800
    },
    {
      "epoch": 0.04345738348472612,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001518382,
      "loss": 3.1617,
      "step": 240810
    },
    {
      "epoch": 0.043476026720884305,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015183620000000002,
      "loss": 3.1279,
      "step": 240820
    },
    {
      "epoch": 0.04349466995704249,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001518342,
      "loss": 3.0845,
      "step": 240830
    },
    {
      "epoch": 0.04351331319320067,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001518322,
      "loss": 3.1727,
      "step": 240840
    },
    {
      "epoch": 0.04353195642935886,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001518302,
      "loss": 3.0624,
      "step": 240850
    },
    {
      "epoch": 0.04355059966551704,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001518282,
      "loss": 3.1592,
      "step": 240860
    },
    {
      "epoch": 0.04356924290167522,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015182620000000002,
      "loss": 3.1607,
      "step": 240870
    },
    {
      "epoch": 0.043587886137833404,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015182420000000002,
      "loss": 3.0453,
      "step": 240880
    },
    {
      "epoch": 0.04360652937399159,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001518222,
      "loss": 3.2103,
      "step": 240890
    },
    {
      "epoch": 0.043625172610149775,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001518202,
      "loss": 3.1134,
      "step": 240900
    },
    {
      "epoch": 0.04364381584630796,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001518182,
      "loss": 3.0711,
      "step": 240910
    },
    {
      "epoch": 0.04366245908246614,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001518162,
      "loss": 3.0562,
      "step": 240920
    },
    {
      "epoch": 0.04368110231862433,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015181420000000002,
      "loss": 3.1695,
      "step": 240930
    },
    {
      "epoch": 0.04369974555478251,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015181220000000002,
      "loss": 3.0889,
      "step": 240940
    },
    {
      "epoch": 0.04371838879094069,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001518102,
      "loss": 3.2227,
      "step": 240950
    },
    {
      "epoch": 0.043737032027098874,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001518082,
      "loss": 3.1952,
      "step": 240960
    },
    {
      "epoch": 0.04375567526325706,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001518062,
      "loss": 3.1724,
      "step": 240970
    },
    {
      "epoch": 0.043774318499415245,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001518042,
      "loss": 3.1322,
      "step": 240980
    },
    {
      "epoch": 0.04379296173557343,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001518022,
      "loss": 3.2198,
      "step": 240990
    },
    {
      "epoch": 0.04381160497173161,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015180020000000002,
      "loss": 3.1077,
      "step": 241000
    },
    {
      "epoch": 0.04383024820788979,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001517982,
      "loss": 3.1125,
      "step": 241010
    },
    {
      "epoch": 0.04384889144404798,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001517962,
      "loss": 3.0941,
      "step": 241020
    },
    {
      "epoch": 0.04386753468020616,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001517942,
      "loss": 3.1346,
      "step": 241030
    },
    {
      "epoch": 0.043886177916364344,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001517922,
      "loss": 3.1181,
      "step": 241040
    },
    {
      "epoch": 0.043904821152522526,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015179020000000002,
      "loss": 3.2233,
      "step": 241050
    },
    {
      "epoch": 0.043923464388680715,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001517882,
      "loss": 3.1395,
      "step": 241060
    },
    {
      "epoch": 0.0439421076248389,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015178620000000001,
      "loss": 3.2177,
      "step": 241070
    },
    {
      "epoch": 0.04396075086099708,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001517842,
      "loss": 3.0968,
      "step": 241080
    },
    {
      "epoch": 0.04397939409715526,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001517822,
      "loss": 3.2327,
      "step": 241090
    },
    {
      "epoch": 0.04399803733331345,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015178020000000003,
      "loss": 3.1761,
      "step": 241100
    },
    {
      "epoch": 0.04401668056947163,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001517782,
      "loss": 3.2252,
      "step": 241110
    },
    {
      "epoch": 0.044035323805629814,
      "grad_norm": 0.625,
      "learning_rate": 0.00015177620000000002,
      "loss": 3.3467,
      "step": 241120
    },
    {
      "epoch": 0.044053967041787996,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001517742,
      "loss": 3.1887,
      "step": 241130
    },
    {
      "epoch": 0.044072610277946185,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001517722,
      "loss": 3.0606,
      "step": 241140
    },
    {
      "epoch": 0.04409125351410437,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001517702,
      "loss": 3.0596,
      "step": 241150
    },
    {
      "epoch": 0.04410989675026255,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001517682,
      "loss": 3.2224,
      "step": 241160
    },
    {
      "epoch": 0.04412853998642073,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015176620000000002,
      "loss": 3.2694,
      "step": 241170
    },
    {
      "epoch": 0.04414718322257891,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001517642,
      "loss": 3.2324,
      "step": 241180
    },
    {
      "epoch": 0.0441658264587371,
      "grad_norm": 0.75,
      "learning_rate": 0.00015176220000000002,
      "loss": 3.1091,
      "step": 241190
    },
    {
      "epoch": 0.044184469694895284,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015176019999999998,
      "loss": 3.1947,
      "step": 241200
    },
    {
      "epoch": 0.044203112931053466,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001517582,
      "loss": 3.1613,
      "step": 241210
    },
    {
      "epoch": 0.04422175616721165,
      "grad_norm": 0.75,
      "learning_rate": 0.0001517562,
      "loss": 3.0853,
      "step": 241220
    },
    {
      "epoch": 0.04424039940336984,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001517542,
      "loss": 3.0723,
      "step": 241230
    },
    {
      "epoch": 0.04425904263952802,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015175220000000002,
      "loss": 3.202,
      "step": 241240
    },
    {
      "epoch": 0.0442776858756862,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001517502,
      "loss": 3.1395,
      "step": 241250
    },
    {
      "epoch": 0.04429632911184438,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001517482,
      "loss": 3.1118,
      "step": 241260
    },
    {
      "epoch": 0.04431497234800257,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001517462,
      "loss": 3.1039,
      "step": 241270
    },
    {
      "epoch": 0.044333615584160754,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001517442,
      "loss": 3.0329,
      "step": 241280
    },
    {
      "epoch": 0.044352258820318936,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001517422,
      "loss": 3.2397,
      "step": 241290
    },
    {
      "epoch": 0.04437090205647712,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001517402,
      "loss": 3.1034,
      "step": 241300
    },
    {
      "epoch": 0.04438954529263531,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015173820000000002,
      "loss": 3.2003,
      "step": 241310
    },
    {
      "epoch": 0.04440818852879349,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001517362,
      "loss": 3.049,
      "step": 241320
    },
    {
      "epoch": 0.04442683176495167,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001517342,
      "loss": 3.0202,
      "step": 241330
    },
    {
      "epoch": 0.04444547500110985,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001517322,
      "loss": 3.1326,
      "step": 241340
    },
    {
      "epoch": 0.044464118237268035,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001517302,
      "loss": 3.1157,
      "step": 241350
    },
    {
      "epoch": 0.044482761473426224,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015172820000000002,
      "loss": 3.1703,
      "step": 241360
    },
    {
      "epoch": 0.044501404709584406,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015172620000000002,
      "loss": 3.2315,
      "step": 241370
    },
    {
      "epoch": 0.04452004794574259,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001517242,
      "loss": 3.1704,
      "step": 241380
    },
    {
      "epoch": 0.04453869118190077,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001517222,
      "loss": 3.1343,
      "step": 241390
    },
    {
      "epoch": 0.04455733441805896,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001517202,
      "loss": 3.1853,
      "step": 241400
    },
    {
      "epoch": 0.04457597765421714,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001517182,
      "loss": 3.2352,
      "step": 241410
    },
    {
      "epoch": 0.044594620890375324,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001517162,
      "loss": 3.1527,
      "step": 241420
    },
    {
      "epoch": 0.044613264126533506,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015171420000000002,
      "loss": 3.1228,
      "step": 241430
    },
    {
      "epoch": 0.044631907362691695,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015171220000000001,
      "loss": 3.1421,
      "step": 241440
    },
    {
      "epoch": 0.04465055059884988,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001517102,
      "loss": 3.2734,
      "step": 241450
    },
    {
      "epoch": 0.04466919383500806,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001517082,
      "loss": 3.0434,
      "step": 241460
    },
    {
      "epoch": 0.04468783707116624,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001517062,
      "loss": 3.0893,
      "step": 241470
    },
    {
      "epoch": 0.04470648030732443,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001517042,
      "loss": 3.2026,
      "step": 241480
    },
    {
      "epoch": 0.04472512354348261,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015170220000000002,
      "loss": 3.1268,
      "step": 241490
    },
    {
      "epoch": 0.044743766779640794,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015170020000000002,
      "loss": 3.1926,
      "step": 241500
    },
    {
      "epoch": 0.044762410015798976,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001516982,
      "loss": 3.2259,
      "step": 241510
    },
    {
      "epoch": 0.04478105325195716,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001516962,
      "loss": 3.1292,
      "step": 241520
    },
    {
      "epoch": 0.04479969648811535,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001516942,
      "loss": 3.0732,
      "step": 241530
    },
    {
      "epoch": 0.04481833972427353,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015169220000000002,
      "loss": 3.3144,
      "step": 241540
    },
    {
      "epoch": 0.04483698296043171,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001516902,
      "loss": 3.1008,
      "step": 241550
    },
    {
      "epoch": 0.04485562619658989,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015168820000000002,
      "loss": 3.1852,
      "step": 241560
    },
    {
      "epoch": 0.04487426943274808,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001516862,
      "loss": 3.0867,
      "step": 241570
    },
    {
      "epoch": 0.044892912668906264,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001516842,
      "loss": 3.1335,
      "step": 241580
    },
    {
      "epoch": 0.044911555905064446,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001516822,
      "loss": 3.1327,
      "step": 241590
    },
    {
      "epoch": 0.04493019914122263,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001516802,
      "loss": 3.1357,
      "step": 241600
    },
    {
      "epoch": 0.04494884237738082,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015167820000000002,
      "loss": 3.0494,
      "step": 241610
    },
    {
      "epoch": 0.044967485613539,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001516762,
      "loss": 3.1639,
      "step": 241620
    },
    {
      "epoch": 0.04498612884969718,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001516742,
      "loss": 3.2422,
      "step": 241630
    },
    {
      "epoch": 0.04500477208585536,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001516722,
      "loss": 3.1735,
      "step": 241640
    },
    {
      "epoch": 0.04502341532201355,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001516702,
      "loss": 3.0232,
      "step": 241650
    },
    {
      "epoch": 0.045042058558171734,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015166820000000003,
      "loss": 3.1444,
      "step": 241660
    },
    {
      "epoch": 0.045060701794329916,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001516662,
      "loss": 3.2152,
      "step": 241670
    },
    {
      "epoch": 0.0450793450304881,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015166420000000002,
      "loss": 3.2256,
      "step": 241680
    },
    {
      "epoch": 0.04509798826664628,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015166219999999999,
      "loss": 3.1206,
      "step": 241690
    },
    {
      "epoch": 0.04511663150280447,
      "grad_norm": 0.75,
      "learning_rate": 0.0001516602,
      "loss": 3.1875,
      "step": 241700
    },
    {
      "epoch": 0.04513527473896265,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001516582,
      "loss": 3.2014,
      "step": 241710
    },
    {
      "epoch": 0.04515391797512083,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001516562,
      "loss": 3.1656,
      "step": 241720
    },
    {
      "epoch": 0.045172561211279015,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015165420000000002,
      "loss": 3.1835,
      "step": 241730
    },
    {
      "epoch": 0.045191204447437204,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001516522,
      "loss": 3.1869,
      "step": 241740
    },
    {
      "epoch": 0.045209847683595386,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015165020000000001,
      "loss": 3.2032,
      "step": 241750
    },
    {
      "epoch": 0.04522849091975357,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001516482,
      "loss": 3.1648,
      "step": 241760
    },
    {
      "epoch": 0.04524713415591175,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001516462,
      "loss": 3.0847,
      "step": 241770
    },
    {
      "epoch": 0.04526577739206994,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001516442,
      "loss": 3.1701,
      "step": 241780
    },
    {
      "epoch": 0.04528442062822812,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001516422,
      "loss": 3.0774,
      "step": 241790
    },
    {
      "epoch": 0.0453030638643863,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015164020000000002,
      "loss": 3.1356,
      "step": 241800
    },
    {
      "epoch": 0.045321707100544485,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001516382,
      "loss": 3.1161,
      "step": 241810
    },
    {
      "epoch": 0.045340350336702674,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001516362,
      "loss": 2.9535,
      "step": 241820
    },
    {
      "epoch": 0.045358993572860856,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001516342,
      "loss": 3.216,
      "step": 241830
    },
    {
      "epoch": 0.04537763680901904,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001516322,
      "loss": 3.1708,
      "step": 241840
    },
    {
      "epoch": 0.04539628004517722,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015163020000000002,
      "loss": 3.0534,
      "step": 241850
    },
    {
      "epoch": 0.0454149232813354,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001516282,
      "loss": 3.0666,
      "step": 241860
    },
    {
      "epoch": 0.04543356651749359,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015162620000000002,
      "loss": 3.0341,
      "step": 241870
    },
    {
      "epoch": 0.04545220975365177,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001516242,
      "loss": 3.0747,
      "step": 241880
    },
    {
      "epoch": 0.045470852989809955,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001516222,
      "loss": 3.1583,
      "step": 241890
    },
    {
      "epoch": 0.04548949622596814,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001516202,
      "loss": 3.1883,
      "step": 241900
    },
    {
      "epoch": 0.045508139462126326,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001516182,
      "loss": 3.0877,
      "step": 241910
    },
    {
      "epoch": 0.04552678269828451,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015161620000000002,
      "loss": 3.1562,
      "step": 241920
    },
    {
      "epoch": 0.04554542593444269,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015161420000000002,
      "loss": 3.2101,
      "step": 241930
    },
    {
      "epoch": 0.04556406917060087,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001516122,
      "loss": 2.9898,
      "step": 241940
    },
    {
      "epoch": 0.04558271240675906,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001516102,
      "loss": 3.234,
      "step": 241950
    },
    {
      "epoch": 0.04560135564291724,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001516082,
      "loss": 3.1583,
      "step": 241960
    },
    {
      "epoch": 0.045619998879075425,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001516062,
      "loss": 3.1537,
      "step": 241970
    },
    {
      "epoch": 0.04563864211523361,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001516042,
      "loss": 3.0363,
      "step": 241980
    },
    {
      "epoch": 0.045657285351391796,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015160220000000002,
      "loss": 2.9784,
      "step": 241990
    },
    {
      "epoch": 0.04567592858754998,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001516002,
      "loss": 3.1596,
      "step": 242000
    },
    {
      "epoch": 0.04569457182370816,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001515982,
      "loss": 3.1459,
      "step": 242010
    },
    {
      "epoch": 0.04571321505986634,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001515962,
      "loss": 3.1148,
      "step": 242020
    },
    {
      "epoch": 0.045731858296024525,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001515942,
      "loss": 2.995,
      "step": 242030
    },
    {
      "epoch": 0.045750501532182714,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001515922,
      "loss": 3.0605,
      "step": 242040
    },
    {
      "epoch": 0.045769144768340896,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015159020000000002,
      "loss": 3.0489,
      "step": 242050
    },
    {
      "epoch": 0.04578778800449908,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001515882,
      "loss": 3.0395,
      "step": 242060
    },
    {
      "epoch": 0.04580643124065726,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001515862,
      "loss": 3.1064,
      "step": 242070
    },
    {
      "epoch": 0.04582507447681545,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001515842,
      "loss": 3.1581,
      "step": 242080
    },
    {
      "epoch": 0.04584371771297363,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001515822,
      "loss": 3.1936,
      "step": 242090
    },
    {
      "epoch": 0.04586236094913181,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015158020000000002,
      "loss": 3.2485,
      "step": 242100
    },
    {
      "epoch": 0.045881004185289995,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001515782,
      "loss": 3.2137,
      "step": 242110
    },
    {
      "epoch": 0.045899647421448184,
      "grad_norm": 0.625,
      "learning_rate": 0.00015157620000000001,
      "loss": 3.1325,
      "step": 242120
    },
    {
      "epoch": 0.045918290657606366,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001515742,
      "loss": 3.1665,
      "step": 242130
    },
    {
      "epoch": 0.04593693389376455,
      "grad_norm": 0.75,
      "learning_rate": 0.0001515722,
      "loss": 3.1919,
      "step": 242140
    },
    {
      "epoch": 0.04595557712992273,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015157020000000003,
      "loss": 3.2164,
      "step": 242150
    },
    {
      "epoch": 0.04597422036608092,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001515682,
      "loss": 3.2389,
      "step": 242160
    },
    {
      "epoch": 0.0459928636022391,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015156620000000002,
      "loss": 3.0394,
      "step": 242170
    },
    {
      "epoch": 0.04601150683839728,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001515642,
      "loss": 3.145,
      "step": 242180
    },
    {
      "epoch": 0.046030150074555465,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001515622,
      "loss": 3.0957,
      "step": 242190
    },
    {
      "epoch": 0.04604879331071365,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001515602,
      "loss": 3.1394,
      "step": 242200
    },
    {
      "epoch": 0.046067436546871836,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001515582,
      "loss": 3.1072,
      "step": 242210
    },
    {
      "epoch": 0.04608607978303002,
      "grad_norm": 0.75,
      "learning_rate": 0.00015155620000000002,
      "loss": 3.1389,
      "step": 242220
    },
    {
      "epoch": 0.0461047230191882,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001515542,
      "loss": 3.1011,
      "step": 242230
    },
    {
      "epoch": 0.04612336625534638,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015155220000000002,
      "loss": 3.1687,
      "step": 242240
    },
    {
      "epoch": 0.04614200949150457,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015155019999999998,
      "loss": 3.1457,
      "step": 242250
    },
    {
      "epoch": 0.04616065272766275,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001515482,
      "loss": 3.188,
      "step": 242260
    },
    {
      "epoch": 0.046179295963820935,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001515462,
      "loss": 3.1465,
      "step": 242270
    },
    {
      "epoch": 0.04619793919997912,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001515442,
      "loss": 3.0684,
      "step": 242280
    },
    {
      "epoch": 0.046216582436137306,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015154220000000002,
      "loss": 3.2022,
      "step": 242290
    },
    {
      "epoch": 0.04623522567229549,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001515402,
      "loss": 3.1703,
      "step": 242300
    },
    {
      "epoch": 0.04625386890845367,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001515382,
      "loss": 3.1194,
      "step": 242310
    },
    {
      "epoch": 0.04627251214461185,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001515362,
      "loss": 3.095,
      "step": 242320
    },
    {
      "epoch": 0.04629115538077004,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001515342,
      "loss": 3.1235,
      "step": 242330
    },
    {
      "epoch": 0.04630979861692822,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015153220000000003,
      "loss": 3.1488,
      "step": 242340
    },
    {
      "epoch": 0.046328441853086405,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001515302,
      "loss": 3.0974,
      "step": 242350
    },
    {
      "epoch": 0.04634708508924459,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015152820000000002,
      "loss": 3.1559,
      "step": 242360
    },
    {
      "epoch": 0.04636572832540277,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001515262,
      "loss": 3.1677,
      "step": 242370
    },
    {
      "epoch": 0.04638437156156096,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001515242,
      "loss": 3.0096,
      "step": 242380
    },
    {
      "epoch": 0.04640301479771914,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001515222,
      "loss": 3.0144,
      "step": 242390
    },
    {
      "epoch": 0.04642165803387732,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001515202,
      "loss": 3.1156,
      "step": 242400
    },
    {
      "epoch": 0.046440301270035504,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015151820000000002,
      "loss": 3.0448,
      "step": 242410
    },
    {
      "epoch": 0.04645894450619369,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001515162,
      "loss": 3.134,
      "step": 242420
    },
    {
      "epoch": 0.046477587742351875,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015151420000000001,
      "loss": 3.062,
      "step": 242430
    },
    {
      "epoch": 0.04649623097851006,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001515122,
      "loss": 3.2434,
      "step": 242440
    },
    {
      "epoch": 0.04651487421466824,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001515102,
      "loss": 3.0123,
      "step": 242450
    },
    {
      "epoch": 0.04653351745082643,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001515082,
      "loss": 3.0203,
      "step": 242460
    },
    {
      "epoch": 0.04655216068698461,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001515062,
      "loss": 3.1173,
      "step": 242470
    },
    {
      "epoch": 0.04657080392314279,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015150420000000002,
      "loss": 2.963,
      "step": 242480
    },
    {
      "epoch": 0.046589447159300974,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015150220000000001,
      "loss": 3.1461,
      "step": 242490
    },
    {
      "epoch": 0.04660809039545916,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001515002,
      "loss": 3.0927,
      "step": 242500
    },
    {
      "epoch": 0.046626733631617345,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001514982,
      "loss": 3.0803,
      "step": 242510
    },
    {
      "epoch": 0.04664537686777553,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001514962,
      "loss": 2.9706,
      "step": 242520
    },
    {
      "epoch": 0.04666402010393371,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001514942,
      "loss": 3.1161,
      "step": 242530
    },
    {
      "epoch": 0.04668266334009189,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015149220000000002,
      "loss": 3.1144,
      "step": 242540
    },
    {
      "epoch": 0.04670130657625008,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015149020000000002,
      "loss": 3.0159,
      "step": 242550
    },
    {
      "epoch": 0.04671994981240826,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001514882,
      "loss": 3.1528,
      "step": 242560
    },
    {
      "epoch": 0.046738593048566444,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001514862,
      "loss": 3.1739,
      "step": 242570
    },
    {
      "epoch": 0.046757236284724626,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001514842,
      "loss": 3.028,
      "step": 242580
    },
    {
      "epoch": 0.046775879520882815,
      "grad_norm": 0.625,
      "learning_rate": 0.0001514822,
      "loss": 3.2113,
      "step": 242590
    },
    {
      "epoch": 0.046794522757041,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001514802,
      "loss": 3.1878,
      "step": 242600
    },
    {
      "epoch": 0.04681316599319918,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015147820000000002,
      "loss": 3.2071,
      "step": 242610
    },
    {
      "epoch": 0.04683180922935736,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001514762,
      "loss": 3.177,
      "step": 242620
    },
    {
      "epoch": 0.04685045246551555,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001514742,
      "loss": 3.1171,
      "step": 242630
    },
    {
      "epoch": 0.04686909570167373,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001514722,
      "loss": 2.9606,
      "step": 242640
    },
    {
      "epoch": 0.046887738937831915,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001514702,
      "loss": 3.1981,
      "step": 242650
    },
    {
      "epoch": 0.0469063821739901,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015146820000000002,
      "loss": 3.179,
      "step": 242660
    },
    {
      "epoch": 0.046925025410148286,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001514662,
      "loss": 3.1743,
      "step": 242670
    },
    {
      "epoch": 0.04694366864630647,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001514642,
      "loss": 3.0583,
      "step": 242680
    },
    {
      "epoch": 0.04696231188246465,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001514622,
      "loss": 3.0832,
      "step": 242690
    },
    {
      "epoch": 0.04698095511862283,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001514602,
      "loss": 3.1494,
      "step": 242700
    },
    {
      "epoch": 0.046999598354781014,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015145820000000003,
      "loss": 3.272,
      "step": 242710
    },
    {
      "epoch": 0.0470182415909392,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001514562,
      "loss": 3.1235,
      "step": 242720
    },
    {
      "epoch": 0.047036884827097385,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015145420000000002,
      "loss": 3.1373,
      "step": 242730
    },
    {
      "epoch": 0.04705552806325557,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015145219999999999,
      "loss": 3.0467,
      "step": 242740
    },
    {
      "epoch": 0.04707417129941375,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001514502,
      "loss": 3.1563,
      "step": 242750
    },
    {
      "epoch": 0.04709281453557194,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001514482,
      "loss": 3.0763,
      "step": 242760
    },
    {
      "epoch": 0.04711145777173012,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001514462,
      "loss": 2.9896,
      "step": 242770
    },
    {
      "epoch": 0.0471301010078883,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015144420000000002,
      "loss": 3.0921,
      "step": 242780
    },
    {
      "epoch": 0.047148744244046484,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001514422,
      "loss": 2.9749,
      "step": 242790
    },
    {
      "epoch": 0.04716738748020467,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015144020000000001,
      "loss": 3.0883,
      "step": 242800
    },
    {
      "epoch": 0.047186030716362855,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001514382,
      "loss": 3.0891,
      "step": 242810
    },
    {
      "epoch": 0.04720467395252104,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001514362,
      "loss": 3.1289,
      "step": 242820
    },
    {
      "epoch": 0.04722331718867922,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015143420000000003,
      "loss": 3.1266,
      "step": 242830
    },
    {
      "epoch": 0.04724196042483741,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001514322,
      "loss": 3.2771,
      "step": 242840
    },
    {
      "epoch": 0.04726060366099559,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015143020000000002,
      "loss": 3.1108,
      "step": 242850
    },
    {
      "epoch": 0.04727924689715377,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001514282,
      "loss": 3.1154,
      "step": 242860
    },
    {
      "epoch": 0.047297890133311954,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001514262,
      "loss": 3.1824,
      "step": 242870
    },
    {
      "epoch": 0.04731653336947014,
      "grad_norm": 0.75,
      "learning_rate": 0.0001514242,
      "loss": 3.0871,
      "step": 242880
    },
    {
      "epoch": 0.047335176605628325,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001514222,
      "loss": 3.2247,
      "step": 242890
    },
    {
      "epoch": 0.04735381984178651,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015142020000000002,
      "loss": 3.2036,
      "step": 242900
    },
    {
      "epoch": 0.04737246307794469,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001514182,
      "loss": 3.1198,
      "step": 242910
    },
    {
      "epoch": 0.04739110631410287,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015141620000000002,
      "loss": 3.1417,
      "step": 242920
    },
    {
      "epoch": 0.04740974955026106,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001514142,
      "loss": 3.1657,
      "step": 242930
    },
    {
      "epoch": 0.04742839278641924,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001514122,
      "loss": 3.1049,
      "step": 242940
    },
    {
      "epoch": 0.047447036022577424,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001514102,
      "loss": 3.0128,
      "step": 242950
    },
    {
      "epoch": 0.047465679258735606,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001514082,
      "loss": 3.1347,
      "step": 242960
    },
    {
      "epoch": 0.047484322494893795,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015140620000000002,
      "loss": 3.0752,
      "step": 242970
    },
    {
      "epoch": 0.04750296573105198,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015140420000000002,
      "loss": 3.1278,
      "step": 242980
    },
    {
      "epoch": 0.04752160896721016,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001514022,
      "loss": 3.1011,
      "step": 242990
    },
    {
      "epoch": 0.04754025220336834,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001514002,
      "loss": 3.1774,
      "step": 243000
    },
    {
      "epoch": 0.04755889543952653,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001513982,
      "loss": 3.1161,
      "step": 243010
    },
    {
      "epoch": 0.04757753867568471,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001513962,
      "loss": 3.0643,
      "step": 243020
    },
    {
      "epoch": 0.047596181911842894,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001513942,
      "loss": 3.1448,
      "step": 243030
    },
    {
      "epoch": 0.047614825148001076,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015139220000000002,
      "loss": 3.1414,
      "step": 243040
    },
    {
      "epoch": 0.047633468384159265,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001513902,
      "loss": 3.1625,
      "step": 243050
    },
    {
      "epoch": 0.04765211162031745,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001513882,
      "loss": 3.1054,
      "step": 243060
    },
    {
      "epoch": 0.04767075485647563,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001513862,
      "loss": 3.1947,
      "step": 243070
    },
    {
      "epoch": 0.04768939809263381,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001513842,
      "loss": 3.1604,
      "step": 243080
    },
    {
      "epoch": 0.04770804132879199,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001513822,
      "loss": 3.0043,
      "step": 243090
    },
    {
      "epoch": 0.04772668456495018,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015138020000000002,
      "loss": 3.1093,
      "step": 243100
    },
    {
      "epoch": 0.047745327801108364,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015137820000000001,
      "loss": 2.9745,
      "step": 243110
    },
    {
      "epoch": 0.047763971037266546,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001513762,
      "loss": 2.9979,
      "step": 243120
    },
    {
      "epoch": 0.04778261427342473,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001513742,
      "loss": 3.1517,
      "step": 243130
    },
    {
      "epoch": 0.04780125750958292,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001513722,
      "loss": 3.3,
      "step": 243140
    },
    {
      "epoch": 0.0478199007457411,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015137020000000002,
      "loss": 3.0761,
      "step": 243150
    },
    {
      "epoch": 0.04783854398189928,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001513682,
      "loss": 3.1001,
      "step": 243160
    },
    {
      "epoch": 0.04785718721805746,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015136620000000001,
      "loss": 3.2003,
      "step": 243170
    },
    {
      "epoch": 0.04787583045421565,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001513642,
      "loss": 3.1057,
      "step": 243180
    },
    {
      "epoch": 0.047894473690373834,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001513622,
      "loss": 3.2068,
      "step": 243190
    },
    {
      "epoch": 0.047913116926532016,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001513602,
      "loss": 3.0733,
      "step": 243200
    },
    {
      "epoch": 0.0479317601626902,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001513582,
      "loss": 3.103,
      "step": 243210
    },
    {
      "epoch": 0.04795040339884839,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015135620000000002,
      "loss": 3.1534,
      "step": 243220
    },
    {
      "epoch": 0.04796904663500657,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001513542,
      "loss": 3.1251,
      "step": 243230
    },
    {
      "epoch": 0.04798768987116475,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001513522,
      "loss": 2.9948,
      "step": 243240
    },
    {
      "epoch": 0.048006333107322934,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001513502,
      "loss": 3.1161,
      "step": 243250
    },
    {
      "epoch": 0.048024976343481116,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001513482,
      "loss": 3.149,
      "step": 243260
    },
    {
      "epoch": 0.048043619579639305,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015134620000000002,
      "loss": 3.1169,
      "step": 243270
    },
    {
      "epoch": 0.04806226281579749,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001513442,
      "loss": 3.1386,
      "step": 243280
    },
    {
      "epoch": 0.04808090605195567,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015134220000000002,
      "loss": 3.0871,
      "step": 243290
    },
    {
      "epoch": 0.04809954928811385,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001513402,
      "loss": 3.1515,
      "step": 243300
    },
    {
      "epoch": 0.04811819252427204,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001513382,
      "loss": 3.1145,
      "step": 243310
    },
    {
      "epoch": 0.04813683576043022,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015133620000000003,
      "loss": 3.1524,
      "step": 243320
    },
    {
      "epoch": 0.048155478996588404,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001513342,
      "loss": 3.134,
      "step": 243330
    },
    {
      "epoch": 0.048174122232746586,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015133220000000002,
      "loss": 3.0264,
      "step": 243340
    },
    {
      "epoch": 0.048192765468904775,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001513302,
      "loss": 3.1824,
      "step": 243350
    },
    {
      "epoch": 0.04821140870506296,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001513282,
      "loss": 3.1278,
      "step": 243360
    },
    {
      "epoch": 0.04823005194122114,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001513262,
      "loss": 3.0326,
      "step": 243370
    },
    {
      "epoch": 0.04824869517737932,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001513242,
      "loss": 3.0836,
      "step": 243380
    },
    {
      "epoch": 0.04826733841353751,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015132220000000003,
      "loss": 3.2206,
      "step": 243390
    },
    {
      "epoch": 0.04828598164969569,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001513202,
      "loss": 3.1669,
      "step": 243400
    },
    {
      "epoch": 0.048304624885853874,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015131820000000002,
      "loss": 3.1935,
      "step": 243410
    },
    {
      "epoch": 0.048323268122012056,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015131619999999999,
      "loss": 3.2224,
      "step": 243420
    },
    {
      "epoch": 0.04834191135817024,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001513142,
      "loss": 3.154,
      "step": 243430
    },
    {
      "epoch": 0.04836055459432843,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001513122,
      "loss": 3.113,
      "step": 243440
    },
    {
      "epoch": 0.04837919783048661,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001513102,
      "loss": 3.2658,
      "step": 243450
    },
    {
      "epoch": 0.04839784106664479,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015130820000000002,
      "loss": 3.0447,
      "step": 243460
    },
    {
      "epoch": 0.04841648430280297,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001513062,
      "loss": 3.1246,
      "step": 243470
    },
    {
      "epoch": 0.04843512753896116,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015130420000000001,
      "loss": 2.99,
      "step": 243480
    },
    {
      "epoch": 0.048453770775119344,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001513022,
      "loss": 3.0539,
      "step": 243490
    },
    {
      "epoch": 0.048472414011277526,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001513002,
      "loss": 3.1651,
      "step": 243500
    },
    {
      "epoch": 0.04849105724743571,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001512982,
      "loss": 3.1199,
      "step": 243510
    },
    {
      "epoch": 0.0485097004835939,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001512962,
      "loss": 3.074,
      "step": 243520
    },
    {
      "epoch": 0.04852834371975208,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015129420000000002,
      "loss": 3.0371,
      "step": 243530
    },
    {
      "epoch": 0.04854698695591026,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015129220000000001,
      "loss": 3.0831,
      "step": 243540
    },
    {
      "epoch": 0.04856563019206844,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001512902,
      "loss": 3.2401,
      "step": 243550
    },
    {
      "epoch": 0.04858427342822663,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001512882,
      "loss": 3.1867,
      "step": 243560
    },
    {
      "epoch": 0.048602916664384814,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001512862,
      "loss": 3.1567,
      "step": 243570
    },
    {
      "epoch": 0.048621559900542996,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001512842,
      "loss": 3.0452,
      "step": 243580
    },
    {
      "epoch": 0.04864020313670118,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001512822,
      "loss": 3.1937,
      "step": 243590
    },
    {
      "epoch": 0.04865884637285936,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015128020000000002,
      "loss": 3.1787,
      "step": 243600
    },
    {
      "epoch": 0.04867748960901755,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001512782,
      "loss": 3.1582,
      "step": 243610
    },
    {
      "epoch": 0.04869613284517573,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001512762,
      "loss": 3.1564,
      "step": 243620
    },
    {
      "epoch": 0.04871477608133391,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001512742,
      "loss": 3.0925,
      "step": 243630
    },
    {
      "epoch": 0.048733419317492095,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001512722,
      "loss": 3.1134,
      "step": 243640
    },
    {
      "epoch": 0.048752062553650284,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001512702,
      "loss": 3.1584,
      "step": 243650
    },
    {
      "epoch": 0.048770705789808466,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015126820000000002,
      "loss": 2.9586,
      "step": 243660
    },
    {
      "epoch": 0.04878934902596665,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001512662,
      "loss": 3.1946,
      "step": 243670
    },
    {
      "epoch": 0.04880799226212483,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001512642,
      "loss": 3.2939,
      "step": 243680
    },
    {
      "epoch": 0.04882663549828302,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001512622,
      "loss": 3.0278,
      "step": 243690
    },
    {
      "epoch": 0.0488452787344412,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001512602,
      "loss": 3.1594,
      "step": 243700
    },
    {
      "epoch": 0.04886392197059938,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015125820000000002,
      "loss": 3.0868,
      "step": 243710
    },
    {
      "epoch": 0.048882565206757565,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001512562,
      "loss": 2.983,
      "step": 243720
    },
    {
      "epoch": 0.048901208442915754,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001512542,
      "loss": 3.269,
      "step": 243730
    },
    {
      "epoch": 0.048919851679073936,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001512522,
      "loss": 2.9987,
      "step": 243740
    },
    {
      "epoch": 0.04893849491523212,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001512502,
      "loss": 3.0861,
      "step": 243750
    },
    {
      "epoch": 0.0489571381513903,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015124820000000003,
      "loss": 3.1693,
      "step": 243760
    },
    {
      "epoch": 0.04897578138754848,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001512462,
      "loss": 3.1076,
      "step": 243770
    },
    {
      "epoch": 0.04899442462370667,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015124420000000002,
      "loss": 3.0084,
      "step": 243780
    },
    {
      "epoch": 0.04901306785986485,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015124220000000001,
      "loss": 3.1275,
      "step": 243790
    },
    {
      "epoch": 0.049031711096023035,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001512402,
      "loss": 3.2086,
      "step": 243800
    },
    {
      "epoch": 0.04905035433218122,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001512382,
      "loss": 3.2465,
      "step": 243810
    },
    {
      "epoch": 0.049068997568339406,
      "grad_norm": 0.75,
      "learning_rate": 0.0001512362,
      "loss": 3.0293,
      "step": 243820
    },
    {
      "epoch": 0.04908764080449759,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015123420000000002,
      "loss": 3.1819,
      "step": 243830
    },
    {
      "epoch": 0.04910628404065577,
      "grad_norm": 0.75,
      "learning_rate": 0.0001512322,
      "loss": 3.0491,
      "step": 243840
    },
    {
      "epoch": 0.04912492727681395,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015123020000000001,
      "loss": 3.0451,
      "step": 243850
    },
    {
      "epoch": 0.04914357051297214,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001512282,
      "loss": 3.1537,
      "step": 243860
    },
    {
      "epoch": 0.049162213749130323,
      "grad_norm": 0.75,
      "learning_rate": 0.0001512262,
      "loss": 3.1404,
      "step": 243870
    },
    {
      "epoch": 0.049180856985288506,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015122420000000003,
      "loss": 3.1757,
      "step": 243880
    },
    {
      "epoch": 0.04919950022144669,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001512222,
      "loss": 3.1663,
      "step": 243890
    },
    {
      "epoch": 0.049218143457604877,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015122020000000002,
      "loss": 3.2087,
      "step": 243900
    },
    {
      "epoch": 0.04923678669376306,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001512182,
      "loss": 3.094,
      "step": 243910
    },
    {
      "epoch": 0.04925542992992124,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001512162,
      "loss": 3.1109,
      "step": 243920
    },
    {
      "epoch": 0.04927407316607942,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001512142,
      "loss": 3.0824,
      "step": 243930
    },
    {
      "epoch": 0.049292716402237605,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001512122,
      "loss": 2.8726,
      "step": 243940
    },
    {
      "epoch": 0.049311359638395794,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015121020000000002,
      "loss": 3.2032,
      "step": 243950
    },
    {
      "epoch": 0.049330002874553976,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001512082,
      "loss": 3.21,
      "step": 243960
    },
    {
      "epoch": 0.04934864611071216,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015120620000000002,
      "loss": 3.0612,
      "step": 243970
    },
    {
      "epoch": 0.04936728934687034,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001512042,
      "loss": 3.0686,
      "step": 243980
    },
    {
      "epoch": 0.04938593258302853,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001512022,
      "loss": 3.2715,
      "step": 243990
    },
    {
      "epoch": 0.04940457581918671,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001512002,
      "loss": 3.1363,
      "step": 244000
    },
    {
      "epoch": 0.04942321905534489,
      "grad_norm": 0.875,
      "learning_rate": 0.0001511982,
      "loss": 3.136,
      "step": 244010
    },
    {
      "epoch": 0.049441862291503075,
      "grad_norm": 0.625,
      "learning_rate": 0.00015119620000000002,
      "loss": 3.1831,
      "step": 244020
    },
    {
      "epoch": 0.049460505527661264,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001511942,
      "loss": 3.1928,
      "step": 244030
    },
    {
      "epoch": 0.049479148763819446,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001511922,
      "loss": 3.2125,
      "step": 244040
    },
    {
      "epoch": 0.04949779199997763,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001511902,
      "loss": 3.1592,
      "step": 244050
    },
    {
      "epoch": 0.04951643523613581,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001511882,
      "loss": 3.2056,
      "step": 244060
    },
    {
      "epoch": 0.049535078472294,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001511862,
      "loss": 3.0649,
      "step": 244070
    },
    {
      "epoch": 0.04955372170845218,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001511842,
      "loss": 3.0792,
      "step": 244080
    },
    {
      "epoch": 0.04957236494461036,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015118220000000002,
      "loss": 3.0759,
      "step": 244090
    },
    {
      "epoch": 0.049591008180768545,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001511802,
      "loss": 3.1777,
      "step": 244100
    },
    {
      "epoch": 0.04960965141692673,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001511782,
      "loss": 3.125,
      "step": 244110
    },
    {
      "epoch": 0.049628294653084916,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001511762,
      "loss": 3.158,
      "step": 244120
    },
    {
      "epoch": 0.0496469378892431,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001511742,
      "loss": 3.2477,
      "step": 244130
    },
    {
      "epoch": 0.04966558112540128,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001511722,
      "loss": 3.2296,
      "step": 244140
    },
    {
      "epoch": 0.04968422436155946,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015117020000000002,
      "loss": 3.1448,
      "step": 244150
    },
    {
      "epoch": 0.04970286759771765,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015116820000000001,
      "loss": 3.1933,
      "step": 244160
    },
    {
      "epoch": 0.04972151083387583,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001511662,
      "loss": 3.2454,
      "step": 244170
    },
    {
      "epoch": 0.049740154070034015,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001511642,
      "loss": 3.1925,
      "step": 244180
    },
    {
      "epoch": 0.0497587973061922,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001511622,
      "loss": 3.2023,
      "step": 244190
    },
    {
      "epoch": 0.049777440542350386,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001511602,
      "loss": 3.1554,
      "step": 244200
    },
    {
      "epoch": 0.04979608377850857,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001511582,
      "loss": 3.039,
      "step": 244210
    },
    {
      "epoch": 0.04981472701466675,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015115620000000001,
      "loss": 3.1141,
      "step": 244220
    },
    {
      "epoch": 0.04983337025082493,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001511542,
      "loss": 3.1385,
      "step": 244230
    },
    {
      "epoch": 0.04985201348698312,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001511522,
      "loss": 3.1267,
      "step": 244240
    },
    {
      "epoch": 0.0498706567231413,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001511502,
      "loss": 3.1134,
      "step": 244250
    },
    {
      "epoch": 0.049889299959299485,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001511482,
      "loss": 3.0394,
      "step": 244260
    },
    {
      "epoch": 0.04990794319545767,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015114620000000002,
      "loss": 3.1925,
      "step": 244270
    },
    {
      "epoch": 0.04992658643161585,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015114420000000002,
      "loss": 3.058,
      "step": 244280
    },
    {
      "epoch": 0.04994522966777404,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001511422,
      "loss": 3.1936,
      "step": 244290
    },
    {
      "epoch": 0.04996387290393222,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001511402,
      "loss": 3.215,
      "step": 244300
    },
    {
      "epoch": 0.0499825161400904,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001511382,
      "loss": 3.187,
      "step": 244310
    },
    {
      "epoch": 0.050001159376248584,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015113620000000003,
      "loss": 3.2457,
      "step": 244320
    },
    {
      "epoch": 0.05001980261240677,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001511342,
      "loss": 3.0547,
      "step": 244330
    },
    {
      "epoch": 0.050038445848564955,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015113220000000002,
      "loss": 3.266,
      "step": 244340
    },
    {
      "epoch": 0.05005708908472314,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001511302,
      "loss": 3.0025,
      "step": 244350
    },
    {
      "epoch": 0.05007573232088132,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001511282,
      "loss": 3.0417,
      "step": 244360
    },
    {
      "epoch": 0.05009437555703951,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015112620000000003,
      "loss": 3.1515,
      "step": 244370
    },
    {
      "epoch": 0.05011301879319769,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001511242,
      "loss": 3.093,
      "step": 244380
    },
    {
      "epoch": 0.05013166202935587,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015112220000000002,
      "loss": 3.2291,
      "step": 244390
    },
    {
      "epoch": 0.050150305265514054,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001511202,
      "loss": 3.1412,
      "step": 244400
    },
    {
      "epoch": 0.05016894850167224,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001511182,
      "loss": 3.1324,
      "step": 244410
    },
    {
      "epoch": 0.050187591737830425,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001511162,
      "loss": 3.1212,
      "step": 244420
    },
    {
      "epoch": 0.05020623497398861,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001511142,
      "loss": 3.2822,
      "step": 244430
    },
    {
      "epoch": 0.05022487821014679,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015111220000000003,
      "loss": 3.1046,
      "step": 244440
    },
    {
      "epoch": 0.05024352144630497,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001511102,
      "loss": 3.0424,
      "step": 244450
    },
    {
      "epoch": 0.05026216468246316,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015110820000000002,
      "loss": 3.0474,
      "step": 244460
    },
    {
      "epoch": 0.05028080791862134,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001511062,
      "loss": 3.0568,
      "step": 244470
    },
    {
      "epoch": 0.050299451154779524,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001511042,
      "loss": 3.1626,
      "step": 244480
    },
    {
      "epoch": 0.050318094390937707,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001511022,
      "loss": 3.1661,
      "step": 244490
    },
    {
      "epoch": 0.050336737627095895,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001511002,
      "loss": 3.0434,
      "step": 244500
    },
    {
      "epoch": 0.05035538086325408,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015109820000000002,
      "loss": 3.2201,
      "step": 244510
    },
    {
      "epoch": 0.05037402409941226,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001510962,
      "loss": 3.1865,
      "step": 244520
    },
    {
      "epoch": 0.05039266733557044,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015109420000000001,
      "loss": 3.1853,
      "step": 244530
    },
    {
      "epoch": 0.05041131057172863,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001510922,
      "loss": 3.1438,
      "step": 244540
    },
    {
      "epoch": 0.05042995380788681,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001510902,
      "loss": 3.0369,
      "step": 244550
    },
    {
      "epoch": 0.050448597044044995,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001510882,
      "loss": 3.1172,
      "step": 244560
    },
    {
      "epoch": 0.05046724028020318,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001510862,
      "loss": 3.1526,
      "step": 244570
    },
    {
      "epoch": 0.050485883516361366,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015108420000000002,
      "loss": 3.0715,
      "step": 244580
    },
    {
      "epoch": 0.05050452675251955,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015108220000000002,
      "loss": 3.0323,
      "step": 244590
    },
    {
      "epoch": 0.05052316998867773,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001510802,
      "loss": 3.1316,
      "step": 244600
    },
    {
      "epoch": 0.05054181322483591,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001510782,
      "loss": 3.2655,
      "step": 244610
    },
    {
      "epoch": 0.050560456460994094,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001510762,
      "loss": 3.1607,
      "step": 244620
    },
    {
      "epoch": 0.05057909969715228,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001510742,
      "loss": 3.2261,
      "step": 244630
    },
    {
      "epoch": 0.050597742933310465,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001510722,
      "loss": 3.0493,
      "step": 244640
    },
    {
      "epoch": 0.05061638616946865,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015107020000000002,
      "loss": 2.9646,
      "step": 244650
    },
    {
      "epoch": 0.05063502940562683,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001510682,
      "loss": 3.0796,
      "step": 244660
    },
    {
      "epoch": 0.05065367264178502,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001510662,
      "loss": 3.223,
      "step": 244670
    },
    {
      "epoch": 0.0506723158779432,
      "grad_norm": 0.75,
      "learning_rate": 0.0001510642,
      "loss": 3.0987,
      "step": 244680
    },
    {
      "epoch": 0.05069095911410138,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001510622,
      "loss": 3.1457,
      "step": 244690
    },
    {
      "epoch": 0.050709602350259564,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001510602,
      "loss": 3.0834,
      "step": 244700
    },
    {
      "epoch": 0.05072824558641775,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015105820000000002,
      "loss": 3.0351,
      "step": 244710
    },
    {
      "epoch": 0.050746888822575935,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001510562,
      "loss": 3.1701,
      "step": 244720
    },
    {
      "epoch": 0.05076553205873412,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001510542,
      "loss": 3.1387,
      "step": 244730
    },
    {
      "epoch": 0.0507841752948923,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001510522,
      "loss": 3.2192,
      "step": 244740
    },
    {
      "epoch": 0.05080281853105049,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001510502,
      "loss": 3.0898,
      "step": 244750
    },
    {
      "epoch": 0.05082146176720867,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015104820000000002,
      "loss": 3.152,
      "step": 244760
    },
    {
      "epoch": 0.05084010500336685,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015104620000000002,
      "loss": 3.1135,
      "step": 244770
    },
    {
      "epoch": 0.050858748239525034,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001510442,
      "loss": 3.0961,
      "step": 244780
    },
    {
      "epoch": 0.050877391475683216,
      "grad_norm": 0.75,
      "learning_rate": 0.0001510422,
      "loss": 3.0792,
      "step": 244790
    },
    {
      "epoch": 0.050896034711841405,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001510402,
      "loss": 3.2703,
      "step": 244800
    },
    {
      "epoch": 0.05091467794799959,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001510382,
      "loss": 3.0948,
      "step": 244810
    },
    {
      "epoch": 0.05093332118415777,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001510362,
      "loss": 3.2028,
      "step": 244820
    },
    {
      "epoch": 0.05095196442031595,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015103420000000002,
      "loss": 3.1939,
      "step": 244830
    },
    {
      "epoch": 0.05097060765647414,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015103220000000001,
      "loss": 3.1567,
      "step": 244840
    },
    {
      "epoch": 0.05098925089263232,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001510302,
      "loss": 3.1203,
      "step": 244850
    },
    {
      "epoch": 0.051007894128790504,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001510282,
      "loss": 3.1241,
      "step": 244860
    },
    {
      "epoch": 0.051026537364948686,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001510262,
      "loss": 3.0201,
      "step": 244870
    },
    {
      "epoch": 0.051045180601106875,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015102420000000002,
      "loss": 3.2569,
      "step": 244880
    },
    {
      "epoch": 0.05106382383726506,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001510222,
      "loss": 3.165,
      "step": 244890
    },
    {
      "epoch": 0.05108246707342324,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015102020000000001,
      "loss": 3.1735,
      "step": 244900
    },
    {
      "epoch": 0.05110111030958142,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001510182,
      "loss": 3.1975,
      "step": 244910
    },
    {
      "epoch": 0.05111975354573961,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001510162,
      "loss": 3.1794,
      "step": 244920
    },
    {
      "epoch": 0.05113839678189779,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015101420000000003,
      "loss": 3.1871,
      "step": 244930
    },
    {
      "epoch": 0.051157040018055974,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001510122,
      "loss": 3.1512,
      "step": 244940
    },
    {
      "epoch": 0.051175683254214156,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015101020000000002,
      "loss": 3.1057,
      "step": 244950
    },
    {
      "epoch": 0.05119432649037234,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001510082,
      "loss": 3.0136,
      "step": 244960
    },
    {
      "epoch": 0.05121296972653053,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001510062,
      "loss": 3.0463,
      "step": 244970
    },
    {
      "epoch": 0.05123161296268871,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001510042,
      "loss": 3.1618,
      "step": 244980
    },
    {
      "epoch": 0.05125025619884689,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001510022,
      "loss": 3.1618,
      "step": 244990
    },
    {
      "epoch": 0.05126889943500507,
      "grad_norm": 0.75,
      "learning_rate": 0.00015100020000000003,
      "loss": 3.0521,
      "step": 245000
    },
    {
      "epoch": 0.05128754267116326,
      "grad_norm": 0.75,
      "learning_rate": 0.0001509982,
      "loss": 3.095,
      "step": 245010
    },
    {
      "epoch": 0.051306185907321444,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015099620000000002,
      "loss": 3.16,
      "step": 245020
    },
    {
      "epoch": 0.051324829143479626,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015099419999999998,
      "loss": 3.1737,
      "step": 245030
    },
    {
      "epoch": 0.05134347237963781,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001509922,
      "loss": 3.208,
      "step": 245040
    },
    {
      "epoch": 0.051362115615796,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001509902,
      "loss": 3.1253,
      "step": 245050
    },
    {
      "epoch": 0.05138075885195418,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001509882,
      "loss": 3.2039,
      "step": 245060
    },
    {
      "epoch": 0.05139940208811236,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015098620000000002,
      "loss": 3.1526,
      "step": 245070
    },
    {
      "epoch": 0.051418045324270543,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001509842,
      "loss": 3.0399,
      "step": 245080
    },
    {
      "epoch": 0.05143668856042873,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001509822,
      "loss": 3.2417,
      "step": 245090
    },
    {
      "epoch": 0.051455331796586914,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001509802,
      "loss": 3.094,
      "step": 245100
    },
    {
      "epoch": 0.051473975032745097,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001509782,
      "loss": 3.1364,
      "step": 245110
    },
    {
      "epoch": 0.05149261826890328,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001509762,
      "loss": 2.9787,
      "step": 245120
    },
    {
      "epoch": 0.05151126150506146,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001509742,
      "loss": 3.2723,
      "step": 245130
    },
    {
      "epoch": 0.05152990474121965,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015097220000000002,
      "loss": 3.1029,
      "step": 245140
    },
    {
      "epoch": 0.05154854797737783,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015097020000000001,
      "loss": 3.1154,
      "step": 245150
    },
    {
      "epoch": 0.051567191213536014,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001509682,
      "loss": 3.1549,
      "step": 245160
    },
    {
      "epoch": 0.051585834449694196,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001509662,
      "loss": 2.9909,
      "step": 245170
    },
    {
      "epoch": 0.051604477685852385,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001509642,
      "loss": 3.0089,
      "step": 245180
    },
    {
      "epoch": 0.05162312092201057,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001509622,
      "loss": 3.0536,
      "step": 245190
    },
    {
      "epoch": 0.05164176415816875,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015096020000000002,
      "loss": 3.1757,
      "step": 245200
    },
    {
      "epoch": 0.05166040739432693,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015095820000000001,
      "loss": 3.1088,
      "step": 245210
    },
    {
      "epoch": 0.05167905063048512,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001509562,
      "loss": 3.0671,
      "step": 245220
    },
    {
      "epoch": 0.0516976938666433,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001509542,
      "loss": 3.1972,
      "step": 245230
    },
    {
      "epoch": 0.051716337102801484,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001509522,
      "loss": 3.1179,
      "step": 245240
    },
    {
      "epoch": 0.051734980338959666,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001509502,
      "loss": 3.1682,
      "step": 245250
    },
    {
      "epoch": 0.051753623575117855,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015094820000000002,
      "loss": 3.0755,
      "step": 245260
    },
    {
      "epoch": 0.05177226681127604,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015094620000000002,
      "loss": 3.0684,
      "step": 245270
    },
    {
      "epoch": 0.05179091004743422,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001509442,
      "loss": 3.2072,
      "step": 245280
    },
    {
      "epoch": 0.0518095532835924,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001509422,
      "loss": 3.0985,
      "step": 245290
    },
    {
      "epoch": 0.05182819651975058,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001509402,
      "loss": 3.0971,
      "step": 245300
    },
    {
      "epoch": 0.05184683975590877,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001509382,
      "loss": 3.1659,
      "step": 245310
    },
    {
      "epoch": 0.051865482992066954,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015093620000000002,
      "loss": 3.0751,
      "step": 245320
    },
    {
      "epoch": 0.051884126228225136,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015093420000000002,
      "loss": 3.2216,
      "step": 245330
    },
    {
      "epoch": 0.05190276946438332,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001509322,
      "loss": 3.0915,
      "step": 245340
    },
    {
      "epoch": 0.05192141270054151,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001509302,
      "loss": 3.1207,
      "step": 245350
    },
    {
      "epoch": 0.05194005593669969,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001509282,
      "loss": 3.1511,
      "step": 245360
    },
    {
      "epoch": 0.05195869917285787,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015092620000000003,
      "loss": 3.1392,
      "step": 245370
    },
    {
      "epoch": 0.05197734240901605,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001509242,
      "loss": 3.0902,
      "step": 245380
    },
    {
      "epoch": 0.05199598564517424,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015092220000000002,
      "loss": 3.1144,
      "step": 245390
    },
    {
      "epoch": 0.052014628881332424,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001509202,
      "loss": 3.23,
      "step": 245400
    },
    {
      "epoch": 0.052033272117490606,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001509182,
      "loss": 3.1156,
      "step": 245410
    },
    {
      "epoch": 0.05205191535364879,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001509162,
      "loss": 3.0414,
      "step": 245420
    },
    {
      "epoch": 0.05207055858980698,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001509142,
      "loss": 3.0313,
      "step": 245430
    },
    {
      "epoch": 0.05208920182596516,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015091220000000002,
      "loss": 3.1611,
      "step": 245440
    },
    {
      "epoch": 0.05210784506212334,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001509102,
      "loss": 3.2152,
      "step": 245450
    },
    {
      "epoch": 0.05212648829828152,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001509082,
      "loss": 3.095,
      "step": 245460
    },
    {
      "epoch": 0.052145131534439705,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001509062,
      "loss": 3.0606,
      "step": 245470
    },
    {
      "epoch": 0.052163774770597894,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001509042,
      "loss": 3.1824,
      "step": 245480
    },
    {
      "epoch": 0.052182418006756076,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015090220000000003,
      "loss": 3.1026,
      "step": 245490
    },
    {
      "epoch": 0.05220106124291426,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001509002,
      "loss": 3.0558,
      "step": 245500
    },
    {
      "epoch": 0.05221970447907244,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015089820000000002,
      "loss": 3.1432,
      "step": 245510
    },
    {
      "epoch": 0.05223834771523063,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001508962,
      "loss": 3.1819,
      "step": 245520
    },
    {
      "epoch": 0.05225699095138881,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001508942,
      "loss": 3.0262,
      "step": 245530
    },
    {
      "epoch": 0.05227563418754699,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001508922,
      "loss": 3.0941,
      "step": 245540
    },
    {
      "epoch": 0.052294277423705175,
      "grad_norm": 0.875,
      "learning_rate": 0.0001508902,
      "loss": 3.2043,
      "step": 245550
    },
    {
      "epoch": 0.052312920659863364,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015088820000000002,
      "loss": 3.116,
      "step": 245560
    },
    {
      "epoch": 0.052331563896021546,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001508862,
      "loss": 3.0517,
      "step": 245570
    },
    {
      "epoch": 0.05235020713217973,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015088420000000001,
      "loss": 3.1118,
      "step": 245580
    },
    {
      "epoch": 0.05236885036833791,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001508822,
      "loss": 3.1623,
      "step": 245590
    },
    {
      "epoch": 0.0523874936044961,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001508802,
      "loss": 3.1446,
      "step": 245600
    },
    {
      "epoch": 0.05240613684065428,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001508782,
      "loss": 3.0898,
      "step": 245610
    },
    {
      "epoch": 0.05242478007681246,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001508762,
      "loss": 3.1096,
      "step": 245620
    },
    {
      "epoch": 0.052443423312970645,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015087420000000002,
      "loss": 3.048,
      "step": 245630
    },
    {
      "epoch": 0.05246206654912883,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001508722,
      "loss": 3.1974,
      "step": 245640
    },
    {
      "epoch": 0.052480709785287016,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001508702,
      "loss": 2.962,
      "step": 245650
    },
    {
      "epoch": 0.0524993530214452,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001508682,
      "loss": 3.0802,
      "step": 245660
    },
    {
      "epoch": 0.05251799625760338,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001508662,
      "loss": 3.1059,
      "step": 245670
    },
    {
      "epoch": 0.05253663949376156,
      "grad_norm": 0.625,
      "learning_rate": 0.0001508642,
      "loss": 3.3185,
      "step": 245680
    },
    {
      "epoch": 0.05255528272991975,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001508622,
      "loss": 3.1137,
      "step": 245690
    },
    {
      "epoch": 0.05257392596607793,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015086020000000002,
      "loss": 3.1519,
      "step": 245700
    },
    {
      "epoch": 0.052592569202236115,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001508582,
      "loss": 3.0808,
      "step": 245710
    },
    {
      "epoch": 0.0526112124383943,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001508562,
      "loss": 3.1347,
      "step": 245720
    },
    {
      "epoch": 0.052629855674552486,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001508542,
      "loss": 3.1794,
      "step": 245730
    },
    {
      "epoch": 0.05264849891071067,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001508522,
      "loss": 3.2001,
      "step": 245740
    },
    {
      "epoch": 0.05266714214686885,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00015085020000000002,
      "loss": 3.2269,
      "step": 245750
    },
    {
      "epoch": 0.05268578538302703,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015084820000000002,
      "loss": 3.0801,
      "step": 245760
    },
    {
      "epoch": 0.05270442861918522,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001508462,
      "loss": 3.1993,
      "step": 245770
    },
    {
      "epoch": 0.052723071855343404,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001508442,
      "loss": 3.0613,
      "step": 245780
    },
    {
      "epoch": 0.052741715091501586,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001508422,
      "loss": 3.0751,
      "step": 245790
    },
    {
      "epoch": 0.05276035832765977,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001508402,
      "loss": 3.1051,
      "step": 245800
    },
    {
      "epoch": 0.05277900156381795,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001508382,
      "loss": 3.2024,
      "step": 245810
    },
    {
      "epoch": 0.05279764479997614,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015083620000000002,
      "loss": 3.1137,
      "step": 245820
    },
    {
      "epoch": 0.05281628803613432,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001508342,
      "loss": 3.1425,
      "step": 245830
    },
    {
      "epoch": 0.0528349312722925,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001508322,
      "loss": 3.1441,
      "step": 245840
    },
    {
      "epoch": 0.052853574508450685,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001508302,
      "loss": 3.2712,
      "step": 245850
    },
    {
      "epoch": 0.052872217744608874,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001508282,
      "loss": 3.0829,
      "step": 245860
    },
    {
      "epoch": 0.052890860980767056,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001508262,
      "loss": 3.0225,
      "step": 245870
    },
    {
      "epoch": 0.05290950421692524,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015082420000000002,
      "loss": 3.1834,
      "step": 245880
    },
    {
      "epoch": 0.05292814745308342,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015082220000000001,
      "loss": 3.1128,
      "step": 245890
    },
    {
      "epoch": 0.05294679068924161,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001508202,
      "loss": 3.1808,
      "step": 245900
    },
    {
      "epoch": 0.05296543392539979,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001508182,
      "loss": 3.1568,
      "step": 245910
    },
    {
      "epoch": 0.05298407716155797,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001508162,
      "loss": 3.1462,
      "step": 245920
    },
    {
      "epoch": 0.053002720397716155,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015081420000000002,
      "loss": 2.9666,
      "step": 245930
    },
    {
      "epoch": 0.053021363633874344,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001508122,
      "loss": 3.1889,
      "step": 245940
    },
    {
      "epoch": 0.053040006870032526,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015081020000000002,
      "loss": 3.0293,
      "step": 245950
    },
    {
      "epoch": 0.05305865010619071,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001508082,
      "loss": 3.0991,
      "step": 245960
    },
    {
      "epoch": 0.05307729334234889,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001508062,
      "loss": 3.1961,
      "step": 245970
    },
    {
      "epoch": 0.05309593657850707,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015080420000000003,
      "loss": 3.1864,
      "step": 245980
    },
    {
      "epoch": 0.05311457981466526,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001508022,
      "loss": 3.1673,
      "step": 245990
    },
    {
      "epoch": 0.05313322305082344,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015080020000000002,
      "loss": 3.0338,
      "step": 246000
    },
    {
      "epoch": 0.053151866286981625,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001507982,
      "loss": 3.0889,
      "step": 246010
    },
    {
      "epoch": 0.05317050952313981,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001507962,
      "loss": 3.0808,
      "step": 246020
    },
    {
      "epoch": 0.053189152759297996,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001507942,
      "loss": 3.1209,
      "step": 246030
    },
    {
      "epoch": 0.05320779599545618,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001507922,
      "loss": 3.2335,
      "step": 246040
    },
    {
      "epoch": 0.05322643923161436,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015079020000000003,
      "loss": 3.1433,
      "step": 246050
    },
    {
      "epoch": 0.05324508246777254,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001507882,
      "loss": 3.0647,
      "step": 246060
    },
    {
      "epoch": 0.05326372570393073,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015078620000000002,
      "loss": 3.1235,
      "step": 246070
    },
    {
      "epoch": 0.05328236894008891,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015078419999999999,
      "loss": 3.1041,
      "step": 246080
    },
    {
      "epoch": 0.053301012176247095,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001507822,
      "loss": 3.1782,
      "step": 246090
    },
    {
      "epoch": 0.05331965541240528,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001507802,
      "loss": 3.2017,
      "step": 246100
    },
    {
      "epoch": 0.053338298648563466,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001507782,
      "loss": 3.1795,
      "step": 246110
    },
    {
      "epoch": 0.05335694188472165,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015077620000000002,
      "loss": 3.2028,
      "step": 246120
    },
    {
      "epoch": 0.05337558512087983,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001507742,
      "loss": 2.9497,
      "step": 246130
    },
    {
      "epoch": 0.05339422835703801,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001507722,
      "loss": 3.1043,
      "step": 246140
    },
    {
      "epoch": 0.053412871593196194,
      "grad_norm": 0.75,
      "learning_rate": 0.0001507702,
      "loss": 3.0539,
      "step": 246150
    },
    {
      "epoch": 0.05343151482935438,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001507682,
      "loss": 3.2047,
      "step": 246160
    },
    {
      "epoch": 0.053450158065512565,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001507662,
      "loss": 3.1427,
      "step": 246170
    },
    {
      "epoch": 0.05346880130167075,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001507642,
      "loss": 3.1103,
      "step": 246180
    },
    {
      "epoch": 0.05348744453782893,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015076220000000002,
      "loss": 3.1133,
      "step": 246190
    },
    {
      "epoch": 0.05350608777398712,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015076020000000001,
      "loss": 3.1299,
      "step": 246200
    },
    {
      "epoch": 0.0535247310101453,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001507582,
      "loss": 3.1372,
      "step": 246210
    },
    {
      "epoch": 0.05354337424630348,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001507562,
      "loss": 3.0201,
      "step": 246220
    },
    {
      "epoch": 0.053562017482461664,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001507542,
      "loss": 3.0102,
      "step": 246230
    },
    {
      "epoch": 0.05358066071861985,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015075220000000002,
      "loss": 3.0652,
      "step": 246240
    },
    {
      "epoch": 0.053599303954778035,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001507502,
      "loss": 3.1076,
      "step": 246250
    },
    {
      "epoch": 0.05361794719093622,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015074820000000001,
      "loss": 3.1639,
      "step": 246260
    },
    {
      "epoch": 0.0536365904270944,
      "grad_norm": 0.625,
      "learning_rate": 0.0001507462,
      "loss": 3.1286,
      "step": 246270
    },
    {
      "epoch": 0.05365523366325259,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001507442,
      "loss": 3.1081,
      "step": 246280
    },
    {
      "epoch": 0.05367387689941077,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001507422,
      "loss": 3.0938,
      "step": 246290
    },
    {
      "epoch": 0.05369252013556895,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001507402,
      "loss": 3.1772,
      "step": 246300
    },
    {
      "epoch": 0.053711163371727134,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015073820000000002,
      "loss": 3.1685,
      "step": 246310
    },
    {
      "epoch": 0.053729806607885316,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015073620000000002,
      "loss": 3.1017,
      "step": 246320
    },
    {
      "epoch": 0.053748449844043505,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001507342,
      "loss": 3.092,
      "step": 246330
    },
    {
      "epoch": 0.05376709308020169,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001507322,
      "loss": 3.257,
      "step": 246340
    },
    {
      "epoch": 0.05378573631635987,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001507302,
      "loss": 3.1595,
      "step": 246350
    },
    {
      "epoch": 0.05380437955251805,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001507282,
      "loss": 3.0552,
      "step": 246360
    },
    {
      "epoch": 0.05382302278867624,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015072620000000002,
      "loss": 3.122,
      "step": 246370
    },
    {
      "epoch": 0.05384166602483442,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015072420000000002,
      "loss": 3.1718,
      "step": 246380
    },
    {
      "epoch": 0.053860309260992605,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001507222,
      "loss": 3.1385,
      "step": 246390
    },
    {
      "epoch": 0.05387895249715079,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001507202,
      "loss": 3.1427,
      "step": 246400
    },
    {
      "epoch": 0.053897595733308976,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001507182,
      "loss": 3.0055,
      "step": 246410
    },
    {
      "epoch": 0.05391623896946716,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001507162,
      "loss": 3.0261,
      "step": 246420
    },
    {
      "epoch": 0.05393488220562534,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001507142,
      "loss": 3.1945,
      "step": 246430
    },
    {
      "epoch": 0.05395352544178352,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015071220000000002,
      "loss": 3.0379,
      "step": 246440
    },
    {
      "epoch": 0.05397216867794171,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001507102,
      "loss": 3.0989,
      "step": 246450
    },
    {
      "epoch": 0.05399081191409989,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001507082,
      "loss": 3.1415,
      "step": 246460
    },
    {
      "epoch": 0.054009455150258075,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001507062,
      "loss": 3.1386,
      "step": 246470
    },
    {
      "epoch": 0.05402809838641626,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001507042,
      "loss": 3.275,
      "step": 246480
    },
    {
      "epoch": 0.05404674162257444,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015070220000000002,
      "loss": 3.1508,
      "step": 246490
    },
    {
      "epoch": 0.05406538485873263,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001507002,
      "loss": 3.1819,
      "step": 246500
    },
    {
      "epoch": 0.05408402809489081,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001506982,
      "loss": 3.0847,
      "step": 246510
    },
    {
      "epoch": 0.05410267133104899,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001506962,
      "loss": 3.0576,
      "step": 246520
    },
    {
      "epoch": 0.054121314567207174,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001506942,
      "loss": 3.118,
      "step": 246530
    },
    {
      "epoch": 0.05413995780336536,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015069220000000003,
      "loss": 3.1072,
      "step": 246540
    },
    {
      "epoch": 0.054158601039523545,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001506902,
      "loss": 3.0394,
      "step": 246550
    },
    {
      "epoch": 0.05417724427568173,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015068820000000002,
      "loss": 3.2339,
      "step": 246560
    },
    {
      "epoch": 0.05419588751183991,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001506862,
      "loss": 3.0295,
      "step": 246570
    },
    {
      "epoch": 0.0542145307479981,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001506842,
      "loss": 3.2122,
      "step": 246580
    },
    {
      "epoch": 0.05423317398415628,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001506822,
      "loss": 3.1851,
      "step": 246590
    },
    {
      "epoch": 0.05425181722031446,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001506802,
      "loss": 3.158,
      "step": 246600
    },
    {
      "epoch": 0.054270460456472644,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015067820000000002,
      "loss": 3.191,
      "step": 246610
    },
    {
      "epoch": 0.05428910369263083,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001506762,
      "loss": 3.1679,
      "step": 246620
    },
    {
      "epoch": 0.054307746928789015,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015067420000000002,
      "loss": 3.1159,
      "step": 246630
    },
    {
      "epoch": 0.0543263901649472,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015067219999999998,
      "loss": 3.0804,
      "step": 246640
    },
    {
      "epoch": 0.05434503340110538,
      "grad_norm": 0.75,
      "learning_rate": 0.0001506702,
      "loss": 3.1967,
      "step": 246650
    },
    {
      "epoch": 0.05436367663726356,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001506682,
      "loss": 3.2388,
      "step": 246660
    },
    {
      "epoch": 0.05438231987342175,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001506662,
      "loss": 3.0319,
      "step": 246670
    },
    {
      "epoch": 0.05440096310957993,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015066420000000002,
      "loss": 3.1815,
      "step": 246680
    },
    {
      "epoch": 0.054419606345738114,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001506622,
      "loss": 2.9694,
      "step": 246690
    },
    {
      "epoch": 0.054438249581896296,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001506602,
      "loss": 3.3381,
      "step": 246700
    },
    {
      "epoch": 0.054456892818054485,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001506582,
      "loss": 3.1179,
      "step": 246710
    },
    {
      "epoch": 0.05447553605421267,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001506562,
      "loss": 3.1412,
      "step": 246720
    },
    {
      "epoch": 0.05449417929037085,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015065420000000003,
      "loss": 3.1594,
      "step": 246730
    },
    {
      "epoch": 0.05451282252652903,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001506522,
      "loss": 3.0999,
      "step": 246740
    },
    {
      "epoch": 0.05453146576268722,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015065020000000002,
      "loss": 3.0248,
      "step": 246750
    },
    {
      "epoch": 0.0545501089988454,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001506482,
      "loss": 3.1583,
      "step": 246760
    },
    {
      "epoch": 0.054568752235003584,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001506462,
      "loss": 3.0939,
      "step": 246770
    },
    {
      "epoch": 0.054587395471161766,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001506442,
      "loss": 3.1976,
      "step": 246780
    },
    {
      "epoch": 0.054606038707319955,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001506422,
      "loss": 3.1434,
      "step": 246790
    },
    {
      "epoch": 0.05462468194347814,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015064020000000002,
      "loss": 3.0479,
      "step": 246800
    },
    {
      "epoch": 0.05464332517963632,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015063820000000002,
      "loss": 3.1927,
      "step": 246810
    },
    {
      "epoch": 0.0546619684157945,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001506362,
      "loss": 3.0646,
      "step": 246820
    },
    {
      "epoch": 0.05468061165195268,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001506342,
      "loss": 3.1515,
      "step": 246830
    },
    {
      "epoch": 0.05469925488811087,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001506322,
      "loss": 3.1481,
      "step": 246840
    },
    {
      "epoch": 0.054717898124269054,
      "grad_norm": 0.625,
      "learning_rate": 0.0001506302,
      "loss": 3.131,
      "step": 246850
    },
    {
      "epoch": 0.054736541360427236,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001506282,
      "loss": 3.1298,
      "step": 246860
    },
    {
      "epoch": 0.05475518459658542,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015062620000000002,
      "loss": 3.0672,
      "step": 246870
    },
    {
      "epoch": 0.05477382783274361,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015062420000000001,
      "loss": 3.1536,
      "step": 246880
    },
    {
      "epoch": 0.05479247106890179,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001506222,
      "loss": 3.2197,
      "step": 246890
    },
    {
      "epoch": 0.05481111430505997,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001506202,
      "loss": 3.2226,
      "step": 246900
    },
    {
      "epoch": 0.05482975754121815,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001506182,
      "loss": 3.1378,
      "step": 246910
    },
    {
      "epoch": 0.05484840077737634,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001506162,
      "loss": 3.1185,
      "step": 246920
    },
    {
      "epoch": 0.054867044013534524,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015061420000000002,
      "loss": 3.1514,
      "step": 246930
    },
    {
      "epoch": 0.054885687249692706,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015061220000000001,
      "loss": 3.2682,
      "step": 246940
    },
    {
      "epoch": 0.05490433048585089,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001506102,
      "loss": 3.1257,
      "step": 246950
    },
    {
      "epoch": 0.05492297372200908,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001506082,
      "loss": 3.0433,
      "step": 246960
    },
    {
      "epoch": 0.05494161695816726,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001506062,
      "loss": 3.0805,
      "step": 246970
    },
    {
      "epoch": 0.05496026019432544,
      "grad_norm": 0.59375,
      "learning_rate": 0.00015060420000000002,
      "loss": 3.0746,
      "step": 246980
    },
    {
      "epoch": 0.054978903430483624,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001506022,
      "loss": 3.2897,
      "step": 246990
    },
    {
      "epoch": 0.054997546666641806,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015060020000000002,
      "loss": 3.2098,
      "step": 247000
    },
    {
      "epoch": 0.055016189902799995,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001505982,
      "loss": 2.9959,
      "step": 247010
    },
    {
      "epoch": 0.05503483313895818,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001505962,
      "loss": 2.9899,
      "step": 247020
    },
    {
      "epoch": 0.05505347637511636,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001505942,
      "loss": 3.2346,
      "step": 247030
    },
    {
      "epoch": 0.05507211961127454,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001505922,
      "loss": 3.2221,
      "step": 247040
    },
    {
      "epoch": 0.05509076284743273,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015059020000000002,
      "loss": 3.1792,
      "step": 247050
    },
    {
      "epoch": 0.05510940608359091,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001505882,
      "loss": 3.1598,
      "step": 247060
    },
    {
      "epoch": 0.055128049319749094,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001505862,
      "loss": 3.1089,
      "step": 247070
    },
    {
      "epoch": 0.055146692555907276,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001505842,
      "loss": 3.0756,
      "step": 247080
    },
    {
      "epoch": 0.055165335792065465,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001505822,
      "loss": 3.2235,
      "step": 247090
    },
    {
      "epoch": 0.05518397902822365,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015058020000000003,
      "loss": 3.1642,
      "step": 247100
    },
    {
      "epoch": 0.05520262226438183,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001505782,
      "loss": 3.0932,
      "step": 247110
    },
    {
      "epoch": 0.05522126550054001,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015057620000000002,
      "loss": 3.06,
      "step": 247120
    },
    {
      "epoch": 0.0552399087366982,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015057419999999999,
      "loss": 3.0844,
      "step": 247130
    },
    {
      "epoch": 0.05525855197285638,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001505722,
      "loss": 3.2306,
      "step": 247140
    },
    {
      "epoch": 0.055277195209014564,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001505702,
      "loss": 3.0889,
      "step": 247150
    },
    {
      "epoch": 0.055295838445172746,
      "grad_norm": 0.75,
      "learning_rate": 0.0001505682,
      "loss": 3.0736,
      "step": 247160
    },
    {
      "epoch": 0.05531448168133093,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015056620000000002,
      "loss": 3.17,
      "step": 247170
    },
    {
      "epoch": 0.05533312491748912,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001505642,
      "loss": 3.0822,
      "step": 247180
    },
    {
      "epoch": 0.0553517681536473,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001505622,
      "loss": 3.1036,
      "step": 247190
    },
    {
      "epoch": 0.05537041138980548,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001505602,
      "loss": 3.1121,
      "step": 247200
    },
    {
      "epoch": 0.05538905462596366,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001505582,
      "loss": 3.2584,
      "step": 247210
    },
    {
      "epoch": 0.05540769786212185,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015055620000000003,
      "loss": 3.069,
      "step": 247220
    },
    {
      "epoch": 0.055426341098280034,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001505542,
      "loss": 3.0896,
      "step": 247230
    },
    {
      "epoch": 0.055444984334438216,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015055220000000002,
      "loss": 3.082,
      "step": 247240
    },
    {
      "epoch": 0.0554636275705964,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001505502,
      "loss": 3.0422,
      "step": 247250
    },
    {
      "epoch": 0.05548227080675459,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001505482,
      "loss": 3.0132,
      "step": 247260
    },
    {
      "epoch": 0.05550091404291277,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001505462,
      "loss": 3.1686,
      "step": 247270
    },
    {
      "epoch": 0.05551955727907095,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001505442,
      "loss": 3.0645,
      "step": 247280
    },
    {
      "epoch": 0.05553820051522913,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015054220000000002,
      "loss": 2.9449,
      "step": 247290
    },
    {
      "epoch": 0.05555684375138732,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001505402,
      "loss": 3.1067,
      "step": 247300
    },
    {
      "epoch": 0.055575486987545504,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015053820000000002,
      "loss": 3.0201,
      "step": 247310
    },
    {
      "epoch": 0.055594130223703686,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001505362,
      "loss": 3.0691,
      "step": 247320
    },
    {
      "epoch": 0.05561277345986187,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001505342,
      "loss": 3.1891,
      "step": 247330
    },
    {
      "epoch": 0.05563141669602005,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001505322,
      "loss": 3.2137,
      "step": 247340
    },
    {
      "epoch": 0.05565005993217824,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001505302,
      "loss": 3.1493,
      "step": 247350
    },
    {
      "epoch": 0.05566870316833642,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015052820000000002,
      "loss": 3.1371,
      "step": 247360
    },
    {
      "epoch": 0.0556873464044946,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015052620000000002,
      "loss": 3.1214,
      "step": 247370
    },
    {
      "epoch": 0.055705989640652785,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001505242,
      "loss": 3.0761,
      "step": 247380
    },
    {
      "epoch": 0.055724632876810974,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001505222,
      "loss": 3.1733,
      "step": 247390
    },
    {
      "epoch": 0.055743276112969156,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001505202,
      "loss": 3.1597,
      "step": 247400
    },
    {
      "epoch": 0.05576191934912734,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001505182,
      "loss": 3.0011,
      "step": 247410
    },
    {
      "epoch": 0.05578056258528552,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015051620000000002,
      "loss": 3.0645,
      "step": 247420
    },
    {
      "epoch": 0.05579920582144371,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015051420000000002,
      "loss": 3.1395,
      "step": 247430
    },
    {
      "epoch": 0.05581784905760189,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001505122,
      "loss": 3.2879,
      "step": 247440
    },
    {
      "epoch": 0.05583649229376007,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001505102,
      "loss": 3.2016,
      "step": 247450
    },
    {
      "epoch": 0.055855135529918255,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001505082,
      "loss": 3.1342,
      "step": 247460
    },
    {
      "epoch": 0.055873778766076444,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001505062,
      "loss": 3.2888,
      "step": 247470
    },
    {
      "epoch": 0.055892422002234626,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001505042,
      "loss": 3.1425,
      "step": 247480
    },
    {
      "epoch": 0.05591106523839281,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015050220000000002,
      "loss": 3.0382,
      "step": 247490
    },
    {
      "epoch": 0.05592970847455099,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001505002,
      "loss": 3.0724,
      "step": 247500
    },
    {
      "epoch": 0.05594835171070918,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001504982,
      "loss": 3.1161,
      "step": 247510
    },
    {
      "epoch": 0.05596699494686736,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001504962,
      "loss": 3.2302,
      "step": 247520
    },
    {
      "epoch": 0.05598563818302554,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001504942,
      "loss": 3.1115,
      "step": 247530
    },
    {
      "epoch": 0.056004281419183725,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015049220000000002,
      "loss": 3.155,
      "step": 247540
    },
    {
      "epoch": 0.05602292465534191,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001504902,
      "loss": 3.1873,
      "step": 247550
    },
    {
      "epoch": 0.056041567891500096,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015048820000000001,
      "loss": 3.2306,
      "step": 247560
    },
    {
      "epoch": 0.05606021112765828,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001504862,
      "loss": 3.0954,
      "step": 247570
    },
    {
      "epoch": 0.05607885436381646,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001504842,
      "loss": 3.1157,
      "step": 247580
    },
    {
      "epoch": 0.05609749759997464,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015048220000000003,
      "loss": 3.0571,
      "step": 247590
    },
    {
      "epoch": 0.05611614083613283,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001504802,
      "loss": 3.1837,
      "step": 247600
    },
    {
      "epoch": 0.056134784072291014,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015047820000000002,
      "loss": 3.2807,
      "step": 247610
    },
    {
      "epoch": 0.056153427308449196,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001504762,
      "loss": 3.064,
      "step": 247620
    },
    {
      "epoch": 0.05617207054460738,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001504742,
      "loss": 3.2283,
      "step": 247630
    },
    {
      "epoch": 0.05619071378076557,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001504722,
      "loss": 3.1244,
      "step": 247640
    },
    {
      "epoch": 0.05620935701692375,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001504702,
      "loss": 3.0864,
      "step": 247650
    },
    {
      "epoch": 0.05622800025308193,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015046820000000002,
      "loss": 3.1169,
      "step": 247660
    },
    {
      "epoch": 0.05624664348924011,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001504662,
      "loss": 3.1205,
      "step": 247670
    },
    {
      "epoch": 0.0562652867253983,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015046420000000002,
      "loss": 3.0912,
      "step": 247680
    },
    {
      "epoch": 0.056283929961556484,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001504622,
      "loss": 3.1832,
      "step": 247690
    },
    {
      "epoch": 0.056302573197714666,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001504602,
      "loss": 3.1504,
      "step": 247700
    },
    {
      "epoch": 0.05632121643387285,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015045820000000003,
      "loss": 3.0992,
      "step": 247710
    },
    {
      "epoch": 0.05633985967003103,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001504562,
      "loss": 3.1725,
      "step": 247720
    },
    {
      "epoch": 0.05635850290618922,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015045420000000002,
      "loss": 3.1858,
      "step": 247730
    },
    {
      "epoch": 0.0563771461423474,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001504522,
      "loss": 3.1702,
      "step": 247740
    },
    {
      "epoch": 0.05639578937850558,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001504502,
      "loss": 3.2271,
      "step": 247750
    },
    {
      "epoch": 0.056414432614663765,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001504482,
      "loss": 3.1176,
      "step": 247760
    },
    {
      "epoch": 0.056433075850821954,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001504462,
      "loss": 3.1657,
      "step": 247770
    },
    {
      "epoch": 0.056451719086980136,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015044420000000003,
      "loss": 2.9008,
      "step": 247780
    },
    {
      "epoch": 0.05647036232313832,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001504422,
      "loss": 3.1252,
      "step": 247790
    },
    {
      "epoch": 0.0564890055592965,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015044020000000002,
      "loss": 3.1061,
      "step": 247800
    },
    {
      "epoch": 0.05650764879545469,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001504382,
      "loss": 3.2013,
      "step": 247810
    },
    {
      "epoch": 0.05652629203161287,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001504362,
      "loss": 3.0413,
      "step": 247820
    },
    {
      "epoch": 0.05654493526777105,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001504342,
      "loss": 3.1208,
      "step": 247830
    },
    {
      "epoch": 0.056563578503929235,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001504322,
      "loss": 3.0381,
      "step": 247840
    },
    {
      "epoch": 0.056582221740087424,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00015043020000000002,
      "loss": 3.1416,
      "step": 247850
    },
    {
      "epoch": 0.056600864976245606,
      "grad_norm": 0.75,
      "learning_rate": 0.0001504282,
      "loss": 3.0771,
      "step": 247860
    },
    {
      "epoch": 0.05661950821240379,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001504262,
      "loss": 3.0699,
      "step": 247870
    },
    {
      "epoch": 0.05663815144856197,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001504242,
      "loss": 3.0223,
      "step": 247880
    },
    {
      "epoch": 0.05665679468472015,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001504222,
      "loss": 3.2309,
      "step": 247890
    },
    {
      "epoch": 0.05667543792087834,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001504202,
      "loss": 3.0942,
      "step": 247900
    },
    {
      "epoch": 0.05669408115703652,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001504182,
      "loss": 3.2462,
      "step": 247910
    },
    {
      "epoch": 0.056712724393194705,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015041620000000002,
      "loss": 3.0506,
      "step": 247920
    },
    {
      "epoch": 0.05673136762935289,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015041420000000001,
      "loss": 3.1771,
      "step": 247930
    },
    {
      "epoch": 0.056750010865511076,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001504122,
      "loss": 3.2147,
      "step": 247940
    },
    {
      "epoch": 0.05676865410166926,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001504102,
      "loss": 3.1028,
      "step": 247950
    },
    {
      "epoch": 0.05678729733782744,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001504082,
      "loss": 3.126,
      "step": 247960
    },
    {
      "epoch": 0.05680594057398562,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001504062,
      "loss": 3.1977,
      "step": 247970
    },
    {
      "epoch": 0.05682458381014381,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015040420000000002,
      "loss": 3.1797,
      "step": 247980
    },
    {
      "epoch": 0.05684322704630199,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015040220000000001,
      "loss": 3.1161,
      "step": 247990
    },
    {
      "epoch": 0.056861870282460175,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001504002,
      "loss": 3.1161,
      "step": 248000
    },
    {
      "epoch": 0.05688051351861836,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001503982,
      "loss": 3.2046,
      "step": 248010
    },
    {
      "epoch": 0.056899156754776546,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001503962,
      "loss": 3.0839,
      "step": 248020
    },
    {
      "epoch": 0.05691779999093473,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001503942,
      "loss": 2.9783,
      "step": 248030
    },
    {
      "epoch": 0.05693644322709291,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001503922,
      "loss": 3.0145,
      "step": 248040
    },
    {
      "epoch": 0.05695508646325109,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015039020000000002,
      "loss": 3.0129,
      "step": 248050
    },
    {
      "epoch": 0.056973729699409274,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001503882,
      "loss": 3.1908,
      "step": 248060
    },
    {
      "epoch": 0.05699237293556746,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001503862,
      "loss": 3.1831,
      "step": 248070
    },
    {
      "epoch": 0.057011016171725645,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001503842,
      "loss": 3.0903,
      "step": 248080
    },
    {
      "epoch": 0.05702965940788383,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001503822,
      "loss": 3.085,
      "step": 248090
    },
    {
      "epoch": 0.05704830264404201,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015038020000000002,
      "loss": 3.1245,
      "step": 248100
    },
    {
      "epoch": 0.0570669458802002,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001503782,
      "loss": 3.4154,
      "step": 248110
    },
    {
      "epoch": 0.05708558911635838,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001503762,
      "loss": 3.1755,
      "step": 248120
    },
    {
      "epoch": 0.05710423235251656,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001503742,
      "loss": 3.1837,
      "step": 248130
    },
    {
      "epoch": 0.057122875588674744,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001503722,
      "loss": 3.0092,
      "step": 248140
    },
    {
      "epoch": 0.05714151882483293,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015037020000000003,
      "loss": 3.0955,
      "step": 248150
    },
    {
      "epoch": 0.057160162060991115,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001503682,
      "loss": 3.1311,
      "step": 248160
    },
    {
      "epoch": 0.0571788052971493,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015036620000000002,
      "loss": 3.1471,
      "step": 248170
    },
    {
      "epoch": 0.05719744853330748,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001503642,
      "loss": 3.2343,
      "step": 248180
    },
    {
      "epoch": 0.05721609176946567,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001503622,
      "loss": 3.0438,
      "step": 248190
    },
    {
      "epoch": 0.05723473500562385,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001503602,
      "loss": 3.0311,
      "step": 248200
    },
    {
      "epoch": 0.05725337824178203,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001503582,
      "loss": 3.0521,
      "step": 248210
    },
    {
      "epoch": 0.057272021477940215,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015035620000000002,
      "loss": 3.0652,
      "step": 248220
    },
    {
      "epoch": 0.0572906647140984,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001503542,
      "loss": 3.0939,
      "step": 248230
    },
    {
      "epoch": 0.057309307950256586,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015035220000000001,
      "loss": 3.0219,
      "step": 248240
    },
    {
      "epoch": 0.05732795118641477,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001503502,
      "loss": 3.1591,
      "step": 248250
    },
    {
      "epoch": 0.05734659442257295,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001503482,
      "loss": 3.1701,
      "step": 248260
    },
    {
      "epoch": 0.05736523765873113,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015034620000000003,
      "loss": 3.2158,
      "step": 248270
    },
    {
      "epoch": 0.05738388089488932,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001503442,
      "loss": 3.0842,
      "step": 248280
    },
    {
      "epoch": 0.0574025241310475,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015034220000000002,
      "loss": 3.2813,
      "step": 248290
    },
    {
      "epoch": 0.057421167367205685,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001503402,
      "loss": 3.0798,
      "step": 248300
    },
    {
      "epoch": 0.05743981060336387,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001503382,
      "loss": 3.1998,
      "step": 248310
    },
    {
      "epoch": 0.057458453839522056,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001503362,
      "loss": 3.2618,
      "step": 248320
    },
    {
      "epoch": 0.05747709707568024,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001503342,
      "loss": 3.075,
      "step": 248330
    },
    {
      "epoch": 0.05749574031183842,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015033220000000002,
      "loss": 3.1168,
      "step": 248340
    },
    {
      "epoch": 0.0575143835479966,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001503302,
      "loss": 3.0551,
      "step": 248350
    },
    {
      "epoch": 0.05753302678415479,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015032820000000002,
      "loss": 3.3027,
      "step": 248360
    },
    {
      "epoch": 0.05755167002031297,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001503262,
      "loss": 3.0892,
      "step": 248370
    },
    {
      "epoch": 0.057570313256471155,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001503242,
      "loss": 3.1677,
      "step": 248380
    },
    {
      "epoch": 0.05758895649262934,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001503222,
      "loss": 3.1295,
      "step": 248390
    },
    {
      "epoch": 0.05760759972878752,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001503202,
      "loss": 2.8651,
      "step": 248400
    },
    {
      "epoch": 0.05762624296494571,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015031820000000002,
      "loss": 3.0094,
      "step": 248410
    },
    {
      "epoch": 0.05764488620110389,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015031620000000002,
      "loss": 3.1866,
      "step": 248420
    },
    {
      "epoch": 0.05766352943726207,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001503142,
      "loss": 3.0083,
      "step": 248430
    },
    {
      "epoch": 0.057682172673420254,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001503122,
      "loss": 3.2734,
      "step": 248440
    },
    {
      "epoch": 0.05770081590957844,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001503102,
      "loss": 3.2303,
      "step": 248450
    },
    {
      "epoch": 0.057719459145736625,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001503082,
      "loss": 3.1635,
      "step": 248460
    },
    {
      "epoch": 0.05773810238189481,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001503062,
      "loss": 3.0978,
      "step": 248470
    },
    {
      "epoch": 0.05775674561805299,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015030420000000002,
      "loss": 3.0643,
      "step": 248480
    },
    {
      "epoch": 0.05777538885421118,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001503022,
      "loss": 3.1573,
      "step": 248490
    },
    {
      "epoch": 0.05779403209036936,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001503002,
      "loss": 3.1947,
      "step": 248500
    },
    {
      "epoch": 0.05781267532652754,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001502982,
      "loss": 3.1355,
      "step": 248510
    },
    {
      "epoch": 0.057831318562685724,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001502962,
      "loss": 3.1761,
      "step": 248520
    },
    {
      "epoch": 0.05784996179884391,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001502942,
      "loss": 3.2021,
      "step": 248530
    },
    {
      "epoch": 0.057868605035002095,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015029220000000002,
      "loss": 3.2223,
      "step": 248540
    },
    {
      "epoch": 0.05788724827116028,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001502902,
      "loss": 3.1786,
      "step": 248550
    },
    {
      "epoch": 0.05790589150731846,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001502882,
      "loss": 3.0183,
      "step": 248560
    },
    {
      "epoch": 0.05792453474347664,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001502862,
      "loss": 3.1571,
      "step": 248570
    },
    {
      "epoch": 0.05794317797963483,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001502842,
      "loss": 3.1534,
      "step": 248580
    },
    {
      "epoch": 0.05796182121579301,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015028220000000002,
      "loss": 3.0272,
      "step": 248590
    },
    {
      "epoch": 0.057980464451951194,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001502802,
      "loss": 3.218,
      "step": 248600
    },
    {
      "epoch": 0.057999107688109376,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015027820000000001,
      "loss": 3.1163,
      "step": 248610
    },
    {
      "epoch": 0.058017750924267565,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001502762,
      "loss": 3.1484,
      "step": 248620
    },
    {
      "epoch": 0.05803639416042575,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001502742,
      "loss": 3.1835,
      "step": 248630
    },
    {
      "epoch": 0.05805503739658393,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001502722,
      "loss": 3.0212,
      "step": 248640
    },
    {
      "epoch": 0.05807368063274211,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001502702,
      "loss": 3.2187,
      "step": 248650
    },
    {
      "epoch": 0.0580923238689003,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015026820000000002,
      "loss": 3.1968,
      "step": 248660
    },
    {
      "epoch": 0.05811096710505848,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015026620000000001,
      "loss": 3.186,
      "step": 248670
    },
    {
      "epoch": 0.058129610341216664,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001502642,
      "loss": 3.0862,
      "step": 248680
    },
    {
      "epoch": 0.058148253577374846,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001502622,
      "loss": 3.2,
      "step": 248690
    },
    {
      "epoch": 0.058166896813533035,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001502602,
      "loss": 3.169,
      "step": 248700
    },
    {
      "epoch": 0.05818554004969122,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015025820000000002,
      "loss": 3.0323,
      "step": 248710
    },
    {
      "epoch": 0.0582041832858494,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001502562,
      "loss": 3.0716,
      "step": 248720
    },
    {
      "epoch": 0.05822282652200758,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015025420000000002,
      "loss": 3.2302,
      "step": 248730
    },
    {
      "epoch": 0.05824146975816576,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001502522,
      "loss": 3.058,
      "step": 248740
    },
    {
      "epoch": 0.05826011299432395,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001502502,
      "loss": 3.1491,
      "step": 248750
    },
    {
      "epoch": 0.058278756230482134,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015024820000000003,
      "loss": 3.1325,
      "step": 248760
    },
    {
      "epoch": 0.058297399466640316,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001502462,
      "loss": 3.2143,
      "step": 248770
    },
    {
      "epoch": 0.0583160427027985,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015024420000000002,
      "loss": 3.1082,
      "step": 248780
    },
    {
      "epoch": 0.05833468593895669,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001502422,
      "loss": 3.1735,
      "step": 248790
    },
    {
      "epoch": 0.05835332917511487,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001502402,
      "loss": 3.0563,
      "step": 248800
    },
    {
      "epoch": 0.05837197241127305,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001502382,
      "loss": 3.1727,
      "step": 248810
    },
    {
      "epoch": 0.058390615647431234,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001502362,
      "loss": 3.2024,
      "step": 248820
    },
    {
      "epoch": 0.05840925888358942,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015023420000000003,
      "loss": 3.276,
      "step": 248830
    },
    {
      "epoch": 0.058427902119747605,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001502322,
      "loss": 3.1156,
      "step": 248840
    },
    {
      "epoch": 0.05844654535590579,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015023020000000002,
      "loss": 2.957,
      "step": 248850
    },
    {
      "epoch": 0.05846518859206397,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015022819999999999,
      "loss": 3.2277,
      "step": 248860
    },
    {
      "epoch": 0.05848383182822216,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001502262,
      "loss": 3.0584,
      "step": 248870
    },
    {
      "epoch": 0.05850247506438034,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001502242,
      "loss": 3.2348,
      "step": 248880
    },
    {
      "epoch": 0.05852111830053852,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001502222,
      "loss": 3.1508,
      "step": 248890
    },
    {
      "epoch": 0.058539761536696704,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015022020000000002,
      "loss": 3.0838,
      "step": 248900
    },
    {
      "epoch": 0.058558404772854886,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001502182,
      "loss": 3.1565,
      "step": 248910
    },
    {
      "epoch": 0.058577048009013075,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015021620000000001,
      "loss": 3.058,
      "step": 248920
    },
    {
      "epoch": 0.05859569124517126,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001502142,
      "loss": 3.0095,
      "step": 248930
    },
    {
      "epoch": 0.05861433448132944,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001502122,
      "loss": 3.1613,
      "step": 248940
    },
    {
      "epoch": 0.05863297771748762,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001502102,
      "loss": 3.007,
      "step": 248950
    },
    {
      "epoch": 0.05865162095364581,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001502082,
      "loss": 3.1991,
      "step": 248960
    },
    {
      "epoch": 0.05867026418980399,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015020620000000002,
      "loss": 3.1913,
      "step": 248970
    },
    {
      "epoch": 0.058688907425962174,
      "grad_norm": 0.75,
      "learning_rate": 0.00015020420000000001,
      "loss": 3.1303,
      "step": 248980
    },
    {
      "epoch": 0.058707550662120356,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001502022,
      "loss": 3.1667,
      "step": 248990
    },
    {
      "epoch": 0.058726193898278545,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001502002,
      "loss": 3.1499,
      "step": 249000
    },
    {
      "epoch": 0.05874483713443673,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001501982,
      "loss": 3.2018,
      "step": 249010
    },
    {
      "epoch": 0.05876348037059491,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001501962,
      "loss": 3.1644,
      "step": 249020
    },
    {
      "epoch": 0.05878212360675309,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015019420000000002,
      "loss": 3.1008,
      "step": 249030
    },
    {
      "epoch": 0.05880076684291128,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015019220000000002,
      "loss": 3.2453,
      "step": 249040
    },
    {
      "epoch": 0.05881941007906946,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001501902,
      "loss": 3.1036,
      "step": 249050
    },
    {
      "epoch": 0.058838053315227644,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001501882,
      "loss": 3.1711,
      "step": 249060
    },
    {
      "epoch": 0.058856696551385826,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001501862,
      "loss": 2.9956,
      "step": 249070
    },
    {
      "epoch": 0.05887533978754401,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001501842,
      "loss": 3.1654,
      "step": 249080
    },
    {
      "epoch": 0.0588939830237022,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001501822,
      "loss": 3.1226,
      "step": 249090
    },
    {
      "epoch": 0.05891262625986038,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015018020000000002,
      "loss": 3.1025,
      "step": 249100
    },
    {
      "epoch": 0.05893126949601856,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001501782,
      "loss": 3.1435,
      "step": 249110
    },
    {
      "epoch": 0.05894991273217674,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001501762,
      "loss": 3.1529,
      "step": 249120
    },
    {
      "epoch": 0.05896855596833493,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001501742,
      "loss": 3.0044,
      "step": 249130
    },
    {
      "epoch": 0.058987199204493114,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001501722,
      "loss": 3.11,
      "step": 249140
    },
    {
      "epoch": 0.059005842440651296,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015017020000000002,
      "loss": 3.2236,
      "step": 249150
    },
    {
      "epoch": 0.05902448567680948,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015016820000000002,
      "loss": 3.0604,
      "step": 249160
    },
    {
      "epoch": 0.05904312891296767,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001501662,
      "loss": 3.1307,
      "step": 249170
    },
    {
      "epoch": 0.05906177214912585,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001501642,
      "loss": 3.1801,
      "step": 249180
    },
    {
      "epoch": 0.05908041538528403,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001501622,
      "loss": 3.0528,
      "step": 249190
    },
    {
      "epoch": 0.05909905862144221,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015016020000000003,
      "loss": 3.108,
      "step": 249200
    },
    {
      "epoch": 0.0591177018576004,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001501582,
      "loss": 3.2151,
      "step": 249210
    },
    {
      "epoch": 0.059136345093758584,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015015620000000002,
      "loss": 3.1696,
      "step": 249220
    },
    {
      "epoch": 0.059154988329916766,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001501542,
      "loss": 3.2722,
      "step": 249230
    },
    {
      "epoch": 0.05917363156607495,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001501522,
      "loss": 3.111,
      "step": 249240
    },
    {
      "epoch": 0.05919227480223313,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001501502,
      "loss": 3.1737,
      "step": 249250
    },
    {
      "epoch": 0.05921091803839132,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001501482,
      "loss": 3.1915,
      "step": 249260
    },
    {
      "epoch": 0.0592295612745495,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015014620000000002,
      "loss": 3.1455,
      "step": 249270
    },
    {
      "epoch": 0.05924820451070768,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001501442,
      "loss": 3.1103,
      "step": 249280
    },
    {
      "epoch": 0.059266847746865865,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015014220000000001,
      "loss": 3.1036,
      "step": 249290
    },
    {
      "epoch": 0.059285490983024054,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001501402,
      "loss": 3.1829,
      "step": 249300
    },
    {
      "epoch": 0.059304134219182236,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001501382,
      "loss": 3.2726,
      "step": 249310
    },
    {
      "epoch": 0.05932277745534042,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015013620000000003,
      "loss": 3.2282,
      "step": 249320
    },
    {
      "epoch": 0.0593414206914986,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001501342,
      "loss": 3.0942,
      "step": 249330
    },
    {
      "epoch": 0.05936006392765679,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015013220000000002,
      "loss": 3.0674,
      "step": 249340
    },
    {
      "epoch": 0.05937870716381497,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001501302,
      "loss": 2.9933,
      "step": 249350
    },
    {
      "epoch": 0.05939735039997315,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001501282,
      "loss": 3.0554,
      "step": 249360
    },
    {
      "epoch": 0.059415993636131335,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001501262,
      "loss": 3.1782,
      "step": 249370
    },
    {
      "epoch": 0.059434636872289524,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001501242,
      "loss": 3.1916,
      "step": 249380
    },
    {
      "epoch": 0.059453280108447706,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015012220000000002,
      "loss": 3.0442,
      "step": 249390
    },
    {
      "epoch": 0.05947192334460589,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001501202,
      "loss": 3.1409,
      "step": 249400
    },
    {
      "epoch": 0.05949056658076407,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015011820000000002,
      "loss": 3.1438,
      "step": 249410
    },
    {
      "epoch": 0.05950920981692225,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001501162,
      "loss": 3.0834,
      "step": 249420
    },
    {
      "epoch": 0.05952785305308044,
      "grad_norm": 0.75,
      "learning_rate": 0.0001501142,
      "loss": 3.1833,
      "step": 249430
    },
    {
      "epoch": 0.059546496289238623,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001501122,
      "loss": 3.1112,
      "step": 249440
    },
    {
      "epoch": 0.059565139525396806,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001501102,
      "loss": 3.116,
      "step": 249450
    },
    {
      "epoch": 0.05958378276155499,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015010820000000002,
      "loss": 3.1816,
      "step": 249460
    },
    {
      "epoch": 0.059602425997713177,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001501062,
      "loss": 3.0487,
      "step": 249470
    },
    {
      "epoch": 0.05962106923387136,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001501042,
      "loss": 3.0919,
      "step": 249480
    },
    {
      "epoch": 0.05963971247002954,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001501022,
      "loss": 3.078,
      "step": 249490
    },
    {
      "epoch": 0.05965835570618772,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001501002,
      "loss": 3.0043,
      "step": 249500
    },
    {
      "epoch": 0.05967699894234591,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001500982,
      "loss": 3.1723,
      "step": 249510
    },
    {
      "epoch": 0.059695642178504094,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001500962,
      "loss": 3.0589,
      "step": 249520
    },
    {
      "epoch": 0.059714285414662276,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015009420000000002,
      "loss": 3.0152,
      "step": 249530
    },
    {
      "epoch": 0.05973292865082046,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001500922,
      "loss": 3.0533,
      "step": 249540
    },
    {
      "epoch": 0.05975157188697865,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001500902,
      "loss": 3.1375,
      "step": 249550
    },
    {
      "epoch": 0.05977021512313683,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001500882,
      "loss": 3.1524,
      "step": 249560
    },
    {
      "epoch": 0.05978885835929501,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001500862,
      "loss": 3.1652,
      "step": 249570
    },
    {
      "epoch": 0.05980750159545319,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001500842,
      "loss": 3.1191,
      "step": 249580
    },
    {
      "epoch": 0.059826144831611375,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015008220000000002,
      "loss": 3.0895,
      "step": 249590
    },
    {
      "epoch": 0.059844788067769564,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015008020000000001,
      "loss": 2.992,
      "step": 249600
    },
    {
      "epoch": 0.059863431303927746,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001500782,
      "loss": 3.2258,
      "step": 249610
    },
    {
      "epoch": 0.05988207454008593,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001500762,
      "loss": 3.3155,
      "step": 249620
    },
    {
      "epoch": 0.05990071777624411,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001500742,
      "loss": 3.1791,
      "step": 249630
    },
    {
      "epoch": 0.0599193610124023,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001500722,
      "loss": 3.0336,
      "step": 249640
    },
    {
      "epoch": 0.05993800424856048,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001500702,
      "loss": 3.1301,
      "step": 249650
    },
    {
      "epoch": 0.05995664748471866,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015006820000000001,
      "loss": 3.2062,
      "step": 249660
    },
    {
      "epoch": 0.059975290720876845,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001500662,
      "loss": 3.0789,
      "step": 249670
    },
    {
      "epoch": 0.059993933957035034,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001500642,
      "loss": 3.0505,
      "step": 249680
    },
    {
      "epoch": 0.060012577193193216,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001500622,
      "loss": 3.3291,
      "step": 249690
    },
    {
      "epoch": 0.0600312204293514,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001500602,
      "loss": 3.1369,
      "step": 249700
    },
    {
      "epoch": 0.06004986366550958,
      "grad_norm": 0.625,
      "learning_rate": 0.00015005820000000002,
      "loss": 3.311,
      "step": 249710
    },
    {
      "epoch": 0.06006850690166777,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015005620000000002,
      "loss": 3.0678,
      "step": 249720
    },
    {
      "epoch": 0.06008715013782595,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001500542,
      "loss": 3.1459,
      "step": 249730
    },
    {
      "epoch": 0.06010579337398413,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001500522,
      "loss": 3.0867,
      "step": 249740
    },
    {
      "epoch": 0.060124436610142315,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001500502,
      "loss": 3.158,
      "step": 249750
    },
    {
      "epoch": 0.0601430798463005,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015004820000000002,
      "loss": 3.0667,
      "step": 249760
    },
    {
      "epoch": 0.060161723082458686,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001500462,
      "loss": 3.1667,
      "step": 249770
    },
    {
      "epoch": 0.06018036631861687,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015004420000000002,
      "loss": 3.1122,
      "step": 249780
    },
    {
      "epoch": 0.06019900955477505,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001500422,
      "loss": 3.1281,
      "step": 249790
    },
    {
      "epoch": 0.06021765279093323,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001500402,
      "loss": 3.2074,
      "step": 249800
    },
    {
      "epoch": 0.06023629602709142,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015003820000000003,
      "loss": 3.0867,
      "step": 249810
    },
    {
      "epoch": 0.0602549392632496,
      "grad_norm": 0.75,
      "learning_rate": 0.0001500362,
      "loss": 3.2113,
      "step": 249820
    },
    {
      "epoch": 0.060273582499407785,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015003420000000002,
      "loss": 3.0397,
      "step": 249830
    },
    {
      "epoch": 0.06029222573556597,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001500322,
      "loss": 3.0853,
      "step": 249840
    },
    {
      "epoch": 0.060310868971724156,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001500302,
      "loss": 3.0825,
      "step": 249850
    },
    {
      "epoch": 0.06032951220788234,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001500282,
      "loss": 3.1474,
      "step": 249860
    },
    {
      "epoch": 0.06034815544404052,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001500262,
      "loss": 3.246,
      "step": 249870
    },
    {
      "epoch": 0.0603667986801987,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015002420000000003,
      "loss": 3.2147,
      "step": 249880
    },
    {
      "epoch": 0.06038544191635689,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001500222,
      "loss": 3.1242,
      "step": 249890
    },
    {
      "epoch": 0.06040408515251507,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015002020000000002,
      "loss": 2.9785,
      "step": 249900
    },
    {
      "epoch": 0.060422728388673255,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015001819999999999,
      "loss": 3.0783,
      "step": 249910
    },
    {
      "epoch": 0.06044137162483144,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001500162,
      "loss": 3.1367,
      "step": 249920
    },
    {
      "epoch": 0.06046001486098962,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001500142,
      "loss": 3.2348,
      "step": 249930
    },
    {
      "epoch": 0.06047865809714781,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001500122,
      "loss": 3.1314,
      "step": 249940
    },
    {
      "epoch": 0.06049730133330599,
      "grad_norm": 0.625,
      "learning_rate": 0.00015001020000000002,
      "loss": 3.1663,
      "step": 249950
    },
    {
      "epoch": 0.06051594456946417,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001500082,
      "loss": 3.188,
      "step": 249960
    },
    {
      "epoch": 0.060534587805622354,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015000620000000001,
      "loss": 3.1326,
      "step": 249970
    },
    {
      "epoch": 0.06055323104178054,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001500042,
      "loss": 3.0214,
      "step": 249980
    },
    {
      "epoch": 0.060571874277938725,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001500022,
      "loss": 3.0679,
      "step": 249990
    },
    {
      "epoch": 0.06059051751409691,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001500002,
      "loss": 3.1433,
      "step": 250000
    },
    {
      "epoch": 0.06060916075025509,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001499982,
      "loss": 3.235,
      "step": 250010
    },
    {
      "epoch": 0.06062780398641328,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014999620000000002,
      "loss": 3.1361,
      "step": 250020
    },
    {
      "epoch": 0.06064644722257146,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014999420000000001,
      "loss": 3.1957,
      "step": 250030
    },
    {
      "epoch": 0.06066509045872964,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001499922,
      "loss": 3.1662,
      "step": 250040
    },
    {
      "epoch": 0.060683733694887824,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001499902,
      "loss": 3.0907,
      "step": 250050
    },
    {
      "epoch": 0.06070237693104601,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001499882,
      "loss": 3.1419,
      "step": 250060
    },
    {
      "epoch": 0.060721020167204195,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001499862,
      "loss": 3.0767,
      "step": 250070
    },
    {
      "epoch": 0.06073966340336238,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001499842,
      "loss": 3.1065,
      "step": 250080
    },
    {
      "epoch": 0.06075830663952056,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014998220000000002,
      "loss": 3.0331,
      "step": 250090
    },
    {
      "epoch": 0.06077694987567874,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001499802,
      "loss": 3.0751,
      "step": 250100
    },
    {
      "epoch": 0.06079559311183693,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001499782,
      "loss": 3.1386,
      "step": 250110
    },
    {
      "epoch": 0.06081423634799511,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001499762,
      "loss": 3.2431,
      "step": 250120
    },
    {
      "epoch": 0.060832879584153295,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001499742,
      "loss": 3.1283,
      "step": 250130
    },
    {
      "epoch": 0.06085152282031148,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001499722,
      "loss": 3.1648,
      "step": 250140
    },
    {
      "epoch": 0.060870166056469666,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014997020000000002,
      "loss": 3.0747,
      "step": 250150
    },
    {
      "epoch": 0.06088880929262785,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001499682,
      "loss": 3.2034,
      "step": 250160
    },
    {
      "epoch": 0.06090745252878603,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001499662,
      "loss": 2.9937,
      "step": 250170
    },
    {
      "epoch": 0.06092609576494421,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001499642,
      "loss": 3.0551,
      "step": 250180
    },
    {
      "epoch": 0.0609447390011024,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001499622,
      "loss": 3.0834,
      "step": 250190
    },
    {
      "epoch": 0.06096338223726058,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014996020000000002,
      "loss": 3.2091,
      "step": 250200
    },
    {
      "epoch": 0.060982025473418765,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014995820000000002,
      "loss": 3.0448,
      "step": 250210
    },
    {
      "epoch": 0.06100066870957695,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001499562,
      "loss": 3.0432,
      "step": 250220
    },
    {
      "epoch": 0.061019311945735136,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001499542,
      "loss": 3.181,
      "step": 250230
    },
    {
      "epoch": 0.06103795518189332,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001499522,
      "loss": 3.1887,
      "step": 250240
    },
    {
      "epoch": 0.0610565984180515,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001499502,
      "loss": 3.1332,
      "step": 250250
    },
    {
      "epoch": 0.06107524165420968,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001499482,
      "loss": 3.1336,
      "step": 250260
    },
    {
      "epoch": 0.061093884890367864,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014994620000000002,
      "loss": 3.2144,
      "step": 250270
    },
    {
      "epoch": 0.06111252812652605,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00014994420000000001,
      "loss": 3.2158,
      "step": 250280
    },
    {
      "epoch": 0.061131171362684235,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001499422,
      "loss": 3.1715,
      "step": 250290
    },
    {
      "epoch": 0.06114981459884242,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001499402,
      "loss": 3.0551,
      "step": 250300
    },
    {
      "epoch": 0.0611684578350006,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001499382,
      "loss": 3.1877,
      "step": 250310
    },
    {
      "epoch": 0.06118710107115879,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014993620000000002,
      "loss": 3.1827,
      "step": 250320
    },
    {
      "epoch": 0.06120574430731697,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001499342,
      "loss": 3.0896,
      "step": 250330
    },
    {
      "epoch": 0.06122438754347515,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014993220000000001,
      "loss": 3.07,
      "step": 250340
    },
    {
      "epoch": 0.061243030779633334,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001499302,
      "loss": 3.1105,
      "step": 250350
    },
    {
      "epoch": 0.06126167401579152,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001499282,
      "loss": 3.1473,
      "step": 250360
    },
    {
      "epoch": 0.061280317251949705,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014992620000000003,
      "loss": 3.0707,
      "step": 250370
    },
    {
      "epoch": 0.06129896048810789,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001499242,
      "loss": 3.1256,
      "step": 250380
    },
    {
      "epoch": 0.06131760372426607,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014992220000000002,
      "loss": 3.1701,
      "step": 250390
    },
    {
      "epoch": 0.06133624696042426,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001499202,
      "loss": 3.1205,
      "step": 250400
    },
    {
      "epoch": 0.06135489019658244,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001499182,
      "loss": 3.0153,
      "step": 250410
    },
    {
      "epoch": 0.06137353343274062,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001499162,
      "loss": 3.1723,
      "step": 250420
    },
    {
      "epoch": 0.061392176668898804,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001499142,
      "loss": 3.16,
      "step": 250430
    },
    {
      "epoch": 0.061410819905056986,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014991220000000002,
      "loss": 3.0365,
      "step": 250440
    },
    {
      "epoch": 0.061429463141215175,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001499102,
      "loss": 3.1678,
      "step": 250450
    },
    {
      "epoch": 0.06144810637737336,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014990820000000002,
      "loss": 3.0834,
      "step": 250460
    },
    {
      "epoch": 0.06146674961353154,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014990619999999998,
      "loss": 3.2514,
      "step": 250470
    },
    {
      "epoch": 0.06148539284968972,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001499042,
      "loss": 3.0876,
      "step": 250480
    },
    {
      "epoch": 0.06150403608584791,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001499022,
      "loss": 3.1697,
      "step": 250490
    },
    {
      "epoch": 0.06152267932200609,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001499002,
      "loss": 3.0995,
      "step": 250500
    },
    {
      "epoch": 0.061541322558164274,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014989820000000002,
      "loss": 3.128,
      "step": 250510
    },
    {
      "epoch": 0.061559965794322456,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001498962,
      "loss": 3.1351,
      "step": 250520
    },
    {
      "epoch": 0.061578609030480645,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001498942,
      "loss": 3.1828,
      "step": 250530
    },
    {
      "epoch": 0.06159725226663883,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001498922,
      "loss": 2.9281,
      "step": 250540
    },
    {
      "epoch": 0.06161589550279701,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001498902,
      "loss": 3.1248,
      "step": 250550
    },
    {
      "epoch": 0.06163453873895519,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001498882,
      "loss": 3.1226,
      "step": 250560
    },
    {
      "epoch": 0.06165318197511338,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001498862,
      "loss": 3.2241,
      "step": 250570
    },
    {
      "epoch": 0.06167182521127156,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014988420000000002,
      "loss": 3.122,
      "step": 250580
    },
    {
      "epoch": 0.061690468447429744,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001498822,
      "loss": 3.2461,
      "step": 250590
    },
    {
      "epoch": 0.061709111683587926,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001498802,
      "loss": 3.1345,
      "step": 250600
    },
    {
      "epoch": 0.06172775491974611,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001498782,
      "loss": 3.1734,
      "step": 250610
    },
    {
      "epoch": 0.0617463981559043,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001498762,
      "loss": 3.2154,
      "step": 250620
    },
    {
      "epoch": 0.06176504139206248,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001498742,
      "loss": 3.1444,
      "step": 250630
    },
    {
      "epoch": 0.06178368462822066,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014987220000000002,
      "loss": 3.1854,
      "step": 250640
    },
    {
      "epoch": 0.06180232786437884,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014987020000000001,
      "loss": 3.1817,
      "step": 250650
    },
    {
      "epoch": 0.06182097110053703,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001498682,
      "loss": 3.1879,
      "step": 250660
    },
    {
      "epoch": 0.061839614336695214,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001498662,
      "loss": 3.1149,
      "step": 250670
    },
    {
      "epoch": 0.061858257572853396,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001498642,
      "loss": 3.1272,
      "step": 250680
    },
    {
      "epoch": 0.06187690080901158,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001498622,
      "loss": 3.1811,
      "step": 250690
    },
    {
      "epoch": 0.06189554404516977,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014986020000000002,
      "loss": 3.0633,
      "step": 250700
    },
    {
      "epoch": 0.06191418728132795,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014985820000000001,
      "loss": 3.1593,
      "step": 250710
    },
    {
      "epoch": 0.06193283051748613,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001498562,
      "loss": 3.1002,
      "step": 250720
    },
    {
      "epoch": 0.061951473753644314,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001498542,
      "loss": 3.2919,
      "step": 250730
    },
    {
      "epoch": 0.0619701169898025,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001498522,
      "loss": 3.1052,
      "step": 250740
    },
    {
      "epoch": 0.061988760225960685,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001498502,
      "loss": 3.1708,
      "step": 250750
    },
    {
      "epoch": 0.06200740346211887,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014984820000000002,
      "loss": 3.2165,
      "step": 250760
    },
    {
      "epoch": 0.06202604669827705,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014984620000000002,
      "loss": 3.0873,
      "step": 250770
    },
    {
      "epoch": 0.06204468993443523,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001498442,
      "loss": 3.1964,
      "step": 250780
    },
    {
      "epoch": 0.06206333317059342,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001498422,
      "loss": 3.1609,
      "step": 250790
    },
    {
      "epoch": 0.0620819764067516,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001498402,
      "loss": 3.0806,
      "step": 250800
    },
    {
      "epoch": 0.062100619642909784,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014983820000000003,
      "loss": 3.1126,
      "step": 250810
    },
    {
      "epoch": 0.062119262879067966,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001498362,
      "loss": 3.2092,
      "step": 250820
    },
    {
      "epoch": 0.062137906115226155,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014983420000000002,
      "loss": 3.238,
      "step": 250830
    },
    {
      "epoch": 0.06215654935138434,
      "grad_norm": 0.75,
      "learning_rate": 0.0001498322,
      "loss": 3.1596,
      "step": 250840
    },
    {
      "epoch": 0.06217519258754252,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001498302,
      "loss": 3.2339,
      "step": 250850
    },
    {
      "epoch": 0.0621938358237007,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001498282,
      "loss": 3.1281,
      "step": 250860
    },
    {
      "epoch": 0.06221247905985889,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001498262,
      "loss": 3.1155,
      "step": 250870
    },
    {
      "epoch": 0.06223112229601707,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014982420000000002,
      "loss": 3.2004,
      "step": 250880
    },
    {
      "epoch": 0.062249765532175254,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001498222,
      "loss": 3.1024,
      "step": 250890
    },
    {
      "epoch": 0.062268408768333436,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001498202,
      "loss": 3.1187,
      "step": 250900
    },
    {
      "epoch": 0.062287052004491625,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001498182,
      "loss": 3.0417,
      "step": 250910
    },
    {
      "epoch": 0.06230569524064981,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001498162,
      "loss": 3.2055,
      "step": 250920
    },
    {
      "epoch": 0.06232433847680799,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014981420000000003,
      "loss": 3.0849,
      "step": 250930
    },
    {
      "epoch": 0.06234298171296617,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001498122,
      "loss": 3.2133,
      "step": 250940
    },
    {
      "epoch": 0.06236162494912435,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014981020000000002,
      "loss": 3.1248,
      "step": 250950
    },
    {
      "epoch": 0.06238026818528254,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014980819999999999,
      "loss": 3.144,
      "step": 250960
    },
    {
      "epoch": 0.062398911421440724,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001498062,
      "loss": 3.3267,
      "step": 250970
    },
    {
      "epoch": 0.062417554657598906,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001498042,
      "loss": 3.1069,
      "step": 250980
    },
    {
      "epoch": 0.06243619789375709,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001498022,
      "loss": 3.0469,
      "step": 250990
    },
    {
      "epoch": 0.06245484112991528,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014980020000000002,
      "loss": 3.2411,
      "step": 251000
    },
    {
      "epoch": 0.06247348436607346,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001497982,
      "loss": 3.1573,
      "step": 251010
    },
    {
      "epoch": 0.06249212760223164,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014979620000000001,
      "loss": 3.181,
      "step": 251020
    },
    {
      "epoch": 0.06251077083838982,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001497942,
      "loss": 3.2088,
      "step": 251030
    },
    {
      "epoch": 0.06252941407454801,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001497922,
      "loss": 3.1954,
      "step": 251040
    },
    {
      "epoch": 0.06254805731070619,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001497902,
      "loss": 3.1567,
      "step": 251050
    },
    {
      "epoch": 0.06256670054686438,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001497882,
      "loss": 3.1956,
      "step": 251060
    },
    {
      "epoch": 0.06258534378302257,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014978620000000002,
      "loss": 3.0146,
      "step": 251070
    },
    {
      "epoch": 0.06260398701918074,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001497842,
      "loss": 3.1835,
      "step": 251080
    },
    {
      "epoch": 0.06262263025533893,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001497822,
      "loss": 2.9188,
      "step": 251090
    },
    {
      "epoch": 0.06264127349149712,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001497802,
      "loss": 3.091,
      "step": 251100
    },
    {
      "epoch": 0.0626599167276553,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001497782,
      "loss": 3.0905,
      "step": 251110
    },
    {
      "epoch": 0.06267855996381348,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001497762,
      "loss": 3.1993,
      "step": 251120
    },
    {
      "epoch": 0.06269720319997166,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001497742,
      "loss": 3.2019,
      "step": 251130
    },
    {
      "epoch": 0.06271584643612985,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014977220000000002,
      "loss": 3.1532,
      "step": 251140
    },
    {
      "epoch": 0.06273448967228804,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001497702,
      "loss": 3.0971,
      "step": 251150
    },
    {
      "epoch": 0.06275313290844621,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001497682,
      "loss": 3.1682,
      "step": 251160
    },
    {
      "epoch": 0.0627717761446044,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001497662,
      "loss": 3.1024,
      "step": 251170
    },
    {
      "epoch": 0.06279041938076257,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001497642,
      "loss": 3.095,
      "step": 251180
    },
    {
      "epoch": 0.06280906261692076,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014976220000000002,
      "loss": 3.0933,
      "step": 251190
    },
    {
      "epoch": 0.06282770585307895,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014976020000000002,
      "loss": 3.1759,
      "step": 251200
    },
    {
      "epoch": 0.06284634908923713,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001497582,
      "loss": 3.1162,
      "step": 251210
    },
    {
      "epoch": 0.06286499232539532,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001497562,
      "loss": 3.1331,
      "step": 251220
    },
    {
      "epoch": 0.0628836355615535,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001497542,
      "loss": 3.1177,
      "step": 251230
    },
    {
      "epoch": 0.06290227879771168,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001497522,
      "loss": 3.1481,
      "step": 251240
    },
    {
      "epoch": 0.06292092203386987,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014975020000000002,
      "loss": 3.105,
      "step": 251250
    },
    {
      "epoch": 0.06293956527002804,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014974820000000002,
      "loss": 3.0592,
      "step": 251260
    },
    {
      "epoch": 0.06295820850618623,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001497462,
      "loss": 3.1168,
      "step": 251270
    },
    {
      "epoch": 0.06297685174234442,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001497442,
      "loss": 3.2298,
      "step": 251280
    },
    {
      "epoch": 0.0629954949785026,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001497422,
      "loss": 3.0954,
      "step": 251290
    },
    {
      "epoch": 0.06301413821466079,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001497402,
      "loss": 3.1457,
      "step": 251300
    },
    {
      "epoch": 0.06303278145081896,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001497382,
      "loss": 3.2281,
      "step": 251310
    },
    {
      "epoch": 0.06305142468697715,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014973620000000002,
      "loss": 3.0433,
      "step": 251320
    },
    {
      "epoch": 0.06307006792313534,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014973420000000001,
      "loss": 3.1326,
      "step": 251330
    },
    {
      "epoch": 0.06308871115929351,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001497322,
      "loss": 3.1189,
      "step": 251340
    },
    {
      "epoch": 0.0631073543954517,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001497302,
      "loss": 3.1196,
      "step": 251350
    },
    {
      "epoch": 0.06312599763160989,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001497282,
      "loss": 3.0313,
      "step": 251360
    },
    {
      "epoch": 0.06314464086776807,
      "grad_norm": 0.75,
      "learning_rate": 0.00014972620000000002,
      "loss": 3.1188,
      "step": 251370
    },
    {
      "epoch": 0.06316328410392626,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001497242,
      "loss": 3.0682,
      "step": 251380
    },
    {
      "epoch": 0.06318192734008443,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014972220000000001,
      "loss": 3.3486,
      "step": 251390
    },
    {
      "epoch": 0.06320057057624262,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001497202,
      "loss": 3.0293,
      "step": 251400
    },
    {
      "epoch": 0.06321921381240081,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001497182,
      "loss": 3.2392,
      "step": 251410
    },
    {
      "epoch": 0.06323785704855898,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014971620000000003,
      "loss": 2.9751,
      "step": 251420
    },
    {
      "epoch": 0.06325650028471717,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001497142,
      "loss": 3.1824,
      "step": 251430
    },
    {
      "epoch": 0.06327514352087536,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014971220000000002,
      "loss": 3.133,
      "step": 251440
    },
    {
      "epoch": 0.06329378675703354,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001497102,
      "loss": 3.068,
      "step": 251450
    },
    {
      "epoch": 0.06331242999319173,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001497082,
      "loss": 3.1539,
      "step": 251460
    },
    {
      "epoch": 0.0633310732293499,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001497062,
      "loss": 3.0853,
      "step": 251470
    },
    {
      "epoch": 0.06334971646550809,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001497042,
      "loss": 3.0383,
      "step": 251480
    },
    {
      "epoch": 0.06336835970166628,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014970220000000003,
      "loss": 3.1383,
      "step": 251490
    },
    {
      "epoch": 0.06338700293782445,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001497002,
      "loss": 3.1528,
      "step": 251500
    },
    {
      "epoch": 0.06340564617398264,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014969820000000002,
      "loss": 3.2007,
      "step": 251510
    },
    {
      "epoch": 0.06342428941014082,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014969619999999998,
      "loss": 3.1168,
      "step": 251520
    },
    {
      "epoch": 0.06344293264629901,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001496942,
      "loss": 3.1562,
      "step": 251530
    },
    {
      "epoch": 0.0634615758824572,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001496922,
      "loss": 2.9832,
      "step": 251540
    },
    {
      "epoch": 0.06348021911861537,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001496902,
      "loss": 3.195,
      "step": 251550
    },
    {
      "epoch": 0.06349886235477356,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014968820000000002,
      "loss": 3.0738,
      "step": 251560
    },
    {
      "epoch": 0.06351750559093175,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001496862,
      "loss": 3.0293,
      "step": 251570
    },
    {
      "epoch": 0.06353614882708992,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001496842,
      "loss": 3.1782,
      "step": 251580
    },
    {
      "epoch": 0.06355479206324811,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001496822,
      "loss": 3.1266,
      "step": 251590
    },
    {
      "epoch": 0.06357343529940629,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001496802,
      "loss": 3.191,
      "step": 251600
    },
    {
      "epoch": 0.06359207853556448,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001496782,
      "loss": 3.0586,
      "step": 251610
    },
    {
      "epoch": 0.06361072177172267,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001496762,
      "loss": 3.0826,
      "step": 251620
    },
    {
      "epoch": 0.06362936500788084,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014967420000000002,
      "loss": 3.1565,
      "step": 251630
    },
    {
      "epoch": 0.06364800824403903,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001496722,
      "loss": 3.1146,
      "step": 251640
    },
    {
      "epoch": 0.0636666514801972,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001496702,
      "loss": 3.0805,
      "step": 251650
    },
    {
      "epoch": 0.0636852947163554,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001496682,
      "loss": 3.1184,
      "step": 251660
    },
    {
      "epoch": 0.06370393795251358,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001496662,
      "loss": 3.1615,
      "step": 251670
    },
    {
      "epoch": 0.06372258118867176,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014966420000000002,
      "loss": 3.158,
      "step": 251680
    },
    {
      "epoch": 0.06374122442482995,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001496622,
      "loss": 3.1841,
      "step": 251690
    },
    {
      "epoch": 0.06375986766098814,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014966020000000001,
      "loss": 3.1665,
      "step": 251700
    },
    {
      "epoch": 0.06377851089714631,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001496582,
      "loss": 3.0792,
      "step": 251710
    },
    {
      "epoch": 0.0637971541333045,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001496562,
      "loss": 3.2637,
      "step": 251720
    },
    {
      "epoch": 0.06381579736946268,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001496542,
      "loss": 3.0203,
      "step": 251730
    },
    {
      "epoch": 0.06383444060562087,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001496522,
      "loss": 3.0816,
      "step": 251740
    },
    {
      "epoch": 0.06385308384177905,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014965020000000002,
      "loss": 3.1846,
      "step": 251750
    },
    {
      "epoch": 0.06387172707793723,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014964820000000002,
      "loss": 3.0946,
      "step": 251760
    },
    {
      "epoch": 0.06389037031409542,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001496462,
      "loss": 3.2187,
      "step": 251770
    },
    {
      "epoch": 0.06390901355025361,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001496442,
      "loss": 3.1182,
      "step": 251780
    },
    {
      "epoch": 0.06392765678641178,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001496422,
      "loss": 3.1719,
      "step": 251790
    },
    {
      "epoch": 0.06394630002256997,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001496402,
      "loss": 3.0789,
      "step": 251800
    },
    {
      "epoch": 0.06396494325872815,
      "grad_norm": 0.75,
      "learning_rate": 0.00014963820000000002,
      "loss": 3.2282,
      "step": 251810
    },
    {
      "epoch": 0.06398358649488634,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014963620000000002,
      "loss": 3.1959,
      "step": 251820
    },
    {
      "epoch": 0.06400222973104452,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001496342,
      "loss": 3.2183,
      "step": 251830
    },
    {
      "epoch": 0.0640208729672027,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001496322,
      "loss": 3.097,
      "step": 251840
    },
    {
      "epoch": 0.06403951620336089,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001496302,
      "loss": 3.0766,
      "step": 251850
    },
    {
      "epoch": 0.06405815943951906,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001496282,
      "loss": 3.1396,
      "step": 251860
    },
    {
      "epoch": 0.06407680267567725,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001496262,
      "loss": 3.033,
      "step": 251870
    },
    {
      "epoch": 0.06409544591183544,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014962420000000002,
      "loss": 3.0844,
      "step": 251880
    },
    {
      "epoch": 0.06411408914799362,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001496222,
      "loss": 3.117,
      "step": 251890
    },
    {
      "epoch": 0.0641327323841518,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001496202,
      "loss": 3.0893,
      "step": 251900
    },
    {
      "epoch": 0.06415137562031,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001496182,
      "loss": 3.1747,
      "step": 251910
    },
    {
      "epoch": 0.06417001885646817,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001496162,
      "loss": 2.9967,
      "step": 251920
    },
    {
      "epoch": 0.06418866209262636,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014961420000000002,
      "loss": 3.0895,
      "step": 251930
    },
    {
      "epoch": 0.06420730532878453,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001496122,
      "loss": 3.1918,
      "step": 251940
    },
    {
      "epoch": 0.06422594856494272,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001496102,
      "loss": 3.2331,
      "step": 251950
    },
    {
      "epoch": 0.06424459180110091,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001496082,
      "loss": 3.0859,
      "step": 251960
    },
    {
      "epoch": 0.06426323503725909,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001496062,
      "loss": 3.115,
      "step": 251970
    },
    {
      "epoch": 0.06428187827341728,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014960420000000003,
      "loss": 3.1439,
      "step": 251980
    },
    {
      "epoch": 0.06430052150957545,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001496022,
      "loss": 3.1743,
      "step": 251990
    },
    {
      "epoch": 0.06431916474573364,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014960020000000002,
      "loss": 3.0392,
      "step": 252000
    },
    {
      "epoch": 0.06433780798189183,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001495982,
      "loss": 3.2771,
      "step": 252010
    },
    {
      "epoch": 0.06435645121805,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001495962,
      "loss": 3.2228,
      "step": 252020
    },
    {
      "epoch": 0.06437509445420819,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001495942,
      "loss": 3.1916,
      "step": 252030
    },
    {
      "epoch": 0.06439373769036638,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001495922,
      "loss": 3.0968,
      "step": 252040
    },
    {
      "epoch": 0.06441238092652456,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014959020000000002,
      "loss": 3.0805,
      "step": 252050
    },
    {
      "epoch": 0.06443102416268275,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001495882,
      "loss": 3.192,
      "step": 252060
    },
    {
      "epoch": 0.06444966739884092,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014958620000000001,
      "loss": 2.9868,
      "step": 252070
    },
    {
      "epoch": 0.06446831063499911,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014958419999999998,
      "loss": 3.0361,
      "step": 252080
    },
    {
      "epoch": 0.0644869538711573,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001495822,
      "loss": 3.1138,
      "step": 252090
    },
    {
      "epoch": 0.06450559710731547,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001495802,
      "loss": 3.1592,
      "step": 252100
    },
    {
      "epoch": 0.06452424034347366,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001495782,
      "loss": 3.0729,
      "step": 252110
    },
    {
      "epoch": 0.06454288357963185,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014957620000000002,
      "loss": 3.0845,
      "step": 252120
    },
    {
      "epoch": 0.06456152681579003,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001495742,
      "loss": 3.0773,
      "step": 252130
    },
    {
      "epoch": 0.06458017005194822,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001495722,
      "loss": 3.248,
      "step": 252140
    },
    {
      "epoch": 0.06459881328810639,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001495702,
      "loss": 3.1214,
      "step": 252150
    },
    {
      "epoch": 0.06461745652426458,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001495682,
      "loss": 3.1821,
      "step": 252160
    },
    {
      "epoch": 0.06463609976042277,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014956620000000002,
      "loss": 3.2388,
      "step": 252170
    },
    {
      "epoch": 0.06465474299658094,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001495642,
      "loss": 3.0301,
      "step": 252180
    },
    {
      "epoch": 0.06467338623273913,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014956220000000002,
      "loss": 3.3404,
      "step": 252190
    },
    {
      "epoch": 0.06469202946889731,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001495602,
      "loss": 3.1663,
      "step": 252200
    },
    {
      "epoch": 0.0647106727050555,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001495582,
      "loss": 3.1457,
      "step": 252210
    },
    {
      "epoch": 0.06472931594121369,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001495562,
      "loss": 3.1189,
      "step": 252220
    },
    {
      "epoch": 0.06474795917737186,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001495542,
      "loss": 3.0932,
      "step": 252230
    },
    {
      "epoch": 0.06476660241353005,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00014955220000000002,
      "loss": 3.1499,
      "step": 252240
    },
    {
      "epoch": 0.06478524564968824,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014955020000000002,
      "loss": 3.1899,
      "step": 252250
    },
    {
      "epoch": 0.06480388888584641,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001495482,
      "loss": 3.1532,
      "step": 252260
    },
    {
      "epoch": 0.0648225321220046,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001495462,
      "loss": 3.0447,
      "step": 252270
    },
    {
      "epoch": 0.06484117535816278,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001495442,
      "loss": 3.1043,
      "step": 252280
    },
    {
      "epoch": 0.06485981859432097,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001495422,
      "loss": 3.082,
      "step": 252290
    },
    {
      "epoch": 0.06487846183047916,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001495402,
      "loss": 3.1552,
      "step": 252300
    },
    {
      "epoch": 0.06489710506663733,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014953820000000002,
      "loss": 3.2394,
      "step": 252310
    },
    {
      "epoch": 0.06491574830279552,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001495362,
      "loss": 3.1852,
      "step": 252320
    },
    {
      "epoch": 0.0649343915389537,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001495342,
      "loss": 3.0258,
      "step": 252330
    },
    {
      "epoch": 0.06495303477511188,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001495322,
      "loss": 3.2854,
      "step": 252340
    },
    {
      "epoch": 0.06497167801127007,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001495302,
      "loss": 3.2229,
      "step": 252350
    },
    {
      "epoch": 0.06499032124742825,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001495282,
      "loss": 3.1495,
      "step": 252360
    },
    {
      "epoch": 0.06500896448358644,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014952620000000002,
      "loss": 3.1307,
      "step": 252370
    },
    {
      "epoch": 0.06502760771974463,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014952420000000001,
      "loss": 3.1447,
      "step": 252380
    },
    {
      "epoch": 0.0650462509559028,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001495222,
      "loss": 3.187,
      "step": 252390
    },
    {
      "epoch": 0.06506489419206099,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001495202,
      "loss": 3.0429,
      "step": 252400
    },
    {
      "epoch": 0.06508353742821917,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001495182,
      "loss": 3.1437,
      "step": 252410
    },
    {
      "epoch": 0.06510218066437735,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014951620000000002,
      "loss": 2.9492,
      "step": 252420
    },
    {
      "epoch": 0.06512082390053554,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001495142,
      "loss": 3.1968,
      "step": 252430
    },
    {
      "epoch": 0.06513946713669372,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014951220000000002,
      "loss": 3.1502,
      "step": 252440
    },
    {
      "epoch": 0.06515811037285191,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001495102,
      "loss": 3.1949,
      "step": 252450
    },
    {
      "epoch": 0.0651767536090101,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001495082,
      "loss": 3.1536,
      "step": 252460
    },
    {
      "epoch": 0.06519539684516827,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001495062,
      "loss": 3.1853,
      "step": 252470
    },
    {
      "epoch": 0.06521404008132646,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001495042,
      "loss": 3.1103,
      "step": 252480
    },
    {
      "epoch": 0.06523268331748464,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014950220000000002,
      "loss": 3.187,
      "step": 252490
    },
    {
      "epoch": 0.06525132655364282,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001495002,
      "loss": 3.1691,
      "step": 252500
    },
    {
      "epoch": 0.06526996978980101,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001494982,
      "loss": 3.1946,
      "step": 252510
    },
    {
      "epoch": 0.06528861302595919,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001494962,
      "loss": 3.1637,
      "step": 252520
    },
    {
      "epoch": 0.06530725626211738,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001494942,
      "loss": 3.1071,
      "step": 252530
    },
    {
      "epoch": 0.06532589949827555,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014949220000000003,
      "loss": 3.0543,
      "step": 252540
    },
    {
      "epoch": 0.06534454273443374,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001494902,
      "loss": 3.1067,
      "step": 252550
    },
    {
      "epoch": 0.06536318597059193,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014948820000000002,
      "loss": 3.128,
      "step": 252560
    },
    {
      "epoch": 0.0653818292067501,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014948619999999999,
      "loss": 3.1113,
      "step": 252570
    },
    {
      "epoch": 0.0654004724429083,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001494842,
      "loss": 2.9985,
      "step": 252580
    },
    {
      "epoch": 0.06541911567906648,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001494822,
      "loss": 3.1514,
      "step": 252590
    },
    {
      "epoch": 0.06543775891522466,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001494802,
      "loss": 3.2539,
      "step": 252600
    },
    {
      "epoch": 0.06545640215138285,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014947820000000002,
      "loss": 3.0407,
      "step": 252610
    },
    {
      "epoch": 0.06547504538754102,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001494762,
      "loss": 3.2231,
      "step": 252620
    },
    {
      "epoch": 0.06549368862369921,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001494742,
      "loss": 3.1494,
      "step": 252630
    },
    {
      "epoch": 0.0655123318598574,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001494722,
      "loss": 3.0381,
      "step": 252640
    },
    {
      "epoch": 0.06553097509601558,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001494702,
      "loss": 3.1924,
      "step": 252650
    },
    {
      "epoch": 0.06554961833217376,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014946820000000003,
      "loss": 3.1981,
      "step": 252660
    },
    {
      "epoch": 0.06556826156833194,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001494662,
      "loss": 3.0601,
      "step": 252670
    },
    {
      "epoch": 0.06558690480449013,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014946420000000002,
      "loss": 3.1337,
      "step": 252680
    },
    {
      "epoch": 0.06560554804064832,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001494622,
      "loss": 3.2336,
      "step": 252690
    },
    {
      "epoch": 0.06562419127680649,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001494602,
      "loss": 3.1897,
      "step": 252700
    },
    {
      "epoch": 0.06564283451296468,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001494582,
      "loss": 3.0935,
      "step": 252710
    },
    {
      "epoch": 0.06566147774912287,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001494562,
      "loss": 3.0359,
      "step": 252720
    },
    {
      "epoch": 0.06568012098528105,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014945420000000002,
      "loss": 3.1227,
      "step": 252730
    },
    {
      "epoch": 0.06569876422143923,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001494522,
      "loss": 3.173,
      "step": 252740
    },
    {
      "epoch": 0.06571740745759741,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014945020000000001,
      "loss": 3.1604,
      "step": 252750
    },
    {
      "epoch": 0.0657360506937556,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001494482,
      "loss": 3.1876,
      "step": 252760
    },
    {
      "epoch": 0.06575469392991379,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001494462,
      "loss": 3.0395,
      "step": 252770
    },
    {
      "epoch": 0.06577333716607196,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001494442,
      "loss": 3.1403,
      "step": 252780
    },
    {
      "epoch": 0.06579198040223015,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001494422,
      "loss": 3.0848,
      "step": 252790
    },
    {
      "epoch": 0.06581062363838834,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014944020000000002,
      "loss": 3.3456,
      "step": 252800
    },
    {
      "epoch": 0.06582926687454652,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014943820000000002,
      "loss": 3.15,
      "step": 252810
    },
    {
      "epoch": 0.0658479101107047,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001494362,
      "loss": 3.1137,
      "step": 252820
    },
    {
      "epoch": 0.06586655334686288,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001494342,
      "loss": 3.1609,
      "step": 252830
    },
    {
      "epoch": 0.06588519658302107,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001494322,
      "loss": 3.1176,
      "step": 252840
    },
    {
      "epoch": 0.06590383981917926,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001494302,
      "loss": 3.1599,
      "step": 252850
    },
    {
      "epoch": 0.06592248305533743,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014942820000000002,
      "loss": 3.0819,
      "step": 252860
    },
    {
      "epoch": 0.06594112629149562,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014942620000000002,
      "loss": 3.1067,
      "step": 252870
    },
    {
      "epoch": 0.0659597695276538,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001494242,
      "loss": 3.2188,
      "step": 252880
    },
    {
      "epoch": 0.06597841276381199,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001494222,
      "loss": 3.1489,
      "step": 252890
    },
    {
      "epoch": 0.06599705599997018,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001494202,
      "loss": 3.0808,
      "step": 252900
    },
    {
      "epoch": 0.06601569923612835,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001494182,
      "loss": 3.0606,
      "step": 252910
    },
    {
      "epoch": 0.06603434247228654,
      "grad_norm": 0.75,
      "learning_rate": 0.0001494162,
      "loss": 3.1022,
      "step": 252920
    },
    {
      "epoch": 0.06605298570844473,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014941420000000002,
      "loss": 3.1208,
      "step": 252930
    },
    {
      "epoch": 0.0660716289446029,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001494122,
      "loss": 2.9969,
      "step": 252940
    },
    {
      "epoch": 0.06609027218076109,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001494102,
      "loss": 3.1143,
      "step": 252950
    },
    {
      "epoch": 0.06610891541691927,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001494082,
      "loss": 3.1453,
      "step": 252960
    },
    {
      "epoch": 0.06612755865307746,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001494062,
      "loss": 3.1355,
      "step": 252970
    },
    {
      "epoch": 0.06614620188923565,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014940420000000002,
      "loss": 3.1466,
      "step": 252980
    },
    {
      "epoch": 0.06616484512539382,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001494022,
      "loss": 3.0878,
      "step": 252990
    },
    {
      "epoch": 0.06618348836155201,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001494002,
      "loss": 3.0547,
      "step": 253000
    },
    {
      "epoch": 0.06620213159771018,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001493982,
      "loss": 3.1126,
      "step": 253010
    },
    {
      "epoch": 0.06622077483386837,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001493962,
      "loss": 3.0832,
      "step": 253020
    },
    {
      "epoch": 0.06623941807002656,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014939420000000003,
      "loss": 2.9511,
      "step": 253030
    },
    {
      "epoch": 0.06625806130618474,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001493922,
      "loss": 3.1921,
      "step": 253040
    },
    {
      "epoch": 0.06627670454234293,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014939020000000002,
      "loss": 3.1512,
      "step": 253050
    },
    {
      "epoch": 0.06629534777850112,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001493882,
      "loss": 3.2469,
      "step": 253060
    },
    {
      "epoch": 0.06631399101465929,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001493862,
      "loss": 3.1613,
      "step": 253070
    },
    {
      "epoch": 0.06633263425081748,
      "grad_norm": 0.625,
      "learning_rate": 0.0001493842,
      "loss": 3.2497,
      "step": 253080
    },
    {
      "epoch": 0.06635127748697565,
      "grad_norm": 0.75,
      "learning_rate": 0.0001493822,
      "loss": 2.9976,
      "step": 253090
    },
    {
      "epoch": 0.06636992072313384,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014938020000000002,
      "loss": 3.0459,
      "step": 253100
    },
    {
      "epoch": 0.06638856395929203,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001493782,
      "loss": 3.2352,
      "step": 253110
    },
    {
      "epoch": 0.06640720719545021,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014937620000000001,
      "loss": 3.1709,
      "step": 253120
    },
    {
      "epoch": 0.0664258504316084,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001493742,
      "loss": 3.1194,
      "step": 253130
    },
    {
      "epoch": 0.06644449366776659,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001493722,
      "loss": 3.1095,
      "step": 253140
    },
    {
      "epoch": 0.06646313690392476,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014937020000000003,
      "loss": 3.1918,
      "step": 253150
    },
    {
      "epoch": 0.06648178014008295,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001493682,
      "loss": 3.0617,
      "step": 253160
    },
    {
      "epoch": 0.06650042337624112,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014936620000000002,
      "loss": 3.2334,
      "step": 253170
    },
    {
      "epoch": 0.06651906661239931,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001493642,
      "loss": 3.0873,
      "step": 253180
    },
    {
      "epoch": 0.0665377098485575,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001493622,
      "loss": 3.1651,
      "step": 253190
    },
    {
      "epoch": 0.06655635308471568,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001493602,
      "loss": 3.1761,
      "step": 253200
    },
    {
      "epoch": 0.06657499632087387,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001493582,
      "loss": 3.02,
      "step": 253210
    },
    {
      "epoch": 0.06659363955703204,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014935620000000003,
      "loss": 3.1426,
      "step": 253220
    },
    {
      "epoch": 0.06661228279319023,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001493542,
      "loss": 3.0091,
      "step": 253230
    },
    {
      "epoch": 0.06663092602934842,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014935220000000002,
      "loss": 3.0473,
      "step": 253240
    },
    {
      "epoch": 0.0666495692655066,
      "grad_norm": 0.625,
      "learning_rate": 0.0001493502,
      "loss": 3.0734,
      "step": 253250
    },
    {
      "epoch": 0.06666821250166478,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001493482,
      "loss": 3.1425,
      "step": 253260
    },
    {
      "epoch": 0.06668685573782297,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001493462,
      "loss": 3.2005,
      "step": 253270
    },
    {
      "epoch": 0.06670549897398115,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001493442,
      "loss": 3.1438,
      "step": 253280
    },
    {
      "epoch": 0.06672414221013934,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014934220000000002,
      "loss": 3.0866,
      "step": 253290
    },
    {
      "epoch": 0.06674278544629751,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001493402,
      "loss": 3.2423,
      "step": 253300
    },
    {
      "epoch": 0.0667614286824557,
      "grad_norm": 0.75,
      "learning_rate": 0.0001493382,
      "loss": 3.3024,
      "step": 253310
    },
    {
      "epoch": 0.06678007191861389,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001493362,
      "loss": 3.0937,
      "step": 253320
    },
    {
      "epoch": 0.06679871515477206,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001493342,
      "loss": 3.1698,
      "step": 253330
    },
    {
      "epoch": 0.06681735839093025,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001493322,
      "loss": 3.1072,
      "step": 253340
    },
    {
      "epoch": 0.06683600162708843,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001493302,
      "loss": 3.0454,
      "step": 253350
    },
    {
      "epoch": 0.06685464486324662,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014932820000000002,
      "loss": 3.1548,
      "step": 253360
    },
    {
      "epoch": 0.0668732880994048,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014932620000000001,
      "loss": 3.235,
      "step": 253370
    },
    {
      "epoch": 0.06689193133556298,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001493242,
      "loss": 3.1253,
      "step": 253380
    },
    {
      "epoch": 0.06691057457172117,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001493222,
      "loss": 3.0525,
      "step": 253390
    },
    {
      "epoch": 0.06692921780787936,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001493202,
      "loss": 3.186,
      "step": 253400
    },
    {
      "epoch": 0.06694786104403753,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001493182,
      "loss": 3.1989,
      "step": 253410
    },
    {
      "epoch": 0.06696650428019572,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00014931620000000002,
      "loss": 3.066,
      "step": 253420
    },
    {
      "epoch": 0.0669851475163539,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014931420000000001,
      "loss": 3.1409,
      "step": 253430
    },
    {
      "epoch": 0.06700379075251209,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001493122,
      "loss": 3.1685,
      "step": 253440
    },
    {
      "epoch": 0.06702243398867028,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001493102,
      "loss": 3.0459,
      "step": 253450
    },
    {
      "epoch": 0.06704107722482845,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001493082,
      "loss": 3.1165,
      "step": 253460
    },
    {
      "epoch": 0.06705972046098664,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001493062,
      "loss": 3.1518,
      "step": 253470
    },
    {
      "epoch": 0.06707836369714483,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001493042,
      "loss": 3.2235,
      "step": 253480
    },
    {
      "epoch": 0.067097006933303,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014930220000000002,
      "loss": 3.0095,
      "step": 253490
    },
    {
      "epoch": 0.0671156501694612,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001493002,
      "loss": 2.9476,
      "step": 253500
    },
    {
      "epoch": 0.06713429340561937,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001492982,
      "loss": 3.0587,
      "step": 253510
    },
    {
      "epoch": 0.06715293664177756,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001492962,
      "loss": 3.0976,
      "step": 253520
    },
    {
      "epoch": 0.06717157987793575,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001492942,
      "loss": 3.1816,
      "step": 253530
    },
    {
      "epoch": 0.06719022311409392,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014929220000000002,
      "loss": 3.0935,
      "step": 253540
    },
    {
      "epoch": 0.06720886635025211,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001492902,
      "loss": 3.1192,
      "step": 253550
    },
    {
      "epoch": 0.06722750958641029,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001492882,
      "loss": 3.0297,
      "step": 253560
    },
    {
      "epoch": 0.06724615282256848,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001492862,
      "loss": 3.1353,
      "step": 253570
    },
    {
      "epoch": 0.06726479605872666,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001492842,
      "loss": 3.1964,
      "step": 253580
    },
    {
      "epoch": 0.06728343929488484,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014928220000000003,
      "loss": 3.0882,
      "step": 253590
    },
    {
      "epoch": 0.06730208253104303,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001492802,
      "loss": 3.0624,
      "step": 253600
    },
    {
      "epoch": 0.06732072576720122,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014927820000000002,
      "loss": 3.1096,
      "step": 253610
    },
    {
      "epoch": 0.06733936900335939,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001492762,
      "loss": 3.2746,
      "step": 253620
    },
    {
      "epoch": 0.06735801223951758,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001492742,
      "loss": 3.0481,
      "step": 253630
    },
    {
      "epoch": 0.06737665547567576,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014927220000000003,
      "loss": 3.2581,
      "step": 253640
    },
    {
      "epoch": 0.06739529871183395,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001492702,
      "loss": 3.1169,
      "step": 253650
    },
    {
      "epoch": 0.06741394194799213,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014926820000000002,
      "loss": 3.2388,
      "step": 253660
    },
    {
      "epoch": 0.06743258518415031,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001492662,
      "loss": 3.2283,
      "step": 253670
    },
    {
      "epoch": 0.0674512284203085,
      "grad_norm": 0.625,
      "learning_rate": 0.0001492642,
      "loss": 3.1052,
      "step": 253680
    },
    {
      "epoch": 0.06746987165646667,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001492622,
      "loss": 3.166,
      "step": 253690
    },
    {
      "epoch": 0.06748851489262486,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001492602,
      "loss": 3.1487,
      "step": 253700
    },
    {
      "epoch": 0.06750715812878305,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014925820000000003,
      "loss": 3.1217,
      "step": 253710
    },
    {
      "epoch": 0.06752580136494123,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001492562,
      "loss": 3.1269,
      "step": 253720
    },
    {
      "epoch": 0.06754444460109942,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014925420000000002,
      "loss": 3.0627,
      "step": 253730
    },
    {
      "epoch": 0.0675630878372576,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001492522,
      "loss": 3.0618,
      "step": 253740
    },
    {
      "epoch": 0.06758173107341578,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001492502,
      "loss": 3.1064,
      "step": 253750
    },
    {
      "epoch": 0.06760037430957397,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001492482,
      "loss": 3.1776,
      "step": 253760
    },
    {
      "epoch": 0.06761901754573214,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001492462,
      "loss": 3.0862,
      "step": 253770
    },
    {
      "epoch": 0.06763766078189033,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014924420000000002,
      "loss": 3.07,
      "step": 253780
    },
    {
      "epoch": 0.06765630401804852,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001492422,
      "loss": 3.1572,
      "step": 253790
    },
    {
      "epoch": 0.0676749472542067,
      "grad_norm": 0.578125,
      "learning_rate": 0.00014924020000000001,
      "loss": 3.07,
      "step": 253800
    },
    {
      "epoch": 0.06769359049036489,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001492382,
      "loss": 3.2343,
      "step": 253810
    },
    {
      "epoch": 0.06771223372652307,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001492362,
      "loss": 3.0792,
      "step": 253820
    },
    {
      "epoch": 0.06773087696268125,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001492342,
      "loss": 3.124,
      "step": 253830
    },
    {
      "epoch": 0.06774952019883944,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001492322,
      "loss": 3.0377,
      "step": 253840
    },
    {
      "epoch": 0.06776816343499761,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014923020000000002,
      "loss": 3.0768,
      "step": 253850
    },
    {
      "epoch": 0.0677868066711558,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014922820000000002,
      "loss": 3.1842,
      "step": 253860
    },
    {
      "epoch": 0.06780544990731399,
      "grad_norm": 0.625,
      "learning_rate": 0.0001492262,
      "loss": 2.9221,
      "step": 253870
    },
    {
      "epoch": 0.06782409314347217,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001492242,
      "loss": 2.965,
      "step": 253880
    },
    {
      "epoch": 0.06784273637963036,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001492222,
      "loss": 3.1193,
      "step": 253890
    },
    {
      "epoch": 0.06786137961578853,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001492202,
      "loss": 3.2644,
      "step": 253900
    },
    {
      "epoch": 0.06788002285194672,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001492182,
      "loss": 3.0947,
      "step": 253910
    },
    {
      "epoch": 0.06789866608810491,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014921620000000002,
      "loss": 3.0888,
      "step": 253920
    },
    {
      "epoch": 0.06791730932426308,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001492142,
      "loss": 3.0783,
      "step": 253930
    },
    {
      "epoch": 0.06793595256042127,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001492122,
      "loss": 3.1518,
      "step": 253940
    },
    {
      "epoch": 0.06795459579657946,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001492102,
      "loss": 3.1512,
      "step": 253950
    },
    {
      "epoch": 0.06797323903273764,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001492082,
      "loss": 3.1718,
      "step": 253960
    },
    {
      "epoch": 0.06799188226889583,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001492062,
      "loss": 3.154,
      "step": 253970
    },
    {
      "epoch": 0.068010525505054,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014920420000000002,
      "loss": 3.1799,
      "step": 253980
    },
    {
      "epoch": 0.06802916874121219,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001492022,
      "loss": 3.1514,
      "step": 253990
    },
    {
      "epoch": 0.06804781197737038,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001492002,
      "loss": 3.1876,
      "step": 254000
    },
    {
      "epoch": 0.06806645521352855,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001491982,
      "loss": 3.0047,
      "step": 254010
    },
    {
      "epoch": 0.06808509844968674,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001491962,
      "loss": 3.1006,
      "step": 254020
    },
    {
      "epoch": 0.06810374168584492,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014919420000000002,
      "loss": 3.0877,
      "step": 254030
    },
    {
      "epoch": 0.0681223849220031,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001491922,
      "loss": 3.2519,
      "step": 254040
    },
    {
      "epoch": 0.0681410281581613,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00014919020000000001,
      "loss": 3.1607,
      "step": 254050
    },
    {
      "epoch": 0.06815967139431947,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001491882,
      "loss": 3.218,
      "step": 254060
    },
    {
      "epoch": 0.06817831463047766,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001491862,
      "loss": 3.0375,
      "step": 254070
    },
    {
      "epoch": 0.06819695786663585,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001491842,
      "loss": 3.3006,
      "step": 254080
    },
    {
      "epoch": 0.06821560110279402,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001491822,
      "loss": 3.2286,
      "step": 254090
    },
    {
      "epoch": 0.06823424433895221,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014918020000000002,
      "loss": 3.2372,
      "step": 254100
    },
    {
      "epoch": 0.06825288757511039,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014917820000000001,
      "loss": 3.0049,
      "step": 254110
    },
    {
      "epoch": 0.06827153081126858,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001491762,
      "loss": 3.0338,
      "step": 254120
    },
    {
      "epoch": 0.06829017404742677,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001491742,
      "loss": 3.0462,
      "step": 254130
    },
    {
      "epoch": 0.06830881728358494,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001491722,
      "loss": 3.1294,
      "step": 254140
    },
    {
      "epoch": 0.06832746051974313,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014917020000000002,
      "loss": 3.2003,
      "step": 254150
    },
    {
      "epoch": 0.06834610375590132,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001491682,
      "loss": 3.204,
      "step": 254160
    },
    {
      "epoch": 0.0683647469920595,
      "grad_norm": 0.75,
      "learning_rate": 0.00014916620000000002,
      "loss": 3.0347,
      "step": 254170
    },
    {
      "epoch": 0.06838339022821768,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001491642,
      "loss": 3.2614,
      "step": 254180
    },
    {
      "epoch": 0.06840203346437586,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001491622,
      "loss": 3.1461,
      "step": 254190
    },
    {
      "epoch": 0.06842067670053405,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014916020000000003,
      "loss": 3.0981,
      "step": 254200
    },
    {
      "epoch": 0.06843931993669224,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001491582,
      "loss": 3.1725,
      "step": 254210
    },
    {
      "epoch": 0.06845796317285041,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014915620000000002,
      "loss": 3.0585,
      "step": 254220
    },
    {
      "epoch": 0.0684766064090086,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001491542,
      "loss": 3.1736,
      "step": 254230
    },
    {
      "epoch": 0.06849524964516678,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001491522,
      "loss": 3.1226,
      "step": 254240
    },
    {
      "epoch": 0.06851389288132496,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001491502,
      "loss": 3.0673,
      "step": 254250
    },
    {
      "epoch": 0.06853253611748315,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001491482,
      "loss": 3.0703,
      "step": 254260
    },
    {
      "epoch": 0.06855117935364133,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014914620000000003,
      "loss": 3.1205,
      "step": 254270
    },
    {
      "epoch": 0.06856982258979952,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001491442,
      "loss": 3.0452,
      "step": 254280
    },
    {
      "epoch": 0.0685884658259577,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014914220000000002,
      "loss": 3.1798,
      "step": 254290
    },
    {
      "epoch": 0.06860710906211588,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014914019999999999,
      "loss": 3.1186,
      "step": 254300
    },
    {
      "epoch": 0.06862575229827407,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001491382,
      "loss": 3.1415,
      "step": 254310
    },
    {
      "epoch": 0.06864439553443225,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001491362,
      "loss": 3.1252,
      "step": 254320
    },
    {
      "epoch": 0.06866303877059043,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001491342,
      "loss": 3.126,
      "step": 254330
    },
    {
      "epoch": 0.06868168200674862,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014913220000000002,
      "loss": 3.1321,
      "step": 254340
    },
    {
      "epoch": 0.0687003252429068,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001491302,
      "loss": 3.2125,
      "step": 254350
    },
    {
      "epoch": 0.06871896847906499,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001491282,
      "loss": 3.1703,
      "step": 254360
    },
    {
      "epoch": 0.06873761171522318,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001491262,
      "loss": 3.0805,
      "step": 254370
    },
    {
      "epoch": 0.06875625495138135,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001491242,
      "loss": 3.1153,
      "step": 254380
    },
    {
      "epoch": 0.06877489818753954,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001491222,
      "loss": 3.13,
      "step": 254390
    },
    {
      "epoch": 0.06879354142369772,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001491202,
      "loss": 3.1505,
      "step": 254400
    },
    {
      "epoch": 0.0688121846598559,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014911820000000002,
      "loss": 3.0656,
      "step": 254410
    },
    {
      "epoch": 0.0688308278960141,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014911620000000001,
      "loss": 3.1116,
      "step": 254420
    },
    {
      "epoch": 0.06884947113217227,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001491142,
      "loss": 3.0521,
      "step": 254430
    },
    {
      "epoch": 0.06886811436833046,
      "grad_norm": 0.75,
      "learning_rate": 0.0001491122,
      "loss": 3.1006,
      "step": 254440
    },
    {
      "epoch": 0.06888675760448863,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001491102,
      "loss": 3.097,
      "step": 254450
    },
    {
      "epoch": 0.06890540084064682,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001491082,
      "loss": 3.0882,
      "step": 254460
    },
    {
      "epoch": 0.06892404407680501,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014910620000000002,
      "loss": 3.2818,
      "step": 254470
    },
    {
      "epoch": 0.06894268731296319,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014910420000000001,
      "loss": 3.118,
      "step": 254480
    },
    {
      "epoch": 0.06896133054912137,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001491022,
      "loss": 3.124,
      "step": 254490
    },
    {
      "epoch": 0.06897997378527956,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001491002,
      "loss": 3.2218,
      "step": 254500
    },
    {
      "epoch": 0.06899861702143774,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001490982,
      "loss": 3.2444,
      "step": 254510
    },
    {
      "epoch": 0.06901726025759593,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001490962,
      "loss": 3.1356,
      "step": 254520
    },
    {
      "epoch": 0.0690359034937541,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001490942,
      "loss": 3.2226,
      "step": 254530
    },
    {
      "epoch": 0.06905454672991229,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014909220000000002,
      "loss": 3.1353,
      "step": 254540
    },
    {
      "epoch": 0.06907318996607048,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001490902,
      "loss": 3.1152,
      "step": 254550
    },
    {
      "epoch": 0.06909183320222866,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001490882,
      "loss": 3.1737,
      "step": 254560
    },
    {
      "epoch": 0.06911047643838684,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001490862,
      "loss": 3.072,
      "step": 254570
    },
    {
      "epoch": 0.06912911967454502,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001490842,
      "loss": 3.2747,
      "step": 254580
    },
    {
      "epoch": 0.06914776291070321,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014908220000000002,
      "loss": 3.1867,
      "step": 254590
    },
    {
      "epoch": 0.0691664061468614,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014908020000000002,
      "loss": 3.1102,
      "step": 254600
    },
    {
      "epoch": 0.06918504938301957,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001490782,
      "loss": 3.0944,
      "step": 254610
    },
    {
      "epoch": 0.06920369261917776,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001490762,
      "loss": 3.0805,
      "step": 254620
    },
    {
      "epoch": 0.06922233585533595,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001490742,
      "loss": 3.0592,
      "step": 254630
    },
    {
      "epoch": 0.06924097909149413,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014907220000000003,
      "loss": 3.1306,
      "step": 254640
    },
    {
      "epoch": 0.06925962232765231,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001490702,
      "loss": 3.1469,
      "step": 254650
    },
    {
      "epoch": 0.06927826556381049,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014906820000000002,
      "loss": 3.0844,
      "step": 254660
    },
    {
      "epoch": 0.06929690879996868,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001490662,
      "loss": 3.1612,
      "step": 254670
    },
    {
      "epoch": 0.06931555203612687,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001490642,
      "loss": 3.2369,
      "step": 254680
    },
    {
      "epoch": 0.06933419527228504,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001490622,
      "loss": 3.2162,
      "step": 254690
    },
    {
      "epoch": 0.06935283850844323,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001490602,
      "loss": 3.0367,
      "step": 254700
    },
    {
      "epoch": 0.06937148174460142,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014905820000000002,
      "loss": 3.178,
      "step": 254710
    },
    {
      "epoch": 0.0693901249807596,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001490562,
      "loss": 3.2055,
      "step": 254720
    },
    {
      "epoch": 0.06940876821691778,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014905420000000001,
      "loss": 3.1216,
      "step": 254730
    },
    {
      "epoch": 0.06942741145307596,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001490522,
      "loss": 3.0171,
      "step": 254740
    },
    {
      "epoch": 0.06944605468923415,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001490502,
      "loss": 3.1301,
      "step": 254750
    },
    {
      "epoch": 0.06946469792539234,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014904820000000003,
      "loss": 3.2463,
      "step": 254760
    },
    {
      "epoch": 0.06948334116155051,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001490462,
      "loss": 3.1146,
      "step": 254770
    },
    {
      "epoch": 0.0695019843977087,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014904420000000002,
      "loss": 3.1457,
      "step": 254780
    },
    {
      "epoch": 0.06952062763386688,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001490422,
      "loss": 3.1631,
      "step": 254790
    },
    {
      "epoch": 0.06953927087002507,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001490402,
      "loss": 3.0404,
      "step": 254800
    },
    {
      "epoch": 0.06955791410618326,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001490382,
      "loss": 3.1121,
      "step": 254810
    },
    {
      "epoch": 0.06957655734234143,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001490362,
      "loss": 3.1746,
      "step": 254820
    },
    {
      "epoch": 0.06959520057849962,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014903420000000002,
      "loss": 3.1077,
      "step": 254830
    },
    {
      "epoch": 0.06961384381465781,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001490322,
      "loss": 3.0923,
      "step": 254840
    },
    {
      "epoch": 0.06963248705081598,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014903020000000002,
      "loss": 3.1321,
      "step": 254850
    },
    {
      "epoch": 0.06965113028697417,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001490282,
      "loss": 3.1325,
      "step": 254860
    },
    {
      "epoch": 0.06966977352313235,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001490262,
      "loss": 3.1141,
      "step": 254870
    },
    {
      "epoch": 0.06968841675929054,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001490242,
      "loss": 3.0323,
      "step": 254880
    },
    {
      "epoch": 0.06970705999544873,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001490222,
      "loss": 3.2278,
      "step": 254890
    },
    {
      "epoch": 0.0697257032316069,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014902020000000002,
      "loss": 3.1994,
      "step": 254900
    },
    {
      "epoch": 0.06974434646776509,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001490182,
      "loss": 3.1618,
      "step": 254910
    },
    {
      "epoch": 0.06976298970392326,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001490162,
      "loss": 3.0625,
      "step": 254920
    },
    {
      "epoch": 0.06978163294008145,
      "grad_norm": 0.75,
      "learning_rate": 0.0001490142,
      "loss": 3.1447,
      "step": 254930
    },
    {
      "epoch": 0.06980027617623964,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001490122,
      "loss": 3.1311,
      "step": 254940
    },
    {
      "epoch": 0.06981891941239782,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001490102,
      "loss": 3.1374,
      "step": 254950
    },
    {
      "epoch": 0.069837562648556,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001490082,
      "loss": 3.0589,
      "step": 254960
    },
    {
      "epoch": 0.0698562058847142,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014900620000000002,
      "loss": 3.0344,
      "step": 254970
    },
    {
      "epoch": 0.06987484912087237,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001490042,
      "loss": 3.2251,
      "step": 254980
    },
    {
      "epoch": 0.06989349235703056,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001490022,
      "loss": 3.1112,
      "step": 254990
    },
    {
      "epoch": 0.06991213559318873,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001490002,
      "loss": 3.0644,
      "step": 255000
    },
    {
      "epoch": 0.06993077882934692,
      "grad_norm": 0.75,
      "learning_rate": 0.0001489982,
      "loss": 3.1081,
      "step": 255010
    },
    {
      "epoch": 0.06994942206550511,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001489962,
      "loss": 3.1413,
      "step": 255020
    },
    {
      "epoch": 0.06996806530166329,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014899420000000002,
      "loss": 3.2473,
      "step": 255030
    },
    {
      "epoch": 0.06998670853782148,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001489922,
      "loss": 3.0962,
      "step": 255040
    },
    {
      "epoch": 0.07000535177397967,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001489902,
      "loss": 3.2541,
      "step": 255050
    },
    {
      "epoch": 0.07002399501013784,
      "grad_norm": 0.625,
      "learning_rate": 0.0001489882,
      "loss": 3.0055,
      "step": 255060
    },
    {
      "epoch": 0.07004263824629603,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001489862,
      "loss": 3.0948,
      "step": 255070
    },
    {
      "epoch": 0.0700612814824542,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014898420000000002,
      "loss": 3.1273,
      "step": 255080
    },
    {
      "epoch": 0.0700799247186124,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014898220000000002,
      "loss": 3.0759,
      "step": 255090
    },
    {
      "epoch": 0.07009856795477058,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014898020000000001,
      "loss": 3.0504,
      "step": 255100
    },
    {
      "epoch": 0.07011721119092876,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001489782,
      "loss": 3.2013,
      "step": 255110
    },
    {
      "epoch": 0.07013585442708695,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001489762,
      "loss": 3.1319,
      "step": 255120
    },
    {
      "epoch": 0.07015449766324512,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001489742,
      "loss": 3.242,
      "step": 255130
    },
    {
      "epoch": 0.07017314089940331,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001489722,
      "loss": 3.0592,
      "step": 255140
    },
    {
      "epoch": 0.0701917841355615,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014897020000000002,
      "loss": 3.2926,
      "step": 255150
    },
    {
      "epoch": 0.07021042737171967,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014896820000000001,
      "loss": 3.1145,
      "step": 255160
    },
    {
      "epoch": 0.07022907060787786,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001489662,
      "loss": 3.2156,
      "step": 255170
    },
    {
      "epoch": 0.07024771384403605,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001489642,
      "loss": 3.0837,
      "step": 255180
    },
    {
      "epoch": 0.07026635708019423,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001489622,
      "loss": 3.0671,
      "step": 255190
    },
    {
      "epoch": 0.07028500031635242,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014896020000000002,
      "loss": 3.1163,
      "step": 255200
    },
    {
      "epoch": 0.07030364355251059,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001489582,
      "loss": 3.1456,
      "step": 255210
    },
    {
      "epoch": 0.07032228678866878,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014895620000000002,
      "loss": 3.1402,
      "step": 255220
    },
    {
      "epoch": 0.07034093002482697,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001489542,
      "loss": 3.0647,
      "step": 255230
    },
    {
      "epoch": 0.07035957326098514,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001489522,
      "loss": 3.0518,
      "step": 255240
    },
    {
      "epoch": 0.07037821649714333,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014895020000000003,
      "loss": 3.0992,
      "step": 255250
    },
    {
      "epoch": 0.07039685973330151,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001489482,
      "loss": 3.1881,
      "step": 255260
    },
    {
      "epoch": 0.0704155029694597,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014894620000000002,
      "loss": 3.1321,
      "step": 255270
    },
    {
      "epoch": 0.07043414620561789,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001489442,
      "loss": 3.1921,
      "step": 255280
    },
    {
      "epoch": 0.07045278944177606,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001489422,
      "loss": 3.0825,
      "step": 255290
    },
    {
      "epoch": 0.07047143267793425,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001489402,
      "loss": 3.1742,
      "step": 255300
    },
    {
      "epoch": 0.07049007591409244,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001489382,
      "loss": 3.028,
      "step": 255310
    },
    {
      "epoch": 0.07050871915025061,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014893620000000003,
      "loss": 3.3654,
      "step": 255320
    },
    {
      "epoch": 0.0705273623864088,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001489342,
      "loss": 3.033,
      "step": 255330
    },
    {
      "epoch": 0.07054600562256698,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014893220000000002,
      "loss": 3.2207,
      "step": 255340
    },
    {
      "epoch": 0.07056464885872517,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014893019999999999,
      "loss": 3.1124,
      "step": 255350
    },
    {
      "epoch": 0.07058329209488336,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001489282,
      "loss": 3.0091,
      "step": 255360
    },
    {
      "epoch": 0.07060193533104153,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001489262,
      "loss": 3.1028,
      "step": 255370
    },
    {
      "epoch": 0.07062057856719972,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001489242,
      "loss": 3.1287,
      "step": 255380
    },
    {
      "epoch": 0.07063922180335791,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014892220000000002,
      "loss": 3.1387,
      "step": 255390
    },
    {
      "epoch": 0.07065786503951609,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001489202,
      "loss": 3.0643,
      "step": 255400
    },
    {
      "epoch": 0.07067650827567427,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014891820000000001,
      "loss": 3.1996,
      "step": 255410
    },
    {
      "epoch": 0.07069515151183245,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001489162,
      "loss": 3.0561,
      "step": 255420
    },
    {
      "epoch": 0.07071379474799064,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001489142,
      "loss": 3.0915,
      "step": 255430
    },
    {
      "epoch": 0.07073243798414883,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001489122,
      "loss": 2.9605,
      "step": 255440
    },
    {
      "epoch": 0.070751081220307,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001489102,
      "loss": 3.0188,
      "step": 255450
    },
    {
      "epoch": 0.07076972445646519,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014890820000000002,
      "loss": 3.2415,
      "step": 255460
    },
    {
      "epoch": 0.07078836769262337,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014890620000000001,
      "loss": 3.0109,
      "step": 255470
    },
    {
      "epoch": 0.07080701092878156,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001489042,
      "loss": 3.1793,
      "step": 255480
    },
    {
      "epoch": 0.07082565416493974,
      "grad_norm": 0.75,
      "learning_rate": 0.0001489022,
      "loss": 3.13,
      "step": 255490
    },
    {
      "epoch": 0.07084429740109792,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001489002,
      "loss": 3.0224,
      "step": 255500
    },
    {
      "epoch": 0.07086294063725611,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001488982,
      "loss": 3.08,
      "step": 255510
    },
    {
      "epoch": 0.0708815838734143,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001488962,
      "loss": 3.223,
      "step": 255520
    },
    {
      "epoch": 0.07090022710957247,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014889420000000002,
      "loss": 3.0092,
      "step": 255530
    },
    {
      "epoch": 0.07091887034573066,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001488922,
      "loss": 3.2158,
      "step": 255540
    },
    {
      "epoch": 0.07093751358188884,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001488902,
      "loss": 3.17,
      "step": 255550
    },
    {
      "epoch": 0.07095615681804703,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001488882,
      "loss": 3.1355,
      "step": 255560
    },
    {
      "epoch": 0.07097480005420521,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001488862,
      "loss": 3.1449,
      "step": 255570
    },
    {
      "epoch": 0.07099344329036339,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014888420000000002,
      "loss": 3.1019,
      "step": 255580
    },
    {
      "epoch": 0.07101208652652158,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014888220000000002,
      "loss": 3.0776,
      "step": 255590
    },
    {
      "epoch": 0.07103072976267975,
      "grad_norm": 0.75,
      "learning_rate": 0.0001488802,
      "loss": 3.1812,
      "step": 255600
    },
    {
      "epoch": 0.07104937299883794,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001488782,
      "loss": 3.2203,
      "step": 255610
    },
    {
      "epoch": 0.07106801623499613,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001488762,
      "loss": 3.0936,
      "step": 255620
    },
    {
      "epoch": 0.0710866594711543,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001488742,
      "loss": 2.9928,
      "step": 255630
    },
    {
      "epoch": 0.0711053027073125,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014887220000000002,
      "loss": 3.1867,
      "step": 255640
    },
    {
      "epoch": 0.07112394594347068,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014887020000000002,
      "loss": 3.091,
      "step": 255650
    },
    {
      "epoch": 0.07114258917962886,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001488682,
      "loss": 3.1687,
      "step": 255660
    },
    {
      "epoch": 0.07116123241578705,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001488662,
      "loss": 3.1721,
      "step": 255670
    },
    {
      "epoch": 0.07117987565194522,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001488642,
      "loss": 3.1375,
      "step": 255680
    },
    {
      "epoch": 0.07119851888810341,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001488622,
      "loss": 3.0863,
      "step": 255690
    },
    {
      "epoch": 0.0712171621242616,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001488602,
      "loss": 3.0677,
      "step": 255700
    },
    {
      "epoch": 0.07123580536041978,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014885820000000002,
      "loss": 3.1581,
      "step": 255710
    },
    {
      "epoch": 0.07125444859657797,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001488562,
      "loss": 3.1692,
      "step": 255720
    },
    {
      "epoch": 0.07127309183273615,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001488542,
      "loss": 3.1712,
      "step": 255730
    },
    {
      "epoch": 0.07129173506889433,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001488522,
      "loss": 3.1309,
      "step": 255740
    },
    {
      "epoch": 0.07131037830505252,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001488502,
      "loss": 3.0956,
      "step": 255750
    },
    {
      "epoch": 0.0713290215412107,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014884820000000002,
      "loss": 3.1671,
      "step": 255760
    },
    {
      "epoch": 0.07134766477736888,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001488462,
      "loss": 3.233,
      "step": 255770
    },
    {
      "epoch": 0.07136630801352707,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014884420000000001,
      "loss": 3.0054,
      "step": 255780
    },
    {
      "epoch": 0.07138495124968525,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001488422,
      "loss": 3.1478,
      "step": 255790
    },
    {
      "epoch": 0.07140359448584344,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001488402,
      "loss": 3.0785,
      "step": 255800
    },
    {
      "epoch": 0.07142223772200161,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014883820000000003,
      "loss": 3.2818,
      "step": 255810
    },
    {
      "epoch": 0.0714408809581598,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001488362,
      "loss": 3.0609,
      "step": 255820
    },
    {
      "epoch": 0.07145952419431799,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014883420000000002,
      "loss": 3.1496,
      "step": 255830
    },
    {
      "epoch": 0.07147816743047616,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001488322,
      "loss": 3.1015,
      "step": 255840
    },
    {
      "epoch": 0.07149681066663435,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001488302,
      "loss": 3.1429,
      "step": 255850
    },
    {
      "epoch": 0.07151545390279254,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001488282,
      "loss": 3.1549,
      "step": 255860
    },
    {
      "epoch": 0.07153409713895072,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001488262,
      "loss": 2.937,
      "step": 255870
    },
    {
      "epoch": 0.0715527403751089,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014882420000000002,
      "loss": 3.1145,
      "step": 255880
    },
    {
      "epoch": 0.07157138361126708,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001488222,
      "loss": 3.0843,
      "step": 255890
    },
    {
      "epoch": 0.07159002684742527,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014882020000000002,
      "loss": 3.1598,
      "step": 255900
    },
    {
      "epoch": 0.07160867008358346,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014881819999999998,
      "loss": 3.1619,
      "step": 255910
    },
    {
      "epoch": 0.07162731331974163,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001488162,
      "loss": 3.0961,
      "step": 255920
    },
    {
      "epoch": 0.07164595655589982,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001488142,
      "loss": 3.0897,
      "step": 255930
    },
    {
      "epoch": 0.071664599792058,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001488122,
      "loss": 3.207,
      "step": 255940
    },
    {
      "epoch": 0.07168324302821619,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014881020000000002,
      "loss": 3.1222,
      "step": 255950
    },
    {
      "epoch": 0.07170188626437438,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001488082,
      "loss": 3.0543,
      "step": 255960
    },
    {
      "epoch": 0.07172052950053255,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001488062,
      "loss": 3.1228,
      "step": 255970
    },
    {
      "epoch": 0.07173917273669074,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001488042,
      "loss": 3.1571,
      "step": 255980
    },
    {
      "epoch": 0.07175781597284893,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001488022,
      "loss": 3.09,
      "step": 255990
    },
    {
      "epoch": 0.0717764592090071,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001488002,
      "loss": 3.0506,
      "step": 256000
    },
    {
      "epoch": 0.07179510244516529,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001487982,
      "loss": 3.1468,
      "step": 256010
    },
    {
      "epoch": 0.07181374568132347,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014879620000000002,
      "loss": 3.0949,
      "step": 256020
    },
    {
      "epoch": 0.07183238891748166,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001487942,
      "loss": 3.0494,
      "step": 256030
    },
    {
      "epoch": 0.07185103215363985,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001487922,
      "loss": 3.1546,
      "step": 256040
    },
    {
      "epoch": 0.07186967538979802,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001487902,
      "loss": 3.2033,
      "step": 256050
    },
    {
      "epoch": 0.07188831862595621,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001487882,
      "loss": 3.1518,
      "step": 256060
    },
    {
      "epoch": 0.0719069618621144,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014878620000000002,
      "loss": 3.1997,
      "step": 256070
    },
    {
      "epoch": 0.07192560509827257,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014878420000000002,
      "loss": 3.3221,
      "step": 256080
    },
    {
      "epoch": 0.07194424833443076,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001487822,
      "loss": 3.0036,
      "step": 256090
    },
    {
      "epoch": 0.07196289157058894,
      "grad_norm": 0.75,
      "learning_rate": 0.0001487802,
      "loss": 3.1213,
      "step": 256100
    },
    {
      "epoch": 0.07198153480674713,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001487782,
      "loss": 3.2376,
      "step": 256110
    },
    {
      "epoch": 0.07200017804290532,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001487762,
      "loss": 3.1463,
      "step": 256120
    },
    {
      "epoch": 0.07201882127906349,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001487742,
      "loss": 3.1397,
      "step": 256130
    },
    {
      "epoch": 0.07203746451522168,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014877220000000002,
      "loss": 2.9159,
      "step": 256140
    },
    {
      "epoch": 0.07205610775137986,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014877020000000001,
      "loss": 2.9682,
      "step": 256150
    },
    {
      "epoch": 0.07207475098753804,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001487682,
      "loss": 3.1396,
      "step": 256160
    },
    {
      "epoch": 0.07209339422369623,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001487662,
      "loss": 3.1167,
      "step": 256170
    },
    {
      "epoch": 0.07211203745985441,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001487642,
      "loss": 3.1462,
      "step": 256180
    },
    {
      "epoch": 0.0721306806960126,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001487622,
      "loss": 3.2051,
      "step": 256190
    },
    {
      "epoch": 0.07214932393217079,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014876020000000002,
      "loss": 3.1438,
      "step": 256200
    },
    {
      "epoch": 0.07216796716832896,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014875820000000002,
      "loss": 3.1772,
      "step": 256210
    },
    {
      "epoch": 0.07218661040448715,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001487562,
      "loss": 3.1303,
      "step": 256220
    },
    {
      "epoch": 0.07220525364064533,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001487542,
      "loss": 3.1266,
      "step": 256230
    },
    {
      "epoch": 0.07222389687680351,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001487522,
      "loss": 3.0927,
      "step": 256240
    },
    {
      "epoch": 0.0722425401129617,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014875020000000002,
      "loss": 3.2101,
      "step": 256250
    },
    {
      "epoch": 0.07226118334911988,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001487482,
      "loss": 3.1204,
      "step": 256260
    },
    {
      "epoch": 0.07227982658527807,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014874620000000002,
      "loss": 3.0726,
      "step": 256270
    },
    {
      "epoch": 0.07229846982143624,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001487442,
      "loss": 3.0675,
      "step": 256280
    },
    {
      "epoch": 0.07231711305759443,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001487422,
      "loss": 3.2624,
      "step": 256290
    },
    {
      "epoch": 0.07233575629375262,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001487402,
      "loss": 3.0468,
      "step": 256300
    },
    {
      "epoch": 0.0723543995299108,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001487382,
      "loss": 3.1492,
      "step": 256310
    },
    {
      "epoch": 0.07237304276606898,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014873620000000002,
      "loss": 3.1213,
      "step": 256320
    },
    {
      "epoch": 0.07239168600222717,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001487342,
      "loss": 3.276,
      "step": 256330
    },
    {
      "epoch": 0.07241032923838535,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001487322,
      "loss": 3.1829,
      "step": 256340
    },
    {
      "epoch": 0.07242897247454354,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001487302,
      "loss": 3.1395,
      "step": 256350
    },
    {
      "epoch": 0.07244761571070171,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001487282,
      "loss": 3.1518,
      "step": 256360
    },
    {
      "epoch": 0.0724662589468599,
      "grad_norm": 0.75,
      "learning_rate": 0.00014872620000000003,
      "loss": 3.2388,
      "step": 256370
    },
    {
      "epoch": 0.07248490218301809,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001487242,
      "loss": 3.0639,
      "step": 256380
    },
    {
      "epoch": 0.07250354541917627,
      "grad_norm": 0.625,
      "learning_rate": 0.00014872220000000002,
      "loss": 3.0553,
      "step": 256390
    },
    {
      "epoch": 0.07252218865533445,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014872019999999999,
      "loss": 3.1638,
      "step": 256400
    },
    {
      "epoch": 0.07254083189149264,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001487182,
      "loss": 3.1146,
      "step": 256410
    },
    {
      "epoch": 0.07255947512765082,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001487162,
      "loss": 3.0183,
      "step": 256420
    },
    {
      "epoch": 0.07257811836380901,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001487142,
      "loss": 3.1752,
      "step": 256430
    },
    {
      "epoch": 0.07259676159996718,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00014871220000000002,
      "loss": 3.2405,
      "step": 256440
    },
    {
      "epoch": 0.07261540483612537,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001487102,
      "loss": 3.208,
      "step": 256450
    },
    {
      "epoch": 0.07263404807228356,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014870820000000001,
      "loss": 3.0343,
      "step": 256460
    },
    {
      "epoch": 0.07265269130844174,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001487062,
      "loss": 3.1948,
      "step": 256470
    },
    {
      "epoch": 0.07267133454459992,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001487042,
      "loss": 3.083,
      "step": 256480
    },
    {
      "epoch": 0.0726899777807581,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001487022,
      "loss": 3.0501,
      "step": 256490
    },
    {
      "epoch": 0.07270862101691629,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001487002,
      "loss": 3.1101,
      "step": 256500
    },
    {
      "epoch": 0.07272726425307448,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014869820000000002,
      "loss": 3.0509,
      "step": 256510
    },
    {
      "epoch": 0.07274590748923265,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001486962,
      "loss": 3.1879,
      "step": 256520
    },
    {
      "epoch": 0.07276455072539084,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001486942,
      "loss": 3.0893,
      "step": 256530
    },
    {
      "epoch": 0.07278319396154903,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001486922,
      "loss": 3.1321,
      "step": 256540
    },
    {
      "epoch": 0.0728018371977072,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001486902,
      "loss": 2.9962,
      "step": 256550
    },
    {
      "epoch": 0.0728204804338654,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014868820000000002,
      "loss": 3.1794,
      "step": 256560
    },
    {
      "epoch": 0.07283912367002357,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001486862,
      "loss": 3.0247,
      "step": 256570
    },
    {
      "epoch": 0.07285776690618176,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014868420000000002,
      "loss": 3.137,
      "step": 256580
    },
    {
      "epoch": 0.07287641014233995,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001486822,
      "loss": 3.0826,
      "step": 256590
    },
    {
      "epoch": 0.07289505337849812,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001486802,
      "loss": 3.0425,
      "step": 256600
    },
    {
      "epoch": 0.07291369661465631,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001486782,
      "loss": 3.028,
      "step": 256610
    },
    {
      "epoch": 0.07293233985081449,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001486762,
      "loss": 3.1735,
      "step": 256620
    },
    {
      "epoch": 0.07295098308697268,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014867420000000002,
      "loss": 3.1593,
      "step": 256630
    },
    {
      "epoch": 0.07296962632313087,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014867220000000002,
      "loss": 3.16,
      "step": 256640
    },
    {
      "epoch": 0.07298826955928904,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001486702,
      "loss": 2.9958,
      "step": 256650
    },
    {
      "epoch": 0.07300691279544723,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001486682,
      "loss": 3.1857,
      "step": 256660
    },
    {
      "epoch": 0.07302555603160542,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001486662,
      "loss": 3.0621,
      "step": 256670
    },
    {
      "epoch": 0.0730441992677636,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001486642,
      "loss": 3.2422,
      "step": 256680
    },
    {
      "epoch": 0.07306284250392178,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014866220000000002,
      "loss": 3.0665,
      "step": 256690
    },
    {
      "epoch": 0.07308148574007996,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014866020000000002,
      "loss": 3.1245,
      "step": 256700
    },
    {
      "epoch": 0.07310012897623815,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001486582,
      "loss": 3.2067,
      "step": 256710
    },
    {
      "epoch": 0.07311877221239634,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001486562,
      "loss": 3.3135,
      "step": 256720
    },
    {
      "epoch": 0.07313741544855451,
      "grad_norm": 0.75,
      "learning_rate": 0.0001486542,
      "loss": 3.1037,
      "step": 256730
    },
    {
      "epoch": 0.0731560586847127,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001486522,
      "loss": 3.1658,
      "step": 256740
    },
    {
      "epoch": 0.07317470192087089,
      "grad_norm": 0.625,
      "learning_rate": 0.0001486502,
      "loss": 3.1662,
      "step": 256750
    },
    {
      "epoch": 0.07319334515702906,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014864820000000002,
      "loss": 3.0552,
      "step": 256760
    },
    {
      "epoch": 0.07321198839318725,
      "grad_norm": 0.75,
      "learning_rate": 0.0001486462,
      "loss": 3.2338,
      "step": 256770
    },
    {
      "epoch": 0.07323063162934543,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001486442,
      "loss": 3.0044,
      "step": 256780
    },
    {
      "epoch": 0.07324927486550362,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001486422,
      "loss": 3.2466,
      "step": 256790
    },
    {
      "epoch": 0.0732679181016618,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001486402,
      "loss": 3.1846,
      "step": 256800
    },
    {
      "epoch": 0.07328656133781998,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014863820000000002,
      "loss": 3.1081,
      "step": 256810
    },
    {
      "epoch": 0.07330520457397817,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001486362,
      "loss": 3.1752,
      "step": 256820
    },
    {
      "epoch": 0.07332384781013634,
      "grad_norm": 0.75,
      "learning_rate": 0.00014863420000000001,
      "loss": 3.1078,
      "step": 256830
    },
    {
      "epoch": 0.07334249104629453,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001486322,
      "loss": 3.0215,
      "step": 256840
    },
    {
      "epoch": 0.07336113428245272,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001486302,
      "loss": 3.0285,
      "step": 256850
    },
    {
      "epoch": 0.0733797775186109,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014862820000000003,
      "loss": 3.0831,
      "step": 256860
    },
    {
      "epoch": 0.07339842075476909,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001486262,
      "loss": 3.16,
      "step": 256870
    },
    {
      "epoch": 0.07341706399092728,
      "grad_norm": 0.75,
      "learning_rate": 0.00014862420000000002,
      "loss": 3.1582,
      "step": 256880
    },
    {
      "epoch": 0.07343570722708545,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001486222,
      "loss": 3.1836,
      "step": 256890
    },
    {
      "epoch": 0.07345435046324364,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001486202,
      "loss": 3.1244,
      "step": 256900
    },
    {
      "epoch": 0.07347299369940181,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001486182,
      "loss": 3.1921,
      "step": 256910
    },
    {
      "epoch": 0.07349163693556,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001486162,
      "loss": 3.1477,
      "step": 256920
    },
    {
      "epoch": 0.07351028017171819,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014861420000000002,
      "loss": 3.1373,
      "step": 256930
    },
    {
      "epoch": 0.07352892340787637,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001486122,
      "loss": 3.2319,
      "step": 256940
    },
    {
      "epoch": 0.07354756664403456,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014861020000000002,
      "loss": 3.1266,
      "step": 256950
    },
    {
      "epoch": 0.07356620988019273,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014860819999999998,
      "loss": 3.0498,
      "step": 256960
    },
    {
      "epoch": 0.07358485311635092,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001486062,
      "loss": 3.0368,
      "step": 256970
    },
    {
      "epoch": 0.07360349635250911,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001486042,
      "loss": 3.0891,
      "step": 256980
    },
    {
      "epoch": 0.07362213958866728,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001486022,
      "loss": 3.124,
      "step": 256990
    },
    {
      "epoch": 0.07364078282482547,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014860020000000002,
      "loss": 3.0968,
      "step": 257000
    },
    {
      "epoch": 0.07365942606098366,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001485982,
      "loss": 3.1873,
      "step": 257010
    },
    {
      "epoch": 0.07367806929714184,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001485962,
      "loss": 3.133,
      "step": 257020
    },
    {
      "epoch": 0.07369671253330003,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001485942,
      "loss": 3.0379,
      "step": 257030
    },
    {
      "epoch": 0.0737153557694582,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001485922,
      "loss": 3.1178,
      "step": 257040
    },
    {
      "epoch": 0.07373399900561639,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014859020000000003,
      "loss": 3.1801,
      "step": 257050
    },
    {
      "epoch": 0.07375264224177458,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001485882,
      "loss": 3.0879,
      "step": 257060
    },
    {
      "epoch": 0.07377128547793275,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014858620000000002,
      "loss": 3.121,
      "step": 257070
    },
    {
      "epoch": 0.07378992871409094,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001485842,
      "loss": 3.2192,
      "step": 257080
    },
    {
      "epoch": 0.07380857195024913,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001485822,
      "loss": 3.0316,
      "step": 257090
    },
    {
      "epoch": 0.07382721518640731,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001485802,
      "loss": 3.0321,
      "step": 257100
    },
    {
      "epoch": 0.0738458584225655,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001485782,
      "loss": 3.1537,
      "step": 257110
    },
    {
      "epoch": 0.07386450165872367,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014857620000000002,
      "loss": 3.0125,
      "step": 257120
    },
    {
      "epoch": 0.07388314489488186,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001485742,
      "loss": 3.2924,
      "step": 257130
    },
    {
      "epoch": 0.07390178813104005,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014857220000000001,
      "loss": 3.1974,
      "step": 257140
    },
    {
      "epoch": 0.07392043136719822,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001485702,
      "loss": 3.18,
      "step": 257150
    },
    {
      "epoch": 0.07393907460335641,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001485682,
      "loss": 3.231,
      "step": 257160
    },
    {
      "epoch": 0.07395771783951459,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001485662,
      "loss": 3.1438,
      "step": 257170
    },
    {
      "epoch": 0.07397636107567278,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001485642,
      "loss": 3.1658,
      "step": 257180
    },
    {
      "epoch": 0.07399500431183097,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014856220000000002,
      "loss": 3.1352,
      "step": 257190
    },
    {
      "epoch": 0.07401364754798914,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014856020000000001,
      "loss": 3.1958,
      "step": 257200
    },
    {
      "epoch": 0.07403229078414733,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001485582,
      "loss": 3.0495,
      "step": 257210
    },
    {
      "epoch": 0.07405093402030552,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001485562,
      "loss": 3.083,
      "step": 257220
    },
    {
      "epoch": 0.0740695772564637,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001485542,
      "loss": 3.0278,
      "step": 257230
    },
    {
      "epoch": 0.07408822049262188,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001485522,
      "loss": 3.0458,
      "step": 257240
    },
    {
      "epoch": 0.07410686372878006,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014855020000000002,
      "loss": 3.0536,
      "step": 257250
    },
    {
      "epoch": 0.07412550696493825,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014854820000000002,
      "loss": 3.1045,
      "step": 257260
    },
    {
      "epoch": 0.07414415020109644,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001485462,
      "loss": 3.0824,
      "step": 257270
    },
    {
      "epoch": 0.07416279343725461,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001485442,
      "loss": 3.2057,
      "step": 257280
    },
    {
      "epoch": 0.0741814366734128,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001485422,
      "loss": 3.1089,
      "step": 257290
    },
    {
      "epoch": 0.07420007990957098,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001485402,
      "loss": 3.1788,
      "step": 257300
    },
    {
      "epoch": 0.07421872314572917,
      "grad_norm": 0.625,
      "learning_rate": 0.0001485382,
      "loss": 3.219,
      "step": 257310
    },
    {
      "epoch": 0.07423736638188735,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014853620000000002,
      "loss": 3.149,
      "step": 257320
    },
    {
      "epoch": 0.07425600961804553,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001485342,
      "loss": 3.1415,
      "step": 257330
    },
    {
      "epoch": 0.07427465285420372,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001485322,
      "loss": 3.1157,
      "step": 257340
    },
    {
      "epoch": 0.07429329609036191,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001485302,
      "loss": 3.1118,
      "step": 257350
    },
    {
      "epoch": 0.07431193932652008,
      "grad_norm": 0.75,
      "learning_rate": 0.0001485282,
      "loss": 3.1435,
      "step": 257360
    },
    {
      "epoch": 0.07433058256267827,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014852620000000002,
      "loss": 3.1993,
      "step": 257370
    },
    {
      "epoch": 0.07434922579883645,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001485242,
      "loss": 3.1978,
      "step": 257380
    },
    {
      "epoch": 0.07436786903499464,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001485222,
      "loss": 2.9635,
      "step": 257390
    },
    {
      "epoch": 0.07438651227115282,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001485202,
      "loss": 3.1449,
      "step": 257400
    },
    {
      "epoch": 0.074405155507311,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001485182,
      "loss": 3.1536,
      "step": 257410
    },
    {
      "epoch": 0.07442379874346919,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014851620000000003,
      "loss": 3.0449,
      "step": 257420
    },
    {
      "epoch": 0.07444244197962738,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001485142,
      "loss": 3.1215,
      "step": 257430
    },
    {
      "epoch": 0.07446108521578555,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014851220000000002,
      "loss": 3.1204,
      "step": 257440
    },
    {
      "epoch": 0.07447972845194374,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014851019999999999,
      "loss": 3.107,
      "step": 257450
    },
    {
      "epoch": 0.07449837168810192,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001485082,
      "loss": 2.986,
      "step": 257460
    },
    {
      "epoch": 0.0745170149242601,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001485062,
      "loss": 3.0047,
      "step": 257470
    },
    {
      "epoch": 0.0745356581604183,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001485042,
      "loss": 3.2074,
      "step": 257480
    },
    {
      "epoch": 0.07455430139657647,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014850220000000002,
      "loss": 3.0542,
      "step": 257490
    },
    {
      "epoch": 0.07457294463273466,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001485002,
      "loss": 3.0794,
      "step": 257500
    },
    {
      "epoch": 0.07459158786889283,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014849820000000001,
      "loss": 3.1762,
      "step": 257510
    },
    {
      "epoch": 0.07461023110505102,
      "grad_norm": 0.75,
      "learning_rate": 0.0001484962,
      "loss": 2.988,
      "step": 257520
    },
    {
      "epoch": 0.07462887434120921,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001484942,
      "loss": 3.121,
      "step": 257530
    },
    {
      "epoch": 0.07464751757736739,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014849220000000003,
      "loss": 3.1824,
      "step": 257540
    },
    {
      "epoch": 0.07466616081352558,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001484902,
      "loss": 3.0365,
      "step": 257550
    },
    {
      "epoch": 0.07468480404968376,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014848820000000002,
      "loss": 3.208,
      "step": 257560
    },
    {
      "epoch": 0.07470344728584194,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001484862,
      "loss": 3.1076,
      "step": 257570
    },
    {
      "epoch": 0.07472209052200013,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001484842,
      "loss": 2.954,
      "step": 257580
    },
    {
      "epoch": 0.0747407337581583,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001484822,
      "loss": 3.2614,
      "step": 257590
    },
    {
      "epoch": 0.07475937699431649,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001484802,
      "loss": 2.9378,
      "step": 257600
    },
    {
      "epoch": 0.07477802023047468,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014847820000000002,
      "loss": 3.001,
      "step": 257610
    },
    {
      "epoch": 0.07479666346663286,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001484762,
      "loss": 3.0635,
      "step": 257620
    },
    {
      "epoch": 0.07481530670279105,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014847420000000002,
      "loss": 3.0743,
      "step": 257630
    },
    {
      "epoch": 0.07483394993894922,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001484722,
      "loss": 3.0715,
      "step": 257640
    },
    {
      "epoch": 0.07485259317510741,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001484702,
      "loss": 3.1658,
      "step": 257650
    },
    {
      "epoch": 0.0748712364112656,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001484682,
      "loss": 3.1429,
      "step": 257660
    },
    {
      "epoch": 0.07488987964742377,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001484662,
      "loss": 3.0706,
      "step": 257670
    },
    {
      "epoch": 0.07490852288358196,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014846420000000002,
      "loss": 3.2419,
      "step": 257680
    },
    {
      "epoch": 0.07492716611974015,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014846220000000002,
      "loss": 3.2264,
      "step": 257690
    },
    {
      "epoch": 0.07494580935589833,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001484602,
      "loss": 3.1696,
      "step": 257700
    },
    {
      "epoch": 0.07496445259205652,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001484582,
      "loss": 3.0716,
      "step": 257710
    },
    {
      "epoch": 0.07498309582821469,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001484562,
      "loss": 3.265,
      "step": 257720
    },
    {
      "epoch": 0.07500173906437288,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001484542,
      "loss": 3.1923,
      "step": 257730
    },
    {
      "epoch": 0.07502038230053107,
      "grad_norm": 0.75,
      "learning_rate": 0.0001484522,
      "loss": 3.0896,
      "step": 257740
    },
    {
      "epoch": 0.07503902553668924,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014845020000000002,
      "loss": 3.1508,
      "step": 257750
    },
    {
      "epoch": 0.07505766877284743,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001484482,
      "loss": 3.052,
      "step": 257760
    },
    {
      "epoch": 0.07507631200900562,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001484462,
      "loss": 3.1278,
      "step": 257770
    },
    {
      "epoch": 0.0750949552451638,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001484442,
      "loss": 3.0894,
      "step": 257780
    },
    {
      "epoch": 0.07511359848132199,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001484422,
      "loss": 3.0697,
      "step": 257790
    },
    {
      "epoch": 0.07513224171748016,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001484402,
      "loss": 3.1936,
      "step": 257800
    },
    {
      "epoch": 0.07515088495363835,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014843820000000002,
      "loss": 3.1227,
      "step": 257810
    },
    {
      "epoch": 0.07516952818979654,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014843620000000001,
      "loss": 3.0309,
      "step": 257820
    },
    {
      "epoch": 0.07518817142595471,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001484342,
      "loss": 3.2361,
      "step": 257830
    },
    {
      "epoch": 0.0752068146621129,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001484322,
      "loss": 3.257,
      "step": 257840
    },
    {
      "epoch": 0.07522545789827108,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001484302,
      "loss": 3.1771,
      "step": 257850
    },
    {
      "epoch": 0.07524410113442927,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014842820000000002,
      "loss": 3.0853,
      "step": 257860
    },
    {
      "epoch": 0.07526274437058746,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001484262,
      "loss": 3.2577,
      "step": 257870
    },
    {
      "epoch": 0.07528138760674563,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014842420000000001,
      "loss": 3.1599,
      "step": 257880
    },
    {
      "epoch": 0.07530003084290382,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001484222,
      "loss": 3.0601,
      "step": 257890
    },
    {
      "epoch": 0.07531867407906201,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001484202,
      "loss": 3.181,
      "step": 257900
    },
    {
      "epoch": 0.07533731731522018,
      "grad_norm": 0.625,
      "learning_rate": 0.0001484182,
      "loss": 3.1123,
      "step": 257910
    },
    {
      "epoch": 0.07535596055137837,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001484162,
      "loss": 3.1304,
      "step": 257920
    },
    {
      "epoch": 0.07537460378753655,
      "grad_norm": 0.75,
      "learning_rate": 0.00014841420000000002,
      "loss": 3.1387,
      "step": 257930
    },
    {
      "epoch": 0.07539324702369474,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001484122,
      "loss": 3.0694,
      "step": 257940
    },
    {
      "epoch": 0.07541189025985293,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001484102,
      "loss": 3.1149,
      "step": 257950
    },
    {
      "epoch": 0.0754305334960111,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001484082,
      "loss": 3.0277,
      "step": 257960
    },
    {
      "epoch": 0.07544917673216929,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001484062,
      "loss": 3.1112,
      "step": 257970
    },
    {
      "epoch": 0.07546781996832747,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014840420000000002,
      "loss": 3.0771,
      "step": 257980
    },
    {
      "epoch": 0.07548646320448565,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001484022,
      "loss": 3.0836,
      "step": 257990
    },
    {
      "epoch": 0.07550510644064384,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014840020000000002,
      "loss": 3.1668,
      "step": 258000
    },
    {
      "epoch": 0.07552374967680202,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001483982,
      "loss": 3.0524,
      "step": 258010
    },
    {
      "epoch": 0.07554239291296021,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001483962,
      "loss": 3.0122,
      "step": 258020
    },
    {
      "epoch": 0.0755610361491184,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001483942,
      "loss": 3.2017,
      "step": 258030
    },
    {
      "epoch": 0.07557967938527657,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001483922,
      "loss": 3.2409,
      "step": 258040
    },
    {
      "epoch": 0.07559832262143476,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014839020000000002,
      "loss": 3.1497,
      "step": 258050
    },
    {
      "epoch": 0.07561696585759294,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001483882,
      "loss": 3.1339,
      "step": 258060
    },
    {
      "epoch": 0.07563560909375112,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001483862,
      "loss": 3.0544,
      "step": 258070
    },
    {
      "epoch": 0.07565425232990931,
      "grad_norm": 0.75,
      "learning_rate": 0.0001483842,
      "loss": 3.0846,
      "step": 258080
    },
    {
      "epoch": 0.07567289556606749,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001483822,
      "loss": 3.0932,
      "step": 258090
    },
    {
      "epoch": 0.07569153880222568,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014838020000000003,
      "loss": 3.1667,
      "step": 258100
    },
    {
      "epoch": 0.07571018203838387,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001483782,
      "loss": 3.1869,
      "step": 258110
    },
    {
      "epoch": 0.07572882527454204,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014837620000000002,
      "loss": 3.1805,
      "step": 258120
    },
    {
      "epoch": 0.07574746851070023,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014837419999999999,
      "loss": 3.1767,
      "step": 258130
    },
    {
      "epoch": 0.0757661117468584,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001483722,
      "loss": 3.1596,
      "step": 258140
    },
    {
      "epoch": 0.0757847549830166,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001483702,
      "loss": 3.1877,
      "step": 258150
    },
    {
      "epoch": 0.07580339821917478,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001483682,
      "loss": 3.1674,
      "step": 258160
    },
    {
      "epoch": 0.07582204145533296,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014836620000000002,
      "loss": 3.1158,
      "step": 258170
    },
    {
      "epoch": 0.07584068469149115,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001483642,
      "loss": 3.176,
      "step": 258180
    },
    {
      "epoch": 0.07585932792764932,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014836220000000001,
      "loss": 3.093,
      "step": 258190
    },
    {
      "epoch": 0.07587797116380751,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001483602,
      "loss": 3.1035,
      "step": 258200
    },
    {
      "epoch": 0.0758966143999657,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001483582,
      "loss": 3.0298,
      "step": 258210
    },
    {
      "epoch": 0.07591525763612388,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001483562,
      "loss": 3.1186,
      "step": 258220
    },
    {
      "epoch": 0.07593390087228206,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001483542,
      "loss": 3.1511,
      "step": 258230
    },
    {
      "epoch": 0.07595254410844025,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014835220000000002,
      "loss": 3.1622,
      "step": 258240
    },
    {
      "epoch": 0.07597118734459843,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014835020000000001,
      "loss": 3.102,
      "step": 258250
    },
    {
      "epoch": 0.07598983058075662,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001483482,
      "loss": 3.2103,
      "step": 258260
    },
    {
      "epoch": 0.07600847381691479,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001483462,
      "loss": 3.1019,
      "step": 258270
    },
    {
      "epoch": 0.07602711705307298,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001483442,
      "loss": 3.2852,
      "step": 258280
    },
    {
      "epoch": 0.07604576028923117,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001483422,
      "loss": 2.9785,
      "step": 258290
    },
    {
      "epoch": 0.07606440352538935,
      "grad_norm": 0.890625,
      "learning_rate": 0.00014834020000000002,
      "loss": 3.0796,
      "step": 258300
    },
    {
      "epoch": 0.07608304676154753,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014833820000000002,
      "loss": 3.0683,
      "step": 258310
    },
    {
      "epoch": 0.07610168999770571,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001483362,
      "loss": 3.1315,
      "step": 258320
    },
    {
      "epoch": 0.0761203332338639,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001483342,
      "loss": 3.0698,
      "step": 258330
    },
    {
      "epoch": 0.07613897647002209,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001483322,
      "loss": 3.2159,
      "step": 258340
    },
    {
      "epoch": 0.07615761970618026,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001483302,
      "loss": 3.1828,
      "step": 258350
    },
    {
      "epoch": 0.07617626294233845,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001483282,
      "loss": 3.0066,
      "step": 258360
    },
    {
      "epoch": 0.07619490617849664,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014832620000000002,
      "loss": 3.0766,
      "step": 258370
    },
    {
      "epoch": 0.07621354941465482,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001483242,
      "loss": 3.0718,
      "step": 258380
    },
    {
      "epoch": 0.076232192650813,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001483222,
      "loss": 2.994,
      "step": 258390
    },
    {
      "epoch": 0.07625083588697118,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001483202,
      "loss": 3.1316,
      "step": 258400
    },
    {
      "epoch": 0.07626947912312937,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001483182,
      "loss": 3.0146,
      "step": 258410
    },
    {
      "epoch": 0.07628812235928756,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014831620000000002,
      "loss": 3.1641,
      "step": 258420
    },
    {
      "epoch": 0.07630676559544573,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001483142,
      "loss": 3.0929,
      "step": 258430
    },
    {
      "epoch": 0.07632540883160392,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001483122,
      "loss": 3.0904,
      "step": 258440
    },
    {
      "epoch": 0.07634405206776211,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001483102,
      "loss": 3.0612,
      "step": 258450
    },
    {
      "epoch": 0.07636269530392029,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001483082,
      "loss": 3.1721,
      "step": 258460
    },
    {
      "epoch": 0.07638133854007848,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014830620000000003,
      "loss": 3.1196,
      "step": 258470
    },
    {
      "epoch": 0.07639998177623665,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001483042,
      "loss": 3.2197,
      "step": 258480
    },
    {
      "epoch": 0.07641862501239484,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014830220000000002,
      "loss": 3.1213,
      "step": 258490
    },
    {
      "epoch": 0.07643726824855303,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014830020000000001,
      "loss": 3.1957,
      "step": 258500
    },
    {
      "epoch": 0.0764559114847112,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001482982,
      "loss": 3.1978,
      "step": 258510
    },
    {
      "epoch": 0.07647455472086939,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001482962,
      "loss": 3.1202,
      "step": 258520
    },
    {
      "epoch": 0.07649319795702757,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001482942,
      "loss": 3.1317,
      "step": 258530
    },
    {
      "epoch": 0.07651184119318576,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014829220000000002,
      "loss": 3.1063,
      "step": 258540
    },
    {
      "epoch": 0.07653048442934395,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001482902,
      "loss": 3.0881,
      "step": 258550
    },
    {
      "epoch": 0.07654912766550212,
      "grad_norm": 0.75,
      "learning_rate": 0.00014828820000000001,
      "loss": 3.0897,
      "step": 258560
    },
    {
      "epoch": 0.07656777090166031,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001482862,
      "loss": 3.1137,
      "step": 258570
    },
    {
      "epoch": 0.0765864141378185,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001482842,
      "loss": 3.1565,
      "step": 258580
    },
    {
      "epoch": 0.07660505737397667,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014828220000000003,
      "loss": 3.1331,
      "step": 258590
    },
    {
      "epoch": 0.07662370061013486,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001482802,
      "loss": 3.1997,
      "step": 258600
    },
    {
      "epoch": 0.07664234384629304,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014827820000000002,
      "loss": 3.0843,
      "step": 258610
    },
    {
      "epoch": 0.07666098708245123,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001482762,
      "loss": 3.0553,
      "step": 258620
    },
    {
      "epoch": 0.07667963031860942,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001482742,
      "loss": 3.1589,
      "step": 258630
    },
    {
      "epoch": 0.07669827355476759,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001482722,
      "loss": 3.1691,
      "step": 258640
    },
    {
      "epoch": 0.07671691679092578,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001482702,
      "loss": 3.1527,
      "step": 258650
    },
    {
      "epoch": 0.07673556002708395,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014826820000000002,
      "loss": 3.1345,
      "step": 258660
    },
    {
      "epoch": 0.07675420326324214,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001482662,
      "loss": 3.1052,
      "step": 258670
    },
    {
      "epoch": 0.07677284649940033,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014826420000000002,
      "loss": 3.2273,
      "step": 258680
    },
    {
      "epoch": 0.07679148973555851,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001482622,
      "loss": 3.2697,
      "step": 258690
    },
    {
      "epoch": 0.0768101329717167,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001482602,
      "loss": 3.3208,
      "step": 258700
    },
    {
      "epoch": 0.07682877620787489,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001482582,
      "loss": 3.166,
      "step": 258710
    },
    {
      "epoch": 0.07684741944403306,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001482562,
      "loss": 3.2063,
      "step": 258720
    },
    {
      "epoch": 0.07686606268019125,
      "grad_norm": 0.921875,
      "learning_rate": 0.00014825420000000002,
      "loss": 2.9868,
      "step": 258730
    },
    {
      "epoch": 0.07688470591634942,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001482522,
      "loss": 3.1331,
      "step": 258740
    },
    {
      "epoch": 0.07690334915250761,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001482502,
      "loss": 3.2039,
      "step": 258750
    },
    {
      "epoch": 0.0769219923886658,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001482482,
      "loss": 3.1445,
      "step": 258760
    },
    {
      "epoch": 0.07694063562482398,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001482462,
      "loss": 3.1911,
      "step": 258770
    },
    {
      "epoch": 0.07695927886098217,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001482442,
      "loss": 3.2147,
      "step": 258780
    },
    {
      "epoch": 0.07697792209714036,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001482422,
      "loss": 3.2419,
      "step": 258790
    },
    {
      "epoch": 0.07699656533329853,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014824020000000002,
      "loss": 3.0211,
      "step": 258800
    },
    {
      "epoch": 0.07701520856945672,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001482382,
      "loss": 3.195,
      "step": 258810
    },
    {
      "epoch": 0.0770338518056149,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001482362,
      "loss": 3.1329,
      "step": 258820
    },
    {
      "epoch": 0.07705249504177308,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001482342,
      "loss": 3.0885,
      "step": 258830
    },
    {
      "epoch": 0.07707113827793127,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001482322,
      "loss": 3.1549,
      "step": 258840
    },
    {
      "epoch": 0.07708978151408945,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001482302,
      "loss": 3.1703,
      "step": 258850
    },
    {
      "epoch": 0.07710842475024764,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014822820000000002,
      "loss": 3.0499,
      "step": 258860
    },
    {
      "epoch": 0.07712706798640581,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014822620000000001,
      "loss": 3.1688,
      "step": 258870
    },
    {
      "epoch": 0.077145711222564,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001482242,
      "loss": 3.2714,
      "step": 258880
    },
    {
      "epoch": 0.07716435445872219,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001482222,
      "loss": 3.1171,
      "step": 258890
    },
    {
      "epoch": 0.07718299769488036,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001482202,
      "loss": 3.0183,
      "step": 258900
    },
    {
      "epoch": 0.07720164093103855,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014821820000000002,
      "loss": 3.132,
      "step": 258910
    },
    {
      "epoch": 0.07722028416719674,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001482162,
      "loss": 3.1494,
      "step": 258920
    },
    {
      "epoch": 0.07723892740335492,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014821420000000001,
      "loss": 3.1168,
      "step": 258930
    },
    {
      "epoch": 0.0772575706395131,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001482122,
      "loss": 3.2183,
      "step": 258940
    },
    {
      "epoch": 0.07727621387567128,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001482102,
      "loss": 3.1391,
      "step": 258950
    },
    {
      "epoch": 0.07729485711182947,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001482082,
      "loss": 3.1004,
      "step": 258960
    },
    {
      "epoch": 0.07731350034798766,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001482062,
      "loss": 3.2673,
      "step": 258970
    },
    {
      "epoch": 0.07733214358414583,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014820420000000002,
      "loss": 3.0908,
      "step": 258980
    },
    {
      "epoch": 0.07735078682030402,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014820220000000002,
      "loss": 3.1009,
      "step": 258990
    },
    {
      "epoch": 0.07736943005646221,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001482002,
      "loss": 3.0737,
      "step": 259000
    },
    {
      "epoch": 0.07738807329262039,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001481982,
      "loss": 3.0397,
      "step": 259010
    },
    {
      "epoch": 0.07740671652877858,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001481962,
      "loss": 3.0981,
      "step": 259020
    },
    {
      "epoch": 0.07742535976493675,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014819420000000003,
      "loss": 3.2492,
      "step": 259030
    },
    {
      "epoch": 0.07744400300109494,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001481922,
      "loss": 3.0766,
      "step": 259040
    },
    {
      "epoch": 0.07746264623725313,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014819020000000002,
      "loss": 3.0494,
      "step": 259050
    },
    {
      "epoch": 0.0774812894734113,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001481882,
      "loss": 3.0807,
      "step": 259060
    },
    {
      "epoch": 0.0774999327095695,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001481862,
      "loss": 3.0471,
      "step": 259070
    },
    {
      "epoch": 0.07751857594572767,
      "grad_norm": 0.578125,
      "learning_rate": 0.00014818420000000003,
      "loss": 3.0483,
      "step": 259080
    },
    {
      "epoch": 0.07753721918188586,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001481822,
      "loss": 2.9922,
      "step": 259090
    },
    {
      "epoch": 0.07755586241804405,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014818020000000002,
      "loss": 3.1051,
      "step": 259100
    },
    {
      "epoch": 0.07757450565420222,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001481782,
      "loss": 3.2198,
      "step": 259110
    },
    {
      "epoch": 0.07759314889036041,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001481762,
      "loss": 3.1053,
      "step": 259120
    },
    {
      "epoch": 0.0776117921265186,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001481742,
      "loss": 3.1115,
      "step": 259130
    },
    {
      "epoch": 0.07763043536267678,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001481722,
      "loss": 3.0715,
      "step": 259140
    },
    {
      "epoch": 0.07764907859883496,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014817020000000003,
      "loss": 3.1516,
      "step": 259150
    },
    {
      "epoch": 0.07766772183499314,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001481682,
      "loss": 3.1434,
      "step": 259160
    },
    {
      "epoch": 0.07768636507115133,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014816620000000002,
      "loss": 3.0381,
      "step": 259170
    },
    {
      "epoch": 0.07770500830730952,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001481642,
      "loss": 3.1062,
      "step": 259180
    },
    {
      "epoch": 0.07772365154346769,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001481622,
      "loss": 3.1187,
      "step": 259190
    },
    {
      "epoch": 0.07774229477962588,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001481602,
      "loss": 3.0884,
      "step": 259200
    },
    {
      "epoch": 0.07776093801578406,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001481582,
      "loss": 3.152,
      "step": 259210
    },
    {
      "epoch": 0.07777958125194225,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014815620000000002,
      "loss": 3.0867,
      "step": 259220
    },
    {
      "epoch": 0.07779822448810043,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001481542,
      "loss": 3.1595,
      "step": 259230
    },
    {
      "epoch": 0.07781686772425861,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014815220000000001,
      "loss": 3.1391,
      "step": 259240
    },
    {
      "epoch": 0.0778355109604168,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001481502,
      "loss": 3.1482,
      "step": 259250
    },
    {
      "epoch": 0.07785415419657499,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001481482,
      "loss": 3.1238,
      "step": 259260
    },
    {
      "epoch": 0.07787279743273316,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001481462,
      "loss": 3.0736,
      "step": 259270
    },
    {
      "epoch": 0.07789144066889135,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001481442,
      "loss": 3.0864,
      "step": 259280
    },
    {
      "epoch": 0.07791008390504953,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014814220000000002,
      "loss": 3.1792,
      "step": 259290
    },
    {
      "epoch": 0.07792872714120772,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014814020000000002,
      "loss": 3.2525,
      "step": 259300
    },
    {
      "epoch": 0.0779473703773659,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001481382,
      "loss": 3.0417,
      "step": 259310
    },
    {
      "epoch": 0.07796601361352408,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001481362,
      "loss": 3.0842,
      "step": 259320
    },
    {
      "epoch": 0.07798465684968227,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001481342,
      "loss": 3.2055,
      "step": 259330
    },
    {
      "epoch": 0.07800330008584046,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001481322,
      "loss": 3.2196,
      "step": 259340
    },
    {
      "epoch": 0.07802194332199863,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001481302,
      "loss": 3.0999,
      "step": 259350
    },
    {
      "epoch": 0.07804058655815682,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014812820000000002,
      "loss": 3.147,
      "step": 259360
    },
    {
      "epoch": 0.078059229794315,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001481262,
      "loss": 3.1448,
      "step": 259370
    },
    {
      "epoch": 0.07807787303047319,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001481242,
      "loss": 3.111,
      "step": 259380
    },
    {
      "epoch": 0.07809651626663137,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001481222,
      "loss": 3.1325,
      "step": 259390
    },
    {
      "epoch": 0.07811515950278955,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001481202,
      "loss": 3.1051,
      "step": 259400
    },
    {
      "epoch": 0.07813380273894774,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001481182,
      "loss": 3.1942,
      "step": 259410
    },
    {
      "epoch": 0.07815244597510591,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014811620000000002,
      "loss": 3.1588,
      "step": 259420
    },
    {
      "epoch": 0.0781710892112641,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001481142,
      "loss": 3.1646,
      "step": 259430
    },
    {
      "epoch": 0.07818973244742229,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001481122,
      "loss": 3.0784,
      "step": 259440
    },
    {
      "epoch": 0.07820837568358047,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001481102,
      "loss": 3.143,
      "step": 259450
    },
    {
      "epoch": 0.07822701891973866,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001481082,
      "loss": 3.1807,
      "step": 259460
    },
    {
      "epoch": 0.07824566215589684,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014810620000000002,
      "loss": 3.204,
      "step": 259470
    },
    {
      "epoch": 0.07826430539205502,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001481042,
      "loss": 3.1081,
      "step": 259480
    },
    {
      "epoch": 0.07828294862821321,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001481022,
      "loss": 3.0885,
      "step": 259490
    },
    {
      "epoch": 0.07830159186437138,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001481002,
      "loss": 3.109,
      "step": 259500
    },
    {
      "epoch": 0.07832023510052957,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001480982,
      "loss": 3.1312,
      "step": 259510
    },
    {
      "epoch": 0.07833887833668776,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001480962,
      "loss": 3.0595,
      "step": 259520
    },
    {
      "epoch": 0.07835752157284594,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001480942,
      "loss": 3.0019,
      "step": 259530
    },
    {
      "epoch": 0.07837616480900413,
      "grad_norm": 0.75,
      "learning_rate": 0.00014809220000000002,
      "loss": 3.1576,
      "step": 259540
    },
    {
      "epoch": 0.0783948080451623,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014809020000000001,
      "loss": 3.0132,
      "step": 259550
    },
    {
      "epoch": 0.07841345128132049,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001480882,
      "loss": 3.0373,
      "step": 259560
    },
    {
      "epoch": 0.07843209451747868,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001480862,
      "loss": 3.0432,
      "step": 259570
    },
    {
      "epoch": 0.07845073775363685,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001480842,
      "loss": 3.1248,
      "step": 259580
    },
    {
      "epoch": 0.07846938098979504,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014808220000000002,
      "loss": 3.0908,
      "step": 259590
    },
    {
      "epoch": 0.07848802422595323,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001480802,
      "loss": 3.0417,
      "step": 259600
    },
    {
      "epoch": 0.0785066674621114,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014807820000000001,
      "loss": 3.0933,
      "step": 259610
    },
    {
      "epoch": 0.0785253106982696,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001480762,
      "loss": 3.0477,
      "step": 259620
    },
    {
      "epoch": 0.07854395393442777,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001480742,
      "loss": 3.121,
      "step": 259630
    },
    {
      "epoch": 0.07856259717058596,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014807220000000003,
      "loss": 3.1563,
      "step": 259640
    },
    {
      "epoch": 0.07858124040674415,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001480702,
      "loss": 3.0912,
      "step": 259650
    },
    {
      "epoch": 0.07859988364290232,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014806820000000002,
      "loss": 3.1083,
      "step": 259660
    },
    {
      "epoch": 0.07861852687906051,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001480662,
      "loss": 3.0968,
      "step": 259670
    },
    {
      "epoch": 0.0786371701152187,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001480642,
      "loss": 3.1168,
      "step": 259680
    },
    {
      "epoch": 0.07865581335137688,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001480622,
      "loss": 3.1767,
      "step": 259690
    },
    {
      "epoch": 0.07867445658753507,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001480602,
      "loss": 3.1238,
      "step": 259700
    },
    {
      "epoch": 0.07869309982369324,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014805820000000003,
      "loss": 3.1683,
      "step": 259710
    },
    {
      "epoch": 0.07871174305985143,
      "grad_norm": 0.75,
      "learning_rate": 0.0001480562,
      "loss": 3.1613,
      "step": 259720
    },
    {
      "epoch": 0.07873038629600962,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014805420000000002,
      "loss": 3.087,
      "step": 259730
    },
    {
      "epoch": 0.0787490295321678,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014805219999999998,
      "loss": 3.1378,
      "step": 259740
    },
    {
      "epoch": 0.07876767276832598,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001480502,
      "loss": 3.0185,
      "step": 259750
    },
    {
      "epoch": 0.07878631600448416,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001480482,
      "loss": 3.0995,
      "step": 259760
    },
    {
      "epoch": 0.07880495924064235,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001480462,
      "loss": 3.0191,
      "step": 259770
    },
    {
      "epoch": 0.07882360247680054,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014804420000000002,
      "loss": 3.2556,
      "step": 259780
    },
    {
      "epoch": 0.07884224571295871,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001480422,
      "loss": 3.1697,
      "step": 259790
    },
    {
      "epoch": 0.0788608889491169,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001480402,
      "loss": 3.2082,
      "step": 259800
    },
    {
      "epoch": 0.07887953218527509,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001480382,
      "loss": 3.2153,
      "step": 259810
    },
    {
      "epoch": 0.07889817542143326,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001480362,
      "loss": 3.1877,
      "step": 259820
    },
    {
      "epoch": 0.07891681865759145,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001480342,
      "loss": 3.2847,
      "step": 259830
    },
    {
      "epoch": 0.07893546189374963,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001480322,
      "loss": 3.1417,
      "step": 259840
    },
    {
      "epoch": 0.07895410512990782,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014803020000000002,
      "loss": 3.1694,
      "step": 259850
    },
    {
      "epoch": 0.078972748366066,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014802820000000001,
      "loss": 3.0531,
      "step": 259860
    },
    {
      "epoch": 0.07899139160222418,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001480262,
      "loss": 3.2289,
      "step": 259870
    },
    {
      "epoch": 0.07901003483838237,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001480242,
      "loss": 3.1714,
      "step": 259880
    },
    {
      "epoch": 0.07902867807454055,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001480222,
      "loss": 3.1258,
      "step": 259890
    },
    {
      "epoch": 0.07904732131069873,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001480202,
      "loss": 3.0839,
      "step": 259900
    },
    {
      "epoch": 0.07906596454685692,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014801820000000002,
      "loss": 3.0534,
      "step": 259910
    },
    {
      "epoch": 0.0790846077830151,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014801620000000001,
      "loss": 3.0511,
      "step": 259920
    },
    {
      "epoch": 0.07910325101917329,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001480142,
      "loss": 3.2602,
      "step": 259930
    },
    {
      "epoch": 0.07912189425533148,
      "grad_norm": 0.75,
      "learning_rate": 0.0001480122,
      "loss": 3.0689,
      "step": 259940
    },
    {
      "epoch": 0.07914053749148965,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001480102,
      "loss": 3.1263,
      "step": 259950
    },
    {
      "epoch": 0.07915918072764784,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001480082,
      "loss": 3.0496,
      "step": 259960
    },
    {
      "epoch": 0.07917782396380602,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001480062,
      "loss": 3.1693,
      "step": 259970
    },
    {
      "epoch": 0.0791964671999642,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014800420000000002,
      "loss": 3.2802,
      "step": 259980
    },
    {
      "epoch": 0.0792151104361224,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001480022,
      "loss": 3.003,
      "step": 259990
    },
    {
      "epoch": 0.07923375367228057,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001480002,
      "loss": 3.0551,
      "step": 260000
    },
    {
      "epoch": 0.07925239690843876,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001479982,
      "loss": 3.0041,
      "step": 260010
    },
    {
      "epoch": 0.07927104014459695,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001479962,
      "loss": 2.9419,
      "step": 260020
    },
    {
      "epoch": 0.07928968338075512,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014799420000000002,
      "loss": 3.0807,
      "step": 260030
    },
    {
      "epoch": 0.07930832661691331,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014799220000000002,
      "loss": 3.0739,
      "step": 260040
    },
    {
      "epoch": 0.07932696985307149,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001479902,
      "loss": 3.0254,
      "step": 260050
    },
    {
      "epoch": 0.07934561308922967,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001479882,
      "loss": 3.1604,
      "step": 260060
    },
    {
      "epoch": 0.07936425632538786,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001479862,
      "loss": 3.1012,
      "step": 260070
    },
    {
      "epoch": 0.07938289956154604,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014798420000000003,
      "loss": 3.0939,
      "step": 260080
    },
    {
      "epoch": 0.07940154279770423,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001479822,
      "loss": 3.0447,
      "step": 260090
    },
    {
      "epoch": 0.0794201860338624,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014798020000000002,
      "loss": 3.1869,
      "step": 260100
    },
    {
      "epoch": 0.07943882927002059,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001479782,
      "loss": 3.1105,
      "step": 260110
    },
    {
      "epoch": 0.07945747250617878,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001479762,
      "loss": 2.9993,
      "step": 260120
    },
    {
      "epoch": 0.07947611574233696,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001479742,
      "loss": 3.2451,
      "step": 260130
    },
    {
      "epoch": 0.07949475897849514,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001479722,
      "loss": 3.0361,
      "step": 260140
    },
    {
      "epoch": 0.07951340221465333,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014797020000000002,
      "loss": 3.0664,
      "step": 260150
    },
    {
      "epoch": 0.07953204545081151,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001479682,
      "loss": 3.1875,
      "step": 260160
    },
    {
      "epoch": 0.0795506886869697,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001479662,
      "loss": 3.1913,
      "step": 260170
    },
    {
      "epoch": 0.07956933192312787,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001479642,
      "loss": 3.0793,
      "step": 260180
    },
    {
      "epoch": 0.07958797515928606,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001479622,
      "loss": 3.0581,
      "step": 260190
    },
    {
      "epoch": 0.07960661839544425,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014796020000000003,
      "loss": 3.1061,
      "step": 260200
    },
    {
      "epoch": 0.07962526163160243,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001479582,
      "loss": 3.1075,
      "step": 260210
    },
    {
      "epoch": 0.07964390486776061,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014795620000000002,
      "loss": 2.951,
      "step": 260220
    },
    {
      "epoch": 0.07966254810391879,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001479542,
      "loss": 3.0515,
      "step": 260230
    },
    {
      "epoch": 0.07968119134007698,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001479522,
      "loss": 3.0755,
      "step": 260240
    },
    {
      "epoch": 0.07969983457623517,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001479502,
      "loss": 3.1217,
      "step": 260250
    },
    {
      "epoch": 0.07971847781239334,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001479482,
      "loss": 3.103,
      "step": 260260
    },
    {
      "epoch": 0.07973712104855153,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014794620000000002,
      "loss": 3.1215,
      "step": 260270
    },
    {
      "epoch": 0.07975576428470972,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001479442,
      "loss": 3.064,
      "step": 260280
    },
    {
      "epoch": 0.0797744075208679,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014794220000000001,
      "loss": 3.0471,
      "step": 260290
    },
    {
      "epoch": 0.07979305075702608,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001479402,
      "loss": 3.153,
      "step": 260300
    },
    {
      "epoch": 0.07981169399318426,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001479382,
      "loss": 3.211,
      "step": 260310
    },
    {
      "epoch": 0.07983033722934245,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001479362,
      "loss": 3.2021,
      "step": 260320
    },
    {
      "epoch": 0.07984898046550064,
      "grad_norm": 0.75,
      "learning_rate": 0.0001479342,
      "loss": 3.0668,
      "step": 260330
    },
    {
      "epoch": 0.07986762370165881,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014793220000000002,
      "loss": 3.1149,
      "step": 260340
    },
    {
      "epoch": 0.079886266937817,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001479302,
      "loss": 3.1485,
      "step": 260350
    },
    {
      "epoch": 0.07990491017397519,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001479282,
      "loss": 3.1363,
      "step": 260360
    },
    {
      "epoch": 0.07992355341013337,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001479262,
      "loss": 3.1095,
      "step": 260370
    },
    {
      "epoch": 0.07994219664629156,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001479242,
      "loss": 3.131,
      "step": 260380
    },
    {
      "epoch": 0.07996083988244973,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001479222,
      "loss": 3.1676,
      "step": 260390
    },
    {
      "epoch": 0.07997948311860792,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001479202,
      "loss": 3.042,
      "step": 260400
    },
    {
      "epoch": 0.07999812635476611,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014791820000000002,
      "loss": 3.0196,
      "step": 260410
    },
    {
      "epoch": 0.08001676959092428,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001479162,
      "loss": 3.0893,
      "step": 260420
    },
    {
      "epoch": 0.08003541282708247,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001479142,
      "loss": 3.1821,
      "step": 260430
    },
    {
      "epoch": 0.08005405606324065,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001479122,
      "loss": 3.0932,
      "step": 260440
    },
    {
      "epoch": 0.08007269929939884,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001479102,
      "loss": 3.1026,
      "step": 260450
    },
    {
      "epoch": 0.08009134253555703,
      "grad_norm": 0.75,
      "learning_rate": 0.0001479082,
      "loss": 3.1722,
      "step": 260460
    },
    {
      "epoch": 0.0801099857717152,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014790620000000002,
      "loss": 3.1909,
      "step": 260470
    },
    {
      "epoch": 0.08012862900787339,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001479042,
      "loss": 3.0381,
      "step": 260480
    },
    {
      "epoch": 0.08014727224403158,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001479022,
      "loss": 2.9084,
      "step": 260490
    },
    {
      "epoch": 0.08016591548018975,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001479002,
      "loss": 3.153,
      "step": 260500
    },
    {
      "epoch": 0.08018455871634794,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001478982,
      "loss": 3.0587,
      "step": 260510
    },
    {
      "epoch": 0.08020320195250612,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014789620000000002,
      "loss": 3.0581,
      "step": 260520
    },
    {
      "epoch": 0.0802218451886643,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014789420000000002,
      "loss": 3.0946,
      "step": 260530
    },
    {
      "epoch": 0.0802404884248225,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014789220000000001,
      "loss": 3.0437,
      "step": 260540
    },
    {
      "epoch": 0.08025913166098067,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001478902,
      "loss": 3.195,
      "step": 260550
    },
    {
      "epoch": 0.08027777489713886,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001478882,
      "loss": 3.0874,
      "step": 260560
    },
    {
      "epoch": 0.08029641813329703,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001478862,
      "loss": 3.1753,
      "step": 260570
    },
    {
      "epoch": 0.08031506136945522,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001478842,
      "loss": 3.1828,
      "step": 260580
    },
    {
      "epoch": 0.08033370460561341,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014788220000000002,
      "loss": 3.1337,
      "step": 260590
    },
    {
      "epoch": 0.08035234784177159,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014788020000000001,
      "loss": 3.0914,
      "step": 260600
    },
    {
      "epoch": 0.08037099107792978,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001478782,
      "loss": 3.1652,
      "step": 260610
    },
    {
      "epoch": 0.08038963431408797,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001478762,
      "loss": 3.2011,
      "step": 260620
    },
    {
      "epoch": 0.08040827755024614,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001478742,
      "loss": 3.0795,
      "step": 260630
    },
    {
      "epoch": 0.08042692078640433,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014787220000000002,
      "loss": 3.0875,
      "step": 260640
    },
    {
      "epoch": 0.0804455640225625,
      "grad_norm": 0.75,
      "learning_rate": 0.0001478702,
      "loss": 3.1622,
      "step": 260650
    },
    {
      "epoch": 0.0804642072587207,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014786820000000002,
      "loss": 3.1981,
      "step": 260660
    },
    {
      "epoch": 0.08048285049487888,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001478662,
      "loss": 3.1175,
      "step": 260670
    },
    {
      "epoch": 0.08050149373103706,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001478642,
      "loss": 3.0519,
      "step": 260680
    },
    {
      "epoch": 0.08052013696719525,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014786220000000003,
      "loss": 3.0846,
      "step": 260690
    },
    {
      "epoch": 0.08053878020335344,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001478602,
      "loss": 3.1141,
      "step": 260700
    },
    {
      "epoch": 0.08055742343951161,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014785820000000002,
      "loss": 3.212,
      "step": 260710
    },
    {
      "epoch": 0.0805760666756698,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001478562,
      "loss": 3.039,
      "step": 260720
    },
    {
      "epoch": 0.08059470991182797,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001478542,
      "loss": 3.0249,
      "step": 260730
    },
    {
      "epoch": 0.08061335314798616,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001478522,
      "loss": 3.0817,
      "step": 260740
    },
    {
      "epoch": 0.08063199638414435,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001478502,
      "loss": 3.1513,
      "step": 260750
    },
    {
      "epoch": 0.08065063962030253,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014784820000000003,
      "loss": 3.0103,
      "step": 260760
    },
    {
      "epoch": 0.08066928285646072,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001478462,
      "loss": 3.2178,
      "step": 260770
    },
    {
      "epoch": 0.08068792609261889,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014784420000000002,
      "loss": 3.0278,
      "step": 260780
    },
    {
      "epoch": 0.08070656932877708,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014784219999999999,
      "loss": 3.1224,
      "step": 260790
    },
    {
      "epoch": 0.08072521256493527,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001478402,
      "loss": 3.0825,
      "step": 260800
    },
    {
      "epoch": 0.08074385580109344,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001478382,
      "loss": 3.1885,
      "step": 260810
    },
    {
      "epoch": 0.08076249903725163,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001478362,
      "loss": 3.1335,
      "step": 260820
    },
    {
      "epoch": 0.08078114227340982,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014783420000000002,
      "loss": 3.0675,
      "step": 260830
    },
    {
      "epoch": 0.080799785509568,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001478322,
      "loss": 3.1256,
      "step": 260840
    },
    {
      "epoch": 0.08081842874572619,
      "grad_norm": 0.75,
      "learning_rate": 0.0001478302,
      "loss": 3.1826,
      "step": 260850
    },
    {
      "epoch": 0.08083707198188436,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001478282,
      "loss": 3.1475,
      "step": 260860
    },
    {
      "epoch": 0.08085571521804255,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001478262,
      "loss": 3.0652,
      "step": 260870
    },
    {
      "epoch": 0.08087435845420074,
      "grad_norm": 0.75,
      "learning_rate": 0.0001478242,
      "loss": 3.1316,
      "step": 260880
    },
    {
      "epoch": 0.08089300169035891,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001478222,
      "loss": 3.0461,
      "step": 260890
    },
    {
      "epoch": 0.0809116449265171,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014782020000000002,
      "loss": 3.1952,
      "step": 260900
    },
    {
      "epoch": 0.08093028816267528,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014781820000000001,
      "loss": 3.0538,
      "step": 260910
    },
    {
      "epoch": 0.08094893139883347,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001478162,
      "loss": 3.0773,
      "step": 260920
    },
    {
      "epoch": 0.08096757463499166,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001478142,
      "loss": 3.1015,
      "step": 260930
    },
    {
      "epoch": 0.08098621787114983,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001478122,
      "loss": 3.1033,
      "step": 260940
    },
    {
      "epoch": 0.08100486110730802,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001478102,
      "loss": 3.1304,
      "step": 260950
    },
    {
      "epoch": 0.08102350434346621,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001478082,
      "loss": 3.0709,
      "step": 260960
    },
    {
      "epoch": 0.08104214757962439,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014780620000000001,
      "loss": 3.1679,
      "step": 260970
    },
    {
      "epoch": 0.08106079081578257,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001478042,
      "loss": 3.0906,
      "step": 260980
    },
    {
      "epoch": 0.08107943405194075,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001478022,
      "loss": 3.202,
      "step": 260990
    },
    {
      "epoch": 0.08109807728809894,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001478002,
      "loss": 3.1154,
      "step": 261000
    },
    {
      "epoch": 0.08111672052425713,
      "grad_norm": 0.75,
      "learning_rate": 0.0001477982,
      "loss": 3.2077,
      "step": 261010
    },
    {
      "epoch": 0.0811353637604153,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014779620000000002,
      "loss": 3.2016,
      "step": 261020
    },
    {
      "epoch": 0.08115400699657349,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014779420000000002,
      "loss": 3.1717,
      "step": 261030
    },
    {
      "epoch": 0.08117265023273168,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001477922,
      "loss": 3.1254,
      "step": 261040
    },
    {
      "epoch": 0.08119129346888986,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001477902,
      "loss": 3.0394,
      "step": 261050
    },
    {
      "epoch": 0.08120993670504804,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001477882,
      "loss": 3.2273,
      "step": 261060
    },
    {
      "epoch": 0.08122857994120622,
      "grad_norm": 0.875,
      "learning_rate": 0.0001477862,
      "loss": 2.994,
      "step": 261070
    },
    {
      "epoch": 0.08124722317736441,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014778420000000002,
      "loss": 2.9934,
      "step": 261080
    },
    {
      "epoch": 0.0812658664135226,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014778220000000002,
      "loss": 3.2833,
      "step": 261090
    },
    {
      "epoch": 0.08128450964968077,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001477802,
      "loss": 3.1421,
      "step": 261100
    },
    {
      "epoch": 0.08130315288583896,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001477782,
      "loss": 3.1775,
      "step": 261110
    },
    {
      "epoch": 0.08132179612199714,
      "grad_norm": 0.75,
      "learning_rate": 0.0001477762,
      "loss": 3.0797,
      "step": 261120
    },
    {
      "epoch": 0.08134043935815533,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014777420000000003,
      "loss": 3.0072,
      "step": 261130
    },
    {
      "epoch": 0.08135908259431351,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001477722,
      "loss": 3.1416,
      "step": 261140
    },
    {
      "epoch": 0.08137772583047169,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014777020000000002,
      "loss": 3.1577,
      "step": 261150
    },
    {
      "epoch": 0.08139636906662988,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001477682,
      "loss": 3.0907,
      "step": 261160
    },
    {
      "epoch": 0.08141501230278807,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001477662,
      "loss": 3.2666,
      "step": 261170
    },
    {
      "epoch": 0.08143365553894624,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001477642,
      "loss": 3.0521,
      "step": 261180
    },
    {
      "epoch": 0.08145229877510443,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001477622,
      "loss": 3.0458,
      "step": 261190
    },
    {
      "epoch": 0.0814709420112626,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014776020000000002,
      "loss": 3.15,
      "step": 261200
    },
    {
      "epoch": 0.0814895852474208,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001477582,
      "loss": 3.2572,
      "step": 261210
    },
    {
      "epoch": 0.08150822848357898,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001477562,
      "loss": 3.0376,
      "step": 261220
    },
    {
      "epoch": 0.08152687171973716,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001477542,
      "loss": 2.9632,
      "step": 261230
    },
    {
      "epoch": 0.08154551495589535,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001477522,
      "loss": 3.1516,
      "step": 261240
    },
    {
      "epoch": 0.08156415819205352,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014775020000000003,
      "loss": 3.1221,
      "step": 261250
    },
    {
      "epoch": 0.08158280142821171,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001477482,
      "loss": 3.1513,
      "step": 261260
    },
    {
      "epoch": 0.0816014446643699,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014774620000000002,
      "loss": 3.148,
      "step": 261270
    },
    {
      "epoch": 0.08162008790052808,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001477442,
      "loss": 3.1294,
      "step": 261280
    },
    {
      "epoch": 0.08163873113668627,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001477422,
      "loss": 3.0707,
      "step": 261290
    },
    {
      "epoch": 0.08165737437284445,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001477402,
      "loss": 3.1858,
      "step": 261300
    },
    {
      "epoch": 0.08167601760900263,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001477382,
      "loss": 3.2302,
      "step": 261310
    },
    {
      "epoch": 0.08169466084516082,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014773620000000002,
      "loss": 2.9728,
      "step": 261320
    },
    {
      "epoch": 0.081713304081319,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001477342,
      "loss": 3.1329,
      "step": 261330
    },
    {
      "epoch": 0.08173194731747718,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014773220000000002,
      "loss": 3.0532,
      "step": 261340
    },
    {
      "epoch": 0.08175059055363537,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014773019999999998,
      "loss": 3.0523,
      "step": 261350
    },
    {
      "epoch": 0.08176923378979355,
      "grad_norm": 0.625,
      "learning_rate": 0.0001477282,
      "loss": 3.0861,
      "step": 261360
    },
    {
      "epoch": 0.08178787702595174,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001477262,
      "loss": 3.1191,
      "step": 261370
    },
    {
      "epoch": 0.08180652026210992,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001477242,
      "loss": 3.0899,
      "step": 261380
    },
    {
      "epoch": 0.0818251634982681,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014772220000000002,
      "loss": 3.1295,
      "step": 261390
    },
    {
      "epoch": 0.08184380673442629,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001477202,
      "loss": 3.1651,
      "step": 261400
    },
    {
      "epoch": 0.08186244997058446,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001477182,
      "loss": 3.1102,
      "step": 261410
    },
    {
      "epoch": 0.08188109320674265,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001477162,
      "loss": 3.2617,
      "step": 261420
    },
    {
      "epoch": 0.08189973644290084,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001477142,
      "loss": 3.1808,
      "step": 261430
    },
    {
      "epoch": 0.08191837967905902,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001477122,
      "loss": 3.0653,
      "step": 261440
    },
    {
      "epoch": 0.0819370229152172,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001477102,
      "loss": 3.0791,
      "step": 261450
    },
    {
      "epoch": 0.08195566615137538,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014770820000000002,
      "loss": 3.2178,
      "step": 261460
    },
    {
      "epoch": 0.08197430938753357,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001477062,
      "loss": 3.1729,
      "step": 261470
    },
    {
      "epoch": 0.08199295262369176,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001477042,
      "loss": 3.3216,
      "step": 261480
    },
    {
      "epoch": 0.08201159585984993,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001477022,
      "loss": 3.2402,
      "step": 261490
    },
    {
      "epoch": 0.08203023909600812,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001477002,
      "loss": 3.1146,
      "step": 261500
    },
    {
      "epoch": 0.08204888233216631,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014769820000000002,
      "loss": 3.2293,
      "step": 261510
    },
    {
      "epoch": 0.08206752556832449,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014769620000000002,
      "loss": 3.2615,
      "step": 261520
    },
    {
      "epoch": 0.08208616880448268,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001476942,
      "loss": 3.1499,
      "step": 261530
    },
    {
      "epoch": 0.08210481204064085,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001476922,
      "loss": 3.2184,
      "step": 261540
    },
    {
      "epoch": 0.08212345527679904,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001476902,
      "loss": 3.202,
      "step": 261550
    },
    {
      "epoch": 0.08214209851295723,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001476882,
      "loss": 3.1624,
      "step": 261560
    },
    {
      "epoch": 0.0821607417491154,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001476862,
      "loss": 3.2082,
      "step": 261570
    },
    {
      "epoch": 0.08217938498527359,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014768420000000002,
      "loss": 3.1677,
      "step": 261580
    },
    {
      "epoch": 0.08219802822143177,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014768220000000001,
      "loss": 3.1672,
      "step": 261590
    },
    {
      "epoch": 0.08221667145758996,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001476802,
      "loss": 3.1894,
      "step": 261600
    },
    {
      "epoch": 0.08223531469374815,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001476782,
      "loss": 3.2411,
      "step": 261610
    },
    {
      "epoch": 0.08225395792990632,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001476762,
      "loss": 3.195,
      "step": 261620
    },
    {
      "epoch": 0.08227260116606451,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001476742,
      "loss": 3.0636,
      "step": 261630
    },
    {
      "epoch": 0.0822912444022227,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014767220000000002,
      "loss": 3.2737,
      "step": 261640
    },
    {
      "epoch": 0.08230988763838087,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014767020000000001,
      "loss": 3.0315,
      "step": 261650
    },
    {
      "epoch": 0.08232853087453906,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001476682,
      "loss": 3.1441,
      "step": 261660
    },
    {
      "epoch": 0.08234717411069724,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001476662,
      "loss": 3.1791,
      "step": 261670
    },
    {
      "epoch": 0.08236581734685543,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001476642,
      "loss": 3.1702,
      "step": 261680
    },
    {
      "epoch": 0.08238446058301362,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014766220000000002,
      "loss": 3.1603,
      "step": 261690
    },
    {
      "epoch": 0.08240310381917179,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001476602,
      "loss": 3.1692,
      "step": 261700
    },
    {
      "epoch": 0.08242174705532998,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014765820000000002,
      "loss": 3.1048,
      "step": 261710
    },
    {
      "epoch": 0.08244039029148817,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001476562,
      "loss": 3.0447,
      "step": 261720
    },
    {
      "epoch": 0.08245903352764634,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001476542,
      "loss": 3.1726,
      "step": 261730
    },
    {
      "epoch": 0.08247767676380453,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001476522,
      "loss": 3.1018,
      "step": 261740
    },
    {
      "epoch": 0.08249631999996271,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001476502,
      "loss": 3.1102,
      "step": 261750
    },
    {
      "epoch": 0.0825149632361209,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014764820000000002,
      "loss": 3.033,
      "step": 261760
    },
    {
      "epoch": 0.08253360647227909,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001476462,
      "loss": 3.2228,
      "step": 261770
    },
    {
      "epoch": 0.08255224970843726,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001476442,
      "loss": 3.17,
      "step": 261780
    },
    {
      "epoch": 0.08257089294459545,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001476422,
      "loss": 3.1264,
      "step": 261790
    },
    {
      "epoch": 0.08258953618075363,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001476402,
      "loss": 3.1893,
      "step": 261800
    },
    {
      "epoch": 0.08260817941691181,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014763820000000003,
      "loss": 3.1384,
      "step": 261810
    },
    {
      "epoch": 0.08262682265307,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001476362,
      "loss": 3.2876,
      "step": 261820
    },
    {
      "epoch": 0.08264546588922818,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014763420000000002,
      "loss": 3.1635,
      "step": 261830
    },
    {
      "epoch": 0.08266410912538637,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014763219999999999,
      "loss": 3.1904,
      "step": 261840
    },
    {
      "epoch": 0.08268275236154456,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001476302,
      "loss": 3.0391,
      "step": 261850
    },
    {
      "epoch": 0.08270139559770273,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001476282,
      "loss": 3.2128,
      "step": 261860
    },
    {
      "epoch": 0.08272003883386092,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001476262,
      "loss": 3.1161,
      "step": 261870
    },
    {
      "epoch": 0.0827386820700191,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014762420000000002,
      "loss": 3.3396,
      "step": 261880
    },
    {
      "epoch": 0.08275732530617728,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001476222,
      "loss": 3.113,
      "step": 261890
    },
    {
      "epoch": 0.08277596854233547,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001476202,
      "loss": 3.0361,
      "step": 261900
    },
    {
      "epoch": 0.08279461177849365,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001476182,
      "loss": 3.2,
      "step": 261910
    },
    {
      "epoch": 0.08281325501465184,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001476162,
      "loss": 3.1344,
      "step": 261920
    },
    {
      "epoch": 0.08283189825081001,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001476142,
      "loss": 2.9778,
      "step": 261930
    },
    {
      "epoch": 0.0828505414869682,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001476122,
      "loss": 3.1312,
      "step": 261940
    },
    {
      "epoch": 0.08286918472312639,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014761020000000002,
      "loss": 3.0543,
      "step": 261950
    },
    {
      "epoch": 0.08288782795928457,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001476082,
      "loss": 3.1876,
      "step": 261960
    },
    {
      "epoch": 0.08290647119544275,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001476062,
      "loss": 3.0827,
      "step": 261970
    },
    {
      "epoch": 0.08292511443160094,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001476042,
      "loss": 3.1188,
      "step": 261980
    },
    {
      "epoch": 0.08294375766775912,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001476022,
      "loss": 3.2338,
      "step": 261990
    },
    {
      "epoch": 0.08296240090391731,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014760020000000002,
      "loss": 3.1807,
      "step": 262000
    },
    {
      "epoch": 0.08298104414007548,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001475982,
      "loss": 3.2145,
      "step": 262010
    },
    {
      "epoch": 0.08299968737623367,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014759620000000002,
      "loss": 3.0666,
      "step": 262020
    },
    {
      "epoch": 0.08301833061239186,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001475942,
      "loss": 2.9917,
      "step": 262030
    },
    {
      "epoch": 0.08303697384855004,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0001475922,
      "loss": 3.0691,
      "step": 262040
    },
    {
      "epoch": 0.08305561708470822,
      "grad_norm": 0.75,
      "learning_rate": 0.0001475902,
      "loss": 3.0364,
      "step": 262050
    },
    {
      "epoch": 0.08307426032086641,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001475882,
      "loss": 3.1428,
      "step": 262060
    },
    {
      "epoch": 0.08309290355702459,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014758620000000002,
      "loss": 3.1176,
      "step": 262070
    },
    {
      "epoch": 0.08311154679318278,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014758420000000002,
      "loss": 3.0477,
      "step": 262080
    },
    {
      "epoch": 0.08313019002934095,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001475822,
      "loss": 3.165,
      "step": 262090
    },
    {
      "epoch": 0.08314883326549914,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001475802,
      "loss": 3.133,
      "step": 262100
    },
    {
      "epoch": 0.08316747650165733,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001475782,
      "loss": 3.0512,
      "step": 262110
    },
    {
      "epoch": 0.0831861197378155,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001475762,
      "loss": 3.1367,
      "step": 262120
    },
    {
      "epoch": 0.0832047629739737,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014757420000000002,
      "loss": 3.1952,
      "step": 262130
    },
    {
      "epoch": 0.08322340621013187,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014757220000000002,
      "loss": 3.1475,
      "step": 262140
    },
    {
      "epoch": 0.08324204944629006,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001475702,
      "loss": 3.1365,
      "step": 262150
    },
    {
      "epoch": 0.08326069268244825,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001475682,
      "loss": 3.2482,
      "step": 262160
    },
    {
      "epoch": 0.08327933591860642,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001475662,
      "loss": 3.106,
      "step": 262170
    },
    {
      "epoch": 0.08329797915476461,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001475642,
      "loss": 3.1697,
      "step": 262180
    },
    {
      "epoch": 0.0833166223909228,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001475622,
      "loss": 3.1373,
      "step": 262190
    },
    {
      "epoch": 0.08333526562708098,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014756020000000002,
      "loss": 3.1089,
      "step": 262200
    },
    {
      "epoch": 0.08335390886323917,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001475582,
      "loss": 3.1216,
      "step": 262210
    },
    {
      "epoch": 0.08337255209939734,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001475562,
      "loss": 3.2397,
      "step": 262220
    },
    {
      "epoch": 0.08339119533555553,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001475542,
      "loss": 3.1447,
      "step": 262230
    },
    {
      "epoch": 0.08340983857171372,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001475522,
      "loss": 3.1987,
      "step": 262240
    },
    {
      "epoch": 0.0834284818078719,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014755020000000002,
      "loss": 3.1686,
      "step": 262250
    },
    {
      "epoch": 0.08344712504403008,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001475482,
      "loss": 3.1235,
      "step": 262260
    },
    {
      "epoch": 0.08346576828018826,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014754620000000001,
      "loss": 3.1243,
      "step": 262270
    },
    {
      "epoch": 0.08348441151634645,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001475442,
      "loss": 3.2253,
      "step": 262280
    },
    {
      "epoch": 0.08350305475250464,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001475422,
      "loss": 3.1326,
      "step": 262290
    },
    {
      "epoch": 0.08352169798866281,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014754020000000003,
      "loss": 3.0701,
      "step": 262300
    },
    {
      "epoch": 0.083540341224821,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001475382,
      "loss": 3.0302,
      "step": 262310
    },
    {
      "epoch": 0.08355898446097919,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014753620000000002,
      "loss": 3.2277,
      "step": 262320
    },
    {
      "epoch": 0.08357762769713736,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001475342,
      "loss": 3.1511,
      "step": 262330
    },
    {
      "epoch": 0.08359627093329555,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001475322,
      "loss": 3.0525,
      "step": 262340
    },
    {
      "epoch": 0.08361491416945373,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001475302,
      "loss": 3.1508,
      "step": 262350
    },
    {
      "epoch": 0.08363355740561192,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001475282,
      "loss": 3.1092,
      "step": 262360
    },
    {
      "epoch": 0.0836522006417701,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014752620000000002,
      "loss": 3.186,
      "step": 262370
    },
    {
      "epoch": 0.08367084387792828,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001475242,
      "loss": 2.9813,
      "step": 262380
    },
    {
      "epoch": 0.08368948711408647,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014752220000000002,
      "loss": 3.1482,
      "step": 262390
    },
    {
      "epoch": 0.08370813035024466,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014752019999999998,
      "loss": 3.135,
      "step": 262400
    },
    {
      "epoch": 0.08372677358640283,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001475182,
      "loss": 3.0567,
      "step": 262410
    },
    {
      "epoch": 0.08374541682256102,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001475162,
      "loss": 3.1489,
      "step": 262420
    },
    {
      "epoch": 0.0837640600587192,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001475142,
      "loss": 3.0115,
      "step": 262430
    },
    {
      "epoch": 0.08378270329487739,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014751220000000002,
      "loss": 3.1598,
      "step": 262440
    },
    {
      "epoch": 0.08380134653103558,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001475102,
      "loss": 3.0569,
      "step": 262450
    },
    {
      "epoch": 0.08381998976719375,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001475082,
      "loss": 3.2138,
      "step": 262460
    },
    {
      "epoch": 0.08383863300335194,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001475062,
      "loss": 3.0817,
      "step": 262470
    },
    {
      "epoch": 0.08385727623951011,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001475042,
      "loss": 3.1924,
      "step": 262480
    },
    {
      "epoch": 0.0838759194756683,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014750220000000003,
      "loss": 3.1223,
      "step": 262490
    },
    {
      "epoch": 0.08389456271182649,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001475002,
      "loss": 3.1995,
      "step": 262500
    },
    {
      "epoch": 0.08391320594798467,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014749820000000002,
      "loss": 3.186,
      "step": 262510
    },
    {
      "epoch": 0.08393184918414286,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001474962,
      "loss": 3.075,
      "step": 262520
    },
    {
      "epoch": 0.08395049242030105,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001474942,
      "loss": 3.2107,
      "step": 262530
    },
    {
      "epoch": 0.08396913565645922,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001474922,
      "loss": 3.0123,
      "step": 262540
    },
    {
      "epoch": 0.08398777889261741,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001474902,
      "loss": 3.1111,
      "step": 262550
    },
    {
      "epoch": 0.08400642212877558,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014748820000000002,
      "loss": 3.1391,
      "step": 262560
    },
    {
      "epoch": 0.08402506536493377,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001474862,
      "loss": 3.2726,
      "step": 262570
    },
    {
      "epoch": 0.08404370860109196,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001474842,
      "loss": 3.2403,
      "step": 262580
    },
    {
      "epoch": 0.08406235183725014,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001474822,
      "loss": 3.0254,
      "step": 262590
    },
    {
      "epoch": 0.08408099507340833,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001474802,
      "loss": 3.1446,
      "step": 262600
    },
    {
      "epoch": 0.0840996383095665,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001474782,
      "loss": 3.2161,
      "step": 262610
    },
    {
      "epoch": 0.08411828154572469,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001474762,
      "loss": 2.9946,
      "step": 262620
    },
    {
      "epoch": 0.08413692478188288,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00014747420000000002,
      "loss": 3.0803,
      "step": 262630
    },
    {
      "epoch": 0.08415556801804105,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014747220000000001,
      "loss": 3.1939,
      "step": 262640
    },
    {
      "epoch": 0.08417421125419924,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001474702,
      "loss": 3.1829,
      "step": 262650
    },
    {
      "epoch": 0.08419285449035743,
      "grad_norm": 0.75,
      "learning_rate": 0.0001474682,
      "loss": 3.1008,
      "step": 262660
    },
    {
      "epoch": 0.08421149772651561,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001474662,
      "loss": 3.0951,
      "step": 262670
    },
    {
      "epoch": 0.0842301409626738,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001474642,
      "loss": 3.1981,
      "step": 262680
    },
    {
      "epoch": 0.08424878419883197,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014746220000000002,
      "loss": 3.1675,
      "step": 262690
    },
    {
      "epoch": 0.08426742743499016,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014746020000000002,
      "loss": 3.1153,
      "step": 262700
    },
    {
      "epoch": 0.08428607067114835,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001474582,
      "loss": 3.0884,
      "step": 262710
    },
    {
      "epoch": 0.08430471390730652,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001474562,
      "loss": 3.098,
      "step": 262720
    },
    {
      "epoch": 0.08432335714346471,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001474542,
      "loss": 3.1935,
      "step": 262730
    },
    {
      "epoch": 0.0843420003796229,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014745220000000002,
      "loss": 3.1517,
      "step": 262740
    },
    {
      "epoch": 0.08436064361578108,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001474502,
      "loss": 3.0555,
      "step": 262750
    },
    {
      "epoch": 0.08437928685193927,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014744820000000002,
      "loss": 3.2124,
      "step": 262760
    },
    {
      "epoch": 0.08439793008809744,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001474462,
      "loss": 3.1093,
      "step": 262770
    },
    {
      "epoch": 0.08441657332425563,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001474442,
      "loss": 3.1815,
      "step": 262780
    },
    {
      "epoch": 0.08443521656041382,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001474422,
      "loss": 3.2188,
      "step": 262790
    },
    {
      "epoch": 0.084453859796572,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001474402,
      "loss": 3.1242,
      "step": 262800
    },
    {
      "epoch": 0.08447250303273018,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014743820000000002,
      "loss": 3.0157,
      "step": 262810
    },
    {
      "epoch": 0.08449114626888836,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001474362,
      "loss": 3.1719,
      "step": 262820
    },
    {
      "epoch": 0.08450978950504655,
      "grad_norm": 0.75,
      "learning_rate": 0.0001474342,
      "loss": 3.0218,
      "step": 262830
    },
    {
      "epoch": 0.08452843274120474,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001474322,
      "loss": 3.0963,
      "step": 262840
    },
    {
      "epoch": 0.08454707597736291,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001474302,
      "loss": 3.0207,
      "step": 262850
    },
    {
      "epoch": 0.0845657192135211,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014742820000000003,
      "loss": 3.16,
      "step": 262860
    },
    {
      "epoch": 0.08458436244967929,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001474262,
      "loss": 3.1584,
      "step": 262870
    },
    {
      "epoch": 0.08460300568583747,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014742420000000002,
      "loss": 3.0917,
      "step": 262880
    },
    {
      "epoch": 0.08462164892199565,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014742219999999999,
      "loss": 3.1203,
      "step": 262890
    },
    {
      "epoch": 0.08464029215815383,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001474202,
      "loss": 3.2392,
      "step": 262900
    },
    {
      "epoch": 0.08465893539431202,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001474182,
      "loss": 3.107,
      "step": 262910
    },
    {
      "epoch": 0.08467757863047021,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001474162,
      "loss": 3.2921,
      "step": 262920
    },
    {
      "epoch": 0.08469622186662838,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014741420000000002,
      "loss": 3.0356,
      "step": 262930
    },
    {
      "epoch": 0.08471486510278657,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001474122,
      "loss": 3.0955,
      "step": 262940
    },
    {
      "epoch": 0.08473350833894475,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014741020000000001,
      "loss": 3.0765,
      "step": 262950
    },
    {
      "epoch": 0.08475215157510294,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001474082,
      "loss": 3.118,
      "step": 262960
    },
    {
      "epoch": 0.08477079481126112,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001474062,
      "loss": 3.118,
      "step": 262970
    },
    {
      "epoch": 0.0847894380474193,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014740420000000003,
      "loss": 3.1017,
      "step": 262980
    },
    {
      "epoch": 0.08480808128357749,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001474022,
      "loss": 3.1693,
      "step": 262990
    },
    {
      "epoch": 0.08482672451973568,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014740020000000002,
      "loss": 3.1565,
      "step": 263000
    },
    {
      "epoch": 0.08484536775589385,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001473982,
      "loss": 3.2,
      "step": 263010
    },
    {
      "epoch": 0.08486401099205204,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001473962,
      "loss": 2.9118,
      "step": 263020
    },
    {
      "epoch": 0.08488265422821022,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001473942,
      "loss": 3.066,
      "step": 263030
    },
    {
      "epoch": 0.0849012974643684,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001473922,
      "loss": 3.1219,
      "step": 263040
    },
    {
      "epoch": 0.0849199407005266,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014739020000000002,
      "loss": 3.0022,
      "step": 263050
    },
    {
      "epoch": 0.08493858393668477,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001473882,
      "loss": 3.09,
      "step": 263060
    },
    {
      "epoch": 0.08495722717284296,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014738620000000002,
      "loss": 3.1353,
      "step": 263070
    },
    {
      "epoch": 0.08497587040900115,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001473842,
      "loss": 3.0639,
      "step": 263080
    },
    {
      "epoch": 0.08499451364515932,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001473822,
      "loss": 3.0914,
      "step": 263090
    },
    {
      "epoch": 0.08501315688131751,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001473802,
      "loss": 3.1175,
      "step": 263100
    },
    {
      "epoch": 0.08503180011747569,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001473782,
      "loss": 3.1285,
      "step": 263110
    },
    {
      "epoch": 0.08505044335363388,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014737620000000002,
      "loss": 3.0754,
      "step": 263120
    },
    {
      "epoch": 0.08506908658979206,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014737420000000002,
      "loss": 3.0706,
      "step": 263130
    },
    {
      "epoch": 0.08508772982595024,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001473722,
      "loss": 3.1325,
      "step": 263140
    },
    {
      "epoch": 0.08510637306210843,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001473702,
      "loss": 3.0265,
      "step": 263150
    },
    {
      "epoch": 0.0851250162982666,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001473682,
      "loss": 3.0343,
      "step": 263160
    },
    {
      "epoch": 0.08514365953442479,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001473662,
      "loss": 3.0639,
      "step": 263170
    },
    {
      "epoch": 0.08516230277058298,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001473642,
      "loss": 3.1535,
      "step": 263180
    },
    {
      "epoch": 0.08518094600674116,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014736220000000002,
      "loss": 3.1203,
      "step": 263190
    },
    {
      "epoch": 0.08519958924289935,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001473602,
      "loss": 3.1709,
      "step": 263200
    },
    {
      "epoch": 0.08521823247905753,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001473582,
      "loss": 3.079,
      "step": 263210
    },
    {
      "epoch": 0.08523687571521571,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001473562,
      "loss": 3.1865,
      "step": 263220
    },
    {
      "epoch": 0.0852555189513739,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001473542,
      "loss": 3.1318,
      "step": 263230
    },
    {
      "epoch": 0.08527416218753207,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001473522,
      "loss": 3.1443,
      "step": 263240
    },
    {
      "epoch": 0.08529280542369026,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014735020000000002,
      "loss": 3.0915,
      "step": 263250
    },
    {
      "epoch": 0.08531144865984845,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001473482,
      "loss": 3.0716,
      "step": 263260
    },
    {
      "epoch": 0.08533009189600663,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001473462,
      "loss": 3.0979,
      "step": 263270
    },
    {
      "epoch": 0.08534873513216482,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001473442,
      "loss": 3.1572,
      "step": 263280
    },
    {
      "epoch": 0.08536737836832299,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001473422,
      "loss": 3.0841,
      "step": 263290
    },
    {
      "epoch": 0.08538602160448118,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00014734020000000002,
      "loss": 3.179,
      "step": 263300
    },
    {
      "epoch": 0.08540466484063937,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001473382,
      "loss": 3.113,
      "step": 263310
    },
    {
      "epoch": 0.08542330807679754,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014733620000000001,
      "loss": 3.0467,
      "step": 263320
    },
    {
      "epoch": 0.08544195131295573,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001473342,
      "loss": 3.2271,
      "step": 263330
    },
    {
      "epoch": 0.08546059454911392,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001473322,
      "loss": 3.2832,
      "step": 263340
    },
    {
      "epoch": 0.0854792377852721,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001473302,
      "loss": 3.1556,
      "step": 263350
    },
    {
      "epoch": 0.08549788102143029,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001473282,
      "loss": 3.1454,
      "step": 263360
    },
    {
      "epoch": 0.08551652425758846,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014732620000000002,
      "loss": 3.0755,
      "step": 263370
    },
    {
      "epoch": 0.08553516749374665,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001473242,
      "loss": 3.1299,
      "step": 263380
    },
    {
      "epoch": 0.08555381072990484,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001473222,
      "loss": 3.109,
      "step": 263390
    },
    {
      "epoch": 0.08557245396606301,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001473202,
      "loss": 3.0994,
      "step": 263400
    },
    {
      "epoch": 0.0855910972022212,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001473182,
      "loss": 3.0403,
      "step": 263410
    },
    {
      "epoch": 0.08560974043837939,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014731620000000002,
      "loss": 3.2149,
      "step": 263420
    },
    {
      "epoch": 0.08562838367453757,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001473142,
      "loss": 3.1466,
      "step": 263430
    },
    {
      "epoch": 0.08564702691069576,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014731220000000002,
      "loss": 3.1673,
      "step": 263440
    },
    {
      "epoch": 0.08566567014685393,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001473102,
      "loss": 3.0808,
      "step": 263450
    },
    {
      "epoch": 0.08568431338301212,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001473082,
      "loss": 3.0484,
      "step": 263460
    },
    {
      "epoch": 0.08570295661917031,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014730620000000003,
      "loss": 3.1774,
      "step": 263470
    },
    {
      "epoch": 0.08572159985532848,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001473042,
      "loss": 3.1488,
      "step": 263480
    },
    {
      "epoch": 0.08574024309148667,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014730220000000002,
      "loss": 3.1373,
      "step": 263490
    },
    {
      "epoch": 0.08575888632764485,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001473002,
      "loss": 3.0647,
      "step": 263500
    },
    {
      "epoch": 0.08577752956380304,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001472982,
      "loss": 3.0563,
      "step": 263510
    },
    {
      "epoch": 0.08579617279996123,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001472962,
      "loss": 3.0355,
      "step": 263520
    },
    {
      "epoch": 0.0858148160361194,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001472942,
      "loss": 3.127,
      "step": 263530
    },
    {
      "epoch": 0.08583345927227759,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014729220000000003,
      "loss": 3.1584,
      "step": 263540
    },
    {
      "epoch": 0.08585210250843578,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001472902,
      "loss": 3.0797,
      "step": 263550
    },
    {
      "epoch": 0.08587074574459395,
      "grad_norm": 0.75,
      "learning_rate": 0.00014728820000000002,
      "loss": 3.0474,
      "step": 263560
    },
    {
      "epoch": 0.08588938898075214,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014728619999999999,
      "loss": 3.1562,
      "step": 263570
    },
    {
      "epoch": 0.08590803221691032,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001472842,
      "loss": 3.1044,
      "step": 263580
    },
    {
      "epoch": 0.08592667545306851,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001472822,
      "loss": 3.1646,
      "step": 263590
    },
    {
      "epoch": 0.0859453186892267,
      "grad_norm": 0.75,
      "learning_rate": 0.0001472802,
      "loss": 3.061,
      "step": 263600
    },
    {
      "epoch": 0.08596396192538487,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014727820000000002,
      "loss": 3.0397,
      "step": 263610
    },
    {
      "epoch": 0.08598260516154306,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001472762,
      "loss": 3.2966,
      "step": 263620
    },
    {
      "epoch": 0.08600124839770125,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014727420000000001,
      "loss": 3.1769,
      "step": 263630
    },
    {
      "epoch": 0.08601989163385942,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001472722,
      "loss": 3.1303,
      "step": 263640
    },
    {
      "epoch": 0.08603853487001761,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001472702,
      "loss": 3.1256,
      "step": 263650
    },
    {
      "epoch": 0.08605717810617579,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001472682,
      "loss": 3.1013,
      "step": 263660
    },
    {
      "epoch": 0.08607582134233398,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001472662,
      "loss": 3.1151,
      "step": 263670
    },
    {
      "epoch": 0.08609446457849217,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014726420000000002,
      "loss": 2.992,
      "step": 263680
    },
    {
      "epoch": 0.08611310781465034,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014726220000000001,
      "loss": 3.2862,
      "step": 263690
    },
    {
      "epoch": 0.08613175105080853,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001472602,
      "loss": 3.1023,
      "step": 263700
    },
    {
      "epoch": 0.0861503942869667,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001472582,
      "loss": 3.1884,
      "step": 263710
    },
    {
      "epoch": 0.0861690375231249,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001472562,
      "loss": 3.0079,
      "step": 263720
    },
    {
      "epoch": 0.08618768075928308,
      "grad_norm": 0.75,
      "learning_rate": 0.0001472542,
      "loss": 3.1454,
      "step": 263730
    },
    {
      "epoch": 0.08620632399544126,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014725220000000002,
      "loss": 3.0848,
      "step": 263740
    },
    {
      "epoch": 0.08622496723159945,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014725020000000002,
      "loss": 3.0597,
      "step": 263750
    },
    {
      "epoch": 0.08624361046775764,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001472482,
      "loss": 3.1767,
      "step": 263760
    },
    {
      "epoch": 0.08626225370391581,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001472462,
      "loss": 3.2505,
      "step": 263770
    },
    {
      "epoch": 0.086280896940074,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001472442,
      "loss": 3.0493,
      "step": 263780
    },
    {
      "epoch": 0.08629954017623218,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001472422,
      "loss": 3.227,
      "step": 263790
    },
    {
      "epoch": 0.08631818341239036,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001472402,
      "loss": 3.0582,
      "step": 263800
    },
    {
      "epoch": 0.08633682664854855,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014723820000000002,
      "loss": 3.1209,
      "step": 263810
    },
    {
      "epoch": 0.08635546988470673,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001472362,
      "loss": 3.1447,
      "step": 263820
    },
    {
      "epoch": 0.08637411312086492,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001472342,
      "loss": 3.2197,
      "step": 263830
    },
    {
      "epoch": 0.08639275635702309,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001472322,
      "loss": 3.2622,
      "step": 263840
    },
    {
      "epoch": 0.08641139959318128,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001472302,
      "loss": 3.2097,
      "step": 263850
    },
    {
      "epoch": 0.08643004282933947,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014722820000000002,
      "loss": 3.2158,
      "step": 263860
    },
    {
      "epoch": 0.08644868606549765,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001472262,
      "loss": 3.142,
      "step": 263870
    },
    {
      "epoch": 0.08646732930165583,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001472242,
      "loss": 3.2871,
      "step": 263880
    },
    {
      "epoch": 0.08648597253781402,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001472222,
      "loss": 3.1835,
      "step": 263890
    },
    {
      "epoch": 0.0865046157739722,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001472202,
      "loss": 3.1961,
      "step": 263900
    },
    {
      "epoch": 0.08652325901013039,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014721820000000003,
      "loss": 3.0805,
      "step": 263910
    },
    {
      "epoch": 0.08654190224628856,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001472162,
      "loss": 3.114,
      "step": 263920
    },
    {
      "epoch": 0.08656054548244675,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014721420000000002,
      "loss": 3.0978,
      "step": 263930
    },
    {
      "epoch": 0.08657918871860494,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001472122,
      "loss": 3.0664,
      "step": 263940
    },
    {
      "epoch": 0.08659783195476312,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001472102,
      "loss": 3.1842,
      "step": 263950
    },
    {
      "epoch": 0.0866164751909213,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001472082,
      "loss": 3.0385,
      "step": 263960
    },
    {
      "epoch": 0.0866351184270795,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001472062,
      "loss": 3.1204,
      "step": 263970
    },
    {
      "epoch": 0.08665376166323767,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014720420000000002,
      "loss": 3.0956,
      "step": 263980
    },
    {
      "epoch": 0.08667240489939586,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001472022,
      "loss": 3.0756,
      "step": 263990
    },
    {
      "epoch": 0.08669104813555403,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014720020000000001,
      "loss": 3.1377,
      "step": 264000
    },
    {
      "epoch": 0.08670969137171222,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001471982,
      "loss": 3.1594,
      "step": 264010
    },
    {
      "epoch": 0.08672833460787041,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001471962,
      "loss": 3.1517,
      "step": 264020
    },
    {
      "epoch": 0.08674697784402859,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014719420000000003,
      "loss": 3.075,
      "step": 264030
    },
    {
      "epoch": 0.08676562108018677,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001471922,
      "loss": 3.1361,
      "step": 264040
    },
    {
      "epoch": 0.08678426431634495,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014719020000000002,
      "loss": 3.0802,
      "step": 264050
    },
    {
      "epoch": 0.08680290755250314,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001471882,
      "loss": 3.1956,
      "step": 264060
    },
    {
      "epoch": 0.08682155078866133,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001471862,
      "loss": 2.9521,
      "step": 264070
    },
    {
      "epoch": 0.0868401940248195,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001471842,
      "loss": 3.1857,
      "step": 264080
    },
    {
      "epoch": 0.08685883726097769,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001471822,
      "loss": 3.0539,
      "step": 264090
    },
    {
      "epoch": 0.08687748049713588,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00014718020000000002,
      "loss": 3.2802,
      "step": 264100
    },
    {
      "epoch": 0.08689612373329406,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001471782,
      "loss": 3.0493,
      "step": 264110
    },
    {
      "epoch": 0.08691476696945225,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014717620000000002,
      "loss": 3.1066,
      "step": 264120
    },
    {
      "epoch": 0.08693341020561042,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001471742,
      "loss": 3.266,
      "step": 264130
    },
    {
      "epoch": 0.08695205344176861,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001471722,
      "loss": 3.1508,
      "step": 264140
    },
    {
      "epoch": 0.0869706966779268,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001471702,
      "loss": 3.1242,
      "step": 264150
    },
    {
      "epoch": 0.08698933991408497,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001471682,
      "loss": 2.9727,
      "step": 264160
    },
    {
      "epoch": 0.08700798315024316,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014716620000000002,
      "loss": 3.1259,
      "step": 264170
    },
    {
      "epoch": 0.08702662638640134,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001471642,
      "loss": 3.0983,
      "step": 264180
    },
    {
      "epoch": 0.08704526962255953,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001471622,
      "loss": 3.1984,
      "step": 264190
    },
    {
      "epoch": 0.08706391285871772,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001471602,
      "loss": 3.0325,
      "step": 264200
    },
    {
      "epoch": 0.08708255609487589,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001471582,
      "loss": 3.1171,
      "step": 264210
    },
    {
      "epoch": 0.08710119933103408,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001471562,
      "loss": 3.03,
      "step": 264220
    },
    {
      "epoch": 0.08711984256719227,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001471542,
      "loss": 3.0747,
      "step": 264230
    },
    {
      "epoch": 0.08713848580335044,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014715220000000002,
      "loss": 3.0933,
      "step": 264240
    },
    {
      "epoch": 0.08715712903950863,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001471502,
      "loss": 3.1071,
      "step": 264250
    },
    {
      "epoch": 0.08717577227566681,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001471482,
      "loss": 3.1182,
      "step": 264260
    },
    {
      "epoch": 0.087194415511825,
      "grad_norm": 0.875,
      "learning_rate": 0.0001471462,
      "loss": 3.0997,
      "step": 264270
    },
    {
      "epoch": 0.08721305874798319,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001471442,
      "loss": 3.1982,
      "step": 264280
    },
    {
      "epoch": 0.08723170198414136,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001471422,
      "loss": 3.0759,
      "step": 264290
    },
    {
      "epoch": 0.08725034522029955,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014714020000000002,
      "loss": 3.2237,
      "step": 264300
    },
    {
      "epoch": 0.08726898845645774,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014713820000000001,
      "loss": 3.2369,
      "step": 264310
    },
    {
      "epoch": 0.08728763169261591,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001471362,
      "loss": 3.1228,
      "step": 264320
    },
    {
      "epoch": 0.0873062749287741,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001471342,
      "loss": 3.1322,
      "step": 264330
    },
    {
      "epoch": 0.08732491816493228,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001471322,
      "loss": 3.0745,
      "step": 264340
    },
    {
      "epoch": 0.08734356140109047,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014713020000000002,
      "loss": 3.0492,
      "step": 264350
    },
    {
      "epoch": 0.08736220463724866,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001471282,
      "loss": 3.2933,
      "step": 264360
    },
    {
      "epoch": 0.08738084787340683,
      "grad_norm": 0.609375,
      "learning_rate": 0.00014712620000000001,
      "loss": 3.2315,
      "step": 264370
    },
    {
      "epoch": 0.08739949110956502,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001471242,
      "loss": 3.2151,
      "step": 264380
    },
    {
      "epoch": 0.0874181343457232,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001471222,
      "loss": 3.1626,
      "step": 264390
    },
    {
      "epoch": 0.08743677758188138,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001471202,
      "loss": 3.2094,
      "step": 264400
    },
    {
      "epoch": 0.08745542081803957,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001471182,
      "loss": 2.9808,
      "step": 264410
    },
    {
      "epoch": 0.08747406405419775,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014711620000000002,
      "loss": 3.1084,
      "step": 264420
    },
    {
      "epoch": 0.08749270729035594,
      "grad_norm": 0.625,
      "learning_rate": 0.00014711420000000002,
      "loss": 3.0792,
      "step": 264430
    },
    {
      "epoch": 0.08751135052651413,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001471122,
      "loss": 3.1266,
      "step": 264440
    },
    {
      "epoch": 0.0875299937626723,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001471102,
      "loss": 3.163,
      "step": 264450
    },
    {
      "epoch": 0.08754863699883049,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001471082,
      "loss": 3.181,
      "step": 264460
    },
    {
      "epoch": 0.08756728023498866,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014710620000000002,
      "loss": 3.1392,
      "step": 264470
    },
    {
      "epoch": 0.08758592347114685,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001471042,
      "loss": 3.1136,
      "step": 264480
    },
    {
      "epoch": 0.08760456670730504,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014710220000000002,
      "loss": 3.1838,
      "step": 264490
    },
    {
      "epoch": 0.08762320994346322,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001471002,
      "loss": 3.0452,
      "step": 264500
    },
    {
      "epoch": 0.0876418531796214,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001470982,
      "loss": 3.1604,
      "step": 264510
    },
    {
      "epoch": 0.08766049641577958,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014709620000000003,
      "loss": 3.1565,
      "step": 264520
    },
    {
      "epoch": 0.08767913965193777,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001470942,
      "loss": 3.1196,
      "step": 264530
    },
    {
      "epoch": 0.08769778288809596,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014709220000000002,
      "loss": 3.1158,
      "step": 264540
    },
    {
      "epoch": 0.08771642612425413,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001470902,
      "loss": 3.1104,
      "step": 264550
    },
    {
      "epoch": 0.08773506936041232,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001470882,
      "loss": 3.2004,
      "step": 264560
    },
    {
      "epoch": 0.08775371259657051,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001470862,
      "loss": 3.1269,
      "step": 264570
    },
    {
      "epoch": 0.08777235583272869,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001470842,
      "loss": 3.0641,
      "step": 264580
    },
    {
      "epoch": 0.08779099906888688,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014708220000000003,
      "loss": 3.1057,
      "step": 264590
    },
    {
      "epoch": 0.08780964230504505,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001470802,
      "loss": 3.1673,
      "step": 264600
    },
    {
      "epoch": 0.08782828554120324,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014707820000000002,
      "loss": 3.0544,
      "step": 264610
    },
    {
      "epoch": 0.08784692877736143,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014707619999999999,
      "loss": 3.1945,
      "step": 264620
    },
    {
      "epoch": 0.0878655720135196,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001470742,
      "loss": 3.1787,
      "step": 264630
    },
    {
      "epoch": 0.0878842152496778,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001470722,
      "loss": 3.2794,
      "step": 264640
    },
    {
      "epoch": 0.08790285848583598,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001470702,
      "loss": 3.0685,
      "step": 264650
    },
    {
      "epoch": 0.08792150172199416,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014706820000000002,
      "loss": 3.1178,
      "step": 264660
    },
    {
      "epoch": 0.08794014495815235,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001470662,
      "loss": 3.0466,
      "step": 264670
    },
    {
      "epoch": 0.08795878819431052,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014706420000000001,
      "loss": 3.2091,
      "step": 264680
    },
    {
      "epoch": 0.08797743143046871,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001470622,
      "loss": 3.1738,
      "step": 264690
    },
    {
      "epoch": 0.0879960746666269,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001470602,
      "loss": 3.0702,
      "step": 264700
    },
    {
      "epoch": 0.08801471790278508,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001470582,
      "loss": 3.1141,
      "step": 264710
    },
    {
      "epoch": 0.08803336113894326,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001470562,
      "loss": 3.295,
      "step": 264720
    },
    {
      "epoch": 0.08805200437510144,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014705420000000002,
      "loss": 3.1261,
      "step": 264730
    },
    {
      "epoch": 0.08807064761125963,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014705220000000001,
      "loss": 3.1671,
      "step": 264740
    },
    {
      "epoch": 0.08808929084741782,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001470502,
      "loss": 3.0931,
      "step": 264750
    },
    {
      "epoch": 0.08810793408357599,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001470482,
      "loss": 3.0303,
      "step": 264760
    },
    {
      "epoch": 0.08812657731973418,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001470462,
      "loss": 3.1626,
      "step": 264770
    },
    {
      "epoch": 0.08814522055589237,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001470442,
      "loss": 3.136,
      "step": 264780
    },
    {
      "epoch": 0.08816386379205055,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001470422,
      "loss": 3.1172,
      "step": 264790
    },
    {
      "epoch": 0.08818250702820873,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014704020000000002,
      "loss": 3.2226,
      "step": 264800
    },
    {
      "epoch": 0.08820115026436691,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001470382,
      "loss": 3.084,
      "step": 264810
    },
    {
      "epoch": 0.0882197935005251,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001470362,
      "loss": 3.2148,
      "step": 264820
    },
    {
      "epoch": 0.08823843673668329,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001470342,
      "loss": 3.0856,
      "step": 264830
    },
    {
      "epoch": 0.08825707997284146,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001470322,
      "loss": 3.0876,
      "step": 264840
    },
    {
      "epoch": 0.08827572320899965,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001470302,
      "loss": 3.1954,
      "step": 264850
    },
    {
      "epoch": 0.08829436644515783,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014702820000000002,
      "loss": 3.1446,
      "step": 264860
    },
    {
      "epoch": 0.08831300968131602,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001470262,
      "loss": 2.965,
      "step": 264870
    },
    {
      "epoch": 0.0883316529174742,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001470242,
      "loss": 3.0111,
      "step": 264880
    },
    {
      "epoch": 0.08835029615363238,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001470222,
      "loss": 3.1093,
      "step": 264890
    },
    {
      "epoch": 0.08836893938979057,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001470202,
      "loss": 3.0903,
      "step": 264900
    },
    {
      "epoch": 0.08838758262594876,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014701820000000002,
      "loss": 3.2227,
      "step": 264910
    },
    {
      "epoch": 0.08840622586210693,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014701620000000002,
      "loss": 3.0816,
      "step": 264920
    },
    {
      "epoch": 0.08842486909826512,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001470142,
      "loss": 3.1747,
      "step": 264930
    },
    {
      "epoch": 0.0884435123344233,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001470122,
      "loss": 3.1186,
      "step": 264940
    },
    {
      "epoch": 0.08846215557058149,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001470102,
      "loss": 3.1212,
      "step": 264950
    },
    {
      "epoch": 0.08848079880673967,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014700820000000003,
      "loss": 3.1818,
      "step": 264960
    },
    {
      "epoch": 0.08849944204289785,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001470062,
      "loss": 3.0948,
      "step": 264970
    },
    {
      "epoch": 0.08851808527905604,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014700420000000002,
      "loss": 3.0672,
      "step": 264980
    },
    {
      "epoch": 0.08853672851521423,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014700220000000001,
      "loss": 3.1659,
      "step": 264990
    },
    {
      "epoch": 0.0885553717513724,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001470002,
      "loss": 3.13,
      "step": 265000
    },
    {
      "epoch": 0.08857401498753059,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001469982,
      "loss": 3.1083,
      "step": 265010
    },
    {
      "epoch": 0.08859265822368877,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001469962,
      "loss": 3.1859,
      "step": 265020
    },
    {
      "epoch": 0.08861130145984696,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014699420000000002,
      "loss": 3.2371,
      "step": 265030
    },
    {
      "epoch": 0.08862994469600514,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001469922,
      "loss": 3.0546,
      "step": 265040
    },
    {
      "epoch": 0.08864858793216332,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014699020000000001,
      "loss": 2.9919,
      "step": 265050
    },
    {
      "epoch": 0.08866723116832151,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001469882,
      "loss": 3.231,
      "step": 265060
    },
    {
      "epoch": 0.08868587440447968,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001469862,
      "loss": 3.0839,
      "step": 265070
    },
    {
      "epoch": 0.08870451764063787,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014698420000000003,
      "loss": 3.1724,
      "step": 265080
    },
    {
      "epoch": 0.08872316087679606,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001469822,
      "loss": 3.0505,
      "step": 265090
    },
    {
      "epoch": 0.08874180411295424,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014698020000000002,
      "loss": 3.3038,
      "step": 265100
    },
    {
      "epoch": 0.08876044734911243,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001469782,
      "loss": 3.1898,
      "step": 265110
    },
    {
      "epoch": 0.08877909058527061,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001469762,
      "loss": 3.0911,
      "step": 265120
    },
    {
      "epoch": 0.08879773382142879,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001469742,
      "loss": 3.1341,
      "step": 265130
    },
    {
      "epoch": 0.08881637705758698,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001469722,
      "loss": 3.0225,
      "step": 265140
    },
    {
      "epoch": 0.08883502029374515,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014697020000000002,
      "loss": 3.0921,
      "step": 265150
    },
    {
      "epoch": 0.08885366352990334,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001469682,
      "loss": 3.0744,
      "step": 265160
    },
    {
      "epoch": 0.08887230676606153,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014696620000000002,
      "loss": 3.1844,
      "step": 265170
    },
    {
      "epoch": 0.0888909500022197,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014696419999999998,
      "loss": 3.0692,
      "step": 265180
    },
    {
      "epoch": 0.0889095932383779,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001469622,
      "loss": 3.1556,
      "step": 265190
    },
    {
      "epoch": 0.08892823647453607,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001469602,
      "loss": 3.2929,
      "step": 265200
    },
    {
      "epoch": 0.08894687971069426,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001469582,
      "loss": 3.174,
      "step": 265210
    },
    {
      "epoch": 0.08896552294685245,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014695620000000002,
      "loss": 3.0802,
      "step": 265220
    },
    {
      "epoch": 0.08898416618301062,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001469542,
      "loss": 3.0943,
      "step": 265230
    },
    {
      "epoch": 0.08900280941916881,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001469522,
      "loss": 3.162,
      "step": 265240
    },
    {
      "epoch": 0.089021452655327,
      "grad_norm": 0.75,
      "learning_rate": 0.0001469502,
      "loss": 3.1019,
      "step": 265250
    },
    {
      "epoch": 0.08904009589148518,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001469482,
      "loss": 3.0554,
      "step": 265260
    },
    {
      "epoch": 0.08905873912764337,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001469462,
      "loss": 3.0813,
      "step": 265270
    },
    {
      "epoch": 0.08907738236380154,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001469442,
      "loss": 3.1847,
      "step": 265280
    },
    {
      "epoch": 0.08909602559995973,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014694220000000002,
      "loss": 2.9582,
      "step": 265290
    },
    {
      "epoch": 0.08911466883611792,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001469402,
      "loss": 3.083,
      "step": 265300
    },
    {
      "epoch": 0.0891333120722761,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001469382,
      "loss": 3.1809,
      "step": 265310
    },
    {
      "epoch": 0.08915195530843428,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001469362,
      "loss": 3.0653,
      "step": 265320
    },
    {
      "epoch": 0.08917059854459247,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001469342,
      "loss": 3.127,
      "step": 265330
    },
    {
      "epoch": 0.08918924178075065,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001469322,
      "loss": 3.1706,
      "step": 265340
    },
    {
      "epoch": 0.08920788501690884,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014693020000000002,
      "loss": 3.0927,
      "step": 265350
    },
    {
      "epoch": 0.08922652825306701,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014692820000000001,
      "loss": 3.1341,
      "step": 265360
    },
    {
      "epoch": 0.0892451714892252,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001469262,
      "loss": 3.1116,
      "step": 265370
    },
    {
      "epoch": 0.08926381472538339,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001469242,
      "loss": 3.2292,
      "step": 265380
    },
    {
      "epoch": 0.08928245796154156,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001469222,
      "loss": 3.2085,
      "step": 265390
    },
    {
      "epoch": 0.08930110119769975,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001469202,
      "loss": 3.1037,
      "step": 265400
    },
    {
      "epoch": 0.08931974443385793,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014691820000000002,
      "loss": 3.2155,
      "step": 265410
    },
    {
      "epoch": 0.08933838767001612,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014691620000000001,
      "loss": 3.2305,
      "step": 265420
    },
    {
      "epoch": 0.0893570309061743,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001469142,
      "loss": 3.0575,
      "step": 265430
    },
    {
      "epoch": 0.08937567414233248,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001469122,
      "loss": 3.0746,
      "step": 265440
    },
    {
      "epoch": 0.08939431737849067,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001469102,
      "loss": 3.0461,
      "step": 265450
    },
    {
      "epoch": 0.08941296061464886,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001469082,
      "loss": 3.1005,
      "step": 265460
    },
    {
      "epoch": 0.08943160385080703,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014690620000000002,
      "loss": 3.2197,
      "step": 265470
    },
    {
      "epoch": 0.08945024708696522,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014690420000000002,
      "loss": 2.9205,
      "step": 265480
    },
    {
      "epoch": 0.0894688903231234,
      "grad_norm": 0.75,
      "learning_rate": 0.0001469022,
      "loss": 3.1524,
      "step": 265490
    },
    {
      "epoch": 0.08948753355928159,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001469002,
      "loss": 3.1415,
      "step": 265500
    },
    {
      "epoch": 0.08950617679543978,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001468982,
      "loss": 3.2206,
      "step": 265510
    },
    {
      "epoch": 0.08952482003159795,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00014689620000000003,
      "loss": 3.2149,
      "step": 265520
    },
    {
      "epoch": 0.08954346326775614,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001468942,
      "loss": 3.1304,
      "step": 265530
    },
    {
      "epoch": 0.08956210650391432,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014689220000000002,
      "loss": 3.0628,
      "step": 265540
    },
    {
      "epoch": 0.0895807497400725,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001468902,
      "loss": 3.2017,
      "step": 265550
    },
    {
      "epoch": 0.0895993929762307,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001468882,
      "loss": 3.0894,
      "step": 265560
    },
    {
      "epoch": 0.08961803621238887,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001468862,
      "loss": 3.0773,
      "step": 265570
    },
    {
      "epoch": 0.08963667944854706,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001468842,
      "loss": 3.1759,
      "step": 265580
    },
    {
      "epoch": 0.08965532268470525,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014688220000000002,
      "loss": 3.1418,
      "step": 265590
    },
    {
      "epoch": 0.08967396592086342,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001468802,
      "loss": 3.1407,
      "step": 265600
    },
    {
      "epoch": 0.08969260915702161,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001468782,
      "loss": 3.1063,
      "step": 265610
    },
    {
      "epoch": 0.08971125239317979,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001468762,
      "loss": 3.0548,
      "step": 265620
    },
    {
      "epoch": 0.08972989562933797,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001468742,
      "loss": 3.0626,
      "step": 265630
    },
    {
      "epoch": 0.08974853886549616,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014687220000000003,
      "loss": 3.1408,
      "step": 265640
    },
    {
      "epoch": 0.08976718210165434,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001468702,
      "loss": 3.128,
      "step": 265650
    },
    {
      "epoch": 0.08978582533781253,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014686820000000002,
      "loss": 3.1413,
      "step": 265660
    },
    {
      "epoch": 0.08980446857397072,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001468662,
      "loss": 3.0042,
      "step": 265670
    },
    {
      "epoch": 0.08982311181012889,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001468642,
      "loss": 3.0209,
      "step": 265680
    },
    {
      "epoch": 0.08984175504628708,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001468622,
      "loss": 2.9511,
      "step": 265690
    },
    {
      "epoch": 0.08986039828244526,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001468602,
      "loss": 3.0917,
      "step": 265700
    },
    {
      "epoch": 0.08987904151860344,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014685820000000002,
      "loss": 3.1654,
      "step": 265710
    },
    {
      "epoch": 0.08989768475476163,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001468562,
      "loss": 3.036,
      "step": 265720
    },
    {
      "epoch": 0.08991632799091981,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014685420000000001,
      "loss": 3.0201,
      "step": 265730
    },
    {
      "epoch": 0.089934971227078,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001468522,
      "loss": 3.1967,
      "step": 265740
    },
    {
      "epoch": 0.08995361446323617,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001468502,
      "loss": 2.9959,
      "step": 265750
    },
    {
      "epoch": 0.08997225769939436,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001468482,
      "loss": 3.0664,
      "step": 265760
    },
    {
      "epoch": 0.08999090093555255,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001468462,
      "loss": 3.1185,
      "step": 265770
    },
    {
      "epoch": 0.09000954417171073,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014684420000000002,
      "loss": 3.0665,
      "step": 265780
    },
    {
      "epoch": 0.09002818740786891,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001468422,
      "loss": 3.1628,
      "step": 265790
    },
    {
      "epoch": 0.0900468306440271,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001468402,
      "loss": 3.1814,
      "step": 265800
    },
    {
      "epoch": 0.09006547388018528,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001468382,
      "loss": 3.1271,
      "step": 265810
    },
    {
      "epoch": 0.09008411711634347,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001468362,
      "loss": 3.0364,
      "step": 265820
    },
    {
      "epoch": 0.09010276035250164,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001468342,
      "loss": 3.1287,
      "step": 265830
    },
    {
      "epoch": 0.09012140358865983,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001468322,
      "loss": 3.0841,
      "step": 265840
    },
    {
      "epoch": 0.09014004682481802,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014683020000000002,
      "loss": 3.1662,
      "step": 265850
    },
    {
      "epoch": 0.0901586900609762,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001468282,
      "loss": 3.1563,
      "step": 265860
    },
    {
      "epoch": 0.09017733329713438,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001468262,
      "loss": 3.1067,
      "step": 265870
    },
    {
      "epoch": 0.09019597653329256,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001468242,
      "loss": 3.0591,
      "step": 265880
    },
    {
      "epoch": 0.09021461976945075,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001468222,
      "loss": 3.1544,
      "step": 265890
    },
    {
      "epoch": 0.09023326300560894,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014682020000000002,
      "loss": 3.1101,
      "step": 265900
    },
    {
      "epoch": 0.09025190624176711,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014681820000000002,
      "loss": 3.1342,
      "step": 265910
    },
    {
      "epoch": 0.0902705494779253,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001468162,
      "loss": 3.145,
      "step": 265920
    },
    {
      "epoch": 0.09028919271408349,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001468142,
      "loss": 3.1285,
      "step": 265930
    },
    {
      "epoch": 0.09030783595024167,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001468122,
      "loss": 3.1095,
      "step": 265940
    },
    {
      "epoch": 0.09032647918639986,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001468102,
      "loss": 3.1373,
      "step": 265950
    },
    {
      "epoch": 0.09034512242255803,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014680820000000002,
      "loss": 3.034,
      "step": 265960
    },
    {
      "epoch": 0.09036376565871622,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014680620000000002,
      "loss": 3.0203,
      "step": 265970
    },
    {
      "epoch": 0.09038240889487441,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001468042,
      "loss": 3.0802,
      "step": 265980
    },
    {
      "epoch": 0.09040105213103258,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001468022,
      "loss": 3.084,
      "step": 265990
    },
    {
      "epoch": 0.09041969536719077,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001468002,
      "loss": 3.1678,
      "step": 266000
    },
    {
      "epoch": 0.09043833860334896,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001467982,
      "loss": 3.1111,
      "step": 266010
    },
    {
      "epoch": 0.09045698183950714,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001467962,
      "loss": 3.0759,
      "step": 266020
    },
    {
      "epoch": 0.09047562507566533,
      "grad_norm": 0.75,
      "learning_rate": 0.00014679420000000002,
      "loss": 3.1172,
      "step": 266030
    },
    {
      "epoch": 0.0904942683118235,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014679220000000001,
      "loss": 3.1821,
      "step": 266040
    },
    {
      "epoch": 0.09051291154798169,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001467902,
      "loss": 3.1329,
      "step": 266050
    },
    {
      "epoch": 0.09053155478413988,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001467882,
      "loss": 3.1266,
      "step": 266060
    },
    {
      "epoch": 0.09055019802029805,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001467862,
      "loss": 3.116,
      "step": 266070
    },
    {
      "epoch": 0.09056884125645624,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014678420000000002,
      "loss": 3.0829,
      "step": 266080
    },
    {
      "epoch": 0.09058748449261442,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001467822,
      "loss": 3.1044,
      "step": 266090
    },
    {
      "epoch": 0.0906061277287726,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014678020000000001,
      "loss": 3.1123,
      "step": 266100
    },
    {
      "epoch": 0.0906247709649308,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001467782,
      "loss": 3.2679,
      "step": 266110
    },
    {
      "epoch": 0.09064341420108897,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001467762,
      "loss": 3.0958,
      "step": 266120
    },
    {
      "epoch": 0.09066205743724716,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014677420000000003,
      "loss": 3.0206,
      "step": 266130
    },
    {
      "epoch": 0.09068070067340535,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001467722,
      "loss": 3.0439,
      "step": 266140
    },
    {
      "epoch": 0.09069934390956352,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014677020000000002,
      "loss": 3.1952,
      "step": 266150
    },
    {
      "epoch": 0.09071798714572171,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001467682,
      "loss": 3.1678,
      "step": 266160
    },
    {
      "epoch": 0.09073663038187989,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001467662,
      "loss": 3.1666,
      "step": 266170
    },
    {
      "epoch": 0.09075527361803808,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001467642,
      "loss": 3.0659,
      "step": 266180
    },
    {
      "epoch": 0.09077391685419627,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001467622,
      "loss": 3.0915,
      "step": 266190
    },
    {
      "epoch": 0.09079256009035444,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014676020000000003,
      "loss": 3.1058,
      "step": 266200
    },
    {
      "epoch": 0.09081120332651263,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001467582,
      "loss": 3.062,
      "step": 266210
    },
    {
      "epoch": 0.0908298465626708,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014675620000000002,
      "loss": 3.1124,
      "step": 266220
    },
    {
      "epoch": 0.090848489798829,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014675419999999998,
      "loss": 3.0877,
      "step": 266230
    },
    {
      "epoch": 0.09086713303498718,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001467522,
      "loss": 3.1742,
      "step": 266240
    },
    {
      "epoch": 0.09088577627114536,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001467502,
      "loss": 3.1941,
      "step": 266250
    },
    {
      "epoch": 0.09090441950730355,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001467482,
      "loss": 3.1599,
      "step": 266260
    },
    {
      "epoch": 0.09092306274346174,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014674620000000002,
      "loss": 3.108,
      "step": 266270
    },
    {
      "epoch": 0.09094170597961991,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001467442,
      "loss": 3.0984,
      "step": 266280
    },
    {
      "epoch": 0.0909603492157781,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001467422,
      "loss": 3.1607,
      "step": 266290
    },
    {
      "epoch": 0.09097899245193627,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001467402,
      "loss": 3.1097,
      "step": 266300
    },
    {
      "epoch": 0.09099763568809446,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001467382,
      "loss": 3.2019,
      "step": 266310
    },
    {
      "epoch": 0.09101627892425265,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001467362,
      "loss": 3.2113,
      "step": 266320
    },
    {
      "epoch": 0.09103492216041083,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001467342,
      "loss": 3.196,
      "step": 266330
    },
    {
      "epoch": 0.09105356539656902,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014673220000000002,
      "loss": 3.1167,
      "step": 266340
    },
    {
      "epoch": 0.0910722086327272,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001467302,
      "loss": 3.1298,
      "step": 266350
    },
    {
      "epoch": 0.09109085186888538,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001467282,
      "loss": 3.1126,
      "step": 266360
    },
    {
      "epoch": 0.09110949510504357,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001467262,
      "loss": 3.2219,
      "step": 266370
    },
    {
      "epoch": 0.09112813834120174,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001467242,
      "loss": 3.2026,
      "step": 266380
    },
    {
      "epoch": 0.09114678157735993,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014672220000000002,
      "loss": 3.0868,
      "step": 266390
    },
    {
      "epoch": 0.09116542481351812,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001467202,
      "loss": 3.1213,
      "step": 266400
    },
    {
      "epoch": 0.0911840680496763,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014671820000000001,
      "loss": 3.0369,
      "step": 266410
    },
    {
      "epoch": 0.09120271128583449,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001467162,
      "loss": 3.1205,
      "step": 266420
    },
    {
      "epoch": 0.09122135452199266,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001467142,
      "loss": 3.2599,
      "step": 266430
    },
    {
      "epoch": 0.09123999775815085,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001467122,
      "loss": 3.0762,
      "step": 266440
    },
    {
      "epoch": 0.09125864099430904,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001467102,
      "loss": 3.0593,
      "step": 266450
    },
    {
      "epoch": 0.09127728423046721,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014670820000000002,
      "loss": 3.2757,
      "step": 266460
    },
    {
      "epoch": 0.0912959274666254,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014670620000000002,
      "loss": 3.1416,
      "step": 266470
    },
    {
      "epoch": 0.09131457070278359,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001467042,
      "loss": 3.1237,
      "step": 266480
    },
    {
      "epoch": 0.09133321393894177,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001467022,
      "loss": 3.1589,
      "step": 266490
    },
    {
      "epoch": 0.09135185717509996,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001467002,
      "loss": 3.1115,
      "step": 266500
    },
    {
      "epoch": 0.09137050041125813,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001466982,
      "loss": 3.1528,
      "step": 266510
    },
    {
      "epoch": 0.09138914364741632,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014669620000000002,
      "loss": 3.126,
      "step": 266520
    },
    {
      "epoch": 0.09140778688357451,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014669420000000002,
      "loss": 2.9951,
      "step": 266530
    },
    {
      "epoch": 0.09142643011973269,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001466922,
      "loss": 3.24,
      "step": 266540
    },
    {
      "epoch": 0.09144507335589087,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001466902,
      "loss": 3.1185,
      "step": 266550
    },
    {
      "epoch": 0.09146371659204905,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001466882,
      "loss": 3.1178,
      "step": 266560
    },
    {
      "epoch": 0.09148235982820724,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014668620000000003,
      "loss": 3.0903,
      "step": 266570
    },
    {
      "epoch": 0.09150100306436543,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001466842,
      "loss": 3.2539,
      "step": 266580
    },
    {
      "epoch": 0.0915196463005236,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014668220000000002,
      "loss": 3.1135,
      "step": 266590
    },
    {
      "epoch": 0.09153828953668179,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001466802,
      "loss": 3.1426,
      "step": 266600
    },
    {
      "epoch": 0.09155693277283998,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001466782,
      "loss": 3.1545,
      "step": 266610
    },
    {
      "epoch": 0.09157557600899816,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001466762,
      "loss": 3.1154,
      "step": 266620
    },
    {
      "epoch": 0.09159421924515634,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001466742,
      "loss": 3.1398,
      "step": 266630
    },
    {
      "epoch": 0.09161286248131452,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014667220000000002,
      "loss": 3.161,
      "step": 266640
    },
    {
      "epoch": 0.09163150571747271,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001466702,
      "loss": 3.0467,
      "step": 266650
    },
    {
      "epoch": 0.0916501489536309,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001466682,
      "loss": 3.2545,
      "step": 266660
    },
    {
      "epoch": 0.09166879218978907,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001466662,
      "loss": 3.1127,
      "step": 266670
    },
    {
      "epoch": 0.09168743542594726,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001466642,
      "loss": 3.2133,
      "step": 266680
    },
    {
      "epoch": 0.09170607866210545,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014666220000000003,
      "loss": 3.1238,
      "step": 266690
    },
    {
      "epoch": 0.09172472189826363,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001466602,
      "loss": 3.1546,
      "step": 266700
    },
    {
      "epoch": 0.09174336513442181,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014665820000000002,
      "loss": 3.1284,
      "step": 266710
    },
    {
      "epoch": 0.09176200837057999,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001466562,
      "loss": 3.1111,
      "step": 266720
    },
    {
      "epoch": 0.09178065160673818,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001466542,
      "loss": 3.1043,
      "step": 266730
    },
    {
      "epoch": 0.09179929484289637,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001466522,
      "loss": 3.1528,
      "step": 266740
    },
    {
      "epoch": 0.09181793807905454,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001466502,
      "loss": 3.0698,
      "step": 266750
    },
    {
      "epoch": 0.09183658131521273,
      "grad_norm": 0.75,
      "learning_rate": 0.00014664820000000002,
      "loss": 3.1436,
      "step": 266760
    },
    {
      "epoch": 0.0918552245513709,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001466462,
      "loss": 3.1877,
      "step": 266770
    },
    {
      "epoch": 0.0918738677875291,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014664420000000001,
      "loss": 3.0747,
      "step": 266780
    },
    {
      "epoch": 0.09189251102368728,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014664219999999998,
      "loss": 3.2534,
      "step": 266790
    },
    {
      "epoch": 0.09191115425984546,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001466402,
      "loss": 3.1616,
      "step": 266800
    },
    {
      "epoch": 0.09192979749600365,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001466382,
      "loss": 2.9828,
      "step": 266810
    },
    {
      "epoch": 0.09194844073216184,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001466362,
      "loss": 3.125,
      "step": 266820
    },
    {
      "epoch": 0.09196708396832001,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014663420000000002,
      "loss": 3.1201,
      "step": 266830
    },
    {
      "epoch": 0.0919857272044782,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001466322,
      "loss": 3.1263,
      "step": 266840
    },
    {
      "epoch": 0.09200437044063638,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001466302,
      "loss": 3.0297,
      "step": 266850
    },
    {
      "epoch": 0.09202301367679457,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001466282,
      "loss": 3.2308,
      "step": 266860
    },
    {
      "epoch": 0.09204165691295275,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001466262,
      "loss": 3.0531,
      "step": 266870
    },
    {
      "epoch": 0.09206030014911093,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014662420000000003,
      "loss": 3.102,
      "step": 266880
    },
    {
      "epoch": 0.09207894338526912,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001466222,
      "loss": 3.16,
      "step": 266890
    },
    {
      "epoch": 0.0920975866214273,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014662020000000002,
      "loss": 3.2012,
      "step": 266900
    },
    {
      "epoch": 0.09211622985758548,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001466182,
      "loss": 3.0501,
      "step": 266910
    },
    {
      "epoch": 0.09213487309374367,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001466162,
      "loss": 3.0676,
      "step": 266920
    },
    {
      "epoch": 0.09215351632990185,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001466142,
      "loss": 3.0724,
      "step": 266930
    },
    {
      "epoch": 0.09217215956606004,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001466122,
      "loss": 3.1338,
      "step": 266940
    },
    {
      "epoch": 0.09219080280221822,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014661020000000002,
      "loss": 3.042,
      "step": 266950
    },
    {
      "epoch": 0.0922094460383764,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014660820000000002,
      "loss": 3.1211,
      "step": 266960
    },
    {
      "epoch": 0.09222808927453459,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001466062,
      "loss": 3.0196,
      "step": 266970
    },
    {
      "epoch": 0.09224673251069276,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001466042,
      "loss": 2.9627,
      "step": 266980
    },
    {
      "epoch": 0.09226537574685095,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001466022,
      "loss": 3.189,
      "step": 266990
    },
    {
      "epoch": 0.09228401898300914,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001466002,
      "loss": 3.1215,
      "step": 267000
    },
    {
      "epoch": 0.09230266221916732,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001465982,
      "loss": 3.1349,
      "step": 267010
    },
    {
      "epoch": 0.0923213054553255,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014659620000000002,
      "loss": 3.1003,
      "step": 267020
    },
    {
      "epoch": 0.0923399486914837,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001465942,
      "loss": 3.036,
      "step": 267030
    },
    {
      "epoch": 0.09235859192764187,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001465922,
      "loss": 3.1613,
      "step": 267040
    },
    {
      "epoch": 0.09237723516380006,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001465902,
      "loss": 3.1805,
      "step": 267050
    },
    {
      "epoch": 0.09239587839995823,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001465882,
      "loss": 3.1015,
      "step": 267060
    },
    {
      "epoch": 0.09241452163611642,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001465862,
      "loss": 3.0115,
      "step": 267070
    },
    {
      "epoch": 0.09243316487227461,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014658420000000002,
      "loss": 3.0709,
      "step": 267080
    },
    {
      "epoch": 0.09245180810843279,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014658220000000001,
      "loss": 3.0385,
      "step": 267090
    },
    {
      "epoch": 0.09247045134459098,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001465802,
      "loss": 3.1574,
      "step": 267100
    },
    {
      "epoch": 0.09248909458074915,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001465782,
      "loss": 3.0456,
      "step": 267110
    },
    {
      "epoch": 0.09250773781690734,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001465762,
      "loss": 3.0857,
      "step": 267120
    },
    {
      "epoch": 0.09252638105306553,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014657420000000002,
      "loss": 3.0751,
      "step": 267130
    },
    {
      "epoch": 0.0925450242892237,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001465722,
      "loss": 3.018,
      "step": 267140
    },
    {
      "epoch": 0.09256366752538189,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014657020000000002,
      "loss": 3.0546,
      "step": 267150
    },
    {
      "epoch": 0.09258231076154008,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001465682,
      "loss": 2.9818,
      "step": 267160
    },
    {
      "epoch": 0.09260095399769826,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001465662,
      "loss": 3.1152,
      "step": 267170
    },
    {
      "epoch": 0.09261959723385645,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001465642,
      "loss": 3.0778,
      "step": 267180
    },
    {
      "epoch": 0.09263824047001462,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001465622,
      "loss": 3.1111,
      "step": 267190
    },
    {
      "epoch": 0.09265688370617281,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014656020000000002,
      "loss": 3.0887,
      "step": 267200
    },
    {
      "epoch": 0.092675526942331,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001465582,
      "loss": 3.0992,
      "step": 267210
    },
    {
      "epoch": 0.09269417017848917,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001465562,
      "loss": 3.1229,
      "step": 267220
    },
    {
      "epoch": 0.09271281341464736,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001465542,
      "loss": 2.9785,
      "step": 267230
    },
    {
      "epoch": 0.09273145665080554,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001465522,
      "loss": 3.0016,
      "step": 267240
    },
    {
      "epoch": 0.09275009988696373,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014655020000000003,
      "loss": 3.1426,
      "step": 267250
    },
    {
      "epoch": 0.09276874312312192,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001465482,
      "loss": 3.147,
      "step": 267260
    },
    {
      "epoch": 0.09278738635928009,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014654620000000002,
      "loss": 3.274,
      "step": 267270
    },
    {
      "epoch": 0.09280602959543828,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014654419999999999,
      "loss": 3.1261,
      "step": 267280
    },
    {
      "epoch": 0.09282467283159647,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001465422,
      "loss": 3.1381,
      "step": 267290
    },
    {
      "epoch": 0.09284331606775464,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001465402,
      "loss": 3.1424,
      "step": 267300
    },
    {
      "epoch": 0.09286195930391283,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001465382,
      "loss": 3.0878,
      "step": 267310
    },
    {
      "epoch": 0.09288060254007101,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014653620000000002,
      "loss": 3.0885,
      "step": 267320
    },
    {
      "epoch": 0.0928992457762292,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001465342,
      "loss": 3.1281,
      "step": 267330
    },
    {
      "epoch": 0.09291788901238739,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001465322,
      "loss": 3.1257,
      "step": 267340
    },
    {
      "epoch": 0.09293653224854556,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001465302,
      "loss": 3.1628,
      "step": 267350
    },
    {
      "epoch": 0.09295517548470375,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001465282,
      "loss": 3.2191,
      "step": 267360
    },
    {
      "epoch": 0.09297381872086194,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014652620000000003,
      "loss": 3.2022,
      "step": 267370
    },
    {
      "epoch": 0.09299246195702011,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001465242,
      "loss": 3.011,
      "step": 267380
    },
    {
      "epoch": 0.0930111051931783,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014652220000000002,
      "loss": 3.0984,
      "step": 267390
    },
    {
      "epoch": 0.09302974842933648,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001465202,
      "loss": 3.1966,
      "step": 267400
    },
    {
      "epoch": 0.09304839166549467,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001465182,
      "loss": 3.0785,
      "step": 267410
    },
    {
      "epoch": 0.09306703490165286,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001465162,
      "loss": 3.135,
      "step": 267420
    },
    {
      "epoch": 0.09308567813781103,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001465142,
      "loss": 3.1858,
      "step": 267430
    },
    {
      "epoch": 0.09310432137396922,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014651220000000002,
      "loss": 3.1546,
      "step": 267440
    },
    {
      "epoch": 0.0931229646101274,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001465102,
      "loss": 3.1561,
      "step": 267450
    },
    {
      "epoch": 0.09314160784628558,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014650820000000001,
      "loss": 3.2227,
      "step": 267460
    },
    {
      "epoch": 0.09316025108244377,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001465062,
      "loss": 3.1178,
      "step": 267470
    },
    {
      "epoch": 0.09317889431860195,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001465042,
      "loss": 2.9985,
      "step": 267480
    },
    {
      "epoch": 0.09319753755476014,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001465022,
      "loss": 3.1085,
      "step": 267490
    },
    {
      "epoch": 0.09321618079091833,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001465002,
      "loss": 3.1592,
      "step": 267500
    },
    {
      "epoch": 0.0932348240270765,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014649820000000002,
      "loss": 3.0709,
      "step": 267510
    },
    {
      "epoch": 0.09325346726323469,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014649620000000002,
      "loss": 3.1812,
      "step": 267520
    },
    {
      "epoch": 0.09327211049939287,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001464942,
      "loss": 3.0063,
      "step": 267530
    },
    {
      "epoch": 0.09329075373555105,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001464922,
      "loss": 3.0743,
      "step": 267540
    },
    {
      "epoch": 0.09330939697170924,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001464902,
      "loss": 3.2186,
      "step": 267550
    },
    {
      "epoch": 0.09332804020786742,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001464882,
      "loss": 3.0841,
      "step": 267560
    },
    {
      "epoch": 0.09334668344402561,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00014648620000000002,
      "loss": 3.1745,
      "step": 267570
    },
    {
      "epoch": 0.09336532668018378,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014648420000000002,
      "loss": 2.9785,
      "step": 267580
    },
    {
      "epoch": 0.09338396991634197,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001464822,
      "loss": 3.122,
      "step": 267590
    },
    {
      "epoch": 0.09340261315250016,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001464802,
      "loss": 3.1236,
      "step": 267600
    },
    {
      "epoch": 0.09342125638865834,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001464782,
      "loss": 3.0703,
      "step": 267610
    },
    {
      "epoch": 0.09343989962481652,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001464762,
      "loss": 3.1331,
      "step": 267620
    },
    {
      "epoch": 0.09345854286097471,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001464742,
      "loss": 3.0773,
      "step": 267630
    },
    {
      "epoch": 0.09347718609713289,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014647220000000002,
      "loss": 3.1331,
      "step": 267640
    },
    {
      "epoch": 0.09349582933329108,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001464702,
      "loss": 3.0551,
      "step": 267650
    },
    {
      "epoch": 0.09351447256944925,
      "grad_norm": 0.75,
      "learning_rate": 0.0001464682,
      "loss": 3.1636,
      "step": 267660
    },
    {
      "epoch": 0.09353311580560744,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001464662,
      "loss": 3.1974,
      "step": 267670
    },
    {
      "epoch": 0.09355175904176563,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001464642,
      "loss": 3.0264,
      "step": 267680
    },
    {
      "epoch": 0.0935704022779238,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014646220000000002,
      "loss": 3.1423,
      "step": 267690
    },
    {
      "epoch": 0.093589045514082,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001464602,
      "loss": 3.2219,
      "step": 267700
    },
    {
      "epoch": 0.09360768875024018,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001464582,
      "loss": 3.1684,
      "step": 267710
    },
    {
      "epoch": 0.09362633198639836,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001464562,
      "loss": 3.1939,
      "step": 267720
    },
    {
      "epoch": 0.09364497522255655,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001464542,
      "loss": 3.0759,
      "step": 267730
    },
    {
      "epoch": 0.09366361845871472,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00014645220000000003,
      "loss": 3.1079,
      "step": 267740
    },
    {
      "epoch": 0.09368226169487291,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001464502,
      "loss": 3.1261,
      "step": 267750
    },
    {
      "epoch": 0.0937009049310311,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014644820000000002,
      "loss": 3.2098,
      "step": 267760
    },
    {
      "epoch": 0.09371954816718928,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001464462,
      "loss": 3.067,
      "step": 267770
    },
    {
      "epoch": 0.09373819140334747,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001464442,
      "loss": 3.0814,
      "step": 267780
    },
    {
      "epoch": 0.09375683463950564,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001464422,
      "loss": 3.2308,
      "step": 267790
    },
    {
      "epoch": 0.09377547787566383,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001464402,
      "loss": 3.0408,
      "step": 267800
    },
    {
      "epoch": 0.09379412111182202,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014643820000000002,
      "loss": 3.2326,
      "step": 267810
    },
    {
      "epoch": 0.0938127643479802,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001464362,
      "loss": 3.1285,
      "step": 267820
    },
    {
      "epoch": 0.09383140758413838,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014643420000000002,
      "loss": 2.9845,
      "step": 267830
    },
    {
      "epoch": 0.09385005082029657,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001464322,
      "loss": 3.1127,
      "step": 267840
    },
    {
      "epoch": 0.09386869405645475,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001464302,
      "loss": 3.1648,
      "step": 267850
    },
    {
      "epoch": 0.09388733729261294,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001464282,
      "loss": 3.2185,
      "step": 267860
    },
    {
      "epoch": 0.09390598052877111,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001464262,
      "loss": 3.1838,
      "step": 267870
    },
    {
      "epoch": 0.0939246237649293,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014642420000000002,
      "loss": 3.2626,
      "step": 267880
    },
    {
      "epoch": 0.09394326700108749,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001464222,
      "loss": 2.9795,
      "step": 267890
    },
    {
      "epoch": 0.09396191023724566,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001464202,
      "loss": 2.9751,
      "step": 267900
    },
    {
      "epoch": 0.09398055347340385,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001464182,
      "loss": 3.0425,
      "step": 267910
    },
    {
      "epoch": 0.09399919670956203,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001464162,
      "loss": 3.1969,
      "step": 267920
    },
    {
      "epoch": 0.09401783994572022,
      "grad_norm": 0.75,
      "learning_rate": 0.00014641420000000003,
      "loss": 3.1575,
      "step": 267930
    },
    {
      "epoch": 0.0940364831818784,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001464122,
      "loss": 2.9174,
      "step": 267940
    },
    {
      "epoch": 0.09405512641803658,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014641020000000002,
      "loss": 3.1766,
      "step": 267950
    },
    {
      "epoch": 0.09407376965419477,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001464082,
      "loss": 3.1228,
      "step": 267960
    },
    {
      "epoch": 0.09409241289035296,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001464062,
      "loss": 3.1426,
      "step": 267970
    },
    {
      "epoch": 0.09411105612651113,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001464042,
      "loss": 3.2361,
      "step": 267980
    },
    {
      "epoch": 0.09412969936266932,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001464022,
      "loss": 3.0815,
      "step": 267990
    },
    {
      "epoch": 0.0941483425988275,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014640020000000002,
      "loss": 3.1216,
      "step": 268000
    },
    {
      "epoch": 0.09416698583498569,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001463982,
      "loss": 3.0702,
      "step": 268010
    },
    {
      "epoch": 0.09418562907114388,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001463962,
      "loss": 3.0268,
      "step": 268020
    },
    {
      "epoch": 0.09420427230730205,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001463942,
      "loss": 3.1284,
      "step": 268030
    },
    {
      "epoch": 0.09422291554346024,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001463922,
      "loss": 3.2019,
      "step": 268040
    },
    {
      "epoch": 0.09424155877961843,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001463902,
      "loss": 3.1105,
      "step": 268050
    },
    {
      "epoch": 0.0942602020157766,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001463882,
      "loss": 3.1343,
      "step": 268060
    },
    {
      "epoch": 0.09427884525193479,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014638620000000002,
      "loss": 3.1877,
      "step": 268070
    },
    {
      "epoch": 0.09429748848809297,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014638420000000001,
      "loss": 3.2262,
      "step": 268080
    },
    {
      "epoch": 0.09431613172425116,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001463822,
      "loss": 3.0733,
      "step": 268090
    },
    {
      "epoch": 0.09433477496040935,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001463802,
      "loss": 3.2255,
      "step": 268100
    },
    {
      "epoch": 0.09435341819656752,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001463782,
      "loss": 3.123,
      "step": 268110
    },
    {
      "epoch": 0.09437206143272571,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001463762,
      "loss": 3.059,
      "step": 268120
    },
    {
      "epoch": 0.09439070466888388,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014637420000000002,
      "loss": 3.1659,
      "step": 268130
    },
    {
      "epoch": 0.09440934790504207,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014637220000000001,
      "loss": 3.1105,
      "step": 268140
    },
    {
      "epoch": 0.09442799114120026,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001463702,
      "loss": 3.0891,
      "step": 268150
    },
    {
      "epoch": 0.09444663437735844,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001463682,
      "loss": 3.1224,
      "step": 268160
    },
    {
      "epoch": 0.09446527761351663,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001463662,
      "loss": 3.08,
      "step": 268170
    },
    {
      "epoch": 0.09448392084967482,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014636420000000002,
      "loss": 3.1856,
      "step": 268180
    },
    {
      "epoch": 0.09450256408583299,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001463622,
      "loss": 3.1092,
      "step": 268190
    },
    {
      "epoch": 0.09452120732199118,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014636020000000002,
      "loss": 3.0997,
      "step": 268200
    },
    {
      "epoch": 0.09453985055814935,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001463582,
      "loss": 3.0698,
      "step": 268210
    },
    {
      "epoch": 0.09455849379430754,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001463562,
      "loss": 3.1864,
      "step": 268220
    },
    {
      "epoch": 0.09457713703046573,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001463542,
      "loss": 3.1088,
      "step": 268230
    },
    {
      "epoch": 0.09459578026662391,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001463522,
      "loss": 3.069,
      "step": 268240
    },
    {
      "epoch": 0.0946144235027821,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014635020000000002,
      "loss": 3.0814,
      "step": 268250
    },
    {
      "epoch": 0.09463306673894029,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001463482,
      "loss": 3.1596,
      "step": 268260
    },
    {
      "epoch": 0.09465170997509846,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001463462,
      "loss": 3.2392,
      "step": 268270
    },
    {
      "epoch": 0.09467035321125665,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001463442,
      "loss": 3.1157,
      "step": 268280
    },
    {
      "epoch": 0.09468899644741482,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001463422,
      "loss": 3.2142,
      "step": 268290
    },
    {
      "epoch": 0.09470763968357301,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014634020000000003,
      "loss": 3.0555,
      "step": 268300
    },
    {
      "epoch": 0.0947262829197312,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001463382,
      "loss": 3.2753,
      "step": 268310
    },
    {
      "epoch": 0.09474492615588938,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014633620000000002,
      "loss": 2.874,
      "step": 268320
    },
    {
      "epoch": 0.09476356939204757,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001463342,
      "loss": 3.2416,
      "step": 268330
    },
    {
      "epoch": 0.09478221262820574,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001463322,
      "loss": 3.1091,
      "step": 268340
    },
    {
      "epoch": 0.09480085586436393,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001463302,
      "loss": 3.1985,
      "step": 268350
    },
    {
      "epoch": 0.09481949910052212,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001463282,
      "loss": 3.1925,
      "step": 268360
    },
    {
      "epoch": 0.0948381423366803,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014632620000000002,
      "loss": 3.1188,
      "step": 268370
    },
    {
      "epoch": 0.09485678557283848,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001463242,
      "loss": 3.1385,
      "step": 268380
    },
    {
      "epoch": 0.09487542880899667,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001463222,
      "loss": 3.1529,
      "step": 268390
    },
    {
      "epoch": 0.09489407204515485,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001463202,
      "loss": 3.0777,
      "step": 268400
    },
    {
      "epoch": 0.09491271528131304,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001463182,
      "loss": 3.0096,
      "step": 268410
    },
    {
      "epoch": 0.09493135851747121,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014631620000000003,
      "loss": 3.0471,
      "step": 268420
    },
    {
      "epoch": 0.0949500017536294,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001463142,
      "loss": 2.9639,
      "step": 268430
    },
    {
      "epoch": 0.09496864498978759,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014631220000000002,
      "loss": 3.1332,
      "step": 268440
    },
    {
      "epoch": 0.09498728822594577,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001463102,
      "loss": 3.1125,
      "step": 268450
    },
    {
      "epoch": 0.09500593146210395,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001463082,
      "loss": 3.162,
      "step": 268460
    },
    {
      "epoch": 0.09502457469826213,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001463062,
      "loss": 3.0501,
      "step": 268470
    },
    {
      "epoch": 0.09504321793442032,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001463042,
      "loss": 3.1332,
      "step": 268480
    },
    {
      "epoch": 0.09506186117057851,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014630220000000002,
      "loss": 3.143,
      "step": 268490
    },
    {
      "epoch": 0.09508050440673668,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001463002,
      "loss": 3.1753,
      "step": 268500
    },
    {
      "epoch": 0.09509914764289487,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014629820000000001,
      "loss": 3.1646,
      "step": 268510
    },
    {
      "epoch": 0.09511779087905306,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001462962,
      "loss": 3.1639,
      "step": 268520
    },
    {
      "epoch": 0.09513643411521124,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001462942,
      "loss": 3.102,
      "step": 268530
    },
    {
      "epoch": 0.09515507735136942,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001462922,
      "loss": 3.1461,
      "step": 268540
    },
    {
      "epoch": 0.0951737205875276,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001462902,
      "loss": 3.0998,
      "step": 268550
    },
    {
      "epoch": 0.09519236382368579,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014628820000000002,
      "loss": 3.1777,
      "step": 268560
    },
    {
      "epoch": 0.09521100705984398,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014628620000000002,
      "loss": 3.015,
      "step": 268570
    },
    {
      "epoch": 0.09522965029600215,
      "grad_norm": 0.75,
      "learning_rate": 0.0001462842,
      "loss": 3.0473,
      "step": 268580
    },
    {
      "epoch": 0.09524829353216034,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001462822,
      "loss": 2.9687,
      "step": 268590
    },
    {
      "epoch": 0.09526693676831853,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001462802,
      "loss": 3.2405,
      "step": 268600
    },
    {
      "epoch": 0.0952855800044767,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001462782,
      "loss": 3.0576,
      "step": 268610
    },
    {
      "epoch": 0.0953042232406349,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001462762,
      "loss": 3.1602,
      "step": 268620
    },
    {
      "epoch": 0.09532286647679307,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014627420000000002,
      "loss": 3.0354,
      "step": 268630
    },
    {
      "epoch": 0.09534150971295126,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001462722,
      "loss": 3.0334,
      "step": 268640
    },
    {
      "epoch": 0.09536015294910945,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001462702,
      "loss": 3.1719,
      "step": 268650
    },
    {
      "epoch": 0.09537879618526762,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001462682,
      "loss": 3.1239,
      "step": 268660
    },
    {
      "epoch": 0.09539743942142581,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001462662,
      "loss": 3.1193,
      "step": 268670
    },
    {
      "epoch": 0.09541608265758399,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001462642,
      "loss": 3.2311,
      "step": 268680
    },
    {
      "epoch": 0.09543472589374218,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014626220000000002,
      "loss": 3.0966,
      "step": 268690
    },
    {
      "epoch": 0.09545336912990036,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001462602,
      "loss": 3.1534,
      "step": 268700
    },
    {
      "epoch": 0.09547201236605854,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001462582,
      "loss": 3.1558,
      "step": 268710
    },
    {
      "epoch": 0.09549065560221673,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001462562,
      "loss": 3.0972,
      "step": 268720
    },
    {
      "epoch": 0.09550929883837492,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001462542,
      "loss": 3.2241,
      "step": 268730
    },
    {
      "epoch": 0.09552794207453309,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014625220000000002,
      "loss": 3.1149,
      "step": 268740
    },
    {
      "epoch": 0.09554658531069128,
      "grad_norm": 0.75,
      "learning_rate": 0.0001462502,
      "loss": 3.1814,
      "step": 268750
    },
    {
      "epoch": 0.09556522854684946,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014624820000000001,
      "loss": 3.1162,
      "step": 268760
    },
    {
      "epoch": 0.09558387178300765,
      "grad_norm": 0.75,
      "learning_rate": 0.0001462462,
      "loss": 3.2526,
      "step": 268770
    },
    {
      "epoch": 0.09560251501916583,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001462442,
      "loss": 3.1458,
      "step": 268780
    },
    {
      "epoch": 0.09562115825532401,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014624220000000003,
      "loss": 3.0257,
      "step": 268790
    },
    {
      "epoch": 0.0956398014914822,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001462402,
      "loss": 3.028,
      "step": 268800
    },
    {
      "epoch": 0.09565844472764037,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014623820000000002,
      "loss": 3.2077,
      "step": 268810
    },
    {
      "epoch": 0.09567708796379856,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014623620000000001,
      "loss": 3.0068,
      "step": 268820
    },
    {
      "epoch": 0.09569573119995675,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001462342,
      "loss": 3.1386,
      "step": 268830
    },
    {
      "epoch": 0.09571437443611493,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001462322,
      "loss": 3.1584,
      "step": 268840
    },
    {
      "epoch": 0.09573301767227312,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001462302,
      "loss": 3.0602,
      "step": 268850
    },
    {
      "epoch": 0.0957516609084313,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014622820000000002,
      "loss": 3.2965,
      "step": 268860
    },
    {
      "epoch": 0.09577030414458948,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001462262,
      "loss": 3.0361,
      "step": 268870
    },
    {
      "epoch": 0.09578894738074767,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014622420000000002,
      "loss": 3.2244,
      "step": 268880
    },
    {
      "epoch": 0.09580759061690584,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001462222,
      "loss": 3.1301,
      "step": 268890
    },
    {
      "epoch": 0.09582623385306403,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001462202,
      "loss": 3.0708,
      "step": 268900
    },
    {
      "epoch": 0.09584487708922222,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014621820000000003,
      "loss": 3.2426,
      "step": 268910
    },
    {
      "epoch": 0.0958635203253804,
      "grad_norm": 0.75,
      "learning_rate": 0.0001462162,
      "loss": 3.0355,
      "step": 268920
    },
    {
      "epoch": 0.09588216356153859,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014621420000000002,
      "loss": 3.14,
      "step": 268930
    },
    {
      "epoch": 0.09590080679769677,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001462122,
      "loss": 2.8334,
      "step": 268940
    },
    {
      "epoch": 0.09591945003385495,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001462102,
      "loss": 3.084,
      "step": 268950
    },
    {
      "epoch": 0.09593809327001314,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001462082,
      "loss": 3.1465,
      "step": 268960
    },
    {
      "epoch": 0.09595673650617131,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001462062,
      "loss": 3.1008,
      "step": 268970
    },
    {
      "epoch": 0.0959753797423295,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014620420000000003,
      "loss": 3.1627,
      "step": 268980
    },
    {
      "epoch": 0.09599402297848769,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001462022,
      "loss": 3.1393,
      "step": 268990
    },
    {
      "epoch": 0.09601266621464587,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014620020000000002,
      "loss": 3.0211,
      "step": 269000
    },
    {
      "epoch": 0.09603130945080406,
      "grad_norm": 0.859375,
      "learning_rate": 0.00014619819999999999,
      "loss": 3.1107,
      "step": 269010
    },
    {
      "epoch": 0.09604995268696223,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001461962,
      "loss": 3.0926,
      "step": 269020
    },
    {
      "epoch": 0.09606859592312042,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001461942,
      "loss": 3.0822,
      "step": 269030
    },
    {
      "epoch": 0.09608723915927861,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001461922,
      "loss": 3.1166,
      "step": 269040
    },
    {
      "epoch": 0.09610588239543678,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014619020000000002,
      "loss": 3.0769,
      "step": 269050
    },
    {
      "epoch": 0.09612452563159497,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001461882,
      "loss": 3.1326,
      "step": 269060
    },
    {
      "epoch": 0.09614316886775316,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001461862,
      "loss": 3.1902,
      "step": 269070
    },
    {
      "epoch": 0.09616181210391134,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001461842,
      "loss": 2.9744,
      "step": 269080
    },
    {
      "epoch": 0.09618045534006953,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001461822,
      "loss": 3.064,
      "step": 269090
    },
    {
      "epoch": 0.0961990985762277,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001461802,
      "loss": 3.2031,
      "step": 269100
    },
    {
      "epoch": 0.09621774181238589,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001461782,
      "loss": 3.0325,
      "step": 269110
    },
    {
      "epoch": 0.09623638504854408,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014617620000000002,
      "loss": 3.0548,
      "step": 269120
    },
    {
      "epoch": 0.09625502828470225,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014617420000000001,
      "loss": 3.1278,
      "step": 269130
    },
    {
      "epoch": 0.09627367152086044,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001461722,
      "loss": 3.1481,
      "step": 269140
    },
    {
      "epoch": 0.09629231475701862,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001461702,
      "loss": 3.0381,
      "step": 269150
    },
    {
      "epoch": 0.09631095799317681,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001461682,
      "loss": 3.2058,
      "step": 269160
    },
    {
      "epoch": 0.096329601229335,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001461662,
      "loss": 3.0417,
      "step": 269170
    },
    {
      "epoch": 0.09634824446549317,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014616420000000002,
      "loss": 3.1626,
      "step": 269180
    },
    {
      "epoch": 0.09636688770165136,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014616220000000001,
      "loss": 2.9692,
      "step": 269190
    },
    {
      "epoch": 0.09638553093780955,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001461602,
      "loss": 3.035,
      "step": 269200
    },
    {
      "epoch": 0.09640417417396772,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001461582,
      "loss": 3.3146,
      "step": 269210
    },
    {
      "epoch": 0.09642281741012591,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001461562,
      "loss": 3.1359,
      "step": 269220
    },
    {
      "epoch": 0.09644146064628409,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001461542,
      "loss": 3.0474,
      "step": 269230
    },
    {
      "epoch": 0.09646010388244228,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001461522,
      "loss": 3.2406,
      "step": 269240
    },
    {
      "epoch": 0.09647874711860047,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014615020000000002,
      "loss": 3.0953,
      "step": 269250
    },
    {
      "epoch": 0.09649739035475864,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001461482,
      "loss": 3.2711,
      "step": 269260
    },
    {
      "epoch": 0.09651603359091683,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001461462,
      "loss": 3.1124,
      "step": 269270
    },
    {
      "epoch": 0.09653467682707502,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001461442,
      "loss": 3.1184,
      "step": 269280
    },
    {
      "epoch": 0.0965533200632332,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001461422,
      "loss": 3.1587,
      "step": 269290
    },
    {
      "epoch": 0.09657196329939138,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014614020000000002,
      "loss": 3.093,
      "step": 269300
    },
    {
      "epoch": 0.09659060653554956,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001461382,
      "loss": 3.0981,
      "step": 269310
    },
    {
      "epoch": 0.09660924977170775,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001461362,
      "loss": 3.1427,
      "step": 269320
    },
    {
      "epoch": 0.09662789300786594,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001461342,
      "loss": 3.0866,
      "step": 269330
    },
    {
      "epoch": 0.09664653624402411,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001461322,
      "loss": 3.1714,
      "step": 269340
    },
    {
      "epoch": 0.0966651794801823,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014613020000000003,
      "loss": 3.0962,
      "step": 269350
    },
    {
      "epoch": 0.09668382271634048,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001461282,
      "loss": 3.16,
      "step": 269360
    },
    {
      "epoch": 0.09670246595249866,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014612620000000002,
      "loss": 3.1028,
      "step": 269370
    },
    {
      "epoch": 0.09672110918865685,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001461242,
      "loss": 3.1372,
      "step": 269380
    },
    {
      "epoch": 0.09673975242481503,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001461222,
      "loss": 3.2226,
      "step": 269390
    },
    {
      "epoch": 0.09675839566097322,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001461202,
      "loss": 3.1734,
      "step": 269400
    },
    {
      "epoch": 0.0967770388971314,
      "grad_norm": 0.75,
      "learning_rate": 0.0001461182,
      "loss": 3.0577,
      "step": 269410
    },
    {
      "epoch": 0.09679568213328958,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014611620000000002,
      "loss": 3.1042,
      "step": 269420
    },
    {
      "epoch": 0.09681432536944777,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001461142,
      "loss": 3.0829,
      "step": 269430
    },
    {
      "epoch": 0.09683296860560595,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014611220000000001,
      "loss": 3.1411,
      "step": 269440
    },
    {
      "epoch": 0.09685161184176413,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001461102,
      "loss": 3.0687,
      "step": 269450
    },
    {
      "epoch": 0.09687025507792232,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001461082,
      "loss": 3.1081,
      "step": 269460
    },
    {
      "epoch": 0.0968888983140805,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014610620000000003,
      "loss": 3.076,
      "step": 269470
    },
    {
      "epoch": 0.09690754155023869,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001461042,
      "loss": 3.0051,
      "step": 269480
    },
    {
      "epoch": 0.09692618478639686,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014610220000000002,
      "loss": 3.1847,
      "step": 269490
    },
    {
      "epoch": 0.09694482802255505,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001461002,
      "loss": 3.1175,
      "step": 269500
    },
    {
      "epoch": 0.09696347125871324,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001460982,
      "loss": 2.9936,
      "step": 269510
    },
    {
      "epoch": 0.09698211449487142,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001460962,
      "loss": 3.1458,
      "step": 269520
    },
    {
      "epoch": 0.0970007577310296,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001460942,
      "loss": 3.2838,
      "step": 269530
    },
    {
      "epoch": 0.0970194009671878,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014609220000000002,
      "loss": 3.1968,
      "step": 269540
    },
    {
      "epoch": 0.09703804420334597,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001460902,
      "loss": 3.1732,
      "step": 269550
    },
    {
      "epoch": 0.09705668743950416,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014608820000000002,
      "loss": 3.1836,
      "step": 269560
    },
    {
      "epoch": 0.09707533067566233,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001460862,
      "loss": 3.0076,
      "step": 269570
    },
    {
      "epoch": 0.09709397391182052,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001460842,
      "loss": 3.0792,
      "step": 269580
    },
    {
      "epoch": 0.09711261714797871,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001460822,
      "loss": 3.1102,
      "step": 269590
    },
    {
      "epoch": 0.09713126038413689,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001460802,
      "loss": 3.1801,
      "step": 269600
    },
    {
      "epoch": 0.09714990362029507,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014607820000000002,
      "loss": 3.1158,
      "step": 269610
    },
    {
      "epoch": 0.09716854685645326,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001460762,
      "loss": 3.089,
      "step": 269620
    },
    {
      "epoch": 0.09718719009261144,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001460742,
      "loss": 3.1604,
      "step": 269630
    },
    {
      "epoch": 0.09720583332876963,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001460722,
      "loss": 3.1297,
      "step": 269640
    },
    {
      "epoch": 0.0972244765649278,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001460702,
      "loss": 3.0866,
      "step": 269650
    },
    {
      "epoch": 0.09724311980108599,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001460682,
      "loss": 3.1102,
      "step": 269660
    },
    {
      "epoch": 0.09726176303724418,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001460662,
      "loss": 3.1549,
      "step": 269670
    },
    {
      "epoch": 0.09728040627340236,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014606420000000002,
      "loss": 3.0385,
      "step": 269680
    },
    {
      "epoch": 0.09729904950956055,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001460622,
      "loss": 3.1199,
      "step": 269690
    },
    {
      "epoch": 0.09731769274571872,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001460602,
      "loss": 3.1267,
      "step": 269700
    },
    {
      "epoch": 0.09733633598187691,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001460582,
      "loss": 3.035,
      "step": 269710
    },
    {
      "epoch": 0.0973549792180351,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001460562,
      "loss": 3.0014,
      "step": 269720
    },
    {
      "epoch": 0.09737362245419327,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001460542,
      "loss": 3.1163,
      "step": 269730
    },
    {
      "epoch": 0.09739226569035146,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014605220000000002,
      "loss": 3.1387,
      "step": 269740
    },
    {
      "epoch": 0.09741090892650965,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001460502,
      "loss": 3.0637,
      "step": 269750
    },
    {
      "epoch": 0.09742955216266783,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001460482,
      "loss": 3.1726,
      "step": 269760
    },
    {
      "epoch": 0.09744819539882602,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001460462,
      "loss": 3.2929,
      "step": 269770
    },
    {
      "epoch": 0.09746683863498419,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001460442,
      "loss": 3.0372,
      "step": 269780
    },
    {
      "epoch": 0.09748548187114238,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014604220000000002,
      "loss": 3.0873,
      "step": 269790
    },
    {
      "epoch": 0.09750412510730057,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001460402,
      "loss": 3.0224,
      "step": 269800
    },
    {
      "epoch": 0.09752276834345874,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014603820000000001,
      "loss": 3.1509,
      "step": 269810
    },
    {
      "epoch": 0.09754141157961693,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001460362,
      "loss": 3.054,
      "step": 269820
    },
    {
      "epoch": 0.09756005481577511,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001460342,
      "loss": 3.1381,
      "step": 269830
    },
    {
      "epoch": 0.0975786980519333,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001460322,
      "loss": 3.2059,
      "step": 269840
    },
    {
      "epoch": 0.09759734128809149,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001460302,
      "loss": 3.1557,
      "step": 269850
    },
    {
      "epoch": 0.09761598452424966,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014602820000000002,
      "loss": 3.1356,
      "step": 269860
    },
    {
      "epoch": 0.09763462776040785,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014602620000000001,
      "loss": 3.1223,
      "step": 269870
    },
    {
      "epoch": 0.09765327099656604,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001460242,
      "loss": 3.1579,
      "step": 269880
    },
    {
      "epoch": 0.09767191423272421,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001460222,
      "loss": 3.3022,
      "step": 269890
    },
    {
      "epoch": 0.0976905574688824,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001460202,
      "loss": 3.1312,
      "step": 269900
    },
    {
      "epoch": 0.09770920070504058,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014601820000000002,
      "loss": 3.0276,
      "step": 269910
    },
    {
      "epoch": 0.09772784394119877,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001460162,
      "loss": 3.1678,
      "step": 269920
    },
    {
      "epoch": 0.09774648717735696,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014601420000000002,
      "loss": 3.1013,
      "step": 269930
    },
    {
      "epoch": 0.09776513041351513,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001460122,
      "loss": 3.1375,
      "step": 269940
    },
    {
      "epoch": 0.09778377364967332,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001460102,
      "loss": 3.087,
      "step": 269950
    },
    {
      "epoch": 0.09780241688583151,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014600820000000003,
      "loss": 3.2094,
      "step": 269960
    },
    {
      "epoch": 0.09782106012198968,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001460062,
      "loss": 3.0959,
      "step": 269970
    },
    {
      "epoch": 0.09783970335814787,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014600420000000002,
      "loss": 3.0466,
      "step": 269980
    },
    {
      "epoch": 0.09785834659430605,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001460022,
      "loss": 3.1804,
      "step": 269990
    },
    {
      "epoch": 0.09787698983046424,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001460002,
      "loss": 3.1191,
      "step": 270000
    },
    {
      "epoch": 0.09789563306662243,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001459982,
      "loss": 3.1634,
      "step": 270010
    },
    {
      "epoch": 0.0979142763027806,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001459962,
      "loss": 3.0657,
      "step": 270020
    },
    {
      "epoch": 0.09793291953893879,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014599420000000003,
      "loss": 3.2281,
      "step": 270030
    },
    {
      "epoch": 0.09795156277509696,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001459922,
      "loss": 3.1391,
      "step": 270040
    },
    {
      "epoch": 0.09797020601125515,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014599020000000002,
      "loss": 3.2483,
      "step": 270050
    },
    {
      "epoch": 0.09798884924741334,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014598819999999999,
      "loss": 3.2674,
      "step": 270060
    },
    {
      "epoch": 0.09800749248357152,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001459862,
      "loss": 3.1437,
      "step": 270070
    },
    {
      "epoch": 0.0980261357197297,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001459842,
      "loss": 3.0324,
      "step": 270080
    },
    {
      "epoch": 0.0980447789558879,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001459822,
      "loss": 3.1456,
      "step": 270090
    },
    {
      "epoch": 0.09806342219204607,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014598020000000002,
      "loss": 3.1714,
      "step": 270100
    },
    {
      "epoch": 0.09808206542820426,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001459782,
      "loss": 3.1716,
      "step": 270110
    },
    {
      "epoch": 0.09810070866436243,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014597620000000001,
      "loss": 3.0751,
      "step": 270120
    },
    {
      "epoch": 0.09811935190052062,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001459742,
      "loss": 3.2304,
      "step": 270130
    },
    {
      "epoch": 0.09813799513667881,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001459722,
      "loss": 3.1107,
      "step": 270140
    },
    {
      "epoch": 0.09815663837283699,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001459702,
      "loss": 3.2113,
      "step": 270150
    },
    {
      "epoch": 0.09817528160899518,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001459682,
      "loss": 3.113,
      "step": 270160
    },
    {
      "epoch": 0.09819392484515335,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014596620000000002,
      "loss": 3.2643,
      "step": 270170
    },
    {
      "epoch": 0.09821256808131154,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014596420000000001,
      "loss": 3.1354,
      "step": 270180
    },
    {
      "epoch": 0.09823121131746973,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001459622,
      "loss": 3.1529,
      "step": 270190
    },
    {
      "epoch": 0.0982498545536279,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001459602,
      "loss": 3.1651,
      "step": 270200
    },
    {
      "epoch": 0.0982684977897861,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001459582,
      "loss": 3.3482,
      "step": 270210
    },
    {
      "epoch": 0.09828714102594428,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001459562,
      "loss": 3.2249,
      "step": 270220
    },
    {
      "epoch": 0.09830578426210246,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001459542,
      "loss": 3.1689,
      "step": 270230
    },
    {
      "epoch": 0.09832442749826065,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014595220000000002,
      "loss": 3.0688,
      "step": 270240
    },
    {
      "epoch": 0.09834307073441882,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001459502,
      "loss": 3.1258,
      "step": 270250
    },
    {
      "epoch": 0.09836171397057701,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001459482,
      "loss": 3.2374,
      "step": 270260
    },
    {
      "epoch": 0.0983803572067352,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001459462,
      "loss": 3.1607,
      "step": 270270
    },
    {
      "epoch": 0.09839900044289338,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001459442,
      "loss": 2.9103,
      "step": 270280
    },
    {
      "epoch": 0.09841764367905156,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001459422,
      "loss": 3.1206,
      "step": 270290
    },
    {
      "epoch": 0.09843628691520975,
      "grad_norm": 0.625,
      "learning_rate": 0.00014594020000000002,
      "loss": 3.1721,
      "step": 270300
    },
    {
      "epoch": 0.09845493015136793,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001459382,
      "loss": 3.0886,
      "step": 270310
    },
    {
      "epoch": 0.09847357338752612,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001459362,
      "loss": 3.0175,
      "step": 270320
    },
    {
      "epoch": 0.09849221662368429,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001459342,
      "loss": 3.218,
      "step": 270330
    },
    {
      "epoch": 0.09851085985984248,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001459322,
      "loss": 3.0884,
      "step": 270340
    },
    {
      "epoch": 0.09852950309600067,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014593020000000002,
      "loss": 3.1539,
      "step": 270350
    },
    {
      "epoch": 0.09854814633215885,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014592820000000002,
      "loss": 3.1431,
      "step": 270360
    },
    {
      "epoch": 0.09856678956831703,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001459262,
      "loss": 3.1635,
      "step": 270370
    },
    {
      "epoch": 0.09858543280447521,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001459242,
      "loss": 3.173,
      "step": 270380
    },
    {
      "epoch": 0.0986040760406334,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001459222,
      "loss": 3.0754,
      "step": 270390
    },
    {
      "epoch": 0.09862271927679159,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014592020000000003,
      "loss": 3.1886,
      "step": 270400
    },
    {
      "epoch": 0.09864136251294976,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001459182,
      "loss": 3.1874,
      "step": 270410
    },
    {
      "epoch": 0.09866000574910795,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014591620000000002,
      "loss": 3.2537,
      "step": 270420
    },
    {
      "epoch": 0.09867864898526614,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001459142,
      "loss": 3.169,
      "step": 270430
    },
    {
      "epoch": 0.09869729222142432,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001459122,
      "loss": 3.2576,
      "step": 270440
    },
    {
      "epoch": 0.0987159354575825,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001459102,
      "loss": 3.1839,
      "step": 270450
    },
    {
      "epoch": 0.09873457869374068,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001459082,
      "loss": 3.1444,
      "step": 270460
    },
    {
      "epoch": 0.09875322192989887,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014590620000000002,
      "loss": 3.1939,
      "step": 270470
    },
    {
      "epoch": 0.09877186516605706,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001459042,
      "loss": 3.1404,
      "step": 270480
    },
    {
      "epoch": 0.09879050840221523,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014590220000000001,
      "loss": 3.1176,
      "step": 270490
    },
    {
      "epoch": 0.09880915163837342,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001459002,
      "loss": 3.1474,
      "step": 270500
    },
    {
      "epoch": 0.0988277948745316,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001458982,
      "loss": 3.104,
      "step": 270510
    },
    {
      "epoch": 0.09884643811068979,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014589620000000003,
      "loss": 3.2527,
      "step": 270520
    },
    {
      "epoch": 0.09886508134684797,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001458942,
      "loss": 3.2457,
      "step": 270530
    },
    {
      "epoch": 0.09888372458300615,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014589220000000002,
      "loss": 3.1128,
      "step": 270540
    },
    {
      "epoch": 0.09890236781916434,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001458902,
      "loss": 3.1887,
      "step": 270550
    },
    {
      "epoch": 0.09892101105532253,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001458882,
      "loss": 3.1201,
      "step": 270560
    },
    {
      "epoch": 0.0989396542914807,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001458862,
      "loss": 3.0763,
      "step": 270570
    },
    {
      "epoch": 0.09895829752763889,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001458842,
      "loss": 3.1447,
      "step": 270580
    },
    {
      "epoch": 0.09897694076379707,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014588220000000002,
      "loss": 3.2116,
      "step": 270590
    },
    {
      "epoch": 0.09899558399995526,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001458802,
      "loss": 3.1245,
      "step": 270600
    },
    {
      "epoch": 0.09901422723611344,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014587820000000002,
      "loss": 3.022,
      "step": 270610
    },
    {
      "epoch": 0.09903287047227162,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014587619999999998,
      "loss": 3.2016,
      "step": 270620
    },
    {
      "epoch": 0.09905151370842981,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001458742,
      "loss": 3.2001,
      "step": 270630
    },
    {
      "epoch": 0.099070156944588,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001458722,
      "loss": 3.1102,
      "step": 270640
    },
    {
      "epoch": 0.09908880018074617,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001458702,
      "loss": 3.1279,
      "step": 270650
    },
    {
      "epoch": 0.09910744341690436,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014586820000000002,
      "loss": 3.2281,
      "step": 270660
    },
    {
      "epoch": 0.09912608665306254,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001458662,
      "loss": 3.1779,
      "step": 270670
    },
    {
      "epoch": 0.09914472988922073,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001458642,
      "loss": 3.03,
      "step": 270680
    },
    {
      "epoch": 0.09916337312537891,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001458622,
      "loss": 3.1816,
      "step": 270690
    },
    {
      "epoch": 0.09918201636153709,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001458602,
      "loss": 3.1622,
      "step": 270700
    },
    {
      "epoch": 0.09920065959769528,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001458582,
      "loss": 3.1932,
      "step": 270710
    },
    {
      "epoch": 0.09921930283385345,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001458562,
      "loss": 3.098,
      "step": 270720
    },
    {
      "epoch": 0.09923794607001164,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00014585420000000002,
      "loss": 3.1783,
      "step": 270730
    },
    {
      "epoch": 0.09925658930616983,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001458522,
      "loss": 3.2448,
      "step": 270740
    },
    {
      "epoch": 0.099275232542328,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001458502,
      "loss": 3.0879,
      "step": 270750
    },
    {
      "epoch": 0.0992938757784862,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001458482,
      "loss": 3.0117,
      "step": 270760
    },
    {
      "epoch": 0.09931251901464438,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001458462,
      "loss": 3.1405,
      "step": 270770
    },
    {
      "epoch": 0.09933116225080256,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001458442,
      "loss": 3.0492,
      "step": 270780
    },
    {
      "epoch": 0.09934980548696075,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014584220000000002,
      "loss": 3.2633,
      "step": 270790
    },
    {
      "epoch": 0.09936844872311892,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014584020000000001,
      "loss": 2.9778,
      "step": 270800
    },
    {
      "epoch": 0.09938709195927711,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001458382,
      "loss": 3.1113,
      "step": 270810
    },
    {
      "epoch": 0.0994057351954353,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001458362,
      "loss": 3.2296,
      "step": 270820
    },
    {
      "epoch": 0.09942437843159348,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001458342,
      "loss": 3.1242,
      "step": 270830
    },
    {
      "epoch": 0.09944302166775167,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001458322,
      "loss": 3.0672,
      "step": 270840
    },
    {
      "epoch": 0.09946166490390984,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014583020000000002,
      "loss": 3.0208,
      "step": 270850
    },
    {
      "epoch": 0.09948030814006803,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014582820000000001,
      "loss": 3.2291,
      "step": 270860
    },
    {
      "epoch": 0.09949895137622622,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001458262,
      "loss": 3.1508,
      "step": 270870
    },
    {
      "epoch": 0.0995175946123844,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001458242,
      "loss": 3.0096,
      "step": 270880
    },
    {
      "epoch": 0.09953623784854258,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001458222,
      "loss": 3.0442,
      "step": 270890
    },
    {
      "epoch": 0.09955488108470077,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001458202,
      "loss": 3.0797,
      "step": 270900
    },
    {
      "epoch": 0.09957352432085895,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014581820000000002,
      "loss": 3.2725,
      "step": 270910
    },
    {
      "epoch": 0.09959216755701714,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014581620000000002,
      "loss": 3.1557,
      "step": 270920
    },
    {
      "epoch": 0.09961081079317531,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001458142,
      "loss": 3.0941,
      "step": 270930
    },
    {
      "epoch": 0.0996294540293335,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001458122,
      "loss": 3.0928,
      "step": 270940
    },
    {
      "epoch": 0.09964809726549169,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001458102,
      "loss": 3.0098,
      "step": 270950
    },
    {
      "epoch": 0.09966674050164986,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014580820000000002,
      "loss": 3.0476,
      "step": 270960
    },
    {
      "epoch": 0.09968538373780805,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001458062,
      "loss": 3.1467,
      "step": 270970
    },
    {
      "epoch": 0.09970402697396624,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014580420000000002,
      "loss": 3.1135,
      "step": 270980
    },
    {
      "epoch": 0.09972267021012442,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001458022,
      "loss": 3.1706,
      "step": 270990
    },
    {
      "epoch": 0.0997413134462826,
      "grad_norm": 0.875,
      "learning_rate": 0.0001458002,
      "loss": 3.1012,
      "step": 271000
    },
    {
      "epoch": 0.09975995668244078,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001457982,
      "loss": 3.173,
      "step": 271010
    },
    {
      "epoch": 0.09977859991859897,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001457962,
      "loss": 3.2527,
      "step": 271020
    },
    {
      "epoch": 0.09979724315475716,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014579420000000002,
      "loss": 3.063,
      "step": 271030
    },
    {
      "epoch": 0.09981588639091533,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001457922,
      "loss": 3.2232,
      "step": 271040
    },
    {
      "epoch": 0.09983452962707352,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001457902,
      "loss": 3.1854,
      "step": 271050
    },
    {
      "epoch": 0.0998531728632317,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001457882,
      "loss": 3.1069,
      "step": 271060
    },
    {
      "epoch": 0.09987181609938989,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001457862,
      "loss": 3.2454,
      "step": 271070
    },
    {
      "epoch": 0.09989045933554808,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014578420000000003,
      "loss": 3.0941,
      "step": 271080
    },
    {
      "epoch": 0.09990910257170625,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001457822,
      "loss": 3.1764,
      "step": 271090
    },
    {
      "epoch": 0.09992774580786444,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014578020000000002,
      "loss": 3.1752,
      "step": 271100
    },
    {
      "epoch": 0.09994638904402263,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014577819999999999,
      "loss": 3.2,
      "step": 271110
    },
    {
      "epoch": 0.0999650322801808,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001457762,
      "loss": 3.1791,
      "step": 271120
    },
    {
      "epoch": 0.099983675516339,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001457742,
      "loss": 3.2725,
      "step": 271130
    },
    {
      "epoch": 0.10000231875249717,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001457722,
      "loss": 3.1533,
      "step": 271140
    },
    {
      "epoch": 0.10002096198865536,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014577020000000002,
      "loss": 3.0615,
      "step": 271150
    },
    {
      "epoch": 0.10003960522481355,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001457682,
      "loss": 3.0379,
      "step": 271160
    },
    {
      "epoch": 0.10005824846097172,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014576620000000001,
      "loss": 3.1677,
      "step": 271170
    },
    {
      "epoch": 0.10007689169712991,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001457642,
      "loss": 3.093,
      "step": 271180
    },
    {
      "epoch": 0.10009553493328809,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001457622,
      "loss": 3.1021,
      "step": 271190
    },
    {
      "epoch": 0.10011417816944627,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001457602,
      "loss": 3.19,
      "step": 271200
    },
    {
      "epoch": 0.10013282140560446,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001457582,
      "loss": 3.144,
      "step": 271210
    },
    {
      "epoch": 0.10015146464176264,
      "grad_norm": 0.859375,
      "learning_rate": 0.00014575620000000002,
      "loss": 3.0236,
      "step": 271220
    },
    {
      "epoch": 0.10017010787792083,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001457542,
      "loss": 3.1771,
      "step": 271230
    },
    {
      "epoch": 0.10018875111407902,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001457522,
      "loss": 3.0612,
      "step": 271240
    },
    {
      "epoch": 0.10020739435023719,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001457502,
      "loss": 3.1058,
      "step": 271250
    },
    {
      "epoch": 0.10022603758639538,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001457482,
      "loss": 3.3,
      "step": 271260
    },
    {
      "epoch": 0.10024468082255356,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001457462,
      "loss": 3.1947,
      "step": 271270
    },
    {
      "epoch": 0.10026332405871174,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001457442,
      "loss": 3.1675,
      "step": 271280
    },
    {
      "epoch": 0.10028196729486993,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014574220000000002,
      "loss": 3.2171,
      "step": 271290
    },
    {
      "epoch": 0.10030061053102811,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001457402,
      "loss": 3.1638,
      "step": 271300
    },
    {
      "epoch": 0.1003192537671863,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001457382,
      "loss": 3.0619,
      "step": 271310
    },
    {
      "epoch": 0.10033789700334449,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001457362,
      "loss": 3.0171,
      "step": 271320
    },
    {
      "epoch": 0.10035654023950266,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001457342,
      "loss": 3.1106,
      "step": 271330
    },
    {
      "epoch": 0.10037518347566085,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014573220000000002,
      "loss": 3.1952,
      "step": 271340
    },
    {
      "epoch": 0.10039382671181903,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014573020000000002,
      "loss": 3.1817,
      "step": 271350
    },
    {
      "epoch": 0.10041246994797721,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001457282,
      "loss": 3.1061,
      "step": 271360
    },
    {
      "epoch": 0.1004311131841354,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001457262,
      "loss": 3.2139,
      "step": 271370
    },
    {
      "epoch": 0.10044975642029358,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001457242,
      "loss": 3.1031,
      "step": 271380
    },
    {
      "epoch": 0.10046839965645177,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001457222,
      "loss": 3.1576,
      "step": 271390
    },
    {
      "epoch": 0.10048704289260994,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014572020000000002,
      "loss": 3.2152,
      "step": 271400
    },
    {
      "epoch": 0.10050568612876813,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014571820000000002,
      "loss": 3.2283,
      "step": 271410
    },
    {
      "epoch": 0.10052432936492632,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001457162,
      "loss": 3.2269,
      "step": 271420
    },
    {
      "epoch": 0.1005429726010845,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001457142,
      "loss": 3.0807,
      "step": 271430
    },
    {
      "epoch": 0.10056161583724268,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001457122,
      "loss": 3.1479,
      "step": 271440
    },
    {
      "epoch": 0.10058025907340087,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001457102,
      "loss": 3.1702,
      "step": 271450
    },
    {
      "epoch": 0.10059890230955905,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001457082,
      "loss": 3.1065,
      "step": 271460
    },
    {
      "epoch": 0.10061754554571724,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014570620000000002,
      "loss": 3.2243,
      "step": 271470
    },
    {
      "epoch": 0.10063618878187541,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001457042,
      "loss": 2.9892,
      "step": 271480
    },
    {
      "epoch": 0.1006548320180336,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001457022,
      "loss": 3.1118,
      "step": 271490
    },
    {
      "epoch": 0.10067347525419179,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001457002,
      "loss": 3.0944,
      "step": 271500
    },
    {
      "epoch": 0.10069211849034997,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001456982,
      "loss": 3.1397,
      "step": 271510
    },
    {
      "epoch": 0.10071076172650816,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014569620000000002,
      "loss": 2.9865,
      "step": 271520
    },
    {
      "epoch": 0.10072940496266633,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001456942,
      "loss": 3.1336,
      "step": 271530
    },
    {
      "epoch": 0.10074804819882452,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014569220000000001,
      "loss": 3.0477,
      "step": 271540
    },
    {
      "epoch": 0.10076669143498271,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001456902,
      "loss": 3.0587,
      "step": 271550
    },
    {
      "epoch": 0.10078533467114088,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001456882,
      "loss": 3.2143,
      "step": 271560
    },
    {
      "epoch": 0.10080397790729907,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014568620000000003,
      "loss": 3.1152,
      "step": 271570
    },
    {
      "epoch": 0.10082262114345726,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001456842,
      "loss": 3.1519,
      "step": 271580
    },
    {
      "epoch": 0.10084126437961544,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014568220000000002,
      "loss": 3.0839,
      "step": 271590
    },
    {
      "epoch": 0.10085990761577363,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001456802,
      "loss": 3.2374,
      "step": 271600
    },
    {
      "epoch": 0.1008785508519318,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001456782,
      "loss": 3.1321,
      "step": 271610
    },
    {
      "epoch": 0.10089719408808999,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001456762,
      "loss": 3.0475,
      "step": 271620
    },
    {
      "epoch": 0.10091583732424818,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001456742,
      "loss": 3.1349,
      "step": 271630
    },
    {
      "epoch": 0.10093448056040635,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014567220000000002,
      "loss": 3.1223,
      "step": 271640
    },
    {
      "epoch": 0.10095312379656454,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001456702,
      "loss": 3.2262,
      "step": 271650
    },
    {
      "epoch": 0.10097176703272273,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014566820000000002,
      "loss": 3.108,
      "step": 271660
    },
    {
      "epoch": 0.1009904102688809,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014566619999999998,
      "loss": 3.1819,
      "step": 271670
    },
    {
      "epoch": 0.1010090535050391,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001456642,
      "loss": 3.0674,
      "step": 271680
    },
    {
      "epoch": 0.10102769674119727,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001456622,
      "loss": 3.0185,
      "step": 271690
    },
    {
      "epoch": 0.10104633997735546,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001456602,
      "loss": 3.0799,
      "step": 271700
    },
    {
      "epoch": 0.10106498321351365,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014565820000000002,
      "loss": 3.1687,
      "step": 271710
    },
    {
      "epoch": 0.10108362644967182,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001456562,
      "loss": 3.082,
      "step": 271720
    },
    {
      "epoch": 0.10110226968583001,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001456542,
      "loss": 3.213,
      "step": 271730
    },
    {
      "epoch": 0.10112091292198819,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001456522,
      "loss": 3.1601,
      "step": 271740
    },
    {
      "epoch": 0.10113955615814638,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001456502,
      "loss": 3.0839,
      "step": 271750
    },
    {
      "epoch": 0.10115819939430457,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001456482,
      "loss": 3.0683,
      "step": 271760
    },
    {
      "epoch": 0.10117684263046274,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001456462,
      "loss": 3.1378,
      "step": 271770
    },
    {
      "epoch": 0.10119548586662093,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014564420000000002,
      "loss": 3.2747,
      "step": 271780
    },
    {
      "epoch": 0.10121412910277912,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001456422,
      "loss": 3.2625,
      "step": 271790
    },
    {
      "epoch": 0.1012327723389373,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001456402,
      "loss": 3.0199,
      "step": 271800
    },
    {
      "epoch": 0.10125141557509548,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001456382,
      "loss": 3.0633,
      "step": 271810
    },
    {
      "epoch": 0.10127005881125366,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001456362,
      "loss": 3.0836,
      "step": 271820
    },
    {
      "epoch": 0.10128870204741185,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014563420000000002,
      "loss": 3.1934,
      "step": 271830
    },
    {
      "epoch": 0.10130734528357004,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001456322,
      "loss": 3.0674,
      "step": 271840
    },
    {
      "epoch": 0.10132598851972821,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014563020000000001,
      "loss": 3.2643,
      "step": 271850
    },
    {
      "epoch": 0.1013446317558864,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001456282,
      "loss": 3.1359,
      "step": 271860
    },
    {
      "epoch": 0.10136327499204457,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001456262,
      "loss": 3.0303,
      "step": 271870
    },
    {
      "epoch": 0.10138191822820276,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001456242,
      "loss": 3.1956,
      "step": 271880
    },
    {
      "epoch": 0.10140056146436095,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001456222,
      "loss": 3.154,
      "step": 271890
    },
    {
      "epoch": 0.10141920470051913,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014562020000000002,
      "loss": 3.1076,
      "step": 271900
    },
    {
      "epoch": 0.10143784793667732,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014561820000000001,
      "loss": 3.1678,
      "step": 271910
    },
    {
      "epoch": 0.1014564911728355,
      "grad_norm": 0.625,
      "learning_rate": 0.0001456162,
      "loss": 3.0379,
      "step": 271920
    },
    {
      "epoch": 0.10147513440899368,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001456142,
      "loss": 3.1262,
      "step": 271930
    },
    {
      "epoch": 0.10149377764515187,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001456122,
      "loss": 3.0694,
      "step": 271940
    },
    {
      "epoch": 0.10151242088131004,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001456102,
      "loss": 3.0814,
      "step": 271950
    },
    {
      "epoch": 0.10153106411746823,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014560820000000002,
      "loss": 3.1193,
      "step": 271960
    },
    {
      "epoch": 0.10154970735362642,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014560620000000002,
      "loss": 3.0639,
      "step": 271970
    },
    {
      "epoch": 0.1015683505897846,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001456042,
      "loss": 3.1576,
      "step": 271980
    },
    {
      "epoch": 0.10158699382594279,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001456022,
      "loss": 3.1126,
      "step": 271990
    },
    {
      "epoch": 0.10160563706210098,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001456002,
      "loss": 3.2375,
      "step": 272000
    },
    {
      "epoch": 0.10162428029825915,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014559820000000003,
      "loss": 3.0711,
      "step": 272010
    },
    {
      "epoch": 0.10164292353441734,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001455962,
      "loss": 3.1171,
      "step": 272020
    },
    {
      "epoch": 0.10166156677057551,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014559420000000002,
      "loss": 3.1929,
      "step": 272030
    },
    {
      "epoch": 0.1016802100067337,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001455922,
      "loss": 3.1387,
      "step": 272040
    },
    {
      "epoch": 0.10169885324289189,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001455902,
      "loss": 3.1093,
      "step": 272050
    },
    {
      "epoch": 0.10171749647905007,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001455882,
      "loss": 3.191,
      "step": 272060
    },
    {
      "epoch": 0.10173613971520826,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001455862,
      "loss": 3.1649,
      "step": 272070
    },
    {
      "epoch": 0.10175478295136643,
      "grad_norm": 0.84375,
      "learning_rate": 0.00014558420000000002,
      "loss": 3.2015,
      "step": 272080
    },
    {
      "epoch": 0.10177342618752462,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001455822,
      "loss": 3.0161,
      "step": 272090
    },
    {
      "epoch": 0.10179206942368281,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001455802,
      "loss": 3.2102,
      "step": 272100
    },
    {
      "epoch": 0.10181071265984099,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001455782,
      "loss": 3.032,
      "step": 272110
    },
    {
      "epoch": 0.10182935589599917,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001455762,
      "loss": 3.0662,
      "step": 272120
    },
    {
      "epoch": 0.10184799913215736,
      "grad_norm": 0.75,
      "learning_rate": 0.00014557420000000003,
      "loss": 3.1588,
      "step": 272130
    },
    {
      "epoch": 0.10186664236831554,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001455722,
      "loss": 3.272,
      "step": 272140
    },
    {
      "epoch": 0.10188528560447373,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014557020000000002,
      "loss": 3.1776,
      "step": 272150
    },
    {
      "epoch": 0.1019039288406319,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00014556819999999999,
      "loss": 3.2995,
      "step": 272160
    },
    {
      "epoch": 0.10192257207679009,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001455662,
      "loss": 3.1742,
      "step": 272170
    },
    {
      "epoch": 0.10194121531294828,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001455642,
      "loss": 3.0771,
      "step": 272180
    },
    {
      "epoch": 0.10195985854910646,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001455622,
      "loss": 3.2035,
      "step": 272190
    },
    {
      "epoch": 0.10197850178526464,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014556020000000002,
      "loss": 3.1103,
      "step": 272200
    },
    {
      "epoch": 0.10199714502142282,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001455582,
      "loss": 3.142,
      "step": 272210
    },
    {
      "epoch": 0.10201578825758101,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014555620000000001,
      "loss": 3.2647,
      "step": 272220
    },
    {
      "epoch": 0.1020344314937392,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014555419999999998,
      "loss": 3.2682,
      "step": 272230
    },
    {
      "epoch": 0.10205307472989737,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001455522,
      "loss": 3.096,
      "step": 272240
    },
    {
      "epoch": 0.10207171796605556,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001455502,
      "loss": 3.3052,
      "step": 272250
    },
    {
      "epoch": 0.10209036120221375,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001455482,
      "loss": 3.1335,
      "step": 272260
    },
    {
      "epoch": 0.10210900443837193,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014554620000000002,
      "loss": 3.1954,
      "step": 272270
    },
    {
      "epoch": 0.10212764767453011,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001455442,
      "loss": 3.2505,
      "step": 272280
    },
    {
      "epoch": 0.10214629091068829,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001455422,
      "loss": 3.1468,
      "step": 272290
    },
    {
      "epoch": 0.10216493414684648,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001455402,
      "loss": 3.1841,
      "step": 272300
    },
    {
      "epoch": 0.10218357738300467,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001455382,
      "loss": 3.0327,
      "step": 272310
    },
    {
      "epoch": 0.10220222061916284,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014553620000000002,
      "loss": 3.0714,
      "step": 272320
    },
    {
      "epoch": 0.10222086385532103,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001455342,
      "loss": 3.1228,
      "step": 272330
    },
    {
      "epoch": 0.10223950709147922,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014553220000000002,
      "loss": 3.1959,
      "step": 272340
    },
    {
      "epoch": 0.1022581503276374,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001455302,
      "loss": 3.1302,
      "step": 272350
    },
    {
      "epoch": 0.10227679356379558,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001455282,
      "loss": 3.1408,
      "step": 272360
    },
    {
      "epoch": 0.10229543679995376,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001455262,
      "loss": 3.1221,
      "step": 272370
    },
    {
      "epoch": 0.10231408003611195,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001455242,
      "loss": 3.0965,
      "step": 272380
    },
    {
      "epoch": 0.10233272327227014,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014552220000000002,
      "loss": 3.213,
      "step": 272390
    },
    {
      "epoch": 0.10235136650842831,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014552020000000002,
      "loss": 3.2143,
      "step": 272400
    },
    {
      "epoch": 0.1023700097445865,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001455182,
      "loss": 3.1847,
      "step": 272410
    },
    {
      "epoch": 0.10238865298074468,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001455162,
      "loss": 3.1365,
      "step": 272420
    },
    {
      "epoch": 0.10240729621690287,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001455142,
      "loss": 3.165,
      "step": 272430
    },
    {
      "epoch": 0.10242593945306105,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001455122,
      "loss": 3.1159,
      "step": 272440
    },
    {
      "epoch": 0.10244458268921923,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001455102,
      "loss": 2.9966,
      "step": 272450
    },
    {
      "epoch": 0.10246322592537742,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014550820000000002,
      "loss": 3.1579,
      "step": 272460
    },
    {
      "epoch": 0.10248186916153561,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001455062,
      "loss": 3.0123,
      "step": 272470
    },
    {
      "epoch": 0.10250051239769378,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001455042,
      "loss": 2.8863,
      "step": 272480
    },
    {
      "epoch": 0.10251915563385197,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001455022,
      "loss": 3.2533,
      "step": 272490
    },
    {
      "epoch": 0.10253779887001015,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001455002,
      "loss": 3.1365,
      "step": 272500
    },
    {
      "epoch": 0.10255644210616834,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001454982,
      "loss": 3.039,
      "step": 272510
    },
    {
      "epoch": 0.10257508534232652,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014549620000000002,
      "loss": 3.0479,
      "step": 272520
    },
    {
      "epoch": 0.1025937285784847,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014549420000000001,
      "loss": 3.1347,
      "step": 272530
    },
    {
      "epoch": 0.10261237181464289,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001454922,
      "loss": 3.1472,
      "step": 272540
    },
    {
      "epoch": 0.10263101505080106,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001454902,
      "loss": 3.1698,
      "step": 272550
    },
    {
      "epoch": 0.10264965828695925,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001454882,
      "loss": 3.1452,
      "step": 272560
    },
    {
      "epoch": 0.10266830152311744,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014548620000000002,
      "loss": 3.1485,
      "step": 272570
    },
    {
      "epoch": 0.10268694475927562,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001454842,
      "loss": 3.018,
      "step": 272580
    },
    {
      "epoch": 0.1027055879954338,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014548220000000001,
      "loss": 3.1269,
      "step": 272590
    },
    {
      "epoch": 0.102724231231592,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001454802,
      "loss": 3.2456,
      "step": 272600
    },
    {
      "epoch": 0.10274287446775017,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001454782,
      "loss": 3.2358,
      "step": 272610
    },
    {
      "epoch": 0.10276151770390836,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014547620000000003,
      "loss": 3.1056,
      "step": 272620
    },
    {
      "epoch": 0.10278016094006653,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001454742,
      "loss": 3.0891,
      "step": 272630
    },
    {
      "epoch": 0.10279880417622472,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014547220000000002,
      "loss": 3.0577,
      "step": 272640
    },
    {
      "epoch": 0.10281744741238291,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001454702,
      "loss": 3.0977,
      "step": 272650
    },
    {
      "epoch": 0.10283609064854109,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001454682,
      "loss": 3.0557,
      "step": 272660
    },
    {
      "epoch": 0.10285473388469928,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001454662,
      "loss": 3.1209,
      "step": 272670
    },
    {
      "epoch": 0.10287337712085746,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001454642,
      "loss": 3.0178,
      "step": 272680
    },
    {
      "epoch": 0.10289202035701564,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014546220000000002,
      "loss": 3.1016,
      "step": 272690
    },
    {
      "epoch": 0.10291066359317383,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001454602,
      "loss": 3.2047,
      "step": 272700
    },
    {
      "epoch": 0.102929306829332,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014545820000000002,
      "loss": 3.0019,
      "step": 272710
    },
    {
      "epoch": 0.10294795006549019,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014545619999999998,
      "loss": 3.1669,
      "step": 272720
    },
    {
      "epoch": 0.10296659330164838,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001454542,
      "loss": 3.1832,
      "step": 272730
    },
    {
      "epoch": 0.10298523653780656,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001454522,
      "loss": 2.994,
      "step": 272740
    },
    {
      "epoch": 0.10300387977396475,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001454502,
      "loss": 3.2329,
      "step": 272750
    },
    {
      "epoch": 0.10302252301012292,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014544820000000002,
      "loss": 3.1571,
      "step": 272760
    },
    {
      "epoch": 0.10304116624628111,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001454462,
      "loss": 3.2735,
      "step": 272770
    },
    {
      "epoch": 0.1030598094824393,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001454442,
      "loss": 3.0705,
      "step": 272780
    },
    {
      "epoch": 0.10307845271859747,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001454422,
      "loss": 3.1754,
      "step": 272790
    },
    {
      "epoch": 0.10309709595475566,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001454402,
      "loss": 3.1198,
      "step": 272800
    },
    {
      "epoch": 0.10311573919091385,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014543820000000003,
      "loss": 3.0866,
      "step": 272810
    },
    {
      "epoch": 0.10313438242707203,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001454362,
      "loss": 3.167,
      "step": 272820
    },
    {
      "epoch": 0.10315302566323022,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014543420000000002,
      "loss": 3.2453,
      "step": 272830
    },
    {
      "epoch": 0.10317166889938839,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014543219999999999,
      "loss": 3.1063,
      "step": 272840
    },
    {
      "epoch": 0.10319031213554658,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001454302,
      "loss": 3.1958,
      "step": 272850
    },
    {
      "epoch": 0.10320895537170477,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001454282,
      "loss": 3.146,
      "step": 272860
    },
    {
      "epoch": 0.10322759860786294,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001454262,
      "loss": 3.2542,
      "step": 272870
    },
    {
      "epoch": 0.10324624184402113,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014542420000000002,
      "loss": 3.1729,
      "step": 272880
    },
    {
      "epoch": 0.10326488508017932,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001454222,
      "loss": 2.9865,
      "step": 272890
    },
    {
      "epoch": 0.1032835283163375,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014542020000000001,
      "loss": 3.1688,
      "step": 272900
    },
    {
      "epoch": 0.10330217155249569,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001454182,
      "loss": 3.0999,
      "step": 272910
    },
    {
      "epoch": 0.10332081478865386,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001454162,
      "loss": 3.1419,
      "step": 272920
    },
    {
      "epoch": 0.10333945802481205,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001454142,
      "loss": 3.0568,
      "step": 272930
    },
    {
      "epoch": 0.10335810126097024,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001454122,
      "loss": 3.1414,
      "step": 272940
    },
    {
      "epoch": 0.10337674449712841,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014541020000000002,
      "loss": 3.108,
      "step": 272950
    },
    {
      "epoch": 0.1033953877332866,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014540820000000002,
      "loss": 3.1923,
      "step": 272960
    },
    {
      "epoch": 0.10341403096944478,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001454062,
      "loss": 3.2419,
      "step": 272970
    },
    {
      "epoch": 0.10343267420560297,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001454042,
      "loss": 3.0503,
      "step": 272980
    },
    {
      "epoch": 0.10345131744176116,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001454022,
      "loss": 3.013,
      "step": 272990
    },
    {
      "epoch": 0.10346996067791933,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001454002,
      "loss": 3.0937,
      "step": 273000
    },
    {
      "epoch": 0.10348860391407752,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014539820000000002,
      "loss": 3.1911,
      "step": 273010
    },
    {
      "epoch": 0.10350724715023571,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014539620000000002,
      "loss": 3.0383,
      "step": 273020
    },
    {
      "epoch": 0.10352589038639388,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001453942,
      "loss": 3.1299,
      "step": 273030
    },
    {
      "epoch": 0.10354453362255207,
      "grad_norm": 0.75,
      "learning_rate": 0.0001453922,
      "loss": 3.1218,
      "step": 273040
    },
    {
      "epoch": 0.10356317685871025,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001453902,
      "loss": 3.2295,
      "step": 273050
    },
    {
      "epoch": 0.10358182009486844,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001453882,
      "loss": 2.973,
      "step": 273060
    },
    {
      "epoch": 0.10360046333102663,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001453862,
      "loss": 3.218,
      "step": 273070
    },
    {
      "epoch": 0.1036191065671848,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014538420000000002,
      "loss": 3.1641,
      "step": 273080
    },
    {
      "epoch": 0.10363774980334299,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001453822,
      "loss": 3.0285,
      "step": 273090
    },
    {
      "epoch": 0.10365639303950117,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001453802,
      "loss": 3.1611,
      "step": 273100
    },
    {
      "epoch": 0.10367503627565935,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001453782,
      "loss": 3.2509,
      "step": 273110
    },
    {
      "epoch": 0.10369367951181754,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001453762,
      "loss": 3.0361,
      "step": 273120
    },
    {
      "epoch": 0.10371232274797572,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014537420000000002,
      "loss": 3.1008,
      "step": 273130
    },
    {
      "epoch": 0.10373096598413391,
      "grad_norm": 0.75,
      "learning_rate": 0.0001453722,
      "loss": 3.0706,
      "step": 273140
    },
    {
      "epoch": 0.1037496092202921,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001453702,
      "loss": 3.1019,
      "step": 273150
    },
    {
      "epoch": 0.10376825245645027,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001453682,
      "loss": 3.1837,
      "step": 273160
    },
    {
      "epoch": 0.10378689569260846,
      "grad_norm": 0.75,
      "learning_rate": 0.0001453662,
      "loss": 3.1994,
      "step": 273170
    },
    {
      "epoch": 0.10380553892876664,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014536420000000003,
      "loss": 3.1527,
      "step": 273180
    },
    {
      "epoch": 0.10382418216492482,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001453622,
      "loss": 3.0706,
      "step": 273190
    },
    {
      "epoch": 0.10384282540108301,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014536020000000002,
      "loss": 3.1221,
      "step": 273200
    },
    {
      "epoch": 0.10386146863724119,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001453582,
      "loss": 3.0817,
      "step": 273210
    },
    {
      "epoch": 0.10388011187339938,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001453562,
      "loss": 3.042,
      "step": 273220
    },
    {
      "epoch": 0.10389875510955757,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001453542,
      "loss": 3.1291,
      "step": 273230
    },
    {
      "epoch": 0.10391739834571574,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001453522,
      "loss": 3.1538,
      "step": 273240
    },
    {
      "epoch": 0.10393604158187393,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014535020000000002,
      "loss": 3.0534,
      "step": 273250
    },
    {
      "epoch": 0.1039546848180321,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001453482,
      "loss": 3.1871,
      "step": 273260
    },
    {
      "epoch": 0.1039733280541903,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014534620000000001,
      "loss": 3.1879,
      "step": 273270
    },
    {
      "epoch": 0.10399197129034848,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001453442,
      "loss": 3.1147,
      "step": 273280
    },
    {
      "epoch": 0.10401061452650666,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001453422,
      "loss": 3.2151,
      "step": 273290
    },
    {
      "epoch": 0.10402925776266485,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014534020000000003,
      "loss": 3.1038,
      "step": 273300
    },
    {
      "epoch": 0.10404790099882302,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001453382,
      "loss": 3.1808,
      "step": 273310
    },
    {
      "epoch": 0.10406654423498121,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014533620000000002,
      "loss": 3.0174,
      "step": 273320
    },
    {
      "epoch": 0.1040851874711394,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001453342,
      "loss": 3.1519,
      "step": 273330
    },
    {
      "epoch": 0.10410383070729758,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001453322,
      "loss": 3.1437,
      "step": 273340
    },
    {
      "epoch": 0.10412247394345577,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001453302,
      "loss": 3.084,
      "step": 273350
    },
    {
      "epoch": 0.10414111717961395,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001453282,
      "loss": 3.1763,
      "step": 273360
    },
    {
      "epoch": 0.10415976041577213,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014532620000000002,
      "loss": 3.1497,
      "step": 273370
    },
    {
      "epoch": 0.10417840365193032,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001453242,
      "loss": 3.0137,
      "step": 273380
    },
    {
      "epoch": 0.1041970468880885,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014532220000000002,
      "loss": 3.1341,
      "step": 273390
    },
    {
      "epoch": 0.10421569012424668,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001453202,
      "loss": 3.163,
      "step": 273400
    },
    {
      "epoch": 0.10423433336040487,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001453182,
      "loss": 3.1375,
      "step": 273410
    },
    {
      "epoch": 0.10425297659656305,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001453162,
      "loss": 3.1392,
      "step": 273420
    },
    {
      "epoch": 0.10427161983272124,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001453142,
      "loss": 3.0813,
      "step": 273430
    },
    {
      "epoch": 0.10429026306887941,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014531220000000002,
      "loss": 3.091,
      "step": 273440
    },
    {
      "epoch": 0.1043089063050376,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001453102,
      "loss": 3.1629,
      "step": 273450
    },
    {
      "epoch": 0.10432754954119579,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001453082,
      "loss": 3.0583,
      "step": 273460
    },
    {
      "epoch": 0.10434619277735396,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001453062,
      "loss": 3.1186,
      "step": 273470
    },
    {
      "epoch": 0.10436483601351215,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001453042,
      "loss": 2.993,
      "step": 273480
    },
    {
      "epoch": 0.10438347924967034,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001453022,
      "loss": 3.0863,
      "step": 273490
    },
    {
      "epoch": 0.10440212248582852,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001453002,
      "loss": 3.1722,
      "step": 273500
    },
    {
      "epoch": 0.1044207657219867,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014529820000000002,
      "loss": 3.0776,
      "step": 273510
    },
    {
      "epoch": 0.10443940895814488,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001452962,
      "loss": 3.1997,
      "step": 273520
    },
    {
      "epoch": 0.10445805219430307,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001452942,
      "loss": 3.0036,
      "step": 273530
    },
    {
      "epoch": 0.10447669543046126,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001452922,
      "loss": 3.1659,
      "step": 273540
    },
    {
      "epoch": 0.10449533866661943,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001452902,
      "loss": 3.083,
      "step": 273550
    },
    {
      "epoch": 0.10451398190277762,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001452882,
      "loss": 3.0143,
      "step": 273560
    },
    {
      "epoch": 0.10453262513893581,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014528620000000002,
      "loss": 3.185,
      "step": 273570
    },
    {
      "epoch": 0.10455126837509399,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014528420000000001,
      "loss": 3.1887,
      "step": 273580
    },
    {
      "epoch": 0.10456991161125218,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001452822,
      "loss": 3.0976,
      "step": 273590
    },
    {
      "epoch": 0.10458855484741035,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001452802,
      "loss": 3.1866,
      "step": 273600
    },
    {
      "epoch": 0.10460719808356854,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001452782,
      "loss": 3.1111,
      "step": 273610
    },
    {
      "epoch": 0.10462584131972673,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014527620000000002,
      "loss": 3.1055,
      "step": 273620
    },
    {
      "epoch": 0.1046444845558849,
      "grad_norm": 0.75,
      "learning_rate": 0.0001452742,
      "loss": 3.0802,
      "step": 273630
    },
    {
      "epoch": 0.10466312779204309,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014527220000000001,
      "loss": 2.9969,
      "step": 273640
    },
    {
      "epoch": 0.10468177102820127,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001452702,
      "loss": 3.1578,
      "step": 273650
    },
    {
      "epoch": 0.10470041426435946,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001452682,
      "loss": 2.9898,
      "step": 273660
    },
    {
      "epoch": 0.10471905750051765,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001452662,
      "loss": 3.0764,
      "step": 273670
    },
    {
      "epoch": 0.10473770073667582,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001452642,
      "loss": 2.9872,
      "step": 273680
    },
    {
      "epoch": 0.10475634397283401,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014526220000000002,
      "loss": 3.2616,
      "step": 273690
    },
    {
      "epoch": 0.1047749872089922,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001452602,
      "loss": 3.1275,
      "step": 273700
    },
    {
      "epoch": 0.10479363044515037,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001452582,
      "loss": 3.1174,
      "step": 273710
    },
    {
      "epoch": 0.10481227368130856,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001452562,
      "loss": 3.1838,
      "step": 273720
    },
    {
      "epoch": 0.10483091691746674,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001452542,
      "loss": 3.1359,
      "step": 273730
    },
    {
      "epoch": 0.10484956015362493,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014525220000000003,
      "loss": 3.1312,
      "step": 273740
    },
    {
      "epoch": 0.10486820338978312,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001452502,
      "loss": 3.0241,
      "step": 273750
    },
    {
      "epoch": 0.10488684662594129,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014524820000000002,
      "loss": 3.051,
      "step": 273760
    },
    {
      "epoch": 0.10490548986209948,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001452462,
      "loss": 3.0534,
      "step": 273770
    },
    {
      "epoch": 0.10492413309825765,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001452442,
      "loss": 3.1997,
      "step": 273780
    },
    {
      "epoch": 0.10494277633441584,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014524220000000003,
      "loss": 3.2356,
      "step": 273790
    },
    {
      "epoch": 0.10496141957057403,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001452402,
      "loss": 3.0905,
      "step": 273800
    },
    {
      "epoch": 0.10498006280673221,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014523820000000002,
      "loss": 3.1388,
      "step": 273810
    },
    {
      "epoch": 0.1049987060428904,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001452362,
      "loss": 3.0601,
      "step": 273820
    },
    {
      "epoch": 0.10501734927904859,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001452342,
      "loss": 3.0245,
      "step": 273830
    },
    {
      "epoch": 0.10503599251520676,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001452322,
      "loss": 3.1303,
      "step": 273840
    },
    {
      "epoch": 0.10505463575136495,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001452302,
      "loss": 3.1104,
      "step": 273850
    },
    {
      "epoch": 0.10507327898752312,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014522820000000003,
      "loss": 3.1837,
      "step": 273860
    },
    {
      "epoch": 0.10509192222368131,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001452262,
      "loss": 3.2262,
      "step": 273870
    },
    {
      "epoch": 0.1051105654598395,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014522420000000002,
      "loss": 3.2048,
      "step": 273880
    },
    {
      "epoch": 0.10512920869599768,
      "grad_norm": 0.75,
      "learning_rate": 0.0001452222,
      "loss": 3.1024,
      "step": 273890
    },
    {
      "epoch": 0.10514785193215587,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001452202,
      "loss": 3.1282,
      "step": 273900
    },
    {
      "epoch": 0.10516649516831406,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001452182,
      "loss": 2.9592,
      "step": 273910
    },
    {
      "epoch": 0.10518513840447223,
      "grad_norm": 0.75,
      "learning_rate": 0.0001452162,
      "loss": 3.1353,
      "step": 273920
    },
    {
      "epoch": 0.10520378164063042,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014521420000000002,
      "loss": 3.119,
      "step": 273930
    },
    {
      "epoch": 0.1052224248767886,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001452122,
      "loss": 3.2103,
      "step": 273940
    },
    {
      "epoch": 0.10524106811294678,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014521020000000001,
      "loss": 3.061,
      "step": 273950
    },
    {
      "epoch": 0.10525971134910497,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001452082,
      "loss": 3.1343,
      "step": 273960
    },
    {
      "epoch": 0.10527835458526315,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001452062,
      "loss": 3.1389,
      "step": 273970
    },
    {
      "epoch": 0.10529699782142134,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001452042,
      "loss": 3.0828,
      "step": 273980
    },
    {
      "epoch": 0.10531564105757951,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001452022,
      "loss": 3.0786,
      "step": 273990
    },
    {
      "epoch": 0.1053342842937377,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014520020000000002,
      "loss": 3.0926,
      "step": 274000
    },
    {
      "epoch": 0.10535292752989589,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014519820000000002,
      "loss": 3.0825,
      "step": 274010
    },
    {
      "epoch": 0.10537157076605407,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001451962,
      "loss": 3.0502,
      "step": 274020
    },
    {
      "epoch": 0.10539021400221225,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001451942,
      "loss": 3.0928,
      "step": 274030
    },
    {
      "epoch": 0.10540885723837044,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001451922,
      "loss": 3.0047,
      "step": 274040
    },
    {
      "epoch": 0.10542750047452862,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001451902,
      "loss": 3.1992,
      "step": 274050
    },
    {
      "epoch": 0.10544614371068681,
      "grad_norm": 0.75,
      "learning_rate": 0.0001451882,
      "loss": 3.1258,
      "step": 274060
    },
    {
      "epoch": 0.10546478694684498,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014518620000000002,
      "loss": 3.1986,
      "step": 274070
    },
    {
      "epoch": 0.10548343018300317,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001451842,
      "loss": 3.1673,
      "step": 274080
    },
    {
      "epoch": 0.10550207341916136,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001451822,
      "loss": 3.0339,
      "step": 274090
    },
    {
      "epoch": 0.10552071665531954,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001451802,
      "loss": 3.053,
      "step": 274100
    },
    {
      "epoch": 0.10553935989147772,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001451782,
      "loss": 3.0722,
      "step": 274110
    },
    {
      "epoch": 0.1055580031276359,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001451762,
      "loss": 3.1592,
      "step": 274120
    },
    {
      "epoch": 0.10557664636379409,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014517420000000002,
      "loss": 3.1361,
      "step": 274130
    },
    {
      "epoch": 0.10559528959995228,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001451722,
      "loss": 3.1118,
      "step": 274140
    },
    {
      "epoch": 0.10561393283611045,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001451702,
      "loss": 3.2192,
      "step": 274150
    },
    {
      "epoch": 0.10563257607226864,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001451682,
      "loss": 3.2193,
      "step": 274160
    },
    {
      "epoch": 0.10565121930842683,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001451662,
      "loss": 3.2204,
      "step": 274170
    },
    {
      "epoch": 0.105669862544585,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014516420000000002,
      "loss": 3.172,
      "step": 274180
    },
    {
      "epoch": 0.1056885057807432,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001451622,
      "loss": 3.1823,
      "step": 274190
    },
    {
      "epoch": 0.10570714901690137,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001451602,
      "loss": 3.1236,
      "step": 274200
    },
    {
      "epoch": 0.10572579225305956,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001451582,
      "loss": 3.0786,
      "step": 274210
    },
    {
      "epoch": 0.10574443548921775,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001451562,
      "loss": 3.1407,
      "step": 274220
    },
    {
      "epoch": 0.10576307872537592,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014515420000000003,
      "loss": 3.1279,
      "step": 274230
    },
    {
      "epoch": 0.10578172196153411,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001451522,
      "loss": 3.1002,
      "step": 274240
    },
    {
      "epoch": 0.1058003651976923,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014515020000000002,
      "loss": 3.2796,
      "step": 274250
    },
    {
      "epoch": 0.10581900843385048,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014514820000000001,
      "loss": 3.1349,
      "step": 274260
    },
    {
      "epoch": 0.10583765167000866,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001451462,
      "loss": 3.2441,
      "step": 274270
    },
    {
      "epoch": 0.10585629490616684,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001451442,
      "loss": 3.0802,
      "step": 274280
    },
    {
      "epoch": 0.10587493814232503,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001451422,
      "loss": 3.1022,
      "step": 274290
    },
    {
      "epoch": 0.10589358137848322,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014514020000000002,
      "loss": 3.212,
      "step": 274300
    },
    {
      "epoch": 0.10591222461464139,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001451382,
      "loss": 3.03,
      "step": 274310
    },
    {
      "epoch": 0.10593086785079958,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014513620000000001,
      "loss": 3.1521,
      "step": 274320
    },
    {
      "epoch": 0.10594951108695776,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001451342,
      "loss": 3.2266,
      "step": 274330
    },
    {
      "epoch": 0.10596815432311595,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001451322,
      "loss": 3.1126,
      "step": 274340
    },
    {
      "epoch": 0.10598679755927413,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00014513020000000003,
      "loss": 3.1011,
      "step": 274350
    },
    {
      "epoch": 0.10600544079543231,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001451282,
      "loss": 3.1002,
      "step": 274360
    },
    {
      "epoch": 0.1060240840315905,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014512620000000002,
      "loss": 3.0674,
      "step": 274370
    },
    {
      "epoch": 0.10604272726774869,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001451242,
      "loss": 3.1216,
      "step": 274380
    },
    {
      "epoch": 0.10606137050390686,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001451222,
      "loss": 3.1457,
      "step": 274390
    },
    {
      "epoch": 0.10608001374006505,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001451202,
      "loss": 3.1266,
      "step": 274400
    },
    {
      "epoch": 0.10609865697622323,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001451182,
      "loss": 3.1624,
      "step": 274410
    },
    {
      "epoch": 0.10611730021238142,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014511620000000003,
      "loss": 2.9888,
      "step": 274420
    },
    {
      "epoch": 0.1061359434485396,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001451142,
      "loss": 3.1449,
      "step": 274430
    },
    {
      "epoch": 0.10615458668469778,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014511220000000002,
      "loss": 3.1315,
      "step": 274440
    },
    {
      "epoch": 0.10617322992085597,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014511019999999998,
      "loss": 3.1507,
      "step": 274450
    },
    {
      "epoch": 0.10619187315701414,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001451082,
      "loss": 2.9927,
      "step": 274460
    },
    {
      "epoch": 0.10621051639317233,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001451062,
      "loss": 3.1917,
      "step": 274470
    },
    {
      "epoch": 0.10622915962933052,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001451042,
      "loss": 3.1487,
      "step": 274480
    },
    {
      "epoch": 0.1062478028654887,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014510220000000002,
      "loss": 3.2509,
      "step": 274490
    },
    {
      "epoch": 0.10626644610164689,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001451002,
      "loss": 2.9923,
      "step": 274500
    },
    {
      "epoch": 0.10628508933780507,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001450982,
      "loss": 3.094,
      "step": 274510
    },
    {
      "epoch": 0.10630373257396325,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001450962,
      "loss": 3.1203,
      "step": 274520
    },
    {
      "epoch": 0.10632237581012144,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001450942,
      "loss": 3.1247,
      "step": 274530
    },
    {
      "epoch": 0.10634101904627961,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001450922,
      "loss": 3.0817,
      "step": 274540
    },
    {
      "epoch": 0.1063596622824378,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001450902,
      "loss": 3.0421,
      "step": 274550
    },
    {
      "epoch": 0.10637830551859599,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014508820000000002,
      "loss": 3.1697,
      "step": 274560
    },
    {
      "epoch": 0.10639694875475417,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014508620000000001,
      "loss": 3.1216,
      "step": 274570
    },
    {
      "epoch": 0.10641559199091236,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001450842,
      "loss": 3.0062,
      "step": 274580
    },
    {
      "epoch": 0.10643423522707054,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001450822,
      "loss": 2.9861,
      "step": 274590
    },
    {
      "epoch": 0.10645287846322872,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001450802,
      "loss": 3.0787,
      "step": 274600
    },
    {
      "epoch": 0.10647152169938691,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001450782,
      "loss": 3.0375,
      "step": 274610
    },
    {
      "epoch": 0.10649016493554508,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014507620000000002,
      "loss": 3.1849,
      "step": 274620
    },
    {
      "epoch": 0.10650880817170327,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014507420000000001,
      "loss": 3.1376,
      "step": 274630
    },
    {
      "epoch": 0.10652745140786146,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001450722,
      "loss": 3.0606,
      "step": 274640
    },
    {
      "epoch": 0.10654609464401964,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001450702,
      "loss": 3.1809,
      "step": 274650
    },
    {
      "epoch": 0.10656473788017783,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001450682,
      "loss": 3.0362,
      "step": 274660
    },
    {
      "epoch": 0.106583381116336,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001450662,
      "loss": 3.1608,
      "step": 274670
    },
    {
      "epoch": 0.10660202435249419,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001450642,
      "loss": 2.9554,
      "step": 274680
    },
    {
      "epoch": 0.10662066758865238,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014506220000000002,
      "loss": 3.1888,
      "step": 274690
    },
    {
      "epoch": 0.10663931082481055,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001450602,
      "loss": 3.0279,
      "step": 274700
    },
    {
      "epoch": 0.10665795406096874,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001450582,
      "loss": 3.1192,
      "step": 274710
    },
    {
      "epoch": 0.10667659729712693,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001450562,
      "loss": 3.121,
      "step": 274720
    },
    {
      "epoch": 0.10669524053328511,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001450542,
      "loss": 3.0841,
      "step": 274730
    },
    {
      "epoch": 0.1067138837694433,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014505220000000002,
      "loss": 3.1633,
      "step": 274740
    },
    {
      "epoch": 0.10673252700560147,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014505020000000002,
      "loss": 3.0919,
      "step": 274750
    },
    {
      "epoch": 0.10675117024175966,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001450482,
      "loss": 2.9814,
      "step": 274760
    },
    {
      "epoch": 0.10676981347791785,
      "grad_norm": 0.75,
      "learning_rate": 0.0001450462,
      "loss": 3.1575,
      "step": 274770
    },
    {
      "epoch": 0.10678845671407602,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001450442,
      "loss": 3.2308,
      "step": 274780
    },
    {
      "epoch": 0.10680709995023421,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014504220000000003,
      "loss": 3.1687,
      "step": 274790
    },
    {
      "epoch": 0.10682574318639239,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001450402,
      "loss": 3.1441,
      "step": 274800
    },
    {
      "epoch": 0.10684438642255058,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014503820000000002,
      "loss": 3.0923,
      "step": 274810
    },
    {
      "epoch": 0.10686302965870877,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001450362,
      "loss": 3.0447,
      "step": 274820
    },
    {
      "epoch": 0.10688167289486694,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001450342,
      "loss": 3.0523,
      "step": 274830
    },
    {
      "epoch": 0.10690031613102513,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001450322,
      "loss": 3.1947,
      "step": 274840
    },
    {
      "epoch": 0.10691895936718332,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001450302,
      "loss": 3.1261,
      "step": 274850
    },
    {
      "epoch": 0.1069376026033415,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014502820000000002,
      "loss": 3.0824,
      "step": 274860
    },
    {
      "epoch": 0.10695624583949968,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001450262,
      "loss": 3.2956,
      "step": 274870
    },
    {
      "epoch": 0.10697488907565786,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001450242,
      "loss": 3.1061,
      "step": 274880
    },
    {
      "epoch": 0.10699353231181605,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001450222,
      "loss": 3.1724,
      "step": 274890
    },
    {
      "epoch": 0.10701217554797424,
      "grad_norm": 0.75,
      "learning_rate": 0.0001450202,
      "loss": 2.9814,
      "step": 274900
    },
    {
      "epoch": 0.10703081878413241,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014501820000000003,
      "loss": 3.0462,
      "step": 274910
    },
    {
      "epoch": 0.1070494620202906,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001450162,
      "loss": 3.1219,
      "step": 274920
    },
    {
      "epoch": 0.10706810525644879,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014501420000000002,
      "loss": 3.1998,
      "step": 274930
    },
    {
      "epoch": 0.10708674849260696,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001450122,
      "loss": 3.1419,
      "step": 274940
    },
    {
      "epoch": 0.10710539172876515,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001450102,
      "loss": 3.0849,
      "step": 274950
    },
    {
      "epoch": 0.10712403496492333,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001450082,
      "loss": 3.0701,
      "step": 274960
    },
    {
      "epoch": 0.10714267820108152,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001450062,
      "loss": 3.0818,
      "step": 274970
    },
    {
      "epoch": 0.1071613214372397,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014500420000000002,
      "loss": 3.1184,
      "step": 274980
    },
    {
      "epoch": 0.10717996467339788,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001450022,
      "loss": 3.1223,
      "step": 274990
    },
    {
      "epoch": 0.10719860790955607,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014500020000000001,
      "loss": 3.1586,
      "step": 275000
    },
    {
      "epoch": 0.10721725114571425,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001449982,
      "loss": 3.1586,
      "step": 275010
    },
    {
      "epoch": 0.10723589438187243,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001449962,
      "loss": 3.1081,
      "step": 275020
    },
    {
      "epoch": 0.10725453761803062,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001449942,
      "loss": 3.0359,
      "step": 275030
    },
    {
      "epoch": 0.1072731808541888,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001449922,
      "loss": 3.0979,
      "step": 275040
    },
    {
      "epoch": 0.10729182409034699,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014499020000000002,
      "loss": 3.2166,
      "step": 275050
    },
    {
      "epoch": 0.10731046732650518,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001449882,
      "loss": 3.1761,
      "step": 275060
    },
    {
      "epoch": 0.10732911056266335,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001449862,
      "loss": 3.1956,
      "step": 275070
    },
    {
      "epoch": 0.10734775379882154,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001449842,
      "loss": 3.0681,
      "step": 275080
    },
    {
      "epoch": 0.10736639703497972,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001449822,
      "loss": 3.0255,
      "step": 275090
    },
    {
      "epoch": 0.1073850402711379,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001449802,
      "loss": 3.1609,
      "step": 275100
    },
    {
      "epoch": 0.1074036835072961,
      "grad_norm": 0.75,
      "learning_rate": 0.0001449782,
      "loss": 3.0578,
      "step": 275110
    },
    {
      "epoch": 0.10742232674345427,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014497620000000002,
      "loss": 3.0084,
      "step": 275120
    },
    {
      "epoch": 0.10744096997961246,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001449742,
      "loss": 3.1274,
      "step": 275130
    },
    {
      "epoch": 0.10745961321577063,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001449722,
      "loss": 3.1923,
      "step": 275140
    },
    {
      "epoch": 0.10747825645192882,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001449702,
      "loss": 3.1652,
      "step": 275150
    },
    {
      "epoch": 0.10749689968808701,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001449682,
      "loss": 3.1097,
      "step": 275160
    },
    {
      "epoch": 0.10751554292424519,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001449662,
      "loss": 3.2155,
      "step": 275170
    },
    {
      "epoch": 0.10753418616040337,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014496420000000002,
      "loss": 3.2067,
      "step": 275180
    },
    {
      "epoch": 0.10755282939656156,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001449622,
      "loss": 2.9281,
      "step": 275190
    },
    {
      "epoch": 0.10757147263271974,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001449602,
      "loss": 3.0201,
      "step": 275200
    },
    {
      "epoch": 0.10759011586887793,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001449582,
      "loss": 3.08,
      "step": 275210
    },
    {
      "epoch": 0.1076087591050361,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001449562,
      "loss": 3.1105,
      "step": 275220
    },
    {
      "epoch": 0.10762740234119429,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014495420000000002,
      "loss": 3.0179,
      "step": 275230
    },
    {
      "epoch": 0.10764604557735248,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014495220000000002,
      "loss": 3.1166,
      "step": 275240
    },
    {
      "epoch": 0.10766468881351066,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014495020000000001,
      "loss": 3.0605,
      "step": 275250
    },
    {
      "epoch": 0.10768333204966885,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001449482,
      "loss": 3.0929,
      "step": 275260
    },
    {
      "epoch": 0.10770197528582703,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001449462,
      "loss": 3.1198,
      "step": 275270
    },
    {
      "epoch": 0.10772061852198521,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001449442,
      "loss": 3.0544,
      "step": 275280
    },
    {
      "epoch": 0.1077392617581434,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001449422,
      "loss": 3.0346,
      "step": 275290
    },
    {
      "epoch": 0.10775790499430157,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014494020000000002,
      "loss": 3.0254,
      "step": 275300
    },
    {
      "epoch": 0.10777654823045976,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014493820000000001,
      "loss": 3.1402,
      "step": 275310
    },
    {
      "epoch": 0.10779519146661795,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001449362,
      "loss": 3.1766,
      "step": 275320
    },
    {
      "epoch": 0.10781383470277613,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001449342,
      "loss": 3.0396,
      "step": 275330
    },
    {
      "epoch": 0.10783247793893432,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001449322,
      "loss": 3.2712,
      "step": 275340
    },
    {
      "epoch": 0.10785112117509249,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014493020000000002,
      "loss": 3.0437,
      "step": 275350
    },
    {
      "epoch": 0.10786976441125068,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001449282,
      "loss": 3.0992,
      "step": 275360
    },
    {
      "epoch": 0.10788840764740887,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014492620000000002,
      "loss": 3.1217,
      "step": 275370
    },
    {
      "epoch": 0.10790705088356704,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001449242,
      "loss": 3.1646,
      "step": 275380
    },
    {
      "epoch": 0.10792569411972523,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001449222,
      "loss": 3.0956,
      "step": 275390
    },
    {
      "epoch": 0.10794433735588342,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014492020000000003,
      "loss": 3.2164,
      "step": 275400
    },
    {
      "epoch": 0.1079629805920416,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001449182,
      "loss": 3.0328,
      "step": 275410
    },
    {
      "epoch": 0.10798162382819979,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014491620000000002,
      "loss": 3.1468,
      "step": 275420
    },
    {
      "epoch": 0.10800026706435796,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001449142,
      "loss": 3.1531,
      "step": 275430
    },
    {
      "epoch": 0.10801891030051615,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001449122,
      "loss": 3.1799,
      "step": 275440
    },
    {
      "epoch": 0.10803755353667434,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001449102,
      "loss": 3.1716,
      "step": 275450
    },
    {
      "epoch": 0.10805619677283251,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001449082,
      "loss": 3.2262,
      "step": 275460
    },
    {
      "epoch": 0.1080748400089907,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014490620000000003,
      "loss": 3.0923,
      "step": 275470
    },
    {
      "epoch": 0.10809348324514888,
      "grad_norm": 0.75,
      "learning_rate": 0.0001449042,
      "loss": 3.0938,
      "step": 275480
    },
    {
      "epoch": 0.10811212648130707,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014490220000000002,
      "loss": 3.0314,
      "step": 275490
    },
    {
      "epoch": 0.10813076971746526,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014490019999999999,
      "loss": 3.0962,
      "step": 275500
    },
    {
      "epoch": 0.10814941295362343,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001448982,
      "loss": 3.2402,
      "step": 275510
    },
    {
      "epoch": 0.10816805618978162,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001448962,
      "loss": 3.1662,
      "step": 275520
    },
    {
      "epoch": 0.10818669942593981,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001448942,
      "loss": 3.0272,
      "step": 275530
    },
    {
      "epoch": 0.10820534266209798,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014489220000000002,
      "loss": 3.2229,
      "step": 275540
    },
    {
      "epoch": 0.10822398589825617,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001448902,
      "loss": 3.0591,
      "step": 275550
    },
    {
      "epoch": 0.10824262913441435,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001448882,
      "loss": 3.1759,
      "step": 275560
    },
    {
      "epoch": 0.10826127237057254,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001448862,
      "loss": 3.1959,
      "step": 275570
    },
    {
      "epoch": 0.10827991560673073,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001448842,
      "loss": 3.1595,
      "step": 275580
    },
    {
      "epoch": 0.1082985588428889,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001448822,
      "loss": 3.151,
      "step": 275590
    },
    {
      "epoch": 0.10831720207904709,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001448802,
      "loss": 3.2174,
      "step": 275600
    },
    {
      "epoch": 0.10833584531520528,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014487820000000002,
      "loss": 3.1843,
      "step": 275610
    },
    {
      "epoch": 0.10835448855136345,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014487620000000001,
      "loss": 3.1959,
      "step": 275620
    },
    {
      "epoch": 0.10837313178752164,
      "grad_norm": 0.75,
      "learning_rate": 0.0001448742,
      "loss": 3.1028,
      "step": 275630
    },
    {
      "epoch": 0.10839177502367982,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001448722,
      "loss": 3.276,
      "step": 275640
    },
    {
      "epoch": 0.108410418259838,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001448702,
      "loss": 3.0505,
      "step": 275650
    },
    {
      "epoch": 0.1084290614959962,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001448682,
      "loss": 3.1009,
      "step": 275660
    },
    {
      "epoch": 0.10844770473215437,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001448662,
      "loss": 3.1788,
      "step": 275670
    },
    {
      "epoch": 0.10846634796831256,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014486420000000001,
      "loss": 3.1385,
      "step": 275680
    },
    {
      "epoch": 0.10848499120447073,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001448622,
      "loss": 3.0785,
      "step": 275690
    },
    {
      "epoch": 0.10850363444062892,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001448602,
      "loss": 3.175,
      "step": 275700
    },
    {
      "epoch": 0.10852227767678711,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001448582,
      "loss": 3.0903,
      "step": 275710
    },
    {
      "epoch": 0.10854092091294529,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001448562,
      "loss": 3.2462,
      "step": 275720
    },
    {
      "epoch": 0.10855956414910348,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00014485420000000002,
      "loss": 3.0929,
      "step": 275730
    },
    {
      "epoch": 0.10857820738526167,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014485220000000002,
      "loss": 3.2519,
      "step": 275740
    },
    {
      "epoch": 0.10859685062141984,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001448502,
      "loss": 3.1829,
      "step": 275750
    },
    {
      "epoch": 0.10861549385757803,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001448482,
      "loss": 3.0278,
      "step": 275760
    },
    {
      "epoch": 0.1086341370937362,
      "grad_norm": 0.625,
      "learning_rate": 0.0001448462,
      "loss": 3.1456,
      "step": 275770
    },
    {
      "epoch": 0.1086527803298944,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001448442,
      "loss": 3.1622,
      "step": 275780
    },
    {
      "epoch": 0.10867142356605258,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014484220000000002,
      "loss": 3.262,
      "step": 275790
    },
    {
      "epoch": 0.10869006680221076,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014484020000000002,
      "loss": 3.0514,
      "step": 275800
    },
    {
      "epoch": 0.10870871003836895,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001448382,
      "loss": 3.2915,
      "step": 275810
    },
    {
      "epoch": 0.10872735327452712,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001448362,
      "loss": 3.1511,
      "step": 275820
    },
    {
      "epoch": 0.10874599651068531,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001448342,
      "loss": 3.238,
      "step": 275830
    },
    {
      "epoch": 0.1087646397468435,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014483220000000003,
      "loss": 3.1241,
      "step": 275840
    },
    {
      "epoch": 0.10878328298300168,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001448302,
      "loss": 3.2555,
      "step": 275850
    },
    {
      "epoch": 0.10880192621915986,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014482820000000002,
      "loss": 3.1296,
      "step": 275860
    },
    {
      "epoch": 0.10882056945531805,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001448262,
      "loss": 3.1928,
      "step": 275870
    },
    {
      "epoch": 0.10883921269147623,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001448242,
      "loss": 3.0653,
      "step": 275880
    },
    {
      "epoch": 0.10885785592763442,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001448222,
      "loss": 3.0985,
      "step": 275890
    },
    {
      "epoch": 0.10887649916379259,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001448202,
      "loss": 3.1406,
      "step": 275900
    },
    {
      "epoch": 0.10889514239995078,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014481820000000002,
      "loss": 3.1189,
      "step": 275910
    },
    {
      "epoch": 0.10891378563610897,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001448162,
      "loss": 3.1394,
      "step": 275920
    },
    {
      "epoch": 0.10893242887226715,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014481420000000001,
      "loss": 3.0145,
      "step": 275930
    },
    {
      "epoch": 0.10895107210842533,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001448122,
      "loss": 3.0698,
      "step": 275940
    },
    {
      "epoch": 0.10896971534458352,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001448102,
      "loss": 3.0801,
      "step": 275950
    },
    {
      "epoch": 0.1089883585807417,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014480820000000003,
      "loss": 3.2326,
      "step": 275960
    },
    {
      "epoch": 0.10900700181689989,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001448062,
      "loss": 3.1958,
      "step": 275970
    },
    {
      "epoch": 0.10902564505305806,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014480420000000002,
      "loss": 3.1518,
      "step": 275980
    },
    {
      "epoch": 0.10904428828921625,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001448022,
      "loss": 3.0697,
      "step": 275990
    },
    {
      "epoch": 0.10906293152537444,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001448002,
      "loss": 3.0736,
      "step": 276000
    },
    {
      "epoch": 0.10908157476153262,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001447982,
      "loss": 3.1556,
      "step": 276010
    },
    {
      "epoch": 0.1091002179976908,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001447962,
      "loss": 3.1161,
      "step": 276020
    },
    {
      "epoch": 0.10911886123384898,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014479420000000002,
      "loss": 3.075,
      "step": 276030
    },
    {
      "epoch": 0.10913750447000717,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001447922,
      "loss": 3.1402,
      "step": 276040
    },
    {
      "epoch": 0.10915614770616536,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014479020000000002,
      "loss": 3.0878,
      "step": 276050
    },
    {
      "epoch": 0.10917479094232353,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014478819999999998,
      "loss": 3.2239,
      "step": 276060
    },
    {
      "epoch": 0.10919343417848172,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001447862,
      "loss": 3.1973,
      "step": 276070
    },
    {
      "epoch": 0.10921207741463991,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001447842,
      "loss": 3.026,
      "step": 276080
    },
    {
      "epoch": 0.10923072065079809,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001447822,
      "loss": 3.1906,
      "step": 276090
    },
    {
      "epoch": 0.10924936388695627,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014478020000000002,
      "loss": 3.1916,
      "step": 276100
    },
    {
      "epoch": 0.10926800712311445,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001447782,
      "loss": 3.0379,
      "step": 276110
    },
    {
      "epoch": 0.10928665035927264,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001447762,
      "loss": 3.0965,
      "step": 276120
    },
    {
      "epoch": 0.10930529359543083,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001447742,
      "loss": 3.1926,
      "step": 276130
    },
    {
      "epoch": 0.109323936831589,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001447722,
      "loss": 3.2281,
      "step": 276140
    },
    {
      "epoch": 0.10934258006774719,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001447702,
      "loss": 3.18,
      "step": 276150
    },
    {
      "epoch": 0.10936122330390537,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001447682,
      "loss": 3.2867,
      "step": 276160
    },
    {
      "epoch": 0.10937986654006356,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014476620000000002,
      "loss": 3.1743,
      "step": 276170
    },
    {
      "epoch": 0.10939850977622174,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001447642,
      "loss": 3.14,
      "step": 276180
    },
    {
      "epoch": 0.10941715301237992,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001447622,
      "loss": 3.1308,
      "step": 276190
    },
    {
      "epoch": 0.10943579624853811,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001447602,
      "loss": 3.2117,
      "step": 276200
    },
    {
      "epoch": 0.1094544394846963,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001447582,
      "loss": 3.1289,
      "step": 276210
    },
    {
      "epoch": 0.10947308272085447,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014475620000000002,
      "loss": 3.1824,
      "step": 276220
    },
    {
      "epoch": 0.10949172595701266,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014475420000000002,
      "loss": 3.0295,
      "step": 276230
    },
    {
      "epoch": 0.10951036919317084,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001447522,
      "loss": 3.2479,
      "step": 276240
    },
    {
      "epoch": 0.10952901242932903,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001447502,
      "loss": 3.0617,
      "step": 276250
    },
    {
      "epoch": 0.10954765566548721,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001447482,
      "loss": 3.1937,
      "step": 276260
    },
    {
      "epoch": 0.10956629890164539,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001447462,
      "loss": 3.1505,
      "step": 276270
    },
    {
      "epoch": 0.10958494213780358,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001447442,
      "loss": 3.0235,
      "step": 276280
    },
    {
      "epoch": 0.10960358537396177,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014474220000000002,
      "loss": 3.1924,
      "step": 276290
    },
    {
      "epoch": 0.10962222861011994,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014474020000000001,
      "loss": 3.1018,
      "step": 276300
    },
    {
      "epoch": 0.10964087184627813,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001447382,
      "loss": 3.1476,
      "step": 276310
    },
    {
      "epoch": 0.1096595150824363,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001447362,
      "loss": 3.1524,
      "step": 276320
    },
    {
      "epoch": 0.1096781583185945,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001447342,
      "loss": 3.1677,
      "step": 276330
    },
    {
      "epoch": 0.10969680155475268,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001447322,
      "loss": 3.0877,
      "step": 276340
    },
    {
      "epoch": 0.10971544479091086,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014473020000000002,
      "loss": 3.2083,
      "step": 276350
    },
    {
      "epoch": 0.10973408802706905,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014472820000000001,
      "loss": 3.0849,
      "step": 276360
    },
    {
      "epoch": 0.10975273126322722,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001447262,
      "loss": 3.1449,
      "step": 276370
    },
    {
      "epoch": 0.10977137449938541,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001447242,
      "loss": 3.1334,
      "step": 276380
    },
    {
      "epoch": 0.1097900177355436,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001447222,
      "loss": 3.0838,
      "step": 276390
    },
    {
      "epoch": 0.10980866097170178,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014472020000000002,
      "loss": 3.158,
      "step": 276400
    },
    {
      "epoch": 0.10982730420785997,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001447182,
      "loss": 3.1005,
      "step": 276410
    },
    {
      "epoch": 0.10984594744401815,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014471620000000002,
      "loss": 3.1591,
      "step": 276420
    },
    {
      "epoch": 0.10986459068017633,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001447142,
      "loss": 3.2489,
      "step": 276430
    },
    {
      "epoch": 0.10988323391633452,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001447122,
      "loss": 3.1551,
      "step": 276440
    },
    {
      "epoch": 0.1099018771524927,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014471020000000003,
      "loss": 3.1914,
      "step": 276450
    },
    {
      "epoch": 0.10992052038865088,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001447082,
      "loss": 3.1788,
      "step": 276460
    },
    {
      "epoch": 0.10993916362480907,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014470620000000002,
      "loss": 3.2238,
      "step": 276470
    },
    {
      "epoch": 0.10995780686096725,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001447042,
      "loss": 3.0451,
      "step": 276480
    },
    {
      "epoch": 0.10997645009712544,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001447022,
      "loss": 3.1312,
      "step": 276490
    },
    {
      "epoch": 0.10999509333328361,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001447002,
      "loss": 3.1695,
      "step": 276500
    },
    {
      "epoch": 0.1100137365694418,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001446982,
      "loss": 3.0735,
      "step": 276510
    },
    {
      "epoch": 0.11003237980559999,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014469620000000003,
      "loss": 3.1279,
      "step": 276520
    },
    {
      "epoch": 0.11005102304175816,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001446942,
      "loss": 3.1765,
      "step": 276530
    },
    {
      "epoch": 0.11006966627791635,
      "grad_norm": 0.75,
      "learning_rate": 0.00014469220000000002,
      "loss": 3.0196,
      "step": 276540
    },
    {
      "epoch": 0.11008830951407454,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014469019999999999,
      "loss": 3.0571,
      "step": 276550
    },
    {
      "epoch": 0.11010695275023272,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001446882,
      "loss": 3.0738,
      "step": 276560
    },
    {
      "epoch": 0.1101255959863909,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001446862,
      "loss": 3.1257,
      "step": 276570
    },
    {
      "epoch": 0.11014423922254908,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001446842,
      "loss": 3.137,
      "step": 276580
    },
    {
      "epoch": 0.11016288245870727,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014468220000000002,
      "loss": 2.9762,
      "step": 276590
    },
    {
      "epoch": 0.11018152569486546,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001446802,
      "loss": 3.1894,
      "step": 276600
    },
    {
      "epoch": 0.11020016893102363,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001446782,
      "loss": 3.0854,
      "step": 276610
    },
    {
      "epoch": 0.11021881216718182,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001446762,
      "loss": 3.0939,
      "step": 276620
    },
    {
      "epoch": 0.11023745540334001,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001446742,
      "loss": 3.0562,
      "step": 276630
    },
    {
      "epoch": 0.11025609863949819,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001446722,
      "loss": 3.2119,
      "step": 276640
    },
    {
      "epoch": 0.11027474187565638,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001446702,
      "loss": 3.1057,
      "step": 276650
    },
    {
      "epoch": 0.11029338511181455,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014466820000000002,
      "loss": 3.0978,
      "step": 276660
    },
    {
      "epoch": 0.11031202834797274,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001446662,
      "loss": 3.1219,
      "step": 276670
    },
    {
      "epoch": 0.11033067158413093,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001446642,
      "loss": 3.2452,
      "step": 276680
    },
    {
      "epoch": 0.1103493148202891,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001446622,
      "loss": 3.1179,
      "step": 276690
    },
    {
      "epoch": 0.1103679580564473,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001446602,
      "loss": 3.0882,
      "step": 276700
    },
    {
      "epoch": 0.11038660129260547,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014465820000000002,
      "loss": 3.0653,
      "step": 276710
    },
    {
      "epoch": 0.11040524452876366,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001446562,
      "loss": 3.2313,
      "step": 276720
    },
    {
      "epoch": 0.11042388776492185,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014465420000000002,
      "loss": 3.1626,
      "step": 276730
    },
    {
      "epoch": 0.11044253100108002,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001446522,
      "loss": 3.1381,
      "step": 276740
    },
    {
      "epoch": 0.11046117423723821,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001446502,
      "loss": 3.3145,
      "step": 276750
    },
    {
      "epoch": 0.1104798174733964,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001446482,
      "loss": 2.9989,
      "step": 276760
    },
    {
      "epoch": 0.11049846070955457,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001446462,
      "loss": 3.2012,
      "step": 276770
    },
    {
      "epoch": 0.11051710394571276,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014464420000000002,
      "loss": 3.1174,
      "step": 276780
    },
    {
      "epoch": 0.11053574718187094,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014464220000000002,
      "loss": 2.9571,
      "step": 276790
    },
    {
      "epoch": 0.11055439041802913,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001446402,
      "loss": 3.1642,
      "step": 276800
    },
    {
      "epoch": 0.11057303365418732,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001446382,
      "loss": 3.0377,
      "step": 276810
    },
    {
      "epoch": 0.11059167689034549,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001446362,
      "loss": 3.2104,
      "step": 276820
    },
    {
      "epoch": 0.11061032012650368,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001446342,
      "loss": 3.0896,
      "step": 276830
    },
    {
      "epoch": 0.11062896336266186,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014463220000000002,
      "loss": 3.1021,
      "step": 276840
    },
    {
      "epoch": 0.11064760659882004,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014463020000000002,
      "loss": 3.0302,
      "step": 276850
    },
    {
      "epoch": 0.11066624983497823,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001446282,
      "loss": 3.1433,
      "step": 276860
    },
    {
      "epoch": 0.11068489307113641,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001446262,
      "loss": 3.1844,
      "step": 276870
    },
    {
      "epoch": 0.1107035363072946,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001446242,
      "loss": 3.2165,
      "step": 276880
    },
    {
      "epoch": 0.11072217954345279,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001446222,
      "loss": 3.2031,
      "step": 276890
    },
    {
      "epoch": 0.11074082277961096,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001446202,
      "loss": 3.1607,
      "step": 276900
    },
    {
      "epoch": 0.11075946601576915,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00014461820000000002,
      "loss": 3.2053,
      "step": 276910
    },
    {
      "epoch": 0.11077810925192733,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001446162,
      "loss": 3.1109,
      "step": 276920
    },
    {
      "epoch": 0.11079675248808551,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001446142,
      "loss": 3.2274,
      "step": 276930
    },
    {
      "epoch": 0.1108153957242437,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001446122,
      "loss": 3.024,
      "step": 276940
    },
    {
      "epoch": 0.11083403896040188,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001446102,
      "loss": 3.0599,
      "step": 276950
    },
    {
      "epoch": 0.11085268219656007,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014460820000000002,
      "loss": 3.1691,
      "step": 276960
    },
    {
      "epoch": 0.11087132543271826,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001446062,
      "loss": 3.1459,
      "step": 276970
    },
    {
      "epoch": 0.11088996866887643,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014460420000000001,
      "loss": 3.0364,
      "step": 276980
    },
    {
      "epoch": 0.11090861190503462,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001446022,
      "loss": 3.2956,
      "step": 276990
    },
    {
      "epoch": 0.1109272551411928,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001446002,
      "loss": 3.0265,
      "step": 277000
    },
    {
      "epoch": 0.11094589837735098,
      "grad_norm": 0.75,
      "learning_rate": 0.00014459820000000003,
      "loss": 3.2057,
      "step": 277010
    },
    {
      "epoch": 0.11096454161350917,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001445962,
      "loss": 3.0807,
      "step": 277020
    },
    {
      "epoch": 0.11098318484966735,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014459420000000002,
      "loss": 3.0692,
      "step": 277030
    },
    {
      "epoch": 0.11100182808582554,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001445922,
      "loss": 3.1308,
      "step": 277040
    },
    {
      "epoch": 0.11102047132198371,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001445902,
      "loss": 3.2122,
      "step": 277050
    },
    {
      "epoch": 0.1110391145581419,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001445882,
      "loss": 3.1074,
      "step": 277060
    },
    {
      "epoch": 0.11105775779430009,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001445862,
      "loss": 3.0769,
      "step": 277070
    },
    {
      "epoch": 0.11107640103045827,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014458420000000002,
      "loss": 3.1889,
      "step": 277080
    },
    {
      "epoch": 0.11109504426661646,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001445822,
      "loss": 3.1178,
      "step": 277090
    },
    {
      "epoch": 0.11111368750277464,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014458020000000002,
      "loss": 3.0835,
      "step": 277100
    },
    {
      "epoch": 0.11113233073893282,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014457819999999998,
      "loss": 3.1816,
      "step": 277110
    },
    {
      "epoch": 0.11115097397509101,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001445762,
      "loss": 3.127,
      "step": 277120
    },
    {
      "epoch": 0.11116961721124918,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001445742,
      "loss": 3.11,
      "step": 277130
    },
    {
      "epoch": 0.11118826044740737,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001445722,
      "loss": 3.1099,
      "step": 277140
    },
    {
      "epoch": 0.11120690368356556,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014457020000000002,
      "loss": 3.1329,
      "step": 277150
    },
    {
      "epoch": 0.11122554691972374,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001445682,
      "loss": 3.0729,
      "step": 277160
    },
    {
      "epoch": 0.11124419015588193,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001445662,
      "loss": 3.1925,
      "step": 277170
    },
    {
      "epoch": 0.1112628333920401,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001445642,
      "loss": 3.1125,
      "step": 277180
    },
    {
      "epoch": 0.11128147662819829,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001445622,
      "loss": 3.1425,
      "step": 277190
    },
    {
      "epoch": 0.11130011986435648,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014456020000000003,
      "loss": 3.0518,
      "step": 277200
    },
    {
      "epoch": 0.11131876310051465,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001445582,
      "loss": 3.0481,
      "step": 277210
    },
    {
      "epoch": 0.11133740633667284,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014455620000000002,
      "loss": 3.0998,
      "step": 277220
    },
    {
      "epoch": 0.11135604957283103,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001445542,
      "loss": 3.1753,
      "step": 277230
    },
    {
      "epoch": 0.1113746928089892,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001445522,
      "loss": 3.116,
      "step": 277240
    },
    {
      "epoch": 0.1113933360451474,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001445502,
      "loss": 3.2533,
      "step": 277250
    },
    {
      "epoch": 0.11141197928130557,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001445482,
      "loss": 3.1163,
      "step": 277260
    },
    {
      "epoch": 0.11143062251746376,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014454620000000002,
      "loss": 3.0528,
      "step": 277270
    },
    {
      "epoch": 0.11144926575362195,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001445442,
      "loss": 3.163,
      "step": 277280
    },
    {
      "epoch": 0.11146790898978012,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001445422,
      "loss": 3.2153,
      "step": 277290
    },
    {
      "epoch": 0.11148655222593831,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001445402,
      "loss": 3.1323,
      "step": 277300
    },
    {
      "epoch": 0.1115051954620965,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001445382,
      "loss": 3.0241,
      "step": 277310
    },
    {
      "epoch": 0.11152383869825468,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001445362,
      "loss": 3.1836,
      "step": 277320
    },
    {
      "epoch": 0.11154248193441287,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001445342,
      "loss": 3.0927,
      "step": 277330
    },
    {
      "epoch": 0.11156112517057104,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014453220000000002,
      "loss": 3.1022,
      "step": 277340
    },
    {
      "epoch": 0.11157976840672923,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014453020000000001,
      "loss": 3.0758,
      "step": 277350
    },
    {
      "epoch": 0.11159841164288742,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001445282,
      "loss": 3.046,
      "step": 277360
    },
    {
      "epoch": 0.1116170548790456,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001445262,
      "loss": 3.1373,
      "step": 277370
    },
    {
      "epoch": 0.11163569811520378,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001445242,
      "loss": 3.2416,
      "step": 277380
    },
    {
      "epoch": 0.11165434135136196,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001445222,
      "loss": 3.096,
      "step": 277390
    },
    {
      "epoch": 0.11167298458752015,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014452020000000002,
      "loss": 3.0621,
      "step": 277400
    },
    {
      "epoch": 0.11169162782367834,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014451820000000002,
      "loss": 3.0834,
      "step": 277410
    },
    {
      "epoch": 0.11171027105983651,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001445162,
      "loss": 3.0567,
      "step": 277420
    },
    {
      "epoch": 0.1117289142959947,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001445142,
      "loss": 3.1894,
      "step": 277430
    },
    {
      "epoch": 0.11174755753215289,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001445122,
      "loss": 3.0482,
      "step": 277440
    },
    {
      "epoch": 0.11176620076831106,
      "grad_norm": 0.625,
      "learning_rate": 0.00014451020000000002,
      "loss": 3.1887,
      "step": 277450
    },
    {
      "epoch": 0.11178484400446925,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001445082,
      "loss": 3.1174,
      "step": 277460
    },
    {
      "epoch": 0.11180348724062743,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014450620000000002,
      "loss": 3.2229,
      "step": 277470
    },
    {
      "epoch": 0.11182213047678562,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001445042,
      "loss": 3.1185,
      "step": 277480
    },
    {
      "epoch": 0.1118407737129438,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001445022,
      "loss": 3.1702,
      "step": 277490
    },
    {
      "epoch": 0.11185941694910198,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001445002,
      "loss": 2.9997,
      "step": 277500
    },
    {
      "epoch": 0.11187806018526017,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001444982,
      "loss": 3.1569,
      "step": 277510
    },
    {
      "epoch": 0.11189670342141836,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014449620000000002,
      "loss": 3.1819,
      "step": 277520
    },
    {
      "epoch": 0.11191534665757653,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001444942,
      "loss": 3.1096,
      "step": 277530
    },
    {
      "epoch": 0.11193398989373472,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001444922,
      "loss": 3.1495,
      "step": 277540
    },
    {
      "epoch": 0.1119526331298929,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001444902,
      "loss": 3.1334,
      "step": 277550
    },
    {
      "epoch": 0.11197127636605109,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001444882,
      "loss": 2.9579,
      "step": 277560
    },
    {
      "epoch": 0.11198991960220928,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014448620000000003,
      "loss": 3.1206,
      "step": 277570
    },
    {
      "epoch": 0.11200856283836745,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001444842,
      "loss": 3.0833,
      "step": 277580
    },
    {
      "epoch": 0.11202720607452564,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014448220000000002,
      "loss": 3.108,
      "step": 277590
    },
    {
      "epoch": 0.11204584931068381,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014448019999999999,
      "loss": 3.2038,
      "step": 277600
    },
    {
      "epoch": 0.112064492546842,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001444782,
      "loss": 3.1745,
      "step": 277610
    },
    {
      "epoch": 0.11208313578300019,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001444762,
      "loss": 3.1341,
      "step": 277620
    },
    {
      "epoch": 0.11210177901915837,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001444742,
      "loss": 3.0419,
      "step": 277630
    },
    {
      "epoch": 0.11212042225531656,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014447220000000002,
      "loss": 3.1061,
      "step": 277640
    },
    {
      "epoch": 0.11213906549147475,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001444702,
      "loss": 3.1058,
      "step": 277650
    },
    {
      "epoch": 0.11215770872763292,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014446820000000001,
      "loss": 3.1493,
      "step": 277660
    },
    {
      "epoch": 0.11217635196379111,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001444662,
      "loss": 3.1364,
      "step": 277670
    },
    {
      "epoch": 0.11219499519994929,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001444642,
      "loss": 3.0712,
      "step": 277680
    },
    {
      "epoch": 0.11221363843610747,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001444622,
      "loss": 3.2937,
      "step": 277690
    },
    {
      "epoch": 0.11223228167226566,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001444602,
      "loss": 2.9502,
      "step": 277700
    },
    {
      "epoch": 0.11225092490842384,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014445820000000002,
      "loss": 3.1138,
      "step": 277710
    },
    {
      "epoch": 0.11226956814458203,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001444562,
      "loss": 3.0911,
      "step": 277720
    },
    {
      "epoch": 0.1122882113807402,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001444542,
      "loss": 3.2533,
      "step": 277730
    },
    {
      "epoch": 0.11230685461689839,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001444522,
      "loss": 3.2675,
      "step": 277740
    },
    {
      "epoch": 0.11232549785305658,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001444502,
      "loss": 2.9138,
      "step": 277750
    },
    {
      "epoch": 0.11234414108921476,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014444820000000002,
      "loss": 3.057,
      "step": 277760
    },
    {
      "epoch": 0.11236278432537294,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001444462,
      "loss": 3.222,
      "step": 277770
    },
    {
      "epoch": 0.11238142756153113,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014444420000000002,
      "loss": 3.1999,
      "step": 277780
    },
    {
      "epoch": 0.11240007079768931,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001444422,
      "loss": 3.1083,
      "step": 277790
    },
    {
      "epoch": 0.1124187140338475,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001444402,
      "loss": 3.1042,
      "step": 277800
    },
    {
      "epoch": 0.11243735727000567,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001444382,
      "loss": 3.2117,
      "step": 277810
    },
    {
      "epoch": 0.11245600050616386,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001444362,
      "loss": 2.9784,
      "step": 277820
    },
    {
      "epoch": 0.11247464374232205,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014443420000000002,
      "loss": 3.1644,
      "step": 277830
    },
    {
      "epoch": 0.11249328697848023,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014443220000000002,
      "loss": 3.1976,
      "step": 277840
    },
    {
      "epoch": 0.11251193021463841,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001444302,
      "loss": 3.0357,
      "step": 277850
    },
    {
      "epoch": 0.1125305734507966,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001444282,
      "loss": 3.0726,
      "step": 277860
    },
    {
      "epoch": 0.11254921668695478,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001444262,
      "loss": 3.0672,
      "step": 277870
    },
    {
      "epoch": 0.11256785992311297,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001444242,
      "loss": 3.1544,
      "step": 277880
    },
    {
      "epoch": 0.11258650315927114,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001444222,
      "loss": 3.1158,
      "step": 277890
    },
    {
      "epoch": 0.11260514639542933,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014442020000000002,
      "loss": 3.1185,
      "step": 277900
    },
    {
      "epoch": 0.11262378963158752,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001444182,
      "loss": 3.1876,
      "step": 277910
    },
    {
      "epoch": 0.1126424328677457,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001444162,
      "loss": 3.1884,
      "step": 277920
    },
    {
      "epoch": 0.11266107610390388,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001444142,
      "loss": 3.1368,
      "step": 277930
    },
    {
      "epoch": 0.11267971934006206,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001444122,
      "loss": 3.0776,
      "step": 277940
    },
    {
      "epoch": 0.11269836257622025,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001444102,
      "loss": 3.0006,
      "step": 277950
    },
    {
      "epoch": 0.11271700581237844,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014440820000000002,
      "loss": 3.0938,
      "step": 277960
    },
    {
      "epoch": 0.11273564904853661,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001444062,
      "loss": 3.1806,
      "step": 277970
    },
    {
      "epoch": 0.1127542922846948,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001444042,
      "loss": 3.174,
      "step": 277980
    },
    {
      "epoch": 0.11277293552085299,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001444022,
      "loss": 3.2073,
      "step": 277990
    },
    {
      "epoch": 0.11279157875701117,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001444002,
      "loss": 3.0329,
      "step": 278000
    },
    {
      "epoch": 0.11281022199316935,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014439820000000002,
      "loss": 3.1905,
      "step": 278010
    },
    {
      "epoch": 0.11282886522932753,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001443962,
      "loss": 3.096,
      "step": 278020
    },
    {
      "epoch": 0.11284750846548572,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00014439420000000001,
      "loss": 3.1323,
      "step": 278030
    },
    {
      "epoch": 0.11286615170164391,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001443922,
      "loss": 3.0385,
      "step": 278040
    },
    {
      "epoch": 0.11288479493780208,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001443902,
      "loss": 3.0848,
      "step": 278050
    },
    {
      "epoch": 0.11290343817396027,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014438820000000003,
      "loss": 3.1957,
      "step": 278060
    },
    {
      "epoch": 0.11292208141011845,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001443862,
      "loss": 3.1275,
      "step": 278070
    },
    {
      "epoch": 0.11294072464627664,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014438420000000002,
      "loss": 3.1146,
      "step": 278080
    },
    {
      "epoch": 0.11295936788243482,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001443822,
      "loss": 3.0906,
      "step": 278090
    },
    {
      "epoch": 0.112978011118593,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001443802,
      "loss": 3.1744,
      "step": 278100
    },
    {
      "epoch": 0.11299665435475119,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001443782,
      "loss": 3.1366,
      "step": 278110
    },
    {
      "epoch": 0.11301529759090938,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001443762,
      "loss": 3.1988,
      "step": 278120
    },
    {
      "epoch": 0.11303394082706755,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014437420000000002,
      "loss": 3.1701,
      "step": 278130
    },
    {
      "epoch": 0.11305258406322574,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001443722,
      "loss": 3.2207,
      "step": 278140
    },
    {
      "epoch": 0.11307122729938392,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014437020000000002,
      "loss": 3.1153,
      "step": 278150
    },
    {
      "epoch": 0.1130898705355421,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001443682,
      "loss": 3.0076,
      "step": 278160
    },
    {
      "epoch": 0.1131085137717003,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001443662,
      "loss": 3.1237,
      "step": 278170
    },
    {
      "epoch": 0.11312715700785847,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001443642,
      "loss": 3.1248,
      "step": 278180
    },
    {
      "epoch": 0.11314580024401666,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001443622,
      "loss": 3.0173,
      "step": 278190
    },
    {
      "epoch": 0.11316444348017485,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014436020000000002,
      "loss": 3.0772,
      "step": 278200
    },
    {
      "epoch": 0.11318308671633302,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001443582,
      "loss": 3.0283,
      "step": 278210
    },
    {
      "epoch": 0.11320172995249121,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001443562,
      "loss": 3.101,
      "step": 278220
    },
    {
      "epoch": 0.11322037318864939,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001443542,
      "loss": 3.2077,
      "step": 278230
    },
    {
      "epoch": 0.11323901642480758,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001443522,
      "loss": 3.1138,
      "step": 278240
    },
    {
      "epoch": 0.11325765966096576,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014435020000000003,
      "loss": 3.0338,
      "step": 278250
    },
    {
      "epoch": 0.11327630289712394,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001443482,
      "loss": 3.0949,
      "step": 278260
    },
    {
      "epoch": 0.11329494613328213,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014434620000000002,
      "loss": 3.2346,
      "step": 278270
    },
    {
      "epoch": 0.1133135893694403,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014434419999999999,
      "loss": 3.1982,
      "step": 278280
    },
    {
      "epoch": 0.11333223260559849,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001443422,
      "loss": 3.0896,
      "step": 278290
    },
    {
      "epoch": 0.11335087584175668,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001443402,
      "loss": 3.2155,
      "step": 278300
    },
    {
      "epoch": 0.11336951907791486,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001443382,
      "loss": 3.223,
      "step": 278310
    },
    {
      "epoch": 0.11338816231407305,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014433620000000002,
      "loss": 3.1729,
      "step": 278320
    },
    {
      "epoch": 0.11340680555023123,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001443342,
      "loss": 3.2009,
      "step": 278330
    },
    {
      "epoch": 0.11342544878638941,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014433220000000001,
      "loss": 3.1221,
      "step": 278340
    },
    {
      "epoch": 0.1134440920225476,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001443302,
      "loss": 3.095,
      "step": 278350
    },
    {
      "epoch": 0.11346273525870577,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001443282,
      "loss": 3.3123,
      "step": 278360
    },
    {
      "epoch": 0.11348137849486396,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001443262,
      "loss": 3.0768,
      "step": 278370
    },
    {
      "epoch": 0.11350002173102215,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001443242,
      "loss": 3.1148,
      "step": 278380
    },
    {
      "epoch": 0.11351866496718033,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014432220000000002,
      "loss": 3.2455,
      "step": 278390
    },
    {
      "epoch": 0.11353730820333852,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014432020000000001,
      "loss": 3.1468,
      "step": 278400
    },
    {
      "epoch": 0.11355595143949669,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001443182,
      "loss": 3.1408,
      "step": 278410
    },
    {
      "epoch": 0.11357459467565488,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001443162,
      "loss": 3.0897,
      "step": 278420
    },
    {
      "epoch": 0.11359323791181307,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001443142,
      "loss": 3.1438,
      "step": 278430
    },
    {
      "epoch": 0.11361188114797124,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001443122,
      "loss": 3.037,
      "step": 278440
    },
    {
      "epoch": 0.11363052438412943,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014431020000000002,
      "loss": 3.1008,
      "step": 278450
    },
    {
      "epoch": 0.11364916762028762,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014430820000000002,
      "loss": 3.1703,
      "step": 278460
    },
    {
      "epoch": 0.1136678108564458,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001443062,
      "loss": 3.1108,
      "step": 278470
    },
    {
      "epoch": 0.11368645409260399,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001443042,
      "loss": 3.1167,
      "step": 278480
    },
    {
      "epoch": 0.11370509732876216,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001443022,
      "loss": 3.1875,
      "step": 278490
    },
    {
      "epoch": 0.11372374056492035,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001443002,
      "loss": 3.0867,
      "step": 278500
    },
    {
      "epoch": 0.11374238380107854,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001442982,
      "loss": 3.1389,
      "step": 278510
    },
    {
      "epoch": 0.11376102703723671,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00014429620000000002,
      "loss": 3.2872,
      "step": 278520
    },
    {
      "epoch": 0.1137796702733949,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001442942,
      "loss": 3.1171,
      "step": 278530
    },
    {
      "epoch": 0.11379831350955309,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001442922,
      "loss": 3.121,
      "step": 278540
    },
    {
      "epoch": 0.11381695674571127,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001442902,
      "loss": 3.1882,
      "step": 278550
    },
    {
      "epoch": 0.11383559998186946,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001442882,
      "loss": 3.1251,
      "step": 278560
    },
    {
      "epoch": 0.11385424321802763,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014428620000000002,
      "loss": 2.9792,
      "step": 278570
    },
    {
      "epoch": 0.11387288645418582,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001442842,
      "loss": 3.144,
      "step": 278580
    },
    {
      "epoch": 0.11389152969034401,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001442822,
      "loss": 3.162,
      "step": 278590
    },
    {
      "epoch": 0.11391017292650218,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001442802,
      "loss": 3.1743,
      "step": 278600
    },
    {
      "epoch": 0.11392881616266037,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001442782,
      "loss": 3.1455,
      "step": 278610
    },
    {
      "epoch": 0.11394745939881855,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014427620000000003,
      "loss": 3.1204,
      "step": 278620
    },
    {
      "epoch": 0.11396610263497674,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001442742,
      "loss": 3.2228,
      "step": 278630
    },
    {
      "epoch": 0.11398474587113493,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014427220000000002,
      "loss": 3.0463,
      "step": 278640
    },
    {
      "epoch": 0.1140033891072931,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014427019999999999,
      "loss": 3.1225,
      "step": 278650
    },
    {
      "epoch": 0.11402203234345129,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001442682,
      "loss": 3.0933,
      "step": 278660
    },
    {
      "epoch": 0.11404067557960948,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001442662,
      "loss": 2.9152,
      "step": 278670
    },
    {
      "epoch": 0.11405931881576765,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001442642,
      "loss": 3.0844,
      "step": 278680
    },
    {
      "epoch": 0.11407796205192584,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014426220000000002,
      "loss": 3.004,
      "step": 278690
    },
    {
      "epoch": 0.11409660528808402,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001442602,
      "loss": 3.0822,
      "step": 278700
    },
    {
      "epoch": 0.11411524852424221,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014425820000000001,
      "loss": 3.1275,
      "step": 278710
    },
    {
      "epoch": 0.1141338917604004,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001442562,
      "loss": 3.0722,
      "step": 278720
    },
    {
      "epoch": 0.11415253499655857,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001442542,
      "loss": 3.1249,
      "step": 278730
    },
    {
      "epoch": 0.11417117823271676,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014425220000000003,
      "loss": 3.1363,
      "step": 278740
    },
    {
      "epoch": 0.11418982146887494,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001442502,
      "loss": 3.2292,
      "step": 278750
    },
    {
      "epoch": 0.11420846470503312,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014424820000000002,
      "loss": 3.1171,
      "step": 278760
    },
    {
      "epoch": 0.11422710794119131,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001442462,
      "loss": 3.1206,
      "step": 278770
    },
    {
      "epoch": 0.11424575117734949,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001442442,
      "loss": 3.1359,
      "step": 278780
    },
    {
      "epoch": 0.11426439441350768,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001442422,
      "loss": 3.2195,
      "step": 278790
    },
    {
      "epoch": 0.11428303764966587,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001442402,
      "loss": 3.1066,
      "step": 278800
    },
    {
      "epoch": 0.11430168088582404,
      "grad_norm": 0.609375,
      "learning_rate": 0.00014423820000000002,
      "loss": 3.0696,
      "step": 278810
    },
    {
      "epoch": 0.11432032412198223,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001442362,
      "loss": 3.1783,
      "step": 278820
    },
    {
      "epoch": 0.1143389673581404,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014423420000000002,
      "loss": 2.989,
      "step": 278830
    },
    {
      "epoch": 0.1143576105942986,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001442322,
      "loss": 3.1032,
      "step": 278840
    },
    {
      "epoch": 0.11437625383045678,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001442302,
      "loss": 3.1382,
      "step": 278850
    },
    {
      "epoch": 0.11439489706661496,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001442282,
      "loss": 3.1163,
      "step": 278860
    },
    {
      "epoch": 0.11441354030277315,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001442262,
      "loss": 3.0037,
      "step": 278870
    },
    {
      "epoch": 0.11443218353893134,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014422420000000002,
      "loss": 3.1428,
      "step": 278880
    },
    {
      "epoch": 0.11445082677508951,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001442222,
      "loss": 2.9842,
      "step": 278890
    },
    {
      "epoch": 0.1144694700112477,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001442202,
      "loss": 3.1726,
      "step": 278900
    },
    {
      "epoch": 0.11448811324740588,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001442182,
      "loss": 3.0858,
      "step": 278910
    },
    {
      "epoch": 0.11450675648356406,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001442162,
      "loss": 3.1424,
      "step": 278920
    },
    {
      "epoch": 0.11452539971972225,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001442142,
      "loss": 3.158,
      "step": 278930
    },
    {
      "epoch": 0.11454404295588043,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001442122,
      "loss": 3.0628,
      "step": 278940
    },
    {
      "epoch": 0.11456268619203862,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014421020000000002,
      "loss": 3.121,
      "step": 278950
    },
    {
      "epoch": 0.1145813294281968,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001442082,
      "loss": 2.9712,
      "step": 278960
    },
    {
      "epoch": 0.11459997266435498,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001442062,
      "loss": 3.1452,
      "step": 278970
    },
    {
      "epoch": 0.11461861590051317,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001442042,
      "loss": 3.2297,
      "step": 278980
    },
    {
      "epoch": 0.11463725913667135,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001442022,
      "loss": 3.261,
      "step": 278990
    },
    {
      "epoch": 0.11465590237282954,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001442002,
      "loss": 3.1017,
      "step": 279000
    },
    {
      "epoch": 0.11467454560898772,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014419820000000002,
      "loss": 3.2769,
      "step": 279010
    },
    {
      "epoch": 0.1146931888451459,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014419620000000001,
      "loss": 3.1248,
      "step": 279020
    },
    {
      "epoch": 0.11471183208130409,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001441942,
      "loss": 3.2443,
      "step": 279030
    },
    {
      "epoch": 0.11473047531746226,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001441922,
      "loss": 3.2295,
      "step": 279040
    },
    {
      "epoch": 0.11474911855362045,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001441902,
      "loss": 3.1133,
      "step": 279050
    },
    {
      "epoch": 0.11476776178977864,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014418820000000002,
      "loss": 3.0414,
      "step": 279060
    },
    {
      "epoch": 0.11478640502593682,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001441862,
      "loss": 3.0812,
      "step": 279070
    },
    {
      "epoch": 0.114805048262095,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014418420000000001,
      "loss": 3.1293,
      "step": 279080
    },
    {
      "epoch": 0.11482369149825318,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001441822,
      "loss": 3.2132,
      "step": 279090
    },
    {
      "epoch": 0.11484233473441137,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001441802,
      "loss": 3.0521,
      "step": 279100
    },
    {
      "epoch": 0.11486097797056956,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001441782,
      "loss": 3.1331,
      "step": 279110
    },
    {
      "epoch": 0.11487962120672773,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001441762,
      "loss": 3.1895,
      "step": 279120
    },
    {
      "epoch": 0.11489826444288592,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014417420000000002,
      "loss": 3.155,
      "step": 279130
    },
    {
      "epoch": 0.11491690767904411,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001441722,
      "loss": 3.078,
      "step": 279140
    },
    {
      "epoch": 0.11493555091520229,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001441702,
      "loss": 3.0033,
      "step": 279150
    },
    {
      "epoch": 0.11495419415136048,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001441682,
      "loss": 3.1303,
      "step": 279160
    },
    {
      "epoch": 0.11497283738751865,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001441662,
      "loss": 3.1406,
      "step": 279170
    },
    {
      "epoch": 0.11499148062367684,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014416420000000002,
      "loss": 3.0655,
      "step": 279180
    },
    {
      "epoch": 0.11501012385983503,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001441622,
      "loss": 3.0613,
      "step": 279190
    },
    {
      "epoch": 0.1150287670959932,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014416020000000002,
      "loss": 3.1501,
      "step": 279200
    },
    {
      "epoch": 0.11504741033215139,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001441582,
      "loss": 3.2407,
      "step": 279210
    },
    {
      "epoch": 0.11506605356830958,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001441562,
      "loss": 3.1433,
      "step": 279220
    },
    {
      "epoch": 0.11508469680446776,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014415420000000003,
      "loss": 3.2144,
      "step": 279230
    },
    {
      "epoch": 0.11510334004062595,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001441522,
      "loss": 3.1124,
      "step": 279240
    },
    {
      "epoch": 0.11512198327678412,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014415020000000002,
      "loss": 3.0421,
      "step": 279250
    },
    {
      "epoch": 0.11514062651294231,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001441482,
      "loss": 3.1677,
      "step": 279260
    },
    {
      "epoch": 0.1151592697491005,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001441462,
      "loss": 3.2051,
      "step": 279270
    },
    {
      "epoch": 0.11517791298525867,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001441442,
      "loss": 3.143,
      "step": 279280
    },
    {
      "epoch": 0.11519655622141686,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001441422,
      "loss": 3.1902,
      "step": 279290
    },
    {
      "epoch": 0.11521519945757504,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014414020000000003,
      "loss": 3.1557,
      "step": 279300
    },
    {
      "epoch": 0.11523384269373323,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001441382,
      "loss": 3.0843,
      "step": 279310
    },
    {
      "epoch": 0.11525248592989142,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014413620000000002,
      "loss": 3.0954,
      "step": 279320
    },
    {
      "epoch": 0.11527112916604959,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014413419999999999,
      "loss": 3.1623,
      "step": 279330
    },
    {
      "epoch": 0.11528977240220778,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001441322,
      "loss": 3.1405,
      "step": 279340
    },
    {
      "epoch": 0.11530841563836597,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001441302,
      "loss": 3.2228,
      "step": 279350
    },
    {
      "epoch": 0.11532705887452414,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001441282,
      "loss": 3.0923,
      "step": 279360
    },
    {
      "epoch": 0.11534570211068233,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014412620000000002,
      "loss": 2.9461,
      "step": 279370
    },
    {
      "epoch": 0.11536434534684051,
      "grad_norm": 0.75,
      "learning_rate": 0.0001441242,
      "loss": 3.1956,
      "step": 279380
    },
    {
      "epoch": 0.1153829885829987,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014412220000000001,
      "loss": 3.0324,
      "step": 279390
    },
    {
      "epoch": 0.11540163181915689,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001441202,
      "loss": 3.138,
      "step": 279400
    },
    {
      "epoch": 0.11542027505531506,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001441182,
      "loss": 3.1509,
      "step": 279410
    },
    {
      "epoch": 0.11543891829147325,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001441162,
      "loss": 3.0636,
      "step": 279420
    },
    {
      "epoch": 0.11545756152763142,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001441142,
      "loss": 3.0693,
      "step": 279430
    },
    {
      "epoch": 0.11547620476378961,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014411220000000002,
      "loss": 3.1498,
      "step": 279440
    },
    {
      "epoch": 0.1154948479999478,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014411020000000001,
      "loss": 3.2051,
      "step": 279450
    },
    {
      "epoch": 0.11551349123610598,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001441082,
      "loss": 3.058,
      "step": 279460
    },
    {
      "epoch": 0.11553213447226417,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001441062,
      "loss": 3.1576,
      "step": 279470
    },
    {
      "epoch": 0.11555077770842236,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001441042,
      "loss": 3.1999,
      "step": 279480
    },
    {
      "epoch": 0.11556942094458053,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001441022,
      "loss": 3.2542,
      "step": 279490
    },
    {
      "epoch": 0.11558806418073872,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001441002,
      "loss": 3.1688,
      "step": 279500
    },
    {
      "epoch": 0.1156067074168969,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014409820000000002,
      "loss": 3.1589,
      "step": 279510
    },
    {
      "epoch": 0.11562535065305508,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001440962,
      "loss": 3.1679,
      "step": 279520
    },
    {
      "epoch": 0.11564399388921327,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001440942,
      "loss": 3.1517,
      "step": 279530
    },
    {
      "epoch": 0.11566263712537145,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001440922,
      "loss": 3.1684,
      "step": 279540
    },
    {
      "epoch": 0.11568128036152964,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001440902,
      "loss": 3.2127,
      "step": 279550
    },
    {
      "epoch": 0.11569992359768783,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001440882,
      "loss": 3.1993,
      "step": 279560
    },
    {
      "epoch": 0.115718566833846,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014408620000000002,
      "loss": 3.2521,
      "step": 279570
    },
    {
      "epoch": 0.11573721007000419,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001440842,
      "loss": 3.1345,
      "step": 279580
    },
    {
      "epoch": 0.11575585330616237,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001440822,
      "loss": 3.086,
      "step": 279590
    },
    {
      "epoch": 0.11577449654232055,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001440802,
      "loss": 3.101,
      "step": 279600
    },
    {
      "epoch": 0.11579313977847874,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001440782,
      "loss": 3.2546,
      "step": 279610
    },
    {
      "epoch": 0.11581178301463692,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014407620000000002,
      "loss": 3.142,
      "step": 279620
    },
    {
      "epoch": 0.11583042625079511,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001440742,
      "loss": 3.189,
      "step": 279630
    },
    {
      "epoch": 0.11584906948695328,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001440722,
      "loss": 3.1598,
      "step": 279640
    },
    {
      "epoch": 0.11586771272311147,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001440702,
      "loss": 3.1669,
      "step": 279650
    },
    {
      "epoch": 0.11588635595926966,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001440682,
      "loss": 3.1627,
      "step": 279660
    },
    {
      "epoch": 0.11590499919542784,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014406620000000003,
      "loss": 3.0988,
      "step": 279670
    },
    {
      "epoch": 0.11592364243158602,
      "grad_norm": 0.75,
      "learning_rate": 0.0001440642,
      "loss": 3.2075,
      "step": 279680
    },
    {
      "epoch": 0.11594228566774421,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014406220000000002,
      "loss": 3.0752,
      "step": 279690
    },
    {
      "epoch": 0.11596092890390239,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014406020000000001,
      "loss": 3.0978,
      "step": 279700
    },
    {
      "epoch": 0.11597957214006058,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001440582,
      "loss": 3.2333,
      "step": 279710
    },
    {
      "epoch": 0.11599821537621875,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001440562,
      "loss": 2.9776,
      "step": 279720
    },
    {
      "epoch": 0.11601685861237694,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001440542,
      "loss": 3.0895,
      "step": 279730
    },
    {
      "epoch": 0.11603550184853513,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014405220000000002,
      "loss": 3.1289,
      "step": 279740
    },
    {
      "epoch": 0.1160541450846933,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001440502,
      "loss": 3.0008,
      "step": 279750
    },
    {
      "epoch": 0.1160727883208515,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014404820000000001,
      "loss": 3.1079,
      "step": 279760
    },
    {
      "epoch": 0.11609143155700967,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001440462,
      "loss": 3.1084,
      "step": 279770
    },
    {
      "epoch": 0.11611007479316786,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001440442,
      "loss": 3.1379,
      "step": 279780
    },
    {
      "epoch": 0.11612871802932605,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014404220000000003,
      "loss": 3.1416,
      "step": 279790
    },
    {
      "epoch": 0.11614736126548422,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001440402,
      "loss": 3.1266,
      "step": 279800
    },
    {
      "epoch": 0.11616600450164241,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014403820000000002,
      "loss": 3.0058,
      "step": 279810
    },
    {
      "epoch": 0.1161846477378006,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001440362,
      "loss": 3.1557,
      "step": 279820
    },
    {
      "epoch": 0.11620329097395878,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001440342,
      "loss": 3.1221,
      "step": 279830
    },
    {
      "epoch": 0.11622193421011696,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001440322,
      "loss": 3.095,
      "step": 279840
    },
    {
      "epoch": 0.11624057744627514,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001440302,
      "loss": 3.1109,
      "step": 279850
    },
    {
      "epoch": 0.11625922068243333,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014402820000000002,
      "loss": 3.1477,
      "step": 279860
    },
    {
      "epoch": 0.11627786391859152,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001440262,
      "loss": 3.0289,
      "step": 279870
    },
    {
      "epoch": 0.11629650715474969,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014402420000000002,
      "loss": 3.1319,
      "step": 279880
    },
    {
      "epoch": 0.11631515039090788,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014402219999999998,
      "loss": 3.0762,
      "step": 279890
    },
    {
      "epoch": 0.11633379362706607,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001440202,
      "loss": 3.0246,
      "step": 279900
    },
    {
      "epoch": 0.11635243686322425,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001440182,
      "loss": 3.1139,
      "step": 279910
    },
    {
      "epoch": 0.11637108009938243,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001440162,
      "loss": 3.1712,
      "step": 279920
    },
    {
      "epoch": 0.11638972333554061,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014401420000000002,
      "loss": 3.1409,
      "step": 279930
    },
    {
      "epoch": 0.1164083665716988,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001440122,
      "loss": 3.0885,
      "step": 279940
    },
    {
      "epoch": 0.11642700980785699,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001440102,
      "loss": 2.9535,
      "step": 279950
    },
    {
      "epoch": 0.11644565304401516,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001440082,
      "loss": 3.1826,
      "step": 279960
    },
    {
      "epoch": 0.11646429628017335,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001440062,
      "loss": 3.1371,
      "step": 279970
    },
    {
      "epoch": 0.11648293951633153,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001440042,
      "loss": 3.0966,
      "step": 279980
    },
    {
      "epoch": 0.11650158275248972,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001440022,
      "loss": 3.0745,
      "step": 279990
    },
    {
      "epoch": 0.1165202259886479,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014400020000000002,
      "loss": 3.2174,
      "step": 280000
    },
    {
      "epoch": 0.11653886922480608,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001439982,
      "loss": 3.0428,
      "step": 280010
    },
    {
      "epoch": 0.11655751246096427,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001439962,
      "loss": 3.0133,
      "step": 280020
    },
    {
      "epoch": 0.11657615569712246,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001439942,
      "loss": 3.2318,
      "step": 280030
    },
    {
      "epoch": 0.11659479893328063,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001439922,
      "loss": 3.1263,
      "step": 280040
    },
    {
      "epoch": 0.11661344216943882,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001439902,
      "loss": 3.1123,
      "step": 280050
    },
    {
      "epoch": 0.116632085405597,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014398820000000002,
      "loss": 3.1797,
      "step": 280060
    },
    {
      "epoch": 0.11665072864175519,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014398620000000001,
      "loss": 3.1477,
      "step": 280070
    },
    {
      "epoch": 0.11666937187791337,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001439842,
      "loss": 3.1653,
      "step": 280080
    },
    {
      "epoch": 0.11668801511407155,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001439822,
      "loss": 3.1678,
      "step": 280090
    },
    {
      "epoch": 0.11670665835022974,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001439802,
      "loss": 3.0498,
      "step": 280100
    },
    {
      "epoch": 0.11672530158638791,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001439782,
      "loss": 3.1301,
      "step": 280110
    },
    {
      "epoch": 0.1167439448225461,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001439762,
      "loss": 2.9849,
      "step": 280120
    },
    {
      "epoch": 0.11676258805870429,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014397420000000001,
      "loss": 3.0506,
      "step": 280130
    },
    {
      "epoch": 0.11678123129486247,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001439722,
      "loss": 3.1799,
      "step": 280140
    },
    {
      "epoch": 0.11679987453102066,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001439702,
      "loss": 3.1179,
      "step": 280150
    },
    {
      "epoch": 0.11681851776717884,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001439682,
      "loss": 3.1283,
      "step": 280160
    },
    {
      "epoch": 0.11683716100333702,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001439662,
      "loss": 3.0596,
      "step": 280170
    },
    {
      "epoch": 0.11685580423949521,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014396420000000002,
      "loss": 3.1796,
      "step": 280180
    },
    {
      "epoch": 0.11687444747565338,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014396220000000002,
      "loss": 3.0317,
      "step": 280190
    },
    {
      "epoch": 0.11689309071181157,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001439602,
      "loss": 3.2624,
      "step": 280200
    },
    {
      "epoch": 0.11691173394796976,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001439582,
      "loss": 3.2037,
      "step": 280210
    },
    {
      "epoch": 0.11693037718412794,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001439562,
      "loss": 3.1913,
      "step": 280220
    },
    {
      "epoch": 0.11694902042028613,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014395420000000003,
      "loss": 3.1833,
      "step": 280230
    },
    {
      "epoch": 0.11696766365644432,
      "grad_norm": 0.875,
      "learning_rate": 0.0001439522,
      "loss": 2.987,
      "step": 280240
    },
    {
      "epoch": 0.11698630689260249,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014395020000000002,
      "loss": 3.1392,
      "step": 280250
    },
    {
      "epoch": 0.11700495012876068,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001439482,
      "loss": 3.0368,
      "step": 280260
    },
    {
      "epoch": 0.11702359336491885,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001439462,
      "loss": 3.1081,
      "step": 280270
    },
    {
      "epoch": 0.11704223660107704,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014394420000000003,
      "loss": 3.0864,
      "step": 280280
    },
    {
      "epoch": 0.11706087983723523,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001439422,
      "loss": 3.0184,
      "step": 280290
    },
    {
      "epoch": 0.11707952307339341,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014394020000000002,
      "loss": 3.0187,
      "step": 280300
    },
    {
      "epoch": 0.1170981663095516,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001439382,
      "loss": 3.1871,
      "step": 280310
    },
    {
      "epoch": 0.11711680954570977,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001439362,
      "loss": 3.1388,
      "step": 280320
    },
    {
      "epoch": 0.11713545278186796,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001439342,
      "loss": 3.2282,
      "step": 280330
    },
    {
      "epoch": 0.11715409601802615,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001439322,
      "loss": 3.1816,
      "step": 280340
    },
    {
      "epoch": 0.11717273925418432,
      "grad_norm": 0.859375,
      "learning_rate": 0.00014393020000000003,
      "loss": 3.1395,
      "step": 280350
    },
    {
      "epoch": 0.11719138249034251,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001439282,
      "loss": 3.2427,
      "step": 280360
    },
    {
      "epoch": 0.1172100257265007,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014392620000000002,
      "loss": 3.2492,
      "step": 280370
    },
    {
      "epoch": 0.11722866896265888,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001439242,
      "loss": 3.2576,
      "step": 280380
    },
    {
      "epoch": 0.11724731219881707,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001439222,
      "loss": 3.0929,
      "step": 280390
    },
    {
      "epoch": 0.11726595543497524,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001439202,
      "loss": 3.2362,
      "step": 280400
    },
    {
      "epoch": 0.11728459867113343,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001439182,
      "loss": 3.0398,
      "step": 280410
    },
    {
      "epoch": 0.11730324190729162,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014391620000000002,
      "loss": 3.2626,
      "step": 280420
    },
    {
      "epoch": 0.1173218851434498,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001439142,
      "loss": 3.0539,
      "step": 280430
    },
    {
      "epoch": 0.11734052837960798,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014391220000000001,
      "loss": 3.15,
      "step": 280440
    },
    {
      "epoch": 0.11735917161576616,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001439102,
      "loss": 3.0314,
      "step": 280450
    },
    {
      "epoch": 0.11737781485192435,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001439082,
      "loss": 3.044,
      "step": 280460
    },
    {
      "epoch": 0.11739645808808254,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001439062,
      "loss": 3.1958,
      "step": 280470
    },
    {
      "epoch": 0.11741510132424071,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001439042,
      "loss": 3.2418,
      "step": 280480
    },
    {
      "epoch": 0.1174337445603989,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014390220000000002,
      "loss": 3.0374,
      "step": 280490
    },
    {
      "epoch": 0.11745238779655709,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001439002,
      "loss": 3.0664,
      "step": 280500
    },
    {
      "epoch": 0.11747103103271526,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001438982,
      "loss": 3.322,
      "step": 280510
    },
    {
      "epoch": 0.11748967426887345,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001438962,
      "loss": 3.1717,
      "step": 280520
    },
    {
      "epoch": 0.11750831750503163,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001438942,
      "loss": 3.2375,
      "step": 280530
    },
    {
      "epoch": 0.11752696074118982,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001438922,
      "loss": 3.1798,
      "step": 280540
    },
    {
      "epoch": 0.117545603977348,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001438902,
      "loss": 3.1387,
      "step": 280550
    },
    {
      "epoch": 0.11756424721350618,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014388820000000002,
      "loss": 3.1449,
      "step": 280560
    },
    {
      "epoch": 0.11758289044966437,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001438862,
      "loss": 3.2897,
      "step": 280570
    },
    {
      "epoch": 0.11760153368582256,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001438842,
      "loss": 2.8913,
      "step": 280580
    },
    {
      "epoch": 0.11762017692198073,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001438822,
      "loss": 3.0473,
      "step": 280590
    },
    {
      "epoch": 0.11763882015813892,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001438802,
      "loss": 3.0393,
      "step": 280600
    },
    {
      "epoch": 0.1176574633942971,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001438782,
      "loss": 3.235,
      "step": 280610
    },
    {
      "epoch": 0.11767610663045529,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014387620000000002,
      "loss": 3.0032,
      "step": 280620
    },
    {
      "epoch": 0.11769474986661348,
      "grad_norm": 0.75,
      "learning_rate": 0.0001438742,
      "loss": 3.0955,
      "step": 280630
    },
    {
      "epoch": 0.11771339310277165,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001438722,
      "loss": 3.1253,
      "step": 280640
    },
    {
      "epoch": 0.11773203633892984,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001438702,
      "loss": 3.0986,
      "step": 280650
    },
    {
      "epoch": 0.11775067957508802,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001438682,
      "loss": 3.1691,
      "step": 280660
    },
    {
      "epoch": 0.1177693228112462,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014386620000000002,
      "loss": 3.0272,
      "step": 280670
    },
    {
      "epoch": 0.1177879660474044,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014386420000000002,
      "loss": 3.2587,
      "step": 280680
    },
    {
      "epoch": 0.11780660928356257,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001438622,
      "loss": 3.0645,
      "step": 280690
    },
    {
      "epoch": 0.11782525251972076,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001438602,
      "loss": 3.2594,
      "step": 280700
    },
    {
      "epoch": 0.11784389575587895,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001438582,
      "loss": 3.0885,
      "step": 280710
    },
    {
      "epoch": 0.11786253899203712,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001438562,
      "loss": 3.1322,
      "step": 280720
    },
    {
      "epoch": 0.11788118222819531,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001438542,
      "loss": 3.2984,
      "step": 280730
    },
    {
      "epoch": 0.11789982546435349,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014385220000000002,
      "loss": 3.1886,
      "step": 280740
    },
    {
      "epoch": 0.11791846870051167,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014385020000000001,
      "loss": 3.2295,
      "step": 280750
    },
    {
      "epoch": 0.11793711193666986,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001438482,
      "loss": 3.1547,
      "step": 280760
    },
    {
      "epoch": 0.11795575517282804,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001438462,
      "loss": 3.0557,
      "step": 280770
    },
    {
      "epoch": 0.11797439840898623,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001438442,
      "loss": 3.1224,
      "step": 280780
    },
    {
      "epoch": 0.1179930416451444,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014384220000000002,
      "loss": 3.1756,
      "step": 280790
    },
    {
      "epoch": 0.11801168488130259,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001438402,
      "loss": 3.2041,
      "step": 280800
    },
    {
      "epoch": 0.11803032811746078,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014383820000000001,
      "loss": 3.0882,
      "step": 280810
    },
    {
      "epoch": 0.11804897135361896,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001438362,
      "loss": 3.0282,
      "step": 280820
    },
    {
      "epoch": 0.11806761458977715,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001438342,
      "loss": 3.0194,
      "step": 280830
    },
    {
      "epoch": 0.11808625782593533,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014383220000000003,
      "loss": 3.2138,
      "step": 280840
    },
    {
      "epoch": 0.11810490106209351,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001438302,
      "loss": 3.1359,
      "step": 280850
    },
    {
      "epoch": 0.1181235442982517,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014382820000000002,
      "loss": 3.0708,
      "step": 280860
    },
    {
      "epoch": 0.11814218753440987,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001438262,
      "loss": 3.0305,
      "step": 280870
    },
    {
      "epoch": 0.11816083077056806,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001438242,
      "loss": 3.1167,
      "step": 280880
    },
    {
      "epoch": 0.11817947400672625,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001438222,
      "loss": 3.3052,
      "step": 280890
    },
    {
      "epoch": 0.11819811724288443,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001438202,
      "loss": 3.1915,
      "step": 280900
    },
    {
      "epoch": 0.11821676047904262,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014381820000000003,
      "loss": 3.0062,
      "step": 280910
    },
    {
      "epoch": 0.1182354037152008,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001438162,
      "loss": 3.1201,
      "step": 280920
    },
    {
      "epoch": 0.11825404695135898,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014381420000000002,
      "loss": 3.1707,
      "step": 280930
    },
    {
      "epoch": 0.11827269018751717,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014381219999999998,
      "loss": 3.0572,
      "step": 280940
    },
    {
      "epoch": 0.11829133342367534,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001438102,
      "loss": 3.11,
      "step": 280950
    },
    {
      "epoch": 0.11830997665983353,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001438082,
      "loss": 3.2491,
      "step": 280960
    },
    {
      "epoch": 0.11832861989599172,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001438062,
      "loss": 3.1307,
      "step": 280970
    },
    {
      "epoch": 0.1183472631321499,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014380420000000002,
      "loss": 3.1544,
      "step": 280980
    },
    {
      "epoch": 0.11836590636830809,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001438022,
      "loss": 3.1407,
      "step": 280990
    },
    {
      "epoch": 0.11838454960446626,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001438002,
      "loss": 3.1843,
      "step": 281000
    },
    {
      "epoch": 0.11840319284062445,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001437982,
      "loss": 3.1101,
      "step": 281010
    },
    {
      "epoch": 0.11842183607678264,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001437962,
      "loss": 3.1315,
      "step": 281020
    },
    {
      "epoch": 0.11844047931294081,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001437942,
      "loss": 3.1346,
      "step": 281030
    },
    {
      "epoch": 0.118459122549099,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001437922,
      "loss": 3.2264,
      "step": 281040
    },
    {
      "epoch": 0.11847776578525719,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014379020000000002,
      "loss": 3.2037,
      "step": 281050
    },
    {
      "epoch": 0.11849640902141537,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014378820000000001,
      "loss": 3.0598,
      "step": 281060
    },
    {
      "epoch": 0.11851505225757356,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001437862,
      "loss": 3.1566,
      "step": 281070
    },
    {
      "epoch": 0.11853369549373173,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001437842,
      "loss": 3.1054,
      "step": 281080
    },
    {
      "epoch": 0.11855233872988992,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001437822,
      "loss": 3.1253,
      "step": 281090
    },
    {
      "epoch": 0.11857098196604811,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001437802,
      "loss": 3.2071,
      "step": 281100
    },
    {
      "epoch": 0.11858962520220628,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001437782,
      "loss": 3.1765,
      "step": 281110
    },
    {
      "epoch": 0.11860826843836447,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014377620000000001,
      "loss": 3.0663,
      "step": 281120
    },
    {
      "epoch": 0.11862691167452265,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001437742,
      "loss": 3.1881,
      "step": 281130
    },
    {
      "epoch": 0.11864555491068084,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001437722,
      "loss": 3.2222,
      "step": 281140
    },
    {
      "epoch": 0.11866419814683903,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001437702,
      "loss": 3.2656,
      "step": 281150
    },
    {
      "epoch": 0.1186828413829972,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001437682,
      "loss": 3.0785,
      "step": 281160
    },
    {
      "epoch": 0.11870148461915539,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014376620000000002,
      "loss": 3.1346,
      "step": 281170
    },
    {
      "epoch": 0.11872012785531358,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014376420000000002,
      "loss": 3.112,
      "step": 281180
    },
    {
      "epoch": 0.11873877109147175,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001437622,
      "loss": 2.9524,
      "step": 281190
    },
    {
      "epoch": 0.11875741432762994,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001437602,
      "loss": 3.0476,
      "step": 281200
    },
    {
      "epoch": 0.11877605756378812,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001437582,
      "loss": 3.1744,
      "step": 281210
    },
    {
      "epoch": 0.1187947007999463,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001437562,
      "loss": 3.2711,
      "step": 281220
    },
    {
      "epoch": 0.1188133440361045,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014375420000000002,
      "loss": 3.1438,
      "step": 281230
    },
    {
      "epoch": 0.11883198727226267,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014375220000000002,
      "loss": 3.1367,
      "step": 281240
    },
    {
      "epoch": 0.11885063050842086,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001437502,
      "loss": 3.0389,
      "step": 281250
    },
    {
      "epoch": 0.11886927374457905,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001437482,
      "loss": 3.2598,
      "step": 281260
    },
    {
      "epoch": 0.11888791698073722,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001437462,
      "loss": 3.2097,
      "step": 281270
    },
    {
      "epoch": 0.11890656021689541,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014374420000000003,
      "loss": 3.1449,
      "step": 281280
    },
    {
      "epoch": 0.11892520345305359,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001437422,
      "loss": 3.1226,
      "step": 281290
    },
    {
      "epoch": 0.11894384668921178,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014374020000000002,
      "loss": 3.1313,
      "step": 281300
    },
    {
      "epoch": 0.11896248992536997,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001437382,
      "loss": 3.048,
      "step": 281310
    },
    {
      "epoch": 0.11898113316152814,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001437362,
      "loss": 3.2273,
      "step": 281320
    },
    {
      "epoch": 0.11899977639768633,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001437342,
      "loss": 3.0895,
      "step": 281330
    },
    {
      "epoch": 0.1190184196338445,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001437322,
      "loss": 3.0964,
      "step": 281340
    },
    {
      "epoch": 0.1190370628700027,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014373020000000002,
      "loss": 3.0624,
      "step": 281350
    },
    {
      "epoch": 0.11905570610616088,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001437282,
      "loss": 3.0731,
      "step": 281360
    },
    {
      "epoch": 0.11907434934231906,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001437262,
      "loss": 3.1758,
      "step": 281370
    },
    {
      "epoch": 0.11909299257847725,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001437242,
      "loss": 3.1062,
      "step": 281380
    },
    {
      "epoch": 0.11911163581463544,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001437222,
      "loss": 3.0925,
      "step": 281390
    },
    {
      "epoch": 0.11913027905079361,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014372020000000003,
      "loss": 3.0865,
      "step": 281400
    },
    {
      "epoch": 0.1191489222869518,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001437182,
      "loss": 2.9759,
      "step": 281410
    },
    {
      "epoch": 0.11916756552310998,
      "grad_norm": 0.84375,
      "learning_rate": 0.00014371620000000002,
      "loss": 3.2276,
      "step": 281420
    },
    {
      "epoch": 0.11918620875926816,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001437142,
      "loss": 3.1063,
      "step": 281430
    },
    {
      "epoch": 0.11920485199542635,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001437122,
      "loss": 3.1475,
      "step": 281440
    },
    {
      "epoch": 0.11922349523158453,
      "grad_norm": 0.75,
      "learning_rate": 0.0001437102,
      "loss": 3.2375,
      "step": 281450
    },
    {
      "epoch": 0.11924213846774272,
      "grad_norm": 0.75,
      "learning_rate": 0.0001437082,
      "loss": 3.0225,
      "step": 281460
    },
    {
      "epoch": 0.11926078170390089,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014370620000000002,
      "loss": 3.1264,
      "step": 281470
    },
    {
      "epoch": 0.11927942494005908,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001437042,
      "loss": 3.1585,
      "step": 281480
    },
    {
      "epoch": 0.11929806817621727,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014370220000000001,
      "loss": 3.109,
      "step": 281490
    },
    {
      "epoch": 0.11931671141237545,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014370019999999998,
      "loss": 3.1337,
      "step": 281500
    },
    {
      "epoch": 0.11933535464853363,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001436982,
      "loss": 3.0461,
      "step": 281510
    },
    {
      "epoch": 0.11935399788469182,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001436962,
      "loss": 2.9205,
      "step": 281520
    },
    {
      "epoch": 0.11937264112085,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001436942,
      "loss": 3.1062,
      "step": 281530
    },
    {
      "epoch": 0.11939128435700819,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014369220000000002,
      "loss": 3.023,
      "step": 281540
    },
    {
      "epoch": 0.11940992759316636,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001436902,
      "loss": 3.1251,
      "step": 281550
    },
    {
      "epoch": 0.11942857082932455,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001436882,
      "loss": 3.1535,
      "step": 281560
    },
    {
      "epoch": 0.11944721406548274,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001436862,
      "loss": 3.1207,
      "step": 281570
    },
    {
      "epoch": 0.11946585730164092,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001436842,
      "loss": 3.0473,
      "step": 281580
    },
    {
      "epoch": 0.1194845005377991,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001436822,
      "loss": 3.1559,
      "step": 281590
    },
    {
      "epoch": 0.1195031437739573,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001436802,
      "loss": 3.0782,
      "step": 281600
    },
    {
      "epoch": 0.11952178701011547,
      "grad_norm": 0.75,
      "learning_rate": 0.00014367820000000002,
      "loss": 3.1453,
      "step": 281610
    },
    {
      "epoch": 0.11954043024627366,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001436762,
      "loss": 3.1719,
      "step": 281620
    },
    {
      "epoch": 0.11955907348243183,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001436742,
      "loss": 3.0069,
      "step": 281630
    },
    {
      "epoch": 0.11957771671859002,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001436722,
      "loss": 3.1732,
      "step": 281640
    },
    {
      "epoch": 0.11959635995474821,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001436702,
      "loss": 3.1625,
      "step": 281650
    },
    {
      "epoch": 0.11961500319090639,
      "grad_norm": 0.84375,
      "learning_rate": 0.00014366820000000002,
      "loss": 3.1151,
      "step": 281660
    },
    {
      "epoch": 0.11963364642706457,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014366620000000002,
      "loss": 3.1167,
      "step": 281670
    },
    {
      "epoch": 0.11965228966322275,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001436642,
      "loss": 3.0673,
      "step": 281680
    },
    {
      "epoch": 0.11967093289938094,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001436622,
      "loss": 3.1852,
      "step": 281690
    },
    {
      "epoch": 0.11968957613553913,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001436602,
      "loss": 3.0435,
      "step": 281700
    },
    {
      "epoch": 0.1197082193716973,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001436582,
      "loss": 3.1738,
      "step": 281710
    },
    {
      "epoch": 0.11972686260785549,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001436562,
      "loss": 2.9946,
      "step": 281720
    },
    {
      "epoch": 0.11974550584401368,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014365420000000002,
      "loss": 3.0991,
      "step": 281730
    },
    {
      "epoch": 0.11976414908017186,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001436522,
      "loss": 3.2025,
      "step": 281740
    },
    {
      "epoch": 0.11978279231633004,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001436502,
      "loss": 3.2405,
      "step": 281750
    },
    {
      "epoch": 0.11980143555248822,
      "grad_norm": 0.75,
      "learning_rate": 0.0001436482,
      "loss": 3.0623,
      "step": 281760
    },
    {
      "epoch": 0.11982007878864641,
      "grad_norm": 0.75,
      "learning_rate": 0.0001436462,
      "loss": 3.16,
      "step": 281770
    },
    {
      "epoch": 0.1198387220248046,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001436442,
      "loss": 3.138,
      "step": 281780
    },
    {
      "epoch": 0.11985736526096277,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014364220000000002,
      "loss": 3.0434,
      "step": 281790
    },
    {
      "epoch": 0.11987600849712096,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014364020000000001,
      "loss": 3.0223,
      "step": 281800
    },
    {
      "epoch": 0.11989465173327915,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001436382,
      "loss": 3.2067,
      "step": 281810
    },
    {
      "epoch": 0.11991329496943733,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001436362,
      "loss": 3.1723,
      "step": 281820
    },
    {
      "epoch": 0.11993193820559551,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001436342,
      "loss": 2.9451,
      "step": 281830
    },
    {
      "epoch": 0.11995058144175369,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014363220000000002,
      "loss": 3.1334,
      "step": 281840
    },
    {
      "epoch": 0.11996922467791188,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001436302,
      "loss": 3.1929,
      "step": 281850
    },
    {
      "epoch": 0.11998786791407007,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014362820000000002,
      "loss": 2.998,
      "step": 281860
    },
    {
      "epoch": 0.12000651115022824,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001436262,
      "loss": 3.0521,
      "step": 281870
    },
    {
      "epoch": 0.12002515438638643,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001436242,
      "loss": 3.0867,
      "step": 281880
    },
    {
      "epoch": 0.1200437976225446,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014362220000000003,
      "loss": 3.1008,
      "step": 281890
    },
    {
      "epoch": 0.1200624408587028,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001436202,
      "loss": 3.1967,
      "step": 281900
    },
    {
      "epoch": 0.12008108409486098,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014361820000000002,
      "loss": 3.1709,
      "step": 281910
    },
    {
      "epoch": 0.12009972733101916,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001436162,
      "loss": 3.1153,
      "step": 281920
    },
    {
      "epoch": 0.12011837056717735,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001436142,
      "loss": 3.192,
      "step": 281930
    },
    {
      "epoch": 0.12013701380333554,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001436122,
      "loss": 3.2048,
      "step": 281940
    },
    {
      "epoch": 0.12015565703949371,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001436102,
      "loss": 3.1691,
      "step": 281950
    },
    {
      "epoch": 0.1201743002756519,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014360820000000003,
      "loss": 3.1498,
      "step": 281960
    },
    {
      "epoch": 0.12019294351181008,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001436062,
      "loss": 2.9777,
      "step": 281970
    },
    {
      "epoch": 0.12021158674796827,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014360420000000002,
      "loss": 3.0768,
      "step": 281980
    },
    {
      "epoch": 0.12023022998412645,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014360219999999999,
      "loss": 3.0131,
      "step": 281990
    },
    {
      "epoch": 0.12024887322028463,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001436002,
      "loss": 3.2416,
      "step": 282000
    },
    {
      "epoch": 0.12026751645644282,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001435982,
      "loss": 3.0968,
      "step": 282010
    },
    {
      "epoch": 0.120286159692601,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001435962,
      "loss": 3.0966,
      "step": 282020
    },
    {
      "epoch": 0.12030480292875918,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014359420000000002,
      "loss": 3.1454,
      "step": 282030
    },
    {
      "epoch": 0.12032344616491737,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001435922,
      "loss": 3.0399,
      "step": 282040
    },
    {
      "epoch": 0.12034208940107555,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001435902,
      "loss": 3.1933,
      "step": 282050
    },
    {
      "epoch": 0.12036073263723374,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001435882,
      "loss": 3.1446,
      "step": 282060
    },
    {
      "epoch": 0.12037937587339193,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001435862,
      "loss": 3.196,
      "step": 282070
    },
    {
      "epoch": 0.1203980191095501,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001435842,
      "loss": 3.1118,
      "step": 282080
    },
    {
      "epoch": 0.12041666234570829,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001435822,
      "loss": 3.2091,
      "step": 282090
    },
    {
      "epoch": 0.12043530558186646,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014358020000000002,
      "loss": 3.1032,
      "step": 282100
    },
    {
      "epoch": 0.12045394881802465,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001435782,
      "loss": 3.0881,
      "step": 282110
    },
    {
      "epoch": 0.12047259205418284,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001435762,
      "loss": 3.0342,
      "step": 282120
    },
    {
      "epoch": 0.12049123529034102,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001435742,
      "loss": 3.092,
      "step": 282130
    },
    {
      "epoch": 0.1205098785264992,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001435722,
      "loss": 2.983,
      "step": 282140
    },
    {
      "epoch": 0.1205285217626574,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014357020000000002,
      "loss": 3.1205,
      "step": 282150
    },
    {
      "epoch": 0.12054716499881557,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001435682,
      "loss": 3.164,
      "step": 282160
    },
    {
      "epoch": 0.12056580823497376,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014356620000000001,
      "loss": 3.1817,
      "step": 282170
    },
    {
      "epoch": 0.12058445147113193,
      "grad_norm": 0.75,
      "learning_rate": 0.0001435642,
      "loss": 3.2503,
      "step": 282180
    },
    {
      "epoch": 0.12060309470729012,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001435622,
      "loss": 3.0827,
      "step": 282190
    },
    {
      "epoch": 0.12062173794344831,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001435602,
      "loss": 3.0696,
      "step": 282200
    },
    {
      "epoch": 0.12064038117960649,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001435582,
      "loss": 3.1517,
      "step": 282210
    },
    {
      "epoch": 0.12065902441576468,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014355620000000002,
      "loss": 3.1666,
      "step": 282220
    },
    {
      "epoch": 0.12067766765192285,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014355420000000002,
      "loss": 3.0947,
      "step": 282230
    },
    {
      "epoch": 0.12069631088808104,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001435522,
      "loss": 3.1069,
      "step": 282240
    },
    {
      "epoch": 0.12071495412423923,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001435502,
      "loss": 3.0923,
      "step": 282250
    },
    {
      "epoch": 0.1207335973603974,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001435482,
      "loss": 3.2263,
      "step": 282260
    },
    {
      "epoch": 0.1207522405965556,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001435462,
      "loss": 3.1497,
      "step": 282270
    },
    {
      "epoch": 0.12077088383271378,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014354420000000002,
      "loss": 3.2002,
      "step": 282280
    },
    {
      "epoch": 0.12078952706887196,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014354220000000002,
      "loss": 3.002,
      "step": 282290
    },
    {
      "epoch": 0.12080817030503015,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001435402,
      "loss": 2.9241,
      "step": 282300
    },
    {
      "epoch": 0.12082681354118832,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001435382,
      "loss": 3.1216,
      "step": 282310
    },
    {
      "epoch": 0.12084545677734651,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001435362,
      "loss": 3.0962,
      "step": 282320
    },
    {
      "epoch": 0.1208641000135047,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001435342,
      "loss": 3.0981,
      "step": 282330
    },
    {
      "epoch": 0.12088274324966287,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001435322,
      "loss": 3.1864,
      "step": 282340
    },
    {
      "epoch": 0.12090138648582106,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014353020000000002,
      "loss": 3.2346,
      "step": 282350
    },
    {
      "epoch": 0.12092002972197924,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001435282,
      "loss": 3.0711,
      "step": 282360
    },
    {
      "epoch": 0.12093867295813743,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001435262,
      "loss": 3.1732,
      "step": 282370
    },
    {
      "epoch": 0.12095731619429562,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001435242,
      "loss": 2.9382,
      "step": 282380
    },
    {
      "epoch": 0.12097595943045379,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001435222,
      "loss": 3.2206,
      "step": 282390
    },
    {
      "epoch": 0.12099460266661198,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014352020000000002,
      "loss": 3.1005,
      "step": 282400
    },
    {
      "epoch": 0.12101324590277017,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001435182,
      "loss": 3.1894,
      "step": 282410
    },
    {
      "epoch": 0.12103188913892834,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001435162,
      "loss": 2.9875,
      "step": 282420
    },
    {
      "epoch": 0.12105053237508653,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001435142,
      "loss": 3.1039,
      "step": 282430
    },
    {
      "epoch": 0.12106917561124471,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001435122,
      "loss": 3.1258,
      "step": 282440
    },
    {
      "epoch": 0.1210878188474029,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014351020000000003,
      "loss": 3.047,
      "step": 282450
    },
    {
      "epoch": 0.12110646208356109,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001435082,
      "loss": 3.099,
      "step": 282460
    },
    {
      "epoch": 0.12112510531971926,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014350620000000002,
      "loss": 3.2271,
      "step": 282470
    },
    {
      "epoch": 0.12114374855587745,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001435042,
      "loss": 3.1941,
      "step": 282480
    },
    {
      "epoch": 0.12116239179203564,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001435022,
      "loss": 3.0648,
      "step": 282490
    },
    {
      "epoch": 0.12118103502819381,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001435002,
      "loss": 3.1184,
      "step": 282500
    },
    {
      "epoch": 0.121199678264352,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001434982,
      "loss": 3.0892,
      "step": 282510
    },
    {
      "epoch": 0.12121832150051018,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014349620000000002,
      "loss": 3.1202,
      "step": 282520
    },
    {
      "epoch": 0.12123696473666837,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001434942,
      "loss": 3.0942,
      "step": 282530
    },
    {
      "epoch": 0.12125560797282656,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014349220000000002,
      "loss": 3.0772,
      "step": 282540
    },
    {
      "epoch": 0.12127425120898473,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014349019999999998,
      "loss": 3.1282,
      "step": 282550
    },
    {
      "epoch": 0.12129289444514292,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001434882,
      "loss": 3.0257,
      "step": 282560
    },
    {
      "epoch": 0.1213115376813011,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001434862,
      "loss": 3.1516,
      "step": 282570
    },
    {
      "epoch": 0.12133018091745928,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001434842,
      "loss": 3.2331,
      "step": 282580
    },
    {
      "epoch": 0.12134882415361747,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014348220000000002,
      "loss": 3.0404,
      "step": 282590
    },
    {
      "epoch": 0.12136746738977565,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001434802,
      "loss": 3.0073,
      "step": 282600
    },
    {
      "epoch": 0.12138611062593384,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001434782,
      "loss": 3.0771,
      "step": 282610
    },
    {
      "epoch": 0.12140475386209203,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001434762,
      "loss": 3.1565,
      "step": 282620
    },
    {
      "epoch": 0.1214233970982502,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001434742,
      "loss": 3.1048,
      "step": 282630
    },
    {
      "epoch": 0.12144204033440839,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014347220000000003,
      "loss": 3.1971,
      "step": 282640
    },
    {
      "epoch": 0.12146068357056657,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001434702,
      "loss": 3.2024,
      "step": 282650
    },
    {
      "epoch": 0.12147932680672476,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014346820000000002,
      "loss": 3.2212,
      "step": 282660
    },
    {
      "epoch": 0.12149797004288294,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001434662,
      "loss": 3.1295,
      "step": 282670
    },
    {
      "epoch": 0.12151661327904112,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001434642,
      "loss": 3.1356,
      "step": 282680
    },
    {
      "epoch": 0.12153525651519931,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001434622,
      "loss": 3.2932,
      "step": 282690
    },
    {
      "epoch": 0.12155389975135748,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001434602,
      "loss": 3.1673,
      "step": 282700
    },
    {
      "epoch": 0.12157254298751567,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014345820000000002,
      "loss": 3.046,
      "step": 282710
    },
    {
      "epoch": 0.12159118622367386,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001434562,
      "loss": 3.1581,
      "step": 282720
    },
    {
      "epoch": 0.12160982945983204,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001434542,
      "loss": 3.1443,
      "step": 282730
    },
    {
      "epoch": 0.12162847269599023,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001434522,
      "loss": 3.1288,
      "step": 282740
    },
    {
      "epoch": 0.12164711593214841,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001434502,
      "loss": 3.0763,
      "step": 282750
    },
    {
      "epoch": 0.12166575916830659,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001434482,
      "loss": 3.1694,
      "step": 282760
    },
    {
      "epoch": 0.12168440240446478,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001434462,
      "loss": 3.1363,
      "step": 282770
    },
    {
      "epoch": 0.12170304564062295,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014344420000000002,
      "loss": 3.0909,
      "step": 282780
    },
    {
      "epoch": 0.12172168887678114,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014344220000000001,
      "loss": 3.0371,
      "step": 282790
    },
    {
      "epoch": 0.12174033211293933,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001434402,
      "loss": 3.1513,
      "step": 282800
    },
    {
      "epoch": 0.1217589753490975,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001434382,
      "loss": 3.1992,
      "step": 282810
    },
    {
      "epoch": 0.1217776185852557,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001434362,
      "loss": 3.2441,
      "step": 282820
    },
    {
      "epoch": 0.12179626182141388,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001434342,
      "loss": 3.1419,
      "step": 282830
    },
    {
      "epoch": 0.12181490505757206,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00014343220000000002,
      "loss": 3.1233,
      "step": 282840
    },
    {
      "epoch": 0.12183354829373025,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014343020000000001,
      "loss": 3.0964,
      "step": 282850
    },
    {
      "epoch": 0.12185219152988842,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001434282,
      "loss": 3.1514,
      "step": 282860
    },
    {
      "epoch": 0.12187083476604661,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001434262,
      "loss": 3.1698,
      "step": 282870
    },
    {
      "epoch": 0.1218894780022048,
      "grad_norm": 0.75,
      "learning_rate": 0.0001434242,
      "loss": 3.1975,
      "step": 282880
    },
    {
      "epoch": 0.12190812123836298,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014342220000000002,
      "loss": 3.2412,
      "step": 282890
    },
    {
      "epoch": 0.12192676447452117,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001434202,
      "loss": 3.1802,
      "step": 282900
    },
    {
      "epoch": 0.12194540771067934,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014341820000000002,
      "loss": 3.0537,
      "step": 282910
    },
    {
      "epoch": 0.12196405094683753,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001434162,
      "loss": 3.1402,
      "step": 282920
    },
    {
      "epoch": 0.12198269418299572,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001434142,
      "loss": 3.0751,
      "step": 282930
    },
    {
      "epoch": 0.1220013374191539,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001434122,
      "loss": 3.0759,
      "step": 282940
    },
    {
      "epoch": 0.12201998065531208,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001434102,
      "loss": 3.1586,
      "step": 282950
    },
    {
      "epoch": 0.12203862389147027,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014340820000000002,
      "loss": 3.0533,
      "step": 282960
    },
    {
      "epoch": 0.12205726712762845,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001434062,
      "loss": 3.0895,
      "step": 282970
    },
    {
      "epoch": 0.12207591036378664,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001434042,
      "loss": 3.1269,
      "step": 282980
    },
    {
      "epoch": 0.12209455359994481,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001434022,
      "loss": 3.1892,
      "step": 282990
    },
    {
      "epoch": 0.122113196836103,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001434002,
      "loss": 3.1097,
      "step": 283000
    },
    {
      "epoch": 0.12213184007226119,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014339820000000003,
      "loss": 3.0383,
      "step": 283010
    },
    {
      "epoch": 0.12215048330841936,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001433962,
      "loss": 3.1229,
      "step": 283020
    },
    {
      "epoch": 0.12216912654457755,
      "grad_norm": 0.75,
      "learning_rate": 0.00014339420000000002,
      "loss": 3.1983,
      "step": 283030
    },
    {
      "epoch": 0.12218776978073573,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014339219999999999,
      "loss": 3.0289,
      "step": 283040
    },
    {
      "epoch": 0.12220641301689392,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001433902,
      "loss": 3.126,
      "step": 283050
    },
    {
      "epoch": 0.1222250562530521,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001433882,
      "loss": 3.1612,
      "step": 283060
    },
    {
      "epoch": 0.12224369948921028,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001433862,
      "loss": 3.1643,
      "step": 283070
    },
    {
      "epoch": 0.12226234272536847,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014338420000000002,
      "loss": 3.0743,
      "step": 283080
    },
    {
      "epoch": 0.12228098596152666,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001433822,
      "loss": 3.0544,
      "step": 283090
    },
    {
      "epoch": 0.12229962919768483,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001433802,
      "loss": 3.2043,
      "step": 283100
    },
    {
      "epoch": 0.12231827243384302,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001433782,
      "loss": 3.2581,
      "step": 283110
    },
    {
      "epoch": 0.1223369156700012,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001433762,
      "loss": 3.1768,
      "step": 283120
    },
    {
      "epoch": 0.12235555890615939,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014337420000000003,
      "loss": 3.1568,
      "step": 283130
    },
    {
      "epoch": 0.12237420214231758,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001433722,
      "loss": 3.1935,
      "step": 283140
    },
    {
      "epoch": 0.12239284537847575,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014337020000000002,
      "loss": 3.201,
      "step": 283150
    },
    {
      "epoch": 0.12241148861463394,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001433682,
      "loss": 3.0341,
      "step": 283160
    },
    {
      "epoch": 0.12243013185079213,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001433662,
      "loss": 3.1932,
      "step": 283170
    },
    {
      "epoch": 0.1224487750869503,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001433642,
      "loss": 3.1246,
      "step": 283180
    },
    {
      "epoch": 0.12246741832310849,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001433622,
      "loss": 3.2897,
      "step": 283190
    },
    {
      "epoch": 0.12248606155926667,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014336020000000002,
      "loss": 3.0701,
      "step": 283200
    },
    {
      "epoch": 0.12250470479542486,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001433582,
      "loss": 3.1229,
      "step": 283210
    },
    {
      "epoch": 0.12252334803158305,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014335620000000001,
      "loss": 3.0978,
      "step": 283220
    },
    {
      "epoch": 0.12254199126774122,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001433542,
      "loss": 3.0083,
      "step": 283230
    },
    {
      "epoch": 0.12256063450389941,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001433522,
      "loss": 3.1605,
      "step": 283240
    },
    {
      "epoch": 0.12257927774005759,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001433502,
      "loss": 3.0815,
      "step": 283250
    },
    {
      "epoch": 0.12259792097621577,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001433482,
      "loss": 2.983,
      "step": 283260
    },
    {
      "epoch": 0.12261656421237396,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014334620000000002,
      "loss": 3.2024,
      "step": 283270
    },
    {
      "epoch": 0.12263520744853214,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014334420000000002,
      "loss": 3.1554,
      "step": 283280
    },
    {
      "epoch": 0.12265385068469033,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001433422,
      "loss": 3.1214,
      "step": 283290
    },
    {
      "epoch": 0.12267249392084852,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001433402,
      "loss": 3.2069,
      "step": 283300
    },
    {
      "epoch": 0.12269113715700669,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001433382,
      "loss": 3.1137,
      "step": 283310
    },
    {
      "epoch": 0.12270978039316488,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001433362,
      "loss": 3.0553,
      "step": 283320
    },
    {
      "epoch": 0.12272842362932306,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001433342,
      "loss": 3.1023,
      "step": 283330
    },
    {
      "epoch": 0.12274706686548124,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014333220000000002,
      "loss": 3.1421,
      "step": 283340
    },
    {
      "epoch": 0.12276571010163943,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001433302,
      "loss": 3.2198,
      "step": 283350
    },
    {
      "epoch": 0.12278435333779761,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001433282,
      "loss": 3.1974,
      "step": 283360
    },
    {
      "epoch": 0.1228029965739558,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001433262,
      "loss": 3.2668,
      "step": 283370
    },
    {
      "epoch": 0.12282163981011397,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001433242,
      "loss": 3.1267,
      "step": 283380
    },
    {
      "epoch": 0.12284028304627216,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001433222,
      "loss": 3.0944,
      "step": 283390
    },
    {
      "epoch": 0.12285892628243035,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014332020000000002,
      "loss": 3.0756,
      "step": 283400
    },
    {
      "epoch": 0.12287756951858853,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001433182,
      "loss": 3.0802,
      "step": 283410
    },
    {
      "epoch": 0.12289621275474671,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001433162,
      "loss": 3.117,
      "step": 283420
    },
    {
      "epoch": 0.1229148559909049,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001433142,
      "loss": 3.1986,
      "step": 283430
    },
    {
      "epoch": 0.12293349922706308,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001433122,
      "loss": 3.0966,
      "step": 283440
    },
    {
      "epoch": 0.12295214246322127,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014331020000000002,
      "loss": 3.2032,
      "step": 283450
    },
    {
      "epoch": 0.12297078569937944,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001433082,
      "loss": 3.1388,
      "step": 283460
    },
    {
      "epoch": 0.12298942893553763,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014330620000000001,
      "loss": 3.1661,
      "step": 283470
    },
    {
      "epoch": 0.12300807217169582,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001433042,
      "loss": 3.1061,
      "step": 283480
    },
    {
      "epoch": 0.123026715407854,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001433022,
      "loss": 3.1243,
      "step": 283490
    },
    {
      "epoch": 0.12304535864401218,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014330020000000003,
      "loss": 3.1674,
      "step": 283500
    },
    {
      "epoch": 0.12306400188017037,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001432982,
      "loss": 3.1908,
      "step": 283510
    },
    {
      "epoch": 0.12308264511632855,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014329620000000002,
      "loss": 3.2276,
      "step": 283520
    },
    {
      "epoch": 0.12310128835248674,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001432942,
      "loss": 3.2535,
      "step": 283530
    },
    {
      "epoch": 0.12311993158864491,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001432922,
      "loss": 3.0942,
      "step": 283540
    },
    {
      "epoch": 0.1231385748248031,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001432902,
      "loss": 3.0982,
      "step": 283550
    },
    {
      "epoch": 0.12315721806096129,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001432882,
      "loss": 3.1271,
      "step": 283560
    },
    {
      "epoch": 0.12317586129711947,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014328620000000002,
      "loss": 3.216,
      "step": 283570
    },
    {
      "epoch": 0.12319450453327765,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001432842,
      "loss": 3.111,
      "step": 283580
    },
    {
      "epoch": 0.12321314776943583,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014328220000000002,
      "loss": 3.2126,
      "step": 283590
    },
    {
      "epoch": 0.12323179100559402,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001432802,
      "loss": 3.0376,
      "step": 283600
    },
    {
      "epoch": 0.12325043424175221,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001432782,
      "loss": 3.248,
      "step": 283610
    },
    {
      "epoch": 0.12326907747791038,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014327620000000003,
      "loss": 3.0927,
      "step": 283620
    },
    {
      "epoch": 0.12328772071406857,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001432742,
      "loss": 3.2434,
      "step": 283630
    },
    {
      "epoch": 0.12330636395022676,
      "grad_norm": 0.84375,
      "learning_rate": 0.00014327220000000002,
      "loss": 3.2617,
      "step": 283640
    },
    {
      "epoch": 0.12332500718638494,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001432702,
      "loss": 3.0767,
      "step": 283650
    },
    {
      "epoch": 0.12334365042254312,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001432682,
      "loss": 3.1645,
      "step": 283660
    },
    {
      "epoch": 0.1233622936587013,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001432662,
      "loss": 3.0152,
      "step": 283670
    },
    {
      "epoch": 0.12338093689485949,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001432642,
      "loss": 3.0623,
      "step": 283680
    },
    {
      "epoch": 0.12339958013101768,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014326220000000003,
      "loss": 3.1888,
      "step": 283690
    },
    {
      "epoch": 0.12341822336717585,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001432602,
      "loss": 3.1975,
      "step": 283700
    },
    {
      "epoch": 0.12343686660333404,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014325820000000002,
      "loss": 3.2298,
      "step": 283710
    },
    {
      "epoch": 0.12345550983949222,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014325619999999999,
      "loss": 3.1792,
      "step": 283720
    },
    {
      "epoch": 0.1234741530756504,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001432542,
      "loss": 3.1209,
      "step": 283730
    },
    {
      "epoch": 0.1234927963118086,
      "grad_norm": 0.875,
      "learning_rate": 0.0001432522,
      "loss": 3.0881,
      "step": 283740
    },
    {
      "epoch": 0.12351143954796677,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001432502,
      "loss": 3.1489,
      "step": 283750
    },
    {
      "epoch": 0.12353008278412496,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014324820000000002,
      "loss": 3.2128,
      "step": 283760
    },
    {
      "epoch": 0.12354872602028315,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001432462,
      "loss": 3.1387,
      "step": 283770
    },
    {
      "epoch": 0.12356736925644132,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001432442,
      "loss": 3.1892,
      "step": 283780
    },
    {
      "epoch": 0.12358601249259951,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001432422,
      "loss": 3.0658,
      "step": 283790
    },
    {
      "epoch": 0.12360465572875769,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001432402,
      "loss": 3.1949,
      "step": 283800
    },
    {
      "epoch": 0.12362329896491588,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001432382,
      "loss": 3.1327,
      "step": 283810
    },
    {
      "epoch": 0.12364194220107406,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001432362,
      "loss": 3.0239,
      "step": 283820
    },
    {
      "epoch": 0.12366058543723224,
      "grad_norm": 0.609375,
      "learning_rate": 0.00014323420000000002,
      "loss": 3.1665,
      "step": 283830
    },
    {
      "epoch": 0.12367922867339043,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014323220000000001,
      "loss": 3.116,
      "step": 283840
    },
    {
      "epoch": 0.12369787190954862,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001432302,
      "loss": 3.1788,
      "step": 283850
    },
    {
      "epoch": 0.12371651514570679,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001432282,
      "loss": 3.119,
      "step": 283860
    },
    {
      "epoch": 0.12373515838186498,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001432262,
      "loss": 3.146,
      "step": 283870
    },
    {
      "epoch": 0.12375380161802316,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001432242,
      "loss": 3.2837,
      "step": 283880
    },
    {
      "epoch": 0.12377244485418135,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014322220000000002,
      "loss": 3.1209,
      "step": 283890
    },
    {
      "epoch": 0.12379108809033953,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014322020000000001,
      "loss": 3.1373,
      "step": 283900
    },
    {
      "epoch": 0.12380973132649771,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001432182,
      "loss": 2.9741,
      "step": 283910
    },
    {
      "epoch": 0.1238283745626559,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001432162,
      "loss": 3.1655,
      "step": 283920
    },
    {
      "epoch": 0.12384701779881407,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001432142,
      "loss": 3.0854,
      "step": 283930
    },
    {
      "epoch": 0.12386566103497226,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001432122,
      "loss": 2.9932,
      "step": 283940
    },
    {
      "epoch": 0.12388430427113045,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001432102,
      "loss": 3.0524,
      "step": 283950
    },
    {
      "epoch": 0.12390294750728863,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014320820000000002,
      "loss": 3.2194,
      "step": 283960
    },
    {
      "epoch": 0.12392159074344682,
      "grad_norm": 0.625,
      "learning_rate": 0.0001432062,
      "loss": 3.0368,
      "step": 283970
    },
    {
      "epoch": 0.123940233979605,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001432042,
      "loss": 3.1756,
      "step": 283980
    },
    {
      "epoch": 0.12395887721576318,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001432022,
      "loss": 3.0351,
      "step": 283990
    },
    {
      "epoch": 0.12397752045192137,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001432002,
      "loss": 3.0707,
      "step": 284000
    },
    {
      "epoch": 0.12399616368807954,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014319820000000002,
      "loss": 3.1507,
      "step": 284010
    },
    {
      "epoch": 0.12401480692423773,
      "grad_norm": 0.75,
      "learning_rate": 0.0001431962,
      "loss": 3.2717,
      "step": 284020
    },
    {
      "epoch": 0.12403345016039592,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001431942,
      "loss": 3.1922,
      "step": 284030
    },
    {
      "epoch": 0.1240520933965541,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001431922,
      "loss": 3.111,
      "step": 284040
    },
    {
      "epoch": 0.12407073663271229,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001431902,
      "loss": 3.2314,
      "step": 284050
    },
    {
      "epoch": 0.12408937986887046,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014318820000000003,
      "loss": 3.1103,
      "step": 284060
    },
    {
      "epoch": 0.12410802310502865,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001431862,
      "loss": 3.0826,
      "step": 284070
    },
    {
      "epoch": 0.12412666634118684,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014318420000000002,
      "loss": 3.0404,
      "step": 284080
    },
    {
      "epoch": 0.12414530957734501,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001431822,
      "loss": 3.231,
      "step": 284090
    },
    {
      "epoch": 0.1241639528135032,
      "grad_norm": 0.75,
      "learning_rate": 0.0001431802,
      "loss": 3.1807,
      "step": 284100
    },
    {
      "epoch": 0.12418259604966139,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014317820000000003,
      "loss": 3.0069,
      "step": 284110
    },
    {
      "epoch": 0.12420123928581957,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001431762,
      "loss": 3.1883,
      "step": 284120
    },
    {
      "epoch": 0.12421988252197776,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014317420000000002,
      "loss": 3.1322,
      "step": 284130
    },
    {
      "epoch": 0.12423852575813593,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001431722,
      "loss": 3.1197,
      "step": 284140
    },
    {
      "epoch": 0.12425716899429412,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014317020000000001,
      "loss": 3.256,
      "step": 284150
    },
    {
      "epoch": 0.12427581223045231,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001431682,
      "loss": 3.2005,
      "step": 284160
    },
    {
      "epoch": 0.12429445546661048,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001431662,
      "loss": 3.1483,
      "step": 284170
    },
    {
      "epoch": 0.12431309870276867,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014316420000000003,
      "loss": 3.0457,
      "step": 284180
    },
    {
      "epoch": 0.12433174193892686,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001431622,
      "loss": 3.1653,
      "step": 284190
    },
    {
      "epoch": 0.12435038517508504,
      "grad_norm": 0.75,
      "learning_rate": 0.00014316020000000002,
      "loss": 3.026,
      "step": 284200
    },
    {
      "epoch": 0.12436902841124323,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001431582,
      "loss": 3.0275,
      "step": 284210
    },
    {
      "epoch": 0.1243876716474014,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001431562,
      "loss": 3.0774,
      "step": 284220
    },
    {
      "epoch": 0.12440631488355959,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001431542,
      "loss": 3.1358,
      "step": 284230
    },
    {
      "epoch": 0.12442495811971778,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001431522,
      "loss": 3.1698,
      "step": 284240
    },
    {
      "epoch": 0.12444360135587595,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014315020000000002,
      "loss": 3.1519,
      "step": 284250
    },
    {
      "epoch": 0.12446224459203414,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001431482,
      "loss": 3.1696,
      "step": 284260
    },
    {
      "epoch": 0.12448088782819232,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014314620000000002,
      "loss": 3.0859,
      "step": 284270
    },
    {
      "epoch": 0.12449953106435051,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001431442,
      "loss": 3.207,
      "step": 284280
    },
    {
      "epoch": 0.1245181743005087,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001431422,
      "loss": 3.1433,
      "step": 284290
    },
    {
      "epoch": 0.12453681753666687,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001431402,
      "loss": 3.027,
      "step": 284300
    },
    {
      "epoch": 0.12455546077282506,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001431382,
      "loss": 2.9993,
      "step": 284310
    },
    {
      "epoch": 0.12457410400898325,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014313620000000002,
      "loss": 3.0686,
      "step": 284320
    },
    {
      "epoch": 0.12459274724514142,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001431342,
      "loss": 3.1333,
      "step": 284330
    },
    {
      "epoch": 0.12461139048129961,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001431322,
      "loss": 3.0779,
      "step": 284340
    },
    {
      "epoch": 0.12463003371745779,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001431302,
      "loss": 3.204,
      "step": 284350
    },
    {
      "epoch": 0.12464867695361598,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001431282,
      "loss": 3.1914,
      "step": 284360
    },
    {
      "epoch": 0.12466732018977417,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001431262,
      "loss": 3.0015,
      "step": 284370
    },
    {
      "epoch": 0.12468596342593234,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001431242,
      "loss": 3.0529,
      "step": 284380
    },
    {
      "epoch": 0.12470460666209053,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014312220000000002,
      "loss": 3.2699,
      "step": 284390
    },
    {
      "epoch": 0.1247232498982487,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001431202,
      "loss": 3.2209,
      "step": 284400
    },
    {
      "epoch": 0.1247418931344069,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001431182,
      "loss": 3.2226,
      "step": 284410
    },
    {
      "epoch": 0.12476053637056508,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001431162,
      "loss": 3.1073,
      "step": 284420
    },
    {
      "epoch": 0.12477917960672326,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001431142,
      "loss": 3.1878,
      "step": 284430
    },
    {
      "epoch": 0.12479782284288145,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001431122,
      "loss": 3.1575,
      "step": 284440
    },
    {
      "epoch": 0.12481646607903964,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014311020000000002,
      "loss": 3.2953,
      "step": 284450
    },
    {
      "epoch": 0.12483510931519781,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001431082,
      "loss": 3.1866,
      "step": 284460
    },
    {
      "epoch": 0.124853752551356,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001431062,
      "loss": 3.0641,
      "step": 284470
    },
    {
      "epoch": 0.12487239578751418,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001431042,
      "loss": 3.1264,
      "step": 284480
    },
    {
      "epoch": 0.12489103902367236,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001431022,
      "loss": 3.105,
      "step": 284490
    },
    {
      "epoch": 0.12490968225983055,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00014310020000000002,
      "loss": 3.0249,
      "step": 284500
    },
    {
      "epoch": 0.12492832549598873,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001430982,
      "loss": 3.1396,
      "step": 284510
    },
    {
      "epoch": 0.12494696873214692,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014309620000000001,
      "loss": 3.0538,
      "step": 284520
    },
    {
      "epoch": 0.12496561196830511,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001430942,
      "loss": 3.1673,
      "step": 284530
    },
    {
      "epoch": 0.12498425520446328,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001430922,
      "loss": 2.9878,
      "step": 284540
    },
    {
      "epoch": 0.12500289844062146,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001430902,
      "loss": 3.0934,
      "step": 284550
    },
    {
      "epoch": 0.12502154167677965,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001430882,
      "loss": 3.098,
      "step": 284560
    },
    {
      "epoch": 0.12504018491293784,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014308620000000002,
      "loss": 3.1801,
      "step": 284570
    },
    {
      "epoch": 0.12505882814909602,
      "grad_norm": 0.875,
      "learning_rate": 0.00014308420000000001,
      "loss": 3.0294,
      "step": 284580
    },
    {
      "epoch": 0.1250774713852542,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001430822,
      "loss": 3.1708,
      "step": 284590
    },
    {
      "epoch": 0.12509611462141237,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001430802,
      "loss": 3.1048,
      "step": 284600
    },
    {
      "epoch": 0.12511475785757056,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001430782,
      "loss": 3.0919,
      "step": 284610
    },
    {
      "epoch": 0.12513340109372875,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014307620000000002,
      "loss": 3.1243,
      "step": 284620
    },
    {
      "epoch": 0.12515204432988694,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001430742,
      "loss": 3.1491,
      "step": 284630
    },
    {
      "epoch": 0.12517068756604513,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014307220000000002,
      "loss": 3.1422,
      "step": 284640
    },
    {
      "epoch": 0.1251893308022033,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001430702,
      "loss": 3.0628,
      "step": 284650
    },
    {
      "epoch": 0.12520797403836148,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001430682,
      "loss": 3.0758,
      "step": 284660
    },
    {
      "epoch": 0.12522661727451967,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014306620000000003,
      "loss": 2.9798,
      "step": 284670
    },
    {
      "epoch": 0.12524526051067786,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001430642,
      "loss": 3.0824,
      "step": 284680
    },
    {
      "epoch": 0.12526390374683605,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014306220000000002,
      "loss": 3.1072,
      "step": 284690
    },
    {
      "epoch": 0.12528254698299424,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001430602,
      "loss": 3.1447,
      "step": 284700
    },
    {
      "epoch": 0.1253011902191524,
      "grad_norm": 0.625,
      "learning_rate": 0.0001430582,
      "loss": 3.1115,
      "step": 284710
    },
    {
      "epoch": 0.1253198334553106,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001430562,
      "loss": 3.0333,
      "step": 284720
    },
    {
      "epoch": 0.12533847669146878,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001430542,
      "loss": 3.1926,
      "step": 284730
    },
    {
      "epoch": 0.12535711992762696,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014305220000000003,
      "loss": 3.1081,
      "step": 284740
    },
    {
      "epoch": 0.12537576316378515,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001430502,
      "loss": 3.1633,
      "step": 284750
    },
    {
      "epoch": 0.12539440639994331,
      "grad_norm": 0.90625,
      "learning_rate": 0.00014304820000000002,
      "loss": 3.0194,
      "step": 284760
    },
    {
      "epoch": 0.1254130496361015,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014304619999999999,
      "loss": 3.0953,
      "step": 284770
    },
    {
      "epoch": 0.1254316928722597,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001430442,
      "loss": 3.1107,
      "step": 284780
    },
    {
      "epoch": 0.12545033610841788,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001430422,
      "loss": 3.1807,
      "step": 284790
    },
    {
      "epoch": 0.12546897934457607,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001430402,
      "loss": 3.1812,
      "step": 284800
    },
    {
      "epoch": 0.12548762258073423,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014303820000000002,
      "loss": 3.1449,
      "step": 284810
    },
    {
      "epoch": 0.12550626581689242,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001430362,
      "loss": 3.0676,
      "step": 284820
    },
    {
      "epoch": 0.1255249090530506,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014303420000000001,
      "loss": 3.222,
      "step": 284830
    },
    {
      "epoch": 0.1255435522892088,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001430322,
      "loss": 3.111,
      "step": 284840
    },
    {
      "epoch": 0.125562195525367,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001430302,
      "loss": 3.1443,
      "step": 284850
    },
    {
      "epoch": 0.12558083876152515,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001430282,
      "loss": 3.1339,
      "step": 284860
    },
    {
      "epoch": 0.12559948199768334,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001430262,
      "loss": 3.1871,
      "step": 284870
    },
    {
      "epoch": 0.12561812523384153,
      "grad_norm": 0.625,
      "learning_rate": 0.00014302420000000002,
      "loss": 3.1466,
      "step": 284880
    },
    {
      "epoch": 0.12563676846999972,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014302220000000001,
      "loss": 3.0744,
      "step": 284890
    },
    {
      "epoch": 0.1256554117061579,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001430202,
      "loss": 3.0829,
      "step": 284900
    },
    {
      "epoch": 0.12567405494231607,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001430182,
      "loss": 3.0679,
      "step": 284910
    },
    {
      "epoch": 0.12569269817847425,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001430162,
      "loss": 3.2255,
      "step": 284920
    },
    {
      "epoch": 0.12571134141463244,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001430142,
      "loss": 3.072,
      "step": 284930
    },
    {
      "epoch": 0.12572998465079063,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001430122,
      "loss": 3.207,
      "step": 284940
    },
    {
      "epoch": 0.12574862788694882,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014301020000000002,
      "loss": 3.0887,
      "step": 284950
    },
    {
      "epoch": 0.125767271123107,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001430082,
      "loss": 3.1261,
      "step": 284960
    },
    {
      "epoch": 0.12578591435926517,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001430062,
      "loss": 3.1009,
      "step": 284970
    },
    {
      "epoch": 0.12580455759542336,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001430042,
      "loss": 3.0908,
      "step": 284980
    },
    {
      "epoch": 0.12582320083158155,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001430022,
      "loss": 3.0663,
      "step": 284990
    },
    {
      "epoch": 0.12584184406773974,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001430002,
      "loss": 3.1212,
      "step": 285000
    },
    {
      "epoch": 0.12586048730389793,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014299820000000002,
      "loss": 3.152,
      "step": 285010
    },
    {
      "epoch": 0.1258791305400561,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001429962,
      "loss": 3.0072,
      "step": 285020
    },
    {
      "epoch": 0.12589777377621428,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001429942,
      "loss": 3.1516,
      "step": 285030
    },
    {
      "epoch": 0.12591641701237247,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001429922,
      "loss": 3.2851,
      "step": 285040
    },
    {
      "epoch": 0.12593506024853066,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001429902,
      "loss": 3.0655,
      "step": 285050
    },
    {
      "epoch": 0.12595370348468884,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014298820000000002,
      "loss": 3.1989,
      "step": 285060
    },
    {
      "epoch": 0.125972346720847,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014298620000000002,
      "loss": 3.1262,
      "step": 285070
    },
    {
      "epoch": 0.1259909899570052,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001429842,
      "loss": 3.09,
      "step": 285080
    },
    {
      "epoch": 0.12600963319316338,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001429822,
      "loss": 3.1965,
      "step": 285090
    },
    {
      "epoch": 0.12602827642932157,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001429802,
      "loss": 3.069,
      "step": 285100
    },
    {
      "epoch": 0.12604691966547976,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014297820000000003,
      "loss": 3.1443,
      "step": 285110
    },
    {
      "epoch": 0.12606556290163792,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001429762,
      "loss": 3.2834,
      "step": 285120
    },
    {
      "epoch": 0.1260842061377961,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014297420000000002,
      "loss": 3.034,
      "step": 285130
    },
    {
      "epoch": 0.1261028493739543,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001429722,
      "loss": 3.1614,
      "step": 285140
    },
    {
      "epoch": 0.1261214926101125,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001429702,
      "loss": 2.9792,
      "step": 285150
    },
    {
      "epoch": 0.12614013584627068,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001429682,
      "loss": 3.0576,
      "step": 285160
    },
    {
      "epoch": 0.12615877908242887,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001429662,
      "loss": 3.0683,
      "step": 285170
    },
    {
      "epoch": 0.12617742231858703,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014296420000000002,
      "loss": 3.245,
      "step": 285180
    },
    {
      "epoch": 0.12619606555474522,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001429622,
      "loss": 3.064,
      "step": 285190
    },
    {
      "epoch": 0.1262147087909034,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014296020000000001,
      "loss": 3.1774,
      "step": 285200
    },
    {
      "epoch": 0.1262333520270616,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001429582,
      "loss": 3.1966,
      "step": 285210
    },
    {
      "epoch": 0.12625199526321979,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001429562,
      "loss": 3.05,
      "step": 285220
    },
    {
      "epoch": 0.12627063849937795,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014295420000000003,
      "loss": 3.1267,
      "step": 285230
    },
    {
      "epoch": 0.12628928173553614,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001429522,
      "loss": 3.1239,
      "step": 285240
    },
    {
      "epoch": 0.12630792497169432,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014295020000000002,
      "loss": 3.1469,
      "step": 285250
    },
    {
      "epoch": 0.1263265682078525,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001429482,
      "loss": 2.9356,
      "step": 285260
    },
    {
      "epoch": 0.1263452114440107,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001429462,
      "loss": 3.0757,
      "step": 285270
    },
    {
      "epoch": 0.12636385468016886,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001429442,
      "loss": 3.1741,
      "step": 285280
    },
    {
      "epoch": 0.12638249791632705,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001429422,
      "loss": 3.2105,
      "step": 285290
    },
    {
      "epoch": 0.12640114115248524,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014294020000000002,
      "loss": 3.0685,
      "step": 285300
    },
    {
      "epoch": 0.12641978438864343,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001429382,
      "loss": 3.1212,
      "step": 285310
    },
    {
      "epoch": 0.12643842762480162,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014293620000000002,
      "loss": 3.2344,
      "step": 285320
    },
    {
      "epoch": 0.12645707086095978,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014293419999999998,
      "loss": 3.0626,
      "step": 285330
    },
    {
      "epoch": 0.12647571409711797,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001429322,
      "loss": 3.1925,
      "step": 285340
    },
    {
      "epoch": 0.12649435733327616,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001429302,
      "loss": 3.1293,
      "step": 285350
    },
    {
      "epoch": 0.12651300056943435,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001429282,
      "loss": 3.0521,
      "step": 285360
    },
    {
      "epoch": 0.12653164380559254,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014292620000000002,
      "loss": 3.1144,
      "step": 285370
    },
    {
      "epoch": 0.12655028704175073,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001429242,
      "loss": 3.1761,
      "step": 285380
    },
    {
      "epoch": 0.1265689302779089,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001429222,
      "loss": 3.1148,
      "step": 285390
    },
    {
      "epoch": 0.12658757351406708,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001429202,
      "loss": 3.0872,
      "step": 285400
    },
    {
      "epoch": 0.12660621675022526,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001429182,
      "loss": 3.1075,
      "step": 285410
    },
    {
      "epoch": 0.12662485998638345,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001429162,
      "loss": 3.0649,
      "step": 285420
    },
    {
      "epoch": 0.12664350322254164,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001429142,
      "loss": 3.1534,
      "step": 285430
    },
    {
      "epoch": 0.1266621464586998,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014291220000000002,
      "loss": 3.1063,
      "step": 285440
    },
    {
      "epoch": 0.126680789694858,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001429102,
      "loss": 3.2579,
      "step": 285450
    },
    {
      "epoch": 0.12669943293101618,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001429082,
      "loss": 2.9995,
      "step": 285460
    },
    {
      "epoch": 0.12671807616717437,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001429062,
      "loss": 3.1231,
      "step": 285470
    },
    {
      "epoch": 0.12673671940333256,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001429042,
      "loss": 3.0938,
      "step": 285480
    },
    {
      "epoch": 0.12675536263949072,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001429022,
      "loss": 3.2718,
      "step": 285490
    },
    {
      "epoch": 0.1267740058756489,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014290020000000002,
      "loss": 3.118,
      "step": 285500
    },
    {
      "epoch": 0.1267926491118071,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014289820000000001,
      "loss": 3.0686,
      "step": 285510
    },
    {
      "epoch": 0.1268112923479653,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001428962,
      "loss": 3.098,
      "step": 285520
    },
    {
      "epoch": 0.12682993558412348,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001428942,
      "loss": 3.1144,
      "step": 285530
    },
    {
      "epoch": 0.12684857882028164,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001428922,
      "loss": 3.1278,
      "step": 285540
    },
    {
      "epoch": 0.12686722205643983,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001428902,
      "loss": 3.2007,
      "step": 285550
    },
    {
      "epoch": 0.12688586529259802,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014288820000000002,
      "loss": 3.2504,
      "step": 285560
    },
    {
      "epoch": 0.1269045085287562,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014288620000000001,
      "loss": 3.2109,
      "step": 285570
    },
    {
      "epoch": 0.1269231517649144,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001428842,
      "loss": 3.1065,
      "step": 285580
    },
    {
      "epoch": 0.12694179500107255,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001428822,
      "loss": 3.1644,
      "step": 285590
    },
    {
      "epoch": 0.12696043823723074,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001428802,
      "loss": 3.1582,
      "step": 285600
    },
    {
      "epoch": 0.12697908147338893,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001428782,
      "loss": 3.0654,
      "step": 285610
    },
    {
      "epoch": 0.12699772470954712,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014287620000000002,
      "loss": 3.0746,
      "step": 285620
    },
    {
      "epoch": 0.1270163679457053,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014287420000000002,
      "loss": 3.0674,
      "step": 285630
    },
    {
      "epoch": 0.1270350111818635,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001428722,
      "loss": 3.0695,
      "step": 285640
    },
    {
      "epoch": 0.12705365441802166,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001428702,
      "loss": 3.0005,
      "step": 285650
    },
    {
      "epoch": 0.12707229765417985,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001428682,
      "loss": 3.1799,
      "step": 285660
    },
    {
      "epoch": 0.12709094089033804,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014286620000000002,
      "loss": 3.116,
      "step": 285670
    },
    {
      "epoch": 0.12710958412649623,
      "grad_norm": 0.75,
      "learning_rate": 0.0001428642,
      "loss": 3.076,
      "step": 285680
    },
    {
      "epoch": 0.12712822736265442,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014286220000000002,
      "loss": 3.1054,
      "step": 285690
    },
    {
      "epoch": 0.12714687059881258,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001428602,
      "loss": 3.1494,
      "step": 285700
    },
    {
      "epoch": 0.12716551383497077,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001428582,
      "loss": 3.0517,
      "step": 285710
    },
    {
      "epoch": 0.12718415707112896,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014285620000000003,
      "loss": 3.202,
      "step": 285720
    },
    {
      "epoch": 0.12720280030728714,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001428542,
      "loss": 3.0371,
      "step": 285730
    },
    {
      "epoch": 0.12722144354344533,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014285220000000002,
      "loss": 3.0579,
      "step": 285740
    },
    {
      "epoch": 0.1272400867796035,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001428502,
      "loss": 3.0503,
      "step": 285750
    },
    {
      "epoch": 0.12725873001576168,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001428482,
      "loss": 2.924,
      "step": 285760
    },
    {
      "epoch": 0.12727737325191987,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001428462,
      "loss": 3.1814,
      "step": 285770
    },
    {
      "epoch": 0.12729601648807806,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001428442,
      "loss": 3.0785,
      "step": 285780
    },
    {
      "epoch": 0.12731465972423625,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014284220000000003,
      "loss": 3.1191,
      "step": 285790
    },
    {
      "epoch": 0.1273333029603944,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001428402,
      "loss": 3.2054,
      "step": 285800
    },
    {
      "epoch": 0.1273519461965526,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014283820000000002,
      "loss": 3.1953,
      "step": 285810
    },
    {
      "epoch": 0.1273705894327108,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014283619999999999,
      "loss": 3.1201,
      "step": 285820
    },
    {
      "epoch": 0.12738923266886898,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001428342,
      "loss": 3.1456,
      "step": 285830
    },
    {
      "epoch": 0.12740787590502717,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001428322,
      "loss": 3.2256,
      "step": 285840
    },
    {
      "epoch": 0.12742651914118536,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001428302,
      "loss": 3.0255,
      "step": 285850
    },
    {
      "epoch": 0.12744516237734352,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014282820000000002,
      "loss": 3.2232,
      "step": 285860
    },
    {
      "epoch": 0.1274638056135017,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001428262,
      "loss": 3.2667,
      "step": 285870
    },
    {
      "epoch": 0.1274824488496599,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014282420000000001,
      "loss": 3.0692,
      "step": 285880
    },
    {
      "epoch": 0.12750109208581809,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001428222,
      "loss": 3.2301,
      "step": 285890
    },
    {
      "epoch": 0.12751973532197627,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001428202,
      "loss": 3.0868,
      "step": 285900
    },
    {
      "epoch": 0.12753837855813444,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001428182,
      "loss": 3.1467,
      "step": 285910
    },
    {
      "epoch": 0.12755702179429262,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001428162,
      "loss": 3.1579,
      "step": 285920
    },
    {
      "epoch": 0.1275756650304508,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014281420000000002,
      "loss": 3.3654,
      "step": 285930
    },
    {
      "epoch": 0.127594308266609,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001428122,
      "loss": 3.1799,
      "step": 285940
    },
    {
      "epoch": 0.1276129515027672,
      "grad_norm": 0.75,
      "learning_rate": 0.0001428102,
      "loss": 3.0023,
      "step": 285950
    },
    {
      "epoch": 0.12763159473892535,
      "grad_norm": 0.75,
      "learning_rate": 0.0001428082,
      "loss": 3.1327,
      "step": 285960
    },
    {
      "epoch": 0.12765023797508354,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001428062,
      "loss": 3.1419,
      "step": 285970
    },
    {
      "epoch": 0.12766888121124173,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001428042,
      "loss": 3.1665,
      "step": 285980
    },
    {
      "epoch": 0.12768752444739992,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001428022,
      "loss": 3.0909,
      "step": 285990
    },
    {
      "epoch": 0.1277061676835581,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014280020000000002,
      "loss": 3.0794,
      "step": 286000
    },
    {
      "epoch": 0.12772481091971627,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001427982,
      "loss": 3.1626,
      "step": 286010
    },
    {
      "epoch": 0.12774345415587446,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001427962,
      "loss": 3.0128,
      "step": 286020
    },
    {
      "epoch": 0.12776209739203265,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001427942,
      "loss": 3.1864,
      "step": 286030
    },
    {
      "epoch": 0.12778074062819084,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001427922,
      "loss": 3.152,
      "step": 286040
    },
    {
      "epoch": 0.12779938386434903,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014279020000000002,
      "loss": 3.0928,
      "step": 286050
    },
    {
      "epoch": 0.12781802710050721,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014278820000000002,
      "loss": 3.0728,
      "step": 286060
    },
    {
      "epoch": 0.12783667033666538,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001427862,
      "loss": 3.1882,
      "step": 286070
    },
    {
      "epoch": 0.12785531357282356,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001427842,
      "loss": 3.1158,
      "step": 286080
    },
    {
      "epoch": 0.12787395680898175,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001427822,
      "loss": 3.0395,
      "step": 286090
    },
    {
      "epoch": 0.12789260004513994,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001427802,
      "loss": 3.2039,
      "step": 286100
    },
    {
      "epoch": 0.12791124328129813,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014277820000000002,
      "loss": 3.1699,
      "step": 286110
    },
    {
      "epoch": 0.1279298865174563,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014277620000000002,
      "loss": 3.0877,
      "step": 286120
    },
    {
      "epoch": 0.12794852975361448,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001427742,
      "loss": 3.0455,
      "step": 286130
    },
    {
      "epoch": 0.12796717298977267,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001427722,
      "loss": 3.0502,
      "step": 286140
    },
    {
      "epoch": 0.12798581622593086,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001427702,
      "loss": 3.1463,
      "step": 286150
    },
    {
      "epoch": 0.12800445946208905,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001427682,
      "loss": 3.091,
      "step": 286160
    },
    {
      "epoch": 0.1280231026982472,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001427662,
      "loss": 3.0653,
      "step": 286170
    },
    {
      "epoch": 0.1280417459344054,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014276420000000002,
      "loss": 3.1241,
      "step": 286180
    },
    {
      "epoch": 0.1280603891705636,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014276220000000001,
      "loss": 3.101,
      "step": 286190
    },
    {
      "epoch": 0.12807903240672178,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001427602,
      "loss": 3.2013,
      "step": 286200
    },
    {
      "epoch": 0.12809767564287997,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001427582,
      "loss": 3.1543,
      "step": 286210
    },
    {
      "epoch": 0.12811631887903813,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001427562,
      "loss": 3.1793,
      "step": 286220
    },
    {
      "epoch": 0.12813496211519632,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014275420000000002,
      "loss": 3.1673,
      "step": 286230
    },
    {
      "epoch": 0.1281536053513545,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001427522,
      "loss": 3.1609,
      "step": 286240
    },
    {
      "epoch": 0.1281722485875127,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014275020000000001,
      "loss": 3.187,
      "step": 286250
    },
    {
      "epoch": 0.12819089182367088,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001427482,
      "loss": 3.1457,
      "step": 286260
    },
    {
      "epoch": 0.12820953505982904,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001427462,
      "loss": 3.0784,
      "step": 286270
    },
    {
      "epoch": 0.12822817829598723,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014274420000000003,
      "loss": 3.0979,
      "step": 286280
    },
    {
      "epoch": 0.12824682153214542,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001427422,
      "loss": 3.0863,
      "step": 286290
    },
    {
      "epoch": 0.1282654647683036,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014274020000000002,
      "loss": 3.1417,
      "step": 286300
    },
    {
      "epoch": 0.1282841080044618,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001427382,
      "loss": 3.1342,
      "step": 286310
    },
    {
      "epoch": 0.12830275124062,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001427362,
      "loss": 3.1676,
      "step": 286320
    },
    {
      "epoch": 0.12832139447677815,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001427342,
      "loss": 3.1833,
      "step": 286330
    },
    {
      "epoch": 0.12834003771293634,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001427322,
      "loss": 3.095,
      "step": 286340
    },
    {
      "epoch": 0.12835868094909453,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014273020000000002,
      "loss": 3.0318,
      "step": 286350
    },
    {
      "epoch": 0.12837732418525272,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001427282,
      "loss": 2.9884,
      "step": 286360
    },
    {
      "epoch": 0.1283959674214109,
      "grad_norm": 0.625,
      "learning_rate": 0.00014272620000000002,
      "loss": 3.004,
      "step": 286370
    },
    {
      "epoch": 0.12841461065756907,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014272419999999998,
      "loss": 3.3121,
      "step": 286380
    },
    {
      "epoch": 0.12843325389372726,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001427222,
      "loss": 3.1779,
      "step": 286390
    },
    {
      "epoch": 0.12845189712988545,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001427202,
      "loss": 3.1953,
      "step": 286400
    },
    {
      "epoch": 0.12847054036604363,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001427182,
      "loss": 3.0866,
      "step": 286410
    },
    {
      "epoch": 0.12848918360220182,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014271620000000002,
      "loss": 3.013,
      "step": 286420
    },
    {
      "epoch": 0.12850782683835998,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001427142,
      "loss": 3.0599,
      "step": 286430
    },
    {
      "epoch": 0.12852647007451817,
      "grad_norm": 0.625,
      "learning_rate": 0.0001427122,
      "loss": 3.0044,
      "step": 286440
    },
    {
      "epoch": 0.12854511331067636,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001427102,
      "loss": 3.1701,
      "step": 286450
    },
    {
      "epoch": 0.12856375654683455,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001427082,
      "loss": 3.2331,
      "step": 286460
    },
    {
      "epoch": 0.12858239978299274,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001427062,
      "loss": 3.1173,
      "step": 286470
    },
    {
      "epoch": 0.1286010430191509,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001427042,
      "loss": 3.1668,
      "step": 286480
    },
    {
      "epoch": 0.1286196862553091,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014270220000000002,
      "loss": 2.9296,
      "step": 286490
    },
    {
      "epoch": 0.12863832949146728,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001427002,
      "loss": 3.0392,
      "step": 286500
    },
    {
      "epoch": 0.12865697272762547,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001426982,
      "loss": 3.1796,
      "step": 286510
    },
    {
      "epoch": 0.12867561596378366,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001426962,
      "loss": 3.11,
      "step": 286520
    },
    {
      "epoch": 0.12869425919994185,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001426942,
      "loss": 3.0573,
      "step": 286530
    },
    {
      "epoch": 0.1287129024361,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014269220000000002,
      "loss": 2.9516,
      "step": 286540
    },
    {
      "epoch": 0.1287315456722582,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001426902,
      "loss": 3.0626,
      "step": 286550
    },
    {
      "epoch": 0.12875018890841639,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014268820000000001,
      "loss": 3.2466,
      "step": 286560
    },
    {
      "epoch": 0.12876883214457457,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001426862,
      "loss": 3.2188,
      "step": 286570
    },
    {
      "epoch": 0.12878747538073276,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001426842,
      "loss": 3.0326,
      "step": 286580
    },
    {
      "epoch": 0.12880611861689092,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001426822,
      "loss": 3.1825,
      "step": 286590
    },
    {
      "epoch": 0.1288247618530491,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001426802,
      "loss": 3.1227,
      "step": 286600
    },
    {
      "epoch": 0.1288434050892073,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014267820000000002,
      "loss": 3.0267,
      "step": 286610
    },
    {
      "epoch": 0.1288620483253655,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014267620000000001,
      "loss": 3.2686,
      "step": 286620
    },
    {
      "epoch": 0.12888069156152368,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001426742,
      "loss": 3.1417,
      "step": 286630
    },
    {
      "epoch": 0.12889933479768184,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001426722,
      "loss": 3.066,
      "step": 286640
    },
    {
      "epoch": 0.12891797803384003,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001426702,
      "loss": 3.0851,
      "step": 286650
    },
    {
      "epoch": 0.12893662126999822,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001426682,
      "loss": 3.0653,
      "step": 286660
    },
    {
      "epoch": 0.1289552645061564,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014266620000000002,
      "loss": 3.1074,
      "step": 286670
    },
    {
      "epoch": 0.1289739077423146,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014266420000000002,
      "loss": 3.0776,
      "step": 286680
    },
    {
      "epoch": 0.12899255097847276,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001426622,
      "loss": 3.113,
      "step": 286690
    },
    {
      "epoch": 0.12901119421463095,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001426602,
      "loss": 3.154,
      "step": 286700
    },
    {
      "epoch": 0.12902983745078914,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001426582,
      "loss": 3.0695,
      "step": 286710
    },
    {
      "epoch": 0.12904848068694733,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014265620000000003,
      "loss": 3.2522,
      "step": 286720
    },
    {
      "epoch": 0.12906712392310551,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001426542,
      "loss": 3.0353,
      "step": 286730
    },
    {
      "epoch": 0.1290857671592637,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014265220000000002,
      "loss": 3.2092,
      "step": 286740
    },
    {
      "epoch": 0.12910441039542186,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001426502,
      "loss": 3.1056,
      "step": 286750
    },
    {
      "epoch": 0.12912305363158005,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001426482,
      "loss": 2.9022,
      "step": 286760
    },
    {
      "epoch": 0.12914169686773824,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001426462,
      "loss": 3.0531,
      "step": 286770
    },
    {
      "epoch": 0.12916034010389643,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001426442,
      "loss": 3.0164,
      "step": 286780
    },
    {
      "epoch": 0.12917898334005462,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014264220000000002,
      "loss": 3.1488,
      "step": 286790
    },
    {
      "epoch": 0.12919762657621278,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001426402,
      "loss": 3.0604,
      "step": 286800
    },
    {
      "epoch": 0.12921626981237097,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001426382,
      "loss": 3.1882,
      "step": 286810
    },
    {
      "epoch": 0.12923491304852916,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001426362,
      "loss": 3.1468,
      "step": 286820
    },
    {
      "epoch": 0.12925355628468735,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001426342,
      "loss": 3.2327,
      "step": 286830
    },
    {
      "epoch": 0.12927219952084554,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014263220000000003,
      "loss": 3.1177,
      "step": 286840
    },
    {
      "epoch": 0.1292908427570037,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001426302,
      "loss": 3.1466,
      "step": 286850
    },
    {
      "epoch": 0.1293094859931619,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014262820000000002,
      "loss": 3.0007,
      "step": 286860
    },
    {
      "epoch": 0.12932812922932008,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014262619999999999,
      "loss": 3.2402,
      "step": 286870
    },
    {
      "epoch": 0.12934677246547827,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001426242,
      "loss": 3.1405,
      "step": 286880
    },
    {
      "epoch": 0.12936541570163645,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001426222,
      "loss": 3.2498,
      "step": 286890
    },
    {
      "epoch": 0.12938405893779462,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001426202,
      "loss": 3.1337,
      "step": 286900
    },
    {
      "epoch": 0.1294027021739528,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014261820000000002,
      "loss": 3.1403,
      "step": 286910
    },
    {
      "epoch": 0.129421345410111,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001426162,
      "loss": 3.0394,
      "step": 286920
    },
    {
      "epoch": 0.12943998864626918,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014261420000000001,
      "loss": 3.0743,
      "step": 286930
    },
    {
      "epoch": 0.12945863188242737,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014261219999999998,
      "loss": 3.0048,
      "step": 286940
    },
    {
      "epoch": 0.12947727511858556,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001426102,
      "loss": 3.1101,
      "step": 286950
    },
    {
      "epoch": 0.12949591835474372,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001426082,
      "loss": 3.2048,
      "step": 286960
    },
    {
      "epoch": 0.1295145615909019,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001426062,
      "loss": 3.1985,
      "step": 286970
    },
    {
      "epoch": 0.1295332048270601,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014260420000000002,
      "loss": 3.0451,
      "step": 286980
    },
    {
      "epoch": 0.1295518480632183,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001426022,
      "loss": 2.9719,
      "step": 286990
    },
    {
      "epoch": 0.12957049129937648,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001426002,
      "loss": 3.1943,
      "step": 287000
    },
    {
      "epoch": 0.12958913453553464,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001425982,
      "loss": 3.2635,
      "step": 287010
    },
    {
      "epoch": 0.12960777777169283,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001425962,
      "loss": 3.0844,
      "step": 287020
    },
    {
      "epoch": 0.12962642100785102,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014259420000000002,
      "loss": 3.1103,
      "step": 287030
    },
    {
      "epoch": 0.1296450642440092,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001425922,
      "loss": 3.1303,
      "step": 287040
    },
    {
      "epoch": 0.1296637074801674,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014259020000000002,
      "loss": 2.9602,
      "step": 287050
    },
    {
      "epoch": 0.12968235071632556,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001425882,
      "loss": 3.2127,
      "step": 287060
    },
    {
      "epoch": 0.12970099395248375,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001425862,
      "loss": 3.0587,
      "step": 287070
    },
    {
      "epoch": 0.12971963718864193,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001425842,
      "loss": 3.1392,
      "step": 287080
    },
    {
      "epoch": 0.12973828042480012,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001425822,
      "loss": 3.0458,
      "step": 287090
    },
    {
      "epoch": 0.1297569236609583,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014258020000000002,
      "loss": 3.111,
      "step": 287100
    },
    {
      "epoch": 0.12977556689711647,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014257820000000002,
      "loss": 3.0455,
      "step": 287110
    },
    {
      "epoch": 0.12979421013327466,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001425762,
      "loss": 3.1362,
      "step": 287120
    },
    {
      "epoch": 0.12981285336943285,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001425742,
      "loss": 3.0213,
      "step": 287130
    },
    {
      "epoch": 0.12983149660559104,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001425722,
      "loss": 3.2578,
      "step": 287140
    },
    {
      "epoch": 0.12985013984174923,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001425702,
      "loss": 3.1797,
      "step": 287150
    },
    {
      "epoch": 0.1298687830779074,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001425682,
      "loss": 3.1586,
      "step": 287160
    },
    {
      "epoch": 0.12988742631406558,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014256620000000002,
      "loss": 3.0944,
      "step": 287170
    },
    {
      "epoch": 0.12990606955022377,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001425642,
      "loss": 3.0971,
      "step": 287180
    },
    {
      "epoch": 0.12992471278638196,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001425622,
      "loss": 3.1293,
      "step": 287190
    },
    {
      "epoch": 0.12994335602254015,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001425602,
      "loss": 3.0065,
      "step": 287200
    },
    {
      "epoch": 0.12996199925869834,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001425582,
      "loss": 3.1333,
      "step": 287210
    },
    {
      "epoch": 0.1299806424948565,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001425562,
      "loss": 3.1661,
      "step": 287220
    },
    {
      "epoch": 0.12999928573101469,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014255420000000002,
      "loss": 3.275,
      "step": 287230
    },
    {
      "epoch": 0.13001792896717287,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014255220000000001,
      "loss": 3.1817,
      "step": 287240
    },
    {
      "epoch": 0.13003657220333106,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001425502,
      "loss": 3.1828,
      "step": 287250
    },
    {
      "epoch": 0.13005521543948925,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001425482,
      "loss": 3.2129,
      "step": 287260
    },
    {
      "epoch": 0.1300738586756474,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001425462,
      "loss": 3.2056,
      "step": 287270
    },
    {
      "epoch": 0.1300925019118056,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014254420000000002,
      "loss": 3.078,
      "step": 287280
    },
    {
      "epoch": 0.1301111451479638,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001425422,
      "loss": 3.0354,
      "step": 287290
    },
    {
      "epoch": 0.13012978838412198,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014254020000000001,
      "loss": 3.1001,
      "step": 287300
    },
    {
      "epoch": 0.13014843162028017,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001425382,
      "loss": 3.1362,
      "step": 287310
    },
    {
      "epoch": 0.13016707485643833,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001425362,
      "loss": 3.1184,
      "step": 287320
    },
    {
      "epoch": 0.13018571809259652,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014253420000000003,
      "loss": 3.1309,
      "step": 287330
    },
    {
      "epoch": 0.1302043613287547,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001425322,
      "loss": 3.1993,
      "step": 287340
    },
    {
      "epoch": 0.1302230045649129,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014253020000000002,
      "loss": 3.2499,
      "step": 287350
    },
    {
      "epoch": 0.1302416478010711,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001425282,
      "loss": 2.9587,
      "step": 287360
    },
    {
      "epoch": 0.13026029103722925,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001425262,
      "loss": 3.0919,
      "step": 287370
    },
    {
      "epoch": 0.13027893427338744,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001425242,
      "loss": 2.9585,
      "step": 287380
    },
    {
      "epoch": 0.13029757750954563,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001425222,
      "loss": 2.999,
      "step": 287390
    },
    {
      "epoch": 0.13031622074570381,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014252020000000003,
      "loss": 3.1158,
      "step": 287400
    },
    {
      "epoch": 0.130334863981862,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001425182,
      "loss": 3.2387,
      "step": 287410
    },
    {
      "epoch": 0.1303535072180202,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014251620000000002,
      "loss": 3.1674,
      "step": 287420
    },
    {
      "epoch": 0.13037215045417835,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014251419999999998,
      "loss": 3.1551,
      "step": 287430
    },
    {
      "epoch": 0.13039079369033654,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001425122,
      "loss": 3.1115,
      "step": 287440
    },
    {
      "epoch": 0.13040943692649473,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001425102,
      "loss": 3.1977,
      "step": 287450
    },
    {
      "epoch": 0.13042808016265292,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001425082,
      "loss": 3.0583,
      "step": 287460
    },
    {
      "epoch": 0.1304467233988111,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014250620000000002,
      "loss": 3.1096,
      "step": 287470
    },
    {
      "epoch": 0.13046536663496927,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001425042,
      "loss": 3.3352,
      "step": 287480
    },
    {
      "epoch": 0.13048400987112746,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001425022,
      "loss": 3.1601,
      "step": 287490
    },
    {
      "epoch": 0.13050265310728565,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001425002,
      "loss": 3.0841,
      "step": 287500
    },
    {
      "epoch": 0.13052129634344384,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001424982,
      "loss": 3.0377,
      "step": 287510
    },
    {
      "epoch": 0.13053993957960203,
      "grad_norm": 0.75,
      "learning_rate": 0.0001424962,
      "loss": 3.1423,
      "step": 287520
    },
    {
      "epoch": 0.1305585828157602,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001424942,
      "loss": 3.2662,
      "step": 287530
    },
    {
      "epoch": 0.13057722605191838,
      "grad_norm": 0.859375,
      "learning_rate": 0.00014249220000000002,
      "loss": 3.1045,
      "step": 287540
    },
    {
      "epoch": 0.13059586928807657,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014249019999999999,
      "loss": 3.129,
      "step": 287550
    },
    {
      "epoch": 0.13061451252423475,
      "grad_norm": 0.625,
      "learning_rate": 0.0001424882,
      "loss": 3.1272,
      "step": 287560
    },
    {
      "epoch": 0.13063315576039294,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001424862,
      "loss": 3.2521,
      "step": 287570
    },
    {
      "epoch": 0.1306517989965511,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001424842,
      "loss": 3.107,
      "step": 287580
    },
    {
      "epoch": 0.1306704422327093,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014248220000000002,
      "loss": 3.2741,
      "step": 287590
    },
    {
      "epoch": 0.13068908546886748,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001424802,
      "loss": 3.0464,
      "step": 287600
    },
    {
      "epoch": 0.13070772870502567,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014247820000000001,
      "loss": 3.2249,
      "step": 287610
    },
    {
      "epoch": 0.13072637194118386,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001424762,
      "loss": 3.1988,
      "step": 287620
    },
    {
      "epoch": 0.13074501517734205,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001424742,
      "loss": 3.1978,
      "step": 287630
    },
    {
      "epoch": 0.1307636584135002,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001424722,
      "loss": 3.0943,
      "step": 287640
    },
    {
      "epoch": 0.1307823016496584,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001424702,
      "loss": 3.1315,
      "step": 287650
    },
    {
      "epoch": 0.1308009448858166,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014246820000000002,
      "loss": 3.0798,
      "step": 287660
    },
    {
      "epoch": 0.13081958812197478,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014246620000000002,
      "loss": 3.134,
      "step": 287670
    },
    {
      "epoch": 0.13083823135813297,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001424642,
      "loss": 3.0636,
      "step": 287680
    },
    {
      "epoch": 0.13085687459429113,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001424622,
      "loss": 3.0562,
      "step": 287690
    },
    {
      "epoch": 0.13087551783044932,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001424602,
      "loss": 3.2087,
      "step": 287700
    },
    {
      "epoch": 0.1308941610666075,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001424582,
      "loss": 3.217,
      "step": 287710
    },
    {
      "epoch": 0.1309128043027657,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014245620000000002,
      "loss": 3.2548,
      "step": 287720
    },
    {
      "epoch": 0.13093144753892388,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014245420000000002,
      "loss": 3.2339,
      "step": 287730
    },
    {
      "epoch": 0.13095009077508205,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001424522,
      "loss": 3.0601,
      "step": 287740
    },
    {
      "epoch": 0.13096873401124023,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001424502,
      "loss": 3.1417,
      "step": 287750
    },
    {
      "epoch": 0.13098737724739842,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001424482,
      "loss": 3.1655,
      "step": 287760
    },
    {
      "epoch": 0.1310060204835566,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001424462,
      "loss": 3.155,
      "step": 287770
    },
    {
      "epoch": 0.1310246637197148,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001424442,
      "loss": 3.1548,
      "step": 287780
    },
    {
      "epoch": 0.13104330695587296,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014244220000000002,
      "loss": 3.0206,
      "step": 287790
    },
    {
      "epoch": 0.13106195019203115,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001424402,
      "loss": 3.1044,
      "step": 287800
    },
    {
      "epoch": 0.13108059342818934,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001424382,
      "loss": 3.1015,
      "step": 287810
    },
    {
      "epoch": 0.13109923666434753,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001424362,
      "loss": 3.1208,
      "step": 287820
    },
    {
      "epoch": 0.13111787990050572,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001424342,
      "loss": 3.2361,
      "step": 287830
    },
    {
      "epoch": 0.13113652313666388,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014243220000000002,
      "loss": 3.1493,
      "step": 287840
    },
    {
      "epoch": 0.13115516637282207,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001424302,
      "loss": 3.0449,
      "step": 287850
    },
    {
      "epoch": 0.13117380960898026,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001424282,
      "loss": 2.9662,
      "step": 287860
    },
    {
      "epoch": 0.13119245284513845,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001424262,
      "loss": 3.1348,
      "step": 287870
    },
    {
      "epoch": 0.13121109608129664,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001424242,
      "loss": 3.1232,
      "step": 287880
    },
    {
      "epoch": 0.13122973931745482,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014242220000000003,
      "loss": 3.034,
      "step": 287890
    },
    {
      "epoch": 0.13124838255361299,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001424202,
      "loss": 3.0847,
      "step": 287900
    },
    {
      "epoch": 0.13126702578977117,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014241820000000002,
      "loss": 3.1127,
      "step": 287910
    },
    {
      "epoch": 0.13128566902592936,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001424162,
      "loss": 2.9846,
      "step": 287920
    },
    {
      "epoch": 0.13130431226208755,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001424142,
      "loss": 3.1825,
      "step": 287930
    },
    {
      "epoch": 0.13132295549824574,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001424122,
      "loss": 3.2201,
      "step": 287940
    },
    {
      "epoch": 0.1313415987344039,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001424102,
      "loss": 3.0531,
      "step": 287950
    },
    {
      "epoch": 0.1313602419705621,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014240820000000002,
      "loss": 3.0748,
      "step": 287960
    },
    {
      "epoch": 0.13137888520672028,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001424062,
      "loss": 3.2025,
      "step": 287970
    },
    {
      "epoch": 0.13139752844287847,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014240420000000001,
      "loss": 3.1835,
      "step": 287980
    },
    {
      "epoch": 0.13141617167903666,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001424022,
      "loss": 3.1501,
      "step": 287990
    },
    {
      "epoch": 0.13143481491519482,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001424002,
      "loss": 3.3703,
      "step": 288000
    },
    {
      "epoch": 0.131453458151353,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001423982,
      "loss": 3.1058,
      "step": 288010
    },
    {
      "epoch": 0.1314721013875112,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001423962,
      "loss": 3.1837,
      "step": 288020
    },
    {
      "epoch": 0.1314907446236694,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014239420000000002,
      "loss": 3.2291,
      "step": 288030
    },
    {
      "epoch": 0.13150938785982758,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001423922,
      "loss": 3.1717,
      "step": 288040
    },
    {
      "epoch": 0.13152803109598574,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001423902,
      "loss": 3.1026,
      "step": 288050
    },
    {
      "epoch": 0.13154667433214393,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001423882,
      "loss": 3.1239,
      "step": 288060
    },
    {
      "epoch": 0.13156531756830211,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001423862,
      "loss": 3.0022,
      "step": 288070
    },
    {
      "epoch": 0.1315839608044603,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014238420000000002,
      "loss": 3.2109,
      "step": 288080
    },
    {
      "epoch": 0.1316026040406185,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001423822,
      "loss": 3.0586,
      "step": 288090
    },
    {
      "epoch": 0.13162124727677668,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014238020000000002,
      "loss": 3.0868,
      "step": 288100
    },
    {
      "epoch": 0.13163989051293484,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001423782,
      "loss": 3.0057,
      "step": 288110
    },
    {
      "epoch": 0.13165853374909303,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001423762,
      "loss": 3.197,
      "step": 288120
    },
    {
      "epoch": 0.13167717698525122,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001423742,
      "loss": 3.1024,
      "step": 288130
    },
    {
      "epoch": 0.1316958202214094,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001423722,
      "loss": 3.0251,
      "step": 288140
    },
    {
      "epoch": 0.1317144634575676,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014237020000000002,
      "loss": 3.117,
      "step": 288150
    },
    {
      "epoch": 0.13173310669372576,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001423682,
      "loss": 3.2212,
      "step": 288160
    },
    {
      "epoch": 0.13175174992988395,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001423662,
      "loss": 3.057,
      "step": 288170
    },
    {
      "epoch": 0.13177039316604214,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001423642,
      "loss": 3.2046,
      "step": 288180
    },
    {
      "epoch": 0.13178903640220033,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001423622,
      "loss": 3.1377,
      "step": 288190
    },
    {
      "epoch": 0.13180767963835852,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001423602,
      "loss": 3.004,
      "step": 288200
    },
    {
      "epoch": 0.13182632287451668,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001423582,
      "loss": 3.076,
      "step": 288210
    },
    {
      "epoch": 0.13184496611067487,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014235620000000002,
      "loss": 3.158,
      "step": 288220
    },
    {
      "epoch": 0.13186360934683306,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001423542,
      "loss": 3.1483,
      "step": 288230
    },
    {
      "epoch": 0.13188225258299124,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001423522,
      "loss": 3.0605,
      "step": 288240
    },
    {
      "epoch": 0.13190089581914943,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001423502,
      "loss": 2.9939,
      "step": 288250
    },
    {
      "epoch": 0.1319195390553076,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001423482,
      "loss": 3.1838,
      "step": 288260
    },
    {
      "epoch": 0.13193818229146578,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001423462,
      "loss": 3.1073,
      "step": 288270
    },
    {
      "epoch": 0.13195682552762397,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014234420000000002,
      "loss": 3.1698,
      "step": 288280
    },
    {
      "epoch": 0.13197546876378216,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014234220000000001,
      "loss": 3.1812,
      "step": 288290
    },
    {
      "epoch": 0.13199411199994035,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001423402,
      "loss": 3.1125,
      "step": 288300
    },
    {
      "epoch": 0.13201275523609854,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001423382,
      "loss": 3.163,
      "step": 288310
    },
    {
      "epoch": 0.1320313984722567,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001423362,
      "loss": 3.2153,
      "step": 288320
    },
    {
      "epoch": 0.1320500417084149,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014233420000000002,
      "loss": 3.0311,
      "step": 288330
    },
    {
      "epoch": 0.13206868494457308,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001423322,
      "loss": 3.1331,
      "step": 288340
    },
    {
      "epoch": 0.13208732818073127,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014233020000000001,
      "loss": 3.0967,
      "step": 288350
    },
    {
      "epoch": 0.13210597141688946,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001423282,
      "loss": 3.1478,
      "step": 288360
    },
    {
      "epoch": 0.13212461465304762,
      "grad_norm": 0.75,
      "learning_rate": 0.0001423262,
      "loss": 3.0714,
      "step": 288370
    },
    {
      "epoch": 0.1321432578892058,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001423242,
      "loss": 3.0997,
      "step": 288380
    },
    {
      "epoch": 0.132161901125364,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001423222,
      "loss": 3.1724,
      "step": 288390
    },
    {
      "epoch": 0.13218054436152218,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014232020000000002,
      "loss": 3.1445,
      "step": 288400
    },
    {
      "epoch": 0.13219918759768037,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001423182,
      "loss": 3.1084,
      "step": 288410
    },
    {
      "epoch": 0.13221783083383853,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001423162,
      "loss": 3.1458,
      "step": 288420
    },
    {
      "epoch": 0.13223647406999672,
      "grad_norm": 0.75,
      "learning_rate": 0.0001423142,
      "loss": 3.2169,
      "step": 288430
    },
    {
      "epoch": 0.1322551173061549,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001423122,
      "loss": 3.0955,
      "step": 288440
    },
    {
      "epoch": 0.1322737605423131,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014231020000000003,
      "loss": 3.1227,
      "step": 288450
    },
    {
      "epoch": 0.1322924037784713,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001423082,
      "loss": 3.0991,
      "step": 288460
    },
    {
      "epoch": 0.13231104701462945,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014230620000000002,
      "loss": 3.1524,
      "step": 288470
    },
    {
      "epoch": 0.13232969025078764,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014230419999999999,
      "loss": 3.1248,
      "step": 288480
    },
    {
      "epoch": 0.13234833348694583,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001423022,
      "loss": 3.1212,
      "step": 288490
    },
    {
      "epoch": 0.13236697672310402,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001423002,
      "loss": 3.2167,
      "step": 288500
    },
    {
      "epoch": 0.1323856199592622,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001422982,
      "loss": 3.0996,
      "step": 288510
    },
    {
      "epoch": 0.13240426319542037,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014229620000000002,
      "loss": 3.1497,
      "step": 288520
    },
    {
      "epoch": 0.13242290643157856,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001422942,
      "loss": 3.1364,
      "step": 288530
    },
    {
      "epoch": 0.13244154966773675,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001422922,
      "loss": 3.1573,
      "step": 288540
    },
    {
      "epoch": 0.13246019290389494,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001422902,
      "loss": 2.9819,
      "step": 288550
    },
    {
      "epoch": 0.13247883614005312,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001422882,
      "loss": 3.2471,
      "step": 288560
    },
    {
      "epoch": 0.1324974793762113,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014228620000000003,
      "loss": 3.1178,
      "step": 288570
    },
    {
      "epoch": 0.13251612261236947,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001422842,
      "loss": 3.1089,
      "step": 288580
    },
    {
      "epoch": 0.13253476584852766,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014228220000000002,
      "loss": 3.104,
      "step": 288590
    },
    {
      "epoch": 0.13255340908468585,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001422802,
      "loss": 3.1668,
      "step": 288600
    },
    {
      "epoch": 0.13257205232084404,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001422782,
      "loss": 3.1472,
      "step": 288610
    },
    {
      "epoch": 0.13259069555700223,
      "grad_norm": 0.75,
      "learning_rate": 0.0001422762,
      "loss": 3.1072,
      "step": 288620
    },
    {
      "epoch": 0.1326093387931604,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001422742,
      "loss": 3.0901,
      "step": 288630
    },
    {
      "epoch": 0.13262798202931858,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014227220000000002,
      "loss": 3.1304,
      "step": 288640
    },
    {
      "epoch": 0.13264662526547677,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001422702,
      "loss": 3.059,
      "step": 288650
    },
    {
      "epoch": 0.13266526850163496,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014226820000000001,
      "loss": 3.0436,
      "step": 288660
    },
    {
      "epoch": 0.13268391173779315,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001422662,
      "loss": 3.1227,
      "step": 288670
    },
    {
      "epoch": 0.1327025549739513,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001422642,
      "loss": 3.0947,
      "step": 288680
    },
    {
      "epoch": 0.1327211982101095,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001422622,
      "loss": 3.0964,
      "step": 288690
    },
    {
      "epoch": 0.1327398414462677,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001422602,
      "loss": 3.16,
      "step": 288700
    },
    {
      "epoch": 0.13275848468242588,
      "grad_norm": 0.859375,
      "learning_rate": 0.00014225820000000002,
      "loss": 2.9591,
      "step": 288710
    },
    {
      "epoch": 0.13277712791858406,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014225620000000002,
      "loss": 3.1469,
      "step": 288720
    },
    {
      "epoch": 0.13279577115474223,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001422542,
      "loss": 3.0861,
      "step": 288730
    },
    {
      "epoch": 0.13281441439090041,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001422522,
      "loss": 3.2474,
      "step": 288740
    },
    {
      "epoch": 0.1328330576270586,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001422502,
      "loss": 3.2308,
      "step": 288750
    },
    {
      "epoch": 0.1328517008632168,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001422482,
      "loss": 3.1853,
      "step": 288760
    },
    {
      "epoch": 0.13287034409937498,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001422462,
      "loss": 3.1535,
      "step": 288770
    },
    {
      "epoch": 0.13288898733553317,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014224420000000002,
      "loss": 3.2066,
      "step": 288780
    },
    {
      "epoch": 0.13290763057169133,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001422422,
      "loss": 3.1308,
      "step": 288790
    },
    {
      "epoch": 0.13292627380784952,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001422402,
      "loss": 3.0741,
      "step": 288800
    },
    {
      "epoch": 0.1329449170440077,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001422382,
      "loss": 3.0217,
      "step": 288810
    },
    {
      "epoch": 0.1329635602801659,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001422362,
      "loss": 3.082,
      "step": 288820
    },
    {
      "epoch": 0.1329822035163241,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001422342,
      "loss": 3.3041,
      "step": 288830
    },
    {
      "epoch": 0.13300084675248225,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014223220000000002,
      "loss": 3.0798,
      "step": 288840
    },
    {
      "epoch": 0.13301948998864044,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001422302,
      "loss": 3.1545,
      "step": 288850
    },
    {
      "epoch": 0.13303813322479863,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001422282,
      "loss": 3.2927,
      "step": 288860
    },
    {
      "epoch": 0.13305677646095682,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001422262,
      "loss": 3.1974,
      "step": 288870
    },
    {
      "epoch": 0.133075419697115,
      "grad_norm": 0.625,
      "learning_rate": 0.0001422242,
      "loss": 3.069,
      "step": 288880
    },
    {
      "epoch": 0.13309406293327317,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014222220000000002,
      "loss": 3.067,
      "step": 288890
    },
    {
      "epoch": 0.13311270616943136,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001422202,
      "loss": 3.1556,
      "step": 288900
    },
    {
      "epoch": 0.13313134940558954,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001422182,
      "loss": 3.132,
      "step": 288910
    },
    {
      "epoch": 0.13314999264174773,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001422162,
      "loss": 3.2308,
      "step": 288920
    },
    {
      "epoch": 0.13316863587790592,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001422142,
      "loss": 3.2425,
      "step": 288930
    },
    {
      "epoch": 0.13318727911406408,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014221220000000003,
      "loss": 3.1305,
      "step": 288940
    },
    {
      "epoch": 0.13320592235022227,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001422102,
      "loss": 3.1138,
      "step": 288950
    },
    {
      "epoch": 0.13322456558638046,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014220820000000002,
      "loss": 3.1481,
      "step": 288960
    },
    {
      "epoch": 0.13324320882253865,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001422062,
      "loss": 3.1961,
      "step": 288970
    },
    {
      "epoch": 0.13326185205869684,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001422042,
      "loss": 3.2074,
      "step": 288980
    },
    {
      "epoch": 0.13328049529485503,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001422022,
      "loss": 3.0151,
      "step": 288990
    },
    {
      "epoch": 0.1332991385310132,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001422002,
      "loss": 3.1648,
      "step": 289000
    },
    {
      "epoch": 0.13331778176717138,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014219820000000002,
      "loss": 3.2155,
      "step": 289010
    },
    {
      "epoch": 0.13333642500332957,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001421962,
      "loss": 3.3469,
      "step": 289020
    },
    {
      "epoch": 0.13335506823948776,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014219420000000001,
      "loss": 3.1444,
      "step": 289030
    },
    {
      "epoch": 0.13337371147564595,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001421922,
      "loss": 3.142,
      "step": 289040
    },
    {
      "epoch": 0.1333923547118041,
      "grad_norm": 0.75,
      "learning_rate": 0.0001421902,
      "loss": 3.151,
      "step": 289050
    },
    {
      "epoch": 0.1334109979479623,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014218820000000003,
      "loss": 3.1912,
      "step": 289060
    },
    {
      "epoch": 0.13342964118412048,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001421862,
      "loss": 3.2814,
      "step": 289070
    },
    {
      "epoch": 0.13344828442027867,
      "grad_norm": 0.75,
      "learning_rate": 0.00014218420000000002,
      "loss": 3.0348,
      "step": 289080
    },
    {
      "epoch": 0.13346692765643686,
      "grad_norm": 0.75,
      "learning_rate": 0.0001421822,
      "loss": 3.0729,
      "step": 289090
    },
    {
      "epoch": 0.13348557089259502,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001421802,
      "loss": 3.1133,
      "step": 289100
    },
    {
      "epoch": 0.1335042141287532,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001421782,
      "loss": 3.0646,
      "step": 289110
    },
    {
      "epoch": 0.1335228573649114,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001421762,
      "loss": 3.1356,
      "step": 289120
    },
    {
      "epoch": 0.1335415006010696,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014217420000000003,
      "loss": 3.1434,
      "step": 289130
    },
    {
      "epoch": 0.13356014383722778,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001421722,
      "loss": 3.0441,
      "step": 289140
    },
    {
      "epoch": 0.13357878707338594,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00014217020000000002,
      "loss": 3.1104,
      "step": 289150
    },
    {
      "epoch": 0.13359743030954413,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014216819999999999,
      "loss": 3.0913,
      "step": 289160
    },
    {
      "epoch": 0.13361607354570232,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001421662,
      "loss": 3.2015,
      "step": 289170
    },
    {
      "epoch": 0.1336347167818605,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001421642,
      "loss": 3.1339,
      "step": 289180
    },
    {
      "epoch": 0.1336533600180187,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001421622,
      "loss": 3.0647,
      "step": 289190
    },
    {
      "epoch": 0.13367200325417686,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014216020000000002,
      "loss": 3.1464,
      "step": 289200
    },
    {
      "epoch": 0.13369064649033505,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001421582,
      "loss": 3.2037,
      "step": 289210
    },
    {
      "epoch": 0.13370928972649324,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001421562,
      "loss": 3.1521,
      "step": 289220
    },
    {
      "epoch": 0.13372793296265142,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001421542,
      "loss": 3.1871,
      "step": 289230
    },
    {
      "epoch": 0.1337465761988096,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001421522,
      "loss": 3.1241,
      "step": 289240
    },
    {
      "epoch": 0.1337652194349678,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001421502,
      "loss": 3.1163,
      "step": 289250
    },
    {
      "epoch": 0.13378386267112596,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001421482,
      "loss": 3.0797,
      "step": 289260
    },
    {
      "epoch": 0.13380250590728415,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014214620000000002,
      "loss": 3.0658,
      "step": 289270
    },
    {
      "epoch": 0.13382114914344234,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014214420000000001,
      "loss": 3.1377,
      "step": 289280
    },
    {
      "epoch": 0.13383979237960053,
      "grad_norm": 0.625,
      "learning_rate": 0.0001421422,
      "loss": 3.1261,
      "step": 289290
    },
    {
      "epoch": 0.13385843561575872,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001421402,
      "loss": 3.1065,
      "step": 289300
    },
    {
      "epoch": 0.13387707885191688,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001421382,
      "loss": 3.0921,
      "step": 289310
    },
    {
      "epoch": 0.13389572208807507,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001421362,
      "loss": 3.2028,
      "step": 289320
    },
    {
      "epoch": 0.13391436532423326,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014213420000000002,
      "loss": 3.1863,
      "step": 289330
    },
    {
      "epoch": 0.13393300856039145,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014213220000000001,
      "loss": 3.128,
      "step": 289340
    },
    {
      "epoch": 0.13395165179654964,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001421302,
      "loss": 3.0454,
      "step": 289350
    },
    {
      "epoch": 0.1339702950327078,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001421282,
      "loss": 3.0557,
      "step": 289360
    },
    {
      "epoch": 0.133988938268866,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001421262,
      "loss": 3.1918,
      "step": 289370
    },
    {
      "epoch": 0.13400758150502418,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001421242,
      "loss": 3.1774,
      "step": 289380
    },
    {
      "epoch": 0.13402622474118236,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001421222,
      "loss": 3.1466,
      "step": 289390
    },
    {
      "epoch": 0.13404486797734055,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014212020000000002,
      "loss": 3.1807,
      "step": 289400
    },
    {
      "epoch": 0.13406351121349872,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001421182,
      "loss": 3.2696,
      "step": 289410
    },
    {
      "epoch": 0.1340821544496569,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001421162,
      "loss": 3.1886,
      "step": 289420
    },
    {
      "epoch": 0.1341007976858151,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001421142,
      "loss": 3.1466,
      "step": 289430
    },
    {
      "epoch": 0.13411944092197328,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001421122,
      "loss": 3.3046,
      "step": 289440
    },
    {
      "epoch": 0.13413808415813147,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014211020000000002,
      "loss": 3.0753,
      "step": 289450
    },
    {
      "epoch": 0.13415672739428966,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001421082,
      "loss": 3.0856,
      "step": 289460
    },
    {
      "epoch": 0.13417537063044782,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001421062,
      "loss": 3.0897,
      "step": 289470
    },
    {
      "epoch": 0.134194013866606,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001421042,
      "loss": 3.0153,
      "step": 289480
    },
    {
      "epoch": 0.1342126571027642,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001421022,
      "loss": 3.042,
      "step": 289490
    },
    {
      "epoch": 0.1342313003389224,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014210020000000003,
      "loss": 3.0873,
      "step": 289500
    },
    {
      "epoch": 0.13424994357508058,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001420982,
      "loss": 3.2038,
      "step": 289510
    },
    {
      "epoch": 0.13426858681123874,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014209620000000002,
      "loss": 3.2803,
      "step": 289520
    },
    {
      "epoch": 0.13428723004739693,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001420942,
      "loss": 3.2083,
      "step": 289530
    },
    {
      "epoch": 0.13430587328355512,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001420922,
      "loss": 3.0444,
      "step": 289540
    },
    {
      "epoch": 0.1343245165197133,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014209020000000003,
      "loss": 3.0446,
      "step": 289550
    },
    {
      "epoch": 0.1343431597558715,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001420882,
      "loss": 3.0648,
      "step": 289560
    },
    {
      "epoch": 0.13436180299202966,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014208620000000002,
      "loss": 3.0229,
      "step": 289570
    },
    {
      "epoch": 0.13438044622818784,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001420842,
      "loss": 3.0536,
      "step": 289580
    },
    {
      "epoch": 0.13439908946434603,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001420822,
      "loss": 3.0631,
      "step": 289590
    },
    {
      "epoch": 0.13441773270050422,
      "grad_norm": 0.75,
      "learning_rate": 0.0001420802,
      "loss": 3.0325,
      "step": 289600
    },
    {
      "epoch": 0.1344363759366624,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001420782,
      "loss": 3.2081,
      "step": 289610
    },
    {
      "epoch": 0.13445501917282057,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014207620000000003,
      "loss": 3.218,
      "step": 289620
    },
    {
      "epoch": 0.13447366240897876,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001420742,
      "loss": 3.0602,
      "step": 289630
    },
    {
      "epoch": 0.13449230564513695,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014207220000000002,
      "loss": 3.0602,
      "step": 289640
    },
    {
      "epoch": 0.13451094888129514,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001420702,
      "loss": 3.0786,
      "step": 289650
    },
    {
      "epoch": 0.13452959211745333,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001420682,
      "loss": 3.1217,
      "step": 289660
    },
    {
      "epoch": 0.13454823535361152,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001420662,
      "loss": 3.1153,
      "step": 289670
    },
    {
      "epoch": 0.13456687858976968,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001420642,
      "loss": 3.1671,
      "step": 289680
    },
    {
      "epoch": 0.13458552182592787,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014206220000000002,
      "loss": 3.1697,
      "step": 289690
    },
    {
      "epoch": 0.13460416506208606,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001420602,
      "loss": 2.9834,
      "step": 289700
    },
    {
      "epoch": 0.13462280829824425,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014205820000000001,
      "loss": 3.1582,
      "step": 289710
    },
    {
      "epoch": 0.13464145153440243,
      "grad_norm": 0.625,
      "learning_rate": 0.0001420562,
      "loss": 3.0832,
      "step": 289720
    },
    {
      "epoch": 0.1346600947705606,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001420542,
      "loss": 3.1032,
      "step": 289730
    },
    {
      "epoch": 0.13467873800671878,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001420522,
      "loss": 3.155,
      "step": 289740
    },
    {
      "epoch": 0.13469738124287697,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001420502,
      "loss": 3.0641,
      "step": 289750
    },
    {
      "epoch": 0.13471602447903516,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014204820000000002,
      "loss": 3.1168,
      "step": 289760
    },
    {
      "epoch": 0.13473466771519335,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001420462,
      "loss": 3.1515,
      "step": 289770
    },
    {
      "epoch": 0.1347533109513515,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001420442,
      "loss": 3.1506,
      "step": 289780
    },
    {
      "epoch": 0.1347719541875097,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001420422,
      "loss": 3.1751,
      "step": 289790
    },
    {
      "epoch": 0.1347905974236679,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001420402,
      "loss": 3.1718,
      "step": 289800
    },
    {
      "epoch": 0.13480924065982608,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001420382,
      "loss": 3.1365,
      "step": 289810
    },
    {
      "epoch": 0.13482788389598427,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001420362,
      "loss": 3.069,
      "step": 289820
    },
    {
      "epoch": 0.13484652713214243,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014203420000000002,
      "loss": 3.1764,
      "step": 289830
    },
    {
      "epoch": 0.13486517036830062,
      "grad_norm": 0.75,
      "learning_rate": 0.0001420322,
      "loss": 3.146,
      "step": 289840
    },
    {
      "epoch": 0.1348838136044588,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001420302,
      "loss": 3.2045,
      "step": 289850
    },
    {
      "epoch": 0.134902456840617,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001420282,
      "loss": 3.0935,
      "step": 289860
    },
    {
      "epoch": 0.13492110007677519,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001420262,
      "loss": 3.0377,
      "step": 289870
    },
    {
      "epoch": 0.13493974331293335,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001420242,
      "loss": 3.06,
      "step": 289880
    },
    {
      "epoch": 0.13495838654909154,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014202220000000002,
      "loss": 3.1425,
      "step": 289890
    },
    {
      "epoch": 0.13497702978524972,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001420202,
      "loss": 3.0386,
      "step": 289900
    },
    {
      "epoch": 0.1349956730214079,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001420182,
      "loss": 3.0811,
      "step": 289910
    },
    {
      "epoch": 0.1350143162575661,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001420162,
      "loss": 3.1638,
      "step": 289920
    },
    {
      "epoch": 0.1350329594937243,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001420142,
      "loss": 3.07,
      "step": 289930
    },
    {
      "epoch": 0.13505160272988245,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014201220000000002,
      "loss": 3.1896,
      "step": 289940
    },
    {
      "epoch": 0.13507024596604064,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001420102,
      "loss": 3.0798,
      "step": 289950
    },
    {
      "epoch": 0.13508888920219883,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014200820000000001,
      "loss": 3.0905,
      "step": 289960
    },
    {
      "epoch": 0.13510753243835702,
      "grad_norm": 0.875,
      "learning_rate": 0.0001420062,
      "loss": 3.1764,
      "step": 289970
    },
    {
      "epoch": 0.1351261756745152,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001420042,
      "loss": 3.3231,
      "step": 289980
    },
    {
      "epoch": 0.13514481891067337,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001420022,
      "loss": 3.0889,
      "step": 289990
    },
    {
      "epoch": 0.13516346214683156,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001420002,
      "loss": 3.0249,
      "step": 290000
    },
    {
      "epoch": 0.13518210538298975,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014199820000000002,
      "loss": 3.1529,
      "step": 290010
    },
    {
      "epoch": 0.13520074861914794,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014199620000000001,
      "loss": 3.0388,
      "step": 290020
    },
    {
      "epoch": 0.13521939185530613,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001419942,
      "loss": 3.0185,
      "step": 290030
    },
    {
      "epoch": 0.1352380350914643,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001419922,
      "loss": 2.9655,
      "step": 290040
    },
    {
      "epoch": 0.13525667832762248,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001419902,
      "loss": 3.0273,
      "step": 290050
    },
    {
      "epoch": 0.13527532156378066,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014198820000000002,
      "loss": 3.1366,
      "step": 290060
    },
    {
      "epoch": 0.13529396479993885,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001419862,
      "loss": 3.0866,
      "step": 290070
    },
    {
      "epoch": 0.13531260803609704,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014198420000000002,
      "loss": 3.1414,
      "step": 290080
    },
    {
      "epoch": 0.1353312512722552,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001419822,
      "loss": 3.0767,
      "step": 290090
    },
    {
      "epoch": 0.1353498945084134,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001419802,
      "loss": 3.2273,
      "step": 290100
    },
    {
      "epoch": 0.13536853774457158,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014197820000000003,
      "loss": 3.125,
      "step": 290110
    },
    {
      "epoch": 0.13538718098072977,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001419762,
      "loss": 3.2566,
      "step": 290120
    },
    {
      "epoch": 0.13540582421688796,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014197420000000002,
      "loss": 3.0844,
      "step": 290130
    },
    {
      "epoch": 0.13542446745304615,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001419722,
      "loss": 3.1114,
      "step": 290140
    },
    {
      "epoch": 0.1354431106892043,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001419702,
      "loss": 3.1295,
      "step": 290150
    },
    {
      "epoch": 0.1354617539253625,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001419682,
      "loss": 3.1161,
      "step": 290160
    },
    {
      "epoch": 0.1354803971615207,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001419662,
      "loss": 2.9734,
      "step": 290170
    },
    {
      "epoch": 0.13549904039767888,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014196420000000003,
      "loss": 3.2717,
      "step": 290180
    },
    {
      "epoch": 0.13551768363383707,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001419622,
      "loss": 3.1154,
      "step": 290190
    },
    {
      "epoch": 0.13553632686999523,
      "grad_norm": 0.75,
      "learning_rate": 0.00014196020000000002,
      "loss": 3.1326,
      "step": 290200
    },
    {
      "epoch": 0.13555497010615342,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014195819999999999,
      "loss": 3.2547,
      "step": 290210
    },
    {
      "epoch": 0.1355736133423116,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001419562,
      "loss": 3.1078,
      "step": 290220
    },
    {
      "epoch": 0.1355922565784698,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001419542,
      "loss": 3.1524,
      "step": 290230
    },
    {
      "epoch": 0.13561089981462798,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001419522,
      "loss": 3.204,
      "step": 290240
    },
    {
      "epoch": 0.13562954305078614,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014195020000000002,
      "loss": 3.1884,
      "step": 290250
    },
    {
      "epoch": 0.13564818628694433,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001419482,
      "loss": 3.2678,
      "step": 290260
    },
    {
      "epoch": 0.13566682952310252,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001419462,
      "loss": 3.0998,
      "step": 290270
    },
    {
      "epoch": 0.1356854727592607,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001419442,
      "loss": 3.0525,
      "step": 290280
    },
    {
      "epoch": 0.1357041159954189,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001419422,
      "loss": 3.1499,
      "step": 290290
    },
    {
      "epoch": 0.13572275923157706,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001419402,
      "loss": 3.1215,
      "step": 290300
    },
    {
      "epoch": 0.13574140246773525,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001419382,
      "loss": 3.0913,
      "step": 290310
    },
    {
      "epoch": 0.13576004570389344,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014193620000000002,
      "loss": 3.1844,
      "step": 290320
    },
    {
      "epoch": 0.13577868894005163,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014193420000000001,
      "loss": 3.0323,
      "step": 290330
    },
    {
      "epoch": 0.13579733217620982,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001419322,
      "loss": 3.0901,
      "step": 290340
    },
    {
      "epoch": 0.135815975412368,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001419302,
      "loss": 2.9949,
      "step": 290350
    },
    {
      "epoch": 0.13583461864852617,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001419282,
      "loss": 2.9944,
      "step": 290360
    },
    {
      "epoch": 0.13585326188468436,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001419262,
      "loss": 3.1698,
      "step": 290370
    },
    {
      "epoch": 0.13587190512084255,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001419242,
      "loss": 3.1807,
      "step": 290380
    },
    {
      "epoch": 0.13589054835700073,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014192220000000001,
      "loss": 3.0727,
      "step": 290390
    },
    {
      "epoch": 0.13590919159315892,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001419202,
      "loss": 3.1938,
      "step": 290400
    },
    {
      "epoch": 0.13592783482931708,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001419182,
      "loss": 3.018,
      "step": 290410
    },
    {
      "epoch": 0.13594647806547527,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001419162,
      "loss": 3.2154,
      "step": 290420
    },
    {
      "epoch": 0.13596512130163346,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001419142,
      "loss": 3.1041,
      "step": 290430
    },
    {
      "epoch": 0.13598376453779165,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001419122,
      "loss": 3.0788,
      "step": 290440
    },
    {
      "epoch": 0.13600240777394984,
      "grad_norm": 0.625,
      "learning_rate": 0.00014191020000000002,
      "loss": 3.159,
      "step": 290450
    },
    {
      "epoch": 0.136021051010108,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001419082,
      "loss": 3.0938,
      "step": 290460
    },
    {
      "epoch": 0.1360396942462662,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001419062,
      "loss": 3.1879,
      "step": 290470
    },
    {
      "epoch": 0.13605833748242438,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001419042,
      "loss": 3.2594,
      "step": 290480
    },
    {
      "epoch": 0.13607698071858257,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001419022,
      "loss": 3.1786,
      "step": 290490
    },
    {
      "epoch": 0.13609562395474076,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014190020000000002,
      "loss": 3.0373,
      "step": 290500
    },
    {
      "epoch": 0.13611426719089892,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014189820000000002,
      "loss": 3.1396,
      "step": 290510
    },
    {
      "epoch": 0.1361329104270571,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001418962,
      "loss": 3.0608,
      "step": 290520
    },
    {
      "epoch": 0.1361515536632153,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001418942,
      "loss": 3.2228,
      "step": 290530
    },
    {
      "epoch": 0.13617019689937349,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001418922,
      "loss": 3.193,
      "step": 290540
    },
    {
      "epoch": 0.13618884013553167,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014189020000000003,
      "loss": 3.163,
      "step": 290550
    },
    {
      "epoch": 0.13620748337168984,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001418882,
      "loss": 3.0968,
      "step": 290560
    },
    {
      "epoch": 0.13622612660784802,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014188620000000002,
      "loss": 3.1567,
      "step": 290570
    },
    {
      "epoch": 0.1362447698440062,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001418842,
      "loss": 3.0858,
      "step": 290580
    },
    {
      "epoch": 0.1362634130801644,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001418822,
      "loss": 3.2057,
      "step": 290590
    },
    {
      "epoch": 0.1362820563163226,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001418802,
      "loss": 3.0037,
      "step": 290600
    },
    {
      "epoch": 0.13630069955248078,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001418782,
      "loss": 3.1518,
      "step": 290610
    },
    {
      "epoch": 0.13631934278863894,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014187620000000002,
      "loss": 3.1537,
      "step": 290620
    },
    {
      "epoch": 0.13633798602479713,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001418742,
      "loss": 3.1511,
      "step": 290630
    },
    {
      "epoch": 0.13635662926095532,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014187220000000001,
      "loss": 3.1977,
      "step": 290640
    },
    {
      "epoch": 0.1363752724971135,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001418702,
      "loss": 3.1925,
      "step": 290650
    },
    {
      "epoch": 0.1363939157332717,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001418682,
      "loss": 3.1883,
      "step": 290660
    },
    {
      "epoch": 0.13641255896942986,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014186620000000003,
      "loss": 3.1612,
      "step": 290670
    },
    {
      "epoch": 0.13643120220558805,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001418642,
      "loss": 3.2266,
      "step": 290680
    },
    {
      "epoch": 0.13644984544174624,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014186220000000002,
      "loss": 3.1985,
      "step": 290690
    },
    {
      "epoch": 0.13646848867790443,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001418602,
      "loss": 3.0924,
      "step": 290700
    },
    {
      "epoch": 0.13648713191406261,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001418582,
      "loss": 3.021,
      "step": 290710
    },
    {
      "epoch": 0.13650577515022078,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001418562,
      "loss": 3.0567,
      "step": 290720
    },
    {
      "epoch": 0.13652441838637897,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001418542,
      "loss": 2.9876,
      "step": 290730
    },
    {
      "epoch": 0.13654306162253715,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014185220000000002,
      "loss": 3.1189,
      "step": 290740
    },
    {
      "epoch": 0.13656170485869534,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001418502,
      "loss": 3.2964,
      "step": 290750
    },
    {
      "epoch": 0.13658034809485353,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014184820000000002,
      "loss": 3.1634,
      "step": 290760
    },
    {
      "epoch": 0.1365989913310117,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014184619999999998,
      "loss": 3.2088,
      "step": 290770
    },
    {
      "epoch": 0.13661763456716988,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001418442,
      "loss": 3.2147,
      "step": 290780
    },
    {
      "epoch": 0.13663627780332807,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001418422,
      "loss": 3.0511,
      "step": 290790
    },
    {
      "epoch": 0.13665492103948626,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001418402,
      "loss": 3.0586,
      "step": 290800
    },
    {
      "epoch": 0.13667356427564445,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014183820000000002,
      "loss": 3.1963,
      "step": 290810
    },
    {
      "epoch": 0.13669220751180264,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001418362,
      "loss": 3.2269,
      "step": 290820
    },
    {
      "epoch": 0.1367108507479608,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001418342,
      "loss": 3.2002,
      "step": 290830
    },
    {
      "epoch": 0.136729493984119,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001418322,
      "loss": 3.1136,
      "step": 290840
    },
    {
      "epoch": 0.13674813722027718,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001418302,
      "loss": 3.1396,
      "step": 290850
    },
    {
      "epoch": 0.13676678045643537,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001418282,
      "loss": 3.1084,
      "step": 290860
    },
    {
      "epoch": 0.13678542369259356,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001418262,
      "loss": 3.2166,
      "step": 290870
    },
    {
      "epoch": 0.13680406692875172,
      "grad_norm": 0.75,
      "learning_rate": 0.00014182420000000002,
      "loss": 3.1163,
      "step": 290880
    },
    {
      "epoch": 0.1368227101649099,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001418222,
      "loss": 3.0488,
      "step": 290890
    },
    {
      "epoch": 0.1368413534010681,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001418202,
      "loss": 3.1063,
      "step": 290900
    },
    {
      "epoch": 0.13685999663722628,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001418182,
      "loss": 3.1575,
      "step": 290910
    },
    {
      "epoch": 0.13687863987338447,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001418162,
      "loss": 3.1209,
      "step": 290920
    },
    {
      "epoch": 0.13689728310954263,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001418142,
      "loss": 3.2698,
      "step": 290930
    },
    {
      "epoch": 0.13691592634570082,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014181220000000002,
      "loss": 3.0625,
      "step": 290940
    },
    {
      "epoch": 0.136934569581859,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001418102,
      "loss": 3.1935,
      "step": 290950
    },
    {
      "epoch": 0.1369532128180172,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001418082,
      "loss": 3.1332,
      "step": 290960
    },
    {
      "epoch": 0.1369718560541754,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001418062,
      "loss": 3.2629,
      "step": 290970
    },
    {
      "epoch": 0.13699049929033355,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001418042,
      "loss": 3.1148,
      "step": 290980
    },
    {
      "epoch": 0.13700914252649174,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001418022,
      "loss": 3.0969,
      "step": 290990
    },
    {
      "epoch": 0.13702778576264993,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014180020000000002,
      "loss": 3.2175,
      "step": 291000
    },
    {
      "epoch": 0.13704642899880812,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014179820000000001,
      "loss": 3.1878,
      "step": 291010
    },
    {
      "epoch": 0.1370650722349663,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001417962,
      "loss": 3.0898,
      "step": 291020
    },
    {
      "epoch": 0.1370837154711245,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001417942,
      "loss": 3.0651,
      "step": 291030
    },
    {
      "epoch": 0.13710235870728266,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001417922,
      "loss": 3.1643,
      "step": 291040
    },
    {
      "epoch": 0.13712100194344085,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001417902,
      "loss": 3.1325,
      "step": 291050
    },
    {
      "epoch": 0.13713964517959903,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014178820000000002,
      "loss": 2.9473,
      "step": 291060
    },
    {
      "epoch": 0.13715828841575722,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014178620000000001,
      "loss": 3.1401,
      "step": 291070
    },
    {
      "epoch": 0.1371769316519154,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001417842,
      "loss": 3.1734,
      "step": 291080
    },
    {
      "epoch": 0.13719557488807357,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001417822,
      "loss": 3.2531,
      "step": 291090
    },
    {
      "epoch": 0.13721421812423176,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001417802,
      "loss": 2.9779,
      "step": 291100
    },
    {
      "epoch": 0.13723286136038995,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014177820000000002,
      "loss": 3.1623,
      "step": 291110
    },
    {
      "epoch": 0.13725150459654814,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001417762,
      "loss": 3.0956,
      "step": 291120
    },
    {
      "epoch": 0.13727014783270633,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014177420000000002,
      "loss": 3.1076,
      "step": 291130
    },
    {
      "epoch": 0.1372887910688645,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001417722,
      "loss": 3.0943,
      "step": 291140
    },
    {
      "epoch": 0.13730743430502268,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001417702,
      "loss": 3.1279,
      "step": 291150
    },
    {
      "epoch": 0.13732607754118087,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014176820000000003,
      "loss": 3.0478,
      "step": 291160
    },
    {
      "epoch": 0.13734472077733906,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001417662,
      "loss": 3.1885,
      "step": 291170
    },
    {
      "epoch": 0.13736336401349725,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014176420000000002,
      "loss": 3.0852,
      "step": 291180
    },
    {
      "epoch": 0.1373820072496554,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001417622,
      "loss": 3.1564,
      "step": 291190
    },
    {
      "epoch": 0.1374006504858136,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001417602,
      "loss": 3.0248,
      "step": 291200
    },
    {
      "epoch": 0.13741929372197179,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001417582,
      "loss": 3.0657,
      "step": 291210
    },
    {
      "epoch": 0.13743793695812997,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001417562,
      "loss": 3.0567,
      "step": 291220
    },
    {
      "epoch": 0.13745658019428816,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014175420000000003,
      "loss": 3.103,
      "step": 291230
    },
    {
      "epoch": 0.13747522343044635,
      "grad_norm": 0.625,
      "learning_rate": 0.0001417522,
      "loss": 3.1023,
      "step": 291240
    },
    {
      "epoch": 0.13749386666660451,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014175020000000002,
      "loss": 3.1524,
      "step": 291250
    },
    {
      "epoch": 0.1375125099027627,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014174819999999999,
      "loss": 3.1917,
      "step": 291260
    },
    {
      "epoch": 0.1375311531389209,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001417462,
      "loss": 3.1692,
      "step": 291270
    },
    {
      "epoch": 0.13754979637507908,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001417442,
      "loss": 2.9948,
      "step": 291280
    },
    {
      "epoch": 0.13756843961123727,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001417422,
      "loss": 3.055,
      "step": 291290
    },
    {
      "epoch": 0.13758708284739543,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014174020000000002,
      "loss": 3.1669,
      "step": 291300
    },
    {
      "epoch": 0.13760572608355362,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001417382,
      "loss": 3.2054,
      "step": 291310
    },
    {
      "epoch": 0.1376243693197118,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014173620000000001,
      "loss": 3.0658,
      "step": 291320
    },
    {
      "epoch": 0.13764301255587,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001417342,
      "loss": 3.1565,
      "step": 291330
    },
    {
      "epoch": 0.1376616557920282,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001417322,
      "loss": 3.1423,
      "step": 291340
    },
    {
      "epoch": 0.13768029902818635,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001417302,
      "loss": 3.2615,
      "step": 291350
    },
    {
      "epoch": 0.13769894226434454,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001417282,
      "loss": 3.2867,
      "step": 291360
    },
    {
      "epoch": 0.13771758550050273,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014172620000000002,
      "loss": 3.0388,
      "step": 291370
    },
    {
      "epoch": 0.13773622873666092,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001417242,
      "loss": 3.1264,
      "step": 291380
    },
    {
      "epoch": 0.1377548719728191,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001417222,
      "loss": 3.2052,
      "step": 291390
    },
    {
      "epoch": 0.13777351520897727,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001417202,
      "loss": 3.2561,
      "step": 291400
    },
    {
      "epoch": 0.13779215844513545,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001417182,
      "loss": 3.2408,
      "step": 291410
    },
    {
      "epoch": 0.13781080168129364,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001417162,
      "loss": 3.2426,
      "step": 291420
    },
    {
      "epoch": 0.13782944491745183,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001417142,
      "loss": 3.2184,
      "step": 291430
    },
    {
      "epoch": 0.13784808815361002,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014171220000000002,
      "loss": 3.1373,
      "step": 291440
    },
    {
      "epoch": 0.13786673138976818,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001417102,
      "loss": 3.0388,
      "step": 291450
    },
    {
      "epoch": 0.13788537462592637,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001417082,
      "loss": 3.2208,
      "step": 291460
    },
    {
      "epoch": 0.13790401786208456,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001417062,
      "loss": 3.0119,
      "step": 291470
    },
    {
      "epoch": 0.13792266109824275,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001417042,
      "loss": 3.1856,
      "step": 291480
    },
    {
      "epoch": 0.13794130433440094,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014170220000000002,
      "loss": 3.1726,
      "step": 291490
    },
    {
      "epoch": 0.13795994757055913,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014170020000000002,
      "loss": 3.034,
      "step": 291500
    },
    {
      "epoch": 0.1379785908067173,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001416982,
      "loss": 3.0764,
      "step": 291510
    },
    {
      "epoch": 0.13799723404287548,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001416962,
      "loss": 2.8835,
      "step": 291520
    },
    {
      "epoch": 0.13801587727903367,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001416942,
      "loss": 3.1639,
      "step": 291530
    },
    {
      "epoch": 0.13803452051519186,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001416922,
      "loss": 3.1077,
      "step": 291540
    },
    {
      "epoch": 0.13805316375135004,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014169020000000002,
      "loss": 3.0961,
      "step": 291550
    },
    {
      "epoch": 0.1380718069875082,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014168820000000002,
      "loss": 3.1675,
      "step": 291560
    },
    {
      "epoch": 0.1380904502236664,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001416862,
      "loss": 3.1182,
      "step": 291570
    },
    {
      "epoch": 0.13810909345982458,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001416842,
      "loss": 3.0307,
      "step": 291580
    },
    {
      "epoch": 0.13812773669598277,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001416822,
      "loss": 3.0031,
      "step": 291590
    },
    {
      "epoch": 0.13814637993214096,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001416802,
      "loss": 3.213,
      "step": 291600
    },
    {
      "epoch": 0.13816502316829912,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001416782,
      "loss": 3.0991,
      "step": 291610
    },
    {
      "epoch": 0.1381836664044573,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014167620000000002,
      "loss": 3.1023,
      "step": 291620
    },
    {
      "epoch": 0.1382023096406155,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001416742,
      "loss": 3.1271,
      "step": 291630
    },
    {
      "epoch": 0.1382209528767737,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001416722,
      "loss": 3.1538,
      "step": 291640
    },
    {
      "epoch": 0.13823959611293188,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001416702,
      "loss": 3.1858,
      "step": 291650
    },
    {
      "epoch": 0.13825823934909004,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001416682,
      "loss": 3.077,
      "step": 291660
    },
    {
      "epoch": 0.13827688258524823,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014166620000000002,
      "loss": 3.1617,
      "step": 291670
    },
    {
      "epoch": 0.13829552582140642,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001416642,
      "loss": 3.1562,
      "step": 291680
    },
    {
      "epoch": 0.1383141690575646,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014166220000000001,
      "loss": 3.0396,
      "step": 291690
    },
    {
      "epoch": 0.1383328122937228,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001416602,
      "loss": 3.0532,
      "step": 291700
    },
    {
      "epoch": 0.13835145552988098,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001416582,
      "loss": 3.1078,
      "step": 291710
    },
    {
      "epoch": 0.13837009876603915,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014165620000000003,
      "loss": 3.1167,
      "step": 291720
    },
    {
      "epoch": 0.13838874200219733,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001416542,
      "loss": 3.2283,
      "step": 291730
    },
    {
      "epoch": 0.13840738523835552,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014165220000000002,
      "loss": 3.1284,
      "step": 291740
    },
    {
      "epoch": 0.1384260284745137,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001416502,
      "loss": 3.1676,
      "step": 291750
    },
    {
      "epoch": 0.1384446717106719,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001416482,
      "loss": 3.0724,
      "step": 291760
    },
    {
      "epoch": 0.13846331494683006,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001416462,
      "loss": 3.1581,
      "step": 291770
    },
    {
      "epoch": 0.13848195818298825,
      "grad_norm": 0.75,
      "learning_rate": 0.0001416442,
      "loss": 3.104,
      "step": 291780
    },
    {
      "epoch": 0.13850060141914644,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014164220000000002,
      "loss": 3.2222,
      "step": 291790
    },
    {
      "epoch": 0.13851924465530463,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001416402,
      "loss": 3.2031,
      "step": 291800
    },
    {
      "epoch": 0.13853788789146282,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014163820000000002,
      "loss": 3.2604,
      "step": 291810
    },
    {
      "epoch": 0.13855653112762098,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014163619999999998,
      "loss": 3.0955,
      "step": 291820
    },
    {
      "epoch": 0.13857517436377917,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001416342,
      "loss": 3.1659,
      "step": 291830
    },
    {
      "epoch": 0.13859381759993736,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001416322,
      "loss": 3.1419,
      "step": 291840
    },
    {
      "epoch": 0.13861246083609555,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001416302,
      "loss": 2.964,
      "step": 291850
    },
    {
      "epoch": 0.13863110407225374,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014162820000000002,
      "loss": 3.1836,
      "step": 291860
    },
    {
      "epoch": 0.1386497473084119,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001416262,
      "loss": 3.1896,
      "step": 291870
    },
    {
      "epoch": 0.13866839054457009,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001416242,
      "loss": 3.1513,
      "step": 291880
    },
    {
      "epoch": 0.13868703378072827,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001416222,
      "loss": 3.1769,
      "step": 291890
    },
    {
      "epoch": 0.13870567701688646,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001416202,
      "loss": 2.9919,
      "step": 291900
    },
    {
      "epoch": 0.13872432025304465,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001416182,
      "loss": 3.0224,
      "step": 291910
    },
    {
      "epoch": 0.13874296348920284,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001416162,
      "loss": 2.9952,
      "step": 291920
    },
    {
      "epoch": 0.138761606725361,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014161420000000002,
      "loss": 3.2028,
      "step": 291930
    },
    {
      "epoch": 0.1387802499615192,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001416122,
      "loss": 3.0376,
      "step": 291940
    },
    {
      "epoch": 0.13879889319767738,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001416102,
      "loss": 3.1301,
      "step": 291950
    },
    {
      "epoch": 0.13881753643383557,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001416082,
      "loss": 3.014,
      "step": 291960
    },
    {
      "epoch": 0.13883617966999376,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001416062,
      "loss": 3.1217,
      "step": 291970
    },
    {
      "epoch": 0.13885482290615192,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014160420000000002,
      "loss": 3.216,
      "step": 291980
    },
    {
      "epoch": 0.1388734661423101,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001416022,
      "loss": 3.3149,
      "step": 291990
    },
    {
      "epoch": 0.1388921093784683,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001416002,
      "loss": 3.0095,
      "step": 292000
    },
    {
      "epoch": 0.1389107526146265,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001415982,
      "loss": 3.1407,
      "step": 292010
    },
    {
      "epoch": 0.13892939585078468,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001415962,
      "loss": 3.1877,
      "step": 292020
    },
    {
      "epoch": 0.13894803908694284,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001415942,
      "loss": 3.1573,
      "step": 292030
    },
    {
      "epoch": 0.13896668232310103,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001415922,
      "loss": 3.1898,
      "step": 292040
    },
    {
      "epoch": 0.13898532555925922,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014159020000000002,
      "loss": 3.0305,
      "step": 292050
    },
    {
      "epoch": 0.1390039687954174,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014158820000000001,
      "loss": 3.1488,
      "step": 292060
    },
    {
      "epoch": 0.1390226120315756,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001415862,
      "loss": 3.1171,
      "step": 292070
    },
    {
      "epoch": 0.13904125526773375,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001415842,
      "loss": 3.1528,
      "step": 292080
    },
    {
      "epoch": 0.13905989850389194,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001415822,
      "loss": 3.129,
      "step": 292090
    },
    {
      "epoch": 0.13907854174005013,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001415802,
      "loss": 3.0928,
      "step": 292100
    },
    {
      "epoch": 0.13909718497620832,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014157820000000002,
      "loss": 3.1864,
      "step": 292110
    },
    {
      "epoch": 0.1391158282123665,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014157620000000002,
      "loss": 3.0421,
      "step": 292120
    },
    {
      "epoch": 0.13913447144852467,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001415742,
      "loss": 3.1736,
      "step": 292130
    },
    {
      "epoch": 0.13915311468468286,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001415722,
      "loss": 3.1797,
      "step": 292140
    },
    {
      "epoch": 0.13917175792084105,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001415702,
      "loss": 3.2517,
      "step": 292150
    },
    {
      "epoch": 0.13919040115699924,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014156820000000002,
      "loss": 3.2026,
      "step": 292160
    },
    {
      "epoch": 0.13920904439315743,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001415662,
      "loss": 3.052,
      "step": 292170
    },
    {
      "epoch": 0.13922768762931562,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014156420000000002,
      "loss": 3.1384,
      "step": 292180
    },
    {
      "epoch": 0.13924633086547378,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001415622,
      "loss": 3.1463,
      "step": 292190
    },
    {
      "epoch": 0.13926497410163197,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001415602,
      "loss": 3.1918,
      "step": 292200
    },
    {
      "epoch": 0.13928361733779016,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001415582,
      "loss": 3.0949,
      "step": 292210
    },
    {
      "epoch": 0.13930226057394834,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001415562,
      "loss": 3.2033,
      "step": 292220
    },
    {
      "epoch": 0.13932090381010653,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014155420000000002,
      "loss": 3.1159,
      "step": 292230
    },
    {
      "epoch": 0.1393395470462647,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001415522,
      "loss": 3.0917,
      "step": 292240
    },
    {
      "epoch": 0.13935819028242288,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001415502,
      "loss": 3.0106,
      "step": 292250
    },
    {
      "epoch": 0.13937683351858107,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001415482,
      "loss": 3.1271,
      "step": 292260
    },
    {
      "epoch": 0.13939547675473926,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001415462,
      "loss": 3.1309,
      "step": 292270
    },
    {
      "epoch": 0.13941411999089745,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014154420000000003,
      "loss": 3.1058,
      "step": 292280
    },
    {
      "epoch": 0.1394327632270556,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001415422,
      "loss": 3.1893,
      "step": 292290
    },
    {
      "epoch": 0.1394514064632138,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014154020000000002,
      "loss": 3.09,
      "step": 292300
    },
    {
      "epoch": 0.139470049699372,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014153819999999999,
      "loss": 3.1196,
      "step": 292310
    },
    {
      "epoch": 0.13948869293553018,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001415362,
      "loss": 3.2366,
      "step": 292320
    },
    {
      "epoch": 0.13950733617168837,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001415342,
      "loss": 3.0959,
      "step": 292330
    },
    {
      "epoch": 0.13952597940784653,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001415322,
      "loss": 3.2098,
      "step": 292340
    },
    {
      "epoch": 0.13954462264400472,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014153020000000002,
      "loss": 3.0687,
      "step": 292350
    },
    {
      "epoch": 0.1395632658801629,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001415282,
      "loss": 3.0797,
      "step": 292360
    },
    {
      "epoch": 0.1395819091163211,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014152620000000001,
      "loss": 3.086,
      "step": 292370
    },
    {
      "epoch": 0.13960055235247928,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001415242,
      "loss": 3.0952,
      "step": 292380
    },
    {
      "epoch": 0.13961919558863747,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001415222,
      "loss": 3.2162,
      "step": 292390
    },
    {
      "epoch": 0.13963783882479563,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001415202,
      "loss": 2.9536,
      "step": 292400
    },
    {
      "epoch": 0.13965648206095382,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001415182,
      "loss": 3.0943,
      "step": 292410
    },
    {
      "epoch": 0.139675125297112,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014151620000000002,
      "loss": 3.2381,
      "step": 292420
    },
    {
      "epoch": 0.1396937685332702,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001415142,
      "loss": 3.2073,
      "step": 292430
    },
    {
      "epoch": 0.1397124117694284,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001415122,
      "loss": 3.0321,
      "step": 292440
    },
    {
      "epoch": 0.13973105500558655,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001415102,
      "loss": 3.1875,
      "step": 292450
    },
    {
      "epoch": 0.13974969824174474,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001415082,
      "loss": 3.1546,
      "step": 292460
    },
    {
      "epoch": 0.13976834147790293,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014150620000000002,
      "loss": 3.2569,
      "step": 292470
    },
    {
      "epoch": 0.13978698471406112,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001415042,
      "loss": 3.0884,
      "step": 292480
    },
    {
      "epoch": 0.1398056279502193,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014150220000000002,
      "loss": 3.1765,
      "step": 292490
    },
    {
      "epoch": 0.13982427118637747,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001415002,
      "loss": 3.1159,
      "step": 292500
    },
    {
      "epoch": 0.13984291442253566,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001414982,
      "loss": 3.1751,
      "step": 292510
    },
    {
      "epoch": 0.13986155765869385,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001414962,
      "loss": 3.1206,
      "step": 292520
    },
    {
      "epoch": 0.13988020089485204,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001414942,
      "loss": 3.1653,
      "step": 292530
    },
    {
      "epoch": 0.13989884413101022,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014149220000000002,
      "loss": 3.107,
      "step": 292540
    },
    {
      "epoch": 0.13991748736716839,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014149020000000002,
      "loss": 2.9841,
      "step": 292550
    },
    {
      "epoch": 0.13993613060332658,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001414882,
      "loss": 2.9782,
      "step": 292560
    },
    {
      "epoch": 0.13995477383948476,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001414862,
      "loss": 3.1954,
      "step": 292570
    },
    {
      "epoch": 0.13997341707564295,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001414842,
      "loss": 3.1621,
      "step": 292580
    },
    {
      "epoch": 0.13999206031180114,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001414822,
      "loss": 3.2354,
      "step": 292590
    },
    {
      "epoch": 0.14001070354795933,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001414802,
      "loss": 3.2427,
      "step": 292600
    },
    {
      "epoch": 0.1400293467841175,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014147820000000002,
      "loss": 3.1495,
      "step": 292610
    },
    {
      "epoch": 0.14004799002027568,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001414762,
      "loss": 3.0499,
      "step": 292620
    },
    {
      "epoch": 0.14006663325643387,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001414742,
      "loss": 3.1431,
      "step": 292630
    },
    {
      "epoch": 0.14008527649259206,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001414722,
      "loss": 3.3201,
      "step": 292640
    },
    {
      "epoch": 0.14010391972875025,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001414702,
      "loss": 3.1497,
      "step": 292650
    },
    {
      "epoch": 0.1401225629649084,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001414682,
      "loss": 3.1488,
      "step": 292660
    },
    {
      "epoch": 0.1401412062010666,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014146620000000002,
      "loss": 3.1707,
      "step": 292670
    },
    {
      "epoch": 0.1401598494372248,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001414642,
      "loss": 3.1156,
      "step": 292680
    },
    {
      "epoch": 0.14017849267338298,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001414622,
      "loss": 3.185,
      "step": 292690
    },
    {
      "epoch": 0.14019713590954117,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001414602,
      "loss": 3.0368,
      "step": 292700
    },
    {
      "epoch": 0.14021577914569933,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001414582,
      "loss": 3.171,
      "step": 292710
    },
    {
      "epoch": 0.14023442238185752,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014145620000000002,
      "loss": 3.0746,
      "step": 292720
    },
    {
      "epoch": 0.1402530656180157,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001414542,
      "loss": 2.9449,
      "step": 292730
    },
    {
      "epoch": 0.1402717088541739,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014145220000000001,
      "loss": 3.2438,
      "step": 292740
    },
    {
      "epoch": 0.14029035209033208,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001414502,
      "loss": 3.1052,
      "step": 292750
    },
    {
      "epoch": 0.14030899532649024,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001414482,
      "loss": 3.1355,
      "step": 292760
    },
    {
      "epoch": 0.14032763856264843,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014144620000000003,
      "loss": 3.1663,
      "step": 292770
    },
    {
      "epoch": 0.14034628179880662,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001414442,
      "loss": 3.147,
      "step": 292780
    },
    {
      "epoch": 0.1403649250349648,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014144220000000002,
      "loss": 3.1183,
      "step": 292790
    },
    {
      "epoch": 0.140383568271123,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001414402,
      "loss": 3.2833,
      "step": 292800
    },
    {
      "epoch": 0.14040221150728116,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001414382,
      "loss": 3.2257,
      "step": 292810
    },
    {
      "epoch": 0.14042085474343935,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001414362,
      "loss": 3.1347,
      "step": 292820
    },
    {
      "epoch": 0.14043949797959754,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001414342,
      "loss": 3.0709,
      "step": 292830
    },
    {
      "epoch": 0.14045814121575573,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014143220000000002,
      "loss": 3.1405,
      "step": 292840
    },
    {
      "epoch": 0.14047678445191392,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001414302,
      "loss": 3.1263,
      "step": 292850
    },
    {
      "epoch": 0.1404954276880721,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014142820000000002,
      "loss": 3.1649,
      "step": 292860
    },
    {
      "epoch": 0.14051407092423027,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014142619999999998,
      "loss": 3.1772,
      "step": 292870
    },
    {
      "epoch": 0.14053271416038846,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001414242,
      "loss": 3.0666,
      "step": 292880
    },
    {
      "epoch": 0.14055135739654664,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001414222,
      "loss": 3.2191,
      "step": 292890
    },
    {
      "epoch": 0.14057000063270483,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001414202,
      "loss": 3.1414,
      "step": 292900
    },
    {
      "epoch": 0.14058864386886302,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014141820000000002,
      "loss": 3.1843,
      "step": 292910
    },
    {
      "epoch": 0.14060728710502118,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001414162,
      "loss": 3.1012,
      "step": 292920
    },
    {
      "epoch": 0.14062593034117937,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001414142,
      "loss": 3.1956,
      "step": 292930
    },
    {
      "epoch": 0.14064457357733756,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001414122,
      "loss": 3.0644,
      "step": 292940
    },
    {
      "epoch": 0.14066321681349575,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001414102,
      "loss": 3.0754,
      "step": 292950
    },
    {
      "epoch": 0.14068186004965394,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014140820000000003,
      "loss": 3.0566,
      "step": 292960
    },
    {
      "epoch": 0.1407005032858121,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001414062,
      "loss": 2.9997,
      "step": 292970
    },
    {
      "epoch": 0.1407191465219703,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014140420000000002,
      "loss": 3.1127,
      "step": 292980
    },
    {
      "epoch": 0.14073778975812848,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014140219999999999,
      "loss": 3.1494,
      "step": 292990
    },
    {
      "epoch": 0.14075643299428667,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001414002,
      "loss": 3.166,
      "step": 293000
    },
    {
      "epoch": 0.14077507623044486,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001413982,
      "loss": 3.1021,
      "step": 293010
    },
    {
      "epoch": 0.14079371946660302,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001413962,
      "loss": 3.2078,
      "step": 293020
    },
    {
      "epoch": 0.1408123627027612,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014139420000000002,
      "loss": 3.2251,
      "step": 293030
    },
    {
      "epoch": 0.1408310059389194,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001413922,
      "loss": 3.1589,
      "step": 293040
    },
    {
      "epoch": 0.14084964917507758,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014139020000000001,
      "loss": 3.118,
      "step": 293050
    },
    {
      "epoch": 0.14086829241123577,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001413882,
      "loss": 3.1406,
      "step": 293060
    },
    {
      "epoch": 0.14088693564739396,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001413862,
      "loss": 3.0929,
      "step": 293070
    },
    {
      "epoch": 0.14090557888355212,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001413842,
      "loss": 3.1227,
      "step": 293080
    },
    {
      "epoch": 0.1409242221197103,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001413822,
      "loss": 3.1492,
      "step": 293090
    },
    {
      "epoch": 0.1409428653558685,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014138020000000002,
      "loss": 3.1031,
      "step": 293100
    },
    {
      "epoch": 0.1409615085920267,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014137820000000001,
      "loss": 3.2291,
      "step": 293110
    },
    {
      "epoch": 0.14098015182818488,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001413762,
      "loss": 3.1584,
      "step": 293120
    },
    {
      "epoch": 0.14099879506434304,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001413742,
      "loss": 2.8838,
      "step": 293130
    },
    {
      "epoch": 0.14101743830050123,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001413722,
      "loss": 3.0917,
      "step": 293140
    },
    {
      "epoch": 0.14103608153665942,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001413702,
      "loss": 3.0979,
      "step": 293150
    },
    {
      "epoch": 0.1410547247728176,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014136820000000002,
      "loss": 3.155,
      "step": 293160
    },
    {
      "epoch": 0.1410733680089758,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014136620000000002,
      "loss": 3.1681,
      "step": 293170
    },
    {
      "epoch": 0.14109201124513396,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001413642,
      "loss": 3.1106,
      "step": 293180
    },
    {
      "epoch": 0.14111065448129215,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001413622,
      "loss": 3.1742,
      "step": 293190
    },
    {
      "epoch": 0.14112929771745034,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001413602,
      "loss": 3.0989,
      "step": 293200
    },
    {
      "epoch": 0.14114794095360852,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001413582,
      "loss": 3.1339,
      "step": 293210
    },
    {
      "epoch": 0.14116658418976671,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001413562,
      "loss": 3.1302,
      "step": 293220
    },
    {
      "epoch": 0.14118522742592488,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014135420000000002,
      "loss": 2.9715,
      "step": 293230
    },
    {
      "epoch": 0.14120387066208306,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001413522,
      "loss": 3.1835,
      "step": 293240
    },
    {
      "epoch": 0.14122251389824125,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001413502,
      "loss": 3.0615,
      "step": 293250
    },
    {
      "epoch": 0.14124115713439944,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001413482,
      "loss": 3.2276,
      "step": 293260
    },
    {
      "epoch": 0.14125980037055763,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001413462,
      "loss": 3.124,
      "step": 293270
    },
    {
      "epoch": 0.14127844360671582,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014134420000000002,
      "loss": 3.1794,
      "step": 293280
    },
    {
      "epoch": 0.14129708684287398,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001413422,
      "loss": 3.024,
      "step": 293290
    },
    {
      "epoch": 0.14131573007903217,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001413402,
      "loss": 3.2815,
      "step": 293300
    },
    {
      "epoch": 0.14133437331519036,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001413382,
      "loss": 3.0868,
      "step": 293310
    },
    {
      "epoch": 0.14135301655134855,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001413362,
      "loss": 3.2344,
      "step": 293320
    },
    {
      "epoch": 0.14137165978750674,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014133420000000003,
      "loss": 3.184,
      "step": 293330
    },
    {
      "epoch": 0.1413903030236649,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001413322,
      "loss": 3.0769,
      "step": 293340
    },
    {
      "epoch": 0.1414089462598231,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014133020000000002,
      "loss": 3.1284,
      "step": 293350
    },
    {
      "epoch": 0.14142758949598128,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014132819999999999,
      "loss": 3.075,
      "step": 293360
    },
    {
      "epoch": 0.14144623273213947,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001413262,
      "loss": 2.9371,
      "step": 293370
    },
    {
      "epoch": 0.14146487596829765,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001413242,
      "loss": 3.1388,
      "step": 293380
    },
    {
      "epoch": 0.14148351920445582,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001413222,
      "loss": 3.1903,
      "step": 293390
    },
    {
      "epoch": 0.141502162440614,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014132020000000002,
      "loss": 3.1877,
      "step": 293400
    },
    {
      "epoch": 0.1415208056767722,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001413182,
      "loss": 3.112,
      "step": 293410
    },
    {
      "epoch": 0.14153944891293038,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014131620000000001,
      "loss": 3.0581,
      "step": 293420
    },
    {
      "epoch": 0.14155809214908857,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001413142,
      "loss": 3.0454,
      "step": 293430
    },
    {
      "epoch": 0.14157673538524673,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001413122,
      "loss": 3.0974,
      "step": 293440
    },
    {
      "epoch": 0.14159537862140492,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014131020000000003,
      "loss": 3.1729,
      "step": 293450
    },
    {
      "epoch": 0.1416140218575631,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001413082,
      "loss": 3.1518,
      "step": 293460
    },
    {
      "epoch": 0.1416326650937213,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014130620000000002,
      "loss": 3.0102,
      "step": 293470
    },
    {
      "epoch": 0.1416513083298795,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001413042,
      "loss": 3.1482,
      "step": 293480
    },
    {
      "epoch": 0.14166995156603765,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001413022,
      "loss": 3.0989,
      "step": 293490
    },
    {
      "epoch": 0.14168859480219584,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001413002,
      "loss": 3.0408,
      "step": 293500
    },
    {
      "epoch": 0.14170723803835403,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001412982,
      "loss": 3.1444,
      "step": 293510
    },
    {
      "epoch": 0.14172588127451222,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014129620000000002,
      "loss": 3.1244,
      "step": 293520
    },
    {
      "epoch": 0.1417445245106704,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001412942,
      "loss": 3.1656,
      "step": 293530
    },
    {
      "epoch": 0.1417631677468286,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014129220000000002,
      "loss": 3.1183,
      "step": 293540
    },
    {
      "epoch": 0.14178181098298676,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001412902,
      "loss": 2.8492,
      "step": 293550
    },
    {
      "epoch": 0.14180045421914494,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001412882,
      "loss": 3.1231,
      "step": 293560
    },
    {
      "epoch": 0.14181909745530313,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001412862,
      "loss": 3.2341,
      "step": 293570
    },
    {
      "epoch": 0.14183774069146132,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001412842,
      "loss": 3.0967,
      "step": 293580
    },
    {
      "epoch": 0.1418563839276195,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014128220000000002,
      "loss": 3.1845,
      "step": 293590
    },
    {
      "epoch": 0.14187502716377767,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001412802,
      "loss": 3.0643,
      "step": 293600
    },
    {
      "epoch": 0.14189367039993586,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001412782,
      "loss": 3.0813,
      "step": 293610
    },
    {
      "epoch": 0.14191231363609405,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001412762,
      "loss": 3.0733,
      "step": 293620
    },
    {
      "epoch": 0.14193095687225224,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001412742,
      "loss": 3.2725,
      "step": 293630
    },
    {
      "epoch": 0.14194960010841043,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001412722,
      "loss": 3.1996,
      "step": 293640
    },
    {
      "epoch": 0.1419682433445686,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001412702,
      "loss": 3.0505,
      "step": 293650
    },
    {
      "epoch": 0.14198688658072678,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014126820000000002,
      "loss": 3.1523,
      "step": 293660
    },
    {
      "epoch": 0.14200552981688497,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001412662,
      "loss": 3.0859,
      "step": 293670
    },
    {
      "epoch": 0.14202417305304316,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001412642,
      "loss": 3.2778,
      "step": 293680
    },
    {
      "epoch": 0.14204281628920135,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001412622,
      "loss": 3.0092,
      "step": 293690
    },
    {
      "epoch": 0.1420614595253595,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001412602,
      "loss": 3.1623,
      "step": 293700
    },
    {
      "epoch": 0.1420801027615177,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001412582,
      "loss": 3.1259,
      "step": 293710
    },
    {
      "epoch": 0.14209874599767588,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014125620000000002,
      "loss": 2.9984,
      "step": 293720
    },
    {
      "epoch": 0.14211738923383407,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014125420000000001,
      "loss": 3.1497,
      "step": 293730
    },
    {
      "epoch": 0.14213603246999226,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001412522,
      "loss": 3.0852,
      "step": 293740
    },
    {
      "epoch": 0.14215467570615045,
      "grad_norm": 0.75,
      "learning_rate": 0.0001412502,
      "loss": 3.0391,
      "step": 293750
    },
    {
      "epoch": 0.1421733189423086,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001412482,
      "loss": 3.1873,
      "step": 293760
    },
    {
      "epoch": 0.1421919621784668,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014124620000000002,
      "loss": 3.0937,
      "step": 293770
    },
    {
      "epoch": 0.142210605414625,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001412442,
      "loss": 3.2327,
      "step": 293780
    },
    {
      "epoch": 0.14222924865078318,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014124220000000001,
      "loss": 3.1787,
      "step": 293790
    },
    {
      "epoch": 0.14224789188694137,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001412402,
      "loss": 3.1144,
      "step": 293800
    },
    {
      "epoch": 0.14226653512309953,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001412382,
      "loss": 2.9863,
      "step": 293810
    },
    {
      "epoch": 0.14228517835925772,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001412362,
      "loss": 3.1997,
      "step": 293820
    },
    {
      "epoch": 0.1423038215954159,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001412342,
      "loss": 3.0489,
      "step": 293830
    },
    {
      "epoch": 0.1423224648315741,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014123220000000002,
      "loss": 3.0924,
      "step": 293840
    },
    {
      "epoch": 0.14234110806773229,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001412302,
      "loss": 3.0571,
      "step": 293850
    },
    {
      "epoch": 0.14235975130389045,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001412282,
      "loss": 3.016,
      "step": 293860
    },
    {
      "epoch": 0.14237839454004864,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001412262,
      "loss": 3.2087,
      "step": 293870
    },
    {
      "epoch": 0.14239703777620683,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001412242,
      "loss": 3.2127,
      "step": 293880
    },
    {
      "epoch": 0.14241568101236501,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014122220000000002,
      "loss": 3.1646,
      "step": 293890
    },
    {
      "epoch": 0.1424343242485232,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001412202,
      "loss": 3.1735,
      "step": 293900
    },
    {
      "epoch": 0.14245296748468136,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014121820000000002,
      "loss": 3.0812,
      "step": 293910
    },
    {
      "epoch": 0.14247161072083955,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001412162,
      "loss": 3.1422,
      "step": 293920
    },
    {
      "epoch": 0.14249025395699774,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001412142,
      "loss": 3.1298,
      "step": 293930
    },
    {
      "epoch": 0.14250889719315593,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014121220000000003,
      "loss": 3.121,
      "step": 293940
    },
    {
      "epoch": 0.14252754042931412,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001412102,
      "loss": 3.0563,
      "step": 293950
    },
    {
      "epoch": 0.1425461836654723,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014120820000000002,
      "loss": 2.9062,
      "step": 293960
    },
    {
      "epoch": 0.14256482690163047,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001412062,
      "loss": 3.1506,
      "step": 293970
    },
    {
      "epoch": 0.14258347013778866,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001412042,
      "loss": 3.2325,
      "step": 293980
    },
    {
      "epoch": 0.14260211337394685,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001412022,
      "loss": 3.0658,
      "step": 293990
    },
    {
      "epoch": 0.14262075661010504,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001412002,
      "loss": 3.0885,
      "step": 294000
    },
    {
      "epoch": 0.14263939984626323,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014119820000000003,
      "loss": 3.1674,
      "step": 294010
    },
    {
      "epoch": 0.1426580430824214,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001411962,
      "loss": 3.21,
      "step": 294020
    },
    {
      "epoch": 0.14267668631857958,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014119420000000002,
      "loss": 3.1126,
      "step": 294030
    },
    {
      "epoch": 0.14269532955473777,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014119219999999999,
      "loss": 3.1904,
      "step": 294040
    },
    {
      "epoch": 0.14271397279089595,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001411902,
      "loss": 3.0753,
      "step": 294050
    },
    {
      "epoch": 0.14273261602705414,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001411882,
      "loss": 3.1621,
      "step": 294060
    },
    {
      "epoch": 0.1427512592632123,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001411862,
      "loss": 3.0454,
      "step": 294070
    },
    {
      "epoch": 0.1427699024993705,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014118420000000002,
      "loss": 3.2002,
      "step": 294080
    },
    {
      "epoch": 0.14278854573552868,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001411822,
      "loss": 3.0871,
      "step": 294090
    },
    {
      "epoch": 0.14280718897168687,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014118020000000001,
      "loss": 3.0037,
      "step": 294100
    },
    {
      "epoch": 0.14282583220784506,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001411782,
      "loss": 3.1444,
      "step": 294110
    },
    {
      "epoch": 0.14284447544400322,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001411762,
      "loss": 3.0935,
      "step": 294120
    },
    {
      "epoch": 0.1428631186801614,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001411742,
      "loss": 3.149,
      "step": 294130
    },
    {
      "epoch": 0.1428817619163196,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001411722,
      "loss": 3.2083,
      "step": 294140
    },
    {
      "epoch": 0.1429004051524778,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014117020000000002,
      "loss": 3.0932,
      "step": 294150
    },
    {
      "epoch": 0.14291904838863598,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014116820000000001,
      "loss": 3.1419,
      "step": 294160
    },
    {
      "epoch": 0.14293769162479414,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001411662,
      "loss": 3.1468,
      "step": 294170
    },
    {
      "epoch": 0.14295633486095233,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001411642,
      "loss": 3.2368,
      "step": 294180
    },
    {
      "epoch": 0.14297497809711052,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001411622,
      "loss": 3.077,
      "step": 294190
    },
    {
      "epoch": 0.1429936213332687,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001411602,
      "loss": 3.1262,
      "step": 294200
    },
    {
      "epoch": 0.1430122645694269,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001411582,
      "loss": 3.0167,
      "step": 294210
    },
    {
      "epoch": 0.14303090780558508,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014115620000000002,
      "loss": 3.0397,
      "step": 294220
    },
    {
      "epoch": 0.14304955104174324,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001411542,
      "loss": 3.0892,
      "step": 294230
    },
    {
      "epoch": 0.14306819427790143,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001411522,
      "loss": 3.085,
      "step": 294240
    },
    {
      "epoch": 0.14308683751405962,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001411502,
      "loss": 3.07,
      "step": 294250
    },
    {
      "epoch": 0.1431054807502178,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001411482,
      "loss": 3.0201,
      "step": 294260
    },
    {
      "epoch": 0.143124123986376,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001411462,
      "loss": 3.1456,
      "step": 294270
    },
    {
      "epoch": 0.14314276722253416,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014114420000000002,
      "loss": 3.1887,
      "step": 294280
    },
    {
      "epoch": 0.14316141045869235,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001411422,
      "loss": 3.1783,
      "step": 294290
    },
    {
      "epoch": 0.14318005369485054,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001411402,
      "loss": 3.0703,
      "step": 294300
    },
    {
      "epoch": 0.14319869693100873,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001411382,
      "loss": 3.1101,
      "step": 294310
    },
    {
      "epoch": 0.14321734016716692,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001411362,
      "loss": 3.0451,
      "step": 294320
    },
    {
      "epoch": 0.14323598340332508,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014113420000000002,
      "loss": 3.1848,
      "step": 294330
    },
    {
      "epoch": 0.14325462663948327,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001411322,
      "loss": 3.1068,
      "step": 294340
    },
    {
      "epoch": 0.14327326987564146,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001411302,
      "loss": 3.1077,
      "step": 294350
    },
    {
      "epoch": 0.14329191311179965,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001411282,
      "loss": 3.0531,
      "step": 294360
    },
    {
      "epoch": 0.14331055634795783,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001411262,
      "loss": 3.0771,
      "step": 294370
    },
    {
      "epoch": 0.143329199584116,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014112420000000003,
      "loss": 3.0405,
      "step": 294380
    },
    {
      "epoch": 0.14334784282027419,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001411222,
      "loss": 3.0732,
      "step": 294390
    },
    {
      "epoch": 0.14336648605643237,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014112020000000002,
      "loss": 3.0904,
      "step": 294400
    },
    {
      "epoch": 0.14338512929259056,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014111820000000001,
      "loss": 3.1599,
      "step": 294410
    },
    {
      "epoch": 0.14340377252874875,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001411162,
      "loss": 3.2143,
      "step": 294420
    },
    {
      "epoch": 0.14342241576490694,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001411142,
      "loss": 3.1105,
      "step": 294430
    },
    {
      "epoch": 0.1434410590010651,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001411122,
      "loss": 3.2451,
      "step": 294440
    },
    {
      "epoch": 0.1434597022372233,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014111020000000002,
      "loss": 3.1288,
      "step": 294450
    },
    {
      "epoch": 0.14347834547338148,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001411082,
      "loss": 3.1301,
      "step": 294460
    },
    {
      "epoch": 0.14349698870953967,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014110620000000001,
      "loss": 2.9799,
      "step": 294470
    },
    {
      "epoch": 0.14351563194569786,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001411042,
      "loss": 3.1969,
      "step": 294480
    },
    {
      "epoch": 0.14353427518185602,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001411022,
      "loss": 3.0607,
      "step": 294490
    },
    {
      "epoch": 0.1435529184180142,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014110020000000003,
      "loss": 3.1675,
      "step": 294500
    },
    {
      "epoch": 0.1435715616541724,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001410982,
      "loss": 3.1419,
      "step": 294510
    },
    {
      "epoch": 0.14359020489033059,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014109620000000002,
      "loss": 3.1242,
      "step": 294520
    },
    {
      "epoch": 0.14360884812648878,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001410942,
      "loss": 3.0878,
      "step": 294530
    },
    {
      "epoch": 0.14362749136264694,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001410922,
      "loss": 3.0534,
      "step": 294540
    },
    {
      "epoch": 0.14364613459880513,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001410902,
      "loss": 3.0718,
      "step": 294550
    },
    {
      "epoch": 0.14366477783496331,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001410882,
      "loss": 3.1803,
      "step": 294560
    },
    {
      "epoch": 0.1436834210711215,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014108620000000002,
      "loss": 3.0721,
      "step": 294570
    },
    {
      "epoch": 0.1437020643072797,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001410842,
      "loss": 3.0374,
      "step": 294580
    },
    {
      "epoch": 0.14372070754343785,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014108220000000002,
      "loss": 3.2148,
      "step": 294590
    },
    {
      "epoch": 0.14373935077959604,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014108019999999998,
      "loss": 3.1198,
      "step": 294600
    },
    {
      "epoch": 0.14375799401575423,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001410782,
      "loss": 3.1563,
      "step": 294610
    },
    {
      "epoch": 0.14377663725191242,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001410762,
      "loss": 3.1185,
      "step": 294620
    },
    {
      "epoch": 0.1437952804880706,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001410742,
      "loss": 3.0855,
      "step": 294630
    },
    {
      "epoch": 0.1438139237242288,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014107220000000002,
      "loss": 3.1552,
      "step": 294640
    },
    {
      "epoch": 0.14383256696038696,
      "grad_norm": 0.75,
      "learning_rate": 0.0001410702,
      "loss": 3.04,
      "step": 294650
    },
    {
      "epoch": 0.14385121019654515,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001410682,
      "loss": 3.1074,
      "step": 294660
    },
    {
      "epoch": 0.14386985343270334,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001410662,
      "loss": 3.1406,
      "step": 294670
    },
    {
      "epoch": 0.14388849666886153,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001410642,
      "loss": 3.2381,
      "step": 294680
    },
    {
      "epoch": 0.14390713990501972,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001410622,
      "loss": 3.0588,
      "step": 294690
    },
    {
      "epoch": 0.14392578314117788,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001410602,
      "loss": 3.2366,
      "step": 294700
    },
    {
      "epoch": 0.14394442637733607,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014105820000000002,
      "loss": 3.1252,
      "step": 294710
    },
    {
      "epoch": 0.14396306961349425,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001410562,
      "loss": 3.0734,
      "step": 294720
    },
    {
      "epoch": 0.14398171284965244,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001410542,
      "loss": 3.0318,
      "step": 294730
    },
    {
      "epoch": 0.14400035608581063,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001410522,
      "loss": 3.182,
      "step": 294740
    },
    {
      "epoch": 0.1440189993219688,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001410502,
      "loss": 3.1309,
      "step": 294750
    },
    {
      "epoch": 0.14403764255812698,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001410482,
      "loss": 2.9579,
      "step": 294760
    },
    {
      "epoch": 0.14405628579428517,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014104620000000002,
      "loss": 3.0096,
      "step": 294770
    },
    {
      "epoch": 0.14407492903044336,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014104420000000001,
      "loss": 3.126,
      "step": 294780
    },
    {
      "epoch": 0.14409357226660155,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001410422,
      "loss": 3.2336,
      "step": 294790
    },
    {
      "epoch": 0.1441122155027597,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001410402,
      "loss": 3.1452,
      "step": 294800
    },
    {
      "epoch": 0.1441308587389179,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001410382,
      "loss": 3.0977,
      "step": 294810
    },
    {
      "epoch": 0.1441495019750761,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001410362,
      "loss": 3.1577,
      "step": 294820
    },
    {
      "epoch": 0.14416814521123428,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001410342,
      "loss": 3.0528,
      "step": 294830
    },
    {
      "epoch": 0.14418678844739247,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014103220000000001,
      "loss": 3.3139,
      "step": 294840
    },
    {
      "epoch": 0.14420543168355063,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001410302,
      "loss": 3.0922,
      "step": 294850
    },
    {
      "epoch": 0.14422407491970882,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001410282,
      "loss": 3.1524,
      "step": 294860
    },
    {
      "epoch": 0.144242718155867,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001410262,
      "loss": 3.1781,
      "step": 294870
    },
    {
      "epoch": 0.1442613613920252,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001410242,
      "loss": 2.9849,
      "step": 294880
    },
    {
      "epoch": 0.14428000462818338,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014102220000000002,
      "loss": 3.2427,
      "step": 294890
    },
    {
      "epoch": 0.14429864786434157,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014102020000000002,
      "loss": 3.1558,
      "step": 294900
    },
    {
      "epoch": 0.14431729110049973,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001410182,
      "loss": 3.0089,
      "step": 294910
    },
    {
      "epoch": 0.14433593433665792,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001410162,
      "loss": 3.2322,
      "step": 294920
    },
    {
      "epoch": 0.1443545775728161,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001410142,
      "loss": 3.0609,
      "step": 294930
    },
    {
      "epoch": 0.1443732208089743,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014101220000000003,
      "loss": 3.2441,
      "step": 294940
    },
    {
      "epoch": 0.1443918640451325,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001410102,
      "loss": 3.0968,
      "step": 294950
    },
    {
      "epoch": 0.14441050728129065,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014100820000000002,
      "loss": 3.0253,
      "step": 294960
    },
    {
      "epoch": 0.14442915051744884,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001410062,
      "loss": 2.9988,
      "step": 294970
    },
    {
      "epoch": 0.14444779375360703,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001410042,
      "loss": 3.1254,
      "step": 294980
    },
    {
      "epoch": 0.14446643698976522,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014100220000000003,
      "loss": 3.0391,
      "step": 294990
    },
    {
      "epoch": 0.1444850802259234,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001410002,
      "loss": 3.0362,
      "step": 295000
    },
    {
      "epoch": 0.14450372346208157,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014099820000000002,
      "loss": 3.2264,
      "step": 295010
    },
    {
      "epoch": 0.14452236669823976,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001409962,
      "loss": 3.1361,
      "step": 295020
    },
    {
      "epoch": 0.14454100993439795,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001409942,
      "loss": 3.0465,
      "step": 295030
    },
    {
      "epoch": 0.14455965317055613,
      "grad_norm": 0.75,
      "learning_rate": 0.0001409922,
      "loss": 3.1844,
      "step": 295040
    },
    {
      "epoch": 0.14457829640671432,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001409902,
      "loss": 3.1772,
      "step": 295050
    },
    {
      "epoch": 0.14459693964287249,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014098820000000003,
      "loss": 3.0238,
      "step": 295060
    },
    {
      "epoch": 0.14461558287903067,
      "grad_norm": 0.75,
      "learning_rate": 0.0001409862,
      "loss": 3.0374,
      "step": 295070
    },
    {
      "epoch": 0.14463422611518886,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014098420000000002,
      "loss": 3.207,
      "step": 295080
    },
    {
      "epoch": 0.14465286935134705,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001409822,
      "loss": 3.2069,
      "step": 295090
    },
    {
      "epoch": 0.14467151258750524,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001409802,
      "loss": 3.1519,
      "step": 295100
    },
    {
      "epoch": 0.14469015582366343,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001409782,
      "loss": 3.1458,
      "step": 295110
    },
    {
      "epoch": 0.1447087990598216,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001409762,
      "loss": 3.1163,
      "step": 295120
    },
    {
      "epoch": 0.14472744229597978,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014097420000000002,
      "loss": 3.1205,
      "step": 295130
    },
    {
      "epoch": 0.14474608553213797,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001409722,
      "loss": 3.1419,
      "step": 295140
    },
    {
      "epoch": 0.14476472876829616,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014097020000000001,
      "loss": 3.008,
      "step": 295150
    },
    {
      "epoch": 0.14478337200445435,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001409682,
      "loss": 3.1813,
      "step": 295160
    },
    {
      "epoch": 0.1448020152406125,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001409662,
      "loss": 3.0872,
      "step": 295170
    },
    {
      "epoch": 0.1448206584767707,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001409642,
      "loss": 3.2083,
      "step": 295180
    },
    {
      "epoch": 0.1448393017129289,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001409622,
      "loss": 3.0119,
      "step": 295190
    },
    {
      "epoch": 0.14485794494908708,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014096020000000002,
      "loss": 3.177,
      "step": 295200
    },
    {
      "epoch": 0.14487658818524526,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001409582,
      "loss": 3.181,
      "step": 295210
    },
    {
      "epoch": 0.14489523142140343,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001409562,
      "loss": 2.9118,
      "step": 295220
    },
    {
      "epoch": 0.14491387465756161,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001409542,
      "loss": 3.1954,
      "step": 295230
    },
    {
      "epoch": 0.1449325178937198,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001409522,
      "loss": 3.1618,
      "step": 295240
    },
    {
      "epoch": 0.144951161129878,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001409502,
      "loss": 3.0148,
      "step": 295250
    },
    {
      "epoch": 0.14496980436603618,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001409482,
      "loss": 3.153,
      "step": 295260
    },
    {
      "epoch": 0.14498844760219434,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014094620000000002,
      "loss": 3.056,
      "step": 295270
    },
    {
      "epoch": 0.14500709083835253,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001409442,
      "loss": 3.1277,
      "step": 295280
    },
    {
      "epoch": 0.14502573407451072,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001409422,
      "loss": 3.1876,
      "step": 295290
    },
    {
      "epoch": 0.1450443773106689,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001409402,
      "loss": 3.1469,
      "step": 295300
    },
    {
      "epoch": 0.1450630205468271,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001409382,
      "loss": 3.088,
      "step": 295310
    },
    {
      "epoch": 0.1450816637829853,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001409362,
      "loss": 3.1249,
      "step": 295320
    },
    {
      "epoch": 0.14510030701914345,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014093420000000002,
      "loss": 3.033,
      "step": 295330
    },
    {
      "epoch": 0.14511895025530164,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001409322,
      "loss": 3.0209,
      "step": 295340
    },
    {
      "epoch": 0.14513759349145983,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001409302,
      "loss": 3.0131,
      "step": 295350
    },
    {
      "epoch": 0.14515623672761802,
      "grad_norm": 0.75,
      "learning_rate": 0.0001409282,
      "loss": 3.0841,
      "step": 295360
    },
    {
      "epoch": 0.1451748799637762,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001409262,
      "loss": 2.9712,
      "step": 295370
    },
    {
      "epoch": 0.14519352319993437,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014092420000000002,
      "loss": 3.1621,
      "step": 295380
    },
    {
      "epoch": 0.14521216643609255,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014092220000000002,
      "loss": 3.0992,
      "step": 295390
    },
    {
      "epoch": 0.14523080967225074,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001409202,
      "loss": 3.1166,
      "step": 295400
    },
    {
      "epoch": 0.14524945290840893,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001409182,
      "loss": 3.316,
      "step": 295410
    },
    {
      "epoch": 0.14526809614456712,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001409162,
      "loss": 3.0241,
      "step": 295420
    },
    {
      "epoch": 0.14528673938072528,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001409142,
      "loss": 3.2039,
      "step": 295430
    },
    {
      "epoch": 0.14530538261688347,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001409122,
      "loss": 3.0048,
      "step": 295440
    },
    {
      "epoch": 0.14532402585304166,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014091020000000002,
      "loss": 3.1809,
      "step": 295450
    },
    {
      "epoch": 0.14534266908919985,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014090820000000001,
      "loss": 3.1321,
      "step": 295460
    },
    {
      "epoch": 0.14536131232535804,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001409062,
      "loss": 2.988,
      "step": 295470
    },
    {
      "epoch": 0.1453799555615162,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001409042,
      "loss": 3.1377,
      "step": 295480
    },
    {
      "epoch": 0.1453985987976744,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001409022,
      "loss": 3.1399,
      "step": 295490
    },
    {
      "epoch": 0.14541724203383258,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014090020000000002,
      "loss": 3.1808,
      "step": 295500
    },
    {
      "epoch": 0.14543588526999077,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001408982,
      "loss": 3.1648,
      "step": 295510
    },
    {
      "epoch": 0.14545452850614896,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014089620000000001,
      "loss": 3.1166,
      "step": 295520
    },
    {
      "epoch": 0.14547317174230712,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001408942,
      "loss": 3.1052,
      "step": 295530
    },
    {
      "epoch": 0.1454918149784653,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001408922,
      "loss": 3.1057,
      "step": 295540
    },
    {
      "epoch": 0.1455104582146235,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014089020000000003,
      "loss": 3.116,
      "step": 295550
    },
    {
      "epoch": 0.14552910145078168,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001408882,
      "loss": 3.1985,
      "step": 295560
    },
    {
      "epoch": 0.14554774468693987,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014088620000000002,
      "loss": 3.1949,
      "step": 295570
    },
    {
      "epoch": 0.14556638792309806,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001408842,
      "loss": 3.0814,
      "step": 295580
    },
    {
      "epoch": 0.14558503115925622,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001408822,
      "loss": 3.1791,
      "step": 295590
    },
    {
      "epoch": 0.1456036743954144,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001408802,
      "loss": 3.1592,
      "step": 295600
    },
    {
      "epoch": 0.1456223176315726,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001408782,
      "loss": 3.0424,
      "step": 295610
    },
    {
      "epoch": 0.1456409608677308,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014087620000000003,
      "loss": 3.1474,
      "step": 295620
    },
    {
      "epoch": 0.14565960410388898,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001408742,
      "loss": 3.0947,
      "step": 295630
    },
    {
      "epoch": 0.14567824734004714,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014087220000000002,
      "loss": 3.1037,
      "step": 295640
    },
    {
      "epoch": 0.14569689057620533,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014087019999999998,
      "loss": 3.1344,
      "step": 295650
    },
    {
      "epoch": 0.14571553381236352,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001408682,
      "loss": 3.1753,
      "step": 295660
    },
    {
      "epoch": 0.1457341770485217,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001408662,
      "loss": 3.2424,
      "step": 295670
    },
    {
      "epoch": 0.1457528202846799,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001408642,
      "loss": 3.2521,
      "step": 295680
    },
    {
      "epoch": 0.14577146352083806,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014086220000000002,
      "loss": 3.075,
      "step": 295690
    },
    {
      "epoch": 0.14579010675699625,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001408602,
      "loss": 3.2377,
      "step": 295700
    },
    {
      "epoch": 0.14580874999315444,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001408582,
      "loss": 3.0114,
      "step": 295710
    },
    {
      "epoch": 0.14582739322931262,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001408562,
      "loss": 3.0336,
      "step": 295720
    },
    {
      "epoch": 0.1458460364654708,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001408542,
      "loss": 3.092,
      "step": 295730
    },
    {
      "epoch": 0.14586467970162897,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001408522,
      "loss": 3.0369,
      "step": 295740
    },
    {
      "epoch": 0.14588332293778716,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001408502,
      "loss": 3.2127,
      "step": 295750
    },
    {
      "epoch": 0.14590196617394535,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014084820000000002,
      "loss": 3.166,
      "step": 295760
    },
    {
      "epoch": 0.14592060941010354,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014084620000000001,
      "loss": 3.103,
      "step": 295770
    },
    {
      "epoch": 0.14593925264626173,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001408442,
      "loss": 3.2296,
      "step": 295780
    },
    {
      "epoch": 0.14595789588241992,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001408422,
      "loss": 3.2018,
      "step": 295790
    },
    {
      "epoch": 0.14597653911857808,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001408402,
      "loss": 3.1865,
      "step": 295800
    },
    {
      "epoch": 0.14599518235473627,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001408382,
      "loss": 3.327,
      "step": 295810
    },
    {
      "epoch": 0.14601382559089446,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001408362,
      "loss": 3.2107,
      "step": 295820
    },
    {
      "epoch": 0.14603246882705265,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014083420000000001,
      "loss": 3.2588,
      "step": 295830
    },
    {
      "epoch": 0.14605111206321084,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001408322,
      "loss": 3.1048,
      "step": 295840
    },
    {
      "epoch": 0.146069755299369,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001408302,
      "loss": 3.1606,
      "step": 295850
    },
    {
      "epoch": 0.1460883985355272,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001408282,
      "loss": 3.1541,
      "step": 295860
    },
    {
      "epoch": 0.14610704177168538,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001408262,
      "loss": 3.1699,
      "step": 295870
    },
    {
      "epoch": 0.14612568500784356,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014082420000000002,
      "loss": 3.1076,
      "step": 295880
    },
    {
      "epoch": 0.14614432824400175,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014082220000000002,
      "loss": 3.0768,
      "step": 295890
    },
    {
      "epoch": 0.14616297148015991,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001408202,
      "loss": 3.1913,
      "step": 295900
    },
    {
      "epoch": 0.1461816147163181,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001408182,
      "loss": 3.0815,
      "step": 295910
    },
    {
      "epoch": 0.1462002579524763,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001408162,
      "loss": 3.2708,
      "step": 295920
    },
    {
      "epoch": 0.14621890118863448,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001408142,
      "loss": 3.1855,
      "step": 295930
    },
    {
      "epoch": 0.14623754442479267,
      "grad_norm": 0.75,
      "learning_rate": 0.00014081220000000002,
      "loss": 3.0971,
      "step": 295940
    },
    {
      "epoch": 0.14625618766095083,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014081020000000002,
      "loss": 3.2006,
      "step": 295950
    },
    {
      "epoch": 0.14627483089710902,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001408082,
      "loss": 3.2258,
      "step": 295960
    },
    {
      "epoch": 0.1462934741332672,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001408062,
      "loss": 3.0211,
      "step": 295970
    },
    {
      "epoch": 0.1463121173694254,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001408042,
      "loss": 3.1096,
      "step": 295980
    },
    {
      "epoch": 0.1463307606055836,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014080220000000003,
      "loss": 3.0602,
      "step": 295990
    },
    {
      "epoch": 0.14634940384174178,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001408002,
      "loss": 3.1216,
      "step": 296000
    },
    {
      "epoch": 0.14636804707789994,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014079820000000002,
      "loss": 3.1618,
      "step": 296010
    },
    {
      "epoch": 0.14638669031405813,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001407962,
      "loss": 3.1145,
      "step": 296020
    },
    {
      "epoch": 0.14640533355021632,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001407942,
      "loss": 3.1166,
      "step": 296030
    },
    {
      "epoch": 0.1464239767863745,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001407922,
      "loss": 3.1698,
      "step": 296040
    },
    {
      "epoch": 0.1464426200225327,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001407902,
      "loss": 3.375,
      "step": 296050
    },
    {
      "epoch": 0.14646126325869085,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014078820000000002,
      "loss": 3.0267,
      "step": 296060
    },
    {
      "epoch": 0.14647990649484904,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001407862,
      "loss": 3.2566,
      "step": 296070
    },
    {
      "epoch": 0.14649854973100723,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001407842,
      "loss": 3.1579,
      "step": 296080
    },
    {
      "epoch": 0.14651719296716542,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001407822,
      "loss": 3.0614,
      "step": 296090
    },
    {
      "epoch": 0.1465358362033236,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001407802,
      "loss": 3.1443,
      "step": 296100
    },
    {
      "epoch": 0.14655447943948177,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014077820000000003,
      "loss": 3.1336,
      "step": 296110
    },
    {
      "epoch": 0.14657312267563996,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001407762,
      "loss": 3.1274,
      "step": 296120
    },
    {
      "epoch": 0.14659176591179815,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014077420000000002,
      "loss": 3.014,
      "step": 296130
    },
    {
      "epoch": 0.14661040914795634,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001407722,
      "loss": 3.1737,
      "step": 296140
    },
    {
      "epoch": 0.14662905238411453,
      "grad_norm": 0.75,
      "learning_rate": 0.0001407702,
      "loss": 3.1336,
      "step": 296150
    },
    {
      "epoch": 0.1466476956202727,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001407682,
      "loss": 3.196,
      "step": 296160
    },
    {
      "epoch": 0.14666633885643088,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001407662,
      "loss": 3.2137,
      "step": 296170
    },
    {
      "epoch": 0.14668498209258907,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014076420000000002,
      "loss": 3.0679,
      "step": 296180
    },
    {
      "epoch": 0.14670362532874726,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001407622,
      "loss": 3.0384,
      "step": 296190
    },
    {
      "epoch": 0.14672226856490544,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00014076020000000001,
      "loss": 3.0717,
      "step": 296200
    },
    {
      "epoch": 0.14674091180106363,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001407582,
      "loss": 3.1004,
      "step": 296210
    },
    {
      "epoch": 0.1467595550372218,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001407562,
      "loss": 3.0996,
      "step": 296220
    },
    {
      "epoch": 0.14677819827337998,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001407542,
      "loss": 3.1073,
      "step": 296230
    },
    {
      "epoch": 0.14679684150953817,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001407522,
      "loss": 3.1557,
      "step": 296240
    },
    {
      "epoch": 0.14681548474569636,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014075020000000002,
      "loss": 3.2225,
      "step": 296250
    },
    {
      "epoch": 0.14683412798185455,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001407482,
      "loss": 3.1163,
      "step": 296260
    },
    {
      "epoch": 0.1468527712180127,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001407462,
      "loss": 3.1928,
      "step": 296270
    },
    {
      "epoch": 0.1468714144541709,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001407442,
      "loss": 3.1499,
      "step": 296280
    },
    {
      "epoch": 0.1468900576903291,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001407422,
      "loss": 3.062,
      "step": 296290
    },
    {
      "epoch": 0.14690870092648728,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001407402,
      "loss": 3.1464,
      "step": 296300
    },
    {
      "epoch": 0.14692734416264547,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001407382,
      "loss": 3.0855,
      "step": 296310
    },
    {
      "epoch": 0.14694598739880363,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014073620000000002,
      "loss": 3.0761,
      "step": 296320
    },
    {
      "epoch": 0.14696463063496182,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001407342,
      "loss": 3.1154,
      "step": 296330
    },
    {
      "epoch": 0.14698327387112,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001407322,
      "loss": 3.2562,
      "step": 296340
    },
    {
      "epoch": 0.1470019171072782,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001407302,
      "loss": 3.0423,
      "step": 296350
    },
    {
      "epoch": 0.14702056034343639,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001407282,
      "loss": 3.2587,
      "step": 296360
    },
    {
      "epoch": 0.14703920357959455,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014072620000000002,
      "loss": 3.0676,
      "step": 296370
    },
    {
      "epoch": 0.14705784681575274,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014072420000000002,
      "loss": 3.098,
      "step": 296380
    },
    {
      "epoch": 0.14707649005191092,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001407222,
      "loss": 3.0096,
      "step": 296390
    },
    {
      "epoch": 0.1470951332880691,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001407202,
      "loss": 3.0461,
      "step": 296400
    },
    {
      "epoch": 0.1471137765242273,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001407182,
      "loss": 3.0771,
      "step": 296410
    },
    {
      "epoch": 0.14713241976038546,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001407162,
      "loss": 3.0702,
      "step": 296420
    },
    {
      "epoch": 0.14715106299654365,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001407142,
      "loss": 3.1251,
      "step": 296430
    },
    {
      "epoch": 0.14716970623270184,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014071220000000002,
      "loss": 3.0692,
      "step": 296440
    },
    {
      "epoch": 0.14718834946886003,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014071020000000001,
      "loss": 3.1039,
      "step": 296450
    },
    {
      "epoch": 0.14720699270501822,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001407082,
      "loss": 3.0586,
      "step": 296460
    },
    {
      "epoch": 0.1472256359411764,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001407062,
      "loss": 3.2063,
      "step": 296470
    },
    {
      "epoch": 0.14724427917733457,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001407042,
      "loss": 3.1243,
      "step": 296480
    },
    {
      "epoch": 0.14726292241349276,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001407022,
      "loss": 3.069,
      "step": 296490
    },
    {
      "epoch": 0.14728156564965095,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014070020000000002,
      "loss": 3.1179,
      "step": 296500
    },
    {
      "epoch": 0.14730020888580914,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014069820000000001,
      "loss": 3.2082,
      "step": 296510
    },
    {
      "epoch": 0.14731885212196733,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001406962,
      "loss": 3.1603,
      "step": 296520
    },
    {
      "epoch": 0.1473374953581255,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001406942,
      "loss": 3.0136,
      "step": 296530
    },
    {
      "epoch": 0.14735613859428368,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001406922,
      "loss": 3.1973,
      "step": 296540
    },
    {
      "epoch": 0.14737478183044186,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014069020000000002,
      "loss": 3.2313,
      "step": 296550
    },
    {
      "epoch": 0.14739342506660005,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001406882,
      "loss": 3.1917,
      "step": 296560
    },
    {
      "epoch": 0.14741206830275824,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014068620000000002,
      "loss": 3.0016,
      "step": 296570
    },
    {
      "epoch": 0.1474307115389164,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001406842,
      "loss": 3.1682,
      "step": 296580
    },
    {
      "epoch": 0.1474493547750746,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001406822,
      "loss": 3.012,
      "step": 296590
    },
    {
      "epoch": 0.14746799801123278,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014068020000000003,
      "loss": 3.0006,
      "step": 296600
    },
    {
      "epoch": 0.14748664124739097,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001406782,
      "loss": 3.1737,
      "step": 296610
    },
    {
      "epoch": 0.14750528448354916,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014067620000000002,
      "loss": 3.1009,
      "step": 296620
    },
    {
      "epoch": 0.14752392771970732,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001406742,
      "loss": 3.0668,
      "step": 296630
    },
    {
      "epoch": 0.1475425709558655,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001406722,
      "loss": 3.0815,
      "step": 296640
    },
    {
      "epoch": 0.1475612141920237,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001406702,
      "loss": 3.0523,
      "step": 296650
    },
    {
      "epoch": 0.1475798574281819,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001406682,
      "loss": 2.9884,
      "step": 296660
    },
    {
      "epoch": 0.14759850066434008,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014066620000000003,
      "loss": 3.0538,
      "step": 296670
    },
    {
      "epoch": 0.14761714390049827,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001406642,
      "loss": 3.012,
      "step": 296680
    },
    {
      "epoch": 0.14763578713665643,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014066220000000002,
      "loss": 3.0239,
      "step": 296690
    },
    {
      "epoch": 0.14765443037281462,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014066019999999999,
      "loss": 3.1258,
      "step": 296700
    },
    {
      "epoch": 0.1476730736089728,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001406582,
      "loss": 3.2502,
      "step": 296710
    },
    {
      "epoch": 0.147691716845131,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001406562,
      "loss": 3.1782,
      "step": 296720
    },
    {
      "epoch": 0.14771036008128918,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001406542,
      "loss": 3.1444,
      "step": 296730
    },
    {
      "epoch": 0.14772900331744734,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014065220000000002,
      "loss": 3.2174,
      "step": 296740
    },
    {
      "epoch": 0.14774764655360553,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001406502,
      "loss": 3.1204,
      "step": 296750
    },
    {
      "epoch": 0.14776628978976372,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001406482,
      "loss": 3.2938,
      "step": 296760
    },
    {
      "epoch": 0.1477849330259219,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001406462,
      "loss": 3.2001,
      "step": 296770
    },
    {
      "epoch": 0.1478035762620801,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001406442,
      "loss": 3.0688,
      "step": 296780
    },
    {
      "epoch": 0.14782221949823826,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001406422,
      "loss": 3.0483,
      "step": 296790
    },
    {
      "epoch": 0.14784086273439645,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001406402,
      "loss": 3.0845,
      "step": 296800
    },
    {
      "epoch": 0.14785950597055464,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014063820000000002,
      "loss": 3.0636,
      "step": 296810
    },
    {
      "epoch": 0.14787814920671283,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001406362,
      "loss": 3.2163,
      "step": 296820
    },
    {
      "epoch": 0.14789679244287102,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001406342,
      "loss": 3.1116,
      "step": 296830
    },
    {
      "epoch": 0.14791543567902918,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001406322,
      "loss": 3.1081,
      "step": 296840
    },
    {
      "epoch": 0.14793407891518737,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001406302,
      "loss": 3.2247,
      "step": 296850
    },
    {
      "epoch": 0.14795272215134556,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001406282,
      "loss": 3.0584,
      "step": 296860
    },
    {
      "epoch": 0.14797136538750374,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001406262,
      "loss": 3.2024,
      "step": 296870
    },
    {
      "epoch": 0.14799000862366193,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014062420000000001,
      "loss": 3.28,
      "step": 296880
    },
    {
      "epoch": 0.14800865185982012,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001406222,
      "loss": 3.0316,
      "step": 296890
    },
    {
      "epoch": 0.14802729509597828,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001406202,
      "loss": 3.2607,
      "step": 296900
    },
    {
      "epoch": 0.14804593833213647,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001406182,
      "loss": 3.0766,
      "step": 296910
    },
    {
      "epoch": 0.14806458156829466,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001406162,
      "loss": 3.0684,
      "step": 296920
    },
    {
      "epoch": 0.14808322480445285,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014061420000000002,
      "loss": 3.223,
      "step": 296930
    },
    {
      "epoch": 0.14810186804061104,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014061220000000002,
      "loss": 3.3042,
      "step": 296940
    },
    {
      "epoch": 0.1481205112767692,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001406102,
      "loss": 3.2715,
      "step": 296950
    },
    {
      "epoch": 0.1481391545129274,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001406082,
      "loss": 3.1923,
      "step": 296960
    },
    {
      "epoch": 0.14815779774908558,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001406062,
      "loss": 3.1014,
      "step": 296970
    },
    {
      "epoch": 0.14817644098524377,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001406042,
      "loss": 3.0782,
      "step": 296980
    },
    {
      "epoch": 0.14819508422140196,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014060220000000002,
      "loss": 3.0883,
      "step": 296990
    },
    {
      "epoch": 0.14821372745756012,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014060020000000002,
      "loss": 3.1248,
      "step": 297000
    },
    {
      "epoch": 0.1482323706937183,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001405982,
      "loss": 3.0411,
      "step": 297010
    },
    {
      "epoch": 0.1482510139298765,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001405962,
      "loss": 3.154,
      "step": 297020
    },
    {
      "epoch": 0.14826965716603469,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001405942,
      "loss": 3.0871,
      "step": 297030
    },
    {
      "epoch": 0.14828830040219287,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001405922,
      "loss": 3.1342,
      "step": 297040
    },
    {
      "epoch": 0.14830694363835104,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001405902,
      "loss": 3.1935,
      "step": 297050
    },
    {
      "epoch": 0.14832558687450922,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014058820000000002,
      "loss": 3.067,
      "step": 297060
    },
    {
      "epoch": 0.1483442301106674,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001405862,
      "loss": 3.2605,
      "step": 297070
    },
    {
      "epoch": 0.1483628733468256,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001405842,
      "loss": 3.0988,
      "step": 297080
    },
    {
      "epoch": 0.1483815165829838,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001405822,
      "loss": 3.1746,
      "step": 297090
    },
    {
      "epoch": 0.14840015981914195,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001405802,
      "loss": 3.0519,
      "step": 297100
    },
    {
      "epoch": 0.14841880305530014,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014057820000000002,
      "loss": 3.1639,
      "step": 297110
    },
    {
      "epoch": 0.14843744629145833,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001405762,
      "loss": 3.1226,
      "step": 297120
    },
    {
      "epoch": 0.14845608952761652,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001405742,
      "loss": 3.1151,
      "step": 297130
    },
    {
      "epoch": 0.1484747327637747,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001405722,
      "loss": 3.1641,
      "step": 297140
    },
    {
      "epoch": 0.1484933759999329,
      "grad_norm": 0.75,
      "learning_rate": 0.0001405702,
      "loss": 3.1746,
      "step": 297150
    },
    {
      "epoch": 0.14851201923609106,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014056820000000003,
      "loss": 3.2128,
      "step": 297160
    },
    {
      "epoch": 0.14853066247224925,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001405662,
      "loss": 3.165,
      "step": 297170
    },
    {
      "epoch": 0.14854930570840744,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014056420000000002,
      "loss": 3.2433,
      "step": 297180
    },
    {
      "epoch": 0.14856794894456563,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001405622,
      "loss": 3.0395,
      "step": 297190
    },
    {
      "epoch": 0.14858659218072381,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001405602,
      "loss": 3.1553,
      "step": 297200
    },
    {
      "epoch": 0.14860523541688198,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001405582,
      "loss": 3.1459,
      "step": 297210
    },
    {
      "epoch": 0.14862387865304016,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001405562,
      "loss": 3.2227,
      "step": 297220
    },
    {
      "epoch": 0.14864252188919835,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014055420000000002,
      "loss": 3.0909,
      "step": 297230
    },
    {
      "epoch": 0.14866116512535654,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001405522,
      "loss": 3.183,
      "step": 297240
    },
    {
      "epoch": 0.14867980836151473,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014055020000000002,
      "loss": 3.0541,
      "step": 297250
    },
    {
      "epoch": 0.1486984515976729,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014054819999999998,
      "loss": 3.0604,
      "step": 297260
    },
    {
      "epoch": 0.14871709483383108,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001405462,
      "loss": 3.0983,
      "step": 297270
    },
    {
      "epoch": 0.14873573806998927,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001405442,
      "loss": 3.1804,
      "step": 297280
    },
    {
      "epoch": 0.14875438130614746,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001405422,
      "loss": 3.1098,
      "step": 297290
    },
    {
      "epoch": 0.14877302454230565,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014054020000000002,
      "loss": 3.0886,
      "step": 297300
    },
    {
      "epoch": 0.1487916677784638,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001405382,
      "loss": 3.1751,
      "step": 297310
    },
    {
      "epoch": 0.148810311014622,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001405362,
      "loss": 3.1406,
      "step": 297320
    },
    {
      "epoch": 0.1488289542507802,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001405342,
      "loss": 3.2405,
      "step": 297330
    },
    {
      "epoch": 0.14884759748693838,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001405322,
      "loss": 3.0732,
      "step": 297340
    },
    {
      "epoch": 0.14886624072309657,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001405302,
      "loss": 3.1128,
      "step": 297350
    },
    {
      "epoch": 0.14888488395925475,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001405282,
      "loss": 3.1931,
      "step": 297360
    },
    {
      "epoch": 0.14890352719541292,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014052620000000002,
      "loss": 3.1103,
      "step": 297370
    },
    {
      "epoch": 0.1489221704315711,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001405242,
      "loss": 3.2937,
      "step": 297380
    },
    {
      "epoch": 0.1489408136677293,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001405222,
      "loss": 3.2517,
      "step": 297390
    },
    {
      "epoch": 0.14895945690388748,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001405202,
      "loss": 3.0477,
      "step": 297400
    },
    {
      "epoch": 0.14897810014004567,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001405182,
      "loss": 3.2096,
      "step": 297410
    },
    {
      "epoch": 0.14899674337620383,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014051620000000002,
      "loss": 3.1106,
      "step": 297420
    },
    {
      "epoch": 0.14901538661236202,
      "grad_norm": 0.75,
      "learning_rate": 0.0001405142,
      "loss": 3.0246,
      "step": 297430
    },
    {
      "epoch": 0.1490340298485202,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001405122,
      "loss": 3.0901,
      "step": 297440
    },
    {
      "epoch": 0.1490526730846784,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001405102,
      "loss": 3.0825,
      "step": 297450
    },
    {
      "epoch": 0.1490713163208366,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001405082,
      "loss": 3.114,
      "step": 297460
    },
    {
      "epoch": 0.14908995955699475,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001405062,
      "loss": 3.0775,
      "step": 297470
    },
    {
      "epoch": 0.14910860279315294,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001405042,
      "loss": 3.0995,
      "step": 297480
    },
    {
      "epoch": 0.14912724602931113,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014050220000000002,
      "loss": 3.1378,
      "step": 297490
    },
    {
      "epoch": 0.14914588926546932,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00014050020000000001,
      "loss": 3.088,
      "step": 297500
    },
    {
      "epoch": 0.1491645325016275,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001404982,
      "loss": 2.9116,
      "step": 297510
    },
    {
      "epoch": 0.14918317573778567,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001404962,
      "loss": 3.0903,
      "step": 297520
    },
    {
      "epoch": 0.14920181897394386,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001404942,
      "loss": 3.1657,
      "step": 297530
    },
    {
      "epoch": 0.14922046221010205,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001404922,
      "loss": 2.9521,
      "step": 297540
    },
    {
      "epoch": 0.14923910544626023,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014049020000000002,
      "loss": 3.1576,
      "step": 297550
    },
    {
      "epoch": 0.14925774868241842,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014048820000000001,
      "loss": 3.265,
      "step": 297560
    },
    {
      "epoch": 0.1492763919185766,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001404862,
      "loss": 3.1523,
      "step": 297570
    },
    {
      "epoch": 0.14929503515473477,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001404842,
      "loss": 3.1703,
      "step": 297580
    },
    {
      "epoch": 0.14931367839089296,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001404822,
      "loss": 3.1133,
      "step": 297590
    },
    {
      "epoch": 0.14933232162705115,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014048020000000002,
      "loss": 3.1424,
      "step": 297600
    },
    {
      "epoch": 0.14935096486320934,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001404782,
      "loss": 3.0641,
      "step": 297610
    },
    {
      "epoch": 0.14936960809936753,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014047620000000002,
      "loss": 3.046,
      "step": 297620
    },
    {
      "epoch": 0.1493882513355257,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001404742,
      "loss": 3.1494,
      "step": 297630
    },
    {
      "epoch": 0.14940689457168388,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001404722,
      "loss": 3.156,
      "step": 297640
    },
    {
      "epoch": 0.14942553780784207,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001404702,
      "loss": 3.1415,
      "step": 297650
    },
    {
      "epoch": 0.14944418104400026,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001404682,
      "loss": 3.1823,
      "step": 297660
    },
    {
      "epoch": 0.14946282428015845,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014046620000000002,
      "loss": 3.1565,
      "step": 297670
    },
    {
      "epoch": 0.1494814675163166,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001404642,
      "loss": 3.2035,
      "step": 297680
    },
    {
      "epoch": 0.1495001107524748,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001404622,
      "loss": 3.1423,
      "step": 297690
    },
    {
      "epoch": 0.14951875398863299,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001404602,
      "loss": 3.1037,
      "step": 297700
    },
    {
      "epoch": 0.14953739722479117,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001404582,
      "loss": 3.0843,
      "step": 297710
    },
    {
      "epoch": 0.14955604046094936,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014045620000000003,
      "loss": 3.1707,
      "step": 297720
    },
    {
      "epoch": 0.14957468369710752,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001404542,
      "loss": 3.0878,
      "step": 297730
    },
    {
      "epoch": 0.1495933269332657,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014045220000000002,
      "loss": 3.2201,
      "step": 297740
    },
    {
      "epoch": 0.1496119701694239,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014045019999999999,
      "loss": 2.9605,
      "step": 297750
    },
    {
      "epoch": 0.1496306134055821,
      "grad_norm": 0.625,
      "learning_rate": 0.0001404482,
      "loss": 3.1873,
      "step": 297760
    },
    {
      "epoch": 0.14964925664174028,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001404462,
      "loss": 3.172,
      "step": 297770
    },
    {
      "epoch": 0.14966789987789844,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001404442,
      "loss": 3.0924,
      "step": 297780
    },
    {
      "epoch": 0.14968654311405663,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014044220000000002,
      "loss": 3.1161,
      "step": 297790
    },
    {
      "epoch": 0.14970518635021482,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001404402,
      "loss": 3.1442,
      "step": 297800
    },
    {
      "epoch": 0.149723829586373,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001404382,
      "loss": 3.1223,
      "step": 297810
    },
    {
      "epoch": 0.1497424728225312,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001404362,
      "loss": 3.1394,
      "step": 297820
    },
    {
      "epoch": 0.1497611160586894,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001404342,
      "loss": 3.1173,
      "step": 297830
    },
    {
      "epoch": 0.14977975929484755,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001404322,
      "loss": 3.1995,
      "step": 297840
    },
    {
      "epoch": 0.14979840253100574,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001404302,
      "loss": 3.2306,
      "step": 297850
    },
    {
      "epoch": 0.14981704576716393,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014042820000000002,
      "loss": 3.2221,
      "step": 297860
    },
    {
      "epoch": 0.14983568900332211,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001404262,
      "loss": 3.0775,
      "step": 297870
    },
    {
      "epoch": 0.1498543322394803,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001404242,
      "loss": 3.1565,
      "step": 297880
    },
    {
      "epoch": 0.14987297547563846,
      "grad_norm": 0.875,
      "learning_rate": 0.0001404222,
      "loss": 3.2044,
      "step": 297890
    },
    {
      "epoch": 0.14989161871179665,
      "grad_norm": 0.75,
      "learning_rate": 0.0001404202,
      "loss": 3.0615,
      "step": 297900
    },
    {
      "epoch": 0.14991026194795484,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014041820000000002,
      "loss": 3.0666,
      "step": 297910
    },
    {
      "epoch": 0.14992890518411303,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001404162,
      "loss": 2.9494,
      "step": 297920
    },
    {
      "epoch": 0.14994754842027122,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014041420000000002,
      "loss": 3.2478,
      "step": 297930
    },
    {
      "epoch": 0.14996619165642938,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001404122,
      "loss": 3.164,
      "step": 297940
    },
    {
      "epoch": 0.14998483489258757,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001404102,
      "loss": 3.2078,
      "step": 297950
    },
    {
      "epoch": 0.15000347812874576,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001404082,
      "loss": 3.1687,
      "step": 297960
    },
    {
      "epoch": 0.15002212136490395,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001404062,
      "loss": 3.1157,
      "step": 297970
    },
    {
      "epoch": 0.15004076460106214,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014040420000000002,
      "loss": 3.0306,
      "step": 297980
    },
    {
      "epoch": 0.1500594078372203,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014040220000000002,
      "loss": 3.0855,
      "step": 297990
    },
    {
      "epoch": 0.1500780510733785,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001404002,
      "loss": 3.2049,
      "step": 298000
    },
    {
      "epoch": 0.15009669430953668,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001403982,
      "loss": 3.1069,
      "step": 298010
    },
    {
      "epoch": 0.15011533754569487,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001403962,
      "loss": 3.101,
      "step": 298020
    },
    {
      "epoch": 0.15013398078185305,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001403942,
      "loss": 3.0895,
      "step": 298030
    },
    {
      "epoch": 0.15015262401801124,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001403922,
      "loss": 3.0826,
      "step": 298040
    },
    {
      "epoch": 0.1501712672541694,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014039020000000002,
      "loss": 3.2779,
      "step": 298050
    },
    {
      "epoch": 0.1501899104903276,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001403882,
      "loss": 3.1469,
      "step": 298060
    },
    {
      "epoch": 0.15020855372648578,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001403862,
      "loss": 3.2712,
      "step": 298070
    },
    {
      "epoch": 0.15022719696264397,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001403842,
      "loss": 3.0625,
      "step": 298080
    },
    {
      "epoch": 0.15024584019880216,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001403822,
      "loss": 3.081,
      "step": 298090
    },
    {
      "epoch": 0.15026448343496032,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001403802,
      "loss": 3.0908,
      "step": 298100
    },
    {
      "epoch": 0.1502831266711185,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014037820000000002,
      "loss": 3.2489,
      "step": 298110
    },
    {
      "epoch": 0.1503017699072767,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001403762,
      "loss": 3.169,
      "step": 298120
    },
    {
      "epoch": 0.1503204131434349,
      "grad_norm": 0.75,
      "learning_rate": 0.0001403742,
      "loss": 2.892,
      "step": 298130
    },
    {
      "epoch": 0.15033905637959308,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001403722,
      "loss": 3.2524,
      "step": 298140
    },
    {
      "epoch": 0.15035769961575124,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001403702,
      "loss": 3.1193,
      "step": 298150
    },
    {
      "epoch": 0.15037634285190943,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014036820000000002,
      "loss": 3.1274,
      "step": 298160
    },
    {
      "epoch": 0.15039498608806762,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001403662,
      "loss": 3.0084,
      "step": 298170
    },
    {
      "epoch": 0.1504136293242258,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014036420000000001,
      "loss": 3.1805,
      "step": 298180
    },
    {
      "epoch": 0.150432272560384,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001403622,
      "loss": 3.0788,
      "step": 298190
    },
    {
      "epoch": 0.15045091579654216,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001403602,
      "loss": 3.1578,
      "step": 298200
    },
    {
      "epoch": 0.15046955903270035,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014035820000000003,
      "loss": 3.1279,
      "step": 298210
    },
    {
      "epoch": 0.15048820226885853,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001403562,
      "loss": 3.0694,
      "step": 298220
    },
    {
      "epoch": 0.15050684550501672,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014035420000000002,
      "loss": 3.0573,
      "step": 298230
    },
    {
      "epoch": 0.1505254887411749,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001403522,
      "loss": 3.1001,
      "step": 298240
    },
    {
      "epoch": 0.1505441319773331,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001403502,
      "loss": 3.1426,
      "step": 298250
    },
    {
      "epoch": 0.15056277521349126,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001403482,
      "loss": 3.0743,
      "step": 298260
    },
    {
      "epoch": 0.15058141844964945,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001403462,
      "loss": 3.1406,
      "step": 298270
    },
    {
      "epoch": 0.15060006168580764,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014034420000000002,
      "loss": 3.0807,
      "step": 298280
    },
    {
      "epoch": 0.15061870492196583,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001403422,
      "loss": 3.1925,
      "step": 298290
    },
    {
      "epoch": 0.15063734815812402,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014034020000000002,
      "loss": 3.1201,
      "step": 298300
    },
    {
      "epoch": 0.15065599139428218,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014033819999999998,
      "loss": 3.1464,
      "step": 298310
    },
    {
      "epoch": 0.15067463463044037,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001403362,
      "loss": 3.0236,
      "step": 298320
    },
    {
      "epoch": 0.15069327786659856,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001403342,
      "loss": 3.1707,
      "step": 298330
    },
    {
      "epoch": 0.15071192110275675,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001403322,
      "loss": 3.1161,
      "step": 298340
    },
    {
      "epoch": 0.15073056433891494,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014033020000000002,
      "loss": 3.1655,
      "step": 298350
    },
    {
      "epoch": 0.1507492075750731,
      "grad_norm": 0.625,
      "learning_rate": 0.0001403282,
      "loss": 3.1222,
      "step": 298360
    },
    {
      "epoch": 0.15076785081123129,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001403262,
      "loss": 3.1427,
      "step": 298370
    },
    {
      "epoch": 0.15078649404738947,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001403242,
      "loss": 3.1022,
      "step": 298380
    },
    {
      "epoch": 0.15080513728354766,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001403222,
      "loss": 3.274,
      "step": 298390
    },
    {
      "epoch": 0.15082378051970585,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014032020000000003,
      "loss": 3.0985,
      "step": 298400
    },
    {
      "epoch": 0.150842423755864,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001403182,
      "loss": 3.1723,
      "step": 298410
    },
    {
      "epoch": 0.1508610669920222,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014031620000000002,
      "loss": 3.1092,
      "step": 298420
    },
    {
      "epoch": 0.1508797102281804,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014031419999999999,
      "loss": 3.181,
      "step": 298430
    },
    {
      "epoch": 0.15089835346433858,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001403122,
      "loss": 3.1796,
      "step": 298440
    },
    {
      "epoch": 0.15091699670049677,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001403102,
      "loss": 3.1601,
      "step": 298450
    },
    {
      "epoch": 0.15093563993665493,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001403082,
      "loss": 3.0112,
      "step": 298460
    },
    {
      "epoch": 0.15095428317281312,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014030620000000002,
      "loss": 3.157,
      "step": 298470
    },
    {
      "epoch": 0.1509729264089713,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001403042,
      "loss": 3.1376,
      "step": 298480
    },
    {
      "epoch": 0.1509915696451295,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001403022,
      "loss": 3.1394,
      "step": 298490
    },
    {
      "epoch": 0.1510102128812877,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001403002,
      "loss": 3.0942,
      "step": 298500
    },
    {
      "epoch": 0.15102885611744588,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001402982,
      "loss": 3.1883,
      "step": 298510
    },
    {
      "epoch": 0.15104749935360404,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001402962,
      "loss": 3.1844,
      "step": 298520
    },
    {
      "epoch": 0.15106614258976223,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001402942,
      "loss": 3.155,
      "step": 298530
    },
    {
      "epoch": 0.15108478582592041,
      "grad_norm": 0.875,
      "learning_rate": 0.00014029220000000002,
      "loss": 3.0794,
      "step": 298540
    },
    {
      "epoch": 0.1511034290620786,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014029020000000001,
      "loss": 3.0858,
      "step": 298550
    },
    {
      "epoch": 0.1511220722982368,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001402882,
      "loss": 3.1856,
      "step": 298560
    },
    {
      "epoch": 0.15114071553439495,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001402862,
      "loss": 3.0589,
      "step": 298570
    },
    {
      "epoch": 0.15115935877055314,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001402842,
      "loss": 3.1193,
      "step": 298580
    },
    {
      "epoch": 0.15117800200671133,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001402822,
      "loss": 3.0997,
      "step": 298590
    },
    {
      "epoch": 0.15119664524286952,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014028020000000002,
      "loss": 3.2043,
      "step": 298600
    },
    {
      "epoch": 0.1512152884790277,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014027820000000001,
      "loss": 3.2541,
      "step": 298610
    },
    {
      "epoch": 0.15123393171518587,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001402762,
      "loss": 3.1755,
      "step": 298620
    },
    {
      "epoch": 0.15125257495134406,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001402742,
      "loss": 3.0596,
      "step": 298630
    },
    {
      "epoch": 0.15127121818750225,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001402722,
      "loss": 3.0579,
      "step": 298640
    },
    {
      "epoch": 0.15128986142366044,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001402702,
      "loss": 3.0735,
      "step": 298650
    },
    {
      "epoch": 0.15130850465981863,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001402682,
      "loss": 3.0548,
      "step": 298660
    },
    {
      "epoch": 0.1513271478959768,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014026620000000002,
      "loss": 3.2591,
      "step": 298670
    },
    {
      "epoch": 0.15134579113213498,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001402642,
      "loss": 3.165,
      "step": 298680
    },
    {
      "epoch": 0.15136443436829317,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001402622,
      "loss": 3.1274,
      "step": 298690
    },
    {
      "epoch": 0.15138307760445135,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001402602,
      "loss": 3.115,
      "step": 298700
    },
    {
      "epoch": 0.15140172084060954,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001402582,
      "loss": 3.083,
      "step": 298710
    },
    {
      "epoch": 0.15142036407676773,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014025620000000002,
      "loss": 3.2583,
      "step": 298720
    },
    {
      "epoch": 0.1514390073129259,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001402542,
      "loss": 3.1456,
      "step": 298730
    },
    {
      "epoch": 0.15145765054908408,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001402522,
      "loss": 3.0688,
      "step": 298740
    },
    {
      "epoch": 0.15147629378524227,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001402502,
      "loss": 3.1045,
      "step": 298750
    },
    {
      "epoch": 0.15149493702140046,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001402482,
      "loss": 3.1052,
      "step": 298760
    },
    {
      "epoch": 0.15151358025755865,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014024620000000003,
      "loss": 3.1743,
      "step": 298770
    },
    {
      "epoch": 0.1515322234937168,
      "grad_norm": 0.75,
      "learning_rate": 0.0001402442,
      "loss": 3.037,
      "step": 298780
    },
    {
      "epoch": 0.151550866729875,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014024220000000002,
      "loss": 3.0161,
      "step": 298790
    },
    {
      "epoch": 0.1515695099660332,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014024019999999999,
      "loss": 3.1786,
      "step": 298800
    },
    {
      "epoch": 0.15158815320219138,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001402382,
      "loss": 3.1442,
      "step": 298810
    },
    {
      "epoch": 0.15160679643834957,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001402362,
      "loss": 3.0353,
      "step": 298820
    },
    {
      "epoch": 0.15162543967450773,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001402342,
      "loss": 3.122,
      "step": 298830
    },
    {
      "epoch": 0.15164408291066592,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014023220000000002,
      "loss": 3.0363,
      "step": 298840
    },
    {
      "epoch": 0.1516627261468241,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001402302,
      "loss": 3.1827,
      "step": 298850
    },
    {
      "epoch": 0.1516813693829823,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014022820000000001,
      "loss": 3.1066,
      "step": 298860
    },
    {
      "epoch": 0.15170001261914048,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001402262,
      "loss": 3.0652,
      "step": 298870
    },
    {
      "epoch": 0.15171865585529865,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001402242,
      "loss": 3.1078,
      "step": 298880
    },
    {
      "epoch": 0.15173729909145683,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014022220000000003,
      "loss": 3.0995,
      "step": 298890
    },
    {
      "epoch": 0.15175594232761502,
      "grad_norm": 0.625,
      "learning_rate": 0.0001402202,
      "loss": 3.1768,
      "step": 298900
    },
    {
      "epoch": 0.1517745855637732,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014021820000000002,
      "loss": 3.1514,
      "step": 298910
    },
    {
      "epoch": 0.1517932287999314,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001402162,
      "loss": 2.998,
      "step": 298920
    },
    {
      "epoch": 0.1518118720360896,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001402142,
      "loss": 3.0868,
      "step": 298930
    },
    {
      "epoch": 0.15183051527224775,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001402122,
      "loss": 3.1215,
      "step": 298940
    },
    {
      "epoch": 0.15184915850840594,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001402102,
      "loss": 3.0662,
      "step": 298950
    },
    {
      "epoch": 0.15186780174456413,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014020820000000002,
      "loss": 3.3269,
      "step": 298960
    },
    {
      "epoch": 0.15188644498072232,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001402062,
      "loss": 3.167,
      "step": 298970
    },
    {
      "epoch": 0.1519050882168805,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014020420000000002,
      "loss": 2.9236,
      "step": 298980
    },
    {
      "epoch": 0.15192373145303867,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001402022,
      "loss": 3.0859,
      "step": 298990
    },
    {
      "epoch": 0.15194237468919686,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001402002,
      "loss": 3.0479,
      "step": 299000
    },
    {
      "epoch": 0.15196101792535505,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001401982,
      "loss": 3.1227,
      "step": 299010
    },
    {
      "epoch": 0.15197966116151324,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001401962,
      "loss": 3.0457,
      "step": 299020
    },
    {
      "epoch": 0.15199830439767142,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014019420000000002,
      "loss": 2.9906,
      "step": 299030
    },
    {
      "epoch": 0.15201694763382959,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001401922,
      "loss": 3.1923,
      "step": 299040
    },
    {
      "epoch": 0.15203559086998777,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001401902,
      "loss": 3.1884,
      "step": 299050
    },
    {
      "epoch": 0.15205423410614596,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001401882,
      "loss": 3.222,
      "step": 299060
    },
    {
      "epoch": 0.15207287734230415,
      "grad_norm": 0.625,
      "learning_rate": 0.0001401862,
      "loss": 3.2335,
      "step": 299070
    },
    {
      "epoch": 0.15209152057846234,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001401842,
      "loss": 3.1402,
      "step": 299080
    },
    {
      "epoch": 0.1521101638146205,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001401822,
      "loss": 3.1643,
      "step": 299090
    },
    {
      "epoch": 0.1521288070507787,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014018020000000002,
      "loss": 3.0278,
      "step": 299100
    },
    {
      "epoch": 0.15214745028693688,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001401782,
      "loss": 3.0985,
      "step": 299110
    },
    {
      "epoch": 0.15216609352309507,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001401762,
      "loss": 3.2377,
      "step": 299120
    },
    {
      "epoch": 0.15218473675925326,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001401742,
      "loss": 3.0722,
      "step": 299130
    },
    {
      "epoch": 0.15220337999541142,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001401722,
      "loss": 3.0773,
      "step": 299140
    },
    {
      "epoch": 0.1522220232315696,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001401702,
      "loss": 3.06,
      "step": 299150
    },
    {
      "epoch": 0.1522406664677278,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014016820000000002,
      "loss": 3.0915,
      "step": 299160
    },
    {
      "epoch": 0.152259309703886,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001401662,
      "loss": 3.1979,
      "step": 299170
    },
    {
      "epoch": 0.15227795294004418,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001401642,
      "loss": 3.0416,
      "step": 299180
    },
    {
      "epoch": 0.15229659617620236,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001401622,
      "loss": 3.2424,
      "step": 299190
    },
    {
      "epoch": 0.15231523941236053,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001401602,
      "loss": 3.1582,
      "step": 299200
    },
    {
      "epoch": 0.15233388264851871,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014015820000000002,
      "loss": 3.1512,
      "step": 299210
    },
    {
      "epoch": 0.1523525258846769,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001401562,
      "loss": 3.178,
      "step": 299220
    },
    {
      "epoch": 0.1523711691208351,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014015420000000001,
      "loss": 3.2295,
      "step": 299230
    },
    {
      "epoch": 0.15238981235699328,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001401522,
      "loss": 3.1003,
      "step": 299240
    },
    {
      "epoch": 0.15240845559315144,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001401502,
      "loss": 3.2126,
      "step": 299250
    },
    {
      "epoch": 0.15242709882930963,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001401482,
      "loss": 3.0915,
      "step": 299260
    },
    {
      "epoch": 0.15244574206546782,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001401462,
      "loss": 3.1243,
      "step": 299270
    },
    {
      "epoch": 0.152464385301626,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014014420000000002,
      "loss": 2.9835,
      "step": 299280
    },
    {
      "epoch": 0.1524830285377842,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001401422,
      "loss": 3.1561,
      "step": 299290
    },
    {
      "epoch": 0.15250167177394236,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001401402,
      "loss": 3.0879,
      "step": 299300
    },
    {
      "epoch": 0.15252031501010055,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001401382,
      "loss": 3.0171,
      "step": 299310
    },
    {
      "epoch": 0.15253895824625874,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001401362,
      "loss": 3.0832,
      "step": 299320
    },
    {
      "epoch": 0.15255760148241693,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014013420000000002,
      "loss": 3.1257,
      "step": 299330
    },
    {
      "epoch": 0.15257624471857512,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001401322,
      "loss": 3.1346,
      "step": 299340
    },
    {
      "epoch": 0.15259488795473328,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014013020000000002,
      "loss": 3.1337,
      "step": 299350
    },
    {
      "epoch": 0.15261353119089147,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001401282,
      "loss": 3.046,
      "step": 299360
    },
    {
      "epoch": 0.15263217442704965,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001401262,
      "loss": 3.1513,
      "step": 299370
    },
    {
      "epoch": 0.15265081766320784,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014012420000000003,
      "loss": 3.0635,
      "step": 299380
    },
    {
      "epoch": 0.15266946089936603,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001401222,
      "loss": 3.1223,
      "step": 299390
    },
    {
      "epoch": 0.15268810413552422,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014012020000000002,
      "loss": 3.1351,
      "step": 299400
    },
    {
      "epoch": 0.15270674737168238,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001401182,
      "loss": 3.1059,
      "step": 299410
    },
    {
      "epoch": 0.15272539060784057,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001401162,
      "loss": 3.1652,
      "step": 299420
    },
    {
      "epoch": 0.15274403384399876,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001401142,
      "loss": 3.0474,
      "step": 299430
    },
    {
      "epoch": 0.15276267708015695,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001401122,
      "loss": 3.1571,
      "step": 299440
    },
    {
      "epoch": 0.15278132031631514,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014011020000000003,
      "loss": 3.0824,
      "step": 299450
    },
    {
      "epoch": 0.1527999635524733,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001401082,
      "loss": 3.1293,
      "step": 299460
    },
    {
      "epoch": 0.1528186067886315,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014010620000000002,
      "loss": 3.0312,
      "step": 299470
    },
    {
      "epoch": 0.15283725002478968,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014010419999999999,
      "loss": 3.0984,
      "step": 299480
    },
    {
      "epoch": 0.15285589326094787,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001401022,
      "loss": 3.0782,
      "step": 299490
    },
    {
      "epoch": 0.15287453649710606,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001401002,
      "loss": 3.1868,
      "step": 299500
    },
    {
      "epoch": 0.15289317973326422,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001400982,
      "loss": 3.0903,
      "step": 299510
    },
    {
      "epoch": 0.1529118229694224,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014009620000000002,
      "loss": 2.9063,
      "step": 299520
    },
    {
      "epoch": 0.1529304662055806,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001400942,
      "loss": 3.2549,
      "step": 299530
    },
    {
      "epoch": 0.15294910944173878,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014009220000000001,
      "loss": 2.998,
      "step": 299540
    },
    {
      "epoch": 0.15296775267789697,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001400902,
      "loss": 3.1494,
      "step": 299550
    },
    {
      "epoch": 0.15298639591405513,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001400882,
      "loss": 3.1092,
      "step": 299560
    },
    {
      "epoch": 0.15300503915021332,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001400862,
      "loss": 3.1892,
      "step": 299570
    },
    {
      "epoch": 0.1530236823863715,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001400842,
      "loss": 3.1398,
      "step": 299580
    },
    {
      "epoch": 0.1530423256225297,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014008220000000002,
      "loss": 3.1005,
      "step": 299590
    },
    {
      "epoch": 0.1530609688586879,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014008020000000001,
      "loss": 2.9782,
      "step": 299600
    },
    {
      "epoch": 0.15307961209484608,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001400782,
      "loss": 3.121,
      "step": 299610
    },
    {
      "epoch": 0.15309825533100424,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001400762,
      "loss": 3.1335,
      "step": 299620
    },
    {
      "epoch": 0.15311689856716243,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001400742,
      "loss": 3.1615,
      "step": 299630
    },
    {
      "epoch": 0.15313554180332062,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001400722,
      "loss": 3.1778,
      "step": 299640
    },
    {
      "epoch": 0.1531541850394788,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001400702,
      "loss": 3.1233,
      "step": 299650
    },
    {
      "epoch": 0.153172828275637,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014006820000000002,
      "loss": 3.0065,
      "step": 299660
    },
    {
      "epoch": 0.15319147151179516,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001400662,
      "loss": 3.124,
      "step": 299670
    },
    {
      "epoch": 0.15321011474795335,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001400642,
      "loss": 3.0764,
      "step": 299680
    },
    {
      "epoch": 0.15322875798411154,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001400622,
      "loss": 3.1987,
      "step": 299690
    },
    {
      "epoch": 0.15324740122026972,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001400602,
      "loss": 3.037,
      "step": 299700
    },
    {
      "epoch": 0.1532660444564279,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001400582,
      "loss": 3.0559,
      "step": 299710
    },
    {
      "epoch": 0.15328468769258607,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014005620000000002,
      "loss": 3.121,
      "step": 299720
    },
    {
      "epoch": 0.15330333092874426,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001400542,
      "loss": 3.2101,
      "step": 299730
    },
    {
      "epoch": 0.15332197416490245,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001400522,
      "loss": 3.1437,
      "step": 299740
    },
    {
      "epoch": 0.15334061740106064,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001400502,
      "loss": 3.1841,
      "step": 299750
    },
    {
      "epoch": 0.15335926063721883,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001400482,
      "loss": 3.2051,
      "step": 299760
    },
    {
      "epoch": 0.153377903873377,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014004620000000002,
      "loss": 3.0898,
      "step": 299770
    },
    {
      "epoch": 0.15339654710953518,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001400442,
      "loss": 3.1989,
      "step": 299780
    },
    {
      "epoch": 0.15341519034569337,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001400422,
      "loss": 3.1006,
      "step": 299790
    },
    {
      "epoch": 0.15343383358185156,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001400402,
      "loss": 3.0981,
      "step": 299800
    },
    {
      "epoch": 0.15345247681800975,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001400382,
      "loss": 3.2168,
      "step": 299810
    },
    {
      "epoch": 0.1534711200541679,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014003620000000003,
      "loss": 3.1908,
      "step": 299820
    },
    {
      "epoch": 0.1534897632903261,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001400342,
      "loss": 3.0955,
      "step": 299830
    },
    {
      "epoch": 0.1535084065264843,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014003220000000002,
      "loss": 3.1887,
      "step": 299840
    },
    {
      "epoch": 0.15352704976264248,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001400302,
      "loss": 3.1007,
      "step": 299850
    },
    {
      "epoch": 0.15354569299880066,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001400282,
      "loss": 3.0876,
      "step": 299860
    },
    {
      "epoch": 0.15356433623495885,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001400262,
      "loss": 2.9491,
      "step": 299870
    },
    {
      "epoch": 0.15358297947111701,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001400242,
      "loss": 3.0974,
      "step": 299880
    },
    {
      "epoch": 0.1536016227072752,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014002220000000002,
      "loss": 3.1137,
      "step": 299890
    },
    {
      "epoch": 0.1536202659434334,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001400202,
      "loss": 3.0941,
      "step": 299900
    },
    {
      "epoch": 0.15363890917959158,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014001820000000001,
      "loss": 3.0617,
      "step": 299910
    },
    {
      "epoch": 0.15365755241574977,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001400162,
      "loss": 3.1823,
      "step": 299920
    },
    {
      "epoch": 0.15367619565190793,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001400142,
      "loss": 3.0069,
      "step": 299930
    },
    {
      "epoch": 0.15369483888806612,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014001220000000003,
      "loss": 3.0268,
      "step": 299940
    },
    {
      "epoch": 0.1537134821242243,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001400102,
      "loss": 3.1957,
      "step": 299950
    },
    {
      "epoch": 0.1537321253603825,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014000820000000002,
      "loss": 3.1488,
      "step": 299960
    },
    {
      "epoch": 0.1537507685965407,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001400062,
      "loss": 3.0088,
      "step": 299970
    },
    {
      "epoch": 0.15376941183269885,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001400042,
      "loss": 3.2037,
      "step": 299980
    },
    {
      "epoch": 0.15378805506885704,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001400022,
      "loss": 3.0555,
      "step": 299990
    },
    {
      "epoch": 0.15380669830501523,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001400002,
      "loss": 3.1387,
      "step": 300000
    },
    {
      "epoch": 0.15382534154117342,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013999820000000002,
      "loss": 3.1916,
      "step": 300010
    },
    {
      "epoch": 0.1538439847773316,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001399962,
      "loss": 3.1512,
      "step": 300020
    },
    {
      "epoch": 0.15386262801348977,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013999420000000002,
      "loss": 3.0204,
      "step": 300030
    },
    {
      "epoch": 0.15388127124964796,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001399922,
      "loss": 3.1033,
      "step": 300040
    },
    {
      "epoch": 0.15389991448580614,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001399902,
      "loss": 3.0063,
      "step": 300050
    },
    {
      "epoch": 0.15391855772196433,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001399882,
      "loss": 3.0346,
      "step": 300060
    },
    {
      "epoch": 0.15393720095812252,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001399862,
      "loss": 3.181,
      "step": 300070
    },
    {
      "epoch": 0.1539558441942807,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013998420000000002,
      "loss": 3.1527,
      "step": 300080
    },
    {
      "epoch": 0.15397448743043887,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001399822,
      "loss": 3.0973,
      "step": 300090
    },
    {
      "epoch": 0.15399313066659706,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001399802,
      "loss": 3.0831,
      "step": 300100
    },
    {
      "epoch": 0.15401177390275525,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001399782,
      "loss": 3.1503,
      "step": 300110
    },
    {
      "epoch": 0.15403041713891344,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001399762,
      "loss": 3.0666,
      "step": 300120
    },
    {
      "epoch": 0.15404906037507163,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001399742,
      "loss": 3.1098,
      "step": 300130
    },
    {
      "epoch": 0.1540677036112298,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001399722,
      "loss": 3.1914,
      "step": 300140
    },
    {
      "epoch": 0.15408634684738798,
      "grad_norm": 0.75,
      "learning_rate": 0.00013997020000000002,
      "loss": 3.1269,
      "step": 300150
    },
    {
      "epoch": 0.15410499008354617,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001399682,
      "loss": 3.0427,
      "step": 300160
    },
    {
      "epoch": 0.15412363331970436,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001399662,
      "loss": 3.1352,
      "step": 300170
    },
    {
      "epoch": 0.15414227655586255,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001399642,
      "loss": 3.1778,
      "step": 300180
    },
    {
      "epoch": 0.1541609197920207,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001399622,
      "loss": 3.1578,
      "step": 300190
    },
    {
      "epoch": 0.1541795630281789,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001399602,
      "loss": 3.1044,
      "step": 300200
    },
    {
      "epoch": 0.15419820626433708,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013995820000000002,
      "loss": 3.1414,
      "step": 300210
    },
    {
      "epoch": 0.15421684950049527,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013995620000000001,
      "loss": 3.055,
      "step": 300220
    },
    {
      "epoch": 0.15423549273665346,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001399542,
      "loss": 3.224,
      "step": 300230
    },
    {
      "epoch": 0.15425413597281162,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001399522,
      "loss": 3.206,
      "step": 300240
    },
    {
      "epoch": 0.1542727792089698,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001399502,
      "loss": 3.1613,
      "step": 300250
    },
    {
      "epoch": 0.154291422445128,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001399482,
      "loss": 3.1636,
      "step": 300260
    },
    {
      "epoch": 0.1543100656812862,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001399462,
      "loss": 3.3121,
      "step": 300270
    },
    {
      "epoch": 0.15432870891744438,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013994420000000001,
      "loss": 3.2034,
      "step": 300280
    },
    {
      "epoch": 0.15434735215360257,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001399422,
      "loss": 3.1053,
      "step": 300290
    },
    {
      "epoch": 0.15436599538976073,
      "grad_norm": 0.75,
      "learning_rate": 0.0001399402,
      "loss": 3.0531,
      "step": 300300
    },
    {
      "epoch": 0.15438463862591892,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001399382,
      "loss": 3.1463,
      "step": 300310
    },
    {
      "epoch": 0.1544032818620771,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001399362,
      "loss": 3.1566,
      "step": 300320
    },
    {
      "epoch": 0.1544219250982353,
      "grad_norm": 0.75,
      "learning_rate": 0.00013993420000000002,
      "loss": 3.0663,
      "step": 300330
    },
    {
      "epoch": 0.15444056833439349,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013993220000000002,
      "loss": 3.0159,
      "step": 300340
    },
    {
      "epoch": 0.15445921157055165,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001399302,
      "loss": 3.1976,
      "step": 300350
    },
    {
      "epoch": 0.15447785480670984,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001399282,
      "loss": 3.116,
      "step": 300360
    },
    {
      "epoch": 0.15449649804286802,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001399262,
      "loss": 3.2963,
      "step": 300370
    },
    {
      "epoch": 0.1545151412790262,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013992420000000002,
      "loss": 3.0567,
      "step": 300380
    },
    {
      "epoch": 0.1545337845151844,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001399222,
      "loss": 3.1006,
      "step": 300390
    },
    {
      "epoch": 0.15455242775134256,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013992020000000002,
      "loss": 3.1204,
      "step": 300400
    },
    {
      "epoch": 0.15457107098750075,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001399182,
      "loss": 3.2087,
      "step": 300410
    },
    {
      "epoch": 0.15458971422365894,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001399162,
      "loss": 3.095,
      "step": 300420
    },
    {
      "epoch": 0.15460835745981713,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013991420000000003,
      "loss": 3.1657,
      "step": 300430
    },
    {
      "epoch": 0.15462700069597532,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001399122,
      "loss": 3.1348,
      "step": 300440
    },
    {
      "epoch": 0.15464564393213348,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013991020000000002,
      "loss": 3.1304,
      "step": 300450
    },
    {
      "epoch": 0.15466428716829167,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001399082,
      "loss": 3.1326,
      "step": 300460
    },
    {
      "epoch": 0.15468293040444986,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001399062,
      "loss": 3.2485,
      "step": 300470
    },
    {
      "epoch": 0.15470157364060805,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001399042,
      "loss": 3.1559,
      "step": 300480
    },
    {
      "epoch": 0.15472021687676624,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001399022,
      "loss": 3.2031,
      "step": 300490
    },
    {
      "epoch": 0.15473886011292443,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013990020000000003,
      "loss": 3.0602,
      "step": 300500
    },
    {
      "epoch": 0.1547575033490826,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001398982,
      "loss": 3.169,
      "step": 300510
    },
    {
      "epoch": 0.15477614658524078,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013989620000000002,
      "loss": 3.1298,
      "step": 300520
    },
    {
      "epoch": 0.15479478982139896,
      "grad_norm": 0.859375,
      "learning_rate": 0.00013989419999999999,
      "loss": 3.1695,
      "step": 300530
    },
    {
      "epoch": 0.15481343305755715,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001398922,
      "loss": 3.1722,
      "step": 300540
    },
    {
      "epoch": 0.15483207629371534,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001398902,
      "loss": 3.176,
      "step": 300550
    },
    {
      "epoch": 0.1548507195298735,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001398882,
      "loss": 3.0881,
      "step": 300560
    },
    {
      "epoch": 0.1548693627660317,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013988620000000002,
      "loss": 3.1656,
      "step": 300570
    },
    {
      "epoch": 0.15488800600218988,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001398842,
      "loss": 3.0892,
      "step": 300580
    },
    {
      "epoch": 0.15490664923834807,
      "grad_norm": 0.75,
      "learning_rate": 0.00013988220000000001,
      "loss": 3.0476,
      "step": 300590
    },
    {
      "epoch": 0.15492529247450626,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001398802,
      "loss": 3.1465,
      "step": 300600
    },
    {
      "epoch": 0.15494393571066442,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001398782,
      "loss": 3.1567,
      "step": 300610
    },
    {
      "epoch": 0.1549625789468226,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001398762,
      "loss": 3.1541,
      "step": 300620
    },
    {
      "epoch": 0.1549812221829808,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001398742,
      "loss": 3.1025,
      "step": 300630
    },
    {
      "epoch": 0.154999865419139,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013987220000000002,
      "loss": 3.2352,
      "step": 300640
    },
    {
      "epoch": 0.15501850865529718,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001398702,
      "loss": 3.1661,
      "step": 300650
    },
    {
      "epoch": 0.15503715189145534,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001398682,
      "loss": 3.1235,
      "step": 300660
    },
    {
      "epoch": 0.15505579512761353,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001398662,
      "loss": 3.2143,
      "step": 300670
    },
    {
      "epoch": 0.15507443836377172,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001398642,
      "loss": 3.1565,
      "step": 300680
    },
    {
      "epoch": 0.1550930815999299,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001398622,
      "loss": 3.1445,
      "step": 300690
    },
    {
      "epoch": 0.1551117248360881,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001398602,
      "loss": 3.2588,
      "step": 300700
    },
    {
      "epoch": 0.15513036807224626,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013985820000000002,
      "loss": 3.0847,
      "step": 300710
    },
    {
      "epoch": 0.15514901130840444,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001398562,
      "loss": 3.1019,
      "step": 300720
    },
    {
      "epoch": 0.15516765454456263,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001398542,
      "loss": 3.073,
      "step": 300730
    },
    {
      "epoch": 0.15518629778072082,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001398522,
      "loss": 3.1915,
      "step": 300740
    },
    {
      "epoch": 0.155204941016879,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001398502,
      "loss": 3.0486,
      "step": 300750
    },
    {
      "epoch": 0.1552235842530372,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001398482,
      "loss": 3.1405,
      "step": 300760
    },
    {
      "epoch": 0.15524222748919536,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013984620000000002,
      "loss": 3.1456,
      "step": 300770
    },
    {
      "epoch": 0.15526087072535355,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001398442,
      "loss": 3.0315,
      "step": 300780
    },
    {
      "epoch": 0.15527951396151174,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001398422,
      "loss": 3.1236,
      "step": 300790
    },
    {
      "epoch": 0.15529815719766993,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001398402,
      "loss": 3.0375,
      "step": 300800
    },
    {
      "epoch": 0.15531680043382812,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001398382,
      "loss": 3.1281,
      "step": 300810
    },
    {
      "epoch": 0.15533544366998628,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013983620000000002,
      "loss": 3.2524,
      "step": 300820
    },
    {
      "epoch": 0.15535408690614447,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013983420000000002,
      "loss": 3.0842,
      "step": 300830
    },
    {
      "epoch": 0.15537273014230266,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001398322,
      "loss": 3.178,
      "step": 300840
    },
    {
      "epoch": 0.15539137337846085,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001398302,
      "loss": 3.1141,
      "step": 300850
    },
    {
      "epoch": 0.15541001661461903,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001398282,
      "loss": 3.0767,
      "step": 300860
    },
    {
      "epoch": 0.1554286598507772,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001398262,
      "loss": 2.8338,
      "step": 300870
    },
    {
      "epoch": 0.15544730308693538,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001398242,
      "loss": 3.1508,
      "step": 300880
    },
    {
      "epoch": 0.15546594632309357,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013982220000000002,
      "loss": 2.9995,
      "step": 300890
    },
    {
      "epoch": 0.15548458955925176,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013982020000000001,
      "loss": 3.0736,
      "step": 300900
    },
    {
      "epoch": 0.15550323279540995,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001398182,
      "loss": 3.3015,
      "step": 300910
    },
    {
      "epoch": 0.1555218760315681,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001398162,
      "loss": 3.1071,
      "step": 300920
    },
    {
      "epoch": 0.1555405192677263,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001398142,
      "loss": 3.1951,
      "step": 300930
    },
    {
      "epoch": 0.1555591625038845,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013981220000000002,
      "loss": 3.1433,
      "step": 300940
    },
    {
      "epoch": 0.15557780574004268,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001398102,
      "loss": 2.9977,
      "step": 300950
    },
    {
      "epoch": 0.15559644897620087,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013980820000000001,
      "loss": 3.1616,
      "step": 300960
    },
    {
      "epoch": 0.15561509221235906,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001398062,
      "loss": 3.1681,
      "step": 300970
    },
    {
      "epoch": 0.15563373544851722,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001398042,
      "loss": 3.1501,
      "step": 300980
    },
    {
      "epoch": 0.1556523786846754,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013980220000000003,
      "loss": 3.0796,
      "step": 300990
    },
    {
      "epoch": 0.1556710219208336,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001398002,
      "loss": 3.1965,
      "step": 301000
    },
    {
      "epoch": 0.15568966515699179,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013979820000000002,
      "loss": 3.0335,
      "step": 301010
    },
    {
      "epoch": 0.15570830839314997,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001397962,
      "loss": 3.003,
      "step": 301020
    },
    {
      "epoch": 0.15572695162930814,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001397942,
      "loss": 3.1586,
      "step": 301030
    },
    {
      "epoch": 0.15574559486546632,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001397922,
      "loss": 3.1203,
      "step": 301040
    },
    {
      "epoch": 0.1557642381016245,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001397902,
      "loss": 3.0187,
      "step": 301050
    },
    {
      "epoch": 0.1557828813377827,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013978820000000002,
      "loss": 3.1635,
      "step": 301060
    },
    {
      "epoch": 0.1558015245739409,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001397862,
      "loss": 3.1093,
      "step": 301070
    },
    {
      "epoch": 0.15582016781009905,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013978420000000002,
      "loss": 3.0899,
      "step": 301080
    },
    {
      "epoch": 0.15583881104625724,
      "grad_norm": 0.75,
      "learning_rate": 0.00013978219999999998,
      "loss": 3.1948,
      "step": 301090
    },
    {
      "epoch": 0.15585745428241543,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001397802,
      "loss": 3.1691,
      "step": 301100
    },
    {
      "epoch": 0.15587609751857362,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001397782,
      "loss": 3.2566,
      "step": 301110
    },
    {
      "epoch": 0.1558947407547318,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001397762,
      "loss": 3.0888,
      "step": 301120
    },
    {
      "epoch": 0.15591338399088997,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013977420000000002,
      "loss": 3.0712,
      "step": 301130
    },
    {
      "epoch": 0.15593202722704816,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001397722,
      "loss": 3.2301,
      "step": 301140
    },
    {
      "epoch": 0.15595067046320635,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001397702,
      "loss": 3.0915,
      "step": 301150
    },
    {
      "epoch": 0.15596931369936454,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001397682,
      "loss": 3.0761,
      "step": 301160
    },
    {
      "epoch": 0.15598795693552273,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001397662,
      "loss": 3.1844,
      "step": 301170
    },
    {
      "epoch": 0.15600660017168091,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001397642,
      "loss": 3.1347,
      "step": 301180
    },
    {
      "epoch": 0.15602524340783908,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001397622,
      "loss": 3.0473,
      "step": 301190
    },
    {
      "epoch": 0.15604388664399726,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013976020000000002,
      "loss": 3.2786,
      "step": 301200
    },
    {
      "epoch": 0.15606252988015545,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001397582,
      "loss": 3.0804,
      "step": 301210
    },
    {
      "epoch": 0.15608117311631364,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001397562,
      "loss": 3.0995,
      "step": 301220
    },
    {
      "epoch": 0.15609981635247183,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001397542,
      "loss": 3.1268,
      "step": 301230
    },
    {
      "epoch": 0.15611845958863,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001397522,
      "loss": 3.0445,
      "step": 301240
    },
    {
      "epoch": 0.15613710282478818,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001397502,
      "loss": 3.1794,
      "step": 301250
    },
    {
      "epoch": 0.15615574606094637,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001397482,
      "loss": 3.0282,
      "step": 301260
    },
    {
      "epoch": 0.15617438929710456,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013974620000000001,
      "loss": 3.0858,
      "step": 301270
    },
    {
      "epoch": 0.15619303253326275,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001397442,
      "loss": 2.9535,
      "step": 301280
    },
    {
      "epoch": 0.1562116757694209,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001397422,
      "loss": 3.1216,
      "step": 301290
    },
    {
      "epoch": 0.1562303190055791,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001397402,
      "loss": 3.0863,
      "step": 301300
    },
    {
      "epoch": 0.1562489622417373,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001397382,
      "loss": 3.0656,
      "step": 301310
    },
    {
      "epoch": 0.15626760547789548,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013973620000000002,
      "loss": 3.0919,
      "step": 301320
    },
    {
      "epoch": 0.15628624871405367,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013973420000000001,
      "loss": 2.963,
      "step": 301330
    },
    {
      "epoch": 0.15630489195021183,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001397322,
      "loss": 3.1814,
      "step": 301340
    },
    {
      "epoch": 0.15632353518637002,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001397302,
      "loss": 3.2818,
      "step": 301350
    },
    {
      "epoch": 0.1563421784225282,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001397282,
      "loss": 3.0872,
      "step": 301360
    },
    {
      "epoch": 0.1563608216586864,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001397262,
      "loss": 3.1898,
      "step": 301370
    },
    {
      "epoch": 0.15637946489484458,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013972420000000002,
      "loss": 3.1264,
      "step": 301380
    },
    {
      "epoch": 0.15639810813100274,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013972220000000002,
      "loss": 3.0251,
      "step": 301390
    },
    {
      "epoch": 0.15641675136716093,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001397202,
      "loss": 3.0525,
      "step": 301400
    },
    {
      "epoch": 0.15643539460331912,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001397182,
      "loss": 3.2338,
      "step": 301410
    },
    {
      "epoch": 0.1564540378394773,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001397162,
      "loss": 3.2369,
      "step": 301420
    },
    {
      "epoch": 0.1564726810756355,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013971420000000003,
      "loss": 3.2312,
      "step": 301430
    },
    {
      "epoch": 0.1564913243117937,
      "grad_norm": 0.75,
      "learning_rate": 0.0001397122,
      "loss": 3.0845,
      "step": 301440
    },
    {
      "epoch": 0.15650996754795185,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013971020000000002,
      "loss": 2.9718,
      "step": 301450
    },
    {
      "epoch": 0.15652861078411004,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001397082,
      "loss": 3.139,
      "step": 301460
    },
    {
      "epoch": 0.15654725402026823,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001397062,
      "loss": 3.212,
      "step": 301470
    },
    {
      "epoch": 0.15656589725642642,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001397042,
      "loss": 3.1541,
      "step": 301480
    },
    {
      "epoch": 0.1565845404925846,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001397022,
      "loss": 3.1241,
      "step": 301490
    },
    {
      "epoch": 0.15660318372874277,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013970020000000002,
      "loss": 3.2355,
      "step": 301500
    },
    {
      "epoch": 0.15662182696490096,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001396982,
      "loss": 3.1324,
      "step": 301510
    },
    {
      "epoch": 0.15664047020105915,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001396962,
      "loss": 3.2807,
      "step": 301520
    },
    {
      "epoch": 0.15665911343721733,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001396942,
      "loss": 3.1366,
      "step": 301530
    },
    {
      "epoch": 0.15667775667337552,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001396922,
      "loss": 3.0804,
      "step": 301540
    },
    {
      "epoch": 0.15669639990953368,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013969020000000003,
      "loss": 3.1358,
      "step": 301550
    },
    {
      "epoch": 0.15671504314569187,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001396882,
      "loss": 3.0115,
      "step": 301560
    },
    {
      "epoch": 0.15673368638185006,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013968620000000002,
      "loss": 3.1236,
      "step": 301570
    },
    {
      "epoch": 0.15675232961800825,
      "grad_norm": 0.625,
      "learning_rate": 0.0001396842,
      "loss": 3.029,
      "step": 301580
    },
    {
      "epoch": 0.15677097285416644,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001396822,
      "loss": 3.0611,
      "step": 301590
    },
    {
      "epoch": 0.1567896160903246,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001396802,
      "loss": 3.0897,
      "step": 301600
    },
    {
      "epoch": 0.1568082593264828,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001396782,
      "loss": 3.1955,
      "step": 301610
    },
    {
      "epoch": 0.15682690256264098,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013967620000000002,
      "loss": 3.2107,
      "step": 301620
    },
    {
      "epoch": 0.15684554579879917,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001396742,
      "loss": 3.1081,
      "step": 301630
    },
    {
      "epoch": 0.15686418903495736,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013967220000000001,
      "loss": 3.0325,
      "step": 301640
    },
    {
      "epoch": 0.15688283227111555,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001396702,
      "loss": 3.1571,
      "step": 301650
    },
    {
      "epoch": 0.1569014755072737,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001396682,
      "loss": 3.0684,
      "step": 301660
    },
    {
      "epoch": 0.1569201187434319,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001396662,
      "loss": 3.0913,
      "step": 301670
    },
    {
      "epoch": 0.15693876197959009,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001396642,
      "loss": 3.1435,
      "step": 301680
    },
    {
      "epoch": 0.15695740521574827,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013966220000000002,
      "loss": 3.1017,
      "step": 301690
    },
    {
      "epoch": 0.15697604845190646,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001396602,
      "loss": 3.1633,
      "step": 301700
    },
    {
      "epoch": 0.15699469168806462,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001396582,
      "loss": 3.1421,
      "step": 301710
    },
    {
      "epoch": 0.1570133349242228,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001396562,
      "loss": 3.1808,
      "step": 301720
    },
    {
      "epoch": 0.157031978160381,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001396542,
      "loss": 3.0716,
      "step": 301730
    },
    {
      "epoch": 0.1570506213965392,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001396522,
      "loss": 3.0631,
      "step": 301740
    },
    {
      "epoch": 0.15706926463269738,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001396502,
      "loss": 3.042,
      "step": 301750
    },
    {
      "epoch": 0.15708790786885554,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013964820000000002,
      "loss": 3.1395,
      "step": 301760
    },
    {
      "epoch": 0.15710655110501373,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001396462,
      "loss": 2.9992,
      "step": 301770
    },
    {
      "epoch": 0.15712519434117192,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001396442,
      "loss": 3.0737,
      "step": 301780
    },
    {
      "epoch": 0.1571438375773301,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001396422,
      "loss": 3.0916,
      "step": 301790
    },
    {
      "epoch": 0.1571624808134883,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001396402,
      "loss": 3.1459,
      "step": 301800
    },
    {
      "epoch": 0.15718112404964646,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013963820000000002,
      "loss": 3.1537,
      "step": 301810
    },
    {
      "epoch": 0.15719976728580465,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013963620000000002,
      "loss": 3.0883,
      "step": 301820
    },
    {
      "epoch": 0.15721841052196284,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001396342,
      "loss": 3.0628,
      "step": 301830
    },
    {
      "epoch": 0.15723705375812103,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001396322,
      "loss": 3.1142,
      "step": 301840
    },
    {
      "epoch": 0.15725569699427921,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001396302,
      "loss": 3.3196,
      "step": 301850
    },
    {
      "epoch": 0.1572743402304374,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001396282,
      "loss": 3.0572,
      "step": 301860
    },
    {
      "epoch": 0.15729298346659557,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001396262,
      "loss": 2.9752,
      "step": 301870
    },
    {
      "epoch": 0.15731162670275375,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013962420000000002,
      "loss": 3.0612,
      "step": 301880
    },
    {
      "epoch": 0.15733026993891194,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001396222,
      "loss": 3.1727,
      "step": 301890
    },
    {
      "epoch": 0.15734891317507013,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001396202,
      "loss": 3.2539,
      "step": 301900
    },
    {
      "epoch": 0.15736755641122832,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001396182,
      "loss": 3.0337,
      "step": 301910
    },
    {
      "epoch": 0.15738619964738648,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001396162,
      "loss": 3.0174,
      "step": 301920
    },
    {
      "epoch": 0.15740484288354467,
      "grad_norm": 0.875,
      "learning_rate": 0.0001396142,
      "loss": 3.1979,
      "step": 301930
    },
    {
      "epoch": 0.15742348611970286,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013961220000000002,
      "loss": 3.1401,
      "step": 301940
    },
    {
      "epoch": 0.15744212935586105,
      "grad_norm": 0.75,
      "learning_rate": 0.00013961020000000001,
      "loss": 3.2929,
      "step": 301950
    },
    {
      "epoch": 0.15746077259201924,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001396082,
      "loss": 3.0983,
      "step": 301960
    },
    {
      "epoch": 0.1574794158281774,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001396062,
      "loss": 3.1355,
      "step": 301970
    },
    {
      "epoch": 0.1574980590643356,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001396042,
      "loss": 3.1634,
      "step": 301980
    },
    {
      "epoch": 0.15751670230049378,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013960220000000002,
      "loss": 3.2171,
      "step": 301990
    },
    {
      "epoch": 0.15753534553665197,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001396002,
      "loss": 3.2394,
      "step": 302000
    },
    {
      "epoch": 0.15755398877281016,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013959820000000001,
      "loss": 2.9488,
      "step": 302010
    },
    {
      "epoch": 0.15757263200896832,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001395962,
      "loss": 3.0868,
      "step": 302020
    },
    {
      "epoch": 0.1575912752451265,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001395942,
      "loss": 3.1288,
      "step": 302030
    },
    {
      "epoch": 0.1576099184812847,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013959220000000003,
      "loss": 3.0579,
      "step": 302040
    },
    {
      "epoch": 0.15762856171744288,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001395902,
      "loss": 3.078,
      "step": 302050
    },
    {
      "epoch": 0.15764720495360107,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013958820000000002,
      "loss": 3.0401,
      "step": 302060
    },
    {
      "epoch": 0.15766584818975923,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001395862,
      "loss": 3.2225,
      "step": 302070
    },
    {
      "epoch": 0.15768449142591742,
      "grad_norm": 0.75,
      "learning_rate": 0.0001395842,
      "loss": 3.0089,
      "step": 302080
    },
    {
      "epoch": 0.1577031346620756,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001395822,
      "loss": 3.2096,
      "step": 302090
    },
    {
      "epoch": 0.1577217778982338,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001395802,
      "loss": 3.1821,
      "step": 302100
    },
    {
      "epoch": 0.157740421134392,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013957820000000003,
      "loss": 3.1276,
      "step": 302110
    },
    {
      "epoch": 0.15775906437055018,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001395762,
      "loss": 3.0897,
      "step": 302120
    },
    {
      "epoch": 0.15777770760670834,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013957420000000002,
      "loss": 3.1999,
      "step": 302130
    },
    {
      "epoch": 0.15779635084286653,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013957219999999998,
      "loss": 3.0031,
      "step": 302140
    },
    {
      "epoch": 0.15781499407902472,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001395702,
      "loss": 2.9966,
      "step": 302150
    },
    {
      "epoch": 0.1578336373151829,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001395682,
      "loss": 3.1433,
      "step": 302160
    },
    {
      "epoch": 0.1578522805513411,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001395662,
      "loss": 3.1556,
      "step": 302170
    },
    {
      "epoch": 0.15787092378749926,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013956420000000002,
      "loss": 3.1435,
      "step": 302180
    },
    {
      "epoch": 0.15788956702365745,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001395622,
      "loss": 3.1594,
      "step": 302190
    },
    {
      "epoch": 0.15790821025981563,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001395602,
      "loss": 3.1033,
      "step": 302200
    },
    {
      "epoch": 0.15792685349597382,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001395582,
      "loss": 3.2428,
      "step": 302210
    },
    {
      "epoch": 0.157945496732132,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001395562,
      "loss": 3.068,
      "step": 302220
    },
    {
      "epoch": 0.15796413996829017,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001395542,
      "loss": 3.1442,
      "step": 302230
    },
    {
      "epoch": 0.15798278320444836,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001395522,
      "loss": 3.054,
      "step": 302240
    },
    {
      "epoch": 0.15800142644060655,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013955020000000002,
      "loss": 3.0423,
      "step": 302250
    },
    {
      "epoch": 0.15802006967676474,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013954819999999999,
      "loss": 3.1718,
      "step": 302260
    },
    {
      "epoch": 0.15803871291292293,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001395462,
      "loss": 3.047,
      "step": 302270
    },
    {
      "epoch": 0.1580573561490811,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001395442,
      "loss": 3.086,
      "step": 302280
    },
    {
      "epoch": 0.15807599938523928,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001395422,
      "loss": 3.1832,
      "step": 302290
    },
    {
      "epoch": 0.15809464262139747,
      "grad_norm": 0.75,
      "learning_rate": 0.00013954020000000002,
      "loss": 3.1192,
      "step": 302300
    },
    {
      "epoch": 0.15811328585755566,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001395382,
      "loss": 3.1333,
      "step": 302310
    },
    {
      "epoch": 0.15813192909371385,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013953620000000001,
      "loss": 3.1101,
      "step": 302320
    },
    {
      "epoch": 0.15815057232987204,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001395342,
      "loss": 3.1658,
      "step": 302330
    },
    {
      "epoch": 0.1581692155660302,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001395322,
      "loss": 3.0387,
      "step": 302340
    },
    {
      "epoch": 0.15818785880218839,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001395302,
      "loss": 3.0719,
      "step": 302350
    },
    {
      "epoch": 0.15820650203834657,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001395282,
      "loss": 3.0366,
      "step": 302360
    },
    {
      "epoch": 0.15822514527450476,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013952620000000002,
      "loss": 3.0322,
      "step": 302370
    },
    {
      "epoch": 0.15824378851066295,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013952420000000002,
      "loss": 3.149,
      "step": 302380
    },
    {
      "epoch": 0.15826243174682111,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001395222,
      "loss": 3.1817,
      "step": 302390
    },
    {
      "epoch": 0.1582810749829793,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001395202,
      "loss": 3.1038,
      "step": 302400
    },
    {
      "epoch": 0.1582997182191375,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001395182,
      "loss": 3.152,
      "step": 302410
    },
    {
      "epoch": 0.15831836145529568,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001395162,
      "loss": 3.0753,
      "step": 302420
    },
    {
      "epoch": 0.15833700469145387,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013951420000000002,
      "loss": 3.046,
      "step": 302430
    },
    {
      "epoch": 0.15835564792761203,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013951220000000002,
      "loss": 3.0522,
      "step": 302440
    },
    {
      "epoch": 0.15837429116377022,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001395102,
      "loss": 3.1253,
      "step": 302450
    },
    {
      "epoch": 0.1583929343999284,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001395082,
      "loss": 3.0492,
      "step": 302460
    },
    {
      "epoch": 0.1584115776360866,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001395062,
      "loss": 3.1025,
      "step": 302470
    },
    {
      "epoch": 0.1584302208722448,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001395042,
      "loss": 3.0922,
      "step": 302480
    },
    {
      "epoch": 0.15844886410840295,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001395022,
      "loss": 3.0985,
      "step": 302490
    },
    {
      "epoch": 0.15846750734456114,
      "grad_norm": 0.859375,
      "learning_rate": 0.00013950020000000002,
      "loss": 3.1927,
      "step": 302500
    },
    {
      "epoch": 0.15848615058071933,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001394982,
      "loss": 3.0694,
      "step": 302510
    },
    {
      "epoch": 0.15850479381687752,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001394962,
      "loss": 2.995,
      "step": 302520
    },
    {
      "epoch": 0.1585234370530357,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001394942,
      "loss": 3.2224,
      "step": 302530
    },
    {
      "epoch": 0.1585420802891939,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001394922,
      "loss": 3.1229,
      "step": 302540
    },
    {
      "epoch": 0.15856072352535205,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013949020000000002,
      "loss": 3.0388,
      "step": 302550
    },
    {
      "epoch": 0.15857936676151024,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001394882,
      "loss": 2.947,
      "step": 302560
    },
    {
      "epoch": 0.15859800999766843,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001394862,
      "loss": 3.193,
      "step": 302570
    },
    {
      "epoch": 0.15861665323382662,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001394842,
      "loss": 3.1858,
      "step": 302580
    },
    {
      "epoch": 0.1586352964699848,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001394822,
      "loss": 3.0822,
      "step": 302590
    },
    {
      "epoch": 0.15865393970614297,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013948020000000003,
      "loss": 3.0498,
      "step": 302600
    },
    {
      "epoch": 0.15867258294230116,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001394782,
      "loss": 3.1669,
      "step": 302610
    },
    {
      "epoch": 0.15869122617845935,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013947620000000002,
      "loss": 3.2095,
      "step": 302620
    },
    {
      "epoch": 0.15870986941461754,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001394742,
      "loss": 3.1225,
      "step": 302630
    },
    {
      "epoch": 0.15872851265077573,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001394722,
      "loss": 3.0722,
      "step": 302640
    },
    {
      "epoch": 0.1587471558869339,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001394702,
      "loss": 3.1505,
      "step": 302650
    },
    {
      "epoch": 0.15876579912309208,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001394682,
      "loss": 3.2036,
      "step": 302660
    },
    {
      "epoch": 0.15878444235925027,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00013946620000000002,
      "loss": 3.0717,
      "step": 302670
    },
    {
      "epoch": 0.15880308559540846,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001394642,
      "loss": 3.2274,
      "step": 302680
    },
    {
      "epoch": 0.15882172883156664,
      "grad_norm": 0.875,
      "learning_rate": 0.00013946220000000001,
      "loss": 3.123,
      "step": 302690
    },
    {
      "epoch": 0.1588403720677248,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013946019999999998,
      "loss": 3.142,
      "step": 302700
    },
    {
      "epoch": 0.158859015303883,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001394582,
      "loss": 3.1526,
      "step": 302710
    },
    {
      "epoch": 0.15887765854004118,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001394562,
      "loss": 3.0971,
      "step": 302720
    },
    {
      "epoch": 0.15889630177619937,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001394542,
      "loss": 3.1094,
      "step": 302730
    },
    {
      "epoch": 0.15891494501235756,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00013945220000000002,
      "loss": 3.0823,
      "step": 302740
    },
    {
      "epoch": 0.15893358824851572,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001394502,
      "loss": 3.1993,
      "step": 302750
    },
    {
      "epoch": 0.1589522314846739,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001394482,
      "loss": 3.0671,
      "step": 302760
    },
    {
      "epoch": 0.1589708747208321,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001394462,
      "loss": 3.2234,
      "step": 302770
    },
    {
      "epoch": 0.1589895179569903,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001394442,
      "loss": 3.0904,
      "step": 302780
    },
    {
      "epoch": 0.15900816119314848,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013944220000000003,
      "loss": 3.1245,
      "step": 302790
    },
    {
      "epoch": 0.15902680442930667,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001394402,
      "loss": 3.1075,
      "step": 302800
    },
    {
      "epoch": 0.15904544766546483,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013943820000000002,
      "loss": 3.0097,
      "step": 302810
    },
    {
      "epoch": 0.15906409090162302,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001394362,
      "loss": 3.0591,
      "step": 302820
    },
    {
      "epoch": 0.1590827341377812,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001394342,
      "loss": 3.1056,
      "step": 302830
    },
    {
      "epoch": 0.1591013773739394,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001394322,
      "loss": 3.1139,
      "step": 302840
    },
    {
      "epoch": 0.15912002061009758,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001394302,
      "loss": 2.9757,
      "step": 302850
    },
    {
      "epoch": 0.15913866384625575,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013942820000000002,
      "loss": 3.0883,
      "step": 302860
    },
    {
      "epoch": 0.15915730708241393,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001394262,
      "loss": 3.1923,
      "step": 302870
    },
    {
      "epoch": 0.15917595031857212,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001394242,
      "loss": 3.178,
      "step": 302880
    },
    {
      "epoch": 0.1591945935547303,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001394222,
      "loss": 3.0449,
      "step": 302890
    },
    {
      "epoch": 0.1592132367908885,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001394202,
      "loss": 3.0769,
      "step": 302900
    },
    {
      "epoch": 0.15923188002704666,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001394182,
      "loss": 3.0756,
      "step": 302910
    },
    {
      "epoch": 0.15925052326320485,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001394162,
      "loss": 3.0531,
      "step": 302920
    },
    {
      "epoch": 0.15926916649936304,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013941420000000002,
      "loss": 3.2313,
      "step": 302930
    },
    {
      "epoch": 0.15928780973552123,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001394122,
      "loss": 3.1606,
      "step": 302940
    },
    {
      "epoch": 0.15930645297167942,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001394102,
      "loss": 3.0562,
      "step": 302950
    },
    {
      "epoch": 0.15932509620783758,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001394082,
      "loss": 3.1856,
      "step": 302960
    },
    {
      "epoch": 0.15934373944399577,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001394062,
      "loss": 2.9973,
      "step": 302970
    },
    {
      "epoch": 0.15936238268015396,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001394042,
      "loss": 3.1776,
      "step": 302980
    },
    {
      "epoch": 0.15938102591631215,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013940220000000002,
      "loss": 2.9328,
      "step": 302990
    },
    {
      "epoch": 0.15939966915247034,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013940020000000001,
      "loss": 3.2519,
      "step": 303000
    },
    {
      "epoch": 0.15941831238862852,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001393982,
      "loss": 3.2502,
      "step": 303010
    },
    {
      "epoch": 0.15943695562478669,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001393962,
      "loss": 3.2008,
      "step": 303020
    },
    {
      "epoch": 0.15945559886094487,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001393942,
      "loss": 3.0928,
      "step": 303030
    },
    {
      "epoch": 0.15947424209710306,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013939220000000002,
      "loss": 3.1428,
      "step": 303040
    },
    {
      "epoch": 0.15949288533326125,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001393902,
      "loss": 3.1454,
      "step": 303050
    },
    {
      "epoch": 0.15951152856941944,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013938820000000002,
      "loss": 3.1416,
      "step": 303060
    },
    {
      "epoch": 0.1595301718055776,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001393862,
      "loss": 3.1122,
      "step": 303070
    },
    {
      "epoch": 0.1595488150417358,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001393842,
      "loss": 3.1743,
      "step": 303080
    },
    {
      "epoch": 0.15956745827789398,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001393822,
      "loss": 3.0997,
      "step": 303090
    },
    {
      "epoch": 0.15958610151405217,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001393802,
      "loss": 3.1399,
      "step": 303100
    },
    {
      "epoch": 0.15960474475021036,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013937820000000002,
      "loss": 3.1868,
      "step": 303110
    },
    {
      "epoch": 0.15962338798636852,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001393762,
      "loss": 3.1004,
      "step": 303120
    },
    {
      "epoch": 0.1596420312225267,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001393742,
      "loss": 3.1384,
      "step": 303130
    },
    {
      "epoch": 0.1596606744586849,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001393722,
      "loss": 3.1546,
      "step": 303140
    },
    {
      "epoch": 0.1596793176948431,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001393702,
      "loss": 3.0438,
      "step": 303150
    },
    {
      "epoch": 0.15969796093100128,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013936820000000003,
      "loss": 3.0719,
      "step": 303160
    },
    {
      "epoch": 0.15971660416715944,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001393662,
      "loss": 3.0371,
      "step": 303170
    },
    {
      "epoch": 0.15973524740331763,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013936420000000002,
      "loss": 3.1508,
      "step": 303180
    },
    {
      "epoch": 0.15975389063947582,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013936219999999999,
      "loss": 3.1138,
      "step": 303190
    },
    {
      "epoch": 0.159772533875634,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001393602,
      "loss": 3.0353,
      "step": 303200
    },
    {
      "epoch": 0.1597911771117922,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001393582,
      "loss": 3.1685,
      "step": 303210
    },
    {
      "epoch": 0.15980982034795038,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001393562,
      "loss": 3.1741,
      "step": 303220
    },
    {
      "epoch": 0.15982846358410854,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013935420000000002,
      "loss": 3.1338,
      "step": 303230
    },
    {
      "epoch": 0.15984710682026673,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001393522,
      "loss": 3.1447,
      "step": 303240
    },
    {
      "epoch": 0.15986575005642492,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001393502,
      "loss": 3.2066,
      "step": 303250
    },
    {
      "epoch": 0.1598843932925831,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001393482,
      "loss": 3.1164,
      "step": 303260
    },
    {
      "epoch": 0.1599030365287413,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001393462,
      "loss": 2.9672,
      "step": 303270
    },
    {
      "epoch": 0.15992167976489946,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013934420000000003,
      "loss": 3.0192,
      "step": 303280
    },
    {
      "epoch": 0.15994032300105765,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001393422,
      "loss": 3.0393,
      "step": 303290
    },
    {
      "epoch": 0.15995896623721584,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013934020000000002,
      "loss": 2.9709,
      "step": 303300
    },
    {
      "epoch": 0.15997760947337403,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001393382,
      "loss": 3.0731,
      "step": 303310
    },
    {
      "epoch": 0.15999625270953222,
      "grad_norm": 0.75,
      "learning_rate": 0.0001393362,
      "loss": 3.1224,
      "step": 303320
    },
    {
      "epoch": 0.16001489594569038,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001393342,
      "loss": 3.0918,
      "step": 303330
    },
    {
      "epoch": 0.16003353918184857,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001393322,
      "loss": 3.1134,
      "step": 303340
    },
    {
      "epoch": 0.16005218241800676,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013933020000000002,
      "loss": 3.1584,
      "step": 303350
    },
    {
      "epoch": 0.16007082565416494,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001393282,
      "loss": 3.0631,
      "step": 303360
    },
    {
      "epoch": 0.16008946889032313,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013932620000000001,
      "loss": 3.1098,
      "step": 303370
    },
    {
      "epoch": 0.1601081121264813,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001393242,
      "loss": 3.2461,
      "step": 303380
    },
    {
      "epoch": 0.16012675536263948,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001393222,
      "loss": 3.1957,
      "step": 303390
    },
    {
      "epoch": 0.16014539859879767,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001393202,
      "loss": 3.0561,
      "step": 303400
    },
    {
      "epoch": 0.16016404183495586,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001393182,
      "loss": 3.1022,
      "step": 303410
    },
    {
      "epoch": 0.16018268507111405,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013931620000000002,
      "loss": 3.1358,
      "step": 303420
    },
    {
      "epoch": 0.1602013283072722,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013931420000000002,
      "loss": 3.1612,
      "step": 303430
    },
    {
      "epoch": 0.1602199715434304,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001393122,
      "loss": 3.1677,
      "step": 303440
    },
    {
      "epoch": 0.1602386147795886,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001393102,
      "loss": 3.1758,
      "step": 303450
    },
    {
      "epoch": 0.16025725801574678,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001393082,
      "loss": 3.19,
      "step": 303460
    },
    {
      "epoch": 0.16027590125190497,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001393062,
      "loss": 3.123,
      "step": 303470
    },
    {
      "epoch": 0.16029454448806316,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001393042,
      "loss": 3.1254,
      "step": 303480
    },
    {
      "epoch": 0.16031318772422132,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013930220000000002,
      "loss": 3.0565,
      "step": 303490
    },
    {
      "epoch": 0.1603318309603795,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001393002,
      "loss": 3.0622,
      "step": 303500
    },
    {
      "epoch": 0.1603504741965377,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001392982,
      "loss": 3.0795,
      "step": 303510
    },
    {
      "epoch": 0.16036911743269588,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001392962,
      "loss": 3.1529,
      "step": 303520
    },
    {
      "epoch": 0.16038776066885407,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001392942,
      "loss": 3.2638,
      "step": 303530
    },
    {
      "epoch": 0.16040640390501223,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001392922,
      "loss": 3.0832,
      "step": 303540
    },
    {
      "epoch": 0.16042504714117042,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013929020000000002,
      "loss": 3.1242,
      "step": 303550
    },
    {
      "epoch": 0.1604436903773286,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001392882,
      "loss": 3.2014,
      "step": 303560
    },
    {
      "epoch": 0.1604623336134868,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001392862,
      "loss": 3.2946,
      "step": 303570
    },
    {
      "epoch": 0.160480976849645,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001392842,
      "loss": 3.1784,
      "step": 303580
    },
    {
      "epoch": 0.16049962008580315,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001392822,
      "loss": 3.1365,
      "step": 303590
    },
    {
      "epoch": 0.16051826332196134,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013928020000000002,
      "loss": 3.2103,
      "step": 303600
    },
    {
      "epoch": 0.16053690655811953,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001392782,
      "loss": 2.9647,
      "step": 303610
    },
    {
      "epoch": 0.16055554979427772,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001392762,
      "loss": 3.2163,
      "step": 303620
    },
    {
      "epoch": 0.1605741930304359,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001392742,
      "loss": 3.2097,
      "step": 303630
    },
    {
      "epoch": 0.16059283626659407,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001392722,
      "loss": 3.1463,
      "step": 303640
    },
    {
      "epoch": 0.16061147950275226,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013927020000000003,
      "loss": 3.1916,
      "step": 303650
    },
    {
      "epoch": 0.16063012273891045,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001392682,
      "loss": 3.0979,
      "step": 303660
    },
    {
      "epoch": 0.16064876597506864,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013926620000000002,
      "loss": 3.1533,
      "step": 303670
    },
    {
      "epoch": 0.16066740921122682,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001392642,
      "loss": 3.0962,
      "step": 303680
    },
    {
      "epoch": 0.160686052447385,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001392622,
      "loss": 3.1629,
      "step": 303690
    },
    {
      "epoch": 0.16070469568354318,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001392602,
      "loss": 3.0515,
      "step": 303700
    },
    {
      "epoch": 0.16072333891970136,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001392582,
      "loss": 3.1375,
      "step": 303710
    },
    {
      "epoch": 0.16074198215585955,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013925620000000002,
      "loss": 3.0935,
      "step": 303720
    },
    {
      "epoch": 0.16076062539201774,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001392542,
      "loss": 3.1683,
      "step": 303730
    },
    {
      "epoch": 0.16077926862817593,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013925220000000001,
      "loss": 3.2444,
      "step": 303740
    },
    {
      "epoch": 0.1607979118643341,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001392502,
      "loss": 3.0791,
      "step": 303750
    },
    {
      "epoch": 0.16081655510049228,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001392482,
      "loss": 3.1743,
      "step": 303760
    },
    {
      "epoch": 0.16083519833665047,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013924620000000003,
      "loss": 3.2611,
      "step": 303770
    },
    {
      "epoch": 0.16085384157280866,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001392442,
      "loss": 3.1228,
      "step": 303780
    },
    {
      "epoch": 0.16087248480896685,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013924220000000002,
      "loss": 3.1264,
      "step": 303790
    },
    {
      "epoch": 0.160891128045125,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001392402,
      "loss": 3.0681,
      "step": 303800
    },
    {
      "epoch": 0.1609097712812832,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001392382,
      "loss": 3.1239,
      "step": 303810
    },
    {
      "epoch": 0.1609284145174414,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001392362,
      "loss": 3.184,
      "step": 303820
    },
    {
      "epoch": 0.16094705775359958,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001392342,
      "loss": 3.1508,
      "step": 303830
    },
    {
      "epoch": 0.16096570098975777,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013923220000000003,
      "loss": 3.1591,
      "step": 303840
    },
    {
      "epoch": 0.16098434422591593,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001392302,
      "loss": 3.0196,
      "step": 303850
    },
    {
      "epoch": 0.16100298746207412,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013922820000000002,
      "loss": 3.0736,
      "step": 303860
    },
    {
      "epoch": 0.1610216306982323,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001392262,
      "loss": 3.1753,
      "step": 303870
    },
    {
      "epoch": 0.1610402739343905,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001392242,
      "loss": 3.119,
      "step": 303880
    },
    {
      "epoch": 0.16105891717054868,
      "grad_norm": 0.75,
      "learning_rate": 0.0001392222,
      "loss": 3.184,
      "step": 303890
    },
    {
      "epoch": 0.16107756040670687,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001392202,
      "loss": 3.1714,
      "step": 303900
    },
    {
      "epoch": 0.16109620364286503,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013921820000000002,
      "loss": 3.116,
      "step": 303910
    },
    {
      "epoch": 0.16111484687902322,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001392162,
      "loss": 3.2533,
      "step": 303920
    },
    {
      "epoch": 0.1611334901151814,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001392142,
      "loss": 3.1219,
      "step": 303930
    },
    {
      "epoch": 0.1611521333513396,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001392122,
      "loss": 3.0316,
      "step": 303940
    },
    {
      "epoch": 0.1611707765874978,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001392102,
      "loss": 3.1649,
      "step": 303950
    },
    {
      "epoch": 0.16118941982365595,
      "grad_norm": 0.75,
      "learning_rate": 0.0001392082,
      "loss": 2.9693,
      "step": 303960
    },
    {
      "epoch": 0.16120806305981414,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001392062,
      "loss": 3.068,
      "step": 303970
    },
    {
      "epoch": 0.16122670629597233,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013920420000000002,
      "loss": 3.1642,
      "step": 303980
    },
    {
      "epoch": 0.16124534953213052,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013920220000000001,
      "loss": 3.042,
      "step": 303990
    },
    {
      "epoch": 0.1612639927682887,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001392002,
      "loss": 3.1817,
      "step": 304000
    },
    {
      "epoch": 0.16128263600444687,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001391982,
      "loss": 3.111,
      "step": 304010
    },
    {
      "epoch": 0.16130127924060506,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001391962,
      "loss": 2.9707,
      "step": 304020
    },
    {
      "epoch": 0.16131992247676324,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001391942,
      "loss": 3.1066,
      "step": 304030
    },
    {
      "epoch": 0.16133856571292143,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013919220000000002,
      "loss": 3.1296,
      "step": 304040
    },
    {
      "epoch": 0.16135720894907962,
      "grad_norm": 0.75,
      "learning_rate": 0.00013919020000000001,
      "loss": 3.2631,
      "step": 304050
    },
    {
      "epoch": 0.16137585218523778,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001391882,
      "loss": 3.2091,
      "step": 304060
    },
    {
      "epoch": 0.16139449542139597,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001391862,
      "loss": 3.1191,
      "step": 304070
    },
    {
      "epoch": 0.16141313865755416,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001391842,
      "loss": 3.1316,
      "step": 304080
    },
    {
      "epoch": 0.16143178189371235,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001391822,
      "loss": 2.9951,
      "step": 304090
    },
    {
      "epoch": 0.16145042512987054,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001391802,
      "loss": 3.0935,
      "step": 304100
    },
    {
      "epoch": 0.1614690683660287,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013917820000000002,
      "loss": 3.1247,
      "step": 304110
    },
    {
      "epoch": 0.1614877116021869,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001391762,
      "loss": 3.1534,
      "step": 304120
    },
    {
      "epoch": 0.16150635483834508,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001391742,
      "loss": 3.0772,
      "step": 304130
    },
    {
      "epoch": 0.16152499807450327,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001391722,
      "loss": 3.1902,
      "step": 304140
    },
    {
      "epoch": 0.16154364131066146,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001391702,
      "loss": 3.0517,
      "step": 304150
    },
    {
      "epoch": 0.16156228454681965,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013916820000000002,
      "loss": 3.3021,
      "step": 304160
    },
    {
      "epoch": 0.1615809277829778,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001391662,
      "loss": 3.0715,
      "step": 304170
    },
    {
      "epoch": 0.161599571019136,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001391642,
      "loss": 3.0628,
      "step": 304180
    },
    {
      "epoch": 0.16161821425529418,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001391622,
      "loss": 3.2478,
      "step": 304190
    },
    {
      "epoch": 0.16163685749145237,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001391602,
      "loss": 3.122,
      "step": 304200
    },
    {
      "epoch": 0.16165550072761056,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013915820000000003,
      "loss": 3.2689,
      "step": 304210
    },
    {
      "epoch": 0.16167414396376872,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001391562,
      "loss": 3.191,
      "step": 304220
    },
    {
      "epoch": 0.1616927871999269,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013915420000000002,
      "loss": 3.1051,
      "step": 304230
    },
    {
      "epoch": 0.1617114304360851,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001391522,
      "loss": 3.117,
      "step": 304240
    },
    {
      "epoch": 0.1617300736722433,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001391502,
      "loss": 3.0413,
      "step": 304250
    },
    {
      "epoch": 0.16174871690840148,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013914820000000003,
      "loss": 3.2234,
      "step": 304260
    },
    {
      "epoch": 0.16176736014455964,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001391462,
      "loss": 3.1087,
      "step": 304270
    },
    {
      "epoch": 0.16178600338071783,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013914420000000002,
      "loss": 3.2126,
      "step": 304280
    },
    {
      "epoch": 0.16180464661687602,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001391422,
      "loss": 3.2856,
      "step": 304290
    },
    {
      "epoch": 0.1618232898530342,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001391402,
      "loss": 3.223,
      "step": 304300
    },
    {
      "epoch": 0.1618419330891924,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001391382,
      "loss": 3.0972,
      "step": 304310
    },
    {
      "epoch": 0.16186057632535056,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001391362,
      "loss": 3.2089,
      "step": 304320
    },
    {
      "epoch": 0.16187921956150875,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013913420000000003,
      "loss": 3.1097,
      "step": 304330
    },
    {
      "epoch": 0.16189786279766694,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001391322,
      "loss": 3.0428,
      "step": 304340
    },
    {
      "epoch": 0.16191650603382512,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013913020000000002,
      "loss": 3.1297,
      "step": 304350
    },
    {
      "epoch": 0.16193514926998331,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001391282,
      "loss": 3.0712,
      "step": 304360
    },
    {
      "epoch": 0.1619537925061415,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001391262,
      "loss": 3.1389,
      "step": 304370
    },
    {
      "epoch": 0.16197243574229966,
      "grad_norm": 0.75,
      "learning_rate": 0.0001391242,
      "loss": 3.04,
      "step": 304380
    },
    {
      "epoch": 0.16199107897845785,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001391222,
      "loss": 3.0796,
      "step": 304390
    },
    {
      "epoch": 0.16200972221461604,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013912020000000002,
      "loss": 3.0199,
      "step": 304400
    },
    {
      "epoch": 0.16202836545077423,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001391182,
      "loss": 3.1229,
      "step": 304410
    },
    {
      "epoch": 0.16204700868693242,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013911620000000001,
      "loss": 3.1203,
      "step": 304420
    },
    {
      "epoch": 0.16206565192309058,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001391142,
      "loss": 3.129,
      "step": 304430
    },
    {
      "epoch": 0.16208429515924877,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001391122,
      "loss": 3.0858,
      "step": 304440
    },
    {
      "epoch": 0.16210293839540696,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001391102,
      "loss": 3.1452,
      "step": 304450
    },
    {
      "epoch": 0.16212158163156515,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001391082,
      "loss": 3.1612,
      "step": 304460
    },
    {
      "epoch": 0.16214022486772334,
      "grad_norm": 0.75,
      "learning_rate": 0.00013910620000000002,
      "loss": 3.0269,
      "step": 304470
    },
    {
      "epoch": 0.1621588681038815,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001391042,
      "loss": 3.1896,
      "step": 304480
    },
    {
      "epoch": 0.1621775113400397,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001391022,
      "loss": 3.1404,
      "step": 304490
    },
    {
      "epoch": 0.16219615457619788,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001391002,
      "loss": 3.0998,
      "step": 304500
    },
    {
      "epoch": 0.16221479781235607,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001390982,
      "loss": 3.0687,
      "step": 304510
    },
    {
      "epoch": 0.16223344104851425,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001390962,
      "loss": 3.1132,
      "step": 304520
    },
    {
      "epoch": 0.16225208428467242,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001390942,
      "loss": 3.0853,
      "step": 304530
    },
    {
      "epoch": 0.1622707275208306,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013909220000000002,
      "loss": 3.0342,
      "step": 304540
    },
    {
      "epoch": 0.1622893707569888,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001390902,
      "loss": 3.1919,
      "step": 304550
    },
    {
      "epoch": 0.16230801399314698,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001390882,
      "loss": 3.2384,
      "step": 304560
    },
    {
      "epoch": 0.16232665722930517,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001390862,
      "loss": 3.1688,
      "step": 304570
    },
    {
      "epoch": 0.16234530046546336,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001390842,
      "loss": 3.0521,
      "step": 304580
    },
    {
      "epoch": 0.16236394370162152,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001390822,
      "loss": 3.1788,
      "step": 304590
    },
    {
      "epoch": 0.1623825869377797,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013908020000000002,
      "loss": 3.0474,
      "step": 304600
    },
    {
      "epoch": 0.1624012301739379,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001390782,
      "loss": 3.189,
      "step": 304610
    },
    {
      "epoch": 0.1624198734100961,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001390762,
      "loss": 3.1705,
      "step": 304620
    },
    {
      "epoch": 0.16243851664625428,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001390742,
      "loss": 3.0819,
      "step": 304630
    },
    {
      "epoch": 0.16245715988241244,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001390722,
      "loss": 3.1002,
      "step": 304640
    },
    {
      "epoch": 0.16247580311857063,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013907020000000002,
      "loss": 3.1162,
      "step": 304650
    },
    {
      "epoch": 0.16249444635472882,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001390682,
      "loss": 3.0672,
      "step": 304660
    },
    {
      "epoch": 0.162513089590887,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013906620000000001,
      "loss": 3.0421,
      "step": 304670
    },
    {
      "epoch": 0.1625317328270452,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001390642,
      "loss": 3.081,
      "step": 304680
    },
    {
      "epoch": 0.16255037606320336,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001390622,
      "loss": 3.2638,
      "step": 304690
    },
    {
      "epoch": 0.16256901929936154,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001390602,
      "loss": 3.0872,
      "step": 304700
    },
    {
      "epoch": 0.16258766253551973,
      "grad_norm": 0.75,
      "learning_rate": 0.0001390582,
      "loss": 3.0763,
      "step": 304710
    },
    {
      "epoch": 0.16260630577167792,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013905620000000002,
      "loss": 3.1143,
      "step": 304720
    },
    {
      "epoch": 0.1626249490078361,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013905420000000001,
      "loss": 3.039,
      "step": 304730
    },
    {
      "epoch": 0.16264359224399427,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001390522,
      "loss": 3.1852,
      "step": 304740
    },
    {
      "epoch": 0.16266223548015246,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001390502,
      "loss": 3.1931,
      "step": 304750
    },
    {
      "epoch": 0.16268087871631065,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001390482,
      "loss": 3.1025,
      "step": 304760
    },
    {
      "epoch": 0.16269952195246884,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013904620000000002,
      "loss": 3.2067,
      "step": 304770
    },
    {
      "epoch": 0.16271816518862703,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001390442,
      "loss": 3.1659,
      "step": 304780
    },
    {
      "epoch": 0.16273680842478522,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013904220000000002,
      "loss": 3.0689,
      "step": 304790
    },
    {
      "epoch": 0.16275545166094338,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001390402,
      "loss": 3.047,
      "step": 304800
    },
    {
      "epoch": 0.16277409489710157,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001390382,
      "loss": 3.2117,
      "step": 304810
    },
    {
      "epoch": 0.16279273813325976,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013903620000000003,
      "loss": 3.3215,
      "step": 304820
    },
    {
      "epoch": 0.16281138136941795,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001390342,
      "loss": 3.1031,
      "step": 304830
    },
    {
      "epoch": 0.16283002460557613,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013903220000000002,
      "loss": 3.0978,
      "step": 304840
    },
    {
      "epoch": 0.1628486678417343,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001390302,
      "loss": 3.0711,
      "step": 304850
    },
    {
      "epoch": 0.16286731107789248,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001390282,
      "loss": 3.0744,
      "step": 304860
    },
    {
      "epoch": 0.16288595431405067,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001390262,
      "loss": 2.9993,
      "step": 304870
    },
    {
      "epoch": 0.16290459755020886,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001390242,
      "loss": 3.1235,
      "step": 304880
    },
    {
      "epoch": 0.16292324078636705,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013902220000000003,
      "loss": 3.2101,
      "step": 304890
    },
    {
      "epoch": 0.1629418840225252,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001390202,
      "loss": 3.1129,
      "step": 304900
    },
    {
      "epoch": 0.1629605272586834,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013901820000000002,
      "loss": 3.038,
      "step": 304910
    },
    {
      "epoch": 0.1629791704948416,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013901619999999999,
      "loss": 3.0624,
      "step": 304920
    },
    {
      "epoch": 0.16299781373099978,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001390142,
      "loss": 3.1376,
      "step": 304930
    },
    {
      "epoch": 0.16301645696715797,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001390122,
      "loss": 3.0637,
      "step": 304940
    },
    {
      "epoch": 0.16303510020331613,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001390102,
      "loss": 3.1674,
      "step": 304950
    },
    {
      "epoch": 0.16305374343947432,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013900820000000002,
      "loss": 3.1254,
      "step": 304960
    },
    {
      "epoch": 0.1630723866756325,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001390062,
      "loss": 3.2268,
      "step": 304970
    },
    {
      "epoch": 0.1630910299117907,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001390042,
      "loss": 2.9461,
      "step": 304980
    },
    {
      "epoch": 0.16310967314794889,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001390022,
      "loss": 3.1859,
      "step": 304990
    },
    {
      "epoch": 0.16312831638410705,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001390002,
      "loss": 3.158,
      "step": 305000
    },
    {
      "epoch": 0.16314695962026524,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001389982,
      "loss": 3.144,
      "step": 305010
    },
    {
      "epoch": 0.16316560285642343,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001389962,
      "loss": 3.0271,
      "step": 305020
    },
    {
      "epoch": 0.16318424609258161,
      "grad_norm": 0.75,
      "learning_rate": 0.00013899420000000002,
      "loss": 3.1629,
      "step": 305030
    },
    {
      "epoch": 0.1632028893287398,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013899220000000001,
      "loss": 3.0243,
      "step": 305040
    },
    {
      "epoch": 0.163221532564898,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001389902,
      "loss": 3.1226,
      "step": 305050
    },
    {
      "epoch": 0.16324017580105615,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001389882,
      "loss": 3.1585,
      "step": 305060
    },
    {
      "epoch": 0.16325881903721434,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001389862,
      "loss": 2.9065,
      "step": 305070
    },
    {
      "epoch": 0.16327746227337253,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001389842,
      "loss": 3.004,
      "step": 305080
    },
    {
      "epoch": 0.16329610550953072,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001389822,
      "loss": 3.2156,
      "step": 305090
    },
    {
      "epoch": 0.1633147487456889,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013898020000000001,
      "loss": 3.1324,
      "step": 305100
    },
    {
      "epoch": 0.16333339198184707,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001389782,
      "loss": 3.0385,
      "step": 305110
    },
    {
      "epoch": 0.16335203521800526,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001389762,
      "loss": 3.1419,
      "step": 305120
    },
    {
      "epoch": 0.16337067845416345,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001389742,
      "loss": 3.1691,
      "step": 305130
    },
    {
      "epoch": 0.16338932169032164,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001389722,
      "loss": 3.2015,
      "step": 305140
    },
    {
      "epoch": 0.16340796492647983,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001389702,
      "loss": 3.2808,
      "step": 305150
    },
    {
      "epoch": 0.163426608162638,
      "grad_norm": 0.84375,
      "learning_rate": 0.00013896820000000002,
      "loss": 3.0712,
      "step": 305160
    },
    {
      "epoch": 0.16344525139879618,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001389662,
      "loss": 3.1425,
      "step": 305170
    },
    {
      "epoch": 0.16346389463495437,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001389642,
      "loss": 3.1234,
      "step": 305180
    },
    {
      "epoch": 0.16348253787111255,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001389622,
      "loss": 3.1325,
      "step": 305190
    },
    {
      "epoch": 0.16350118110727074,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001389602,
      "loss": 3.1412,
      "step": 305200
    },
    {
      "epoch": 0.1635198243434289,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013895820000000002,
      "loss": 3.2081,
      "step": 305210
    },
    {
      "epoch": 0.1635384675795871,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013895620000000002,
      "loss": 3.24,
      "step": 305220
    },
    {
      "epoch": 0.16355711081574528,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001389542,
      "loss": 3.0261,
      "step": 305230
    },
    {
      "epoch": 0.16357575405190347,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001389522,
      "loss": 3.2534,
      "step": 305240
    },
    {
      "epoch": 0.16359439728806166,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001389502,
      "loss": 3.0149,
      "step": 305250
    },
    {
      "epoch": 0.16361304052421985,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013894820000000003,
      "loss": 3.1033,
      "step": 305260
    },
    {
      "epoch": 0.163631683760378,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001389462,
      "loss": 3.1896,
      "step": 305270
    },
    {
      "epoch": 0.1636503269965362,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013894420000000002,
      "loss": 3.1089,
      "step": 305280
    },
    {
      "epoch": 0.1636689702326944,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001389422,
      "loss": 3.0646,
      "step": 305290
    },
    {
      "epoch": 0.16368761346885258,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001389402,
      "loss": 3.0595,
      "step": 305300
    },
    {
      "epoch": 0.16370625670501077,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001389382,
      "loss": 3.1383,
      "step": 305310
    },
    {
      "epoch": 0.16372489994116893,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001389362,
      "loss": 3.0325,
      "step": 305320
    },
    {
      "epoch": 0.16374354317732712,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013893420000000002,
      "loss": 3.0959,
      "step": 305330
    },
    {
      "epoch": 0.1637621864134853,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001389322,
      "loss": 3.1705,
      "step": 305340
    },
    {
      "epoch": 0.1637808296496435,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013893020000000001,
      "loss": 3.0319,
      "step": 305350
    },
    {
      "epoch": 0.16379947288580168,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001389282,
      "loss": 3.1362,
      "step": 305360
    },
    {
      "epoch": 0.16381811612195984,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001389262,
      "loss": 3.0917,
      "step": 305370
    },
    {
      "epoch": 0.16383675935811803,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013892420000000003,
      "loss": 3.0628,
      "step": 305380
    },
    {
      "epoch": 0.16385540259427622,
      "grad_norm": 0.75,
      "learning_rate": 0.0001389222,
      "loss": 3.2306,
      "step": 305390
    },
    {
      "epoch": 0.1638740458304344,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013892020000000002,
      "loss": 2.9943,
      "step": 305400
    },
    {
      "epoch": 0.1638926890665926,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001389182,
      "loss": 3.1455,
      "step": 305410
    },
    {
      "epoch": 0.16391133230275076,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001389162,
      "loss": 3.1914,
      "step": 305420
    },
    {
      "epoch": 0.16392997553890895,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001389142,
      "loss": 3.0363,
      "step": 305430
    },
    {
      "epoch": 0.16394861877506714,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001389122,
      "loss": 3.2317,
      "step": 305440
    },
    {
      "epoch": 0.16396726201122533,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013891020000000002,
      "loss": 3.1863,
      "step": 305450
    },
    {
      "epoch": 0.16398590524738352,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001389082,
      "loss": 3.0022,
      "step": 305460
    },
    {
      "epoch": 0.1640045484835417,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013890620000000002,
      "loss": 3.1705,
      "step": 305470
    },
    {
      "epoch": 0.16402319171969987,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001389042,
      "loss": 3.1279,
      "step": 305480
    },
    {
      "epoch": 0.16404183495585806,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001389022,
      "loss": 3.1101,
      "step": 305490
    },
    {
      "epoch": 0.16406047819201625,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001389002,
      "loss": 3.0247,
      "step": 305500
    },
    {
      "epoch": 0.16407912142817443,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001388982,
      "loss": 3.0486,
      "step": 305510
    },
    {
      "epoch": 0.16409776466433262,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013889620000000002,
      "loss": 3.1309,
      "step": 305520
    },
    {
      "epoch": 0.16411640790049078,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001388942,
      "loss": 3.1389,
      "step": 305530
    },
    {
      "epoch": 0.16413505113664897,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001388922,
      "loss": 3.0897,
      "step": 305540
    },
    {
      "epoch": 0.16415369437280716,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001388902,
      "loss": 3.0919,
      "step": 305550
    },
    {
      "epoch": 0.16417233760896535,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001388882,
      "loss": 3.1128,
      "step": 305560
    },
    {
      "epoch": 0.16419098084512354,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001388862,
      "loss": 3.1331,
      "step": 305570
    },
    {
      "epoch": 0.1642096240812817,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001388842,
      "loss": 3.1798,
      "step": 305580
    },
    {
      "epoch": 0.1642282673174399,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013888220000000002,
      "loss": 3.2132,
      "step": 305590
    },
    {
      "epoch": 0.16424691055359808,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001388802,
      "loss": 3.0743,
      "step": 305600
    },
    {
      "epoch": 0.16426555378975627,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001388782,
      "loss": 3.1274,
      "step": 305610
    },
    {
      "epoch": 0.16428419702591446,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001388762,
      "loss": 3.1171,
      "step": 305620
    },
    {
      "epoch": 0.16430284026207262,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001388742,
      "loss": 3.2063,
      "step": 305630
    },
    {
      "epoch": 0.1643214834982308,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001388722,
      "loss": 3.1466,
      "step": 305640
    },
    {
      "epoch": 0.164340126734389,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013887020000000002,
      "loss": 2.9868,
      "step": 305650
    },
    {
      "epoch": 0.16435876997054719,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001388682,
      "loss": 3.0613,
      "step": 305660
    },
    {
      "epoch": 0.16437741320670538,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001388662,
      "loss": 3.2804,
      "step": 305670
    },
    {
      "epoch": 0.16439605644286354,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001388642,
      "loss": 3.2021,
      "step": 305680
    },
    {
      "epoch": 0.16441469967902173,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001388622,
      "loss": 2.9715,
      "step": 305690
    },
    {
      "epoch": 0.16443334291517991,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001388602,
      "loss": 3.039,
      "step": 305700
    },
    {
      "epoch": 0.1644519861513381,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013885820000000002,
      "loss": 3.2748,
      "step": 305710
    },
    {
      "epoch": 0.1644706293874963,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013885620000000001,
      "loss": 3.0403,
      "step": 305720
    },
    {
      "epoch": 0.16448927262365448,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001388542,
      "loss": 3.1279,
      "step": 305730
    },
    {
      "epoch": 0.16450791585981264,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001388522,
      "loss": 3.0015,
      "step": 305740
    },
    {
      "epoch": 0.16452655909597083,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001388502,
      "loss": 3.1606,
      "step": 305750
    },
    {
      "epoch": 0.16454520233212902,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001388482,
      "loss": 3.0954,
      "step": 305760
    },
    {
      "epoch": 0.1645638455682872,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013884620000000002,
      "loss": 3.0105,
      "step": 305770
    },
    {
      "epoch": 0.1645824888044454,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013884420000000001,
      "loss": 3.1222,
      "step": 305780
    },
    {
      "epoch": 0.16460113204060356,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001388422,
      "loss": 3.0003,
      "step": 305790
    },
    {
      "epoch": 0.16461977527676175,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001388402,
      "loss": 2.9882,
      "step": 305800
    },
    {
      "epoch": 0.16463841851291994,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001388382,
      "loss": 3.1128,
      "step": 305810
    },
    {
      "epoch": 0.16465706174907813,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013883620000000002,
      "loss": 3.1565,
      "step": 305820
    },
    {
      "epoch": 0.16467570498523632,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001388342,
      "loss": 3.2339,
      "step": 305830
    },
    {
      "epoch": 0.16469434822139448,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013883220000000002,
      "loss": 3.0586,
      "step": 305840
    },
    {
      "epoch": 0.16471299145755267,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001388302,
      "loss": 3.1531,
      "step": 305850
    },
    {
      "epoch": 0.16473163469371085,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001388282,
      "loss": 3.1683,
      "step": 305860
    },
    {
      "epoch": 0.16475027792986904,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013882620000000003,
      "loss": 3.0232,
      "step": 305870
    },
    {
      "epoch": 0.16476892116602723,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001388242,
      "loss": 3.0429,
      "step": 305880
    },
    {
      "epoch": 0.1647875644021854,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013882220000000002,
      "loss": 3.1367,
      "step": 305890
    },
    {
      "epoch": 0.16480620763834358,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001388202,
      "loss": 3.1042,
      "step": 305900
    },
    {
      "epoch": 0.16482485087450177,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001388182,
      "loss": 3.1042,
      "step": 305910
    },
    {
      "epoch": 0.16484349411065996,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001388162,
      "loss": 3.0417,
      "step": 305920
    },
    {
      "epoch": 0.16486213734681815,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001388142,
      "loss": 3.0185,
      "step": 305930
    },
    {
      "epoch": 0.16488078058297634,
      "grad_norm": 0.890625,
      "learning_rate": 0.00013881220000000003,
      "loss": 3.0634,
      "step": 305940
    },
    {
      "epoch": 0.1648994238191345,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001388102,
      "loss": 3.1221,
      "step": 305950
    },
    {
      "epoch": 0.1649180670552927,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013880820000000002,
      "loss": 3.1528,
      "step": 305960
    },
    {
      "epoch": 0.16493671029145088,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013880619999999999,
      "loss": 3.0335,
      "step": 305970
    },
    {
      "epoch": 0.16495535352760907,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001388042,
      "loss": 3.1706,
      "step": 305980
    },
    {
      "epoch": 0.16497399676376726,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001388022,
      "loss": 3.16,
      "step": 305990
    },
    {
      "epoch": 0.16499263999992542,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001388002,
      "loss": 3.3001,
      "step": 306000
    },
    {
      "epoch": 0.1650112832360836,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013879820000000002,
      "loss": 3.2728,
      "step": 306010
    },
    {
      "epoch": 0.1650299264722418,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001387962,
      "loss": 3.1001,
      "step": 306020
    },
    {
      "epoch": 0.16504856970839998,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013879420000000001,
      "loss": 3.1747,
      "step": 306030
    },
    {
      "epoch": 0.16506721294455817,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001387922,
      "loss": 3.0158,
      "step": 306040
    },
    {
      "epoch": 0.16508585618071633,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001387902,
      "loss": 3.1258,
      "step": 306050
    },
    {
      "epoch": 0.16510449941687452,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001387882,
      "loss": 3.0902,
      "step": 306060
    },
    {
      "epoch": 0.1651231426530327,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001387862,
      "loss": 3.1784,
      "step": 306070
    },
    {
      "epoch": 0.1651417858891909,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013878420000000002,
      "loss": 3.0842,
      "step": 306080
    },
    {
      "epoch": 0.1651604291253491,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001387822,
      "loss": 3.0098,
      "step": 306090
    },
    {
      "epoch": 0.16517907236150725,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001387802,
      "loss": 3.1357,
      "step": 306100
    },
    {
      "epoch": 0.16519771559766544,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001387782,
      "loss": 3.2078,
      "step": 306110
    },
    {
      "epoch": 0.16521635883382363,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001387762,
      "loss": 3.1236,
      "step": 306120
    },
    {
      "epoch": 0.16523500206998182,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001387742,
      "loss": 3.1012,
      "step": 306130
    },
    {
      "epoch": 0.16525364530614,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001387722,
      "loss": 3.1622,
      "step": 306140
    },
    {
      "epoch": 0.1652722885422982,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013877020000000002,
      "loss": 3.188,
      "step": 306150
    },
    {
      "epoch": 0.16529093177845636,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001387682,
      "loss": 3.3089,
      "step": 306160
    },
    {
      "epoch": 0.16530957501461455,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001387662,
      "loss": 3.1141,
      "step": 306170
    },
    {
      "epoch": 0.16532821825077273,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001387642,
      "loss": 3.1532,
      "step": 306180
    },
    {
      "epoch": 0.16534686148693092,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001387622,
      "loss": 3.0604,
      "step": 306190
    },
    {
      "epoch": 0.1653655047230891,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013876020000000002,
      "loss": 3.0834,
      "step": 306200
    },
    {
      "epoch": 0.16538414795924727,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013875820000000002,
      "loss": 3.0879,
      "step": 306210
    },
    {
      "epoch": 0.16540279119540546,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001387562,
      "loss": 3.1249,
      "step": 306220
    },
    {
      "epoch": 0.16542143443156365,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001387542,
      "loss": 3.1915,
      "step": 306230
    },
    {
      "epoch": 0.16544007766772184,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001387522,
      "loss": 3.1852,
      "step": 306240
    },
    {
      "epoch": 0.16545872090388003,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001387502,
      "loss": 3.0067,
      "step": 306250
    },
    {
      "epoch": 0.1654773641400382,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013874820000000002,
      "loss": 3.1824,
      "step": 306260
    },
    {
      "epoch": 0.16549600737619638,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013874620000000002,
      "loss": 2.9658,
      "step": 306270
    },
    {
      "epoch": 0.16551465061235457,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001387442,
      "loss": 3.1639,
      "step": 306280
    },
    {
      "epoch": 0.16553329384851276,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001387422,
      "loss": 3.1007,
      "step": 306290
    },
    {
      "epoch": 0.16555193708467095,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001387402,
      "loss": 3.193,
      "step": 306300
    },
    {
      "epoch": 0.1655705803208291,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001387382,
      "loss": 3.0636,
      "step": 306310
    },
    {
      "epoch": 0.1655892235569873,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001387362,
      "loss": 3.1183,
      "step": 306320
    },
    {
      "epoch": 0.1656078667931455,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013873420000000002,
      "loss": 3.1038,
      "step": 306330
    },
    {
      "epoch": 0.16562651002930368,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001387322,
      "loss": 3.1358,
      "step": 306340
    },
    {
      "epoch": 0.16564515326546186,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001387302,
      "loss": 3.2079,
      "step": 306350
    },
    {
      "epoch": 0.16566379650162003,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001387282,
      "loss": 3.1199,
      "step": 306360
    },
    {
      "epoch": 0.16568243973777821,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001387262,
      "loss": 3.0947,
      "step": 306370
    },
    {
      "epoch": 0.1657010829739364,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013872420000000002,
      "loss": 3.17,
      "step": 306380
    },
    {
      "epoch": 0.1657197262100946,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001387222,
      "loss": 3.2106,
      "step": 306390
    },
    {
      "epoch": 0.16573836944625278,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013872020000000001,
      "loss": 3.1745,
      "step": 306400
    },
    {
      "epoch": 0.16575701268241097,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001387182,
      "loss": 3.0773,
      "step": 306410
    },
    {
      "epoch": 0.16577565591856913,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001387162,
      "loss": 3.2838,
      "step": 306420
    },
    {
      "epoch": 0.16579429915472732,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00013871420000000003,
      "loss": 3.1332,
      "step": 306430
    },
    {
      "epoch": 0.1658129423908855,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001387122,
      "loss": 3.1939,
      "step": 306440
    },
    {
      "epoch": 0.1658315856270437,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013871020000000002,
      "loss": 3.1775,
      "step": 306450
    },
    {
      "epoch": 0.1658502288632019,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001387082,
      "loss": 3.072,
      "step": 306460
    },
    {
      "epoch": 0.16586887209936005,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001387062,
      "loss": 3.1405,
      "step": 306470
    },
    {
      "epoch": 0.16588751533551824,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001387042,
      "loss": 3.1514,
      "step": 306480
    },
    {
      "epoch": 0.16590615857167643,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001387022,
      "loss": 3.1424,
      "step": 306490
    },
    {
      "epoch": 0.16592480180783462,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013870020000000002,
      "loss": 3.1555,
      "step": 306500
    },
    {
      "epoch": 0.1659434450439928,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001386982,
      "loss": 3.0354,
      "step": 306510
    },
    {
      "epoch": 0.16596208828015097,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013869620000000002,
      "loss": 3.1064,
      "step": 306520
    },
    {
      "epoch": 0.16598073151630915,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013869419999999998,
      "loss": 3.1907,
      "step": 306530
    },
    {
      "epoch": 0.16599937475246734,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001386922,
      "loss": 3.0655,
      "step": 306540
    },
    {
      "epoch": 0.16601801798862553,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001386902,
      "loss": 3.1886,
      "step": 306550
    },
    {
      "epoch": 0.16603666122478372,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001386882,
      "loss": 3.1282,
      "step": 306560
    },
    {
      "epoch": 0.16605530446094188,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013868620000000002,
      "loss": 3.0434,
      "step": 306570
    },
    {
      "epoch": 0.16607394769710007,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001386842,
      "loss": 3.183,
      "step": 306580
    },
    {
      "epoch": 0.16609259093325826,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001386822,
      "loss": 3.1835,
      "step": 306590
    },
    {
      "epoch": 0.16611123416941645,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001386802,
      "loss": 3.1923,
      "step": 306600
    },
    {
      "epoch": 0.16612987740557464,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001386782,
      "loss": 3.0748,
      "step": 306610
    },
    {
      "epoch": 0.16614852064173283,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001386762,
      "loss": 3.2232,
      "step": 306620
    },
    {
      "epoch": 0.166167163877891,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001386742,
      "loss": 3.0743,
      "step": 306630
    },
    {
      "epoch": 0.16618580711404918,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013867220000000002,
      "loss": 3.058,
      "step": 306640
    },
    {
      "epoch": 0.16620445035020737,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001386702,
      "loss": 3.1287,
      "step": 306650
    },
    {
      "epoch": 0.16622309358636556,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001386682,
      "loss": 3.1496,
      "step": 306660
    },
    {
      "epoch": 0.16624173682252374,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001386662,
      "loss": 3.1115,
      "step": 306670
    },
    {
      "epoch": 0.1662603800586819,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001386642,
      "loss": 3.1475,
      "step": 306680
    },
    {
      "epoch": 0.1662790232948401,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001386622,
      "loss": 3.2463,
      "step": 306690
    },
    {
      "epoch": 0.16629766653099828,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001386602,
      "loss": 3.0364,
      "step": 306700
    },
    {
      "epoch": 0.16631630976715647,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013865820000000001,
      "loss": 3.0825,
      "step": 306710
    },
    {
      "epoch": 0.16633495300331466,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001386562,
      "loss": 3.2628,
      "step": 306720
    },
    {
      "epoch": 0.16635359623947282,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001386542,
      "loss": 3.1561,
      "step": 306730
    },
    {
      "epoch": 0.166372239475631,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001386522,
      "loss": 3.2473,
      "step": 306740
    },
    {
      "epoch": 0.1663908827117892,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001386502,
      "loss": 2.9595,
      "step": 306750
    },
    {
      "epoch": 0.1664095259479474,
      "grad_norm": 0.75,
      "learning_rate": 0.00013864820000000002,
      "loss": 3.1626,
      "step": 306760
    },
    {
      "epoch": 0.16642816918410558,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013864620000000001,
      "loss": 3.1279,
      "step": 306770
    },
    {
      "epoch": 0.16644681242026374,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001386442,
      "loss": 2.9987,
      "step": 306780
    },
    {
      "epoch": 0.16646545565642193,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001386422,
      "loss": 3.225,
      "step": 306790
    },
    {
      "epoch": 0.16648409889258012,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001386402,
      "loss": 3.0447,
      "step": 306800
    },
    {
      "epoch": 0.1665027421287383,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001386382,
      "loss": 3.2538,
      "step": 306810
    },
    {
      "epoch": 0.1665213853648965,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013863620000000002,
      "loss": 3.23,
      "step": 306820
    },
    {
      "epoch": 0.16654002860105468,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013863420000000002,
      "loss": 3.1025,
      "step": 306830
    },
    {
      "epoch": 0.16655867183721285,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001386322,
      "loss": 3.1591,
      "step": 306840
    },
    {
      "epoch": 0.16657731507337104,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001386302,
      "loss": 3.067,
      "step": 306850
    },
    {
      "epoch": 0.16659595830952922,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001386282,
      "loss": 3.1574,
      "step": 306860
    },
    {
      "epoch": 0.1666146015456874,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013862620000000002,
      "loss": 3.1789,
      "step": 306870
    },
    {
      "epoch": 0.1666332447818456,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001386242,
      "loss": 3.1364,
      "step": 306880
    },
    {
      "epoch": 0.16665188801800376,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013862220000000002,
      "loss": 2.9495,
      "step": 306890
    },
    {
      "epoch": 0.16667053125416195,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001386202,
      "loss": 3.1334,
      "step": 306900
    },
    {
      "epoch": 0.16668917449032014,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001386182,
      "loss": 3.2161,
      "step": 306910
    },
    {
      "epoch": 0.16670781772647833,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001386162,
      "loss": 3.1364,
      "step": 306920
    },
    {
      "epoch": 0.16672646096263652,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001386142,
      "loss": 3.0677,
      "step": 306930
    },
    {
      "epoch": 0.16674510419879468,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013861220000000002,
      "loss": 3.0482,
      "step": 306940
    },
    {
      "epoch": 0.16676374743495287,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001386102,
      "loss": 3.1068,
      "step": 306950
    },
    {
      "epoch": 0.16678239067111106,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001386082,
      "loss": 3.2022,
      "step": 306960
    },
    {
      "epoch": 0.16680103390726925,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001386062,
      "loss": 3.1231,
      "step": 306970
    },
    {
      "epoch": 0.16681967714342744,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001386042,
      "loss": 3.1512,
      "step": 306980
    },
    {
      "epoch": 0.1668383203795856,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013860220000000003,
      "loss": 3.0593,
      "step": 306990
    },
    {
      "epoch": 0.1668569636157438,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001386002,
      "loss": 3.1353,
      "step": 307000
    },
    {
      "epoch": 0.16687560685190198,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013859820000000002,
      "loss": 3.0955,
      "step": 307010
    },
    {
      "epoch": 0.16689425008806016,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013859619999999999,
      "loss": 3.0612,
      "step": 307020
    },
    {
      "epoch": 0.16691289332421835,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001385942,
      "loss": 3.1894,
      "step": 307030
    },
    {
      "epoch": 0.16693153656037651,
      "grad_norm": 0.75,
      "learning_rate": 0.0001385922,
      "loss": 3.0434,
      "step": 307040
    },
    {
      "epoch": 0.1669501797965347,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001385902,
      "loss": 3.0855,
      "step": 307050
    },
    {
      "epoch": 0.1669688230326929,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013858820000000002,
      "loss": 3.1004,
      "step": 307060
    },
    {
      "epoch": 0.16698746626885108,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001385862,
      "loss": 3.0053,
      "step": 307070
    },
    {
      "epoch": 0.16700610950500927,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013858420000000001,
      "loss": 3.1865,
      "step": 307080
    },
    {
      "epoch": 0.16702475274116746,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001385822,
      "loss": 3.1101,
      "step": 307090
    },
    {
      "epoch": 0.16704339597732562,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001385802,
      "loss": 3.0232,
      "step": 307100
    },
    {
      "epoch": 0.1670620392134838,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001385782,
      "loss": 3.0893,
      "step": 307110
    },
    {
      "epoch": 0.167080682449642,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001385762,
      "loss": 3.1895,
      "step": 307120
    },
    {
      "epoch": 0.1670993256858002,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013857420000000002,
      "loss": 3.1652,
      "step": 307130
    },
    {
      "epoch": 0.16711796892195838,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001385722,
      "loss": 3.0769,
      "step": 307140
    },
    {
      "epoch": 0.16713661215811654,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001385702,
      "loss": 3.1269,
      "step": 307150
    },
    {
      "epoch": 0.16715525539427473,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001385682,
      "loss": 3.1718,
      "step": 307160
    },
    {
      "epoch": 0.16717389863043292,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001385662,
      "loss": 3.0908,
      "step": 307170
    },
    {
      "epoch": 0.1671925418665911,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001385642,
      "loss": 3.1375,
      "step": 307180
    },
    {
      "epoch": 0.1672111851027493,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001385622,
      "loss": 3.152,
      "step": 307190
    },
    {
      "epoch": 0.16722982833890745,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013856020000000002,
      "loss": 3.0901,
      "step": 307200
    },
    {
      "epoch": 0.16724847157506564,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001385582,
      "loss": 3.0588,
      "step": 307210
    },
    {
      "epoch": 0.16726711481122383,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001385562,
      "loss": 3.158,
      "step": 307220
    },
    {
      "epoch": 0.16728575804738202,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001385542,
      "loss": 3.0666,
      "step": 307230
    },
    {
      "epoch": 0.1673044012835402,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001385522,
      "loss": 3.0642,
      "step": 307240
    },
    {
      "epoch": 0.16732304451969837,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013855020000000002,
      "loss": 3.1185,
      "step": 307250
    },
    {
      "epoch": 0.16734168775585656,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013854820000000002,
      "loss": 3.1154,
      "step": 307260
    },
    {
      "epoch": 0.16736033099201475,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001385462,
      "loss": 3.1662,
      "step": 307270
    },
    {
      "epoch": 0.16737897422817294,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001385442,
      "loss": 3.1829,
      "step": 307280
    },
    {
      "epoch": 0.16739761746433113,
      "grad_norm": 1.25,
      "learning_rate": 0.0001385422,
      "loss": 3.1236,
      "step": 307290
    },
    {
      "epoch": 0.16741626070048932,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001385402,
      "loss": 3.0597,
      "step": 307300
    },
    {
      "epoch": 0.16743490393664748,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001385382,
      "loss": 3.0701,
      "step": 307310
    },
    {
      "epoch": 0.16745354717280567,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013853620000000002,
      "loss": 3.1426,
      "step": 307320
    },
    {
      "epoch": 0.16747219040896386,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001385342,
      "loss": 3.0965,
      "step": 307330
    },
    {
      "epoch": 0.16749083364512204,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001385322,
      "loss": 3.1342,
      "step": 307340
    },
    {
      "epoch": 0.16750947688128023,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001385302,
      "loss": 3.0827,
      "step": 307350
    },
    {
      "epoch": 0.1675281201174384,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001385282,
      "loss": 3.0323,
      "step": 307360
    },
    {
      "epoch": 0.16754676335359658,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001385262,
      "loss": 3.1923,
      "step": 307370
    },
    {
      "epoch": 0.16756540658975477,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013852420000000002,
      "loss": 3.0951,
      "step": 307380
    },
    {
      "epoch": 0.16758404982591296,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001385222,
      "loss": 3.0142,
      "step": 307390
    },
    {
      "epoch": 0.16760269306207115,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001385202,
      "loss": 3.0062,
      "step": 307400
    },
    {
      "epoch": 0.1676213362982293,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001385182,
      "loss": 3.1878,
      "step": 307410
    },
    {
      "epoch": 0.1676399795343875,
      "grad_norm": 0.75,
      "learning_rate": 0.0001385162,
      "loss": 3.0372,
      "step": 307420
    },
    {
      "epoch": 0.1676586227705457,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013851420000000002,
      "loss": 3.1756,
      "step": 307430
    },
    {
      "epoch": 0.16767726600670388,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001385122,
      "loss": 3.2462,
      "step": 307440
    },
    {
      "epoch": 0.16769590924286207,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013851020000000001,
      "loss": 3.1192,
      "step": 307450
    },
    {
      "epoch": 0.16771455247902023,
      "grad_norm": 0.625,
      "learning_rate": 0.0001385082,
      "loss": 3.2499,
      "step": 307460
    },
    {
      "epoch": 0.16773319571517842,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001385062,
      "loss": 3.0983,
      "step": 307470
    },
    {
      "epoch": 0.1677518389513366,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013850420000000003,
      "loss": 3.0871,
      "step": 307480
    },
    {
      "epoch": 0.1677704821874948,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001385022,
      "loss": 3.1315,
      "step": 307490
    },
    {
      "epoch": 0.16778912542365298,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013850020000000002,
      "loss": 3.1621,
      "step": 307500
    },
    {
      "epoch": 0.16780776865981117,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001384982,
      "loss": 3.2286,
      "step": 307510
    },
    {
      "epoch": 0.16782641189596934,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001384962,
      "loss": 3.1617,
      "step": 307520
    },
    {
      "epoch": 0.16784505513212752,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001384942,
      "loss": 3.1667,
      "step": 307530
    },
    {
      "epoch": 0.1678636983682857,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001384922,
      "loss": 3.1677,
      "step": 307540
    },
    {
      "epoch": 0.1678823416044439,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013849020000000002,
      "loss": 3.1943,
      "step": 307550
    },
    {
      "epoch": 0.1679009848406021,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001384882,
      "loss": 3.1208,
      "step": 307560
    },
    {
      "epoch": 0.16791962807676025,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013848620000000002,
      "loss": 3.1084,
      "step": 307570
    },
    {
      "epoch": 0.16793827131291844,
      "grad_norm": 0.75,
      "learning_rate": 0.00013848419999999998,
      "loss": 3.1106,
      "step": 307580
    },
    {
      "epoch": 0.16795691454907663,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001384822,
      "loss": 3.107,
      "step": 307590
    },
    {
      "epoch": 0.16797555778523482,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001384802,
      "loss": 3.1681,
      "step": 307600
    },
    {
      "epoch": 0.167994201021393,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001384782,
      "loss": 3.2012,
      "step": 307610
    },
    {
      "epoch": 0.16801284425755117,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013847620000000002,
      "loss": 2.9597,
      "step": 307620
    },
    {
      "epoch": 0.16803148749370936,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001384742,
      "loss": 3.1047,
      "step": 307630
    },
    {
      "epoch": 0.16805013072986755,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001384722,
      "loss": 3.0802,
      "step": 307640
    },
    {
      "epoch": 0.16806877396602574,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001384702,
      "loss": 3.1712,
      "step": 307650
    },
    {
      "epoch": 0.16808741720218393,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001384682,
      "loss": 3.0641,
      "step": 307660
    },
    {
      "epoch": 0.1681060604383421,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001384662,
      "loss": 3.1787,
      "step": 307670
    },
    {
      "epoch": 0.16812470367450028,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001384642,
      "loss": 3.1954,
      "step": 307680
    },
    {
      "epoch": 0.16814334691065846,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013846220000000002,
      "loss": 3.1466,
      "step": 307690
    },
    {
      "epoch": 0.16816199014681665,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001384602,
      "loss": 3.1941,
      "step": 307700
    },
    {
      "epoch": 0.16818063338297484,
      "grad_norm": 0.75,
      "learning_rate": 0.0001384582,
      "loss": 3.1405,
      "step": 307710
    },
    {
      "epoch": 0.168199276619133,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001384562,
      "loss": 3.2143,
      "step": 307720
    },
    {
      "epoch": 0.1682179198552912,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001384542,
      "loss": 3.0412,
      "step": 307730
    },
    {
      "epoch": 0.16823656309144938,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013845220000000002,
      "loss": 3.2131,
      "step": 307740
    },
    {
      "epoch": 0.16825520632760757,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001384502,
      "loss": 3.0886,
      "step": 307750
    },
    {
      "epoch": 0.16827384956376576,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013844820000000001,
      "loss": 2.9318,
      "step": 307760
    },
    {
      "epoch": 0.16829249279992395,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001384462,
      "loss": 3.1755,
      "step": 307770
    },
    {
      "epoch": 0.1683111360360821,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001384442,
      "loss": 3.0155,
      "step": 307780
    },
    {
      "epoch": 0.1683297792722403,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001384422,
      "loss": 3.1601,
      "step": 307790
    },
    {
      "epoch": 0.1683484225083985,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001384402,
      "loss": 3.0363,
      "step": 307800
    },
    {
      "epoch": 0.16836706574455668,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013843820000000002,
      "loss": 2.994,
      "step": 307810
    },
    {
      "epoch": 0.16838570898071487,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013843620000000001,
      "loss": 3.0226,
      "step": 307820
    },
    {
      "epoch": 0.16840435221687303,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001384342,
      "loss": 3.1145,
      "step": 307830
    },
    {
      "epoch": 0.16842299545303122,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001384322,
      "loss": 3.214,
      "step": 307840
    },
    {
      "epoch": 0.1684416386891894,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001384302,
      "loss": 3.0681,
      "step": 307850
    },
    {
      "epoch": 0.1684602819253476,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001384282,
      "loss": 3.0636,
      "step": 307860
    },
    {
      "epoch": 0.16847892516150578,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013842620000000002,
      "loss": 3.0928,
      "step": 307870
    },
    {
      "epoch": 0.16849756839766394,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013842420000000002,
      "loss": 3.2486,
      "step": 307880
    },
    {
      "epoch": 0.16851621163382213,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001384222,
      "loss": 3.089,
      "step": 307890
    },
    {
      "epoch": 0.16853485486998032,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001384202,
      "loss": 3.0923,
      "step": 307900
    },
    {
      "epoch": 0.1685534981061385,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001384182,
      "loss": 3.176,
      "step": 307910
    },
    {
      "epoch": 0.1685721413422967,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001384162,
      "loss": 3.0081,
      "step": 307920
    },
    {
      "epoch": 0.16859078457845486,
      "grad_norm": 0.75,
      "learning_rate": 0.0001384142,
      "loss": 3.1859,
      "step": 307930
    },
    {
      "epoch": 0.16860942781461305,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00013841220000000002,
      "loss": 2.9849,
      "step": 307940
    },
    {
      "epoch": 0.16862807105077124,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001384102,
      "loss": 3.1683,
      "step": 307950
    },
    {
      "epoch": 0.16864671428692943,
      "grad_norm": 0.75,
      "learning_rate": 0.0001384082,
      "loss": 3.0968,
      "step": 307960
    },
    {
      "epoch": 0.16866535752308762,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001384062,
      "loss": 3.1113,
      "step": 307970
    },
    {
      "epoch": 0.1686840007592458,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001384042,
      "loss": 3.2122,
      "step": 307980
    },
    {
      "epoch": 0.16870264399540397,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013840220000000002,
      "loss": 3.0952,
      "step": 307990
    },
    {
      "epoch": 0.16872128723156216,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001384002,
      "loss": 3.1375,
      "step": 308000
    },
    {
      "epoch": 0.16873993046772034,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001383982,
      "loss": 2.9909,
      "step": 308010
    },
    {
      "epoch": 0.16875857370387853,
      "grad_norm": 0.75,
      "learning_rate": 0.0001383962,
      "loss": 3.2143,
      "step": 308020
    },
    {
      "epoch": 0.16877721694003672,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001383942,
      "loss": 3.209,
      "step": 308030
    },
    {
      "epoch": 0.16879586017619488,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013839220000000003,
      "loss": 3.1827,
      "step": 308040
    },
    {
      "epoch": 0.16881450341235307,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001383902,
      "loss": 3.0483,
      "step": 308050
    },
    {
      "epoch": 0.16883314664851126,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013838820000000002,
      "loss": 3.1054,
      "step": 308060
    },
    {
      "epoch": 0.16885178988466945,
      "grad_norm": 0.625,
      "learning_rate": 0.00013838619999999999,
      "loss": 3.0407,
      "step": 308070
    },
    {
      "epoch": 0.16887043312082764,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001383842,
      "loss": 3.0674,
      "step": 308080
    },
    {
      "epoch": 0.1688890763569858,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001383822,
      "loss": 2.9831,
      "step": 308090
    },
    {
      "epoch": 0.168907719593144,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001383802,
      "loss": 3.1398,
      "step": 308100
    },
    {
      "epoch": 0.16892636282930218,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013837820000000002,
      "loss": 3.2091,
      "step": 308110
    },
    {
      "epoch": 0.16894500606546037,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001383762,
      "loss": 3.1628,
      "step": 308120
    },
    {
      "epoch": 0.16896364930161856,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013837420000000001,
      "loss": 3.113,
      "step": 308130
    },
    {
      "epoch": 0.16898229253777672,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013837219999999998,
      "loss": 3.1178,
      "step": 308140
    },
    {
      "epoch": 0.1690009357739349,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001383702,
      "loss": 3.15,
      "step": 308150
    },
    {
      "epoch": 0.1690195790100931,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001383682,
      "loss": 3.1325,
      "step": 308160
    },
    {
      "epoch": 0.16903822224625129,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001383662,
      "loss": 3.1067,
      "step": 308170
    },
    {
      "epoch": 0.16905686548240947,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013836420000000002,
      "loss": 3.0738,
      "step": 308180
    },
    {
      "epoch": 0.16907550871856766,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001383622,
      "loss": 3.2416,
      "step": 308190
    },
    {
      "epoch": 0.16909415195472582,
      "grad_norm": 0.625,
      "learning_rate": 0.0001383602,
      "loss": 3.1087,
      "step": 308200
    },
    {
      "epoch": 0.169112795190884,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001383582,
      "loss": 3.2175,
      "step": 308210
    },
    {
      "epoch": 0.1691314384270422,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001383562,
      "loss": 3.1641,
      "step": 308220
    },
    {
      "epoch": 0.1691500816632004,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013835420000000002,
      "loss": 3.1606,
      "step": 308230
    },
    {
      "epoch": 0.16916872489935858,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001383522,
      "loss": 3.068,
      "step": 308240
    },
    {
      "epoch": 0.16918736813551674,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013835020000000002,
      "loss": 3.2104,
      "step": 308250
    },
    {
      "epoch": 0.16920601137167493,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001383482,
      "loss": 3.0378,
      "step": 308260
    },
    {
      "epoch": 0.16922465460783312,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001383462,
      "loss": 3.1302,
      "step": 308270
    },
    {
      "epoch": 0.1692432978439913,
      "grad_norm": 0.75,
      "learning_rate": 0.0001383442,
      "loss": 3.1193,
      "step": 308280
    },
    {
      "epoch": 0.1692619410801495,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001383422,
      "loss": 3.0891,
      "step": 308290
    },
    {
      "epoch": 0.16928058431630766,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013834020000000002,
      "loss": 3.2859,
      "step": 308300
    },
    {
      "epoch": 0.16929922755246585,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001383382,
      "loss": 3.178,
      "step": 308310
    },
    {
      "epoch": 0.16931787078862404,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001383362,
      "loss": 3.0255,
      "step": 308320
    },
    {
      "epoch": 0.16933651402478223,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001383342,
      "loss": 3.1052,
      "step": 308330
    },
    {
      "epoch": 0.16935515726094041,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001383322,
      "loss": 3.1782,
      "step": 308340
    },
    {
      "epoch": 0.16937380049709858,
      "grad_norm": 0.75,
      "learning_rate": 0.0001383302,
      "loss": 3.1848,
      "step": 308350
    },
    {
      "epoch": 0.16939244373325676,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001383282,
      "loss": 3.1458,
      "step": 308360
    },
    {
      "epoch": 0.16941108696941495,
      "grad_norm": 0.75,
      "learning_rate": 0.00013832620000000002,
      "loss": 3.0415,
      "step": 308370
    },
    {
      "epoch": 0.16942973020557314,
      "grad_norm": 0.75,
      "learning_rate": 0.0001383242,
      "loss": 3.249,
      "step": 308380
    },
    {
      "epoch": 0.16944837344173133,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001383222,
      "loss": 3.1282,
      "step": 308390
    },
    {
      "epoch": 0.1694670166778895,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001383202,
      "loss": 3.0141,
      "step": 308400
    },
    {
      "epoch": 0.16948565991404768,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001383182,
      "loss": 3.0769,
      "step": 308410
    },
    {
      "epoch": 0.16950430315020587,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001383162,
      "loss": 3.0287,
      "step": 308420
    },
    {
      "epoch": 0.16952294638636406,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013831420000000002,
      "loss": 3.0901,
      "step": 308430
    },
    {
      "epoch": 0.16954158962252225,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013831220000000001,
      "loss": 3.0946,
      "step": 308440
    },
    {
      "epoch": 0.16956023285868044,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001383102,
      "loss": 3.1233,
      "step": 308450
    },
    {
      "epoch": 0.1695788760948386,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001383082,
      "loss": 3.2497,
      "step": 308460
    },
    {
      "epoch": 0.1695975193309968,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001383062,
      "loss": 3.2197,
      "step": 308470
    },
    {
      "epoch": 0.16961616256715498,
      "grad_norm": 0.75,
      "learning_rate": 0.00013830420000000002,
      "loss": 3.1439,
      "step": 308480
    },
    {
      "epoch": 0.16963480580331317,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001383022,
      "loss": 3.0826,
      "step": 308490
    },
    {
      "epoch": 0.16965344903947135,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013830020000000001,
      "loss": 3.2802,
      "step": 308500
    },
    {
      "epoch": 0.16967209227562952,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001382982,
      "loss": 3.0365,
      "step": 308510
    },
    {
      "epoch": 0.1696907355117877,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001382962,
      "loss": 3.1959,
      "step": 308520
    },
    {
      "epoch": 0.1697093787479459,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001382942,
      "loss": 3.1478,
      "step": 308530
    },
    {
      "epoch": 0.16972802198410408,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001382922,
      "loss": 3.114,
      "step": 308540
    },
    {
      "epoch": 0.16974666522026227,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013829020000000002,
      "loss": 3.1786,
      "step": 308550
    },
    {
      "epoch": 0.16976530845642043,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001382882,
      "loss": 3.1546,
      "step": 308560
    },
    {
      "epoch": 0.16978395169257862,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001382862,
      "loss": 3.1237,
      "step": 308570
    },
    {
      "epoch": 0.1698025949287368,
      "grad_norm": 0.75,
      "learning_rate": 0.0001382842,
      "loss": 2.9855,
      "step": 308580
    },
    {
      "epoch": 0.169821238164895,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001382822,
      "loss": 3.1657,
      "step": 308590
    },
    {
      "epoch": 0.1698398814010532,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013828020000000002,
      "loss": 3.2279,
      "step": 308600
    },
    {
      "epoch": 0.16985852463721135,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001382782,
      "loss": 3.2648,
      "step": 308610
    },
    {
      "epoch": 0.16987716787336954,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013827620000000002,
      "loss": 3.2033,
      "step": 308620
    },
    {
      "epoch": 0.16989581110952773,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013827419999999998,
      "loss": 3.155,
      "step": 308630
    },
    {
      "epoch": 0.16991445434568592,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001382722,
      "loss": 3.0474,
      "step": 308640
    },
    {
      "epoch": 0.1699330975818441,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001382702,
      "loss": 3.1037,
      "step": 308650
    },
    {
      "epoch": 0.1699517408180023,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001382682,
      "loss": 3.1084,
      "step": 308660
    },
    {
      "epoch": 0.16997038405416046,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013826620000000002,
      "loss": 3.1176,
      "step": 308670
    },
    {
      "epoch": 0.16998902729031865,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001382642,
      "loss": 3.0702,
      "step": 308680
    },
    {
      "epoch": 0.17000767052647683,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001382622,
      "loss": 3.21,
      "step": 308690
    },
    {
      "epoch": 0.17002631376263502,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001382602,
      "loss": 3.1004,
      "step": 308700
    },
    {
      "epoch": 0.1700449569987932,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001382582,
      "loss": 3.1202,
      "step": 308710
    },
    {
      "epoch": 0.17006360023495137,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013825620000000003,
      "loss": 3.2108,
      "step": 308720
    },
    {
      "epoch": 0.17008224347110956,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001382542,
      "loss": 3.2194,
      "step": 308730
    },
    {
      "epoch": 0.17010088670726775,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013825220000000002,
      "loss": 3.1518,
      "step": 308740
    },
    {
      "epoch": 0.17011952994342594,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013825019999999999,
      "loss": 3.1819,
      "step": 308750
    },
    {
      "epoch": 0.17013817317958413,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001382482,
      "loss": 3.0264,
      "step": 308760
    },
    {
      "epoch": 0.1701568164157423,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001382462,
      "loss": 3.0622,
      "step": 308770
    },
    {
      "epoch": 0.17017545965190048,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001382442,
      "loss": 3.1789,
      "step": 308780
    },
    {
      "epoch": 0.17019410288805867,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013824220000000002,
      "loss": 3.1903,
      "step": 308790
    },
    {
      "epoch": 0.17021274612421686,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001382402,
      "loss": 3.066,
      "step": 308800
    },
    {
      "epoch": 0.17023138936037505,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013823820000000001,
      "loss": 3.1883,
      "step": 308810
    },
    {
      "epoch": 0.1702500325965332,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001382362,
      "loss": 3.1758,
      "step": 308820
    },
    {
      "epoch": 0.1702686758326914,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001382342,
      "loss": 3.1674,
      "step": 308830
    },
    {
      "epoch": 0.17028731906884959,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001382322,
      "loss": 3.1745,
      "step": 308840
    },
    {
      "epoch": 0.17030596230500777,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001382302,
      "loss": 3.1279,
      "step": 308850
    },
    {
      "epoch": 0.17032460554116596,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013822820000000002,
      "loss": 3.0363,
      "step": 308860
    },
    {
      "epoch": 0.17034324877732415,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013822620000000001,
      "loss": 3.1586,
      "step": 308870
    },
    {
      "epoch": 0.1703618920134823,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001382242,
      "loss": 3.1445,
      "step": 308880
    },
    {
      "epoch": 0.1703805352496405,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001382222,
      "loss": 3.1375,
      "step": 308890
    },
    {
      "epoch": 0.1703991784857987,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001382202,
      "loss": 3.1204,
      "step": 308900
    },
    {
      "epoch": 0.17041782172195688,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001382182,
      "loss": 2.9893,
      "step": 308910
    },
    {
      "epoch": 0.17043646495811507,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001382162,
      "loss": 3.2428,
      "step": 308920
    },
    {
      "epoch": 0.17045510819427323,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013821420000000002,
      "loss": 3.0445,
      "step": 308930
    },
    {
      "epoch": 0.17047375143043142,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001382122,
      "loss": 3.1593,
      "step": 308940
    },
    {
      "epoch": 0.1704923946665896,
      "grad_norm": 0.875,
      "learning_rate": 0.0001382102,
      "loss": 3.0307,
      "step": 308950
    },
    {
      "epoch": 0.1705110379027478,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001382082,
      "loss": 3.065,
      "step": 308960
    },
    {
      "epoch": 0.170529681138906,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001382062,
      "loss": 3.0264,
      "step": 308970
    },
    {
      "epoch": 0.17054832437506415,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001382042,
      "loss": 3.1841,
      "step": 308980
    },
    {
      "epoch": 0.17056696761122234,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013820220000000002,
      "loss": 3.015,
      "step": 308990
    },
    {
      "epoch": 0.17058561084738053,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001382002,
      "loss": 3.0874,
      "step": 309000
    },
    {
      "epoch": 0.17060425408353871,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001381982,
      "loss": 3.1381,
      "step": 309010
    },
    {
      "epoch": 0.1706228973196969,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001381962,
      "loss": 3.111,
      "step": 309020
    },
    {
      "epoch": 0.17064154055585506,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001381942,
      "loss": 3.2187,
      "step": 309030
    },
    {
      "epoch": 0.17066018379201325,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013819220000000002,
      "loss": 3.1503,
      "step": 309040
    },
    {
      "epoch": 0.17067882702817144,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001381902,
      "loss": 3.1322,
      "step": 309050
    },
    {
      "epoch": 0.17069747026432963,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001381882,
      "loss": 3.0872,
      "step": 309060
    },
    {
      "epoch": 0.17071611350048782,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001381862,
      "loss": 3.0286,
      "step": 309070
    },
    {
      "epoch": 0.17073475673664598,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001381842,
      "loss": 3.2707,
      "step": 309080
    },
    {
      "epoch": 0.17075339997280417,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013818220000000003,
      "loss": 3.0806,
      "step": 309090
    },
    {
      "epoch": 0.17077204320896236,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001381802,
      "loss": 3.2461,
      "step": 309100
    },
    {
      "epoch": 0.17079068644512055,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013817820000000002,
      "loss": 3.1486,
      "step": 309110
    },
    {
      "epoch": 0.17080932968127874,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001381762,
      "loss": 3.2112,
      "step": 309120
    },
    {
      "epoch": 0.17082797291743693,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001381742,
      "loss": 3.1076,
      "step": 309130
    },
    {
      "epoch": 0.1708466161535951,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001381722,
      "loss": 3.0682,
      "step": 309140
    },
    {
      "epoch": 0.17086525938975328,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001381702,
      "loss": 3.2054,
      "step": 309150
    },
    {
      "epoch": 0.17088390262591147,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013816820000000002,
      "loss": 2.9896,
      "step": 309160
    },
    {
      "epoch": 0.17090254586206965,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001381662,
      "loss": 3.1643,
      "step": 309170
    },
    {
      "epoch": 0.17092118909822784,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013816420000000001,
      "loss": 3.1062,
      "step": 309180
    },
    {
      "epoch": 0.170939832334386,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001381622,
      "loss": 2.936,
      "step": 309190
    },
    {
      "epoch": 0.1709584755705442,
      "grad_norm": 0.75,
      "learning_rate": 0.0001381602,
      "loss": 3.1167,
      "step": 309200
    },
    {
      "epoch": 0.17097711880670238,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013815820000000003,
      "loss": 3.0852,
      "step": 309210
    },
    {
      "epoch": 0.17099576204286057,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001381562,
      "loss": 3.133,
      "step": 309220
    },
    {
      "epoch": 0.17101440527901876,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013815420000000002,
      "loss": 3.1188,
      "step": 309230
    },
    {
      "epoch": 0.17103304851517692,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001381522,
      "loss": 3.0991,
      "step": 309240
    },
    {
      "epoch": 0.1710516917513351,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001381502,
      "loss": 3.1965,
      "step": 309250
    },
    {
      "epoch": 0.1710703349874933,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001381482,
      "loss": 3.1924,
      "step": 309260
    },
    {
      "epoch": 0.1710889782236515,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001381462,
      "loss": 3.1756,
      "step": 309270
    },
    {
      "epoch": 0.17110762145980968,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013814420000000002,
      "loss": 3.1563,
      "step": 309280
    },
    {
      "epoch": 0.17112626469596784,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001381422,
      "loss": 3.232,
      "step": 309290
    },
    {
      "epoch": 0.17114490793212603,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013814020000000002,
      "loss": 3.2833,
      "step": 309300
    },
    {
      "epoch": 0.17116355116828422,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001381382,
      "loss": 3.1087,
      "step": 309310
    },
    {
      "epoch": 0.1711821944044424,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001381362,
      "loss": 3.2397,
      "step": 309320
    },
    {
      "epoch": 0.1712008376406006,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001381342,
      "loss": 3.172,
      "step": 309330
    },
    {
      "epoch": 0.17121948087675878,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001381322,
      "loss": 3.0444,
      "step": 309340
    },
    {
      "epoch": 0.17123812411291695,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013813020000000002,
      "loss": 3.2188,
      "step": 309350
    },
    {
      "epoch": 0.17125676734907513,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001381282,
      "loss": 3.1485,
      "step": 309360
    },
    {
      "epoch": 0.17127541058523332,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001381262,
      "loss": 3.1372,
      "step": 309370
    },
    {
      "epoch": 0.1712940538213915,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001381242,
      "loss": 3.1223,
      "step": 309380
    },
    {
      "epoch": 0.1713126970575497,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001381222,
      "loss": 3.104,
      "step": 309390
    },
    {
      "epoch": 0.17133134029370786,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001381202,
      "loss": 3.078,
      "step": 309400
    },
    {
      "epoch": 0.17134998352986605,
      "grad_norm": 0.75,
      "learning_rate": 0.0001381182,
      "loss": 3.0578,
      "step": 309410
    },
    {
      "epoch": 0.17136862676602424,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013811620000000002,
      "loss": 3.1609,
      "step": 309420
    },
    {
      "epoch": 0.17138727000218243,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001381142,
      "loss": 3.2127,
      "step": 309430
    },
    {
      "epoch": 0.17140591323834062,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001381122,
      "loss": 3.0245,
      "step": 309440
    },
    {
      "epoch": 0.17142455647449878,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001381102,
      "loss": 3.1984,
      "step": 309450
    },
    {
      "epoch": 0.17144319971065697,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001381082,
      "loss": 3.1413,
      "step": 309460
    },
    {
      "epoch": 0.17146184294681516,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001381062,
      "loss": 3.0741,
      "step": 309470
    },
    {
      "epoch": 0.17148048618297335,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013810420000000002,
      "loss": 3.052,
      "step": 309480
    },
    {
      "epoch": 0.17149912941913154,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013810220000000001,
      "loss": 3.1088,
      "step": 309490
    },
    {
      "epoch": 0.1715177726552897,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001381002,
      "loss": 2.9555,
      "step": 309500
    },
    {
      "epoch": 0.17153641589144789,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001380982,
      "loss": 3.0734,
      "step": 309510
    },
    {
      "epoch": 0.17155505912760607,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001380962,
      "loss": 3.2289,
      "step": 309520
    },
    {
      "epoch": 0.17157370236376426,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001380942,
      "loss": 2.9783,
      "step": 309530
    },
    {
      "epoch": 0.17159234559992245,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001380922,
      "loss": 3.1525,
      "step": 309540
    },
    {
      "epoch": 0.17161098883608064,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013809020000000001,
      "loss": 2.9944,
      "step": 309550
    },
    {
      "epoch": 0.1716296320722388,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001380882,
      "loss": 3.1173,
      "step": 309560
    },
    {
      "epoch": 0.171648275308397,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001380862,
      "loss": 3.1317,
      "step": 309570
    },
    {
      "epoch": 0.17166691854455518,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001380842,
      "loss": 3.0906,
      "step": 309580
    },
    {
      "epoch": 0.17168556178071337,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001380822,
      "loss": 3.0604,
      "step": 309590
    },
    {
      "epoch": 0.17170420501687156,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013808020000000002,
      "loss": 3.1154,
      "step": 309600
    },
    {
      "epoch": 0.17172284825302972,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001380782,
      "loss": 3.1729,
      "step": 309610
    },
    {
      "epoch": 0.1717414914891879,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001380762,
      "loss": 3.0926,
      "step": 309620
    },
    {
      "epoch": 0.1717601347253461,
      "grad_norm": 0.625,
      "learning_rate": 0.0001380742,
      "loss": 3.0939,
      "step": 309630
    },
    {
      "epoch": 0.1717787779615043,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001380722,
      "loss": 3.0761,
      "step": 309640
    },
    {
      "epoch": 0.17179742119766248,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013807020000000003,
      "loss": 3.1706,
      "step": 309650
    },
    {
      "epoch": 0.17181606443382064,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001380682,
      "loss": 3.0527,
      "step": 309660
    },
    {
      "epoch": 0.17183470766997883,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013806620000000002,
      "loss": 3.1652,
      "step": 309670
    },
    {
      "epoch": 0.17185335090613701,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001380642,
      "loss": 3.1224,
      "step": 309680
    },
    {
      "epoch": 0.1718719941422952,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001380622,
      "loss": 3.0506,
      "step": 309690
    },
    {
      "epoch": 0.1718906373784534,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013806020000000003,
      "loss": 3.1062,
      "step": 309700
    },
    {
      "epoch": 0.17190928061461155,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001380582,
      "loss": 2.9833,
      "step": 309710
    },
    {
      "epoch": 0.17192792385076974,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013805620000000002,
      "loss": 3.1388,
      "step": 309720
    },
    {
      "epoch": 0.17194656708692793,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001380542,
      "loss": 3.1138,
      "step": 309730
    },
    {
      "epoch": 0.17196521032308612,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001380522,
      "loss": 3.0279,
      "step": 309740
    },
    {
      "epoch": 0.1719838535592443,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001380502,
      "loss": 3.1859,
      "step": 309750
    },
    {
      "epoch": 0.1720024967954025,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001380482,
      "loss": 3.2789,
      "step": 309760
    },
    {
      "epoch": 0.17202114003156066,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013804620000000003,
      "loss": 3.139,
      "step": 309770
    },
    {
      "epoch": 0.17203978326771885,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001380442,
      "loss": 3.1453,
      "step": 309780
    },
    {
      "epoch": 0.17205842650387704,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013804220000000002,
      "loss": 3.149,
      "step": 309790
    },
    {
      "epoch": 0.17207706974003523,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001380402,
      "loss": 3.1594,
      "step": 309800
    },
    {
      "epoch": 0.17209571297619342,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001380382,
      "loss": 3.1341,
      "step": 309810
    },
    {
      "epoch": 0.17211435621235158,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001380362,
      "loss": 3.1628,
      "step": 309820
    },
    {
      "epoch": 0.17213299944850977,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001380342,
      "loss": 3.1337,
      "step": 309830
    },
    {
      "epoch": 0.17215164268466795,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013803220000000002,
      "loss": 3.1325,
      "step": 309840
    },
    {
      "epoch": 0.17217028592082614,
      "grad_norm": 0.75,
      "learning_rate": 0.0001380302,
      "loss": 3.0766,
      "step": 309850
    },
    {
      "epoch": 0.17218892915698433,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013802820000000001,
      "loss": 3.172,
      "step": 309860
    },
    {
      "epoch": 0.1722075723931425,
      "grad_norm": 0.625,
      "learning_rate": 0.0001380262,
      "loss": 3.257,
      "step": 309870
    },
    {
      "epoch": 0.17222621562930068,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001380242,
      "loss": 3.1391,
      "step": 309880
    },
    {
      "epoch": 0.17224485886545887,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001380222,
      "loss": 3.1141,
      "step": 309890
    },
    {
      "epoch": 0.17226350210161706,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001380202,
      "loss": 3.1479,
      "step": 309900
    },
    {
      "epoch": 0.17228214533777525,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013801820000000002,
      "loss": 3.0655,
      "step": 309910
    },
    {
      "epoch": 0.1723007885739334,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013801620000000002,
      "loss": 2.9826,
      "step": 309920
    },
    {
      "epoch": 0.1723194318100916,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001380142,
      "loss": 3.1473,
      "step": 309930
    },
    {
      "epoch": 0.1723380750462498,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001380122,
      "loss": 3.1166,
      "step": 309940
    },
    {
      "epoch": 0.17235671828240798,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001380102,
      "loss": 3.185,
      "step": 309950
    },
    {
      "epoch": 0.17237536151856617,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001380082,
      "loss": 3.1097,
      "step": 309960
    },
    {
      "epoch": 0.17239400475472433,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001380062,
      "loss": 3.0786,
      "step": 309970
    },
    {
      "epoch": 0.17241264799088252,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013800420000000002,
      "loss": 3.1351,
      "step": 309980
    },
    {
      "epoch": 0.1724312912270407,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001380022,
      "loss": 3.1396,
      "step": 309990
    },
    {
      "epoch": 0.1724499344631989,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001380002,
      "loss": 3.0715,
      "step": 310000
    },
    {
      "epoch": 0.17246857769935708,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001379982,
      "loss": 3.1338,
      "step": 310010
    },
    {
      "epoch": 0.17248722093551527,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001379962,
      "loss": 3.1096,
      "step": 310020
    },
    {
      "epoch": 0.17250586417167343,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001379942,
      "loss": 3.2741,
      "step": 310030
    },
    {
      "epoch": 0.17252450740783162,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013799220000000002,
      "loss": 3.0607,
      "step": 310040
    },
    {
      "epoch": 0.1725431506439898,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001379902,
      "loss": 3.2046,
      "step": 310050
    },
    {
      "epoch": 0.172561793880148,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001379882,
      "loss": 3.1286,
      "step": 310060
    },
    {
      "epoch": 0.1725804371163062,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001379862,
      "loss": 3.1079,
      "step": 310070
    },
    {
      "epoch": 0.17259908035246435,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001379842,
      "loss": 3.2479,
      "step": 310080
    },
    {
      "epoch": 0.17261772358862254,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013798220000000002,
      "loss": 3.0687,
      "step": 310090
    },
    {
      "epoch": 0.17263636682478073,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001379802,
      "loss": 3.0418,
      "step": 310100
    },
    {
      "epoch": 0.17265501006093892,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001379782,
      "loss": 3.1884,
      "step": 310110
    },
    {
      "epoch": 0.1726736532970971,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001379762,
      "loss": 3.0463,
      "step": 310120
    },
    {
      "epoch": 0.17269229653325527,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001379742,
      "loss": 3.1415,
      "step": 310130
    },
    {
      "epoch": 0.17271093976941346,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001379722,
      "loss": 3.0516,
      "step": 310140
    },
    {
      "epoch": 0.17272958300557165,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001379702,
      "loss": 3.1632,
      "step": 310150
    },
    {
      "epoch": 0.17274822624172984,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013796820000000002,
      "loss": 3.0098,
      "step": 310160
    },
    {
      "epoch": 0.17276686947788802,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013796620000000001,
      "loss": 3.1337,
      "step": 310170
    },
    {
      "epoch": 0.17278551271404619,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001379642,
      "loss": 3.1579,
      "step": 310180
    },
    {
      "epoch": 0.17280415595020437,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001379622,
      "loss": 3.0707,
      "step": 310190
    },
    {
      "epoch": 0.17282279918636256,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001379602,
      "loss": 3.206,
      "step": 310200
    },
    {
      "epoch": 0.17284144242252075,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013795820000000002,
      "loss": 3.1224,
      "step": 310210
    },
    {
      "epoch": 0.17286008565867894,
      "grad_norm": 0.75,
      "learning_rate": 0.0001379562,
      "loss": 3.1739,
      "step": 310220
    },
    {
      "epoch": 0.17287872889483713,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013795420000000001,
      "loss": 3.1581,
      "step": 310230
    },
    {
      "epoch": 0.1728973721309953,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001379522,
      "loss": 3.0885,
      "step": 310240
    },
    {
      "epoch": 0.17291601536715348,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001379502,
      "loss": 3.0398,
      "step": 310250
    },
    {
      "epoch": 0.17293465860331167,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013794820000000003,
      "loss": 3.2274,
      "step": 310260
    },
    {
      "epoch": 0.17295330183946986,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001379462,
      "loss": 3.0496,
      "step": 310270
    },
    {
      "epoch": 0.17297194507562805,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013794420000000002,
      "loss": 2.9878,
      "step": 310280
    },
    {
      "epoch": 0.1729905883117862,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001379422,
      "loss": 3.1,
      "step": 310290
    },
    {
      "epoch": 0.1730092315479444,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001379402,
      "loss": 3.1977,
      "step": 310300
    },
    {
      "epoch": 0.1730278747841026,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001379382,
      "loss": 3.1187,
      "step": 310310
    },
    {
      "epoch": 0.17304651802026078,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001379362,
      "loss": 3.2824,
      "step": 310320
    },
    {
      "epoch": 0.17306516125641896,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013793420000000003,
      "loss": 3.163,
      "step": 310330
    },
    {
      "epoch": 0.17308380449257713,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001379322,
      "loss": 3.031,
      "step": 310340
    },
    {
      "epoch": 0.17310244772873531,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013793020000000002,
      "loss": 3.1455,
      "step": 310350
    },
    {
      "epoch": 0.1731210909648935,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013792819999999998,
      "loss": 3.0348,
      "step": 310360
    },
    {
      "epoch": 0.1731397342010517,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001379262,
      "loss": 3.0654,
      "step": 310370
    },
    {
      "epoch": 0.17315837743720988,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001379242,
      "loss": 3.1054,
      "step": 310380
    },
    {
      "epoch": 0.17317702067336804,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001379222,
      "loss": 3.1493,
      "step": 310390
    },
    {
      "epoch": 0.17319566390952623,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013792020000000002,
      "loss": 2.9913,
      "step": 310400
    },
    {
      "epoch": 0.17321430714568442,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001379182,
      "loss": 3.086,
      "step": 310410
    },
    {
      "epoch": 0.1732329503818426,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001379162,
      "loss": 3.1749,
      "step": 310420
    },
    {
      "epoch": 0.1732515936180008,
      "grad_norm": 0.75,
      "learning_rate": 0.0001379142,
      "loss": 3.1841,
      "step": 310430
    },
    {
      "epoch": 0.173270236854159,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001379122,
      "loss": 3.0547,
      "step": 310440
    },
    {
      "epoch": 0.17328888009031715,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001379102,
      "loss": 2.9932,
      "step": 310450
    },
    {
      "epoch": 0.17330752332647534,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001379082,
      "loss": 3.1027,
      "step": 310460
    },
    {
      "epoch": 0.17332616656263353,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013790620000000002,
      "loss": 3.0808,
      "step": 310470
    },
    {
      "epoch": 0.17334480979879172,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013790420000000001,
      "loss": 3.148,
      "step": 310480
    },
    {
      "epoch": 0.1733634530349499,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001379022,
      "loss": 3.0356,
      "step": 310490
    },
    {
      "epoch": 0.17338209627110807,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001379002,
      "loss": 3.0678,
      "step": 310500
    },
    {
      "epoch": 0.17340073950726625,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001378982,
      "loss": 3.1476,
      "step": 310510
    },
    {
      "epoch": 0.17341938274342444,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001378962,
      "loss": 3.1341,
      "step": 310520
    },
    {
      "epoch": 0.17343802597958263,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001378942,
      "loss": 3.1901,
      "step": 310530
    },
    {
      "epoch": 0.17345666921574082,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013789220000000001,
      "loss": 3.1482,
      "step": 310540
    },
    {
      "epoch": 0.17347531245189898,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001378902,
      "loss": 3.1297,
      "step": 310550
    },
    {
      "epoch": 0.17349395568805717,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001378882,
      "loss": 3.2445,
      "step": 310560
    },
    {
      "epoch": 0.17351259892421536,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001378862,
      "loss": 3.0842,
      "step": 310570
    },
    {
      "epoch": 0.17353124216037355,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001378842,
      "loss": 3.1086,
      "step": 310580
    },
    {
      "epoch": 0.17354988539653174,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001378822,
      "loss": 3.0751,
      "step": 310590
    },
    {
      "epoch": 0.1735685286326899,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013788020000000002,
      "loss": 2.9799,
      "step": 310600
    },
    {
      "epoch": 0.1735871718688481,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001378782,
      "loss": 3.025,
      "step": 310610
    },
    {
      "epoch": 0.17360581510500628,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001378762,
      "loss": 3.0986,
      "step": 310620
    },
    {
      "epoch": 0.17362445834116447,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001378742,
      "loss": 2.9675,
      "step": 310630
    },
    {
      "epoch": 0.17364310157732266,
      "grad_norm": 0.75,
      "learning_rate": 0.0001378722,
      "loss": 3.1613,
      "step": 310640
    },
    {
      "epoch": 0.17366174481348082,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013787020000000002,
      "loss": 3.1324,
      "step": 310650
    },
    {
      "epoch": 0.173680388049639,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013786820000000002,
      "loss": 3.0576,
      "step": 310660
    },
    {
      "epoch": 0.1736990312857972,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001378662,
      "loss": 3.0512,
      "step": 310670
    },
    {
      "epoch": 0.17371767452195538,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001378642,
      "loss": 3.1064,
      "step": 310680
    },
    {
      "epoch": 0.17373631775811357,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001378622,
      "loss": 3.1719,
      "step": 310690
    },
    {
      "epoch": 0.17375496099427176,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013786020000000003,
      "loss": 3.2303,
      "step": 310700
    },
    {
      "epoch": 0.17377360423042992,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001378582,
      "loss": 3.0624,
      "step": 310710
    },
    {
      "epoch": 0.1737922474665881,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013785620000000002,
      "loss": 3.1837,
      "step": 310720
    },
    {
      "epoch": 0.1738108907027463,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001378542,
      "loss": 3.2239,
      "step": 310730
    },
    {
      "epoch": 0.1738295339389045,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001378522,
      "loss": 3.0979,
      "step": 310740
    },
    {
      "epoch": 0.17384817717506268,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001378502,
      "loss": 3.0795,
      "step": 310750
    },
    {
      "epoch": 0.17386682041122084,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001378482,
      "loss": 3.0475,
      "step": 310760
    },
    {
      "epoch": 0.17388546364737903,
      "grad_norm": 0.75,
      "learning_rate": 0.00013784620000000002,
      "loss": 3.1987,
      "step": 310770
    },
    {
      "epoch": 0.17390410688353722,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001378442,
      "loss": 3.2026,
      "step": 310780
    },
    {
      "epoch": 0.1739227501196954,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001378422,
      "loss": 3.115,
      "step": 310790
    },
    {
      "epoch": 0.1739413933558536,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001378402,
      "loss": 3.1303,
      "step": 310800
    },
    {
      "epoch": 0.17396003659201176,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001378382,
      "loss": 3.1531,
      "step": 310810
    },
    {
      "epoch": 0.17397867982816995,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013783620000000003,
      "loss": 3.0787,
      "step": 310820
    },
    {
      "epoch": 0.17399732306432814,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001378342,
      "loss": 3.1491,
      "step": 310830
    },
    {
      "epoch": 0.17401596630048632,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013783220000000002,
      "loss": 3.1594,
      "step": 310840
    },
    {
      "epoch": 0.1740346095366445,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001378302,
      "loss": 3.0155,
      "step": 310850
    },
    {
      "epoch": 0.17405325277280267,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001378282,
      "loss": 3.1676,
      "step": 310860
    },
    {
      "epoch": 0.17407189600896086,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001378262,
      "loss": 3.149,
      "step": 310870
    },
    {
      "epoch": 0.17409053924511905,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001378242,
      "loss": 3.1353,
      "step": 310880
    },
    {
      "epoch": 0.17410918248127724,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013782220000000002,
      "loss": 3.1364,
      "step": 310890
    },
    {
      "epoch": 0.17412782571743543,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001378202,
      "loss": 3.0211,
      "step": 310900
    },
    {
      "epoch": 0.17414646895359362,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013781820000000001,
      "loss": 3.2037,
      "step": 310910
    },
    {
      "epoch": 0.17416511218975178,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001378162,
      "loss": 3.2105,
      "step": 310920
    },
    {
      "epoch": 0.17418375542590997,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001378142,
      "loss": 3.1147,
      "step": 310930
    },
    {
      "epoch": 0.17420239866206816,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001378122,
      "loss": 3.0638,
      "step": 310940
    },
    {
      "epoch": 0.17422104189822635,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001378102,
      "loss": 3.0819,
      "step": 310950
    },
    {
      "epoch": 0.17423968513438454,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013780820000000002,
      "loss": 3.1296,
      "step": 310960
    },
    {
      "epoch": 0.1742583283705427,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001378062,
      "loss": 3.0697,
      "step": 310970
    },
    {
      "epoch": 0.1742769716067009,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001378042,
      "loss": 3.0482,
      "step": 310980
    },
    {
      "epoch": 0.17429561484285908,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001378022,
      "loss": 3.1795,
      "step": 310990
    },
    {
      "epoch": 0.17431425807901726,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001378002,
      "loss": 3.1823,
      "step": 311000
    },
    {
      "epoch": 0.17433290131517545,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001377982,
      "loss": 3.1705,
      "step": 311010
    },
    {
      "epoch": 0.17435154455133361,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001377962,
      "loss": 3.0688,
      "step": 311020
    },
    {
      "epoch": 0.1743701877874918,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013779420000000002,
      "loss": 3.1331,
      "step": 311030
    },
    {
      "epoch": 0.17438883102365,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001377922,
      "loss": 3.0381,
      "step": 311040
    },
    {
      "epoch": 0.17440747425980818,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001377902,
      "loss": 3.1433,
      "step": 311050
    },
    {
      "epoch": 0.17442611749596637,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001377882,
      "loss": 3.1691,
      "step": 311060
    },
    {
      "epoch": 0.17444476073212453,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001377862,
      "loss": 3.1371,
      "step": 311070
    },
    {
      "epoch": 0.17446340396828272,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001377842,
      "loss": 3.1547,
      "step": 311080
    },
    {
      "epoch": 0.1744820472044409,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013778220000000002,
      "loss": 3.0841,
      "step": 311090
    },
    {
      "epoch": 0.1745006904405991,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001377802,
      "loss": 3.1436,
      "step": 311100
    },
    {
      "epoch": 0.1745193336767573,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001377782,
      "loss": 3.1898,
      "step": 311110
    },
    {
      "epoch": 0.17453797691291548,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001377762,
      "loss": 3.0245,
      "step": 311120
    },
    {
      "epoch": 0.17455662014907364,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001377742,
      "loss": 3.0802,
      "step": 311130
    },
    {
      "epoch": 0.17457526338523183,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001377722,
      "loss": 3.2121,
      "step": 311140
    },
    {
      "epoch": 0.17459390662139002,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013777020000000002,
      "loss": 3.1178,
      "step": 311150
    },
    {
      "epoch": 0.1746125498575482,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013776820000000001,
      "loss": 3.1161,
      "step": 311160
    },
    {
      "epoch": 0.1746311930937064,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001377662,
      "loss": 3.1303,
      "step": 311170
    },
    {
      "epoch": 0.17464983632986456,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001377642,
      "loss": 3.1104,
      "step": 311180
    },
    {
      "epoch": 0.17466847956602274,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001377622,
      "loss": 3.0462,
      "step": 311190
    },
    {
      "epoch": 0.17468712280218093,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001377602,
      "loss": 3.1792,
      "step": 311200
    },
    {
      "epoch": 0.17470576603833912,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013775820000000002,
      "loss": 3.2295,
      "step": 311210
    },
    {
      "epoch": 0.1747244092744973,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013775620000000001,
      "loss": 3.2027,
      "step": 311220
    },
    {
      "epoch": 0.17474305251065547,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001377542,
      "loss": 3.0894,
      "step": 311230
    },
    {
      "epoch": 0.17476169574681366,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001377522,
      "loss": 3.1364,
      "step": 311240
    },
    {
      "epoch": 0.17478033898297185,
      "grad_norm": 0.625,
      "learning_rate": 0.0001377502,
      "loss": 3.0625,
      "step": 311250
    },
    {
      "epoch": 0.17479898221913004,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013774820000000002,
      "loss": 3.1342,
      "step": 311260
    },
    {
      "epoch": 0.17481762545528823,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001377462,
      "loss": 3.1888,
      "step": 311270
    },
    {
      "epoch": 0.1748362686914464,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013774420000000002,
      "loss": 3.0485,
      "step": 311280
    },
    {
      "epoch": 0.17485491192760458,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001377422,
      "loss": 3.0934,
      "step": 311290
    },
    {
      "epoch": 0.17487355516376277,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001377402,
      "loss": 3.0624,
      "step": 311300
    },
    {
      "epoch": 0.17489219839992096,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013773820000000003,
      "loss": 3.1562,
      "step": 311310
    },
    {
      "epoch": 0.17491084163607915,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001377362,
      "loss": 3.1375,
      "step": 311320
    },
    {
      "epoch": 0.1749294848722373,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013773420000000002,
      "loss": 3.0065,
      "step": 311330
    },
    {
      "epoch": 0.1749481281083955,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001377322,
      "loss": 3.1047,
      "step": 311340
    },
    {
      "epoch": 0.17496677134455368,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001377302,
      "loss": 3.1163,
      "step": 311350
    },
    {
      "epoch": 0.17498541458071187,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001377282,
      "loss": 3.1757,
      "step": 311360
    },
    {
      "epoch": 0.17500405781687006,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001377262,
      "loss": 3.1418,
      "step": 311370
    },
    {
      "epoch": 0.17502270105302825,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013772420000000003,
      "loss": 3.0454,
      "step": 311380
    },
    {
      "epoch": 0.1750413442891864,
      "grad_norm": 0.625,
      "learning_rate": 0.0001377222,
      "loss": 3.1319,
      "step": 311390
    },
    {
      "epoch": 0.1750599875253446,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013772020000000002,
      "loss": 3.0735,
      "step": 311400
    },
    {
      "epoch": 0.1750786307615028,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013771819999999999,
      "loss": 3.1168,
      "step": 311410
    },
    {
      "epoch": 0.17509727399766098,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001377162,
      "loss": 3.1086,
      "step": 311420
    },
    {
      "epoch": 0.17511591723381917,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001377142,
      "loss": 3.1072,
      "step": 311430
    },
    {
      "epoch": 0.17513456046997733,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001377122,
      "loss": 3.1535,
      "step": 311440
    },
    {
      "epoch": 0.17515320370613552,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013771020000000002,
      "loss": 3.0836,
      "step": 311450
    },
    {
      "epoch": 0.1751718469422937,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001377082,
      "loss": 3.2696,
      "step": 311460
    },
    {
      "epoch": 0.1751904901784519,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001377062,
      "loss": 3.0618,
      "step": 311470
    },
    {
      "epoch": 0.17520913341461009,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001377042,
      "loss": 3.0483,
      "step": 311480
    },
    {
      "epoch": 0.17522777665076825,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001377022,
      "loss": 3.1295,
      "step": 311490
    },
    {
      "epoch": 0.17524641988692644,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001377002,
      "loss": 3.1548,
      "step": 311500
    },
    {
      "epoch": 0.17526506312308462,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001376982,
      "loss": 3.0507,
      "step": 311510
    },
    {
      "epoch": 0.1752837063592428,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013769620000000002,
      "loss": 3.083,
      "step": 311520
    },
    {
      "epoch": 0.175302349595401,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013769420000000001,
      "loss": 3.0807,
      "step": 311530
    },
    {
      "epoch": 0.17532099283155916,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001376922,
      "loss": 2.9825,
      "step": 311540
    },
    {
      "epoch": 0.17533963606771735,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001376902,
      "loss": 3.1928,
      "step": 311550
    },
    {
      "epoch": 0.17535827930387554,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001376882,
      "loss": 3.2213,
      "step": 311560
    },
    {
      "epoch": 0.17537692254003373,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001376862,
      "loss": 3.0538,
      "step": 311570
    },
    {
      "epoch": 0.17539556577619192,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001376842,
      "loss": 3.14,
      "step": 311580
    },
    {
      "epoch": 0.1754142090123501,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013768220000000001,
      "loss": 3.1521,
      "step": 311590
    },
    {
      "epoch": 0.17543285224850827,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001376802,
      "loss": 3.1521,
      "step": 311600
    },
    {
      "epoch": 0.17545149548466646,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001376782,
      "loss": 3.0734,
      "step": 311610
    },
    {
      "epoch": 0.17547013872082465,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001376762,
      "loss": 3.1064,
      "step": 311620
    },
    {
      "epoch": 0.17548878195698284,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001376742,
      "loss": 3.12,
      "step": 311630
    },
    {
      "epoch": 0.17550742519314103,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013767220000000002,
      "loss": 3.2002,
      "step": 311640
    },
    {
      "epoch": 0.1755260684292992,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013767020000000002,
      "loss": 3.0283,
      "step": 311650
    },
    {
      "epoch": 0.17554471166545738,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001376682,
      "loss": 3.1388,
      "step": 311660
    },
    {
      "epoch": 0.17556335490161556,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001376662,
      "loss": 3.0545,
      "step": 311670
    },
    {
      "epoch": 0.17558199813777375,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001376642,
      "loss": 3.1329,
      "step": 311680
    },
    {
      "epoch": 0.17560064137393194,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001376622,
      "loss": 3.048,
      "step": 311690
    },
    {
      "epoch": 0.1756192846100901,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013766020000000002,
      "loss": 3.1689,
      "step": 311700
    },
    {
      "epoch": 0.1756379278462483,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013765820000000002,
      "loss": 3.3141,
      "step": 311710
    },
    {
      "epoch": 0.17565657108240648,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001376562,
      "loss": 3.1602,
      "step": 311720
    },
    {
      "epoch": 0.17567521431856467,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001376542,
      "loss": 3.059,
      "step": 311730
    },
    {
      "epoch": 0.17569385755472286,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001376522,
      "loss": 3.0379,
      "step": 311740
    },
    {
      "epoch": 0.17571250079088102,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001376502,
      "loss": 3.1115,
      "step": 311750
    },
    {
      "epoch": 0.1757311440270392,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001376482,
      "loss": 3.1559,
      "step": 311760
    },
    {
      "epoch": 0.1757497872631974,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013764620000000002,
      "loss": 3.2492,
      "step": 311770
    },
    {
      "epoch": 0.1757684304993556,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001376442,
      "loss": 3.0542,
      "step": 311780
    },
    {
      "epoch": 0.17578707373551378,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001376422,
      "loss": 3.1357,
      "step": 311790
    },
    {
      "epoch": 0.17580571697167197,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001376402,
      "loss": 3.0994,
      "step": 311800
    },
    {
      "epoch": 0.17582436020783013,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001376382,
      "loss": 3.2315,
      "step": 311810
    },
    {
      "epoch": 0.17584300344398832,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013763620000000002,
      "loss": 3.1081,
      "step": 311820
    },
    {
      "epoch": 0.1758616466801465,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001376342,
      "loss": 3.2429,
      "step": 311830
    },
    {
      "epoch": 0.1758802899163047,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013763220000000001,
      "loss": 3.1969,
      "step": 311840
    },
    {
      "epoch": 0.17589893315246288,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001376302,
      "loss": 3.1687,
      "step": 311850
    },
    {
      "epoch": 0.17591757638862104,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001376282,
      "loss": 3.1577,
      "step": 311860
    },
    {
      "epoch": 0.17593621962477923,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013762620000000003,
      "loss": 3.0545,
      "step": 311870
    },
    {
      "epoch": 0.17595486286093742,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001376242,
      "loss": 3.176,
      "step": 311880
    },
    {
      "epoch": 0.1759735060970956,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013762220000000002,
      "loss": 3.1437,
      "step": 311890
    },
    {
      "epoch": 0.1759921493332538,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001376202,
      "loss": 3.211,
      "step": 311900
    },
    {
      "epoch": 0.17601079256941196,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001376182,
      "loss": 3.0055,
      "step": 311910
    },
    {
      "epoch": 0.17602943580557015,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001376162,
      "loss": 3.2169,
      "step": 311920
    },
    {
      "epoch": 0.17604807904172834,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001376142,
      "loss": 3.2045,
      "step": 311930
    },
    {
      "epoch": 0.17606672227788653,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013761220000000002,
      "loss": 3.2033,
      "step": 311940
    },
    {
      "epoch": 0.17608536551404472,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001376102,
      "loss": 3.1012,
      "step": 311950
    },
    {
      "epoch": 0.17610400875020288,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013760820000000002,
      "loss": 3.1845,
      "step": 311960
    },
    {
      "epoch": 0.17612265198636107,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013760619999999998,
      "loss": 3.0366,
      "step": 311970
    },
    {
      "epoch": 0.17614129522251926,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001376042,
      "loss": 3.0045,
      "step": 311980
    },
    {
      "epoch": 0.17615993845867745,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001376022,
      "loss": 3.0453,
      "step": 311990
    },
    {
      "epoch": 0.17617858169483563,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001376002,
      "loss": 3.1314,
      "step": 312000
    },
    {
      "epoch": 0.1761972249309938,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013759820000000002,
      "loss": 3.0598,
      "step": 312010
    },
    {
      "epoch": 0.17621586816715198,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001375962,
      "loss": 3.0881,
      "step": 312020
    },
    {
      "epoch": 0.17623451140331017,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001375942,
      "loss": 3.0601,
      "step": 312030
    },
    {
      "epoch": 0.17625315463946836,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001375922,
      "loss": 3.271,
      "step": 312040
    },
    {
      "epoch": 0.17627179787562655,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001375902,
      "loss": 3.1325,
      "step": 312050
    },
    {
      "epoch": 0.17629044111178474,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001375882,
      "loss": 3.1895,
      "step": 312060
    },
    {
      "epoch": 0.1763090843479429,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001375862,
      "loss": 3.0912,
      "step": 312070
    },
    {
      "epoch": 0.1763277275841011,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013758420000000002,
      "loss": 3.1692,
      "step": 312080
    },
    {
      "epoch": 0.17634637082025928,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001375822,
      "loss": 3.0118,
      "step": 312090
    },
    {
      "epoch": 0.17636501405641747,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001375802,
      "loss": 3.1403,
      "step": 312100
    },
    {
      "epoch": 0.17638365729257566,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001375782,
      "loss": 3.1381,
      "step": 312110
    },
    {
      "epoch": 0.17640230052873382,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001375762,
      "loss": 3.2038,
      "step": 312120
    },
    {
      "epoch": 0.176420943764892,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013757420000000002,
      "loss": 2.8875,
      "step": 312130
    },
    {
      "epoch": 0.1764395870010502,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001375722,
      "loss": 3.1855,
      "step": 312140
    },
    {
      "epoch": 0.17645823023720839,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001375702,
      "loss": 3.0943,
      "step": 312150
    },
    {
      "epoch": 0.17647687347336657,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001375682,
      "loss": 3.009,
      "step": 312160
    },
    {
      "epoch": 0.17649551670952474,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001375662,
      "loss": 3.1239,
      "step": 312170
    },
    {
      "epoch": 0.17651415994568292,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001375642,
      "loss": 3.0902,
      "step": 312180
    },
    {
      "epoch": 0.1765328031818411,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001375622,
      "loss": 3.1354,
      "step": 312190
    },
    {
      "epoch": 0.1765514464179993,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013756020000000002,
      "loss": 3.0717,
      "step": 312200
    },
    {
      "epoch": 0.1765700896541575,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013755820000000001,
      "loss": 3.212,
      "step": 312210
    },
    {
      "epoch": 0.17658873289031565,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001375562,
      "loss": 3.0553,
      "step": 312220
    },
    {
      "epoch": 0.17660737612647384,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001375542,
      "loss": 3.1933,
      "step": 312230
    },
    {
      "epoch": 0.17662601936263203,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001375522,
      "loss": 3.0349,
      "step": 312240
    },
    {
      "epoch": 0.17664466259879022,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001375502,
      "loss": 3.0933,
      "step": 312250
    },
    {
      "epoch": 0.1766633058349484,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013754820000000002,
      "loss": 3.0754,
      "step": 312260
    },
    {
      "epoch": 0.1766819490711066,
      "grad_norm": 0.75,
      "learning_rate": 0.00013754620000000001,
      "loss": 3.0573,
      "step": 312270
    },
    {
      "epoch": 0.17670059230726476,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001375442,
      "loss": 3.1928,
      "step": 312280
    },
    {
      "epoch": 0.17671923554342295,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001375422,
      "loss": 3.0956,
      "step": 312290
    },
    {
      "epoch": 0.17673787877958114,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001375402,
      "loss": 3.1589,
      "step": 312300
    },
    {
      "epoch": 0.17675652201573933,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013753820000000002,
      "loss": 3.0662,
      "step": 312310
    },
    {
      "epoch": 0.17677516525189751,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001375362,
      "loss": 3.1672,
      "step": 312320
    },
    {
      "epoch": 0.17679380848805568,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013753420000000002,
      "loss": 3.1735,
      "step": 312330
    },
    {
      "epoch": 0.17681245172421386,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001375322,
      "loss": 3.2039,
      "step": 312340
    },
    {
      "epoch": 0.17683109496037205,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001375302,
      "loss": 2.9756,
      "step": 312350
    },
    {
      "epoch": 0.17684973819653024,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001375282,
      "loss": 3.1836,
      "step": 312360
    },
    {
      "epoch": 0.17686838143268843,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001375262,
      "loss": 3.1218,
      "step": 312370
    },
    {
      "epoch": 0.1768870246688466,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013752420000000002,
      "loss": 3.1591,
      "step": 312380
    },
    {
      "epoch": 0.17690566790500478,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001375222,
      "loss": 3.0821,
      "step": 312390
    },
    {
      "epoch": 0.17692431114116297,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001375202,
      "loss": 3.3018,
      "step": 312400
    },
    {
      "epoch": 0.17694295437732116,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001375182,
      "loss": 3.0886,
      "step": 312410
    },
    {
      "epoch": 0.17696159761347935,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001375162,
      "loss": 3.2786,
      "step": 312420
    },
    {
      "epoch": 0.1769802408496375,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013751420000000003,
      "loss": 3.288,
      "step": 312430
    },
    {
      "epoch": 0.1769988840857957,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001375122,
      "loss": 3.1157,
      "step": 312440
    },
    {
      "epoch": 0.1770175273219539,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013751020000000002,
      "loss": 3.2182,
      "step": 312450
    },
    {
      "epoch": 0.17703617055811208,
      "grad_norm": 0.90625,
      "learning_rate": 0.00013750819999999999,
      "loss": 3.183,
      "step": 312460
    },
    {
      "epoch": 0.17705481379427027,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001375062,
      "loss": 3.1622,
      "step": 312470
    },
    {
      "epoch": 0.17707345703042845,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001375042,
      "loss": 3.0996,
      "step": 312480
    },
    {
      "epoch": 0.17709210026658662,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001375022,
      "loss": 3.1856,
      "step": 312490
    },
    {
      "epoch": 0.1771107435027448,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013750020000000002,
      "loss": 3.1032,
      "step": 312500
    },
    {
      "epoch": 0.177129386738903,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001374982,
      "loss": 3.0707,
      "step": 312510
    },
    {
      "epoch": 0.17714802997506118,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001374962,
      "loss": 3.1089,
      "step": 312520
    },
    {
      "epoch": 0.17716667321121937,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001374942,
      "loss": 3.0351,
      "step": 312530
    },
    {
      "epoch": 0.17718531644737753,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001374922,
      "loss": 3.1147,
      "step": 312540
    },
    {
      "epoch": 0.17720395968353572,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001374902,
      "loss": 3.0305,
      "step": 312550
    },
    {
      "epoch": 0.1772226029196939,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001374882,
      "loss": 3.2127,
      "step": 312560
    },
    {
      "epoch": 0.1772412461558521,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013748620000000002,
      "loss": 3.0553,
      "step": 312570
    },
    {
      "epoch": 0.1772598893920103,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001374842,
      "loss": 3.0836,
      "step": 312580
    },
    {
      "epoch": 0.17727853262816845,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001374822,
      "loss": 3.0709,
      "step": 312590
    },
    {
      "epoch": 0.17729717586432664,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001374802,
      "loss": 3.1556,
      "step": 312600
    },
    {
      "epoch": 0.17731581910048483,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001374782,
      "loss": 3.1614,
      "step": 312610
    },
    {
      "epoch": 0.17733446233664302,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013747620000000002,
      "loss": 3.0302,
      "step": 312620
    },
    {
      "epoch": 0.1773531055728012,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001374742,
      "loss": 3.0137,
      "step": 312630
    },
    {
      "epoch": 0.17737174880895937,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00013747220000000002,
      "loss": 3.1937,
      "step": 312640
    },
    {
      "epoch": 0.17739039204511756,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001374702,
      "loss": 3.1669,
      "step": 312650
    },
    {
      "epoch": 0.17740903528127575,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001374682,
      "loss": 3.2414,
      "step": 312660
    },
    {
      "epoch": 0.17742767851743393,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001374662,
      "loss": 3.214,
      "step": 312670
    },
    {
      "epoch": 0.17744632175359212,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001374642,
      "loss": 3.2635,
      "step": 312680
    },
    {
      "epoch": 0.17746496498975028,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013746220000000002,
      "loss": 3.3255,
      "step": 312690
    },
    {
      "epoch": 0.17748360822590847,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013746020000000002,
      "loss": 3.0283,
      "step": 312700
    },
    {
      "epoch": 0.17750225146206666,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001374582,
      "loss": 3.1054,
      "step": 312710
    },
    {
      "epoch": 0.17752089469822485,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001374562,
      "loss": 3.0972,
      "step": 312720
    },
    {
      "epoch": 0.17753953793438304,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001374542,
      "loss": 3.0931,
      "step": 312730
    },
    {
      "epoch": 0.17755818117054123,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001374522,
      "loss": 3.1554,
      "step": 312740
    },
    {
      "epoch": 0.1775768244066994,
      "grad_norm": 0.75,
      "learning_rate": 0.0001374502,
      "loss": 2.9007,
      "step": 312750
    },
    {
      "epoch": 0.17759546764285758,
      "grad_norm": 0.75,
      "learning_rate": 0.00013744820000000002,
      "loss": 3.0126,
      "step": 312760
    },
    {
      "epoch": 0.17761411087901577,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001374462,
      "loss": 3.0874,
      "step": 312770
    },
    {
      "epoch": 0.17763275411517396,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001374442,
      "loss": 3.1638,
      "step": 312780
    },
    {
      "epoch": 0.17765139735133215,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001374422,
      "loss": 2.9815,
      "step": 312790
    },
    {
      "epoch": 0.1776700405874903,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001374402,
      "loss": 3.0937,
      "step": 312800
    },
    {
      "epoch": 0.1776886838236485,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001374382,
      "loss": 3.1712,
      "step": 312810
    },
    {
      "epoch": 0.17770732705980669,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013743620000000002,
      "loss": 3.1401,
      "step": 312820
    },
    {
      "epoch": 0.17772597029596487,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001374342,
      "loss": 2.9827,
      "step": 312830
    },
    {
      "epoch": 0.17774461353212306,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001374322,
      "loss": 3.1474,
      "step": 312840
    },
    {
      "epoch": 0.17776325676828122,
      "grad_norm": 0.75,
      "learning_rate": 0.0001374302,
      "loss": 3.2638,
      "step": 312850
    },
    {
      "epoch": 0.1777819000044394,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001374282,
      "loss": 3.0762,
      "step": 312860
    },
    {
      "epoch": 0.1778005432405976,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013742620000000002,
      "loss": 3.0596,
      "step": 312870
    },
    {
      "epoch": 0.1778191864767558,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001374242,
      "loss": 3.2084,
      "step": 312880
    },
    {
      "epoch": 0.17783782971291398,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013742220000000001,
      "loss": 3.1062,
      "step": 312890
    },
    {
      "epoch": 0.17785647294907214,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001374202,
      "loss": 3.1075,
      "step": 312900
    },
    {
      "epoch": 0.17787511618523033,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001374182,
      "loss": 3.2783,
      "step": 312910
    },
    {
      "epoch": 0.17789375942138852,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013741620000000003,
      "loss": 3.0806,
      "step": 312920
    },
    {
      "epoch": 0.1779124026575467,
      "grad_norm": 0.75,
      "learning_rate": 0.0001374142,
      "loss": 3.131,
      "step": 312930
    },
    {
      "epoch": 0.1779310458937049,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013741220000000002,
      "loss": 3.0829,
      "step": 312940
    },
    {
      "epoch": 0.1779496891298631,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001374102,
      "loss": 3.0489,
      "step": 312950
    },
    {
      "epoch": 0.17796833236602125,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001374082,
      "loss": 3.1273,
      "step": 312960
    },
    {
      "epoch": 0.17798697560217944,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001374062,
      "loss": 3.171,
      "step": 312970
    },
    {
      "epoch": 0.17800561883833763,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001374042,
      "loss": 3.1089,
      "step": 312980
    },
    {
      "epoch": 0.17802426207449581,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013740220000000002,
      "loss": 3.0212,
      "step": 312990
    },
    {
      "epoch": 0.178042905310654,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001374002,
      "loss": 3.1656,
      "step": 313000
    },
    {
      "epoch": 0.17806154854681217,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013739820000000002,
      "loss": 3.1851,
      "step": 313010
    },
    {
      "epoch": 0.17808019178297035,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013739619999999998,
      "loss": 3.2533,
      "step": 313020
    },
    {
      "epoch": 0.17809883501912854,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001373942,
      "loss": 3.0584,
      "step": 313030
    },
    {
      "epoch": 0.17811747825528673,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001373922,
      "loss": 3.0649,
      "step": 313040
    },
    {
      "epoch": 0.17813612149144492,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001373902,
      "loss": 3.0655,
      "step": 313050
    },
    {
      "epoch": 0.17815476472760308,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013738820000000002,
      "loss": 3.1016,
      "step": 313060
    },
    {
      "epoch": 0.17817340796376127,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001373862,
      "loss": 3.0853,
      "step": 313070
    },
    {
      "epoch": 0.17819205119991946,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001373842,
      "loss": 3.1401,
      "step": 313080
    },
    {
      "epoch": 0.17821069443607765,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001373822,
      "loss": 3.227,
      "step": 313090
    },
    {
      "epoch": 0.17822933767223584,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001373802,
      "loss": 3.3052,
      "step": 313100
    },
    {
      "epoch": 0.178247980908394,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013737820000000003,
      "loss": 3.2084,
      "step": 313110
    },
    {
      "epoch": 0.1782666241445522,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001373762,
      "loss": 3.1528,
      "step": 313120
    },
    {
      "epoch": 0.17828526738071038,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013737420000000002,
      "loss": 3.172,
      "step": 313130
    },
    {
      "epoch": 0.17830391061686857,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001373722,
      "loss": 3.0932,
      "step": 313140
    },
    {
      "epoch": 0.17832255385302676,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001373702,
      "loss": 3.1496,
      "step": 313150
    },
    {
      "epoch": 0.17834119708918494,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001373682,
      "loss": 3.1907,
      "step": 313160
    },
    {
      "epoch": 0.1783598403253431,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001373662,
      "loss": 3.0407,
      "step": 313170
    },
    {
      "epoch": 0.1783784835615013,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013736420000000002,
      "loss": 3.1248,
      "step": 313180
    },
    {
      "epoch": 0.17839712679765948,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001373622,
      "loss": 3.1384,
      "step": 313190
    },
    {
      "epoch": 0.17841577003381767,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001373602,
      "loss": 3.1231,
      "step": 313200
    },
    {
      "epoch": 0.17843441326997586,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001373582,
      "loss": 3.1096,
      "step": 313210
    },
    {
      "epoch": 0.17845305650613402,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001373562,
      "loss": 3.1167,
      "step": 313220
    },
    {
      "epoch": 0.1784716997422922,
      "grad_norm": 0.625,
      "learning_rate": 0.0001373542,
      "loss": 3.1356,
      "step": 313230
    },
    {
      "epoch": 0.1784903429784504,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001373522,
      "loss": 3.164,
      "step": 313240
    },
    {
      "epoch": 0.1785089862146086,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013735020000000002,
      "loss": 3.1478,
      "step": 313250
    },
    {
      "epoch": 0.17852762945076678,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013734820000000001,
      "loss": 3.0919,
      "step": 313260
    },
    {
      "epoch": 0.17854627268692494,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001373462,
      "loss": 3.1642,
      "step": 313270
    },
    {
      "epoch": 0.17856491592308313,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001373442,
      "loss": 3.0124,
      "step": 313280
    },
    {
      "epoch": 0.17858355915924132,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001373422,
      "loss": 3.071,
      "step": 313290
    },
    {
      "epoch": 0.1786022023953995,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001373402,
      "loss": 3.0816,
      "step": 313300
    },
    {
      "epoch": 0.1786208456315577,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013733820000000002,
      "loss": 3.1959,
      "step": 313310
    },
    {
      "epoch": 0.17863948886771586,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013733620000000002,
      "loss": 3.1519,
      "step": 313320
    },
    {
      "epoch": 0.17865813210387405,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001373342,
      "loss": 2.9684,
      "step": 313330
    },
    {
      "epoch": 0.17867677534003223,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001373322,
      "loss": 3.0751,
      "step": 313340
    },
    {
      "epoch": 0.17869541857619042,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001373302,
      "loss": 3.149,
      "step": 313350
    },
    {
      "epoch": 0.1787140618123486,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001373282,
      "loss": 3.1211,
      "step": 313360
    },
    {
      "epoch": 0.17873270504850677,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001373262,
      "loss": 3.1512,
      "step": 313370
    },
    {
      "epoch": 0.17875134828466496,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013732420000000002,
      "loss": 3.1142,
      "step": 313380
    },
    {
      "epoch": 0.17876999152082315,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001373222,
      "loss": 3.0406,
      "step": 313390
    },
    {
      "epoch": 0.17878863475698134,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001373202,
      "loss": 3.0681,
      "step": 313400
    },
    {
      "epoch": 0.17880727799313953,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001373182,
      "loss": 3.3223,
      "step": 313410
    },
    {
      "epoch": 0.17882592122929772,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001373162,
      "loss": 3.1398,
      "step": 313420
    },
    {
      "epoch": 0.17884456446545588,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013731420000000002,
      "loss": 3.1362,
      "step": 313430
    },
    {
      "epoch": 0.17886320770161407,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001373122,
      "loss": 2.9427,
      "step": 313440
    },
    {
      "epoch": 0.17888185093777226,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001373102,
      "loss": 3.1539,
      "step": 313450
    },
    {
      "epoch": 0.17890049417393045,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001373082,
      "loss": 3.0998,
      "step": 313460
    },
    {
      "epoch": 0.17891913741008864,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001373062,
      "loss": 3.1406,
      "step": 313470
    },
    {
      "epoch": 0.1789377806462468,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013730420000000003,
      "loss": 3.1261,
      "step": 313480
    },
    {
      "epoch": 0.17895642388240499,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001373022,
      "loss": 2.9255,
      "step": 313490
    },
    {
      "epoch": 0.17897506711856317,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013730020000000002,
      "loss": 3.2363,
      "step": 313500
    },
    {
      "epoch": 0.17899371035472136,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013729819999999999,
      "loss": 3.1819,
      "step": 313510
    },
    {
      "epoch": 0.17901235359087955,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001372962,
      "loss": 3.03,
      "step": 313520
    },
    {
      "epoch": 0.17903099682703771,
      "grad_norm": 0.75,
      "learning_rate": 0.0001372942,
      "loss": 3.1598,
      "step": 313530
    },
    {
      "epoch": 0.1790496400631959,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001372922,
      "loss": 3.164,
      "step": 313540
    },
    {
      "epoch": 0.1790682832993541,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013729020000000002,
      "loss": 3.2056,
      "step": 313550
    },
    {
      "epoch": 0.17908692653551228,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001372882,
      "loss": 3.1676,
      "step": 313560
    },
    {
      "epoch": 0.17910556977167047,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013728620000000001,
      "loss": 3.0264,
      "step": 313570
    },
    {
      "epoch": 0.17912421300782863,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001372842,
      "loss": 3.1469,
      "step": 313580
    },
    {
      "epoch": 0.17914285624398682,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001372822,
      "loss": 3.1128,
      "step": 313590
    },
    {
      "epoch": 0.179161499480145,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013728020000000003,
      "loss": 3.1558,
      "step": 313600
    },
    {
      "epoch": 0.1791801427163032,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001372782,
      "loss": 3.0987,
      "step": 313610
    },
    {
      "epoch": 0.1791987859524614,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013727620000000002,
      "loss": 3.1806,
      "step": 313620
    },
    {
      "epoch": 0.17921742918861958,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001372742,
      "loss": 3.255,
      "step": 313630
    },
    {
      "epoch": 0.17923607242477774,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001372722,
      "loss": 3.2038,
      "step": 313640
    },
    {
      "epoch": 0.17925471566093593,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001372702,
      "loss": 3.1961,
      "step": 313650
    },
    {
      "epoch": 0.17927335889709411,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001372682,
      "loss": 2.9652,
      "step": 313660
    },
    {
      "epoch": 0.1792920021332523,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013726620000000002,
      "loss": 3.0806,
      "step": 313670
    },
    {
      "epoch": 0.1793106453694105,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001372642,
      "loss": 3.1909,
      "step": 313680
    },
    {
      "epoch": 0.17932928860556865,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013726220000000002,
      "loss": 3.118,
      "step": 313690
    },
    {
      "epoch": 0.17934793184172684,
      "grad_norm": 0.75,
      "learning_rate": 0.0001372602,
      "loss": 3.0717,
      "step": 313700
    },
    {
      "epoch": 0.17936657507788503,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001372582,
      "loss": 3.0656,
      "step": 313710
    },
    {
      "epoch": 0.17938521831404322,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001372562,
      "loss": 3.1655,
      "step": 313720
    },
    {
      "epoch": 0.1794038615502014,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001372542,
      "loss": 3.1438,
      "step": 313730
    },
    {
      "epoch": 0.17942250478635957,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013725220000000002,
      "loss": 3.1325,
      "step": 313740
    },
    {
      "epoch": 0.17944114802251776,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00013725020000000002,
      "loss": 3.1376,
      "step": 313750
    },
    {
      "epoch": 0.17945979125867595,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001372482,
      "loss": 3.1248,
      "step": 313760
    },
    {
      "epoch": 0.17947843449483414,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001372462,
      "loss": 3.0372,
      "step": 313770
    },
    {
      "epoch": 0.17949707773099233,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001372442,
      "loss": 3.1342,
      "step": 313780
    },
    {
      "epoch": 0.1795157209671505,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001372422,
      "loss": 3.1455,
      "step": 313790
    },
    {
      "epoch": 0.17953436420330868,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001372402,
      "loss": 3.2028,
      "step": 313800
    },
    {
      "epoch": 0.17955300743946687,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013723820000000002,
      "loss": 3.1006,
      "step": 313810
    },
    {
      "epoch": 0.17957165067562506,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001372362,
      "loss": 3.1112,
      "step": 313820
    },
    {
      "epoch": 0.17959029391178324,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001372342,
      "loss": 3.2689,
      "step": 313830
    },
    {
      "epoch": 0.17960893714794143,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001372322,
      "loss": 3.1778,
      "step": 313840
    },
    {
      "epoch": 0.1796275803840996,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001372302,
      "loss": 3.0144,
      "step": 313850
    },
    {
      "epoch": 0.17964622362025778,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001372282,
      "loss": 3.0428,
      "step": 313860
    },
    {
      "epoch": 0.17966486685641597,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013722620000000002,
      "loss": 3.0559,
      "step": 313870
    },
    {
      "epoch": 0.17968351009257416,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001372242,
      "loss": 3.12,
      "step": 313880
    },
    {
      "epoch": 0.17970215332873235,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001372222,
      "loss": 3.0099,
      "step": 313890
    },
    {
      "epoch": 0.1797207965648905,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001372202,
      "loss": 3.3235,
      "step": 313900
    },
    {
      "epoch": 0.1797394398010487,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001372182,
      "loss": 3.1279,
      "step": 313910
    },
    {
      "epoch": 0.1797580830372069,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013721620000000002,
      "loss": 3.1587,
      "step": 313920
    },
    {
      "epoch": 0.17977672627336508,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001372142,
      "loss": 3.1036,
      "step": 313930
    },
    {
      "epoch": 0.17979536950952327,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013721220000000001,
      "loss": 2.9876,
      "step": 313940
    },
    {
      "epoch": 0.17981401274568143,
      "grad_norm": 0.75,
      "learning_rate": 0.0001372102,
      "loss": 3.0851,
      "step": 313950
    },
    {
      "epoch": 0.17983265598183962,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001372082,
      "loss": 3.1084,
      "step": 313960
    },
    {
      "epoch": 0.1798512992179978,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001372062,
      "loss": 3.0076,
      "step": 313970
    },
    {
      "epoch": 0.179869942454156,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001372042,
      "loss": 3.1131,
      "step": 313980
    },
    {
      "epoch": 0.17988858569031418,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013720220000000002,
      "loss": 3.121,
      "step": 313990
    },
    {
      "epoch": 0.17990722892647235,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001372002,
      "loss": 3.0867,
      "step": 314000
    },
    {
      "epoch": 0.17992587216263053,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001371982,
      "loss": 3.1526,
      "step": 314010
    },
    {
      "epoch": 0.17994451539878872,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001371962,
      "loss": 3.3235,
      "step": 314020
    },
    {
      "epoch": 0.1799631586349469,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001371942,
      "loss": 3.1331,
      "step": 314030
    },
    {
      "epoch": 0.1799818018711051,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013719220000000002,
      "loss": 3.0574,
      "step": 314040
    },
    {
      "epoch": 0.1800004451072633,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001371902,
      "loss": 3.0844,
      "step": 314050
    },
    {
      "epoch": 0.18001908834342145,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013718820000000002,
      "loss": 3.1611,
      "step": 314060
    },
    {
      "epoch": 0.18003773157957964,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001371862,
      "loss": 3.1526,
      "step": 314070
    },
    {
      "epoch": 0.18005637481573783,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001371842,
      "loss": 3.1487,
      "step": 314080
    },
    {
      "epoch": 0.18007501805189602,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013718220000000003,
      "loss": 3.1488,
      "step": 314090
    },
    {
      "epoch": 0.1800936612880542,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001371802,
      "loss": 2.9902,
      "step": 314100
    },
    {
      "epoch": 0.18011230452421237,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013717820000000002,
      "loss": 3.1806,
      "step": 314110
    },
    {
      "epoch": 0.18013094776037056,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001371762,
      "loss": 3.1784,
      "step": 314120
    },
    {
      "epoch": 0.18014959099652875,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001371742,
      "loss": 3.0828,
      "step": 314130
    },
    {
      "epoch": 0.18016823423268694,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001371722,
      "loss": 3.2223,
      "step": 314140
    },
    {
      "epoch": 0.18018687746884512,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001371702,
      "loss": 3.1176,
      "step": 314150
    },
    {
      "epoch": 0.18020552070500329,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013716820000000003,
      "loss": 3.1534,
      "step": 314160
    },
    {
      "epoch": 0.18022416394116147,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001371662,
      "loss": 3.1027,
      "step": 314170
    },
    {
      "epoch": 0.18024280717731966,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013716420000000002,
      "loss": 3.1194,
      "step": 314180
    },
    {
      "epoch": 0.18026145041347785,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013716219999999999,
      "loss": 3.136,
      "step": 314190
    },
    {
      "epoch": 0.18028009364963604,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001371602,
      "loss": 3.208,
      "step": 314200
    },
    {
      "epoch": 0.1802987368857942,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001371582,
      "loss": 2.9915,
      "step": 314210
    },
    {
      "epoch": 0.1803173801219524,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001371562,
      "loss": 3.1909,
      "step": 314220
    },
    {
      "epoch": 0.18033602335811058,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013715420000000002,
      "loss": 3.0344,
      "step": 314230
    },
    {
      "epoch": 0.18035466659426877,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001371522,
      "loss": 3.0523,
      "step": 314240
    },
    {
      "epoch": 0.18037330983042696,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013715020000000001,
      "loss": 3.0441,
      "step": 314250
    },
    {
      "epoch": 0.18039195306658512,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001371482,
      "loss": 3.2303,
      "step": 314260
    },
    {
      "epoch": 0.1804105963027433,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001371462,
      "loss": 3.2104,
      "step": 314270
    },
    {
      "epoch": 0.1804292395389015,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001371442,
      "loss": 3.0342,
      "step": 314280
    },
    {
      "epoch": 0.1804478827750597,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001371422,
      "loss": 3.1878,
      "step": 314290
    },
    {
      "epoch": 0.18046652601121788,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013714020000000002,
      "loss": 3.0031,
      "step": 314300
    },
    {
      "epoch": 0.18048516924737606,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013713820000000001,
      "loss": 3.1632,
      "step": 314310
    },
    {
      "epoch": 0.18050381248353423,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001371362,
      "loss": 3.0701,
      "step": 314320
    },
    {
      "epoch": 0.18052245571969242,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001371342,
      "loss": 3.1622,
      "step": 314330
    },
    {
      "epoch": 0.1805410989558506,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001371322,
      "loss": 3.1648,
      "step": 314340
    },
    {
      "epoch": 0.1805597421920088,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001371302,
      "loss": 3.2525,
      "step": 314350
    },
    {
      "epoch": 0.18057838542816698,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001371282,
      "loss": 3.171,
      "step": 314360
    },
    {
      "epoch": 0.18059702866432514,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013712620000000002,
      "loss": 3.1739,
      "step": 314370
    },
    {
      "epoch": 0.18061567190048333,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001371242,
      "loss": 3.1382,
      "step": 314380
    },
    {
      "epoch": 0.18063431513664152,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001371222,
      "loss": 3.1049,
      "step": 314390
    },
    {
      "epoch": 0.1806529583727997,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001371202,
      "loss": 3.156,
      "step": 314400
    },
    {
      "epoch": 0.1806716016089579,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001371182,
      "loss": 3.2371,
      "step": 314410
    },
    {
      "epoch": 0.18069024484511606,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001371162,
      "loss": 3.0683,
      "step": 314420
    },
    {
      "epoch": 0.18070888808127425,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013711420000000002,
      "loss": 3.179,
      "step": 314430
    },
    {
      "epoch": 0.18072753131743244,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001371122,
      "loss": 3.1973,
      "step": 314440
    },
    {
      "epoch": 0.18074617455359063,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001371102,
      "loss": 3.1541,
      "step": 314450
    },
    {
      "epoch": 0.18076481778974882,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001371082,
      "loss": 3.1831,
      "step": 314460
    },
    {
      "epoch": 0.18078346102590698,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001371062,
      "loss": 3.0553,
      "step": 314470
    },
    {
      "epoch": 0.18080210426206517,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013710420000000002,
      "loss": 3.1323,
      "step": 314480
    },
    {
      "epoch": 0.18082074749822336,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001371022,
      "loss": 3.1697,
      "step": 314490
    },
    {
      "epoch": 0.18083939073438154,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001371002,
      "loss": 3.0109,
      "step": 314500
    },
    {
      "epoch": 0.18085803397053973,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001370982,
      "loss": 3.0719,
      "step": 314510
    },
    {
      "epoch": 0.18087667720669792,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001370962,
      "loss": 3.1887,
      "step": 314520
    },
    {
      "epoch": 0.18089532044285608,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013709420000000003,
      "loss": 3.09,
      "step": 314530
    },
    {
      "epoch": 0.18091396367901427,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001370922,
      "loss": 3.1253,
      "step": 314540
    },
    {
      "epoch": 0.18093260691517246,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013709020000000002,
      "loss": 3.1049,
      "step": 314550
    },
    {
      "epoch": 0.18095125015133065,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001370882,
      "loss": 3.155,
      "step": 314560
    },
    {
      "epoch": 0.18096989338748884,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001370862,
      "loss": 3.2057,
      "step": 314570
    },
    {
      "epoch": 0.180988536623647,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001370842,
      "loss": 3.0759,
      "step": 314580
    },
    {
      "epoch": 0.1810071798598052,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001370822,
      "loss": 3.1357,
      "step": 314590
    },
    {
      "epoch": 0.18102582309596338,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00013708020000000002,
      "loss": 3.0999,
      "step": 314600
    },
    {
      "epoch": 0.18104446633212157,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001370782,
      "loss": 3.0551,
      "step": 314610
    },
    {
      "epoch": 0.18106310956827976,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013707620000000001,
      "loss": 3.0452,
      "step": 314620
    },
    {
      "epoch": 0.18108175280443792,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001370742,
      "loss": 3.1705,
      "step": 314630
    },
    {
      "epoch": 0.1811003960405961,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001370722,
      "loss": 3.0096,
      "step": 314640
    },
    {
      "epoch": 0.1811190392767543,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013707020000000003,
      "loss": 3.1569,
      "step": 314650
    },
    {
      "epoch": 0.18113768251291248,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001370682,
      "loss": 3.1566,
      "step": 314660
    },
    {
      "epoch": 0.18115632574907067,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013706620000000002,
      "loss": 3.0651,
      "step": 314670
    },
    {
      "epoch": 0.18117496898522883,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001370642,
      "loss": 3.1546,
      "step": 314680
    },
    {
      "epoch": 0.18119361222138702,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001370622,
      "loss": 3.0868,
      "step": 314690
    },
    {
      "epoch": 0.1812122554575452,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001370602,
      "loss": 3.1223,
      "step": 314700
    },
    {
      "epoch": 0.1812308986937034,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001370582,
      "loss": 3.1666,
      "step": 314710
    },
    {
      "epoch": 0.1812495419298616,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013705620000000002,
      "loss": 3.0222,
      "step": 314720
    },
    {
      "epoch": 0.18126818516601978,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001370542,
      "loss": 3.105,
      "step": 314730
    },
    {
      "epoch": 0.18128682840217794,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013705220000000002,
      "loss": 3.2339,
      "step": 314740
    },
    {
      "epoch": 0.18130547163833613,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001370502,
      "loss": 3.1831,
      "step": 314750
    },
    {
      "epoch": 0.18132411487449432,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001370482,
      "loss": 3.1771,
      "step": 314760
    },
    {
      "epoch": 0.1813427581106525,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001370462,
      "loss": 3.1103,
      "step": 314770
    },
    {
      "epoch": 0.1813614013468107,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001370442,
      "loss": 2.9932,
      "step": 314780
    },
    {
      "epoch": 0.18138004458296886,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013704220000000002,
      "loss": 3.081,
      "step": 314790
    },
    {
      "epoch": 0.18139868781912705,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001370402,
      "loss": 3.216,
      "step": 314800
    },
    {
      "epoch": 0.18141733105528524,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001370382,
      "loss": 3.044,
      "step": 314810
    },
    {
      "epoch": 0.18143597429144342,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001370362,
      "loss": 2.9913,
      "step": 314820
    },
    {
      "epoch": 0.1814546175276016,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001370342,
      "loss": 3.0397,
      "step": 314830
    },
    {
      "epoch": 0.18147326076375977,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001370322,
      "loss": 3.1615,
      "step": 314840
    },
    {
      "epoch": 0.18149190399991796,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001370302,
      "loss": 3.0028,
      "step": 314850
    },
    {
      "epoch": 0.18151054723607615,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013702820000000002,
      "loss": 3.2317,
      "step": 314860
    },
    {
      "epoch": 0.18152919047223434,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001370262,
      "loss": 3.0975,
      "step": 314870
    },
    {
      "epoch": 0.18154783370839253,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001370242,
      "loss": 3.0951,
      "step": 314880
    },
    {
      "epoch": 0.1815664769445507,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001370222,
      "loss": 3.0576,
      "step": 314890
    },
    {
      "epoch": 0.18158512018070888,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001370202,
      "loss": 3.0252,
      "step": 314900
    },
    {
      "epoch": 0.18160376341686707,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001370182,
      "loss": 3.1416,
      "step": 314910
    },
    {
      "epoch": 0.18162240665302526,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013701620000000002,
      "loss": 2.9985,
      "step": 314920
    },
    {
      "epoch": 0.18164104988918345,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013701420000000001,
      "loss": 3.0647,
      "step": 314930
    },
    {
      "epoch": 0.1816596931253416,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001370122,
      "loss": 3.1995,
      "step": 314940
    },
    {
      "epoch": 0.1816783363614998,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001370102,
      "loss": 3.1443,
      "step": 314950
    },
    {
      "epoch": 0.181696979597658,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001370082,
      "loss": 3.0766,
      "step": 314960
    },
    {
      "epoch": 0.18171562283381618,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001370062,
      "loss": 3.22,
      "step": 314970
    },
    {
      "epoch": 0.18173426606997437,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001370042,
      "loss": 3.2172,
      "step": 314980
    },
    {
      "epoch": 0.18175290930613255,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013700220000000001,
      "loss": 3.1693,
      "step": 314990
    },
    {
      "epoch": 0.18177155254229072,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001370002,
      "loss": 3.3048,
      "step": 315000
    },
    {
      "epoch": 0.1817901957784489,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001369982,
      "loss": 3.0995,
      "step": 315010
    },
    {
      "epoch": 0.1818088390146071,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001369962,
      "loss": 3.1476,
      "step": 315020
    },
    {
      "epoch": 0.18182748225076528,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001369942,
      "loss": 3.258,
      "step": 315030
    },
    {
      "epoch": 0.18184612548692347,
      "grad_norm": 0.75,
      "learning_rate": 0.00013699220000000002,
      "loss": 3.2152,
      "step": 315040
    },
    {
      "epoch": 0.18186476872308163,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013699020000000002,
      "loss": 3.1128,
      "step": 315050
    },
    {
      "epoch": 0.18188341195923982,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001369882,
      "loss": 3.0967,
      "step": 315060
    },
    {
      "epoch": 0.181902055195398,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001369862,
      "loss": 3.0499,
      "step": 315070
    },
    {
      "epoch": 0.1819206984315562,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001369842,
      "loss": 3.0438,
      "step": 315080
    },
    {
      "epoch": 0.1819393416677144,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013698220000000002,
      "loss": 3.164,
      "step": 315090
    },
    {
      "epoch": 0.18195798490387255,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001369802,
      "loss": 3.2044,
      "step": 315100
    },
    {
      "epoch": 0.18197662814003074,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013697820000000002,
      "loss": 2.99,
      "step": 315110
    },
    {
      "epoch": 0.18199527137618893,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001369762,
      "loss": 3.1201,
      "step": 315120
    },
    {
      "epoch": 0.18201391461234712,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001369742,
      "loss": 3.1356,
      "step": 315130
    },
    {
      "epoch": 0.1820325578485053,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013697220000000003,
      "loss": 3.1582,
      "step": 315140
    },
    {
      "epoch": 0.18205120108466347,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001369702,
      "loss": 3.0635,
      "step": 315150
    },
    {
      "epoch": 0.18206984432082166,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013696820000000002,
      "loss": 3.2144,
      "step": 315160
    },
    {
      "epoch": 0.18208848755697984,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001369662,
      "loss": 3.2698,
      "step": 315170
    },
    {
      "epoch": 0.18210713079313803,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001369642,
      "loss": 3.1288,
      "step": 315180
    },
    {
      "epoch": 0.18212577402929622,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001369622,
      "loss": 3.1847,
      "step": 315190
    },
    {
      "epoch": 0.1821444172654544,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001369602,
      "loss": 3.1851,
      "step": 315200
    },
    {
      "epoch": 0.18216306050161257,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013695820000000003,
      "loss": 3.1895,
      "step": 315210
    },
    {
      "epoch": 0.18218170373777076,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001369562,
      "loss": 3.0644,
      "step": 315220
    },
    {
      "epoch": 0.18220034697392895,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013695420000000002,
      "loss": 3.14,
      "step": 315230
    },
    {
      "epoch": 0.18221899021008714,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013695219999999999,
      "loss": 3.1074,
      "step": 315240
    },
    {
      "epoch": 0.18223763344624533,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001369502,
      "loss": 3.1259,
      "step": 315250
    },
    {
      "epoch": 0.1822562766824035,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001369482,
      "loss": 3.1733,
      "step": 315260
    },
    {
      "epoch": 0.18227491991856168,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001369462,
      "loss": 3.1248,
      "step": 315270
    },
    {
      "epoch": 0.18229356315471987,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013694420000000002,
      "loss": 3.182,
      "step": 315280
    },
    {
      "epoch": 0.18231220639087806,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001369422,
      "loss": 3.1682,
      "step": 315290
    },
    {
      "epoch": 0.18233084962703625,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013694020000000001,
      "loss": 3.316,
      "step": 315300
    },
    {
      "epoch": 0.1823494928631944,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001369382,
      "loss": 3.1243,
      "step": 315310
    },
    {
      "epoch": 0.1823681360993526,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001369362,
      "loss": 3.0047,
      "step": 315320
    },
    {
      "epoch": 0.18238677933551078,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001369342,
      "loss": 3.2008,
      "step": 315330
    },
    {
      "epoch": 0.18240542257166897,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001369322,
      "loss": 3.2173,
      "step": 315340
    },
    {
      "epoch": 0.18242406580782716,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013693020000000002,
      "loss": 3.1243,
      "step": 315350
    },
    {
      "epoch": 0.18244270904398532,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013692820000000001,
      "loss": 3.2607,
      "step": 315360
    },
    {
      "epoch": 0.1824613522801435,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001369262,
      "loss": 3.0929,
      "step": 315370
    },
    {
      "epoch": 0.1824799955163017,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001369242,
      "loss": 3.1361,
      "step": 315380
    },
    {
      "epoch": 0.1824986387524599,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001369222,
      "loss": 3.1879,
      "step": 315390
    },
    {
      "epoch": 0.18251728198861808,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001369202,
      "loss": 3.2158,
      "step": 315400
    },
    {
      "epoch": 0.18253592522477627,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001369182,
      "loss": 3.0584,
      "step": 315410
    },
    {
      "epoch": 0.18255456846093443,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013691620000000002,
      "loss": 3.15,
      "step": 315420
    },
    {
      "epoch": 0.18257321169709262,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001369142,
      "loss": 3.0441,
      "step": 315430
    },
    {
      "epoch": 0.1825918549332508,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001369122,
      "loss": 3.1365,
      "step": 315440
    },
    {
      "epoch": 0.182610498169409,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001369102,
      "loss": 3.013,
      "step": 315450
    },
    {
      "epoch": 0.18262914140556719,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001369082,
      "loss": 2.934,
      "step": 315460
    },
    {
      "epoch": 0.18264778464172535,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001369062,
      "loss": 3.231,
      "step": 315470
    },
    {
      "epoch": 0.18266642787788354,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013690420000000002,
      "loss": 3.0862,
      "step": 315480
    },
    {
      "epoch": 0.18268507111404172,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001369022,
      "loss": 3.0775,
      "step": 315490
    },
    {
      "epoch": 0.18270371435019991,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001369002,
      "loss": 3.1195,
      "step": 315500
    },
    {
      "epoch": 0.1827223575863581,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001368982,
      "loss": 3.0972,
      "step": 315510
    },
    {
      "epoch": 0.18274100082251626,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001368962,
      "loss": 3.3097,
      "step": 315520
    },
    {
      "epoch": 0.18275964405867445,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013689420000000002,
      "loss": 3.0959,
      "step": 315530
    },
    {
      "epoch": 0.18277828729483264,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013689220000000002,
      "loss": 3.1977,
      "step": 315540
    },
    {
      "epoch": 0.18279693053099083,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001368902,
      "loss": 3.0073,
      "step": 315550
    },
    {
      "epoch": 0.18281557376714902,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001368882,
      "loss": 3.1842,
      "step": 315560
    },
    {
      "epoch": 0.18283421700330718,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001368862,
      "loss": 3.0008,
      "step": 315570
    },
    {
      "epoch": 0.18285286023946537,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001368842,
      "loss": 3.1213,
      "step": 315580
    },
    {
      "epoch": 0.18287150347562356,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001368822,
      "loss": 3.0508,
      "step": 315590
    },
    {
      "epoch": 0.18289014671178175,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013688020000000002,
      "loss": 3.1126,
      "step": 315600
    },
    {
      "epoch": 0.18290878994793994,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013687820000000001,
      "loss": 3.1156,
      "step": 315610
    },
    {
      "epoch": 0.1829274331840981,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001368762,
      "loss": 3.08,
      "step": 315620
    },
    {
      "epoch": 0.1829460764202563,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001368742,
      "loss": 3.037,
      "step": 315630
    },
    {
      "epoch": 0.18296471965641448,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001368722,
      "loss": 3.1286,
      "step": 315640
    },
    {
      "epoch": 0.18298336289257267,
      "grad_norm": 0.75,
      "learning_rate": 0.00013687020000000002,
      "loss": 3.0894,
      "step": 315650
    },
    {
      "epoch": 0.18300200612873085,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001368682,
      "loss": 3.1039,
      "step": 315660
    },
    {
      "epoch": 0.18302064936488904,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013686620000000001,
      "loss": 3.143,
      "step": 315670
    },
    {
      "epoch": 0.1830392926010472,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001368642,
      "loss": 3.1407,
      "step": 315680
    },
    {
      "epoch": 0.1830579358372054,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001368622,
      "loss": 3.1962,
      "step": 315690
    },
    {
      "epoch": 0.18307657907336358,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013686020000000003,
      "loss": 3.1485,
      "step": 315700
    },
    {
      "epoch": 0.18309522230952177,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001368582,
      "loss": 3.0642,
      "step": 315710
    },
    {
      "epoch": 0.18311386554567996,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013685620000000002,
      "loss": 3.1499,
      "step": 315720
    },
    {
      "epoch": 0.18313250878183812,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001368542,
      "loss": 3.1817,
      "step": 315730
    },
    {
      "epoch": 0.1831511520179963,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001368522,
      "loss": 3.2324,
      "step": 315740
    },
    {
      "epoch": 0.1831697952541545,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001368502,
      "loss": 3.0481,
      "step": 315750
    },
    {
      "epoch": 0.1831884384903127,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001368482,
      "loss": 3.1293,
      "step": 315760
    },
    {
      "epoch": 0.18320708172647088,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013684620000000002,
      "loss": 3.111,
      "step": 315770
    },
    {
      "epoch": 0.18322572496262904,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001368442,
      "loss": 3.1801,
      "step": 315780
    },
    {
      "epoch": 0.18324436819878723,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013684220000000002,
      "loss": 3.1258,
      "step": 315790
    },
    {
      "epoch": 0.18326301143494542,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013684019999999998,
      "loss": 2.996,
      "step": 315800
    },
    {
      "epoch": 0.1832816546711036,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001368382,
      "loss": 3.0759,
      "step": 315810
    },
    {
      "epoch": 0.1833002979072618,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001368362,
      "loss": 3.258,
      "step": 315820
    },
    {
      "epoch": 0.18331894114341996,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001368342,
      "loss": 3.2202,
      "step": 315830
    },
    {
      "epoch": 0.18333758437957814,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013683220000000002,
      "loss": 3.1313,
      "step": 315840
    },
    {
      "epoch": 0.18335622761573633,
      "grad_norm": 0.75,
      "learning_rate": 0.0001368302,
      "loss": 3.1558,
      "step": 315850
    },
    {
      "epoch": 0.18337487085189452,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001368282,
      "loss": 3.0366,
      "step": 315860
    },
    {
      "epoch": 0.1833935140880527,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001368262,
      "loss": 3.0859,
      "step": 315870
    },
    {
      "epoch": 0.1834121573242109,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001368242,
      "loss": 3.2219,
      "step": 315880
    },
    {
      "epoch": 0.18343080056036906,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001368222,
      "loss": 3.0889,
      "step": 315890
    },
    {
      "epoch": 0.18344944379652725,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001368202,
      "loss": 3.1331,
      "step": 315900
    },
    {
      "epoch": 0.18346808703268544,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013681820000000002,
      "loss": 3.0748,
      "step": 315910
    },
    {
      "epoch": 0.18348673026884363,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001368162,
      "loss": 3.2412,
      "step": 315920
    },
    {
      "epoch": 0.18350537350500182,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001368142,
      "loss": 3.1301,
      "step": 315930
    },
    {
      "epoch": 0.18352401674115998,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001368122,
      "loss": 3.0993,
      "step": 315940
    },
    {
      "epoch": 0.18354265997731817,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001368102,
      "loss": 3.1494,
      "step": 315950
    },
    {
      "epoch": 0.18356130321347636,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001368082,
      "loss": 3.1201,
      "step": 315960
    },
    {
      "epoch": 0.18357994644963455,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001368062,
      "loss": 3.2312,
      "step": 315970
    },
    {
      "epoch": 0.18359858968579273,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00013680420000000001,
      "loss": 3.2025,
      "step": 315980
    },
    {
      "epoch": 0.1836172329219509,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001368022,
      "loss": 3.0167,
      "step": 315990
    },
    {
      "epoch": 0.18363587615810908,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001368002,
      "loss": 3.0731,
      "step": 316000
    },
    {
      "epoch": 0.18365451939426727,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001367982,
      "loss": 3.106,
      "step": 316010
    },
    {
      "epoch": 0.18367316263042546,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001367962,
      "loss": 3.2549,
      "step": 316020
    },
    {
      "epoch": 0.18369180586658365,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013679420000000002,
      "loss": 2.9951,
      "step": 316030
    },
    {
      "epoch": 0.1837104491027418,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013679220000000001,
      "loss": 3.2511,
      "step": 316040
    },
    {
      "epoch": 0.1837290923389,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001367902,
      "loss": 3.0669,
      "step": 316050
    },
    {
      "epoch": 0.1837477355750582,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001367882,
      "loss": 3.1544,
      "step": 316060
    },
    {
      "epoch": 0.18376637881121638,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001367862,
      "loss": 3.1523,
      "step": 316070
    },
    {
      "epoch": 0.18378502204737457,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001367842,
      "loss": 3.1695,
      "step": 316080
    },
    {
      "epoch": 0.18380366528353276,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013678220000000002,
      "loss": 3.0434,
      "step": 316090
    },
    {
      "epoch": 0.18382230851969092,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013678020000000002,
      "loss": 3.1743,
      "step": 316100
    },
    {
      "epoch": 0.1838409517558491,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001367782,
      "loss": 3.0634,
      "step": 316110
    },
    {
      "epoch": 0.1838595949920073,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001367762,
      "loss": 3.1349,
      "step": 316120
    },
    {
      "epoch": 0.18387823822816549,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001367742,
      "loss": 3.0627,
      "step": 316130
    },
    {
      "epoch": 0.18389688146432367,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013677220000000003,
      "loss": 3.1644,
      "step": 316140
    },
    {
      "epoch": 0.18391552470048184,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001367702,
      "loss": 3.218,
      "step": 316150
    },
    {
      "epoch": 0.18393416793664003,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013676820000000002,
      "loss": 3.0428,
      "step": 316160
    },
    {
      "epoch": 0.18395281117279821,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001367662,
      "loss": 3.0959,
      "step": 316170
    },
    {
      "epoch": 0.1839714544089564,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001367642,
      "loss": 2.9462,
      "step": 316180
    },
    {
      "epoch": 0.1839900976451146,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001367622,
      "loss": 3.2602,
      "step": 316190
    },
    {
      "epoch": 0.18400874088127275,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001367602,
      "loss": 3.1661,
      "step": 316200
    },
    {
      "epoch": 0.18402738411743094,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013675820000000002,
      "loss": 3.1101,
      "step": 316210
    },
    {
      "epoch": 0.18404602735358913,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001367562,
      "loss": 3.1218,
      "step": 316220
    },
    {
      "epoch": 0.18406467058974732,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001367542,
      "loss": 3.1275,
      "step": 316230
    },
    {
      "epoch": 0.1840833138259055,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001367522,
      "loss": 3.11,
      "step": 316240
    },
    {
      "epoch": 0.18410195706206367,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001367502,
      "loss": 3.1104,
      "step": 316250
    },
    {
      "epoch": 0.18412060029822186,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013674820000000003,
      "loss": 3.1522,
      "step": 316260
    },
    {
      "epoch": 0.18413924353438005,
      "grad_norm": 0.75,
      "learning_rate": 0.0001367462,
      "loss": 2.9704,
      "step": 316270
    },
    {
      "epoch": 0.18415788677053824,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013674420000000002,
      "loss": 3.068,
      "step": 316280
    },
    {
      "epoch": 0.18417653000669643,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001367422,
      "loss": 3.0789,
      "step": 316290
    },
    {
      "epoch": 0.1841951732428546,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001367402,
      "loss": 3.1611,
      "step": 316300
    },
    {
      "epoch": 0.18421381647901278,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001367382,
      "loss": 3.1513,
      "step": 316310
    },
    {
      "epoch": 0.18423245971517097,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001367362,
      "loss": 3.1847,
      "step": 316320
    },
    {
      "epoch": 0.18425110295132915,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013673420000000002,
      "loss": 3.0756,
      "step": 316330
    },
    {
      "epoch": 0.18426974618748734,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001367322,
      "loss": 3.152,
      "step": 316340
    },
    {
      "epoch": 0.18428838942364553,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013673020000000001,
      "loss": 3.2254,
      "step": 316350
    },
    {
      "epoch": 0.1843070326598037,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001367282,
      "loss": 3.2075,
      "step": 316360
    },
    {
      "epoch": 0.18432567589596188,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001367262,
      "loss": 3.0821,
      "step": 316370
    },
    {
      "epoch": 0.18434431913212007,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001367242,
      "loss": 3.1597,
      "step": 316380
    },
    {
      "epoch": 0.18436296236827826,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001367222,
      "loss": 3.0869,
      "step": 316390
    },
    {
      "epoch": 0.18438160560443645,
      "grad_norm": 0.84375,
      "learning_rate": 0.00013672020000000002,
      "loss": 3.1328,
      "step": 316400
    },
    {
      "epoch": 0.1844002488405946,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001367182,
      "loss": 3.1209,
      "step": 316410
    },
    {
      "epoch": 0.1844188920767528,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001367162,
      "loss": 3.0706,
      "step": 316420
    },
    {
      "epoch": 0.184437535312911,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001367142,
      "loss": 3.0964,
      "step": 316430
    },
    {
      "epoch": 0.18445617854906918,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001367122,
      "loss": 3.0879,
      "step": 316440
    },
    {
      "epoch": 0.18447482178522737,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001367102,
      "loss": 3.1932,
      "step": 316450
    },
    {
      "epoch": 0.18449346502138553,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001367082,
      "loss": 3.1738,
      "step": 316460
    },
    {
      "epoch": 0.18451210825754372,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013670620000000002,
      "loss": 3.0779,
      "step": 316470
    },
    {
      "epoch": 0.1845307514937019,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001367042,
      "loss": 3.0309,
      "step": 316480
    },
    {
      "epoch": 0.1845493947298601,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001367022,
      "loss": 3.1223,
      "step": 316490
    },
    {
      "epoch": 0.18456803796601828,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001367002,
      "loss": 3.2613,
      "step": 316500
    },
    {
      "epoch": 0.18458668120217644,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001366982,
      "loss": 3.014,
      "step": 316510
    },
    {
      "epoch": 0.18460532443833463,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001366962,
      "loss": 3.2162,
      "step": 316520
    },
    {
      "epoch": 0.18462396767449282,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013669420000000002,
      "loss": 3.1767,
      "step": 316530
    },
    {
      "epoch": 0.184642610910651,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001366922,
      "loss": 3.1114,
      "step": 316540
    },
    {
      "epoch": 0.1846612541468092,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001366902,
      "loss": 3.168,
      "step": 316550
    },
    {
      "epoch": 0.1846798973829674,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001366882,
      "loss": 3.0668,
      "step": 316560
    },
    {
      "epoch": 0.18469854061912555,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001366862,
      "loss": 3.1999,
      "step": 316570
    },
    {
      "epoch": 0.18471718385528374,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001366842,
      "loss": 3.0737,
      "step": 316580
    },
    {
      "epoch": 0.18473582709144193,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013668220000000002,
      "loss": 3.3151,
      "step": 316590
    },
    {
      "epoch": 0.18475447032760012,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001366802,
      "loss": 3.2255,
      "step": 316600
    },
    {
      "epoch": 0.1847731135637583,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001366782,
      "loss": 3.2142,
      "step": 316610
    },
    {
      "epoch": 0.18479175679991647,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001366762,
      "loss": 3.0611,
      "step": 316620
    },
    {
      "epoch": 0.18481040003607466,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001366742,
      "loss": 3.0259,
      "step": 316630
    },
    {
      "epoch": 0.18482904327223285,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001366722,
      "loss": 3.1842,
      "step": 316640
    },
    {
      "epoch": 0.18484768650839103,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013667020000000002,
      "loss": 3.0655,
      "step": 316650
    },
    {
      "epoch": 0.18486632974454922,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013666820000000001,
      "loss": 3.1167,
      "step": 316660
    },
    {
      "epoch": 0.18488497298070738,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001366662,
      "loss": 3.0838,
      "step": 316670
    },
    {
      "epoch": 0.18490361621686557,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001366642,
      "loss": 3.1898,
      "step": 316680
    },
    {
      "epoch": 0.18492225945302376,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001366622,
      "loss": 3.0511,
      "step": 316690
    },
    {
      "epoch": 0.18494090268918195,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013666020000000002,
      "loss": 3.1059,
      "step": 316700
    },
    {
      "epoch": 0.18495954592534014,
      "grad_norm": 0.75,
      "learning_rate": 0.0001366582,
      "loss": 3.0828,
      "step": 316710
    },
    {
      "epoch": 0.1849781891614983,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013665620000000001,
      "loss": 2.9937,
      "step": 316720
    },
    {
      "epoch": 0.1849968323976565,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001366542,
      "loss": 3.0576,
      "step": 316730
    },
    {
      "epoch": 0.18501547563381468,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001366522,
      "loss": 3.0561,
      "step": 316740
    },
    {
      "epoch": 0.18503411886997287,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013665020000000003,
      "loss": 3.1078,
      "step": 316750
    },
    {
      "epoch": 0.18505276210613106,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001366482,
      "loss": 3.1919,
      "step": 316760
    },
    {
      "epoch": 0.18507140534228925,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013664620000000002,
      "loss": 3.3107,
      "step": 316770
    },
    {
      "epoch": 0.1850900485784474,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001366442,
      "loss": 3.1644,
      "step": 316780
    },
    {
      "epoch": 0.1851086918146056,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001366422,
      "loss": 3.1062,
      "step": 316790
    },
    {
      "epoch": 0.18512733505076379,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001366402,
      "loss": 3.2207,
      "step": 316800
    },
    {
      "epoch": 0.18514597828692198,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001366382,
      "loss": 3.1809,
      "step": 316810
    },
    {
      "epoch": 0.18516462152308016,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013663620000000003,
      "loss": 3.1885,
      "step": 316820
    },
    {
      "epoch": 0.18518326475923833,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001366342,
      "loss": 3.0856,
      "step": 316830
    },
    {
      "epoch": 0.18520190799539651,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00013663220000000002,
      "loss": 3.0748,
      "step": 316840
    },
    {
      "epoch": 0.1852205512315547,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013663019999999998,
      "loss": 3.284,
      "step": 316850
    },
    {
      "epoch": 0.1852391944677129,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001366282,
      "loss": 3.0931,
      "step": 316860
    },
    {
      "epoch": 0.18525783770387108,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001366262,
      "loss": 3.1059,
      "step": 316870
    },
    {
      "epoch": 0.18527648094002924,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001366242,
      "loss": 3.155,
      "step": 316880
    },
    {
      "epoch": 0.18529512417618743,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013662220000000002,
      "loss": 3.2118,
      "step": 316890
    },
    {
      "epoch": 0.18531376741234562,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001366202,
      "loss": 2.9965,
      "step": 316900
    },
    {
      "epoch": 0.1853324106485038,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001366182,
      "loss": 3.1477,
      "step": 316910
    },
    {
      "epoch": 0.185351053884662,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001366162,
      "loss": 3.1326,
      "step": 316920
    },
    {
      "epoch": 0.18536969712082016,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001366142,
      "loss": 3.0097,
      "step": 316930
    },
    {
      "epoch": 0.18538834035697835,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001366122,
      "loss": 3.0685,
      "step": 316940
    },
    {
      "epoch": 0.18540698359313654,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001366102,
      "loss": 3.0813,
      "step": 316950
    },
    {
      "epoch": 0.18542562682929473,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013660820000000002,
      "loss": 3.086,
      "step": 316960
    },
    {
      "epoch": 0.18544427006545292,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013660620000000001,
      "loss": 2.9455,
      "step": 316970
    },
    {
      "epoch": 0.18546291330161108,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001366042,
      "loss": 3.3129,
      "step": 316980
    },
    {
      "epoch": 0.18548155653776927,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001366022,
      "loss": 3.0998,
      "step": 316990
    },
    {
      "epoch": 0.18550019977392745,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001366002,
      "loss": 3.1538,
      "step": 317000
    },
    {
      "epoch": 0.18551884301008564,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001365982,
      "loss": 3.0173,
      "step": 317010
    },
    {
      "epoch": 0.18553748624624383,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001365962,
      "loss": 3.1376,
      "step": 317020
    },
    {
      "epoch": 0.18555612948240202,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013659420000000001,
      "loss": 3.0074,
      "step": 317030
    },
    {
      "epoch": 0.18557477271856018,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001365922,
      "loss": 3.0998,
      "step": 317040
    },
    {
      "epoch": 0.18559341595471837,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001365902,
      "loss": 3.1717,
      "step": 317050
    },
    {
      "epoch": 0.18561205919087656,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001365882,
      "loss": 3.2361,
      "step": 317060
    },
    {
      "epoch": 0.18563070242703475,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001365862,
      "loss": 3.0758,
      "step": 317070
    },
    {
      "epoch": 0.18564934566319294,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013658420000000002,
      "loss": 3.0174,
      "step": 317080
    },
    {
      "epoch": 0.1856679888993511,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013658220000000002,
      "loss": 3.1013,
      "step": 317090
    },
    {
      "epoch": 0.1856866321355093,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001365802,
      "loss": 3.297,
      "step": 317100
    },
    {
      "epoch": 0.18570527537166748,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001365782,
      "loss": 3.0692,
      "step": 317110
    },
    {
      "epoch": 0.18572391860782567,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001365762,
      "loss": 3.0419,
      "step": 317120
    },
    {
      "epoch": 0.18574256184398386,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001365742,
      "loss": 3.0132,
      "step": 317130
    },
    {
      "epoch": 0.18576120508014202,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013657220000000002,
      "loss": 3.2394,
      "step": 317140
    },
    {
      "epoch": 0.1857798483163002,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013657020000000002,
      "loss": 3.128,
      "step": 317150
    },
    {
      "epoch": 0.1857984915524584,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001365682,
      "loss": 3.1777,
      "step": 317160
    },
    {
      "epoch": 0.18581713478861658,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001365662,
      "loss": 3.0645,
      "step": 317170
    },
    {
      "epoch": 0.18583577802477477,
      "grad_norm": 0.75,
      "learning_rate": 0.0001365642,
      "loss": 3.0949,
      "step": 317180
    },
    {
      "epoch": 0.18585442126093293,
      "grad_norm": 0.75,
      "learning_rate": 0.0001365622,
      "loss": 3.0154,
      "step": 317190
    },
    {
      "epoch": 0.18587306449709112,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001365602,
      "loss": 3.0588,
      "step": 317200
    },
    {
      "epoch": 0.1858917077332493,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013655820000000002,
      "loss": 3.0768,
      "step": 317210
    },
    {
      "epoch": 0.1859103509694075,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001365562,
      "loss": 3.1759,
      "step": 317220
    },
    {
      "epoch": 0.1859289942055657,
      "grad_norm": 0.75,
      "learning_rate": 0.0001365542,
      "loss": 3.1265,
      "step": 317230
    },
    {
      "epoch": 0.18594763744172388,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001365522,
      "loss": 3.1151,
      "step": 317240
    },
    {
      "epoch": 0.18596628067788204,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001365502,
      "loss": 3.0733,
      "step": 317250
    },
    {
      "epoch": 0.18598492391404023,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013654820000000002,
      "loss": 3.2017,
      "step": 317260
    },
    {
      "epoch": 0.18600356715019842,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001365462,
      "loss": 3.0091,
      "step": 317270
    },
    {
      "epoch": 0.1860222103863566,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001365442,
      "loss": 3.2095,
      "step": 317280
    },
    {
      "epoch": 0.1860408536225148,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001365422,
      "loss": 3.175,
      "step": 317290
    },
    {
      "epoch": 0.18605949685867296,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001365402,
      "loss": 3.1966,
      "step": 317300
    },
    {
      "epoch": 0.18607814009483115,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00013653820000000003,
      "loss": 3.0637,
      "step": 317310
    },
    {
      "epoch": 0.18609678333098933,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001365362,
      "loss": 3.19,
      "step": 317320
    },
    {
      "epoch": 0.18611542656714752,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013653420000000002,
      "loss": 3.0582,
      "step": 317330
    },
    {
      "epoch": 0.1861340698033057,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001365322,
      "loss": 3.1836,
      "step": 317340
    },
    {
      "epoch": 0.18615271303946387,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001365302,
      "loss": 3.094,
      "step": 317350
    },
    {
      "epoch": 0.18617135627562206,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001365282,
      "loss": 3.1871,
      "step": 317360
    },
    {
      "epoch": 0.18618999951178025,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001365262,
      "loss": 3.0373,
      "step": 317370
    },
    {
      "epoch": 0.18620864274793844,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013652420000000002,
      "loss": 3.0057,
      "step": 317380
    },
    {
      "epoch": 0.18622728598409663,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001365222,
      "loss": 3.0608,
      "step": 317390
    },
    {
      "epoch": 0.1862459292202548,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013652020000000001,
      "loss": 3.1072,
      "step": 317400
    },
    {
      "epoch": 0.18626457245641298,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013651819999999998,
      "loss": 3.134,
      "step": 317410
    },
    {
      "epoch": 0.18628321569257117,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001365162,
      "loss": 3.1312,
      "step": 317420
    },
    {
      "epoch": 0.18630185892872936,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001365142,
      "loss": 3.0217,
      "step": 317430
    },
    {
      "epoch": 0.18632050216488755,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001365122,
      "loss": 3.2492,
      "step": 317440
    },
    {
      "epoch": 0.18633914540104574,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013651020000000002,
      "loss": 2.9864,
      "step": 317450
    },
    {
      "epoch": 0.1863577886372039,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001365082,
      "loss": 2.9846,
      "step": 317460
    },
    {
      "epoch": 0.1863764318733621,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001365062,
      "loss": 3.1211,
      "step": 317470
    },
    {
      "epoch": 0.18639507510952028,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001365042,
      "loss": 3.0639,
      "step": 317480
    },
    {
      "epoch": 0.18641371834567846,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001365022,
      "loss": 2.9126,
      "step": 317490
    },
    {
      "epoch": 0.18643236158183665,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001365002,
      "loss": 3.0743,
      "step": 317500
    },
    {
      "epoch": 0.18645100481799481,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001364982,
      "loss": 3.0656,
      "step": 317510
    },
    {
      "epoch": 0.186469648054153,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013649620000000002,
      "loss": 3.1806,
      "step": 317520
    },
    {
      "epoch": 0.1864882912903112,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001364942,
      "loss": 3.2401,
      "step": 317530
    },
    {
      "epoch": 0.18650693452646938,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001364922,
      "loss": 3.2188,
      "step": 317540
    },
    {
      "epoch": 0.18652557776262757,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001364902,
      "loss": 3.171,
      "step": 317550
    },
    {
      "epoch": 0.18654422099878573,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001364882,
      "loss": 3.097,
      "step": 317560
    },
    {
      "epoch": 0.18656286423494392,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013648620000000002,
      "loss": 3.1171,
      "step": 317570
    },
    {
      "epoch": 0.1865815074711021,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013648420000000002,
      "loss": 3.1472,
      "step": 317580
    },
    {
      "epoch": 0.1866001507072603,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001364822,
      "loss": 3.2136,
      "step": 317590
    },
    {
      "epoch": 0.1866187939434185,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001364802,
      "loss": 3.2021,
      "step": 317600
    },
    {
      "epoch": 0.18663743717957665,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001364782,
      "loss": 3.207,
      "step": 317610
    },
    {
      "epoch": 0.18665608041573484,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001364762,
      "loss": 3.1731,
      "step": 317620
    },
    {
      "epoch": 0.18667472365189303,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001364742,
      "loss": 3.1275,
      "step": 317630
    },
    {
      "epoch": 0.18669336688805122,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013647220000000002,
      "loss": 3.1103,
      "step": 317640
    },
    {
      "epoch": 0.1867120101242094,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001364702,
      "loss": 3.1213,
      "step": 317650
    },
    {
      "epoch": 0.18673065336036757,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001364682,
      "loss": 3.1091,
      "step": 317660
    },
    {
      "epoch": 0.18674929659652575,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001364662,
      "loss": 3.048,
      "step": 317670
    },
    {
      "epoch": 0.18676793983268394,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001364642,
      "loss": 3.0723,
      "step": 317680
    },
    {
      "epoch": 0.18678658306884213,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001364622,
      "loss": 3.1768,
      "step": 317690
    },
    {
      "epoch": 0.18680522630500032,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013646020000000002,
      "loss": 3.17,
      "step": 317700
    },
    {
      "epoch": 0.1868238695411585,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013645820000000001,
      "loss": 3.1248,
      "step": 317710
    },
    {
      "epoch": 0.18684251277731667,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001364562,
      "loss": 3.2602,
      "step": 317720
    },
    {
      "epoch": 0.18686115601347486,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001364542,
      "loss": 3.2909,
      "step": 317730
    },
    {
      "epoch": 0.18687979924963305,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001364522,
      "loss": 3.0625,
      "step": 317740
    },
    {
      "epoch": 0.18689844248579124,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013645020000000002,
      "loss": 3.0832,
      "step": 317750
    },
    {
      "epoch": 0.18691708572194943,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001364482,
      "loss": 3.1335,
      "step": 317760
    },
    {
      "epoch": 0.1869357289581076,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013644620000000002,
      "loss": 3.0671,
      "step": 317770
    },
    {
      "epoch": 0.18695437219426578,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001364442,
      "loss": 3.0743,
      "step": 317780
    },
    {
      "epoch": 0.18697301543042397,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001364422,
      "loss": 3.3426,
      "step": 317790
    },
    {
      "epoch": 0.18699165866658216,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001364402,
      "loss": 3.2546,
      "step": 317800
    },
    {
      "epoch": 0.18701030190274034,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001364382,
      "loss": 3.1131,
      "step": 317810
    },
    {
      "epoch": 0.1870289451388985,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013643620000000002,
      "loss": 3.1112,
      "step": 317820
    },
    {
      "epoch": 0.1870475883750567,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001364342,
      "loss": 3.1641,
      "step": 317830
    },
    {
      "epoch": 0.18706623161121488,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001364322,
      "loss": 3.0229,
      "step": 317840
    },
    {
      "epoch": 0.18708487484737307,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001364302,
      "loss": 3.1087,
      "step": 317850
    },
    {
      "epoch": 0.18710351808353126,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001364282,
      "loss": 2.9534,
      "step": 317860
    },
    {
      "epoch": 0.18712216131968942,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013642620000000003,
      "loss": 3.2558,
      "step": 317870
    },
    {
      "epoch": 0.1871408045558476,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001364242,
      "loss": 3.1277,
      "step": 317880
    },
    {
      "epoch": 0.1871594477920058,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013642220000000002,
      "loss": 3.2406,
      "step": 317890
    },
    {
      "epoch": 0.187178091028164,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013642019999999999,
      "loss": 3.1004,
      "step": 317900
    },
    {
      "epoch": 0.18719673426432218,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001364182,
      "loss": 3.1024,
      "step": 317910
    },
    {
      "epoch": 0.18721537750048037,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001364162,
      "loss": 3.1034,
      "step": 317920
    },
    {
      "epoch": 0.18723402073663853,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001364142,
      "loss": 3.1679,
      "step": 317930
    },
    {
      "epoch": 0.18725266397279672,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013641220000000002,
      "loss": 3.1409,
      "step": 317940
    },
    {
      "epoch": 0.1872713072089549,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001364102,
      "loss": 3.1767,
      "step": 317950
    },
    {
      "epoch": 0.1872899504451131,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001364082,
      "loss": 3.0997,
      "step": 317960
    },
    {
      "epoch": 0.18730859368127128,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001364062,
      "loss": 3.1825,
      "step": 317970
    },
    {
      "epoch": 0.18732723691742945,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001364042,
      "loss": 3.149,
      "step": 317980
    },
    {
      "epoch": 0.18734588015358764,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001364022,
      "loss": 3.1121,
      "step": 317990
    },
    {
      "epoch": 0.18736452338974582,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001364002,
      "loss": 3.1492,
      "step": 318000
    },
    {
      "epoch": 0.187383166625904,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013639820000000002,
      "loss": 3.0333,
      "step": 318010
    },
    {
      "epoch": 0.1874018098620622,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001363962,
      "loss": 3.0663,
      "step": 318020
    },
    {
      "epoch": 0.18742045309822036,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001363942,
      "loss": 3.2733,
      "step": 318030
    },
    {
      "epoch": 0.18743909633437855,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001363922,
      "loss": 3.1091,
      "step": 318040
    },
    {
      "epoch": 0.18745773957053674,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001363902,
      "loss": 3.165,
      "step": 318050
    },
    {
      "epoch": 0.18747638280669493,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013638820000000002,
      "loss": 3.2845,
      "step": 318060
    },
    {
      "epoch": 0.18749502604285312,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001363862,
      "loss": 3.0656,
      "step": 318070
    },
    {
      "epoch": 0.18751366927901128,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013638420000000001,
      "loss": 3.2422,
      "step": 318080
    },
    {
      "epoch": 0.18753231251516947,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001363822,
      "loss": 3.0651,
      "step": 318090
    },
    {
      "epoch": 0.18755095575132766,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001363802,
      "loss": 3.0313,
      "step": 318100
    },
    {
      "epoch": 0.18756959898748585,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001363782,
      "loss": 3.2154,
      "step": 318110
    },
    {
      "epoch": 0.18758824222364404,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001363762,
      "loss": 2.9522,
      "step": 318120
    },
    {
      "epoch": 0.18760688545980223,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013637420000000002,
      "loss": 3.1133,
      "step": 318130
    },
    {
      "epoch": 0.1876255286959604,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013637220000000002,
      "loss": 3.0043,
      "step": 318140
    },
    {
      "epoch": 0.18764417193211858,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001363702,
      "loss": 3.2444,
      "step": 318150
    },
    {
      "epoch": 0.18766281516827676,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001363682,
      "loss": 3.0284,
      "step": 318160
    },
    {
      "epoch": 0.18768145840443495,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001363662,
      "loss": 3.0321,
      "step": 318170
    },
    {
      "epoch": 0.18770010164059314,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001363642,
      "loss": 3.0653,
      "step": 318180
    },
    {
      "epoch": 0.1877187448767513,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001363622,
      "loss": 3.2044,
      "step": 318190
    },
    {
      "epoch": 0.1877373881129095,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013636020000000002,
      "loss": 3.1811,
      "step": 318200
    },
    {
      "epoch": 0.18775603134906768,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001363582,
      "loss": 3.1802,
      "step": 318210
    },
    {
      "epoch": 0.18777467458522587,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001363562,
      "loss": 3.1071,
      "step": 318220
    },
    {
      "epoch": 0.18779331782138406,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001363542,
      "loss": 3.0815,
      "step": 318230
    },
    {
      "epoch": 0.18781196105754222,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001363522,
      "loss": 3.0293,
      "step": 318240
    },
    {
      "epoch": 0.1878306042937004,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001363502,
      "loss": 3.1875,
      "step": 318250
    },
    {
      "epoch": 0.1878492475298586,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013634820000000002,
      "loss": 3.1469,
      "step": 318260
    },
    {
      "epoch": 0.1878678907660168,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001363462,
      "loss": 3.0867,
      "step": 318270
    },
    {
      "epoch": 0.18788653400217498,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001363442,
      "loss": 3.0443,
      "step": 318280
    },
    {
      "epoch": 0.18790517723833314,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001363422,
      "loss": 3.1971,
      "step": 318290
    },
    {
      "epoch": 0.18792382047449133,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001363402,
      "loss": 3.1338,
      "step": 318300
    },
    {
      "epoch": 0.18794246371064952,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013633820000000002,
      "loss": 3.0185,
      "step": 318310
    },
    {
      "epoch": 0.1879611069468077,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001363362,
      "loss": 3.1761,
      "step": 318320
    },
    {
      "epoch": 0.1879797501829659,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001363342,
      "loss": 3.1269,
      "step": 318330
    },
    {
      "epoch": 0.18799839341912405,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001363322,
      "loss": 3.1259,
      "step": 318340
    },
    {
      "epoch": 0.18801703665528224,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001363302,
      "loss": 3.1695,
      "step": 318350
    },
    {
      "epoch": 0.18803567989144043,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013632820000000003,
      "loss": 3.1433,
      "step": 318360
    },
    {
      "epoch": 0.18805432312759862,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001363262,
      "loss": 3.2043,
      "step": 318370
    },
    {
      "epoch": 0.1880729663637568,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013632420000000002,
      "loss": 3.1715,
      "step": 318380
    },
    {
      "epoch": 0.188091609599915,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001363222,
      "loss": 3.1593,
      "step": 318390
    },
    {
      "epoch": 0.18811025283607316,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001363202,
      "loss": 2.9827,
      "step": 318400
    },
    {
      "epoch": 0.18812889607223135,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001363182,
      "loss": 3.1768,
      "step": 318410
    },
    {
      "epoch": 0.18814753930838954,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001363162,
      "loss": 3.1313,
      "step": 318420
    },
    {
      "epoch": 0.18816618254454773,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013631420000000002,
      "loss": 3.1182,
      "step": 318430
    },
    {
      "epoch": 0.18818482578070592,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001363122,
      "loss": 3.0264,
      "step": 318440
    },
    {
      "epoch": 0.18820346901686408,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013631020000000002,
      "loss": 3.2221,
      "step": 318450
    },
    {
      "epoch": 0.18822211225302227,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013630819999999998,
      "loss": 3.1234,
      "step": 318460
    },
    {
      "epoch": 0.18824075548918046,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001363062,
      "loss": 3.0334,
      "step": 318470
    },
    {
      "epoch": 0.18825939872533864,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001363042,
      "loss": 3.1329,
      "step": 318480
    },
    {
      "epoch": 0.18827804196149683,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001363022,
      "loss": 3.1875,
      "step": 318490
    },
    {
      "epoch": 0.188296685197655,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013630020000000002,
      "loss": 3.3296,
      "step": 318500
    },
    {
      "epoch": 0.18831532843381318,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001362982,
      "loss": 3.0948,
      "step": 318510
    },
    {
      "epoch": 0.18833397166997137,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001362962,
      "loss": 3.034,
      "step": 318520
    },
    {
      "epoch": 0.18835261490612956,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001362942,
      "loss": 3.059,
      "step": 318530
    },
    {
      "epoch": 0.18837125814228775,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001362922,
      "loss": 3.0736,
      "step": 318540
    },
    {
      "epoch": 0.1883899013784459,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013629020000000003,
      "loss": 3.1576,
      "step": 318550
    },
    {
      "epoch": 0.1884085446146041,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001362882,
      "loss": 3.0973,
      "step": 318560
    },
    {
      "epoch": 0.1884271878507623,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013628620000000002,
      "loss": 2.9609,
      "step": 318570
    },
    {
      "epoch": 0.18844583108692048,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001362842,
      "loss": 3.0229,
      "step": 318580
    },
    {
      "epoch": 0.18846447432307867,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001362822,
      "loss": 3.0985,
      "step": 318590
    },
    {
      "epoch": 0.18848311755923686,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001362802,
      "loss": 3.1238,
      "step": 318600
    },
    {
      "epoch": 0.18850176079539502,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001362782,
      "loss": 3.0376,
      "step": 318610
    },
    {
      "epoch": 0.1885204040315532,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013627620000000002,
      "loss": 3.1018,
      "step": 318620
    },
    {
      "epoch": 0.1885390472677114,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001362742,
      "loss": 3.1719,
      "step": 318630
    },
    {
      "epoch": 0.18855769050386958,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001362722,
      "loss": 3.0773,
      "step": 318640
    },
    {
      "epoch": 0.18857633374002777,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001362702,
      "loss": 3.1932,
      "step": 318650
    },
    {
      "epoch": 0.18859497697618594,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001362682,
      "loss": 3.0916,
      "step": 318660
    },
    {
      "epoch": 0.18861362021234412,
      "grad_norm": 0.75,
      "learning_rate": 0.0001362662,
      "loss": 3.0754,
      "step": 318670
    },
    {
      "epoch": 0.1886322634485023,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001362642,
      "loss": 3.0289,
      "step": 318680
    },
    {
      "epoch": 0.1886509066846605,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013626220000000002,
      "loss": 3.1572,
      "step": 318690
    },
    {
      "epoch": 0.1886695499208187,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013626020000000001,
      "loss": 3.1019,
      "step": 318700
    },
    {
      "epoch": 0.18868819315697685,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001362582,
      "loss": 3.1265,
      "step": 318710
    },
    {
      "epoch": 0.18870683639313504,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001362562,
      "loss": 3.1905,
      "step": 318720
    },
    {
      "epoch": 0.18872547962929323,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001362542,
      "loss": 3.1327,
      "step": 318730
    },
    {
      "epoch": 0.18874412286545142,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001362522,
      "loss": 3.0651,
      "step": 318740
    },
    {
      "epoch": 0.1887627661016096,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013625020000000002,
      "loss": 3.0675,
      "step": 318750
    },
    {
      "epoch": 0.18878140933776777,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013624820000000001,
      "loss": 3.2163,
      "step": 318760
    },
    {
      "epoch": 0.18880005257392596,
      "grad_norm": 0.75,
      "learning_rate": 0.0001362462,
      "loss": 3.0278,
      "step": 318770
    },
    {
      "epoch": 0.18881869581008415,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001362442,
      "loss": 3.1335,
      "step": 318780
    },
    {
      "epoch": 0.18883733904624234,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001362422,
      "loss": 3.2096,
      "step": 318790
    },
    {
      "epoch": 0.18885598228240053,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001362402,
      "loss": 3.0627,
      "step": 318800
    },
    {
      "epoch": 0.18887462551855871,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001362382,
      "loss": 3.0713,
      "step": 318810
    },
    {
      "epoch": 0.18889326875471688,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013623620000000002,
      "loss": 3.1026,
      "step": 318820
    },
    {
      "epoch": 0.18891191199087506,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001362342,
      "loss": 3.0828,
      "step": 318830
    },
    {
      "epoch": 0.18893055522703325,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001362322,
      "loss": 3.2564,
      "step": 318840
    },
    {
      "epoch": 0.18894919846319144,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001362302,
      "loss": 3.0312,
      "step": 318850
    },
    {
      "epoch": 0.18896784169934963,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001362282,
      "loss": 3.1197,
      "step": 318860
    },
    {
      "epoch": 0.1889864849355078,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013622620000000002,
      "loss": 3.0949,
      "step": 318870
    },
    {
      "epoch": 0.18900512817166598,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001362242,
      "loss": 3.1558,
      "step": 318880
    },
    {
      "epoch": 0.18902377140782417,
      "grad_norm": 0.75,
      "learning_rate": 0.0001362222,
      "loss": 3.0593,
      "step": 318890
    },
    {
      "epoch": 0.18904241464398236,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001362202,
      "loss": 3.0704,
      "step": 318900
    },
    {
      "epoch": 0.18906105788014055,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001362182,
      "loss": 3.0352,
      "step": 318910
    },
    {
      "epoch": 0.1890797011162987,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013621620000000003,
      "loss": 3.1472,
      "step": 318920
    },
    {
      "epoch": 0.1890983443524569,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001362142,
      "loss": 3.1491,
      "step": 318930
    },
    {
      "epoch": 0.1891169875886151,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013621220000000002,
      "loss": 3.2098,
      "step": 318940
    },
    {
      "epoch": 0.18913563082477328,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013621019999999999,
      "loss": 3.0026,
      "step": 318950
    },
    {
      "epoch": 0.18915427406093147,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001362082,
      "loss": 3.1238,
      "step": 318960
    },
    {
      "epoch": 0.18917291729708963,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001362062,
      "loss": 3.0247,
      "step": 318970
    },
    {
      "epoch": 0.18919156053324782,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001362042,
      "loss": 3.0207,
      "step": 318980
    },
    {
      "epoch": 0.189210203769406,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013620220000000002,
      "loss": 3.1224,
      "step": 318990
    },
    {
      "epoch": 0.1892288470055642,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001362002,
      "loss": 3.1869,
      "step": 319000
    },
    {
      "epoch": 0.18924749024172238,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001361982,
      "loss": 3.0952,
      "step": 319010
    },
    {
      "epoch": 0.18926613347788057,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001361962,
      "loss": 3.1878,
      "step": 319020
    },
    {
      "epoch": 0.18928477671403873,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001361942,
      "loss": 3.005,
      "step": 319030
    },
    {
      "epoch": 0.18930341995019692,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013619220000000003,
      "loss": 2.9786,
      "step": 319040
    },
    {
      "epoch": 0.1893220631863551,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001361902,
      "loss": 3.1369,
      "step": 319050
    },
    {
      "epoch": 0.1893407064225133,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013618820000000002,
      "loss": 3.1984,
      "step": 319060
    },
    {
      "epoch": 0.1893593496586715,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001361862,
      "loss": 3.2703,
      "step": 319070
    },
    {
      "epoch": 0.18937799289482965,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001361842,
      "loss": 3.057,
      "step": 319080
    },
    {
      "epoch": 0.18939663613098784,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001361822,
      "loss": 2.9854,
      "step": 319090
    },
    {
      "epoch": 0.18941527936714603,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001361802,
      "loss": 2.9237,
      "step": 319100
    },
    {
      "epoch": 0.18943392260330422,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013617820000000002,
      "loss": 3.1381,
      "step": 319110
    },
    {
      "epoch": 0.1894525658394624,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001361762,
      "loss": 3.0379,
      "step": 319120
    },
    {
      "epoch": 0.18947120907562057,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013617420000000001,
      "loss": 3.054,
      "step": 319130
    },
    {
      "epoch": 0.18948985231177876,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001361722,
      "loss": 3.1104,
      "step": 319140
    },
    {
      "epoch": 0.18950849554793694,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001361702,
      "loss": 3.2376,
      "step": 319150
    },
    {
      "epoch": 0.18952713878409513,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001361682,
      "loss": 3.1591,
      "step": 319160
    },
    {
      "epoch": 0.18954578202025332,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001361662,
      "loss": 3.071,
      "step": 319170
    },
    {
      "epoch": 0.18956442525641148,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013616420000000002,
      "loss": 3.0979,
      "step": 319180
    },
    {
      "epoch": 0.18958306849256967,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013616220000000002,
      "loss": 3.0686,
      "step": 319190
    },
    {
      "epoch": 0.18960171172872786,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001361602,
      "loss": 3.111,
      "step": 319200
    },
    {
      "epoch": 0.18962035496488605,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001361582,
      "loss": 3.0512,
      "step": 319210
    },
    {
      "epoch": 0.18963899820104424,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001361562,
      "loss": 3.0261,
      "step": 319220
    },
    {
      "epoch": 0.1896576414372024,
      "grad_norm": 0.75,
      "learning_rate": 0.0001361542,
      "loss": 3.1885,
      "step": 319230
    },
    {
      "epoch": 0.1896762846733606,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001361522,
      "loss": 3.1781,
      "step": 319240
    },
    {
      "epoch": 0.18969492790951878,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013615020000000002,
      "loss": 3.0936,
      "step": 319250
    },
    {
      "epoch": 0.18971357114567697,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001361482,
      "loss": 3.2118,
      "step": 319260
    },
    {
      "epoch": 0.18973221438183516,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001361462,
      "loss": 3.1596,
      "step": 319270
    },
    {
      "epoch": 0.18975085761799335,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001361442,
      "loss": 3.2378,
      "step": 319280
    },
    {
      "epoch": 0.1897695008541515,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001361422,
      "loss": 3.1929,
      "step": 319290
    },
    {
      "epoch": 0.1897881440903097,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001361402,
      "loss": 3.0085,
      "step": 319300
    },
    {
      "epoch": 0.18980678732646789,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013613820000000002,
      "loss": 3.0315,
      "step": 319310
    },
    {
      "epoch": 0.18982543056262607,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001361362,
      "loss": 3.0363,
      "step": 319320
    },
    {
      "epoch": 0.18984407379878426,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001361342,
      "loss": 3.1081,
      "step": 319330
    },
    {
      "epoch": 0.18986271703494242,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001361322,
      "loss": 3.2647,
      "step": 319340
    },
    {
      "epoch": 0.1898813602711006,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001361302,
      "loss": 3.1901,
      "step": 319350
    },
    {
      "epoch": 0.1899000035072588,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013612820000000002,
      "loss": 3.1614,
      "step": 319360
    },
    {
      "epoch": 0.189918646743417,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001361262,
      "loss": 3.1457,
      "step": 319370
    },
    {
      "epoch": 0.18993728997957518,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013612420000000001,
      "loss": 3.0955,
      "step": 319380
    },
    {
      "epoch": 0.18995593321573334,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001361222,
      "loss": 3.0542,
      "step": 319390
    },
    {
      "epoch": 0.18997457645189153,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001361202,
      "loss": 3.1514,
      "step": 319400
    },
    {
      "epoch": 0.18999321968804972,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001361182,
      "loss": 3.1349,
      "step": 319410
    },
    {
      "epoch": 0.1900118629242079,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001361162,
      "loss": 3.085,
      "step": 319420
    },
    {
      "epoch": 0.1900305061603661,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013611420000000002,
      "loss": 3.1776,
      "step": 319430
    },
    {
      "epoch": 0.19004914939652426,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001361122,
      "loss": 3.1717,
      "step": 319440
    },
    {
      "epoch": 0.19006779263268245,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001361102,
      "loss": 3.0992,
      "step": 319450
    },
    {
      "epoch": 0.19008643586884064,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001361082,
      "loss": 3.0402,
      "step": 319460
    },
    {
      "epoch": 0.19010507910499883,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001361062,
      "loss": 3.1549,
      "step": 319470
    },
    {
      "epoch": 0.19012372234115701,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013610420000000002,
      "loss": 3.1618,
      "step": 319480
    },
    {
      "epoch": 0.1901423655773152,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001361022,
      "loss": 2.9616,
      "step": 319490
    },
    {
      "epoch": 0.19016100881347336,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013610020000000002,
      "loss": 3.152,
      "step": 319500
    },
    {
      "epoch": 0.19017965204963155,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001360982,
      "loss": 2.9914,
      "step": 319510
    },
    {
      "epoch": 0.19019829528578974,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001360962,
      "loss": 3.1644,
      "step": 319520
    },
    {
      "epoch": 0.19021693852194793,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013609420000000003,
      "loss": 3.2925,
      "step": 319530
    },
    {
      "epoch": 0.19023558175810612,
      "grad_norm": 0.625,
      "learning_rate": 0.0001360922,
      "loss": 3.1523,
      "step": 319540
    },
    {
      "epoch": 0.19025422499426428,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013609020000000002,
      "loss": 3.0132,
      "step": 319550
    },
    {
      "epoch": 0.19027286823042247,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001360882,
      "loss": 3.1008,
      "step": 319560
    },
    {
      "epoch": 0.19029151146658066,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001360862,
      "loss": 3.1743,
      "step": 319570
    },
    {
      "epoch": 0.19031015470273885,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001360842,
      "loss": 3.0689,
      "step": 319580
    },
    {
      "epoch": 0.19032879793889704,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001360822,
      "loss": 3.1201,
      "step": 319590
    },
    {
      "epoch": 0.1903474411750552,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013608020000000003,
      "loss": 3.1975,
      "step": 319600
    },
    {
      "epoch": 0.1903660844112134,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001360782,
      "loss": 2.9939,
      "step": 319610
    },
    {
      "epoch": 0.19038472764737158,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013607620000000002,
      "loss": 3.1743,
      "step": 319620
    },
    {
      "epoch": 0.19040337088352977,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013607419999999999,
      "loss": 3.1053,
      "step": 319630
    },
    {
      "epoch": 0.19042201411968795,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001360722,
      "loss": 3.1211,
      "step": 319640
    },
    {
      "epoch": 0.19044065735584612,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001360702,
      "loss": 3.1209,
      "step": 319650
    },
    {
      "epoch": 0.1904593005920043,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001360682,
      "loss": 3.0727,
      "step": 319660
    },
    {
      "epoch": 0.1904779438281625,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013606620000000002,
      "loss": 3.1394,
      "step": 319670
    },
    {
      "epoch": 0.19049658706432068,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001360642,
      "loss": 3.0507,
      "step": 319680
    },
    {
      "epoch": 0.19051523030047887,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001360622,
      "loss": 3.0967,
      "step": 319690
    },
    {
      "epoch": 0.19053387353663706,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001360602,
      "loss": 3.0781,
      "step": 319700
    },
    {
      "epoch": 0.19055251677279522,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001360582,
      "loss": 2.9504,
      "step": 319710
    },
    {
      "epoch": 0.1905711600089534,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001360562,
      "loss": 3.0702,
      "step": 319720
    },
    {
      "epoch": 0.1905898032451116,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001360542,
      "loss": 3.0981,
      "step": 319730
    },
    {
      "epoch": 0.1906084464812698,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013605220000000002,
      "loss": 3.1025,
      "step": 319740
    },
    {
      "epoch": 0.19062708971742798,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013605020000000001,
      "loss": 3.1249,
      "step": 319750
    },
    {
      "epoch": 0.19064573295358614,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001360482,
      "loss": 3.1737,
      "step": 319760
    },
    {
      "epoch": 0.19066437618974433,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001360462,
      "loss": 3.1028,
      "step": 319770
    },
    {
      "epoch": 0.19068301942590252,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001360442,
      "loss": 3.1201,
      "step": 319780
    },
    {
      "epoch": 0.1907016626620607,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001360422,
      "loss": 3.229,
      "step": 319790
    },
    {
      "epoch": 0.1907203058982189,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001360402,
      "loss": 3.2141,
      "step": 319800
    },
    {
      "epoch": 0.19073894913437706,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013603820000000001,
      "loss": 3.121,
      "step": 319810
    },
    {
      "epoch": 0.19075759237053524,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001360362,
      "loss": 3.058,
      "step": 319820
    },
    {
      "epoch": 0.19077623560669343,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001360342,
      "loss": 3.1509,
      "step": 319830
    },
    {
      "epoch": 0.19079487884285162,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001360322,
      "loss": 3.2236,
      "step": 319840
    },
    {
      "epoch": 0.1908135220790098,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001360302,
      "loss": 3.0244,
      "step": 319850
    },
    {
      "epoch": 0.19083216531516797,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001360282,
      "loss": 3.0032,
      "step": 319860
    },
    {
      "epoch": 0.19085080855132616,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013602620000000002,
      "loss": 2.9773,
      "step": 319870
    },
    {
      "epoch": 0.19086945178748435,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001360242,
      "loss": 3.074,
      "step": 319880
    },
    {
      "epoch": 0.19088809502364254,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001360222,
      "loss": 3.1293,
      "step": 319890
    },
    {
      "epoch": 0.19090673825980073,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001360202,
      "loss": 3.0457,
      "step": 319900
    },
    {
      "epoch": 0.1909253814959589,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001360182,
      "loss": 3.1848,
      "step": 319910
    },
    {
      "epoch": 0.19094402473211708,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013601620000000002,
      "loss": 3.117,
      "step": 319920
    },
    {
      "epoch": 0.19096266796827527,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001360142,
      "loss": 3.1062,
      "step": 319930
    },
    {
      "epoch": 0.19098131120443346,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001360122,
      "loss": 3.1267,
      "step": 319940
    },
    {
      "epoch": 0.19099995444059165,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001360102,
      "loss": 3.0595,
      "step": 319950
    },
    {
      "epoch": 0.19101859767674984,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001360082,
      "loss": 3.1855,
      "step": 319960
    },
    {
      "epoch": 0.191037240912908,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013600620000000003,
      "loss": 3.2789,
      "step": 319970
    },
    {
      "epoch": 0.19105588414906619,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001360042,
      "loss": 3.0808,
      "step": 319980
    },
    {
      "epoch": 0.19107452738522437,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013600220000000002,
      "loss": 3.0801,
      "step": 319990
    },
    {
      "epoch": 0.19109317062138256,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001360002,
      "loss": 3.2256,
      "step": 320000
    },
    {
      "epoch": 0.19111181385754075,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001359982,
      "loss": 3.1575,
      "step": 320010
    },
    {
      "epoch": 0.1911304570936989,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001359962,
      "loss": 3.0557,
      "step": 320020
    },
    {
      "epoch": 0.1911491003298571,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001359942,
      "loss": 3.1168,
      "step": 320030
    },
    {
      "epoch": 0.1911677435660153,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013599220000000002,
      "loss": 3.1056,
      "step": 320040
    },
    {
      "epoch": 0.19118638680217348,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001359902,
      "loss": 3.1855,
      "step": 320050
    },
    {
      "epoch": 0.19120503003833167,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013598820000000001,
      "loss": 3.077,
      "step": 320060
    },
    {
      "epoch": 0.19122367327448983,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001359862,
      "loss": 3.118,
      "step": 320070
    },
    {
      "epoch": 0.19124231651064802,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001359842,
      "loss": 3.1037,
      "step": 320080
    },
    {
      "epoch": 0.1912609597468062,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013598220000000003,
      "loss": 3.0375,
      "step": 320090
    },
    {
      "epoch": 0.1912796029829644,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001359802,
      "loss": 3.0975,
      "step": 320100
    },
    {
      "epoch": 0.1912982462191226,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013597820000000002,
      "loss": 3.136,
      "step": 320110
    },
    {
      "epoch": 0.19131688945528075,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001359762,
      "loss": 3.0601,
      "step": 320120
    },
    {
      "epoch": 0.19133553269143894,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001359742,
      "loss": 3.1058,
      "step": 320130
    },
    {
      "epoch": 0.19135417592759713,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001359722,
      "loss": 3.1632,
      "step": 320140
    },
    {
      "epoch": 0.19137281916375531,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001359702,
      "loss": 3.1405,
      "step": 320150
    },
    {
      "epoch": 0.1913914623999135,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00013596820000000002,
      "loss": 3.117,
      "step": 320160
    },
    {
      "epoch": 0.1914101056360717,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001359662,
      "loss": 3.1339,
      "step": 320170
    },
    {
      "epoch": 0.19142874887222985,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013596420000000002,
      "loss": 3.131,
      "step": 320180
    },
    {
      "epoch": 0.19144739210838804,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001359622,
      "loss": 3.0877,
      "step": 320190
    },
    {
      "epoch": 0.19146603534454623,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001359602,
      "loss": 3.1493,
      "step": 320200
    },
    {
      "epoch": 0.19148467858070442,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001359582,
      "loss": 3.1987,
      "step": 320210
    },
    {
      "epoch": 0.1915033218168626,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001359562,
      "loss": 3.2109,
      "step": 320220
    },
    {
      "epoch": 0.19152196505302077,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013595420000000002,
      "loss": 3.1928,
      "step": 320230
    },
    {
      "epoch": 0.19154060828917896,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001359522,
      "loss": 3.1791,
      "step": 320240
    },
    {
      "epoch": 0.19155925152533715,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001359502,
      "loss": 3.0658,
      "step": 320250
    },
    {
      "epoch": 0.19157789476149534,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001359482,
      "loss": 3.0249,
      "step": 320260
    },
    {
      "epoch": 0.19159653799765353,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001359462,
      "loss": 3.0883,
      "step": 320270
    },
    {
      "epoch": 0.1916151812338117,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001359442,
      "loss": 3.0332,
      "step": 320280
    },
    {
      "epoch": 0.19163382446996988,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001359422,
      "loss": 3.1449,
      "step": 320290
    },
    {
      "epoch": 0.19165246770612807,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013594020000000002,
      "loss": 3.1261,
      "step": 320300
    },
    {
      "epoch": 0.19167111094228625,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001359382,
      "loss": 3.1523,
      "step": 320310
    },
    {
      "epoch": 0.19168975417844444,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001359362,
      "loss": 3.0579,
      "step": 320320
    },
    {
      "epoch": 0.1917083974146026,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001359342,
      "loss": 3.2185,
      "step": 320330
    },
    {
      "epoch": 0.1917270406507608,
      "grad_norm": 0.75,
      "learning_rate": 0.0001359322,
      "loss": 2.9932,
      "step": 320340
    },
    {
      "epoch": 0.19174568388691898,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001359302,
      "loss": 3.1492,
      "step": 320350
    },
    {
      "epoch": 0.19176432712307717,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013592820000000002,
      "loss": 3.1055,
      "step": 320360
    },
    {
      "epoch": 0.19178297035923536,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001359262,
      "loss": 3.053,
      "step": 320370
    },
    {
      "epoch": 0.19180161359539355,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001359242,
      "loss": 3.1425,
      "step": 320380
    },
    {
      "epoch": 0.1918202568315517,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001359222,
      "loss": 2.9461,
      "step": 320390
    },
    {
      "epoch": 0.1918389000677099,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001359202,
      "loss": 3.0971,
      "step": 320400
    },
    {
      "epoch": 0.1918575433038681,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001359182,
      "loss": 3.1823,
      "step": 320410
    },
    {
      "epoch": 0.19187618654002628,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001359162,
      "loss": 3.3089,
      "step": 320420
    },
    {
      "epoch": 0.19189482977618447,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013591420000000001,
      "loss": 3.1594,
      "step": 320430
    },
    {
      "epoch": 0.19191347301234263,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001359122,
      "loss": 3.1232,
      "step": 320440
    },
    {
      "epoch": 0.19193211624850082,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001359102,
      "loss": 3.1248,
      "step": 320450
    },
    {
      "epoch": 0.191950759484659,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001359082,
      "loss": 3.1414,
      "step": 320460
    },
    {
      "epoch": 0.1919694027208172,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001359062,
      "loss": 3.1327,
      "step": 320470
    },
    {
      "epoch": 0.19198804595697538,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013590420000000002,
      "loss": 3.1708,
      "step": 320480
    },
    {
      "epoch": 0.19200668919313355,
      "grad_norm": 0.859375,
      "learning_rate": 0.00013590220000000001,
      "loss": 3.1354,
      "step": 320490
    },
    {
      "epoch": 0.19202533242929173,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001359002,
      "loss": 3.0353,
      "step": 320500
    },
    {
      "epoch": 0.19204397566544992,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001358982,
      "loss": 3.1775,
      "step": 320510
    },
    {
      "epoch": 0.1920626189016081,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001358962,
      "loss": 3.1322,
      "step": 320520
    },
    {
      "epoch": 0.1920812621377663,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013589420000000002,
      "loss": 3.2475,
      "step": 320530
    },
    {
      "epoch": 0.19209990537392446,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001358922,
      "loss": 3.1705,
      "step": 320540
    },
    {
      "epoch": 0.19211854861008265,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013589020000000002,
      "loss": 3.1267,
      "step": 320550
    },
    {
      "epoch": 0.19213719184624084,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001358882,
      "loss": 3.0637,
      "step": 320560
    },
    {
      "epoch": 0.19215583508239903,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001358862,
      "loss": 3.1058,
      "step": 320570
    },
    {
      "epoch": 0.19217447831855722,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013588420000000003,
      "loss": 3.1504,
      "step": 320580
    },
    {
      "epoch": 0.19219312155471538,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001358822,
      "loss": 3.1329,
      "step": 320590
    },
    {
      "epoch": 0.19221176479087357,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013588020000000002,
      "loss": 3.1392,
      "step": 320600
    },
    {
      "epoch": 0.19223040802703176,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001358782,
      "loss": 3.1082,
      "step": 320610
    },
    {
      "epoch": 0.19224905126318995,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001358762,
      "loss": 3.0176,
      "step": 320620
    },
    {
      "epoch": 0.19226769449934814,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001358742,
      "loss": 3.1836,
      "step": 320630
    },
    {
      "epoch": 0.19228633773550632,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001358722,
      "loss": 3.1628,
      "step": 320640
    },
    {
      "epoch": 0.19230498097166449,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013587020000000003,
      "loss": 3.2321,
      "step": 320650
    },
    {
      "epoch": 0.19232362420782267,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001358682,
      "loss": 3.0449,
      "step": 320660
    },
    {
      "epoch": 0.19234226744398086,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013586620000000002,
      "loss": 3.1382,
      "step": 320670
    },
    {
      "epoch": 0.19236091068013905,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013586419999999999,
      "loss": 3.2191,
      "step": 320680
    },
    {
      "epoch": 0.19237955391629724,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001358622,
      "loss": 3.0308,
      "step": 320690
    },
    {
      "epoch": 0.1923981971524554,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001358602,
      "loss": 3.0751,
      "step": 320700
    },
    {
      "epoch": 0.1924168403886136,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001358582,
      "loss": 3.2231,
      "step": 320710
    },
    {
      "epoch": 0.19243548362477178,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013585620000000002,
      "loss": 3.257,
      "step": 320720
    },
    {
      "epoch": 0.19245412686092997,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001358542,
      "loss": 3.2651,
      "step": 320730
    },
    {
      "epoch": 0.19247277009708816,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00013585220000000001,
      "loss": 2.9405,
      "step": 320740
    },
    {
      "epoch": 0.19249141333324632,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001358502,
      "loss": 3.1128,
      "step": 320750
    },
    {
      "epoch": 0.1925100565694045,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001358482,
      "loss": 3.23,
      "step": 320760
    },
    {
      "epoch": 0.1925286998055627,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001358462,
      "loss": 3.2147,
      "step": 320770
    },
    {
      "epoch": 0.1925473430417209,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001358442,
      "loss": 3.2492,
      "step": 320780
    },
    {
      "epoch": 0.19256598627787908,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013584220000000002,
      "loss": 3.1268,
      "step": 320790
    },
    {
      "epoch": 0.19258462951403724,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013584020000000001,
      "loss": 3.1419,
      "step": 320800
    },
    {
      "epoch": 0.19260327275019543,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001358382,
      "loss": 3.0651,
      "step": 320810
    },
    {
      "epoch": 0.19262191598635361,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001358362,
      "loss": 3.033,
      "step": 320820
    },
    {
      "epoch": 0.1926405592225118,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001358342,
      "loss": 3.2601,
      "step": 320830
    },
    {
      "epoch": 0.19265920245867,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001358322,
      "loss": 3.177,
      "step": 320840
    },
    {
      "epoch": 0.19267784569482818,
      "grad_norm": 0.75,
      "learning_rate": 0.0001358302,
      "loss": 3.2014,
      "step": 320850
    },
    {
      "epoch": 0.19269648893098634,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013582820000000002,
      "loss": 3.1408,
      "step": 320860
    },
    {
      "epoch": 0.19271513216714453,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001358262,
      "loss": 3.268,
      "step": 320870
    },
    {
      "epoch": 0.19273377540330272,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001358242,
      "loss": 3.0709,
      "step": 320880
    },
    {
      "epoch": 0.1927524186394609,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001358222,
      "loss": 3.1624,
      "step": 320890
    },
    {
      "epoch": 0.1927710618756191,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001358202,
      "loss": 3.1855,
      "step": 320900
    },
    {
      "epoch": 0.19278970511177726,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001358182,
      "loss": 3.1072,
      "step": 320910
    },
    {
      "epoch": 0.19280834834793545,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013581620000000002,
      "loss": 3.1277,
      "step": 320920
    },
    {
      "epoch": 0.19282699158409364,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001358142,
      "loss": 3.1877,
      "step": 320930
    },
    {
      "epoch": 0.19284563482025183,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001358122,
      "loss": 3.1336,
      "step": 320940
    },
    {
      "epoch": 0.19286427805641002,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001358102,
      "loss": 3.0201,
      "step": 320950
    },
    {
      "epoch": 0.19288292129256818,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001358082,
      "loss": 3.0984,
      "step": 320960
    },
    {
      "epoch": 0.19290156452872637,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013580620000000002,
      "loss": 3.0187,
      "step": 320970
    },
    {
      "epoch": 0.19292020776488455,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013580420000000002,
      "loss": 3.1269,
      "step": 320980
    },
    {
      "epoch": 0.19293885100104274,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001358022,
      "loss": 3.1505,
      "step": 320990
    },
    {
      "epoch": 0.19295749423720093,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001358002,
      "loss": 3.2085,
      "step": 321000
    },
    {
      "epoch": 0.1929761374733591,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001357982,
      "loss": 3.107,
      "step": 321010
    },
    {
      "epoch": 0.19299478070951728,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001357962,
      "loss": 3.1312,
      "step": 321020
    },
    {
      "epoch": 0.19301342394567547,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001357942,
      "loss": 3.1589,
      "step": 321030
    },
    {
      "epoch": 0.19303206718183366,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013579220000000002,
      "loss": 3.0994,
      "step": 321040
    },
    {
      "epoch": 0.19305071041799185,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001357902,
      "loss": 2.9306,
      "step": 321050
    },
    {
      "epoch": 0.19306935365415004,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001357882,
      "loss": 3.1131,
      "step": 321060
    },
    {
      "epoch": 0.1930879968903082,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001357862,
      "loss": 3.1321,
      "step": 321070
    },
    {
      "epoch": 0.1931066401264664,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001357842,
      "loss": 3.2185,
      "step": 321080
    },
    {
      "epoch": 0.19312528336262458,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013578220000000002,
      "loss": 3.205,
      "step": 321090
    },
    {
      "epoch": 0.19314392659878277,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001357802,
      "loss": 3.1028,
      "step": 321100
    },
    {
      "epoch": 0.19316256983494096,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013577820000000001,
      "loss": 3.058,
      "step": 321110
    },
    {
      "epoch": 0.19318121307109912,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001357762,
      "loss": 3.1111,
      "step": 321120
    },
    {
      "epoch": 0.1931998563072573,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001357742,
      "loss": 3.0239,
      "step": 321130
    },
    {
      "epoch": 0.1932184995434155,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013577220000000003,
      "loss": 3.1482,
      "step": 321140
    },
    {
      "epoch": 0.19323714277957368,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001357702,
      "loss": 3.1155,
      "step": 321150
    },
    {
      "epoch": 0.19325578601573187,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013576820000000002,
      "loss": 3.0802,
      "step": 321160
    },
    {
      "epoch": 0.19327442925189003,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001357662,
      "loss": 3.2015,
      "step": 321170
    },
    {
      "epoch": 0.19329307248804822,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001357642,
      "loss": 3.1165,
      "step": 321180
    },
    {
      "epoch": 0.1933117157242064,
      "grad_norm": 0.75,
      "learning_rate": 0.0001357622,
      "loss": 3.1305,
      "step": 321190
    },
    {
      "epoch": 0.1933303589603646,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001357602,
      "loss": 3.1181,
      "step": 321200
    },
    {
      "epoch": 0.1933490021965228,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013575820000000002,
      "loss": 3.2504,
      "step": 321210
    },
    {
      "epoch": 0.19336764543268095,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001357562,
      "loss": 3.1486,
      "step": 321220
    },
    {
      "epoch": 0.19338628866883914,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013575420000000002,
      "loss": 3.2087,
      "step": 321230
    },
    {
      "epoch": 0.19340493190499733,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013575219999999998,
      "loss": 3.1498,
      "step": 321240
    },
    {
      "epoch": 0.19342357514115552,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001357502,
      "loss": 3.2915,
      "step": 321250
    },
    {
      "epoch": 0.1934422183773137,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001357482,
      "loss": 3.086,
      "step": 321260
    },
    {
      "epoch": 0.19346086161347187,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001357462,
      "loss": 3.0865,
      "step": 321270
    },
    {
      "epoch": 0.19347950484963006,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013574420000000002,
      "loss": 3.0754,
      "step": 321280
    },
    {
      "epoch": 0.19349814808578825,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001357422,
      "loss": 3.2972,
      "step": 321290
    },
    {
      "epoch": 0.19351679132194644,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001357402,
      "loss": 3.0781,
      "step": 321300
    },
    {
      "epoch": 0.19353543455810462,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001357382,
      "loss": 3.0658,
      "step": 321310
    },
    {
      "epoch": 0.1935540777942628,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001357362,
      "loss": 3.1364,
      "step": 321320
    },
    {
      "epoch": 0.19357272103042097,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001357342,
      "loss": 3.1289,
      "step": 321330
    },
    {
      "epoch": 0.19359136426657916,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001357322,
      "loss": 3.1124,
      "step": 321340
    },
    {
      "epoch": 0.19361000750273735,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013573020000000002,
      "loss": 3.127,
      "step": 321350
    },
    {
      "epoch": 0.19362865073889554,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001357282,
      "loss": 3.17,
      "step": 321360
    },
    {
      "epoch": 0.19364729397505373,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001357262,
      "loss": 2.9869,
      "step": 321370
    },
    {
      "epoch": 0.1936659372112119,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001357242,
      "loss": 2.9565,
      "step": 321380
    },
    {
      "epoch": 0.19368458044737008,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001357222,
      "loss": 3.0424,
      "step": 321390
    },
    {
      "epoch": 0.19370322368352827,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001357202,
      "loss": 2.9693,
      "step": 321400
    },
    {
      "epoch": 0.19372186691968646,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013571820000000002,
      "loss": 3.0914,
      "step": 321410
    },
    {
      "epoch": 0.19374051015584465,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013571620000000001,
      "loss": 2.9744,
      "step": 321420
    },
    {
      "epoch": 0.1937591533920028,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001357142,
      "loss": 3.1334,
      "step": 321430
    },
    {
      "epoch": 0.193777796628161,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001357122,
      "loss": 3.1204,
      "step": 321440
    },
    {
      "epoch": 0.1937964398643192,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001357102,
      "loss": 3.2301,
      "step": 321450
    },
    {
      "epoch": 0.19381508310047738,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001357082,
      "loss": 3.1776,
      "step": 321460
    },
    {
      "epoch": 0.19383372633663556,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013570620000000002,
      "loss": 3.1557,
      "step": 321470
    },
    {
      "epoch": 0.19385236957279373,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013570420000000001,
      "loss": 3.137,
      "step": 321480
    },
    {
      "epoch": 0.19387101280895191,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001357022,
      "loss": 3.2598,
      "step": 321490
    },
    {
      "epoch": 0.1938896560451101,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001357002,
      "loss": 3.0901,
      "step": 321500
    },
    {
      "epoch": 0.1939082992812683,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001356982,
      "loss": 3.0142,
      "step": 321510
    },
    {
      "epoch": 0.19392694251742648,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001356962,
      "loss": 3.1323,
      "step": 321520
    },
    {
      "epoch": 0.19394558575358467,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013569420000000002,
      "loss": 3.0869,
      "step": 321530
    },
    {
      "epoch": 0.19396422898974283,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013569220000000002,
      "loss": 3.0484,
      "step": 321540
    },
    {
      "epoch": 0.19398287222590102,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001356902,
      "loss": 3.0416,
      "step": 321550
    },
    {
      "epoch": 0.1940015154620592,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001356882,
      "loss": 3.1165,
      "step": 321560
    },
    {
      "epoch": 0.1940201586982174,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001356862,
      "loss": 3.0514,
      "step": 321570
    },
    {
      "epoch": 0.1940388019343756,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013568420000000002,
      "loss": 3.1767,
      "step": 321580
    },
    {
      "epoch": 0.19405744517053375,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001356822,
      "loss": 3.1177,
      "step": 321590
    },
    {
      "epoch": 0.19407608840669194,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013568020000000002,
      "loss": 3.1565,
      "step": 321600
    },
    {
      "epoch": 0.19409473164285013,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001356782,
      "loss": 3.1455,
      "step": 321610
    },
    {
      "epoch": 0.19411337487900832,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001356762,
      "loss": 3.098,
      "step": 321620
    },
    {
      "epoch": 0.1941320181151665,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001356742,
      "loss": 3.1054,
      "step": 321630
    },
    {
      "epoch": 0.19415066135132467,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001356722,
      "loss": 3.141,
      "step": 321640
    },
    {
      "epoch": 0.19416930458748285,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013567020000000002,
      "loss": 3.2103,
      "step": 321650
    },
    {
      "epoch": 0.19418794782364104,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001356682,
      "loss": 3.1807,
      "step": 321660
    },
    {
      "epoch": 0.19420659105979923,
      "grad_norm": 0.75,
      "learning_rate": 0.0001356662,
      "loss": 3.2072,
      "step": 321670
    },
    {
      "epoch": 0.19422523429595742,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001356642,
      "loss": 3.1667,
      "step": 321680
    },
    {
      "epoch": 0.19424387753211558,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001356622,
      "loss": 3.1414,
      "step": 321690
    },
    {
      "epoch": 0.19426252076827377,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013566020000000003,
      "loss": 3.0916,
      "step": 321700
    },
    {
      "epoch": 0.19428116400443196,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001356582,
      "loss": 3.0976,
      "step": 321710
    },
    {
      "epoch": 0.19429980724059015,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013565620000000002,
      "loss": 3.2726,
      "step": 321720
    },
    {
      "epoch": 0.19431845047674834,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013565419999999999,
      "loss": 3.0395,
      "step": 321730
    },
    {
      "epoch": 0.19433709371290653,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001356522,
      "loss": 3.0065,
      "step": 321740
    },
    {
      "epoch": 0.1943557369490647,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001356502,
      "loss": 3.2024,
      "step": 321750
    },
    {
      "epoch": 0.19437438018522288,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001356482,
      "loss": 3.1839,
      "step": 321760
    },
    {
      "epoch": 0.19439302342138107,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013564620000000002,
      "loss": 3.1963,
      "step": 321770
    },
    {
      "epoch": 0.19441166665753926,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001356442,
      "loss": 3.103,
      "step": 321780
    },
    {
      "epoch": 0.19443030989369744,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013564220000000001,
      "loss": 3.0591,
      "step": 321790
    },
    {
      "epoch": 0.1944489531298556,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001356402,
      "loss": 3.143,
      "step": 321800
    },
    {
      "epoch": 0.1944675963660138,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001356382,
      "loss": 3.1529,
      "step": 321810
    },
    {
      "epoch": 0.19448623960217198,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001356362,
      "loss": 3.1489,
      "step": 321820
    },
    {
      "epoch": 0.19450488283833017,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001356342,
      "loss": 3.11,
      "step": 321830
    },
    {
      "epoch": 0.19452352607448836,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013563220000000002,
      "loss": 3.0844,
      "step": 321840
    },
    {
      "epoch": 0.19454216931064652,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001356302,
      "loss": 3.0743,
      "step": 321850
    },
    {
      "epoch": 0.1945608125468047,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001356282,
      "loss": 3.1095,
      "step": 321860
    },
    {
      "epoch": 0.1945794557829629,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001356262,
      "loss": 3.2691,
      "step": 321870
    },
    {
      "epoch": 0.1945980990191211,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001356242,
      "loss": 3.0932,
      "step": 321880
    },
    {
      "epoch": 0.19461674225527928,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001356222,
      "loss": 3.075,
      "step": 321890
    },
    {
      "epoch": 0.19463538549143744,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001356202,
      "loss": 3.093,
      "step": 321900
    },
    {
      "epoch": 0.19465402872759563,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013561820000000002,
      "loss": 3.2531,
      "step": 321910
    },
    {
      "epoch": 0.19467267196375382,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001356162,
      "loss": 2.9764,
      "step": 321920
    },
    {
      "epoch": 0.194691315199912,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001356142,
      "loss": 3.2397,
      "step": 321930
    },
    {
      "epoch": 0.1947099584360702,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001356122,
      "loss": 3.06,
      "step": 321940
    },
    {
      "epoch": 0.19472860167222836,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001356102,
      "loss": 3.0515,
      "step": 321950
    },
    {
      "epoch": 0.19474724490838655,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013560820000000002,
      "loss": 3.0203,
      "step": 321960
    },
    {
      "epoch": 0.19476588814454474,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013560620000000002,
      "loss": 3.128,
      "step": 321970
    },
    {
      "epoch": 0.19478453138070292,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001356042,
      "loss": 3.2384,
      "step": 321980
    },
    {
      "epoch": 0.1948031746168611,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001356022,
      "loss": 3.1949,
      "step": 321990
    },
    {
      "epoch": 0.1948218178530193,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001356002,
      "loss": 3.159,
      "step": 322000
    },
    {
      "epoch": 0.19484046108917746,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001355982,
      "loss": 3.1335,
      "step": 322010
    },
    {
      "epoch": 0.19485910432533565,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001355962,
      "loss": 3.103,
      "step": 322020
    },
    {
      "epoch": 0.19487774756149384,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013559420000000002,
      "loss": 3.2126,
      "step": 322030
    },
    {
      "epoch": 0.19489639079765203,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001355922,
      "loss": 3.1273,
      "step": 322040
    },
    {
      "epoch": 0.19491503403381022,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001355902,
      "loss": 3.2242,
      "step": 322050
    },
    {
      "epoch": 0.19493367726996838,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001355882,
      "loss": 3.0786,
      "step": 322060
    },
    {
      "epoch": 0.19495232050612657,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001355862,
      "loss": 3.1077,
      "step": 322070
    },
    {
      "epoch": 0.19497096374228476,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001355842,
      "loss": 3.1767,
      "step": 322080
    },
    {
      "epoch": 0.19498960697844295,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013558220000000002,
      "loss": 3.1153,
      "step": 322090
    },
    {
      "epoch": 0.19500825021460114,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001355802,
      "loss": 3.1826,
      "step": 322100
    },
    {
      "epoch": 0.1950268934507593,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001355782,
      "loss": 3.2304,
      "step": 322110
    },
    {
      "epoch": 0.1950455366869175,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001355762,
      "loss": 3.1714,
      "step": 322120
    },
    {
      "epoch": 0.19506417992307568,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001355742,
      "loss": 3.1263,
      "step": 322130
    },
    {
      "epoch": 0.19508282315923386,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013557220000000002,
      "loss": 3.2362,
      "step": 322140
    },
    {
      "epoch": 0.19510146639539205,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001355702,
      "loss": 3.1153,
      "step": 322150
    },
    {
      "epoch": 0.19512010963155021,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00013556820000000001,
      "loss": 2.9888,
      "step": 322160
    },
    {
      "epoch": 0.1951387528677084,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001355662,
      "loss": 3.07,
      "step": 322170
    },
    {
      "epoch": 0.1951573961038666,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001355642,
      "loss": 3.1696,
      "step": 322180
    },
    {
      "epoch": 0.19517603934002478,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013556220000000003,
      "loss": 3.2435,
      "step": 322190
    },
    {
      "epoch": 0.19519468257618297,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001355602,
      "loss": 3.1566,
      "step": 322200
    },
    {
      "epoch": 0.19521332581234116,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013555820000000002,
      "loss": 3.0881,
      "step": 322210
    },
    {
      "epoch": 0.19523196904849932,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001355562,
      "loss": 3.1437,
      "step": 322220
    },
    {
      "epoch": 0.1952506122846575,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001355542,
      "loss": 3.2139,
      "step": 322230
    },
    {
      "epoch": 0.1952692555208157,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001355522,
      "loss": 3.1468,
      "step": 322240
    },
    {
      "epoch": 0.1952878987569739,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001355502,
      "loss": 3.1051,
      "step": 322250
    },
    {
      "epoch": 0.19530654199313208,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013554820000000002,
      "loss": 3.2291,
      "step": 322260
    },
    {
      "epoch": 0.19532518522929024,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001355462,
      "loss": 3.1344,
      "step": 322270
    },
    {
      "epoch": 0.19534382846544843,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013554420000000002,
      "loss": 3.1901,
      "step": 322280
    },
    {
      "epoch": 0.19536247170160662,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013554219999999998,
      "loss": 3.1384,
      "step": 322290
    },
    {
      "epoch": 0.1953811149377648,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001355402,
      "loss": 3.0706,
      "step": 322300
    },
    {
      "epoch": 0.195399758173923,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001355382,
      "loss": 3.134,
      "step": 322310
    },
    {
      "epoch": 0.19541840141008116,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001355362,
      "loss": 3.1501,
      "step": 322320
    },
    {
      "epoch": 0.19543704464623934,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013553420000000002,
      "loss": 3.1133,
      "step": 322330
    },
    {
      "epoch": 0.19545568788239753,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001355322,
      "loss": 3.1854,
      "step": 322340
    },
    {
      "epoch": 0.19547433111855572,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001355302,
      "loss": 3.2011,
      "step": 322350
    },
    {
      "epoch": 0.1954929743547139,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001355282,
      "loss": 3.181,
      "step": 322360
    },
    {
      "epoch": 0.19551161759087207,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001355262,
      "loss": 3.0856,
      "step": 322370
    },
    {
      "epoch": 0.19553026082703026,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001355242,
      "loss": 3.1907,
      "step": 322380
    },
    {
      "epoch": 0.19554890406318845,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001355222,
      "loss": 3.1153,
      "step": 322390
    },
    {
      "epoch": 0.19556754729934664,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013552020000000002,
      "loss": 3.2205,
      "step": 322400
    },
    {
      "epoch": 0.19558619053550483,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001355182,
      "loss": 3.1129,
      "step": 322410
    },
    {
      "epoch": 0.19560483377166302,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001355162,
      "loss": 3.1174,
      "step": 322420
    },
    {
      "epoch": 0.19562347700782118,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001355142,
      "loss": 3.2442,
      "step": 322430
    },
    {
      "epoch": 0.19564212024397937,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001355122,
      "loss": 3.2798,
      "step": 322440
    },
    {
      "epoch": 0.19566076348013756,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013551020000000002,
      "loss": 3.1575,
      "step": 322450
    },
    {
      "epoch": 0.19567940671629575,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001355082,
      "loss": 3.0818,
      "step": 322460
    },
    {
      "epoch": 0.19569804995245393,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013550620000000001,
      "loss": 3.1661,
      "step": 322470
    },
    {
      "epoch": 0.1957166931886121,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001355042,
      "loss": 3.0854,
      "step": 322480
    },
    {
      "epoch": 0.19573533642477028,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001355022,
      "loss": 3.1574,
      "step": 322490
    },
    {
      "epoch": 0.19575397966092847,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001355002,
      "loss": 3.1101,
      "step": 322500
    },
    {
      "epoch": 0.19577262289708666,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001354982,
      "loss": 3.195,
      "step": 322510
    },
    {
      "epoch": 0.19579126613324485,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013549620000000002,
      "loss": 3.1761,
      "step": 322520
    },
    {
      "epoch": 0.195809909369403,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013549420000000001,
      "loss": 3.1893,
      "step": 322530
    },
    {
      "epoch": 0.1958285526055612,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001354922,
      "loss": 2.9254,
      "step": 322540
    },
    {
      "epoch": 0.1958471958417194,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001354902,
      "loss": 3.1937,
      "step": 322550
    },
    {
      "epoch": 0.19586583907787758,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001354882,
      "loss": 3.0942,
      "step": 322560
    },
    {
      "epoch": 0.19588448231403577,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001354862,
      "loss": 3.1086,
      "step": 322570
    },
    {
      "epoch": 0.19590312555019393,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013548420000000002,
      "loss": 3.2166,
      "step": 322580
    },
    {
      "epoch": 0.19592176878635212,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013548220000000002,
      "loss": 3.2648,
      "step": 322590
    },
    {
      "epoch": 0.1959404120225103,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001354802,
      "loss": 3.0944,
      "step": 322600
    },
    {
      "epoch": 0.1959590552586685,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001354782,
      "loss": 3.1519,
      "step": 322610
    },
    {
      "epoch": 0.19597769849482669,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001354762,
      "loss": 3.1127,
      "step": 322620
    },
    {
      "epoch": 0.19599634173098485,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001354742,
      "loss": 3.1721,
      "step": 322630
    },
    {
      "epoch": 0.19601498496714304,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001354722,
      "loss": 3.1723,
      "step": 322640
    },
    {
      "epoch": 0.19603362820330122,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013547020000000002,
      "loss": 3.0247,
      "step": 322650
    },
    {
      "epoch": 0.1960522714394594,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001354682,
      "loss": 3.139,
      "step": 322660
    },
    {
      "epoch": 0.1960709146756176,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001354662,
      "loss": 3.1669,
      "step": 322670
    },
    {
      "epoch": 0.1960895579117758,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001354642,
      "loss": 3.2678,
      "step": 322680
    },
    {
      "epoch": 0.19610820114793395,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001354622,
      "loss": 3.0263,
      "step": 322690
    },
    {
      "epoch": 0.19612684438409214,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013546020000000002,
      "loss": 3.0536,
      "step": 322700
    },
    {
      "epoch": 0.19614548762025033,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001354582,
      "loss": 3.075,
      "step": 322710
    },
    {
      "epoch": 0.19616413085640852,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001354562,
      "loss": 3.1921,
      "step": 322720
    },
    {
      "epoch": 0.1961827740925667,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001354542,
      "loss": 3.0227,
      "step": 322730
    },
    {
      "epoch": 0.19620141732872487,
      "grad_norm": 0.625,
      "learning_rate": 0.0001354522,
      "loss": 3.1018,
      "step": 322740
    },
    {
      "epoch": 0.19622006056488306,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013545020000000003,
      "loss": 3.1285,
      "step": 322750
    },
    {
      "epoch": 0.19623870380104125,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001354482,
      "loss": 3.1221,
      "step": 322760
    },
    {
      "epoch": 0.19625734703719944,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013544620000000002,
      "loss": 3.1021,
      "step": 322770
    },
    {
      "epoch": 0.19627599027335763,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013544419999999999,
      "loss": 3.1415,
      "step": 322780
    },
    {
      "epoch": 0.1962946335095158,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001354422,
      "loss": 3.0716,
      "step": 322790
    },
    {
      "epoch": 0.19631327674567398,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001354402,
      "loss": 2.9957,
      "step": 322800
    },
    {
      "epoch": 0.19633191998183216,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001354382,
      "loss": 3.1272,
      "step": 322810
    },
    {
      "epoch": 0.19635056321799035,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013543620000000002,
      "loss": 3.1336,
      "step": 322820
    },
    {
      "epoch": 0.19636920645414854,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001354342,
      "loss": 3.1858,
      "step": 322830
    },
    {
      "epoch": 0.1963878496903067,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013543220000000001,
      "loss": 3.1635,
      "step": 322840
    },
    {
      "epoch": 0.1964064929264649,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013543019999999998,
      "loss": 3.1929,
      "step": 322850
    },
    {
      "epoch": 0.19642513616262308,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001354282,
      "loss": 3.1971,
      "step": 322860
    },
    {
      "epoch": 0.19644377939878127,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001354262,
      "loss": 3.1505,
      "step": 322870
    },
    {
      "epoch": 0.19646242263493946,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001354242,
      "loss": 3.1612,
      "step": 322880
    },
    {
      "epoch": 0.19648106587109765,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013542220000000002,
      "loss": 3.1133,
      "step": 322890
    },
    {
      "epoch": 0.1964997091072558,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001354202,
      "loss": 3.2698,
      "step": 322900
    },
    {
      "epoch": 0.196518352343414,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001354182,
      "loss": 3.0816,
      "step": 322910
    },
    {
      "epoch": 0.1965369955795722,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001354162,
      "loss": 3.0649,
      "step": 322920
    },
    {
      "epoch": 0.19655563881573038,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001354142,
      "loss": 3.0112,
      "step": 322930
    },
    {
      "epoch": 0.19657428205188857,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013541220000000002,
      "loss": 3.0832,
      "step": 322940
    },
    {
      "epoch": 0.19659292528804673,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001354102,
      "loss": 3.0973,
      "step": 322950
    },
    {
      "epoch": 0.19661156852420492,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013540820000000002,
      "loss": 3.2329,
      "step": 322960
    },
    {
      "epoch": 0.1966302117603631,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001354062,
      "loss": 3.2199,
      "step": 322970
    },
    {
      "epoch": 0.1966488549965213,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001354042,
      "loss": 3.2336,
      "step": 322980
    },
    {
      "epoch": 0.19666749823267948,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001354022,
      "loss": 3.0104,
      "step": 322990
    },
    {
      "epoch": 0.19668614146883764,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001354002,
      "loss": 3.0131,
      "step": 323000
    },
    {
      "epoch": 0.19670478470499583,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013539820000000002,
      "loss": 3.2168,
      "step": 323010
    },
    {
      "epoch": 0.19672342794115402,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013539620000000002,
      "loss": 3.1404,
      "step": 323020
    },
    {
      "epoch": 0.1967420711773122,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001353942,
      "loss": 3.1174,
      "step": 323030
    },
    {
      "epoch": 0.1967607144134704,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001353922,
      "loss": 3.0538,
      "step": 323040
    },
    {
      "epoch": 0.19677935764962856,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001353902,
      "loss": 3.1835,
      "step": 323050
    },
    {
      "epoch": 0.19679800088578675,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001353882,
      "loss": 3.1129,
      "step": 323060
    },
    {
      "epoch": 0.19681664412194494,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001353862,
      "loss": 3.1467,
      "step": 323070
    },
    {
      "epoch": 0.19683528735810313,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013538420000000002,
      "loss": 3.1794,
      "step": 323080
    },
    {
      "epoch": 0.19685393059426132,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001353822,
      "loss": 3.1635,
      "step": 323090
    },
    {
      "epoch": 0.1968725738304195,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001353802,
      "loss": 3.2262,
      "step": 323100
    },
    {
      "epoch": 0.19689121706657767,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001353782,
      "loss": 3.1434,
      "step": 323110
    },
    {
      "epoch": 0.19690986030273586,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001353762,
      "loss": 3.0452,
      "step": 323120
    },
    {
      "epoch": 0.19692850353889405,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001353742,
      "loss": 3.1949,
      "step": 323130
    },
    {
      "epoch": 0.19694714677505223,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013537220000000002,
      "loss": 3.0819,
      "step": 323140
    },
    {
      "epoch": 0.19696579001121042,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013537020000000001,
      "loss": 3.0579,
      "step": 323150
    },
    {
      "epoch": 0.19698443324736858,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001353682,
      "loss": 3.0879,
      "step": 323160
    },
    {
      "epoch": 0.19700307648352677,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001353662,
      "loss": 3.1363,
      "step": 323170
    },
    {
      "epoch": 0.19702171971968496,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001353642,
      "loss": 3.045,
      "step": 323180
    },
    {
      "epoch": 0.19704036295584315,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013536220000000002,
      "loss": 3.1446,
      "step": 323190
    },
    {
      "epoch": 0.19705900619200134,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001353602,
      "loss": 3.1375,
      "step": 323200
    },
    {
      "epoch": 0.1970776494281595,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013535820000000001,
      "loss": 3.1076,
      "step": 323210
    },
    {
      "epoch": 0.1970962926643177,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001353562,
      "loss": 3.121,
      "step": 323220
    },
    {
      "epoch": 0.19711493590047588,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001353542,
      "loss": 3.0994,
      "step": 323230
    },
    {
      "epoch": 0.19713357913663407,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001353522,
      "loss": 3.0686,
      "step": 323240
    },
    {
      "epoch": 0.19715222237279226,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001353502,
      "loss": 3.0404,
      "step": 323250
    },
    {
      "epoch": 0.19717086560895042,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013534820000000002,
      "loss": 3.119,
      "step": 323260
    },
    {
      "epoch": 0.1971895088451086,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001353462,
      "loss": 3.0462,
      "step": 323270
    },
    {
      "epoch": 0.1972081520812668,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001353442,
      "loss": 3.0435,
      "step": 323280
    },
    {
      "epoch": 0.19722679531742499,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001353422,
      "loss": 3.2321,
      "step": 323290
    },
    {
      "epoch": 0.19724543855358317,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001353402,
      "loss": 3.1409,
      "step": 323300
    },
    {
      "epoch": 0.19726408178974136,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013533820000000003,
      "loss": 3.1209,
      "step": 323310
    },
    {
      "epoch": 0.19728272502589952,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001353362,
      "loss": 3.1823,
      "step": 323320
    },
    {
      "epoch": 0.1973013682620577,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013533420000000002,
      "loss": 3.0338,
      "step": 323330
    },
    {
      "epoch": 0.1973200114982159,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013533219999999998,
      "loss": 3.2055,
      "step": 323340
    },
    {
      "epoch": 0.1973386547343741,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001353302,
      "loss": 3.0973,
      "step": 323350
    },
    {
      "epoch": 0.19735729797053228,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001353282,
      "loss": 3.1463,
      "step": 323360
    },
    {
      "epoch": 0.19737594120669044,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001353262,
      "loss": 3.142,
      "step": 323370
    },
    {
      "epoch": 0.19739458444284863,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013532420000000002,
      "loss": 3.0946,
      "step": 323380
    },
    {
      "epoch": 0.19741322767900682,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001353222,
      "loss": 3.2133,
      "step": 323390
    },
    {
      "epoch": 0.197431870915165,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001353202,
      "loss": 3.1233,
      "step": 323400
    },
    {
      "epoch": 0.1974505141513232,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001353182,
      "loss": 3.2147,
      "step": 323410
    },
    {
      "epoch": 0.19746915738748136,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001353162,
      "loss": 3.0491,
      "step": 323420
    },
    {
      "epoch": 0.19748780062363955,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013531420000000003,
      "loss": 3.0444,
      "step": 323430
    },
    {
      "epoch": 0.19750644385979774,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001353122,
      "loss": 3.0263,
      "step": 323440
    },
    {
      "epoch": 0.19752508709595593,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013531020000000002,
      "loss": 3.1802,
      "step": 323450
    },
    {
      "epoch": 0.19754373033211411,
      "grad_norm": 0.84375,
      "learning_rate": 0.00013530819999999999,
      "loss": 3.1024,
      "step": 323460
    },
    {
      "epoch": 0.19756237356827228,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001353062,
      "loss": 3.1478,
      "step": 323470
    },
    {
      "epoch": 0.19758101680443046,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001353042,
      "loss": 3.1148,
      "step": 323480
    },
    {
      "epoch": 0.19759966004058865,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001353022,
      "loss": 3.0485,
      "step": 323490
    },
    {
      "epoch": 0.19761830327674684,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013530020000000002,
      "loss": 3.1529,
      "step": 323500
    },
    {
      "epoch": 0.19763694651290503,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001352982,
      "loss": 3.0848,
      "step": 323510
    },
    {
      "epoch": 0.1976555897490632,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013529620000000001,
      "loss": 3.0798,
      "step": 323520
    },
    {
      "epoch": 0.19767423298522138,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001352942,
      "loss": 3.0802,
      "step": 323530
    },
    {
      "epoch": 0.19769287622137957,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001352922,
      "loss": 3.114,
      "step": 323540
    },
    {
      "epoch": 0.19771151945753776,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001352902,
      "loss": 3.0694,
      "step": 323550
    },
    {
      "epoch": 0.19773016269369595,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001352882,
      "loss": 3.2109,
      "step": 323560
    },
    {
      "epoch": 0.19774880592985414,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013528620000000002,
      "loss": 2.9766,
      "step": 323570
    },
    {
      "epoch": 0.1977674491660123,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013528420000000002,
      "loss": 3.1599,
      "step": 323580
    },
    {
      "epoch": 0.1977860924021705,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001352822,
      "loss": 3.0657,
      "step": 323590
    },
    {
      "epoch": 0.19780473563832868,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001352802,
      "loss": 3.193,
      "step": 323600
    },
    {
      "epoch": 0.19782337887448687,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001352782,
      "loss": 3.1283,
      "step": 323610
    },
    {
      "epoch": 0.19784202211064505,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001352762,
      "loss": 2.9979,
      "step": 323620
    },
    {
      "epoch": 0.19786066534680322,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013527420000000002,
      "loss": 3.2069,
      "step": 323630
    },
    {
      "epoch": 0.1978793085829614,
      "grad_norm": 1.015625,
      "learning_rate": 0.00013527220000000002,
      "loss": 3.1207,
      "step": 323640
    },
    {
      "epoch": 0.1978979518191196,
      "grad_norm": 0.75,
      "learning_rate": 0.0001352702,
      "loss": 3.056,
      "step": 323650
    },
    {
      "epoch": 0.19791659505527778,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001352682,
      "loss": 3.1008,
      "step": 323660
    },
    {
      "epoch": 0.19793523829143597,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001352662,
      "loss": 3.0981,
      "step": 323670
    },
    {
      "epoch": 0.19795388152759413,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001352642,
      "loss": 3.1017,
      "step": 323680
    },
    {
      "epoch": 0.19797252476375232,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001352622,
      "loss": 3.248,
      "step": 323690
    },
    {
      "epoch": 0.1979911679999105,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013526020000000002,
      "loss": 3.2253,
      "step": 323700
    },
    {
      "epoch": 0.1980098112360687,
      "grad_norm": 0.75,
      "learning_rate": 0.0001352582,
      "loss": 3.2239,
      "step": 323710
    },
    {
      "epoch": 0.1980284544722269,
      "grad_norm": 0.75,
      "learning_rate": 0.0001352562,
      "loss": 3.1496,
      "step": 323720
    },
    {
      "epoch": 0.19804709770838505,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001352542,
      "loss": 3.074,
      "step": 323730
    },
    {
      "epoch": 0.19806574094454324,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001352522,
      "loss": 3.0047,
      "step": 323740
    },
    {
      "epoch": 0.19808438418070143,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013525020000000002,
      "loss": 3.1693,
      "step": 323750
    },
    {
      "epoch": 0.19810302741685962,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001352482,
      "loss": 3.0537,
      "step": 323760
    },
    {
      "epoch": 0.1981216706530178,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001352462,
      "loss": 3.1838,
      "step": 323770
    },
    {
      "epoch": 0.198140313889176,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001352442,
      "loss": 3.1669,
      "step": 323780
    },
    {
      "epoch": 0.19815895712533416,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001352422,
      "loss": 3.0166,
      "step": 323790
    },
    {
      "epoch": 0.19817760036149235,
      "grad_norm": 0.75,
      "learning_rate": 0.00013524020000000003,
      "loss": 3.1509,
      "step": 323800
    },
    {
      "epoch": 0.19819624359765053,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001352382,
      "loss": 3.1415,
      "step": 323810
    },
    {
      "epoch": 0.19821488683380872,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013523620000000002,
      "loss": 3.0288,
      "step": 323820
    },
    {
      "epoch": 0.1982335300699669,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001352342,
      "loss": 3.0941,
      "step": 323830
    },
    {
      "epoch": 0.19825217330612507,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001352322,
      "loss": 3.1417,
      "step": 323840
    },
    {
      "epoch": 0.19827081654228326,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001352302,
      "loss": 3.0416,
      "step": 323850
    },
    {
      "epoch": 0.19828945977844145,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001352282,
      "loss": 3.1254,
      "step": 323860
    },
    {
      "epoch": 0.19830810301459964,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013522620000000002,
      "loss": 3.0202,
      "step": 323870
    },
    {
      "epoch": 0.19832674625075783,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001352242,
      "loss": 3.2401,
      "step": 323880
    },
    {
      "epoch": 0.198345389486916,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013522220000000001,
      "loss": 3.0196,
      "step": 323890
    },
    {
      "epoch": 0.19836403272307418,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001352202,
      "loss": 3.0819,
      "step": 323900
    },
    {
      "epoch": 0.19838267595923237,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001352182,
      "loss": 3.1985,
      "step": 323910
    },
    {
      "epoch": 0.19840131919539056,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013521620000000003,
      "loss": 3.2503,
      "step": 323920
    },
    {
      "epoch": 0.19841996243154875,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001352142,
      "loss": 3.1994,
      "step": 323930
    },
    {
      "epoch": 0.1984386056677069,
      "grad_norm": 0.578125,
      "learning_rate": 0.00013521220000000002,
      "loss": 3.0186,
      "step": 323940
    },
    {
      "epoch": 0.1984572489038651,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001352102,
      "loss": 3.0883,
      "step": 323950
    },
    {
      "epoch": 0.19847589214002329,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001352082,
      "loss": 3.1127,
      "step": 323960
    },
    {
      "epoch": 0.19849453537618147,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001352062,
      "loss": 3.1376,
      "step": 323970
    },
    {
      "epoch": 0.19851317861233966,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001352042,
      "loss": 3.1097,
      "step": 323980
    },
    {
      "epoch": 0.19853182184849785,
      "grad_norm": 0.625,
      "learning_rate": 0.00013520220000000002,
      "loss": 3.1701,
      "step": 323990
    },
    {
      "epoch": 0.198550465084656,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001352002,
      "loss": 3.1128,
      "step": 324000
    },
    {
      "epoch": 0.1985691083208142,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013519820000000002,
      "loss": 3.0348,
      "step": 324010
    },
    {
      "epoch": 0.1985877515569724,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001351962,
      "loss": 3.0676,
      "step": 324020
    },
    {
      "epoch": 0.19860639479313058,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001351942,
      "loss": 3.1496,
      "step": 324030
    },
    {
      "epoch": 0.19862503802928877,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001351922,
      "loss": 3.0166,
      "step": 324040
    },
    {
      "epoch": 0.19864368126544693,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001351902,
      "loss": 3.1729,
      "step": 324050
    },
    {
      "epoch": 0.19866232450160512,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013518820000000002,
      "loss": 3.0213,
      "step": 324060
    },
    {
      "epoch": 0.1986809677377633,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001351862,
      "loss": 3.2024,
      "step": 324070
    },
    {
      "epoch": 0.1986996109739215,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001351842,
      "loss": 3.1188,
      "step": 324080
    },
    {
      "epoch": 0.1987182542100797,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001351822,
      "loss": 3.1657,
      "step": 324090
    },
    {
      "epoch": 0.19873689744623785,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001351802,
      "loss": 3.1346,
      "step": 324100
    },
    {
      "epoch": 0.19875554068239604,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001351782,
      "loss": 3.1393,
      "step": 324110
    },
    {
      "epoch": 0.19877418391855423,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001351762,
      "loss": 3.2109,
      "step": 324120
    },
    {
      "epoch": 0.19879282715471241,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013517420000000002,
      "loss": 2.9689,
      "step": 324130
    },
    {
      "epoch": 0.1988114703908706,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001351722,
      "loss": 3.0915,
      "step": 324140
    },
    {
      "epoch": 0.19883011362702877,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001351702,
      "loss": 3.0576,
      "step": 324150
    },
    {
      "epoch": 0.19884875686318695,
      "grad_norm": 0.625,
      "learning_rate": 0.0001351682,
      "loss": 3.1378,
      "step": 324160
    },
    {
      "epoch": 0.19886740009934514,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001351662,
      "loss": 3.0333,
      "step": 324170
    },
    {
      "epoch": 0.19888604333550333,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001351642,
      "loss": 3.2304,
      "step": 324180
    },
    {
      "epoch": 0.19890468657166152,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013516220000000002,
      "loss": 3.0607,
      "step": 324190
    },
    {
      "epoch": 0.19892332980781968,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013516020000000001,
      "loss": 3.211,
      "step": 324200
    },
    {
      "epoch": 0.19894197304397787,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001351582,
      "loss": 3.0324,
      "step": 324210
    },
    {
      "epoch": 0.19896061628013606,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001351562,
      "loss": 3.0221,
      "step": 324220
    },
    {
      "epoch": 0.19897925951629425,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001351542,
      "loss": 3.1246,
      "step": 324230
    },
    {
      "epoch": 0.19899790275245244,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001351522,
      "loss": 3.1168,
      "step": 324240
    },
    {
      "epoch": 0.19901654598861063,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001351502,
      "loss": 3.072,
      "step": 324250
    },
    {
      "epoch": 0.1990351892247688,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013514820000000001,
      "loss": 3.1863,
      "step": 324260
    },
    {
      "epoch": 0.19905383246092698,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001351462,
      "loss": 3.0997,
      "step": 324270
    },
    {
      "epoch": 0.19907247569708517,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001351442,
      "loss": 3.1312,
      "step": 324280
    },
    {
      "epoch": 0.19909111893324336,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001351422,
      "loss": 3.1539,
      "step": 324290
    },
    {
      "epoch": 0.19910976216940154,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001351402,
      "loss": 3.1855,
      "step": 324300
    },
    {
      "epoch": 0.1991284054055597,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013513820000000002,
      "loss": 3.1695,
      "step": 324310
    },
    {
      "epoch": 0.1991470486417179,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001351362,
      "loss": 3.308,
      "step": 324320
    },
    {
      "epoch": 0.19916569187787608,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001351342,
      "loss": 3.1745,
      "step": 324330
    },
    {
      "epoch": 0.19918433511403427,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001351322,
      "loss": 3.1976,
      "step": 324340
    },
    {
      "epoch": 0.19920297835019246,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001351302,
      "loss": 3.1924,
      "step": 324350
    },
    {
      "epoch": 0.19922162158635062,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013512820000000003,
      "loss": 3.1716,
      "step": 324360
    },
    {
      "epoch": 0.1992402648225088,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001351262,
      "loss": 3.1871,
      "step": 324370
    },
    {
      "epoch": 0.199258908058667,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013512420000000002,
      "loss": 3.2256,
      "step": 324380
    },
    {
      "epoch": 0.1992775512948252,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001351222,
      "loss": 3.1792,
      "step": 324390
    },
    {
      "epoch": 0.19929619453098338,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001351202,
      "loss": 3.1177,
      "step": 324400
    },
    {
      "epoch": 0.19931483776714154,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013511820000000003,
      "loss": 3.133,
      "step": 324410
    },
    {
      "epoch": 0.19933348100329973,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001351162,
      "loss": 3.191,
      "step": 324420
    },
    {
      "epoch": 0.19935212423945792,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013511420000000002,
      "loss": 3.1469,
      "step": 324430
    },
    {
      "epoch": 0.1993707674756161,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001351122,
      "loss": 3.2133,
      "step": 324440
    },
    {
      "epoch": 0.1993894107117743,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001351102,
      "loss": 2.9768,
      "step": 324450
    },
    {
      "epoch": 0.19940805394793248,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001351082,
      "loss": 3.0201,
      "step": 324460
    },
    {
      "epoch": 0.19942669718409065,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001351062,
      "loss": 3.0967,
      "step": 324470
    },
    {
      "epoch": 0.19944534042024883,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013510420000000003,
      "loss": 3.1063,
      "step": 324480
    },
    {
      "epoch": 0.19946398365640702,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001351022,
      "loss": 3.1384,
      "step": 324490
    },
    {
      "epoch": 0.1994826268925652,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013510020000000002,
      "loss": 3.1053,
      "step": 324500
    },
    {
      "epoch": 0.1995012701287234,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001350982,
      "loss": 3.1369,
      "step": 324510
    },
    {
      "epoch": 0.19951991336488156,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001350962,
      "loss": 3.1408,
      "step": 324520
    },
    {
      "epoch": 0.19953855660103975,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001350942,
      "loss": 3.0972,
      "step": 324530
    },
    {
      "epoch": 0.19955719983719794,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001350922,
      "loss": 3.2386,
      "step": 324540
    },
    {
      "epoch": 0.19957584307335613,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013509020000000002,
      "loss": 3.0378,
      "step": 324550
    },
    {
      "epoch": 0.19959448630951432,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001350882,
      "loss": 3.0485,
      "step": 324560
    },
    {
      "epoch": 0.19961312954567248,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013508620000000001,
      "loss": 3.109,
      "step": 324570
    },
    {
      "epoch": 0.19963177278183067,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001350842,
      "loss": 3.1465,
      "step": 324580
    },
    {
      "epoch": 0.19965041601798886,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001350822,
      "loss": 3.2881,
      "step": 324590
    },
    {
      "epoch": 0.19966905925414705,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001350802,
      "loss": 3.1456,
      "step": 324600
    },
    {
      "epoch": 0.19968770249030524,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001350782,
      "loss": 3.1571,
      "step": 324610
    },
    {
      "epoch": 0.1997063457264634,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013507620000000002,
      "loss": 3.0855,
      "step": 324620
    },
    {
      "epoch": 0.19972498896262159,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013507420000000002,
      "loss": 3.0009,
      "step": 324630
    },
    {
      "epoch": 0.19974363219877977,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001350722,
      "loss": 3.0358,
      "step": 324640
    },
    {
      "epoch": 0.19976227543493796,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001350702,
      "loss": 3.162,
      "step": 324650
    },
    {
      "epoch": 0.19978091867109615,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001350682,
      "loss": 3.145,
      "step": 324660
    },
    {
      "epoch": 0.19979956190725434,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001350662,
      "loss": 3.0099,
      "step": 324670
    },
    {
      "epoch": 0.1998182051434125,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001350642,
      "loss": 3.1129,
      "step": 324680
    },
    {
      "epoch": 0.1998368483795707,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013506220000000002,
      "loss": 3.164,
      "step": 324690
    },
    {
      "epoch": 0.19985549161572888,
      "grad_norm": 0.75,
      "learning_rate": 0.0001350602,
      "loss": 3.185,
      "step": 324700
    },
    {
      "epoch": 0.19987413485188707,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001350582,
      "loss": 3.0812,
      "step": 324710
    },
    {
      "epoch": 0.19989277808804526,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001350562,
      "loss": 3.0728,
      "step": 324720
    },
    {
      "epoch": 0.19991142132420342,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001350542,
      "loss": 3.1133,
      "step": 324730
    },
    {
      "epoch": 0.1999300645603616,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001350522,
      "loss": 3.0943,
      "step": 324740
    },
    {
      "epoch": 0.1999487077965198,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013505020000000002,
      "loss": 3.2024,
      "step": 324750
    },
    {
      "epoch": 0.199967351032678,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001350482,
      "loss": 3.1017,
      "step": 324760
    },
    {
      "epoch": 0.19998599426883618,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001350462,
      "loss": 3.1775,
      "step": 324770
    },
    {
      "epoch": 0.20000463750499434,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001350442,
      "loss": 3.0853,
      "step": 324780
    },
    {
      "epoch": 0.20002328074115253,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001350422,
      "loss": 3.155,
      "step": 324790
    },
    {
      "epoch": 0.20004192397731071,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013504020000000002,
      "loss": 3.2538,
      "step": 324800
    },
    {
      "epoch": 0.2000605672134689,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001350382,
      "loss": 3.1623,
      "step": 324810
    },
    {
      "epoch": 0.2000792104496271,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001350362,
      "loss": 3.1786,
      "step": 324820
    },
    {
      "epoch": 0.20009785368578525,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001350342,
      "loss": 3.074,
      "step": 324830
    },
    {
      "epoch": 0.20011649692194344,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001350322,
      "loss": 3.1906,
      "step": 324840
    },
    {
      "epoch": 0.20013514015810163,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001350302,
      "loss": 3.0433,
      "step": 324850
    },
    {
      "epoch": 0.20015378339425982,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001350282,
      "loss": 3.1055,
      "step": 324860
    },
    {
      "epoch": 0.200172426630418,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013502620000000002,
      "loss": 3.0976,
      "step": 324870
    },
    {
      "epoch": 0.20019106986657617,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013502420000000001,
      "loss": 2.9984,
      "step": 324880
    },
    {
      "epoch": 0.20020971310273436,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001350222,
      "loss": 3.0955,
      "step": 324890
    },
    {
      "epoch": 0.20022835633889255,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001350202,
      "loss": 3.1281,
      "step": 324900
    },
    {
      "epoch": 0.20024699957505074,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001350182,
      "loss": 3.2685,
      "step": 324910
    },
    {
      "epoch": 0.20026564281120893,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013501620000000002,
      "loss": 3.2164,
      "step": 324920
    },
    {
      "epoch": 0.20028428604736712,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001350142,
      "loss": 3.2034,
      "step": 324930
    },
    {
      "epoch": 0.20030292928352528,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013501220000000001,
      "loss": 3.0843,
      "step": 324940
    },
    {
      "epoch": 0.20032157251968347,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001350102,
      "loss": 3.0955,
      "step": 324950
    },
    {
      "epoch": 0.20034021575584166,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001350082,
      "loss": 3.0042,
      "step": 324960
    },
    {
      "epoch": 0.20035885899199984,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013500620000000003,
      "loss": 3.2671,
      "step": 324970
    },
    {
      "epoch": 0.20037750222815803,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001350042,
      "loss": 3.0285,
      "step": 324980
    },
    {
      "epoch": 0.2003961454643162,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013500220000000002,
      "loss": 3.0028,
      "step": 324990
    },
    {
      "epoch": 0.20041478870047438,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001350002,
      "loss": 3.1683,
      "step": 325000
    },
    {
      "epoch": 0.20043343193663257,
      "grad_norm": 0.75,
      "learning_rate": 0.0001349982,
      "loss": 3.136,
      "step": 325010
    },
    {
      "epoch": 0.20045207517279076,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001349962,
      "loss": 3.1178,
      "step": 325020
    },
    {
      "epoch": 0.20047071840894895,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001349942,
      "loss": 3.1919,
      "step": 325030
    },
    {
      "epoch": 0.2004893616451071,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013499220000000003,
      "loss": 2.9878,
      "step": 325040
    },
    {
      "epoch": 0.2005080048812653,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001349902,
      "loss": 3.2925,
      "step": 325050
    },
    {
      "epoch": 0.2005266481174235,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013498820000000002,
      "loss": 3.1392,
      "step": 325060
    },
    {
      "epoch": 0.20054529135358168,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013498619999999999,
      "loss": 3.0911,
      "step": 325070
    },
    {
      "epoch": 0.20056393458973987,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001349842,
      "loss": 3.1991,
      "step": 325080
    },
    {
      "epoch": 0.20058257782589803,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001349822,
      "loss": 3.1615,
      "step": 325090
    },
    {
      "epoch": 0.20060122106205622,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001349802,
      "loss": 3.0859,
      "step": 325100
    },
    {
      "epoch": 0.2006198642982144,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013497820000000002,
      "loss": 3.1311,
      "step": 325110
    },
    {
      "epoch": 0.2006385075343726,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001349762,
      "loss": 3.1099,
      "step": 325120
    },
    {
      "epoch": 0.20065715077053078,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001349742,
      "loss": 3.0408,
      "step": 325130
    },
    {
      "epoch": 0.20067579400668897,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001349722,
      "loss": 3.1896,
      "step": 325140
    },
    {
      "epoch": 0.20069443724284713,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001349702,
      "loss": 3.0749,
      "step": 325150
    },
    {
      "epoch": 0.20071308047900532,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001349682,
      "loss": 3.1682,
      "step": 325160
    },
    {
      "epoch": 0.2007317237151635,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001349662,
      "loss": 3.0664,
      "step": 325170
    },
    {
      "epoch": 0.2007503669513217,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013496420000000002,
      "loss": 3.1813,
      "step": 325180
    },
    {
      "epoch": 0.2007690101874799,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013496220000000001,
      "loss": 3.1431,
      "step": 325190
    },
    {
      "epoch": 0.20078765342363805,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001349602,
      "loss": 3.0705,
      "step": 325200
    },
    {
      "epoch": 0.20080629665979624,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001349582,
      "loss": 2.9915,
      "step": 325210
    },
    {
      "epoch": 0.20082493989595443,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001349562,
      "loss": 2.9894,
      "step": 325220
    },
    {
      "epoch": 0.20084358313211262,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001349542,
      "loss": 3.0344,
      "step": 325230
    },
    {
      "epoch": 0.2008622263682708,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013495220000000002,
      "loss": 3.1669,
      "step": 325240
    },
    {
      "epoch": 0.20088086960442897,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013495020000000001,
      "loss": 3.0952,
      "step": 325250
    },
    {
      "epoch": 0.20089951284058716,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001349482,
      "loss": 3.0287,
      "step": 325260
    },
    {
      "epoch": 0.20091815607674535,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001349462,
      "loss": 3.1739,
      "step": 325270
    },
    {
      "epoch": 0.20093679931290354,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001349442,
      "loss": 2.9985,
      "step": 325280
    },
    {
      "epoch": 0.20095544254906172,
      "grad_norm": 0.75,
      "learning_rate": 0.0001349422,
      "loss": 2.9615,
      "step": 325290
    },
    {
      "epoch": 0.20097408578521989,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001349402,
      "loss": 3.0986,
      "step": 325300
    },
    {
      "epoch": 0.20099272902137807,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013493820000000002,
      "loss": 3.2791,
      "step": 325310
    },
    {
      "epoch": 0.20101137225753626,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001349362,
      "loss": 3.1222,
      "step": 325320
    },
    {
      "epoch": 0.20103001549369445,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001349342,
      "loss": 3.1519,
      "step": 325330
    },
    {
      "epoch": 0.20104865872985264,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001349322,
      "loss": 3.0097,
      "step": 325340
    },
    {
      "epoch": 0.20106730196601083,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001349302,
      "loss": 3.2441,
      "step": 325350
    },
    {
      "epoch": 0.201085945202169,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013492820000000002,
      "loss": 3.1814,
      "step": 325360
    },
    {
      "epoch": 0.20110458843832718,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013492620000000002,
      "loss": 3.1736,
      "step": 325370
    },
    {
      "epoch": 0.20112323167448537,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001349242,
      "loss": 3.0984,
      "step": 325380
    },
    {
      "epoch": 0.20114187491064356,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001349222,
      "loss": 3.2199,
      "step": 325390
    },
    {
      "epoch": 0.20116051814680175,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001349202,
      "loss": 3.1486,
      "step": 325400
    },
    {
      "epoch": 0.2011791613829599,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013491820000000003,
      "loss": 3.097,
      "step": 325410
    },
    {
      "epoch": 0.2011978046191181,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001349162,
      "loss": 3.111,
      "step": 325420
    },
    {
      "epoch": 0.2012164478552763,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013491420000000002,
      "loss": 3.2779,
      "step": 325430
    },
    {
      "epoch": 0.20123509109143448,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001349122,
      "loss": 3.0755,
      "step": 325440
    },
    {
      "epoch": 0.20125373432759266,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001349102,
      "loss": 2.9444,
      "step": 325450
    },
    {
      "epoch": 0.20127237756375083,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001349082,
      "loss": 3.0952,
      "step": 325460
    },
    {
      "epoch": 0.20129102079990902,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001349062,
      "loss": 3.1689,
      "step": 325470
    },
    {
      "epoch": 0.2013096640360672,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013490420000000002,
      "loss": 3.146,
      "step": 325480
    },
    {
      "epoch": 0.2013283072722254,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001349022,
      "loss": 3.2597,
      "step": 325490
    },
    {
      "epoch": 0.20134695050838358,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001349002,
      "loss": 3.1102,
      "step": 325500
    },
    {
      "epoch": 0.20136559374454174,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001348982,
      "loss": 3.1727,
      "step": 325510
    },
    {
      "epoch": 0.20138423698069993,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001348962,
      "loss": 3.1038,
      "step": 325520
    },
    {
      "epoch": 0.20140288021685812,
      "grad_norm": 0.75,
      "learning_rate": 0.00013489420000000003,
      "loss": 3.0431,
      "step": 325530
    },
    {
      "epoch": 0.2014215234530163,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001348922,
      "loss": 3.0215,
      "step": 325540
    },
    {
      "epoch": 0.2014401666891745,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013489020000000002,
      "loss": 3.1549,
      "step": 325550
    },
    {
      "epoch": 0.20145880992533266,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001348882,
      "loss": 3.156,
      "step": 325560
    },
    {
      "epoch": 0.20147745316149085,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001348862,
      "loss": 3.189,
      "step": 325570
    },
    {
      "epoch": 0.20149609639764904,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001348842,
      "loss": 3.066,
      "step": 325580
    },
    {
      "epoch": 0.20151473963380723,
      "grad_norm": 0.625,
      "learning_rate": 0.0001348822,
      "loss": 3.1945,
      "step": 325590
    },
    {
      "epoch": 0.20153338286996542,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013488020000000002,
      "loss": 3.1147,
      "step": 325600
    },
    {
      "epoch": 0.2015520261061236,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001348782,
      "loss": 3.1006,
      "step": 325610
    },
    {
      "epoch": 0.20157066934228177,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013487620000000001,
      "loss": 3.2098,
      "step": 325620
    },
    {
      "epoch": 0.20158931257843996,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001348742,
      "loss": 3.0741,
      "step": 325630
    },
    {
      "epoch": 0.20160795581459814,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001348722,
      "loss": 3.1147,
      "step": 325640
    },
    {
      "epoch": 0.20162659905075633,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001348702,
      "loss": 3.01,
      "step": 325650
    },
    {
      "epoch": 0.20164524228691452,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001348682,
      "loss": 3.1556,
      "step": 325660
    },
    {
      "epoch": 0.20166388552307268,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013486620000000002,
      "loss": 3.2066,
      "step": 325670
    },
    {
      "epoch": 0.20168252875923087,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001348642,
      "loss": 3.0918,
      "step": 325680
    },
    {
      "epoch": 0.20170117199538906,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001348622,
      "loss": 3.1454,
      "step": 325690
    },
    {
      "epoch": 0.20171981523154725,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001348602,
      "loss": 3.2557,
      "step": 325700
    },
    {
      "epoch": 0.20173845846770544,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001348582,
      "loss": 3.084,
      "step": 325710
    },
    {
      "epoch": 0.2017571017038636,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001348562,
      "loss": 2.9902,
      "step": 325720
    },
    {
      "epoch": 0.2017757449400218,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001348542,
      "loss": 3.2074,
      "step": 325730
    },
    {
      "epoch": 0.20179438817617998,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013485220000000002,
      "loss": 3.0791,
      "step": 325740
    },
    {
      "epoch": 0.20181303141233817,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001348502,
      "loss": 3.1206,
      "step": 325750
    },
    {
      "epoch": 0.20183167464849636,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001348482,
      "loss": 3.1256,
      "step": 325760
    },
    {
      "epoch": 0.20185031788465452,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001348462,
      "loss": 3.1386,
      "step": 325770
    },
    {
      "epoch": 0.2018689611208127,
      "grad_norm": 0.625,
      "learning_rate": 0.0001348442,
      "loss": 3.1297,
      "step": 325780
    },
    {
      "epoch": 0.2018876043569709,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001348422,
      "loss": 3.1954,
      "step": 325790
    },
    {
      "epoch": 0.20190624759312908,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013484020000000002,
      "loss": 3.1389,
      "step": 325800
    },
    {
      "epoch": 0.20192489082928727,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001348382,
      "loss": 3.2604,
      "step": 325810
    },
    {
      "epoch": 0.20194353406544546,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001348362,
      "loss": 3.0812,
      "step": 325820
    },
    {
      "epoch": 0.20196217730160362,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001348342,
      "loss": 3.2247,
      "step": 325830
    },
    {
      "epoch": 0.2019808205377618,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001348322,
      "loss": 3.1304,
      "step": 325840
    },
    {
      "epoch": 0.20199946377392,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001348302,
      "loss": 3.0923,
      "step": 325850
    },
    {
      "epoch": 0.2020181070100782,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013482820000000002,
      "loss": 3.1731,
      "step": 325860
    },
    {
      "epoch": 0.20203675024623638,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013482620000000001,
      "loss": 3.0584,
      "step": 325870
    },
    {
      "epoch": 0.20205539348239454,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001348242,
      "loss": 3.2718,
      "step": 325880
    },
    {
      "epoch": 0.20207403671855273,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001348222,
      "loss": 3.0526,
      "step": 325890
    },
    {
      "epoch": 0.20209267995471092,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001348202,
      "loss": 3.0489,
      "step": 325900
    },
    {
      "epoch": 0.2021113231908691,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001348182,
      "loss": 3.1503,
      "step": 325910
    },
    {
      "epoch": 0.2021299664270273,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013481620000000002,
      "loss": 3.0743,
      "step": 325920
    },
    {
      "epoch": 0.20214860966318546,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013481420000000001,
      "loss": 2.9804,
      "step": 325930
    },
    {
      "epoch": 0.20216725289934365,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001348122,
      "loss": 3.0225,
      "step": 325940
    },
    {
      "epoch": 0.20218589613550184,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001348102,
      "loss": 3.1048,
      "step": 325950
    },
    {
      "epoch": 0.20220453937166002,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001348082,
      "loss": 2.9362,
      "step": 325960
    },
    {
      "epoch": 0.2022231826078182,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013480620000000002,
      "loss": 3.2017,
      "step": 325970
    },
    {
      "epoch": 0.20224182584397637,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001348042,
      "loss": 3.1198,
      "step": 325980
    },
    {
      "epoch": 0.20226046908013456,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013480220000000002,
      "loss": 3.1243,
      "step": 325990
    },
    {
      "epoch": 0.20227911231629275,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001348002,
      "loss": 3.0616,
      "step": 326000
    },
    {
      "epoch": 0.20229775555245094,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001347982,
      "loss": 3.3154,
      "step": 326010
    },
    {
      "epoch": 0.20231639878860913,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013479620000000003,
      "loss": 3.2019,
      "step": 326020
    },
    {
      "epoch": 0.20233504202476732,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001347942,
      "loss": 3.1245,
      "step": 326030
    },
    {
      "epoch": 0.20235368526092548,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013479220000000002,
      "loss": 3.0727,
      "step": 326040
    },
    {
      "epoch": 0.20237232849708367,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001347902,
      "loss": 3.1243,
      "step": 326050
    },
    {
      "epoch": 0.20239097173324186,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001347882,
      "loss": 3.1674,
      "step": 326060
    },
    {
      "epoch": 0.20240961496940005,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001347862,
      "loss": 3.2538,
      "step": 326070
    },
    {
      "epoch": 0.20242825820555824,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001347842,
      "loss": 3.1774,
      "step": 326080
    },
    {
      "epoch": 0.2024469014417164,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013478220000000003,
      "loss": 3.0913,
      "step": 326090
    },
    {
      "epoch": 0.2024655446778746,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001347802,
      "loss": 3.0361,
      "step": 326100
    },
    {
      "epoch": 0.20248418791403278,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013477820000000002,
      "loss": 3.1349,
      "step": 326110
    },
    {
      "epoch": 0.20250283115019097,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013477619999999999,
      "loss": 3.0652,
      "step": 326120
    },
    {
      "epoch": 0.20252147438634915,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001347742,
      "loss": 3.0867,
      "step": 326130
    },
    {
      "epoch": 0.20254011762250732,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001347722,
      "loss": 3.0261,
      "step": 326140
    },
    {
      "epoch": 0.2025587608586655,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001347702,
      "loss": 3.0263,
      "step": 326150
    },
    {
      "epoch": 0.2025774040948237,
      "grad_norm": 0.75,
      "learning_rate": 0.00013476820000000002,
      "loss": 3.0402,
      "step": 326160
    },
    {
      "epoch": 0.20259604733098188,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001347662,
      "loss": 3.1112,
      "step": 326170
    },
    {
      "epoch": 0.20261469056714007,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001347642,
      "loss": 3.0852,
      "step": 326180
    },
    {
      "epoch": 0.20263333380329823,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001347622,
      "loss": 3.0918,
      "step": 326190
    },
    {
      "epoch": 0.20265197703945642,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001347602,
      "loss": 3.1195,
      "step": 326200
    },
    {
      "epoch": 0.2026706202756146,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001347582,
      "loss": 3.2069,
      "step": 326210
    },
    {
      "epoch": 0.2026892635117728,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001347562,
      "loss": 3.1173,
      "step": 326220
    },
    {
      "epoch": 0.202707906747931,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013475420000000002,
      "loss": 3.1314,
      "step": 326230
    },
    {
      "epoch": 0.20272654998408915,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013475220000000001,
      "loss": 2.9955,
      "step": 326240
    },
    {
      "epoch": 0.20274519322024734,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001347502,
      "loss": 3.231,
      "step": 326250
    },
    {
      "epoch": 0.20276383645640553,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001347482,
      "loss": 3.2481,
      "step": 326260
    },
    {
      "epoch": 0.20278247969256372,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001347462,
      "loss": 3.0867,
      "step": 326270
    },
    {
      "epoch": 0.2028011229287219,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001347442,
      "loss": 3.1207,
      "step": 326280
    },
    {
      "epoch": 0.2028197661648801,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001347422,
      "loss": 3.1086,
      "step": 326290
    },
    {
      "epoch": 0.20283840940103826,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013474020000000001,
      "loss": 3.2176,
      "step": 326300
    },
    {
      "epoch": 0.20285705263719644,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001347382,
      "loss": 3.1442,
      "step": 326310
    },
    {
      "epoch": 0.20287569587335463,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001347362,
      "loss": 3.1302,
      "step": 326320
    },
    {
      "epoch": 0.20289433910951282,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001347342,
      "loss": 3.136,
      "step": 326330
    },
    {
      "epoch": 0.202912982345671,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001347322,
      "loss": 3.1124,
      "step": 326340
    },
    {
      "epoch": 0.20293162558182917,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001347302,
      "loss": 3.2617,
      "step": 326350
    },
    {
      "epoch": 0.20295026881798736,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013472820000000002,
      "loss": 3.1884,
      "step": 326360
    },
    {
      "epoch": 0.20296891205414555,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001347262,
      "loss": 3.245,
      "step": 326370
    },
    {
      "epoch": 0.20298755529030374,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001347242,
      "loss": 3.1112,
      "step": 326380
    },
    {
      "epoch": 0.20300619852646193,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001347222,
      "loss": 3.18,
      "step": 326390
    },
    {
      "epoch": 0.2030248417626201,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001347202,
      "loss": 3.208,
      "step": 326400
    },
    {
      "epoch": 0.20304348499877828,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013471820000000002,
      "loss": 3.1016,
      "step": 326410
    },
    {
      "epoch": 0.20306212823493647,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013471620000000002,
      "loss": 3.1901,
      "step": 326420
    },
    {
      "epoch": 0.20308077147109466,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001347142,
      "loss": 3.0315,
      "step": 326430
    },
    {
      "epoch": 0.20309941470725285,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001347122,
      "loss": 3.136,
      "step": 326440
    },
    {
      "epoch": 0.203118057943411,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001347102,
      "loss": 3.1939,
      "step": 326450
    },
    {
      "epoch": 0.2031367011795692,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001347082,
      "loss": 3.1841,
      "step": 326460
    },
    {
      "epoch": 0.20315534441572738,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001347062,
      "loss": 3.0808,
      "step": 326470
    },
    {
      "epoch": 0.20317398765188557,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013470420000000002,
      "loss": 3.1531,
      "step": 326480
    },
    {
      "epoch": 0.20319263088804376,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001347022,
      "loss": 3.2442,
      "step": 326490
    },
    {
      "epoch": 0.20321127412420195,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001347002,
      "loss": 3.1538,
      "step": 326500
    },
    {
      "epoch": 0.2032299173603601,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001346982,
      "loss": 3.168,
      "step": 326510
    },
    {
      "epoch": 0.2032485605965183,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001346962,
      "loss": 3.0835,
      "step": 326520
    },
    {
      "epoch": 0.2032672038326765,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013469420000000002,
      "loss": 2.9758,
      "step": 326530
    },
    {
      "epoch": 0.20328584706883468,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001346922,
      "loss": 3.2,
      "step": 326540
    },
    {
      "epoch": 0.20330449030499287,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013469020000000001,
      "loss": 3.0585,
      "step": 326550
    },
    {
      "epoch": 0.20332313354115103,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001346882,
      "loss": 3.1286,
      "step": 326560
    },
    {
      "epoch": 0.20334177677730922,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001346862,
      "loss": 3.1165,
      "step": 326570
    },
    {
      "epoch": 0.2033604200134674,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013468420000000003,
      "loss": 2.9909,
      "step": 326580
    },
    {
      "epoch": 0.2033790632496256,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001346822,
      "loss": 3.1319,
      "step": 326590
    },
    {
      "epoch": 0.20339770648578379,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013468020000000002,
      "loss": 3.1771,
      "step": 326600
    },
    {
      "epoch": 0.20341634972194195,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001346782,
      "loss": 3.2076,
      "step": 326610
    },
    {
      "epoch": 0.20343499295810014,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001346762,
      "loss": 3.195,
      "step": 326620
    },
    {
      "epoch": 0.20345363619425832,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001346742,
      "loss": 3.1299,
      "step": 326630
    },
    {
      "epoch": 0.20347227943041651,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001346722,
      "loss": 3.0945,
      "step": 326640
    },
    {
      "epoch": 0.2034909226665747,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00013467020000000002,
      "loss": 3.192,
      "step": 326650
    },
    {
      "epoch": 0.20350956590273286,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001346682,
      "loss": 3.1822,
      "step": 326660
    },
    {
      "epoch": 0.20352820913889105,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013466620000000002,
      "loss": 3.0503,
      "step": 326670
    },
    {
      "epoch": 0.20354685237504924,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013466419999999998,
      "loss": 3.1091,
      "step": 326680
    },
    {
      "epoch": 0.20356549561120743,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001346622,
      "loss": 3.174,
      "step": 326690
    },
    {
      "epoch": 0.20358413884736562,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001346602,
      "loss": 3.0815,
      "step": 326700
    },
    {
      "epoch": 0.2036027820835238,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001346582,
      "loss": 3.0352,
      "step": 326710
    },
    {
      "epoch": 0.20362142531968197,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013465620000000002,
      "loss": 3.0835,
      "step": 326720
    },
    {
      "epoch": 0.20364006855584016,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001346542,
      "loss": 3.14,
      "step": 326730
    },
    {
      "epoch": 0.20365871179199835,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001346522,
      "loss": 3.0659,
      "step": 326740
    },
    {
      "epoch": 0.20367735502815654,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001346502,
      "loss": 3.154,
      "step": 326750
    },
    {
      "epoch": 0.20369599826431473,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001346482,
      "loss": 3.1318,
      "step": 326760
    },
    {
      "epoch": 0.2037146415004729,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001346462,
      "loss": 3.1094,
      "step": 326770
    },
    {
      "epoch": 0.20373328473663108,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001346442,
      "loss": 3.2693,
      "step": 326780
    },
    {
      "epoch": 0.20375192797278927,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013464220000000002,
      "loss": 3.0416,
      "step": 326790
    },
    {
      "epoch": 0.20377057120894745,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001346402,
      "loss": 3.2133,
      "step": 326800
    },
    {
      "epoch": 0.20378921444510564,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001346382,
      "loss": 3.1205,
      "step": 326810
    },
    {
      "epoch": 0.2038078576812638,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001346362,
      "loss": 3.0929,
      "step": 326820
    },
    {
      "epoch": 0.203826500917422,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001346342,
      "loss": 3.1085,
      "step": 326830
    },
    {
      "epoch": 0.20384514415358018,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001346322,
      "loss": 3.1485,
      "step": 326840
    },
    {
      "epoch": 0.20386378738973837,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013463020000000002,
      "loss": 3.131,
      "step": 326850
    },
    {
      "epoch": 0.20388243062589656,
      "grad_norm": 0.75,
      "learning_rate": 0.0001346282,
      "loss": 2.9381,
      "step": 326860
    },
    {
      "epoch": 0.20390107386205472,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001346262,
      "loss": 3.0147,
      "step": 326870
    },
    {
      "epoch": 0.2039197170982129,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001346242,
      "loss": 3.1815,
      "step": 326880
    },
    {
      "epoch": 0.2039383603343711,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001346222,
      "loss": 3.1538,
      "step": 326890
    },
    {
      "epoch": 0.2039570035705293,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001346202,
      "loss": 3.091,
      "step": 326900
    },
    {
      "epoch": 0.20397564680668748,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013461820000000002,
      "loss": 3.1973,
      "step": 326910
    },
    {
      "epoch": 0.20399429004284564,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013461620000000001,
      "loss": 3.0889,
      "step": 326920
    },
    {
      "epoch": 0.20401293327900383,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001346142,
      "loss": 3.1799,
      "step": 326930
    },
    {
      "epoch": 0.20403157651516202,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001346122,
      "loss": 3.1535,
      "step": 326940
    },
    {
      "epoch": 0.2040502197513202,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001346102,
      "loss": 3.0673,
      "step": 326950
    },
    {
      "epoch": 0.2040688629874784,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001346082,
      "loss": 3.1216,
      "step": 326960
    },
    {
      "epoch": 0.20408750622363658,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013460620000000002,
      "loss": 3.1554,
      "step": 326970
    },
    {
      "epoch": 0.20410614945979474,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013460420000000001,
      "loss": 3.2168,
      "step": 326980
    },
    {
      "epoch": 0.20412479269595293,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001346022,
      "loss": 3.1156,
      "step": 326990
    },
    {
      "epoch": 0.20414343593211112,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001346002,
      "loss": 3.0366,
      "step": 327000
    },
    {
      "epoch": 0.2041620791682693,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001345982,
      "loss": 3.2048,
      "step": 327010
    },
    {
      "epoch": 0.2041807224044275,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013459620000000002,
      "loss": 3.1713,
      "step": 327020
    },
    {
      "epoch": 0.20419936564058566,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001345942,
      "loss": 3.105,
      "step": 327030
    },
    {
      "epoch": 0.20421800887674385,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013459220000000002,
      "loss": 3.0726,
      "step": 327040
    },
    {
      "epoch": 0.20423665211290204,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001345902,
      "loss": 3.1317,
      "step": 327050
    },
    {
      "epoch": 0.20425529534906023,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001345882,
      "loss": 3.2303,
      "step": 327060
    },
    {
      "epoch": 0.20427393858521842,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001345862,
      "loss": 2.9974,
      "step": 327070
    },
    {
      "epoch": 0.20429258182137658,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001345842,
      "loss": 3.1905,
      "step": 327080
    },
    {
      "epoch": 0.20431122505753477,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013458220000000002,
      "loss": 3.213,
      "step": 327090
    },
    {
      "epoch": 0.20432986829369296,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001345802,
      "loss": 3.0276,
      "step": 327100
    },
    {
      "epoch": 0.20434851152985115,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001345782,
      "loss": 3.0728,
      "step": 327110
    },
    {
      "epoch": 0.20436715476600933,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001345762,
      "loss": 3.1289,
      "step": 327120
    },
    {
      "epoch": 0.2043857980021675,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001345742,
      "loss": 3.216,
      "step": 327130
    },
    {
      "epoch": 0.20440444123832568,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013457220000000003,
      "loss": 3.1703,
      "step": 327140
    },
    {
      "epoch": 0.20442308447448387,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001345702,
      "loss": 3.1685,
      "step": 327150
    },
    {
      "epoch": 0.20444172771064206,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013456820000000002,
      "loss": 3.0074,
      "step": 327160
    },
    {
      "epoch": 0.20446037094680025,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013456619999999999,
      "loss": 3.1153,
      "step": 327170
    },
    {
      "epoch": 0.20447901418295844,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001345642,
      "loss": 3.1037,
      "step": 327180
    },
    {
      "epoch": 0.2044976574191166,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001345622,
      "loss": 3.1249,
      "step": 327190
    },
    {
      "epoch": 0.2045163006552748,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001345602,
      "loss": 3.0788,
      "step": 327200
    },
    {
      "epoch": 0.20453494389143298,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013455820000000002,
      "loss": 3.1801,
      "step": 327210
    },
    {
      "epoch": 0.20455358712759117,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001345562,
      "loss": 3.0825,
      "step": 327220
    },
    {
      "epoch": 0.20457223036374936,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001345542,
      "loss": 3.0799,
      "step": 327230
    },
    {
      "epoch": 0.20459087359990752,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001345522,
      "loss": 3.1554,
      "step": 327240
    },
    {
      "epoch": 0.2046095168360657,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001345502,
      "loss": 3.039,
      "step": 327250
    },
    {
      "epoch": 0.2046281600722239,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001345482,
      "loss": 3.2283,
      "step": 327260
    },
    {
      "epoch": 0.20464680330838209,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001345462,
      "loss": 3.0621,
      "step": 327270
    },
    {
      "epoch": 0.20466544654454027,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013454420000000002,
      "loss": 3.1809,
      "step": 327280
    },
    {
      "epoch": 0.20468408978069844,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001345422,
      "loss": 3.1548,
      "step": 327290
    },
    {
      "epoch": 0.20470273301685663,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001345402,
      "loss": 3.1848,
      "step": 327300
    },
    {
      "epoch": 0.20472137625301481,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001345382,
      "loss": 3.1284,
      "step": 327310
    },
    {
      "epoch": 0.204740019489173,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001345362,
      "loss": 3.173,
      "step": 327320
    },
    {
      "epoch": 0.2047586627253312,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001345342,
      "loss": 3.1134,
      "step": 327330
    },
    {
      "epoch": 0.20477730596148935,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001345322,
      "loss": 3.2569,
      "step": 327340
    },
    {
      "epoch": 0.20479594919764754,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013453020000000002,
      "loss": 3.0193,
      "step": 327350
    },
    {
      "epoch": 0.20481459243380573,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001345282,
      "loss": 3.1671,
      "step": 327360
    },
    {
      "epoch": 0.20483323566996392,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001345262,
      "loss": 3.0246,
      "step": 327370
    },
    {
      "epoch": 0.2048518789061221,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001345242,
      "loss": 3.1184,
      "step": 327380
    },
    {
      "epoch": 0.2048705221422803,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001345222,
      "loss": 3.2418,
      "step": 327390
    },
    {
      "epoch": 0.20488916537843846,
      "grad_norm": 0.75,
      "learning_rate": 0.00013452020000000002,
      "loss": 3.1311,
      "step": 327400
    },
    {
      "epoch": 0.20490780861459665,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013451820000000002,
      "loss": 3.0517,
      "step": 327410
    },
    {
      "epoch": 0.20492645185075484,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001345162,
      "loss": 3.0681,
      "step": 327420
    },
    {
      "epoch": 0.20494509508691303,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001345142,
      "loss": 3.121,
      "step": 327430
    },
    {
      "epoch": 0.20496373832307122,
      "grad_norm": 0.875,
      "learning_rate": 0.0001345122,
      "loss": 3.0654,
      "step": 327440
    },
    {
      "epoch": 0.20498238155922938,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001345102,
      "loss": 3.122,
      "step": 327450
    },
    {
      "epoch": 0.20500102479538757,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013450820000000002,
      "loss": 2.9967,
      "step": 327460
    },
    {
      "epoch": 0.20501966803154575,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013450620000000002,
      "loss": 3.193,
      "step": 327470
    },
    {
      "epoch": 0.20503831126770394,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001345042,
      "loss": 3.1465,
      "step": 327480
    },
    {
      "epoch": 0.20505695450386213,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001345022,
      "loss": 3.0151,
      "step": 327490
    },
    {
      "epoch": 0.2050755977400203,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001345002,
      "loss": 3.1164,
      "step": 327500
    },
    {
      "epoch": 0.20509424097617848,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001344982,
      "loss": 3.0553,
      "step": 327510
    },
    {
      "epoch": 0.20511288421233667,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001344962,
      "loss": 3.1408,
      "step": 327520
    },
    {
      "epoch": 0.20513152744849486,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013449420000000002,
      "loss": 3.1999,
      "step": 327530
    },
    {
      "epoch": 0.20515017068465305,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001344922,
      "loss": 3.1746,
      "step": 327540
    },
    {
      "epoch": 0.2051688139208112,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001344902,
      "loss": 3.183,
      "step": 327550
    },
    {
      "epoch": 0.2051874571569694,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001344882,
      "loss": 3.1172,
      "step": 327560
    },
    {
      "epoch": 0.2052061003931276,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001344862,
      "loss": 3.1631,
      "step": 327570
    },
    {
      "epoch": 0.20522474362928578,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00013448420000000002,
      "loss": 3.1706,
      "step": 327580
    },
    {
      "epoch": 0.20524338686544397,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001344822,
      "loss": 3.1938,
      "step": 327590
    },
    {
      "epoch": 0.20526203010160213,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013448020000000001,
      "loss": 3.1563,
      "step": 327600
    },
    {
      "epoch": 0.20528067333776032,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001344782,
      "loss": 3.1406,
      "step": 327610
    },
    {
      "epoch": 0.2052993165739185,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001344762,
      "loss": 3.1274,
      "step": 327620
    },
    {
      "epoch": 0.2053179598100767,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013447420000000003,
      "loss": 3.1029,
      "step": 327630
    },
    {
      "epoch": 0.20533660304623488,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001344722,
      "loss": 3.0706,
      "step": 327640
    },
    {
      "epoch": 0.20535524628239307,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013447020000000002,
      "loss": 3.1243,
      "step": 327650
    },
    {
      "epoch": 0.20537388951855123,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001344682,
      "loss": 3.1073,
      "step": 327660
    },
    {
      "epoch": 0.20539253275470942,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001344662,
      "loss": 3.1768,
      "step": 327670
    },
    {
      "epoch": 0.2054111759908676,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001344642,
      "loss": 3.2207,
      "step": 327680
    },
    {
      "epoch": 0.2054298192270258,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001344622,
      "loss": 3.1391,
      "step": 327690
    },
    {
      "epoch": 0.205448462463184,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013446020000000002,
      "loss": 3.0564,
      "step": 327700
    },
    {
      "epoch": 0.20546710569934215,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001344582,
      "loss": 2.9457,
      "step": 327710
    },
    {
      "epoch": 0.20548574893550034,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013445620000000002,
      "loss": 3.1489,
      "step": 327720
    },
    {
      "epoch": 0.20550439217165853,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013445419999999998,
      "loss": 3.1766,
      "step": 327730
    },
    {
      "epoch": 0.20552303540781672,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001344522,
      "loss": 3.1621,
      "step": 327740
    },
    {
      "epoch": 0.2055416786439749,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001344502,
      "loss": 3.0178,
      "step": 327750
    },
    {
      "epoch": 0.20556032188013307,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001344482,
      "loss": 3.0169,
      "step": 327760
    },
    {
      "epoch": 0.20557896511629126,
      "grad_norm": 1.03125,
      "learning_rate": 0.00013444620000000002,
      "loss": 3.1634,
      "step": 327770
    },
    {
      "epoch": 0.20559760835244945,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001344442,
      "loss": 3.0768,
      "step": 327780
    },
    {
      "epoch": 0.20561625158860763,
      "grad_norm": 0.75,
      "learning_rate": 0.0001344422,
      "loss": 3.2076,
      "step": 327790
    },
    {
      "epoch": 0.20563489482476582,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001344402,
      "loss": 3.137,
      "step": 327800
    },
    {
      "epoch": 0.20565353806092398,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001344382,
      "loss": 3.1246,
      "step": 327810
    },
    {
      "epoch": 0.20567218129708217,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001344362,
      "loss": 3.2264,
      "step": 327820
    },
    {
      "epoch": 0.20569082453324036,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001344342,
      "loss": 2.9975,
      "step": 327830
    },
    {
      "epoch": 0.20570946776939855,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013443220000000002,
      "loss": 2.9869,
      "step": 327840
    },
    {
      "epoch": 0.20572811100555674,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001344302,
      "loss": 3.1237,
      "step": 327850
    },
    {
      "epoch": 0.20574675424171493,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001344282,
      "loss": 3.0608,
      "step": 327860
    },
    {
      "epoch": 0.2057653974778731,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001344262,
      "loss": 3.1519,
      "step": 327870
    },
    {
      "epoch": 0.20578404071403128,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001344242,
      "loss": 3.0793,
      "step": 327880
    },
    {
      "epoch": 0.20580268395018947,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013442220000000002,
      "loss": 3.1379,
      "step": 327890
    },
    {
      "epoch": 0.20582132718634766,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001344202,
      "loss": 3.1692,
      "step": 327900
    },
    {
      "epoch": 0.20583997042250585,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001344182,
      "loss": 3.1301,
      "step": 327910
    },
    {
      "epoch": 0.205858613658664,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001344162,
      "loss": 3.102,
      "step": 327920
    },
    {
      "epoch": 0.2058772568948222,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001344142,
      "loss": 3.0812,
      "step": 327930
    },
    {
      "epoch": 0.20589590013098039,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001344122,
      "loss": 3.064,
      "step": 327940
    },
    {
      "epoch": 0.20591454336713857,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001344102,
      "loss": 3.0611,
      "step": 327950
    },
    {
      "epoch": 0.20593318660329676,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013440820000000002,
      "loss": 3.1214,
      "step": 327960
    },
    {
      "epoch": 0.20595182983945493,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013440620000000001,
      "loss": 3.2366,
      "step": 327970
    },
    {
      "epoch": 0.20597047307561311,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001344042,
      "loss": 3.061,
      "step": 327980
    },
    {
      "epoch": 0.2059891163117713,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001344022,
      "loss": 3.0898,
      "step": 327990
    },
    {
      "epoch": 0.2060077595479295,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001344002,
      "loss": 3.0974,
      "step": 328000
    },
    {
      "epoch": 0.20602640278408768,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001343982,
      "loss": 3.015,
      "step": 328010
    },
    {
      "epoch": 0.20604504602024584,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013439620000000002,
      "loss": 3.1179,
      "step": 328020
    },
    {
      "epoch": 0.20606368925640403,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013439420000000002,
      "loss": 3.1587,
      "step": 328030
    },
    {
      "epoch": 0.20608233249256222,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001343922,
      "loss": 3.1647,
      "step": 328040
    },
    {
      "epoch": 0.2061009757287204,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001343902,
      "loss": 3.1619,
      "step": 328050
    },
    {
      "epoch": 0.2061196189648786,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001343882,
      "loss": 3.0935,
      "step": 328060
    },
    {
      "epoch": 0.2061382622010368,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001343862,
      "loss": 3.1939,
      "step": 328070
    },
    {
      "epoch": 0.20615690543719495,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001343842,
      "loss": 3.1023,
      "step": 328080
    },
    {
      "epoch": 0.20617554867335314,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013438220000000002,
      "loss": 3.0916,
      "step": 328090
    },
    {
      "epoch": 0.20619419190951133,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001343802,
      "loss": 2.908,
      "step": 328100
    },
    {
      "epoch": 0.20621283514566952,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001343782,
      "loss": 3.1177,
      "step": 328110
    },
    {
      "epoch": 0.2062314783818277,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001343762,
      "loss": 3.0897,
      "step": 328120
    },
    {
      "epoch": 0.20625012161798587,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001343742,
      "loss": 3.161,
      "step": 328130
    },
    {
      "epoch": 0.20626876485414405,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013437220000000002,
      "loss": 3.1685,
      "step": 328140
    },
    {
      "epoch": 0.20628740809030224,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001343702,
      "loss": 3.1088,
      "step": 328150
    },
    {
      "epoch": 0.20630605132646043,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001343682,
      "loss": 3.1143,
      "step": 328160
    },
    {
      "epoch": 0.20632469456261862,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001343662,
      "loss": 3.0648,
      "step": 328170
    },
    {
      "epoch": 0.20634333779877678,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001343642,
      "loss": 3.0546,
      "step": 328180
    },
    {
      "epoch": 0.20636198103493497,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013436220000000003,
      "loss": 3.1946,
      "step": 328190
    },
    {
      "epoch": 0.20638062427109316,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001343602,
      "loss": 3.1435,
      "step": 328200
    },
    {
      "epoch": 0.20639926750725135,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013435820000000002,
      "loss": 3.1964,
      "step": 328210
    },
    {
      "epoch": 0.20641791074340954,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013435619999999999,
      "loss": 3.1154,
      "step": 328220
    },
    {
      "epoch": 0.2064365539795677,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001343542,
      "loss": 3.1023,
      "step": 328230
    },
    {
      "epoch": 0.2064551972157259,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001343522,
      "loss": 3.0121,
      "step": 328240
    },
    {
      "epoch": 0.20647384045188408,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001343502,
      "loss": 3.2344,
      "step": 328250
    },
    {
      "epoch": 0.20649248368804227,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013434820000000002,
      "loss": 3.1548,
      "step": 328260
    },
    {
      "epoch": 0.20651112692420046,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001343462,
      "loss": 3.1761,
      "step": 328270
    },
    {
      "epoch": 0.20652977016035864,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013434420000000001,
      "loss": 3.0787,
      "step": 328280
    },
    {
      "epoch": 0.2065484133965168,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013434219999999998,
      "loss": 3.1247,
      "step": 328290
    },
    {
      "epoch": 0.206567056632675,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001343402,
      "loss": 3.0376,
      "step": 328300
    },
    {
      "epoch": 0.20658569986883318,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001343382,
      "loss": 3.1331,
      "step": 328310
    },
    {
      "epoch": 0.20660434310499137,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001343362,
      "loss": 3.1343,
      "step": 328320
    },
    {
      "epoch": 0.20662298634114956,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013433420000000002,
      "loss": 3.1356,
      "step": 328330
    },
    {
      "epoch": 0.20664162957730772,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001343322,
      "loss": 3.1434,
      "step": 328340
    },
    {
      "epoch": 0.2066602728134659,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001343302,
      "loss": 3.0785,
      "step": 328350
    },
    {
      "epoch": 0.2066789160496241,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001343282,
      "loss": 3.2182,
      "step": 328360
    },
    {
      "epoch": 0.2066975592857823,
      "grad_norm": 0.75,
      "learning_rate": 0.0001343262,
      "loss": 3.1351,
      "step": 328370
    },
    {
      "epoch": 0.20671620252194048,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013432420000000002,
      "loss": 3.1636,
      "step": 328380
    },
    {
      "epoch": 0.20673484575809864,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001343222,
      "loss": 3.1186,
      "step": 328390
    },
    {
      "epoch": 0.20675348899425683,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013432020000000002,
      "loss": 3.0845,
      "step": 328400
    },
    {
      "epoch": 0.20677213223041502,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001343182,
      "loss": 3.1585,
      "step": 328410
    },
    {
      "epoch": 0.2067907754665732,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001343162,
      "loss": 3.1363,
      "step": 328420
    },
    {
      "epoch": 0.2068094187027314,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001343142,
      "loss": 3.0824,
      "step": 328430
    },
    {
      "epoch": 0.20682806193888956,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001343122,
      "loss": 2.9323,
      "step": 328440
    },
    {
      "epoch": 0.20684670517504775,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013431020000000002,
      "loss": 3.0335,
      "step": 328450
    },
    {
      "epoch": 0.20686534841120593,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013430820000000002,
      "loss": 3.1735,
      "step": 328460
    },
    {
      "epoch": 0.20688399164736412,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001343062,
      "loss": 3.0961,
      "step": 328470
    },
    {
      "epoch": 0.2069026348835223,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001343042,
      "loss": 3.1478,
      "step": 328480
    },
    {
      "epoch": 0.20692127811968047,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001343022,
      "loss": 3.1433,
      "step": 328490
    },
    {
      "epoch": 0.20693992135583866,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001343002,
      "loss": 3.1131,
      "step": 328500
    },
    {
      "epoch": 0.20695856459199685,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001342982,
      "loss": 3.04,
      "step": 328510
    },
    {
      "epoch": 0.20697720782815504,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013429620000000002,
      "loss": 3.0818,
      "step": 328520
    },
    {
      "epoch": 0.20699585106431323,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001342942,
      "loss": 3.1038,
      "step": 328530
    },
    {
      "epoch": 0.20701449430047142,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001342922,
      "loss": 3.1497,
      "step": 328540
    },
    {
      "epoch": 0.20703313753662958,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001342902,
      "loss": 3.1203,
      "step": 328550
    },
    {
      "epoch": 0.20705178077278777,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001342882,
      "loss": 3.015,
      "step": 328560
    },
    {
      "epoch": 0.20707042400894596,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001342862,
      "loss": 3.0742,
      "step": 328570
    },
    {
      "epoch": 0.20708906724510415,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013428420000000002,
      "loss": 3.1026,
      "step": 328580
    },
    {
      "epoch": 0.20710771048126234,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001342822,
      "loss": 3.2274,
      "step": 328590
    },
    {
      "epoch": 0.2071263537174205,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001342802,
      "loss": 3.0969,
      "step": 328600
    },
    {
      "epoch": 0.2071449969535787,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001342782,
      "loss": 2.9699,
      "step": 328610
    },
    {
      "epoch": 0.20716364018973688,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001342762,
      "loss": 3.1571,
      "step": 328620
    },
    {
      "epoch": 0.20718228342589506,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013427420000000002,
      "loss": 3.0894,
      "step": 328630
    },
    {
      "epoch": 0.20720092666205325,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001342722,
      "loss": 3.1564,
      "step": 328640
    },
    {
      "epoch": 0.20721956989821141,
      "grad_norm": 0.625,
      "learning_rate": 0.00013427020000000001,
      "loss": 3.0329,
      "step": 328650
    },
    {
      "epoch": 0.2072382131343696,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001342682,
      "loss": 3.1328,
      "step": 328660
    },
    {
      "epoch": 0.2072568563705278,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001342662,
      "loss": 3.1683,
      "step": 328670
    },
    {
      "epoch": 0.20727549960668598,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001342642,
      "loss": 3.0537,
      "step": 328680
    },
    {
      "epoch": 0.20729414284284417,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001342622,
      "loss": 3.1973,
      "step": 328690
    },
    {
      "epoch": 0.20731278607900233,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013426020000000002,
      "loss": 3.205,
      "step": 328700
    },
    {
      "epoch": 0.20733142931516052,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001342582,
      "loss": 3.1345,
      "step": 328710
    },
    {
      "epoch": 0.2073500725513187,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001342562,
      "loss": 3.3134,
      "step": 328720
    },
    {
      "epoch": 0.2073687157874769,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001342542,
      "loss": 3.1899,
      "step": 328730
    },
    {
      "epoch": 0.2073873590236351,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001342522,
      "loss": 3.0618,
      "step": 328740
    },
    {
      "epoch": 0.20740600225979328,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013425020000000002,
      "loss": 3.1162,
      "step": 328750
    },
    {
      "epoch": 0.20742464549595144,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001342482,
      "loss": 3.1046,
      "step": 328760
    },
    {
      "epoch": 0.20744328873210963,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013424620000000002,
      "loss": 3.1795,
      "step": 328770
    },
    {
      "epoch": 0.20746193196826782,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013424419999999998,
      "loss": 3.1634,
      "step": 328780
    },
    {
      "epoch": 0.207480575204426,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001342422,
      "loss": 3.1525,
      "step": 328790
    },
    {
      "epoch": 0.2074992184405842,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001342402,
      "loss": 3.1135,
      "step": 328800
    },
    {
      "epoch": 0.20751786167674235,
      "grad_norm": 0.75,
      "learning_rate": 0.0001342382,
      "loss": 3.2549,
      "step": 328810
    },
    {
      "epoch": 0.20753650491290054,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013423620000000002,
      "loss": 3.1756,
      "step": 328820
    },
    {
      "epoch": 0.20755514814905873,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001342342,
      "loss": 3.0754,
      "step": 328830
    },
    {
      "epoch": 0.20757379138521692,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001342322,
      "loss": 3.1826,
      "step": 328840
    },
    {
      "epoch": 0.2075924346213751,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001342302,
      "loss": 3.0962,
      "step": 328850
    },
    {
      "epoch": 0.20761107785753327,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001342282,
      "loss": 3.1501,
      "step": 328860
    },
    {
      "epoch": 0.20762972109369146,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013422620000000003,
      "loss": 3.1041,
      "step": 328870
    },
    {
      "epoch": 0.20764836432984965,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001342242,
      "loss": 3.0994,
      "step": 328880
    },
    {
      "epoch": 0.20766700756600784,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013422220000000002,
      "loss": 3.0889,
      "step": 328890
    },
    {
      "epoch": 0.20768565080216603,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013422019999999999,
      "loss": 3.195,
      "step": 328900
    },
    {
      "epoch": 0.2077042940383242,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001342182,
      "loss": 3.0441,
      "step": 328910
    },
    {
      "epoch": 0.20772293727448238,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001342162,
      "loss": 3.1545,
      "step": 328920
    },
    {
      "epoch": 0.20774158051064057,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001342142,
      "loss": 3.1295,
      "step": 328930
    },
    {
      "epoch": 0.20776022374679876,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013421220000000002,
      "loss": 3.0782,
      "step": 328940
    },
    {
      "epoch": 0.20777886698295694,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001342102,
      "loss": 3.1427,
      "step": 328950
    },
    {
      "epoch": 0.20779751021911513,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013420820000000001,
      "loss": 3.1858,
      "step": 328960
    },
    {
      "epoch": 0.2078161534552733,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001342062,
      "loss": 3.2461,
      "step": 328970
    },
    {
      "epoch": 0.20783479669143148,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001342042,
      "loss": 3.0646,
      "step": 328980
    },
    {
      "epoch": 0.20785343992758967,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001342022,
      "loss": 3.1149,
      "step": 328990
    },
    {
      "epoch": 0.20787208316374786,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001342002,
      "loss": 3.1599,
      "step": 329000
    },
    {
      "epoch": 0.20789072639990605,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013419820000000002,
      "loss": 3.2245,
      "step": 329010
    },
    {
      "epoch": 0.2079093696360642,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013419620000000001,
      "loss": 3.1519,
      "step": 329020
    },
    {
      "epoch": 0.2079280128722224,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001341942,
      "loss": 3.1996,
      "step": 329030
    },
    {
      "epoch": 0.2079466561083806,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001341922,
      "loss": 3.2956,
      "step": 329040
    },
    {
      "epoch": 0.20796529934453878,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001341902,
      "loss": 3.1707,
      "step": 329050
    },
    {
      "epoch": 0.20798394258069697,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001341882,
      "loss": 3.1446,
      "step": 329060
    },
    {
      "epoch": 0.20800258581685513,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013418620000000002,
      "loss": 3.0593,
      "step": 329070
    },
    {
      "epoch": 0.20802122905301332,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013418420000000002,
      "loss": 3.2076,
      "step": 329080
    },
    {
      "epoch": 0.2080398722891715,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001341822,
      "loss": 3.1878,
      "step": 329090
    },
    {
      "epoch": 0.2080585155253297,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001341802,
      "loss": 3.0344,
      "step": 329100
    },
    {
      "epoch": 0.20807715876148788,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001341782,
      "loss": 3.0214,
      "step": 329110
    },
    {
      "epoch": 0.20809580199764605,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001341762,
      "loss": 3.0589,
      "step": 329120
    },
    {
      "epoch": 0.20811444523380423,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001341742,
      "loss": 3.1237,
      "step": 329130
    },
    {
      "epoch": 0.20813308846996242,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013417220000000002,
      "loss": 3.0935,
      "step": 329140
    },
    {
      "epoch": 0.2081517317061206,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001341702,
      "loss": 3.1319,
      "step": 329150
    },
    {
      "epoch": 0.2081703749422788,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001341682,
      "loss": 3.2273,
      "step": 329160
    },
    {
      "epoch": 0.20818901817843696,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001341662,
      "loss": 3.197,
      "step": 329170
    },
    {
      "epoch": 0.20820766141459515,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001341642,
      "loss": 3.2081,
      "step": 329180
    },
    {
      "epoch": 0.20822630465075334,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013416220000000002,
      "loss": 3.0743,
      "step": 329190
    },
    {
      "epoch": 0.20824494788691153,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001341602,
      "loss": 3.1056,
      "step": 329200
    },
    {
      "epoch": 0.20826359112306972,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001341582,
      "loss": 3.2178,
      "step": 329210
    },
    {
      "epoch": 0.2082822343592279,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001341562,
      "loss": 3.1399,
      "step": 329220
    },
    {
      "epoch": 0.20830087759538607,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001341542,
      "loss": 3.1442,
      "step": 329230
    },
    {
      "epoch": 0.20831952083154426,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013415220000000003,
      "loss": 3.0662,
      "step": 329240
    },
    {
      "epoch": 0.20833816406770245,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001341502,
      "loss": 3.1586,
      "step": 329250
    },
    {
      "epoch": 0.20835680730386064,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013414820000000002,
      "loss": 3.2274,
      "step": 329260
    },
    {
      "epoch": 0.20837545054001883,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013414619999999999,
      "loss": 3.1492,
      "step": 329270
    },
    {
      "epoch": 0.208394093776177,
      "grad_norm": 0.75,
      "learning_rate": 0.0001341442,
      "loss": 3.0102,
      "step": 329280
    },
    {
      "epoch": 0.20841273701233518,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001341422,
      "loss": 3.1137,
      "step": 329290
    },
    {
      "epoch": 0.20843138024849336,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001341402,
      "loss": 3.209,
      "step": 329300
    },
    {
      "epoch": 0.20845002348465155,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013413820000000002,
      "loss": 3.1096,
      "step": 329310
    },
    {
      "epoch": 0.20846866672080974,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001341362,
      "loss": 3.168,
      "step": 329320
    },
    {
      "epoch": 0.2084873099569679,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013413420000000001,
      "loss": 3.1562,
      "step": 329330
    },
    {
      "epoch": 0.2085059531931261,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001341322,
      "loss": 3.139,
      "step": 329340
    },
    {
      "epoch": 0.20852459642928428,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001341302,
      "loss": 3.1035,
      "step": 329350
    },
    {
      "epoch": 0.20854323966544247,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013412820000000003,
      "loss": 3.1958,
      "step": 329360
    },
    {
      "epoch": 0.20856188290160066,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001341262,
      "loss": 3.1929,
      "step": 329370
    },
    {
      "epoch": 0.20858052613775882,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013412420000000002,
      "loss": 3.303,
      "step": 329380
    },
    {
      "epoch": 0.208599169373917,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001341222,
      "loss": 3.1403,
      "step": 329390
    },
    {
      "epoch": 0.2086178126100752,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001341202,
      "loss": 3.1264,
      "step": 329400
    },
    {
      "epoch": 0.2086364558462334,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001341182,
      "loss": 3.0787,
      "step": 329410
    },
    {
      "epoch": 0.20865509908239158,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001341162,
      "loss": 3.0858,
      "step": 329420
    },
    {
      "epoch": 0.20867374231854977,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013411420000000002,
      "loss": 3.1223,
      "step": 329430
    },
    {
      "epoch": 0.20869238555470793,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001341122,
      "loss": 3.1933,
      "step": 329440
    },
    {
      "epoch": 0.20871102879086612,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013411020000000002,
      "loss": 3.1281,
      "step": 329450
    },
    {
      "epoch": 0.2087296720270243,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001341082,
      "loss": 3.0895,
      "step": 329460
    },
    {
      "epoch": 0.2087483152631825,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001341062,
      "loss": 3.2046,
      "step": 329470
    },
    {
      "epoch": 0.20876695849934068,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001341042,
      "loss": 3.1229,
      "step": 329480
    },
    {
      "epoch": 0.20878560173549884,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001341022,
      "loss": 3.1065,
      "step": 329490
    },
    {
      "epoch": 0.20880424497165703,
      "grad_norm": 0.75,
      "learning_rate": 0.00013410020000000002,
      "loss": 3.1354,
      "step": 329500
    },
    {
      "epoch": 0.20882288820781522,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001340982,
      "loss": 3.1878,
      "step": 329510
    },
    {
      "epoch": 0.2088415314439734,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001340962,
      "loss": 3.1495,
      "step": 329520
    },
    {
      "epoch": 0.2088601746801316,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001340942,
      "loss": 2.9281,
      "step": 329530
    },
    {
      "epoch": 0.20887881791628976,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001340922,
      "loss": 3.1538,
      "step": 329540
    },
    {
      "epoch": 0.20889746115244795,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001340902,
      "loss": 2.9816,
      "step": 329550
    },
    {
      "epoch": 0.20891610438860614,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001340882,
      "loss": 3.178,
      "step": 329560
    },
    {
      "epoch": 0.20893474762476433,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013408620000000002,
      "loss": 3.1635,
      "step": 329570
    },
    {
      "epoch": 0.20895339086092252,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001340842,
      "loss": 3.1488,
      "step": 329580
    },
    {
      "epoch": 0.20897203409708068,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001340822,
      "loss": 3.0285,
      "step": 329590
    },
    {
      "epoch": 0.20899067733323887,
      "grad_norm": 0.75,
      "learning_rate": 0.0001340802,
      "loss": 3.1005,
      "step": 329600
    },
    {
      "epoch": 0.20900932056939706,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001340782,
      "loss": 3.0629,
      "step": 329610
    },
    {
      "epoch": 0.20902796380555524,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001340762,
      "loss": 3.0734,
      "step": 329620
    },
    {
      "epoch": 0.20904660704171343,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013407420000000002,
      "loss": 3.1394,
      "step": 329630
    },
    {
      "epoch": 0.20906525027787162,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013407220000000001,
      "loss": 3.2253,
      "step": 329640
    },
    {
      "epoch": 0.20908389351402978,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001340702,
      "loss": 3.071,
      "step": 329650
    },
    {
      "epoch": 0.20910253675018797,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001340682,
      "loss": 3.1617,
      "step": 329660
    },
    {
      "epoch": 0.20912117998634616,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001340662,
      "loss": 3.1143,
      "step": 329670
    },
    {
      "epoch": 0.20913982322250435,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001340642,
      "loss": 3.2042,
      "step": 329680
    },
    {
      "epoch": 0.20915846645866254,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001340622,
      "loss": 3.0827,
      "step": 329690
    },
    {
      "epoch": 0.2091771096948207,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013406020000000001,
      "loss": 3.0666,
      "step": 329700
    },
    {
      "epoch": 0.2091957529309789,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001340582,
      "loss": 3.1261,
      "step": 329710
    },
    {
      "epoch": 0.20921439616713708,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001340562,
      "loss": 3.212,
      "step": 329720
    },
    {
      "epoch": 0.20923303940329527,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001340542,
      "loss": 3.0389,
      "step": 329730
    },
    {
      "epoch": 0.20925168263945346,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001340522,
      "loss": 3.1074,
      "step": 329740
    },
    {
      "epoch": 0.20927032587561162,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013405020000000002,
      "loss": 3.1203,
      "step": 329750
    },
    {
      "epoch": 0.2092889691117698,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001340482,
      "loss": 3.0931,
      "step": 329760
    },
    {
      "epoch": 0.209307612347928,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001340462,
      "loss": 3.0795,
      "step": 329770
    },
    {
      "epoch": 0.20932625558408618,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001340442,
      "loss": 3.1219,
      "step": 329780
    },
    {
      "epoch": 0.20934489882024437,
      "grad_norm": 0.875,
      "learning_rate": 0.0001340422,
      "loss": 3.1747,
      "step": 329790
    },
    {
      "epoch": 0.20936354205640254,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013404020000000002,
      "loss": 3.052,
      "step": 329800
    },
    {
      "epoch": 0.20938218529256072,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001340382,
      "loss": 3.2122,
      "step": 329810
    },
    {
      "epoch": 0.2094008285287189,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013403620000000002,
      "loss": 3.1407,
      "step": 329820
    },
    {
      "epoch": 0.2094194717648771,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001340342,
      "loss": 3.1308,
      "step": 329830
    },
    {
      "epoch": 0.2094381150010353,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001340322,
      "loss": 3.2403,
      "step": 329840
    },
    {
      "epoch": 0.20945675823719345,
      "grad_norm": 0.75,
      "learning_rate": 0.00013403020000000003,
      "loss": 3.1686,
      "step": 329850
    },
    {
      "epoch": 0.20947540147335164,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001340282,
      "loss": 3.2123,
      "step": 329860
    },
    {
      "epoch": 0.20949404470950983,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013402620000000002,
      "loss": 3.0133,
      "step": 329870
    },
    {
      "epoch": 0.20951268794566802,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001340242,
      "loss": 3.0382,
      "step": 329880
    },
    {
      "epoch": 0.2095313311818262,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001340222,
      "loss": 3.0918,
      "step": 329890
    },
    {
      "epoch": 0.2095499744179844,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001340202,
      "loss": 3.0839,
      "step": 329900
    },
    {
      "epoch": 0.20956861765414256,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001340182,
      "loss": 3.0926,
      "step": 329910
    },
    {
      "epoch": 0.20958726089030075,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013401620000000003,
      "loss": 3.0497,
      "step": 329920
    },
    {
      "epoch": 0.20960590412645894,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001340142,
      "loss": 3.1452,
      "step": 329930
    },
    {
      "epoch": 0.20962454736261713,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013401220000000002,
      "loss": 3.177,
      "step": 329940
    },
    {
      "epoch": 0.20964319059877531,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013401019999999999,
      "loss": 3.2283,
      "step": 329950
    },
    {
      "epoch": 0.20966183383493348,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001340082,
      "loss": 3.1517,
      "step": 329960
    },
    {
      "epoch": 0.20968047707109166,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001340062,
      "loss": 3.0982,
      "step": 329970
    },
    {
      "epoch": 0.20969912030724985,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001340042,
      "loss": 3.0954,
      "step": 329980
    },
    {
      "epoch": 0.20971776354340804,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013400220000000002,
      "loss": 3.0393,
      "step": 329990
    },
    {
      "epoch": 0.20973640677956623,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001340002,
      "loss": 3.1958,
      "step": 330000
    },
    {
      "epoch": 0.2097550500157244,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013399820000000001,
      "loss": 3.1189,
      "step": 330010
    },
    {
      "epoch": 0.20977369325188258,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001339962,
      "loss": 3.082,
      "step": 330020
    },
    {
      "epoch": 0.20979233648804077,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001339942,
      "loss": 3.108,
      "step": 330030
    },
    {
      "epoch": 0.20981097972419896,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001339922,
      "loss": 3.1872,
      "step": 330040
    },
    {
      "epoch": 0.20982962296035715,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001339902,
      "loss": 3.1756,
      "step": 330050
    },
    {
      "epoch": 0.2098482661965153,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013398820000000002,
      "loss": 3.1145,
      "step": 330060
    },
    {
      "epoch": 0.2098669094326735,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013398620000000001,
      "loss": 3.1652,
      "step": 330070
    },
    {
      "epoch": 0.2098855526688317,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001339842,
      "loss": 3.0825,
      "step": 330080
    },
    {
      "epoch": 0.20990419590498988,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001339822,
      "loss": 3.0484,
      "step": 330090
    },
    {
      "epoch": 0.20992283914114807,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001339802,
      "loss": 3.1315,
      "step": 330100
    },
    {
      "epoch": 0.20994148237730625,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001339782,
      "loss": 3.2382,
      "step": 330110
    },
    {
      "epoch": 0.20996012561346442,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001339762,
      "loss": 3.1031,
      "step": 330120
    },
    {
      "epoch": 0.2099787688496226,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013397420000000002,
      "loss": 3.0622,
      "step": 330130
    },
    {
      "epoch": 0.2099974120857808,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001339722,
      "loss": 3.1547,
      "step": 330140
    },
    {
      "epoch": 0.21001605532193898,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001339702,
      "loss": 3.0421,
      "step": 330150
    },
    {
      "epoch": 0.21003469855809717,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001339682,
      "loss": 3.1189,
      "step": 330160
    },
    {
      "epoch": 0.21005334179425533,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001339662,
      "loss": 3.0833,
      "step": 330170
    },
    {
      "epoch": 0.21007198503041352,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001339642,
      "loss": 3.1015,
      "step": 330180
    },
    {
      "epoch": 0.2100906282665717,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013396220000000002,
      "loss": 3.0947,
      "step": 330190
    },
    {
      "epoch": 0.2101092715027299,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001339602,
      "loss": 3.1898,
      "step": 330200
    },
    {
      "epoch": 0.2101279147388881,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001339582,
      "loss": 3.0057,
      "step": 330210
    },
    {
      "epoch": 0.21014655797504625,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001339562,
      "loss": 3.2798,
      "step": 330220
    },
    {
      "epoch": 0.21016520121120444,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001339542,
      "loss": 3.169,
      "step": 330230
    },
    {
      "epoch": 0.21018384444736263,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013395220000000002,
      "loss": 3.0968,
      "step": 330240
    },
    {
      "epoch": 0.21020248768352082,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001339502,
      "loss": 3.1235,
      "step": 330250
    },
    {
      "epoch": 0.210221130919679,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001339482,
      "loss": 3.1623,
      "step": 330260
    },
    {
      "epoch": 0.21023977415583717,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001339462,
      "loss": 3.0632,
      "step": 330270
    },
    {
      "epoch": 0.21025841739199536,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001339442,
      "loss": 3.1797,
      "step": 330280
    },
    {
      "epoch": 0.21027706062815354,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001339422,
      "loss": 3.2222,
      "step": 330290
    },
    {
      "epoch": 0.21029570386431173,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001339402,
      "loss": 3.2039,
      "step": 330300
    },
    {
      "epoch": 0.21031434710046992,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013393820000000002,
      "loss": 3.2156,
      "step": 330310
    },
    {
      "epoch": 0.2103329903366281,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013393620000000001,
      "loss": 3.0697,
      "step": 330320
    },
    {
      "epoch": 0.21035163357278627,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001339342,
      "loss": 3.0865,
      "step": 330330
    },
    {
      "epoch": 0.21037027680894446,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001339322,
      "loss": 3.1612,
      "step": 330340
    },
    {
      "epoch": 0.21038892004510265,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001339302,
      "loss": 3.0742,
      "step": 330350
    },
    {
      "epoch": 0.21040756328126084,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013392820000000002,
      "loss": 3.2166,
      "step": 330360
    },
    {
      "epoch": 0.21042620651741903,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001339262,
      "loss": 3.059,
      "step": 330370
    },
    {
      "epoch": 0.2104448497535772,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013392420000000001,
      "loss": 3.1576,
      "step": 330380
    },
    {
      "epoch": 0.21046349298973538,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001339222,
      "loss": 3.145,
      "step": 330390
    },
    {
      "epoch": 0.21048213622589357,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001339202,
      "loss": 3.2555,
      "step": 330400
    },
    {
      "epoch": 0.21050077946205176,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013391820000000003,
      "loss": 3.1994,
      "step": 330410
    },
    {
      "epoch": 0.21051942269820995,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001339162,
      "loss": 2.9929,
      "step": 330420
    },
    {
      "epoch": 0.2105380659343681,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013391420000000002,
      "loss": 3.1105,
      "step": 330430
    },
    {
      "epoch": 0.2105567091705263,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001339122,
      "loss": 3.254,
      "step": 330440
    },
    {
      "epoch": 0.21057535240668449,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001339102,
      "loss": 3.1565,
      "step": 330450
    },
    {
      "epoch": 0.21059399564284267,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001339082,
      "loss": 2.9978,
      "step": 330460
    },
    {
      "epoch": 0.21061263887900086,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001339062,
      "loss": 3.0898,
      "step": 330470
    },
    {
      "epoch": 0.21063128211515902,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013390420000000002,
      "loss": 3.0848,
      "step": 330480
    },
    {
      "epoch": 0.2106499253513172,
      "grad_norm": 0.625,
      "learning_rate": 0.0001339022,
      "loss": 3.0175,
      "step": 330490
    },
    {
      "epoch": 0.2106685685874754,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013390020000000002,
      "loss": 2.9291,
      "step": 330500
    },
    {
      "epoch": 0.2106872118236336,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013389819999999998,
      "loss": 3.1398,
      "step": 330510
    },
    {
      "epoch": 0.21070585505979178,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001338962,
      "loss": 3.1385,
      "step": 330520
    },
    {
      "epoch": 0.21072449829594994,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001338942,
      "loss": 3.0469,
      "step": 330530
    },
    {
      "epoch": 0.21074314153210813,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001338922,
      "loss": 3.056,
      "step": 330540
    },
    {
      "epoch": 0.21076178476826632,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013389020000000002,
      "loss": 3.1672,
      "step": 330550
    },
    {
      "epoch": 0.2107804280044245,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001338882,
      "loss": 3.0741,
      "step": 330560
    },
    {
      "epoch": 0.2107990712405827,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001338862,
      "loss": 3.1184,
      "step": 330570
    },
    {
      "epoch": 0.2108177144767409,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001338842,
      "loss": 3.0552,
      "step": 330580
    },
    {
      "epoch": 0.21083635771289905,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001338822,
      "loss": 3.1924,
      "step": 330590
    },
    {
      "epoch": 0.21085500094905724,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001338802,
      "loss": 3.1143,
      "step": 330600
    },
    {
      "epoch": 0.21087364418521543,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001338782,
      "loss": 3.1504,
      "step": 330610
    },
    {
      "epoch": 0.21089228742137361,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013387620000000002,
      "loss": 3.2043,
      "step": 330620
    },
    {
      "epoch": 0.2109109306575318,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001338742,
      "loss": 3.15,
      "step": 330630
    },
    {
      "epoch": 0.21092957389368996,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001338722,
      "loss": 3.0221,
      "step": 330640
    },
    {
      "epoch": 0.21094821712984815,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001338702,
      "loss": 3.2536,
      "step": 330650
    },
    {
      "epoch": 0.21096686036600634,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001338682,
      "loss": 3.1668,
      "step": 330660
    },
    {
      "epoch": 0.21098550360216453,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001338662,
      "loss": 3.1433,
      "step": 330670
    },
    {
      "epoch": 0.21100414683832272,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013386420000000002,
      "loss": 3.0772,
      "step": 330680
    },
    {
      "epoch": 0.21102279007448088,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013386220000000001,
      "loss": 3.207,
      "step": 330690
    },
    {
      "epoch": 0.21104143331063907,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001338602,
      "loss": 3.0655,
      "step": 330700
    },
    {
      "epoch": 0.21106007654679726,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001338582,
      "loss": 3.1607,
      "step": 330710
    },
    {
      "epoch": 0.21107871978295545,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001338562,
      "loss": 3.1786,
      "step": 330720
    },
    {
      "epoch": 0.21109736301911364,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001338542,
      "loss": 3.0544,
      "step": 330730
    },
    {
      "epoch": 0.2111160062552718,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001338522,
      "loss": 3.109,
      "step": 330740
    },
    {
      "epoch": 0.21113464949143,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013385020000000001,
      "loss": 3.0849,
      "step": 330750
    },
    {
      "epoch": 0.21115329272758818,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001338482,
      "loss": 3.0925,
      "step": 330760
    },
    {
      "epoch": 0.21117193596374637,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001338462,
      "loss": 3.206,
      "step": 330770
    },
    {
      "epoch": 0.21119057919990455,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001338442,
      "loss": 3.1088,
      "step": 330780
    },
    {
      "epoch": 0.21120922243606274,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001338422,
      "loss": 3.1372,
      "step": 330790
    },
    {
      "epoch": 0.2112278656722209,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013384020000000002,
      "loss": 3.2204,
      "step": 330800
    },
    {
      "epoch": 0.2112465089083791,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013383820000000002,
      "loss": 3.0263,
      "step": 330810
    },
    {
      "epoch": 0.21126515214453728,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001338362,
      "loss": 3.067,
      "step": 330820
    },
    {
      "epoch": 0.21128379538069547,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001338342,
      "loss": 3.068,
      "step": 330830
    },
    {
      "epoch": 0.21130243861685366,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001338322,
      "loss": 3.1923,
      "step": 330840
    },
    {
      "epoch": 0.21132108185301182,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013383020000000003,
      "loss": 3.2425,
      "step": 330850
    },
    {
      "epoch": 0.21133972508917,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001338282,
      "loss": 3.1007,
      "step": 330860
    },
    {
      "epoch": 0.2113583683253282,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013382620000000002,
      "loss": 3.0294,
      "step": 330870
    },
    {
      "epoch": 0.2113770115614864,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001338242,
      "loss": 3.0641,
      "step": 330880
    },
    {
      "epoch": 0.21139565479764458,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001338222,
      "loss": 3.1626,
      "step": 330890
    },
    {
      "epoch": 0.21141429803380274,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001338202,
      "loss": 3.2313,
      "step": 330900
    },
    {
      "epoch": 0.21143294126996093,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001338182,
      "loss": 3.0934,
      "step": 330910
    },
    {
      "epoch": 0.21145158450611912,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013381620000000002,
      "loss": 3.081,
      "step": 330920
    },
    {
      "epoch": 0.2114702277422773,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001338142,
      "loss": 3.1207,
      "step": 330930
    },
    {
      "epoch": 0.2114888709784355,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001338122,
      "loss": 3.1043,
      "step": 330940
    },
    {
      "epoch": 0.21150751421459366,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001338102,
      "loss": 3.0342,
      "step": 330950
    },
    {
      "epoch": 0.21152615745075184,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001338082,
      "loss": 2.9548,
      "step": 330960
    },
    {
      "epoch": 0.21154480068691003,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013380620000000003,
      "loss": 3.2127,
      "step": 330970
    },
    {
      "epoch": 0.21156344392306822,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001338042,
      "loss": 3.2927,
      "step": 330980
    },
    {
      "epoch": 0.2115820871592264,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013380220000000002,
      "loss": 3.0207,
      "step": 330990
    },
    {
      "epoch": 0.2116007303953846,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001338002,
      "loss": 3.1443,
      "step": 331000
    },
    {
      "epoch": 0.21161937363154276,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001337982,
      "loss": 3.0741,
      "step": 331010
    },
    {
      "epoch": 0.21163801686770095,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001337962,
      "loss": 3.0373,
      "step": 331020
    },
    {
      "epoch": 0.21165666010385914,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001337942,
      "loss": 3.0815,
      "step": 331030
    },
    {
      "epoch": 0.21167530334001733,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013379220000000002,
      "loss": 3.1666,
      "step": 331040
    },
    {
      "epoch": 0.21169394657617552,
      "grad_norm": 0.75,
      "learning_rate": 0.0001337902,
      "loss": 3.2226,
      "step": 331050
    },
    {
      "epoch": 0.21171258981233368,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013378820000000001,
      "loss": 3.1804,
      "step": 331060
    },
    {
      "epoch": 0.21173123304849187,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001337862,
      "loss": 3.1345,
      "step": 331070
    },
    {
      "epoch": 0.21174987628465006,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001337842,
      "loss": 3.1154,
      "step": 331080
    },
    {
      "epoch": 0.21176851952080825,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001337822,
      "loss": 3.0601,
      "step": 331090
    },
    {
      "epoch": 0.21178716275696644,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001337802,
      "loss": 2.9703,
      "step": 331100
    },
    {
      "epoch": 0.2118058059931246,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013377820000000002,
      "loss": 3.0583,
      "step": 331110
    },
    {
      "epoch": 0.21182444922928279,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001337762,
      "loss": 3.133,
      "step": 331120
    },
    {
      "epoch": 0.21184309246544097,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001337742,
      "loss": 3.1144,
      "step": 331130
    },
    {
      "epoch": 0.21186173570159916,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001337722,
      "loss": 3.129,
      "step": 331140
    },
    {
      "epoch": 0.21188037893775735,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001337702,
      "loss": 3.1681,
      "step": 331150
    },
    {
      "epoch": 0.2118990221739155,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001337682,
      "loss": 3.1695,
      "step": 331160
    },
    {
      "epoch": 0.2119176654100737,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001337662,
      "loss": 3.2044,
      "step": 331170
    },
    {
      "epoch": 0.2119363086462319,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013376420000000002,
      "loss": 3.2517,
      "step": 331180
    },
    {
      "epoch": 0.21195495188239008,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001337622,
      "loss": 3.2544,
      "step": 331190
    },
    {
      "epoch": 0.21197359511854827,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001337602,
      "loss": 3.1699,
      "step": 331200
    },
    {
      "epoch": 0.21199223835470643,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001337582,
      "loss": 3.1315,
      "step": 331210
    },
    {
      "epoch": 0.21201088159086462,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001337562,
      "loss": 3.0085,
      "step": 331220
    },
    {
      "epoch": 0.2120295248270228,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001337542,
      "loss": 3.1244,
      "step": 331230
    },
    {
      "epoch": 0.212048168063181,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013375220000000002,
      "loss": 3.2287,
      "step": 331240
    },
    {
      "epoch": 0.2120668112993392,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001337502,
      "loss": 3.2052,
      "step": 331250
    },
    {
      "epoch": 0.21208545453549738,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001337482,
      "loss": 2.9683,
      "step": 331260
    },
    {
      "epoch": 0.21210409777165554,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001337462,
      "loss": 3.0842,
      "step": 331270
    },
    {
      "epoch": 0.21212274100781373,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001337442,
      "loss": 3.2773,
      "step": 331280
    },
    {
      "epoch": 0.21214138424397191,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013374220000000002,
      "loss": 3.1477,
      "step": 331290
    },
    {
      "epoch": 0.2121600274801301,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013374020000000002,
      "loss": 3.2218,
      "step": 331300
    },
    {
      "epoch": 0.2121786707162883,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001337382,
      "loss": 3.1727,
      "step": 331310
    },
    {
      "epoch": 0.21219731395244645,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001337362,
      "loss": 3.1891,
      "step": 331320
    },
    {
      "epoch": 0.21221595718860464,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001337342,
      "loss": 3.028,
      "step": 331330
    },
    {
      "epoch": 0.21223460042476283,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001337322,
      "loss": 3.0889,
      "step": 331340
    },
    {
      "epoch": 0.21225324366092102,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001337302,
      "loss": 3.192,
      "step": 331350
    },
    {
      "epoch": 0.2122718868970792,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013372820000000002,
      "loss": 3.0695,
      "step": 331360
    },
    {
      "epoch": 0.21229053013323737,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013372620000000001,
      "loss": 3.1388,
      "step": 331370
    },
    {
      "epoch": 0.21230917336939556,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001337242,
      "loss": 3.0536,
      "step": 331380
    },
    {
      "epoch": 0.21232781660555375,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001337222,
      "loss": 3.0791,
      "step": 331390
    },
    {
      "epoch": 0.21234645984171194,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001337202,
      "loss": 3.1478,
      "step": 331400
    },
    {
      "epoch": 0.21236510307787013,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013371820000000002,
      "loss": 3.1335,
      "step": 331410
    },
    {
      "epoch": 0.2123837463140283,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001337162,
      "loss": 3.2444,
      "step": 331420
    },
    {
      "epoch": 0.21240238955018648,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013371420000000001,
      "loss": 3.1297,
      "step": 331430
    },
    {
      "epoch": 0.21242103278634467,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001337122,
      "loss": 3.1422,
      "step": 331440
    },
    {
      "epoch": 0.21243967602250285,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001337102,
      "loss": 3.0859,
      "step": 331450
    },
    {
      "epoch": 0.21245831925866104,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013370820000000003,
      "loss": 3.2112,
      "step": 331460
    },
    {
      "epoch": 0.21247696249481923,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001337062,
      "loss": 3.1484,
      "step": 331470
    },
    {
      "epoch": 0.2124956057309774,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013370420000000002,
      "loss": 3.2449,
      "step": 331480
    },
    {
      "epoch": 0.21251424896713558,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001337022,
      "loss": 3.1711,
      "step": 331490
    },
    {
      "epoch": 0.21253289220329377,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001337002,
      "loss": 3.1666,
      "step": 331500
    },
    {
      "epoch": 0.21255153543945196,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001336982,
      "loss": 3.0624,
      "step": 331510
    },
    {
      "epoch": 0.21257017867561015,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001336962,
      "loss": 3.1951,
      "step": 331520
    },
    {
      "epoch": 0.2125888219117683,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013369420000000003,
      "loss": 3.0593,
      "step": 331530
    },
    {
      "epoch": 0.2126074651479265,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001336922,
      "loss": 3.13,
      "step": 331540
    },
    {
      "epoch": 0.2126261083840847,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013369020000000002,
      "loss": 3.1832,
      "step": 331550
    },
    {
      "epoch": 0.21264475162024288,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013368819999999998,
      "loss": 3.0564,
      "step": 331560
    },
    {
      "epoch": 0.21266339485640107,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001336862,
      "loss": 3.1531,
      "step": 331570
    },
    {
      "epoch": 0.21268203809255923,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001336842,
      "loss": 3.1639,
      "step": 331580
    },
    {
      "epoch": 0.21270068132871742,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001336822,
      "loss": 3.207,
      "step": 331590
    },
    {
      "epoch": 0.2127193245648756,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013368020000000002,
      "loss": 3.1833,
      "step": 331600
    },
    {
      "epoch": 0.2127379678010338,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001336782,
      "loss": 3.1474,
      "step": 331610
    },
    {
      "epoch": 0.21275661103719198,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001336762,
      "loss": 3.2186,
      "step": 331620
    },
    {
      "epoch": 0.21277525427335015,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001336742,
      "loss": 3.2206,
      "step": 331630
    },
    {
      "epoch": 0.21279389750950833,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001336722,
      "loss": 3.0117,
      "step": 331640
    },
    {
      "epoch": 0.21281254074566652,
      "grad_norm": 0.5,
      "learning_rate": 0.0001336702,
      "loss": 2.8702,
      "step": 331650
    },
    {
      "epoch": 0.2128311839818247,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001336682,
      "loss": 3.0524,
      "step": 331660
    },
    {
      "epoch": 0.2128498272179829,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013366620000000002,
      "loss": 3.0356,
      "step": 331670
    },
    {
      "epoch": 0.2128684704541411,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013366420000000001,
      "loss": 3.1028,
      "step": 331680
    },
    {
      "epoch": 0.21288711369029925,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001336622,
      "loss": 2.9778,
      "step": 331690
    },
    {
      "epoch": 0.21290575692645744,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001336602,
      "loss": 3.1272,
      "step": 331700
    },
    {
      "epoch": 0.21292440016261563,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001336582,
      "loss": 3.0898,
      "step": 331710
    },
    {
      "epoch": 0.21294304339877382,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001336562,
      "loss": 3.1719,
      "step": 331720
    },
    {
      "epoch": 0.212961686634932,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001336542,
      "loss": 3.1627,
      "step": 331730
    },
    {
      "epoch": 0.21298032987109017,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013365220000000001,
      "loss": 3.3133,
      "step": 331740
    },
    {
      "epoch": 0.21299897310724836,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001336502,
      "loss": 3.1493,
      "step": 331750
    },
    {
      "epoch": 0.21301761634340655,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001336482,
      "loss": 3.1836,
      "step": 331760
    },
    {
      "epoch": 0.21303625957956474,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001336462,
      "loss": 3.1545,
      "step": 331770
    },
    {
      "epoch": 0.21305490281572292,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001336442,
      "loss": 3.1298,
      "step": 331780
    },
    {
      "epoch": 0.21307354605188109,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013364220000000002,
      "loss": 3.0557,
      "step": 331790
    },
    {
      "epoch": 0.21309218928803927,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013364020000000002,
      "loss": 3.0272,
      "step": 331800
    },
    {
      "epoch": 0.21311083252419746,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001336382,
      "loss": 3.0478,
      "step": 331810
    },
    {
      "epoch": 0.21312947576035565,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001336362,
      "loss": 3.0883,
      "step": 331820
    },
    {
      "epoch": 0.21314811899651384,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001336342,
      "loss": 3.1601,
      "step": 331830
    },
    {
      "epoch": 0.213166762232672,
      "grad_norm": 0.75,
      "learning_rate": 0.0001336322,
      "loss": 3.1285,
      "step": 331840
    },
    {
      "epoch": 0.2131854054688302,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013363020000000002,
      "loss": 3.057,
      "step": 331850
    },
    {
      "epoch": 0.21320404870498838,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013362820000000002,
      "loss": 3.1429,
      "step": 331860
    },
    {
      "epoch": 0.21322269194114657,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001336262,
      "loss": 3.1566,
      "step": 331870
    },
    {
      "epoch": 0.21324133517730476,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001336242,
      "loss": 3.1985,
      "step": 331880
    },
    {
      "epoch": 0.21325997841346292,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001336222,
      "loss": 2.9902,
      "step": 331890
    },
    {
      "epoch": 0.2132786216496211,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001336202,
      "loss": 3.0435,
      "step": 331900
    },
    {
      "epoch": 0.2132972648857793,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001336182,
      "loss": 3.027,
      "step": 331910
    },
    {
      "epoch": 0.2133159081219375,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013361620000000002,
      "loss": 3.0986,
      "step": 331920
    },
    {
      "epoch": 0.21333455135809568,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001336142,
      "loss": 3.1259,
      "step": 331930
    },
    {
      "epoch": 0.21335319459425386,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001336122,
      "loss": 3.2455,
      "step": 331940
    },
    {
      "epoch": 0.21337183783041203,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001336102,
      "loss": 3.1907,
      "step": 331950
    },
    {
      "epoch": 0.21339048106657021,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001336082,
      "loss": 3.0869,
      "step": 331960
    },
    {
      "epoch": 0.2134091243027284,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013360620000000002,
      "loss": 3.1246,
      "step": 331970
    },
    {
      "epoch": 0.2134277675388866,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001336042,
      "loss": 3.0826,
      "step": 331980
    },
    {
      "epoch": 0.21344641077504478,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001336022,
      "loss": 3.1068,
      "step": 331990
    },
    {
      "epoch": 0.21346505401120294,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001336002,
      "loss": 3.0724,
      "step": 332000
    },
    {
      "epoch": 0.21348369724736113,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001335982,
      "loss": 3.1211,
      "step": 332010
    },
    {
      "epoch": 0.21350234048351932,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013359620000000003,
      "loss": 3.0014,
      "step": 332020
    },
    {
      "epoch": 0.2135209837196775,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001335942,
      "loss": 3.1505,
      "step": 332030
    },
    {
      "epoch": 0.2135396269558357,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013359220000000002,
      "loss": 3.0571,
      "step": 332040
    },
    {
      "epoch": 0.21355827019199386,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001335902,
      "loss": 3.1343,
      "step": 332050
    },
    {
      "epoch": 0.21357691342815205,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001335882,
      "loss": 3.1305,
      "step": 332060
    },
    {
      "epoch": 0.21359555666431024,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001335862,
      "loss": 3.0498,
      "step": 332070
    },
    {
      "epoch": 0.21361419990046843,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001335842,
      "loss": 3.0998,
      "step": 332080
    },
    {
      "epoch": 0.21363284313662662,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013358220000000002,
      "loss": 3.0545,
      "step": 332090
    },
    {
      "epoch": 0.21365148637278478,
      "grad_norm": 0.75,
      "learning_rate": 0.0001335802,
      "loss": 3.1157,
      "step": 332100
    },
    {
      "epoch": 0.21367012960894297,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013357820000000001,
      "loss": 3.0868,
      "step": 332110
    },
    {
      "epoch": 0.21368877284510115,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013357619999999998,
      "loss": 3.1223,
      "step": 332120
    },
    {
      "epoch": 0.21370741608125934,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001335742,
      "loss": 3.2853,
      "step": 332130
    },
    {
      "epoch": 0.21372605931741753,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001335722,
      "loss": 3.0694,
      "step": 332140
    },
    {
      "epoch": 0.21374470255357572,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001335702,
      "loss": 3.1627,
      "step": 332150
    },
    {
      "epoch": 0.21376334578973388,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013356820000000002,
      "loss": 3.1585,
      "step": 332160
    },
    {
      "epoch": 0.21378198902589207,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001335662,
      "loss": 3.1031,
      "step": 332170
    },
    {
      "epoch": 0.21380063226205026,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001335642,
      "loss": 3.0299,
      "step": 332180
    },
    {
      "epoch": 0.21381927549820845,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001335622,
      "loss": 3.0248,
      "step": 332190
    },
    {
      "epoch": 0.21383791873436664,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001335602,
      "loss": 3.0386,
      "step": 332200
    },
    {
      "epoch": 0.2138565619705248,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001335582,
      "loss": 3.1047,
      "step": 332210
    },
    {
      "epoch": 0.213875205206683,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001335562,
      "loss": 3.1857,
      "step": 332220
    },
    {
      "epoch": 0.21389384844284118,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013355420000000002,
      "loss": 3.0282,
      "step": 332230
    },
    {
      "epoch": 0.21391249167899937,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001335522,
      "loss": 3.0374,
      "step": 332240
    },
    {
      "epoch": 0.21393113491515756,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001335502,
      "loss": 3.0563,
      "step": 332250
    },
    {
      "epoch": 0.21394977815131572,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001335482,
      "loss": 3.1061,
      "step": 332260
    },
    {
      "epoch": 0.2139684213874739,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001335462,
      "loss": 3.0862,
      "step": 332270
    },
    {
      "epoch": 0.2139870646236321,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013354420000000002,
      "loss": 3.1453,
      "step": 332280
    },
    {
      "epoch": 0.21400570785979028,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013354220000000002,
      "loss": 3.1405,
      "step": 332290
    },
    {
      "epoch": 0.21402435109594847,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001335402,
      "loss": 3.1697,
      "step": 332300
    },
    {
      "epoch": 0.21404299433210663,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001335382,
      "loss": 3.2173,
      "step": 332310
    },
    {
      "epoch": 0.21406163756826482,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001335362,
      "loss": 3.1651,
      "step": 332320
    },
    {
      "epoch": 0.214080280804423,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001335342,
      "loss": 3.1309,
      "step": 332330
    },
    {
      "epoch": 0.2140989240405812,
      "grad_norm": 0.75,
      "learning_rate": 0.0001335322,
      "loss": 2.9562,
      "step": 332340
    },
    {
      "epoch": 0.2141175672767394,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013353020000000002,
      "loss": 3.059,
      "step": 332350
    },
    {
      "epoch": 0.21413621051289758,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001335282,
      "loss": 3.1687,
      "step": 332360
    },
    {
      "epoch": 0.21415485374905574,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001335262,
      "loss": 3.1451,
      "step": 332370
    },
    {
      "epoch": 0.21417349698521393,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001335242,
      "loss": 3.1315,
      "step": 332380
    },
    {
      "epoch": 0.21419214022137212,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001335222,
      "loss": 3.171,
      "step": 332390
    },
    {
      "epoch": 0.2142107834575303,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001335202,
      "loss": 3.1542,
      "step": 332400
    },
    {
      "epoch": 0.2142294266936885,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013351820000000002,
      "loss": 3.1883,
      "step": 332410
    },
    {
      "epoch": 0.21424806992984666,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013351620000000001,
      "loss": 3.1656,
      "step": 332420
    },
    {
      "epoch": 0.21426671316600485,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001335142,
      "loss": 3.0635,
      "step": 332430
    },
    {
      "epoch": 0.21428535640216304,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001335122,
      "loss": 3.0333,
      "step": 332440
    },
    {
      "epoch": 0.21430399963832122,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001335102,
      "loss": 3.0781,
      "step": 332450
    },
    {
      "epoch": 0.2143226428744794,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013350820000000002,
      "loss": 2.967,
      "step": 332460
    },
    {
      "epoch": 0.21434128611063757,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001335062,
      "loss": 3.1128,
      "step": 332470
    },
    {
      "epoch": 0.21435992934679576,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013350420000000002,
      "loss": 2.9952,
      "step": 332480
    },
    {
      "epoch": 0.21437857258295395,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001335022,
      "loss": 3.1292,
      "step": 332490
    },
    {
      "epoch": 0.21439721581911214,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001335002,
      "loss": 3.1165,
      "step": 332500
    },
    {
      "epoch": 0.21441585905527033,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001334982,
      "loss": 3.2115,
      "step": 332510
    },
    {
      "epoch": 0.2144345022914285,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001334962,
      "loss": 3.1657,
      "step": 332520
    },
    {
      "epoch": 0.21445314552758668,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013349420000000002,
      "loss": 3.0561,
      "step": 332530
    },
    {
      "epoch": 0.21447178876374487,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001334922,
      "loss": 3.1846,
      "step": 332540
    },
    {
      "epoch": 0.21449043199990306,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001334902,
      "loss": 3.1122,
      "step": 332550
    },
    {
      "epoch": 0.21450907523606125,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001334882,
      "loss": 3.1479,
      "step": 332560
    },
    {
      "epoch": 0.21452771847221944,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001334862,
      "loss": 3.1042,
      "step": 332570
    },
    {
      "epoch": 0.2145463617083776,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013348420000000003,
      "loss": 3.0655,
      "step": 332580
    },
    {
      "epoch": 0.2145650049445358,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001334822,
      "loss": 3.0701,
      "step": 332590
    },
    {
      "epoch": 0.21458364818069398,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00013348020000000002,
      "loss": 3.1985,
      "step": 332600
    },
    {
      "epoch": 0.21460229141685216,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013347819999999999,
      "loss": 3.1073,
      "step": 332610
    },
    {
      "epoch": 0.21462093465301035,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001334762,
      "loss": 3.1337,
      "step": 332620
    },
    {
      "epoch": 0.21463957788916851,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001334742,
      "loss": 3.2481,
      "step": 332630
    },
    {
      "epoch": 0.2146582211253267,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001334722,
      "loss": 3.018,
      "step": 332640
    },
    {
      "epoch": 0.2146768643614849,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013347020000000002,
      "loss": 3.2144,
      "step": 332650
    },
    {
      "epoch": 0.21469550759764308,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001334682,
      "loss": 3.1102,
      "step": 332660
    },
    {
      "epoch": 0.21471415083380127,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001334662,
      "loss": 3.0927,
      "step": 332670
    },
    {
      "epoch": 0.21473279406995943,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001334642,
      "loss": 2.9886,
      "step": 332680
    },
    {
      "epoch": 0.21475143730611762,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001334622,
      "loss": 3.1733,
      "step": 332690
    },
    {
      "epoch": 0.2147700805422758,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001334602,
      "loss": 3.1146,
      "step": 332700
    },
    {
      "epoch": 0.214788723778434,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001334582,
      "loss": 3.1457,
      "step": 332710
    },
    {
      "epoch": 0.2148073670145922,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013345620000000002,
      "loss": 3.1994,
      "step": 332720
    },
    {
      "epoch": 0.21482601025075035,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001334542,
      "loss": 3.153,
      "step": 332730
    },
    {
      "epoch": 0.21484465348690854,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001334522,
      "loss": 3.0599,
      "step": 332740
    },
    {
      "epoch": 0.21486329672306673,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001334502,
      "loss": 3.2821,
      "step": 332750
    },
    {
      "epoch": 0.21488193995922492,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001334482,
      "loss": 3.0008,
      "step": 332760
    },
    {
      "epoch": 0.2149005831953831,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013344620000000002,
      "loss": 3.1272,
      "step": 332770
    },
    {
      "epoch": 0.21491922643154127,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001334442,
      "loss": 3.1232,
      "step": 332780
    },
    {
      "epoch": 0.21493786966769945,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013344220000000001,
      "loss": 3.226,
      "step": 332790
    },
    {
      "epoch": 0.21495651290385764,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001334402,
      "loss": 3.163,
      "step": 332800
    },
    {
      "epoch": 0.21497515614001583,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001334382,
      "loss": 3.1661,
      "step": 332810
    },
    {
      "epoch": 0.21499379937617402,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001334362,
      "loss": 3.1748,
      "step": 332820
    },
    {
      "epoch": 0.2150124426123322,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001334342,
      "loss": 3.2296,
      "step": 332830
    },
    {
      "epoch": 0.21503108584849037,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013343220000000002,
      "loss": 3.0391,
      "step": 332840
    },
    {
      "epoch": 0.21504972908464856,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013343020000000002,
      "loss": 3.2036,
      "step": 332850
    },
    {
      "epoch": 0.21506837232080675,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001334282,
      "loss": 3.1746,
      "step": 332860
    },
    {
      "epoch": 0.21508701555696494,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001334262,
      "loss": 3.2226,
      "step": 332870
    },
    {
      "epoch": 0.21510565879312313,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001334242,
      "loss": 3.1742,
      "step": 332880
    },
    {
      "epoch": 0.2151243020292813,
      "grad_norm": 0.75,
      "learning_rate": 0.0001334222,
      "loss": 3.0192,
      "step": 332890
    },
    {
      "epoch": 0.21514294526543948,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013342020000000002,
      "loss": 3.0746,
      "step": 332900
    },
    {
      "epoch": 0.21516158850159767,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013341820000000002,
      "loss": 3.1299,
      "step": 332910
    },
    {
      "epoch": 0.21518023173775586,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001334162,
      "loss": 3.2002,
      "step": 332920
    },
    {
      "epoch": 0.21519887497391404,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001334142,
      "loss": 3.0709,
      "step": 332930
    },
    {
      "epoch": 0.2152175182100722,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001334122,
      "loss": 3.03,
      "step": 332940
    },
    {
      "epoch": 0.2152361614462304,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001334102,
      "loss": 3.2186,
      "step": 332950
    },
    {
      "epoch": 0.21525480468238858,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001334082,
      "loss": 3.0431,
      "step": 332960
    },
    {
      "epoch": 0.21527344791854677,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013340620000000002,
      "loss": 3.1616,
      "step": 332970
    },
    {
      "epoch": 0.21529209115470496,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001334042,
      "loss": 3.1216,
      "step": 332980
    },
    {
      "epoch": 0.21531073439086312,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001334022,
      "loss": 3.2192,
      "step": 332990
    },
    {
      "epoch": 0.2153293776270213,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001334002,
      "loss": 3.1162,
      "step": 333000
    },
    {
      "epoch": 0.2153480208631795,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001333982,
      "loss": 3.1111,
      "step": 333010
    },
    {
      "epoch": 0.2153666640993377,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013339620000000002,
      "loss": 3.0959,
      "step": 333020
    },
    {
      "epoch": 0.21538530733549588,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001333942,
      "loss": 3.1548,
      "step": 333030
    },
    {
      "epoch": 0.21540395057165407,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001333922,
      "loss": 3.2354,
      "step": 333040
    },
    {
      "epoch": 0.21542259380781223,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001333902,
      "loss": 3.2069,
      "step": 333050
    },
    {
      "epoch": 0.21544123704397042,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001333882,
      "loss": 3.1712,
      "step": 333060
    },
    {
      "epoch": 0.2154598802801286,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013338620000000003,
      "loss": 3.1773,
      "step": 333070
    },
    {
      "epoch": 0.2154785235162868,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001333842,
      "loss": 3.1373,
      "step": 333080
    },
    {
      "epoch": 0.21549716675244499,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013338220000000002,
      "loss": 3.0378,
      "step": 333090
    },
    {
      "epoch": 0.21551580998860315,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001333802,
      "loss": 3.2182,
      "step": 333100
    },
    {
      "epoch": 0.21553445322476134,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001333782,
      "loss": 2.9663,
      "step": 333110
    },
    {
      "epoch": 0.21555309646091952,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001333762,
      "loss": 3.1749,
      "step": 333120
    },
    {
      "epoch": 0.2155717396970777,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001333742,
      "loss": 2.9676,
      "step": 333130
    },
    {
      "epoch": 0.2155903829332359,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013337220000000002,
      "loss": 3.1472,
      "step": 333140
    },
    {
      "epoch": 0.21560902616939406,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001333702,
      "loss": 3.2002,
      "step": 333150
    },
    {
      "epoch": 0.21562766940555225,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013336820000000002,
      "loss": 3.1461,
      "step": 333160
    },
    {
      "epoch": 0.21564631264171044,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013336619999999998,
      "loss": 3.0054,
      "step": 333170
    },
    {
      "epoch": 0.21566495587786863,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001333642,
      "loss": 3.124,
      "step": 333180
    },
    {
      "epoch": 0.21568359911402682,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001333622,
      "loss": 3.1659,
      "step": 333190
    },
    {
      "epoch": 0.21570224235018498,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001333602,
      "loss": 3.2085,
      "step": 333200
    },
    {
      "epoch": 0.21572088558634317,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013335820000000002,
      "loss": 3.0827,
      "step": 333210
    },
    {
      "epoch": 0.21573952882250136,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001333562,
      "loss": 3.1601,
      "step": 333220
    },
    {
      "epoch": 0.21575817205865955,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001333542,
      "loss": 3.1557,
      "step": 333230
    },
    {
      "epoch": 0.21577681529481774,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001333522,
      "loss": 3.1663,
      "step": 333240
    },
    {
      "epoch": 0.21579545853097593,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001333502,
      "loss": 3.0726,
      "step": 333250
    },
    {
      "epoch": 0.2158141017671341,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013334820000000003,
      "loss": 3.1209,
      "step": 333260
    },
    {
      "epoch": 0.21583274500329228,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001333462,
      "loss": 3.0462,
      "step": 333270
    },
    {
      "epoch": 0.21585138823945046,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013334420000000002,
      "loss": 3.1505,
      "step": 333280
    },
    {
      "epoch": 0.21587003147560865,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001333422,
      "loss": 3.1585,
      "step": 333290
    },
    {
      "epoch": 0.21588867471176684,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001333402,
      "loss": 3.0956,
      "step": 333300
    },
    {
      "epoch": 0.215907317947925,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001333382,
      "loss": 3.1305,
      "step": 333310
    },
    {
      "epoch": 0.2159259611840832,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001333362,
      "loss": 2.9774,
      "step": 333320
    },
    {
      "epoch": 0.21594460442024138,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013333420000000002,
      "loss": 3.1052,
      "step": 333330
    },
    {
      "epoch": 0.21596324765639957,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001333322,
      "loss": 3.071,
      "step": 333340
    },
    {
      "epoch": 0.21598189089255776,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001333302,
      "loss": 3.0963,
      "step": 333350
    },
    {
      "epoch": 0.21600053412871592,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001333282,
      "loss": 2.9582,
      "step": 333360
    },
    {
      "epoch": 0.2160191773648741,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001333262,
      "loss": 3.1346,
      "step": 333370
    },
    {
      "epoch": 0.2160378206010323,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001333242,
      "loss": 3.1284,
      "step": 333380
    },
    {
      "epoch": 0.2160564638371905,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001333222,
      "loss": 3.2925,
      "step": 333390
    },
    {
      "epoch": 0.21607510707334868,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00013332020000000002,
      "loss": 3.2877,
      "step": 333400
    },
    {
      "epoch": 0.21609375030950684,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013331820000000001,
      "loss": 3.2627,
      "step": 333410
    },
    {
      "epoch": 0.21611239354566503,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001333162,
      "loss": 3.2472,
      "step": 333420
    },
    {
      "epoch": 0.21613103678182322,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001333142,
      "loss": 3.2667,
      "step": 333430
    },
    {
      "epoch": 0.2161496800179814,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001333122,
      "loss": 3.0879,
      "step": 333440
    },
    {
      "epoch": 0.2161683232541396,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001333102,
      "loss": 2.983,
      "step": 333450
    },
    {
      "epoch": 0.21618696649029776,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013330820000000002,
      "loss": 3.2627,
      "step": 333460
    },
    {
      "epoch": 0.21620560972645594,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013330620000000001,
      "loss": 3.1313,
      "step": 333470
    },
    {
      "epoch": 0.21622425296261413,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001333042,
      "loss": 3.1411,
      "step": 333480
    },
    {
      "epoch": 0.21624289619877232,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001333022,
      "loss": 3.1832,
      "step": 333490
    },
    {
      "epoch": 0.2162615394349305,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001333002,
      "loss": 3.267,
      "step": 333500
    },
    {
      "epoch": 0.2162801826710887,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001332982,
      "loss": 3.1206,
      "step": 333510
    },
    {
      "epoch": 0.21629882590724686,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001332962,
      "loss": 3.1928,
      "step": 333520
    },
    {
      "epoch": 0.21631746914340505,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013329420000000002,
      "loss": 3.2173,
      "step": 333530
    },
    {
      "epoch": 0.21633611237956324,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001332922,
      "loss": 3.1245,
      "step": 333540
    },
    {
      "epoch": 0.21635475561572143,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001332902,
      "loss": 3.1195,
      "step": 333550
    },
    {
      "epoch": 0.21637339885187962,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001332882,
      "loss": 3.0296,
      "step": 333560
    },
    {
      "epoch": 0.21639204208803778,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001332862,
      "loss": 3.1048,
      "step": 333570
    },
    {
      "epoch": 0.21641068532419597,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013328420000000002,
      "loss": 3.1111,
      "step": 333580
    },
    {
      "epoch": 0.21642932856035416,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001332822,
      "loss": 3.1262,
      "step": 333590
    },
    {
      "epoch": 0.21644797179651235,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001332802,
      "loss": 3.1394,
      "step": 333600
    },
    {
      "epoch": 0.21646661503267053,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001332782,
      "loss": 3.1806,
      "step": 333610
    },
    {
      "epoch": 0.2164852582688287,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001332762,
      "loss": 3.1019,
      "step": 333620
    },
    {
      "epoch": 0.21650390150498688,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013327420000000003,
      "loss": 3.1779,
      "step": 333630
    },
    {
      "epoch": 0.21652254474114507,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001332722,
      "loss": 3.2034,
      "step": 333640
    },
    {
      "epoch": 0.21654118797730326,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013327020000000002,
      "loss": 3.1427,
      "step": 333650
    },
    {
      "epoch": 0.21655983121346145,
      "grad_norm": 0.890625,
      "learning_rate": 0.00013326819999999999,
      "loss": 3.1452,
      "step": 333660
    },
    {
      "epoch": 0.2165784744496196,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001332662,
      "loss": 3.1875,
      "step": 333670
    },
    {
      "epoch": 0.2165971176857778,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001332642,
      "loss": 3.1321,
      "step": 333680
    },
    {
      "epoch": 0.216615760921936,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001332622,
      "loss": 3.2251,
      "step": 333690
    },
    {
      "epoch": 0.21663440415809418,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013326020000000002,
      "loss": 3.0029,
      "step": 333700
    },
    {
      "epoch": 0.21665304739425237,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001332582,
      "loss": 3.0874,
      "step": 333710
    },
    {
      "epoch": 0.21667169063041056,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001332562,
      "loss": 3.1698,
      "step": 333720
    },
    {
      "epoch": 0.21669033386656872,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001332542,
      "loss": 3.1587,
      "step": 333730
    },
    {
      "epoch": 0.2167089771027269,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001332522,
      "loss": 3.1634,
      "step": 333740
    },
    {
      "epoch": 0.2167276203388851,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013325020000000003,
      "loss": 3.1968,
      "step": 333750
    },
    {
      "epoch": 0.21674626357504329,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001332482,
      "loss": 3.0883,
      "step": 333760
    },
    {
      "epoch": 0.21676490681120147,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013324620000000002,
      "loss": 3.2275,
      "step": 333770
    },
    {
      "epoch": 0.21678355004735964,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001332442,
      "loss": 3.1911,
      "step": 333780
    },
    {
      "epoch": 0.21680219328351782,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001332422,
      "loss": 3.1885,
      "step": 333790
    },
    {
      "epoch": 0.216820836519676,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001332402,
      "loss": 3.0172,
      "step": 333800
    },
    {
      "epoch": 0.2168394797558342,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001332382,
      "loss": 3.2248,
      "step": 333810
    },
    {
      "epoch": 0.2168581229919924,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013323620000000002,
      "loss": 3.2114,
      "step": 333820
    },
    {
      "epoch": 0.21687676622815055,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001332342,
      "loss": 3.1231,
      "step": 333830
    },
    {
      "epoch": 0.21689540946430874,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013323220000000002,
      "loss": 3.0719,
      "step": 333840
    },
    {
      "epoch": 0.21691405270046693,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001332302,
      "loss": 3.0419,
      "step": 333850
    },
    {
      "epoch": 0.21693269593662512,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001332282,
      "loss": 3.0909,
      "step": 333860
    },
    {
      "epoch": 0.2169513391727833,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001332262,
      "loss": 3.1011,
      "step": 333870
    },
    {
      "epoch": 0.21696998240894147,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001332242,
      "loss": 3.0651,
      "step": 333880
    },
    {
      "epoch": 0.21698862564509966,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013322220000000002,
      "loss": 3.1817,
      "step": 333890
    },
    {
      "epoch": 0.21700726888125785,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013322020000000002,
      "loss": 3.1083,
      "step": 333900
    },
    {
      "epoch": 0.21702591211741604,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001332182,
      "loss": 3.0112,
      "step": 333910
    },
    {
      "epoch": 0.21704455535357423,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001332162,
      "loss": 3.0722,
      "step": 333920
    },
    {
      "epoch": 0.21706319858973241,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001332142,
      "loss": 3.1002,
      "step": 333930
    },
    {
      "epoch": 0.21708184182589058,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001332122,
      "loss": 3.1233,
      "step": 333940
    },
    {
      "epoch": 0.21710048506204876,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001332102,
      "loss": 3.179,
      "step": 333950
    },
    {
      "epoch": 0.21711912829820695,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013320820000000002,
      "loss": 3.1437,
      "step": 333960
    },
    {
      "epoch": 0.21713777153436514,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001332062,
      "loss": 3.0986,
      "step": 333970
    },
    {
      "epoch": 0.21715641477052333,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001332042,
      "loss": 3.148,
      "step": 333980
    },
    {
      "epoch": 0.2171750580066815,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001332022,
      "loss": 3.0168,
      "step": 333990
    },
    {
      "epoch": 0.21719370124283968,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001332002,
      "loss": 3.1986,
      "step": 334000
    },
    {
      "epoch": 0.21721234447899787,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001331982,
      "loss": 3.186,
      "step": 334010
    },
    {
      "epoch": 0.21723098771515606,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013319620000000002,
      "loss": 3.099,
      "step": 334020
    },
    {
      "epoch": 0.21724963095131425,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001331942,
      "loss": 3.0732,
      "step": 334030
    },
    {
      "epoch": 0.2172682741874724,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001331922,
      "loss": 3.0191,
      "step": 334040
    },
    {
      "epoch": 0.2172869174236306,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001331902,
      "loss": 3.0845,
      "step": 334050
    },
    {
      "epoch": 0.2173055606597888,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001331882,
      "loss": 3.1472,
      "step": 334060
    },
    {
      "epoch": 0.21732420389594698,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013318620000000002,
      "loss": 3.0054,
      "step": 334070
    },
    {
      "epoch": 0.21734284713210517,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001331842,
      "loss": 3.182,
      "step": 334080
    },
    {
      "epoch": 0.21736149036826333,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013318220000000001,
      "loss": 3.264,
      "step": 334090
    },
    {
      "epoch": 0.21738013360442152,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001331802,
      "loss": 3.1375,
      "step": 334100
    },
    {
      "epoch": 0.2173987768405797,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001331782,
      "loss": 3.1624,
      "step": 334110
    },
    {
      "epoch": 0.2174174200767379,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001331762,
      "loss": 3.1543,
      "step": 334120
    },
    {
      "epoch": 0.21743606331289608,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001331742,
      "loss": 2.9983,
      "step": 334130
    },
    {
      "epoch": 0.21745470654905424,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013317220000000002,
      "loss": 3.1398,
      "step": 334140
    },
    {
      "epoch": 0.21747334978521243,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001331702,
      "loss": 3.2007,
      "step": 334150
    },
    {
      "epoch": 0.21749199302137062,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001331682,
      "loss": 3.0744,
      "step": 334160
    },
    {
      "epoch": 0.2175106362575288,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001331662,
      "loss": 3.2011,
      "step": 334170
    },
    {
      "epoch": 0.217529279493687,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001331642,
      "loss": 3.132,
      "step": 334180
    },
    {
      "epoch": 0.2175479227298452,
      "grad_norm": 0.75,
      "learning_rate": 0.00013316220000000002,
      "loss": 3.1145,
      "step": 334190
    },
    {
      "epoch": 0.21756656596600335,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001331602,
      "loss": 3.0477,
      "step": 334200
    },
    {
      "epoch": 0.21758520920216154,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013315820000000002,
      "loss": 3.1,
      "step": 334210
    },
    {
      "epoch": 0.21760385243831973,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001331562,
      "loss": 3.0058,
      "step": 334220
    },
    {
      "epoch": 0.21762249567447792,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001331542,
      "loss": 3.0904,
      "step": 334230
    },
    {
      "epoch": 0.2176411389106361,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013315220000000003,
      "loss": 3.112,
      "step": 334240
    },
    {
      "epoch": 0.21765978214679427,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001331502,
      "loss": 3.1185,
      "step": 334250
    },
    {
      "epoch": 0.21767842538295246,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013314820000000002,
      "loss": 3.1255,
      "step": 334260
    },
    {
      "epoch": 0.21769706861911065,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001331462,
      "loss": 3.2041,
      "step": 334270
    },
    {
      "epoch": 0.21771571185526883,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001331442,
      "loss": 3.0934,
      "step": 334280
    },
    {
      "epoch": 0.21773435509142702,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001331422,
      "loss": 3.0587,
      "step": 334290
    },
    {
      "epoch": 0.21775299832758518,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001331402,
      "loss": 3.1901,
      "step": 334300
    },
    {
      "epoch": 0.21777164156374337,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013313820000000003,
      "loss": 3.1436,
      "step": 334310
    },
    {
      "epoch": 0.21779028479990156,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001331362,
      "loss": 3.0461,
      "step": 334320
    },
    {
      "epoch": 0.21780892803605975,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013313420000000002,
      "loss": 2.9912,
      "step": 334330
    },
    {
      "epoch": 0.21782757127221794,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013313219999999999,
      "loss": 3.0729,
      "step": 334340
    },
    {
      "epoch": 0.2178462145083761,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001331302,
      "loss": 3.188,
      "step": 334350
    },
    {
      "epoch": 0.2178648577445343,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001331282,
      "loss": 3.0327,
      "step": 334360
    },
    {
      "epoch": 0.21788350098069248,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001331262,
      "loss": 3.1322,
      "step": 334370
    },
    {
      "epoch": 0.21790214421685067,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013312420000000002,
      "loss": 3.0776,
      "step": 334380
    },
    {
      "epoch": 0.21792078745300886,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001331222,
      "loss": 3.1818,
      "step": 334390
    },
    {
      "epoch": 0.21793943068916705,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001331202,
      "loss": 3.2238,
      "step": 334400
    },
    {
      "epoch": 0.2179580739253252,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001331182,
      "loss": 3.1301,
      "step": 334410
    },
    {
      "epoch": 0.2179767171614834,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001331162,
      "loss": 3.1697,
      "step": 334420
    },
    {
      "epoch": 0.21799536039764159,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001331142,
      "loss": 3.2047,
      "step": 334430
    },
    {
      "epoch": 0.21801400363379977,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001331122,
      "loss": 3.0891,
      "step": 334440
    },
    {
      "epoch": 0.21803264686995796,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013311020000000002,
      "loss": 3.1867,
      "step": 334450
    },
    {
      "epoch": 0.21805129010611612,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013310820000000001,
      "loss": 3.0266,
      "step": 334460
    },
    {
      "epoch": 0.2180699333422743,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001331062,
      "loss": 3.1183,
      "step": 334470
    },
    {
      "epoch": 0.2180885765784325,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001331042,
      "loss": 3.227,
      "step": 334480
    },
    {
      "epoch": 0.2181072198145907,
      "grad_norm": 0.75,
      "learning_rate": 0.0001331022,
      "loss": 3.1454,
      "step": 334490
    },
    {
      "epoch": 0.21812586305074888,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001331002,
      "loss": 3.1092,
      "step": 334500
    },
    {
      "epoch": 0.21814450628690704,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013309820000000002,
      "loss": 3.1069,
      "step": 334510
    },
    {
      "epoch": 0.21816314952306523,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013309620000000001,
      "loss": 3.0567,
      "step": 334520
    },
    {
      "epoch": 0.21818179275922342,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001330942,
      "loss": 3.1842,
      "step": 334530
    },
    {
      "epoch": 0.2182004359953816,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001330922,
      "loss": 3.0361,
      "step": 334540
    },
    {
      "epoch": 0.2182190792315398,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001330902,
      "loss": 3.3355,
      "step": 334550
    },
    {
      "epoch": 0.21823772246769796,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001330882,
      "loss": 3.242,
      "step": 334560
    },
    {
      "epoch": 0.21825636570385615,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001330862,
      "loss": 3.1485,
      "step": 334570
    },
    {
      "epoch": 0.21827500894001434,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013308420000000002,
      "loss": 3.1082,
      "step": 334580
    },
    {
      "epoch": 0.21829365217617253,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001330822,
      "loss": 3.1033,
      "step": 334590
    },
    {
      "epoch": 0.21831229541233071,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001330802,
      "loss": 3.2044,
      "step": 334600
    },
    {
      "epoch": 0.2183309386484889,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001330782,
      "loss": 3.1797,
      "step": 334610
    },
    {
      "epoch": 0.21834958188464706,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001330762,
      "loss": 3.2154,
      "step": 334620
    },
    {
      "epoch": 0.21836822512080525,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013307420000000002,
      "loss": 3.1204,
      "step": 334630
    },
    {
      "epoch": 0.21838686835696344,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001330722,
      "loss": 2.9439,
      "step": 334640
    },
    {
      "epoch": 0.21840551159312163,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001330702,
      "loss": 3.1376,
      "step": 334650
    },
    {
      "epoch": 0.21842415482927982,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001330682,
      "loss": 3.0145,
      "step": 334660
    },
    {
      "epoch": 0.21844279806543798,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001330662,
      "loss": 3.0856,
      "step": 334670
    },
    {
      "epoch": 0.21846144130159617,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013306420000000003,
      "loss": 3.0708,
      "step": 334680
    },
    {
      "epoch": 0.21848008453775436,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001330622,
      "loss": 3.1508,
      "step": 334690
    },
    {
      "epoch": 0.21849872777391255,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013306020000000002,
      "loss": 3.1693,
      "step": 334700
    },
    {
      "epoch": 0.21851737101007074,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001330582,
      "loss": 2.9971,
      "step": 334710
    },
    {
      "epoch": 0.2185360142462289,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001330562,
      "loss": 3.0285,
      "step": 334720
    },
    {
      "epoch": 0.2185546574823871,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001330542,
      "loss": 3.2532,
      "step": 334730
    },
    {
      "epoch": 0.21857330071854528,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001330522,
      "loss": 3.2332,
      "step": 334740
    },
    {
      "epoch": 0.21859194395470347,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013305020000000002,
      "loss": 3.14,
      "step": 334750
    },
    {
      "epoch": 0.21861058719086165,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001330482,
      "loss": 3.0445,
      "step": 334760
    },
    {
      "epoch": 0.21862923042701982,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013304620000000001,
      "loss": 3.0537,
      "step": 334770
    },
    {
      "epoch": 0.218647873663178,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001330442,
      "loss": 2.9664,
      "step": 334780
    },
    {
      "epoch": 0.2186665168993362,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001330422,
      "loss": 3.0837,
      "step": 334790
    },
    {
      "epoch": 0.21868516013549438,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013304020000000003,
      "loss": 3.109,
      "step": 334800
    },
    {
      "epoch": 0.21870380337165257,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001330382,
      "loss": 3.1898,
      "step": 334810
    },
    {
      "epoch": 0.21872244660781073,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013303620000000002,
      "loss": 3.1506,
      "step": 334820
    },
    {
      "epoch": 0.21874108984396892,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001330342,
      "loss": 3.0869,
      "step": 334830
    },
    {
      "epoch": 0.2187597330801271,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001330322,
      "loss": 3.0815,
      "step": 334840
    },
    {
      "epoch": 0.2187783763162853,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001330302,
      "loss": 3.0991,
      "step": 334850
    },
    {
      "epoch": 0.2187970195524435,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001330282,
      "loss": 3.144,
      "step": 334860
    },
    {
      "epoch": 0.21881566278860168,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013302620000000002,
      "loss": 3.0947,
      "step": 334870
    },
    {
      "epoch": 0.21883430602475984,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001330242,
      "loss": 3.2388,
      "step": 334880
    },
    {
      "epoch": 0.21885294926091803,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013302220000000002,
      "loss": 3.2082,
      "step": 334890
    },
    {
      "epoch": 0.21887159249707622,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001330202,
      "loss": 3.1259,
      "step": 334900
    },
    {
      "epoch": 0.2188902357332344,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001330182,
      "loss": 3.0906,
      "step": 334910
    },
    {
      "epoch": 0.2189088789693926,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001330162,
      "loss": 3.0342,
      "step": 334920
    },
    {
      "epoch": 0.21892752220555076,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001330142,
      "loss": 3.1642,
      "step": 334930
    },
    {
      "epoch": 0.21894616544170895,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013301220000000002,
      "loss": 3.1483,
      "step": 334940
    },
    {
      "epoch": 0.21896480867786713,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001330102,
      "loss": 2.9694,
      "step": 334950
    },
    {
      "epoch": 0.21898345191402532,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001330082,
      "loss": 3.1447,
      "step": 334960
    },
    {
      "epoch": 0.2190020951501835,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001330062,
      "loss": 3.174,
      "step": 334970
    },
    {
      "epoch": 0.21902073838634167,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001330042,
      "loss": 3.0408,
      "step": 334980
    },
    {
      "epoch": 0.21903938162249986,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001330022,
      "loss": 3.0851,
      "step": 334990
    },
    {
      "epoch": 0.21905802485865805,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001330002,
      "loss": 3.0967,
      "step": 335000
    },
    {
      "epoch": 0.21907666809481624,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013299820000000002,
      "loss": 3.1019,
      "step": 335010
    },
    {
      "epoch": 0.21909531133097443,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001329962,
      "loss": 3.0156,
      "step": 335020
    },
    {
      "epoch": 0.2191139545671326,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001329942,
      "loss": 3.1924,
      "step": 335030
    },
    {
      "epoch": 0.21913259780329078,
      "grad_norm": 0.75,
      "learning_rate": 0.0001329922,
      "loss": 3.0748,
      "step": 335040
    },
    {
      "epoch": 0.21915124103944897,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001329902,
      "loss": 3.1044,
      "step": 335050
    },
    {
      "epoch": 0.21916988427560716,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001329882,
      "loss": 3.1943,
      "step": 335060
    },
    {
      "epoch": 0.21918852751176535,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013298620000000002,
      "loss": 3.1418,
      "step": 335070
    },
    {
      "epoch": 0.21920717074792354,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001329842,
      "loss": 3.1444,
      "step": 335080
    },
    {
      "epoch": 0.2192258139840817,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001329822,
      "loss": 3.1206,
      "step": 335090
    },
    {
      "epoch": 0.21924445722023989,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001329802,
      "loss": 3.107,
      "step": 335100
    },
    {
      "epoch": 0.21926310045639807,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001329782,
      "loss": 3.133,
      "step": 335110
    },
    {
      "epoch": 0.21928174369255626,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013297620000000002,
      "loss": 3.2246,
      "step": 335120
    },
    {
      "epoch": 0.21930038692871445,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001329742,
      "loss": 3.1054,
      "step": 335130
    },
    {
      "epoch": 0.2193190301648726,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013297220000000001,
      "loss": 3.0075,
      "step": 335140
    },
    {
      "epoch": 0.2193376734010308,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001329702,
      "loss": 3.1651,
      "step": 335150
    },
    {
      "epoch": 0.219356316637189,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001329682,
      "loss": 3.1321,
      "step": 335160
    },
    {
      "epoch": 0.21937495987334718,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001329662,
      "loss": 3.1479,
      "step": 335170
    },
    {
      "epoch": 0.21939360310950537,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001329642,
      "loss": 3.096,
      "step": 335180
    },
    {
      "epoch": 0.21941224634566353,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013296220000000002,
      "loss": 3.031,
      "step": 335190
    },
    {
      "epoch": 0.21943088958182172,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013296020000000001,
      "loss": 3.0415,
      "step": 335200
    },
    {
      "epoch": 0.2194495328179799,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001329582,
      "loss": 2.9758,
      "step": 335210
    },
    {
      "epoch": 0.2194681760541381,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001329562,
      "loss": 3.2487,
      "step": 335220
    },
    {
      "epoch": 0.2194868192902963,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001329542,
      "loss": 2.9633,
      "step": 335230
    },
    {
      "epoch": 0.21950546252645445,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013295220000000002,
      "loss": 3.152,
      "step": 335240
    },
    {
      "epoch": 0.21952410576261264,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001329502,
      "loss": 3.1242,
      "step": 335250
    },
    {
      "epoch": 0.21954274899877083,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013294820000000002,
      "loss": 3.0648,
      "step": 335260
    },
    {
      "epoch": 0.21956139223492901,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001329462,
      "loss": 3.0155,
      "step": 335270
    },
    {
      "epoch": 0.2195800354710872,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001329442,
      "loss": 3.109,
      "step": 335280
    },
    {
      "epoch": 0.2195986787072454,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013294220000000003,
      "loss": 3.085,
      "step": 335290
    },
    {
      "epoch": 0.21961732194340355,
      "grad_norm": 0.75,
      "learning_rate": 0.0001329402,
      "loss": 3.1204,
      "step": 335300
    },
    {
      "epoch": 0.21963596517956174,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013293820000000002,
      "loss": 3.2362,
      "step": 335310
    },
    {
      "epoch": 0.21965460841571993,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001329362,
      "loss": 3.1636,
      "step": 335320
    },
    {
      "epoch": 0.21967325165187812,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001329342,
      "loss": 3.1018,
      "step": 335330
    },
    {
      "epoch": 0.2196918948880363,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001329322,
      "loss": 3.1018,
      "step": 335340
    },
    {
      "epoch": 0.21971053812419447,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001329302,
      "loss": 3.0896,
      "step": 335350
    },
    {
      "epoch": 0.21972918136035266,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013292820000000003,
      "loss": 3.1172,
      "step": 335360
    },
    {
      "epoch": 0.21974782459651085,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001329262,
      "loss": 3.1539,
      "step": 335370
    },
    {
      "epoch": 0.21976646783266904,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013292420000000002,
      "loss": 3.2406,
      "step": 335380
    },
    {
      "epoch": 0.21978511106882723,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013292219999999999,
      "loss": 3.2094,
      "step": 335390
    },
    {
      "epoch": 0.2198037543049854,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001329202,
      "loss": 3.2245,
      "step": 335400
    },
    {
      "epoch": 0.21982239754114358,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001329182,
      "loss": 3.1463,
      "step": 335410
    },
    {
      "epoch": 0.21984104077730177,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001329162,
      "loss": 3.15,
      "step": 335420
    },
    {
      "epoch": 0.21985968401345996,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013291420000000002,
      "loss": 3.1764,
      "step": 335430
    },
    {
      "epoch": 0.21987832724961814,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001329122,
      "loss": 3.1474,
      "step": 335440
    },
    {
      "epoch": 0.2198969704857763,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013291020000000001,
      "loss": 3.0231,
      "step": 335450
    },
    {
      "epoch": 0.2199156137219345,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001329082,
      "loss": 3.0398,
      "step": 335460
    },
    {
      "epoch": 0.21993425695809268,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001329062,
      "loss": 3.0594,
      "step": 335470
    },
    {
      "epoch": 0.21995290019425087,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001329042,
      "loss": 3.1236,
      "step": 335480
    },
    {
      "epoch": 0.21997154343040906,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001329022,
      "loss": 3.1454,
      "step": 335490
    },
    {
      "epoch": 0.21999018666656722,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013290020000000002,
      "loss": 2.9763,
      "step": 335500
    },
    {
      "epoch": 0.2200088299027254,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013289820000000001,
      "loss": 3.2541,
      "step": 335510
    },
    {
      "epoch": 0.2200274731388836,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001328962,
      "loss": 3.0875,
      "step": 335520
    },
    {
      "epoch": 0.2200461163750418,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001328942,
      "loss": 3.2538,
      "step": 335530
    },
    {
      "epoch": 0.22006475961119998,
      "grad_norm": 0.75,
      "learning_rate": 0.0001328922,
      "loss": 3.1148,
      "step": 335540
    },
    {
      "epoch": 0.22008340284735817,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001328902,
      "loss": 3.1962,
      "step": 335550
    },
    {
      "epoch": 0.22010204608351633,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001328882,
      "loss": 3.1618,
      "step": 335560
    },
    {
      "epoch": 0.22012068931967452,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013288620000000002,
      "loss": 3.0347,
      "step": 335570
    },
    {
      "epoch": 0.2201393325558327,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001328842,
      "loss": 3.0509,
      "step": 335580
    },
    {
      "epoch": 0.2201579757919909,
      "grad_norm": 0.75,
      "learning_rate": 0.0001328822,
      "loss": 3.1608,
      "step": 335590
    },
    {
      "epoch": 0.22017661902814908,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001328802,
      "loss": 3.0788,
      "step": 335600
    },
    {
      "epoch": 0.22019526226430725,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001328782,
      "loss": 3.1464,
      "step": 335610
    },
    {
      "epoch": 0.22021390550046543,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001328762,
      "loss": 3.0784,
      "step": 335620
    },
    {
      "epoch": 0.22023254873662362,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013287420000000002,
      "loss": 3.1112,
      "step": 335630
    },
    {
      "epoch": 0.2202511919727818,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001328722,
      "loss": 3.0631,
      "step": 335640
    },
    {
      "epoch": 0.22026983520894,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001328702,
      "loss": 3.1929,
      "step": 335650
    },
    {
      "epoch": 0.22028847844509816,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001328682,
      "loss": 3.0877,
      "step": 335660
    },
    {
      "epoch": 0.22030712168125635,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001328662,
      "loss": 3.1489,
      "step": 335670
    },
    {
      "epoch": 0.22032576491741454,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013286420000000002,
      "loss": 3.1279,
      "step": 335680
    },
    {
      "epoch": 0.22034440815357273,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013286220000000002,
      "loss": 3.295,
      "step": 335690
    },
    {
      "epoch": 0.22036305138973092,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001328602,
      "loss": 3.2429,
      "step": 335700
    },
    {
      "epoch": 0.22038169462588908,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001328582,
      "loss": 3.0,
      "step": 335710
    },
    {
      "epoch": 0.22040033786204727,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001328562,
      "loss": 3.0485,
      "step": 335720
    },
    {
      "epoch": 0.22041898109820546,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001328542,
      "loss": 2.9449,
      "step": 335730
    },
    {
      "epoch": 0.22043762433436365,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001328522,
      "loss": 3.1338,
      "step": 335740
    },
    {
      "epoch": 0.22045626757052184,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013285020000000002,
      "loss": 3.1539,
      "step": 335750
    },
    {
      "epoch": 0.22047491080668002,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001328482,
      "loss": 3.2195,
      "step": 335760
    },
    {
      "epoch": 0.22049355404283819,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001328462,
      "loss": 3.1245,
      "step": 335770
    },
    {
      "epoch": 0.22051219727899637,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001328442,
      "loss": 3.15,
      "step": 335780
    },
    {
      "epoch": 0.22053084051515456,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001328422,
      "loss": 3.3395,
      "step": 335790
    },
    {
      "epoch": 0.22054948375131275,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013284020000000002,
      "loss": 3.1047,
      "step": 335800
    },
    {
      "epoch": 0.22056812698747094,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001328382,
      "loss": 3.1768,
      "step": 335810
    },
    {
      "epoch": 0.2205867702236291,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013283620000000001,
      "loss": 3.1533,
      "step": 335820
    },
    {
      "epoch": 0.2206054134597873,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001328342,
      "loss": 3.0809,
      "step": 335830
    },
    {
      "epoch": 0.22062405669594548,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001328322,
      "loss": 3.0161,
      "step": 335840
    },
    {
      "epoch": 0.22064269993210367,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013283020000000003,
      "loss": 3.2359,
      "step": 335850
    },
    {
      "epoch": 0.22066134316826186,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001328282,
      "loss": 3.1013,
      "step": 335860
    },
    {
      "epoch": 0.22067998640442002,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013282620000000002,
      "loss": 3.1073,
      "step": 335870
    },
    {
      "epoch": 0.2206986296405782,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001328242,
      "loss": 3.1076,
      "step": 335880
    },
    {
      "epoch": 0.2207172728767364,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001328222,
      "loss": 3.0324,
      "step": 335890
    },
    {
      "epoch": 0.2207359161128946,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001328202,
      "loss": 3.1487,
      "step": 335900
    },
    {
      "epoch": 0.22075455934905278,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001328182,
      "loss": 3.139,
      "step": 335910
    },
    {
      "epoch": 0.22077320258521094,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013281620000000002,
      "loss": 3.0681,
      "step": 335920
    },
    {
      "epoch": 0.22079184582136913,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001328142,
      "loss": 3.2325,
      "step": 335930
    },
    {
      "epoch": 0.22081048905752731,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013281220000000002,
      "loss": 3.0538,
      "step": 335940
    },
    {
      "epoch": 0.2208291322936855,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013281019999999998,
      "loss": 3.2381,
      "step": 335950
    },
    {
      "epoch": 0.2208477755298437,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001328082,
      "loss": 3.217,
      "step": 335960
    },
    {
      "epoch": 0.22086641876600188,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001328062,
      "loss": 3.1043,
      "step": 335970
    },
    {
      "epoch": 0.22088506200216004,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001328042,
      "loss": 3.119,
      "step": 335980
    },
    {
      "epoch": 0.22090370523831823,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013280220000000002,
      "loss": 3.1114,
      "step": 335990
    },
    {
      "epoch": 0.22092234847447642,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001328002,
      "loss": 3.0776,
      "step": 336000
    },
    {
      "epoch": 0.2209409917106346,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001327982,
      "loss": 3.1129,
      "step": 336010
    },
    {
      "epoch": 0.2209596349467928,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001327962,
      "loss": 3.1907,
      "step": 336020
    },
    {
      "epoch": 0.22097827818295096,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001327942,
      "loss": 3.0786,
      "step": 336030
    },
    {
      "epoch": 0.22099692141910915,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001327922,
      "loss": 3.0922,
      "step": 336040
    },
    {
      "epoch": 0.22101556465526734,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001327902,
      "loss": 3.0243,
      "step": 336050
    },
    {
      "epoch": 0.22103420789142553,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013278820000000002,
      "loss": 3.1667,
      "step": 336060
    },
    {
      "epoch": 0.22105285112758372,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001327862,
      "loss": 3.1026,
      "step": 336070
    },
    {
      "epoch": 0.22107149436374188,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001327842,
      "loss": 3.1324,
      "step": 336080
    },
    {
      "epoch": 0.22109013759990007,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001327822,
      "loss": 3.1221,
      "step": 336090
    },
    {
      "epoch": 0.22110878083605826,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001327802,
      "loss": 3.1099,
      "step": 336100
    },
    {
      "epoch": 0.22112742407221644,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001327782,
      "loss": 3.2127,
      "step": 336110
    },
    {
      "epoch": 0.22114606730837463,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013277620000000002,
      "loss": 3.1331,
      "step": 336120
    },
    {
      "epoch": 0.2211647105445328,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013277420000000001,
      "loss": 3.0838,
      "step": 336130
    },
    {
      "epoch": 0.22118335378069098,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001327722,
      "loss": 3.0828,
      "step": 336140
    },
    {
      "epoch": 0.22120199701684917,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001327702,
      "loss": 3.0564,
      "step": 336150
    },
    {
      "epoch": 0.22122064025300736,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001327682,
      "loss": 3.1791,
      "step": 336160
    },
    {
      "epoch": 0.22123928348916555,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001327662,
      "loss": 3.2144,
      "step": 336170
    },
    {
      "epoch": 0.2212579267253237,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001327642,
      "loss": 3.1415,
      "step": 336180
    },
    {
      "epoch": 0.2212765699614819,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013276220000000001,
      "loss": 3.0197,
      "step": 336190
    },
    {
      "epoch": 0.2212952131976401,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001327602,
      "loss": 3.0535,
      "step": 336200
    },
    {
      "epoch": 0.22131385643379828,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001327582,
      "loss": 3.08,
      "step": 336210
    },
    {
      "epoch": 0.22133249966995647,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001327562,
      "loss": 3.1979,
      "step": 336220
    },
    {
      "epoch": 0.22135114290611466,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001327542,
      "loss": 3.1741,
      "step": 336230
    },
    {
      "epoch": 0.22136978614227282,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013275220000000002,
      "loss": 3.2147,
      "step": 336240
    },
    {
      "epoch": 0.221388429378431,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013275020000000002,
      "loss": 3.191,
      "step": 336250
    },
    {
      "epoch": 0.2214070726145892,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001327482,
      "loss": 3.0899,
      "step": 336260
    },
    {
      "epoch": 0.22142571585074738,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001327462,
      "loss": 3.0883,
      "step": 336270
    },
    {
      "epoch": 0.22144435908690557,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001327442,
      "loss": 3.2065,
      "step": 336280
    },
    {
      "epoch": 0.22146300232306373,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013274220000000002,
      "loss": 3.1201,
      "step": 336290
    },
    {
      "epoch": 0.22148164555922192,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001327402,
      "loss": 3.1909,
      "step": 336300
    },
    {
      "epoch": 0.2215002887953801,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013273820000000002,
      "loss": 3.1905,
      "step": 336310
    },
    {
      "epoch": 0.2215189320315383,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001327362,
      "loss": 3.0833,
      "step": 336320
    },
    {
      "epoch": 0.2215375752676965,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001327342,
      "loss": 3.168,
      "step": 336330
    },
    {
      "epoch": 0.22155621850385465,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001327322,
      "loss": 3.1521,
      "step": 336340
    },
    {
      "epoch": 0.22157486174001284,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001327302,
      "loss": 3.216,
      "step": 336350
    },
    {
      "epoch": 0.22159350497617103,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013272820000000002,
      "loss": 3.0974,
      "step": 336360
    },
    {
      "epoch": 0.22161214821232922,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001327262,
      "loss": 3.1964,
      "step": 336370
    },
    {
      "epoch": 0.2216307914484874,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001327242,
      "loss": 3.1277,
      "step": 336380
    },
    {
      "epoch": 0.22164943468464557,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001327222,
      "loss": 3.0953,
      "step": 336390
    },
    {
      "epoch": 0.22166807792080376,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001327202,
      "loss": 3.1199,
      "step": 336400
    },
    {
      "epoch": 0.22168672115696195,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013271820000000003,
      "loss": 3.2284,
      "step": 336410
    },
    {
      "epoch": 0.22170536439312014,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001327162,
      "loss": 2.9835,
      "step": 336420
    },
    {
      "epoch": 0.22172400762927832,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013271420000000002,
      "loss": 3.0489,
      "step": 336430
    },
    {
      "epoch": 0.2217426508654365,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013271219999999999,
      "loss": 3.1726,
      "step": 336440
    },
    {
      "epoch": 0.22176129410159467,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001327102,
      "loss": 3.0591,
      "step": 336450
    },
    {
      "epoch": 0.22177993733775286,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001327082,
      "loss": 3.0488,
      "step": 336460
    },
    {
      "epoch": 0.22179858057391105,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001327062,
      "loss": 3.1438,
      "step": 336470
    },
    {
      "epoch": 0.22181722381006924,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013270420000000002,
      "loss": 2.9546,
      "step": 336480
    },
    {
      "epoch": 0.22183586704622743,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001327022,
      "loss": 3.2265,
      "step": 336490
    },
    {
      "epoch": 0.2218545102823856,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013270020000000001,
      "loss": 3.0856,
      "step": 336500
    },
    {
      "epoch": 0.22187315351854378,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001326982,
      "loss": 3.1944,
      "step": 336510
    },
    {
      "epoch": 0.22189179675470197,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001326962,
      "loss": 3.0895,
      "step": 336520
    },
    {
      "epoch": 0.22191043999086016,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001326942,
      "loss": 3.0396,
      "step": 336530
    },
    {
      "epoch": 0.22192908322701835,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001326922,
      "loss": 3.114,
      "step": 336540
    },
    {
      "epoch": 0.2219477264631765,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013269020000000002,
      "loss": 3.09,
      "step": 336550
    },
    {
      "epoch": 0.2219663696993347,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001326882,
      "loss": 3.2305,
      "step": 336560
    },
    {
      "epoch": 0.2219850129354929,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001326862,
      "loss": 2.9676,
      "step": 336570
    },
    {
      "epoch": 0.22200365617165108,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001326842,
      "loss": 3.2702,
      "step": 336580
    },
    {
      "epoch": 0.22202229940780926,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001326822,
      "loss": 3.0592,
      "step": 336590
    },
    {
      "epoch": 0.22204094264396743,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001326802,
      "loss": 3.1296,
      "step": 336600
    },
    {
      "epoch": 0.22205958588012562,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001326782,
      "loss": 3.2599,
      "step": 336610
    },
    {
      "epoch": 0.2220782291162838,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013267620000000002,
      "loss": 3.065,
      "step": 336620
    },
    {
      "epoch": 0.222096872352442,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001326742,
      "loss": 3.0769,
      "step": 336630
    },
    {
      "epoch": 0.22211551558860018,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001326722,
      "loss": 3.0387,
      "step": 336640
    },
    {
      "epoch": 0.22213415882475837,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001326702,
      "loss": 3.0635,
      "step": 336650
    },
    {
      "epoch": 0.22215280206091653,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001326682,
      "loss": 3.0804,
      "step": 336660
    },
    {
      "epoch": 0.22217144529707472,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001326662,
      "loss": 3.0187,
      "step": 336670
    },
    {
      "epoch": 0.2221900885332329,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013266420000000002,
      "loss": 3.0463,
      "step": 336680
    },
    {
      "epoch": 0.2222087317693911,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001326622,
      "loss": 3.0198,
      "step": 336690
    },
    {
      "epoch": 0.2222273750055493,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001326602,
      "loss": 3.2077,
      "step": 336700
    },
    {
      "epoch": 0.22224601824170745,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001326582,
      "loss": 3.1021,
      "step": 336710
    },
    {
      "epoch": 0.22226466147786564,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001326562,
      "loss": 3.1492,
      "step": 336720
    },
    {
      "epoch": 0.22228330471402383,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013265420000000002,
      "loss": 3.1814,
      "step": 336730
    },
    {
      "epoch": 0.22230194795018202,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013265220000000002,
      "loss": 3.1543,
      "step": 336740
    },
    {
      "epoch": 0.2223205911863402,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001326502,
      "loss": 3.0777,
      "step": 336750
    },
    {
      "epoch": 0.22233923442249837,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001326482,
      "loss": 3.0743,
      "step": 336760
    },
    {
      "epoch": 0.22235787765865656,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001326462,
      "loss": 3.1334,
      "step": 336770
    },
    {
      "epoch": 0.22237652089481474,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001326442,
      "loss": 3.0567,
      "step": 336780
    },
    {
      "epoch": 0.22239516413097293,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001326422,
      "loss": 2.998,
      "step": 336790
    },
    {
      "epoch": 0.22241380736713112,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013264020000000002,
      "loss": 3.0971,
      "step": 336800
    },
    {
      "epoch": 0.22243245060328928,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013263820000000001,
      "loss": 3.1796,
      "step": 336810
    },
    {
      "epoch": 0.22245109383944747,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001326362,
      "loss": 3.1858,
      "step": 336820
    },
    {
      "epoch": 0.22246973707560566,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001326342,
      "loss": 3.1834,
      "step": 336830
    },
    {
      "epoch": 0.22248838031176385,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001326322,
      "loss": 3.0497,
      "step": 336840
    },
    {
      "epoch": 0.22250702354792204,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013263020000000002,
      "loss": 3.2138,
      "step": 336850
    },
    {
      "epoch": 0.2225256667840802,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001326282,
      "loss": 3.0992,
      "step": 336860
    },
    {
      "epoch": 0.2225443100202384,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013262620000000001,
      "loss": 3.0885,
      "step": 336870
    },
    {
      "epoch": 0.22256295325639658,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001326242,
      "loss": 3.1391,
      "step": 336880
    },
    {
      "epoch": 0.22258159649255477,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001326222,
      "loss": 3.0983,
      "step": 336890
    },
    {
      "epoch": 0.22260023972871296,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013262020000000003,
      "loss": 3.0245,
      "step": 336900
    },
    {
      "epoch": 0.22261888296487115,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001326182,
      "loss": 3.1536,
      "step": 336910
    },
    {
      "epoch": 0.2226375262010293,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013261620000000002,
      "loss": 3.0562,
      "step": 336920
    },
    {
      "epoch": 0.2226561694371875,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001326142,
      "loss": 3.1241,
      "step": 336930
    },
    {
      "epoch": 0.22267481267334568,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001326122,
      "loss": 3.0997,
      "step": 336940
    },
    {
      "epoch": 0.22269345590950387,
      "grad_norm": 0.75,
      "learning_rate": 0.0001326102,
      "loss": 3.2746,
      "step": 336950
    },
    {
      "epoch": 0.22271209914566206,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001326082,
      "loss": 3.018,
      "step": 336960
    },
    {
      "epoch": 0.22273074238182022,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013260620000000002,
      "loss": 3.1396,
      "step": 336970
    },
    {
      "epoch": 0.2227493856179784,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001326042,
      "loss": 3.0983,
      "step": 336980
    },
    {
      "epoch": 0.2227680288541366,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013260220000000002,
      "loss": 3.1656,
      "step": 336990
    },
    {
      "epoch": 0.2227866720902948,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013260019999999998,
      "loss": 3.1364,
      "step": 337000
    },
    {
      "epoch": 0.22280531532645298,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001325982,
      "loss": 3.0292,
      "step": 337010
    },
    {
      "epoch": 0.22282395856261114,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001325962,
      "loss": 3.2029,
      "step": 337020
    },
    {
      "epoch": 0.22284260179876933,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001325942,
      "loss": 3.0299,
      "step": 337030
    },
    {
      "epoch": 0.22286124503492752,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013259220000000002,
      "loss": 3.0751,
      "step": 337040
    },
    {
      "epoch": 0.2228798882710857,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001325902,
      "loss": 3.0736,
      "step": 337050
    },
    {
      "epoch": 0.2228985315072439,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001325882,
      "loss": 3.1126,
      "step": 337060
    },
    {
      "epoch": 0.22291717474340206,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001325862,
      "loss": 3.2227,
      "step": 337070
    },
    {
      "epoch": 0.22293581797956025,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001325842,
      "loss": 2.9518,
      "step": 337080
    },
    {
      "epoch": 0.22295446121571844,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001325822,
      "loss": 3.2157,
      "step": 337090
    },
    {
      "epoch": 0.22297310445187662,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001325802,
      "loss": 3.1467,
      "step": 337100
    },
    {
      "epoch": 0.2229917476880348,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013257820000000002,
      "loss": 3.2192,
      "step": 337110
    },
    {
      "epoch": 0.223010390924193,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001325762,
      "loss": 3.1595,
      "step": 337120
    },
    {
      "epoch": 0.22302903416035116,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001325742,
      "loss": 2.9692,
      "step": 337130
    },
    {
      "epoch": 0.22304767739650935,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001325722,
      "loss": 3.1928,
      "step": 337140
    },
    {
      "epoch": 0.22306632063266754,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001325702,
      "loss": 3.0899,
      "step": 337150
    },
    {
      "epoch": 0.22308496386882573,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001325682,
      "loss": 3.1068,
      "step": 337160
    },
    {
      "epoch": 0.22310360710498392,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001325662,
      "loss": 3.1876,
      "step": 337170
    },
    {
      "epoch": 0.22312225034114208,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013256420000000001,
      "loss": 3.2258,
      "step": 337180
    },
    {
      "epoch": 0.22314089357730027,
      "grad_norm": 0.625,
      "learning_rate": 0.0001325622,
      "loss": 3.0974,
      "step": 337190
    },
    {
      "epoch": 0.22315953681345846,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001325602,
      "loss": 3.1457,
      "step": 337200
    },
    {
      "epoch": 0.22317818004961665,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001325582,
      "loss": 3.0943,
      "step": 337210
    },
    {
      "epoch": 0.22319682328577484,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001325562,
      "loss": 3.1129,
      "step": 337220
    },
    {
      "epoch": 0.223215466521933,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013255420000000002,
      "loss": 3.0898,
      "step": 337230
    },
    {
      "epoch": 0.2232341097580912,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013255220000000001,
      "loss": 3.194,
      "step": 337240
    },
    {
      "epoch": 0.22325275299424938,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001325502,
      "loss": 3.1118,
      "step": 337250
    },
    {
      "epoch": 0.22327139623040756,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001325482,
      "loss": 3.1125,
      "step": 337260
    },
    {
      "epoch": 0.22329003946656575,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001325462,
      "loss": 3.1167,
      "step": 337270
    },
    {
      "epoch": 0.22330868270272392,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001325442,
      "loss": 3.0156,
      "step": 337280
    },
    {
      "epoch": 0.2233273259388821,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013254220000000002,
      "loss": 3.1955,
      "step": 337290
    },
    {
      "epoch": 0.2233459691750403,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013254020000000002,
      "loss": 3.0993,
      "step": 337300
    },
    {
      "epoch": 0.22336461241119848,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001325382,
      "loss": 3.1737,
      "step": 337310
    },
    {
      "epoch": 0.22338325564735667,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001325362,
      "loss": 3.1402,
      "step": 337320
    },
    {
      "epoch": 0.22340189888351486,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001325342,
      "loss": 3.2435,
      "step": 337330
    },
    {
      "epoch": 0.22342054211967302,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001325322,
      "loss": 3.0176,
      "step": 337340
    },
    {
      "epoch": 0.2234391853558312,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001325302,
      "loss": 3.0992,
      "step": 337350
    },
    {
      "epoch": 0.2234578285919894,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013252820000000002,
      "loss": 3.1583,
      "step": 337360
    },
    {
      "epoch": 0.2234764718281476,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001325262,
      "loss": 3.092,
      "step": 337370
    },
    {
      "epoch": 0.22349511506430578,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001325242,
      "loss": 3.1622,
      "step": 337380
    },
    {
      "epoch": 0.22351375830046394,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001325222,
      "loss": 3.1455,
      "step": 337390
    },
    {
      "epoch": 0.22353240153662213,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001325202,
      "loss": 3.2482,
      "step": 337400
    },
    {
      "epoch": 0.22355104477278032,
      "grad_norm": 0.828125,
      "learning_rate": 0.00013251820000000002,
      "loss": 3.106,
      "step": 337410
    },
    {
      "epoch": 0.2235696880089385,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001325162,
      "loss": 3.1218,
      "step": 337420
    },
    {
      "epoch": 0.2235883312450967,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001325142,
      "loss": 3.2107,
      "step": 337430
    },
    {
      "epoch": 0.22360697448125486,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001325122,
      "loss": 3.1566,
      "step": 337440
    },
    {
      "epoch": 0.22362561771741304,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001325102,
      "loss": 3.1061,
      "step": 337450
    },
    {
      "epoch": 0.22364426095357123,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013250820000000003,
      "loss": 3.0245,
      "step": 337460
    },
    {
      "epoch": 0.22366290418972942,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001325062,
      "loss": 3.136,
      "step": 337470
    },
    {
      "epoch": 0.2236815474258876,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013250420000000002,
      "loss": 3.1456,
      "step": 337480
    },
    {
      "epoch": 0.22370019066204577,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013250219999999999,
      "loss": 3.0608,
      "step": 337490
    },
    {
      "epoch": 0.22371883389820396,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001325002,
      "loss": 3.0186,
      "step": 337500
    },
    {
      "epoch": 0.22373747713436215,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001324982,
      "loss": 3.1958,
      "step": 337510
    },
    {
      "epoch": 0.22375612037052034,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001324962,
      "loss": 3.1501,
      "step": 337520
    },
    {
      "epoch": 0.22377476360667853,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013249420000000002,
      "loss": 3.3141,
      "step": 337530
    },
    {
      "epoch": 0.22379340684283672,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001324922,
      "loss": 3.1393,
      "step": 337540
    },
    {
      "epoch": 0.22381205007899488,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013249020000000001,
      "loss": 3.1519,
      "step": 337550
    },
    {
      "epoch": 0.22383069331515307,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013248819999999998,
      "loss": 3.1187,
      "step": 337560
    },
    {
      "epoch": 0.22384933655131126,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001324862,
      "loss": 3.1418,
      "step": 337570
    },
    {
      "epoch": 0.22386797978746945,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001324842,
      "loss": 3.1504,
      "step": 337580
    },
    {
      "epoch": 0.22388662302362763,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001324822,
      "loss": 3.0038,
      "step": 337590
    },
    {
      "epoch": 0.2239052662597858,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013248020000000002,
      "loss": 3.1548,
      "step": 337600
    },
    {
      "epoch": 0.22392390949594398,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001324782,
      "loss": 3.1421,
      "step": 337610
    },
    {
      "epoch": 0.22394255273210217,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001324762,
      "loss": 3.099,
      "step": 337620
    },
    {
      "epoch": 0.22396119596826036,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001324742,
      "loss": 3.0395,
      "step": 337630
    },
    {
      "epoch": 0.22397983920441855,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001324722,
      "loss": 3.0898,
      "step": 337640
    },
    {
      "epoch": 0.2239984824405767,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001324702,
      "loss": 3.1322,
      "step": 337650
    },
    {
      "epoch": 0.2240171256767349,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001324682,
      "loss": 3.0014,
      "step": 337660
    },
    {
      "epoch": 0.2240357689128931,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013246620000000002,
      "loss": 3.1376,
      "step": 337670
    },
    {
      "epoch": 0.22405441214905128,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001324642,
      "loss": 3.0635,
      "step": 337680
    },
    {
      "epoch": 0.22407305538520947,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001324622,
      "loss": 3.1881,
      "step": 337690
    },
    {
      "epoch": 0.22409169862136763,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001324602,
      "loss": 3.1156,
      "step": 337700
    },
    {
      "epoch": 0.22411034185752582,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001324582,
      "loss": 3.171,
      "step": 337710
    },
    {
      "epoch": 0.224128985093684,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013245620000000002,
      "loss": 2.996,
      "step": 337720
    },
    {
      "epoch": 0.2241476283298422,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013245420000000002,
      "loss": 3.1384,
      "step": 337730
    },
    {
      "epoch": 0.22416627156600039,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001324522,
      "loss": 3.196,
      "step": 337740
    },
    {
      "epoch": 0.22418491480215855,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001324502,
      "loss": 2.9877,
      "step": 337750
    },
    {
      "epoch": 0.22420355803831674,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001324482,
      "loss": 3.0593,
      "step": 337760
    },
    {
      "epoch": 0.22422220127447492,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001324462,
      "loss": 3.1697,
      "step": 337770
    },
    {
      "epoch": 0.2242408445106331,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001324442,
      "loss": 3.1974,
      "step": 337780
    },
    {
      "epoch": 0.2242594877467913,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013244220000000002,
      "loss": 3.182,
      "step": 337790
    },
    {
      "epoch": 0.2242781309829495,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001324402,
      "loss": 3.1167,
      "step": 337800
    },
    {
      "epoch": 0.22429677421910765,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001324382,
      "loss": 3.1174,
      "step": 337810
    },
    {
      "epoch": 0.22431541745526584,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001324362,
      "loss": 3.0188,
      "step": 337820
    },
    {
      "epoch": 0.22433406069142403,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001324342,
      "loss": 3.1059,
      "step": 337830
    },
    {
      "epoch": 0.22435270392758222,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001324322,
      "loss": 3.1221,
      "step": 337840
    },
    {
      "epoch": 0.2243713471637404,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013243020000000002,
      "loss": 3.1179,
      "step": 337850
    },
    {
      "epoch": 0.22438999039989857,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013242820000000001,
      "loss": 3.0645,
      "step": 337860
    },
    {
      "epoch": 0.22440863363605676,
      "grad_norm": 0.75,
      "learning_rate": 0.0001324262,
      "loss": 3.1984,
      "step": 337870
    },
    {
      "epoch": 0.22442727687221495,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001324242,
      "loss": 3.0488,
      "step": 337880
    },
    {
      "epoch": 0.22444592010837314,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001324222,
      "loss": 3.1934,
      "step": 337890
    },
    {
      "epoch": 0.22446456334453133,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013242020000000002,
      "loss": 3.1488,
      "step": 337900
    },
    {
      "epoch": 0.2244832065806895,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001324182,
      "loss": 3.1315,
      "step": 337910
    },
    {
      "epoch": 0.22450184981684768,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013241620000000001,
      "loss": 2.9948,
      "step": 337920
    },
    {
      "epoch": 0.22452049305300587,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001324142,
      "loss": 3.1307,
      "step": 337930
    },
    {
      "epoch": 0.22453913628916405,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001324122,
      "loss": 3.1738,
      "step": 337940
    },
    {
      "epoch": 0.22455777952532224,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001324102,
      "loss": 3.2801,
      "step": 337950
    },
    {
      "epoch": 0.2245764227614804,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001324082,
      "loss": 3.1061,
      "step": 337960
    },
    {
      "epoch": 0.2245950659976386,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013240620000000002,
      "loss": 3.0984,
      "step": 337970
    },
    {
      "epoch": 0.22461370923379678,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001324042,
      "loss": 3.1067,
      "step": 337980
    },
    {
      "epoch": 0.22463235246995497,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001324022,
      "loss": 3.012,
      "step": 337990
    },
    {
      "epoch": 0.22465099570611316,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001324002,
      "loss": 3.1032,
      "step": 338000
    },
    {
      "epoch": 0.22466963894227135,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001323982,
      "loss": 3.0214,
      "step": 338010
    },
    {
      "epoch": 0.2246882821784295,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013239620000000003,
      "loss": 3.1161,
      "step": 338020
    },
    {
      "epoch": 0.2247069254145877,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001323942,
      "loss": 3.1547,
      "step": 338030
    },
    {
      "epoch": 0.2247255686507459,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013239220000000002,
      "loss": 3.2044,
      "step": 338040
    },
    {
      "epoch": 0.22474421188690408,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013239019999999998,
      "loss": 3.0253,
      "step": 338050
    },
    {
      "epoch": 0.22476285512306227,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001323882,
      "loss": 3.106,
      "step": 338060
    },
    {
      "epoch": 0.22478149835922043,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001323862,
      "loss": 3.0512,
      "step": 338070
    },
    {
      "epoch": 0.22480014159537862,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001323842,
      "loss": 3.0582,
      "step": 338080
    },
    {
      "epoch": 0.2248187848315368,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013238220000000002,
      "loss": 3.13,
      "step": 338090
    },
    {
      "epoch": 0.224837428067695,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001323802,
      "loss": 3.0188,
      "step": 338100
    },
    {
      "epoch": 0.22485607130385318,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001323782,
      "loss": 3.2534,
      "step": 338110
    },
    {
      "epoch": 0.22487471454001134,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001323762,
      "loss": 3.1956,
      "step": 338120
    },
    {
      "epoch": 0.22489335777616953,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001323742,
      "loss": 3.1469,
      "step": 338130
    },
    {
      "epoch": 0.22491200101232772,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001323722,
      "loss": 3.0983,
      "step": 338140
    },
    {
      "epoch": 0.2249306442484859,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001323702,
      "loss": 3.0661,
      "step": 338150
    },
    {
      "epoch": 0.2249492874846441,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013236820000000002,
      "loss": 3.1491,
      "step": 338160
    },
    {
      "epoch": 0.22496793072080226,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00013236619999999999,
      "loss": 3.0099,
      "step": 338170
    },
    {
      "epoch": 0.22498657395696045,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001323642,
      "loss": 3.1674,
      "step": 338180
    },
    {
      "epoch": 0.22500521719311864,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001323622,
      "loss": 3.0947,
      "step": 338190
    },
    {
      "epoch": 0.22502386042927683,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001323602,
      "loss": 3.1883,
      "step": 338200
    },
    {
      "epoch": 0.22504250366543502,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00013235820000000002,
      "loss": 3.0919,
      "step": 338210
    },
    {
      "epoch": 0.2250611469015932,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001323562,
      "loss": 3.1064,
      "step": 338220
    },
    {
      "epoch": 0.22507979013775137,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013235420000000001,
      "loss": 3.0255,
      "step": 338230
    },
    {
      "epoch": 0.22509843337390956,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001323522,
      "loss": 3.1315,
      "step": 338240
    },
    {
      "epoch": 0.22511707661006775,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001323502,
      "loss": 3.1272,
      "step": 338250
    },
    {
      "epoch": 0.22513571984622593,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001323482,
      "loss": 3.1375,
      "step": 338260
    },
    {
      "epoch": 0.22515436308238412,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001323462,
      "loss": 3.1458,
      "step": 338270
    },
    {
      "epoch": 0.22517300631854228,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013234420000000002,
      "loss": 3.0682,
      "step": 338280
    },
    {
      "epoch": 0.22519164955470047,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013234220000000002,
      "loss": 3.1361,
      "step": 338290
    },
    {
      "epoch": 0.22521029279085866,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001323402,
      "loss": 3.2147,
      "step": 338300
    },
    {
      "epoch": 0.22522893602701685,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001323382,
      "loss": 3.1799,
      "step": 338310
    },
    {
      "epoch": 0.22524757926317504,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001323362,
      "loss": 2.9921,
      "step": 338320
    },
    {
      "epoch": 0.2252662224993332,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001323342,
      "loss": 3.1099,
      "step": 338330
    },
    {
      "epoch": 0.2252848657354914,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013233220000000002,
      "loss": 3.1193,
      "step": 338340
    },
    {
      "epoch": 0.22530350897164958,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013233020000000002,
      "loss": 3.0141,
      "step": 338350
    },
    {
      "epoch": 0.22532215220780777,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001323282,
      "loss": 3.0637,
      "step": 338360
    },
    {
      "epoch": 0.22534079544396596,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001323262,
      "loss": 3.1835,
      "step": 338370
    },
    {
      "epoch": 0.22535943868012412,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001323242,
      "loss": 3.1415,
      "step": 338380
    },
    {
      "epoch": 0.2253780819162823,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001323222,
      "loss": 3.1861,
      "step": 338390
    },
    {
      "epoch": 0.2253967251524405,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001323202,
      "loss": 3.0209,
      "step": 338400
    },
    {
      "epoch": 0.22541536838859869,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013231820000000002,
      "loss": 3.022,
      "step": 338410
    },
    {
      "epoch": 0.22543401162475687,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001323162,
      "loss": 3.1578,
      "step": 338420
    },
    {
      "epoch": 0.22545265486091504,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001323142,
      "loss": 3.1762,
      "step": 338430
    },
    {
      "epoch": 0.22547129809707323,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001323122,
      "loss": 3.0593,
      "step": 338440
    },
    {
      "epoch": 0.22548994133323141,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001323102,
      "loss": 3.0527,
      "step": 338450
    },
    {
      "epoch": 0.2255085845693896,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013230820000000002,
      "loss": 3.2984,
      "step": 338460
    },
    {
      "epoch": 0.2255272278055478,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001323062,
      "loss": 3.1787,
      "step": 338470
    },
    {
      "epoch": 0.22554587104170598,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001323042,
      "loss": 3.1168,
      "step": 338480
    },
    {
      "epoch": 0.22556451427786414,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001323022,
      "loss": 3.0161,
      "step": 338490
    },
    {
      "epoch": 0.22558315751402233,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001323002,
      "loss": 3.19,
      "step": 338500
    },
    {
      "epoch": 0.22560180075018052,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013229820000000003,
      "loss": 3.1518,
      "step": 338510
    },
    {
      "epoch": 0.2256204439863387,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001322962,
      "loss": 3.0966,
      "step": 338520
    },
    {
      "epoch": 0.2256390872224969,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013229420000000002,
      "loss": 3.0536,
      "step": 338530
    },
    {
      "epoch": 0.22565773045865506,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001322922,
      "loss": 3.1746,
      "step": 338540
    },
    {
      "epoch": 0.22567637369481325,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001322902,
      "loss": 3.1145,
      "step": 338550
    },
    {
      "epoch": 0.22569501693097144,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001322882,
      "loss": 3.2044,
      "step": 338560
    },
    {
      "epoch": 0.22571366016712963,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001322862,
      "loss": 3.1651,
      "step": 338570
    },
    {
      "epoch": 0.22573230340328782,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013228420000000002,
      "loss": 3.1665,
      "step": 338580
    },
    {
      "epoch": 0.22575094663944598,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001322822,
      "loss": 3.0067,
      "step": 338590
    },
    {
      "epoch": 0.22576958987560417,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013228020000000001,
      "loss": 3.0533,
      "step": 338600
    },
    {
      "epoch": 0.22578823311176235,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013227819999999998,
      "loss": 3.1248,
      "step": 338610
    },
    {
      "epoch": 0.22580687634792054,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001322762,
      "loss": 3.2066,
      "step": 338620
    },
    {
      "epoch": 0.22582551958407873,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001322742,
      "loss": 3.1802,
      "step": 338630
    },
    {
      "epoch": 0.2258441628202369,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001322722,
      "loss": 3.1726,
      "step": 338640
    },
    {
      "epoch": 0.22586280605639508,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013227020000000002,
      "loss": 3.1226,
      "step": 338650
    },
    {
      "epoch": 0.22588144929255327,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001322682,
      "loss": 3.1089,
      "step": 338660
    },
    {
      "epoch": 0.22590009252871146,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001322662,
      "loss": 3.017,
      "step": 338670
    },
    {
      "epoch": 0.22591873576486965,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001322642,
      "loss": 3.0532,
      "step": 338680
    },
    {
      "epoch": 0.22593737900102784,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001322622,
      "loss": 3.2184,
      "step": 338690
    },
    {
      "epoch": 0.225956022237186,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013226020000000003,
      "loss": 3.1539,
      "step": 338700
    },
    {
      "epoch": 0.2259746654733442,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001322582,
      "loss": 3.0787,
      "step": 338710
    },
    {
      "epoch": 0.22599330870950238,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013225620000000002,
      "loss": 3.1955,
      "step": 338720
    },
    {
      "epoch": 0.22601195194566057,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001322542,
      "loss": 3.2268,
      "step": 338730
    },
    {
      "epoch": 0.22603059518181876,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001322522,
      "loss": 3.0187,
      "step": 338740
    },
    {
      "epoch": 0.22604923841797692,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001322502,
      "loss": 3.0982,
      "step": 338750
    },
    {
      "epoch": 0.2260678816541351,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001322482,
      "loss": 3.0377,
      "step": 338760
    },
    {
      "epoch": 0.2260865248902933,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013224620000000002,
      "loss": 3.1498,
      "step": 338770
    },
    {
      "epoch": 0.22610516812645148,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001322442,
      "loss": 2.9968,
      "step": 338780
    },
    {
      "epoch": 0.22612381136260967,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001322422,
      "loss": 3.1968,
      "step": 338790
    },
    {
      "epoch": 0.22614245459876783,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001322402,
      "loss": 3.1579,
      "step": 338800
    },
    {
      "epoch": 0.22616109783492602,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001322382,
      "loss": 3.1331,
      "step": 338810
    },
    {
      "epoch": 0.2261797410710842,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001322362,
      "loss": 3.0667,
      "step": 338820
    },
    {
      "epoch": 0.2261983843072424,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001322342,
      "loss": 3.0006,
      "step": 338830
    },
    {
      "epoch": 0.2262170275434006,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013223220000000002,
      "loss": 3.1251,
      "step": 338840
    },
    {
      "epoch": 0.22623567077955875,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001322302,
      "loss": 3.0667,
      "step": 338850
    },
    {
      "epoch": 0.22625431401571694,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001322282,
      "loss": 3.1198,
      "step": 338860
    },
    {
      "epoch": 0.22627295725187513,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001322262,
      "loss": 3.2564,
      "step": 338870
    },
    {
      "epoch": 0.22629160048803332,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001322242,
      "loss": 3.0914,
      "step": 338880
    },
    {
      "epoch": 0.2263102437241915,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001322222,
      "loss": 3.118,
      "step": 338890
    },
    {
      "epoch": 0.2263288869603497,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013222020000000002,
      "loss": 3.2652,
      "step": 338900
    },
    {
      "epoch": 0.22634753019650786,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013221820000000001,
      "loss": 2.9617,
      "step": 338910
    },
    {
      "epoch": 0.22636617343266605,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001322162,
      "loss": 3.236,
      "step": 338920
    },
    {
      "epoch": 0.22638481666882423,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001322142,
      "loss": 3.0664,
      "step": 338930
    },
    {
      "epoch": 0.22640345990498242,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001322122,
      "loss": 3.0321,
      "step": 338940
    },
    {
      "epoch": 0.2264221031411406,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013221020000000002,
      "loss": 3.1841,
      "step": 338950
    },
    {
      "epoch": 0.22644074637729877,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001322082,
      "loss": 3.0656,
      "step": 338960
    },
    {
      "epoch": 0.22645938961345696,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013220620000000002,
      "loss": 3.1355,
      "step": 338970
    },
    {
      "epoch": 0.22647803284961515,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001322042,
      "loss": 3.2344,
      "step": 338980
    },
    {
      "epoch": 0.22649667608577334,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001322022,
      "loss": 3.2358,
      "step": 338990
    },
    {
      "epoch": 0.22651531932193153,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001322002,
      "loss": 2.966,
      "step": 339000
    },
    {
      "epoch": 0.2265339625580897,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001321982,
      "loss": 3.0682,
      "step": 339010
    },
    {
      "epoch": 0.22655260579424788,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013219620000000002,
      "loss": 3.0844,
      "step": 339020
    },
    {
      "epoch": 0.22657124903040607,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001321942,
      "loss": 3.1795,
      "step": 339030
    },
    {
      "epoch": 0.22658989226656426,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001321922,
      "loss": 3.2229,
      "step": 339040
    },
    {
      "epoch": 0.22660853550272245,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001321902,
      "loss": 3.0263,
      "step": 339050
    },
    {
      "epoch": 0.2266271787388806,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001321882,
      "loss": 3.0567,
      "step": 339060
    },
    {
      "epoch": 0.2266458219750388,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013218620000000003,
      "loss": 3.2297,
      "step": 339070
    },
    {
      "epoch": 0.22666446521119699,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001321842,
      "loss": 3.0683,
      "step": 339080
    },
    {
      "epoch": 0.22668310844735517,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013218220000000002,
      "loss": 3.1939,
      "step": 339090
    },
    {
      "epoch": 0.22670175168351336,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013218019999999999,
      "loss": 3.1126,
      "step": 339100
    },
    {
      "epoch": 0.22672039491967153,
      "grad_norm": 0.75,
      "learning_rate": 0.0001321782,
      "loss": 3.0426,
      "step": 339110
    },
    {
      "epoch": 0.22673903815582971,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001321762,
      "loss": 3.2037,
      "step": 339120
    },
    {
      "epoch": 0.2267576813919879,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001321742,
      "loss": 3.0032,
      "step": 339130
    },
    {
      "epoch": 0.2267763246281461,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00013217220000000002,
      "loss": 3.2047,
      "step": 339140
    },
    {
      "epoch": 0.22679496786430428,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001321702,
      "loss": 3.1046,
      "step": 339150
    },
    {
      "epoch": 0.22681361110046247,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001321682,
      "loss": 3.0994,
      "step": 339160
    },
    {
      "epoch": 0.22683225433662063,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001321662,
      "loss": 3.0561,
      "step": 339170
    },
    {
      "epoch": 0.22685089757277882,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001321642,
      "loss": 3.1256,
      "step": 339180
    },
    {
      "epoch": 0.226869540808937,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013216220000000003,
      "loss": 3.0868,
      "step": 339190
    },
    {
      "epoch": 0.2268881840450952,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001321602,
      "loss": 3.1491,
      "step": 339200
    },
    {
      "epoch": 0.2269068272812534,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013215820000000002,
      "loss": 3.2572,
      "step": 339210
    },
    {
      "epoch": 0.22692547051741155,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001321562,
      "loss": 3.3369,
      "step": 339220
    },
    {
      "epoch": 0.22694411375356974,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001321542,
      "loss": 3.0564,
      "step": 339230
    },
    {
      "epoch": 0.22696275698972793,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001321522,
      "loss": 3.2043,
      "step": 339240
    },
    {
      "epoch": 0.22698140022588612,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001321502,
      "loss": 3.1278,
      "step": 339250
    },
    {
      "epoch": 0.2270000434620443,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013214820000000002,
      "loss": 3.1958,
      "step": 339260
    },
    {
      "epoch": 0.22701868669820247,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001321462,
      "loss": 3.1806,
      "step": 339270
    },
    {
      "epoch": 0.22703732993436065,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013214420000000001,
      "loss": 2.9872,
      "step": 339280
    },
    {
      "epoch": 0.22705597317051884,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001321422,
      "loss": 3.1638,
      "step": 339290
    },
    {
      "epoch": 0.22707461640667703,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001321402,
      "loss": 3.0065,
      "step": 339300
    },
    {
      "epoch": 0.22709325964283522,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001321382,
      "loss": 3.2342,
      "step": 339310
    },
    {
      "epoch": 0.22711190287899338,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001321362,
      "loss": 3.1871,
      "step": 339320
    },
    {
      "epoch": 0.22713054611515157,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013213420000000002,
      "loss": 3.1029,
      "step": 339330
    },
    {
      "epoch": 0.22714918935130976,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013213220000000002,
      "loss": 3.0989,
      "step": 339340
    },
    {
      "epoch": 0.22716783258746795,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001321302,
      "loss": 3.0115,
      "step": 339350
    },
    {
      "epoch": 0.22718647582362614,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001321282,
      "loss": 3.1238,
      "step": 339360
    },
    {
      "epoch": 0.22720511905978433,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001321262,
      "loss": 3.1633,
      "step": 339370
    },
    {
      "epoch": 0.2272237622959425,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001321242,
      "loss": 3.1517,
      "step": 339380
    },
    {
      "epoch": 0.22724240553210068,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001321222,
      "loss": 3.0866,
      "step": 339390
    },
    {
      "epoch": 0.22726104876825887,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013212020000000002,
      "loss": 3.0454,
      "step": 339400
    },
    {
      "epoch": 0.22727969200441706,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001321182,
      "loss": 3.0762,
      "step": 339410
    },
    {
      "epoch": 0.22729833524057524,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001321162,
      "loss": 3.1166,
      "step": 339420
    },
    {
      "epoch": 0.2273169784767334,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001321142,
      "loss": 3.1068,
      "step": 339430
    },
    {
      "epoch": 0.2273356217128916,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001321122,
      "loss": 3.0919,
      "step": 339440
    },
    {
      "epoch": 0.22735426494904978,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001321102,
      "loss": 3.2769,
      "step": 339450
    },
    {
      "epoch": 0.22737290818520797,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013210820000000002,
      "loss": 3.2003,
      "step": 339460
    },
    {
      "epoch": 0.22739155142136616,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001321062,
      "loss": 3.099,
      "step": 339470
    },
    {
      "epoch": 0.22741019465752432,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001321042,
      "loss": 3.0213,
      "step": 339480
    },
    {
      "epoch": 0.2274288378936825,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001321022,
      "loss": 3.1775,
      "step": 339490
    },
    {
      "epoch": 0.2274474811298407,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001321002,
      "loss": 3.0869,
      "step": 339500
    },
    {
      "epoch": 0.2274661243659989,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013209820000000002,
      "loss": 3.2091,
      "step": 339510
    },
    {
      "epoch": 0.22748476760215708,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001320962,
      "loss": 3.1689,
      "step": 339520
    },
    {
      "epoch": 0.22750341083831524,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001320942,
      "loss": 3.163,
      "step": 339530
    },
    {
      "epoch": 0.22752205407447343,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001320922,
      "loss": 3.1559,
      "step": 339540
    },
    {
      "epoch": 0.22754069731063162,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001320902,
      "loss": 3.2675,
      "step": 339550
    },
    {
      "epoch": 0.2275593405467898,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001320882,
      "loss": 3.119,
      "step": 339560
    },
    {
      "epoch": 0.227577983782948,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001320862,
      "loss": 3.1025,
      "step": 339570
    },
    {
      "epoch": 0.22759662701910618,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013208420000000002,
      "loss": 3.185,
      "step": 339580
    },
    {
      "epoch": 0.22761527025526435,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001320822,
      "loss": 3.1337,
      "step": 339590
    },
    {
      "epoch": 0.22763391349142253,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001320802,
      "loss": 3.0252,
      "step": 339600
    },
    {
      "epoch": 0.22765255672758072,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001320782,
      "loss": 3.1327,
      "step": 339610
    },
    {
      "epoch": 0.2276711999637389,
      "grad_norm": 0.75,
      "learning_rate": 0.0001320762,
      "loss": 3.1119,
      "step": 339620
    },
    {
      "epoch": 0.2276898431998971,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013207420000000002,
      "loss": 3.0507,
      "step": 339630
    },
    {
      "epoch": 0.22770848643605526,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001320722,
      "loss": 3.1881,
      "step": 339640
    },
    {
      "epoch": 0.22772712967221345,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013207020000000001,
      "loss": 3.1807,
      "step": 339650
    },
    {
      "epoch": 0.22774577290837164,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001320682,
      "loss": 3.2209,
      "step": 339660
    },
    {
      "epoch": 0.22776441614452983,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001320662,
      "loss": 3.0171,
      "step": 339670
    },
    {
      "epoch": 0.22778305938068802,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013206420000000003,
      "loss": 2.9854,
      "step": 339680
    },
    {
      "epoch": 0.22780170261684618,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001320622,
      "loss": 3.0516,
      "step": 339690
    },
    {
      "epoch": 0.22782034585300437,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013206020000000002,
      "loss": 3.2143,
      "step": 339700
    },
    {
      "epoch": 0.22783898908916256,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001320582,
      "loss": 3.1157,
      "step": 339710
    },
    {
      "epoch": 0.22785763232532075,
      "grad_norm": 0.75,
      "learning_rate": 0.0001320562,
      "loss": 3.0919,
      "step": 339720
    },
    {
      "epoch": 0.22787627556147894,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001320542,
      "loss": 3.0366,
      "step": 339730
    },
    {
      "epoch": 0.2278949187976371,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001320522,
      "loss": 3.1199,
      "step": 339740
    },
    {
      "epoch": 0.22791356203379529,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013205020000000003,
      "loss": 3.0692,
      "step": 339750
    },
    {
      "epoch": 0.22793220526995348,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001320482,
      "loss": 3.1982,
      "step": 339760
    },
    {
      "epoch": 0.22795084850611166,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013204620000000002,
      "loss": 3.1036,
      "step": 339770
    },
    {
      "epoch": 0.22796949174226985,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013204419999999999,
      "loss": 3.0422,
      "step": 339780
    },
    {
      "epoch": 0.22798813497842801,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001320422,
      "loss": 3.1716,
      "step": 339790
    },
    {
      "epoch": 0.2280067782145862,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001320402,
      "loss": 3.1475,
      "step": 339800
    },
    {
      "epoch": 0.2280254214507444,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001320382,
      "loss": 3.2228,
      "step": 339810
    },
    {
      "epoch": 0.22804406468690258,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013203620000000002,
      "loss": 3.1331,
      "step": 339820
    },
    {
      "epoch": 0.22806270792306077,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001320342,
      "loss": 3.3092,
      "step": 339830
    },
    {
      "epoch": 0.22808135115921896,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001320322,
      "loss": 3.1558,
      "step": 339840
    },
    {
      "epoch": 0.22809999439537712,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001320302,
      "loss": 3.2611,
      "step": 339850
    },
    {
      "epoch": 0.2281186376315353,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001320282,
      "loss": 3.054,
      "step": 339860
    },
    {
      "epoch": 0.2281372808676935,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001320262,
      "loss": 3.0932,
      "step": 339870
    },
    {
      "epoch": 0.2281559241038517,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001320242,
      "loss": 3.0923,
      "step": 339880
    },
    {
      "epoch": 0.22817456734000988,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013202220000000002,
      "loss": 3.1499,
      "step": 339890
    },
    {
      "epoch": 0.22819321057616804,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013202020000000001,
      "loss": 2.9572,
      "step": 339900
    },
    {
      "epoch": 0.22821185381232623,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001320182,
      "loss": 3.1796,
      "step": 339910
    },
    {
      "epoch": 0.22823049704848442,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001320162,
      "loss": 3.1791,
      "step": 339920
    },
    {
      "epoch": 0.2282491402846426,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001320142,
      "loss": 2.9684,
      "step": 339930
    },
    {
      "epoch": 0.2282677835208008,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001320122,
      "loss": 3.1069,
      "step": 339940
    },
    {
      "epoch": 0.22828642675695895,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013201020000000002,
      "loss": 3.1596,
      "step": 339950
    },
    {
      "epoch": 0.22830506999311714,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013200820000000001,
      "loss": 3.0633,
      "step": 339960
    },
    {
      "epoch": 0.22832371322927533,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001320062,
      "loss": 3.0816,
      "step": 339970
    },
    {
      "epoch": 0.22834235646543352,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001320042,
      "loss": 3.2401,
      "step": 339980
    },
    {
      "epoch": 0.2283609997015917,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001320022,
      "loss": 3.0195,
      "step": 339990
    },
    {
      "epoch": 0.22837964293774987,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001320002,
      "loss": 3.0564,
      "step": 340000
    },
    {
      "epoch": 0.22839828617390806,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001319982,
      "loss": 3.2483,
      "step": 340010
    },
    {
      "epoch": 0.22841692941006625,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013199620000000002,
      "loss": 3.1078,
      "step": 340020
    },
    {
      "epoch": 0.22843557264622444,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001319942,
      "loss": 3.0674,
      "step": 340030
    },
    {
      "epoch": 0.22845421588238263,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001319922,
      "loss": 3.0155,
      "step": 340040
    },
    {
      "epoch": 0.22847285911854082,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001319902,
      "loss": 3.004,
      "step": 340050
    },
    {
      "epoch": 0.22849150235469898,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001319882,
      "loss": 3.1003,
      "step": 340060
    },
    {
      "epoch": 0.22851014559085717,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013198620000000002,
      "loss": 3.0973,
      "step": 340070
    },
    {
      "epoch": 0.22852878882701536,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001319842,
      "loss": 3.214,
      "step": 340080
    },
    {
      "epoch": 0.22854743206317354,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001319822,
      "loss": 2.9837,
      "step": 340090
    },
    {
      "epoch": 0.22856607529933173,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001319802,
      "loss": 3.2359,
      "step": 340100
    },
    {
      "epoch": 0.2285847185354899,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001319782,
      "loss": 3.1472,
      "step": 340110
    },
    {
      "epoch": 0.22860336177164808,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013197620000000003,
      "loss": 3.1668,
      "step": 340120
    },
    {
      "epoch": 0.22862200500780627,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001319742,
      "loss": 3.1755,
      "step": 340130
    },
    {
      "epoch": 0.22864064824396446,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013197220000000002,
      "loss": 3.1131,
      "step": 340140
    },
    {
      "epoch": 0.22865929148012265,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001319702,
      "loss": 3.1166,
      "step": 340150
    },
    {
      "epoch": 0.2286779347162808,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001319682,
      "loss": 3.1374,
      "step": 340160
    },
    {
      "epoch": 0.228696577952439,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001319662,
      "loss": 3.1136,
      "step": 340170
    },
    {
      "epoch": 0.2287152211885972,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001319642,
      "loss": 3.1452,
      "step": 340180
    },
    {
      "epoch": 0.22873386442475538,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013196220000000002,
      "loss": 3.1527,
      "step": 340190
    },
    {
      "epoch": 0.22875250766091357,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001319602,
      "loss": 3.1409,
      "step": 340200
    },
    {
      "epoch": 0.22877115089707173,
      "grad_norm": 0.625,
      "learning_rate": 0.0001319582,
      "loss": 3.2079,
      "step": 340210
    },
    {
      "epoch": 0.22878979413322992,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001319562,
      "loss": 3.2284,
      "step": 340220
    },
    {
      "epoch": 0.2288084373693881,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001319542,
      "loss": 3.1775,
      "step": 340230
    },
    {
      "epoch": 0.2288270806055463,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013195220000000003,
      "loss": 3.117,
      "step": 340240
    },
    {
      "epoch": 0.22884572384170448,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001319502,
      "loss": 3.1359,
      "step": 340250
    },
    {
      "epoch": 0.22886436707786267,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013194820000000002,
      "loss": 3.2963,
      "step": 340260
    },
    {
      "epoch": 0.22888301031402083,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001319462,
      "loss": 3.0857,
      "step": 340270
    },
    {
      "epoch": 0.22890165355017902,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001319442,
      "loss": 3.1539,
      "step": 340280
    },
    {
      "epoch": 0.2289202967863372,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001319422,
      "loss": 3.1126,
      "step": 340290
    },
    {
      "epoch": 0.2289389400224954,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001319402,
      "loss": 3.0954,
      "step": 340300
    },
    {
      "epoch": 0.2289575832586536,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013193820000000002,
      "loss": 3.1125,
      "step": 340310
    },
    {
      "epoch": 0.22897622649481175,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001319362,
      "loss": 3.1033,
      "step": 340320
    },
    {
      "epoch": 0.22899486973096994,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013193420000000001,
      "loss": 3.1931,
      "step": 340330
    },
    {
      "epoch": 0.22901351296712813,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001319322,
      "loss": 3.1251,
      "step": 340340
    },
    {
      "epoch": 0.22903215620328632,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001319302,
      "loss": 3.2269,
      "step": 340350
    },
    {
      "epoch": 0.2290507994394445,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001319282,
      "loss": 3.116,
      "step": 340360
    },
    {
      "epoch": 0.22906944267560267,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001319262,
      "loss": 3.0249,
      "step": 340370
    },
    {
      "epoch": 0.22908808591176086,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013192420000000002,
      "loss": 3.1599,
      "step": 340380
    },
    {
      "epoch": 0.22910672914791905,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001319222,
      "loss": 3.1294,
      "step": 340390
    },
    {
      "epoch": 0.22912537238407724,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001319202,
      "loss": 3.2601,
      "step": 340400
    },
    {
      "epoch": 0.22914401562023543,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001319182,
      "loss": 3.1648,
      "step": 340410
    },
    {
      "epoch": 0.2291626588563936,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001319162,
      "loss": 3.0633,
      "step": 340420
    },
    {
      "epoch": 0.22918130209255178,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001319142,
      "loss": 3.0573,
      "step": 340430
    },
    {
      "epoch": 0.22919994532870996,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001319122,
      "loss": 3.2125,
      "step": 340440
    },
    {
      "epoch": 0.22921858856486815,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013191020000000002,
      "loss": 3.0766,
      "step": 340450
    },
    {
      "epoch": 0.22923723180102634,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001319082,
      "loss": 3.1386,
      "step": 340460
    },
    {
      "epoch": 0.2292558750371845,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001319062,
      "loss": 3.1713,
      "step": 340470
    },
    {
      "epoch": 0.2292745182733427,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001319042,
      "loss": 3.1787,
      "step": 340480
    },
    {
      "epoch": 0.22929316150950088,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001319022,
      "loss": 3.1691,
      "step": 340490
    },
    {
      "epoch": 0.22931180474565907,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001319002,
      "loss": 3.1617,
      "step": 340500
    },
    {
      "epoch": 0.22933044798181726,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013189820000000002,
      "loss": 3.1501,
      "step": 340510
    },
    {
      "epoch": 0.22934909121797545,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001318962,
      "loss": 3.1942,
      "step": 340520
    },
    {
      "epoch": 0.2293677344541336,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001318942,
      "loss": 3.2519,
      "step": 340530
    },
    {
      "epoch": 0.2293863776902918,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001318922,
      "loss": 3.0742,
      "step": 340540
    },
    {
      "epoch": 0.22940502092645,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001318902,
      "loss": 3.1182,
      "step": 340550
    },
    {
      "epoch": 0.22942366416260818,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013188820000000002,
      "loss": 3.2654,
      "step": 340560
    },
    {
      "epoch": 0.22944230739876637,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001318862,
      "loss": 3.0938,
      "step": 340570
    },
    {
      "epoch": 0.22946095063492453,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013188420000000001,
      "loss": 3.1325,
      "step": 340580
    },
    {
      "epoch": 0.22947959387108272,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001318822,
      "loss": 3.1084,
      "step": 340590
    },
    {
      "epoch": 0.2294982371072409,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001318802,
      "loss": 3.0975,
      "step": 340600
    },
    {
      "epoch": 0.2295168803433991,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001318782,
      "loss": 3.1742,
      "step": 340610
    },
    {
      "epoch": 0.22953552357955728,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001318762,
      "loss": 3.0869,
      "step": 340620
    },
    {
      "epoch": 0.22955416681571544,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013187420000000002,
      "loss": 3.1804,
      "step": 340630
    },
    {
      "epoch": 0.22957281005187363,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013187220000000001,
      "loss": 3.0709,
      "step": 340640
    },
    {
      "epoch": 0.22959145328803182,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001318702,
      "loss": 3.2629,
      "step": 340650
    },
    {
      "epoch": 0.22961009652419,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001318682,
      "loss": 3.1092,
      "step": 340660
    },
    {
      "epoch": 0.2296287397603482,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001318662,
      "loss": 3.1734,
      "step": 340670
    },
    {
      "epoch": 0.22964738299650636,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013186420000000002,
      "loss": 3.0668,
      "step": 340680
    },
    {
      "epoch": 0.22966602623266455,
      "grad_norm": 0.75,
      "learning_rate": 0.0001318622,
      "loss": 3.1085,
      "step": 340690
    },
    {
      "epoch": 0.22968466946882274,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013186020000000002,
      "loss": 3.2128,
      "step": 340700
    },
    {
      "epoch": 0.22970331270498093,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001318582,
      "loss": 3.0791,
      "step": 340710
    },
    {
      "epoch": 0.22972195594113912,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001318562,
      "loss": 3.1373,
      "step": 340720
    },
    {
      "epoch": 0.2297405991772973,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013185420000000003,
      "loss": 3.0489,
      "step": 340730
    },
    {
      "epoch": 0.22975924241345547,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001318522,
      "loss": 3.2261,
      "step": 340740
    },
    {
      "epoch": 0.22977788564961366,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013185020000000002,
      "loss": 3.0643,
      "step": 340750
    },
    {
      "epoch": 0.22979652888577184,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001318482,
      "loss": 3.1978,
      "step": 340760
    },
    {
      "epoch": 0.22981517212193003,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001318462,
      "loss": 3.2175,
      "step": 340770
    },
    {
      "epoch": 0.22983381535808822,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001318442,
      "loss": 3.1551,
      "step": 340780
    },
    {
      "epoch": 0.22985245859424638,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001318422,
      "loss": 3.1661,
      "step": 340790
    },
    {
      "epoch": 0.22987110183040457,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013184020000000003,
      "loss": 3.0156,
      "step": 340800
    },
    {
      "epoch": 0.22988974506656276,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001318382,
      "loss": 3.2343,
      "step": 340810
    },
    {
      "epoch": 0.22990838830272095,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013183620000000002,
      "loss": 3.1482,
      "step": 340820
    },
    {
      "epoch": 0.22992703153887914,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013183419999999999,
      "loss": 3.0783,
      "step": 340830
    },
    {
      "epoch": 0.2299456747750373,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001318322,
      "loss": 3.0944,
      "step": 340840
    },
    {
      "epoch": 0.2299643180111955,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001318302,
      "loss": 3.0325,
      "step": 340850
    },
    {
      "epoch": 0.22998296124735368,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001318282,
      "loss": 3.1671,
      "step": 340860
    },
    {
      "epoch": 0.23000160448351187,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013182620000000002,
      "loss": 3.1878,
      "step": 340870
    },
    {
      "epoch": 0.23002024771967006,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001318242,
      "loss": 3.1174,
      "step": 340880
    },
    {
      "epoch": 0.23003889095582822,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001318222,
      "loss": 3.1882,
      "step": 340890
    },
    {
      "epoch": 0.2300575341919864,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001318202,
      "loss": 3.0304,
      "step": 340900
    },
    {
      "epoch": 0.2300761774281446,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001318182,
      "loss": 3.0138,
      "step": 340910
    },
    {
      "epoch": 0.23009482066430278,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001318162,
      "loss": 3.1056,
      "step": 340920
    },
    {
      "epoch": 0.23011346390046097,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001318142,
      "loss": 3.1082,
      "step": 340930
    },
    {
      "epoch": 0.23013210713661916,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00013181220000000002,
      "loss": 3.306,
      "step": 340940
    },
    {
      "epoch": 0.23015075037277732,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013181020000000001,
      "loss": 3.1426,
      "step": 340950
    },
    {
      "epoch": 0.2301693936089355,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001318082,
      "loss": 3.0657,
      "step": 340960
    },
    {
      "epoch": 0.2301880368450937,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001318062,
      "loss": 3.0738,
      "step": 340970
    },
    {
      "epoch": 0.2302066800812519,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001318042,
      "loss": 3.0075,
      "step": 340980
    },
    {
      "epoch": 0.23022532331741008,
      "grad_norm": 0.75,
      "learning_rate": 0.0001318022,
      "loss": 3.0487,
      "step": 340990
    },
    {
      "epoch": 0.23024396655356824,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001318002,
      "loss": 3.1682,
      "step": 341000
    },
    {
      "epoch": 0.23026260978972643,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013179820000000001,
      "loss": 3.0575,
      "step": 341010
    },
    {
      "epoch": 0.23028125302588462,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001317962,
      "loss": 3.121,
      "step": 341020
    },
    {
      "epoch": 0.2302998962620428,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001317942,
      "loss": 3.1395,
      "step": 341030
    },
    {
      "epoch": 0.230318539498201,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001317922,
      "loss": 3.1358,
      "step": 341040
    },
    {
      "epoch": 0.23033718273435916,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001317902,
      "loss": 3.1972,
      "step": 341050
    },
    {
      "epoch": 0.23035582597051735,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001317882,
      "loss": 3.0175,
      "step": 341060
    },
    {
      "epoch": 0.23037446920667554,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013178620000000002,
      "loss": 3.0311,
      "step": 341070
    },
    {
      "epoch": 0.23039311244283373,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001317842,
      "loss": 3.1311,
      "step": 341080
    },
    {
      "epoch": 0.23041175567899191,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001317822,
      "loss": 2.9875,
      "step": 341090
    },
    {
      "epoch": 0.23043039891515008,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001317802,
      "loss": 3.1376,
      "step": 341100
    },
    {
      "epoch": 0.23044904215130826,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001317782,
      "loss": 3.1156,
      "step": 341110
    },
    {
      "epoch": 0.23046768538746645,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013177620000000002,
      "loss": 3.1247,
      "step": 341120
    },
    {
      "epoch": 0.23048632862362464,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00013177420000000002,
      "loss": 3.1333,
      "step": 341130
    },
    {
      "epoch": 0.23050497185978283,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001317722,
      "loss": 3.0134,
      "step": 341140
    },
    {
      "epoch": 0.230523615095941,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001317702,
      "loss": 3.1438,
      "step": 341150
    },
    {
      "epoch": 0.23054225833209918,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001317682,
      "loss": 3.1735,
      "step": 341160
    },
    {
      "epoch": 0.23056090156825737,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013176620000000003,
      "loss": 3.1017,
      "step": 341170
    },
    {
      "epoch": 0.23057954480441556,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001317642,
      "loss": 3.1395,
      "step": 341180
    },
    {
      "epoch": 0.23059818804057375,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013176220000000002,
      "loss": 3.2219,
      "step": 341190
    },
    {
      "epoch": 0.23061683127673194,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001317602,
      "loss": 3.1302,
      "step": 341200
    },
    {
      "epoch": 0.2306354745128901,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001317582,
      "loss": 3.1044,
      "step": 341210
    },
    {
      "epoch": 0.2306541177490483,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001317562,
      "loss": 3.1207,
      "step": 341220
    },
    {
      "epoch": 0.23067276098520648,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001317542,
      "loss": 3.1647,
      "step": 341230
    },
    {
      "epoch": 0.23069140422136467,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013175220000000002,
      "loss": 3.1243,
      "step": 341240
    },
    {
      "epoch": 0.23071004745752285,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001317502,
      "loss": 3.0302,
      "step": 341250
    },
    {
      "epoch": 0.23072869069368102,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00013174820000000001,
      "loss": 3.0529,
      "step": 341260
    },
    {
      "epoch": 0.2307473339298392,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001317462,
      "loss": 3.1806,
      "step": 341270
    },
    {
      "epoch": 0.2307659771659974,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001317442,
      "loss": 3.0711,
      "step": 341280
    },
    {
      "epoch": 0.23078462040215558,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013174220000000003,
      "loss": 3.1317,
      "step": 341290
    },
    {
      "epoch": 0.23080326363831377,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001317402,
      "loss": 3.0715,
      "step": 341300
    },
    {
      "epoch": 0.23082190687447193,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013173820000000002,
      "loss": 3.102,
      "step": 341310
    },
    {
      "epoch": 0.23084055011063012,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001317362,
      "loss": 3.0201,
      "step": 341320
    },
    {
      "epoch": 0.2308591933467883,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001317342,
      "loss": 2.9829,
      "step": 341330
    },
    {
      "epoch": 0.2308778365829465,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001317322,
      "loss": 3.1991,
      "step": 341340
    },
    {
      "epoch": 0.2308964798191047,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001317302,
      "loss": 3.0069,
      "step": 341350
    },
    {
      "epoch": 0.23091512305526285,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013172820000000002,
      "loss": 3.1531,
      "step": 341360
    },
    {
      "epoch": 0.23093376629142104,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001317262,
      "loss": 3.0442,
      "step": 341370
    },
    {
      "epoch": 0.23095240952757923,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013172420000000002,
      "loss": 3.1833,
      "step": 341380
    },
    {
      "epoch": 0.23097105276373742,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013172219999999998,
      "loss": 3.0576,
      "step": 341390
    },
    {
      "epoch": 0.2309896959998956,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001317202,
      "loss": 3.0583,
      "step": 341400
    },
    {
      "epoch": 0.2310083392360538,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001317182,
      "loss": 3.0967,
      "step": 341410
    },
    {
      "epoch": 0.23102698247221196,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001317162,
      "loss": 3.0543,
      "step": 341420
    },
    {
      "epoch": 0.23104562570837014,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013171420000000002,
      "loss": 3.2178,
      "step": 341430
    },
    {
      "epoch": 0.23106426894452833,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001317122,
      "loss": 3.0754,
      "step": 341440
    },
    {
      "epoch": 0.23108291218068652,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001317102,
      "loss": 3.2217,
      "step": 341450
    },
    {
      "epoch": 0.2311015554168447,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001317082,
      "loss": 3.1805,
      "step": 341460
    },
    {
      "epoch": 0.23112019865300287,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001317062,
      "loss": 3.2022,
      "step": 341470
    },
    {
      "epoch": 0.23113884188916106,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001317042,
      "loss": 3.011,
      "step": 341480
    },
    {
      "epoch": 0.23115748512531925,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001317022,
      "loss": 3.2325,
      "step": 341490
    },
    {
      "epoch": 0.23117612836147744,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013170020000000002,
      "loss": 3.1413,
      "step": 341500
    },
    {
      "epoch": 0.23119477159763563,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001316982,
      "loss": 3.1167,
      "step": 341510
    },
    {
      "epoch": 0.2312134148337938,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001316962,
      "loss": 3.2237,
      "step": 341520
    },
    {
      "epoch": 0.23123205806995198,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001316942,
      "loss": 3.0378,
      "step": 341530
    },
    {
      "epoch": 0.23125070130611017,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001316922,
      "loss": 3.0759,
      "step": 341540
    },
    {
      "epoch": 0.23126934454226836,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001316902,
      "loss": 3.0327,
      "step": 341550
    },
    {
      "epoch": 0.23128798777842655,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013168820000000002,
      "loss": 3.0038,
      "step": 341560
    },
    {
      "epoch": 0.2313066310145847,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001316862,
      "loss": 3.22,
      "step": 341570
    },
    {
      "epoch": 0.2313252742507429,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001316842,
      "loss": 3.1343,
      "step": 341580
    },
    {
      "epoch": 0.23134391748690109,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001316822,
      "loss": 3.2428,
      "step": 341590
    },
    {
      "epoch": 0.23136256072305927,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001316802,
      "loss": 3.105,
      "step": 341600
    },
    {
      "epoch": 0.23138120395921746,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001316782,
      "loss": 3.1208,
      "step": 341610
    },
    {
      "epoch": 0.23139984719537565,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013167620000000002,
      "loss": 3.0589,
      "step": 341620
    },
    {
      "epoch": 0.2314184904315338,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013167420000000001,
      "loss": 3.0181,
      "step": 341630
    },
    {
      "epoch": 0.231437133667692,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001316722,
      "loss": 3.1637,
      "step": 341640
    },
    {
      "epoch": 0.2314557769038502,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001316702,
      "loss": 3.2196,
      "step": 341650
    },
    {
      "epoch": 0.23147442014000838,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001316682,
      "loss": 3.1285,
      "step": 341660
    },
    {
      "epoch": 0.23149306337616657,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001316662,
      "loss": 3.1328,
      "step": 341670
    },
    {
      "epoch": 0.23151170661232473,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013166420000000002,
      "loss": 3.1158,
      "step": 341680
    },
    {
      "epoch": 0.23153034984848292,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013166220000000001,
      "loss": 3.1425,
      "step": 341690
    },
    {
      "epoch": 0.2315489930846411,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001316602,
      "loss": 3.1081,
      "step": 341700
    },
    {
      "epoch": 0.2315676363207993,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001316582,
      "loss": 3.2308,
      "step": 341710
    },
    {
      "epoch": 0.2315862795569575,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001316562,
      "loss": 2.9905,
      "step": 341720
    },
    {
      "epoch": 0.23160492279311565,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013165420000000002,
      "loss": 3.055,
      "step": 341730
    },
    {
      "epoch": 0.23162356602927384,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001316522,
      "loss": 3.1932,
      "step": 341740
    },
    {
      "epoch": 0.23164220926543203,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00013165020000000002,
      "loss": 3.05,
      "step": 341750
    },
    {
      "epoch": 0.23166085250159021,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001316482,
      "loss": 3.0009,
      "step": 341760
    },
    {
      "epoch": 0.2316794957377484,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001316462,
      "loss": 3.0827,
      "step": 341770
    },
    {
      "epoch": 0.23169813897390656,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001316442,
      "loss": 3.0698,
      "step": 341780
    },
    {
      "epoch": 0.23171678221006475,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001316422,
      "loss": 2.9404,
      "step": 341790
    },
    {
      "epoch": 0.23173542544622294,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013164020000000002,
      "loss": 3.2554,
      "step": 341800
    },
    {
      "epoch": 0.23175406868238113,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001316382,
      "loss": 3.0312,
      "step": 341810
    },
    {
      "epoch": 0.23177271191853932,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001316362,
      "loss": 3.1151,
      "step": 341820
    },
    {
      "epoch": 0.2317913551546975,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001316342,
      "loss": 3.2223,
      "step": 341830
    },
    {
      "epoch": 0.23180999839085567,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001316322,
      "loss": 3.0704,
      "step": 341840
    },
    {
      "epoch": 0.23182864162701386,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013163020000000003,
      "loss": 3.1569,
      "step": 341850
    },
    {
      "epoch": 0.23184728486317205,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001316282,
      "loss": 3.1322,
      "step": 341860
    },
    {
      "epoch": 0.23186592809933024,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013162620000000002,
      "loss": 3.1575,
      "step": 341870
    },
    {
      "epoch": 0.23188457133548843,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013162419999999999,
      "loss": 3.2384,
      "step": 341880
    },
    {
      "epoch": 0.2319032145716466,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001316222,
      "loss": 3.2343,
      "step": 341890
    },
    {
      "epoch": 0.23192185780780478,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001316202,
      "loss": 3.214,
      "step": 341900
    },
    {
      "epoch": 0.23194050104396297,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001316182,
      "loss": 3.1594,
      "step": 341910
    },
    {
      "epoch": 0.23195914428012115,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013161620000000002,
      "loss": 3.0072,
      "step": 341920
    },
    {
      "epoch": 0.23197778751627934,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001316142,
      "loss": 3.26,
      "step": 341930
    },
    {
      "epoch": 0.2319964307524375,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013161220000000001,
      "loss": 3.2293,
      "step": 341940
    },
    {
      "epoch": 0.2320150739885957,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001316102,
      "loss": 3.1234,
      "step": 341950
    },
    {
      "epoch": 0.23203371722475388,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001316082,
      "loss": 3.1285,
      "step": 341960
    },
    {
      "epoch": 0.23205236046091207,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001316062,
      "loss": 3.1216,
      "step": 341970
    },
    {
      "epoch": 0.23207100369707026,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001316042,
      "loss": 3.1411,
      "step": 341980
    },
    {
      "epoch": 0.23208964693322842,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013160220000000002,
      "loss": 3.0454,
      "step": 341990
    },
    {
      "epoch": 0.2321082901693866,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001316002,
      "loss": 3.1114,
      "step": 342000
    },
    {
      "epoch": 0.2321269334055448,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001315982,
      "loss": 3.0877,
      "step": 342010
    },
    {
      "epoch": 0.232145576641703,
      "grad_norm": 0.625,
      "learning_rate": 0.0001315962,
      "loss": 3.0823,
      "step": 342020
    },
    {
      "epoch": 0.23216421987786118,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001315942,
      "loss": 3.0277,
      "step": 342030
    },
    {
      "epoch": 0.23218286311401934,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001315922,
      "loss": 3.0653,
      "step": 342040
    },
    {
      "epoch": 0.23220150635017753,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001315902,
      "loss": 3.2273,
      "step": 342050
    },
    {
      "epoch": 0.23222014958633572,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013158820000000002,
      "loss": 3.043,
      "step": 342060
    },
    {
      "epoch": 0.2322387928224939,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001315862,
      "loss": 3.0757,
      "step": 342070
    },
    {
      "epoch": 0.2322574360586521,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001315842,
      "loss": 3.0931,
      "step": 342080
    },
    {
      "epoch": 0.23227607929481028,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001315822,
      "loss": 3.1304,
      "step": 342090
    },
    {
      "epoch": 0.23229472253096844,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001315802,
      "loss": 3.0795,
      "step": 342100
    },
    {
      "epoch": 0.23231336576712663,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013157820000000002,
      "loss": 3.0527,
      "step": 342110
    },
    {
      "epoch": 0.23233200900328482,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013157620000000002,
      "loss": 3.1238,
      "step": 342120
    },
    {
      "epoch": 0.232350652239443,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001315742,
      "loss": 3.1341,
      "step": 342130
    },
    {
      "epoch": 0.2323692954756012,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001315722,
      "loss": 3.1953,
      "step": 342140
    },
    {
      "epoch": 0.23238793871175936,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001315702,
      "loss": 2.9811,
      "step": 342150
    },
    {
      "epoch": 0.23240658194791755,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001315682,
      "loss": 3.0893,
      "step": 342160
    },
    {
      "epoch": 0.23242522518407574,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013156620000000002,
      "loss": 3.0751,
      "step": 342170
    },
    {
      "epoch": 0.23244386842023393,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013156420000000002,
      "loss": 3.0881,
      "step": 342180
    },
    {
      "epoch": 0.23246251165639212,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001315622,
      "loss": 3.1603,
      "step": 342190
    },
    {
      "epoch": 0.23248115489255028,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001315602,
      "loss": 3.0827,
      "step": 342200
    },
    {
      "epoch": 0.23249979812870847,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001315582,
      "loss": 3.0077,
      "step": 342210
    },
    {
      "epoch": 0.23251844136486666,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001315562,
      "loss": 3.1656,
      "step": 342220
    },
    {
      "epoch": 0.23253708460102485,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001315542,
      "loss": 3.0813,
      "step": 342230
    },
    {
      "epoch": 0.23255572783718303,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013155220000000002,
      "loss": 3.2278,
      "step": 342240
    },
    {
      "epoch": 0.2325743710733412,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001315502,
      "loss": 3.3137,
      "step": 342250
    },
    {
      "epoch": 0.23259301430949939,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001315482,
      "loss": 3.201,
      "step": 342260
    },
    {
      "epoch": 0.23261165754565757,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001315462,
      "loss": 3.0338,
      "step": 342270
    },
    {
      "epoch": 0.23263030078181576,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001315442,
      "loss": 3.1981,
      "step": 342280
    },
    {
      "epoch": 0.23264894401797395,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013154220000000002,
      "loss": 3.1277,
      "step": 342290
    },
    {
      "epoch": 0.23266758725413214,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001315402,
      "loss": 2.9855,
      "step": 342300
    },
    {
      "epoch": 0.2326862304902903,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013153820000000001,
      "loss": 3.1004,
      "step": 342310
    },
    {
      "epoch": 0.2327048737264485,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001315362,
      "loss": 3.1608,
      "step": 342320
    },
    {
      "epoch": 0.23272351696260668,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001315342,
      "loss": 3.235,
      "step": 342330
    },
    {
      "epoch": 0.23274216019876487,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013153220000000003,
      "loss": 3.0124,
      "step": 342340
    },
    {
      "epoch": 0.23276080343492306,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001315302,
      "loss": 3.1719,
      "step": 342350
    },
    {
      "epoch": 0.23277944667108122,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013152820000000002,
      "loss": 3.0916,
      "step": 342360
    },
    {
      "epoch": 0.2327980899072394,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001315262,
      "loss": 3.1437,
      "step": 342370
    },
    {
      "epoch": 0.2328167331433976,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001315242,
      "loss": 3.1437,
      "step": 342380
    },
    {
      "epoch": 0.2328353763795558,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001315222,
      "loss": 3.0541,
      "step": 342390
    },
    {
      "epoch": 0.23285401961571398,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001315202,
      "loss": 3.1793,
      "step": 342400
    },
    {
      "epoch": 0.23287266285187214,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013151820000000002,
      "loss": 3.1096,
      "step": 342410
    },
    {
      "epoch": 0.23289130608803033,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001315162,
      "loss": 3.0828,
      "step": 342420
    },
    {
      "epoch": 0.23290994932418851,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00013151420000000002,
      "loss": 3.0507,
      "step": 342430
    },
    {
      "epoch": 0.2329285925603467,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013151219999999998,
      "loss": 3.1419,
      "step": 342440
    },
    {
      "epoch": 0.2329472357965049,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001315102,
      "loss": 3.2061,
      "step": 342450
    },
    {
      "epoch": 0.23296587903266305,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001315082,
      "loss": 3.1672,
      "step": 342460
    },
    {
      "epoch": 0.23298452226882124,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001315062,
      "loss": 3.1495,
      "step": 342470
    },
    {
      "epoch": 0.23300316550497943,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013150420000000002,
      "loss": 3.1318,
      "step": 342480
    },
    {
      "epoch": 0.23302180874113762,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001315022,
      "loss": 3.1057,
      "step": 342490
    },
    {
      "epoch": 0.2330404519772958,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001315002,
      "loss": 3.0736,
      "step": 342500
    },
    {
      "epoch": 0.233059095213454,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001314982,
      "loss": 2.9964,
      "step": 342510
    },
    {
      "epoch": 0.23307773844961216,
      "grad_norm": 0.75,
      "learning_rate": 0.0001314962,
      "loss": 3.026,
      "step": 342520
    },
    {
      "epoch": 0.23309638168577035,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001314942,
      "loss": 3.1508,
      "step": 342530
    },
    {
      "epoch": 0.23311502492192854,
      "grad_norm": 0.75,
      "learning_rate": 0.0001314922,
      "loss": 3.0899,
      "step": 342540
    },
    {
      "epoch": 0.23313366815808673,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013149020000000002,
      "loss": 3.0783,
      "step": 342550
    },
    {
      "epoch": 0.23315231139424492,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001314882,
      "loss": 3.12,
      "step": 342560
    },
    {
      "epoch": 0.23317095463040308,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001314862,
      "loss": 3.0716,
      "step": 342570
    },
    {
      "epoch": 0.23318959786656127,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001314842,
      "loss": 3.081,
      "step": 342580
    },
    {
      "epoch": 0.23320824110271945,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001314822,
      "loss": 3.0979,
      "step": 342590
    },
    {
      "epoch": 0.23322688433887764,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013148020000000002,
      "loss": 3.0754,
      "step": 342600
    },
    {
      "epoch": 0.23324552757503583,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001314782,
      "loss": 3.1286,
      "step": 342610
    },
    {
      "epoch": 0.233264170811194,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001314762,
      "loss": 3.0171,
      "step": 342620
    },
    {
      "epoch": 0.23328281404735218,
      "grad_norm": 0.75,
      "learning_rate": 0.0001314742,
      "loss": 2.9934,
      "step": 342630
    },
    {
      "epoch": 0.23330145728351037,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001314722,
      "loss": 3.1627,
      "step": 342640
    },
    {
      "epoch": 0.23332010051966856,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001314702,
      "loss": 2.9891,
      "step": 342650
    },
    {
      "epoch": 0.23333874375582675,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001314682,
      "loss": 3.0871,
      "step": 342660
    },
    {
      "epoch": 0.2333573869919849,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013146620000000002,
      "loss": 3.0432,
      "step": 342670
    },
    {
      "epoch": 0.2333760302281431,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013146420000000001,
      "loss": 3.1951,
      "step": 342680
    },
    {
      "epoch": 0.2333946734643013,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001314622,
      "loss": 3.114,
      "step": 342690
    },
    {
      "epoch": 0.23341331670045948,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001314602,
      "loss": 3.2213,
      "step": 342700
    },
    {
      "epoch": 0.23343195993661767,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001314582,
      "loss": 3.2112,
      "step": 342710
    },
    {
      "epoch": 0.23345060317277583,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001314562,
      "loss": 3.1836,
      "step": 342720
    },
    {
      "epoch": 0.23346924640893402,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00013145420000000002,
      "loss": 2.9752,
      "step": 342730
    },
    {
      "epoch": 0.2334878896450922,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013145220000000002,
      "loss": 3.1663,
      "step": 342740
    },
    {
      "epoch": 0.2335065328812504,
      "grad_norm": 0.75,
      "learning_rate": 0.0001314502,
      "loss": 3.0912,
      "step": 342750
    },
    {
      "epoch": 0.23352517611740858,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001314482,
      "loss": 3.2211,
      "step": 342760
    },
    {
      "epoch": 0.23354381935356677,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001314462,
      "loss": 3.1609,
      "step": 342770
    },
    {
      "epoch": 0.23356246258972493,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013144420000000002,
      "loss": 3.1281,
      "step": 342780
    },
    {
      "epoch": 0.23358110582588312,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001314422,
      "loss": 3.0075,
      "step": 342790
    },
    {
      "epoch": 0.2335997490620413,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013144020000000002,
      "loss": 3.0087,
      "step": 342800
    },
    {
      "epoch": 0.2336183922981995,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001314382,
      "loss": 3.1517,
      "step": 342810
    },
    {
      "epoch": 0.2336370355343577,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001314362,
      "loss": 3.1718,
      "step": 342820
    },
    {
      "epoch": 0.23365567877051585,
      "grad_norm": 0.625,
      "learning_rate": 0.0001314342,
      "loss": 3.0894,
      "step": 342830
    },
    {
      "epoch": 0.23367432200667404,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001314322,
      "loss": 3.2474,
      "step": 342840
    },
    {
      "epoch": 0.23369296524283223,
      "grad_norm": 0.828125,
      "learning_rate": 0.00013143020000000002,
      "loss": 3.1995,
      "step": 342850
    },
    {
      "epoch": 0.23371160847899042,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001314282,
      "loss": 3.134,
      "step": 342860
    },
    {
      "epoch": 0.2337302517151486,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001314262,
      "loss": 3.1497,
      "step": 342870
    },
    {
      "epoch": 0.23374889495130677,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001314242,
      "loss": 3.1219,
      "step": 342880
    },
    {
      "epoch": 0.23376753818746496,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001314222,
      "loss": 3.1919,
      "step": 342890
    },
    {
      "epoch": 0.23378618142362315,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013142020000000003,
      "loss": 3.1507,
      "step": 342900
    },
    {
      "epoch": 0.23380482465978134,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001314182,
      "loss": 2.9955,
      "step": 342910
    },
    {
      "epoch": 0.23382346789593952,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013141620000000002,
      "loss": 3.2426,
      "step": 342920
    },
    {
      "epoch": 0.23384211113209769,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013141419999999999,
      "loss": 3.0963,
      "step": 342930
    },
    {
      "epoch": 0.23386075436825587,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001314122,
      "loss": 3.0097,
      "step": 342940
    },
    {
      "epoch": 0.23387939760441406,
      "grad_norm": 0.75,
      "learning_rate": 0.0001314102,
      "loss": 3.1252,
      "step": 342950
    },
    {
      "epoch": 0.23389804084057225,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001314082,
      "loss": 3.1635,
      "step": 342960
    },
    {
      "epoch": 0.23391668407673044,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013140620000000002,
      "loss": 3.1312,
      "step": 342970
    },
    {
      "epoch": 0.23393532731288863,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001314042,
      "loss": 3.1288,
      "step": 342980
    },
    {
      "epoch": 0.2339539705490468,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013140220000000001,
      "loss": 3.0953,
      "step": 342990
    },
    {
      "epoch": 0.23397261378520498,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013140019999999998,
      "loss": 3.1068,
      "step": 343000
    },
    {
      "epoch": 0.23399125702136317,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001313982,
      "loss": 3.2535,
      "step": 343010
    },
    {
      "epoch": 0.23400990025752136,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001313962,
      "loss": 3.0936,
      "step": 343020
    },
    {
      "epoch": 0.23402854349367955,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001313942,
      "loss": 3.0875,
      "step": 343030
    },
    {
      "epoch": 0.2340471867298377,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013139220000000002,
      "loss": 2.9997,
      "step": 343040
    },
    {
      "epoch": 0.2340658299659959,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001313902,
      "loss": 3.1258,
      "step": 343050
    },
    {
      "epoch": 0.2340844732021541,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001313882,
      "loss": 3.1512,
      "step": 343060
    },
    {
      "epoch": 0.23410311643831228,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001313862,
      "loss": 3.2353,
      "step": 343070
    },
    {
      "epoch": 0.23412175967447046,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001313842,
      "loss": 3.1778,
      "step": 343080
    },
    {
      "epoch": 0.23414040291062863,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013138220000000002,
      "loss": 3.2015,
      "step": 343090
    },
    {
      "epoch": 0.23415904614678681,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001313802,
      "loss": 3.0294,
      "step": 343100
    },
    {
      "epoch": 0.234177689382945,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013137820000000002,
      "loss": 3.1415,
      "step": 343110
    },
    {
      "epoch": 0.2341963326191032,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001313762,
      "loss": 2.962,
      "step": 343120
    },
    {
      "epoch": 0.23421497585526138,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001313742,
      "loss": 3.1567,
      "step": 343130
    },
    {
      "epoch": 0.23423361909141954,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001313722,
      "loss": 3.1562,
      "step": 343140
    },
    {
      "epoch": 0.23425226232757773,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001313702,
      "loss": 3.0743,
      "step": 343150
    },
    {
      "epoch": 0.23427090556373592,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013136820000000002,
      "loss": 3.2718,
      "step": 343160
    },
    {
      "epoch": 0.2342895487998941,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013136620000000002,
      "loss": 3.2413,
      "step": 343170
    },
    {
      "epoch": 0.2343081920360523,
      "grad_norm": 0.75,
      "learning_rate": 0.0001313642,
      "loss": 3.1097,
      "step": 343180
    },
    {
      "epoch": 0.2343268352722105,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001313622,
      "loss": 3.1889,
      "step": 343190
    },
    {
      "epoch": 0.23434547850836865,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001313602,
      "loss": 3.1802,
      "step": 343200
    },
    {
      "epoch": 0.23436412174452684,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001313582,
      "loss": 3.1069,
      "step": 343210
    },
    {
      "epoch": 0.23438276498068503,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001313562,
      "loss": 3.2024,
      "step": 343220
    },
    {
      "epoch": 0.23440140821684322,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013135420000000002,
      "loss": 3.1258,
      "step": 343230
    },
    {
      "epoch": 0.2344200514530014,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001313522,
      "loss": 3.0867,
      "step": 343240
    },
    {
      "epoch": 0.23443869468915957,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001313502,
      "loss": 3.071,
      "step": 343250
    },
    {
      "epoch": 0.23445733792531775,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001313482,
      "loss": 3.1103,
      "step": 343260
    },
    {
      "epoch": 0.23447598116147594,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001313462,
      "loss": 3.1969,
      "step": 343270
    },
    {
      "epoch": 0.23449462439763413,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001313442,
      "loss": 3.134,
      "step": 343280
    },
    {
      "epoch": 0.23451326763379232,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00013134220000000002,
      "loss": 3.126,
      "step": 343290
    },
    {
      "epoch": 0.23453191086995048,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001313402,
      "loss": 3.291,
      "step": 343300
    },
    {
      "epoch": 0.23455055410610867,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001313382,
      "loss": 3.127,
      "step": 343310
    },
    {
      "epoch": 0.23456919734226686,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001313362,
      "loss": 3.0436,
      "step": 343320
    },
    {
      "epoch": 0.23458784057842505,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001313342,
      "loss": 3.0925,
      "step": 343330
    },
    {
      "epoch": 0.23460648381458324,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013133220000000002,
      "loss": 3.1077,
      "step": 343340
    },
    {
      "epoch": 0.2346251270507414,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001313302,
      "loss": 2.9662,
      "step": 343350
    },
    {
      "epoch": 0.2346437702868996,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013132820000000001,
      "loss": 3.1393,
      "step": 343360
    },
    {
      "epoch": 0.23466241352305778,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001313262,
      "loss": 3.1394,
      "step": 343370
    },
    {
      "epoch": 0.23468105675921597,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001313242,
      "loss": 3.0678,
      "step": 343380
    },
    {
      "epoch": 0.23469969999537416,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001313222,
      "loss": 3.2492,
      "step": 343390
    },
    {
      "epoch": 0.23471834323153232,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001313202,
      "loss": 3.2113,
      "step": 343400
    },
    {
      "epoch": 0.2347369864676905,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013131820000000002,
      "loss": 3.1668,
      "step": 343410
    },
    {
      "epoch": 0.2347556297038487,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001313162,
      "loss": 3.0794,
      "step": 343420
    },
    {
      "epoch": 0.23477427294000688,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001313142,
      "loss": 3.1795,
      "step": 343430
    },
    {
      "epoch": 0.23479291617616507,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001313122,
      "loss": 3.166,
      "step": 343440
    },
    {
      "epoch": 0.23481155941232326,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001313102,
      "loss": 3.1445,
      "step": 343450
    },
    {
      "epoch": 0.23483020264848142,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013130820000000002,
      "loss": 3.1842,
      "step": 343460
    },
    {
      "epoch": 0.2348488458846396,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001313062,
      "loss": 3.1595,
      "step": 343470
    },
    {
      "epoch": 0.2348674891207978,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013130420000000002,
      "loss": 3.1506,
      "step": 343480
    },
    {
      "epoch": 0.234886132356956,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013130219999999998,
      "loss": 3.0654,
      "step": 343490
    },
    {
      "epoch": 0.23490477559311418,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001313002,
      "loss": 3.1428,
      "step": 343500
    },
    {
      "epoch": 0.23492341882927234,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001312982,
      "loss": 3.2532,
      "step": 343510
    },
    {
      "epoch": 0.23494206206543053,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001312962,
      "loss": 3.1621,
      "step": 343520
    },
    {
      "epoch": 0.23496070530158872,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013129420000000002,
      "loss": 3.1293,
      "step": 343530
    },
    {
      "epoch": 0.2349793485377469,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001312922,
      "loss": 3.081,
      "step": 343540
    },
    {
      "epoch": 0.2349979917739051,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001312902,
      "loss": 3.1318,
      "step": 343550
    },
    {
      "epoch": 0.23501663501006326,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001312882,
      "loss": 3.0911,
      "step": 343560
    },
    {
      "epoch": 0.23503527824622145,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001312862,
      "loss": 3.058,
      "step": 343570
    },
    {
      "epoch": 0.23505392148237964,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013128420000000003,
      "loss": 3.1035,
      "step": 343580
    },
    {
      "epoch": 0.23507256471853782,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001312822,
      "loss": 3.079,
      "step": 343590
    },
    {
      "epoch": 0.235091207954696,
      "grad_norm": 0.75,
      "learning_rate": 0.00013128020000000002,
      "loss": 3.256,
      "step": 343600
    },
    {
      "epoch": 0.23510985119085417,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013127819999999999,
      "loss": 3.1856,
      "step": 343610
    },
    {
      "epoch": 0.23512849442701236,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001312762,
      "loss": 3.1246,
      "step": 343620
    },
    {
      "epoch": 0.23514713766317055,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001312742,
      "loss": 3.0376,
      "step": 343630
    },
    {
      "epoch": 0.23516578089932874,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001312722,
      "loss": 3.1196,
      "step": 343640
    },
    {
      "epoch": 0.23518442413548693,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013127020000000002,
      "loss": 3.1625,
      "step": 343650
    },
    {
      "epoch": 0.23520306737164512,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001312682,
      "loss": 3.086,
      "step": 343660
    },
    {
      "epoch": 0.23522171060780328,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013126620000000001,
      "loss": 3.0327,
      "step": 343670
    },
    {
      "epoch": 0.23524035384396147,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001312642,
      "loss": 3.283,
      "step": 343680
    },
    {
      "epoch": 0.23525899708011966,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001312622,
      "loss": 3.0619,
      "step": 343690
    },
    {
      "epoch": 0.23527764031627785,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001312602,
      "loss": 2.9571,
      "step": 343700
    },
    {
      "epoch": 0.23529628355243604,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001312582,
      "loss": 3.135,
      "step": 343710
    },
    {
      "epoch": 0.2353149267885942,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013125620000000002,
      "loss": 3.2416,
      "step": 343720
    },
    {
      "epoch": 0.2353335700247524,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013125420000000001,
      "loss": 3.1178,
      "step": 343730
    },
    {
      "epoch": 0.23535221326091058,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001312522,
      "loss": 3.0811,
      "step": 343740
    },
    {
      "epoch": 0.23537085649706876,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001312502,
      "loss": 3.0917,
      "step": 343750
    },
    {
      "epoch": 0.23538949973322695,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001312482,
      "loss": 3.0541,
      "step": 343760
    },
    {
      "epoch": 0.23540814296938511,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001312462,
      "loss": 3.253,
      "step": 343770
    },
    {
      "epoch": 0.2354267862055433,
      "grad_norm": 0.90625,
      "learning_rate": 0.00013124420000000002,
      "loss": 3.1062,
      "step": 343780
    },
    {
      "epoch": 0.2354454294417015,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013124220000000002,
      "loss": 3.1413,
      "step": 343790
    },
    {
      "epoch": 0.23546407267785968,
      "grad_norm": 0.75,
      "learning_rate": 0.0001312402,
      "loss": 3.1211,
      "step": 343800
    },
    {
      "epoch": 0.23548271591401787,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001312382,
      "loss": 2.9751,
      "step": 343810
    },
    {
      "epoch": 0.23550135915017603,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001312362,
      "loss": 3.0599,
      "step": 343820
    },
    {
      "epoch": 0.23552000238633422,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001312342,
      "loss": 3.0168,
      "step": 343830
    },
    {
      "epoch": 0.2355386456224924,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001312322,
      "loss": 3.2142,
      "step": 343840
    },
    {
      "epoch": 0.2355572888586506,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013123020000000002,
      "loss": 3.2119,
      "step": 343850
    },
    {
      "epoch": 0.2355759320948088,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001312282,
      "loss": 3.1585,
      "step": 343860
    },
    {
      "epoch": 0.23559457533096698,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001312262,
      "loss": 3.1398,
      "step": 343870
    },
    {
      "epoch": 0.23561321856712514,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001312242,
      "loss": 3.0547,
      "step": 343880
    },
    {
      "epoch": 0.23563186180328333,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001312222,
      "loss": 3.0727,
      "step": 343890
    },
    {
      "epoch": 0.23565050503944152,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013122020000000002,
      "loss": 3.0439,
      "step": 343900
    },
    {
      "epoch": 0.2356691482755997,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001312182,
      "loss": 3.1112,
      "step": 343910
    },
    {
      "epoch": 0.2356877915117579,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001312162,
      "loss": 3.113,
      "step": 343920
    },
    {
      "epoch": 0.23570643474791605,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001312142,
      "loss": 3.2327,
      "step": 343930
    },
    {
      "epoch": 0.23572507798407424,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001312122,
      "loss": 3.1074,
      "step": 343940
    },
    {
      "epoch": 0.23574372122023243,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013121020000000003,
      "loss": 3.0589,
      "step": 343950
    },
    {
      "epoch": 0.23576236445639062,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001312082,
      "loss": 3.1482,
      "step": 343960
    },
    {
      "epoch": 0.2357810076925488,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013120620000000002,
      "loss": 3.2515,
      "step": 343970
    },
    {
      "epoch": 0.23579965092870697,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013120419999999999,
      "loss": 3.1338,
      "step": 343980
    },
    {
      "epoch": 0.23581829416486516,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001312022,
      "loss": 3.0295,
      "step": 343990
    },
    {
      "epoch": 0.23583693740102335,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001312002,
      "loss": 3.0489,
      "step": 344000
    },
    {
      "epoch": 0.23585558063718154,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001311982,
      "loss": 3.2061,
      "step": 344010
    },
    {
      "epoch": 0.23587422387333973,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013119620000000002,
      "loss": 3.1184,
      "step": 344020
    },
    {
      "epoch": 0.2358928671094979,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001311942,
      "loss": 3.292,
      "step": 344030
    },
    {
      "epoch": 0.23591151034565608,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013119220000000001,
      "loss": 3.1796,
      "step": 344040
    },
    {
      "epoch": 0.23593015358181427,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001311902,
      "loss": 2.9563,
      "step": 344050
    },
    {
      "epoch": 0.23594879681797246,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001311882,
      "loss": 3.2399,
      "step": 344060
    },
    {
      "epoch": 0.23596744005413064,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013118620000000003,
      "loss": 3.2061,
      "step": 344070
    },
    {
      "epoch": 0.2359860832902888,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001311842,
      "loss": 3.1695,
      "step": 344080
    },
    {
      "epoch": 0.236004726526447,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013118220000000002,
      "loss": 3.1222,
      "step": 344090
    },
    {
      "epoch": 0.23602336976260518,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001311802,
      "loss": 3.0108,
      "step": 344100
    },
    {
      "epoch": 0.23604201299876337,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001311782,
      "loss": 3.1453,
      "step": 344110
    },
    {
      "epoch": 0.23606065623492156,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001311762,
      "loss": 3.0406,
      "step": 344120
    },
    {
      "epoch": 0.23607929947107975,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001311742,
      "loss": 3.042,
      "step": 344130
    },
    {
      "epoch": 0.2360979427072379,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013117220000000002,
      "loss": 3.2084,
      "step": 344140
    },
    {
      "epoch": 0.2361165859433961,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001311702,
      "loss": 3.1715,
      "step": 344150
    },
    {
      "epoch": 0.2361352291795543,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013116820000000002,
      "loss": 3.213,
      "step": 344160
    },
    {
      "epoch": 0.23615387241571248,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001311662,
      "loss": 3.1307,
      "step": 344170
    },
    {
      "epoch": 0.23617251565187067,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001311642,
      "loss": 3.0235,
      "step": 344180
    },
    {
      "epoch": 0.23619115888802883,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001311622,
      "loss": 3.2299,
      "step": 344190
    },
    {
      "epoch": 0.23620980212418702,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001311602,
      "loss": 3.0888,
      "step": 344200
    },
    {
      "epoch": 0.2362284453603452,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013115820000000002,
      "loss": 3.1078,
      "step": 344210
    },
    {
      "epoch": 0.2362470885965034,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001311562,
      "loss": 3.2459,
      "step": 344220
    },
    {
      "epoch": 0.23626573183266159,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001311542,
      "loss": 3.2416,
      "step": 344230
    },
    {
      "epoch": 0.23628437506881975,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001311522,
      "loss": 3.0862,
      "step": 344240
    },
    {
      "epoch": 0.23630301830497794,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001311502,
      "loss": 3.0233,
      "step": 344250
    },
    {
      "epoch": 0.23632166154113612,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001311482,
      "loss": 3.0451,
      "step": 344260
    },
    {
      "epoch": 0.2363403047772943,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001311462,
      "loss": 3.1295,
      "step": 344270
    },
    {
      "epoch": 0.2363589480134525,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013114420000000002,
      "loss": 3.0789,
      "step": 344280
    },
    {
      "epoch": 0.23637759124961066,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001311422,
      "loss": 3.177,
      "step": 344290
    },
    {
      "epoch": 0.23639623448576885,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001311402,
      "loss": 3.1605,
      "step": 344300
    },
    {
      "epoch": 0.23641487772192704,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001311382,
      "loss": 3.1476,
      "step": 344310
    },
    {
      "epoch": 0.23643352095808523,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001311362,
      "loss": 3.1404,
      "step": 344320
    },
    {
      "epoch": 0.23645216419424342,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001311342,
      "loss": 3.0019,
      "step": 344330
    },
    {
      "epoch": 0.2364708074304016,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013113220000000002,
      "loss": 3.0499,
      "step": 344340
    },
    {
      "epoch": 0.23648945066655977,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013113020000000001,
      "loss": 3.0293,
      "step": 344350
    },
    {
      "epoch": 0.23650809390271796,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001311282,
      "loss": 3.1526,
      "step": 344360
    },
    {
      "epoch": 0.23652673713887615,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001311262,
      "loss": 3.173,
      "step": 344370
    },
    {
      "epoch": 0.23654538037503434,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001311242,
      "loss": 3.0401,
      "step": 344380
    },
    {
      "epoch": 0.23656402361119253,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013112220000000002,
      "loss": 3.0279,
      "step": 344390
    },
    {
      "epoch": 0.2365826668473507,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001311202,
      "loss": 3.2478,
      "step": 344400
    },
    {
      "epoch": 0.23660131008350888,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013111820000000001,
      "loss": 3.0888,
      "step": 344410
    },
    {
      "epoch": 0.23661995331966706,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001311162,
      "loss": 3.1062,
      "step": 344420
    },
    {
      "epoch": 0.23663859655582525,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001311142,
      "loss": 3.1598,
      "step": 344430
    },
    {
      "epoch": 0.23665723979198344,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001311122,
      "loss": 3.2305,
      "step": 344440
    },
    {
      "epoch": 0.2366758830281416,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001311102,
      "loss": 3.1393,
      "step": 344450
    },
    {
      "epoch": 0.2366945262642998,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013110820000000002,
      "loss": 3.1194,
      "step": 344460
    },
    {
      "epoch": 0.23671316950045798,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001311062,
      "loss": 3.1692,
      "step": 344470
    },
    {
      "epoch": 0.23673181273661617,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001311042,
      "loss": 3.0313,
      "step": 344480
    },
    {
      "epoch": 0.23675045597277436,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001311022,
      "loss": 3.006,
      "step": 344490
    },
    {
      "epoch": 0.23676909920893252,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001311002,
      "loss": 3.262,
      "step": 344500
    },
    {
      "epoch": 0.2367877424450907,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013109820000000002,
      "loss": 3.0175,
      "step": 344510
    },
    {
      "epoch": 0.2368063856812489,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001310962,
      "loss": 3.1094,
      "step": 344520
    },
    {
      "epoch": 0.2368250289174071,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013109420000000002,
      "loss": 3.1102,
      "step": 344530
    },
    {
      "epoch": 0.23684367215356528,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001310922,
      "loss": 3.2058,
      "step": 344540
    },
    {
      "epoch": 0.23686231538972347,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001310902,
      "loss": 3.1384,
      "step": 344550
    },
    {
      "epoch": 0.23688095862588163,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001310882,
      "loss": 3.1147,
      "step": 344560
    },
    {
      "epoch": 0.23689960186203982,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001310862,
      "loss": 3.1628,
      "step": 344570
    },
    {
      "epoch": 0.236918245098198,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013108420000000002,
      "loss": 3.0574,
      "step": 344580
    },
    {
      "epoch": 0.2369368883343562,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001310822,
      "loss": 3.0933,
      "step": 344590
    },
    {
      "epoch": 0.23695553157051438,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001310802,
      "loss": 3.0522,
      "step": 344600
    },
    {
      "epoch": 0.23697417480667254,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001310782,
      "loss": 3.2343,
      "step": 344610
    },
    {
      "epoch": 0.23699281804283073,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001310762,
      "loss": 3.1179,
      "step": 344620
    },
    {
      "epoch": 0.23701146127898892,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013107420000000003,
      "loss": 3.1208,
      "step": 344630
    },
    {
      "epoch": 0.2370301045151471,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001310722,
      "loss": 3.138,
      "step": 344640
    },
    {
      "epoch": 0.2370487477513053,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013107020000000002,
      "loss": 3.0937,
      "step": 344650
    },
    {
      "epoch": 0.23706739098746346,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013106819999999999,
      "loss": 3.1865,
      "step": 344660
    },
    {
      "epoch": 0.23708603422362165,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001310662,
      "loss": 3.23,
      "step": 344670
    },
    {
      "epoch": 0.23710467745977984,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001310642,
      "loss": 3.0644,
      "step": 344680
    },
    {
      "epoch": 0.23712332069593803,
      "grad_norm": 0.625,
      "learning_rate": 0.0001310622,
      "loss": 3.1775,
      "step": 344690
    },
    {
      "epoch": 0.23714196393209622,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013106020000000002,
      "loss": 3.2431,
      "step": 344700
    },
    {
      "epoch": 0.23716060716825438,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001310582,
      "loss": 3.2279,
      "step": 344710
    },
    {
      "epoch": 0.23717925040441257,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013105620000000001,
      "loss": 3.1851,
      "step": 344720
    },
    {
      "epoch": 0.23719789364057076,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001310542,
      "loss": 3.1721,
      "step": 344730
    },
    {
      "epoch": 0.23721653687672895,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001310522,
      "loss": 3.0933,
      "step": 344740
    },
    {
      "epoch": 0.23723518011288713,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001310502,
      "loss": 3.1665,
      "step": 344750
    },
    {
      "epoch": 0.2372538233490453,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001310482,
      "loss": 3.0671,
      "step": 344760
    },
    {
      "epoch": 0.23727246658520348,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013104620000000002,
      "loss": 3.2336,
      "step": 344770
    },
    {
      "epoch": 0.23729110982136167,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013104420000000001,
      "loss": 3.0127,
      "step": 344780
    },
    {
      "epoch": 0.23730975305751986,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001310422,
      "loss": 3.1058,
      "step": 344790
    },
    {
      "epoch": 0.23732839629367805,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001310402,
      "loss": 3.182,
      "step": 344800
    },
    {
      "epoch": 0.23734703952983624,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001310382,
      "loss": 3.1475,
      "step": 344810
    },
    {
      "epoch": 0.2373656827659944,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001310362,
      "loss": 3.1457,
      "step": 344820
    },
    {
      "epoch": 0.2373843260021526,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001310342,
      "loss": 3.1404,
      "step": 344830
    },
    {
      "epoch": 0.23740296923831078,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013103220000000002,
      "loss": 3.1158,
      "step": 344840
    },
    {
      "epoch": 0.23742161247446897,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001310302,
      "loss": 3.0363,
      "step": 344850
    },
    {
      "epoch": 0.23744025571062716,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001310282,
      "loss": 3.1628,
      "step": 344860
    },
    {
      "epoch": 0.23745889894678532,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001310262,
      "loss": 3.3301,
      "step": 344870
    },
    {
      "epoch": 0.2374775421829435,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001310242,
      "loss": 3.1624,
      "step": 344880
    },
    {
      "epoch": 0.2374961854191017,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001310222,
      "loss": 3.2075,
      "step": 344890
    },
    {
      "epoch": 0.23751482865525989,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013102020000000002,
      "loss": 3.0951,
      "step": 344900
    },
    {
      "epoch": 0.23753347189141807,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001310182,
      "loss": 3.1582,
      "step": 344910
    },
    {
      "epoch": 0.23755211512757624,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001310162,
      "loss": 3.1794,
      "step": 344920
    },
    {
      "epoch": 0.23757075836373442,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001310142,
      "loss": 3.1394,
      "step": 344930
    },
    {
      "epoch": 0.2375894015998926,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001310122,
      "loss": 3.2031,
      "step": 344940
    },
    {
      "epoch": 0.2376080448360508,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013101020000000002,
      "loss": 3.0229,
      "step": 344950
    },
    {
      "epoch": 0.237626688072209,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001310082,
      "loss": 3.1791,
      "step": 344960
    },
    {
      "epoch": 0.23764533130836715,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001310062,
      "loss": 3.2585,
      "step": 344970
    },
    {
      "epoch": 0.23766397454452534,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001310042,
      "loss": 3.1645,
      "step": 344980
    },
    {
      "epoch": 0.23768261778068353,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001310022,
      "loss": 3.103,
      "step": 344990
    },
    {
      "epoch": 0.23770126101684172,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013100020000000003,
      "loss": 3.2242,
      "step": 345000
    },
    {
      "epoch": 0.2377199042529999,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001309982,
      "loss": 3.1495,
      "step": 345010
    },
    {
      "epoch": 0.2377385474891581,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013099620000000002,
      "loss": 3.0887,
      "step": 345020
    },
    {
      "epoch": 0.23775719072531626,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013099420000000001,
      "loss": 3.0287,
      "step": 345030
    },
    {
      "epoch": 0.23777583396147445,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001309922,
      "loss": 3.201,
      "step": 345040
    },
    {
      "epoch": 0.23779447719763264,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001309902,
      "loss": 3.0478,
      "step": 345050
    },
    {
      "epoch": 0.23781312043379083,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001309882,
      "loss": 3.2342,
      "step": 345060
    },
    {
      "epoch": 0.23783176366994901,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013098620000000002,
      "loss": 3.1536,
      "step": 345070
    },
    {
      "epoch": 0.23785040690610718,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001309842,
      "loss": 3.104,
      "step": 345080
    },
    {
      "epoch": 0.23786905014226536,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013098220000000001,
      "loss": 3.1035,
      "step": 345090
    },
    {
      "epoch": 0.23788769337842355,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001309802,
      "loss": 3.2337,
      "step": 345100
    },
    {
      "epoch": 0.23790633661458174,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001309782,
      "loss": 3.1452,
      "step": 345110
    },
    {
      "epoch": 0.23792497985073993,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013097620000000003,
      "loss": 3.0452,
      "step": 345120
    },
    {
      "epoch": 0.2379436230868981,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001309742,
      "loss": 3.0998,
      "step": 345130
    },
    {
      "epoch": 0.23796226632305628,
      "grad_norm": 0.875,
      "learning_rate": 0.00013097220000000002,
      "loss": 3.1771,
      "step": 345140
    },
    {
      "epoch": 0.23798090955921447,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001309702,
      "loss": 3.0939,
      "step": 345150
    },
    {
      "epoch": 0.23799955279537266,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001309682,
      "loss": 3.1586,
      "step": 345160
    },
    {
      "epoch": 0.23801819603153085,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001309662,
      "loss": 2.9904,
      "step": 345170
    },
    {
      "epoch": 0.238036839267689,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001309642,
      "loss": 3.1533,
      "step": 345180
    },
    {
      "epoch": 0.2380554825038472,
      "grad_norm": 0.828125,
      "learning_rate": 0.00013096220000000002,
      "loss": 3.129,
      "step": 345190
    },
    {
      "epoch": 0.2380741257400054,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001309602,
      "loss": 3.1716,
      "step": 345200
    },
    {
      "epoch": 0.23809276897616358,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013095820000000002,
      "loss": 3.096,
      "step": 345210
    },
    {
      "epoch": 0.23811141221232177,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013095619999999998,
      "loss": 3.1626,
      "step": 345220
    },
    {
      "epoch": 0.23813005544847995,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001309542,
      "loss": 3.113,
      "step": 345230
    },
    {
      "epoch": 0.23814869868463812,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001309522,
      "loss": 3.1048,
      "step": 345240
    },
    {
      "epoch": 0.2381673419207963,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001309502,
      "loss": 3.1548,
      "step": 345250
    },
    {
      "epoch": 0.2381859851569545,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013094820000000002,
      "loss": 3.0979,
      "step": 345260
    },
    {
      "epoch": 0.23820462839311268,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001309462,
      "loss": 3.2279,
      "step": 345270
    },
    {
      "epoch": 0.23822327162927087,
      "grad_norm": 0.75,
      "learning_rate": 0.0001309442,
      "loss": 3.1121,
      "step": 345280
    },
    {
      "epoch": 0.23824191486542903,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001309422,
      "loss": 3.178,
      "step": 345290
    },
    {
      "epoch": 0.23826055810158722,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001309402,
      "loss": 3.12,
      "step": 345300
    },
    {
      "epoch": 0.2382792013377454,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001309382,
      "loss": 3.2436,
      "step": 345310
    },
    {
      "epoch": 0.2382978445739036,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001309362,
      "loss": 3.148,
      "step": 345320
    },
    {
      "epoch": 0.2383164878100618,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013093420000000002,
      "loss": 3.0924,
      "step": 345330
    },
    {
      "epoch": 0.23833513104621995,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001309322,
      "loss": 3.175,
      "step": 345340
    },
    {
      "epoch": 0.23835377428237814,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001309302,
      "loss": 3.155,
      "step": 345350
    },
    {
      "epoch": 0.23837241751853633,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001309282,
      "loss": 3.1275,
      "step": 345360
    },
    {
      "epoch": 0.23839106075469452,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001309262,
      "loss": 3.1152,
      "step": 345370
    },
    {
      "epoch": 0.2384097039908527,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001309242,
      "loss": 3.0641,
      "step": 345380
    },
    {
      "epoch": 0.23842834722701087,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013092220000000002,
      "loss": 3.1035,
      "step": 345390
    },
    {
      "epoch": 0.23844699046316906,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013092020000000001,
      "loss": 3.0342,
      "step": 345400
    },
    {
      "epoch": 0.23846563369932725,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001309182,
      "loss": 3.254,
      "step": 345410
    },
    {
      "epoch": 0.23848427693548543,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001309162,
      "loss": 3.0567,
      "step": 345420
    },
    {
      "epoch": 0.23850292017164362,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001309142,
      "loss": 3.1642,
      "step": 345430
    },
    {
      "epoch": 0.23852156340780178,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001309122,
      "loss": 3.0811,
      "step": 345440
    },
    {
      "epoch": 0.23854020664395997,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001309102,
      "loss": 3.1067,
      "step": 345450
    },
    {
      "epoch": 0.23855884988011816,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013090820000000001,
      "loss": 3.1263,
      "step": 345460
    },
    {
      "epoch": 0.23857749311627635,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001309062,
      "loss": 3.1259,
      "step": 345470
    },
    {
      "epoch": 0.23859613635243454,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001309042,
      "loss": 3.0991,
      "step": 345480
    },
    {
      "epoch": 0.23861477958859273,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001309022,
      "loss": 3.1871,
      "step": 345490
    },
    {
      "epoch": 0.2386334228247509,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001309002,
      "loss": 3.1113,
      "step": 345500
    },
    {
      "epoch": 0.23865206606090908,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013089820000000002,
      "loss": 3.0569,
      "step": 345510
    },
    {
      "epoch": 0.23867070929706727,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013089620000000002,
      "loss": 3.0662,
      "step": 345520
    },
    {
      "epoch": 0.23868935253322546,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001308942,
      "loss": 3.1592,
      "step": 345530
    },
    {
      "epoch": 0.23870799576938365,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001308922,
      "loss": 3.2036,
      "step": 345540
    },
    {
      "epoch": 0.2387266390055418,
      "grad_norm": 0.75,
      "learning_rate": 0.0001308902,
      "loss": 2.9849,
      "step": 345550
    },
    {
      "epoch": 0.2387452822417,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013088820000000003,
      "loss": 3.0577,
      "step": 345560
    },
    {
      "epoch": 0.23876392547785819,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001308862,
      "loss": 3.1868,
      "step": 345570
    },
    {
      "epoch": 0.23878256871401637,
      "grad_norm": 0.828125,
      "learning_rate": 0.00013088420000000002,
      "loss": 3.3059,
      "step": 345580
    },
    {
      "epoch": 0.23880121195017456,
      "grad_norm": 0.625,
      "learning_rate": 0.0001308822,
      "loss": 3.1756,
      "step": 345590
    },
    {
      "epoch": 0.23881985518633272,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001308802,
      "loss": 3.2672,
      "step": 345600
    },
    {
      "epoch": 0.2388384984224909,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001308782,
      "loss": 3.1213,
      "step": 345610
    },
    {
      "epoch": 0.2388571416586491,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001308762,
      "loss": 3.1163,
      "step": 345620
    },
    {
      "epoch": 0.2388757848948073,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013087420000000002,
      "loss": 3.0479,
      "step": 345630
    },
    {
      "epoch": 0.23889442813096548,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001308722,
      "loss": 3.2085,
      "step": 345640
    },
    {
      "epoch": 0.23891307136712364,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001308702,
      "loss": 3.2421,
      "step": 345650
    },
    {
      "epoch": 0.23893171460328183,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001308682,
      "loss": 3.0186,
      "step": 345660
    },
    {
      "epoch": 0.23895035783944002,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001308662,
      "loss": 3.0507,
      "step": 345670
    },
    {
      "epoch": 0.2389690010755982,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013086420000000003,
      "loss": 3.0488,
      "step": 345680
    },
    {
      "epoch": 0.2389876443117564,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001308622,
      "loss": 3.2132,
      "step": 345690
    },
    {
      "epoch": 0.2390062875479146,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013086020000000002,
      "loss": 3.1872,
      "step": 345700
    },
    {
      "epoch": 0.23902493078407275,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001308582,
      "loss": 3.1418,
      "step": 345710
    },
    {
      "epoch": 0.23904357402023094,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001308562,
      "loss": 3.196,
      "step": 345720
    },
    {
      "epoch": 0.23906221725638913,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001308542,
      "loss": 3.0968,
      "step": 345730
    },
    {
      "epoch": 0.23908086049254731,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001308522,
      "loss": 3.1443,
      "step": 345740
    },
    {
      "epoch": 0.2390995037287055,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013085020000000002,
      "loss": 3.1955,
      "step": 345750
    },
    {
      "epoch": 0.23911814696486366,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001308482,
      "loss": 3.1365,
      "step": 345760
    },
    {
      "epoch": 0.23913679020102185,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013084620000000001,
      "loss": 3.1373,
      "step": 345770
    },
    {
      "epoch": 0.23915543343718004,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001308442,
      "loss": 3.065,
      "step": 345780
    },
    {
      "epoch": 0.23917407667333823,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001308422,
      "loss": 3.1534,
      "step": 345790
    },
    {
      "epoch": 0.23919271990949642,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001308402,
      "loss": 3.1,
      "step": 345800
    },
    {
      "epoch": 0.23921136314565458,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001308382,
      "loss": 3.1071,
      "step": 345810
    },
    {
      "epoch": 0.23923000638181277,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013083620000000002,
      "loss": 3.2184,
      "step": 345820
    },
    {
      "epoch": 0.23924864961797096,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001308342,
      "loss": 3.2307,
      "step": 345830
    },
    {
      "epoch": 0.23926729285412915,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001308322,
      "loss": 3.1522,
      "step": 345840
    },
    {
      "epoch": 0.23928593609028734,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001308302,
      "loss": 3.1813,
      "step": 345850
    },
    {
      "epoch": 0.2393045793264455,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001308282,
      "loss": 3.1549,
      "step": 345860
    },
    {
      "epoch": 0.2393232225626037,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001308262,
      "loss": 3.0831,
      "step": 345870
    },
    {
      "epoch": 0.23934186579876188,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001308242,
      "loss": 3.1627,
      "step": 345880
    },
    {
      "epoch": 0.23936050903492007,
      "grad_norm": 0.625,
      "learning_rate": 0.00013082220000000002,
      "loss": 3.1755,
      "step": 345890
    },
    {
      "epoch": 0.23937915227107825,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001308202,
      "loss": 3.2275,
      "step": 345900
    },
    {
      "epoch": 0.23939779550723644,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001308182,
      "loss": 3.1586,
      "step": 345910
    },
    {
      "epoch": 0.2394164387433946,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001308162,
      "loss": 3.1231,
      "step": 345920
    },
    {
      "epoch": 0.2394350819795528,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001308142,
      "loss": 3.0421,
      "step": 345930
    },
    {
      "epoch": 0.23945372521571098,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001308122,
      "loss": 2.9986,
      "step": 345940
    },
    {
      "epoch": 0.23947236845186917,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013081020000000002,
      "loss": 2.9486,
      "step": 345950
    },
    {
      "epoch": 0.23949101168802736,
      "grad_norm": 0.875,
      "learning_rate": 0.0001308082,
      "loss": 3.1218,
      "step": 345960
    },
    {
      "epoch": 0.23950965492418552,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001308062,
      "loss": 3.1472,
      "step": 345970
    },
    {
      "epoch": 0.2395282981603437,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001308042,
      "loss": 2.9733,
      "step": 345980
    },
    {
      "epoch": 0.2395469413965019,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001308022,
      "loss": 3.0296,
      "step": 345990
    },
    {
      "epoch": 0.2395655846326601,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00013080020000000002,
      "loss": 3.088,
      "step": 346000
    },
    {
      "epoch": 0.23958422786881828,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001307982,
      "loss": 3.1329,
      "step": 346010
    },
    {
      "epoch": 0.23960287110497644,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001307962,
      "loss": 3.0926,
      "step": 346020
    },
    {
      "epoch": 0.23962151434113463,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001307942,
      "loss": 3.1582,
      "step": 346030
    },
    {
      "epoch": 0.23964015757729282,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001307922,
      "loss": 3.0468,
      "step": 346040
    },
    {
      "epoch": 0.239658800813451,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001307902,
      "loss": 3.2235,
      "step": 346050
    },
    {
      "epoch": 0.2396774440496092,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001307882,
      "loss": 3.1463,
      "step": 346060
    },
    {
      "epoch": 0.23969608728576736,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013078620000000002,
      "loss": 3.1532,
      "step": 346070
    },
    {
      "epoch": 0.23971473052192555,
      "grad_norm": 0.859375,
      "learning_rate": 0.00013078420000000001,
      "loss": 3.0765,
      "step": 346080
    },
    {
      "epoch": 0.23973337375808373,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001307822,
      "loss": 3.1516,
      "step": 346090
    },
    {
      "epoch": 0.23975201699424192,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001307802,
      "loss": 2.9857,
      "step": 346100
    },
    {
      "epoch": 0.2397706602304001,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001307782,
      "loss": 3.129,
      "step": 346110
    },
    {
      "epoch": 0.2397893034665583,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013077620000000002,
      "loss": 3.1328,
      "step": 346120
    },
    {
      "epoch": 0.23980794670271646,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001307742,
      "loss": 3.126,
      "step": 346130
    },
    {
      "epoch": 0.23982658993887465,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013077220000000001,
      "loss": 3.0349,
      "step": 346140
    },
    {
      "epoch": 0.23984523317503284,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001307702,
      "loss": 3.1398,
      "step": 346150
    },
    {
      "epoch": 0.23986387641119103,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001307682,
      "loss": 3.094,
      "step": 346160
    },
    {
      "epoch": 0.23988251964734922,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013076620000000003,
      "loss": 3.0769,
      "step": 346170
    },
    {
      "epoch": 0.23990116288350738,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001307642,
      "loss": 3.1922,
      "step": 346180
    },
    {
      "epoch": 0.23991980611966557,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013076220000000002,
      "loss": 3.1258,
      "step": 346190
    },
    {
      "epoch": 0.23993844935582376,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001307602,
      "loss": 3.1818,
      "step": 346200
    },
    {
      "epoch": 0.23995709259198195,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001307582,
      "loss": 3.1228,
      "step": 346210
    },
    {
      "epoch": 0.23997573582814014,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001307562,
      "loss": 3.0663,
      "step": 346220
    },
    {
      "epoch": 0.2399943790642983,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001307542,
      "loss": 3.1234,
      "step": 346230
    },
    {
      "epoch": 0.24001302230045649,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013075220000000003,
      "loss": 3.1543,
      "step": 346240
    },
    {
      "epoch": 0.24003166553661467,
      "grad_norm": 0.75,
      "learning_rate": 0.0001307502,
      "loss": 3.1785,
      "step": 346250
    },
    {
      "epoch": 0.24005030877277286,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013074820000000002,
      "loss": 3.1289,
      "step": 346260
    },
    {
      "epoch": 0.24006895200893105,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013074619999999998,
      "loss": 3.1113,
      "step": 346270
    },
    {
      "epoch": 0.2400875952450892,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001307442,
      "loss": 3.1523,
      "step": 346280
    },
    {
      "epoch": 0.2401062384812474,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001307422,
      "loss": 3.1034,
      "step": 346290
    },
    {
      "epoch": 0.2401248817174056,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001307402,
      "loss": 3.11,
      "step": 346300
    },
    {
      "epoch": 0.24014352495356378,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013073820000000002,
      "loss": 3.1529,
      "step": 346310
    },
    {
      "epoch": 0.24016216818972197,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001307362,
      "loss": 3.2343,
      "step": 346320
    },
    {
      "epoch": 0.24018081142588013,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001307342,
      "loss": 3.0559,
      "step": 346330
    },
    {
      "epoch": 0.24019945466203832,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001307322,
      "loss": 3.048,
      "step": 346340
    },
    {
      "epoch": 0.2402180978981965,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001307302,
      "loss": 3.0916,
      "step": 346350
    },
    {
      "epoch": 0.2402367411343547,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001307282,
      "loss": 3.124,
      "step": 346360
    },
    {
      "epoch": 0.2402553843705129,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001307262,
      "loss": 3.1683,
      "step": 346370
    },
    {
      "epoch": 0.24027402760667108,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013072420000000002,
      "loss": 3.0319,
      "step": 346380
    },
    {
      "epoch": 0.24029267084282924,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013072220000000001,
      "loss": 3.0925,
      "step": 346390
    },
    {
      "epoch": 0.24031131407898743,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001307202,
      "loss": 3.1509,
      "step": 346400
    },
    {
      "epoch": 0.24032995731514561,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001307182,
      "loss": 3.2149,
      "step": 346410
    },
    {
      "epoch": 0.2403486005513038,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001307162,
      "loss": 3.1335,
      "step": 346420
    },
    {
      "epoch": 0.240367243787462,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001307142,
      "loss": 3.0063,
      "step": 346430
    },
    {
      "epoch": 0.24038588702362015,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001307122,
      "loss": 3.171,
      "step": 346440
    },
    {
      "epoch": 0.24040453025977834,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013071020000000001,
      "loss": 3.1678,
      "step": 346450
    },
    {
      "epoch": 0.24042317349593653,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001307082,
      "loss": 3.0369,
      "step": 346460
    },
    {
      "epoch": 0.24044181673209472,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001307062,
      "loss": 3.1965,
      "step": 346470
    },
    {
      "epoch": 0.2404604599682529,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001307042,
      "loss": 3.1252,
      "step": 346480
    },
    {
      "epoch": 0.24047910320441107,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001307022,
      "loss": 3.1668,
      "step": 346490
    },
    {
      "epoch": 0.24049774644056926,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001307002,
      "loss": 3.1372,
      "step": 346500
    },
    {
      "epoch": 0.24051638967672745,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013069820000000002,
      "loss": 3.2341,
      "step": 346510
    },
    {
      "epoch": 0.24053503291288564,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001306962,
      "loss": 3.068,
      "step": 346520
    },
    {
      "epoch": 0.24055367614904383,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001306942,
      "loss": 3.2046,
      "step": 346530
    },
    {
      "epoch": 0.240572319385202,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001306922,
      "loss": 3.1258,
      "step": 346540
    },
    {
      "epoch": 0.24059096262136018,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001306902,
      "loss": 3.0667,
      "step": 346550
    },
    {
      "epoch": 0.24060960585751837,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013068820000000002,
      "loss": 3.1464,
      "step": 346560
    },
    {
      "epoch": 0.24062824909367656,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013068620000000002,
      "loss": 3.1347,
      "step": 346570
    },
    {
      "epoch": 0.24064689232983474,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001306842,
      "loss": 3.1335,
      "step": 346580
    },
    {
      "epoch": 0.24066553556599293,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001306822,
      "loss": 3.1868,
      "step": 346590
    },
    {
      "epoch": 0.2406841788021511,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001306802,
      "loss": 2.9979,
      "step": 346600
    },
    {
      "epoch": 0.24070282203830928,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013067820000000003,
      "loss": 3.1884,
      "step": 346610
    },
    {
      "epoch": 0.24072146527446747,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001306762,
      "loss": 3.0768,
      "step": 346620
    },
    {
      "epoch": 0.24074010851062566,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013067420000000002,
      "loss": 3.088,
      "step": 346630
    },
    {
      "epoch": 0.24075875174678385,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001306722,
      "loss": 3.2913,
      "step": 346640
    },
    {
      "epoch": 0.240777394982942,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001306702,
      "loss": 3.1531,
      "step": 346650
    },
    {
      "epoch": 0.2407960382191002,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001306682,
      "loss": 3.1469,
      "step": 346660
    },
    {
      "epoch": 0.2408146814552584,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001306662,
      "loss": 3.1319,
      "step": 346670
    },
    {
      "epoch": 0.24083332469141658,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013066420000000002,
      "loss": 3.0993,
      "step": 346680
    },
    {
      "epoch": 0.24085196792757477,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001306622,
      "loss": 3.1755,
      "step": 346690
    },
    {
      "epoch": 0.24087061116373293,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001306602,
      "loss": 3.1825,
      "step": 346700
    },
    {
      "epoch": 0.24088925439989112,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001306582,
      "loss": 3.0934,
      "step": 346710
    },
    {
      "epoch": 0.2409078976360493,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001306562,
      "loss": 3.0321,
      "step": 346720
    },
    {
      "epoch": 0.2409265408722075,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013065420000000003,
      "loss": 3.1957,
      "step": 346730
    },
    {
      "epoch": 0.24094518410836568,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001306522,
      "loss": 3.1047,
      "step": 346740
    },
    {
      "epoch": 0.24096382734452385,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013065020000000002,
      "loss": 3.2171,
      "step": 346750
    },
    {
      "epoch": 0.24098247058068203,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001306482,
      "loss": 3.0851,
      "step": 346760
    },
    {
      "epoch": 0.24100111381684022,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001306462,
      "loss": 3.1308,
      "step": 346770
    },
    {
      "epoch": 0.2410197570529984,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001306442,
      "loss": 3.1738,
      "step": 346780
    },
    {
      "epoch": 0.2410384002891566,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001306422,
      "loss": 3.064,
      "step": 346790
    },
    {
      "epoch": 0.2410570435253148,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013064020000000002,
      "loss": 3.0874,
      "step": 346800
    },
    {
      "epoch": 0.24107568676147295,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001306382,
      "loss": 3.0515,
      "step": 346810
    },
    {
      "epoch": 0.24109432999763114,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013063620000000001,
      "loss": 3.143,
      "step": 346820
    },
    {
      "epoch": 0.24111297323378933,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013063419999999998,
      "loss": 3.193,
      "step": 346830
    },
    {
      "epoch": 0.24113161646994752,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001306322,
      "loss": 3.0074,
      "step": 346840
    },
    {
      "epoch": 0.2411502597061057,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001306302,
      "loss": 3.0945,
      "step": 346850
    },
    {
      "epoch": 0.24116890294226387,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001306282,
      "loss": 3.1813,
      "step": 346860
    },
    {
      "epoch": 0.24118754617842206,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013062620000000002,
      "loss": 3.151,
      "step": 346870
    },
    {
      "epoch": 0.24120618941458025,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001306242,
      "loss": 3.1929,
      "step": 346880
    },
    {
      "epoch": 0.24122483265073844,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001306222,
      "loss": 3.0332,
      "step": 346890
    },
    {
      "epoch": 0.24124347588689662,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001306202,
      "loss": 3.1819,
      "step": 346900
    },
    {
      "epoch": 0.24126211912305479,
      "grad_norm": 0.625,
      "learning_rate": 0.0001306182,
      "loss": 3.1209,
      "step": 346910
    },
    {
      "epoch": 0.24128076235921297,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001306162,
      "loss": 3.1263,
      "step": 346920
    },
    {
      "epoch": 0.24129940559537116,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001306142,
      "loss": 3.2009,
      "step": 346930
    },
    {
      "epoch": 0.24131804883152935,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013061220000000002,
      "loss": 3.1396,
      "step": 346940
    },
    {
      "epoch": 0.24133669206768754,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001306102,
      "loss": 3.0444,
      "step": 346950
    },
    {
      "epoch": 0.2413553353038457,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001306082,
      "loss": 3.1722,
      "step": 346960
    },
    {
      "epoch": 0.2413739785400039,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001306062,
      "loss": 3.233,
      "step": 346970
    },
    {
      "epoch": 0.24139262177616208,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001306042,
      "loss": 3.1517,
      "step": 346980
    },
    {
      "epoch": 0.24141126501232027,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001306022,
      "loss": 3.0918,
      "step": 346990
    },
    {
      "epoch": 0.24142990824847846,
      "grad_norm": 0.625,
      "learning_rate": 0.00013060020000000002,
      "loss": 3.1797,
      "step": 347000
    },
    {
      "epoch": 0.24144855148463662,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001305982,
      "loss": 3.1467,
      "step": 347010
    },
    {
      "epoch": 0.2414671947207948,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001305962,
      "loss": 3.0996,
      "step": 347020
    },
    {
      "epoch": 0.241485837956953,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001305942,
      "loss": 3.0097,
      "step": 347030
    },
    {
      "epoch": 0.2415044811931112,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001305922,
      "loss": 3.2133,
      "step": 347040
    },
    {
      "epoch": 0.24152312442926938,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001305902,
      "loss": 3.0409,
      "step": 347050
    },
    {
      "epoch": 0.24154176766542756,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013058820000000002,
      "loss": 3.1366,
      "step": 347060
    },
    {
      "epoch": 0.24156041090158573,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013058620000000001,
      "loss": 3.0184,
      "step": 347070
    },
    {
      "epoch": 0.24157905413774391,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001305842,
      "loss": 3.1782,
      "step": 347080
    },
    {
      "epoch": 0.2415976973739021,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001305822,
      "loss": 3.1499,
      "step": 347090
    },
    {
      "epoch": 0.2416163406100603,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001305802,
      "loss": 3.0893,
      "step": 347100
    },
    {
      "epoch": 0.24163498384621848,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001305782,
      "loss": 3.0784,
      "step": 347110
    },
    {
      "epoch": 0.24165362708237664,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013057620000000002,
      "loss": 3.1418,
      "step": 347120
    },
    {
      "epoch": 0.24167227031853483,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013057420000000001,
      "loss": 3.2013,
      "step": 347130
    },
    {
      "epoch": 0.24169091355469302,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001305722,
      "loss": 3.1351,
      "step": 347140
    },
    {
      "epoch": 0.2417095567908512,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001305702,
      "loss": 3.1308,
      "step": 347150
    },
    {
      "epoch": 0.2417282000270094,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001305682,
      "loss": 3.0402,
      "step": 347160
    },
    {
      "epoch": 0.24174684326316756,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013056620000000002,
      "loss": 3.1532,
      "step": 347170
    },
    {
      "epoch": 0.24176548649932575,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001305642,
      "loss": 3.1298,
      "step": 347180
    },
    {
      "epoch": 0.24178412973548394,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013056220000000002,
      "loss": 2.9804,
      "step": 347190
    },
    {
      "epoch": 0.24180277297164213,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001305602,
      "loss": 2.9883,
      "step": 347200
    },
    {
      "epoch": 0.24182141620780032,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001305582,
      "loss": 3.0907,
      "step": 347210
    },
    {
      "epoch": 0.24184005944395848,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001305562,
      "loss": 3.1485,
      "step": 347220
    },
    {
      "epoch": 0.24185870268011667,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001305542,
      "loss": 3.107,
      "step": 347230
    },
    {
      "epoch": 0.24187734591627486,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013055220000000002,
      "loss": 3.3728,
      "step": 347240
    },
    {
      "epoch": 0.24189598915243304,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001305502,
      "loss": 3.1621,
      "step": 347250
    },
    {
      "epoch": 0.24191463238859123,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001305482,
      "loss": 3.1629,
      "step": 347260
    },
    {
      "epoch": 0.24193327562474942,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001305462,
      "loss": 3.0951,
      "step": 347270
    },
    {
      "epoch": 0.24195191886090758,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001305442,
      "loss": 3.0962,
      "step": 347280
    },
    {
      "epoch": 0.24197056209706577,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013054220000000003,
      "loss": 3.0664,
      "step": 347290
    },
    {
      "epoch": 0.24198920533322396,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001305402,
      "loss": 3.0908,
      "step": 347300
    },
    {
      "epoch": 0.24200784856938215,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013053820000000002,
      "loss": 3.1729,
      "step": 347310
    },
    {
      "epoch": 0.24202649180554034,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013053619999999999,
      "loss": 3.0489,
      "step": 347320
    },
    {
      "epoch": 0.2420451350416985,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001305342,
      "loss": 3.1414,
      "step": 347330
    },
    {
      "epoch": 0.2420637782778567,
      "grad_norm": 0.625,
      "learning_rate": 0.0001305322,
      "loss": 3.2626,
      "step": 347340
    },
    {
      "epoch": 0.24208242151401488,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001305302,
      "loss": 3.0693,
      "step": 347350
    },
    {
      "epoch": 0.24210106475017307,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00013052820000000002,
      "loss": 3.1369,
      "step": 347360
    },
    {
      "epoch": 0.24211970798633126,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001305262,
      "loss": 2.9856,
      "step": 347370
    },
    {
      "epoch": 0.24213835122248942,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001305242,
      "loss": 3.1765,
      "step": 347380
    },
    {
      "epoch": 0.2421569944586476,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001305222,
      "loss": 3.1289,
      "step": 347390
    },
    {
      "epoch": 0.2421756376948058,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001305202,
      "loss": 3.1005,
      "step": 347400
    },
    {
      "epoch": 0.24219428093096398,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001305182,
      "loss": 3.1653,
      "step": 347410
    },
    {
      "epoch": 0.24221292416712217,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001305162,
      "loss": 3.1951,
      "step": 347420
    },
    {
      "epoch": 0.24223156740328033,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013051420000000002,
      "loss": 3.0739,
      "step": 347430
    },
    {
      "epoch": 0.24225021063943852,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001305122,
      "loss": 3.1853,
      "step": 347440
    },
    {
      "epoch": 0.2422688538755967,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001305102,
      "loss": 3.1505,
      "step": 347450
    },
    {
      "epoch": 0.2422874971117549,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001305082,
      "loss": 3.0684,
      "step": 347460
    },
    {
      "epoch": 0.2423061403479131,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001305062,
      "loss": 3.0941,
      "step": 347470
    },
    {
      "epoch": 0.24232478358407128,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001305042,
      "loss": 3.0683,
      "step": 347480
    },
    {
      "epoch": 0.24234342682022944,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001305022,
      "loss": 3.0618,
      "step": 347490
    },
    {
      "epoch": 0.24236207005638763,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013050020000000001,
      "loss": 3.0307,
      "step": 347500
    },
    {
      "epoch": 0.24238071329254582,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001304982,
      "loss": 3.1361,
      "step": 347510
    },
    {
      "epoch": 0.242399356528704,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001304962,
      "loss": 3.135,
      "step": 347520
    },
    {
      "epoch": 0.2424179997648622,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001304942,
      "loss": 2.981,
      "step": 347530
    },
    {
      "epoch": 0.24243664300102036,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001304922,
      "loss": 3.1898,
      "step": 347540
    },
    {
      "epoch": 0.24245528623717855,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013049020000000002,
      "loss": 3.2083,
      "step": 347550
    },
    {
      "epoch": 0.24247392947333674,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013048820000000002,
      "loss": 3.0854,
      "step": 347560
    },
    {
      "epoch": 0.24249257270949492,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001304862,
      "loss": 3.1122,
      "step": 347570
    },
    {
      "epoch": 0.2425112159456531,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001304842,
      "loss": 3.1877,
      "step": 347580
    },
    {
      "epoch": 0.24252985918181127,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001304822,
      "loss": 3.1879,
      "step": 347590
    },
    {
      "epoch": 0.24254850241796946,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001304802,
      "loss": 3.0298,
      "step": 347600
    },
    {
      "epoch": 0.24256714565412765,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013047820000000002,
      "loss": 3.1457,
      "step": 347610
    },
    {
      "epoch": 0.24258578889028584,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013047620000000002,
      "loss": 3.2559,
      "step": 347620
    },
    {
      "epoch": 0.24260443212644403,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001304742,
      "loss": 3.1412,
      "step": 347630
    },
    {
      "epoch": 0.2426230753626022,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001304722,
      "loss": 3.2004,
      "step": 347640
    },
    {
      "epoch": 0.24264171859876038,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001304702,
      "loss": 3.1097,
      "step": 347650
    },
    {
      "epoch": 0.24266036183491857,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001304682,
      "loss": 3.1094,
      "step": 347660
    },
    {
      "epoch": 0.24267900507107676,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001304662,
      "loss": 2.9832,
      "step": 347670
    },
    {
      "epoch": 0.24269764830723495,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013046420000000002,
      "loss": 3.1232,
      "step": 347680
    },
    {
      "epoch": 0.2427162915433931,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001304622,
      "loss": 2.9785,
      "step": 347690
    },
    {
      "epoch": 0.2427349347795513,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001304602,
      "loss": 3.1631,
      "step": 347700
    },
    {
      "epoch": 0.2427535780157095,
      "grad_norm": 0.75,
      "learning_rate": 0.0001304582,
      "loss": 3.0927,
      "step": 347710
    },
    {
      "epoch": 0.24277222125186768,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001304562,
      "loss": 3.1715,
      "step": 347720
    },
    {
      "epoch": 0.24279086448802586,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013045420000000002,
      "loss": 3.2641,
      "step": 347730
    },
    {
      "epoch": 0.24280950772418405,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001304522,
      "loss": 3.0224,
      "step": 347740
    },
    {
      "epoch": 0.24282815096034222,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001304502,
      "loss": 3.1826,
      "step": 347750
    },
    {
      "epoch": 0.2428467941965004,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001304482,
      "loss": 3.1956,
      "step": 347760
    },
    {
      "epoch": 0.2428654374326586,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001304462,
      "loss": 3.2013,
      "step": 347770
    },
    {
      "epoch": 0.24288408066881678,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013044420000000003,
      "loss": 3.1957,
      "step": 347780
    },
    {
      "epoch": 0.24290272390497497,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001304422,
      "loss": 3.1548,
      "step": 347790
    },
    {
      "epoch": 0.24292136714113313,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013044020000000002,
      "loss": 3.2116,
      "step": 347800
    },
    {
      "epoch": 0.24294001037729132,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001304382,
      "loss": 3.1335,
      "step": 347810
    },
    {
      "epoch": 0.2429586536134495,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001304362,
      "loss": 3.0617,
      "step": 347820
    },
    {
      "epoch": 0.2429772968496077,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001304342,
      "loss": 3.2071,
      "step": 347830
    },
    {
      "epoch": 0.2429959400857659,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001304322,
      "loss": 3.167,
      "step": 347840
    },
    {
      "epoch": 0.24301458332192405,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013043020000000002,
      "loss": 3.1602,
      "step": 347850
    },
    {
      "epoch": 0.24303322655808224,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001304282,
      "loss": 3.0149,
      "step": 347860
    },
    {
      "epoch": 0.24305186979424043,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013042620000000002,
      "loss": 3.2344,
      "step": 347870
    },
    {
      "epoch": 0.24307051303039862,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013042419999999998,
      "loss": 3.1572,
      "step": 347880
    },
    {
      "epoch": 0.2430891562665568,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001304222,
      "loss": 3.1071,
      "step": 347890
    },
    {
      "epoch": 0.24310779950271497,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001304202,
      "loss": 3.1414,
      "step": 347900
    },
    {
      "epoch": 0.24312644273887316,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001304182,
      "loss": 3.1059,
      "step": 347910
    },
    {
      "epoch": 0.24314508597503134,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013041620000000002,
      "loss": 3.1707,
      "step": 347920
    },
    {
      "epoch": 0.24316372921118953,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001304142,
      "loss": 3.0612,
      "step": 347930
    },
    {
      "epoch": 0.24318237244734772,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001304122,
      "loss": 3.0907,
      "step": 347940
    },
    {
      "epoch": 0.2432010156835059,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001304102,
      "loss": 3.2119,
      "step": 347950
    },
    {
      "epoch": 0.24321965891966407,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001304082,
      "loss": 3.1448,
      "step": 347960
    },
    {
      "epoch": 0.24323830215582226,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001304062,
      "loss": 3.0661,
      "step": 347970
    },
    {
      "epoch": 0.24325694539198045,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001304042,
      "loss": 3.19,
      "step": 347980
    },
    {
      "epoch": 0.24327558862813864,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013040220000000002,
      "loss": 2.9685,
      "step": 347990
    },
    {
      "epoch": 0.24329423186429683,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001304002,
      "loss": 3.0192,
      "step": 348000
    },
    {
      "epoch": 0.243312875100455,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001303982,
      "loss": 3.1906,
      "step": 348010
    },
    {
      "epoch": 0.24333151833661318,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001303962,
      "loss": 3.159,
      "step": 348020
    },
    {
      "epoch": 0.24335016157277137,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001303942,
      "loss": 3.1895,
      "step": 348030
    },
    {
      "epoch": 0.24336880480892956,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013039220000000002,
      "loss": 3.0737,
      "step": 348040
    },
    {
      "epoch": 0.24338744804508775,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001303902,
      "loss": 3.0253,
      "step": 348050
    },
    {
      "epoch": 0.2434060912812459,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001303882,
      "loss": 3.0877,
      "step": 348060
    },
    {
      "epoch": 0.2434247345174041,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001303862,
      "loss": 3.1446,
      "step": 348070
    },
    {
      "epoch": 0.24344337775356228,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001303842,
      "loss": 3.3698,
      "step": 348080
    },
    {
      "epoch": 0.24346202098972047,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001303822,
      "loss": 3.15,
      "step": 348090
    },
    {
      "epoch": 0.24348066422587866,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001303802,
      "loss": 3.1643,
      "step": 348100
    },
    {
      "epoch": 0.24349930746203682,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013037820000000002,
      "loss": 3.1371,
      "step": 348110
    },
    {
      "epoch": 0.243517950698195,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013037620000000001,
      "loss": 3.1449,
      "step": 348120
    },
    {
      "epoch": 0.2435365939343532,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001303742,
      "loss": 3.1499,
      "step": 348130
    },
    {
      "epoch": 0.2435552371705114,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001303722,
      "loss": 3.1434,
      "step": 348140
    },
    {
      "epoch": 0.24357388040666958,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001303702,
      "loss": 3.0513,
      "step": 348150
    },
    {
      "epoch": 0.24359252364282777,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001303682,
      "loss": 3.0807,
      "step": 348160
    },
    {
      "epoch": 0.24361116687898593,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013036620000000002,
      "loss": 3.0863,
      "step": 348170
    },
    {
      "epoch": 0.24362981011514412,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013036420000000001,
      "loss": 3.2058,
      "step": 348180
    },
    {
      "epoch": 0.2436484533513023,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001303622,
      "loss": 3.1644,
      "step": 348190
    },
    {
      "epoch": 0.2436670965874605,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001303602,
      "loss": 3.046,
      "step": 348200
    },
    {
      "epoch": 0.24368573982361869,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001303582,
      "loss": 3.0624,
      "step": 348210
    },
    {
      "epoch": 0.24370438305977685,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013035620000000002,
      "loss": 3.1207,
      "step": 348220
    },
    {
      "epoch": 0.24372302629593504,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001303542,
      "loss": 3.0416,
      "step": 348230
    },
    {
      "epoch": 0.24374166953209322,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013035220000000002,
      "loss": 3.037,
      "step": 348240
    },
    {
      "epoch": 0.2437603127682514,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001303502,
      "loss": 3.1836,
      "step": 348250
    },
    {
      "epoch": 0.2437789560044096,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001303482,
      "loss": 3.2338,
      "step": 348260
    },
    {
      "epoch": 0.24379759924056776,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001303462,
      "loss": 3.1839,
      "step": 348270
    },
    {
      "epoch": 0.24381624247672595,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001303442,
      "loss": 2.9741,
      "step": 348280
    },
    {
      "epoch": 0.24383488571288414,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013034220000000002,
      "loss": 3.1044,
      "step": 348290
    },
    {
      "epoch": 0.24385352894904233,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001303402,
      "loss": 3.0936,
      "step": 348300
    },
    {
      "epoch": 0.24387217218520052,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001303382,
      "loss": 3.1055,
      "step": 348310
    },
    {
      "epoch": 0.24389081542135868,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001303362,
      "loss": 3.0878,
      "step": 348320
    },
    {
      "epoch": 0.24390945865751687,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001303342,
      "loss": 3.1346,
      "step": 348330
    },
    {
      "epoch": 0.24392810189367506,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013033220000000003,
      "loss": 3.0282,
      "step": 348340
    },
    {
      "epoch": 0.24394674512983325,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001303302,
      "loss": 3.1523,
      "step": 348350
    },
    {
      "epoch": 0.24396538836599144,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00013032820000000002,
      "loss": 3.048,
      "step": 348360
    },
    {
      "epoch": 0.2439840316021496,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013032619999999999,
      "loss": 3.099,
      "step": 348370
    },
    {
      "epoch": 0.2440026748383078,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001303242,
      "loss": 3.1568,
      "step": 348380
    },
    {
      "epoch": 0.24402131807446598,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001303222,
      "loss": 3.0961,
      "step": 348390
    },
    {
      "epoch": 0.24403996131062416,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001303202,
      "loss": 3.0961,
      "step": 348400
    },
    {
      "epoch": 0.24405860454678235,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013031820000000002,
      "loss": 3.1518,
      "step": 348410
    },
    {
      "epoch": 0.24407724778294054,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001303162,
      "loss": 3.2676,
      "step": 348420
    },
    {
      "epoch": 0.2440958910190987,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001303142,
      "loss": 3.1122,
      "step": 348430
    },
    {
      "epoch": 0.2441145342552569,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013031219999999998,
      "loss": 3.1112,
      "step": 348440
    },
    {
      "epoch": 0.24413317749141508,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001303102,
      "loss": 3.0683,
      "step": 348450
    },
    {
      "epoch": 0.24415182072757327,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001303082,
      "loss": 3.0322,
      "step": 348460
    },
    {
      "epoch": 0.24417046396373146,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001303062,
      "loss": 3.1991,
      "step": 348470
    },
    {
      "epoch": 0.24418910719988962,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013030420000000002,
      "loss": 3.1931,
      "step": 348480
    },
    {
      "epoch": 0.2442077504360478,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001303022,
      "loss": 3.2486,
      "step": 348490
    },
    {
      "epoch": 0.244226393672206,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001303002,
      "loss": 3.1389,
      "step": 348500
    },
    {
      "epoch": 0.2442450369083642,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001302982,
      "loss": 3.0187,
      "step": 348510
    },
    {
      "epoch": 0.24426368014452238,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001302962,
      "loss": 3.2356,
      "step": 348520
    },
    {
      "epoch": 0.24428232338068054,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00013029420000000002,
      "loss": 3.2167,
      "step": 348530
    },
    {
      "epoch": 0.24430096661683873,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001302922,
      "loss": 2.9749,
      "step": 348540
    },
    {
      "epoch": 0.24431960985299692,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013029020000000002,
      "loss": 3.0765,
      "step": 348550
    },
    {
      "epoch": 0.2443382530891551,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001302882,
      "loss": 3.1845,
      "step": 348560
    },
    {
      "epoch": 0.2443568963253133,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001302862,
      "loss": 3.2313,
      "step": 348570
    },
    {
      "epoch": 0.24437553956147146,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001302842,
      "loss": 3.1204,
      "step": 348580
    },
    {
      "epoch": 0.24439418279762964,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001302822,
      "loss": 3.2556,
      "step": 348590
    },
    {
      "epoch": 0.24441282603378783,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013028020000000002,
      "loss": 3.0536,
      "step": 348600
    },
    {
      "epoch": 0.24443146926994602,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013027820000000002,
      "loss": 3.0931,
      "step": 348610
    },
    {
      "epoch": 0.2444501125061042,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001302762,
      "loss": 3.088,
      "step": 348620
    },
    {
      "epoch": 0.2444687557422624,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001302742,
      "loss": 3.2052,
      "step": 348630
    },
    {
      "epoch": 0.24448739897842056,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001302722,
      "loss": 3.092,
      "step": 348640
    },
    {
      "epoch": 0.24450604221457875,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001302702,
      "loss": 3.1609,
      "step": 348650
    },
    {
      "epoch": 0.24452468545073694,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001302682,
      "loss": 3.1442,
      "step": 348660
    },
    {
      "epoch": 0.24454332868689513,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013026620000000002,
      "loss": 3.1624,
      "step": 348670
    },
    {
      "epoch": 0.24456197192305332,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001302642,
      "loss": 3.0737,
      "step": 348680
    },
    {
      "epoch": 0.24458061515921148,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001302622,
      "loss": 3.0875,
      "step": 348690
    },
    {
      "epoch": 0.24459925839536967,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001302602,
      "loss": 3.0999,
      "step": 348700
    },
    {
      "epoch": 0.24461790163152786,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001302582,
      "loss": 3.094,
      "step": 348710
    },
    {
      "epoch": 0.24463654486768605,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001302562,
      "loss": 3.0765,
      "step": 348720
    },
    {
      "epoch": 0.24465518810384423,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013025420000000002,
      "loss": 3.0701,
      "step": 348730
    },
    {
      "epoch": 0.2446738313400024,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001302522,
      "loss": 2.9847,
      "step": 348740
    },
    {
      "epoch": 0.24469247457616058,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001302502,
      "loss": 3.1203,
      "step": 348750
    },
    {
      "epoch": 0.24471111781231877,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001302482,
      "loss": 3.0384,
      "step": 348760
    },
    {
      "epoch": 0.24472976104847696,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001302462,
      "loss": 3.055,
      "step": 348770
    },
    {
      "epoch": 0.24474840428463515,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013024420000000002,
      "loss": 3.118,
      "step": 348780
    },
    {
      "epoch": 0.2447670475207933,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001302422,
      "loss": 3.0638,
      "step": 348790
    },
    {
      "epoch": 0.2447856907569515,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013024020000000001,
      "loss": 3.0202,
      "step": 348800
    },
    {
      "epoch": 0.2448043339931097,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001302382,
      "loss": 3.1126,
      "step": 348810
    },
    {
      "epoch": 0.24482297722926788,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001302362,
      "loss": 3.0232,
      "step": 348820
    },
    {
      "epoch": 0.24484162046542607,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013023420000000003,
      "loss": 3.1983,
      "step": 348830
    },
    {
      "epoch": 0.24486026370158426,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001302322,
      "loss": 3.1886,
      "step": 348840
    },
    {
      "epoch": 0.24487890693774242,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013023020000000002,
      "loss": 2.9402,
      "step": 348850
    },
    {
      "epoch": 0.2448975501739006,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001302282,
      "loss": 3.183,
      "step": 348860
    },
    {
      "epoch": 0.2449161934100588,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001302262,
      "loss": 2.9568,
      "step": 348870
    },
    {
      "epoch": 0.24493483664621699,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001302242,
      "loss": 3.1192,
      "step": 348880
    },
    {
      "epoch": 0.24495347988237517,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001302222,
      "loss": 3.1443,
      "step": 348890
    },
    {
      "epoch": 0.24497212311853334,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013022020000000002,
      "loss": 3.0513,
      "step": 348900
    },
    {
      "epoch": 0.24499076635469152,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001302182,
      "loss": 3.0735,
      "step": 348910
    },
    {
      "epoch": 0.2450094095908497,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013021620000000002,
      "loss": 3.1655,
      "step": 348920
    },
    {
      "epoch": 0.2450280528270079,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013021419999999998,
      "loss": 3.1643,
      "step": 348930
    },
    {
      "epoch": 0.2450466960631661,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001302122,
      "loss": 3.1227,
      "step": 348940
    },
    {
      "epoch": 0.24506533929932425,
      "grad_norm": 0.75,
      "learning_rate": 0.0001302102,
      "loss": 3.1973,
      "step": 348950
    },
    {
      "epoch": 0.24508398253548244,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001302082,
      "loss": 3.1226,
      "step": 348960
    },
    {
      "epoch": 0.24510262577164063,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013020620000000002,
      "loss": 3.2317,
      "step": 348970
    },
    {
      "epoch": 0.24512126900779882,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001302042,
      "loss": 3.2191,
      "step": 348980
    },
    {
      "epoch": 0.245139912243957,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001302022,
      "loss": 3.0902,
      "step": 348990
    },
    {
      "epoch": 0.24515855548011517,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001302002,
      "loss": 3.086,
      "step": 349000
    },
    {
      "epoch": 0.24517719871627336,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001301982,
      "loss": 3.2478,
      "step": 349010
    },
    {
      "epoch": 0.24519584195243155,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013019620000000003,
      "loss": 3.2,
      "step": 349020
    },
    {
      "epoch": 0.24521448518858974,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001301942,
      "loss": 3.1543,
      "step": 349030
    },
    {
      "epoch": 0.24523312842474793,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013019220000000002,
      "loss": 3.1414,
      "step": 349040
    },
    {
      "epoch": 0.2452517716609061,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00013019019999999999,
      "loss": 3.048,
      "step": 349050
    },
    {
      "epoch": 0.24527041489706428,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001301882,
      "loss": 3.1084,
      "step": 349060
    },
    {
      "epoch": 0.24528905813322247,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001301862,
      "loss": 3.1045,
      "step": 349070
    },
    {
      "epoch": 0.24530770136938065,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001301842,
      "loss": 3.0855,
      "step": 349080
    },
    {
      "epoch": 0.24532634460553884,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013018220000000002,
      "loss": 3.1711,
      "step": 349090
    },
    {
      "epoch": 0.24534498784169703,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001301802,
      "loss": 2.9995,
      "step": 349100
    },
    {
      "epoch": 0.2453636310778552,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001301782,
      "loss": 3.0866,
      "step": 349110
    },
    {
      "epoch": 0.24538227431401338,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001301762,
      "loss": 3.0503,
      "step": 349120
    },
    {
      "epoch": 0.24540091755017157,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001301742,
      "loss": 3.2644,
      "step": 349130
    },
    {
      "epoch": 0.24541956078632976,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001301722,
      "loss": 3.1031,
      "step": 349140
    },
    {
      "epoch": 0.24543820402248795,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001301702,
      "loss": 3.1756,
      "step": 349150
    },
    {
      "epoch": 0.2454568472586461,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013016820000000002,
      "loss": 3.0617,
      "step": 349160
    },
    {
      "epoch": 0.2454754904948043,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013016620000000001,
      "loss": 3.0135,
      "step": 349170
    },
    {
      "epoch": 0.2454941337309625,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001301642,
      "loss": 3.1909,
      "step": 349180
    },
    {
      "epoch": 0.24551277696712068,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001301622,
      "loss": 2.9899,
      "step": 349190
    },
    {
      "epoch": 0.24553142020327887,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001301602,
      "loss": 3.0304,
      "step": 349200
    },
    {
      "epoch": 0.24555006343943703,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001301582,
      "loss": 3.2033,
      "step": 349210
    },
    {
      "epoch": 0.24556870667559522,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013015620000000002,
      "loss": 3.085,
      "step": 349220
    },
    {
      "epoch": 0.2455873499117534,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013015420000000002,
      "loss": 3.1527,
      "step": 349230
    },
    {
      "epoch": 0.2456059931479116,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001301522,
      "loss": 3.0135,
      "step": 349240
    },
    {
      "epoch": 0.24562463638406978,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001301502,
      "loss": 3.1098,
      "step": 349250
    },
    {
      "epoch": 0.24564327962022794,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001301482,
      "loss": 3.1906,
      "step": 349260
    },
    {
      "epoch": 0.24566192285638613,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001301462,
      "loss": 3.2135,
      "step": 349270
    },
    {
      "epoch": 0.24568056609254432,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001301442,
      "loss": 2.9633,
      "step": 349280
    },
    {
      "epoch": 0.2456992093287025,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013014220000000002,
      "loss": 3.0484,
      "step": 349290
    },
    {
      "epoch": 0.2457178525648607,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001301402,
      "loss": 3.0856,
      "step": 349300
    },
    {
      "epoch": 0.2457364958010189,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001301382,
      "loss": 3.2286,
      "step": 349310
    },
    {
      "epoch": 0.24575513903717705,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001301362,
      "loss": 3.0862,
      "step": 349320
    },
    {
      "epoch": 0.24577378227333524,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001301342,
      "loss": 3.1446,
      "step": 349330
    },
    {
      "epoch": 0.24579242550949343,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013013220000000002,
      "loss": 3.2093,
      "step": 349340
    },
    {
      "epoch": 0.24581106874565162,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001301302,
      "loss": 3.0272,
      "step": 349350
    },
    {
      "epoch": 0.2458297119818098,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001301282,
      "loss": 3.0834,
      "step": 349360
    },
    {
      "epoch": 0.24584835521796797,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001301262,
      "loss": 2.9675,
      "step": 349370
    },
    {
      "epoch": 0.24586699845412616,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001301242,
      "loss": 3.0365,
      "step": 349380
    },
    {
      "epoch": 0.24588564169028435,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013012220000000003,
      "loss": 3.1462,
      "step": 349390
    },
    {
      "epoch": 0.24590428492644253,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001301202,
      "loss": 3.1408,
      "step": 349400
    },
    {
      "epoch": 0.24592292816260072,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013011820000000002,
      "loss": 2.984,
      "step": 349410
    },
    {
      "epoch": 0.24594157139875888,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013011619999999999,
      "loss": 3.1863,
      "step": 349420
    },
    {
      "epoch": 0.24596021463491707,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001301142,
      "loss": 3.0601,
      "step": 349430
    },
    {
      "epoch": 0.24597885787107526,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001301122,
      "loss": 3.0223,
      "step": 349440
    },
    {
      "epoch": 0.24599750110723345,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001301102,
      "loss": 3.2187,
      "step": 349450
    },
    {
      "epoch": 0.24601614434339164,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013010820000000002,
      "loss": 3.0511,
      "step": 349460
    },
    {
      "epoch": 0.2460347875795498,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001301062,
      "loss": 3.1995,
      "step": 349470
    },
    {
      "epoch": 0.246053430815708,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013010420000000001,
      "loss": 3.1927,
      "step": 349480
    },
    {
      "epoch": 0.24607207405186618,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001301022,
      "loss": 2.9885,
      "step": 349490
    },
    {
      "epoch": 0.24609071728802437,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001301002,
      "loss": 3.1258,
      "step": 349500
    },
    {
      "epoch": 0.24610936052418256,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013009820000000003,
      "loss": 2.9432,
      "step": 349510
    },
    {
      "epoch": 0.24612800376034075,
      "grad_norm": 0.75,
      "learning_rate": 0.0001300962,
      "loss": 3.1207,
      "step": 349520
    },
    {
      "epoch": 0.2461466469964989,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013009420000000002,
      "loss": 3.1011,
      "step": 349530
    },
    {
      "epoch": 0.2461652902326571,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001300922,
      "loss": 3.2379,
      "step": 349540
    },
    {
      "epoch": 0.24618393346881529,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001300902,
      "loss": 3.0788,
      "step": 349550
    },
    {
      "epoch": 0.24620257670497347,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001300882,
      "loss": 2.8891,
      "step": 349560
    },
    {
      "epoch": 0.24622121994113166,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001300862,
      "loss": 3.1536,
      "step": 349570
    },
    {
      "epoch": 0.24623986317728982,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013008420000000002,
      "loss": 3.1736,
      "step": 349580
    },
    {
      "epoch": 0.24625850641344801,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001300822,
      "loss": 3.1198,
      "step": 349590
    },
    {
      "epoch": 0.2462771496496062,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013008020000000002,
      "loss": 3.1782,
      "step": 349600
    },
    {
      "epoch": 0.2462957928857644,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001300782,
      "loss": 3.2099,
      "step": 349610
    },
    {
      "epoch": 0.24631443612192258,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001300762,
      "loss": 3.0491,
      "step": 349620
    },
    {
      "epoch": 0.24633307935808074,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001300742,
      "loss": 3.0786,
      "step": 349630
    },
    {
      "epoch": 0.24635172259423893,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001300722,
      "loss": 3.1701,
      "step": 349640
    },
    {
      "epoch": 0.24637036583039712,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013007020000000002,
      "loss": 3.1016,
      "step": 349650
    },
    {
      "epoch": 0.2463890090665553,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001300682,
      "loss": 3.129,
      "step": 349660
    },
    {
      "epoch": 0.2464076523027135,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001300662,
      "loss": 3.1467,
      "step": 349670
    },
    {
      "epoch": 0.24642629553887166,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001300642,
      "loss": 3.2222,
      "step": 349680
    },
    {
      "epoch": 0.24644493877502985,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001300622,
      "loss": 3.0984,
      "step": 349690
    },
    {
      "epoch": 0.24646358201118804,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001300602,
      "loss": 3.109,
      "step": 349700
    },
    {
      "epoch": 0.24648222524734623,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001300582,
      "loss": 3.0315,
      "step": 349710
    },
    {
      "epoch": 0.24650086848350442,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013005620000000002,
      "loss": 3.1089,
      "step": 349720
    },
    {
      "epoch": 0.24651951171966258,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001300542,
      "loss": 3.1697,
      "step": 349730
    },
    {
      "epoch": 0.24653815495582077,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001300522,
      "loss": 3.1536,
      "step": 349740
    },
    {
      "epoch": 0.24655679819197895,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001300502,
      "loss": 3.0436,
      "step": 349750
    },
    {
      "epoch": 0.24657544142813714,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001300482,
      "loss": 3.0225,
      "step": 349760
    },
    {
      "epoch": 0.24659408466429533,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001300462,
      "loss": 3.163,
      "step": 349770
    },
    {
      "epoch": 0.24661272790045352,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013004420000000002,
      "loss": 3.2076,
      "step": 349780
    },
    {
      "epoch": 0.24663137113661168,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001300422,
      "loss": 2.9882,
      "step": 349790
    },
    {
      "epoch": 0.24665001437276987,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001300402,
      "loss": 3.0813,
      "step": 349800
    },
    {
      "epoch": 0.24666865760892806,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001300382,
      "loss": 3.0401,
      "step": 349810
    },
    {
      "epoch": 0.24668730084508625,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001300362,
      "loss": 3.1718,
      "step": 349820
    },
    {
      "epoch": 0.24670594408124444,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00013003420000000002,
      "loss": 3.1635,
      "step": 349830
    },
    {
      "epoch": 0.2467245873174026,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001300322,
      "loss": 3.1534,
      "step": 349840
    },
    {
      "epoch": 0.2467432305535608,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013003020000000001,
      "loss": 3.1955,
      "step": 349850
    },
    {
      "epoch": 0.24676187378971898,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001300282,
      "loss": 3.0803,
      "step": 349860
    },
    {
      "epoch": 0.24678051702587717,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001300262,
      "loss": 3.0293,
      "step": 349870
    },
    {
      "epoch": 0.24679916026203536,
      "grad_norm": 0.75,
      "learning_rate": 0.0001300242,
      "loss": 3.2219,
      "step": 349880
    },
    {
      "epoch": 0.24681780349819352,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001300222,
      "loss": 3.174,
      "step": 349890
    },
    {
      "epoch": 0.2468364467343517,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013002020000000002,
      "loss": 3.1781,
      "step": 349900
    },
    {
      "epoch": 0.2468550899705099,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001300182,
      "loss": 3.2035,
      "step": 349910
    },
    {
      "epoch": 0.24687373320666808,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001300162,
      "loss": 3.1128,
      "step": 349920
    },
    {
      "epoch": 0.24689237644282627,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001300142,
      "loss": 3.0334,
      "step": 349930
    },
    {
      "epoch": 0.24691101967898443,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001300122,
      "loss": 3.0625,
      "step": 349940
    },
    {
      "epoch": 0.24692966291514262,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00013001020000000002,
      "loss": 2.9763,
      "step": 349950
    },
    {
      "epoch": 0.2469483061513008,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001300082,
      "loss": 3.1336,
      "step": 349960
    },
    {
      "epoch": 0.246966949387459,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00013000620000000002,
      "loss": 3.1828,
      "step": 349970
    },
    {
      "epoch": 0.2469855926236172,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001300042,
      "loss": 3.0435,
      "step": 349980
    },
    {
      "epoch": 0.24700423585977538,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001300022,
      "loss": 3.0854,
      "step": 349990
    },
    {
      "epoch": 0.24702287909593354,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00013000020000000003,
      "loss": 3.1464,
      "step": 350000
    },
    {
      "epoch": 0.24704152233209173,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001299982,
      "loss": 3.0556,
      "step": 350010
    },
    {
      "epoch": 0.24706016556824992,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012999620000000002,
      "loss": 3.0552,
      "step": 350020
    },
    {
      "epoch": 0.2470788088044081,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001299942,
      "loss": 3.0328,
      "step": 350030
    },
    {
      "epoch": 0.2470974520405663,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001299922,
      "loss": 3.1213,
      "step": 350040
    },
    {
      "epoch": 0.24711609527672446,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001299902,
      "loss": 3.2998,
      "step": 350050
    },
    {
      "epoch": 0.24713473851288265,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001299882,
      "loss": 3.1162,
      "step": 350060
    },
    {
      "epoch": 0.24715338174904083,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012998620000000003,
      "loss": 3.1392,
      "step": 350070
    },
    {
      "epoch": 0.24717202498519902,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001299842,
      "loss": 3.0973,
      "step": 350080
    },
    {
      "epoch": 0.2471906682213572,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012998220000000002,
      "loss": 3.1717,
      "step": 350090
    },
    {
      "epoch": 0.24720931145751537,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012998019999999999,
      "loss": 3.1856,
      "step": 350100
    },
    {
      "epoch": 0.24722795469367356,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001299782,
      "loss": 3.0636,
      "step": 350110
    },
    {
      "epoch": 0.24724659792983175,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001299762,
      "loss": 3.1402,
      "step": 350120
    },
    {
      "epoch": 0.24726524116598994,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001299742,
      "loss": 3.1336,
      "step": 350130
    },
    {
      "epoch": 0.24728388440214813,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012997220000000002,
      "loss": 3.0463,
      "step": 350140
    },
    {
      "epoch": 0.2473025276383063,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001299702,
      "loss": 3.1985,
      "step": 350150
    },
    {
      "epoch": 0.24732117087446448,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012996820000000001,
      "loss": 3.109,
      "step": 350160
    },
    {
      "epoch": 0.24733981411062267,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001299662,
      "loss": 3.1917,
      "step": 350170
    },
    {
      "epoch": 0.24735845734678086,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001299642,
      "loss": 3.0046,
      "step": 350180
    },
    {
      "epoch": 0.24737710058293905,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001299622,
      "loss": 3.06,
      "step": 350190
    },
    {
      "epoch": 0.24739574381909724,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001299602,
      "loss": 3.2174,
      "step": 350200
    },
    {
      "epoch": 0.2474143870552554,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012995820000000002,
      "loss": 3.1737,
      "step": 350210
    },
    {
      "epoch": 0.24743303029141359,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012995620000000001,
      "loss": 3.1866,
      "step": 350220
    },
    {
      "epoch": 0.24745167352757177,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001299542,
      "loss": 3.2453,
      "step": 350230
    },
    {
      "epoch": 0.24747031676372996,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001299522,
      "loss": 3.2143,
      "step": 350240
    },
    {
      "epoch": 0.24748895999988815,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001299502,
      "loss": 2.9961,
      "step": 350250
    },
    {
      "epoch": 0.24750760323604631,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001299482,
      "loss": 3.2138,
      "step": 350260
    },
    {
      "epoch": 0.2475262464722045,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001299462,
      "loss": 3.1457,
      "step": 350270
    },
    {
      "epoch": 0.2475448897083627,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012994420000000002,
      "loss": 3.0774,
      "step": 350280
    },
    {
      "epoch": 0.24756353294452088,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001299422,
      "loss": 3.1904,
      "step": 350290
    },
    {
      "epoch": 0.24758217618067907,
      "grad_norm": 0.75,
      "learning_rate": 0.0001299402,
      "loss": 3.0212,
      "step": 350300
    },
    {
      "epoch": 0.24760081941683723,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001299382,
      "loss": 3.1108,
      "step": 350310
    },
    {
      "epoch": 0.24761946265299542,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001299362,
      "loss": 3.1396,
      "step": 350320
    },
    {
      "epoch": 0.2476381058891536,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001299342,
      "loss": 3.1199,
      "step": 350330
    },
    {
      "epoch": 0.2476567491253118,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012993220000000002,
      "loss": 3.1542,
      "step": 350340
    },
    {
      "epoch": 0.24767539236147,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001299302,
      "loss": 3.2041,
      "step": 350350
    },
    {
      "epoch": 0.24769403559762815,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001299282,
      "loss": 3.0691,
      "step": 350360
    },
    {
      "epoch": 0.24771267883378634,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001299262,
      "loss": 3.1387,
      "step": 350370
    },
    {
      "epoch": 0.24773132206994453,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001299242,
      "loss": 3.0474,
      "step": 350380
    },
    {
      "epoch": 0.24774996530610272,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012992220000000002,
      "loss": 3.057,
      "step": 350390
    },
    {
      "epoch": 0.2477686085422609,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001299202,
      "loss": 3.2257,
      "step": 350400
    },
    {
      "epoch": 0.24778725177841907,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001299182,
      "loss": 3.2025,
      "step": 350410
    },
    {
      "epoch": 0.24780589501457725,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001299162,
      "loss": 3.1322,
      "step": 350420
    },
    {
      "epoch": 0.24782453825073544,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001299142,
      "loss": 3.2092,
      "step": 350430
    },
    {
      "epoch": 0.24784318148689363,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012991220000000003,
      "loss": 3.2042,
      "step": 350440
    },
    {
      "epoch": 0.24786182472305182,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001299102,
      "loss": 3.0591,
      "step": 350450
    },
    {
      "epoch": 0.24788046795921,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012990820000000002,
      "loss": 3.0074,
      "step": 350460
    },
    {
      "epoch": 0.24789911119536817,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001299062,
      "loss": 3.0306,
      "step": 350470
    },
    {
      "epoch": 0.24791775443152636,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001299042,
      "loss": 3.1629,
      "step": 350480
    },
    {
      "epoch": 0.24793639766768455,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001299022,
      "loss": 3.1124,
      "step": 350490
    },
    {
      "epoch": 0.24795504090384274,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001299002,
      "loss": 3.1012,
      "step": 350500
    },
    {
      "epoch": 0.24797368414000093,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012989820000000002,
      "loss": 3.0473,
      "step": 350510
    },
    {
      "epoch": 0.2479923273761591,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001298962,
      "loss": 3.0944,
      "step": 350520
    },
    {
      "epoch": 0.24801097061231728,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012989420000000001,
      "loss": 3.0826,
      "step": 350530
    },
    {
      "epoch": 0.24802961384847547,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001298922,
      "loss": 3.1173,
      "step": 350540
    },
    {
      "epoch": 0.24804825708463366,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001298902,
      "loss": 3.1551,
      "step": 350550
    },
    {
      "epoch": 0.24806690032079184,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012988820000000003,
      "loss": 3.0585,
      "step": 350560
    },
    {
      "epoch": 0.24808554355695,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001298862,
      "loss": 3.1535,
      "step": 350570
    },
    {
      "epoch": 0.2481041867931082,
      "grad_norm": 0.75,
      "learning_rate": 0.00012988420000000002,
      "loss": 3.1164,
      "step": 350580
    },
    {
      "epoch": 0.24812283002926638,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001298822,
      "loss": 3.1443,
      "step": 350590
    },
    {
      "epoch": 0.24814147326542457,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001298802,
      "loss": 3.1906,
      "step": 350600
    },
    {
      "epoch": 0.24816011650158276,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001298782,
      "loss": 3.1666,
      "step": 350610
    },
    {
      "epoch": 0.24817875973774092,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001298762,
      "loss": 3.1894,
      "step": 350620
    },
    {
      "epoch": 0.2481974029738991,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012987420000000002,
      "loss": 3.1964,
      "step": 350630
    },
    {
      "epoch": 0.2482160462100573,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001298722,
      "loss": 3.1272,
      "step": 350640
    },
    {
      "epoch": 0.2482346894462155,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012987020000000002,
      "loss": 2.9746,
      "step": 350650
    },
    {
      "epoch": 0.24825333268237368,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012986819999999998,
      "loss": 3.2495,
      "step": 350660
    },
    {
      "epoch": 0.24827197591853187,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001298662,
      "loss": 3.1302,
      "step": 350670
    },
    {
      "epoch": 0.24829061915469003,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001298642,
      "loss": 3.0876,
      "step": 350680
    },
    {
      "epoch": 0.24830926239084822,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001298622,
      "loss": 3.2144,
      "step": 350690
    },
    {
      "epoch": 0.2483279056270064,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012986020000000002,
      "loss": 2.9974,
      "step": 350700
    },
    {
      "epoch": 0.2483465488631646,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001298582,
      "loss": 3.0783,
      "step": 350710
    },
    {
      "epoch": 0.24836519209932278,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001298562,
      "loss": 3.1543,
      "step": 350720
    },
    {
      "epoch": 0.24838383533548095,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001298542,
      "loss": 3.1208,
      "step": 350730
    },
    {
      "epoch": 0.24840247857163913,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001298522,
      "loss": 3.109,
      "step": 350740
    },
    {
      "epoch": 0.24842112180779732,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001298502,
      "loss": 3.131,
      "step": 350750
    },
    {
      "epoch": 0.2484397650439555,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001298482,
      "loss": 3.0304,
      "step": 350760
    },
    {
      "epoch": 0.2484584082801137,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012984620000000002,
      "loss": 3.1419,
      "step": 350770
    },
    {
      "epoch": 0.24847705151627186,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001298442,
      "loss": 3.1205,
      "step": 350780
    },
    {
      "epoch": 0.24849569475243005,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001298422,
      "loss": 3.1418,
      "step": 350790
    },
    {
      "epoch": 0.24851433798858824,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001298402,
      "loss": 3.0216,
      "step": 350800
    },
    {
      "epoch": 0.24853298122474643,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001298382,
      "loss": 3.1575,
      "step": 350810
    },
    {
      "epoch": 0.24855162446090462,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001298362,
      "loss": 2.9623,
      "step": 350820
    },
    {
      "epoch": 0.24857026769706278,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012983420000000002,
      "loss": 3.168,
      "step": 350830
    },
    {
      "epoch": 0.24858891093322097,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012983220000000001,
      "loss": 3.1129,
      "step": 350840
    },
    {
      "epoch": 0.24860755416937916,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001298302,
      "loss": 3.0969,
      "step": 350850
    },
    {
      "epoch": 0.24862619740553735,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001298282,
      "loss": 3.1875,
      "step": 350860
    },
    {
      "epoch": 0.24864484064169554,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001298262,
      "loss": 3.1811,
      "step": 350870
    },
    {
      "epoch": 0.24866348387785372,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001298242,
      "loss": 3.0199,
      "step": 350880
    },
    {
      "epoch": 0.24868212711401189,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001298222,
      "loss": 3.1788,
      "step": 350890
    },
    {
      "epoch": 0.24870077035017008,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012982020000000001,
      "loss": 3.1031,
      "step": 350900
    },
    {
      "epoch": 0.24871941358632826,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001298182,
      "loss": 3.1531,
      "step": 350910
    },
    {
      "epoch": 0.24873805682248645,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001298162,
      "loss": 3.1097,
      "step": 350920
    },
    {
      "epoch": 0.24875670005864464,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001298142,
      "loss": 3.1559,
      "step": 350930
    },
    {
      "epoch": 0.2487753432948028,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001298122,
      "loss": 3.1282,
      "step": 350940
    },
    {
      "epoch": 0.248793986530961,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012981020000000002,
      "loss": 3.1054,
      "step": 350950
    },
    {
      "epoch": 0.24881262976711918,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00012980820000000002,
      "loss": 3.1782,
      "step": 350960
    },
    {
      "epoch": 0.24883127300327737,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001298062,
      "loss": 3.066,
      "step": 350970
    },
    {
      "epoch": 0.24884991623943556,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001298042,
      "loss": 3.1198,
      "step": 350980
    },
    {
      "epoch": 0.24886855947559372,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001298022,
      "loss": 3.0982,
      "step": 350990
    },
    {
      "epoch": 0.2488872027117519,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012980020000000002,
      "loss": 3.0833,
      "step": 351000
    },
    {
      "epoch": 0.2489058459479101,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001297982,
      "loss": 3.2146,
      "step": 351010
    },
    {
      "epoch": 0.2489244891840683,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012979620000000002,
      "loss": 2.9938,
      "step": 351020
    },
    {
      "epoch": 0.24894313242022648,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001297942,
      "loss": 3.1955,
      "step": 351030
    },
    {
      "epoch": 0.24896177565638464,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001297922,
      "loss": 3.1493,
      "step": 351040
    },
    {
      "epoch": 0.24898041889254283,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001297902,
      "loss": 3.2479,
      "step": 351050
    },
    {
      "epoch": 0.24899906212870102,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001297882,
      "loss": 3.0805,
      "step": 351060
    },
    {
      "epoch": 0.2490177053648592,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012978620000000002,
      "loss": 3.0573,
      "step": 351070
    },
    {
      "epoch": 0.2490363486010174,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001297842,
      "loss": 3.3148,
      "step": 351080
    },
    {
      "epoch": 0.24905499183717558,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001297822,
      "loss": 3.2277,
      "step": 351090
    },
    {
      "epoch": 0.24907363507333374,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001297802,
      "loss": 3.2869,
      "step": 351100
    },
    {
      "epoch": 0.24909227830949193,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001297782,
      "loss": 3.0409,
      "step": 351110
    },
    {
      "epoch": 0.24911092154565012,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012977620000000003,
      "loss": 3.1975,
      "step": 351120
    },
    {
      "epoch": 0.2491295647818083,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001297742,
      "loss": 2.9726,
      "step": 351130
    },
    {
      "epoch": 0.2491482080179665,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012977220000000002,
      "loss": 3.1864,
      "step": 351140
    },
    {
      "epoch": 0.24916685125412466,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012977019999999999,
      "loss": 3.0605,
      "step": 351150
    },
    {
      "epoch": 0.24918549449028285,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001297682,
      "loss": 3.2493,
      "step": 351160
    },
    {
      "epoch": 0.24920413772644104,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001297662,
      "loss": 3.0593,
      "step": 351170
    },
    {
      "epoch": 0.24922278096259923,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001297642,
      "loss": 3.2012,
      "step": 351180
    },
    {
      "epoch": 0.24924142419875742,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012976220000000002,
      "loss": 3.0955,
      "step": 351190
    },
    {
      "epoch": 0.24926006743491558,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001297602,
      "loss": 3.0838,
      "step": 351200
    },
    {
      "epoch": 0.24927871067107377,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012975820000000001,
      "loss": 2.9659,
      "step": 351210
    },
    {
      "epoch": 0.24929735390723196,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001297562,
      "loss": 3.0291,
      "step": 351220
    },
    {
      "epoch": 0.24931599714339014,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001297542,
      "loss": 2.9802,
      "step": 351230
    },
    {
      "epoch": 0.24933464037954833,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001297522,
      "loss": 3.1367,
      "step": 351240
    },
    {
      "epoch": 0.2493532836157065,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001297502,
      "loss": 3.1554,
      "step": 351250
    },
    {
      "epoch": 0.24937192685186468,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012974820000000002,
      "loss": 3.1575,
      "step": 351260
    },
    {
      "epoch": 0.24939057008802287,
      "grad_norm": 0.625,
      "learning_rate": 0.0001297462,
      "loss": 3.2517,
      "step": 351270
    },
    {
      "epoch": 0.24940921332418106,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001297442,
      "loss": 3.1361,
      "step": 351280
    },
    {
      "epoch": 0.24942785656033925,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001297422,
      "loss": 3.1516,
      "step": 351290
    },
    {
      "epoch": 0.2494464997964974,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001297402,
      "loss": 3.0685,
      "step": 351300
    },
    {
      "epoch": 0.2494651430326556,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001297382,
      "loss": 3.205,
      "step": 351310
    },
    {
      "epoch": 0.2494837862688138,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001297362,
      "loss": 3.0975,
      "step": 351320
    },
    {
      "epoch": 0.24950242950497198,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012973420000000002,
      "loss": 3.0732,
      "step": 351330
    },
    {
      "epoch": 0.24952107274113017,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001297322,
      "loss": 3.0958,
      "step": 351340
    },
    {
      "epoch": 0.24953971597728836,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001297302,
      "loss": 3.155,
      "step": 351350
    },
    {
      "epoch": 0.24955835921344652,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001297282,
      "loss": 3.1423,
      "step": 351360
    },
    {
      "epoch": 0.2495770024496047,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001297262,
      "loss": 2.9495,
      "step": 351370
    },
    {
      "epoch": 0.2495956456857629,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001297242,
      "loss": 3.1455,
      "step": 351380
    },
    {
      "epoch": 0.24961428892192108,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012972220000000002,
      "loss": 3.1077,
      "step": 351390
    },
    {
      "epoch": 0.24963293215807927,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001297202,
      "loss": 3.0095,
      "step": 351400
    },
    {
      "epoch": 0.24965157539423743,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001297182,
      "loss": 3.1949,
      "step": 351410
    },
    {
      "epoch": 0.24967021863039562,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001297162,
      "loss": 3.1651,
      "step": 351420
    },
    {
      "epoch": 0.2496888618665538,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001297142,
      "loss": 3.0725,
      "step": 351430
    },
    {
      "epoch": 0.249707505102712,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012971220000000002,
      "loss": 3.133,
      "step": 351440
    },
    {
      "epoch": 0.2497261483388702,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012971020000000002,
      "loss": 3.1219,
      "step": 351450
    },
    {
      "epoch": 0.24974479157502835,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001297082,
      "loss": 3.1252,
      "step": 351460
    },
    {
      "epoch": 0.24976343481118654,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001297062,
      "loss": 3.2238,
      "step": 351470
    },
    {
      "epoch": 0.24978207804734473,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001297042,
      "loss": 3.1111,
      "step": 351480
    },
    {
      "epoch": 0.24980072128350292,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001297022,
      "loss": 3.12,
      "step": 351490
    },
    {
      "epoch": 0.2498193645196611,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001297002,
      "loss": 3.1418,
      "step": 351500
    },
    {
      "epoch": 0.24983800775581927,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012969820000000002,
      "loss": 3.1092,
      "step": 351510
    },
    {
      "epoch": 0.24985665099197746,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012969620000000001,
      "loss": 3.0413,
      "step": 351520
    },
    {
      "epoch": 0.24987529422813565,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001296942,
      "loss": 3.1083,
      "step": 351530
    },
    {
      "epoch": 0.24989393746429384,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001296922,
      "loss": 3.1144,
      "step": 351540
    },
    {
      "epoch": 0.24991258070045202,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001296902,
      "loss": 3.1295,
      "step": 351550
    },
    {
      "epoch": 0.24993122393661021,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012968820000000002,
      "loss": 3.1644,
      "step": 351560
    },
    {
      "epoch": 0.24994986717276838,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001296862,
      "loss": 3.1136,
      "step": 351570
    },
    {
      "epoch": 0.24996851040892656,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012968420000000001,
      "loss": 2.9791,
      "step": 351580
    },
    {
      "epoch": 0.24998715364508475,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001296822,
      "loss": 3.0713,
      "step": 351590
    },
    {
      "epoch": 0.2500057968812429,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001296802,
      "loss": 3.1812,
      "step": 351600
    },
    {
      "epoch": 0.2500244401174011,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012967820000000003,
      "loss": 3.1573,
      "step": 351610
    },
    {
      "epoch": 0.2500430833535593,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001296762,
      "loss": 3.0994,
      "step": 351620
    },
    {
      "epoch": 0.2500617265897175,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012967420000000002,
      "loss": 3.1581,
      "step": 351630
    },
    {
      "epoch": 0.25008036982587567,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001296722,
      "loss": 3.1906,
      "step": 351640
    },
    {
      "epoch": 0.25009901306203386,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001296702,
      "loss": 3.026,
      "step": 351650
    },
    {
      "epoch": 0.25011765629819205,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001296682,
      "loss": 3.1194,
      "step": 351660
    },
    {
      "epoch": 0.25013629953435024,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001296662,
      "loss": 3.0144,
      "step": 351670
    },
    {
      "epoch": 0.2501549427705084,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012966420000000002,
      "loss": 3.1769,
      "step": 351680
    },
    {
      "epoch": 0.2501735860066666,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001296622,
      "loss": 3.2459,
      "step": 351690
    },
    {
      "epoch": 0.25019222924282475,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012966020000000002,
      "loss": 3.1487,
      "step": 351700
    },
    {
      "epoch": 0.25021087247898294,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012965819999999998,
      "loss": 3.0725,
      "step": 351710
    },
    {
      "epoch": 0.2502295157151411,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001296562,
      "loss": 3.0156,
      "step": 351720
    },
    {
      "epoch": 0.2502481589512993,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001296542,
      "loss": 3.1649,
      "step": 351730
    },
    {
      "epoch": 0.2502668021874575,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001296522,
      "loss": 3.1587,
      "step": 351740
    },
    {
      "epoch": 0.2502854454236157,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012965020000000002,
      "loss": 3.152,
      "step": 351750
    },
    {
      "epoch": 0.2503040886597739,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001296482,
      "loss": 3.07,
      "step": 351760
    },
    {
      "epoch": 0.25032273189593207,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001296462,
      "loss": 3.0633,
      "step": 351770
    },
    {
      "epoch": 0.25034137513209026,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001296442,
      "loss": 3.0933,
      "step": 351780
    },
    {
      "epoch": 0.25036001836824845,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001296422,
      "loss": 3.1482,
      "step": 351790
    },
    {
      "epoch": 0.2503786616044066,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001296402,
      "loss": 3.1877,
      "step": 351800
    },
    {
      "epoch": 0.25039730484056477,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001296382,
      "loss": 3.0885,
      "step": 351810
    },
    {
      "epoch": 0.25041594807672296,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012963620000000002,
      "loss": 3.1688,
      "step": 351820
    },
    {
      "epoch": 0.25043459131288115,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001296342,
      "loss": 3.1313,
      "step": 351830
    },
    {
      "epoch": 0.25045323454903934,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001296322,
      "loss": 3.2035,
      "step": 351840
    },
    {
      "epoch": 0.2504718777851975,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001296302,
      "loss": 3.116,
      "step": 351850
    },
    {
      "epoch": 0.2504905210213557,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001296282,
      "loss": 3.1635,
      "step": 351860
    },
    {
      "epoch": 0.2505091642575139,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001296262,
      "loss": 3.0264,
      "step": 351870
    },
    {
      "epoch": 0.2505278074936721,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001296242,
      "loss": 3.0708,
      "step": 351880
    },
    {
      "epoch": 0.2505464507298303,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012962220000000001,
      "loss": 2.9875,
      "step": 351890
    },
    {
      "epoch": 0.2505650939659885,
      "grad_norm": 0.75,
      "learning_rate": 0.0001296202,
      "loss": 3.2019,
      "step": 351900
    },
    {
      "epoch": 0.2505837372021466,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001296182,
      "loss": 3.111,
      "step": 351910
    },
    {
      "epoch": 0.2506023804383048,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001296162,
      "loss": 3.2265,
      "step": 351920
    },
    {
      "epoch": 0.250621023674463,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001296142,
      "loss": 3.2006,
      "step": 351930
    },
    {
      "epoch": 0.2506396669106212,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012961220000000002,
      "loss": 3.1873,
      "step": 351940
    },
    {
      "epoch": 0.25065831014677936,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012961020000000001,
      "loss": 3.1677,
      "step": 351950
    },
    {
      "epoch": 0.25067695338293755,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001296082,
      "loss": 3.1925,
      "step": 351960
    },
    {
      "epoch": 0.25069559661909574,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001296062,
      "loss": 3.1678,
      "step": 351970
    },
    {
      "epoch": 0.25071423985525393,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001296042,
      "loss": 3.262,
      "step": 351980
    },
    {
      "epoch": 0.2507328830914121,
      "grad_norm": 0.75,
      "learning_rate": 0.0001296022,
      "loss": 3.1142,
      "step": 351990
    },
    {
      "epoch": 0.2507515263275703,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012960020000000002,
      "loss": 3.049,
      "step": 352000
    },
    {
      "epoch": 0.25077016956372844,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012959820000000002,
      "loss": 3.2024,
      "step": 352010
    },
    {
      "epoch": 0.25078881279988663,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001295962,
      "loss": 3.189,
      "step": 352020
    },
    {
      "epoch": 0.2508074560360448,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001295942,
      "loss": 3.0952,
      "step": 352030
    },
    {
      "epoch": 0.250826099272203,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001295922,
      "loss": 3.1215,
      "step": 352040
    },
    {
      "epoch": 0.2508447425083612,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012959020000000003,
      "loss": 3.3029,
      "step": 352050
    },
    {
      "epoch": 0.2508633857445194,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001295882,
      "loss": 3.1876,
      "step": 352060
    },
    {
      "epoch": 0.2508820289806776,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012958620000000002,
      "loss": 3.3022,
      "step": 352070
    },
    {
      "epoch": 0.25090067221683576,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001295842,
      "loss": 3.169,
      "step": 352080
    },
    {
      "epoch": 0.25091931545299395,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001295822,
      "loss": 3.0836,
      "step": 352090
    },
    {
      "epoch": 0.25093795868915214,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001295802,
      "loss": 3.1885,
      "step": 352100
    },
    {
      "epoch": 0.25095660192531033,
      "grad_norm": 0.75,
      "learning_rate": 0.0001295782,
      "loss": 3.1024,
      "step": 352110
    },
    {
      "epoch": 0.25097524516146846,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012957620000000002,
      "loss": 3.3139,
      "step": 352120
    },
    {
      "epoch": 0.25099388839762665,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001295742,
      "loss": 3.147,
      "step": 352130
    },
    {
      "epoch": 0.25101253163378484,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001295722,
      "loss": 3.1001,
      "step": 352140
    },
    {
      "epoch": 0.25103117486994303,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001295702,
      "loss": 2.9717,
      "step": 352150
    },
    {
      "epoch": 0.2510498181061012,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001295682,
      "loss": 3.2097,
      "step": 352160
    },
    {
      "epoch": 0.2510684613422594,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012956620000000003,
      "loss": 3.1398,
      "step": 352170
    },
    {
      "epoch": 0.2510871045784176,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001295642,
      "loss": 3.2541,
      "step": 352180
    },
    {
      "epoch": 0.2511057478145758,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012956220000000002,
      "loss": 3.1267,
      "step": 352190
    },
    {
      "epoch": 0.251124391050734,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001295602,
      "loss": 3.1347,
      "step": 352200
    },
    {
      "epoch": 0.25114303428689216,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001295582,
      "loss": 3.2051,
      "step": 352210
    },
    {
      "epoch": 0.2511616775230503,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001295562,
      "loss": 3.1963,
      "step": 352220
    },
    {
      "epoch": 0.2511803207592085,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001295542,
      "loss": 3.1176,
      "step": 352230
    },
    {
      "epoch": 0.2511989639953667,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012955220000000002,
      "loss": 3.1007,
      "step": 352240
    },
    {
      "epoch": 0.25121760723152486,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001295502,
      "loss": 3.0984,
      "step": 352250
    },
    {
      "epoch": 0.25123625046768305,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012954820000000001,
      "loss": 3.2985,
      "step": 352260
    },
    {
      "epoch": 0.25125489370384124,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012954619999999998,
      "loss": 3.2866,
      "step": 352270
    },
    {
      "epoch": 0.25127353693999943,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001295442,
      "loss": 3.3106,
      "step": 352280
    },
    {
      "epoch": 0.2512921801761576,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001295422,
      "loss": 3.1362,
      "step": 352290
    },
    {
      "epoch": 0.2513108234123158,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001295402,
      "loss": 3.0346,
      "step": 352300
    },
    {
      "epoch": 0.251329466648474,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012953820000000002,
      "loss": 3.1584,
      "step": 352310
    },
    {
      "epoch": 0.25134810988463213,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001295362,
      "loss": 3.1618,
      "step": 352320
    },
    {
      "epoch": 0.2513667531207903,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001295342,
      "loss": 3.0072,
      "step": 352330
    },
    {
      "epoch": 0.2513853963569485,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001295322,
      "loss": 3.1141,
      "step": 352340
    },
    {
      "epoch": 0.2514040395931067,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001295302,
      "loss": 3.1648,
      "step": 352350
    },
    {
      "epoch": 0.2514226828292649,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001295282,
      "loss": 3.2189,
      "step": 352360
    },
    {
      "epoch": 0.2514413260654231,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001295262,
      "loss": 3.2263,
      "step": 352370
    },
    {
      "epoch": 0.25145996930158127,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012952420000000002,
      "loss": 3.2095,
      "step": 352380
    },
    {
      "epoch": 0.25147861253773945,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001295222,
      "loss": 3.0746,
      "step": 352390
    },
    {
      "epoch": 0.25149725577389764,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001295202,
      "loss": 3.0913,
      "step": 352400
    },
    {
      "epoch": 0.25151589901005583,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001295182,
      "loss": 3.1969,
      "step": 352410
    },
    {
      "epoch": 0.251534542246214,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001295162,
      "loss": 3.1141,
      "step": 352420
    },
    {
      "epoch": 0.25155318548237215,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012951420000000002,
      "loss": 3.0378,
      "step": 352430
    },
    {
      "epoch": 0.25157182871853034,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012951220000000002,
      "loss": 3.0675,
      "step": 352440
    },
    {
      "epoch": 0.25159047195468853,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001295102,
      "loss": 3.1077,
      "step": 352450
    },
    {
      "epoch": 0.2516091151908467,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001295082,
      "loss": 3.1256,
      "step": 352460
    },
    {
      "epoch": 0.2516277584270049,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001295062,
      "loss": 3.146,
      "step": 352470
    },
    {
      "epoch": 0.2516464016631631,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001295042,
      "loss": 3.0898,
      "step": 352480
    },
    {
      "epoch": 0.2516650448993213,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001295022,
      "loss": 3.109,
      "step": 352490
    },
    {
      "epoch": 0.2516836881354795,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012950020000000002,
      "loss": 3.1097,
      "step": 352500
    },
    {
      "epoch": 0.25170233137163767,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001294982,
      "loss": 3.046,
      "step": 352510
    },
    {
      "epoch": 0.25172097460779586,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001294962,
      "loss": 3.0914,
      "step": 352520
    },
    {
      "epoch": 0.251739617843954,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001294942,
      "loss": 3.2542,
      "step": 352530
    },
    {
      "epoch": 0.2517582610801122,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001294922,
      "loss": 3.1792,
      "step": 352540
    },
    {
      "epoch": 0.25177690431627037,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001294902,
      "loss": 3.1209,
      "step": 352550
    },
    {
      "epoch": 0.25179554755242856,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012948820000000002,
      "loss": 3.082,
      "step": 352560
    },
    {
      "epoch": 0.25181419078858674,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012948620000000001,
      "loss": 3.1804,
      "step": 352570
    },
    {
      "epoch": 0.25183283402474493,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001294842,
      "loss": 3.058,
      "step": 352580
    },
    {
      "epoch": 0.2518514772609031,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001294822,
      "loss": 3.1616,
      "step": 352590
    },
    {
      "epoch": 0.2518701204970613,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001294802,
      "loss": 3.1323,
      "step": 352600
    },
    {
      "epoch": 0.2518887637332195,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012947820000000002,
      "loss": 3.0962,
      "step": 352610
    },
    {
      "epoch": 0.2519074069693777,
      "grad_norm": 0.625,
      "learning_rate": 0.0001294762,
      "loss": 3.1111,
      "step": 352620
    },
    {
      "epoch": 0.2519260502055359,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012947420000000001,
      "loss": 3.0999,
      "step": 352630
    },
    {
      "epoch": 0.251944693441694,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001294722,
      "loss": 3.1051,
      "step": 352640
    },
    {
      "epoch": 0.2519633366778522,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001294702,
      "loss": 3.2421,
      "step": 352650
    },
    {
      "epoch": 0.2519819799140104,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012946820000000003,
      "loss": 3.0562,
      "step": 352660
    },
    {
      "epoch": 0.2520006231501686,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001294662,
      "loss": 3.0862,
      "step": 352670
    },
    {
      "epoch": 0.25201926638632677,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012946420000000002,
      "loss": 3.0202,
      "step": 352680
    },
    {
      "epoch": 0.25203790962248496,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001294622,
      "loss": 3.1262,
      "step": 352690
    },
    {
      "epoch": 0.25205655285864315,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001294602,
      "loss": 3.1113,
      "step": 352700
    },
    {
      "epoch": 0.25207519609480133,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001294582,
      "loss": 3.1709,
      "step": 352710
    },
    {
      "epoch": 0.2520938393309595,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001294562,
      "loss": 3.0507,
      "step": 352720
    },
    {
      "epoch": 0.2521124825671177,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012945420000000003,
      "loss": 3.0835,
      "step": 352730
    },
    {
      "epoch": 0.25213112580327585,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001294522,
      "loss": 3.0318,
      "step": 352740
    },
    {
      "epoch": 0.25214976903943404,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012945020000000002,
      "loss": 3.1163,
      "step": 352750
    },
    {
      "epoch": 0.2521684122755922,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012944819999999998,
      "loss": 3.1525,
      "step": 352760
    },
    {
      "epoch": 0.2521870555117504,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001294462,
      "loss": 3.0859,
      "step": 352770
    },
    {
      "epoch": 0.2522056987479086,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001294442,
      "loss": 3.1871,
      "step": 352780
    },
    {
      "epoch": 0.2522243419840668,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001294422,
      "loss": 3.0432,
      "step": 352790
    },
    {
      "epoch": 0.252242985220225,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012944020000000002,
      "loss": 3.2195,
      "step": 352800
    },
    {
      "epoch": 0.25226162845638317,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001294382,
      "loss": 3.176,
      "step": 352810
    },
    {
      "epoch": 0.25228027169254136,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001294362,
      "loss": 3.1403,
      "step": 352820
    },
    {
      "epoch": 0.25229891492869955,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001294342,
      "loss": 3.0391,
      "step": 352830
    },
    {
      "epoch": 0.25231755816485774,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001294322,
      "loss": 3.0953,
      "step": 352840
    },
    {
      "epoch": 0.25233620140101587,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001294302,
      "loss": 3.1403,
      "step": 352850
    },
    {
      "epoch": 0.25235484463717406,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001294282,
      "loss": 3.2762,
      "step": 352860
    },
    {
      "epoch": 0.25237348787333225,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012942620000000002,
      "loss": 3.1183,
      "step": 352870
    },
    {
      "epoch": 0.25239213110949044,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012942419999999999,
      "loss": 3.12,
      "step": 352880
    },
    {
      "epoch": 0.2524107743456486,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001294222,
      "loss": 3.1843,
      "step": 352890
    },
    {
      "epoch": 0.2524294175818068,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001294202,
      "loss": 3.2874,
      "step": 352900
    },
    {
      "epoch": 0.252448060817965,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001294182,
      "loss": 3.0589,
      "step": 352910
    },
    {
      "epoch": 0.2524667040541232,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012941620000000002,
      "loss": 3.2379,
      "step": 352920
    },
    {
      "epoch": 0.2524853472902814,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001294142,
      "loss": 3.1558,
      "step": 352930
    },
    {
      "epoch": 0.25250399052643957,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012941220000000001,
      "loss": 3.1917,
      "step": 352940
    },
    {
      "epoch": 0.2525226337625977,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001294102,
      "loss": 3.2965,
      "step": 352950
    },
    {
      "epoch": 0.2525412769987559,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001294082,
      "loss": 3.1592,
      "step": 352960
    },
    {
      "epoch": 0.2525599202349141,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001294062,
      "loss": 3.1487,
      "step": 352970
    },
    {
      "epoch": 0.25257856347107227,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001294042,
      "loss": 3.0548,
      "step": 352980
    },
    {
      "epoch": 0.25259720670723046,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012940220000000002,
      "loss": 3.0456,
      "step": 352990
    },
    {
      "epoch": 0.25261584994338865,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012940020000000002,
      "loss": 3.069,
      "step": 353000
    },
    {
      "epoch": 0.25263449317954684,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001293982,
      "loss": 2.9406,
      "step": 353010
    },
    {
      "epoch": 0.252653136415705,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001293962,
      "loss": 3.0495,
      "step": 353020
    },
    {
      "epoch": 0.2526717796518632,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001293942,
      "loss": 3.2543,
      "step": 353030
    },
    {
      "epoch": 0.2526904228880214,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001293922,
      "loss": 3.1849,
      "step": 353040
    },
    {
      "epoch": 0.2527090661241796,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012939020000000002,
      "loss": 3.13,
      "step": 353050
    },
    {
      "epoch": 0.2527277093603377,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012938820000000002,
      "loss": 3.089,
      "step": 353060
    },
    {
      "epoch": 0.2527463525964959,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001293862,
      "loss": 3.0664,
      "step": 353070
    },
    {
      "epoch": 0.2527649958326541,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001293842,
      "loss": 3.2315,
      "step": 353080
    },
    {
      "epoch": 0.2527836390688123,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001293822,
      "loss": 3.1307,
      "step": 353090
    },
    {
      "epoch": 0.2528022823049705,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001293802,
      "loss": 3.2136,
      "step": 353100
    },
    {
      "epoch": 0.25282092554112867,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001293782,
      "loss": 3.068,
      "step": 353110
    },
    {
      "epoch": 0.25283956877728686,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012937620000000002,
      "loss": 3.2392,
      "step": 353120
    },
    {
      "epoch": 0.25285821201344505,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001293742,
      "loss": 3.1917,
      "step": 353130
    },
    {
      "epoch": 0.25287685524960324,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001293722,
      "loss": 3.0563,
      "step": 353140
    },
    {
      "epoch": 0.2528954984857614,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001293702,
      "loss": 3.1986,
      "step": 353150
    },
    {
      "epoch": 0.25291414172191956,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001293682,
      "loss": 3.1495,
      "step": 353160
    },
    {
      "epoch": 0.25293278495807775,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012936620000000002,
      "loss": 3.087,
      "step": 353170
    },
    {
      "epoch": 0.25295142819423594,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001293642,
      "loss": 3.222,
      "step": 353180
    },
    {
      "epoch": 0.25297007143039413,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001293622,
      "loss": 3.0591,
      "step": 353190
    },
    {
      "epoch": 0.2529887146665523,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001293602,
      "loss": 3.1529,
      "step": 353200
    },
    {
      "epoch": 0.2530073579027105,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001293582,
      "loss": 3.1633,
      "step": 353210
    },
    {
      "epoch": 0.2530260011388687,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012935620000000003,
      "loss": 3.1516,
      "step": 353220
    },
    {
      "epoch": 0.2530446443750269,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001293542,
      "loss": 3.1755,
      "step": 353230
    },
    {
      "epoch": 0.2530632876111851,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012935220000000002,
      "loss": 2.9688,
      "step": 353240
    },
    {
      "epoch": 0.25308193084734326,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001293502,
      "loss": 3.0712,
      "step": 353250
    },
    {
      "epoch": 0.25310057408350145,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001293482,
      "loss": 3.1247,
      "step": 353260
    },
    {
      "epoch": 0.2531192173196596,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001293462,
      "loss": 3.1703,
      "step": 353270
    },
    {
      "epoch": 0.2531378605558178,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001293442,
      "loss": 3.2961,
      "step": 353280
    },
    {
      "epoch": 0.25315650379197596,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012934220000000002,
      "loss": 3.0667,
      "step": 353290
    },
    {
      "epoch": 0.25317514702813415,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001293402,
      "loss": 3.0987,
      "step": 353300
    },
    {
      "epoch": 0.25319379026429234,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012933820000000001,
      "loss": 3.1817,
      "step": 353310
    },
    {
      "epoch": 0.25321243350045053,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012933619999999998,
      "loss": 3.0764,
      "step": 353320
    },
    {
      "epoch": 0.2532310767366087,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001293342,
      "loss": 3.0909,
      "step": 353330
    },
    {
      "epoch": 0.2532497199727669,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001293322,
      "loss": 3.0382,
      "step": 353340
    },
    {
      "epoch": 0.2532683632089251,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001293302,
      "loss": 3.2255,
      "step": 353350
    },
    {
      "epoch": 0.2532870064450833,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012932820000000002,
      "loss": 3.0976,
      "step": 353360
    },
    {
      "epoch": 0.2533056496812414,
      "grad_norm": 0.75,
      "learning_rate": 0.0001293262,
      "loss": 2.9613,
      "step": 353370
    },
    {
      "epoch": 0.2533242929173996,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001293242,
      "loss": 3.097,
      "step": 353380
    },
    {
      "epoch": 0.2533429361535578,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001293222,
      "loss": 3.0223,
      "step": 353390
    },
    {
      "epoch": 0.253361579389716,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001293202,
      "loss": 3.2609,
      "step": 353400
    },
    {
      "epoch": 0.2533802226258742,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012931820000000003,
      "loss": 3.1303,
      "step": 353410
    },
    {
      "epoch": 0.25339886586203236,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001293162,
      "loss": 3.1866,
      "step": 353420
    },
    {
      "epoch": 0.25341750909819055,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012931420000000002,
      "loss": 3.154,
      "step": 353430
    },
    {
      "epoch": 0.25343615233434874,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001293122,
      "loss": 3.174,
      "step": 353440
    },
    {
      "epoch": 0.25345479557050693,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001293102,
      "loss": 3.1295,
      "step": 353450
    },
    {
      "epoch": 0.2534734388066651,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001293082,
      "loss": 3.1796,
      "step": 353460
    },
    {
      "epoch": 0.2534920820428233,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001293062,
      "loss": 3.1907,
      "step": 353470
    },
    {
      "epoch": 0.25351072527898144,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012930420000000002,
      "loss": 3.149,
      "step": 353480
    },
    {
      "epoch": 0.25352936851513963,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001293022,
      "loss": 3.0744,
      "step": 353490
    },
    {
      "epoch": 0.2535480117512978,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001293002,
      "loss": 3.2148,
      "step": 353500
    },
    {
      "epoch": 0.253566654987456,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001292982,
      "loss": 3.0424,
      "step": 353510
    },
    {
      "epoch": 0.2535852982236142,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001292962,
      "loss": 3.1017,
      "step": 353520
    },
    {
      "epoch": 0.2536039414597724,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001292942,
      "loss": 3.2108,
      "step": 353530
    },
    {
      "epoch": 0.2536225846959306,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001292922,
      "loss": 3.0465,
      "step": 353540
    },
    {
      "epoch": 0.25364122793208876,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012929020000000002,
      "loss": 3.0575,
      "step": 353550
    },
    {
      "epoch": 0.25365987116824695,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001292882,
      "loss": 3.0078,
      "step": 353560
    },
    {
      "epoch": 0.25367851440440514,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001292862,
      "loss": 3.1666,
      "step": 353570
    },
    {
      "epoch": 0.2536971576405633,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001292842,
      "loss": 3.1526,
      "step": 353580
    },
    {
      "epoch": 0.25371580087672146,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001292822,
      "loss": 3.1108,
      "step": 353590
    },
    {
      "epoch": 0.25373444411287965,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001292802,
      "loss": 3.1233,
      "step": 353600
    },
    {
      "epoch": 0.25375308734903784,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012927820000000002,
      "loss": 3.132,
      "step": 353610
    },
    {
      "epoch": 0.25377173058519603,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012927620000000001,
      "loss": 3.1721,
      "step": 353620
    },
    {
      "epoch": 0.2537903738213542,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001292742,
      "loss": 3.0788,
      "step": 353630
    },
    {
      "epoch": 0.2538090170575124,
      "grad_norm": 0.75,
      "learning_rate": 0.0001292722,
      "loss": 3.0831,
      "step": 353640
    },
    {
      "epoch": 0.2538276602936706,
      "grad_norm": 0.75,
      "learning_rate": 0.0001292702,
      "loss": 3.0607,
      "step": 353650
    },
    {
      "epoch": 0.2538463035298288,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012926820000000002,
      "loss": 3.1623,
      "step": 353660
    },
    {
      "epoch": 0.253864946765987,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001292662,
      "loss": 3.1084,
      "step": 353670
    },
    {
      "epoch": 0.2538835900021451,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012926420000000002,
      "loss": 3.1596,
      "step": 353680
    },
    {
      "epoch": 0.2539022332383033,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001292622,
      "loss": 3.1187,
      "step": 353690
    },
    {
      "epoch": 0.2539208764744615,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001292602,
      "loss": 3.2199,
      "step": 353700
    },
    {
      "epoch": 0.2539395197106197,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001292582,
      "loss": 3.0937,
      "step": 353710
    },
    {
      "epoch": 0.25395816294677787,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001292562,
      "loss": 3.1849,
      "step": 353720
    },
    {
      "epoch": 0.25397680618293605,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012925420000000002,
      "loss": 3.124,
      "step": 353730
    },
    {
      "epoch": 0.25399544941909424,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001292522,
      "loss": 3.1163,
      "step": 353740
    },
    {
      "epoch": 0.25401409265525243,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001292502,
      "loss": 3.111,
      "step": 353750
    },
    {
      "epoch": 0.2540327358914106,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001292482,
      "loss": 3.062,
      "step": 353760
    },
    {
      "epoch": 0.2540513791275688,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001292462,
      "loss": 3.1965,
      "step": 353770
    },
    {
      "epoch": 0.254070022363727,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012924420000000003,
      "loss": 3.0503,
      "step": 353780
    },
    {
      "epoch": 0.25408866559988513,
      "grad_norm": 0.75,
      "learning_rate": 0.0001292422,
      "loss": 3.1526,
      "step": 353790
    },
    {
      "epoch": 0.2541073088360433,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012924020000000002,
      "loss": 3.1306,
      "step": 353800
    },
    {
      "epoch": 0.2541259520722015,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012923819999999999,
      "loss": 3.1771,
      "step": 353810
    },
    {
      "epoch": 0.2541445953083597,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001292362,
      "loss": 3.1422,
      "step": 353820
    },
    {
      "epoch": 0.2541632385445179,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001292342,
      "loss": 3.1992,
      "step": 353830
    },
    {
      "epoch": 0.2541818817806761,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001292322,
      "loss": 3.2104,
      "step": 353840
    },
    {
      "epoch": 0.25420052501683427,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012923020000000002,
      "loss": 3.0967,
      "step": 353850
    },
    {
      "epoch": 0.25421916825299246,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001292282,
      "loss": 3.0699,
      "step": 353860
    },
    {
      "epoch": 0.25423781148915064,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001292262,
      "loss": 3.1446,
      "step": 353870
    },
    {
      "epoch": 0.25425645472530883,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001292242,
      "loss": 3.1611,
      "step": 353880
    },
    {
      "epoch": 0.25427509796146697,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001292222,
      "loss": 3.2336,
      "step": 353890
    },
    {
      "epoch": 0.25429374119762516,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012922020000000003,
      "loss": 3.2282,
      "step": 353900
    },
    {
      "epoch": 0.25431238443378335,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001292182,
      "loss": 3.2527,
      "step": 353910
    },
    {
      "epoch": 0.25433102766994153,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012921620000000002,
      "loss": 3.0985,
      "step": 353920
    },
    {
      "epoch": 0.2543496709060997,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001292142,
      "loss": 3.1378,
      "step": 353930
    },
    {
      "epoch": 0.2543683141422579,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001292122,
      "loss": 3.1201,
      "step": 353940
    },
    {
      "epoch": 0.2543869573784161,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001292102,
      "loss": 3.0721,
      "step": 353950
    },
    {
      "epoch": 0.2544056006145743,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001292082,
      "loss": 3.1163,
      "step": 353960
    },
    {
      "epoch": 0.2544242438507325,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012920620000000002,
      "loss": 3.1719,
      "step": 353970
    },
    {
      "epoch": 0.25444288708689067,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001292042,
      "loss": 3.1818,
      "step": 353980
    },
    {
      "epoch": 0.25446153032304886,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012920220000000001,
      "loss": 3.1868,
      "step": 353990
    },
    {
      "epoch": 0.254480173559207,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001292002,
      "loss": 3.0495,
      "step": 354000
    },
    {
      "epoch": 0.2544988167953652,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001291982,
      "loss": 3.1721,
      "step": 354010
    },
    {
      "epoch": 0.25451746003152337,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001291962,
      "loss": 3.1533,
      "step": 354020
    },
    {
      "epoch": 0.25453610326768156,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001291942,
      "loss": 3.1027,
      "step": 354030
    },
    {
      "epoch": 0.25455474650383975,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012919220000000002,
      "loss": 3.0161,
      "step": 354040
    },
    {
      "epoch": 0.25457338973999794,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012919020000000002,
      "loss": 3.1056,
      "step": 354050
    },
    {
      "epoch": 0.2545920329761561,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001291882,
      "loss": 3.0499,
      "step": 354060
    },
    {
      "epoch": 0.2546106762123143,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001291862,
      "loss": 3.1037,
      "step": 354070
    },
    {
      "epoch": 0.2546293194484725,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001291842,
      "loss": 3.1416,
      "step": 354080
    },
    {
      "epoch": 0.2546479626846307,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001291822,
      "loss": 3.1956,
      "step": 354090
    },
    {
      "epoch": 0.2546666059207888,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001291802,
      "loss": 3.0677,
      "step": 354100
    },
    {
      "epoch": 0.254685249156947,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012917820000000002,
      "loss": 3.129,
      "step": 354110
    },
    {
      "epoch": 0.2547038923931052,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001291762,
      "loss": 3.1219,
      "step": 354120
    },
    {
      "epoch": 0.2547225356292634,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001291742,
      "loss": 3.1503,
      "step": 354130
    },
    {
      "epoch": 0.2547411788654216,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001291722,
      "loss": 3.142,
      "step": 354140
    },
    {
      "epoch": 0.25475982210157977,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001291702,
      "loss": 3.0984,
      "step": 354150
    },
    {
      "epoch": 0.25477846533773796,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001291682,
      "loss": 3.1624,
      "step": 354160
    },
    {
      "epoch": 0.25479710857389615,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012916620000000002,
      "loss": 3.016,
      "step": 354170
    },
    {
      "epoch": 0.25481575181005434,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001291642,
      "loss": 3.0716,
      "step": 354180
    },
    {
      "epoch": 0.2548343950462125,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001291622,
      "loss": 3.0852,
      "step": 354190
    },
    {
      "epoch": 0.2548530382823707,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001291602,
      "loss": 3.0027,
      "step": 354200
    },
    {
      "epoch": 0.25487168151852885,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001291582,
      "loss": 3.0128,
      "step": 354210
    },
    {
      "epoch": 0.25489032475468704,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012915620000000002,
      "loss": 3.1911,
      "step": 354220
    },
    {
      "epoch": 0.2549089679908452,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001291542,
      "loss": 3.1355,
      "step": 354230
    },
    {
      "epoch": 0.2549276112270034,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001291522,
      "loss": 3.1216,
      "step": 354240
    },
    {
      "epoch": 0.2549462544631616,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001291502,
      "loss": 3.1227,
      "step": 354250
    },
    {
      "epoch": 0.2549648976993198,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001291482,
      "loss": 3.1201,
      "step": 354260
    },
    {
      "epoch": 0.254983540935478,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012914620000000003,
      "loss": 3.1785,
      "step": 354270
    },
    {
      "epoch": 0.25500218417163617,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001291442,
      "loss": 3.1364,
      "step": 354280
    },
    {
      "epoch": 0.25502082740779436,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012914220000000002,
      "loss": 3.136,
      "step": 354290
    },
    {
      "epoch": 0.25503947064395255,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001291402,
      "loss": 3.2234,
      "step": 354300
    },
    {
      "epoch": 0.2550581138801107,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001291382,
      "loss": 3.0965,
      "step": 354310
    },
    {
      "epoch": 0.25507675711626887,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001291362,
      "loss": 3.2207,
      "step": 354320
    },
    {
      "epoch": 0.25509540035242706,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001291342,
      "loss": 3.0453,
      "step": 354330
    },
    {
      "epoch": 0.25511404358858525,
      "grad_norm": 0.75,
      "learning_rate": 0.00012913220000000002,
      "loss": 3.2196,
      "step": 354340
    },
    {
      "epoch": 0.25513268682474344,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001291302,
      "loss": 3.1851,
      "step": 354350
    },
    {
      "epoch": 0.2551513300609016,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012912820000000002,
      "loss": 3.1135,
      "step": 354360
    },
    {
      "epoch": 0.2551699732970598,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001291262,
      "loss": 3.0384,
      "step": 354370
    },
    {
      "epoch": 0.255188616533218,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001291242,
      "loss": 3.1393,
      "step": 354380
    },
    {
      "epoch": 0.2552072597693762,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001291222,
      "loss": 3.2637,
      "step": 354390
    },
    {
      "epoch": 0.2552259030055344,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001291202,
      "loss": 3.1137,
      "step": 354400
    },
    {
      "epoch": 0.25524454624169257,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012911820000000002,
      "loss": 3.1829,
      "step": 354410
    },
    {
      "epoch": 0.2552631894778507,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001291162,
      "loss": 3.1591,
      "step": 354420
    },
    {
      "epoch": 0.2552818327140089,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001291142,
      "loss": 3.1645,
      "step": 354430
    },
    {
      "epoch": 0.2553004759501671,
      "grad_norm": 0.75,
      "learning_rate": 0.0001291122,
      "loss": 3.1619,
      "step": 354440
    },
    {
      "epoch": 0.25531911918632527,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001291102,
      "loss": 3.0308,
      "step": 354450
    },
    {
      "epoch": 0.25533776242248346,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012910820000000003,
      "loss": 3.1518,
      "step": 354460
    },
    {
      "epoch": 0.25535640565864165,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001291062,
      "loss": 3.0794,
      "step": 354470
    },
    {
      "epoch": 0.25537504889479984,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012910420000000002,
      "loss": 3.0955,
      "step": 354480
    },
    {
      "epoch": 0.255393692130958,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012910219999999999,
      "loss": 3.0051,
      "step": 354490
    },
    {
      "epoch": 0.2554123353671162,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001291002,
      "loss": 3.0209,
      "step": 354500
    },
    {
      "epoch": 0.2554309786032744,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001290982,
      "loss": 3.1192,
      "step": 354510
    },
    {
      "epoch": 0.25544962183943254,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001290962,
      "loss": 3.1493,
      "step": 354520
    },
    {
      "epoch": 0.25546826507559073,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012909420000000002,
      "loss": 3.2174,
      "step": 354530
    },
    {
      "epoch": 0.2554869083117489,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001290922,
      "loss": 3.1016,
      "step": 354540
    },
    {
      "epoch": 0.2555055515479071,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001290902,
      "loss": 3.1668,
      "step": 354550
    },
    {
      "epoch": 0.2555241947840653,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001290882,
      "loss": 3.1157,
      "step": 354560
    },
    {
      "epoch": 0.2555428380202235,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001290862,
      "loss": 3.0757,
      "step": 354570
    },
    {
      "epoch": 0.2555614812563817,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001290842,
      "loss": 3.1699,
      "step": 354580
    },
    {
      "epoch": 0.25558012449253986,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001290822,
      "loss": 3.1481,
      "step": 354590
    },
    {
      "epoch": 0.25559876772869805,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012908020000000002,
      "loss": 3.2003,
      "step": 354600
    },
    {
      "epoch": 0.25561741096485624,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012907820000000001,
      "loss": 3.066,
      "step": 354610
    },
    {
      "epoch": 0.25563605420101443,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001290762,
      "loss": 3.164,
      "step": 354620
    },
    {
      "epoch": 0.25565469743717256,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001290742,
      "loss": 3.1057,
      "step": 354630
    },
    {
      "epoch": 0.25567334067333075,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001290722,
      "loss": 3.2579,
      "step": 354640
    },
    {
      "epoch": 0.25569198390948894,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001290702,
      "loss": 3.0853,
      "step": 354650
    },
    {
      "epoch": 0.25571062714564713,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012906820000000002,
      "loss": 3.1229,
      "step": 354660
    },
    {
      "epoch": 0.2557292703818053,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012906620000000001,
      "loss": 3.1926,
      "step": 354670
    },
    {
      "epoch": 0.2557479136179635,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001290642,
      "loss": 3.1318,
      "step": 354680
    },
    {
      "epoch": 0.2557665568541217,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001290622,
      "loss": 3.0442,
      "step": 354690
    },
    {
      "epoch": 0.2557852000902799,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001290602,
      "loss": 3.1647,
      "step": 354700
    },
    {
      "epoch": 0.2558038433264381,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001290582,
      "loss": 3.0215,
      "step": 354710
    },
    {
      "epoch": 0.25582248656259626,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001290562,
      "loss": 3.2498,
      "step": 354720
    },
    {
      "epoch": 0.2558411297987544,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012905420000000002,
      "loss": 3.0808,
      "step": 354730
    },
    {
      "epoch": 0.2558597730349126,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001290522,
      "loss": 3.073,
      "step": 354740
    },
    {
      "epoch": 0.2558784162710708,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001290502,
      "loss": 3.0477,
      "step": 354750
    },
    {
      "epoch": 0.25589705950722896,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001290482,
      "loss": 3.2221,
      "step": 354760
    },
    {
      "epoch": 0.25591570274338715,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001290462,
      "loss": 3.125,
      "step": 354770
    },
    {
      "epoch": 0.25593434597954534,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012904420000000002,
      "loss": 3.2001,
      "step": 354780
    },
    {
      "epoch": 0.25595298921570353,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001290422,
      "loss": 2.9937,
      "step": 354790
    },
    {
      "epoch": 0.2559716324518617,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001290402,
      "loss": 3.2057,
      "step": 354800
    },
    {
      "epoch": 0.2559902756880199,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001290382,
      "loss": 3.0292,
      "step": 354810
    },
    {
      "epoch": 0.2560089189241781,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001290362,
      "loss": 3.0041,
      "step": 354820
    },
    {
      "epoch": 0.2560275621603363,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012903420000000003,
      "loss": 3.025,
      "step": 354830
    },
    {
      "epoch": 0.2560462053964944,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001290322,
      "loss": 3.1554,
      "step": 354840
    },
    {
      "epoch": 0.2560648486326526,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012903020000000002,
      "loss": 3.0929,
      "step": 354850
    },
    {
      "epoch": 0.2560834918688108,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001290282,
      "loss": 3.0602,
      "step": 354860
    },
    {
      "epoch": 0.256102135104969,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001290262,
      "loss": 3.1381,
      "step": 354870
    },
    {
      "epoch": 0.2561207783411272,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001290242,
      "loss": 3.2005,
      "step": 354880
    },
    {
      "epoch": 0.25613942157728536,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001290222,
      "loss": 3.1589,
      "step": 354890
    },
    {
      "epoch": 0.25615806481344355,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012902020000000002,
      "loss": 3.0856,
      "step": 354900
    },
    {
      "epoch": 0.25617670804960174,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001290182,
      "loss": 3.1275,
      "step": 354910
    },
    {
      "epoch": 0.25619535128575993,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001290162,
      "loss": 3.1522,
      "step": 354920
    },
    {
      "epoch": 0.2562139945219181,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001290142,
      "loss": 3.1589,
      "step": 354930
    },
    {
      "epoch": 0.25623263775807625,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001290122,
      "loss": 3.1587,
      "step": 354940
    },
    {
      "epoch": 0.25625128099423444,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012901020000000003,
      "loss": 3.1641,
      "step": 354950
    },
    {
      "epoch": 0.25626992423039263,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001290082,
      "loss": 3.1054,
      "step": 354960
    },
    {
      "epoch": 0.2562885674665508,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012900620000000002,
      "loss": 3.0961,
      "step": 354970
    },
    {
      "epoch": 0.256307210702709,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001290042,
      "loss": 3.1446,
      "step": 354980
    },
    {
      "epoch": 0.2563258539388672,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001290022,
      "loss": 3.2887,
      "step": 354990
    },
    {
      "epoch": 0.2563444971750254,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001290002,
      "loss": 3.0845,
      "step": 355000
    },
    {
      "epoch": 0.2563631404111836,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001289982,
      "loss": 3.1898,
      "step": 355010
    },
    {
      "epoch": 0.25638178364734177,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012899620000000002,
      "loss": 3.1642,
      "step": 355020
    },
    {
      "epoch": 0.25640042688349995,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001289942,
      "loss": 3.1269,
      "step": 355030
    },
    {
      "epoch": 0.2564190701196581,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012899220000000001,
      "loss": 3.1195,
      "step": 355040
    },
    {
      "epoch": 0.2564377133558163,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001289902,
      "loss": 3.1956,
      "step": 355050
    },
    {
      "epoch": 0.25645635659197447,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001289882,
      "loss": 3.0823,
      "step": 355060
    },
    {
      "epoch": 0.25647499982813265,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001289862,
      "loss": 2.9712,
      "step": 355070
    },
    {
      "epoch": 0.25649364306429084,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001289842,
      "loss": 3.1197,
      "step": 355080
    },
    {
      "epoch": 0.25651228630044903,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012898220000000002,
      "loss": 3.0984,
      "step": 355090
    },
    {
      "epoch": 0.2565309295366072,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001289802,
      "loss": 3.1252,
      "step": 355100
    },
    {
      "epoch": 0.2565495727727654,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001289782,
      "loss": 3.151,
      "step": 355110
    },
    {
      "epoch": 0.2565682160089236,
      "grad_norm": 0.625,
      "learning_rate": 0.0001289762,
      "loss": 3.1778,
      "step": 355120
    },
    {
      "epoch": 0.2565868592450818,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001289742,
      "loss": 3.0834,
      "step": 355130
    },
    {
      "epoch": 0.25660550248124,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001289722,
      "loss": 3.1779,
      "step": 355140
    },
    {
      "epoch": 0.2566241457173981,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001289702,
      "loss": 3.0001,
      "step": 355150
    },
    {
      "epoch": 0.2566427889535563,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012896820000000002,
      "loss": 3.2383,
      "step": 355160
    },
    {
      "epoch": 0.2566614321897145,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001289662,
      "loss": 3.0488,
      "step": 355170
    },
    {
      "epoch": 0.2566800754258727,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001289642,
      "loss": 3.124,
      "step": 355180
    },
    {
      "epoch": 0.25669871866203087,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001289622,
      "loss": 3.1176,
      "step": 355190
    },
    {
      "epoch": 0.25671736189818906,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001289602,
      "loss": 3.128,
      "step": 355200
    },
    {
      "epoch": 0.25673600513434724,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001289582,
      "loss": 3.1718,
      "step": 355210
    },
    {
      "epoch": 0.25675464837050543,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012895620000000002,
      "loss": 3.0523,
      "step": 355220
    },
    {
      "epoch": 0.2567732916066636,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001289542,
      "loss": 3.1232,
      "step": 355230
    },
    {
      "epoch": 0.2567919348428218,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001289522,
      "loss": 3.1074,
      "step": 355240
    },
    {
      "epoch": 0.25681057807897995,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001289502,
      "loss": 3.1449,
      "step": 355250
    },
    {
      "epoch": 0.25682922131513813,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001289482,
      "loss": 3.1954,
      "step": 355260
    },
    {
      "epoch": 0.2568478645512963,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012894620000000002,
      "loss": 3.1511,
      "step": 355270
    },
    {
      "epoch": 0.2568665077874545,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001289442,
      "loss": 3.1228,
      "step": 355280
    },
    {
      "epoch": 0.2568851510236127,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012894220000000001,
      "loss": 3.0855,
      "step": 355290
    },
    {
      "epoch": 0.2569037942597709,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001289402,
      "loss": 3.1571,
      "step": 355300
    },
    {
      "epoch": 0.2569224374959291,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001289382,
      "loss": 3.0436,
      "step": 355310
    },
    {
      "epoch": 0.25694108073208727,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001289362,
      "loss": 3.1458,
      "step": 355320
    },
    {
      "epoch": 0.25695972396824546,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001289342,
      "loss": 3.1216,
      "step": 355330
    },
    {
      "epoch": 0.25697836720440365,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012893220000000002,
      "loss": 3.1609,
      "step": 355340
    },
    {
      "epoch": 0.25699701044056183,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001289302,
      "loss": 3.2011,
      "step": 355350
    },
    {
      "epoch": 0.25701565367671997,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001289282,
      "loss": 3.185,
      "step": 355360
    },
    {
      "epoch": 0.25703429691287816,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001289262,
      "loss": 3.26,
      "step": 355370
    },
    {
      "epoch": 0.25705294014903635,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001289242,
      "loss": 3.1482,
      "step": 355380
    },
    {
      "epoch": 0.25707158338519454,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012892220000000002,
      "loss": 3.1783,
      "step": 355390
    },
    {
      "epoch": 0.2570902266213527,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001289202,
      "loss": 3.0863,
      "step": 355400
    },
    {
      "epoch": 0.2571088698575109,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012891820000000002,
      "loss": 3.2295,
      "step": 355410
    },
    {
      "epoch": 0.2571275130936691,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001289162,
      "loss": 3.0043,
      "step": 355420
    },
    {
      "epoch": 0.2571461563298273,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001289142,
      "loss": 2.9823,
      "step": 355430
    },
    {
      "epoch": 0.2571647995659855,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012891220000000003,
      "loss": 3.1283,
      "step": 355440
    },
    {
      "epoch": 0.25718344280214367,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001289102,
      "loss": 3.1588,
      "step": 355450
    },
    {
      "epoch": 0.2572020860383018,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012890820000000002,
      "loss": 3.185,
      "step": 355460
    },
    {
      "epoch": 0.25722072927446,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001289062,
      "loss": 3.1048,
      "step": 355470
    },
    {
      "epoch": 0.2572393725106182,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001289042,
      "loss": 3.1635,
      "step": 355480
    },
    {
      "epoch": 0.25725801574677637,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001289022,
      "loss": 3.0417,
      "step": 355490
    },
    {
      "epoch": 0.25727665898293456,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001289002,
      "loss": 3.1452,
      "step": 355500
    },
    {
      "epoch": 0.25729530221909275,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012889820000000003,
      "loss": 3.0875,
      "step": 355510
    },
    {
      "epoch": 0.25731394545525094,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001288962,
      "loss": 3.1087,
      "step": 355520
    },
    {
      "epoch": 0.2573325886914091,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012889420000000002,
      "loss": 3.1376,
      "step": 355530
    },
    {
      "epoch": 0.2573512319275673,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012889219999999999,
      "loss": 3.1739,
      "step": 355540
    },
    {
      "epoch": 0.2573698751637255,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001288902,
      "loss": 3.1011,
      "step": 355550
    },
    {
      "epoch": 0.2573885183998837,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001288882,
      "loss": 3.1308,
      "step": 355560
    },
    {
      "epoch": 0.2574071616360418,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001288862,
      "loss": 3.2173,
      "step": 355570
    },
    {
      "epoch": 0.2574258048722,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012888420000000002,
      "loss": 3.0413,
      "step": 355580
    },
    {
      "epoch": 0.2574444481083582,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001288822,
      "loss": 3.1592,
      "step": 355590
    },
    {
      "epoch": 0.2574630913445164,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001288802,
      "loss": 3.1886,
      "step": 355600
    },
    {
      "epoch": 0.2574817345806746,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001288782,
      "loss": 3.0726,
      "step": 355610
    },
    {
      "epoch": 0.25750037781683277,
      "grad_norm": 0.75,
      "learning_rate": 0.0001288762,
      "loss": 3.057,
      "step": 355620
    },
    {
      "epoch": 0.25751902105299096,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001288742,
      "loss": 3.1181,
      "step": 355630
    },
    {
      "epoch": 0.25753766428914915,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001288722,
      "loss": 3.1315,
      "step": 355640
    },
    {
      "epoch": 0.25755630752530734,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012887020000000002,
      "loss": 3.1691,
      "step": 355650
    },
    {
      "epoch": 0.2575749507614655,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012886820000000001,
      "loss": 2.9327,
      "step": 355660
    },
    {
      "epoch": 0.25759359399762366,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001288662,
      "loss": 2.9662,
      "step": 355670
    },
    {
      "epoch": 0.25761223723378185,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001288642,
      "loss": 3.1052,
      "step": 355680
    },
    {
      "epoch": 0.25763088046994004,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001288622,
      "loss": 3.1168,
      "step": 355690
    },
    {
      "epoch": 0.2576495237060982,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001288602,
      "loss": 3.1064,
      "step": 355700
    },
    {
      "epoch": 0.2576681669422564,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001288582,
      "loss": 3.2681,
      "step": 355710
    },
    {
      "epoch": 0.2576868101784146,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012885620000000001,
      "loss": 3.1766,
      "step": 355720
    },
    {
      "epoch": 0.2577054534145728,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001288542,
      "loss": 3.2417,
      "step": 355730
    },
    {
      "epoch": 0.257724096650731,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001288522,
      "loss": 3.1225,
      "step": 355740
    },
    {
      "epoch": 0.25774273988688917,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001288502,
      "loss": 2.9602,
      "step": 355750
    },
    {
      "epoch": 0.25776138312304736,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001288482,
      "loss": 3.0888,
      "step": 355760
    },
    {
      "epoch": 0.25778002635920555,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001288462,
      "loss": 3.1396,
      "step": 355770
    },
    {
      "epoch": 0.2577986695953637,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012884420000000002,
      "loss": 3.2713,
      "step": 355780
    },
    {
      "epoch": 0.25781731283152187,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001288422,
      "loss": 3.0822,
      "step": 355790
    },
    {
      "epoch": 0.25783595606768006,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001288402,
      "loss": 3.0733,
      "step": 355800
    },
    {
      "epoch": 0.25785459930383825,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001288382,
      "loss": 3.2619,
      "step": 355810
    },
    {
      "epoch": 0.25787324253999644,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001288362,
      "loss": 3.0737,
      "step": 355820
    },
    {
      "epoch": 0.25789188577615463,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012883420000000002,
      "loss": 2.9569,
      "step": 355830
    },
    {
      "epoch": 0.2579105290123128,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001288322,
      "loss": 3.112,
      "step": 355840
    },
    {
      "epoch": 0.257929172248471,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001288302,
      "loss": 3.1866,
      "step": 355850
    },
    {
      "epoch": 0.2579478154846292,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001288282,
      "loss": 3.1225,
      "step": 355860
    },
    {
      "epoch": 0.2579664587207874,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001288262,
      "loss": 3.1879,
      "step": 355870
    },
    {
      "epoch": 0.2579851019569455,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012882420000000003,
      "loss": 3.1108,
      "step": 355880
    },
    {
      "epoch": 0.2580037451931037,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001288222,
      "loss": 3.0444,
      "step": 355890
    },
    {
      "epoch": 0.2580223884292619,
      "grad_norm": 0.75,
      "learning_rate": 0.00012882020000000002,
      "loss": 3.2509,
      "step": 355900
    },
    {
      "epoch": 0.2580410316654201,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001288182,
      "loss": 3.3002,
      "step": 355910
    },
    {
      "epoch": 0.2580596749015783,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001288162,
      "loss": 3.2466,
      "step": 355920
    },
    {
      "epoch": 0.25807831813773646,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001288142,
      "loss": 3.215,
      "step": 355930
    },
    {
      "epoch": 0.25809696137389465,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001288122,
      "loss": 3.144,
      "step": 355940
    },
    {
      "epoch": 0.25811560461005284,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012881020000000002,
      "loss": 3.1627,
      "step": 355950
    },
    {
      "epoch": 0.25813424784621103,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001288082,
      "loss": 3.1752,
      "step": 355960
    },
    {
      "epoch": 0.2581528910823692,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012880620000000001,
      "loss": 3.1676,
      "step": 355970
    },
    {
      "epoch": 0.2581715343185274,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001288042,
      "loss": 3.2178,
      "step": 355980
    },
    {
      "epoch": 0.25819017755468554,
      "grad_norm": 0.75,
      "learning_rate": 0.0001288022,
      "loss": 3.0333,
      "step": 355990
    },
    {
      "epoch": 0.25820882079084373,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012880020000000003,
      "loss": 3.0484,
      "step": 356000
    },
    {
      "epoch": 0.2582274640270019,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001287982,
      "loss": 3.2211,
      "step": 356010
    },
    {
      "epoch": 0.2582461072631601,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012879620000000002,
      "loss": 3.1432,
      "step": 356020
    },
    {
      "epoch": 0.2582647504993183,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001287942,
      "loss": 3.2721,
      "step": 356030
    },
    {
      "epoch": 0.2582833937354765,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001287922,
      "loss": 3.0211,
      "step": 356040
    },
    {
      "epoch": 0.2583020369716347,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001287902,
      "loss": 3.0344,
      "step": 356050
    },
    {
      "epoch": 0.25832068020779286,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001287882,
      "loss": 3.0661,
      "step": 356060
    },
    {
      "epoch": 0.25833932344395105,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012878620000000002,
      "loss": 3.0742,
      "step": 356070
    },
    {
      "epoch": 0.25835796668010924,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001287842,
      "loss": 3.1269,
      "step": 356080
    },
    {
      "epoch": 0.2583766099162674,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012878220000000002,
      "loss": 2.9732,
      "step": 356090
    },
    {
      "epoch": 0.25839525315242556,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012878019999999998,
      "loss": 3.1176,
      "step": 356100
    },
    {
      "epoch": 0.25841389638858375,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001287782,
      "loss": 3.1366,
      "step": 356110
    },
    {
      "epoch": 0.25843253962474194,
      "grad_norm": 0.75,
      "learning_rate": 0.0001287762,
      "loss": 3.0819,
      "step": 356120
    },
    {
      "epoch": 0.25845118286090013,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001287742,
      "loss": 3.0637,
      "step": 356130
    },
    {
      "epoch": 0.2584698260970583,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012877220000000002,
      "loss": 3.1907,
      "step": 356140
    },
    {
      "epoch": 0.2584884693332165,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001287702,
      "loss": 3.3003,
      "step": 356150
    },
    {
      "epoch": 0.2585071125693747,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001287682,
      "loss": 3.1753,
      "step": 356160
    },
    {
      "epoch": 0.2585257558055329,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001287662,
      "loss": 3.0735,
      "step": 356170
    },
    {
      "epoch": 0.2585443990416911,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001287642,
      "loss": 3.2512,
      "step": 356180
    },
    {
      "epoch": 0.25856304227784926,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001287622,
      "loss": 3.0644,
      "step": 356190
    },
    {
      "epoch": 0.2585816855140074,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001287602,
      "loss": 3.1876,
      "step": 356200
    },
    {
      "epoch": 0.2586003287501656,
      "grad_norm": 0.75,
      "learning_rate": 0.00012875820000000002,
      "loss": 3.1142,
      "step": 356210
    },
    {
      "epoch": 0.2586189719863238,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001287562,
      "loss": 3.08,
      "step": 356220
    },
    {
      "epoch": 0.25863761522248196,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001287542,
      "loss": 2.9665,
      "step": 356230
    },
    {
      "epoch": 0.25865625845864015,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001287522,
      "loss": 3.1114,
      "step": 356240
    },
    {
      "epoch": 0.25867490169479834,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001287502,
      "loss": 3.1899,
      "step": 356250
    },
    {
      "epoch": 0.25869354493095653,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001287482,
      "loss": 3.195,
      "step": 356260
    },
    {
      "epoch": 0.2587121881671147,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012874620000000002,
      "loss": 2.9779,
      "step": 356270
    },
    {
      "epoch": 0.2587308314032729,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001287442,
      "loss": 3.1493,
      "step": 356280
    },
    {
      "epoch": 0.2587494746394311,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001287422,
      "loss": 3.2492,
      "step": 356290
    },
    {
      "epoch": 0.25876811787558923,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001287402,
      "loss": 3.0646,
      "step": 356300
    },
    {
      "epoch": 0.2587867611117474,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001287382,
      "loss": 3.1162,
      "step": 356310
    },
    {
      "epoch": 0.2588054043479056,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001287362,
      "loss": 3.0936,
      "step": 356320
    },
    {
      "epoch": 0.2588240475840638,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001287342,
      "loss": 3.0065,
      "step": 356330
    },
    {
      "epoch": 0.258842690820222,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012873220000000001,
      "loss": 3.151,
      "step": 356340
    },
    {
      "epoch": 0.2588613340563802,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001287302,
      "loss": 3.12,
      "step": 356350
    },
    {
      "epoch": 0.25887997729253837,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001287282,
      "loss": 3.1099,
      "step": 356360
    },
    {
      "epoch": 0.25889862052869655,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001287262,
      "loss": 3.2007,
      "step": 356370
    },
    {
      "epoch": 0.25891726376485474,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001287242,
      "loss": 3.0481,
      "step": 356380
    },
    {
      "epoch": 0.25893590700101293,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012872220000000002,
      "loss": 3.0709,
      "step": 356390
    },
    {
      "epoch": 0.2589545502371711,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012872020000000001,
      "loss": 3.187,
      "step": 356400
    },
    {
      "epoch": 0.25897319347332926,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001287182,
      "loss": 3.0739,
      "step": 356410
    },
    {
      "epoch": 0.25899183670948744,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001287162,
      "loss": 3.0334,
      "step": 356420
    },
    {
      "epoch": 0.25901047994564563,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001287142,
      "loss": 3.2572,
      "step": 356430
    },
    {
      "epoch": 0.2590291231818038,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012871220000000002,
      "loss": 3.1772,
      "step": 356440
    },
    {
      "epoch": 0.259047766417962,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001287102,
      "loss": 3.0941,
      "step": 356450
    },
    {
      "epoch": 0.2590664096541202,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012870820000000002,
      "loss": 2.9797,
      "step": 356460
    },
    {
      "epoch": 0.2590850528902784,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001287062,
      "loss": 3.0723,
      "step": 356470
    },
    {
      "epoch": 0.2591036961264366,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001287042,
      "loss": 3.0339,
      "step": 356480
    },
    {
      "epoch": 0.25912233936259477,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012870220000000003,
      "loss": 3.1417,
      "step": 356490
    },
    {
      "epoch": 0.25914098259875296,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001287002,
      "loss": 3.0621,
      "step": 356500
    },
    {
      "epoch": 0.2591596258349111,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012869820000000002,
      "loss": 3.1083,
      "step": 356510
    },
    {
      "epoch": 0.2591782690710693,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001286962,
      "loss": 3.0667,
      "step": 356520
    },
    {
      "epoch": 0.25919691230722747,
      "grad_norm": 0.75,
      "learning_rate": 0.0001286942,
      "loss": 2.9992,
      "step": 356530
    },
    {
      "epoch": 0.25921555554338566,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001286922,
      "loss": 3.1272,
      "step": 356540
    },
    {
      "epoch": 0.25923419877954385,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001286902,
      "loss": 3.1867,
      "step": 356550
    },
    {
      "epoch": 0.25925284201570203,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012868820000000003,
      "loss": 3.133,
      "step": 356560
    },
    {
      "epoch": 0.2592714852518602,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001286862,
      "loss": 3.0469,
      "step": 356570
    },
    {
      "epoch": 0.2592901284880184,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012868420000000002,
      "loss": 3.0493,
      "step": 356580
    },
    {
      "epoch": 0.2593087717241766,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012868219999999999,
      "loss": 3.2945,
      "step": 356590
    },
    {
      "epoch": 0.2593274149603348,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001286802,
      "loss": 3.0787,
      "step": 356600
    },
    {
      "epoch": 0.2593460581964929,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001286782,
      "loss": 3.1255,
      "step": 356610
    },
    {
      "epoch": 0.2593647014326511,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001286762,
      "loss": 3.2159,
      "step": 356620
    },
    {
      "epoch": 0.2593833446688093,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012867420000000002,
      "loss": 3.0855,
      "step": 356630
    },
    {
      "epoch": 0.2594019879049675,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001286722,
      "loss": 3.1311,
      "step": 356640
    },
    {
      "epoch": 0.2594206311411257,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012867020000000001,
      "loss": 3.0975,
      "step": 356650
    },
    {
      "epoch": 0.25943927437728387,
      "grad_norm": 0.75,
      "learning_rate": 0.0001286682,
      "loss": 3.0367,
      "step": 356660
    },
    {
      "epoch": 0.25945791761344206,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001286662,
      "loss": 3.1205,
      "step": 356670
    },
    {
      "epoch": 0.25947656084960025,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001286642,
      "loss": 3.0612,
      "step": 356680
    },
    {
      "epoch": 0.25949520408575844,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001286622,
      "loss": 3.1315,
      "step": 356690
    },
    {
      "epoch": 0.2595138473219166,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012866020000000002,
      "loss": 3.213,
      "step": 356700
    },
    {
      "epoch": 0.2595324905580748,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001286582,
      "loss": 3.0876,
      "step": 356710
    },
    {
      "epoch": 0.25955113379423295,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001286562,
      "loss": 3.1318,
      "step": 356720
    },
    {
      "epoch": 0.25956977703039114,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001286542,
      "loss": 3.0727,
      "step": 356730
    },
    {
      "epoch": 0.2595884202665493,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001286522,
      "loss": 3.0452,
      "step": 356740
    },
    {
      "epoch": 0.2596070635027075,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001286502,
      "loss": 3.1253,
      "step": 356750
    },
    {
      "epoch": 0.2596257067388657,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001286482,
      "loss": 3.1098,
      "step": 356760
    },
    {
      "epoch": 0.2596443499750239,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012864620000000002,
      "loss": 3.1672,
      "step": 356770
    },
    {
      "epoch": 0.2596629932111821,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001286442,
      "loss": 2.9832,
      "step": 356780
    },
    {
      "epoch": 0.25968163644734027,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001286422,
      "loss": 3.0022,
      "step": 356790
    },
    {
      "epoch": 0.25970027968349846,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001286402,
      "loss": 3.0673,
      "step": 356800
    },
    {
      "epoch": 0.25971892291965665,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001286382,
      "loss": 3.1675,
      "step": 356810
    },
    {
      "epoch": 0.2597375661558148,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001286362,
      "loss": 3.185,
      "step": 356820
    },
    {
      "epoch": 0.25975620939197297,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012863420000000002,
      "loss": 3.141,
      "step": 356830
    },
    {
      "epoch": 0.25977485262813116,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001286322,
      "loss": 3.0863,
      "step": 356840
    },
    {
      "epoch": 0.25979349586428935,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001286302,
      "loss": 3.0085,
      "step": 356850
    },
    {
      "epoch": 0.25981213910044754,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001286282,
      "loss": 3.0347,
      "step": 356860
    },
    {
      "epoch": 0.2598307823366057,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001286262,
      "loss": 3.0279,
      "step": 356870
    },
    {
      "epoch": 0.2598494255727639,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012862420000000002,
      "loss": 2.9647,
      "step": 356880
    },
    {
      "epoch": 0.2598680688089221,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012862220000000002,
      "loss": 3.217,
      "step": 356890
    },
    {
      "epoch": 0.2598867120450803,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001286202,
      "loss": 2.9674,
      "step": 356900
    },
    {
      "epoch": 0.2599053552812385,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001286182,
      "loss": 3.1307,
      "step": 356910
    },
    {
      "epoch": 0.25992399851739667,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001286162,
      "loss": 3.037,
      "step": 356920
    },
    {
      "epoch": 0.2599426417535548,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001286142,
      "loss": 3.106,
      "step": 356930
    },
    {
      "epoch": 0.259961284989713,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001286122,
      "loss": 3.1448,
      "step": 356940
    },
    {
      "epoch": 0.2599799282258712,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012861020000000002,
      "loss": 3.1165,
      "step": 356950
    },
    {
      "epoch": 0.25999857146202937,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001286082,
      "loss": 3.1533,
      "step": 356960
    },
    {
      "epoch": 0.26001721469818756,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001286062,
      "loss": 3.0966,
      "step": 356970
    },
    {
      "epoch": 0.26003585793434575,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001286042,
      "loss": 3.1108,
      "step": 356980
    },
    {
      "epoch": 0.26005450117050394,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001286022,
      "loss": 3.0626,
      "step": 356990
    },
    {
      "epoch": 0.2600731444066621,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012860020000000002,
      "loss": 3.1176,
      "step": 357000
    },
    {
      "epoch": 0.2600917876428203,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001285982,
      "loss": 3.1241,
      "step": 357010
    },
    {
      "epoch": 0.2601104308789785,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012859620000000001,
      "loss": 3.1259,
      "step": 357020
    },
    {
      "epoch": 0.26012907411513664,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001285942,
      "loss": 3.2547,
      "step": 357030
    },
    {
      "epoch": 0.2601477173512948,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001285922,
      "loss": 2.9889,
      "step": 357040
    },
    {
      "epoch": 0.260166360587453,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012859020000000003,
      "loss": 3.1401,
      "step": 357050
    },
    {
      "epoch": 0.2601850038236112,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001285882,
      "loss": 3.0377,
      "step": 357060
    },
    {
      "epoch": 0.2602036470597694,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012858620000000002,
      "loss": 3.1739,
      "step": 357070
    },
    {
      "epoch": 0.2602222902959276,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001285842,
      "loss": 3.0445,
      "step": 357080
    },
    {
      "epoch": 0.26024093353208577,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001285822,
      "loss": 3.2257,
      "step": 357090
    },
    {
      "epoch": 0.26025957676824396,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001285802,
      "loss": 3.1446,
      "step": 357100
    },
    {
      "epoch": 0.26027822000440215,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001285782,
      "loss": 3.1562,
      "step": 357110
    },
    {
      "epoch": 0.26029686324056034,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012857620000000002,
      "loss": 3.143,
      "step": 357120
    },
    {
      "epoch": 0.26031550647671853,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001285742,
      "loss": 3.154,
      "step": 357130
    },
    {
      "epoch": 0.26033414971287666,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012857220000000002,
      "loss": 3.1354,
      "step": 357140
    },
    {
      "epoch": 0.26035279294903485,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012857019999999998,
      "loss": 3.07,
      "step": 357150
    },
    {
      "epoch": 0.26037143618519304,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001285682,
      "loss": 3.1383,
      "step": 357160
    },
    {
      "epoch": 0.26039007942135123,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001285662,
      "loss": 3.0847,
      "step": 357170
    },
    {
      "epoch": 0.2604087226575094,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001285642,
      "loss": 3.1351,
      "step": 357180
    },
    {
      "epoch": 0.2604273658936676,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012856220000000002,
      "loss": 3.1058,
      "step": 357190
    },
    {
      "epoch": 0.2604460091298258,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001285602,
      "loss": 3.134,
      "step": 357200
    },
    {
      "epoch": 0.260464652365984,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001285582,
      "loss": 3.2248,
      "step": 357210
    },
    {
      "epoch": 0.2604832956021422,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001285562,
      "loss": 3.0194,
      "step": 357220
    },
    {
      "epoch": 0.26050193883830036,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001285542,
      "loss": 3.1027,
      "step": 357230
    },
    {
      "epoch": 0.2605205820744585,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001285522,
      "loss": 3.147,
      "step": 357240
    },
    {
      "epoch": 0.2605392253106167,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001285502,
      "loss": 3.173,
      "step": 357250
    },
    {
      "epoch": 0.2605578685467749,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012854820000000002,
      "loss": 3.0892,
      "step": 357260
    },
    {
      "epoch": 0.26057651178293306,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001285462,
      "loss": 3.0988,
      "step": 357270
    },
    {
      "epoch": 0.26059515501909125,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001285442,
      "loss": 3.1618,
      "step": 357280
    },
    {
      "epoch": 0.26061379825524944,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001285422,
      "loss": 3.0802,
      "step": 357290
    },
    {
      "epoch": 0.26063244149140763,
      "grad_norm": 0.75,
      "learning_rate": 0.0001285402,
      "loss": 3.169,
      "step": 357300
    },
    {
      "epoch": 0.2606510847275658,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001285382,
      "loss": 3.0943,
      "step": 357310
    },
    {
      "epoch": 0.260669727963724,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001285362,
      "loss": 3.0611,
      "step": 357320
    },
    {
      "epoch": 0.2606883711998822,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012853420000000001,
      "loss": 3.1779,
      "step": 357330
    },
    {
      "epoch": 0.2607070144360404,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001285322,
      "loss": 3.1296,
      "step": 357340
    },
    {
      "epoch": 0.2607256576721985,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001285302,
      "loss": 3.1986,
      "step": 357350
    },
    {
      "epoch": 0.2607443009083567,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001285282,
      "loss": 2.9519,
      "step": 357360
    },
    {
      "epoch": 0.2607629441445149,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001285262,
      "loss": 3.0999,
      "step": 357370
    },
    {
      "epoch": 0.2607815873806731,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012852420000000002,
      "loss": 3.1846,
      "step": 357380
    },
    {
      "epoch": 0.2608002306168313,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012852220000000001,
      "loss": 3.0698,
      "step": 357390
    },
    {
      "epoch": 0.26081887385298946,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001285202,
      "loss": 2.9539,
      "step": 357400
    },
    {
      "epoch": 0.26083751708914765,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001285182,
      "loss": 3.0998,
      "step": 357410
    },
    {
      "epoch": 0.26085616032530584,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001285162,
      "loss": 2.9324,
      "step": 357420
    },
    {
      "epoch": 0.26087480356146403,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001285142,
      "loss": 3.2412,
      "step": 357430
    },
    {
      "epoch": 0.2608934467976222,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012851220000000002,
      "loss": 3.1408,
      "step": 357440
    },
    {
      "epoch": 0.26091209003378035,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012851020000000002,
      "loss": 3.0891,
      "step": 357450
    },
    {
      "epoch": 0.26093073326993854,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001285082,
      "loss": 3.0124,
      "step": 357460
    },
    {
      "epoch": 0.26094937650609673,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001285062,
      "loss": 3.1196,
      "step": 357470
    },
    {
      "epoch": 0.2609680197422549,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001285042,
      "loss": 2.9649,
      "step": 357480
    },
    {
      "epoch": 0.2609866629784131,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012850220000000002,
      "loss": 3.0705,
      "step": 357490
    },
    {
      "epoch": 0.2610053062145713,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001285002,
      "loss": 3.0793,
      "step": 357500
    },
    {
      "epoch": 0.2610239494507295,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012849820000000002,
      "loss": 3.2076,
      "step": 357510
    },
    {
      "epoch": 0.2610425926868877,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001284962,
      "loss": 2.9895,
      "step": 357520
    },
    {
      "epoch": 0.26106123592304586,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001284942,
      "loss": 3.1612,
      "step": 357530
    },
    {
      "epoch": 0.26107987915920405,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001284922,
      "loss": 3.2129,
      "step": 357540
    },
    {
      "epoch": 0.26109852239536224,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001284902,
      "loss": 3.2002,
      "step": 357550
    },
    {
      "epoch": 0.2611171656315204,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012848820000000002,
      "loss": 3.1532,
      "step": 357560
    },
    {
      "epoch": 0.26113580886767856,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001284862,
      "loss": 3.2466,
      "step": 357570
    },
    {
      "epoch": 0.26115445210383675,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001284842,
      "loss": 3.0887,
      "step": 357580
    },
    {
      "epoch": 0.26117309533999494,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001284822,
      "loss": 3.0531,
      "step": 357590
    },
    {
      "epoch": 0.26119173857615313,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001284802,
      "loss": 3.0005,
      "step": 357600
    },
    {
      "epoch": 0.2612103818123113,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012847820000000003,
      "loss": 3.0451,
      "step": 357610
    },
    {
      "epoch": 0.2612290250484695,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001284762,
      "loss": 3.1365,
      "step": 357620
    },
    {
      "epoch": 0.2612476682846277,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012847420000000002,
      "loss": 3.1568,
      "step": 357630
    },
    {
      "epoch": 0.2612663115207859,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012847219999999999,
      "loss": 3.2087,
      "step": 357640
    },
    {
      "epoch": 0.2612849547569441,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001284702,
      "loss": 3.1929,
      "step": 357650
    },
    {
      "epoch": 0.2613035979931022,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001284682,
      "loss": 3.2796,
      "step": 357660
    },
    {
      "epoch": 0.2613222412292604,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001284662,
      "loss": 3.048,
      "step": 357670
    },
    {
      "epoch": 0.2613408844654186,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012846420000000002,
      "loss": 3.1142,
      "step": 357680
    },
    {
      "epoch": 0.2613595277015768,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001284622,
      "loss": 3.0961,
      "step": 357690
    },
    {
      "epoch": 0.26137817093773497,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012846020000000001,
      "loss": 3.0099,
      "step": 357700
    },
    {
      "epoch": 0.26139681417389315,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012845819999999998,
      "loss": 2.999,
      "step": 357710
    },
    {
      "epoch": 0.26141545741005134,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001284562,
      "loss": 3.0886,
      "step": 357720
    },
    {
      "epoch": 0.26143410064620953,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001284542,
      "loss": 3.0643,
      "step": 357730
    },
    {
      "epoch": 0.2614527438823677,
      "grad_norm": 0.875,
      "learning_rate": 0.0001284522,
      "loss": 3.1289,
      "step": 357740
    },
    {
      "epoch": 0.2614713871185259,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012845020000000002,
      "loss": 3.184,
      "step": 357750
    },
    {
      "epoch": 0.2614900303546841,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001284482,
      "loss": 3.1123,
      "step": 357760
    },
    {
      "epoch": 0.26150867359084223,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001284462,
      "loss": 3.0358,
      "step": 357770
    },
    {
      "epoch": 0.2615273168270004,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001284442,
      "loss": 3.1599,
      "step": 357780
    },
    {
      "epoch": 0.2615459600631586,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001284422,
      "loss": 3.0379,
      "step": 357790
    },
    {
      "epoch": 0.2615646032993168,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001284402,
      "loss": 3.2035,
      "step": 357800
    },
    {
      "epoch": 0.261583246535475,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001284382,
      "loss": 3.3479,
      "step": 357810
    },
    {
      "epoch": 0.2616018897716332,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012843620000000002,
      "loss": 3.1038,
      "step": 357820
    },
    {
      "epoch": 0.26162053300779137,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001284342,
      "loss": 3.1247,
      "step": 357830
    },
    {
      "epoch": 0.26163917624394956,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001284322,
      "loss": 3.1693,
      "step": 357840
    },
    {
      "epoch": 0.26165781948010775,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001284302,
      "loss": 3.0875,
      "step": 357850
    },
    {
      "epoch": 0.26167646271626593,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001284282,
      "loss": 3.089,
      "step": 357860
    },
    {
      "epoch": 0.26169510595242407,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012842620000000002,
      "loss": 3.0212,
      "step": 357870
    },
    {
      "epoch": 0.26171374918858226,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012842420000000002,
      "loss": 3.1516,
      "step": 357880
    },
    {
      "epoch": 0.26173239242474045,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001284222,
      "loss": 3.1036,
      "step": 357890
    },
    {
      "epoch": 0.26175103566089863,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001284202,
      "loss": 3.0824,
      "step": 357900
    },
    {
      "epoch": 0.2617696788970568,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001284182,
      "loss": 3.1034,
      "step": 357910
    },
    {
      "epoch": 0.261788322133215,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001284162,
      "loss": 3.129,
      "step": 357920
    },
    {
      "epoch": 0.2618069653693732,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001284142,
      "loss": 3.0357,
      "step": 357930
    },
    {
      "epoch": 0.2618256086055314,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012841220000000002,
      "loss": 3.1658,
      "step": 357940
    },
    {
      "epoch": 0.2618442518416896,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001284102,
      "loss": 3.1349,
      "step": 357950
    },
    {
      "epoch": 0.26186289507784777,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001284082,
      "loss": 3.1494,
      "step": 357960
    },
    {
      "epoch": 0.2618815383140059,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001284062,
      "loss": 3.1083,
      "step": 357970
    },
    {
      "epoch": 0.2619001815501641,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001284042,
      "loss": 3.0537,
      "step": 357980
    },
    {
      "epoch": 0.2619188247863223,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001284022,
      "loss": 3.1084,
      "step": 357990
    },
    {
      "epoch": 0.26193746802248047,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012840020000000002,
      "loss": 3.0217,
      "step": 358000
    },
    {
      "epoch": 0.26195611125863866,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001283982,
      "loss": 3.1931,
      "step": 358010
    },
    {
      "epoch": 0.26197475449479685,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001283962,
      "loss": 3.205,
      "step": 358020
    },
    {
      "epoch": 0.26199339773095504,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001283942,
      "loss": 3.1155,
      "step": 358030
    },
    {
      "epoch": 0.2620120409671132,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001283922,
      "loss": 3.1403,
      "step": 358040
    },
    {
      "epoch": 0.2620306842032714,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012839020000000002,
      "loss": 3.1061,
      "step": 358050
    },
    {
      "epoch": 0.2620493274394296,
      "grad_norm": 0.625,
      "learning_rate": 0.0001283882,
      "loss": 3.2231,
      "step": 358060
    },
    {
      "epoch": 0.2620679706755878,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012838620000000001,
      "loss": 3.1245,
      "step": 358070
    },
    {
      "epoch": 0.2620866139117459,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001283842,
      "loss": 3.117,
      "step": 358080
    },
    {
      "epoch": 0.2621052571479041,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001283822,
      "loss": 3.0804,
      "step": 358090
    },
    {
      "epoch": 0.2621239003840623,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012838020000000003,
      "loss": 3.0146,
      "step": 358100
    },
    {
      "epoch": 0.2621425436202205,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001283782,
      "loss": 3.125,
      "step": 358110
    },
    {
      "epoch": 0.2621611868563787,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012837620000000002,
      "loss": 3.1975,
      "step": 358120
    },
    {
      "epoch": 0.26217983009253687,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001283742,
      "loss": 3.1662,
      "step": 358130
    },
    {
      "epoch": 0.26219847332869506,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001283722,
      "loss": 3.1443,
      "step": 358140
    },
    {
      "epoch": 0.26221711656485325,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001283702,
      "loss": 3.1298,
      "step": 358150
    },
    {
      "epoch": 0.26223575980101144,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001283682,
      "loss": 3.1746,
      "step": 358160
    },
    {
      "epoch": 0.2622544030371696,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012836620000000002,
      "loss": 2.9382,
      "step": 358170
    },
    {
      "epoch": 0.26227304627332776,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001283642,
      "loss": 3.016,
      "step": 358180
    },
    {
      "epoch": 0.26229168950948595,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012836220000000002,
      "loss": 3.1497,
      "step": 358190
    },
    {
      "epoch": 0.26231033274564414,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012836019999999998,
      "loss": 3.0836,
      "step": 358200
    },
    {
      "epoch": 0.2623289759818023,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001283582,
      "loss": 3.0258,
      "step": 358210
    },
    {
      "epoch": 0.2623476192179605,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001283562,
      "loss": 3.09,
      "step": 358220
    },
    {
      "epoch": 0.2623662624541187,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001283542,
      "loss": 3.1456,
      "step": 358230
    },
    {
      "epoch": 0.2623849056902769,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012835220000000002,
      "loss": 3.1554,
      "step": 358240
    },
    {
      "epoch": 0.2624035489264351,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001283502,
      "loss": 3.1535,
      "step": 358250
    },
    {
      "epoch": 0.26242219216259327,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001283482,
      "loss": 3.134,
      "step": 358260
    },
    {
      "epoch": 0.26244083539875146,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001283462,
      "loss": 3.1983,
      "step": 358270
    },
    {
      "epoch": 0.26245947863490965,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001283442,
      "loss": 3.1015,
      "step": 358280
    },
    {
      "epoch": 0.2624781218710678,
      "grad_norm": 0.75,
      "learning_rate": 0.0001283422,
      "loss": 3.1083,
      "step": 358290
    },
    {
      "epoch": 0.26249676510722597,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001283402,
      "loss": 3.0012,
      "step": 358300
    },
    {
      "epoch": 0.26251540834338416,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012833820000000002,
      "loss": 3.1659,
      "step": 358310
    },
    {
      "epoch": 0.26253405157954235,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00012833619999999999,
      "loss": 3.145,
      "step": 358320
    },
    {
      "epoch": 0.26255269481570054,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001283342,
      "loss": 3.02,
      "step": 358330
    },
    {
      "epoch": 0.2625713380518587,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001283322,
      "loss": 3.1388,
      "step": 358340
    },
    {
      "epoch": 0.2625899812880169,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001283302,
      "loss": 3.1554,
      "step": 358350
    },
    {
      "epoch": 0.2626086245241751,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012832820000000002,
      "loss": 3.115,
      "step": 358360
    },
    {
      "epoch": 0.2626272677603333,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001283262,
      "loss": 3.2091,
      "step": 358370
    },
    {
      "epoch": 0.2626459109964915,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012832420000000001,
      "loss": 3.1356,
      "step": 358380
    },
    {
      "epoch": 0.2626645542326496,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001283222,
      "loss": 3.3331,
      "step": 358390
    },
    {
      "epoch": 0.2626831974688078,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001283202,
      "loss": 3.1114,
      "step": 358400
    },
    {
      "epoch": 0.262701840704966,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001283182,
      "loss": 3.1059,
      "step": 358410
    },
    {
      "epoch": 0.2627204839411242,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001283162,
      "loss": 3.1653,
      "step": 358420
    },
    {
      "epoch": 0.26273912717728237,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012831420000000002,
      "loss": 3.1098,
      "step": 358430
    },
    {
      "epoch": 0.26275777041344056,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012831220000000001,
      "loss": 3.1227,
      "step": 358440
    },
    {
      "epoch": 0.26277641364959875,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001283102,
      "loss": 3.2283,
      "step": 358450
    },
    {
      "epoch": 0.26279505688575694,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001283082,
      "loss": 3.1637,
      "step": 358460
    },
    {
      "epoch": 0.26281370012191513,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001283062,
      "loss": 3.1846,
      "step": 358470
    },
    {
      "epoch": 0.2628323433580733,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001283042,
      "loss": 3.1801,
      "step": 358480
    },
    {
      "epoch": 0.2628509865942315,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012830220000000002,
      "loss": 3.2132,
      "step": 358490
    },
    {
      "epoch": 0.26286962983038964,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012830020000000002,
      "loss": 3.092,
      "step": 358500
    },
    {
      "epoch": 0.26288827306654783,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001282982,
      "loss": 3.1079,
      "step": 358510
    },
    {
      "epoch": 0.262906916302706,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001282962,
      "loss": 3.0195,
      "step": 358520
    },
    {
      "epoch": 0.2629255595388642,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001282942,
      "loss": 3.1248,
      "step": 358530
    },
    {
      "epoch": 0.2629442027750224,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001282922,
      "loss": 3.1099,
      "step": 358540
    },
    {
      "epoch": 0.2629628460111806,
      "grad_norm": 0.75,
      "learning_rate": 0.0001282902,
      "loss": 3.1742,
      "step": 358550
    },
    {
      "epoch": 0.2629814892473388,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012828820000000002,
      "loss": 3.1129,
      "step": 358560
    },
    {
      "epoch": 0.26300013248349696,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001282862,
      "loss": 3.1199,
      "step": 358570
    },
    {
      "epoch": 0.26301877571965515,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001282842,
      "loss": 3.0256,
      "step": 358580
    },
    {
      "epoch": 0.26303741895581334,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001282822,
      "loss": 3.0549,
      "step": 358590
    },
    {
      "epoch": 0.2630560621919715,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001282802,
      "loss": 3.103,
      "step": 358600
    },
    {
      "epoch": 0.26307470542812966,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012827820000000002,
      "loss": 3.1055,
      "step": 358610
    },
    {
      "epoch": 0.26309334866428785,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001282762,
      "loss": 3.2123,
      "step": 358620
    },
    {
      "epoch": 0.26311199190044604,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001282742,
      "loss": 3.0164,
      "step": 358630
    },
    {
      "epoch": 0.26313063513660423,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001282722,
      "loss": 3.1263,
      "step": 358640
    },
    {
      "epoch": 0.2631492783727624,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001282702,
      "loss": 3.129,
      "step": 358650
    },
    {
      "epoch": 0.2631679216089206,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012826820000000003,
      "loss": 3.0839,
      "step": 358660
    },
    {
      "epoch": 0.2631865648450788,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001282662,
      "loss": 3.1458,
      "step": 358670
    },
    {
      "epoch": 0.263205208081237,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012826420000000002,
      "loss": 3.1209,
      "step": 358680
    },
    {
      "epoch": 0.2632238513173952,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012826219999999999,
      "loss": 3.1092,
      "step": 358690
    },
    {
      "epoch": 0.26324249455355336,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001282602,
      "loss": 3.0967,
      "step": 358700
    },
    {
      "epoch": 0.2632611377897115,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001282582,
      "loss": 3.0369,
      "step": 358710
    },
    {
      "epoch": 0.2632797810258697,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001282562,
      "loss": 3.1472,
      "step": 358720
    },
    {
      "epoch": 0.2632984242620279,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012825420000000002,
      "loss": 3.1116,
      "step": 358730
    },
    {
      "epoch": 0.26331706749818606,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001282522,
      "loss": 3.1073,
      "step": 358740
    },
    {
      "epoch": 0.26333571073434425,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012825020000000001,
      "loss": 3.1681,
      "step": 358750
    },
    {
      "epoch": 0.26335435397050244,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012824819999999998,
      "loss": 3.0624,
      "step": 358760
    },
    {
      "epoch": 0.26337299720666063,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001282462,
      "loss": 3.0169,
      "step": 358770
    },
    {
      "epoch": 0.2633916404428188,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001282442,
      "loss": 3.1776,
      "step": 358780
    },
    {
      "epoch": 0.263410283678977,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001282422,
      "loss": 2.8713,
      "step": 358790
    },
    {
      "epoch": 0.2634289269151352,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012824020000000002,
      "loss": 3.1259,
      "step": 358800
    },
    {
      "epoch": 0.26344757015129333,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001282382,
      "loss": 3.174,
      "step": 358810
    },
    {
      "epoch": 0.2634662133874515,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001282362,
      "loss": 3.2488,
      "step": 358820
    },
    {
      "epoch": 0.2634848566236097,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001282342,
      "loss": 3.2871,
      "step": 358830
    },
    {
      "epoch": 0.2635034998597679,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001282322,
      "loss": 3.3012,
      "step": 358840
    },
    {
      "epoch": 0.2635221430959261,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012823020000000002,
      "loss": 3.1025,
      "step": 358850
    },
    {
      "epoch": 0.2635407863320843,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001282282,
      "loss": 3.1419,
      "step": 358860
    },
    {
      "epoch": 0.26355942956824246,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012822620000000002,
      "loss": 3.0265,
      "step": 358870
    },
    {
      "epoch": 0.26357807280440065,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001282242,
      "loss": 3.2578,
      "step": 358880
    },
    {
      "epoch": 0.26359671604055884,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001282222,
      "loss": 3.1875,
      "step": 358890
    },
    {
      "epoch": 0.26361535927671703,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001282202,
      "loss": 3.1401,
      "step": 358900
    },
    {
      "epoch": 0.2636340025128752,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001282182,
      "loss": 2.9895,
      "step": 358910
    },
    {
      "epoch": 0.26365264574903335,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012821620000000002,
      "loss": 3.2071,
      "step": 358920
    },
    {
      "epoch": 0.26367128898519154,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001282142,
      "loss": 3.0911,
      "step": 358930
    },
    {
      "epoch": 0.26368993222134973,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001282122,
      "loss": 3.0574,
      "step": 358940
    },
    {
      "epoch": 0.2637085754575079,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001282102,
      "loss": 3.1021,
      "step": 358950
    },
    {
      "epoch": 0.2637272186936661,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001282082,
      "loss": 3.1802,
      "step": 358960
    },
    {
      "epoch": 0.2637458619298243,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001282062,
      "loss": 3.1464,
      "step": 358970
    },
    {
      "epoch": 0.2637645051659825,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001282042,
      "loss": 3.0891,
      "step": 358980
    },
    {
      "epoch": 0.2637831484021407,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012820220000000002,
      "loss": 3.1323,
      "step": 358990
    },
    {
      "epoch": 0.26380179163829887,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001282002,
      "loss": 2.9951,
      "step": 359000
    },
    {
      "epoch": 0.26382043487445705,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001281982,
      "loss": 2.9647,
      "step": 359010
    },
    {
      "epoch": 0.2638390781106152,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001281962,
      "loss": 3.1481,
      "step": 359020
    },
    {
      "epoch": 0.2638577213467734,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001281942,
      "loss": 3.0975,
      "step": 359030
    },
    {
      "epoch": 0.26387636458293157,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001281922,
      "loss": 3.0647,
      "step": 359040
    },
    {
      "epoch": 0.26389500781908976,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012819020000000002,
      "loss": 3.1767,
      "step": 359050
    },
    {
      "epoch": 0.26391365105524794,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012818820000000001,
      "loss": 3.2115,
      "step": 359060
    },
    {
      "epoch": 0.26393229429140613,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001281862,
      "loss": 3.1136,
      "step": 359070
    },
    {
      "epoch": 0.2639509375275643,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001281842,
      "loss": 2.997,
      "step": 359080
    },
    {
      "epoch": 0.2639695807637225,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001281822,
      "loss": 3.0487,
      "step": 359090
    },
    {
      "epoch": 0.2639882239998807,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012818020000000002,
      "loss": 3.013,
      "step": 359100
    },
    {
      "epoch": 0.2640068672360389,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001281782,
      "loss": 3.0394,
      "step": 359110
    },
    {
      "epoch": 0.2640255104721971,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012817620000000001,
      "loss": 3.1298,
      "step": 359120
    },
    {
      "epoch": 0.2640441537083552,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001281742,
      "loss": 3.1518,
      "step": 359130
    },
    {
      "epoch": 0.2640627969445134,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001281722,
      "loss": 3.0941,
      "step": 359140
    },
    {
      "epoch": 0.2640814401806716,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001281702,
      "loss": 3.0807,
      "step": 359150
    },
    {
      "epoch": 0.2641000834168298,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001281682,
      "loss": 3.0878,
      "step": 359160
    },
    {
      "epoch": 0.26411872665298797,
      "grad_norm": 0.75,
      "learning_rate": 0.00012816620000000002,
      "loss": 3.2786,
      "step": 359170
    },
    {
      "epoch": 0.26413736988914616,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001281642,
      "loss": 3.0661,
      "step": 359180
    },
    {
      "epoch": 0.26415601312530435,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001281622,
      "loss": 3.1673,
      "step": 359190
    },
    {
      "epoch": 0.26417465636146253,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001281602,
      "loss": 3.1513,
      "step": 359200
    },
    {
      "epoch": 0.2641932995976207,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001281582,
      "loss": 3.0927,
      "step": 359210
    },
    {
      "epoch": 0.2642119428337789,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012815620000000003,
      "loss": 3.1441,
      "step": 359220
    },
    {
      "epoch": 0.26423058606993705,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001281542,
      "loss": 3.1798,
      "step": 359230
    },
    {
      "epoch": 0.26424922930609523,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00012815220000000002,
      "loss": 3.2038,
      "step": 359240
    },
    {
      "epoch": 0.2642678725422534,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012815019999999998,
      "loss": 3.0649,
      "step": 359250
    },
    {
      "epoch": 0.2642865157784116,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001281482,
      "loss": 3.1387,
      "step": 359260
    },
    {
      "epoch": 0.2643051590145698,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001281462,
      "loss": 3.0887,
      "step": 359270
    },
    {
      "epoch": 0.264323802250728,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001281442,
      "loss": 3.1509,
      "step": 359280
    },
    {
      "epoch": 0.2643424454868862,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012814220000000002,
      "loss": 3.069,
      "step": 359290
    },
    {
      "epoch": 0.26436108872304437,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001281402,
      "loss": 3.2928,
      "step": 359300
    },
    {
      "epoch": 0.26437973195920256,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001281382,
      "loss": 3.1541,
      "step": 359310
    },
    {
      "epoch": 0.26439837519536075,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001281362,
      "loss": 3.1286,
      "step": 359320
    },
    {
      "epoch": 0.2644170184315189,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001281342,
      "loss": 3.0898,
      "step": 359330
    },
    {
      "epoch": 0.26443566166767707,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012813220000000003,
      "loss": 3.0347,
      "step": 359340
    },
    {
      "epoch": 0.26445430490383526,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001281302,
      "loss": 3.0453,
      "step": 359350
    },
    {
      "epoch": 0.26447294813999345,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012812820000000002,
      "loss": 3.0927,
      "step": 359360
    },
    {
      "epoch": 0.26449159137615164,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012812619999999999,
      "loss": 3.1239,
      "step": 359370
    },
    {
      "epoch": 0.2645102346123098,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001281242,
      "loss": 3.0172,
      "step": 359380
    },
    {
      "epoch": 0.264528877848468,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001281222,
      "loss": 3.1829,
      "step": 359390
    },
    {
      "epoch": 0.2645475210846262,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001281202,
      "loss": 3.1762,
      "step": 359400
    },
    {
      "epoch": 0.2645661643207844,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012811820000000002,
      "loss": 3.1426,
      "step": 359410
    },
    {
      "epoch": 0.2645848075569426,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001281162,
      "loss": 3.1129,
      "step": 359420
    },
    {
      "epoch": 0.26460345079310077,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012811420000000001,
      "loss": 3.1677,
      "step": 359430
    },
    {
      "epoch": 0.2646220940292589,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001281122,
      "loss": 3.1265,
      "step": 359440
    },
    {
      "epoch": 0.2646407372654171,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001281102,
      "loss": 3.0779,
      "step": 359450
    },
    {
      "epoch": 0.2646593805015753,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001281082,
      "loss": 3.0612,
      "step": 359460
    },
    {
      "epoch": 0.26467802373773347,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001281062,
      "loss": 3.213,
      "step": 359470
    },
    {
      "epoch": 0.26469666697389166,
      "grad_norm": 1.203125,
      "learning_rate": 0.00012810420000000002,
      "loss": 3.0778,
      "step": 359480
    },
    {
      "epoch": 0.26471531021004985,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012810220000000002,
      "loss": 3.1164,
      "step": 359490
    },
    {
      "epoch": 0.26473395344620804,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001281002,
      "loss": 3.1799,
      "step": 359500
    },
    {
      "epoch": 0.2647525966823662,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001280982,
      "loss": 3.1608,
      "step": 359510
    },
    {
      "epoch": 0.2647712399185244,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001280962,
      "loss": 3.0604,
      "step": 359520
    },
    {
      "epoch": 0.2647898831546826,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001280942,
      "loss": 3.0031,
      "step": 359530
    },
    {
      "epoch": 0.26480852639084074,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001280922,
      "loss": 3.296,
      "step": 359540
    },
    {
      "epoch": 0.2648271696269989,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012809020000000002,
      "loss": 3.1628,
      "step": 359550
    },
    {
      "epoch": 0.2648458128631571,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001280882,
      "loss": 3.201,
      "step": 359560
    },
    {
      "epoch": 0.2648644560993153,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001280862,
      "loss": 3.2044,
      "step": 359570
    },
    {
      "epoch": 0.2648830993354735,
      "grad_norm": 0.75,
      "learning_rate": 0.0001280842,
      "loss": 3.0671,
      "step": 359580
    },
    {
      "epoch": 0.2649017425716317,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001280822,
      "loss": 2.9956,
      "step": 359590
    },
    {
      "epoch": 0.26492038580778987,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001280802,
      "loss": 3.1322,
      "step": 359600
    },
    {
      "epoch": 0.26493902904394806,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012807820000000002,
      "loss": 3.2329,
      "step": 359610
    },
    {
      "epoch": 0.26495767228010625,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001280762,
      "loss": 3.0918,
      "step": 359620
    },
    {
      "epoch": 0.26497631551626444,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001280742,
      "loss": 3.171,
      "step": 359630
    },
    {
      "epoch": 0.2649949587524226,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001280722,
      "loss": 3.274,
      "step": 359640
    },
    {
      "epoch": 0.26501360198858076,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001280702,
      "loss": 3.1368,
      "step": 359650
    },
    {
      "epoch": 0.26503224522473895,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012806820000000002,
      "loss": 3.2834,
      "step": 359660
    },
    {
      "epoch": 0.26505088846089714,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001280662,
      "loss": 3.0536,
      "step": 359670
    },
    {
      "epoch": 0.2650695316970553,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001280642,
      "loss": 3.1808,
      "step": 359680
    },
    {
      "epoch": 0.2650881749332135,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001280622,
      "loss": 3.1863,
      "step": 359690
    },
    {
      "epoch": 0.2651068181693717,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001280602,
      "loss": 3.2115,
      "step": 359700
    },
    {
      "epoch": 0.2651254614055299,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012805820000000003,
      "loss": 3.2505,
      "step": 359710
    },
    {
      "epoch": 0.2651441046416881,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001280562,
      "loss": 3.1478,
      "step": 359720
    },
    {
      "epoch": 0.26516274787784627,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012805420000000002,
      "loss": 3.0806,
      "step": 359730
    },
    {
      "epoch": 0.26518139111400446,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001280522,
      "loss": 3.1003,
      "step": 359740
    },
    {
      "epoch": 0.2652000343501626,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001280502,
      "loss": 3.0517,
      "step": 359750
    },
    {
      "epoch": 0.2652186775863208,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001280482,
      "loss": 3.3019,
      "step": 359760
    },
    {
      "epoch": 0.265237320822479,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001280462,
      "loss": 3.1941,
      "step": 359770
    },
    {
      "epoch": 0.26525596405863716,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012804420000000002,
      "loss": 3.0159,
      "step": 359780
    },
    {
      "epoch": 0.26527460729479535,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001280422,
      "loss": 3.1228,
      "step": 359790
    },
    {
      "epoch": 0.26529325053095354,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012804020000000001,
      "loss": 3.0,
      "step": 359800
    },
    {
      "epoch": 0.26531189376711173,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001280382,
      "loss": 3.1377,
      "step": 359810
    },
    {
      "epoch": 0.2653305370032699,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001280362,
      "loss": 3.151,
      "step": 359820
    },
    {
      "epoch": 0.2653491802394281,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012803420000000003,
      "loss": 2.9659,
      "step": 359830
    },
    {
      "epoch": 0.2653678234755863,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001280322,
      "loss": 3.1311,
      "step": 359840
    },
    {
      "epoch": 0.2653864667117445,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012803020000000002,
      "loss": 3.0363,
      "step": 359850
    },
    {
      "epoch": 0.2654051099479026,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001280282,
      "loss": 3.0871,
      "step": 359860
    },
    {
      "epoch": 0.2654237531840608,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001280262,
      "loss": 3.1193,
      "step": 359870
    },
    {
      "epoch": 0.265442396420219,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001280242,
      "loss": 3.185,
      "step": 359880
    },
    {
      "epoch": 0.2654610396563772,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001280222,
      "loss": 3.1219,
      "step": 359890
    },
    {
      "epoch": 0.2654796828925354,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012802020000000002,
      "loss": 3.1922,
      "step": 359900
    },
    {
      "epoch": 0.26549832612869356,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001280182,
      "loss": 3.086,
      "step": 359910
    },
    {
      "epoch": 0.26551696936485175,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012801620000000002,
      "loss": 2.9944,
      "step": 359920
    },
    {
      "epoch": 0.26553561260100994,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012801419999999998,
      "loss": 3.1577,
      "step": 359930
    },
    {
      "epoch": 0.26555425583716813,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001280122,
      "loss": 3.3685,
      "step": 359940
    },
    {
      "epoch": 0.2655728990733263,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001280102,
      "loss": 3.0703,
      "step": 359950
    },
    {
      "epoch": 0.26559154230948445,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001280082,
      "loss": 3.1709,
      "step": 359960
    },
    {
      "epoch": 0.26561018554564264,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012800620000000002,
      "loss": 3.1936,
      "step": 359970
    },
    {
      "epoch": 0.26562882878180083,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001280042,
      "loss": 3.0558,
      "step": 359980
    },
    {
      "epoch": 0.265647472017959,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001280022,
      "loss": 3.099,
      "step": 359990
    },
    {
      "epoch": 0.2656661152541172,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001280002,
      "loss": 3.2481,
      "step": 360000
    },
    {
      "epoch": 0.2656847584902754,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001279982,
      "loss": 3.1874,
      "step": 360010
    },
    {
      "epoch": 0.2657034017264336,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001279962,
      "loss": 3.1864,
      "step": 360020
    },
    {
      "epoch": 0.2657220449625918,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001279942,
      "loss": 3.2092,
      "step": 360030
    },
    {
      "epoch": 0.26574068819874996,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012799220000000002,
      "loss": 3.0595,
      "step": 360040
    },
    {
      "epoch": 0.26575933143490815,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001279902,
      "loss": 3.1771,
      "step": 360050
    },
    {
      "epoch": 0.26577797467106634,
      "grad_norm": 0.75,
      "learning_rate": 0.0001279882,
      "loss": 3.0989,
      "step": 360060
    },
    {
      "epoch": 0.2657966179072245,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001279862,
      "loss": 3.1137,
      "step": 360070
    },
    {
      "epoch": 0.26581526114338266,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001279842,
      "loss": 3.1684,
      "step": 360080
    },
    {
      "epoch": 0.26583390437954085,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001279822,
      "loss": 3.2019,
      "step": 360090
    },
    {
      "epoch": 0.26585254761569904,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012798020000000002,
      "loss": 3.0573,
      "step": 360100
    },
    {
      "epoch": 0.26587119085185723,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012797820000000001,
      "loss": 3.1934,
      "step": 360110
    },
    {
      "epoch": 0.2658898340880154,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001279762,
      "loss": 3.2586,
      "step": 360120
    },
    {
      "epoch": 0.2659084773241736,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001279742,
      "loss": 3.0721,
      "step": 360130
    },
    {
      "epoch": 0.2659271205603318,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001279722,
      "loss": 3.1454,
      "step": 360140
    },
    {
      "epoch": 0.26594576379649,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001279702,
      "loss": 3.1264,
      "step": 360150
    },
    {
      "epoch": 0.2659644070326482,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001279682,
      "loss": 3.0431,
      "step": 360160
    },
    {
      "epoch": 0.2659830502688063,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012796620000000001,
      "loss": 3.1604,
      "step": 360170
    },
    {
      "epoch": 0.2660016935049645,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001279642,
      "loss": 3.0256,
      "step": 360180
    },
    {
      "epoch": 0.2660203367411227,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001279622,
      "loss": 3.1472,
      "step": 360190
    },
    {
      "epoch": 0.2660389799772809,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001279602,
      "loss": 3.0624,
      "step": 360200
    },
    {
      "epoch": 0.26605762321343907,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001279582,
      "loss": 3.1752,
      "step": 360210
    },
    {
      "epoch": 0.26607626644959725,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012795620000000002,
      "loss": 3.1273,
      "step": 360220
    },
    {
      "epoch": 0.26609490968575544,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001279542,
      "loss": 3.1198,
      "step": 360230
    },
    {
      "epoch": 0.26611355292191363,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001279522,
      "loss": 3.0189,
      "step": 360240
    },
    {
      "epoch": 0.2661321961580718,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001279502,
      "loss": 3.1893,
      "step": 360250
    },
    {
      "epoch": 0.26615083939423,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001279482,
      "loss": 3.1076,
      "step": 360260
    },
    {
      "epoch": 0.2661694826303882,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012794620000000003,
      "loss": 3.0765,
      "step": 360270
    },
    {
      "epoch": 0.26618812586654633,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001279442,
      "loss": 3.0949,
      "step": 360280
    },
    {
      "epoch": 0.2662067691027045,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012794220000000002,
      "loss": 3.1374,
      "step": 360290
    },
    {
      "epoch": 0.2662254123388627,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001279402,
      "loss": 3.0539,
      "step": 360300
    },
    {
      "epoch": 0.2662440555750209,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001279382,
      "loss": 3.103,
      "step": 360310
    },
    {
      "epoch": 0.2662626988111791,
      "grad_norm": 0.75,
      "learning_rate": 0.00012793620000000003,
      "loss": 3.1184,
      "step": 360320
    },
    {
      "epoch": 0.2662813420473373,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001279342,
      "loss": 3.0922,
      "step": 360330
    },
    {
      "epoch": 0.26629998528349547,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012793220000000002,
      "loss": 3.1713,
      "step": 360340
    },
    {
      "epoch": 0.26631862851965366,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001279302,
      "loss": 3.2349,
      "step": 360350
    },
    {
      "epoch": 0.26633727175581184,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001279282,
      "loss": 3.1017,
      "step": 360360
    },
    {
      "epoch": 0.26635591499197003,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001279262,
      "loss": 3.1374,
      "step": 360370
    },
    {
      "epoch": 0.26637455822812817,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001279242,
      "loss": 3.0832,
      "step": 360380
    },
    {
      "epoch": 0.26639320146428636,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012792220000000003,
      "loss": 3.163,
      "step": 360390
    },
    {
      "epoch": 0.26641184470044454,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001279202,
      "loss": 3.0437,
      "step": 360400
    },
    {
      "epoch": 0.26643048793660273,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012791820000000002,
      "loss": 3.029,
      "step": 360410
    },
    {
      "epoch": 0.2664491311727609,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001279162,
      "loss": 3.0693,
      "step": 360420
    },
    {
      "epoch": 0.2664677744089191,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001279142,
      "loss": 3.2243,
      "step": 360430
    },
    {
      "epoch": 0.2664864176450773,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001279122,
      "loss": 3.1063,
      "step": 360440
    },
    {
      "epoch": 0.2665050608812355,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001279102,
      "loss": 3.0039,
      "step": 360450
    },
    {
      "epoch": 0.2665237041173937,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012790820000000002,
      "loss": 3.0552,
      "step": 360460
    },
    {
      "epoch": 0.26654234735355187,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001279062,
      "loss": 3.064,
      "step": 360470
    },
    {
      "epoch": 0.26656099058971006,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012790420000000001,
      "loss": 3.0436,
      "step": 360480
    },
    {
      "epoch": 0.2665796338258682,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001279022,
      "loss": 3.0984,
      "step": 360490
    },
    {
      "epoch": 0.2665982770620264,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001279002,
      "loss": 3.0981,
      "step": 360500
    },
    {
      "epoch": 0.26661692029818457,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001278982,
      "loss": 3.2005,
      "step": 360510
    },
    {
      "epoch": 0.26663556353434276,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001278962,
      "loss": 3.0816,
      "step": 360520
    },
    {
      "epoch": 0.26665420677050095,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012789420000000002,
      "loss": 3.2813,
      "step": 360530
    },
    {
      "epoch": 0.26667285000665913,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001278922,
      "loss": 2.9909,
      "step": 360540
    },
    {
      "epoch": 0.2666914932428173,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001278902,
      "loss": 3.0849,
      "step": 360550
    },
    {
      "epoch": 0.2667101364789755,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001278882,
      "loss": 3.2062,
      "step": 360560
    },
    {
      "epoch": 0.2667287797151337,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001278862,
      "loss": 3.0602,
      "step": 360570
    },
    {
      "epoch": 0.2667474229512919,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001278842,
      "loss": 2.9849,
      "step": 360580
    },
    {
      "epoch": 0.26676606618745,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001278822,
      "loss": 2.9819,
      "step": 360590
    },
    {
      "epoch": 0.2667847094236082,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012788020000000002,
      "loss": 3.1488,
      "step": 360600
    },
    {
      "epoch": 0.2668033526597664,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001278782,
      "loss": 3.0958,
      "step": 360610
    },
    {
      "epoch": 0.2668219958959246,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001278762,
      "loss": 3.054,
      "step": 360620
    },
    {
      "epoch": 0.2668406391320828,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001278742,
      "loss": 3.1848,
      "step": 360630
    },
    {
      "epoch": 0.26685928236824097,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001278722,
      "loss": 2.9607,
      "step": 360640
    },
    {
      "epoch": 0.26687792560439916,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001278702,
      "loss": 3.0778,
      "step": 360650
    },
    {
      "epoch": 0.26689656884055735,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012786820000000002,
      "loss": 3.1483,
      "step": 360660
    },
    {
      "epoch": 0.26691521207671554,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001278662,
      "loss": 3.0829,
      "step": 360670
    },
    {
      "epoch": 0.2669338553128737,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001278642,
      "loss": 3.0989,
      "step": 360680
    },
    {
      "epoch": 0.2669524985490319,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001278622,
      "loss": 3.1552,
      "step": 360690
    },
    {
      "epoch": 0.26697114178519005,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001278602,
      "loss": 3.1368,
      "step": 360700
    },
    {
      "epoch": 0.26698978502134824,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012785820000000002,
      "loss": 3.2579,
      "step": 360710
    },
    {
      "epoch": 0.2670084282575064,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001278562,
      "loss": 3.294,
      "step": 360720
    },
    {
      "epoch": 0.2670270714936646,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001278542,
      "loss": 3.1857,
      "step": 360730
    },
    {
      "epoch": 0.2670457147298228,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001278522,
      "loss": 3.0339,
      "step": 360740
    },
    {
      "epoch": 0.267064357965981,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001278502,
      "loss": 3.1553,
      "step": 360750
    },
    {
      "epoch": 0.2670830012021392,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001278482,
      "loss": 3.1117,
      "step": 360760
    },
    {
      "epoch": 0.26710164443829737,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001278462,
      "loss": 3.0898,
      "step": 360770
    },
    {
      "epoch": 0.26712028767445556,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012784420000000002,
      "loss": 3.1171,
      "step": 360780
    },
    {
      "epoch": 0.26713893091061375,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012784220000000001,
      "loss": 3.077,
      "step": 360790
    },
    {
      "epoch": 0.2671575741467719,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001278402,
      "loss": 3.1427,
      "step": 360800
    },
    {
      "epoch": 0.26717621738293007,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001278382,
      "loss": 3.1406,
      "step": 360810
    },
    {
      "epoch": 0.26719486061908826,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001278362,
      "loss": 3.2356,
      "step": 360820
    },
    {
      "epoch": 0.26721350385524645,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012783420000000002,
      "loss": 3.1059,
      "step": 360830
    },
    {
      "epoch": 0.26723214709140464,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001278322,
      "loss": 3.1262,
      "step": 360840
    },
    {
      "epoch": 0.2672507903275628,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012783020000000001,
      "loss": 3.0567,
      "step": 360850
    },
    {
      "epoch": 0.267269433563721,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001278282,
      "loss": 3.1509,
      "step": 360860
    },
    {
      "epoch": 0.2672880767998792,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001278262,
      "loss": 3.0367,
      "step": 360870
    },
    {
      "epoch": 0.2673067200360374,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012782420000000003,
      "loss": 3.2069,
      "step": 360880
    },
    {
      "epoch": 0.2673253632721956,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001278222,
      "loss": 3.1443,
      "step": 360890
    },
    {
      "epoch": 0.2673440065083537,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012782020000000002,
      "loss": 3.1671,
      "step": 360900
    },
    {
      "epoch": 0.2673626497445119,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001278182,
      "loss": 3.1782,
      "step": 360910
    },
    {
      "epoch": 0.2673812929806701,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001278162,
      "loss": 3.2238,
      "step": 360920
    },
    {
      "epoch": 0.2673999362168283,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001278142,
      "loss": 3.0673,
      "step": 360930
    },
    {
      "epoch": 0.26741857945298647,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001278122,
      "loss": 3.0063,
      "step": 360940
    },
    {
      "epoch": 0.26743722268914466,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012781020000000003,
      "loss": 3.1785,
      "step": 360950
    },
    {
      "epoch": 0.26745586592530285,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001278082,
      "loss": 3.1345,
      "step": 360960
    },
    {
      "epoch": 0.26747450916146104,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012780620000000002,
      "loss": 3.1496,
      "step": 360970
    },
    {
      "epoch": 0.2674931523976192,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012780419999999999,
      "loss": 3.0199,
      "step": 360980
    },
    {
      "epoch": 0.2675117956337774,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001278022,
      "loss": 3.2421,
      "step": 360990
    },
    {
      "epoch": 0.2675304388699356,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001278002,
      "loss": 3.1997,
      "step": 361000
    },
    {
      "epoch": 0.26754908210609374,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001277982,
      "loss": 3.0613,
      "step": 361010
    },
    {
      "epoch": 0.2675677253422519,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012779620000000002,
      "loss": 3.0855,
      "step": 361020
    },
    {
      "epoch": 0.2675863685784101,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001277942,
      "loss": 3.1453,
      "step": 361030
    },
    {
      "epoch": 0.2676050118145683,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001277922,
      "loss": 3.065,
      "step": 361040
    },
    {
      "epoch": 0.2676236550507265,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001277902,
      "loss": 3.1281,
      "step": 361050
    },
    {
      "epoch": 0.2676422982868847,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001277882,
      "loss": 2.9898,
      "step": 361060
    },
    {
      "epoch": 0.2676609415230429,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001277862,
      "loss": 3.1982,
      "step": 361070
    },
    {
      "epoch": 0.26767958475920106,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001277842,
      "loss": 3.1245,
      "step": 361080
    },
    {
      "epoch": 0.26769822799535925,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012778220000000002,
      "loss": 3.1295,
      "step": 361090
    },
    {
      "epoch": 0.26771687123151744,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012778020000000001,
      "loss": 3.1547,
      "step": 361100
    },
    {
      "epoch": 0.2677355144676756,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001277782,
      "loss": 3.1398,
      "step": 361110
    },
    {
      "epoch": 0.26775415770383376,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001277762,
      "loss": 3.1616,
      "step": 361120
    },
    {
      "epoch": 0.26777280093999195,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001277742,
      "loss": 3.2089,
      "step": 361130
    },
    {
      "epoch": 0.26779144417615014,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001277722,
      "loss": 3.1247,
      "step": 361140
    },
    {
      "epoch": 0.26781008741230833,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001277702,
      "loss": 3.0442,
      "step": 361150
    },
    {
      "epoch": 0.2678287306484665,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012776820000000001,
      "loss": 2.9859,
      "step": 361160
    },
    {
      "epoch": 0.2678473738846247,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001277662,
      "loss": 3.2162,
      "step": 361170
    },
    {
      "epoch": 0.2678660171207829,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001277642,
      "loss": 3.2294,
      "step": 361180
    },
    {
      "epoch": 0.2678846603569411,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001277622,
      "loss": 3.2115,
      "step": 361190
    },
    {
      "epoch": 0.2679033035930993,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001277602,
      "loss": 3.1142,
      "step": 361200
    },
    {
      "epoch": 0.26792194682925746,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001277582,
      "loss": 3.2376,
      "step": 361210
    },
    {
      "epoch": 0.2679405900654156,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012775620000000002,
      "loss": 3.1732,
      "step": 361220
    },
    {
      "epoch": 0.2679592333015738,
      "grad_norm": 0.625,
      "learning_rate": 0.0001277542,
      "loss": 3.2247,
      "step": 361230
    },
    {
      "epoch": 0.267977876537732,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001277522,
      "loss": 3.0856,
      "step": 361240
    },
    {
      "epoch": 0.26799651977389016,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001277502,
      "loss": 3.0608,
      "step": 361250
    },
    {
      "epoch": 0.26801516301004835,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001277482,
      "loss": 3.1556,
      "step": 361260
    },
    {
      "epoch": 0.26803380624620654,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012774620000000002,
      "loss": 3.1242,
      "step": 361270
    },
    {
      "epoch": 0.26805244948236473,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012774420000000002,
      "loss": 2.9956,
      "step": 361280
    },
    {
      "epoch": 0.2680710927185229,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001277422,
      "loss": 3.3018,
      "step": 361290
    },
    {
      "epoch": 0.2680897359546811,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001277402,
      "loss": 3.1558,
      "step": 361300
    },
    {
      "epoch": 0.2681083791908393,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001277382,
      "loss": 3.2763,
      "step": 361310
    },
    {
      "epoch": 0.26812702242699743,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012773620000000003,
      "loss": 2.9707,
      "step": 361320
    },
    {
      "epoch": 0.2681456656631556,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001277342,
      "loss": 3.2445,
      "step": 361330
    },
    {
      "epoch": 0.2681643088993138,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012773220000000002,
      "loss": 3.0397,
      "step": 361340
    },
    {
      "epoch": 0.268182952135472,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001277302,
      "loss": 3.237,
      "step": 361350
    },
    {
      "epoch": 0.2682015953716302,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001277282,
      "loss": 3.1131,
      "step": 361360
    },
    {
      "epoch": 0.2682202386077884,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001277262,
      "loss": 3.0942,
      "step": 361370
    },
    {
      "epoch": 0.26823888184394656,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001277242,
      "loss": 3.1344,
      "step": 361380
    },
    {
      "epoch": 0.26825752508010475,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012772220000000002,
      "loss": 2.9893,
      "step": 361390
    },
    {
      "epoch": 0.26827616831626294,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001277202,
      "loss": 3.1769,
      "step": 361400
    },
    {
      "epoch": 0.26829481155242113,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001277182,
      "loss": 3.2973,
      "step": 361410
    },
    {
      "epoch": 0.2683134547885793,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001277162,
      "loss": 3.1207,
      "step": 361420
    },
    {
      "epoch": 0.26833209802473745,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001277142,
      "loss": 3.2673,
      "step": 361430
    },
    {
      "epoch": 0.26835074126089564,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012771220000000003,
      "loss": 3.0226,
      "step": 361440
    },
    {
      "epoch": 0.26836938449705383,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001277102,
      "loss": 3.1361,
      "step": 361450
    },
    {
      "epoch": 0.268388027733212,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012770820000000002,
      "loss": 3.0382,
      "step": 361460
    },
    {
      "epoch": 0.2684066709693702,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001277062,
      "loss": 3.168,
      "step": 361470
    },
    {
      "epoch": 0.2684253142055284,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001277042,
      "loss": 3.0579,
      "step": 361480
    },
    {
      "epoch": 0.2684439574416866,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001277022,
      "loss": 3.1465,
      "step": 361490
    },
    {
      "epoch": 0.2684626006778448,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001277002,
      "loss": 3.0542,
      "step": 361500
    },
    {
      "epoch": 0.26848124391400296,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012769820000000002,
      "loss": 3.0192,
      "step": 361510
    },
    {
      "epoch": 0.26849988715016115,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001276962,
      "loss": 3.0933,
      "step": 361520
    },
    {
      "epoch": 0.2685185303863193,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012769420000000001,
      "loss": 3.1962,
      "step": 361530
    },
    {
      "epoch": 0.2685371736224775,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012769219999999998,
      "loss": 3.0701,
      "step": 361540
    },
    {
      "epoch": 0.26855581685863567,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001276902,
      "loss": 3.1858,
      "step": 361550
    },
    {
      "epoch": 0.26857446009479385,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001276882,
      "loss": 3.1463,
      "step": 361560
    },
    {
      "epoch": 0.26859310333095204,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001276862,
      "loss": 3.1227,
      "step": 361570
    },
    {
      "epoch": 0.26861174656711023,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012768420000000002,
      "loss": 3.2163,
      "step": 361580
    },
    {
      "epoch": 0.2686303898032684,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001276822,
      "loss": 3.2658,
      "step": 361590
    },
    {
      "epoch": 0.2686490330394266,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001276802,
      "loss": 3.1443,
      "step": 361600
    },
    {
      "epoch": 0.2686676762755848,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001276782,
      "loss": 2.9379,
      "step": 361610
    },
    {
      "epoch": 0.268686319511743,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001276762,
      "loss": 3.1301,
      "step": 361620
    },
    {
      "epoch": 0.2687049627479012,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001276742,
      "loss": 3.1527,
      "step": 361630
    },
    {
      "epoch": 0.2687236059840593,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001276722,
      "loss": 3.0808,
      "step": 361640
    },
    {
      "epoch": 0.2687422492202175,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012767020000000002,
      "loss": 3.2269,
      "step": 361650
    },
    {
      "epoch": 0.2687608924563757,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001276682,
      "loss": 3.0615,
      "step": 361660
    },
    {
      "epoch": 0.2687795356925339,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001276662,
      "loss": 3.0476,
      "step": 361670
    },
    {
      "epoch": 0.26879817892869207,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001276642,
      "loss": 3.0467,
      "step": 361680
    },
    {
      "epoch": 0.26881682216485026,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001276622,
      "loss": 3.1226,
      "step": 361690
    },
    {
      "epoch": 0.26883546540100844,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001276602,
      "loss": 3.2098,
      "step": 361700
    },
    {
      "epoch": 0.26885410863716663,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012765820000000002,
      "loss": 3.0769,
      "step": 361710
    },
    {
      "epoch": 0.2688727518733248,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001276562,
      "loss": 3.1985,
      "step": 361720
    },
    {
      "epoch": 0.268891395109483,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001276542,
      "loss": 3.0974,
      "step": 361730
    },
    {
      "epoch": 0.26891003834564114,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001276522,
      "loss": 3.0495,
      "step": 361740
    },
    {
      "epoch": 0.26892868158179933,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001276502,
      "loss": 3.118,
      "step": 361750
    },
    {
      "epoch": 0.2689473248179575,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001276482,
      "loss": 3.0911,
      "step": 361760
    },
    {
      "epoch": 0.2689659680541157,
      "grad_norm": 0.875,
      "learning_rate": 0.00012764620000000002,
      "loss": 3.1364,
      "step": 361770
    },
    {
      "epoch": 0.2689846112902739,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012764420000000001,
      "loss": 3.1176,
      "step": 361780
    },
    {
      "epoch": 0.2690032545264321,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001276422,
      "loss": 3.2272,
      "step": 361790
    },
    {
      "epoch": 0.2690218977625903,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001276402,
      "loss": 3.0545,
      "step": 361800
    },
    {
      "epoch": 0.26904054099874847,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001276382,
      "loss": 3.1522,
      "step": 361810
    },
    {
      "epoch": 0.26905918423490666,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001276362,
      "loss": 2.9827,
      "step": 361820
    },
    {
      "epoch": 0.26907782747106485,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012763420000000002,
      "loss": 3.2034,
      "step": 361830
    },
    {
      "epoch": 0.26909647070722303,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012763220000000001,
      "loss": 3.1761,
      "step": 361840
    },
    {
      "epoch": 0.26911511394338117,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001276302,
      "loss": 3.2379,
      "step": 361850
    },
    {
      "epoch": 0.26913375717953936,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001276282,
      "loss": 3.1453,
      "step": 361860
    },
    {
      "epoch": 0.26915240041569755,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001276262,
      "loss": 3.0237,
      "step": 361870
    },
    {
      "epoch": 0.26917104365185573,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012762420000000002,
      "loss": 3.2132,
      "step": 361880
    },
    {
      "epoch": 0.2691896868880139,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001276222,
      "loss": 3.1923,
      "step": 361890
    },
    {
      "epoch": 0.2692083301241721,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012762020000000002,
      "loss": 3.1558,
      "step": 361900
    },
    {
      "epoch": 0.2692269733603303,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001276182,
      "loss": 3.1843,
      "step": 361910
    },
    {
      "epoch": 0.2692456165964885,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001276162,
      "loss": 3.0904,
      "step": 361920
    },
    {
      "epoch": 0.2692642598326467,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012761420000000003,
      "loss": 3.1892,
      "step": 361930
    },
    {
      "epoch": 0.26928290306880487,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001276122,
      "loss": 3.1978,
      "step": 361940
    },
    {
      "epoch": 0.269301546304963,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012761020000000002,
      "loss": 3.048,
      "step": 361950
    },
    {
      "epoch": 0.2693201895411212,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001276082,
      "loss": 3.1566,
      "step": 361960
    },
    {
      "epoch": 0.2693388327772794,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001276062,
      "loss": 3.1361,
      "step": 361970
    },
    {
      "epoch": 0.26935747601343757,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001276042,
      "loss": 3.2198,
      "step": 361980
    },
    {
      "epoch": 0.26937611924959576,
      "grad_norm": 0.75,
      "learning_rate": 0.0001276022,
      "loss": 3.0943,
      "step": 361990
    },
    {
      "epoch": 0.26939476248575395,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012760020000000003,
      "loss": 3.0783,
      "step": 362000
    },
    {
      "epoch": 0.26941340572191214,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001275982,
      "loss": 3.0813,
      "step": 362010
    },
    {
      "epoch": 0.2694320489580703,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012759620000000002,
      "loss": 3.0632,
      "step": 362020
    },
    {
      "epoch": 0.2694506921942285,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012759419999999999,
      "loss": 3.0714,
      "step": 362030
    },
    {
      "epoch": 0.2694693354303867,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001275922,
      "loss": 3.2458,
      "step": 362040
    },
    {
      "epoch": 0.2694879786665449,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001275902,
      "loss": 3.0845,
      "step": 362050
    },
    {
      "epoch": 0.269506621902703,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001275882,
      "loss": 3.2859,
      "step": 362060
    },
    {
      "epoch": 0.2695252651388612,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012758620000000002,
      "loss": 3.0723,
      "step": 362070
    },
    {
      "epoch": 0.2695439083750194,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001275842,
      "loss": 3.2264,
      "step": 362080
    },
    {
      "epoch": 0.2695625516111776,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001275822,
      "loss": 3.2272,
      "step": 362090
    },
    {
      "epoch": 0.2695811948473358,
      "grad_norm": 0.75,
      "learning_rate": 0.0001275802,
      "loss": 3.2563,
      "step": 362100
    },
    {
      "epoch": 0.26959983808349397,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001275782,
      "loss": 3.2927,
      "step": 362110
    },
    {
      "epoch": 0.26961848131965216,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001275762,
      "loss": 3.2157,
      "step": 362120
    },
    {
      "epoch": 0.26963712455581035,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001275742,
      "loss": 3.1104,
      "step": 362130
    },
    {
      "epoch": 0.26965576779196854,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012757220000000002,
      "loss": 3.0064,
      "step": 362140
    },
    {
      "epoch": 0.2696744110281267,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001275702,
      "loss": 3.0821,
      "step": 362150
    },
    {
      "epoch": 0.26969305426428486,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001275682,
      "loss": 3.1035,
      "step": 362160
    },
    {
      "epoch": 0.26971169750044305,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001275662,
      "loss": 3.1971,
      "step": 362170
    },
    {
      "epoch": 0.26973034073660124,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001275642,
      "loss": 3.1925,
      "step": 362180
    },
    {
      "epoch": 0.2697489839727594,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001275622,
      "loss": 3.0119,
      "step": 362190
    },
    {
      "epoch": 0.2697676272089176,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001275602,
      "loss": 3.0524,
      "step": 362200
    },
    {
      "epoch": 0.2697862704450758,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012755820000000001,
      "loss": 3.0564,
      "step": 362210
    },
    {
      "epoch": 0.269804913681234,
      "grad_norm": 0.75,
      "learning_rate": 0.0001275562,
      "loss": 3.0889,
      "step": 362220
    },
    {
      "epoch": 0.2698235569173922,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001275542,
      "loss": 3.0988,
      "step": 362230
    },
    {
      "epoch": 0.26984220015355037,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001275522,
      "loss": 3.038,
      "step": 362240
    },
    {
      "epoch": 0.26986084338970856,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001275502,
      "loss": 3.2521,
      "step": 362250
    },
    {
      "epoch": 0.2698794866258667,
      "grad_norm": 0.75,
      "learning_rate": 0.00012754820000000002,
      "loss": 3.1829,
      "step": 362260
    },
    {
      "epoch": 0.2698981298620249,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012754620000000002,
      "loss": 3.2876,
      "step": 362270
    },
    {
      "epoch": 0.26991677309818307,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001275442,
      "loss": 3.0515,
      "step": 362280
    },
    {
      "epoch": 0.26993541633434126,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001275422,
      "loss": 3.1014,
      "step": 362290
    },
    {
      "epoch": 0.26995405957049945,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001275402,
      "loss": 3.1824,
      "step": 362300
    },
    {
      "epoch": 0.26997270280665764,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001275382,
      "loss": 3.1373,
      "step": 362310
    },
    {
      "epoch": 0.2699913460428158,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012753620000000002,
      "loss": 3.0667,
      "step": 362320
    },
    {
      "epoch": 0.270009989278974,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012753420000000002,
      "loss": 3.0275,
      "step": 362330
    },
    {
      "epoch": 0.2700286325151322,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001275322,
      "loss": 3.3202,
      "step": 362340
    },
    {
      "epoch": 0.2700472757512904,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001275302,
      "loss": 3.0607,
      "step": 362350
    },
    {
      "epoch": 0.2700659189874486,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001275282,
      "loss": 3.0942,
      "step": 362360
    },
    {
      "epoch": 0.2700845622236067,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001275262,
      "loss": 3.2072,
      "step": 362370
    },
    {
      "epoch": 0.2701032054597649,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001275242,
      "loss": 3.2038,
      "step": 362380
    },
    {
      "epoch": 0.2701218486959231,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012752220000000002,
      "loss": 3.1118,
      "step": 362390
    },
    {
      "epoch": 0.2701404919320813,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001275202,
      "loss": 3.0516,
      "step": 362400
    },
    {
      "epoch": 0.2701591351682395,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001275182,
      "loss": 3.078,
      "step": 362410
    },
    {
      "epoch": 0.27017777840439766,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001275162,
      "loss": 3.0712,
      "step": 362420
    },
    {
      "epoch": 0.27019642164055585,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001275142,
      "loss": 3.1007,
      "step": 362430
    },
    {
      "epoch": 0.27021506487671404,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012751220000000002,
      "loss": 3.0965,
      "step": 362440
    },
    {
      "epoch": 0.27023370811287223,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001275102,
      "loss": 3.0148,
      "step": 362450
    },
    {
      "epoch": 0.2702523513490304,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012750820000000001,
      "loss": 3.1265,
      "step": 362460
    },
    {
      "epoch": 0.27027099458518855,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001275062,
      "loss": 3.1263,
      "step": 362470
    },
    {
      "epoch": 0.27028963782134674,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001275042,
      "loss": 3.129,
      "step": 362480
    },
    {
      "epoch": 0.27030828105750493,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012750220000000003,
      "loss": 3.1247,
      "step": 362490
    },
    {
      "epoch": 0.2703269242936631,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001275002,
      "loss": 3.1756,
      "step": 362500
    },
    {
      "epoch": 0.2703455675298213,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012749820000000002,
      "loss": 3.0566,
      "step": 362510
    },
    {
      "epoch": 0.2703642107659795,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001274962,
      "loss": 3.0142,
      "step": 362520
    },
    {
      "epoch": 0.2703828540021377,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001274942,
      "loss": 3.137,
      "step": 362530
    },
    {
      "epoch": 0.2704014972382959,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001274922,
      "loss": 3.271,
      "step": 362540
    },
    {
      "epoch": 0.27042014047445406,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001274902,
      "loss": 3.0643,
      "step": 362550
    },
    {
      "epoch": 0.27043878371061225,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012748820000000002,
      "loss": 3.0994,
      "step": 362560
    },
    {
      "epoch": 0.27045742694677044,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001274862,
      "loss": 3.1401,
      "step": 362570
    },
    {
      "epoch": 0.2704760701829286,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012748420000000002,
      "loss": 3.1064,
      "step": 362580
    },
    {
      "epoch": 0.27049471341908676,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012748219999999998,
      "loss": 3.1642,
      "step": 362590
    },
    {
      "epoch": 0.27051335665524495,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001274802,
      "loss": 3.197,
      "step": 362600
    },
    {
      "epoch": 0.27053199989140314,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001274782,
      "loss": 3.0817,
      "step": 362610
    },
    {
      "epoch": 0.27055064312756133,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001274762,
      "loss": 3.0837,
      "step": 362620
    },
    {
      "epoch": 0.2705692863637195,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012747420000000002,
      "loss": 3.208,
      "step": 362630
    },
    {
      "epoch": 0.2705879295998777,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001274722,
      "loss": 3.2742,
      "step": 362640
    },
    {
      "epoch": 0.2706065728360359,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001274702,
      "loss": 3.003,
      "step": 362650
    },
    {
      "epoch": 0.2706252160721941,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001274682,
      "loss": 3.0781,
      "step": 362660
    },
    {
      "epoch": 0.2706438593083523,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001274662,
      "loss": 3.1385,
      "step": 362670
    },
    {
      "epoch": 0.2706625025445104,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001274642,
      "loss": 3.0797,
      "step": 362680
    },
    {
      "epoch": 0.2706811457806686,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001274622,
      "loss": 3.0077,
      "step": 362690
    },
    {
      "epoch": 0.2706997890168268,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012746020000000002,
      "loss": 3.25,
      "step": 362700
    },
    {
      "epoch": 0.270718432252985,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001274582,
      "loss": 3.1692,
      "step": 362710
    },
    {
      "epoch": 0.27073707548914316,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001274562,
      "loss": 3.1376,
      "step": 362720
    },
    {
      "epoch": 0.27075571872530135,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001274542,
      "loss": 3.1496,
      "step": 362730
    },
    {
      "epoch": 0.27077436196145954,
      "grad_norm": 0.75,
      "learning_rate": 0.0001274522,
      "loss": 3.0718,
      "step": 362740
    },
    {
      "epoch": 0.27079300519761773,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012745020000000002,
      "loss": 3.0755,
      "step": 362750
    },
    {
      "epoch": 0.2708116484337759,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001274482,
      "loss": 3.0613,
      "step": 362760
    },
    {
      "epoch": 0.2708302916699341,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001274462,
      "loss": 3.1242,
      "step": 362770
    },
    {
      "epoch": 0.2708489349060923,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001274442,
      "loss": 3.1288,
      "step": 362780
    },
    {
      "epoch": 0.27086757814225043,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001274422,
      "loss": 3.1924,
      "step": 362790
    },
    {
      "epoch": 0.2708862213784086,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001274402,
      "loss": 3.1542,
      "step": 362800
    },
    {
      "epoch": 0.2709048646145668,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001274382,
      "loss": 3.2069,
      "step": 362810
    },
    {
      "epoch": 0.270923507850725,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012743620000000002,
      "loss": 3.1142,
      "step": 362820
    },
    {
      "epoch": 0.2709421510868832,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012743420000000001,
      "loss": 3.0765,
      "step": 362830
    },
    {
      "epoch": 0.2709607943230414,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001274322,
      "loss": 3.2073,
      "step": 362840
    },
    {
      "epoch": 0.27097943755919957,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001274302,
      "loss": 3.2951,
      "step": 362850
    },
    {
      "epoch": 0.27099808079535775,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001274282,
      "loss": 3.0908,
      "step": 362860
    },
    {
      "epoch": 0.27101672403151594,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001274262,
      "loss": 3.0578,
      "step": 362870
    },
    {
      "epoch": 0.27103536726767413,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012742420000000002,
      "loss": 3.2175,
      "step": 362880
    },
    {
      "epoch": 0.27105401050383227,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012742220000000001,
      "loss": 3.0607,
      "step": 362890
    },
    {
      "epoch": 0.27107265373999045,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001274202,
      "loss": 3.1029,
      "step": 362900
    },
    {
      "epoch": 0.27109129697614864,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001274182,
      "loss": 3.0459,
      "step": 362910
    },
    {
      "epoch": 0.27110994021230683,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001274162,
      "loss": 3.0134,
      "step": 362920
    },
    {
      "epoch": 0.271128583448465,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012741420000000002,
      "loss": 3.3086,
      "step": 362930
    },
    {
      "epoch": 0.2711472266846232,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001274122,
      "loss": 3.1692,
      "step": 362940
    },
    {
      "epoch": 0.2711658699207814,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012741020000000002,
      "loss": 2.9981,
      "step": 362950
    },
    {
      "epoch": 0.2711845131569396,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001274082,
      "loss": 3.1974,
      "step": 362960
    },
    {
      "epoch": 0.2712031563930978,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001274062,
      "loss": 3.2588,
      "step": 362970
    },
    {
      "epoch": 0.27122179962925597,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001274042,
      "loss": 2.9999,
      "step": 362980
    },
    {
      "epoch": 0.27124044286541416,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001274022,
      "loss": 3.1607,
      "step": 362990
    },
    {
      "epoch": 0.2712590861015723,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012740020000000002,
      "loss": 3.1344,
      "step": 363000
    },
    {
      "epoch": 0.2712777293377305,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001273982,
      "loss": 3.1765,
      "step": 363010
    },
    {
      "epoch": 0.27129637257388867,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001273962,
      "loss": 3.1513,
      "step": 363020
    },
    {
      "epoch": 0.27131501581004686,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001273942,
      "loss": 3.0733,
      "step": 363030
    },
    {
      "epoch": 0.27133365904620504,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001273922,
      "loss": 3.2148,
      "step": 363040
    },
    {
      "epoch": 0.27135230228236323,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012739020000000003,
      "loss": 3.0458,
      "step": 363050
    },
    {
      "epoch": 0.2713709455185214,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001273882,
      "loss": 3.076,
      "step": 363060
    },
    {
      "epoch": 0.2713895887546796,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012738620000000002,
      "loss": 3.2586,
      "step": 363070
    },
    {
      "epoch": 0.2714082319908378,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012738419999999999,
      "loss": 3.2263,
      "step": 363080
    },
    {
      "epoch": 0.271426875226996,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001273822,
      "loss": 3.0911,
      "step": 363090
    },
    {
      "epoch": 0.2714455184631541,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001273802,
      "loss": 3.0981,
      "step": 363100
    },
    {
      "epoch": 0.2714641616993123,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001273782,
      "loss": 3.1844,
      "step": 363110
    },
    {
      "epoch": 0.2714828049354705,
      "grad_norm": 0.75,
      "learning_rate": 0.00012737620000000002,
      "loss": 3.1882,
      "step": 363120
    },
    {
      "epoch": 0.2715014481716287,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001273742,
      "loss": 3.1232,
      "step": 363130
    },
    {
      "epoch": 0.2715200914077869,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001273722,
      "loss": 3.1156,
      "step": 363140
    },
    {
      "epoch": 0.27153873464394507,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012737019999999998,
      "loss": 3.0686,
      "step": 363150
    },
    {
      "epoch": 0.27155737788010326,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001273682,
      "loss": 3.0931,
      "step": 363160
    },
    {
      "epoch": 0.27157602111626145,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001273662,
      "loss": 3.1084,
      "step": 363170
    },
    {
      "epoch": 0.27159466435241963,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001273642,
      "loss": 3.1248,
      "step": 363180
    },
    {
      "epoch": 0.2716133075885778,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012736220000000002,
      "loss": 3.156,
      "step": 363190
    },
    {
      "epoch": 0.271631950824736,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001273602,
      "loss": 3.0838,
      "step": 363200
    },
    {
      "epoch": 0.27165059406089415,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001273582,
      "loss": 3.1247,
      "step": 363210
    },
    {
      "epoch": 0.27166923729705234,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001273562,
      "loss": 3.0527,
      "step": 363220
    },
    {
      "epoch": 0.2716878805332105,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001273542,
      "loss": 3.1711,
      "step": 363230
    },
    {
      "epoch": 0.2717065237693687,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012735220000000002,
      "loss": 3.0462,
      "step": 363240
    },
    {
      "epoch": 0.2717251670055269,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001273502,
      "loss": 3.2133,
      "step": 363250
    },
    {
      "epoch": 0.2717438102416851,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012734820000000002,
      "loss": 3.0851,
      "step": 363260
    },
    {
      "epoch": 0.2717624534778433,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001273462,
      "loss": 3.2237,
      "step": 363270
    },
    {
      "epoch": 0.27178109671400147,
      "grad_norm": 0.75,
      "learning_rate": 0.0001273442,
      "loss": 3.0859,
      "step": 363280
    },
    {
      "epoch": 0.27179973995015966,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001273422,
      "loss": 3.1771,
      "step": 363290
    },
    {
      "epoch": 0.27181838318631785,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001273402,
      "loss": 2.9895,
      "step": 363300
    },
    {
      "epoch": 0.271837026422476,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012733820000000002,
      "loss": 3.1026,
      "step": 363310
    },
    {
      "epoch": 0.27185566965863417,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012733620000000002,
      "loss": 3.1151,
      "step": 363320
    },
    {
      "epoch": 0.27187431289479236,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001273342,
      "loss": 3.1237,
      "step": 363330
    },
    {
      "epoch": 0.27189295613095055,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001273322,
      "loss": 3.0687,
      "step": 363340
    },
    {
      "epoch": 0.27191159936710874,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001273302,
      "loss": 3.1239,
      "step": 363350
    },
    {
      "epoch": 0.2719302426032669,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001273282,
      "loss": 3.2065,
      "step": 363360
    },
    {
      "epoch": 0.2719488858394251,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001273262,
      "loss": 3.1954,
      "step": 363370
    },
    {
      "epoch": 0.2719675290755833,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012732420000000002,
      "loss": 3.2394,
      "step": 363380
    },
    {
      "epoch": 0.2719861723117415,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001273222,
      "loss": 3.0989,
      "step": 363390
    },
    {
      "epoch": 0.2720048155478997,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001273202,
      "loss": 3.1773,
      "step": 363400
    },
    {
      "epoch": 0.27202345878405787,
      "grad_norm": 0.75,
      "learning_rate": 0.0001273182,
      "loss": 3.1483,
      "step": 363410
    },
    {
      "epoch": 0.272042102020216,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001273162,
      "loss": 3.1725,
      "step": 363420
    },
    {
      "epoch": 0.2720607452563742,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001273142,
      "loss": 3.1893,
      "step": 363430
    },
    {
      "epoch": 0.2720793884925324,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012731220000000002,
      "loss": 3.0997,
      "step": 363440
    },
    {
      "epoch": 0.27209803172869057,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001273102,
      "loss": 3.1528,
      "step": 363450
    },
    {
      "epoch": 0.27211667496484876,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001273082,
      "loss": 3.1131,
      "step": 363460
    },
    {
      "epoch": 0.27213531820100695,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001273062,
      "loss": 3.0474,
      "step": 363470
    },
    {
      "epoch": 0.27215396143716514,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001273042,
      "loss": 3.0872,
      "step": 363480
    },
    {
      "epoch": 0.2721726046733233,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012730220000000002,
      "loss": 3.1652,
      "step": 363490
    },
    {
      "epoch": 0.2721912479094815,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001273002,
      "loss": 3.1421,
      "step": 363500
    },
    {
      "epoch": 0.2722098911456397,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012729820000000001,
      "loss": 3.0766,
      "step": 363510
    },
    {
      "epoch": 0.27222853438179784,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001272962,
      "loss": 3.0809,
      "step": 363520
    },
    {
      "epoch": 0.272247177617956,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001272942,
      "loss": 3.2052,
      "step": 363530
    },
    {
      "epoch": 0.2722658208541142,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012729220000000003,
      "loss": 3.0885,
      "step": 363540
    },
    {
      "epoch": 0.2722844640902724,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001272902,
      "loss": 3.2919,
      "step": 363550
    },
    {
      "epoch": 0.2723031073264306,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012728820000000002,
      "loss": 3.1903,
      "step": 363560
    },
    {
      "epoch": 0.2723217505625888,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001272862,
      "loss": 3.1047,
      "step": 363570
    },
    {
      "epoch": 0.27234039379874697,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001272842,
      "loss": 3.1467,
      "step": 363580
    },
    {
      "epoch": 0.27235903703490516,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001272822,
      "loss": 3.1584,
      "step": 363590
    },
    {
      "epoch": 0.27237768027106335,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001272802,
      "loss": 3.1968,
      "step": 363600
    },
    {
      "epoch": 0.27239632350722154,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012727820000000002,
      "loss": 3.1947,
      "step": 363610
    },
    {
      "epoch": 0.27241496674337967,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001272762,
      "loss": 3.2337,
      "step": 363620
    },
    {
      "epoch": 0.27243360997953786,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00012727420000000002,
      "loss": 3.1692,
      "step": 363630
    },
    {
      "epoch": 0.27245225321569605,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012727219999999998,
      "loss": 3.1207,
      "step": 363640
    },
    {
      "epoch": 0.27247089645185424,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001272702,
      "loss": 3.2368,
      "step": 363650
    },
    {
      "epoch": 0.2724895396880124,
      "grad_norm": 0.75,
      "learning_rate": 0.0001272682,
      "loss": 3.2581,
      "step": 363660
    },
    {
      "epoch": 0.2725081829241706,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001272662,
      "loss": 3.1857,
      "step": 363670
    },
    {
      "epoch": 0.2725268261603288,
      "grad_norm": 0.75,
      "learning_rate": 0.00012726420000000002,
      "loss": 3.1608,
      "step": 363680
    },
    {
      "epoch": 0.272545469396487,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001272622,
      "loss": 3.1533,
      "step": 363690
    },
    {
      "epoch": 0.2725641126326452,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001272602,
      "loss": 3.121,
      "step": 363700
    },
    {
      "epoch": 0.2725827558688034,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001272582,
      "loss": 3.0943,
      "step": 363710
    },
    {
      "epoch": 0.27260139910496156,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001272562,
      "loss": 2.9959,
      "step": 363720
    },
    {
      "epoch": 0.2726200423411197,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012725420000000003,
      "loss": 3.0489,
      "step": 363730
    },
    {
      "epoch": 0.2726386855772779,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001272522,
      "loss": 2.9993,
      "step": 363740
    },
    {
      "epoch": 0.2726573288134361,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012725020000000002,
      "loss": 3.138,
      "step": 363750
    },
    {
      "epoch": 0.27267597204959426,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012724819999999999,
      "loss": 3.2035,
      "step": 363760
    },
    {
      "epoch": 0.27269461528575245,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001272462,
      "loss": 3.1359,
      "step": 363770
    },
    {
      "epoch": 0.27271325852191064,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001272442,
      "loss": 3.15,
      "step": 363780
    },
    {
      "epoch": 0.27273190175806883,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001272422,
      "loss": 2.9923,
      "step": 363790
    },
    {
      "epoch": 0.272750544994227,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012724020000000002,
      "loss": 3.1558,
      "step": 363800
    },
    {
      "epoch": 0.2727691882303852,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001272382,
      "loss": 3.1436,
      "step": 363810
    },
    {
      "epoch": 0.2727878314665434,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001272362,
      "loss": 3.1135,
      "step": 363820
    },
    {
      "epoch": 0.27280647470270153,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001272342,
      "loss": 3.0191,
      "step": 363830
    },
    {
      "epoch": 0.2728251179388597,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001272322,
      "loss": 3.298,
      "step": 363840
    },
    {
      "epoch": 0.2728437611750179,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001272302,
      "loss": 2.968,
      "step": 363850
    },
    {
      "epoch": 0.2728624044111761,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001272282,
      "loss": 3.117,
      "step": 363860
    },
    {
      "epoch": 0.2728810476473343,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012722620000000002,
      "loss": 3.2157,
      "step": 363870
    },
    {
      "epoch": 0.2728996908834925,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012722420000000001,
      "loss": 3.1231,
      "step": 363880
    },
    {
      "epoch": 0.27291833411965066,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001272222,
      "loss": 2.9836,
      "step": 363890
    },
    {
      "epoch": 0.27293697735580885,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001272202,
      "loss": 3.2278,
      "step": 363900
    },
    {
      "epoch": 0.27295562059196704,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001272182,
      "loss": 3.0442,
      "step": 363910
    },
    {
      "epoch": 0.27297426382812523,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001272162,
      "loss": 3.279,
      "step": 363920
    },
    {
      "epoch": 0.2729929070642834,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012721420000000002,
      "loss": 3.0685,
      "step": 363930
    },
    {
      "epoch": 0.27301155030044155,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012721220000000002,
      "loss": 3.079,
      "step": 363940
    },
    {
      "epoch": 0.27303019353659974,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001272102,
      "loss": 3.0556,
      "step": 363950
    },
    {
      "epoch": 0.27304883677275793,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001272082,
      "loss": 3.2385,
      "step": 363960
    },
    {
      "epoch": 0.2730674800089161,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001272062,
      "loss": 3.1691,
      "step": 363970
    },
    {
      "epoch": 0.2730861232450743,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001272042,
      "loss": 3.073,
      "step": 363980
    },
    {
      "epoch": 0.2731047664812325,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001272022,
      "loss": 3.1614,
      "step": 363990
    },
    {
      "epoch": 0.2731234097173907,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012720020000000002,
      "loss": 3.2916,
      "step": 364000
    },
    {
      "epoch": 0.2731420529535489,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001271982,
      "loss": 3.1618,
      "step": 364010
    },
    {
      "epoch": 0.27316069618970706,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001271962,
      "loss": 3.1381,
      "step": 364020
    },
    {
      "epoch": 0.27317933942586525,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001271942,
      "loss": 3.166,
      "step": 364030
    },
    {
      "epoch": 0.2731979826620234,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001271922,
      "loss": 3.158,
      "step": 364040
    },
    {
      "epoch": 0.2732166258981816,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012719020000000002,
      "loss": 3.2031,
      "step": 364050
    },
    {
      "epoch": 0.27323526913433976,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001271882,
      "loss": 3.2166,
      "step": 364060
    },
    {
      "epoch": 0.27325391237049795,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001271862,
      "loss": 3.1968,
      "step": 364070
    },
    {
      "epoch": 0.27327255560665614,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001271842,
      "loss": 3.077,
      "step": 364080
    },
    {
      "epoch": 0.27329119884281433,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001271822,
      "loss": 3.0125,
      "step": 364090
    },
    {
      "epoch": 0.2733098420789725,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012718020000000003,
      "loss": 3.2156,
      "step": 364100
    },
    {
      "epoch": 0.2733284853151307,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001271782,
      "loss": 3.2621,
      "step": 364110
    },
    {
      "epoch": 0.2733471285512889,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012717620000000002,
      "loss": 3.2005,
      "step": 364120
    },
    {
      "epoch": 0.2733657717874471,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012717419999999999,
      "loss": 3.2973,
      "step": 364130
    },
    {
      "epoch": 0.2733844150236053,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001271722,
      "loss": 3.1198,
      "step": 364140
    },
    {
      "epoch": 0.2734030582597634,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001271702,
      "loss": 3.1578,
      "step": 364150
    },
    {
      "epoch": 0.2734217014959216,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001271682,
      "loss": 3.1113,
      "step": 364160
    },
    {
      "epoch": 0.2734403447320798,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012716620000000002,
      "loss": 3.1326,
      "step": 364170
    },
    {
      "epoch": 0.273458987968238,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001271642,
      "loss": 3.2416,
      "step": 364180
    },
    {
      "epoch": 0.27347763120439617,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012716220000000001,
      "loss": 3.0856,
      "step": 364190
    },
    {
      "epoch": 0.27349627444055435,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001271602,
      "loss": 3.1142,
      "step": 364200
    },
    {
      "epoch": 0.27351491767671254,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001271582,
      "loss": 3.0659,
      "step": 364210
    },
    {
      "epoch": 0.27353356091287073,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001271562,
      "loss": 3.2086,
      "step": 364220
    },
    {
      "epoch": 0.2735522041490289,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001271542,
      "loss": 3.1929,
      "step": 364230
    },
    {
      "epoch": 0.2735708473851871,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012715220000000002,
      "loss": 3.2167,
      "step": 364240
    },
    {
      "epoch": 0.27358949062134524,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001271502,
      "loss": 3.2468,
      "step": 364250
    },
    {
      "epoch": 0.27360813385750343,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001271482,
      "loss": 3.0534,
      "step": 364260
    },
    {
      "epoch": 0.2736267770936616,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001271462,
      "loss": 3.033,
      "step": 364270
    },
    {
      "epoch": 0.2736454203298198,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001271442,
      "loss": 3.145,
      "step": 364280
    },
    {
      "epoch": 0.273664063565978,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012714220000000002,
      "loss": 3.1604,
      "step": 364290
    },
    {
      "epoch": 0.2736827068021362,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001271402,
      "loss": 3.1447,
      "step": 364300
    },
    {
      "epoch": 0.2737013500382944,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012713820000000002,
      "loss": 3.1992,
      "step": 364310
    },
    {
      "epoch": 0.27371999327445257,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001271362,
      "loss": 3.1757,
      "step": 364320
    },
    {
      "epoch": 0.27373863651061076,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001271342,
      "loss": 3.2198,
      "step": 364330
    },
    {
      "epoch": 0.27375727974676894,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001271322,
      "loss": 3.1396,
      "step": 364340
    },
    {
      "epoch": 0.27377592298292713,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001271302,
      "loss": 3.2468,
      "step": 364350
    },
    {
      "epoch": 0.27379456621908527,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012712820000000002,
      "loss": 3.0528,
      "step": 364360
    },
    {
      "epoch": 0.27381320945524346,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001271262,
      "loss": 3.1497,
      "step": 364370
    },
    {
      "epoch": 0.27383185269140164,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001271242,
      "loss": 3.1051,
      "step": 364380
    },
    {
      "epoch": 0.27385049592755983,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001271222,
      "loss": 3.1437,
      "step": 364390
    },
    {
      "epoch": 0.273869139163718,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001271202,
      "loss": 3.128,
      "step": 364400
    },
    {
      "epoch": 0.2738877823998762,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001271182,
      "loss": 3.1206,
      "step": 364410
    },
    {
      "epoch": 0.2739064256360344,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001271162,
      "loss": 3.2094,
      "step": 364420
    },
    {
      "epoch": 0.2739250688721926,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012711420000000002,
      "loss": 3.1736,
      "step": 364430
    },
    {
      "epoch": 0.2739437121083508,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001271122,
      "loss": 3.096,
      "step": 364440
    },
    {
      "epoch": 0.27396235534450897,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001271102,
      "loss": 3.2015,
      "step": 364450
    },
    {
      "epoch": 0.2739809985806671,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001271082,
      "loss": 3.1617,
      "step": 364460
    },
    {
      "epoch": 0.2739996418168253,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001271062,
      "loss": 3.0586,
      "step": 364470
    },
    {
      "epoch": 0.2740182850529835,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001271042,
      "loss": 3.1059,
      "step": 364480
    },
    {
      "epoch": 0.27403692828914167,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012710220000000002,
      "loss": 3.0803,
      "step": 364490
    },
    {
      "epoch": 0.27405557152529986,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001271002,
      "loss": 3.0002,
      "step": 364500
    },
    {
      "epoch": 0.27407421476145805,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001270982,
      "loss": 3.1132,
      "step": 364510
    },
    {
      "epoch": 0.27409285799761623,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001270962,
      "loss": 3.0758,
      "step": 364520
    },
    {
      "epoch": 0.2741115012337744,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001270942,
      "loss": 3.1463,
      "step": 364530
    },
    {
      "epoch": 0.2741301444699326,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012709220000000002,
      "loss": 3.0398,
      "step": 364540
    },
    {
      "epoch": 0.2741487877060908,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001270902,
      "loss": 3.1336,
      "step": 364550
    },
    {
      "epoch": 0.274167430942249,
      "grad_norm": 0.859375,
      "learning_rate": 0.00012708820000000001,
      "loss": 3.0728,
      "step": 364560
    },
    {
      "epoch": 0.2741860741784071,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001270862,
      "loss": 3.033,
      "step": 364570
    },
    {
      "epoch": 0.2742047174145653,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001270842,
      "loss": 3.0755,
      "step": 364580
    },
    {
      "epoch": 0.2742233606507235,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001270822,
      "loss": 3.1249,
      "step": 364590
    },
    {
      "epoch": 0.2742420038868817,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001270802,
      "loss": 3.0899,
      "step": 364600
    },
    {
      "epoch": 0.2742606471230399,
      "grad_norm": 0.625,
      "learning_rate": 0.00012707820000000002,
      "loss": 3.2871,
      "step": 364610
    },
    {
      "epoch": 0.27427929035919807,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001270762,
      "loss": 3.2181,
      "step": 364620
    },
    {
      "epoch": 0.27429793359535626,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001270742,
      "loss": 3.1608,
      "step": 364630
    },
    {
      "epoch": 0.27431657683151445,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001270722,
      "loss": 3.141,
      "step": 364640
    },
    {
      "epoch": 0.27433522006767264,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001270702,
      "loss": 3.1333,
      "step": 364650
    },
    {
      "epoch": 0.2743538633038308,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012706820000000002,
      "loss": 3.0447,
      "step": 364660
    },
    {
      "epoch": 0.27437250653998896,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001270662,
      "loss": 3.2897,
      "step": 364670
    },
    {
      "epoch": 0.27439114977614715,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012706420000000002,
      "loss": 3.1347,
      "step": 364680
    },
    {
      "epoch": 0.27440979301230534,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001270622,
      "loss": 3.1238,
      "step": 364690
    },
    {
      "epoch": 0.2744284362484635,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001270602,
      "loss": 3.1024,
      "step": 364700
    },
    {
      "epoch": 0.2744470794846217,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001270582,
      "loss": 3.1359,
      "step": 364710
    },
    {
      "epoch": 0.2744657227207799,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001270562,
      "loss": 3.0852,
      "step": 364720
    },
    {
      "epoch": 0.2744843659569381,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012705420000000002,
      "loss": 3.1916,
      "step": 364730
    },
    {
      "epoch": 0.2745030091930963,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001270522,
      "loss": 2.9498,
      "step": 364740
    },
    {
      "epoch": 0.27452165242925447,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001270502,
      "loss": 3.073,
      "step": 364750
    },
    {
      "epoch": 0.27454029566541266,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001270482,
      "loss": 3.2014,
      "step": 364760
    },
    {
      "epoch": 0.27455893890157085,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001270462,
      "loss": 3.1787,
      "step": 364770
    },
    {
      "epoch": 0.274577582137729,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012704420000000003,
      "loss": 3.0881,
      "step": 364780
    },
    {
      "epoch": 0.27459622537388717,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001270422,
      "loss": 3.1307,
      "step": 364790
    },
    {
      "epoch": 0.27461486861004536,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012704020000000002,
      "loss": 3.1403,
      "step": 364800
    },
    {
      "epoch": 0.27463351184620355,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00012703819999999999,
      "loss": 3.1642,
      "step": 364810
    },
    {
      "epoch": 0.27465215508236174,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001270362,
      "loss": 3.1923,
      "step": 364820
    },
    {
      "epoch": 0.2746707983185199,
      "grad_norm": 0.625,
      "learning_rate": 0.0001270342,
      "loss": 3.2194,
      "step": 364830
    },
    {
      "epoch": 0.2746894415546781,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001270322,
      "loss": 3.088,
      "step": 364840
    },
    {
      "epoch": 0.2747080847908363,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012703020000000002,
      "loss": 3.0343,
      "step": 364850
    },
    {
      "epoch": 0.2747267280269945,
      "grad_norm": 0.75,
      "learning_rate": 0.0001270282,
      "loss": 3.0778,
      "step": 364860
    },
    {
      "epoch": 0.2747453712631527,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012702620000000001,
      "loss": 3.123,
      "step": 364870
    },
    {
      "epoch": 0.2747640144993108,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001270242,
      "loss": 3.1108,
      "step": 364880
    },
    {
      "epoch": 0.274782657735469,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001270222,
      "loss": 3.0226,
      "step": 364890
    },
    {
      "epoch": 0.2748013009716272,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001270202,
      "loss": 3.1727,
      "step": 364900
    },
    {
      "epoch": 0.2748199442077854,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001270182,
      "loss": 3.0483,
      "step": 364910
    },
    {
      "epoch": 0.27483858744394357,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012701620000000002,
      "loss": 3.1037,
      "step": 364920
    },
    {
      "epoch": 0.27485723068010176,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012701420000000001,
      "loss": 3.1791,
      "step": 364930
    },
    {
      "epoch": 0.27487587391625995,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001270122,
      "loss": 2.9795,
      "step": 364940
    },
    {
      "epoch": 0.27489451715241814,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001270102,
      "loss": 3.1715,
      "step": 364950
    },
    {
      "epoch": 0.2749131603885763,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001270082,
      "loss": 3.1196,
      "step": 364960
    },
    {
      "epoch": 0.2749318036247345,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001270062,
      "loss": 2.9635,
      "step": 364970
    },
    {
      "epoch": 0.2749504468608927,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001270042,
      "loss": 3.1042,
      "step": 364980
    },
    {
      "epoch": 0.27496909009705084,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012700220000000002,
      "loss": 3.105,
      "step": 364990
    },
    {
      "epoch": 0.27498773333320903,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001270002,
      "loss": 3.2445,
      "step": 365000
    },
    {
      "epoch": 0.2750063765693672,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001269982,
      "loss": 3.1501,
      "step": 365010
    },
    {
      "epoch": 0.2750250198055254,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001269962,
      "loss": 3.1319,
      "step": 365020
    },
    {
      "epoch": 0.2750436630416836,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001269942,
      "loss": 3.1427,
      "step": 365030
    },
    {
      "epoch": 0.2750623062778418,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001269922,
      "loss": 3.1399,
      "step": 365040
    },
    {
      "epoch": 0.275080949514,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012699020000000002,
      "loss": 3.1287,
      "step": 365050
    },
    {
      "epoch": 0.27509959275015816,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001269882,
      "loss": 3.1094,
      "step": 365060
    },
    {
      "epoch": 0.27511823598631635,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001269862,
      "loss": 3.0321,
      "step": 365070
    },
    {
      "epoch": 0.27513687922247454,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001269842,
      "loss": 3.0728,
      "step": 365080
    },
    {
      "epoch": 0.2751555224586327,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001269822,
      "loss": 3.0874,
      "step": 365090
    },
    {
      "epoch": 0.27517416569479086,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012698020000000002,
      "loss": 3.0101,
      "step": 365100
    },
    {
      "epoch": 0.27519280893094905,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001269782,
      "loss": 3.2247,
      "step": 365110
    },
    {
      "epoch": 0.27521145216710724,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001269762,
      "loss": 3.0894,
      "step": 365120
    },
    {
      "epoch": 0.27523009540326543,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001269742,
      "loss": 3.0486,
      "step": 365130
    },
    {
      "epoch": 0.2752487386394236,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001269722,
      "loss": 3.0979,
      "step": 365140
    },
    {
      "epoch": 0.2752673818755818,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012697020000000003,
      "loss": 3.0998,
      "step": 365150
    },
    {
      "epoch": 0.27528602511174,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001269682,
      "loss": 3.0418,
      "step": 365160
    },
    {
      "epoch": 0.2753046683478982,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012696620000000002,
      "loss": 3.137,
      "step": 365170
    },
    {
      "epoch": 0.2753233115840564,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00012696419999999999,
      "loss": 3.2329,
      "step": 365180
    },
    {
      "epoch": 0.2753419548202145,
      "grad_norm": 0.75,
      "learning_rate": 0.0001269622,
      "loss": 3.1207,
      "step": 365190
    },
    {
      "epoch": 0.2753605980563727,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001269602,
      "loss": 3.1531,
      "step": 365200
    },
    {
      "epoch": 0.2753792412925309,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001269582,
      "loss": 3.1338,
      "step": 365210
    },
    {
      "epoch": 0.2753978845286891,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012695620000000002,
      "loss": 3.0789,
      "step": 365220
    },
    {
      "epoch": 0.27541652776484726,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001269542,
      "loss": 3.1018,
      "step": 365230
    },
    {
      "epoch": 0.27543517100100545,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012695220000000001,
      "loss": 3.1467,
      "step": 365240
    },
    {
      "epoch": 0.27545381423716364,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001269502,
      "loss": 3.2188,
      "step": 365250
    },
    {
      "epoch": 0.27547245747332183,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001269482,
      "loss": 3.108,
      "step": 365260
    },
    {
      "epoch": 0.27549110070948,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012694620000000003,
      "loss": 3.1458,
      "step": 365270
    },
    {
      "epoch": 0.2755097439456382,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001269442,
      "loss": 3.0589,
      "step": 365280
    },
    {
      "epoch": 0.2755283871817964,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012694220000000002,
      "loss": 3.1575,
      "step": 365290
    },
    {
      "epoch": 0.27554703041795453,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001269402,
      "loss": 3.0445,
      "step": 365300
    },
    {
      "epoch": 0.2755656736541127,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001269382,
      "loss": 3.0063,
      "step": 365310
    },
    {
      "epoch": 0.2755843168902709,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001269362,
      "loss": 3.0933,
      "step": 365320
    },
    {
      "epoch": 0.2756029601264291,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001269342,
      "loss": 3.1198,
      "step": 365330
    },
    {
      "epoch": 0.2756216033625873,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012693220000000002,
      "loss": 3.0402,
      "step": 365340
    },
    {
      "epoch": 0.2756402465987455,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001269302,
      "loss": 3.0686,
      "step": 365350
    },
    {
      "epoch": 0.27565888983490366,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012692820000000002,
      "loss": 3.2246,
      "step": 365360
    },
    {
      "epoch": 0.27567753307106185,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012692619999999998,
      "loss": 2.9901,
      "step": 365370
    },
    {
      "epoch": 0.27569617630722004,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001269242,
      "loss": 2.9763,
      "step": 365380
    },
    {
      "epoch": 0.27571481954337823,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001269222,
      "loss": 3.0623,
      "step": 365390
    },
    {
      "epoch": 0.27573346277953636,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001269202,
      "loss": 3.208,
      "step": 365400
    },
    {
      "epoch": 0.27575210601569455,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012691820000000002,
      "loss": 3.0765,
      "step": 365410
    },
    {
      "epoch": 0.27577074925185274,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001269162,
      "loss": 3.0924,
      "step": 365420
    },
    {
      "epoch": 0.27578939248801093,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001269142,
      "loss": 3.1305,
      "step": 365430
    },
    {
      "epoch": 0.2758080357241691,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001269122,
      "loss": 3.1243,
      "step": 365440
    },
    {
      "epoch": 0.2758266789603273,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001269102,
      "loss": 3.2264,
      "step": 365450
    },
    {
      "epoch": 0.2758453221964855,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001269082,
      "loss": 3.1486,
      "step": 365460
    },
    {
      "epoch": 0.2758639654326437,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001269062,
      "loss": 3.1409,
      "step": 365470
    },
    {
      "epoch": 0.2758826086688019,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012690420000000002,
      "loss": 3.1053,
      "step": 365480
    },
    {
      "epoch": 0.27590125190496007,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001269022,
      "loss": 2.998,
      "step": 365490
    },
    {
      "epoch": 0.27591989514111825,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001269002,
      "loss": 3.1477,
      "step": 365500
    },
    {
      "epoch": 0.2759385383772764,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001268982,
      "loss": 3.1491,
      "step": 365510
    },
    {
      "epoch": 0.2759571816134346,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001268962,
      "loss": 3.1174,
      "step": 365520
    },
    {
      "epoch": 0.27597582484959277,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001268942,
      "loss": 2.9431,
      "step": 365530
    },
    {
      "epoch": 0.27599446808575095,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012689220000000002,
      "loss": 3.1546,
      "step": 365540
    },
    {
      "epoch": 0.27601311132190914,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012689020000000001,
      "loss": 3.0129,
      "step": 365550
    },
    {
      "epoch": 0.27603175455806733,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001268882,
      "loss": 3.0699,
      "step": 365560
    },
    {
      "epoch": 0.2760503977942255,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001268862,
      "loss": 3.1525,
      "step": 365570
    },
    {
      "epoch": 0.2760690410303837,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001268842,
      "loss": 3.1376,
      "step": 365580
    },
    {
      "epoch": 0.2760876842665419,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001268822,
      "loss": 3.0719,
      "step": 365590
    },
    {
      "epoch": 0.2761063275027001,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001268802,
      "loss": 3.174,
      "step": 365600
    },
    {
      "epoch": 0.2761249707388582,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012687820000000001,
      "loss": 3.1687,
      "step": 365610
    },
    {
      "epoch": 0.2761436139750164,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001268762,
      "loss": 3.1005,
      "step": 365620
    },
    {
      "epoch": 0.2761622572111746,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001268742,
      "loss": 3.118,
      "step": 365630
    },
    {
      "epoch": 0.2761809004473328,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001268722,
      "loss": 3.0458,
      "step": 365640
    },
    {
      "epoch": 0.276199543683491,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001268702,
      "loss": 2.9526,
      "step": 365650
    },
    {
      "epoch": 0.27621818691964917,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012686820000000002,
      "loss": 3.0118,
      "step": 365660
    },
    {
      "epoch": 0.27623683015580736,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001268662,
      "loss": 3.1071,
      "step": 365670
    },
    {
      "epoch": 0.27625547339196554,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001268642,
      "loss": 3.1371,
      "step": 365680
    },
    {
      "epoch": 0.27627411662812373,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001268622,
      "loss": 3.1052,
      "step": 365690
    },
    {
      "epoch": 0.2762927598642819,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001268602,
      "loss": 3.089,
      "step": 365700
    },
    {
      "epoch": 0.2763114031004401,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012685820000000002,
      "loss": 3.1709,
      "step": 365710
    },
    {
      "epoch": 0.27633004633659825,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001268562,
      "loss": 3.1199,
      "step": 365720
    },
    {
      "epoch": 0.27634868957275643,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012685420000000002,
      "loss": 3.0353,
      "step": 365730
    },
    {
      "epoch": 0.2763673328089146,
      "grad_norm": 0.75,
      "learning_rate": 0.0001268522,
      "loss": 3.1883,
      "step": 365740
    },
    {
      "epoch": 0.2763859760450728,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001268502,
      "loss": 3.02,
      "step": 365750
    },
    {
      "epoch": 0.276404619281231,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012684820000000003,
      "loss": 3.0588,
      "step": 365760
    },
    {
      "epoch": 0.2764232625173892,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001268462,
      "loss": 3.2254,
      "step": 365770
    },
    {
      "epoch": 0.2764419057535474,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012684420000000002,
      "loss": 3.1384,
      "step": 365780
    },
    {
      "epoch": 0.27646054898970557,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001268422,
      "loss": 3.1731,
      "step": 365790
    },
    {
      "epoch": 0.27647919222586376,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001268402,
      "loss": 3.2672,
      "step": 365800
    },
    {
      "epoch": 0.27649783546202195,
      "grad_norm": 0.625,
      "learning_rate": 0.0001268382,
      "loss": 3.1678,
      "step": 365810
    },
    {
      "epoch": 0.2765164786981801,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001268362,
      "loss": 3.051,
      "step": 365820
    },
    {
      "epoch": 0.27653512193433827,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012683420000000003,
      "loss": 3.0672,
      "step": 365830
    },
    {
      "epoch": 0.27655376517049646,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001268322,
      "loss": 3.1771,
      "step": 365840
    },
    {
      "epoch": 0.27657240840665465,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012683020000000002,
      "loss": 3.1439,
      "step": 365850
    },
    {
      "epoch": 0.27659105164281284,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012682819999999999,
      "loss": 3.298,
      "step": 365860
    },
    {
      "epoch": 0.276609694878971,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001268262,
      "loss": 3.0749,
      "step": 365870
    },
    {
      "epoch": 0.2766283381151292,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001268242,
      "loss": 3.1894,
      "step": 365880
    },
    {
      "epoch": 0.2766469813512874,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001268222,
      "loss": 3.0195,
      "step": 365890
    },
    {
      "epoch": 0.2766656245874456,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012682020000000002,
      "loss": 3.0199,
      "step": 365900
    },
    {
      "epoch": 0.2766842678236038,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001268182,
      "loss": 3.2043,
      "step": 365910
    },
    {
      "epoch": 0.27670291105976197,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012681620000000001,
      "loss": 3.0911,
      "step": 365920
    },
    {
      "epoch": 0.2767215542959201,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001268142,
      "loss": 3.0787,
      "step": 365930
    },
    {
      "epoch": 0.2767401975320783,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001268122,
      "loss": 3.0999,
      "step": 365940
    },
    {
      "epoch": 0.2767588407682365,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001268102,
      "loss": 3.1504,
      "step": 365950
    },
    {
      "epoch": 0.27677748400439467,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001268082,
      "loss": 3.2627,
      "step": 365960
    },
    {
      "epoch": 0.27679612724055286,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012680620000000002,
      "loss": 3.2453,
      "step": 365970
    },
    {
      "epoch": 0.27681477047671105,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001268042,
      "loss": 3.1416,
      "step": 365980
    },
    {
      "epoch": 0.27683341371286924,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001268022,
      "loss": 3.197,
      "step": 365990
    },
    {
      "epoch": 0.2768520569490274,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001268002,
      "loss": 3.1121,
      "step": 366000
    },
    {
      "epoch": 0.2768707001851856,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001267982,
      "loss": 3.1349,
      "step": 366010
    },
    {
      "epoch": 0.2768893434213438,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001267962,
      "loss": 3.1896,
      "step": 366020
    },
    {
      "epoch": 0.27690798665750194,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001267942,
      "loss": 3.2186,
      "step": 366030
    },
    {
      "epoch": 0.2769266298936601,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012679220000000002,
      "loss": 3.2016,
      "step": 366040
    },
    {
      "epoch": 0.2769452731298183,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001267902,
      "loss": 3.1695,
      "step": 366050
    },
    {
      "epoch": 0.2769639163659765,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001267882,
      "loss": 3.114,
      "step": 366060
    },
    {
      "epoch": 0.2769825596021347,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001267862,
      "loss": 3.0911,
      "step": 366070
    },
    {
      "epoch": 0.2770012028382929,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001267842,
      "loss": 3.1037,
      "step": 366080
    },
    {
      "epoch": 0.27701984607445107,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001267822,
      "loss": 3.0878,
      "step": 366090
    },
    {
      "epoch": 0.27703848931060926,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012678020000000002,
      "loss": 3.108,
      "step": 366100
    },
    {
      "epoch": 0.27705713254676745,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001267782,
      "loss": 3.0907,
      "step": 366110
    },
    {
      "epoch": 0.27707577578292564,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001267762,
      "loss": 3.2201,
      "step": 366120
    },
    {
      "epoch": 0.2770944190190838,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001267742,
      "loss": 3.1616,
      "step": 366130
    },
    {
      "epoch": 0.27711306225524196,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001267722,
      "loss": 3.0265,
      "step": 366140
    },
    {
      "epoch": 0.27713170549140015,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012677020000000002,
      "loss": 3.0637,
      "step": 366150
    },
    {
      "epoch": 0.27715034872755834,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001267682,
      "loss": 3.2168,
      "step": 366160
    },
    {
      "epoch": 0.2771689919637165,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001267662,
      "loss": 3.0916,
      "step": 366170
    },
    {
      "epoch": 0.2771876351998747,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001267642,
      "loss": 3.1576,
      "step": 366180
    },
    {
      "epoch": 0.2772062784360329,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001267622,
      "loss": 3.0967,
      "step": 366190
    },
    {
      "epoch": 0.2772249216721911,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001267602,
      "loss": 2.9986,
      "step": 366200
    },
    {
      "epoch": 0.2772435649083493,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001267582,
      "loss": 2.9552,
      "step": 366210
    },
    {
      "epoch": 0.27726220814450747,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012675620000000002,
      "loss": 3.1352,
      "step": 366220
    },
    {
      "epoch": 0.27728085138066566,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012675420000000001,
      "loss": 3.2097,
      "step": 366230
    },
    {
      "epoch": 0.2772994946168238,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001267522,
      "loss": 3.2063,
      "step": 366240
    },
    {
      "epoch": 0.277318137852982,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001267502,
      "loss": 3.0746,
      "step": 366250
    },
    {
      "epoch": 0.27733678108914017,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001267482,
      "loss": 3.1053,
      "step": 366260
    },
    {
      "epoch": 0.27735542432529836,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012674620000000002,
      "loss": 3.2158,
      "step": 366270
    },
    {
      "epoch": 0.27737406756145655,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001267442,
      "loss": 3.3544,
      "step": 366280
    },
    {
      "epoch": 0.27739271079761474,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012674220000000001,
      "loss": 3.0789,
      "step": 366290
    },
    {
      "epoch": 0.27741135403377293,
      "grad_norm": 0.75,
      "learning_rate": 0.0001267402,
      "loss": 3.2214,
      "step": 366300
    },
    {
      "epoch": 0.2774299972699311,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001267382,
      "loss": 3.1537,
      "step": 366310
    },
    {
      "epoch": 0.2774486405060893,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012673620000000003,
      "loss": 3.1493,
      "step": 366320
    },
    {
      "epoch": 0.2774672837422475,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001267342,
      "loss": 3.1094,
      "step": 366330
    },
    {
      "epoch": 0.2774859269784057,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012673220000000002,
      "loss": 3.1174,
      "step": 366340
    },
    {
      "epoch": 0.2775045702145638,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001267302,
      "loss": 3.0379,
      "step": 366350
    },
    {
      "epoch": 0.277523213450722,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001267282,
      "loss": 3.0985,
      "step": 366360
    },
    {
      "epoch": 0.2775418566868802,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001267262,
      "loss": 3.1037,
      "step": 366370
    },
    {
      "epoch": 0.2775604999230384,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001267242,
      "loss": 3.1583,
      "step": 366380
    },
    {
      "epoch": 0.2775791431591966,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012672220000000002,
      "loss": 3.1457,
      "step": 366390
    },
    {
      "epoch": 0.27759778639535476,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001267202,
      "loss": 3.0508,
      "step": 366400
    },
    {
      "epoch": 0.27761642963151295,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012671820000000002,
      "loss": 3.0998,
      "step": 366410
    },
    {
      "epoch": 0.27763507286767114,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012671619999999998,
      "loss": 3.0517,
      "step": 366420
    },
    {
      "epoch": 0.27765371610382933,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001267142,
      "loss": 3.1898,
      "step": 366430
    },
    {
      "epoch": 0.2776723593399875,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001267122,
      "loss": 3.1319,
      "step": 366440
    },
    {
      "epoch": 0.27769100257614565,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001267102,
      "loss": 3.0825,
      "step": 366450
    },
    {
      "epoch": 0.27770964581230384,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012670820000000002,
      "loss": 3.1129,
      "step": 366460
    },
    {
      "epoch": 0.27772828904846203,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001267062,
      "loss": 3.2154,
      "step": 366470
    },
    {
      "epoch": 0.2777469322846202,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001267042,
      "loss": 3.173,
      "step": 366480
    },
    {
      "epoch": 0.2777655755207784,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001267022,
      "loss": 3.0229,
      "step": 366490
    },
    {
      "epoch": 0.2777842187569366,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001267002,
      "loss": 3.0683,
      "step": 366500
    },
    {
      "epoch": 0.2778028619930948,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001266982,
      "loss": 3.1158,
      "step": 366510
    },
    {
      "epoch": 0.277821505229253,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001266962,
      "loss": 3.1154,
      "step": 366520
    },
    {
      "epoch": 0.27784014846541116,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012669420000000002,
      "loss": 3.0151,
      "step": 366530
    },
    {
      "epoch": 0.27785879170156935,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001266922,
      "loss": 3.0213,
      "step": 366540
    },
    {
      "epoch": 0.2778774349377275,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001266902,
      "loss": 3.0636,
      "step": 366550
    },
    {
      "epoch": 0.2778960781738857,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001266882,
      "loss": 3.1707,
      "step": 366560
    },
    {
      "epoch": 0.27791472141004386,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001266862,
      "loss": 3.1465,
      "step": 366570
    },
    {
      "epoch": 0.27793336464620205,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001266842,
      "loss": 3.2451,
      "step": 366580
    },
    {
      "epoch": 0.27795200788236024,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001266822,
      "loss": 3.0859,
      "step": 366590
    },
    {
      "epoch": 0.27797065111851843,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012668020000000001,
      "loss": 3.188,
      "step": 366600
    },
    {
      "epoch": 0.2779892943546766,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001266782,
      "loss": 3.0656,
      "step": 366610
    },
    {
      "epoch": 0.2780079375908348,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001266762,
      "loss": 3.0133,
      "step": 366620
    },
    {
      "epoch": 0.278026580826993,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001266742,
      "loss": 3.1837,
      "step": 366630
    },
    {
      "epoch": 0.2780452240631512,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001266722,
      "loss": 3.1131,
      "step": 366640
    },
    {
      "epoch": 0.2780638672993094,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001266702,
      "loss": 3.1705,
      "step": 366650
    },
    {
      "epoch": 0.2780825105354675,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012666820000000001,
      "loss": 3.0966,
      "step": 366660
    },
    {
      "epoch": 0.2781011537716257,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001266662,
      "loss": 3.3069,
      "step": 366670
    },
    {
      "epoch": 0.2781197970077839,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001266642,
      "loss": 3.0066,
      "step": 366680
    },
    {
      "epoch": 0.2781384402439421,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001266622,
      "loss": 3.3504,
      "step": 366690
    },
    {
      "epoch": 0.27815708348010026,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001266602,
      "loss": 3.2343,
      "step": 366700
    },
    {
      "epoch": 0.27817572671625845,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012665820000000002,
      "loss": 3.1595,
      "step": 366710
    },
    {
      "epoch": 0.27819436995241664,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012665620000000002,
      "loss": 3.1213,
      "step": 366720
    },
    {
      "epoch": 0.27821301318857483,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001266542,
      "loss": 3.0727,
      "step": 366730
    },
    {
      "epoch": 0.278231656424733,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001266522,
      "loss": 3.1131,
      "step": 366740
    },
    {
      "epoch": 0.2782502996608912,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001266502,
      "loss": 3.042,
      "step": 366750
    },
    {
      "epoch": 0.27826894289704934,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012664820000000003,
      "loss": 3.1937,
      "step": 366760
    },
    {
      "epoch": 0.27828758613320753,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001266462,
      "loss": 2.9911,
      "step": 366770
    },
    {
      "epoch": 0.2783062293693657,
      "grad_norm": 0.75,
      "learning_rate": 0.00012664420000000002,
      "loss": 3.146,
      "step": 366780
    },
    {
      "epoch": 0.2783248726055239,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001266422,
      "loss": 3.0917,
      "step": 366790
    },
    {
      "epoch": 0.2783435158416821,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001266402,
      "loss": 3.1436,
      "step": 366800
    },
    {
      "epoch": 0.2783621590778403,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001266382,
      "loss": 3.1562,
      "step": 366810
    },
    {
      "epoch": 0.2783808023139985,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001266362,
      "loss": 3.1335,
      "step": 366820
    },
    {
      "epoch": 0.27839944555015667,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012663420000000002,
      "loss": 3.0308,
      "step": 366830
    },
    {
      "epoch": 0.27841808878631485,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001266322,
      "loss": 3.1287,
      "step": 366840
    },
    {
      "epoch": 0.27843673202247304,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001266302,
      "loss": 3.1367,
      "step": 366850
    },
    {
      "epoch": 0.27845537525863123,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001266282,
      "loss": 3.1032,
      "step": 366860
    },
    {
      "epoch": 0.27847401849478937,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001266262,
      "loss": 3.187,
      "step": 366870
    },
    {
      "epoch": 0.27849266173094755,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012662420000000003,
      "loss": 3.2467,
      "step": 366880
    },
    {
      "epoch": 0.27851130496710574,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001266222,
      "loss": 3.1726,
      "step": 366890
    },
    {
      "epoch": 0.27852994820326393,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012662020000000002,
      "loss": 3.2095,
      "step": 366900
    },
    {
      "epoch": 0.2785485914394221,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001266182,
      "loss": 3.0013,
      "step": 366910
    },
    {
      "epoch": 0.2785672346755803,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001266162,
      "loss": 3.1754,
      "step": 366920
    },
    {
      "epoch": 0.2785858779117385,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001266142,
      "loss": 3.1563,
      "step": 366930
    },
    {
      "epoch": 0.2786045211478967,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001266122,
      "loss": 3.1181,
      "step": 366940
    },
    {
      "epoch": 0.2786231643840549,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012661020000000002,
      "loss": 3.0772,
      "step": 366950
    },
    {
      "epoch": 0.27864180762021307,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001266082,
      "loss": 3.0948,
      "step": 366960
    },
    {
      "epoch": 0.2786604508563712,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012660620000000001,
      "loss": 3.1401,
      "step": 366970
    },
    {
      "epoch": 0.2786790940925294,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012660419999999998,
      "loss": 3.1851,
      "step": 366980
    },
    {
      "epoch": 0.2786977373286876,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001266022,
      "loss": 3.0252,
      "step": 366990
    },
    {
      "epoch": 0.27871638056484577,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001266002,
      "loss": 3.1726,
      "step": 367000
    },
    {
      "epoch": 0.27873502380100396,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001265982,
      "loss": 3.1282,
      "step": 367010
    },
    {
      "epoch": 0.27875366703716215,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012659620000000002,
      "loss": 3.271,
      "step": 367020
    },
    {
      "epoch": 0.27877231027332033,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001265942,
      "loss": 3.0093,
      "step": 367030
    },
    {
      "epoch": 0.2787909535094785,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001265922,
      "loss": 3.1241,
      "step": 367040
    },
    {
      "epoch": 0.2788095967456367,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001265902,
      "loss": 3.1553,
      "step": 367050
    },
    {
      "epoch": 0.2788282399817949,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001265882,
      "loss": 3.0816,
      "step": 367060
    },
    {
      "epoch": 0.2788468832179531,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001265862,
      "loss": 3.1935,
      "step": 367070
    },
    {
      "epoch": 0.2788655264541112,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001265842,
      "loss": 3.0834,
      "step": 367080
    },
    {
      "epoch": 0.2788841696902694,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012658220000000002,
      "loss": 3.1552,
      "step": 367090
    },
    {
      "epoch": 0.2789028129264276,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001265802,
      "loss": 3.111,
      "step": 367100
    },
    {
      "epoch": 0.2789214561625858,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001265782,
      "loss": 3.1683,
      "step": 367110
    },
    {
      "epoch": 0.278940099398744,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001265762,
      "loss": 3.0973,
      "step": 367120
    },
    {
      "epoch": 0.27895874263490217,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001265742,
      "loss": 3.0314,
      "step": 367130
    },
    {
      "epoch": 0.27897738587106036,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001265722,
      "loss": 3.014,
      "step": 367140
    },
    {
      "epoch": 0.27899602910721855,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012657020000000002,
      "loss": 3.1336,
      "step": 367150
    },
    {
      "epoch": 0.27901467234337674,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001265682,
      "loss": 3.0737,
      "step": 367160
    },
    {
      "epoch": 0.2790333155795349,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001265662,
      "loss": 3.1322,
      "step": 367170
    },
    {
      "epoch": 0.27905195881569306,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001265642,
      "loss": 3.1669,
      "step": 367180
    },
    {
      "epoch": 0.27907060205185125,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001265622,
      "loss": 3.1266,
      "step": 367190
    },
    {
      "epoch": 0.27908924528800944,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001265602,
      "loss": 3.0712,
      "step": 367200
    },
    {
      "epoch": 0.2791078885241676,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012655820000000002,
      "loss": 3.2249,
      "step": 367210
    },
    {
      "epoch": 0.2791265317603258,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001265562,
      "loss": 3.1196,
      "step": 367220
    },
    {
      "epoch": 0.279145174996484,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001265542,
      "loss": 3.0769,
      "step": 367230
    },
    {
      "epoch": 0.2791638182326422,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001265522,
      "loss": 3.2106,
      "step": 367240
    },
    {
      "epoch": 0.2791824614688004,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001265502,
      "loss": 3.1215,
      "step": 367250
    },
    {
      "epoch": 0.27920110470495857,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001265482,
      "loss": 2.9562,
      "step": 367260
    },
    {
      "epoch": 0.27921974794111676,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012654620000000002,
      "loss": 3.0947,
      "step": 367270
    },
    {
      "epoch": 0.27923839117727495,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012654420000000001,
      "loss": 3.1102,
      "step": 367280
    },
    {
      "epoch": 0.2792570344134331,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001265422,
      "loss": 3.1373,
      "step": 367290
    },
    {
      "epoch": 0.27927567764959127,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001265402,
      "loss": 3.24,
      "step": 367300
    },
    {
      "epoch": 0.27929432088574946,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001265382,
      "loss": 3.2175,
      "step": 367310
    },
    {
      "epoch": 0.27931296412190765,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012653620000000002,
      "loss": 3.1078,
      "step": 367320
    },
    {
      "epoch": 0.27933160735806584,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001265342,
      "loss": 3.1317,
      "step": 367330
    },
    {
      "epoch": 0.279350250594224,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012653220000000001,
      "loss": 3.0244,
      "step": 367340
    },
    {
      "epoch": 0.2793688938303822,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001265302,
      "loss": 3.0221,
      "step": 367350
    },
    {
      "epoch": 0.2793875370665404,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001265282,
      "loss": 3.266,
      "step": 367360
    },
    {
      "epoch": 0.2794061803026986,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012652620000000003,
      "loss": 3.093,
      "step": 367370
    },
    {
      "epoch": 0.2794248235388568,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001265242,
      "loss": 3.0799,
      "step": 367380
    },
    {
      "epoch": 0.2794434667750149,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012652220000000002,
      "loss": 3.1639,
      "step": 367390
    },
    {
      "epoch": 0.2794621100111731,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001265202,
      "loss": 2.9809,
      "step": 367400
    },
    {
      "epoch": 0.2794807532473313,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001265182,
      "loss": 3.1727,
      "step": 367410
    },
    {
      "epoch": 0.2794993964834895,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001265162,
      "loss": 3.1192,
      "step": 367420
    },
    {
      "epoch": 0.27951803971964767,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001265142,
      "loss": 3.2207,
      "step": 367430
    },
    {
      "epoch": 0.27953668295580586,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012651220000000003,
      "loss": 3.1812,
      "step": 367440
    },
    {
      "epoch": 0.27955532619196405,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001265102,
      "loss": 3.2152,
      "step": 367450
    },
    {
      "epoch": 0.27957396942812224,
      "grad_norm": 0.75,
      "learning_rate": 0.00012650820000000002,
      "loss": 3.1506,
      "step": 367460
    },
    {
      "epoch": 0.2795926126642804,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012650619999999998,
      "loss": 3.0739,
      "step": 367470
    },
    {
      "epoch": 0.2796112559004386,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001265042,
      "loss": 3.3098,
      "step": 367480
    },
    {
      "epoch": 0.2796298991365968,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001265022,
      "loss": 3.0745,
      "step": 367490
    },
    {
      "epoch": 0.27964854237275494,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001265002,
      "loss": 3.0744,
      "step": 367500
    },
    {
      "epoch": 0.2796671856089131,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012649820000000002,
      "loss": 3.1403,
      "step": 367510
    },
    {
      "epoch": 0.2796858288450713,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001264962,
      "loss": 3.0132,
      "step": 367520
    },
    {
      "epoch": 0.2797044720812295,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001264942,
      "loss": 3.1547,
      "step": 367530
    },
    {
      "epoch": 0.2797231153173877,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001264922,
      "loss": 3.0632,
      "step": 367540
    },
    {
      "epoch": 0.2797417585535459,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001264902,
      "loss": 3.1823,
      "step": 367550
    },
    {
      "epoch": 0.27976040178970407,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001264882,
      "loss": 3.0129,
      "step": 367560
    },
    {
      "epoch": 0.27977904502586226,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001264862,
      "loss": 2.9594,
      "step": 367570
    },
    {
      "epoch": 0.27979768826202045,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012648420000000002,
      "loss": 3.2052,
      "step": 367580
    },
    {
      "epoch": 0.27981633149817864,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001264822,
      "loss": 3.1416,
      "step": 367590
    },
    {
      "epoch": 0.27983497473433677,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001264802,
      "loss": 3.1761,
      "step": 367600
    },
    {
      "epoch": 0.27985361797049496,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001264782,
      "loss": 3.2214,
      "step": 367610
    },
    {
      "epoch": 0.27987226120665315,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001264762,
      "loss": 3.2338,
      "step": 367620
    },
    {
      "epoch": 0.27989090444281134,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001264742,
      "loss": 3.0914,
      "step": 367630
    },
    {
      "epoch": 0.27990954767896953,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001264722,
      "loss": 3.1367,
      "step": 367640
    },
    {
      "epoch": 0.2799281909151277,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012647020000000001,
      "loss": 3.0194,
      "step": 367650
    },
    {
      "epoch": 0.2799468341512859,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001264682,
      "loss": 3.112,
      "step": 367660
    },
    {
      "epoch": 0.2799654773874441,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001264662,
      "loss": 3.1441,
      "step": 367670
    },
    {
      "epoch": 0.2799841206236023,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001264642,
      "loss": 3.1395,
      "step": 367680
    },
    {
      "epoch": 0.2800027638597605,
      "grad_norm": 0.75,
      "learning_rate": 0.0001264622,
      "loss": 3.0608,
      "step": 367690
    },
    {
      "epoch": 0.28002140709591866,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012646020000000002,
      "loss": 3.061,
      "step": 367700
    },
    {
      "epoch": 0.2800400503320768,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012645820000000002,
      "loss": 3.1876,
      "step": 367710
    },
    {
      "epoch": 0.280058693568235,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001264562,
      "loss": 3.1463,
      "step": 367720
    },
    {
      "epoch": 0.2800773368043932,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001264542,
      "loss": 3.121,
      "step": 367730
    },
    {
      "epoch": 0.28009598004055136,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001264522,
      "loss": 3.1616,
      "step": 367740
    },
    {
      "epoch": 0.28011462327670955,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001264502,
      "loss": 3.196,
      "step": 367750
    },
    {
      "epoch": 0.28013326651286774,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012644820000000002,
      "loss": 3.2151,
      "step": 367760
    },
    {
      "epoch": 0.28015190974902593,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012644620000000002,
      "loss": 3.1445,
      "step": 367770
    },
    {
      "epoch": 0.2801705529851841,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001264442,
      "loss": 3.0773,
      "step": 367780
    },
    {
      "epoch": 0.2801891962213423,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001264422,
      "loss": 3.1748,
      "step": 367790
    },
    {
      "epoch": 0.2802078394575005,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001264402,
      "loss": 3.0825,
      "step": 367800
    },
    {
      "epoch": 0.28022648269365863,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001264382,
      "loss": 3.3391,
      "step": 367810
    },
    {
      "epoch": 0.2802451259298168,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001264362,
      "loss": 3.2482,
      "step": 367820
    },
    {
      "epoch": 0.280263769165975,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012643420000000002,
      "loss": 3.1663,
      "step": 367830
    },
    {
      "epoch": 0.2802824124021332,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001264322,
      "loss": 3.0789,
      "step": 367840
    },
    {
      "epoch": 0.2803010556382914,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001264302,
      "loss": 3.0538,
      "step": 367850
    },
    {
      "epoch": 0.2803196988744496,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001264282,
      "loss": 3.056,
      "step": 367860
    },
    {
      "epoch": 0.28033834211060776,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001264262,
      "loss": 2.9877,
      "step": 367870
    },
    {
      "epoch": 0.28035698534676595,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012642420000000002,
      "loss": 3.019,
      "step": 367880
    },
    {
      "epoch": 0.28037562858292414,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001264222,
      "loss": 3.0636,
      "step": 367890
    },
    {
      "epoch": 0.28039427181908233,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001264202,
      "loss": 3.0757,
      "step": 367900
    },
    {
      "epoch": 0.28041291505524046,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001264182,
      "loss": 3.2928,
      "step": 367910
    },
    {
      "epoch": 0.28043155829139865,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001264162,
      "loss": 2.9699,
      "step": 367920
    },
    {
      "epoch": 0.28045020152755684,
      "grad_norm": 0.75,
      "learning_rate": 0.00012641420000000003,
      "loss": 3.0993,
      "step": 367930
    },
    {
      "epoch": 0.28046884476371503,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001264122,
      "loss": 3.0331,
      "step": 367940
    },
    {
      "epoch": 0.2804874879998732,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012641020000000002,
      "loss": 3.1215,
      "step": 367950
    },
    {
      "epoch": 0.2805061312360314,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001264082,
      "loss": 3.1857,
      "step": 367960
    },
    {
      "epoch": 0.2805247744721896,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001264062,
      "loss": 3.1768,
      "step": 367970
    },
    {
      "epoch": 0.2805434177083478,
      "grad_norm": 0.75,
      "learning_rate": 0.0001264042,
      "loss": 3.1199,
      "step": 367980
    },
    {
      "epoch": 0.280562060944506,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001264022,
      "loss": 3.1646,
      "step": 367990
    },
    {
      "epoch": 0.28058070418066416,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012640020000000002,
      "loss": 3.1458,
      "step": 368000
    },
    {
      "epoch": 0.28059934741682235,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001263982,
      "loss": 3.0635,
      "step": 368010
    },
    {
      "epoch": 0.2806179906529805,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012639620000000001,
      "loss": 3.201,
      "step": 368020
    },
    {
      "epoch": 0.2806366338891387,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012639419999999998,
      "loss": 3.1256,
      "step": 368030
    },
    {
      "epoch": 0.28065527712529686,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001263922,
      "loss": 3.146,
      "step": 368040
    },
    {
      "epoch": 0.28067392036145505,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001263902,
      "loss": 3.1474,
      "step": 368050
    },
    {
      "epoch": 0.28069256359761324,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001263882,
      "loss": 3.0584,
      "step": 368060
    },
    {
      "epoch": 0.28071120683377143,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012638620000000002,
      "loss": 3.1405,
      "step": 368070
    },
    {
      "epoch": 0.2807298500699296,
      "grad_norm": 0.75,
      "learning_rate": 0.0001263842,
      "loss": 3.1571,
      "step": 368080
    },
    {
      "epoch": 0.2807484933060878,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001263822,
      "loss": 3.0486,
      "step": 368090
    },
    {
      "epoch": 0.280767136542246,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001263802,
      "loss": 3.1898,
      "step": 368100
    },
    {
      "epoch": 0.2807857797784042,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001263782,
      "loss": 3.1873,
      "step": 368110
    },
    {
      "epoch": 0.2808044230145623,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001263762,
      "loss": 3.0191,
      "step": 368120
    },
    {
      "epoch": 0.2808230662507205,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001263742,
      "loss": 3.1095,
      "step": 368130
    },
    {
      "epoch": 0.2808417094868787,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012637220000000002,
      "loss": 3.1157,
      "step": 368140
    },
    {
      "epoch": 0.2808603527230369,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001263702,
      "loss": 3.2017,
      "step": 368150
    },
    {
      "epoch": 0.2808789959591951,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001263682,
      "loss": 3.1457,
      "step": 368160
    },
    {
      "epoch": 0.28089763919535327,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001263662,
      "loss": 3.0169,
      "step": 368170
    },
    {
      "epoch": 0.28091628243151145,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001263642,
      "loss": 3.062,
      "step": 368180
    },
    {
      "epoch": 0.28093492566766964,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012636220000000002,
      "loss": 3.0129,
      "step": 368190
    },
    {
      "epoch": 0.28095356890382783,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001263602,
      "loss": 3.1613,
      "step": 368200
    },
    {
      "epoch": 0.280972212139986,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001263582,
      "loss": 3.129,
      "step": 368210
    },
    {
      "epoch": 0.2809908553761442,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001263562,
      "loss": 3.0396,
      "step": 368220
    },
    {
      "epoch": 0.28100949861230234,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001263542,
      "loss": 3.0667,
      "step": 368230
    },
    {
      "epoch": 0.28102814184846053,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001263522,
      "loss": 3.0783,
      "step": 368240
    },
    {
      "epoch": 0.2810467850846187,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001263502,
      "loss": 3.1268,
      "step": 368250
    },
    {
      "epoch": 0.2810654283207769,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012634820000000002,
      "loss": 3.1736,
      "step": 368260
    },
    {
      "epoch": 0.2810840715569351,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001263462,
      "loss": 3.272,
      "step": 368270
    },
    {
      "epoch": 0.2811027147930933,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001263442,
      "loss": 3.2876,
      "step": 368280
    },
    {
      "epoch": 0.2811213580292515,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001263422,
      "loss": 2.9933,
      "step": 368290
    },
    {
      "epoch": 0.28114000126540967,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001263402,
      "loss": 3.124,
      "step": 368300
    },
    {
      "epoch": 0.28115864450156786,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001263382,
      "loss": 3.1,
      "step": 368310
    },
    {
      "epoch": 0.28117728773772604,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012633620000000002,
      "loss": 3.028,
      "step": 368320
    },
    {
      "epoch": 0.2811959309738842,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012633420000000001,
      "loss": 3.1161,
      "step": 368330
    },
    {
      "epoch": 0.28121457421004237,
      "grad_norm": 0.625,
      "learning_rate": 0.0001263322,
      "loss": 3.2517,
      "step": 368340
    },
    {
      "epoch": 0.28123321744620056,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001263302,
      "loss": 3.0043,
      "step": 368350
    },
    {
      "epoch": 0.28125186068235875,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001263282,
      "loss": 3.1343,
      "step": 368360
    },
    {
      "epoch": 0.28127050391851693,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012632620000000002,
      "loss": 3.1906,
      "step": 368370
    },
    {
      "epoch": 0.2812891471546751,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001263242,
      "loss": 2.9852,
      "step": 368380
    },
    {
      "epoch": 0.2813077903908333,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012632220000000002,
      "loss": 2.9419,
      "step": 368390
    },
    {
      "epoch": 0.2813264336269915,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001263202,
      "loss": 3.1529,
      "step": 368400
    },
    {
      "epoch": 0.2813450768631497,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001263182,
      "loss": 3.2928,
      "step": 368410
    },
    {
      "epoch": 0.2813637200993079,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001263162,
      "loss": 3.0782,
      "step": 368420
    },
    {
      "epoch": 0.28138236333546607,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001263142,
      "loss": 3.1274,
      "step": 368430
    },
    {
      "epoch": 0.2814010065716242,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012631220000000002,
      "loss": 3.1024,
      "step": 368440
    },
    {
      "epoch": 0.2814196498077824,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001263102,
      "loss": 3.1388,
      "step": 368450
    },
    {
      "epoch": 0.2814382930439406,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001263082,
      "loss": 3.0932,
      "step": 368460
    },
    {
      "epoch": 0.28145693628009877,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001263062,
      "loss": 3.2078,
      "step": 368470
    },
    {
      "epoch": 0.28147557951625696,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001263042,
      "loss": 3.2545,
      "step": 368480
    },
    {
      "epoch": 0.28149422275241515,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012630220000000003,
      "loss": 3.071,
      "step": 368490
    },
    {
      "epoch": 0.28151286598857334,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001263002,
      "loss": 3.1024,
      "step": 368500
    },
    {
      "epoch": 0.2815315092247315,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012629820000000002,
      "loss": 3.177,
      "step": 368510
    },
    {
      "epoch": 0.2815501524608897,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012629619999999999,
      "loss": 3.0168,
      "step": 368520
    },
    {
      "epoch": 0.2815687956970479,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001262942,
      "loss": 3.0544,
      "step": 368530
    },
    {
      "epoch": 0.28158743893320604,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001262922,
      "loss": 3.0634,
      "step": 368540
    },
    {
      "epoch": 0.2816060821693642,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001262902,
      "loss": 3.1619,
      "step": 368550
    },
    {
      "epoch": 0.2816247254055224,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012628820000000002,
      "loss": 3.2019,
      "step": 368560
    },
    {
      "epoch": 0.2816433686416806,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001262862,
      "loss": 3.1071,
      "step": 368570
    },
    {
      "epoch": 0.2816620118778388,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001262842,
      "loss": 3.0735,
      "step": 368580
    },
    {
      "epoch": 0.281680655113997,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012628219999999998,
      "loss": 3.1064,
      "step": 368590
    },
    {
      "epoch": 0.28169929835015517,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001262802,
      "loss": 3.167,
      "step": 368600
    },
    {
      "epoch": 0.28171794158631336,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001262782,
      "loss": 3.0025,
      "step": 368610
    },
    {
      "epoch": 0.28173658482247155,
      "grad_norm": 0.75,
      "learning_rate": 0.0001262762,
      "loss": 2.9849,
      "step": 368620
    },
    {
      "epoch": 0.28175522805862974,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012627420000000002,
      "loss": 3.0484,
      "step": 368630
    },
    {
      "epoch": 0.2817738712947879,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001262722,
      "loss": 3.1817,
      "step": 368640
    },
    {
      "epoch": 0.28179251453094606,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001262702,
      "loss": 3.1192,
      "step": 368650
    },
    {
      "epoch": 0.28181115776710425,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001262682,
      "loss": 3.1335,
      "step": 368660
    },
    {
      "epoch": 0.28182980100326244,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001262662,
      "loss": 3.1239,
      "step": 368670
    },
    {
      "epoch": 0.2818484442394206,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012626420000000002,
      "loss": 3.1433,
      "step": 368680
    },
    {
      "epoch": 0.2818670874755788,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001262622,
      "loss": 3.0769,
      "step": 368690
    },
    {
      "epoch": 0.281885730711737,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012626020000000001,
      "loss": 3.137,
      "step": 368700
    },
    {
      "epoch": 0.2819043739478952,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001262582,
      "loss": 3.1971,
      "step": 368710
    },
    {
      "epoch": 0.2819230171840534,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001262562,
      "loss": 3.0273,
      "step": 368720
    },
    {
      "epoch": 0.28194166042021157,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001262542,
      "loss": 3.2088,
      "step": 368730
    },
    {
      "epoch": 0.28196030365636976,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001262522,
      "loss": 3.0988,
      "step": 368740
    },
    {
      "epoch": 0.2819789468925279,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012625020000000002,
      "loss": 3.0766,
      "step": 368750
    },
    {
      "epoch": 0.2819975901286861,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012624820000000002,
      "loss": 3.0651,
      "step": 368760
    },
    {
      "epoch": 0.28201623336484427,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001262462,
      "loss": 3.0956,
      "step": 368770
    },
    {
      "epoch": 0.28203487660100246,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001262442,
      "loss": 3.2774,
      "step": 368780
    },
    {
      "epoch": 0.28205351983716065,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001262422,
      "loss": 3.2428,
      "step": 368790
    },
    {
      "epoch": 0.28207216307331884,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001262402,
      "loss": 3.1545,
      "step": 368800
    },
    {
      "epoch": 0.282090806309477,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001262382,
      "loss": 3.0473,
      "step": 368810
    },
    {
      "epoch": 0.2821094495456352,
      "grad_norm": 0.625,
      "learning_rate": 0.00012623620000000002,
      "loss": 3.227,
      "step": 368820
    },
    {
      "epoch": 0.2821280927817934,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001262342,
      "loss": 3.3281,
      "step": 368830
    },
    {
      "epoch": 0.2821467360179516,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001262322,
      "loss": 3.1394,
      "step": 368840
    },
    {
      "epoch": 0.2821653792541098,
      "grad_norm": 0.75,
      "learning_rate": 0.0001262302,
      "loss": 3.1759,
      "step": 368850
    },
    {
      "epoch": 0.2821840224902679,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001262282,
      "loss": 3.1797,
      "step": 368860
    },
    {
      "epoch": 0.2822026657264261,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001262262,
      "loss": 3.105,
      "step": 368870
    },
    {
      "epoch": 0.2822213089625843,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012622420000000002,
      "loss": 3.1667,
      "step": 368880
    },
    {
      "epoch": 0.2822399521987425,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001262222,
      "loss": 3.097,
      "step": 368890
    },
    {
      "epoch": 0.28225859543490067,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001262202,
      "loss": 3.0467,
      "step": 368900
    },
    {
      "epoch": 0.28227723867105886,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001262182,
      "loss": 3.0793,
      "step": 368910
    },
    {
      "epoch": 0.28229588190721705,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001262162,
      "loss": 3.1082,
      "step": 368920
    },
    {
      "epoch": 0.28231452514337524,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012621420000000002,
      "loss": 3.0229,
      "step": 368930
    },
    {
      "epoch": 0.28233316837953343,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001262122,
      "loss": 3.1549,
      "step": 368940
    },
    {
      "epoch": 0.2823518116156916,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001262102,
      "loss": 3.1461,
      "step": 368950
    },
    {
      "epoch": 0.28237045485184975,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001262082,
      "loss": 3.1518,
      "step": 368960
    },
    {
      "epoch": 0.28238909808800794,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001262062,
      "loss": 3.0541,
      "step": 368970
    },
    {
      "epoch": 0.28240774132416613,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012620420000000003,
      "loss": 3.1227,
      "step": 368980
    },
    {
      "epoch": 0.2824263845603243,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001262022,
      "loss": 3.079,
      "step": 368990
    },
    {
      "epoch": 0.2824450277964825,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012620020000000002,
      "loss": 3.2445,
      "step": 369000
    },
    {
      "epoch": 0.2824636710326407,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001261982,
      "loss": 3.0999,
      "step": 369010
    },
    {
      "epoch": 0.2824823142687989,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001261962,
      "loss": 3.0565,
      "step": 369020
    },
    {
      "epoch": 0.2825009575049571,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001261942,
      "loss": 3.034,
      "step": 369030
    },
    {
      "epoch": 0.28251960074111526,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001261922,
      "loss": 3.2249,
      "step": 369040
    },
    {
      "epoch": 0.28253824397727345,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012619020000000002,
      "loss": 3.1316,
      "step": 369050
    },
    {
      "epoch": 0.28255688721343164,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001261882,
      "loss": 3.1625,
      "step": 369060
    },
    {
      "epoch": 0.2825755304495898,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012618620000000002,
      "loss": 3.1307,
      "step": 369070
    },
    {
      "epoch": 0.28259417368574796,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012618419999999998,
      "loss": 3.0252,
      "step": 369080
    },
    {
      "epoch": 0.28261281692190615,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001261822,
      "loss": 2.9736,
      "step": 369090
    },
    {
      "epoch": 0.28263146015806434,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001261802,
      "loss": 3.1953,
      "step": 369100
    },
    {
      "epoch": 0.28265010339422253,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001261782,
      "loss": 3.1704,
      "step": 369110
    },
    {
      "epoch": 0.2826687466303807,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012617620000000002,
      "loss": 3.2306,
      "step": 369120
    },
    {
      "epoch": 0.2826873898665389,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001261742,
      "loss": 3.1113,
      "step": 369130
    },
    {
      "epoch": 0.2827060331026971,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001261722,
      "loss": 3.0511,
      "step": 369140
    },
    {
      "epoch": 0.2827246763388553,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001261702,
      "loss": 3.1398,
      "step": 369150
    },
    {
      "epoch": 0.2827433195750135,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001261682,
      "loss": 3.21,
      "step": 369160
    },
    {
      "epoch": 0.2827619628111716,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012616620000000003,
      "loss": 3.2212,
      "step": 369170
    },
    {
      "epoch": 0.2827806060473298,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001261642,
      "loss": 3.1136,
      "step": 369180
    },
    {
      "epoch": 0.282799249283488,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012616220000000002,
      "loss": 3.1575,
      "step": 369190
    },
    {
      "epoch": 0.2828178925196462,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012616019999999999,
      "loss": 3.0714,
      "step": 369200
    },
    {
      "epoch": 0.28283653575580436,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001261582,
      "loss": 3.1419,
      "step": 369210
    },
    {
      "epoch": 0.28285517899196255,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001261562,
      "loss": 3.1932,
      "step": 369220
    },
    {
      "epoch": 0.28287382222812074,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001261542,
      "loss": 3.1311,
      "step": 369230
    },
    {
      "epoch": 0.28289246546427893,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012615220000000002,
      "loss": 3.2035,
      "step": 369240
    },
    {
      "epoch": 0.2829111087004371,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001261502,
      "loss": 2.9883,
      "step": 369250
    },
    {
      "epoch": 0.2829297519365953,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001261482,
      "loss": 3.0769,
      "step": 369260
    },
    {
      "epoch": 0.2829483951727535,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001261462,
      "loss": 3.1532,
      "step": 369270
    },
    {
      "epoch": 0.28296703840891163,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001261442,
      "loss": 3.125,
      "step": 369280
    },
    {
      "epoch": 0.2829856816450698,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001261422,
      "loss": 3.199,
      "step": 369290
    },
    {
      "epoch": 0.283004324881228,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001261402,
      "loss": 3.1249,
      "step": 369300
    },
    {
      "epoch": 0.2830229681173862,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012613820000000002,
      "loss": 3.1194,
      "step": 369310
    },
    {
      "epoch": 0.2830416113535444,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012613620000000001,
      "loss": 3.0836,
      "step": 369320
    },
    {
      "epoch": 0.2830602545897026,
      "grad_norm": 0.75,
      "learning_rate": 0.0001261342,
      "loss": 3.2448,
      "step": 369330
    },
    {
      "epoch": 0.28307889782586076,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001261322,
      "loss": 3.0538,
      "step": 369340
    },
    {
      "epoch": 0.28309754106201895,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001261302,
      "loss": 3.096,
      "step": 369350
    },
    {
      "epoch": 0.28311618429817714,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001261282,
      "loss": 3.1137,
      "step": 369360
    },
    {
      "epoch": 0.28313482753433533,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012612620000000002,
      "loss": 3.1406,
      "step": 369370
    },
    {
      "epoch": 0.28315347077049347,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012612420000000001,
      "loss": 3.0839,
      "step": 369380
    },
    {
      "epoch": 0.28317211400665165,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001261222,
      "loss": 3.1501,
      "step": 369390
    },
    {
      "epoch": 0.28319075724280984,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001261202,
      "loss": 3.2013,
      "step": 369400
    },
    {
      "epoch": 0.28320940047896803,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001261182,
      "loss": 3.1403,
      "step": 369410
    },
    {
      "epoch": 0.2832280437151262,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001261162,
      "loss": 3.1116,
      "step": 369420
    },
    {
      "epoch": 0.2832466869512844,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001261142,
      "loss": 3.051,
      "step": 369430
    },
    {
      "epoch": 0.2832653301874426,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012611220000000002,
      "loss": 3.1118,
      "step": 369440
    },
    {
      "epoch": 0.2832839734236008,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001261102,
      "loss": 3.1884,
      "step": 369450
    },
    {
      "epoch": 0.283302616659759,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001261082,
      "loss": 3.2043,
      "step": 369460
    },
    {
      "epoch": 0.28332125989591717,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001261062,
      "loss": 3.1149,
      "step": 369470
    },
    {
      "epoch": 0.2833399031320753,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001261042,
      "loss": 3.0428,
      "step": 369480
    },
    {
      "epoch": 0.2833585463682335,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012610220000000002,
      "loss": 3.0844,
      "step": 369490
    },
    {
      "epoch": 0.2833771896043917,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001261002,
      "loss": 3.1871,
      "step": 369500
    },
    {
      "epoch": 0.28339583284054987,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001260982,
      "loss": 3.2328,
      "step": 369510
    },
    {
      "epoch": 0.28341447607670806,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001260962,
      "loss": 3.1597,
      "step": 369520
    },
    {
      "epoch": 0.28343311931286624,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001260942,
      "loss": 3.1516,
      "step": 369530
    },
    {
      "epoch": 0.28345176254902443,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012609220000000003,
      "loss": 3.2714,
      "step": 369540
    },
    {
      "epoch": 0.2834704057851826,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001260902,
      "loss": 3.1246,
      "step": 369550
    },
    {
      "epoch": 0.2834890490213408,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012608820000000002,
      "loss": 3.2288,
      "step": 369560
    },
    {
      "epoch": 0.283507692257499,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012608619999999999,
      "loss": 3.1484,
      "step": 369570
    },
    {
      "epoch": 0.2835263354936572,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001260842,
      "loss": 3.1872,
      "step": 369580
    },
    {
      "epoch": 0.2835449787298153,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001260822,
      "loss": 3.0805,
      "step": 369590
    },
    {
      "epoch": 0.2835636219659735,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001260802,
      "loss": 2.9733,
      "step": 369600
    },
    {
      "epoch": 0.2835822652021317,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012607820000000002,
      "loss": 3.1711,
      "step": 369610
    },
    {
      "epoch": 0.2836009084382899,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001260762,
      "loss": 3.2299,
      "step": 369620
    },
    {
      "epoch": 0.2836195516744481,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001260742,
      "loss": 3.1063,
      "step": 369630
    },
    {
      "epoch": 0.28363819491060627,
      "grad_norm": 0.75,
      "learning_rate": 0.0001260722,
      "loss": 3.1149,
      "step": 369640
    },
    {
      "epoch": 0.28365683814676446,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001260702,
      "loss": 3.1177,
      "step": 369650
    },
    {
      "epoch": 0.28367548138292265,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012606820000000003,
      "loss": 3.2653,
      "step": 369660
    },
    {
      "epoch": 0.28369412461908083,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001260662,
      "loss": 3.0388,
      "step": 369670
    },
    {
      "epoch": 0.283712767855239,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012606420000000002,
      "loss": 3.044,
      "step": 369680
    },
    {
      "epoch": 0.28373141109139716,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001260622,
      "loss": 3.1629,
      "step": 369690
    },
    {
      "epoch": 0.28375005432755535,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001260602,
      "loss": 3.1852,
      "step": 369700
    },
    {
      "epoch": 0.28376869756371353,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001260582,
      "loss": 3.1598,
      "step": 369710
    },
    {
      "epoch": 0.2837873407998717,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001260562,
      "loss": 3.1801,
      "step": 369720
    },
    {
      "epoch": 0.2838059840360299,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012605420000000002,
      "loss": 3.1978,
      "step": 369730
    },
    {
      "epoch": 0.2838246272721881,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001260522,
      "loss": 3.2283,
      "step": 369740
    },
    {
      "epoch": 0.2838432705083463,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012605020000000002,
      "loss": 3.0271,
      "step": 369750
    },
    {
      "epoch": 0.2838619137445045,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001260482,
      "loss": 3.0172,
      "step": 369760
    },
    {
      "epoch": 0.28388055698066267,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001260462,
      "loss": 3.0357,
      "step": 369770
    },
    {
      "epoch": 0.28389920021682086,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001260442,
      "loss": 3.055,
      "step": 369780
    },
    {
      "epoch": 0.28391784345297905,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001260422,
      "loss": 3.2395,
      "step": 369790
    },
    {
      "epoch": 0.2839364866891372,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012604020000000002,
      "loss": 3.1081,
      "step": 369800
    },
    {
      "epoch": 0.28395512992529537,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001260382,
      "loss": 3.102,
      "step": 369810
    },
    {
      "epoch": 0.28397377316145356,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001260362,
      "loss": 3.0793,
      "step": 369820
    },
    {
      "epoch": 0.28399241639761175,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001260342,
      "loss": 3.0819,
      "step": 369830
    },
    {
      "epoch": 0.28401105963376994,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001260322,
      "loss": 3.1584,
      "step": 369840
    },
    {
      "epoch": 0.2840297028699281,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001260302,
      "loss": 3.2051,
      "step": 369850
    },
    {
      "epoch": 0.2840483461060863,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001260282,
      "loss": 3.0858,
      "step": 369860
    },
    {
      "epoch": 0.2840669893422445,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012602620000000002,
      "loss": 3.1063,
      "step": 369870
    },
    {
      "epoch": 0.2840856325784027,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001260242,
      "loss": 3.1266,
      "step": 369880
    },
    {
      "epoch": 0.2841042758145609,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001260222,
      "loss": 3.1199,
      "step": 369890
    },
    {
      "epoch": 0.284122919050719,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001260202,
      "loss": 3.1595,
      "step": 369900
    },
    {
      "epoch": 0.2841415622868772,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001260182,
      "loss": 3.1344,
      "step": 369910
    },
    {
      "epoch": 0.2841602055230354,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001260162,
      "loss": 3.1651,
      "step": 369920
    },
    {
      "epoch": 0.2841788487591936,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012601420000000002,
      "loss": 3.2479,
      "step": 369930
    },
    {
      "epoch": 0.28419749199535177,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001260122,
      "loss": 3.1471,
      "step": 369940
    },
    {
      "epoch": 0.28421613523150996,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001260102,
      "loss": 3.1868,
      "step": 369950
    },
    {
      "epoch": 0.28423477846766815,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001260082,
      "loss": 3.0833,
      "step": 369960
    },
    {
      "epoch": 0.28425342170382634,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001260062,
      "loss": 3.1363,
      "step": 369970
    },
    {
      "epoch": 0.2842720649399845,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012600420000000002,
      "loss": 3.1649,
      "step": 369980
    },
    {
      "epoch": 0.2842907081761427,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001260022,
      "loss": 3.1246,
      "step": 369990
    },
    {
      "epoch": 0.2843093514123009,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012600020000000001,
      "loss": 3.0979,
      "step": 370000
    },
    {
      "epoch": 0.28432799464845904,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001259982,
      "loss": 3.1069,
      "step": 370010
    },
    {
      "epoch": 0.2843466378846172,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001259962,
      "loss": 3.1817,
      "step": 370020
    },
    {
      "epoch": 0.2843652811207754,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001259942,
      "loss": 3.148,
      "step": 370030
    },
    {
      "epoch": 0.2843839243569336,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001259922,
      "loss": 3.1163,
      "step": 370040
    },
    {
      "epoch": 0.2844025675930918,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012599020000000002,
      "loss": 3.0852,
      "step": 370050
    },
    {
      "epoch": 0.28442121082925,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001259882,
      "loss": 3.0792,
      "step": 370060
    },
    {
      "epoch": 0.28443985406540817,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001259862,
      "loss": 2.9924,
      "step": 370070
    },
    {
      "epoch": 0.28445849730156636,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001259842,
      "loss": 3.1556,
      "step": 370080
    },
    {
      "epoch": 0.28447714053772455,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001259822,
      "loss": 3.172,
      "step": 370090
    },
    {
      "epoch": 0.28449578377388274,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012598020000000002,
      "loss": 3.1442,
      "step": 370100
    },
    {
      "epoch": 0.28451442701004087,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001259782,
      "loss": 3.0655,
      "step": 370110
    },
    {
      "epoch": 0.28453307024619906,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012597620000000002,
      "loss": 3.1304,
      "step": 370120
    },
    {
      "epoch": 0.28455171348235725,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001259742,
      "loss": 3.1155,
      "step": 370130
    },
    {
      "epoch": 0.28457035671851544,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001259722,
      "loss": 3.1128,
      "step": 370140
    },
    {
      "epoch": 0.2845889999546736,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012597020000000003,
      "loss": 3.1375,
      "step": 370150
    },
    {
      "epoch": 0.2846076431908318,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001259682,
      "loss": 3.2239,
      "step": 370160
    },
    {
      "epoch": 0.28462628642699,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012596620000000002,
      "loss": 3.0864,
      "step": 370170
    },
    {
      "epoch": 0.2846449296631482,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001259642,
      "loss": 3.174,
      "step": 370180
    },
    {
      "epoch": 0.2846635728993064,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001259622,
      "loss": 3.0867,
      "step": 370190
    },
    {
      "epoch": 0.28468221613546457,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001259602,
      "loss": 3.1171,
      "step": 370200
    },
    {
      "epoch": 0.28470085937162276,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001259582,
      "loss": 3.2455,
      "step": 370210
    },
    {
      "epoch": 0.2847195026077809,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012595620000000003,
      "loss": 3.0129,
      "step": 370220
    },
    {
      "epoch": 0.2847381458439391,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001259542,
      "loss": 3.1296,
      "step": 370230
    },
    {
      "epoch": 0.2847567890800973,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012595220000000002,
      "loss": 3.1387,
      "step": 370240
    },
    {
      "epoch": 0.28477543231625546,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012595019999999999,
      "loss": 3.1162,
      "step": 370250
    },
    {
      "epoch": 0.28479407555241365,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001259482,
      "loss": 3.1317,
      "step": 370260
    },
    {
      "epoch": 0.28481271878857184,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001259462,
      "loss": 3.1606,
      "step": 370270
    },
    {
      "epoch": 0.28483136202473003,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001259442,
      "loss": 3.1268,
      "step": 370280
    },
    {
      "epoch": 0.2848500052608882,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012594220000000002,
      "loss": 3.0177,
      "step": 370290
    },
    {
      "epoch": 0.2848686484970464,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001259402,
      "loss": 3.0442,
      "step": 370300
    },
    {
      "epoch": 0.2848872917332046,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001259382,
      "loss": 3.0985,
      "step": 370310
    },
    {
      "epoch": 0.28490593496936273,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001259362,
      "loss": 3.1432,
      "step": 370320
    },
    {
      "epoch": 0.2849245782055209,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001259342,
      "loss": 3.0885,
      "step": 370330
    },
    {
      "epoch": 0.2849432214416791,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001259322,
      "loss": 3.009,
      "step": 370340
    },
    {
      "epoch": 0.2849618646778373,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001259302,
      "loss": 3.1377,
      "step": 370350
    },
    {
      "epoch": 0.2849805079139955,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012592820000000002,
      "loss": 3.0078,
      "step": 370360
    },
    {
      "epoch": 0.2849991511501537,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012592620000000001,
      "loss": 3.0293,
      "step": 370370
    },
    {
      "epoch": 0.28501779438631186,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001259242,
      "loss": 3.0353,
      "step": 370380
    },
    {
      "epoch": 0.28503643762247005,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001259222,
      "loss": 3.1634,
      "step": 370390
    },
    {
      "epoch": 0.28505508085862824,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001259202,
      "loss": 3.1466,
      "step": 370400
    },
    {
      "epoch": 0.28507372409478643,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001259182,
      "loss": 3.2726,
      "step": 370410
    },
    {
      "epoch": 0.2850923673309446,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001259162,
      "loss": 3.021,
      "step": 370420
    },
    {
      "epoch": 0.28511101056710275,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012591420000000001,
      "loss": 3.0409,
      "step": 370430
    },
    {
      "epoch": 0.28512965380326094,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001259122,
      "loss": 2.9516,
      "step": 370440
    },
    {
      "epoch": 0.28514829703941913,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001259102,
      "loss": 3.0693,
      "step": 370450
    },
    {
      "epoch": 0.2851669402755773,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001259082,
      "loss": 3.0823,
      "step": 370460
    },
    {
      "epoch": 0.2851855835117355,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001259062,
      "loss": 3.0744,
      "step": 370470
    },
    {
      "epoch": 0.2852042267478937,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001259042,
      "loss": 3.1338,
      "step": 370480
    },
    {
      "epoch": 0.2852228699840519,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012590220000000002,
      "loss": 3.1294,
      "step": 370490
    },
    {
      "epoch": 0.2852415132202101,
      "grad_norm": 0.75,
      "learning_rate": 0.0001259002,
      "loss": 3.0628,
      "step": 370500
    },
    {
      "epoch": 1.8643236158183664e-05,
      "grad_norm": 0.75,
      "learning_rate": 0.00012591079107910792,
      "loss": 3.135,
      "step": 370510
    },
    {
      "epoch": 3.728647231636733e-05,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001259087908790879,
      "loss": 3.1809,
      "step": 370520
    },
    {
      "epoch": 5.592970847455099e-05,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001259067906790679,
      "loss": 3.2657,
      "step": 370530
    },
    {
      "epoch": 7.457294463273465e-05,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012590479047904792,
      "loss": 3.0789,
      "step": 370540
    },
    {
      "epoch": 9.321618079091832e-05,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001259027902790279,
      "loss": 3.1528,
      "step": 370550
    },
    {
      "epoch": 0.00011185941694910198,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001259007900790079,
      "loss": 3.2107,
      "step": 370560
    },
    {
      "epoch": 0.00013050265310728565,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012589878987898792,
      "loss": 3.2072,
      "step": 370570
    },
    {
      "epoch": 0.0001491458892654693,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001258967896789679,
      "loss": 3.1584,
      "step": 370580
    },
    {
      "epoch": 0.00016778912542365297,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001258947894789479,
      "loss": 3.1765,
      "step": 370590
    },
    {
      "epoch": 0.00018643236158183665,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001258927892789279,
      "loss": 3.2329,
      "step": 370600
    },
    {
      "epoch": 0.0002050755977400203,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001258907890789079,
      "loss": 3.249,
      "step": 370610
    },
    {
      "epoch": 0.00022371883389820396,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012588878887888788,
      "loss": 3.1737,
      "step": 370620
    },
    {
      "epoch": 0.00024236207005638762,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001258867886788679,
      "loss": 3.2238,
      "step": 370630
    },
    {
      "epoch": 0.0002610053062145713,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012588478847884788,
      "loss": 3.2679,
      "step": 370640
    },
    {
      "epoch": 0.00027964854237275496,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012588278827882789,
      "loss": 3.2335,
      "step": 370650
    },
    {
      "epoch": 0.0002982917785309386,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012588078807880787,
      "loss": 3.2078,
      "step": 370660
    },
    {
      "epoch": 0.0003169350146891223,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012587878787878788,
      "loss": 3.1736,
      "step": 370670
    },
    {
      "epoch": 0.00033557825084730593,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012587678767876789,
      "loss": 3.1184,
      "step": 370680
    },
    {
      "epoch": 0.00035422148700548964,
      "grad_norm": 0.859375,
      "learning_rate": 0.00012587478747874787,
      "loss": 3.1428,
      "step": 370690
    },
    {
      "epoch": 0.0003728647231636733,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00012587278727872788,
      "loss": 3.1612,
      "step": 370700
    },
    {
      "epoch": 0.00039150795932185696,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012587078707870789,
      "loss": 3.1319,
      "step": 370710
    },
    {
      "epoch": 0.0004101511954800406,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001258687868786879,
      "loss": 3.3028,
      "step": 370720
    },
    {
      "epoch": 0.00042879443163822427,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00012586678667866788,
      "loss": 3.1912,
      "step": 370730
    },
    {
      "epoch": 0.0004474376677964079,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012586478647864789,
      "loss": 3.2163,
      "step": 370740
    },
    {
      "epoch": 0.0004660809039545916,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012586278627862787,
      "loss": 3.1991,
      "step": 370750
    },
    {
      "epoch": 0.00048472414011277524,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012586078607860788,
      "loss": 3.2029,
      "step": 370760
    },
    {
      "epoch": 0.000503367376270959,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012585878587858786,
      "loss": 3.1871,
      "step": 370770
    },
    {
      "epoch": 0.0005220106124291426,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012585678567856787,
      "loss": 3.199,
      "step": 370780
    },
    {
      "epoch": 0.0005406538485873263,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012585478547854785,
      "loss": 3.3132,
      "step": 370790
    },
    {
      "epoch": 0.0005592970847455099,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012585278527852786,
      "loss": 3.307,
      "step": 370800
    },
    {
      "epoch": 0.0005779403209036936,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012585078507850784,
      "loss": 3.2101,
      "step": 370810
    },
    {
      "epoch": 0.0005965835570618772,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012584878487848785,
      "loss": 3.2708,
      "step": 370820
    },
    {
      "epoch": 0.0006152267932200609,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012584678467846783,
      "loss": 3.0273,
      "step": 370830
    },
    {
      "epoch": 0.0006338700293782445,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012584478447844784,
      "loss": 3.2191,
      "step": 370840
    },
    {
      "epoch": 0.0006525132655364282,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012584278427842785,
      "loss": 3.2161,
      "step": 370850
    },
    {
      "epoch": 0.0006711565016946119,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012584078407840786,
      "loss": 3.2823,
      "step": 370860
    },
    {
      "epoch": 0.0006897997378527955,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012583878387838784,
      "loss": 3.1971,
      "step": 370870
    },
    {
      "epoch": 0.0007084429740109793,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012583678367836785,
      "loss": 3.2413,
      "step": 370880
    },
    {
      "epoch": 0.0007270862101691629,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012583478347834786,
      "loss": 3.342,
      "step": 370890
    },
    {
      "epoch": 0.0007457294463273466,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012583278327832784,
      "loss": 3.2172,
      "step": 370900
    },
    {
      "epoch": 0.0007643726824855303,
      "grad_norm": 0.75,
      "learning_rate": 0.00012583078307830785,
      "loss": 3.315,
      "step": 370910
    },
    {
      "epoch": 0.0007830159186437139,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012582878287828783,
      "loss": 3.3901,
      "step": 370920
    },
    {
      "epoch": 0.0008016591548018976,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012582678267826784,
      "loss": 3.1692,
      "step": 370930
    },
    {
      "epoch": 0.0008203023909600812,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012582478247824783,
      "loss": 3.245,
      "step": 370940
    },
    {
      "epoch": 0.0008389456271182649,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012582278227822783,
      "loss": 3.2778,
      "step": 370950
    },
    {
      "epoch": 0.0008575888632764485,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012582078207820782,
      "loss": 3.3641,
      "step": 370960
    },
    {
      "epoch": 0.0008762320994346322,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012581878187818783,
      "loss": 3.1831,
      "step": 370970
    },
    {
      "epoch": 0.0008948753355928159,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001258167816781678,
      "loss": 3.1901,
      "step": 370980
    },
    {
      "epoch": 0.0009135185717509995,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012581478147814782,
      "loss": 3.2013,
      "step": 370990
    },
    {
      "epoch": 0.0009321618079091832,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001258127812781278,
      "loss": 3.0274,
      "step": 371000
    },
    {
      "epoch": 0.0009508050440673668,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001258107810781078,
      "loss": 3.1826,
      "step": 371010
    },
    {
      "epoch": 0.0009694482802255505,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012580878087808782,
      "loss": 3.1895,
      "step": 371020
    },
    {
      "epoch": 0.0009880915163837341,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012580678067806783,
      "loss": 3.2167,
      "step": 371030
    },
    {
      "epoch": 0.001006734752541918,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001258047804780478,
      "loss": 3.1804,
      "step": 371040
    },
    {
      "epoch": 0.0010253779887001014,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012580278027802782,
      "loss": 3.2875,
      "step": 371050
    },
    {
      "epoch": 0.0010440212248582852,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001258007800780078,
      "loss": 3.2163,
      "step": 371060
    },
    {
      "epoch": 0.0010626644610164688,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001257987798779878,
      "loss": 3.1785,
      "step": 371070
    },
    {
      "epoch": 0.0010813076971746525,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012579677967796782,
      "loss": 3.2261,
      "step": 371080
    },
    {
      "epoch": 0.001099950933332836,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001257947794779478,
      "loss": 3.2145,
      "step": 371090
    },
    {
      "epoch": 0.0011185941694910198,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001257927792779278,
      "loss": 3.2402,
      "step": 371100
    },
    {
      "epoch": 0.0011372374056492036,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001257907790779078,
      "loss": 3.2919,
      "step": 371110
    },
    {
      "epoch": 0.0011558806418073872,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001257887788778878,
      "loss": 3.1606,
      "step": 371120
    },
    {
      "epoch": 0.001174523877965571,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012578677867786778,
      "loss": 3.1418,
      "step": 371130
    },
    {
      "epoch": 0.0011931671141237545,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001257847784778478,
      "loss": 3.2524,
      "step": 371140
    },
    {
      "epoch": 0.0012118103502819382,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012578277827782777,
      "loss": 3.1448,
      "step": 371150
    },
    {
      "epoch": 0.0012304535864401218,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012578077807780778,
      "loss": 3.1693,
      "step": 371160
    },
    {
      "epoch": 0.0012490968225983055,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001257787778777878,
      "loss": 3.192,
      "step": 371170
    },
    {
      "epoch": 0.001267740058756489,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001257767776777678,
      "loss": 3.1816,
      "step": 371180
    },
    {
      "epoch": 0.0012863832949146729,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012577477747774778,
      "loss": 3.1418,
      "step": 371190
    },
    {
      "epoch": 0.0013050265310728564,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001257727772777278,
      "loss": 3.1946,
      "step": 371200
    },
    {
      "epoch": 0.0013236697672310402,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012577077707770777,
      "loss": 3.1691,
      "step": 371210
    },
    {
      "epoch": 0.0013423130033892237,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012576877687768778,
      "loss": 3.4189,
      "step": 371220
    },
    {
      "epoch": 0.0013609562395474075,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012576677667766777,
      "loss": 3.2254,
      "step": 371230
    },
    {
      "epoch": 0.001379599475705591,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012576477647764777,
      "loss": 3.3077,
      "step": 371240
    },
    {
      "epoch": 0.0013982427118637748,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012576277627762776,
      "loss": 3.3319,
      "step": 371250
    },
    {
      "epoch": 0.0014168859480219586,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012576077607760777,
      "loss": 3.2349,
      "step": 371260
    },
    {
      "epoch": 0.0014355291841801421,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012575877587758777,
      "loss": 3.2848,
      "step": 371270
    },
    {
      "epoch": 0.0014541724203383259,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012575677567756776,
      "loss": 3.193,
      "step": 371280
    },
    {
      "epoch": 0.0014728156564965094,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012575477547754777,
      "loss": 3.1808,
      "step": 371290
    },
    {
      "epoch": 0.0014914588926546932,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012575277527752775,
      "loss": 3.1244,
      "step": 371300
    },
    {
      "epoch": 0.0015101021288128767,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012575077507750776,
      "loss": 3.0878,
      "step": 371310
    },
    {
      "epoch": 0.0015287453649710605,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012574877487748774,
      "loss": 3.1419,
      "step": 371320
    },
    {
      "epoch": 0.001547388601129244,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012574677467746775,
      "loss": 3.3422,
      "step": 371330
    },
    {
      "epoch": 0.0015660318372874278,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012574477447744776,
      "loss": 3.1663,
      "step": 371340
    },
    {
      "epoch": 0.0015846750734456114,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012574277427742777,
      "loss": 3.15,
      "step": 371350
    },
    {
      "epoch": 0.0016033183096037951,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012574077407740775,
      "loss": 3.2536,
      "step": 371360
    },
    {
      "epoch": 0.0016219615457619787,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012573877387738776,
      "loss": 3.2082,
      "step": 371370
    },
    {
      "epoch": 0.0016406047819201624,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012573677367736774,
      "loss": 3.1624,
      "step": 371380
    },
    {
      "epoch": 0.001659248018078346,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012573477347734775,
      "loss": 3.1726,
      "step": 371390
    },
    {
      "epoch": 0.0016778912542365298,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012573277327732773,
      "loss": 3.0704,
      "step": 371400
    },
    {
      "epoch": 0.0016965344903947135,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012573077307730774,
      "loss": 3.0457,
      "step": 371410
    },
    {
      "epoch": 0.001715177726552897,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012572877287728772,
      "loss": 3.2805,
      "step": 371420
    },
    {
      "epoch": 0.0017338209627110808,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012572677267726773,
      "loss": 3.1717,
      "step": 371430
    },
    {
      "epoch": 0.0017524641988692644,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001257247724772477,
      "loss": 3.2016,
      "step": 371440
    },
    {
      "epoch": 0.0017711074350274482,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012572277227722772,
      "loss": 3.2792,
      "step": 371450
    },
    {
      "epoch": 0.0017897506711856317,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001257207720772077,
      "loss": 3.281,
      "step": 371460
    },
    {
      "epoch": 0.0018083939073438155,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012571877187718771,
      "loss": 3.2757,
      "step": 371470
    },
    {
      "epoch": 0.001827037143501999,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012571677167716772,
      "loss": 3.0925,
      "step": 371480
    },
    {
      "epoch": 0.0018456803796601828,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012571477147714773,
      "loss": 3.2594,
      "step": 371490
    },
    {
      "epoch": 0.0018643236158183663,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00012571277127712774,
      "loss": 3.1623,
      "step": 371500
    },
    {
      "epoch": 0.00188296685197655,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012571077107710772,
      "loss": 3.1902,
      "step": 371510
    },
    {
      "epoch": 0.0019016100881347336,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012570877087708773,
      "loss": 3.2685,
      "step": 371520
    },
    {
      "epoch": 0.0019202533242929174,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012570677067706771,
      "loss": 3.181,
      "step": 371530
    },
    {
      "epoch": 0.001938896560451101,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012570477047704772,
      "loss": 3.1482,
      "step": 371540
    },
    {
      "epoch": 0.001957539796609285,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001257027702770277,
      "loss": 3.1146,
      "step": 371550
    },
    {
      "epoch": 0.0019761830327674683,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00012570077007700771,
      "loss": 3.1959,
      "step": 371560
    },
    {
      "epoch": 0.001994826268925652,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001256987698769877,
      "loss": 3.2824,
      "step": 371570
    },
    {
      "epoch": 0.002013469505083836,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001256967696769677,
      "loss": 3.1957,
      "step": 371580
    },
    {
      "epoch": 0.0020321127412420196,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001256947694769477,
      "loss": 3.388,
      "step": 371590
    },
    {
      "epoch": 0.002050755977400203,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001256927692769277,
      "loss": 3.1811,
      "step": 371600
    },
    {
      "epoch": 0.0020693992135583867,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012569076907690768,
      "loss": 3.176,
      "step": 371610
    },
    {
      "epoch": 0.0020880424497165704,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001256887688768877,
      "loss": 3.3043,
      "step": 371620
    },
    {
      "epoch": 0.002106685685874754,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00012568676867686767,
      "loss": 3.1108,
      "step": 371630
    },
    {
      "epoch": 0.0021253289220329375,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012568476847684768,
      "loss": 3.1938,
      "step": 371640
    },
    {
      "epoch": 0.0021439721581911213,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001256827682768277,
      "loss": 3.2913,
      "step": 371650
    },
    {
      "epoch": 0.002162615394349305,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001256807680768077,
      "loss": 3.2086,
      "step": 371660
    },
    {
      "epoch": 0.002181258630507489,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001256787678767877,
      "loss": 3.2335,
      "step": 371670
    },
    {
      "epoch": 0.002199901866665672,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001256767676767677,
      "loss": 3.2708,
      "step": 371680
    },
    {
      "epoch": 0.002218545102823856,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001256747674767477,
      "loss": 3.1987,
      "step": 371690
    },
    {
      "epoch": 0.0022371883389820397,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012567276727672768,
      "loss": 3.2792,
      "step": 371700
    },
    {
      "epoch": 0.0022558315751402234,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001256707670767077,
      "loss": 3.3215,
      "step": 371710
    },
    {
      "epoch": 0.002274474811298407,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012566876687668767,
      "loss": 3.1233,
      "step": 371720
    },
    {
      "epoch": 0.0022931180474565905,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012566676667666768,
      "loss": 3.2402,
      "step": 371730
    },
    {
      "epoch": 0.0023117612836147743,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012566476647664766,
      "loss": 3.2313,
      "step": 371740
    },
    {
      "epoch": 0.002330404519772958,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012566276627662767,
      "loss": 3.2233,
      "step": 371750
    },
    {
      "epoch": 0.002349047755931142,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012566076607660765,
      "loss": 3.1952,
      "step": 371760
    },
    {
      "epoch": 0.002367690992089325,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012565876587658766,
      "loss": 3.1155,
      "step": 371770
    },
    {
      "epoch": 0.002386334228247509,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012565676567656764,
      "loss": 3.2117,
      "step": 371780
    },
    {
      "epoch": 0.0024049774644056927,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012565476547654765,
      "loss": 3.1675,
      "step": 371790
    },
    {
      "epoch": 0.0024236207005638765,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012565276527652766,
      "loss": 3.1938,
      "step": 371800
    },
    {
      "epoch": 0.00244226393672206,
      "grad_norm": 0.75,
      "learning_rate": 0.00012565076507650767,
      "loss": 3.2239,
      "step": 371810
    },
    {
      "epoch": 0.0024609071728802436,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012564876487648765,
      "loss": 3.2565,
      "step": 371820
    },
    {
      "epoch": 0.0024795504090384273,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012564676467646766,
      "loss": 3.3143,
      "step": 371830
    },
    {
      "epoch": 0.002498193645196611,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012564476447644765,
      "loss": 3.1454,
      "step": 371840
    },
    {
      "epoch": 0.002516836881354795,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012564276427642765,
      "loss": 3.2925,
      "step": 371850
    },
    {
      "epoch": 0.002535480117512978,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012564076407640766,
      "loss": 3.3369,
      "step": 371860
    },
    {
      "epoch": 0.002554123353671162,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012563876387638765,
      "loss": 3.2875,
      "step": 371870
    },
    {
      "epoch": 0.0025727665898293457,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012563676367636766,
      "loss": 3.2458,
      "step": 371880
    },
    {
      "epoch": 0.0025914098259875295,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012563476347634764,
      "loss": 3.1434,
      "step": 371890
    },
    {
      "epoch": 0.002610053062145713,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012563276327632765,
      "loss": 3.1565,
      "step": 371900
    },
    {
      "epoch": 0.0026286962983038966,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012563076307630763,
      "loss": 3.1059,
      "step": 371910
    },
    {
      "epoch": 0.0026473395344620804,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012562876287628764,
      "loss": 3.2696,
      "step": 371920
    },
    {
      "epoch": 0.002665982770620264,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012562676267626762,
      "loss": 3.2358,
      "step": 371930
    },
    {
      "epoch": 0.0026846260067784474,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012562476247624763,
      "loss": 3.1927,
      "step": 371940
    },
    {
      "epoch": 0.002703269242936631,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001256227622762276,
      "loss": 3.2752,
      "step": 371950
    },
    {
      "epoch": 0.002721912479094815,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012562076207620762,
      "loss": 3.2981,
      "step": 371960
    },
    {
      "epoch": 0.0027405557152529987,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012561876187618763,
      "loss": 3.186,
      "step": 371970
    },
    {
      "epoch": 0.002759198951411182,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012561676167616764,
      "loss": 3.3003,
      "step": 371980
    },
    {
      "epoch": 0.002777842187569366,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012561476147614762,
      "loss": 3.2379,
      "step": 371990
    },
    {
      "epoch": 0.0027964854237275496,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012561276127612763,
      "loss": 3.1212,
      "step": 372000
    },
    {
      "epoch": 0.0028151286598857334,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001256107610761076,
      "loss": 3.1889,
      "step": 372010
    },
    {
      "epoch": 0.002833771896043917,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012560876087608762,
      "loss": 3.2499,
      "step": 372020
    },
    {
      "epoch": 0.0028524151322021005,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001256067606760676,
      "loss": 3.2527,
      "step": 372030
    },
    {
      "epoch": 0.0028710583683602842,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001256047604760476,
      "loss": 3.2124,
      "step": 372040
    },
    {
      "epoch": 0.002889701604518468,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001256027602760276,
      "loss": 3.2317,
      "step": 372050
    },
    {
      "epoch": 0.0029083448406766518,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001256007600760076,
      "loss": 3.2629,
      "step": 372060
    },
    {
      "epoch": 0.002926988076834835,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001255987598759876,
      "loss": 3.1867,
      "step": 372070
    },
    {
      "epoch": 0.002945631312993019,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001255967596759676,
      "loss": 3.2762,
      "step": 372080
    },
    {
      "epoch": 0.0029642745491512026,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001255947594759476,
      "loss": 3.1612,
      "step": 372090
    },
    {
      "epoch": 0.0029829177853093864,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012559275927592759,
      "loss": 3.181,
      "step": 372100
    },
    {
      "epoch": 0.0030015610214675697,
      "grad_norm": 0.75,
      "learning_rate": 0.0001255907590759076,
      "loss": 2.98,
      "step": 372110
    },
    {
      "epoch": 0.0030202042576257535,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001255887588758876,
      "loss": 3.2063,
      "step": 372120
    },
    {
      "epoch": 0.0030388474937839373,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001255867586758676,
      "loss": 3.1929,
      "step": 372130
    },
    {
      "epoch": 0.003057490729942121,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001255847584758476,
      "loss": 3.1531,
      "step": 372140
    },
    {
      "epoch": 0.0030761339661003048,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001255827582758276,
      "loss": 3.2784,
      "step": 372150
    },
    {
      "epoch": 0.003094777202258488,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012558075807580759,
      "loss": 3.2281,
      "step": 372160
    },
    {
      "epoch": 0.003113420438416672,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001255787578757876,
      "loss": 3.2414,
      "step": 372170
    },
    {
      "epoch": 0.0031320636745748556,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012557675767576758,
      "loss": 3.2852,
      "step": 372180
    },
    {
      "epoch": 0.0031507069107330394,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012557475747574759,
      "loss": 3.1187,
      "step": 372190
    },
    {
      "epoch": 0.0031693501468912227,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012557275727572757,
      "loss": 3.1517,
      "step": 372200
    },
    {
      "epoch": 0.0031879933830494065,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012557075707570758,
      "loss": 3.2263,
      "step": 372210
    },
    {
      "epoch": 0.0032066366192075903,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012556875687568756,
      "loss": 3.08,
      "step": 372220
    },
    {
      "epoch": 0.003225279855365774,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012556675667566757,
      "loss": 3.1404,
      "step": 372230
    },
    {
      "epoch": 0.0032439230915239574,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012556475647564755,
      "loss": 3.3095,
      "step": 372240
    },
    {
      "epoch": 0.003262566327682141,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012556275627562756,
      "loss": 3.1748,
      "step": 372250
    },
    {
      "epoch": 0.003281209563840325,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012556075607560757,
      "loss": 3.1871,
      "step": 372260
    },
    {
      "epoch": 0.0032998527999985087,
      "grad_norm": 0.75,
      "learning_rate": 0.00012555875587558755,
      "loss": 3.3228,
      "step": 372270
    },
    {
      "epoch": 0.003318496036156692,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012555675567556756,
      "loss": 3.244,
      "step": 372280
    },
    {
      "epoch": 0.0033371392723148758,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012555475547554757,
      "loss": 3.1868,
      "step": 372290
    },
    {
      "epoch": 0.0033557825084730595,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012555275527552758,
      "loss": 3.1289,
      "step": 372300
    },
    {
      "epoch": 0.0033744257446312433,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012555075507550756,
      "loss": 3.1927,
      "step": 372310
    },
    {
      "epoch": 0.003393068980789427,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012554875487548757,
      "loss": 3.1351,
      "step": 372320
    },
    {
      "epoch": 0.0034117122169476104,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012554675467546755,
      "loss": 3.2877,
      "step": 372330
    },
    {
      "epoch": 0.003430355453105794,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012554475447544756,
      "loss": 3.1869,
      "step": 372340
    },
    {
      "epoch": 0.003448998689263978,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012554275427542754,
      "loss": 3.2957,
      "step": 372350
    },
    {
      "epoch": 0.0034676419254221617,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012554075407540755,
      "loss": 3.113,
      "step": 372360
    },
    {
      "epoch": 0.003486285161580345,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012553875387538753,
      "loss": 3.2939,
      "step": 372370
    },
    {
      "epoch": 0.0035049283977385288,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012553675367536754,
      "loss": 3.1199,
      "step": 372380
    },
    {
      "epoch": 0.0035235716338967125,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012553475347534753,
      "loss": 3.1647,
      "step": 372390
    },
    {
      "epoch": 0.0035422148700548963,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012553275327532753,
      "loss": 3.4132,
      "step": 372400
    },
    {
      "epoch": 0.0035608581062130796,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012553075307530752,
      "loss": 3.2711,
      "step": 372410
    },
    {
      "epoch": 0.0035795013423712634,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012552875287528753,
      "loss": 3.2827,
      "step": 372420
    },
    {
      "epoch": 0.003598144578529447,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012552675267526753,
      "loss": 3.1855,
      "step": 372430
    },
    {
      "epoch": 0.003616787814687631,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012552475247524754,
      "loss": 3.1399,
      "step": 372440
    },
    {
      "epoch": 0.0036354310508458147,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012552275227522755,
      "loss": 3.2329,
      "step": 372450
    },
    {
      "epoch": 0.003654074287003998,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012552075207520754,
      "loss": 3.2965,
      "step": 372460
    },
    {
      "epoch": 0.003672717523162182,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012551875187518754,
      "loss": 3.2878,
      "step": 372470
    },
    {
      "epoch": 0.0036913607593203656,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012551675167516753,
      "loss": 3.2522,
      "step": 372480
    },
    {
      "epoch": 0.0037100039954785493,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012551475147514754,
      "loss": 3.2727,
      "step": 372490
    },
    {
      "epoch": 0.0037286472316367327,
      "grad_norm": 0.75,
      "learning_rate": 0.00012551275127512752,
      "loss": 3.2962,
      "step": 372500
    },
    {
      "epoch": 0.0037472904677949164,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012551075107510753,
      "loss": 3.2374,
      "step": 372510
    },
    {
      "epoch": 0.0037659337039531,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001255087508750875,
      "loss": 3.2502,
      "step": 372520
    },
    {
      "epoch": 0.003784576940111284,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012550675067506752,
      "loss": 3.2305,
      "step": 372530
    },
    {
      "epoch": 0.0038032201762694673,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001255047504750475,
      "loss": 3.1651,
      "step": 372540
    },
    {
      "epoch": 0.003821863412427651,
      "grad_norm": 0.75,
      "learning_rate": 0.0001255027502750275,
      "loss": 3.1577,
      "step": 372550
    },
    {
      "epoch": 0.003840506648585835,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001255007500750075,
      "loss": 3.239,
      "step": 372560
    },
    {
      "epoch": 0.0038591498847440186,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001254987498749875,
      "loss": 3.1401,
      "step": 372570
    },
    {
      "epoch": 0.003877793120902202,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012549674967496748,
      "loss": 3.3818,
      "step": 372580
    },
    {
      "epoch": 0.0038964363570603857,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001254947494749475,
      "loss": 3.186,
      "step": 372590
    },
    {
      "epoch": 0.00391507959321857,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001254927492749275,
      "loss": 3.1979,
      "step": 372600
    },
    {
      "epoch": 0.003933722829376753,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001254907490749075,
      "loss": 3.25,
      "step": 372610
    },
    {
      "epoch": 0.0039523660655349365,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001254887488748875,
      "loss": 3.0278,
      "step": 372620
    },
    {
      "epoch": 0.00397100930169312,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001254867486748675,
      "loss": 3.3968,
      "step": 372630
    },
    {
      "epoch": 0.003989652537851304,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012548474847484748,
      "loss": 3.1932,
      "step": 372640
    },
    {
      "epoch": 0.004008295774009488,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001254827482748275,
      "loss": 3.2367,
      "step": 372650
    },
    {
      "epoch": 0.004026939010167672,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001254807480748075,
      "loss": 3.3308,
      "step": 372660
    },
    {
      "epoch": 0.004045582246325855,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00012547874787478748,
      "loss": 3.086,
      "step": 372670
    },
    {
      "epoch": 0.004064225482484039,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001254767476747675,
      "loss": 3.25,
      "step": 372680
    },
    {
      "epoch": 0.004082868718642222,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00012547474747474747,
      "loss": 3.1537,
      "step": 372690
    },
    {
      "epoch": 0.004101511954800406,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012547274727472748,
      "loss": 3.2567,
      "step": 372700
    },
    {
      "epoch": 0.0041201551909585896,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012547074707470747,
      "loss": 3.2186,
      "step": 372710
    },
    {
      "epoch": 0.004138798427116773,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012546874687468747,
      "loss": 3.2812,
      "step": 372720
    },
    {
      "epoch": 0.004157441663274957,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012546674667466746,
      "loss": 3.2504,
      "step": 372730
    },
    {
      "epoch": 0.004176084899433141,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012546474647464747,
      "loss": 3.203,
      "step": 372740
    },
    {
      "epoch": 0.004194728135591325,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012546274627462748,
      "loss": 3.305,
      "step": 372750
    },
    {
      "epoch": 0.004213371371749508,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012546074607460748,
      "loss": 3.1906,
      "step": 372760
    },
    {
      "epoch": 0.004232014607907692,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012545874587458747,
      "loss": 3.3266,
      "step": 372770
    },
    {
      "epoch": 0.004250657844065875,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012545674567456748,
      "loss": 3.1424,
      "step": 372780
    },
    {
      "epoch": 0.004269301080224059,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012545474547454746,
      "loss": 3.2218,
      "step": 372790
    },
    {
      "epoch": 0.004287944316382243,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012545274527452747,
      "loss": 3.0966,
      "step": 372800
    },
    {
      "epoch": 0.004306587552540426,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012545074507450745,
      "loss": 3.0837,
      "step": 372810
    },
    {
      "epoch": 0.00432523078869861,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012544874487448746,
      "loss": 3.2684,
      "step": 372820
    },
    {
      "epoch": 0.004343874024856794,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012544674467446744,
      "loss": 3.2573,
      "step": 372830
    },
    {
      "epoch": 0.004362517261014978,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012544474447444745,
      "loss": 3.2289,
      "step": 372840
    },
    {
      "epoch": 0.004381160497173161,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012544274427442746,
      "loss": 3.2523,
      "step": 372850
    },
    {
      "epoch": 0.004399803733331344,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012544074407440744,
      "loss": 3.252,
      "step": 372860
    },
    {
      "epoch": 0.004418446969489528,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012543874387438745,
      "loss": 3.2319,
      "step": 372870
    },
    {
      "epoch": 0.004437090205647712,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012543674367436743,
      "loss": 3.1964,
      "step": 372880
    },
    {
      "epoch": 0.004455733441805896,
      "grad_norm": 0.625,
      "learning_rate": 0.00012543474347434744,
      "loss": 3.1401,
      "step": 372890
    },
    {
      "epoch": 0.004474376677964079,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012543274327432742,
      "loss": 3.1199,
      "step": 372900
    },
    {
      "epoch": 0.004493019914122263,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012543074307430743,
      "loss": 3.329,
      "step": 372910
    },
    {
      "epoch": 0.004511663150280447,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012542874287428744,
      "loss": 3.1955,
      "step": 372920
    },
    {
      "epoch": 0.004530306386438631,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012542674267426745,
      "loss": 3.1631,
      "step": 372930
    },
    {
      "epoch": 0.004548949622596814,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012542474247424743,
      "loss": 3.1575,
      "step": 372940
    },
    {
      "epoch": 0.004567592858754997,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012542274227422744,
      "loss": 3.2107,
      "step": 372950
    },
    {
      "epoch": 0.004586236094913181,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012542074207420742,
      "loss": 3.2226,
      "step": 372960
    },
    {
      "epoch": 0.004604879331071365,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012541874187418743,
      "loss": 3.1267,
      "step": 372970
    },
    {
      "epoch": 0.004623522567229549,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012541674167416741,
      "loss": 3.2754,
      "step": 372980
    },
    {
      "epoch": 0.004642165803387732,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012541474147414742,
      "loss": 3.2205,
      "step": 372990
    },
    {
      "epoch": 0.004660809039545916,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001254127412741274,
      "loss": 3.3268,
      "step": 373000
    },
    {
      "epoch": 0.0046794522757041,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012541074107410741,
      "loss": 3.2327,
      "step": 373010
    },
    {
      "epoch": 0.004698095511862284,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001254087408740874,
      "loss": 3.2384,
      "step": 373020
    },
    {
      "epoch": 0.004716738748020467,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001254067406740674,
      "loss": 3.2168,
      "step": 373030
    },
    {
      "epoch": 0.00473538198417865,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001254047404740474,
      "loss": 3.2155,
      "step": 373040
    },
    {
      "epoch": 0.004754025220336834,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001254027402740274,
      "loss": 3.1811,
      "step": 373050
    },
    {
      "epoch": 0.004772668456495018,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001254007400740074,
      "loss": 3.2505,
      "step": 373060
    },
    {
      "epoch": 0.004791311692653202,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012539873987398742,
      "loss": 3.2427,
      "step": 373070
    },
    {
      "epoch": 0.004809954928811385,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012539673967396742,
      "loss": 3.2246,
      "step": 373080
    },
    {
      "epoch": 0.004828598164969569,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001253947394739474,
      "loss": 3.1288,
      "step": 373090
    },
    {
      "epoch": 0.004847241401127753,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012539273927392742,
      "loss": 3.189,
      "step": 373100
    },
    {
      "epoch": 0.004865884637285937,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001253907390739074,
      "loss": 3.1645,
      "step": 373110
    },
    {
      "epoch": 0.00488452787344412,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001253887388738874,
      "loss": 3.3337,
      "step": 373120
    },
    {
      "epoch": 0.004903171109602303,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001253867386738674,
      "loss": 3.1332,
      "step": 373130
    },
    {
      "epoch": 0.004921814345760487,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001253847384738474,
      "loss": 3.1064,
      "step": 373140
    },
    {
      "epoch": 0.004940457581918671,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012538273827382738,
      "loss": 3.2857,
      "step": 373150
    },
    {
      "epoch": 0.004959100818076855,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001253807380738074,
      "loss": 3.2263,
      "step": 373160
    },
    {
      "epoch": 0.004977744054235038,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012537873787378737,
      "loss": 3.1699,
      "step": 373170
    },
    {
      "epoch": 0.004996387290393222,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012537673767376738,
      "loss": 3.2724,
      "step": 373180
    },
    {
      "epoch": 0.005015030526551406,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012537473747374736,
      "loss": 3.1572,
      "step": 373190
    },
    {
      "epoch": 0.00503367376270959,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012537273727372737,
      "loss": 3.1902,
      "step": 373200
    },
    {
      "epoch": 0.005052316998867773,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012537073707370735,
      "loss": 3.1595,
      "step": 373210
    },
    {
      "epoch": 0.005070960235025956,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012536873687368736,
      "loss": 3.2309,
      "step": 373220
    },
    {
      "epoch": 0.00508960347118414,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012536673667366737,
      "loss": 3.193,
      "step": 373230
    },
    {
      "epoch": 0.005108246707342324,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012536473647364738,
      "loss": 3.2356,
      "step": 373240
    },
    {
      "epoch": 0.005126889943500508,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001253627362736274,
      "loss": 3.1824,
      "step": 373250
    },
    {
      "epoch": 0.0051455331796586914,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012536073607360737,
      "loss": 3.2274,
      "step": 373260
    },
    {
      "epoch": 0.005164176415816875,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012535873587358738,
      "loss": 3.0474,
      "step": 373270
    },
    {
      "epoch": 0.005182819651975059,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012535673567356736,
      "loss": 3.2054,
      "step": 373280
    },
    {
      "epoch": 0.005201462888133242,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012535473547354737,
      "loss": 3.0668,
      "step": 373290
    },
    {
      "epoch": 0.005220106124291426,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012535273527352735,
      "loss": 3.1928,
      "step": 373300
    },
    {
      "epoch": 0.005238749360449609,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012535073507350736,
      "loss": 3.1512,
      "step": 373310
    },
    {
      "epoch": 0.005257392596607793,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012534873487348735,
      "loss": 3.1479,
      "step": 373320
    },
    {
      "epoch": 0.005276035832765977,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012534673467346736,
      "loss": 3.1849,
      "step": 373330
    },
    {
      "epoch": 0.005294679068924161,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012534473447344734,
      "loss": 3.2301,
      "step": 373340
    },
    {
      "epoch": 0.0053133223050823445,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012534273427342735,
      "loss": 3.2525,
      "step": 373350
    },
    {
      "epoch": 0.005331965541240528,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012534073407340733,
      "loss": 3.1837,
      "step": 373360
    },
    {
      "epoch": 0.005350608777398712,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012533873387338734,
      "loss": 3.2148,
      "step": 373370
    },
    {
      "epoch": 0.005369252013556895,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012533673367336735,
      "loss": 3.2298,
      "step": 373380
    },
    {
      "epoch": 0.005387895249715079,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012533473347334736,
      "loss": 3.2708,
      "step": 373390
    },
    {
      "epoch": 0.005406538485873262,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012533273327332734,
      "loss": 3.222,
      "step": 373400
    },
    {
      "epoch": 0.005425181722031446,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012533073307330735,
      "loss": 3.2103,
      "step": 373410
    },
    {
      "epoch": 0.00544382495818963,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012532873287328733,
      "loss": 3.3513,
      "step": 373420
    },
    {
      "epoch": 0.005462468194347814,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012532673267326734,
      "loss": 3.1308,
      "step": 373430
    },
    {
      "epoch": 0.0054811114305059975,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012532473247324735,
      "loss": 3.2264,
      "step": 373440
    },
    {
      "epoch": 0.005499754666664181,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012532273227322733,
      "loss": 3.214,
      "step": 373450
    },
    {
      "epoch": 0.005518397902822364,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012532073207320734,
      "loss": 3.2499,
      "step": 373460
    },
    {
      "epoch": 0.005537041138980548,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012531873187318732,
      "loss": 3.2311,
      "step": 373470
    },
    {
      "epoch": 0.005555684375138732,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012531673167316733,
      "loss": 3.1725,
      "step": 373480
    },
    {
      "epoch": 0.0055743276112969154,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001253147314731473,
      "loss": 3.281,
      "step": 373490
    },
    {
      "epoch": 0.005592970847455099,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012531273127312732,
      "loss": 3.2322,
      "step": 373500
    },
    {
      "epoch": 0.005611614083613283,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001253107310731073,
      "loss": 3.2118,
      "step": 373510
    },
    {
      "epoch": 0.005630257319771467,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001253087308730873,
      "loss": 3.1479,
      "step": 373520
    },
    {
      "epoch": 0.0056489005559296505,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001253067306730673,
      "loss": 3.1918,
      "step": 373530
    },
    {
      "epoch": 0.005667543792087834,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001253047304730473,
      "loss": 3.1031,
      "step": 373540
    },
    {
      "epoch": 0.005686187028246017,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001253027302730273,
      "loss": 3.1865,
      "step": 373550
    },
    {
      "epoch": 0.005704830264404201,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012530073007300732,
      "loss": 3.2214,
      "step": 373560
    },
    {
      "epoch": 0.005723473500562385,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001252987298729873,
      "loss": 3.1702,
      "step": 373570
    },
    {
      "epoch": 0.0057421167367205685,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001252967296729673,
      "loss": 3.3583,
      "step": 373580
    },
    {
      "epoch": 0.005760759972878752,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001252947294729473,
      "loss": 3.1172,
      "step": 373590
    },
    {
      "epoch": 0.005779403209036936,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001252927292729273,
      "loss": 3.1876,
      "step": 373600
    },
    {
      "epoch": 0.00579804644519512,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012529072907290729,
      "loss": 3.2124,
      "step": 373610
    },
    {
      "epoch": 0.0058166896813533035,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001252887288728873,
      "loss": 3.1784,
      "step": 373620
    },
    {
      "epoch": 0.005835332917511486,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012528672867286728,
      "loss": 3.2798,
      "step": 373630
    },
    {
      "epoch": 0.00585397615366967,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012528472847284729,
      "loss": 3.2171,
      "step": 373640
    },
    {
      "epoch": 0.005872619389827854,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001252827282728273,
      "loss": 3.2197,
      "step": 373650
    },
    {
      "epoch": 0.005891262625986038,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012528072807280728,
      "loss": 3.1648,
      "step": 373660
    },
    {
      "epoch": 0.0059099058621442215,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001252787278727873,
      "loss": 3.2406,
      "step": 373670
    },
    {
      "epoch": 0.005928549098302405,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012527672767276727,
      "loss": 3.2256,
      "step": 373680
    },
    {
      "epoch": 0.005947192334460589,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012527472747274728,
      "loss": 3.0821,
      "step": 373690
    },
    {
      "epoch": 0.005965835570618773,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001252727272727273,
      "loss": 3.2432,
      "step": 373700
    },
    {
      "epoch": 0.0059844788067769565,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001252707270727073,
      "loss": 3.1922,
      "step": 373710
    },
    {
      "epoch": 0.0060031220429351394,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012526872687268728,
      "loss": 3.3122,
      "step": 373720
    },
    {
      "epoch": 0.006021765279093323,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001252667266726673,
      "loss": 3.167,
      "step": 373730
    },
    {
      "epoch": 0.006040408515251507,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012526472647264727,
      "loss": 3.1426,
      "step": 373740
    },
    {
      "epoch": 0.006059051751409691,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012526272627262728,
      "loss": 3.0752,
      "step": 373750
    },
    {
      "epoch": 0.0060776949875678745,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012526072607260726,
      "loss": 3.1923,
      "step": 373760
    },
    {
      "epoch": 0.006096338223726058,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012525872587258727,
      "loss": 3.1005,
      "step": 373770
    },
    {
      "epoch": 0.006114981459884242,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012525672567256725,
      "loss": 3.1942,
      "step": 373780
    },
    {
      "epoch": 0.006133624696042426,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012525472547254726,
      "loss": 3.2814,
      "step": 373790
    },
    {
      "epoch": 0.0061522679322006096,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012525272527252724,
      "loss": 3.107,
      "step": 373800
    },
    {
      "epoch": 0.0061709111683587925,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012525072507250725,
      "loss": 3.2995,
      "step": 373810
    },
    {
      "epoch": 0.006189554404516976,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012524872487248723,
      "loss": 3.2906,
      "step": 373820
    },
    {
      "epoch": 0.00620819764067516,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012524672467246724,
      "loss": 3.1954,
      "step": 373830
    },
    {
      "epoch": 0.006226840876833344,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012524472447244725,
      "loss": 3.1458,
      "step": 373840
    },
    {
      "epoch": 0.0062454841129915275,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012524272427242723,
      "loss": 3.1256,
      "step": 373850
    },
    {
      "epoch": 0.006264127349149711,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012524072407240724,
      "loss": 3.1032,
      "step": 373860
    },
    {
      "epoch": 0.006282770585307895,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012523872387238725,
      "loss": 3.1924,
      "step": 373870
    },
    {
      "epoch": 0.006301413821466079,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012523672367236726,
      "loss": 3.2282,
      "step": 373880
    },
    {
      "epoch": 0.006320057057624262,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012523472347234724,
      "loss": 3.2259,
      "step": 373890
    },
    {
      "epoch": 0.0063387002937824455,
      "grad_norm": 0.75,
      "learning_rate": 0.00012523272327232725,
      "loss": 3.1388,
      "step": 373900
    },
    {
      "epoch": 0.006357343529940629,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012523072307230724,
      "loss": 3.2646,
      "step": 373910
    },
    {
      "epoch": 0.006375986766098813,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012522872287228724,
      "loss": 3.1082,
      "step": 373920
    },
    {
      "epoch": 0.006394630002256997,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012522672267226723,
      "loss": 3.0843,
      "step": 373930
    },
    {
      "epoch": 0.0064132732384151805,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012522472247224724,
      "loss": 3.2343,
      "step": 373940
    },
    {
      "epoch": 0.006431916474573364,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012522272227222722,
      "loss": 3.2624,
      "step": 373950
    },
    {
      "epoch": 0.006450559710731548,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012522072207220723,
      "loss": 3.1847,
      "step": 373960
    },
    {
      "epoch": 0.006469202946889732,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001252187218721872,
      "loss": 3.3037,
      "step": 373970
    },
    {
      "epoch": 0.006487846183047915,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012521672167216722,
      "loss": 3.1035,
      "step": 373980
    },
    {
      "epoch": 0.0065064894192060985,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001252147214721472,
      "loss": 3.3666,
      "step": 373990
    },
    {
      "epoch": 0.006525132655364282,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001252127212721272,
      "loss": 3.1761,
      "step": 374000
    },
    {
      "epoch": 0.006543775891522466,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012521072107210722,
      "loss": 3.1203,
      "step": 374010
    },
    {
      "epoch": 0.00656241912768065,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012520872087208723,
      "loss": 3.1037,
      "step": 374020
    },
    {
      "epoch": 0.0065810623638388336,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012520672067206724,
      "loss": 3.0033,
      "step": 374030
    },
    {
      "epoch": 0.006599705599997017,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012520472047204722,
      "loss": 3.2724,
      "step": 374040
    },
    {
      "epoch": 0.006618348836155201,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012520272027202723,
      "loss": 3.172,
      "step": 374050
    },
    {
      "epoch": 0.006636992072313384,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001252007200720072,
      "loss": 3.3043,
      "step": 374060
    },
    {
      "epoch": 0.006655635308471568,
      "grad_norm": 0.75,
      "learning_rate": 0.00012519871987198722,
      "loss": 3.0631,
      "step": 374070
    },
    {
      "epoch": 0.0066742785446297515,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001251967196719672,
      "loss": 3.3051,
      "step": 374080
    },
    {
      "epoch": 0.006692921780787935,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001251947194719472,
      "loss": 3.2338,
      "step": 374090
    },
    {
      "epoch": 0.006711565016946119,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001251927192719272,
      "loss": 3.0821,
      "step": 374100
    },
    {
      "epoch": 0.006730208253104303,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001251907190719072,
      "loss": 3.2223,
      "step": 374110
    },
    {
      "epoch": 0.006748851489262487,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012518871887188718,
      "loss": 3.3294,
      "step": 374120
    },
    {
      "epoch": 0.00676749472542067,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001251867186718672,
      "loss": 3.248,
      "step": 374130
    },
    {
      "epoch": 0.006786137961578854,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012518471847184717,
      "loss": 3.0673,
      "step": 374140
    },
    {
      "epoch": 0.006804781197737037,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012518271827182718,
      "loss": 3.1264,
      "step": 374150
    },
    {
      "epoch": 0.006823424433895221,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001251807180718072,
      "loss": 3.2855,
      "step": 374160
    },
    {
      "epoch": 0.0068420676700534045,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012517871787178717,
      "loss": 3.2111,
      "step": 374170
    },
    {
      "epoch": 0.006860710906211588,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012517671767176718,
      "loss": 3.2396,
      "step": 374180
    },
    {
      "epoch": 0.006879354142369772,
      "grad_norm": 0.625,
      "learning_rate": 0.0001251747174717472,
      "loss": 3.0789,
      "step": 374190
    },
    {
      "epoch": 0.006897997378527956,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012517271727172718,
      "loss": 3.105,
      "step": 374200
    },
    {
      "epoch": 0.00691664061468614,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012517071707170718,
      "loss": 3.0578,
      "step": 374210
    },
    {
      "epoch": 0.006935283850844323,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012516871687168717,
      "loss": 3.3528,
      "step": 374220
    },
    {
      "epoch": 0.006953927087002506,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012516671667166718,
      "loss": 3.2781,
      "step": 374230
    },
    {
      "epoch": 0.00697257032316069,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012516471647164718,
      "loss": 3.1683,
      "step": 374240
    },
    {
      "epoch": 0.006991213559318874,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012516271627162717,
      "loss": 3.2483,
      "step": 374250
    },
    {
      "epoch": 0.0070098567954770576,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012516071607160718,
      "loss": 3.2049,
      "step": 374260
    },
    {
      "epoch": 0.007028500031635241,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012515871587158716,
      "loss": 3.185,
      "step": 374270
    },
    {
      "epoch": 0.007047143267793425,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012515671567156717,
      "loss": 3.1779,
      "step": 374280
    },
    {
      "epoch": 0.007065786503951609,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012515471547154715,
      "loss": 3.2025,
      "step": 374290
    },
    {
      "epoch": 0.007084429740109793,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012515271527152716,
      "loss": 3.1296,
      "step": 374300
    },
    {
      "epoch": 0.007103072976267976,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012515071507150714,
      "loss": 3.2148,
      "step": 374310
    },
    {
      "epoch": 0.007121716212426159,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012514871487148715,
      "loss": 3.2072,
      "step": 374320
    },
    {
      "epoch": 0.007140359448584343,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012514671467146716,
      "loss": 3.2683,
      "step": 374330
    },
    {
      "epoch": 0.007159002684742527,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012514471447144717,
      "loss": 3.2418,
      "step": 374340
    },
    {
      "epoch": 0.007177645920900711,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012514271427142715,
      "loss": 3.0581,
      "step": 374350
    },
    {
      "epoch": 0.007196289157058894,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012514071407140716,
      "loss": 3.1903,
      "step": 374360
    },
    {
      "epoch": 0.007214932393217078,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012513871387138714,
      "loss": 3.2986,
      "step": 374370
    },
    {
      "epoch": 0.007233575629375262,
      "grad_norm": 0.625,
      "learning_rate": 0.00012513671367136715,
      "loss": 3.3376,
      "step": 374380
    },
    {
      "epoch": 0.007252218865533446,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012513471347134713,
      "loss": 3.2232,
      "step": 374390
    },
    {
      "epoch": 0.007270862101691629,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012513271327132714,
      "loss": 3.2613,
      "step": 374400
    },
    {
      "epoch": 0.007289505337849812,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012513071307130712,
      "loss": 3.1147,
      "step": 374410
    },
    {
      "epoch": 0.007308148574007996,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012512871287128713,
      "loss": 3.2834,
      "step": 374420
    },
    {
      "epoch": 0.00732679181016618,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012512671267126714,
      "loss": 3.1755,
      "step": 374430
    },
    {
      "epoch": 0.007345435046324364,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012512471247124712,
      "loss": 3.115,
      "step": 374440
    },
    {
      "epoch": 0.007364078282482547,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012512271227122713,
      "loss": 3.2663,
      "step": 374450
    },
    {
      "epoch": 0.007382721518640731,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012512071207120711,
      "loss": 3.1978,
      "step": 374460
    },
    {
      "epoch": 0.007401364754798915,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012511871187118712,
      "loss": 3.2492,
      "step": 374470
    },
    {
      "epoch": 0.007420007990957099,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012511671167116713,
      "loss": 3.2509,
      "step": 374480
    },
    {
      "epoch": 0.0074386512271152816,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012511471147114712,
      "loss": 3.2384,
      "step": 374490
    },
    {
      "epoch": 0.007457294463273465,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012511271127112712,
      "loss": 3.2004,
      "step": 374500
    },
    {
      "epoch": 0.007475937699431649,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012511071107110713,
      "loss": 3.1972,
      "step": 374510
    },
    {
      "epoch": 0.007494580935589833,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012510871087108712,
      "loss": 3.1401,
      "step": 374520
    },
    {
      "epoch": 0.007513224171748017,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012510671067106712,
      "loss": 3.1922,
      "step": 374530
    },
    {
      "epoch": 0.0075318674079062,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001251047104710471,
      "loss": 3.2775,
      "step": 374540
    },
    {
      "epoch": 0.007550510644064384,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012510271027102712,
      "loss": 3.2728,
      "step": 374550
    },
    {
      "epoch": 0.007569153880222568,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001251007100710071,
      "loss": 3.1307,
      "step": 374560
    },
    {
      "epoch": 0.007587797116380752,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001250987098709871,
      "loss": 3.2047,
      "step": 374570
    },
    {
      "epoch": 0.007606440352538935,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001250967096709671,
      "loss": 3.2041,
      "step": 374580
    },
    {
      "epoch": 0.007625083588697118,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001250947094709471,
      "loss": 3.2391,
      "step": 374590
    },
    {
      "epoch": 0.007643726824855302,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012509270927092708,
      "loss": 3.1413,
      "step": 374600
    },
    {
      "epoch": 0.007662370061013486,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001250907090709071,
      "loss": 3.2428,
      "step": 374610
    },
    {
      "epoch": 0.00768101329717167,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001250887088708871,
      "loss": 3.2034,
      "step": 374620
    },
    {
      "epoch": 0.007699656533329853,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012508670867086708,
      "loss": 3.14,
      "step": 374630
    },
    {
      "epoch": 0.007718299769488037,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001250847084708471,
      "loss": 3.281,
      "step": 374640
    },
    {
      "epoch": 0.007736943005646221,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001250827082708271,
      "loss": 3.2253,
      "step": 374650
    },
    {
      "epoch": 0.007755586241804404,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001250807080708071,
      "loss": 3.3595,
      "step": 374660
    },
    {
      "epoch": 0.007774229477962588,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001250787078707871,
      "loss": 3.0564,
      "step": 374670
    },
    {
      "epoch": 0.007792872714120771,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001250767076707671,
      "loss": 3.255,
      "step": 374680
    },
    {
      "epoch": 0.007811515950278955,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012507470747074708,
      "loss": 3.2259,
      "step": 374690
    },
    {
      "epoch": 0.00783015918643714,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001250727072707271,
      "loss": 3.319,
      "step": 374700
    },
    {
      "epoch": 0.007848802422595322,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00012507070707070707,
      "loss": 3.2629,
      "step": 374710
    },
    {
      "epoch": 0.007867445658753506,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012506870687068708,
      "loss": 3.2056,
      "step": 374720
    },
    {
      "epoch": 0.00788608889491169,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012506670667066706,
      "loss": 3.268,
      "step": 374730
    },
    {
      "epoch": 0.007904732131069873,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012506470647064707,
      "loss": 3.2078,
      "step": 374740
    },
    {
      "epoch": 0.007923375367228057,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012506270627062705,
      "loss": 3.2131,
      "step": 374750
    },
    {
      "epoch": 0.00794201860338624,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012506070607060706,
      "loss": 3.352,
      "step": 374760
    },
    {
      "epoch": 0.007960661839544424,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012505870587058705,
      "loss": 3.2886,
      "step": 374770
    },
    {
      "epoch": 0.007979305075702608,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012505670567056706,
      "loss": 3.2509,
      "step": 374780
    },
    {
      "epoch": 0.007997948311860792,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012505470547054706,
      "loss": 3.2196,
      "step": 374790
    },
    {
      "epoch": 0.008016591548018976,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012505270527052705,
      "loss": 3.1887,
      "step": 374800
    },
    {
      "epoch": 0.00803523478417716,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012505070507050706,
      "loss": 3.307,
      "step": 374810
    },
    {
      "epoch": 0.008053878020335343,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012504870487048706,
      "loss": 3.2672,
      "step": 374820
    },
    {
      "epoch": 0.008072521256493527,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012504670467046707,
      "loss": 3.186,
      "step": 374830
    },
    {
      "epoch": 0.00809116449265171,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012504470447044706,
      "loss": 3.275,
      "step": 374840
    },
    {
      "epoch": 0.008109807728809895,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012504270427042706,
      "loss": 3.2211,
      "step": 374850
    },
    {
      "epoch": 0.008128450964968078,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012504070407040705,
      "loss": 3.3315,
      "step": 374860
    },
    {
      "epoch": 0.008147094201126262,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012503870387038706,
      "loss": 3.2062,
      "step": 374870
    },
    {
      "epoch": 0.008165737437284444,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012503670367036704,
      "loss": 3.3165,
      "step": 374880
    },
    {
      "epoch": 0.008184380673442628,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012503470347034705,
      "loss": 3.2728,
      "step": 374890
    },
    {
      "epoch": 0.008203023909600812,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012503270327032703,
      "loss": 3.1437,
      "step": 374900
    },
    {
      "epoch": 0.008221667145758995,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012503070307030704,
      "loss": 3.3039,
      "step": 374910
    },
    {
      "epoch": 0.008240310381917179,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012502870287028702,
      "loss": 3.1926,
      "step": 374920
    },
    {
      "epoch": 0.008258953618075363,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012502670267026703,
      "loss": 3.1259,
      "step": 374930
    },
    {
      "epoch": 0.008277596854233547,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000125024702470247,
      "loss": 3.2525,
      "step": 374940
    },
    {
      "epoch": 0.00829624009039173,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012502270227022702,
      "loss": 3.1364,
      "step": 374950
    },
    {
      "epoch": 0.008314883326549914,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012502070207020703,
      "loss": 3.1597,
      "step": 374960
    },
    {
      "epoch": 0.008333526562708098,
      "grad_norm": 0.90625,
      "learning_rate": 0.00012501870187018704,
      "loss": 3.1666,
      "step": 374970
    },
    {
      "epoch": 0.008352169798866282,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012501670167016702,
      "loss": 3.1611,
      "step": 374980
    },
    {
      "epoch": 0.008370813035024465,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012501470147014703,
      "loss": 3.1342,
      "step": 374990
    },
    {
      "epoch": 0.00838945627118265,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000125012701270127,
      "loss": 3.1215,
      "step": 375000
    },
    {
      "epoch": 0.008408099507340833,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012501070107010702,
      "loss": 3.1726,
      "step": 375010
    },
    {
      "epoch": 0.008426742743499017,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012500870087008703,
      "loss": 3.1341,
      "step": 375020
    },
    {
      "epoch": 0.0084453859796572,
      "grad_norm": 0.6875,
      "learning_rate": 0.000125006700670067,
      "loss": 3.2949,
      "step": 375030
    },
    {
      "epoch": 0.008464029215815384,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012500470047004702,
      "loss": 3.2534,
      "step": 375040
    },
    {
      "epoch": 0.008482672451973566,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000125002700270027,
      "loss": 3.2699,
      "step": 375050
    },
    {
      "epoch": 0.00850131568813175,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000125000700070007,
      "loss": 3.181,
      "step": 375060
    },
    {
      "epoch": 0.008519958924289934,
      "grad_norm": 0.76953125,
      "learning_rate": 0.000124998699869987,
      "loss": 3.2725,
      "step": 375070
    },
    {
      "epoch": 0.008538602160448118,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000124996699669967,
      "loss": 3.3296,
      "step": 375080
    },
    {
      "epoch": 0.008557245396606301,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012499469946994699,
      "loss": 3.1721,
      "step": 375090
    },
    {
      "epoch": 0.008575888632764485,
      "grad_norm": 0.6875,
      "learning_rate": 0.000124992699269927,
      "loss": 3.1254,
      "step": 375100
    },
    {
      "epoch": 0.008594531868922669,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000124990699069907,
      "loss": 3.0855,
      "step": 375110
    },
    {
      "epoch": 0.008613175105080853,
      "grad_norm": 0.703125,
      "learning_rate": 0.000124988698869887,
      "loss": 3.2716,
      "step": 375120
    },
    {
      "epoch": 0.008631818341239036,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000124986698669867,
      "loss": 3.2299,
      "step": 375130
    },
    {
      "epoch": 0.00865046157739722,
      "grad_norm": 0.71875,
      "learning_rate": 0.000124984698469847,
      "loss": 3.304,
      "step": 375140
    },
    {
      "epoch": 0.008669104813555404,
      "grad_norm": 0.6875,
      "learning_rate": 0.000124982698269827,
      "loss": 3.2507,
      "step": 375150
    },
    {
      "epoch": 0.008687748049713588,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000124980698069807,
      "loss": 3.2436,
      "step": 375160
    },
    {
      "epoch": 0.008706391285871772,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012497869786978698,
      "loss": 3.0985,
      "step": 375170
    },
    {
      "epoch": 0.008725034522029955,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000124976697669767,
      "loss": 3.3643,
      "step": 375180
    },
    {
      "epoch": 0.008743677758188139,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012497469746974697,
      "loss": 3.1283,
      "step": 375190
    },
    {
      "epoch": 0.008762320994346323,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012497269726972698,
      "loss": 3.149,
      "step": 375200
    },
    {
      "epoch": 0.008780964230504507,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012497069706970696,
      "loss": 3.1673,
      "step": 375210
    },
    {
      "epoch": 0.008799607466662689,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012496869686968697,
      "loss": 3.1515,
      "step": 375220
    },
    {
      "epoch": 0.008818250702820872,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012496669666966698,
      "loss": 3.0314,
      "step": 375230
    },
    {
      "epoch": 0.008836893938979056,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012496469646964696,
      "loss": 3.2198,
      "step": 375240
    },
    {
      "epoch": 0.00885553717513724,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012496269626962697,
      "loss": 3.1977,
      "step": 375250
    },
    {
      "epoch": 0.008874180411295424,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012496069606960695,
      "loss": 3.0789,
      "step": 375260
    },
    {
      "epoch": 0.008892823647453607,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012495869586958696,
      "loss": 3.2127,
      "step": 375270
    },
    {
      "epoch": 0.008911466883611791,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012495669566956697,
      "loss": 3.1964,
      "step": 375280
    },
    {
      "epoch": 0.008930110119769975,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012495469546954698,
      "loss": 3.2247,
      "step": 375290
    },
    {
      "epoch": 0.008948753355928159,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012495269526952696,
      "loss": 3.1367,
      "step": 375300
    },
    {
      "epoch": 0.008967396592086343,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012495069506950697,
      "loss": 3.2972,
      "step": 375310
    },
    {
      "epoch": 0.008986039828244526,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012494869486948695,
      "loss": 3.0885,
      "step": 375320
    },
    {
      "epoch": 0.00900468306440271,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012494669466946696,
      "loss": 3.2624,
      "step": 375330
    },
    {
      "epoch": 0.009023326300560894,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012494469446944694,
      "loss": 3.282,
      "step": 375340
    },
    {
      "epoch": 0.009041969536719078,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012494269426942695,
      "loss": 3.1855,
      "step": 375350
    },
    {
      "epoch": 0.009060612772877261,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012494069406940693,
      "loss": 3.1358,
      "step": 375360
    },
    {
      "epoch": 0.009079256009035445,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012493869386938694,
      "loss": 3.2676,
      "step": 375370
    },
    {
      "epoch": 0.009097899245193629,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012493669366936693,
      "loss": 3.3759,
      "step": 375380
    },
    {
      "epoch": 0.009116542481351811,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012493469346934694,
      "loss": 3.2498,
      "step": 375390
    },
    {
      "epoch": 0.009135185717509995,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012493269326932692,
      "loss": 3.0912,
      "step": 375400
    },
    {
      "epoch": 0.009153828953668178,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012493069306930693,
      "loss": 3.24,
      "step": 375410
    },
    {
      "epoch": 0.009172472189826362,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012492869286928694,
      "loss": 3.1607,
      "step": 375420
    },
    {
      "epoch": 0.009191115425984546,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012492669266926694,
      "loss": 3.14,
      "step": 375430
    },
    {
      "epoch": 0.00920975866214273,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012492469246924693,
      "loss": 3.1762,
      "step": 375440
    },
    {
      "epoch": 0.009228401898300913,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012492269226922694,
      "loss": 3.123,
      "step": 375450
    },
    {
      "epoch": 0.009247045134459097,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012492069206920695,
      "loss": 3.237,
      "step": 375460
    },
    {
      "epoch": 0.009265688370617281,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012491869186918693,
      "loss": 3.1726,
      "step": 375470
    },
    {
      "epoch": 0.009284331606775465,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012491669166916694,
      "loss": 3.2322,
      "step": 375480
    },
    {
      "epoch": 0.009302974842933649,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012491469146914692,
      "loss": 3.3225,
      "step": 375490
    },
    {
      "epoch": 0.009321618079091832,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012491269126912693,
      "loss": 3.2185,
      "step": 375500
    },
    {
      "epoch": 0.009340261315250016,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001249106910691069,
      "loss": 3.2396,
      "step": 375510
    },
    {
      "epoch": 0.0093589045514082,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012490869086908692,
      "loss": 3.2392,
      "step": 375520
    },
    {
      "epoch": 0.009377547787566384,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001249066906690669,
      "loss": 3.1954,
      "step": 375530
    },
    {
      "epoch": 0.009396191023724567,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001249046904690469,
      "loss": 3.1794,
      "step": 375540
    },
    {
      "epoch": 0.009414834259882751,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001249026902690269,
      "loss": 3.0951,
      "step": 375550
    },
    {
      "epoch": 0.009433477496040933,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001249006900690069,
      "loss": 3.2799,
      "step": 375560
    },
    {
      "epoch": 0.009452120732199117,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012489868986898688,
      "loss": 3.3191,
      "step": 375570
    },
    {
      "epoch": 0.0094707639683573,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001248966896689669,
      "loss": 3.2062,
      "step": 375580
    },
    {
      "epoch": 0.009489407204515484,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001248946894689469,
      "loss": 3.2933,
      "step": 375590
    },
    {
      "epoch": 0.009508050440673668,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001248926892689269,
      "loss": 3.1319,
      "step": 375600
    },
    {
      "epoch": 0.009526693676831852,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012489068906890692,
      "loss": 3.2894,
      "step": 375610
    },
    {
      "epoch": 0.009545336912990036,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001248886888688869,
      "loss": 3.2492,
      "step": 375620
    },
    {
      "epoch": 0.00956398014914822,
      "grad_norm": 0.75,
      "learning_rate": 0.0001248866886688669,
      "loss": 3.1317,
      "step": 375630
    },
    {
      "epoch": 0.009582623385306403,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001248846884688469,
      "loss": 3.2081,
      "step": 375640
    },
    {
      "epoch": 0.009601266621464587,
      "grad_norm": 0.75,
      "learning_rate": 0.0001248826882688269,
      "loss": 3.1493,
      "step": 375650
    },
    {
      "epoch": 0.00961990985762277,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012488068806880688,
      "loss": 3.3011,
      "step": 375660
    },
    {
      "epoch": 0.009638553093780955,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001248786878687869,
      "loss": 3.1873,
      "step": 375670
    },
    {
      "epoch": 0.009657196329939138,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012487668766876688,
      "loss": 3.2398,
      "step": 375680
    },
    {
      "epoch": 0.009675839566097322,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012487468746874688,
      "loss": 3.2845,
      "step": 375690
    },
    {
      "epoch": 0.009694482802255506,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012487268726872687,
      "loss": 3.1294,
      "step": 375700
    },
    {
      "epoch": 0.00971312603841369,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012487068706870688,
      "loss": 3.1652,
      "step": 375710
    },
    {
      "epoch": 0.009731769274571873,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012486868686868686,
      "loss": 3.2015,
      "step": 375720
    },
    {
      "epoch": 0.009750412510730057,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012486668666866687,
      "loss": 3.2302,
      "step": 375730
    },
    {
      "epoch": 0.00976905574688824,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012486468646864688,
      "loss": 3.3324,
      "step": 375740
    },
    {
      "epoch": 0.009787698983046423,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012486268626862689,
      "loss": 3.1802,
      "step": 375750
    },
    {
      "epoch": 0.009806342219204607,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012486068606860687,
      "loss": 3.2575,
      "step": 375760
    },
    {
      "epoch": 0.00982498545536279,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012485868586858688,
      "loss": 3.1322,
      "step": 375770
    },
    {
      "epoch": 0.009843628691520974,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012485668566856686,
      "loss": 3.4258,
      "step": 375780
    },
    {
      "epoch": 0.009862271927679158,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012485468546854687,
      "loss": 3.1503,
      "step": 375790
    },
    {
      "epoch": 0.009880915163837342,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012485268526852685,
      "loss": 3.1546,
      "step": 375800
    },
    {
      "epoch": 0.009899558399995526,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012485068506850686,
      "loss": 3.1202,
      "step": 375810
    },
    {
      "epoch": 0.00991820163615371,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012484868486848687,
      "loss": 3.2851,
      "step": 375820
    },
    {
      "epoch": 0.009936844872311893,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012484668466846685,
      "loss": 3.1096,
      "step": 375830
    },
    {
      "epoch": 0.009955488108470077,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012484468446844686,
      "loss": 3.1147,
      "step": 375840
    },
    {
      "epoch": 0.00997413134462826,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012484268426842684,
      "loss": 3.4285,
      "step": 375850
    },
    {
      "epoch": 0.009992774580786444,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012484068406840685,
      "loss": 3.2422,
      "step": 375860
    },
    {
      "epoch": 0.010011417816944628,
      "grad_norm": 0.75,
      "learning_rate": 0.00012483868386838683,
      "loss": 3.243,
      "step": 375870
    },
    {
      "epoch": 0.010030061053102812,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012483668366836684,
      "loss": 3.1964,
      "step": 375880
    },
    {
      "epoch": 0.010048704289260996,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012483468346834682,
      "loss": 3.2755,
      "step": 375890
    },
    {
      "epoch": 0.01006734752541918,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012483268326832683,
      "loss": 3.1159,
      "step": 375900
    },
    {
      "epoch": 0.010085990761577361,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012483068306830684,
      "loss": 3.1849,
      "step": 375910
    },
    {
      "epoch": 0.010104633997735545,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012482868286828685,
      "loss": 3.2272,
      "step": 375920
    },
    {
      "epoch": 0.010123277233893729,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012482668266826683,
      "loss": 3.2785,
      "step": 375930
    },
    {
      "epoch": 0.010141920470051913,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012482468246824684,
      "loss": 3.1582,
      "step": 375940
    },
    {
      "epoch": 0.010160563706210097,
      "grad_norm": 0.75,
      "learning_rate": 0.00012482268226822682,
      "loss": 3.1632,
      "step": 375950
    },
    {
      "epoch": 0.01017920694236828,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012482068206820683,
      "loss": 3.1478,
      "step": 375960
    },
    {
      "epoch": 0.010197850178526464,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00012481868186818682,
      "loss": 3.0938,
      "step": 375970
    },
    {
      "epoch": 0.010216493414684648,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012481668166816682,
      "loss": 3.1761,
      "step": 375980
    },
    {
      "epoch": 0.010235136650842832,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001248146814681468,
      "loss": 3.2423,
      "step": 375990
    },
    {
      "epoch": 0.010253779887001015,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012481268126812682,
      "loss": 3.3254,
      "step": 376000
    },
    {
      "epoch": 0.010272423123159199,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012481068106810682,
      "loss": 3.2904,
      "step": 376010
    },
    {
      "epoch": 0.010291066359317383,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001248086808680868,
      "loss": 3.1813,
      "step": 376020
    },
    {
      "epoch": 0.010309709595475567,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012480668066806682,
      "loss": 3.2775,
      "step": 376030
    },
    {
      "epoch": 0.01032835283163375,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001248046804680468,
      "loss": 3.1586,
      "step": 376040
    },
    {
      "epoch": 0.010346996067791934,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001248026802680268,
      "loss": 3.2272,
      "step": 376050
    },
    {
      "epoch": 0.010365639303950118,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012480068006800682,
      "loss": 3.1169,
      "step": 376060
    },
    {
      "epoch": 0.010384282540108302,
      "grad_norm": 0.75,
      "learning_rate": 0.0001247986798679868,
      "loss": 3.1186,
      "step": 376070
    },
    {
      "epoch": 0.010402925776266484,
      "grad_norm": 0.75,
      "learning_rate": 0.0001247966796679668,
      "loss": 3.31,
      "step": 376080
    },
    {
      "epoch": 0.010421569012424668,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012479467946794682,
      "loss": 3.3008,
      "step": 376090
    },
    {
      "epoch": 0.010440212248582851,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001247926792679268,
      "loss": 3.1368,
      "step": 376100
    },
    {
      "epoch": 0.010458855484741035,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001247906790679068,
      "loss": 3.2597,
      "step": 376110
    },
    {
      "epoch": 0.010477498720899219,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001247886788678868,
      "loss": 3.267,
      "step": 376120
    },
    {
      "epoch": 0.010496141957057403,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001247866786678668,
      "loss": 3.1457,
      "step": 376130
    },
    {
      "epoch": 0.010514785193215586,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012478467846784678,
      "loss": 3.1778,
      "step": 376140
    },
    {
      "epoch": 0.01053342842937377,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001247826782678268,
      "loss": 3.1889,
      "step": 376150
    },
    {
      "epoch": 0.010552071665531954,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012478067806780677,
      "loss": 3.1879,
      "step": 376160
    },
    {
      "epoch": 0.010570714901690138,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012477867786778678,
      "loss": 3.1063,
      "step": 376170
    },
    {
      "epoch": 0.010589358137848321,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012477667766776676,
      "loss": 3.1724,
      "step": 376180
    },
    {
      "epoch": 0.010608001374006505,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012477467746774677,
      "loss": 3.1501,
      "step": 376190
    },
    {
      "epoch": 0.010626644610164689,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012477267726772678,
      "loss": 3.2452,
      "step": 376200
    },
    {
      "epoch": 0.010645287846322873,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012477067706770676,
      "loss": 3.1505,
      "step": 376210
    },
    {
      "epoch": 0.010663931082481056,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012476867686768677,
      "loss": 3.2575,
      "step": 376220
    },
    {
      "epoch": 0.01068257431863924,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012476667666766678,
      "loss": 3.1986,
      "step": 376230
    },
    {
      "epoch": 0.010701217554797424,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001247646764676468,
      "loss": 3.2305,
      "step": 376240
    },
    {
      "epoch": 0.010719860790955606,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012476267626762677,
      "loss": 3.2,
      "step": 376250
    },
    {
      "epoch": 0.01073850402711379,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012476067606760678,
      "loss": 3.1852,
      "step": 376260
    },
    {
      "epoch": 0.010757147263271974,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012475867586758676,
      "loss": 3.2365,
      "step": 376270
    },
    {
      "epoch": 0.010775790499430157,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012475667566756677,
      "loss": 3.129,
      "step": 376280
    },
    {
      "epoch": 0.010794433735588341,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012475467546754676,
      "loss": 3.2722,
      "step": 376290
    },
    {
      "epoch": 0.010813076971746525,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012475267526752676,
      "loss": 3.1677,
      "step": 376300
    },
    {
      "epoch": 0.010831720207904709,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012475067506750675,
      "loss": 3.1506,
      "step": 376310
    },
    {
      "epoch": 0.010850363444062892,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012474867486748676,
      "loss": 3.2409,
      "step": 376320
    },
    {
      "epoch": 0.010869006680221076,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012474667466746674,
      "loss": 3.1507,
      "step": 376330
    },
    {
      "epoch": 0.01088764991637926,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012474467446744675,
      "loss": 3.1319,
      "step": 376340
    },
    {
      "epoch": 0.010906293152537444,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012474267426742673,
      "loss": 3.3273,
      "step": 376350
    },
    {
      "epoch": 0.010924936388695627,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012474067406740674,
      "loss": 3.22,
      "step": 376360
    },
    {
      "epoch": 0.010943579624853811,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012473867386738675,
      "loss": 3.1107,
      "step": 376370
    },
    {
      "epoch": 0.010962222861011995,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012473667366736676,
      "loss": 3.0953,
      "step": 376380
    },
    {
      "epoch": 0.010980866097170179,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012473467346734674,
      "loss": 3.224,
      "step": 376390
    },
    {
      "epoch": 0.010999509333328363,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012473267326732675,
      "loss": 3.2422,
      "step": 376400
    },
    {
      "epoch": 0.011018152569486546,
      "grad_norm": 0.75,
      "learning_rate": 0.00012473067306730676,
      "loss": 3.2666,
      "step": 376410
    },
    {
      "epoch": 0.011036795805644728,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012472867286728674,
      "loss": 3.2057,
      "step": 376420
    },
    {
      "epoch": 0.011055439041802912,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012472667266726675,
      "loss": 3.2059,
      "step": 376430
    },
    {
      "epoch": 0.011074082277961096,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012472467246724673,
      "loss": 3.1485,
      "step": 376440
    },
    {
      "epoch": 0.01109272551411928,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012472267226722674,
      "loss": 3.0683,
      "step": 376450
    },
    {
      "epoch": 0.011111368750277463,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012472067206720672,
      "loss": 3.1762,
      "step": 376460
    },
    {
      "epoch": 0.011130011986435647,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012471867186718673,
      "loss": 3.2047,
      "step": 376470
    },
    {
      "epoch": 0.011148655222593831,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001247166716671667,
      "loss": 3.2103,
      "step": 376480
    },
    {
      "epoch": 0.011167298458752015,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012471467146714672,
      "loss": 3.1628,
      "step": 376490
    },
    {
      "epoch": 0.011185941694910198,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001247126712671267,
      "loss": 3.2689,
      "step": 376500
    },
    {
      "epoch": 0.011204584931068382,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001247106710671067,
      "loss": 3.2594,
      "step": 376510
    },
    {
      "epoch": 0.011223228167226566,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001247086708670867,
      "loss": 3.2,
      "step": 376520
    },
    {
      "epoch": 0.01124187140338475,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001247066706670667,
      "loss": 3.1984,
      "step": 376530
    },
    {
      "epoch": 0.011260514639542933,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001247046704670467,
      "loss": 3.1612,
      "step": 376540
    },
    {
      "epoch": 0.011279157875701117,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012470267026702672,
      "loss": 3.3108,
      "step": 376550
    },
    {
      "epoch": 0.011297801111859301,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001247006700670067,
      "loss": 3.1341,
      "step": 376560
    },
    {
      "epoch": 0.011316444348017485,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012469866986698671,
      "loss": 3.2467,
      "step": 376570
    },
    {
      "epoch": 0.011335087584175669,
      "grad_norm": 0.625,
      "learning_rate": 0.0001246966696669667,
      "loss": 3.2048,
      "step": 376580
    },
    {
      "epoch": 0.01135373082033385,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001246946694669467,
      "loss": 3.0803,
      "step": 376590
    },
    {
      "epoch": 0.011372374056492034,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012469266926692671,
      "loss": 3.2249,
      "step": 376600
    },
    {
      "epoch": 0.011391017292650218,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001246906690669067,
      "loss": 3.1606,
      "step": 376610
    },
    {
      "epoch": 0.011409660528808402,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001246886688668867,
      "loss": 3.153,
      "step": 376620
    },
    {
      "epoch": 0.011428303764966586,
      "grad_norm": 0.75,
      "learning_rate": 0.0001246866686668667,
      "loss": 3.2628,
      "step": 376630
    },
    {
      "epoch": 0.01144694700112477,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001246846684668467,
      "loss": 3.2052,
      "step": 376640
    },
    {
      "epoch": 0.011465590237282953,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012468266826682668,
      "loss": 3.0832,
      "step": 376650
    },
    {
      "epoch": 0.011484233473441137,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001246806680668067,
      "loss": 3.2891,
      "step": 376660
    },
    {
      "epoch": 0.01150287670959932,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012467866786678667,
      "loss": 3.1071,
      "step": 376670
    },
    {
      "epoch": 0.011521519945757504,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012467666766676668,
      "loss": 3.3704,
      "step": 376680
    },
    {
      "epoch": 0.011540163181915688,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001246746674667467,
      "loss": 3.2098,
      "step": 376690
    },
    {
      "epoch": 0.011558806418073872,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001246726672667267,
      "loss": 3.1615,
      "step": 376700
    },
    {
      "epoch": 0.011577449654232056,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012467066706670668,
      "loss": 3.1366,
      "step": 376710
    },
    {
      "epoch": 0.01159609289039024,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001246686668666867,
      "loss": 3.2184,
      "step": 376720
    },
    {
      "epoch": 0.011614736126548423,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012466666666666667,
      "loss": 3.1742,
      "step": 376730
    },
    {
      "epoch": 0.011633379362706607,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012466466646664668,
      "loss": 3.2953,
      "step": 376740
    },
    {
      "epoch": 0.01165202259886479,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012466266626662666,
      "loss": 3.1793,
      "step": 376750
    },
    {
      "epoch": 0.011670665835022973,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012466066606660667,
      "loss": 3.1892,
      "step": 376760
    },
    {
      "epoch": 0.011689309071181157,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012465866586658665,
      "loss": 3.4079,
      "step": 376770
    },
    {
      "epoch": 0.01170795230733934,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012465666566656666,
      "loss": 3.3316,
      "step": 376780
    },
    {
      "epoch": 0.011726595543497524,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012465466546654664,
      "loss": 3.1827,
      "step": 376790
    },
    {
      "epoch": 0.011745238779655708,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012465266526652665,
      "loss": 3.2263,
      "step": 376800
    },
    {
      "epoch": 0.011763882015813892,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012465066506650666,
      "loss": 3.1808,
      "step": 376810
    },
    {
      "epoch": 0.011782525251972075,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012464866486648664,
      "loss": 3.2281,
      "step": 376820
    },
    {
      "epoch": 0.01180116848813026,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012464666466646665,
      "loss": 3.2626,
      "step": 376830
    },
    {
      "epoch": 0.011819811724288443,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012464466446644664,
      "loss": 3.2693,
      "step": 376840
    },
    {
      "epoch": 0.011838454960446627,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012464266426642664,
      "loss": 3.1376,
      "step": 376850
    },
    {
      "epoch": 0.01185709819660481,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012464066406640665,
      "loss": 3.1996,
      "step": 376860
    },
    {
      "epoch": 0.011875741432762994,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012463866386638666,
      "loss": 3.3231,
      "step": 376870
    },
    {
      "epoch": 0.011894384668921178,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012463666366636664,
      "loss": 3.1062,
      "step": 376880
    },
    {
      "epoch": 0.011913027905079362,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012463466346634665,
      "loss": 3.0805,
      "step": 376890
    },
    {
      "epoch": 0.011931671141237546,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012463266326632664,
      "loss": 3.2414,
      "step": 376900
    },
    {
      "epoch": 0.01195031437739573,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012463066306630665,
      "loss": 3.2919,
      "step": 376910
    },
    {
      "epoch": 0.011968957613553913,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012462866286628663,
      "loss": 3.1992,
      "step": 376920
    },
    {
      "epoch": 0.011987600849712097,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012462666266626664,
      "loss": 3.2459,
      "step": 376930
    },
    {
      "epoch": 0.012006244085870279,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012462466246624662,
      "loss": 3.3366,
      "step": 376940
    },
    {
      "epoch": 0.012024887322028463,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012462266226622663,
      "loss": 3.202,
      "step": 376950
    },
    {
      "epoch": 0.012043530558186646,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001246206620662066,
      "loss": 3.2093,
      "step": 376960
    },
    {
      "epoch": 0.01206217379434483,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012461866186618662,
      "loss": 3.1723,
      "step": 376970
    },
    {
      "epoch": 0.012080817030503014,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001246166616661666,
      "loss": 3.1361,
      "step": 376980
    },
    {
      "epoch": 0.012099460266661198,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001246146614661466,
      "loss": 3.21,
      "step": 376990
    },
    {
      "epoch": 0.012118103502819381,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012461266126612662,
      "loss": 3.2233,
      "step": 377000
    },
    {
      "epoch": 0.012136746738977565,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012461066106610663,
      "loss": 3.1623,
      "step": 377010
    },
    {
      "epoch": 0.012155389975135749,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012460866086608664,
      "loss": 3.2378,
      "step": 377020
    },
    {
      "epoch": 0.012174033211293933,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012460666066606662,
      "loss": 3.2886,
      "step": 377030
    },
    {
      "epoch": 0.012192676447452117,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012460466046604663,
      "loss": 3.2203,
      "step": 377040
    },
    {
      "epoch": 0.0122113196836103,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001246026602660266,
      "loss": 3.1683,
      "step": 377050
    },
    {
      "epoch": 0.012229962919768484,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012460066006600662,
      "loss": 3.241,
      "step": 377060
    },
    {
      "epoch": 0.012248606155926668,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001245986598659866,
      "loss": 3.2177,
      "step": 377070
    },
    {
      "epoch": 0.012267249392084852,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001245966596659666,
      "loss": 3.2336,
      "step": 377080
    },
    {
      "epoch": 0.012285892628243035,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001245946594659466,
      "loss": 3.1281,
      "step": 377090
    },
    {
      "epoch": 0.012304535864401219,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001245926592659266,
      "loss": 3.1823,
      "step": 377100
    },
    {
      "epoch": 0.012323179100559401,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012459065906590658,
      "loss": 3.1642,
      "step": 377110
    },
    {
      "epoch": 0.012341822336717585,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001245886588658866,
      "loss": 3.0736,
      "step": 377120
    },
    {
      "epoch": 0.012360465572875769,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012458665866586658,
      "loss": 3.1545,
      "step": 377130
    },
    {
      "epoch": 0.012379108809033952,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012458465846584658,
      "loss": 3.1945,
      "step": 377140
    },
    {
      "epoch": 0.012397752045192136,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012458265826582657,
      "loss": 3.1666,
      "step": 377150
    },
    {
      "epoch": 0.01241639528135032,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012458065806580658,
      "loss": 3.2166,
      "step": 377160
    },
    {
      "epoch": 0.012435038517508504,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012457865786578658,
      "loss": 3.1569,
      "step": 377170
    },
    {
      "epoch": 0.012453681753666688,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001245766576657666,
      "loss": 3.1756,
      "step": 377180
    },
    {
      "epoch": 0.012472324989824871,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001245746574657466,
      "loss": 3.1248,
      "step": 377190
    },
    {
      "epoch": 0.012490968225983055,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00012457265726572659,
      "loss": 3.2269,
      "step": 377200
    },
    {
      "epoch": 0.012509611462141239,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001245706570657066,
      "loss": 3.2719,
      "step": 377210
    },
    {
      "epoch": 0.012528254698299423,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012456865686568658,
      "loss": 3.223,
      "step": 377220
    },
    {
      "epoch": 0.012546897934457606,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012456665666566659,
      "loss": 3.1617,
      "step": 377230
    },
    {
      "epoch": 0.01256554117061579,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00012456465646564657,
      "loss": 3.2346,
      "step": 377240
    },
    {
      "epoch": 0.012584184406773974,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012456265626562658,
      "loss": 3.27,
      "step": 377250
    },
    {
      "epoch": 0.012602827642932158,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012456065606560656,
      "loss": 3.1051,
      "step": 377260
    },
    {
      "epoch": 0.012621470879090341,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012455865586558657,
      "loss": 3.2075,
      "step": 377270
    },
    {
      "epoch": 0.012640114115248523,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012455665566556655,
      "loss": 3.2592,
      "step": 377280
    },
    {
      "epoch": 0.012658757351406707,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012455465546554656,
      "loss": 3.3802,
      "step": 377290
    },
    {
      "epoch": 0.012677400587564891,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012455265526552654,
      "loss": 3.3036,
      "step": 377300
    },
    {
      "epoch": 0.012696043823723075,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012455065506550655,
      "loss": 3.1498,
      "step": 377310
    },
    {
      "epoch": 0.012714687059881258,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012454865486548656,
      "loss": 3.1375,
      "step": 377320
    },
    {
      "epoch": 0.012733330296039442,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012454665466546657,
      "loss": 3.1479,
      "step": 377330
    },
    {
      "epoch": 0.012751973532197626,
      "grad_norm": 0.890625,
      "learning_rate": 0.00012454465446544655,
      "loss": 3.2422,
      "step": 377340
    },
    {
      "epoch": 0.01277061676835581,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012454265426542656,
      "loss": 3.2224,
      "step": 377350
    },
    {
      "epoch": 0.012789260004513994,
      "grad_norm": 0.625,
      "learning_rate": 0.00012454065406540654,
      "loss": 3.1924,
      "step": 377360
    },
    {
      "epoch": 0.012807903240672177,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012453865386538655,
      "loss": 3.1408,
      "step": 377370
    },
    {
      "epoch": 0.012826546476830361,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012453665366536653,
      "loss": 3.2359,
      "step": 377380
    },
    {
      "epoch": 0.012845189712988545,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012453465346534654,
      "loss": 3.304,
      "step": 377390
    },
    {
      "epoch": 0.012863832949146729,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012453265326532655,
      "loss": 3.0801,
      "step": 377400
    },
    {
      "epoch": 0.012882476185304912,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012453065306530653,
      "loss": 3.3523,
      "step": 377410
    },
    {
      "epoch": 0.012901119421463096,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012452865286528654,
      "loss": 3.0737,
      "step": 377420
    },
    {
      "epoch": 0.01291976265762128,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012452665266526652,
      "loss": 3.1436,
      "step": 377430
    },
    {
      "epoch": 0.012938405893779464,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012452465246524653,
      "loss": 3.2221,
      "step": 377440
    },
    {
      "epoch": 0.012957049129937646,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012452265226522652,
      "loss": 3.235,
      "step": 377450
    },
    {
      "epoch": 0.01297569236609583,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012452065206520652,
      "loss": 3.1406,
      "step": 377460
    },
    {
      "epoch": 0.012994335602254013,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001245186518651865,
      "loss": 3.2019,
      "step": 377470
    },
    {
      "epoch": 0.013012978838412197,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012451665166516652,
      "loss": 3.2929,
      "step": 377480
    },
    {
      "epoch": 0.01303162207457038,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012451465146514653,
      "loss": 3.1388,
      "step": 377490
    },
    {
      "epoch": 0.013050265310728565,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012451265126512653,
      "loss": 3.2774,
      "step": 377500
    },
    {
      "epoch": 0.013068908546886748,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012451065106510652,
      "loss": 3.2035,
      "step": 377510
    },
    {
      "epoch": 0.013087551783044932,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012450865086508653,
      "loss": 3.225,
      "step": 377520
    },
    {
      "epoch": 0.013106195019203116,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001245066506650665,
      "loss": 3.1838,
      "step": 377530
    },
    {
      "epoch": 0.0131248382553613,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012450465046504652,
      "loss": 3.282,
      "step": 377540
    },
    {
      "epoch": 0.013143481491519483,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001245026502650265,
      "loss": 3.2796,
      "step": 377550
    },
    {
      "epoch": 0.013162124727677667,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001245006500650065,
      "loss": 3.1609,
      "step": 377560
    },
    {
      "epoch": 0.013180767963835851,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001244986498649865,
      "loss": 3.1766,
      "step": 377570
    },
    {
      "epoch": 0.013199411199994035,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001244966496649665,
      "loss": 3.2114,
      "step": 377580
    },
    {
      "epoch": 0.013218054436152218,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001244946494649465,
      "loss": 3.2745,
      "step": 377590
    },
    {
      "epoch": 0.013236697672310402,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001244926492649265,
      "loss": 3.0195,
      "step": 377600
    },
    {
      "epoch": 0.013255340908468586,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001244906490649065,
      "loss": 3.1423,
      "step": 377610
    },
    {
      "epoch": 0.013273984144626768,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012448864886488648,
      "loss": 3.2238,
      "step": 377620
    },
    {
      "epoch": 0.013292627380784952,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001244866486648665,
      "loss": 3.2195,
      "step": 377630
    },
    {
      "epoch": 0.013311270616943136,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001244846484648465,
      "loss": 3.2307,
      "step": 377640
    },
    {
      "epoch": 0.01332991385310132,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001244826482648265,
      "loss": 3.0579,
      "step": 377650
    },
    {
      "epoch": 0.013348557089259503,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001244806480648065,
      "loss": 3.2403,
      "step": 377660
    },
    {
      "epoch": 0.013367200325417687,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001244786478647865,
      "loss": 3.2209,
      "step": 377670
    },
    {
      "epoch": 0.01338584356157587,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012447664766476648,
      "loss": 3.1919,
      "step": 377680
    },
    {
      "epoch": 0.013404486797734054,
      "grad_norm": 0.75,
      "learning_rate": 0.0001244746474647465,
      "loss": 3.1218,
      "step": 377690
    },
    {
      "epoch": 0.013423130033892238,
      "grad_norm": 0.859375,
      "learning_rate": 0.00012447264726472647,
      "loss": 3.3033,
      "step": 377700
    },
    {
      "epoch": 0.013441773270050422,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012447064706470648,
      "loss": 3.2175,
      "step": 377710
    },
    {
      "epoch": 0.013460416506208606,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012446864686468646,
      "loss": 3.1828,
      "step": 377720
    },
    {
      "epoch": 0.01347905974236679,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012446664666466647,
      "loss": 3.2462,
      "step": 377730
    },
    {
      "epoch": 0.013497702978524973,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012446464646464646,
      "loss": 3.1859,
      "step": 377740
    },
    {
      "epoch": 0.013516346214683157,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012446264626462646,
      "loss": 3.1942,
      "step": 377750
    },
    {
      "epoch": 0.01353498945084134,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012446064606460645,
      "loss": 3.2337,
      "step": 377760
    },
    {
      "epoch": 0.013553632686999524,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012445864586458646,
      "loss": 3.1313,
      "step": 377770
    },
    {
      "epoch": 0.013572275923157708,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012445664566456647,
      "loss": 3.2746,
      "step": 377780
    },
    {
      "epoch": 0.01359091915931589,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012445464546454645,
      "loss": 3.3809,
      "step": 377790
    },
    {
      "epoch": 0.013609562395474074,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012445264526452646,
      "loss": 3.1629,
      "step": 377800
    },
    {
      "epoch": 0.013628205631632258,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012445064506450647,
      "loss": 3.2204,
      "step": 377810
    },
    {
      "epoch": 0.013646848867790442,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012444864486448647,
      "loss": 3.187,
      "step": 377820
    },
    {
      "epoch": 0.013665492103948625,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012444664466446646,
      "loss": 3.2886,
      "step": 377830
    },
    {
      "epoch": 0.013684135340106809,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012444464446444647,
      "loss": 3.0921,
      "step": 377840
    },
    {
      "epoch": 0.013702778576264993,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012444264426442645,
      "loss": 3.0662,
      "step": 377850
    },
    {
      "epoch": 0.013721421812423177,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012444064406440646,
      "loss": 3.1122,
      "step": 377860
    },
    {
      "epoch": 0.01374006504858136,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012443864386438644,
      "loss": 3.1787,
      "step": 377870
    },
    {
      "epoch": 0.013758708284739544,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012443664366436645,
      "loss": 3.1141,
      "step": 377880
    },
    {
      "epoch": 0.013777351520897728,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012443464346434643,
      "loss": 3.1724,
      "step": 377890
    },
    {
      "epoch": 0.013795994757055912,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012443264326432644,
      "loss": 3.1937,
      "step": 377900
    },
    {
      "epoch": 0.013814637993214095,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012443064306430642,
      "loss": 3.0975,
      "step": 377910
    },
    {
      "epoch": 0.01383328122937228,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012442864286428643,
      "loss": 3.1982,
      "step": 377920
    },
    {
      "epoch": 0.013851924465530463,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001244266426642664,
      "loss": 3.1882,
      "step": 377930
    },
    {
      "epoch": 0.013870567701688647,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012442464246424642,
      "loss": 3.2531,
      "step": 377940
    },
    {
      "epoch": 0.01388921093784683,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012442264226422643,
      "loss": 3.2354,
      "step": 377950
    },
    {
      "epoch": 0.013907854174005013,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012442064206420644,
      "loss": 3.2631,
      "step": 377960
    },
    {
      "epoch": 0.013926497410163196,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012441864186418642,
      "loss": 3.1762,
      "step": 377970
    },
    {
      "epoch": 0.01394514064632138,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00012441664166416643,
      "loss": 3.3731,
      "step": 377980
    },
    {
      "epoch": 0.013963783882479564,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012441464146414644,
      "loss": 3.2196,
      "step": 377990
    },
    {
      "epoch": 0.013982427118637748,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012441264126412642,
      "loss": 3.2396,
      "step": 378000
    },
    {
      "epoch": 0.014001070354795931,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00012441064106410643,
      "loss": 3.2595,
      "step": 378010
    },
    {
      "epoch": 0.014019713590954115,
      "grad_norm": 0.90625,
      "learning_rate": 0.00012440864086408641,
      "loss": 3.1624,
      "step": 378020
    },
    {
      "epoch": 0.014038356827112299,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012440664066406642,
      "loss": 3.2388,
      "step": 378030
    },
    {
      "epoch": 0.014057000063270483,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001244046404640464,
      "loss": 3.2209,
      "step": 378040
    },
    {
      "epoch": 0.014075643299428666,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012440264026402641,
      "loss": 3.3118,
      "step": 378050
    },
    {
      "epoch": 0.01409428653558685,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001244006400640064,
      "loss": 3.2807,
      "step": 378060
    },
    {
      "epoch": 0.014112929771745034,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001243986398639864,
      "loss": 3.0777,
      "step": 378070
    },
    {
      "epoch": 0.014131573007903218,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001243966396639664,
      "loss": 3.2563,
      "step": 378080
    },
    {
      "epoch": 0.014150216244061401,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001243946394639464,
      "loss": 3.2015,
      "step": 378090
    },
    {
      "epoch": 0.014168859480219585,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012439263926392638,
      "loss": 3.2293,
      "step": 378100
    },
    {
      "epoch": 0.014187502716377769,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001243906390639064,
      "loss": 3.1525,
      "step": 378110
    },
    {
      "epoch": 0.014206145952535953,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001243886388638864,
      "loss": 3.239,
      "step": 378120
    },
    {
      "epoch": 0.014224789188694137,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001243866386638664,
      "loss": 3.1192,
      "step": 378130
    },
    {
      "epoch": 0.014243432424852319,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001243846384638464,
      "loss": 3.2483,
      "step": 378140
    },
    {
      "epoch": 0.014262075661010502,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001243826382638264,
      "loss": 3.2238,
      "step": 378150
    },
    {
      "epoch": 0.014280718897168686,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012438063806380638,
      "loss": 3.0551,
      "step": 378160
    },
    {
      "epoch": 0.01429936213332687,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001243786378637864,
      "loss": 3.2094,
      "step": 378170
    },
    {
      "epoch": 0.014318005369485054,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001243766376637664,
      "loss": 3.1546,
      "step": 378180
    },
    {
      "epoch": 0.014336648605643237,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012437463746374638,
      "loss": 3.2361,
      "step": 378190
    },
    {
      "epoch": 0.014355291841801421,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001243726372637264,
      "loss": 3.2773,
      "step": 378200
    },
    {
      "epoch": 0.014373935077959605,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012437063706370637,
      "loss": 3.2254,
      "step": 378210
    },
    {
      "epoch": 0.014392578314117789,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012436863686368638,
      "loss": 3.1005,
      "step": 378220
    },
    {
      "epoch": 0.014411221550275972,
      "grad_norm": 0.75,
      "learning_rate": 0.00012436663666366636,
      "loss": 3.3395,
      "step": 378230
    },
    {
      "epoch": 0.014429864786434156,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012436463646364637,
      "loss": 3.1548,
      "step": 378240
    },
    {
      "epoch": 0.01444850802259234,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012436263626362635,
      "loss": 3.1632,
      "step": 378250
    },
    {
      "epoch": 0.014467151258750524,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012436063606360636,
      "loss": 3.1108,
      "step": 378260
    },
    {
      "epoch": 0.014485794494908708,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012435863586358637,
      "loss": 3.1964,
      "step": 378270
    },
    {
      "epoch": 0.014504437731066891,
      "grad_norm": 0.75,
      "learning_rate": 0.00012435663566356638,
      "loss": 3.257,
      "step": 378280
    },
    {
      "epoch": 0.014523080967225075,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012435463546354636,
      "loss": 3.294,
      "step": 378290
    },
    {
      "epoch": 0.014541724203383259,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012435263526352637,
      "loss": 3.305,
      "step": 378300
    },
    {
      "epoch": 0.01456036743954144,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012435063506350635,
      "loss": 3.1839,
      "step": 378310
    },
    {
      "epoch": 0.014579010675699625,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012434863486348636,
      "loss": 3.332,
      "step": 378320
    },
    {
      "epoch": 0.014597653911857808,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012434663466346634,
      "loss": 3.1097,
      "step": 378330
    },
    {
      "epoch": 0.014616297148015992,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012434463446344635,
      "loss": 3.1705,
      "step": 378340
    },
    {
      "epoch": 0.014634940384174176,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012434263426342634,
      "loss": 3.1644,
      "step": 378350
    },
    {
      "epoch": 0.01465358362033236,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012434063406340635,
      "loss": 3.164,
      "step": 378360
    },
    {
      "epoch": 0.014672226856490543,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012433863386338635,
      "loss": 3.3547,
      "step": 378370
    },
    {
      "epoch": 0.014690870092648727,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012433663366336634,
      "loss": 3.1302,
      "step": 378380
    },
    {
      "epoch": 0.014709513328806911,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012433463346334635,
      "loss": 3.2315,
      "step": 378390
    },
    {
      "epoch": 0.014728156564965095,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012433263326332633,
      "loss": 3.2246,
      "step": 378400
    },
    {
      "epoch": 0.014746799801123278,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012433063306330634,
      "loss": 3.2327,
      "step": 378410
    },
    {
      "epoch": 0.014765443037281462,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012432863286328632,
      "loss": 3.2666,
      "step": 378420
    },
    {
      "epoch": 0.014784086273439646,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012432663266326633,
      "loss": 3.344,
      "step": 378430
    },
    {
      "epoch": 0.01480272950959783,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012432463246324634,
      "loss": 3.1337,
      "step": 378440
    },
    {
      "epoch": 0.014821372745756014,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012432263226322635,
      "loss": 3.2923,
      "step": 378450
    },
    {
      "epoch": 0.014840015981914197,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012432063206320633,
      "loss": 3.4011,
      "step": 378460
    },
    {
      "epoch": 0.014858659218072381,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012431863186318634,
      "loss": 3.3118,
      "step": 378470
    },
    {
      "epoch": 0.014877302454230563,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012431663166316632,
      "loss": 3.308,
      "step": 378480
    },
    {
      "epoch": 0.014895945690388747,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012431463146314633,
      "loss": 3.1569,
      "step": 378490
    },
    {
      "epoch": 0.01491458892654693,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001243126312631263,
      "loss": 3.241,
      "step": 378500
    },
    {
      "epoch": 0.014933232162705114,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012431063106310632,
      "loss": 3.1589,
      "step": 378510
    },
    {
      "epoch": 0.014951875398863298,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001243086308630863,
      "loss": 3.2267,
      "step": 378520
    },
    {
      "epoch": 0.014970518635021482,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001243066306630663,
      "loss": 3.2557,
      "step": 378530
    },
    {
      "epoch": 0.014989161871179666,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001243046304630463,
      "loss": 3.2072,
      "step": 378540
    },
    {
      "epoch": 0.01500780510733785,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001243026302630263,
      "loss": 3.1661,
      "step": 378550
    },
    {
      "epoch": 0.015026448343496033,
      "grad_norm": 0.75,
      "learning_rate": 0.00012430063006300628,
      "loss": 3.1155,
      "step": 378560
    },
    {
      "epoch": 0.015045091579654217,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001242986298629863,
      "loss": 3.17,
      "step": 378570
    },
    {
      "epoch": 0.0150637348158124,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001242966296629663,
      "loss": 3.1373,
      "step": 378580
    },
    {
      "epoch": 0.015082378051970585,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001242946294629463,
      "loss": 3.2829,
      "step": 378590
    },
    {
      "epoch": 0.015101021288128768,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012429262926292632,
      "loss": 3.1988,
      "step": 378600
    },
    {
      "epoch": 0.015119664524286952,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001242906290629063,
      "loss": 3.3796,
      "step": 378610
    },
    {
      "epoch": 0.015138307760445136,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001242886288628863,
      "loss": 3.2213,
      "step": 378620
    },
    {
      "epoch": 0.01515695099660332,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001242866286628663,
      "loss": 3.1659,
      "step": 378630
    },
    {
      "epoch": 0.015175594232761503,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001242846284628463,
      "loss": 3.224,
      "step": 378640
    },
    {
      "epoch": 0.015194237468919685,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012428262826282629,
      "loss": 3.2573,
      "step": 378650
    },
    {
      "epoch": 0.01521288070507787,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001242806280628063,
      "loss": 3.1452,
      "step": 378660
    },
    {
      "epoch": 0.015231523941236053,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012427862786278628,
      "loss": 3.0687,
      "step": 378670
    },
    {
      "epoch": 0.015250167177394237,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012427662766276629,
      "loss": 3.2847,
      "step": 378680
    },
    {
      "epoch": 0.01526881041355242,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012427462746274627,
      "loss": 3.2505,
      "step": 378690
    },
    {
      "epoch": 0.015287453649710604,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012427262726272628,
      "loss": 3.3138,
      "step": 378700
    },
    {
      "epoch": 0.015306096885868788,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012427062706270626,
      "loss": 3.0713,
      "step": 378710
    },
    {
      "epoch": 0.015324740122026972,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012426862686268627,
      "loss": 3.2099,
      "step": 378720
    },
    {
      "epoch": 0.015343383358185156,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012426662666266628,
      "loss": 3.1818,
      "step": 378730
    },
    {
      "epoch": 0.01536202659434334,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012426462646264626,
      "loss": 3.1861,
      "step": 378740
    },
    {
      "epoch": 0.015380669830501523,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012426262626262627,
      "loss": 3.1524,
      "step": 378750
    },
    {
      "epoch": 0.015399313066659707,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012426062606260628,
      "loss": 3.1898,
      "step": 378760
    },
    {
      "epoch": 0.01541795630281789,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012425862586258629,
      "loss": 3.1891,
      "step": 378770
    },
    {
      "epoch": 0.015436599538976074,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012425662566256627,
      "loss": 3.2119,
      "step": 378780
    },
    {
      "epoch": 0.015455242775134258,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012425462546254628,
      "loss": 3.2317,
      "step": 378790
    },
    {
      "epoch": 0.015473886011292442,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012425262526252626,
      "loss": 3.0845,
      "step": 378800
    },
    {
      "epoch": 0.015492529247450626,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012425062506250627,
      "loss": 3.1864,
      "step": 378810
    },
    {
      "epoch": 0.015511172483608808,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012424862486248625,
      "loss": 3.2377,
      "step": 378820
    },
    {
      "epoch": 0.015529815719766991,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012424662466246626,
      "loss": 3.1603,
      "step": 378830
    },
    {
      "epoch": 0.015548458955925175,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012424462446244624,
      "loss": 3.2928,
      "step": 378840
    },
    {
      "epoch": 0.015567102192083359,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012424262426242625,
      "loss": 3.2316,
      "step": 378850
    },
    {
      "epoch": 0.015585745428241543,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012424062406240623,
      "loss": 3.2869,
      "step": 378860
    },
    {
      "epoch": 0.015604388664399726,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012423862386238624,
      "loss": 3.1943,
      "step": 378870
    },
    {
      "epoch": 0.01562303190055791,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012423662366236622,
      "loss": 3.1991,
      "step": 378880
    },
    {
      "epoch": 0.015641675136716094,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012423462346234623,
      "loss": 3.1525,
      "step": 378890
    },
    {
      "epoch": 0.01566031837287428,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012423262326232624,
      "loss": 3.1929,
      "step": 378900
    },
    {
      "epoch": 0.01567896160903246,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012423062306230625,
      "loss": 3.2585,
      "step": 378910
    },
    {
      "epoch": 0.015697604845190644,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012422862286228623,
      "loss": 3.0918,
      "step": 378920
    },
    {
      "epoch": 0.01571624808134883,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012422662266226624,
      "loss": 3.2378,
      "step": 378930
    },
    {
      "epoch": 0.01573489131750701,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012422462246224623,
      "loss": 3.1149,
      "step": 378940
    },
    {
      "epoch": 0.015753534553665197,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012422262226222623,
      "loss": 3.1286,
      "step": 378950
    },
    {
      "epoch": 0.01577217778982338,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012422062206220622,
      "loss": 3.2946,
      "step": 378960
    },
    {
      "epoch": 0.015790821025981564,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012421862186218623,
      "loss": 3.3232,
      "step": 378970
    },
    {
      "epoch": 0.015809464262139746,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012421662166216623,
      "loss": 3.1428,
      "step": 378980
    },
    {
      "epoch": 0.01582810749829793,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012421462146214622,
      "loss": 3.1374,
      "step": 378990
    },
    {
      "epoch": 0.015846750734456114,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012421262126212623,
      "loss": 3.1993,
      "step": 379000
    },
    {
      "epoch": 0.0158653939706143,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001242106210621062,
      "loss": 3.1942,
      "step": 379010
    },
    {
      "epoch": 0.01588403720677248,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012420862086208622,
      "loss": 3.1874,
      "step": 379020
    },
    {
      "epoch": 0.015902680442930667,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001242066206620662,
      "loss": 3.2207,
      "step": 379030
    },
    {
      "epoch": 0.01592132367908885,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001242046204620462,
      "loss": 3.1103,
      "step": 379040
    },
    {
      "epoch": 0.015939966915247034,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001242026202620262,
      "loss": 3.1688,
      "step": 379050
    },
    {
      "epoch": 0.015958610151405216,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001242006200620062,
      "loss": 3.0722,
      "step": 379060
    },
    {
      "epoch": 0.015977253387563402,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001241986198619862,
      "loss": 3.2013,
      "step": 379070
    },
    {
      "epoch": 0.015995896623721584,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012419661966196622,
      "loss": 3.1175,
      "step": 379080
    },
    {
      "epoch": 0.016014539859879766,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001241946194619462,
      "loss": 3.1991,
      "step": 379090
    },
    {
      "epoch": 0.01603318309603795,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001241926192619262,
      "loss": 3.3182,
      "step": 379100
    },
    {
      "epoch": 0.016051826332196133,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001241906190619062,
      "loss": 3.3152,
      "step": 379110
    },
    {
      "epoch": 0.01607046956835432,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001241886188618862,
      "loss": 3.1921,
      "step": 379120
    },
    {
      "epoch": 0.0160891128045125,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012418661866186618,
      "loss": 3.1038,
      "step": 379130
    },
    {
      "epoch": 0.016107756040670686,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001241846184618462,
      "loss": 3.3114,
      "step": 379140
    },
    {
      "epoch": 0.01612639927682887,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012418261826182617,
      "loss": 3.2011,
      "step": 379150
    },
    {
      "epoch": 0.016145042512987054,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012418061806180618,
      "loss": 3.2545,
      "step": 379160
    },
    {
      "epoch": 0.016163685749145236,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001241786178617862,
      "loss": 3.0832,
      "step": 379170
    },
    {
      "epoch": 0.01618232898530342,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012417661766176617,
      "loss": 3.1672,
      "step": 379180
    },
    {
      "epoch": 0.016200972221461604,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012417461746174618,
      "loss": 3.1928,
      "step": 379190
    },
    {
      "epoch": 0.01621961545761979,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012417261726172616,
      "loss": 3.4074,
      "step": 379200
    },
    {
      "epoch": 0.01623825869377797,
      "grad_norm": 0.75,
      "learning_rate": 0.00012417061706170617,
      "loss": 3.2389,
      "step": 379210
    },
    {
      "epoch": 0.016256901929936157,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012416861686168618,
      "loss": 3.2629,
      "step": 379220
    },
    {
      "epoch": 0.01627554516609434,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001241666166616662,
      "loss": 3.2181,
      "step": 379230
    },
    {
      "epoch": 0.016294188402252524,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012416461646164617,
      "loss": 3.1952,
      "step": 379240
    },
    {
      "epoch": 0.016312831638410706,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012416261626162618,
      "loss": 3.2007,
      "step": 379250
    },
    {
      "epoch": 0.016331474874568888,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012416061606160617,
      "loss": 3.2793,
      "step": 379260
    },
    {
      "epoch": 0.016350118110727074,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012415861586158617,
      "loss": 3.2677,
      "step": 379270
    },
    {
      "epoch": 0.016368761346885256,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012415661566156616,
      "loss": 3.0825,
      "step": 379280
    },
    {
      "epoch": 0.01638740458304344,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012415461546154617,
      "loss": 3.2491,
      "step": 379290
    },
    {
      "epoch": 0.016406047819201623,
      "grad_norm": 0.859375,
      "learning_rate": 0.00012415261526152615,
      "loss": 3.3493,
      "step": 379300
    },
    {
      "epoch": 0.01642469105535981,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012415061506150616,
      "loss": 3.2193,
      "step": 379310
    },
    {
      "epoch": 0.01644333429151799,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012414861486148614,
      "loss": 3.1268,
      "step": 379320
    },
    {
      "epoch": 0.016461977527676176,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012414661466146615,
      "loss": 3.2766,
      "step": 379330
    },
    {
      "epoch": 0.016480620763834358,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012414461446144613,
      "loss": 3.2451,
      "step": 379340
    },
    {
      "epoch": 0.016499263999992544,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012414261426142614,
      "loss": 3.2253,
      "step": 379350
    },
    {
      "epoch": 0.016517907236150726,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012414061406140615,
      "loss": 3.2091,
      "step": 379360
    },
    {
      "epoch": 0.01653655047230891,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012413861386138613,
      "loss": 3.2292,
      "step": 379370
    },
    {
      "epoch": 0.016555193708467093,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012413661366136614,
      "loss": 3.3064,
      "step": 379380
    },
    {
      "epoch": 0.01657383694462528,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012413461346134615,
      "loss": 3.1921,
      "step": 379390
    },
    {
      "epoch": 0.01659248018078346,
      "grad_norm": 0.75,
      "learning_rate": 0.00012413261326132616,
      "loss": 3.1279,
      "step": 379400
    },
    {
      "epoch": 0.016611123416941646,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012413061306130614,
      "loss": 3.1879,
      "step": 379410
    },
    {
      "epoch": 0.01662976665309983,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012412861286128615,
      "loss": 3.1917,
      "step": 379420
    },
    {
      "epoch": 0.01664840988925801,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012412661266126613,
      "loss": 3.0924,
      "step": 379430
    },
    {
      "epoch": 0.016667053125416196,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012412461246124614,
      "loss": 3.2461,
      "step": 379440
    },
    {
      "epoch": 0.016685696361574378,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012412261226122612,
      "loss": 3.2552,
      "step": 379450
    },
    {
      "epoch": 0.016704339597732563,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012412061206120613,
      "loss": 3.4151,
      "step": 379460
    },
    {
      "epoch": 0.016722982833890745,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001241186118611861,
      "loss": 3.2088,
      "step": 379470
    },
    {
      "epoch": 0.01674162607004893,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012411661166116612,
      "loss": 3.2773,
      "step": 379480
    },
    {
      "epoch": 0.016760269306207113,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001241146114611461,
      "loss": 3.2265,
      "step": 379490
    },
    {
      "epoch": 0.0167789125423653,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012411261126112611,
      "loss": 3.2391,
      "step": 379500
    },
    {
      "epoch": 0.01679755577852348,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001241106110611061,
      "loss": 3.1674,
      "step": 379510
    },
    {
      "epoch": 0.016816199014681666,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001241086108610861,
      "loss": 3.0694,
      "step": 379520
    },
    {
      "epoch": 0.016834842250839848,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012410661066106611,
      "loss": 3.258,
      "step": 379530
    },
    {
      "epoch": 0.016853485486998034,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012410461046104612,
      "loss": 3.1685,
      "step": 379540
    },
    {
      "epoch": 0.016872128723156216,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001241026102610261,
      "loss": 3.1675,
      "step": 379550
    },
    {
      "epoch": 0.0168907719593144,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012410061006100611,
      "loss": 3.2096,
      "step": 379560
    },
    {
      "epoch": 0.016909415195472583,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012409860986098612,
      "loss": 3.1172,
      "step": 379570
    },
    {
      "epoch": 0.01692805843163077,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001240966096609661,
      "loss": 3.3186,
      "step": 379580
    },
    {
      "epoch": 0.01694670166778895,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012409460946094611,
      "loss": 3.0144,
      "step": 379590
    },
    {
      "epoch": 0.016965344903947133,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001240926092609261,
      "loss": 3.1149,
      "step": 379600
    },
    {
      "epoch": 0.016983988140105318,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001240906090609061,
      "loss": 3.2359,
      "step": 379610
    },
    {
      "epoch": 0.0170026313762635,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001240886088608861,
      "loss": 3.2223,
      "step": 379620
    },
    {
      "epoch": 0.017021274612421686,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001240866086608661,
      "loss": 3.1938,
      "step": 379630
    },
    {
      "epoch": 0.017039917848579868,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012408460846084608,
      "loss": 3.2507,
      "step": 379640
    },
    {
      "epoch": 0.017058561084738053,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001240826082608261,
      "loss": 3.0874,
      "step": 379650
    },
    {
      "epoch": 0.017077204320896235,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012408060806080607,
      "loss": 3.1998,
      "step": 379660
    },
    {
      "epoch": 0.01709584755705442,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012407860786078608,
      "loss": 3.1861,
      "step": 379670
    },
    {
      "epoch": 0.017114490793212603,
      "grad_norm": 0.875,
      "learning_rate": 0.0001240766076607661,
      "loss": 3.0269,
      "step": 379680
    },
    {
      "epoch": 0.01713313402937079,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00012407460746074607,
      "loss": 3.2614,
      "step": 379690
    },
    {
      "epoch": 0.01715177726552897,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012407260726072608,
      "loss": 3.1484,
      "step": 379700
    },
    {
      "epoch": 0.017170420501687156,
      "grad_norm": 0.75,
      "learning_rate": 0.0001240706070607061,
      "loss": 3.1919,
      "step": 379710
    },
    {
      "epoch": 0.017189063737845338,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012406860686068607,
      "loss": 3.1805,
      "step": 379720
    },
    {
      "epoch": 0.017207706974003523,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012406660666066608,
      "loss": 3.1349,
      "step": 379730
    },
    {
      "epoch": 0.017226350210161705,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012406460646064606,
      "loss": 3.1484,
      "step": 379740
    },
    {
      "epoch": 0.01724499344631989,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012406260626062607,
      "loss": 3.2537,
      "step": 379750
    },
    {
      "epoch": 0.017263636682478073,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012406060606060608,
      "loss": 3.1818,
      "step": 379760
    },
    {
      "epoch": 0.017282279918636255,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012405860586058606,
      "loss": 3.1529,
      "step": 379770
    },
    {
      "epoch": 0.01730092315479444,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012405660566056607,
      "loss": 3.2054,
      "step": 379780
    },
    {
      "epoch": 0.017319566390952622,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012405460546054605,
      "loss": 3.2252,
      "step": 379790
    },
    {
      "epoch": 0.017338209627110808,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012405260526052606,
      "loss": 3.1382,
      "step": 379800
    },
    {
      "epoch": 0.01735685286326899,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012405060506050605,
      "loss": 3.2899,
      "step": 379810
    },
    {
      "epoch": 0.017375496099427176,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012404860486048605,
      "loss": 3.2934,
      "step": 379820
    },
    {
      "epoch": 0.017394139335585358,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012404660466046604,
      "loss": 3.1396,
      "step": 379830
    },
    {
      "epoch": 0.017412782571743543,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012404460446044605,
      "loss": 3.195,
      "step": 379840
    },
    {
      "epoch": 0.017431425807901725,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012404260426042605,
      "loss": 3.0727,
      "step": 379850
    },
    {
      "epoch": 0.01745006904405991,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012404060406040606,
      "loss": 3.1963,
      "step": 379860
    },
    {
      "epoch": 0.017468712280218093,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012403860386038605,
      "loss": 3.1967,
      "step": 379870
    },
    {
      "epoch": 0.017487355516376278,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012403660366036605,
      "loss": 3.1667,
      "step": 379880
    },
    {
      "epoch": 0.01750599875253446,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012403460346034604,
      "loss": 3.1688,
      "step": 379890
    },
    {
      "epoch": 0.017524641988692646,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012403260326032605,
      "loss": 3.2003,
      "step": 379900
    },
    {
      "epoch": 0.017543285224850828,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012403060306030603,
      "loss": 3.2509,
      "step": 379910
    },
    {
      "epoch": 0.017561928461009013,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012402860286028604,
      "loss": 3.2501,
      "step": 379920
    },
    {
      "epoch": 0.017580571697167195,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012402660266026602,
      "loss": 3.2967,
      "step": 379930
    },
    {
      "epoch": 0.017599214933325377,
      "grad_norm": 0.75,
      "learning_rate": 0.00012402460246024603,
      "loss": 3.2462,
      "step": 379940
    },
    {
      "epoch": 0.017617858169483563,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012402260226022604,
      "loss": 3.2367,
      "step": 379950
    },
    {
      "epoch": 0.017636501405641745,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012402060206020602,
      "loss": 3.2493,
      "step": 379960
    },
    {
      "epoch": 0.01765514464179993,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012401860186018603,
      "loss": 3.219,
      "step": 379970
    },
    {
      "epoch": 0.017673787877958112,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000124016601660166,
      "loss": 3.1318,
      "step": 379980
    },
    {
      "epoch": 0.017692431114116298,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012401460146014602,
      "loss": 3.1441,
      "step": 379990
    },
    {
      "epoch": 0.01771107435027448,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012401260126012603,
      "loss": 3.0764,
      "step": 380000
    },
    {
      "epoch": 0.017729717586432665,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000124010601060106,
      "loss": 3.2276,
      "step": 380010
    },
    {
      "epoch": 0.017748360822590847,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012400860086008602,
      "loss": 3.1564,
      "step": 380020
    },
    {
      "epoch": 0.017767004058749033,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012400660066006603,
      "loss": 3.2605,
      "step": 380030
    },
    {
      "epoch": 0.017785647294907215,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000124004600460046,
      "loss": 3.0588,
      "step": 380040
    },
    {
      "epoch": 0.0178042905310654,
      "grad_norm": 0.75,
      "learning_rate": 0.00012400260026002602,
      "loss": 3.1228,
      "step": 380050
    },
    {
      "epoch": 0.017822933767223582,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000124000600060006,
      "loss": 3.2968,
      "step": 380060
    },
    {
      "epoch": 0.017841577003381768,
      "grad_norm": 0.92578125,
      "learning_rate": 0.000123998599859986,
      "loss": 3.1717,
      "step": 380070
    },
    {
      "epoch": 0.01786022023953995,
      "grad_norm": 0.80859375,
      "learning_rate": 0.000123996599659966,
      "loss": 3.2286,
      "step": 380080
    },
    {
      "epoch": 0.017878863475698135,
      "grad_norm": 0.78125,
      "learning_rate": 0.000123994599459946,
      "loss": 3.2846,
      "step": 380090
    },
    {
      "epoch": 0.017897506711856317,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012399259925992599,
      "loss": 3.2303,
      "step": 380100
    },
    {
      "epoch": 0.0179161499480145,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000123990599059906,
      "loss": 3.3,
      "step": 380110
    },
    {
      "epoch": 0.017934793184172685,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012398859885988598,
      "loss": 3.1862,
      "step": 380120
    },
    {
      "epoch": 0.017953436420330867,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012398659865986599,
      "loss": 3.3381,
      "step": 380130
    },
    {
      "epoch": 0.017972079656489053,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012398459845984597,
      "loss": 3.1776,
      "step": 380140
    },
    {
      "epoch": 0.017990722892647235,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012398259825982598,
      "loss": 3.1452,
      "step": 380150
    },
    {
      "epoch": 0.01800936612880542,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012398059805980599,
      "loss": 3.3056,
      "step": 380160
    },
    {
      "epoch": 0.018028009364963602,
      "grad_norm": 0.796875,
      "learning_rate": 0.000123978597859786,
      "loss": 3.226,
      "step": 380170
    },
    {
      "epoch": 0.018046652601121788,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000123976597659766,
      "loss": 3.1783,
      "step": 380180
    },
    {
      "epoch": 0.01806529583727997,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012397459745974599,
      "loss": 3.1692,
      "step": 380190
    },
    {
      "epoch": 0.018083939073438155,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000123972597259726,
      "loss": 3.2363,
      "step": 380200
    },
    {
      "epoch": 0.018102582309596337,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012397059705970598,
      "loss": 3.0694,
      "step": 380210
    },
    {
      "epoch": 0.018121225545754523,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012396859685968599,
      "loss": 3.2659,
      "step": 380220
    },
    {
      "epoch": 0.018139868781912705,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012396659665966597,
      "loss": 3.2089,
      "step": 380230
    },
    {
      "epoch": 0.01815851201807089,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012396459645964598,
      "loss": 3.2938,
      "step": 380240
    },
    {
      "epoch": 0.018177155254229072,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012396259625962596,
      "loss": 3.1918,
      "step": 380250
    },
    {
      "epoch": 0.018195798490387258,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012396059605960597,
      "loss": 3.188,
      "step": 380260
    },
    {
      "epoch": 0.01821444172654544,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012395859585958595,
      "loss": 3.1756,
      "step": 380270
    },
    {
      "epoch": 0.018233084962703622,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012395659565956596,
      "loss": 3.0643,
      "step": 380280
    },
    {
      "epoch": 0.018251728198861807,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012395459545954594,
      "loss": 3.2094,
      "step": 380290
    },
    {
      "epoch": 0.01827037143501999,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012395259525952595,
      "loss": 3.1458,
      "step": 380300
    },
    {
      "epoch": 0.018289014671178175,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012395059505950596,
      "loss": 3.2512,
      "step": 380310
    },
    {
      "epoch": 0.018307657907336357,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012394859485948594,
      "loss": 3.1313,
      "step": 380320
    },
    {
      "epoch": 0.018326301143494542,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012394659465946595,
      "loss": 3.1777,
      "step": 380330
    },
    {
      "epoch": 0.018344944379652724,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012394459445944596,
      "loss": 3.0856,
      "step": 380340
    },
    {
      "epoch": 0.01836358761581091,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012394259425942597,
      "loss": 3.1936,
      "step": 380350
    },
    {
      "epoch": 0.018382230851969092,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012394059405940595,
      "loss": 3.1976,
      "step": 380360
    },
    {
      "epoch": 0.018400874088127277,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012393859385938596,
      "loss": 3.2388,
      "step": 380370
    },
    {
      "epoch": 0.01841951732428546,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012393659365936594,
      "loss": 3.1955,
      "step": 380380
    },
    {
      "epoch": 0.018438160560443645,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012393459345934595,
      "loss": 3.3171,
      "step": 380390
    },
    {
      "epoch": 0.018456803796601827,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012393259325932593,
      "loss": 3.3841,
      "step": 380400
    },
    {
      "epoch": 0.018475447032760012,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012393059305930594,
      "loss": 3.1744,
      "step": 380410
    },
    {
      "epoch": 0.018494090268918194,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012392859285928593,
      "loss": 3.2039,
      "step": 380420
    },
    {
      "epoch": 0.01851273350507638,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012392659265926593,
      "loss": 3.2771,
      "step": 380430
    },
    {
      "epoch": 0.018531376741234562,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012392459245924592,
      "loss": 3.1672,
      "step": 380440
    },
    {
      "epoch": 0.018550019977392744,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012392259225922593,
      "loss": 3.0312,
      "step": 380450
    },
    {
      "epoch": 0.01856866321355093,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001239205920592059,
      "loss": 3.1077,
      "step": 380460
    },
    {
      "epoch": 0.01858730644970911,
      "grad_norm": 0.75,
      "learning_rate": 0.00012391859185918592,
      "loss": 3.235,
      "step": 380470
    },
    {
      "epoch": 0.018605949685867297,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012391659165916593,
      "loss": 3.2128,
      "step": 380480
    },
    {
      "epoch": 0.01862459292202548,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00012391459145914594,
      "loss": 3.1174,
      "step": 380490
    },
    {
      "epoch": 0.018643236158183665,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012391259125912592,
      "loss": 3.1812,
      "step": 380500
    },
    {
      "epoch": 0.018661879394341847,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012391059105910593,
      "loss": 3.331,
      "step": 380510
    },
    {
      "epoch": 0.018680522630500032,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001239085908590859,
      "loss": 3.1654,
      "step": 380520
    },
    {
      "epoch": 0.018699165866658214,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012390659065906592,
      "loss": 3.1716,
      "step": 380530
    },
    {
      "epoch": 0.0187178091028164,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001239045904590459,
      "loss": 3.0707,
      "step": 380540
    },
    {
      "epoch": 0.01873645233897458,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001239025902590259,
      "loss": 3.2707,
      "step": 380550
    },
    {
      "epoch": 0.018755095575132767,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012390059005900592,
      "loss": 3.2114,
      "step": 380560
    },
    {
      "epoch": 0.01877373881129095,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001238985898589859,
      "loss": 3.1982,
      "step": 380570
    },
    {
      "epoch": 0.018792382047449135,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001238965896589659,
      "loss": 3.0578,
      "step": 380580
    },
    {
      "epoch": 0.018811025283607317,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001238945894589459,
      "loss": 3.2064,
      "step": 380590
    },
    {
      "epoch": 0.018829668519765502,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001238925892589259,
      "loss": 3.2825,
      "step": 380600
    },
    {
      "epoch": 0.018848311755923684,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012389058905890588,
      "loss": 3.0485,
      "step": 380610
    },
    {
      "epoch": 0.018866954992081866,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001238885888588859,
      "loss": 3.2297,
      "step": 380620
    },
    {
      "epoch": 0.018885598228240052,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001238865886588659,
      "loss": 3.2667,
      "step": 380630
    },
    {
      "epoch": 0.018904241464398234,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012388458845884588,
      "loss": 3.2769,
      "step": 380640
    },
    {
      "epoch": 0.01892288470055642,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001238825882588259,
      "loss": 3.2374,
      "step": 380650
    },
    {
      "epoch": 0.0189415279367146,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001238805880588059,
      "loss": 3.2507,
      "step": 380660
    },
    {
      "epoch": 0.018960171172872787,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012387858785878588,
      "loss": 3.0299,
      "step": 380670
    },
    {
      "epoch": 0.01897881440903097,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001238765876587659,
      "loss": 3.1271,
      "step": 380680
    },
    {
      "epoch": 0.018997457645189154,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012387458745874587,
      "loss": 3.1417,
      "step": 380690
    },
    {
      "epoch": 0.019016100881347336,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012387258725872588,
      "loss": 3.175,
      "step": 380700
    },
    {
      "epoch": 0.019034744117505522,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012387058705870587,
      "loss": 3.3763,
      "step": 380710
    },
    {
      "epoch": 0.019053387353663704,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012386858685868587,
      "loss": 3.1033,
      "step": 380720
    },
    {
      "epoch": 0.01907203058982189,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012386658665866586,
      "loss": 3.1833,
      "step": 380730
    },
    {
      "epoch": 0.01909067382598007,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00012386458645864587,
      "loss": 3.1498,
      "step": 380740
    },
    {
      "epoch": 0.019109317062138257,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012386258625862587,
      "loss": 3.1191,
      "step": 380750
    },
    {
      "epoch": 0.01912796029829644,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012386058605860586,
      "loss": 3.2407,
      "step": 380760
    },
    {
      "epoch": 0.019146603534454625,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012385858585858587,
      "loss": 3.173,
      "step": 380770
    },
    {
      "epoch": 0.019165246770612807,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012385658565856585,
      "loss": 3.1912,
      "step": 380780
    },
    {
      "epoch": 0.01918389000677099,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012385458545854586,
      "loss": 3.2141,
      "step": 380790
    },
    {
      "epoch": 0.019202533242929174,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012385258525852587,
      "loss": 3.2475,
      "step": 380800
    },
    {
      "epoch": 0.019221176479087356,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012385058505850588,
      "loss": 3.0454,
      "step": 380810
    },
    {
      "epoch": 0.01923981971524554,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012384858485848586,
      "loss": 3.2099,
      "step": 380820
    },
    {
      "epoch": 0.019258462951403724,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012384658465846587,
      "loss": 3.2117,
      "step": 380830
    },
    {
      "epoch": 0.01927710618756191,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012384458445844585,
      "loss": 3.1524,
      "step": 380840
    },
    {
      "epoch": 0.01929574942372009,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012384258425842586,
      "loss": 3.2727,
      "step": 380850
    },
    {
      "epoch": 0.019314392659878277,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012384058405840584,
      "loss": 3.1054,
      "step": 380860
    },
    {
      "epoch": 0.01933303589603646,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012383858385838585,
      "loss": 3.263,
      "step": 380870
    },
    {
      "epoch": 0.019351679132194644,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012383658365836583,
      "loss": 3.2602,
      "step": 380880
    },
    {
      "epoch": 0.019370322368352826,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012383458345834584,
      "loss": 3.1738,
      "step": 380890
    },
    {
      "epoch": 0.019388965604511012,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012383258325832582,
      "loss": 3.2361,
      "step": 380900
    },
    {
      "epoch": 0.019407608840669194,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012383058305830583,
      "loss": 3.2018,
      "step": 380910
    },
    {
      "epoch": 0.01942625207682738,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001238285828582858,
      "loss": 3.2672,
      "step": 380920
    },
    {
      "epoch": 0.01944489531298556,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012382658265826582,
      "loss": 3.253,
      "step": 380930
    },
    {
      "epoch": 0.019463538549143747,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012382458245824583,
      "loss": 3.1751,
      "step": 380940
    },
    {
      "epoch": 0.01948218178530193,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012382258225822584,
      "loss": 3.1683,
      "step": 380950
    },
    {
      "epoch": 0.019500825021460114,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012382058205820582,
      "loss": 3.1823,
      "step": 380960
    },
    {
      "epoch": 0.019519468257618296,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012381858185818583,
      "loss": 3.1877,
      "step": 380970
    },
    {
      "epoch": 0.01953811149377648,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012381658165816584,
      "loss": 2.9946,
      "step": 380980
    },
    {
      "epoch": 0.019556754729934664,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012381458145814582,
      "loss": 3.2772,
      "step": 380990
    },
    {
      "epoch": 0.019575397966092846,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012381258125812583,
      "loss": 3.101,
      "step": 381000
    },
    {
      "epoch": 0.01959404120225103,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012381058105810581,
      "loss": 3.2272,
      "step": 381010
    },
    {
      "epoch": 0.019612684438409213,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012380858085808582,
      "loss": 3.1849,
      "step": 381020
    },
    {
      "epoch": 0.0196313276745674,
      "grad_norm": 0.75,
      "learning_rate": 0.0001238065806580658,
      "loss": 3.2406,
      "step": 381030
    },
    {
      "epoch": 0.01964997091072558,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012380458045804581,
      "loss": 3.2413,
      "step": 381040
    },
    {
      "epoch": 0.019668614146883766,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001238025802580258,
      "loss": 3.2694,
      "step": 381050
    },
    {
      "epoch": 0.01968725738304195,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001238005800580058,
      "loss": 3.2774,
      "step": 381060
    },
    {
      "epoch": 0.019705900619200134,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001237985798579858,
      "loss": 3.1042,
      "step": 381070
    },
    {
      "epoch": 0.019724543855358316,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001237965796579658,
      "loss": 3.2343,
      "step": 381080
    },
    {
      "epoch": 0.0197431870915165,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012379457945794578,
      "loss": 3.229,
      "step": 381090
    },
    {
      "epoch": 0.019761830327674684,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001237925792579258,
      "loss": 3.1693,
      "step": 381100
    },
    {
      "epoch": 0.01978047356383287,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001237905790579058,
      "loss": 3.216,
      "step": 381110
    },
    {
      "epoch": 0.01979911679999105,
      "grad_norm": 0.75,
      "learning_rate": 0.0001237885788578858,
      "loss": 3.1929,
      "step": 381120
    },
    {
      "epoch": 0.019817760036149237,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001237865786578658,
      "loss": 3.1393,
      "step": 381130
    },
    {
      "epoch": 0.01983640327230742,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001237845784578458,
      "loss": 3.1229,
      "step": 381140
    },
    {
      "epoch": 0.0198550465084656,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001237825782578258,
      "loss": 3.2062,
      "step": 381150
    },
    {
      "epoch": 0.019873689744623786,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001237805780578058,
      "loss": 3.0545,
      "step": 381160
    },
    {
      "epoch": 0.019892332980781968,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001237785778577858,
      "loss": 3.2396,
      "step": 381170
    },
    {
      "epoch": 0.019910976216940154,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012377657765776578,
      "loss": 3.1577,
      "step": 381180
    },
    {
      "epoch": 0.019929619453098336,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001237745774577458,
      "loss": 3.2623,
      "step": 381190
    },
    {
      "epoch": 0.01994826268925652,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012377257725772577,
      "loss": 3.1981,
      "step": 381200
    },
    {
      "epoch": 0.019966905925414703,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012377057705770578,
      "loss": 3.189,
      "step": 381210
    },
    {
      "epoch": 0.01998554916157289,
      "grad_norm": 0.75,
      "learning_rate": 0.00012376857685768576,
      "loss": 3.2781,
      "step": 381220
    },
    {
      "epoch": 0.02000419239773107,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012376657665766577,
      "loss": 3.2098,
      "step": 381230
    },
    {
      "epoch": 0.020022835633889256,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012376457645764575,
      "loss": 3.1053,
      "step": 381240
    },
    {
      "epoch": 0.02004147887004744,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012376257625762576,
      "loss": 3.1539,
      "step": 381250
    },
    {
      "epoch": 0.020060122106205624,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012376057605760577,
      "loss": 3.1977,
      "step": 381260
    },
    {
      "epoch": 0.020078765342363806,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012375857585758578,
      "loss": 3.1624,
      "step": 381270
    },
    {
      "epoch": 0.02009740857852199,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012375657565756576,
      "loss": 3.158,
      "step": 381280
    },
    {
      "epoch": 0.020116051814680173,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012375457545754577,
      "loss": 3.0916,
      "step": 381290
    },
    {
      "epoch": 0.02013469505083836,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012375257525752575,
      "loss": 2.9893,
      "step": 381300
    },
    {
      "epoch": 0.02015333828699654,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012375057505750576,
      "loss": 3.2039,
      "step": 381310
    },
    {
      "epoch": 0.020171981523154723,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00012374857485748575,
      "loss": 3.1497,
      "step": 381320
    },
    {
      "epoch": 0.02019062475931291,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012374657465746575,
      "loss": 3.1396,
      "step": 381330
    },
    {
      "epoch": 0.02020926799547109,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012374457445744576,
      "loss": 3.2493,
      "step": 381340
    },
    {
      "epoch": 0.020227911231629276,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012374257425742575,
      "loss": 3.0288,
      "step": 381350
    },
    {
      "epoch": 0.020246554467787458,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012374057405740576,
      "loss": 3.1236,
      "step": 381360
    },
    {
      "epoch": 0.020265197703945644,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012373857385738574,
      "loss": 3.2587,
      "step": 381370
    },
    {
      "epoch": 0.020283840940103826,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012373657365736575,
      "loss": 3.3121,
      "step": 381380
    },
    {
      "epoch": 0.02030248417626201,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012373457345734573,
      "loss": 3.2148,
      "step": 381390
    },
    {
      "epoch": 0.020321127412420193,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012373257325732574,
      "loss": 3.2431,
      "step": 381400
    },
    {
      "epoch": 0.02033977064857838,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012373057305730572,
      "loss": 3.2381,
      "step": 381410
    },
    {
      "epoch": 0.02035841388473656,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012372857285728573,
      "loss": 3.1244,
      "step": 381420
    },
    {
      "epoch": 0.020377057120894746,
      "grad_norm": 0.75,
      "learning_rate": 0.00012372657265726574,
      "loss": 3.0229,
      "step": 381430
    },
    {
      "epoch": 0.020395700357052928,
      "grad_norm": 0.875,
      "learning_rate": 0.00012372457245724575,
      "loss": 3.2711,
      "step": 381440
    },
    {
      "epoch": 0.020414343593211114,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012372257225722573,
      "loss": 3.1771,
      "step": 381450
    },
    {
      "epoch": 0.020432986829369296,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012372057205720574,
      "loss": 3.1918,
      "step": 381460
    },
    {
      "epoch": 0.02045163006552748,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012371857185718572,
      "loss": 3.2105,
      "step": 381470
    },
    {
      "epoch": 0.020470273301685663,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012371657165716573,
      "loss": 3.2239,
      "step": 381480
    },
    {
      "epoch": 0.020488916537843845,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001237145714571457,
      "loss": 3.2072,
      "step": 381490
    },
    {
      "epoch": 0.02050755977400203,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012371257125712572,
      "loss": 3.1821,
      "step": 381500
    },
    {
      "epoch": 0.020526203010160213,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001237105710571057,
      "loss": 3.1056,
      "step": 381510
    },
    {
      "epoch": 0.020544846246318398,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001237085708570857,
      "loss": 3.1807,
      "step": 381520
    },
    {
      "epoch": 0.02056348948247658,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012370657065706572,
      "loss": 3.2282,
      "step": 381530
    },
    {
      "epoch": 0.020582132718634766,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001237045704570457,
      "loss": 3.1768,
      "step": 381540
    },
    {
      "epoch": 0.020600775954792948,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001237025702570257,
      "loss": 3.1423,
      "step": 381550
    },
    {
      "epoch": 0.020619419190951133,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001237005700570057,
      "loss": 3.1992,
      "step": 381560
    },
    {
      "epoch": 0.020638062427109315,
      "grad_norm": 0.875,
      "learning_rate": 0.0001236985698569857,
      "loss": 3.0845,
      "step": 381570
    },
    {
      "epoch": 0.0206567056632675,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001236965696569657,
      "loss": 3.1677,
      "step": 381580
    },
    {
      "epoch": 0.020675348899425683,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012369456945694572,
      "loss": 3.137,
      "step": 381590
    },
    {
      "epoch": 0.02069399213558387,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001236925692569257,
      "loss": 3.107,
      "step": 381600
    },
    {
      "epoch": 0.02071263537174205,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001236905690569057,
      "loss": 3.1704,
      "step": 381610
    },
    {
      "epoch": 0.020731278607900236,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001236885688568857,
      "loss": 3.226,
      "step": 381620
    },
    {
      "epoch": 0.020749921844058418,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001236865686568657,
      "loss": 3.2045,
      "step": 381630
    },
    {
      "epoch": 0.020768565080216603,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012368456845684569,
      "loss": 3.2309,
      "step": 381640
    },
    {
      "epoch": 0.020787208316374785,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001236825682568257,
      "loss": 3.0946,
      "step": 381650
    },
    {
      "epoch": 0.020805851552532968,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012368056805680568,
      "loss": 3.1615,
      "step": 381660
    },
    {
      "epoch": 0.020824494788691153,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012367856785678569,
      "loss": 3.1824,
      "step": 381670
    },
    {
      "epoch": 0.020843138024849335,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012367656765676567,
      "loss": 3.1299,
      "step": 381680
    },
    {
      "epoch": 0.02086178126100752,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012367456745674568,
      "loss": 3.1838,
      "step": 381690
    },
    {
      "epoch": 0.020880424497165703,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012367256725672566,
      "loss": 3.3065,
      "step": 381700
    },
    {
      "epoch": 0.020899067733323888,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012367056705670567,
      "loss": 3.2053,
      "step": 381710
    },
    {
      "epoch": 0.02091771096948207,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012366856685668565,
      "loss": 3.2812,
      "step": 381720
    },
    {
      "epoch": 0.020936354205640256,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012366656665666566,
      "loss": 3.2737,
      "step": 381730
    },
    {
      "epoch": 0.020954997441798438,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012366456645664567,
      "loss": 3.2782,
      "step": 381740
    },
    {
      "epoch": 0.020973640677956623,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012366256625662568,
      "loss": 3.1185,
      "step": 381750
    },
    {
      "epoch": 0.020992283914114805,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001236605660566057,
      "loss": 3.2461,
      "step": 381760
    },
    {
      "epoch": 0.02101092715027299,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012365856585658567,
      "loss": 3.2418,
      "step": 381770
    },
    {
      "epoch": 0.021029570386431173,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012365656565656568,
      "loss": 3.2155,
      "step": 381780
    },
    {
      "epoch": 0.021048213622589358,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012365456545654566,
      "loss": 3.1999,
      "step": 381790
    },
    {
      "epoch": 0.02106685685874754,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012365256525652567,
      "loss": 3.2034,
      "step": 381800
    },
    {
      "epoch": 0.021085500094905726,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012365056505650565,
      "loss": 3.2766,
      "step": 381810
    },
    {
      "epoch": 0.021104143331063908,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012364856485648566,
      "loss": 3.1934,
      "step": 381820
    },
    {
      "epoch": 0.02112278656722209,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012364656465646564,
      "loss": 3.218,
      "step": 381830
    },
    {
      "epoch": 0.021141429803380275,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012364456445644565,
      "loss": 3.1098,
      "step": 381840
    },
    {
      "epoch": 0.021160073039538457,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012364256425642563,
      "loss": 3.1941,
      "step": 381850
    },
    {
      "epoch": 0.021178716275696643,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012364056405640564,
      "loss": 3.2716,
      "step": 381860
    },
    {
      "epoch": 0.021197359511854825,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012363856385638563,
      "loss": 3.2179,
      "step": 381870
    },
    {
      "epoch": 0.02121600274801301,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012363656365636563,
      "loss": 3.1827,
      "step": 381880
    },
    {
      "epoch": 0.021234645984171192,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012363456345634564,
      "loss": 3.3393,
      "step": 381890
    },
    {
      "epoch": 0.021253289220329378,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012363256325632565,
      "loss": 3.171,
      "step": 381900
    },
    {
      "epoch": 0.02127193245648756,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012363056305630563,
      "loss": 3.187,
      "step": 381910
    },
    {
      "epoch": 0.021290575692645745,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012362856285628564,
      "loss": 3.2503,
      "step": 381920
    },
    {
      "epoch": 0.021309218928803927,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012362656265626565,
      "loss": 3.1846,
      "step": 381930
    },
    {
      "epoch": 0.021327862164962113,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012362456245624564,
      "loss": 3.2147,
      "step": 381940
    },
    {
      "epoch": 0.021346505401120295,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012362256225622564,
      "loss": 3.2322,
      "step": 381950
    },
    {
      "epoch": 0.02136514863727848,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012362056205620563,
      "loss": 3.2123,
      "step": 381960
    },
    {
      "epoch": 0.021383791873436662,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012361856185618564,
      "loss": 3.1543,
      "step": 381970
    },
    {
      "epoch": 0.021402435109594848,
      "grad_norm": 0.75,
      "learning_rate": 0.00012361656165616562,
      "loss": 3.2892,
      "step": 381980
    },
    {
      "epoch": 0.02142107834575303,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012361456145614563,
      "loss": 3.1946,
      "step": 381990
    },
    {
      "epoch": 0.021439721581911212,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001236125612561256,
      "loss": 3.1971,
      "step": 382000
    },
    {
      "epoch": 0.021458364818069398,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012361056105610562,
      "loss": 3.2194,
      "step": 382010
    },
    {
      "epoch": 0.02147700805422758,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001236085608560856,
      "loss": 3.1664,
      "step": 382020
    },
    {
      "epoch": 0.021495651290385765,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001236065606560656,
      "loss": 3.1848,
      "step": 382030
    },
    {
      "epoch": 0.021514294526543947,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001236045604560456,
      "loss": 3.1078,
      "step": 382040
    },
    {
      "epoch": 0.021532937762702133,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001236025602560256,
      "loss": 3.2131,
      "step": 382050
    },
    {
      "epoch": 0.021551580998860315,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001236005600560056,
      "loss": 3.2711,
      "step": 382060
    },
    {
      "epoch": 0.0215702242350185,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012359855985598562,
      "loss": 3.2212,
      "step": 382070
    },
    {
      "epoch": 0.021588867471176682,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001235965596559656,
      "loss": 3.171,
      "step": 382080
    },
    {
      "epoch": 0.021607510707334868,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001235945594559456,
      "loss": 3.1726,
      "step": 382090
    },
    {
      "epoch": 0.02162615394349305,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001235925592559256,
      "loss": 3.2244,
      "step": 382100
    },
    {
      "epoch": 0.021644797179651235,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001235905590559056,
      "loss": 3.0248,
      "step": 382110
    },
    {
      "epoch": 0.021663440415809417,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001235885588558856,
      "loss": 3.1395,
      "step": 382120
    },
    {
      "epoch": 0.021682083651967603,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001235865586558656,
      "loss": 3.1346,
      "step": 382130
    },
    {
      "epoch": 0.021700726888125785,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001235845584558456,
      "loss": 3.3528,
      "step": 382140
    },
    {
      "epoch": 0.02171937012428397,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012358255825582558,
      "loss": 3.1007,
      "step": 382150
    },
    {
      "epoch": 0.021738013360442152,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001235805580558056,
      "loss": 3.1578,
      "step": 382160
    },
    {
      "epoch": 0.021756656596600334,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012357855785578557,
      "loss": 3.2256,
      "step": 382170
    },
    {
      "epoch": 0.02177529983275852,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012357655765576558,
      "loss": 3.1853,
      "step": 382180
    },
    {
      "epoch": 0.021793943068916702,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012357455745574557,
      "loss": 3.1687,
      "step": 382190
    },
    {
      "epoch": 0.021812586305074887,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012357255725572557,
      "loss": 3.2101,
      "step": 382200
    },
    {
      "epoch": 0.02183122954123307,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012357055705570558,
      "loss": 3.1902,
      "step": 382210
    },
    {
      "epoch": 0.021849872777391255,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001235685568556856,
      "loss": 3.2613,
      "step": 382220
    },
    {
      "epoch": 0.021868516013549437,
      "grad_norm": 0.5625,
      "learning_rate": 0.00012356655665566558,
      "loss": 3.3228,
      "step": 382230
    },
    {
      "epoch": 0.021887159249707622,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012356455645564558,
      "loss": 3.1695,
      "step": 382240
    },
    {
      "epoch": 0.021905802485865804,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012356255625562557,
      "loss": 3.2128,
      "step": 382250
    },
    {
      "epoch": 0.02192444572202399,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012356055605560558,
      "loss": 3.1886,
      "step": 382260
    },
    {
      "epoch": 0.021943088958182172,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012355855585558556,
      "loss": 3.1163,
      "step": 382270
    },
    {
      "epoch": 0.021961732194340357,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012355655565556557,
      "loss": 3.2417,
      "step": 382280
    },
    {
      "epoch": 0.02198037543049854,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012355455545554555,
      "loss": 3.1785,
      "step": 382290
    },
    {
      "epoch": 0.021999018666656725,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012355255525552556,
      "loss": 3.2444,
      "step": 382300
    },
    {
      "epoch": 0.022017661902814907,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012355055505550554,
      "loss": 3.3281,
      "step": 382310
    },
    {
      "epoch": 0.022036305138973093,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00012354855485548555,
      "loss": 3.2868,
      "step": 382320
    },
    {
      "epoch": 0.022054948375131275,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012354655465546556,
      "loss": 3.0689,
      "step": 382330
    },
    {
      "epoch": 0.022073591611289457,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012354455445544554,
      "loss": 3.1646,
      "step": 382340
    },
    {
      "epoch": 0.022092234847447642,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012354255425542555,
      "loss": 3.184,
      "step": 382350
    },
    {
      "epoch": 0.022110878083605824,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012354055405540553,
      "loss": 3.0427,
      "step": 382360
    },
    {
      "epoch": 0.02212952131976401,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012353855385538554,
      "loss": 3.1859,
      "step": 382370
    },
    {
      "epoch": 0.02214816455592219,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012353655365536555,
      "loss": 3.24,
      "step": 382380
    },
    {
      "epoch": 0.022166807792080377,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012353455345534556,
      "loss": 3.2143,
      "step": 382390
    },
    {
      "epoch": 0.02218545102823856,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012353255325532554,
      "loss": 3.363,
      "step": 382400
    },
    {
      "epoch": 0.022204094264396745,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012353055305530555,
      "loss": 3.1779,
      "step": 382410
    },
    {
      "epoch": 0.022222737500554927,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012352855285528553,
      "loss": 3.1191,
      "step": 382420
    },
    {
      "epoch": 0.022241380736713112,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012352655265526554,
      "loss": 3.1548,
      "step": 382430
    },
    {
      "epoch": 0.022260023972871294,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012352455245524552,
      "loss": 3.2591,
      "step": 382440
    },
    {
      "epoch": 0.02227866720902948,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012352255225522553,
      "loss": 3.3046,
      "step": 382450
    },
    {
      "epoch": 0.022297310445187662,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012352055205520551,
      "loss": 3.2987,
      "step": 382460
    },
    {
      "epoch": 0.022315953681345847,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012351855185518552,
      "loss": 3.0365,
      "step": 382470
    },
    {
      "epoch": 0.02233459691750403,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001235165516551655,
      "loss": 3.1046,
      "step": 382480
    },
    {
      "epoch": 0.022353240153662215,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012351455145514551,
      "loss": 3.2853,
      "step": 382490
    },
    {
      "epoch": 0.022371883389820397,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001235125512551255,
      "loss": 3.3301,
      "step": 382500
    },
    {
      "epoch": 0.02239052662597858,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001235105510551055,
      "loss": 3.269,
      "step": 382510
    },
    {
      "epoch": 0.022409169862136764,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012350855085508552,
      "loss": 3.1511,
      "step": 382520
    },
    {
      "epoch": 0.022427813098294946,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012350655065506552,
      "loss": 3.2266,
      "step": 382530
    },
    {
      "epoch": 0.022446456334453132,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012350455045504553,
      "loss": 3.2676,
      "step": 382540
    },
    {
      "epoch": 0.022465099570611314,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012350255025502552,
      "loss": 3.151,
      "step": 382550
    },
    {
      "epoch": 0.0224837428067695,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012350055005500552,
      "loss": 3.1564,
      "step": 382560
    },
    {
      "epoch": 0.02250238604292768,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001234985498549855,
      "loss": 3.0309,
      "step": 382570
    },
    {
      "epoch": 0.022521029279085867,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012349654965496552,
      "loss": 3.1547,
      "step": 382580
    },
    {
      "epoch": 0.02253967251524405,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001234945494549455,
      "loss": 3.0269,
      "step": 382590
    },
    {
      "epoch": 0.022558315751402234,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001234925492549255,
      "loss": 3.1761,
      "step": 382600
    },
    {
      "epoch": 0.022576958987560417,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001234905490549055,
      "loss": 3.3188,
      "step": 382610
    },
    {
      "epoch": 0.022595602223718602,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001234885488548855,
      "loss": 3.3321,
      "step": 382620
    },
    {
      "epoch": 0.022614245459876784,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012348654865486548,
      "loss": 3.2401,
      "step": 382630
    },
    {
      "epoch": 0.02263288869603497,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001234845484548455,
      "loss": 3.3148,
      "step": 382640
    },
    {
      "epoch": 0.02265153193219315,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012348254825482547,
      "loss": 3.161,
      "step": 382650
    },
    {
      "epoch": 0.022670175168351337,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012348054805480548,
      "loss": 3.1859,
      "step": 382660
    },
    {
      "epoch": 0.02268881840450952,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012347854785478546,
      "loss": 3.1023,
      "step": 382670
    },
    {
      "epoch": 0.0227074616406677,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012347654765476547,
      "loss": 3.165,
      "step": 382680
    },
    {
      "epoch": 0.022726104876825887,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012347454745474548,
      "loss": 3.1657,
      "step": 382690
    },
    {
      "epoch": 0.02274474811298407,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001234725472547255,
      "loss": 3.288,
      "step": 382700
    },
    {
      "epoch": 0.022763391349142254,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012347054705470547,
      "loss": 3.1742,
      "step": 382710
    },
    {
      "epoch": 0.022782034585300436,
      "grad_norm": 0.75,
      "learning_rate": 0.00012346854685468548,
      "loss": 3.1115,
      "step": 382720
    },
    {
      "epoch": 0.02280067782145862,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001234665466546655,
      "loss": 3.3277,
      "step": 382730
    },
    {
      "epoch": 0.022819321057616804,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012346454645464547,
      "loss": 3.2477,
      "step": 382740
    },
    {
      "epoch": 0.02283796429377499,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012346254625462548,
      "loss": 3.2931,
      "step": 382750
    },
    {
      "epoch": 0.02285660752993317,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012346054605460546,
      "loss": 3.216,
      "step": 382760
    },
    {
      "epoch": 0.022875250766091357,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012345854585458547,
      "loss": 3.2852,
      "step": 382770
    },
    {
      "epoch": 0.02289389400224954,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012345654565456545,
      "loss": 3.1806,
      "step": 382780
    },
    {
      "epoch": 0.022912537238407724,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012345454545454546,
      "loss": 3.1101,
      "step": 382790
    },
    {
      "epoch": 0.022931180474565906,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012345254525452545,
      "loss": 3.1779,
      "step": 382800
    },
    {
      "epoch": 0.022949823710724092,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012345054505450546,
      "loss": 3.1833,
      "step": 382810
    },
    {
      "epoch": 0.022968466946882274,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012344854485448544,
      "loss": 3.2693,
      "step": 382820
    },
    {
      "epoch": 0.02298711018304046,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012344654465446545,
      "loss": 3.2434,
      "step": 382830
    },
    {
      "epoch": 0.02300575341919864,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012344454445444546,
      "loss": 3.1874,
      "step": 382840
    },
    {
      "epoch": 0.023024396655356823,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012344254425442546,
      "loss": 3.1289,
      "step": 382850
    },
    {
      "epoch": 0.02304303989151501,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012344054405440545,
      "loss": 3.2146,
      "step": 382860
    },
    {
      "epoch": 0.02306168312767319,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012343854385438546,
      "loss": 3.1999,
      "step": 382870
    },
    {
      "epoch": 0.023080326363831376,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012343654365436544,
      "loss": 3.1885,
      "step": 382880
    },
    {
      "epoch": 0.02309896959998956,
      "grad_norm": 0.625,
      "learning_rate": 0.00012343454345434545,
      "loss": 3.2856,
      "step": 382890
    },
    {
      "epoch": 0.023117612836147744,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012343254325432543,
      "loss": 3.204,
      "step": 382900
    },
    {
      "epoch": 0.023136256072305926,
      "grad_norm": 0.921875,
      "learning_rate": 0.00012343054305430544,
      "loss": 3.2388,
      "step": 382910
    },
    {
      "epoch": 0.02315489930846411,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012342854285428545,
      "loss": 3.1489,
      "step": 382920
    },
    {
      "epoch": 0.023173542544622294,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012342654265426543,
      "loss": 3.2844,
      "step": 382930
    },
    {
      "epoch": 0.02319218578078048,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012342454245424544,
      "loss": 3.2157,
      "step": 382940
    },
    {
      "epoch": 0.02321082901693866,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012342254225422542,
      "loss": 3.2005,
      "step": 382950
    },
    {
      "epoch": 0.023229472253096847,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012342054205420543,
      "loss": 3.3134,
      "step": 382960
    },
    {
      "epoch": 0.02324811548925503,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001234185418541854,
      "loss": 3.1696,
      "step": 382970
    },
    {
      "epoch": 0.023266758725413214,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012341654165416542,
      "loss": 3.2728,
      "step": 382980
    },
    {
      "epoch": 0.023285401961571396,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001234145414541454,
      "loss": 3.2889,
      "step": 382990
    },
    {
      "epoch": 0.02330404519772958,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001234125412541254,
      "loss": 3.0768,
      "step": 383000
    },
    {
      "epoch": 0.023322688433887764,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012341054105410542,
      "loss": 3.216,
      "step": 383010
    },
    {
      "epoch": 0.023341331670045946,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012340854085408543,
      "loss": 3.0492,
      "step": 383020
    },
    {
      "epoch": 0.02335997490620413,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001234065406540654,
      "loss": 3.2099,
      "step": 383030
    },
    {
      "epoch": 0.023378618142362313,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012340454045404542,
      "loss": 3.026,
      "step": 383040
    },
    {
      "epoch": 0.0233972613785205,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001234025402540254,
      "loss": 3.3151,
      "step": 383050
    },
    {
      "epoch": 0.02341590461467868,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001234005400540054,
      "loss": 3.0636,
      "step": 383060
    },
    {
      "epoch": 0.023434547850836866,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001233985398539854,
      "loss": 3.264,
      "step": 383070
    },
    {
      "epoch": 0.02345319108699505,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001233965396539654,
      "loss": 3.2209,
      "step": 383080
    },
    {
      "epoch": 0.023471834323153234,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012339453945394539,
      "loss": 3.2974,
      "step": 383090
    },
    {
      "epoch": 0.023490477559311416,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001233925392539254,
      "loss": 3.1804,
      "step": 383100
    },
    {
      "epoch": 0.0235091207954696,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001233905390539054,
      "loss": 3.0934,
      "step": 383110
    },
    {
      "epoch": 0.023527764031627783,
      "grad_norm": 0.75,
      "learning_rate": 0.00012338853885388539,
      "loss": 3.301,
      "step": 383120
    },
    {
      "epoch": 0.02354640726778597,
      "grad_norm": 0.75,
      "learning_rate": 0.0001233865386538654,
      "loss": 3.2172,
      "step": 383130
    },
    {
      "epoch": 0.02356505050394415,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012338453845384538,
      "loss": 3.236,
      "step": 383140
    },
    {
      "epoch": 0.023583693740102336,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001233825382538254,
      "loss": 3.1756,
      "step": 383150
    },
    {
      "epoch": 0.02360233697626052,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001233805380538054,
      "loss": 3.2249,
      "step": 383160
    },
    {
      "epoch": 0.023620980212418704,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001233785378537854,
      "loss": 3.2341,
      "step": 383170
    },
    {
      "epoch": 0.023639623448576886,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001233765376537654,
      "loss": 3.2102,
      "step": 383180
    },
    {
      "epoch": 0.02365826668473507,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001233745374537454,
      "loss": 3.1059,
      "step": 383190
    },
    {
      "epoch": 0.023676909920893253,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012337253725372538,
      "loss": 3.2576,
      "step": 383200
    },
    {
      "epoch": 0.023695553157051435,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001233705370537054,
      "loss": 3.1216,
      "step": 383210
    },
    {
      "epoch": 0.02371419639320962,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012336853685368537,
      "loss": 3.1387,
      "step": 383220
    },
    {
      "epoch": 0.023732839629367803,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012336653665366538,
      "loss": 3.1342,
      "step": 383230
    },
    {
      "epoch": 0.02375148286552599,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012336453645364536,
      "loss": 3.3276,
      "step": 383240
    },
    {
      "epoch": 0.02377012610168417,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012336253625362537,
      "loss": 3.133,
      "step": 383250
    },
    {
      "epoch": 0.023788769337842356,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012336053605360535,
      "loss": 3.1623,
      "step": 383260
    },
    {
      "epoch": 0.023807412574000538,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012335853585358536,
      "loss": 3.1711,
      "step": 383270
    },
    {
      "epoch": 0.023826055810158724,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012335653565356534,
      "loss": 3.1783,
      "step": 383280
    },
    {
      "epoch": 0.023844699046316906,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012335453545354535,
      "loss": 3.0906,
      "step": 383290
    },
    {
      "epoch": 0.02386334228247509,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012335253525352533,
      "loss": 3.2926,
      "step": 383300
    },
    {
      "epoch": 0.023881985518633273,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012335053505350534,
      "loss": 3.3117,
      "step": 383310
    },
    {
      "epoch": 0.02390062875479146,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012334853485348535,
      "loss": 3.2,
      "step": 383320
    },
    {
      "epoch": 0.02391927199094964,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012334653465346536,
      "loss": 3.0863,
      "step": 383330
    },
    {
      "epoch": 0.023937915227107826,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012334453445344537,
      "loss": 3.3661,
      "step": 383340
    },
    {
      "epoch": 0.023956558463266008,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012334253425342535,
      "loss": 3.2727,
      "step": 383350
    },
    {
      "epoch": 0.023975201699424194,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012334053405340536,
      "loss": 3.1369,
      "step": 383360
    },
    {
      "epoch": 0.023993844935582376,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012333853385338534,
      "loss": 3.3209,
      "step": 383370
    },
    {
      "epoch": 0.024012488171740558,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012333653365336535,
      "loss": 3.187,
      "step": 383380
    },
    {
      "epoch": 0.024031131407898743,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012333453345334534,
      "loss": 3.3016,
      "step": 383390
    },
    {
      "epoch": 0.024049774644056925,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012333253325332534,
      "loss": 3.189,
      "step": 383400
    },
    {
      "epoch": 0.02406841788021511,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012333053305330533,
      "loss": 3.1469,
      "step": 383410
    },
    {
      "epoch": 0.024087061116373293,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012332853285328534,
      "loss": 3.2636,
      "step": 383420
    },
    {
      "epoch": 0.02410570435253148,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012332653265326532,
      "loss": 3.2254,
      "step": 383430
    },
    {
      "epoch": 0.02412434758868966,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012332453245324533,
      "loss": 3.2264,
      "step": 383440
    },
    {
      "epoch": 0.024142990824847846,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001233225322532253,
      "loss": 3.2821,
      "step": 383450
    },
    {
      "epoch": 0.024161634061006028,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012332053205320532,
      "loss": 3.2269,
      "step": 383460
    },
    {
      "epoch": 0.024180277297164213,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00012331853185318533,
      "loss": 3.1694,
      "step": 383470
    },
    {
      "epoch": 0.024198920533322395,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012331653165316534,
      "loss": 3.2455,
      "step": 383480
    },
    {
      "epoch": 0.02421756376948058,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012331453145314532,
      "loss": 3.2609,
      "step": 383490
    },
    {
      "epoch": 0.024236207005638763,
      "grad_norm": 0.75,
      "learning_rate": 0.00012331253125312533,
      "loss": 3.1622,
      "step": 383500
    },
    {
      "epoch": 0.02425485024179695,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012331053105310534,
      "loss": 3.1108,
      "step": 383510
    },
    {
      "epoch": 0.02427349347795513,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012330853085308532,
      "loss": 3.2638,
      "step": 383520
    },
    {
      "epoch": 0.024292136714113316,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012330653065306533,
      "loss": 3.1446,
      "step": 383530
    },
    {
      "epoch": 0.024310779950271498,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001233045304530453,
      "loss": 3.3125,
      "step": 383540
    },
    {
      "epoch": 0.02432942318642968,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012330253025302532,
      "loss": 3.1912,
      "step": 383550
    },
    {
      "epoch": 0.024348066422587866,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001233005300530053,
      "loss": 3.0623,
      "step": 383560
    },
    {
      "epoch": 0.024366709658746048,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001232985298529853,
      "loss": 3.1648,
      "step": 383570
    },
    {
      "epoch": 0.024385352894904233,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001232965296529653,
      "loss": 3.187,
      "step": 383580
    },
    {
      "epoch": 0.024403996131062415,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001232945294529453,
      "loss": 3.1481,
      "step": 383590
    },
    {
      "epoch": 0.0244226393672206,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012329252925292528,
      "loss": 3.2355,
      "step": 383600
    },
    {
      "epoch": 0.024441282603378783,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001232905290529053,
      "loss": 3.2832,
      "step": 383610
    },
    {
      "epoch": 0.024459925839536968,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012328852885288527,
      "loss": 3.2171,
      "step": 383620
    },
    {
      "epoch": 0.02447856907569515,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012328652865286528,
      "loss": 3.1342,
      "step": 383630
    },
    {
      "epoch": 0.024497212311853336,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001232845284528453,
      "loss": 3.2032,
      "step": 383640
    },
    {
      "epoch": 0.024515855548011518,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001232825282528253,
      "loss": 3.2612,
      "step": 383650
    },
    {
      "epoch": 0.024534498784169703,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012328052805280528,
      "loss": 3.2445,
      "step": 383660
    },
    {
      "epoch": 0.024553142020327885,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001232785278527853,
      "loss": 3.2163,
      "step": 383670
    },
    {
      "epoch": 0.02457178525648607,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012327652765276528,
      "loss": 3.219,
      "step": 383680
    },
    {
      "epoch": 0.024590428492644253,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012327452745274528,
      "loss": 3.2404,
      "step": 383690
    },
    {
      "epoch": 0.024609071728802438,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001232725272527253,
      "loss": 3.1088,
      "step": 383700
    },
    {
      "epoch": 0.02462771496496062,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012327052705270528,
      "loss": 3.1451,
      "step": 383710
    },
    {
      "epoch": 0.024646358201118802,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012326852685268528,
      "loss": 3.143,
      "step": 383720
    },
    {
      "epoch": 0.024665001437276988,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012326652665266527,
      "loss": 3.1654,
      "step": 383730
    },
    {
      "epoch": 0.02468364467343517,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012326452645264528,
      "loss": 3.1513,
      "step": 383740
    },
    {
      "epoch": 0.024702287909593355,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012326252625262526,
      "loss": 3.185,
      "step": 383750
    },
    {
      "epoch": 0.024720931145751537,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012326052605260527,
      "loss": 3.2523,
      "step": 383760
    },
    {
      "epoch": 0.024739574381909723,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012325852585258525,
      "loss": 3.0397,
      "step": 383770
    },
    {
      "epoch": 0.024758217618067905,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00012325652565256526,
      "loss": 3.1938,
      "step": 383780
    },
    {
      "epoch": 0.02477686085422609,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012325452545254527,
      "loss": 3.3177,
      "step": 383790
    },
    {
      "epoch": 0.024795504090384272,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012325252525252528,
      "loss": 3.2477,
      "step": 383800
    },
    {
      "epoch": 0.024814147326542458,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012325052505250526,
      "loss": 3.2765,
      "step": 383810
    },
    {
      "epoch": 0.02483279056270064,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012324852485248527,
      "loss": 3.1534,
      "step": 383820
    },
    {
      "epoch": 0.024851433798858825,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012324652465246525,
      "loss": 3.1325,
      "step": 383830
    },
    {
      "epoch": 0.024870077035017008,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012324452445244526,
      "loss": 3.1676,
      "step": 383840
    },
    {
      "epoch": 0.024888720271175193,
      "grad_norm": 0.75,
      "learning_rate": 0.00012324252425242524,
      "loss": 3.212,
      "step": 383850
    },
    {
      "epoch": 0.024907363507333375,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012324052405240525,
      "loss": 3.2114,
      "step": 383860
    },
    {
      "epoch": 0.02492600674349156,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012323852385238523,
      "loss": 3.0985,
      "step": 383870
    },
    {
      "epoch": 0.024944649979649743,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012323652365236524,
      "loss": 3.3129,
      "step": 383880
    },
    {
      "epoch": 0.024963293215807925,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012323452345234522,
      "loss": 3.198,
      "step": 383890
    },
    {
      "epoch": 0.02498193645196611,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012323252325232523,
      "loss": 3.209,
      "step": 383900
    },
    {
      "epoch": 0.025000579688124292,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012323052305230524,
      "loss": 3.1727,
      "step": 383910
    },
    {
      "epoch": 0.025019222924282478,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012322852285228522,
      "loss": 3.1916,
      "step": 383920
    },
    {
      "epoch": 0.02503786616044066,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012322652265226523,
      "loss": 3.2403,
      "step": 383930
    },
    {
      "epoch": 0.025056509396598845,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012322452245224521,
      "loss": 3.2047,
      "step": 383940
    },
    {
      "epoch": 0.025075152632757027,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012322252225222522,
      "loss": 3.167,
      "step": 383950
    },
    {
      "epoch": 0.025093795868915213,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012322052205220523,
      "loss": 3.2248,
      "step": 383960
    },
    {
      "epoch": 0.025112439105073395,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012321852185218524,
      "loss": 3.1856,
      "step": 383970
    },
    {
      "epoch": 0.02513108234123158,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012321652165216522,
      "loss": 3.2089,
      "step": 383980
    },
    {
      "epoch": 0.025149725577389762,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012321452145214523,
      "loss": 3.2259,
      "step": 383990
    },
    {
      "epoch": 0.025168368813547948,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012321252125212522,
      "loss": 3.1422,
      "step": 384000
    },
    {
      "epoch": 0.02518701204970613,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012321052105210522,
      "loss": 3.1041,
      "step": 384010
    },
    {
      "epoch": 0.025205655285864315,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001232085208520852,
      "loss": 3.2005,
      "step": 384020
    },
    {
      "epoch": 0.025224298522022497,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012320652065206522,
      "loss": 3.1017,
      "step": 384030
    },
    {
      "epoch": 0.025242941758180683,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001232045204520452,
      "loss": 3.3002,
      "step": 384040
    },
    {
      "epoch": 0.025261584994338865,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001232025202520252,
      "loss": 3.1416,
      "step": 384050
    },
    {
      "epoch": 0.025280228230497047,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001232005200520052,
      "loss": 3.2663,
      "step": 384060
    },
    {
      "epoch": 0.025298871466655232,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001231985198519852,
      "loss": 3.3037,
      "step": 384070
    },
    {
      "epoch": 0.025317514702813414,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012319651965196518,
      "loss": 3.1645,
      "step": 384080
    },
    {
      "epoch": 0.0253361579389716,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001231945194519452,
      "loss": 3.3391,
      "step": 384090
    },
    {
      "epoch": 0.025354801175129782,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001231925192519252,
      "loss": 3.3098,
      "step": 384100
    },
    {
      "epoch": 0.025373444411287967,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001231905190519052,
      "loss": 3.2802,
      "step": 384110
    },
    {
      "epoch": 0.02539208764744615,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012318851885188522,
      "loss": 3.38,
      "step": 384120
    },
    {
      "epoch": 0.025410730883604335,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001231865186518652,
      "loss": 3.1529,
      "step": 384130
    },
    {
      "epoch": 0.025429374119762517,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001231845184518452,
      "loss": 3.2751,
      "step": 384140
    },
    {
      "epoch": 0.025448017355920702,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001231825182518252,
      "loss": 3.2388,
      "step": 384150
    },
    {
      "epoch": 0.025466660592078885,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001231805180518052,
      "loss": 3.2105,
      "step": 384160
    },
    {
      "epoch": 0.02548530382823707,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012317851785178518,
      "loss": 3.2056,
      "step": 384170
    },
    {
      "epoch": 0.025503947064395252,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001231765176517652,
      "loss": 3.2383,
      "step": 384180
    },
    {
      "epoch": 0.025522590300553438,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012317451745174517,
      "loss": 3.159,
      "step": 384190
    },
    {
      "epoch": 0.02554123353671162,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012317251725172518,
      "loss": 3.2891,
      "step": 384200
    },
    {
      "epoch": 0.025559876772869805,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012317051705170516,
      "loss": 3.151,
      "step": 384210
    },
    {
      "epoch": 0.025578520009027987,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012316851685168517,
      "loss": 3.2696,
      "step": 384220
    },
    {
      "epoch": 0.02559716324518617,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012316651665166515,
      "loss": 3.0613,
      "step": 384230
    },
    {
      "epoch": 0.025615806481344355,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012316451645164516,
      "loss": 3.1853,
      "step": 384240
    },
    {
      "epoch": 0.025634449717502537,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012316251625162517,
      "loss": 3.07,
      "step": 384250
    },
    {
      "epoch": 0.025653092953660722,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012316051605160516,
      "loss": 3.301,
      "step": 384260
    },
    {
      "epoch": 0.025671736189818904,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012315851585158516,
      "loss": 3.2122,
      "step": 384270
    },
    {
      "epoch": 0.02569037942597709,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012315651565156517,
      "loss": 3.2772,
      "step": 384280
    },
    {
      "epoch": 0.025709022662135272,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012315451545154516,
      "loss": 3.2936,
      "step": 384290
    },
    {
      "epoch": 0.025727665898293457,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012315251525152516,
      "loss": 3.2191,
      "step": 384300
    },
    {
      "epoch": 0.02574630913445164,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012315051505150517,
      "loss": 3.3504,
      "step": 384310
    },
    {
      "epoch": 0.025764952370609825,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012314851485148516,
      "loss": 3.3097,
      "step": 384320
    },
    {
      "epoch": 0.025783595606768007,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012314651465146516,
      "loss": 3.0181,
      "step": 384330
    },
    {
      "epoch": 0.025802238842926192,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012314451445144515,
      "loss": 3.1847,
      "step": 384340
    },
    {
      "epoch": 0.025820882079084374,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012314251425142516,
      "loss": 3.1579,
      "step": 384350
    },
    {
      "epoch": 0.02583952531524256,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012314051405140514,
      "loss": 3.2123,
      "step": 384360
    },
    {
      "epoch": 0.025858168551400742,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012313851385138515,
      "loss": 3.1432,
      "step": 384370
    },
    {
      "epoch": 0.025876811787558927,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012313651365136513,
      "loss": 3.1502,
      "step": 384380
    },
    {
      "epoch": 0.02589545502371711,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012313451345134514,
      "loss": 3.0967,
      "step": 384390
    },
    {
      "epoch": 0.02591409825987529,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012313251325132512,
      "loss": 3.1448,
      "step": 384400
    },
    {
      "epoch": 0.025932741496033477,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012313051305130513,
      "loss": 3.2609,
      "step": 384410
    },
    {
      "epoch": 0.02595138473219166,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012312851285128514,
      "loss": 3.2048,
      "step": 384420
    },
    {
      "epoch": 0.025970027968349844,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012312651265126515,
      "loss": 3.2504,
      "step": 384430
    },
    {
      "epoch": 0.025988671204508026,
      "grad_norm": 0.75,
      "learning_rate": 0.00012312451245124513,
      "loss": 3.1277,
      "step": 384440
    },
    {
      "epoch": 0.026007314440666212,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012312251225122514,
      "loss": 3.2649,
      "step": 384450
    },
    {
      "epoch": 0.026025957676824394,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012312051205120512,
      "loss": 3.224,
      "step": 384460
    },
    {
      "epoch": 0.02604460091298258,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012311851185118513,
      "loss": 3.2315,
      "step": 384470
    },
    {
      "epoch": 0.02606324414914076,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001231165116511651,
      "loss": 3.2049,
      "step": 384480
    },
    {
      "epoch": 0.026081887385298947,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012311451145114512,
      "loss": 3.2619,
      "step": 384490
    },
    {
      "epoch": 0.02610053062145713,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012311251125112513,
      "loss": 3.2599,
      "step": 384500
    },
    {
      "epoch": 0.026119173857615315,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001231105110511051,
      "loss": 3.161,
      "step": 384510
    },
    {
      "epoch": 0.026137817093773497,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012310851085108512,
      "loss": 3.2362,
      "step": 384520
    },
    {
      "epoch": 0.026156460329931682,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001231065106510651,
      "loss": 3.2462,
      "step": 384530
    },
    {
      "epoch": 0.026175103566089864,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001231045104510451,
      "loss": 3.1928,
      "step": 384540
    },
    {
      "epoch": 0.02619374680224805,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001231025102510251,
      "loss": 3.2476,
      "step": 384550
    },
    {
      "epoch": 0.02621239003840623,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001231005100510051,
      "loss": 3.2148,
      "step": 384560
    },
    {
      "epoch": 0.026231033274564414,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012309850985098509,
      "loss": 3.0374,
      "step": 384570
    },
    {
      "epoch": 0.0262496765107226,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001230965096509651,
      "loss": 3.1912,
      "step": 384580
    },
    {
      "epoch": 0.02626831974688078,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001230945094509451,
      "loss": 3.2251,
      "step": 384590
    },
    {
      "epoch": 0.026286962983038967,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012309250925092511,
      "loss": 3.1919,
      "step": 384600
    },
    {
      "epoch": 0.02630560621919715,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001230905090509051,
      "loss": 3.1518,
      "step": 384610
    },
    {
      "epoch": 0.026324249455355334,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001230885088508851,
      "loss": 3.1944,
      "step": 384620
    },
    {
      "epoch": 0.026342892691513516,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001230865086508651,
      "loss": 3.2733,
      "step": 384630
    },
    {
      "epoch": 0.026361535927671702,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001230845084508451,
      "loss": 3.2021,
      "step": 384640
    },
    {
      "epoch": 0.026380179163829884,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012308250825082508,
      "loss": 3.2127,
      "step": 384650
    },
    {
      "epoch": 0.02639882239998807,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001230805080508051,
      "loss": 3.2104,
      "step": 384660
    },
    {
      "epoch": 0.02641746563614625,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012307850785078507,
      "loss": 3.2127,
      "step": 384670
    },
    {
      "epoch": 0.026436108872304437,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012307650765076508,
      "loss": 3.2903,
      "step": 384680
    },
    {
      "epoch": 0.02645475210846262,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001230745074507451,
      "loss": 3.1653,
      "step": 384690
    },
    {
      "epoch": 0.026473395344620804,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012307250725072507,
      "loss": 3.1268,
      "step": 384700
    },
    {
      "epoch": 0.026492038580778986,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00012307050705070508,
      "loss": 3.1758,
      "step": 384710
    },
    {
      "epoch": 0.026510681816937172,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012306850685068506,
      "loss": 3.2009,
      "step": 384720
    },
    {
      "epoch": 0.026529325053095354,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012306650665066507,
      "loss": 3.1316,
      "step": 384730
    },
    {
      "epoch": 0.026547968289253536,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012306450645064508,
      "loss": 3.2274,
      "step": 384740
    },
    {
      "epoch": 0.02656661152541172,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001230625062506251,
      "loss": 3.2059,
      "step": 384750
    },
    {
      "epoch": 0.026585254761569903,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012306050605060507,
      "loss": 3.2541,
      "step": 384760
    },
    {
      "epoch": 0.02660389799772809,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012305850585058508,
      "loss": 3.1756,
      "step": 384770
    },
    {
      "epoch": 0.02662254123388627,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012305650565056506,
      "loss": 3.1983,
      "step": 384780
    },
    {
      "epoch": 0.026641184470044457,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012305450545054507,
      "loss": 3.1868,
      "step": 384790
    },
    {
      "epoch": 0.02665982770620264,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012305250525052505,
      "loss": 3.132,
      "step": 384800
    },
    {
      "epoch": 0.026678470942360824,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012305050505050506,
      "loss": 3.3249,
      "step": 384810
    },
    {
      "epoch": 0.026697114178519006,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012304850485048504,
      "loss": 3.0906,
      "step": 384820
    },
    {
      "epoch": 0.02671575741467719,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012304650465046505,
      "loss": 3.1653,
      "step": 384830
    },
    {
      "epoch": 0.026734400650835374,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012304450445044504,
      "loss": 3.1683,
      "step": 384840
    },
    {
      "epoch": 0.02675304388699356,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012304250425042504,
      "loss": 3.2833,
      "step": 384850
    },
    {
      "epoch": 0.02677168712315174,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012304050405040503,
      "loss": 3.263,
      "step": 384860
    },
    {
      "epoch": 0.026790330359309927,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012303850385038504,
      "loss": 3.2633,
      "step": 384870
    },
    {
      "epoch": 0.02680897359546811,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012303650365036504,
      "loss": 3.1418,
      "step": 384880
    },
    {
      "epoch": 0.026827616831626294,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012303450345034503,
      "loss": 3.2956,
      "step": 384890
    },
    {
      "epoch": 0.026846260067784476,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012303250325032504,
      "loss": 3.1219,
      "step": 384900
    },
    {
      "epoch": 0.026864903303942658,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012303050305030504,
      "loss": 3.0891,
      "step": 384910
    },
    {
      "epoch": 0.026883546540100844,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012302850285028505,
      "loss": 3.1493,
      "step": 384920
    },
    {
      "epoch": 0.026902189776259026,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012302650265026504,
      "loss": 3.2485,
      "step": 384930
    },
    {
      "epoch": 0.02692083301241721,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012302450245024505,
      "loss": 3.1097,
      "step": 384940
    },
    {
      "epoch": 0.026939476248575393,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012302250225022503,
      "loss": 3.3045,
      "step": 384950
    },
    {
      "epoch": 0.02695811948473358,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012302050205020504,
      "loss": 3.1904,
      "step": 384960
    },
    {
      "epoch": 0.02697676272089176,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012301850185018502,
      "loss": 3.1909,
      "step": 384970
    },
    {
      "epoch": 0.026995405957049946,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012301650165016503,
      "loss": 3.1407,
      "step": 384980
    },
    {
      "epoch": 0.02701404919320813,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000123014501450145,
      "loss": 3.0957,
      "step": 384990
    },
    {
      "epoch": 0.027032692429366314,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012301250125012502,
      "loss": 3.115,
      "step": 385000
    },
    {
      "epoch": 0.027051335665524496,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000123010501050105,
      "loss": 3.2535,
      "step": 385010
    },
    {
      "epoch": 0.02706997890168268,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000123008500850085,
      "loss": 3.2034,
      "step": 385020
    },
    {
      "epoch": 0.027088622137840863,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000123006500650065,
      "loss": 3.1916,
      "step": 385030
    },
    {
      "epoch": 0.02710726537399905,
      "grad_norm": 0.765625,
      "learning_rate": 0.000123004500450045,
      "loss": 3.245,
      "step": 385040
    },
    {
      "epoch": 0.02712590861015723,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000123002500250025,
      "loss": 3.3234,
      "step": 385050
    },
    {
      "epoch": 0.027144551846315416,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012300050005000502,
      "loss": 3.3514,
      "step": 385060
    },
    {
      "epoch": 0.0271631950824736,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000122998499849985,
      "loss": 3.3452,
      "step": 385070
    },
    {
      "epoch": 0.02718183831863178,
      "grad_norm": 0.6875,
      "learning_rate": 0.000122996499649965,
      "loss": 3.2247,
      "step": 385080
    },
    {
      "epoch": 0.027200481554789966,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012299449944994502,
      "loss": 3.2155,
      "step": 385090
    },
    {
      "epoch": 0.027219124790948148,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000122992499249925,
      "loss": 3.2287,
      "step": 385100
    },
    {
      "epoch": 0.027237768027106334,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000122990499049905,
      "loss": 3.1825,
      "step": 385110
    },
    {
      "epoch": 0.027256411263264516,
      "grad_norm": 0.78125,
      "learning_rate": 0.000122988498849885,
      "loss": 3.0599,
      "step": 385120
    },
    {
      "epoch": 0.0272750544994227,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000122986498649865,
      "loss": 3.1529,
      "step": 385130
    },
    {
      "epoch": 0.027293697735580883,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012298449844984498,
      "loss": 3.2408,
      "step": 385140
    },
    {
      "epoch": 0.02731234097173907,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000122982498249825,
      "loss": 3.1119,
      "step": 385150
    },
    {
      "epoch": 0.02733098420789725,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012298049804980498,
      "loss": 3.1929,
      "step": 385160
    },
    {
      "epoch": 0.027349627444055436,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012297849784978498,
      "loss": 3.2154,
      "step": 385170
    },
    {
      "epoch": 0.027368270680213618,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012297649764976497,
      "loss": 3.1249,
      "step": 385180
    },
    {
      "epoch": 0.027386913916371804,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012297449744974498,
      "loss": 3.1236,
      "step": 385190
    },
    {
      "epoch": 0.027405557152529986,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012297249724972498,
      "loss": 3.3302,
      "step": 385200
    },
    {
      "epoch": 0.02742420038868817,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012297049704970497,
      "loss": 3.0818,
      "step": 385210
    },
    {
      "epoch": 0.027442843624846353,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012296849684968498,
      "loss": 3.0894,
      "step": 385220
    },
    {
      "epoch": 0.02746148686100454,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012296649664966499,
      "loss": 3.1899,
      "step": 385230
    },
    {
      "epoch": 0.02748013009716272,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012296449644964497,
      "loss": 3.2112,
      "step": 385240
    },
    {
      "epoch": 0.027498773333320903,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012296249624962498,
      "loss": 3.1692,
      "step": 385250
    },
    {
      "epoch": 0.02751741656947909,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012296049604960496,
      "loss": 3.1312,
      "step": 385260
    },
    {
      "epoch": 0.02753605980563727,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012295849584958497,
      "loss": 3.0552,
      "step": 385270
    },
    {
      "epoch": 0.027554703041795456,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012295649564956498,
      "loss": 3.1011,
      "step": 385280
    },
    {
      "epoch": 0.027573346277953638,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012295449544954496,
      "loss": 3.3705,
      "step": 385290
    },
    {
      "epoch": 0.027591989514111823,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012295249524952497,
      "loss": 3.0456,
      "step": 385300
    },
    {
      "epoch": 0.027610632750270005,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012295049504950495,
      "loss": 3.0984,
      "step": 385310
    },
    {
      "epoch": 0.02762927598642819,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012294849484948496,
      "loss": 3.2296,
      "step": 385320
    },
    {
      "epoch": 0.027647919222586373,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012294649464946494,
      "loss": 3.1627,
      "step": 385330
    },
    {
      "epoch": 0.02766656245874456,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012294449444944495,
      "loss": 3.3549,
      "step": 385340
    },
    {
      "epoch": 0.02768520569490274,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012294249424942493,
      "loss": 3.115,
      "step": 385350
    },
    {
      "epoch": 0.027703848931060926,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012294049404940494,
      "loss": 3.2464,
      "step": 385360
    },
    {
      "epoch": 0.027722492167219108,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012293849384938495,
      "loss": 3.2121,
      "step": 385370
    },
    {
      "epoch": 0.027741135403377293,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012293649364936496,
      "loss": 3.1472,
      "step": 385380
    },
    {
      "epoch": 0.027759778639535476,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012293449344934494,
      "loss": 3.1338,
      "step": 385390
    },
    {
      "epoch": 0.02777842187569366,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012293249324932495,
      "loss": 3.2056,
      "step": 385400
    },
    {
      "epoch": 0.027797065111851843,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012293049304930493,
      "loss": 3.1778,
      "step": 385410
    },
    {
      "epoch": 0.027815708348010025,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012292849284928494,
      "loss": 3.3197,
      "step": 385420
    },
    {
      "epoch": 0.02783435158416821,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012292649264926492,
      "loss": 3.1726,
      "step": 385430
    },
    {
      "epoch": 0.027852994820326393,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012292449244924493,
      "loss": 3.2472,
      "step": 385440
    },
    {
      "epoch": 0.027871638056484578,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012292249224922492,
      "loss": 3.2828,
      "step": 385450
    },
    {
      "epoch": 0.02789028129264276,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012292049204920492,
      "loss": 3.1732,
      "step": 385460
    },
    {
      "epoch": 0.027908924528800946,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001229184918491849,
      "loss": 3.1332,
      "step": 385470
    },
    {
      "epoch": 0.027927567764959128,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012291649164916492,
      "loss": 3.1667,
      "step": 385480
    },
    {
      "epoch": 0.027946211001117313,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012291449144914492,
      "loss": 3.3116,
      "step": 385490
    },
    {
      "epoch": 0.027964854237275495,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001229124912491249,
      "loss": 3.1718,
      "step": 385500
    },
    {
      "epoch": 0.02798349747343368,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012291049104910492,
      "loss": 3.2085,
      "step": 385510
    },
    {
      "epoch": 0.028002140709591863,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012290849084908493,
      "loss": 3.1307,
      "step": 385520
    },
    {
      "epoch": 0.028020783945750048,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001229064906490649,
      "loss": 3.1246,
      "step": 385530
    },
    {
      "epoch": 0.02803942718190823,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012290449044904492,
      "loss": 3.1114,
      "step": 385540
    },
    {
      "epoch": 0.028058070418066416,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012290249024902493,
      "loss": 3.2598,
      "step": 385550
    },
    {
      "epoch": 0.028076713654224598,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001229004900490049,
      "loss": 3.099,
      "step": 385560
    },
    {
      "epoch": 0.028095356890382783,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012289848984898492,
      "loss": 3.2547,
      "step": 385570
    },
    {
      "epoch": 0.028114000126540965,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001228964896489649,
      "loss": 3.0244,
      "step": 385580
    },
    {
      "epoch": 0.02813264336269915,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001228944894489449,
      "loss": 3.1975,
      "step": 385590
    },
    {
      "epoch": 0.028151286598857333,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001228924892489249,
      "loss": 3.049,
      "step": 385600
    },
    {
      "epoch": 0.028169929835015515,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001228904890489049,
      "loss": 3.1434,
      "step": 385610
    },
    {
      "epoch": 0.0281885730711737,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012288848884888488,
      "loss": 3.3348,
      "step": 385620
    },
    {
      "epoch": 0.028207216307331882,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001228864886488649,
      "loss": 3.2049,
      "step": 385630
    },
    {
      "epoch": 0.028225859543490068,
      "grad_norm": 0.90625,
      "learning_rate": 0.00012288448844884487,
      "loss": 3.0727,
      "step": 385640
    },
    {
      "epoch": 0.02824450277964825,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012288248824882488,
      "loss": 3.2104,
      "step": 385650
    },
    {
      "epoch": 0.028263146015806435,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012288048804880486,
      "loss": 3.2776,
      "step": 385660
    },
    {
      "epoch": 0.028281789251964617,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012287848784878487,
      "loss": 3.1585,
      "step": 385670
    },
    {
      "epoch": 0.028300432488122803,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012287648764876488,
      "loss": 3.2369,
      "step": 385680
    },
    {
      "epoch": 0.028319075724280985,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001228744874487449,
      "loss": 3.1454,
      "step": 385690
    },
    {
      "epoch": 0.02833771896043917,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001228724872487249,
      "loss": 3.1424,
      "step": 385700
    },
    {
      "epoch": 0.028356362196597353,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012287048704870488,
      "loss": 3.1422,
      "step": 385710
    },
    {
      "epoch": 0.028375005432755538,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001228684868486849,
      "loss": 3.1975,
      "step": 385720
    },
    {
      "epoch": 0.02839364866891372,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012286648664866487,
      "loss": 3.088,
      "step": 385730
    },
    {
      "epoch": 0.028412291905071906,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012286448644864488,
      "loss": 3.0838,
      "step": 385740
    },
    {
      "epoch": 0.028430935141230088,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012286248624862486,
      "loss": 3.1962,
      "step": 385750
    },
    {
      "epoch": 0.028449578377388273,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012286048604860487,
      "loss": 3.2174,
      "step": 385760
    },
    {
      "epoch": 0.028468221613546455,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012285848584858486,
      "loss": 3.2848,
      "step": 385770
    },
    {
      "epoch": 0.028486864849704637,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012285648564856486,
      "loss": 3.2375,
      "step": 385780
    },
    {
      "epoch": 0.028505508085862823,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012285448544854485,
      "loss": 3.2316,
      "step": 385790
    },
    {
      "epoch": 0.028524151322021005,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012285248524852486,
      "loss": 3.3053,
      "step": 385800
    },
    {
      "epoch": 0.02854279455817919,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012285048504850484,
      "loss": 3.2007,
      "step": 385810
    },
    {
      "epoch": 0.028561437794337372,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012284848484848485,
      "loss": 3.1024,
      "step": 385820
    },
    {
      "epoch": 0.028580081030495558,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012284648464846486,
      "loss": 3.1509,
      "step": 385830
    },
    {
      "epoch": 0.02859872426665374,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012284448444844487,
      "loss": 3.1631,
      "step": 385840
    },
    {
      "epoch": 0.028617367502811925,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012284248424842485,
      "loss": 3.1287,
      "step": 385850
    },
    {
      "epoch": 0.028636010738970107,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012284048404840486,
      "loss": 3.1614,
      "step": 385860
    },
    {
      "epoch": 0.028654653975128293,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012283848384838487,
      "loss": 3.2002,
      "step": 385870
    },
    {
      "epoch": 0.028673297211286475,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012283648364836485,
      "loss": 3.2187,
      "step": 385880
    },
    {
      "epoch": 0.02869194044744466,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012283448344834486,
      "loss": 3.1914,
      "step": 385890
    },
    {
      "epoch": 0.028710583683602842,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012283248324832484,
      "loss": 3.1667,
      "step": 385900
    },
    {
      "epoch": 0.028729226919761028,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012283048304830485,
      "loss": 3.2257,
      "step": 385910
    },
    {
      "epoch": 0.02874787015591921,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012282848284828483,
      "loss": 3.2337,
      "step": 385920
    },
    {
      "epoch": 0.028766513392077395,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012282648264826484,
      "loss": 3.1507,
      "step": 385930
    },
    {
      "epoch": 0.028785156628235577,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012282448244824482,
      "loss": 3.1335,
      "step": 385940
    },
    {
      "epoch": 0.02880379986439376,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012282248224822483,
      "loss": 3.2653,
      "step": 385950
    },
    {
      "epoch": 0.028822443100551945,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001228204820482048,
      "loss": 3.0282,
      "step": 385960
    },
    {
      "epoch": 0.028841086336710127,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012281848184818482,
      "loss": 3.2114,
      "step": 385970
    },
    {
      "epoch": 0.028859729572868312,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001228164816481648,
      "loss": 3.1624,
      "step": 385980
    },
    {
      "epoch": 0.028878372809026494,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001228144814481448,
      "loss": 3.3909,
      "step": 385990
    },
    {
      "epoch": 0.02889701604518468,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012281248124812482,
      "loss": 3.1112,
      "step": 386000
    },
    {
      "epoch": 0.028915659281342862,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012281048104810483,
      "loss": 3.1495,
      "step": 386010
    },
    {
      "epoch": 0.028934302517501048,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001228084808480848,
      "loss": 3.279,
      "step": 386020
    },
    {
      "epoch": 0.02895294575365923,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012280648064806482,
      "loss": 3.1707,
      "step": 386030
    },
    {
      "epoch": 0.028971588989817415,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001228044804480448,
      "loss": 3.2274,
      "step": 386040
    },
    {
      "epoch": 0.028990232225975597,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012280248024802481,
      "loss": 3.2065,
      "step": 386050
    },
    {
      "epoch": 0.029008875462133783,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001228004800480048,
      "loss": 3.1518,
      "step": 386060
    },
    {
      "epoch": 0.029027518698291965,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001227984798479848,
      "loss": 3.1099,
      "step": 386070
    },
    {
      "epoch": 0.02904616193445015,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012279647964796481,
      "loss": 3.0547,
      "step": 386080
    },
    {
      "epoch": 0.029064805170608332,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001227944794479448,
      "loss": 3.2246,
      "step": 386090
    },
    {
      "epoch": 0.029083448406766518,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001227924792479248,
      "loss": 3.1358,
      "step": 386100
    },
    {
      "epoch": 0.0291020916429247,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001227904790479048,
      "loss": 3.1079,
      "step": 386110
    },
    {
      "epoch": 0.02912073487908288,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001227884788478848,
      "loss": 3.1342,
      "step": 386120
    },
    {
      "epoch": 0.029139378115241067,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012278647864786478,
      "loss": 3.0768,
      "step": 386130
    },
    {
      "epoch": 0.02915802135139925,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001227844784478448,
      "loss": 3.2865,
      "step": 386140
    },
    {
      "epoch": 0.029176664587557435,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001227824782478248,
      "loss": 3.1634,
      "step": 386150
    },
    {
      "epoch": 0.029195307823715617,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012278047804780478,
      "loss": 3.1176,
      "step": 386160
    },
    {
      "epoch": 0.029213951059873802,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001227784778477848,
      "loss": 3.177,
      "step": 386170
    },
    {
      "epoch": 0.029232594296031984,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001227764776477648,
      "loss": 3.1249,
      "step": 386180
    },
    {
      "epoch": 0.02925123753219017,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012277447744774478,
      "loss": 3.1708,
      "step": 386190
    },
    {
      "epoch": 0.029269880768348352,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001227724772477248,
      "loss": 3.2134,
      "step": 386200
    },
    {
      "epoch": 0.029288524004506537,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012277047704770477,
      "loss": 3.2318,
      "step": 386210
    },
    {
      "epoch": 0.02930716724066472,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012276847684768478,
      "loss": 3.2365,
      "step": 386220
    },
    {
      "epoch": 0.029325810476822905,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012276647664766476,
      "loss": 3.3584,
      "step": 386230
    },
    {
      "epoch": 0.029344453712981087,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012276447644764477,
      "loss": 3.1439,
      "step": 386240
    },
    {
      "epoch": 0.029363096949139272,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012276247624762475,
      "loss": 3.2292,
      "step": 386250
    },
    {
      "epoch": 0.029381740185297454,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012276047604760476,
      "loss": 3.0904,
      "step": 386260
    },
    {
      "epoch": 0.02940038342145564,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012275847584758477,
      "loss": 3.3539,
      "step": 386270
    },
    {
      "epoch": 0.029419026657613822,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012275647564756475,
      "loss": 3.1117,
      "step": 386280
    },
    {
      "epoch": 0.029437669893772004,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012275447544754476,
      "loss": 3.0199,
      "step": 386290
    },
    {
      "epoch": 0.02945631312993019,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012275247524752474,
      "loss": 3.2229,
      "step": 386300
    },
    {
      "epoch": 0.02947495636608837,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012275047504750475,
      "loss": 3.2455,
      "step": 386310
    },
    {
      "epoch": 0.029493599602246557,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012274847484748476,
      "loss": 3.2024,
      "step": 386320
    },
    {
      "epoch": 0.02951224283840474,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012274647464746477,
      "loss": 3.1233,
      "step": 386330
    },
    {
      "epoch": 0.029530886074562925,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012274447444744475,
      "loss": 3.1425,
      "step": 386340
    },
    {
      "epoch": 0.029549529310721107,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012274247424742476,
      "loss": 3.2847,
      "step": 386350
    },
    {
      "epoch": 0.029568172546879292,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012274047404740474,
      "loss": 3.1986,
      "step": 386360
    },
    {
      "epoch": 0.029586815783037474,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012273847384738475,
      "loss": 3.1806,
      "step": 386370
    },
    {
      "epoch": 0.02960545901919566,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012273647364736474,
      "loss": 3.1608,
      "step": 386380
    },
    {
      "epoch": 0.02962410225535384,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012273447344734475,
      "loss": 3.2519,
      "step": 386390
    },
    {
      "epoch": 0.029642745491512027,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012273247324732473,
      "loss": 3.1367,
      "step": 386400
    },
    {
      "epoch": 0.02966138872767021,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012273047304730474,
      "loss": 3.1706,
      "step": 386410
    },
    {
      "epoch": 0.029680031963828395,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012272847284728472,
      "loss": 3.2228,
      "step": 386420
    },
    {
      "epoch": 0.029698675199986577,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012272647264726473,
      "loss": 3.1814,
      "step": 386430
    },
    {
      "epoch": 0.029717318436144762,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001227244724472447,
      "loss": 3.3405,
      "step": 386440
    },
    {
      "epoch": 0.029735961672302944,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012272247224722472,
      "loss": 3.1998,
      "step": 386450
    },
    {
      "epoch": 0.029754604908461126,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012272047204720473,
      "loss": 3.2028,
      "step": 386460
    },
    {
      "epoch": 0.029773248144619312,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00012271847184718474,
      "loss": 3.2099,
      "step": 386470
    },
    {
      "epoch": 0.029791891380777494,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012271647164716472,
      "loss": 3.2045,
      "step": 386480
    },
    {
      "epoch": 0.02981053461693568,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00012271447144714473,
      "loss": 3.1801,
      "step": 386490
    },
    {
      "epoch": 0.02982917785309386,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012271247124712474,
      "loss": 3.203,
      "step": 386500
    },
    {
      "epoch": 0.029847821089252047,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012271047104710472,
      "loss": 3.2136,
      "step": 386510
    },
    {
      "epoch": 0.02986646432541023,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012270847084708473,
      "loss": 3.1447,
      "step": 386520
    },
    {
      "epoch": 0.029885107561568414,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001227064706470647,
      "loss": 3.239,
      "step": 386530
    },
    {
      "epoch": 0.029903750797726596,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012270447044704472,
      "loss": 3.2691,
      "step": 386540
    },
    {
      "epoch": 0.029922394033884782,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001227024702470247,
      "loss": 3.1898,
      "step": 386550
    },
    {
      "epoch": 0.029941037270042964,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001227004700470047,
      "loss": 3.2846,
      "step": 386560
    },
    {
      "epoch": 0.02995968050620115,
      "grad_norm": 0.875,
      "learning_rate": 0.0001226984698469847,
      "loss": 3.2671,
      "step": 386570
    },
    {
      "epoch": 0.02997832374235933,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001226964696469647,
      "loss": 3.2402,
      "step": 386580
    },
    {
      "epoch": 0.029996966978517517,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012269446944694468,
      "loss": 3.2255,
      "step": 386590
    },
    {
      "epoch": 0.0300156102146757,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001226924692469247,
      "loss": 3.1749,
      "step": 386600
    },
    {
      "epoch": 0.030034253450833884,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012269046904690468,
      "loss": 3.1509,
      "step": 386610
    },
    {
      "epoch": 0.030052896686992066,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012268846884688468,
      "loss": 3.2976,
      "step": 386620
    },
    {
      "epoch": 0.03007153992315025,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001226864686468647,
      "loss": 3.2174,
      "step": 386630
    },
    {
      "epoch": 0.030090183159308434,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001226844684468447,
      "loss": 3.2543,
      "step": 386640
    },
    {
      "epoch": 0.030108826395466616,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012268246824682468,
      "loss": 3.1735,
      "step": 386650
    },
    {
      "epoch": 0.0301274696316248,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001226804680468047,
      "loss": 3.2026,
      "step": 386660
    },
    {
      "epoch": 0.030146112867782984,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001226784678467847,
      "loss": 3.2272,
      "step": 386670
    },
    {
      "epoch": 0.03016475610394117,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012267646764676469,
      "loss": 3.2671,
      "step": 386680
    },
    {
      "epoch": 0.03018339934009935,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001226744674467447,
      "loss": 3.122,
      "step": 386690
    },
    {
      "epoch": 0.030202042576257537,
      "grad_norm": 0.625,
      "learning_rate": 0.00012267246724672468,
      "loss": 3.2669,
      "step": 386700
    },
    {
      "epoch": 0.03022068581241572,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012267046704670469,
      "loss": 3.0996,
      "step": 386710
    },
    {
      "epoch": 0.030239329048573904,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012266846684668467,
      "loss": 3.2128,
      "step": 386720
    },
    {
      "epoch": 0.030257972284732086,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012266646664666468,
      "loss": 3.2366,
      "step": 386730
    },
    {
      "epoch": 0.03027661552089027,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012266446644664466,
      "loss": 3.0944,
      "step": 386740
    },
    {
      "epoch": 0.030295258757048454,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012266246624662467,
      "loss": 3.1564,
      "step": 386750
    },
    {
      "epoch": 0.03031390199320664,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012266046604660465,
      "loss": 3.2049,
      "step": 386760
    },
    {
      "epoch": 0.03033254522936482,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012265846584658466,
      "loss": 3.2891,
      "step": 386770
    },
    {
      "epoch": 0.030351188465523007,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012265646564656467,
      "loss": 3.119,
      "step": 386780
    },
    {
      "epoch": 0.03036983170168119,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012265446544654468,
      "loss": 3.2517,
      "step": 386790
    },
    {
      "epoch": 0.03038847493783937,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012265246524652466,
      "loss": 3.1962,
      "step": 386800
    },
    {
      "epoch": 0.030407118173997556,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012265046504650467,
      "loss": 3.1572,
      "step": 386810
    },
    {
      "epoch": 0.03042576141015574,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012264846484648465,
      "loss": 3.2581,
      "step": 386820
    },
    {
      "epoch": 0.030444404646313924,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012264646464646466,
      "loss": 3.1894,
      "step": 386830
    },
    {
      "epoch": 0.030463047882472106,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012264446444644464,
      "loss": 3.2004,
      "step": 386840
    },
    {
      "epoch": 0.03048169111863029,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012264246424642465,
      "loss": 3.1468,
      "step": 386850
    },
    {
      "epoch": 0.030500334354788473,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012264046404640466,
      "loss": 3.1847,
      "step": 386860
    },
    {
      "epoch": 0.03051897759094666,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012263846384638464,
      "loss": 3.215,
      "step": 386870
    },
    {
      "epoch": 0.03053762082710484,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012263646364636465,
      "loss": 3.2542,
      "step": 386880
    },
    {
      "epoch": 0.030556264063263026,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012263446344634463,
      "loss": 3.1187,
      "step": 386890
    },
    {
      "epoch": 0.03057490729942121,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012263246324632464,
      "loss": 3.266,
      "step": 386900
    },
    {
      "epoch": 0.030593550535579394,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012263046304630462,
      "loss": 3.0153,
      "step": 386910
    },
    {
      "epoch": 0.030612193771737576,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012262846284628463,
      "loss": 3.1816,
      "step": 386920
    },
    {
      "epoch": 0.03063083700789576,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012262646264626462,
      "loss": 3.0855,
      "step": 386930
    },
    {
      "epoch": 0.030649480244053944,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012262446244624462,
      "loss": 3.1597,
      "step": 386940
    },
    {
      "epoch": 0.03066812348021213,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012262246224622463,
      "loss": 3.1698,
      "step": 386950
    },
    {
      "epoch": 0.03068676671637031,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012262046204620464,
      "loss": 3.1689,
      "step": 386960
    },
    {
      "epoch": 0.030705409952528493,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012261846184618463,
      "loss": 3.1543,
      "step": 386970
    },
    {
      "epoch": 0.03072405318868668,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012261646164616463,
      "loss": 3.2165,
      "step": 386980
    },
    {
      "epoch": 0.03074269642484486,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012261446144614462,
      "loss": 3.2395,
      "step": 386990
    },
    {
      "epoch": 0.030761339661003046,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012261246124612463,
      "loss": 3.1445,
      "step": 387000
    },
    {
      "epoch": 0.030779982897161228,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001226104610461046,
      "loss": 3.2181,
      "step": 387010
    },
    {
      "epoch": 0.030798626133319414,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012260846084608462,
      "loss": 3.1541,
      "step": 387020
    },
    {
      "epoch": 0.030817269369477596,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001226064606460646,
      "loss": 3.2718,
      "step": 387030
    },
    {
      "epoch": 0.03083591260563578,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001226044604460446,
      "loss": 3.1118,
      "step": 387040
    },
    {
      "epoch": 0.030854555841793963,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001226024602460246,
      "loss": 3.1906,
      "step": 387050
    },
    {
      "epoch": 0.03087319907795215,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001226004600460046,
      "loss": 3.2349,
      "step": 387060
    },
    {
      "epoch": 0.03089184231411033,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001225984598459846,
      "loss": 3.2701,
      "step": 387070
    },
    {
      "epoch": 0.030910485550268516,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001225964596459646,
      "loss": 3.2243,
      "step": 387080
    },
    {
      "epoch": 0.030929128786426698,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001225944594459446,
      "loss": 3.1039,
      "step": 387090
    },
    {
      "epoch": 0.030947772022584884,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001225924592459246,
      "loss": 3.1401,
      "step": 387100
    },
    {
      "epoch": 0.030966415258743066,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012259045904590462,
      "loss": 3.1175,
      "step": 387110
    },
    {
      "epoch": 0.03098505849490125,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001225884588458846,
      "loss": 3.1392,
      "step": 387120
    },
    {
      "epoch": 0.031003701731059433,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001225864586458646,
      "loss": 3.1453,
      "step": 387130
    },
    {
      "epoch": 0.031022344967217615,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001225844584458446,
      "loss": 3.2611,
      "step": 387140
    },
    {
      "epoch": 0.0310409882033758,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001225824582458246,
      "loss": 3.1875,
      "step": 387150
    },
    {
      "epoch": 0.031059631439533983,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012258045804580458,
      "loss": 3.1601,
      "step": 387160
    },
    {
      "epoch": 0.03107827467569217,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001225784578457846,
      "loss": 3.1842,
      "step": 387170
    },
    {
      "epoch": 0.03109691791185035,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012257645764576457,
      "loss": 3.075,
      "step": 387180
    },
    {
      "epoch": 0.031115561148008536,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012257445744574458,
      "loss": 3.2299,
      "step": 387190
    },
    {
      "epoch": 0.031134204384166718,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00012257245724572456,
      "loss": 3.3017,
      "step": 387200
    },
    {
      "epoch": 0.031152847620324903,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012257045704570457,
      "loss": 3.0673,
      "step": 387210
    },
    {
      "epoch": 0.031171490856483085,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012256845684568456,
      "loss": 3.0884,
      "step": 387220
    },
    {
      "epoch": 0.03119013409264127,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012256645664566456,
      "loss": 3.1817,
      "step": 387230
    },
    {
      "epoch": 0.031208777328799453,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012256445644564455,
      "loss": 3.2273,
      "step": 387240
    },
    {
      "epoch": 0.03122742056495764,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012256245624562456,
      "loss": 3.1027,
      "step": 387250
    },
    {
      "epoch": 0.03124606380111582,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012256045604560457,
      "loss": 3.1603,
      "step": 387260
    },
    {
      "epoch": 0.031264707037274006,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012255845584558457,
      "loss": 3.2848,
      "step": 387270
    },
    {
      "epoch": 0.03128335027343219,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012255645564556458,
      "loss": 3.1416,
      "step": 387280
    },
    {
      "epoch": 0.03130199350959037,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012255445544554457,
      "loss": 3.1105,
      "step": 387290
    },
    {
      "epoch": 0.03132063674574856,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012255245524552457,
      "loss": 3.2117,
      "step": 387300
    },
    {
      "epoch": 0.03133927998190674,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012255045504550456,
      "loss": 3.2301,
      "step": 387310
    },
    {
      "epoch": 0.03135792321806492,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012254845484548457,
      "loss": 3.2355,
      "step": 387320
    },
    {
      "epoch": 0.031376566454223105,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012254645464546455,
      "loss": 3.1635,
      "step": 387330
    },
    {
      "epoch": 0.03139520969038129,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012254445444544456,
      "loss": 3.1537,
      "step": 387340
    },
    {
      "epoch": 0.031413852926539476,
      "grad_norm": 0.75,
      "learning_rate": 0.00012254245424542454,
      "loss": 3.1673,
      "step": 387350
    },
    {
      "epoch": 0.03143249616269766,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012254045404540455,
      "loss": 3.231,
      "step": 387360
    },
    {
      "epoch": 0.03145113939885584,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012253845384538453,
      "loss": 3.0709,
      "step": 387370
    },
    {
      "epoch": 0.03146978263501402,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012253645364536454,
      "loss": 3.0779,
      "step": 387380
    },
    {
      "epoch": 0.03148842587117221,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012253445344534452,
      "loss": 3.1332,
      "step": 387390
    },
    {
      "epoch": 0.03150706910733039,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012253245324532453,
      "loss": 3.2363,
      "step": 387400
    },
    {
      "epoch": 0.031525712343488575,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012253045304530454,
      "loss": 3.1566,
      "step": 387410
    },
    {
      "epoch": 0.03154435557964676,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012252845284528455,
      "loss": 3.1143,
      "step": 387420
    },
    {
      "epoch": 0.031562998815804946,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012252645264526453,
      "loss": 3.1905,
      "step": 387430
    },
    {
      "epoch": 0.03158164205196313,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012252445244524454,
      "loss": 3.1843,
      "step": 387440
    },
    {
      "epoch": 0.03160028528812131,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012252245224522455,
      "loss": 3.2724,
      "step": 387450
    },
    {
      "epoch": 0.03161892852427949,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012252045204520453,
      "loss": 3.2827,
      "step": 387460
    },
    {
      "epoch": 0.03163757176043768,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012251845184518454,
      "loss": 3.2121,
      "step": 387470
    },
    {
      "epoch": 0.03165621499659586,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012251645164516452,
      "loss": 3.1841,
      "step": 387480
    },
    {
      "epoch": 0.031674858232754045,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012251445144514453,
      "loss": 3.1528,
      "step": 387490
    },
    {
      "epoch": 0.03169350146891223,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012251245124512451,
      "loss": 3.0724,
      "step": 387500
    },
    {
      "epoch": 0.03171214470507041,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012251045104510452,
      "loss": 3.2207,
      "step": 387510
    },
    {
      "epoch": 0.0317307879412286,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001225084508450845,
      "loss": 3.1571,
      "step": 387520
    },
    {
      "epoch": 0.03174943117738678,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012250645064506451,
      "loss": 3.1296,
      "step": 387530
    },
    {
      "epoch": 0.03176807441354496,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001225044504450445,
      "loss": 3.1156,
      "step": 387540
    },
    {
      "epoch": 0.031786717649703145,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001225024502450245,
      "loss": 3.1597,
      "step": 387550
    },
    {
      "epoch": 0.031805360885861333,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001225004500450045,
      "loss": 3.1145,
      "step": 387560
    },
    {
      "epoch": 0.031824004122019516,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001224984498449845,
      "loss": 3.2411,
      "step": 387570
    },
    {
      "epoch": 0.0318426473581777,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001224964496449645,
      "loss": 3.2016,
      "step": 387580
    },
    {
      "epoch": 0.03186129059433588,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012249444944494451,
      "loss": 3.1343,
      "step": 387590
    },
    {
      "epoch": 0.03187993383049407,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001224924492449245,
      "loss": 3.154,
      "step": 387600
    },
    {
      "epoch": 0.03189857706665225,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001224904490449045,
      "loss": 3.3274,
      "step": 387610
    },
    {
      "epoch": 0.03191722030281043,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001224884488448845,
      "loss": 3.2196,
      "step": 387620
    },
    {
      "epoch": 0.031935863538968615,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001224864486448645,
      "loss": 3.2243,
      "step": 387630
    },
    {
      "epoch": 0.031954506775126804,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012248444844484448,
      "loss": 3.2218,
      "step": 387640
    },
    {
      "epoch": 0.031973150011284986,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001224824482448245,
      "loss": 3.3024,
      "step": 387650
    },
    {
      "epoch": 0.03199179324744317,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001224804480448045,
      "loss": 3.2243,
      "step": 387660
    },
    {
      "epoch": 0.03201043648360135,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00012247844784478448,
      "loss": 3.3115,
      "step": 387670
    },
    {
      "epoch": 0.03202907971975953,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001224764476447645,
      "loss": 3.1847,
      "step": 387680
    },
    {
      "epoch": 0.03204772295591772,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012247444744474447,
      "loss": 3.1203,
      "step": 387690
    },
    {
      "epoch": 0.0320663661920759,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012247244724472448,
      "loss": 3.3218,
      "step": 387700
    },
    {
      "epoch": 0.032085009428234085,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012247044704470446,
      "loss": 3.2916,
      "step": 387710
    },
    {
      "epoch": 0.03210365266439227,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012246844684468447,
      "loss": 3.1348,
      "step": 387720
    },
    {
      "epoch": 0.032122295900550456,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012246644664466448,
      "loss": 3.3296,
      "step": 387730
    },
    {
      "epoch": 0.03214093913670864,
      "grad_norm": 0.875,
      "learning_rate": 0.0001224644464446445,
      "loss": 3.1567,
      "step": 387740
    },
    {
      "epoch": 0.03215958237286682,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012246244624462447,
      "loss": 3.1339,
      "step": 387750
    },
    {
      "epoch": 0.032178225609025,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012246044604460448,
      "loss": 3.1735,
      "step": 387760
    },
    {
      "epoch": 0.03219686884518319,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012245844584458446,
      "loss": 3.1466,
      "step": 387770
    },
    {
      "epoch": 0.03221551208134137,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012245644564456447,
      "loss": 3.1428,
      "step": 387780
    },
    {
      "epoch": 0.032234155317499555,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012245444544454445,
      "loss": 3.1268,
      "step": 387790
    },
    {
      "epoch": 0.03225279855365774,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012245244524452446,
      "loss": 3.1268,
      "step": 387800
    },
    {
      "epoch": 0.032271441789815926,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012245044504450444,
      "loss": 3.1348,
      "step": 387810
    },
    {
      "epoch": 0.03229008502597411,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012244844484448445,
      "loss": 3.1432,
      "step": 387820
    },
    {
      "epoch": 0.03230872826213229,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012244644464446444,
      "loss": 3.1088,
      "step": 387830
    },
    {
      "epoch": 0.03232737149829047,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012244444444444445,
      "loss": 3.0804,
      "step": 387840
    },
    {
      "epoch": 0.032346014734448654,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012244244424442445,
      "loss": 3.2192,
      "step": 387850
    },
    {
      "epoch": 0.03236465797060684,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012244044404440444,
      "loss": 3.1722,
      "step": 387860
    },
    {
      "epoch": 0.032383301206765025,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012243844384438445,
      "loss": 3.3106,
      "step": 387870
    },
    {
      "epoch": 0.03240194444292321,
      "grad_norm": 0.75,
      "learning_rate": 0.00012243644364436443,
      "loss": 3.153,
      "step": 387880
    },
    {
      "epoch": 0.03242058767908139,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012243444344434444,
      "loss": 3.1882,
      "step": 387890
    },
    {
      "epoch": 0.03243923091523958,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012243244324432445,
      "loss": 3.2477,
      "step": 387900
    },
    {
      "epoch": 0.03245787415139776,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012243044304430445,
      "loss": 3.171,
      "step": 387910
    },
    {
      "epoch": 0.03247651738755594,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012242844284428444,
      "loss": 3.1212,
      "step": 387920
    },
    {
      "epoch": 0.032495160623714124,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012242644264426445,
      "loss": 3.0809,
      "step": 387930
    },
    {
      "epoch": 0.03251380385987231,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012242444244424443,
      "loss": 3.0738,
      "step": 387940
    },
    {
      "epoch": 0.032532447096030495,
      "grad_norm": 0.75,
      "learning_rate": 0.00012242244224422444,
      "loss": 3.1132,
      "step": 387950
    },
    {
      "epoch": 0.03255109033218868,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012242044204420442,
      "loss": 3.1202,
      "step": 387960
    },
    {
      "epoch": 0.03256973356834686,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012241844184418443,
      "loss": 3.189,
      "step": 387970
    },
    {
      "epoch": 0.03258837680450505,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001224164416441644,
      "loss": 3.1668,
      "step": 387980
    },
    {
      "epoch": 0.03260702004066323,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012241444144414442,
      "loss": 3.3287,
      "step": 387990
    },
    {
      "epoch": 0.03262566327682141,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001224124412441244,
      "loss": 3.1734,
      "step": 388000
    },
    {
      "epoch": 0.032644306512979594,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001224104410441044,
      "loss": 3.1825,
      "step": 388010
    },
    {
      "epoch": 0.032662949749137776,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001224084408440844,
      "loss": 3.2174,
      "step": 388020
    },
    {
      "epoch": 0.032681592985295965,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001224064406440644,
      "loss": 3.1762,
      "step": 388030
    },
    {
      "epoch": 0.03270023622145415,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001224044404440444,
      "loss": 3.2111,
      "step": 388040
    },
    {
      "epoch": 0.03271887945761233,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012240244024402442,
      "loss": 3.2793,
      "step": 388050
    },
    {
      "epoch": 0.03273752269377051,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012240044004400443,
      "loss": 3.1503,
      "step": 388060
    },
    {
      "epoch": 0.0327561659299287,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001223984398439844,
      "loss": 3.2414,
      "step": 388070
    },
    {
      "epoch": 0.03277480916608688,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012239643964396442,
      "loss": 3.23,
      "step": 388080
    },
    {
      "epoch": 0.032793452402245064,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001223944394439444,
      "loss": 3.1377,
      "step": 388090
    },
    {
      "epoch": 0.032812095638403246,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001223924392439244,
      "loss": 3.1779,
      "step": 388100
    },
    {
      "epoch": 0.032830738874561435,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001223904390439044,
      "loss": 3.2969,
      "step": 388110
    },
    {
      "epoch": 0.03284938211071962,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001223884388438844,
      "loss": 3.2168,
      "step": 388120
    },
    {
      "epoch": 0.0328680253468778,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012238643864386439,
      "loss": 3.1847,
      "step": 388130
    },
    {
      "epoch": 0.03288666858303598,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001223844384438444,
      "loss": 3.1064,
      "step": 388140
    },
    {
      "epoch": 0.03290531181919417,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012238243824382438,
      "loss": 3.2057,
      "step": 388150
    },
    {
      "epoch": 0.03292395505535235,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012238043804380439,
      "loss": 3.1444,
      "step": 388160
    },
    {
      "epoch": 0.032942598291510534,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012237843784378437,
      "loss": 3.1799,
      "step": 388170
    },
    {
      "epoch": 0.032961241527668717,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012237643764376438,
      "loss": 3.2626,
      "step": 388180
    },
    {
      "epoch": 0.0329798847638269,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012237443744374436,
      "loss": 3.2341,
      "step": 388190
    },
    {
      "epoch": 0.03299852799998509,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012237243724372437,
      "loss": 3.1561,
      "step": 388200
    },
    {
      "epoch": 0.03301717123614327,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012237043704370438,
      "loss": 3.2036,
      "step": 388210
    },
    {
      "epoch": 0.03303581447230145,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012236843684368439,
      "loss": 3.2265,
      "step": 388220
    },
    {
      "epoch": 0.033054457708459634,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012236643664366437,
      "loss": 3.2745,
      "step": 388230
    },
    {
      "epoch": 0.03307310094461782,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012236443644364438,
      "loss": 3.2335,
      "step": 388240
    },
    {
      "epoch": 0.033091744180776005,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001223624362436244,
      "loss": 3.173,
      "step": 388250
    },
    {
      "epoch": 0.03311038741693419,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012236043604360437,
      "loss": 3.2255,
      "step": 388260
    },
    {
      "epoch": 0.03312903065309237,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012235843584358438,
      "loss": 3.203,
      "step": 388270
    },
    {
      "epoch": 0.03314767388925056,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012235643564356436,
      "loss": 3.1553,
      "step": 388280
    },
    {
      "epoch": 0.03316631712540874,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012235443544354437,
      "loss": 3.1989,
      "step": 388290
    },
    {
      "epoch": 0.03318496036156692,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012235243524352435,
      "loss": 3.1666,
      "step": 388300
    },
    {
      "epoch": 0.033203603597725104,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012235043504350436,
      "loss": 3.1791,
      "step": 388310
    },
    {
      "epoch": 0.03322224683388329,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012234843484348434,
      "loss": 3.2599,
      "step": 388320
    },
    {
      "epoch": 0.033240890070041475,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012234643464346435,
      "loss": 3.0903,
      "step": 388330
    },
    {
      "epoch": 0.03325953330619966,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012234443444344433,
      "loss": 3.0676,
      "step": 388340
    },
    {
      "epoch": 0.03327817654235784,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012234243424342434,
      "loss": 3.1834,
      "step": 388350
    },
    {
      "epoch": 0.03329681977851602,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012234043404340435,
      "loss": 3.3188,
      "step": 388360
    },
    {
      "epoch": 0.03331546301467421,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012233843384338436,
      "loss": 3.1102,
      "step": 388370
    },
    {
      "epoch": 0.03333410625083239,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00012233643364336434,
      "loss": 3.1419,
      "step": 388380
    },
    {
      "epoch": 0.033352749486990574,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012233443344334435,
      "loss": 3.2479,
      "step": 388390
    },
    {
      "epoch": 0.033371392723148756,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012233243324332433,
      "loss": 3.1942,
      "step": 388400
    },
    {
      "epoch": 0.033390035959306945,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012233043304330434,
      "loss": 3.2294,
      "step": 388410
    },
    {
      "epoch": 0.03340867919546513,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012232843284328433,
      "loss": 3.2951,
      "step": 388420
    },
    {
      "epoch": 0.03342732243162331,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012232643264326433,
      "loss": 3.3637,
      "step": 388430
    },
    {
      "epoch": 0.03344596566778149,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012232443244324434,
      "loss": 3.2225,
      "step": 388440
    },
    {
      "epoch": 0.03346460890393968,
      "grad_norm": 0.75,
      "learning_rate": 0.00012232243224322433,
      "loss": 3.2543,
      "step": 388450
    },
    {
      "epoch": 0.03348325214009786,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012232043204320433,
      "loss": 3.2076,
      "step": 388460
    },
    {
      "epoch": 0.033501895376256044,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012231843184318432,
      "loss": 3.1711,
      "step": 388470
    },
    {
      "epoch": 0.033520538612414226,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012231643164316433,
      "loss": 3.2691,
      "step": 388480
    },
    {
      "epoch": 0.033539181848572415,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001223144314431443,
      "loss": 3.2551,
      "step": 388490
    },
    {
      "epoch": 0.0335578250847306,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012231243124312432,
      "loss": 3.0991,
      "step": 388500
    },
    {
      "epoch": 0.03357646832088878,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001223104310431043,
      "loss": 2.9948,
      "step": 388510
    },
    {
      "epoch": 0.03359511155704696,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001223084308430843,
      "loss": 3.1987,
      "step": 388520
    },
    {
      "epoch": 0.03361375479320514,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012230643064306432,
      "loss": 3.1333,
      "step": 388530
    },
    {
      "epoch": 0.03363239802936333,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012230443044304433,
      "loss": 3.0855,
      "step": 388540
    },
    {
      "epoch": 0.033651041265521514,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001223024302430243,
      "loss": 3.1776,
      "step": 388550
    },
    {
      "epoch": 0.033669684501679696,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012230043004300432,
      "loss": 3.2047,
      "step": 388560
    },
    {
      "epoch": 0.03368832773783788,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001222984298429843,
      "loss": 3.1348,
      "step": 388570
    },
    {
      "epoch": 0.03370697097399607,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001222964296429643,
      "loss": 3.1082,
      "step": 388580
    },
    {
      "epoch": 0.03372561421015425,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001222944294429443,
      "loss": 3.2104,
      "step": 388590
    },
    {
      "epoch": 0.03374425744631243,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001222924292429243,
      "loss": 3.244,
      "step": 388600
    },
    {
      "epoch": 0.03376290068247061,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012229042904290428,
      "loss": 3.1129,
      "step": 388610
    },
    {
      "epoch": 0.0337815439186288,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001222884288428843,
      "loss": 3.124,
      "step": 388620
    },
    {
      "epoch": 0.033800187154786984,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012228642864286427,
      "loss": 3.1876,
      "step": 388630
    },
    {
      "epoch": 0.033818830390945166,
      "grad_norm": 0.75,
      "learning_rate": 0.00012228442844284428,
      "loss": 3.237,
      "step": 388640
    },
    {
      "epoch": 0.03383747362710335,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001222824282428243,
      "loss": 3.2217,
      "step": 388650
    },
    {
      "epoch": 0.03385611686326154,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012228042804280427,
      "loss": 3.096,
      "step": 388660
    },
    {
      "epoch": 0.03387476009941972,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012227842784278428,
      "loss": 3.131,
      "step": 388670
    },
    {
      "epoch": 0.0338934033355779,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001222764276427643,
      "loss": 3.2773,
      "step": 388680
    },
    {
      "epoch": 0.03391204657173608,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001222744274427443,
      "loss": 3.0937,
      "step": 388690
    },
    {
      "epoch": 0.033930689807894265,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012227242724272428,
      "loss": 3.1797,
      "step": 388700
    },
    {
      "epoch": 0.033949333044052454,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001222704270427043,
      "loss": 3.1307,
      "step": 388710
    },
    {
      "epoch": 0.033967976280210636,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012226842684268427,
      "loss": 3.2065,
      "step": 388720
    },
    {
      "epoch": 0.03398661951636882,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012226642664266428,
      "loss": 3.3444,
      "step": 388730
    },
    {
      "epoch": 0.034005262752527,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012226442644264427,
      "loss": 3.1053,
      "step": 388740
    },
    {
      "epoch": 0.03402390598868519,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012226242624262427,
      "loss": 3.1961,
      "step": 388750
    },
    {
      "epoch": 0.03404254922484337,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012226042604260426,
      "loss": 3.2088,
      "step": 388760
    },
    {
      "epoch": 0.03406119246100155,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012225842584258427,
      "loss": 3.3337,
      "step": 388770
    },
    {
      "epoch": 0.034079835697159735,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012225642564256425,
      "loss": 3.238,
      "step": 388780
    },
    {
      "epoch": 0.034098478933317924,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012225442544254426,
      "loss": 3.1751,
      "step": 388790
    },
    {
      "epoch": 0.034117122169476106,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012225242524252424,
      "loss": 3.1422,
      "step": 388800
    },
    {
      "epoch": 0.03413576540563429,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012225042504250425,
      "loss": 3.2936,
      "step": 388810
    },
    {
      "epoch": 0.03415440864179247,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012224842484248423,
      "loss": 3.1525,
      "step": 388820
    },
    {
      "epoch": 0.03417305187795066,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012224642464246424,
      "loss": 3.203,
      "step": 388830
    },
    {
      "epoch": 0.03419169511410884,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012224442444244425,
      "loss": 3.1786,
      "step": 388840
    },
    {
      "epoch": 0.034210338350267024,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012224242424242426,
      "loss": 3.2194,
      "step": 388850
    },
    {
      "epoch": 0.034228981586425206,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012224042404240427,
      "loss": 3.2064,
      "step": 388860
    },
    {
      "epoch": 0.03424762482258339,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012223842384238425,
      "loss": 3.163,
      "step": 388870
    },
    {
      "epoch": 0.03426626805874158,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012223642364236426,
      "loss": 3.1053,
      "step": 388880
    },
    {
      "epoch": 0.03428491129489976,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012223442344234424,
      "loss": 3.1672,
      "step": 388890
    },
    {
      "epoch": 0.03430355453105794,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012223242324232425,
      "loss": 3.1675,
      "step": 388900
    },
    {
      "epoch": 0.03432219776721612,
      "grad_norm": 0.9375,
      "learning_rate": 0.00012223042304230423,
      "loss": 3.2301,
      "step": 388910
    },
    {
      "epoch": 0.03434084100337431,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012222842284228424,
      "loss": 3.1848,
      "step": 388920
    },
    {
      "epoch": 0.034359484239532494,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012222642264226422,
      "loss": 3.1943,
      "step": 388930
    },
    {
      "epoch": 0.034378127475690676,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012222442244224423,
      "loss": 3.2771,
      "step": 388940
    },
    {
      "epoch": 0.03439677071184886,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012222242224222421,
      "loss": 3.1892,
      "step": 388950
    },
    {
      "epoch": 0.03441541394800705,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012222042204220422,
      "loss": 3.2016,
      "step": 388960
    },
    {
      "epoch": 0.03443405718416523,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001222184218421842,
      "loss": 3.15,
      "step": 388970
    },
    {
      "epoch": 0.03445270042032341,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012221642164216421,
      "loss": 3.1128,
      "step": 388980
    },
    {
      "epoch": 0.03447134365648159,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012221442144214422,
      "loss": 3.1656,
      "step": 388990
    },
    {
      "epoch": 0.03448998689263978,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012221242124212423,
      "loss": 3.0847,
      "step": 389000
    },
    {
      "epoch": 0.034508630128797964,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012221042104210421,
      "loss": 3.1578,
      "step": 389010
    },
    {
      "epoch": 0.034527273364956146,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012220842084208422,
      "loss": 3.2428,
      "step": 389020
    },
    {
      "epoch": 0.03454591660111433,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012220642064206423,
      "loss": 3.1608,
      "step": 389030
    },
    {
      "epoch": 0.03456455983727251,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012220442044204421,
      "loss": 3.2287,
      "step": 389040
    },
    {
      "epoch": 0.0345832030734307,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012220242024202422,
      "loss": 3.1719,
      "step": 389050
    },
    {
      "epoch": 0.03460184630958888,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001222004200420042,
      "loss": 3.1886,
      "step": 389060
    },
    {
      "epoch": 0.03462048954574706,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012219841984198421,
      "loss": 3.1348,
      "step": 389070
    },
    {
      "epoch": 0.034639132781905245,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001221964196419642,
      "loss": 3.2584,
      "step": 389080
    },
    {
      "epoch": 0.034657776018063434,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001221944194419442,
      "loss": 3.2955,
      "step": 389090
    },
    {
      "epoch": 0.034676419254221616,
      "grad_norm": 0.625,
      "learning_rate": 0.0001221924192419242,
      "loss": 3.2538,
      "step": 389100
    },
    {
      "epoch": 0.0346950624903798,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001221904190419042,
      "loss": 3.176,
      "step": 389110
    },
    {
      "epoch": 0.03471370572653798,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012218841884188418,
      "loss": 3.2344,
      "step": 389120
    },
    {
      "epoch": 0.03473234896269617,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001221864186418642,
      "loss": 3.2936,
      "step": 389130
    },
    {
      "epoch": 0.03475099219885435,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012218441844184417,
      "loss": 3.2643,
      "step": 389140
    },
    {
      "epoch": 0.03476963543501253,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012218241824182418,
      "loss": 3.1799,
      "step": 389150
    },
    {
      "epoch": 0.034788278671170715,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001221804180418042,
      "loss": 3.1714,
      "step": 389160
    },
    {
      "epoch": 0.034806921907328904,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001221784178417842,
      "loss": 3.0411,
      "step": 389170
    },
    {
      "epoch": 0.034825565143487086,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012217641764176418,
      "loss": 3.1449,
      "step": 389180
    },
    {
      "epoch": 0.03484420837964527,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001221744174417442,
      "loss": 3.1364,
      "step": 389190
    },
    {
      "epoch": 0.03486285161580345,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012217241724172417,
      "loss": 3.1618,
      "step": 389200
    },
    {
      "epoch": 0.03488149485196163,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012217041704170418,
      "loss": 3.1632,
      "step": 389210
    },
    {
      "epoch": 0.03490013808811982,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012216841684168416,
      "loss": 3.2034,
      "step": 389220
    },
    {
      "epoch": 0.034918781324278,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012216641664166417,
      "loss": 3.0724,
      "step": 389230
    },
    {
      "epoch": 0.034937424560436185,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012216441644164418,
      "loss": 3.2317,
      "step": 389240
    },
    {
      "epoch": 0.03495606779659437,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012216241624162416,
      "loss": 3.0996,
      "step": 389250
    },
    {
      "epoch": 0.034974711032752556,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012216041604160417,
      "loss": 3.1346,
      "step": 389260
    },
    {
      "epoch": 0.03499335426891074,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012215841584158415,
      "loss": 3.212,
      "step": 389270
    },
    {
      "epoch": 0.03501199750506892,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012215641564156416,
      "loss": 3.1059,
      "step": 389280
    },
    {
      "epoch": 0.0350306407412271,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012215441544154415,
      "loss": 3.1919,
      "step": 389290
    },
    {
      "epoch": 0.03504928397738529,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012215241524152415,
      "loss": 3.1364,
      "step": 389300
    },
    {
      "epoch": 0.03506792721354347,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012215041504150416,
      "loss": 3.1437,
      "step": 389310
    },
    {
      "epoch": 0.035086570449701655,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012214841484148417,
      "loss": 3.137,
      "step": 389320
    },
    {
      "epoch": 0.03510521368585984,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012214641464146415,
      "loss": 3.2579,
      "step": 389330
    },
    {
      "epoch": 0.035123856922018026,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012214441444144416,
      "loss": 3.2721,
      "step": 389340
    },
    {
      "epoch": 0.03514250015817621,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012214241424142415,
      "loss": 3.1407,
      "step": 389350
    },
    {
      "epoch": 0.03516114339433439,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012214041404140415,
      "loss": 3.0655,
      "step": 389360
    },
    {
      "epoch": 0.03517978663049257,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012213841384138414,
      "loss": 3.1407,
      "step": 389370
    },
    {
      "epoch": 0.035198429866650754,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012213641364136415,
      "loss": 3.0317,
      "step": 389380
    },
    {
      "epoch": 0.03521707310280894,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012213441344134413,
      "loss": 3.1839,
      "step": 389390
    },
    {
      "epoch": 0.035235716338967125,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012213241324132414,
      "loss": 3.2328,
      "step": 389400
    },
    {
      "epoch": 0.03525435957512531,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012213041304130412,
      "loss": 3.1429,
      "step": 389410
    },
    {
      "epoch": 0.03527300281128349,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012212841284128413,
      "loss": 3.3859,
      "step": 389420
    },
    {
      "epoch": 0.03529164604744168,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012212641264126414,
      "loss": 3.1365,
      "step": 389430
    },
    {
      "epoch": 0.03531028928359986,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012212441244124412,
      "loss": 3.1281,
      "step": 389440
    },
    {
      "epoch": 0.03532893251975804,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012212241224122413,
      "loss": 3.2785,
      "step": 389450
    },
    {
      "epoch": 0.035347575755916225,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001221204120412041,
      "loss": 3.1622,
      "step": 389460
    },
    {
      "epoch": 0.035366218992074414,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012211841184118412,
      "loss": 3.0591,
      "step": 389470
    },
    {
      "epoch": 0.035384862228232596,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012211641164116413,
      "loss": 3.3186,
      "step": 389480
    },
    {
      "epoch": 0.03540350546439078,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012211441144114414,
      "loss": 3.1474,
      "step": 389490
    },
    {
      "epoch": 0.03542214870054896,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012211241124112412,
      "loss": 3.2155,
      "step": 389500
    },
    {
      "epoch": 0.03544079193670715,
      "grad_norm": 0.9375,
      "learning_rate": 0.00012211041104110413,
      "loss": 3.2782,
      "step": 389510
    },
    {
      "epoch": 0.03545943517286533,
      "grad_norm": 0.75,
      "learning_rate": 0.0001221084108410841,
      "loss": 3.282,
      "step": 389520
    },
    {
      "epoch": 0.03547807840902351,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012210641064106412,
      "loss": 3.1781,
      "step": 389530
    },
    {
      "epoch": 0.035496721645181695,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001221044104410441,
      "loss": 3.1523,
      "step": 389540
    },
    {
      "epoch": 0.03551536488133988,
      "grad_norm": 0.75,
      "learning_rate": 0.0001221024102410241,
      "loss": 3.0877,
      "step": 389550
    },
    {
      "epoch": 0.035534008117498066,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001221004100410041,
      "loss": 3.2046,
      "step": 389560
    },
    {
      "epoch": 0.03555265135365625,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001220984098409841,
      "loss": 3.2677,
      "step": 389570
    },
    {
      "epoch": 0.03557129458981443,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012209640964096409,
      "loss": 3.1472,
      "step": 389580
    },
    {
      "epoch": 0.03558993782597261,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001220944094409441,
      "loss": 3.2362,
      "step": 389590
    },
    {
      "epoch": 0.0356085810621308,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012209240924092408,
      "loss": 3.2618,
      "step": 389600
    },
    {
      "epoch": 0.03562722429828898,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012209040904090409,
      "loss": 3.2126,
      "step": 389610
    },
    {
      "epoch": 0.035645867534447165,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001220884088408841,
      "loss": 3.1547,
      "step": 389620
    },
    {
      "epoch": 0.03566451077060535,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001220864086408641,
      "loss": 3.1581,
      "step": 389630
    },
    {
      "epoch": 0.035683154006763536,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001220844084408441,
      "loss": 3.3335,
      "step": 389640
    },
    {
      "epoch": 0.03570179724292172,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001220824082408241,
      "loss": 3.2305,
      "step": 389650
    },
    {
      "epoch": 0.0357204404790799,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001220804080408041,
      "loss": 3.1064,
      "step": 389660
    },
    {
      "epoch": 0.03573908371523808,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012207840784078409,
      "loss": 3.1922,
      "step": 389670
    },
    {
      "epoch": 0.03575772695139627,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001220764076407641,
      "loss": 3.1389,
      "step": 389680
    },
    {
      "epoch": 0.03577637018755445,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012207440744074408,
      "loss": 3.0688,
      "step": 389690
    },
    {
      "epoch": 0.035795013423712635,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001220724072407241,
      "loss": 3.21,
      "step": 389700
    },
    {
      "epoch": 0.03581365665987082,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012207040704070407,
      "loss": 3.1276,
      "step": 389710
    },
    {
      "epoch": 0.035832299896029,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012206840684068408,
      "loss": 3.194,
      "step": 389720
    },
    {
      "epoch": 0.03585094313218719,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012206640664066406,
      "loss": 3.1033,
      "step": 389730
    },
    {
      "epoch": 0.03586958636834537,
      "grad_norm": 0.859375,
      "learning_rate": 0.00012206440644064407,
      "loss": 3.2181,
      "step": 389740
    },
    {
      "epoch": 0.03588822960450355,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012206240624062406,
      "loss": 3.1975,
      "step": 389750
    },
    {
      "epoch": 0.035906872840661734,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012206040604060407,
      "loss": 3.1275,
      "step": 389760
    },
    {
      "epoch": 0.03592551607681992,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012205840584058406,
      "loss": 3.1289,
      "step": 389770
    },
    {
      "epoch": 0.035944159312978105,
      "grad_norm": 0.75,
      "learning_rate": 0.00012205640564056406,
      "loss": 3.311,
      "step": 389780
    },
    {
      "epoch": 0.03596280254913629,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012205440544054405,
      "loss": 3.2777,
      "step": 389790
    },
    {
      "epoch": 0.03598144578529447,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012205240524052406,
      "loss": 3.281,
      "step": 389800
    },
    {
      "epoch": 0.03600008902145266,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012205040504050405,
      "loss": 3.2622,
      "step": 389810
    },
    {
      "epoch": 0.03601873225761084,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012204840484048406,
      "loss": 3.1812,
      "step": 389820
    },
    {
      "epoch": 0.03603737549376902,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012204640464046406,
      "loss": 3.0663,
      "step": 389830
    },
    {
      "epoch": 0.036056018729927204,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012204440444044405,
      "loss": 3.2183,
      "step": 389840
    },
    {
      "epoch": 0.03607466196608539,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012204240424042406,
      "loss": 3.1924,
      "step": 389850
    },
    {
      "epoch": 0.036093305202243575,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012204040404040404,
      "loss": 3.2471,
      "step": 389860
    },
    {
      "epoch": 0.03611194843840176,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012203840384038405,
      "loss": 3.2583,
      "step": 389870
    },
    {
      "epoch": 0.03613059167455994,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012203640364036403,
      "loss": 3.2049,
      "step": 389880
    },
    {
      "epoch": 0.03614923491071812,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012203440344034404,
      "loss": 3.1644,
      "step": 389890
    },
    {
      "epoch": 0.03616787814687631,
      "grad_norm": 0.875,
      "learning_rate": 0.00012203240324032404,
      "loss": 3.2583,
      "step": 389900
    },
    {
      "epoch": 0.03618652138303449,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012203040304030405,
      "loss": 3.2352,
      "step": 389910
    },
    {
      "epoch": 0.036205164619192674,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012202840284028403,
      "loss": 3.2933,
      "step": 389920
    },
    {
      "epoch": 0.036223807855350856,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012202640264026404,
      "loss": 3.1957,
      "step": 389930
    },
    {
      "epoch": 0.036242451091509045,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012202440244024402,
      "loss": 3.1932,
      "step": 389940
    },
    {
      "epoch": 0.03626109432766723,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012202240224022403,
      "loss": 3.2873,
      "step": 389950
    },
    {
      "epoch": 0.03627973756382541,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012202040204020401,
      "loss": 3.1671,
      "step": 389960
    },
    {
      "epoch": 0.03629838079998359,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012201840184018402,
      "loss": 3.1762,
      "step": 389970
    },
    {
      "epoch": 0.03631702403614178,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00012201640164016402,
      "loss": 3.2586,
      "step": 389980
    },
    {
      "epoch": 0.03633566727229996,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012201440144014403,
      "loss": 3.0835,
      "step": 389990
    },
    {
      "epoch": 0.036354310508458144,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012201240124012401,
      "loss": 3.0921,
      "step": 390000
    },
    {
      "epoch": 0.036372953744616326,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012201040104010402,
      "loss": 3.095,
      "step": 390010
    },
    {
      "epoch": 0.036391596980774515,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012200840084008403,
      "loss": 3.0934,
      "step": 390020
    },
    {
      "epoch": 0.0364102402169327,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012200640064006401,
      "loss": 3.0923,
      "step": 390030
    },
    {
      "epoch": 0.03642888345309088,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012200440044004402,
      "loss": 3.1746,
      "step": 390040
    },
    {
      "epoch": 0.03644752668924906,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000122002400240024,
      "loss": 3.1293,
      "step": 390050
    },
    {
      "epoch": 0.036466169925407244,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012200040004000401,
      "loss": 3.192,
      "step": 390060
    },
    {
      "epoch": 0.03648481316156543,
      "grad_norm": 0.8125,
      "learning_rate": 0.000121998399839984,
      "loss": 3.1402,
      "step": 390070
    },
    {
      "epoch": 0.036503456397723615,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012199639963996401,
      "loss": 3.2685,
      "step": 390080
    },
    {
      "epoch": 0.0365220996338818,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000121994399439944,
      "loss": 3.279,
      "step": 390090
    },
    {
      "epoch": 0.03654074287003998,
      "grad_norm": 0.65625,
      "learning_rate": 0.000121992399239924,
      "loss": 3.2668,
      "step": 390100
    },
    {
      "epoch": 0.03655938610619817,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012199039903990399,
      "loss": 3.2176,
      "step": 390110
    },
    {
      "epoch": 0.03657802934235635,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000121988398839884,
      "loss": 3.0925,
      "step": 390120
    },
    {
      "epoch": 0.03659667257851453,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012198639863986399,
      "loss": 3.2304,
      "step": 390130
    },
    {
      "epoch": 0.036615315814672714,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000121984398439844,
      "loss": 3.2233,
      "step": 390140
    },
    {
      "epoch": 0.0366339590508309,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012198239823982398,
      "loss": 3.3196,
      "step": 390150
    },
    {
      "epoch": 0.036652602286989085,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012198039803980399,
      "loss": 3.0813,
      "step": 390160
    },
    {
      "epoch": 0.03667124552314727,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012197839783978397,
      "loss": 3.0601,
      "step": 390170
    },
    {
      "epoch": 0.03668988875930545,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012197639763976398,
      "loss": 3.1688,
      "step": 390180
    },
    {
      "epoch": 0.03670853199546364,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012197439743974397,
      "loss": 3.0924,
      "step": 390190
    },
    {
      "epoch": 0.03672717523162182,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012197239723972397,
      "loss": 3.1782,
      "step": 390200
    },
    {
      "epoch": 0.03674581846778,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012197039703970397,
      "loss": 3.1446,
      "step": 390210
    },
    {
      "epoch": 0.036764461703938184,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012196839683968398,
      "loss": 3.1799,
      "step": 390220
    },
    {
      "epoch": 0.036783104940096366,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012196639663966399,
      "loss": 3.2082,
      "step": 390230
    },
    {
      "epoch": 0.036801748176254555,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012196439643964397,
      "loss": 3.2086,
      "step": 390240
    },
    {
      "epoch": 0.03682039141241274,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012196239623962398,
      "loss": 3.2319,
      "step": 390250
    },
    {
      "epoch": 0.03683903464857092,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012196039603960396,
      "loss": 3.065,
      "step": 390260
    },
    {
      "epoch": 0.0368576778847291,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012195839583958397,
      "loss": 3.1994,
      "step": 390270
    },
    {
      "epoch": 0.03687632112088729,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012195639563956395,
      "loss": 3.3322,
      "step": 390280
    },
    {
      "epoch": 0.03689496435704547,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012195439543954396,
      "loss": 3.0784,
      "step": 390290
    },
    {
      "epoch": 0.036913607593203654,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012195239523952396,
      "loss": 3.1362,
      "step": 390300
    },
    {
      "epoch": 0.036932250829361836,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012195039503950397,
      "loss": 3.0984,
      "step": 390310
    },
    {
      "epoch": 0.036950894065520025,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012194839483948395,
      "loss": 3.3114,
      "step": 390320
    },
    {
      "epoch": 0.03696953730167821,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012194639463946396,
      "loss": 3.2488,
      "step": 390330
    },
    {
      "epoch": 0.03698818053783639,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012194439443944394,
      "loss": 3.1074,
      "step": 390340
    },
    {
      "epoch": 0.03700682377399457,
      "grad_norm": 0.75,
      "learning_rate": 0.00012194239423942395,
      "loss": 3.1484,
      "step": 390350
    },
    {
      "epoch": 0.03702546701015276,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012194039403940393,
      "loss": 3.1522,
      "step": 390360
    },
    {
      "epoch": 0.03704411024631094,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012193839383938394,
      "loss": 3.1597,
      "step": 390370
    },
    {
      "epoch": 0.037062753482469124,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012193639363936394,
      "loss": 3.2275,
      "step": 390380
    },
    {
      "epoch": 0.037081396718627306,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012193439343934395,
      "loss": 3.1796,
      "step": 390390
    },
    {
      "epoch": 0.03710003995478549,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012193239323932393,
      "loss": 3.2101,
      "step": 390400
    },
    {
      "epoch": 0.03711868319094368,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012193039303930394,
      "loss": 3.2049,
      "step": 390410
    },
    {
      "epoch": 0.03713732642710186,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012192839283928395,
      "loss": 3.1147,
      "step": 390420
    },
    {
      "epoch": 0.03715596966326004,
      "grad_norm": 0.75,
      "learning_rate": 0.00012192639263926393,
      "loss": 3.2432,
      "step": 390430
    },
    {
      "epoch": 0.03717461289941822,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012192439243924394,
      "loss": 3.2065,
      "step": 390440
    },
    {
      "epoch": 0.03719325613557641,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012192239223922393,
      "loss": 3.0881,
      "step": 390450
    },
    {
      "epoch": 0.037211899371734594,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012192039203920393,
      "loss": 3.1545,
      "step": 390460
    },
    {
      "epoch": 0.037230542607892776,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012191839183918392,
      "loss": 3.1889,
      "step": 390470
    },
    {
      "epoch": 0.03724918584405096,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012191639163916393,
      "loss": 3.2347,
      "step": 390480
    },
    {
      "epoch": 0.03726782908020915,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012191439143914391,
      "loss": 3.2002,
      "step": 390490
    },
    {
      "epoch": 0.03728647231636733,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012191239123912392,
      "loss": 3.1571,
      "step": 390500
    },
    {
      "epoch": 0.03730511555252551,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001219103910391039,
      "loss": 3.1786,
      "step": 390510
    },
    {
      "epoch": 0.03732375878868369,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012190839083908391,
      "loss": 3.2883,
      "step": 390520
    },
    {
      "epoch": 0.03734240202484188,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012190639063906391,
      "loss": 3.093,
      "step": 390530
    },
    {
      "epoch": 0.037361045261000064,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012190439043904392,
      "loss": 3.161,
      "step": 390540
    },
    {
      "epoch": 0.037379688497158246,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001219023902390239,
      "loss": 3.1899,
      "step": 390550
    },
    {
      "epoch": 0.03739833173331643,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012190039003900391,
      "loss": 3.1664,
      "step": 390560
    },
    {
      "epoch": 0.03741697496947461,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012189838983898389,
      "loss": 3.1802,
      "step": 390570
    },
    {
      "epoch": 0.0374356182056328,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001218963896389639,
      "loss": 3.131,
      "step": 390580
    },
    {
      "epoch": 0.03745426144179098,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012189438943894388,
      "loss": 3.1133,
      "step": 390590
    },
    {
      "epoch": 0.03747290467794916,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001218923892389239,
      "loss": 3.3803,
      "step": 390600
    },
    {
      "epoch": 0.037491547914107345,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001218903890389039,
      "loss": 3.0489,
      "step": 390610
    },
    {
      "epoch": 0.037510191150265534,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001218883888388839,
      "loss": 3.1007,
      "step": 390620
    },
    {
      "epoch": 0.037528834386423716,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012188638863886391,
      "loss": 3.1985,
      "step": 390630
    },
    {
      "epoch": 0.0375474776225819,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012188438843884389,
      "loss": 3.1687,
      "step": 390640
    },
    {
      "epoch": 0.03756612085874008,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001218823882388239,
      "loss": 3.2037,
      "step": 390650
    },
    {
      "epoch": 0.03758476409489827,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012188038803880388,
      "loss": 3.2479,
      "step": 390660
    },
    {
      "epoch": 0.03760340733105645,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012187838783878389,
      "loss": 3.1904,
      "step": 390670
    },
    {
      "epoch": 0.037622050567214634,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012187638763876387,
      "loss": 3.2503,
      "step": 390680
    },
    {
      "epoch": 0.037640693803372816,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012187438743874388,
      "loss": 3.2145,
      "step": 390690
    },
    {
      "epoch": 0.037659337039531005,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012187238723872388,
      "loss": 3.138,
      "step": 390700
    },
    {
      "epoch": 0.03767798027568919,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012187038703870389,
      "loss": 3.2335,
      "step": 390710
    },
    {
      "epoch": 0.03769662351184737,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012186838683868387,
      "loss": 3.3702,
      "step": 390720
    },
    {
      "epoch": 0.03771526674800555,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012186638663866388,
      "loss": 3.1422,
      "step": 390730
    },
    {
      "epoch": 0.03773390998416373,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012186438643864386,
      "loss": 3.1512,
      "step": 390740
    },
    {
      "epoch": 0.03775255322032192,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012186238623862387,
      "loss": 3.1761,
      "step": 390750
    },
    {
      "epoch": 0.037771196456480104,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012186038603860386,
      "loss": 3.3205,
      "step": 390760
    },
    {
      "epoch": 0.037789839692638286,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012185838583858387,
      "loss": 3.1411,
      "step": 390770
    },
    {
      "epoch": 0.03780848292879647,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012185638563856385,
      "loss": 3.2489,
      "step": 390780
    },
    {
      "epoch": 0.03782712616495466,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012185438543854386,
      "loss": 3.1105,
      "step": 390790
    },
    {
      "epoch": 0.03784576940111284,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012185238523852385,
      "loss": 3.1611,
      "step": 390800
    },
    {
      "epoch": 0.03786441263727102,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012185038503850385,
      "loss": 3.1414,
      "step": 390810
    },
    {
      "epoch": 0.0378830558734292,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012184838483848386,
      "loss": 3.2219,
      "step": 390820
    },
    {
      "epoch": 0.03790169910958739,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012184638463846385,
      "loss": 3.1751,
      "step": 390830
    },
    {
      "epoch": 0.037920342345745574,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012184438443844386,
      "loss": 3.2328,
      "step": 390840
    },
    {
      "epoch": 0.037938985581903756,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012184238423842385,
      "loss": 3.2218,
      "step": 390850
    },
    {
      "epoch": 0.03795762881806194,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012184038403840386,
      "loss": 3.0624,
      "step": 390860
    },
    {
      "epoch": 0.03797627205422013,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012183838383838384,
      "loss": 3.1841,
      "step": 390870
    },
    {
      "epoch": 0.03799491529037831,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012183638363836385,
      "loss": 3.0848,
      "step": 390880
    },
    {
      "epoch": 0.03801355852653649,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012183438343834383,
      "loss": 3.2593,
      "step": 390890
    },
    {
      "epoch": 0.03803220176269467,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012183238323832384,
      "loss": 3.2545,
      "step": 390900
    },
    {
      "epoch": 0.038050844998852855,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012183038303830382,
      "loss": 3.1872,
      "step": 390910
    },
    {
      "epoch": 0.038069488235011044,
      "grad_norm": 0.890625,
      "learning_rate": 0.00012182838283828383,
      "loss": 3.1901,
      "step": 390920
    },
    {
      "epoch": 0.038088131471169226,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012182638263826383,
      "loss": 3.0683,
      "step": 390930
    },
    {
      "epoch": 0.03810677470732741,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012182438243824384,
      "loss": 3.0938,
      "step": 390940
    },
    {
      "epoch": 0.03812541794348559,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012182238223822382,
      "loss": 3.2249,
      "step": 390950
    },
    {
      "epoch": 0.03814406117964378,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012182038203820383,
      "loss": 3.1981,
      "step": 390960
    },
    {
      "epoch": 0.03816270441580196,
      "grad_norm": 0.890625,
      "learning_rate": 0.00012181838183818381,
      "loss": 3.1607,
      "step": 390970
    },
    {
      "epoch": 0.03818134765196014,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012181638163816382,
      "loss": 3.2318,
      "step": 390980
    },
    {
      "epoch": 0.038199990888118325,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012181438143814382,
      "loss": 3.1724,
      "step": 390990
    },
    {
      "epoch": 0.038218634124276514,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012181238123812381,
      "loss": 3.1144,
      "step": 391000
    },
    {
      "epoch": 0.038237277360434696,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012181038103810382,
      "loss": 3.0988,
      "step": 391010
    },
    {
      "epoch": 0.03825592059659288,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012180838083808382,
      "loss": 3.1385,
      "step": 391020
    },
    {
      "epoch": 0.03827456383275106,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012180638063806383,
      "loss": 3.2637,
      "step": 391030
    },
    {
      "epoch": 0.03829320706890925,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012180438043804381,
      "loss": 3.1699,
      "step": 391040
    },
    {
      "epoch": 0.03831185030506743,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012180238023802382,
      "loss": 3.1783,
      "step": 391050
    },
    {
      "epoch": 0.03833049354122561,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001218003800380038,
      "loss": 3.1746,
      "step": 391060
    },
    {
      "epoch": 0.038349136777383795,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012179837983798381,
      "loss": 3.2329,
      "step": 391070
    },
    {
      "epoch": 0.03836778001354198,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001217963796379638,
      "loss": 3.0861,
      "step": 391080
    },
    {
      "epoch": 0.038386423249700166,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012179437943794381,
      "loss": 3.1835,
      "step": 391090
    },
    {
      "epoch": 0.03840506648585835,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001217923792379238,
      "loss": 3.3617,
      "step": 391100
    },
    {
      "epoch": 0.03842370972201653,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001217903790379038,
      "loss": 3.3672,
      "step": 391110
    },
    {
      "epoch": 0.03844235295817471,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012178837883788379,
      "loss": 3.1219,
      "step": 391120
    },
    {
      "epoch": 0.0384609961943329,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001217863786378638,
      "loss": 3.226,
      "step": 391130
    },
    {
      "epoch": 0.03847963943049108,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012178437843784378,
      "loss": 3.1006,
      "step": 391140
    },
    {
      "epoch": 0.038498282666649265,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00012178237823782379,
      "loss": 3.2129,
      "step": 391150
    },
    {
      "epoch": 0.03851692590280745,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012178037803780378,
      "loss": 3.2454,
      "step": 391160
    },
    {
      "epoch": 0.038535569138965636,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012177837783778379,
      "loss": 3.1592,
      "step": 391170
    },
    {
      "epoch": 0.03855421237512382,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012177637763776377,
      "loss": 3.2595,
      "step": 391180
    },
    {
      "epoch": 0.038572855611282,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012177437743774378,
      "loss": 3.2035,
      "step": 391190
    },
    {
      "epoch": 0.03859149884744018,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012177237723772379,
      "loss": 3.1783,
      "step": 391200
    },
    {
      "epoch": 0.03861014208359837,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012177037703770377,
      "loss": 3.2533,
      "step": 391210
    },
    {
      "epoch": 0.03862878531975655,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012176837683768378,
      "loss": 3.1661,
      "step": 391220
    },
    {
      "epoch": 0.038647428555914735,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012176637663766376,
      "loss": 3.1831,
      "step": 391230
    },
    {
      "epoch": 0.03866607179207292,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012176437643764377,
      "loss": 3.1006,
      "step": 391240
    },
    {
      "epoch": 0.038684715028231106,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012176237623762377,
      "loss": 3.3072,
      "step": 391250
    },
    {
      "epoch": 0.03870335826438929,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012176037603760378,
      "loss": 3.1126,
      "step": 391260
    },
    {
      "epoch": 0.03872200150054747,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012175837583758376,
      "loss": 3.1765,
      "step": 391270
    },
    {
      "epoch": 0.03874064473670565,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012175637563756377,
      "loss": 3.165,
      "step": 391280
    },
    {
      "epoch": 0.038759287972863835,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012175437543754375,
      "loss": 3.2325,
      "step": 391290
    },
    {
      "epoch": 0.038777931209022024,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012175237523752376,
      "loss": 3.1963,
      "step": 391300
    },
    {
      "epoch": 0.038796574445180206,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012175037503750376,
      "loss": 3.1203,
      "step": 391310
    },
    {
      "epoch": 0.03881521768133839,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012174837483748375,
      "loss": 3.2143,
      "step": 391320
    },
    {
      "epoch": 0.03883386091749657,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012174637463746375,
      "loss": 3.1586,
      "step": 391330
    },
    {
      "epoch": 0.03885250415365476,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012174437443744376,
      "loss": 3.1991,
      "step": 391340
    },
    {
      "epoch": 0.03887114738981294,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012174237423742374,
      "loss": 3.2073,
      "step": 391350
    },
    {
      "epoch": 0.03888979062597112,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00012174037403740375,
      "loss": 3.0594,
      "step": 391360
    },
    {
      "epoch": 0.038908433862129305,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012173837383738373,
      "loss": 3.2156,
      "step": 391370
    },
    {
      "epoch": 0.038927077098287494,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012173637363736374,
      "loss": 3.1629,
      "step": 391380
    },
    {
      "epoch": 0.038945720334445676,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012173437343734373,
      "loss": 3.1455,
      "step": 391390
    },
    {
      "epoch": 0.03896436357060386,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012173237323732374,
      "loss": 3.1546,
      "step": 391400
    },
    {
      "epoch": 0.03898300680676204,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012173037303730375,
      "loss": 3.1382,
      "step": 391410
    },
    {
      "epoch": 0.03900165004292023,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012172837283728374,
      "loss": 3.1923,
      "step": 391420
    },
    {
      "epoch": 0.03902029327907841,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012172637263726374,
      "loss": 3.0805,
      "step": 391430
    },
    {
      "epoch": 0.03903893651523659,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012172437243724373,
      "loss": 3.2761,
      "step": 391440
    },
    {
      "epoch": 0.039057579751394775,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012172237223722374,
      "loss": 3.2652,
      "step": 391450
    },
    {
      "epoch": 0.03907622298755296,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012172037203720372,
      "loss": 3.0564,
      "step": 391460
    },
    {
      "epoch": 0.039094866223711146,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012171837183718373,
      "loss": 3.164,
      "step": 391470
    },
    {
      "epoch": 0.03911350945986933,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012171637163716372,
      "loss": 3.2339,
      "step": 391480
    },
    {
      "epoch": 0.03913215269602751,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012171437143714373,
      "loss": 3.2786,
      "step": 391490
    },
    {
      "epoch": 0.03915079593218569,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012171237123712371,
      "loss": 3.1727,
      "step": 391500
    },
    {
      "epoch": 0.03916943916834388,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012171037103710372,
      "loss": 3.252,
      "step": 391510
    },
    {
      "epoch": 0.03918808240450206,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001217083708370837,
      "loss": 3.189,
      "step": 391520
    },
    {
      "epoch": 0.039206725640660245,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012170637063706371,
      "loss": 3.2808,
      "step": 391530
    },
    {
      "epoch": 0.03922536887681843,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001217043704370437,
      "loss": 3.078,
      "step": 391540
    },
    {
      "epoch": 0.039244012112976616,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001217023702370237,
      "loss": 3.2954,
      "step": 391550
    },
    {
      "epoch": 0.0392626553491348,
      "grad_norm": 0.75,
      "learning_rate": 0.0001217003700370037,
      "loss": 3.1504,
      "step": 391560
    },
    {
      "epoch": 0.03928129858529298,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012169836983698371,
      "loss": 3.1327,
      "step": 391570
    },
    {
      "epoch": 0.03929994182145116,
      "grad_norm": 0.75,
      "learning_rate": 0.00012169636963696369,
      "loss": 3.2206,
      "step": 391580
    },
    {
      "epoch": 0.03931858505760935,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001216943694369437,
      "loss": 3.0399,
      "step": 391590
    },
    {
      "epoch": 0.03933722829376753,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012169236923692371,
      "loss": 3.2572,
      "step": 391600
    },
    {
      "epoch": 0.039355871529925715,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012169036903690369,
      "loss": 3.291,
      "step": 391610
    },
    {
      "epoch": 0.0393745147660839,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001216883688368837,
      "loss": 3.2937,
      "step": 391620
    },
    {
      "epoch": 0.03939315800224208,
      "grad_norm": 0.75,
      "learning_rate": 0.0001216863686368637,
      "loss": 3.1977,
      "step": 391630
    },
    {
      "epoch": 0.03941180123840027,
      "grad_norm": 0.625,
      "learning_rate": 0.00012168436843684369,
      "loss": 3.1375,
      "step": 391640
    },
    {
      "epoch": 0.03943044447455845,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012168236823682369,
      "loss": 3.1362,
      "step": 391650
    },
    {
      "epoch": 0.03944908771071663,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001216803680368037,
      "loss": 3.0808,
      "step": 391660
    },
    {
      "epoch": 0.039467730946874814,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012167836783678368,
      "loss": 3.2044,
      "step": 391670
    },
    {
      "epoch": 0.039486374183033,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012167636763676369,
      "loss": 3.1197,
      "step": 391680
    },
    {
      "epoch": 0.039505017419191185,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012167436743674367,
      "loss": 3.2759,
      "step": 391690
    },
    {
      "epoch": 0.03952366065534937,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012167236723672368,
      "loss": 3.0748,
      "step": 391700
    },
    {
      "epoch": 0.03954230389150755,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012167036703670368,
      "loss": 3.1286,
      "step": 391710
    },
    {
      "epoch": 0.03956094712766574,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012166836683668368,
      "loss": 3.1988,
      "step": 391720
    },
    {
      "epoch": 0.03957959036382392,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012166636663666367,
      "loss": 3.1618,
      "step": 391730
    },
    {
      "epoch": 0.0395982335999821,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012166436643664368,
      "loss": 3.2045,
      "step": 391740
    },
    {
      "epoch": 0.039616876836140284,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012166236623662366,
      "loss": 3.2511,
      "step": 391750
    },
    {
      "epoch": 0.03963552007229847,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012166036603660367,
      "loss": 3.1994,
      "step": 391760
    },
    {
      "epoch": 0.039654163308456655,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012165836583658365,
      "loss": 3.2049,
      "step": 391770
    },
    {
      "epoch": 0.03967280654461484,
      "grad_norm": 0.75,
      "learning_rate": 0.00012165636563656366,
      "loss": 3.1711,
      "step": 391780
    },
    {
      "epoch": 0.03969144978077302,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012165436543654365,
      "loss": 3.2431,
      "step": 391790
    },
    {
      "epoch": 0.0397100930169312,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012165236523652366,
      "loss": 3.1083,
      "step": 391800
    },
    {
      "epoch": 0.03972873625308939,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012165036503650367,
      "loss": 3.2891,
      "step": 391810
    },
    {
      "epoch": 0.03974737948924757,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012164836483648365,
      "loss": 3.1628,
      "step": 391820
    },
    {
      "epoch": 0.039766022725405754,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012164636463646366,
      "loss": 3.243,
      "step": 391830
    },
    {
      "epoch": 0.039784665961563936,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012164436443644365,
      "loss": 3.0672,
      "step": 391840
    },
    {
      "epoch": 0.039803309197722125,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012164236423642365,
      "loss": 3.1417,
      "step": 391850
    },
    {
      "epoch": 0.03982195243388031,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012164036403640364,
      "loss": 3.2255,
      "step": 391860
    },
    {
      "epoch": 0.03984059567003849,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012163836383638365,
      "loss": 3.2397,
      "step": 391870
    },
    {
      "epoch": 0.03985923890619667,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00012163636363636364,
      "loss": 3.1819,
      "step": 391880
    },
    {
      "epoch": 0.03987788214235486,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012163436343634365,
      "loss": 3.0589,
      "step": 391890
    },
    {
      "epoch": 0.03989652537851304,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012163236323632363,
      "loss": 3.1352,
      "step": 391900
    },
    {
      "epoch": 0.039915168614671225,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012163036303630364,
      "loss": 3.1866,
      "step": 391910
    },
    {
      "epoch": 0.03993381185082941,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012162836283628362,
      "loss": 3.0867,
      "step": 391920
    },
    {
      "epoch": 0.039952455086987596,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012162636263626363,
      "loss": 3.2927,
      "step": 391930
    },
    {
      "epoch": 0.03997109832314578,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00012162436243624363,
      "loss": 3.1285,
      "step": 391940
    },
    {
      "epoch": 0.03998974155930396,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012162236223622362,
      "loss": 3.2019,
      "step": 391950
    },
    {
      "epoch": 0.04000838479546214,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012162036203620362,
      "loss": 3.1621,
      "step": 391960
    },
    {
      "epoch": 0.040027028031620324,
      "grad_norm": 0.75,
      "learning_rate": 0.00012161836183618363,
      "loss": 3.2875,
      "step": 391970
    },
    {
      "epoch": 0.04004567126777851,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012161636163616361,
      "loss": 3.1245,
      "step": 391980
    },
    {
      "epoch": 0.040064314503936695,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012161436143614362,
      "loss": 3.2023,
      "step": 391990
    },
    {
      "epoch": 0.04008295774009488,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012161236123612363,
      "loss": 3.1529,
      "step": 392000
    },
    {
      "epoch": 0.04010160097625306,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012161036103610361,
      "loss": 3.1852,
      "step": 392010
    },
    {
      "epoch": 0.04012024421241125,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012160836083608362,
      "loss": 3.2755,
      "step": 392020
    },
    {
      "epoch": 0.04013888744856943,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012160636063606362,
      "loss": 3.0639,
      "step": 392030
    },
    {
      "epoch": 0.04015753068472761,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012160436043604362,
      "loss": 3.1171,
      "step": 392040
    },
    {
      "epoch": 0.040176173920885794,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001216023602360236,
      "loss": 3.1746,
      "step": 392050
    },
    {
      "epoch": 0.04019481715704398,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012160036003600362,
      "loss": 3.2827,
      "step": 392060
    },
    {
      "epoch": 0.040213460393202165,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001215983598359836,
      "loss": 3.2018,
      "step": 392070
    },
    {
      "epoch": 0.04023210362936035,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012159635963596361,
      "loss": 3.3091,
      "step": 392080
    },
    {
      "epoch": 0.04025074686551853,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012159435943594359,
      "loss": 3.2186,
      "step": 392090
    },
    {
      "epoch": 0.04026939010167672,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001215923592359236,
      "loss": 3.2621,
      "step": 392100
    },
    {
      "epoch": 0.0402880333378349,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001215903590359036,
      "loss": 3.1891,
      "step": 392110
    },
    {
      "epoch": 0.04030667657399308,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001215883588358836,
      "loss": 3.1758,
      "step": 392120
    },
    {
      "epoch": 0.040325319810151264,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012158635863586359,
      "loss": 3.2429,
      "step": 392130
    },
    {
      "epoch": 0.040343963046309446,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001215843584358436,
      "loss": 3.1816,
      "step": 392140
    },
    {
      "epoch": 0.040362606282467635,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012158235823582358,
      "loss": 3.1432,
      "step": 392150
    },
    {
      "epoch": 0.04038124951862582,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012158035803580359,
      "loss": 3.1062,
      "step": 392160
    },
    {
      "epoch": 0.040399892754784,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012157835783578357,
      "loss": 3.2654,
      "step": 392170
    },
    {
      "epoch": 0.04041853599094218,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012157635763576358,
      "loss": 3.2043,
      "step": 392180
    },
    {
      "epoch": 0.04043717922710037,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012157435743574359,
      "loss": 3.1464,
      "step": 392190
    },
    {
      "epoch": 0.04045582246325855,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012157235723572358,
      "loss": 3.308,
      "step": 392200
    },
    {
      "epoch": 0.040474465699416734,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012157035703570359,
      "loss": 3.1575,
      "step": 392210
    },
    {
      "epoch": 0.040493108935574916,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012156835683568357,
      "loss": 3.1192,
      "step": 392220
    },
    {
      "epoch": 0.040511752171733105,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012156635663566358,
      "loss": 3.0749,
      "step": 392230
    },
    {
      "epoch": 0.04053039540789129,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012156435643564356,
      "loss": 3.1409,
      "step": 392240
    },
    {
      "epoch": 0.04054903864404947,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012156235623562357,
      "loss": 3.0853,
      "step": 392250
    },
    {
      "epoch": 0.04056768188020765,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012156035603560357,
      "loss": 3.1373,
      "step": 392260
    },
    {
      "epoch": 0.04058632511636584,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012155835583558356,
      "loss": 3.1473,
      "step": 392270
    },
    {
      "epoch": 0.04060496835252402,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012155635563556356,
      "loss": 3.1279,
      "step": 392280
    },
    {
      "epoch": 0.040623611588682204,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012155435543554357,
      "loss": 3.1055,
      "step": 392290
    },
    {
      "epoch": 0.040642254824840386,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012155235523552355,
      "loss": 3.2704,
      "step": 392300
    },
    {
      "epoch": 0.04066089806099857,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012155035503550356,
      "loss": 3.1867,
      "step": 392310
    },
    {
      "epoch": 0.04067954129715676,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012154835483548354,
      "loss": 3.1628,
      "step": 392320
    },
    {
      "epoch": 0.04069818453331494,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012154635463546355,
      "loss": 3.1678,
      "step": 392330
    },
    {
      "epoch": 0.04071682776947312,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012154435443544355,
      "loss": 3.2014,
      "step": 392340
    },
    {
      "epoch": 0.0407354710056313,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00012154235423542356,
      "loss": 3.1,
      "step": 392350
    },
    {
      "epoch": 0.04075411424178949,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012154035403540354,
      "loss": 3.148,
      "step": 392360
    },
    {
      "epoch": 0.040772757477947674,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012153835383538355,
      "loss": 3.274,
      "step": 392370
    },
    {
      "epoch": 0.040791400714105856,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012153635363536353,
      "loss": 3.2133,
      "step": 392380
    },
    {
      "epoch": 0.04081004395026404,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012153435343534354,
      "loss": 3.2263,
      "step": 392390
    },
    {
      "epoch": 0.04082868718642223,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012153235323532355,
      "loss": 3.1883,
      "step": 392400
    },
    {
      "epoch": 0.04084733042258041,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012153035303530353,
      "loss": 3.1781,
      "step": 392410
    },
    {
      "epoch": 0.04086597365873859,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012152835283528354,
      "loss": 3.1533,
      "step": 392420
    },
    {
      "epoch": 0.04088461689489677,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012152635263526353,
      "loss": 3.2233,
      "step": 392430
    },
    {
      "epoch": 0.04090326013105496,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012152435243524354,
      "loss": 3.1759,
      "step": 392440
    },
    {
      "epoch": 0.040921903367213144,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012152235223522353,
      "loss": 3.2091,
      "step": 392450
    },
    {
      "epoch": 0.040940546603371326,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012152035203520353,
      "loss": 3.2196,
      "step": 392460
    },
    {
      "epoch": 0.04095918983952951,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012151835183518352,
      "loss": 3.1365,
      "step": 392470
    },
    {
      "epoch": 0.04097783307568769,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012151635163516353,
      "loss": 3.1746,
      "step": 392480
    },
    {
      "epoch": 0.04099647631184588,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012151435143514351,
      "loss": 3.1923,
      "step": 392490
    },
    {
      "epoch": 0.04101511954800406,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012151235123512352,
      "loss": 3.2054,
      "step": 392500
    },
    {
      "epoch": 0.041033762784162243,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012151035103510351,
      "loss": 3.205,
      "step": 392510
    },
    {
      "epoch": 0.041052406020320426,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012150835083508352,
      "loss": 3.1271,
      "step": 392520
    },
    {
      "epoch": 0.041071049256478614,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001215063506350635,
      "loss": 3.1759,
      "step": 392530
    },
    {
      "epoch": 0.041089692492636797,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012150435043504351,
      "loss": 3.2197,
      "step": 392540
    },
    {
      "epoch": 0.04110833572879498,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001215023502350235,
      "loss": 3.0953,
      "step": 392550
    },
    {
      "epoch": 0.04112697896495316,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001215003500350035,
      "loss": 3.2404,
      "step": 392560
    },
    {
      "epoch": 0.04114562220111135,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001214983498349835,
      "loss": 3.1238,
      "step": 392570
    },
    {
      "epoch": 0.04116426543726953,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012149634963496351,
      "loss": 3.1036,
      "step": 392580
    },
    {
      "epoch": 0.041182908673427714,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001214943494349435,
      "loss": 3.1118,
      "step": 392590
    },
    {
      "epoch": 0.041201551909585896,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001214923492349235,
      "loss": 3.3119,
      "step": 392600
    },
    {
      "epoch": 0.041220195145744085,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012149034903490351,
      "loss": 2.99,
      "step": 392610
    },
    {
      "epoch": 0.04123883838190227,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012148834883488349,
      "loss": 3.2578,
      "step": 392620
    },
    {
      "epoch": 0.04125748161806045,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001214863486348635,
      "loss": 3.1234,
      "step": 392630
    },
    {
      "epoch": 0.04127612485421863,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012148434843484348,
      "loss": 3.1813,
      "step": 392640
    },
    {
      "epoch": 0.04129476809037681,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012148234823482349,
      "loss": 3.2382,
      "step": 392650
    },
    {
      "epoch": 0.041313411326535,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012148034803480349,
      "loss": 3.1096,
      "step": 392660
    },
    {
      "epoch": 0.041332054562693184,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001214783478347835,
      "loss": 3.1571,
      "step": 392670
    },
    {
      "epoch": 0.041350697798851366,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012147634763476348,
      "loss": 3.313,
      "step": 392680
    },
    {
      "epoch": 0.04136934103500955,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012147434743474349,
      "loss": 3.166,
      "step": 392690
    },
    {
      "epoch": 0.04138798427116774,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012147234723472347,
      "loss": 3.0396,
      "step": 392700
    },
    {
      "epoch": 0.04140662750732592,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012147034703470348,
      "loss": 2.9919,
      "step": 392710
    },
    {
      "epoch": 0.0414252707434841,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012146834683468346,
      "loss": 3.1514,
      "step": 392720
    },
    {
      "epoch": 0.04144391397964228,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012146634663466347,
      "loss": 3.2276,
      "step": 392730
    },
    {
      "epoch": 0.04146255721580047,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012146434643464347,
      "loss": 3.1513,
      "step": 392740
    },
    {
      "epoch": 0.041481200451958654,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012146234623462347,
      "loss": 3.2117,
      "step": 392750
    },
    {
      "epoch": 0.041499843688116836,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012146034603460346,
      "loss": 3.0859,
      "step": 392760
    },
    {
      "epoch": 0.04151848692427502,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012145834583458347,
      "loss": 3.0394,
      "step": 392770
    },
    {
      "epoch": 0.04153713016043321,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012145634563456347,
      "loss": 3.1196,
      "step": 392780
    },
    {
      "epoch": 0.04155577339659139,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012145434543454346,
      "loss": 3.2875,
      "step": 392790
    },
    {
      "epoch": 0.04157441663274957,
      "grad_norm": 0.75,
      "learning_rate": 0.00012145234523452347,
      "loss": 3.1949,
      "step": 392800
    },
    {
      "epoch": 0.04159305986890775,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012145034503450345,
      "loss": 3.28,
      "step": 392810
    },
    {
      "epoch": 0.041611703105065935,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012144834483448346,
      "loss": 3.1865,
      "step": 392820
    },
    {
      "epoch": 0.041630346341224124,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012144634463446345,
      "loss": 3.2107,
      "step": 392830
    },
    {
      "epoch": 0.041648989577382306,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00012144434443444346,
      "loss": 3.087,
      "step": 392840
    },
    {
      "epoch": 0.04166763281354049,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012144234423442344,
      "loss": 3.1838,
      "step": 392850
    },
    {
      "epoch": 0.04168627604969867,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012144034403440345,
      "loss": 3.1739,
      "step": 392860
    },
    {
      "epoch": 0.04170491928585686,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012143834383438344,
      "loss": 3.357,
      "step": 392870
    },
    {
      "epoch": 0.04172356252201504,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012143634363436344,
      "loss": 3.1648,
      "step": 392880
    },
    {
      "epoch": 0.04174220575817322,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012143434343434344,
      "loss": 3.1185,
      "step": 392890
    },
    {
      "epoch": 0.041760848994331405,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012143234323432345,
      "loss": 3.236,
      "step": 392900
    },
    {
      "epoch": 0.041779492230489594,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012143034303430343,
      "loss": 3.2198,
      "step": 392910
    },
    {
      "epoch": 0.041798135466647776,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012142834283428344,
      "loss": 3.2688,
      "step": 392920
    },
    {
      "epoch": 0.04181677870280596,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012142634263426342,
      "loss": 3.1368,
      "step": 392930
    },
    {
      "epoch": 0.04183542193896414,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012142434243424343,
      "loss": 3.2282,
      "step": 392940
    },
    {
      "epoch": 0.04185406517512233,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012142234223422341,
      "loss": 3.1172,
      "step": 392950
    },
    {
      "epoch": 0.04187270841128051,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00012142034203420342,
      "loss": 3.11,
      "step": 392960
    },
    {
      "epoch": 0.04189135164743869,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012141834183418342,
      "loss": 3.0661,
      "step": 392970
    },
    {
      "epoch": 0.041909994883596875,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012141634163416343,
      "loss": 3.1661,
      "step": 392980
    },
    {
      "epoch": 0.04192863811975506,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012141434143414344,
      "loss": 3.2005,
      "step": 392990
    },
    {
      "epoch": 0.041947281355913246,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012141234123412342,
      "loss": 3.2085,
      "step": 393000
    },
    {
      "epoch": 0.04196592459207143,
      "grad_norm": 0.75,
      "learning_rate": 0.00012141034103410343,
      "loss": 2.9588,
      "step": 393010
    },
    {
      "epoch": 0.04198456782822961,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012140834083408341,
      "loss": 3.1319,
      "step": 393020
    },
    {
      "epoch": 0.04200321106438779,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012140634063406342,
      "loss": 3.2361,
      "step": 393030
    },
    {
      "epoch": 0.04202185430054598,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001214043404340434,
      "loss": 3.1857,
      "step": 393040
    },
    {
      "epoch": 0.04204049753670416,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012140234023402341,
      "loss": 3.1591,
      "step": 393050
    },
    {
      "epoch": 0.042059140772862345,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001214003400340034,
      "loss": 3.2434,
      "step": 393060
    },
    {
      "epoch": 0.04207778400902053,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012139833983398341,
      "loss": 3.08,
      "step": 393070
    },
    {
      "epoch": 0.042096427245178716,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001213963396339634,
      "loss": 3.2737,
      "step": 393080
    },
    {
      "epoch": 0.0421150704813369,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001213943394339434,
      "loss": 3.2681,
      "step": 393090
    },
    {
      "epoch": 0.04213371371749508,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012139233923392339,
      "loss": 3.2786,
      "step": 393100
    },
    {
      "epoch": 0.04215235695365326,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001213903390339034,
      "loss": 3.1788,
      "step": 393110
    },
    {
      "epoch": 0.04217100018981145,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012138833883388338,
      "loss": 3.2561,
      "step": 393120
    },
    {
      "epoch": 0.04218964342596963,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012138633863386339,
      "loss": 3.1068,
      "step": 393130
    },
    {
      "epoch": 0.042208286662127815,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012138433843384338,
      "loss": 3.1317,
      "step": 393140
    },
    {
      "epoch": 0.042226929898286,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012138233823382339,
      "loss": 3.1329,
      "step": 393150
    },
    {
      "epoch": 0.04224557313444418,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012138033803380338,
      "loss": 3.1826,
      "step": 393160
    },
    {
      "epoch": 0.04226421637060237,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012137833783378338,
      "loss": 3.1746,
      "step": 393170
    },
    {
      "epoch": 0.04228285960676055,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001213763376337634,
      "loss": 3.2387,
      "step": 393180
    },
    {
      "epoch": 0.04230150284291873,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012137433743374338,
      "loss": 3.1364,
      "step": 393190
    },
    {
      "epoch": 0.042320146079076915,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012137233723372338,
      "loss": 3.1953,
      "step": 393200
    },
    {
      "epoch": 0.042338789315235104,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012137033703370338,
      "loss": 3.1179,
      "step": 393210
    },
    {
      "epoch": 0.042357432551393286,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012136833683368338,
      "loss": 3.1969,
      "step": 393220
    },
    {
      "epoch": 0.04237607578755147,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012136633663366337,
      "loss": 3.1654,
      "step": 393230
    },
    {
      "epoch": 0.04239471902370965,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00012136433643364338,
      "loss": 3.151,
      "step": 393240
    },
    {
      "epoch": 0.04241336225986784,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012136233623362336,
      "loss": 3.2687,
      "step": 393250
    },
    {
      "epoch": 0.04243200549602602,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012136033603360337,
      "loss": 3.3326,
      "step": 393260
    },
    {
      "epoch": 0.0424506487321842,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012135833583358335,
      "loss": 3.2613,
      "step": 393270
    },
    {
      "epoch": 0.042469291968342385,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012135633563356336,
      "loss": 3.282,
      "step": 393280
    },
    {
      "epoch": 0.042487935204500574,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012135433543354336,
      "loss": 3.2955,
      "step": 393290
    },
    {
      "epoch": 0.042506578440658756,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012135233523352337,
      "loss": 3.2281,
      "step": 393300
    },
    {
      "epoch": 0.04252522167681694,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012135033503350335,
      "loss": 3.0183,
      "step": 393310
    },
    {
      "epoch": 0.04254386491297512,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012134833483348336,
      "loss": 3.2377,
      "step": 393320
    },
    {
      "epoch": 0.0425625081491333,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012134633463346334,
      "loss": 3.0115,
      "step": 393330
    },
    {
      "epoch": 0.04258115138529149,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012134433443344335,
      "loss": 3.1716,
      "step": 393340
    },
    {
      "epoch": 0.04259979462144967,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012134233423342333,
      "loss": 3.2479,
      "step": 393350
    },
    {
      "epoch": 0.042618437857607855,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012134033403340334,
      "loss": 3.2106,
      "step": 393360
    },
    {
      "epoch": 0.04263708109376604,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012133833383338335,
      "loss": 3.2407,
      "step": 393370
    },
    {
      "epoch": 0.042655724329924226,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012133633363336335,
      "loss": 3.2643,
      "step": 393380
    },
    {
      "epoch": 0.04267436756608241,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012133433343334336,
      "loss": 3.3065,
      "step": 393390
    },
    {
      "epoch": 0.04269301080224059,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012133233323332334,
      "loss": 3.2734,
      "step": 393400
    },
    {
      "epoch": 0.04271165403839877,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012133033303330335,
      "loss": 3.1401,
      "step": 393410
    },
    {
      "epoch": 0.04273029727455696,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012132833283328333,
      "loss": 3.1716,
      "step": 393420
    },
    {
      "epoch": 0.04274894051071514,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012132633263326334,
      "loss": 3.0922,
      "step": 393430
    },
    {
      "epoch": 0.042767583746873325,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012132433243324332,
      "loss": 3.2031,
      "step": 393440
    },
    {
      "epoch": 0.04278622698303151,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012132233223322333,
      "loss": 3.1239,
      "step": 393450
    },
    {
      "epoch": 0.042804870219189696,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012132033203320332,
      "loss": 3.0673,
      "step": 393460
    },
    {
      "epoch": 0.04282351345534788,
      "grad_norm": 0.75,
      "learning_rate": 0.00012131833183318333,
      "loss": 3.0825,
      "step": 393470
    },
    {
      "epoch": 0.04284215669150606,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012131633163316332,
      "loss": 3.266,
      "step": 393480
    },
    {
      "epoch": 0.04286079992766424,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012131433143314332,
      "loss": 3.1161,
      "step": 393490
    },
    {
      "epoch": 0.042879443163822424,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001213123312331233,
      "loss": 3.3061,
      "step": 393500
    },
    {
      "epoch": 0.04289808639998061,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012131033103310332,
      "loss": 3.1437,
      "step": 393510
    },
    {
      "epoch": 0.042916729636138795,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00012130833083308331,
      "loss": 3.1973,
      "step": 393520
    },
    {
      "epoch": 0.04293537287229698,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012130633063306332,
      "loss": 3.1553,
      "step": 393530
    },
    {
      "epoch": 0.04295401610845516,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001213043304330433,
      "loss": 3.1942,
      "step": 393540
    },
    {
      "epoch": 0.04297265934461335,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012130233023302331,
      "loss": 3.1658,
      "step": 393550
    },
    {
      "epoch": 0.04299130258077153,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001213003300330033,
      "loss": 3.1774,
      "step": 393560
    },
    {
      "epoch": 0.04300994581692971,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001212983298329833,
      "loss": 3.1361,
      "step": 393570
    },
    {
      "epoch": 0.043028589053087894,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012129632963296331,
      "loss": 3.2409,
      "step": 393580
    },
    {
      "epoch": 0.04304723228924608,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001212943294329433,
      "loss": 3.1513,
      "step": 393590
    },
    {
      "epoch": 0.043065875525404265,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001212923292329233,
      "loss": 3.3089,
      "step": 393600
    },
    {
      "epoch": 0.04308451876156245,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001212903290329033,
      "loss": 3.3128,
      "step": 393610
    },
    {
      "epoch": 0.04310316199772063,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012128832883288331,
      "loss": 3.0891,
      "step": 393620
    },
    {
      "epoch": 0.04312180523387882,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012128632863286329,
      "loss": 3.1188,
      "step": 393630
    },
    {
      "epoch": 0.043140448470037,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001212843284328433,
      "loss": 3.2223,
      "step": 393640
    },
    {
      "epoch": 0.04315909170619518,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012128232823282328,
      "loss": 3.2176,
      "step": 393650
    },
    {
      "epoch": 0.043177734942353364,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012128032803280329,
      "loss": 2.8689,
      "step": 393660
    },
    {
      "epoch": 0.043196378178511546,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012127832783278327,
      "loss": 3.2233,
      "step": 393670
    },
    {
      "epoch": 0.043215021414669735,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012127632763276328,
      "loss": 3.2202,
      "step": 393680
    },
    {
      "epoch": 0.04323366465082792,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012127432743274328,
      "loss": 3.2789,
      "step": 393690
    },
    {
      "epoch": 0.0432523078869861,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012127232723272329,
      "loss": 3.1765,
      "step": 393700
    },
    {
      "epoch": 0.04327095112314428,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012127032703270327,
      "loss": 3.1958,
      "step": 393710
    },
    {
      "epoch": 0.04328959435930247,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012126832683268328,
      "loss": 3.0746,
      "step": 393720
    },
    {
      "epoch": 0.04330823759546065,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012126632663266326,
      "loss": 3.334,
      "step": 393730
    },
    {
      "epoch": 0.043326880831618834,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012126432643264327,
      "loss": 3.1625,
      "step": 393740
    },
    {
      "epoch": 0.043345524067777016,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012126232623262326,
      "loss": 3.2024,
      "step": 393750
    },
    {
      "epoch": 0.043364167303935205,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012126032603260326,
      "loss": 3.1873,
      "step": 393760
    },
    {
      "epoch": 0.04338281054009339,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012125832583258327,
      "loss": 3.1997,
      "step": 393770
    },
    {
      "epoch": 0.04340145377625157,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012125632563256326,
      "loss": 3.2299,
      "step": 393780
    },
    {
      "epoch": 0.04342009701240975,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012125432543254327,
      "loss": 3.063,
      "step": 393790
    },
    {
      "epoch": 0.04343874024856794,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012125232523252326,
      "loss": 3.1256,
      "step": 393800
    },
    {
      "epoch": 0.04345738348472612,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012125032503250326,
      "loss": 3.1745,
      "step": 393810
    },
    {
      "epoch": 0.043476026720884305,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012124832483248325,
      "loss": 3.1592,
      "step": 393820
    },
    {
      "epoch": 0.04349466995704249,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012124632463246326,
      "loss": 3.217,
      "step": 393830
    },
    {
      "epoch": 0.04351331319320067,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012124432443244325,
      "loss": 3.1722,
      "step": 393840
    },
    {
      "epoch": 0.04353195642935886,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012124232423242326,
      "loss": 3.324,
      "step": 393850
    },
    {
      "epoch": 0.04355059966551704,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012124032403240324,
      "loss": 3.3056,
      "step": 393860
    },
    {
      "epoch": 0.04356924290167522,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012123832383238325,
      "loss": 3.2134,
      "step": 393870
    },
    {
      "epoch": 0.043587886137833404,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012123632363236323,
      "loss": 3.2751,
      "step": 393880
    },
    {
      "epoch": 0.04360652937399159,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012123432343234324,
      "loss": 3.1615,
      "step": 393890
    },
    {
      "epoch": 0.043625172610149775,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012123232323232323,
      "loss": 3.1786,
      "step": 393900
    },
    {
      "epoch": 0.04364381584630796,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012123032303230323,
      "loss": 3.1753,
      "step": 393910
    },
    {
      "epoch": 0.04366245908246614,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012122832283228323,
      "loss": 3.3094,
      "step": 393920
    },
    {
      "epoch": 0.04368110231862433,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012122632263226324,
      "loss": 3.1879,
      "step": 393930
    },
    {
      "epoch": 0.04369974555478251,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012122432243224322,
      "loss": 3.1259,
      "step": 393940
    },
    {
      "epoch": 0.04371838879094069,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012122232223222323,
      "loss": 3.1915,
      "step": 393950
    },
    {
      "epoch": 0.043737032027098874,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012122032203220321,
      "loss": 3.162,
      "step": 393960
    },
    {
      "epoch": 0.04375567526325706,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012121832183218322,
      "loss": 3.1711,
      "step": 393970
    },
    {
      "epoch": 0.043774318499415245,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012121632163216323,
      "loss": 3.2424,
      "step": 393980
    },
    {
      "epoch": 0.04379296173557343,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012121432143214321,
      "loss": 3.2264,
      "step": 393990
    },
    {
      "epoch": 0.04381160497173161,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012121232123212322,
      "loss": 3.3067,
      "step": 394000
    },
    {
      "epoch": 0.04383024820788979,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012121032103210322,
      "loss": 3.1763,
      "step": 394010
    },
    {
      "epoch": 0.04384889144404798,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012120832083208323,
      "loss": 3.2017,
      "step": 394020
    },
    {
      "epoch": 0.04386753468020616,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012120632063206321,
      "loss": 3.2491,
      "step": 394030
    },
    {
      "epoch": 0.043886177916364344,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012120432043204322,
      "loss": 3.2002,
      "step": 394040
    },
    {
      "epoch": 0.043904821152522526,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001212023202320232,
      "loss": 3.0952,
      "step": 394050
    },
    {
      "epoch": 0.043923464388680715,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012120032003200321,
      "loss": 3.112,
      "step": 394060
    },
    {
      "epoch": 0.0439421076248389,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001211983198319832,
      "loss": 3.2433,
      "step": 394070
    },
    {
      "epoch": 0.04396075086099708,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001211963196319632,
      "loss": 3.2401,
      "step": 394080
    },
    {
      "epoch": 0.04397939409715526,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001211943194319432,
      "loss": 3.1683,
      "step": 394090
    },
    {
      "epoch": 0.04399803733331345,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001211923192319232,
      "loss": 3.126,
      "step": 394100
    },
    {
      "epoch": 0.04401668056947163,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012119031903190319,
      "loss": 3.2754,
      "step": 394110
    },
    {
      "epoch": 0.044035323805629814,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001211883188318832,
      "loss": 3.2212,
      "step": 394120
    },
    {
      "epoch": 0.044053967041787996,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012118631863186318,
      "loss": 3.2555,
      "step": 394130
    },
    {
      "epoch": 0.044072610277946185,
      "grad_norm": 0.75,
      "learning_rate": 0.00012118431843184319,
      "loss": 3.1981,
      "step": 394140
    },
    {
      "epoch": 0.04409125351410437,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012118231823182318,
      "loss": 3.3308,
      "step": 394150
    },
    {
      "epoch": 0.04410989675026255,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012118031803180319,
      "loss": 3.1581,
      "step": 394160
    },
    {
      "epoch": 0.04412853998642073,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001211783178317832,
      "loss": 3.2871,
      "step": 394170
    },
    {
      "epoch": 0.04414718322257891,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012117631763176318,
      "loss": 3.1974,
      "step": 394180
    },
    {
      "epoch": 0.0441658264587371,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012117431743174319,
      "loss": 3.137,
      "step": 394190
    },
    {
      "epoch": 0.044184469694895284,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012117231723172317,
      "loss": 3.1568,
      "step": 394200
    },
    {
      "epoch": 0.044203112931053466,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012117031703170318,
      "loss": 3.1655,
      "step": 394210
    },
    {
      "epoch": 0.04422175616721165,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012116831683168317,
      "loss": 3.2173,
      "step": 394220
    },
    {
      "epoch": 0.04424039940336984,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012116631663166317,
      "loss": 3.2834,
      "step": 394230
    },
    {
      "epoch": 0.04425904263952802,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012116431643164317,
      "loss": 3.2034,
      "step": 394240
    },
    {
      "epoch": 0.0442776858756862,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012116231623162318,
      "loss": 3.0996,
      "step": 394250
    },
    {
      "epoch": 0.04429632911184438,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012116031603160316,
      "loss": 3.163,
      "step": 394260
    },
    {
      "epoch": 0.04431497234800257,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012115831583158317,
      "loss": 3.2507,
      "step": 394270
    },
    {
      "epoch": 0.044333615584160754,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012115631563156315,
      "loss": 3.2389,
      "step": 394280
    },
    {
      "epoch": 0.044352258820318936,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012115431543154316,
      "loss": 3.1352,
      "step": 394290
    },
    {
      "epoch": 0.04437090205647712,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012115231523152314,
      "loss": 3.3102,
      "step": 394300
    },
    {
      "epoch": 0.04438954529263531,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012115031503150315,
      "loss": 3.09,
      "step": 394310
    },
    {
      "epoch": 0.04440818852879349,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012114831483148315,
      "loss": 3.1216,
      "step": 394320
    },
    {
      "epoch": 0.04442683176495167,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012114631463146316,
      "loss": 3.145,
      "step": 394330
    },
    {
      "epoch": 0.04444547500110985,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012114431443144314,
      "loss": 3.1042,
      "step": 394340
    },
    {
      "epoch": 0.044464118237268035,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012114231423142315,
      "loss": 3.1695,
      "step": 394350
    },
    {
      "epoch": 0.044482761473426224,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012114031403140316,
      "loss": 3.1457,
      "step": 394360
    },
    {
      "epoch": 0.044501404709584406,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012113831383138314,
      "loss": 3.2845,
      "step": 394370
    },
    {
      "epoch": 0.04452004794574259,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012113631363136315,
      "loss": 3.1262,
      "step": 394380
    },
    {
      "epoch": 0.04453869118190077,
      "grad_norm": 0.75,
      "learning_rate": 0.00012113431343134314,
      "loss": 3.2019,
      "step": 394390
    },
    {
      "epoch": 0.04455733441805896,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012113231323132314,
      "loss": 3.2939,
      "step": 394400
    },
    {
      "epoch": 0.04457597765421714,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012113031303130314,
      "loss": 3.2251,
      "step": 394410
    },
    {
      "epoch": 0.044594620890375324,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012112831283128315,
      "loss": 3.0743,
      "step": 394420
    },
    {
      "epoch": 0.044613264126533506,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012112631263126313,
      "loss": 3.1853,
      "step": 394430
    },
    {
      "epoch": 0.044631907362691695,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012112431243124314,
      "loss": 3.23,
      "step": 394440
    },
    {
      "epoch": 0.04465055059884988,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012112231223122312,
      "loss": 3.1362,
      "step": 394450
    },
    {
      "epoch": 0.04466919383500806,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012112031203120313,
      "loss": 3.2619,
      "step": 394460
    },
    {
      "epoch": 0.04468783707116624,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012111831183118312,
      "loss": 3.2583,
      "step": 394470
    },
    {
      "epoch": 0.04470648030732443,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012111631163116313,
      "loss": 3.1604,
      "step": 394480
    },
    {
      "epoch": 0.04472512354348261,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012111431143114311,
      "loss": 3.2326,
      "step": 394490
    },
    {
      "epoch": 0.044743766779640794,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012111231123112312,
      "loss": 3.3495,
      "step": 394500
    },
    {
      "epoch": 0.044762410015798976,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001211103110311031,
      "loss": 3.2026,
      "step": 394510
    },
    {
      "epoch": 0.04478105325195716,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012110831083108311,
      "loss": 3.1748,
      "step": 394520
    },
    {
      "epoch": 0.04479969648811535,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001211063106310631,
      "loss": 3.1836,
      "step": 394530
    },
    {
      "epoch": 0.04481833972427353,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001211043104310431,
      "loss": 3.266,
      "step": 394540
    },
    {
      "epoch": 0.04483698296043171,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001211023102310231,
      "loss": 3.316,
      "step": 394550
    },
    {
      "epoch": 0.04485562619658989,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012110031003100311,
      "loss": 3.1163,
      "step": 394560
    },
    {
      "epoch": 0.04487426943274808,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012109830983098312,
      "loss": 3.176,
      "step": 394570
    },
    {
      "epoch": 0.044892912668906264,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001210963096309631,
      "loss": 3.1395,
      "step": 394580
    },
    {
      "epoch": 0.044911555905064446,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012109430943094311,
      "loss": 3.2937,
      "step": 394590
    },
    {
      "epoch": 0.04493019914122263,
      "grad_norm": 0.75,
      "learning_rate": 0.00012109230923092309,
      "loss": 3.2183,
      "step": 394600
    },
    {
      "epoch": 0.04494884237738082,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001210903090309031,
      "loss": 3.1499,
      "step": 394610
    },
    {
      "epoch": 0.044967485613539,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012108830883088308,
      "loss": 3.229,
      "step": 394620
    },
    {
      "epoch": 0.04498612884969718,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001210863086308631,
      "loss": 3.2569,
      "step": 394630
    },
    {
      "epoch": 0.04500477208585536,
      "grad_norm": 0.75,
      "learning_rate": 0.00012108430843084309,
      "loss": 3.2598,
      "step": 394640
    },
    {
      "epoch": 0.04502341532201355,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001210823082308231,
      "loss": 3.3275,
      "step": 394650
    },
    {
      "epoch": 0.045042058558171734,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012108030803080308,
      "loss": 3.2021,
      "step": 394660
    },
    {
      "epoch": 0.045060701794329916,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012107830783078309,
      "loss": 3.2117,
      "step": 394670
    },
    {
      "epoch": 0.0450793450304881,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012107630763076307,
      "loss": 3.1555,
      "step": 394680
    },
    {
      "epoch": 0.04509798826664628,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012107430743074308,
      "loss": 3.1606,
      "step": 394690
    },
    {
      "epoch": 0.04511663150280447,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012107230723072308,
      "loss": 3.1452,
      "step": 394700
    },
    {
      "epoch": 0.04513527473896265,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012107030703070307,
      "loss": 3.2781,
      "step": 394710
    },
    {
      "epoch": 0.04515391797512083,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012106830683068307,
      "loss": 3.1508,
      "step": 394720
    },
    {
      "epoch": 0.045172561211279015,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012106630663066308,
      "loss": 3.1515,
      "step": 394730
    },
    {
      "epoch": 0.045191204447437204,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012106430643064306,
      "loss": 3.1687,
      "step": 394740
    },
    {
      "epoch": 0.045209847683595386,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012106230623062307,
      "loss": 3.1584,
      "step": 394750
    },
    {
      "epoch": 0.04522849091975357,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012106030603060308,
      "loss": 3.1603,
      "step": 394760
    },
    {
      "epoch": 0.04524713415591175,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012105830583058306,
      "loss": 3.1854,
      "step": 394770
    },
    {
      "epoch": 0.04526577739206994,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012105630563056307,
      "loss": 2.9944,
      "step": 394780
    },
    {
      "epoch": 0.04528442062822812,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012105430543054306,
      "loss": 3.2018,
      "step": 394790
    },
    {
      "epoch": 0.0453030638643863,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012105230523052307,
      "loss": 3.1771,
      "step": 394800
    },
    {
      "epoch": 0.045321707100544485,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012105030503050305,
      "loss": 3.1514,
      "step": 394810
    },
    {
      "epoch": 0.045340350336702674,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012104830483048306,
      "loss": 3.2757,
      "step": 394820
    },
    {
      "epoch": 0.045358993572860856,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012104630463046305,
      "loss": 3.2466,
      "step": 394830
    },
    {
      "epoch": 0.04537763680901904,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012104430443044306,
      "loss": 3.2916,
      "step": 394840
    },
    {
      "epoch": 0.04539628004517722,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012104230423042304,
      "loss": 3.1278,
      "step": 394850
    },
    {
      "epoch": 0.0454149232813354,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012104030403040305,
      "loss": 3.3188,
      "step": 394860
    },
    {
      "epoch": 0.04543356651749359,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012103830383038304,
      "loss": 3.1581,
      "step": 394870
    },
    {
      "epoch": 0.04545220975365177,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012103630363036305,
      "loss": 3.1808,
      "step": 394880
    },
    {
      "epoch": 0.045470852989809955,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012103430343034303,
      "loss": 3.2536,
      "step": 394890
    },
    {
      "epoch": 0.04548949622596814,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012103230323032304,
      "loss": 3.1465,
      "step": 394900
    },
    {
      "epoch": 0.045508139462126326,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012103030303030302,
      "loss": 3.2119,
      "step": 394910
    },
    {
      "epoch": 0.04552678269828451,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012102830283028303,
      "loss": 3.1137,
      "step": 394920
    },
    {
      "epoch": 0.04554542593444269,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012102630263026302,
      "loss": 3.1864,
      "step": 394930
    },
    {
      "epoch": 0.04556406917060087,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012102430243024302,
      "loss": 3.0594,
      "step": 394940
    },
    {
      "epoch": 0.04558271240675906,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012102230223022303,
      "loss": 3.1595,
      "step": 394950
    },
    {
      "epoch": 0.04560135564291724,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012102030203020303,
      "loss": 3.2629,
      "step": 394960
    },
    {
      "epoch": 0.045619998879075425,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012101830183018304,
      "loss": 3.0513,
      "step": 394970
    },
    {
      "epoch": 0.04563864211523361,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012101630163016302,
      "loss": 3.2966,
      "step": 394980
    },
    {
      "epoch": 0.045657285351391796,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012101430143014303,
      "loss": 3.2809,
      "step": 394990
    },
    {
      "epoch": 0.04567592858754998,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012101230123012301,
      "loss": 3.0972,
      "step": 395000
    },
    {
      "epoch": 0.04569457182370816,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012101030103010302,
      "loss": 3.0827,
      "step": 395010
    },
    {
      "epoch": 0.04571321505986634,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012100830083008302,
      "loss": 3.2484,
      "step": 395020
    },
    {
      "epoch": 0.045731858296024525,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012100630063006301,
      "loss": 3.137,
      "step": 395030
    },
    {
      "epoch": 0.045750501532182714,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012100430043004301,
      "loss": 3.2689,
      "step": 395040
    },
    {
      "epoch": 0.045769144768340896,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012100230023002302,
      "loss": 3.1347,
      "step": 395050
    },
    {
      "epoch": 0.04578778800449908,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000121000300030003,
      "loss": 3.2635,
      "step": 395060
    },
    {
      "epoch": 0.04580643124065726,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012099829982998301,
      "loss": 3.2764,
      "step": 395070
    },
    {
      "epoch": 0.04582507447681545,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012099629962996299,
      "loss": 3.2307,
      "step": 395080
    },
    {
      "epoch": 0.04584371771297363,
      "grad_norm": 0.796875,
      "learning_rate": 0.000120994299429943,
      "loss": 3.2449,
      "step": 395090
    },
    {
      "epoch": 0.04586236094913181,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000120992299229923,
      "loss": 3.1761,
      "step": 395100
    },
    {
      "epoch": 0.045881004185289995,
      "grad_norm": 0.75,
      "learning_rate": 0.000120990299029903,
      "loss": 3.2144,
      "step": 395110
    },
    {
      "epoch": 0.045899647421448184,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012098829882988299,
      "loss": 3.1789,
      "step": 395120
    },
    {
      "epoch": 0.045918290657606366,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000120986298629863,
      "loss": 3.2157,
      "step": 395130
    },
    {
      "epoch": 0.04593693389376455,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012098429842984298,
      "loss": 3.0968,
      "step": 395140
    },
    {
      "epoch": 0.04595557712992273,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012098229822982299,
      "loss": 3.1357,
      "step": 395150
    },
    {
      "epoch": 0.04597422036608092,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000120980298029803,
      "loss": 3.3494,
      "step": 395160
    },
    {
      "epoch": 0.0459928636022391,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012097829782978298,
      "loss": 3.1021,
      "step": 395170
    },
    {
      "epoch": 0.04601150683839728,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012097629762976299,
      "loss": 3.154,
      "step": 395180
    },
    {
      "epoch": 0.046030150074555465,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012097429742974298,
      "loss": 3.2225,
      "step": 395190
    },
    {
      "epoch": 0.04604879331071365,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012097229722972299,
      "loss": 3.2152,
      "step": 395200
    },
    {
      "epoch": 0.046067436546871836,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012097029702970297,
      "loss": 3.2283,
      "step": 395210
    },
    {
      "epoch": 0.04608607978303002,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00012096829682968298,
      "loss": 3.1459,
      "step": 395220
    },
    {
      "epoch": 0.0461047230191882,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012096629662966296,
      "loss": 3.1722,
      "step": 395230
    },
    {
      "epoch": 0.04612336625534638,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012096429642964297,
      "loss": 3.253,
      "step": 395240
    },
    {
      "epoch": 0.04614200949150457,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012096229622962296,
      "loss": 3.2396,
      "step": 395250
    },
    {
      "epoch": 0.04616065272766275,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012096029602960296,
      "loss": 3.1206,
      "step": 395260
    },
    {
      "epoch": 0.046179295963820935,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012095829582958296,
      "loss": 3.1543,
      "step": 395270
    },
    {
      "epoch": 0.04619793919997912,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012095629562956297,
      "loss": 3.1062,
      "step": 395280
    },
    {
      "epoch": 0.046216582436137306,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012095429542954295,
      "loss": 3.2696,
      "step": 395290
    },
    {
      "epoch": 0.04623522567229549,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012095229522952296,
      "loss": 3.3427,
      "step": 395300
    },
    {
      "epoch": 0.04625386890845367,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012095029502950294,
      "loss": 3.2284,
      "step": 395310
    },
    {
      "epoch": 0.04627251214461185,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012094829482948295,
      "loss": 3.1728,
      "step": 395320
    },
    {
      "epoch": 0.04629115538077004,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012094629462946295,
      "loss": 3.1496,
      "step": 395330
    },
    {
      "epoch": 0.04630979861692822,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00012094429442944296,
      "loss": 3.0349,
      "step": 395340
    },
    {
      "epoch": 0.046328441853086405,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012094229422942295,
      "loss": 3.2092,
      "step": 395350
    },
    {
      "epoch": 0.04634708508924459,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012094029402940295,
      "loss": 3.2185,
      "step": 395360
    },
    {
      "epoch": 0.04636572832540277,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012093829382938296,
      "loss": 3.1977,
      "step": 395370
    },
    {
      "epoch": 0.04638437156156096,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012093629362936294,
      "loss": 3.3421,
      "step": 395380
    },
    {
      "epoch": 0.04640301479771914,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012093429342934295,
      "loss": 3.3186,
      "step": 395390
    },
    {
      "epoch": 0.04642165803387732,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012093229322932293,
      "loss": 2.9903,
      "step": 395400
    },
    {
      "epoch": 0.046440301270035504,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012093029302930294,
      "loss": 3.2286,
      "step": 395410
    },
    {
      "epoch": 0.04645894450619369,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012092829282928294,
      "loss": 3.1322,
      "step": 395420
    },
    {
      "epoch": 0.046477587742351875,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012092629262926294,
      "loss": 3.2122,
      "step": 395430
    },
    {
      "epoch": 0.04649623097851006,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012092429242924293,
      "loss": 3.2861,
      "step": 395440
    },
    {
      "epoch": 0.04651487421466824,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012092229222922294,
      "loss": 2.9559,
      "step": 395450
    },
    {
      "epoch": 0.04653351745082643,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012092029202920292,
      "loss": 3.2369,
      "step": 395460
    },
    {
      "epoch": 0.04655216068698461,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012091829182918293,
      "loss": 3.238,
      "step": 395470
    },
    {
      "epoch": 0.04657080392314279,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012091629162916291,
      "loss": 3.1953,
      "step": 395480
    },
    {
      "epoch": 0.046589447159300974,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012091429142914292,
      "loss": 3.1771,
      "step": 395490
    },
    {
      "epoch": 0.04660809039545916,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012091229122912291,
      "loss": 3.2134,
      "step": 395500
    },
    {
      "epoch": 0.046626733631617345,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012091029102910292,
      "loss": 3.102,
      "step": 395510
    },
    {
      "epoch": 0.04664537686777553,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001209082908290829,
      "loss": 3.1604,
      "step": 395520
    },
    {
      "epoch": 0.04666402010393371,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012090629062906291,
      "loss": 3.1736,
      "step": 395530
    },
    {
      "epoch": 0.04668266334009189,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001209042904290429,
      "loss": 3.2149,
      "step": 395540
    },
    {
      "epoch": 0.04670130657625008,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001209022902290229,
      "loss": 3.0554,
      "step": 395550
    },
    {
      "epoch": 0.04671994981240826,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012090029002900291,
      "loss": 3.2291,
      "step": 395560
    },
    {
      "epoch": 0.046738593048566444,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001208982898289829,
      "loss": 3.2323,
      "step": 395570
    },
    {
      "epoch": 0.046757236284724626,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001208962896289629,
      "loss": 2.9675,
      "step": 395580
    },
    {
      "epoch": 0.046775879520882815,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001208942894289429,
      "loss": 3.1727,
      "step": 395590
    },
    {
      "epoch": 0.046794522757041,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012089228922892291,
      "loss": 3.1228,
      "step": 395600
    },
    {
      "epoch": 0.04681316599319918,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012089028902890289,
      "loss": 3.2099,
      "step": 395610
    },
    {
      "epoch": 0.04683180922935736,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001208882888288829,
      "loss": 3.1622,
      "step": 395620
    },
    {
      "epoch": 0.04685045246551555,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012088628862886288,
      "loss": 3.2296,
      "step": 395630
    },
    {
      "epoch": 0.04686909570167373,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012088428842884289,
      "loss": 3.3336,
      "step": 395640
    },
    {
      "epoch": 0.046887738937831915,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012088228822882289,
      "loss": 3.1556,
      "step": 395650
    },
    {
      "epoch": 0.0469063821739901,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001208802880288029,
      "loss": 3.1143,
      "step": 395660
    },
    {
      "epoch": 0.046925025410148286,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012087828782878288,
      "loss": 3.1487,
      "step": 395670
    },
    {
      "epoch": 0.04694366864630647,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012087628762876289,
      "loss": 3.1866,
      "step": 395680
    },
    {
      "epoch": 0.04696231188246465,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012087428742874287,
      "loss": 3.1415,
      "step": 395690
    },
    {
      "epoch": 0.04698095511862283,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012087228722872288,
      "loss": 3.1019,
      "step": 395700
    },
    {
      "epoch": 0.046999598354781014,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012087028702870286,
      "loss": 3.1228,
      "step": 395710
    },
    {
      "epoch": 0.0470182415909392,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012086828682868287,
      "loss": 3.1683,
      "step": 395720
    },
    {
      "epoch": 0.047036884827097385,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012086628662866287,
      "loss": 3.1296,
      "step": 395730
    },
    {
      "epoch": 0.04705552806325557,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012086428642864288,
      "loss": 3.1579,
      "step": 395740
    },
    {
      "epoch": 0.04707417129941375,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012086228622862288,
      "loss": 3.2117,
      "step": 395750
    },
    {
      "epoch": 0.04709281453557194,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012086028602860287,
      "loss": 3.2034,
      "step": 395760
    },
    {
      "epoch": 0.04711145777173012,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012085828582858288,
      "loss": 3.0252,
      "step": 395770
    },
    {
      "epoch": 0.0471301010078883,
      "grad_norm": 0.75,
      "learning_rate": 0.00012085628562856286,
      "loss": 3.1392,
      "step": 395780
    },
    {
      "epoch": 0.047148744244046484,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012085428542854287,
      "loss": 2.9979,
      "step": 395790
    },
    {
      "epoch": 0.04716738748020467,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012085228522852285,
      "loss": 3.3154,
      "step": 395800
    },
    {
      "epoch": 0.047186030716362855,
      "grad_norm": 0.75,
      "learning_rate": 0.00012085028502850286,
      "loss": 3.1382,
      "step": 395810
    },
    {
      "epoch": 0.04720467395252104,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012084828482848285,
      "loss": 3.19,
      "step": 395820
    },
    {
      "epoch": 0.04722331718867922,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012084628462846286,
      "loss": 3.3377,
      "step": 395830
    },
    {
      "epoch": 0.04724196042483741,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012084428442844284,
      "loss": 3.1144,
      "step": 395840
    },
    {
      "epoch": 0.04726060366099559,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012084228422842285,
      "loss": 3.1272,
      "step": 395850
    },
    {
      "epoch": 0.04727924689715377,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012084028402840284,
      "loss": 3.2399,
      "step": 395860
    },
    {
      "epoch": 0.047297890133311954,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012083828382838285,
      "loss": 3.3066,
      "step": 395870
    },
    {
      "epoch": 0.04731653336947014,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012083628362836284,
      "loss": 3.085,
      "step": 395880
    },
    {
      "epoch": 0.047335176605628325,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012083428342834284,
      "loss": 2.9295,
      "step": 395890
    },
    {
      "epoch": 0.04735381984178651,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012083228322832283,
      "loss": 3.0313,
      "step": 395900
    },
    {
      "epoch": 0.04737246307794469,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012083028302830284,
      "loss": 3.1307,
      "step": 395910
    },
    {
      "epoch": 0.04739110631410287,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012082828282828282,
      "loss": 3.0441,
      "step": 395920
    },
    {
      "epoch": 0.04740974955026106,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012082628262826283,
      "loss": 3.2872,
      "step": 395930
    },
    {
      "epoch": 0.04742839278641924,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012082428242824284,
      "loss": 3.2438,
      "step": 395940
    },
    {
      "epoch": 0.047447036022577424,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012082228222822282,
      "loss": 3.1891,
      "step": 395950
    },
    {
      "epoch": 0.047465679258735606,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012082028202820283,
      "loss": 3.2651,
      "step": 395960
    },
    {
      "epoch": 0.047484322494893795,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012081828182818283,
      "loss": 3.1571,
      "step": 395970
    },
    {
      "epoch": 0.04750296573105198,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012081628162816282,
      "loss": 3.2588,
      "step": 395980
    },
    {
      "epoch": 0.04752160896721016,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012081428142814282,
      "loss": 3.2498,
      "step": 395990
    },
    {
      "epoch": 0.04754025220336834,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012081228122812283,
      "loss": 3.2544,
      "step": 396000
    },
    {
      "epoch": 0.04755889543952653,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012081028102810281,
      "loss": 3.1528,
      "step": 396010
    },
    {
      "epoch": 0.04757753867568471,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012080828082808282,
      "loss": 3.2649,
      "step": 396020
    },
    {
      "epoch": 0.047596181911842894,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001208062806280628,
      "loss": 3.2239,
      "step": 396030
    },
    {
      "epoch": 0.047614825148001076,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012080428042804281,
      "loss": 3.2393,
      "step": 396040
    },
    {
      "epoch": 0.047633468384159265,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001208022802280228,
      "loss": 3.1001,
      "step": 396050
    },
    {
      "epoch": 0.04765211162031745,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012080028002800282,
      "loss": 3.1701,
      "step": 396060
    },
    {
      "epoch": 0.04767075485647563,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001207982798279828,
      "loss": 3.2078,
      "step": 396070
    },
    {
      "epoch": 0.04768939809263381,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012079627962796281,
      "loss": 3.1465,
      "step": 396080
    },
    {
      "epoch": 0.04770804132879199,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012079427942794279,
      "loss": 3.1592,
      "step": 396090
    },
    {
      "epoch": 0.04772668456495018,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001207922792279228,
      "loss": 3.2444,
      "step": 396100
    },
    {
      "epoch": 0.047745327801108364,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012079027902790278,
      "loss": 3.1669,
      "step": 396110
    },
    {
      "epoch": 0.047763971037266546,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012078827882788279,
      "loss": 3.1336,
      "step": 396120
    },
    {
      "epoch": 0.04778261427342473,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012078627862786278,
      "loss": 3.2474,
      "step": 396130
    },
    {
      "epoch": 0.04780125750958292,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001207842784278428,
      "loss": 3.158,
      "step": 396140
    },
    {
      "epoch": 0.0478199007457411,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001207822782278228,
      "loss": 3.0769,
      "step": 396150
    },
    {
      "epoch": 0.04783854398189928,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012078027802780279,
      "loss": 3.1629,
      "step": 396160
    },
    {
      "epoch": 0.04785718721805746,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001207782778277828,
      "loss": 3.1198,
      "step": 396170
    },
    {
      "epoch": 0.04787583045421565,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012077627762776278,
      "loss": 3.3246,
      "step": 396180
    },
    {
      "epoch": 0.047894473690373834,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012077427742774279,
      "loss": 3.1009,
      "step": 396190
    },
    {
      "epoch": 0.047913116926532016,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012077227722772277,
      "loss": 3.2244,
      "step": 396200
    },
    {
      "epoch": 0.0479317601626902,
      "grad_norm": 0.75,
      "learning_rate": 0.00012077027702770278,
      "loss": 3.0767,
      "step": 396210
    },
    {
      "epoch": 0.04795040339884839,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012076827682768277,
      "loss": 3.1717,
      "step": 396220
    },
    {
      "epoch": 0.04796904663500657,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012076627662766278,
      "loss": 3.2635,
      "step": 396230
    },
    {
      "epoch": 0.04798768987116475,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012076427642764276,
      "loss": 3.1927,
      "step": 396240
    },
    {
      "epoch": 0.048006333107322934,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012076227622762277,
      "loss": 3.1817,
      "step": 396250
    },
    {
      "epoch": 0.048024976343481116,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012076027602760275,
      "loss": 3.1671,
      "step": 396260
    },
    {
      "epoch": 0.048043619579639305,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012075827582758276,
      "loss": 3.2187,
      "step": 396270
    },
    {
      "epoch": 0.04806226281579749,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012075627562756276,
      "loss": 3.0712,
      "step": 396280
    },
    {
      "epoch": 0.04808090605195567,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012075427542754277,
      "loss": 3.2324,
      "step": 396290
    },
    {
      "epoch": 0.04809954928811385,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012075227522752275,
      "loss": 3.1512,
      "step": 396300
    },
    {
      "epoch": 0.04811819252427204,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012075027502750276,
      "loss": 3.2161,
      "step": 396310
    },
    {
      "epoch": 0.04813683576043022,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012074827482748274,
      "loss": 3.088,
      "step": 396320
    },
    {
      "epoch": 0.048155478996588404,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012074627462746275,
      "loss": 3.2385,
      "step": 396330
    },
    {
      "epoch": 0.048174122232746586,
      "grad_norm": 0.75,
      "learning_rate": 0.00012074427442744276,
      "loss": 3.2325,
      "step": 396340
    },
    {
      "epoch": 0.048192765468904775,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012074227422742274,
      "loss": 3.2216,
      "step": 396350
    },
    {
      "epoch": 0.04821140870506296,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012074027402740275,
      "loss": 3.1772,
      "step": 396360
    },
    {
      "epoch": 0.04823005194122114,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012073827382738275,
      "loss": 3.1478,
      "step": 396370
    },
    {
      "epoch": 0.04824869517737932,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012073627362736276,
      "loss": 3.3027,
      "step": 396380
    },
    {
      "epoch": 0.04826733841353751,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012073427342734274,
      "loss": 3.228,
      "step": 396390
    },
    {
      "epoch": 0.04828598164969569,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012073227322732275,
      "loss": 3.0895,
      "step": 396400
    },
    {
      "epoch": 0.048304624885853874,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012073027302730273,
      "loss": 3.1489,
      "step": 396410
    },
    {
      "epoch": 0.048323268122012056,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012072827282728274,
      "loss": 3.2422,
      "step": 396420
    },
    {
      "epoch": 0.04834191135817024,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012072627262726272,
      "loss": 3.0855,
      "step": 396430
    },
    {
      "epoch": 0.04836055459432843,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012072427242724273,
      "loss": 3.2475,
      "step": 396440
    },
    {
      "epoch": 0.04837919783048661,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012072227222722273,
      "loss": 3.1078,
      "step": 396450
    },
    {
      "epoch": 0.04839784106664479,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00012072027202720273,
      "loss": 3.2888,
      "step": 396460
    },
    {
      "epoch": 0.04841648430280297,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012071827182718272,
      "loss": 3.1865,
      "step": 396470
    },
    {
      "epoch": 0.04843512753896116,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012071627162716273,
      "loss": 3.0602,
      "step": 396480
    },
    {
      "epoch": 0.048453770775119344,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012071427142714271,
      "loss": 3.0977,
      "step": 396490
    },
    {
      "epoch": 0.048472414011277526,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012071227122712272,
      "loss": 3.1372,
      "step": 396500
    },
    {
      "epoch": 0.04849105724743571,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012071027102710271,
      "loss": 3.2056,
      "step": 396510
    },
    {
      "epoch": 0.0485097004835939,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012070827082708271,
      "loss": 3.1227,
      "step": 396520
    },
    {
      "epoch": 0.04852834371975208,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012070627062706272,
      "loss": 3.1487,
      "step": 396530
    },
    {
      "epoch": 0.04854698695591026,
      "grad_norm": 1.453125,
      "learning_rate": 0.00012070427042704271,
      "loss": 3.3336,
      "step": 396540
    },
    {
      "epoch": 0.04856563019206844,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012070227022702272,
      "loss": 3.216,
      "step": 396550
    },
    {
      "epoch": 0.04858427342822663,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001207002700270027,
      "loss": 3.1901,
      "step": 396560
    },
    {
      "epoch": 0.048602916664384814,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012069826982698271,
      "loss": 3.1407,
      "step": 396570
    },
    {
      "epoch": 0.048621559900542996,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001206962696269627,
      "loss": 3.158,
      "step": 396580
    },
    {
      "epoch": 0.04864020313670118,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001206942694269427,
      "loss": 3.1779,
      "step": 396590
    },
    {
      "epoch": 0.04865884637285936,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001206922692269227,
      "loss": 3.2153,
      "step": 396600
    },
    {
      "epoch": 0.04867748960901755,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012069026902690271,
      "loss": 3.1377,
      "step": 396610
    },
    {
      "epoch": 0.04869613284517573,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012068826882688269,
      "loss": 3.1905,
      "step": 396620
    },
    {
      "epoch": 0.04871477608133391,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001206862686268627,
      "loss": 3.1914,
      "step": 396630
    },
    {
      "epoch": 0.048733419317492095,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012068426842684268,
      "loss": 3.1041,
      "step": 396640
    },
    {
      "epoch": 0.048752062553650284,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012068226822682269,
      "loss": 3.2597,
      "step": 396650
    },
    {
      "epoch": 0.048770705789808466,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012068026802680267,
      "loss": 3.0881,
      "step": 396660
    },
    {
      "epoch": 0.04878934902596665,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012067826782678268,
      "loss": 3.2155,
      "step": 396670
    },
    {
      "epoch": 0.04880799226212483,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012067626762676268,
      "loss": 3.2481,
      "step": 396680
    },
    {
      "epoch": 0.04882663549828302,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012067426742674269,
      "loss": 3.1798,
      "step": 396690
    },
    {
      "epoch": 0.0488452787344412,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012067226722672267,
      "loss": 3.0899,
      "step": 396700
    },
    {
      "epoch": 0.04886392197059938,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012067026702670268,
      "loss": 3.1889,
      "step": 396710
    },
    {
      "epoch": 0.048882565206757565,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012066826682668266,
      "loss": 3.1747,
      "step": 396720
    },
    {
      "epoch": 0.048901208442915754,
      "grad_norm": 0.75,
      "learning_rate": 0.00012066626662666267,
      "loss": 3.2225,
      "step": 396730
    },
    {
      "epoch": 0.048919851679073936,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012066426642664268,
      "loss": 3.1543,
      "step": 396740
    },
    {
      "epoch": 0.04893849491523212,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012066226622662266,
      "loss": 3.2274,
      "step": 396750
    },
    {
      "epoch": 0.0489571381513903,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012066026602660267,
      "loss": 3.1627,
      "step": 396760
    },
    {
      "epoch": 0.04897578138754848,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012065826582658267,
      "loss": 3.1358,
      "step": 396770
    },
    {
      "epoch": 0.04899442462370667,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012065626562656267,
      "loss": 3.2599,
      "step": 396780
    },
    {
      "epoch": 0.04901306785986485,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012065426542654266,
      "loss": 3.1472,
      "step": 396790
    },
    {
      "epoch": 0.049031711096023035,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012065226522652267,
      "loss": 3.2842,
      "step": 396800
    },
    {
      "epoch": 0.04905035433218122,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012065026502650265,
      "loss": 3.0149,
      "step": 396810
    },
    {
      "epoch": 0.049068997568339406,
      "grad_norm": 0.890625,
      "learning_rate": 0.00012064826482648266,
      "loss": 3.2508,
      "step": 396820
    },
    {
      "epoch": 0.04908764080449759,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012064626462646265,
      "loss": 3.2167,
      "step": 396830
    },
    {
      "epoch": 0.04910628404065577,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012064426442644265,
      "loss": 3.2196,
      "step": 396840
    },
    {
      "epoch": 0.04912492727681395,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012064226422642264,
      "loss": 3.0663,
      "step": 396850
    },
    {
      "epoch": 0.04914357051297214,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012064026402640265,
      "loss": 3.1688,
      "step": 396860
    },
    {
      "epoch": 0.049162213749130323,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012063826382638264,
      "loss": 3.0943,
      "step": 396870
    },
    {
      "epoch": 0.049180856985288506,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012063626362636264,
      "loss": 3.3746,
      "step": 396880
    },
    {
      "epoch": 0.04919950022144669,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012063426342634263,
      "loss": 3.0034,
      "step": 396890
    },
    {
      "epoch": 0.049218143457604877,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012063226322632264,
      "loss": 3.0021,
      "step": 396900
    },
    {
      "epoch": 0.04923678669376306,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012063026302630263,
      "loss": 3.2458,
      "step": 396910
    },
    {
      "epoch": 0.04925542992992124,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012062826282628264,
      "loss": 3.2157,
      "step": 396920
    },
    {
      "epoch": 0.04927407316607942,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012062626262626265,
      "loss": 3.3195,
      "step": 396930
    },
    {
      "epoch": 0.049292716402237605,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012062426242624263,
      "loss": 3.2258,
      "step": 396940
    },
    {
      "epoch": 0.049311359638395794,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012062226222622264,
      "loss": 3.0035,
      "step": 396950
    },
    {
      "epoch": 0.049330002874553976,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012062026202620262,
      "loss": 3.186,
      "step": 396960
    },
    {
      "epoch": 0.04934864611071216,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012061826182618263,
      "loss": 3.3087,
      "step": 396970
    },
    {
      "epoch": 0.04936728934687034,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012061626162616261,
      "loss": 3.1338,
      "step": 396980
    },
    {
      "epoch": 0.04938593258302853,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012061426142614262,
      "loss": 3.0899,
      "step": 396990
    },
    {
      "epoch": 0.04940457581918671,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012061226122612262,
      "loss": 3.1754,
      "step": 397000
    },
    {
      "epoch": 0.04942321905534489,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012061026102610263,
      "loss": 3.2083,
      "step": 397010
    },
    {
      "epoch": 0.049441862291503075,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012060826082608261,
      "loss": 3.2033,
      "step": 397020
    },
    {
      "epoch": 0.049460505527661264,
      "grad_norm": 0.75,
      "learning_rate": 0.00012060626062606262,
      "loss": 3.1298,
      "step": 397030
    },
    {
      "epoch": 0.049479148763819446,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001206042604260426,
      "loss": 3.1993,
      "step": 397040
    },
    {
      "epoch": 0.04949779199997763,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012060226022602261,
      "loss": 3.1211,
      "step": 397050
    },
    {
      "epoch": 0.04951643523613581,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012060026002600259,
      "loss": 3.3364,
      "step": 397060
    },
    {
      "epoch": 0.049535078472294,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001205982598259826,
      "loss": 3.1136,
      "step": 397070
    },
    {
      "epoch": 0.04955372170845218,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001205962596259626,
      "loss": 3.0788,
      "step": 397080
    },
    {
      "epoch": 0.04957236494461036,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001205942594259426,
      "loss": 3.1259,
      "step": 397090
    },
    {
      "epoch": 0.049591008180768545,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012059225922592259,
      "loss": 3.2084,
      "step": 397100
    },
    {
      "epoch": 0.04960965141692673,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001205902590259026,
      "loss": 3.1493,
      "step": 397110
    },
    {
      "epoch": 0.049628294653084916,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012058825882588258,
      "loss": 3.2047,
      "step": 397120
    },
    {
      "epoch": 0.0496469378892431,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012058625862586259,
      "loss": 3.3601,
      "step": 397130
    },
    {
      "epoch": 0.04966558112540128,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001205842584258426,
      "loss": 3.046,
      "step": 397140
    },
    {
      "epoch": 0.04968422436155946,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012058225822582259,
      "loss": 3.2026,
      "step": 397150
    },
    {
      "epoch": 0.04970286759771765,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012058025802580259,
      "loss": 3.2339,
      "step": 397160
    },
    {
      "epoch": 0.04972151083387583,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012057825782578258,
      "loss": 3.1174,
      "step": 397170
    },
    {
      "epoch": 0.049740154070034015,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001205762576257626,
      "loss": 3.2252,
      "step": 397180
    },
    {
      "epoch": 0.0497587973061922,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012057425742574258,
      "loss": 3.22,
      "step": 397190
    },
    {
      "epoch": 0.049777440542350386,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012057225722572258,
      "loss": 3.1486,
      "step": 397200
    },
    {
      "epoch": 0.04979608377850857,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00012057025702570257,
      "loss": 3.2897,
      "step": 397210
    },
    {
      "epoch": 0.04981472701466675,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012056825682568258,
      "loss": 3.1967,
      "step": 397220
    },
    {
      "epoch": 0.04983337025082493,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012056625662566257,
      "loss": 3.1701,
      "step": 397230
    },
    {
      "epoch": 0.04985201348698312,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012056425642564258,
      "loss": 3.1788,
      "step": 397240
    },
    {
      "epoch": 0.0498706567231413,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012056225622562256,
      "loss": 3.1127,
      "step": 397250
    },
    {
      "epoch": 0.049889299959299485,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012056025602560257,
      "loss": 3.2735,
      "step": 397260
    },
    {
      "epoch": 0.04990794319545767,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012055825582558255,
      "loss": 3.1997,
      "step": 397270
    },
    {
      "epoch": 0.04992658643161585,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012055625562556256,
      "loss": 3.2293,
      "step": 397280
    },
    {
      "epoch": 0.04994522966777404,
      "grad_norm": 0.75,
      "learning_rate": 0.00012055425542554254,
      "loss": 3.2065,
      "step": 397290
    },
    {
      "epoch": 0.04996387290393222,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012055225522552255,
      "loss": 3.1837,
      "step": 397300
    },
    {
      "epoch": 0.0499825161400904,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012055025502550255,
      "loss": 3.2722,
      "step": 397310
    },
    {
      "epoch": 0.050001159376248584,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012054825482548256,
      "loss": 3.2681,
      "step": 397320
    },
    {
      "epoch": 0.05001980261240677,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012054625462546257,
      "loss": 3.2592,
      "step": 397330
    },
    {
      "epoch": 0.050038445848564955,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012054425442544255,
      "loss": 3.0396,
      "step": 397340
    },
    {
      "epoch": 0.05005708908472314,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012054225422542256,
      "loss": 3.1848,
      "step": 397350
    },
    {
      "epoch": 0.05007573232088132,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00012054025402540254,
      "loss": 3.1304,
      "step": 397360
    },
    {
      "epoch": 0.05009437555703951,
      "grad_norm": 0.75,
      "learning_rate": 0.00012053825382538255,
      "loss": 3.1175,
      "step": 397370
    },
    {
      "epoch": 0.05011301879319769,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012053625362536253,
      "loss": 3.2122,
      "step": 397380
    },
    {
      "epoch": 0.05013166202935587,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012053425342534254,
      "loss": 3.1349,
      "step": 397390
    },
    {
      "epoch": 0.050150305265514054,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00012053225322532254,
      "loss": 3.0862,
      "step": 397400
    },
    {
      "epoch": 0.05016894850167224,
      "grad_norm": 0.875,
      "learning_rate": 0.00012053025302530255,
      "loss": 3.1507,
      "step": 397410
    },
    {
      "epoch": 0.050187591737830425,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012052825282528253,
      "loss": 3.3541,
      "step": 397420
    },
    {
      "epoch": 0.05020623497398861,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012052625262526254,
      "loss": 3.1822,
      "step": 397430
    },
    {
      "epoch": 0.05022487821014679,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012052425242524252,
      "loss": 3.1331,
      "step": 397440
    },
    {
      "epoch": 0.05024352144630497,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012052225222522253,
      "loss": 3.2197,
      "step": 397450
    },
    {
      "epoch": 0.05026216468246316,
      "grad_norm": 0.75,
      "learning_rate": 0.00012052025202520252,
      "loss": 3.2522,
      "step": 397460
    },
    {
      "epoch": 0.05028080791862134,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012051825182518252,
      "loss": 3.0489,
      "step": 397470
    },
    {
      "epoch": 0.050299451154779524,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012051625162516252,
      "loss": 3.3382,
      "step": 397480
    },
    {
      "epoch": 0.050318094390937707,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012051425142514252,
      "loss": 3.0655,
      "step": 397490
    },
    {
      "epoch": 0.050336737627095895,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001205122512251225,
      "loss": 3.1337,
      "step": 397500
    },
    {
      "epoch": 0.05035538086325408,
      "grad_norm": 0.75,
      "learning_rate": 0.00012051025102510252,
      "loss": 3.2558,
      "step": 397510
    },
    {
      "epoch": 0.05037402409941226,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012050825082508252,
      "loss": 3.1513,
      "step": 397520
    },
    {
      "epoch": 0.05039266733557044,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012050625062506251,
      "loss": 3.245,
      "step": 397530
    },
    {
      "epoch": 0.05041131057172863,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012050425042504252,
      "loss": 3.1537,
      "step": 397540
    },
    {
      "epoch": 0.05042995380788681,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012050225022502251,
      "loss": 3.1626,
      "step": 397550
    },
    {
      "epoch": 0.050448597044044995,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012050025002500252,
      "loss": 3.0807,
      "step": 397560
    },
    {
      "epoch": 0.05046724028020318,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001204982498249825,
      "loss": 3.1673,
      "step": 397570
    },
    {
      "epoch": 0.050485883516361366,
      "grad_norm": 0.75,
      "learning_rate": 0.00012049624962496251,
      "loss": 3.1793,
      "step": 397580
    },
    {
      "epoch": 0.05050452675251955,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001204942494249425,
      "loss": 3.2149,
      "step": 397590
    },
    {
      "epoch": 0.05052316998867773,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001204922492249225,
      "loss": 3.2181,
      "step": 397600
    },
    {
      "epoch": 0.05054181322483591,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012049024902490249,
      "loss": 3.2256,
      "step": 397610
    },
    {
      "epoch": 0.050560456460994094,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001204882488248825,
      "loss": 3.3319,
      "step": 397620
    },
    {
      "epoch": 0.05057909969715228,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012048624862486249,
      "loss": 3.0937,
      "step": 397630
    },
    {
      "epoch": 0.050597742933310465,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001204842484248425,
      "loss": 3.1067,
      "step": 397640
    },
    {
      "epoch": 0.05061638616946865,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012048224822482248,
      "loss": 3.0698,
      "step": 397650
    },
    {
      "epoch": 0.05063502940562683,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012048024802480249,
      "loss": 3.2441,
      "step": 397660
    },
    {
      "epoch": 0.05065367264178502,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012047824782478247,
      "loss": 3.168,
      "step": 397670
    },
    {
      "epoch": 0.0506723158779432,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012047624762476248,
      "loss": 3.226,
      "step": 397680
    },
    {
      "epoch": 0.05069095911410138,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012047424742474246,
      "loss": 3.1643,
      "step": 397690
    },
    {
      "epoch": 0.050709602350259564,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012047224722472247,
      "loss": 3.1152,
      "step": 397700
    },
    {
      "epoch": 0.05072824558641775,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012047024702470247,
      "loss": 3.214,
      "step": 397710
    },
    {
      "epoch": 0.050746888822575935,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012046824682468248,
      "loss": 3.2064,
      "step": 397720
    },
    {
      "epoch": 0.05076553205873412,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012046624662466249,
      "loss": 3.1734,
      "step": 397730
    },
    {
      "epoch": 0.0507841752948923,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012046424642464247,
      "loss": 3.1673,
      "step": 397740
    },
    {
      "epoch": 0.05080281853105049,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012046224622462248,
      "loss": 3.1993,
      "step": 397750
    },
    {
      "epoch": 0.05082146176720867,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012046024602460246,
      "loss": 3.1462,
      "step": 397760
    },
    {
      "epoch": 0.05084010500336685,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012045824582458247,
      "loss": 3.1982,
      "step": 397770
    },
    {
      "epoch": 0.050858748239525034,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012045624562456246,
      "loss": 3.2515,
      "step": 397780
    },
    {
      "epoch": 0.050877391475683216,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012045424542454246,
      "loss": 3.1334,
      "step": 397790
    },
    {
      "epoch": 0.050896034711841405,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012045224522452246,
      "loss": 3.1261,
      "step": 397800
    },
    {
      "epoch": 0.05091467794799959,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012045024502450246,
      "loss": 3.1226,
      "step": 397810
    },
    {
      "epoch": 0.05093332118415777,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012044824482448245,
      "loss": 3.1503,
      "step": 397820
    },
    {
      "epoch": 0.05095196442031595,
      "grad_norm": 0.75,
      "learning_rate": 0.00012044624462446246,
      "loss": 3.3496,
      "step": 397830
    },
    {
      "epoch": 0.05097060765647414,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012044424442444244,
      "loss": 3.2088,
      "step": 397840
    },
    {
      "epoch": 0.05098925089263232,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012044224422442245,
      "loss": 3.32,
      "step": 397850
    },
    {
      "epoch": 0.051007894128790504,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012044024402440244,
      "loss": 3.2639,
      "step": 397860
    },
    {
      "epoch": 0.051026537364948686,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00012043824382438245,
      "loss": 3.1499,
      "step": 397870
    },
    {
      "epoch": 0.051045180601106875,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012043624362436243,
      "loss": 3.0923,
      "step": 397880
    },
    {
      "epoch": 0.05106382383726506,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012043424342434244,
      "loss": 3.1429,
      "step": 397890
    },
    {
      "epoch": 0.05108246707342324,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012043224322432243,
      "loss": 3.1922,
      "step": 397900
    },
    {
      "epoch": 0.05110111030958142,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012043024302430243,
      "loss": 3.2187,
      "step": 397910
    },
    {
      "epoch": 0.05111975354573961,
      "grad_norm": 0.75,
      "learning_rate": 0.00012042824282428244,
      "loss": 3.1445,
      "step": 397920
    },
    {
      "epoch": 0.05113839678189779,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012042624262426243,
      "loss": 3.1517,
      "step": 397930
    },
    {
      "epoch": 0.051157040018055974,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00012042424242424243,
      "loss": 3.0948,
      "step": 397940
    },
    {
      "epoch": 0.051175683254214156,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012042224222422243,
      "loss": 3.1092,
      "step": 397950
    },
    {
      "epoch": 0.05119432649037234,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012042024202420244,
      "loss": 3.1513,
      "step": 397960
    },
    {
      "epoch": 0.05121296972653053,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012041824182418242,
      "loss": 3.2123,
      "step": 397970
    },
    {
      "epoch": 0.05123161296268871,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012041624162416243,
      "loss": 3.3294,
      "step": 397980
    },
    {
      "epoch": 0.05125025619884689,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012041424142414241,
      "loss": 3.2031,
      "step": 397990
    },
    {
      "epoch": 0.05126889943500507,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012041224122412242,
      "loss": 3.1943,
      "step": 398000
    },
    {
      "epoch": 0.05128754267116326,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001204102410241024,
      "loss": 3.1888,
      "step": 398010
    },
    {
      "epoch": 0.051306185907321444,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012040824082408241,
      "loss": 3.1648,
      "step": 398020
    },
    {
      "epoch": 0.051324829143479626,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012040624062406241,
      "loss": 3.1673,
      "step": 398030
    },
    {
      "epoch": 0.05134347237963781,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012040424042404242,
      "loss": 3.2213,
      "step": 398040
    },
    {
      "epoch": 0.051362115615796,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001204022402240224,
      "loss": 3.1625,
      "step": 398050
    },
    {
      "epoch": 0.05138075885195418,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012040024002400241,
      "loss": 3.2452,
      "step": 398060
    },
    {
      "epoch": 0.05139940208811236,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012039823982398239,
      "loss": 3.1754,
      "step": 398070
    },
    {
      "epoch": 0.051418045324270543,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001203962396239624,
      "loss": 3.2019,
      "step": 398080
    },
    {
      "epoch": 0.05143668856042873,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001203942394239424,
      "loss": 3.1651,
      "step": 398090
    },
    {
      "epoch": 0.051455331796586914,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001203922392239224,
      "loss": 3.0469,
      "step": 398100
    },
    {
      "epoch": 0.051473975032745097,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001203902390239024,
      "loss": 3.1724,
      "step": 398110
    },
    {
      "epoch": 0.05149261826890328,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001203882388238824,
      "loss": 3.1175,
      "step": 398120
    },
    {
      "epoch": 0.05151126150506146,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001203862386238624,
      "loss": 3.2176,
      "step": 398130
    },
    {
      "epoch": 0.05152990474121965,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012038423842384239,
      "loss": 3.2322,
      "step": 398140
    },
    {
      "epoch": 0.05154854797737783,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001203822382238224,
      "loss": 3.1503,
      "step": 398150
    },
    {
      "epoch": 0.051567191213536014,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012038023802380238,
      "loss": 3.1685,
      "step": 398160
    },
    {
      "epoch": 0.051585834449694196,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012037823782378239,
      "loss": 3.271,
      "step": 398170
    },
    {
      "epoch": 0.051604477685852385,
      "grad_norm": 0.625,
      "learning_rate": 0.00012037623762376238,
      "loss": 3.1499,
      "step": 398180
    },
    {
      "epoch": 0.05162312092201057,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012037423742374239,
      "loss": 3.27,
      "step": 398190
    },
    {
      "epoch": 0.05164176415816875,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012037223722372237,
      "loss": 3.2148,
      "step": 398200
    },
    {
      "epoch": 0.05166040739432693,
      "grad_norm": 0.578125,
      "learning_rate": 0.00012037023702370238,
      "loss": 3.1874,
      "step": 398210
    },
    {
      "epoch": 0.05167905063048512,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012036823682368237,
      "loss": 3.2146,
      "step": 398220
    },
    {
      "epoch": 0.0516976938666433,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012036623662366237,
      "loss": 3.1495,
      "step": 398230
    },
    {
      "epoch": 0.051716337102801484,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012036423642364236,
      "loss": 3.2571,
      "step": 398240
    },
    {
      "epoch": 0.051734980338959666,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012036223622362237,
      "loss": 3.267,
      "step": 398250
    },
    {
      "epoch": 0.051753623575117855,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012036023602360236,
      "loss": 3.2054,
      "step": 398260
    },
    {
      "epoch": 0.05177226681127604,
      "grad_norm": 0.75,
      "learning_rate": 0.00012035823582358237,
      "loss": 3.2329,
      "step": 398270
    },
    {
      "epoch": 0.05179091004743422,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012035623562356235,
      "loss": 3.2345,
      "step": 398280
    },
    {
      "epoch": 0.0518095532835924,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012035423542354236,
      "loss": 3.2572,
      "step": 398290
    },
    {
      "epoch": 0.05182819651975058,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012035223522352234,
      "loss": 3.2352,
      "step": 398300
    },
    {
      "epoch": 0.05184683975590877,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012035023502350235,
      "loss": 3.0961,
      "step": 398310
    },
    {
      "epoch": 0.051865482992066954,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012034823482348236,
      "loss": 3.1058,
      "step": 398320
    },
    {
      "epoch": 0.051884126228225136,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012034623462346234,
      "loss": 3.1471,
      "step": 398330
    },
    {
      "epoch": 0.05190276946438332,
      "grad_norm": 0.75,
      "learning_rate": 0.00012034423442344235,
      "loss": 3.2052,
      "step": 398340
    },
    {
      "epoch": 0.05192141270054151,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012034223422342235,
      "loss": 3.1531,
      "step": 398350
    },
    {
      "epoch": 0.05194005593669969,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012034023402340236,
      "loss": 3.1027,
      "step": 398360
    },
    {
      "epoch": 0.05195869917285787,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012033823382338234,
      "loss": 3.3447,
      "step": 398370
    },
    {
      "epoch": 0.05197734240901605,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012033623362336235,
      "loss": 3.1501,
      "step": 398380
    },
    {
      "epoch": 0.05199598564517424,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012033423342334233,
      "loss": 3.2078,
      "step": 398390
    },
    {
      "epoch": 0.052014628881332424,
      "grad_norm": 0.75,
      "learning_rate": 0.00012033223322332234,
      "loss": 3.1579,
      "step": 398400
    },
    {
      "epoch": 0.052033272117490606,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012033023302330234,
      "loss": 3.3663,
      "step": 398410
    },
    {
      "epoch": 0.05205191535364879,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012032823282328235,
      "loss": 3.2315,
      "step": 398420
    },
    {
      "epoch": 0.05207055858980698,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012032623262326233,
      "loss": 3.2114,
      "step": 398430
    },
    {
      "epoch": 0.05208920182596516,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012032423242324234,
      "loss": 3.1115,
      "step": 398440
    },
    {
      "epoch": 0.05210784506212334,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012032223222322232,
      "loss": 3.1793,
      "step": 398450
    },
    {
      "epoch": 0.05212648829828152,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012032023202320233,
      "loss": 3.2016,
      "step": 398460
    },
    {
      "epoch": 0.052145131534439705,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012031823182318231,
      "loss": 3.2107,
      "step": 398470
    },
    {
      "epoch": 0.052163774770597894,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012031623162316232,
      "loss": 3.3017,
      "step": 398480
    },
    {
      "epoch": 0.052182418006756076,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012031423142314231,
      "loss": 3.1953,
      "step": 398490
    },
    {
      "epoch": 0.05220106124291426,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012031223122312232,
      "loss": 3.0567,
      "step": 398500
    },
    {
      "epoch": 0.05221970447907244,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012031023102310233,
      "loss": 3.1827,
      "step": 398510
    },
    {
      "epoch": 0.05223834771523063,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012030823082308231,
      "loss": 3.2467,
      "step": 398520
    },
    {
      "epoch": 0.05225699095138881,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012030623062306232,
      "loss": 3.1799,
      "step": 398530
    },
    {
      "epoch": 0.05227563418754699,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001203042304230423,
      "loss": 3.28,
      "step": 398540
    },
    {
      "epoch": 0.052294277423705175,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012030223022302232,
      "loss": 3.1463,
      "step": 398550
    },
    {
      "epoch": 0.052312920659863364,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001203002300230023,
      "loss": 3.0529,
      "step": 398560
    },
    {
      "epoch": 0.052331563896021546,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001202982298229823,
      "loss": 3.2092,
      "step": 398570
    },
    {
      "epoch": 0.05235020713217973,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001202962296229623,
      "loss": 3.0514,
      "step": 398580
    },
    {
      "epoch": 0.05236885036833791,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012029422942294231,
      "loss": 3.1552,
      "step": 398590
    },
    {
      "epoch": 0.0523874936044961,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012029222922292229,
      "loss": 3.1432,
      "step": 398600
    },
    {
      "epoch": 0.05240613684065428,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001202902290229023,
      "loss": 3.2435,
      "step": 398610
    },
    {
      "epoch": 0.05242478007681246,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012028822882288228,
      "loss": 3.2088,
      "step": 398620
    },
    {
      "epoch": 0.052443423312970645,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001202862286228623,
      "loss": 3.2019,
      "step": 398630
    },
    {
      "epoch": 0.05246206654912883,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00012028422842284229,
      "loss": 3.2705,
      "step": 398640
    },
    {
      "epoch": 0.052480709785287016,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012028222822282228,
      "loss": 3.2496,
      "step": 398650
    },
    {
      "epoch": 0.0524993530214452,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012028022802280228,
      "loss": 3.1959,
      "step": 398660
    },
    {
      "epoch": 0.05251799625760338,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012027822782278229,
      "loss": 3.1114,
      "step": 398670
    },
    {
      "epoch": 0.05253663949376156,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012027622762276227,
      "loss": 3.1118,
      "step": 398680
    },
    {
      "epoch": 0.05255528272991975,
      "grad_norm": 0.75,
      "learning_rate": 0.00012027422742274228,
      "loss": 3.1752,
      "step": 398690
    },
    {
      "epoch": 0.05257392596607793,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012027222722272229,
      "loss": 3.0882,
      "step": 398700
    },
    {
      "epoch": 0.052592569202236115,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012027022702270227,
      "loss": 3.0525,
      "step": 398710
    },
    {
      "epoch": 0.0526112124383943,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012026822682268228,
      "loss": 3.2685,
      "step": 398720
    },
    {
      "epoch": 0.052629855674552486,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012026622662266228,
      "loss": 3.1825,
      "step": 398730
    },
    {
      "epoch": 0.05264849891071067,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012026422642264229,
      "loss": 3.1411,
      "step": 398740
    },
    {
      "epoch": 0.05266714214686885,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00012026222622262227,
      "loss": 3.2028,
      "step": 398750
    },
    {
      "epoch": 0.05268578538302703,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00012026022602260228,
      "loss": 3.218,
      "step": 398760
    },
    {
      "epoch": 0.05270442861918522,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012025822582258226,
      "loss": 3.1461,
      "step": 398770
    },
    {
      "epoch": 0.052723071855343404,
      "grad_norm": 0.75,
      "learning_rate": 0.00012025622562256227,
      "loss": 3.1283,
      "step": 398780
    },
    {
      "epoch": 0.052741715091501586,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012025422542254225,
      "loss": 2.9811,
      "step": 398790
    },
    {
      "epoch": 0.05276035832765977,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012025222522252226,
      "loss": 3.2397,
      "step": 398800
    },
    {
      "epoch": 0.05277900156381795,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012025022502250225,
      "loss": 3.2876,
      "step": 398810
    },
    {
      "epoch": 0.05279764479997614,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012024822482248226,
      "loss": 3.064,
      "step": 398820
    },
    {
      "epoch": 0.05281628803613432,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00012024622462246225,
      "loss": 3.1036,
      "step": 398830
    },
    {
      "epoch": 0.0528349312722925,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012024422442244226,
      "loss": 3.1909,
      "step": 398840
    },
    {
      "epoch": 0.052853574508450685,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012024222422242224,
      "loss": 3.1983,
      "step": 398850
    },
    {
      "epoch": 0.052872217744608874,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012024022402240225,
      "loss": 3.168,
      "step": 398860
    },
    {
      "epoch": 0.052890860980767056,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012023822382238223,
      "loss": 3.0562,
      "step": 398870
    },
    {
      "epoch": 0.05290950421692524,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012023622362236224,
      "loss": 3.1971,
      "step": 398880
    },
    {
      "epoch": 0.05292814745308342,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012023422342234223,
      "loss": 3.2488,
      "step": 398890
    },
    {
      "epoch": 0.05294679068924161,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012023222322232224,
      "loss": 3.1123,
      "step": 398900
    },
    {
      "epoch": 0.05296543392539979,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012023022302230225,
      "loss": 3.1131,
      "step": 398910
    },
    {
      "epoch": 0.05298407716155797,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012022822282228223,
      "loss": 3.0745,
      "step": 398920
    },
    {
      "epoch": 0.053002720397716155,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012022622262226224,
      "loss": 3.2381,
      "step": 398930
    },
    {
      "epoch": 0.053021363633874344,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012022422242224222,
      "loss": 3.1989,
      "step": 398940
    },
    {
      "epoch": 0.053040006870032526,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012022222222222223,
      "loss": 3.1898,
      "step": 398950
    },
    {
      "epoch": 0.05305865010619071,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012022022202220222,
      "loss": 3.2622,
      "step": 398960
    },
    {
      "epoch": 0.05307729334234889,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012021822182218222,
      "loss": 3.1538,
      "step": 398970
    },
    {
      "epoch": 0.05309593657850707,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012021622162216222,
      "loss": 3.1227,
      "step": 398980
    },
    {
      "epoch": 0.05311457981466526,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012021422142214223,
      "loss": 3.0054,
      "step": 398990
    },
    {
      "epoch": 0.05313322305082344,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012021222122212221,
      "loss": 3.2274,
      "step": 399000
    },
    {
      "epoch": 0.053151866286981625,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00012021022102210222,
      "loss": 3.2783,
      "step": 399010
    },
    {
      "epoch": 0.05317050952313981,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001202082208220822,
      "loss": 3.2359,
      "step": 399020
    },
    {
      "epoch": 0.053189152759297996,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00012020622062206221,
      "loss": 3.1777,
      "step": 399030
    },
    {
      "epoch": 0.05320779599545618,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012020422042204221,
      "loss": 3.2364,
      "step": 399040
    },
    {
      "epoch": 0.05322643923161436,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012020222022202222,
      "loss": 3.1527,
      "step": 399050
    },
    {
      "epoch": 0.05324508246777254,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001202002200220022,
      "loss": 3.143,
      "step": 399060
    },
    {
      "epoch": 0.05326372570393073,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012019821982198221,
      "loss": 3.1629,
      "step": 399070
    },
    {
      "epoch": 0.05328236894008891,
      "grad_norm": 0.875,
      "learning_rate": 0.00012019621962196219,
      "loss": 3.1603,
      "step": 399080
    },
    {
      "epoch": 0.053301012176247095,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001201942194219422,
      "loss": 3.2516,
      "step": 399090
    },
    {
      "epoch": 0.05331965541240528,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012019221922192221,
      "loss": 3.1819,
      "step": 399100
    },
    {
      "epoch": 0.053338298648563466,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012019021902190219,
      "loss": 3.2499,
      "step": 399110
    },
    {
      "epoch": 0.05335694188472165,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001201882188218822,
      "loss": 3.2062,
      "step": 399120
    },
    {
      "epoch": 0.05337558512087983,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001201862186218622,
      "loss": 3.145,
      "step": 399130
    },
    {
      "epoch": 0.05339422835703801,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001201842184218422,
      "loss": 3.3149,
      "step": 399140
    },
    {
      "epoch": 0.053412871593196194,
      "grad_norm": 0.75,
      "learning_rate": 0.00012018221822182219,
      "loss": 3.0653,
      "step": 399150
    },
    {
      "epoch": 0.05343151482935438,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001201802180218022,
      "loss": 3.205,
      "step": 399160
    },
    {
      "epoch": 0.053450158065512565,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012017821782178218,
      "loss": 3.1647,
      "step": 399170
    },
    {
      "epoch": 0.05346880130167075,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012017621762176219,
      "loss": 3.2163,
      "step": 399180
    },
    {
      "epoch": 0.05348744453782893,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012017421742174217,
      "loss": 3.1721,
      "step": 399190
    },
    {
      "epoch": 0.05350608777398712,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012017221722172218,
      "loss": 3.1086,
      "step": 399200
    },
    {
      "epoch": 0.0535247310101453,
      "grad_norm": 0.828125,
      "learning_rate": 0.00012017021702170217,
      "loss": 3.3074,
      "step": 399210
    },
    {
      "epoch": 0.05354337424630348,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012016821682168218,
      "loss": 3.3096,
      "step": 399220
    },
    {
      "epoch": 0.053562017482461664,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012016621662166216,
      "loss": 3.0581,
      "step": 399230
    },
    {
      "epoch": 0.05358066071861985,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012016421642164217,
      "loss": 3.2939,
      "step": 399240
    },
    {
      "epoch": 0.053599303954778035,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00012016221622162216,
      "loss": 3.2289,
      "step": 399250
    },
    {
      "epoch": 0.05361794719093622,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012016021602160216,
      "loss": 3.1796,
      "step": 399260
    },
    {
      "epoch": 0.0536365904270944,
      "grad_norm": 0.625,
      "learning_rate": 0.00012015821582158216,
      "loss": 3.1677,
      "step": 399270
    },
    {
      "epoch": 0.05365523366325259,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00012015621562156216,
      "loss": 3.1586,
      "step": 399280
    },
    {
      "epoch": 0.05367387689941077,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012015421542154215,
      "loss": 3.1468,
      "step": 399290
    },
    {
      "epoch": 0.05369252013556895,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012015221522152216,
      "loss": 2.9478,
      "step": 399300
    },
    {
      "epoch": 0.053711163371727134,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012015021502150217,
      "loss": 3.2116,
      "step": 399310
    },
    {
      "epoch": 0.053729806607885316,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012014821482148215,
      "loss": 3.1789,
      "step": 399320
    },
    {
      "epoch": 0.053748449844043505,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012014621462146216,
      "loss": 3.0181,
      "step": 399330
    },
    {
      "epoch": 0.05376709308020169,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012014421442144214,
      "loss": 3.1353,
      "step": 399340
    },
    {
      "epoch": 0.05378573631635987,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012014221422142215,
      "loss": 3.3275,
      "step": 399350
    },
    {
      "epoch": 0.05380437955251805,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012014021402140215,
      "loss": 3.2108,
      "step": 399360
    },
    {
      "epoch": 0.05382302278867624,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012013821382138216,
      "loss": 3.1748,
      "step": 399370
    },
    {
      "epoch": 0.05384166602483442,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012013621362136214,
      "loss": 3.2723,
      "step": 399380
    },
    {
      "epoch": 0.053860309260992605,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012013421342134215,
      "loss": 3.1899,
      "step": 399390
    },
    {
      "epoch": 0.05387895249715079,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012013221322132213,
      "loss": 3.0247,
      "step": 399400
    },
    {
      "epoch": 0.053897595733308976,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012013021302130214,
      "loss": 3.2461,
      "step": 399410
    },
    {
      "epoch": 0.05391623896946716,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012012821282128212,
      "loss": 3.1815,
      "step": 399420
    },
    {
      "epoch": 0.05393488220562534,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012012621262126213,
      "loss": 3.1038,
      "step": 399430
    },
    {
      "epoch": 0.05395352544178352,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012012421242124213,
      "loss": 3.2337,
      "step": 399440
    },
    {
      "epoch": 0.05397216867794171,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012012221222122214,
      "loss": 3.2644,
      "step": 399450
    },
    {
      "epoch": 0.05399081191409989,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012012021202120212,
      "loss": 3.1482,
      "step": 399460
    },
    {
      "epoch": 0.054009455150258075,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012011821182118213,
      "loss": 3.2243,
      "step": 399470
    },
    {
      "epoch": 0.05402809838641626,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012011621162116211,
      "loss": 3.1874,
      "step": 399480
    },
    {
      "epoch": 0.05404674162257444,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012011421142114212,
      "loss": 3.2895,
      "step": 399490
    },
    {
      "epoch": 0.05406538485873263,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012011221122112213,
      "loss": 3.3327,
      "step": 399500
    },
    {
      "epoch": 0.05408402809489081,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012011021102110211,
      "loss": 3.3124,
      "step": 399510
    },
    {
      "epoch": 0.05410267133104899,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012010821082108212,
      "loss": 3.306,
      "step": 399520
    },
    {
      "epoch": 0.054121314567207174,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012010621062106211,
      "loss": 3.2203,
      "step": 399530
    },
    {
      "epoch": 0.05413995780336536,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012010421042104212,
      "loss": 3.2737,
      "step": 399540
    },
    {
      "epoch": 0.054158601039523545,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001201022102210221,
      "loss": 3.1959,
      "step": 399550
    },
    {
      "epoch": 0.05417724427568173,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012010021002100211,
      "loss": 3.2331,
      "step": 399560
    },
    {
      "epoch": 0.05419588751183991,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001200982098209821,
      "loss": 3.3603,
      "step": 399570
    },
    {
      "epoch": 0.0542145307479981,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001200962096209621,
      "loss": 3.1228,
      "step": 399580
    },
    {
      "epoch": 0.05423317398415628,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001200942094209421,
      "loss": 3.2557,
      "step": 399590
    },
    {
      "epoch": 0.05425181722031446,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001200922092209221,
      "loss": 3.2137,
      "step": 399600
    },
    {
      "epoch": 0.054270460456472644,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012009020902090209,
      "loss": 3.0502,
      "step": 399610
    },
    {
      "epoch": 0.05428910369263083,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001200882088208821,
      "loss": 3.2498,
      "step": 399620
    },
    {
      "epoch": 0.054307746928789015,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012008620862086208,
      "loss": 3.2953,
      "step": 399630
    },
    {
      "epoch": 0.0543263901649472,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00012008420842084209,
      "loss": 3.2869,
      "step": 399640
    },
    {
      "epoch": 0.05434503340110538,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012008220822082207,
      "loss": 3.0982,
      "step": 399650
    },
    {
      "epoch": 0.05436367663726356,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012008020802080208,
      "loss": 3.1663,
      "step": 399660
    },
    {
      "epoch": 0.05438231987342175,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012007820782078208,
      "loss": 3.1828,
      "step": 399670
    },
    {
      "epoch": 0.05440096310957993,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00012007620762076209,
      "loss": 3.1208,
      "step": 399680
    },
    {
      "epoch": 0.054419606345738114,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001200742074207421,
      "loss": 3.2243,
      "step": 399690
    },
    {
      "epoch": 0.054438249581896296,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00012007220722072208,
      "loss": 3.0935,
      "step": 399700
    },
    {
      "epoch": 0.054456892818054485,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012007020702070209,
      "loss": 3.1536,
      "step": 399710
    },
    {
      "epoch": 0.05447553605421267,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012006820682068207,
      "loss": 3.2076,
      "step": 399720
    },
    {
      "epoch": 0.05449417929037085,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00012006620662066208,
      "loss": 3.1728,
      "step": 399730
    },
    {
      "epoch": 0.05451282252652903,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00012006420642064206,
      "loss": 3.1557,
      "step": 399740
    },
    {
      "epoch": 0.05453146576268722,
      "grad_norm": 0.875,
      "learning_rate": 0.00012006220622062207,
      "loss": 3.1798,
      "step": 399750
    },
    {
      "epoch": 0.0545501089988454,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012006020602060207,
      "loss": 3.2095,
      "step": 399760
    },
    {
      "epoch": 0.054568752235003584,
      "grad_norm": 0.75,
      "learning_rate": 0.00012005820582058208,
      "loss": 3.1185,
      "step": 399770
    },
    {
      "epoch": 0.054587395471161766,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00012005620562056206,
      "loss": 3.2235,
      "step": 399780
    },
    {
      "epoch": 0.054606038707319955,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012005420542054207,
      "loss": 3.1142,
      "step": 399790
    },
    {
      "epoch": 0.05462468194347814,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012005220522052205,
      "loss": 3.2252,
      "step": 399800
    },
    {
      "epoch": 0.05464332517963632,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012005020502050206,
      "loss": 3.3384,
      "step": 399810
    },
    {
      "epoch": 0.0546619684157945,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012004820482048204,
      "loss": 3.1738,
      "step": 399820
    },
    {
      "epoch": 0.05468061165195268,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012004620462046205,
      "loss": 3.2242,
      "step": 399830
    },
    {
      "epoch": 0.05469925488811087,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012004420442044204,
      "loss": 3.2444,
      "step": 399840
    },
    {
      "epoch": 0.054717898124269054,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00012004220422042205,
      "loss": 3.2655,
      "step": 399850
    },
    {
      "epoch": 0.054736541360427236,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012004020402040204,
      "loss": 3.1217,
      "step": 399860
    },
    {
      "epoch": 0.05475518459658542,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012003820382038205,
      "loss": 3.2816,
      "step": 399870
    },
    {
      "epoch": 0.05477382783274361,
      "grad_norm": 0.78125,
      "learning_rate": 0.00012003620362036203,
      "loss": 3.2012,
      "step": 399880
    },
    {
      "epoch": 0.05479247106890179,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012003420342034204,
      "loss": 3.1985,
      "step": 399890
    },
    {
      "epoch": 0.05481111430505997,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012003220322032205,
      "loss": 3.0757,
      "step": 399900
    },
    {
      "epoch": 0.05482975754121815,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012003020302030204,
      "loss": 3.2157,
      "step": 399910
    },
    {
      "epoch": 0.05484840077737634,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012002820282028204,
      "loss": 3.1022,
      "step": 399920
    },
    {
      "epoch": 0.054867044013534524,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012002620262026203,
      "loss": 3.2059,
      "step": 399930
    },
    {
      "epoch": 0.054885687249692706,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012002420242024204,
      "loss": 3.1845,
      "step": 399940
    },
    {
      "epoch": 0.05490433048585089,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012002220222022202,
      "loss": 3.0892,
      "step": 399950
    },
    {
      "epoch": 0.05492297372200908,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00012002020202020203,
      "loss": 3.1833,
      "step": 399960
    },
    {
      "epoch": 0.05494161695816726,
      "grad_norm": 0.796875,
      "learning_rate": 0.00012001820182018201,
      "loss": 3.2518,
      "step": 399970
    },
    {
      "epoch": 0.05496026019432544,
      "grad_norm": 0.765625,
      "learning_rate": 0.00012001620162016202,
      "loss": 3.2694,
      "step": 399980
    },
    {
      "epoch": 0.054978903430483624,
      "grad_norm": 0.75,
      "learning_rate": 0.00012001420142014202,
      "loss": 3.1954,
      "step": 399990
    },
    {
      "epoch": 0.054997546666641806,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012001220122012203,
      "loss": 3.2071,
      "step": 400000
    },
    {
      "epoch": 0.055016189902799995,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012001020102010201,
      "loss": 3.256,
      "step": 400010
    },
    {
      "epoch": 0.05503483313895818,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012000820082008202,
      "loss": 3.2357,
      "step": 400020
    },
    {
      "epoch": 0.05505347637511636,
      "grad_norm": 0.84375,
      "learning_rate": 0.000120006200620062,
      "loss": 3.1429,
      "step": 400030
    },
    {
      "epoch": 0.05507211961127454,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00012000420042004201,
      "loss": 3.187,
      "step": 400040
    },
    {
      "epoch": 0.05509076284743273,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012000220022002199,
      "loss": 3.1418,
      "step": 400050
    },
    {
      "epoch": 0.05510940608359091,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000120000200020002,
      "loss": 3.2862,
      "step": 400060
    },
    {
      "epoch": 0.055128049319749094,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000119998199819982,
      "loss": 3.2013,
      "step": 400070
    },
    {
      "epoch": 0.055146692555907276,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011999619961996201,
      "loss": 3.2272,
      "step": 400080
    },
    {
      "epoch": 0.055165335792065465,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011999419941994202,
      "loss": 3.1322,
      "step": 400090
    },
    {
      "epoch": 0.05518397902822365,
      "grad_norm": 0.6875,
      "learning_rate": 0.000119992199219922,
      "loss": 3.0919,
      "step": 400100
    },
    {
      "epoch": 0.05520262226438183,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011999019901990201,
      "loss": 3.1474,
      "step": 400110
    },
    {
      "epoch": 0.05522126550054001,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011998819881988199,
      "loss": 3.1174,
      "step": 400120
    },
    {
      "epoch": 0.0552399087366982,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000119986198619862,
      "loss": 3.1618,
      "step": 400130
    },
    {
      "epoch": 0.05525855197285638,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011998419841984198,
      "loss": 3.197,
      "step": 400140
    },
    {
      "epoch": 0.055277195209014564,
      "grad_norm": 0.75,
      "learning_rate": 0.00011998219821982199,
      "loss": 3.1923,
      "step": 400150
    },
    {
      "epoch": 0.055295838445172746,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011998019801980199,
      "loss": 3.197,
      "step": 400160
    },
    {
      "epoch": 0.05531448168133093,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000119978197819782,
      "loss": 3.2331,
      "step": 400170
    },
    {
      "epoch": 0.05533312491748912,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011997619761976198,
      "loss": 3.1131,
      "step": 400180
    },
    {
      "epoch": 0.0553517681536473,
      "grad_norm": 0.75,
      "learning_rate": 0.00011997419741974199,
      "loss": 3.2183,
      "step": 400190
    },
    {
      "epoch": 0.05537041138980548,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011997219721972197,
      "loss": 3.0981,
      "step": 400200
    },
    {
      "epoch": 0.05538905462596366,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011997019701970198,
      "loss": 3.2076,
      "step": 400210
    },
    {
      "epoch": 0.05540769786212185,
      "grad_norm": 0.75,
      "learning_rate": 0.00011996819681968197,
      "loss": 3.2373,
      "step": 400220
    },
    {
      "epoch": 0.055426341098280034,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011996619661966197,
      "loss": 3.2018,
      "step": 400230
    },
    {
      "epoch": 0.055444984334438216,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011996419641964196,
      "loss": 3.0612,
      "step": 400240
    },
    {
      "epoch": 0.0554636275705964,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011996219621962197,
      "loss": 2.9916,
      "step": 400250
    },
    {
      "epoch": 0.05548227080675459,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011996019601960195,
      "loss": 3.2468,
      "step": 400260
    },
    {
      "epoch": 0.05550091404291277,
      "grad_norm": 0.75,
      "learning_rate": 0.00011995819581958196,
      "loss": 3.2321,
      "step": 400270
    },
    {
      "epoch": 0.05551955727907095,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011995619561956197,
      "loss": 3.1885,
      "step": 400280
    },
    {
      "epoch": 0.05553820051522913,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011995419541954196,
      "loss": 3.1229,
      "step": 400290
    },
    {
      "epoch": 0.05555684375138732,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011995219521952196,
      "loss": 3.2105,
      "step": 400300
    },
    {
      "epoch": 0.055575486987545504,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011995019501950196,
      "loss": 3.0943,
      "step": 400310
    },
    {
      "epoch": 0.055594130223703686,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011994819481948197,
      "loss": 3.2167,
      "step": 400320
    },
    {
      "epoch": 0.05561277345986187,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011994619461946195,
      "loss": 3.1938,
      "step": 400330
    },
    {
      "epoch": 0.05563141669602005,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011994419441944196,
      "loss": 3.3195,
      "step": 400340
    },
    {
      "epoch": 0.05565005993217824,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011994219421942194,
      "loss": 3.2155,
      "step": 400350
    },
    {
      "epoch": 0.05566870316833642,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011994019401940195,
      "loss": 3.1786,
      "step": 400360
    },
    {
      "epoch": 0.0556873464044946,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011993819381938193,
      "loss": 3.3255,
      "step": 400370
    },
    {
      "epoch": 0.055705989640652785,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011993619361936194,
      "loss": 3.2446,
      "step": 400380
    },
    {
      "epoch": 0.055724632876810974,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011993419341934194,
      "loss": 3.2577,
      "step": 400390
    },
    {
      "epoch": 0.055743276112969156,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011993219321932195,
      "loss": 3.1466,
      "step": 400400
    },
    {
      "epoch": 0.05576191934912734,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011993019301930193,
      "loss": 3.2233,
      "step": 400410
    },
    {
      "epoch": 0.05578056258528552,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011992819281928194,
      "loss": 3.1915,
      "step": 400420
    },
    {
      "epoch": 0.05579920582144371,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011992619261926192,
      "loss": 3.165,
      "step": 400430
    },
    {
      "epoch": 0.05581784905760189,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011992419241924193,
      "loss": 3.2303,
      "step": 400440
    },
    {
      "epoch": 0.05583649229376007,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011992219221922191,
      "loss": 3.2655,
      "step": 400450
    },
    {
      "epoch": 0.055855135529918255,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011992019201920192,
      "loss": 3.2324,
      "step": 400460
    },
    {
      "epoch": 0.055873778766076444,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011991819181918192,
      "loss": 3.2893,
      "step": 400470
    },
    {
      "epoch": 0.055892422002234626,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011991619161916193,
      "loss": 3.172,
      "step": 400480
    },
    {
      "epoch": 0.05591106523839281,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011991419141914193,
      "loss": 3.248,
      "step": 400490
    },
    {
      "epoch": 0.05592970847455099,
      "grad_norm": 0.75,
      "learning_rate": 0.00011991219121912192,
      "loss": 3.0796,
      "step": 400500
    },
    {
      "epoch": 0.05594835171070918,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011991019101910193,
      "loss": 3.133,
      "step": 400510
    },
    {
      "epoch": 0.05596699494686736,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011990819081908191,
      "loss": 3.298,
      "step": 400520
    },
    {
      "epoch": 0.05598563818302554,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011990619061906192,
      "loss": 3.2138,
      "step": 400530
    },
    {
      "epoch": 0.056004281419183725,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011990419041904191,
      "loss": 3.1082,
      "step": 400540
    },
    {
      "epoch": 0.05602292465534191,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011990219021902191,
      "loss": 3.1591,
      "step": 400550
    },
    {
      "epoch": 0.056041567891500096,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001199001900190019,
      "loss": 3.1813,
      "step": 400560
    },
    {
      "epoch": 0.05606021112765828,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011989818981898191,
      "loss": 3.2117,
      "step": 400570
    },
    {
      "epoch": 0.05607885436381646,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001198961896189619,
      "loss": 3.3411,
      "step": 400580
    },
    {
      "epoch": 0.05609749759997464,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001198941894189419,
      "loss": 3.2002,
      "step": 400590
    },
    {
      "epoch": 0.05611614083613283,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011989218921892189,
      "loss": 3.1811,
      "step": 400600
    },
    {
      "epoch": 0.056134784072291014,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001198901890189019,
      "loss": 3.1818,
      "step": 400610
    },
    {
      "epoch": 0.056153427308449196,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011988818881888189,
      "loss": 3.1366,
      "step": 400620
    },
    {
      "epoch": 0.05617207054460738,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001198861886188619,
      "loss": 3.3324,
      "step": 400630
    },
    {
      "epoch": 0.05619071378076557,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011988418841884188,
      "loss": 3.1639,
      "step": 400640
    },
    {
      "epoch": 0.05620935701692375,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011988218821882189,
      "loss": 3.328,
      "step": 400650
    },
    {
      "epoch": 0.05622800025308193,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011988018801880187,
      "loss": 3.271,
      "step": 400660
    },
    {
      "epoch": 0.05624664348924011,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011987818781878188,
      "loss": 3.0576,
      "step": 400670
    },
    {
      "epoch": 0.0562652867253983,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011987618761876189,
      "loss": 3.1588,
      "step": 400680
    },
    {
      "epoch": 0.056283929961556484,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011987418741874187,
      "loss": 3.1311,
      "step": 400690
    },
    {
      "epoch": 0.056302573197714666,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011987218721872188,
      "loss": 2.9566,
      "step": 400700
    },
    {
      "epoch": 0.05632121643387285,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011987018701870188,
      "loss": 3.2034,
      "step": 400710
    },
    {
      "epoch": 0.05633985967003103,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011986818681868189,
      "loss": 3.0601,
      "step": 400720
    },
    {
      "epoch": 0.05635850290618922,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011986618661866187,
      "loss": 3.2155,
      "step": 400730
    },
    {
      "epoch": 0.0563771461423474,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011986418641864188,
      "loss": 3.2348,
      "step": 400740
    },
    {
      "epoch": 0.05639578937850558,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011986218621862186,
      "loss": 3.2263,
      "step": 400750
    },
    {
      "epoch": 0.056414432614663765,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011986018601860187,
      "loss": 3.1998,
      "step": 400760
    },
    {
      "epoch": 0.056433075850821954,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011985818581858185,
      "loss": 3.1899,
      "step": 400770
    },
    {
      "epoch": 0.056451719086980136,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011985618561856186,
      "loss": 3.0349,
      "step": 400780
    },
    {
      "epoch": 0.05647036232313832,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011985418541854186,
      "loss": 3.0977,
      "step": 400790
    },
    {
      "epoch": 0.0564890055592965,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011985218521852187,
      "loss": 3.1752,
      "step": 400800
    },
    {
      "epoch": 0.05650764879545469,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011985018501850185,
      "loss": 3.1796,
      "step": 400810
    },
    {
      "epoch": 0.05652629203161287,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011984818481848186,
      "loss": 3.042,
      "step": 400820
    },
    {
      "epoch": 0.05654493526777105,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011984618461846184,
      "loss": 3.2336,
      "step": 400830
    },
    {
      "epoch": 0.056563578503929235,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011984418441844185,
      "loss": 3.1972,
      "step": 400840
    },
    {
      "epoch": 0.056582221740087424,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011984218421842184,
      "loss": 3.1971,
      "step": 400850
    },
    {
      "epoch": 0.056600864976245606,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011984018401840185,
      "loss": 3.1966,
      "step": 400860
    },
    {
      "epoch": 0.05661950821240379,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011983818381838183,
      "loss": 3.2573,
      "step": 400870
    },
    {
      "epoch": 0.05663815144856197,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011983618361836184,
      "loss": 3.1243,
      "step": 400880
    },
    {
      "epoch": 0.05665679468472015,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011983418341834185,
      "loss": 3.1981,
      "step": 400890
    },
    {
      "epoch": 0.05667543792087834,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011983218321832184,
      "loss": 3.1585,
      "step": 400900
    },
    {
      "epoch": 0.05669408115703652,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011983018301830184,
      "loss": 3.2095,
      "step": 400910
    },
    {
      "epoch": 0.056712724393194705,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011982818281828183,
      "loss": 3.2269,
      "step": 400920
    },
    {
      "epoch": 0.05673136762935289,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011982618261826184,
      "loss": 3.3302,
      "step": 400930
    },
    {
      "epoch": 0.056750010865511076,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011982418241824183,
      "loss": 3.1208,
      "step": 400940
    },
    {
      "epoch": 0.05676865410166926,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011982218221822184,
      "loss": 3.2219,
      "step": 400950
    },
    {
      "epoch": 0.05678729733782744,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011982018201820182,
      "loss": 3.2328,
      "step": 400960
    },
    {
      "epoch": 0.05680594057398562,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011981818181818183,
      "loss": 3.1507,
      "step": 400970
    },
    {
      "epoch": 0.05682458381014381,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011981618161816181,
      "loss": 3.1104,
      "step": 400980
    },
    {
      "epoch": 0.05684322704630199,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011981418141814182,
      "loss": 3.1578,
      "step": 400990
    },
    {
      "epoch": 0.056861870282460175,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001198121812181218,
      "loss": 3.2291,
      "step": 401000
    },
    {
      "epoch": 0.05688051351861836,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011981018101810181,
      "loss": 3.0961,
      "step": 401010
    },
    {
      "epoch": 0.056899156754776546,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011980818081808181,
      "loss": 3.2517,
      "step": 401020
    },
    {
      "epoch": 0.05691779999093473,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011980618061806182,
      "loss": 3.1734,
      "step": 401030
    },
    {
      "epoch": 0.05693644322709291,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001198041804180418,
      "loss": 3.2378,
      "step": 401040
    },
    {
      "epoch": 0.05695508646325109,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011980218021802181,
      "loss": 3.2297,
      "step": 401050
    },
    {
      "epoch": 0.056973729699409274,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011980018001800179,
      "loss": 3.2063,
      "step": 401060
    },
    {
      "epoch": 0.05699237293556746,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001197981798179818,
      "loss": 3.1555,
      "step": 401070
    },
    {
      "epoch": 0.057011016171725645,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011979617961796181,
      "loss": 3.0749,
      "step": 401080
    },
    {
      "epoch": 0.05702965940788383,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011979417941794179,
      "loss": 3.2439,
      "step": 401090
    },
    {
      "epoch": 0.05704830264404201,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001197921792179218,
      "loss": 3.1315,
      "step": 401100
    },
    {
      "epoch": 0.0570669458802002,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001197901790179018,
      "loss": 3.2019,
      "step": 401110
    },
    {
      "epoch": 0.05708558911635838,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001197881788178818,
      "loss": 3.0788,
      "step": 401120
    },
    {
      "epoch": 0.05710423235251656,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011978617861786179,
      "loss": 3.3593,
      "step": 401130
    },
    {
      "epoch": 0.057122875588674744,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001197841784178418,
      "loss": 3.1758,
      "step": 401140
    },
    {
      "epoch": 0.05714151882483293,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011978217821782178,
      "loss": 3.1528,
      "step": 401150
    },
    {
      "epoch": 0.057160162060991115,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011978017801780179,
      "loss": 3.1033,
      "step": 401160
    },
    {
      "epoch": 0.0571788052971493,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011977817781778178,
      "loss": 3.0837,
      "step": 401170
    },
    {
      "epoch": 0.05719744853330748,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001197761776177618,
      "loss": 3.2355,
      "step": 401180
    },
    {
      "epoch": 0.05721609176946567,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011977417741774178,
      "loss": 3.2058,
      "step": 401190
    },
    {
      "epoch": 0.05723473500562385,
      "grad_norm": 0.75,
      "learning_rate": 0.00011977217721772178,
      "loss": 3.1347,
      "step": 401200
    },
    {
      "epoch": 0.05725337824178203,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011977017701770177,
      "loss": 3.1482,
      "step": 401210
    },
    {
      "epoch": 0.057272021477940215,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011976817681768178,
      "loss": 2.9865,
      "step": 401220
    },
    {
      "epoch": 0.0572906647140984,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011976617661766176,
      "loss": 3.1321,
      "step": 401230
    },
    {
      "epoch": 0.057309307950256586,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011976417641764177,
      "loss": 3.13,
      "step": 401240
    },
    {
      "epoch": 0.05732795118641477,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011976217621762176,
      "loss": 3.1791,
      "step": 401250
    },
    {
      "epoch": 0.05734659442257295,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011976017601760177,
      "loss": 3.2581,
      "step": 401260
    },
    {
      "epoch": 0.05736523765873113,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011975817581758178,
      "loss": 3.2145,
      "step": 401270
    },
    {
      "epoch": 0.05738388089488932,
      "grad_norm": 0.75,
      "learning_rate": 0.00011975617561756176,
      "loss": 3.1493,
      "step": 401280
    },
    {
      "epoch": 0.0574025241310475,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011975417541754177,
      "loss": 3.0019,
      "step": 401290
    },
    {
      "epoch": 0.057421167367205685,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011975217521752175,
      "loss": 3.1151,
      "step": 401300
    },
    {
      "epoch": 0.05743981060336387,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011975017501750176,
      "loss": 3.1979,
      "step": 401310
    },
    {
      "epoch": 0.057458453839522056,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011974817481748175,
      "loss": 3.1835,
      "step": 401320
    },
    {
      "epoch": 0.05747709707568024,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011974617461746175,
      "loss": 3.1003,
      "step": 401330
    },
    {
      "epoch": 0.05749574031183842,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011974417441744175,
      "loss": 3.0495,
      "step": 401340
    },
    {
      "epoch": 0.0575143835479966,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011974217421742176,
      "loss": 3.3201,
      "step": 401350
    },
    {
      "epoch": 0.05753302678415479,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011974017401740174,
      "loss": 3.1742,
      "step": 401360
    },
    {
      "epoch": 0.05755167002031297,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011973817381738175,
      "loss": 3.3776,
      "step": 401370
    },
    {
      "epoch": 0.057570313256471155,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011973617361736173,
      "loss": 3.1317,
      "step": 401380
    },
    {
      "epoch": 0.05758895649262934,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011973417341734174,
      "loss": 3.0664,
      "step": 401390
    },
    {
      "epoch": 0.05760759972878752,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011973217321732174,
      "loss": 3.1569,
      "step": 401400
    },
    {
      "epoch": 0.05762624296494571,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011973017301730173,
      "loss": 3.0941,
      "step": 401410
    },
    {
      "epoch": 0.05764488620110389,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011972817281728173,
      "loss": 3.1944,
      "step": 401420
    },
    {
      "epoch": 0.05766352943726207,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011972617261726174,
      "loss": 3.1713,
      "step": 401430
    },
    {
      "epoch": 0.057682172673420254,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011972417241724172,
      "loss": 3.126,
      "step": 401440
    },
    {
      "epoch": 0.05770081590957844,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011972217221722173,
      "loss": 3.265,
      "step": 401450
    },
    {
      "epoch": 0.057719459145736625,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011972017201720171,
      "loss": 3.2181,
      "step": 401460
    },
    {
      "epoch": 0.05773810238189481,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011971817181718172,
      "loss": 3.1822,
      "step": 401470
    },
    {
      "epoch": 0.05775674561805299,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011971617161716173,
      "loss": 3.1734,
      "step": 401480
    },
    {
      "epoch": 0.05777538885421118,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011971417141714172,
      "loss": 3.0457,
      "step": 401490
    },
    {
      "epoch": 0.05779403209036936,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011971217121712173,
      "loss": 3.1198,
      "step": 401500
    },
    {
      "epoch": 0.05781267532652754,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011971017101710172,
      "loss": 3.1485,
      "step": 401510
    },
    {
      "epoch": 0.057831318562685724,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011970817081708172,
      "loss": 3.1518,
      "step": 401520
    },
    {
      "epoch": 0.05784996179884391,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011970617061706171,
      "loss": 3.1323,
      "step": 401530
    },
    {
      "epoch": 0.057868605035002095,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011970417041704172,
      "loss": 3.124,
      "step": 401540
    },
    {
      "epoch": 0.05788724827116028,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001197021702170217,
      "loss": 3.2443,
      "step": 401550
    },
    {
      "epoch": 0.05790589150731846,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011970017001700171,
      "loss": 3.137,
      "step": 401560
    },
    {
      "epoch": 0.05792453474347664,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001196981698169817,
      "loss": 3.1273,
      "step": 401570
    },
    {
      "epoch": 0.05794317797963483,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011969616961696171,
      "loss": 3.2539,
      "step": 401580
    },
    {
      "epoch": 0.05796182121579301,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001196941694169417,
      "loss": 3.2175,
      "step": 401590
    },
    {
      "epoch": 0.057980464451951194,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001196921692169217,
      "loss": 3.1609,
      "step": 401600
    },
    {
      "epoch": 0.057999107688109376,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011969016901690169,
      "loss": 3.1694,
      "step": 401610
    },
    {
      "epoch": 0.058017750924267565,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001196881688168817,
      "loss": 3.0493,
      "step": 401620
    },
    {
      "epoch": 0.05803639416042575,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011968616861686168,
      "loss": 3.283,
      "step": 401630
    },
    {
      "epoch": 0.05805503739658393,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011968416841684169,
      "loss": 3.2327,
      "step": 401640
    },
    {
      "epoch": 0.05807368063274211,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011968216821682168,
      "loss": 3.1179,
      "step": 401650
    },
    {
      "epoch": 0.0580923238689003,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011968016801680169,
      "loss": 3.1085,
      "step": 401660
    },
    {
      "epoch": 0.05811096710505848,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001196781678167817,
      "loss": 3.0689,
      "step": 401670
    },
    {
      "epoch": 0.058129610341216664,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011967616761676168,
      "loss": 3.2587,
      "step": 401680
    },
    {
      "epoch": 0.058148253577374846,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011967416741674169,
      "loss": 3.1276,
      "step": 401690
    },
    {
      "epoch": 0.058166896813533035,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011967216721672167,
      "loss": 3.2528,
      "step": 401700
    },
    {
      "epoch": 0.05818554004969122,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011967016701670168,
      "loss": 3.1137,
      "step": 401710
    },
    {
      "epoch": 0.0582041832858494,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011966816681668166,
      "loss": 3.1018,
      "step": 401720
    },
    {
      "epoch": 0.05822282652200758,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00011966616661666167,
      "loss": 3.2636,
      "step": 401730
    },
    {
      "epoch": 0.05824146975816576,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011966416641664167,
      "loss": 3.2189,
      "step": 401740
    },
    {
      "epoch": 0.05826011299432395,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011966216621662168,
      "loss": 3.066,
      "step": 401750
    },
    {
      "epoch": 0.058278756230482134,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011966016601660166,
      "loss": 3.144,
      "step": 401760
    },
    {
      "epoch": 0.058297399466640316,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011965816581658167,
      "loss": 3.304,
      "step": 401770
    },
    {
      "epoch": 0.0583160427027985,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011965616561656165,
      "loss": 3.2469,
      "step": 401780
    },
    {
      "epoch": 0.05833468593895669,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011965416541654166,
      "loss": 3.1231,
      "step": 401790
    },
    {
      "epoch": 0.05835332917511487,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011965216521652166,
      "loss": 3.1232,
      "step": 401800
    },
    {
      "epoch": 0.05837197241127305,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011965016501650166,
      "loss": 3.1446,
      "step": 401810
    },
    {
      "epoch": 0.058390615647431234,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011964816481648165,
      "loss": 3.2123,
      "step": 401820
    },
    {
      "epoch": 0.05840925888358942,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011964616461646166,
      "loss": 3.0512,
      "step": 401830
    },
    {
      "epoch": 0.058427902119747605,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011964416441644164,
      "loss": 3.1319,
      "step": 401840
    },
    {
      "epoch": 0.05844654535590579,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011964216421642165,
      "loss": 3.127,
      "step": 401850
    },
    {
      "epoch": 0.05846518859206397,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011964016401640166,
      "loss": 3.0981,
      "step": 401860
    },
    {
      "epoch": 0.05848383182822216,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011963816381638164,
      "loss": 3.1975,
      "step": 401870
    },
    {
      "epoch": 0.05850247506438034,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011963616361636165,
      "loss": 3.1531,
      "step": 401880
    },
    {
      "epoch": 0.05852111830053852,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011963416341634164,
      "loss": 3.1091,
      "step": 401890
    },
    {
      "epoch": 0.058539761536696704,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011963216321632165,
      "loss": 3.1567,
      "step": 401900
    },
    {
      "epoch": 0.058558404772854886,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011963016301630163,
      "loss": 3.2441,
      "step": 401910
    },
    {
      "epoch": 0.058577048009013075,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011962816281628164,
      "loss": 3.1694,
      "step": 401920
    },
    {
      "epoch": 0.05859569124517126,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011962616261626163,
      "loss": 3.1856,
      "step": 401930
    },
    {
      "epoch": 0.05861433448132944,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011962416241624163,
      "loss": 3.1179,
      "step": 401940
    },
    {
      "epoch": 0.05863297771748762,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011962216221622162,
      "loss": 3.1211,
      "step": 401950
    },
    {
      "epoch": 0.05865162095364581,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011962016201620163,
      "loss": 3.1565,
      "step": 401960
    },
    {
      "epoch": 0.05867026418980399,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011961816181618162,
      "loss": 3.1467,
      "step": 401970
    },
    {
      "epoch": 0.058688907425962174,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011961616161616163,
      "loss": 3.1543,
      "step": 401980
    },
    {
      "epoch": 0.058707550662120356,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011961416141614161,
      "loss": 3.1398,
      "step": 401990
    },
    {
      "epoch": 0.058726193898278545,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011961216121612162,
      "loss": 3.1706,
      "step": 402000
    },
    {
      "epoch": 0.05874483713443673,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001196101610161016,
      "loss": 3.0685,
      "step": 402010
    },
    {
      "epoch": 0.05876348037059491,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011960816081608161,
      "loss": 3.1642,
      "step": 402020
    },
    {
      "epoch": 0.05878212360675309,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00011960616061606161,
      "loss": 3.2323,
      "step": 402030
    },
    {
      "epoch": 0.05880076684291128,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001196041604160416,
      "loss": 3.137,
      "step": 402040
    },
    {
      "epoch": 0.05881941007906946,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001196021602160216,
      "loss": 3.1839,
      "step": 402050
    },
    {
      "epoch": 0.058838053315227644,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011960016001600161,
      "loss": 3.2475,
      "step": 402060
    },
    {
      "epoch": 0.058856696551385826,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011959815981598162,
      "loss": 3.0716,
      "step": 402070
    },
    {
      "epoch": 0.05887533978754401,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001195961596159616,
      "loss": 3.2723,
      "step": 402080
    },
    {
      "epoch": 0.0588939830237022,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011959415941594161,
      "loss": 3.2352,
      "step": 402090
    },
    {
      "epoch": 0.05891262625986038,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011959215921592159,
      "loss": 3.1166,
      "step": 402100
    },
    {
      "epoch": 0.05893126949601856,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001195901590159016,
      "loss": 3.1936,
      "step": 402110
    },
    {
      "epoch": 0.05894991273217674,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001195881588158816,
      "loss": 3.0657,
      "step": 402120
    },
    {
      "epoch": 0.05896855596833493,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001195861586158616,
      "loss": 3.1123,
      "step": 402130
    },
    {
      "epoch": 0.058987199204493114,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011958415841584159,
      "loss": 3.2236,
      "step": 402140
    },
    {
      "epoch": 0.059005842440651296,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001195821582158216,
      "loss": 3.2139,
      "step": 402150
    },
    {
      "epoch": 0.05902448567680948,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011958015801580158,
      "loss": 3.0531,
      "step": 402160
    },
    {
      "epoch": 0.05904312891296767,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011957815781578159,
      "loss": 3.2549,
      "step": 402170
    },
    {
      "epoch": 0.05906177214912585,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011957615761576157,
      "loss": 3.1683,
      "step": 402180
    },
    {
      "epoch": 0.05908041538528403,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011957415741574158,
      "loss": 3.2423,
      "step": 402190
    },
    {
      "epoch": 0.05909905862144221,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011957215721572157,
      "loss": 3.1473,
      "step": 402200
    },
    {
      "epoch": 0.0591177018576004,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011957015701570158,
      "loss": 3.1402,
      "step": 402210
    },
    {
      "epoch": 0.059136345093758584,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011956815681568157,
      "loss": 3.1984,
      "step": 402220
    },
    {
      "epoch": 0.059154988329916766,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011956615661566157,
      "loss": 3.0955,
      "step": 402230
    },
    {
      "epoch": 0.05917363156607495,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011956415641564156,
      "loss": 3.217,
      "step": 402240
    },
    {
      "epoch": 0.05919227480223313,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011956215621562157,
      "loss": 3.0656,
      "step": 402250
    },
    {
      "epoch": 0.05921091803839132,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011956015601560157,
      "loss": 3.1742,
      "step": 402260
    },
    {
      "epoch": 0.0592295612745495,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011955815581558156,
      "loss": 3.3323,
      "step": 402270
    },
    {
      "epoch": 0.05924820451070768,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011955615561556157,
      "loss": 3.1574,
      "step": 402280
    },
    {
      "epoch": 0.059266847746865865,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011955415541554156,
      "loss": 3.2298,
      "step": 402290
    },
    {
      "epoch": 0.059285490983024054,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011955215521552157,
      "loss": 3.1276,
      "step": 402300
    },
    {
      "epoch": 0.059304134219182236,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011955015501550155,
      "loss": 3.1724,
      "step": 402310
    },
    {
      "epoch": 0.05932277745534042,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011954815481548156,
      "loss": 3.3214,
      "step": 402320
    },
    {
      "epoch": 0.0593414206914986,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011954615461546154,
      "loss": 3.265,
      "step": 402330
    },
    {
      "epoch": 0.05936006392765679,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011954415441544155,
      "loss": 3.1323,
      "step": 402340
    },
    {
      "epoch": 0.05937870716381497,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011954215421542155,
      "loss": 3.1809,
      "step": 402350
    },
    {
      "epoch": 0.05939735039997315,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011954015401540154,
      "loss": 3.1416,
      "step": 402360
    },
    {
      "epoch": 0.059415993636131335,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011953815381538154,
      "loss": 3.1626,
      "step": 402370
    },
    {
      "epoch": 0.059434636872289524,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011953615361536155,
      "loss": 3.2458,
      "step": 402380
    },
    {
      "epoch": 0.059453280108447706,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011953415341534153,
      "loss": 3.0938,
      "step": 402390
    },
    {
      "epoch": 0.05947192334460589,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00011953215321532154,
      "loss": 3.0869,
      "step": 402400
    },
    {
      "epoch": 0.05949056658076407,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011953015301530152,
      "loss": 3.2117,
      "step": 402410
    },
    {
      "epoch": 0.05950920981692225,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011952815281528153,
      "loss": 3.0578,
      "step": 402420
    },
    {
      "epoch": 0.05952785305308044,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011952615261526153,
      "loss": 3.1675,
      "step": 402430
    },
    {
      "epoch": 0.059546496289238623,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011952415241524154,
      "loss": 3.218,
      "step": 402440
    },
    {
      "epoch": 0.059565139525396806,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011952215221522155,
      "loss": 3.2588,
      "step": 402450
    },
    {
      "epoch": 0.05958378276155499,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011952015201520153,
      "loss": 3.1165,
      "step": 402460
    },
    {
      "epoch": 0.059602425997713177,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011951815181518154,
      "loss": 3.1445,
      "step": 402470
    },
    {
      "epoch": 0.05962106923387136,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011951615161516152,
      "loss": 3.2265,
      "step": 402480
    },
    {
      "epoch": 0.05963971247002954,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011951415141514153,
      "loss": 3.1897,
      "step": 402490
    },
    {
      "epoch": 0.05965835570618772,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011951215121512151,
      "loss": 3.2137,
      "step": 402500
    },
    {
      "epoch": 0.05967699894234591,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011951015101510152,
      "loss": 3.236,
      "step": 402510
    },
    {
      "epoch": 0.059695642178504094,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011950815081508151,
      "loss": 3.26,
      "step": 402520
    },
    {
      "epoch": 0.059714285414662276,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011950615061506152,
      "loss": 3.2422,
      "step": 402530
    },
    {
      "epoch": 0.05973292865082046,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001195041504150415,
      "loss": 3.1975,
      "step": 402540
    },
    {
      "epoch": 0.05975157188697865,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011950215021502151,
      "loss": 3.1969,
      "step": 402550
    },
    {
      "epoch": 0.05977021512313683,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001195001500150015,
      "loss": 3.1436,
      "step": 402560
    },
    {
      "epoch": 0.05978885835929501,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001194981498149815,
      "loss": 3.1526,
      "step": 402570
    },
    {
      "epoch": 0.05980750159545319,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011949614961496149,
      "loss": 3.1499,
      "step": 402580
    },
    {
      "epoch": 0.059826144831611375,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001194941494149415,
      "loss": 3.2102,
      "step": 402590
    },
    {
      "epoch": 0.059844788067769564,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011949214921492149,
      "loss": 3.3,
      "step": 402600
    },
    {
      "epoch": 0.059863431303927746,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001194901490149015,
      "loss": 3.2897,
      "step": 402610
    },
    {
      "epoch": 0.05988207454008593,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011948814881488148,
      "loss": 3.2235,
      "step": 402620
    },
    {
      "epoch": 0.05990071777624411,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001194861486148615,
      "loss": 3.2353,
      "step": 402630
    },
    {
      "epoch": 0.0599193610124023,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011948414841484148,
      "loss": 3.1019,
      "step": 402640
    },
    {
      "epoch": 0.05993800424856048,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011948214821482148,
      "loss": 3.236,
      "step": 402650
    },
    {
      "epoch": 0.05995664748471866,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001194801480148015,
      "loss": 3.1502,
      "step": 402660
    },
    {
      "epoch": 0.059975290720876845,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011947814781478149,
      "loss": 3.1653,
      "step": 402670
    },
    {
      "epoch": 0.059993933957035034,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011947614761476148,
      "loss": 3.1624,
      "step": 402680
    },
    {
      "epoch": 0.060012577193193216,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011947414741474148,
      "loss": 3.1047,
      "step": 402690
    },
    {
      "epoch": 0.0600312204293514,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011947214721472149,
      "loss": 3.3096,
      "step": 402700
    },
    {
      "epoch": 0.06004986366550958,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011947014701470147,
      "loss": 3.0103,
      "step": 402710
    },
    {
      "epoch": 0.06006850690166777,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011946814681468148,
      "loss": 3.1791,
      "step": 402720
    },
    {
      "epoch": 0.06008715013782595,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011946614661466146,
      "loss": 3.2046,
      "step": 402730
    },
    {
      "epoch": 0.06010579337398413,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011946414641464147,
      "loss": 3.1794,
      "step": 402740
    },
    {
      "epoch": 0.060124436610142315,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011946214621462147,
      "loss": 3.1954,
      "step": 402750
    },
    {
      "epoch": 0.0601430798463005,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011946014601460148,
      "loss": 3.3068,
      "step": 402760
    },
    {
      "epoch": 0.060161723082458686,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011945814581458146,
      "loss": 3.1795,
      "step": 402770
    },
    {
      "epoch": 0.06018036631861687,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011945614561456147,
      "loss": 3.1905,
      "step": 402780
    },
    {
      "epoch": 0.06019900955477505,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011945414541454145,
      "loss": 3.1326,
      "step": 402790
    },
    {
      "epoch": 0.06021765279093323,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011945214521452146,
      "loss": 3.0278,
      "step": 402800
    },
    {
      "epoch": 0.06023629602709142,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011945014501450144,
      "loss": 3.1141,
      "step": 402810
    },
    {
      "epoch": 0.0602549392632496,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011944814481448145,
      "loss": 3.1871,
      "step": 402820
    },
    {
      "epoch": 0.060273582499407785,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011944614461446145,
      "loss": 3.1686,
      "step": 402830
    },
    {
      "epoch": 0.06029222573556597,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011944414441444145,
      "loss": 3.2075,
      "step": 402840
    },
    {
      "epoch": 0.060310868971724156,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011944214421442146,
      "loss": 3.2644,
      "step": 402850
    },
    {
      "epoch": 0.06032951220788234,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011944014401440145,
      "loss": 3.2406,
      "step": 402860
    },
    {
      "epoch": 0.06034815544404052,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011943814381438146,
      "loss": 3.1714,
      "step": 402870
    },
    {
      "epoch": 0.0603667986801987,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011943614361436144,
      "loss": 3.2049,
      "step": 402880
    },
    {
      "epoch": 0.06038544191635689,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011943414341434145,
      "loss": 3.098,
      "step": 402890
    },
    {
      "epoch": 0.06040408515251507,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011943214321432143,
      "loss": 3.1899,
      "step": 402900
    },
    {
      "epoch": 0.060422728388673255,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011943014301430144,
      "loss": 3.1711,
      "step": 402910
    },
    {
      "epoch": 0.06044137162483144,
      "grad_norm": 0.75,
      "learning_rate": 0.00011942814281428143,
      "loss": 3.1944,
      "step": 402920
    },
    {
      "epoch": 0.06046001486098962,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011942614261426144,
      "loss": 3.3884,
      "step": 402930
    },
    {
      "epoch": 0.06047865809714781,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011942414241424142,
      "loss": 3.2682,
      "step": 402940
    },
    {
      "epoch": 0.06049730133330599,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011942214221422143,
      "loss": 3.0679,
      "step": 402950
    },
    {
      "epoch": 0.06051594456946417,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011942014201420142,
      "loss": 3.2593,
      "step": 402960
    },
    {
      "epoch": 0.060534587805622354,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011941814181418142,
      "loss": 3.2323,
      "step": 402970
    },
    {
      "epoch": 0.06055323104178054,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011941614161416142,
      "loss": 3.1422,
      "step": 402980
    },
    {
      "epoch": 0.060571874277938725,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011941414141414143,
      "loss": 3.0917,
      "step": 402990
    },
    {
      "epoch": 0.06059051751409691,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011941214121412141,
      "loss": 3.1707,
      "step": 403000
    },
    {
      "epoch": 0.06060916075025509,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011941014101410142,
      "loss": 3.2315,
      "step": 403010
    },
    {
      "epoch": 0.06062780398641328,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001194081408140814,
      "loss": 3.1872,
      "step": 403020
    },
    {
      "epoch": 0.06064644722257146,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011940614061406141,
      "loss": 3.0925,
      "step": 403030
    },
    {
      "epoch": 0.06066509045872964,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001194041404140414,
      "loss": 3.0571,
      "step": 403040
    },
    {
      "epoch": 0.060683733694887824,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001194021402140214,
      "loss": 3.0219,
      "step": 403050
    },
    {
      "epoch": 0.06070237693104601,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011940014001400141,
      "loss": 3.1586,
      "step": 403060
    },
    {
      "epoch": 0.060721020167204195,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011939813981398141,
      "loss": 3.2,
      "step": 403070
    },
    {
      "epoch": 0.06073966340336238,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011939613961396142,
      "loss": 3.1697,
      "step": 403080
    },
    {
      "epoch": 0.06075830663952056,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001193941394139414,
      "loss": 3.1704,
      "step": 403090
    },
    {
      "epoch": 0.06077694987567874,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011939213921392141,
      "loss": 3.2712,
      "step": 403100
    },
    {
      "epoch": 0.06079559311183693,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011939013901390139,
      "loss": 3.1973,
      "step": 403110
    },
    {
      "epoch": 0.06081423634799511,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001193881388138814,
      "loss": 3.1493,
      "step": 403120
    },
    {
      "epoch": 0.060832879584153295,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011938613861386138,
      "loss": 3.2412,
      "step": 403130
    },
    {
      "epoch": 0.06085152282031148,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011938413841384139,
      "loss": 3.2046,
      "step": 403140
    },
    {
      "epoch": 0.060870166056469666,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011938213821382139,
      "loss": 3.1379,
      "step": 403150
    },
    {
      "epoch": 0.06088880929262785,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001193801380138014,
      "loss": 3.1975,
      "step": 403160
    },
    {
      "epoch": 0.06090745252878603,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011937813781378138,
      "loss": 2.9983,
      "step": 403170
    },
    {
      "epoch": 0.06092609576494421,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011937613761376139,
      "loss": 3.1866,
      "step": 403180
    },
    {
      "epoch": 0.0609447390011024,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011937413741374137,
      "loss": 3.2154,
      "step": 403190
    },
    {
      "epoch": 0.06096338223726058,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011937213721372138,
      "loss": 3.2263,
      "step": 403200
    },
    {
      "epoch": 0.060982025473418765,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011937013701370136,
      "loss": 3.1164,
      "step": 403210
    },
    {
      "epoch": 0.06100066870957695,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011936813681368137,
      "loss": 3.2893,
      "step": 403220
    },
    {
      "epoch": 0.061019311945735136,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011936613661366136,
      "loss": 3.1431,
      "step": 403230
    },
    {
      "epoch": 0.06103795518189332,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011936413641364137,
      "loss": 3.0226,
      "step": 403240
    },
    {
      "epoch": 0.0610565984180515,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011936213621362138,
      "loss": 3.253,
      "step": 403250
    },
    {
      "epoch": 0.06107524165420968,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011936013601360136,
      "loss": 3.1572,
      "step": 403260
    },
    {
      "epoch": 0.061093884890367864,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011935813581358137,
      "loss": 3.1874,
      "step": 403270
    },
    {
      "epoch": 0.06111252812652605,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011935613561356136,
      "loss": 3.2398,
      "step": 403280
    },
    {
      "epoch": 0.061131171362684235,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011935413541354137,
      "loss": 3.1396,
      "step": 403290
    },
    {
      "epoch": 0.06114981459884242,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00011935213521352136,
      "loss": 3.1441,
      "step": 403300
    },
    {
      "epoch": 0.0611684578350006,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011935013501350136,
      "loss": 3.1588,
      "step": 403310
    },
    {
      "epoch": 0.06118710107115879,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011934813481348135,
      "loss": 3.2779,
      "step": 403320
    },
    {
      "epoch": 0.06120574430731697,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011934613461346136,
      "loss": 3.1865,
      "step": 403330
    },
    {
      "epoch": 0.06122438754347515,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011934413441344134,
      "loss": 3.1457,
      "step": 403340
    },
    {
      "epoch": 0.061243030779633334,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011934213421342135,
      "loss": 3.2193,
      "step": 403350
    },
    {
      "epoch": 0.06126167401579152,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011934013401340133,
      "loss": 3.1833,
      "step": 403360
    },
    {
      "epoch": 0.061280317251949705,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011933813381338134,
      "loss": 3.1718,
      "step": 403370
    },
    {
      "epoch": 0.06129896048810789,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011933613361336134,
      "loss": 3.0687,
      "step": 403380
    },
    {
      "epoch": 0.06131760372426607,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011933413341334135,
      "loss": 3.2316,
      "step": 403390
    },
    {
      "epoch": 0.06133624696042426,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011933213321332133,
      "loss": 3.2101,
      "step": 403400
    },
    {
      "epoch": 0.06135489019658244,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011933013301330134,
      "loss": 3.1902,
      "step": 403410
    },
    {
      "epoch": 0.06137353343274062,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011932813281328132,
      "loss": 3.3065,
      "step": 403420
    },
    {
      "epoch": 0.061392176668898804,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011932613261326133,
      "loss": 3.157,
      "step": 403430
    },
    {
      "epoch": 0.061410819905056986,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011932413241324134,
      "loss": 3.0552,
      "step": 403440
    },
    {
      "epoch": 0.061429463141215175,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011932213221322132,
      "loss": 3.177,
      "step": 403450
    },
    {
      "epoch": 0.06144810637737336,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011932013201320133,
      "loss": 3.2656,
      "step": 403460
    },
    {
      "epoch": 0.06146674961353154,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00011931813181318133,
      "loss": 3.2035,
      "step": 403470
    },
    {
      "epoch": 0.06148539284968972,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011931613161316134,
      "loss": 3.2317,
      "step": 403480
    },
    {
      "epoch": 0.06150403608584791,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011931413141314132,
      "loss": 3.175,
      "step": 403490
    },
    {
      "epoch": 0.06152267932200609,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011931213121312133,
      "loss": 3.201,
      "step": 403500
    },
    {
      "epoch": 0.061541322558164274,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011931013101310131,
      "loss": 3.252,
      "step": 403510
    },
    {
      "epoch": 0.061559965794322456,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011930813081308132,
      "loss": 3.2079,
      "step": 403520
    },
    {
      "epoch": 0.061578609030480645,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001193061306130613,
      "loss": 3.2457,
      "step": 403530
    },
    {
      "epoch": 0.06159725226663883,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011930413041304131,
      "loss": 3.136,
      "step": 403540
    },
    {
      "epoch": 0.06161589550279701,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001193021302130213,
      "loss": 3.0783,
      "step": 403550
    },
    {
      "epoch": 0.06163453873895519,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011930013001300131,
      "loss": 3.2353,
      "step": 403560
    },
    {
      "epoch": 0.06165318197511338,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001192981298129813,
      "loss": 3.1736,
      "step": 403570
    },
    {
      "epoch": 0.06167182521127156,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001192961296129613,
      "loss": 3.1761,
      "step": 403580
    },
    {
      "epoch": 0.061690468447429744,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011929412941294129,
      "loss": 3.1579,
      "step": 403590
    },
    {
      "epoch": 0.061709111683587926,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001192921292129213,
      "loss": 3.1053,
      "step": 403600
    },
    {
      "epoch": 0.06172775491974611,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011929012901290129,
      "loss": 3.172,
      "step": 403610
    },
    {
      "epoch": 0.0617463981559043,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001192881288128813,
      "loss": 3.0633,
      "step": 403620
    },
    {
      "epoch": 0.06176504139206248,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011928612861286128,
      "loss": 3.2777,
      "step": 403630
    },
    {
      "epoch": 0.06178368462822066,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011928412841284129,
      "loss": 3.1749,
      "step": 403640
    },
    {
      "epoch": 0.06180232786437884,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001192821282128213,
      "loss": 3.1643,
      "step": 403650
    },
    {
      "epoch": 0.06182097110053703,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011928012801280128,
      "loss": 3.3347,
      "step": 403660
    },
    {
      "epoch": 0.061839614336695214,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011927812781278129,
      "loss": 3.195,
      "step": 403670
    },
    {
      "epoch": 0.061858257572853396,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011927612761276127,
      "loss": 3.3309,
      "step": 403680
    },
    {
      "epoch": 0.06187690080901158,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011927412741274128,
      "loss": 3.2028,
      "step": 403690
    },
    {
      "epoch": 0.06189554404516977,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011927212721272128,
      "loss": 3.1429,
      "step": 403700
    },
    {
      "epoch": 0.06191418728132795,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011927012701270129,
      "loss": 3.1824,
      "step": 403710
    },
    {
      "epoch": 0.06193283051748613,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011926812681268127,
      "loss": 3.2707,
      "step": 403720
    },
    {
      "epoch": 0.061951473753644314,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011926612661266128,
      "loss": 3.1555,
      "step": 403730
    },
    {
      "epoch": 0.0619701169898025,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011926412641264126,
      "loss": 3.3296,
      "step": 403740
    },
    {
      "epoch": 0.061988760225960685,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011926212621262127,
      "loss": 3.203,
      "step": 403750
    },
    {
      "epoch": 0.06200740346211887,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011926012601260125,
      "loss": 3.2513,
      "step": 403760
    },
    {
      "epoch": 0.06202604669827705,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011925812581258126,
      "loss": 3.0608,
      "step": 403770
    },
    {
      "epoch": 0.06204468993443523,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011925612561256126,
      "loss": 3.2281,
      "step": 403780
    },
    {
      "epoch": 0.06206333317059342,
      "grad_norm": 0.75,
      "learning_rate": 0.00011925412541254127,
      "loss": 3.0865,
      "step": 403790
    },
    {
      "epoch": 0.0620819764067516,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011925212521252125,
      "loss": 3.2911,
      "step": 403800
    },
    {
      "epoch": 0.062100619642909784,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011925012501250126,
      "loss": 3.1806,
      "step": 403810
    },
    {
      "epoch": 0.062119262879067966,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011924812481248124,
      "loss": 3.2274,
      "step": 403820
    },
    {
      "epoch": 0.062137906115226155,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011924612461246125,
      "loss": 3.2539,
      "step": 403830
    },
    {
      "epoch": 0.06215654935138434,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011924412441244126,
      "loss": 3.1091,
      "step": 403840
    },
    {
      "epoch": 0.06217519258754252,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011924212421242124,
      "loss": 3.0608,
      "step": 403850
    },
    {
      "epoch": 0.0621938358237007,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011924012401240125,
      "loss": 3.1194,
      "step": 403860
    },
    {
      "epoch": 0.06221247905985889,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011923812381238125,
      "loss": 3.2993,
      "step": 403870
    },
    {
      "epoch": 0.06223112229601707,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011923612361236125,
      "loss": 3.1611,
      "step": 403880
    },
    {
      "epoch": 0.062249765532175254,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011923412341234124,
      "loss": 3.0075,
      "step": 403890
    },
    {
      "epoch": 0.062268408768333436,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011923212321232125,
      "loss": 3.2244,
      "step": 403900
    },
    {
      "epoch": 0.062287052004491625,
      "grad_norm": 0.75,
      "learning_rate": 0.00011923012301230123,
      "loss": 3.24,
      "step": 403910
    },
    {
      "epoch": 0.06230569524064981,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011922812281228124,
      "loss": 3.2023,
      "step": 403920
    },
    {
      "epoch": 0.06232433847680799,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011922612261226123,
      "loss": 3.3154,
      "step": 403930
    },
    {
      "epoch": 0.06234298171296617,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011922412241224124,
      "loss": 3.1698,
      "step": 403940
    },
    {
      "epoch": 0.06236162494912435,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011922212221222122,
      "loss": 3.0759,
      "step": 403950
    },
    {
      "epoch": 0.06238026818528254,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011922012201220123,
      "loss": 3.2466,
      "step": 403960
    },
    {
      "epoch": 0.062398911421440724,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011921812181218121,
      "loss": 3.0433,
      "step": 403970
    },
    {
      "epoch": 0.062417554657598906,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011921612161216122,
      "loss": 3.2377,
      "step": 403980
    },
    {
      "epoch": 0.06243619789375709,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001192141214121412,
      "loss": 3.1434,
      "step": 403990
    },
    {
      "epoch": 0.06245484112991528,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011921212121212121,
      "loss": 3.0777,
      "step": 404000
    },
    {
      "epoch": 0.06247348436607346,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011921012101210121,
      "loss": 3.2216,
      "step": 404010
    },
    {
      "epoch": 0.06249212760223164,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011920812081208122,
      "loss": 3.2912,
      "step": 404020
    },
    {
      "epoch": 0.06251077083838982,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011920612061206123,
      "loss": 3.1442,
      "step": 404030
    },
    {
      "epoch": 0.06252941407454801,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011920412041204121,
      "loss": 3.0955,
      "step": 404040
    },
    {
      "epoch": 0.06254805731070619,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011920212021202122,
      "loss": 3.2059,
      "step": 404050
    },
    {
      "epoch": 0.06256670054686438,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001192001200120012,
      "loss": 3.2816,
      "step": 404060
    },
    {
      "epoch": 0.06258534378302257,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011919811981198121,
      "loss": 3.3102,
      "step": 404070
    },
    {
      "epoch": 0.06260398701918074,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011919611961196119,
      "loss": 3.1214,
      "step": 404080
    },
    {
      "epoch": 0.06262263025533893,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001191941194119412,
      "loss": 3.2469,
      "step": 404090
    },
    {
      "epoch": 0.06264127349149712,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001191921192119212,
      "loss": 3.1195,
      "step": 404100
    },
    {
      "epoch": 0.0626599167276553,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011919011901190121,
      "loss": 3.4013,
      "step": 404110
    },
    {
      "epoch": 0.06267855996381348,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011918811881188119,
      "loss": 3.1336,
      "step": 404120
    },
    {
      "epoch": 0.06269720319997166,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001191861186118612,
      "loss": 3.2139,
      "step": 404130
    },
    {
      "epoch": 0.06271584643612985,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011918411841184118,
      "loss": 3.1874,
      "step": 404140
    },
    {
      "epoch": 0.06273448967228804,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011918211821182119,
      "loss": 3.1687,
      "step": 404150
    },
    {
      "epoch": 0.06275313290844621,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011918011801180119,
      "loss": 3.1464,
      "step": 404160
    },
    {
      "epoch": 0.0627717761446044,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011917811781178118,
      "loss": 3.1311,
      "step": 404170
    },
    {
      "epoch": 0.06279041938076257,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011917611761176118,
      "loss": 3.1311,
      "step": 404180
    },
    {
      "epoch": 0.06280906261692076,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011917411741174119,
      "loss": 3.2075,
      "step": 404190
    },
    {
      "epoch": 0.06282770585307895,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011917211721172117,
      "loss": 3.2007,
      "step": 404200
    },
    {
      "epoch": 0.06284634908923713,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011917011701170118,
      "loss": 3.19,
      "step": 404210
    },
    {
      "epoch": 0.06286499232539532,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011916811681168116,
      "loss": 3.2006,
      "step": 404220
    },
    {
      "epoch": 0.0628836355615535,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011916611661166117,
      "loss": 3.1699,
      "step": 404230
    },
    {
      "epoch": 0.06290227879771168,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011916411641164118,
      "loss": 3.259,
      "step": 404240
    },
    {
      "epoch": 0.06292092203386987,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011916211621162117,
      "loss": 3.1478,
      "step": 404250
    },
    {
      "epoch": 0.06293956527002804,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011916011601160118,
      "loss": 3.1056,
      "step": 404260
    },
    {
      "epoch": 0.06295820850618623,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011915811581158116,
      "loss": 3.1751,
      "step": 404270
    },
    {
      "epoch": 0.06297685174234442,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011915611561156117,
      "loss": 2.912,
      "step": 404280
    },
    {
      "epoch": 0.0629954949785026,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011915411541154115,
      "loss": 3.0626,
      "step": 404290
    },
    {
      "epoch": 0.06301413821466079,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011915211521152116,
      "loss": 3.2155,
      "step": 404300
    },
    {
      "epoch": 0.06303278145081896,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011915011501150115,
      "loss": 3.1466,
      "step": 404310
    },
    {
      "epoch": 0.06305142468697715,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011914811481148116,
      "loss": 3.1946,
      "step": 404320
    },
    {
      "epoch": 0.06307006792313534,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011914611461146115,
      "loss": 3.2191,
      "step": 404330
    },
    {
      "epoch": 0.06308871115929351,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011914411441144116,
      "loss": 3.2637,
      "step": 404340
    },
    {
      "epoch": 0.0631073543954517,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011914211421142114,
      "loss": 3.1182,
      "step": 404350
    },
    {
      "epoch": 0.06312599763160989,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011914011401140115,
      "loss": 3.1276,
      "step": 404360
    },
    {
      "epoch": 0.06314464086776807,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011913811381138113,
      "loss": 3.2415,
      "step": 404370
    },
    {
      "epoch": 0.06316328410392626,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011913611361136114,
      "loss": 3.1463,
      "step": 404380
    },
    {
      "epoch": 0.06318192734008443,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011913411341134112,
      "loss": 3.1842,
      "step": 404390
    },
    {
      "epoch": 0.06320057057624262,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011913211321132113,
      "loss": 3.275,
      "step": 404400
    },
    {
      "epoch": 0.06321921381240081,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011913011301130113,
      "loss": 3.1826,
      "step": 404410
    },
    {
      "epoch": 0.06323785704855898,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011912811281128114,
      "loss": 3.146,
      "step": 404420
    },
    {
      "epoch": 0.06325650028471717,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011912611261126115,
      "loss": 3.0941,
      "step": 404430
    },
    {
      "epoch": 0.06327514352087536,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011912411241124113,
      "loss": 3.201,
      "step": 404440
    },
    {
      "epoch": 0.06329378675703354,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011912211221122114,
      "loss": 3.1027,
      "step": 404450
    },
    {
      "epoch": 0.06331242999319173,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011912011201120112,
      "loss": 3.0792,
      "step": 404460
    },
    {
      "epoch": 0.0633310732293499,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011911811181118113,
      "loss": 3.0677,
      "step": 404470
    },
    {
      "epoch": 0.06334971646550809,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011911611161116111,
      "loss": 3.2348,
      "step": 404480
    },
    {
      "epoch": 0.06336835970166628,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011911411141114112,
      "loss": 3.0941,
      "step": 404490
    },
    {
      "epoch": 0.06338700293782445,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011911211121112112,
      "loss": 3.1897,
      "step": 404500
    },
    {
      "epoch": 0.06340564617398264,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011911011101110113,
      "loss": 3.214,
      "step": 404510
    },
    {
      "epoch": 0.06342428941014082,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011910811081108111,
      "loss": 3.3304,
      "step": 404520
    },
    {
      "epoch": 0.06344293264629901,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011910611061106112,
      "loss": 3.1667,
      "step": 404530
    },
    {
      "epoch": 0.0634615758824572,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001191041104110411,
      "loss": 3.191,
      "step": 404540
    },
    {
      "epoch": 0.06348021911861537,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011910211021102111,
      "loss": 3.2539,
      "step": 404550
    },
    {
      "epoch": 0.06349886235477356,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001191001100110011,
      "loss": 3.1327,
      "step": 404560
    },
    {
      "epoch": 0.06351750559093175,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011909810981098111,
      "loss": 3.1724,
      "step": 404570
    },
    {
      "epoch": 0.06353614882708992,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001190961096109611,
      "loss": 3.0159,
      "step": 404580
    },
    {
      "epoch": 0.06355479206324811,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001190941094109411,
      "loss": 3.1427,
      "step": 404590
    },
    {
      "epoch": 0.06357343529940629,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011909210921092109,
      "loss": 3.1803,
      "step": 404600
    },
    {
      "epoch": 0.06359207853556448,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001190901090109011,
      "loss": 3.167,
      "step": 404610
    },
    {
      "epoch": 0.06361072177172267,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011908810881088108,
      "loss": 3.1242,
      "step": 404620
    },
    {
      "epoch": 0.06362936500788084,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011908610861086109,
      "loss": 3.2788,
      "step": 404630
    },
    {
      "epoch": 0.06364800824403903,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001190841084108411,
      "loss": 3.2732,
      "step": 404640
    },
    {
      "epoch": 0.0636666514801972,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011908210821082109,
      "loss": 3.2352,
      "step": 404650
    },
    {
      "epoch": 0.0636852947163554,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001190801080108011,
      "loss": 3.1731,
      "step": 404660
    },
    {
      "epoch": 0.06370393795251358,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011907810781078108,
      "loss": 3.3221,
      "step": 404670
    },
    {
      "epoch": 0.06372258118867176,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011907610761076109,
      "loss": 3.1951,
      "step": 404680
    },
    {
      "epoch": 0.06374122442482995,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011907410741074107,
      "loss": 3.1513,
      "step": 404690
    },
    {
      "epoch": 0.06375986766098814,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011907210721072108,
      "loss": 3.016,
      "step": 404700
    },
    {
      "epoch": 0.06377851089714631,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011907010701070106,
      "loss": 3.2317,
      "step": 404710
    },
    {
      "epoch": 0.0637971541333045,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011906810681068107,
      "loss": 3.2808,
      "step": 404720
    },
    {
      "epoch": 0.06381579736946268,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011906610661066107,
      "loss": 3.1862,
      "step": 404730
    },
    {
      "epoch": 0.06383444060562087,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011906410641064108,
      "loss": 3.1715,
      "step": 404740
    },
    {
      "epoch": 0.06385308384177905,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011906210621062106,
      "loss": 3.1024,
      "step": 404750
    },
    {
      "epoch": 0.06387172707793723,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011906010601060107,
      "loss": 3.0846,
      "step": 404760
    },
    {
      "epoch": 0.06389037031409542,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011905810581058105,
      "loss": 3.1891,
      "step": 404770
    },
    {
      "epoch": 0.06390901355025361,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011905610561056106,
      "loss": 3.1414,
      "step": 404780
    },
    {
      "epoch": 0.06392765678641178,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011905410541054106,
      "loss": 3.2961,
      "step": 404790
    },
    {
      "epoch": 0.06394630002256997,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011905210521052105,
      "loss": 3.112,
      "step": 404800
    },
    {
      "epoch": 0.06396494325872815,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011905010501050105,
      "loss": 3.1332,
      "step": 404810
    },
    {
      "epoch": 0.06398358649488634,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011904810481048106,
      "loss": 3.1199,
      "step": 404820
    },
    {
      "epoch": 0.06400222973104452,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011904610461046107,
      "loss": 3.2897,
      "step": 404830
    },
    {
      "epoch": 0.0640208729672027,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011904410441044105,
      "loss": 3.2836,
      "step": 404840
    },
    {
      "epoch": 0.06403951620336089,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011904210421042106,
      "loss": 3.1035,
      "step": 404850
    },
    {
      "epoch": 0.06405815943951906,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011904010401040104,
      "loss": 3.3214,
      "step": 404860
    },
    {
      "epoch": 0.06407680267567725,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011903810381038105,
      "loss": 3.092,
      "step": 404870
    },
    {
      "epoch": 0.06409544591183544,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011903610361036104,
      "loss": 3.2733,
      "step": 404880
    },
    {
      "epoch": 0.06411408914799362,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011903410341034105,
      "loss": 3.2392,
      "step": 404890
    },
    {
      "epoch": 0.0641327323841518,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011903210321032104,
      "loss": 3.0872,
      "step": 404900
    },
    {
      "epoch": 0.06415137562031,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011903010301030104,
      "loss": 3.1515,
      "step": 404910
    },
    {
      "epoch": 0.06417001885646817,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011902810281028103,
      "loss": 3.2902,
      "step": 404920
    },
    {
      "epoch": 0.06418866209262636,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011902610261026104,
      "loss": 3.1296,
      "step": 404930
    },
    {
      "epoch": 0.06420730532878453,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011902410241024102,
      "loss": 3.1793,
      "step": 404940
    },
    {
      "epoch": 0.06422594856494272,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011902210221022103,
      "loss": 3.1125,
      "step": 404950
    },
    {
      "epoch": 0.06424459180110091,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011902010201020102,
      "loss": 3.1508,
      "step": 404960
    },
    {
      "epoch": 0.06426323503725909,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011901810181018103,
      "loss": 3.1834,
      "step": 404970
    },
    {
      "epoch": 0.06428187827341728,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011901610161016101,
      "loss": 3.2263,
      "step": 404980
    },
    {
      "epoch": 0.06430052150957545,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011901410141014102,
      "loss": 3.0347,
      "step": 404990
    },
    {
      "epoch": 0.06431916474573364,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000119012101210121,
      "loss": 3.231,
      "step": 405000
    },
    {
      "epoch": 0.06433780798189183,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011901010101010101,
      "loss": 3.0953,
      "step": 405010
    },
    {
      "epoch": 0.06435645121805,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011900810081008102,
      "loss": 3.0858,
      "step": 405020
    },
    {
      "epoch": 0.06437509445420819,
      "grad_norm": 0.6875,
      "learning_rate": 0.000119006100610061,
      "loss": 3.0632,
      "step": 405030
    },
    {
      "epoch": 0.06439373769036638,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011900410041004101,
      "loss": 3.0939,
      "step": 405040
    },
    {
      "epoch": 0.06441238092652456,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011900210021002101,
      "loss": 3.2081,
      "step": 405050
    },
    {
      "epoch": 0.06443102416268275,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011900010001000102,
      "loss": 3.2214,
      "step": 405060
    },
    {
      "epoch": 0.06444966739884092,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000118998099809981,
      "loss": 3.2304,
      "step": 405070
    },
    {
      "epoch": 0.06446831063499911,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011899609960996101,
      "loss": 3.1547,
      "step": 405080
    },
    {
      "epoch": 0.0644869538711573,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011899409940994099,
      "loss": 3.1646,
      "step": 405090
    },
    {
      "epoch": 0.06450559710731547,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000118992099209921,
      "loss": 3.1271,
      "step": 405100
    },
    {
      "epoch": 0.06452424034347366,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000118990099009901,
      "loss": 3.2576,
      "step": 405110
    },
    {
      "epoch": 0.06454288357963185,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011898809880988099,
      "loss": 3.1409,
      "step": 405120
    },
    {
      "epoch": 0.06456152681579003,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011898609860986099,
      "loss": 3.1995,
      "step": 405130
    },
    {
      "epoch": 0.06458017005194822,
      "grad_norm": 0.71875,
      "learning_rate": 0.000118984098409841,
      "loss": 3.0805,
      "step": 405140
    },
    {
      "epoch": 0.06459881328810639,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011898209820982098,
      "loss": 3.1156,
      "step": 405150
    },
    {
      "epoch": 0.06461745652426458,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011898009800980099,
      "loss": 3.24,
      "step": 405160
    },
    {
      "epoch": 0.06463609976042277,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011897809780978097,
      "loss": 3.1215,
      "step": 405170
    },
    {
      "epoch": 0.06465474299658094,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011897609760976098,
      "loss": 3.1319,
      "step": 405180
    },
    {
      "epoch": 0.06467338623273913,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011897409740974098,
      "loss": 3.1595,
      "step": 405190
    },
    {
      "epoch": 0.06469202946889731,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011897209720972098,
      "loss": 3.0621,
      "step": 405200
    },
    {
      "epoch": 0.0647106727050555,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011897009700970097,
      "loss": 3.2779,
      "step": 405210
    },
    {
      "epoch": 0.06472931594121369,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011896809680968098,
      "loss": 3.1711,
      "step": 405220
    },
    {
      "epoch": 0.06474795917737186,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011896609660966098,
      "loss": 3.2959,
      "step": 405230
    },
    {
      "epoch": 0.06476660241353005,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011896409640964097,
      "loss": 3.1175,
      "step": 405240
    },
    {
      "epoch": 0.06478524564968824,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011896209620962098,
      "loss": 3.1534,
      "step": 405250
    },
    {
      "epoch": 0.06480388888584641,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011896009600960096,
      "loss": 3.0424,
      "step": 405260
    },
    {
      "epoch": 0.0648225321220046,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011895809580958097,
      "loss": 3.2388,
      "step": 405270
    },
    {
      "epoch": 0.06484117535816278,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011895609560956096,
      "loss": 3.2146,
      "step": 405280
    },
    {
      "epoch": 0.06485981859432097,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011895409540954097,
      "loss": 3.227,
      "step": 405290
    },
    {
      "epoch": 0.06487846183047916,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011895209520952095,
      "loss": 3.1668,
      "step": 405300
    },
    {
      "epoch": 0.06489710506663733,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011895009500950096,
      "loss": 3.2259,
      "step": 405310
    },
    {
      "epoch": 0.06491574830279552,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011894809480948095,
      "loss": 3.0918,
      "step": 405320
    },
    {
      "epoch": 0.0649343915389537,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011894609460946095,
      "loss": 3.1671,
      "step": 405330
    },
    {
      "epoch": 0.06495303477511188,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011894409440944094,
      "loss": 3.2387,
      "step": 405340
    },
    {
      "epoch": 0.06497167801127007,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011894209420942095,
      "loss": 3.2037,
      "step": 405350
    },
    {
      "epoch": 0.06499032124742825,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011894009400940094,
      "loss": 3.229,
      "step": 405360
    },
    {
      "epoch": 0.06500896448358644,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011893809380938095,
      "loss": 3.2415,
      "step": 405370
    },
    {
      "epoch": 0.06502760771974463,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011893609360936093,
      "loss": 3.2644,
      "step": 405380
    },
    {
      "epoch": 0.0650462509559028,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011893409340934094,
      "loss": 3.1055,
      "step": 405390
    },
    {
      "epoch": 0.06506489419206099,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011893209320932092,
      "loss": 3.2107,
      "step": 405400
    },
    {
      "epoch": 0.06508353742821917,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011893009300930093,
      "loss": 3.171,
      "step": 405410
    },
    {
      "epoch": 0.06510218066437735,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011892809280928094,
      "loss": 3.3198,
      "step": 405420
    },
    {
      "epoch": 0.06512082390053554,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00011892609260926094,
      "loss": 3.1363,
      "step": 405430
    },
    {
      "epoch": 0.06513946713669372,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011892409240924093,
      "loss": 3.2104,
      "step": 405440
    },
    {
      "epoch": 0.06515811037285191,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011892209220922093,
      "loss": 3.1118,
      "step": 405450
    },
    {
      "epoch": 0.0651767536090101,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011892009200920094,
      "loss": 3.2292,
      "step": 405460
    },
    {
      "epoch": 0.06519539684516827,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011891809180918092,
      "loss": 3.2102,
      "step": 405470
    },
    {
      "epoch": 0.06521404008132646,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011891609160916093,
      "loss": 3.2559,
      "step": 405480
    },
    {
      "epoch": 0.06523268331748464,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011891409140914091,
      "loss": 3.109,
      "step": 405490
    },
    {
      "epoch": 0.06525132655364282,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011891209120912092,
      "loss": 3.1843,
      "step": 405500
    },
    {
      "epoch": 0.06526996978980101,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011891009100910092,
      "loss": 3.2526,
      "step": 405510
    },
    {
      "epoch": 0.06528861302595919,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011890809080908092,
      "loss": 3.1946,
      "step": 405520
    },
    {
      "epoch": 0.06530725626211738,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001189060906090609,
      "loss": 3.0921,
      "step": 405530
    },
    {
      "epoch": 0.06532589949827555,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011890409040904092,
      "loss": 3.232,
      "step": 405540
    },
    {
      "epoch": 0.06534454273443374,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001189020902090209,
      "loss": 3.2219,
      "step": 405550
    },
    {
      "epoch": 0.06536318597059193,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011890009000900091,
      "loss": 3.1826,
      "step": 405560
    },
    {
      "epoch": 0.0653818292067501,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011889808980898089,
      "loss": 3.1557,
      "step": 405570
    },
    {
      "epoch": 0.0654004724429083,
      "grad_norm": 0.875,
      "learning_rate": 0.0001188960896089609,
      "loss": 3.1994,
      "step": 405580
    },
    {
      "epoch": 0.06541911567906648,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001188940894089409,
      "loss": 3.1462,
      "step": 405590
    },
    {
      "epoch": 0.06543775891522466,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001188920892089209,
      "loss": 3.234,
      "step": 405600
    },
    {
      "epoch": 0.06545640215138285,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011889008900890091,
      "loss": 3.2,
      "step": 405610
    },
    {
      "epoch": 0.06547504538754102,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001188880888088809,
      "loss": 3.1364,
      "step": 405620
    },
    {
      "epoch": 0.06549368862369921,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001188860886088609,
      "loss": 3.1193,
      "step": 405630
    },
    {
      "epoch": 0.0655123318598574,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011888408840884089,
      "loss": 3.2114,
      "step": 405640
    },
    {
      "epoch": 0.06553097509601558,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001188820882088209,
      "loss": 3.2247,
      "step": 405650
    },
    {
      "epoch": 0.06554961833217376,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011888008800880088,
      "loss": 3.2091,
      "step": 405660
    },
    {
      "epoch": 0.06556826156833194,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011887808780878089,
      "loss": 3.1649,
      "step": 405670
    },
    {
      "epoch": 0.06558690480449013,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011887608760876088,
      "loss": 3.2757,
      "step": 405680
    },
    {
      "epoch": 0.06560554804064832,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011887408740874089,
      "loss": 3.1593,
      "step": 405690
    },
    {
      "epoch": 0.06562419127680649,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011887208720872087,
      "loss": 3.2119,
      "step": 405700
    },
    {
      "epoch": 0.06564283451296468,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011887008700870088,
      "loss": 3.2485,
      "step": 405710
    },
    {
      "epoch": 0.06566147774912287,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011886808680868086,
      "loss": 3.0613,
      "step": 405720
    },
    {
      "epoch": 0.06568012098528105,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011886608660866087,
      "loss": 3.2724,
      "step": 405730
    },
    {
      "epoch": 0.06569876422143923,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011886408640864087,
      "loss": 3.2394,
      "step": 405740
    },
    {
      "epoch": 0.06571740745759741,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011886208620862088,
      "loss": 3.1725,
      "step": 405750
    },
    {
      "epoch": 0.0657360506937556,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011886008600860086,
      "loss": 3.1658,
      "step": 405760
    },
    {
      "epoch": 0.06575469392991379,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011885808580858087,
      "loss": 3.119,
      "step": 405770
    },
    {
      "epoch": 0.06577333716607196,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011885608560856085,
      "loss": 3.2504,
      "step": 405780
    },
    {
      "epoch": 0.06579198040223015,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011885408540854086,
      "loss": 3.3091,
      "step": 405790
    },
    {
      "epoch": 0.06581062363838834,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011885208520852084,
      "loss": 3.1575,
      "step": 405800
    },
    {
      "epoch": 0.06582926687454652,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011885008500850085,
      "loss": 3.2421,
      "step": 405810
    },
    {
      "epoch": 0.0658479101107047,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011884808480848086,
      "loss": 3.1533,
      "step": 405820
    },
    {
      "epoch": 0.06586655334686288,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011884608460846086,
      "loss": 3.0133,
      "step": 405830
    },
    {
      "epoch": 0.06588519658302107,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011884408440844087,
      "loss": 3.3503,
      "step": 405840
    },
    {
      "epoch": 0.06590383981917926,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011884208420842085,
      "loss": 3.0551,
      "step": 405850
    },
    {
      "epoch": 0.06592248305533743,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011884008400840086,
      "loss": 3.2671,
      "step": 405860
    },
    {
      "epoch": 0.06594112629149562,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011883808380838084,
      "loss": 3.1351,
      "step": 405870
    },
    {
      "epoch": 0.0659597695276538,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011883608360836085,
      "loss": 3.1083,
      "step": 405880
    },
    {
      "epoch": 0.06597841276381199,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011883408340834083,
      "loss": 3.1251,
      "step": 405890
    },
    {
      "epoch": 0.06599705599997018,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011883208320832084,
      "loss": 3.1543,
      "step": 405900
    },
    {
      "epoch": 0.06601569923612835,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011883008300830083,
      "loss": 3.3026,
      "step": 405910
    },
    {
      "epoch": 0.06603434247228654,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011882808280828084,
      "loss": 3.2551,
      "step": 405920
    },
    {
      "epoch": 0.06605298570844473,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011882608260826083,
      "loss": 3.2794,
      "step": 405930
    },
    {
      "epoch": 0.0660716289446029,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011882408240824083,
      "loss": 3.1878,
      "step": 405940
    },
    {
      "epoch": 0.06609027218076109,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011882208220822082,
      "loss": 3.2874,
      "step": 405950
    },
    {
      "epoch": 0.06610891541691927,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011882008200820083,
      "loss": 3.1622,
      "step": 405960
    },
    {
      "epoch": 0.06612755865307746,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011881808180818081,
      "loss": 3.1846,
      "step": 405970
    },
    {
      "epoch": 0.06614620188923565,
      "grad_norm": 0.75,
      "learning_rate": 0.00011881608160816082,
      "loss": 3.1337,
      "step": 405980
    },
    {
      "epoch": 0.06616484512539382,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011881408140814081,
      "loss": 3.2203,
      "step": 405990
    },
    {
      "epoch": 0.06618348836155201,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011881208120812082,
      "loss": 3.2102,
      "step": 406000
    },
    {
      "epoch": 0.06620213159771018,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011881008100810083,
      "loss": 3.2247,
      "step": 406010
    },
    {
      "epoch": 0.06622077483386837,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011880808080808081,
      "loss": 3.1858,
      "step": 406020
    },
    {
      "epoch": 0.06623941807002656,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011880608060806082,
      "loss": 3.2474,
      "step": 406030
    },
    {
      "epoch": 0.06625806130618474,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001188040804080408,
      "loss": 3.1869,
      "step": 406040
    },
    {
      "epoch": 0.06627670454234293,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011880208020802081,
      "loss": 3.1955,
      "step": 406050
    },
    {
      "epoch": 0.06629534777850112,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011880008000800081,
      "loss": 3.2367,
      "step": 406060
    },
    {
      "epoch": 0.06631399101465929,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001187980798079808,
      "loss": 3.1739,
      "step": 406070
    },
    {
      "epoch": 0.06633263425081748,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001187960796079608,
      "loss": 3.1654,
      "step": 406080
    },
    {
      "epoch": 0.06635127748697565,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011879407940794081,
      "loss": 3.2255,
      "step": 406090
    },
    {
      "epoch": 0.06636992072313384,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011879207920792079,
      "loss": 3.3446,
      "step": 406100
    },
    {
      "epoch": 0.06638856395929203,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001187900790079008,
      "loss": 3.0604,
      "step": 406110
    },
    {
      "epoch": 0.06640720719545021,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011878807880788078,
      "loss": 3.2504,
      "step": 406120
    },
    {
      "epoch": 0.0664258504316084,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011878607860786079,
      "loss": 3.2346,
      "step": 406130
    },
    {
      "epoch": 0.06644449366776659,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011878407840784079,
      "loss": 3.2883,
      "step": 406140
    },
    {
      "epoch": 0.06646313690392476,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001187820782078208,
      "loss": 3.261,
      "step": 406150
    },
    {
      "epoch": 0.06648178014008295,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011878007800780078,
      "loss": 3.3646,
      "step": 406160
    },
    {
      "epoch": 0.06650042337624112,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011877807780778079,
      "loss": 3.1495,
      "step": 406170
    },
    {
      "epoch": 0.06651906661239931,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011877607760776077,
      "loss": 3.1805,
      "step": 406180
    },
    {
      "epoch": 0.0665377098485575,
      "grad_norm": 0.75,
      "learning_rate": 0.00011877407740774078,
      "loss": 3.2584,
      "step": 406190
    },
    {
      "epoch": 0.06655635308471568,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011877207720772079,
      "loss": 3.05,
      "step": 406200
    },
    {
      "epoch": 0.06657499632087387,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011877007700770077,
      "loss": 3.1027,
      "step": 406210
    },
    {
      "epoch": 0.06659363955703204,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011876807680768078,
      "loss": 3.2388,
      "step": 406220
    },
    {
      "epoch": 0.06661228279319023,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011876607660766077,
      "loss": 3.294,
      "step": 406230
    },
    {
      "epoch": 0.06663092602934842,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011876407640764078,
      "loss": 3.2354,
      "step": 406240
    },
    {
      "epoch": 0.0666495692655066,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011876207620762077,
      "loss": 3.0789,
      "step": 406250
    },
    {
      "epoch": 0.06666821250166478,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011876007600760077,
      "loss": 3.1937,
      "step": 406260
    },
    {
      "epoch": 0.06668685573782297,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011875807580758076,
      "loss": 3.1393,
      "step": 406270
    },
    {
      "epoch": 0.06670549897398115,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011875607560756077,
      "loss": 3.0053,
      "step": 406280
    },
    {
      "epoch": 0.06672414221013934,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011875407540754075,
      "loss": 3.1359,
      "step": 406290
    },
    {
      "epoch": 0.06674278544629751,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011875207520752076,
      "loss": 3.2821,
      "step": 406300
    },
    {
      "epoch": 0.0667614286824557,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011875007500750075,
      "loss": 3.0818,
      "step": 406310
    },
    {
      "epoch": 0.06678007191861389,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011874807480748076,
      "loss": 3.2086,
      "step": 406320
    },
    {
      "epoch": 0.06679871515477206,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011874607460746074,
      "loss": 3.0689,
      "step": 406330
    },
    {
      "epoch": 0.06681735839093025,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011874407440744075,
      "loss": 3.0707,
      "step": 406340
    },
    {
      "epoch": 0.06683600162708843,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011874207420742074,
      "loss": 3.2139,
      "step": 406350
    },
    {
      "epoch": 0.06685464486324662,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011874007400740074,
      "loss": 3.1453,
      "step": 406360
    },
    {
      "epoch": 0.0668732880994048,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011873807380738074,
      "loss": 3.2278,
      "step": 406370
    },
    {
      "epoch": 0.06689193133556298,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011873607360736075,
      "loss": 3.0956,
      "step": 406380
    },
    {
      "epoch": 0.06691057457172117,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011873407340734073,
      "loss": 3.1589,
      "step": 406390
    },
    {
      "epoch": 0.06692921780787936,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011873207320732074,
      "loss": 3.215,
      "step": 406400
    },
    {
      "epoch": 0.06694786104403753,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011873007300730075,
      "loss": 3.2448,
      "step": 406410
    },
    {
      "epoch": 0.06696650428019572,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011872807280728073,
      "loss": 3.2542,
      "step": 406420
    },
    {
      "epoch": 0.0669851475163539,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011872607260726074,
      "loss": 3.2473,
      "step": 406430
    },
    {
      "epoch": 0.06700379075251209,
      "grad_norm": 0.625,
      "learning_rate": 0.00011872407240724072,
      "loss": 3.2484,
      "step": 406440
    },
    {
      "epoch": 0.06702243398867028,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011872207220722073,
      "loss": 3.2712,
      "step": 406450
    },
    {
      "epoch": 0.06704107722482845,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011872007200720073,
      "loss": 3.0876,
      "step": 406460
    },
    {
      "epoch": 0.06705972046098664,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011871807180718074,
      "loss": 3.0511,
      "step": 406470
    },
    {
      "epoch": 0.06707836369714483,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011871607160716072,
      "loss": 3.2234,
      "step": 406480
    },
    {
      "epoch": 0.067097006933303,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011871407140714073,
      "loss": 3.1973,
      "step": 406490
    },
    {
      "epoch": 0.0671156501694612,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011871207120712071,
      "loss": 3.3656,
      "step": 406500
    },
    {
      "epoch": 0.06713429340561937,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011871007100710072,
      "loss": 3.1956,
      "step": 406510
    },
    {
      "epoch": 0.06715293664177756,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001187080708070807,
      "loss": 3.175,
      "step": 406520
    },
    {
      "epoch": 0.06717157987793575,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011870607060706071,
      "loss": 3.1692,
      "step": 406530
    },
    {
      "epoch": 0.06719022311409392,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001187040704070407,
      "loss": 3.1229,
      "step": 406540
    },
    {
      "epoch": 0.06720886635025211,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011870207020702071,
      "loss": 3.2069,
      "step": 406550
    },
    {
      "epoch": 0.06722750958641029,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001187000700070007,
      "loss": 3.0122,
      "step": 406560
    },
    {
      "epoch": 0.06724615282256848,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001186980698069807,
      "loss": 3.1878,
      "step": 406570
    },
    {
      "epoch": 0.06726479605872666,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011869606960696069,
      "loss": 3.1468,
      "step": 406580
    },
    {
      "epoch": 0.06728343929488484,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001186940694069407,
      "loss": 3.0564,
      "step": 406590
    },
    {
      "epoch": 0.06730208253104303,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001186920692069207,
      "loss": 3.2059,
      "step": 406600
    },
    {
      "epoch": 0.06732072576720122,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00011869006900690069,
      "loss": 3.0817,
      "step": 406610
    },
    {
      "epoch": 0.06733936900335939,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001186880688068807,
      "loss": 3.1839,
      "step": 406620
    },
    {
      "epoch": 0.06735801223951758,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011868606860686069,
      "loss": 3.1474,
      "step": 406630
    },
    {
      "epoch": 0.06737665547567576,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001186840684068407,
      "loss": 3.213,
      "step": 406640
    },
    {
      "epoch": 0.06739529871183395,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011868206820682068,
      "loss": 3.1813,
      "step": 406650
    },
    {
      "epoch": 0.06741394194799213,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001186800680068007,
      "loss": 3.1363,
      "step": 406660
    },
    {
      "epoch": 0.06743258518415031,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011867806780678068,
      "loss": 3.1599,
      "step": 406670
    },
    {
      "epoch": 0.0674512284203085,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011867606760676068,
      "loss": 3.1593,
      "step": 406680
    },
    {
      "epoch": 0.06746987165646667,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011867406740674068,
      "loss": 3.1127,
      "step": 406690
    },
    {
      "epoch": 0.06748851489262486,
      "grad_norm": 0.875,
      "learning_rate": 0.00011867206720672069,
      "loss": 3.1925,
      "step": 406700
    },
    {
      "epoch": 0.06750715812878305,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011867006700670067,
      "loss": 3.3163,
      "step": 406710
    },
    {
      "epoch": 0.06752580136494123,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011866806680668068,
      "loss": 3.1032,
      "step": 406720
    },
    {
      "epoch": 0.06754444460109942,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011866606660666066,
      "loss": 3.2708,
      "step": 406730
    },
    {
      "epoch": 0.0675630878372576,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011866406640664067,
      "loss": 3.1891,
      "step": 406740
    },
    {
      "epoch": 0.06758173107341578,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011866206620662065,
      "loss": 3.0264,
      "step": 406750
    },
    {
      "epoch": 0.06760037430957397,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011866006600660066,
      "loss": 3.1889,
      "step": 406760
    },
    {
      "epoch": 0.06761901754573214,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011865806580658066,
      "loss": 3.135,
      "step": 406770
    },
    {
      "epoch": 0.06763766078189033,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011865606560656067,
      "loss": 3.134,
      "step": 406780
    },
    {
      "epoch": 0.06765630401804852,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011865406540654065,
      "loss": 3.13,
      "step": 406790
    },
    {
      "epoch": 0.0676749472542067,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011865206520652066,
      "loss": 3.0938,
      "step": 406800
    },
    {
      "epoch": 0.06769359049036489,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011865006500650067,
      "loss": 3.1017,
      "step": 406810
    },
    {
      "epoch": 0.06771223372652307,
      "grad_norm": 0.75,
      "learning_rate": 0.00011864806480648065,
      "loss": 3.183,
      "step": 406820
    },
    {
      "epoch": 0.06773087696268125,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011864606460646066,
      "loss": 3.1071,
      "step": 406830
    },
    {
      "epoch": 0.06774952019883944,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011864406440644064,
      "loss": 3.1009,
      "step": 406840
    },
    {
      "epoch": 0.06776816343499761,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011864206420642065,
      "loss": 3.2565,
      "step": 406850
    },
    {
      "epoch": 0.0677868066711558,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011864006400640065,
      "loss": 3.3095,
      "step": 406860
    },
    {
      "epoch": 0.06780544990731399,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011863806380638066,
      "loss": 3.2586,
      "step": 406870
    },
    {
      "epoch": 0.06782409314347217,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011863606360636064,
      "loss": 3.2818,
      "step": 406880
    },
    {
      "epoch": 0.06784273637963036,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011863406340634065,
      "loss": 3.2798,
      "step": 406890
    },
    {
      "epoch": 0.06786137961578853,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011863206320632063,
      "loss": 3.2109,
      "step": 406900
    },
    {
      "epoch": 0.06788002285194672,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011863006300630064,
      "loss": 3.1438,
      "step": 406910
    },
    {
      "epoch": 0.06789866608810491,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011862806280628063,
      "loss": 3.0824,
      "step": 406920
    },
    {
      "epoch": 0.06791730932426308,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011862606260626063,
      "loss": 3.0885,
      "step": 406930
    },
    {
      "epoch": 0.06793595256042127,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011862406240624062,
      "loss": 3.1924,
      "step": 406940
    },
    {
      "epoch": 0.06795459579657946,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011862206220622063,
      "loss": 3.219,
      "step": 406950
    },
    {
      "epoch": 0.06797323903273764,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011862006200620062,
      "loss": 3.3054,
      "step": 406960
    },
    {
      "epoch": 0.06799188226889583,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011861806180618062,
      "loss": 3.0479,
      "step": 406970
    },
    {
      "epoch": 0.068010525505054,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001186160616061606,
      "loss": 3.0828,
      "step": 406980
    },
    {
      "epoch": 0.06802916874121219,
      "grad_norm": 0.75,
      "learning_rate": 0.00011861406140614062,
      "loss": 3.1841,
      "step": 406990
    },
    {
      "epoch": 0.06804781197737038,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011861206120612062,
      "loss": 3.2854,
      "step": 407000
    },
    {
      "epoch": 0.06806645521352855,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011861006100610062,
      "loss": 3.153,
      "step": 407010
    },
    {
      "epoch": 0.06808509844968674,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011860806080608063,
      "loss": 3.0501,
      "step": 407020
    },
    {
      "epoch": 0.06810374168584492,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011860606060606061,
      "loss": 3.1997,
      "step": 407030
    },
    {
      "epoch": 0.0681223849220031,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011860406040604062,
      "loss": 3.2907,
      "step": 407040
    },
    {
      "epoch": 0.0681410281581613,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001186020602060206,
      "loss": 3.1839,
      "step": 407050
    },
    {
      "epoch": 0.06815967139431947,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011860006000600061,
      "loss": 3.1548,
      "step": 407060
    },
    {
      "epoch": 0.06817831463047766,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001185980598059806,
      "loss": 3.1483,
      "step": 407070
    },
    {
      "epoch": 0.06819695786663585,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001185960596059606,
      "loss": 3.1937,
      "step": 407080
    },
    {
      "epoch": 0.06821560110279402,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001185940594059406,
      "loss": 3.1783,
      "step": 407090
    },
    {
      "epoch": 0.06823424433895221,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011859205920592061,
      "loss": 3.1672,
      "step": 407100
    },
    {
      "epoch": 0.06825288757511039,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011859005900590059,
      "loss": 3.0964,
      "step": 407110
    },
    {
      "epoch": 0.06827153081126858,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001185880588058806,
      "loss": 3.1333,
      "step": 407120
    },
    {
      "epoch": 0.06829017404742677,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011858605860586058,
      "loss": 3.1469,
      "step": 407130
    },
    {
      "epoch": 0.06830881728358494,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011858405840584059,
      "loss": 2.9877,
      "step": 407140
    },
    {
      "epoch": 0.06832746051974313,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00011858205820582057,
      "loss": 3.0313,
      "step": 407150
    },
    {
      "epoch": 0.06834610375590132,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011858005800580058,
      "loss": 3.191,
      "step": 407160
    },
    {
      "epoch": 0.0683647469920595,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011857805780578058,
      "loss": 3.2005,
      "step": 407170
    },
    {
      "epoch": 0.06838339022821768,
      "grad_norm": 0.75,
      "learning_rate": 0.00011857605760576059,
      "loss": 3.1714,
      "step": 407180
    },
    {
      "epoch": 0.06840203346437586,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001185740574057406,
      "loss": 3.2224,
      "step": 407190
    },
    {
      "epoch": 0.06842067670053405,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011857205720572058,
      "loss": 3.2597,
      "step": 407200
    },
    {
      "epoch": 0.06843931993669224,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011857005700570059,
      "loss": 3.0425,
      "step": 407210
    },
    {
      "epoch": 0.06845796317285041,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011856805680568057,
      "loss": 3.2948,
      "step": 407220
    },
    {
      "epoch": 0.0684766064090086,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011856605660566058,
      "loss": 3.2672,
      "step": 407230
    },
    {
      "epoch": 0.06849524964516678,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011856405640564057,
      "loss": 3.0966,
      "step": 407240
    },
    {
      "epoch": 0.06851389288132496,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011856205620562057,
      "loss": 3.1233,
      "step": 407250
    },
    {
      "epoch": 0.06853253611748315,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011856005600560056,
      "loss": 3.1607,
      "step": 407260
    },
    {
      "epoch": 0.06855117935364133,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011855805580558057,
      "loss": 3.0854,
      "step": 407270
    },
    {
      "epoch": 0.06856982258979952,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011855605560556056,
      "loss": 3.1118,
      "step": 407280
    },
    {
      "epoch": 0.0685884658259577,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011855405540554056,
      "loss": 3.2962,
      "step": 407290
    },
    {
      "epoch": 0.06860710906211588,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011855205520552055,
      "loss": 3.1171,
      "step": 407300
    },
    {
      "epoch": 0.06862575229827407,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011855005500550056,
      "loss": 3.2547,
      "step": 407310
    },
    {
      "epoch": 0.06864439553443225,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011854805480548055,
      "loss": 3.2044,
      "step": 407320
    },
    {
      "epoch": 0.06866303877059043,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011854605460546056,
      "loss": 3.2727,
      "step": 407330
    },
    {
      "epoch": 0.06868168200674862,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011854405440544054,
      "loss": 3.1876,
      "step": 407340
    },
    {
      "epoch": 0.0687003252429068,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011854205420542055,
      "loss": 3.1934,
      "step": 407350
    },
    {
      "epoch": 0.06871896847906499,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011854005400540053,
      "loss": 3.2257,
      "step": 407360
    },
    {
      "epoch": 0.06873761171522318,
      "grad_norm": 0.75,
      "learning_rate": 0.00011853805380538054,
      "loss": 3.1983,
      "step": 407370
    },
    {
      "epoch": 0.06875625495138135,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011853605360536053,
      "loss": 3.0595,
      "step": 407380
    },
    {
      "epoch": 0.06877489818753954,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011853405340534053,
      "loss": 3.1724,
      "step": 407390
    },
    {
      "epoch": 0.06879354142369772,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011853205320532054,
      "loss": 3.2107,
      "step": 407400
    },
    {
      "epoch": 0.0688121846598559,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011853005300530054,
      "loss": 3.2289,
      "step": 407410
    },
    {
      "epoch": 0.0688308278960141,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011852805280528055,
      "loss": 3.25,
      "step": 407420
    },
    {
      "epoch": 0.06884947113217227,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011852605260526053,
      "loss": 3.1343,
      "step": 407430
    },
    {
      "epoch": 0.06886811436833046,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011852405240524054,
      "loss": 3.2442,
      "step": 407440
    },
    {
      "epoch": 0.06888675760448863,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011852205220522052,
      "loss": 3.2459,
      "step": 407450
    },
    {
      "epoch": 0.06890540084064682,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011852005200520053,
      "loss": 3.1299,
      "step": 407460
    },
    {
      "epoch": 0.06892404407680501,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011851805180518051,
      "loss": 3.1877,
      "step": 407470
    },
    {
      "epoch": 0.06894268731296319,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011851605160516052,
      "loss": 3.1766,
      "step": 407480
    },
    {
      "epoch": 0.06896133054912137,
      "grad_norm": 0.75,
      "learning_rate": 0.00011851405140514052,
      "loss": 3.1702,
      "step": 407490
    },
    {
      "epoch": 0.06897997378527956,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011851205120512053,
      "loss": 3.2319,
      "step": 407500
    },
    {
      "epoch": 0.06899861702143774,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011851005100510051,
      "loss": 3.1113,
      "step": 407510
    },
    {
      "epoch": 0.06901726025759593,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011850805080508052,
      "loss": 3.1961,
      "step": 407520
    },
    {
      "epoch": 0.0690359034937541,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001185060506050605,
      "loss": 3.3016,
      "step": 407530
    },
    {
      "epoch": 0.06905454672991229,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011850405040504051,
      "loss": 3.237,
      "step": 407540
    },
    {
      "epoch": 0.06907318996607048,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001185020502050205,
      "loss": 3.1155,
      "step": 407550
    },
    {
      "epoch": 0.06909183320222866,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001185000500050005,
      "loss": 3.148,
      "step": 407560
    },
    {
      "epoch": 0.06911047643838684,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001184980498049805,
      "loss": 3.2598,
      "step": 407570
    },
    {
      "epoch": 0.06912911967454502,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001184960496049605,
      "loss": 3.1493,
      "step": 407580
    },
    {
      "epoch": 0.06914776291070321,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011849404940494051,
      "loss": 3.1764,
      "step": 407590
    },
    {
      "epoch": 0.0691664061468614,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001184920492049205,
      "loss": 3.1584,
      "step": 407600
    },
    {
      "epoch": 0.06918504938301957,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001184900490049005,
      "loss": 3.1841,
      "step": 407610
    },
    {
      "epoch": 0.06920369261917776,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011848804880488049,
      "loss": 3.1835,
      "step": 407620
    },
    {
      "epoch": 0.06922233585533595,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001184860486048605,
      "loss": 3.2362,
      "step": 407630
    },
    {
      "epoch": 0.06924097909149413,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011848404840484049,
      "loss": 3.2428,
      "step": 407640
    },
    {
      "epoch": 0.06925962232765231,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001184820482048205,
      "loss": 3.2854,
      "step": 407650
    },
    {
      "epoch": 0.06927826556381049,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011848004800480048,
      "loss": 3.0647,
      "step": 407660
    },
    {
      "epoch": 0.06929690879996868,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011847804780478049,
      "loss": 3.1289,
      "step": 407670
    },
    {
      "epoch": 0.06931555203612687,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00011847604760476047,
      "loss": 3.202,
      "step": 407680
    },
    {
      "epoch": 0.06933419527228504,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011847404740474048,
      "loss": 3.0908,
      "step": 407690
    },
    {
      "epoch": 0.06935283850844323,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011847204720472047,
      "loss": 3.1796,
      "step": 407700
    },
    {
      "epoch": 0.06937148174460142,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011847004700470047,
      "loss": 3.1171,
      "step": 407710
    },
    {
      "epoch": 0.0693901249807596,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011846804680468047,
      "loss": 3.1558,
      "step": 407720
    },
    {
      "epoch": 0.06940876821691778,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011846604660466048,
      "loss": 3.1931,
      "step": 407730
    },
    {
      "epoch": 0.06942741145307596,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011846404640464046,
      "loss": 3.2445,
      "step": 407740
    },
    {
      "epoch": 0.06944605468923415,
      "grad_norm": 0.75,
      "learning_rate": 0.00011846204620462047,
      "loss": 3.1504,
      "step": 407750
    },
    {
      "epoch": 0.06946469792539234,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00011846004600460045,
      "loss": 3.2228,
      "step": 407760
    },
    {
      "epoch": 0.06948334116155051,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011845804580458046,
      "loss": 3.1292,
      "step": 407770
    },
    {
      "epoch": 0.0695019843977087,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011845604560456047,
      "loss": 3.1535,
      "step": 407780
    },
    {
      "epoch": 0.06952062763386688,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011845404540454045,
      "loss": 3.2496,
      "step": 407790
    },
    {
      "epoch": 0.06953927087002507,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011845204520452046,
      "loss": 3.3106,
      "step": 407800
    },
    {
      "epoch": 0.06955791410618326,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011845004500450046,
      "loss": 3.2478,
      "step": 407810
    },
    {
      "epoch": 0.06957655734234143,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011844804480448047,
      "loss": 3.1863,
      "step": 407820
    },
    {
      "epoch": 0.06959520057849962,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011844604460446045,
      "loss": 3.0802,
      "step": 407830
    },
    {
      "epoch": 0.06961384381465781,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011844404440444046,
      "loss": 3.1537,
      "step": 407840
    },
    {
      "epoch": 0.06963248705081598,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011844204420442044,
      "loss": 3.1363,
      "step": 407850
    },
    {
      "epoch": 0.06965113028697417,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011844004400440045,
      "loss": 3.147,
      "step": 407860
    },
    {
      "epoch": 0.06966977352313235,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011843804380438044,
      "loss": 3.1654,
      "step": 407870
    },
    {
      "epoch": 0.06968841675929054,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011843604360436044,
      "loss": 3.1116,
      "step": 407880
    },
    {
      "epoch": 0.06970705999544873,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011843404340434044,
      "loss": 3.1636,
      "step": 407890
    },
    {
      "epoch": 0.0697257032316069,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011843204320432045,
      "loss": 3.2613,
      "step": 407900
    },
    {
      "epoch": 0.06974434646776509,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011843004300430043,
      "loss": 3.0213,
      "step": 407910
    },
    {
      "epoch": 0.06976298970392326,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011842804280428044,
      "loss": 3.2811,
      "step": 407920
    },
    {
      "epoch": 0.06978163294008145,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011842604260426042,
      "loss": 3.1888,
      "step": 407930
    },
    {
      "epoch": 0.06980027617623964,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011842404240424043,
      "loss": 3.145,
      "step": 407940
    },
    {
      "epoch": 0.06981891941239782,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011842204220422042,
      "loss": 3.1863,
      "step": 407950
    },
    {
      "epoch": 0.069837562648556,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011842004200420043,
      "loss": 3.2687,
      "step": 407960
    },
    {
      "epoch": 0.0698562058847142,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011841804180418041,
      "loss": 3.1652,
      "step": 407970
    },
    {
      "epoch": 0.06987484912087237,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011841604160416042,
      "loss": 3.0949,
      "step": 407980
    },
    {
      "epoch": 0.06989349235703056,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011841404140414043,
      "loss": 3.0754,
      "step": 407990
    },
    {
      "epoch": 0.06991213559318873,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011841204120412041,
      "loss": 3.1353,
      "step": 408000
    },
    {
      "epoch": 0.06993077882934692,
      "grad_norm": 0.75,
      "learning_rate": 0.00011841004100410042,
      "loss": 3.1637,
      "step": 408010
    },
    {
      "epoch": 0.06994942206550511,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001184080408040804,
      "loss": 3.3259,
      "step": 408020
    },
    {
      "epoch": 0.06996806530166329,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011840604060406042,
      "loss": 3.2085,
      "step": 408030
    },
    {
      "epoch": 0.06998670853782148,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011840404040404041,
      "loss": 3.2625,
      "step": 408040
    },
    {
      "epoch": 0.07000535177397967,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011840204020402042,
      "loss": 3.0863,
      "step": 408050
    },
    {
      "epoch": 0.07002399501013784,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001184000400040004,
      "loss": 3.293,
      "step": 408060
    },
    {
      "epoch": 0.07004263824629603,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011839803980398041,
      "loss": 3.0533,
      "step": 408070
    },
    {
      "epoch": 0.0700612814824542,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011839603960396039,
      "loss": 3.2601,
      "step": 408080
    },
    {
      "epoch": 0.0700799247186124,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001183940394039404,
      "loss": 3.1186,
      "step": 408090
    },
    {
      "epoch": 0.07009856795477058,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011839203920392038,
      "loss": 3.1518,
      "step": 408100
    },
    {
      "epoch": 0.07011721119092876,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001183900390039004,
      "loss": 3.1163,
      "step": 408110
    },
    {
      "epoch": 0.07013585442708695,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011838803880388039,
      "loss": 3.1571,
      "step": 408120
    },
    {
      "epoch": 0.07015449766324512,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001183860386038604,
      "loss": 3.1345,
      "step": 408130
    },
    {
      "epoch": 0.07017314089940331,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011838403840384038,
      "loss": 3.2897,
      "step": 408140
    },
    {
      "epoch": 0.0701917841355615,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011838203820382039,
      "loss": 3.0704,
      "step": 408150
    },
    {
      "epoch": 0.07021042737171967,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011838003800380037,
      "loss": 3.3107,
      "step": 408160
    },
    {
      "epoch": 0.07022907060787786,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011837803780378038,
      "loss": 3.1891,
      "step": 408170
    },
    {
      "epoch": 0.07024771384403605,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011837603760376039,
      "loss": 3.2337,
      "step": 408180
    },
    {
      "epoch": 0.07026635708019423,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011837403740374039,
      "loss": 3.1612,
      "step": 408190
    },
    {
      "epoch": 0.07028500031635242,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011837203720372038,
      "loss": 3.1639,
      "step": 408200
    },
    {
      "epoch": 0.07030364355251059,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011837003700370038,
      "loss": 3.0073,
      "step": 408210
    },
    {
      "epoch": 0.07032228678866878,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011836803680368039,
      "loss": 3.1376,
      "step": 408220
    },
    {
      "epoch": 0.07034093002482697,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011836603660366037,
      "loss": 3.3049,
      "step": 408230
    },
    {
      "epoch": 0.07035957326098514,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011836403640364038,
      "loss": 3.1227,
      "step": 408240
    },
    {
      "epoch": 0.07037821649714333,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011836203620362036,
      "loss": 3.2354,
      "step": 408250
    },
    {
      "epoch": 0.07039685973330151,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011836003600360037,
      "loss": 3.1299,
      "step": 408260
    },
    {
      "epoch": 0.0704155029694597,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011835803580358036,
      "loss": 3.0797,
      "step": 408270
    },
    {
      "epoch": 0.07043414620561789,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011835603560356037,
      "loss": 3.0291,
      "step": 408280
    },
    {
      "epoch": 0.07045278944177606,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011835403540354035,
      "loss": 3.2886,
      "step": 408290
    },
    {
      "epoch": 0.07047143267793425,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011835203520352036,
      "loss": 3.1311,
      "step": 408300
    },
    {
      "epoch": 0.07049007591409244,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011835003500350035,
      "loss": 3.0643,
      "step": 408310
    },
    {
      "epoch": 0.07050871915025061,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011834803480348036,
      "loss": 3.2641,
      "step": 408320
    },
    {
      "epoch": 0.0705273623864088,
      "grad_norm": 0.625,
      "learning_rate": 0.00011834603460346034,
      "loss": 3.0155,
      "step": 408330
    },
    {
      "epoch": 0.07054600562256698,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011834403440344035,
      "loss": 3.169,
      "step": 408340
    },
    {
      "epoch": 0.07056464885872517,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011834203420342034,
      "loss": 3.0942,
      "step": 408350
    },
    {
      "epoch": 0.07058329209488336,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011834003400340035,
      "loss": 3.0021,
      "step": 408360
    },
    {
      "epoch": 0.07060193533104153,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011833803380338033,
      "loss": 3.2245,
      "step": 408370
    },
    {
      "epoch": 0.07062057856719972,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011833603360336034,
      "loss": 3.1385,
      "step": 408380
    },
    {
      "epoch": 0.07063922180335791,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011833403340334035,
      "loss": 3.1506,
      "step": 408390
    },
    {
      "epoch": 0.07065786503951609,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011833203320332033,
      "loss": 3.1826,
      "step": 408400
    },
    {
      "epoch": 0.07067650827567427,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011833003300330034,
      "loss": 3.2174,
      "step": 408410
    },
    {
      "epoch": 0.07069515151183245,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011832803280328032,
      "loss": 3.2131,
      "step": 408420
    },
    {
      "epoch": 0.07071379474799064,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011832603260326033,
      "loss": 3.1137,
      "step": 408430
    },
    {
      "epoch": 0.07073243798414883,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011832403240324033,
      "loss": 3.1752,
      "step": 408440
    },
    {
      "epoch": 0.070751081220307,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011832203220322034,
      "loss": 3.0292,
      "step": 408450
    },
    {
      "epoch": 0.07076972445646519,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011832003200320032,
      "loss": 3.0543,
      "step": 408460
    },
    {
      "epoch": 0.07078836769262337,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011831803180318033,
      "loss": 3.1426,
      "step": 408470
    },
    {
      "epoch": 0.07080701092878156,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011831603160316031,
      "loss": 3.2634,
      "step": 408480
    },
    {
      "epoch": 0.07082565416493974,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011831403140314032,
      "loss": 3.2409,
      "step": 408490
    },
    {
      "epoch": 0.07084429740109792,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011831203120312032,
      "loss": 3.2606,
      "step": 408500
    },
    {
      "epoch": 0.07086294063725611,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011831003100310033,
      "loss": 3.1245,
      "step": 408510
    },
    {
      "epoch": 0.0708815838734143,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011830803080308031,
      "loss": 3.1567,
      "step": 408520
    },
    {
      "epoch": 0.07090022710957247,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011830603060306032,
      "loss": 3.1798,
      "step": 408530
    },
    {
      "epoch": 0.07091887034573066,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001183040304030403,
      "loss": 3.2488,
      "step": 408540
    },
    {
      "epoch": 0.07093751358188884,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011830203020302031,
      "loss": 3.1831,
      "step": 408550
    },
    {
      "epoch": 0.07095615681804703,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011830003000300029,
      "loss": 3.126,
      "step": 408560
    },
    {
      "epoch": 0.07097480005420521,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001182980298029803,
      "loss": 3.2617,
      "step": 408570
    },
    {
      "epoch": 0.07099344329036339,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011829602960296031,
      "loss": 3.0794,
      "step": 408580
    },
    {
      "epoch": 0.07101208652652158,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001182940294029403,
      "loss": 3.3023,
      "step": 408590
    },
    {
      "epoch": 0.07103072976267975,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011829202920292031,
      "loss": 3.0904,
      "step": 408600
    },
    {
      "epoch": 0.07104937299883794,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001182900290029003,
      "loss": 3.3049,
      "step": 408610
    },
    {
      "epoch": 0.07106801623499613,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001182880288028803,
      "loss": 3.1488,
      "step": 408620
    },
    {
      "epoch": 0.0710866594711543,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011828602860286029,
      "loss": 3.1029,
      "step": 408630
    },
    {
      "epoch": 0.0711053027073125,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001182840284028403,
      "loss": 3.2074,
      "step": 408640
    },
    {
      "epoch": 0.07112394594347068,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011828202820282028,
      "loss": 3.3033,
      "step": 408650
    },
    {
      "epoch": 0.07114258917962886,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011828002800280029,
      "loss": 3.2208,
      "step": 408660
    },
    {
      "epoch": 0.07116123241578705,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011827802780278028,
      "loss": 3.1695,
      "step": 408670
    },
    {
      "epoch": 0.07117987565194522,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011827602760276029,
      "loss": 3.2315,
      "step": 408680
    },
    {
      "epoch": 0.07119851888810341,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011827402740274027,
      "loss": 3.0394,
      "step": 408690
    },
    {
      "epoch": 0.0712171621242616,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011827202720272028,
      "loss": 3.0558,
      "step": 408700
    },
    {
      "epoch": 0.07123580536041978,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011827002700270026,
      "loss": 3.1241,
      "step": 408710
    },
    {
      "epoch": 0.07125444859657797,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011826802680268027,
      "loss": 3.1684,
      "step": 408720
    },
    {
      "epoch": 0.07127309183273615,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011826602660266026,
      "loss": 3.2134,
      "step": 408730
    },
    {
      "epoch": 0.07129173506889433,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011826402640264026,
      "loss": 3.1334,
      "step": 408740
    },
    {
      "epoch": 0.07131037830505252,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011826202620262026,
      "loss": 3.184,
      "step": 408750
    },
    {
      "epoch": 0.0713290215412107,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011826002600260027,
      "loss": 3.1726,
      "step": 408760
    },
    {
      "epoch": 0.07134766477736888,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011825802580258028,
      "loss": 3.1835,
      "step": 408770
    },
    {
      "epoch": 0.07136630801352707,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011825602560256026,
      "loss": 3.2728,
      "step": 408780
    },
    {
      "epoch": 0.07138495124968525,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011825402540254027,
      "loss": 3.1375,
      "step": 408790
    },
    {
      "epoch": 0.07140359448584344,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011825202520252025,
      "loss": 3.216,
      "step": 408800
    },
    {
      "epoch": 0.07142223772200161,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011825002500250026,
      "loss": 3.1965,
      "step": 408810
    },
    {
      "epoch": 0.0714408809581598,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011824802480248026,
      "loss": 3.2333,
      "step": 408820
    },
    {
      "epoch": 0.07145952419431799,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011824602460246025,
      "loss": 3.1307,
      "step": 408830
    },
    {
      "epoch": 0.07147816743047616,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011824402440244025,
      "loss": 3.1785,
      "step": 408840
    },
    {
      "epoch": 0.07149681066663435,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011824202420242026,
      "loss": 3.0259,
      "step": 408850
    },
    {
      "epoch": 0.07151545390279254,
      "grad_norm": 0.75,
      "learning_rate": 0.00011824002400240024,
      "loss": 3.1624,
      "step": 408860
    },
    {
      "epoch": 0.07153409713895072,
      "grad_norm": 0.75,
      "learning_rate": 0.00011823802380238025,
      "loss": 3.2479,
      "step": 408870
    },
    {
      "epoch": 0.0715527403751089,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011823602360236023,
      "loss": 3.1774,
      "step": 408880
    },
    {
      "epoch": 0.07157138361126708,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011823402340234024,
      "loss": 3.1618,
      "step": 408890
    },
    {
      "epoch": 0.07159002684742527,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011823202320232024,
      "loss": 3.2263,
      "step": 408900
    },
    {
      "epoch": 0.07160867008358346,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011823002300230024,
      "loss": 3.2001,
      "step": 408910
    },
    {
      "epoch": 0.07162731331974163,
      "grad_norm": 0.90625,
      "learning_rate": 0.00011822802280228023,
      "loss": 3.1315,
      "step": 408920
    },
    {
      "epoch": 0.07164595655589982,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011822602260226024,
      "loss": 3.1776,
      "step": 408930
    },
    {
      "epoch": 0.071664599792058,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011822402240224022,
      "loss": 3.2109,
      "step": 408940
    },
    {
      "epoch": 0.07168324302821619,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011822202220222023,
      "loss": 3.1275,
      "step": 408950
    },
    {
      "epoch": 0.07170188626437438,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011822002200220021,
      "loss": 3.2201,
      "step": 408960
    },
    {
      "epoch": 0.07172052950053255,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011821802180218022,
      "loss": 3.1254,
      "step": 408970
    },
    {
      "epoch": 0.07173917273669074,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011821602160216023,
      "loss": 3.1708,
      "step": 408980
    },
    {
      "epoch": 0.07175781597284893,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011821402140214022,
      "loss": 3.2357,
      "step": 408990
    },
    {
      "epoch": 0.0717764592090071,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011821202120212023,
      "loss": 3.1901,
      "step": 409000
    },
    {
      "epoch": 0.07179510244516529,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011821002100210021,
      "loss": 3.1581,
      "step": 409010
    },
    {
      "epoch": 0.07181374568132347,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011820802080208022,
      "loss": 3.1842,
      "step": 409020
    },
    {
      "epoch": 0.07183238891748166,
      "grad_norm": 0.75,
      "learning_rate": 0.0001182060206020602,
      "loss": 3.1864,
      "step": 409030
    },
    {
      "epoch": 0.07185103215363985,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011820402040204021,
      "loss": 3.1512,
      "step": 409040
    },
    {
      "epoch": 0.07186967538979802,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001182020202020202,
      "loss": 3.1766,
      "step": 409050
    },
    {
      "epoch": 0.07188831862595621,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001182000200020002,
      "loss": 3.1875,
      "step": 409060
    },
    {
      "epoch": 0.0719069618621144,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001181980198019802,
      "loss": 3.0959,
      "step": 409070
    },
    {
      "epoch": 0.07192560509827257,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011819601960196021,
      "loss": 3.1263,
      "step": 409080
    },
    {
      "epoch": 0.07194424833443076,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011819401940194019,
      "loss": 3.1534,
      "step": 409090
    },
    {
      "epoch": 0.07196289157058894,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001181920192019202,
      "loss": 3.1608,
      "step": 409100
    },
    {
      "epoch": 0.07198153480674713,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011819001900190018,
      "loss": 3.0156,
      "step": 409110
    },
    {
      "epoch": 0.07200017804290532,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011818801880188019,
      "loss": 3.14,
      "step": 409120
    },
    {
      "epoch": 0.07201882127906349,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011818601860186019,
      "loss": 3.0983,
      "step": 409130
    },
    {
      "epoch": 0.07203746451522168,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001181840184018402,
      "loss": 3.1135,
      "step": 409140
    },
    {
      "epoch": 0.07205610775137986,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011818201820182018,
      "loss": 3.209,
      "step": 409150
    },
    {
      "epoch": 0.07207475098753804,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011818001800180019,
      "loss": 3.306,
      "step": 409160
    },
    {
      "epoch": 0.07209339422369623,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001181780178017802,
      "loss": 3.2735,
      "step": 409170
    },
    {
      "epoch": 0.07211203745985441,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011817601760176018,
      "loss": 3.2098,
      "step": 409180
    },
    {
      "epoch": 0.0721306806960126,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011817401740174019,
      "loss": 3.1681,
      "step": 409190
    },
    {
      "epoch": 0.07214932393217079,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011817201720172017,
      "loss": 3.0624,
      "step": 409200
    },
    {
      "epoch": 0.07216796716832896,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011817001700170018,
      "loss": 3.434,
      "step": 409210
    },
    {
      "epoch": 0.07218661040448715,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011816801680168018,
      "loss": 3.1866,
      "step": 409220
    },
    {
      "epoch": 0.07220525364064533,
      "grad_norm": 0.75,
      "learning_rate": 0.00011816601660166018,
      "loss": 3.0793,
      "step": 409230
    },
    {
      "epoch": 0.07222389687680351,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011816401640164017,
      "loss": 3.1511,
      "step": 409240
    },
    {
      "epoch": 0.0722425401129617,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011816201620162018,
      "loss": 3.2668,
      "step": 409250
    },
    {
      "epoch": 0.07226118334911988,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011816001600160016,
      "loss": 3.1448,
      "step": 409260
    },
    {
      "epoch": 0.07227982658527807,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011815801580158017,
      "loss": 3.2241,
      "step": 409270
    },
    {
      "epoch": 0.07229846982143624,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011815601560156015,
      "loss": 3.2147,
      "step": 409280
    },
    {
      "epoch": 0.07231711305759443,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00011815401540154016,
      "loss": 3.0743,
      "step": 409290
    },
    {
      "epoch": 0.07233575629375262,
      "grad_norm": 0.75,
      "learning_rate": 0.00011815201520152015,
      "loss": 3.1158,
      "step": 409300
    },
    {
      "epoch": 0.0723543995299108,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011815001500150016,
      "loss": 3.2843,
      "step": 409310
    },
    {
      "epoch": 0.07237304276606898,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011814801480148014,
      "loss": 3.024,
      "step": 409320
    },
    {
      "epoch": 0.07239168600222717,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011814601460146015,
      "loss": 3.1516,
      "step": 409330
    },
    {
      "epoch": 0.07241032923838535,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011814401440144014,
      "loss": 3.1958,
      "step": 409340
    },
    {
      "epoch": 0.07242897247454354,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011814201420142015,
      "loss": 3.2487,
      "step": 409350
    },
    {
      "epoch": 0.07244761571070171,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011814001400140015,
      "loss": 3.1583,
      "step": 409360
    },
    {
      "epoch": 0.0724662589468599,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011813801380138014,
      "loss": 3.0934,
      "step": 409370
    },
    {
      "epoch": 0.07248490218301809,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011813601360136015,
      "loss": 3.1521,
      "step": 409380
    },
    {
      "epoch": 0.07250354541917627,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011813401340134014,
      "loss": 3.2397,
      "step": 409390
    },
    {
      "epoch": 0.07252218865533445,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011813201320132015,
      "loss": 3.2529,
      "step": 409400
    },
    {
      "epoch": 0.07254083189149264,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011813001300130013,
      "loss": 3.2044,
      "step": 409410
    },
    {
      "epoch": 0.07255947512765082,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011812801280128014,
      "loss": 3.2239,
      "step": 409420
    },
    {
      "epoch": 0.07257811836380901,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011812601260126012,
      "loss": 3.1818,
      "step": 409430
    },
    {
      "epoch": 0.07259676159996718,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011812401240124013,
      "loss": 3.0893,
      "step": 409440
    },
    {
      "epoch": 0.07261540483612537,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011812201220122013,
      "loss": 3.1843,
      "step": 409450
    },
    {
      "epoch": 0.07263404807228356,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011812001200120014,
      "loss": 3.1076,
      "step": 409460
    },
    {
      "epoch": 0.07265269130844174,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00011811801180118012,
      "loss": 3.1977,
      "step": 409470
    },
    {
      "epoch": 0.07267133454459992,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011811601160116013,
      "loss": 3.0691,
      "step": 409480
    },
    {
      "epoch": 0.0726899777807581,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011811401140114011,
      "loss": 3.1383,
      "step": 409490
    },
    {
      "epoch": 0.07270862101691629,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011811201120112012,
      "loss": 3.0899,
      "step": 409500
    },
    {
      "epoch": 0.07272726425307448,
      "grad_norm": 0.75,
      "learning_rate": 0.0001181100110011001,
      "loss": 3.2078,
      "step": 409510
    },
    {
      "epoch": 0.07274590748923265,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011810801080108011,
      "loss": 3.1381,
      "step": 409520
    },
    {
      "epoch": 0.07276455072539084,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001181060106010601,
      "loss": 3.3143,
      "step": 409530
    },
    {
      "epoch": 0.07278319396154903,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011810401040104012,
      "loss": 3.2603,
      "step": 409540
    },
    {
      "epoch": 0.0728018371977072,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001181020102010201,
      "loss": 3.132,
      "step": 409550
    },
    {
      "epoch": 0.0728204804338654,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011810001000100011,
      "loss": 3.2377,
      "step": 409560
    },
    {
      "epoch": 0.07283912367002357,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011809800980098012,
      "loss": 3.2505,
      "step": 409570
    },
    {
      "epoch": 0.07285776690618176,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001180960096009601,
      "loss": 3.2887,
      "step": 409580
    },
    {
      "epoch": 0.07287641014233995,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011809400940094011,
      "loss": 3.1356,
      "step": 409590
    },
    {
      "epoch": 0.07289505337849812,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011809200920092009,
      "loss": 3.1429,
      "step": 409600
    },
    {
      "epoch": 0.07291369661465631,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001180900090009001,
      "loss": 3.1702,
      "step": 409610
    },
    {
      "epoch": 0.07293233985081449,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001180880088008801,
      "loss": 3.2791,
      "step": 409620
    },
    {
      "epoch": 0.07295098308697268,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001180860086008601,
      "loss": 3.2212,
      "step": 409630
    },
    {
      "epoch": 0.07296962632313087,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011808400840084009,
      "loss": 3.0511,
      "step": 409640
    },
    {
      "epoch": 0.07298826955928904,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001180820082008201,
      "loss": 3.0993,
      "step": 409650
    },
    {
      "epoch": 0.07300691279544723,
      "grad_norm": 0.75,
      "learning_rate": 0.00011808000800080008,
      "loss": 3.1537,
      "step": 409660
    },
    {
      "epoch": 0.07302555603160542,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011807800780078009,
      "loss": 3.2223,
      "step": 409670
    },
    {
      "epoch": 0.0730441992677636,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011807600760076008,
      "loss": 3.1384,
      "step": 409680
    },
    {
      "epoch": 0.07306284250392178,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011807400740074008,
      "loss": 3.2875,
      "step": 409690
    },
    {
      "epoch": 0.07308148574007996,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011807200720072007,
      "loss": 3.0699,
      "step": 409700
    },
    {
      "epoch": 0.07310012897623815,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011807000700070008,
      "loss": 3.2028,
      "step": 409710
    },
    {
      "epoch": 0.07311877221239634,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011806800680068006,
      "loss": 2.9839,
      "step": 409720
    },
    {
      "epoch": 0.07313741544855451,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011806600660066007,
      "loss": 3.1034,
      "step": 409730
    },
    {
      "epoch": 0.0731560586847127,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011806400640064005,
      "loss": 3.2025,
      "step": 409740
    },
    {
      "epoch": 0.07317470192087089,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011806200620062006,
      "loss": 3.1585,
      "step": 409750
    },
    {
      "epoch": 0.07319334515702906,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011806000600060007,
      "loss": 3.1743,
      "step": 409760
    },
    {
      "epoch": 0.07321198839318725,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011805800580058007,
      "loss": 3.1588,
      "step": 409770
    },
    {
      "epoch": 0.07323063162934543,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011805600560056008,
      "loss": 3.223,
      "step": 409780
    },
    {
      "epoch": 0.07324927486550362,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011805400540054006,
      "loss": 3.33,
      "step": 409790
    },
    {
      "epoch": 0.0732679181016618,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011805200520052007,
      "loss": 3.2308,
      "step": 409800
    },
    {
      "epoch": 0.07328656133781998,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011805000500050005,
      "loss": 3.2702,
      "step": 409810
    },
    {
      "epoch": 0.07330520457397817,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011804800480048006,
      "loss": 3.3033,
      "step": 409820
    },
    {
      "epoch": 0.07332384781013634,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011804600460046004,
      "loss": 3.1322,
      "step": 409830
    },
    {
      "epoch": 0.07334249104629453,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011804400440044005,
      "loss": 3.1015,
      "step": 409840
    },
    {
      "epoch": 0.07336113428245272,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011804200420042005,
      "loss": 3.2087,
      "step": 409850
    },
    {
      "epoch": 0.0733797775186109,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011804000400040006,
      "loss": 3.2834,
      "step": 409860
    },
    {
      "epoch": 0.07339842075476909,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011803800380038004,
      "loss": 3.249,
      "step": 409870
    },
    {
      "epoch": 0.07341706399092728,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011803600360036005,
      "loss": 3.2275,
      "step": 409880
    },
    {
      "epoch": 0.07343570722708545,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011803400340034003,
      "loss": 3.1603,
      "step": 409890
    },
    {
      "epoch": 0.07345435046324364,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011803200320032004,
      "loss": 3.3009,
      "step": 409900
    },
    {
      "epoch": 0.07347299369940181,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011803000300030002,
      "loss": 3.2317,
      "step": 409910
    },
    {
      "epoch": 0.07349163693556,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011802800280028003,
      "loss": 3.3154,
      "step": 409920
    },
    {
      "epoch": 0.07351028017171819,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011802600260026003,
      "loss": 3.121,
      "step": 409930
    },
    {
      "epoch": 0.07352892340787637,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011802400240024003,
      "loss": 3.0793,
      "step": 409940
    },
    {
      "epoch": 0.07354756664403456,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011802200220022004,
      "loss": 3.2203,
      "step": 409950
    },
    {
      "epoch": 0.07356620988019273,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011802000200020003,
      "loss": 3.0991,
      "step": 409960
    },
    {
      "epoch": 0.07358485311635092,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011801800180018003,
      "loss": 3.0578,
      "step": 409970
    },
    {
      "epoch": 0.07360349635250911,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011801600160016002,
      "loss": 3.0123,
      "step": 409980
    },
    {
      "epoch": 0.07362213958866728,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011801400140014003,
      "loss": 3.2394,
      "step": 409990
    },
    {
      "epoch": 0.07364078282482547,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011801200120012002,
      "loss": 3.2477,
      "step": 410000
    },
    {
      "epoch": 0.07365942606098366,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011801000100010002,
      "loss": 3.1432,
      "step": 410010
    },
    {
      "epoch": 0.07367806929714184,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011800800080008001,
      "loss": 3.226,
      "step": 410020
    },
    {
      "epoch": 0.07369671253330003,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011800600060006002,
      "loss": 3.1504,
      "step": 410030
    },
    {
      "epoch": 0.0737153557694582,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011800400040004,
      "loss": 3.1656,
      "step": 410040
    },
    {
      "epoch": 0.07373399900561639,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011800200020002001,
      "loss": 3.0497,
      "step": 410050
    },
    {
      "epoch": 0.07375264224177458,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000118,
      "loss": 3.1368,
      "step": 410060
    },
    {
      "epoch": 0.07377128547793275,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011799799979998,
      "loss": 3.0679,
      "step": 410070
    },
    {
      "epoch": 0.07378992871409094,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011799599959996,
      "loss": 3.1838,
      "step": 410080
    },
    {
      "epoch": 0.07380857195024913,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011799399939994001,
      "loss": 3.1941,
      "step": 410090
    },
    {
      "epoch": 0.07382721518640731,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011799199919991999,
      "loss": 2.9538,
      "step": 410100
    },
    {
      "epoch": 0.0738458584225655,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001179899989999,
      "loss": 3.1401,
      "step": 410110
    },
    {
      "epoch": 0.07386450165872367,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011798799879987998,
      "loss": 3.2557,
      "step": 410120
    },
    {
      "epoch": 0.07388314489488186,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011798599859985999,
      "loss": 3.221,
      "step": 410130
    },
    {
      "epoch": 0.07390178813104005,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011798399839983997,
      "loss": 3.171,
      "step": 410140
    },
    {
      "epoch": 0.07392043136719822,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011798199819981998,
      "loss": 3.206,
      "step": 410150
    },
    {
      "epoch": 0.07393907460335641,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011797999799979999,
      "loss": 3.3043,
      "step": 410160
    },
    {
      "epoch": 0.07395771783951459,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011797799779977999,
      "loss": 3.0983,
      "step": 410170
    },
    {
      "epoch": 0.07397636107567278,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011797599759976,
      "loss": 3.1386,
      "step": 410180
    },
    {
      "epoch": 0.07399500431183097,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011797399739973998,
      "loss": 3.0402,
      "step": 410190
    },
    {
      "epoch": 0.07401364754798914,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011797199719971999,
      "loss": 3.1674,
      "step": 410200
    },
    {
      "epoch": 0.07403229078414733,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011796999699969997,
      "loss": 3.2382,
      "step": 410210
    },
    {
      "epoch": 0.07405093402030552,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011796799679967998,
      "loss": 3.3224,
      "step": 410220
    },
    {
      "epoch": 0.0740695772564637,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011796599659965996,
      "loss": 3.2132,
      "step": 410230
    },
    {
      "epoch": 0.07408822049262188,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011796399639963997,
      "loss": 3.0456,
      "step": 410240
    },
    {
      "epoch": 0.07410686372878006,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011796199619961997,
      "loss": 3.1092,
      "step": 410250
    },
    {
      "epoch": 0.07412550696493825,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011795999599959997,
      "loss": 3.2039,
      "step": 410260
    },
    {
      "epoch": 0.07414415020109644,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011795799579957996,
      "loss": 3.2512,
      "step": 410270
    },
    {
      "epoch": 0.07416279343725461,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011795599559955997,
      "loss": 3.2453,
      "step": 410280
    },
    {
      "epoch": 0.0741814366734128,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011795399539953995,
      "loss": 3.1828,
      "step": 410290
    },
    {
      "epoch": 0.07420007990957098,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011795199519951996,
      "loss": 3.1986,
      "step": 410300
    },
    {
      "epoch": 0.07421872314572917,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011794999499949995,
      "loss": 3.1108,
      "step": 410310
    },
    {
      "epoch": 0.07423736638188735,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011794799479947995,
      "loss": 3.1765,
      "step": 410320
    },
    {
      "epoch": 0.07425600961804553,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011794599459945994,
      "loss": 3.2611,
      "step": 410330
    },
    {
      "epoch": 0.07427465285420372,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011794399439943995,
      "loss": 3.1317,
      "step": 410340
    },
    {
      "epoch": 0.07429329609036191,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011794199419941996,
      "loss": 3.1319,
      "step": 410350
    },
    {
      "epoch": 0.07431193932652008,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011793999399939994,
      "loss": 3.1563,
      "step": 410360
    },
    {
      "epoch": 0.07433058256267827,
      "grad_norm": 0.75,
      "learning_rate": 0.00011793799379937995,
      "loss": 3.1747,
      "step": 410370
    },
    {
      "epoch": 0.07434922579883645,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011793599359935994,
      "loss": 3.0661,
      "step": 410380
    },
    {
      "epoch": 0.07436786903499464,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011793399339933994,
      "loss": 3.0878,
      "step": 410390
    },
    {
      "epoch": 0.07438651227115282,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011793199319931994,
      "loss": 3.1891,
      "step": 410400
    },
    {
      "epoch": 0.074405155507311,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011792999299929995,
      "loss": 3.0079,
      "step": 410410
    },
    {
      "epoch": 0.07442379874346919,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011792799279927993,
      "loss": 3.0722,
      "step": 410420
    },
    {
      "epoch": 0.07444244197962738,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011792599259925994,
      "loss": 3.0838,
      "step": 410430
    },
    {
      "epoch": 0.07446108521578555,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011792399239923992,
      "loss": 3.1357,
      "step": 410440
    },
    {
      "epoch": 0.07447972845194374,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011792199219921993,
      "loss": 3.2946,
      "step": 410450
    },
    {
      "epoch": 0.07449837168810192,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011791999199919991,
      "loss": 3.2263,
      "step": 410460
    },
    {
      "epoch": 0.0745170149242601,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011791799179917992,
      "loss": 3.0689,
      "step": 410470
    },
    {
      "epoch": 0.0745356581604183,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011791599159915992,
      "loss": 3.1676,
      "step": 410480
    },
    {
      "epoch": 0.07455430139657647,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011791399139913993,
      "loss": 3.2215,
      "step": 410490
    },
    {
      "epoch": 0.07457294463273466,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011791199119911991,
      "loss": 3.1367,
      "step": 410500
    },
    {
      "epoch": 0.07459158786889283,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011790999099909992,
      "loss": 3.0727,
      "step": 410510
    },
    {
      "epoch": 0.07461023110505102,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001179079907990799,
      "loss": 3.1297,
      "step": 410520
    },
    {
      "epoch": 0.07462887434120921,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011790599059905991,
      "loss": 3.0848,
      "step": 410530
    },
    {
      "epoch": 0.07464751757736739,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011790399039903989,
      "loss": 3.0904,
      "step": 410540
    },
    {
      "epoch": 0.07466616081352558,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001179019901990199,
      "loss": 3.2062,
      "step": 410550
    },
    {
      "epoch": 0.07468480404968376,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011789998999899991,
      "loss": 3.1037,
      "step": 410560
    },
    {
      "epoch": 0.07470344728584194,
      "grad_norm": 0.75,
      "learning_rate": 0.0001178979897989799,
      "loss": 3.2291,
      "step": 410570
    },
    {
      "epoch": 0.07472209052200013,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011789598959895992,
      "loss": 3.1099,
      "step": 410580
    },
    {
      "epoch": 0.0747407337581583,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001178939893989399,
      "loss": 3.1665,
      "step": 410590
    },
    {
      "epoch": 0.07475937699431649,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001178919891989199,
      "loss": 3.2512,
      "step": 410600
    },
    {
      "epoch": 0.07477802023047468,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011788998899889989,
      "loss": 3.1811,
      "step": 410610
    },
    {
      "epoch": 0.07479666346663286,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001178879887988799,
      "loss": 3.2305,
      "step": 410620
    },
    {
      "epoch": 0.07481530670279105,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011788598859885989,
      "loss": 3.0632,
      "step": 410630
    },
    {
      "epoch": 0.07483394993894922,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011788398839883989,
      "loss": 3.1155,
      "step": 410640
    },
    {
      "epoch": 0.07485259317510741,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011788198819881988,
      "loss": 3.0813,
      "step": 410650
    },
    {
      "epoch": 0.0748712364112656,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001178799879987999,
      "loss": 3.1145,
      "step": 410660
    },
    {
      "epoch": 0.07488987964742377,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011787798779877988,
      "loss": 3.19,
      "step": 410670
    },
    {
      "epoch": 0.07490852288358196,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011787598759875988,
      "loss": 3.2439,
      "step": 410680
    },
    {
      "epoch": 0.07492716611974015,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011787398739873987,
      "loss": 3.1167,
      "step": 410690
    },
    {
      "epoch": 0.07494580935589833,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011787198719871988,
      "loss": 3.2632,
      "step": 410700
    },
    {
      "epoch": 0.07496445259205652,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011786998699869987,
      "loss": 3.1033,
      "step": 410710
    },
    {
      "epoch": 0.07498309582821469,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011786798679867988,
      "loss": 3.2372,
      "step": 410720
    },
    {
      "epoch": 0.07500173906437288,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011786598659865986,
      "loss": 3.2321,
      "step": 410730
    },
    {
      "epoch": 0.07502038230053107,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011786398639863987,
      "loss": 3.185,
      "step": 410740
    },
    {
      "epoch": 0.07503902553668924,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011786198619861988,
      "loss": 3.1268,
      "step": 410750
    },
    {
      "epoch": 0.07505766877284743,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011785998599859986,
      "loss": 3.2044,
      "step": 410760
    },
    {
      "epoch": 0.07507631200900562,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011785798579857987,
      "loss": 3.129,
      "step": 410770
    },
    {
      "epoch": 0.0750949552451638,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011785598559855985,
      "loss": 3.0646,
      "step": 410780
    },
    {
      "epoch": 0.07511359848132199,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011785398539853986,
      "loss": 3.3025,
      "step": 410790
    },
    {
      "epoch": 0.07513224171748016,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011785198519851986,
      "loss": 3.189,
      "step": 410800
    },
    {
      "epoch": 0.07515088495363835,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011784998499849987,
      "loss": 3.0485,
      "step": 410810
    },
    {
      "epoch": 0.07516952818979654,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011784798479847985,
      "loss": 3.1185,
      "step": 410820
    },
    {
      "epoch": 0.07518817142595471,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011784598459845986,
      "loss": 3.31,
      "step": 410830
    },
    {
      "epoch": 0.0752068146621129,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011784398439843984,
      "loss": 3.236,
      "step": 410840
    },
    {
      "epoch": 0.07522545789827108,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011784198419841985,
      "loss": 3.1112,
      "step": 410850
    },
    {
      "epoch": 0.07524410113442927,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011783998399839983,
      "loss": 3.1442,
      "step": 410860
    },
    {
      "epoch": 0.07526274437058746,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011783798379837984,
      "loss": 3.1969,
      "step": 410870
    },
    {
      "epoch": 0.07528138760674563,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011783598359835984,
      "loss": 3.1809,
      "step": 410880
    },
    {
      "epoch": 0.07530003084290382,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011783398339833985,
      "loss": 3.1343,
      "step": 410890
    },
    {
      "epoch": 0.07531867407906201,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011783198319831983,
      "loss": 3.176,
      "step": 410900
    },
    {
      "epoch": 0.07533731731522018,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011782998299829984,
      "loss": 3.2495,
      "step": 410910
    },
    {
      "epoch": 0.07535596055137837,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011782798279827982,
      "loss": 3.3397,
      "step": 410920
    },
    {
      "epoch": 0.07537460378753655,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011782598259825983,
      "loss": 3.0597,
      "step": 410930
    },
    {
      "epoch": 0.07539324702369474,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011782398239823984,
      "loss": 3.1506,
      "step": 410940
    },
    {
      "epoch": 0.07541189025985293,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011782198219821983,
      "loss": 3.0571,
      "step": 410950
    },
    {
      "epoch": 0.0754305334960111,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011781998199819983,
      "loss": 3.1116,
      "step": 410960
    },
    {
      "epoch": 0.07544917673216929,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011781798179817982,
      "loss": 3.0826,
      "step": 410970
    },
    {
      "epoch": 0.07546781996832747,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011781598159815983,
      "loss": 3.1254,
      "step": 410980
    },
    {
      "epoch": 0.07548646320448565,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011781398139813982,
      "loss": 3.0401,
      "step": 410990
    },
    {
      "epoch": 0.07550510644064384,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011781198119811982,
      "loss": 3.1908,
      "step": 411000
    },
    {
      "epoch": 0.07552374967680202,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011780998099809981,
      "loss": 3.1203,
      "step": 411010
    },
    {
      "epoch": 0.07554239291296021,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011780798079807982,
      "loss": 3.1856,
      "step": 411020
    },
    {
      "epoch": 0.0755610361491184,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011780598059805981,
      "loss": 3.1366,
      "step": 411030
    },
    {
      "epoch": 0.07557967938527657,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011780398039803982,
      "loss": 3.1639,
      "step": 411040
    },
    {
      "epoch": 0.07559832262143476,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001178019801980198,
      "loss": 3.1424,
      "step": 411050
    },
    {
      "epoch": 0.07561696585759294,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011779997999799981,
      "loss": 3.0769,
      "step": 411060
    },
    {
      "epoch": 0.07563560909375112,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001177979797979798,
      "loss": 3.1456,
      "step": 411070
    },
    {
      "epoch": 0.07565425232990931,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001177959795979598,
      "loss": 3.1592,
      "step": 411080
    },
    {
      "epoch": 0.07567289556606749,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011779397939793979,
      "loss": 3.1601,
      "step": 411090
    },
    {
      "epoch": 0.07569153880222568,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001177919791979198,
      "loss": 3.1851,
      "step": 411100
    },
    {
      "epoch": 0.07571018203838387,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011778997899789979,
      "loss": 3.1972,
      "step": 411110
    },
    {
      "epoch": 0.07572882527454204,
      "grad_norm": 0.75,
      "learning_rate": 0.0001177879787978798,
      "loss": 3.1237,
      "step": 411120
    },
    {
      "epoch": 0.07574746851070023,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011778597859785978,
      "loss": 3.1257,
      "step": 411130
    },
    {
      "epoch": 0.0757661117468584,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011778397839783979,
      "loss": 3.244,
      "step": 411140
    },
    {
      "epoch": 0.0757847549830166,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001177819781978198,
      "loss": 3.0857,
      "step": 411150
    },
    {
      "epoch": 0.07580339821917478,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011777997799779978,
      "loss": 3.0331,
      "step": 411160
    },
    {
      "epoch": 0.07582204145533296,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011777797779777979,
      "loss": 3.1138,
      "step": 411170
    },
    {
      "epoch": 0.07584068469149115,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011777597759775977,
      "loss": 3.163,
      "step": 411180
    },
    {
      "epoch": 0.07585932792764932,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011777397739773978,
      "loss": 3.1243,
      "step": 411190
    },
    {
      "epoch": 0.07587797116380751,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011777197719771978,
      "loss": 3.1462,
      "step": 411200
    },
    {
      "epoch": 0.0758966143999657,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011776997699769979,
      "loss": 3.1314,
      "step": 411210
    },
    {
      "epoch": 0.07591525763612388,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011776797679767977,
      "loss": 3.113,
      "step": 411220
    },
    {
      "epoch": 0.07593390087228206,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011776597659765978,
      "loss": 3.176,
      "step": 411230
    },
    {
      "epoch": 0.07595254410844025,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011776397639763976,
      "loss": 3.2073,
      "step": 411240
    },
    {
      "epoch": 0.07597118734459843,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011776197619761977,
      "loss": 3.1572,
      "step": 411250
    },
    {
      "epoch": 0.07598983058075662,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011775997599759976,
      "loss": 3.1943,
      "step": 411260
    },
    {
      "epoch": 0.07600847381691479,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011775797579757977,
      "loss": 3.1212,
      "step": 411270
    },
    {
      "epoch": 0.07602711705307298,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011775597559755976,
      "loss": 3.2961,
      "step": 411280
    },
    {
      "epoch": 0.07604576028923117,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011775397539753976,
      "loss": 3.1475,
      "step": 411290
    },
    {
      "epoch": 0.07606440352538935,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011775197519751975,
      "loss": 3.1155,
      "step": 411300
    },
    {
      "epoch": 0.07608304676154753,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011774997499749976,
      "loss": 3.2525,
      "step": 411310
    },
    {
      "epoch": 0.07610168999770571,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011774797479747974,
      "loss": 3.2551,
      "step": 411320
    },
    {
      "epoch": 0.0761203332338639,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011774597459745975,
      "loss": 3.0882,
      "step": 411330
    },
    {
      "epoch": 0.07613897647002209,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011774397439743976,
      "loss": 3.2367,
      "step": 411340
    },
    {
      "epoch": 0.07615761970618026,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011774197419741975,
      "loss": 3.207,
      "step": 411350
    },
    {
      "epoch": 0.07617626294233845,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011773997399739976,
      "loss": 3.2527,
      "step": 411360
    },
    {
      "epoch": 0.07619490617849664,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011773797379737974,
      "loss": 3.2377,
      "step": 411370
    },
    {
      "epoch": 0.07621354941465482,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011773597359735975,
      "loss": 3.0801,
      "step": 411380
    },
    {
      "epoch": 0.076232192650813,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011773397339733973,
      "loss": 3.1962,
      "step": 411390
    },
    {
      "epoch": 0.07625083588697118,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011773197319731974,
      "loss": 2.9599,
      "step": 411400
    },
    {
      "epoch": 0.07626947912312937,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011772997299729973,
      "loss": 3.1535,
      "step": 411410
    },
    {
      "epoch": 0.07628812235928756,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011772797279727973,
      "loss": 3.2195,
      "step": 411420
    },
    {
      "epoch": 0.07630676559544573,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011772597259725973,
      "loss": 3.1554,
      "step": 411430
    },
    {
      "epoch": 0.07632540883160392,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011772397239723974,
      "loss": 3.1689,
      "step": 411440
    },
    {
      "epoch": 0.07634405206776211,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011772197219721972,
      "loss": 3.134,
      "step": 411450
    },
    {
      "epoch": 0.07636269530392029,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011771997199719973,
      "loss": 3.115,
      "step": 411460
    },
    {
      "epoch": 0.07638133854007848,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011771797179717971,
      "loss": 3.151,
      "step": 411470
    },
    {
      "epoch": 0.07639998177623665,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011771597159715972,
      "loss": 3.1714,
      "step": 411480
    },
    {
      "epoch": 0.07641862501239484,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011771397139713972,
      "loss": 3.1754,
      "step": 411490
    },
    {
      "epoch": 0.07643726824855303,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011771197119711971,
      "loss": 3.1174,
      "step": 411500
    },
    {
      "epoch": 0.0764559114847112,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011770997099709971,
      "loss": 3.1913,
      "step": 411510
    },
    {
      "epoch": 0.07647455472086939,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011770797079707972,
      "loss": 3.1001,
      "step": 411520
    },
    {
      "epoch": 0.07649319795702757,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011770597059705973,
      "loss": 3.0903,
      "step": 411530
    },
    {
      "epoch": 0.07651184119318576,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011770397039703971,
      "loss": 3.0235,
      "step": 411540
    },
    {
      "epoch": 0.07653048442934395,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011770197019701972,
      "loss": 3.0873,
      "step": 411550
    },
    {
      "epoch": 0.07654912766550212,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001176999699969997,
      "loss": 3.1899,
      "step": 411560
    },
    {
      "epoch": 0.07656777090166031,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011769796979697971,
      "loss": 3.1771,
      "step": 411570
    },
    {
      "epoch": 0.0765864141378185,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001176959695969597,
      "loss": 3.2525,
      "step": 411580
    },
    {
      "epoch": 0.07660505737397667,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011769396939693971,
      "loss": 3.1874,
      "step": 411590
    },
    {
      "epoch": 0.07662370061013486,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001176919691969197,
      "loss": 3.1999,
      "step": 411600
    },
    {
      "epoch": 0.07664234384629304,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001176899689968997,
      "loss": 3.2414,
      "step": 411610
    },
    {
      "epoch": 0.07666098708245123,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011768796879687969,
      "loss": 3.1073,
      "step": 411620
    },
    {
      "epoch": 0.07667963031860942,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001176859685968597,
      "loss": 3.1126,
      "step": 411630
    },
    {
      "epoch": 0.07669827355476759,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011768396839683968,
      "loss": 3.1372,
      "step": 411640
    },
    {
      "epoch": 0.07671691679092578,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011768196819681969,
      "loss": 3.1566,
      "step": 411650
    },
    {
      "epoch": 0.07673556002708395,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011767996799679968,
      "loss": 3.2177,
      "step": 411660
    },
    {
      "epoch": 0.07675420326324214,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00011767796779677969,
      "loss": 3.1403,
      "step": 411670
    },
    {
      "epoch": 0.07677284649940033,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011767596759675967,
      "loss": 3.0752,
      "step": 411680
    },
    {
      "epoch": 0.07679148973555851,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011767396739673968,
      "loss": 3.2005,
      "step": 411690
    },
    {
      "epoch": 0.0768101329717167,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011767196719671967,
      "loss": 3.1784,
      "step": 411700
    },
    {
      "epoch": 0.07682877620787489,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011766996699669967,
      "loss": 3.1698,
      "step": 411710
    },
    {
      "epoch": 0.07684741944403306,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011766796679667966,
      "loss": 3.1985,
      "step": 411720
    },
    {
      "epoch": 0.07686606268019125,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011766596659665967,
      "loss": 3.3373,
      "step": 411730
    },
    {
      "epoch": 0.07688470591634942,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011766396639663968,
      "loss": 3.2658,
      "step": 411740
    },
    {
      "epoch": 0.07690334915250761,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011766196619661967,
      "loss": 3.1233,
      "step": 411750
    },
    {
      "epoch": 0.0769219923886658,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011765996599659968,
      "loss": 3.2298,
      "step": 411760
    },
    {
      "epoch": 0.07694063562482398,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011765796579657966,
      "loss": 3.1118,
      "step": 411770
    },
    {
      "epoch": 0.07695927886098217,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011765596559655967,
      "loss": 3.2009,
      "step": 411780
    },
    {
      "epoch": 0.07697792209714036,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011765396539653965,
      "loss": 3.2122,
      "step": 411790
    },
    {
      "epoch": 0.07699656533329853,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00011765196519651966,
      "loss": 3.0488,
      "step": 411800
    },
    {
      "epoch": 0.07701520856945672,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011764996499649964,
      "loss": 3.1786,
      "step": 411810
    },
    {
      "epoch": 0.0770338518056149,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011764796479647965,
      "loss": 3.1396,
      "step": 411820
    },
    {
      "epoch": 0.07705249504177308,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011764596459645965,
      "loss": 3.1604,
      "step": 411830
    },
    {
      "epoch": 0.07707113827793127,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011764396439643966,
      "loss": 3.1343,
      "step": 411840
    },
    {
      "epoch": 0.07708978151408945,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011764196419641964,
      "loss": 3.1825,
      "step": 411850
    },
    {
      "epoch": 0.07710842475024764,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011763996399639965,
      "loss": 3.2181,
      "step": 411860
    },
    {
      "epoch": 0.07712706798640581,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011763796379637963,
      "loss": 3.0616,
      "step": 411870
    },
    {
      "epoch": 0.077145711222564,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011763596359635964,
      "loss": 3.0312,
      "step": 411880
    },
    {
      "epoch": 0.07716435445872219,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011763396339633964,
      "loss": 3.0905,
      "step": 411890
    },
    {
      "epoch": 0.07718299769488036,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011763196319631965,
      "loss": 3.2358,
      "step": 411900
    },
    {
      "epoch": 0.07720164093103855,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011762996299629963,
      "loss": 3.2052,
      "step": 411910
    },
    {
      "epoch": 0.07722028416719674,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011762796279627964,
      "loss": 3.187,
      "step": 411920
    },
    {
      "epoch": 0.07723892740335492,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011762596259625965,
      "loss": 3.1836,
      "step": 411930
    },
    {
      "epoch": 0.0772575706395131,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011762396239623963,
      "loss": 3.1945,
      "step": 411940
    },
    {
      "epoch": 0.07727621387567128,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011762196219621964,
      "loss": 3.1624,
      "step": 411950
    },
    {
      "epoch": 0.07729485711182947,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011761996199619962,
      "loss": 3.1928,
      "step": 411960
    },
    {
      "epoch": 0.07731350034798766,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011761796179617963,
      "loss": 3.0397,
      "step": 411970
    },
    {
      "epoch": 0.07733214358414583,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011761596159615962,
      "loss": 3.1768,
      "step": 411980
    },
    {
      "epoch": 0.07735078682030402,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011761396139613963,
      "loss": 3.2077,
      "step": 411990
    },
    {
      "epoch": 0.07736943005646221,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011761196119611961,
      "loss": 3.1317,
      "step": 412000
    },
    {
      "epoch": 0.07738807329262039,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011760996099609962,
      "loss": 3.1105,
      "step": 412010
    },
    {
      "epoch": 0.07740671652877858,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001176079607960796,
      "loss": 3.1869,
      "step": 412020
    },
    {
      "epoch": 0.07742535976493675,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011760596059605962,
      "loss": 3.2451,
      "step": 412030
    },
    {
      "epoch": 0.07744400300109494,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001176039603960396,
      "loss": 3.1057,
      "step": 412040
    },
    {
      "epoch": 0.07746264623725313,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001176019601960196,
      "loss": 3.0755,
      "step": 412050
    },
    {
      "epoch": 0.0774812894734113,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001175999599959996,
      "loss": 3.1844,
      "step": 412060
    },
    {
      "epoch": 0.0774999327095695,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011759795979597961,
      "loss": 3.108,
      "step": 412070
    },
    {
      "epoch": 0.07751857594572767,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011759595959595959,
      "loss": 3.2097,
      "step": 412080
    },
    {
      "epoch": 0.07753721918188586,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001175939593959396,
      "loss": 3.2287,
      "step": 412090
    },
    {
      "epoch": 0.07755586241804405,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011759195919591958,
      "loss": 3.1613,
      "step": 412100
    },
    {
      "epoch": 0.07757450565420222,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001175899589958996,
      "loss": 3.248,
      "step": 412110
    },
    {
      "epoch": 0.07759314889036041,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011758795879587959,
      "loss": 3.1604,
      "step": 412120
    },
    {
      "epoch": 0.0776117921265186,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011758595859585958,
      "loss": 3.1198,
      "step": 412130
    },
    {
      "epoch": 0.07763043536267678,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001175839583958396,
      "loss": 3.1638,
      "step": 412140
    },
    {
      "epoch": 0.07764907859883496,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011758195819581959,
      "loss": 3.0962,
      "step": 412150
    },
    {
      "epoch": 0.07766772183499314,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001175799579957996,
      "loss": 3.1805,
      "step": 412160
    },
    {
      "epoch": 0.07768636507115133,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011757795779577958,
      "loss": 3.2234,
      "step": 412170
    },
    {
      "epoch": 0.07770500830730952,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011757595759575959,
      "loss": 3.2478,
      "step": 412180
    },
    {
      "epoch": 0.07772365154346769,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011757395739573957,
      "loss": 3.1443,
      "step": 412190
    },
    {
      "epoch": 0.07774229477962588,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011757195719571958,
      "loss": 3.1888,
      "step": 412200
    },
    {
      "epoch": 0.07776093801578406,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011756995699569958,
      "loss": 3.3585,
      "step": 412210
    },
    {
      "epoch": 0.07777958125194225,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011756795679567959,
      "loss": 3.1329,
      "step": 412220
    },
    {
      "epoch": 0.07779822448810043,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011756595659565957,
      "loss": 3.2503,
      "step": 412230
    },
    {
      "epoch": 0.07781686772425861,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011756395639563958,
      "loss": 3.2009,
      "step": 412240
    },
    {
      "epoch": 0.0778355109604168,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011756195619561956,
      "loss": 3.17,
      "step": 412250
    },
    {
      "epoch": 0.07785415419657499,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011755995599559957,
      "loss": 3.2357,
      "step": 412260
    },
    {
      "epoch": 0.07787279743273316,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011755795579557955,
      "loss": 3.1566,
      "step": 412270
    },
    {
      "epoch": 0.07789144066889135,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011755595559555956,
      "loss": 2.9981,
      "step": 412280
    },
    {
      "epoch": 0.07791008390504953,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011755395539553955,
      "loss": 3.0993,
      "step": 412290
    },
    {
      "epoch": 0.07792872714120772,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011755195519551956,
      "loss": 3.1376,
      "step": 412300
    },
    {
      "epoch": 0.0779473703773659,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011754995499549955,
      "loss": 3.2022,
      "step": 412310
    },
    {
      "epoch": 0.07796601361352408,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011754795479547956,
      "loss": 3.1533,
      "step": 412320
    },
    {
      "epoch": 0.07798465684968227,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011754595459545956,
      "loss": 3.231,
      "step": 412330
    },
    {
      "epoch": 0.07800330008584046,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011754395439543955,
      "loss": 3.2488,
      "step": 412340
    },
    {
      "epoch": 0.07802194332199863,
      "grad_norm": 0.75,
      "learning_rate": 0.00011754195419541956,
      "loss": 3.0938,
      "step": 412350
    },
    {
      "epoch": 0.07804058655815682,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011753995399539954,
      "loss": 3.1377,
      "step": 412360
    },
    {
      "epoch": 0.078059229794315,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011753795379537955,
      "loss": 3.1886,
      "step": 412370
    },
    {
      "epoch": 0.07807787303047319,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011753595359535954,
      "loss": 3.1885,
      "step": 412380
    },
    {
      "epoch": 0.07809651626663137,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011753395339533955,
      "loss": 3.1873,
      "step": 412390
    },
    {
      "epoch": 0.07811515950278955,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011753195319531953,
      "loss": 3.1392,
      "step": 412400
    },
    {
      "epoch": 0.07813380273894774,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011752995299529954,
      "loss": 3.231,
      "step": 412410
    },
    {
      "epoch": 0.07815244597510591,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011752795279527952,
      "loss": 3.0544,
      "step": 412420
    },
    {
      "epoch": 0.0781710892112641,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00011752595259525953,
      "loss": 3.2252,
      "step": 412430
    },
    {
      "epoch": 0.07818973244742229,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011752395239523953,
      "loss": 3.2996,
      "step": 412440
    },
    {
      "epoch": 0.07820837568358047,
      "grad_norm": 0.75,
      "learning_rate": 0.00011752195219521952,
      "loss": 3.0813,
      "step": 412450
    },
    {
      "epoch": 0.07822701891973866,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011751995199519952,
      "loss": 3.1759,
      "step": 412460
    },
    {
      "epoch": 0.07824566215589684,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011751795179517953,
      "loss": 3.1278,
      "step": 412470
    },
    {
      "epoch": 0.07826430539205502,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011751595159515951,
      "loss": 3.2012,
      "step": 412480
    },
    {
      "epoch": 0.07828294862821321,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011751395139513952,
      "loss": 3.3264,
      "step": 412490
    },
    {
      "epoch": 0.07830159186437138,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001175119511951195,
      "loss": 3.2621,
      "step": 412500
    },
    {
      "epoch": 0.07832023510052957,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011750995099509951,
      "loss": 3.1005,
      "step": 412510
    },
    {
      "epoch": 0.07833887833668776,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011750795079507952,
      "loss": 3.2174,
      "step": 412520
    },
    {
      "epoch": 0.07835752157284594,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011750595059505952,
      "loss": 3.1141,
      "step": 412530
    },
    {
      "epoch": 0.07837616480900413,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011750395039503953,
      "loss": 3.1075,
      "step": 412540
    },
    {
      "epoch": 0.0783948080451623,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011750195019501951,
      "loss": 3.1672,
      "step": 412550
    },
    {
      "epoch": 0.07841345128132049,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011749994999499952,
      "loss": 3.1151,
      "step": 412560
    },
    {
      "epoch": 0.07843209451747868,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001174979497949795,
      "loss": 3.0516,
      "step": 412570
    },
    {
      "epoch": 0.07845073775363685,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011749594959495951,
      "loss": 3.1225,
      "step": 412580
    },
    {
      "epoch": 0.07846938098979504,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011749394939493949,
      "loss": 3.1594,
      "step": 412590
    },
    {
      "epoch": 0.07848802422595323,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001174919491949195,
      "loss": 3.165,
      "step": 412600
    },
    {
      "epoch": 0.0785066674621114,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001174899489948995,
      "loss": 3.0782,
      "step": 412610
    },
    {
      "epoch": 0.0785253106982696,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001174879487948795,
      "loss": 3.1517,
      "step": 412620
    },
    {
      "epoch": 0.07854395393442777,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011748594859485949,
      "loss": 3.0621,
      "step": 412630
    },
    {
      "epoch": 0.07856259717058596,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001174839483948395,
      "loss": 3.1415,
      "step": 412640
    },
    {
      "epoch": 0.07858124040674415,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011748194819481948,
      "loss": 3.2483,
      "step": 412650
    },
    {
      "epoch": 0.07859988364290232,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011747994799479949,
      "loss": 3.177,
      "step": 412660
    },
    {
      "epoch": 0.07861852687906051,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011747794779477947,
      "loss": 3.1325,
      "step": 412670
    },
    {
      "epoch": 0.0786371701152187,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011747594759475948,
      "loss": 3.1255,
      "step": 412680
    },
    {
      "epoch": 0.07865581335137688,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011747394739473947,
      "loss": 3.1477,
      "step": 412690
    },
    {
      "epoch": 0.07867445658753507,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011747194719471948,
      "loss": 3.1149,
      "step": 412700
    },
    {
      "epoch": 0.07869309982369324,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011746994699469946,
      "loss": 3.1741,
      "step": 412710
    },
    {
      "epoch": 0.07871174305985143,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011746794679467947,
      "loss": 3.2246,
      "step": 412720
    },
    {
      "epoch": 0.07873038629600962,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011746594659465948,
      "loss": 3.1697,
      "step": 412730
    },
    {
      "epoch": 0.0787490295321678,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011746394639463946,
      "loss": 3.1208,
      "step": 412740
    },
    {
      "epoch": 0.07876767276832598,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011746194619461947,
      "loss": 3.1558,
      "step": 412750
    },
    {
      "epoch": 0.07878631600448416,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011745994599459947,
      "loss": 3.1496,
      "step": 412760
    },
    {
      "epoch": 0.07880495924064235,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011745794579457947,
      "loss": 3.1538,
      "step": 412770
    },
    {
      "epoch": 0.07882360247680054,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011745594559455946,
      "loss": 3.0755,
      "step": 412780
    },
    {
      "epoch": 0.07884224571295871,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011745394539453947,
      "loss": 3.0925,
      "step": 412790
    },
    {
      "epoch": 0.0788608889491169,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011745194519451945,
      "loss": 3.1761,
      "step": 412800
    },
    {
      "epoch": 0.07887953218527509,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011744994499449946,
      "loss": 3.1866,
      "step": 412810
    },
    {
      "epoch": 0.07889817542143326,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011744794479447944,
      "loss": 3.1642,
      "step": 412820
    },
    {
      "epoch": 0.07891681865759145,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011744594459445945,
      "loss": 3.1986,
      "step": 412830
    },
    {
      "epoch": 0.07893546189374963,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011744394439443945,
      "loss": 3.2652,
      "step": 412840
    },
    {
      "epoch": 0.07895410512990782,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011744194419441946,
      "loss": 3.2281,
      "step": 412850
    },
    {
      "epoch": 0.078972748366066,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011743994399439944,
      "loss": 3.1948,
      "step": 412860
    },
    {
      "epoch": 0.07899139160222418,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011743794379437945,
      "loss": 3.1841,
      "step": 412870
    },
    {
      "epoch": 0.07901003483838237,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011743594359435943,
      "loss": 3.1506,
      "step": 412880
    },
    {
      "epoch": 0.07902867807454055,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011743394339433944,
      "loss": 3.0307,
      "step": 412890
    },
    {
      "epoch": 0.07904732131069873,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011743194319431942,
      "loss": 3.1626,
      "step": 412900
    },
    {
      "epoch": 0.07906596454685692,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011742994299429943,
      "loss": 3.1007,
      "step": 412910
    },
    {
      "epoch": 0.0790846077830151,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011742794279427944,
      "loss": 3.1656,
      "step": 412920
    },
    {
      "epoch": 0.07910325101917329,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011742594259425944,
      "loss": 3.2304,
      "step": 412930
    },
    {
      "epoch": 0.07912189425533148,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011742394239423944,
      "loss": 3.1796,
      "step": 412940
    },
    {
      "epoch": 0.07914053749148965,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011742194219421943,
      "loss": 3.1611,
      "step": 412950
    },
    {
      "epoch": 0.07915918072764784,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011741994199419944,
      "loss": 3.1543,
      "step": 412960
    },
    {
      "epoch": 0.07917782396380602,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011741794179417942,
      "loss": 3.0932,
      "step": 412970
    },
    {
      "epoch": 0.0791964671999642,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011741594159415943,
      "loss": 3.1257,
      "step": 412980
    },
    {
      "epoch": 0.0792151104361224,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011741394139413941,
      "loss": 3.177,
      "step": 412990
    },
    {
      "epoch": 0.07923375367228057,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011741194119411942,
      "loss": 3.2793,
      "step": 413000
    },
    {
      "epoch": 0.07925239690843876,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011740994099409941,
      "loss": 3.3658,
      "step": 413010
    },
    {
      "epoch": 0.07927104014459695,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011740794079407942,
      "loss": 3.1368,
      "step": 413020
    },
    {
      "epoch": 0.07928968338075512,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001174059405940594,
      "loss": 3.025,
      "step": 413030
    },
    {
      "epoch": 0.07930832661691331,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011740394039403941,
      "loss": 3.3176,
      "step": 413040
    },
    {
      "epoch": 0.07932696985307149,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001174019401940194,
      "loss": 3.1452,
      "step": 413050
    },
    {
      "epoch": 0.07934561308922967,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001173999399939994,
      "loss": 3.1018,
      "step": 413060
    },
    {
      "epoch": 0.07936425632538786,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001173979397939794,
      "loss": 3.1831,
      "step": 413070
    },
    {
      "epoch": 0.07938289956154604,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001173959395939594,
      "loss": 3.0847,
      "step": 413080
    },
    {
      "epoch": 0.07940154279770423,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011739393939393939,
      "loss": 3.2393,
      "step": 413090
    },
    {
      "epoch": 0.0794201860338624,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001173919391939194,
      "loss": 3.131,
      "step": 413100
    },
    {
      "epoch": 0.07943882927002059,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011738993899389941,
      "loss": 3.1418,
      "step": 413110
    },
    {
      "epoch": 0.07945747250617878,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011738793879387939,
      "loss": 3.1984,
      "step": 413120
    },
    {
      "epoch": 0.07947611574233696,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001173859385938594,
      "loss": 3.2241,
      "step": 413130
    },
    {
      "epoch": 0.07949475897849514,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011738393839383938,
      "loss": 3.1399,
      "step": 413140
    },
    {
      "epoch": 0.07951340221465333,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011738193819381939,
      "loss": 3.2388,
      "step": 413150
    },
    {
      "epoch": 0.07953204545081151,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011737993799379939,
      "loss": 3.1833,
      "step": 413160
    },
    {
      "epoch": 0.0795506886869697,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001173779377937794,
      "loss": 3.1278,
      "step": 413170
    },
    {
      "epoch": 0.07956933192312787,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011737593759375938,
      "loss": 3.1879,
      "step": 413180
    },
    {
      "epoch": 0.07958797515928606,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011737393739373939,
      "loss": 3.3032,
      "step": 413190
    },
    {
      "epoch": 0.07960661839544425,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011737193719371937,
      "loss": 3.2163,
      "step": 413200
    },
    {
      "epoch": 0.07962526163160243,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011736993699369938,
      "loss": 3.2287,
      "step": 413210
    },
    {
      "epoch": 0.07964390486776061,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011736793679367936,
      "loss": 3.194,
      "step": 413220
    },
    {
      "epoch": 0.07966254810391879,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011736593659365937,
      "loss": 3.2477,
      "step": 413230
    },
    {
      "epoch": 0.07968119134007698,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011736393639363937,
      "loss": 3.0544,
      "step": 413240
    },
    {
      "epoch": 0.07969983457623517,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011736193619361938,
      "loss": 3.1246,
      "step": 413250
    },
    {
      "epoch": 0.07971847781239334,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011735993599359936,
      "loss": 3.1546,
      "step": 413260
    },
    {
      "epoch": 0.07973712104855153,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011735793579357937,
      "loss": 3.1511,
      "step": 413270
    },
    {
      "epoch": 0.07975576428470972,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011735593559355935,
      "loss": 3.2028,
      "step": 413280
    },
    {
      "epoch": 0.0797744075208679,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011735393539353936,
      "loss": 3.1243,
      "step": 413290
    },
    {
      "epoch": 0.07979305075702608,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011735193519351934,
      "loss": 3.1329,
      "step": 413300
    },
    {
      "epoch": 0.07981169399318426,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011734993499349935,
      "loss": 3.1535,
      "step": 413310
    },
    {
      "epoch": 0.07983033722934245,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011734793479347936,
      "loss": 3.2304,
      "step": 413320
    },
    {
      "epoch": 0.07984898046550064,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011734593459345935,
      "loss": 3.0807,
      "step": 413330
    },
    {
      "epoch": 0.07986762370165881,
      "grad_norm": 0.875,
      "learning_rate": 0.00011734393439343936,
      "loss": 3.1223,
      "step": 413340
    },
    {
      "epoch": 0.079886266937817,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00011734193419341935,
      "loss": 3.1458,
      "step": 413350
    },
    {
      "epoch": 0.07990491017397519,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011733993399339935,
      "loss": 3.1719,
      "step": 413360
    },
    {
      "epoch": 0.07992355341013337,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011733793379337934,
      "loss": 3.114,
      "step": 413370
    },
    {
      "epoch": 0.07994219664629156,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011733593359335935,
      "loss": 3.1457,
      "step": 413380
    },
    {
      "epoch": 0.07996083988244973,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011733393339333934,
      "loss": 3.2256,
      "step": 413390
    },
    {
      "epoch": 0.07997948311860792,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011733193319331934,
      "loss": 3.2236,
      "step": 413400
    },
    {
      "epoch": 0.07999812635476611,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011732993299329933,
      "loss": 3.1545,
      "step": 413410
    },
    {
      "epoch": 0.08001676959092428,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011732793279327934,
      "loss": 3.2271,
      "step": 413420
    },
    {
      "epoch": 0.08003541282708247,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011732593259325932,
      "loss": 3.1782,
      "step": 413430
    },
    {
      "epoch": 0.08005405606324065,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011732393239323933,
      "loss": 3.1707,
      "step": 413440
    },
    {
      "epoch": 0.08007269929939884,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011732193219321931,
      "loss": 3.204,
      "step": 413450
    },
    {
      "epoch": 0.08009134253555703,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011731993199319932,
      "loss": 3.129,
      "step": 413460
    },
    {
      "epoch": 0.0801099857717152,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011731793179317932,
      "loss": 3.1,
      "step": 413470
    },
    {
      "epoch": 0.08012862900787339,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011731593159315933,
      "loss": 3.1902,
      "step": 413480
    },
    {
      "epoch": 0.08014727224403158,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011731393139313931,
      "loss": 3.1047,
      "step": 413490
    },
    {
      "epoch": 0.08016591548018975,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011731193119311932,
      "loss": 3.2039,
      "step": 413500
    },
    {
      "epoch": 0.08018455871634794,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011730993099309933,
      "loss": 3.1711,
      "step": 413510
    },
    {
      "epoch": 0.08020320195250612,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011730793079307931,
      "loss": 3.2893,
      "step": 413520
    },
    {
      "epoch": 0.0802218451886643,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011730593059305932,
      "loss": 3.137,
      "step": 413530
    },
    {
      "epoch": 0.0802404884248225,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001173039303930393,
      "loss": 3.1351,
      "step": 413540
    },
    {
      "epoch": 0.08025913166098067,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011730193019301931,
      "loss": 3.3199,
      "step": 413550
    },
    {
      "epoch": 0.08027777489713886,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011729992999299931,
      "loss": 3.2244,
      "step": 413560
    },
    {
      "epoch": 0.08029641813329703,
      "grad_norm": 0.875,
      "learning_rate": 0.00011729792979297932,
      "loss": 3.1425,
      "step": 413570
    },
    {
      "epoch": 0.08031506136945522,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001172959295929593,
      "loss": 3.1952,
      "step": 413580
    },
    {
      "epoch": 0.08033370460561341,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011729392939293931,
      "loss": 3.2358,
      "step": 413590
    },
    {
      "epoch": 0.08035234784177159,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011729192919291929,
      "loss": 3.1472,
      "step": 413600
    },
    {
      "epoch": 0.08037099107792978,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001172899289928993,
      "loss": 3.2795,
      "step": 413610
    },
    {
      "epoch": 0.08038963431408797,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011728792879287928,
      "loss": 3.2436,
      "step": 413620
    },
    {
      "epoch": 0.08040827755024614,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011728592859285929,
      "loss": 3.0197,
      "step": 413630
    },
    {
      "epoch": 0.08042692078640433,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011728392839283929,
      "loss": 3.0834,
      "step": 413640
    },
    {
      "epoch": 0.0804455640225625,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001172819281928193,
      "loss": 3.1907,
      "step": 413650
    },
    {
      "epoch": 0.0804642072587207,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011727992799279928,
      "loss": 3.2449,
      "step": 413660
    },
    {
      "epoch": 0.08048285049487888,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011727792779277929,
      "loss": 3.1277,
      "step": 413670
    },
    {
      "epoch": 0.08050149373103706,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011727592759275927,
      "loss": 3.1298,
      "step": 413680
    },
    {
      "epoch": 0.08052013696719525,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011727392739273928,
      "loss": 3.2127,
      "step": 413690
    },
    {
      "epoch": 0.08053878020335344,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011727192719271929,
      "loss": 3.211,
      "step": 413700
    },
    {
      "epoch": 0.08055742343951161,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011726992699269928,
      "loss": 3.2121,
      "step": 413710
    },
    {
      "epoch": 0.0805760666756698,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011726792679267928,
      "loss": 3.1854,
      "step": 413720
    },
    {
      "epoch": 0.08059470991182797,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011726592659265927,
      "loss": 3.0626,
      "step": 413730
    },
    {
      "epoch": 0.08061335314798616,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011726392639263928,
      "loss": 3.1646,
      "step": 413740
    },
    {
      "epoch": 0.08063199638414435,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011726192619261926,
      "loss": 3.1727,
      "step": 413750
    },
    {
      "epoch": 0.08065063962030253,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011725992599259927,
      "loss": 3.2909,
      "step": 413760
    },
    {
      "epoch": 0.08066928285646072,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011725792579257925,
      "loss": 3.0588,
      "step": 413770
    },
    {
      "epoch": 0.08068792609261889,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011725592559255926,
      "loss": 3.1728,
      "step": 413780
    },
    {
      "epoch": 0.08070656932877708,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011725392539253926,
      "loss": 3.2498,
      "step": 413790
    },
    {
      "epoch": 0.08072521256493527,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011725192519251927,
      "loss": 3.1522,
      "step": 413800
    },
    {
      "epoch": 0.08074385580109344,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011724992499249925,
      "loss": 3.1621,
      "step": 413810
    },
    {
      "epoch": 0.08076249903725163,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011724792479247926,
      "loss": 3.2395,
      "step": 413820
    },
    {
      "epoch": 0.08078114227340982,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011724592459245924,
      "loss": 3.0173,
      "step": 413830
    },
    {
      "epoch": 0.080799785509568,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011724392439243925,
      "loss": 3.1928,
      "step": 413840
    },
    {
      "epoch": 0.08081842874572619,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011724192419241923,
      "loss": 3.1395,
      "step": 413850
    },
    {
      "epoch": 0.08083707198188436,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011723992399239924,
      "loss": 3.2391,
      "step": 413860
    },
    {
      "epoch": 0.08085571521804255,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011723792379237924,
      "loss": 3.3136,
      "step": 413870
    },
    {
      "epoch": 0.08087435845420074,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011723592359235925,
      "loss": 3.2337,
      "step": 413880
    },
    {
      "epoch": 0.08089300169035891,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011723392339233923,
      "loss": 3.255,
      "step": 413890
    },
    {
      "epoch": 0.0809116449265171,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011723192319231924,
      "loss": 3.0645,
      "step": 413900
    },
    {
      "epoch": 0.08093028816267528,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011722992299229925,
      "loss": 3.1372,
      "step": 413910
    },
    {
      "epoch": 0.08094893139883347,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011722792279227923,
      "loss": 3.1784,
      "step": 413920
    },
    {
      "epoch": 0.08096757463499166,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011722592259225924,
      "loss": 3.0907,
      "step": 413930
    },
    {
      "epoch": 0.08098621787114983,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011722392239223922,
      "loss": 3.2851,
      "step": 413940
    },
    {
      "epoch": 0.08100486110730802,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011722192219221923,
      "loss": 3.1914,
      "step": 413950
    },
    {
      "epoch": 0.08102350434346621,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011721992199219923,
      "loss": 3.1741,
      "step": 413960
    },
    {
      "epoch": 0.08104214757962439,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011721792179217923,
      "loss": 3.0705,
      "step": 413970
    },
    {
      "epoch": 0.08106079081578257,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011721592159215922,
      "loss": 3.0229,
      "step": 413980
    },
    {
      "epoch": 0.08107943405194075,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00011721392139213923,
      "loss": 3.2447,
      "step": 413990
    },
    {
      "epoch": 0.08109807728809894,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011721192119211921,
      "loss": 3.1168,
      "step": 414000
    },
    {
      "epoch": 0.08111672052425713,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011720992099209922,
      "loss": 3.1843,
      "step": 414010
    },
    {
      "epoch": 0.0811353637604153,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011720792079207921,
      "loss": 3.1699,
      "step": 414020
    },
    {
      "epoch": 0.08115400699657349,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011720592059205922,
      "loss": 3.1928,
      "step": 414030
    },
    {
      "epoch": 0.08117265023273168,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001172039203920392,
      "loss": 3.2126,
      "step": 414040
    },
    {
      "epoch": 0.08119129346888986,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011720192019201921,
      "loss": 3.134,
      "step": 414050
    },
    {
      "epoch": 0.08120993670504804,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001171999199919992,
      "loss": 3.255,
      "step": 414060
    },
    {
      "epoch": 0.08122857994120622,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001171979197919792,
      "loss": 2.9988,
      "step": 414070
    },
    {
      "epoch": 0.08124722317736441,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011719591959195919,
      "loss": 3.07,
      "step": 414080
    },
    {
      "epoch": 0.0812658664135226,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001171939193919392,
      "loss": 3.1481,
      "step": 414090
    },
    {
      "epoch": 0.08128450964968077,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001171919191919192,
      "loss": 3.1494,
      "step": 414100
    },
    {
      "epoch": 0.08130315288583896,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001171899189918992,
      "loss": 3.2203,
      "step": 414110
    },
    {
      "epoch": 0.08132179612199714,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011718791879187921,
      "loss": 3.2491,
      "step": 414120
    },
    {
      "epoch": 0.08134043935815533,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011718591859185919,
      "loss": 3.2615,
      "step": 414130
    },
    {
      "epoch": 0.08135908259431351,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001171839183918392,
      "loss": 3.2112,
      "step": 414140
    },
    {
      "epoch": 0.08137772583047169,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011718191819181918,
      "loss": 3.0841,
      "step": 414150
    },
    {
      "epoch": 0.08139636906662988,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011717991799179919,
      "loss": 3.137,
      "step": 414160
    },
    {
      "epoch": 0.08141501230278807,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011717791779177917,
      "loss": 3.127,
      "step": 414170
    },
    {
      "epoch": 0.08143365553894624,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011717591759175918,
      "loss": 3.21,
      "step": 414180
    },
    {
      "epoch": 0.08145229877510443,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011717391739173918,
      "loss": 3.316,
      "step": 414190
    },
    {
      "epoch": 0.0814709420112626,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011717191719171919,
      "loss": 3.2261,
      "step": 414200
    },
    {
      "epoch": 0.0814895852474208,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011716991699169917,
      "loss": 3.2374,
      "step": 414210
    },
    {
      "epoch": 0.08150822848357898,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011716791679167918,
      "loss": 3.156,
      "step": 414220
    },
    {
      "epoch": 0.08152687171973716,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011716591659165916,
      "loss": 3.168,
      "step": 414230
    },
    {
      "epoch": 0.08154551495589535,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011716391639163917,
      "loss": 3.0951,
      "step": 414240
    },
    {
      "epoch": 0.08156415819205352,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011716191619161917,
      "loss": 3.1173,
      "step": 414250
    },
    {
      "epoch": 0.08158280142821171,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011715991599159916,
      "loss": 3.2014,
      "step": 414260
    },
    {
      "epoch": 0.0816014446643699,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011715791579157916,
      "loss": 3.2049,
      "step": 414270
    },
    {
      "epoch": 0.08162008790052808,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011715591559155917,
      "loss": 3.1116,
      "step": 414280
    },
    {
      "epoch": 0.08163873113668627,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011715391539153915,
      "loss": 3.2478,
      "step": 414290
    },
    {
      "epoch": 0.08165737437284445,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011715191519151916,
      "loss": 3.1829,
      "step": 414300
    },
    {
      "epoch": 0.08167601760900263,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011714991499149917,
      "loss": 3.1541,
      "step": 414310
    },
    {
      "epoch": 0.08169466084516082,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011714791479147915,
      "loss": 3.1907,
      "step": 414320
    },
    {
      "epoch": 0.081713304081319,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011714591459145916,
      "loss": 3.2775,
      "step": 414330
    },
    {
      "epoch": 0.08173194731747718,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011714391439143915,
      "loss": 3.2219,
      "step": 414340
    },
    {
      "epoch": 0.08175059055363537,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011714191419141916,
      "loss": 3.1213,
      "step": 414350
    },
    {
      "epoch": 0.08176923378979355,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00011713991399139914,
      "loss": 3.1314,
      "step": 414360
    },
    {
      "epoch": 0.08178787702595174,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011713791379137915,
      "loss": 3.1964,
      "step": 414370
    },
    {
      "epoch": 0.08180652026210992,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011713591359135914,
      "loss": 3.1888,
      "step": 414380
    },
    {
      "epoch": 0.0818251634982681,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011713391339133914,
      "loss": 3.3095,
      "step": 414390
    },
    {
      "epoch": 0.08184380673442629,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011713191319131913,
      "loss": 3.1975,
      "step": 414400
    },
    {
      "epoch": 0.08186244997058446,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011712991299129914,
      "loss": 3.033,
      "step": 414410
    },
    {
      "epoch": 0.08188109320674265,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011712791279127913,
      "loss": 3.2358,
      "step": 414420
    },
    {
      "epoch": 0.08189973644290084,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011712591259125914,
      "loss": 3.2351,
      "step": 414430
    },
    {
      "epoch": 0.08191837967905902,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011712391239123912,
      "loss": 3.0944,
      "step": 414440
    },
    {
      "epoch": 0.0819370229152172,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011712191219121913,
      "loss": 3.285,
      "step": 414450
    },
    {
      "epoch": 0.08195566615137538,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011711991199119911,
      "loss": 3.15,
      "step": 414460
    },
    {
      "epoch": 0.08197430938753357,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011711791179117912,
      "loss": 3.3057,
      "step": 414470
    },
    {
      "epoch": 0.08199295262369176,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001171159115911591,
      "loss": 3.0456,
      "step": 414480
    },
    {
      "epoch": 0.08201159585984993,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011711391139113911,
      "loss": 3.1874,
      "step": 414490
    },
    {
      "epoch": 0.08203023909600812,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011711191119111912,
      "loss": 3.1595,
      "step": 414500
    },
    {
      "epoch": 0.08204888233216631,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011710991099109912,
      "loss": 3.1853,
      "step": 414510
    },
    {
      "epoch": 0.08206752556832449,
      "grad_norm": 0.75,
      "learning_rate": 0.00011710791079107913,
      "loss": 3.2458,
      "step": 414520
    },
    {
      "epoch": 0.08208616880448268,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011710591059105911,
      "loss": 3.1131,
      "step": 414530
    },
    {
      "epoch": 0.08210481204064085,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011710391039103912,
      "loss": 3.0714,
      "step": 414540
    },
    {
      "epoch": 0.08212345527679904,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001171019101910191,
      "loss": 3.0898,
      "step": 414550
    },
    {
      "epoch": 0.08214209851295723,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011709990999099911,
      "loss": 3.1784,
      "step": 414560
    },
    {
      "epoch": 0.0821607417491154,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011709790979097909,
      "loss": 3.1866,
      "step": 414570
    },
    {
      "epoch": 0.08217938498527359,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001170959095909591,
      "loss": 3.2099,
      "step": 414580
    },
    {
      "epoch": 0.08219802822143177,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001170939093909391,
      "loss": 3.1812,
      "step": 414590
    },
    {
      "epoch": 0.08221667145758996,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001170919091909191,
      "loss": 3.1382,
      "step": 414600
    },
    {
      "epoch": 0.08223531469374815,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011708990899089909,
      "loss": 3.2131,
      "step": 414610
    },
    {
      "epoch": 0.08225395792990632,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001170879087908791,
      "loss": 2.9991,
      "step": 414620
    },
    {
      "epoch": 0.08227260116606451,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011708590859085908,
      "loss": 3.2181,
      "step": 414630
    },
    {
      "epoch": 0.0822912444022227,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011708390839083909,
      "loss": 3.2144,
      "step": 414640
    },
    {
      "epoch": 0.08230988763838087,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011708190819081908,
      "loss": 3.1334,
      "step": 414650
    },
    {
      "epoch": 0.08232853087453906,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001170799079907991,
      "loss": 3.2394,
      "step": 414660
    },
    {
      "epoch": 0.08234717411069724,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011707790779077908,
      "loss": 3.2114,
      "step": 414670
    },
    {
      "epoch": 0.08236581734685543,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011707590759075908,
      "loss": 3.2478,
      "step": 414680
    },
    {
      "epoch": 0.08238446058301362,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001170739073907391,
      "loss": 3.129,
      "step": 414690
    },
    {
      "epoch": 0.08240310381917179,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011707190719071908,
      "loss": 3.1446,
      "step": 414700
    },
    {
      "epoch": 0.08242174705532998,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011706990699069908,
      "loss": 3.2038,
      "step": 414710
    },
    {
      "epoch": 0.08244039029148817,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011706790679067907,
      "loss": 3.2527,
      "step": 414720
    },
    {
      "epoch": 0.08245903352764634,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011706590659065908,
      "loss": 3.1341,
      "step": 414730
    },
    {
      "epoch": 0.08247767676380453,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011706390639063907,
      "loss": 3.1888,
      "step": 414740
    },
    {
      "epoch": 0.08249631999996271,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011706190619061908,
      "loss": 3.0419,
      "step": 414750
    },
    {
      "epoch": 0.0825149632361209,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011705990599059906,
      "loss": 3.2453,
      "step": 414760
    },
    {
      "epoch": 0.08253360647227909,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011705790579057907,
      "loss": 3.1369,
      "step": 414770
    },
    {
      "epoch": 0.08255224970843726,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011705590559055905,
      "loss": 3.2398,
      "step": 414780
    },
    {
      "epoch": 0.08257089294459545,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011705390539053906,
      "loss": 3.1358,
      "step": 414790
    },
    {
      "epoch": 0.08258953618075363,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011705190519051905,
      "loss": 3.2611,
      "step": 414800
    },
    {
      "epoch": 0.08260817941691181,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011704990499049905,
      "loss": 3.1902,
      "step": 414810
    },
    {
      "epoch": 0.08262682265307,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011704790479047905,
      "loss": 3.0287,
      "step": 414820
    },
    {
      "epoch": 0.08264546588922818,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011704590459045906,
      "loss": 3.2869,
      "step": 414830
    },
    {
      "epoch": 0.08266410912538637,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011704390439043904,
      "loss": 3.2046,
      "step": 414840
    },
    {
      "epoch": 0.08268275236154456,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011704190419041905,
      "loss": 3.1786,
      "step": 414850
    },
    {
      "epoch": 0.08270139559770273,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011703990399039903,
      "loss": 3.2342,
      "step": 414860
    },
    {
      "epoch": 0.08272003883386092,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011703790379037904,
      "loss": 3.273,
      "step": 414870
    },
    {
      "epoch": 0.0827386820700191,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011703590359035904,
      "loss": 3.1681,
      "step": 414880
    },
    {
      "epoch": 0.08275732530617728,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011703390339033903,
      "loss": 3.1492,
      "step": 414890
    },
    {
      "epoch": 0.08277596854233547,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011703190319031904,
      "loss": 3.2132,
      "step": 414900
    },
    {
      "epoch": 0.08279461177849365,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011702990299029904,
      "loss": 3.1599,
      "step": 414910
    },
    {
      "epoch": 0.08281325501465184,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011702790279027905,
      "loss": 3.1807,
      "step": 414920
    },
    {
      "epoch": 0.08283189825081001,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011702590259025903,
      "loss": 3.3011,
      "step": 414930
    },
    {
      "epoch": 0.0828505414869682,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011702390239023904,
      "loss": 3.1434,
      "step": 414940
    },
    {
      "epoch": 0.08286918472312639,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011702190219021902,
      "loss": 3.0397,
      "step": 414950
    },
    {
      "epoch": 0.08288782795928457,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011701990199019903,
      "loss": 3.2008,
      "step": 414960
    },
    {
      "epoch": 0.08290647119544275,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011701790179017902,
      "loss": 3.2531,
      "step": 414970
    },
    {
      "epoch": 0.08292511443160094,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011701590159015903,
      "loss": 3.2403,
      "step": 414980
    },
    {
      "epoch": 0.08294375766775912,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011701390139013902,
      "loss": 3.1021,
      "step": 414990
    },
    {
      "epoch": 0.08296240090391731,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011701190119011902,
      "loss": 3.0748,
      "step": 415000
    },
    {
      "epoch": 0.08298104414007548,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011700990099009901,
      "loss": 3.2558,
      "step": 415010
    },
    {
      "epoch": 0.08299968737623367,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011700790079007902,
      "loss": 3.158,
      "step": 415020
    },
    {
      "epoch": 0.08301833061239186,
      "grad_norm": 0.76953125,
      "learning_rate": 0.000117005900590059,
      "loss": 3.1898,
      "step": 415030
    },
    {
      "epoch": 0.08303697384855004,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011700390039003901,
      "loss": 3.1035,
      "step": 415040
    },
    {
      "epoch": 0.08305561708470822,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000117001900190019,
      "loss": 2.9211,
      "step": 415050
    },
    {
      "epoch": 0.08307426032086641,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011699989998999901,
      "loss": 3.1309,
      "step": 415060
    },
    {
      "epoch": 0.08309290355702459,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000116997899789979,
      "loss": 3.1586,
      "step": 415070
    },
    {
      "epoch": 0.08311154679318278,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000116995899589959,
      "loss": 3.097,
      "step": 415080
    },
    {
      "epoch": 0.08313019002934095,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011699389938993901,
      "loss": 3.2342,
      "step": 415090
    },
    {
      "epoch": 0.08314883326549914,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000116991899189919,
      "loss": 3.129,
      "step": 415100
    },
    {
      "epoch": 0.08316747650165733,
      "grad_norm": 0.71875,
      "learning_rate": 0.000116989898989899,
      "loss": 3.1698,
      "step": 415110
    },
    {
      "epoch": 0.0831861197378155,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011698789878987899,
      "loss": 3.1989,
      "step": 415120
    },
    {
      "epoch": 0.0832047629739737,
      "grad_norm": 0.765625,
      "learning_rate": 0.000116985898589859,
      "loss": 3.2649,
      "step": 415130
    },
    {
      "epoch": 0.08322340621013187,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011698389838983899,
      "loss": 3.0384,
      "step": 415140
    },
    {
      "epoch": 0.08324204944629006,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000116981898189819,
      "loss": 3.2944,
      "step": 415150
    },
    {
      "epoch": 0.08326069268244825,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011697989798979898,
      "loss": 3.1239,
      "step": 415160
    },
    {
      "epoch": 0.08327933591860642,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011697789778977899,
      "loss": 3.1054,
      "step": 415170
    },
    {
      "epoch": 0.08329797915476461,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011697589758975897,
      "loss": 3.2508,
      "step": 415180
    },
    {
      "epoch": 0.0833166223909228,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011697389738973898,
      "loss": 2.9969,
      "step": 415190
    },
    {
      "epoch": 0.08333526562708098,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011697189718971898,
      "loss": 3.0101,
      "step": 415200
    },
    {
      "epoch": 0.08335390886323917,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011696989698969897,
      "loss": 3.2976,
      "step": 415210
    },
    {
      "epoch": 0.08337255209939734,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011696789678967897,
      "loss": 3.2716,
      "step": 415220
    },
    {
      "epoch": 0.08339119533555553,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011696589658965898,
      "loss": 3.0718,
      "step": 415230
    },
    {
      "epoch": 0.08340983857171372,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011696389638963896,
      "loss": 3.1399,
      "step": 415240
    },
    {
      "epoch": 0.0834284818078719,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011696189618961897,
      "loss": 3.3297,
      "step": 415250
    },
    {
      "epoch": 0.08344712504403008,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011695989598959895,
      "loss": 3.1646,
      "step": 415260
    },
    {
      "epoch": 0.08346576828018826,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011695789578957896,
      "loss": 3.1339,
      "step": 415270
    },
    {
      "epoch": 0.08348441151634645,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011695589558955897,
      "loss": 3.1471,
      "step": 415280
    },
    {
      "epoch": 0.08350305475250464,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011695389538953896,
      "loss": 3.1672,
      "step": 415290
    },
    {
      "epoch": 0.08352169798866281,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011695189518951897,
      "loss": 3.1486,
      "step": 415300
    },
    {
      "epoch": 0.083540341224821,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011694989498949896,
      "loss": 3.0894,
      "step": 415310
    },
    {
      "epoch": 0.08355898446097919,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011694789478947897,
      "loss": 3.18,
      "step": 415320
    },
    {
      "epoch": 0.08357762769713736,
      "grad_norm": 0.625,
      "learning_rate": 0.00011694589458945895,
      "loss": 3.1767,
      "step": 415330
    },
    {
      "epoch": 0.08359627093329555,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011694389438943896,
      "loss": 3.174,
      "step": 415340
    },
    {
      "epoch": 0.08361491416945373,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011694189418941894,
      "loss": 3.0475,
      "step": 415350
    },
    {
      "epoch": 0.08363355740561192,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011693989398939895,
      "loss": 3.1624,
      "step": 415360
    },
    {
      "epoch": 0.0836522006417701,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011693789378937894,
      "loss": 3.1352,
      "step": 415370
    },
    {
      "epoch": 0.08367084387792828,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011693589358935895,
      "loss": 3.2171,
      "step": 415380
    },
    {
      "epoch": 0.08368948711408647,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011693389338933893,
      "loss": 3.1274,
      "step": 415390
    },
    {
      "epoch": 0.08370813035024466,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011693189318931894,
      "loss": 3.1834,
      "step": 415400
    },
    {
      "epoch": 0.08372677358640283,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011692989298929893,
      "loss": 3.1547,
      "step": 415410
    },
    {
      "epoch": 0.08374541682256102,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011692789278927893,
      "loss": 3.0169,
      "step": 415420
    },
    {
      "epoch": 0.0837640600587192,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011692589258925892,
      "loss": 3.1169,
      "step": 415430
    },
    {
      "epoch": 0.08378270329487739,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011692389238923893,
      "loss": 3.1397,
      "step": 415440
    },
    {
      "epoch": 0.08380134653103558,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011692189218921892,
      "loss": 3.1544,
      "step": 415450
    },
    {
      "epoch": 0.08381998976719375,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011691989198919893,
      "loss": 3.1245,
      "step": 415460
    },
    {
      "epoch": 0.08383863300335194,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011691789178917891,
      "loss": 3.1489,
      "step": 415470
    },
    {
      "epoch": 0.08385727623951011,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011691589158915892,
      "loss": 3.0583,
      "step": 415480
    },
    {
      "epoch": 0.0838759194756683,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011691389138913893,
      "loss": 3.0921,
      "step": 415490
    },
    {
      "epoch": 0.08389456271182649,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011691189118911891,
      "loss": 3.0598,
      "step": 415500
    },
    {
      "epoch": 0.08391320594798467,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011690989098909892,
      "loss": 3.1262,
      "step": 415510
    },
    {
      "epoch": 0.08393184918414286,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011690789078907892,
      "loss": 3.1083,
      "step": 415520
    },
    {
      "epoch": 0.08395049242030105,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011690589058905891,
      "loss": 3.2535,
      "step": 415530
    },
    {
      "epoch": 0.08396913565645922,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011690389038903891,
      "loss": 3.0694,
      "step": 415540
    },
    {
      "epoch": 0.08398777889261741,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011690189018901892,
      "loss": 3.1167,
      "step": 415550
    },
    {
      "epoch": 0.08400642212877558,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001168998899889989,
      "loss": 3.2268,
      "step": 415560
    },
    {
      "epoch": 0.08402506536493377,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011689788978897891,
      "loss": 3.1608,
      "step": 415570
    },
    {
      "epoch": 0.08404370860109196,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011689588958895889,
      "loss": 3.1166,
      "step": 415580
    },
    {
      "epoch": 0.08406235183725014,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001168938893889389,
      "loss": 3.2726,
      "step": 415590
    },
    {
      "epoch": 0.08408099507340833,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001168918891889189,
      "loss": 3.0395,
      "step": 415600
    },
    {
      "epoch": 0.0840996383095665,
      "grad_norm": 0.75,
      "learning_rate": 0.0001168898889888989,
      "loss": 3.1559,
      "step": 415610
    },
    {
      "epoch": 0.08411828154572469,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011688788878887889,
      "loss": 3.1839,
      "step": 415620
    },
    {
      "epoch": 0.08413692478188288,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001168858885888589,
      "loss": 3.168,
      "step": 415630
    },
    {
      "epoch": 0.08415556801804105,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011688388838883888,
      "loss": 3.1596,
      "step": 415640
    },
    {
      "epoch": 0.08417421125419924,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011688188818881889,
      "loss": 3.2195,
      "step": 415650
    },
    {
      "epoch": 0.08419285449035743,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011687988798879887,
      "loss": 3.2606,
      "step": 415660
    },
    {
      "epoch": 0.08421149772651561,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011687788778877888,
      "loss": 3.1934,
      "step": 415670
    },
    {
      "epoch": 0.0842301409626738,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011687588758875889,
      "loss": 3.17,
      "step": 415680
    },
    {
      "epoch": 0.08424878419883197,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011687388738873888,
      "loss": 3.3786,
      "step": 415690
    },
    {
      "epoch": 0.08426742743499016,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011687188718871889,
      "loss": 3.098,
      "step": 415700
    },
    {
      "epoch": 0.08428607067114835,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011686988698869887,
      "loss": 3.3168,
      "step": 415710
    },
    {
      "epoch": 0.08430471390730652,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011686788678867888,
      "loss": 3.2671,
      "step": 415720
    },
    {
      "epoch": 0.08432335714346471,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011686588658865887,
      "loss": 3.175,
      "step": 415730
    },
    {
      "epoch": 0.0843420003796229,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011686388638863887,
      "loss": 3.1573,
      "step": 415740
    },
    {
      "epoch": 0.08436064361578108,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011686188618861886,
      "loss": 3.1644,
      "step": 415750
    },
    {
      "epoch": 0.08437928685193927,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011685988598859887,
      "loss": 3.276,
      "step": 415760
    },
    {
      "epoch": 0.08439793008809744,
      "grad_norm": 0.75,
      "learning_rate": 0.00011685788578857886,
      "loss": 3.2024,
      "step": 415770
    },
    {
      "epoch": 0.08441657332425563,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011685588558855887,
      "loss": 3.0682,
      "step": 415780
    },
    {
      "epoch": 0.08443521656041382,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011685388538853885,
      "loss": 3.2257,
      "step": 415790
    },
    {
      "epoch": 0.084453859796572,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011685188518851886,
      "loss": 3.0324,
      "step": 415800
    },
    {
      "epoch": 0.08447250303273018,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011684988498849884,
      "loss": 3.2073,
      "step": 415810
    },
    {
      "epoch": 0.08449114626888836,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011684788478847885,
      "loss": 3.1977,
      "step": 415820
    },
    {
      "epoch": 0.08450978950504655,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011684588458845885,
      "loss": 3.2213,
      "step": 415830
    },
    {
      "epoch": 0.08452843274120474,
      "grad_norm": 0.875,
      "learning_rate": 0.00011684388438843886,
      "loss": 3.1514,
      "step": 415840
    },
    {
      "epoch": 0.08454707597736291,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011684188418841884,
      "loss": 3.0131,
      "step": 415850
    },
    {
      "epoch": 0.0845657192135211,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011683988398839885,
      "loss": 3.1126,
      "step": 415860
    },
    {
      "epoch": 0.08458436244967929,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011683788378837883,
      "loss": 3.1646,
      "step": 415870
    },
    {
      "epoch": 0.08460300568583747,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011683588358835884,
      "loss": 3.1324,
      "step": 415880
    },
    {
      "epoch": 0.08462164892199565,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011683388338833885,
      "loss": 3.1474,
      "step": 415890
    },
    {
      "epoch": 0.08464029215815383,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011683188318831883,
      "loss": 3.1478,
      "step": 415900
    },
    {
      "epoch": 0.08465893539431202,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011682988298829884,
      "loss": 3.1909,
      "step": 415910
    },
    {
      "epoch": 0.08467757863047021,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011682788278827884,
      "loss": 3.0631,
      "step": 415920
    },
    {
      "epoch": 0.08469622186662838,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00011682588258825885,
      "loss": 3.1253,
      "step": 415930
    },
    {
      "epoch": 0.08471486510278657,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011682388238823883,
      "loss": 3.2755,
      "step": 415940
    },
    {
      "epoch": 0.08473350833894475,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011682188218821884,
      "loss": 3.169,
      "step": 415950
    },
    {
      "epoch": 0.08475215157510294,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011681988198819882,
      "loss": 3.1579,
      "step": 415960
    },
    {
      "epoch": 0.08477079481126112,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011681788178817883,
      "loss": 3.1319,
      "step": 415970
    },
    {
      "epoch": 0.0847894380474193,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011681588158815881,
      "loss": 3.1948,
      "step": 415980
    },
    {
      "epoch": 0.08480808128357749,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011681388138813882,
      "loss": 3.3654,
      "step": 415990
    },
    {
      "epoch": 0.08482672451973568,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011681188118811881,
      "loss": 3.1098,
      "step": 416000
    },
    {
      "epoch": 0.08484536775589385,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011680988098809882,
      "loss": 3.2091,
      "step": 416010
    },
    {
      "epoch": 0.08486401099205204,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001168078807880788,
      "loss": 3.081,
      "step": 416020
    },
    {
      "epoch": 0.08488265422821022,
      "grad_norm": 0.75,
      "learning_rate": 0.00011680588058805881,
      "loss": 3.2245,
      "step": 416030
    },
    {
      "epoch": 0.0849012974643684,
      "grad_norm": 0.875,
      "learning_rate": 0.0001168038803880388,
      "loss": 3.1623,
      "step": 416040
    },
    {
      "epoch": 0.0849199407005266,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001168018801880188,
      "loss": 3.2202,
      "step": 416050
    },
    {
      "epoch": 0.08493858393668477,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011679987998799879,
      "loss": 3.1939,
      "step": 416060
    },
    {
      "epoch": 0.08495722717284296,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001167978797879788,
      "loss": 3.1207,
      "step": 416070
    },
    {
      "epoch": 0.08497587040900115,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001167958795879588,
      "loss": 3.105,
      "step": 416080
    },
    {
      "epoch": 0.08499451364515932,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001167938793879388,
      "loss": 3.2116,
      "step": 416090
    },
    {
      "epoch": 0.08501315688131751,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011679187918791881,
      "loss": 3.2884,
      "step": 416100
    },
    {
      "epoch": 0.08503180011747569,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001167898789878988,
      "loss": 3.2022,
      "step": 416110
    },
    {
      "epoch": 0.08505044335363388,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001167878787878788,
      "loss": 3.1405,
      "step": 416120
    },
    {
      "epoch": 0.08506908658979206,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011678587858785878,
      "loss": 3.2611,
      "step": 416130
    },
    {
      "epoch": 0.08508772982595024,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001167838783878388,
      "loss": 3.1723,
      "step": 416140
    },
    {
      "epoch": 0.08510637306210843,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011678187818781879,
      "loss": 3.2118,
      "step": 416150
    },
    {
      "epoch": 0.0851250162982666,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011677987798779878,
      "loss": 3.28,
      "step": 416160
    },
    {
      "epoch": 0.08514365953442479,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011677787778777878,
      "loss": 3.1561,
      "step": 416170
    },
    {
      "epoch": 0.08516230277058298,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011677587758775879,
      "loss": 3.1403,
      "step": 416180
    },
    {
      "epoch": 0.08518094600674116,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011677387738773877,
      "loss": 3.2637,
      "step": 416190
    },
    {
      "epoch": 0.08519958924289935,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011677187718771878,
      "loss": 3.068,
      "step": 416200
    },
    {
      "epoch": 0.08521823247905753,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011676987698769876,
      "loss": 3.1083,
      "step": 416210
    },
    {
      "epoch": 0.08523687571521571,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011676787678767877,
      "loss": 3.2776,
      "step": 416220
    },
    {
      "epoch": 0.0852555189513739,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011676587658765877,
      "loss": 3.1454,
      "step": 416230
    },
    {
      "epoch": 0.08527416218753207,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011676387638763878,
      "loss": 3.1426,
      "step": 416240
    },
    {
      "epoch": 0.08529280542369026,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011676187618761876,
      "loss": 3.2057,
      "step": 416250
    },
    {
      "epoch": 0.08531144865984845,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011675987598759877,
      "loss": 3.1195,
      "step": 416260
    },
    {
      "epoch": 0.08533009189600663,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011675787578757878,
      "loss": 3.0691,
      "step": 416270
    },
    {
      "epoch": 0.08534873513216482,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011675587558755876,
      "loss": 3.065,
      "step": 416280
    },
    {
      "epoch": 0.08536737836832299,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011675387538753877,
      "loss": 3.0396,
      "step": 416290
    },
    {
      "epoch": 0.08538602160448118,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011675187518751875,
      "loss": 3.1865,
      "step": 416300
    },
    {
      "epoch": 0.08540466484063937,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011674987498749876,
      "loss": 3.1965,
      "step": 416310
    },
    {
      "epoch": 0.08542330807679754,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011674787478747875,
      "loss": 3.2327,
      "step": 416320
    },
    {
      "epoch": 0.08544195131295573,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011674587458745876,
      "loss": 3.1493,
      "step": 416330
    },
    {
      "epoch": 0.08546059454911392,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011674387438743875,
      "loss": 3.0791,
      "step": 416340
    },
    {
      "epoch": 0.0854792377852721,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011674187418741876,
      "loss": 3.2283,
      "step": 416350
    },
    {
      "epoch": 0.08549788102143029,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011673987398739874,
      "loss": 3.0812,
      "step": 416360
    },
    {
      "epoch": 0.08551652425758846,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011673787378737875,
      "loss": 3.1712,
      "step": 416370
    },
    {
      "epoch": 0.08553516749374665,
      "grad_norm": 0.625,
      "learning_rate": 0.00011673587358735873,
      "loss": 3.1417,
      "step": 416380
    },
    {
      "epoch": 0.08555381072990484,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011673387338733874,
      "loss": 3.1613,
      "step": 416390
    },
    {
      "epoch": 0.08557245396606301,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011673187318731873,
      "loss": 3.1912,
      "step": 416400
    },
    {
      "epoch": 0.0855910972022212,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011672987298729874,
      "loss": 3.1505,
      "step": 416410
    },
    {
      "epoch": 0.08560974043837939,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011672787278727872,
      "loss": 3.1957,
      "step": 416420
    },
    {
      "epoch": 0.08562838367453757,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011672587258725873,
      "loss": 3.1031,
      "step": 416430
    },
    {
      "epoch": 0.08564702691069576,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011672387238723872,
      "loss": 3.2072,
      "step": 416440
    },
    {
      "epoch": 0.08566567014685393,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011672187218721872,
      "loss": 3.174,
      "step": 416450
    },
    {
      "epoch": 0.08568431338301212,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011671987198719872,
      "loss": 3.1553,
      "step": 416460
    },
    {
      "epoch": 0.08570295661917031,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011671787178717873,
      "loss": 3.2601,
      "step": 416470
    },
    {
      "epoch": 0.08572159985532848,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011671587158715873,
      "loss": 3.1811,
      "step": 416480
    },
    {
      "epoch": 0.08574024309148667,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011671387138713872,
      "loss": 3.1223,
      "step": 416490
    },
    {
      "epoch": 0.08575888632764485,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011671187118711873,
      "loss": 3.2202,
      "step": 416500
    },
    {
      "epoch": 0.08577752956380304,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011670987098709871,
      "loss": 3.2143,
      "step": 416510
    },
    {
      "epoch": 0.08579617279996123,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011670787078707872,
      "loss": 3.1681,
      "step": 416520
    },
    {
      "epoch": 0.0858148160361194,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001167058705870587,
      "loss": 3.1983,
      "step": 416530
    },
    {
      "epoch": 0.08583345927227759,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011670387038703871,
      "loss": 3.088,
      "step": 416540
    },
    {
      "epoch": 0.08585210250843578,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011670187018701871,
      "loss": 3.0873,
      "step": 416550
    },
    {
      "epoch": 0.08587074574459395,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011669986998699872,
      "loss": 3.2605,
      "step": 416560
    },
    {
      "epoch": 0.08588938898075214,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001166978697869787,
      "loss": 3.2089,
      "step": 416570
    },
    {
      "epoch": 0.08590803221691032,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011669586958695871,
      "loss": 2.9992,
      "step": 416580
    },
    {
      "epoch": 0.08592667545306851,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011669386938693869,
      "loss": 3.1738,
      "step": 416590
    },
    {
      "epoch": 0.0859453186892267,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001166918691869187,
      "loss": 3.1494,
      "step": 416600
    },
    {
      "epoch": 0.08596396192538487,
      "grad_norm": 0.625,
      "learning_rate": 0.00011668986898689868,
      "loss": 3.1769,
      "step": 416610
    },
    {
      "epoch": 0.08598260516154306,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011668786878687869,
      "loss": 3.1337,
      "step": 416620
    },
    {
      "epoch": 0.08600124839770125,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011668586858685869,
      "loss": 3.203,
      "step": 416630
    },
    {
      "epoch": 0.08601989163385942,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001166838683868387,
      "loss": 3.0786,
      "step": 416640
    },
    {
      "epoch": 0.08603853487001761,
      "grad_norm": 0.75,
      "learning_rate": 0.00011668186818681868,
      "loss": 3.1928,
      "step": 416650
    },
    {
      "epoch": 0.08605717810617579,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011667986798679869,
      "loss": 3.2351,
      "step": 416660
    },
    {
      "epoch": 0.08607582134233398,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001166778677867787,
      "loss": 2.9663,
      "step": 416670
    },
    {
      "epoch": 0.08609446457849217,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011667586758675868,
      "loss": 3.2124,
      "step": 416680
    },
    {
      "epoch": 0.08611310781465034,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011667386738673869,
      "loss": 3.0788,
      "step": 416690
    },
    {
      "epoch": 0.08613175105080853,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011667186718671867,
      "loss": 3.229,
      "step": 416700
    },
    {
      "epoch": 0.0861503942869667,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011666986698669868,
      "loss": 3.065,
      "step": 416710
    },
    {
      "epoch": 0.0861690375231249,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011666786678667867,
      "loss": 3.1813,
      "step": 416720
    },
    {
      "epoch": 0.08618768075928308,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011666586658665868,
      "loss": 3.3473,
      "step": 416730
    },
    {
      "epoch": 0.08620632399544126,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011666386638663866,
      "loss": 3.2073,
      "step": 416740
    },
    {
      "epoch": 0.08622496723159945,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011666186618661867,
      "loss": 3.3203,
      "step": 416750
    },
    {
      "epoch": 0.08624361046775764,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011665986598659866,
      "loss": 3.06,
      "step": 416760
    },
    {
      "epoch": 0.08626225370391581,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011665786578657867,
      "loss": 3.107,
      "step": 416770
    },
    {
      "epoch": 0.086280896940074,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011665586558655866,
      "loss": 3.151,
      "step": 416780
    },
    {
      "epoch": 0.08629954017623218,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011665386538653867,
      "loss": 3.168,
      "step": 416790
    },
    {
      "epoch": 0.08631818341239036,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011665186518651865,
      "loss": 3.1276,
      "step": 416800
    },
    {
      "epoch": 0.08633682664854855,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011664986498649866,
      "loss": 3.0752,
      "step": 416810
    },
    {
      "epoch": 0.08635546988470673,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011664786478647864,
      "loss": 3.2276,
      "step": 416820
    },
    {
      "epoch": 0.08637411312086492,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011664586458645865,
      "loss": 3.1503,
      "step": 416830
    },
    {
      "epoch": 0.08639275635702309,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011664386438643863,
      "loss": 3.2038,
      "step": 416840
    },
    {
      "epoch": 0.08641139959318128,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011664186418641864,
      "loss": 3.2098,
      "step": 416850
    },
    {
      "epoch": 0.08643004282933947,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011663986398639865,
      "loss": 3.2084,
      "step": 416860
    },
    {
      "epoch": 0.08644868606549765,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011663786378637865,
      "loss": 3.1845,
      "step": 416870
    },
    {
      "epoch": 0.08646732930165583,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011663586358635866,
      "loss": 3.0529,
      "step": 416880
    },
    {
      "epoch": 0.08648597253781402,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011663386338633864,
      "loss": 3.0648,
      "step": 416890
    },
    {
      "epoch": 0.0865046157739722,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011663186318631865,
      "loss": 3.0024,
      "step": 416900
    },
    {
      "epoch": 0.08652325901013039,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011662986298629863,
      "loss": 3.1946,
      "step": 416910
    },
    {
      "epoch": 0.08654190224628856,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011662786278627864,
      "loss": 3.1174,
      "step": 416920
    },
    {
      "epoch": 0.08656054548244675,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011662586258625862,
      "loss": 3.2524,
      "step": 416930
    },
    {
      "epoch": 0.08657918871860494,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011662386238623863,
      "loss": 3.1453,
      "step": 416940
    },
    {
      "epoch": 0.08659783195476312,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011662186218621863,
      "loss": 3.275,
      "step": 416950
    },
    {
      "epoch": 0.0866164751909213,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011661986198619864,
      "loss": 3.2101,
      "step": 416960
    },
    {
      "epoch": 0.0866351184270795,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011661786178617862,
      "loss": 3.0381,
      "step": 416970
    },
    {
      "epoch": 0.08665376166323767,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011661586158615863,
      "loss": 3.1923,
      "step": 416980
    },
    {
      "epoch": 0.08667240489939586,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011661386138613861,
      "loss": 3.001,
      "step": 416990
    },
    {
      "epoch": 0.08669104813555403,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011661186118611862,
      "loss": 3.2175,
      "step": 417000
    },
    {
      "epoch": 0.08670969137171222,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011660986098609861,
      "loss": 3.1807,
      "step": 417010
    },
    {
      "epoch": 0.08672833460787041,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011660786078607861,
      "loss": 3.167,
      "step": 417020
    },
    {
      "epoch": 0.08674697784402859,
      "grad_norm": 0.75,
      "learning_rate": 0.0001166058605860586,
      "loss": 3.0987,
      "step": 417030
    },
    {
      "epoch": 0.08676562108018677,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011660386038603861,
      "loss": 3.1416,
      "step": 417040
    },
    {
      "epoch": 0.08678426431634495,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001166018601860186,
      "loss": 3.2982,
      "step": 417050
    },
    {
      "epoch": 0.08680290755250314,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001165998599859986,
      "loss": 3.1782,
      "step": 417060
    },
    {
      "epoch": 0.08682155078866133,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011659785978597861,
      "loss": 3.0807,
      "step": 417070
    },
    {
      "epoch": 0.0868401940248195,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001165958595859586,
      "loss": 3.2958,
      "step": 417080
    },
    {
      "epoch": 0.08685883726097769,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001165938593859386,
      "loss": 3.0854,
      "step": 417090
    },
    {
      "epoch": 0.08687748049713588,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001165918591859186,
      "loss": 3.0709,
      "step": 417100
    },
    {
      "epoch": 0.08689612373329406,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011658985898589861,
      "loss": 3.2244,
      "step": 417110
    },
    {
      "epoch": 0.08691476696945225,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011658785878587859,
      "loss": 3.2871,
      "step": 417120
    },
    {
      "epoch": 0.08693341020561042,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001165858585858586,
      "loss": 3.0364,
      "step": 417130
    },
    {
      "epoch": 0.08695205344176861,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011658385838583858,
      "loss": 3.1913,
      "step": 417140
    },
    {
      "epoch": 0.0869706966779268,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011658185818581859,
      "loss": 3.2184,
      "step": 417150
    },
    {
      "epoch": 0.08698933991408497,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011657985798579857,
      "loss": 2.9827,
      "step": 417160
    },
    {
      "epoch": 0.08700798315024316,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011657785778577858,
      "loss": 3.2081,
      "step": 417170
    },
    {
      "epoch": 0.08702662638640134,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011657585758575858,
      "loss": 3.2234,
      "step": 417180
    },
    {
      "epoch": 0.08704526962255953,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011657385738573859,
      "loss": 3.1713,
      "step": 417190
    },
    {
      "epoch": 0.08706391285871772,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011657185718571857,
      "loss": 3.2638,
      "step": 417200
    },
    {
      "epoch": 0.08708255609487589,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011656985698569858,
      "loss": 3.1493,
      "step": 417210
    },
    {
      "epoch": 0.08710119933103408,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011656785678567856,
      "loss": 3.1045,
      "step": 417220
    },
    {
      "epoch": 0.08711984256719227,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011656585658565857,
      "loss": 3.1497,
      "step": 417230
    },
    {
      "epoch": 0.08713848580335044,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011656385638563855,
      "loss": 3.1211,
      "step": 417240
    },
    {
      "epoch": 0.08715712903950863,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011656185618561856,
      "loss": 3.1799,
      "step": 417250
    },
    {
      "epoch": 0.08717577227566681,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011655985598559857,
      "loss": 3.2017,
      "step": 417260
    },
    {
      "epoch": 0.087194415511825,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011655785578557857,
      "loss": 3.225,
      "step": 417270
    },
    {
      "epoch": 0.08721305874798319,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011655585558555858,
      "loss": 3.2631,
      "step": 417280
    },
    {
      "epoch": 0.08723170198414136,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011655385538553856,
      "loss": 3.1777,
      "step": 417290
    },
    {
      "epoch": 0.08725034522029955,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011655185518551857,
      "loss": 3.0604,
      "step": 417300
    },
    {
      "epoch": 0.08726898845645774,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011654985498549855,
      "loss": 3.2023,
      "step": 417310
    },
    {
      "epoch": 0.08728763169261591,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011654785478547856,
      "loss": 3.1578,
      "step": 417320
    },
    {
      "epoch": 0.0873062749287741,
      "grad_norm": 0.921875,
      "learning_rate": 0.00011654585458545854,
      "loss": 3.062,
      "step": 417330
    },
    {
      "epoch": 0.08732491816493228,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011654385438543855,
      "loss": 3.1629,
      "step": 417340
    },
    {
      "epoch": 0.08734356140109047,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011654185418541855,
      "loss": 3.0863,
      "step": 417350
    },
    {
      "epoch": 0.08736220463724866,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011653985398539855,
      "loss": 3.1867,
      "step": 417360
    },
    {
      "epoch": 0.08738084787340683,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011653785378537854,
      "loss": 3.3107,
      "step": 417370
    },
    {
      "epoch": 0.08739949110956502,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011653585358535855,
      "loss": 3.0556,
      "step": 417380
    },
    {
      "epoch": 0.0874181343457232,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011653385338533853,
      "loss": 3.1563,
      "step": 417390
    },
    {
      "epoch": 0.08743677758188138,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011653185318531854,
      "loss": 3.0721,
      "step": 417400
    },
    {
      "epoch": 0.08745542081803957,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011652985298529853,
      "loss": 3.3323,
      "step": 417410
    },
    {
      "epoch": 0.08747406405419775,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011652785278527854,
      "loss": 3.0943,
      "step": 417420
    },
    {
      "epoch": 0.08749270729035594,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011652585258525852,
      "loss": 3.163,
      "step": 417430
    },
    {
      "epoch": 0.08751135052651413,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011652385238523853,
      "loss": 3.0532,
      "step": 417440
    },
    {
      "epoch": 0.0875299937626723,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011652185218521854,
      "loss": 3.1702,
      "step": 417450
    },
    {
      "epoch": 0.08754863699883049,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011651985198519852,
      "loss": 3.1779,
      "step": 417460
    },
    {
      "epoch": 0.08756728023498866,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011651785178517853,
      "loss": 3.1223,
      "step": 417470
    },
    {
      "epoch": 0.08758592347114685,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011651585158515851,
      "loss": 3.245,
      "step": 417480
    },
    {
      "epoch": 0.08760456670730504,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011651385138513852,
      "loss": 3.1634,
      "step": 417490
    },
    {
      "epoch": 0.08762320994346322,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011651185118511852,
      "loss": 3.1869,
      "step": 417500
    },
    {
      "epoch": 0.0876418531796214,
      "grad_norm": 0.75,
      "learning_rate": 0.00011650985098509853,
      "loss": 3.1219,
      "step": 417510
    },
    {
      "epoch": 0.08766049641577958,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011650785078507851,
      "loss": 2.9534,
      "step": 417520
    },
    {
      "epoch": 0.08767913965193777,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011650585058505852,
      "loss": 3.2766,
      "step": 417530
    },
    {
      "epoch": 0.08769778288809596,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001165038503850385,
      "loss": 3.2586,
      "step": 417540
    },
    {
      "epoch": 0.08771642612425413,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011650185018501851,
      "loss": 3.0052,
      "step": 417550
    },
    {
      "epoch": 0.08773506936041232,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011649984998499849,
      "loss": 3.1053,
      "step": 417560
    },
    {
      "epoch": 0.08775371259657051,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001164978497849785,
      "loss": 2.9868,
      "step": 417570
    },
    {
      "epoch": 0.08777235583272869,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001164958495849585,
      "loss": 3.0666,
      "step": 417580
    },
    {
      "epoch": 0.08779099906888688,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011649384938493851,
      "loss": 3.1297,
      "step": 417590
    },
    {
      "epoch": 0.08780964230504505,
      "grad_norm": 0.578125,
      "learning_rate": 0.00011649184918491849,
      "loss": 3.3772,
      "step": 417600
    },
    {
      "epoch": 0.08782828554120324,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001164898489848985,
      "loss": 3.1609,
      "step": 417610
    },
    {
      "epoch": 0.08784692877736143,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011648784878487848,
      "loss": 3.0118,
      "step": 417620
    },
    {
      "epoch": 0.0878655720135196,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011648584858485849,
      "loss": 3.1594,
      "step": 417630
    },
    {
      "epoch": 0.0878842152496778,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011648384838483849,
      "loss": 3.146,
      "step": 417640
    },
    {
      "epoch": 0.08790285848583598,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011648184818481848,
      "loss": 3.2543,
      "step": 417650
    },
    {
      "epoch": 0.08792150172199416,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011647984798479849,
      "loss": 3.1775,
      "step": 417660
    },
    {
      "epoch": 0.08794014495815235,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011647784778477849,
      "loss": 3.0846,
      "step": 417670
    },
    {
      "epoch": 0.08795878819431052,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001164758475847585,
      "loss": 3.1982,
      "step": 417680
    },
    {
      "epoch": 0.08797743143046871,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011647384738473848,
      "loss": 3.187,
      "step": 417690
    },
    {
      "epoch": 0.0879960746666269,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011647184718471849,
      "loss": 3.092,
      "step": 417700
    },
    {
      "epoch": 0.08801471790278508,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011646984698469847,
      "loss": 3.1223,
      "step": 417710
    },
    {
      "epoch": 0.08803336113894326,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011646784678467848,
      "loss": 3.1195,
      "step": 417720
    },
    {
      "epoch": 0.08805200437510144,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011646584658465847,
      "loss": 3.1469,
      "step": 417730
    },
    {
      "epoch": 0.08807064761125963,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011646384638463848,
      "loss": 3.1506,
      "step": 417740
    },
    {
      "epoch": 0.08808929084741782,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011646184618461846,
      "loss": 3.2123,
      "step": 417750
    },
    {
      "epoch": 0.08810793408357599,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011645984598459847,
      "loss": 3.298,
      "step": 417760
    },
    {
      "epoch": 0.08812657731973418,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011645784578457845,
      "loss": 3.1133,
      "step": 417770
    },
    {
      "epoch": 0.08814522055589237,
      "grad_norm": 0.75,
      "learning_rate": 0.00011645584558455846,
      "loss": 3.1244,
      "step": 417780
    },
    {
      "epoch": 0.08816386379205055,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011645384538453845,
      "loss": 3.0402,
      "step": 417790
    },
    {
      "epoch": 0.08818250702820873,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011645184518451846,
      "loss": 3.2202,
      "step": 417800
    },
    {
      "epoch": 0.08820115026436691,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00011644984498449845,
      "loss": 3.1843,
      "step": 417810
    },
    {
      "epoch": 0.0882197935005251,
      "grad_norm": 0.75,
      "learning_rate": 0.00011644784478447846,
      "loss": 3.1931,
      "step": 417820
    },
    {
      "epoch": 0.08823843673668329,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011644584458445844,
      "loss": 3.1288,
      "step": 417830
    },
    {
      "epoch": 0.08825707997284146,
      "grad_norm": 0.75,
      "learning_rate": 0.00011644384438443845,
      "loss": 3.2302,
      "step": 417840
    },
    {
      "epoch": 0.08827572320899965,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011644184418441846,
      "loss": 3.2151,
      "step": 417850
    },
    {
      "epoch": 0.08829436644515783,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011643984398439844,
      "loss": 3.2796,
      "step": 417860
    },
    {
      "epoch": 0.08831300968131602,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011643784378437845,
      "loss": 3.2479,
      "step": 417870
    },
    {
      "epoch": 0.0883316529174742,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011643584358435843,
      "loss": 3.1255,
      "step": 417880
    },
    {
      "epoch": 0.08835029615363238,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011643384338433844,
      "loss": 3.0696,
      "step": 417890
    },
    {
      "epoch": 0.08836893938979057,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011643184318431844,
      "loss": 3.1107,
      "step": 417900
    },
    {
      "epoch": 0.08838758262594876,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011642984298429845,
      "loss": 3.2069,
      "step": 417910
    },
    {
      "epoch": 0.08840622586210693,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011642784278427843,
      "loss": 3.1426,
      "step": 417920
    },
    {
      "epoch": 0.08842486909826512,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011642584258425844,
      "loss": 3.0806,
      "step": 417930
    },
    {
      "epoch": 0.0884435123344233,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011642384238423842,
      "loss": 3.2952,
      "step": 417940
    },
    {
      "epoch": 0.08846215557058149,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011642184218421843,
      "loss": 3.2029,
      "step": 417950
    },
    {
      "epoch": 0.08848079880673967,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011641984198419843,
      "loss": 3.234,
      "step": 417960
    },
    {
      "epoch": 0.08849944204289785,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011641784178417842,
      "loss": 3.1742,
      "step": 417970
    },
    {
      "epoch": 0.08851808527905604,
      "grad_norm": 0.75,
      "learning_rate": 0.00011641584158415842,
      "loss": 3.1983,
      "step": 417980
    },
    {
      "epoch": 0.08853672851521423,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011641384138413843,
      "loss": 3.0986,
      "step": 417990
    },
    {
      "epoch": 0.0885553717513724,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011641184118411841,
      "loss": 3.1414,
      "step": 418000
    },
    {
      "epoch": 0.08857401498753059,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011640984098409842,
      "loss": 3.1994,
      "step": 418010
    },
    {
      "epoch": 0.08859265822368877,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001164078407840784,
      "loss": 3.1505,
      "step": 418020
    },
    {
      "epoch": 0.08861130145984696,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011640584058405841,
      "loss": 3.1019,
      "step": 418030
    },
    {
      "epoch": 0.08862994469600514,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001164038403840384,
      "loss": 3.194,
      "step": 418040
    },
    {
      "epoch": 0.08864858793216332,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011640184018401841,
      "loss": 3.0244,
      "step": 418050
    },
    {
      "epoch": 0.08866723116832151,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011639983998399842,
      "loss": 3.2954,
      "step": 418060
    },
    {
      "epoch": 0.08868587440447968,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001163978397839784,
      "loss": 2.9857,
      "step": 418070
    },
    {
      "epoch": 0.08870451764063787,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011639583958395841,
      "loss": 3.2252,
      "step": 418080
    },
    {
      "epoch": 0.08872316087679606,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001163938393839384,
      "loss": 3.1597,
      "step": 418090
    },
    {
      "epoch": 0.08874180411295424,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001163918391839184,
      "loss": 3.1552,
      "step": 418100
    },
    {
      "epoch": 0.08876044734911243,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011638983898389839,
      "loss": 3.0657,
      "step": 418110
    },
    {
      "epoch": 0.08877909058527061,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001163878387838784,
      "loss": 3.204,
      "step": 418120
    },
    {
      "epoch": 0.08879773382142879,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011638583858385839,
      "loss": 3.0845,
      "step": 418130
    },
    {
      "epoch": 0.08881637705758698,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001163838383838384,
      "loss": 3.1007,
      "step": 418140
    },
    {
      "epoch": 0.08883502029374515,
      "grad_norm": 0.625,
      "learning_rate": 0.00011638183818381838,
      "loss": 3.2362,
      "step": 418150
    },
    {
      "epoch": 0.08885366352990334,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011637983798379839,
      "loss": 3.259,
      "step": 418160
    },
    {
      "epoch": 0.08887230676606153,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011637783778377837,
      "loss": 3.2167,
      "step": 418170
    },
    {
      "epoch": 0.0888909500022197,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011637583758375838,
      "loss": 3.1417,
      "step": 418180
    },
    {
      "epoch": 0.0889095932383779,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011637383738373836,
      "loss": 3.1218,
      "step": 418190
    },
    {
      "epoch": 0.08892823647453607,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011637183718371837,
      "loss": 3.2312,
      "step": 418200
    },
    {
      "epoch": 0.08894687971069426,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011636983698369837,
      "loss": 3.1631,
      "step": 418210
    },
    {
      "epoch": 0.08896552294685245,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011636783678367838,
      "loss": 3.1851,
      "step": 418220
    },
    {
      "epoch": 0.08898416618301062,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011636583658365836,
      "loss": 3.1427,
      "step": 418230
    },
    {
      "epoch": 0.08900280941916881,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011636383638363837,
      "loss": 3.1152,
      "step": 418240
    },
    {
      "epoch": 0.089021452655327,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011636183618361838,
      "loss": 3.1383,
      "step": 418250
    },
    {
      "epoch": 0.08904009589148518,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011635983598359836,
      "loss": 3.1743,
      "step": 418260
    },
    {
      "epoch": 0.08905873912764337,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011635783578357837,
      "loss": 3.2182,
      "step": 418270
    },
    {
      "epoch": 0.08907738236380154,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011635583558355837,
      "loss": 3.202,
      "step": 418280
    },
    {
      "epoch": 0.08909602559995973,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011635383538353836,
      "loss": 3.2735,
      "step": 418290
    },
    {
      "epoch": 0.08911466883611792,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011635183518351836,
      "loss": 3.1973,
      "step": 418300
    },
    {
      "epoch": 0.0891333120722761,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011634983498349837,
      "loss": 3.1961,
      "step": 418310
    },
    {
      "epoch": 0.08915195530843428,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011634783478347835,
      "loss": 3.1013,
      "step": 418320
    },
    {
      "epoch": 0.08917059854459247,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011634583458345836,
      "loss": 3.1172,
      "step": 418330
    },
    {
      "epoch": 0.08918924178075065,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011634383438343834,
      "loss": 3.1804,
      "step": 418340
    },
    {
      "epoch": 0.08920788501690884,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011634183418341835,
      "loss": 3.2156,
      "step": 418350
    },
    {
      "epoch": 0.08922652825306701,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011633983398339834,
      "loss": 3.096,
      "step": 418360
    },
    {
      "epoch": 0.0892451714892252,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011633783378337835,
      "loss": 3.2358,
      "step": 418370
    },
    {
      "epoch": 0.08926381472538339,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011633583358335834,
      "loss": 3.2093,
      "step": 418380
    },
    {
      "epoch": 0.08928245796154156,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011633383338333834,
      "loss": 3.1893,
      "step": 418390
    },
    {
      "epoch": 0.08930110119769975,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011633183318331833,
      "loss": 2.9991,
      "step": 418400
    },
    {
      "epoch": 0.08931974443385793,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011632983298329834,
      "loss": 3.2532,
      "step": 418410
    },
    {
      "epoch": 0.08933838767001612,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011632783278327832,
      "loss": 3.2849,
      "step": 418420
    },
    {
      "epoch": 0.0893570309061743,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011632583258325833,
      "loss": 3.1396,
      "step": 418430
    },
    {
      "epoch": 0.08937567414233248,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011632383238323834,
      "loss": 3.0507,
      "step": 418440
    },
    {
      "epoch": 0.08939431737849067,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011632183218321833,
      "loss": 3.1805,
      "step": 418450
    },
    {
      "epoch": 0.08941296061464886,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011631983198319834,
      "loss": 3.0291,
      "step": 418460
    },
    {
      "epoch": 0.08943160385080703,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011631783178317832,
      "loss": 3.1849,
      "step": 418470
    },
    {
      "epoch": 0.08945024708696522,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011631583158315833,
      "loss": 3.0008,
      "step": 418480
    },
    {
      "epoch": 0.0894688903231234,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011631383138313831,
      "loss": 3.305,
      "step": 418490
    },
    {
      "epoch": 0.08948753355928159,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011631183118311832,
      "loss": 3.1706,
      "step": 418500
    },
    {
      "epoch": 0.08950617679543978,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001163098309830983,
      "loss": 3.2019,
      "step": 418510
    },
    {
      "epoch": 0.08952482003159795,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011630783078307831,
      "loss": 3.1999,
      "step": 418520
    },
    {
      "epoch": 0.08954346326775614,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011630583058305831,
      "loss": 3.0137,
      "step": 418530
    },
    {
      "epoch": 0.08956210650391432,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011630383038303832,
      "loss": 3.0396,
      "step": 418540
    },
    {
      "epoch": 0.0895807497400725,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001163018301830183,
      "loss": 3.3025,
      "step": 418550
    },
    {
      "epoch": 0.0895993929762307,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011629982998299831,
      "loss": 3.3753,
      "step": 418560
    },
    {
      "epoch": 0.08961803621238887,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011629782978297829,
      "loss": 3.2529,
      "step": 418570
    },
    {
      "epoch": 0.08963667944854706,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001162958295829583,
      "loss": 3.2153,
      "step": 418580
    },
    {
      "epoch": 0.08965532268470525,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001162938293829383,
      "loss": 3.1797,
      "step": 418590
    },
    {
      "epoch": 0.08967396592086342,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001162918291829183,
      "loss": 3.2231,
      "step": 418600
    },
    {
      "epoch": 0.08969260915702161,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011628982898289829,
      "loss": 3.186,
      "step": 418610
    },
    {
      "epoch": 0.08971125239317979,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001162878287828783,
      "loss": 3.2126,
      "step": 418620
    },
    {
      "epoch": 0.08972989562933797,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011628582858285828,
      "loss": 2.9959,
      "step": 418630
    },
    {
      "epoch": 0.08974853886549616,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011628382838283829,
      "loss": 3.1927,
      "step": 418640
    },
    {
      "epoch": 0.08976718210165434,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001162818281828183,
      "loss": 3.1753,
      "step": 418650
    },
    {
      "epoch": 0.08978582533781253,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011627982798279828,
      "loss": 3.1689,
      "step": 418660
    },
    {
      "epoch": 0.08980446857397072,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011627782778277829,
      "loss": 3.0702,
      "step": 418670
    },
    {
      "epoch": 0.08982311181012889,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011627582758275828,
      "loss": 3.2392,
      "step": 418680
    },
    {
      "epoch": 0.08984175504628708,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001162738273827383,
      "loss": 3.1663,
      "step": 418690
    },
    {
      "epoch": 0.08986039828244526,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011627182718271828,
      "loss": 3.1017,
      "step": 418700
    },
    {
      "epoch": 0.08987904151860344,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011626982698269828,
      "loss": 3.1544,
      "step": 418710
    },
    {
      "epoch": 0.08989768475476163,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011626782678267827,
      "loss": 3.1488,
      "step": 418720
    },
    {
      "epoch": 0.08991632799091981,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011626582658265828,
      "loss": 3.221,
      "step": 418730
    },
    {
      "epoch": 0.089934971227078,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011626382638263826,
      "loss": 3.1597,
      "step": 418740
    },
    {
      "epoch": 0.08995361446323617,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011626182618261827,
      "loss": 3.1207,
      "step": 418750
    },
    {
      "epoch": 0.08997225769939436,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011625982598259826,
      "loss": 3.0643,
      "step": 418760
    },
    {
      "epoch": 0.08999090093555255,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011625782578257827,
      "loss": 3.2844,
      "step": 418770
    },
    {
      "epoch": 0.09000954417171073,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011625582558255825,
      "loss": 3.0097,
      "step": 418780
    },
    {
      "epoch": 0.09002818740786891,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011625382538253826,
      "loss": 3.2276,
      "step": 418790
    },
    {
      "epoch": 0.0900468306440271,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011625182518251825,
      "loss": 3.0359,
      "step": 418800
    },
    {
      "epoch": 0.09006547388018528,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011624982498249825,
      "loss": 3.1831,
      "step": 418810
    },
    {
      "epoch": 0.09008411711634347,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011624782478247824,
      "loss": 3.1777,
      "step": 418820
    },
    {
      "epoch": 0.09010276035250164,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011624582458245825,
      "loss": 3.2148,
      "step": 418830
    },
    {
      "epoch": 0.09012140358865983,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011624382438243825,
      "loss": 3.2173,
      "step": 418840
    },
    {
      "epoch": 0.09014004682481802,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011624182418241825,
      "loss": 3.1825,
      "step": 418850
    },
    {
      "epoch": 0.0901586900609762,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011623982398239826,
      "loss": 3.0884,
      "step": 418860
    },
    {
      "epoch": 0.09017733329713438,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011623782378237824,
      "loss": 3.2367,
      "step": 418870
    },
    {
      "epoch": 0.09019597653329256,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011623582358235825,
      "loss": 3.142,
      "step": 418880
    },
    {
      "epoch": 0.09021461976945075,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011623382338233823,
      "loss": 3.1369,
      "step": 418890
    },
    {
      "epoch": 0.09023326300560894,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011623182318231824,
      "loss": 3.0643,
      "step": 418900
    },
    {
      "epoch": 0.09025190624176711,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011622982298229824,
      "loss": 3.2099,
      "step": 418910
    },
    {
      "epoch": 0.0902705494779253,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011622782278227823,
      "loss": 3.0813,
      "step": 418920
    },
    {
      "epoch": 0.09028919271408349,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011622582258225823,
      "loss": 3.0766,
      "step": 418930
    },
    {
      "epoch": 0.09030783595024167,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011622382238223824,
      "loss": 3.1961,
      "step": 418940
    },
    {
      "epoch": 0.09032647918639986,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011622182218221822,
      "loss": 2.8544,
      "step": 418950
    },
    {
      "epoch": 0.09034512242255803,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011621982198219823,
      "loss": 3.2229,
      "step": 418960
    },
    {
      "epoch": 0.09036376565871622,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011621782178217821,
      "loss": 3.1257,
      "step": 418970
    },
    {
      "epoch": 0.09038240889487441,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011621582158215822,
      "loss": 3.2233,
      "step": 418980
    },
    {
      "epoch": 0.09040105213103258,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011621382138213822,
      "loss": 3.1103,
      "step": 418990
    },
    {
      "epoch": 0.09041969536719077,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011621182118211822,
      "loss": 3.0744,
      "step": 419000
    },
    {
      "epoch": 0.09043833860334896,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001162098209820982,
      "loss": 3.1013,
      "step": 419010
    },
    {
      "epoch": 0.09045698183950714,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011620782078207822,
      "loss": 3.2232,
      "step": 419020
    },
    {
      "epoch": 0.09047562507566533,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011620582058205823,
      "loss": 3.1493,
      "step": 419030
    },
    {
      "epoch": 0.0904942683118235,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011620382038203821,
      "loss": 3.1353,
      "step": 419040
    },
    {
      "epoch": 0.09051291154798169,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011620182018201822,
      "loss": 3.1173,
      "step": 419050
    },
    {
      "epoch": 0.09053155478413988,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001161998199819982,
      "loss": 3.2282,
      "step": 419060
    },
    {
      "epoch": 0.09055019802029805,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011619781978197821,
      "loss": 3.1445,
      "step": 419070
    },
    {
      "epoch": 0.09056884125645624,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001161958195819582,
      "loss": 3.147,
      "step": 419080
    },
    {
      "epoch": 0.09058748449261442,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011619381938193821,
      "loss": 3.131,
      "step": 419090
    },
    {
      "epoch": 0.0906061277287726,
      "grad_norm": 0.75,
      "learning_rate": 0.0001161918191819182,
      "loss": 3.0537,
      "step": 419100
    },
    {
      "epoch": 0.0906247709649308,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001161898189818982,
      "loss": 3.2634,
      "step": 419110
    },
    {
      "epoch": 0.09064341420108897,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011618781878187819,
      "loss": 3.0574,
      "step": 419120
    },
    {
      "epoch": 0.09066205743724716,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001161858185818582,
      "loss": 3.1107,
      "step": 419130
    },
    {
      "epoch": 0.09068070067340535,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011618381838183818,
      "loss": 3.233,
      "step": 419140
    },
    {
      "epoch": 0.09069934390956352,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011618181818181819,
      "loss": 3.0959,
      "step": 419150
    },
    {
      "epoch": 0.09071798714572171,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011617981798179818,
      "loss": 3.0524,
      "step": 419160
    },
    {
      "epoch": 0.09073663038187989,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011617781778177819,
      "loss": 3.1427,
      "step": 419170
    },
    {
      "epoch": 0.09075527361803808,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011617581758175817,
      "loss": 3.0698,
      "step": 419180
    },
    {
      "epoch": 0.09077391685419627,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011617381738173818,
      "loss": 3.0698,
      "step": 419190
    },
    {
      "epoch": 0.09079256009035444,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011617181718171816,
      "loss": 3.1501,
      "step": 419200
    },
    {
      "epoch": 0.09081120332651263,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011616981698169817,
      "loss": 3.0923,
      "step": 419210
    },
    {
      "epoch": 0.0908298465626708,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011616781678167817,
      "loss": 3.1214,
      "step": 419220
    },
    {
      "epoch": 0.090848489798829,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011616581658165818,
      "loss": 3.1786,
      "step": 419230
    },
    {
      "epoch": 0.09086713303498718,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011616381638163817,
      "loss": 3.1265,
      "step": 419240
    },
    {
      "epoch": 0.09088577627114536,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011616181618161817,
      "loss": 3.1404,
      "step": 419250
    },
    {
      "epoch": 0.09090441950730355,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011615981598159818,
      "loss": 3.0712,
      "step": 419260
    },
    {
      "epoch": 0.09092306274346174,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011615781578157816,
      "loss": 2.9789,
      "step": 419270
    },
    {
      "epoch": 0.09094170597961991,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011615581558155817,
      "loss": 3.162,
      "step": 419280
    },
    {
      "epoch": 0.0909603492157781,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011615381538153815,
      "loss": 3.2264,
      "step": 419290
    },
    {
      "epoch": 0.09097899245193627,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011615181518151816,
      "loss": 3.0973,
      "step": 419300
    },
    {
      "epoch": 0.09099763568809446,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011614981498149816,
      "loss": 3.0823,
      "step": 419310
    },
    {
      "epoch": 0.09101627892425265,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011614781478147816,
      "loss": 3.1599,
      "step": 419320
    },
    {
      "epoch": 0.09103492216041083,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011614581458145815,
      "loss": 2.983,
      "step": 419330
    },
    {
      "epoch": 0.09105356539656902,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011614381438143816,
      "loss": 3.1813,
      "step": 419340
    },
    {
      "epoch": 0.0910722086327272,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011614181418141814,
      "loss": 3.1229,
      "step": 419350
    },
    {
      "epoch": 0.09109085186888538,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011613981398139815,
      "loss": 3.2093,
      "step": 419360
    },
    {
      "epoch": 0.09110949510504357,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011613781378137813,
      "loss": 3.0934,
      "step": 419370
    },
    {
      "epoch": 0.09112813834120174,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011613581358135814,
      "loss": 3.0616,
      "step": 419380
    },
    {
      "epoch": 0.09114678157735993,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011613381338133813,
      "loss": 3.1712,
      "step": 419390
    },
    {
      "epoch": 0.09116542481351812,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011613181318131814,
      "loss": 3.2142,
      "step": 419400
    },
    {
      "epoch": 0.0911840680496763,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011612981298129813,
      "loss": 3.2159,
      "step": 419410
    },
    {
      "epoch": 0.09120271128583449,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011612781278127813,
      "loss": 3.1762,
      "step": 419420
    },
    {
      "epoch": 0.09122135452199266,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011612581258125814,
      "loss": 3.1097,
      "step": 419430
    },
    {
      "epoch": 0.09123999775815085,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011612381238123813,
      "loss": 3.1084,
      "step": 419440
    },
    {
      "epoch": 0.09125864099430904,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011612181218121813,
      "loss": 3.1593,
      "step": 419450
    },
    {
      "epoch": 0.09127728423046721,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011611981198119812,
      "loss": 3.2041,
      "step": 419460
    },
    {
      "epoch": 0.0912959274666254,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011611781178117813,
      "loss": 3.2149,
      "step": 419470
    },
    {
      "epoch": 0.09131457070278359,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011611581158115812,
      "loss": 3.168,
      "step": 419480
    },
    {
      "epoch": 0.09133321393894177,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011611381138113813,
      "loss": 3.1041,
      "step": 419490
    },
    {
      "epoch": 0.09135185717509996,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011611181118111811,
      "loss": 3.0364,
      "step": 419500
    },
    {
      "epoch": 0.09137050041125813,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011610981098109812,
      "loss": 3.3867,
      "step": 419510
    },
    {
      "epoch": 0.09138914364741632,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001161078107810781,
      "loss": 3.1156,
      "step": 419520
    },
    {
      "epoch": 0.09140778688357451,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011610581058105811,
      "loss": 3.2441,
      "step": 419530
    },
    {
      "epoch": 0.09142643011973269,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011610381038103811,
      "loss": 3.1808,
      "step": 419540
    },
    {
      "epoch": 0.09144507335589087,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011610181018101812,
      "loss": 3.2184,
      "step": 419550
    },
    {
      "epoch": 0.09146371659204905,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001160998099809981,
      "loss": 3.1212,
      "step": 419560
    },
    {
      "epoch": 0.09148235982820724,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011609780978097811,
      "loss": 3.1406,
      "step": 419570
    },
    {
      "epoch": 0.09150100306436543,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011609580958095809,
      "loss": 3.1364,
      "step": 419580
    },
    {
      "epoch": 0.0915196463005236,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001160938093809381,
      "loss": 3.26,
      "step": 419590
    },
    {
      "epoch": 0.09153828953668179,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011609180918091808,
      "loss": 3.1813,
      "step": 419600
    },
    {
      "epoch": 0.09155693277283998,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011608980898089809,
      "loss": 3.1047,
      "step": 419610
    },
    {
      "epoch": 0.09157557600899816,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011608780878087809,
      "loss": 3.1729,
      "step": 419620
    },
    {
      "epoch": 0.09159421924515634,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001160858085808581,
      "loss": 3.1806,
      "step": 419630
    },
    {
      "epoch": 0.09161286248131452,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001160838083808381,
      "loss": 3.2252,
      "step": 419640
    },
    {
      "epoch": 0.09163150571747271,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011608180818081809,
      "loss": 3.1844,
      "step": 419650
    },
    {
      "epoch": 0.0916501489536309,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001160798079807981,
      "loss": 3.2109,
      "step": 419660
    },
    {
      "epoch": 0.09166879218978907,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011607780778077808,
      "loss": 3.2084,
      "step": 419670
    },
    {
      "epoch": 0.09168743542594726,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011607580758075809,
      "loss": 3.1672,
      "step": 419680
    },
    {
      "epoch": 0.09170607866210545,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011607380738073807,
      "loss": 3.0971,
      "step": 419690
    },
    {
      "epoch": 0.09172472189826363,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011607180718071808,
      "loss": 3.149,
      "step": 419700
    },
    {
      "epoch": 0.09174336513442181,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011606980698069807,
      "loss": 3.0972,
      "step": 419710
    },
    {
      "epoch": 0.09176200837057999,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011606780678067808,
      "loss": 3.404,
      "step": 419720
    },
    {
      "epoch": 0.09178065160673818,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011606580658065807,
      "loss": 3.158,
      "step": 419730
    },
    {
      "epoch": 0.09179929484289637,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011606380638063807,
      "loss": 3.1669,
      "step": 419740
    },
    {
      "epoch": 0.09181793807905454,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011606180618061806,
      "loss": 3.1687,
      "step": 419750
    },
    {
      "epoch": 0.09183658131521273,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011605980598059807,
      "loss": 3.093,
      "step": 419760
    },
    {
      "epoch": 0.0918552245513709,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011605780578057806,
      "loss": 3.0683,
      "step": 419770
    },
    {
      "epoch": 0.0918738677875291,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011605580558055806,
      "loss": 3.131,
      "step": 419780
    },
    {
      "epoch": 0.09189251102368728,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011605380538053805,
      "loss": 3.0871,
      "step": 419790
    },
    {
      "epoch": 0.09191115425984546,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011605180518051806,
      "loss": 3.1978,
      "step": 419800
    },
    {
      "epoch": 0.09192979749600365,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011604980498049804,
      "loss": 3.1503,
      "step": 419810
    },
    {
      "epoch": 0.09194844073216184,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011604780478047805,
      "loss": 3.0717,
      "step": 419820
    },
    {
      "epoch": 0.09196708396832001,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011604580458045806,
      "loss": 3.0478,
      "step": 419830
    },
    {
      "epoch": 0.0919857272044782,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011604380438043804,
      "loss": 3.1411,
      "step": 419840
    },
    {
      "epoch": 0.09200437044063638,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011604180418041805,
      "loss": 3.1208,
      "step": 419850
    },
    {
      "epoch": 0.09202301367679457,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011603980398039805,
      "loss": 3.3844,
      "step": 419860
    },
    {
      "epoch": 0.09204165691295275,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011603780378037806,
      "loss": 3.1036,
      "step": 419870
    },
    {
      "epoch": 0.09206030014911093,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011603580358035804,
      "loss": 3.2611,
      "step": 419880
    },
    {
      "epoch": 0.09207894338526912,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011603380338033805,
      "loss": 3.15,
      "step": 419890
    },
    {
      "epoch": 0.0920975866214273,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011603180318031803,
      "loss": 3.2456,
      "step": 419900
    },
    {
      "epoch": 0.09211622985758548,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011602980298029804,
      "loss": 3.2052,
      "step": 419910
    },
    {
      "epoch": 0.09213487309374367,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011602780278027802,
      "loss": 3.0826,
      "step": 419920
    },
    {
      "epoch": 0.09215351632990185,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011602580258025803,
      "loss": 3.2599,
      "step": 419930
    },
    {
      "epoch": 0.09217215956606004,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011602380238023803,
      "loss": 3.1157,
      "step": 419940
    },
    {
      "epoch": 0.09219080280221822,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011602180218021804,
      "loss": 3.1237,
      "step": 419950
    },
    {
      "epoch": 0.0922094460383764,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011601980198019802,
      "loss": 3.1862,
      "step": 419960
    },
    {
      "epoch": 0.09222808927453459,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011601780178017803,
      "loss": 3.139,
      "step": 419970
    },
    {
      "epoch": 0.09224673251069276,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011601580158015801,
      "loss": 3.1412,
      "step": 419980
    },
    {
      "epoch": 0.09226537574685095,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011601380138013802,
      "loss": 3.081,
      "step": 419990
    },
    {
      "epoch": 0.09228401898300914,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000116011801180118,
      "loss": 3.107,
      "step": 420000
    },
    {
      "epoch": 0.09230266221916732,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011600980098009801,
      "loss": 3.123,
      "step": 420010
    },
    {
      "epoch": 0.0923213054553255,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011600780078007802,
      "loss": 3.1803,
      "step": 420020
    },
    {
      "epoch": 0.0923399486914837,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011600580058005801,
      "loss": 3.0876,
      "step": 420030
    },
    {
      "epoch": 0.09235859192764187,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011600380038003802,
      "loss": 3.1067,
      "step": 420040
    },
    {
      "epoch": 0.09237723516380006,
      "grad_norm": 0.796875,
      "learning_rate": 0.000116001800180018,
      "loss": 3.1553,
      "step": 420050
    },
    {
      "epoch": 0.09239587839995823,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011599979997999802,
      "loss": 3.1886,
      "step": 420060
    },
    {
      "epoch": 0.09241452163611642,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000115997799779978,
      "loss": 3.1533,
      "step": 420070
    },
    {
      "epoch": 0.09243316487227461,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000115995799579958,
      "loss": 3.1299,
      "step": 420080
    },
    {
      "epoch": 0.09245180810843279,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000115993799379938,
      "loss": 3.2378,
      "step": 420090
    },
    {
      "epoch": 0.09247045134459098,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000115991799179918,
      "loss": 3.1291,
      "step": 420100
    },
    {
      "epoch": 0.09248909458074915,
      "grad_norm": 0.75,
      "learning_rate": 0.00011598979897989799,
      "loss": 3.0673,
      "step": 420110
    },
    {
      "epoch": 0.09250773781690734,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000115987798779878,
      "loss": 3.1488,
      "step": 420120
    },
    {
      "epoch": 0.09252638105306553,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011598579857985798,
      "loss": 3.0824,
      "step": 420130
    },
    {
      "epoch": 0.0925450242892237,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000115983798379838,
      "loss": 3.1222,
      "step": 420140
    },
    {
      "epoch": 0.09256366752538189,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011598179817981798,
      "loss": 3.194,
      "step": 420150
    },
    {
      "epoch": 0.09258231076154008,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011597979797979798,
      "loss": 3.1001,
      "step": 420160
    },
    {
      "epoch": 0.09260095399769826,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011597779777977798,
      "loss": 3.0197,
      "step": 420170
    },
    {
      "epoch": 0.09261959723385645,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011597579757975799,
      "loss": 3.1527,
      "step": 420180
    },
    {
      "epoch": 0.09263824047001462,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011597379737973797,
      "loss": 3.3044,
      "step": 420190
    },
    {
      "epoch": 0.09265688370617281,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011597179717971798,
      "loss": 3.0462,
      "step": 420200
    },
    {
      "epoch": 0.092675526942331,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011596979697969796,
      "loss": 3.1343,
      "step": 420210
    },
    {
      "epoch": 0.09269417017848917,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011596779677967797,
      "loss": 3.1006,
      "step": 420220
    },
    {
      "epoch": 0.09271281341464736,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011596579657965798,
      "loss": 3.1618,
      "step": 420230
    },
    {
      "epoch": 0.09273145665080554,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011596379637963796,
      "loss": 2.8445,
      "step": 420240
    },
    {
      "epoch": 0.09275009988696373,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011596179617961797,
      "loss": 3.0742,
      "step": 420250
    },
    {
      "epoch": 0.09276874312312192,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011595979597959797,
      "loss": 3.11,
      "step": 420260
    },
    {
      "epoch": 0.09278738635928009,
      "grad_norm": 0.75,
      "learning_rate": 0.00011595779577957798,
      "loss": 3.1866,
      "step": 420270
    },
    {
      "epoch": 0.09280602959543828,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011595579557955796,
      "loss": 3.222,
      "step": 420280
    },
    {
      "epoch": 0.09282467283159647,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011595379537953797,
      "loss": 3.1702,
      "step": 420290
    },
    {
      "epoch": 0.09284331606775464,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011595179517951795,
      "loss": 3.0503,
      "step": 420300
    },
    {
      "epoch": 0.09286195930391283,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011594979497949796,
      "loss": 3.0311,
      "step": 420310
    },
    {
      "epoch": 0.09288060254007101,
      "grad_norm": 0.625,
      "learning_rate": 0.00011594779477947794,
      "loss": 3.4243,
      "step": 420320
    },
    {
      "epoch": 0.0928992457762292,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011594579457945795,
      "loss": 3.0651,
      "step": 420330
    },
    {
      "epoch": 0.09291788901238739,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011594379437943795,
      "loss": 3.1202,
      "step": 420340
    },
    {
      "epoch": 0.09293653224854556,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011594179417941796,
      "loss": 3.2125,
      "step": 420350
    },
    {
      "epoch": 0.09295517548470375,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011593979397939794,
      "loss": 3.0888,
      "step": 420360
    },
    {
      "epoch": 0.09297381872086194,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011593779377937795,
      "loss": 3.1659,
      "step": 420370
    },
    {
      "epoch": 0.09299246195702011,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011593579357935793,
      "loss": 3.1585,
      "step": 420380
    },
    {
      "epoch": 0.0930111051931783,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011593379337933794,
      "loss": 3.1671,
      "step": 420390
    },
    {
      "epoch": 0.09302974842933648,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011593179317931793,
      "loss": 3.1104,
      "step": 420400
    },
    {
      "epoch": 0.09304839166549467,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011592979297929793,
      "loss": 3.2049,
      "step": 420410
    },
    {
      "epoch": 0.09306703490165286,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011592779277927794,
      "loss": 3.1335,
      "step": 420420
    },
    {
      "epoch": 0.09308567813781103,
      "grad_norm": 0.75,
      "learning_rate": 0.00011592579257925793,
      "loss": 3.0989,
      "step": 420430
    },
    {
      "epoch": 0.09310432137396922,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011592379237923794,
      "loss": 3.0899,
      "step": 420440
    },
    {
      "epoch": 0.0931229646101274,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011592179217921792,
      "loss": 3.2048,
      "step": 420450
    },
    {
      "epoch": 0.09314160784628558,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011591979197919793,
      "loss": 3.1185,
      "step": 420460
    },
    {
      "epoch": 0.09316025108244377,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011591779177917792,
      "loss": 3.1684,
      "step": 420470
    },
    {
      "epoch": 0.09317889431860195,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011591579157915792,
      "loss": 3.1457,
      "step": 420480
    },
    {
      "epoch": 0.09319753755476014,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011591379137913792,
      "loss": 3.1632,
      "step": 420490
    },
    {
      "epoch": 0.09321618079091833,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011591179117911793,
      "loss": 3.1593,
      "step": 420500
    },
    {
      "epoch": 0.0932348240270765,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011590979097909791,
      "loss": 3.092,
      "step": 420510
    },
    {
      "epoch": 0.09325346726323469,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011590779077907792,
      "loss": 3.1146,
      "step": 420520
    },
    {
      "epoch": 0.09327211049939287,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001159057905790579,
      "loss": 3.1871,
      "step": 420530
    },
    {
      "epoch": 0.09329075373555105,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011590379037903791,
      "loss": 3.1617,
      "step": 420540
    },
    {
      "epoch": 0.09330939697170924,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001159017901790179,
      "loss": 3.2964,
      "step": 420550
    },
    {
      "epoch": 0.09332804020786742,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001158997899789979,
      "loss": 2.9929,
      "step": 420560
    },
    {
      "epoch": 0.09334668344402561,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001158977897789779,
      "loss": 3.1822,
      "step": 420570
    },
    {
      "epoch": 0.09336532668018378,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011589578957895791,
      "loss": 3.2231,
      "step": 420580
    },
    {
      "epoch": 0.09338396991634197,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011589378937893789,
      "loss": 3.0946,
      "step": 420590
    },
    {
      "epoch": 0.09340261315250016,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001158917891789179,
      "loss": 3.2303,
      "step": 420600
    },
    {
      "epoch": 0.09342125638865834,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011588978897889791,
      "loss": 3.0644,
      "step": 420610
    },
    {
      "epoch": 0.09343989962481652,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011588778877887789,
      "loss": 3.2687,
      "step": 420620
    },
    {
      "epoch": 0.09345854286097471,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001158857885788579,
      "loss": 3.1221,
      "step": 420630
    },
    {
      "epoch": 0.09347718609713289,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011588378837883788,
      "loss": 3.1698,
      "step": 420640
    },
    {
      "epoch": 0.09349582933329108,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011588178817881789,
      "loss": 3.1302,
      "step": 420650
    },
    {
      "epoch": 0.09351447256944925,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011587978797879789,
      "loss": 3.1882,
      "step": 420660
    },
    {
      "epoch": 0.09353311580560744,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001158777877787779,
      "loss": 3.0575,
      "step": 420670
    },
    {
      "epoch": 0.09355175904176563,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011587578757875788,
      "loss": 3.1478,
      "step": 420680
    },
    {
      "epoch": 0.0935704022779238,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011587378737873789,
      "loss": 3.0338,
      "step": 420690
    },
    {
      "epoch": 0.093589045514082,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011587178717871787,
      "loss": 3.1514,
      "step": 420700
    },
    {
      "epoch": 0.09360768875024018,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011586978697869788,
      "loss": 3.0187,
      "step": 420710
    },
    {
      "epoch": 0.09362633198639836,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011586778677867787,
      "loss": 3.1092,
      "step": 420720
    },
    {
      "epoch": 0.09364497522255655,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011586578657865787,
      "loss": 3.075,
      "step": 420730
    },
    {
      "epoch": 0.09366361845871472,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011586378637863786,
      "loss": 3.225,
      "step": 420740
    },
    {
      "epoch": 0.09368226169487291,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011586178617861787,
      "loss": 3.029,
      "step": 420750
    },
    {
      "epoch": 0.0937009049310311,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011585978597859786,
      "loss": 3.1748,
      "step": 420760
    },
    {
      "epoch": 0.09371954816718928,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011585778577857786,
      "loss": 3.1011,
      "step": 420770
    },
    {
      "epoch": 0.09373819140334747,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011585578557855785,
      "loss": 3.2869,
      "step": 420780
    },
    {
      "epoch": 0.09375683463950564,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011585378537853786,
      "loss": 3.1874,
      "step": 420790
    },
    {
      "epoch": 0.09377547787566383,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011585178517851785,
      "loss": 3.2052,
      "step": 420800
    },
    {
      "epoch": 0.09379412111182202,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011584978497849786,
      "loss": 3.109,
      "step": 420810
    },
    {
      "epoch": 0.0938127643479802,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011584778477847787,
      "loss": 3.2584,
      "step": 420820
    },
    {
      "epoch": 0.09383140758413838,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011584578457845785,
      "loss": 3.1191,
      "step": 420830
    },
    {
      "epoch": 0.09385005082029657,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011584378437843786,
      "loss": 3.1084,
      "step": 420840
    },
    {
      "epoch": 0.09386869405645475,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011584178417841784,
      "loss": 2.9862,
      "step": 420850
    },
    {
      "epoch": 0.09388733729261294,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011583978397839785,
      "loss": 3.275,
      "step": 420860
    },
    {
      "epoch": 0.09390598052877111,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011583778377837783,
      "loss": 3.1607,
      "step": 420870
    },
    {
      "epoch": 0.0939246237649293,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011583578357835784,
      "loss": 3.125,
      "step": 420880
    },
    {
      "epoch": 0.09394326700108749,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011583378337833784,
      "loss": 3.1707,
      "step": 420890
    },
    {
      "epoch": 0.09396191023724566,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011583178317831785,
      "loss": 3.2368,
      "step": 420900
    },
    {
      "epoch": 0.09398055347340385,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011582978297829783,
      "loss": 3.1083,
      "step": 420910
    },
    {
      "epoch": 0.09399919670956203,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011582778277827784,
      "loss": 3.1141,
      "step": 420920
    },
    {
      "epoch": 0.09401783994572022,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011582578257825782,
      "loss": 3.0991,
      "step": 420930
    },
    {
      "epoch": 0.0940364831818784,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011582378237823783,
      "loss": 3.2074,
      "step": 420940
    },
    {
      "epoch": 0.09405512641803658,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011582178217821781,
      "loss": 3.1283,
      "step": 420950
    },
    {
      "epoch": 0.09407376965419477,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00011581978197819782,
      "loss": 3.1224,
      "step": 420960
    },
    {
      "epoch": 0.09409241289035296,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011581778177817782,
      "loss": 3.2365,
      "step": 420970
    },
    {
      "epoch": 0.09411105612651113,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011581578157815783,
      "loss": 3.0674,
      "step": 420980
    },
    {
      "epoch": 0.09412969936266932,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011581378137813781,
      "loss": 3.0709,
      "step": 420990
    },
    {
      "epoch": 0.0941483425988275,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011581178117811782,
      "loss": 3.147,
      "step": 421000
    },
    {
      "epoch": 0.09416698583498569,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011580978097809783,
      "loss": 3.0713,
      "step": 421010
    },
    {
      "epoch": 0.09418562907114388,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011580778077807781,
      "loss": 3.2286,
      "step": 421020
    },
    {
      "epoch": 0.09420427230730205,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011580578057805782,
      "loss": 3.1491,
      "step": 421030
    },
    {
      "epoch": 0.09422291554346024,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011580378037803781,
      "loss": 3.1455,
      "step": 421040
    },
    {
      "epoch": 0.09424155877961843,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011580178017801781,
      "loss": 3.2292,
      "step": 421050
    },
    {
      "epoch": 0.0942602020157766,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001157997799779978,
      "loss": 3.16,
      "step": 421060
    },
    {
      "epoch": 0.09427884525193479,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011579777977797781,
      "loss": 3.0459,
      "step": 421070
    },
    {
      "epoch": 0.09429748848809297,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001157957795779578,
      "loss": 3.0253,
      "step": 421080
    },
    {
      "epoch": 0.09431613172425116,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001157937793779378,
      "loss": 3.0267,
      "step": 421090
    },
    {
      "epoch": 0.09433477496040935,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011579177917791779,
      "loss": 2.9949,
      "step": 421100
    },
    {
      "epoch": 0.09435341819656752,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001157897789778978,
      "loss": 3.2673,
      "step": 421110
    },
    {
      "epoch": 0.09437206143272571,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011578777877787779,
      "loss": 3.2036,
      "step": 421120
    },
    {
      "epoch": 0.09439070466888388,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001157857785778578,
      "loss": 3.2241,
      "step": 421130
    },
    {
      "epoch": 0.09440934790504207,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011578377837783778,
      "loss": 3.1519,
      "step": 421140
    },
    {
      "epoch": 0.09442799114120026,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011578177817781779,
      "loss": 3.1953,
      "step": 421150
    },
    {
      "epoch": 0.09444663437735844,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011577977797779777,
      "loss": 3.2486,
      "step": 421160
    },
    {
      "epoch": 0.09446527761351663,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011577777777777778,
      "loss": 3.1154,
      "step": 421170
    },
    {
      "epoch": 0.09448392084967482,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011577577757775777,
      "loss": 3.1485,
      "step": 421180
    },
    {
      "epoch": 0.09450256408583299,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011577377737773777,
      "loss": 3.2379,
      "step": 421190
    },
    {
      "epoch": 0.09452120732199118,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011577177717771778,
      "loss": 3.1404,
      "step": 421200
    },
    {
      "epoch": 0.09453985055814935,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011576977697769778,
      "loss": 3.1892,
      "step": 421210
    },
    {
      "epoch": 0.09455849379430754,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011576777677767779,
      "loss": 3.1743,
      "step": 421220
    },
    {
      "epoch": 0.09457713703046573,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011576577657765777,
      "loss": 3.2432,
      "step": 421230
    },
    {
      "epoch": 0.09459578026662391,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011576377637763778,
      "loss": 3.0801,
      "step": 421240
    },
    {
      "epoch": 0.0946144235027821,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011576177617761776,
      "loss": 3.2479,
      "step": 421250
    },
    {
      "epoch": 0.09463306673894029,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011575977597759777,
      "loss": 3.2437,
      "step": 421260
    },
    {
      "epoch": 0.09465170997509846,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011575777577757775,
      "loss": 3.2367,
      "step": 421270
    },
    {
      "epoch": 0.09467035321125665,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011575577557755776,
      "loss": 3.1253,
      "step": 421280
    },
    {
      "epoch": 0.09468899644741482,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011575377537753776,
      "loss": 3.169,
      "step": 421290
    },
    {
      "epoch": 0.09470763968357301,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011575177517751777,
      "loss": 3.2718,
      "step": 421300
    },
    {
      "epoch": 0.0947262829197312,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011574977497749775,
      "loss": 2.9921,
      "step": 421310
    },
    {
      "epoch": 0.09474492615588938,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011574777477747776,
      "loss": 3.1382,
      "step": 421320
    },
    {
      "epoch": 0.09476356939204757,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011574577457745774,
      "loss": 3.124,
      "step": 421330
    },
    {
      "epoch": 0.09478221262820574,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011574377437743775,
      "loss": 3.2122,
      "step": 421340
    },
    {
      "epoch": 0.09480085586436393,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011574177417741774,
      "loss": 3.0685,
      "step": 421350
    },
    {
      "epoch": 0.09481949910052212,
      "grad_norm": 0.75,
      "learning_rate": 0.00011573977397739775,
      "loss": 3.2185,
      "step": 421360
    },
    {
      "epoch": 0.0948381423366803,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011573777377737774,
      "loss": 3.2544,
      "step": 421370
    },
    {
      "epoch": 0.09485678557283848,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011573577357735775,
      "loss": 3.1929,
      "step": 421380
    },
    {
      "epoch": 0.09487542880899667,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011573377337733773,
      "loss": 3.1159,
      "step": 421390
    },
    {
      "epoch": 0.09489407204515485,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011573177317731774,
      "loss": 3.1982,
      "step": 421400
    },
    {
      "epoch": 0.09491271528131304,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011572977297729775,
      "loss": 3.0986,
      "step": 421410
    },
    {
      "epoch": 0.09493135851747121,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011572777277727773,
      "loss": 3.1794,
      "step": 421420
    },
    {
      "epoch": 0.0949500017536294,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011572577257725774,
      "loss": 3.1522,
      "step": 421430
    },
    {
      "epoch": 0.09496864498978759,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011572377237723773,
      "loss": 3.2977,
      "step": 421440
    },
    {
      "epoch": 0.09498728822594577,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011572177217721774,
      "loss": 3.1306,
      "step": 421450
    },
    {
      "epoch": 0.09500593146210395,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011571977197719772,
      "loss": 3.0668,
      "step": 421460
    },
    {
      "epoch": 0.09502457469826213,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011571777177717773,
      "loss": 3.3402,
      "step": 421470
    },
    {
      "epoch": 0.09504321793442032,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011571577157715771,
      "loss": 3.079,
      "step": 421480
    },
    {
      "epoch": 0.09506186117057851,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011571377137713772,
      "loss": 3.1976,
      "step": 421490
    },
    {
      "epoch": 0.09508050440673668,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001157117711771177,
      "loss": 3.1594,
      "step": 421500
    },
    {
      "epoch": 0.09509914764289487,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011570977097709772,
      "loss": 3.1675,
      "step": 421510
    },
    {
      "epoch": 0.09511779087905306,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011570777077707771,
      "loss": 3.0678,
      "step": 421520
    },
    {
      "epoch": 0.09513643411521124,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011570577057705772,
      "loss": 3.0497,
      "step": 421530
    },
    {
      "epoch": 0.09515507735136942,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001157037703770377,
      "loss": 3.1267,
      "step": 421540
    },
    {
      "epoch": 0.0951737205875276,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011570177017701771,
      "loss": 3.2033,
      "step": 421550
    },
    {
      "epoch": 0.09519236382368579,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011569976997699769,
      "loss": 3.1177,
      "step": 421560
    },
    {
      "epoch": 0.09521100705984398,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001156977697769777,
      "loss": 3.181,
      "step": 421570
    },
    {
      "epoch": 0.09522965029600215,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011569576957695768,
      "loss": 3.0826,
      "step": 421580
    },
    {
      "epoch": 0.09524829353216034,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001156937693769377,
      "loss": 3.176,
      "step": 421590
    },
    {
      "epoch": 0.09526693676831853,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001156917691769177,
      "loss": 3.1113,
      "step": 421600
    },
    {
      "epoch": 0.0952855800044767,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001156897689768977,
      "loss": 3.1531,
      "step": 421610
    },
    {
      "epoch": 0.0953042232406349,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011568776877687771,
      "loss": 3.0464,
      "step": 421620
    },
    {
      "epoch": 0.09532286647679307,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011568576857685769,
      "loss": 3.1347,
      "step": 421630
    },
    {
      "epoch": 0.09534150971295126,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001156837683768377,
      "loss": 3.0835,
      "step": 421640
    },
    {
      "epoch": 0.09536015294910945,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011568176817681768,
      "loss": 3.2033,
      "step": 421650
    },
    {
      "epoch": 0.09537879618526762,
      "grad_norm": 0.75,
      "learning_rate": 0.00011567976797679769,
      "loss": 3.0812,
      "step": 421660
    },
    {
      "epoch": 0.09539743942142581,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011567776777677769,
      "loss": 3.1606,
      "step": 421670
    },
    {
      "epoch": 0.09541608265758399,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011567576757675768,
      "loss": 3.0653,
      "step": 421680
    },
    {
      "epoch": 0.09543472589374218,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011567376737673768,
      "loss": 3.0786,
      "step": 421690
    },
    {
      "epoch": 0.09545336912990036,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011567176717671769,
      "loss": 3.0959,
      "step": 421700
    },
    {
      "epoch": 0.09547201236605854,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011566976697669767,
      "loss": 3.2189,
      "step": 421710
    },
    {
      "epoch": 0.09549065560221673,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011566776677667768,
      "loss": 3.3388,
      "step": 421720
    },
    {
      "epoch": 0.09550929883837492,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011566576657665766,
      "loss": 3.2149,
      "step": 421730
    },
    {
      "epoch": 0.09552794207453309,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011566376637663767,
      "loss": 3.1594,
      "step": 421740
    },
    {
      "epoch": 0.09554658531069128,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011566176617661766,
      "loss": 3.1533,
      "step": 421750
    },
    {
      "epoch": 0.09556522854684946,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011565976597659767,
      "loss": 3.1965,
      "step": 421760
    },
    {
      "epoch": 0.09558387178300765,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011565776577657765,
      "loss": 3.04,
      "step": 421770
    },
    {
      "epoch": 0.09560251501916583,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011565576557655766,
      "loss": 3.2153,
      "step": 421780
    },
    {
      "epoch": 0.09562115825532401,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011565376537653765,
      "loss": 3.2097,
      "step": 421790
    },
    {
      "epoch": 0.0956398014914822,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011565176517651766,
      "loss": 3.1707,
      "step": 421800
    },
    {
      "epoch": 0.09565844472764037,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011564976497649766,
      "loss": 3.0826,
      "step": 421810
    },
    {
      "epoch": 0.09567708796379856,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011564776477647765,
      "loss": 3.2781,
      "step": 421820
    },
    {
      "epoch": 0.09569573119995675,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011564576457645766,
      "loss": 3.1372,
      "step": 421830
    },
    {
      "epoch": 0.09571437443611493,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011564376437643765,
      "loss": 3.0982,
      "step": 421840
    },
    {
      "epoch": 0.09573301767227312,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011564176417641766,
      "loss": 3.1673,
      "step": 421850
    },
    {
      "epoch": 0.0957516609084313,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011563976397639764,
      "loss": 3.1482,
      "step": 421860
    },
    {
      "epoch": 0.09577030414458948,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011563776377637765,
      "loss": 3.2517,
      "step": 421870
    },
    {
      "epoch": 0.09578894738074767,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011563576357635763,
      "loss": 3.1611,
      "step": 421880
    },
    {
      "epoch": 0.09580759061690584,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011563376337633764,
      "loss": 3.0897,
      "step": 421890
    },
    {
      "epoch": 0.09582623385306403,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011563176317631762,
      "loss": 3.2164,
      "step": 421900
    },
    {
      "epoch": 0.09584487708922222,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011562976297629763,
      "loss": 3.1706,
      "step": 421910
    },
    {
      "epoch": 0.0958635203253804,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011562776277627763,
      "loss": 3.1975,
      "step": 421920
    },
    {
      "epoch": 0.09588216356153859,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011562576257625764,
      "loss": 3.204,
      "step": 421930
    },
    {
      "epoch": 0.09590080679769677,
      "grad_norm": 0.875,
      "learning_rate": 0.00011562376237623762,
      "loss": 3.2591,
      "step": 421940
    },
    {
      "epoch": 0.09591945003385495,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011562176217621763,
      "loss": 3.0761,
      "step": 421950
    },
    {
      "epoch": 0.09593809327001314,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011561976197619761,
      "loss": 3.2376,
      "step": 421960
    },
    {
      "epoch": 0.09595673650617131,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011561776177617762,
      "loss": 3.1451,
      "step": 421970
    },
    {
      "epoch": 0.0959753797423295,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011561576157615762,
      "loss": 3.051,
      "step": 421980
    },
    {
      "epoch": 0.09599402297848769,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011561376137613763,
      "loss": 3.1477,
      "step": 421990
    },
    {
      "epoch": 0.09601266621464587,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011561176117611762,
      "loss": 3.0784,
      "step": 422000
    },
    {
      "epoch": 0.09603130945080406,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011560976097609762,
      "loss": 3.2711,
      "step": 422010
    },
    {
      "epoch": 0.09604995268696223,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011560776077607763,
      "loss": 3.0951,
      "step": 422020
    },
    {
      "epoch": 0.09606859592312042,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011560576057605761,
      "loss": 3.1364,
      "step": 422030
    },
    {
      "epoch": 0.09608723915927861,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011560376037603762,
      "loss": 3.2278,
      "step": 422040
    },
    {
      "epoch": 0.09610588239543678,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001156017601760176,
      "loss": 3.1086,
      "step": 422050
    },
    {
      "epoch": 0.09612452563159497,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011559975997599761,
      "loss": 3.2275,
      "step": 422060
    },
    {
      "epoch": 0.09614316886775316,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001155977597759776,
      "loss": 3.2399,
      "step": 422070
    },
    {
      "epoch": 0.09616181210391134,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011559575957595761,
      "loss": 3.2124,
      "step": 422080
    },
    {
      "epoch": 0.09618045534006953,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001155937593759376,
      "loss": 3.1471,
      "step": 422090
    },
    {
      "epoch": 0.0961990985762277,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001155917591759176,
      "loss": 3.2852,
      "step": 422100
    },
    {
      "epoch": 0.09621774181238589,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011558975897589759,
      "loss": 3.2248,
      "step": 422110
    },
    {
      "epoch": 0.09623638504854408,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001155877587758776,
      "loss": 3.1683,
      "step": 422120
    },
    {
      "epoch": 0.09625502828470225,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011558575857585758,
      "loss": 3.0685,
      "step": 422130
    },
    {
      "epoch": 0.09627367152086044,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011558375837583759,
      "loss": 3.1399,
      "step": 422140
    },
    {
      "epoch": 0.09629231475701862,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011558175817581758,
      "loss": 3.2031,
      "step": 422150
    },
    {
      "epoch": 0.09631095799317681,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011557975797579759,
      "loss": 3.1722,
      "step": 422160
    },
    {
      "epoch": 0.096329601229335,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011557775777577757,
      "loss": 3.0824,
      "step": 422170
    },
    {
      "epoch": 0.09634824446549317,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011557575757575758,
      "loss": 3.2001,
      "step": 422180
    },
    {
      "epoch": 0.09636688770165136,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011557375737573759,
      "loss": 3.2639,
      "step": 422190
    },
    {
      "epoch": 0.09638553093780955,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011557175717571757,
      "loss": 3.211,
      "step": 422200
    },
    {
      "epoch": 0.09640417417396772,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011556975697569758,
      "loss": 3.1071,
      "step": 422210
    },
    {
      "epoch": 0.09642281741012591,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011556775677567756,
      "loss": 3.1329,
      "step": 422220
    },
    {
      "epoch": 0.09644146064628409,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011556575657565757,
      "loss": 3.1638,
      "step": 422230
    },
    {
      "epoch": 0.09646010388244228,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011556375637563757,
      "loss": 3.202,
      "step": 422240
    },
    {
      "epoch": 0.09647874711860047,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011556175617561758,
      "loss": 3.0718,
      "step": 422250
    },
    {
      "epoch": 0.09649739035475864,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011555975597559756,
      "loss": 3.0859,
      "step": 422260
    },
    {
      "epoch": 0.09651603359091683,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011555775577557757,
      "loss": 3.1487,
      "step": 422270
    },
    {
      "epoch": 0.09653467682707502,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011555575557555755,
      "loss": 3.0631,
      "step": 422280
    },
    {
      "epoch": 0.0965533200632332,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011555375537553756,
      "loss": 3.2108,
      "step": 422290
    },
    {
      "epoch": 0.09657196329939138,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011555175517551756,
      "loss": 3.1304,
      "step": 422300
    },
    {
      "epoch": 0.09659060653554956,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011554975497549757,
      "loss": 3.1751,
      "step": 422310
    },
    {
      "epoch": 0.09660924977170775,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011554775477547755,
      "loss": 3.2071,
      "step": 422320
    },
    {
      "epoch": 0.09662789300786594,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011554575457545756,
      "loss": 3.0561,
      "step": 422330
    },
    {
      "epoch": 0.09664653624402411,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011554375437543754,
      "loss": 3.1851,
      "step": 422340
    },
    {
      "epoch": 0.0966651794801823,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011554175417541755,
      "loss": 3.2552,
      "step": 422350
    },
    {
      "epoch": 0.09668382271634048,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011553975397539753,
      "loss": 3.2083,
      "step": 422360
    },
    {
      "epoch": 0.09670246595249866,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011553775377537754,
      "loss": 3.2006,
      "step": 422370
    },
    {
      "epoch": 0.09672110918865685,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011553575357535754,
      "loss": 3.1895,
      "step": 422380
    },
    {
      "epoch": 0.09673975242481503,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011553375337533754,
      "loss": 3.1464,
      "step": 422390
    },
    {
      "epoch": 0.09675839566097322,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011553175317531755,
      "loss": 3.0573,
      "step": 422400
    },
    {
      "epoch": 0.0967770388971314,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011552975297529754,
      "loss": 3.2187,
      "step": 422410
    },
    {
      "epoch": 0.09679568213328958,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011552775277527754,
      "loss": 3.1028,
      "step": 422420
    },
    {
      "epoch": 0.09681432536944777,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011552575257525753,
      "loss": 3.2204,
      "step": 422430
    },
    {
      "epoch": 0.09683296860560595,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011552375237523754,
      "loss": 3.0697,
      "step": 422440
    },
    {
      "epoch": 0.09685161184176413,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011552175217521752,
      "loss": 3.2874,
      "step": 422450
    },
    {
      "epoch": 0.09687025507792232,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011551975197519753,
      "loss": 3.2682,
      "step": 422460
    },
    {
      "epoch": 0.0968888983140805,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011551775177517752,
      "loss": 3.1376,
      "step": 422470
    },
    {
      "epoch": 0.09690754155023869,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011551575157515753,
      "loss": 3.191,
      "step": 422480
    },
    {
      "epoch": 0.09692618478639686,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011551375137513751,
      "loss": 3.1302,
      "step": 422490
    },
    {
      "epoch": 0.09694482802255505,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011551175117511752,
      "loss": 3.2964,
      "step": 422500
    },
    {
      "epoch": 0.09696347125871324,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001155097509750975,
      "loss": 3.1403,
      "step": 422510
    },
    {
      "epoch": 0.09698211449487142,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011550775077507751,
      "loss": 3.2273,
      "step": 422520
    },
    {
      "epoch": 0.0970007577310296,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011550575057505751,
      "loss": 3.3638,
      "step": 422530
    },
    {
      "epoch": 0.0970194009671878,
      "grad_norm": 0.75,
      "learning_rate": 0.0001155037503750375,
      "loss": 3.2507,
      "step": 422540
    },
    {
      "epoch": 0.09703804420334597,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001155017501750175,
      "loss": 3.1397,
      "step": 422550
    },
    {
      "epoch": 0.09705668743950416,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011549974997499751,
      "loss": 3.3803,
      "step": 422560
    },
    {
      "epoch": 0.09707533067566233,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011549774977497749,
      "loss": 3.0866,
      "step": 422570
    },
    {
      "epoch": 0.09709397391182052,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001154957495749575,
      "loss": 3.068,
      "step": 422580
    },
    {
      "epoch": 0.09711261714797871,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011549374937493751,
      "loss": 3.2186,
      "step": 422590
    },
    {
      "epoch": 0.09713126038413689,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011549174917491749,
      "loss": 3.2068,
      "step": 422600
    },
    {
      "epoch": 0.09714990362029507,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001154897489748975,
      "loss": 3.2636,
      "step": 422610
    },
    {
      "epoch": 0.09716854685645326,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001154877487748775,
      "loss": 3.1056,
      "step": 422620
    },
    {
      "epoch": 0.09718719009261144,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001154857485748575,
      "loss": 3.197,
      "step": 422630
    },
    {
      "epoch": 0.09720583332876963,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011548374837483749,
      "loss": 3.1782,
      "step": 422640
    },
    {
      "epoch": 0.0972244765649278,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001154817481748175,
      "loss": 3.0751,
      "step": 422650
    },
    {
      "epoch": 0.09724311980108599,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011547974797479748,
      "loss": 3.1767,
      "step": 422660
    },
    {
      "epoch": 0.09726176303724418,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011547774777477749,
      "loss": 3.1685,
      "step": 422670
    },
    {
      "epoch": 0.09728040627340236,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011547574757475747,
      "loss": 3.2204,
      "step": 422680
    },
    {
      "epoch": 0.09729904950956055,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011547374737473748,
      "loss": 3.0783,
      "step": 422690
    },
    {
      "epoch": 0.09731769274571872,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011547174717471748,
      "loss": 3.3091,
      "step": 422700
    },
    {
      "epoch": 0.09733633598187691,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011546974697469748,
      "loss": 3.2386,
      "step": 422710
    },
    {
      "epoch": 0.0973549792180351,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011546774677467747,
      "loss": 3.2928,
      "step": 422720
    },
    {
      "epoch": 0.09737362245419327,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011546574657465748,
      "loss": 3.2607,
      "step": 422730
    },
    {
      "epoch": 0.09739226569035146,
      "grad_norm": 0.75,
      "learning_rate": 0.00011546374637463746,
      "loss": 3.0809,
      "step": 422740
    },
    {
      "epoch": 0.09741090892650965,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011546174617461747,
      "loss": 3.0905,
      "step": 422750
    },
    {
      "epoch": 0.09742955216266783,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011545974597459745,
      "loss": 3.1915,
      "step": 422760
    },
    {
      "epoch": 0.09744819539882602,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011545774577457746,
      "loss": 3.2302,
      "step": 422770
    },
    {
      "epoch": 0.09746683863498419,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011545574557455747,
      "loss": 3.2492,
      "step": 422780
    },
    {
      "epoch": 0.09748548187114238,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011545374537453746,
      "loss": 3.1874,
      "step": 422790
    },
    {
      "epoch": 0.09750412510730057,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011545174517451747,
      "loss": 2.9735,
      "step": 422800
    },
    {
      "epoch": 0.09752276834345874,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011544974497449745,
      "loss": 3.0874,
      "step": 422810
    },
    {
      "epoch": 0.09754141157961693,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011544774477447746,
      "loss": 3.0711,
      "step": 422820
    },
    {
      "epoch": 0.09756005481577511,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011544574457445745,
      "loss": 3.0844,
      "step": 422830
    },
    {
      "epoch": 0.0975786980519333,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011544374437443745,
      "loss": 3.2861,
      "step": 422840
    },
    {
      "epoch": 0.09759734128809149,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011544174417441745,
      "loss": 3.0973,
      "step": 422850
    },
    {
      "epoch": 0.09761598452424966,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011543974397439745,
      "loss": 3.0975,
      "step": 422860
    },
    {
      "epoch": 0.09763462776040785,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011543774377437744,
      "loss": 3.0428,
      "step": 422870
    },
    {
      "epoch": 0.09765327099656604,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011543574357435745,
      "loss": 3.1361,
      "step": 422880
    },
    {
      "epoch": 0.09767191423272421,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011543374337433743,
      "loss": 3.1659,
      "step": 422890
    },
    {
      "epoch": 0.0976905574688824,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011543174317431744,
      "loss": 3.2447,
      "step": 422900
    },
    {
      "epoch": 0.09770920070504058,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011542974297429742,
      "loss": 3.1882,
      "step": 422910
    },
    {
      "epoch": 0.09772784394119877,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011542774277427743,
      "loss": 3.1976,
      "step": 422920
    },
    {
      "epoch": 0.09774648717735696,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011542574257425743,
      "loss": 3.2846,
      "step": 422930
    },
    {
      "epoch": 0.09776513041351513,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011542374237423744,
      "loss": 2.9839,
      "step": 422940
    },
    {
      "epoch": 0.09778377364967332,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011542174217421742,
      "loss": 3.0854,
      "step": 422950
    },
    {
      "epoch": 0.09780241688583151,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011541974197419743,
      "loss": 3.2053,
      "step": 422960
    },
    {
      "epoch": 0.09782106012198968,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011541774177417741,
      "loss": 3.2051,
      "step": 422970
    },
    {
      "epoch": 0.09783970335814787,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011541574157415742,
      "loss": 3.1979,
      "step": 422980
    },
    {
      "epoch": 0.09785834659430605,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011541374137413743,
      "loss": 3.1305,
      "step": 422990
    },
    {
      "epoch": 0.09787698983046424,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011541174117411741,
      "loss": 3.238,
      "step": 423000
    },
    {
      "epoch": 0.09789563306662243,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011540974097409742,
      "loss": 3.2024,
      "step": 423010
    },
    {
      "epoch": 0.0979142763027806,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011540774077407742,
      "loss": 3.2091,
      "step": 423020
    },
    {
      "epoch": 0.09793291953893879,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011540574057405742,
      "loss": 3.1432,
      "step": 423030
    },
    {
      "epoch": 0.09795156277509696,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011540374037403741,
      "loss": 3.1444,
      "step": 423040
    },
    {
      "epoch": 0.09797020601125515,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011540174017401742,
      "loss": 3.2647,
      "step": 423050
    },
    {
      "epoch": 0.09798884924741334,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001153997399739974,
      "loss": 3.2425,
      "step": 423060
    },
    {
      "epoch": 0.09800749248357152,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011539773977397741,
      "loss": 3.0653,
      "step": 423070
    },
    {
      "epoch": 0.0980261357197297,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011539573957395739,
      "loss": 3.3216,
      "step": 423080
    },
    {
      "epoch": 0.0980447789558879,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001153937393739374,
      "loss": 3.1673,
      "step": 423090
    },
    {
      "epoch": 0.09806342219204607,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001153917391739174,
      "loss": 3.1718,
      "step": 423100
    },
    {
      "epoch": 0.09808206542820426,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001153897389738974,
      "loss": 3.169,
      "step": 423110
    },
    {
      "epoch": 0.09810070866436243,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011538773877387739,
      "loss": 3.1835,
      "step": 423120
    },
    {
      "epoch": 0.09811935190052062,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001153857385738574,
      "loss": 3.3803,
      "step": 423130
    },
    {
      "epoch": 0.09813799513667881,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011538373837383738,
      "loss": 2.9769,
      "step": 423140
    },
    {
      "epoch": 0.09815663837283699,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011538173817381739,
      "loss": 3.1285,
      "step": 423150
    },
    {
      "epoch": 0.09817528160899518,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011537973797379738,
      "loss": 3.1106,
      "step": 423160
    },
    {
      "epoch": 0.09819392484515335,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011537773777377738,
      "loss": 3.1924,
      "step": 423170
    },
    {
      "epoch": 0.09821256808131154,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011537573757375739,
      "loss": 3.1327,
      "step": 423180
    },
    {
      "epoch": 0.09823121131746973,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011537373737373738,
      "loss": 3.2793,
      "step": 423190
    },
    {
      "epoch": 0.0982498545536279,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011537173717371739,
      "loss": 3.0675,
      "step": 423200
    },
    {
      "epoch": 0.0982684977897861,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011536973697369737,
      "loss": 3.0342,
      "step": 423210
    },
    {
      "epoch": 0.09828714102594428,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011536773677367738,
      "loss": 3.1396,
      "step": 423220
    },
    {
      "epoch": 0.09830578426210246,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011536573657365736,
      "loss": 3.2715,
      "step": 423230
    },
    {
      "epoch": 0.09832442749826065,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011536373637363737,
      "loss": 3.0977,
      "step": 423240
    },
    {
      "epoch": 0.09834307073441882,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011536173617361737,
      "loss": 3.1064,
      "step": 423250
    },
    {
      "epoch": 0.09836171397057701,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011535973597359738,
      "loss": 3.0829,
      "step": 423260
    },
    {
      "epoch": 0.0983803572067352,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011535773577357736,
      "loss": 3.1124,
      "step": 423270
    },
    {
      "epoch": 0.09839900044289338,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011535573557355737,
      "loss": 3.2349,
      "step": 423280
    },
    {
      "epoch": 0.09841764367905156,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011535373537353735,
      "loss": 3.2929,
      "step": 423290
    },
    {
      "epoch": 0.09843628691520975,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011535173517351736,
      "loss": 3.2484,
      "step": 423300
    },
    {
      "epoch": 0.09845493015136793,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011534973497349734,
      "loss": 3.1771,
      "step": 423310
    },
    {
      "epoch": 0.09847357338752612,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011534773477347735,
      "loss": 3.1447,
      "step": 423320
    },
    {
      "epoch": 0.09849221662368429,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011534573457345735,
      "loss": 2.9963,
      "step": 423330
    },
    {
      "epoch": 0.09851085985984248,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011534373437343736,
      "loss": 3.0688,
      "step": 423340
    },
    {
      "epoch": 0.09852950309600067,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011534173417341734,
      "loss": 3.0717,
      "step": 423350
    },
    {
      "epoch": 0.09854814633215885,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011533973397339735,
      "loss": 3.2883,
      "step": 423360
    },
    {
      "epoch": 0.09856678956831703,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011533773377337733,
      "loss": 3.1715,
      "step": 423370
    },
    {
      "epoch": 0.09858543280447521,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011533573357335734,
      "loss": 3.1619,
      "step": 423380
    },
    {
      "epoch": 0.0986040760406334,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011533373337333735,
      "loss": 3.0222,
      "step": 423390
    },
    {
      "epoch": 0.09862271927679159,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011533173317331733,
      "loss": 3.0202,
      "step": 423400
    },
    {
      "epoch": 0.09864136251294976,
      "grad_norm": 0.75,
      "learning_rate": 0.00011532973297329734,
      "loss": 3.1839,
      "step": 423410
    },
    {
      "epoch": 0.09866000574910795,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011532773277327733,
      "loss": 3.2062,
      "step": 423420
    },
    {
      "epoch": 0.09867864898526614,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011532573257325734,
      "loss": 3.1561,
      "step": 423430
    },
    {
      "epoch": 0.09869729222142432,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011532373237323733,
      "loss": 3.0922,
      "step": 423440
    },
    {
      "epoch": 0.0987159354575825,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011532173217321733,
      "loss": 3.1763,
      "step": 423450
    },
    {
      "epoch": 0.09873457869374068,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011531973197319732,
      "loss": 3.1894,
      "step": 423460
    },
    {
      "epoch": 0.09875322192989887,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011531773177317733,
      "loss": 3.2389,
      "step": 423470
    },
    {
      "epoch": 0.09877186516605706,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011531573157315732,
      "loss": 3.1892,
      "step": 423480
    },
    {
      "epoch": 0.09879050840221523,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011531373137313732,
      "loss": 3.2749,
      "step": 423490
    },
    {
      "epoch": 0.09880915163837342,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011531173117311731,
      "loss": 3.3019,
      "step": 423500
    },
    {
      "epoch": 0.0988277948745316,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011530973097309732,
      "loss": 3.0055,
      "step": 423510
    },
    {
      "epoch": 0.09884643811068979,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001153077307730773,
      "loss": 3.0911,
      "step": 423520
    },
    {
      "epoch": 0.09886508134684797,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011530573057305731,
      "loss": 3.1746,
      "step": 423530
    },
    {
      "epoch": 0.09888372458300615,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001153037303730373,
      "loss": 3.2402,
      "step": 423540
    },
    {
      "epoch": 0.09890236781916434,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001153017301730173,
      "loss": 3.1023,
      "step": 423550
    },
    {
      "epoch": 0.09892101105532253,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001152997299729973,
      "loss": 3.2117,
      "step": 423560
    },
    {
      "epoch": 0.0989396542914807,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011529772977297731,
      "loss": 3.1584,
      "step": 423570
    },
    {
      "epoch": 0.09895829752763889,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011529572957295732,
      "loss": 3.1689,
      "step": 423580
    },
    {
      "epoch": 0.09897694076379707,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001152937293729373,
      "loss": 3.2148,
      "step": 423590
    },
    {
      "epoch": 0.09899558399995526,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011529172917291731,
      "loss": 3.1526,
      "step": 423600
    },
    {
      "epoch": 0.09901422723611344,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011528972897289729,
      "loss": 3.1559,
      "step": 423610
    },
    {
      "epoch": 0.09903287047227162,
      "grad_norm": 0.625,
      "learning_rate": 0.0001152877287728773,
      "loss": 3.0999,
      "step": 423620
    },
    {
      "epoch": 0.09905151370842981,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011528572857285728,
      "loss": 3.1059,
      "step": 423630
    },
    {
      "epoch": 0.099070156944588,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011528372837283729,
      "loss": 3.1705,
      "step": 423640
    },
    {
      "epoch": 0.09908880018074617,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011528172817281729,
      "loss": 3.1253,
      "step": 423650
    },
    {
      "epoch": 0.09910744341690436,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001152797279727973,
      "loss": 3.1365,
      "step": 423660
    },
    {
      "epoch": 0.09912608665306254,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011527772777277728,
      "loss": 3.0802,
      "step": 423670
    },
    {
      "epoch": 0.09914472988922073,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011527572757275729,
      "loss": 3.2114,
      "step": 423680
    },
    {
      "epoch": 0.09916337312537891,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011527372737273727,
      "loss": 3.1465,
      "step": 423690
    },
    {
      "epoch": 0.09918201636153709,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011527172717271728,
      "loss": 3.1691,
      "step": 423700
    },
    {
      "epoch": 0.09920065959769528,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011526972697269726,
      "loss": 3.1413,
      "step": 423710
    },
    {
      "epoch": 0.09921930283385345,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011526772677267727,
      "loss": 3.119,
      "step": 423720
    },
    {
      "epoch": 0.09923794607001164,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011526572657265727,
      "loss": 3.1007,
      "step": 423730
    },
    {
      "epoch": 0.09925658930616983,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011526372637263727,
      "loss": 3.1604,
      "step": 423740
    },
    {
      "epoch": 0.099275232542328,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011526172617261726,
      "loss": 3.2211,
      "step": 423750
    },
    {
      "epoch": 0.0992938757784862,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011525972597259727,
      "loss": 3.0203,
      "step": 423760
    },
    {
      "epoch": 0.09931251901464438,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011525772577257728,
      "loss": 3.2035,
      "step": 423770
    },
    {
      "epoch": 0.09933116225080256,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011525572557255726,
      "loss": 3.0047,
      "step": 423780
    },
    {
      "epoch": 0.09934980548696075,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011525372537253727,
      "loss": 3.1843,
      "step": 423790
    },
    {
      "epoch": 0.09936844872311892,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011525172517251726,
      "loss": 3.1472,
      "step": 423800
    },
    {
      "epoch": 0.09938709195927711,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011524972497249726,
      "loss": 3.1445,
      "step": 423810
    },
    {
      "epoch": 0.0994057351954353,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011524772477247725,
      "loss": 3.1514,
      "step": 423820
    },
    {
      "epoch": 0.09942437843159348,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011524572457245726,
      "loss": 3.0891,
      "step": 423830
    },
    {
      "epoch": 0.09944302166775167,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011524372437243724,
      "loss": 3.185,
      "step": 423840
    },
    {
      "epoch": 0.09946166490390984,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011524172417241725,
      "loss": 3.2694,
      "step": 423850
    },
    {
      "epoch": 0.09948030814006803,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011523972397239724,
      "loss": 3.1879,
      "step": 423860
    },
    {
      "epoch": 0.09949895137622622,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011523772377237724,
      "loss": 3.2765,
      "step": 423870
    },
    {
      "epoch": 0.0995175946123844,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011523572357235724,
      "loss": 3.1972,
      "step": 423880
    },
    {
      "epoch": 0.09953623784854258,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011523372337233725,
      "loss": 3.1237,
      "step": 423890
    },
    {
      "epoch": 0.09955488108470077,
      "grad_norm": 0.75,
      "learning_rate": 0.00011523172317231723,
      "loss": 3.229,
      "step": 423900
    },
    {
      "epoch": 0.09957352432085895,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011522972297229724,
      "loss": 3.0973,
      "step": 423910
    },
    {
      "epoch": 0.09959216755701714,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011522772277227722,
      "loss": 3.1428,
      "step": 423920
    },
    {
      "epoch": 0.09961081079317531,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011522572257225723,
      "loss": 3.0994,
      "step": 423930
    },
    {
      "epoch": 0.0996294540293335,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011522372237223721,
      "loss": 3.1317,
      "step": 423940
    },
    {
      "epoch": 0.09964809726549169,
      "grad_norm": 0.75,
      "learning_rate": 0.00011522172217221722,
      "loss": 3.2192,
      "step": 423950
    },
    {
      "epoch": 0.09966674050164986,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011521972197219722,
      "loss": 3.0848,
      "step": 423960
    },
    {
      "epoch": 0.09968538373780805,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011521772177217723,
      "loss": 2.9606,
      "step": 423970
    },
    {
      "epoch": 0.09970402697396624,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011521572157215724,
      "loss": 3.0485,
      "step": 423980
    },
    {
      "epoch": 0.09972267021012442,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011521372137213722,
      "loss": 3.0392,
      "step": 423990
    },
    {
      "epoch": 0.0997413134462826,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011521172117211723,
      "loss": 3.167,
      "step": 424000
    },
    {
      "epoch": 0.09975995668244078,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011520972097209721,
      "loss": 3.1579,
      "step": 424010
    },
    {
      "epoch": 0.09977859991859897,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011520772077207722,
      "loss": 3.1705,
      "step": 424020
    },
    {
      "epoch": 0.09979724315475716,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001152057205720572,
      "loss": 3.2656,
      "step": 424030
    },
    {
      "epoch": 0.09981588639091533,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011520372037203721,
      "loss": 3.2116,
      "step": 424040
    },
    {
      "epoch": 0.09983452962707352,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001152017201720172,
      "loss": 3.1775,
      "step": 424050
    },
    {
      "epoch": 0.0998531728632317,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011519971997199722,
      "loss": 3.1664,
      "step": 424060
    },
    {
      "epoch": 0.09987181609938989,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001151977197719772,
      "loss": 3.1576,
      "step": 424070
    },
    {
      "epoch": 0.09989045933554808,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001151957195719572,
      "loss": 3.178,
      "step": 424080
    },
    {
      "epoch": 0.09990910257170625,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011519371937193719,
      "loss": 3.1566,
      "step": 424090
    },
    {
      "epoch": 0.09992774580786444,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001151917191719172,
      "loss": 3.011,
      "step": 424100
    },
    {
      "epoch": 0.09994638904402263,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011518971897189719,
      "loss": 3.2079,
      "step": 424110
    },
    {
      "epoch": 0.0999650322801808,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001151877187718772,
      "loss": 3.2002,
      "step": 424120
    },
    {
      "epoch": 0.099983675516339,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011518571857185718,
      "loss": 3.2843,
      "step": 424130
    },
    {
      "epoch": 0.10000231875249717,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001151837183718372,
      "loss": 3.0968,
      "step": 424140
    },
    {
      "epoch": 0.10002096198865536,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011518171817181718,
      "loss": 3.2549,
      "step": 424150
    },
    {
      "epoch": 0.10003960522481355,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011517971797179718,
      "loss": 3.1755,
      "step": 424160
    },
    {
      "epoch": 0.10005824846097172,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001151777177717772,
      "loss": 3.2036,
      "step": 424170
    },
    {
      "epoch": 0.10007689169712991,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011517571757175718,
      "loss": 3.2215,
      "step": 424180
    },
    {
      "epoch": 0.10009553493328809,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011517371737173718,
      "loss": 3.107,
      "step": 424190
    },
    {
      "epoch": 0.10011417816944627,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011517171717171718,
      "loss": 3.1705,
      "step": 424200
    },
    {
      "epoch": 0.10013282140560446,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011516971697169719,
      "loss": 3.1057,
      "step": 424210
    },
    {
      "epoch": 0.10015146464176264,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011516771677167717,
      "loss": 3.1655,
      "step": 424220
    },
    {
      "epoch": 0.10017010787792083,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011516571657165718,
      "loss": 3.0351,
      "step": 424230
    },
    {
      "epoch": 0.10018875111407902,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011516371637163716,
      "loss": 3.074,
      "step": 424240
    },
    {
      "epoch": 0.10020739435023719,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011516171617161717,
      "loss": 3.0588,
      "step": 424250
    },
    {
      "epoch": 0.10022603758639538,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011515971597159715,
      "loss": 3.1297,
      "step": 424260
    },
    {
      "epoch": 0.10024468082255356,
      "grad_norm": 0.75,
      "learning_rate": 0.00011515771577157716,
      "loss": 3.1568,
      "step": 424270
    },
    {
      "epoch": 0.10026332405871174,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011515571557155716,
      "loss": 3.1265,
      "step": 424280
    },
    {
      "epoch": 0.10028196729486993,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011515371537153717,
      "loss": 3.0689,
      "step": 424290
    },
    {
      "epoch": 0.10030061053102811,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011515171517151715,
      "loss": 3.0765,
      "step": 424300
    },
    {
      "epoch": 0.1003192537671863,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011514971497149716,
      "loss": 3.328,
      "step": 424310
    },
    {
      "epoch": 0.10033789700334449,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011514771477147714,
      "loss": 3.2349,
      "step": 424320
    },
    {
      "epoch": 0.10035654023950266,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011514571457145715,
      "loss": 3.2843,
      "step": 424330
    },
    {
      "epoch": 0.10037518347566085,
      "grad_norm": 0.75,
      "learning_rate": 0.00011514371437143715,
      "loss": 2.9374,
      "step": 424340
    },
    {
      "epoch": 0.10039382671181903,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011514171417141714,
      "loss": 3.1664,
      "step": 424350
    },
    {
      "epoch": 0.10041246994797721,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011513971397139715,
      "loss": 3.1158,
      "step": 424360
    },
    {
      "epoch": 0.1004311131841354,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011513771377137715,
      "loss": 3.1448,
      "step": 424370
    },
    {
      "epoch": 0.10044975642029358,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011513571357135716,
      "loss": 3.1836,
      "step": 424380
    },
    {
      "epoch": 0.10046839965645177,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011513371337133714,
      "loss": 3.2732,
      "step": 424390
    },
    {
      "epoch": 0.10048704289260994,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011513171317131715,
      "loss": 3.1619,
      "step": 424400
    },
    {
      "epoch": 0.10050568612876813,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011512971297129713,
      "loss": 3.0488,
      "step": 424410
    },
    {
      "epoch": 0.10052432936492632,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011512771277127714,
      "loss": 3.2428,
      "step": 424420
    },
    {
      "epoch": 0.1005429726010845,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011512571257125713,
      "loss": 3.1501,
      "step": 424430
    },
    {
      "epoch": 0.10056161583724268,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011512371237123713,
      "loss": 3.2961,
      "step": 424440
    },
    {
      "epoch": 0.10058025907340087,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011512171217121712,
      "loss": 3.2449,
      "step": 424450
    },
    {
      "epoch": 0.10059890230955905,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011511971197119713,
      "loss": 3.0881,
      "step": 424460
    },
    {
      "epoch": 0.10061754554571724,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011511771177117712,
      "loss": 3.1514,
      "step": 424470
    },
    {
      "epoch": 0.10063618878187541,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011511571157115712,
      "loss": 3.2892,
      "step": 424480
    },
    {
      "epoch": 0.1006548320180336,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011511371137113711,
      "loss": 3.2048,
      "step": 424490
    },
    {
      "epoch": 0.10067347525419179,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011511171117111712,
      "loss": 3.1581,
      "step": 424500
    },
    {
      "epoch": 0.10069211849034997,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011510971097109711,
      "loss": 3.2827,
      "step": 424510
    },
    {
      "epoch": 0.10071076172650816,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011510771077107712,
      "loss": 3.0895,
      "step": 424520
    },
    {
      "epoch": 0.10072940496266633,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001151057105710571,
      "loss": 3.1549,
      "step": 424530
    },
    {
      "epoch": 0.10074804819882452,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011510371037103711,
      "loss": 3.0677,
      "step": 424540
    },
    {
      "epoch": 0.10076669143498271,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001151017101710171,
      "loss": 3.1681,
      "step": 424550
    },
    {
      "epoch": 0.10078533467114088,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001150997099709971,
      "loss": 3.1539,
      "step": 424560
    },
    {
      "epoch": 0.10080397790729907,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011509770977097711,
      "loss": 3.0204,
      "step": 424570
    },
    {
      "epoch": 0.10082262114345726,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001150957095709571,
      "loss": 3.0951,
      "step": 424580
    },
    {
      "epoch": 0.10084126437961544,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001150937093709371,
      "loss": 3.0868,
      "step": 424590
    },
    {
      "epoch": 0.10085990761577363,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001150917091709171,
      "loss": 3.0696,
      "step": 424600
    },
    {
      "epoch": 0.1008785508519318,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011508970897089711,
      "loss": 3.2021,
      "step": 424610
    },
    {
      "epoch": 0.10089719408808999,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011508770877087709,
      "loss": 3.0978,
      "step": 424620
    },
    {
      "epoch": 0.10091583732424818,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001150857085708571,
      "loss": 3.2479,
      "step": 424630
    },
    {
      "epoch": 0.10093448056040635,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011508370837083708,
      "loss": 3.1936,
      "step": 424640
    },
    {
      "epoch": 0.10095312379656454,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011508170817081709,
      "loss": 3.2656,
      "step": 424650
    },
    {
      "epoch": 0.10097176703272273,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011507970797079707,
      "loss": 3.145,
      "step": 424660
    },
    {
      "epoch": 0.1009904102688809,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011507770777077708,
      "loss": 3.1378,
      "step": 424670
    },
    {
      "epoch": 0.1010090535050391,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011507570757075708,
      "loss": 3.1467,
      "step": 424680
    },
    {
      "epoch": 0.10102769674119727,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011507370737073709,
      "loss": 2.9666,
      "step": 424690
    },
    {
      "epoch": 0.10104633997735546,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011507170717071707,
      "loss": 3.1723,
      "step": 424700
    },
    {
      "epoch": 0.10106498321351365,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011506970697069708,
      "loss": 3.1954,
      "step": 424710
    },
    {
      "epoch": 0.10108362644967182,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011506770677067706,
      "loss": 3.0335,
      "step": 424720
    },
    {
      "epoch": 0.10110226968583001,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011506570657065707,
      "loss": 3.1905,
      "step": 424730
    },
    {
      "epoch": 0.10112091292198819,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011506370637063706,
      "loss": 3.1835,
      "step": 424740
    },
    {
      "epoch": 0.10113955615814638,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011506170617061707,
      "loss": 3.0323,
      "step": 424750
    },
    {
      "epoch": 0.10115819939430457,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011505970597059707,
      "loss": 3.2104,
      "step": 424760
    },
    {
      "epoch": 0.10117684263046274,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011505770577057706,
      "loss": 3.1607,
      "step": 424770
    },
    {
      "epoch": 0.10119548586662093,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011505570557055707,
      "loss": 3.1794,
      "step": 424780
    },
    {
      "epoch": 0.10121412910277912,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011505370537053706,
      "loss": 3.1403,
      "step": 424790
    },
    {
      "epoch": 0.1012327723389373,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011505170517051707,
      "loss": 3.1921,
      "step": 424800
    },
    {
      "epoch": 0.10125141557509548,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011504970497049705,
      "loss": 3.1849,
      "step": 424810
    },
    {
      "epoch": 0.10127005881125366,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011504770477047706,
      "loss": 3.1571,
      "step": 424820
    },
    {
      "epoch": 0.10128870204741185,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011504570457045705,
      "loss": 3.1084,
      "step": 424830
    },
    {
      "epoch": 0.10130734528357004,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011504370437043706,
      "loss": 3.2067,
      "step": 424840
    },
    {
      "epoch": 0.10132598851972821,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011504170417041704,
      "loss": 3.1686,
      "step": 424850
    },
    {
      "epoch": 0.1013446317558864,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011503970397039705,
      "loss": 3.0847,
      "step": 424860
    },
    {
      "epoch": 0.10136327499204457,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011503770377037703,
      "loss": 3.2477,
      "step": 424870
    },
    {
      "epoch": 0.10138191822820276,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011503570357035704,
      "loss": 3.1163,
      "step": 424880
    },
    {
      "epoch": 0.10140056146436095,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011503370337033703,
      "loss": 3.1659,
      "step": 424890
    },
    {
      "epoch": 0.10141920470051913,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011503170317031703,
      "loss": 3.1941,
      "step": 424900
    },
    {
      "epoch": 0.10143784793667732,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011502970297029703,
      "loss": 3.1136,
      "step": 424910
    },
    {
      "epoch": 0.1014564911728355,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011502770277027704,
      "loss": 3.1104,
      "step": 424920
    },
    {
      "epoch": 0.10147513440899368,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011502570257025702,
      "loss": 3.0891,
      "step": 424930
    },
    {
      "epoch": 0.10149377764515187,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011502370237023703,
      "loss": 3.2011,
      "step": 424940
    },
    {
      "epoch": 0.10151242088131004,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011502170217021704,
      "loss": 3.1767,
      "step": 424950
    },
    {
      "epoch": 0.10153106411746823,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011501970197019702,
      "loss": 3.2921,
      "step": 424960
    },
    {
      "epoch": 0.10154970735362642,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011501770177017703,
      "loss": 3.2172,
      "step": 424970
    },
    {
      "epoch": 0.1015683505897846,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011501570157015701,
      "loss": 3.1528,
      "step": 424980
    },
    {
      "epoch": 0.10158699382594279,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011501370137013702,
      "loss": 3.1296,
      "step": 424990
    },
    {
      "epoch": 0.10160563706210098,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011501170117011702,
      "loss": 3.0853,
      "step": 425000
    },
    {
      "epoch": 0.10162428029825915,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011500970097009703,
      "loss": 3.3586,
      "step": 425010
    },
    {
      "epoch": 0.10164292353441734,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011500770077007701,
      "loss": 3.2346,
      "step": 425020
    },
    {
      "epoch": 0.10166156677057551,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011500570057005702,
      "loss": 3.1766,
      "step": 425030
    },
    {
      "epoch": 0.1016802100067337,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000115003700370037,
      "loss": 3.2058,
      "step": 425040
    },
    {
      "epoch": 0.10169885324289189,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011500170017001701,
      "loss": 3.2685,
      "step": 425050
    },
    {
      "epoch": 0.10171749647905007,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000114999699969997,
      "loss": 3.1797,
      "step": 425060
    },
    {
      "epoch": 0.10173613971520826,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011499769976997701,
      "loss": 3.2121,
      "step": 425070
    },
    {
      "epoch": 0.10175478295136643,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000114995699569957,
      "loss": 3.0922,
      "step": 425080
    },
    {
      "epoch": 0.10177342618752462,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000114993699369937,
      "loss": 3.1294,
      "step": 425090
    },
    {
      "epoch": 0.10179206942368281,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011499169916991699,
      "loss": 3.1209,
      "step": 425100
    },
    {
      "epoch": 0.10181071265984099,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000114989698969897,
      "loss": 3.2793,
      "step": 425110
    },
    {
      "epoch": 0.10182935589599917,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011498769876987698,
      "loss": 3.1,
      "step": 425120
    },
    {
      "epoch": 0.10184799913215736,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011498569856985699,
      "loss": 3.2192,
      "step": 425130
    },
    {
      "epoch": 0.10186664236831554,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011498369836983698,
      "loss": 3.2404,
      "step": 425140
    },
    {
      "epoch": 0.10188528560447373,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011498169816981699,
      "loss": 3.1377,
      "step": 425150
    },
    {
      "epoch": 0.1019039288406319,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000114979697969797,
      "loss": 3.166,
      "step": 425160
    },
    {
      "epoch": 0.10192257207679009,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011497769776977698,
      "loss": 3.3074,
      "step": 425170
    },
    {
      "epoch": 0.10194121531294828,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011497569756975699,
      "loss": 3.224,
      "step": 425180
    },
    {
      "epoch": 0.10195985854910646,
      "grad_norm": 0.75,
      "learning_rate": 0.00011497369736973697,
      "loss": 3.0996,
      "step": 425190
    },
    {
      "epoch": 0.10197850178526464,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011497169716971698,
      "loss": 3.3096,
      "step": 425200
    },
    {
      "epoch": 0.10199714502142282,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011496969696969697,
      "loss": 3.2524,
      "step": 425210
    },
    {
      "epoch": 0.10201578825758101,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011496769676967698,
      "loss": 3.0751,
      "step": 425220
    },
    {
      "epoch": 0.1020344314937392,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011496569656965697,
      "loss": 3.1932,
      "step": 425230
    },
    {
      "epoch": 0.10205307472989737,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011496369636963698,
      "loss": 3.2558,
      "step": 425240
    },
    {
      "epoch": 0.10207171796605556,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011496169616961696,
      "loss": 3.124,
      "step": 425250
    },
    {
      "epoch": 0.10209036120221375,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011495969596959697,
      "loss": 3.0794,
      "step": 425260
    },
    {
      "epoch": 0.10210900443837193,
      "grad_norm": 0.90625,
      "learning_rate": 0.00011495769576957695,
      "loss": 3.2524,
      "step": 425270
    },
    {
      "epoch": 0.10212764767453011,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011495569556955696,
      "loss": 3.2645,
      "step": 425280
    },
    {
      "epoch": 0.10214629091068829,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011495369536953696,
      "loss": 3.0815,
      "step": 425290
    },
    {
      "epoch": 0.10216493414684648,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011495169516951695,
      "loss": 3.1104,
      "step": 425300
    },
    {
      "epoch": 0.10218357738300467,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011494969496949695,
      "loss": 3.246,
      "step": 425310
    },
    {
      "epoch": 0.10220222061916284,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011494769476947696,
      "loss": 3.0856,
      "step": 425320
    },
    {
      "epoch": 0.10222086385532103,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011494569456945694,
      "loss": 3.1334,
      "step": 425330
    },
    {
      "epoch": 0.10223950709147922,
      "grad_norm": 0.625,
      "learning_rate": 0.00011494369436943695,
      "loss": 3.1181,
      "step": 425340
    },
    {
      "epoch": 0.1022581503276374,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011494169416941696,
      "loss": 3.2207,
      "step": 425350
    },
    {
      "epoch": 0.10227679356379558,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011493969396939694,
      "loss": 3.2015,
      "step": 425360
    },
    {
      "epoch": 0.10229543679995376,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011493769376937695,
      "loss": 3.1001,
      "step": 425370
    },
    {
      "epoch": 0.10231408003611195,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011493569356935695,
      "loss": 3.1616,
      "step": 425380
    },
    {
      "epoch": 0.10233272327227014,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011493369336933695,
      "loss": 3.1655,
      "step": 425390
    },
    {
      "epoch": 0.10235136650842831,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011493169316931694,
      "loss": 3.0967,
      "step": 425400
    },
    {
      "epoch": 0.1023700097445865,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011492969296929695,
      "loss": 3.1059,
      "step": 425410
    },
    {
      "epoch": 0.10238865298074468,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011492769276927693,
      "loss": 3.1984,
      "step": 425420
    },
    {
      "epoch": 0.10240729621690287,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011492569256925694,
      "loss": 3.2138,
      "step": 425430
    },
    {
      "epoch": 0.10242593945306105,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011492369236923692,
      "loss": 3.0907,
      "step": 425440
    },
    {
      "epoch": 0.10244458268921923,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011492169216921693,
      "loss": 3.0219,
      "step": 425450
    },
    {
      "epoch": 0.10246322592537742,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011491969196919692,
      "loss": 3.1149,
      "step": 425460
    },
    {
      "epoch": 0.10248186916153561,
      "grad_norm": 0.75,
      "learning_rate": 0.00011491769176917693,
      "loss": 2.9679,
      "step": 425470
    },
    {
      "epoch": 0.10250051239769378,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011491569156915691,
      "loss": 3.268,
      "step": 425480
    },
    {
      "epoch": 0.10251915563385197,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011491369136913692,
      "loss": 3.2347,
      "step": 425490
    },
    {
      "epoch": 0.10253779887001015,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001149116911691169,
      "loss": 3.1141,
      "step": 425500
    },
    {
      "epoch": 0.10255644210616834,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011490969096909692,
      "loss": 3.1266,
      "step": 425510
    },
    {
      "epoch": 0.10257508534232652,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001149076907690769,
      "loss": 3.1124,
      "step": 425520
    },
    {
      "epoch": 0.1025937285784847,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001149056905690569,
      "loss": 3.2081,
      "step": 425530
    },
    {
      "epoch": 0.10261237181464289,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001149036903690369,
      "loss": 3.1678,
      "step": 425540
    },
    {
      "epoch": 0.10263101505080106,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011490169016901691,
      "loss": 3.2195,
      "step": 425550
    },
    {
      "epoch": 0.10264965828695925,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011489968996899692,
      "loss": 3.2255,
      "step": 425560
    },
    {
      "epoch": 0.10266830152311744,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001148976897689769,
      "loss": 3.0625,
      "step": 425570
    },
    {
      "epoch": 0.10268694475927562,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011489568956895691,
      "loss": 3.2147,
      "step": 425580
    },
    {
      "epoch": 0.1027055879954338,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001148936893689369,
      "loss": 3.1867,
      "step": 425590
    },
    {
      "epoch": 0.102724231231592,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001148916891689169,
      "loss": 3.1749,
      "step": 425600
    },
    {
      "epoch": 0.10274287446775017,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001148896889688969,
      "loss": 3.1588,
      "step": 425610
    },
    {
      "epoch": 0.10276151770390836,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001148876887688769,
      "loss": 3.3345,
      "step": 425620
    },
    {
      "epoch": 0.10278016094006653,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011488568856885689,
      "loss": 3.222,
      "step": 425630
    },
    {
      "epoch": 0.10279880417622472,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001148836883688369,
      "loss": 3.2471,
      "step": 425640
    },
    {
      "epoch": 0.10281744741238291,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011488168816881688,
      "loss": 3.1191,
      "step": 425650
    },
    {
      "epoch": 0.10283609064854109,
      "grad_norm": 0.625,
      "learning_rate": 0.00011487968796879689,
      "loss": 3.24,
      "step": 425660
    },
    {
      "epoch": 0.10285473388469928,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011487768776877687,
      "loss": 2.9624,
      "step": 425670
    },
    {
      "epoch": 0.10287337712085746,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011487568756875688,
      "loss": 3.1414,
      "step": 425680
    },
    {
      "epoch": 0.10289202035701564,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011487368736873688,
      "loss": 3.1564,
      "step": 425690
    },
    {
      "epoch": 0.10291066359317383,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011487168716871689,
      "loss": 3.3187,
      "step": 425700
    },
    {
      "epoch": 0.102929306829332,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011486968696869687,
      "loss": 3.066,
      "step": 425710
    },
    {
      "epoch": 0.10294795006549019,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011486768676867688,
      "loss": 3.4232,
      "step": 425720
    },
    {
      "epoch": 0.10296659330164838,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011486568656865686,
      "loss": 3.1038,
      "step": 425730
    },
    {
      "epoch": 0.10298523653780656,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011486368636863687,
      "loss": 3.2413,
      "step": 425740
    },
    {
      "epoch": 0.10300387977396475,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011486168616861688,
      "loss": 3.125,
      "step": 425750
    },
    {
      "epoch": 0.10302252301012292,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011485968596859686,
      "loss": 3.1052,
      "step": 425760
    },
    {
      "epoch": 0.10304116624628111,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011485768576857687,
      "loss": 3.1615,
      "step": 425770
    },
    {
      "epoch": 0.1030598094824393,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011485568556855686,
      "loss": 3.115,
      "step": 425780
    },
    {
      "epoch": 0.10307845271859747,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011485368536853687,
      "loss": 3.2456,
      "step": 425790
    },
    {
      "epoch": 0.10309709595475566,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011485168516851686,
      "loss": 3.1267,
      "step": 425800
    },
    {
      "epoch": 0.10311573919091385,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011484968496849686,
      "loss": 3.2162,
      "step": 425810
    },
    {
      "epoch": 0.10313438242707203,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011484768476847685,
      "loss": 3.1658,
      "step": 425820
    },
    {
      "epoch": 0.10315302566323022,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011484568456845686,
      "loss": 3.2046,
      "step": 425830
    },
    {
      "epoch": 0.10317166889938839,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011484368436843684,
      "loss": 3.1414,
      "step": 425840
    },
    {
      "epoch": 0.10319031213554658,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011484168416841685,
      "loss": 3.1439,
      "step": 425850
    },
    {
      "epoch": 0.10320895537170477,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011483968396839684,
      "loss": 3.0656,
      "step": 425860
    },
    {
      "epoch": 0.10322759860786294,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011483768376837685,
      "loss": 3.1214,
      "step": 425870
    },
    {
      "epoch": 0.10324624184402113,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011483568356835683,
      "loss": 3.2801,
      "step": 425880
    },
    {
      "epoch": 0.10326488508017932,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011483368336833684,
      "loss": 3.2315,
      "step": 425890
    },
    {
      "epoch": 0.1032835283163375,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011483168316831682,
      "loss": 3.1316,
      "step": 425900
    },
    {
      "epoch": 0.10330217155249569,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011482968296829683,
      "loss": 3.1213,
      "step": 425910
    },
    {
      "epoch": 0.10332081478865386,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011482768276827683,
      "loss": 3.2236,
      "step": 425920
    },
    {
      "epoch": 0.10333945802481205,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011482568256825682,
      "loss": 3.0822,
      "step": 425930
    },
    {
      "epoch": 0.10335810126097024,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011482368236823683,
      "loss": 3.1787,
      "step": 425940
    },
    {
      "epoch": 0.10337674449712841,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011482168216821683,
      "loss": 3.1372,
      "step": 425950
    },
    {
      "epoch": 0.1033953877332866,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011481968196819684,
      "loss": 3.1688,
      "step": 425960
    },
    {
      "epoch": 0.10341403096944478,
      "grad_norm": 0.75,
      "learning_rate": 0.00011481768176817682,
      "loss": 3.2074,
      "step": 425970
    },
    {
      "epoch": 0.10343267420560297,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011481568156815683,
      "loss": 3.2088,
      "step": 425980
    },
    {
      "epoch": 0.10345131744176116,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011481368136813681,
      "loss": 3.1344,
      "step": 425990
    },
    {
      "epoch": 0.10346996067791933,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011481168116811682,
      "loss": 3.2679,
      "step": 426000
    },
    {
      "epoch": 0.10348860391407752,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011480968096809682,
      "loss": 3.1892,
      "step": 426010
    },
    {
      "epoch": 0.10350724715023571,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011480768076807683,
      "loss": 3.0904,
      "step": 426020
    },
    {
      "epoch": 0.10352589038639388,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011480568056805681,
      "loss": 3.0735,
      "step": 426030
    },
    {
      "epoch": 0.10354453362255207,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011480368036803682,
      "loss": 3.1699,
      "step": 426040
    },
    {
      "epoch": 0.10356317685871025,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001148016801680168,
      "loss": 3.2308,
      "step": 426050
    },
    {
      "epoch": 0.10358182009486844,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011479967996799681,
      "loss": 3.18,
      "step": 426060
    },
    {
      "epoch": 0.10360046333102663,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011479767976797679,
      "loss": 3.2326,
      "step": 426070
    },
    {
      "epoch": 0.1036191065671848,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001147956795679568,
      "loss": 3.1792,
      "step": 426080
    },
    {
      "epoch": 0.10363774980334299,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001147936793679368,
      "loss": 3.1371,
      "step": 426090
    },
    {
      "epoch": 0.10365639303950117,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001147916791679168,
      "loss": 3.2001,
      "step": 426100
    },
    {
      "epoch": 0.10367503627565935,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011478967896789679,
      "loss": 3.1401,
      "step": 426110
    },
    {
      "epoch": 0.10369367951181754,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001147876787678768,
      "loss": 3.1826,
      "step": 426120
    },
    {
      "epoch": 0.10371232274797572,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011478567856785678,
      "loss": 3.1194,
      "step": 426130
    },
    {
      "epoch": 0.10373096598413391,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011478367836783679,
      "loss": 3.0417,
      "step": 426140
    },
    {
      "epoch": 0.1037496092202921,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001147816781678168,
      "loss": 3.2764,
      "step": 426150
    },
    {
      "epoch": 0.10376825245645027,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011477967796779678,
      "loss": 3.1899,
      "step": 426160
    },
    {
      "epoch": 0.10378689569260846,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011477767776777679,
      "loss": 3.2872,
      "step": 426170
    },
    {
      "epoch": 0.10380553892876664,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011477567756775678,
      "loss": 3.1162,
      "step": 426180
    },
    {
      "epoch": 0.10382418216492482,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011477367736773679,
      "loss": 3.0552,
      "step": 426190
    },
    {
      "epoch": 0.10384282540108301,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011477167716771677,
      "loss": 3.091,
      "step": 426200
    },
    {
      "epoch": 0.10386146863724119,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011476967696769678,
      "loss": 3.1608,
      "step": 426210
    },
    {
      "epoch": 0.10388011187339938,
      "grad_norm": 0.75,
      "learning_rate": 0.00011476767676767676,
      "loss": 3.0048,
      "step": 426220
    },
    {
      "epoch": 0.10389875510955757,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011476567656765677,
      "loss": 3.1226,
      "step": 426230
    },
    {
      "epoch": 0.10391739834571574,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011476367636763677,
      "loss": 3.0834,
      "step": 426240
    },
    {
      "epoch": 0.10393604158187393,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011476167616761677,
      "loss": 3.168,
      "step": 426250
    },
    {
      "epoch": 0.1039546848180321,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011475967596759676,
      "loss": 3.1524,
      "step": 426260
    },
    {
      "epoch": 0.1039733280541903,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011475767576757677,
      "loss": 3.0887,
      "step": 426270
    },
    {
      "epoch": 0.10399197129034848,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011475567556755675,
      "loss": 3.1018,
      "step": 426280
    },
    {
      "epoch": 0.10401061452650666,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011475367536753676,
      "loss": 3.1709,
      "step": 426290
    },
    {
      "epoch": 0.10402925776266485,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011475167516751674,
      "loss": 3.176,
      "step": 426300
    },
    {
      "epoch": 0.10404790099882302,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011474967496749675,
      "loss": 3.106,
      "step": 426310
    },
    {
      "epoch": 0.10406654423498121,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011474767476747675,
      "loss": 3.1342,
      "step": 426320
    },
    {
      "epoch": 0.1040851874711394,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011474567456745676,
      "loss": 3.0519,
      "step": 426330
    },
    {
      "epoch": 0.10410383070729758,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011474367436743677,
      "loss": 3.066,
      "step": 426340
    },
    {
      "epoch": 0.10412247394345577,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011474167416741675,
      "loss": 3.1293,
      "step": 426350
    },
    {
      "epoch": 0.10414111717961395,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011473967396739676,
      "loss": 3.2086,
      "step": 426360
    },
    {
      "epoch": 0.10415976041577213,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011473767376737674,
      "loss": 3.1008,
      "step": 426370
    },
    {
      "epoch": 0.10417840365193032,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011473567356735675,
      "loss": 3.1965,
      "step": 426380
    },
    {
      "epoch": 0.1041970468880885,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011473367336733673,
      "loss": 3.0909,
      "step": 426390
    },
    {
      "epoch": 0.10421569012424668,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011473167316731674,
      "loss": 2.9947,
      "step": 426400
    },
    {
      "epoch": 0.10423433336040487,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011472967296729674,
      "loss": 3.1868,
      "step": 426410
    },
    {
      "epoch": 0.10425297659656305,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011472767276727674,
      "loss": 3.0862,
      "step": 426420
    },
    {
      "epoch": 0.10427161983272124,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011472567256725673,
      "loss": 3.1289,
      "step": 426430
    },
    {
      "epoch": 0.10429026306887941,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011472367236723674,
      "loss": 3.1466,
      "step": 426440
    },
    {
      "epoch": 0.1043089063050376,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011472167216721672,
      "loss": 3.1433,
      "step": 426450
    },
    {
      "epoch": 0.10432754954119579,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011471967196719673,
      "loss": 3.1752,
      "step": 426460
    },
    {
      "epoch": 0.10434619277735396,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011471767176717671,
      "loss": 3.0528,
      "step": 426470
    },
    {
      "epoch": 0.10436483601351215,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011471567156715672,
      "loss": 3.1174,
      "step": 426480
    },
    {
      "epoch": 0.10438347924967034,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011471367136713671,
      "loss": 3.16,
      "step": 426490
    },
    {
      "epoch": 0.10440212248582852,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011471167116711672,
      "loss": 3.2123,
      "step": 426500
    },
    {
      "epoch": 0.1044207657219867,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001147096709670967,
      "loss": 3.2844,
      "step": 426510
    },
    {
      "epoch": 0.10443940895814488,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011470767076707671,
      "loss": 3.114,
      "step": 426520
    },
    {
      "epoch": 0.10445805219430307,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011470567056705672,
      "loss": 3.0767,
      "step": 426530
    },
    {
      "epoch": 0.10447669543046126,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001147036703670367,
      "loss": 3.0625,
      "step": 426540
    },
    {
      "epoch": 0.10449533866661943,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011470167016701671,
      "loss": 3.1343,
      "step": 426550
    },
    {
      "epoch": 0.10451398190277762,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011469966996699671,
      "loss": 3.0961,
      "step": 426560
    },
    {
      "epoch": 0.10453262513893581,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001146976697669767,
      "loss": 3.0543,
      "step": 426570
    },
    {
      "epoch": 0.10455126837509399,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001146956695669567,
      "loss": 3.2577,
      "step": 426580
    },
    {
      "epoch": 0.10456991161125218,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011469366936693671,
      "loss": 3.238,
      "step": 426590
    },
    {
      "epoch": 0.10458855484741035,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011469166916691669,
      "loss": 3.2248,
      "step": 426600
    },
    {
      "epoch": 0.10460719808356854,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001146896689668967,
      "loss": 3.1449,
      "step": 426610
    },
    {
      "epoch": 0.10462584131972673,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011468766876687668,
      "loss": 3.1654,
      "step": 426620
    },
    {
      "epoch": 0.1046444845558849,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011468566856685669,
      "loss": 3.1497,
      "step": 426630
    },
    {
      "epoch": 0.10466312779204309,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011468366836683669,
      "loss": 3.1514,
      "step": 426640
    },
    {
      "epoch": 0.10468177102820127,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001146816681668167,
      "loss": 3.0751,
      "step": 426650
    },
    {
      "epoch": 0.10470041426435946,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011467966796679668,
      "loss": 3.1333,
      "step": 426660
    },
    {
      "epoch": 0.10471905750051765,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011467766776677669,
      "loss": 3.1084,
      "step": 426670
    },
    {
      "epoch": 0.10473770073667582,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011467566756675667,
      "loss": 3.2264,
      "step": 426680
    },
    {
      "epoch": 0.10475634397283401,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011467366736673668,
      "loss": 3.0548,
      "step": 426690
    },
    {
      "epoch": 0.1047749872089922,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011467166716671666,
      "loss": 3.1632,
      "step": 426700
    },
    {
      "epoch": 0.10479363044515037,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011466966696669667,
      "loss": 3.1826,
      "step": 426710
    },
    {
      "epoch": 0.10481227368130856,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011466766676667667,
      "loss": 3.2028,
      "step": 426720
    },
    {
      "epoch": 0.10483091691746674,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011466566656665668,
      "loss": 3.2635,
      "step": 426730
    },
    {
      "epoch": 0.10484956015362493,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011466366636663668,
      "loss": 3.0883,
      "step": 426740
    },
    {
      "epoch": 0.10486820338978312,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011466166616661667,
      "loss": 3.1433,
      "step": 426750
    },
    {
      "epoch": 0.10488684662594129,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011465966596659668,
      "loss": 3.1226,
      "step": 426760
    },
    {
      "epoch": 0.10490548986209948,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011465766576657666,
      "loss": 3.2001,
      "step": 426770
    },
    {
      "epoch": 0.10492413309825765,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011465566556655667,
      "loss": 3.1895,
      "step": 426780
    },
    {
      "epoch": 0.10494277633441584,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011465366536653665,
      "loss": 3.1311,
      "step": 426790
    },
    {
      "epoch": 0.10496141957057403,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011465166516651666,
      "loss": 3.2994,
      "step": 426800
    },
    {
      "epoch": 0.10498006280673221,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011464966496649665,
      "loss": 3.0925,
      "step": 426810
    },
    {
      "epoch": 0.1049987060428904,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011464766476647666,
      "loss": 3.1743,
      "step": 426820
    },
    {
      "epoch": 0.10501734927904859,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011464566456645665,
      "loss": 3.0778,
      "step": 426830
    },
    {
      "epoch": 0.10503599251520676,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011464366436643665,
      "loss": 3.1125,
      "step": 426840
    },
    {
      "epoch": 0.10505463575136495,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011464166416641664,
      "loss": 3.1547,
      "step": 426850
    },
    {
      "epoch": 0.10507327898752312,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011463966396639665,
      "loss": 3.2015,
      "step": 426860
    },
    {
      "epoch": 0.10509192222368131,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011463766376637664,
      "loss": 3.2202,
      "step": 426870
    },
    {
      "epoch": 0.1051105654598395,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011463566356635665,
      "loss": 3.0563,
      "step": 426880
    },
    {
      "epoch": 0.10512920869599768,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011463366336633663,
      "loss": 3.188,
      "step": 426890
    },
    {
      "epoch": 0.10514785193215587,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011463166316631664,
      "loss": 3.1236,
      "step": 426900
    },
    {
      "epoch": 0.10516649516831406,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011462966296629662,
      "loss": 3.2454,
      "step": 426910
    },
    {
      "epoch": 0.10518513840447223,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011462766276627663,
      "loss": 3.0882,
      "step": 426920
    },
    {
      "epoch": 0.10520378164063042,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011462566256625664,
      "loss": 3.1356,
      "step": 426930
    },
    {
      "epoch": 0.1052224248767886,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011462366236623662,
      "loss": 3.132,
      "step": 426940
    },
    {
      "epoch": 0.10524106811294678,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011462166216621663,
      "loss": 3.3159,
      "step": 426950
    },
    {
      "epoch": 0.10525971134910497,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011461966196619663,
      "loss": 3.0262,
      "step": 426960
    },
    {
      "epoch": 0.10527835458526315,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011461766176617664,
      "loss": 3.2523,
      "step": 426970
    },
    {
      "epoch": 0.10529699782142134,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011461566156615662,
      "loss": 3.2148,
      "step": 426980
    },
    {
      "epoch": 0.10531564105757951,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011461366136613663,
      "loss": 3.156,
      "step": 426990
    },
    {
      "epoch": 0.1053342842937377,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011461166116611661,
      "loss": 3.14,
      "step": 427000
    },
    {
      "epoch": 0.10535292752989589,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011460966096609662,
      "loss": 3.0619,
      "step": 427010
    },
    {
      "epoch": 0.10537157076605407,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001146076607660766,
      "loss": 3.2484,
      "step": 427020
    },
    {
      "epoch": 0.10539021400221225,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011460566056605661,
      "loss": 3.2299,
      "step": 427030
    },
    {
      "epoch": 0.10540885723837044,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011460366036603661,
      "loss": 3.0717,
      "step": 427040
    },
    {
      "epoch": 0.10542750047452862,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011460166016601662,
      "loss": 3.1682,
      "step": 427050
    },
    {
      "epoch": 0.10544614371068681,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001145996599659966,
      "loss": 3.2282,
      "step": 427060
    },
    {
      "epoch": 0.10546478694684498,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011459765976597661,
      "loss": 2.9645,
      "step": 427070
    },
    {
      "epoch": 0.10548343018300317,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011459565956595659,
      "loss": 3.0925,
      "step": 427080
    },
    {
      "epoch": 0.10550207341916136,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001145936593659366,
      "loss": 3.1834,
      "step": 427090
    },
    {
      "epoch": 0.10552071665531954,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001145916591659166,
      "loss": 3.0536,
      "step": 427100
    },
    {
      "epoch": 0.10553935989147772,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011458965896589659,
      "loss": 3.1415,
      "step": 427110
    },
    {
      "epoch": 0.1055580031276359,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011458765876587659,
      "loss": 3.2566,
      "step": 427120
    },
    {
      "epoch": 0.10557664636379409,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001145856585658566,
      "loss": 3.1181,
      "step": 427130
    },
    {
      "epoch": 0.10559528959995228,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001145836583658366,
      "loss": 3.1201,
      "step": 427140
    },
    {
      "epoch": 0.10561393283611045,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011458165816581659,
      "loss": 3.069,
      "step": 427150
    },
    {
      "epoch": 0.10563257607226864,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001145796579657966,
      "loss": 3.0736,
      "step": 427160
    },
    {
      "epoch": 0.10565121930842683,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011457765776577658,
      "loss": 3.0483,
      "step": 427170
    },
    {
      "epoch": 0.105669862544585,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011457565756575659,
      "loss": 3.1743,
      "step": 427180
    },
    {
      "epoch": 0.1056885057807432,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011457365736573658,
      "loss": 3.1861,
      "step": 427190
    },
    {
      "epoch": 0.10570714901690137,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011457165716571659,
      "loss": 3.1488,
      "step": 427200
    },
    {
      "epoch": 0.10572579225305956,
      "grad_norm": 0.625,
      "learning_rate": 0.00011456965696569657,
      "loss": 3.2719,
      "step": 427210
    },
    {
      "epoch": 0.10574443548921775,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011456765676567658,
      "loss": 3.1927,
      "step": 427220
    },
    {
      "epoch": 0.10576307872537592,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011456565656565656,
      "loss": 3.2956,
      "step": 427230
    },
    {
      "epoch": 0.10578172196153411,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011456365636563657,
      "loss": 3.2245,
      "step": 427240
    },
    {
      "epoch": 0.1058003651976923,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011456165616561655,
      "loss": 3.2093,
      "step": 427250
    },
    {
      "epoch": 0.10581900843385048,
      "grad_norm": 0.75,
      "learning_rate": 0.00011455965596559656,
      "loss": 3.2466,
      "step": 427260
    },
    {
      "epoch": 0.10583765167000866,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011455765576557656,
      "loss": 3.004,
      "step": 427270
    },
    {
      "epoch": 0.10585629490616684,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011455565556555657,
      "loss": 3.1467,
      "step": 427280
    },
    {
      "epoch": 0.10587493814232503,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011455365536553655,
      "loss": 3.2332,
      "step": 427290
    },
    {
      "epoch": 0.10589358137848322,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011455165516551656,
      "loss": 3.2447,
      "step": 427300
    },
    {
      "epoch": 0.10591222461464139,
      "grad_norm": 0.75,
      "learning_rate": 0.00011454965496549654,
      "loss": 3.2493,
      "step": 427310
    },
    {
      "epoch": 0.10593086785079958,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011454765476547655,
      "loss": 3.1993,
      "step": 427320
    },
    {
      "epoch": 0.10594951108695776,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011454565456545656,
      "loss": 3.1805,
      "step": 427330
    },
    {
      "epoch": 0.10596815432311595,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011454365436543654,
      "loss": 3.2069,
      "step": 427340
    },
    {
      "epoch": 0.10598679755927413,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011454165416541655,
      "loss": 3.0979,
      "step": 427350
    },
    {
      "epoch": 0.10600544079543231,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011453965396539655,
      "loss": 3.1633,
      "step": 427360
    },
    {
      "epoch": 0.1060240840315905,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011453765376537656,
      "loss": 3.2299,
      "step": 427370
    },
    {
      "epoch": 0.10604272726774869,
      "grad_norm": 0.875,
      "learning_rate": 0.00011453565356535654,
      "loss": 3.2225,
      "step": 427380
    },
    {
      "epoch": 0.10606137050390686,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011453365336533655,
      "loss": 3.0533,
      "step": 427390
    },
    {
      "epoch": 0.10608001374006505,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011453165316531653,
      "loss": 3.2728,
      "step": 427400
    },
    {
      "epoch": 0.10609865697622323,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011452965296529654,
      "loss": 3.2052,
      "step": 427410
    },
    {
      "epoch": 0.10611730021238142,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011452765276527652,
      "loss": 3.2494,
      "step": 427420
    },
    {
      "epoch": 0.1061359434485396,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011452565256525653,
      "loss": 3.1144,
      "step": 427430
    },
    {
      "epoch": 0.10615458668469778,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011452365236523653,
      "loss": 3.1641,
      "step": 427440
    },
    {
      "epoch": 0.10617322992085597,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011452165216521653,
      "loss": 3.2112,
      "step": 427450
    },
    {
      "epoch": 0.10619187315701414,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011451965196519652,
      "loss": 3.0939,
      "step": 427460
    },
    {
      "epoch": 0.10621051639317233,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011451765176517653,
      "loss": 3.1676,
      "step": 427470
    },
    {
      "epoch": 0.10622915962933052,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011451565156515651,
      "loss": 3.2057,
      "step": 427480
    },
    {
      "epoch": 0.1062478028654887,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011451365136513652,
      "loss": 3.2465,
      "step": 427490
    },
    {
      "epoch": 0.10626644610164689,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011451165116511651,
      "loss": 3.2434,
      "step": 427500
    },
    {
      "epoch": 0.10628508933780507,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011450965096509652,
      "loss": 3.1936,
      "step": 427510
    },
    {
      "epoch": 0.10630373257396325,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011450765076507652,
      "loss": 3.1365,
      "step": 427520
    },
    {
      "epoch": 0.10632237581012144,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011450565056505651,
      "loss": 3.2116,
      "step": 427530
    },
    {
      "epoch": 0.10634101904627961,
      "grad_norm": 0.75,
      "learning_rate": 0.00011450365036503652,
      "loss": 3.0364,
      "step": 427540
    },
    {
      "epoch": 0.1063596622824378,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001145016501650165,
      "loss": 3.2659,
      "step": 427550
    },
    {
      "epoch": 0.10637830551859599,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011449964996499651,
      "loss": 3.0755,
      "step": 427560
    },
    {
      "epoch": 0.10639694875475417,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001144976497649765,
      "loss": 3.1557,
      "step": 427570
    },
    {
      "epoch": 0.10641559199091236,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001144956495649565,
      "loss": 3.1912,
      "step": 427580
    },
    {
      "epoch": 0.10643423522707054,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001144936493649365,
      "loss": 3.2255,
      "step": 427590
    },
    {
      "epoch": 0.10645287846322872,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011449164916491651,
      "loss": 3.1544,
      "step": 427600
    },
    {
      "epoch": 0.10647152169938691,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011448964896489649,
      "loss": 3.0631,
      "step": 427610
    },
    {
      "epoch": 0.10649016493554508,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001144876487648765,
      "loss": 3.2016,
      "step": 427620
    },
    {
      "epoch": 0.10650880817170327,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011448564856485648,
      "loss": 3.1222,
      "step": 427630
    },
    {
      "epoch": 0.10652745140786146,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011448364836483649,
      "loss": 3.0526,
      "step": 427640
    },
    {
      "epoch": 0.10654609464401964,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011448164816481647,
      "loss": 3.1001,
      "step": 427650
    },
    {
      "epoch": 0.10656473788017783,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011447964796479648,
      "loss": 3.1244,
      "step": 427660
    },
    {
      "epoch": 0.106583381116336,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011447764776477648,
      "loss": 3.0313,
      "step": 427670
    },
    {
      "epoch": 0.10660202435249419,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011447564756475649,
      "loss": 3.099,
      "step": 427680
    },
    {
      "epoch": 0.10662066758865238,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011447364736473647,
      "loss": 3.1282,
      "step": 427690
    },
    {
      "epoch": 0.10663931082481055,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011447164716471648,
      "loss": 3.1797,
      "step": 427700
    },
    {
      "epoch": 0.10665795406096874,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011446964696469646,
      "loss": 3.106,
      "step": 427710
    },
    {
      "epoch": 0.10667659729712693,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011446764676467647,
      "loss": 3.3572,
      "step": 427720
    },
    {
      "epoch": 0.10669524053328511,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011446564656465648,
      "loss": 3.1355,
      "step": 427730
    },
    {
      "epoch": 0.1067138837694433,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011446364636463646,
      "loss": 3.1726,
      "step": 427740
    },
    {
      "epoch": 0.10673252700560147,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011446164616461647,
      "loss": 3.1332,
      "step": 427750
    },
    {
      "epoch": 0.10675117024175966,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011445964596459647,
      "loss": 3.2402,
      "step": 427760
    },
    {
      "epoch": 0.10676981347791785,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011445764576457647,
      "loss": 3.2751,
      "step": 427770
    },
    {
      "epoch": 0.10678845671407602,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00011445564556455646,
      "loss": 3.1688,
      "step": 427780
    },
    {
      "epoch": 0.10680709995023421,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011445364536453647,
      "loss": 3.0774,
      "step": 427790
    },
    {
      "epoch": 0.10682574318639239,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011445164516451645,
      "loss": 3.3243,
      "step": 427800
    },
    {
      "epoch": 0.10684438642255058,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011444964496449646,
      "loss": 3.2077,
      "step": 427810
    },
    {
      "epoch": 0.10686302965870877,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011444764476447645,
      "loss": 2.9435,
      "step": 427820
    },
    {
      "epoch": 0.10688167289486694,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011444564456445646,
      "loss": 2.9789,
      "step": 427830
    },
    {
      "epoch": 0.10690031613102513,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011444364436443644,
      "loss": 3.2236,
      "step": 427840
    },
    {
      "epoch": 0.10691895936718332,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011444164416441645,
      "loss": 3.232,
      "step": 427850
    },
    {
      "epoch": 0.1069376026033415,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011443964396439644,
      "loss": 3.2395,
      "step": 427860
    },
    {
      "epoch": 0.10695624583949968,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011443764376437644,
      "loss": 3.169,
      "step": 427870
    },
    {
      "epoch": 0.10697488907565786,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011443564356435643,
      "loss": 3.1129,
      "step": 427880
    },
    {
      "epoch": 0.10699353231181605,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011443364336433644,
      "loss": 3.2234,
      "step": 427890
    },
    {
      "epoch": 0.10701217554797424,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011443164316431643,
      "loss": 3.1453,
      "step": 427900
    },
    {
      "epoch": 0.10703081878413241,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011442964296429644,
      "loss": 3.198,
      "step": 427910
    },
    {
      "epoch": 0.1070494620202906,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011442764276427645,
      "loss": 3.1199,
      "step": 427920
    },
    {
      "epoch": 0.10706810525644879,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00011442564256425643,
      "loss": 3.1811,
      "step": 427930
    },
    {
      "epoch": 0.10708674849260696,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011442364236423644,
      "loss": 3.1247,
      "step": 427940
    },
    {
      "epoch": 0.10710539172876515,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011442164216421642,
      "loss": 3.1953,
      "step": 427950
    },
    {
      "epoch": 0.10712403496492333,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011441964196419643,
      "loss": 3.1358,
      "step": 427960
    },
    {
      "epoch": 0.10714267820108152,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011441764176417641,
      "loss": 3.0806,
      "step": 427970
    },
    {
      "epoch": 0.1071613214372397,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011441564156415642,
      "loss": 3.1073,
      "step": 427980
    },
    {
      "epoch": 0.10717996467339788,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011441364136413642,
      "loss": 3.1324,
      "step": 427990
    },
    {
      "epoch": 0.10719860790955607,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011441164116411643,
      "loss": 3.0763,
      "step": 428000
    },
    {
      "epoch": 0.10721725114571425,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011440964096409641,
      "loss": 3.2426,
      "step": 428010
    },
    {
      "epoch": 0.10723589438187243,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011440764076407642,
      "loss": 3.1131,
      "step": 428020
    },
    {
      "epoch": 0.10725453761803062,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001144056405640564,
      "loss": 3.2539,
      "step": 428030
    },
    {
      "epoch": 0.1072731808541888,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011440364036403641,
      "loss": 3.1431,
      "step": 428040
    },
    {
      "epoch": 0.10729182409034699,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001144016401640164,
      "loss": 3.2175,
      "step": 428050
    },
    {
      "epoch": 0.10731046732650518,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001143996399639964,
      "loss": 3.2059,
      "step": 428060
    },
    {
      "epoch": 0.10732911056266335,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001143976397639764,
      "loss": 3.2513,
      "step": 428070
    },
    {
      "epoch": 0.10734775379882154,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001143956395639564,
      "loss": 3.1887,
      "step": 428080
    },
    {
      "epoch": 0.10736639703497972,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011439363936393639,
      "loss": 3.3126,
      "step": 428090
    },
    {
      "epoch": 0.1073850402711379,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001143916391639164,
      "loss": 3.1363,
      "step": 428100
    },
    {
      "epoch": 0.1074036835072961,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001143896389638964,
      "loss": 3.2019,
      "step": 428110
    },
    {
      "epoch": 0.10742232674345427,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011438763876387639,
      "loss": 3.1875,
      "step": 428120
    },
    {
      "epoch": 0.10744096997961246,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001143856385638564,
      "loss": 3.2189,
      "step": 428130
    },
    {
      "epoch": 0.10745961321577063,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001143836383638364,
      "loss": 3.2096,
      "step": 428140
    },
    {
      "epoch": 0.10747825645192882,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001143816381638164,
      "loss": 3.3292,
      "step": 428150
    },
    {
      "epoch": 0.10749689968808701,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011437963796379638,
      "loss": 3.2807,
      "step": 428160
    },
    {
      "epoch": 0.10751554292424519,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001143776377637764,
      "loss": 3.2029,
      "step": 428170
    },
    {
      "epoch": 0.10753418616040337,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011437563756375638,
      "loss": 3.1176,
      "step": 428180
    },
    {
      "epoch": 0.10755282939656156,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011437363736373638,
      "loss": 3.1642,
      "step": 428190
    },
    {
      "epoch": 0.10757147263271974,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011437163716371637,
      "loss": 3.1502,
      "step": 428200
    },
    {
      "epoch": 0.10759011586887793,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011436963696369638,
      "loss": 3.0199,
      "step": 428210
    },
    {
      "epoch": 0.1076087591050361,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011436763676367637,
      "loss": 3.2418,
      "step": 428220
    },
    {
      "epoch": 0.10762740234119429,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011436563656365638,
      "loss": 3.1543,
      "step": 428230
    },
    {
      "epoch": 0.10764604557735248,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011436363636363636,
      "loss": 3.1808,
      "step": 428240
    },
    {
      "epoch": 0.10766468881351066,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011436163616361637,
      "loss": 3.0714,
      "step": 428250
    },
    {
      "epoch": 0.10768333204966885,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011435963596359635,
      "loss": 3.037,
      "step": 428260
    },
    {
      "epoch": 0.10770197528582703,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011435763576357636,
      "loss": 3.1722,
      "step": 428270
    },
    {
      "epoch": 0.10772061852198521,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011435563556355635,
      "loss": 3.2196,
      "step": 428280
    },
    {
      "epoch": 0.1077392617581434,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011435363536353635,
      "loss": 3.252,
      "step": 428290
    },
    {
      "epoch": 0.10775790499430157,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011435163516351635,
      "loss": 3.2555,
      "step": 428300
    },
    {
      "epoch": 0.10777654823045976,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011434963496349636,
      "loss": 3.1148,
      "step": 428310
    },
    {
      "epoch": 0.10779519146661795,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011434763476347637,
      "loss": 3.1174,
      "step": 428320
    },
    {
      "epoch": 0.10781383470277613,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011434563456345635,
      "loss": 3.1539,
      "step": 428330
    },
    {
      "epoch": 0.10783247793893432,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011434363436343636,
      "loss": 3.1382,
      "step": 428340
    },
    {
      "epoch": 0.10785112117509249,
      "grad_norm": 0.75,
      "learning_rate": 0.00011434163416341634,
      "loss": 3.0175,
      "step": 428350
    },
    {
      "epoch": 0.10786976441125068,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011433963396339635,
      "loss": 3.1984,
      "step": 428360
    },
    {
      "epoch": 0.10788840764740887,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011433763376337635,
      "loss": 3.0786,
      "step": 428370
    },
    {
      "epoch": 0.10790705088356704,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011433563356335634,
      "loss": 3.1469,
      "step": 428380
    },
    {
      "epoch": 0.10792569411972523,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011433363336333634,
      "loss": 3.1276,
      "step": 428390
    },
    {
      "epoch": 0.10794433735588342,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011433163316331635,
      "loss": 3.1801,
      "step": 428400
    },
    {
      "epoch": 0.1079629805920416,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011432963296329633,
      "loss": 3.188,
      "step": 428410
    },
    {
      "epoch": 0.10798162382819979,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011432763276327634,
      "loss": 3.1101,
      "step": 428420
    },
    {
      "epoch": 0.10800026706435796,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011432563256325632,
      "loss": 3.0454,
      "step": 428430
    },
    {
      "epoch": 0.10801891030051615,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011432363236323633,
      "loss": 3.1593,
      "step": 428440
    },
    {
      "epoch": 0.10803755353667434,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011432163216321632,
      "loss": 3.1537,
      "step": 428450
    },
    {
      "epoch": 0.10805619677283251,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011431963196319633,
      "loss": 3.1005,
      "step": 428460
    },
    {
      "epoch": 0.1080748400089907,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011431763176317632,
      "loss": 3.252,
      "step": 428470
    },
    {
      "epoch": 0.10809348324514888,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011431563156315632,
      "loss": 3.2651,
      "step": 428480
    },
    {
      "epoch": 0.10811212648130707,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011431363136313631,
      "loss": 3.1716,
      "step": 428490
    },
    {
      "epoch": 0.10813076971746526,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011431163116311632,
      "loss": 3.3068,
      "step": 428500
    },
    {
      "epoch": 0.10814941295362343,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011430963096309633,
      "loss": 3.2001,
      "step": 428510
    },
    {
      "epoch": 0.10816805618978162,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011430763076307631,
      "loss": 3.1239,
      "step": 428520
    },
    {
      "epoch": 0.10818669942593981,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011430563056305632,
      "loss": 3.028,
      "step": 428530
    },
    {
      "epoch": 0.10820534266209798,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011430363036303631,
      "loss": 3.0754,
      "step": 428540
    },
    {
      "epoch": 0.10822398589825617,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011430163016301632,
      "loss": 3.101,
      "step": 428550
    },
    {
      "epoch": 0.10824262913441435,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001142996299629963,
      "loss": 3.181,
      "step": 428560
    },
    {
      "epoch": 0.10826127237057254,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011429762976297631,
      "loss": 3.254,
      "step": 428570
    },
    {
      "epoch": 0.10827991560673073,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001142956295629563,
      "loss": 3.1007,
      "step": 428580
    },
    {
      "epoch": 0.1082985588428889,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001142936293629363,
      "loss": 3.2498,
      "step": 428590
    },
    {
      "epoch": 0.10831720207904709,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011429162916291629,
      "loss": 3.2834,
      "step": 428600
    },
    {
      "epoch": 0.10833584531520528,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001142896289628963,
      "loss": 3.0367,
      "step": 428610
    },
    {
      "epoch": 0.10835448855136345,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011428762876287629,
      "loss": 3.1027,
      "step": 428620
    },
    {
      "epoch": 0.10837313178752164,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001142856285628563,
      "loss": 3.3166,
      "step": 428630
    },
    {
      "epoch": 0.10839177502367982,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011428362836283628,
      "loss": 3.2058,
      "step": 428640
    },
    {
      "epoch": 0.108410418259838,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011428162816281629,
      "loss": 3.0478,
      "step": 428650
    },
    {
      "epoch": 0.1084290614959962,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011427962796279627,
      "loss": 3.115,
      "step": 428660
    },
    {
      "epoch": 0.10844770473215437,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011427762776277628,
      "loss": 3.0945,
      "step": 428670
    },
    {
      "epoch": 0.10846634796831256,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011427562756275628,
      "loss": 3.1302,
      "step": 428680
    },
    {
      "epoch": 0.10848499120447073,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011427362736273629,
      "loss": 3.1234,
      "step": 428690
    },
    {
      "epoch": 0.10850363444062892,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011427162716271627,
      "loss": 3.0528,
      "step": 428700
    },
    {
      "epoch": 0.10852227767678711,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011426962696269628,
      "loss": 3.0618,
      "step": 428710
    },
    {
      "epoch": 0.10854092091294529,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011426762676267629,
      "loss": 3.1349,
      "step": 428720
    },
    {
      "epoch": 0.10855956414910348,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011426562656265627,
      "loss": 3.2036,
      "step": 428730
    },
    {
      "epoch": 0.10857820738526167,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011426362636263628,
      "loss": 3.0784,
      "step": 428740
    },
    {
      "epoch": 0.10859685062141984,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011426162616261626,
      "loss": 3.1631,
      "step": 428750
    },
    {
      "epoch": 0.10861549385757803,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011425962596259627,
      "loss": 3.2239,
      "step": 428760
    },
    {
      "epoch": 0.1086341370937362,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011425762576257626,
      "loss": 3.1542,
      "step": 428770
    },
    {
      "epoch": 0.1086527803298944,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011425562556255627,
      "loss": 3.1467,
      "step": 428780
    },
    {
      "epoch": 0.10867142356605258,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011425362536253626,
      "loss": 3.1303,
      "step": 428790
    },
    {
      "epoch": 0.10869006680221076,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011425162516251627,
      "loss": 3.1167,
      "step": 428800
    },
    {
      "epoch": 0.10870871003836895,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011424962496249625,
      "loss": 3.196,
      "step": 428810
    },
    {
      "epoch": 0.10872735327452712,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011424762476247626,
      "loss": 3.1847,
      "step": 428820
    },
    {
      "epoch": 0.10874599651068531,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011424562456245624,
      "loss": 3.0287,
      "step": 428830
    },
    {
      "epoch": 0.1087646397468435,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011424362436243625,
      "loss": 3.2652,
      "step": 428840
    },
    {
      "epoch": 0.10878328298300168,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011424162416241624,
      "loss": 3.305,
      "step": 428850
    },
    {
      "epoch": 0.10880192621915986,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011423962396239625,
      "loss": 3.1295,
      "step": 428860
    },
    {
      "epoch": 0.10882056945531805,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011423762376237623,
      "loss": 3.1966,
      "step": 428870
    },
    {
      "epoch": 0.10883921269147623,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011423562356235624,
      "loss": 3.1218,
      "step": 428880
    },
    {
      "epoch": 0.10885785592763442,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011423362336233623,
      "loss": 3.1631,
      "step": 428890
    },
    {
      "epoch": 0.10887649916379259,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011423162316231623,
      "loss": 3.1444,
      "step": 428900
    },
    {
      "epoch": 0.10889514239995078,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011422962296229624,
      "loss": 3.2147,
      "step": 428910
    },
    {
      "epoch": 0.10891378563610897,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011422762276227623,
      "loss": 3.1551,
      "step": 428920
    },
    {
      "epoch": 0.10893242887226715,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011422562256225623,
      "loss": 3.1689,
      "step": 428930
    },
    {
      "epoch": 0.10895107210842533,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011422362236223623,
      "loss": 3.1285,
      "step": 428940
    },
    {
      "epoch": 0.10896971534458352,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011422162216221624,
      "loss": 3.1954,
      "step": 428950
    },
    {
      "epoch": 0.1089883585807417,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011421962196219622,
      "loss": 3.104,
      "step": 428960
    },
    {
      "epoch": 0.10900700181689989,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011421762176217623,
      "loss": 3.1366,
      "step": 428970
    },
    {
      "epoch": 0.10902564505305806,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011421562156215621,
      "loss": 3.2776,
      "step": 428980
    },
    {
      "epoch": 0.10904428828921625,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011421362136213622,
      "loss": 3.2362,
      "step": 428990
    },
    {
      "epoch": 0.10906293152537444,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011421162116211622,
      "loss": 3.2043,
      "step": 429000
    },
    {
      "epoch": 0.10908157476153262,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011420962096209621,
      "loss": 3.1173,
      "step": 429010
    },
    {
      "epoch": 0.1091002179976908,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011420762076207621,
      "loss": 3.1843,
      "step": 429020
    },
    {
      "epoch": 0.10911886123384898,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011420562056205622,
      "loss": 3.4103,
      "step": 429030
    },
    {
      "epoch": 0.10913750447000717,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001142036203620362,
      "loss": 3.1634,
      "step": 429040
    },
    {
      "epoch": 0.10915614770616536,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011420162016201621,
      "loss": 3.03,
      "step": 429050
    },
    {
      "epoch": 0.10917479094232353,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011419961996199619,
      "loss": 3.0065,
      "step": 429060
    },
    {
      "epoch": 0.10919343417848172,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001141976197619762,
      "loss": 3.1045,
      "step": 429070
    },
    {
      "epoch": 0.10921207741463991,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001141956195619562,
      "loss": 3.0922,
      "step": 429080
    },
    {
      "epoch": 0.10923072065079809,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001141936193619362,
      "loss": 3.1738,
      "step": 429090
    },
    {
      "epoch": 0.10924936388695627,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011419161916191621,
      "loss": 3.2454,
      "step": 429100
    },
    {
      "epoch": 0.10926800712311445,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001141896189618962,
      "loss": 3.0587,
      "step": 429110
    },
    {
      "epoch": 0.10928665035927264,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001141876187618762,
      "loss": 3.1139,
      "step": 429120
    },
    {
      "epoch": 0.10930529359543083,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011418561856185619,
      "loss": 3.143,
      "step": 429130
    },
    {
      "epoch": 0.109323936831589,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001141836183618362,
      "loss": 3.1917,
      "step": 429140
    },
    {
      "epoch": 0.10934258006774719,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011418161816181618,
      "loss": 3.1551,
      "step": 429150
    },
    {
      "epoch": 0.10936122330390537,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011417961796179619,
      "loss": 3.1408,
      "step": 429160
    },
    {
      "epoch": 0.10937986654006356,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011417761776177618,
      "loss": 2.9628,
      "step": 429170
    },
    {
      "epoch": 0.10939850977622174,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011417561756175619,
      "loss": 2.9896,
      "step": 429180
    },
    {
      "epoch": 0.10941715301237992,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011417361736173617,
      "loss": 3.139,
      "step": 429190
    },
    {
      "epoch": 0.10943579624853811,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011417161716171618,
      "loss": 3.2543,
      "step": 429200
    },
    {
      "epoch": 0.1094544394846963,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011416961696169617,
      "loss": 3.0999,
      "step": 429210
    },
    {
      "epoch": 0.10947308272085447,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011416761676167617,
      "loss": 3.0692,
      "step": 429220
    },
    {
      "epoch": 0.10949172595701266,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011416561656165616,
      "loss": 3.1072,
      "step": 429230
    },
    {
      "epoch": 0.10951036919317084,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011416361636163617,
      "loss": 3.1758,
      "step": 429240
    },
    {
      "epoch": 0.10952901242932903,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011416161616161616,
      "loss": 3.1166,
      "step": 429250
    },
    {
      "epoch": 0.10954765566548721,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011415961596159617,
      "loss": 3.1017,
      "step": 429260
    },
    {
      "epoch": 0.10956629890164539,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011415761576157615,
      "loss": 3.2291,
      "step": 429270
    },
    {
      "epoch": 0.10958494213780358,
      "grad_norm": 0.625,
      "learning_rate": 0.00011415561556155616,
      "loss": 3.1556,
      "step": 429280
    },
    {
      "epoch": 0.10960358537396177,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011415361536153614,
      "loss": 3.2239,
      "step": 429290
    },
    {
      "epoch": 0.10962222861011994,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011415161516151615,
      "loss": 3.1215,
      "step": 429300
    },
    {
      "epoch": 0.10964087184627813,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011414961496149616,
      "loss": 3.1757,
      "step": 429310
    },
    {
      "epoch": 0.1096595150824363,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011414761476147616,
      "loss": 3.1186,
      "step": 429320
    },
    {
      "epoch": 0.1096781583185945,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011414561456145615,
      "loss": 3.1149,
      "step": 429330
    },
    {
      "epoch": 0.10969680155475268,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011414361436143615,
      "loss": 3.1411,
      "step": 429340
    },
    {
      "epoch": 0.10971544479091086,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011414161416141616,
      "loss": 3.2122,
      "step": 429350
    },
    {
      "epoch": 0.10973408802706905,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011413961396139614,
      "loss": 3.1922,
      "step": 429360
    },
    {
      "epoch": 0.10975273126322722,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011413761376137615,
      "loss": 3.1504,
      "step": 429370
    },
    {
      "epoch": 0.10977137449938541,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011413561356135613,
      "loss": 3.194,
      "step": 429380
    },
    {
      "epoch": 0.1097900177355436,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011413361336133614,
      "loss": 3.1011,
      "step": 429390
    },
    {
      "epoch": 0.10980866097170178,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011413161316131614,
      "loss": 3.1897,
      "step": 429400
    },
    {
      "epoch": 0.10982730420785997,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011412961296129615,
      "loss": 3.0746,
      "step": 429410
    },
    {
      "epoch": 0.10984594744401815,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011412761276127613,
      "loss": 3.2149,
      "step": 429420
    },
    {
      "epoch": 0.10986459068017633,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011412561256125614,
      "loss": 3.0598,
      "step": 429430
    },
    {
      "epoch": 0.10988323391633452,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011412361236123612,
      "loss": 3.0798,
      "step": 429440
    },
    {
      "epoch": 0.1099018771524927,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011412161216121613,
      "loss": 3.0409,
      "step": 429450
    },
    {
      "epoch": 0.10992052038865088,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011411961196119611,
      "loss": 3.1452,
      "step": 429460
    },
    {
      "epoch": 0.10993916362480907,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011411761176117612,
      "loss": 2.9871,
      "step": 429470
    },
    {
      "epoch": 0.10995780686096725,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011411561156115611,
      "loss": 3.0853,
      "step": 429480
    },
    {
      "epoch": 0.10997645009712544,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00011411361136113612,
      "loss": 3.0761,
      "step": 429490
    },
    {
      "epoch": 0.10999509333328361,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011411161116111613,
      "loss": 3.1663,
      "step": 429500
    },
    {
      "epoch": 0.1100137365694418,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011410961096109611,
      "loss": 3.1893,
      "step": 429510
    },
    {
      "epoch": 0.11003237980559999,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011410761076107612,
      "loss": 3.2155,
      "step": 429520
    },
    {
      "epoch": 0.11005102304175816,
      "grad_norm": 0.625,
      "learning_rate": 0.0001141056105610561,
      "loss": 3.1781,
      "step": 429530
    },
    {
      "epoch": 0.11006966627791635,
      "grad_norm": 0.75,
      "learning_rate": 0.00011410361036103612,
      "loss": 3.1639,
      "step": 429540
    },
    {
      "epoch": 0.11008830951407454,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001141016101610161,
      "loss": 3.1719,
      "step": 429550
    },
    {
      "epoch": 0.11010695275023272,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001140996099609961,
      "loss": 3.0519,
      "step": 429560
    },
    {
      "epoch": 0.1101255959863909,
      "grad_norm": 0.875,
      "learning_rate": 0.0001140976097609761,
      "loss": 3.1259,
      "step": 429570
    },
    {
      "epoch": 0.11014423922254908,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011409560956095611,
      "loss": 3.1993,
      "step": 429580
    },
    {
      "epoch": 0.11016288245870727,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011409360936093609,
      "loss": 3.0032,
      "step": 429590
    },
    {
      "epoch": 0.11018152569486546,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001140916091609161,
      "loss": 3.0452,
      "step": 429600
    },
    {
      "epoch": 0.11020016893102363,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011408960896089608,
      "loss": 3.2185,
      "step": 429610
    },
    {
      "epoch": 0.11021881216718182,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001140876087608761,
      "loss": 3.1763,
      "step": 429620
    },
    {
      "epoch": 0.11023745540334001,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011408560856085609,
      "loss": 3.1348,
      "step": 429630
    },
    {
      "epoch": 0.11025609863949819,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001140836083608361,
      "loss": 3.069,
      "step": 429640
    },
    {
      "epoch": 0.11027474187565638,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011408160816081608,
      "loss": 3.0152,
      "step": 429650
    },
    {
      "epoch": 0.11029338511181455,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011407960796079609,
      "loss": 3.1508,
      "step": 429660
    },
    {
      "epoch": 0.11031202834797274,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011407760776077607,
      "loss": 2.9639,
      "step": 429670
    },
    {
      "epoch": 0.11033067158413093,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011407560756075608,
      "loss": 3.0861,
      "step": 429680
    },
    {
      "epoch": 0.1103493148202891,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011407360736073609,
      "loss": 3.0964,
      "step": 429690
    },
    {
      "epoch": 0.1103679580564473,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011407160716071607,
      "loss": 3.1491,
      "step": 429700
    },
    {
      "epoch": 0.11038660129260547,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011406960696069608,
      "loss": 3.0315,
      "step": 429710
    },
    {
      "epoch": 0.11040524452876366,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011406760676067608,
      "loss": 3.2498,
      "step": 429720
    },
    {
      "epoch": 0.11042388776492185,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011406560656065609,
      "loss": 3.171,
      "step": 429730
    },
    {
      "epoch": 0.11044253100108002,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011406360636063607,
      "loss": 3.2029,
      "step": 429740
    },
    {
      "epoch": 0.11046117423723821,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011406160616061608,
      "loss": 3.1253,
      "step": 429750
    },
    {
      "epoch": 0.1104798174733964,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011405960596059606,
      "loss": 3.1739,
      "step": 429760
    },
    {
      "epoch": 0.11049846070955457,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011405760576057607,
      "loss": 3.1343,
      "step": 429770
    },
    {
      "epoch": 0.11051710394571276,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011405560556055605,
      "loss": 3.163,
      "step": 429780
    },
    {
      "epoch": 0.11053574718187094,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011405360536053606,
      "loss": 3.1097,
      "step": 429790
    },
    {
      "epoch": 0.11055439041802913,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011405160516051605,
      "loss": 3.2345,
      "step": 429800
    },
    {
      "epoch": 0.11057303365418732,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011404960496049606,
      "loss": 3.1348,
      "step": 429810
    },
    {
      "epoch": 0.11059167689034549,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011404760476047605,
      "loss": 3.1841,
      "step": 429820
    },
    {
      "epoch": 0.11061032012650368,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011404560456045606,
      "loss": 3.1544,
      "step": 429830
    },
    {
      "epoch": 0.11062896336266186,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011404360436043604,
      "loss": 3.1215,
      "step": 429840
    },
    {
      "epoch": 0.11064760659882004,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011404160416041605,
      "loss": 3.2537,
      "step": 429850
    },
    {
      "epoch": 0.11066624983497823,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011403960396039604,
      "loss": 3.2142,
      "step": 429860
    },
    {
      "epoch": 0.11068489307113641,
      "grad_norm": 0.75,
      "learning_rate": 0.00011403760376037604,
      "loss": 3.1158,
      "step": 429870
    },
    {
      "epoch": 0.1107035363072946,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011403560356035603,
      "loss": 3.2365,
      "step": 429880
    },
    {
      "epoch": 0.11072217954345279,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011403360336033604,
      "loss": 3.2207,
      "step": 429890
    },
    {
      "epoch": 0.11074082277961096,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011403160316031605,
      "loss": 3.1174,
      "step": 429900
    },
    {
      "epoch": 0.11075946601576915,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011402960296029603,
      "loss": 3.1587,
      "step": 429910
    },
    {
      "epoch": 0.11077810925192733,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011402760276027604,
      "loss": 3.2669,
      "step": 429920
    },
    {
      "epoch": 0.11079675248808551,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011402560256025602,
      "loss": 2.9424,
      "step": 429930
    },
    {
      "epoch": 0.1108153957242437,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011402360236023603,
      "loss": 3.1559,
      "step": 429940
    },
    {
      "epoch": 0.11083403896040188,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011402160216021603,
      "loss": 3.085,
      "step": 429950
    },
    {
      "epoch": 0.11085268219656007,
      "grad_norm": 0.75,
      "learning_rate": 0.00011401960196019604,
      "loss": 3.1348,
      "step": 429960
    },
    {
      "epoch": 0.11087132543271826,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011401760176017602,
      "loss": 3.0948,
      "step": 429970
    },
    {
      "epoch": 0.11088996866887643,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011401560156015603,
      "loss": 3.1437,
      "step": 429980
    },
    {
      "epoch": 0.11090861190503462,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011401360136013601,
      "loss": 2.98,
      "step": 429990
    },
    {
      "epoch": 0.1109272551411928,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011401160116011602,
      "loss": 3.1118,
      "step": 430000
    },
    {
      "epoch": 0.11094589837735098,
      "grad_norm": 0.796875,
      "learning_rate": 0.000114009600960096,
      "loss": 3.2204,
      "step": 430010
    },
    {
      "epoch": 0.11096454161350917,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011400760076007601,
      "loss": 3.113,
      "step": 430020
    },
    {
      "epoch": 0.11098318484966735,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011400560056005601,
      "loss": 3.0983,
      "step": 430030
    },
    {
      "epoch": 0.11100182808582554,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011400360036003602,
      "loss": 3.1426,
      "step": 430040
    },
    {
      "epoch": 0.11102047132198371,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000114001600160016,
      "loss": 3.0923,
      "step": 430050
    },
    {
      "epoch": 0.1110391145581419,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011399959995999601,
      "loss": 3.1973,
      "step": 430060
    },
    {
      "epoch": 0.11105775779430009,
      "grad_norm": 0.875,
      "learning_rate": 0.00011399759975997599,
      "loss": 3.1062,
      "step": 430070
    },
    {
      "epoch": 0.11107640103045827,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000113995599559956,
      "loss": 3.2243,
      "step": 430080
    },
    {
      "epoch": 0.11109504426661646,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011399359935993601,
      "loss": 3.0781,
      "step": 430090
    },
    {
      "epoch": 0.11111368750277464,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011399159915991599,
      "loss": 3.1242,
      "step": 430100
    },
    {
      "epoch": 0.11113233073893282,
      "grad_norm": 0.796875,
      "learning_rate": 0.000113989598959896,
      "loss": 3.0756,
      "step": 430110
    },
    {
      "epoch": 0.11115097397509101,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000113987598759876,
      "loss": 3.2857,
      "step": 430120
    },
    {
      "epoch": 0.11116961721124918,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000113985598559856,
      "loss": 3.2561,
      "step": 430130
    },
    {
      "epoch": 0.11118826044740737,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011398359835983599,
      "loss": 3.4268,
      "step": 430140
    },
    {
      "epoch": 0.11120690368356556,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000113981598159816,
      "loss": 3.2247,
      "step": 430150
    },
    {
      "epoch": 0.11122554691972374,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011397959795979598,
      "loss": 3.0624,
      "step": 430160
    },
    {
      "epoch": 0.11124419015588193,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011397759775977599,
      "loss": 3.2026,
      "step": 430170
    },
    {
      "epoch": 0.1112628333920401,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011397559755975597,
      "loss": 3.1612,
      "step": 430180
    },
    {
      "epoch": 0.11128147662819829,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011397359735973598,
      "loss": 3.1986,
      "step": 430190
    },
    {
      "epoch": 0.11130011986435648,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011397159715971597,
      "loss": 3.0277,
      "step": 430200
    },
    {
      "epoch": 0.11131876310051465,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011396959695969598,
      "loss": 3.162,
      "step": 430210
    },
    {
      "epoch": 0.11133740633667284,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011396759675967596,
      "loss": 3.2403,
      "step": 430220
    },
    {
      "epoch": 0.11135604957283103,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011396559655965597,
      "loss": 3.1051,
      "step": 430230
    },
    {
      "epoch": 0.1113746928089892,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011396359635963596,
      "loss": 3.1213,
      "step": 430240
    },
    {
      "epoch": 0.1113933360451474,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011396159615961597,
      "loss": 3.1767,
      "step": 430250
    },
    {
      "epoch": 0.11141197928130557,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011395959595959596,
      "loss": 3.0924,
      "step": 430260
    },
    {
      "epoch": 0.11143062251746376,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011395759575957597,
      "loss": 3.0218,
      "step": 430270
    },
    {
      "epoch": 0.11144926575362195,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011395559555955597,
      "loss": 3.2893,
      "step": 430280
    },
    {
      "epoch": 0.11146790898978012,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011395359535953596,
      "loss": 3.1741,
      "step": 430290
    },
    {
      "epoch": 0.11148655222593831,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011395159515951597,
      "loss": 3.1508,
      "step": 430300
    },
    {
      "epoch": 0.1115051954620965,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011394959495949595,
      "loss": 3.1862,
      "step": 430310
    },
    {
      "epoch": 0.11152383869825468,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011394759475947596,
      "loss": 3.1593,
      "step": 430320
    },
    {
      "epoch": 0.11154248193441287,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011394559455945594,
      "loss": 3.0684,
      "step": 430330
    },
    {
      "epoch": 0.11156112517057104,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011394359435943595,
      "loss": 3.0144,
      "step": 430340
    },
    {
      "epoch": 0.11157976840672923,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011394159415941595,
      "loss": 3.1961,
      "step": 430350
    },
    {
      "epoch": 0.11159841164288742,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011393959395939596,
      "loss": 3.1041,
      "step": 430360
    },
    {
      "epoch": 0.1116170548790456,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011393759375937594,
      "loss": 3.2552,
      "step": 430370
    },
    {
      "epoch": 0.11163569811520378,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011393559355935595,
      "loss": 3.1182,
      "step": 430380
    },
    {
      "epoch": 0.11165434135136196,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011393359335933593,
      "loss": 3.1319,
      "step": 430390
    },
    {
      "epoch": 0.11167298458752015,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011393159315931594,
      "loss": 3.1069,
      "step": 430400
    },
    {
      "epoch": 0.11169162782367834,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011392959295929592,
      "loss": 3.0461,
      "step": 430410
    },
    {
      "epoch": 0.11171027105983651,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011392759275927593,
      "loss": 3.1612,
      "step": 430420
    },
    {
      "epoch": 0.1117289142959947,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011392559255925593,
      "loss": 3.1589,
      "step": 430430
    },
    {
      "epoch": 0.11174755753215289,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011392359235923594,
      "loss": 3.1294,
      "step": 430440
    },
    {
      "epoch": 0.11176620076831106,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011392159215921592,
      "loss": 3.1582,
      "step": 430450
    },
    {
      "epoch": 0.11178484400446925,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011391959195919593,
      "loss": 3.1597,
      "step": 430460
    },
    {
      "epoch": 0.11180348724062743,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011391759175917591,
      "loss": 3.0976,
      "step": 430470
    },
    {
      "epoch": 0.11182213047678562,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011391559155915592,
      "loss": 3.1549,
      "step": 430480
    },
    {
      "epoch": 0.1118407737129438,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011391359135913593,
      "loss": 3.0537,
      "step": 430490
    },
    {
      "epoch": 0.11185941694910198,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011391159115911591,
      "loss": 3.1506,
      "step": 430500
    },
    {
      "epoch": 0.11187806018526017,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011390959095909592,
      "loss": 3.1116,
      "step": 430510
    },
    {
      "epoch": 0.11189670342141836,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011390759075907591,
      "loss": 3.1736,
      "step": 430520
    },
    {
      "epoch": 0.11191534665757653,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011390559055905592,
      "loss": 3.1312,
      "step": 430530
    },
    {
      "epoch": 0.11193398989373472,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001139035903590359,
      "loss": 3.1903,
      "step": 430540
    },
    {
      "epoch": 0.1119526331298929,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011390159015901591,
      "loss": 3.0756,
      "step": 430550
    },
    {
      "epoch": 0.11197127636605109,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001138995899589959,
      "loss": 3.2162,
      "step": 430560
    },
    {
      "epoch": 0.11198991960220928,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001138975897589759,
      "loss": 3.304,
      "step": 430570
    },
    {
      "epoch": 0.11200856283836745,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001138955895589559,
      "loss": 3.2098,
      "step": 430580
    },
    {
      "epoch": 0.11202720607452564,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011389358935893591,
      "loss": 3.1306,
      "step": 430590
    },
    {
      "epoch": 0.11204584931068381,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011389158915891589,
      "loss": 3.0531,
      "step": 430600
    },
    {
      "epoch": 0.112064492546842,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001138895889588959,
      "loss": 3.1079,
      "step": 430610
    },
    {
      "epoch": 0.11208313578300019,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011388758875887588,
      "loss": 3.0705,
      "step": 430620
    },
    {
      "epoch": 0.11210177901915837,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011388558855885589,
      "loss": 3.0586,
      "step": 430630
    },
    {
      "epoch": 0.11212042225531656,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011388358835883587,
      "loss": 3.1226,
      "step": 430640
    },
    {
      "epoch": 0.11213906549147475,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011388158815881588,
      "loss": 3.3136,
      "step": 430650
    },
    {
      "epoch": 0.11215770872763292,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011387958795879588,
      "loss": 3.1587,
      "step": 430660
    },
    {
      "epoch": 0.11217635196379111,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011387758775877589,
      "loss": 3.2044,
      "step": 430670
    },
    {
      "epoch": 0.11219499519994929,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001138755875587559,
      "loss": 3.1954,
      "step": 430680
    },
    {
      "epoch": 0.11221363843610747,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011387358735873588,
      "loss": 3.2429,
      "step": 430690
    },
    {
      "epoch": 0.11223228167226566,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011387158715871589,
      "loss": 3.1939,
      "step": 430700
    },
    {
      "epoch": 0.11225092490842384,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011386958695869587,
      "loss": 3.2094,
      "step": 430710
    },
    {
      "epoch": 0.11226956814458203,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011386758675867588,
      "loss": 3.006,
      "step": 430720
    },
    {
      "epoch": 0.1122882113807402,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011386558655865586,
      "loss": 3.1588,
      "step": 430730
    },
    {
      "epoch": 0.11230685461689839,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011386358635863587,
      "loss": 3.0855,
      "step": 430740
    },
    {
      "epoch": 0.11232549785305658,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011386158615861587,
      "loss": 3.1761,
      "step": 430750
    },
    {
      "epoch": 0.11234414108921476,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011385958595859588,
      "loss": 3.06,
      "step": 430760
    },
    {
      "epoch": 0.11236278432537294,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011385758575857586,
      "loss": 3.1443,
      "step": 430770
    },
    {
      "epoch": 0.11238142756153113,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011385558555855587,
      "loss": 3.0674,
      "step": 430780
    },
    {
      "epoch": 0.11240007079768931,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011385358535853585,
      "loss": 3.294,
      "step": 430790
    },
    {
      "epoch": 0.1124187140338475,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011385158515851586,
      "loss": 3.1952,
      "step": 430800
    },
    {
      "epoch": 0.11243735727000567,
      "grad_norm": 0.75,
      "learning_rate": 0.00011384958495849585,
      "loss": 3.1143,
      "step": 430810
    },
    {
      "epoch": 0.11245600050616386,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011384758475847585,
      "loss": 3.1006,
      "step": 430820
    },
    {
      "epoch": 0.11247464374232205,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011384558455845585,
      "loss": 3.1573,
      "step": 430830
    },
    {
      "epoch": 0.11249328697848023,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011384358435843585,
      "loss": 3.2957,
      "step": 430840
    },
    {
      "epoch": 0.11251193021463841,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011384158415841584,
      "loss": 3.0863,
      "step": 430850
    },
    {
      "epoch": 0.1125305734507966,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011383958395839585,
      "loss": 3.2339,
      "step": 430860
    },
    {
      "epoch": 0.11254921668695478,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011383758375837583,
      "loss": 3.241,
      "step": 430870
    },
    {
      "epoch": 0.11256785992311297,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011383558355835584,
      "loss": 3.1055,
      "step": 430880
    },
    {
      "epoch": 0.11258650315927114,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011383358335833585,
      "loss": 3.1549,
      "step": 430890
    },
    {
      "epoch": 0.11260514639542933,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011383158315831584,
      "loss": 3.2822,
      "step": 430900
    },
    {
      "epoch": 0.11262378963158752,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011382958295829585,
      "loss": 3.2282,
      "step": 430910
    },
    {
      "epoch": 0.1126424328677457,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011382758275827583,
      "loss": 3.1776,
      "step": 430920
    },
    {
      "epoch": 0.11266107610390388,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011382558255825584,
      "loss": 3.2015,
      "step": 430930
    },
    {
      "epoch": 0.11267971934006206,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011382358235823582,
      "loss": 3.1986,
      "step": 430940
    },
    {
      "epoch": 0.11269836257622025,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011382158215821583,
      "loss": 3.1532,
      "step": 430950
    },
    {
      "epoch": 0.11271700581237844,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011381958195819581,
      "loss": 3.0238,
      "step": 430960
    },
    {
      "epoch": 0.11273564904853661,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011381758175817582,
      "loss": 3.1143,
      "step": 430970
    },
    {
      "epoch": 0.1127542922846948,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011381558155815582,
      "loss": 3.0176,
      "step": 430980
    },
    {
      "epoch": 0.11277293552085299,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011381358135813583,
      "loss": 3.1229,
      "step": 430990
    },
    {
      "epoch": 0.11279157875701117,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011381158115811581,
      "loss": 3.279,
      "step": 431000
    },
    {
      "epoch": 0.11281022199316935,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011380958095809582,
      "loss": 3.1664,
      "step": 431010
    },
    {
      "epoch": 0.11282886522932753,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001138075807580758,
      "loss": 3.2351,
      "step": 431020
    },
    {
      "epoch": 0.11284750846548572,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011380558055805581,
      "loss": 3.2067,
      "step": 431030
    },
    {
      "epoch": 0.11286615170164391,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011380358035803579,
      "loss": 3.1086,
      "step": 431040
    },
    {
      "epoch": 0.11288479493780208,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001138015801580158,
      "loss": 3.2345,
      "step": 431050
    },
    {
      "epoch": 0.11290343817396027,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001137995799579958,
      "loss": 3.0722,
      "step": 431060
    },
    {
      "epoch": 0.11292208141011845,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011379757975797581,
      "loss": 3.1678,
      "step": 431070
    },
    {
      "epoch": 0.11294072464627664,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011379557955795582,
      "loss": 3.0824,
      "step": 431080
    },
    {
      "epoch": 0.11295936788243482,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001137935793579358,
      "loss": 3.2555,
      "step": 431090
    },
    {
      "epoch": 0.112978011118593,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011379157915791581,
      "loss": 3.1037,
      "step": 431100
    },
    {
      "epoch": 0.11299665435475119,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011378957895789579,
      "loss": 3.1274,
      "step": 431110
    },
    {
      "epoch": 0.11301529759090938,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001137875787578758,
      "loss": 3.27,
      "step": 431120
    },
    {
      "epoch": 0.11303394082706755,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001137855785578558,
      "loss": 3.12,
      "step": 431130
    },
    {
      "epoch": 0.11305258406322574,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011378357835783579,
      "loss": 3.1291,
      "step": 431140
    },
    {
      "epoch": 0.11307122729938392,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011378157815781579,
      "loss": 3.0124,
      "step": 431150
    },
    {
      "epoch": 0.1130898705355421,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001137795779577958,
      "loss": 3.1049,
      "step": 431160
    },
    {
      "epoch": 0.1131085137717003,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011377757775777578,
      "loss": 3.1569,
      "step": 431170
    },
    {
      "epoch": 0.11312715700785847,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011377557755775579,
      "loss": 3.1088,
      "step": 431180
    },
    {
      "epoch": 0.11314580024401666,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011377357735773577,
      "loss": 3.0474,
      "step": 431190
    },
    {
      "epoch": 0.11316444348017485,
      "grad_norm": 0.75,
      "learning_rate": 0.00011377157715771578,
      "loss": 3.1471,
      "step": 431200
    },
    {
      "epoch": 0.11318308671633302,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011376957695769577,
      "loss": 3.2127,
      "step": 431210
    },
    {
      "epoch": 0.11320172995249121,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011376757675767578,
      "loss": 3.0415,
      "step": 431220
    },
    {
      "epoch": 0.11322037318864939,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011376557655765576,
      "loss": 3.0933,
      "step": 431230
    },
    {
      "epoch": 0.11323901642480758,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011376357635763577,
      "loss": 3.1954,
      "step": 431240
    },
    {
      "epoch": 0.11325765966096576,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011376157615761575,
      "loss": 3.0548,
      "step": 431250
    },
    {
      "epoch": 0.11327630289712394,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011375957595759576,
      "loss": 3.2973,
      "step": 431260
    },
    {
      "epoch": 0.11329494613328213,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011375757575757577,
      "loss": 3.2098,
      "step": 431270
    },
    {
      "epoch": 0.1133135893694403,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011375557555755576,
      "loss": 3.0194,
      "step": 431280
    },
    {
      "epoch": 0.11333223260559849,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011375357535753576,
      "loss": 3.0715,
      "step": 431290
    },
    {
      "epoch": 0.11335087584175668,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011375157515751576,
      "loss": 3.1925,
      "step": 431300
    },
    {
      "epoch": 0.11336951907791486,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011374957495749577,
      "loss": 3.1411,
      "step": 431310
    },
    {
      "epoch": 0.11338816231407305,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011374757475747575,
      "loss": 3.239,
      "step": 431320
    },
    {
      "epoch": 0.11340680555023123,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011374557455745576,
      "loss": 3.0867,
      "step": 431330
    },
    {
      "epoch": 0.11342544878638941,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011374357435743574,
      "loss": 3.1198,
      "step": 431340
    },
    {
      "epoch": 0.1134440920225476,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011374157415741575,
      "loss": 3.0823,
      "step": 431350
    },
    {
      "epoch": 0.11346273525870577,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011373957395739573,
      "loss": 3.1072,
      "step": 431360
    },
    {
      "epoch": 0.11348137849486396,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011373757375737574,
      "loss": 3.0994,
      "step": 431370
    },
    {
      "epoch": 0.11350002173102215,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011373557355735574,
      "loss": 3.1252,
      "step": 431380
    },
    {
      "epoch": 0.11351866496718033,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011373357335733575,
      "loss": 3.1824,
      "step": 431390
    },
    {
      "epoch": 0.11353730820333852,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011373157315731573,
      "loss": 3.0663,
      "step": 431400
    },
    {
      "epoch": 0.11355595143949669,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011372957295729574,
      "loss": 3.1335,
      "step": 431410
    },
    {
      "epoch": 0.11357459467565488,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011372757275727572,
      "loss": 3.0873,
      "step": 431420
    },
    {
      "epoch": 0.11359323791181307,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011372557255725573,
      "loss": 3.2061,
      "step": 431430
    },
    {
      "epoch": 0.11361188114797124,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011372357235723573,
      "loss": 3.2438,
      "step": 431440
    },
    {
      "epoch": 0.11363052438412943,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011372157215721573,
      "loss": 3.1486,
      "step": 431450
    },
    {
      "epoch": 0.11364916762028762,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011371957195719572,
      "loss": 3.1093,
      "step": 431460
    },
    {
      "epoch": 0.1136678108564458,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011371757175717573,
      "loss": 3.2627,
      "step": 431470
    },
    {
      "epoch": 0.11368645409260399,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011371557155715573,
      "loss": 3.1438,
      "step": 431480
    },
    {
      "epoch": 0.11370509732876216,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011371357135713572,
      "loss": 3.1374,
      "step": 431490
    },
    {
      "epoch": 0.11372374056492035,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011371157115711573,
      "loss": 3.1862,
      "step": 431500
    },
    {
      "epoch": 0.11374238380107854,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011370957095709571,
      "loss": 3.1735,
      "step": 431510
    },
    {
      "epoch": 0.11376102703723671,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011370757075707572,
      "loss": 3.1789,
      "step": 431520
    },
    {
      "epoch": 0.1137796702733949,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011370557055705571,
      "loss": 3.238,
      "step": 431530
    },
    {
      "epoch": 0.11379831350955309,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011370357035703572,
      "loss": 2.9903,
      "step": 431540
    },
    {
      "epoch": 0.11381695674571127,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001137015701570157,
      "loss": 3.0852,
      "step": 431550
    },
    {
      "epoch": 0.11383559998186946,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011369956995699571,
      "loss": 3.1455,
      "step": 431560
    },
    {
      "epoch": 0.11385424321802763,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001136975697569757,
      "loss": 3.1369,
      "step": 431570
    },
    {
      "epoch": 0.11387288645418582,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001136955695569557,
      "loss": 3.1057,
      "step": 431580
    },
    {
      "epoch": 0.11389152969034401,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011369356935693569,
      "loss": 3.0538,
      "step": 431590
    },
    {
      "epoch": 0.11391017292650218,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001136915691569157,
      "loss": 3.1062,
      "step": 431600
    },
    {
      "epoch": 0.11392881616266037,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00011368956895689569,
      "loss": 3.284,
      "step": 431610
    },
    {
      "epoch": 0.11394745939881855,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001136875687568757,
      "loss": 3.179,
      "step": 431620
    },
    {
      "epoch": 0.11396610263497674,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011368556855685568,
      "loss": 3.0764,
      "step": 431630
    },
    {
      "epoch": 0.11398474587113493,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011368356835683569,
      "loss": 3.0478,
      "step": 431640
    },
    {
      "epoch": 0.1140033891072931,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011368156815681567,
      "loss": 3.1343,
      "step": 431650
    },
    {
      "epoch": 0.11402203234345129,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011367956795679568,
      "loss": 3.1122,
      "step": 431660
    },
    {
      "epoch": 0.11404067557960948,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011367756775677569,
      "loss": 3.1398,
      "step": 431670
    },
    {
      "epoch": 0.11405931881576765,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011367556755675567,
      "loss": 3.1058,
      "step": 431680
    },
    {
      "epoch": 0.11407796205192584,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011367356735673568,
      "loss": 3.0898,
      "step": 431690
    },
    {
      "epoch": 0.11409660528808402,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011367156715671568,
      "loss": 3.0986,
      "step": 431700
    },
    {
      "epoch": 0.11411524852424221,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011366956695669569,
      "loss": 3.1627,
      "step": 431710
    },
    {
      "epoch": 0.1141338917604004,
      "grad_norm": 0.75,
      "learning_rate": 0.00011366756675667567,
      "loss": 3.1608,
      "step": 431720
    },
    {
      "epoch": 0.11415253499655857,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011366556655665568,
      "loss": 3.072,
      "step": 431730
    },
    {
      "epoch": 0.11417117823271676,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011366356635663566,
      "loss": 3.1478,
      "step": 431740
    },
    {
      "epoch": 0.11418982146887494,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011366156615661567,
      "loss": 3.2873,
      "step": 431750
    },
    {
      "epoch": 0.11420846470503312,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011365956595659567,
      "loss": 3.1613,
      "step": 431760
    },
    {
      "epoch": 0.11422710794119131,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011365756575657566,
      "loss": 3.2915,
      "step": 431770
    },
    {
      "epoch": 0.11424575117734949,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011365556555655566,
      "loss": 3.0561,
      "step": 431780
    },
    {
      "epoch": 0.11426439441350768,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011365356535653567,
      "loss": 3.2115,
      "step": 431790
    },
    {
      "epoch": 0.11428303764966587,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011365156515651565,
      "loss": 3.1891,
      "step": 431800
    },
    {
      "epoch": 0.11430168088582404,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011364956495649566,
      "loss": 3.2114,
      "step": 431810
    },
    {
      "epoch": 0.11432032412198223,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011364756475647564,
      "loss": 2.985,
      "step": 431820
    },
    {
      "epoch": 0.1143389673581404,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011364556455645565,
      "loss": 3.1012,
      "step": 431830
    },
    {
      "epoch": 0.1143576105942986,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011364356435643564,
      "loss": 3.0431,
      "step": 431840
    },
    {
      "epoch": 0.11437625383045678,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011364156415641565,
      "loss": 3.1759,
      "step": 431850
    },
    {
      "epoch": 0.11439489706661496,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00011363956395639566,
      "loss": 3.1335,
      "step": 431860
    },
    {
      "epoch": 0.11441354030277315,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011363756375637564,
      "loss": 3.1668,
      "step": 431870
    },
    {
      "epoch": 0.11443218353893134,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011363556355635565,
      "loss": 3.1763,
      "step": 431880
    },
    {
      "epoch": 0.11445082677508951,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011363356335633564,
      "loss": 3.2173,
      "step": 431890
    },
    {
      "epoch": 0.1144694700112477,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011363156315631564,
      "loss": 3.0984,
      "step": 431900
    },
    {
      "epoch": 0.11448811324740588,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011362956295629563,
      "loss": 3.1674,
      "step": 431910
    },
    {
      "epoch": 0.11450675648356406,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011362756275627564,
      "loss": 3.2937,
      "step": 431920
    },
    {
      "epoch": 0.11452539971972225,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011362556255625563,
      "loss": 3.2285,
      "step": 431930
    },
    {
      "epoch": 0.11454404295588043,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011362356235623564,
      "loss": 3.2403,
      "step": 431940
    },
    {
      "epoch": 0.11456268619203862,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011362156215621562,
      "loss": 3.1328,
      "step": 431950
    },
    {
      "epoch": 0.1145813294281968,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011361956195619563,
      "loss": 3.0964,
      "step": 431960
    },
    {
      "epoch": 0.11459997266435498,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011361756175617561,
      "loss": 3.1284,
      "step": 431970
    },
    {
      "epoch": 0.11461861590051317,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011361556155615562,
      "loss": 2.9922,
      "step": 431980
    },
    {
      "epoch": 0.11463725913667135,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001136135613561356,
      "loss": 3.2125,
      "step": 431990
    },
    {
      "epoch": 0.11465590237282954,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011361156115611561,
      "loss": 3.1201,
      "step": 432000
    },
    {
      "epoch": 0.11467454560898772,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011360956095609561,
      "loss": 3.0681,
      "step": 432010
    },
    {
      "epoch": 0.1146931888451459,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011360756075607562,
      "loss": 3.0781,
      "step": 432020
    },
    {
      "epoch": 0.11471183208130409,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001136055605560556,
      "loss": 3.1313,
      "step": 432030
    },
    {
      "epoch": 0.11473047531746226,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011360356035603561,
      "loss": 3.3054,
      "step": 432040
    },
    {
      "epoch": 0.11474911855362045,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011360156015601559,
      "loss": 3.3124,
      "step": 432050
    },
    {
      "epoch": 0.11476776178977864,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001135995599559956,
      "loss": 3.1682,
      "step": 432060
    },
    {
      "epoch": 0.11478640502593682,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011359755975597561,
      "loss": 3.1388,
      "step": 432070
    },
    {
      "epoch": 0.114805048262095,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001135955595559556,
      "loss": 3.1965,
      "step": 432080
    },
    {
      "epoch": 0.11482369149825318,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001135935593559356,
      "loss": 3.1034,
      "step": 432090
    },
    {
      "epoch": 0.11484233473441137,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001135915591559156,
      "loss": 3.1329,
      "step": 432100
    },
    {
      "epoch": 0.11486097797056956,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001135895589558956,
      "loss": 3.0993,
      "step": 432110
    },
    {
      "epoch": 0.11487962120672773,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011358755875587559,
      "loss": 3.2149,
      "step": 432120
    },
    {
      "epoch": 0.11489826444288592,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001135855585558556,
      "loss": 3.2079,
      "step": 432130
    },
    {
      "epoch": 0.11491690767904411,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011358355835583558,
      "loss": 3.0126,
      "step": 432140
    },
    {
      "epoch": 0.11493555091520229,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011358155815581559,
      "loss": 3.1295,
      "step": 432150
    },
    {
      "epoch": 0.11495419415136048,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011357955795579558,
      "loss": 3.1147,
      "step": 432160
    },
    {
      "epoch": 0.11497283738751865,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001135775577557756,
      "loss": 3.1936,
      "step": 432170
    },
    {
      "epoch": 0.11499148062367684,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011357555755575558,
      "loss": 3.1881,
      "step": 432180
    },
    {
      "epoch": 0.11501012385983503,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011357355735573558,
      "loss": 3.1291,
      "step": 432190
    },
    {
      "epoch": 0.1150287670959932,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011357155715571557,
      "loss": 3.0678,
      "step": 432200
    },
    {
      "epoch": 0.11504741033215139,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011356955695569558,
      "loss": 3.1454,
      "step": 432210
    },
    {
      "epoch": 0.11506605356830958,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011356755675567556,
      "loss": 3.1413,
      "step": 432220
    },
    {
      "epoch": 0.11508469680446776,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011356555655565557,
      "loss": 3.0746,
      "step": 432230
    },
    {
      "epoch": 0.11510334004062595,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011356355635563556,
      "loss": 3.0962,
      "step": 432240
    },
    {
      "epoch": 0.11512198327678412,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011356155615561557,
      "loss": 3.1067,
      "step": 432250
    },
    {
      "epoch": 0.11514062651294231,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011355955595559558,
      "loss": 3.1566,
      "step": 432260
    },
    {
      "epoch": 0.1151592697491005,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011355755575557556,
      "loss": 3.3048,
      "step": 432270
    },
    {
      "epoch": 0.11517791298525867,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011355555555555557,
      "loss": 3.1086,
      "step": 432280
    },
    {
      "epoch": 0.11519655622141686,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011355355535553555,
      "loss": 3.0983,
      "step": 432290
    },
    {
      "epoch": 0.11521519945757504,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011355155515551556,
      "loss": 3.2211,
      "step": 432300
    },
    {
      "epoch": 0.11523384269373323,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011354955495549555,
      "loss": 3.2414,
      "step": 432310
    },
    {
      "epoch": 0.11525248592989142,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011354755475547555,
      "loss": 3.1728,
      "step": 432320
    },
    {
      "epoch": 0.11527112916604959,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011354555455545555,
      "loss": 3.1172,
      "step": 432330
    },
    {
      "epoch": 0.11528977240220778,
      "grad_norm": 0.625,
      "learning_rate": 0.00011354355435543556,
      "loss": 3.215,
      "step": 432340
    },
    {
      "epoch": 0.11530841563836597,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011354155415541554,
      "loss": 2.9716,
      "step": 432350
    },
    {
      "epoch": 0.11532705887452414,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011353955395539555,
      "loss": 3.1393,
      "step": 432360
    },
    {
      "epoch": 0.11534570211068233,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011353755375537553,
      "loss": 3.1858,
      "step": 432370
    },
    {
      "epoch": 0.11536434534684051,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011353555355535554,
      "loss": 2.988,
      "step": 432380
    },
    {
      "epoch": 0.1153829885829987,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011353355335533554,
      "loss": 3.0341,
      "step": 432390
    },
    {
      "epoch": 0.11540163181915689,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011353155315531555,
      "loss": 3.2666,
      "step": 432400
    },
    {
      "epoch": 0.11542027505531506,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011352955295529553,
      "loss": 3.173,
      "step": 432410
    },
    {
      "epoch": 0.11543891829147325,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011352755275527554,
      "loss": 3.0289,
      "step": 432420
    },
    {
      "epoch": 0.11545756152763142,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011352555255525552,
      "loss": 3.0464,
      "step": 432430
    },
    {
      "epoch": 0.11547620476378961,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011352355235523553,
      "loss": 3.1333,
      "step": 432440
    },
    {
      "epoch": 0.1154948479999478,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011352155215521551,
      "loss": 3.1027,
      "step": 432450
    },
    {
      "epoch": 0.11551349123610598,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011351955195519552,
      "loss": 3.1046,
      "step": 432460
    },
    {
      "epoch": 0.11553213447226417,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011351755175517553,
      "loss": 3.0295,
      "step": 432470
    },
    {
      "epoch": 0.11555077770842236,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011351555155515552,
      "loss": 3.0832,
      "step": 432480
    },
    {
      "epoch": 0.11556942094458053,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011351355135513553,
      "loss": 3.1292,
      "step": 432490
    },
    {
      "epoch": 0.11558806418073872,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011351155115511552,
      "loss": 3.1973,
      "step": 432500
    },
    {
      "epoch": 0.1156067074168969,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011350955095509552,
      "loss": 3.0491,
      "step": 432510
    },
    {
      "epoch": 0.11562535065305508,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011350755075507551,
      "loss": 3.1863,
      "step": 432520
    },
    {
      "epoch": 0.11564399388921327,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011350555055505552,
      "loss": 3.1365,
      "step": 432530
    },
    {
      "epoch": 0.11566263712537145,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001135035503550355,
      "loss": 3.1648,
      "step": 432540
    },
    {
      "epoch": 0.11568128036152964,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011350155015501551,
      "loss": 3.1247,
      "step": 432550
    },
    {
      "epoch": 0.11569992359768783,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001134995499549955,
      "loss": 3.1703,
      "step": 432560
    },
    {
      "epoch": 0.115718566833846,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011349754975497551,
      "loss": 3.0,
      "step": 432570
    },
    {
      "epoch": 0.11573721007000419,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001134955495549555,
      "loss": 3.1245,
      "step": 432580
    },
    {
      "epoch": 0.11575585330616237,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001134935493549355,
      "loss": 3.2555,
      "step": 432590
    },
    {
      "epoch": 0.11577449654232055,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011349154915491549,
      "loss": 3.1315,
      "step": 432600
    },
    {
      "epoch": 0.11579313977847874,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001134895489548955,
      "loss": 2.9759,
      "step": 432610
    },
    {
      "epoch": 0.11581178301463692,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011348754875487549,
      "loss": 3.1632,
      "step": 432620
    },
    {
      "epoch": 0.11583042625079511,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011348554855485549,
      "loss": 3.1999,
      "step": 432630
    },
    {
      "epoch": 0.11584906948695328,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011348354835483548,
      "loss": 3.126,
      "step": 432640
    },
    {
      "epoch": 0.11586771272311147,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011348154815481549,
      "loss": 3.1877,
      "step": 432650
    },
    {
      "epoch": 0.11588635595926966,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001134795479547955,
      "loss": 3.117,
      "step": 432660
    },
    {
      "epoch": 0.11590499919542784,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011347754775477548,
      "loss": 3.1066,
      "step": 432670
    },
    {
      "epoch": 0.11592364243158602,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011347554755475549,
      "loss": 3.1818,
      "step": 432680
    },
    {
      "epoch": 0.11594228566774421,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011347354735473547,
      "loss": 3.2672,
      "step": 432690
    },
    {
      "epoch": 0.11596092890390239,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011347154715471548,
      "loss": 3.1541,
      "step": 432700
    },
    {
      "epoch": 0.11597957214006058,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011346954695469548,
      "loss": 3.252,
      "step": 432710
    },
    {
      "epoch": 0.11599821537621875,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00011346754675467549,
      "loss": 3.1363,
      "step": 432720
    },
    {
      "epoch": 0.11601685861237694,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011346554655465547,
      "loss": 3.1032,
      "step": 432730
    },
    {
      "epoch": 0.11603550184853513,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011346354635463548,
      "loss": 3.2161,
      "step": 432740
    },
    {
      "epoch": 0.1160541450846933,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011346154615461546,
      "loss": 3.1758,
      "step": 432750
    },
    {
      "epoch": 0.1160727883208515,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011345954595459547,
      "loss": 3.0689,
      "step": 432760
    },
    {
      "epoch": 0.11609143155700967,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011345754575457545,
      "loss": 3.0794,
      "step": 432770
    },
    {
      "epoch": 0.11611007479316786,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011345554555455546,
      "loss": 3.2475,
      "step": 432780
    },
    {
      "epoch": 0.11612871802932605,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011345354535453546,
      "loss": 3.2537,
      "step": 432790
    },
    {
      "epoch": 0.11614736126548422,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011345154515451546,
      "loss": 3.1075,
      "step": 432800
    },
    {
      "epoch": 0.11616600450164241,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011344954495449545,
      "loss": 3.1085,
      "step": 432810
    },
    {
      "epoch": 0.1161846477378006,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011344754475447546,
      "loss": 3.1927,
      "step": 432820
    },
    {
      "epoch": 0.11620329097395878,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011344554455445544,
      "loss": 3.083,
      "step": 432830
    },
    {
      "epoch": 0.11622193421011696,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011344354435443545,
      "loss": 3.2376,
      "step": 432840
    },
    {
      "epoch": 0.11624057744627514,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011344154415441546,
      "loss": 3.1191,
      "step": 432850
    },
    {
      "epoch": 0.11625922068243333,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011343954395439544,
      "loss": 3.0468,
      "step": 432860
    },
    {
      "epoch": 0.11627786391859152,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011343754375437545,
      "loss": 2.9603,
      "step": 432870
    },
    {
      "epoch": 0.11629650715474969,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011343554355435544,
      "loss": 3.0559,
      "step": 432880
    },
    {
      "epoch": 0.11631515039090788,
      "grad_norm": 0.75,
      "learning_rate": 0.00011343354335433545,
      "loss": 3.1188,
      "step": 432890
    },
    {
      "epoch": 0.11633379362706607,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011343154315431543,
      "loss": 3.2079,
      "step": 432900
    },
    {
      "epoch": 0.11635243686322425,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011342954295429544,
      "loss": 3.2288,
      "step": 432910
    },
    {
      "epoch": 0.11637108009938243,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011342754275427543,
      "loss": 3.1481,
      "step": 432920
    },
    {
      "epoch": 0.11638972333554061,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011342554255425543,
      "loss": 3.1334,
      "step": 432930
    },
    {
      "epoch": 0.1164083665716988,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011342354235423543,
      "loss": 3.2422,
      "step": 432940
    },
    {
      "epoch": 0.11642700980785699,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011342154215421543,
      "loss": 3.1145,
      "step": 432950
    },
    {
      "epoch": 0.11644565304401516,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011341954195419542,
      "loss": 3.183,
      "step": 432960
    },
    {
      "epoch": 0.11646429628017335,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011341754175417543,
      "loss": 3.0944,
      "step": 432970
    },
    {
      "epoch": 0.11648293951633153,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011341554155415541,
      "loss": 3.1186,
      "step": 432980
    },
    {
      "epoch": 0.11650158275248972,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011341354135413542,
      "loss": 3.0534,
      "step": 432990
    },
    {
      "epoch": 0.1165202259886479,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001134115411541154,
      "loss": 2.9761,
      "step": 433000
    },
    {
      "epoch": 0.11653886922480608,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011340954095409541,
      "loss": 3.2738,
      "step": 433010
    },
    {
      "epoch": 0.11655751246096427,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011340754075407541,
      "loss": 3.0772,
      "step": 433020
    },
    {
      "epoch": 0.11657615569712246,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011340554055405542,
      "loss": 3.135,
      "step": 433030
    },
    {
      "epoch": 0.11659479893328063,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001134035403540354,
      "loss": 3.294,
      "step": 433040
    },
    {
      "epoch": 0.11661344216943882,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011340154015401541,
      "loss": 3.2406,
      "step": 433050
    },
    {
      "epoch": 0.116632085405597,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011339953995399542,
      "loss": 3.1255,
      "step": 433060
    },
    {
      "epoch": 0.11665072864175519,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001133975397539754,
      "loss": 3.1873,
      "step": 433070
    },
    {
      "epoch": 0.11666937187791337,
      "grad_norm": 0.953125,
      "learning_rate": 0.00011339553955395541,
      "loss": 3.0499,
      "step": 433080
    },
    {
      "epoch": 0.11668801511407155,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011339353935393539,
      "loss": 3.1562,
      "step": 433090
    },
    {
      "epoch": 0.11670665835022974,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001133915391539154,
      "loss": 3.1606,
      "step": 433100
    },
    {
      "epoch": 0.11672530158638791,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001133895389538954,
      "loss": 3.1391,
      "step": 433110
    },
    {
      "epoch": 0.1167439448225461,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001133875387538754,
      "loss": 3.2147,
      "step": 433120
    },
    {
      "epoch": 0.11676258805870429,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011338553855385539,
      "loss": 3.1085,
      "step": 433130
    },
    {
      "epoch": 0.11678123129486247,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001133835383538354,
      "loss": 3.0975,
      "step": 433140
    },
    {
      "epoch": 0.11679987453102066,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011338153815381538,
      "loss": 2.9894,
      "step": 433150
    },
    {
      "epoch": 0.11681851776717884,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011337953795379539,
      "loss": 3.1346,
      "step": 433160
    },
    {
      "epoch": 0.11683716100333702,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011337753775377537,
      "loss": 3.1029,
      "step": 433170
    },
    {
      "epoch": 0.11685580423949521,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011337553755375538,
      "loss": 3.1802,
      "step": 433180
    },
    {
      "epoch": 0.11687444747565338,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011337353735373537,
      "loss": 3.2271,
      "step": 433190
    },
    {
      "epoch": 0.11689309071181157,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011337153715371538,
      "loss": 3.081,
      "step": 433200
    },
    {
      "epoch": 0.11691173394796976,
      "grad_norm": 0.75,
      "learning_rate": 0.00011336953695369537,
      "loss": 3.213,
      "step": 433210
    },
    {
      "epoch": 0.11693037718412794,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011336753675367537,
      "loss": 3.0821,
      "step": 433220
    },
    {
      "epoch": 0.11694902042028613,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011336553655365536,
      "loss": 3.2054,
      "step": 433230
    },
    {
      "epoch": 0.11696766365644432,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011336353635363537,
      "loss": 3.1503,
      "step": 433240
    },
    {
      "epoch": 0.11698630689260249,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011336153615361538,
      "loss": 3.2008,
      "step": 433250
    },
    {
      "epoch": 0.11700495012876068,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011335953595359536,
      "loss": 3.0667,
      "step": 433260
    },
    {
      "epoch": 0.11702359336491885,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011335753575357537,
      "loss": 3.3092,
      "step": 433270
    },
    {
      "epoch": 0.11704223660107704,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011335553555355536,
      "loss": 3.1678,
      "step": 433280
    },
    {
      "epoch": 0.11706087983723523,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011335353535353537,
      "loss": 3.1283,
      "step": 433290
    },
    {
      "epoch": 0.11707952307339341,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011335153515351535,
      "loss": 3.2671,
      "step": 433300
    },
    {
      "epoch": 0.1170981663095516,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011334953495349536,
      "loss": 3.0613,
      "step": 433310
    },
    {
      "epoch": 0.11711680954570977,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011334753475347534,
      "loss": 3.062,
      "step": 433320
    },
    {
      "epoch": 0.11713545278186796,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011334553455345535,
      "loss": 3.0735,
      "step": 433330
    },
    {
      "epoch": 0.11715409601802615,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011334353435343535,
      "loss": 3.1808,
      "step": 433340
    },
    {
      "epoch": 0.11717273925418432,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011334153415341536,
      "loss": 3.0758,
      "step": 433350
    },
    {
      "epoch": 0.11719138249034251,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011333953395339534,
      "loss": 3.068,
      "step": 433360
    },
    {
      "epoch": 0.1172100257265007,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011333753375337535,
      "loss": 3.2323,
      "step": 433370
    },
    {
      "epoch": 0.11722866896265888,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011333553355335533,
      "loss": 3.182,
      "step": 433380
    },
    {
      "epoch": 0.11724731219881707,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011333353335333534,
      "loss": 3.1631,
      "step": 433390
    },
    {
      "epoch": 0.11726595543497524,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011333153315331532,
      "loss": 3.2027,
      "step": 433400
    },
    {
      "epoch": 0.11728459867113343,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011332953295329533,
      "loss": 3.0961,
      "step": 433410
    },
    {
      "epoch": 0.11730324190729162,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011332753275327533,
      "loss": 3.1039,
      "step": 433420
    },
    {
      "epoch": 0.1173218851434498,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011332553255325534,
      "loss": 3.1507,
      "step": 433430
    },
    {
      "epoch": 0.11734052837960798,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011332353235323535,
      "loss": 3.2133,
      "step": 433440
    },
    {
      "epoch": 0.11735917161576616,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011332153215321533,
      "loss": 3.0401,
      "step": 433450
    },
    {
      "epoch": 0.11737781485192435,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011331953195319534,
      "loss": 3.1189,
      "step": 433460
    },
    {
      "epoch": 0.11739645808808254,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011331753175317532,
      "loss": 3.2113,
      "step": 433470
    },
    {
      "epoch": 0.11741510132424071,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011331553155315533,
      "loss": 2.9908,
      "step": 433480
    },
    {
      "epoch": 0.1174337445603989,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011331353135313531,
      "loss": 3.1738,
      "step": 433490
    },
    {
      "epoch": 0.11745238779655709,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011331153115311532,
      "loss": 3.0829,
      "step": 433500
    },
    {
      "epoch": 0.11747103103271526,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011330953095309531,
      "loss": 3.2492,
      "step": 433510
    },
    {
      "epoch": 0.11748967426887345,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011330753075307532,
      "loss": 3.1464,
      "step": 433520
    },
    {
      "epoch": 0.11750831750503163,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001133055305530553,
      "loss": 3.1006,
      "step": 433530
    },
    {
      "epoch": 0.11752696074118982,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011330353035303532,
      "loss": 3.2471,
      "step": 433540
    },
    {
      "epoch": 0.117545603977348,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001133015301530153,
      "loss": 3.1638,
      "step": 433550
    },
    {
      "epoch": 0.11756424721350618,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001132995299529953,
      "loss": 3.1449,
      "step": 433560
    },
    {
      "epoch": 0.11758289044966437,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001132975297529753,
      "loss": 3.1142,
      "step": 433570
    },
    {
      "epoch": 0.11760153368582256,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001132955295529553,
      "loss": 2.9913,
      "step": 433580
    },
    {
      "epoch": 0.11762017692198073,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011329352935293529,
      "loss": 3.1311,
      "step": 433590
    },
    {
      "epoch": 0.11763882015813892,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001132915291529153,
      "loss": 3.1451,
      "step": 433600
    },
    {
      "epoch": 0.1176574633942971,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011328952895289528,
      "loss": 3.068,
      "step": 433610
    },
    {
      "epoch": 0.11767610663045529,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001132875287528753,
      "loss": 3.1264,
      "step": 433620
    },
    {
      "epoch": 0.11769474986661348,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011328552855285528,
      "loss": 3.2193,
      "step": 433630
    },
    {
      "epoch": 0.11771339310277165,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011328352835283528,
      "loss": 3.1862,
      "step": 433640
    },
    {
      "epoch": 0.11773203633892984,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001132815281528153,
      "loss": 3.055,
      "step": 433650
    },
    {
      "epoch": 0.11775067957508802,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011327952795279529,
      "loss": 3.1315,
      "step": 433660
    },
    {
      "epoch": 0.1177693228112462,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001132775277527753,
      "loss": 3.1814,
      "step": 433670
    },
    {
      "epoch": 0.1177879660474044,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011327552755275528,
      "loss": 3.1079,
      "step": 433680
    },
    {
      "epoch": 0.11780660928356257,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011327352735273529,
      "loss": 2.9764,
      "step": 433690
    },
    {
      "epoch": 0.11782525251972076,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011327152715271527,
      "loss": 3.1475,
      "step": 433700
    },
    {
      "epoch": 0.11784389575587895,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011326952695269528,
      "loss": 3.1835,
      "step": 433710
    },
    {
      "epoch": 0.11786253899203712,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011326752675267526,
      "loss": 3.0495,
      "step": 433720
    },
    {
      "epoch": 0.11788118222819531,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011326552655265527,
      "loss": 3.0712,
      "step": 433730
    },
    {
      "epoch": 0.11789982546435349,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011326352635263527,
      "loss": 3.1659,
      "step": 433740
    },
    {
      "epoch": 0.11791846870051167,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011326152615261528,
      "loss": 3.2906,
      "step": 433750
    },
    {
      "epoch": 0.11793711193666986,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011325952595259526,
      "loss": 3.2348,
      "step": 433760
    },
    {
      "epoch": 0.11795575517282804,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011325752575257527,
      "loss": 3.1833,
      "step": 433770
    },
    {
      "epoch": 0.11797439840898623,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011325552555255525,
      "loss": 3.162,
      "step": 433780
    },
    {
      "epoch": 0.1179930416451444,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011325352535253526,
      "loss": 3.0928,
      "step": 433790
    },
    {
      "epoch": 0.11801168488130259,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011325152515251524,
      "loss": 3.177,
      "step": 433800
    },
    {
      "epoch": 0.11803032811746078,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011324952495249525,
      "loss": 3.0396,
      "step": 433810
    },
    {
      "epoch": 0.11804897135361896,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011324752475247525,
      "loss": 3.1908,
      "step": 433820
    },
    {
      "epoch": 0.11806761458977715,
      "grad_norm": 0.75,
      "learning_rate": 0.00011324552455245526,
      "loss": 3.1758,
      "step": 433830
    },
    {
      "epoch": 0.11808625782593533,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011324352435243526,
      "loss": 3.076,
      "step": 433840
    },
    {
      "epoch": 0.11810490106209351,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011324152415241525,
      "loss": 3.1927,
      "step": 433850
    },
    {
      "epoch": 0.1181235442982517,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011323952395239526,
      "loss": 3.1801,
      "step": 433860
    },
    {
      "epoch": 0.11814218753440987,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011323752375237524,
      "loss": 3.1534,
      "step": 433870
    },
    {
      "epoch": 0.11816083077056806,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011323552355235525,
      "loss": 3.1598,
      "step": 433880
    },
    {
      "epoch": 0.11817947400672625,
      "grad_norm": 0.75,
      "learning_rate": 0.00011323352335233524,
      "loss": 3.1217,
      "step": 433890
    },
    {
      "epoch": 0.11819811724288443,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011323152315231524,
      "loss": 3.0104,
      "step": 433900
    },
    {
      "epoch": 0.11821676047904262,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011322952295229523,
      "loss": 3.206,
      "step": 433910
    },
    {
      "epoch": 0.1182354037152008,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011322752275227524,
      "loss": 2.9869,
      "step": 433920
    },
    {
      "epoch": 0.11825404695135898,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011322552255225522,
      "loss": 3.1694,
      "step": 433930
    },
    {
      "epoch": 0.11827269018751717,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011322352235223523,
      "loss": 3.077,
      "step": 433940
    },
    {
      "epoch": 0.11829133342367534,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011322152215221522,
      "loss": 2.9708,
      "step": 433950
    },
    {
      "epoch": 0.11830997665983353,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011321952195219522,
      "loss": 3.2401,
      "step": 433960
    },
    {
      "epoch": 0.11832861989599172,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011321752175217522,
      "loss": 3.1368,
      "step": 433970
    },
    {
      "epoch": 0.1183472631321499,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011321552155215523,
      "loss": 3.1862,
      "step": 433980
    },
    {
      "epoch": 0.11836590636830809,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011321352135213521,
      "loss": 3.0039,
      "step": 433990
    },
    {
      "epoch": 0.11838454960446626,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011321152115211522,
      "loss": 3.137,
      "step": 434000
    },
    {
      "epoch": 0.11840319284062445,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001132095209520952,
      "loss": 3.0999,
      "step": 434010
    },
    {
      "epoch": 0.11842183607678264,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011320752075207521,
      "loss": 3.0685,
      "step": 434020
    },
    {
      "epoch": 0.11844047931294081,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011320552055205522,
      "loss": 3.1924,
      "step": 434030
    },
    {
      "epoch": 0.118459122549099,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001132035203520352,
      "loss": 3.0626,
      "step": 434040
    },
    {
      "epoch": 0.11847776578525719,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011320152015201521,
      "loss": 3.1364,
      "step": 434050
    },
    {
      "epoch": 0.11849640902141537,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011319951995199521,
      "loss": 2.9996,
      "step": 434060
    },
    {
      "epoch": 0.11851505225757356,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011319751975197522,
      "loss": 3.1793,
      "step": 434070
    },
    {
      "epoch": 0.11853369549373173,
      "grad_norm": 0.75,
      "learning_rate": 0.0001131955195519552,
      "loss": 3.1633,
      "step": 434080
    },
    {
      "epoch": 0.11855233872988992,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011319351935193521,
      "loss": 3.2523,
      "step": 434090
    },
    {
      "epoch": 0.11857098196604811,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011319151915191519,
      "loss": 3.0722,
      "step": 434100
    },
    {
      "epoch": 0.11858962520220628,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001131895189518952,
      "loss": 3.2138,
      "step": 434110
    },
    {
      "epoch": 0.11860826843836447,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011318751875187518,
      "loss": 3.2082,
      "step": 434120
    },
    {
      "epoch": 0.11862691167452265,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011318551855185519,
      "loss": 3.0577,
      "step": 434130
    },
    {
      "epoch": 0.11864555491068084,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011318351835183519,
      "loss": 3.2856,
      "step": 434140
    },
    {
      "epoch": 0.11866419814683903,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001131815181518152,
      "loss": 3.1598,
      "step": 434150
    },
    {
      "epoch": 0.1186828413829972,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011317951795179518,
      "loss": 3.3111,
      "step": 434160
    },
    {
      "epoch": 0.11870148461915539,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011317751775177519,
      "loss": 3.1584,
      "step": 434170
    },
    {
      "epoch": 0.11872012785531358,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011317551755175517,
      "loss": 3.0871,
      "step": 434180
    },
    {
      "epoch": 0.11873877109147175,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011317351735173518,
      "loss": 3.1783,
      "step": 434190
    },
    {
      "epoch": 0.11875741432762994,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011317151715171517,
      "loss": 3.1432,
      "step": 434200
    },
    {
      "epoch": 0.11877605756378812,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011316951695169518,
      "loss": 3.2009,
      "step": 434210
    },
    {
      "epoch": 0.1187947007999463,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011316751675167516,
      "loss": 3.1119,
      "step": 434220
    },
    {
      "epoch": 0.1188133440361045,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011316551655165517,
      "loss": 3.0993,
      "step": 434230
    },
    {
      "epoch": 0.11883198727226267,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011316351635163518,
      "loss": 3.1904,
      "step": 434240
    },
    {
      "epoch": 0.11885063050842086,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011316151615161516,
      "loss": 3.2741,
      "step": 434250
    },
    {
      "epoch": 0.11886927374457905,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011315951595159517,
      "loss": 3.1932,
      "step": 434260
    },
    {
      "epoch": 0.11888791698073722,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011315751575157516,
      "loss": 3.015,
      "step": 434270
    },
    {
      "epoch": 0.11890656021689541,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011315551555155517,
      "loss": 3.1903,
      "step": 434280
    },
    {
      "epoch": 0.11892520345305359,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011315351535153516,
      "loss": 3.1833,
      "step": 434290
    },
    {
      "epoch": 0.11894384668921178,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011315151515151517,
      "loss": 3.0476,
      "step": 434300
    },
    {
      "epoch": 0.11896248992536997,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011314951495149515,
      "loss": 2.9693,
      "step": 434310
    },
    {
      "epoch": 0.11898113316152814,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011314751475147516,
      "loss": 3.0311,
      "step": 434320
    },
    {
      "epoch": 0.11899977639768633,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011314551455145514,
      "loss": 3.2128,
      "step": 434330
    },
    {
      "epoch": 0.1190184196338445,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011314351435143515,
      "loss": 3.1385,
      "step": 434340
    },
    {
      "epoch": 0.1190370628700027,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011314151415141513,
      "loss": 3.0965,
      "step": 434350
    },
    {
      "epoch": 0.11905570610616088,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011313951395139514,
      "loss": 3.0807,
      "step": 434360
    },
    {
      "epoch": 0.11907434934231906,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011313751375137514,
      "loss": 3.1698,
      "step": 434370
    },
    {
      "epoch": 0.11909299257847725,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011313551355135515,
      "loss": 3.0942,
      "step": 434380
    },
    {
      "epoch": 0.11911163581463544,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011313351335133513,
      "loss": 3.0687,
      "step": 434390
    },
    {
      "epoch": 0.11913027905079361,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011313151315131514,
      "loss": 3.2128,
      "step": 434400
    },
    {
      "epoch": 0.1191489222869518,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011312951295129512,
      "loss": 3.1195,
      "step": 434410
    },
    {
      "epoch": 0.11916756552310998,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011312751275127513,
      "loss": 3.002,
      "step": 434420
    },
    {
      "epoch": 0.11918620875926816,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011312551255125514,
      "loss": 3.119,
      "step": 434430
    },
    {
      "epoch": 0.11920485199542635,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011312351235123512,
      "loss": 3.1105,
      "step": 434440
    },
    {
      "epoch": 0.11922349523158453,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011312151215121513,
      "loss": 3.1923,
      "step": 434450
    },
    {
      "epoch": 0.11924213846774272,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011311951195119513,
      "loss": 3.1103,
      "step": 434460
    },
    {
      "epoch": 0.11926078170390089,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011311751175117514,
      "loss": 3.2487,
      "step": 434470
    },
    {
      "epoch": 0.11927942494005908,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011311551155115512,
      "loss": 3.0724,
      "step": 434480
    },
    {
      "epoch": 0.11929806817621727,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011311351135113513,
      "loss": 3.086,
      "step": 434490
    },
    {
      "epoch": 0.11931671141237545,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011311151115111511,
      "loss": 3.1455,
      "step": 434500
    },
    {
      "epoch": 0.11933535464853363,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011310951095109512,
      "loss": 3.1755,
      "step": 434510
    },
    {
      "epoch": 0.11935399788469182,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011310751075107511,
      "loss": 3.0337,
      "step": 434520
    },
    {
      "epoch": 0.11937264112085,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011310551055105511,
      "loss": 3.1999,
      "step": 434530
    },
    {
      "epoch": 0.11939128435700819,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001131035103510351,
      "loss": 3.1268,
      "step": 434540
    },
    {
      "epoch": 0.11940992759316636,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011310151015101511,
      "loss": 3.2284,
      "step": 434550
    },
    {
      "epoch": 0.11942857082932455,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001130995099509951,
      "loss": 3.1388,
      "step": 434560
    },
    {
      "epoch": 0.11944721406548274,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001130975097509751,
      "loss": 3.0409,
      "step": 434570
    },
    {
      "epoch": 0.11946585730164092,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011309550955095509,
      "loss": 3.171,
      "step": 434580
    },
    {
      "epoch": 0.1194845005377991,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001130935093509351,
      "loss": 3.0821,
      "step": 434590
    },
    {
      "epoch": 0.1195031437739573,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011309150915091509,
      "loss": 3.2748,
      "step": 434600
    },
    {
      "epoch": 0.11952178701011547,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001130895089508951,
      "loss": 3.0819,
      "step": 434610
    },
    {
      "epoch": 0.11954043024627366,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011308750875087508,
      "loss": 3.101,
      "step": 434620
    },
    {
      "epoch": 0.11955907348243183,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011308550855085509,
      "loss": 3.2291,
      "step": 434630
    },
    {
      "epoch": 0.11957771671859002,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001130835083508351,
      "loss": 3.0989,
      "step": 434640
    },
    {
      "epoch": 0.11959635995474821,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011308150815081508,
      "loss": 3.0814,
      "step": 434650
    },
    {
      "epoch": 0.11961500319090639,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011307950795079509,
      "loss": 3.2715,
      "step": 434660
    },
    {
      "epoch": 0.11963364642706457,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011307750775077507,
      "loss": 3.1808,
      "step": 434670
    },
    {
      "epoch": 0.11965228966322275,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011307550755075508,
      "loss": 3.1273,
      "step": 434680
    },
    {
      "epoch": 0.11967093289938094,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011307350735073508,
      "loss": 3.165,
      "step": 434690
    },
    {
      "epoch": 0.11968957613553913,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011307150715071509,
      "loss": 3.0514,
      "step": 434700
    },
    {
      "epoch": 0.1197082193716973,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011306950695069507,
      "loss": 3.1938,
      "step": 434710
    },
    {
      "epoch": 0.11972686260785549,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011306750675067508,
      "loss": 3.2186,
      "step": 434720
    },
    {
      "epoch": 0.11974550584401368,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011306550655065506,
      "loss": 2.9801,
      "step": 434730
    },
    {
      "epoch": 0.11976414908017186,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011306350635063507,
      "loss": 3.13,
      "step": 434740
    },
    {
      "epoch": 0.11978279231633004,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011306150615061505,
      "loss": 3.2282,
      "step": 434750
    },
    {
      "epoch": 0.11980143555248822,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011305950595059506,
      "loss": 3.1661,
      "step": 434760
    },
    {
      "epoch": 0.11982007878864641,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011305750575057506,
      "loss": 3.1306,
      "step": 434770
    },
    {
      "epoch": 0.1198387220248046,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011305550555055507,
      "loss": 3.2008,
      "step": 434780
    },
    {
      "epoch": 0.11985736526096277,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011305350535053505,
      "loss": 3.1674,
      "step": 434790
    },
    {
      "epoch": 0.11987600849712096,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011305150515051506,
      "loss": 3.016,
      "step": 434800
    },
    {
      "epoch": 0.11989465173327915,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011304950495049504,
      "loss": 3.2403,
      "step": 434810
    },
    {
      "epoch": 0.11991329496943733,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011304750475047505,
      "loss": 3.162,
      "step": 434820
    },
    {
      "epoch": 0.11993193820559551,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011304550455045506,
      "loss": 3.0879,
      "step": 434830
    },
    {
      "epoch": 0.11995058144175369,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011304350435043505,
      "loss": 3.0828,
      "step": 434840
    },
    {
      "epoch": 0.11996922467791188,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011304150415041505,
      "loss": 3.12,
      "step": 434850
    },
    {
      "epoch": 0.11998786791407007,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011303950395039505,
      "loss": 3.0546,
      "step": 434860
    },
    {
      "epoch": 0.12000651115022824,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011303750375037505,
      "loss": 3.1102,
      "step": 434870
    },
    {
      "epoch": 0.12002515438638643,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011303550355035504,
      "loss": 3.0135,
      "step": 434880
    },
    {
      "epoch": 0.1200437976225446,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011303350335033505,
      "loss": 3.2346,
      "step": 434890
    },
    {
      "epoch": 0.1200624408587028,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011303150315031503,
      "loss": 3.1729,
      "step": 434900
    },
    {
      "epoch": 0.12008108409486098,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011302950295029504,
      "loss": 3.0741,
      "step": 434910
    },
    {
      "epoch": 0.12009972733101916,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011302750275027503,
      "loss": 3.1104,
      "step": 434920
    },
    {
      "epoch": 0.12011837056717735,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011302550255025504,
      "loss": 3.1522,
      "step": 434930
    },
    {
      "epoch": 0.12013701380333554,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011302350235023502,
      "loss": 3.4347,
      "step": 434940
    },
    {
      "epoch": 0.12015565703949371,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011302150215021503,
      "loss": 3.1641,
      "step": 434950
    },
    {
      "epoch": 0.1201743002756519,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011301950195019501,
      "loss": 3.1311,
      "step": 434960
    },
    {
      "epoch": 0.12019294351181008,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011301750175017502,
      "loss": 3.1479,
      "step": 434970
    },
    {
      "epoch": 0.12021158674796827,
      "grad_norm": 0.7890625,
      "learning_rate": 0.000113015501550155,
      "loss": 3.2033,
      "step": 434980
    },
    {
      "epoch": 0.12023022998412645,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011301350135013502,
      "loss": 3.0011,
      "step": 434990
    },
    {
      "epoch": 0.12024887322028463,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011301150115011501,
      "loss": 3.1303,
      "step": 435000
    },
    {
      "epoch": 0.12026751645644282,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011300950095009502,
      "loss": 3.1596,
      "step": 435010
    },
    {
      "epoch": 0.120286159692601,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011300750075007503,
      "loss": 3.1708,
      "step": 435020
    },
    {
      "epoch": 0.12030480292875918,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011300550055005501,
      "loss": 3.1888,
      "step": 435030
    },
    {
      "epoch": 0.12032344616491737,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011300350035003502,
      "loss": 3.1842,
      "step": 435040
    },
    {
      "epoch": 0.12034208940107555,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000113001500150015,
      "loss": 3.2103,
      "step": 435050
    },
    {
      "epoch": 0.12036073263723374,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011299949994999501,
      "loss": 3.1129,
      "step": 435060
    },
    {
      "epoch": 0.12037937587339193,
      "grad_norm": 0.78125,
      "learning_rate": 0.000112997499749975,
      "loss": 3.2622,
      "step": 435070
    },
    {
      "epoch": 0.1203980191095501,
      "grad_norm": 0.76953125,
      "learning_rate": 0.000112995499549955,
      "loss": 3.2628,
      "step": 435080
    },
    {
      "epoch": 0.12041666234570829,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000112993499349935,
      "loss": 2.9646,
      "step": 435090
    },
    {
      "epoch": 0.12043530558186646,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011299149914991501,
      "loss": 3.137,
      "step": 435100
    },
    {
      "epoch": 0.12045394881802465,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011298949894989499,
      "loss": 3.2242,
      "step": 435110
    },
    {
      "epoch": 0.12047259205418284,
      "grad_norm": 0.625,
      "learning_rate": 0.000112987498749875,
      "loss": 3.2611,
      "step": 435120
    },
    {
      "epoch": 0.12049123529034102,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011298549854985498,
      "loss": 3.1133,
      "step": 435130
    },
    {
      "epoch": 0.1205098785264992,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011298349834983499,
      "loss": 2.9847,
      "step": 435140
    },
    {
      "epoch": 0.1205285217626574,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011298149814981499,
      "loss": 3.1225,
      "step": 435150
    },
    {
      "epoch": 0.12054716499881557,
      "grad_norm": 0.6875,
      "learning_rate": 0.000112979497949795,
      "loss": 3.1405,
      "step": 435160
    },
    {
      "epoch": 0.12056580823497376,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011297749774977498,
      "loss": 3.2381,
      "step": 435170
    },
    {
      "epoch": 0.12058445147113193,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011297549754975499,
      "loss": 3.1532,
      "step": 435180
    },
    {
      "epoch": 0.12060309470729012,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011297349734973497,
      "loss": 3.1334,
      "step": 435190
    },
    {
      "epoch": 0.12062173794344831,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011297149714971498,
      "loss": 3.1098,
      "step": 435200
    },
    {
      "epoch": 0.12064038117960649,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011296949694969496,
      "loss": 3.1164,
      "step": 435210
    },
    {
      "epoch": 0.12065902441576468,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011296749674967497,
      "loss": 3.313,
      "step": 435220
    },
    {
      "epoch": 0.12067766765192285,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011296549654965498,
      "loss": 3.2115,
      "step": 435230
    },
    {
      "epoch": 0.12069631088808104,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011296349634963497,
      "loss": 3.0793,
      "step": 435240
    },
    {
      "epoch": 0.12071495412423923,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011296149614961498,
      "loss": 3.0901,
      "step": 435250
    },
    {
      "epoch": 0.1207335973603974,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011295949594959496,
      "loss": 3.1521,
      "step": 435260
    },
    {
      "epoch": 0.1207522405965556,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011295749574957497,
      "loss": 3.1241,
      "step": 435270
    },
    {
      "epoch": 0.12077088383271378,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011295549554955496,
      "loss": 2.9966,
      "step": 435280
    },
    {
      "epoch": 0.12078952706887196,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011295349534953496,
      "loss": 3.2065,
      "step": 435290
    },
    {
      "epoch": 0.12080817030503015,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011295149514951495,
      "loss": 3.1058,
      "step": 435300
    },
    {
      "epoch": 0.12082681354118832,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011294949494949496,
      "loss": 3.1906,
      "step": 435310
    },
    {
      "epoch": 0.12084545677734651,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011294749474947495,
      "loss": 3.1267,
      "step": 435320
    },
    {
      "epoch": 0.1208641000135047,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011294549454945496,
      "loss": 3.188,
      "step": 435330
    },
    {
      "epoch": 0.12088274324966287,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011294349434943494,
      "loss": 3.2331,
      "step": 435340
    },
    {
      "epoch": 0.12090138648582106,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011294149414941495,
      "loss": 3.1978,
      "step": 435350
    },
    {
      "epoch": 0.12092002972197924,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011293949394939493,
      "loss": 3.1951,
      "step": 435360
    },
    {
      "epoch": 0.12093867295813743,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011293749374937494,
      "loss": 3.0558,
      "step": 435370
    },
    {
      "epoch": 0.12095731619429562,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011293549354935494,
      "loss": 3.1208,
      "step": 435380
    },
    {
      "epoch": 0.12097595943045379,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011293349334933493,
      "loss": 3.1332,
      "step": 435390
    },
    {
      "epoch": 0.12099460266661198,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011293149314931493,
      "loss": 3.0607,
      "step": 435400
    },
    {
      "epoch": 0.12101324590277017,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011292949294929494,
      "loss": 3.1069,
      "step": 435410
    },
    {
      "epoch": 0.12103188913892834,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011292749274927495,
      "loss": 3.0669,
      "step": 435420
    },
    {
      "epoch": 0.12105053237508653,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011292549254925493,
      "loss": 3.1279,
      "step": 435430
    },
    {
      "epoch": 0.12106917561124471,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011292349234923494,
      "loss": 3.1282,
      "step": 435440
    },
    {
      "epoch": 0.1210878188474029,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011292149214921492,
      "loss": 3.03,
      "step": 435450
    },
    {
      "epoch": 0.12110646208356109,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011291949194919493,
      "loss": 2.9978,
      "step": 435460
    },
    {
      "epoch": 0.12112510531971926,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011291749174917493,
      "loss": 3.1909,
      "step": 435470
    },
    {
      "epoch": 0.12114374855587745,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011291549154915493,
      "loss": 3.0423,
      "step": 435480
    },
    {
      "epoch": 0.12116239179203564,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011291349134913492,
      "loss": 3.0031,
      "step": 435490
    },
    {
      "epoch": 0.12118103502819381,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011291149114911493,
      "loss": 3.1466,
      "step": 435500
    },
    {
      "epoch": 0.121199678264352,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011290949094909491,
      "loss": 3.0594,
      "step": 435510
    },
    {
      "epoch": 0.12121832150051018,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011290749074907492,
      "loss": 3.1713,
      "step": 435520
    },
    {
      "epoch": 0.12123696473666837,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001129054905490549,
      "loss": 3.1081,
      "step": 435530
    },
    {
      "epoch": 0.12125560797282656,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011290349034903491,
      "loss": 3.0465,
      "step": 435540
    },
    {
      "epoch": 0.12127425120898473,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001129014901490149,
      "loss": 3.0656,
      "step": 435550
    },
    {
      "epoch": 0.12129289444514292,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011289948994899491,
      "loss": 3.0346,
      "step": 435560
    },
    {
      "epoch": 0.1213115376813011,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001128974897489749,
      "loss": 3.1115,
      "step": 435570
    },
    {
      "epoch": 0.12133018091745928,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001128954895489549,
      "loss": 3.0721,
      "step": 435580
    },
    {
      "epoch": 0.12134882415361747,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011289348934893489,
      "loss": 3.1079,
      "step": 435590
    },
    {
      "epoch": 0.12136746738977565,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001128914891489149,
      "loss": 3.1069,
      "step": 435600
    },
    {
      "epoch": 0.12138611062593384,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001128894889488949,
      "loss": 3.2266,
      "step": 435610
    },
    {
      "epoch": 0.12140475386209203,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011288748874887489,
      "loss": 3.1398,
      "step": 435620
    },
    {
      "epoch": 0.1214233970982502,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001128854885488549,
      "loss": 3.1308,
      "step": 435630
    },
    {
      "epoch": 0.12144204033440839,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011288348834883489,
      "loss": 3.1503,
      "step": 435640
    },
    {
      "epoch": 0.12146068357056657,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001128814881488149,
      "loss": 3.1687,
      "step": 435650
    },
    {
      "epoch": 0.12147932680672476,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011287948794879488,
      "loss": 3.0658,
      "step": 435660
    },
    {
      "epoch": 0.12149797004288294,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011287748774877489,
      "loss": 2.9731,
      "step": 435670
    },
    {
      "epoch": 0.12151661327904112,
      "grad_norm": 0.75,
      "learning_rate": 0.00011287548754875487,
      "loss": 3.2053,
      "step": 435680
    },
    {
      "epoch": 0.12153525651519931,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011287348734873488,
      "loss": 3.2703,
      "step": 435690
    },
    {
      "epoch": 0.12155389975135748,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011287148714871488,
      "loss": 3.1756,
      "step": 435700
    },
    {
      "epoch": 0.12157254298751567,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011286948694869487,
      "loss": 3.0869,
      "step": 435710
    },
    {
      "epoch": 0.12159118622367386,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011286748674867487,
      "loss": 2.9961,
      "step": 435720
    },
    {
      "epoch": 0.12160982945983204,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011286548654865488,
      "loss": 3.0743,
      "step": 435730
    },
    {
      "epoch": 0.12162847269599023,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011286348634863486,
      "loss": 3.0741,
      "step": 435740
    },
    {
      "epoch": 0.12164711593214841,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011286148614861487,
      "loss": 3.1773,
      "step": 435750
    },
    {
      "epoch": 0.12166575916830659,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011285948594859485,
      "loss": 3.1728,
      "step": 435760
    },
    {
      "epoch": 0.12168440240446478,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011285748574857486,
      "loss": 3.1255,
      "step": 435770
    },
    {
      "epoch": 0.12170304564062295,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011285548554855486,
      "loss": 3.1683,
      "step": 435780
    },
    {
      "epoch": 0.12172168887678114,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011285348534853487,
      "loss": 3.2231,
      "step": 435790
    },
    {
      "epoch": 0.12174033211293933,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011285148514851485,
      "loss": 3.0326,
      "step": 435800
    },
    {
      "epoch": 0.1217589753490975,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011284948494849486,
      "loss": 3.1538,
      "step": 435810
    },
    {
      "epoch": 0.1217776185852557,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011284748474847487,
      "loss": 3.2229,
      "step": 435820
    },
    {
      "epoch": 0.12179626182141388,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011284548454845485,
      "loss": 3.2409,
      "step": 435830
    },
    {
      "epoch": 0.12181490505757206,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011284348434843486,
      "loss": 3.1161,
      "step": 435840
    },
    {
      "epoch": 0.12183354829373025,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011284148414841484,
      "loss": 3.117,
      "step": 435850
    },
    {
      "epoch": 0.12185219152988842,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011283948394839485,
      "loss": 3.2283,
      "step": 435860
    },
    {
      "epoch": 0.12187083476604661,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011283748374837484,
      "loss": 3.2085,
      "step": 435870
    },
    {
      "epoch": 0.1218894780022048,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011283548354835485,
      "loss": 3.0431,
      "step": 435880
    },
    {
      "epoch": 0.12190812123836298,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011283348334833484,
      "loss": 3.0438,
      "step": 435890
    },
    {
      "epoch": 0.12192676447452117,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011283148314831484,
      "loss": 3.2317,
      "step": 435900
    },
    {
      "epoch": 0.12194540771067934,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011282948294829483,
      "loss": 3.2154,
      "step": 435910
    },
    {
      "epoch": 0.12196405094683753,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011282748274827484,
      "loss": 3.0346,
      "step": 435920
    },
    {
      "epoch": 0.12198269418299572,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011282548254825482,
      "loss": 3.1359,
      "step": 435930
    },
    {
      "epoch": 0.1220013374191539,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011282348234823483,
      "loss": 3.1264,
      "step": 435940
    },
    {
      "epoch": 0.12201998065531208,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011282148214821482,
      "loss": 3.0531,
      "step": 435950
    },
    {
      "epoch": 0.12203862389147027,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011281948194819483,
      "loss": 3.1332,
      "step": 435960
    },
    {
      "epoch": 0.12205726712762845,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011281748174817481,
      "loss": 3.0475,
      "step": 435970
    },
    {
      "epoch": 0.12207591036378664,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011281548154815482,
      "loss": 3.2161,
      "step": 435980
    },
    {
      "epoch": 0.12209455359994481,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001128134813481348,
      "loss": 3.0638,
      "step": 435990
    },
    {
      "epoch": 0.122113196836103,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011281148114811481,
      "loss": 3.0435,
      "step": 436000
    },
    {
      "epoch": 0.12213184007226119,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011280948094809482,
      "loss": 3.162,
      "step": 436010
    },
    {
      "epoch": 0.12215048330841936,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001128074807480748,
      "loss": 3.137,
      "step": 436020
    },
    {
      "epoch": 0.12216912654457755,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011280548054805481,
      "loss": 2.9877,
      "step": 436030
    },
    {
      "epoch": 0.12218776978073573,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011280348034803481,
      "loss": 3.0989,
      "step": 436040
    },
    {
      "epoch": 0.12220641301689392,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011280148014801482,
      "loss": 3.1568,
      "step": 436050
    },
    {
      "epoch": 0.1222250562530521,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001127994799479948,
      "loss": 3.0724,
      "step": 436060
    },
    {
      "epoch": 0.12224369948921028,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011279747974797481,
      "loss": 3.1192,
      "step": 436070
    },
    {
      "epoch": 0.12226234272536847,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011279547954795479,
      "loss": 3.2818,
      "step": 436080
    },
    {
      "epoch": 0.12228098596152666,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001127934793479348,
      "loss": 3.2864,
      "step": 436090
    },
    {
      "epoch": 0.12229962919768483,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001127914791479148,
      "loss": 3.1372,
      "step": 436100
    },
    {
      "epoch": 0.12231827243384302,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001127894789478948,
      "loss": 3.2361,
      "step": 436110
    },
    {
      "epoch": 0.1223369156700012,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011278747874787479,
      "loss": 3.008,
      "step": 436120
    },
    {
      "epoch": 0.12235555890615939,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001127854785478548,
      "loss": 3.2226,
      "step": 436130
    },
    {
      "epoch": 0.12237420214231758,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011278347834783478,
      "loss": 3.1512,
      "step": 436140
    },
    {
      "epoch": 0.12239284537847575,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011278147814781479,
      "loss": 3.2037,
      "step": 436150
    },
    {
      "epoch": 0.12241148861463394,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011277947794779477,
      "loss": 3.1028,
      "step": 436160
    },
    {
      "epoch": 0.12243013185079213,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011277747774777478,
      "loss": 3.2402,
      "step": 436170
    },
    {
      "epoch": 0.1224487750869503,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011277547754775478,
      "loss": 3.147,
      "step": 436180
    },
    {
      "epoch": 0.12246741832310849,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011277347734773478,
      "loss": 3.1596,
      "step": 436190
    },
    {
      "epoch": 0.12248606155926667,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011277147714771477,
      "loss": 3.1739,
      "step": 436200
    },
    {
      "epoch": 0.12250470479542486,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011276947694769478,
      "loss": 3.1734,
      "step": 436210
    },
    {
      "epoch": 0.12252334803158305,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011276747674767478,
      "loss": 2.9906,
      "step": 436220
    },
    {
      "epoch": 0.12254199126774122,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011276547654765477,
      "loss": 3.1559,
      "step": 436230
    },
    {
      "epoch": 0.12256063450389941,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011276347634763478,
      "loss": 3.229,
      "step": 436240
    },
    {
      "epoch": 0.12257927774005759,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011276147614761476,
      "loss": 3.1567,
      "step": 436250
    },
    {
      "epoch": 0.12259792097621577,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011275947594759477,
      "loss": 3.207,
      "step": 436260
    },
    {
      "epoch": 0.12261656421237396,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011275747574757476,
      "loss": 2.9665,
      "step": 436270
    },
    {
      "epoch": 0.12263520744853214,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011275547554755477,
      "loss": 3.1621,
      "step": 436280
    },
    {
      "epoch": 0.12265385068469033,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011275347534753475,
      "loss": 3.0378,
      "step": 436290
    },
    {
      "epoch": 0.12267249392084852,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011275147514751476,
      "loss": 3.1639,
      "step": 436300
    },
    {
      "epoch": 0.12269113715700669,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011274947494749475,
      "loss": 3.1566,
      "step": 436310
    },
    {
      "epoch": 0.12270978039316488,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011274747474747475,
      "loss": 3.2144,
      "step": 436320
    },
    {
      "epoch": 0.12272842362932306,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011274547454745475,
      "loss": 3.011,
      "step": 436330
    },
    {
      "epoch": 0.12274706686548124,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00011274347434743475,
      "loss": 3.0269,
      "step": 436340
    },
    {
      "epoch": 0.12276571010163943,
      "grad_norm": 0.625,
      "learning_rate": 0.00011274147414741474,
      "loss": 3.247,
      "step": 436350
    },
    {
      "epoch": 0.12278435333779761,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011273947394739475,
      "loss": 3.0762,
      "step": 436360
    },
    {
      "epoch": 0.1228029965739558,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011273747374737473,
      "loss": 3.1013,
      "step": 436370
    },
    {
      "epoch": 0.12282163981011397,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011273547354735474,
      "loss": 3.2564,
      "step": 436380
    },
    {
      "epoch": 0.12284028304627216,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011273347334733472,
      "loss": 3.0669,
      "step": 436390
    },
    {
      "epoch": 0.12285892628243035,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011273147314731473,
      "loss": 3.1977,
      "step": 436400
    },
    {
      "epoch": 0.12287756951858853,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011272947294729474,
      "loss": 3.1718,
      "step": 436410
    },
    {
      "epoch": 0.12289621275474671,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011272747274727474,
      "loss": 3.1851,
      "step": 436420
    },
    {
      "epoch": 0.1229148559909049,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011272547254725475,
      "loss": 3.0678,
      "step": 436430
    },
    {
      "epoch": 0.12293349922706308,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011272347234723473,
      "loss": 3.0213,
      "step": 436440
    },
    {
      "epoch": 0.12295214246322127,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011272147214721474,
      "loss": 3.1887,
      "step": 436450
    },
    {
      "epoch": 0.12297078569937944,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011271947194719472,
      "loss": 3.1428,
      "step": 436460
    },
    {
      "epoch": 0.12298942893553763,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011271747174717473,
      "loss": 2.9987,
      "step": 436470
    },
    {
      "epoch": 0.12300807217169582,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011271547154715471,
      "loss": 3.0912,
      "step": 436480
    },
    {
      "epoch": 0.123026715407854,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011271347134713472,
      "loss": 3.1732,
      "step": 436490
    },
    {
      "epoch": 0.12304535864401218,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011271147114711472,
      "loss": 3.2228,
      "step": 436500
    },
    {
      "epoch": 0.12306400188017037,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011270947094709472,
      "loss": 3.1325,
      "step": 436510
    },
    {
      "epoch": 0.12308264511632855,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011270747074707471,
      "loss": 3.0259,
      "step": 436520
    },
    {
      "epoch": 0.12310128835248674,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011270547054705472,
      "loss": 3.0625,
      "step": 436530
    },
    {
      "epoch": 0.12311993158864491,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001127034703470347,
      "loss": 3.1747,
      "step": 436540
    },
    {
      "epoch": 0.1231385748248031,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011270147014701471,
      "loss": 3.2761,
      "step": 436550
    },
    {
      "epoch": 0.12315721806096129,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011269946994699469,
      "loss": 3.0933,
      "step": 436560
    },
    {
      "epoch": 0.12317586129711947,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001126974697469747,
      "loss": 3.1843,
      "step": 436570
    },
    {
      "epoch": 0.12319450453327765,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001126954695469547,
      "loss": 3.1479,
      "step": 436580
    },
    {
      "epoch": 0.12321314776943583,
      "grad_norm": 0.75,
      "learning_rate": 0.0001126934693469347,
      "loss": 3.1883,
      "step": 436590
    },
    {
      "epoch": 0.12323179100559402,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011269146914691471,
      "loss": 3.0686,
      "step": 436600
    },
    {
      "epoch": 0.12325043424175221,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001126894689468947,
      "loss": 3.1297,
      "step": 436610
    },
    {
      "epoch": 0.12326907747791038,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001126874687468747,
      "loss": 3.144,
      "step": 436620
    },
    {
      "epoch": 0.12328772071406857,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011268546854685469,
      "loss": 3.1658,
      "step": 436630
    },
    {
      "epoch": 0.12330636395022676,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001126834683468347,
      "loss": 3.209,
      "step": 436640
    },
    {
      "epoch": 0.12332500718638494,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011268146814681469,
      "loss": 3.2114,
      "step": 436650
    },
    {
      "epoch": 0.12334365042254312,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011267946794679469,
      "loss": 3.1671,
      "step": 436660
    },
    {
      "epoch": 0.1233622936587013,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011267746774677468,
      "loss": 3.1049,
      "step": 436670
    },
    {
      "epoch": 0.12338093689485949,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011267546754675469,
      "loss": 3.1678,
      "step": 436680
    },
    {
      "epoch": 0.12339958013101768,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011267346734673467,
      "loss": 3.1467,
      "step": 436690
    },
    {
      "epoch": 0.12341822336717585,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011267146714671468,
      "loss": 3.0926,
      "step": 436700
    },
    {
      "epoch": 0.12343686660333404,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011266946694669466,
      "loss": 3.2518,
      "step": 436710
    },
    {
      "epoch": 0.12345550983949222,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011266746674667467,
      "loss": 3.2063,
      "step": 436720
    },
    {
      "epoch": 0.1234741530756504,
      "grad_norm": 0.953125,
      "learning_rate": 0.00011266546654665467,
      "loss": 3.0336,
      "step": 436730
    },
    {
      "epoch": 0.1234927963118086,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011266346634663468,
      "loss": 3.1466,
      "step": 436740
    },
    {
      "epoch": 0.12351143954796677,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011266146614661466,
      "loss": 3.1401,
      "step": 436750
    },
    {
      "epoch": 0.12353008278412496,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011265946594659467,
      "loss": 3.1165,
      "step": 436760
    },
    {
      "epoch": 0.12354872602028315,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011265746574657465,
      "loss": 3.168,
      "step": 436770
    },
    {
      "epoch": 0.12356736925644132,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011265546554655466,
      "loss": 3.0118,
      "step": 436780
    },
    {
      "epoch": 0.12358601249259951,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011265346534653464,
      "loss": 3.235,
      "step": 436790
    },
    {
      "epoch": 0.12360465572875769,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011265146514651465,
      "loss": 3.1944,
      "step": 436800
    },
    {
      "epoch": 0.12362329896491588,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011264946494649466,
      "loss": 3.1661,
      "step": 436810
    },
    {
      "epoch": 0.12364194220107406,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011264746474647466,
      "loss": 3.1433,
      "step": 436820
    },
    {
      "epoch": 0.12366058543723224,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00011264546454645467,
      "loss": 3.1201,
      "step": 436830
    },
    {
      "epoch": 0.12367922867339043,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011264346434643465,
      "loss": 3.2053,
      "step": 436840
    },
    {
      "epoch": 0.12369787190954862,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011264146414641466,
      "loss": 3.1032,
      "step": 436850
    },
    {
      "epoch": 0.12371651514570679,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011263946394639464,
      "loss": 3.0862,
      "step": 436860
    },
    {
      "epoch": 0.12373515838186498,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011263746374637465,
      "loss": 3.1258,
      "step": 436870
    },
    {
      "epoch": 0.12375380161802316,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011263546354635463,
      "loss": 3.1679,
      "step": 436880
    },
    {
      "epoch": 0.12377244485418135,
      "grad_norm": 0.75,
      "learning_rate": 0.00011263346334633464,
      "loss": 3.1137,
      "step": 436890
    },
    {
      "epoch": 0.12379108809033953,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011263146314631463,
      "loss": 3.0723,
      "step": 436900
    },
    {
      "epoch": 0.12380973132649771,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00011262946294629464,
      "loss": 3.197,
      "step": 436910
    },
    {
      "epoch": 0.1238283745626559,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011262746274627463,
      "loss": 3.0283,
      "step": 436920
    },
    {
      "epoch": 0.12384701779881407,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011262546254625463,
      "loss": 3.1424,
      "step": 436930
    },
    {
      "epoch": 0.12386566103497226,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011262346234623462,
      "loss": 3.1658,
      "step": 436940
    },
    {
      "epoch": 0.12388430427113045,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011262146214621463,
      "loss": 3.1035,
      "step": 436950
    },
    {
      "epoch": 0.12390294750728863,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011261946194619462,
      "loss": 3.2259,
      "step": 436960
    },
    {
      "epoch": 0.12392159074344682,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011261746174617463,
      "loss": 3.1435,
      "step": 436970
    },
    {
      "epoch": 0.123940233979605,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011261546154615461,
      "loss": 3.1083,
      "step": 436980
    },
    {
      "epoch": 0.12395887721576318,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011261346134613462,
      "loss": 3.1988,
      "step": 436990
    },
    {
      "epoch": 0.12397752045192137,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011261146114611463,
      "loss": 3.1432,
      "step": 437000
    },
    {
      "epoch": 0.12399616368807954,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011260946094609461,
      "loss": 3.1369,
      "step": 437010
    },
    {
      "epoch": 0.12401480692423773,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011260746074607462,
      "loss": 3.2945,
      "step": 437020
    },
    {
      "epoch": 0.12403345016039592,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001126054605460546,
      "loss": 3.1069,
      "step": 437030
    },
    {
      "epoch": 0.1240520933965541,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011260346034603461,
      "loss": 3.1959,
      "step": 437040
    },
    {
      "epoch": 0.12407073663271229,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011260146014601461,
      "loss": 3.1222,
      "step": 437050
    },
    {
      "epoch": 0.12408937986887046,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011259945994599462,
      "loss": 3.1151,
      "step": 437060
    },
    {
      "epoch": 0.12410802310502865,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001125974597459746,
      "loss": 3.2203,
      "step": 437070
    },
    {
      "epoch": 0.12412666634118684,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011259545954595461,
      "loss": 3.1836,
      "step": 437080
    },
    {
      "epoch": 0.12414530957734501,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011259345934593459,
      "loss": 3.2435,
      "step": 437090
    },
    {
      "epoch": 0.1241639528135032,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001125914591459146,
      "loss": 3.278,
      "step": 437100
    },
    {
      "epoch": 0.12418259604966139,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011258945894589458,
      "loss": 3.1129,
      "step": 437110
    },
    {
      "epoch": 0.12420123928581957,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011258745874587459,
      "loss": 3.0486,
      "step": 437120
    },
    {
      "epoch": 0.12421988252197776,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011258545854585459,
      "loss": 3.1639,
      "step": 437130
    },
    {
      "epoch": 0.12423852575813593,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001125834583458346,
      "loss": 3.1067,
      "step": 437140
    },
    {
      "epoch": 0.12425716899429412,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011258145814581458,
      "loss": 3.0813,
      "step": 437150
    },
    {
      "epoch": 0.12427581223045231,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011257945794579459,
      "loss": 3.1754,
      "step": 437160
    },
    {
      "epoch": 0.12429445546661048,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011257745774577457,
      "loss": 3.2105,
      "step": 437170
    },
    {
      "epoch": 0.12431309870276867,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011257545754575458,
      "loss": 3.1957,
      "step": 437180
    },
    {
      "epoch": 0.12433174193892686,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011257345734573459,
      "loss": 3.1434,
      "step": 437190
    },
    {
      "epoch": 0.12435038517508504,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011257145714571457,
      "loss": 3.1325,
      "step": 437200
    },
    {
      "epoch": 0.12436902841124323,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011256945694569458,
      "loss": 3.1408,
      "step": 437210
    },
    {
      "epoch": 0.1243876716474014,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011256745674567457,
      "loss": 3.1723,
      "step": 437220
    },
    {
      "epoch": 0.12440631488355959,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011256545654565458,
      "loss": 3.1736,
      "step": 437230
    },
    {
      "epoch": 0.12442495811971778,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011256345634563457,
      "loss": 3.1831,
      "step": 437240
    },
    {
      "epoch": 0.12444360135587595,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011256145614561458,
      "loss": 3.2235,
      "step": 437250
    },
    {
      "epoch": 0.12446224459203414,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011255945594559456,
      "loss": 3.1619,
      "step": 437260
    },
    {
      "epoch": 0.12448088782819232,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011255745574557457,
      "loss": 3.1524,
      "step": 437270
    },
    {
      "epoch": 0.12449953106435051,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011255545554555456,
      "loss": 3.2076,
      "step": 437280
    },
    {
      "epoch": 0.1245181743005087,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011255345534553456,
      "loss": 3.1909,
      "step": 437290
    },
    {
      "epoch": 0.12453681753666687,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011255145514551455,
      "loss": 3.1402,
      "step": 437300
    },
    {
      "epoch": 0.12455546077282506,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011254945494549456,
      "loss": 3.0649,
      "step": 437310
    },
    {
      "epoch": 0.12457410400898325,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011254745474547454,
      "loss": 3.0476,
      "step": 437320
    },
    {
      "epoch": 0.12459274724514142,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011254545454545455,
      "loss": 3.1901,
      "step": 437330
    },
    {
      "epoch": 0.12461139048129961,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011254345434543454,
      "loss": 3.1629,
      "step": 437340
    },
    {
      "epoch": 0.12463003371745779,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011254145414541454,
      "loss": 3.1973,
      "step": 437350
    },
    {
      "epoch": 0.12464867695361598,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011253945394539454,
      "loss": 3.1597,
      "step": 437360
    },
    {
      "epoch": 0.12466732018977417,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011253745374537455,
      "loss": 3.1863,
      "step": 437370
    },
    {
      "epoch": 0.12468596342593234,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011253545354535453,
      "loss": 3.1433,
      "step": 437380
    },
    {
      "epoch": 0.12470460666209053,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011253345334533454,
      "loss": 3.0943,
      "step": 437390
    },
    {
      "epoch": 0.1247232498982487,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011253145314531455,
      "loss": 3.0588,
      "step": 437400
    },
    {
      "epoch": 0.1247418931344069,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011252945294529453,
      "loss": 3.0517,
      "step": 437410
    },
    {
      "epoch": 0.12476053637056508,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011252745274527454,
      "loss": 3.1604,
      "step": 437420
    },
    {
      "epoch": 0.12477917960672326,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011252545254525452,
      "loss": 2.9456,
      "step": 437430
    },
    {
      "epoch": 0.12479782284288145,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011252345234523453,
      "loss": 3.0819,
      "step": 437440
    },
    {
      "epoch": 0.12481646607903964,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011252145214521453,
      "loss": 3.3097,
      "step": 437450
    },
    {
      "epoch": 0.12483510931519781,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011251945194519454,
      "loss": 3.2244,
      "step": 437460
    },
    {
      "epoch": 0.124853752551356,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011251745174517452,
      "loss": 3.1874,
      "step": 437470
    },
    {
      "epoch": 0.12487239578751418,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011251545154515453,
      "loss": 3.1444,
      "step": 437480
    },
    {
      "epoch": 0.12489103902367236,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011251345134513451,
      "loss": 3.0804,
      "step": 437490
    },
    {
      "epoch": 0.12490968225983055,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011251145114511452,
      "loss": 3.0919,
      "step": 437500
    },
    {
      "epoch": 0.12492832549598873,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001125094509450945,
      "loss": 3.1712,
      "step": 437510
    },
    {
      "epoch": 0.12494696873214692,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011250745074507451,
      "loss": 3.1299,
      "step": 437520
    },
    {
      "epoch": 0.12496561196830511,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001125054505450545,
      "loss": 3.0614,
      "step": 437530
    },
    {
      "epoch": 0.12498425520446328,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011250345034503452,
      "loss": 3.0251,
      "step": 437540
    },
    {
      "epoch": 0.12500289844062146,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001125014501450145,
      "loss": 3.238,
      "step": 437550
    },
    {
      "epoch": 0.12502154167677965,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001124994499449945,
      "loss": 3.3088,
      "step": 437560
    },
    {
      "epoch": 0.12504018491293784,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011249744974497449,
      "loss": 3.112,
      "step": 437570
    },
    {
      "epoch": 0.12505882814909602,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001124954495449545,
      "loss": 3.1244,
      "step": 437580
    },
    {
      "epoch": 0.1250774713852542,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001124934493449345,
      "loss": 3.0251,
      "step": 437590
    },
    {
      "epoch": 0.12509611462141237,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001124914491449145,
      "loss": 3.0491,
      "step": 437600
    },
    {
      "epoch": 0.12511475785757056,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001124894489448945,
      "loss": 3.1545,
      "step": 437610
    },
    {
      "epoch": 0.12513340109372875,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001124874487448745,
      "loss": 3.0741,
      "step": 437620
    },
    {
      "epoch": 0.12515204432988694,
      "grad_norm": 0.75,
      "learning_rate": 0.0001124854485448545,
      "loss": 3.1239,
      "step": 437630
    },
    {
      "epoch": 0.12517068756604513,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011248344834483448,
      "loss": 3.099,
      "step": 437640
    },
    {
      "epoch": 0.1251893308022033,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001124814481448145,
      "loss": 3.1949,
      "step": 437650
    },
    {
      "epoch": 0.12520797403836148,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011247944794479448,
      "loss": 3.1674,
      "step": 437660
    },
    {
      "epoch": 0.12522661727451967,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011247744774477448,
      "loss": 3.1207,
      "step": 437670
    },
    {
      "epoch": 0.12524526051067786,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011247544754475448,
      "loss": 3.0917,
      "step": 437680
    },
    {
      "epoch": 0.12526390374683605,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011247344734473449,
      "loss": 3.2024,
      "step": 437690
    },
    {
      "epoch": 0.12528254698299424,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011247144714471447,
      "loss": 3.0907,
      "step": 437700
    },
    {
      "epoch": 0.1253011902191524,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011246944694469448,
      "loss": 3.0845,
      "step": 437710
    },
    {
      "epoch": 0.1253198334553106,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011246744674467446,
      "loss": 3.1207,
      "step": 437720
    },
    {
      "epoch": 0.12533847669146878,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011246544654465447,
      "loss": 3.2021,
      "step": 437730
    },
    {
      "epoch": 0.12535711992762696,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011246344634463445,
      "loss": 3.1253,
      "step": 437740
    },
    {
      "epoch": 0.12537576316378515,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011246144614461446,
      "loss": 3.2948,
      "step": 437750
    },
    {
      "epoch": 0.12539440639994331,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011245944594459446,
      "loss": 3.12,
      "step": 437760
    },
    {
      "epoch": 0.1254130496361015,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011245744574457447,
      "loss": 3.1892,
      "step": 437770
    },
    {
      "epoch": 0.1254316928722597,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011245544554455448,
      "loss": 3.132,
      "step": 437780
    },
    {
      "epoch": 0.12545033610841788,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011245344534453446,
      "loss": 3.1258,
      "step": 437790
    },
    {
      "epoch": 0.12546897934457607,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011245144514451447,
      "loss": 3.1274,
      "step": 437800
    },
    {
      "epoch": 0.12548762258073423,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011244944494449445,
      "loss": 3.1361,
      "step": 437810
    },
    {
      "epoch": 0.12550626581689242,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011244744474447446,
      "loss": 3.0816,
      "step": 437820
    },
    {
      "epoch": 0.1255249090530506,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011244544454445444,
      "loss": 3.0882,
      "step": 437830
    },
    {
      "epoch": 0.1255435522892088,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011244344434443445,
      "loss": 3.1016,
      "step": 437840
    },
    {
      "epoch": 0.125562195525367,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011244144414441445,
      "loss": 3.3092,
      "step": 437850
    },
    {
      "epoch": 0.12558083876152515,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011243944394439446,
      "loss": 2.9855,
      "step": 437860
    },
    {
      "epoch": 0.12559948199768334,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011243744374437444,
      "loss": 3.0597,
      "step": 437870
    },
    {
      "epoch": 0.12561812523384153,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011243544354435445,
      "loss": 3.0356,
      "step": 437880
    },
    {
      "epoch": 0.12563676846999972,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011243344334433443,
      "loss": 3.1355,
      "step": 437890
    },
    {
      "epoch": 0.1256554117061579,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011243144314431444,
      "loss": 3.0247,
      "step": 437900
    },
    {
      "epoch": 0.12567405494231607,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011242944294429443,
      "loss": 3.0981,
      "step": 437910
    },
    {
      "epoch": 0.12569269817847425,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011242744274427444,
      "loss": 3.109,
      "step": 437920
    },
    {
      "epoch": 0.12571134141463244,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011242544254425442,
      "loss": 3.019,
      "step": 437930
    },
    {
      "epoch": 0.12572998465079063,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011242344234423443,
      "loss": 3.0861,
      "step": 437940
    },
    {
      "epoch": 0.12574862788694882,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011242144214421442,
      "loss": 3.0996,
      "step": 437950
    },
    {
      "epoch": 0.125767271123107,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011241944194419442,
      "loss": 3.1121,
      "step": 437960
    },
    {
      "epoch": 0.12578591435926517,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011241744174417441,
      "loss": 3.0213,
      "step": 437970
    },
    {
      "epoch": 0.12580455759542336,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011241544154415442,
      "loss": 3.0961,
      "step": 437980
    },
    {
      "epoch": 0.12582320083158155,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011241344134413443,
      "loss": 3.1738,
      "step": 437990
    },
    {
      "epoch": 0.12584184406773974,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011241144114411442,
      "loss": 3.0846,
      "step": 438000
    },
    {
      "epoch": 0.12586048730389793,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011240944094409443,
      "loss": 3.223,
      "step": 438010
    },
    {
      "epoch": 0.1258791305400561,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011240744074407441,
      "loss": 3.1523,
      "step": 438020
    },
    {
      "epoch": 0.12589777377621428,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011240544054405442,
      "loss": 3.148,
      "step": 438030
    },
    {
      "epoch": 0.12591641701237247,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001124034403440344,
      "loss": 3.1113,
      "step": 438040
    },
    {
      "epoch": 0.12593506024853066,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011240144014401441,
      "loss": 3.1572,
      "step": 438050
    },
    {
      "epoch": 0.12595370348468884,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001123994399439944,
      "loss": 3.1619,
      "step": 438060
    },
    {
      "epoch": 0.125972346720847,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001123974397439744,
      "loss": 3.2087,
      "step": 438070
    },
    {
      "epoch": 0.1259909899570052,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001123954395439544,
      "loss": 3.0858,
      "step": 438080
    },
    {
      "epoch": 0.12600963319316338,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011239343934393441,
      "loss": 3.1516,
      "step": 438090
    },
    {
      "epoch": 0.12602827642932157,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011239143914391439,
      "loss": 3.2218,
      "step": 438100
    },
    {
      "epoch": 0.12604691966547976,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001123894389438944,
      "loss": 3.1165,
      "step": 438110
    },
    {
      "epoch": 0.12606556290163792,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011238743874387438,
      "loss": 3.0784,
      "step": 438120
    },
    {
      "epoch": 0.1260842061377961,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011238543854385439,
      "loss": 3.2223,
      "step": 438130
    },
    {
      "epoch": 0.1261028493739543,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011238343834383439,
      "loss": 3.1479,
      "step": 438140
    },
    {
      "epoch": 0.1261214926101125,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011238143814381438,
      "loss": 3.0544,
      "step": 438150
    },
    {
      "epoch": 0.12614013584627068,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011237943794379438,
      "loss": 3.2117,
      "step": 438160
    },
    {
      "epoch": 0.12615877908242887,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00011237743774377439,
      "loss": 3.0243,
      "step": 438170
    },
    {
      "epoch": 0.12617742231858703,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001123754375437544,
      "loss": 3.1268,
      "step": 438180
    },
    {
      "epoch": 0.12619606555474522,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011237343734373438,
      "loss": 3.2853,
      "step": 438190
    },
    {
      "epoch": 0.1262147087909034,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011237143714371439,
      "loss": 3.0976,
      "step": 438200
    },
    {
      "epoch": 0.1262333520270616,
      "grad_norm": 0.75,
      "learning_rate": 0.00011236943694369437,
      "loss": 3.0618,
      "step": 438210
    },
    {
      "epoch": 0.12625199526321979,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011236743674367438,
      "loss": 3.1747,
      "step": 438220
    },
    {
      "epoch": 0.12627063849937795,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011236543654365437,
      "loss": 3.0041,
      "step": 438230
    },
    {
      "epoch": 0.12628928173553614,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011236343634363438,
      "loss": 3.1227,
      "step": 438240
    },
    {
      "epoch": 0.12630792497169432,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011236143614361436,
      "loss": 3.0565,
      "step": 438250
    },
    {
      "epoch": 0.1263265682078525,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011235943594359437,
      "loss": 3.1594,
      "step": 438260
    },
    {
      "epoch": 0.1263452114440107,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011235743574357436,
      "loss": 3.1399,
      "step": 438270
    },
    {
      "epoch": 0.12636385468016886,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011235543554355437,
      "loss": 3.0536,
      "step": 438280
    },
    {
      "epoch": 0.12638249791632705,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011235343534353435,
      "loss": 3.1069,
      "step": 438290
    },
    {
      "epoch": 0.12640114115248524,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011235143514351436,
      "loss": 3.1737,
      "step": 438300
    },
    {
      "epoch": 0.12641978438864343,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011234943494349435,
      "loss": 3.1124,
      "step": 438310
    },
    {
      "epoch": 0.12643842762480162,
      "grad_norm": 0.875,
      "learning_rate": 0.00011234743474347436,
      "loss": 3.1338,
      "step": 438320
    },
    {
      "epoch": 0.12645707086095978,
      "grad_norm": 0.75,
      "learning_rate": 0.00011234543454345434,
      "loss": 3.0803,
      "step": 438330
    },
    {
      "epoch": 0.12647571409711797,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011234343434343435,
      "loss": 3.1771,
      "step": 438340
    },
    {
      "epoch": 0.12649435733327616,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011234143414341433,
      "loss": 3.1958,
      "step": 438350
    },
    {
      "epoch": 0.12651300056943435,
      "grad_norm": 0.75,
      "learning_rate": 0.00011233943394339434,
      "loss": 3.152,
      "step": 438360
    },
    {
      "epoch": 0.12653164380559254,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011233743374337433,
      "loss": 3.0511,
      "step": 438370
    },
    {
      "epoch": 0.12655028704175073,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011233543354335433,
      "loss": 3.0941,
      "step": 438380
    },
    {
      "epoch": 0.1265689302779089,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011233343334333434,
      "loss": 3.1191,
      "step": 438390
    },
    {
      "epoch": 0.12658757351406708,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011233143314331434,
      "loss": 3.1558,
      "step": 438400
    },
    {
      "epoch": 0.12660621675022526,
      "grad_norm": 0.75,
      "learning_rate": 0.00011232943294329435,
      "loss": 3.1785,
      "step": 438410
    },
    {
      "epoch": 0.12662485998638345,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011232743274327433,
      "loss": 3.1208,
      "step": 438420
    },
    {
      "epoch": 0.12664350322254164,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011232543254325434,
      "loss": 3.1267,
      "step": 438430
    },
    {
      "epoch": 0.1266621464586998,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011232343234323432,
      "loss": 3.154,
      "step": 438440
    },
    {
      "epoch": 0.126680789694858,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011232143214321433,
      "loss": 3.1993,
      "step": 438450
    },
    {
      "epoch": 0.12669943293101618,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011231943194319433,
      "loss": 3.1094,
      "step": 438460
    },
    {
      "epoch": 0.12671807616717437,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011231743174317432,
      "loss": 3.1159,
      "step": 438470
    },
    {
      "epoch": 0.12673671940333256,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011231543154315432,
      "loss": 3.143,
      "step": 438480
    },
    {
      "epoch": 0.12675536263949072,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011231343134313433,
      "loss": 3.2776,
      "step": 438490
    },
    {
      "epoch": 0.1267740058756489,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011231143114311431,
      "loss": 3.1182,
      "step": 438500
    },
    {
      "epoch": 0.1267926491118071,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011230943094309432,
      "loss": 3.0749,
      "step": 438510
    },
    {
      "epoch": 0.1268112923479653,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001123074307430743,
      "loss": 3.1056,
      "step": 438520
    },
    {
      "epoch": 0.12682993558412348,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011230543054305431,
      "loss": 3.1553,
      "step": 438530
    },
    {
      "epoch": 0.12684857882028164,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001123034303430343,
      "loss": 3.1409,
      "step": 438540
    },
    {
      "epoch": 0.12686722205643983,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011230143014301431,
      "loss": 3.0855,
      "step": 438550
    },
    {
      "epoch": 0.12688586529259802,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001122994299429943,
      "loss": 3.2441,
      "step": 438560
    },
    {
      "epoch": 0.1269045085287562,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001122974297429743,
      "loss": 3.164,
      "step": 438570
    },
    {
      "epoch": 0.1269231517649144,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011229542954295431,
      "loss": 3.0145,
      "step": 438580
    },
    {
      "epoch": 0.12694179500107255,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001122934293429343,
      "loss": 2.9123,
      "step": 438590
    },
    {
      "epoch": 0.12696043823723074,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001122914291429143,
      "loss": 3.1267,
      "step": 438600
    },
    {
      "epoch": 0.12697908147338893,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011228942894289429,
      "loss": 2.9064,
      "step": 438610
    },
    {
      "epoch": 0.12699772470954712,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001122874287428743,
      "loss": 3.0251,
      "step": 438620
    },
    {
      "epoch": 0.1270163679457053,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011228542854285429,
      "loss": 3.0722,
      "step": 438630
    },
    {
      "epoch": 0.1270350111818635,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001122834283428343,
      "loss": 3.0453,
      "step": 438640
    },
    {
      "epoch": 0.12705365441802166,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011228142814281428,
      "loss": 3.2053,
      "step": 438650
    },
    {
      "epoch": 0.12707229765417985,
      "grad_norm": 0.75,
      "learning_rate": 0.00011227942794279429,
      "loss": 3.1326,
      "step": 438660
    },
    {
      "epoch": 0.12709094089033804,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011227742774277427,
      "loss": 3.2171,
      "step": 438670
    },
    {
      "epoch": 0.12710958412649623,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011227542754275428,
      "loss": 3.2444,
      "step": 438680
    },
    {
      "epoch": 0.12712822736265442,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011227342734273427,
      "loss": 3.1818,
      "step": 438690
    },
    {
      "epoch": 0.12714687059881258,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011227142714271427,
      "loss": 3.2046,
      "step": 438700
    },
    {
      "epoch": 0.12716551383497077,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011226942694269427,
      "loss": 2.9956,
      "step": 438710
    },
    {
      "epoch": 0.12718415707112896,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011226742674267428,
      "loss": 3.2142,
      "step": 438720
    },
    {
      "epoch": 0.12720280030728714,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011226542654265426,
      "loss": 3.1779,
      "step": 438730
    },
    {
      "epoch": 0.12722144354344533,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011226342634263427,
      "loss": 3.1729,
      "step": 438740
    },
    {
      "epoch": 0.1272400867796035,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011226142614261425,
      "loss": 3.1716,
      "step": 438750
    },
    {
      "epoch": 0.12725873001576168,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011225942594259426,
      "loss": 3.2861,
      "step": 438760
    },
    {
      "epoch": 0.12727737325191987,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011225742574257427,
      "loss": 3.1445,
      "step": 438770
    },
    {
      "epoch": 0.12729601648807806,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011225542554255425,
      "loss": 3.1729,
      "step": 438780
    },
    {
      "epoch": 0.12731465972423625,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011225342534253426,
      "loss": 3.1302,
      "step": 438790
    },
    {
      "epoch": 0.1273333029603944,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011225142514251426,
      "loss": 3.1379,
      "step": 438800
    },
    {
      "epoch": 0.1273519461965526,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011224942494249427,
      "loss": 3.1652,
      "step": 438810
    },
    {
      "epoch": 0.1273705894327108,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011224742474247425,
      "loss": 3.111,
      "step": 438820
    },
    {
      "epoch": 0.12738923266886898,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011224542454245426,
      "loss": 3.1851,
      "step": 438830
    },
    {
      "epoch": 0.12740787590502717,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011224342434243424,
      "loss": 3.2001,
      "step": 438840
    },
    {
      "epoch": 0.12742651914118536,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011224142414241425,
      "loss": 3.0215,
      "step": 438850
    },
    {
      "epoch": 0.12744516237734352,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011223942394239425,
      "loss": 3.1921,
      "step": 438860
    },
    {
      "epoch": 0.1274638056135017,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011223742374237425,
      "loss": 3.0925,
      "step": 438870
    },
    {
      "epoch": 0.1274824488496599,
      "grad_norm": 0.625,
      "learning_rate": 0.00011223542354235424,
      "loss": 3.2411,
      "step": 438880
    },
    {
      "epoch": 0.12750109208581809,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011223342334233425,
      "loss": 3.117,
      "step": 438890
    },
    {
      "epoch": 0.12751973532197627,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011223142314231423,
      "loss": 3.0391,
      "step": 438900
    },
    {
      "epoch": 0.12753837855813444,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011222942294229424,
      "loss": 3.1201,
      "step": 438910
    },
    {
      "epoch": 0.12755702179429262,
      "grad_norm": 0.75,
      "learning_rate": 0.00011222742274227422,
      "loss": 3.1426,
      "step": 438920
    },
    {
      "epoch": 0.1275756650304508,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011222542254225423,
      "loss": 3.1647,
      "step": 438930
    },
    {
      "epoch": 0.127594308266609,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011222342234223422,
      "loss": 3.2205,
      "step": 438940
    },
    {
      "epoch": 0.1276129515027672,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011222142214221423,
      "loss": 3.0529,
      "step": 438950
    },
    {
      "epoch": 0.12763159473892535,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011221942194219421,
      "loss": 3.0138,
      "step": 438960
    },
    {
      "epoch": 0.12765023797508354,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011221742174217422,
      "loss": 3.0077,
      "step": 438970
    },
    {
      "epoch": 0.12766888121124173,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011221542154215423,
      "loss": 3.0416,
      "step": 438980
    },
    {
      "epoch": 0.12768752444739992,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011221342134213421,
      "loss": 3.1317,
      "step": 438990
    },
    {
      "epoch": 0.1277061676835581,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011221142114211422,
      "loss": 3.205,
      "step": 439000
    },
    {
      "epoch": 0.12772481091971627,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001122094209420942,
      "loss": 3.0783,
      "step": 439010
    },
    {
      "epoch": 0.12774345415587446,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011220742074207422,
      "loss": 3.1376,
      "step": 439020
    },
    {
      "epoch": 0.12776209739203265,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011220542054205421,
      "loss": 3.0594,
      "step": 439030
    },
    {
      "epoch": 0.12778074062819084,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011220342034203422,
      "loss": 3.0,
      "step": 439040
    },
    {
      "epoch": 0.12779938386434903,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001122014201420142,
      "loss": 3.2136,
      "step": 439050
    },
    {
      "epoch": 0.12781802710050721,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011219941994199421,
      "loss": 3.1675,
      "step": 439060
    },
    {
      "epoch": 0.12783667033666538,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001121974197419742,
      "loss": 3.1031,
      "step": 439070
    },
    {
      "epoch": 0.12785531357282356,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001121954195419542,
      "loss": 3.3529,
      "step": 439080
    },
    {
      "epoch": 0.12787395680898175,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001121934193419342,
      "loss": 3.1276,
      "step": 439090
    },
    {
      "epoch": 0.12789260004513994,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001121914191419142,
      "loss": 3.1135,
      "step": 439100
    },
    {
      "epoch": 0.12791124328129813,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011218941894189419,
      "loss": 3.172,
      "step": 439110
    },
    {
      "epoch": 0.1279298865174563,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001121874187418742,
      "loss": 3.1118,
      "step": 439120
    },
    {
      "epoch": 0.12794852975361448,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011218541854185418,
      "loss": 3.0649,
      "step": 439130
    },
    {
      "epoch": 0.12796717298977267,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011218341834183419,
      "loss": 3.2261,
      "step": 439140
    },
    {
      "epoch": 0.12798581622593086,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011218141814181417,
      "loss": 3.1365,
      "step": 439150
    },
    {
      "epoch": 0.12800445946208905,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011217941794179418,
      "loss": 3.1548,
      "step": 439160
    },
    {
      "epoch": 0.1280231026982472,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011217741774177419,
      "loss": 3.0044,
      "step": 439170
    },
    {
      "epoch": 0.1280417459344054,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011217541754175419,
      "loss": 3.0787,
      "step": 439180
    },
    {
      "epoch": 0.1280603891705636,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001121734173417342,
      "loss": 3.1129,
      "step": 439190
    },
    {
      "epoch": 0.12807903240672178,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011217141714171418,
      "loss": 3.0997,
      "step": 439200
    },
    {
      "epoch": 0.12809767564287997,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011216941694169419,
      "loss": 2.9714,
      "step": 439210
    },
    {
      "epoch": 0.12811631887903813,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011216741674167417,
      "loss": 3.098,
      "step": 439220
    },
    {
      "epoch": 0.12813496211519632,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011216541654165418,
      "loss": 3.1009,
      "step": 439230
    },
    {
      "epoch": 0.1281536053513545,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011216341634163416,
      "loss": 3.0723,
      "step": 439240
    },
    {
      "epoch": 0.1281722485875127,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011216141614161417,
      "loss": 3.0288,
      "step": 439250
    },
    {
      "epoch": 0.12819089182367088,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011215941594159416,
      "loss": 3.0645,
      "step": 439260
    },
    {
      "epoch": 0.12820953505982904,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011215741574157417,
      "loss": 3.046,
      "step": 439270
    },
    {
      "epoch": 0.12822817829598723,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011215541554155415,
      "loss": 3.1606,
      "step": 439280
    },
    {
      "epoch": 0.12824682153214542,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011215341534153416,
      "loss": 3.0712,
      "step": 439290
    },
    {
      "epoch": 0.1282654647683036,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011215141514151415,
      "loss": 3.1959,
      "step": 439300
    },
    {
      "epoch": 0.1282841080044618,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011214941494149416,
      "loss": 3.118,
      "step": 439310
    },
    {
      "epoch": 0.12830275124062,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011214741474147414,
      "loss": 3.2091,
      "step": 439320
    },
    {
      "epoch": 0.12832139447677815,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011214541454145415,
      "loss": 3.1714,
      "step": 439330
    },
    {
      "epoch": 0.12834003771293634,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011214341434143414,
      "loss": 3.1526,
      "step": 439340
    },
    {
      "epoch": 0.12835868094909453,
      "grad_norm": 0.921875,
      "learning_rate": 0.00011214141414141415,
      "loss": 3.0852,
      "step": 439350
    },
    {
      "epoch": 0.12837732418525272,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011213941394139416,
      "loss": 3.2091,
      "step": 439360
    },
    {
      "epoch": 0.1283959674214109,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011213741374137414,
      "loss": 3.227,
      "step": 439370
    },
    {
      "epoch": 0.12841461065756907,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011213541354135415,
      "loss": 3.1106,
      "step": 439380
    },
    {
      "epoch": 0.12843325389372726,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011213341334133413,
      "loss": 3.1609,
      "step": 439390
    },
    {
      "epoch": 0.12845189712988545,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011213141314131414,
      "loss": 3.0296,
      "step": 439400
    },
    {
      "epoch": 0.12847054036604363,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011212941294129414,
      "loss": 3.0456,
      "step": 439410
    },
    {
      "epoch": 0.12848918360220182,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011212741274127413,
      "loss": 3.0691,
      "step": 439420
    },
    {
      "epoch": 0.12850782683835998,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011212541254125413,
      "loss": 3.1695,
      "step": 439430
    },
    {
      "epoch": 0.12852647007451817,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011212341234123414,
      "loss": 3.0744,
      "step": 439440
    },
    {
      "epoch": 0.12854511331067636,
      "grad_norm": 0.75,
      "learning_rate": 0.00011212141214121412,
      "loss": 3.1113,
      "step": 439450
    },
    {
      "epoch": 0.12856375654683455,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011211941194119413,
      "loss": 3.3689,
      "step": 439460
    },
    {
      "epoch": 0.12858239978299274,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011211741174117411,
      "loss": 3.0316,
      "step": 439470
    },
    {
      "epoch": 0.1286010430191509,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011211541154115412,
      "loss": 3.1132,
      "step": 439480
    },
    {
      "epoch": 0.1286196862553091,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011211341134113412,
      "loss": 3.1525,
      "step": 439490
    },
    {
      "epoch": 0.12863832949146728,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011211141114111413,
      "loss": 3.1956,
      "step": 439500
    },
    {
      "epoch": 0.12865697272762547,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011210941094109411,
      "loss": 3.0338,
      "step": 439510
    },
    {
      "epoch": 0.12867561596378366,
      "grad_norm": 0.75,
      "learning_rate": 0.00011210741074107412,
      "loss": 3.3098,
      "step": 439520
    },
    {
      "epoch": 0.12869425919994185,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001121054105410541,
      "loss": 3.0713,
      "step": 439530
    },
    {
      "epoch": 0.1287129024361,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011210341034103411,
      "loss": 3.0987,
      "step": 439540
    },
    {
      "epoch": 0.1287315456722582,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011210141014101409,
      "loss": 3.2481,
      "step": 439550
    },
    {
      "epoch": 0.12875018890841639,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001120994099409941,
      "loss": 3.1311,
      "step": 439560
    },
    {
      "epoch": 0.12876883214457457,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011209740974097411,
      "loss": 3.1269,
      "step": 439570
    },
    {
      "epoch": 0.12878747538073276,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001120954095409541,
      "loss": 3.1111,
      "step": 439580
    },
    {
      "epoch": 0.12880611861689092,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011209340934093411,
      "loss": 3.2079,
      "step": 439590
    },
    {
      "epoch": 0.1288247618530491,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001120914091409141,
      "loss": 3.1765,
      "step": 439600
    },
    {
      "epoch": 0.1288434050892073,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001120894089408941,
      "loss": 3.1058,
      "step": 439610
    },
    {
      "epoch": 0.1288620483253655,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011208740874087409,
      "loss": 3.1887,
      "step": 439620
    },
    {
      "epoch": 0.12888069156152368,
      "grad_norm": 0.75,
      "learning_rate": 0.0001120854085408541,
      "loss": 3.0921,
      "step": 439630
    },
    {
      "epoch": 0.12889933479768184,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011208340834083408,
      "loss": 3.1144,
      "step": 439640
    },
    {
      "epoch": 0.12891797803384003,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011208140814081409,
      "loss": 3.2074,
      "step": 439650
    },
    {
      "epoch": 0.12893662126999822,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011207940794079408,
      "loss": 3.0573,
      "step": 439660
    },
    {
      "epoch": 0.1289552645061564,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011207740774077409,
      "loss": 3.0934,
      "step": 439670
    },
    {
      "epoch": 0.1289739077423146,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011207540754075407,
      "loss": 3.2487,
      "step": 439680
    },
    {
      "epoch": 0.12899255097847276,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011207340734073408,
      "loss": 3.2064,
      "step": 439690
    },
    {
      "epoch": 0.12901119421463095,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011207140714071406,
      "loss": 3.2286,
      "step": 439700
    },
    {
      "epoch": 0.12902983745078914,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011206940694069407,
      "loss": 3.119,
      "step": 439710
    },
    {
      "epoch": 0.12904848068694733,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011206740674067407,
      "loss": 3.1916,
      "step": 439720
    },
    {
      "epoch": 0.12906712392310551,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011206540654065408,
      "loss": 3.1642,
      "step": 439730
    },
    {
      "epoch": 0.1290857671592637,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011206340634063406,
      "loss": 3.1808,
      "step": 439740
    },
    {
      "epoch": 0.12910441039542186,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011206140614061407,
      "loss": 3.2108,
      "step": 439750
    },
    {
      "epoch": 0.12912305363158005,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011205940594059408,
      "loss": 3.1958,
      "step": 439760
    },
    {
      "epoch": 0.12914169686773824,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011205740574057406,
      "loss": 3.2413,
      "step": 439770
    },
    {
      "epoch": 0.12916034010389643,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011205540554055407,
      "loss": 3.16,
      "step": 439780
    },
    {
      "epoch": 0.12917898334005462,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011205340534053405,
      "loss": 3.2036,
      "step": 439790
    },
    {
      "epoch": 0.12919762657621278,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011205140514051406,
      "loss": 2.9541,
      "step": 439800
    },
    {
      "epoch": 0.12921626981237097,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011204940494049406,
      "loss": 3.3008,
      "step": 439810
    },
    {
      "epoch": 0.12923491304852916,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011204740474047407,
      "loss": 3.117,
      "step": 439820
    },
    {
      "epoch": 0.12925355628468735,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011204540454045405,
      "loss": 3.227,
      "step": 439830
    },
    {
      "epoch": 0.12927219952084554,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011204340434043406,
      "loss": 3.1048,
      "step": 439840
    },
    {
      "epoch": 0.1292908427570037,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011204140414041404,
      "loss": 2.9698,
      "step": 439850
    },
    {
      "epoch": 0.1293094859931619,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011203940394039405,
      "loss": 3.0367,
      "step": 439860
    },
    {
      "epoch": 0.12932812922932008,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011203740374037403,
      "loss": 3.2226,
      "step": 439870
    },
    {
      "epoch": 0.12934677246547827,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011203540354035404,
      "loss": 3.1376,
      "step": 439880
    },
    {
      "epoch": 0.12936541570163645,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011203340334033404,
      "loss": 3.1984,
      "step": 439890
    },
    {
      "epoch": 0.12938405893779462,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011203140314031404,
      "loss": 3.1499,
      "step": 439900
    },
    {
      "epoch": 0.1294027021739528,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011202940294029403,
      "loss": 3.2733,
      "step": 439910
    },
    {
      "epoch": 0.129421345410111,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011202740274027404,
      "loss": 3.2354,
      "step": 439920
    },
    {
      "epoch": 0.12943998864626918,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011202540254025402,
      "loss": 3.0567,
      "step": 439930
    },
    {
      "epoch": 0.12945863188242737,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011202340234023403,
      "loss": 3.0669,
      "step": 439940
    },
    {
      "epoch": 0.12947727511858556,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011202140214021402,
      "loss": 3.1556,
      "step": 439950
    },
    {
      "epoch": 0.12949591835474372,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011201940194019402,
      "loss": 3.1465,
      "step": 439960
    },
    {
      "epoch": 0.1295145615909019,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011201740174017403,
      "loss": 3.1209,
      "step": 439970
    },
    {
      "epoch": 0.1295332048270601,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011201540154015402,
      "loss": 3.212,
      "step": 439980
    },
    {
      "epoch": 0.1295518480632183,
      "grad_norm": 0.75,
      "learning_rate": 0.00011201340134013403,
      "loss": 3.1464,
      "step": 439990
    },
    {
      "epoch": 0.12957049129937648,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011201140114011401,
      "loss": 3.0902,
      "step": 440000
    },
    {
      "epoch": 0.12958913453553464,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011200940094009402,
      "loss": 3.0596,
      "step": 440010
    },
    {
      "epoch": 0.12960777777169283,
      "grad_norm": 0.78125,
      "learning_rate": 0.000112007400740074,
      "loss": 3.1985,
      "step": 440020
    },
    {
      "epoch": 0.12962642100785102,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011200540054005401,
      "loss": 3.1808,
      "step": 440030
    },
    {
      "epoch": 0.1296450642440092,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011200340034003401,
      "loss": 3.0537,
      "step": 440040
    },
    {
      "epoch": 0.1296637074801674,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011200140014001402,
      "loss": 3.2138,
      "step": 440050
    },
    {
      "epoch": 0.12968235071632556,
      "grad_norm": 0.78125,
      "learning_rate": 0.000111999399939994,
      "loss": 3.1601,
      "step": 440060
    },
    {
      "epoch": 0.12970099395248375,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011199739973997401,
      "loss": 3.191,
      "step": 440070
    },
    {
      "epoch": 0.12971963718864193,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011199539953995399,
      "loss": 3.1123,
      "step": 440080
    },
    {
      "epoch": 0.12973828042480012,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000111993399339934,
      "loss": 3.1815,
      "step": 440090
    },
    {
      "epoch": 0.1297569236609583,
      "grad_norm": 0.75,
      "learning_rate": 0.00011199139913991398,
      "loss": 3.1938,
      "step": 440100
    },
    {
      "epoch": 0.12977556689711647,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011198939893989399,
      "loss": 3.0984,
      "step": 440110
    },
    {
      "epoch": 0.12979421013327466,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011198739873987399,
      "loss": 3.0697,
      "step": 440120
    },
    {
      "epoch": 0.12981285336943285,
      "grad_norm": 0.82421875,
      "learning_rate": 0.000111985398539854,
      "loss": 3.131,
      "step": 440130
    },
    {
      "epoch": 0.12983149660559104,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011198339833983398,
      "loss": 3.1796,
      "step": 440140
    },
    {
      "epoch": 0.12985013984174923,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011198139813981399,
      "loss": 3.0485,
      "step": 440150
    },
    {
      "epoch": 0.1298687830779074,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000111979397939794,
      "loss": 3.2845,
      "step": 440160
    },
    {
      "epoch": 0.12988742631406558,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011197739773977398,
      "loss": 3.1382,
      "step": 440170
    },
    {
      "epoch": 0.12990606955022377,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011197539753975399,
      "loss": 3.0964,
      "step": 440180
    },
    {
      "epoch": 0.12992471278638196,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011197339733973397,
      "loss": 3.2785,
      "step": 440190
    },
    {
      "epoch": 0.12994335602254015,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011197139713971398,
      "loss": 3.1897,
      "step": 440200
    },
    {
      "epoch": 0.12996199925869834,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011196939693969398,
      "loss": 3.2449,
      "step": 440210
    },
    {
      "epoch": 0.1299806424948565,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011196739673967398,
      "loss": 3.0609,
      "step": 440220
    },
    {
      "epoch": 0.12999928573101469,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011196539653965397,
      "loss": 3.0086,
      "step": 440230
    },
    {
      "epoch": 0.13001792896717287,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011196339633963398,
      "loss": 3.2033,
      "step": 440240
    },
    {
      "epoch": 0.13003657220333106,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011196139613961396,
      "loss": 3.1457,
      "step": 440250
    },
    {
      "epoch": 0.13005521543948925,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011195939593959397,
      "loss": 3.2365,
      "step": 440260
    },
    {
      "epoch": 0.1300738586756474,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011195739573957395,
      "loss": 3.1184,
      "step": 440270
    },
    {
      "epoch": 0.1300925019118056,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011195539553955396,
      "loss": 3.2804,
      "step": 440280
    },
    {
      "epoch": 0.1301111451479638,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011195339533953395,
      "loss": 3.1035,
      "step": 440290
    },
    {
      "epoch": 0.13012978838412198,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011195139513951396,
      "loss": 3.1745,
      "step": 440300
    },
    {
      "epoch": 0.13014843162028017,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011194939493949395,
      "loss": 3.1028,
      "step": 440310
    },
    {
      "epoch": 0.13016707485643833,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011194739473947395,
      "loss": 3.2177,
      "step": 440320
    },
    {
      "epoch": 0.13018571809259652,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011194539453945394,
      "loss": 3.1255,
      "step": 440330
    },
    {
      "epoch": 0.1302043613287547,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011194339433943395,
      "loss": 3.1647,
      "step": 440340
    },
    {
      "epoch": 0.1302230045649129,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011194139413941395,
      "loss": 3.069,
      "step": 440350
    },
    {
      "epoch": 0.1302416478010711,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011193939393939395,
      "loss": 2.9968,
      "step": 440360
    },
    {
      "epoch": 0.13026029103722925,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011193739373937395,
      "loss": 3.0942,
      "step": 440370
    },
    {
      "epoch": 0.13027893427338744,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011193539353935394,
      "loss": 3.2051,
      "step": 440380
    },
    {
      "epoch": 0.13029757750954563,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011193339333933395,
      "loss": 3.2274,
      "step": 440390
    },
    {
      "epoch": 0.13031622074570381,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011193139313931393,
      "loss": 3.0941,
      "step": 440400
    },
    {
      "epoch": 0.130334863981862,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011192939293929394,
      "loss": 3.0165,
      "step": 440410
    },
    {
      "epoch": 0.1303535072180202,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011192739273927392,
      "loss": 3.0383,
      "step": 440420
    },
    {
      "epoch": 0.13037215045417835,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011192539253925393,
      "loss": 3.1645,
      "step": 440430
    },
    {
      "epoch": 0.13039079369033654,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011192339233923393,
      "loss": 3.2696,
      "step": 440440
    },
    {
      "epoch": 0.13040943692649473,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011192139213921394,
      "loss": 3.0887,
      "step": 440450
    },
    {
      "epoch": 0.13042808016265292,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011191939193919392,
      "loss": 2.9564,
      "step": 440460
    },
    {
      "epoch": 0.1304467233988111,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011191739173917393,
      "loss": 3.2668,
      "step": 440470
    },
    {
      "epoch": 0.13046536663496927,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011191539153915391,
      "loss": 3.1402,
      "step": 440480
    },
    {
      "epoch": 0.13048400987112746,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011191339133913392,
      "loss": 3.3571,
      "step": 440490
    },
    {
      "epoch": 0.13050265310728565,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001119113911391139,
      "loss": 3.028,
      "step": 440500
    },
    {
      "epoch": 0.13052129634344384,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011190939093909391,
      "loss": 3.1518,
      "step": 440510
    },
    {
      "epoch": 0.13053993957960203,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011190739073907391,
      "loss": 3.1494,
      "step": 440520
    },
    {
      "epoch": 0.1305585828157602,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011190539053905392,
      "loss": 3.0748,
      "step": 440530
    },
    {
      "epoch": 0.13057722605191838,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001119033903390339,
      "loss": 2.9938,
      "step": 440540
    },
    {
      "epoch": 0.13059586928807657,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011190139013901391,
      "loss": 3.0461,
      "step": 440550
    },
    {
      "epoch": 0.13061451252423475,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011189938993899392,
      "loss": 3.0401,
      "step": 440560
    },
    {
      "epoch": 0.13063315576039294,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001118973897389739,
      "loss": 3.1085,
      "step": 440570
    },
    {
      "epoch": 0.1306517989965511,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011189538953895391,
      "loss": 3.1557,
      "step": 440580
    },
    {
      "epoch": 0.1306704422327093,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011189338933893389,
      "loss": 3.0154,
      "step": 440590
    },
    {
      "epoch": 0.13068908546886748,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001118913891389139,
      "loss": 3.1941,
      "step": 440600
    },
    {
      "epoch": 0.13070772870502567,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001118893889388939,
      "loss": 3.0001,
      "step": 440610
    },
    {
      "epoch": 0.13072637194118386,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001118873887388739,
      "loss": 3.0849,
      "step": 440620
    },
    {
      "epoch": 0.13074501517734205,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011188538853885389,
      "loss": 3.1364,
      "step": 440630
    },
    {
      "epoch": 0.1307636584135002,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001118833883388339,
      "loss": 3.1783,
      "step": 440640
    },
    {
      "epoch": 0.1307823016496584,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011188138813881388,
      "loss": 3.1109,
      "step": 440650
    },
    {
      "epoch": 0.1308009448858166,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011187938793879389,
      "loss": 3.2695,
      "step": 440660
    },
    {
      "epoch": 0.13081958812197478,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011187738773877388,
      "loss": 3.0499,
      "step": 440670
    },
    {
      "epoch": 0.13083823135813297,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011187538753875389,
      "loss": 3.1806,
      "step": 440680
    },
    {
      "epoch": 0.13085687459429113,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011187338733873387,
      "loss": 3.2888,
      "step": 440690
    },
    {
      "epoch": 0.13087551783044932,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011187138713871388,
      "loss": 3.1693,
      "step": 440700
    },
    {
      "epoch": 0.1308941610666075,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011186938693869386,
      "loss": 3.1219,
      "step": 440710
    },
    {
      "epoch": 0.1309128043027657,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011186738673867387,
      "loss": 3.0547,
      "step": 440720
    },
    {
      "epoch": 0.13093144753892388,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011186538653865385,
      "loss": 3.0953,
      "step": 440730
    },
    {
      "epoch": 0.13095009077508205,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011186338633863386,
      "loss": 2.9719,
      "step": 440740
    },
    {
      "epoch": 0.13096873401124023,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011186138613861387,
      "loss": 3.1774,
      "step": 440750
    },
    {
      "epoch": 0.13098737724739842,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011185938593859387,
      "loss": 3.1363,
      "step": 440760
    },
    {
      "epoch": 0.1310060204835566,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011185738573857388,
      "loss": 3.2077,
      "step": 440770
    },
    {
      "epoch": 0.1310246637197148,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011185538553855386,
      "loss": 3.048,
      "step": 440780
    },
    {
      "epoch": 0.13104330695587296,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011185338533853387,
      "loss": 3.1945,
      "step": 440790
    },
    {
      "epoch": 0.13106195019203115,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011185138513851385,
      "loss": 3.2611,
      "step": 440800
    },
    {
      "epoch": 0.13108059342818934,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011184938493849386,
      "loss": 3.0954,
      "step": 440810
    },
    {
      "epoch": 0.13109923666434753,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011184738473847384,
      "loss": 3.2126,
      "step": 440820
    },
    {
      "epoch": 0.13111787990050572,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011184538453845385,
      "loss": 2.8481,
      "step": 440830
    },
    {
      "epoch": 0.13113652313666388,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011184338433843385,
      "loss": 3.2352,
      "step": 440840
    },
    {
      "epoch": 0.13115516637282207,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011184138413841386,
      "loss": 3.2757,
      "step": 440850
    },
    {
      "epoch": 0.13117380960898026,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011183938393839384,
      "loss": 3.2321,
      "step": 440860
    },
    {
      "epoch": 0.13119245284513845,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011183738373837385,
      "loss": 3.2817,
      "step": 440870
    },
    {
      "epoch": 0.13121109608129664,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011183538353835383,
      "loss": 3.217,
      "step": 440880
    },
    {
      "epoch": 0.13122973931745482,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011183338333833384,
      "loss": 3.0927,
      "step": 440890
    },
    {
      "epoch": 0.13124838255361299,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011183138313831383,
      "loss": 3.1075,
      "step": 440900
    },
    {
      "epoch": 0.13126702578977117,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011182938293829383,
      "loss": 3.068,
      "step": 440910
    },
    {
      "epoch": 0.13128566902592936,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011182738273827383,
      "loss": 3.1074,
      "step": 440920
    },
    {
      "epoch": 0.13130431226208755,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011182538253825383,
      "loss": 3.1268,
      "step": 440930
    },
    {
      "epoch": 0.13132295549824574,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011182338233823384,
      "loss": 3.1709,
      "step": 440940
    },
    {
      "epoch": 0.1313415987344039,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011182138213821383,
      "loss": 3.0553,
      "step": 440950
    },
    {
      "epoch": 0.1313602419705621,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011181938193819383,
      "loss": 3.034,
      "step": 440960
    },
    {
      "epoch": 0.13137888520672028,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011181738173817382,
      "loss": 3.0325,
      "step": 440970
    },
    {
      "epoch": 0.13139752844287847,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011181538153815383,
      "loss": 3.1487,
      "step": 440980
    },
    {
      "epoch": 0.13141617167903666,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011181338133813382,
      "loss": 3.117,
      "step": 440990
    },
    {
      "epoch": 0.13143481491519482,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011181138113811383,
      "loss": 3.2642,
      "step": 441000
    },
    {
      "epoch": 0.131453458151353,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011180938093809381,
      "loss": 3.3396,
      "step": 441010
    },
    {
      "epoch": 0.1314721013875112,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011180738073807382,
      "loss": 3.0661,
      "step": 441020
    },
    {
      "epoch": 0.1314907446236694,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001118053805380538,
      "loss": 3.0973,
      "step": 441030
    },
    {
      "epoch": 0.13150938785982758,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011180338033803381,
      "loss": 3.0152,
      "step": 441040
    },
    {
      "epoch": 0.13152803109598574,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001118013801380138,
      "loss": 3.0746,
      "step": 441050
    },
    {
      "epoch": 0.13154667433214393,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001117993799379938,
      "loss": 3.1336,
      "step": 441060
    },
    {
      "epoch": 0.13156531756830211,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001117973797379738,
      "loss": 3.1977,
      "step": 441070
    },
    {
      "epoch": 0.1315839608044603,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011179537953795381,
      "loss": 3.0854,
      "step": 441080
    },
    {
      "epoch": 0.1316026040406185,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011179337933793379,
      "loss": 3.2399,
      "step": 441090
    },
    {
      "epoch": 0.13162124727677668,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001117913791379138,
      "loss": 3.1234,
      "step": 441100
    },
    {
      "epoch": 0.13163989051293484,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011178937893789378,
      "loss": 3.0907,
      "step": 441110
    },
    {
      "epoch": 0.13165853374909303,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011178737873787379,
      "loss": 3.1048,
      "step": 441120
    },
    {
      "epoch": 0.13167717698525122,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011178537853785377,
      "loss": 3.0895,
      "step": 441130
    },
    {
      "epoch": 0.1316958202214094,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011178337833783378,
      "loss": 3.0776,
      "step": 441140
    },
    {
      "epoch": 0.1317144634575676,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011178137813781379,
      "loss": 3.1566,
      "step": 441150
    },
    {
      "epoch": 0.13173310669372576,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011177937793779379,
      "loss": 2.8634,
      "step": 441160
    },
    {
      "epoch": 0.13175174992988395,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001117773777377738,
      "loss": 3.2752,
      "step": 441170
    },
    {
      "epoch": 0.13177039316604214,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011177537753775378,
      "loss": 3.2087,
      "step": 441180
    },
    {
      "epoch": 0.13178903640220033,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011177337733773379,
      "loss": 3.2074,
      "step": 441190
    },
    {
      "epoch": 0.13180767963835852,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011177137713771377,
      "loss": 3.0967,
      "step": 441200
    },
    {
      "epoch": 0.13182632287451668,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011176937693769378,
      "loss": 3.1628,
      "step": 441210
    },
    {
      "epoch": 0.13184496611067487,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011176737673767377,
      "loss": 3.0942,
      "step": 441220
    },
    {
      "epoch": 0.13186360934683306,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011176537653765377,
      "loss": 3.061,
      "step": 441230
    },
    {
      "epoch": 0.13188225258299124,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011176337633763377,
      "loss": 3.1331,
      "step": 441240
    },
    {
      "epoch": 0.13190089581914943,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011176137613761377,
      "loss": 3.1838,
      "step": 441250
    },
    {
      "epoch": 0.1319195390553076,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011175937593759376,
      "loss": 3.094,
      "step": 441260
    },
    {
      "epoch": 0.13193818229146578,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011175737573757377,
      "loss": 3.1541,
      "step": 441270
    },
    {
      "epoch": 0.13195682552762397,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011175537553755375,
      "loss": 3.202,
      "step": 441280
    },
    {
      "epoch": 0.13197546876378216,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011175337533753376,
      "loss": 3.2018,
      "step": 441290
    },
    {
      "epoch": 0.13199411199994035,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011175137513751375,
      "loss": 3.1469,
      "step": 441300
    },
    {
      "epoch": 0.13201275523609854,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011174937493749376,
      "loss": 3.0654,
      "step": 441310
    },
    {
      "epoch": 0.1320313984722567,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00011174737473747374,
      "loss": 3.1753,
      "step": 441320
    },
    {
      "epoch": 0.1320500417084149,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011174537453745375,
      "loss": 3.0709,
      "step": 441330
    },
    {
      "epoch": 0.13206868494457308,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011174337433743376,
      "loss": 2.9088,
      "step": 441340
    },
    {
      "epoch": 0.13208732818073127,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011174137413741374,
      "loss": 3.1309,
      "step": 441350
    },
    {
      "epoch": 0.13210597141688946,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011173937393739375,
      "loss": 3.1394,
      "step": 441360
    },
    {
      "epoch": 0.13212461465304762,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011173737373737374,
      "loss": 3.1746,
      "step": 441370
    },
    {
      "epoch": 0.1321432578892058,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011173537353735374,
      "loss": 3.078,
      "step": 441380
    },
    {
      "epoch": 0.132161901125364,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011173337333733374,
      "loss": 3.0322,
      "step": 441390
    },
    {
      "epoch": 0.13218054436152218,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011173137313731375,
      "loss": 3.1719,
      "step": 441400
    },
    {
      "epoch": 0.13219918759768037,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011172937293729373,
      "loss": 3.2047,
      "step": 441410
    },
    {
      "epoch": 0.13221783083383853,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011172737273727374,
      "loss": 3.0739,
      "step": 441420
    },
    {
      "epoch": 0.13223647406999672,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011172537253725372,
      "loss": 3.1764,
      "step": 441430
    },
    {
      "epoch": 0.1322551173061549,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011172337233723373,
      "loss": 3.1118,
      "step": 441440
    },
    {
      "epoch": 0.1322737605423131,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011172137213721371,
      "loss": 3.1025,
      "step": 441450
    },
    {
      "epoch": 0.1322924037784713,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011171937193719372,
      "loss": 3.1453,
      "step": 441460
    },
    {
      "epoch": 0.13231104701462945,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011171737173717372,
      "loss": 3.1143,
      "step": 441470
    },
    {
      "epoch": 0.13232969025078764,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011171537153715373,
      "loss": 3.1555,
      "step": 441480
    },
    {
      "epoch": 0.13234833348694583,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011171337133713371,
      "loss": 3.2623,
      "step": 441490
    },
    {
      "epoch": 0.13236697672310402,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011171137113711372,
      "loss": 3.0086,
      "step": 441500
    },
    {
      "epoch": 0.1323856199592622,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001117093709370937,
      "loss": 3.0476,
      "step": 441510
    },
    {
      "epoch": 0.13240426319542037,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011170737073707371,
      "loss": 3.1812,
      "step": 441520
    },
    {
      "epoch": 0.13242290643157856,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011170537053705372,
      "loss": 3.0262,
      "step": 441530
    },
    {
      "epoch": 0.13244154966773675,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001117033703370337,
      "loss": 3.1573,
      "step": 441540
    },
    {
      "epoch": 0.13246019290389494,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011170137013701371,
      "loss": 3.0777,
      "step": 441550
    },
    {
      "epoch": 0.13247883614005312,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001116993699369937,
      "loss": 3.2354,
      "step": 441560
    },
    {
      "epoch": 0.1324974793762113,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011169736973697372,
      "loss": 3.0484,
      "step": 441570
    },
    {
      "epoch": 0.13251612261236947,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001116953695369537,
      "loss": 3.1814,
      "step": 441580
    },
    {
      "epoch": 0.13253476584852766,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001116933693369337,
      "loss": 3.0736,
      "step": 441590
    },
    {
      "epoch": 0.13255340908468585,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011169136913691369,
      "loss": 3.2092,
      "step": 441600
    },
    {
      "epoch": 0.13257205232084404,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001116893689368937,
      "loss": 3.1192,
      "step": 441610
    },
    {
      "epoch": 0.13259069555700223,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001116873687368737,
      "loss": 3.0601,
      "step": 441620
    },
    {
      "epoch": 0.1326093387931604,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001116853685368537,
      "loss": 3.065,
      "step": 441630
    },
    {
      "epoch": 0.13262798202931858,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011168336833683368,
      "loss": 2.9678,
      "step": 441640
    },
    {
      "epoch": 0.13264662526547677,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001116813681368137,
      "loss": 3.1064,
      "step": 441650
    },
    {
      "epoch": 0.13266526850163496,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011167936793679368,
      "loss": 2.9853,
      "step": 441660
    },
    {
      "epoch": 0.13268391173779315,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011167736773677368,
      "loss": 3.085,
      "step": 441670
    },
    {
      "epoch": 0.1327025549739513,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011167536753675367,
      "loss": 3.2329,
      "step": 441680
    },
    {
      "epoch": 0.1327211982101095,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011167336733673368,
      "loss": 3.1144,
      "step": 441690
    },
    {
      "epoch": 0.1327398414462677,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011167136713671367,
      "loss": 3.1114,
      "step": 441700
    },
    {
      "epoch": 0.13275848468242588,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011166936693669368,
      "loss": 3.1682,
      "step": 441710
    },
    {
      "epoch": 0.13277712791858406,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011166736673667366,
      "loss": 3.047,
      "step": 441720
    },
    {
      "epoch": 0.13279577115474223,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011166536653665367,
      "loss": 3.1692,
      "step": 441730
    },
    {
      "epoch": 0.13281441439090041,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011166336633663368,
      "loss": 3.0709,
      "step": 441740
    },
    {
      "epoch": 0.1328330576270586,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011166136613661366,
      "loss": 3.1324,
      "step": 441750
    },
    {
      "epoch": 0.1328517008632168,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011165936593659367,
      "loss": 3.056,
      "step": 441760
    },
    {
      "epoch": 0.13287034409937498,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011165736573657365,
      "loss": 3.1452,
      "step": 441770
    },
    {
      "epoch": 0.13288898733553317,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011165536553655366,
      "loss": 3.142,
      "step": 441780
    },
    {
      "epoch": 0.13290763057169133,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011165336533653366,
      "loss": 3.1434,
      "step": 441790
    },
    {
      "epoch": 0.13292627380784952,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011165136513651367,
      "loss": 3.2116,
      "step": 441800
    },
    {
      "epoch": 0.1329449170440077,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011164936493649365,
      "loss": 3.0642,
      "step": 441810
    },
    {
      "epoch": 0.1329635602801659,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011164736473647366,
      "loss": 3.1256,
      "step": 441820
    },
    {
      "epoch": 0.1329822035163241,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011164536453645364,
      "loss": 3.1439,
      "step": 441830
    },
    {
      "epoch": 0.13300084675248225,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011164336433643365,
      "loss": 3.0448,
      "step": 441840
    },
    {
      "epoch": 0.13301948998864044,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011164136413641365,
      "loss": 3.0667,
      "step": 441850
    },
    {
      "epoch": 0.13303813322479863,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011163936393639364,
      "loss": 3.1743,
      "step": 441860
    },
    {
      "epoch": 0.13305677646095682,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011163736373637364,
      "loss": 3.1379,
      "step": 441870
    },
    {
      "epoch": 0.133075419697115,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011163536353635365,
      "loss": 3.1796,
      "step": 441880
    },
    {
      "epoch": 0.13309406293327317,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011163336333633363,
      "loss": 3.0869,
      "step": 441890
    },
    {
      "epoch": 0.13311270616943136,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011163136313631364,
      "loss": 3.2248,
      "step": 441900
    },
    {
      "epoch": 0.13313134940558954,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011162936293629362,
      "loss": 3.1359,
      "step": 441910
    },
    {
      "epoch": 0.13314999264174773,
      "grad_norm": 0.75,
      "learning_rate": 0.00011162736273627363,
      "loss": 3.2114,
      "step": 441920
    },
    {
      "epoch": 0.13316863587790592,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011162536253625364,
      "loss": 3.0689,
      "step": 441930
    },
    {
      "epoch": 0.13318727911406408,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011162336233623363,
      "loss": 3.1618,
      "step": 441940
    },
    {
      "epoch": 0.13320592235022227,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011162136213621364,
      "loss": 3.0851,
      "step": 441950
    },
    {
      "epoch": 0.13322456558638046,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011161936193619362,
      "loss": 2.9353,
      "step": 441960
    },
    {
      "epoch": 0.13324320882253865,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011161736173617363,
      "loss": 3.1358,
      "step": 441970
    },
    {
      "epoch": 0.13326185205869684,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011161536153615362,
      "loss": 3.1789,
      "step": 441980
    },
    {
      "epoch": 0.13328049529485503,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011161336133613363,
      "loss": 3.1547,
      "step": 441990
    },
    {
      "epoch": 0.1332991385310132,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011161136113611361,
      "loss": 3.112,
      "step": 442000
    },
    {
      "epoch": 0.13331778176717138,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011160936093609362,
      "loss": 3.2172,
      "step": 442010
    },
    {
      "epoch": 0.13333642500332957,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011160736073607361,
      "loss": 3.1411,
      "step": 442020
    },
    {
      "epoch": 0.13335506823948776,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011160536053605362,
      "loss": 3.2738,
      "step": 442030
    },
    {
      "epoch": 0.13337371147564595,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001116033603360336,
      "loss": 3.1433,
      "step": 442040
    },
    {
      "epoch": 0.1333923547118041,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011160136013601361,
      "loss": 3.1461,
      "step": 442050
    },
    {
      "epoch": 0.1334109979479623,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001115993599359936,
      "loss": 3.1519,
      "step": 442060
    },
    {
      "epoch": 0.13342964118412048,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001115973597359736,
      "loss": 3.1109,
      "step": 442070
    },
    {
      "epoch": 0.13344828442027867,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011159535953595359,
      "loss": 3.0879,
      "step": 442080
    },
    {
      "epoch": 0.13346692765643686,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001115933593359336,
      "loss": 2.9988,
      "step": 442090
    },
    {
      "epoch": 0.13348557089259502,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011159135913591359,
      "loss": 3.1113,
      "step": 442100
    },
    {
      "epoch": 0.1335042141287532,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001115893589358936,
      "loss": 3.0727,
      "step": 442110
    },
    {
      "epoch": 0.1335228573649114,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011158735873587358,
      "loss": 3.1705,
      "step": 442120
    },
    {
      "epoch": 0.1335415006010696,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011158535853585359,
      "loss": 3.121,
      "step": 442130
    },
    {
      "epoch": 0.13356014383722778,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001115833583358336,
      "loss": 3.0152,
      "step": 442140
    },
    {
      "epoch": 0.13357878707338594,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011158135813581358,
      "loss": 3.0484,
      "step": 442150
    },
    {
      "epoch": 0.13359743030954413,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011157935793579359,
      "loss": 3.0834,
      "step": 442160
    },
    {
      "epoch": 0.13361607354570232,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011157735773577359,
      "loss": 3.0186,
      "step": 442170
    },
    {
      "epoch": 0.1336347167818605,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011157535753575358,
      "loss": 3.1752,
      "step": 442180
    },
    {
      "epoch": 0.1336533600180187,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011157335733573358,
      "loss": 3.087,
      "step": 442190
    },
    {
      "epoch": 0.13367200325417686,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011157135713571359,
      "loss": 3.1923,
      "step": 442200
    },
    {
      "epoch": 0.13369064649033505,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011156935693569357,
      "loss": 3.2081,
      "step": 442210
    },
    {
      "epoch": 0.13370928972649324,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011156735673567358,
      "loss": 3.047,
      "step": 442220
    },
    {
      "epoch": 0.13372793296265142,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011156535653565356,
      "loss": 3.16,
      "step": 442230
    },
    {
      "epoch": 0.1337465761988096,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011156335633563357,
      "loss": 2.9848,
      "step": 442240
    },
    {
      "epoch": 0.1337652194349678,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011156135613561356,
      "loss": 3.0915,
      "step": 442250
    },
    {
      "epoch": 0.13378386267112596,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00011155935593559357,
      "loss": 3.1662,
      "step": 442260
    },
    {
      "epoch": 0.13380250590728415,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011155735573557356,
      "loss": 3.0249,
      "step": 442270
    },
    {
      "epoch": 0.13382114914344234,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011155535553555357,
      "loss": 3.2597,
      "step": 442280
    },
    {
      "epoch": 0.13383979237960053,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011155335533553355,
      "loss": 3.2574,
      "step": 442290
    },
    {
      "epoch": 0.13385843561575872,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011155135513551356,
      "loss": 3.1826,
      "step": 442300
    },
    {
      "epoch": 0.13387707885191688,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011154935493549354,
      "loss": 3.1307,
      "step": 442310
    },
    {
      "epoch": 0.13389572208807507,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011154735473547355,
      "loss": 3.0268,
      "step": 442320
    },
    {
      "epoch": 0.13391436532423326,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011154535453545356,
      "loss": 3.1315,
      "step": 442330
    },
    {
      "epoch": 0.13393300856039145,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011154335433543355,
      "loss": 3.026,
      "step": 442340
    },
    {
      "epoch": 0.13395165179654964,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011154135413541356,
      "loss": 3.2093,
      "step": 442350
    },
    {
      "epoch": 0.1339702950327078,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011153935393539354,
      "loss": 3.2155,
      "step": 442360
    },
    {
      "epoch": 0.133988938268866,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011153735373537355,
      "loss": 3.0918,
      "step": 442370
    },
    {
      "epoch": 0.13400758150502418,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011153535353535353,
      "loss": 3.1415,
      "step": 442380
    },
    {
      "epoch": 0.13402622474118236,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011153335333533354,
      "loss": 3.0474,
      "step": 442390
    },
    {
      "epoch": 0.13404486797734055,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011153135313531353,
      "loss": 3.1315,
      "step": 442400
    },
    {
      "epoch": 0.13406351121349872,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011152935293529353,
      "loss": 3.1474,
      "step": 442410
    },
    {
      "epoch": 0.1340821544496569,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011152735273527353,
      "loss": 3.1349,
      "step": 442420
    },
    {
      "epoch": 0.1341007976858151,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011152535253525354,
      "loss": 3.0187,
      "step": 442430
    },
    {
      "epoch": 0.13411944092197328,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011152335233523352,
      "loss": 3.0801,
      "step": 442440
    },
    {
      "epoch": 0.13413808415813147,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011152135213521353,
      "loss": 3.1281,
      "step": 442450
    },
    {
      "epoch": 0.13415672739428966,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011151935193519351,
      "loss": 3.2659,
      "step": 442460
    },
    {
      "epoch": 0.13417537063044782,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011151735173517352,
      "loss": 3.1498,
      "step": 442470
    },
    {
      "epoch": 0.134194013866606,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011151535153515352,
      "loss": 3.0689,
      "step": 442480
    },
    {
      "epoch": 0.1342126571027642,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011151335133513353,
      "loss": 3.1175,
      "step": 442490
    },
    {
      "epoch": 0.1342313003389224,
      "grad_norm": 0.75,
      "learning_rate": 0.00011151135113511351,
      "loss": 2.9514,
      "step": 442500
    },
    {
      "epoch": 0.13424994357508058,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011150935093509352,
      "loss": 3.1127,
      "step": 442510
    },
    {
      "epoch": 0.13426858681123874,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011150735073507353,
      "loss": 3.0619,
      "step": 442520
    },
    {
      "epoch": 0.13428723004739693,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011150535053505351,
      "loss": 3.1416,
      "step": 442530
    },
    {
      "epoch": 0.13430587328355512,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011150335033503352,
      "loss": 3.1164,
      "step": 442540
    },
    {
      "epoch": 0.1343245165197133,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001115013501350135,
      "loss": 3.1088,
      "step": 442550
    },
    {
      "epoch": 0.1343431597558715,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011149934993499351,
      "loss": 3.16,
      "step": 442560
    },
    {
      "epoch": 0.13436180299202966,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001114973497349735,
      "loss": 3.1793,
      "step": 442570
    },
    {
      "epoch": 0.13438044622818784,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011149534953495351,
      "loss": 3.0951,
      "step": 442580
    },
    {
      "epoch": 0.13439908946434603,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001114933493349335,
      "loss": 3.2802,
      "step": 442590
    },
    {
      "epoch": 0.13441773270050422,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001114913491349135,
      "loss": 3.1109,
      "step": 442600
    },
    {
      "epoch": 0.1344363759366624,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011148934893489349,
      "loss": 2.9911,
      "step": 442610
    },
    {
      "epoch": 0.13445501917282057,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001114873487348735,
      "loss": 3.1558,
      "step": 442620
    },
    {
      "epoch": 0.13447366240897876,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011148534853485348,
      "loss": 3.0553,
      "step": 442630
    },
    {
      "epoch": 0.13449230564513695,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011148334833483349,
      "loss": 3.0495,
      "step": 442640
    },
    {
      "epoch": 0.13451094888129514,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011148134813481348,
      "loss": 3.0365,
      "step": 442650
    },
    {
      "epoch": 0.13452959211745333,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011147934793479349,
      "loss": 3.0944,
      "step": 442660
    },
    {
      "epoch": 0.13454823535361152,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011147734773477347,
      "loss": 3.0737,
      "step": 442670
    },
    {
      "epoch": 0.13456687858976968,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011147534753475348,
      "loss": 3.1198,
      "step": 442680
    },
    {
      "epoch": 0.13458552182592787,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011147334733473347,
      "loss": 3.179,
      "step": 442690
    },
    {
      "epoch": 0.13460416506208606,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011147134713471347,
      "loss": 3.1642,
      "step": 442700
    },
    {
      "epoch": 0.13462280829824425,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011146934693469347,
      "loss": 3.0283,
      "step": 442710
    },
    {
      "epoch": 0.13464145153440243,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011146734673467347,
      "loss": 3.0621,
      "step": 442720
    },
    {
      "epoch": 0.1346600947705606,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011146534653465348,
      "loss": 3.1767,
      "step": 442730
    },
    {
      "epoch": 0.13467873800671878,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011146334633463347,
      "loss": 3.1553,
      "step": 442740
    },
    {
      "epoch": 0.13469738124287697,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011146134613461348,
      "loss": 3.1748,
      "step": 442750
    },
    {
      "epoch": 0.13471602447903516,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011145934593459346,
      "loss": 3.2356,
      "step": 442760
    },
    {
      "epoch": 0.13473466771519335,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011145734573457347,
      "loss": 3.0736,
      "step": 442770
    },
    {
      "epoch": 0.1347533109513515,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011145534553455345,
      "loss": 3.1893,
      "step": 442780
    },
    {
      "epoch": 0.1347719541875097,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011145334533453346,
      "loss": 3.039,
      "step": 442790
    },
    {
      "epoch": 0.1347905974236679,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011145134513451346,
      "loss": 3.1364,
      "step": 442800
    },
    {
      "epoch": 0.13480924065982608,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011144934493449347,
      "loss": 3.1568,
      "step": 442810
    },
    {
      "epoch": 0.13482788389598427,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011144734473447345,
      "loss": 3.0467,
      "step": 442820
    },
    {
      "epoch": 0.13484652713214243,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011144534453445346,
      "loss": 3.1345,
      "step": 442830
    },
    {
      "epoch": 0.13486517036830062,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011144334433443344,
      "loss": 3.2044,
      "step": 442840
    },
    {
      "epoch": 0.1348838136044588,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011144134413441345,
      "loss": 3.1025,
      "step": 442850
    },
    {
      "epoch": 0.134902456840617,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011143934393439343,
      "loss": 3.1501,
      "step": 442860
    },
    {
      "epoch": 0.13492110007677519,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011143734373437344,
      "loss": 3.1164,
      "step": 442870
    },
    {
      "epoch": 0.13493974331293335,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011143534353435344,
      "loss": 3.1798,
      "step": 442880
    },
    {
      "epoch": 0.13495838654909154,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011143334333433345,
      "loss": 3.1417,
      "step": 442890
    },
    {
      "epoch": 0.13497702978524972,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011143134313431343,
      "loss": 3.1093,
      "step": 442900
    },
    {
      "epoch": 0.1349956730214079,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011142934293429344,
      "loss": 3.0654,
      "step": 442910
    },
    {
      "epoch": 0.1350143162575661,
      "grad_norm": 0.75,
      "learning_rate": 0.00011142734273427345,
      "loss": 3.119,
      "step": 442920
    },
    {
      "epoch": 0.1350329594937243,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011142534253425343,
      "loss": 3.0881,
      "step": 442930
    },
    {
      "epoch": 0.13505160272988245,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011142334233423344,
      "loss": 3.1775,
      "step": 442940
    },
    {
      "epoch": 0.13507024596604064,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011142134213421342,
      "loss": 3.1934,
      "step": 442950
    },
    {
      "epoch": 0.13508888920219883,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011141934193419343,
      "loss": 3.2426,
      "step": 442960
    },
    {
      "epoch": 0.13510753243835702,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011141734173417342,
      "loss": 3.2012,
      "step": 442970
    },
    {
      "epoch": 0.1351261756745152,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011141534153415343,
      "loss": 3.1255,
      "step": 442980
    },
    {
      "epoch": 0.13514481891067337,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011141334133413341,
      "loss": 3.1188,
      "step": 442990
    },
    {
      "epoch": 0.13516346214683156,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011141134113411342,
      "loss": 3.1035,
      "step": 443000
    },
    {
      "epoch": 0.13518210538298975,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001114093409340934,
      "loss": 3.2003,
      "step": 443010
    },
    {
      "epoch": 0.13520074861914794,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011140734073407342,
      "loss": 3.1218,
      "step": 443020
    },
    {
      "epoch": 0.13521939185530613,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001114053405340534,
      "loss": 3.0254,
      "step": 443030
    },
    {
      "epoch": 0.1352380350914643,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001114033403340334,
      "loss": 3.1692,
      "step": 443040
    },
    {
      "epoch": 0.13525667832762248,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001114013401340134,
      "loss": 3.1079,
      "step": 443050
    },
    {
      "epoch": 0.13527532156378066,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011139933993399341,
      "loss": 3.1299,
      "step": 443060
    },
    {
      "epoch": 0.13529396479993885,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011139733973397339,
      "loss": 3.105,
      "step": 443070
    },
    {
      "epoch": 0.13531260803609704,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001113953395339534,
      "loss": 3.1506,
      "step": 443080
    },
    {
      "epoch": 0.1353312512722552,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011139333933393338,
      "loss": 2.9502,
      "step": 443090
    },
    {
      "epoch": 0.1353498945084134,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001113913391339134,
      "loss": 3.2177,
      "step": 443100
    },
    {
      "epoch": 0.13536853774457158,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001113893389338934,
      "loss": 3.1483,
      "step": 443110
    },
    {
      "epoch": 0.13538718098072977,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001113873387338734,
      "loss": 3.1708,
      "step": 443120
    },
    {
      "epoch": 0.13540582421688796,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001113853385338534,
      "loss": 3.2965,
      "step": 443130
    },
    {
      "epoch": 0.13542446745304615,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011138333833383339,
      "loss": 3.0914,
      "step": 443140
    },
    {
      "epoch": 0.1354431106892043,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001113813381338134,
      "loss": 3.1963,
      "step": 443150
    },
    {
      "epoch": 0.1354617539253625,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011137933793379338,
      "loss": 3.1343,
      "step": 443160
    },
    {
      "epoch": 0.1354803971615207,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011137733773377339,
      "loss": 3.1161,
      "step": 443170
    },
    {
      "epoch": 0.13549904039767888,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011137533753375337,
      "loss": 3.113,
      "step": 443180
    },
    {
      "epoch": 0.13551768363383707,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011137333733373338,
      "loss": 3.0993,
      "step": 443190
    },
    {
      "epoch": 0.13553632686999523,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011137133713371338,
      "loss": 3.0961,
      "step": 443200
    },
    {
      "epoch": 0.13555497010615342,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011136933693369339,
      "loss": 3.2184,
      "step": 443210
    },
    {
      "epoch": 0.1355736133423116,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011136733673367337,
      "loss": 3.1902,
      "step": 443220
    },
    {
      "epoch": 0.1355922565784698,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011136533653365338,
      "loss": 3.1722,
      "step": 443230
    },
    {
      "epoch": 0.13561089981462798,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011136333633363336,
      "loss": 2.9545,
      "step": 443240
    },
    {
      "epoch": 0.13562954305078614,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011136133613361337,
      "loss": 3.2235,
      "step": 443250
    },
    {
      "epoch": 0.13564818628694433,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011135933593359335,
      "loss": 3.0854,
      "step": 443260
    },
    {
      "epoch": 0.13566682952310252,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011135733573357336,
      "loss": 3.145,
      "step": 443270
    },
    {
      "epoch": 0.1356854727592607,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011135533553355335,
      "loss": 2.9954,
      "step": 443280
    },
    {
      "epoch": 0.1357041159954189,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011135333533353336,
      "loss": 3.0616,
      "step": 443290
    },
    {
      "epoch": 0.13572275923157706,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011135133513351335,
      "loss": 3.1094,
      "step": 443300
    },
    {
      "epoch": 0.13574140246773525,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011134933493349336,
      "loss": 3.1516,
      "step": 443310
    },
    {
      "epoch": 0.13576004570389344,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011134733473347336,
      "loss": 3.1177,
      "step": 443320
    },
    {
      "epoch": 0.13577868894005163,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011134533453345335,
      "loss": 3.2033,
      "step": 443330
    },
    {
      "epoch": 0.13579733217620982,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011134333433343336,
      "loss": 3.1634,
      "step": 443340
    },
    {
      "epoch": 0.135815975412368,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011134133413341334,
      "loss": 3.0716,
      "step": 443350
    },
    {
      "epoch": 0.13583461864852617,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011133933393339335,
      "loss": 3.1037,
      "step": 443360
    },
    {
      "epoch": 0.13585326188468436,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011133733373337334,
      "loss": 3.08,
      "step": 443370
    },
    {
      "epoch": 0.13587190512084255,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011133533353335335,
      "loss": 3.0419,
      "step": 443380
    },
    {
      "epoch": 0.13589054835700073,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011133333333333333,
      "loss": 3.1214,
      "step": 443390
    },
    {
      "epoch": 0.13590919159315892,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011133133313331334,
      "loss": 3.1137,
      "step": 443400
    },
    {
      "epoch": 0.13592783482931708,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011132933293329332,
      "loss": 3.1574,
      "step": 443410
    },
    {
      "epoch": 0.13594647806547527,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011132733273327333,
      "loss": 3.1102,
      "step": 443420
    },
    {
      "epoch": 0.13596512130163346,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011132533253325333,
      "loss": 3.0242,
      "step": 443430
    },
    {
      "epoch": 0.13598376453779165,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011132333233323334,
      "loss": 3.1618,
      "step": 443440
    },
    {
      "epoch": 0.13600240777394984,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011132133213321332,
      "loss": 3.0985,
      "step": 443450
    },
    {
      "epoch": 0.136021051010108,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011131933193319333,
      "loss": 3.045,
      "step": 443460
    },
    {
      "epoch": 0.1360396942462662,
      "grad_norm": 0.875,
      "learning_rate": 0.00011131733173317331,
      "loss": 3.1188,
      "step": 443470
    },
    {
      "epoch": 0.13605833748242438,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011131533153315332,
      "loss": 3.0978,
      "step": 443480
    },
    {
      "epoch": 0.13607698071858257,
      "grad_norm": 0.75,
      "learning_rate": 0.0001113133313331333,
      "loss": 3.1829,
      "step": 443490
    },
    {
      "epoch": 0.13609562395474076,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011131133113311331,
      "loss": 3.0117,
      "step": 443500
    },
    {
      "epoch": 0.13611426719089892,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011130933093309332,
      "loss": 3.0858,
      "step": 443510
    },
    {
      "epoch": 0.1361329104270571,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011130733073307332,
      "loss": 3.1273,
      "step": 443520
    },
    {
      "epoch": 0.1361515536632153,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011130533053305333,
      "loss": 3.1971,
      "step": 443530
    },
    {
      "epoch": 0.13617019689937349,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011130333033303331,
      "loss": 3.1933,
      "step": 443540
    },
    {
      "epoch": 0.13618884013553167,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011130133013301332,
      "loss": 3.1547,
      "step": 443550
    },
    {
      "epoch": 0.13620748337168984,
      "grad_norm": 0.75,
      "learning_rate": 0.0001112993299329933,
      "loss": 3.0425,
      "step": 443560
    },
    {
      "epoch": 0.13622612660784802,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011129732973297331,
      "loss": 3.1785,
      "step": 443570
    },
    {
      "epoch": 0.1362447698440062,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011129532953295329,
      "loss": 3.1283,
      "step": 443580
    },
    {
      "epoch": 0.1362634130801644,
      "grad_norm": 0.75,
      "learning_rate": 0.0001112933293329333,
      "loss": 3.1959,
      "step": 443590
    },
    {
      "epoch": 0.1362820563163226,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001112913291329133,
      "loss": 3.0852,
      "step": 443600
    },
    {
      "epoch": 0.13630069955248078,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001112893289328933,
      "loss": 3.1561,
      "step": 443610
    },
    {
      "epoch": 0.13631934278863894,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011128732873287329,
      "loss": 3.1332,
      "step": 443620
    },
    {
      "epoch": 0.13633798602479713,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001112853285328533,
      "loss": 3.1674,
      "step": 443630
    },
    {
      "epoch": 0.13635662926095532,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011128332833283328,
      "loss": 3.1246,
      "step": 443640
    },
    {
      "epoch": 0.1363752724971135,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011128132813281329,
      "loss": 3.1269,
      "step": 443650
    },
    {
      "epoch": 0.1363939157332717,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011127932793279328,
      "loss": 3.2544,
      "step": 443660
    },
    {
      "epoch": 0.13641255896942986,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011127732773277328,
      "loss": 3.1108,
      "step": 443670
    },
    {
      "epoch": 0.13643120220558805,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011127532753275327,
      "loss": 3.0096,
      "step": 443680
    },
    {
      "epoch": 0.13644984544174624,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011127332733273328,
      "loss": 3.047,
      "step": 443690
    },
    {
      "epoch": 0.13646848867790443,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011127132713271326,
      "loss": 3.1062,
      "step": 443700
    },
    {
      "epoch": 0.13648713191406261,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011126932693269327,
      "loss": 3.0091,
      "step": 443710
    },
    {
      "epoch": 0.13650577515022078,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011126732673267328,
      "loss": 3.1091,
      "step": 443720
    },
    {
      "epoch": 0.13652441838637897,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011126532653265327,
      "loss": 3.0778,
      "step": 443730
    },
    {
      "epoch": 0.13654306162253715,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011126332633263327,
      "loss": 3.2017,
      "step": 443740
    },
    {
      "epoch": 0.13656170485869534,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011126132613261327,
      "loss": 3.0461,
      "step": 443750
    },
    {
      "epoch": 0.13658034809485353,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011125932593259328,
      "loss": 3.1817,
      "step": 443760
    },
    {
      "epoch": 0.1365989913310117,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011125732573257326,
      "loss": 3.1247,
      "step": 443770
    },
    {
      "epoch": 0.13661763456716988,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011125532553255327,
      "loss": 3.209,
      "step": 443780
    },
    {
      "epoch": 0.13663627780332807,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011125332533253325,
      "loss": 3.1701,
      "step": 443790
    },
    {
      "epoch": 0.13665492103948626,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011125132513251326,
      "loss": 2.8874,
      "step": 443800
    },
    {
      "epoch": 0.13667356427564445,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011124932493249324,
      "loss": 3.133,
      "step": 443810
    },
    {
      "epoch": 0.13669220751180264,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011124732473247325,
      "loss": 3.2123,
      "step": 443820
    },
    {
      "epoch": 0.1367108507479608,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011124532453245325,
      "loss": 3.1024,
      "step": 443830
    },
    {
      "epoch": 0.136729493984119,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011124332433243326,
      "loss": 3.1433,
      "step": 443840
    },
    {
      "epoch": 0.13674813722027718,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011124132413241324,
      "loss": 2.9606,
      "step": 443850
    },
    {
      "epoch": 0.13676678045643537,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011123932393239325,
      "loss": 3.0968,
      "step": 443860
    },
    {
      "epoch": 0.13678542369259356,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011123732373237323,
      "loss": 3.0822,
      "step": 443870
    },
    {
      "epoch": 0.13680406692875172,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011123532353235324,
      "loss": 3.096,
      "step": 443880
    },
    {
      "epoch": 0.1368227101649099,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011123332333233322,
      "loss": 3.0864,
      "step": 443890
    },
    {
      "epoch": 0.1368413534010681,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011123132313231323,
      "loss": 3.2739,
      "step": 443900
    },
    {
      "epoch": 0.13685999663722628,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011122932293229324,
      "loss": 3.1753,
      "step": 443910
    },
    {
      "epoch": 0.13687863987338447,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011122732273227324,
      "loss": 3.0986,
      "step": 443920
    },
    {
      "epoch": 0.13689728310954263,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011122532253225324,
      "loss": 3.1864,
      "step": 443930
    },
    {
      "epoch": 0.13691592634570082,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011122332233223323,
      "loss": 3.1395,
      "step": 443940
    },
    {
      "epoch": 0.136934569581859,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011122132213221324,
      "loss": 3.0338,
      "step": 443950
    },
    {
      "epoch": 0.1369532128180172,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011121932193219322,
      "loss": 3.1414,
      "step": 443960
    },
    {
      "epoch": 0.1369718560541754,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011121732173217323,
      "loss": 3.0538,
      "step": 443970
    },
    {
      "epoch": 0.13699049929033355,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011121532153215322,
      "loss": 3.177,
      "step": 443980
    },
    {
      "epoch": 0.13700914252649174,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011121332133213322,
      "loss": 3.2203,
      "step": 443990
    },
    {
      "epoch": 0.13702778576264993,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011121132113211321,
      "loss": 3.1077,
      "step": 444000
    },
    {
      "epoch": 0.13704642899880812,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011120932093209322,
      "loss": 3.1514,
      "step": 444010
    },
    {
      "epoch": 0.1370650722349663,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001112073207320732,
      "loss": 3.1605,
      "step": 444020
    },
    {
      "epoch": 0.1370837154711245,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011120532053205321,
      "loss": 3.1371,
      "step": 444030
    },
    {
      "epoch": 0.13710235870728266,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001112033203320332,
      "loss": 3.209,
      "step": 444040
    },
    {
      "epoch": 0.13712100194344085,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001112013201320132,
      "loss": 3.0644,
      "step": 444050
    },
    {
      "epoch": 0.13713964517959903,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001111993199319932,
      "loss": 3.1565,
      "step": 444060
    },
    {
      "epoch": 0.13715828841575722,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011119731973197321,
      "loss": 3.0737,
      "step": 444070
    },
    {
      "epoch": 0.1371769316519154,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011119531953195319,
      "loss": 3.0254,
      "step": 444080
    },
    {
      "epoch": 0.13719557488807357,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001111933193319332,
      "loss": 3.1035,
      "step": 444090
    },
    {
      "epoch": 0.13721421812423176,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011119131913191321,
      "loss": 3.1425,
      "step": 444100
    },
    {
      "epoch": 0.13723286136038995,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011118931893189319,
      "loss": 3.1005,
      "step": 444110
    },
    {
      "epoch": 0.13725150459654814,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001111873187318732,
      "loss": 3.213,
      "step": 444120
    },
    {
      "epoch": 0.13727014783270633,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011118531853185318,
      "loss": 3.1201,
      "step": 444130
    },
    {
      "epoch": 0.1372887910688645,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00011118331833183319,
      "loss": 3.1824,
      "step": 444140
    },
    {
      "epoch": 0.13730743430502268,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011118131813181319,
      "loss": 3.1901,
      "step": 444150
    },
    {
      "epoch": 0.13732607754118087,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001111793179317932,
      "loss": 3.1422,
      "step": 444160
    },
    {
      "epoch": 0.13734472077733906,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011117731773177318,
      "loss": 3.151,
      "step": 444170
    },
    {
      "epoch": 0.13736336401349725,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011117531753175319,
      "loss": 3.1498,
      "step": 444180
    },
    {
      "epoch": 0.1373820072496554,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011117331733173317,
      "loss": 3.1534,
      "step": 444190
    },
    {
      "epoch": 0.1374006504858136,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011117131713171318,
      "loss": 3.1023,
      "step": 444200
    },
    {
      "epoch": 0.13741929372197179,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011116931693169316,
      "loss": 3.071,
      "step": 444210
    },
    {
      "epoch": 0.13743793695812997,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011116731673167317,
      "loss": 3.1123,
      "step": 444220
    },
    {
      "epoch": 0.13745658019428816,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011116531653165317,
      "loss": 3.0848,
      "step": 444230
    },
    {
      "epoch": 0.13747522343044635,
      "grad_norm": 0.75,
      "learning_rate": 0.00011116331633163318,
      "loss": 3.1343,
      "step": 444240
    },
    {
      "epoch": 0.13749386666660451,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011116131613161316,
      "loss": 3.0694,
      "step": 444250
    },
    {
      "epoch": 0.1375125099027627,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011115931593159317,
      "loss": 3.0971,
      "step": 444260
    },
    {
      "epoch": 0.1375311531389209,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011115731573157315,
      "loss": 3.1174,
      "step": 444270
    },
    {
      "epoch": 0.13754979637507908,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011115531553155316,
      "loss": 3.1312,
      "step": 444280
    },
    {
      "epoch": 0.13756843961123727,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011115331533153315,
      "loss": 3.0452,
      "step": 444290
    },
    {
      "epoch": 0.13758708284739543,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011115131513151316,
      "loss": 3.0573,
      "step": 444300
    },
    {
      "epoch": 0.13760572608355362,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011114931493149316,
      "loss": 3.0855,
      "step": 444310
    },
    {
      "epoch": 0.1376243693197118,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011114731473147315,
      "loss": 3.0496,
      "step": 444320
    },
    {
      "epoch": 0.13764301255587,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011114531453145316,
      "loss": 3.1896,
      "step": 444330
    },
    {
      "epoch": 0.1376616557920282,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011114331433143315,
      "loss": 3.1094,
      "step": 444340
    },
    {
      "epoch": 0.13768029902818635,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011114131413141315,
      "loss": 3.1633,
      "step": 444350
    },
    {
      "epoch": 0.13769894226434454,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011113931393139314,
      "loss": 3.0891,
      "step": 444360
    },
    {
      "epoch": 0.13771758550050273,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011113731373137315,
      "loss": 3.1002,
      "step": 444370
    },
    {
      "epoch": 0.13773622873666092,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011113531353135314,
      "loss": 3.1989,
      "step": 444380
    },
    {
      "epoch": 0.1377548719728191,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011113331333133315,
      "loss": 3.1736,
      "step": 444390
    },
    {
      "epoch": 0.13777351520897727,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011113131313131313,
      "loss": 3.0567,
      "step": 444400
    },
    {
      "epoch": 0.13779215844513545,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011112931293129314,
      "loss": 3.1759,
      "step": 444410
    },
    {
      "epoch": 0.13781080168129364,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011112731273127312,
      "loss": 3.0928,
      "step": 444420
    },
    {
      "epoch": 0.13782944491745183,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011112531253125313,
      "loss": 3.162,
      "step": 444430
    },
    {
      "epoch": 0.13784808815361002,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011112331233123311,
      "loss": 3.1526,
      "step": 444440
    },
    {
      "epoch": 0.13786673138976818,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011112131213121312,
      "loss": 3.1933,
      "step": 444450
    },
    {
      "epoch": 0.13788537462592637,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011111931193119312,
      "loss": 3.1646,
      "step": 444460
    },
    {
      "epoch": 0.13790401786208456,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011111731173117313,
      "loss": 3.1135,
      "step": 444470
    },
    {
      "epoch": 0.13792266109824275,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011111531153115311,
      "loss": 3.1256,
      "step": 444480
    },
    {
      "epoch": 0.13794130433440094,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011111331133113312,
      "loss": 3.1091,
      "step": 444490
    },
    {
      "epoch": 0.13795994757055913,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011111131113111313,
      "loss": 3.129,
      "step": 444500
    },
    {
      "epoch": 0.1379785908067173,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011110931093109311,
      "loss": 3.1151,
      "step": 444510
    },
    {
      "epoch": 0.13799723404287548,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011110731073107312,
      "loss": 3.2302,
      "step": 444520
    },
    {
      "epoch": 0.13801587727903367,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001111053105310531,
      "loss": 3.026,
      "step": 444530
    },
    {
      "epoch": 0.13803452051519186,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011110331033103311,
      "loss": 3.1477,
      "step": 444540
    },
    {
      "epoch": 0.13805316375135004,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011110131013101311,
      "loss": 3.0716,
      "step": 444550
    },
    {
      "epoch": 0.1380718069875082,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011109930993099312,
      "loss": 3.0212,
      "step": 444560
    },
    {
      "epoch": 0.1380904502236664,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001110973097309731,
      "loss": 3.0157,
      "step": 444570
    },
    {
      "epoch": 0.13810909345982458,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011109530953095311,
      "loss": 3.1187,
      "step": 444580
    },
    {
      "epoch": 0.13812773669598277,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011109330933093309,
      "loss": 3.2249,
      "step": 444590
    },
    {
      "epoch": 0.13814637993214096,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001110913091309131,
      "loss": 3.0097,
      "step": 444600
    },
    {
      "epoch": 0.13816502316829912,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001110893089308931,
      "loss": 3.1752,
      "step": 444610
    },
    {
      "epoch": 0.1381836664044573,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011108730873087309,
      "loss": 3.1304,
      "step": 444620
    },
    {
      "epoch": 0.1382023096406155,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011108530853085309,
      "loss": 3.0955,
      "step": 444630
    },
    {
      "epoch": 0.1382209528767737,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001110833083308331,
      "loss": 3.1177,
      "step": 444640
    },
    {
      "epoch": 0.13823959611293188,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011108130813081308,
      "loss": 2.9824,
      "step": 444650
    },
    {
      "epoch": 0.13825823934909004,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011107930793079309,
      "loss": 3.1669,
      "step": 444660
    },
    {
      "epoch": 0.13827688258524823,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011107730773077307,
      "loss": 2.996,
      "step": 444670
    },
    {
      "epoch": 0.13829552582140642,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011107530753075308,
      "loss": 3.1382,
      "step": 444680
    },
    {
      "epoch": 0.1383141690575646,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011107330733073309,
      "loss": 3.1423,
      "step": 444690
    },
    {
      "epoch": 0.1383328122937228,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011107130713071308,
      "loss": 3.0417,
      "step": 444700
    },
    {
      "epoch": 0.13835145552988098,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011106930693069309,
      "loss": 3.0437,
      "step": 444710
    },
    {
      "epoch": 0.13837009876603915,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011106730673067307,
      "loss": 3.2273,
      "step": 444720
    },
    {
      "epoch": 0.13838874200219733,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011106530653065308,
      "loss": 3.0858,
      "step": 444730
    },
    {
      "epoch": 0.13840738523835552,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011106330633063306,
      "loss": 3.1301,
      "step": 444740
    },
    {
      "epoch": 0.1384260284745137,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011106130613061307,
      "loss": 3.0818,
      "step": 444750
    },
    {
      "epoch": 0.1384446717106719,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011105930593059306,
      "loss": 3.0525,
      "step": 444760
    },
    {
      "epoch": 0.13846331494683006,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011105730573057306,
      "loss": 2.9772,
      "step": 444770
    },
    {
      "epoch": 0.13848195818298825,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011105530553055306,
      "loss": 3.1737,
      "step": 444780
    },
    {
      "epoch": 0.13850060141914644,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011105330533053307,
      "loss": 3.1459,
      "step": 444790
    },
    {
      "epoch": 0.13851924465530463,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011105130513051305,
      "loss": 3.014,
      "step": 444800
    },
    {
      "epoch": 0.13853788789146282,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011104930493049306,
      "loss": 3.2522,
      "step": 444810
    },
    {
      "epoch": 0.13855653112762098,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011104730473047304,
      "loss": 3.1294,
      "step": 444820
    },
    {
      "epoch": 0.13857517436377917,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011104530453045305,
      "loss": 3.2273,
      "step": 444830
    },
    {
      "epoch": 0.13859381759993736,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011104330433043303,
      "loss": 3.1582,
      "step": 444840
    },
    {
      "epoch": 0.13861246083609555,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011104130413041304,
      "loss": 3.0716,
      "step": 444850
    },
    {
      "epoch": 0.13863110407225374,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011103930393039304,
      "loss": 3.0848,
      "step": 444860
    },
    {
      "epoch": 0.1386497473084119,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011103730373037305,
      "loss": 3.1264,
      "step": 444870
    },
    {
      "epoch": 0.13866839054457009,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011103530353035303,
      "loss": 3.1805,
      "step": 444880
    },
    {
      "epoch": 0.13868703378072827,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011103330333033304,
      "loss": 3.1314,
      "step": 444890
    },
    {
      "epoch": 0.13870567701688646,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011103130313031305,
      "loss": 3.0757,
      "step": 444900
    },
    {
      "epoch": 0.13872432025304465,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011102930293029303,
      "loss": 3.0999,
      "step": 444910
    },
    {
      "epoch": 0.13874296348920284,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011102730273027304,
      "loss": 3.2204,
      "step": 444920
    },
    {
      "epoch": 0.138761606725361,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011102530253025303,
      "loss": 3.1573,
      "step": 444930
    },
    {
      "epoch": 0.1387802499615192,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00011102330233023303,
      "loss": 3.2746,
      "step": 444940
    },
    {
      "epoch": 0.13879889319767738,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011102130213021303,
      "loss": 2.9098,
      "step": 444950
    },
    {
      "epoch": 0.13881753643383557,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011101930193019303,
      "loss": 3.108,
      "step": 444960
    },
    {
      "epoch": 0.13883617966999376,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011101730173017302,
      "loss": 3.1528,
      "step": 444970
    },
    {
      "epoch": 0.13885482290615192,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011101530153015303,
      "loss": 3.1485,
      "step": 444980
    },
    {
      "epoch": 0.1388734661423101,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011101330133013301,
      "loss": 3.0015,
      "step": 444990
    },
    {
      "epoch": 0.1388921093784683,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011101130113011302,
      "loss": 3.2249,
      "step": 445000
    },
    {
      "epoch": 0.1389107526146265,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011100930093009301,
      "loss": 3.1349,
      "step": 445010
    },
    {
      "epoch": 0.13892939585078468,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011100730073007302,
      "loss": 3.0531,
      "step": 445020
    },
    {
      "epoch": 0.13894803908694284,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000111005300530053,
      "loss": 3.2332,
      "step": 445030
    },
    {
      "epoch": 0.13896668232310103,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011100330033003301,
      "loss": 3.118,
      "step": 445040
    },
    {
      "epoch": 0.13898532555925922,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000111001300130013,
      "loss": 3.1929,
      "step": 445050
    },
    {
      "epoch": 0.1390039687954174,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000110999299929993,
      "loss": 3.0405,
      "step": 445060
    },
    {
      "epoch": 0.1390226120315756,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011099729972997299,
      "loss": 3.1714,
      "step": 445070
    },
    {
      "epoch": 0.13904125526773375,
      "grad_norm": 0.796875,
      "learning_rate": 0.000110995299529953,
      "loss": 3.0996,
      "step": 445080
    },
    {
      "epoch": 0.13905989850389194,
      "grad_norm": 0.81640625,
      "learning_rate": 0.000110993299329933,
      "loss": 3.0771,
      "step": 445090
    },
    {
      "epoch": 0.13907854174005013,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000110991299129913,
      "loss": 3.136,
      "step": 445100
    },
    {
      "epoch": 0.13909718497620832,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011098929892989301,
      "loss": 3.114,
      "step": 445110
    },
    {
      "epoch": 0.1391158282123665,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011098729872987299,
      "loss": 3.1419,
      "step": 445120
    },
    {
      "epoch": 0.13913447144852467,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000110985298529853,
      "loss": 3.1938,
      "step": 445130
    },
    {
      "epoch": 0.13915311468468286,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011098329832983298,
      "loss": 3.1194,
      "step": 445140
    },
    {
      "epoch": 0.13917175792084105,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011098129812981299,
      "loss": 3.086,
      "step": 445150
    },
    {
      "epoch": 0.13919040115699924,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011097929792979297,
      "loss": 3.1129,
      "step": 445160
    },
    {
      "epoch": 0.13920904439315743,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011097729772977298,
      "loss": 3.0789,
      "step": 445170
    },
    {
      "epoch": 0.13922768762931562,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011097529752975298,
      "loss": 2.9961,
      "step": 445180
    },
    {
      "epoch": 0.13924633086547378,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011097329732973299,
      "loss": 3.1751,
      "step": 445190
    },
    {
      "epoch": 0.13926497410163197,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011097129712971297,
      "loss": 3.2416,
      "step": 445200
    },
    {
      "epoch": 0.13928361733779016,
      "grad_norm": 0.75,
      "learning_rate": 0.00011096929692969298,
      "loss": 3.1435,
      "step": 445210
    },
    {
      "epoch": 0.13930226057394834,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011096729672967296,
      "loss": 3.0622,
      "step": 445220
    },
    {
      "epoch": 0.13932090381010653,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011096529652965297,
      "loss": 3.0115,
      "step": 445230
    },
    {
      "epoch": 0.1393395470462647,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011096329632963297,
      "loss": 3.0025,
      "step": 445240
    },
    {
      "epoch": 0.13935819028242288,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011096129612961297,
      "loss": 3.0208,
      "step": 445250
    },
    {
      "epoch": 0.13937683351858107,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011095929592959296,
      "loss": 3.1707,
      "step": 445260
    },
    {
      "epoch": 0.13939547675473926,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011095729572957297,
      "loss": 3.1132,
      "step": 445270
    },
    {
      "epoch": 0.13941411999089745,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011095529552955298,
      "loss": 3.1246,
      "step": 445280
    },
    {
      "epoch": 0.1394327632270556,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011095329532953296,
      "loss": 3.0866,
      "step": 445290
    },
    {
      "epoch": 0.1394514064632138,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011095129512951297,
      "loss": 3.0788,
      "step": 445300
    },
    {
      "epoch": 0.139470049699372,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011094929492949295,
      "loss": 2.9822,
      "step": 445310
    },
    {
      "epoch": 0.13948869293553018,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011094729472947296,
      "loss": 3.0603,
      "step": 445320
    },
    {
      "epoch": 0.13950733617168837,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011094529452945295,
      "loss": 2.9652,
      "step": 445330
    },
    {
      "epoch": 0.13952597940784653,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011094329432943296,
      "loss": 3.1116,
      "step": 445340
    },
    {
      "epoch": 0.13954462264400472,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011094129412941294,
      "loss": 3.1903,
      "step": 445350
    },
    {
      "epoch": 0.1395632658801629,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011093929392939295,
      "loss": 3.0582,
      "step": 445360
    },
    {
      "epoch": 0.1395819091163211,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011093729372937294,
      "loss": 3.1728,
      "step": 445370
    },
    {
      "epoch": 0.13960055235247928,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011093529352935294,
      "loss": 3.1151,
      "step": 445380
    },
    {
      "epoch": 0.13961919558863747,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011093329332933293,
      "loss": 3.1792,
      "step": 445390
    },
    {
      "epoch": 0.13963783882479563,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011093129312931294,
      "loss": 3.025,
      "step": 445400
    },
    {
      "epoch": 0.13965648206095382,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011092929292929293,
      "loss": 3.054,
      "step": 445410
    },
    {
      "epoch": 0.139675125297112,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011092729272927294,
      "loss": 3.146,
      "step": 445420
    },
    {
      "epoch": 0.1396937685332702,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011092529252925292,
      "loss": 3.1658,
      "step": 445430
    },
    {
      "epoch": 0.1397124117694284,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011092329232923293,
      "loss": 3.1214,
      "step": 445440
    },
    {
      "epoch": 0.13973105500558655,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011092129212921291,
      "loss": 3.1236,
      "step": 445450
    },
    {
      "epoch": 0.13974969824174474,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011091929192919292,
      "loss": 3.1424,
      "step": 445460
    },
    {
      "epoch": 0.13976834147790293,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011091729172917292,
      "loss": 3.1371,
      "step": 445470
    },
    {
      "epoch": 0.13978698471406112,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00011091529152915291,
      "loss": 3.1751,
      "step": 445480
    },
    {
      "epoch": 0.1398056279502193,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011091329132913292,
      "loss": 3.1693,
      "step": 445490
    },
    {
      "epoch": 0.13982427118637747,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011091129112911292,
      "loss": 3.0771,
      "step": 445500
    },
    {
      "epoch": 0.13984291442253566,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011090929092909293,
      "loss": 3.0602,
      "step": 445510
    },
    {
      "epoch": 0.13986155765869385,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011090729072907291,
      "loss": 3.2809,
      "step": 445520
    },
    {
      "epoch": 0.13988020089485204,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011090529052905292,
      "loss": 2.973,
      "step": 445530
    },
    {
      "epoch": 0.13989884413101022,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001109032903290329,
      "loss": 3.0786,
      "step": 445540
    },
    {
      "epoch": 0.13991748736716839,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011090129012901291,
      "loss": 3.1173,
      "step": 445550
    },
    {
      "epoch": 0.13993613060332658,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001108992899289929,
      "loss": 3.1185,
      "step": 445560
    },
    {
      "epoch": 0.13995477383948476,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011089728972897292,
      "loss": 3.1166,
      "step": 445570
    },
    {
      "epoch": 0.13997341707564295,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001108952895289529,
      "loss": 3.1369,
      "step": 445580
    },
    {
      "epoch": 0.13999206031180114,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001108932893289329,
      "loss": 3.2121,
      "step": 445590
    },
    {
      "epoch": 0.14001070354795933,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011089128912891289,
      "loss": 3.0089,
      "step": 445600
    },
    {
      "epoch": 0.1400293467841175,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001108892889288929,
      "loss": 3.154,
      "step": 445610
    },
    {
      "epoch": 0.14004799002027568,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011088728872887288,
      "loss": 2.9472,
      "step": 445620
    },
    {
      "epoch": 0.14006663325643387,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011088528852885289,
      "loss": 3.1006,
      "step": 445630
    },
    {
      "epoch": 0.14008527649259206,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011088328832883288,
      "loss": 3.2103,
      "step": 445640
    },
    {
      "epoch": 0.14010391972875025,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001108812881288129,
      "loss": 3.1846,
      "step": 445650
    },
    {
      "epoch": 0.1401225629649084,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011087928792879288,
      "loss": 3.081,
      "step": 445660
    },
    {
      "epoch": 0.1401412062010666,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011087728772877288,
      "loss": 3.2172,
      "step": 445670
    },
    {
      "epoch": 0.1401598494372248,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001108752875287529,
      "loss": 3.1396,
      "step": 445680
    },
    {
      "epoch": 0.14017849267338298,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011087328732873288,
      "loss": 3.0188,
      "step": 445690
    },
    {
      "epoch": 0.14019713590954117,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011087128712871288,
      "loss": 3.0722,
      "step": 445700
    },
    {
      "epoch": 0.14021577914569933,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011086928692869287,
      "loss": 3.1363,
      "step": 445710
    },
    {
      "epoch": 0.14023442238185752,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011086728672867288,
      "loss": 3.1382,
      "step": 445720
    },
    {
      "epoch": 0.1402530656180157,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011086528652865287,
      "loss": 3.0865,
      "step": 445730
    },
    {
      "epoch": 0.1402717088541739,
      "grad_norm": 0.875,
      "learning_rate": 0.00011086328632863288,
      "loss": 3.1589,
      "step": 445740
    },
    {
      "epoch": 0.14029035209033208,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011086128612861286,
      "loss": 3.0543,
      "step": 445750
    },
    {
      "epoch": 0.14030899532649024,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011085928592859287,
      "loss": 3.0192,
      "step": 445760
    },
    {
      "epoch": 0.14032763856264843,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011085728572857285,
      "loss": 3.1896,
      "step": 445770
    },
    {
      "epoch": 0.14034628179880662,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011085528552855286,
      "loss": 2.9995,
      "step": 445780
    },
    {
      "epoch": 0.1403649250349648,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011085328532853285,
      "loss": 3.1623,
      "step": 445790
    },
    {
      "epoch": 0.140383568271123,
      "grad_norm": 0.75,
      "learning_rate": 0.00011085128512851285,
      "loss": 3.1614,
      "step": 445800
    },
    {
      "epoch": 0.14040221150728116,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011084928492849285,
      "loss": 3.2887,
      "step": 445810
    },
    {
      "epoch": 0.14042085474343935,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011084728472847286,
      "loss": 3.1131,
      "step": 445820
    },
    {
      "epoch": 0.14043949797959754,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011084528452845284,
      "loss": 3.1463,
      "step": 445830
    },
    {
      "epoch": 0.14045814121575573,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011084328432843285,
      "loss": 3.2186,
      "step": 445840
    },
    {
      "epoch": 0.14047678445191392,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011084128412841283,
      "loss": 3.1912,
      "step": 445850
    },
    {
      "epoch": 0.1404954276880721,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011083928392839284,
      "loss": 3.1255,
      "step": 445860
    },
    {
      "epoch": 0.14051407092423027,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011083728372837284,
      "loss": 3.1507,
      "step": 445870
    },
    {
      "epoch": 0.14053271416038846,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011083528352835285,
      "loss": 3.1533,
      "step": 445880
    },
    {
      "epoch": 0.14055135739654664,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011083328332833284,
      "loss": 3.1845,
      "step": 445890
    },
    {
      "epoch": 0.14057000063270483,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011083128312831284,
      "loss": 3.1459,
      "step": 445900
    },
    {
      "epoch": 0.14058864386886302,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011082928292829285,
      "loss": 3.0144,
      "step": 445910
    },
    {
      "epoch": 0.14060728710502118,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011082728272827283,
      "loss": 2.9952,
      "step": 445920
    },
    {
      "epoch": 0.14062593034117937,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011082528252825284,
      "loss": 3.1748,
      "step": 445930
    },
    {
      "epoch": 0.14064457357733756,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011082328232823282,
      "loss": 3.0975,
      "step": 445940
    },
    {
      "epoch": 0.14066321681349575,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011082128212821283,
      "loss": 3.1933,
      "step": 445950
    },
    {
      "epoch": 0.14068186004965394,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011081928192819282,
      "loss": 3.2058,
      "step": 445960
    },
    {
      "epoch": 0.1407005032858121,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011081728172817283,
      "loss": 3.1154,
      "step": 445970
    },
    {
      "epoch": 0.1407191465219703,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011081528152815282,
      "loss": 3.1176,
      "step": 445980
    },
    {
      "epoch": 0.14073778975812848,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011081328132813282,
      "loss": 3.1212,
      "step": 445990
    },
    {
      "epoch": 0.14075643299428667,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011081128112811281,
      "loss": 3.2436,
      "step": 446000
    },
    {
      "epoch": 0.14077507623044486,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011080928092809282,
      "loss": 3.261,
      "step": 446010
    },
    {
      "epoch": 0.14079371946660302,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001108072807280728,
      "loss": 3.2444,
      "step": 446020
    },
    {
      "epoch": 0.1408123627027612,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011080528052805281,
      "loss": 3.0717,
      "step": 446030
    },
    {
      "epoch": 0.1408310059389194,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001108032803280328,
      "loss": 3.1247,
      "step": 446040
    },
    {
      "epoch": 0.14084964917507758,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011080128012801281,
      "loss": 3.1934,
      "step": 446050
    },
    {
      "epoch": 0.14086829241123577,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001107992799279928,
      "loss": 3.0157,
      "step": 446060
    },
    {
      "epoch": 0.14088693564739396,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001107972797279728,
      "loss": 3.3214,
      "step": 446070
    },
    {
      "epoch": 0.14090557888355212,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011079527952795281,
      "loss": 3.0004,
      "step": 446080
    },
    {
      "epoch": 0.1409242221197103,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001107932793279328,
      "loss": 3.0942,
      "step": 446090
    },
    {
      "epoch": 0.1409428653558685,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001107912791279128,
      "loss": 3.155,
      "step": 446100
    },
    {
      "epoch": 0.1409615085920267,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011078927892789279,
      "loss": 3.2017,
      "step": 446110
    },
    {
      "epoch": 0.14098015182818488,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001107872787278728,
      "loss": 3.1341,
      "step": 446120
    },
    {
      "epoch": 0.14099879506434304,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011078527852785279,
      "loss": 3.098,
      "step": 446130
    },
    {
      "epoch": 0.14101743830050123,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001107832783278328,
      "loss": 3.1231,
      "step": 446140
    },
    {
      "epoch": 0.14103608153665942,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011078127812781278,
      "loss": 3.0342,
      "step": 446150
    },
    {
      "epoch": 0.1410547247728176,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011077927792779279,
      "loss": 3.2144,
      "step": 446160
    },
    {
      "epoch": 0.1410733680089758,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011077727772777277,
      "loss": 3.2119,
      "step": 446170
    },
    {
      "epoch": 0.14109201124513396,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011077527752775278,
      "loss": 3.2403,
      "step": 446180
    },
    {
      "epoch": 0.14111065448129215,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011077327732773278,
      "loss": 2.9999,
      "step": 446190
    },
    {
      "epoch": 0.14112929771745034,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011077127712771279,
      "loss": 3.0635,
      "step": 446200
    },
    {
      "epoch": 0.14114794095360852,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011076927692769277,
      "loss": 3.1014,
      "step": 446210
    },
    {
      "epoch": 0.14116658418976671,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011076727672767278,
      "loss": 3.0906,
      "step": 446220
    },
    {
      "epoch": 0.14118522742592488,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011076527652765276,
      "loss": 3.1302,
      "step": 446230
    },
    {
      "epoch": 0.14120387066208306,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011076327632763277,
      "loss": 3.2271,
      "step": 446240
    },
    {
      "epoch": 0.14122251389824125,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011076127612761275,
      "loss": 3.0174,
      "step": 446250
    },
    {
      "epoch": 0.14124115713439944,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011075927592759276,
      "loss": 2.9879,
      "step": 446260
    },
    {
      "epoch": 0.14125980037055763,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011075727572757277,
      "loss": 2.9539,
      "step": 446270
    },
    {
      "epoch": 0.14127844360671582,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011075527552755276,
      "loss": 3.099,
      "step": 446280
    },
    {
      "epoch": 0.14129708684287398,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011075327532753277,
      "loss": 3.1076,
      "step": 446290
    },
    {
      "epoch": 0.14131573007903217,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011075127512751276,
      "loss": 3.1987,
      "step": 446300
    },
    {
      "epoch": 0.14133437331519036,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011074927492749277,
      "loss": 2.9922,
      "step": 446310
    },
    {
      "epoch": 0.14135301655134855,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011074727472747275,
      "loss": 3.2157,
      "step": 446320
    },
    {
      "epoch": 0.14137165978750674,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011074527452745276,
      "loss": 3.017,
      "step": 446330
    },
    {
      "epoch": 0.1413903030236649,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011074327432743274,
      "loss": 3.1423,
      "step": 446340
    },
    {
      "epoch": 0.1414089462598231,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011074127412741275,
      "loss": 3.1247,
      "step": 446350
    },
    {
      "epoch": 0.14142758949598128,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011073927392739274,
      "loss": 3.1798,
      "step": 446360
    },
    {
      "epoch": 0.14144623273213947,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011073727372737275,
      "loss": 3.0619,
      "step": 446370
    },
    {
      "epoch": 0.14146487596829765,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011073527352735273,
      "loss": 3.2159,
      "step": 446380
    },
    {
      "epoch": 0.14148351920445582,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011073327332733274,
      "loss": 3.0615,
      "step": 446390
    },
    {
      "epoch": 0.141502162440614,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011073127312731273,
      "loss": 3.0745,
      "step": 446400
    },
    {
      "epoch": 0.1415208056767722,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011072927292729273,
      "loss": 3.0852,
      "step": 446410
    },
    {
      "epoch": 0.14153944891293038,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011072727272727273,
      "loss": 3.032,
      "step": 446420
    },
    {
      "epoch": 0.14155809214908857,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011072527252725273,
      "loss": 3.0929,
      "step": 446430
    },
    {
      "epoch": 0.14157673538524673,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011072327232723272,
      "loss": 3.084,
      "step": 446440
    },
    {
      "epoch": 0.14159537862140492,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011072127212721273,
      "loss": 3.1921,
      "step": 446450
    },
    {
      "epoch": 0.1416140218575631,
      "grad_norm": 0.75,
      "learning_rate": 0.00011071927192719271,
      "loss": 3.0159,
      "step": 446460
    },
    {
      "epoch": 0.1416326650937213,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011071727172717272,
      "loss": 3.0953,
      "step": 446470
    },
    {
      "epoch": 0.1416513083298795,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011071527152715273,
      "loss": 3.1548,
      "step": 446480
    },
    {
      "epoch": 0.14166995156603765,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011071327132713271,
      "loss": 3.0332,
      "step": 446490
    },
    {
      "epoch": 0.14168859480219584,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011071127112711272,
      "loss": 3.1101,
      "step": 446500
    },
    {
      "epoch": 0.14170723803835403,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011070927092709272,
      "loss": 3.1235,
      "step": 446510
    },
    {
      "epoch": 0.14172588127451222,
      "grad_norm": 0.875,
      "learning_rate": 0.00011070727072707273,
      "loss": 3.1897,
      "step": 446520
    },
    {
      "epoch": 0.1417445245106704,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011070527052705271,
      "loss": 3.0073,
      "step": 446530
    },
    {
      "epoch": 0.1417631677468286,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011070327032703272,
      "loss": 3.1778,
      "step": 446540
    },
    {
      "epoch": 0.14178181098298676,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001107012701270127,
      "loss": 3.0204,
      "step": 446550
    },
    {
      "epoch": 0.14180045421914494,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011069926992699271,
      "loss": 2.9874,
      "step": 446560
    },
    {
      "epoch": 0.14181909745530313,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011069726972697269,
      "loss": 3.1182,
      "step": 446570
    },
    {
      "epoch": 0.14183774069146132,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001106952695269527,
      "loss": 3.0955,
      "step": 446580
    },
    {
      "epoch": 0.1418563839276195,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001106932693269327,
      "loss": 3.136,
      "step": 446590
    },
    {
      "epoch": 0.14187502716377767,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001106912691269127,
      "loss": 2.967,
      "step": 446600
    },
    {
      "epoch": 0.14189367039993586,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011068926892689269,
      "loss": 3.003,
      "step": 446610
    },
    {
      "epoch": 0.14191231363609405,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001106872687268727,
      "loss": 3.1051,
      "step": 446620
    },
    {
      "epoch": 0.14193095687225224,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011068526852685268,
      "loss": 3.0887,
      "step": 446630
    },
    {
      "epoch": 0.14194960010841043,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011068326832683269,
      "loss": 3.1426,
      "step": 446640
    },
    {
      "epoch": 0.1419682433445686,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011068126812681267,
      "loss": 3.1238,
      "step": 446650
    },
    {
      "epoch": 0.14198688658072678,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011067926792679268,
      "loss": 3.083,
      "step": 446660
    },
    {
      "epoch": 0.14200552981688497,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011067726772677269,
      "loss": 3.0312,
      "step": 446670
    },
    {
      "epoch": 0.14202417305304316,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011067526752675268,
      "loss": 3.1291,
      "step": 446680
    },
    {
      "epoch": 0.14204281628920135,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011067326732673269,
      "loss": 3.029,
      "step": 446690
    },
    {
      "epoch": 0.1420614595253595,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00011067126712671267,
      "loss": 3.1883,
      "step": 446700
    },
    {
      "epoch": 0.1420801027615177,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011066926692669268,
      "loss": 3.118,
      "step": 446710
    },
    {
      "epoch": 0.14209874599767588,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011066726672667267,
      "loss": 3.2699,
      "step": 446720
    },
    {
      "epoch": 0.14211738923383407,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011066526652665268,
      "loss": 3.118,
      "step": 446730
    },
    {
      "epoch": 0.14213603246999226,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011066326632663267,
      "loss": 3.1676,
      "step": 446740
    },
    {
      "epoch": 0.14215467570615045,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011066126612661267,
      "loss": 3.1239,
      "step": 446750
    },
    {
      "epoch": 0.1421733189423086,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011065926592659266,
      "loss": 3.1314,
      "step": 446760
    },
    {
      "epoch": 0.1421919621784668,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011065726572657267,
      "loss": 3.1889,
      "step": 446770
    },
    {
      "epoch": 0.142210605414625,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011065526552655265,
      "loss": 3.1337,
      "step": 446780
    },
    {
      "epoch": 0.14222924865078318,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011065326532653266,
      "loss": 3.1934,
      "step": 446790
    },
    {
      "epoch": 0.14224789188694137,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011065126512651264,
      "loss": 3.1552,
      "step": 446800
    },
    {
      "epoch": 0.14226653512309953,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011064926492649265,
      "loss": 3.1444,
      "step": 446810
    },
    {
      "epoch": 0.14228517835925772,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011064726472647265,
      "loss": 3.1408,
      "step": 446820
    },
    {
      "epoch": 0.1423038215954159,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011064526452645266,
      "loss": 3.1008,
      "step": 446830
    },
    {
      "epoch": 0.1423224648315741,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011064326432643264,
      "loss": 3.1505,
      "step": 446840
    },
    {
      "epoch": 0.14234110806773229,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011064126412641265,
      "loss": 2.9771,
      "step": 446850
    },
    {
      "epoch": 0.14235975130389045,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011063926392639266,
      "loss": 3.1612,
      "step": 446860
    },
    {
      "epoch": 0.14237839454004864,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011063726372637264,
      "loss": 3.165,
      "step": 446870
    },
    {
      "epoch": 0.14239703777620683,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011063526352635265,
      "loss": 3.0255,
      "step": 446880
    },
    {
      "epoch": 0.14241568101236501,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011063326332633263,
      "loss": 3.2569,
      "step": 446890
    },
    {
      "epoch": 0.1424343242485232,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011063126312631264,
      "loss": 3.0262,
      "step": 446900
    },
    {
      "epoch": 0.14245296748468136,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011062926292629264,
      "loss": 3.1746,
      "step": 446910
    },
    {
      "epoch": 0.14247161072083955,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011062726272627265,
      "loss": 3.2241,
      "step": 446920
    },
    {
      "epoch": 0.14249025395699774,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011062526252625263,
      "loss": 3.1257,
      "step": 446930
    },
    {
      "epoch": 0.14250889719315593,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011062326232623264,
      "loss": 3.0887,
      "step": 446940
    },
    {
      "epoch": 0.14252754042931412,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011062126212621262,
      "loss": 3.2622,
      "step": 446950
    },
    {
      "epoch": 0.1425461836654723,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011061926192619263,
      "loss": 3.1016,
      "step": 446960
    },
    {
      "epoch": 0.14256482690163047,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011061726172617261,
      "loss": 3.1924,
      "step": 446970
    },
    {
      "epoch": 0.14258347013778866,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011061526152615262,
      "loss": 3.1767,
      "step": 446980
    },
    {
      "epoch": 0.14260211337394685,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011061326132613261,
      "loss": 3.2516,
      "step": 446990
    },
    {
      "epoch": 0.14262075661010504,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011061126112611262,
      "loss": 3.039,
      "step": 447000
    },
    {
      "epoch": 0.14263939984626323,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001106092609260926,
      "loss": 3.1211,
      "step": 447010
    },
    {
      "epoch": 0.1426580430824214,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011060726072607262,
      "loss": 3.2869,
      "step": 447020
    },
    {
      "epoch": 0.14267668631857958,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001106052605260526,
      "loss": 3.1483,
      "step": 447030
    },
    {
      "epoch": 0.14269532955473777,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001106032603260326,
      "loss": 3.0648,
      "step": 447040
    },
    {
      "epoch": 0.14271397279089595,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001106012601260126,
      "loss": 3.1108,
      "step": 447050
    },
    {
      "epoch": 0.14273261602705414,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011059925992599261,
      "loss": 3.1256,
      "step": 447060
    },
    {
      "epoch": 0.1427512592632123,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001105972597259726,
      "loss": 3.0877,
      "step": 447070
    },
    {
      "epoch": 0.1427699024993705,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001105952595259526,
      "loss": 3.0687,
      "step": 447080
    },
    {
      "epoch": 0.14278854573552868,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011059325932593261,
      "loss": 3.138,
      "step": 447090
    },
    {
      "epoch": 0.14280718897168687,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001105912591259126,
      "loss": 3.1368,
      "step": 447100
    },
    {
      "epoch": 0.14282583220784506,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001105892589258926,
      "loss": 3.1624,
      "step": 447110
    },
    {
      "epoch": 0.14284447544400322,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011058725872587258,
      "loss": 3.0246,
      "step": 447120
    },
    {
      "epoch": 0.1428631186801614,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001105852585258526,
      "loss": 3.2258,
      "step": 447130
    },
    {
      "epoch": 0.1428817619163196,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011058325832583259,
      "loss": 3.1561,
      "step": 447140
    },
    {
      "epoch": 0.1429004051524778,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001105812581258126,
      "loss": 3.1105,
      "step": 447150
    },
    {
      "epoch": 0.14291904838863598,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011057925792579258,
      "loss": 3.0857,
      "step": 447160
    },
    {
      "epoch": 0.14293769162479414,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011057725772577259,
      "loss": 3.0849,
      "step": 447170
    },
    {
      "epoch": 0.14295633486095233,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011057525752575257,
      "loss": 3.0836,
      "step": 447180
    },
    {
      "epoch": 0.14297497809711052,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011057325732573258,
      "loss": 3.0921,
      "step": 447190
    },
    {
      "epoch": 0.1429936213332687,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011057125712571256,
      "loss": 3.174,
      "step": 447200
    },
    {
      "epoch": 0.1430122645694269,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011056925692569257,
      "loss": 3.1682,
      "step": 447210
    },
    {
      "epoch": 0.14303090780558508,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011056725672567257,
      "loss": 3.2123,
      "step": 447220
    },
    {
      "epoch": 0.14304955104174324,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011056525652565258,
      "loss": 3.1093,
      "step": 447230
    },
    {
      "epoch": 0.14306819427790143,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011056325632563256,
      "loss": 3.1585,
      "step": 447240
    },
    {
      "epoch": 0.14308683751405962,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011056125612561257,
      "loss": 3.0461,
      "step": 447250
    },
    {
      "epoch": 0.1431054807502178,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011055925592559258,
      "loss": 3.122,
      "step": 447260
    },
    {
      "epoch": 0.143124123986376,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011055725572557256,
      "loss": 3.0491,
      "step": 447270
    },
    {
      "epoch": 0.14314276722253416,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011055525552555257,
      "loss": 2.9696,
      "step": 447280
    },
    {
      "epoch": 0.14316141045869235,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011055325532553255,
      "loss": 3.0798,
      "step": 447290
    },
    {
      "epoch": 0.14318005369485054,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011055125512551256,
      "loss": 3.0984,
      "step": 447300
    },
    {
      "epoch": 0.14319869693100873,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011054925492549256,
      "loss": 3.1606,
      "step": 447310
    },
    {
      "epoch": 0.14321734016716692,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011054725472547256,
      "loss": 3.2625,
      "step": 447320
    },
    {
      "epoch": 0.14323598340332508,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011054525452545255,
      "loss": 3.1384,
      "step": 447330
    },
    {
      "epoch": 0.14325462663948327,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011054325432543256,
      "loss": 3.0369,
      "step": 447340
    },
    {
      "epoch": 0.14327326987564146,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011054125412541254,
      "loss": 3.1487,
      "step": 447350
    },
    {
      "epoch": 0.14329191311179965,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00011053925392539255,
      "loss": 3.0901,
      "step": 447360
    },
    {
      "epoch": 0.14331055634795783,
      "grad_norm": 0.75,
      "learning_rate": 0.00011053725372537254,
      "loss": 3.193,
      "step": 447370
    },
    {
      "epoch": 0.143329199584116,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011053525352535254,
      "loss": 3.0073,
      "step": 447380
    },
    {
      "epoch": 0.14334784282027419,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011053325332533253,
      "loss": 3.2893,
      "step": 447390
    },
    {
      "epoch": 0.14336648605643237,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011053125312531254,
      "loss": 3.2314,
      "step": 447400
    },
    {
      "epoch": 0.14338512929259056,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011052925292529252,
      "loss": 3.1188,
      "step": 447410
    },
    {
      "epoch": 0.14340377252874875,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011052725272527253,
      "loss": 3.1237,
      "step": 447420
    },
    {
      "epoch": 0.14342241576490694,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011052525252525252,
      "loss": 3.1128,
      "step": 447430
    },
    {
      "epoch": 0.1434410590010651,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011052325232523252,
      "loss": 3.0369,
      "step": 447440
    },
    {
      "epoch": 0.1434597022372233,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011052125212521252,
      "loss": 3.104,
      "step": 447450
    },
    {
      "epoch": 0.14347834547338148,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011051925192519253,
      "loss": 3.2371,
      "step": 447460
    },
    {
      "epoch": 0.14349698870953967,
      "grad_norm": 0.625,
      "learning_rate": 0.00011051725172517254,
      "loss": 3.21,
      "step": 447470
    },
    {
      "epoch": 0.14351563194569786,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011051525152515252,
      "loss": 3.0801,
      "step": 447480
    },
    {
      "epoch": 0.14353427518185602,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011051325132513253,
      "loss": 2.9465,
      "step": 447490
    },
    {
      "epoch": 0.1435529184180142,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011051125112511251,
      "loss": 3.0147,
      "step": 447500
    },
    {
      "epoch": 0.1435715616541724,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011050925092509252,
      "loss": 3.1029,
      "step": 447510
    },
    {
      "epoch": 0.14359020489033059,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001105072507250725,
      "loss": 3.1271,
      "step": 447520
    },
    {
      "epoch": 0.14360884812648878,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011050525052505251,
      "loss": 3.1342,
      "step": 447530
    },
    {
      "epoch": 0.14362749136264694,
      "grad_norm": 0.75,
      "learning_rate": 0.00011050325032503251,
      "loss": 3.2582,
      "step": 447540
    },
    {
      "epoch": 0.14364613459880513,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011050125012501252,
      "loss": 3.0676,
      "step": 447550
    },
    {
      "epoch": 0.14366477783496331,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001104992499249925,
      "loss": 3.012,
      "step": 447560
    },
    {
      "epoch": 0.1436834210711215,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011049724972497251,
      "loss": 3.2203,
      "step": 447570
    },
    {
      "epoch": 0.1437020643072797,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011049524952495249,
      "loss": 3.106,
      "step": 447580
    },
    {
      "epoch": 0.14372070754343785,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001104932493249325,
      "loss": 3.0953,
      "step": 447590
    },
    {
      "epoch": 0.14373935077959604,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011049124912491248,
      "loss": 3.2005,
      "step": 447600
    },
    {
      "epoch": 0.14375799401575423,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011048924892489249,
      "loss": 3.0864,
      "step": 447610
    },
    {
      "epoch": 0.14377663725191242,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011048724872487249,
      "loss": 3.1268,
      "step": 447620
    },
    {
      "epoch": 0.1437952804880706,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001104852485248525,
      "loss": 3.1976,
      "step": 447630
    },
    {
      "epoch": 0.1438139237242288,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011048324832483248,
      "loss": 3.0476,
      "step": 447640
    },
    {
      "epoch": 0.14383256696038696,
      "grad_norm": 0.75,
      "learning_rate": 0.00011048124812481249,
      "loss": 3.113,
      "step": 447650
    },
    {
      "epoch": 0.14385121019654515,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001104792479247925,
      "loss": 3.1261,
      "step": 447660
    },
    {
      "epoch": 0.14386985343270334,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011047724772477248,
      "loss": 3.0704,
      "step": 447670
    },
    {
      "epoch": 0.14388849666886153,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011047524752475249,
      "loss": 3.0007,
      "step": 447680
    },
    {
      "epoch": 0.14390713990501972,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011047324732473248,
      "loss": 3.09,
      "step": 447690
    },
    {
      "epoch": 0.14392578314117788,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00011047124712471248,
      "loss": 3.0772,
      "step": 447700
    },
    {
      "epoch": 0.14394442637733607,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011046924692469247,
      "loss": 3.2675,
      "step": 447710
    },
    {
      "epoch": 0.14396306961349425,
      "grad_norm": 0.75,
      "learning_rate": 0.00011046724672467248,
      "loss": 2.9939,
      "step": 447720
    },
    {
      "epoch": 0.14398171284965244,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011046524652465246,
      "loss": 3.1474,
      "step": 447730
    },
    {
      "epoch": 0.14400035608581063,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011046324632463247,
      "loss": 3.1991,
      "step": 447740
    },
    {
      "epoch": 0.1440189993219688,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011046124612461246,
      "loss": 3.1656,
      "step": 447750
    },
    {
      "epoch": 0.14403764255812698,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011045924592459247,
      "loss": 3.1351,
      "step": 447760
    },
    {
      "epoch": 0.14405628579428517,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011045724572457246,
      "loss": 3.1675,
      "step": 447770
    },
    {
      "epoch": 0.14407492903044336,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011045524552455247,
      "loss": 3.1405,
      "step": 447780
    },
    {
      "epoch": 0.14409357226660155,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011045324532453245,
      "loss": 3.0996,
      "step": 447790
    },
    {
      "epoch": 0.1441122155027597,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011045124512451246,
      "loss": 3.1501,
      "step": 447800
    },
    {
      "epoch": 0.1441308587389179,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011044924492449244,
      "loss": 3.0474,
      "step": 447810
    },
    {
      "epoch": 0.1441495019750761,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011044724472447245,
      "loss": 3.0586,
      "step": 447820
    },
    {
      "epoch": 0.14416814521123428,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011044524452445243,
      "loss": 3.023,
      "step": 447830
    },
    {
      "epoch": 0.14418678844739247,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011044324432443244,
      "loss": 3.0917,
      "step": 447840
    },
    {
      "epoch": 0.14420543168355063,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011044124412441245,
      "loss": 2.9947,
      "step": 447850
    },
    {
      "epoch": 0.14422407491970882,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011043924392439245,
      "loss": 3.113,
      "step": 447860
    },
    {
      "epoch": 0.144242718155867,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011043724372437246,
      "loss": 3.1805,
      "step": 447870
    },
    {
      "epoch": 0.1442613613920252,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011043524352435244,
      "loss": 3.1437,
      "step": 447880
    },
    {
      "epoch": 0.14428000462818338,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011043324332433245,
      "loss": 3.2064,
      "step": 447890
    },
    {
      "epoch": 0.14429864786434157,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011043124312431243,
      "loss": 3.0583,
      "step": 447900
    },
    {
      "epoch": 0.14431729110049973,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011042924292429244,
      "loss": 3.1307,
      "step": 447910
    },
    {
      "epoch": 0.14433593433665792,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011042724272427242,
      "loss": 3.0771,
      "step": 447920
    },
    {
      "epoch": 0.1443545775728161,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011042524252425243,
      "loss": 3.1309,
      "step": 447930
    },
    {
      "epoch": 0.1443732208089743,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011042324232423243,
      "loss": 3.1243,
      "step": 447940
    },
    {
      "epoch": 0.1443918640451325,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011042124212421244,
      "loss": 3.0795,
      "step": 447950
    },
    {
      "epoch": 0.14441050728129065,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011041924192419242,
      "loss": 3.2452,
      "step": 447960
    },
    {
      "epoch": 0.14442915051744884,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011041724172417243,
      "loss": 3.0527,
      "step": 447970
    },
    {
      "epoch": 0.14444779375360703,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011041524152415241,
      "loss": 3.108,
      "step": 447980
    },
    {
      "epoch": 0.14446643698976522,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011041324132413242,
      "loss": 3.1518,
      "step": 447990
    },
    {
      "epoch": 0.1444850802259234,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011041124112411241,
      "loss": 3.131,
      "step": 448000
    },
    {
      "epoch": 0.14450372346208157,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011040924092409242,
      "loss": 3.1396,
      "step": 448010
    },
    {
      "epoch": 0.14452236669823976,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001104072407240724,
      "loss": 3.1692,
      "step": 448020
    },
    {
      "epoch": 0.14454100993439795,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011040524052405241,
      "loss": 3.1619,
      "step": 448030
    },
    {
      "epoch": 0.14455965317055613,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001104032403240324,
      "loss": 3.1158,
      "step": 448040
    },
    {
      "epoch": 0.14457829640671432,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001104012401240124,
      "loss": 3.1559,
      "step": 448050
    },
    {
      "epoch": 0.14459693964287249,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011039923992399241,
      "loss": 3.0514,
      "step": 448060
    },
    {
      "epoch": 0.14461558287903067,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001103972397239724,
      "loss": 3.0106,
      "step": 448070
    },
    {
      "epoch": 0.14463422611518886,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001103952395239524,
      "loss": 3.0406,
      "step": 448080
    },
    {
      "epoch": 0.14465286935134705,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001103932393239324,
      "loss": 3.1797,
      "step": 448090
    },
    {
      "epoch": 0.14467151258750524,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011039123912391241,
      "loss": 3.1556,
      "step": 448100
    },
    {
      "epoch": 0.14469015582366343,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011038923892389239,
      "loss": 3.2005,
      "step": 448110
    },
    {
      "epoch": 0.1447087990598216,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001103872387238724,
      "loss": 3.0254,
      "step": 448120
    },
    {
      "epoch": 0.14472744229597978,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011038523852385238,
      "loss": 3.1296,
      "step": 448130
    },
    {
      "epoch": 0.14474608553213797,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011038323832383239,
      "loss": 3.119,
      "step": 448140
    },
    {
      "epoch": 0.14476472876829616,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011038123812381237,
      "loss": 3.0918,
      "step": 448150
    },
    {
      "epoch": 0.14478337200445435,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011037923792379238,
      "loss": 3.1777,
      "step": 448160
    },
    {
      "epoch": 0.1448020152406125,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011037723772377238,
      "loss": 3.1885,
      "step": 448170
    },
    {
      "epoch": 0.1448206584767707,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011037523752375239,
      "loss": 3.1053,
      "step": 448180
    },
    {
      "epoch": 0.1448393017129289,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011037323732373237,
      "loss": 3.0907,
      "step": 448190
    },
    {
      "epoch": 0.14485794494908708,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011037123712371238,
      "loss": 3.1578,
      "step": 448200
    },
    {
      "epoch": 0.14487658818524526,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011036923692369236,
      "loss": 3.1595,
      "step": 448210
    },
    {
      "epoch": 0.14489523142140343,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011036723672367237,
      "loss": 3.1626,
      "step": 448220
    },
    {
      "epoch": 0.14491387465756161,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011036523652365237,
      "loss": 3.0634,
      "step": 448230
    },
    {
      "epoch": 0.1449325178937198,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011036323632363236,
      "loss": 2.9517,
      "step": 448240
    },
    {
      "epoch": 0.144951161129878,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011036123612361237,
      "loss": 3.2603,
      "step": 448250
    },
    {
      "epoch": 0.14496980436603618,
      "grad_norm": 0.75,
      "learning_rate": 0.00011035923592359237,
      "loss": 3.3373,
      "step": 448260
    },
    {
      "epoch": 0.14498844760219434,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011035723572357238,
      "loss": 3.0439,
      "step": 448270
    },
    {
      "epoch": 0.14500709083835253,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011035523552355236,
      "loss": 3.1141,
      "step": 448280
    },
    {
      "epoch": 0.14502573407451072,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011035323532353237,
      "loss": 3.2192,
      "step": 448290
    },
    {
      "epoch": 0.1450443773106689,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011035123512351235,
      "loss": 3.248,
      "step": 448300
    },
    {
      "epoch": 0.1450630205468271,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011034923492349236,
      "loss": 3.0337,
      "step": 448310
    },
    {
      "epoch": 0.1450816637829853,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011034723472347235,
      "loss": 3.2013,
      "step": 448320
    },
    {
      "epoch": 0.14510030701914345,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011034523452345236,
      "loss": 3.0989,
      "step": 448330
    },
    {
      "epoch": 0.14511895025530164,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011034323432343235,
      "loss": 3.1945,
      "step": 448340
    },
    {
      "epoch": 0.14513759349145983,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011034123412341235,
      "loss": 3.0177,
      "step": 448350
    },
    {
      "epoch": 0.14515623672761802,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011033923392339234,
      "loss": 3.0016,
      "step": 448360
    },
    {
      "epoch": 0.1451748799637762,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011033723372337235,
      "loss": 3.1832,
      "step": 448370
    },
    {
      "epoch": 0.14519352319993437,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011033523352335233,
      "loss": 3.153,
      "step": 448380
    },
    {
      "epoch": 0.14521216643609255,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011033323332333234,
      "loss": 3.0486,
      "step": 448390
    },
    {
      "epoch": 0.14523080967225074,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011033123312331233,
      "loss": 3.1347,
      "step": 448400
    },
    {
      "epoch": 0.14524945290840893,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011032923292329234,
      "loss": 3.099,
      "step": 448410
    },
    {
      "epoch": 0.14526809614456712,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011032723272327232,
      "loss": 3.1025,
      "step": 448420
    },
    {
      "epoch": 0.14528673938072528,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011032523252325233,
      "loss": 3.0952,
      "step": 448430
    },
    {
      "epoch": 0.14530538261688347,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011032323232323234,
      "loss": 3.124,
      "step": 448440
    },
    {
      "epoch": 0.14532402585304166,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011032123212321232,
      "loss": 3.0984,
      "step": 448450
    },
    {
      "epoch": 0.14534266908919985,
      "grad_norm": 0.75,
      "learning_rate": 0.00011031923192319233,
      "loss": 3.0906,
      "step": 448460
    },
    {
      "epoch": 0.14536131232535804,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011031723172317232,
      "loss": 3.212,
      "step": 448470
    },
    {
      "epoch": 0.1453799555615162,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011031523152315232,
      "loss": 3.2427,
      "step": 448480
    },
    {
      "epoch": 0.1453985987976744,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011031323132313232,
      "loss": 3.047,
      "step": 448490
    },
    {
      "epoch": 0.14541724203383258,
      "grad_norm": 0.75,
      "learning_rate": 0.00011031123112311233,
      "loss": 3.0242,
      "step": 448500
    },
    {
      "epoch": 0.14543588526999077,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011030923092309231,
      "loss": 3.1937,
      "step": 448510
    },
    {
      "epoch": 0.14545452850614896,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011030723072307232,
      "loss": 2.9842,
      "step": 448520
    },
    {
      "epoch": 0.14547317174230712,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001103052305230523,
      "loss": 3.2016,
      "step": 448530
    },
    {
      "epoch": 0.1454918149784653,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011030323032303231,
      "loss": 3.0373,
      "step": 448540
    },
    {
      "epoch": 0.1455104582146235,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011030123012301231,
      "loss": 3.2296,
      "step": 448550
    },
    {
      "epoch": 0.14552910145078168,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001102992299229923,
      "loss": 3.1249,
      "step": 448560
    },
    {
      "epoch": 0.14554774468693987,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001102972297229723,
      "loss": 3.1564,
      "step": 448570
    },
    {
      "epoch": 0.14556638792309806,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011029522952295231,
      "loss": 3.0322,
      "step": 448580
    },
    {
      "epoch": 0.14558503115925622,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011029322932293229,
      "loss": 3.1886,
      "step": 448590
    },
    {
      "epoch": 0.1456036743954144,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001102912291229123,
      "loss": 3.1606,
      "step": 448600
    },
    {
      "epoch": 0.1456223176315726,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011028922892289228,
      "loss": 3.1035,
      "step": 448610
    },
    {
      "epoch": 0.1456409608677308,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011028722872287229,
      "loss": 3.1212,
      "step": 448620
    },
    {
      "epoch": 0.14565960410388898,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011028522852285229,
      "loss": 3.0694,
      "step": 448630
    },
    {
      "epoch": 0.14567824734004714,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001102832283228323,
      "loss": 3.1357,
      "step": 448640
    },
    {
      "epoch": 0.14569689057620533,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001102812281228123,
      "loss": 3.0568,
      "step": 448650
    },
    {
      "epoch": 0.14571553381236352,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011027922792279229,
      "loss": 3.1093,
      "step": 448660
    },
    {
      "epoch": 0.1457341770485217,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001102772277227723,
      "loss": 3.1042,
      "step": 448670
    },
    {
      "epoch": 0.1457528202846799,
      "grad_norm": 0.890625,
      "learning_rate": 0.00011027522752275228,
      "loss": 3.0904,
      "step": 448680
    },
    {
      "epoch": 0.14577146352083806,
      "grad_norm": 0.75,
      "learning_rate": 0.00011027322732273229,
      "loss": 3.0992,
      "step": 448690
    },
    {
      "epoch": 0.14579010675699625,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011027122712271227,
      "loss": 3.1253,
      "step": 448700
    },
    {
      "epoch": 0.14580874999315444,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011026922692269228,
      "loss": 3.0607,
      "step": 448710
    },
    {
      "epoch": 0.14582739322931262,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011026722672267227,
      "loss": 3.1227,
      "step": 448720
    },
    {
      "epoch": 0.1458460364654708,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011026522652265228,
      "loss": 3.0862,
      "step": 448730
    },
    {
      "epoch": 0.14586467970162897,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011026322632263226,
      "loss": 3.2117,
      "step": 448740
    },
    {
      "epoch": 0.14588332293778716,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011026122612261227,
      "loss": 3.2303,
      "step": 448750
    },
    {
      "epoch": 0.14590196617394535,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011025922592259226,
      "loss": 3.0291,
      "step": 448760
    },
    {
      "epoch": 0.14592060941010354,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011025722572257226,
      "loss": 3.0283,
      "step": 448770
    },
    {
      "epoch": 0.14593925264626173,
      "grad_norm": 0.75,
      "learning_rate": 0.00011025522552255225,
      "loss": 3.0525,
      "step": 448780
    },
    {
      "epoch": 0.14595789588241992,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011025322532253226,
      "loss": 3.0898,
      "step": 448790
    },
    {
      "epoch": 0.14597653911857808,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011025122512251225,
      "loss": 2.9873,
      "step": 448800
    },
    {
      "epoch": 0.14599518235473627,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011024922492249226,
      "loss": 3.1024,
      "step": 448810
    },
    {
      "epoch": 0.14601382559089446,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011024722472247224,
      "loss": 3.1344,
      "step": 448820
    },
    {
      "epoch": 0.14603246882705265,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00011024522452245225,
      "loss": 3.1602,
      "step": 448830
    },
    {
      "epoch": 0.14605111206321084,
      "grad_norm": 0.75,
      "learning_rate": 0.00011024322432243226,
      "loss": 3.1076,
      "step": 448840
    },
    {
      "epoch": 0.146069755299369,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011024122412241224,
      "loss": 3.1005,
      "step": 448850
    },
    {
      "epoch": 0.1460883985355272,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011023922392239225,
      "loss": 3.1265,
      "step": 448860
    },
    {
      "epoch": 0.14610704177168538,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011023722372237223,
      "loss": 3.0965,
      "step": 448870
    },
    {
      "epoch": 0.14612568500784356,
      "grad_norm": 0.75,
      "learning_rate": 0.00011023522352235224,
      "loss": 3.1049,
      "step": 448880
    },
    {
      "epoch": 0.14614432824400175,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00011023322332233224,
      "loss": 3.1265,
      "step": 448890
    },
    {
      "epoch": 0.14616297148015991,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011023122312231225,
      "loss": 3.2109,
      "step": 448900
    },
    {
      "epoch": 0.1461816147163181,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011022922292229223,
      "loss": 3.1943,
      "step": 448910
    },
    {
      "epoch": 0.1462002579524763,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011022722272227224,
      "loss": 3.1557,
      "step": 448920
    },
    {
      "epoch": 0.14621890118863448,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011022522252225222,
      "loss": 3.089,
      "step": 448930
    },
    {
      "epoch": 0.14623754442479267,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011022322232223223,
      "loss": 3.1459,
      "step": 448940
    },
    {
      "epoch": 0.14625618766095083,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011022122212221223,
      "loss": 3.0677,
      "step": 448950
    },
    {
      "epoch": 0.14627483089710902,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011021922192219223,
      "loss": 3.1408,
      "step": 448960
    },
    {
      "epoch": 0.1462934741332672,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011021722172217222,
      "loss": 3.1535,
      "step": 448970
    },
    {
      "epoch": 0.1463121173694254,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011021522152215223,
      "loss": 3.1477,
      "step": 448980
    },
    {
      "epoch": 0.1463307606055836,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011021322132213221,
      "loss": 3.0205,
      "step": 448990
    },
    {
      "epoch": 0.14634940384174178,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011021122112211222,
      "loss": 3.187,
      "step": 449000
    },
    {
      "epoch": 0.14636804707789994,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001102092209220922,
      "loss": 3.1234,
      "step": 449010
    },
    {
      "epoch": 0.14638669031405813,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011020722072207221,
      "loss": 3.0998,
      "step": 449020
    },
    {
      "epoch": 0.14640533355021632,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011020522052205222,
      "loss": 3.0856,
      "step": 449030
    },
    {
      "epoch": 0.1464239767863745,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011020322032203221,
      "loss": 3.0775,
      "step": 449040
    },
    {
      "epoch": 0.1464426200225327,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011020122012201222,
      "loss": 3.2112,
      "step": 449050
    },
    {
      "epoch": 0.14646126325869085,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001101992199219922,
      "loss": 3.1532,
      "step": 449060
    },
    {
      "epoch": 0.14647990649484904,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00011019721972197221,
      "loss": 3.1243,
      "step": 449070
    },
    {
      "epoch": 0.14649854973100723,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001101952195219522,
      "loss": 3.2923,
      "step": 449080
    },
    {
      "epoch": 0.14651719296716542,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001101932193219322,
      "loss": 3.169,
      "step": 449090
    },
    {
      "epoch": 0.1465358362033236,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011019121912191219,
      "loss": 3.1302,
      "step": 449100
    },
    {
      "epoch": 0.14655447943948177,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001101892189218922,
      "loss": 3.1206,
      "step": 449110
    },
    {
      "epoch": 0.14657312267563996,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011018721872187219,
      "loss": 2.9831,
      "step": 449120
    },
    {
      "epoch": 0.14659176591179815,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001101852185218522,
      "loss": 3.0915,
      "step": 449130
    },
    {
      "epoch": 0.14661040914795634,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00011018321832183218,
      "loss": 3.0737,
      "step": 449140
    },
    {
      "epoch": 0.14662905238411453,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011018121812181219,
      "loss": 3.0875,
      "step": 449150
    },
    {
      "epoch": 0.1466476956202727,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011017921792179217,
      "loss": 3.0102,
      "step": 449160
    },
    {
      "epoch": 0.14666633885643088,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011017721772177218,
      "loss": 3.1697,
      "step": 449170
    },
    {
      "epoch": 0.14668498209258907,
      "grad_norm": 0.75,
      "learning_rate": 0.00011017521752175218,
      "loss": 3.1321,
      "step": 449180
    },
    {
      "epoch": 0.14670362532874726,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011017321732173217,
      "loss": 3.1038,
      "step": 449190
    },
    {
      "epoch": 0.14672226856490544,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00011017121712171217,
      "loss": 2.9624,
      "step": 449200
    },
    {
      "epoch": 0.14674091180106363,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011016921692169218,
      "loss": 3.157,
      "step": 449210
    },
    {
      "epoch": 0.1467595550372218,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011016721672167216,
      "loss": 3.0118,
      "step": 449220
    },
    {
      "epoch": 0.14677819827337998,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00011016521652165217,
      "loss": 3.1425,
      "step": 449230
    },
    {
      "epoch": 0.14679684150953817,
      "grad_norm": 0.828125,
      "learning_rate": 0.00011016321632163218,
      "loss": 3.1091,
      "step": 449240
    },
    {
      "epoch": 0.14681548474569636,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011016121612161216,
      "loss": 3.1124,
      "step": 449250
    },
    {
      "epoch": 0.14683412798185455,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011015921592159217,
      "loss": 3.1435,
      "step": 449260
    },
    {
      "epoch": 0.1468527712180127,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011015721572157217,
      "loss": 3.0195,
      "step": 449270
    },
    {
      "epoch": 0.1468714144541709,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011015521552155218,
      "loss": 3.2074,
      "step": 449280
    },
    {
      "epoch": 0.1468900576903291,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011015321532153216,
      "loss": 2.9548,
      "step": 449290
    },
    {
      "epoch": 0.14690870092648728,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00011015121512151217,
      "loss": 3.0853,
      "step": 449300
    },
    {
      "epoch": 0.14692734416264547,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011014921492149215,
      "loss": 3.1078,
      "step": 449310
    },
    {
      "epoch": 0.14694598739880363,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011014721472147216,
      "loss": 3.0433,
      "step": 449320
    },
    {
      "epoch": 0.14696463063496182,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011014521452145214,
      "loss": 3.2294,
      "step": 449330
    },
    {
      "epoch": 0.14698327387112,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011014321432143215,
      "loss": 2.9937,
      "step": 449340
    },
    {
      "epoch": 0.1470019171072782,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011014121412141214,
      "loss": 3.158,
      "step": 449350
    },
    {
      "epoch": 0.14702056034343639,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011013921392139215,
      "loss": 3.0365,
      "step": 449360
    },
    {
      "epoch": 0.14703920357959455,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00011013721372137214,
      "loss": 3.221,
      "step": 449370
    },
    {
      "epoch": 0.14705784681575274,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011013521352135214,
      "loss": 3.0769,
      "step": 449380
    },
    {
      "epoch": 0.14707649005191092,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011013321332133213,
      "loss": 3.0636,
      "step": 449390
    },
    {
      "epoch": 0.1470951332880691,
      "grad_norm": 0.671875,
      "learning_rate": 0.00011013121312131214,
      "loss": 3.195,
      "step": 449400
    },
    {
      "epoch": 0.1471137765242273,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011012921292129212,
      "loss": 3.0561,
      "step": 449410
    },
    {
      "epoch": 0.14713241976038546,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011012721272127213,
      "loss": 3.0987,
      "step": 449420
    },
    {
      "epoch": 0.14715106299654365,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011012521252125214,
      "loss": 3.1225,
      "step": 449430
    },
    {
      "epoch": 0.14716970623270184,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011012321232123213,
      "loss": 3.1656,
      "step": 449440
    },
    {
      "epoch": 0.14718834946886003,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00011012121212121214,
      "loss": 3.1764,
      "step": 449450
    },
    {
      "epoch": 0.14720699270501822,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011011921192119212,
      "loss": 3.1816,
      "step": 449460
    },
    {
      "epoch": 0.1472256359411764,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011011721172117213,
      "loss": 3.1587,
      "step": 449470
    },
    {
      "epoch": 0.14724427917733457,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011011521152115211,
      "loss": 3.2003,
      "step": 449480
    },
    {
      "epoch": 0.14726292241349276,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011011321132113212,
      "loss": 3.0027,
      "step": 449490
    },
    {
      "epoch": 0.14728156564965095,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011011121112111212,
      "loss": 3.0964,
      "step": 449500
    },
    {
      "epoch": 0.14730020888580914,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011010921092109211,
      "loss": 3.054,
      "step": 449510
    },
    {
      "epoch": 0.14731885212196733,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011010721072107211,
      "loss": 3.0914,
      "step": 449520
    },
    {
      "epoch": 0.1473374953581255,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011010521052105212,
      "loss": 3.0797,
      "step": 449530
    },
    {
      "epoch": 0.14735613859428368,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001101032103210321,
      "loss": 3.0768,
      "step": 449540
    },
    {
      "epoch": 0.14737478183044186,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011010121012101211,
      "loss": 3.0861,
      "step": 449550
    },
    {
      "epoch": 0.14739342506660005,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011009920992099209,
      "loss": 3.0041,
      "step": 449560
    },
    {
      "epoch": 0.14741206830275824,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001100972097209721,
      "loss": 3.0737,
      "step": 449570
    },
    {
      "epoch": 0.1474307115389164,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001100952095209521,
      "loss": 3.2522,
      "step": 449580
    },
    {
      "epoch": 0.1474493547750746,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001100932093209321,
      "loss": 3.1087,
      "step": 449590
    },
    {
      "epoch": 0.14746799801123278,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011009120912091209,
      "loss": 3.0646,
      "step": 449600
    },
    {
      "epoch": 0.14748664124739097,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001100892089208921,
      "loss": 3.1466,
      "step": 449610
    },
    {
      "epoch": 0.14750528448354916,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00011008720872087208,
      "loss": 3.0191,
      "step": 449620
    },
    {
      "epoch": 0.14752392771970732,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011008520852085209,
      "loss": 3.0957,
      "step": 449630
    },
    {
      "epoch": 0.1475425709558655,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001100832083208321,
      "loss": 3.0174,
      "step": 449640
    },
    {
      "epoch": 0.1475612141920237,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011008120812081208,
      "loss": 2.9059,
      "step": 449650
    },
    {
      "epoch": 0.1475798574281819,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011007920792079209,
      "loss": 2.9967,
      "step": 449660
    },
    {
      "epoch": 0.14759850066434008,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011007720772077208,
      "loss": 3.1298,
      "step": 449670
    },
    {
      "epoch": 0.14761714390049827,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001100752075207521,
      "loss": 3.1419,
      "step": 449680
    },
    {
      "epoch": 0.14763578713665643,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011007320732073208,
      "loss": 3.2545,
      "step": 449690
    },
    {
      "epoch": 0.14765443037281462,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00011007120712071208,
      "loss": 3.0247,
      "step": 449700
    },
    {
      "epoch": 0.1476730736089728,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011006920692069207,
      "loss": 3.0967,
      "step": 449710
    },
    {
      "epoch": 0.147691716845131,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011006720672067208,
      "loss": 3.1381,
      "step": 449720
    },
    {
      "epoch": 0.14771036008128918,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011006520652065206,
      "loss": 3.122,
      "step": 449730
    },
    {
      "epoch": 0.14772900331744734,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011006320632063207,
      "loss": 3.0512,
      "step": 449740
    },
    {
      "epoch": 0.14774764655360553,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00011006120612061206,
      "loss": 3.0258,
      "step": 449750
    },
    {
      "epoch": 0.14776628978976372,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011005920592059207,
      "loss": 2.9568,
      "step": 449760
    },
    {
      "epoch": 0.1477849330259219,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011005720572057205,
      "loss": 3.1918,
      "step": 449770
    },
    {
      "epoch": 0.1478035762620801,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011005520552055206,
      "loss": 3.1472,
      "step": 449780
    },
    {
      "epoch": 0.14782221949823826,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011005320532053205,
      "loss": 3.0091,
      "step": 449790
    },
    {
      "epoch": 0.14784086273439645,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011005120512051205,
      "loss": 3.1775,
      "step": 449800
    },
    {
      "epoch": 0.14785950597055464,
      "grad_norm": 0.75,
      "learning_rate": 0.00011004920492049205,
      "loss": 3.2168,
      "step": 449810
    },
    {
      "epoch": 0.14787814920671283,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011004720472047206,
      "loss": 3.0837,
      "step": 449820
    },
    {
      "epoch": 0.14789679244287102,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011004520452045205,
      "loss": 3.0124,
      "step": 449830
    },
    {
      "epoch": 0.14791543567902918,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011004320432043205,
      "loss": 3.0391,
      "step": 449840
    },
    {
      "epoch": 0.14793407891518737,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011004120412041206,
      "loss": 3.0748,
      "step": 449850
    },
    {
      "epoch": 0.14795272215134556,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011003920392039204,
      "loss": 3.0682,
      "step": 449860
    },
    {
      "epoch": 0.14797136538750374,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011003720372037205,
      "loss": 3.1679,
      "step": 449870
    },
    {
      "epoch": 0.14799000862366193,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00011003520352035203,
      "loss": 3.2229,
      "step": 449880
    },
    {
      "epoch": 0.14800865185982012,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00011003320332033204,
      "loss": 3.0863,
      "step": 449890
    },
    {
      "epoch": 0.14802729509597828,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011003120312031204,
      "loss": 3.2548,
      "step": 449900
    },
    {
      "epoch": 0.14804593833213647,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011002920292029205,
      "loss": 3.1244,
      "step": 449910
    },
    {
      "epoch": 0.14806458156829466,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011002720272027203,
      "loss": 3.2215,
      "step": 449920
    },
    {
      "epoch": 0.14808322480445285,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00011002520252025204,
      "loss": 2.988,
      "step": 449930
    },
    {
      "epoch": 0.14810186804061104,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011002320232023202,
      "loss": 3.0727,
      "step": 449940
    },
    {
      "epoch": 0.1481205112767692,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011002120212021203,
      "loss": 3.1145,
      "step": 449950
    },
    {
      "epoch": 0.1481391545129274,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011001920192019201,
      "loss": 2.9701,
      "step": 449960
    },
    {
      "epoch": 0.14815779774908558,
      "grad_norm": 0.9375,
      "learning_rate": 0.00011001720172017202,
      "loss": 3.1318,
      "step": 449970
    },
    {
      "epoch": 0.14817644098524377,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011001520152015202,
      "loss": 3.0291,
      "step": 449980
    },
    {
      "epoch": 0.14819508422140196,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011001320132013202,
      "loss": 2.9324,
      "step": 449990
    },
    {
      "epoch": 0.14821372745756012,
      "grad_norm": 0.75,
      "learning_rate": 0.00011001120112011201,
      "loss": 3.0386,
      "step": 450000
    },
    {
      "epoch": 0.1482323706937183,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011000920092009202,
      "loss": 3.1475,
      "step": 450010
    },
    {
      "epoch": 0.1482510139298765,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011000720072007203,
      "loss": 3.1437,
      "step": 450020
    },
    {
      "epoch": 0.14826965716603469,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011000520052005201,
      "loss": 3.0502,
      "step": 450030
    },
    {
      "epoch": 0.14828830040219287,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00011000320032003202,
      "loss": 3.1463,
      "step": 450040
    },
    {
      "epoch": 0.14830694363835104,
      "grad_norm": 0.734375,
      "learning_rate": 0.000110001200120012,
      "loss": 2.9631,
      "step": 450050
    },
    {
      "epoch": 0.14832558687450922,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010999919991999201,
      "loss": 3.1948,
      "step": 450060
    },
    {
      "epoch": 0.1483442301106674,
      "grad_norm": 0.6875,
      "learning_rate": 0.000109997199719972,
      "loss": 3.1088,
      "step": 450070
    },
    {
      "epoch": 0.1483628733468256,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010999519951995201,
      "loss": 3.1628,
      "step": 450080
    },
    {
      "epoch": 0.1483815165829838,
      "grad_norm": 0.734375,
      "learning_rate": 0.000109993199319932,
      "loss": 3.1747,
      "step": 450090
    },
    {
      "epoch": 0.14840015981914195,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000109991199119912,
      "loss": 3.0257,
      "step": 450100
    },
    {
      "epoch": 0.14841880305530014,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010998919891989199,
      "loss": 2.9615,
      "step": 450110
    },
    {
      "epoch": 0.14843744629145833,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000109987198719872,
      "loss": 3.183,
      "step": 450120
    },
    {
      "epoch": 0.14845608952761652,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010998519851985199,
      "loss": 3.175,
      "step": 450130
    },
    {
      "epoch": 0.1484747327637747,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010998319831983199,
      "loss": 3.0404,
      "step": 450140
    },
    {
      "epoch": 0.1484933759999329,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010998119811981198,
      "loss": 3.1542,
      "step": 450150
    },
    {
      "epoch": 0.14851201923609106,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010997919791979199,
      "loss": 3.1416,
      "step": 450160
    },
    {
      "epoch": 0.14853066247224925,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010997719771977197,
      "loss": 3.1154,
      "step": 450170
    },
    {
      "epoch": 0.14854930570840744,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010997519751975198,
      "loss": 3.1098,
      "step": 450180
    },
    {
      "epoch": 0.14856794894456563,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010997319731973196,
      "loss": 3.1049,
      "step": 450190
    },
    {
      "epoch": 0.14858659218072381,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010997119711971197,
      "loss": 3.0917,
      "step": 450200
    },
    {
      "epoch": 0.14860523541688198,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010996919691969197,
      "loss": 2.9608,
      "step": 450210
    },
    {
      "epoch": 0.14862387865304016,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010996719671967198,
      "loss": 3.0982,
      "step": 450220
    },
    {
      "epoch": 0.14864252188919835,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010996519651965199,
      "loss": 3.168,
      "step": 450230
    },
    {
      "epoch": 0.14866116512535654,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010996319631963197,
      "loss": 3.0632,
      "step": 450240
    },
    {
      "epoch": 0.14867980836151473,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010996119611961198,
      "loss": 3.0841,
      "step": 450250
    },
    {
      "epoch": 0.1486984515976729,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010995919591959196,
      "loss": 3.1422,
      "step": 450260
    },
    {
      "epoch": 0.14871709483383108,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010995719571957197,
      "loss": 3.0136,
      "step": 450270
    },
    {
      "epoch": 0.14873573806998927,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010995519551955195,
      "loss": 3.1667,
      "step": 450280
    },
    {
      "epoch": 0.14875438130614746,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010995319531953196,
      "loss": 3.2118,
      "step": 450290
    },
    {
      "epoch": 0.14877302454230565,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010995119511951196,
      "loss": 3.1071,
      "step": 450300
    },
    {
      "epoch": 0.1487916677784638,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010994919491949197,
      "loss": 3.1247,
      "step": 450310
    },
    {
      "epoch": 0.148810311014622,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010994719471947195,
      "loss": 2.9945,
      "step": 450320
    },
    {
      "epoch": 0.1488289542507802,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010994519451945196,
      "loss": 3.0493,
      "step": 450330
    },
    {
      "epoch": 0.14884759748693838,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010994319431943194,
      "loss": 3.0694,
      "step": 450340
    },
    {
      "epoch": 0.14886624072309657,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010994119411941195,
      "loss": 3.1167,
      "step": 450350
    },
    {
      "epoch": 0.14888488395925475,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010993919391939193,
      "loss": 3.1102,
      "step": 450360
    },
    {
      "epoch": 0.14890352719541292,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010993719371937194,
      "loss": 3.1623,
      "step": 450370
    },
    {
      "epoch": 0.1489221704315711,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010993519351935193,
      "loss": 3.1008,
      "step": 450380
    },
    {
      "epoch": 0.1489408136677293,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010993319331933194,
      "loss": 3.2036,
      "step": 450390
    },
    {
      "epoch": 0.14895945690388748,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010993119311931193,
      "loss": 2.9259,
      "step": 450400
    },
    {
      "epoch": 0.14897810014004567,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010992919291929193,
      "loss": 3.1905,
      "step": 450410
    },
    {
      "epoch": 0.14899674337620383,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010992719271927194,
      "loss": 3.07,
      "step": 450420
    },
    {
      "epoch": 0.14901538661236202,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010992519251925193,
      "loss": 3.1224,
      "step": 450430
    },
    {
      "epoch": 0.1490340298485202,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010992319231923194,
      "loss": 3.2042,
      "step": 450440
    },
    {
      "epoch": 0.1490526730846784,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010992119211921193,
      "loss": 3.0129,
      "step": 450450
    },
    {
      "epoch": 0.1490713163208366,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010991919191919193,
      "loss": 3.0369,
      "step": 450460
    },
    {
      "epoch": 0.14908995955699475,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010991719171917192,
      "loss": 3.2587,
      "step": 450470
    },
    {
      "epoch": 0.14910860279315294,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010991519151915193,
      "loss": 2.9659,
      "step": 450480
    },
    {
      "epoch": 0.14912724602931113,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010991319131913191,
      "loss": 2.9825,
      "step": 450490
    },
    {
      "epoch": 0.14914588926546932,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010991119111911192,
      "loss": 3.2139,
      "step": 450500
    },
    {
      "epoch": 0.1491645325016275,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001099091909190919,
      "loss": 3.1951,
      "step": 450510
    },
    {
      "epoch": 0.14918317573778567,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010990719071907191,
      "loss": 3.082,
      "step": 450520
    },
    {
      "epoch": 0.14920181897394386,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010990519051905191,
      "loss": 3.0816,
      "step": 450530
    },
    {
      "epoch": 0.14922046221010205,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010990319031903192,
      "loss": 3.2165,
      "step": 450540
    },
    {
      "epoch": 0.14923910544626023,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001099011901190119,
      "loss": 3.022,
      "step": 450550
    },
    {
      "epoch": 0.14925774868241842,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010989918991899191,
      "loss": 3.111,
      "step": 450560
    },
    {
      "epoch": 0.1492763919185766,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010989718971897189,
      "loss": 3.0759,
      "step": 450570
    },
    {
      "epoch": 0.14929503515473477,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001098951895189519,
      "loss": 3.0335,
      "step": 450580
    },
    {
      "epoch": 0.14931367839089296,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010989318931893188,
      "loss": 3.1039,
      "step": 450590
    },
    {
      "epoch": 0.14933232162705115,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010989118911891189,
      "loss": 3.1774,
      "step": 450600
    },
    {
      "epoch": 0.14935096486320934,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001098891889188919,
      "loss": 3.1976,
      "step": 450610
    },
    {
      "epoch": 0.14936960809936753,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001098871887188719,
      "loss": 3.1509,
      "step": 450620
    },
    {
      "epoch": 0.1493882513355257,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001098851885188519,
      "loss": 3.1703,
      "step": 450630
    },
    {
      "epoch": 0.14940689457168388,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010988318831883189,
      "loss": 3.1355,
      "step": 450640
    },
    {
      "epoch": 0.14942553780784207,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001098811881188119,
      "loss": 3.1063,
      "step": 450650
    },
    {
      "epoch": 0.14944418104400026,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010987918791879188,
      "loss": 2.9942,
      "step": 450660
    },
    {
      "epoch": 0.14946282428015845,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010987718771877189,
      "loss": 3.0243,
      "step": 450670
    },
    {
      "epoch": 0.1494814675163166,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010987518751875187,
      "loss": 3.1168,
      "step": 450680
    },
    {
      "epoch": 0.1495001107524748,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010987318731873188,
      "loss": 3.1521,
      "step": 450690
    },
    {
      "epoch": 0.14951875398863299,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010987118711871187,
      "loss": 3.1718,
      "step": 450700
    },
    {
      "epoch": 0.14953739722479117,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010986918691869188,
      "loss": 3.1003,
      "step": 450710
    },
    {
      "epoch": 0.14955604046094936,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010986718671867187,
      "loss": 3.1525,
      "step": 450720
    },
    {
      "epoch": 0.14957468369710752,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010986518651865188,
      "loss": 3.1661,
      "step": 450730
    },
    {
      "epoch": 0.1495933269332657,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010986318631863186,
      "loss": 3.0621,
      "step": 450740
    },
    {
      "epoch": 0.1496119701694239,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010986118611861187,
      "loss": 3.0391,
      "step": 450750
    },
    {
      "epoch": 0.1496306134055821,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010985918591859186,
      "loss": 3.0836,
      "step": 450760
    },
    {
      "epoch": 0.14964925664174028,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010985718571857187,
      "loss": 3.1442,
      "step": 450770
    },
    {
      "epoch": 0.14966789987789844,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010985518551855185,
      "loss": 3.1031,
      "step": 450780
    },
    {
      "epoch": 0.14968654311405663,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010985318531853186,
      "loss": 3.2296,
      "step": 450790
    },
    {
      "epoch": 0.14970518635021482,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010985118511851184,
      "loss": 3.0903,
      "step": 450800
    },
    {
      "epoch": 0.149723829586373,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010984918491849185,
      "loss": 3.1093,
      "step": 450810
    },
    {
      "epoch": 0.1497424728225312,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010984718471847186,
      "loss": 3.0531,
      "step": 450820
    },
    {
      "epoch": 0.1497611160586894,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010984518451845184,
      "loss": 3.1359,
      "step": 450830
    },
    {
      "epoch": 0.14977975929484755,
      "grad_norm": 0.75,
      "learning_rate": 0.00010984318431843185,
      "loss": 3.1179,
      "step": 450840
    },
    {
      "epoch": 0.14979840253100574,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010984118411841185,
      "loss": 3.0229,
      "step": 450850
    },
    {
      "epoch": 0.14981704576716393,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010983918391839186,
      "loss": 3.2346,
      "step": 450860
    },
    {
      "epoch": 0.14983568900332211,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010983718371837184,
      "loss": 3.0193,
      "step": 450870
    },
    {
      "epoch": 0.1498543322394803,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010983518351835185,
      "loss": 2.9766,
      "step": 450880
    },
    {
      "epoch": 0.14987297547563846,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010983318331833183,
      "loss": 3.0739,
      "step": 450890
    },
    {
      "epoch": 0.14989161871179665,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010983118311831184,
      "loss": 3.1813,
      "step": 450900
    },
    {
      "epoch": 0.14991026194795484,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010982918291829182,
      "loss": 3.1021,
      "step": 450910
    },
    {
      "epoch": 0.14992890518411303,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010982718271827183,
      "loss": 3.2444,
      "step": 450920
    },
    {
      "epoch": 0.14994754842027122,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010982518251825183,
      "loss": 3.2287,
      "step": 450930
    },
    {
      "epoch": 0.14996619165642938,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010982318231823184,
      "loss": 3.137,
      "step": 450940
    },
    {
      "epoch": 0.14998483489258757,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010982118211821182,
      "loss": 3.0852,
      "step": 450950
    },
    {
      "epoch": 0.15000347812874576,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010981918191819183,
      "loss": 3.0693,
      "step": 450960
    },
    {
      "epoch": 0.15002212136490395,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010981718171817181,
      "loss": 3.1045,
      "step": 450970
    },
    {
      "epoch": 0.15004076460106214,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010981518151815182,
      "loss": 3.134,
      "step": 450980
    },
    {
      "epoch": 0.1500594078372203,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010981318131813181,
      "loss": 3.0979,
      "step": 450990
    },
    {
      "epoch": 0.1500780510733785,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010981118111811181,
      "loss": 3.1271,
      "step": 451000
    },
    {
      "epoch": 0.15009669430953668,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010980918091809182,
      "loss": 3.0491,
      "step": 451010
    },
    {
      "epoch": 0.15011533754569487,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010980718071807182,
      "loss": 3.0853,
      "step": 451020
    },
    {
      "epoch": 0.15013398078185305,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010980518051805182,
      "loss": 3.1496,
      "step": 451030
    },
    {
      "epoch": 0.15015262401801124,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001098031803180318,
      "loss": 3.0547,
      "step": 451040
    },
    {
      "epoch": 0.1501712672541694,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010980118011801182,
      "loss": 3.162,
      "step": 451050
    },
    {
      "epoch": 0.1501899104903276,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001097991799179918,
      "loss": 3.2649,
      "step": 451060
    },
    {
      "epoch": 0.15020855372648578,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001097971797179718,
      "loss": 3.1077,
      "step": 451070
    },
    {
      "epoch": 0.15022719696264397,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001097951795179518,
      "loss": 3.1177,
      "step": 451080
    },
    {
      "epoch": 0.15024584019880216,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010979317931793181,
      "loss": 3.0892,
      "step": 451090
    },
    {
      "epoch": 0.15026448343496032,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001097911791179118,
      "loss": 2.9675,
      "step": 451100
    },
    {
      "epoch": 0.1502831266711185,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001097891789178918,
      "loss": 3.1574,
      "step": 451110
    },
    {
      "epoch": 0.1503017699072767,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010978717871787178,
      "loss": 3.0066,
      "step": 451120
    },
    {
      "epoch": 0.1503204131434349,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001097851785178518,
      "loss": 3.2068,
      "step": 451130
    },
    {
      "epoch": 0.15033905637959308,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010978317831783178,
      "loss": 3.2048,
      "step": 451140
    },
    {
      "epoch": 0.15035769961575124,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010978117811781178,
      "loss": 3.0408,
      "step": 451150
    },
    {
      "epoch": 0.15037634285190943,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010977917791779178,
      "loss": 3.1404,
      "step": 451160
    },
    {
      "epoch": 0.15039498608806762,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010977717771777179,
      "loss": 3.044,
      "step": 451170
    },
    {
      "epoch": 0.1504136293242258,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010977517751775177,
      "loss": 3.1395,
      "step": 451180
    },
    {
      "epoch": 0.150432272560384,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010977317731773178,
      "loss": 3.1654,
      "step": 451190
    },
    {
      "epoch": 0.15045091579654216,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010977117711771176,
      "loss": 3.2241,
      "step": 451200
    },
    {
      "epoch": 0.15046955903270035,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010976917691769177,
      "loss": 3.0956,
      "step": 451210
    },
    {
      "epoch": 0.15048820226885853,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010976717671767178,
      "loss": 3.161,
      "step": 451220
    },
    {
      "epoch": 0.15050684550501672,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010976517651765176,
      "loss": 3.1138,
      "step": 451230
    },
    {
      "epoch": 0.1505254887411749,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010976317631763177,
      "loss": 2.9437,
      "step": 451240
    },
    {
      "epoch": 0.1505441319773331,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010976117611761177,
      "loss": 3.0569,
      "step": 451250
    },
    {
      "epoch": 0.15056277521349126,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010975917591759178,
      "loss": 3.1215,
      "step": 451260
    },
    {
      "epoch": 0.15058141844964945,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010975717571757176,
      "loss": 3.0886,
      "step": 451270
    },
    {
      "epoch": 0.15060006168580764,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010975517551755177,
      "loss": 3.2122,
      "step": 451280
    },
    {
      "epoch": 0.15061870492196583,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010975317531753175,
      "loss": 3.1389,
      "step": 451290
    },
    {
      "epoch": 0.15063734815812402,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010975117511751176,
      "loss": 3.182,
      "step": 451300
    },
    {
      "epoch": 0.15065599139428218,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010974917491749176,
      "loss": 3.2172,
      "step": 451310
    },
    {
      "epoch": 0.15067463463044037,
      "grad_norm": 1.03125,
      "learning_rate": 0.00010974717471747175,
      "loss": 3.1353,
      "step": 451320
    },
    {
      "epoch": 0.15069327786659856,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010974517451745175,
      "loss": 3.1261,
      "step": 451330
    },
    {
      "epoch": 0.15071192110275675,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010974317431743176,
      "loss": 3.1403,
      "step": 451340
    },
    {
      "epoch": 0.15073056433891494,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010974117411741174,
      "loss": 3.1465,
      "step": 451350
    },
    {
      "epoch": 0.1507492075750731,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010973917391739175,
      "loss": 3.0304,
      "step": 451360
    },
    {
      "epoch": 0.15076785081123129,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010973717371737173,
      "loss": 3.0885,
      "step": 451370
    },
    {
      "epoch": 0.15078649404738947,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010973517351735174,
      "loss": 3.3509,
      "step": 451380
    },
    {
      "epoch": 0.15080513728354766,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010973317331733173,
      "loss": 3.1659,
      "step": 451390
    },
    {
      "epoch": 0.15082378051970585,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010973117311731174,
      "loss": 3.1805,
      "step": 451400
    },
    {
      "epoch": 0.150842423755864,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010972917291729175,
      "loss": 3.1091,
      "step": 451410
    },
    {
      "epoch": 0.1508610669920222,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010972717271727173,
      "loss": 3.158,
      "step": 451420
    },
    {
      "epoch": 0.1508797102281804,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010972517251725174,
      "loss": 2.9501,
      "step": 451430
    },
    {
      "epoch": 0.15089835346433858,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010972317231723172,
      "loss": 3.0744,
      "step": 451440
    },
    {
      "epoch": 0.15091699670049677,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010972117211721173,
      "loss": 3.0948,
      "step": 451450
    },
    {
      "epoch": 0.15093563993665493,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010971917191719172,
      "loss": 3.0099,
      "step": 451460
    },
    {
      "epoch": 0.15095428317281312,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010971717171717173,
      "loss": 3.124,
      "step": 451470
    },
    {
      "epoch": 0.1509729264089713,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010971517151715172,
      "loss": 3.043,
      "step": 451480
    },
    {
      "epoch": 0.1509915696451295,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010971317131713173,
      "loss": 3.113,
      "step": 451490
    },
    {
      "epoch": 0.1510102128812877,
      "grad_norm": 0.875,
      "learning_rate": 0.00010971117111711171,
      "loss": 3.0477,
      "step": 451500
    },
    {
      "epoch": 0.15102885611744588,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010970917091709172,
      "loss": 3.1818,
      "step": 451510
    },
    {
      "epoch": 0.15104749935360404,
      "grad_norm": 0.75,
      "learning_rate": 0.0001097071707170717,
      "loss": 3.1961,
      "step": 451520
    },
    {
      "epoch": 0.15106614258976223,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010970517051705171,
      "loss": 3.1457,
      "step": 451530
    },
    {
      "epoch": 0.15108478582592041,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001097031703170317,
      "loss": 3.1138,
      "step": 451540
    },
    {
      "epoch": 0.1511034290620786,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001097011701170117,
      "loss": 3.0207,
      "step": 451550
    },
    {
      "epoch": 0.1511220722982368,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001096991699169917,
      "loss": 3.1119,
      "step": 451560
    },
    {
      "epoch": 0.15114071553439495,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010969716971697171,
      "loss": 3.1345,
      "step": 451570
    },
    {
      "epoch": 0.15115935877055314,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010969516951695169,
      "loss": 3.1823,
      "step": 451580
    },
    {
      "epoch": 0.15117800200671133,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001096931693169317,
      "loss": 3.1665,
      "step": 451590
    },
    {
      "epoch": 0.15119664524286952,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010969116911691171,
      "loss": 3.0859,
      "step": 451600
    },
    {
      "epoch": 0.1512152884790277,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010968916891689169,
      "loss": 2.9599,
      "step": 451610
    },
    {
      "epoch": 0.15123393171518587,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001096871687168717,
      "loss": 3.2046,
      "step": 451620
    },
    {
      "epoch": 0.15125257495134406,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010968516851685168,
      "loss": 3.0493,
      "step": 451630
    },
    {
      "epoch": 0.15127121818750225,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010968316831683169,
      "loss": 3.0838,
      "step": 451640
    },
    {
      "epoch": 0.15128986142366044,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010968116811681169,
      "loss": 3.1283,
      "step": 451650
    },
    {
      "epoch": 0.15130850465981863,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001096791679167917,
      "loss": 3.1632,
      "step": 451660
    },
    {
      "epoch": 0.1513271478959768,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010967716771677168,
      "loss": 3.0683,
      "step": 451670
    },
    {
      "epoch": 0.15134579113213498,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010967516751675169,
      "loss": 3.1292,
      "step": 451680
    },
    {
      "epoch": 0.15136443436829317,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010967316731673167,
      "loss": 3.0408,
      "step": 451690
    },
    {
      "epoch": 0.15138307760445135,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010967116711671168,
      "loss": 3.039,
      "step": 451700
    },
    {
      "epoch": 0.15140172084060954,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010966916691669167,
      "loss": 3.095,
      "step": 451710
    },
    {
      "epoch": 0.15142036407676773,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010966716671667168,
      "loss": 3.0242,
      "step": 451720
    },
    {
      "epoch": 0.1514390073129259,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010966516651665166,
      "loss": 3.1287,
      "step": 451730
    },
    {
      "epoch": 0.15145765054908408,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010966316631663167,
      "loss": 3.0805,
      "step": 451740
    },
    {
      "epoch": 0.15147629378524227,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010966116611661166,
      "loss": 3.0792,
      "step": 451750
    },
    {
      "epoch": 0.15149493702140046,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010965916591659167,
      "loss": 3.0501,
      "step": 451760
    },
    {
      "epoch": 0.15151358025755865,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010965716571657165,
      "loss": 2.9688,
      "step": 451770
    },
    {
      "epoch": 0.1515322234937168,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010965516551655166,
      "loss": 3.1841,
      "step": 451780
    },
    {
      "epoch": 0.151550866729875,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010965316531653165,
      "loss": 3.0184,
      "step": 451790
    },
    {
      "epoch": 0.1515695099660332,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010965116511651166,
      "loss": 3.1496,
      "step": 451800
    },
    {
      "epoch": 0.15158815320219138,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010964916491649167,
      "loss": 3.1489,
      "step": 451810
    },
    {
      "epoch": 0.15160679643834957,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010964716471647165,
      "loss": 3.068,
      "step": 451820
    },
    {
      "epoch": 0.15162543967450773,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010964516451645166,
      "loss": 3.1551,
      "step": 451830
    },
    {
      "epoch": 0.15164408291066592,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010964316431643164,
      "loss": 3.0744,
      "step": 451840
    },
    {
      "epoch": 0.1516627261468241,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010964116411641165,
      "loss": 3.1324,
      "step": 451850
    },
    {
      "epoch": 0.1516813693829823,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010963916391639163,
      "loss": 3.1221,
      "step": 451860
    },
    {
      "epoch": 0.15170001261914048,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010963716371637164,
      "loss": 2.9654,
      "step": 451870
    },
    {
      "epoch": 0.15171865585529865,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010963516351635164,
      "loss": 3.1262,
      "step": 451880
    },
    {
      "epoch": 0.15173729909145683,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010963316331633165,
      "loss": 2.9734,
      "step": 451890
    },
    {
      "epoch": 0.15175594232761502,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010963116311631163,
      "loss": 3.1675,
      "step": 451900
    },
    {
      "epoch": 0.1517745855637732,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010962916291629164,
      "loss": 3.0378,
      "step": 451910
    },
    {
      "epoch": 0.1517932287999314,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010962716271627162,
      "loss": 3.1276,
      "step": 451920
    },
    {
      "epoch": 0.1518118720360896,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010962516251625163,
      "loss": 3.1883,
      "step": 451930
    },
    {
      "epoch": 0.15183051527224775,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010962316231623163,
      "loss": 3.1331,
      "step": 451940
    },
    {
      "epoch": 0.15184915850840594,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010962116211621162,
      "loss": 3.1674,
      "step": 451950
    },
    {
      "epoch": 0.15186780174456413,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010961916191619162,
      "loss": 3.0527,
      "step": 451960
    },
    {
      "epoch": 0.15188644498072232,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010961716171617163,
      "loss": 3.0195,
      "step": 451970
    },
    {
      "epoch": 0.1519050882168805,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010961516151615161,
      "loss": 3.1308,
      "step": 451980
    },
    {
      "epoch": 0.15192373145303867,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010961316131613162,
      "loss": 3.2141,
      "step": 451990
    },
    {
      "epoch": 0.15194237468919686,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010961116111611163,
      "loss": 3.0906,
      "step": 452000
    },
    {
      "epoch": 0.15196101792535505,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010960916091609161,
      "loss": 3.0581,
      "step": 452010
    },
    {
      "epoch": 0.15197966116151324,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010960716071607162,
      "loss": 3.1094,
      "step": 452020
    },
    {
      "epoch": 0.15199830439767142,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010960516051605161,
      "loss": 3.0436,
      "step": 452030
    },
    {
      "epoch": 0.15201694763382959,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010960316031603162,
      "loss": 3.0803,
      "step": 452040
    },
    {
      "epoch": 0.15203559086998777,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001096011601160116,
      "loss": 3.1075,
      "step": 452050
    },
    {
      "epoch": 0.15205423410614596,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010959915991599161,
      "loss": 3.0133,
      "step": 452060
    },
    {
      "epoch": 0.15207287734230415,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001095971597159716,
      "loss": 3.0597,
      "step": 452070
    },
    {
      "epoch": 0.15209152057846234,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001095951595159516,
      "loss": 3.1699,
      "step": 452080
    },
    {
      "epoch": 0.1521101638146205,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00010959315931593159,
      "loss": 2.9958,
      "step": 452090
    },
    {
      "epoch": 0.1521288070507787,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001095911591159116,
      "loss": 3.2267,
      "step": 452100
    },
    {
      "epoch": 0.15214745028693688,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010958915891589159,
      "loss": 3.1141,
      "step": 452110
    },
    {
      "epoch": 0.15216609352309507,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001095871587158716,
      "loss": 3.1367,
      "step": 452120
    },
    {
      "epoch": 0.15218473675925326,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010958515851585158,
      "loss": 3.1095,
      "step": 452130
    },
    {
      "epoch": 0.15220337999541142,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010958315831583159,
      "loss": 3.0808,
      "step": 452140
    },
    {
      "epoch": 0.1522220232315696,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010958115811581157,
      "loss": 3.1163,
      "step": 452150
    },
    {
      "epoch": 0.1522406664677278,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010957915791579158,
      "loss": 3.2319,
      "step": 452160
    },
    {
      "epoch": 0.152259309703886,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010957715771577157,
      "loss": 2.9532,
      "step": 452170
    },
    {
      "epoch": 0.15227795294004418,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010957515751575157,
      "loss": 3.2135,
      "step": 452180
    },
    {
      "epoch": 0.15229659617620236,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010957315731573158,
      "loss": 3.0611,
      "step": 452190
    },
    {
      "epoch": 0.15231523941236053,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010957115711571158,
      "loss": 3.1151,
      "step": 452200
    },
    {
      "epoch": 0.15233388264851871,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010956915691569159,
      "loss": 3.1833,
      "step": 452210
    },
    {
      "epoch": 0.1523525258846769,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010956715671567157,
      "loss": 2.9985,
      "step": 452220
    },
    {
      "epoch": 0.1523711691208351,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010956515651565158,
      "loss": 3.1768,
      "step": 452230
    },
    {
      "epoch": 0.15238981235699328,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010956315631563156,
      "loss": 3.1692,
      "step": 452240
    },
    {
      "epoch": 0.15240845559315144,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010956115611561157,
      "loss": 3.166,
      "step": 452250
    },
    {
      "epoch": 0.15242709882930963,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010955915591559157,
      "loss": 3.1066,
      "step": 452260
    },
    {
      "epoch": 0.15244574206546782,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010955715571557156,
      "loss": 3.0506,
      "step": 452270
    },
    {
      "epoch": 0.152464385301626,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010955515551555156,
      "loss": 3.0879,
      "step": 452280
    },
    {
      "epoch": 0.1524830285377842,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010955315531553157,
      "loss": 3.088,
      "step": 452290
    },
    {
      "epoch": 0.15250167177394236,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010955115511551155,
      "loss": 2.9347,
      "step": 452300
    },
    {
      "epoch": 0.15252031501010055,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010954915491549156,
      "loss": 3.1592,
      "step": 452310
    },
    {
      "epoch": 0.15253895824625874,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010954715471547154,
      "loss": 3.0668,
      "step": 452320
    },
    {
      "epoch": 0.15255760148241693,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010954515451545155,
      "loss": 3.17,
      "step": 452330
    },
    {
      "epoch": 0.15257624471857512,
      "grad_norm": 0.75,
      "learning_rate": 0.00010954315431543155,
      "loss": 3.067,
      "step": 452340
    },
    {
      "epoch": 0.15259488795473328,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010954115411541155,
      "loss": 3.2271,
      "step": 452350
    },
    {
      "epoch": 0.15261353119089147,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010953915391539154,
      "loss": 3.1138,
      "step": 452360
    },
    {
      "epoch": 0.15263217442704965,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010953715371537155,
      "loss": 3.1958,
      "step": 452370
    },
    {
      "epoch": 0.15265081766320784,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010953515351535153,
      "loss": 3.1195,
      "step": 452380
    },
    {
      "epoch": 0.15266946089936603,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010953315331533154,
      "loss": 3.1092,
      "step": 452390
    },
    {
      "epoch": 0.15268810413552422,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010953115311531155,
      "loss": 3.1285,
      "step": 452400
    },
    {
      "epoch": 0.15270674737168238,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010952915291529153,
      "loss": 3.2683,
      "step": 452410
    },
    {
      "epoch": 0.15272539060784057,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010952715271527154,
      "loss": 3.0537,
      "step": 452420
    },
    {
      "epoch": 0.15274403384399876,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010952515251525153,
      "loss": 3.1739,
      "step": 452430
    },
    {
      "epoch": 0.15276267708015695,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010952315231523154,
      "loss": 3.1814,
      "step": 452440
    },
    {
      "epoch": 0.15278132031631514,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010952115211521152,
      "loss": 3.1213,
      "step": 452450
    },
    {
      "epoch": 0.1527999635524733,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010951915191519153,
      "loss": 3.1892,
      "step": 452460
    },
    {
      "epoch": 0.1528186067886315,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010951715171517151,
      "loss": 3.119,
      "step": 452470
    },
    {
      "epoch": 0.15283725002478968,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010951515151515152,
      "loss": 3.0661,
      "step": 452480
    },
    {
      "epoch": 0.15285589326094787,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001095131513151315,
      "loss": 3.0355,
      "step": 452490
    },
    {
      "epoch": 0.15287453649710606,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010951115111511152,
      "loss": 3.0543,
      "step": 452500
    },
    {
      "epoch": 0.15289317973326422,
      "grad_norm": 0.75,
      "learning_rate": 0.00010950915091509151,
      "loss": 3.2204,
      "step": 452510
    },
    {
      "epoch": 0.1529118229694224,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010950715071507152,
      "loss": 2.9439,
      "step": 452520
    },
    {
      "epoch": 0.1529304662055806,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001095051505150515,
      "loss": 3.0925,
      "step": 452530
    },
    {
      "epoch": 0.15294910944173878,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010950315031503151,
      "loss": 3.2451,
      "step": 452540
    },
    {
      "epoch": 0.15296775267789697,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001095011501150115,
      "loss": 3.0912,
      "step": 452550
    },
    {
      "epoch": 0.15298639591405513,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001094991499149915,
      "loss": 3.0537,
      "step": 452560
    },
    {
      "epoch": 0.15300503915021332,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001094971497149715,
      "loss": 3.0859,
      "step": 452570
    },
    {
      "epoch": 0.1530236823863715,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010949514951495151,
      "loss": 3.027,
      "step": 452580
    },
    {
      "epoch": 0.1530423256225297,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001094931493149315,
      "loss": 3.2008,
      "step": 452590
    },
    {
      "epoch": 0.1530609688586879,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001094911491149115,
      "loss": 3.1449,
      "step": 452600
    },
    {
      "epoch": 0.15307961209484608,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010948914891489151,
      "loss": 3.1856,
      "step": 452610
    },
    {
      "epoch": 0.15309825533100424,
      "grad_norm": 0.75,
      "learning_rate": 0.00010948714871487149,
      "loss": 3.1193,
      "step": 452620
    },
    {
      "epoch": 0.15311689856716243,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001094851485148515,
      "loss": 3.2731,
      "step": 452630
    },
    {
      "epoch": 0.15313554180332062,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010948314831483148,
      "loss": 3.2362,
      "step": 452640
    },
    {
      "epoch": 0.1531541850394788,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010948114811481149,
      "loss": 3.0545,
      "step": 452650
    },
    {
      "epoch": 0.153172828275637,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010947914791479149,
      "loss": 3.08,
      "step": 452660
    },
    {
      "epoch": 0.15319147151179516,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001094771477147715,
      "loss": 3.0224,
      "step": 452670
    },
    {
      "epoch": 0.15321011474795335,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010947514751475148,
      "loss": 3.1734,
      "step": 452680
    },
    {
      "epoch": 0.15322875798411154,
      "grad_norm": 1.9765625,
      "learning_rate": 0.00010947314731473149,
      "loss": 3.0555,
      "step": 452690
    },
    {
      "epoch": 0.15324740122026972,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010947114711471147,
      "loss": 3.076,
      "step": 452700
    },
    {
      "epoch": 0.1532660444564279,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010946914691469148,
      "loss": 3.0158,
      "step": 452710
    },
    {
      "epoch": 0.15328468769258607,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010946714671467146,
      "loss": 3.087,
      "step": 452720
    },
    {
      "epoch": 0.15330333092874426,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010946514651465147,
      "loss": 3.1156,
      "step": 452730
    },
    {
      "epoch": 0.15332197416490245,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010946314631463146,
      "loss": 3.1119,
      "step": 452740
    },
    {
      "epoch": 0.15334061740106064,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010946114611461147,
      "loss": 3.0697,
      "step": 452750
    },
    {
      "epoch": 0.15335926063721883,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010945914591459145,
      "loss": 3.2425,
      "step": 452760
    },
    {
      "epoch": 0.153377903873377,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010945714571457146,
      "loss": 3.1923,
      "step": 452770
    },
    {
      "epoch": 0.15339654710953518,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010945514551455147,
      "loss": 3.2053,
      "step": 452780
    },
    {
      "epoch": 0.15341519034569337,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010945314531453146,
      "loss": 3.1166,
      "step": 452790
    },
    {
      "epoch": 0.15343383358185156,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010945114511451146,
      "loss": 3.0739,
      "step": 452800
    },
    {
      "epoch": 0.15345247681800975,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010944914491449145,
      "loss": 3.1506,
      "step": 452810
    },
    {
      "epoch": 0.1534711200541679,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010944714471447146,
      "loss": 3.1693,
      "step": 452820
    },
    {
      "epoch": 0.1534897632903261,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010944514451445145,
      "loss": 3.2368,
      "step": 452830
    },
    {
      "epoch": 0.1535084065264843,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010944314431443146,
      "loss": 3.1977,
      "step": 452840
    },
    {
      "epoch": 0.15352704976264248,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010944114411441144,
      "loss": 3.2061,
      "step": 452850
    },
    {
      "epoch": 0.15354569299880066,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010943914391439145,
      "loss": 3.1386,
      "step": 452860
    },
    {
      "epoch": 0.15356433623495885,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010943714371437143,
      "loss": 3.0678,
      "step": 452870
    },
    {
      "epoch": 0.15358297947111701,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010943514351435144,
      "loss": 3.1561,
      "step": 452880
    },
    {
      "epoch": 0.1536016227072752,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010943314331433144,
      "loss": 3.0236,
      "step": 452890
    },
    {
      "epoch": 0.1536202659434334,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010943114311431145,
      "loss": 3.1433,
      "step": 452900
    },
    {
      "epoch": 0.15363890917959158,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010942914291429143,
      "loss": 3.1734,
      "step": 452910
    },
    {
      "epoch": 0.15365755241574977,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010942714271427144,
      "loss": 3.1231,
      "step": 452920
    },
    {
      "epoch": 0.15367619565190793,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010942514251425142,
      "loss": 3.1454,
      "step": 452930
    },
    {
      "epoch": 0.15369483888806612,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010942314231423143,
      "loss": 3.1661,
      "step": 452940
    },
    {
      "epoch": 0.1537134821242243,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010942114211421141,
      "loss": 3.1778,
      "step": 452950
    },
    {
      "epoch": 0.1537321253603825,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010941914191419142,
      "loss": 2.9743,
      "step": 452960
    },
    {
      "epoch": 0.1537507685965407,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010941714171417142,
      "loss": 3.1056,
      "step": 452970
    },
    {
      "epoch": 0.15376941183269885,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010941514151415143,
      "loss": 2.9671,
      "step": 452980
    },
    {
      "epoch": 0.15378805506885704,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010941314131413143,
      "loss": 3.1501,
      "step": 452990
    },
    {
      "epoch": 0.15380669830501523,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010941114111411142,
      "loss": 3.2292,
      "step": 453000
    },
    {
      "epoch": 0.15382534154117342,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010940914091409143,
      "loss": 3.0808,
      "step": 453010
    },
    {
      "epoch": 0.1538439847773316,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010940714071407141,
      "loss": 3.16,
      "step": 453020
    },
    {
      "epoch": 0.15386262801348977,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00010940514051405142,
      "loss": 3.1014,
      "step": 453030
    },
    {
      "epoch": 0.15388127124964796,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001094031403140314,
      "loss": 3.1285,
      "step": 453040
    },
    {
      "epoch": 0.15389991448580614,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010940114011401141,
      "loss": 3.0537,
      "step": 453050
    },
    {
      "epoch": 0.15391855772196433,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001093991399139914,
      "loss": 3.201,
      "step": 453060
    },
    {
      "epoch": 0.15393720095812252,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010939713971397141,
      "loss": 3.1824,
      "step": 453070
    },
    {
      "epoch": 0.1539558441942807,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001093951395139514,
      "loss": 3.2135,
      "step": 453080
    },
    {
      "epoch": 0.15397448743043887,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001093931393139314,
      "loss": 2.9829,
      "step": 453090
    },
    {
      "epoch": 0.15399313066659706,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010939113911391139,
      "loss": 3.1182,
      "step": 453100
    },
    {
      "epoch": 0.15401177390275525,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001093891389138914,
      "loss": 3.1381,
      "step": 453110
    },
    {
      "epoch": 0.15403041713891344,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010938713871387138,
      "loss": 3.1456,
      "step": 453120
    },
    {
      "epoch": 0.15404906037507163,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010938513851385139,
      "loss": 3.2031,
      "step": 453130
    },
    {
      "epoch": 0.1540677036112298,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010938313831383138,
      "loss": 3.133,
      "step": 453140
    },
    {
      "epoch": 0.15408634684738798,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010938113811381139,
      "loss": 3.0275,
      "step": 453150
    },
    {
      "epoch": 0.15410499008354617,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010937913791379137,
      "loss": 2.953,
      "step": 453160
    },
    {
      "epoch": 0.15412363331970436,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010937713771377138,
      "loss": 3.1456,
      "step": 453170
    },
    {
      "epoch": 0.15414227655586255,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010937513751375139,
      "loss": 3.2092,
      "step": 453180
    },
    {
      "epoch": 0.1541609197920207,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010937313731373137,
      "loss": 3.0767,
      "step": 453190
    },
    {
      "epoch": 0.1541795630281789,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010937113711371138,
      "loss": 3.0168,
      "step": 453200
    },
    {
      "epoch": 0.15419820626433708,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010936913691369138,
      "loss": 3.0467,
      "step": 453210
    },
    {
      "epoch": 0.15421684950049527,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010936713671367137,
      "loss": 3.1021,
      "step": 453220
    },
    {
      "epoch": 0.15423549273665346,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010936513651365137,
      "loss": 3.0518,
      "step": 453230
    },
    {
      "epoch": 0.15425413597281162,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010936313631363138,
      "loss": 2.9462,
      "step": 453240
    },
    {
      "epoch": 0.1542727792089698,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010936113611361136,
      "loss": 3.1494,
      "step": 453250
    },
    {
      "epoch": 0.154291422445128,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010935913591359137,
      "loss": 3.1449,
      "step": 453260
    },
    {
      "epoch": 0.1543100656812862,
      "grad_norm": 0.75,
      "learning_rate": 0.00010935713571357135,
      "loss": 3.0919,
      "step": 453270
    },
    {
      "epoch": 0.15432870891744438,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010935513551355136,
      "loss": 3.1869,
      "step": 453280
    },
    {
      "epoch": 0.15434735215360257,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010935313531353136,
      "loss": 3.0734,
      "step": 453290
    },
    {
      "epoch": 0.15436599538976073,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010935113511351137,
      "loss": 3.0411,
      "step": 453300
    },
    {
      "epoch": 0.15438463862591892,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010934913491349135,
      "loss": 3.1441,
      "step": 453310
    },
    {
      "epoch": 0.1544032818620771,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010934713471347136,
      "loss": 3.1416,
      "step": 453320
    },
    {
      "epoch": 0.1544219250982353,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010934513451345134,
      "loss": 3.0076,
      "step": 453330
    },
    {
      "epoch": 0.15444056833439349,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010934313431343135,
      "loss": 3.0339,
      "step": 453340
    },
    {
      "epoch": 0.15445921157055165,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010934113411341133,
      "loss": 3.0418,
      "step": 453350
    },
    {
      "epoch": 0.15447785480670984,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010933913391339134,
      "loss": 3.1025,
      "step": 453360
    },
    {
      "epoch": 0.15449649804286802,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010933713371337134,
      "loss": 3.0993,
      "step": 453370
    },
    {
      "epoch": 0.1545151412790262,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010933513351335134,
      "loss": 3.2312,
      "step": 453380
    },
    {
      "epoch": 0.1545337845151844,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010933313331333135,
      "loss": 3.0268,
      "step": 453390
    },
    {
      "epoch": 0.15455242775134256,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010933113311331134,
      "loss": 3.1628,
      "step": 453400
    },
    {
      "epoch": 0.15457107098750075,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010932913291329134,
      "loss": 3.1838,
      "step": 453410
    },
    {
      "epoch": 0.15458971422365894,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010932713271327133,
      "loss": 3.1516,
      "step": 453420
    },
    {
      "epoch": 0.15460835745981713,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010932513251325134,
      "loss": 3.0943,
      "step": 453430
    },
    {
      "epoch": 0.15462700069597532,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010932313231323132,
      "loss": 2.9591,
      "step": 453440
    },
    {
      "epoch": 0.15464564393213348,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00010932113211321133,
      "loss": 3.0762,
      "step": 453450
    },
    {
      "epoch": 0.15466428716829167,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010931913191319132,
      "loss": 3.0278,
      "step": 453460
    },
    {
      "epoch": 0.15468293040444986,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010931713171317133,
      "loss": 3.0876,
      "step": 453470
    },
    {
      "epoch": 0.15470157364060805,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010931513151315131,
      "loss": 3.0773,
      "step": 453480
    },
    {
      "epoch": 0.15472021687676624,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010931313131313132,
      "loss": 2.8961,
      "step": 453490
    },
    {
      "epoch": 0.15473886011292443,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001093111311131113,
      "loss": 3.1248,
      "step": 453500
    },
    {
      "epoch": 0.1547575033490826,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010930913091309131,
      "loss": 3.0191,
      "step": 453510
    },
    {
      "epoch": 0.15477614658524078,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010930713071307131,
      "loss": 3.0337,
      "step": 453520
    },
    {
      "epoch": 0.15479478982139896,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010930513051305132,
      "loss": 2.9757,
      "step": 453530
    },
    {
      "epoch": 0.15481343305755715,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001093031303130313,
      "loss": 3.0297,
      "step": 453540
    },
    {
      "epoch": 0.15483207629371534,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010930113011301131,
      "loss": 3.0659,
      "step": 453550
    },
    {
      "epoch": 0.1548507195298735,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010929912991299129,
      "loss": 3.0369,
      "step": 453560
    },
    {
      "epoch": 0.1548693627660317,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001092971297129713,
      "loss": 3.1915,
      "step": 453570
    },
    {
      "epoch": 0.15488800600218988,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010929512951295131,
      "loss": 3.0667,
      "step": 453580
    },
    {
      "epoch": 0.15490664923834807,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010929312931293129,
      "loss": 3.1104,
      "step": 453590
    },
    {
      "epoch": 0.15492529247450626,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001092911291129113,
      "loss": 3.0679,
      "step": 453600
    },
    {
      "epoch": 0.15494393571066442,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001092891289128913,
      "loss": 3.0854,
      "step": 453610
    },
    {
      "epoch": 0.1549625789468226,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001092871287128713,
      "loss": 2.9879,
      "step": 453620
    },
    {
      "epoch": 0.1549812221829808,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010928512851285129,
      "loss": 3.2503,
      "step": 453630
    },
    {
      "epoch": 0.154999865419139,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001092831283128313,
      "loss": 3.1643,
      "step": 453640
    },
    {
      "epoch": 0.15501850865529718,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010928112811281128,
      "loss": 3.0729,
      "step": 453650
    },
    {
      "epoch": 0.15503715189145534,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010927912791279129,
      "loss": 3.0174,
      "step": 453660
    },
    {
      "epoch": 0.15505579512761353,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010927712771277127,
      "loss": 3.1383,
      "step": 453670
    },
    {
      "epoch": 0.15507443836377172,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010927512751275128,
      "loss": 3.1595,
      "step": 453680
    },
    {
      "epoch": 0.1550930815999299,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010927312731273128,
      "loss": 3.1001,
      "step": 453690
    },
    {
      "epoch": 0.1551117248360881,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010927112711271128,
      "loss": 3.1751,
      "step": 453700
    },
    {
      "epoch": 0.15513036807224626,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010926912691269127,
      "loss": 3.0222,
      "step": 453710
    },
    {
      "epoch": 0.15514901130840444,
      "grad_norm": 0.75,
      "learning_rate": 0.00010926712671267128,
      "loss": 3.0682,
      "step": 453720
    },
    {
      "epoch": 0.15516765454456263,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010926512651265126,
      "loss": 3.0778,
      "step": 453730
    },
    {
      "epoch": 0.15518629778072082,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010926312631263127,
      "loss": 2.9929,
      "step": 453740
    },
    {
      "epoch": 0.155204941016879,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010926112611261126,
      "loss": 3.0886,
      "step": 453750
    },
    {
      "epoch": 0.1552235842530372,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010925912591259126,
      "loss": 3.1663,
      "step": 453760
    },
    {
      "epoch": 0.15524222748919536,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010925712571257127,
      "loss": 3.0499,
      "step": 453770
    },
    {
      "epoch": 0.15526087072535355,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010925512551255126,
      "loss": 3.1441,
      "step": 453780
    },
    {
      "epoch": 0.15527951396151174,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010925312531253127,
      "loss": 3.1351,
      "step": 453790
    },
    {
      "epoch": 0.15529815719766993,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010925112511251125,
      "loss": 3.0383,
      "step": 453800
    },
    {
      "epoch": 0.15531680043382812,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010924912491249126,
      "loss": 3.0603,
      "step": 453810
    },
    {
      "epoch": 0.15533544366998628,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010924712471247125,
      "loss": 3.1742,
      "step": 453820
    },
    {
      "epoch": 0.15535408690614447,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010924512451245125,
      "loss": 3.08,
      "step": 453830
    },
    {
      "epoch": 0.15537273014230266,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010924312431243125,
      "loss": 3.2585,
      "step": 453840
    },
    {
      "epoch": 0.15539137337846085,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010924112411241126,
      "loss": 3.1794,
      "step": 453850
    },
    {
      "epoch": 0.15541001661461903,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010923912391239124,
      "loss": 3.1884,
      "step": 453860
    },
    {
      "epoch": 0.1554286598507772,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010923712371237125,
      "loss": 3.0465,
      "step": 453870
    },
    {
      "epoch": 0.15544730308693538,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010923512351235123,
      "loss": 3.0928,
      "step": 453880
    },
    {
      "epoch": 0.15546594632309357,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010923312331233124,
      "loss": 3.1169,
      "step": 453890
    },
    {
      "epoch": 0.15548458955925176,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010923112311231122,
      "loss": 3.0989,
      "step": 453900
    },
    {
      "epoch": 0.15550323279540995,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010922912291229123,
      "loss": 3.0969,
      "step": 453910
    },
    {
      "epoch": 0.1555218760315681,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010922712271227123,
      "loss": 2.9704,
      "step": 453920
    },
    {
      "epoch": 0.1555405192677263,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010922512251225124,
      "loss": 3.0356,
      "step": 453930
    },
    {
      "epoch": 0.1555591625038845,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010922312231223122,
      "loss": 3.2074,
      "step": 453940
    },
    {
      "epoch": 0.15557780574004268,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010922112211221123,
      "loss": 3.1386,
      "step": 453950
    },
    {
      "epoch": 0.15559644897620087,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010921912191219121,
      "loss": 3.1,
      "step": 453960
    },
    {
      "epoch": 0.15561509221235906,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010921712171217122,
      "loss": 3.1411,
      "step": 453970
    },
    {
      "epoch": 0.15563373544851722,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010921512151215123,
      "loss": 3.179,
      "step": 453980
    },
    {
      "epoch": 0.1556523786846754,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010921312131213121,
      "loss": 3.329,
      "step": 453990
    },
    {
      "epoch": 0.1556710219208336,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010921112111211122,
      "loss": 3.0567,
      "step": 454000
    },
    {
      "epoch": 0.15568966515699179,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010920912091209122,
      "loss": 3.0169,
      "step": 454010
    },
    {
      "epoch": 0.15570830839314997,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010920712071207123,
      "loss": 3.2043,
      "step": 454020
    },
    {
      "epoch": 0.15572695162930814,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010920512051205121,
      "loss": 3.1811,
      "step": 454030
    },
    {
      "epoch": 0.15574559486546632,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010920312031203122,
      "loss": 3.0939,
      "step": 454040
    },
    {
      "epoch": 0.1557642381016245,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001092011201120112,
      "loss": 3.2167,
      "step": 454050
    },
    {
      "epoch": 0.1557828813377827,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010919911991199121,
      "loss": 3.2986,
      "step": 454060
    },
    {
      "epoch": 0.1558015245739409,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001091971197119712,
      "loss": 3.0511,
      "step": 454070
    },
    {
      "epoch": 0.15582016781009905,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001091951195119512,
      "loss": 3.1648,
      "step": 454080
    },
    {
      "epoch": 0.15583881104625724,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001091931193119312,
      "loss": 3.0584,
      "step": 454090
    },
    {
      "epoch": 0.15585745428241543,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001091911191119112,
      "loss": 3.125,
      "step": 454100
    },
    {
      "epoch": 0.15587609751857362,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010918911891189119,
      "loss": 3.0274,
      "step": 454110
    },
    {
      "epoch": 0.1558947407547318,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001091871187118712,
      "loss": 3.0268,
      "step": 454120
    },
    {
      "epoch": 0.15591338399088997,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010918511851185118,
      "loss": 3.1239,
      "step": 454130
    },
    {
      "epoch": 0.15593202722704816,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010918311831183119,
      "loss": 3.0573,
      "step": 454140
    },
    {
      "epoch": 0.15595067046320635,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010918111811181118,
      "loss": 3.0325,
      "step": 454150
    },
    {
      "epoch": 0.15596931369936454,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010917911791179119,
      "loss": 3.1191,
      "step": 454160
    },
    {
      "epoch": 0.15598795693552273,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001091771177117712,
      "loss": 3.1218,
      "step": 454170
    },
    {
      "epoch": 0.15600660017168091,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010917511751175118,
      "loss": 3.1063,
      "step": 454180
    },
    {
      "epoch": 0.15602524340783908,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010917311731173119,
      "loss": 3.0785,
      "step": 454190
    },
    {
      "epoch": 0.15604388664399726,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00010917111711171117,
      "loss": 3.0934,
      "step": 454200
    },
    {
      "epoch": 0.15606252988015545,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010916911691169118,
      "loss": 2.9263,
      "step": 454210
    },
    {
      "epoch": 0.15608117311631364,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010916711671167116,
      "loss": 3.088,
      "step": 454220
    },
    {
      "epoch": 0.15609981635247183,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010916511651165117,
      "loss": 3.0525,
      "step": 454230
    },
    {
      "epoch": 0.15611845958863,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010916311631163117,
      "loss": 3.069,
      "step": 454240
    },
    {
      "epoch": 0.15613710282478818,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010916111611161118,
      "loss": 3.1219,
      "step": 454250
    },
    {
      "epoch": 0.15615574606094637,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010915911591159116,
      "loss": 3.0549,
      "step": 454260
    },
    {
      "epoch": 0.15617438929710456,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010915711571157117,
      "loss": 3.2297,
      "step": 454270
    },
    {
      "epoch": 0.15619303253326275,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010915511551155115,
      "loss": 2.9935,
      "step": 454280
    },
    {
      "epoch": 0.1562116757694209,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010915311531153116,
      "loss": 3.1265,
      "step": 454290
    },
    {
      "epoch": 0.1562303190055791,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010915111511151114,
      "loss": 3.1955,
      "step": 454300
    },
    {
      "epoch": 0.1562489622417373,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010914911491149115,
      "loss": 3.103,
      "step": 454310
    },
    {
      "epoch": 0.15626760547789548,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010914711471147115,
      "loss": 3.2139,
      "step": 454320
    },
    {
      "epoch": 0.15628624871405367,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010914511451145116,
      "loss": 3.0935,
      "step": 454330
    },
    {
      "epoch": 0.15630489195021183,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010914311431143114,
      "loss": 3.1682,
      "step": 454340
    },
    {
      "epoch": 0.15632353518637002,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010914111411141115,
      "loss": 3.0156,
      "step": 454350
    },
    {
      "epoch": 0.1563421784225282,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010913911391139116,
      "loss": 3.1057,
      "step": 454360
    },
    {
      "epoch": 0.1563608216586864,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010913711371137114,
      "loss": 3.1482,
      "step": 454370
    },
    {
      "epoch": 0.15637946489484458,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010913511351135115,
      "loss": 3.1158,
      "step": 454380
    },
    {
      "epoch": 0.15639810813100274,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010913311331133113,
      "loss": 3.0636,
      "step": 454390
    },
    {
      "epoch": 0.15641675136716093,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010913111311131114,
      "loss": 3.063,
      "step": 454400
    },
    {
      "epoch": 0.15643539460331912,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010912911291129113,
      "loss": 3.0696,
      "step": 454410
    },
    {
      "epoch": 0.1564540378394773,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010912711271127114,
      "loss": 3.0938,
      "step": 454420
    },
    {
      "epoch": 0.1564726810756355,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010912511251125113,
      "loss": 2.999,
      "step": 454430
    },
    {
      "epoch": 0.1564913243117937,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010912311231123113,
      "loss": 3.0557,
      "step": 454440
    },
    {
      "epoch": 0.15650996754795185,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010912111211121112,
      "loss": 3.0587,
      "step": 454450
    },
    {
      "epoch": 0.15652861078411004,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010911911191119113,
      "loss": 3.1388,
      "step": 454460
    },
    {
      "epoch": 0.15654725402026823,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010911711171117112,
      "loss": 3.205,
      "step": 454470
    },
    {
      "epoch": 0.15656589725642642,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010911511151115113,
      "loss": 3.0768,
      "step": 454480
    },
    {
      "epoch": 0.1565845404925846,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010911311131113111,
      "loss": 3.0558,
      "step": 454490
    },
    {
      "epoch": 0.15660318372874277,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010911111111111112,
      "loss": 3.046,
      "step": 454500
    },
    {
      "epoch": 0.15662182696490096,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001091091109110911,
      "loss": 3.074,
      "step": 454510
    },
    {
      "epoch": 0.15664047020105915,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010910711071107111,
      "loss": 3.1209,
      "step": 454520
    },
    {
      "epoch": 0.15665911343721733,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001091051105110511,
      "loss": 3.1839,
      "step": 454530
    },
    {
      "epoch": 0.15667775667337552,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001091031103110311,
      "loss": 3.1222,
      "step": 454540
    },
    {
      "epoch": 0.15669639990953368,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001091011101110111,
      "loss": 3.2402,
      "step": 454550
    },
    {
      "epoch": 0.15671504314569187,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010909910991099111,
      "loss": 3.0623,
      "step": 454560
    },
    {
      "epoch": 0.15673368638185006,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010909710971097112,
      "loss": 3.1008,
      "step": 454570
    },
    {
      "epoch": 0.15675232961800825,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001090951095109511,
      "loss": 3.1433,
      "step": 454580
    },
    {
      "epoch": 0.15677097285416644,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010909310931093111,
      "loss": 3.0273,
      "step": 454590
    },
    {
      "epoch": 0.1567896160903246,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010909110911091109,
      "loss": 3.1093,
      "step": 454600
    },
    {
      "epoch": 0.1568082593264828,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001090891089108911,
      "loss": 3.1179,
      "step": 454610
    },
    {
      "epoch": 0.15682690256264098,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010908710871087108,
      "loss": 3.0177,
      "step": 454620
    },
    {
      "epoch": 0.15684554579879917,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010908510851085109,
      "loss": 3.0927,
      "step": 454630
    },
    {
      "epoch": 0.15686418903495736,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010908310831083109,
      "loss": 3.1268,
      "step": 454640
    },
    {
      "epoch": 0.15688283227111555,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001090811081108111,
      "loss": 3.2347,
      "step": 454650
    },
    {
      "epoch": 0.1569014755072737,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010907910791079108,
      "loss": 3.0655,
      "step": 454660
    },
    {
      "epoch": 0.1569201187434319,
      "grad_norm": 0.75,
      "learning_rate": 0.00010907710771077109,
      "loss": 3.0108,
      "step": 454670
    },
    {
      "epoch": 0.15693876197959009,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010907510751075107,
      "loss": 3.2438,
      "step": 454680
    },
    {
      "epoch": 0.15695740521574827,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010907310731073108,
      "loss": 3.0553,
      "step": 454690
    },
    {
      "epoch": 0.15697604845190646,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010907110711071107,
      "loss": 3.1739,
      "step": 454700
    },
    {
      "epoch": 0.15699469168806462,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010906910691069107,
      "loss": 3.1781,
      "step": 454710
    },
    {
      "epoch": 0.1570133349242228,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010906710671067107,
      "loss": 3.0988,
      "step": 454720
    },
    {
      "epoch": 0.157031978160381,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010906510651065107,
      "loss": 3.1392,
      "step": 454730
    },
    {
      "epoch": 0.1570506213965392,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010906310631063106,
      "loss": 3.0245,
      "step": 454740
    },
    {
      "epoch": 0.15706926463269738,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010906110611061107,
      "loss": 3.2281,
      "step": 454750
    },
    {
      "epoch": 0.15708790786885554,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010905910591059108,
      "loss": 3.0607,
      "step": 454760
    },
    {
      "epoch": 0.15710655110501373,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010905710571057106,
      "loss": 3.1712,
      "step": 454770
    },
    {
      "epoch": 0.15712519434117192,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010905510551055107,
      "loss": 3.1998,
      "step": 454780
    },
    {
      "epoch": 0.1571438375773301,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010905310531053106,
      "loss": 3.2051,
      "step": 454790
    },
    {
      "epoch": 0.1571624808134883,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010905110511051107,
      "loss": 3.147,
      "step": 454800
    },
    {
      "epoch": 0.15718112404964646,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010904910491049105,
      "loss": 3.1595,
      "step": 454810
    },
    {
      "epoch": 0.15719976728580465,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010904710471047106,
      "loss": 3.125,
      "step": 454820
    },
    {
      "epoch": 0.15721841052196284,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010904510451045104,
      "loss": 3.0887,
      "step": 454830
    },
    {
      "epoch": 0.15723705375812103,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010904310431043105,
      "loss": 3.09,
      "step": 454840
    },
    {
      "epoch": 0.15725569699427921,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010904110411041104,
      "loss": 3.0438,
      "step": 454850
    },
    {
      "epoch": 0.1572743402304374,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010903910391039104,
      "loss": 3.0642,
      "step": 454860
    },
    {
      "epoch": 0.15729298346659557,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010903710371037104,
      "loss": 3.0389,
      "step": 454870
    },
    {
      "epoch": 0.15731162670275375,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010903510351035105,
      "loss": 3.0796,
      "step": 454880
    },
    {
      "epoch": 0.15733026993891194,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010903310331033103,
      "loss": 3.1283,
      "step": 454890
    },
    {
      "epoch": 0.15734891317507013,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00010903110311031104,
      "loss": 3.1175,
      "step": 454900
    },
    {
      "epoch": 0.15736755641122832,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010902910291029102,
      "loss": 2.9838,
      "step": 454910
    },
    {
      "epoch": 0.15738619964738648,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010902710271027103,
      "loss": 3.1077,
      "step": 454920
    },
    {
      "epoch": 0.15740484288354467,
      "grad_norm": 0.625,
      "learning_rate": 0.00010902510251025101,
      "loss": 3.1303,
      "step": 454930
    },
    {
      "epoch": 0.15742348611970286,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010902310231023102,
      "loss": 2.9684,
      "step": 454940
    },
    {
      "epoch": 0.15744212935586105,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010902110211021102,
      "loss": 3.1303,
      "step": 454950
    },
    {
      "epoch": 0.15746077259201924,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010901910191019103,
      "loss": 3.1863,
      "step": 454960
    },
    {
      "epoch": 0.1574794158281774,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010901710171017104,
      "loss": 3.0813,
      "step": 454970
    },
    {
      "epoch": 0.1574980590643356,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010901510151015102,
      "loss": 3.0993,
      "step": 454980
    },
    {
      "epoch": 0.15751670230049378,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010901310131013103,
      "loss": 3.1467,
      "step": 454990
    },
    {
      "epoch": 0.15753534553665197,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010901110111011101,
      "loss": 3.0957,
      "step": 455000
    },
    {
      "epoch": 0.15755398877281016,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010900910091009102,
      "loss": 2.9925,
      "step": 455010
    },
    {
      "epoch": 0.15757263200896832,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010900710071007101,
      "loss": 3.0372,
      "step": 455020
    },
    {
      "epoch": 0.1575912752451265,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010900510051005101,
      "loss": 3.063,
      "step": 455030
    },
    {
      "epoch": 0.1576099184812847,
      "grad_norm": 0.765625,
      "learning_rate": 0.000109003100310031,
      "loss": 3.0487,
      "step": 455040
    },
    {
      "epoch": 0.15762856171744288,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010900110011001102,
      "loss": 3.0299,
      "step": 455050
    },
    {
      "epoch": 0.15764720495360107,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000108999099909991,
      "loss": 3.1188,
      "step": 455060
    },
    {
      "epoch": 0.15766584818975923,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000108997099709971,
      "loss": 3.119,
      "step": 455070
    },
    {
      "epoch": 0.15768449142591742,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010899509950995099,
      "loss": 3.1409,
      "step": 455080
    },
    {
      "epoch": 0.1577031346620756,
      "grad_norm": 0.828125,
      "learning_rate": 0.000108993099309931,
      "loss": 3.0936,
      "step": 455090
    },
    {
      "epoch": 0.1577217778982338,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010899109910991099,
      "loss": 3.1194,
      "step": 455100
    },
    {
      "epoch": 0.157740421134392,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000108989098909891,
      "loss": 3.1134,
      "step": 455110
    },
    {
      "epoch": 0.15775906437055018,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010898709870987098,
      "loss": 3.2131,
      "step": 455120
    },
    {
      "epoch": 0.15777770760670834,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000108985098509851,
      "loss": 3.1492,
      "step": 455130
    },
    {
      "epoch": 0.15779635084286653,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010898309830983098,
      "loss": 3.1153,
      "step": 455140
    },
    {
      "epoch": 0.15781499407902472,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010898109810981098,
      "loss": 3.1663,
      "step": 455150
    },
    {
      "epoch": 0.1578336373151829,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000108979097909791,
      "loss": 3.1342,
      "step": 455160
    },
    {
      "epoch": 0.1578522805513411,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010897709770977098,
      "loss": 3.1846,
      "step": 455170
    },
    {
      "epoch": 0.15787092378749926,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010897509750975099,
      "loss": 3.1063,
      "step": 455180
    },
    {
      "epoch": 0.15788956702365745,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010897309730973098,
      "loss": 3.1632,
      "step": 455190
    },
    {
      "epoch": 0.15790821025981563,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010897109710971099,
      "loss": 3.0832,
      "step": 455200
    },
    {
      "epoch": 0.15792685349597382,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010896909690969097,
      "loss": 3.0327,
      "step": 455210
    },
    {
      "epoch": 0.157945496732132,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010896709670967098,
      "loss": 3.164,
      "step": 455220
    },
    {
      "epoch": 0.15796413996829017,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010896509650965096,
      "loss": 3.1541,
      "step": 455230
    },
    {
      "epoch": 0.15798278320444836,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010896309630963097,
      "loss": 3.1562,
      "step": 455240
    },
    {
      "epoch": 0.15800142644060655,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010896109610961095,
      "loss": 3.2166,
      "step": 455250
    },
    {
      "epoch": 0.15802006967676474,
      "grad_norm": 0.75,
      "learning_rate": 0.00010895909590959096,
      "loss": 3.0228,
      "step": 455260
    },
    {
      "epoch": 0.15803871291292293,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010895709570957096,
      "loss": 3.065,
      "step": 455270
    },
    {
      "epoch": 0.1580573561490811,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010895509550955097,
      "loss": 3.1894,
      "step": 455280
    },
    {
      "epoch": 0.15807599938523928,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010895309530953095,
      "loss": 3.0563,
      "step": 455290
    },
    {
      "epoch": 0.15809464262139747,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010895109510951096,
      "loss": 3.0739,
      "step": 455300
    },
    {
      "epoch": 0.15811328585755566,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010894909490949094,
      "loss": 3.0635,
      "step": 455310
    },
    {
      "epoch": 0.15813192909371385,
      "grad_norm": 0.75,
      "learning_rate": 0.00010894709470947095,
      "loss": 3.039,
      "step": 455320
    },
    {
      "epoch": 0.15815057232987204,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010894509450945095,
      "loss": 3.1912,
      "step": 455330
    },
    {
      "epoch": 0.1581692155660302,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010894309430943096,
      "loss": 2.9505,
      "step": 455340
    },
    {
      "epoch": 0.15818785880218839,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010894109410941095,
      "loss": 3.1152,
      "step": 455350
    },
    {
      "epoch": 0.15820650203834657,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010893909390939095,
      "loss": 3.0533,
      "step": 455360
    },
    {
      "epoch": 0.15822514527450476,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010893709370937096,
      "loss": 3.0905,
      "step": 455370
    },
    {
      "epoch": 0.15824378851066295,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010893509350935094,
      "loss": 2.8779,
      "step": 455380
    },
    {
      "epoch": 0.15826243174682111,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010893309330933095,
      "loss": 2.9329,
      "step": 455390
    },
    {
      "epoch": 0.1582810749829793,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010893109310931093,
      "loss": 3.0909,
      "step": 455400
    },
    {
      "epoch": 0.1582997182191375,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010892909290929094,
      "loss": 3.1222,
      "step": 455410
    },
    {
      "epoch": 0.15831836145529568,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010892709270927093,
      "loss": 3.2123,
      "step": 455420
    },
    {
      "epoch": 0.15833700469145387,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010892509250925094,
      "loss": 3.2205,
      "step": 455430
    },
    {
      "epoch": 0.15835564792761203,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010892309230923092,
      "loss": 3.1424,
      "step": 455440
    },
    {
      "epoch": 0.15837429116377022,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010892109210921093,
      "loss": 3.0764,
      "step": 455450
    },
    {
      "epoch": 0.1583929343999284,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010891909190919092,
      "loss": 3.1461,
      "step": 455460
    },
    {
      "epoch": 0.1584115776360866,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010891709170917093,
      "loss": 3.0068,
      "step": 455470
    },
    {
      "epoch": 0.1584302208722448,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010891509150915091,
      "loss": 3.0612,
      "step": 455480
    },
    {
      "epoch": 0.15844886410840295,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010891309130913092,
      "loss": 3.0352,
      "step": 455490
    },
    {
      "epoch": 0.15846750734456114,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010891109110911091,
      "loss": 3.1307,
      "step": 455500
    },
    {
      "epoch": 0.15848615058071933,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010890909090909092,
      "loss": 3.1244,
      "step": 455510
    },
    {
      "epoch": 0.15850479381687752,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001089070907090709,
      "loss": 3.1072,
      "step": 455520
    },
    {
      "epoch": 0.1585234370530357,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010890509050905091,
      "loss": 3.132,
      "step": 455530
    },
    {
      "epoch": 0.1585420802891939,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001089030903090309,
      "loss": 2.9809,
      "step": 455540
    },
    {
      "epoch": 0.15856072352535205,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001089010901090109,
      "loss": 3.1801,
      "step": 455550
    },
    {
      "epoch": 0.15857936676151024,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010889908990899091,
      "loss": 3.0113,
      "step": 455560
    },
    {
      "epoch": 0.15859800999766843,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001088970897089709,
      "loss": 3.18,
      "step": 455570
    },
    {
      "epoch": 0.15861665323382662,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001088950895089509,
      "loss": 3.059,
      "step": 455580
    },
    {
      "epoch": 0.1586352964699848,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001088930893089309,
      "loss": 3.0852,
      "step": 455590
    },
    {
      "epoch": 0.15865393970614297,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010889108910891091,
      "loss": 3.1371,
      "step": 455600
    },
    {
      "epoch": 0.15867258294230116,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010888908890889089,
      "loss": 3.0961,
      "step": 455610
    },
    {
      "epoch": 0.15869122617845935,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001088870887088709,
      "loss": 3.1931,
      "step": 455620
    },
    {
      "epoch": 0.15870986941461754,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010888508850885088,
      "loss": 3.0904,
      "step": 455630
    },
    {
      "epoch": 0.15872851265077573,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010888308830883089,
      "loss": 3.0989,
      "step": 455640
    },
    {
      "epoch": 0.1587471558869339,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010888108810881089,
      "loss": 3.1714,
      "step": 455650
    },
    {
      "epoch": 0.15876579912309208,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001088790879087909,
      "loss": 3.0719,
      "step": 455660
    },
    {
      "epoch": 0.15878444235925027,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010887708770877088,
      "loss": 3.145,
      "step": 455670
    },
    {
      "epoch": 0.15880308559540846,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010887508750875089,
      "loss": 3.1676,
      "step": 455680
    },
    {
      "epoch": 0.15882172883156664,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010887308730873087,
      "loss": 3.0081,
      "step": 455690
    },
    {
      "epoch": 0.1588403720677248,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010887108710871088,
      "loss": 3.0549,
      "step": 455700
    },
    {
      "epoch": 0.158859015303883,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010886908690869086,
      "loss": 3.0484,
      "step": 455710
    },
    {
      "epoch": 0.15887765854004118,
      "grad_norm": 0.75,
      "learning_rate": 0.00010886708670867087,
      "loss": 3.0723,
      "step": 455720
    },
    {
      "epoch": 0.15889630177619937,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010886508650865086,
      "loss": 3.0684,
      "step": 455730
    },
    {
      "epoch": 0.15891494501235756,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010886308630863087,
      "loss": 3.1473,
      "step": 455740
    },
    {
      "epoch": 0.15893358824851572,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010886108610861088,
      "loss": 3.0015,
      "step": 455750
    },
    {
      "epoch": 0.1589522314846739,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010885908590859087,
      "loss": 3.0491,
      "step": 455760
    },
    {
      "epoch": 0.1589708747208321,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010885708570857087,
      "loss": 3.0569,
      "step": 455770
    },
    {
      "epoch": 0.1589895179569903,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010885508550855086,
      "loss": 3.2009,
      "step": 455780
    },
    {
      "epoch": 0.15900816119314848,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010885308530853087,
      "loss": 3.1969,
      "step": 455790
    },
    {
      "epoch": 0.15902680442930667,
      "grad_norm": 0.953125,
      "learning_rate": 0.00010885108510851085,
      "loss": 3.1201,
      "step": 455800
    },
    {
      "epoch": 0.15904544766546483,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010884908490849086,
      "loss": 3.0874,
      "step": 455810
    },
    {
      "epoch": 0.15906409090162302,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010884708470847085,
      "loss": 3.1667,
      "step": 455820
    },
    {
      "epoch": 0.1590827341377812,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010884508450845086,
      "loss": 2.999,
      "step": 455830
    },
    {
      "epoch": 0.1591013773739394,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010884308430843084,
      "loss": 3.1137,
      "step": 455840
    },
    {
      "epoch": 0.15912002061009758,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010884108410841085,
      "loss": 3.1926,
      "step": 455850
    },
    {
      "epoch": 0.15913866384625575,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010883908390839083,
      "loss": 3.1564,
      "step": 455860
    },
    {
      "epoch": 0.15915730708241393,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010883708370837084,
      "loss": 2.9373,
      "step": 455870
    },
    {
      "epoch": 0.15917595031857212,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010883508350835083,
      "loss": 3.0648,
      "step": 455880
    },
    {
      "epoch": 0.1591945935547303,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010883308330833083,
      "loss": 3.2184,
      "step": 455890
    },
    {
      "epoch": 0.1592132367908885,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010883108310831083,
      "loss": 3.0845,
      "step": 455900
    },
    {
      "epoch": 0.15923188002704666,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010882908290829084,
      "loss": 2.9965,
      "step": 455910
    },
    {
      "epoch": 0.15925052326320485,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010882708270827082,
      "loss": 3.1681,
      "step": 455920
    },
    {
      "epoch": 0.15926916649936304,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010882508250825083,
      "loss": 3.1357,
      "step": 455930
    },
    {
      "epoch": 0.15928780973552123,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010882308230823084,
      "loss": 3.0467,
      "step": 455940
    },
    {
      "epoch": 0.15930645297167942,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010882108210821082,
      "loss": 3.1188,
      "step": 455950
    },
    {
      "epoch": 0.15932509620783758,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010881908190819083,
      "loss": 3.0624,
      "step": 455960
    },
    {
      "epoch": 0.15934373944399577,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010881708170817083,
      "loss": 3.1473,
      "step": 455970
    },
    {
      "epoch": 0.15936238268015396,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010881508150815082,
      "loss": 3.1979,
      "step": 455980
    },
    {
      "epoch": 0.15938102591631215,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010881308130813082,
      "loss": 3.0297,
      "step": 455990
    },
    {
      "epoch": 0.15939966915247034,
      "grad_norm": 0.75,
      "learning_rate": 0.00010881108110811083,
      "loss": 3.1377,
      "step": 456000
    },
    {
      "epoch": 0.15941831238862852,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010880908090809081,
      "loss": 3.1677,
      "step": 456010
    },
    {
      "epoch": 0.15943695562478669,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010880708070807082,
      "loss": 3.0131,
      "step": 456020
    },
    {
      "epoch": 0.15945559886094487,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001088050805080508,
      "loss": 3.0446,
      "step": 456030
    },
    {
      "epoch": 0.15947424209710306,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010880308030803081,
      "loss": 3.0786,
      "step": 456040
    },
    {
      "epoch": 0.15949288533326125,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001088010801080108,
      "loss": 3.1267,
      "step": 456050
    },
    {
      "epoch": 0.15951152856941944,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010879907990799081,
      "loss": 3.0938,
      "step": 456060
    },
    {
      "epoch": 0.1595301718055776,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001087970797079708,
      "loss": 3.164,
      "step": 456070
    },
    {
      "epoch": 0.1595488150417358,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001087950795079508,
      "loss": 3.0603,
      "step": 456080
    },
    {
      "epoch": 0.15956745827789398,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010879307930793079,
      "loss": 3.1032,
      "step": 456090
    },
    {
      "epoch": 0.15958610151405217,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001087910791079108,
      "loss": 2.9683,
      "step": 456100
    },
    {
      "epoch": 0.15960474475021036,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010878907890789078,
      "loss": 3.0101,
      "step": 456110
    },
    {
      "epoch": 0.15962338798636852,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010878707870787079,
      "loss": 3.0977,
      "step": 456120
    },
    {
      "epoch": 0.1596420312225267,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010878507850785078,
      "loss": 3.0433,
      "step": 456130
    },
    {
      "epoch": 0.1596606744586849,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010878307830783079,
      "loss": 3.0982,
      "step": 456140
    },
    {
      "epoch": 0.1596793176948431,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001087810781078108,
      "loss": 2.9638,
      "step": 456150
    },
    {
      "epoch": 0.15969796093100128,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010877907790779078,
      "loss": 3.0405,
      "step": 456160
    },
    {
      "epoch": 0.15971660416715944,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010877707770777079,
      "loss": 3.0817,
      "step": 456170
    },
    {
      "epoch": 0.15973524740331763,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010877507750775077,
      "loss": 3.0787,
      "step": 456180
    },
    {
      "epoch": 0.15975389063947582,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010877307730773078,
      "loss": 2.9652,
      "step": 456190
    },
    {
      "epoch": 0.159772533875634,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010877107710771077,
      "loss": 3.1991,
      "step": 456200
    },
    {
      "epoch": 0.1597911771117922,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010876907690769078,
      "loss": 3.1588,
      "step": 456210
    },
    {
      "epoch": 0.15980982034795038,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010876707670767077,
      "loss": 2.9546,
      "step": 456220
    },
    {
      "epoch": 0.15982846358410854,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010876507650765078,
      "loss": 3.0113,
      "step": 456230
    },
    {
      "epoch": 0.15984710682026673,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010876307630763076,
      "loss": 3.1561,
      "step": 456240
    },
    {
      "epoch": 0.15986575005642492,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010876107610761077,
      "loss": 3.1658,
      "step": 456250
    },
    {
      "epoch": 0.1598843932925831,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010875907590759075,
      "loss": 3.047,
      "step": 456260
    },
    {
      "epoch": 0.1599030365287413,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010875707570757076,
      "loss": 3.0903,
      "step": 456270
    },
    {
      "epoch": 0.15992167976489946,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010875507550755076,
      "loss": 3.0194,
      "step": 456280
    },
    {
      "epoch": 0.15994032300105765,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010875307530753077,
      "loss": 3.0967,
      "step": 456290
    },
    {
      "epoch": 0.15995896623721584,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010875107510751075,
      "loss": 3.0689,
      "step": 456300
    },
    {
      "epoch": 0.15997760947337403,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010874907490749076,
      "loss": 3.1737,
      "step": 456310
    },
    {
      "epoch": 0.15999625270953222,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010874707470747074,
      "loss": 3.1371,
      "step": 456320
    },
    {
      "epoch": 0.16001489594569038,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010874507450745075,
      "loss": 2.9625,
      "step": 456330
    },
    {
      "epoch": 0.16003353918184857,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010874307430743076,
      "loss": 3.0154,
      "step": 456340
    },
    {
      "epoch": 0.16005218241800676,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010874107410741074,
      "loss": 3.1858,
      "step": 456350
    },
    {
      "epoch": 0.16007082565416494,
      "grad_norm": 0.75,
      "learning_rate": 0.00010873907390739075,
      "loss": 3.0097,
      "step": 456360
    },
    {
      "epoch": 0.16008946889032313,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010873707370737075,
      "loss": 3.0772,
      "step": 456370
    },
    {
      "epoch": 0.1601081121264813,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010873507350735075,
      "loss": 3.0226,
      "step": 456380
    },
    {
      "epoch": 0.16012675536263948,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010873307330733074,
      "loss": 2.9906,
      "step": 456390
    },
    {
      "epoch": 0.16014539859879767,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010873107310731075,
      "loss": 3.3342,
      "step": 456400
    },
    {
      "epoch": 0.16016404183495586,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010872907290729073,
      "loss": 3.1778,
      "step": 456410
    },
    {
      "epoch": 0.16018268507111405,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010872707270727074,
      "loss": 2.9663,
      "step": 456420
    },
    {
      "epoch": 0.1602013283072722,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010872507250725072,
      "loss": 3.0341,
      "step": 456430
    },
    {
      "epoch": 0.1602199715434304,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010872307230723073,
      "loss": 3.0841,
      "step": 456440
    },
    {
      "epoch": 0.1602386147795886,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010872107210721072,
      "loss": 3.1216,
      "step": 456450
    },
    {
      "epoch": 0.16025725801574678,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010871907190719073,
      "loss": 3.072,
      "step": 456460
    },
    {
      "epoch": 0.16027590125190497,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010871707170717071,
      "loss": 3.1417,
      "step": 456470
    },
    {
      "epoch": 0.16029454448806316,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010871507150715072,
      "loss": 2.971,
      "step": 456480
    },
    {
      "epoch": 0.16031318772422132,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001087130713071307,
      "loss": 3.0732,
      "step": 456490
    },
    {
      "epoch": 0.1603318309603795,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010871107110711072,
      "loss": 2.9058,
      "step": 456500
    },
    {
      "epoch": 0.1603504741965377,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010870907090709071,
      "loss": 3.0517,
      "step": 456510
    },
    {
      "epoch": 0.16036911743269588,
      "grad_norm": 0.75,
      "learning_rate": 0.0001087070707070707,
      "loss": 3.0313,
      "step": 456520
    },
    {
      "epoch": 0.16038776066885407,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010870507050705072,
      "loss": 3.1884,
      "step": 456530
    },
    {
      "epoch": 0.16040640390501223,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010870307030703071,
      "loss": 3.1456,
      "step": 456540
    },
    {
      "epoch": 0.16042504714117042,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010870107010701072,
      "loss": 3.0468,
      "step": 456550
    },
    {
      "epoch": 0.1604436903773286,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001086990699069907,
      "loss": 3.108,
      "step": 456560
    },
    {
      "epoch": 0.1604623336134868,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010869706970697071,
      "loss": 3.0752,
      "step": 456570
    },
    {
      "epoch": 0.160480976849645,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001086950695069507,
      "loss": 3.0307,
      "step": 456580
    },
    {
      "epoch": 0.16049962008580315,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001086930693069307,
      "loss": 2.9498,
      "step": 456590
    },
    {
      "epoch": 0.16051826332196134,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001086910691069107,
      "loss": 3.0433,
      "step": 456600
    },
    {
      "epoch": 0.16053690655811953,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010868906890689071,
      "loss": 2.9499,
      "step": 456610
    },
    {
      "epoch": 0.16055554979427772,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010868706870687069,
      "loss": 3.0643,
      "step": 456620
    },
    {
      "epoch": 0.1605741930304359,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001086850685068507,
      "loss": 3.1061,
      "step": 456630
    },
    {
      "epoch": 0.16059283626659407,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010868306830683068,
      "loss": 3.1461,
      "step": 456640
    },
    {
      "epoch": 0.16061147950275226,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010868106810681069,
      "loss": 2.9708,
      "step": 456650
    },
    {
      "epoch": 0.16063012273891045,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010867906790679067,
      "loss": 3.1841,
      "step": 456660
    },
    {
      "epoch": 0.16064876597506864,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010867706770677068,
      "loss": 3.076,
      "step": 456670
    },
    {
      "epoch": 0.16066740921122682,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010867506750675068,
      "loss": 3.1063,
      "step": 456680
    },
    {
      "epoch": 0.160686052447385,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010867306730673069,
      "loss": 3.1745,
      "step": 456690
    },
    {
      "epoch": 0.16070469568354318,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010867106710671067,
      "loss": 3.2036,
      "step": 456700
    },
    {
      "epoch": 0.16072333891970136,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010866906690669068,
      "loss": 3.0129,
      "step": 456710
    },
    {
      "epoch": 0.16074198215585955,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010866706670667066,
      "loss": 3.1577,
      "step": 456720
    },
    {
      "epoch": 0.16076062539201774,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010866506650665067,
      "loss": 3.0529,
      "step": 456730
    },
    {
      "epoch": 0.16077926862817593,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010866306630663068,
      "loss": 3.1067,
      "step": 456740
    },
    {
      "epoch": 0.1607979118643341,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010866106610661066,
      "loss": 3.1443,
      "step": 456750
    },
    {
      "epoch": 0.16081655510049228,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010865906590659067,
      "loss": 3.1674,
      "step": 456760
    },
    {
      "epoch": 0.16083519833665047,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010865706570657066,
      "loss": 3.0079,
      "step": 456770
    },
    {
      "epoch": 0.16085384157280866,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010865506550655067,
      "loss": 3.1085,
      "step": 456780
    },
    {
      "epoch": 0.16087248480896685,
      "grad_norm": 0.75,
      "learning_rate": 0.00010865306530653066,
      "loss": 3.1197,
      "step": 456790
    },
    {
      "epoch": 0.160891128045125,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010865106510651066,
      "loss": 3.1547,
      "step": 456800
    },
    {
      "epoch": 0.1609097712812832,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010864906490649065,
      "loss": 3.0948,
      "step": 456810
    },
    {
      "epoch": 0.1609284145174414,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010864706470647066,
      "loss": 3.0217,
      "step": 456820
    },
    {
      "epoch": 0.16094705775359958,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010864506450645065,
      "loss": 3.0808,
      "step": 456830
    },
    {
      "epoch": 0.16096570098975777,
      "grad_norm": 0.625,
      "learning_rate": 0.00010864306430643065,
      "loss": 2.992,
      "step": 456840
    },
    {
      "epoch": 0.16098434422591593,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010864106410641064,
      "loss": 3.035,
      "step": 456850
    },
    {
      "epoch": 0.16100298746207412,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010863906390639065,
      "loss": 3.0912,
      "step": 456860
    },
    {
      "epoch": 0.1610216306982323,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010863706370637063,
      "loss": 3.1559,
      "step": 456870
    },
    {
      "epoch": 0.1610402739343905,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010863506350635064,
      "loss": 3.132,
      "step": 456880
    },
    {
      "epoch": 0.16105891717054868,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010863306330633062,
      "loss": 3.1322,
      "step": 456890
    },
    {
      "epoch": 0.16107756040670687,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010863106310631063,
      "loss": 3.2869,
      "step": 456900
    },
    {
      "epoch": 0.16109620364286503,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010862906290629063,
      "loss": 3.0546,
      "step": 456910
    },
    {
      "epoch": 0.16111484687902322,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010862706270627064,
      "loss": 3.1591,
      "step": 456920
    },
    {
      "epoch": 0.1611334901151814,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010862506250625065,
      "loss": 3.0594,
      "step": 456930
    },
    {
      "epoch": 0.1611521333513396,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010862306230623063,
      "loss": 3.1409,
      "step": 456940
    },
    {
      "epoch": 0.1611707765874978,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010862106210621064,
      "loss": 3.0481,
      "step": 456950
    },
    {
      "epoch": 0.16118941982365595,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010861906190619062,
      "loss": 3.0365,
      "step": 456960
    },
    {
      "epoch": 0.16120806305981414,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010861706170617063,
      "loss": 3.2308,
      "step": 456970
    },
    {
      "epoch": 0.16122670629597233,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010861506150615061,
      "loss": 3.0567,
      "step": 456980
    },
    {
      "epoch": 0.16124534953213052,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010861306130613062,
      "loss": 2.9824,
      "step": 456990
    },
    {
      "epoch": 0.1612639927682887,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010861106110611062,
      "loss": 3.0575,
      "step": 457000
    },
    {
      "epoch": 0.16128263600444687,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010860906090609063,
      "loss": 3.0831,
      "step": 457010
    },
    {
      "epoch": 0.16130127924060506,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010860706070607061,
      "loss": 3.2102,
      "step": 457020
    },
    {
      "epoch": 0.16131992247676324,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010860506050605062,
      "loss": 3.1293,
      "step": 457030
    },
    {
      "epoch": 0.16133856571292143,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001086030603060306,
      "loss": 3.1304,
      "step": 457040
    },
    {
      "epoch": 0.16135720894907962,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010860106010601061,
      "loss": 3.0601,
      "step": 457050
    },
    {
      "epoch": 0.16137585218523778,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010859905990599059,
      "loss": 3.1776,
      "step": 457060
    },
    {
      "epoch": 0.16139449542139597,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001085970597059706,
      "loss": 2.9963,
      "step": 457070
    },
    {
      "epoch": 0.16141313865755416,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001085950595059506,
      "loss": 2.9566,
      "step": 457080
    },
    {
      "epoch": 0.16143178189371235,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001085930593059306,
      "loss": 3.1337,
      "step": 457090
    },
    {
      "epoch": 0.16145042512987054,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010859105910591059,
      "loss": 2.9331,
      "step": 457100
    },
    {
      "epoch": 0.1614690683660287,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001085890589058906,
      "loss": 3.1107,
      "step": 457110
    },
    {
      "epoch": 0.1614877116021869,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010858705870587058,
      "loss": 3.119,
      "step": 457120
    },
    {
      "epoch": 0.16150635483834508,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010858505850585059,
      "loss": 3.0071,
      "step": 457130
    },
    {
      "epoch": 0.16152499807450327,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001085830583058306,
      "loss": 2.9864,
      "step": 457140
    },
    {
      "epoch": 0.16154364131066146,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010858105810581059,
      "loss": 3.1837,
      "step": 457150
    },
    {
      "epoch": 0.16156228454681965,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010857905790579059,
      "loss": 3.1019,
      "step": 457160
    },
    {
      "epoch": 0.1615809277829778,
      "grad_norm": 0.75,
      "learning_rate": 0.00010857705770577058,
      "loss": 3.1247,
      "step": 457170
    },
    {
      "epoch": 0.161599571019136,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010857505750575059,
      "loss": 3.1074,
      "step": 457180
    },
    {
      "epoch": 0.16161821425529418,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010857305730573057,
      "loss": 3.1396,
      "step": 457190
    },
    {
      "epoch": 0.16163685749145237,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010857105710571058,
      "loss": 3.019,
      "step": 457200
    },
    {
      "epoch": 0.16165550072761056,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010856905690569057,
      "loss": 3.1241,
      "step": 457210
    },
    {
      "epoch": 0.16167414396376872,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00010856705670567057,
      "loss": 2.7425,
      "step": 457220
    },
    {
      "epoch": 0.1616927871999269,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010856505650565057,
      "loss": 3.1879,
      "step": 457230
    },
    {
      "epoch": 0.1617114304360851,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010856305630563058,
      "loss": 2.9903,
      "step": 457240
    },
    {
      "epoch": 0.1617300736722433,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010856105610561056,
      "loss": 3.115,
      "step": 457250
    },
    {
      "epoch": 0.16174871690840148,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010855905590559057,
      "loss": 3.1116,
      "step": 457260
    },
    {
      "epoch": 0.16176736014455964,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010855705570557055,
      "loss": 3.0459,
      "step": 457270
    },
    {
      "epoch": 0.16178600338071783,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010855505550555056,
      "loss": 3.0588,
      "step": 457280
    },
    {
      "epoch": 0.16180464661687602,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010855305530553054,
      "loss": 3.0056,
      "step": 457290
    },
    {
      "epoch": 0.1618232898530342,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010855105510551055,
      "loss": 2.9717,
      "step": 457300
    },
    {
      "epoch": 0.1618419330891924,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010854905490549055,
      "loss": 3.004,
      "step": 457310
    },
    {
      "epoch": 0.16186057632535056,
      "grad_norm": 0.75,
      "learning_rate": 0.00010854705470547056,
      "loss": 3.16,
      "step": 457320
    },
    {
      "epoch": 0.16187921956150875,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010854505450545057,
      "loss": 3.0965,
      "step": 457330
    },
    {
      "epoch": 0.16189786279766694,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010854305430543055,
      "loss": 3.179,
      "step": 457340
    },
    {
      "epoch": 0.16191650603382512,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010854105410541056,
      "loss": 3.1598,
      "step": 457350
    },
    {
      "epoch": 0.16193514926998331,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010853905390539054,
      "loss": 3.0863,
      "step": 457360
    },
    {
      "epoch": 0.1619537925061415,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010853705370537055,
      "loss": 3.1416,
      "step": 457370
    },
    {
      "epoch": 0.16197243574229966,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010853505350535053,
      "loss": 3.015,
      "step": 457380
    },
    {
      "epoch": 0.16199107897845785,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010853305330533054,
      "loss": 3.0076,
      "step": 457390
    },
    {
      "epoch": 0.16200972221461604,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010853105310531054,
      "loss": 2.9674,
      "step": 457400
    },
    {
      "epoch": 0.16202836545077423,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010852905290529054,
      "loss": 2.9954,
      "step": 457410
    },
    {
      "epoch": 0.16204700868693242,
      "grad_norm": 0.75,
      "learning_rate": 0.00010852705270527053,
      "loss": 3.1689,
      "step": 457420
    },
    {
      "epoch": 0.16206565192309058,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010852505250525054,
      "loss": 3.1152,
      "step": 457430
    },
    {
      "epoch": 0.16208429515924877,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010852305230523052,
      "loss": 3.0467,
      "step": 457440
    },
    {
      "epoch": 0.16210293839540696,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010852105210521053,
      "loss": 3.1251,
      "step": 457450
    },
    {
      "epoch": 0.16212158163156515,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010851905190519052,
      "loss": 3.0937,
      "step": 457460
    },
    {
      "epoch": 0.16214022486772334,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010851705170517052,
      "loss": 3.1849,
      "step": 457470
    },
    {
      "epoch": 0.1621588681038815,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010851505150515051,
      "loss": 3.1054,
      "step": 457480
    },
    {
      "epoch": 0.1621775113400397,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010851305130513052,
      "loss": 3.0019,
      "step": 457490
    },
    {
      "epoch": 0.16219615457619788,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001085110511051105,
      "loss": 3.1198,
      "step": 457500
    },
    {
      "epoch": 0.16221479781235607,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010850905090509051,
      "loss": 3.1075,
      "step": 457510
    },
    {
      "epoch": 0.16223344104851425,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010850705070507052,
      "loss": 3.0622,
      "step": 457520
    },
    {
      "epoch": 0.16225208428467242,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001085050505050505,
      "loss": 3.0546,
      "step": 457530
    },
    {
      "epoch": 0.1622707275208306,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010850305030503051,
      "loss": 3.2069,
      "step": 457540
    },
    {
      "epoch": 0.1622893707569888,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010850105010501051,
      "loss": 3.1399,
      "step": 457550
    },
    {
      "epoch": 0.16230801399314698,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010849904990499052,
      "loss": 3.1631,
      "step": 457560
    },
    {
      "epoch": 0.16232665722930517,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001084970497049705,
      "loss": 3.1684,
      "step": 457570
    },
    {
      "epoch": 0.16234530046546336,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010849504950495051,
      "loss": 3.14,
      "step": 457580
    },
    {
      "epoch": 0.16236394370162152,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010849304930493049,
      "loss": 3.1002,
      "step": 457590
    },
    {
      "epoch": 0.1623825869377797,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001084910491049105,
      "loss": 3.1291,
      "step": 457600
    },
    {
      "epoch": 0.1624012301739379,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010848904890489048,
      "loss": 3.1268,
      "step": 457610
    },
    {
      "epoch": 0.1624198734100961,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010848704870487049,
      "loss": 3.1752,
      "step": 457620
    },
    {
      "epoch": 0.16243851664625428,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010848504850485049,
      "loss": 3.189,
      "step": 457630
    },
    {
      "epoch": 0.16245715988241244,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001084830483048305,
      "loss": 3.0711,
      "step": 457640
    },
    {
      "epoch": 0.16247580311857063,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010848104810481048,
      "loss": 2.9714,
      "step": 457650
    },
    {
      "epoch": 0.16249444635472882,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00010847904790479049,
      "loss": 3.011,
      "step": 457660
    },
    {
      "epoch": 0.162513089590887,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010847704770477047,
      "loss": 3.0955,
      "step": 457670
    },
    {
      "epoch": 0.1625317328270452,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010847504750475048,
      "loss": 3.0591,
      "step": 457680
    },
    {
      "epoch": 0.16255037606320336,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010847304730473046,
      "loss": 2.9092,
      "step": 457690
    },
    {
      "epoch": 0.16256901929936154,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010847104710471047,
      "loss": 3.172,
      "step": 457700
    },
    {
      "epoch": 0.16258766253551973,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010846904690469047,
      "loss": 2.9447,
      "step": 457710
    },
    {
      "epoch": 0.16260630577167792,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010846704670467048,
      "loss": 3.1459,
      "step": 457720
    },
    {
      "epoch": 0.1626249490078361,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010846504650465048,
      "loss": 3.1856,
      "step": 457730
    },
    {
      "epoch": 0.16264359224399427,
      "grad_norm": 0.75,
      "learning_rate": 0.00010846304630463047,
      "loss": 3.0786,
      "step": 457740
    },
    {
      "epoch": 0.16266223548015246,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010846104610461048,
      "loss": 3.0449,
      "step": 457750
    },
    {
      "epoch": 0.16268087871631065,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010845904590459046,
      "loss": 3.2611,
      "step": 457760
    },
    {
      "epoch": 0.16269952195246884,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010845704570457047,
      "loss": 3.1627,
      "step": 457770
    },
    {
      "epoch": 0.16271816518862703,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010845504550455046,
      "loss": 3.1778,
      "step": 457780
    },
    {
      "epoch": 0.16273680842478522,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010845304530453046,
      "loss": 3.096,
      "step": 457790
    },
    {
      "epoch": 0.16275545166094338,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010845104510451045,
      "loss": 3.0865,
      "step": 457800
    },
    {
      "epoch": 0.16277409489710157,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010844904490449046,
      "loss": 3.1613,
      "step": 457810
    },
    {
      "epoch": 0.16279273813325976,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010844704470447045,
      "loss": 3.0946,
      "step": 457820
    },
    {
      "epoch": 0.16281138136941795,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010844504450445045,
      "loss": 3.1491,
      "step": 457830
    },
    {
      "epoch": 0.16283002460557613,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010844304430443044,
      "loss": 3.0251,
      "step": 457840
    },
    {
      "epoch": 0.1628486678417343,
      "grad_norm": 0.875,
      "learning_rate": 0.00010844104410441045,
      "loss": 2.9957,
      "step": 457850
    },
    {
      "epoch": 0.16286731107789248,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010843904390439044,
      "loss": 3.1058,
      "step": 457860
    },
    {
      "epoch": 0.16288595431405067,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010843704370437045,
      "loss": 3.1075,
      "step": 457870
    },
    {
      "epoch": 0.16290459755020886,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010843504350435043,
      "loss": 3.0034,
      "step": 457880
    },
    {
      "epoch": 0.16292324078636705,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010843304330433044,
      "loss": 3.1571,
      "step": 457890
    },
    {
      "epoch": 0.1629418840225252,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010843104310431042,
      "loss": 3.0148,
      "step": 457900
    },
    {
      "epoch": 0.1629605272586834,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010842904290429043,
      "loss": 3.1027,
      "step": 457910
    },
    {
      "epoch": 0.1629791704948416,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010842704270427044,
      "loss": 2.9569,
      "step": 457920
    },
    {
      "epoch": 0.16299781373099978,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010842504250425042,
      "loss": 3.084,
      "step": 457930
    },
    {
      "epoch": 0.16301645696715797,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010842304230423043,
      "loss": 3.1548,
      "step": 457940
    },
    {
      "epoch": 0.16303510020331613,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010842104210421043,
      "loss": 3.2127,
      "step": 457950
    },
    {
      "epoch": 0.16305374343947432,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010841904190419044,
      "loss": 3.1074,
      "step": 457960
    },
    {
      "epoch": 0.1630723866756325,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010841704170417042,
      "loss": 3.0924,
      "step": 457970
    },
    {
      "epoch": 0.1630910299117907,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010841504150415043,
      "loss": 3.1791,
      "step": 457980
    },
    {
      "epoch": 0.16310967314794889,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010841304130413041,
      "loss": 3.1336,
      "step": 457990
    },
    {
      "epoch": 0.16312831638410705,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010841104110411042,
      "loss": 3.107,
      "step": 458000
    },
    {
      "epoch": 0.16314695962026524,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001084090409040904,
      "loss": 3.0849,
      "step": 458010
    },
    {
      "epoch": 0.16316560285642343,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010840704070407041,
      "loss": 3.0034,
      "step": 458020
    },
    {
      "epoch": 0.16318424609258161,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010840504050405041,
      "loss": 2.9805,
      "step": 458030
    },
    {
      "epoch": 0.1632028893287398,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010840304030403042,
      "loss": 3.0316,
      "step": 458040
    },
    {
      "epoch": 0.163221532564898,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001084010401040104,
      "loss": 3.0428,
      "step": 458050
    },
    {
      "epoch": 0.16324017580105615,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010839903990399041,
      "loss": 3.1939,
      "step": 458060
    },
    {
      "epoch": 0.16325881903721434,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010839703970397039,
      "loss": 3.146,
      "step": 458070
    },
    {
      "epoch": 0.16327746227337253,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001083950395039504,
      "loss": 3.0861,
      "step": 458080
    },
    {
      "epoch": 0.16329610550953072,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001083930393039304,
      "loss": 3.1955,
      "step": 458090
    },
    {
      "epoch": 0.1633147487456889,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001083910391039104,
      "loss": 3.1135,
      "step": 458100
    },
    {
      "epoch": 0.16333339198184707,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001083890389038904,
      "loss": 3.1721,
      "step": 458110
    },
    {
      "epoch": 0.16335203521800526,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001083870387038704,
      "loss": 3.1663,
      "step": 458120
    },
    {
      "epoch": 0.16337067845416345,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001083850385038504,
      "loss": 3.0731,
      "step": 458130
    },
    {
      "epoch": 0.16338932169032164,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010838303830383039,
      "loss": 3.1284,
      "step": 458140
    },
    {
      "epoch": 0.16340796492647983,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001083810381038104,
      "loss": 3.1317,
      "step": 458150
    },
    {
      "epoch": 0.163426608162638,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010837903790379038,
      "loss": 3.0293,
      "step": 458160
    },
    {
      "epoch": 0.16344525139879618,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010837703770377039,
      "loss": 3.1516,
      "step": 458170
    },
    {
      "epoch": 0.16346389463495437,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010837503750375038,
      "loss": 3.1013,
      "step": 458180
    },
    {
      "epoch": 0.16348253787111255,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010837303730373039,
      "loss": 3.1166,
      "step": 458190
    },
    {
      "epoch": 0.16350118110727074,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010837103710371037,
      "loss": 3.1639,
      "step": 458200
    },
    {
      "epoch": 0.1635198243434289,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010836903690369038,
      "loss": 3.1826,
      "step": 458210
    },
    {
      "epoch": 0.1635384675795871,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010836703670367036,
      "loss": 2.9367,
      "step": 458220
    },
    {
      "epoch": 0.16355711081574528,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010836503650365037,
      "loss": 3.0687,
      "step": 458230
    },
    {
      "epoch": 0.16357575405190347,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010836303630363036,
      "loss": 3.2321,
      "step": 458240
    },
    {
      "epoch": 0.16359439728806166,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010836103610361036,
      "loss": 3.1188,
      "step": 458250
    },
    {
      "epoch": 0.16361304052421985,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010835903590359036,
      "loss": 3.1214,
      "step": 458260
    },
    {
      "epoch": 0.163631683760378,
      "grad_norm": 0.75,
      "learning_rate": 0.00010835703570357037,
      "loss": 3.2068,
      "step": 458270
    },
    {
      "epoch": 0.1636503269965362,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010835503550355035,
      "loss": 2.9839,
      "step": 458280
    },
    {
      "epoch": 0.1636689702326944,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010835303530353036,
      "loss": 3.1181,
      "step": 458290
    },
    {
      "epoch": 0.16368761346885258,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010835103510351034,
      "loss": 3.0719,
      "step": 458300
    },
    {
      "epoch": 0.16370625670501077,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010834903490349035,
      "loss": 3.0673,
      "step": 458310
    },
    {
      "epoch": 0.16372489994116893,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010834703470347036,
      "loss": 3.1077,
      "step": 458320
    },
    {
      "epoch": 0.16374354317732712,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010834503450345034,
      "loss": 2.9805,
      "step": 458330
    },
    {
      "epoch": 0.1637621864134853,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010834303430343035,
      "loss": 3.0247,
      "step": 458340
    },
    {
      "epoch": 0.1637808296496435,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010834103410341035,
      "loss": 3.0037,
      "step": 458350
    },
    {
      "epoch": 0.16379947288580168,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010833903390339036,
      "loss": 3.0654,
      "step": 458360
    },
    {
      "epoch": 0.16381811612195984,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010833703370337034,
      "loss": 3.1113,
      "step": 458370
    },
    {
      "epoch": 0.16383675935811803,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010833503350335035,
      "loss": 2.9293,
      "step": 458380
    },
    {
      "epoch": 0.16385540259427622,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010833303330333033,
      "loss": 3.198,
      "step": 458390
    },
    {
      "epoch": 0.1638740458304344,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010833103310331034,
      "loss": 3.0984,
      "step": 458400
    },
    {
      "epoch": 0.1638926890665926,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010832903290329033,
      "loss": 3.0317,
      "step": 458410
    },
    {
      "epoch": 0.16391133230275076,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010832703270327034,
      "loss": 3.009,
      "step": 458420
    },
    {
      "epoch": 0.16392997553890895,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010832503250325033,
      "loss": 3.0807,
      "step": 458430
    },
    {
      "epoch": 0.16394861877506714,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010832303230323033,
      "loss": 3.0777,
      "step": 458440
    },
    {
      "epoch": 0.16396726201122533,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010832103210321032,
      "loss": 3.1339,
      "step": 458450
    },
    {
      "epoch": 0.16398590524738352,
      "grad_norm": 0.75,
      "learning_rate": 0.00010831903190319033,
      "loss": 3.0504,
      "step": 458460
    },
    {
      "epoch": 0.1640045484835417,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010831703170317031,
      "loss": 3.1337,
      "step": 458470
    },
    {
      "epoch": 0.16402319171969987,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010831503150315032,
      "loss": 2.938,
      "step": 458480
    },
    {
      "epoch": 0.16404183495585806,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010831303130313031,
      "loss": 3.0365,
      "step": 458490
    },
    {
      "epoch": 0.16406047819201625,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010831103110311032,
      "loss": 3.1387,
      "step": 458500
    },
    {
      "epoch": 0.16407912142817443,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010830903090309033,
      "loss": 3.0039,
      "step": 458510
    },
    {
      "epoch": 0.16409776466433262,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010830703070307031,
      "loss": 3.1469,
      "step": 458520
    },
    {
      "epoch": 0.16411640790049078,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010830503050305032,
      "loss": 3.1948,
      "step": 458530
    },
    {
      "epoch": 0.16413505113664897,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001083030303030303,
      "loss": 3.1081,
      "step": 458540
    },
    {
      "epoch": 0.16415369437280716,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010830103010301031,
      "loss": 3.1402,
      "step": 458550
    },
    {
      "epoch": 0.16417233760896535,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001082990299029903,
      "loss": 3.0298,
      "step": 458560
    },
    {
      "epoch": 0.16419098084512354,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001082970297029703,
      "loss": 3.0927,
      "step": 458570
    },
    {
      "epoch": 0.1642096240812817,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001082950295029503,
      "loss": 3.0401,
      "step": 458580
    },
    {
      "epoch": 0.1642282673174399,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010829302930293031,
      "loss": 3.0423,
      "step": 458590
    },
    {
      "epoch": 0.16424691055359808,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010829102910291029,
      "loss": 3.1648,
      "step": 458600
    },
    {
      "epoch": 0.16426555378975627,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001082890289028903,
      "loss": 3.0382,
      "step": 458610
    },
    {
      "epoch": 0.16428419702591446,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010828702870287028,
      "loss": 3.125,
      "step": 458620
    },
    {
      "epoch": 0.16430284026207262,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010828502850285029,
      "loss": 3.0356,
      "step": 458630
    },
    {
      "epoch": 0.1643214834982308,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010828302830283027,
      "loss": 2.9697,
      "step": 458640
    },
    {
      "epoch": 0.164340126734389,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010828102810281028,
      "loss": 3.098,
      "step": 458650
    },
    {
      "epoch": 0.16435876997054719,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010827902790279028,
      "loss": 3.1691,
      "step": 458660
    },
    {
      "epoch": 0.16437741320670538,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010827702770277029,
      "loss": 3.0958,
      "step": 458670
    },
    {
      "epoch": 0.16439605644286354,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010827502750275027,
      "loss": 3.2225,
      "step": 458680
    },
    {
      "epoch": 0.16441469967902173,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010827302730273028,
      "loss": 3.0755,
      "step": 458690
    },
    {
      "epoch": 0.16443334291517991,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010827102710271026,
      "loss": 3.0818,
      "step": 458700
    },
    {
      "epoch": 0.1644519861513381,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010826902690269027,
      "loss": 3.0978,
      "step": 458710
    },
    {
      "epoch": 0.1644706293874963,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010826702670267028,
      "loss": 3.0641,
      "step": 458720
    },
    {
      "epoch": 0.16448927262365448,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010826502650265027,
      "loss": 3.1489,
      "step": 458730
    },
    {
      "epoch": 0.16450791585981264,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010826302630263027,
      "loss": 3.1517,
      "step": 458740
    },
    {
      "epoch": 0.16452655909597083,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010826102610261027,
      "loss": 2.9815,
      "step": 458750
    },
    {
      "epoch": 0.16454520233212902,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010825902590259028,
      "loss": 3.1972,
      "step": 458760
    },
    {
      "epoch": 0.1645638455682872,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010825702570257026,
      "loss": 3.0401,
      "step": 458770
    },
    {
      "epoch": 0.1645824888044454,
      "grad_norm": 0.625,
      "learning_rate": 0.00010825502550255027,
      "loss": 3.1583,
      "step": 458780
    },
    {
      "epoch": 0.16460113204060356,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010825302530253025,
      "loss": 3.0792,
      "step": 458790
    },
    {
      "epoch": 0.16461977527676175,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010825102510251026,
      "loss": 3.1314,
      "step": 458800
    },
    {
      "epoch": 0.16463841851291994,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010824902490249025,
      "loss": 3.095,
      "step": 458810
    },
    {
      "epoch": 0.16465706174907813,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010824702470247026,
      "loss": 3.1149,
      "step": 458820
    },
    {
      "epoch": 0.16467570498523632,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010824502450245024,
      "loss": 3.1893,
      "step": 458830
    },
    {
      "epoch": 0.16469434822139448,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010824302430243025,
      "loss": 2.9926,
      "step": 458840
    },
    {
      "epoch": 0.16471299145755267,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010824102410241024,
      "loss": 2.9669,
      "step": 458850
    },
    {
      "epoch": 0.16473163469371085,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010823902390239024,
      "loss": 3.0187,
      "step": 458860
    },
    {
      "epoch": 0.16475027792986904,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010823702370237023,
      "loss": 3.1273,
      "step": 458870
    },
    {
      "epoch": 0.16476892116602723,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010823502350235024,
      "loss": 2.9749,
      "step": 458880
    },
    {
      "epoch": 0.1647875644021854,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010823302330233023,
      "loss": 3.1361,
      "step": 458890
    },
    {
      "epoch": 0.16480620763834358,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010823102310231024,
      "loss": 2.9887,
      "step": 458900
    },
    {
      "epoch": 0.16482485087450177,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010822902290229025,
      "loss": 3.0658,
      "step": 458910
    },
    {
      "epoch": 0.16484349411065996,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010822702270227023,
      "loss": 2.8577,
      "step": 458920
    },
    {
      "epoch": 0.16486213734681815,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010822502250225024,
      "loss": 2.9324,
      "step": 458930
    },
    {
      "epoch": 0.16488078058297634,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010822302230223022,
      "loss": 3.0021,
      "step": 458940
    },
    {
      "epoch": 0.1648994238191345,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010822102210221023,
      "loss": 3.1527,
      "step": 458950
    },
    {
      "epoch": 0.1649180670552927,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010821902190219021,
      "loss": 3.0202,
      "step": 458960
    },
    {
      "epoch": 0.16493671029145088,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010821702170217022,
      "loss": 3.1076,
      "step": 458970
    },
    {
      "epoch": 0.16495535352760907,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010821502150215022,
      "loss": 3.1152,
      "step": 458980
    },
    {
      "epoch": 0.16497399676376726,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010821302130213023,
      "loss": 3.0293,
      "step": 458990
    },
    {
      "epoch": 0.16499263999992542,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010821102110211021,
      "loss": 2.9969,
      "step": 459000
    },
    {
      "epoch": 0.1650112832360836,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010820902090209022,
      "loss": 3.2732,
      "step": 459010
    },
    {
      "epoch": 0.1650299264722418,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001082070207020702,
      "loss": 3.1706,
      "step": 459020
    },
    {
      "epoch": 0.16504856970839998,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010820502050205021,
      "loss": 3.0553,
      "step": 459030
    },
    {
      "epoch": 0.16506721294455817,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001082030203020302,
      "loss": 3.1113,
      "step": 459040
    },
    {
      "epoch": 0.16508585618071633,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010820102010201022,
      "loss": 3.0466,
      "step": 459050
    },
    {
      "epoch": 0.16510449941687452,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001081990199019902,
      "loss": 2.9603,
      "step": 459060
    },
    {
      "epoch": 0.1651231426530327,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001081970197019702,
      "loss": 3.119,
      "step": 459070
    },
    {
      "epoch": 0.1651417858891909,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010819501950195019,
      "loss": 3.0407,
      "step": 459080
    },
    {
      "epoch": 0.1651604291253491,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001081930193019302,
      "loss": 3.1441,
      "step": 459090
    },
    {
      "epoch": 0.16517907236150725,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001081910191019102,
      "loss": 3.1426,
      "step": 459100
    },
    {
      "epoch": 0.16519771559766544,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010818901890189019,
      "loss": 2.9985,
      "step": 459110
    },
    {
      "epoch": 0.16521635883382363,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001081870187018702,
      "loss": 3.0891,
      "step": 459120
    },
    {
      "epoch": 0.16523500206998182,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001081850185018502,
      "loss": 3.0438,
      "step": 459130
    },
    {
      "epoch": 0.16525364530614,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001081830183018302,
      "loss": 3.1385,
      "step": 459140
    },
    {
      "epoch": 0.1652722885422982,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010818101810181018,
      "loss": 2.9851,
      "step": 459150
    },
    {
      "epoch": 0.16529093177845636,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001081790179017902,
      "loss": 3.1347,
      "step": 459160
    },
    {
      "epoch": 0.16530957501461455,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010817701770177018,
      "loss": 3.128,
      "step": 459170
    },
    {
      "epoch": 0.16532821825077273,
      "grad_norm": 0.984375,
      "learning_rate": 0.00010817501750175018,
      "loss": 3.037,
      "step": 459180
    },
    {
      "epoch": 0.16534686148693092,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010817301730173017,
      "loss": 3.1218,
      "step": 459190
    },
    {
      "epoch": 0.1653655047230891,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010817101710171018,
      "loss": 3.201,
      "step": 459200
    },
    {
      "epoch": 0.16538414795924727,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010816901690169017,
      "loss": 3.1466,
      "step": 459210
    },
    {
      "epoch": 0.16540279119540546,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010816701670167018,
      "loss": 3.0722,
      "step": 459220
    },
    {
      "epoch": 0.16542143443156365,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010816501650165016,
      "loss": 3.1039,
      "step": 459230
    },
    {
      "epoch": 0.16544007766772184,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010816301630163017,
      "loss": 3.038,
      "step": 459240
    },
    {
      "epoch": 0.16545872090388003,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010816101610161015,
      "loss": 3.0588,
      "step": 459250
    },
    {
      "epoch": 0.1654773641400382,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010815901590159016,
      "loss": 3.0486,
      "step": 459260
    },
    {
      "epoch": 0.16549600737619638,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010815701570157016,
      "loss": 3.2136,
      "step": 459270
    },
    {
      "epoch": 0.16551465061235457,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010815501550155015,
      "loss": 3.01,
      "step": 459280
    },
    {
      "epoch": 0.16553329384851276,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010815301530153015,
      "loss": 2.9918,
      "step": 459290
    },
    {
      "epoch": 0.16555193708467095,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010815101510151016,
      "loss": 3.0826,
      "step": 459300
    },
    {
      "epoch": 0.1655705803208291,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010814901490149017,
      "loss": 3.134,
      "step": 459310
    },
    {
      "epoch": 0.1655892235569873,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010814701470147015,
      "loss": 3.0036,
      "step": 459320
    },
    {
      "epoch": 0.1656078667931455,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010814501450145016,
      "loss": 3.1182,
      "step": 459330
    },
    {
      "epoch": 0.16562651002930368,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010814301430143014,
      "loss": 2.985,
      "step": 459340
    },
    {
      "epoch": 0.16564515326546186,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010814101410141015,
      "loss": 3.166,
      "step": 459350
    },
    {
      "epoch": 0.16566379650162003,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010813901390139015,
      "loss": 3.0747,
      "step": 459360
    },
    {
      "epoch": 0.16568243973777821,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010813701370137016,
      "loss": 3.0831,
      "step": 459370
    },
    {
      "epoch": 0.1657010829739364,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010813501350135014,
      "loss": 3.0617,
      "step": 459380
    },
    {
      "epoch": 0.1657197262100946,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010813301330133015,
      "loss": 2.9081,
      "step": 459390
    },
    {
      "epoch": 0.16573836944625278,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010813101310131013,
      "loss": 3.0277,
      "step": 459400
    },
    {
      "epoch": 0.16575701268241097,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010812901290129014,
      "loss": 3.1529,
      "step": 459410
    },
    {
      "epoch": 0.16577565591856913,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010812701270127012,
      "loss": 3.1833,
      "step": 459420
    },
    {
      "epoch": 0.16579429915472732,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010812501250125013,
      "loss": 3.0358,
      "step": 459430
    },
    {
      "epoch": 0.1658129423908855,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010812301230123012,
      "loss": 3.0424,
      "step": 459440
    },
    {
      "epoch": 0.1658315856270437,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010812101210121013,
      "loss": 3.0066,
      "step": 459450
    },
    {
      "epoch": 0.1658502288632019,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010811901190119012,
      "loss": 3.0752,
      "step": 459460
    },
    {
      "epoch": 0.16586887209936005,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010811701170117012,
      "loss": 3.0835,
      "step": 459470
    },
    {
      "epoch": 0.16588751533551824,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010811501150115011,
      "loss": 3.0674,
      "step": 459480
    },
    {
      "epoch": 0.16590615857167643,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010811301130113012,
      "loss": 3.1519,
      "step": 459490
    },
    {
      "epoch": 0.16592480180783462,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010811101110111013,
      "loss": 3.2019,
      "step": 459500
    },
    {
      "epoch": 0.1659434450439928,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010810901090109011,
      "loss": 3.2049,
      "step": 459510
    },
    {
      "epoch": 0.16596208828015097,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010810701070107012,
      "loss": 3.0324,
      "step": 459520
    },
    {
      "epoch": 0.16598073151630915,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010810501050105011,
      "loss": 3.2179,
      "step": 459530
    },
    {
      "epoch": 0.16599937475246734,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010810301030103012,
      "loss": 3.0544,
      "step": 459540
    },
    {
      "epoch": 0.16601801798862553,
      "grad_norm": 0.75,
      "learning_rate": 0.0001081010101010101,
      "loss": 3.0423,
      "step": 459550
    },
    {
      "epoch": 0.16603666122478372,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010809900990099011,
      "loss": 3.1029,
      "step": 459560
    },
    {
      "epoch": 0.16605530446094188,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001080970097009701,
      "loss": 2.9765,
      "step": 459570
    },
    {
      "epoch": 0.16607394769710007,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001080950095009501,
      "loss": 3.121,
      "step": 459580
    },
    {
      "epoch": 0.16609259093325826,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001080930093009301,
      "loss": 3.0953,
      "step": 459590
    },
    {
      "epoch": 0.16611123416941645,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001080910091009101,
      "loss": 3.0965,
      "step": 459600
    },
    {
      "epoch": 0.16612987740557464,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010808900890089009,
      "loss": 3.2388,
      "step": 459610
    },
    {
      "epoch": 0.16614852064173283,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001080870087008701,
      "loss": 2.9484,
      "step": 459620
    },
    {
      "epoch": 0.166167163877891,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010808500850085008,
      "loss": 3.1662,
      "step": 459630
    },
    {
      "epoch": 0.16618580711404918,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010808300830083009,
      "loss": 3.0503,
      "step": 459640
    },
    {
      "epoch": 0.16620445035020737,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010808100810081007,
      "loss": 3.1589,
      "step": 459650
    },
    {
      "epoch": 0.16622309358636556,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010807900790079008,
      "loss": 3.0814,
      "step": 459660
    },
    {
      "epoch": 0.16624173682252374,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010807700770077008,
      "loss": 2.9712,
      "step": 459670
    },
    {
      "epoch": 0.1662603800586819,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010807500750075009,
      "loss": 3.1417,
      "step": 459680
    },
    {
      "epoch": 0.1662790232948401,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001080730073007301,
      "loss": 3.1251,
      "step": 459690
    },
    {
      "epoch": 0.16629766653099828,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010807100710071008,
      "loss": 3.1312,
      "step": 459700
    },
    {
      "epoch": 0.16631630976715647,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010806900690069009,
      "loss": 3.0809,
      "step": 459710
    },
    {
      "epoch": 0.16633495300331466,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010806700670067007,
      "loss": 3.1163,
      "step": 459720
    },
    {
      "epoch": 0.16635359623947282,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010806500650065008,
      "loss": 3.1077,
      "step": 459730
    },
    {
      "epoch": 0.166372239475631,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010806300630063006,
      "loss": 3.0498,
      "step": 459740
    },
    {
      "epoch": 0.1663908827117892,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010806100610061007,
      "loss": 3.1153,
      "step": 459750
    },
    {
      "epoch": 0.1664095259479474,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010805900590059006,
      "loss": 2.9021,
      "step": 459760
    },
    {
      "epoch": 0.16642816918410558,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010805700570057007,
      "loss": 3.1486,
      "step": 459770
    },
    {
      "epoch": 0.16644681242026374,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010805500550055006,
      "loss": 2.9773,
      "step": 459780
    },
    {
      "epoch": 0.16646545565642193,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00010805300530053007,
      "loss": 3.1938,
      "step": 459790
    },
    {
      "epoch": 0.16648409889258012,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010805100510051005,
      "loss": 3.0428,
      "step": 459800
    },
    {
      "epoch": 0.1665027421287383,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010804900490049006,
      "loss": 3.0066,
      "step": 459810
    },
    {
      "epoch": 0.1665213853648965,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010804700470047004,
      "loss": 2.9345,
      "step": 459820
    },
    {
      "epoch": 0.16654002860105468,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010804500450045005,
      "loss": 3.004,
      "step": 459830
    },
    {
      "epoch": 0.16655867183721285,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010804300430043004,
      "loss": 3.0549,
      "step": 459840
    },
    {
      "epoch": 0.16657731507337104,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010804100410041005,
      "loss": 3.0603,
      "step": 459850
    },
    {
      "epoch": 0.16659595830952922,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010803900390039003,
      "loss": 3.113,
      "step": 459860
    },
    {
      "epoch": 0.1666146015456874,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010803700370037004,
      "loss": 3.0808,
      "step": 459870
    },
    {
      "epoch": 0.1666332447818456,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010803500350035003,
      "loss": 3.1624,
      "step": 459880
    },
    {
      "epoch": 0.16665188801800376,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010803300330033003,
      "loss": 3.1822,
      "step": 459890
    },
    {
      "epoch": 0.16667053125416195,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010803100310031004,
      "loss": 3.0585,
      "step": 459900
    },
    {
      "epoch": 0.16668917449032014,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010802900290029004,
      "loss": 3.1449,
      "step": 459910
    },
    {
      "epoch": 0.16670781772647833,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010802700270027004,
      "loss": 3.0545,
      "step": 459920
    },
    {
      "epoch": 0.16672646096263652,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010802500250025003,
      "loss": 3.0596,
      "step": 459930
    },
    {
      "epoch": 0.16674510419879468,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010802300230023004,
      "loss": 3.1347,
      "step": 459940
    },
    {
      "epoch": 0.16676374743495287,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010802100210021002,
      "loss": 2.9871,
      "step": 459950
    },
    {
      "epoch": 0.16678239067111106,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010801900190019003,
      "loss": 2.9889,
      "step": 459960
    },
    {
      "epoch": 0.16680103390726925,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010801700170017001,
      "loss": 3.0483,
      "step": 459970
    },
    {
      "epoch": 0.16681967714342744,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010801500150015002,
      "loss": 3.1653,
      "step": 459980
    },
    {
      "epoch": 0.1668383203795856,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010801300130013002,
      "loss": 3.1322,
      "step": 459990
    },
    {
      "epoch": 0.1668569636157438,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010801100110011003,
      "loss": 3.0114,
      "step": 460000
    },
    {
      "epoch": 0.16687560685190198,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010800900090009001,
      "loss": 2.9909,
      "step": 460010
    },
    {
      "epoch": 0.16689425008806016,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010800700070007002,
      "loss": 2.9997,
      "step": 460020
    },
    {
      "epoch": 0.16691289332421835,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010800500050005,
      "loss": 3.0503,
      "step": 460030
    },
    {
      "epoch": 0.16693153656037651,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010800300030003001,
      "loss": 3.0628,
      "step": 460040
    },
    {
      "epoch": 0.1669501797965347,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010800100010000999,
      "loss": 3.0639,
      "step": 460050
    },
    {
      "epoch": 0.1669688230326929,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010799899989999,
      "loss": 3.0769,
      "step": 460060
    },
    {
      "epoch": 0.16698746626885108,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010799699969997,
      "loss": 3.0677,
      "step": 460070
    },
    {
      "epoch": 0.16700610950500927,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010799499949995,
      "loss": 3.1696,
      "step": 460080
    },
    {
      "epoch": 0.16702475274116746,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010799299929993001,
      "loss": 3.09,
      "step": 460090
    },
    {
      "epoch": 0.16704339597732562,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010799099909991,
      "loss": 3.0924,
      "step": 460100
    },
    {
      "epoch": 0.1670620392134838,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010798899889989,
      "loss": 3.0645,
      "step": 460110
    },
    {
      "epoch": 0.167080682449642,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010798699869986999,
      "loss": 3.1462,
      "step": 460120
    },
    {
      "epoch": 0.1670993256858002,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010798499849985,
      "loss": 3.1348,
      "step": 460130
    },
    {
      "epoch": 0.16711796892195838,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010798299829982998,
      "loss": 3.1046,
      "step": 460140
    },
    {
      "epoch": 0.16713661215811654,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010798099809980999,
      "loss": 3.2205,
      "step": 460150
    },
    {
      "epoch": 0.16715525539427473,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010797899789978998,
      "loss": 3.0515,
      "step": 460160
    },
    {
      "epoch": 0.16717389863043292,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010797699769976999,
      "loss": 3.1299,
      "step": 460170
    },
    {
      "epoch": 0.1671925418665911,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010797499749974997,
      "loss": 3.1056,
      "step": 460180
    },
    {
      "epoch": 0.1672111851027493,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010797299729972998,
      "loss": 3.0846,
      "step": 460190
    },
    {
      "epoch": 0.16722982833890745,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010797099709970997,
      "loss": 3.0319,
      "step": 460200
    },
    {
      "epoch": 0.16724847157506564,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010796899689968998,
      "loss": 3.3136,
      "step": 460210
    },
    {
      "epoch": 0.16726711481122383,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010796699669966997,
      "loss": 3.0609,
      "step": 460220
    },
    {
      "epoch": 0.16728575804738202,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010796499649964997,
      "loss": 3.1211,
      "step": 460230
    },
    {
      "epoch": 0.1673044012835402,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010796299629962996,
      "loss": 3.0227,
      "step": 460240
    },
    {
      "epoch": 0.16732304451969837,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010796099609960997,
      "loss": 3.0556,
      "step": 460250
    },
    {
      "epoch": 0.16734168775585656,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010795899589958995,
      "loss": 3.1314,
      "step": 460260
    },
    {
      "epoch": 0.16736033099201475,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010795699569956996,
      "loss": 3.0955,
      "step": 460270
    },
    {
      "epoch": 0.16737897422817294,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010795499549954994,
      "loss": 3.0221,
      "step": 460280
    },
    {
      "epoch": 0.16739761746433113,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010795299529952995,
      "loss": 3.0823,
      "step": 460290
    },
    {
      "epoch": 0.16741626070048932,
      "grad_norm": 0.75,
      "learning_rate": 0.00010795099509950996,
      "loss": 3.1768,
      "step": 460300
    },
    {
      "epoch": 0.16743490393664748,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010794899489948996,
      "loss": 3.0913,
      "step": 460310
    },
    {
      "epoch": 0.16745354717280567,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010794699469946997,
      "loss": 3.1355,
      "step": 460320
    },
    {
      "epoch": 0.16747219040896386,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010794499449944995,
      "loss": 3.1404,
      "step": 460330
    },
    {
      "epoch": 0.16749083364512204,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010794299429942996,
      "loss": 3.0897,
      "step": 460340
    },
    {
      "epoch": 0.16750947688128023,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010794099409940994,
      "loss": 3.1753,
      "step": 460350
    },
    {
      "epoch": 0.1675281201174384,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010793899389938995,
      "loss": 3.0541,
      "step": 460360
    },
    {
      "epoch": 0.16754676335359658,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010793699369936993,
      "loss": 2.9858,
      "step": 460370
    },
    {
      "epoch": 0.16756540658975477,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010793499349934994,
      "loss": 2.974,
      "step": 460380
    },
    {
      "epoch": 0.16758404982591296,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010793299329932994,
      "loss": 3.0705,
      "step": 460390
    },
    {
      "epoch": 0.16760269306207115,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010793099309930995,
      "loss": 3.0084,
      "step": 460400
    },
    {
      "epoch": 0.1676213362982293,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010792899289928993,
      "loss": 3.2144,
      "step": 460410
    },
    {
      "epoch": 0.1676399795343875,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010792699269926994,
      "loss": 2.983,
      "step": 460420
    },
    {
      "epoch": 0.1676586227705457,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010792499249924992,
      "loss": 2.9598,
      "step": 460430
    },
    {
      "epoch": 0.16767726600670388,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010792299229922993,
      "loss": 3.1925,
      "step": 460440
    },
    {
      "epoch": 0.16769590924286207,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010792099209920991,
      "loss": 3.0553,
      "step": 460450
    },
    {
      "epoch": 0.16771455247902023,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010791899189918992,
      "loss": 3.1326,
      "step": 460460
    },
    {
      "epoch": 0.16773319571517842,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010791699169916991,
      "loss": 3.0972,
      "step": 460470
    },
    {
      "epoch": 0.1677518389513366,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010791499149914992,
      "loss": 3.0313,
      "step": 460480
    },
    {
      "epoch": 0.1677704821874948,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010791299129912993,
      "loss": 3.1727,
      "step": 460490
    },
    {
      "epoch": 0.16778912542365298,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010791099109910992,
      "loss": 3.2121,
      "step": 460500
    },
    {
      "epoch": 0.16780776865981117,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010790899089908992,
      "loss": 3.1832,
      "step": 460510
    },
    {
      "epoch": 0.16782641189596934,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001079069906990699,
      "loss": 3.1576,
      "step": 460520
    },
    {
      "epoch": 0.16784505513212752,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010790499049904992,
      "loss": 2.9573,
      "step": 460530
    },
    {
      "epoch": 0.1678636983682857,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010790299029902991,
      "loss": 3.1344,
      "step": 460540
    },
    {
      "epoch": 0.1678823416044439,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001079009900990099,
      "loss": 3.1443,
      "step": 460550
    },
    {
      "epoch": 0.1679009848406021,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001078989898989899,
      "loss": 3.0403,
      "step": 460560
    },
    {
      "epoch": 0.16791962807676025,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010789698969896991,
      "loss": 3.0517,
      "step": 460570
    },
    {
      "epoch": 0.16793827131291844,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001078949894989499,
      "loss": 3.0128,
      "step": 460580
    },
    {
      "epoch": 0.16795691454907663,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001078929892989299,
      "loss": 3.2197,
      "step": 460590
    },
    {
      "epoch": 0.16797555778523482,
      "grad_norm": 0.625,
      "learning_rate": 0.00010789098909890988,
      "loss": 3.1128,
      "step": 460600
    },
    {
      "epoch": 0.167994201021393,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001078889888988899,
      "loss": 3.0272,
      "step": 460610
    },
    {
      "epoch": 0.16801284425755117,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010788698869886989,
      "loss": 3.1536,
      "step": 460620
    },
    {
      "epoch": 0.16803148749370936,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001078849884988499,
      "loss": 3.1017,
      "step": 460630
    },
    {
      "epoch": 0.16805013072986755,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010788298829882988,
      "loss": 3.166,
      "step": 460640
    },
    {
      "epoch": 0.16806877396602574,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010788098809880989,
      "loss": 3.0941,
      "step": 460650
    },
    {
      "epoch": 0.16808741720218393,
      "grad_norm": 0.75,
      "learning_rate": 0.00010787898789878987,
      "loss": 2.9917,
      "step": 460660
    },
    {
      "epoch": 0.1681060604383421,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010787698769876988,
      "loss": 3.0388,
      "step": 460670
    },
    {
      "epoch": 0.16812470367450028,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010787498749874989,
      "loss": 3.0845,
      "step": 460680
    },
    {
      "epoch": 0.16814334691065846,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010787298729872987,
      "loss": 3.1693,
      "step": 460690
    },
    {
      "epoch": 0.16816199014681665,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010787098709870988,
      "loss": 2.9722,
      "step": 460700
    },
    {
      "epoch": 0.16818063338297484,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010786898689868988,
      "loss": 3.037,
      "step": 460710
    },
    {
      "epoch": 0.168199276619133,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010786698669866989,
      "loss": 3.2407,
      "step": 460720
    },
    {
      "epoch": 0.1682179198552912,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010786498649864987,
      "loss": 3.0204,
      "step": 460730
    },
    {
      "epoch": 0.16823656309144938,
      "grad_norm": 0.75,
      "learning_rate": 0.00010786298629862988,
      "loss": 3.1059,
      "step": 460740
    },
    {
      "epoch": 0.16825520632760757,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010786098609860986,
      "loss": 3.0249,
      "step": 460750
    },
    {
      "epoch": 0.16827384956376576,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010785898589858987,
      "loss": 3.0687,
      "step": 460760
    },
    {
      "epoch": 0.16829249279992395,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010785698569856985,
      "loss": 3.0142,
      "step": 460770
    },
    {
      "epoch": 0.1683111360360821,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010785498549854986,
      "loss": 2.9574,
      "step": 460780
    },
    {
      "epoch": 0.1683297792722403,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010785298529852986,
      "loss": 3.1541,
      "step": 460790
    },
    {
      "epoch": 0.1683484225083985,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010785098509850986,
      "loss": 3.0873,
      "step": 460800
    },
    {
      "epoch": 0.16836706574455668,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010784898489848985,
      "loss": 3.2224,
      "step": 460810
    },
    {
      "epoch": 0.16838570898071487,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010784698469846986,
      "loss": 3.0477,
      "step": 460820
    },
    {
      "epoch": 0.16840435221687303,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010784498449844984,
      "loss": 3.1429,
      "step": 460830
    },
    {
      "epoch": 0.16842299545303122,
      "grad_norm": 0.75,
      "learning_rate": 0.00010784298429842985,
      "loss": 3.0752,
      "step": 460840
    },
    {
      "epoch": 0.1684416386891894,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010784098409840984,
      "loss": 2.9943,
      "step": 460850
    },
    {
      "epoch": 0.1684602819253476,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010783898389838985,
      "loss": 3.1201,
      "step": 460860
    },
    {
      "epoch": 0.16847892516150578,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010783698369836983,
      "loss": 2.9656,
      "step": 460870
    },
    {
      "epoch": 0.16849756839766394,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010783498349834984,
      "loss": 3.0932,
      "step": 460880
    },
    {
      "epoch": 0.16851621163382213,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010783298329832985,
      "loss": 3.1913,
      "step": 460890
    },
    {
      "epoch": 0.16853485486998032,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010783098309830983,
      "loss": 3.0719,
      "step": 460900
    },
    {
      "epoch": 0.1685534981061385,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010782898289828984,
      "loss": 3.008,
      "step": 460910
    },
    {
      "epoch": 0.1685721413422967,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010782698269826982,
      "loss": 3.1432,
      "step": 460920
    },
    {
      "epoch": 0.16859078457845486,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010782498249824983,
      "loss": 3.0588,
      "step": 460930
    },
    {
      "epoch": 0.16860942781461305,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010782298229822983,
      "loss": 3.2227,
      "step": 460940
    },
    {
      "epoch": 0.16862807105077124,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010782098209820984,
      "loss": 3.2389,
      "step": 460950
    },
    {
      "epoch": 0.16864671428692943,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010781898189818982,
      "loss": 3.1807,
      "step": 460960
    },
    {
      "epoch": 0.16866535752308762,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010781698169816983,
      "loss": 3.1517,
      "step": 460970
    },
    {
      "epoch": 0.1686840007592458,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010781498149814981,
      "loss": 3.0895,
      "step": 460980
    },
    {
      "epoch": 0.16870264399540397,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010781298129812982,
      "loss": 3.0084,
      "step": 460990
    },
    {
      "epoch": 0.16872128723156216,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001078109810981098,
      "loss": 3.0173,
      "step": 461000
    },
    {
      "epoch": 0.16873993046772034,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010780898089808981,
      "loss": 3.0661,
      "step": 461010
    },
    {
      "epoch": 0.16875857370387853,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010780698069806981,
      "loss": 3.0656,
      "step": 461020
    },
    {
      "epoch": 0.16877721694003672,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010780498049804982,
      "loss": 3.0839,
      "step": 461030
    },
    {
      "epoch": 0.16879586017619488,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001078029802980298,
      "loss": 3.0903,
      "step": 461040
    },
    {
      "epoch": 0.16881450341235307,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010780098009800981,
      "loss": 3.0838,
      "step": 461050
    },
    {
      "epoch": 0.16883314664851126,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010779897989798979,
      "loss": 3.1163,
      "step": 461060
    },
    {
      "epoch": 0.16885178988466945,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001077969796979698,
      "loss": 3.0958,
      "step": 461070
    },
    {
      "epoch": 0.16887043312082764,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010779497949794981,
      "loss": 3.0677,
      "step": 461080
    },
    {
      "epoch": 0.1688890763569858,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010779297929792979,
      "loss": 3.0765,
      "step": 461090
    },
    {
      "epoch": 0.168907719593144,
      "grad_norm": 0.75,
      "learning_rate": 0.0001077909790979098,
      "loss": 3.0703,
      "step": 461100
    },
    {
      "epoch": 0.16892636282930218,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001077889788978898,
      "loss": 3.1699,
      "step": 461110
    },
    {
      "epoch": 0.16894500606546037,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001077869786978698,
      "loss": 3.2603,
      "step": 461120
    },
    {
      "epoch": 0.16896364930161856,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010778497849784979,
      "loss": 3.0637,
      "step": 461130
    },
    {
      "epoch": 0.16898229253777672,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001077829782978298,
      "loss": 3.1443,
      "step": 461140
    },
    {
      "epoch": 0.1690009357739349,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010778097809780978,
      "loss": 3.0643,
      "step": 461150
    },
    {
      "epoch": 0.1690195790100931,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010777897789778979,
      "loss": 3.0322,
      "step": 461160
    },
    {
      "epoch": 0.16903822224625129,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010777697769776978,
      "loss": 3.0109,
      "step": 461170
    },
    {
      "epoch": 0.16905686548240947,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010777497749774979,
      "loss": 3.0608,
      "step": 461180
    },
    {
      "epoch": 0.16907550871856766,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010777297729772977,
      "loss": 3.0974,
      "step": 461190
    },
    {
      "epoch": 0.16909415195472582,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010777097709770978,
      "loss": 2.9839,
      "step": 461200
    },
    {
      "epoch": 0.169112795190884,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010776897689768976,
      "loss": 2.9112,
      "step": 461210
    },
    {
      "epoch": 0.1691314384270422,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010776697669766977,
      "loss": 2.8972,
      "step": 461220
    },
    {
      "epoch": 0.1691500816632004,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010776497649764976,
      "loss": 2.9766,
      "step": 461230
    },
    {
      "epoch": 0.16916872489935858,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010776297629762977,
      "loss": 2.9986,
      "step": 461240
    },
    {
      "epoch": 0.16918736813551674,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010776097609760976,
      "loss": 3.0635,
      "step": 461250
    },
    {
      "epoch": 0.16920601137167493,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010775897589758977,
      "loss": 3.0165,
      "step": 461260
    },
    {
      "epoch": 0.16922465460783312,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010775697569756978,
      "loss": 3.0653,
      "step": 461270
    },
    {
      "epoch": 0.1692432978439913,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010775497549754976,
      "loss": 3.06,
      "step": 461280
    },
    {
      "epoch": 0.1692619410801495,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010775297529752977,
      "loss": 3.1706,
      "step": 461290
    },
    {
      "epoch": 0.16928058431630766,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010775097509750975,
      "loss": 3.0627,
      "step": 461300
    },
    {
      "epoch": 0.16929922755246585,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010774897489748976,
      "loss": 3.1971,
      "step": 461310
    },
    {
      "epoch": 0.16931787078862404,
      "grad_norm": 0.75,
      "learning_rate": 0.00010774697469746974,
      "loss": 3.2131,
      "step": 461320
    },
    {
      "epoch": 0.16933651402478223,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010774497449744975,
      "loss": 3.1039,
      "step": 461330
    },
    {
      "epoch": 0.16935515726094041,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010774297429742975,
      "loss": 2.9161,
      "step": 461340
    },
    {
      "epoch": 0.16937380049709858,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010774097409740976,
      "loss": 3.1871,
      "step": 461350
    },
    {
      "epoch": 0.16939244373325676,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010773897389738974,
      "loss": 3.0984,
      "step": 461360
    },
    {
      "epoch": 0.16941108696941495,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010773697369736975,
      "loss": 2.8797,
      "step": 461370
    },
    {
      "epoch": 0.16942973020557314,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010773497349734973,
      "loss": 3.142,
      "step": 461380
    },
    {
      "epoch": 0.16944837344173133,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010773297329732974,
      "loss": 3.0201,
      "step": 461390
    },
    {
      "epoch": 0.1694670166778895,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010773097309730974,
      "loss": 3.0,
      "step": 461400
    },
    {
      "epoch": 0.16948565991404768,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010772897289728973,
      "loss": 3.0618,
      "step": 461410
    },
    {
      "epoch": 0.16950430315020587,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010772697269726973,
      "loss": 3.0707,
      "step": 461420
    },
    {
      "epoch": 0.16952294638636406,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010772497249724974,
      "loss": 2.9498,
      "step": 461430
    },
    {
      "epoch": 0.16954158962252225,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010772297229722972,
      "loss": 3.0345,
      "step": 461440
    },
    {
      "epoch": 0.16956023285868044,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010772097209720973,
      "loss": 3.1448,
      "step": 461450
    },
    {
      "epoch": 0.1695788760948386,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010771897189718971,
      "loss": 3.1632,
      "step": 461460
    },
    {
      "epoch": 0.1695975193309968,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010771697169716972,
      "loss": 2.9174,
      "step": 461470
    },
    {
      "epoch": 0.16961616256715498,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010771497149714973,
      "loss": 3.0527,
      "step": 461480
    },
    {
      "epoch": 0.16963480580331317,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010771297129712972,
      "loss": 2.9394,
      "step": 461490
    },
    {
      "epoch": 0.16965344903947135,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010771097109710972,
      "loss": 3.1467,
      "step": 461500
    },
    {
      "epoch": 0.16967209227562952,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010770897089708971,
      "loss": 2.9372,
      "step": 461510
    },
    {
      "epoch": 0.1696907355117877,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010770697069706972,
      "loss": 3.0582,
      "step": 461520
    },
    {
      "epoch": 0.1697093787479459,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001077049704970497,
      "loss": 3.0377,
      "step": 461530
    },
    {
      "epoch": 0.16972802198410408,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010770297029702971,
      "loss": 3.0831,
      "step": 461540
    },
    {
      "epoch": 0.16974666522026227,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001077009700970097,
      "loss": 3.0379,
      "step": 461550
    },
    {
      "epoch": 0.16976530845642043,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001076989698969897,
      "loss": 3.1969,
      "step": 461560
    },
    {
      "epoch": 0.16978395169257862,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001076969696969697,
      "loss": 3.0958,
      "step": 461570
    },
    {
      "epoch": 0.1698025949287368,
      "grad_norm": 0.75,
      "learning_rate": 0.00010769496949694971,
      "loss": 3.1316,
      "step": 461580
    },
    {
      "epoch": 0.169821238164895,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010769296929692969,
      "loss": 2.9478,
      "step": 461590
    },
    {
      "epoch": 0.1698398814010532,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001076909690969097,
      "loss": 3.0186,
      "step": 461600
    },
    {
      "epoch": 0.16985852463721135,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010768896889688968,
      "loss": 2.9724,
      "step": 461610
    },
    {
      "epoch": 0.16987716787336954,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010768696869686969,
      "loss": 3.0299,
      "step": 461620
    },
    {
      "epoch": 0.16989581110952773,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010768496849684967,
      "loss": 3.0817,
      "step": 461630
    },
    {
      "epoch": 0.16991445434568592,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010768296829682968,
      "loss": 3.0665,
      "step": 461640
    },
    {
      "epoch": 0.1699330975818441,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010768096809680968,
      "loss": 3.0792,
      "step": 461650
    },
    {
      "epoch": 0.1699517408180023,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010767896789678969,
      "loss": 3.0579,
      "step": 461660
    },
    {
      "epoch": 0.16997038405416046,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001076769676967697,
      "loss": 3.1367,
      "step": 461670
    },
    {
      "epoch": 0.16998902729031865,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010767496749674968,
      "loss": 2.969,
      "step": 461680
    },
    {
      "epoch": 0.17000767052647683,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010767296729672969,
      "loss": 3.0438,
      "step": 461690
    },
    {
      "epoch": 0.17002631376263502,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010767096709670967,
      "loss": 3.0807,
      "step": 461700
    },
    {
      "epoch": 0.1700449569987932,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010766896689668968,
      "loss": 3.2121,
      "step": 461710
    },
    {
      "epoch": 0.17006360023495137,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010766696669666966,
      "loss": 3.1497,
      "step": 461720
    },
    {
      "epoch": 0.17008224347110956,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010766496649664967,
      "loss": 3.0055,
      "step": 461730
    },
    {
      "epoch": 0.17010088670726775,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010766296629662967,
      "loss": 3.0795,
      "step": 461740
    },
    {
      "epoch": 0.17011952994342594,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010766096609660968,
      "loss": 2.9878,
      "step": 461750
    },
    {
      "epoch": 0.17013817317958413,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010765896589658966,
      "loss": 3.0725,
      "step": 461760
    },
    {
      "epoch": 0.1701568164157423,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010765696569656967,
      "loss": 2.896,
      "step": 461770
    },
    {
      "epoch": 0.17017545965190048,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010765496549654965,
      "loss": 3.0102,
      "step": 461780
    },
    {
      "epoch": 0.17019410288805867,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010765296529652966,
      "loss": 3.0282,
      "step": 461790
    },
    {
      "epoch": 0.17021274612421686,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010765096509650965,
      "loss": 3.079,
      "step": 461800
    },
    {
      "epoch": 0.17023138936037505,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010764896489648966,
      "loss": 3.0579,
      "step": 461810
    },
    {
      "epoch": 0.1702500325965332,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010764696469646965,
      "loss": 3.1382,
      "step": 461820
    },
    {
      "epoch": 0.1702686758326914,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010764496449644965,
      "loss": 3.1074,
      "step": 461830
    },
    {
      "epoch": 0.17028731906884959,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010764296429642964,
      "loss": 3.0953,
      "step": 461840
    },
    {
      "epoch": 0.17030596230500777,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010764096409640965,
      "loss": 3.0958,
      "step": 461850
    },
    {
      "epoch": 0.17032460554116596,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010763896389638965,
      "loss": 3.0733,
      "step": 461860
    },
    {
      "epoch": 0.17034324877732415,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010763696369636964,
      "loss": 3.0562,
      "step": 461870
    },
    {
      "epoch": 0.1703618920134823,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010763496349634965,
      "loss": 3.0351,
      "step": 461880
    },
    {
      "epoch": 0.1703805352496405,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010763296329632964,
      "loss": 3.0888,
      "step": 461890
    },
    {
      "epoch": 0.1703991784857987,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010763096309630965,
      "loss": 3.0908,
      "step": 461900
    },
    {
      "epoch": 0.17041782172195688,
      "grad_norm": 0.625,
      "learning_rate": 0.00010762896289628963,
      "loss": 3.1241,
      "step": 461910
    },
    {
      "epoch": 0.17043646495811507,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010762696269626964,
      "loss": 3.1038,
      "step": 461920
    },
    {
      "epoch": 0.17045510819427323,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010762496249624962,
      "loss": 3.0985,
      "step": 461930
    },
    {
      "epoch": 0.17047375143043142,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010762296229622963,
      "loss": 3.0361,
      "step": 461940
    },
    {
      "epoch": 0.1704923946665896,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010762096209620962,
      "loss": 3.1885,
      "step": 461950
    },
    {
      "epoch": 0.1705110379027478,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010761896189618962,
      "loss": 3.0282,
      "step": 461960
    },
    {
      "epoch": 0.170529681138906,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010761696169616962,
      "loss": 3.1353,
      "step": 461970
    },
    {
      "epoch": 0.17054832437506415,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010761496149614963,
      "loss": 2.9886,
      "step": 461980
    },
    {
      "epoch": 0.17056696761122234,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010761296129612961,
      "loss": 2.8857,
      "step": 461990
    },
    {
      "epoch": 0.17058561084738053,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010761096109610962,
      "loss": 3.0604,
      "step": 462000
    },
    {
      "epoch": 0.17060425408353871,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001076089608960896,
      "loss": 2.958,
      "step": 462010
    },
    {
      "epoch": 0.1706228973196969,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010760696069606961,
      "loss": 3.1294,
      "step": 462020
    },
    {
      "epoch": 0.17064154055585506,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010760496049604961,
      "loss": 3.2382,
      "step": 462030
    },
    {
      "epoch": 0.17066018379201325,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001076029602960296,
      "loss": 3.0425,
      "step": 462040
    },
    {
      "epoch": 0.17067882702817144,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001076009600960096,
      "loss": 3.1656,
      "step": 462050
    },
    {
      "epoch": 0.17069747026432963,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010759895989598961,
      "loss": 3.0207,
      "step": 462060
    },
    {
      "epoch": 0.17071611350048782,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010759695969596962,
      "loss": 3.0657,
      "step": 462070
    },
    {
      "epoch": 0.17073475673664598,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001075949594959496,
      "loss": 3.15,
      "step": 462080
    },
    {
      "epoch": 0.17075339997280417,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010759295929592961,
      "loss": 2.9665,
      "step": 462090
    },
    {
      "epoch": 0.17077204320896236,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010759095909590959,
      "loss": 3.0371,
      "step": 462100
    },
    {
      "epoch": 0.17079068644512055,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001075889588958896,
      "loss": 3.1291,
      "step": 462110
    },
    {
      "epoch": 0.17080932968127874,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001075869586958696,
      "loss": 3.1054,
      "step": 462120
    },
    {
      "epoch": 0.17082797291743693,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001075849584958496,
      "loss": 3.1472,
      "step": 462130
    },
    {
      "epoch": 0.1708466161535951,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010758295829582959,
      "loss": 3.1721,
      "step": 462140
    },
    {
      "epoch": 0.17086525938975328,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001075809580958096,
      "loss": 2.9698,
      "step": 462150
    },
    {
      "epoch": 0.17088390262591147,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010757895789578958,
      "loss": 3.1397,
      "step": 462160
    },
    {
      "epoch": 0.17090254586206965,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010757695769576959,
      "loss": 3.1567,
      "step": 462170
    },
    {
      "epoch": 0.17092118909822784,
      "grad_norm": 0.75,
      "learning_rate": 0.00010757495749574957,
      "loss": 3.0787,
      "step": 462180
    },
    {
      "epoch": 0.170939832334386,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010757295729572958,
      "loss": 3.0288,
      "step": 462190
    },
    {
      "epoch": 0.1709584755705442,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010757095709570957,
      "loss": 2.9646,
      "step": 462200
    },
    {
      "epoch": 0.17097711880670238,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010756895689568958,
      "loss": 3.147,
      "step": 462210
    },
    {
      "epoch": 0.17099576204286057,
      "grad_norm": 0.75,
      "learning_rate": 0.00010756695669566956,
      "loss": 3.0704,
      "step": 462220
    },
    {
      "epoch": 0.17101440527901876,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010756495649564957,
      "loss": 3.1544,
      "step": 462230
    },
    {
      "epoch": 0.17103304851517692,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010756295629562956,
      "loss": 3.1405,
      "step": 462240
    },
    {
      "epoch": 0.1710516917513351,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010756095609560956,
      "loss": 3.0713,
      "step": 462250
    },
    {
      "epoch": 0.1710703349874933,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010755895589558957,
      "loss": 3.0803,
      "step": 462260
    },
    {
      "epoch": 0.1710889782236515,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010755695569556956,
      "loss": 3.0052,
      "step": 462270
    },
    {
      "epoch": 0.17110762145980968,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010755495549554956,
      "loss": 3.0665,
      "step": 462280
    },
    {
      "epoch": 0.17112626469596784,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010755295529552956,
      "loss": 3.1342,
      "step": 462290
    },
    {
      "epoch": 0.17114490793212603,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010755095509550957,
      "loss": 3.1581,
      "step": 462300
    },
    {
      "epoch": 0.17116355116828422,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010754895489548955,
      "loss": 3.0884,
      "step": 462310
    },
    {
      "epoch": 0.1711821944044424,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010754695469546956,
      "loss": 2.9542,
      "step": 462320
    },
    {
      "epoch": 0.1712008376406006,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010754495449544954,
      "loss": 3.0117,
      "step": 462330
    },
    {
      "epoch": 0.17121948087675878,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010754295429542955,
      "loss": 3.228,
      "step": 462340
    },
    {
      "epoch": 0.17123812411291695,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010754095409540955,
      "loss": 3.1425,
      "step": 462350
    },
    {
      "epoch": 0.17125676734907513,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010753895389538954,
      "loss": 3.1081,
      "step": 462360
    },
    {
      "epoch": 0.17127541058523332,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010753695369536954,
      "loss": 3.0758,
      "step": 462370
    },
    {
      "epoch": 0.1712940538213915,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010753495349534955,
      "loss": 3.104,
      "step": 462380
    },
    {
      "epoch": 0.1713126970575497,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010753295329532953,
      "loss": 3.1108,
      "step": 462390
    },
    {
      "epoch": 0.17133134029370786,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010753095309530954,
      "loss": 3.0071,
      "step": 462400
    },
    {
      "epoch": 0.17134998352986605,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010752895289528952,
      "loss": 3.1584,
      "step": 462410
    },
    {
      "epoch": 0.17136862676602424,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010752695269526953,
      "loss": 3.1394,
      "step": 462420
    },
    {
      "epoch": 0.17138727000218243,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010752495249524953,
      "loss": 3.116,
      "step": 462430
    },
    {
      "epoch": 0.17140591323834062,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010752295229522953,
      "loss": 3.1001,
      "step": 462440
    },
    {
      "epoch": 0.17142455647449878,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010752095209520952,
      "loss": 3.1252,
      "step": 462450
    },
    {
      "epoch": 0.17144319971065697,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010751895189518953,
      "loss": 3.1284,
      "step": 462460
    },
    {
      "epoch": 0.17146184294681516,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010751695169516954,
      "loss": 3.117,
      "step": 462470
    },
    {
      "epoch": 0.17148048618297335,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010751495149514952,
      "loss": 3.0552,
      "step": 462480
    },
    {
      "epoch": 0.17149912941913154,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010751295129512953,
      "loss": 2.9627,
      "step": 462490
    },
    {
      "epoch": 0.1715177726552897,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010751095109510951,
      "loss": 3.127,
      "step": 462500
    },
    {
      "epoch": 0.17153641589144789,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010750895089508952,
      "loss": 3.0354,
      "step": 462510
    },
    {
      "epoch": 0.17155505912760607,
      "grad_norm": 0.75,
      "learning_rate": 0.00010750695069506951,
      "loss": 3.063,
      "step": 462520
    },
    {
      "epoch": 0.17157370236376426,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010750495049504952,
      "loss": 3.046,
      "step": 462530
    },
    {
      "epoch": 0.17159234559992245,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001075029502950295,
      "loss": 3.0945,
      "step": 462540
    },
    {
      "epoch": 0.17161098883608064,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010750095009500951,
      "loss": 3.0707,
      "step": 462550
    },
    {
      "epoch": 0.1716296320722388,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001074989498949895,
      "loss": 2.9937,
      "step": 462560
    },
    {
      "epoch": 0.171648275308397,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001074969496949695,
      "loss": 3.1604,
      "step": 462570
    },
    {
      "epoch": 0.17166691854455518,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010749494949494949,
      "loss": 3.186,
      "step": 462580
    },
    {
      "epoch": 0.17168556178071337,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001074929492949295,
      "loss": 3.1386,
      "step": 462590
    },
    {
      "epoch": 0.17170420501687156,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010749094909490949,
      "loss": 3.0118,
      "step": 462600
    },
    {
      "epoch": 0.17172284825302972,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001074889488948895,
      "loss": 3.0791,
      "step": 462610
    },
    {
      "epoch": 0.1717414914891879,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010748694869486948,
      "loss": 3.1075,
      "step": 462620
    },
    {
      "epoch": 0.1717601347253461,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010748494849484949,
      "loss": 3.1028,
      "step": 462630
    },
    {
      "epoch": 0.1717787779615043,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010748294829482947,
      "loss": 3.0374,
      "step": 462640
    },
    {
      "epoch": 0.17179742119766248,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010748094809480948,
      "loss": 3.0712,
      "step": 462650
    },
    {
      "epoch": 0.17181606443382064,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010747894789478949,
      "loss": 3.1086,
      "step": 462660
    },
    {
      "epoch": 0.17183470766997883,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010747694769476949,
      "loss": 3.0258,
      "step": 462670
    },
    {
      "epoch": 0.17185335090613701,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010747494749474948,
      "loss": 3.0089,
      "step": 462680
    },
    {
      "epoch": 0.1718719941422952,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010747294729472948,
      "loss": 3.0685,
      "step": 462690
    },
    {
      "epoch": 0.1718906373784534,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010747094709470949,
      "loss": 3.1294,
      "step": 462700
    },
    {
      "epoch": 0.17190928061461155,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010746894689468947,
      "loss": 3.0588,
      "step": 462710
    },
    {
      "epoch": 0.17192792385076974,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010746694669466948,
      "loss": 3.1149,
      "step": 462720
    },
    {
      "epoch": 0.17194656708692793,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010746494649464946,
      "loss": 3.019,
      "step": 462730
    },
    {
      "epoch": 0.17196521032308612,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010746294629462947,
      "loss": 2.8851,
      "step": 462740
    },
    {
      "epoch": 0.1719838535592443,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010746094609460947,
      "loss": 3.1186,
      "step": 462750
    },
    {
      "epoch": 0.1720024967954025,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010745894589458948,
      "loss": 3.0869,
      "step": 462760
    },
    {
      "epoch": 0.17202114003156066,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010745694569456946,
      "loss": 3.1349,
      "step": 462770
    },
    {
      "epoch": 0.17203978326771885,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010745494549454947,
      "loss": 2.9969,
      "step": 462780
    },
    {
      "epoch": 0.17205842650387704,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010745294529452945,
      "loss": 3.0701,
      "step": 462790
    },
    {
      "epoch": 0.17207706974003523,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010745094509450946,
      "loss": 3.0879,
      "step": 462800
    },
    {
      "epoch": 0.17209571297619342,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010744894489448944,
      "loss": 3.0453,
      "step": 462810
    },
    {
      "epoch": 0.17211435621235158,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010744694469446945,
      "loss": 3.1514,
      "step": 462820
    },
    {
      "epoch": 0.17213299944850977,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010744494449444944,
      "loss": 3.1803,
      "step": 462830
    },
    {
      "epoch": 0.17215164268466795,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010744294429442945,
      "loss": 3.0595,
      "step": 462840
    },
    {
      "epoch": 0.17217028592082614,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010744094409440946,
      "loss": 3.0828,
      "step": 462850
    },
    {
      "epoch": 0.17218892915698433,
      "grad_norm": 0.953125,
      "learning_rate": 0.00010743894389438944,
      "loss": 3.0821,
      "step": 462860
    },
    {
      "epoch": 0.1722075723931425,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010743694369436945,
      "loss": 3.071,
      "step": 462870
    },
    {
      "epoch": 0.17222621562930068,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010743494349434944,
      "loss": 3.1719,
      "step": 462880
    },
    {
      "epoch": 0.17224485886545887,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010743294329432944,
      "loss": 3.1395,
      "step": 462890
    },
    {
      "epoch": 0.17226350210161706,
      "grad_norm": 0.75,
      "learning_rate": 0.00010743094309430943,
      "loss": 3.0061,
      "step": 462900
    },
    {
      "epoch": 0.17228214533777525,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010742894289428944,
      "loss": 3.0589,
      "step": 462910
    },
    {
      "epoch": 0.1723007885739334,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010742694269426943,
      "loss": 3.0066,
      "step": 462920
    },
    {
      "epoch": 0.1723194318100916,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010742494249424944,
      "loss": 3.1788,
      "step": 462930
    },
    {
      "epoch": 0.1723380750462498,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010742294229422942,
      "loss": 3.1051,
      "step": 462940
    },
    {
      "epoch": 0.17235671828240798,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010742094209420943,
      "loss": 2.9947,
      "step": 462950
    },
    {
      "epoch": 0.17237536151856617,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010741894189418941,
      "loss": 3.0247,
      "step": 462960
    },
    {
      "epoch": 0.17239400475472433,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010741694169416942,
      "loss": 3.0874,
      "step": 462970
    },
    {
      "epoch": 0.17241264799088252,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010741494149414942,
      "loss": 2.9962,
      "step": 462980
    },
    {
      "epoch": 0.1724312912270407,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010741294129412941,
      "loss": 3.0288,
      "step": 462990
    },
    {
      "epoch": 0.1724499344631989,
      "grad_norm": 0.75,
      "learning_rate": 0.00010741094109410941,
      "loss": 3.0061,
      "step": 463000
    },
    {
      "epoch": 0.17246857769935708,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010740894089408942,
      "loss": 3.0989,
      "step": 463010
    },
    {
      "epoch": 0.17248722093551527,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001074069406940694,
      "loss": 3.1082,
      "step": 463020
    },
    {
      "epoch": 0.17250586417167343,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010740494049404941,
      "loss": 3.1222,
      "step": 463030
    },
    {
      "epoch": 0.17252450740783162,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010740294029402939,
      "loss": 3.0956,
      "step": 463040
    },
    {
      "epoch": 0.1725431506439898,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001074009400940094,
      "loss": 3.0701,
      "step": 463050
    },
    {
      "epoch": 0.172561793880148,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010739893989398941,
      "loss": 3.1719,
      "step": 463060
    },
    {
      "epoch": 0.1725804371163062,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001073969396939694,
      "loss": 3.2015,
      "step": 463070
    },
    {
      "epoch": 0.17259908035246435,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010739493949394942,
      "loss": 3.1538,
      "step": 463080
    },
    {
      "epoch": 0.17261772358862254,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001073929392939294,
      "loss": 2.9699,
      "step": 463090
    },
    {
      "epoch": 0.17263636682478073,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001073909390939094,
      "loss": 3.0949,
      "step": 463100
    },
    {
      "epoch": 0.17265501006093892,
      "grad_norm": 0.875,
      "learning_rate": 0.00010738893889388939,
      "loss": 3.0162,
      "step": 463110
    },
    {
      "epoch": 0.1726736532970971,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001073869386938694,
      "loss": 3.1484,
      "step": 463120
    },
    {
      "epoch": 0.17269229653325527,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010738493849384938,
      "loss": 2.991,
      "step": 463130
    },
    {
      "epoch": 0.17271093976941346,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010738293829382939,
      "loss": 3.2052,
      "step": 463140
    },
    {
      "epoch": 0.17272958300557165,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010738093809380938,
      "loss": 3.27,
      "step": 463150
    },
    {
      "epoch": 0.17274822624172984,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001073789378937894,
      "loss": 3.162,
      "step": 463160
    },
    {
      "epoch": 0.17276686947788802,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010737693769376938,
      "loss": 2.9593,
      "step": 463170
    },
    {
      "epoch": 0.17278551271404619,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010737493749374938,
      "loss": 3.2055,
      "step": 463180
    },
    {
      "epoch": 0.17280415595020437,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010737293729372937,
      "loss": 3.1187,
      "step": 463190
    },
    {
      "epoch": 0.17282279918636256,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010737093709370938,
      "loss": 3.0959,
      "step": 463200
    },
    {
      "epoch": 0.17284144242252075,
      "grad_norm": 0.75,
      "learning_rate": 0.00010736893689368936,
      "loss": 3.0789,
      "step": 463210
    },
    {
      "epoch": 0.17286008565867894,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010736693669366937,
      "loss": 3.0962,
      "step": 463220
    },
    {
      "epoch": 0.17287872889483713,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010736493649364936,
      "loss": 3.0303,
      "step": 463230
    },
    {
      "epoch": 0.1728973721309953,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010736293629362937,
      "loss": 3.0019,
      "step": 463240
    },
    {
      "epoch": 0.17291601536715348,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010736093609360938,
      "loss": 3.0704,
      "step": 463250
    },
    {
      "epoch": 0.17293465860331167,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010735893589358936,
      "loss": 3.0768,
      "step": 463260
    },
    {
      "epoch": 0.17295330183946986,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010735693569356937,
      "loss": 3.176,
      "step": 463270
    },
    {
      "epoch": 0.17297194507562805,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010735493549354935,
      "loss": 3.1504,
      "step": 463280
    },
    {
      "epoch": 0.1729905883117862,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010735293529352936,
      "loss": 2.9801,
      "step": 463290
    },
    {
      "epoch": 0.1730092315479444,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010735093509350936,
      "loss": 2.9451,
      "step": 463300
    },
    {
      "epoch": 0.1730278747841026,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010734893489348935,
      "loss": 3.1454,
      "step": 463310
    },
    {
      "epoch": 0.17304651802026078,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010734693469346935,
      "loss": 3.0787,
      "step": 463320
    },
    {
      "epoch": 0.17306516125641896,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010734493449344936,
      "loss": 3.0531,
      "step": 463330
    },
    {
      "epoch": 0.17308380449257713,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010734293429342934,
      "loss": 3.0229,
      "step": 463340
    },
    {
      "epoch": 0.17310244772873531,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010734093409340935,
      "loss": 3.1684,
      "step": 463350
    },
    {
      "epoch": 0.1731210909648935,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010733893389338933,
      "loss": 3.0822,
      "step": 463360
    },
    {
      "epoch": 0.1731397342010517,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010733693369336934,
      "loss": 3.1643,
      "step": 463370
    },
    {
      "epoch": 0.17315837743720988,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010733493349334934,
      "loss": 3.0144,
      "step": 463380
    },
    {
      "epoch": 0.17317702067336804,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010733293329332935,
      "loss": 3.1316,
      "step": 463390
    },
    {
      "epoch": 0.17319566390952623,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010733093309330933,
      "loss": 2.9967,
      "step": 463400
    },
    {
      "epoch": 0.17321430714568442,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010732893289328934,
      "loss": 3.1394,
      "step": 463410
    },
    {
      "epoch": 0.1732329503818426,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010732693269326932,
      "loss": 3.146,
      "step": 463420
    },
    {
      "epoch": 0.1732515936180008,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010732493249324933,
      "loss": 3.1539,
      "step": 463430
    },
    {
      "epoch": 0.173270236854159,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010732293229322934,
      "loss": 3.1053,
      "step": 463440
    },
    {
      "epoch": 0.17328888009031715,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010732093209320932,
      "loss": 3.11,
      "step": 463450
    },
    {
      "epoch": 0.17330752332647534,
      "grad_norm": 0.75,
      "learning_rate": 0.00010731893189318933,
      "loss": 3.1458,
      "step": 463460
    },
    {
      "epoch": 0.17332616656263353,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010731693169316932,
      "loss": 3.0757,
      "step": 463470
    },
    {
      "epoch": 0.17334480979879172,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010731493149314933,
      "loss": 3.0215,
      "step": 463480
    },
    {
      "epoch": 0.1733634530349499,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010731293129312932,
      "loss": 3.1465,
      "step": 463490
    },
    {
      "epoch": 0.17338209627110807,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010731093109310933,
      "loss": 3.1538,
      "step": 463500
    },
    {
      "epoch": 0.17340073950726625,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010730893089308931,
      "loss": 2.9667,
      "step": 463510
    },
    {
      "epoch": 0.17341938274342444,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010730693069306932,
      "loss": 3.1143,
      "step": 463520
    },
    {
      "epoch": 0.17343802597958263,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001073049304930493,
      "loss": 3.0184,
      "step": 463530
    },
    {
      "epoch": 0.17345666921574082,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010730293029302931,
      "loss": 3.0928,
      "step": 463540
    },
    {
      "epoch": 0.17347531245189898,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001073009300930093,
      "loss": 3.1543,
      "step": 463550
    },
    {
      "epoch": 0.17349395568805717,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010729892989298931,
      "loss": 3.0831,
      "step": 463560
    },
    {
      "epoch": 0.17351259892421536,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001072969296929693,
      "loss": 3.0349,
      "step": 463570
    },
    {
      "epoch": 0.17353124216037355,
      "grad_norm": 0.75,
      "learning_rate": 0.0001072949294929493,
      "loss": 3.251,
      "step": 463580
    },
    {
      "epoch": 0.17354988539653174,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010729292929292929,
      "loss": 3.0008,
      "step": 463590
    },
    {
      "epoch": 0.1735685286326899,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001072909290929093,
      "loss": 2.9695,
      "step": 463600
    },
    {
      "epoch": 0.1735871718688481,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010728892889288929,
      "loss": 3.0031,
      "step": 463610
    },
    {
      "epoch": 0.17360581510500628,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001072869286928693,
      "loss": 3.098,
      "step": 463620
    },
    {
      "epoch": 0.17362445834116447,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010728492849284928,
      "loss": 3.1225,
      "step": 463630
    },
    {
      "epoch": 0.17364310157732266,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010728292829282929,
      "loss": 3.0652,
      "step": 463640
    },
    {
      "epoch": 0.17366174481348082,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001072809280928093,
      "loss": 3.0627,
      "step": 463650
    },
    {
      "epoch": 0.173680388049639,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010727892789278928,
      "loss": 3.2053,
      "step": 463660
    },
    {
      "epoch": 0.1736990312857972,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010727692769276929,
      "loss": 3.0862,
      "step": 463670
    },
    {
      "epoch": 0.17371767452195538,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010727492749274927,
      "loss": 3.0803,
      "step": 463680
    },
    {
      "epoch": 0.17373631775811357,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010727292729272928,
      "loss": 3.015,
      "step": 463690
    },
    {
      "epoch": 0.17375496099427176,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010727092709270928,
      "loss": 3.1306,
      "step": 463700
    },
    {
      "epoch": 0.17377360423042992,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010726892689268929,
      "loss": 3.0764,
      "step": 463710
    },
    {
      "epoch": 0.1737922474665881,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010726692669266927,
      "loss": 3.1736,
      "step": 463720
    },
    {
      "epoch": 0.1738108907027463,
      "grad_norm": 0.75,
      "learning_rate": 0.00010726492649264928,
      "loss": 3.0992,
      "step": 463730
    },
    {
      "epoch": 0.1738295339389045,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010726292629262926,
      "loss": 3.0888,
      "step": 463740
    },
    {
      "epoch": 0.17384817717506268,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010726092609260927,
      "loss": 3.1802,
      "step": 463750
    },
    {
      "epoch": 0.17386682041122084,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010725892589258925,
      "loss": 3.1853,
      "step": 463760
    },
    {
      "epoch": 0.17388546364737903,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010725692569256926,
      "loss": 3.1567,
      "step": 463770
    },
    {
      "epoch": 0.17390410688353722,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010725492549254926,
      "loss": 3.093,
      "step": 463780
    },
    {
      "epoch": 0.1739227501196954,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010725292529252927,
      "loss": 3.002,
      "step": 463790
    },
    {
      "epoch": 0.1739413933558536,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010725092509250925,
      "loss": 3.1122,
      "step": 463800
    },
    {
      "epoch": 0.17396003659201176,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010724892489248926,
      "loss": 3.0111,
      "step": 463810
    },
    {
      "epoch": 0.17397867982816995,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010724692469246924,
      "loss": 3.0946,
      "step": 463820
    },
    {
      "epoch": 0.17399732306432814,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010724492449244925,
      "loss": 3.1115,
      "step": 463830
    },
    {
      "epoch": 0.17401596630048632,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010724292429242926,
      "loss": 3.1494,
      "step": 463840
    },
    {
      "epoch": 0.1740346095366445,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010724092409240924,
      "loss": 3.024,
      "step": 463850
    },
    {
      "epoch": 0.17405325277280267,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010723892389238925,
      "loss": 3.1843,
      "step": 463860
    },
    {
      "epoch": 0.17407189600896086,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010723692369236924,
      "loss": 3.0503,
      "step": 463870
    },
    {
      "epoch": 0.17409053924511905,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010723492349234925,
      "loss": 3.0585,
      "step": 463880
    },
    {
      "epoch": 0.17410918248127724,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010723292329232923,
      "loss": 3.0248,
      "step": 463890
    },
    {
      "epoch": 0.17412782571743543,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010723092309230924,
      "loss": 3.2008,
      "step": 463900
    },
    {
      "epoch": 0.17414646895359362,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010722892289228923,
      "loss": 3.0784,
      "step": 463910
    },
    {
      "epoch": 0.17416511218975178,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010722692269226924,
      "loss": 2.9333,
      "step": 463920
    },
    {
      "epoch": 0.17418375542590997,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010722492249224923,
      "loss": 3.2293,
      "step": 463930
    },
    {
      "epoch": 0.17420239866206816,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010722292229222924,
      "loss": 3.0724,
      "step": 463940
    },
    {
      "epoch": 0.17422104189822635,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010722092209220922,
      "loss": 3.1684,
      "step": 463950
    },
    {
      "epoch": 0.17423968513438454,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010721892189218923,
      "loss": 3.0856,
      "step": 463960
    },
    {
      "epoch": 0.1742583283705427,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010721692169216921,
      "loss": 3.2527,
      "step": 463970
    },
    {
      "epoch": 0.1742769716067009,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010721492149214922,
      "loss": 3.0766,
      "step": 463980
    },
    {
      "epoch": 0.17429561484285908,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001072129212921292,
      "loss": 3.1519,
      "step": 463990
    },
    {
      "epoch": 0.17431425807901726,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010721092109210921,
      "loss": 3.1229,
      "step": 464000
    },
    {
      "epoch": 0.17433290131517545,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010720892089208921,
      "loss": 3.2144,
      "step": 464010
    },
    {
      "epoch": 0.17435154455133361,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010720692069206922,
      "loss": 3.1026,
      "step": 464020
    },
    {
      "epoch": 0.1743701877874918,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001072049204920492,
      "loss": 3.0944,
      "step": 464030
    },
    {
      "epoch": 0.17438883102365,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010720292029202921,
      "loss": 3.1014,
      "step": 464040
    },
    {
      "epoch": 0.17440747425980818,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010720092009200922,
      "loss": 3.0549,
      "step": 464050
    },
    {
      "epoch": 0.17442611749596637,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001071989198919892,
      "loss": 3.0865,
      "step": 464060
    },
    {
      "epoch": 0.17444476073212453,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010719691969196921,
      "loss": 3.0721,
      "step": 464070
    },
    {
      "epoch": 0.17446340396828272,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010719491949194919,
      "loss": 3.0642,
      "step": 464080
    },
    {
      "epoch": 0.1744820472044409,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001071929192919292,
      "loss": 3.1443,
      "step": 464090
    },
    {
      "epoch": 0.1745006904405991,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001071909190919092,
      "loss": 3.1259,
      "step": 464100
    },
    {
      "epoch": 0.1745193336767573,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001071889188918892,
      "loss": 3.0699,
      "step": 464110
    },
    {
      "epoch": 0.17453797691291548,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010718691869186919,
      "loss": 3.0456,
      "step": 464120
    },
    {
      "epoch": 0.17455662014907364,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001071849184918492,
      "loss": 3.1047,
      "step": 464130
    },
    {
      "epoch": 0.17457526338523183,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010718291829182918,
      "loss": 3.2103,
      "step": 464140
    },
    {
      "epoch": 0.17459390662139002,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010718091809180919,
      "loss": 3.2258,
      "step": 464150
    },
    {
      "epoch": 0.1746125498575482,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010717891789178918,
      "loss": 3.0072,
      "step": 464160
    },
    {
      "epoch": 0.1746311930937064,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010717691769176918,
      "loss": 3.118,
      "step": 464170
    },
    {
      "epoch": 0.17464983632986456,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010717491749174917,
      "loss": 3.0027,
      "step": 464180
    },
    {
      "epoch": 0.17466847956602274,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010717291729172918,
      "loss": 3.1384,
      "step": 464190
    },
    {
      "epoch": 0.17468712280218093,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010717091709170917,
      "loss": 3.2378,
      "step": 464200
    },
    {
      "epoch": 0.17470576603833912,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010716891689168917,
      "loss": 3.1015,
      "step": 464210
    },
    {
      "epoch": 0.1747244092744973,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010716691669166916,
      "loss": 3.0349,
      "step": 464220
    },
    {
      "epoch": 0.17474305251065547,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010716491649164917,
      "loss": 3.0878,
      "step": 464230
    },
    {
      "epoch": 0.17476169574681366,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010716291629162918,
      "loss": 3.048,
      "step": 464240
    },
    {
      "epoch": 0.17478033898297185,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010716091609160917,
      "loss": 3.2006,
      "step": 464250
    },
    {
      "epoch": 0.17479898221913004,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010715891589158918,
      "loss": 3.0292,
      "step": 464260
    },
    {
      "epoch": 0.17481762545528823,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010715691569156916,
      "loss": 2.9585,
      "step": 464270
    },
    {
      "epoch": 0.1748362686914464,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010715491549154917,
      "loss": 3.0657,
      "step": 464280
    },
    {
      "epoch": 0.17485491192760458,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010715291529152915,
      "loss": 3.1603,
      "step": 464290
    },
    {
      "epoch": 0.17487355516376277,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010715091509150916,
      "loss": 3.1357,
      "step": 464300
    },
    {
      "epoch": 0.17489219839992096,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010714891489148914,
      "loss": 3.0618,
      "step": 464310
    },
    {
      "epoch": 0.17491084163607915,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010714691469146915,
      "loss": 3.1036,
      "step": 464320
    },
    {
      "epoch": 0.1749294848722373,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010714491449144915,
      "loss": 3.1488,
      "step": 464330
    },
    {
      "epoch": 0.1749481281083955,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010714291429142916,
      "loss": 2.9197,
      "step": 464340
    },
    {
      "epoch": 0.17496677134455368,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010714091409140914,
      "loss": 3.0684,
      "step": 464350
    },
    {
      "epoch": 0.17498541458071187,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010713891389138915,
      "loss": 3.1257,
      "step": 464360
    },
    {
      "epoch": 0.17500405781687006,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010713691369136913,
      "loss": 3.0998,
      "step": 464370
    },
    {
      "epoch": 0.17502270105302825,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010713491349134914,
      "loss": 2.9897,
      "step": 464380
    },
    {
      "epoch": 0.1750413442891864,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010713291329132912,
      "loss": 3.0653,
      "step": 464390
    },
    {
      "epoch": 0.1750599875253446,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010713091309130913,
      "loss": 3.1629,
      "step": 464400
    },
    {
      "epoch": 0.1750786307615028,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010712891289128913,
      "loss": 3.0622,
      "step": 464410
    },
    {
      "epoch": 0.17509727399766098,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010712691269126914,
      "loss": 3.0234,
      "step": 464420
    },
    {
      "epoch": 0.17511591723381917,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010712491249124915,
      "loss": 3.0897,
      "step": 464430
    },
    {
      "epoch": 0.17513456046997733,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010712291229122913,
      "loss": 3.1372,
      "step": 464440
    },
    {
      "epoch": 0.17515320370613552,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010712091209120914,
      "loss": 3.182,
      "step": 464450
    },
    {
      "epoch": 0.1751718469422937,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010711891189118912,
      "loss": 3.0619,
      "step": 464460
    },
    {
      "epoch": 0.1751904901784519,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010711691169116913,
      "loss": 3.0853,
      "step": 464470
    },
    {
      "epoch": 0.17520913341461009,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010711491149114911,
      "loss": 2.9516,
      "step": 464480
    },
    {
      "epoch": 0.17522777665076825,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010711291129112912,
      "loss": 3.064,
      "step": 464490
    },
    {
      "epoch": 0.17524641988692644,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010711091109110911,
      "loss": 3.1009,
      "step": 464500
    },
    {
      "epoch": 0.17526506312308462,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010710891089108912,
      "loss": 3.0161,
      "step": 464510
    },
    {
      "epoch": 0.1752837063592428,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001071069106910691,
      "loss": 3.1086,
      "step": 464520
    },
    {
      "epoch": 0.175302349595401,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010710491049104912,
      "loss": 3.0212,
      "step": 464530
    },
    {
      "epoch": 0.17532099283155916,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001071029102910291,
      "loss": 3.0554,
      "step": 464540
    },
    {
      "epoch": 0.17533963606771735,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001071009100910091,
      "loss": 3.0007,
      "step": 464550
    },
    {
      "epoch": 0.17535827930387554,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001070989098909891,
      "loss": 2.9489,
      "step": 464560
    },
    {
      "epoch": 0.17537692254003373,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010709690969096911,
      "loss": 3.1405,
      "step": 464570
    },
    {
      "epoch": 0.17539556577619192,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001070949094909491,
      "loss": 2.9194,
      "step": 464580
    },
    {
      "epoch": 0.1754142090123501,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001070929092909291,
      "loss": 3.0908,
      "step": 464590
    },
    {
      "epoch": 0.17543285224850827,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010709090909090908,
      "loss": 3.0116,
      "step": 464600
    },
    {
      "epoch": 0.17545149548466646,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001070889088908891,
      "loss": 3.0659,
      "step": 464610
    },
    {
      "epoch": 0.17547013872082465,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010708690869086908,
      "loss": 3.0168,
      "step": 464620
    },
    {
      "epoch": 0.17548878195698284,
      "grad_norm": 0.75,
      "learning_rate": 0.00010708490849084908,
      "loss": 3.0838,
      "step": 464630
    },
    {
      "epoch": 0.17550742519314103,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001070829082908291,
      "loss": 3.0108,
      "step": 464640
    },
    {
      "epoch": 0.1755260684292992,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010708090809080909,
      "loss": 3.0242,
      "step": 464650
    },
    {
      "epoch": 0.17554471166545738,
      "grad_norm": 0.75,
      "learning_rate": 0.0001070789078907891,
      "loss": 3.1329,
      "step": 464660
    },
    {
      "epoch": 0.17556335490161556,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010707690769076908,
      "loss": 3.0631,
      "step": 464670
    },
    {
      "epoch": 0.17558199813777375,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010707490749074909,
      "loss": 3.1621,
      "step": 464680
    },
    {
      "epoch": 0.17560064137393194,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010707290729072907,
      "loss": 2.9326,
      "step": 464690
    },
    {
      "epoch": 0.1756192846100901,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010707090709070908,
      "loss": 3.1193,
      "step": 464700
    },
    {
      "epoch": 0.1756379278462483,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010706890689068906,
      "loss": 3.1172,
      "step": 464710
    },
    {
      "epoch": 0.17565657108240648,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010706690669066907,
      "loss": 3.138,
      "step": 464720
    },
    {
      "epoch": 0.17567521431856467,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010706490649064907,
      "loss": 3.1109,
      "step": 464730
    },
    {
      "epoch": 0.17569385755472286,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010706290629062908,
      "loss": 3.0765,
      "step": 464740
    },
    {
      "epoch": 0.17571250079088102,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010706090609060906,
      "loss": 3.0246,
      "step": 464750
    },
    {
      "epoch": 0.1757311440270392,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010705890589058907,
      "loss": 3.1075,
      "step": 464760
    },
    {
      "epoch": 0.1757497872631974,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010705690569056905,
      "loss": 3.0769,
      "step": 464770
    },
    {
      "epoch": 0.1757684304993556,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010705490549054906,
      "loss": 3.0635,
      "step": 464780
    },
    {
      "epoch": 0.17578707373551378,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010705290529052905,
      "loss": 3.0042,
      "step": 464790
    },
    {
      "epoch": 0.17580571697167197,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010705090509050905,
      "loss": 3.0413,
      "step": 464800
    },
    {
      "epoch": 0.17582436020783013,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010704890489048905,
      "loss": 3.0534,
      "step": 464810
    },
    {
      "epoch": 0.17584300344398832,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010704690469046906,
      "loss": 3.0897,
      "step": 464820
    },
    {
      "epoch": 0.1758616466801465,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010704490449044906,
      "loss": 3.0424,
      "step": 464830
    },
    {
      "epoch": 0.1758802899163047,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010704290429042905,
      "loss": 3.0977,
      "step": 464840
    },
    {
      "epoch": 0.17589893315246288,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010704090409040906,
      "loss": 3.1493,
      "step": 464850
    },
    {
      "epoch": 0.17591757638862104,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010703890389038904,
      "loss": 3.0425,
      "step": 464860
    },
    {
      "epoch": 0.17593621962477923,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010703690369036905,
      "loss": 3.1881,
      "step": 464870
    },
    {
      "epoch": 0.17595486286093742,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010703490349034904,
      "loss": 3.1357,
      "step": 464880
    },
    {
      "epoch": 0.1759735060970956,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010703290329032905,
      "loss": 3.0037,
      "step": 464890
    },
    {
      "epoch": 0.1759921493332538,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010703090309030903,
      "loss": 3.1693,
      "step": 464900
    },
    {
      "epoch": 0.17601079256941196,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010702890289028904,
      "loss": 3.069,
      "step": 464910
    },
    {
      "epoch": 0.17602943580557015,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010702690269026902,
      "loss": 3.0883,
      "step": 464920
    },
    {
      "epoch": 0.17604807904172834,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010702490249024903,
      "loss": 3.0792,
      "step": 464930
    },
    {
      "epoch": 0.17606672227788653,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010702290229022902,
      "loss": 2.932,
      "step": 464940
    },
    {
      "epoch": 0.17608536551404472,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010702090209020903,
      "loss": 3.0316,
      "step": 464950
    },
    {
      "epoch": 0.17610400875020288,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010701890189018902,
      "loss": 3.107,
      "step": 464960
    },
    {
      "epoch": 0.17612265198636107,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010701690169016903,
      "loss": 3.0299,
      "step": 464970
    },
    {
      "epoch": 0.17614129522251926,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010701490149014901,
      "loss": 3.0844,
      "step": 464980
    },
    {
      "epoch": 0.17615993845867745,
      "grad_norm": 0.75,
      "learning_rate": 0.00010701290129012902,
      "loss": 3.0654,
      "step": 464990
    },
    {
      "epoch": 0.17617858169483563,
      "grad_norm": 0.703125,
      "learning_rate": 0.000107010901090109,
      "loss": 2.9981,
      "step": 465000
    },
    {
      "epoch": 0.1761972249309938,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010700890089008901,
      "loss": 3.058,
      "step": 465010
    },
    {
      "epoch": 0.17621586816715198,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010700690069006902,
      "loss": 3.0659,
      "step": 465020
    },
    {
      "epoch": 0.17623451140331017,
      "grad_norm": 0.75,
      "learning_rate": 0.000107004900490049,
      "loss": 3.1721,
      "step": 465030
    },
    {
      "epoch": 0.17625315463946836,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010700290029002901,
      "loss": 3.1051,
      "step": 465040
    },
    {
      "epoch": 0.17627179787562655,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010700090009000901,
      "loss": 3.1452,
      "step": 465050
    },
    {
      "epoch": 0.17629044111178474,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010699889988998902,
      "loss": 2.9574,
      "step": 465060
    },
    {
      "epoch": 0.1763090843479429,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000106996899689969,
      "loss": 3.1145,
      "step": 465070
    },
    {
      "epoch": 0.1763277275841011,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010699489948994901,
      "loss": 2.9985,
      "step": 465080
    },
    {
      "epoch": 0.17634637082025928,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010699289928992899,
      "loss": 3.0067,
      "step": 465090
    },
    {
      "epoch": 0.17636501405641747,
      "grad_norm": 0.734375,
      "learning_rate": 0.000106990899089909,
      "loss": 3.0752,
      "step": 465100
    },
    {
      "epoch": 0.17638365729257566,
      "grad_norm": 0.765625,
      "learning_rate": 0.000106988898889889,
      "loss": 3.1042,
      "step": 465110
    },
    {
      "epoch": 0.17640230052873382,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010698689868986899,
      "loss": 3.0599,
      "step": 465120
    },
    {
      "epoch": 0.176420943764892,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010698489848984899,
      "loss": 2.9852,
      "step": 465130
    },
    {
      "epoch": 0.1764395870010502,
      "grad_norm": 0.71875,
      "learning_rate": 0.000106982898289829,
      "loss": 3.0207,
      "step": 465140
    },
    {
      "epoch": 0.17645823023720839,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010698089808980898,
      "loss": 3.1094,
      "step": 465150
    },
    {
      "epoch": 0.17647687347336657,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010697889788978899,
      "loss": 3.0534,
      "step": 465160
    },
    {
      "epoch": 0.17649551670952474,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010697689768976897,
      "loss": 3.0672,
      "step": 465170
    },
    {
      "epoch": 0.17651415994568292,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010697489748974898,
      "loss": 3.0422,
      "step": 465180
    },
    {
      "epoch": 0.1765328031818411,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010697289728972897,
      "loss": 2.9902,
      "step": 465190
    },
    {
      "epoch": 0.1765514464179993,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010697089708970898,
      "loss": 3.1414,
      "step": 465200
    },
    {
      "epoch": 0.1765700896541575,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010696889688968896,
      "loss": 3.0185,
      "step": 465210
    },
    {
      "epoch": 0.17658873289031565,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00010696689668966897,
      "loss": 3.0305,
      "step": 465220
    },
    {
      "epoch": 0.17660737612647384,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010696489648964898,
      "loss": 3.1215,
      "step": 465230
    },
    {
      "epoch": 0.17662601936263203,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010696289628962897,
      "loss": 2.993,
      "step": 465240
    },
    {
      "epoch": 0.17664466259879022,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010696089608960897,
      "loss": 3.0253,
      "step": 465250
    },
    {
      "epoch": 0.1766633058349484,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010695889588958896,
      "loss": 3.044,
      "step": 465260
    },
    {
      "epoch": 0.1766819490711066,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010695689568956897,
      "loss": 3.0896,
      "step": 465270
    },
    {
      "epoch": 0.17670059230726476,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010695489548954896,
      "loss": 3.1002,
      "step": 465280
    },
    {
      "epoch": 0.17671923554342295,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010695289528952897,
      "loss": 3.0312,
      "step": 465290
    },
    {
      "epoch": 0.17673787877958114,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010695089508950895,
      "loss": 2.9894,
      "step": 465300
    },
    {
      "epoch": 0.17675652201573933,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010694889488948896,
      "loss": 3.1544,
      "step": 465310
    },
    {
      "epoch": 0.17677516525189751,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010694689468946894,
      "loss": 2.9841,
      "step": 465320
    },
    {
      "epoch": 0.17679380848805568,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010694489448944895,
      "loss": 3.0402,
      "step": 465330
    },
    {
      "epoch": 0.17681245172421386,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010694289428942893,
      "loss": 3.0859,
      "step": 465340
    },
    {
      "epoch": 0.17683109496037205,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010694089408940894,
      "loss": 3.136,
      "step": 465350
    },
    {
      "epoch": 0.17684973819653024,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010693889388938894,
      "loss": 3.084,
      "step": 465360
    },
    {
      "epoch": 0.17686838143268843,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010693689368936895,
      "loss": 3.0485,
      "step": 465370
    },
    {
      "epoch": 0.1768870246688466,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010693489348934893,
      "loss": 3.0748,
      "step": 465380
    },
    {
      "epoch": 0.17690566790500478,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010693289328932894,
      "loss": 3.0055,
      "step": 465390
    },
    {
      "epoch": 0.17692431114116297,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010693089308930892,
      "loss": 2.8728,
      "step": 465400
    },
    {
      "epoch": 0.17694295437732116,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010692889288928893,
      "loss": 3.1575,
      "step": 465410
    },
    {
      "epoch": 0.17696159761347935,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010692689268926894,
      "loss": 3.0075,
      "step": 465420
    },
    {
      "epoch": 0.1769802408496375,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010692489248924894,
      "loss": 2.969,
      "step": 465430
    },
    {
      "epoch": 0.1769988840857957,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010692289228922893,
      "loss": 3.2263,
      "step": 465440
    },
    {
      "epoch": 0.1770175273219539,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010692089208920893,
      "loss": 2.9055,
      "step": 465450
    },
    {
      "epoch": 0.17703617055811208,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010691889188918894,
      "loss": 3.0602,
      "step": 465460
    },
    {
      "epoch": 0.17705481379427027,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010691689168916892,
      "loss": 3.0361,
      "step": 465470
    },
    {
      "epoch": 0.17707345703042845,
      "grad_norm": 0.75,
      "learning_rate": 0.00010691489148914893,
      "loss": 3.0377,
      "step": 465480
    },
    {
      "epoch": 0.17709210026658662,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010691289128912891,
      "loss": 3.0959,
      "step": 465490
    },
    {
      "epoch": 0.1771107435027448,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010691089108910892,
      "loss": 3.0295,
      "step": 465500
    },
    {
      "epoch": 0.177129386738903,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010690889088908891,
      "loss": 3.1221,
      "step": 465510
    },
    {
      "epoch": 0.17714802997506118,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010690689068906892,
      "loss": 3.1537,
      "step": 465520
    },
    {
      "epoch": 0.17716667321121937,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001069048904890489,
      "loss": 3.06,
      "step": 465530
    },
    {
      "epoch": 0.17718531644737753,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010690289028902891,
      "loss": 3.2114,
      "step": 465540
    },
    {
      "epoch": 0.17720395968353572,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001069008900890089,
      "loss": 3.0764,
      "step": 465550
    },
    {
      "epoch": 0.1772226029196939,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001068988898889889,
      "loss": 3.0917,
      "step": 465560
    },
    {
      "epoch": 0.1772412461558521,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010689688968896889,
      "loss": 3.0168,
      "step": 465570
    },
    {
      "epoch": 0.1772598893920103,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001068948894889489,
      "loss": 2.9656,
      "step": 465580
    },
    {
      "epoch": 0.17727853262816845,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010689288928892889,
      "loss": 3.0085,
      "step": 465590
    },
    {
      "epoch": 0.17729717586432664,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001068908890889089,
      "loss": 3.1543,
      "step": 465600
    },
    {
      "epoch": 0.17731581910048483,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010688888888888891,
      "loss": 3.1891,
      "step": 465610
    },
    {
      "epoch": 0.17733446233664302,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010688688868886889,
      "loss": 2.9925,
      "step": 465620
    },
    {
      "epoch": 0.1773531055728012,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001068848884888489,
      "loss": 3.0036,
      "step": 465630
    },
    {
      "epoch": 0.17737174880895937,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010688288828882888,
      "loss": 3.1101,
      "step": 465640
    },
    {
      "epoch": 0.17739039204511756,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010688088808880889,
      "loss": 3.2464,
      "step": 465650
    },
    {
      "epoch": 0.17740903528127575,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010687888788878887,
      "loss": 3.0892,
      "step": 465660
    },
    {
      "epoch": 0.17742767851743393,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010687688768876888,
      "loss": 3.04,
      "step": 465670
    },
    {
      "epoch": 0.17744632175359212,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010687488748874888,
      "loss": 2.9202,
      "step": 465680
    },
    {
      "epoch": 0.17746496498975028,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010687288728872889,
      "loss": 3.1299,
      "step": 465690
    },
    {
      "epoch": 0.17748360822590847,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010687088708870887,
      "loss": 3.0678,
      "step": 465700
    },
    {
      "epoch": 0.17750225146206666,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010686888688868888,
      "loss": 2.9544,
      "step": 465710
    },
    {
      "epoch": 0.17752089469822485,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010686688668866886,
      "loss": 3.0457,
      "step": 465720
    },
    {
      "epoch": 0.17753953793438304,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010686488648864887,
      "loss": 3.1163,
      "step": 465730
    },
    {
      "epoch": 0.17755818117054123,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010686288628862887,
      "loss": 3.0619,
      "step": 465740
    },
    {
      "epoch": 0.1775768244066994,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010686088608860886,
      "loss": 3.1527,
      "step": 465750
    },
    {
      "epoch": 0.17759546764285758,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010685888588858886,
      "loss": 3.0568,
      "step": 465760
    },
    {
      "epoch": 0.17761411087901577,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010685688568856887,
      "loss": 2.9565,
      "step": 465770
    },
    {
      "epoch": 0.17763275411517396,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010685488548854885,
      "loss": 3.0102,
      "step": 465780
    },
    {
      "epoch": 0.17765139735133215,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010685288528852886,
      "loss": 2.9419,
      "step": 465790
    },
    {
      "epoch": 0.1776700405874903,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010685088508850884,
      "loss": 3.0167,
      "step": 465800
    },
    {
      "epoch": 0.1776886838236485,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010684888488848885,
      "loss": 3.1542,
      "step": 465810
    },
    {
      "epoch": 0.17770732705980669,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010684688468846886,
      "loss": 3.1652,
      "step": 465820
    },
    {
      "epoch": 0.17772597029596487,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010684488448844885,
      "loss": 3.0753,
      "step": 465830
    },
    {
      "epoch": 0.17774461353212306,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010684288428842886,
      "loss": 3.1236,
      "step": 465840
    },
    {
      "epoch": 0.17776325676828122,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010684088408840885,
      "loss": 3.0207,
      "step": 465850
    },
    {
      "epoch": 0.1777819000044394,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010683888388838885,
      "loss": 2.9779,
      "step": 465860
    },
    {
      "epoch": 0.1778005432405976,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010683688368836884,
      "loss": 3.0934,
      "step": 465870
    },
    {
      "epoch": 0.1778191864767558,
      "grad_norm": 0.75,
      "learning_rate": 0.00010683488348834885,
      "loss": 3.1014,
      "step": 465880
    },
    {
      "epoch": 0.17783782971291398,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010683288328832883,
      "loss": 3.0671,
      "step": 465890
    },
    {
      "epoch": 0.17785647294907214,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010683088308830884,
      "loss": 2.9503,
      "step": 465900
    },
    {
      "epoch": 0.17787511618523033,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010682888288828883,
      "loss": 3.1113,
      "step": 465910
    },
    {
      "epoch": 0.17789375942138852,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010682688268826884,
      "loss": 3.1062,
      "step": 465920
    },
    {
      "epoch": 0.1779124026575467,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010682488248824882,
      "loss": 3.1413,
      "step": 465930
    },
    {
      "epoch": 0.1779310458937049,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010682288228822883,
      "loss": 2.964,
      "step": 465940
    },
    {
      "epoch": 0.1779496891298631,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010682088208820881,
      "loss": 3.1185,
      "step": 465950
    },
    {
      "epoch": 0.17796833236602125,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010681888188818882,
      "loss": 3.0823,
      "step": 465960
    },
    {
      "epoch": 0.17798697560217944,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001068168816881688,
      "loss": 3.0997,
      "step": 465970
    },
    {
      "epoch": 0.17800561883833763,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010681488148814882,
      "loss": 3.1184,
      "step": 465980
    },
    {
      "epoch": 0.17802426207449581,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010681288128812881,
      "loss": 3.0951,
      "step": 465990
    },
    {
      "epoch": 0.178042905310654,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010681088108810882,
      "loss": 3.1094,
      "step": 466000
    },
    {
      "epoch": 0.17806154854681217,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010680888088808883,
      "loss": 3.038,
      "step": 466010
    },
    {
      "epoch": 0.17808019178297035,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010680688068806881,
      "loss": 3.0363,
      "step": 466020
    },
    {
      "epoch": 0.17809883501912854,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010680488048804882,
      "loss": 3.1536,
      "step": 466030
    },
    {
      "epoch": 0.17811747825528673,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001068028802880288,
      "loss": 2.943,
      "step": 466040
    },
    {
      "epoch": 0.17813612149144492,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010680088008800881,
      "loss": 3.0844,
      "step": 466050
    },
    {
      "epoch": 0.17815476472760308,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010679887988798881,
      "loss": 3.0829,
      "step": 466060
    },
    {
      "epoch": 0.17817340796376127,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001067968796879688,
      "loss": 3.0277,
      "step": 466070
    },
    {
      "epoch": 0.17819205119991946,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001067948794879488,
      "loss": 3.2064,
      "step": 466080
    },
    {
      "epoch": 0.17821069443607765,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010679287928792881,
      "loss": 3.0342,
      "step": 466090
    },
    {
      "epoch": 0.17822933767223584,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010679087908790879,
      "loss": 3.0311,
      "step": 466100
    },
    {
      "epoch": 0.178247980908394,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001067888788878888,
      "loss": 3.1828,
      "step": 466110
    },
    {
      "epoch": 0.1782666241445522,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010678687868786878,
      "loss": 3.1822,
      "step": 466120
    },
    {
      "epoch": 0.17828526738071038,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010678487848784879,
      "loss": 3.0696,
      "step": 466130
    },
    {
      "epoch": 0.17830391061686857,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010678287828782879,
      "loss": 3.1655,
      "step": 466140
    },
    {
      "epoch": 0.17832255385302676,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001067808780878088,
      "loss": 3.0939,
      "step": 466150
    },
    {
      "epoch": 0.17834119708918494,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010677887788778878,
      "loss": 3.0945,
      "step": 466160
    },
    {
      "epoch": 0.1783598403253431,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010677687768776879,
      "loss": 3.1205,
      "step": 466170
    },
    {
      "epoch": 0.1783784835615013,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010677487748774877,
      "loss": 2.9507,
      "step": 466180
    },
    {
      "epoch": 0.17839712679765948,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010677287728772878,
      "loss": 2.9744,
      "step": 466190
    },
    {
      "epoch": 0.17841577003381767,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010677087708770876,
      "loss": 3.1476,
      "step": 466200
    },
    {
      "epoch": 0.17843441326997586,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010676887688768877,
      "loss": 3.1104,
      "step": 466210
    },
    {
      "epoch": 0.17845305650613402,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010676687668766878,
      "loss": 2.988,
      "step": 466220
    },
    {
      "epoch": 0.1784716997422922,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010676487648764877,
      "loss": 3.0965,
      "step": 466230
    },
    {
      "epoch": 0.1784903429784504,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010676287628762878,
      "loss": 3.1314,
      "step": 466240
    },
    {
      "epoch": 0.1785089862146086,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010676087608760876,
      "loss": 3.0182,
      "step": 466250
    },
    {
      "epoch": 0.17852762945076678,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010675887588758877,
      "loss": 3.0705,
      "step": 466260
    },
    {
      "epoch": 0.17854627268692494,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010675687568756876,
      "loss": 2.9873,
      "step": 466270
    },
    {
      "epoch": 0.17856491592308313,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010675487548754876,
      "loss": 3.0874,
      "step": 466280
    },
    {
      "epoch": 0.17858355915924132,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010675287528752875,
      "loss": 2.8677,
      "step": 466290
    },
    {
      "epoch": 0.1786022023953995,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010675087508750876,
      "loss": 3.0198,
      "step": 466300
    },
    {
      "epoch": 0.1786208456315577,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010674887488748875,
      "loss": 3.0304,
      "step": 466310
    },
    {
      "epoch": 0.17863948886771586,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010674687468746876,
      "loss": 3.1126,
      "step": 466320
    },
    {
      "epoch": 0.17865813210387405,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010674487448744874,
      "loss": 2.9674,
      "step": 466330
    },
    {
      "epoch": 0.17867677534003223,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010674287428742875,
      "loss": 2.9861,
      "step": 466340
    },
    {
      "epoch": 0.17869541857619042,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010674087408740873,
      "loss": 3.14,
      "step": 466350
    },
    {
      "epoch": 0.1787140618123486,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010673887388738874,
      "loss": 3.1624,
      "step": 466360
    },
    {
      "epoch": 0.17873270504850677,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010673687368736874,
      "loss": 3.1174,
      "step": 466370
    },
    {
      "epoch": 0.17875134828466496,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010673487348734875,
      "loss": 3.0106,
      "step": 466380
    },
    {
      "epoch": 0.17876999152082315,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010673287328732873,
      "loss": 3.0671,
      "step": 466390
    },
    {
      "epoch": 0.17878863475698134,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010673087308730874,
      "loss": 3.0756,
      "step": 466400
    },
    {
      "epoch": 0.17880727799313953,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010672887288728875,
      "loss": 3.0574,
      "step": 466410
    },
    {
      "epoch": 0.17882592122929772,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010672687268726873,
      "loss": 3.0101,
      "step": 466420
    },
    {
      "epoch": 0.17884456446545588,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010672487248724874,
      "loss": 3.0552,
      "step": 466430
    },
    {
      "epoch": 0.17886320770161407,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010672287228722872,
      "loss": 3.0237,
      "step": 466440
    },
    {
      "epoch": 0.17888185093777226,
      "grad_norm": 0.75,
      "learning_rate": 0.00010672087208720873,
      "loss": 2.9983,
      "step": 466450
    },
    {
      "epoch": 0.17890049417393045,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010671887188718873,
      "loss": 3.1615,
      "step": 466460
    },
    {
      "epoch": 0.17891913741008864,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010671687168716873,
      "loss": 3.0291,
      "step": 466470
    },
    {
      "epoch": 0.1789377806462468,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010671487148714872,
      "loss": 3.0098,
      "step": 466480
    },
    {
      "epoch": 0.17895642388240499,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010671287128712873,
      "loss": 2.9293,
      "step": 466490
    },
    {
      "epoch": 0.17897506711856317,
      "grad_norm": 0.75,
      "learning_rate": 0.00010671087108710871,
      "loss": 2.9794,
      "step": 466500
    },
    {
      "epoch": 0.17899371035472136,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010670887088708872,
      "loss": 3.1476,
      "step": 466510
    },
    {
      "epoch": 0.17901235359087955,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001067068706870687,
      "loss": 3.0163,
      "step": 466520
    },
    {
      "epoch": 0.17903099682703771,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010670487048704871,
      "loss": 3.0268,
      "step": 466530
    },
    {
      "epoch": 0.1790496400631959,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001067028702870287,
      "loss": 3.1158,
      "step": 466540
    },
    {
      "epoch": 0.1790682832993541,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010670087008700871,
      "loss": 3.0357,
      "step": 466550
    },
    {
      "epoch": 0.17908692653551228,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001066988698869887,
      "loss": 3.0754,
      "step": 466560
    },
    {
      "epoch": 0.17910556977167047,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001066968696869687,
      "loss": 2.982,
      "step": 466570
    },
    {
      "epoch": 0.17912421300782863,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010669486948694869,
      "loss": 3.0332,
      "step": 466580
    },
    {
      "epoch": 0.17914285624398682,
      "grad_norm": 0.75,
      "learning_rate": 0.0001066928692869287,
      "loss": 3.0434,
      "step": 466590
    },
    {
      "epoch": 0.179161499480145,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001066908690869087,
      "loss": 3.0559,
      "step": 466600
    },
    {
      "epoch": 0.1791801427163032,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010668886888688869,
      "loss": 2.9069,
      "step": 466610
    },
    {
      "epoch": 0.1791987859524614,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001066868686868687,
      "loss": 3.1023,
      "step": 466620
    },
    {
      "epoch": 0.17921742918861958,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010668486848684869,
      "loss": 3.004,
      "step": 466630
    },
    {
      "epoch": 0.17923607242477774,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001066828682868287,
      "loss": 3.078,
      "step": 466640
    },
    {
      "epoch": 0.17925471566093593,
      "grad_norm": 0.75,
      "learning_rate": 0.00010668086808680868,
      "loss": 2.9279,
      "step": 466650
    },
    {
      "epoch": 0.17927335889709411,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010667886788678869,
      "loss": 3.0389,
      "step": 466660
    },
    {
      "epoch": 0.1792920021332523,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010667686768676867,
      "loss": 3.1056,
      "step": 466670
    },
    {
      "epoch": 0.1793106453694105,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010667486748674868,
      "loss": 3.0779,
      "step": 466680
    },
    {
      "epoch": 0.17932928860556865,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010667286728672868,
      "loss": 3.1416,
      "step": 466690
    },
    {
      "epoch": 0.17934793184172684,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010667086708670869,
      "loss": 3.0162,
      "step": 466700
    },
    {
      "epoch": 0.17936657507788503,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010666886688668867,
      "loss": 3.1986,
      "step": 466710
    },
    {
      "epoch": 0.17938521831404322,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010666686668666868,
      "loss": 3.0359,
      "step": 466720
    },
    {
      "epoch": 0.1794038615502014,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010666486648664866,
      "loss": 3.0142,
      "step": 466730
    },
    {
      "epoch": 0.17942250478635957,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010666286628662867,
      "loss": 3.106,
      "step": 466740
    },
    {
      "epoch": 0.17944114802251776,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010666086608660865,
      "loss": 3.0217,
      "step": 466750
    },
    {
      "epoch": 0.17945979125867595,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010665886588658866,
      "loss": 3.0797,
      "step": 466760
    },
    {
      "epoch": 0.17947843449483414,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010665686568656866,
      "loss": 3.0639,
      "step": 466770
    },
    {
      "epoch": 0.17949707773099233,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010665486548654867,
      "loss": 3.015,
      "step": 466780
    },
    {
      "epoch": 0.1795157209671505,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010665286528652865,
      "loss": 3.1403,
      "step": 466790
    },
    {
      "epoch": 0.17953436420330868,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010665086508650866,
      "loss": 3.0723,
      "step": 466800
    },
    {
      "epoch": 0.17955300743946687,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010664886488648867,
      "loss": 3.06,
      "step": 466810
    },
    {
      "epoch": 0.17957165067562506,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010664686468646865,
      "loss": 3.0195,
      "step": 466820
    },
    {
      "epoch": 0.17959029391178324,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010664486448644866,
      "loss": 3.1807,
      "step": 466830
    },
    {
      "epoch": 0.17960893714794143,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010664286428642864,
      "loss": 3.1014,
      "step": 466840
    },
    {
      "epoch": 0.1796275803840996,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010664086408640865,
      "loss": 3.0224,
      "step": 466850
    },
    {
      "epoch": 0.17964622362025778,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010663886388638864,
      "loss": 3.0642,
      "step": 466860
    },
    {
      "epoch": 0.17966486685641597,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010663686368636865,
      "loss": 3.046,
      "step": 466870
    },
    {
      "epoch": 0.17968351009257416,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010663486348634864,
      "loss": 3.173,
      "step": 466880
    },
    {
      "epoch": 0.17970215332873235,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010663286328632864,
      "loss": 3.1699,
      "step": 466890
    },
    {
      "epoch": 0.1797207965648905,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010663086308630863,
      "loss": 3.1926,
      "step": 466900
    },
    {
      "epoch": 0.1797394398010487,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010662886288628864,
      "loss": 2.8865,
      "step": 466910
    },
    {
      "epoch": 0.1797580830372069,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010662686268626863,
      "loss": 3.0245,
      "step": 466920
    },
    {
      "epoch": 0.17977672627336508,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010662486248624863,
      "loss": 3.254,
      "step": 466930
    },
    {
      "epoch": 0.17979536950952327,
      "grad_norm": 0.75,
      "learning_rate": 0.00010662286228622862,
      "loss": 3.0743,
      "step": 466940
    },
    {
      "epoch": 0.17981401274568143,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010662086208620863,
      "loss": 3.1747,
      "step": 466950
    },
    {
      "epoch": 0.17983265598183962,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010661886188618861,
      "loss": 2.9391,
      "step": 466960
    },
    {
      "epoch": 0.1798512992179978,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010661686168616862,
      "loss": 2.9962,
      "step": 466970
    },
    {
      "epoch": 0.179869942454156,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001066148614861486,
      "loss": 3.0613,
      "step": 466980
    },
    {
      "epoch": 0.17988858569031418,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010661286128612861,
      "loss": 3.0723,
      "step": 466990
    },
    {
      "epoch": 0.17990722892647235,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010661086108610862,
      "loss": 2.9922,
      "step": 467000
    },
    {
      "epoch": 0.17992587216263053,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010660886088608862,
      "loss": 2.9551,
      "step": 467010
    },
    {
      "epoch": 0.17994451539878872,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010660686068606863,
      "loss": 3.0652,
      "step": 467020
    },
    {
      "epoch": 0.1799631586349469,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010660486048604861,
      "loss": 3.0965,
      "step": 467030
    },
    {
      "epoch": 0.1799818018711051,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010660286028602862,
      "loss": 3.1425,
      "step": 467040
    },
    {
      "epoch": 0.1800004451072633,
      "grad_norm": 0.875,
      "learning_rate": 0.0001066008600860086,
      "loss": 3.2393,
      "step": 467050
    },
    {
      "epoch": 0.18001908834342145,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010659885988598861,
      "loss": 3.0127,
      "step": 467060
    },
    {
      "epoch": 0.18003773157957964,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010659685968596859,
      "loss": 3.0853,
      "step": 467070
    },
    {
      "epoch": 0.18005637481573783,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001065948594859486,
      "loss": 2.9789,
      "step": 467080
    },
    {
      "epoch": 0.18007501805189602,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001065928592859286,
      "loss": 3.2134,
      "step": 467090
    },
    {
      "epoch": 0.1800936612880542,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001065908590859086,
      "loss": 3.0838,
      "step": 467100
    },
    {
      "epoch": 0.18011230452421237,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010658885888588859,
      "loss": 3.0218,
      "step": 467110
    },
    {
      "epoch": 0.18013094776037056,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001065868586858686,
      "loss": 3.069,
      "step": 467120
    },
    {
      "epoch": 0.18014959099652875,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010658485848584858,
      "loss": 3.048,
      "step": 467130
    },
    {
      "epoch": 0.18016823423268694,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010658285828582859,
      "loss": 3.1114,
      "step": 467140
    },
    {
      "epoch": 0.18018687746884512,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010658085808580857,
      "loss": 3.0611,
      "step": 467150
    },
    {
      "epoch": 0.18020552070500329,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010657885788578858,
      "loss": 3.0304,
      "step": 467160
    },
    {
      "epoch": 0.18022416394116147,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010657685768576858,
      "loss": 2.991,
      "step": 467170
    },
    {
      "epoch": 0.18024280717731966,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010657485748574858,
      "loss": 3.0692,
      "step": 467180
    },
    {
      "epoch": 0.18026145041347785,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001065728572857286,
      "loss": 3.0204,
      "step": 467190
    },
    {
      "epoch": 0.18028009364963604,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010657085708570858,
      "loss": 3.0376,
      "step": 467200
    },
    {
      "epoch": 0.1802987368857942,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010656885688568859,
      "loss": 3.0316,
      "step": 467210
    },
    {
      "epoch": 0.1803173801219524,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010656685668566857,
      "loss": 3.1398,
      "step": 467220
    },
    {
      "epoch": 0.18033602335811058,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010656485648564858,
      "loss": 3.035,
      "step": 467230
    },
    {
      "epoch": 0.18035466659426877,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010656285628562856,
      "loss": 3.2015,
      "step": 467240
    },
    {
      "epoch": 0.18037330983042696,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010656085608560857,
      "loss": 3.1541,
      "step": 467250
    },
    {
      "epoch": 0.18039195306658512,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010655885588558856,
      "loss": 2.9749,
      "step": 467260
    },
    {
      "epoch": 0.1804105963027433,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010655685568556857,
      "loss": 3.1247,
      "step": 467270
    },
    {
      "epoch": 0.1804292395389015,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010655485548554855,
      "loss": 3.0844,
      "step": 467280
    },
    {
      "epoch": 0.1804478827750597,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010655285528552856,
      "loss": 3.0565,
      "step": 467290
    },
    {
      "epoch": 0.18046652601121788,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010655085508550855,
      "loss": 3.2697,
      "step": 467300
    },
    {
      "epoch": 0.18048516924737606,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010654885488548855,
      "loss": 3.1256,
      "step": 467310
    },
    {
      "epoch": 0.18050381248353423,
      "grad_norm": 0.75,
      "learning_rate": 0.00010654685468546855,
      "loss": 3.1289,
      "step": 467320
    },
    {
      "epoch": 0.18052245571969242,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010654485448544856,
      "loss": 2.968,
      "step": 467330
    },
    {
      "epoch": 0.1805410989558506,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010654285428542854,
      "loss": 3.1089,
      "step": 467340
    },
    {
      "epoch": 0.1805597421920088,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010654085408540855,
      "loss": 3.1282,
      "step": 467350
    },
    {
      "epoch": 0.18057838542816698,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010653885388538853,
      "loss": 3.0198,
      "step": 467360
    },
    {
      "epoch": 0.18059702866432514,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010653685368536854,
      "loss": 2.9402,
      "step": 467370
    },
    {
      "epoch": 0.18061567190048333,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010653485348534852,
      "loss": 3.08,
      "step": 467380
    },
    {
      "epoch": 0.18063431513664152,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010653285328532853,
      "loss": 3.0075,
      "step": 467390
    },
    {
      "epoch": 0.1806529583727997,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010653085308530854,
      "loss": 3.1498,
      "step": 467400
    },
    {
      "epoch": 0.1806716016089579,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010652885288528854,
      "loss": 3.0707,
      "step": 467410
    },
    {
      "epoch": 0.18069024484511606,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010652685268526855,
      "loss": 3.0963,
      "step": 467420
    },
    {
      "epoch": 0.18070888808127425,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010652485248524853,
      "loss": 2.9718,
      "step": 467430
    },
    {
      "epoch": 0.18072753131743244,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010652285228522854,
      "loss": 3.0098,
      "step": 467440
    },
    {
      "epoch": 0.18074617455359063,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010652085208520852,
      "loss": 3.0085,
      "step": 467450
    },
    {
      "epoch": 0.18076481778974882,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010651885188518853,
      "loss": 3.1183,
      "step": 467460
    },
    {
      "epoch": 0.18078346102590698,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010651685168516851,
      "loss": 3.1316,
      "step": 467470
    },
    {
      "epoch": 0.18080210426206517,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010651485148514852,
      "loss": 3.0378,
      "step": 467480
    },
    {
      "epoch": 0.18082074749822336,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010651285128512852,
      "loss": 2.9948,
      "step": 467490
    },
    {
      "epoch": 0.18083939073438154,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010651085108510853,
      "loss": 2.9806,
      "step": 467500
    },
    {
      "epoch": 0.18085803397053973,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010650885088508851,
      "loss": 3.1831,
      "step": 467510
    },
    {
      "epoch": 0.18087667720669792,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010650685068506852,
      "loss": 3.1371,
      "step": 467520
    },
    {
      "epoch": 0.18089532044285608,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001065048504850485,
      "loss": 3.117,
      "step": 467530
    },
    {
      "epoch": 0.18091396367901427,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010650285028502851,
      "loss": 3.1504,
      "step": 467540
    },
    {
      "epoch": 0.18093260691517246,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001065008500850085,
      "loss": 3.102,
      "step": 467550
    },
    {
      "epoch": 0.18095125015133065,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001064988498849885,
      "loss": 3.1105,
      "step": 467560
    },
    {
      "epoch": 0.18096989338748884,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001064968496849685,
      "loss": 3.0763,
      "step": 467570
    },
    {
      "epoch": 0.180988536623647,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001064948494849485,
      "loss": 3.0618,
      "step": 467580
    },
    {
      "epoch": 0.1810071798598052,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010649284928492851,
      "loss": 3.0072,
      "step": 467590
    },
    {
      "epoch": 0.18102582309596338,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001064908490849085,
      "loss": 3.0038,
      "step": 467600
    },
    {
      "epoch": 0.18104446633212157,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001064888488848885,
      "loss": 2.9733,
      "step": 467610
    },
    {
      "epoch": 0.18106310956827976,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010648684868486849,
      "loss": 3.0953,
      "step": 467620
    },
    {
      "epoch": 0.18108175280443792,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001064848484848485,
      "loss": 3.0084,
      "step": 467630
    },
    {
      "epoch": 0.1811003960405961,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010648284828482849,
      "loss": 3.0581,
      "step": 467640
    },
    {
      "epoch": 0.1811190392767543,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001064808480848085,
      "loss": 3.136,
      "step": 467650
    },
    {
      "epoch": 0.18113768251291248,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010647884788478848,
      "loss": 3.007,
      "step": 467660
    },
    {
      "epoch": 0.18115632574907067,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010647684768476849,
      "loss": 2.9234,
      "step": 467670
    },
    {
      "epoch": 0.18117496898522883,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010647484748474847,
      "loss": 3.0431,
      "step": 467680
    },
    {
      "epoch": 0.18119361222138702,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010647284728472848,
      "loss": 3.1559,
      "step": 467690
    },
    {
      "epoch": 0.1812122554575452,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010647084708470846,
      "loss": 2.9688,
      "step": 467700
    },
    {
      "epoch": 0.1812308986937034,
      "grad_norm": 0.75,
      "learning_rate": 0.00010646884688468847,
      "loss": 3.1014,
      "step": 467710
    },
    {
      "epoch": 0.1812495419298616,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010646684668466847,
      "loss": 3.1396,
      "step": 467720
    },
    {
      "epoch": 0.18126818516601978,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010646484648464848,
      "loss": 3.1053,
      "step": 467730
    },
    {
      "epoch": 0.18128682840217794,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010646284628462846,
      "loss": 3.1485,
      "step": 467740
    },
    {
      "epoch": 0.18130547163833613,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010646084608460847,
      "loss": 3.1829,
      "step": 467750
    },
    {
      "epoch": 0.18132411487449432,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010645884588458845,
      "loss": 2.9951,
      "step": 467760
    },
    {
      "epoch": 0.1813427581106525,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010645684568456846,
      "loss": 3.1204,
      "step": 467770
    },
    {
      "epoch": 0.1813614013468107,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010645484548454844,
      "loss": 3.1713,
      "step": 467780
    },
    {
      "epoch": 0.18138004458296886,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010645284528452845,
      "loss": 3.2329,
      "step": 467790
    },
    {
      "epoch": 0.18139868781912705,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010645084508450846,
      "loss": 3.033,
      "step": 467800
    },
    {
      "epoch": 0.18141733105528524,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010644884488448846,
      "loss": 3.0425,
      "step": 467810
    },
    {
      "epoch": 0.18143597429144342,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010644684468446847,
      "loss": 3.1274,
      "step": 467820
    },
    {
      "epoch": 0.1814546175276016,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010644484448444845,
      "loss": 3.1103,
      "step": 467830
    },
    {
      "epoch": 0.18147326076375977,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010644284428442846,
      "loss": 3.0527,
      "step": 467840
    },
    {
      "epoch": 0.18149190399991796,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010644084408440844,
      "loss": 3.05,
      "step": 467850
    },
    {
      "epoch": 0.18151054723607615,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010643884388438845,
      "loss": 3.2041,
      "step": 467860
    },
    {
      "epoch": 0.18152919047223434,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010643684368436844,
      "loss": 2.973,
      "step": 467870
    },
    {
      "epoch": 0.18154783370839253,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00010643484348434844,
      "loss": 3.1093,
      "step": 467880
    },
    {
      "epoch": 0.1815664769445507,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010643284328432843,
      "loss": 3.0273,
      "step": 467890
    },
    {
      "epoch": 0.18158512018070888,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010643084308430844,
      "loss": 2.9962,
      "step": 467900
    },
    {
      "epoch": 0.18160376341686707,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010642884288428843,
      "loss": 3.1717,
      "step": 467910
    },
    {
      "epoch": 0.18162240665302526,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010642684268426843,
      "loss": 3.0624,
      "step": 467920
    },
    {
      "epoch": 0.18164104988918345,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010642484248424842,
      "loss": 3.0897,
      "step": 467930
    },
    {
      "epoch": 0.1816596931253416,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010642284228422843,
      "loss": 3.0741,
      "step": 467940
    },
    {
      "epoch": 0.1816783363614998,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010642084208420842,
      "loss": 3.0658,
      "step": 467950
    },
    {
      "epoch": 0.181696979597658,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010641884188418843,
      "loss": 3.0735,
      "step": 467960
    },
    {
      "epoch": 0.18171562283381618,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010641684168416841,
      "loss": 3.067,
      "step": 467970
    },
    {
      "epoch": 0.18173426606997437,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010641484148414842,
      "loss": 3.1643,
      "step": 467980
    },
    {
      "epoch": 0.18175290930613255,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010641284128412843,
      "loss": 3.0644,
      "step": 467990
    },
    {
      "epoch": 0.18177155254229072,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010641084108410841,
      "loss": 3.1033,
      "step": 468000
    },
    {
      "epoch": 0.1817901957784489,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010640884088408842,
      "loss": 3.2754,
      "step": 468010
    },
    {
      "epoch": 0.1818088390146071,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001064068406840684,
      "loss": 3.194,
      "step": 468020
    },
    {
      "epoch": 0.18182748225076528,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010640484048404841,
      "loss": 3.2895,
      "step": 468030
    },
    {
      "epoch": 0.18184612548692347,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010640284028402841,
      "loss": 3.274,
      "step": 468040
    },
    {
      "epoch": 0.18186476872308163,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00010640084008400842,
      "loss": 3.2936,
      "step": 468050
    },
    {
      "epoch": 0.18188341195923982,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001063988398839884,
      "loss": 3.2742,
      "step": 468060
    },
    {
      "epoch": 0.181902055195398,
      "grad_norm": 0.9375,
      "learning_rate": 0.00010639683968396841,
      "loss": 3.2544,
      "step": 468070
    },
    {
      "epoch": 0.1819206984315562,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010639483948394839,
      "loss": 3.1351,
      "step": 468080
    },
    {
      "epoch": 0.1819393416677144,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001063928392839284,
      "loss": 3.1741,
      "step": 468090
    },
    {
      "epoch": 0.18195798490387255,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010639083908390838,
      "loss": 3.1847,
      "step": 468100
    },
    {
      "epoch": 0.18197662814003074,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010638883888388839,
      "loss": 3.152,
      "step": 468110
    },
    {
      "epoch": 0.18199527137618893,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010638683868386839,
      "loss": 3.0954,
      "step": 468120
    },
    {
      "epoch": 0.18201391461234712,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001063848384838484,
      "loss": 3.2507,
      "step": 468130
    },
    {
      "epoch": 0.1820325578485053,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010638283828382838,
      "loss": 3.1873,
      "step": 468140
    },
    {
      "epoch": 0.18205120108466347,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010638083808380839,
      "loss": 3.1721,
      "step": 468150
    },
    {
      "epoch": 0.18206984432082166,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010637883788378837,
      "loss": 3.2438,
      "step": 468160
    },
    {
      "epoch": 0.18208848755697984,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010637683768376838,
      "loss": 3.2451,
      "step": 468170
    },
    {
      "epoch": 0.18210713079313803,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010637483748374839,
      "loss": 3.1817,
      "step": 468180
    },
    {
      "epoch": 0.18212577402929622,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010637283728372838,
      "loss": 3.2594,
      "step": 468190
    },
    {
      "epoch": 0.1821444172654544,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010637083708370838,
      "loss": 3.2545,
      "step": 468200
    },
    {
      "epoch": 0.18216306050161257,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010636883688368837,
      "loss": 3.2363,
      "step": 468210
    },
    {
      "epoch": 0.18218170373777076,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010636683668366838,
      "loss": 3.1996,
      "step": 468220
    },
    {
      "epoch": 0.18220034697392895,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010636483648364837,
      "loss": 3.1204,
      "step": 468230
    },
    {
      "epoch": 0.18221899021008714,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010636283628362838,
      "loss": 3.2926,
      "step": 468240
    },
    {
      "epoch": 0.18223763344624533,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010636083608360836,
      "loss": 3.2302,
      "step": 468250
    },
    {
      "epoch": 0.1822562766824035,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010635883588358837,
      "loss": 3.2359,
      "step": 468260
    },
    {
      "epoch": 0.18227491991856168,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010635683568356836,
      "loss": 3.1248,
      "step": 468270
    },
    {
      "epoch": 0.18229356315471987,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010635483548354837,
      "loss": 3.3682,
      "step": 468280
    },
    {
      "epoch": 0.18231220639087806,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010635283528352835,
      "loss": 3.1332,
      "step": 468290
    },
    {
      "epoch": 0.18233084962703625,
      "grad_norm": 0.875,
      "learning_rate": 0.00010635083508350836,
      "loss": 3.1413,
      "step": 468300
    },
    {
      "epoch": 0.1823494928631944,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010634883488348834,
      "loss": 3.0642,
      "step": 468310
    },
    {
      "epoch": 0.1823681360993526,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010634683468346835,
      "loss": 3.2111,
      "step": 468320
    },
    {
      "epoch": 0.18238677933551078,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010634483448344834,
      "loss": 3.2573,
      "step": 468330
    },
    {
      "epoch": 0.18240542257166897,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010634283428342834,
      "loss": 3.1341,
      "step": 468340
    },
    {
      "epoch": 0.18242406580782716,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010634083408340834,
      "loss": 3.3896,
      "step": 468350
    },
    {
      "epoch": 0.18244270904398532,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010633883388338835,
      "loss": 3.2818,
      "step": 468360
    },
    {
      "epoch": 0.1824613522801435,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010633683368336833,
      "loss": 3.1808,
      "step": 468370
    },
    {
      "epoch": 0.1824799955163017,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010633483348334834,
      "loss": 3.3879,
      "step": 468380
    },
    {
      "epoch": 0.1824986387524599,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010633283328332835,
      "loss": 3.0465,
      "step": 468390
    },
    {
      "epoch": 0.18251728198861808,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010633083308330833,
      "loss": 3.1197,
      "step": 468400
    },
    {
      "epoch": 0.18253592522477627,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010632883288328834,
      "loss": 3.0825,
      "step": 468410
    },
    {
      "epoch": 0.18255456846093443,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010632683268326832,
      "loss": 3.2609,
      "step": 468420
    },
    {
      "epoch": 0.18257321169709262,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010632483248324833,
      "loss": 3.2199,
      "step": 468430
    },
    {
      "epoch": 0.1825918549332508,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010632283228322833,
      "loss": 3.2453,
      "step": 468440
    },
    {
      "epoch": 0.182610498169409,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010632083208320834,
      "loss": 3.2463,
      "step": 468450
    },
    {
      "epoch": 0.18262914140556719,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010631883188318832,
      "loss": 3.2122,
      "step": 468460
    },
    {
      "epoch": 0.18264778464172535,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010631683168316833,
      "loss": 3.3034,
      "step": 468470
    },
    {
      "epoch": 0.18266642787788354,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010631483148314831,
      "loss": 3.136,
      "step": 468480
    },
    {
      "epoch": 0.18268507111404172,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010631283128312832,
      "loss": 3.2273,
      "step": 468490
    },
    {
      "epoch": 0.18270371435019991,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010631083108310831,
      "loss": 3.2655,
      "step": 468500
    },
    {
      "epoch": 0.1827223575863581,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010630883088308832,
      "loss": 3.1409,
      "step": 468510
    },
    {
      "epoch": 0.18274100082251626,
      "grad_norm": 0.625,
      "learning_rate": 0.0001063068306830683,
      "loss": 3.1794,
      "step": 468520
    },
    {
      "epoch": 0.18275964405867445,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010630483048304832,
      "loss": 3.1398,
      "step": 468530
    },
    {
      "epoch": 0.18277828729483264,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001063028302830283,
      "loss": 3.3141,
      "step": 468540
    },
    {
      "epoch": 0.18279693053099083,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001063008300830083,
      "loss": 3.2308,
      "step": 468550
    },
    {
      "epoch": 0.18281557376714902,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010629882988298829,
      "loss": 3.111,
      "step": 468560
    },
    {
      "epoch": 0.18283421700330718,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001062968296829683,
      "loss": 3.2941,
      "step": 468570
    },
    {
      "epoch": 0.18285286023946537,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001062948294829483,
      "loss": 3.2586,
      "step": 468580
    },
    {
      "epoch": 0.18287150347562356,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001062928292829283,
      "loss": 3.2328,
      "step": 468590
    },
    {
      "epoch": 0.18289014671178175,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010629082908290831,
      "loss": 3.3213,
      "step": 468600
    },
    {
      "epoch": 0.18290878994793994,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001062888288828883,
      "loss": 3.2063,
      "step": 468610
    },
    {
      "epoch": 0.1829274331840981,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001062868286828683,
      "loss": 3.1428,
      "step": 468620
    },
    {
      "epoch": 0.1829460764202563,
      "grad_norm": 0.75,
      "learning_rate": 0.00010628482848284828,
      "loss": 3.221,
      "step": 468630
    },
    {
      "epoch": 0.18296471965641448,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001062828282828283,
      "loss": 3.1101,
      "step": 468640
    },
    {
      "epoch": 0.18298336289257267,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010628082808280828,
      "loss": 3.2154,
      "step": 468650
    },
    {
      "epoch": 0.18300200612873085,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010627882788278829,
      "loss": 3.1533,
      "step": 468660
    },
    {
      "epoch": 0.18302064936488904,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010627682768276828,
      "loss": 3.1519,
      "step": 468670
    },
    {
      "epoch": 0.1830392926010472,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010627482748274829,
      "loss": 3.1658,
      "step": 468680
    },
    {
      "epoch": 0.1830579358372054,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010627282728272827,
      "loss": 3.2604,
      "step": 468690
    },
    {
      "epoch": 0.18307657907336358,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010627082708270828,
      "loss": 3.1693,
      "step": 468700
    },
    {
      "epoch": 0.18309522230952177,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010626882688268826,
      "loss": 3.2005,
      "step": 468710
    },
    {
      "epoch": 0.18311386554567996,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010626682668266827,
      "loss": 3.2167,
      "step": 468720
    },
    {
      "epoch": 0.18313250878183812,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010626482648264825,
      "loss": 3.2018,
      "step": 468730
    },
    {
      "epoch": 0.1831511520179963,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010626282628262826,
      "loss": 3.2937,
      "step": 468740
    },
    {
      "epoch": 0.1831697952541545,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010626082608260826,
      "loss": 3.0892,
      "step": 468750
    },
    {
      "epoch": 0.1831884384903127,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00010625882588258827,
      "loss": 3.1691,
      "step": 468760
    },
    {
      "epoch": 0.18320708172647088,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010625682568256828,
      "loss": 3.1119,
      "step": 468770
    },
    {
      "epoch": 0.18322572496262904,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010625482548254826,
      "loss": 3.1892,
      "step": 468780
    },
    {
      "epoch": 0.18324436819878723,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010625282528252827,
      "loss": 3.2875,
      "step": 468790
    },
    {
      "epoch": 0.18326301143494542,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010625082508250825,
      "loss": 3.174,
      "step": 468800
    },
    {
      "epoch": 0.1832816546711036,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010624882488248826,
      "loss": 3.138,
      "step": 468810
    },
    {
      "epoch": 0.1833002979072618,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010624682468246826,
      "loss": 3.1508,
      "step": 468820
    },
    {
      "epoch": 0.18331894114341996,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010624482448244825,
      "loss": 3.1083,
      "step": 468830
    },
    {
      "epoch": 0.18333758437957814,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010624282428242825,
      "loss": 3.2041,
      "step": 468840
    },
    {
      "epoch": 0.18335622761573633,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010624082408240826,
      "loss": 3.3177,
      "step": 468850
    },
    {
      "epoch": 0.18337487085189452,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010623882388238824,
      "loss": 3.1467,
      "step": 468860
    },
    {
      "epoch": 0.1833935140880527,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010623682368236825,
      "loss": 3.1301,
      "step": 468870
    },
    {
      "epoch": 0.1834121573242109,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010623482348234823,
      "loss": 3.197,
      "step": 468880
    },
    {
      "epoch": 0.18343080056036906,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010623282328232824,
      "loss": 3.1778,
      "step": 468890
    },
    {
      "epoch": 0.18344944379652725,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010623082308230823,
      "loss": 3.1611,
      "step": 468900
    },
    {
      "epoch": 0.18346808703268544,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010622882288228824,
      "loss": 3.1581,
      "step": 468910
    },
    {
      "epoch": 0.18348673026884363,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010622682268226822,
      "loss": 3.2112,
      "step": 468920
    },
    {
      "epoch": 0.18350537350500182,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010622482248224823,
      "loss": 3.3273,
      "step": 468930
    },
    {
      "epoch": 0.18352401674115998,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010622282228222822,
      "loss": 3.1754,
      "step": 468940
    },
    {
      "epoch": 0.18354265997731817,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010622082208220823,
      "loss": 3.2128,
      "step": 468950
    },
    {
      "epoch": 0.18356130321347636,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010621882188218821,
      "loss": 3.1777,
      "step": 468960
    },
    {
      "epoch": 0.18357994644963455,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010621682168216822,
      "loss": 3.1757,
      "step": 468970
    },
    {
      "epoch": 0.18359858968579273,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010621482148214823,
      "loss": 3.1847,
      "step": 468980
    },
    {
      "epoch": 0.1836172329219509,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010621282128212822,
      "loss": 3.2043,
      "step": 468990
    },
    {
      "epoch": 0.18363587615810908,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010621082108210823,
      "loss": 3.1777,
      "step": 469000
    },
    {
      "epoch": 0.18365451939426727,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010620882088208821,
      "loss": 3.2445,
      "step": 469010
    },
    {
      "epoch": 0.18367316263042546,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010620682068206822,
      "loss": 3.3149,
      "step": 469020
    },
    {
      "epoch": 0.18369180586658365,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001062048204820482,
      "loss": 3.3335,
      "step": 469030
    },
    {
      "epoch": 0.1837104491027418,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010620282028202821,
      "loss": 3.3365,
      "step": 469040
    },
    {
      "epoch": 0.1837290923389,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001062008200820082,
      "loss": 3.2978,
      "step": 469050
    },
    {
      "epoch": 0.1837477355750582,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001061988198819882,
      "loss": 3.1679,
      "step": 469060
    },
    {
      "epoch": 0.18376637881121638,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001061968196819682,
      "loss": 3.2357,
      "step": 469070
    },
    {
      "epoch": 0.18378502204737457,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010619481948194821,
      "loss": 3.2838,
      "step": 469080
    },
    {
      "epoch": 0.18380366528353276,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010619281928192819,
      "loss": 3.112,
      "step": 469090
    },
    {
      "epoch": 0.18382230851969092,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001061908190819082,
      "loss": 3.1648,
      "step": 469100
    },
    {
      "epoch": 0.1838409517558491,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010618881888188818,
      "loss": 3.1275,
      "step": 469110
    },
    {
      "epoch": 0.1838595949920073,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010618681868186819,
      "loss": 3.0937,
      "step": 469120
    },
    {
      "epoch": 0.18387823822816549,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010618481848184819,
      "loss": 3.3025,
      "step": 469130
    },
    {
      "epoch": 0.18389688146432367,
      "grad_norm": 0.75,
      "learning_rate": 0.0001061828182818282,
      "loss": 3.1617,
      "step": 469140
    },
    {
      "epoch": 0.18391552470048184,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010618081808180818,
      "loss": 3.1455,
      "step": 469150
    },
    {
      "epoch": 0.18393416793664003,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010617881788178819,
      "loss": 3.1423,
      "step": 469160
    },
    {
      "epoch": 0.18395281117279821,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001061768176817682,
      "loss": 3.2149,
      "step": 469170
    },
    {
      "epoch": 0.1839714544089564,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00010617481748174818,
      "loss": 3.0263,
      "step": 469180
    },
    {
      "epoch": 0.1839900976451146,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010617281728172819,
      "loss": 3.2006,
      "step": 469190
    },
    {
      "epoch": 0.18400874088127275,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010617081708170817,
      "loss": 3.2912,
      "step": 469200
    },
    {
      "epoch": 0.18402738411743094,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010616881688168818,
      "loss": 3.2254,
      "step": 469210
    },
    {
      "epoch": 0.18404602735358913,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010616681668166817,
      "loss": 3.216,
      "step": 469220
    },
    {
      "epoch": 0.18406467058974732,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010616481648164818,
      "loss": 3.2264,
      "step": 469230
    },
    {
      "epoch": 0.1840833138259055,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010616281628162817,
      "loss": 3.1294,
      "step": 469240
    },
    {
      "epoch": 0.18410195706206367,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010616081608160817,
      "loss": 3.1891,
      "step": 469250
    },
    {
      "epoch": 0.18412060029822186,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010615881588158816,
      "loss": 2.9904,
      "step": 469260
    },
    {
      "epoch": 0.18413924353438005,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010615681568156817,
      "loss": 3.2451,
      "step": 469270
    },
    {
      "epoch": 0.18415788677053824,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010615481548154815,
      "loss": 3.309,
      "step": 469280
    },
    {
      "epoch": 0.18417653000669643,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010615281528152816,
      "loss": 3.2391,
      "step": 469290
    },
    {
      "epoch": 0.1841951732428546,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010615081508150815,
      "loss": 3.2477,
      "step": 469300
    },
    {
      "epoch": 0.18421381647901278,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010614881488148816,
      "loss": 3.2495,
      "step": 469310
    },
    {
      "epoch": 0.18423245971517097,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010614681468146814,
      "loss": 3.1588,
      "step": 469320
    },
    {
      "epoch": 0.18425110295132915,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010614481448144815,
      "loss": 3.1544,
      "step": 469330
    },
    {
      "epoch": 0.18426974618748734,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010614281428142813,
      "loss": 3.1923,
      "step": 469340
    },
    {
      "epoch": 0.18428838942364553,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010614081408140814,
      "loss": 3.2792,
      "step": 469350
    },
    {
      "epoch": 0.1843070326598037,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010613881388138815,
      "loss": 3.2746,
      "step": 469360
    },
    {
      "epoch": 0.18432567589596188,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010613681368136813,
      "loss": 3.1309,
      "step": 469370
    },
    {
      "epoch": 0.18434431913212007,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010613481348134814,
      "loss": 3.182,
      "step": 469380
    },
    {
      "epoch": 0.18436296236827826,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010613281328132814,
      "loss": 3.264,
      "step": 469390
    },
    {
      "epoch": 0.18438160560443645,
      "grad_norm": 0.75,
      "learning_rate": 0.00010613081308130815,
      "loss": 3.1016,
      "step": 469400
    },
    {
      "epoch": 0.1844002488405946,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010612881288128813,
      "loss": 3.1736,
      "step": 469410
    },
    {
      "epoch": 0.1844188920767528,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010612681268126814,
      "loss": 3.1238,
      "step": 469420
    },
    {
      "epoch": 0.184437535312911,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010612481248124812,
      "loss": 3.142,
      "step": 469430
    },
    {
      "epoch": 0.18445617854906918,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010612281228122813,
      "loss": 3.3386,
      "step": 469440
    },
    {
      "epoch": 0.18447482178522737,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010612081208120813,
      "loss": 3.2175,
      "step": 469450
    },
    {
      "epoch": 0.18449346502138553,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010611881188118814,
      "loss": 3.3666,
      "step": 469460
    },
    {
      "epoch": 0.18451210825754372,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010611681168116812,
      "loss": 3.2664,
      "step": 469470
    },
    {
      "epoch": 0.1845307514937019,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010611481148114813,
      "loss": 3.1445,
      "step": 469480
    },
    {
      "epoch": 0.1845493947298601,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010611281128112811,
      "loss": 3.2292,
      "step": 469490
    },
    {
      "epoch": 0.18456803796601828,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010611081108110812,
      "loss": 3.2696,
      "step": 469500
    },
    {
      "epoch": 0.18458668120217644,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001061088108810881,
      "loss": 3.2078,
      "step": 469510
    },
    {
      "epoch": 0.18460532443833463,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010610681068106811,
      "loss": 3.1708,
      "step": 469520
    },
    {
      "epoch": 0.18462396767449282,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001061048104810481,
      "loss": 3.0921,
      "step": 469530
    },
    {
      "epoch": 0.184642610910651,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010610281028102811,
      "loss": 3.2262,
      "step": 469540
    },
    {
      "epoch": 0.1846612541468092,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001061008100810081,
      "loss": 3.2225,
      "step": 469550
    },
    {
      "epoch": 0.1846798973829674,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001060988098809881,
      "loss": 3.149,
      "step": 469560
    },
    {
      "epoch": 0.18469854061912555,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010609680968096811,
      "loss": 3.163,
      "step": 469570
    },
    {
      "epoch": 0.18471718385528374,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001060948094809481,
      "loss": 3.2329,
      "step": 469580
    },
    {
      "epoch": 0.18473582709144193,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001060928092809281,
      "loss": 3.2639,
      "step": 469590
    },
    {
      "epoch": 0.18475447032760012,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010609080908090809,
      "loss": 3.138,
      "step": 469600
    },
    {
      "epoch": 0.1847731135637583,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001060888088808881,
      "loss": 3.2215,
      "step": 469610
    },
    {
      "epoch": 0.18479175679991647,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010608680868086809,
      "loss": 3.2363,
      "step": 469620
    },
    {
      "epoch": 0.18481040003607466,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001060848084808481,
      "loss": 3.3013,
      "step": 469630
    },
    {
      "epoch": 0.18482904327223285,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010608280828082808,
      "loss": 3.1575,
      "step": 469640
    },
    {
      "epoch": 0.18484768650839103,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010608080808080809,
      "loss": 3.3744,
      "step": 469650
    },
    {
      "epoch": 0.18486632974454922,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010607880788078807,
      "loss": 3.2845,
      "step": 469660
    },
    {
      "epoch": 0.18488497298070738,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010607680768076808,
      "loss": 3.2018,
      "step": 469670
    },
    {
      "epoch": 0.18490361621686557,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010607480748074808,
      "loss": 3.1804,
      "step": 469680
    },
    {
      "epoch": 0.18492225945302376,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010607280728072808,
      "loss": 3.0828,
      "step": 469690
    },
    {
      "epoch": 0.18494090268918195,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010607080708070807,
      "loss": 3.1985,
      "step": 469700
    },
    {
      "epoch": 0.18495954592534014,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010606880688068808,
      "loss": 3.1617,
      "step": 469710
    },
    {
      "epoch": 0.1849781891614983,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010606680668066806,
      "loss": 3.2243,
      "step": 469720
    },
    {
      "epoch": 0.1849968323976565,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010606480648064807,
      "loss": 3.1491,
      "step": 469730
    },
    {
      "epoch": 0.18501547563381468,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010606280628062805,
      "loss": 3.1196,
      "step": 469740
    },
    {
      "epoch": 0.18503411886997287,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010606080608060806,
      "loss": 3.1832,
      "step": 469750
    },
    {
      "epoch": 0.18505276210613106,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010605880588058807,
      "loss": 3.2902,
      "step": 469760
    },
    {
      "epoch": 0.18507140534228925,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010605680568056807,
      "loss": 3.1841,
      "step": 469770
    },
    {
      "epoch": 0.1850900485784474,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010605480548054808,
      "loss": 3.1773,
      "step": 469780
    },
    {
      "epoch": 0.1851086918146056,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010605280528052806,
      "loss": 3.1484,
      "step": 469790
    },
    {
      "epoch": 0.18512733505076379,
      "grad_norm": 0.75,
      "learning_rate": 0.00010605080508050807,
      "loss": 3.1852,
      "step": 469800
    },
    {
      "epoch": 0.18514597828692198,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010604880488048805,
      "loss": 3.1931,
      "step": 469810
    },
    {
      "epoch": 0.18516462152308016,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010604680468046806,
      "loss": 3.1558,
      "step": 469820
    },
    {
      "epoch": 0.18518326475923833,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010604480448044804,
      "loss": 3.1181,
      "step": 469830
    },
    {
      "epoch": 0.18520190799539651,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010604280428042805,
      "loss": 3.2443,
      "step": 469840
    },
    {
      "epoch": 0.1852205512315547,
      "grad_norm": 0.875,
      "learning_rate": 0.00010604080408040805,
      "loss": 3.1483,
      "step": 469850
    },
    {
      "epoch": 0.1852391944677129,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010603880388038805,
      "loss": 3.1319,
      "step": 469860
    },
    {
      "epoch": 0.18525783770387108,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010603680368036804,
      "loss": 3.3254,
      "step": 469870
    },
    {
      "epoch": 0.18527648094002924,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010603480348034805,
      "loss": 3.075,
      "step": 469880
    },
    {
      "epoch": 0.18529512417618743,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010603280328032803,
      "loss": 3.1915,
      "step": 469890
    },
    {
      "epoch": 0.18531376741234562,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00010603080308030804,
      "loss": 3.1863,
      "step": 469900
    },
    {
      "epoch": 0.1853324106485038,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010602880288028802,
      "loss": 3.1907,
      "step": 469910
    },
    {
      "epoch": 0.185351053884662,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010602680268026803,
      "loss": 3.2469,
      "step": 469920
    },
    {
      "epoch": 0.18536969712082016,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010602480248024802,
      "loss": 3.2926,
      "step": 469930
    },
    {
      "epoch": 0.18538834035697835,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010602280228022803,
      "loss": 3.1838,
      "step": 469940
    },
    {
      "epoch": 0.18540698359313654,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010602080208020801,
      "loss": 3.1982,
      "step": 469950
    },
    {
      "epoch": 0.18542562682929473,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010601880188018802,
      "loss": 3.1531,
      "step": 469960
    },
    {
      "epoch": 0.18544427006545292,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010601680168016803,
      "loss": 3.2572,
      "step": 469970
    },
    {
      "epoch": 0.18546291330161108,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010601480148014802,
      "loss": 3.0742,
      "step": 469980
    },
    {
      "epoch": 0.18548155653776927,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010601280128012802,
      "loss": 3.2064,
      "step": 469990
    },
    {
      "epoch": 0.18550019977392745,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000106010801080108,
      "loss": 3.2842,
      "step": 470000
    },
    {
      "epoch": 0.18551884301008564,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010600880088008802,
      "loss": 3.3682,
      "step": 470010
    },
    {
      "epoch": 0.18553748624624383,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010600680068006801,
      "loss": 3.2182,
      "step": 470020
    },
    {
      "epoch": 0.18555612948240202,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010600480048004802,
      "loss": 3.2949,
      "step": 470030
    },
    {
      "epoch": 0.18557477271856018,
      "grad_norm": 0.8046875,
      "learning_rate": 0.000106002800280028,
      "loss": 3.154,
      "step": 470040
    },
    {
      "epoch": 0.18559341595471837,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010600080008000801,
      "loss": 3.2186,
      "step": 470050
    },
    {
      "epoch": 0.18561205919087656,
      "grad_norm": 0.703125,
      "learning_rate": 0.000105998799879988,
      "loss": 3.1638,
      "step": 470060
    },
    {
      "epoch": 0.18563070242703475,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000105996799679968,
      "loss": 3.068,
      "step": 470070
    },
    {
      "epoch": 0.18564934566319294,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000105994799479948,
      "loss": 3.2341,
      "step": 470080
    },
    {
      "epoch": 0.1856679888993511,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010599279927992801,
      "loss": 3.195,
      "step": 470090
    },
    {
      "epoch": 0.1856866321355093,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010599079907990799,
      "loss": 3.0947,
      "step": 470100
    },
    {
      "epoch": 0.18570527537166748,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000105988798879888,
      "loss": 3.1658,
      "step": 470110
    },
    {
      "epoch": 0.18572391860782567,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010598679867986798,
      "loss": 3.2896,
      "step": 470120
    },
    {
      "epoch": 0.18574256184398386,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010598479847984799,
      "loss": 3.1009,
      "step": 470130
    },
    {
      "epoch": 0.18576120508014202,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010598279827982797,
      "loss": 3.2847,
      "step": 470140
    },
    {
      "epoch": 0.1857798483163002,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010598079807980798,
      "loss": 3.2636,
      "step": 470150
    },
    {
      "epoch": 0.1857984915524584,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010597879787978799,
      "loss": 3.2215,
      "step": 470160
    },
    {
      "epoch": 0.18581713478861658,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010597679767976799,
      "loss": 3.1782,
      "step": 470170
    },
    {
      "epoch": 0.18583577802477477,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000105974797479748,
      "loss": 3.1416,
      "step": 470180
    },
    {
      "epoch": 0.18585442126093293,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010597279727972798,
      "loss": 3.0637,
      "step": 470190
    },
    {
      "epoch": 0.18587306449709112,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010597079707970799,
      "loss": 3.3277,
      "step": 470200
    },
    {
      "epoch": 0.1858917077332493,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010596879687968797,
      "loss": 3.1153,
      "step": 470210
    },
    {
      "epoch": 0.1859103509694075,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010596679667966798,
      "loss": 3.2233,
      "step": 470220
    },
    {
      "epoch": 0.1859289942055657,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010596479647964796,
      "loss": 3.2944,
      "step": 470230
    },
    {
      "epoch": 0.18594763744172388,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010596279627962797,
      "loss": 3.1516,
      "step": 470240
    },
    {
      "epoch": 0.18596628067788204,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010596079607960796,
      "loss": 3.2432,
      "step": 470250
    },
    {
      "epoch": 0.18598492391404023,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010595879587958797,
      "loss": 3.1975,
      "step": 470260
    },
    {
      "epoch": 0.18600356715019842,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010595679567956796,
      "loss": 3.2873,
      "step": 470270
    },
    {
      "epoch": 0.1860222103863566,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010595479547954796,
      "loss": 3.1663,
      "step": 470280
    },
    {
      "epoch": 0.1860408536225148,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010595279527952795,
      "loss": 3.1278,
      "step": 470290
    },
    {
      "epoch": 0.18605949685867296,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010595079507950796,
      "loss": 3.2142,
      "step": 470300
    },
    {
      "epoch": 0.18607814009483115,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010594879487948795,
      "loss": 3.2011,
      "step": 470310
    },
    {
      "epoch": 0.18609678333098933,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010594679467946795,
      "loss": 3.1938,
      "step": 470320
    },
    {
      "epoch": 0.18611542656714752,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010594479447944794,
      "loss": 3.223,
      "step": 470330
    },
    {
      "epoch": 0.1861340698033057,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010594279427942795,
      "loss": 3.2414,
      "step": 470340
    },
    {
      "epoch": 0.18615271303946387,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010594079407940796,
      "loss": 3.1748,
      "step": 470350
    },
    {
      "epoch": 0.18617135627562206,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010593879387938794,
      "loss": 3.2993,
      "step": 470360
    },
    {
      "epoch": 0.18618999951178025,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010593679367936795,
      "loss": 3.1396,
      "step": 470370
    },
    {
      "epoch": 0.18620864274793844,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010593479347934793,
      "loss": 3.1288,
      "step": 470380
    },
    {
      "epoch": 0.18622728598409663,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010593279327932794,
      "loss": 3.2482,
      "step": 470390
    },
    {
      "epoch": 0.1862459292202548,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010593079307930794,
      "loss": 3.1718,
      "step": 470400
    },
    {
      "epoch": 0.18626457245641298,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010592879287928795,
      "loss": 3.2105,
      "step": 470410
    },
    {
      "epoch": 0.18628321569257117,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010592679267926793,
      "loss": 3.2241,
      "step": 470420
    },
    {
      "epoch": 0.18630185892872936,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010592479247924794,
      "loss": 3.1542,
      "step": 470430
    },
    {
      "epoch": 0.18632050216488755,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010592279227922792,
      "loss": 3.2643,
      "step": 470440
    },
    {
      "epoch": 0.18633914540104574,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010592079207920793,
      "loss": 3.3265,
      "step": 470450
    },
    {
      "epoch": 0.1863577886372039,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010591879187918791,
      "loss": 3.1896,
      "step": 470460
    },
    {
      "epoch": 0.1863764318733621,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010591679167916792,
      "loss": 3.2844,
      "step": 470470
    },
    {
      "epoch": 0.18639507510952028,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010591479147914792,
      "loss": 3.2135,
      "step": 470480
    },
    {
      "epoch": 0.18641371834567846,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010591279127912793,
      "loss": 3.1984,
      "step": 470490
    },
    {
      "epoch": 0.18643236158183665,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010591079107910791,
      "loss": 3.1452,
      "step": 470500
    },
    {
      "epoch": 0.18645100481799481,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010590879087908792,
      "loss": 3.0894,
      "step": 470510
    },
    {
      "epoch": 0.186469648054153,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001059067906790679,
      "loss": 3.2673,
      "step": 470520
    },
    {
      "epoch": 0.1864882912903112,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010590479047904791,
      "loss": 3.2877,
      "step": 470530
    },
    {
      "epoch": 0.18650693452646938,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010590279027902789,
      "loss": 3.3003,
      "step": 470540
    },
    {
      "epoch": 0.18652557776262757,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001059007900790079,
      "loss": 3.3174,
      "step": 470550
    },
    {
      "epoch": 0.18654422099878573,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010589878987898791,
      "loss": 3.2549,
      "step": 470560
    },
    {
      "epoch": 0.18656286423494392,
      "grad_norm": 0.75,
      "learning_rate": 0.0001058967896789679,
      "loss": 3.2429,
      "step": 470570
    },
    {
      "epoch": 0.1865815074711021,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010589478947894791,
      "loss": 3.3124,
      "step": 470580
    },
    {
      "epoch": 0.1866001507072603,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001058927892789279,
      "loss": 3.1398,
      "step": 470590
    },
    {
      "epoch": 0.1866187939434185,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001058907890789079,
      "loss": 3.2659,
      "step": 470600
    },
    {
      "epoch": 0.18663743717957665,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010588878887888789,
      "loss": 3.2542,
      "step": 470610
    },
    {
      "epoch": 0.18665608041573484,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001058867886788679,
      "loss": 3.2612,
      "step": 470620
    },
    {
      "epoch": 0.18667472365189303,
      "grad_norm": 0.75,
      "learning_rate": 0.00010588478847884789,
      "loss": 3.18,
      "step": 470630
    },
    {
      "epoch": 0.18669336688805122,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010588278827882789,
      "loss": 3.3053,
      "step": 470640
    },
    {
      "epoch": 0.1867120101242094,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010588078807880788,
      "loss": 3.0531,
      "step": 470650
    },
    {
      "epoch": 0.18673065336036757,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010587878787878789,
      "loss": 3.293,
      "step": 470660
    },
    {
      "epoch": 0.18674929659652575,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010587678767876787,
      "loss": 3.1284,
      "step": 470670
    },
    {
      "epoch": 0.18676793983268394,
      "grad_norm": 0.75,
      "learning_rate": 0.00010587478747874788,
      "loss": 3.2532,
      "step": 470680
    },
    {
      "epoch": 0.18678658306884213,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010587278727872787,
      "loss": 3.1241,
      "step": 470690
    },
    {
      "epoch": 0.18680522630500032,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010587078707870787,
      "loss": 3.2097,
      "step": 470700
    },
    {
      "epoch": 0.1868238695411585,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010586878687868787,
      "loss": 3.3055,
      "step": 470710
    },
    {
      "epoch": 0.18684251277731667,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010586678667866788,
      "loss": 3.2523,
      "step": 470720
    },
    {
      "epoch": 0.18686115601347486,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010586478647864786,
      "loss": 3.1885,
      "step": 470730
    },
    {
      "epoch": 0.18687979924963305,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010586278627862787,
      "loss": 3.1924,
      "step": 470740
    },
    {
      "epoch": 0.18689844248579124,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010586078607860788,
      "loss": 3.1187,
      "step": 470750
    },
    {
      "epoch": 0.18691708572194943,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010585878587858786,
      "loss": 3.1326,
      "step": 470760
    },
    {
      "epoch": 0.1869357289581076,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010585678567856787,
      "loss": 3.2537,
      "step": 470770
    },
    {
      "epoch": 0.18695437219426578,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010585478547854785,
      "loss": 3.1899,
      "step": 470780
    },
    {
      "epoch": 0.18697301543042397,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010585278527852786,
      "loss": 3.2357,
      "step": 470790
    },
    {
      "epoch": 0.18699165866658216,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010585078507850786,
      "loss": 3.1621,
      "step": 470800
    },
    {
      "epoch": 0.18701030190274034,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010584878487848787,
      "loss": 3.2341,
      "step": 470810
    },
    {
      "epoch": 0.1870289451388985,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010584678467846785,
      "loss": 3.0596,
      "step": 470820
    },
    {
      "epoch": 0.1870475883750567,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010584478447844786,
      "loss": 3.2656,
      "step": 470830
    },
    {
      "epoch": 0.18706623161121488,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010584278427842784,
      "loss": 3.1876,
      "step": 470840
    },
    {
      "epoch": 0.18708487484737307,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010584078407840785,
      "loss": 3.1417,
      "step": 470850
    },
    {
      "epoch": 0.18710351808353126,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010583878387838783,
      "loss": 3.2678,
      "step": 470860
    },
    {
      "epoch": 0.18712216131968942,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010583678367836784,
      "loss": 3.2218,
      "step": 470870
    },
    {
      "epoch": 0.1871408045558476,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010583478347834784,
      "loss": 3.1971,
      "step": 470880
    },
    {
      "epoch": 0.1871594477920058,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010583278327832784,
      "loss": 3.1634,
      "step": 470890
    },
    {
      "epoch": 0.187178091028164,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010583078307830783,
      "loss": 3.232,
      "step": 470900
    },
    {
      "epoch": 0.18719673426432218,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010582878287828784,
      "loss": 3.2026,
      "step": 470910
    },
    {
      "epoch": 0.18721537750048037,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010582678267826782,
      "loss": 3.1299,
      "step": 470920
    },
    {
      "epoch": 0.18723402073663853,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010582478247824783,
      "loss": 3.1941,
      "step": 470930
    },
    {
      "epoch": 0.18725266397279672,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010582278227822784,
      "loss": 3.2617,
      "step": 470940
    },
    {
      "epoch": 0.1872713072089549,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010582078207820783,
      "loss": 3.152,
      "step": 470950
    },
    {
      "epoch": 0.1872899504451131,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010581878187818783,
      "loss": 3.1469,
      "step": 470960
    },
    {
      "epoch": 0.18730859368127128,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010581678167816782,
      "loss": 3.1817,
      "step": 470970
    },
    {
      "epoch": 0.18732723691742945,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010581478147814783,
      "loss": 3.257,
      "step": 470980
    },
    {
      "epoch": 0.18734588015358764,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010581278127812781,
      "loss": 3.176,
      "step": 470990
    },
    {
      "epoch": 0.18736452338974582,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010581078107810782,
      "loss": 3.3093,
      "step": 471000
    },
    {
      "epoch": 0.187383166625904,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001058087808780878,
      "loss": 3.1969,
      "step": 471010
    },
    {
      "epoch": 0.1874018098620622,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010580678067806781,
      "loss": 3.1822,
      "step": 471020
    },
    {
      "epoch": 0.18742045309822036,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010580478047804781,
      "loss": 3.1921,
      "step": 471030
    },
    {
      "epoch": 0.18743909633437855,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010580278027802782,
      "loss": 3.0907,
      "step": 471040
    },
    {
      "epoch": 0.18745773957053674,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001058007800780078,
      "loss": 3.2423,
      "step": 471050
    },
    {
      "epoch": 0.18747638280669493,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010579877987798781,
      "loss": 3.3229,
      "step": 471060
    },
    {
      "epoch": 0.18749502604285312,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010579677967796779,
      "loss": 3.2383,
      "step": 471070
    },
    {
      "epoch": 0.18751366927901128,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001057947794779478,
      "loss": 3.1491,
      "step": 471080
    },
    {
      "epoch": 0.18753231251516947,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010579277927792778,
      "loss": 3.0964,
      "step": 471090
    },
    {
      "epoch": 0.18755095575132766,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010579077907790779,
      "loss": 3.2153,
      "step": 471100
    },
    {
      "epoch": 0.18756959898748585,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010578877887788779,
      "loss": 3.2517,
      "step": 471110
    },
    {
      "epoch": 0.18758824222364404,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001057867786778678,
      "loss": 3.23,
      "step": 471120
    },
    {
      "epoch": 0.18760688545980223,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010578477847784778,
      "loss": 3.2145,
      "step": 471130
    },
    {
      "epoch": 0.1876255286959604,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010578277827782779,
      "loss": 3.3151,
      "step": 471140
    },
    {
      "epoch": 0.18764417193211858,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001057807780778078,
      "loss": 3.1885,
      "step": 471150
    },
    {
      "epoch": 0.18766281516827676,
      "grad_norm": 0.75,
      "learning_rate": 0.00010577877787778778,
      "loss": 3.1638,
      "step": 471160
    },
    {
      "epoch": 0.18768145840443495,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010577677767776779,
      "loss": 3.1827,
      "step": 471170
    },
    {
      "epoch": 0.18770010164059314,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010577477747774777,
      "loss": 3.1859,
      "step": 471180
    },
    {
      "epoch": 0.1877187448767513,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010577277727772778,
      "loss": 3.1417,
      "step": 471190
    },
    {
      "epoch": 0.1877373881129095,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010577077707770778,
      "loss": 3.1717,
      "step": 471200
    },
    {
      "epoch": 0.18775603134906768,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010576877687768778,
      "loss": 3.2258,
      "step": 471210
    },
    {
      "epoch": 0.18777467458522587,
      "grad_norm": 0.53125,
      "learning_rate": 0.00010576677667766777,
      "loss": 3.124,
      "step": 471220
    },
    {
      "epoch": 0.18779331782138406,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010576477647764778,
      "loss": 3.1934,
      "step": 471230
    },
    {
      "epoch": 0.18781196105754222,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010576277627762776,
      "loss": 3.2252,
      "step": 471240
    },
    {
      "epoch": 0.1878306042937004,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010576077607760777,
      "loss": 3.0988,
      "step": 471250
    },
    {
      "epoch": 0.1878492475298586,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010575877587758776,
      "loss": 3.1785,
      "step": 471260
    },
    {
      "epoch": 0.1878678907660168,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010575677567756777,
      "loss": 3.3072,
      "step": 471270
    },
    {
      "epoch": 0.18788653400217498,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010575477547754775,
      "loss": 3.1409,
      "step": 471280
    },
    {
      "epoch": 0.18790517723833314,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010575277527752776,
      "loss": 3.3493,
      "step": 471290
    },
    {
      "epoch": 0.18792382047449133,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010575077507750775,
      "loss": 3.1986,
      "step": 471300
    },
    {
      "epoch": 0.18794246371064952,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010574877487748775,
      "loss": 3.2603,
      "step": 471310
    },
    {
      "epoch": 0.1879611069468077,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010574677467746774,
      "loss": 3.1487,
      "step": 471320
    },
    {
      "epoch": 0.1879797501829659,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010574477447744775,
      "loss": 3.1316,
      "step": 471330
    },
    {
      "epoch": 0.18799839341912405,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010574277427742775,
      "loss": 3.3163,
      "step": 471340
    },
    {
      "epoch": 0.18801703665528224,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010574077407740775,
      "loss": 3.276,
      "step": 471350
    },
    {
      "epoch": 0.18803567989144043,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010573877387738776,
      "loss": 3.175,
      "step": 471360
    },
    {
      "epoch": 0.18805432312759862,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010573677367736774,
      "loss": 3.251,
      "step": 471370
    },
    {
      "epoch": 0.1880729663637568,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010573477347734775,
      "loss": 3.2249,
      "step": 471380
    },
    {
      "epoch": 0.188091609599915,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010573277327732773,
      "loss": 3.1023,
      "step": 471390
    },
    {
      "epoch": 0.18811025283607316,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010573077307730774,
      "loss": 3.269,
      "step": 471400
    },
    {
      "epoch": 0.18812889607223135,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010572877287728772,
      "loss": 3.2392,
      "step": 471410
    },
    {
      "epoch": 0.18814753930838954,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010572677267726773,
      "loss": 3.1769,
      "step": 471420
    },
    {
      "epoch": 0.18816618254454773,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010572477247724773,
      "loss": 3.3033,
      "step": 471430
    },
    {
      "epoch": 0.18818482578070592,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010572277227722774,
      "loss": 3.3032,
      "step": 471440
    },
    {
      "epoch": 0.18820346901686408,
      "grad_norm": 0.75,
      "learning_rate": 0.00010572077207720772,
      "loss": 3.2738,
      "step": 471450
    },
    {
      "epoch": 0.18822211225302227,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010571877187718773,
      "loss": 3.215,
      "step": 471460
    },
    {
      "epoch": 0.18824075548918046,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010571677167716771,
      "loss": 3.1502,
      "step": 471470
    },
    {
      "epoch": 0.18825939872533864,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010571477147714772,
      "loss": 3.1272,
      "step": 471480
    },
    {
      "epoch": 0.18827804196149683,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001057127712771277,
      "loss": 3.1186,
      "step": 471490
    },
    {
      "epoch": 0.188296685197655,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010571077107710771,
      "loss": 3.2484,
      "step": 471500
    },
    {
      "epoch": 0.18831532843381318,
      "grad_norm": 0.75,
      "learning_rate": 0.00010570877087708771,
      "loss": 3.2614,
      "step": 471510
    },
    {
      "epoch": 0.18833397166997137,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010570677067706772,
      "loss": 3.1641,
      "step": 471520
    },
    {
      "epoch": 0.18835261490612956,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001057047704770477,
      "loss": 3.2706,
      "step": 471530
    },
    {
      "epoch": 0.18837125814228775,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010570277027702771,
      "loss": 3.3517,
      "step": 471540
    },
    {
      "epoch": 0.1883899013784459,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010570077007700772,
      "loss": 3.2134,
      "step": 471550
    },
    {
      "epoch": 0.1884085446146041,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001056987698769877,
      "loss": 3.1582,
      "step": 471560
    },
    {
      "epoch": 0.1884271878507623,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010569676967696771,
      "loss": 3.1973,
      "step": 471570
    },
    {
      "epoch": 0.18844583108692048,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001056947694769477,
      "loss": 3.1363,
      "step": 471580
    },
    {
      "epoch": 0.18846447432307867,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001056927692769277,
      "loss": 3.3429,
      "step": 471590
    },
    {
      "epoch": 0.18848311755923686,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001056907690769077,
      "loss": 3.2616,
      "step": 471600
    },
    {
      "epoch": 0.18850176079539502,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001056887688768877,
      "loss": 3.2803,
      "step": 471610
    },
    {
      "epoch": 0.1885204040315532,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010568676867686769,
      "loss": 3.2298,
      "step": 471620
    },
    {
      "epoch": 0.1885390472677114,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001056847684768477,
      "loss": 3.0992,
      "step": 471630
    },
    {
      "epoch": 0.18855769050386958,
      "grad_norm": 1.140625,
      "learning_rate": 0.00010568276827682768,
      "loss": 3.1998,
      "step": 471640
    },
    {
      "epoch": 0.18857633374002777,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010568076807680769,
      "loss": 3.4188,
      "step": 471650
    },
    {
      "epoch": 0.18859497697618594,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010567876787678768,
      "loss": 3.3387,
      "step": 471660
    },
    {
      "epoch": 0.18861362021234412,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010567676767676769,
      "loss": 3.2659,
      "step": 471670
    },
    {
      "epoch": 0.1886322634485023,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010567476747674767,
      "loss": 3.08,
      "step": 471680
    },
    {
      "epoch": 0.1886509066846605,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010567276727672768,
      "loss": 3.2525,
      "step": 471690
    },
    {
      "epoch": 0.1886695499208187,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010567076707670766,
      "loss": 3.257,
      "step": 471700
    },
    {
      "epoch": 0.18868819315697685,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010566876687668767,
      "loss": 3.3682,
      "step": 471710
    },
    {
      "epoch": 0.18870683639313504,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010566676667666766,
      "loss": 3.1593,
      "step": 471720
    },
    {
      "epoch": 0.18872547962929323,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010566476647664766,
      "loss": 3.4033,
      "step": 471730
    },
    {
      "epoch": 0.18874412286545142,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010566276627662767,
      "loss": 3.2239,
      "step": 471740
    },
    {
      "epoch": 0.1887627661016096,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010566076607660767,
      "loss": 3.1796,
      "step": 471750
    },
    {
      "epoch": 0.18878140933776777,
      "grad_norm": 0.625,
      "learning_rate": 0.00010565876587658768,
      "loss": 3.2685,
      "step": 471760
    },
    {
      "epoch": 0.18880005257392596,
      "grad_norm": 0.75,
      "learning_rate": 0.00010565676567656766,
      "loss": 3.2837,
      "step": 471770
    },
    {
      "epoch": 0.18881869581008415,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010565476547654767,
      "loss": 3.2119,
      "step": 471780
    },
    {
      "epoch": 0.18883733904624234,
      "grad_norm": 0.75,
      "learning_rate": 0.00010565276527652765,
      "loss": 3.1849,
      "step": 471790
    },
    {
      "epoch": 0.18885598228240053,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010565076507650766,
      "loss": 3.2232,
      "step": 471800
    },
    {
      "epoch": 0.18887462551855871,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010564876487648764,
      "loss": 3.3306,
      "step": 471810
    },
    {
      "epoch": 0.18889326875471688,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010564676467646765,
      "loss": 3.0123,
      "step": 471820
    },
    {
      "epoch": 0.18891191199087506,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010564476447644765,
      "loss": 3.2168,
      "step": 471830
    },
    {
      "epoch": 0.18893055522703325,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010564276427642766,
      "loss": 3.2153,
      "step": 471840
    },
    {
      "epoch": 0.18894919846319144,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010564076407640764,
      "loss": 3.2548,
      "step": 471850
    },
    {
      "epoch": 0.18896784169934963,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010563876387638765,
      "loss": 3.1803,
      "step": 471860
    },
    {
      "epoch": 0.1889864849355078,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010563676367636763,
      "loss": 3.0496,
      "step": 471870
    },
    {
      "epoch": 0.18900512817166598,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010563476347634764,
      "loss": 3.2344,
      "step": 471880
    },
    {
      "epoch": 0.18902377140782417,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010563276327632763,
      "loss": 3.26,
      "step": 471890
    },
    {
      "epoch": 0.18904241464398236,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010563076307630764,
      "loss": 3.2396,
      "step": 471900
    },
    {
      "epoch": 0.18906105788014055,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010562876287628763,
      "loss": 3.226,
      "step": 471910
    },
    {
      "epoch": 0.1890797011162987,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010562676267626763,
      "loss": 3.1505,
      "step": 471920
    },
    {
      "epoch": 0.1890983443524569,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010562476247624764,
      "loss": 3.2462,
      "step": 471930
    },
    {
      "epoch": 0.1891169875886151,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010562276227622763,
      "loss": 3.2065,
      "step": 471940
    },
    {
      "epoch": 0.18913563082477328,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010562076207620764,
      "loss": 3.1791,
      "step": 471950
    },
    {
      "epoch": 0.18915427406093147,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010561876187618762,
      "loss": 3.1601,
      "step": 471960
    },
    {
      "epoch": 0.18917291729708963,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010561676167616763,
      "loss": 3.2084,
      "step": 471970
    },
    {
      "epoch": 0.18919156053324782,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010561476147614762,
      "loss": 3.189,
      "step": 471980
    },
    {
      "epoch": 0.189210203769406,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010561276127612763,
      "loss": 3.1997,
      "step": 471990
    },
    {
      "epoch": 0.1892288470055642,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010561076107610761,
      "loss": 3.1745,
      "step": 472000
    },
    {
      "epoch": 0.18924749024172238,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010560876087608762,
      "loss": 3.2338,
      "step": 472010
    },
    {
      "epoch": 0.18926613347788057,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001056067606760676,
      "loss": 3.2842,
      "step": 472020
    },
    {
      "epoch": 0.18928477671403873,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010560476047604761,
      "loss": 3.3185,
      "step": 472030
    },
    {
      "epoch": 0.18930341995019692,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001056027602760276,
      "loss": 3.0868,
      "step": 472040
    },
    {
      "epoch": 0.1893220631863551,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001056007600760076,
      "loss": 3.1598,
      "step": 472050
    },
    {
      "epoch": 0.1893407064225133,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001055987598759876,
      "loss": 3.247,
      "step": 472060
    },
    {
      "epoch": 0.1893593496586715,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010559675967596761,
      "loss": 3.5138,
      "step": 472070
    },
    {
      "epoch": 0.18937799289482965,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010559475947594759,
      "loss": 3.1579,
      "step": 472080
    },
    {
      "epoch": 0.18939663613098784,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001055927592759276,
      "loss": 3.1592,
      "step": 472090
    },
    {
      "epoch": 0.18941527936714603,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010559075907590758,
      "loss": 3.063,
      "step": 472100
    },
    {
      "epoch": 0.18943392260330422,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010558875887588759,
      "loss": 3.2699,
      "step": 472110
    },
    {
      "epoch": 0.1894525658394624,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010558675867586759,
      "loss": 3.1598,
      "step": 472120
    },
    {
      "epoch": 0.18947120907562057,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010558475847584758,
      "loss": 3.3043,
      "step": 472130
    },
    {
      "epoch": 0.18948985231177876,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010558275827582759,
      "loss": 3.2485,
      "step": 472140
    },
    {
      "epoch": 0.18950849554793694,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010558075807580759,
      "loss": 3.2986,
      "step": 472150
    },
    {
      "epoch": 0.18952713878409513,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001055787578757876,
      "loss": 3.2603,
      "step": 472160
    },
    {
      "epoch": 0.18954578202025332,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010557675767576758,
      "loss": 3.1769,
      "step": 472170
    },
    {
      "epoch": 0.18956442525641148,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010557475747574759,
      "loss": 3.269,
      "step": 472180
    },
    {
      "epoch": 0.18958306849256967,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010557275727572757,
      "loss": 3.1333,
      "step": 472190
    },
    {
      "epoch": 0.18960171172872786,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010557075707570758,
      "loss": 3.1975,
      "step": 472200
    },
    {
      "epoch": 0.18962035496488605,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010556875687568757,
      "loss": 3.2426,
      "step": 472210
    },
    {
      "epoch": 0.18963899820104424,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010556675667566758,
      "loss": 3.2366,
      "step": 472220
    },
    {
      "epoch": 0.1896576414372024,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010556475647564757,
      "loss": 3.1579,
      "step": 472230
    },
    {
      "epoch": 0.1896762846733606,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010556275627562758,
      "loss": 3.0827,
      "step": 472240
    },
    {
      "epoch": 0.18969492790951878,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010556075607560756,
      "loss": 3.2301,
      "step": 472250
    },
    {
      "epoch": 0.18971357114567697,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010555875587558757,
      "loss": 3.2819,
      "step": 472260
    },
    {
      "epoch": 0.18973221438183516,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010555675567556755,
      "loss": 3.2866,
      "step": 472270
    },
    {
      "epoch": 0.18975085761799335,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010555475547554756,
      "loss": 3.2018,
      "step": 472280
    },
    {
      "epoch": 0.1897695008541515,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010555275527552755,
      "loss": 3.2728,
      "step": 472290
    },
    {
      "epoch": 0.1897881440903097,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010555075507550756,
      "loss": 3.1732,
      "step": 472300
    },
    {
      "epoch": 0.18980678732646789,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010554875487548754,
      "loss": 3.3125,
      "step": 472310
    },
    {
      "epoch": 0.18982543056262607,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010554675467546755,
      "loss": 3.1982,
      "step": 472320
    },
    {
      "epoch": 0.18984407379878426,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010554475447544756,
      "loss": 3.2361,
      "step": 472330
    },
    {
      "epoch": 0.18986271703494242,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010554275427542754,
      "loss": 3.1456,
      "step": 472340
    },
    {
      "epoch": 0.1898813602711006,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010554075407540755,
      "loss": 3.2074,
      "step": 472350
    },
    {
      "epoch": 0.1899000035072588,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010553875387538754,
      "loss": 3.2218,
      "step": 472360
    },
    {
      "epoch": 0.189918646743417,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010553675367536754,
      "loss": 3.2353,
      "step": 472370
    },
    {
      "epoch": 0.18993728997957518,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010553475347534754,
      "loss": 3.1592,
      "step": 472380
    },
    {
      "epoch": 0.18995593321573334,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010553275327532755,
      "loss": 3.186,
      "step": 472390
    },
    {
      "epoch": 0.18997457645189153,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010553075307530753,
      "loss": 3.1869,
      "step": 472400
    },
    {
      "epoch": 0.18999321968804972,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010552875287528754,
      "loss": 3.1762,
      "step": 472410
    },
    {
      "epoch": 0.1900118629242079,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010552675267526752,
      "loss": 3.2201,
      "step": 472420
    },
    {
      "epoch": 0.1900305061603661,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010552475247524753,
      "loss": 3.2276,
      "step": 472430
    },
    {
      "epoch": 0.19004914939652426,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010552275227522753,
      "loss": 3.1734,
      "step": 472440
    },
    {
      "epoch": 0.19006779263268245,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010552075207520752,
      "loss": 3.2434,
      "step": 472450
    },
    {
      "epoch": 0.19008643586884064,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010551875187518752,
      "loss": 3.1992,
      "step": 472460
    },
    {
      "epoch": 0.19010507910499883,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010551675167516753,
      "loss": 3.2033,
      "step": 472470
    },
    {
      "epoch": 0.19012372234115701,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010551475147514751,
      "loss": 3.0747,
      "step": 472480
    },
    {
      "epoch": 0.1901423655773152,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010551275127512752,
      "loss": 3.09,
      "step": 472490
    },
    {
      "epoch": 0.19016100881347336,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001055107510751075,
      "loss": 3.2651,
      "step": 472500
    },
    {
      "epoch": 0.19017965204963155,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010550875087508751,
      "loss": 3.2026,
      "step": 472510
    },
    {
      "epoch": 0.19019829528578974,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010550675067506752,
      "loss": 3.3383,
      "step": 472520
    },
    {
      "epoch": 0.19021693852194793,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010550475047504752,
      "loss": 3.2754,
      "step": 472530
    },
    {
      "epoch": 0.19023558175810612,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010550275027502752,
      "loss": 3.1989,
      "step": 472540
    },
    {
      "epoch": 0.19025422499426428,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001055007500750075,
      "loss": 3.2082,
      "step": 472550
    },
    {
      "epoch": 0.19027286823042247,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010549874987498752,
      "loss": 3.1118,
      "step": 472560
    },
    {
      "epoch": 0.19029151146658066,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001054967496749675,
      "loss": 3.2289,
      "step": 472570
    },
    {
      "epoch": 0.19031015470273885,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001054947494749475,
      "loss": 3.2404,
      "step": 472580
    },
    {
      "epoch": 0.19032879793889704,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010549274927492749,
      "loss": 3.3937,
      "step": 472590
    },
    {
      "epoch": 0.1903474411750552,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001054907490749075,
      "loss": 3.2122,
      "step": 472600
    },
    {
      "epoch": 0.1903660844112134,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001054887488748875,
      "loss": 3.2125,
      "step": 472610
    },
    {
      "epoch": 0.19038472764737158,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001054867486748675,
      "loss": 3.0693,
      "step": 472620
    },
    {
      "epoch": 0.19040337088352977,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010548474847484748,
      "loss": 3.2438,
      "step": 472630
    },
    {
      "epoch": 0.19042201411968795,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001054827482748275,
      "loss": 3.1909,
      "step": 472640
    },
    {
      "epoch": 0.19044065735584612,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010548074807480748,
      "loss": 3.1905,
      "step": 472650
    },
    {
      "epoch": 0.1904593005920043,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010547874787478748,
      "loss": 3.1292,
      "step": 472660
    },
    {
      "epoch": 0.1904779438281625,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010547674767476747,
      "loss": 3.0887,
      "step": 472670
    },
    {
      "epoch": 0.19049658706432068,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010547474747474748,
      "loss": 3.1118,
      "step": 472680
    },
    {
      "epoch": 0.19051523030047887,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010547274727472747,
      "loss": 3.1489,
      "step": 472690
    },
    {
      "epoch": 0.19053387353663706,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010547074707470748,
      "loss": 3.3134,
      "step": 472700
    },
    {
      "epoch": 0.19055251677279522,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010546874687468746,
      "loss": 3.4202,
      "step": 472710
    },
    {
      "epoch": 0.1905711600089534,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010546674667466747,
      "loss": 3.1726,
      "step": 472720
    },
    {
      "epoch": 0.1905898032451116,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010546474647464748,
      "loss": 3.1474,
      "step": 472730
    },
    {
      "epoch": 0.1906084464812698,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010546274627462746,
      "loss": 3.1981,
      "step": 472740
    },
    {
      "epoch": 0.19062708971742798,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010546074607460747,
      "loss": 3.1632,
      "step": 472750
    },
    {
      "epoch": 0.19064573295358614,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010545874587458747,
      "loss": 3.3104,
      "step": 472760
    },
    {
      "epoch": 0.19066437618974433,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010545674567456746,
      "loss": 3.2486,
      "step": 472770
    },
    {
      "epoch": 0.19068301942590252,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010545474547454746,
      "loss": 3.2511,
      "step": 472780
    },
    {
      "epoch": 0.1907016626620607,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010545274527452747,
      "loss": 3.3401,
      "step": 472790
    },
    {
      "epoch": 0.1907203058982189,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010545074507450745,
      "loss": 3.1415,
      "step": 472800
    },
    {
      "epoch": 0.19073894913437706,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010544874487448746,
      "loss": 3.0071,
      "step": 472810
    },
    {
      "epoch": 0.19075759237053524,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010544674467446744,
      "loss": 3.2357,
      "step": 472820
    },
    {
      "epoch": 0.19077623560669343,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010544474447444745,
      "loss": 3.2659,
      "step": 472830
    },
    {
      "epoch": 0.19079487884285162,
      "grad_norm": 1.15625,
      "learning_rate": 0.00010544274427442745,
      "loss": 3.3251,
      "step": 472840
    },
    {
      "epoch": 0.1908135220790098,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010544074407440746,
      "loss": 3.1698,
      "step": 472850
    },
    {
      "epoch": 0.19083216531516797,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010543874387438744,
      "loss": 3.1204,
      "step": 472860
    },
    {
      "epoch": 0.19085080855132616,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010543674367436745,
      "loss": 3.2163,
      "step": 472870
    },
    {
      "epoch": 0.19086945178748435,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010543474347434743,
      "loss": 3.2808,
      "step": 472880
    },
    {
      "epoch": 0.19088809502364254,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010543274327432744,
      "loss": 3.2607,
      "step": 472890
    },
    {
      "epoch": 0.19090673825980073,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010543074307430742,
      "loss": 3.149,
      "step": 472900
    },
    {
      "epoch": 0.1909253814959589,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010542874287428743,
      "loss": 3.3195,
      "step": 472910
    },
    {
      "epoch": 0.19094402473211708,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010542674267426744,
      "loss": 3.1987,
      "step": 472920
    },
    {
      "epoch": 0.19096266796827527,
      "grad_norm": 0.96875,
      "learning_rate": 0.00010542474247424743,
      "loss": 3.1593,
      "step": 472930
    },
    {
      "epoch": 0.19098131120443346,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010542274227422744,
      "loss": 3.0805,
      "step": 472940
    },
    {
      "epoch": 0.19099995444059165,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010542074207420742,
      "loss": 3.2494,
      "step": 472950
    },
    {
      "epoch": 0.19101859767674984,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010541874187418743,
      "loss": 3.3174,
      "step": 472960
    },
    {
      "epoch": 0.191037240912908,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010541674167416742,
      "loss": 3.2483,
      "step": 472970
    },
    {
      "epoch": 0.19105588414906619,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010541474147414743,
      "loss": 3.1279,
      "step": 472980
    },
    {
      "epoch": 0.19107452738522437,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010541274127412741,
      "loss": 3.0965,
      "step": 472990
    },
    {
      "epoch": 0.19109317062138256,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010541074107410742,
      "loss": 3.2735,
      "step": 473000
    },
    {
      "epoch": 0.19111181385754075,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010540874087408741,
      "loss": 3.1418,
      "step": 473010
    },
    {
      "epoch": 0.1911304570936989,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010540674067406742,
      "loss": 3.2046,
      "step": 473020
    },
    {
      "epoch": 0.1911491003298571,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001054047404740474,
      "loss": 3.1458,
      "step": 473030
    },
    {
      "epoch": 0.1911677435660153,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010540274027402741,
      "loss": 3.2891,
      "step": 473040
    },
    {
      "epoch": 0.19118638680217348,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001054007400740074,
      "loss": 3.0833,
      "step": 473050
    },
    {
      "epoch": 0.19120503003833167,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001053987398739874,
      "loss": 3.1192,
      "step": 473060
    },
    {
      "epoch": 0.19122367327448983,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001053967396739674,
      "loss": 3.2023,
      "step": 473070
    },
    {
      "epoch": 0.19124231651064802,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001053947394739474,
      "loss": 3.206,
      "step": 473080
    },
    {
      "epoch": 0.1912609597468062,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010539273927392739,
      "loss": 3.1277,
      "step": 473090
    },
    {
      "epoch": 0.1912796029829644,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001053907390739074,
      "loss": 3.2449,
      "step": 473100
    },
    {
      "epoch": 0.1912982462191226,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010538873887388741,
      "loss": 3.2676,
      "step": 473110
    },
    {
      "epoch": 0.19131688945528075,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010538673867386739,
      "loss": 3.2694,
      "step": 473120
    },
    {
      "epoch": 0.19133553269143894,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001053847384738474,
      "loss": 3.1523,
      "step": 473130
    },
    {
      "epoch": 0.19135417592759713,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010538273827382738,
      "loss": 3.2412,
      "step": 473140
    },
    {
      "epoch": 0.19137281916375531,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010538073807380739,
      "loss": 3.2981,
      "step": 473150
    },
    {
      "epoch": 0.1913914623999135,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010537873787378739,
      "loss": 3.2351,
      "step": 473160
    },
    {
      "epoch": 0.1914101056360717,
      "grad_norm": 0.75,
      "learning_rate": 0.0001053767376737674,
      "loss": 3.0605,
      "step": 473170
    },
    {
      "epoch": 0.19142874887222985,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010537473747374738,
      "loss": 3.1988,
      "step": 473180
    },
    {
      "epoch": 0.19144739210838804,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010537273727372739,
      "loss": 3.0849,
      "step": 473190
    },
    {
      "epoch": 0.19146603534454623,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010537073707370737,
      "loss": 3.2747,
      "step": 473200
    },
    {
      "epoch": 0.19148467858070442,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010536873687368738,
      "loss": 3.3289,
      "step": 473210
    },
    {
      "epoch": 0.1915033218168626,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010536673667366736,
      "loss": 3.3001,
      "step": 473220
    },
    {
      "epoch": 0.19152196505302077,
      "grad_norm": 0.75,
      "learning_rate": 0.00010536473647364737,
      "loss": 3.2363,
      "step": 473230
    },
    {
      "epoch": 0.19154060828917896,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010536273627362736,
      "loss": 3.1504,
      "step": 473240
    },
    {
      "epoch": 0.19155925152533715,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010536073607360737,
      "loss": 3.318,
      "step": 473250
    },
    {
      "epoch": 0.19157789476149534,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010535873587358736,
      "loss": 3.1124,
      "step": 473260
    },
    {
      "epoch": 0.19159653799765353,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010535673567356737,
      "loss": 3.1451,
      "step": 473270
    },
    {
      "epoch": 0.1916151812338117,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010535473547354735,
      "loss": 3.1856,
      "step": 473280
    },
    {
      "epoch": 0.19163382446996988,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010535273527352736,
      "loss": 3.263,
      "step": 473290
    },
    {
      "epoch": 0.19165246770612807,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010535073507350734,
      "loss": 3.2594,
      "step": 473300
    },
    {
      "epoch": 0.19167111094228625,
      "grad_norm": 0.75,
      "learning_rate": 0.00010534873487348735,
      "loss": 3.254,
      "step": 473310
    },
    {
      "epoch": 0.19168975417844444,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010534673467346736,
      "loss": 3.2074,
      "step": 473320
    },
    {
      "epoch": 0.1917083974146026,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010534473447344735,
      "loss": 3.3719,
      "step": 473330
    },
    {
      "epoch": 0.1917270406507608,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010534273427342736,
      "loss": 3.2019,
      "step": 473340
    },
    {
      "epoch": 0.19174568388691898,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010534073407340734,
      "loss": 3.2268,
      "step": 473350
    },
    {
      "epoch": 0.19176432712307717,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010533873387338735,
      "loss": 3.2643,
      "step": 473360
    },
    {
      "epoch": 0.19178297035923536,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010533673367336733,
      "loss": 3.098,
      "step": 473370
    },
    {
      "epoch": 0.19180161359539355,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010533473347334734,
      "loss": 3.4077,
      "step": 473380
    },
    {
      "epoch": 0.1918202568315517,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010533273327332734,
      "loss": 3.2696,
      "step": 473390
    },
    {
      "epoch": 0.1918389000677099,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010533073307330734,
      "loss": 3.1729,
      "step": 473400
    },
    {
      "epoch": 0.1918575433038681,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010532873287328733,
      "loss": 3.1565,
      "step": 473410
    },
    {
      "epoch": 0.19187618654002628,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010532673267326734,
      "loss": 3.191,
      "step": 473420
    },
    {
      "epoch": 0.19189482977618447,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010532473247324732,
      "loss": 3.1943,
      "step": 473430
    },
    {
      "epoch": 0.19191347301234263,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010532273227322733,
      "loss": 3.1438,
      "step": 473440
    },
    {
      "epoch": 0.19193211624850082,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010532073207320731,
      "loss": 3.1693,
      "step": 473450
    },
    {
      "epoch": 0.191950759484659,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010531873187318732,
      "loss": 3.169,
      "step": 473460
    },
    {
      "epoch": 0.1919694027208172,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010531673167316732,
      "loss": 3.3549,
      "step": 473470
    },
    {
      "epoch": 0.19198804595697538,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010531473147314733,
      "loss": 3.3154,
      "step": 473480
    },
    {
      "epoch": 0.19200668919313355,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010531273127312731,
      "loss": 3.1715,
      "step": 473490
    },
    {
      "epoch": 0.19202533242929173,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010531073107310732,
      "loss": 3.32,
      "step": 473500
    },
    {
      "epoch": 0.19204397566544992,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010530873087308733,
      "loss": 3.2294,
      "step": 473510
    },
    {
      "epoch": 0.1920626189016081,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010530673067306731,
      "loss": 3.2318,
      "step": 473520
    },
    {
      "epoch": 0.1920812621377663,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010530473047304732,
      "loss": 3.1084,
      "step": 473530
    },
    {
      "epoch": 0.19209990537392446,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001053027302730273,
      "loss": 3.0664,
      "step": 473540
    },
    {
      "epoch": 0.19211854861008265,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010530073007300731,
      "loss": 3.2771,
      "step": 473550
    },
    {
      "epoch": 0.19213719184624084,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001052987298729873,
      "loss": 3.2036,
      "step": 473560
    },
    {
      "epoch": 0.19215583508239903,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010529672967296731,
      "loss": 3.222,
      "step": 473570
    },
    {
      "epoch": 0.19217447831855722,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001052947294729473,
      "loss": 3.2329,
      "step": 473580
    },
    {
      "epoch": 0.19219312155471538,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001052927292729273,
      "loss": 3.3694,
      "step": 473590
    },
    {
      "epoch": 0.19221176479087357,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010529072907290729,
      "loss": 3.3863,
      "step": 473600
    },
    {
      "epoch": 0.19223040802703176,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001052887288728873,
      "loss": 3.1945,
      "step": 473610
    },
    {
      "epoch": 0.19224905126318995,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010528672867286728,
      "loss": 3.1813,
      "step": 473620
    },
    {
      "epoch": 0.19226769449934814,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010528472847284729,
      "loss": 3.1047,
      "step": 473630
    },
    {
      "epoch": 0.19228633773550632,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010528272827282728,
      "loss": 3.24,
      "step": 473640
    },
    {
      "epoch": 0.19230498097166449,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010528072807280729,
      "loss": 3.2925,
      "step": 473650
    },
    {
      "epoch": 0.19232362420782267,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010527872787278727,
      "loss": 3.2568,
      "step": 473660
    },
    {
      "epoch": 0.19234226744398086,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010527672767276728,
      "loss": 3.2097,
      "step": 473670
    },
    {
      "epoch": 0.19236091068013905,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010527472747274727,
      "loss": 3.2114,
      "step": 473680
    },
    {
      "epoch": 0.19237955391629724,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010527272727272728,
      "loss": 3.2527,
      "step": 473690
    },
    {
      "epoch": 0.1923981971524554,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010527072707270727,
      "loss": 3.1781,
      "step": 473700
    },
    {
      "epoch": 0.1924168403886136,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010526872687268728,
      "loss": 3.3459,
      "step": 473710
    },
    {
      "epoch": 0.19243548362477178,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010526672667266728,
      "loss": 3.2164,
      "step": 473720
    },
    {
      "epoch": 0.19245412686092997,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010526472647264727,
      "loss": 3.1201,
      "step": 473730
    },
    {
      "epoch": 0.19247277009708816,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00010526272627262728,
      "loss": 3.205,
      "step": 473740
    },
    {
      "epoch": 0.19249141333324632,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010526072607260726,
      "loss": 3.1878,
      "step": 473750
    },
    {
      "epoch": 0.1925100565694045,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010525872587258727,
      "loss": 3.2317,
      "step": 473760
    },
    {
      "epoch": 0.1925286998055627,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010525672567256725,
      "loss": 3.1465,
      "step": 473770
    },
    {
      "epoch": 0.1925473430417209,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010525472547254726,
      "loss": 3.2201,
      "step": 473780
    },
    {
      "epoch": 0.19256598627787908,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010525272527252726,
      "loss": 3.1255,
      "step": 473790
    },
    {
      "epoch": 0.19258462951403724,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010525072507250727,
      "loss": 3.3198,
      "step": 473800
    },
    {
      "epoch": 0.19260327275019543,
      "grad_norm": 0.75,
      "learning_rate": 0.00010524872487248725,
      "loss": 3.2724,
      "step": 473810
    },
    {
      "epoch": 0.19262191598635361,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010524672467246726,
      "loss": 3.1136,
      "step": 473820
    },
    {
      "epoch": 0.1926405592225118,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010524472447244724,
      "loss": 3.148,
      "step": 473830
    },
    {
      "epoch": 0.19265920245867,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010524272427242725,
      "loss": 3.014,
      "step": 473840
    },
    {
      "epoch": 0.19267784569482818,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010524072407240723,
      "loss": 3.2749,
      "step": 473850
    },
    {
      "epoch": 0.19269648893098634,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010523872387238724,
      "loss": 3.0095,
      "step": 473860
    },
    {
      "epoch": 0.19271513216714453,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010523672367236724,
      "loss": 3.1244,
      "step": 473870
    },
    {
      "epoch": 0.19273377540330272,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010523472347234725,
      "loss": 3.1677,
      "step": 473880
    },
    {
      "epoch": 0.1927524186394609,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010523272327232723,
      "loss": 3.1691,
      "step": 473890
    },
    {
      "epoch": 0.1927710618756191,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010523072307230724,
      "loss": 3.257,
      "step": 473900
    },
    {
      "epoch": 0.19278970511177726,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010522872287228725,
      "loss": 3.2361,
      "step": 473910
    },
    {
      "epoch": 0.19280834834793545,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010522672267226723,
      "loss": 3.3062,
      "step": 473920
    },
    {
      "epoch": 0.19282699158409364,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010522472247224724,
      "loss": 3.2518,
      "step": 473930
    },
    {
      "epoch": 0.19284563482025183,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010522272227222722,
      "loss": 3.1569,
      "step": 473940
    },
    {
      "epoch": 0.19286427805641002,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010522072207220723,
      "loss": 3.287,
      "step": 473950
    },
    {
      "epoch": 0.19288292129256818,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010521872187218722,
      "loss": 3.1702,
      "step": 473960
    },
    {
      "epoch": 0.19290156452872637,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010521672167216723,
      "loss": 3.2283,
      "step": 473970
    },
    {
      "epoch": 0.19292020776488455,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010521472147214722,
      "loss": 3.0704,
      "step": 473980
    },
    {
      "epoch": 0.19293885100104274,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010521272127212722,
      "loss": 3.2295,
      "step": 473990
    },
    {
      "epoch": 0.19295749423720093,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001052107210721072,
      "loss": 3.0637,
      "step": 474000
    },
    {
      "epoch": 0.1929761374733591,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010520872087208722,
      "loss": 3.0729,
      "step": 474010
    },
    {
      "epoch": 0.19299478070951728,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010520672067206721,
      "loss": 3.2444,
      "step": 474020
    },
    {
      "epoch": 0.19301342394567547,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010520472047204722,
      "loss": 3.2122,
      "step": 474030
    },
    {
      "epoch": 0.19303206718183366,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001052027202720272,
      "loss": 3.1069,
      "step": 474040
    },
    {
      "epoch": 0.19305071041799185,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010520072007200721,
      "loss": 3.1782,
      "step": 474050
    },
    {
      "epoch": 0.19306935365415004,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001051987198719872,
      "loss": 3.1731,
      "step": 474060
    },
    {
      "epoch": 0.1930879968903082,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001051967196719672,
      "loss": 3.208,
      "step": 474070
    },
    {
      "epoch": 0.1931066401264664,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010519471947194718,
      "loss": 3.1842,
      "step": 474080
    },
    {
      "epoch": 0.19312528336262458,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001051927192719272,
      "loss": 3.204,
      "step": 474090
    },
    {
      "epoch": 0.19314392659878277,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001051907190719072,
      "loss": 3.2593,
      "step": 474100
    },
    {
      "epoch": 0.19316256983494096,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001051887188718872,
      "loss": 3.3565,
      "step": 474110
    },
    {
      "epoch": 0.19318121307109912,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010518671867186721,
      "loss": 3.2703,
      "step": 474120
    },
    {
      "epoch": 0.1931998563072573,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010518471847184719,
      "loss": 3.3157,
      "step": 474130
    },
    {
      "epoch": 0.1932184995434155,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001051827182718272,
      "loss": 3.0825,
      "step": 474140
    },
    {
      "epoch": 0.19323714277957368,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010518071807180718,
      "loss": 3.1716,
      "step": 474150
    },
    {
      "epoch": 0.19325578601573187,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010517871787178719,
      "loss": 3.1632,
      "step": 474160
    },
    {
      "epoch": 0.19327442925189003,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010517671767176717,
      "loss": 3.1261,
      "step": 474170
    },
    {
      "epoch": 0.19329307248804822,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010517471747174718,
      "loss": 3.1132,
      "step": 474180
    },
    {
      "epoch": 0.1933117157242064,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010517271727172718,
      "loss": 3.2241,
      "step": 474190
    },
    {
      "epoch": 0.1933303589603646,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010517071707170719,
      "loss": 3.2604,
      "step": 474200
    },
    {
      "epoch": 0.1933490021965228,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010516871687168717,
      "loss": 3.1729,
      "step": 474210
    },
    {
      "epoch": 0.19336764543268095,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010516671667166718,
      "loss": 3.2099,
      "step": 474220
    },
    {
      "epoch": 0.19338628866883914,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010516471647164716,
      "loss": 3.2387,
      "step": 474230
    },
    {
      "epoch": 0.19340493190499733,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010516271627162717,
      "loss": 3.2386,
      "step": 474240
    },
    {
      "epoch": 0.19342357514115552,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010516071607160715,
      "loss": 3.15,
      "step": 474250
    },
    {
      "epoch": 0.1934422183773137,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010515871587158716,
      "loss": 3.2545,
      "step": 474260
    },
    {
      "epoch": 0.19346086161347187,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010515671567156716,
      "loss": 3.1836,
      "step": 474270
    },
    {
      "epoch": 0.19347950484963006,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010515471547154716,
      "loss": 3.2725,
      "step": 474280
    },
    {
      "epoch": 0.19349814808578825,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010515271527152715,
      "loss": 3.3869,
      "step": 474290
    },
    {
      "epoch": 0.19351679132194644,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010515071507150716,
      "loss": 3.1044,
      "step": 474300
    },
    {
      "epoch": 0.19353543455810462,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010514871487148716,
      "loss": 3.3072,
      "step": 474310
    },
    {
      "epoch": 0.1935540777942628,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010514671467146715,
      "loss": 3.1881,
      "step": 474320
    },
    {
      "epoch": 0.19357272103042097,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010514471447144716,
      "loss": 3.154,
      "step": 474330
    },
    {
      "epoch": 0.19359136426657916,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010514271427142715,
      "loss": 3.1955,
      "step": 474340
    },
    {
      "epoch": 0.19361000750273735,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010514071407140715,
      "loss": 3.2922,
      "step": 474350
    },
    {
      "epoch": 0.19362865073889554,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010513871387138714,
      "loss": 3.0064,
      "step": 474360
    },
    {
      "epoch": 0.19364729397505373,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010513671367136715,
      "loss": 3.3052,
      "step": 474370
    },
    {
      "epoch": 0.1936659372112119,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010513471347134713,
      "loss": 3.3989,
      "step": 474380
    },
    {
      "epoch": 0.19368458044737008,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010513271327132714,
      "loss": 3.267,
      "step": 474390
    },
    {
      "epoch": 0.19370322368352827,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010513071307130712,
      "loss": 3.1188,
      "step": 474400
    },
    {
      "epoch": 0.19372186691968646,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010512871287128713,
      "loss": 3.2111,
      "step": 474410
    },
    {
      "epoch": 0.19374051015584465,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010512671267126713,
      "loss": 3.1137,
      "step": 474420
    },
    {
      "epoch": 0.1937591533920028,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010512471247124714,
      "loss": 3.213,
      "step": 474430
    },
    {
      "epoch": 0.193777796628161,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010512271227122712,
      "loss": 3.089,
      "step": 474440
    },
    {
      "epoch": 0.1937964398643192,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010512071207120713,
      "loss": 3.3469,
      "step": 474450
    },
    {
      "epoch": 0.19381508310047738,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010511871187118711,
      "loss": 3.251,
      "step": 474460
    },
    {
      "epoch": 0.19383372633663556,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010511671167116712,
      "loss": 3.2087,
      "step": 474470
    },
    {
      "epoch": 0.19385236957279373,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001051147114711471,
      "loss": 3.2582,
      "step": 474480
    },
    {
      "epoch": 0.19387101280895191,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010511271127112711,
      "loss": 3.1897,
      "step": 474490
    },
    {
      "epoch": 0.1938896560451101,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010511071107110712,
      "loss": 3.3034,
      "step": 474500
    },
    {
      "epoch": 0.1939082992812683,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010510871087108712,
      "loss": 3.2524,
      "step": 474510
    },
    {
      "epoch": 0.19392694251742648,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010510671067106713,
      "loss": 3.1504,
      "step": 474520
    },
    {
      "epoch": 0.19394558575358467,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010510471047104711,
      "loss": 3.2697,
      "step": 474530
    },
    {
      "epoch": 0.19396422898974283,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010510271027102712,
      "loss": 3.0588,
      "step": 474540
    },
    {
      "epoch": 0.19398287222590102,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001051007100710071,
      "loss": 3.2264,
      "step": 474550
    },
    {
      "epoch": 0.1940015154620592,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010509870987098711,
      "loss": 3.2299,
      "step": 474560
    },
    {
      "epoch": 0.1940201586982174,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010509670967096709,
      "loss": 3.1974,
      "step": 474570
    },
    {
      "epoch": 0.1940388019343756,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001050947094709471,
      "loss": 3.0705,
      "step": 474580
    },
    {
      "epoch": 0.19405744517053375,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001050927092709271,
      "loss": 3.1708,
      "step": 474590
    },
    {
      "epoch": 0.19407608840669194,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001050907090709071,
      "loss": 3.2019,
      "step": 474600
    },
    {
      "epoch": 0.19409473164285013,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010508870887088709,
      "loss": 3.2336,
      "step": 474610
    },
    {
      "epoch": 0.19411337487900832,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001050867086708671,
      "loss": 3.3618,
      "step": 474620
    },
    {
      "epoch": 0.1941320181151665,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010508470847084708,
      "loss": 3.1392,
      "step": 474630
    },
    {
      "epoch": 0.19415066135132467,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010508270827082709,
      "loss": 3.1022,
      "step": 474640
    },
    {
      "epoch": 0.19416930458748285,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010508070807080708,
      "loss": 3.2156,
      "step": 474650
    },
    {
      "epoch": 0.19418794782364104,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010507870787078709,
      "loss": 3.1487,
      "step": 474660
    },
    {
      "epoch": 0.19420659105979923,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010507670767076707,
      "loss": 3.2597,
      "step": 474670
    },
    {
      "epoch": 0.19422523429595742,
      "grad_norm": 0.75,
      "learning_rate": 0.00010507470747074708,
      "loss": 3.2003,
      "step": 474680
    },
    {
      "epoch": 0.19424387753211558,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010507270727072709,
      "loss": 3.2259,
      "step": 474690
    },
    {
      "epoch": 0.19426252076827377,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010507070707070707,
      "loss": 3.2399,
      "step": 474700
    },
    {
      "epoch": 0.19428116400443196,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010506870687068708,
      "loss": 3.1366,
      "step": 474710
    },
    {
      "epoch": 0.19429980724059015,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010506670667066707,
      "loss": 3.2601,
      "step": 474720
    },
    {
      "epoch": 0.19431845047674834,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010506470647064707,
      "loss": 3.2281,
      "step": 474730
    },
    {
      "epoch": 0.19433709371290653,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010506270627062707,
      "loss": 3.2225,
      "step": 474740
    },
    {
      "epoch": 0.1943557369490647,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010506070607060708,
      "loss": 3.1973,
      "step": 474750
    },
    {
      "epoch": 0.19437438018522288,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010505870587058706,
      "loss": 3.2482,
      "step": 474760
    },
    {
      "epoch": 0.19439302342138107,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010505670567056707,
      "loss": 3.2163,
      "step": 474770
    },
    {
      "epoch": 0.19441166665753926,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010505470547054705,
      "loss": 3.2636,
      "step": 474780
    },
    {
      "epoch": 0.19443030989369744,
      "grad_norm": 0.75,
      "learning_rate": 0.00010505270527052706,
      "loss": 3.185,
      "step": 474790
    },
    {
      "epoch": 0.1944489531298556,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010505070507050704,
      "loss": 3.1336,
      "step": 474800
    },
    {
      "epoch": 0.1944675963660138,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010504870487048705,
      "loss": 3.1094,
      "step": 474810
    },
    {
      "epoch": 0.19448623960217198,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010504670467046705,
      "loss": 3.039,
      "step": 474820
    },
    {
      "epoch": 0.19450488283833017,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010504470447044706,
      "loss": 3.2435,
      "step": 474830
    },
    {
      "epoch": 0.19452352607448836,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010504270427042704,
      "loss": 3.2136,
      "step": 474840
    },
    {
      "epoch": 0.19454216931064652,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010504070407040705,
      "loss": 3.1411,
      "step": 474850
    },
    {
      "epoch": 0.1945608125468047,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010503870387038703,
      "loss": 3.1285,
      "step": 474860
    },
    {
      "epoch": 0.1945794557829629,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010503670367036704,
      "loss": 3.2284,
      "step": 474870
    },
    {
      "epoch": 0.1945980990191211,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010503470347034704,
      "loss": 3.194,
      "step": 474880
    },
    {
      "epoch": 0.19461674225527928,
      "grad_norm": 0.75,
      "learning_rate": 0.00010503270327032703,
      "loss": 3.2068,
      "step": 474890
    },
    {
      "epoch": 0.19463538549143744,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010503070307030704,
      "loss": 3.2726,
      "step": 474900
    },
    {
      "epoch": 0.19465402872759563,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010502870287028704,
      "loss": 3.2447,
      "step": 474910
    },
    {
      "epoch": 0.19467267196375382,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010502670267026704,
      "loss": 3.2261,
      "step": 474920
    },
    {
      "epoch": 0.194691315199912,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010502470247024703,
      "loss": 3.297,
      "step": 474930
    },
    {
      "epoch": 0.1947099584360702,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010502270227022704,
      "loss": 3.2891,
      "step": 474940
    },
    {
      "epoch": 0.19472860167222836,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010502070207020702,
      "loss": 3.247,
      "step": 474950
    },
    {
      "epoch": 0.19474724490838655,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010501870187018703,
      "loss": 3.1954,
      "step": 474960
    },
    {
      "epoch": 0.19476588814454474,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010501670167016702,
      "loss": 3.1821,
      "step": 474970
    },
    {
      "epoch": 0.19478453138070292,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010501470147014703,
      "loss": 3.168,
      "step": 474980
    },
    {
      "epoch": 0.1948031746168611,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010501270127012701,
      "loss": 3.2666,
      "step": 474990
    },
    {
      "epoch": 0.1948218178530193,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010501070107010702,
      "loss": 3.3056,
      "step": 475000
    },
    {
      "epoch": 0.19484046108917746,
      "grad_norm": 0.7890625,
      "learning_rate": 0.000105008700870087,
      "loss": 3.2506,
      "step": 475010
    },
    {
      "epoch": 0.19485910432533565,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010500670067006701,
      "loss": 3.1062,
      "step": 475020
    },
    {
      "epoch": 0.19487774756149384,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000105004700470047,
      "loss": 3.3291,
      "step": 475030
    },
    {
      "epoch": 0.19489639079765203,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000105002700270027,
      "loss": 3.3333,
      "step": 475040
    },
    {
      "epoch": 0.19491503403381022,
      "grad_norm": 0.703125,
      "learning_rate": 0.000105000700070007,
      "loss": 3.1761,
      "step": 475050
    },
    {
      "epoch": 0.19493367726996838,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010499869986998701,
      "loss": 3.2063,
      "step": 475060
    },
    {
      "epoch": 0.19495232050612657,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010499669966996699,
      "loss": 3.2066,
      "step": 475070
    },
    {
      "epoch": 0.19497096374228476,
      "grad_norm": 0.828125,
      "learning_rate": 0.000104994699469947,
      "loss": 3.2651,
      "step": 475080
    },
    {
      "epoch": 0.19498960697844295,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010499269926992701,
      "loss": 3.1384,
      "step": 475090
    },
    {
      "epoch": 0.19500825021460114,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010499069906990699,
      "loss": 3.2191,
      "step": 475100
    },
    {
      "epoch": 0.1950268934507593,
      "grad_norm": 0.734375,
      "learning_rate": 0.000104988698869887,
      "loss": 3.0648,
      "step": 475110
    },
    {
      "epoch": 0.1950455366869175,
      "grad_norm": 0.875,
      "learning_rate": 0.00010498669866986698,
      "loss": 3.2485,
      "step": 475120
    },
    {
      "epoch": 0.19506417992307568,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010498469846984699,
      "loss": 3.3065,
      "step": 475130
    },
    {
      "epoch": 0.19508282315923386,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010498269826982699,
      "loss": 3.1643,
      "step": 475140
    },
    {
      "epoch": 0.19510146639539205,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000104980698069807,
      "loss": 3.2188,
      "step": 475150
    },
    {
      "epoch": 0.19512010963155021,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010497869786978698,
      "loss": 3.1886,
      "step": 475160
    },
    {
      "epoch": 0.1951387528677084,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010497669766976699,
      "loss": 3.2944,
      "step": 475170
    },
    {
      "epoch": 0.1951573961038666,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010497469746974697,
      "loss": 3.2628,
      "step": 475180
    },
    {
      "epoch": 0.19517603934002478,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010497269726972698,
      "loss": 3.1183,
      "step": 475190
    },
    {
      "epoch": 0.19519468257618297,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010497069706970698,
      "loss": 3.2609,
      "step": 475200
    },
    {
      "epoch": 0.19521332581234116,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010496869686968697,
      "loss": 3.1102,
      "step": 475210
    },
    {
      "epoch": 0.19523196904849932,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010496669666966697,
      "loss": 3.234,
      "step": 475220
    },
    {
      "epoch": 0.1952506122846575,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010496469646964698,
      "loss": 3.0234,
      "step": 475230
    },
    {
      "epoch": 0.1952692555208157,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010496269626962696,
      "loss": 3.2693,
      "step": 475240
    },
    {
      "epoch": 0.1952878987569739,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010496069606960697,
      "loss": 3.1667,
      "step": 475250
    },
    {
      "epoch": 0.19530654199313208,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010495869586958695,
      "loss": 3.2123,
      "step": 475260
    },
    {
      "epoch": 0.19532518522929024,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010495669566956696,
      "loss": 3.2382,
      "step": 475270
    },
    {
      "epoch": 0.19534382846544843,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010495469546954695,
      "loss": 3.2463,
      "step": 475280
    },
    {
      "epoch": 0.19536247170160662,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010495269526952696,
      "loss": 3.239,
      "step": 475290
    },
    {
      "epoch": 0.1953811149377648,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010495069506950697,
      "loss": 3.1324,
      "step": 475300
    },
    {
      "epoch": 0.195399758173923,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010494869486948695,
      "loss": 3.2485,
      "step": 475310
    },
    {
      "epoch": 0.19541840141008116,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010494669466946696,
      "loss": 3.1051,
      "step": 475320
    },
    {
      "epoch": 0.19543704464623934,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010494469446944695,
      "loss": 3.228,
      "step": 475330
    },
    {
      "epoch": 0.19545568788239753,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010494269426942695,
      "loss": 3.2545,
      "step": 475340
    },
    {
      "epoch": 0.19547433111855572,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010494069406940694,
      "loss": 3.082,
      "step": 475350
    },
    {
      "epoch": 0.1954929743547139,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010493869386938695,
      "loss": 3.1174,
      "step": 475360
    },
    {
      "epoch": 0.19551161759087207,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010493669366936694,
      "loss": 3.0592,
      "step": 475370
    },
    {
      "epoch": 0.19553026082703026,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010493469346934695,
      "loss": 3.2504,
      "step": 475380
    },
    {
      "epoch": 0.19554890406318845,
      "grad_norm": 0.75,
      "learning_rate": 0.00010493269326932693,
      "loss": 3.1447,
      "step": 475390
    },
    {
      "epoch": 0.19556754729934664,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010493069306930694,
      "loss": 3.2166,
      "step": 475400
    },
    {
      "epoch": 0.19558619053550483,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010492869286928692,
      "loss": 3.1243,
      "step": 475410
    },
    {
      "epoch": 0.19560483377166302,
      "grad_norm": 0.75,
      "learning_rate": 0.00010492669266926693,
      "loss": 3.0758,
      "step": 475420
    },
    {
      "epoch": 0.19562347700782118,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010492469246924692,
      "loss": 3.222,
      "step": 475430
    },
    {
      "epoch": 0.19564212024397937,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010492269226922692,
      "loss": 3.2061,
      "step": 475440
    },
    {
      "epoch": 0.19566076348013756,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010492069206920692,
      "loss": 3.2621,
      "step": 475450
    },
    {
      "epoch": 0.19567940671629575,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010491869186918693,
      "loss": 3.1862,
      "step": 475460
    },
    {
      "epoch": 0.19569804995245393,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010491669166916691,
      "loss": 3.1958,
      "step": 475470
    },
    {
      "epoch": 0.1957166931886121,
      "grad_norm": 0.625,
      "learning_rate": 0.00010491469146914692,
      "loss": 3.1274,
      "step": 475480
    },
    {
      "epoch": 0.19573533642477028,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010491269126912693,
      "loss": 3.2705,
      "step": 475490
    },
    {
      "epoch": 0.19575397966092847,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010491069106910691,
      "loss": 3.3724,
      "step": 475500
    },
    {
      "epoch": 0.19577262289708666,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010490869086908692,
      "loss": 3.2099,
      "step": 475510
    },
    {
      "epoch": 0.19579126613324485,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010490669066906692,
      "loss": 3.2375,
      "step": 475520
    },
    {
      "epoch": 0.195809909369403,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010490469046904691,
      "loss": 3.2899,
      "step": 475530
    },
    {
      "epoch": 0.1958285526055612,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010490269026902691,
      "loss": 3.2829,
      "step": 475540
    },
    {
      "epoch": 0.1958471958417194,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010490069006900692,
      "loss": 3.2973,
      "step": 475550
    },
    {
      "epoch": 0.19586583907787758,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001048986898689869,
      "loss": 3.3069,
      "step": 475560
    },
    {
      "epoch": 0.19588448231403577,
      "grad_norm": 0.75,
      "learning_rate": 0.00010489668966896691,
      "loss": 3.1977,
      "step": 475570
    },
    {
      "epoch": 0.19590312555019393,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010489468946894689,
      "loss": 3.1862,
      "step": 475580
    },
    {
      "epoch": 0.19592176878635212,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001048926892689269,
      "loss": 3.1296,
      "step": 475590
    },
    {
      "epoch": 0.1959404120225103,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001048906890689069,
      "loss": 3.1298,
      "step": 475600
    },
    {
      "epoch": 0.1959590552586685,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001048886888688869,
      "loss": 3.2662,
      "step": 475610
    },
    {
      "epoch": 0.19597769849482669,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010488668866886689,
      "loss": 3.1848,
      "step": 475620
    },
    {
      "epoch": 0.19599634173098485,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001048846884688469,
      "loss": 3.2308,
      "step": 475630
    },
    {
      "epoch": 0.19601498496714304,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010488268826882688,
      "loss": 3.178,
      "step": 475640
    },
    {
      "epoch": 0.19603362820330122,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010488068806880689,
      "loss": 3.1867,
      "step": 475650
    },
    {
      "epoch": 0.1960522714394594,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010487868786878687,
      "loss": 3.1458,
      "step": 475660
    },
    {
      "epoch": 0.1960709146756176,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010487668766876688,
      "loss": 3.286,
      "step": 475670
    },
    {
      "epoch": 0.1960895579117758,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010487468746874689,
      "loss": 3.232,
      "step": 475680
    },
    {
      "epoch": 0.19610820114793395,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010487268726872688,
      "loss": 3.1625,
      "step": 475690
    },
    {
      "epoch": 0.19612684438409214,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010487068706870689,
      "loss": 3.1862,
      "step": 475700
    },
    {
      "epoch": 0.19614548762025033,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010486868686868687,
      "loss": 3.2664,
      "step": 475710
    },
    {
      "epoch": 0.19616413085640852,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010486668666866688,
      "loss": 3.2931,
      "step": 475720
    },
    {
      "epoch": 0.1961827740925667,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010486468646864686,
      "loss": 3.2547,
      "step": 475730
    },
    {
      "epoch": 0.19620141732872487,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010486268626862687,
      "loss": 3.2595,
      "step": 475740
    },
    {
      "epoch": 0.19622006056488306,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010486068606860686,
      "loss": 3.2886,
      "step": 475750
    },
    {
      "epoch": 0.19623870380104125,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010485868586858686,
      "loss": 3.1636,
      "step": 475760
    },
    {
      "epoch": 0.19625734703719944,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010485668566856686,
      "loss": 3.2568,
      "step": 475770
    },
    {
      "epoch": 0.19627599027335763,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010485468546854687,
      "loss": 3.1358,
      "step": 475780
    },
    {
      "epoch": 0.1962946335095158,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010485268526852685,
      "loss": 3.162,
      "step": 475790
    },
    {
      "epoch": 0.19631327674567398,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010485068506850686,
      "loss": 3.107,
      "step": 475800
    },
    {
      "epoch": 0.19633191998183216,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010484868486848684,
      "loss": 3.2932,
      "step": 475810
    },
    {
      "epoch": 0.19635056321799035,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010484668466846685,
      "loss": 3.3295,
      "step": 475820
    },
    {
      "epoch": 0.19636920645414854,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010484468446844685,
      "loss": 3.1795,
      "step": 475830
    },
    {
      "epoch": 0.1963878496903067,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010484268426842684,
      "loss": 3.1957,
      "step": 475840
    },
    {
      "epoch": 0.1964064929264649,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010484068406840684,
      "loss": 3.2682,
      "step": 475850
    },
    {
      "epoch": 0.19642513616262308,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010483868386838685,
      "loss": 3.1982,
      "step": 475860
    },
    {
      "epoch": 0.19644377939878127,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010483668366836683,
      "loss": 3.2772,
      "step": 475870
    },
    {
      "epoch": 0.19646242263493946,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010483468346834684,
      "loss": 3.1499,
      "step": 475880
    },
    {
      "epoch": 0.19648106587109765,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010483268326832685,
      "loss": 3.2346,
      "step": 475890
    },
    {
      "epoch": 0.1964997091072558,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010483068306830683,
      "loss": 3.1721,
      "step": 475900
    },
    {
      "epoch": 0.196518352343414,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010482868286828684,
      "loss": 3.2265,
      "step": 475910
    },
    {
      "epoch": 0.1965369955795722,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010482668266826683,
      "loss": 3.1956,
      "step": 475920
    },
    {
      "epoch": 0.19655563881573038,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010482468246824684,
      "loss": 3.2,
      "step": 475930
    },
    {
      "epoch": 0.19657428205188857,
      "grad_norm": 0.75,
      "learning_rate": 0.00010482268226822683,
      "loss": 3.2329,
      "step": 475940
    },
    {
      "epoch": 0.19659292528804673,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010482068206820684,
      "loss": 3.2346,
      "step": 475950
    },
    {
      "epoch": 0.19661156852420492,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010481868186818682,
      "loss": 3.1376,
      "step": 475960
    },
    {
      "epoch": 0.1966302117603631,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010481668166816683,
      "loss": 3.2502,
      "step": 475970
    },
    {
      "epoch": 0.1966488549965213,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010481468146814681,
      "loss": 3.2919,
      "step": 475980
    },
    {
      "epoch": 0.19666749823267948,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010481268126812682,
      "loss": 3.3958,
      "step": 475990
    },
    {
      "epoch": 0.19668614146883764,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010481068106810681,
      "loss": 3.2349,
      "step": 476000
    },
    {
      "epoch": 0.19670478470499583,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010480868086808682,
      "loss": 3.2801,
      "step": 476010
    },
    {
      "epoch": 0.19672342794115402,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001048066806680668,
      "loss": 3.2029,
      "step": 476020
    },
    {
      "epoch": 0.1967420711773122,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010480468046804681,
      "loss": 3.2163,
      "step": 476030
    },
    {
      "epoch": 0.1967607144134704,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001048026802680268,
      "loss": 3.2385,
      "step": 476040
    },
    {
      "epoch": 0.19677935764962856,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001048006800680068,
      "loss": 3.2141,
      "step": 476050
    },
    {
      "epoch": 0.19679800088578675,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010479867986798679,
      "loss": 3.1871,
      "step": 476060
    },
    {
      "epoch": 0.19681664412194494,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001047966796679668,
      "loss": 3.205,
      "step": 476070
    },
    {
      "epoch": 0.19683528735810313,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001047946794679468,
      "loss": 3.146,
      "step": 476080
    },
    {
      "epoch": 0.19685393059426132,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001047926792679268,
      "loss": 3.2469,
      "step": 476090
    },
    {
      "epoch": 0.1968725738304195,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010479067906790681,
      "loss": 3.1449,
      "step": 476100
    },
    {
      "epoch": 0.19689121706657767,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010478867886788679,
      "loss": 3.1945,
      "step": 476110
    },
    {
      "epoch": 0.19690986030273586,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001047866786678668,
      "loss": 3.144,
      "step": 476120
    },
    {
      "epoch": 0.19692850353889405,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010478467846784678,
      "loss": 3.2009,
      "step": 476130
    },
    {
      "epoch": 0.19694714677505223,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010478267826782679,
      "loss": 3.2087,
      "step": 476140
    },
    {
      "epoch": 0.19696579001121042,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010478067806780679,
      "loss": 3.1524,
      "step": 476150
    },
    {
      "epoch": 0.19698443324736858,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010477867786778678,
      "loss": 3.2635,
      "step": 476160
    },
    {
      "epoch": 0.19700307648352677,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010477667766776678,
      "loss": 3.1985,
      "step": 476170
    },
    {
      "epoch": 0.19702171971968496,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010477467746774679,
      "loss": 3.1439,
      "step": 476180
    },
    {
      "epoch": 0.19704036295584315,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010477267726772677,
      "loss": 3.2296,
      "step": 476190
    },
    {
      "epoch": 0.19705900619200134,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010477067706770678,
      "loss": 3.3149,
      "step": 476200
    },
    {
      "epoch": 0.1970776494281595,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010476867686768676,
      "loss": 3.1088,
      "step": 476210
    },
    {
      "epoch": 0.1970962926643177,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010476667666766677,
      "loss": 3.0709,
      "step": 476220
    },
    {
      "epoch": 0.19711493590047588,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010476467646764677,
      "loss": 3.2691,
      "step": 476230
    },
    {
      "epoch": 0.19713357913663407,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010476267626762678,
      "loss": 3.287,
      "step": 476240
    },
    {
      "epoch": 0.19715222237279226,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010476067606760676,
      "loss": 3.2436,
      "step": 476250
    },
    {
      "epoch": 0.19717086560895042,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010475867586758677,
      "loss": 3.187,
      "step": 476260
    },
    {
      "epoch": 0.1971895088451086,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010475667566756678,
      "loss": 3.1916,
      "step": 476270
    },
    {
      "epoch": 0.1972081520812668,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010475467546754676,
      "loss": 3.2355,
      "step": 476280
    },
    {
      "epoch": 0.19722679531742499,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010475267526752677,
      "loss": 3.2109,
      "step": 476290
    },
    {
      "epoch": 0.19724543855358317,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010475067506750675,
      "loss": 3.2044,
      "step": 476300
    },
    {
      "epoch": 0.19726408178974136,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010474867486748676,
      "loss": 3.2383,
      "step": 476310
    },
    {
      "epoch": 0.19728272502589952,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010474667466746675,
      "loss": 3.2405,
      "step": 476320
    },
    {
      "epoch": 0.1973013682620577,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010474467446744676,
      "loss": 3.3808,
      "step": 476330
    },
    {
      "epoch": 0.1973200114982159,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010474267426742674,
      "loss": 3.2633,
      "step": 476340
    },
    {
      "epoch": 0.1973386547343741,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010474067406740675,
      "loss": 3.1407,
      "step": 476350
    },
    {
      "epoch": 0.19735729797053228,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010473867386738674,
      "loss": 3.1511,
      "step": 476360
    },
    {
      "epoch": 0.19737594120669044,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010473667366736674,
      "loss": 3.1955,
      "step": 476370
    },
    {
      "epoch": 0.19739458444284863,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010473467346734673,
      "loss": 3.1646,
      "step": 476380
    },
    {
      "epoch": 0.19741322767900682,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010473267326732674,
      "loss": 3.2632,
      "step": 476390
    },
    {
      "epoch": 0.197431870915165,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010473067306730673,
      "loss": 3.1069,
      "step": 476400
    },
    {
      "epoch": 0.1974505141513232,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010472867286728674,
      "loss": 3.2941,
      "step": 476410
    },
    {
      "epoch": 0.19746915738748136,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010472667266726672,
      "loss": 3.1752,
      "step": 476420
    },
    {
      "epoch": 0.19748780062363955,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010472467246724673,
      "loss": 3.2514,
      "step": 476430
    },
    {
      "epoch": 0.19750644385979774,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010472267226722671,
      "loss": 3.122,
      "step": 476440
    },
    {
      "epoch": 0.19752508709595593,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010472067206720672,
      "loss": 3.1786,
      "step": 476450
    },
    {
      "epoch": 0.19754373033211411,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010471867186718672,
      "loss": 3.2626,
      "step": 476460
    },
    {
      "epoch": 0.19756237356827228,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010471667166716673,
      "loss": 3.2599,
      "step": 476470
    },
    {
      "epoch": 0.19758101680443046,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010471467146714672,
      "loss": 3.2428,
      "step": 476480
    },
    {
      "epoch": 0.19759966004058865,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010471267126712672,
      "loss": 3.2057,
      "step": 476490
    },
    {
      "epoch": 0.19761830327674684,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010471067106710673,
      "loss": 3.1339,
      "step": 476500
    },
    {
      "epoch": 0.19763694651290503,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010470867086708671,
      "loss": 3.1335,
      "step": 476510
    },
    {
      "epoch": 0.1976555897490632,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010470667066706672,
      "loss": 3.2236,
      "step": 476520
    },
    {
      "epoch": 0.19767423298522138,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001047046704670467,
      "loss": 3.1536,
      "step": 476530
    },
    {
      "epoch": 0.19769287622137957,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010470267026702671,
      "loss": 3.1893,
      "step": 476540
    },
    {
      "epoch": 0.19771151945753776,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001047006700670067,
      "loss": 3.2573,
      "step": 476550
    },
    {
      "epoch": 0.19773016269369595,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010469866986698672,
      "loss": 3.0,
      "step": 476560
    },
    {
      "epoch": 0.19774880592985414,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001046966696669667,
      "loss": 3.2369,
      "step": 476570
    },
    {
      "epoch": 0.1977674491660123,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001046946694669467,
      "loss": 3.2303,
      "step": 476580
    },
    {
      "epoch": 0.1977860924021705,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010469266926692669,
      "loss": 3.1692,
      "step": 476590
    },
    {
      "epoch": 0.19780473563832868,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001046906690669067,
      "loss": 3.2659,
      "step": 476600
    },
    {
      "epoch": 0.19782337887448687,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010468866886688668,
      "loss": 3.1559,
      "step": 476610
    },
    {
      "epoch": 0.19784202211064505,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010468666866686669,
      "loss": 3.1641,
      "step": 476620
    },
    {
      "epoch": 0.19786066534680322,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010468466846684668,
      "loss": 3.1248,
      "step": 476630
    },
    {
      "epoch": 0.1978793085829614,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001046826682668267,
      "loss": 3.1847,
      "step": 476640
    },
    {
      "epoch": 0.1978979518191196,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010468066806680668,
      "loss": 3.2061,
      "step": 476650
    },
    {
      "epoch": 0.19791659505527778,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010467866786678668,
      "loss": 3.0393,
      "step": 476660
    },
    {
      "epoch": 0.19793523829143597,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001046766676667667,
      "loss": 3.3014,
      "step": 476670
    },
    {
      "epoch": 0.19795388152759413,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010467466746674668,
      "loss": 3.2541,
      "step": 476680
    },
    {
      "epoch": 0.19797252476375232,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010467266726672669,
      "loss": 3.3227,
      "step": 476690
    },
    {
      "epoch": 0.1979911679999105,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010467066706670667,
      "loss": 3.3369,
      "step": 476700
    },
    {
      "epoch": 0.1980098112360687,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010466866686668668,
      "loss": 3.2346,
      "step": 476710
    },
    {
      "epoch": 0.1980284544722269,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010466666666666667,
      "loss": 3.2651,
      "step": 476720
    },
    {
      "epoch": 0.19804709770838505,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010466466646664668,
      "loss": 3.2063,
      "step": 476730
    },
    {
      "epoch": 0.19806574094454324,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010466266626662666,
      "loss": 3.1262,
      "step": 476740
    },
    {
      "epoch": 0.19808438418070143,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010466066606660667,
      "loss": 3.1794,
      "step": 476750
    },
    {
      "epoch": 0.19810302741685962,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010465866586658665,
      "loss": 3.4428,
      "step": 476760
    },
    {
      "epoch": 0.1981216706530178,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010465666566656666,
      "loss": 3.0629,
      "step": 476770
    },
    {
      "epoch": 0.198140313889176,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010465466546654666,
      "loss": 3.2776,
      "step": 476780
    },
    {
      "epoch": 0.19815895712533416,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010465266526652667,
      "loss": 3.2222,
      "step": 476790
    },
    {
      "epoch": 0.19817760036149235,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010465066506650665,
      "loss": 3.2803,
      "step": 476800
    },
    {
      "epoch": 0.19819624359765053,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010464866486648666,
      "loss": 3.2897,
      "step": 476810
    },
    {
      "epoch": 0.19821488683380872,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010464666466646664,
      "loss": 3.146,
      "step": 476820
    },
    {
      "epoch": 0.1982335300699669,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010464466446644665,
      "loss": 3.1658,
      "step": 476830
    },
    {
      "epoch": 0.19825217330612507,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010464266426642663,
      "loss": 3.1586,
      "step": 476840
    },
    {
      "epoch": 0.19827081654228326,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010464066406640664,
      "loss": 3.1642,
      "step": 476850
    },
    {
      "epoch": 0.19828945977844145,
      "grad_norm": 0.75,
      "learning_rate": 0.00010463866386638665,
      "loss": 3.0477,
      "step": 476860
    },
    {
      "epoch": 0.19830810301459964,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010463666366636665,
      "loss": 3.2329,
      "step": 476870
    },
    {
      "epoch": 0.19832674625075783,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010463466346634666,
      "loss": 3.2737,
      "step": 476880
    },
    {
      "epoch": 0.198345389486916,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010463266326632664,
      "loss": 3.2233,
      "step": 476890
    },
    {
      "epoch": 0.19836403272307418,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010463066306630665,
      "loss": 3.222,
      "step": 476900
    },
    {
      "epoch": 0.19838267595923237,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010462866286628663,
      "loss": 3.1122,
      "step": 476910
    },
    {
      "epoch": 0.19840131919539056,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010462666266626664,
      "loss": 3.2872,
      "step": 476920
    },
    {
      "epoch": 0.19841996243154875,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010462466246624662,
      "loss": 3.1611,
      "step": 476930
    },
    {
      "epoch": 0.1984386056677069,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010462266226622663,
      "loss": 3.2491,
      "step": 476940
    },
    {
      "epoch": 0.1984572489038651,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010462066206620662,
      "loss": 3.1585,
      "step": 476950
    },
    {
      "epoch": 0.19847589214002329,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010461866186618663,
      "loss": 3.2986,
      "step": 476960
    },
    {
      "epoch": 0.19849453537618147,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010461666166616662,
      "loss": 3.1492,
      "step": 476970
    },
    {
      "epoch": 0.19851317861233966,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010461466146614663,
      "loss": 3.1173,
      "step": 476980
    },
    {
      "epoch": 0.19853182184849785,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010461266126612661,
      "loss": 3.1069,
      "step": 476990
    },
    {
      "epoch": 0.198550465084656,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010461066106610662,
      "loss": 3.1235,
      "step": 477000
    },
    {
      "epoch": 0.1985691083208142,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010460866086608661,
      "loss": 3.2055,
      "step": 477010
    },
    {
      "epoch": 0.1985877515569724,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010460666066606661,
      "loss": 3.2451,
      "step": 477020
    },
    {
      "epoch": 0.19860639479313058,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001046046604660466,
      "loss": 3.2013,
      "step": 477030
    },
    {
      "epoch": 0.19862503802928877,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010460266026602661,
      "loss": 3.292,
      "step": 477040
    },
    {
      "epoch": 0.19864368126544693,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001046006600660066,
      "loss": 3.1878,
      "step": 477050
    },
    {
      "epoch": 0.19866232450160512,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001045986598659866,
      "loss": 3.2208,
      "step": 477060
    },
    {
      "epoch": 0.1986809677377633,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010459665966596661,
      "loss": 3.2492,
      "step": 477070
    },
    {
      "epoch": 0.1986996109739215,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001045946594659466,
      "loss": 3.1636,
      "step": 477080
    },
    {
      "epoch": 0.1987182542100797,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001045926592659266,
      "loss": 3.1433,
      "step": 477090
    },
    {
      "epoch": 0.19873689744623785,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001045906590659066,
      "loss": 3.2461,
      "step": 477100
    },
    {
      "epoch": 0.19875554068239604,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010458865886588661,
      "loss": 3.2582,
      "step": 477110
    },
    {
      "epoch": 0.19877418391855423,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010458665866586659,
      "loss": 3.2356,
      "step": 477120
    },
    {
      "epoch": 0.19879282715471241,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001045846584658466,
      "loss": 3.23,
      "step": 477130
    },
    {
      "epoch": 0.1988114703908706,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010458265826582658,
      "loss": 3.0606,
      "step": 477140
    },
    {
      "epoch": 0.19883011362702877,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010458065806580659,
      "loss": 3.1617,
      "step": 477150
    },
    {
      "epoch": 0.19884875686318695,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010457865786578657,
      "loss": 3.2366,
      "step": 477160
    },
    {
      "epoch": 0.19886740009934514,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010457665766576658,
      "loss": 3.1091,
      "step": 477170
    },
    {
      "epoch": 0.19888604333550333,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010457465746574658,
      "loss": 3.1464,
      "step": 477180
    },
    {
      "epoch": 0.19890468657166152,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010457265726572659,
      "loss": 3.3095,
      "step": 477190
    },
    {
      "epoch": 0.19892332980781968,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010457065706570657,
      "loss": 3.2875,
      "step": 477200
    },
    {
      "epoch": 0.19894197304397787,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010456865686568658,
      "loss": 3.2826,
      "step": 477210
    },
    {
      "epoch": 0.19896061628013606,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010456665666566656,
      "loss": 3.2509,
      "step": 477220
    },
    {
      "epoch": 0.19897925951629425,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010456465646564657,
      "loss": 3.3392,
      "step": 477230
    },
    {
      "epoch": 0.19899790275245244,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010456265626562655,
      "loss": 3.2657,
      "step": 477240
    },
    {
      "epoch": 0.19901654598861063,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010456065606560656,
      "loss": 3.1534,
      "step": 477250
    },
    {
      "epoch": 0.1990351892247688,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010455865586558657,
      "loss": 3.1079,
      "step": 477260
    },
    {
      "epoch": 0.19905383246092698,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010455665566556657,
      "loss": 3.2535,
      "step": 477270
    },
    {
      "epoch": 0.19907247569708517,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010455465546554657,
      "loss": 3.1046,
      "step": 477280
    },
    {
      "epoch": 0.19909111893324336,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010455265526552656,
      "loss": 3.2439,
      "step": 477290
    },
    {
      "epoch": 0.19910976216940154,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010455065506550657,
      "loss": 3.2444,
      "step": 477300
    },
    {
      "epoch": 0.1991284054055597,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010454865486548655,
      "loss": 3.2436,
      "step": 477310
    },
    {
      "epoch": 0.1991470486417179,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010454665466546656,
      "loss": 3.1984,
      "step": 477320
    },
    {
      "epoch": 0.19916569187787608,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010454465446544654,
      "loss": 3.191,
      "step": 477330
    },
    {
      "epoch": 0.19918433511403427,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010454265426542655,
      "loss": 3.2496,
      "step": 477340
    },
    {
      "epoch": 0.19920297835019246,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010454065406540654,
      "loss": 3.2409,
      "step": 477350
    },
    {
      "epoch": 0.19922162158635062,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010453865386538655,
      "loss": 3.1914,
      "step": 477360
    },
    {
      "epoch": 0.1992402648225088,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010453665366536653,
      "loss": 3.0163,
      "step": 477370
    },
    {
      "epoch": 0.199258908058667,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010453465346534654,
      "loss": 3.1351,
      "step": 477380
    },
    {
      "epoch": 0.1992775512948252,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010453265326532653,
      "loss": 3.3449,
      "step": 477390
    },
    {
      "epoch": 0.19929619453098338,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010453065306530653,
      "loss": 3.1815,
      "step": 477400
    },
    {
      "epoch": 0.19931483776714154,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010452865286528653,
      "loss": 3.1291,
      "step": 477410
    },
    {
      "epoch": 0.19933348100329973,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010452665266526654,
      "loss": 3.179,
      "step": 477420
    },
    {
      "epoch": 0.19935212423945792,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010452465246524652,
      "loss": 3.1679,
      "step": 477430
    },
    {
      "epoch": 0.1993707674756161,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010452265226522653,
      "loss": 3.168,
      "step": 477440
    },
    {
      "epoch": 0.1993894107117743,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010452065206520651,
      "loss": 3.1124,
      "step": 477450
    },
    {
      "epoch": 0.19940805394793248,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010451865186518652,
      "loss": 3.1716,
      "step": 477460
    },
    {
      "epoch": 0.19942669718409065,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010451665166516653,
      "loss": 3.0904,
      "step": 477470
    },
    {
      "epoch": 0.19944534042024883,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010451465146514651,
      "loss": 3.194,
      "step": 477480
    },
    {
      "epoch": 0.19946398365640702,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010451265126512652,
      "loss": 3.1383,
      "step": 477490
    },
    {
      "epoch": 0.1994826268925652,
      "grad_norm": 0.875,
      "learning_rate": 0.00010451065106510652,
      "loss": 3.1113,
      "step": 477500
    },
    {
      "epoch": 0.1995012701287234,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010450865086508653,
      "loss": 3.2181,
      "step": 477510
    },
    {
      "epoch": 0.19951991336488156,
      "grad_norm": 0.75,
      "learning_rate": 0.00010450665066506651,
      "loss": 3.2405,
      "step": 477520
    },
    {
      "epoch": 0.19953855660103975,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010450465046504652,
      "loss": 3.2977,
      "step": 477530
    },
    {
      "epoch": 0.19955719983719794,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001045026502650265,
      "loss": 3.1361,
      "step": 477540
    },
    {
      "epoch": 0.19957584307335613,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010450065006500651,
      "loss": 3.2531,
      "step": 477550
    },
    {
      "epoch": 0.19959448630951432,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010449864986498649,
      "loss": 3.2739,
      "step": 477560
    },
    {
      "epoch": 0.19961312954567248,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001044966496649665,
      "loss": 3.2566,
      "step": 477570
    },
    {
      "epoch": 0.19963177278183067,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001044946494649465,
      "loss": 3.2855,
      "step": 477580
    },
    {
      "epoch": 0.19965041601798886,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001044926492649265,
      "loss": 3.2491,
      "step": 477590
    },
    {
      "epoch": 0.19966905925414705,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010449064906490649,
      "loss": 3.0959,
      "step": 477600
    },
    {
      "epoch": 0.19968770249030524,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001044886488648865,
      "loss": 3.1632,
      "step": 477610
    },
    {
      "epoch": 0.1997063457264634,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010448664866486648,
      "loss": 3.2421,
      "step": 477620
    },
    {
      "epoch": 0.19972498896262159,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010448464846484649,
      "loss": 3.2449,
      "step": 477630
    },
    {
      "epoch": 0.19974363219877977,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010448264826482648,
      "loss": 3.1609,
      "step": 477640
    },
    {
      "epoch": 0.19976227543493796,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010448064806480648,
      "loss": 3.2091,
      "step": 477650
    },
    {
      "epoch": 0.19978091867109615,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010447864786478649,
      "loss": 3.1702,
      "step": 477660
    },
    {
      "epoch": 0.19979956190725434,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010447664766476648,
      "loss": 3.2052,
      "step": 477670
    },
    {
      "epoch": 0.1998182051434125,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010447464746474649,
      "loss": 3.039,
      "step": 477680
    },
    {
      "epoch": 0.1998368483795707,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010447264726472647,
      "loss": 3.1184,
      "step": 477690
    },
    {
      "epoch": 0.19985549161572888,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010447064706470648,
      "loss": 3.2869,
      "step": 477700
    },
    {
      "epoch": 0.19987413485188707,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010446864686468647,
      "loss": 3.1015,
      "step": 477710
    },
    {
      "epoch": 0.19989277808804526,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010446664666466648,
      "loss": 3.2113,
      "step": 477720
    },
    {
      "epoch": 0.19991142132420342,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010446464646464647,
      "loss": 3.1491,
      "step": 477730
    },
    {
      "epoch": 0.1999300645603616,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00010446264626462648,
      "loss": 3.1892,
      "step": 477740
    },
    {
      "epoch": 0.1999487077965198,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010446064606460646,
      "loss": 3.1721,
      "step": 477750
    },
    {
      "epoch": 0.199967351032678,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010445864586458647,
      "loss": 3.2057,
      "step": 477760
    },
    {
      "epoch": 0.19998599426883618,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010445664566456645,
      "loss": 3.082,
      "step": 477770
    },
    {
      "epoch": 0.20000463750499434,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010445464546454646,
      "loss": 3.2295,
      "step": 477780
    },
    {
      "epoch": 0.20002328074115253,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010445264526452644,
      "loss": 3.2089,
      "step": 477790
    },
    {
      "epoch": 0.20004192397731071,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010445064506450645,
      "loss": 3.2091,
      "step": 477800
    },
    {
      "epoch": 0.2000605672134689,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010444864486448645,
      "loss": 3.2497,
      "step": 477810
    },
    {
      "epoch": 0.2000792104496271,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010444664466446646,
      "loss": 3.1498,
      "step": 477820
    },
    {
      "epoch": 0.20009785368578525,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010444464446444644,
      "loss": 3.1501,
      "step": 477830
    },
    {
      "epoch": 0.20011649692194344,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010444264426442645,
      "loss": 3.2507,
      "step": 477840
    },
    {
      "epoch": 0.20013514015810163,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010444064406440646,
      "loss": 3.1218,
      "step": 477850
    },
    {
      "epoch": 0.20015378339425982,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010443864386438644,
      "loss": 3.2293,
      "step": 477860
    },
    {
      "epoch": 0.200172426630418,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010443664366436645,
      "loss": 3.0064,
      "step": 477870
    },
    {
      "epoch": 0.20019106986657617,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010443464346434643,
      "loss": 3.2856,
      "step": 477880
    },
    {
      "epoch": 0.20020971310273436,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010443264326432644,
      "loss": 3.1954,
      "step": 477890
    },
    {
      "epoch": 0.20022835633889255,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010443064306430644,
      "loss": 3.2188,
      "step": 477900
    },
    {
      "epoch": 0.20024699957505074,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010442864286428645,
      "loss": 3.1763,
      "step": 477910
    },
    {
      "epoch": 0.20026564281120893,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010442664266426643,
      "loss": 3.2729,
      "step": 477920
    },
    {
      "epoch": 0.20028428604736712,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010442464246424644,
      "loss": 3.2864,
      "step": 477930
    },
    {
      "epoch": 0.20030292928352528,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010442264226422642,
      "loss": 3.2735,
      "step": 477940
    },
    {
      "epoch": 0.20032157251968347,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010442064206420643,
      "loss": 3.2004,
      "step": 477950
    },
    {
      "epoch": 0.20034021575584166,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010441864186418642,
      "loss": 3.1364,
      "step": 477960
    },
    {
      "epoch": 0.20035885899199984,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010441664166416642,
      "loss": 3.1219,
      "step": 477970
    },
    {
      "epoch": 0.20037750222815803,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010441464146414641,
      "loss": 3.225,
      "step": 477980
    },
    {
      "epoch": 0.2003961454643162,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010441264126412642,
      "loss": 3.3109,
      "step": 477990
    },
    {
      "epoch": 0.20041478870047438,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001044106410641064,
      "loss": 3.2189,
      "step": 478000
    },
    {
      "epoch": 0.20043343193663257,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010440864086408642,
      "loss": 3.3326,
      "step": 478010
    },
    {
      "epoch": 0.20045207517279076,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001044066406640664,
      "loss": 3.2358,
      "step": 478020
    },
    {
      "epoch": 0.20047071840894895,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001044046404640464,
      "loss": 3.1952,
      "step": 478030
    },
    {
      "epoch": 0.2004893616451071,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001044026402640264,
      "loss": 3.1585,
      "step": 478040
    },
    {
      "epoch": 0.2005080048812653,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010440064006400641,
      "loss": 3.2413,
      "step": 478050
    },
    {
      "epoch": 0.2005266481174235,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010439863986398642,
      "loss": 3.1278,
      "step": 478060
    },
    {
      "epoch": 0.20054529135358168,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001043966396639664,
      "loss": 3.2762,
      "step": 478070
    },
    {
      "epoch": 0.20056393458973987,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010439463946394641,
      "loss": 3.2184,
      "step": 478080
    },
    {
      "epoch": 0.20058257782589803,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001043926392639264,
      "loss": 3.2603,
      "step": 478090
    },
    {
      "epoch": 0.20060122106205622,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001043906390639064,
      "loss": 3.2277,
      "step": 478100
    },
    {
      "epoch": 0.2006198642982144,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010438863886388638,
      "loss": 3.2574,
      "step": 478110
    },
    {
      "epoch": 0.2006385075343726,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001043866386638664,
      "loss": 3.2782,
      "step": 478120
    },
    {
      "epoch": 0.20065715077053078,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010438463846384639,
      "loss": 3.2795,
      "step": 478130
    },
    {
      "epoch": 0.20067579400668897,
      "grad_norm": 0.625,
      "learning_rate": 0.0001043826382638264,
      "loss": 3.1834,
      "step": 478140
    },
    {
      "epoch": 0.20069443724284713,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010438063806380638,
      "loss": 3.1936,
      "step": 478150
    },
    {
      "epoch": 0.20071308047900532,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010437863786378639,
      "loss": 3.2093,
      "step": 478160
    },
    {
      "epoch": 0.2007317237151635,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010437663766376637,
      "loss": 3.2748,
      "step": 478170
    },
    {
      "epoch": 0.2007503669513217,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010437463746374638,
      "loss": 3.2294,
      "step": 478180
    },
    {
      "epoch": 0.2007690101874799,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010437263726372636,
      "loss": 3.3038,
      "step": 478190
    },
    {
      "epoch": 0.20078765342363805,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010437063706370637,
      "loss": 3.2318,
      "step": 478200
    },
    {
      "epoch": 0.20080629665979624,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010436863686368637,
      "loss": 3.2096,
      "step": 478210
    },
    {
      "epoch": 0.20082493989595443,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010436663666366638,
      "loss": 3.2674,
      "step": 478220
    },
    {
      "epoch": 0.20084358313211262,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010436463646364636,
      "loss": 3.291,
      "step": 478230
    },
    {
      "epoch": 0.2008622263682708,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010436263626362637,
      "loss": 3.1819,
      "step": 478240
    },
    {
      "epoch": 0.20088086960442897,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010436063606360638,
      "loss": 3.2024,
      "step": 478250
    },
    {
      "epoch": 0.20089951284058716,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010435863586358636,
      "loss": 3.2372,
      "step": 478260
    },
    {
      "epoch": 0.20091815607674535,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010435663566356637,
      "loss": 3.3215,
      "step": 478270
    },
    {
      "epoch": 0.20093679931290354,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010435463546354636,
      "loss": 3.1387,
      "step": 478280
    },
    {
      "epoch": 0.20095544254906172,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010435263526352636,
      "loss": 3.2215,
      "step": 478290
    },
    {
      "epoch": 0.20097408578521989,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010435063506350636,
      "loss": 3.1338,
      "step": 478300
    },
    {
      "epoch": 0.20099272902137807,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010434863486348636,
      "loss": 3.148,
      "step": 478310
    },
    {
      "epoch": 0.20101137225753626,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010434663466346635,
      "loss": 3.1442,
      "step": 478320
    },
    {
      "epoch": 0.20103001549369445,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010434463446344636,
      "loss": 3.1875,
      "step": 478330
    },
    {
      "epoch": 0.20104865872985264,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010434263426342634,
      "loss": 3.1667,
      "step": 478340
    },
    {
      "epoch": 0.20106730196601083,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010434063406340635,
      "loss": 3.1067,
      "step": 478350
    },
    {
      "epoch": 0.201085945202169,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010433863386338634,
      "loss": 3.2126,
      "step": 478360
    },
    {
      "epoch": 0.20110458843832718,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010433663366336635,
      "loss": 3.3285,
      "step": 478370
    },
    {
      "epoch": 0.20112323167448537,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010433463346334633,
      "loss": 3.2346,
      "step": 478380
    },
    {
      "epoch": 0.20114187491064356,
      "grad_norm": 0.75,
      "learning_rate": 0.00010433263326332634,
      "loss": 3.1977,
      "step": 478390
    },
    {
      "epoch": 0.20116051814680175,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010433063306330632,
      "loss": 3.2363,
      "step": 478400
    },
    {
      "epoch": 0.2011791613829599,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010432863286328633,
      "loss": 3.3261,
      "step": 478410
    },
    {
      "epoch": 0.2011978046191181,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010432663266326632,
      "loss": 3.1829,
      "step": 478420
    },
    {
      "epoch": 0.2012164478552763,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010432463246324633,
      "loss": 3.2563,
      "step": 478430
    },
    {
      "epoch": 0.20123509109143448,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010432263226322633,
      "loss": 3.1132,
      "step": 478440
    },
    {
      "epoch": 0.20125373432759266,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010432063206320633,
      "loss": 3.1662,
      "step": 478450
    },
    {
      "epoch": 0.20127237756375083,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010431863186318634,
      "loss": 3.1439,
      "step": 478460
    },
    {
      "epoch": 0.20129102079990902,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010431663166316632,
      "loss": 3.3328,
      "step": 478470
    },
    {
      "epoch": 0.2013096640360672,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010431463146314633,
      "loss": 3.2132,
      "step": 478480
    },
    {
      "epoch": 0.2013283072722254,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010431263126312631,
      "loss": 3.2634,
      "step": 478490
    },
    {
      "epoch": 0.20134695050838358,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010431063106310632,
      "loss": 3.0867,
      "step": 478500
    },
    {
      "epoch": 0.20136559374454174,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001043086308630863,
      "loss": 3.1257,
      "step": 478510
    },
    {
      "epoch": 0.20138423698069993,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010430663066306631,
      "loss": 3.1273,
      "step": 478520
    },
    {
      "epoch": 0.20140288021685812,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010430463046304631,
      "loss": 3.2247,
      "step": 478530
    },
    {
      "epoch": 0.2014215234530163,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010430263026302632,
      "loss": 3.2028,
      "step": 478540
    },
    {
      "epoch": 0.2014401666891745,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001043006300630063,
      "loss": 3.2164,
      "step": 478550
    },
    {
      "epoch": 0.20145880992533266,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010429862986298631,
      "loss": 3.159,
      "step": 478560
    },
    {
      "epoch": 0.20147745316149085,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010429662966296629,
      "loss": 3.1656,
      "step": 478570
    },
    {
      "epoch": 0.20149609639764904,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001042946294629463,
      "loss": 3.309,
      "step": 478580
    },
    {
      "epoch": 0.20151473963380723,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001042926292629263,
      "loss": 3.2283,
      "step": 478590
    },
    {
      "epoch": 0.20153338286996542,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010429062906290629,
      "loss": 3.2021,
      "step": 478600
    },
    {
      "epoch": 0.2015520261061236,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010428862886288629,
      "loss": 3.216,
      "step": 478610
    },
    {
      "epoch": 0.20157066934228177,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001042866286628663,
      "loss": 3.2768,
      "step": 478620
    },
    {
      "epoch": 0.20158931257843996,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010428462846284628,
      "loss": 3.2002,
      "step": 478630
    },
    {
      "epoch": 0.20160795581459814,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010428262826282629,
      "loss": 3.2251,
      "step": 478640
    },
    {
      "epoch": 0.20162659905075633,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001042806280628063,
      "loss": 3.296,
      "step": 478650
    },
    {
      "epoch": 0.20164524228691452,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010427862786278628,
      "loss": 3.4455,
      "step": 478660
    },
    {
      "epoch": 0.20166388552307268,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010427662766276629,
      "loss": 3.3598,
      "step": 478670
    },
    {
      "epoch": 0.20168252875923087,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010427462746274628,
      "loss": 3.1382,
      "step": 478680
    },
    {
      "epoch": 0.20170117199538906,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010427262726272629,
      "loss": 3.222,
      "step": 478690
    },
    {
      "epoch": 0.20171981523154725,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010427062706270627,
      "loss": 3.1738,
      "step": 478700
    },
    {
      "epoch": 0.20173845846770544,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010426862686268628,
      "loss": 3.269,
      "step": 478710
    },
    {
      "epoch": 0.2017571017038636,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010426662666266627,
      "loss": 3.3624,
      "step": 478720
    },
    {
      "epoch": 0.2017757449400218,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010426462646264627,
      "loss": 3.2498,
      "step": 478730
    },
    {
      "epoch": 0.20179438817617998,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010426262626262626,
      "loss": 3.0927,
      "step": 478740
    },
    {
      "epoch": 0.20181303141233817,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010426062606260627,
      "loss": 3.3321,
      "step": 478750
    },
    {
      "epoch": 0.20183167464849636,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010425862586258626,
      "loss": 3.0496,
      "step": 478760
    },
    {
      "epoch": 0.20185031788465452,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010425662566256627,
      "loss": 3.2299,
      "step": 478770
    },
    {
      "epoch": 0.2018689611208127,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010425462546254625,
      "loss": 3.2096,
      "step": 478780
    },
    {
      "epoch": 0.2018876043569709,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010425262526252626,
      "loss": 3.2733,
      "step": 478790
    },
    {
      "epoch": 0.20190624759312908,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010425062506250624,
      "loss": 3.2179,
      "step": 478800
    },
    {
      "epoch": 0.20192489082928727,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010424862486248625,
      "loss": 3.2631,
      "step": 478810
    },
    {
      "epoch": 0.20194353406544546,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010424662466246623,
      "loss": 3.3221,
      "step": 478820
    },
    {
      "epoch": 0.20196217730160362,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010424462446244624,
      "loss": 3.2483,
      "step": 478830
    },
    {
      "epoch": 0.2019808205377618,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010424262426242625,
      "loss": 3.2888,
      "step": 478840
    },
    {
      "epoch": 0.20199946377392,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010424062406240625,
      "loss": 3.1937,
      "step": 478850
    },
    {
      "epoch": 0.2020181070100782,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010423862386238626,
      "loss": 3.2074,
      "step": 478860
    },
    {
      "epoch": 0.20203675024623638,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010423662366236624,
      "loss": 3.1615,
      "step": 478870
    },
    {
      "epoch": 0.20205539348239454,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010423462346234625,
      "loss": 3.1611,
      "step": 478880
    },
    {
      "epoch": 0.20207403671855273,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010423262326232623,
      "loss": 3.2417,
      "step": 478890
    },
    {
      "epoch": 0.20209267995471092,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010423062306230624,
      "loss": 3.2345,
      "step": 478900
    },
    {
      "epoch": 0.2021113231908691,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010422862286228624,
      "loss": 3.2248,
      "step": 478910
    },
    {
      "epoch": 0.2021299664270273,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010422662266226623,
      "loss": 3.1908,
      "step": 478920
    },
    {
      "epoch": 0.20214860966318546,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010422462246224623,
      "loss": 3.113,
      "step": 478930
    },
    {
      "epoch": 0.20216725289934365,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010422262226222624,
      "loss": 3.101,
      "step": 478940
    },
    {
      "epoch": 0.20218589613550184,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010422062206220622,
      "loss": 3.2996,
      "step": 478950
    },
    {
      "epoch": 0.20220453937166002,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010421862186218623,
      "loss": 3.1817,
      "step": 478960
    },
    {
      "epoch": 0.2022231826078182,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010421662166216621,
      "loss": 3.2361,
      "step": 478970
    },
    {
      "epoch": 0.20224182584397637,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010421462146214622,
      "loss": 3.1895,
      "step": 478980
    },
    {
      "epoch": 0.20226046908013456,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010421262126212621,
      "loss": 3.1135,
      "step": 478990
    },
    {
      "epoch": 0.20227911231629275,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010421062106210622,
      "loss": 3.241,
      "step": 479000
    },
    {
      "epoch": 0.20229775555245094,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001042086208620862,
      "loss": 3.2972,
      "step": 479010
    },
    {
      "epoch": 0.20231639878860913,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010420662066206621,
      "loss": 3.1657,
      "step": 479020
    },
    {
      "epoch": 0.20233504202476732,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001042046204620462,
      "loss": 3.2029,
      "step": 479030
    },
    {
      "epoch": 0.20235368526092548,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001042026202620262,
      "loss": 3.1547,
      "step": 479040
    },
    {
      "epoch": 0.20237232849708367,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010420062006200621,
      "loss": 3.2472,
      "step": 479050
    },
    {
      "epoch": 0.20239097173324186,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001041986198619862,
      "loss": 3.2052,
      "step": 479060
    },
    {
      "epoch": 0.20240961496940005,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001041966196619662,
      "loss": 3.1315,
      "step": 479070
    },
    {
      "epoch": 0.20242825820555824,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001041946194619462,
      "loss": 3.1975,
      "step": 479080
    },
    {
      "epoch": 0.2024469014417164,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010419261926192621,
      "loss": 3.1953,
      "step": 479090
    },
    {
      "epoch": 0.2024655446778746,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010419061906190619,
      "loss": 3.1499,
      "step": 479100
    },
    {
      "epoch": 0.20248418791403278,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001041886188618862,
      "loss": 3.2425,
      "step": 479110
    },
    {
      "epoch": 0.20250283115019097,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010418661866186618,
      "loss": 3.2439,
      "step": 479120
    },
    {
      "epoch": 0.20252147438634915,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010418461846184619,
      "loss": 3.1988,
      "step": 479130
    },
    {
      "epoch": 0.20254011762250732,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010418261826182617,
      "loss": 3.2377,
      "step": 479140
    },
    {
      "epoch": 0.2025587608586655,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010418061806180618,
      "loss": 3.2567,
      "step": 479150
    },
    {
      "epoch": 0.2025774040948237,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010417861786178618,
      "loss": 3.1478,
      "step": 479160
    },
    {
      "epoch": 0.20259604733098188,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010417661766176619,
      "loss": 3.1735,
      "step": 479170
    },
    {
      "epoch": 0.20261469056714007,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010417461746174617,
      "loss": 3.2558,
      "step": 479180
    },
    {
      "epoch": 0.20263333380329823,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010417261726172618,
      "loss": 3.1578,
      "step": 479190
    },
    {
      "epoch": 0.20265197703945642,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010417061706170616,
      "loss": 3.165,
      "step": 479200
    },
    {
      "epoch": 0.2026706202756146,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010416861686168617,
      "loss": 3.2489,
      "step": 479210
    },
    {
      "epoch": 0.2026892635117728,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010416661666166617,
      "loss": 3.2621,
      "step": 479220
    },
    {
      "epoch": 0.202707906747931,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010416461646164618,
      "loss": 3.2507,
      "step": 479230
    },
    {
      "epoch": 0.20272654998408915,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010416261626162617,
      "loss": 3.2177,
      "step": 479240
    },
    {
      "epoch": 0.20274519322024734,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010416061606160617,
      "loss": 3.2791,
      "step": 479250
    },
    {
      "epoch": 0.20276383645640553,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010415861586158618,
      "loss": 3.2838,
      "step": 479260
    },
    {
      "epoch": 0.20278247969256372,
      "grad_norm": 0.75,
      "learning_rate": 0.00010415661566156616,
      "loss": 3.3333,
      "step": 479270
    },
    {
      "epoch": 0.2028011229287219,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010415461546154617,
      "loss": 3.3225,
      "step": 479280
    },
    {
      "epoch": 0.2028197661648801,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010415261526152615,
      "loss": 3.2801,
      "step": 479290
    },
    {
      "epoch": 0.20283840940103826,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010415061506150616,
      "loss": 3.2635,
      "step": 479300
    },
    {
      "epoch": 0.20285705263719644,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010414861486148615,
      "loss": 3.1971,
      "step": 479310
    },
    {
      "epoch": 0.20287569587335463,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010414661466146616,
      "loss": 3.149,
      "step": 479320
    },
    {
      "epoch": 0.20289433910951282,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010414461446144615,
      "loss": 3.2401,
      "step": 479330
    },
    {
      "epoch": 0.202912982345671,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010414261426142615,
      "loss": 3.2282,
      "step": 479340
    },
    {
      "epoch": 0.20293162558182917,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010414061406140614,
      "loss": 3.2254,
      "step": 479350
    },
    {
      "epoch": 0.20295026881798736,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010413861386138615,
      "loss": 3.1059,
      "step": 479360
    },
    {
      "epoch": 0.20296891205414555,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010413661366136613,
      "loss": 3.2035,
      "step": 479370
    },
    {
      "epoch": 0.20298755529030374,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010413461346134614,
      "loss": 3.3186,
      "step": 479380
    },
    {
      "epoch": 0.20300619852646193,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010413261326132613,
      "loss": 3.0944,
      "step": 479390
    },
    {
      "epoch": 0.2030248417626201,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010413061306130614,
      "loss": 3.2187,
      "step": 479400
    },
    {
      "epoch": 0.20304348499877828,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010412861286128612,
      "loss": 3.2094,
      "step": 479410
    },
    {
      "epoch": 0.20306212823493647,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010412661266126613,
      "loss": 3.1606,
      "step": 479420
    },
    {
      "epoch": 0.20308077147109466,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010412461246124614,
      "loss": 3.1915,
      "step": 479430
    },
    {
      "epoch": 0.20309941470725285,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010412261226122612,
      "loss": 3.1617,
      "step": 479440
    },
    {
      "epoch": 0.203118057943411,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010412061206120613,
      "loss": 3.309,
      "step": 479450
    },
    {
      "epoch": 0.2031367011795692,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010411861186118612,
      "loss": 3.1047,
      "step": 479460
    },
    {
      "epoch": 0.20315534441572738,
      "grad_norm": 0.625,
      "learning_rate": 0.00010411661166116612,
      "loss": 3.2126,
      "step": 479470
    },
    {
      "epoch": 0.20317398765188557,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010411461146114612,
      "loss": 3.1433,
      "step": 479480
    },
    {
      "epoch": 0.20319263088804376,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010411261126112613,
      "loss": 3.1145,
      "step": 479490
    },
    {
      "epoch": 0.20321127412420195,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010411061106110611,
      "loss": 3.1416,
      "step": 479500
    },
    {
      "epoch": 0.2032299173603601,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010410861086108612,
      "loss": 3.1817,
      "step": 479510
    },
    {
      "epoch": 0.2032485605965183,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001041066106610661,
      "loss": 3.2336,
      "step": 479520
    },
    {
      "epoch": 0.2032672038326765,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010410461046104611,
      "loss": 3.2349,
      "step": 479530
    },
    {
      "epoch": 0.20328584706883468,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010410261026102611,
      "loss": 3.1765,
      "step": 479540
    },
    {
      "epoch": 0.20330449030499287,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010410061006100612,
      "loss": 3.2122,
      "step": 479550
    },
    {
      "epoch": 0.20332313354115103,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001040986098609861,
      "loss": 3.3186,
      "step": 479560
    },
    {
      "epoch": 0.20334177677730922,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010409660966096611,
      "loss": 3.2283,
      "step": 479570
    },
    {
      "epoch": 0.2033604200134674,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010409460946094609,
      "loss": 3.2507,
      "step": 479580
    },
    {
      "epoch": 0.2033790632496256,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001040926092609261,
      "loss": 3.1036,
      "step": 479590
    },
    {
      "epoch": 0.20339770648578379,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010409060906090608,
      "loss": 3.2421,
      "step": 479600
    },
    {
      "epoch": 0.20341634972194195,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010408860886088609,
      "loss": 3.2597,
      "step": 479610
    },
    {
      "epoch": 0.20343499295810014,
      "grad_norm": 0.75,
      "learning_rate": 0.00010408660866086609,
      "loss": 3.2654,
      "step": 479620
    },
    {
      "epoch": 0.20345363619425832,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001040846084608461,
      "loss": 3.1519,
      "step": 479630
    },
    {
      "epoch": 0.20347227943041651,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001040826082608261,
      "loss": 3.1702,
      "step": 479640
    },
    {
      "epoch": 0.2034909226665747,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010408060806080609,
      "loss": 3.1582,
      "step": 479650
    },
    {
      "epoch": 0.20350956590273286,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001040786078607861,
      "loss": 3.1989,
      "step": 479660
    },
    {
      "epoch": 0.20352820913889105,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010407660766076608,
      "loss": 3.1953,
      "step": 479670
    },
    {
      "epoch": 0.20354685237504924,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010407460746074609,
      "loss": 3.1614,
      "step": 479680
    },
    {
      "epoch": 0.20356549561120743,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010407260726072607,
      "loss": 3.306,
      "step": 479690
    },
    {
      "epoch": 0.20358413884736562,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010407060706070608,
      "loss": 3.2359,
      "step": 479700
    },
    {
      "epoch": 0.2036027820835238,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010406860686068607,
      "loss": 3.1907,
      "step": 479710
    },
    {
      "epoch": 0.20362142531968197,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010406660666066608,
      "loss": 3.0973,
      "step": 479720
    },
    {
      "epoch": 0.20364006855584016,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010406460646064606,
      "loss": 3.2047,
      "step": 479730
    },
    {
      "epoch": 0.20365871179199835,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010406260626062607,
      "loss": 3.361,
      "step": 479740
    },
    {
      "epoch": 0.20367735502815654,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010406060606060606,
      "loss": 3.1864,
      "step": 479750
    },
    {
      "epoch": 0.20369599826431473,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010405860586058606,
      "loss": 3.1795,
      "step": 479760
    },
    {
      "epoch": 0.2037146415004729,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010405660566056606,
      "loss": 3.321,
      "step": 479770
    },
    {
      "epoch": 0.20373328473663108,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010405460546054606,
      "loss": 3.1966,
      "step": 479780
    },
    {
      "epoch": 0.20375192797278927,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010405260526052605,
      "loss": 3.2947,
      "step": 479790
    },
    {
      "epoch": 0.20377057120894745,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010405060506050606,
      "loss": 3.2783,
      "step": 479800
    },
    {
      "epoch": 0.20378921444510564,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010404860486048604,
      "loss": 3.2409,
      "step": 479810
    },
    {
      "epoch": 0.2038078576812638,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010404660466046605,
      "loss": 3.0914,
      "step": 479820
    },
    {
      "epoch": 0.203826500917422,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010404460446044606,
      "loss": 3.2094,
      "step": 479830
    },
    {
      "epoch": 0.20384514415358018,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010404260426042604,
      "loss": 3.2536,
      "step": 479840
    },
    {
      "epoch": 0.20386378738973837,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010404060406040605,
      "loss": 3.1055,
      "step": 479850
    },
    {
      "epoch": 0.20388243062589656,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010403860386038605,
      "loss": 3.1324,
      "step": 479860
    },
    {
      "epoch": 0.20390107386205472,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010403660366036606,
      "loss": 3.387,
      "step": 479870
    },
    {
      "epoch": 0.2039197170982129,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010403460346034604,
      "loss": 3.2287,
      "step": 479880
    },
    {
      "epoch": 0.2039383603343711,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010403260326032605,
      "loss": 3.2141,
      "step": 479890
    },
    {
      "epoch": 0.2039570035705293,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010403060306030603,
      "loss": 3.1979,
      "step": 479900
    },
    {
      "epoch": 0.20397564680668748,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010402860286028604,
      "loss": 3.1688,
      "step": 479910
    },
    {
      "epoch": 0.20399429004284564,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010402660266026602,
      "loss": 3.1515,
      "step": 479920
    },
    {
      "epoch": 0.20401293327900383,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010402460246024603,
      "loss": 3.1579,
      "step": 479930
    },
    {
      "epoch": 0.20403157651516202,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010402260226022603,
      "loss": 3.2212,
      "step": 479940
    },
    {
      "epoch": 0.2040502197513202,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010402060206020603,
      "loss": 3.3298,
      "step": 479950
    },
    {
      "epoch": 0.2040688629874784,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010401860186018602,
      "loss": 3.2485,
      "step": 479960
    },
    {
      "epoch": 0.20408750622363658,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010401660166016603,
      "loss": 3.1719,
      "step": 479970
    },
    {
      "epoch": 0.20410614945979474,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010401460146014601,
      "loss": 3.26,
      "step": 479980
    },
    {
      "epoch": 0.20412479269595293,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010401260126012602,
      "loss": 3.1777,
      "step": 479990
    },
    {
      "epoch": 0.20414343593211112,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000104010601060106,
      "loss": 3.2398,
      "step": 480000
    },
    {
      "epoch": 0.2041620791682693,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010400860086008601,
      "loss": 3.2098,
      "step": 480010
    },
    {
      "epoch": 0.2041807224044275,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010400660066006602,
      "loss": 3.2291,
      "step": 480020
    },
    {
      "epoch": 0.20419936564058566,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010400460046004601,
      "loss": 3.2758,
      "step": 480030
    },
    {
      "epoch": 0.20421800887674385,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010400260026002602,
      "loss": 3.291,
      "step": 480040
    },
    {
      "epoch": 0.20423665211290204,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000104000600060006,
      "loss": 3.3282,
      "step": 480050
    },
    {
      "epoch": 0.20425529534906023,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010399859985998601,
      "loss": 3.1723,
      "step": 480060
    },
    {
      "epoch": 0.20427393858521842,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000103996599659966,
      "loss": 3.1872,
      "step": 480070
    },
    {
      "epoch": 0.20429258182137658,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000103994599459946,
      "loss": 3.3321,
      "step": 480080
    },
    {
      "epoch": 0.20431122505753477,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010399259925992599,
      "loss": 3.2237,
      "step": 480090
    },
    {
      "epoch": 0.20432986829369296,
      "grad_norm": 0.80859375,
      "learning_rate": 0.000103990599059906,
      "loss": 3.2689,
      "step": 480100
    },
    {
      "epoch": 0.20434851152985115,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010398859885988599,
      "loss": 3.2429,
      "step": 480110
    },
    {
      "epoch": 0.20436715476600933,
      "grad_norm": 0.84375,
      "learning_rate": 0.000103986598659866,
      "loss": 3.1869,
      "step": 480120
    },
    {
      "epoch": 0.2043857980021675,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00010398459845984598,
      "loss": 3.1743,
      "step": 480130
    },
    {
      "epoch": 0.20440444123832568,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010398259825982599,
      "loss": 3.2642,
      "step": 480140
    },
    {
      "epoch": 0.20442308447448387,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010398059805980597,
      "loss": 3.237,
      "step": 480150
    },
    {
      "epoch": 0.20444172771064206,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010397859785978598,
      "loss": 3.2287,
      "step": 480160
    },
    {
      "epoch": 0.20446037094680025,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010397659765976598,
      "loss": 3.2156,
      "step": 480170
    },
    {
      "epoch": 0.20447901418295844,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010397459745974599,
      "loss": 3.2351,
      "step": 480180
    },
    {
      "epoch": 0.2044976574191166,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010397259725972597,
      "loss": 3.1672,
      "step": 480190
    },
    {
      "epoch": 0.2045163006552748,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010397059705970598,
      "loss": 3.1901,
      "step": 480200
    },
    {
      "epoch": 0.20453494389143298,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010396859685968596,
      "loss": 3.0232,
      "step": 480210
    },
    {
      "epoch": 0.20455358712759117,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010396659665966597,
      "loss": 3.1935,
      "step": 480220
    },
    {
      "epoch": 0.20457223036374936,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010396459645964598,
      "loss": 3.2503,
      "step": 480230
    },
    {
      "epoch": 0.20459087359990752,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010396259625962596,
      "loss": 3.203,
      "step": 480240
    },
    {
      "epoch": 0.2046095168360657,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010396059605960597,
      "loss": 3.2393,
      "step": 480250
    },
    {
      "epoch": 0.2046281600722239,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010395859585958597,
      "loss": 3.1784,
      "step": 480260
    },
    {
      "epoch": 0.20464680330838209,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010395659565956598,
      "loss": 3.1031,
      "step": 480270
    },
    {
      "epoch": 0.20466544654454027,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010395459545954596,
      "loss": 3.1174,
      "step": 480280
    },
    {
      "epoch": 0.20468408978069844,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010395259525952597,
      "loss": 3.0754,
      "step": 480290
    },
    {
      "epoch": 0.20470273301685663,
      "grad_norm": 0.625,
      "learning_rate": 0.00010395059505950595,
      "loss": 3.1904,
      "step": 480300
    },
    {
      "epoch": 0.20472137625301481,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010394859485948596,
      "loss": 3.0901,
      "step": 480310
    },
    {
      "epoch": 0.204740019489173,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010394659465946594,
      "loss": 3.1933,
      "step": 480320
    },
    {
      "epoch": 0.2047586627253312,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010394459445944595,
      "loss": 3.3218,
      "step": 480330
    },
    {
      "epoch": 0.20477730596148935,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010394259425942594,
      "loss": 3.3429,
      "step": 480340
    },
    {
      "epoch": 0.20479594919764754,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010394059405940595,
      "loss": 3.1973,
      "step": 480350
    },
    {
      "epoch": 0.20481459243380573,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010393859385938594,
      "loss": 3.2354,
      "step": 480360
    },
    {
      "epoch": 0.20483323566996392,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010393659365936594,
      "loss": 3.2212,
      "step": 480370
    },
    {
      "epoch": 0.2048518789061221,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010393459345934593,
      "loss": 3.3213,
      "step": 480380
    },
    {
      "epoch": 0.2048705221422803,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010393259325932594,
      "loss": 3.123,
      "step": 480390
    },
    {
      "epoch": 0.20488916537843846,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010393059305930593,
      "loss": 3.0304,
      "step": 480400
    },
    {
      "epoch": 0.20490780861459665,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010392859285928593,
      "loss": 3.1515,
      "step": 480410
    },
    {
      "epoch": 0.20492645185075484,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010392659265926594,
      "loss": 3.1618,
      "step": 480420
    },
    {
      "epoch": 0.20494509508691303,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010392459245924593,
      "loss": 3.2256,
      "step": 480430
    },
    {
      "epoch": 0.20496373832307122,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010392259225922594,
      "loss": 3.1063,
      "step": 480440
    },
    {
      "epoch": 0.20498238155922938,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010392059205920592,
      "loss": 3.2127,
      "step": 480450
    },
    {
      "epoch": 0.20500102479538757,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010391859185918593,
      "loss": 3.1949,
      "step": 480460
    },
    {
      "epoch": 0.20501966803154575,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010391659165916591,
      "loss": 3.2006,
      "step": 480470
    },
    {
      "epoch": 0.20503831126770394,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010391459145914592,
      "loss": 3.2113,
      "step": 480480
    },
    {
      "epoch": 0.20505695450386213,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010391259125912592,
      "loss": 3.2259,
      "step": 480490
    },
    {
      "epoch": 0.2050755977400203,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010391059105910593,
      "loss": 3.2186,
      "step": 480500
    },
    {
      "epoch": 0.20509424097617848,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010390859085908591,
      "loss": 3.0953,
      "step": 480510
    },
    {
      "epoch": 0.20511288421233667,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010390659065906592,
      "loss": 3.1463,
      "step": 480520
    },
    {
      "epoch": 0.20513152744849486,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001039045904590459,
      "loss": 3.1084,
      "step": 480530
    },
    {
      "epoch": 0.20515017068465305,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010390259025902591,
      "loss": 3.2246,
      "step": 480540
    },
    {
      "epoch": 0.2051688139208112,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010390059005900589,
      "loss": 3.2121,
      "step": 480550
    },
    {
      "epoch": 0.2051874571569694,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001038985898589859,
      "loss": 3.1798,
      "step": 480560
    },
    {
      "epoch": 0.2052061003931276,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001038965896589659,
      "loss": 3.1966,
      "step": 480570
    },
    {
      "epoch": 0.20522474362928578,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001038945894589459,
      "loss": 3.1704,
      "step": 480580
    },
    {
      "epoch": 0.20524338686544397,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010389258925892589,
      "loss": 3.286,
      "step": 480590
    },
    {
      "epoch": 0.20526203010160213,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001038905890589059,
      "loss": 3.2433,
      "step": 480600
    },
    {
      "epoch": 0.20528067333776032,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001038885888588859,
      "loss": 3.2152,
      "step": 480610
    },
    {
      "epoch": 0.2052993165739185,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010388658865886589,
      "loss": 3.2596,
      "step": 480620
    },
    {
      "epoch": 0.2053179598100767,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001038845884588459,
      "loss": 3.1545,
      "step": 480630
    },
    {
      "epoch": 0.20533660304623488,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010388258825882588,
      "loss": 3.0864,
      "step": 480640
    },
    {
      "epoch": 0.20535524628239307,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010388058805880589,
      "loss": 3.2884,
      "step": 480650
    },
    {
      "epoch": 0.20537388951855123,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010387858785878588,
      "loss": 3.3796,
      "step": 480660
    },
    {
      "epoch": 0.20539253275470942,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001038765876587659,
      "loss": 3.2084,
      "step": 480670
    },
    {
      "epoch": 0.2054111759908676,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010387458745874588,
      "loss": 3.2575,
      "step": 480680
    },
    {
      "epoch": 0.2054298192270258,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010387258725872589,
      "loss": 3.2978,
      "step": 480690
    },
    {
      "epoch": 0.205448462463184,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010387058705870587,
      "loss": 3.1107,
      "step": 480700
    },
    {
      "epoch": 0.20546710569934215,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010386858685868588,
      "loss": 3.2075,
      "step": 480710
    },
    {
      "epoch": 0.20548574893550034,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010386658665866587,
      "loss": 3.2619,
      "step": 480720
    },
    {
      "epoch": 0.20550439217165853,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010386458645864587,
      "loss": 3.277,
      "step": 480730
    },
    {
      "epoch": 0.20552303540781672,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010386258625862586,
      "loss": 3.2145,
      "step": 480740
    },
    {
      "epoch": 0.2055416786439749,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010386058605860587,
      "loss": 3.2842,
      "step": 480750
    },
    {
      "epoch": 0.20556032188013307,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010385858585858585,
      "loss": 3.2059,
      "step": 480760
    },
    {
      "epoch": 0.20557896511629126,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010385658565856586,
      "loss": 3.3683,
      "step": 480770
    },
    {
      "epoch": 0.20559760835244945,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010385458545854585,
      "loss": 3.1496,
      "step": 480780
    },
    {
      "epoch": 0.20561625158860763,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010385258525852585,
      "loss": 3.2011,
      "step": 480790
    },
    {
      "epoch": 0.20563489482476582,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010385058505850585,
      "loss": 3.1259,
      "step": 480800
    },
    {
      "epoch": 0.20565353806092398,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010384858485848586,
      "loss": 3.1912,
      "step": 480810
    },
    {
      "epoch": 0.20567218129708217,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010384658465846587,
      "loss": 3.2047,
      "step": 480820
    },
    {
      "epoch": 0.20569082453324036,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010384458445844585,
      "loss": 3.2587,
      "step": 480830
    },
    {
      "epoch": 0.20570946776939855,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010384258425842586,
      "loss": 3.2141,
      "step": 480840
    },
    {
      "epoch": 0.20572811100555674,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010384058405840584,
      "loss": 3.1757,
      "step": 480850
    },
    {
      "epoch": 0.20574675424171493,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010383858385838585,
      "loss": 3.2495,
      "step": 480860
    },
    {
      "epoch": 0.2057653974778731,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010383658365836583,
      "loss": 3.2046,
      "step": 480870
    },
    {
      "epoch": 0.20578404071403128,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010383458345834584,
      "loss": 3.1481,
      "step": 480880
    },
    {
      "epoch": 0.20580268395018947,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010383258325832584,
      "loss": 3.3836,
      "step": 480890
    },
    {
      "epoch": 0.20582132718634766,
      "grad_norm": 0.75,
      "learning_rate": 0.00010383058305830585,
      "loss": 3.2191,
      "step": 480900
    },
    {
      "epoch": 0.20583997042250585,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010382858285828583,
      "loss": 3.1742,
      "step": 480910
    },
    {
      "epoch": 0.205858613658664,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010382658265826584,
      "loss": 3.2572,
      "step": 480920
    },
    {
      "epoch": 0.2058772568948222,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010382458245824582,
      "loss": 3.046,
      "step": 480930
    },
    {
      "epoch": 0.20589590013098039,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010382258225822583,
      "loss": 3.0553,
      "step": 480940
    },
    {
      "epoch": 0.20591454336713857,
      "grad_norm": 0.75,
      "learning_rate": 0.00010382058205820581,
      "loss": 3.2105,
      "step": 480950
    },
    {
      "epoch": 0.20593318660329676,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010381858185818582,
      "loss": 3.3259,
      "step": 480960
    },
    {
      "epoch": 0.20595182983945493,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010381658165816582,
      "loss": 3.1536,
      "step": 480970
    },
    {
      "epoch": 0.20597047307561311,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010381458145814583,
      "loss": 3.3239,
      "step": 480980
    },
    {
      "epoch": 0.2059891163117713,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010381258125812581,
      "loss": 3.2144,
      "step": 480990
    },
    {
      "epoch": 0.2060077595479295,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010381058105810582,
      "loss": 3.2437,
      "step": 481000
    },
    {
      "epoch": 0.20602640278408768,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010380858085808583,
      "loss": 3.1926,
      "step": 481010
    },
    {
      "epoch": 0.20604504602024584,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010380658065806581,
      "loss": 3.1681,
      "step": 481020
    },
    {
      "epoch": 0.20606368925640403,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010380458045804582,
      "loss": 3.2245,
      "step": 481030
    },
    {
      "epoch": 0.20608233249256222,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010380258025802581,
      "loss": 3.2732,
      "step": 481040
    },
    {
      "epoch": 0.2061009757287204,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010380058005800581,
      "loss": 2.9944,
      "step": 481050
    },
    {
      "epoch": 0.2061196189648786,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001037985798579858,
      "loss": 3.0899,
      "step": 481060
    },
    {
      "epoch": 0.2061382622010368,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010379657965796581,
      "loss": 3.14,
      "step": 481070
    },
    {
      "epoch": 0.20615690543719495,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001037945794579458,
      "loss": 3.2487,
      "step": 481080
    },
    {
      "epoch": 0.20617554867335314,
      "grad_norm": 0.75,
      "learning_rate": 0.0001037925792579258,
      "loss": 3.087,
      "step": 481090
    },
    {
      "epoch": 0.20619419190951133,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010379057905790579,
      "loss": 3.295,
      "step": 481100
    },
    {
      "epoch": 0.20621283514566952,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001037885788578858,
      "loss": 3.2144,
      "step": 481110
    },
    {
      "epoch": 0.2062314783818277,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010378657865786579,
      "loss": 3.1728,
      "step": 481120
    },
    {
      "epoch": 0.20625012161798587,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001037845784578458,
      "loss": 3.2096,
      "step": 481130
    },
    {
      "epoch": 0.20626876485414405,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010378257825782578,
      "loss": 3.2242,
      "step": 481140
    },
    {
      "epoch": 0.20628740809030224,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010378057805780579,
      "loss": 3.3259,
      "step": 481150
    },
    {
      "epoch": 0.20630605132646043,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010377857785778577,
      "loss": 3.3681,
      "step": 481160
    },
    {
      "epoch": 0.20632469456261862,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010377657765776578,
      "loss": 3.2006,
      "step": 481170
    },
    {
      "epoch": 0.20634333779877678,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010377457745774576,
      "loss": 3.2406,
      "step": 481180
    },
    {
      "epoch": 0.20636198103493497,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010377257725772577,
      "loss": 3.1879,
      "step": 481190
    },
    {
      "epoch": 0.20638062427109316,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010377057705770577,
      "loss": 3.0611,
      "step": 481200
    },
    {
      "epoch": 0.20639926750725135,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010376857685768578,
      "loss": 3.4568,
      "step": 481210
    },
    {
      "epoch": 0.20641791074340954,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010376657665766579,
      "loss": 3.3246,
      "step": 481220
    },
    {
      "epoch": 0.2064365539795677,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010376457645764577,
      "loss": 3.2985,
      "step": 481230
    },
    {
      "epoch": 0.2064551972157259,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010376257625762578,
      "loss": 3.1884,
      "step": 481240
    },
    {
      "epoch": 0.20647384045188408,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010376057605760576,
      "loss": 3.1274,
      "step": 481250
    },
    {
      "epoch": 0.20649248368804227,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010375857585758577,
      "loss": 3.1427,
      "step": 481260
    },
    {
      "epoch": 0.20651112692420046,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010375657565756575,
      "loss": 3.178,
      "step": 481270
    },
    {
      "epoch": 0.20652977016035864,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010375457545754576,
      "loss": 3.2611,
      "step": 481280
    },
    {
      "epoch": 0.2065484133965168,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010375257525752576,
      "loss": 3.2628,
      "step": 481290
    },
    {
      "epoch": 0.206567056632675,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010375057505750577,
      "loss": 3.1402,
      "step": 481300
    },
    {
      "epoch": 0.20658569986883318,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010374857485748575,
      "loss": 3.2329,
      "step": 481310
    },
    {
      "epoch": 0.20660434310499137,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010374657465746576,
      "loss": 3.1724,
      "step": 481320
    },
    {
      "epoch": 0.20662298634114956,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010374457445744574,
      "loss": 3.1963,
      "step": 481330
    },
    {
      "epoch": 0.20664162957730772,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010374257425742575,
      "loss": 3.2222,
      "step": 481340
    },
    {
      "epoch": 0.2066602728134659,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010374057405740574,
      "loss": 3.1891,
      "step": 481350
    },
    {
      "epoch": 0.2066789160496241,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010373857385738575,
      "loss": 3.1292,
      "step": 481360
    },
    {
      "epoch": 0.2066975592857823,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010373657365736573,
      "loss": 3.2551,
      "step": 481370
    },
    {
      "epoch": 0.20671620252194048,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010373457345734574,
      "loss": 3.3817,
      "step": 481380
    },
    {
      "epoch": 0.20673484575809864,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010373257325732573,
      "loss": 3.2135,
      "step": 481390
    },
    {
      "epoch": 0.20675348899425683,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010373057305730573,
      "loss": 3.1809,
      "step": 481400
    },
    {
      "epoch": 0.20677213223041502,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010372857285728574,
      "loss": 3.2098,
      "step": 481410
    },
    {
      "epoch": 0.2067907754665732,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010372657265726573,
      "loss": 3.2184,
      "step": 481420
    },
    {
      "epoch": 0.2068094187027314,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010372457245724574,
      "loss": 3.2976,
      "step": 481430
    },
    {
      "epoch": 0.20682806193888956,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010372257225722573,
      "loss": 3.2351,
      "step": 481440
    },
    {
      "epoch": 0.20684670517504775,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010372057205720574,
      "loss": 3.1906,
      "step": 481450
    },
    {
      "epoch": 0.20686534841120593,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010371857185718572,
      "loss": 3.2343,
      "step": 481460
    },
    {
      "epoch": 0.20688399164736412,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010371657165716573,
      "loss": 3.1256,
      "step": 481470
    },
    {
      "epoch": 0.2069026348835223,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010371457145714571,
      "loss": 3.0929,
      "step": 481480
    },
    {
      "epoch": 0.20692127811968047,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010371257125712572,
      "loss": 3.2037,
      "step": 481490
    },
    {
      "epoch": 0.20693992135583866,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001037105710571057,
      "loss": 3.2189,
      "step": 481500
    },
    {
      "epoch": 0.20695856459199685,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010370857085708571,
      "loss": 3.3014,
      "step": 481510
    },
    {
      "epoch": 0.20697720782815504,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010370657065706571,
      "loss": 3.2162,
      "step": 481520
    },
    {
      "epoch": 0.20699585106431323,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010370457045704572,
      "loss": 3.157,
      "step": 481530
    },
    {
      "epoch": 0.20701449430047142,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001037025702570257,
      "loss": 3.2384,
      "step": 481540
    },
    {
      "epoch": 0.20703313753662958,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010370057005700571,
      "loss": 3.2025,
      "step": 481550
    },
    {
      "epoch": 0.20705178077278777,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010369856985698569,
      "loss": 3.1871,
      "step": 481560
    },
    {
      "epoch": 0.20707042400894596,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001036965696569657,
      "loss": 3.1367,
      "step": 481570
    },
    {
      "epoch": 0.20708906724510415,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010369456945694568,
      "loss": 3.1391,
      "step": 481580
    },
    {
      "epoch": 0.20710771048126234,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010369256925692569,
      "loss": 3.3396,
      "step": 481590
    },
    {
      "epoch": 0.2071263537174205,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001036905690569057,
      "loss": 3.18,
      "step": 481600
    },
    {
      "epoch": 0.2071449969535787,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001036885688568857,
      "loss": 3.1897,
      "step": 481610
    },
    {
      "epoch": 0.20716364018973688,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001036865686568657,
      "loss": 3.2774,
      "step": 481620
    },
    {
      "epoch": 0.20718228342589506,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010368456845684569,
      "loss": 3.1869,
      "step": 481630
    },
    {
      "epoch": 0.20720092666205325,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001036825682568257,
      "loss": 3.2479,
      "step": 481640
    },
    {
      "epoch": 0.20721956989821141,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010368056805680568,
      "loss": 3.1734,
      "step": 481650
    },
    {
      "epoch": 0.2072382131343696,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00010367856785678569,
      "loss": 3.2034,
      "step": 481660
    },
    {
      "epoch": 0.2072568563705278,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010367656765676568,
      "loss": 3.2635,
      "step": 481670
    },
    {
      "epoch": 0.20727549960668598,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010367456745674568,
      "loss": 3.0784,
      "step": 481680
    },
    {
      "epoch": 0.20729414284284417,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010367256725672567,
      "loss": 3.1849,
      "step": 481690
    },
    {
      "epoch": 0.20731278607900233,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010367056705670568,
      "loss": 3.2029,
      "step": 481700
    },
    {
      "epoch": 0.20733142931516052,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010366856685668567,
      "loss": 3.1666,
      "step": 481710
    },
    {
      "epoch": 0.2073500725513187,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010366656665666568,
      "loss": 3.2156,
      "step": 481720
    },
    {
      "epoch": 0.2073687157874769,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010366456645664566,
      "loss": 3.0735,
      "step": 481730
    },
    {
      "epoch": 0.2073873590236351,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010366256625662567,
      "loss": 3.321,
      "step": 481740
    },
    {
      "epoch": 0.20740600225979328,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010366056605660566,
      "loss": 3.1548,
      "step": 481750
    },
    {
      "epoch": 0.20742464549595144,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010365856585658567,
      "loss": 3.1168,
      "step": 481760
    },
    {
      "epoch": 0.20744328873210963,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010365656565656565,
      "loss": 3.1997,
      "step": 481770
    },
    {
      "epoch": 0.20746193196826782,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010365456545654566,
      "loss": 3.2284,
      "step": 481780
    },
    {
      "epoch": 0.207480575204426,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010365256525652564,
      "loss": 3.2088,
      "step": 481790
    },
    {
      "epoch": 0.2074992184405842,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010365056505650565,
      "loss": 3.3356,
      "step": 481800
    },
    {
      "epoch": 0.20751786167674235,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010364856485648566,
      "loss": 3.2443,
      "step": 481810
    },
    {
      "epoch": 0.20753650491290054,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010364656465646564,
      "loss": 3.0954,
      "step": 481820
    },
    {
      "epoch": 0.20755514814905873,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010364456445644565,
      "loss": 3.074,
      "step": 481830
    },
    {
      "epoch": 0.20757379138521692,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010364256425642565,
      "loss": 3.253,
      "step": 481840
    },
    {
      "epoch": 0.2075924346213751,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010364056405640566,
      "loss": 3.2319,
      "step": 481850
    },
    {
      "epoch": 0.20761107785753327,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010363856385638564,
      "loss": 3.1842,
      "step": 481860
    },
    {
      "epoch": 0.20762972109369146,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010363656365636565,
      "loss": 3.1682,
      "step": 481870
    },
    {
      "epoch": 0.20764836432984965,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010363456345634563,
      "loss": 3.2844,
      "step": 481880
    },
    {
      "epoch": 0.20766700756600784,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010363256325632564,
      "loss": 3.2339,
      "step": 481890
    },
    {
      "epoch": 0.20768565080216603,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010363056305630562,
      "loss": 3.2046,
      "step": 481900
    },
    {
      "epoch": 0.2077042940383242,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010362856285628563,
      "loss": 3.2804,
      "step": 481910
    },
    {
      "epoch": 0.20772293727448238,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010362656265626563,
      "loss": 3.0573,
      "step": 481920
    },
    {
      "epoch": 0.20774158051064057,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010362456245624564,
      "loss": 3.2765,
      "step": 481930
    },
    {
      "epoch": 0.20776022374679876,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010362256225622562,
      "loss": 3.3193,
      "step": 481940
    },
    {
      "epoch": 0.20777886698295694,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010362056205620563,
      "loss": 3.1327,
      "step": 481950
    },
    {
      "epoch": 0.20779751021911513,
      "grad_norm": 0.75,
      "learning_rate": 0.00010361856185618561,
      "loss": 3.244,
      "step": 481960
    },
    {
      "epoch": 0.2078161534552733,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010361656165616562,
      "loss": 3.343,
      "step": 481970
    },
    {
      "epoch": 0.20783479669143148,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010361456145614561,
      "loss": 3.2107,
      "step": 481980
    },
    {
      "epoch": 0.20785343992758967,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010361256125612562,
      "loss": 3.1699,
      "step": 481990
    },
    {
      "epoch": 0.20787208316374786,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010361056105610562,
      "loss": 3.1995,
      "step": 482000
    },
    {
      "epoch": 0.20789072639990605,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010360856085608562,
      "loss": 3.2533,
      "step": 482010
    },
    {
      "epoch": 0.2079093696360642,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010360656065606562,
      "loss": 3.1813,
      "step": 482020
    },
    {
      "epoch": 0.2079280128722224,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001036045604560456,
      "loss": 3.1903,
      "step": 482030
    },
    {
      "epoch": 0.2079466561083806,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010360256025602562,
      "loss": 3.3007,
      "step": 482040
    },
    {
      "epoch": 0.20796529934453878,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001036005600560056,
      "loss": 3.1957,
      "step": 482050
    },
    {
      "epoch": 0.20798394258069697,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001035985598559856,
      "loss": 3.1514,
      "step": 482060
    },
    {
      "epoch": 0.20800258581685513,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001035965596559656,
      "loss": 3.3397,
      "step": 482070
    },
    {
      "epoch": 0.20802122905301332,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010359455945594561,
      "loss": 3.1647,
      "step": 482080
    },
    {
      "epoch": 0.2080398722891715,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001035925592559256,
      "loss": 3.0956,
      "step": 482090
    },
    {
      "epoch": 0.2080585155253297,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001035905590559056,
      "loss": 3.0993,
      "step": 482100
    },
    {
      "epoch": 0.20807715876148788,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010358855885588558,
      "loss": 3.1624,
      "step": 482110
    },
    {
      "epoch": 0.20809580199764605,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001035865586558656,
      "loss": 3.1975,
      "step": 482120
    },
    {
      "epoch": 0.20811444523380423,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010358455845584558,
      "loss": 3.2535,
      "step": 482130
    },
    {
      "epoch": 0.20813308846996242,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010358255825582559,
      "loss": 3.209,
      "step": 482140
    },
    {
      "epoch": 0.2081517317061206,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010358055805580558,
      "loss": 3.182,
      "step": 482150
    },
    {
      "epoch": 0.2081703749422788,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010357855785578559,
      "loss": 3.2631,
      "step": 482160
    },
    {
      "epoch": 0.20818901817843696,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010357655765576557,
      "loss": 3.1482,
      "step": 482170
    },
    {
      "epoch": 0.20820766141459515,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010357455745574558,
      "loss": 3.2682,
      "step": 482180
    },
    {
      "epoch": 0.20822630465075334,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010357255725572559,
      "loss": 3.2368,
      "step": 482190
    },
    {
      "epoch": 0.20824494788691153,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010357055705570557,
      "loss": 3.3162,
      "step": 482200
    },
    {
      "epoch": 0.20826359112306972,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010356855685568558,
      "loss": 3.2707,
      "step": 482210
    },
    {
      "epoch": 0.2082822343592279,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010356655665566556,
      "loss": 3.1238,
      "step": 482220
    },
    {
      "epoch": 0.20830087759538607,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010356455645564557,
      "loss": 3.2052,
      "step": 482230
    },
    {
      "epoch": 0.20831952083154426,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010356255625562557,
      "loss": 3.1928,
      "step": 482240
    },
    {
      "epoch": 0.20833816406770245,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010356055605560558,
      "loss": 3.0507,
      "step": 482250
    },
    {
      "epoch": 0.20835680730386064,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010355855585558556,
      "loss": 3.205,
      "step": 482260
    },
    {
      "epoch": 0.20837545054001883,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010355655565556557,
      "loss": 3.2368,
      "step": 482270
    },
    {
      "epoch": 0.208394093776177,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010355455545554555,
      "loss": 3.0896,
      "step": 482280
    },
    {
      "epoch": 0.20841273701233518,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010355255525552556,
      "loss": 3.2506,
      "step": 482290
    },
    {
      "epoch": 0.20843138024849336,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010355055505550556,
      "loss": 3.3004,
      "step": 482300
    },
    {
      "epoch": 0.20845002348465155,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010354855485548556,
      "loss": 3.2601,
      "step": 482310
    },
    {
      "epoch": 0.20846866672080974,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010354655465546555,
      "loss": 3.303,
      "step": 482320
    },
    {
      "epoch": 0.2084873099569679,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010354455445544556,
      "loss": 3.2984,
      "step": 482330
    },
    {
      "epoch": 0.2085059531931261,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010354255425542554,
      "loss": 3.2263,
      "step": 482340
    },
    {
      "epoch": 0.20852459642928428,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010354055405540555,
      "loss": 3.2746,
      "step": 482350
    },
    {
      "epoch": 0.20854323966544247,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010353855385538553,
      "loss": 3.1679,
      "step": 482360
    },
    {
      "epoch": 0.20856188290160066,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010353655365536554,
      "loss": 3.1561,
      "step": 482370
    },
    {
      "epoch": 0.20858052613775882,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010353455345534553,
      "loss": 3.2762,
      "step": 482380
    },
    {
      "epoch": 0.208599169373917,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010353255325532554,
      "loss": 3.278,
      "step": 482390
    },
    {
      "epoch": 0.2086178126100752,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010353055305530555,
      "loss": 3.2046,
      "step": 482400
    },
    {
      "epoch": 0.2086364558462334,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010352855285528553,
      "loss": 3.2571,
      "step": 482410
    },
    {
      "epoch": 0.20865509908239158,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010352655265526554,
      "loss": 3.2427,
      "step": 482420
    },
    {
      "epoch": 0.20867374231854977,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010352455245524553,
      "loss": 3.274,
      "step": 482430
    },
    {
      "epoch": 0.20869238555470793,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010352255225522553,
      "loss": 3.2922,
      "step": 482440
    },
    {
      "epoch": 0.20871102879086612,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010352055205520552,
      "loss": 3.2704,
      "step": 482450
    },
    {
      "epoch": 0.2087296720270243,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010351855185518553,
      "loss": 3.2759,
      "step": 482460
    },
    {
      "epoch": 0.2087483152631825,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010351655165516552,
      "loss": 3.2343,
      "step": 482470
    },
    {
      "epoch": 0.20876695849934068,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010351455145514553,
      "loss": 3.1773,
      "step": 482480
    },
    {
      "epoch": 0.20878560173549884,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010351255125512551,
      "loss": 3.2821,
      "step": 482490
    },
    {
      "epoch": 0.20880424497165703,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010351055105510552,
      "loss": 3.2553,
      "step": 482500
    },
    {
      "epoch": 0.20882288820781522,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001035085508550855,
      "loss": 3.314,
      "step": 482510
    },
    {
      "epoch": 0.2088415314439734,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010350655065506551,
      "loss": 3.1889,
      "step": 482520
    },
    {
      "epoch": 0.2088601746801316,
      "grad_norm": 0.75,
      "learning_rate": 0.00010350455045504551,
      "loss": 3.2015,
      "step": 482530
    },
    {
      "epoch": 0.20887881791628976,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001035025502550255,
      "loss": 3.1208,
      "step": 482540
    },
    {
      "epoch": 0.20889746115244795,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001035005500550055,
      "loss": 3.2537,
      "step": 482550
    },
    {
      "epoch": 0.20891610438860614,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010349854985498551,
      "loss": 3.2388,
      "step": 482560
    },
    {
      "epoch": 0.20893474762476433,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010349654965496549,
      "loss": 3.2271,
      "step": 482570
    },
    {
      "epoch": 0.20895339086092252,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001034945494549455,
      "loss": 3.2262,
      "step": 482580
    },
    {
      "epoch": 0.20897203409708068,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010349254925492551,
      "loss": 3.1858,
      "step": 482590
    },
    {
      "epoch": 0.20899067733323887,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010349054905490549,
      "loss": 3.2765,
      "step": 482600
    },
    {
      "epoch": 0.20900932056939706,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001034885488548855,
      "loss": 3.241,
      "step": 482610
    },
    {
      "epoch": 0.20902796380555524,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001034865486548655,
      "loss": 3.3437,
      "step": 482620
    },
    {
      "epoch": 0.20904660704171343,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001034845484548455,
      "loss": 3.2889,
      "step": 482630
    },
    {
      "epoch": 0.20906525027787162,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010348254825482549,
      "loss": 3.2471,
      "step": 482640
    },
    {
      "epoch": 0.20908389351402978,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001034805480548055,
      "loss": 3.1748,
      "step": 482650
    },
    {
      "epoch": 0.20910253675018797,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010347854785478548,
      "loss": 3.2302,
      "step": 482660
    },
    {
      "epoch": 0.20912117998634616,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010347654765476549,
      "loss": 3.3115,
      "step": 482670
    },
    {
      "epoch": 0.20913982322250435,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010347454745474547,
      "loss": 3.2071,
      "step": 482680
    },
    {
      "epoch": 0.20915846645866254,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010347254725472548,
      "loss": 3.202,
      "step": 482690
    },
    {
      "epoch": 0.2091771096948207,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010347054705470547,
      "loss": 3.2091,
      "step": 482700
    },
    {
      "epoch": 0.2091957529309789,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010346854685468548,
      "loss": 3.2124,
      "step": 482710
    },
    {
      "epoch": 0.20921439616713708,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010346654665466547,
      "loss": 3.2631,
      "step": 482720
    },
    {
      "epoch": 0.20923303940329527,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010346454645464547,
      "loss": 3.1832,
      "step": 482730
    },
    {
      "epoch": 0.20925168263945346,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010346254625462546,
      "loss": 3.1529,
      "step": 482740
    },
    {
      "epoch": 0.20927032587561162,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00010346054605460547,
      "loss": 3.2821,
      "step": 482750
    },
    {
      "epoch": 0.2092889691117698,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010345854585458545,
      "loss": 3.3389,
      "step": 482760
    },
    {
      "epoch": 0.209307612347928,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010345654565456546,
      "loss": 3.0485,
      "step": 482770
    },
    {
      "epoch": 0.20932625558408618,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010345454545454545,
      "loss": 3.0373,
      "step": 482780
    },
    {
      "epoch": 0.20934489882024437,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010345254525452546,
      "loss": 3.3178,
      "step": 482790
    },
    {
      "epoch": 0.20936354205640254,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010345054505450547,
      "loss": 3.0513,
      "step": 482800
    },
    {
      "epoch": 0.20938218529256072,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010344854485448545,
      "loss": 3.3015,
      "step": 482810
    },
    {
      "epoch": 0.2094008285287189,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010344654465446546,
      "loss": 3.0651,
      "step": 482820
    },
    {
      "epoch": 0.2094194717648771,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010344454445444544,
      "loss": 3.1641,
      "step": 482830
    },
    {
      "epoch": 0.2094381150010353,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010344254425442545,
      "loss": 3.198,
      "step": 482840
    },
    {
      "epoch": 0.20945675823719345,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010344054405440543,
      "loss": 3.1902,
      "step": 482850
    },
    {
      "epoch": 0.20947540147335164,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010343854385438544,
      "loss": 3.305,
      "step": 482860
    },
    {
      "epoch": 0.20949404470950983,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010343654365436544,
      "loss": 3.2395,
      "step": 482870
    },
    {
      "epoch": 0.20951268794566802,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010343454345434545,
      "loss": 3.242,
      "step": 482880
    },
    {
      "epoch": 0.2095313311818262,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010343254325432543,
      "loss": 3.2369,
      "step": 482890
    },
    {
      "epoch": 0.2095499744179844,
      "grad_norm": 0.875,
      "learning_rate": 0.00010343054305430544,
      "loss": 3.0448,
      "step": 482900
    },
    {
      "epoch": 0.20956861765414256,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010342854285428542,
      "loss": 3.1808,
      "step": 482910
    },
    {
      "epoch": 0.20958726089030075,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010342654265426543,
      "loss": 3.2313,
      "step": 482920
    },
    {
      "epoch": 0.20960590412645894,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010342454245424543,
      "loss": 3.2387,
      "step": 482930
    },
    {
      "epoch": 0.20962454736261713,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010342254225422544,
      "loss": 3.0779,
      "step": 482940
    },
    {
      "epoch": 0.20964319059877531,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010342054205420542,
      "loss": 3.142,
      "step": 482950
    },
    {
      "epoch": 0.20966183383493348,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010341854185418543,
      "loss": 3.2331,
      "step": 482960
    },
    {
      "epoch": 0.20968047707109166,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010341654165416541,
      "loss": 3.0965,
      "step": 482970
    },
    {
      "epoch": 0.20969912030724985,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010341454145414542,
      "loss": 3.2704,
      "step": 482980
    },
    {
      "epoch": 0.20971776354340804,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010341254125412543,
      "loss": 3.2685,
      "step": 482990
    },
    {
      "epoch": 0.20973640677956623,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010341054105410541,
      "loss": 3.2076,
      "step": 483000
    },
    {
      "epoch": 0.2097550500157244,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010340854085408542,
      "loss": 3.2717,
      "step": 483010
    },
    {
      "epoch": 0.20977369325188258,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010340654065406541,
      "loss": 3.1556,
      "step": 483020
    },
    {
      "epoch": 0.20979233648804077,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010340454045404542,
      "loss": 3.2414,
      "step": 483030
    },
    {
      "epoch": 0.20981097972419896,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001034025402540254,
      "loss": 3.2261,
      "step": 483040
    },
    {
      "epoch": 0.20982962296035715,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010340054005400541,
      "loss": 3.1729,
      "step": 483050
    },
    {
      "epoch": 0.2098482661965153,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001033985398539854,
      "loss": 3.2609,
      "step": 483060
    },
    {
      "epoch": 0.2098669094326735,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001033965396539654,
      "loss": 3.2056,
      "step": 483070
    },
    {
      "epoch": 0.2098855526688317,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010339453945394539,
      "loss": 3.3085,
      "step": 483080
    },
    {
      "epoch": 0.20990419590498988,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001033925392539254,
      "loss": 3.3508,
      "step": 483090
    },
    {
      "epoch": 0.20992283914114807,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010339053905390539,
      "loss": 3.0845,
      "step": 483100
    },
    {
      "epoch": 0.20994148237730625,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001033885388538854,
      "loss": 3.162,
      "step": 483110
    },
    {
      "epoch": 0.20996012561346442,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010338653865386538,
      "loss": 3.281,
      "step": 483120
    },
    {
      "epoch": 0.2099787688496226,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010338453845384539,
      "loss": 3.3077,
      "step": 483130
    },
    {
      "epoch": 0.2099974120857808,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010338253825382537,
      "loss": 3.1095,
      "step": 483140
    },
    {
      "epoch": 0.21001605532193898,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010338053805380538,
      "loss": 3.2209,
      "step": 483150
    },
    {
      "epoch": 0.21003469855809717,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010337853785378538,
      "loss": 3.1613,
      "step": 483160
    },
    {
      "epoch": 0.21005334179425533,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010337653765376538,
      "loss": 3.2311,
      "step": 483170
    },
    {
      "epoch": 0.21007198503041352,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010337453745374538,
      "loss": 3.2552,
      "step": 483180
    },
    {
      "epoch": 0.2100906282665717,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010337253725372538,
      "loss": 3.0471,
      "step": 483190
    },
    {
      "epoch": 0.2101092715027299,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010337053705370539,
      "loss": 3.2034,
      "step": 483200
    },
    {
      "epoch": 0.2101279147388881,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010336853685368537,
      "loss": 3.1909,
      "step": 483210
    },
    {
      "epoch": 0.21014655797504625,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010336653665366538,
      "loss": 3.1841,
      "step": 483220
    },
    {
      "epoch": 0.21016520121120444,
      "grad_norm": 0.984375,
      "learning_rate": 0.00010336453645364536,
      "loss": 3.1838,
      "step": 483230
    },
    {
      "epoch": 0.21018384444736263,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010336253625362537,
      "loss": 3.1054,
      "step": 483240
    },
    {
      "epoch": 0.21020248768352082,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010336053605360537,
      "loss": 3.1958,
      "step": 483250
    },
    {
      "epoch": 0.210221130919679,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010335853585358538,
      "loss": 3.1702,
      "step": 483260
    },
    {
      "epoch": 0.21023977415583717,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010335653565356536,
      "loss": 3.0486,
      "step": 483270
    },
    {
      "epoch": 0.21025841739199536,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010335453545354537,
      "loss": 3.1262,
      "step": 483280
    },
    {
      "epoch": 0.21027706062815354,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010335253525352535,
      "loss": 3.2001,
      "step": 483290
    },
    {
      "epoch": 0.21029570386431173,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010335053505350536,
      "loss": 3.1565,
      "step": 483300
    },
    {
      "epoch": 0.21031434710046992,
      "grad_norm": 0.75,
      "learning_rate": 0.00010334853485348534,
      "loss": 3.1968,
      "step": 483310
    },
    {
      "epoch": 0.2103329903366281,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010334653465346535,
      "loss": 3.0607,
      "step": 483320
    },
    {
      "epoch": 0.21035163357278627,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010334453445344535,
      "loss": 3.1846,
      "step": 483330
    },
    {
      "epoch": 0.21037027680894446,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010334253425342535,
      "loss": 3.0759,
      "step": 483340
    },
    {
      "epoch": 0.21038892004510265,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010334053405340534,
      "loss": 3.1305,
      "step": 483350
    },
    {
      "epoch": 0.21040756328126084,
      "grad_norm": 0.875,
      "learning_rate": 0.00010333853385338535,
      "loss": 3.2543,
      "step": 483360
    },
    {
      "epoch": 0.21042620651741903,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010333653365336533,
      "loss": 3.3536,
      "step": 483370
    },
    {
      "epoch": 0.2104448497535772,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010333453345334534,
      "loss": 3.1491,
      "step": 483380
    },
    {
      "epoch": 0.21046349298973538,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010333253325332535,
      "loss": 3.2207,
      "step": 483390
    },
    {
      "epoch": 0.21048213622589357,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010333053305330533,
      "loss": 3.2131,
      "step": 483400
    },
    {
      "epoch": 0.21050077946205176,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010332853285328534,
      "loss": 3.2628,
      "step": 483410
    },
    {
      "epoch": 0.21051942269820995,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010332653265326533,
      "loss": 3.1502,
      "step": 483420
    },
    {
      "epoch": 0.2105380659343681,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010332453245324534,
      "loss": 3.1904,
      "step": 483430
    },
    {
      "epoch": 0.2105567091705263,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010332253225322532,
      "loss": 3.226,
      "step": 483440
    },
    {
      "epoch": 0.21057535240668449,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010332053205320533,
      "loss": 3.2114,
      "step": 483450
    },
    {
      "epoch": 0.21059399564284267,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010331853185318532,
      "loss": 3.2262,
      "step": 483460
    },
    {
      "epoch": 0.21061263887900086,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010331653165316532,
      "loss": 3.2075,
      "step": 483470
    },
    {
      "epoch": 0.21063128211515902,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010331453145314532,
      "loss": 3.2196,
      "step": 483480
    },
    {
      "epoch": 0.2106499253513172,
      "grad_norm": 0.75,
      "learning_rate": 0.00010331253125312532,
      "loss": 3.1868,
      "step": 483490
    },
    {
      "epoch": 0.2106685685874754,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010331053105310531,
      "loss": 3.1729,
      "step": 483500
    },
    {
      "epoch": 0.2106872118236336,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010330853085308532,
      "loss": 3.2475,
      "step": 483510
    },
    {
      "epoch": 0.21070585505979178,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001033065306530653,
      "loss": 3.2655,
      "step": 483520
    },
    {
      "epoch": 0.21072449829594994,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010330453045304531,
      "loss": 3.1585,
      "step": 483530
    },
    {
      "epoch": 0.21074314153210813,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001033025302530253,
      "loss": 3.3002,
      "step": 483540
    },
    {
      "epoch": 0.21076178476826632,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001033005300530053,
      "loss": 3.2859,
      "step": 483550
    },
    {
      "epoch": 0.2107804280044245,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001032985298529853,
      "loss": 3.182,
      "step": 483560
    },
    {
      "epoch": 0.2107990712405827,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010329652965296531,
      "loss": 3.4285,
      "step": 483570
    },
    {
      "epoch": 0.2108177144767409,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00010329452945294532,
      "loss": 3.1165,
      "step": 483580
    },
    {
      "epoch": 0.21083635771289905,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001032925292529253,
      "loss": 3.3225,
      "step": 483590
    },
    {
      "epoch": 0.21085500094905724,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010329052905290531,
      "loss": 3.3141,
      "step": 483600
    },
    {
      "epoch": 0.21087364418521543,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010328852885288529,
      "loss": 3.1973,
      "step": 483610
    },
    {
      "epoch": 0.21089228742137361,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001032865286528653,
      "loss": 3.2824,
      "step": 483620
    },
    {
      "epoch": 0.2109109306575318,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010328452845284528,
      "loss": 3.1407,
      "step": 483630
    },
    {
      "epoch": 0.21092957389368996,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010328252825282529,
      "loss": 3.1539,
      "step": 483640
    },
    {
      "epoch": 0.21094821712984815,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010328052805280529,
      "loss": 3.1385,
      "step": 483650
    },
    {
      "epoch": 0.21096686036600634,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001032785278527853,
      "loss": 3.1512,
      "step": 483660
    },
    {
      "epoch": 0.21098550360216453,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010327652765276528,
      "loss": 3.2136,
      "step": 483670
    },
    {
      "epoch": 0.21100414683832272,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010327452745274529,
      "loss": 3.2091,
      "step": 483680
    },
    {
      "epoch": 0.21102279007448088,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010327252725272527,
      "loss": 3.307,
      "step": 483690
    },
    {
      "epoch": 0.21104143331063907,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010327052705270528,
      "loss": 3.2482,
      "step": 483700
    },
    {
      "epoch": 0.21106007654679726,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010326852685268526,
      "loss": 3.242,
      "step": 483710
    },
    {
      "epoch": 0.21107871978295545,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010326652665266527,
      "loss": 3.2971,
      "step": 483720
    },
    {
      "epoch": 0.21109736301911364,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010326452645264526,
      "loss": 3.1707,
      "step": 483730
    },
    {
      "epoch": 0.2111160062552718,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010326252625262527,
      "loss": 3.255,
      "step": 483740
    },
    {
      "epoch": 0.21113464949143,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010326052605260526,
      "loss": 3.2786,
      "step": 483750
    },
    {
      "epoch": 0.21115329272758818,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010325852585258526,
      "loss": 3.1572,
      "step": 483760
    },
    {
      "epoch": 0.21117193596374637,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010325652565256527,
      "loss": 3.1719,
      "step": 483770
    },
    {
      "epoch": 0.21119057919990455,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010325452545254526,
      "loss": 3.1599,
      "step": 483780
    },
    {
      "epoch": 0.21120922243606274,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010325252525252526,
      "loss": 3.17,
      "step": 483790
    },
    {
      "epoch": 0.2112278656722209,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010325052505250526,
      "loss": 3.22,
      "step": 483800
    },
    {
      "epoch": 0.2112465089083791,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010324852485248526,
      "loss": 3.1122,
      "step": 483810
    },
    {
      "epoch": 0.21126515214453728,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010324652465246525,
      "loss": 3.305,
      "step": 483820
    },
    {
      "epoch": 0.21128379538069547,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010324452445244526,
      "loss": 3.2555,
      "step": 483830
    },
    {
      "epoch": 0.21130243861685366,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010324252425242524,
      "loss": 3.3246,
      "step": 483840
    },
    {
      "epoch": 0.21132108185301182,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010324052405240525,
      "loss": 3.0924,
      "step": 483850
    },
    {
      "epoch": 0.21133972508917,
      "grad_norm": 0.75,
      "learning_rate": 0.00010323852385238523,
      "loss": 3.1979,
      "step": 483860
    },
    {
      "epoch": 0.2113583683253282,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010323652365236524,
      "loss": 3.2445,
      "step": 483870
    },
    {
      "epoch": 0.2113770115614864,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010323452345234524,
      "loss": 3.1908,
      "step": 483880
    },
    {
      "epoch": 0.21139565479764458,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010323252325232525,
      "loss": 3.2132,
      "step": 483890
    },
    {
      "epoch": 0.21141429803380274,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010323052305230523,
      "loss": 3.1194,
      "step": 483900
    },
    {
      "epoch": 0.21143294126996093,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010322852285228524,
      "loss": 3.315,
      "step": 483910
    },
    {
      "epoch": 0.21145158450611912,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010322652265226522,
      "loss": 3.2098,
      "step": 483920
    },
    {
      "epoch": 0.2114702277422773,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010322452245224523,
      "loss": 3.2723,
      "step": 483930
    },
    {
      "epoch": 0.2114888709784355,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010322252225222521,
      "loss": 3.2032,
      "step": 483940
    },
    {
      "epoch": 0.21150751421459366,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010322052205220522,
      "loss": 3.0563,
      "step": 483950
    },
    {
      "epoch": 0.21152615745075184,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010321852185218522,
      "loss": 3.2572,
      "step": 483960
    },
    {
      "epoch": 0.21154480068691003,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010321652165216523,
      "loss": 3.2525,
      "step": 483970
    },
    {
      "epoch": 0.21156344392306822,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010321452145214524,
      "loss": 3.1711,
      "step": 483980
    },
    {
      "epoch": 0.2115820871592264,
      "grad_norm": 0.5625,
      "learning_rate": 0.00010321252125212522,
      "loss": 3.2736,
      "step": 483990
    },
    {
      "epoch": 0.2116007303953846,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010321052105210523,
      "loss": 3.2026,
      "step": 484000
    },
    {
      "epoch": 0.21161937363154276,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010320852085208521,
      "loss": 3.158,
      "step": 484010
    },
    {
      "epoch": 0.21163801686770095,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010320652065206522,
      "loss": 3.1782,
      "step": 484020
    },
    {
      "epoch": 0.21165666010385914,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001032045204520452,
      "loss": 3.3236,
      "step": 484030
    },
    {
      "epoch": 0.21167530334001733,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010320252025202521,
      "loss": 3.2026,
      "step": 484040
    },
    {
      "epoch": 0.21169394657617552,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001032005200520052,
      "loss": 3.2859,
      "step": 484050
    },
    {
      "epoch": 0.21171258981233368,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010319851985198521,
      "loss": 3.2185,
      "step": 484060
    },
    {
      "epoch": 0.21173123304849187,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001031965196519652,
      "loss": 3.289,
      "step": 484070
    },
    {
      "epoch": 0.21174987628465006,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001031945194519452,
      "loss": 3.2689,
      "step": 484080
    },
    {
      "epoch": 0.21176851952080825,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010319251925192519,
      "loss": 3.3538,
      "step": 484090
    },
    {
      "epoch": 0.21178716275696644,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001031905190519052,
      "loss": 3.2917,
      "step": 484100
    },
    {
      "epoch": 0.2118058059931246,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010318851885188519,
      "loss": 3.2352,
      "step": 484110
    },
    {
      "epoch": 0.21182444922928279,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001031865186518652,
      "loss": 3.1948,
      "step": 484120
    },
    {
      "epoch": 0.21184309246544097,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010318451845184518,
      "loss": 3.1804,
      "step": 484130
    },
    {
      "epoch": 0.21186173570159916,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010318251825182519,
      "loss": 3.3378,
      "step": 484140
    },
    {
      "epoch": 0.21188037893775735,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010318051805180517,
      "loss": 3.1672,
      "step": 484150
    },
    {
      "epoch": 0.2118990221739155,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010317851785178518,
      "loss": 3.1801,
      "step": 484160
    },
    {
      "epoch": 0.2119176654100737,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010317651765176519,
      "loss": 3.2424,
      "step": 484170
    },
    {
      "epoch": 0.2119363086462319,
      "grad_norm": 0.75,
      "learning_rate": 0.00010317451745174517,
      "loss": 3.1403,
      "step": 484180
    },
    {
      "epoch": 0.21195495188239008,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010317251725172518,
      "loss": 3.1419,
      "step": 484190
    },
    {
      "epoch": 0.21197359511854827,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010317051705170518,
      "loss": 3.114,
      "step": 484200
    },
    {
      "epoch": 0.21199223835470643,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010316851685168519,
      "loss": 3.1053,
      "step": 484210
    },
    {
      "epoch": 0.21201088159086462,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010316651665166517,
      "loss": 3.2074,
      "step": 484220
    },
    {
      "epoch": 0.2120295248270228,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010316451645164518,
      "loss": 3.3069,
      "step": 484230
    },
    {
      "epoch": 0.212048168063181,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010316251625162516,
      "loss": 3.238,
      "step": 484240
    },
    {
      "epoch": 0.2120668112993392,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010316051605160517,
      "loss": 3.2294,
      "step": 484250
    },
    {
      "epoch": 0.21208545453549738,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010315851585158515,
      "loss": 3.1936,
      "step": 484260
    },
    {
      "epoch": 0.21210409777165554,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010315651565156516,
      "loss": 3.2009,
      "step": 484270
    },
    {
      "epoch": 0.21212274100781373,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010315451545154516,
      "loss": 3.0826,
      "step": 484280
    },
    {
      "epoch": 0.21214138424397191,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010315251525152517,
      "loss": 3.1843,
      "step": 484290
    },
    {
      "epoch": 0.2121600274801301,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010315051505150515,
      "loss": 3.1956,
      "step": 484300
    },
    {
      "epoch": 0.2121786707162883,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010314851485148516,
      "loss": 3.1314,
      "step": 484310
    },
    {
      "epoch": 0.21219731395244645,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010314651465146514,
      "loss": 3.1265,
      "step": 484320
    },
    {
      "epoch": 0.21221595718860464,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010314451445144515,
      "loss": 3.1905,
      "step": 484330
    },
    {
      "epoch": 0.21223460042476283,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010314251425142513,
      "loss": 3.2281,
      "step": 484340
    },
    {
      "epoch": 0.21225324366092102,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010314051405140514,
      "loss": 3.081,
      "step": 484350
    },
    {
      "epoch": 0.2122718868970792,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010313851385138515,
      "loss": 3.2462,
      "step": 484360
    },
    {
      "epoch": 0.21229053013323737,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010313651365136514,
      "loss": 3.2233,
      "step": 484370
    },
    {
      "epoch": 0.21230917336939556,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010313451345134515,
      "loss": 3.1408,
      "step": 484380
    },
    {
      "epoch": 0.21232781660555375,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010313251325132514,
      "loss": 3.247,
      "step": 484390
    },
    {
      "epoch": 0.21234645984171194,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010313051305130514,
      "loss": 3.0977,
      "step": 484400
    },
    {
      "epoch": 0.21236510307787013,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010312851285128513,
      "loss": 3.2248,
      "step": 484410
    },
    {
      "epoch": 0.2123837463140283,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010312651265126514,
      "loss": 3.2069,
      "step": 484420
    },
    {
      "epoch": 0.21240238955018648,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010312451245124513,
      "loss": 3.1802,
      "step": 484430
    },
    {
      "epoch": 0.21242103278634467,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010312251225122513,
      "loss": 3.2444,
      "step": 484440
    },
    {
      "epoch": 0.21243967602250285,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010312051205120512,
      "loss": 3.2907,
      "step": 484450
    },
    {
      "epoch": 0.21245831925866104,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010311851185118513,
      "loss": 3.1635,
      "step": 484460
    },
    {
      "epoch": 0.21247696249481923,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010311651165116511,
      "loss": 3.33,
      "step": 484470
    },
    {
      "epoch": 0.2124956057309774,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010311451145114512,
      "loss": 3.2313,
      "step": 484480
    },
    {
      "epoch": 0.21251424896713558,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001031125112511251,
      "loss": 3.2476,
      "step": 484490
    },
    {
      "epoch": 0.21253289220329377,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010311051105110511,
      "loss": 3.2199,
      "step": 484500
    },
    {
      "epoch": 0.21255153543945196,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010310851085108511,
      "loss": 3.2649,
      "step": 484510
    },
    {
      "epoch": 0.21257017867561015,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010310651065106512,
      "loss": 3.1286,
      "step": 484520
    },
    {
      "epoch": 0.2125888219117683,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001031045104510451,
      "loss": 3.2099,
      "step": 484530
    },
    {
      "epoch": 0.2126074651479265,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010310251025102511,
      "loss": 3.2486,
      "step": 484540
    },
    {
      "epoch": 0.2126261083840847,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010310051005100509,
      "loss": 3.2539,
      "step": 484550
    },
    {
      "epoch": 0.21264475162024288,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001030985098509851,
      "loss": 3.125,
      "step": 484560
    },
    {
      "epoch": 0.21266339485640107,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010309650965096511,
      "loss": 3.356,
      "step": 484570
    },
    {
      "epoch": 0.21268203809255923,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010309450945094509,
      "loss": 3.1963,
      "step": 484580
    },
    {
      "epoch": 0.21270068132871742,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001030925092509251,
      "loss": 3.2123,
      "step": 484590
    },
    {
      "epoch": 0.2127193245648756,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001030905090509051,
      "loss": 3.2464,
      "step": 484600
    },
    {
      "epoch": 0.2127379678010338,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001030885088508851,
      "loss": 3.264,
      "step": 484610
    },
    {
      "epoch": 0.21275661103719198,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010308650865086509,
      "loss": 3.3256,
      "step": 484620
    },
    {
      "epoch": 0.21277525427335015,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001030845084508451,
      "loss": 3.0435,
      "step": 484630
    },
    {
      "epoch": 0.21279389750950833,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010308250825082508,
      "loss": 3.1172,
      "step": 484640
    },
    {
      "epoch": 0.21281254074566652,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010308050805080509,
      "loss": 3.2454,
      "step": 484650
    },
    {
      "epoch": 0.2128311839818247,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010307850785078507,
      "loss": 3.1095,
      "step": 484660
    },
    {
      "epoch": 0.2128498272179829,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010307650765076508,
      "loss": 3.2272,
      "step": 484670
    },
    {
      "epoch": 0.2128684704541411,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010307450745074508,
      "loss": 3.1825,
      "step": 484680
    },
    {
      "epoch": 0.21288711369029925,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010307250725072508,
      "loss": 3.1871,
      "step": 484690
    },
    {
      "epoch": 0.21290575692645744,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010307050705070507,
      "loss": 3.1455,
      "step": 484700
    },
    {
      "epoch": 0.21292440016261563,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010306850685068508,
      "loss": 3.1606,
      "step": 484710
    },
    {
      "epoch": 0.21294304339877382,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010306650665066506,
      "loss": 3.2621,
      "step": 484720
    },
    {
      "epoch": 0.212961686634932,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010306450645064507,
      "loss": 3.1321,
      "step": 484730
    },
    {
      "epoch": 0.21298032987109017,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010306250625062506,
      "loss": 3.2264,
      "step": 484740
    },
    {
      "epoch": 0.21299897310724836,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010306050605060507,
      "loss": 3.3818,
      "step": 484750
    },
    {
      "epoch": 0.21301761634340655,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010305850585058507,
      "loss": 3.1845,
      "step": 484760
    },
    {
      "epoch": 0.21303625957956474,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010305650565056506,
      "loss": 3.3125,
      "step": 484770
    },
    {
      "epoch": 0.21305490281572292,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010305450545054507,
      "loss": 3.1448,
      "step": 484780
    },
    {
      "epoch": 0.21307354605188109,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010305250525052505,
      "loss": 3.2909,
      "step": 484790
    },
    {
      "epoch": 0.21309218928803927,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010305050505050506,
      "loss": 3.255,
      "step": 484800
    },
    {
      "epoch": 0.21311083252419746,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010304850485048505,
      "loss": 3.292,
      "step": 484810
    },
    {
      "epoch": 0.21312947576035565,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010304650465046505,
      "loss": 3.2623,
      "step": 484820
    },
    {
      "epoch": 0.21314811899651384,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010304450445044505,
      "loss": 3.2075,
      "step": 484830
    },
    {
      "epoch": 0.213166762232672,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010304250425042506,
      "loss": 3.2102,
      "step": 484840
    },
    {
      "epoch": 0.2131854054688302,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010304050405040504,
      "loss": 3.2366,
      "step": 484850
    },
    {
      "epoch": 0.21320404870498838,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010303850385038505,
      "loss": 3.2824,
      "step": 484860
    },
    {
      "epoch": 0.21322269194114657,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010303650365036503,
      "loss": 3.1331,
      "step": 484870
    },
    {
      "epoch": 0.21324133517730476,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010303450345034504,
      "loss": 3.2561,
      "step": 484880
    },
    {
      "epoch": 0.21325997841346292,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010303250325032502,
      "loss": 3.0893,
      "step": 484890
    },
    {
      "epoch": 0.2132786216496211,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010303050305030503,
      "loss": 3.3036,
      "step": 484900
    },
    {
      "epoch": 0.2132972648857793,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00010302850285028503,
      "loss": 3.1864,
      "step": 484910
    },
    {
      "epoch": 0.2133159081219375,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010302650265026504,
      "loss": 3.181,
      "step": 484920
    },
    {
      "epoch": 0.21333455135809568,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010302450245024502,
      "loss": 3.1826,
      "step": 484930
    },
    {
      "epoch": 0.21335319459425386,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010302250225022503,
      "loss": 3.2113,
      "step": 484940
    },
    {
      "epoch": 0.21337183783041203,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010302050205020501,
      "loss": 3.2535,
      "step": 484950
    },
    {
      "epoch": 0.21339048106657021,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010301850185018502,
      "loss": 3.259,
      "step": 484960
    },
    {
      "epoch": 0.2134091243027284,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010301650165016503,
      "loss": 3.1582,
      "step": 484970
    },
    {
      "epoch": 0.2134277675388866,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010301450145014501,
      "loss": 3.1809,
      "step": 484980
    },
    {
      "epoch": 0.21344641077504478,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010301250125012502,
      "loss": 3.1898,
      "step": 484990
    },
    {
      "epoch": 0.21346505401120294,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010301050105010502,
      "loss": 3.0512,
      "step": 485000
    },
    {
      "epoch": 0.21348369724736113,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010300850085008503,
      "loss": 3.26,
      "step": 485010
    },
    {
      "epoch": 0.21350234048351932,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010300650065006501,
      "loss": 3.0457,
      "step": 485020
    },
    {
      "epoch": 0.2135209837196775,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010300450045004502,
      "loss": 3.1953,
      "step": 485030
    },
    {
      "epoch": 0.2135396269558357,
      "grad_norm": 0.703125,
      "learning_rate": 0.000103002500250025,
      "loss": 3.2487,
      "step": 485040
    },
    {
      "epoch": 0.21355827019199386,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010300050005000501,
      "loss": 3.3154,
      "step": 485050
    },
    {
      "epoch": 0.21357691342815205,
      "grad_norm": 0.78515625,
      "learning_rate": 0.000102998499849985,
      "loss": 3.2837,
      "step": 485060
    },
    {
      "epoch": 0.21359555666431024,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010299649964996501,
      "loss": 3.2314,
      "step": 485070
    },
    {
      "epoch": 0.21361419990046843,
      "grad_norm": 0.78125,
      "learning_rate": 0.000102994499449945,
      "loss": 3.253,
      "step": 485080
    },
    {
      "epoch": 0.21363284313662662,
      "grad_norm": 0.703125,
      "learning_rate": 0.000102992499249925,
      "loss": 3.22,
      "step": 485090
    },
    {
      "epoch": 0.21365148637278478,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010299049904990499,
      "loss": 3.0619,
      "step": 485100
    },
    {
      "epoch": 0.21367012960894297,
      "grad_norm": 0.71875,
      "learning_rate": 0.000102988498849885,
      "loss": 3.2495,
      "step": 485110
    },
    {
      "epoch": 0.21368877284510115,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010298649864986498,
      "loss": 3.2178,
      "step": 485120
    },
    {
      "epoch": 0.21370741608125934,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010298449844984499,
      "loss": 3.2685,
      "step": 485130
    },
    {
      "epoch": 0.21372605931741753,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010298249824982498,
      "loss": 3.1499,
      "step": 485140
    },
    {
      "epoch": 0.21374470255357572,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010298049804980499,
      "loss": 3.251,
      "step": 485150
    },
    {
      "epoch": 0.21376334578973388,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000102978497849785,
      "loss": 3.1346,
      "step": 485160
    },
    {
      "epoch": 0.21378198902589207,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010297649764976498,
      "loss": 3.2259,
      "step": 485170
    },
    {
      "epoch": 0.21380063226205026,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010297449744974499,
      "loss": 3.238,
      "step": 485180
    },
    {
      "epoch": 0.21381927549820845,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010297249724972497,
      "loss": 3.2329,
      "step": 485190
    },
    {
      "epoch": 0.21383791873436664,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010297049704970498,
      "loss": 3.2463,
      "step": 485200
    },
    {
      "epoch": 0.2138565619705248,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010296849684968496,
      "loss": 3.2422,
      "step": 485210
    },
    {
      "epoch": 0.213875205206683,
      "grad_norm": 0.9375,
      "learning_rate": 0.00010296649664966497,
      "loss": 3.1178,
      "step": 485220
    },
    {
      "epoch": 0.21389384844284118,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010296449644964497,
      "loss": 3.1948,
      "step": 485230
    },
    {
      "epoch": 0.21391249167899937,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010296249624962498,
      "loss": 3.2162,
      "step": 485240
    },
    {
      "epoch": 0.21393113491515756,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010296049604960496,
      "loss": 3.2808,
      "step": 485250
    },
    {
      "epoch": 0.21394977815131572,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010295849584958497,
      "loss": 3.2104,
      "step": 485260
    },
    {
      "epoch": 0.2139684213874739,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010295649564956495,
      "loss": 3.2695,
      "step": 485270
    },
    {
      "epoch": 0.2139870646236321,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010295449544954496,
      "loss": 3.2002,
      "step": 485280
    },
    {
      "epoch": 0.21400570785979028,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010295249524952496,
      "loss": 3.3184,
      "step": 485290
    },
    {
      "epoch": 0.21402435109594847,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010295049504950495,
      "loss": 3.2416,
      "step": 485300
    },
    {
      "epoch": 0.21404299433210663,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010294849484948495,
      "loss": 3.2078,
      "step": 485310
    },
    {
      "epoch": 0.21406163756826482,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010294649464946496,
      "loss": 3.0997,
      "step": 485320
    },
    {
      "epoch": 0.214080280804423,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010294449444944494,
      "loss": 3.0641,
      "step": 485330
    },
    {
      "epoch": 0.2140989240405812,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010294249424942495,
      "loss": 3.2609,
      "step": 485340
    },
    {
      "epoch": 0.2141175672767394,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010294049404940496,
      "loss": 3.241,
      "step": 485350
    },
    {
      "epoch": 0.21413621051289758,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010293849384938494,
      "loss": 3.2713,
      "step": 485360
    },
    {
      "epoch": 0.21415485374905574,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010293649364936495,
      "loss": 3.1397,
      "step": 485370
    },
    {
      "epoch": 0.21417349698521393,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010293449344934494,
      "loss": 3.1453,
      "step": 485380
    },
    {
      "epoch": 0.21419214022137212,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010293249324932495,
      "loss": 3.3403,
      "step": 485390
    },
    {
      "epoch": 0.2142107834575303,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010293049304930493,
      "loss": 3.1806,
      "step": 485400
    },
    {
      "epoch": 0.2142294266936885,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010292849284928494,
      "loss": 3.2175,
      "step": 485410
    },
    {
      "epoch": 0.21424806992984666,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010292649264926493,
      "loss": 3.1964,
      "step": 485420
    },
    {
      "epoch": 0.21426671316600485,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010292449244924494,
      "loss": 3.3682,
      "step": 485430
    },
    {
      "epoch": 0.21428535640216304,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010292249224922492,
      "loss": 3.0987,
      "step": 485440
    },
    {
      "epoch": 0.21430399963832122,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010292049204920493,
      "loss": 3.2375,
      "step": 485450
    },
    {
      "epoch": 0.2143226428744794,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010291849184918492,
      "loss": 3.1658,
      "step": 485460
    },
    {
      "epoch": 0.21434128611063757,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010291649164916493,
      "loss": 3.1939,
      "step": 485470
    },
    {
      "epoch": 0.21435992934679576,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010291449144914491,
      "loss": 3.2691,
      "step": 485480
    },
    {
      "epoch": 0.21437857258295395,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010291249124912492,
      "loss": 3.2528,
      "step": 485490
    },
    {
      "epoch": 0.21439721581911214,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001029104910491049,
      "loss": 3.2202,
      "step": 485500
    },
    {
      "epoch": 0.21441585905527033,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010290849084908491,
      "loss": 3.2061,
      "step": 485510
    },
    {
      "epoch": 0.2144345022914285,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001029064906490649,
      "loss": 3.2234,
      "step": 485520
    },
    {
      "epoch": 0.21445314552758668,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001029044904490449,
      "loss": 3.2663,
      "step": 485530
    },
    {
      "epoch": 0.21447178876374487,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001029024902490249,
      "loss": 3.168,
      "step": 485540
    },
    {
      "epoch": 0.21449043199990306,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010290049004900491,
      "loss": 3.214,
      "step": 485550
    },
    {
      "epoch": 0.21450907523606125,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010289848984898492,
      "loss": 3.1643,
      "step": 485560
    },
    {
      "epoch": 0.21452771847221944,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001028964896489649,
      "loss": 3.1493,
      "step": 485570
    },
    {
      "epoch": 0.2145463617083776,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010289448944894491,
      "loss": 3.176,
      "step": 485580
    },
    {
      "epoch": 0.2145650049445358,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010289248924892489,
      "loss": 3.1137,
      "step": 485590
    },
    {
      "epoch": 0.21458364818069398,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001028904890489049,
      "loss": 3.3262,
      "step": 485600
    },
    {
      "epoch": 0.21460229141685216,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001028884888488849,
      "loss": 3.1193,
      "step": 485610
    },
    {
      "epoch": 0.21462093465301035,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010288648864886489,
      "loss": 3.1309,
      "step": 485620
    },
    {
      "epoch": 0.21463957788916851,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010288448844884489,
      "loss": 3.3117,
      "step": 485630
    },
    {
      "epoch": 0.2146582211253267,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001028824882488249,
      "loss": 3.2382,
      "step": 485640
    },
    {
      "epoch": 0.2146768643614849,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010288048804880488,
      "loss": 3.2335,
      "step": 485650
    },
    {
      "epoch": 0.21469550759764308,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010287848784878489,
      "loss": 3.1298,
      "step": 485660
    },
    {
      "epoch": 0.21471415083380127,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010287648764876487,
      "loss": 3.1729,
      "step": 485670
    },
    {
      "epoch": 0.21473279406995943,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010287448744874488,
      "loss": 3.176,
      "step": 485680
    },
    {
      "epoch": 0.21475143730611762,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010287248724872487,
      "loss": 3.2668,
      "step": 485690
    },
    {
      "epoch": 0.2147700805422758,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010287048704870488,
      "loss": 3.0549,
      "step": 485700
    },
    {
      "epoch": 0.214788723778434,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010286848684868487,
      "loss": 3.2352,
      "step": 485710
    },
    {
      "epoch": 0.2148073670145922,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010286648664866488,
      "loss": 3.2369,
      "step": 485720
    },
    {
      "epoch": 0.21482601025075035,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010286448644864486,
      "loss": 3.2337,
      "step": 485730
    },
    {
      "epoch": 0.21484465348690854,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010286248624862487,
      "loss": 3.2853,
      "step": 485740
    },
    {
      "epoch": 0.21486329672306673,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010286048604860488,
      "loss": 3.3652,
      "step": 485750
    },
    {
      "epoch": 0.21488193995922492,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010285848584858486,
      "loss": 3.2874,
      "step": 485760
    },
    {
      "epoch": 0.2149005831953831,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010285648564856487,
      "loss": 3.1871,
      "step": 485770
    },
    {
      "epoch": 0.21491922643154127,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010285448544854486,
      "loss": 3.1021,
      "step": 485780
    },
    {
      "epoch": 0.21493786966769945,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010285248524852487,
      "loss": 3.1866,
      "step": 485790
    },
    {
      "epoch": 0.21495651290385764,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010285048504850485,
      "loss": 3.3136,
      "step": 485800
    },
    {
      "epoch": 0.21497515614001583,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010284848484848486,
      "loss": 3.2266,
      "step": 485810
    },
    {
      "epoch": 0.21499379937617402,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010284648464846484,
      "loss": 3.1844,
      "step": 485820
    },
    {
      "epoch": 0.2150124426123322,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010284448444844485,
      "loss": 3.2039,
      "step": 485830
    },
    {
      "epoch": 0.21503108584849037,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010284248424842484,
      "loss": 3.1843,
      "step": 485840
    },
    {
      "epoch": 0.21504972908464856,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010284048404840484,
      "loss": 3.1654,
      "step": 485850
    },
    {
      "epoch": 0.21506837232080675,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010283848384838484,
      "loss": 3.1445,
      "step": 485860
    },
    {
      "epoch": 0.21508701555696494,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010283648364836485,
      "loss": 3.2207,
      "step": 485870
    },
    {
      "epoch": 0.21510565879312313,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010283448344834483,
      "loss": 3.2421,
      "step": 485880
    },
    {
      "epoch": 0.2151243020292813,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010283248324832484,
      "loss": 3.3184,
      "step": 485890
    },
    {
      "epoch": 0.21514294526543948,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010283048304830482,
      "loss": 3.2665,
      "step": 485900
    },
    {
      "epoch": 0.21516158850159767,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010282848284828483,
      "loss": 3.1855,
      "step": 485910
    },
    {
      "epoch": 0.21518023173775586,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010282648264826483,
      "loss": 3.1091,
      "step": 485920
    },
    {
      "epoch": 0.21519887497391404,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010282448244824482,
      "loss": 3.1608,
      "step": 485930
    },
    {
      "epoch": 0.2152175182100722,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010282248224822483,
      "loss": 3.1474,
      "step": 485940
    },
    {
      "epoch": 0.2152361614462304,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010282048204820483,
      "loss": 3.1412,
      "step": 485950
    },
    {
      "epoch": 0.21525480468238858,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010281848184818484,
      "loss": 2.9016,
      "step": 485960
    },
    {
      "epoch": 0.21527344791854677,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010281648164816482,
      "loss": 3.2162,
      "step": 485970
    },
    {
      "epoch": 0.21529209115470496,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010281448144814483,
      "loss": 3.2093,
      "step": 485980
    },
    {
      "epoch": 0.21531073439086312,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010281248124812481,
      "loss": 3.2773,
      "step": 485990
    },
    {
      "epoch": 0.2153293776270213,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010281048104810482,
      "loss": 3.2565,
      "step": 486000
    },
    {
      "epoch": 0.2153480208631795,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010280848084808482,
      "loss": 3.1536,
      "step": 486010
    },
    {
      "epoch": 0.2153666640993377,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010280648064806482,
      "loss": 3.1573,
      "step": 486020
    },
    {
      "epoch": 0.21538530733549588,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001028044804480448,
      "loss": 3.308,
      "step": 486030
    },
    {
      "epoch": 0.21540395057165407,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010280248024802482,
      "loss": 3.2457,
      "step": 486040
    },
    {
      "epoch": 0.21542259380781223,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001028004800480048,
      "loss": 3.1376,
      "step": 486050
    },
    {
      "epoch": 0.21544123704397042,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001027984798479848,
      "loss": 3.3266,
      "step": 486060
    },
    {
      "epoch": 0.2154598802801286,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010279647964796479,
      "loss": 3.3279,
      "step": 486070
    },
    {
      "epoch": 0.2154785235162868,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001027944794479448,
      "loss": 3.1459,
      "step": 486080
    },
    {
      "epoch": 0.21549716675244499,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001027924792479248,
      "loss": 3.0911,
      "step": 486090
    },
    {
      "epoch": 0.21551580998860315,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001027904790479048,
      "loss": 3.1193,
      "step": 486100
    },
    {
      "epoch": 0.21553445322476134,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010278847884788478,
      "loss": 3.2423,
      "step": 486110
    },
    {
      "epoch": 0.21555309646091952,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001027864786478648,
      "loss": 3.1548,
      "step": 486120
    },
    {
      "epoch": 0.2155717396970777,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010278447844784478,
      "loss": 3.2433,
      "step": 486130
    },
    {
      "epoch": 0.2155903829332359,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010278247824782478,
      "loss": 3.2155,
      "step": 486140
    },
    {
      "epoch": 0.21560902616939406,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001027804780478048,
      "loss": 3.2172,
      "step": 486150
    },
    {
      "epoch": 0.21562766940555225,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010277847784778478,
      "loss": 3.225,
      "step": 486160
    },
    {
      "epoch": 0.21564631264171044,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010277647764776479,
      "loss": 3.2815,
      "step": 486170
    },
    {
      "epoch": 0.21566495587786863,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010277447744774478,
      "loss": 3.1672,
      "step": 486180
    },
    {
      "epoch": 0.21568359911402682,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010277247724772479,
      "loss": 3.2764,
      "step": 486190
    },
    {
      "epoch": 0.21570224235018498,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010277047704770477,
      "loss": 3.1259,
      "step": 486200
    },
    {
      "epoch": 0.21572088558634317,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010276847684768478,
      "loss": 3.1872,
      "step": 486210
    },
    {
      "epoch": 0.21573952882250136,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010276647664766476,
      "loss": 3.2086,
      "step": 486220
    },
    {
      "epoch": 0.21575817205865955,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010276447644764477,
      "loss": 3.161,
      "step": 486230
    },
    {
      "epoch": 0.21577681529481774,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010276247624762477,
      "loss": 3.2253,
      "step": 486240
    },
    {
      "epoch": 0.21579545853097593,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010276047604760476,
      "loss": 3.1493,
      "step": 486250
    },
    {
      "epoch": 0.2158141017671341,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010275847584758476,
      "loss": 3.2678,
      "step": 486260
    },
    {
      "epoch": 0.21583274500329228,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010275647564756477,
      "loss": 3.1984,
      "step": 486270
    },
    {
      "epoch": 0.21585138823945046,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010275447544754475,
      "loss": 3.2467,
      "step": 486280
    },
    {
      "epoch": 0.21587003147560865,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010275247524752476,
      "loss": 3.1186,
      "step": 486290
    },
    {
      "epoch": 0.21588867471176684,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010275047504750474,
      "loss": 3.0644,
      "step": 486300
    },
    {
      "epoch": 0.215907317947925,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010274847484748475,
      "loss": 3.1359,
      "step": 486310
    },
    {
      "epoch": 0.2159259611840832,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010274647464746475,
      "loss": 3.2381,
      "step": 486320
    },
    {
      "epoch": 0.21594460442024138,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010274447444744476,
      "loss": 3.2837,
      "step": 486330
    },
    {
      "epoch": 0.21596324765639957,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010274247424742476,
      "loss": 3.1694,
      "step": 486340
    },
    {
      "epoch": 0.21598189089255776,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010274047404740475,
      "loss": 3.2622,
      "step": 486350
    },
    {
      "epoch": 0.21600053412871592,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010273847384738476,
      "loss": 3.2814,
      "step": 486360
    },
    {
      "epoch": 0.2160191773648741,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010273647364736474,
      "loss": 3.1622,
      "step": 486370
    },
    {
      "epoch": 0.2160378206010323,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010273447344734475,
      "loss": 3.2165,
      "step": 486380
    },
    {
      "epoch": 0.2160564638371905,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010273247324732473,
      "loss": 3.1793,
      "step": 486390
    },
    {
      "epoch": 0.21607510707334868,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010273047304730474,
      "loss": 3.2323,
      "step": 486400
    },
    {
      "epoch": 0.21609375030950684,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010272847284728473,
      "loss": 3.131,
      "step": 486410
    },
    {
      "epoch": 0.21611239354566503,
      "grad_norm": 0.75,
      "learning_rate": 0.00010272647264726474,
      "loss": 3.237,
      "step": 486420
    },
    {
      "epoch": 0.21613103678182322,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010272447244724472,
      "loss": 3.1457,
      "step": 486430
    },
    {
      "epoch": 0.2161496800179814,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010272247224722473,
      "loss": 3.2028,
      "step": 486440
    },
    {
      "epoch": 0.2161683232541396,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010272047204720472,
      "loss": 3.123,
      "step": 486450
    },
    {
      "epoch": 0.21618696649029776,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010271847184718473,
      "loss": 3.2604,
      "step": 486460
    },
    {
      "epoch": 0.21620560972645594,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010271647164716471,
      "loss": 3.186,
      "step": 486470
    },
    {
      "epoch": 0.21622425296261413,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010271447144714472,
      "loss": 3.3092,
      "step": 486480
    },
    {
      "epoch": 0.21624289619877232,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010271247124712471,
      "loss": 3.2819,
      "step": 486490
    },
    {
      "epoch": 0.2162615394349305,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010271047104710472,
      "loss": 3.34,
      "step": 486500
    },
    {
      "epoch": 0.2162801826710887,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001027084708470847,
      "loss": 3.2221,
      "step": 486510
    },
    {
      "epoch": 0.21629882590724686,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010270647064706471,
      "loss": 3.2681,
      "step": 486520
    },
    {
      "epoch": 0.21631746914340505,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001027044704470447,
      "loss": 3.3519,
      "step": 486530
    },
    {
      "epoch": 0.21633611237956324,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001027024702470247,
      "loss": 3.3091,
      "step": 486540
    },
    {
      "epoch": 0.21635475561572143,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010270047004700471,
      "loss": 3.2293,
      "step": 486550
    },
    {
      "epoch": 0.21637339885187962,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010269846984698471,
      "loss": 3.185,
      "step": 486560
    },
    {
      "epoch": 0.21639204208803778,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001026964696469647,
      "loss": 3.2073,
      "step": 486570
    },
    {
      "epoch": 0.21641068532419597,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001026944694469447,
      "loss": 3.2188,
      "step": 486580
    },
    {
      "epoch": 0.21642932856035416,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010269246924692471,
      "loss": 3.141,
      "step": 486590
    },
    {
      "epoch": 0.21644797179651235,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010269046904690469,
      "loss": 3.2153,
      "step": 486600
    },
    {
      "epoch": 0.21646661503267053,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001026884688468847,
      "loss": 3.2621,
      "step": 486610
    },
    {
      "epoch": 0.2164852582688287,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010268646864686468,
      "loss": 3.2018,
      "step": 486620
    },
    {
      "epoch": 0.21650390150498688,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010268446844684469,
      "loss": 3.2763,
      "step": 486630
    },
    {
      "epoch": 0.21652254474114507,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010268246824682469,
      "loss": 3.1747,
      "step": 486640
    },
    {
      "epoch": 0.21654118797730326,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001026804680468047,
      "loss": 3.2801,
      "step": 486650
    },
    {
      "epoch": 0.21655983121346145,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010267846784678468,
      "loss": 3.141,
      "step": 486660
    },
    {
      "epoch": 0.2165784744496196,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010267646764676469,
      "loss": 2.9791,
      "step": 486670
    },
    {
      "epoch": 0.2165971176857778,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010267446744674467,
      "loss": 3.3031,
      "step": 486680
    },
    {
      "epoch": 0.216615760921936,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010267246724672468,
      "loss": 3.1953,
      "step": 486690
    },
    {
      "epoch": 0.21663440415809418,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010267046704670466,
      "loss": 3.3032,
      "step": 486700
    },
    {
      "epoch": 0.21665304739425237,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010266846684668467,
      "loss": 3.1645,
      "step": 486710
    },
    {
      "epoch": 0.21667169063041056,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010266646664666466,
      "loss": 3.1549,
      "step": 486720
    },
    {
      "epoch": 0.21669033386656872,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010266446644664467,
      "loss": 3.0617,
      "step": 486730
    },
    {
      "epoch": 0.2167089771027269,
      "grad_norm": 0.75,
      "learning_rate": 0.00010266246624662468,
      "loss": 3.2869,
      "step": 486740
    },
    {
      "epoch": 0.2167276203388851,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010266046604660467,
      "loss": 3.1948,
      "step": 486750
    },
    {
      "epoch": 0.21674626357504329,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010265846584658467,
      "loss": 3.0731,
      "step": 486760
    },
    {
      "epoch": 0.21676490681120147,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010265646564656466,
      "loss": 3.1584,
      "step": 486770
    },
    {
      "epoch": 0.21678355004735964,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010265446544654467,
      "loss": 3.1008,
      "step": 486780
    },
    {
      "epoch": 0.21680219328351782,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010265246524652465,
      "loss": 2.9863,
      "step": 486790
    },
    {
      "epoch": 0.216820836519676,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010265046504650466,
      "loss": 3.2693,
      "step": 486800
    },
    {
      "epoch": 0.2168394797558342,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010264846484648465,
      "loss": 3.2223,
      "step": 486810
    },
    {
      "epoch": 0.2168581229919924,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010264646464646466,
      "loss": 3.2222,
      "step": 486820
    },
    {
      "epoch": 0.21687676622815055,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010264446444644464,
      "loss": 3.2054,
      "step": 486830
    },
    {
      "epoch": 0.21689540946430874,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010264246424642465,
      "loss": 3.2217,
      "step": 486840
    },
    {
      "epoch": 0.21691405270046693,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010264046404640463,
      "loss": 3.3213,
      "step": 486850
    },
    {
      "epoch": 0.21693269593662512,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010263846384638464,
      "loss": 3.2575,
      "step": 486860
    },
    {
      "epoch": 0.2169513391727833,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010263646364636464,
      "loss": 3.2386,
      "step": 486870
    },
    {
      "epoch": 0.21696998240894147,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010263446344634465,
      "loss": 3.292,
      "step": 486880
    },
    {
      "epoch": 0.21698862564509966,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010263246324632463,
      "loss": 3.1412,
      "step": 486890
    },
    {
      "epoch": 0.21700726888125785,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010263046304630464,
      "loss": 3.194,
      "step": 486900
    },
    {
      "epoch": 0.21702591211741604,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010262846284628462,
      "loss": 3.3704,
      "step": 486910
    },
    {
      "epoch": 0.21704455535357423,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010262646264626463,
      "loss": 3.1581,
      "step": 486920
    },
    {
      "epoch": 0.21706319858973241,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010262446244624464,
      "loss": 3.422,
      "step": 486930
    },
    {
      "epoch": 0.21708184182589058,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010262246224622462,
      "loss": 3.2873,
      "step": 486940
    },
    {
      "epoch": 0.21710048506204876,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010262046204620463,
      "loss": 3.126,
      "step": 486950
    },
    {
      "epoch": 0.21711912829820695,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010261846184618463,
      "loss": 3.2652,
      "step": 486960
    },
    {
      "epoch": 0.21713777153436514,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010261646164616464,
      "loss": 3.2124,
      "step": 486970
    },
    {
      "epoch": 0.21715641477052333,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010261446144614462,
      "loss": 3.3699,
      "step": 486980
    },
    {
      "epoch": 0.2171750580066815,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010261246124612463,
      "loss": 3.2544,
      "step": 486990
    },
    {
      "epoch": 0.21719370124283968,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010261046104610461,
      "loss": 3.0151,
      "step": 487000
    },
    {
      "epoch": 0.21721234447899787,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010260846084608462,
      "loss": 3.1637,
      "step": 487010
    },
    {
      "epoch": 0.21723098771515606,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001026064606460646,
      "loss": 3.1521,
      "step": 487020
    },
    {
      "epoch": 0.21724963095131425,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010260446044604461,
      "loss": 3.1963,
      "step": 487030
    },
    {
      "epoch": 0.2172682741874724,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001026024602460246,
      "loss": 3.1076,
      "step": 487040
    },
    {
      "epoch": 0.2172869174236306,
      "grad_norm": 0.75,
      "learning_rate": 0.00010260046004600461,
      "loss": 3.3615,
      "step": 487050
    },
    {
      "epoch": 0.2173055606597888,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001025984598459846,
      "loss": 3.1261,
      "step": 487060
    },
    {
      "epoch": 0.21732420389594698,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001025964596459646,
      "loss": 3.1206,
      "step": 487070
    },
    {
      "epoch": 0.21734284713210517,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010259445944594459,
      "loss": 3.1959,
      "step": 487080
    },
    {
      "epoch": 0.21736149036826333,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001025924592459246,
      "loss": 3.2576,
      "step": 487090
    },
    {
      "epoch": 0.21738013360442152,
      "grad_norm": 0.75,
      "learning_rate": 0.00010259045904590458,
      "loss": 3.0745,
      "step": 487100
    },
    {
      "epoch": 0.2173987768405797,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010258845884588459,
      "loss": 3.2238,
      "step": 487110
    },
    {
      "epoch": 0.2174174200767379,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010258645864586458,
      "loss": 3.2067,
      "step": 487120
    },
    {
      "epoch": 0.21743606331289608,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010258445844584459,
      "loss": 3.2465,
      "step": 487130
    },
    {
      "epoch": 0.21745470654905424,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001025824582458246,
      "loss": 3.1778,
      "step": 487140
    },
    {
      "epoch": 0.21747334978521243,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010258045804580458,
      "loss": 3.3245,
      "step": 487150
    },
    {
      "epoch": 0.21749199302137062,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010257845784578459,
      "loss": 3.1426,
      "step": 487160
    },
    {
      "epoch": 0.2175106362575288,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010257645764576458,
      "loss": 3.2418,
      "step": 487170
    },
    {
      "epoch": 0.217529279493687,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010257445744574458,
      "loss": 3.1909,
      "step": 487180
    },
    {
      "epoch": 0.2175479227298452,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010257245724572458,
      "loss": 3.255,
      "step": 487190
    },
    {
      "epoch": 0.21756656596600335,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010257045704570458,
      "loss": 3.2201,
      "step": 487200
    },
    {
      "epoch": 0.21758520920216154,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010256845684568457,
      "loss": 3.2654,
      "step": 487210
    },
    {
      "epoch": 0.21760385243831973,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010256645664566458,
      "loss": 3.1752,
      "step": 487220
    },
    {
      "epoch": 0.21762249567447792,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010256445644564456,
      "loss": 3.2401,
      "step": 487230
    },
    {
      "epoch": 0.2176411389106361,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010256245624562457,
      "loss": 3.141,
      "step": 487240
    },
    {
      "epoch": 0.21765978214679427,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010256045604560455,
      "loss": 3.2224,
      "step": 487250
    },
    {
      "epoch": 0.21767842538295246,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010255845584558456,
      "loss": 3.2025,
      "step": 487260
    },
    {
      "epoch": 0.21769706861911065,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010255645564556456,
      "loss": 3.2508,
      "step": 487270
    },
    {
      "epoch": 0.21771571185526883,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010255445544554457,
      "loss": 3.1133,
      "step": 487280
    },
    {
      "epoch": 0.21773435509142702,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010255245524552455,
      "loss": 3.2812,
      "step": 487290
    },
    {
      "epoch": 0.21775299832758518,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010255045504550456,
      "loss": 3.2541,
      "step": 487300
    },
    {
      "epoch": 0.21777164156374337,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010254845484548454,
      "loss": 3.1861,
      "step": 487310
    },
    {
      "epoch": 0.21779028479990156,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010254645464546455,
      "loss": 3.1773,
      "step": 487320
    },
    {
      "epoch": 0.21780892803605975,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010254445444544456,
      "loss": 3.0995,
      "step": 487330
    },
    {
      "epoch": 0.21782757127221794,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010254245424542454,
      "loss": 3.2598,
      "step": 487340
    },
    {
      "epoch": 0.2178462145083761,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010254045404540455,
      "loss": 3.1437,
      "step": 487350
    },
    {
      "epoch": 0.2178648577445343,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010253845384538455,
      "loss": 3.2232,
      "step": 487360
    },
    {
      "epoch": 0.21788350098069248,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010253645364536455,
      "loss": 3.0762,
      "step": 487370
    },
    {
      "epoch": 0.21790214421685067,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010253445344534454,
      "loss": 3.2953,
      "step": 487380
    },
    {
      "epoch": 0.21792078745300886,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010253245324532455,
      "loss": 3.2842,
      "step": 487390
    },
    {
      "epoch": 0.21793943068916705,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010253045304530453,
      "loss": 3.1365,
      "step": 487400
    },
    {
      "epoch": 0.2179580739253252,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010252845284528454,
      "loss": 3.1882,
      "step": 487410
    },
    {
      "epoch": 0.2179767171614834,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010252645264526452,
      "loss": 3.2583,
      "step": 487420
    },
    {
      "epoch": 0.21799536039764159,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010252445244524453,
      "loss": 3.2512,
      "step": 487430
    },
    {
      "epoch": 0.21801400363379977,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010252245224522452,
      "loss": 3.1017,
      "step": 487440
    },
    {
      "epoch": 0.21803264686995796,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010252045204520453,
      "loss": 3.2319,
      "step": 487450
    },
    {
      "epoch": 0.21805129010611612,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010251845184518452,
      "loss": 3.2727,
      "step": 487460
    },
    {
      "epoch": 0.2180699333422743,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010251645164516452,
      "loss": 3.2874,
      "step": 487470
    },
    {
      "epoch": 0.2180885765784325,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001025144514451445,
      "loss": 3.1753,
      "step": 487480
    },
    {
      "epoch": 0.2181072198145907,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010251245124512452,
      "loss": 3.2035,
      "step": 487490
    },
    {
      "epoch": 0.21812586305074888,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010251045104510451,
      "loss": 3.2999,
      "step": 487500
    },
    {
      "epoch": 0.21814450628690704,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010250845084508452,
      "loss": 3.3294,
      "step": 487510
    },
    {
      "epoch": 0.21816314952306523,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010250645064506452,
      "loss": 3.2655,
      "step": 487520
    },
    {
      "epoch": 0.21818179275922342,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010250445044504451,
      "loss": 3.2906,
      "step": 487530
    },
    {
      "epoch": 0.2182004359953816,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010250245024502452,
      "loss": 3.148,
      "step": 487540
    },
    {
      "epoch": 0.2182190792315398,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001025004500450045,
      "loss": 3.1779,
      "step": 487550
    },
    {
      "epoch": 0.21823772246769796,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010249844984498451,
      "loss": 3.2516,
      "step": 487560
    },
    {
      "epoch": 0.21825636570385615,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001024964496449645,
      "loss": 3.2323,
      "step": 487570
    },
    {
      "epoch": 0.21827500894001434,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001024944494449445,
      "loss": 3.2044,
      "step": 487580
    },
    {
      "epoch": 0.21829365217617253,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001024924492449245,
      "loss": 3.2976,
      "step": 487590
    },
    {
      "epoch": 0.21831229541233071,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010249044904490451,
      "loss": 3.2942,
      "step": 487600
    },
    {
      "epoch": 0.2183309386484889,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010248844884488449,
      "loss": 3.2389,
      "step": 487610
    },
    {
      "epoch": 0.21834958188464706,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001024864486448645,
      "loss": 3.2617,
      "step": 487620
    },
    {
      "epoch": 0.21836822512080525,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010248444844484448,
      "loss": 3.1037,
      "step": 487630
    },
    {
      "epoch": 0.21838686835696344,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010248244824482449,
      "loss": 3.2475,
      "step": 487640
    },
    {
      "epoch": 0.21840551159312163,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010248044804480447,
      "loss": 3.1213,
      "step": 487650
    },
    {
      "epoch": 0.21842415482927982,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010247844784478448,
      "loss": 3.2038,
      "step": 487660
    },
    {
      "epoch": 0.21844279806543798,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010247644764476448,
      "loss": 3.2333,
      "step": 487670
    },
    {
      "epoch": 0.21846144130159617,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00010247444744474449,
      "loss": 3.1434,
      "step": 487680
    },
    {
      "epoch": 0.21848008453775436,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010247244724472447,
      "loss": 3.2996,
      "step": 487690
    },
    {
      "epoch": 0.21849872777391255,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010247044704470448,
      "loss": 3.2538,
      "step": 487700
    },
    {
      "epoch": 0.21851737101007074,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010246844684468446,
      "loss": 3.2985,
      "step": 487710
    },
    {
      "epoch": 0.2185360142462289,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010246644664466447,
      "loss": 3.1852,
      "step": 487720
    },
    {
      "epoch": 0.2185546574823871,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010246444644464448,
      "loss": 3.3093,
      "step": 487730
    },
    {
      "epoch": 0.21857330071854528,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010246244624462446,
      "loss": 3.1275,
      "step": 487740
    },
    {
      "epoch": 0.21859194395470347,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010246044604460447,
      "loss": 3.1809,
      "step": 487750
    },
    {
      "epoch": 0.21861058719086165,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010245844584458446,
      "loss": 3.347,
      "step": 487760
    },
    {
      "epoch": 0.21862923042701982,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010245644564456447,
      "loss": 3.2139,
      "step": 487770
    },
    {
      "epoch": 0.218647873663178,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010245444544454446,
      "loss": 3.0973,
      "step": 487780
    },
    {
      "epoch": 0.2186665168993362,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010245244524452446,
      "loss": 3.1195,
      "step": 487790
    },
    {
      "epoch": 0.21868516013549438,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010245044504450445,
      "loss": 3.265,
      "step": 487800
    },
    {
      "epoch": 0.21870380337165257,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010244844484448446,
      "loss": 3.1842,
      "step": 487810
    },
    {
      "epoch": 0.21872244660781073,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010244644464446445,
      "loss": 3.257,
      "step": 487820
    },
    {
      "epoch": 0.21874108984396892,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010244444444444446,
      "loss": 3.2578,
      "step": 487830
    },
    {
      "epoch": 0.2187597330801271,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010244244424442444,
      "loss": 3.1591,
      "step": 487840
    },
    {
      "epoch": 0.2187783763162853,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010244044404440445,
      "loss": 3.2342,
      "step": 487850
    },
    {
      "epoch": 0.2187970195524435,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010243844384438443,
      "loss": 3.2407,
      "step": 487860
    },
    {
      "epoch": 0.21881566278860168,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010243644364436444,
      "loss": 3.2128,
      "step": 487870
    },
    {
      "epoch": 0.21883430602475984,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010243444344434442,
      "loss": 3.2545,
      "step": 487880
    },
    {
      "epoch": 0.21885294926091803,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010243244324432443,
      "loss": 3.2539,
      "step": 487890
    },
    {
      "epoch": 0.21887159249707622,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010243044304430443,
      "loss": 3.1937,
      "step": 487900
    },
    {
      "epoch": 0.2188902357332344,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010242844284428444,
      "loss": 3.33,
      "step": 487910
    },
    {
      "epoch": 0.2189088789693926,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010242644264426445,
      "loss": 3.2016,
      "step": 487920
    },
    {
      "epoch": 0.21892752220555076,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010242444244424443,
      "loss": 3.1483,
      "step": 487930
    },
    {
      "epoch": 0.21894616544170895,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010242244224422444,
      "loss": 3.2739,
      "step": 487940
    },
    {
      "epoch": 0.21896480867786713,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010242044204420442,
      "loss": 3.2711,
      "step": 487950
    },
    {
      "epoch": 0.21898345191402532,
      "grad_norm": 0.75,
      "learning_rate": 0.00010241844184418443,
      "loss": 3.2508,
      "step": 487960
    },
    {
      "epoch": 0.2190020951501835,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010241644164416441,
      "loss": 3.171,
      "step": 487970
    },
    {
      "epoch": 0.21902073838634167,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010241444144414442,
      "loss": 3.3209,
      "step": 487980
    },
    {
      "epoch": 0.21903938162249986,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010241244124412442,
      "loss": 3.128,
      "step": 487990
    },
    {
      "epoch": 0.21905802485865805,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010241044104410443,
      "loss": 3.3226,
      "step": 488000
    },
    {
      "epoch": 0.21907666809481624,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010240844084408441,
      "loss": 3.1206,
      "step": 488010
    },
    {
      "epoch": 0.21909531133097443,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010240644064406442,
      "loss": 3.1183,
      "step": 488020
    },
    {
      "epoch": 0.2191139545671326,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001024044404440444,
      "loss": 3.0993,
      "step": 488030
    },
    {
      "epoch": 0.21913259780329078,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010240244024402441,
      "loss": 3.2736,
      "step": 488040
    },
    {
      "epoch": 0.21915124103944897,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001024004400440044,
      "loss": 3.0499,
      "step": 488050
    },
    {
      "epoch": 0.21916988427560716,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001023984398439844,
      "loss": 3.2748,
      "step": 488060
    },
    {
      "epoch": 0.21918852751176535,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001023964396439644,
      "loss": 3.1589,
      "step": 488070
    },
    {
      "epoch": 0.21920717074792354,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001023944394439444,
      "loss": 3.1881,
      "step": 488080
    },
    {
      "epoch": 0.2192258139840817,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010239243924392439,
      "loss": 3.2321,
      "step": 488090
    },
    {
      "epoch": 0.21924445722023989,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001023904390439044,
      "loss": 3.2521,
      "step": 488100
    },
    {
      "epoch": 0.21926310045639807,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001023884388438844,
      "loss": 3.1888,
      "step": 488110
    },
    {
      "epoch": 0.21928174369255626,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010238643864386439,
      "loss": 3.4099,
      "step": 488120
    },
    {
      "epoch": 0.21930038692871445,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001023844384438444,
      "loss": 3.1836,
      "step": 488130
    },
    {
      "epoch": 0.2193190301648726,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010238243824382439,
      "loss": 3.2442,
      "step": 488140
    },
    {
      "epoch": 0.2193376734010308,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001023804380438044,
      "loss": 3.3917,
      "step": 488150
    },
    {
      "epoch": 0.219356316637189,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010237843784378438,
      "loss": 3.2258,
      "step": 488160
    },
    {
      "epoch": 0.21937495987334718,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010237643764376439,
      "loss": 3.2177,
      "step": 488170
    },
    {
      "epoch": 0.21939360310950537,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010237443744374437,
      "loss": 3.1971,
      "step": 488180
    },
    {
      "epoch": 0.21941224634566353,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010237243724372438,
      "loss": 3.1575,
      "step": 488190
    },
    {
      "epoch": 0.21943088958182172,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010237043704370437,
      "loss": 3.2206,
      "step": 488200
    },
    {
      "epoch": 0.2194495328179799,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010236843684368437,
      "loss": 3.2827,
      "step": 488210
    },
    {
      "epoch": 0.2194681760541381,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010236643664366437,
      "loss": 3.1448,
      "step": 488220
    },
    {
      "epoch": 0.2194868192902963,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010236443644364438,
      "loss": 3.1665,
      "step": 488230
    },
    {
      "epoch": 0.21950546252645445,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010236243624362436,
      "loss": 3.0385,
      "step": 488240
    },
    {
      "epoch": 0.21952410576261264,
      "grad_norm": 0.75,
      "learning_rate": 0.00010236043604360437,
      "loss": 3.2621,
      "step": 488250
    },
    {
      "epoch": 0.21954274899877083,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010235843584358435,
      "loss": 3.1657,
      "step": 488260
    },
    {
      "epoch": 0.21956139223492901,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010235643564356436,
      "loss": 3.3033,
      "step": 488270
    },
    {
      "epoch": 0.2195800354710872,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010235443544354434,
      "loss": 3.3428,
      "step": 488280
    },
    {
      "epoch": 0.2195986787072454,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010235243524352435,
      "loss": 3.2479,
      "step": 488290
    },
    {
      "epoch": 0.21961732194340355,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010235043504350435,
      "loss": 3.1622,
      "step": 488300
    },
    {
      "epoch": 0.21963596517956174,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010234843484348436,
      "loss": 3.2719,
      "step": 488310
    },
    {
      "epoch": 0.21965460841571993,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010234643464346437,
      "loss": 3.1245,
      "step": 488320
    },
    {
      "epoch": 0.21967325165187812,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010234443444344435,
      "loss": 3.2001,
      "step": 488330
    },
    {
      "epoch": 0.2196918948880363,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010234243424342436,
      "loss": 3.144,
      "step": 488340
    },
    {
      "epoch": 0.21971053812419447,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010234043404340434,
      "loss": 3.377,
      "step": 488350
    },
    {
      "epoch": 0.21972918136035266,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010233843384338435,
      "loss": 3.2259,
      "step": 488360
    },
    {
      "epoch": 0.21974782459651085,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010233643364336434,
      "loss": 3.1936,
      "step": 488370
    },
    {
      "epoch": 0.21976646783266904,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010233443344334434,
      "loss": 3.2976,
      "step": 488380
    },
    {
      "epoch": 0.21978511106882723,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010233243324332434,
      "loss": 3.1968,
      "step": 488390
    },
    {
      "epoch": 0.2198037543049854,
      "grad_norm": 0.75,
      "learning_rate": 0.00010233043304330434,
      "loss": 3.216,
      "step": 488400
    },
    {
      "epoch": 0.21982239754114358,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010232843284328433,
      "loss": 3.1285,
      "step": 488410
    },
    {
      "epoch": 0.21984104077730177,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010232643264326434,
      "loss": 3.1005,
      "step": 488420
    },
    {
      "epoch": 0.21985968401345996,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010232443244324432,
      "loss": 3.3594,
      "step": 488430
    },
    {
      "epoch": 0.21987832724961814,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010232243224322433,
      "loss": 3.2564,
      "step": 488440
    },
    {
      "epoch": 0.2198969704857763,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010232043204320432,
      "loss": 3.0335,
      "step": 488450
    },
    {
      "epoch": 0.2199156137219345,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010231843184318433,
      "loss": 3.2193,
      "step": 488460
    },
    {
      "epoch": 0.21993425695809268,
      "grad_norm": 0.625,
      "learning_rate": 0.00010231643164316431,
      "loss": 3.2435,
      "step": 488470
    },
    {
      "epoch": 0.21995290019425087,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010231443144314432,
      "loss": 3.2051,
      "step": 488480
    },
    {
      "epoch": 0.21997154343040906,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001023124312431243,
      "loss": 3.1983,
      "step": 488490
    },
    {
      "epoch": 0.21999018666656722,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010231043104310431,
      "loss": 3.2596,
      "step": 488500
    },
    {
      "epoch": 0.2200088299027254,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010230843084308432,
      "loss": 3.3103,
      "step": 488510
    },
    {
      "epoch": 0.2200274731388836,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001023064306430643,
      "loss": 3.146,
      "step": 488520
    },
    {
      "epoch": 0.2200461163750418,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010230443044304431,
      "loss": 3.2269,
      "step": 488530
    },
    {
      "epoch": 0.22006475961119998,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010230243024302431,
      "loss": 3.2098,
      "step": 488540
    },
    {
      "epoch": 0.22008340284735817,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010230043004300432,
      "loss": 3.2827,
      "step": 488550
    },
    {
      "epoch": 0.22010204608351633,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001022984298429843,
      "loss": 3.1683,
      "step": 488560
    },
    {
      "epoch": 0.22012068931967452,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010229642964296431,
      "loss": 3.1406,
      "step": 488570
    },
    {
      "epoch": 0.2201393325558327,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010229442944294429,
      "loss": 3.1828,
      "step": 488580
    },
    {
      "epoch": 0.2201579757919909,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001022924292429243,
      "loss": 3.2746,
      "step": 488590
    },
    {
      "epoch": 0.22017661902814908,
      "grad_norm": 0.75,
      "learning_rate": 0.00010229042904290428,
      "loss": 3.2503,
      "step": 488600
    },
    {
      "epoch": 0.22019526226430725,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010228842884288429,
      "loss": 3.1737,
      "step": 488610
    },
    {
      "epoch": 0.22021390550046543,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010228642864286429,
      "loss": 3.2257,
      "step": 488620
    },
    {
      "epoch": 0.22023254873662362,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001022844284428443,
      "loss": 3.2644,
      "step": 488630
    },
    {
      "epoch": 0.2202511919727818,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010228242824282428,
      "loss": 2.9991,
      "step": 488640
    },
    {
      "epoch": 0.22026983520894,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010228042804280429,
      "loss": 3.0505,
      "step": 488650
    },
    {
      "epoch": 0.22028847844509816,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010227842784278427,
      "loss": 3.1848,
      "step": 488660
    },
    {
      "epoch": 0.22030712168125635,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010227642764276428,
      "loss": 3.2709,
      "step": 488670
    },
    {
      "epoch": 0.22032576491741454,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010227442744274428,
      "loss": 3.1519,
      "step": 488680
    },
    {
      "epoch": 0.22034440815357273,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010227242724272427,
      "loss": 3.2584,
      "step": 488690
    },
    {
      "epoch": 0.22036305138973092,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010227042704270427,
      "loss": 3.2192,
      "step": 488700
    },
    {
      "epoch": 0.22038169462588908,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010226842684268428,
      "loss": 3.2554,
      "step": 488710
    },
    {
      "epoch": 0.22040033786204727,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010226642664266429,
      "loss": 3.2066,
      "step": 488720
    },
    {
      "epoch": 0.22041898109820546,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010226442644264427,
      "loss": 3.221,
      "step": 488730
    },
    {
      "epoch": 0.22043762433436365,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010226242624262428,
      "loss": 3.0726,
      "step": 488740
    },
    {
      "epoch": 0.22045626757052184,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010226042604260426,
      "loss": 3.1307,
      "step": 488750
    },
    {
      "epoch": 0.22047491080668002,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010225842584258427,
      "loss": 3.2371,
      "step": 488760
    },
    {
      "epoch": 0.22049355404283819,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010225642564256426,
      "loss": 3.2278,
      "step": 488770
    },
    {
      "epoch": 0.22051219727899637,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010225442544254427,
      "loss": 3.2168,
      "step": 488780
    },
    {
      "epoch": 0.22053084051515456,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010225242524252425,
      "loss": 3.2495,
      "step": 488790
    },
    {
      "epoch": 0.22054948375131275,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010225042504250426,
      "loss": 3.242,
      "step": 488800
    },
    {
      "epoch": 0.22056812698747094,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010224842484248425,
      "loss": 3.2052,
      "step": 488810
    },
    {
      "epoch": 0.2205867702236291,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010224642464246425,
      "loss": 3.2129,
      "step": 488820
    },
    {
      "epoch": 0.2206054134597873,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010224442444244424,
      "loss": 3.2064,
      "step": 488830
    },
    {
      "epoch": 0.22062405669594548,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010224242424242425,
      "loss": 3.1821,
      "step": 488840
    },
    {
      "epoch": 0.22064269993210367,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010224042404240424,
      "loss": 3.1943,
      "step": 488850
    },
    {
      "epoch": 0.22066134316826186,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010223842384238425,
      "loss": 3.1777,
      "step": 488860
    },
    {
      "epoch": 0.22067998640442002,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010223642364236423,
      "loss": 3.0869,
      "step": 488870
    },
    {
      "epoch": 0.2206986296405782,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010223442344234424,
      "loss": 3.2099,
      "step": 488880
    },
    {
      "epoch": 0.2207172728767364,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010223242324232422,
      "loss": 3.2926,
      "step": 488890
    },
    {
      "epoch": 0.2207359161128946,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010223042304230423,
      "loss": 3.2939,
      "step": 488900
    },
    {
      "epoch": 0.22075455934905278,
      "grad_norm": 0.875,
      "learning_rate": 0.00010222842284228424,
      "loss": 3.1937,
      "step": 488910
    },
    {
      "epoch": 0.22077320258521094,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010222642264226422,
      "loss": 3.1452,
      "step": 488920
    },
    {
      "epoch": 0.22079184582136913,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010222442244224423,
      "loss": 3.1581,
      "step": 488930
    },
    {
      "epoch": 0.22081048905752731,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010222242224222423,
      "loss": 3.0848,
      "step": 488940
    },
    {
      "epoch": 0.2208291322936855,
      "grad_norm": 0.75,
      "learning_rate": 0.00010222042204220424,
      "loss": 3.2304,
      "step": 488950
    },
    {
      "epoch": 0.2208477755298437,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010221842184218422,
      "loss": 3.1899,
      "step": 488960
    },
    {
      "epoch": 0.22086641876600188,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010221642164216423,
      "loss": 3.1145,
      "step": 488970
    },
    {
      "epoch": 0.22088506200216004,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010221442144214421,
      "loss": 3.2515,
      "step": 488980
    },
    {
      "epoch": 0.22090370523831823,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010221242124212422,
      "loss": 3.258,
      "step": 488990
    },
    {
      "epoch": 0.22092234847447642,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010221042104210422,
      "loss": 3.2827,
      "step": 489000
    },
    {
      "epoch": 0.2209409917106346,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010220842084208421,
      "loss": 3.1056,
      "step": 489010
    },
    {
      "epoch": 0.2209596349467928,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010220642064206421,
      "loss": 3.1889,
      "step": 489020
    },
    {
      "epoch": 0.22097827818295096,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010220442044204422,
      "loss": 3.2287,
      "step": 489030
    },
    {
      "epoch": 0.22099692141910915,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001022024202420242,
      "loss": 3.2953,
      "step": 489040
    },
    {
      "epoch": 0.22101556465526734,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010220042004200421,
      "loss": 3.3535,
      "step": 489050
    },
    {
      "epoch": 0.22103420789142553,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010219841984198419,
      "loss": 3.2079,
      "step": 489060
    },
    {
      "epoch": 0.22105285112758372,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001021964196419642,
      "loss": 3.1732,
      "step": 489070
    },
    {
      "epoch": 0.22107149436374188,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001021944194419442,
      "loss": 3.2213,
      "step": 489080
    },
    {
      "epoch": 0.22109013759990007,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001021924192419242,
      "loss": 3.2478,
      "step": 489090
    },
    {
      "epoch": 0.22110878083605826,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010219041904190421,
      "loss": 3.2168,
      "step": 489100
    },
    {
      "epoch": 0.22112742407221644,
      "grad_norm": 0.75,
      "learning_rate": 0.0001021884188418842,
      "loss": 3.282,
      "step": 489110
    },
    {
      "epoch": 0.22114606730837463,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001021864186418642,
      "loss": 3.1943,
      "step": 489120
    },
    {
      "epoch": 0.2211647105445328,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010218441844184419,
      "loss": 3.2726,
      "step": 489130
    },
    {
      "epoch": 0.22118335378069098,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001021824182418242,
      "loss": 3.1618,
      "step": 489140
    },
    {
      "epoch": 0.22120199701684917,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010218041804180418,
      "loss": 3.378,
      "step": 489150
    },
    {
      "epoch": 0.22122064025300736,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010217841784178419,
      "loss": 3.1189,
      "step": 489160
    },
    {
      "epoch": 0.22123928348916555,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010217641764176418,
      "loss": 3.233,
      "step": 489170
    },
    {
      "epoch": 0.2212579267253237,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010217441744174419,
      "loss": 3.2042,
      "step": 489180
    },
    {
      "epoch": 0.2212765699614819,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010217241724172417,
      "loss": 3.3489,
      "step": 489190
    },
    {
      "epoch": 0.2212952131976401,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010217041704170418,
      "loss": 3.3411,
      "step": 489200
    },
    {
      "epoch": 0.22131385643379828,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010216841684168416,
      "loss": 3.3,
      "step": 489210
    },
    {
      "epoch": 0.22133249966995647,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010216641664166417,
      "loss": 3.1008,
      "step": 489220
    },
    {
      "epoch": 0.22135114290611466,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010216441644164416,
      "loss": 3.2461,
      "step": 489230
    },
    {
      "epoch": 0.22136978614227282,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010216241624162416,
      "loss": 3.2379,
      "step": 489240
    },
    {
      "epoch": 0.221388429378431,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010216041604160416,
      "loss": 3.1905,
      "step": 489250
    },
    {
      "epoch": 0.2214070726145892,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010215841584158417,
      "loss": 3.175,
      "step": 489260
    },
    {
      "epoch": 0.22142571585074738,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010215641564156415,
      "loss": 3.2596,
      "step": 489270
    },
    {
      "epoch": 0.22144435908690557,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010215441544154416,
      "loss": 3.2038,
      "step": 489280
    },
    {
      "epoch": 0.22146300232306373,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010215241524152414,
      "loss": 3.1525,
      "step": 489290
    },
    {
      "epoch": 0.22148164555922192,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010215041504150415,
      "loss": 3.2925,
      "step": 489300
    },
    {
      "epoch": 0.2215002887953801,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010214841484148416,
      "loss": 3.2156,
      "step": 489310
    },
    {
      "epoch": 0.2215189320315383,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010214641464146416,
      "loss": 3.3122,
      "step": 489320
    },
    {
      "epoch": 0.2215375752676965,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010214441444144415,
      "loss": 3.2413,
      "step": 489330
    },
    {
      "epoch": 0.22155621850385465,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010214241424142415,
      "loss": 3.1945,
      "step": 489340
    },
    {
      "epoch": 0.22157486174001284,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010214041404140416,
      "loss": 3.2582,
      "step": 489350
    },
    {
      "epoch": 0.22159350497617103,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010213841384138414,
      "loss": 3.1921,
      "step": 489360
    },
    {
      "epoch": 0.22161214821232922,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010213641364136415,
      "loss": 3.0882,
      "step": 489370
    },
    {
      "epoch": 0.2216307914484874,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010213441344134413,
      "loss": 3.3018,
      "step": 489380
    },
    {
      "epoch": 0.22164943468464557,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010213241324132414,
      "loss": 3.2137,
      "step": 489390
    },
    {
      "epoch": 0.22166807792080376,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010213041304130413,
      "loss": 3.2147,
      "step": 489400
    },
    {
      "epoch": 0.22168672115696195,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010212841284128414,
      "loss": 3.1543,
      "step": 489410
    },
    {
      "epoch": 0.22170536439312014,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010212641264126413,
      "loss": 3.2393,
      "step": 489420
    },
    {
      "epoch": 0.22172400762927832,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010212441244124414,
      "loss": 3.1746,
      "step": 489430
    },
    {
      "epoch": 0.2217426508654365,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010212241224122412,
      "loss": 3.1663,
      "step": 489440
    },
    {
      "epoch": 0.22176129410159467,
      "grad_norm": 0.75,
      "learning_rate": 0.00010212041204120413,
      "loss": 3.2458,
      "step": 489450
    },
    {
      "epoch": 0.22177993733775286,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010211841184118411,
      "loss": 3.2722,
      "step": 489460
    },
    {
      "epoch": 0.22179858057391105,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010211641164116412,
      "loss": 3.0794,
      "step": 489470
    },
    {
      "epoch": 0.22181722381006924,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010211441144114411,
      "loss": 3.2313,
      "step": 489480
    },
    {
      "epoch": 0.22183586704622743,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010211241124112412,
      "loss": 3.1734,
      "step": 489490
    },
    {
      "epoch": 0.2218545102823856,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010211041104110413,
      "loss": 3.1872,
      "step": 489500
    },
    {
      "epoch": 0.22187315351854378,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010210841084108411,
      "loss": 3.1848,
      "step": 489510
    },
    {
      "epoch": 0.22189179675470197,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010210641064106412,
      "loss": 3.1375,
      "step": 489520
    },
    {
      "epoch": 0.22191043999086016,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001021044104410441,
      "loss": 3.2527,
      "step": 489530
    },
    {
      "epoch": 0.22192908322701835,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010210241024102411,
      "loss": 3.1037,
      "step": 489540
    },
    {
      "epoch": 0.2219477264631765,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001021004100410041,
      "loss": 3.285,
      "step": 489550
    },
    {
      "epoch": 0.2219663696993347,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001020984098409841,
      "loss": 3.2483,
      "step": 489560
    },
    {
      "epoch": 0.2219850129354929,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001020964096409641,
      "loss": 3.2182,
      "step": 489570
    },
    {
      "epoch": 0.22200365617165108,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010209440944094411,
      "loss": 3.186,
      "step": 489580
    },
    {
      "epoch": 0.22202229940780926,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010209240924092409,
      "loss": 3.1762,
      "step": 489590
    },
    {
      "epoch": 0.22204094264396743,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001020904090409041,
      "loss": 3.2654,
      "step": 489600
    },
    {
      "epoch": 0.22205958588012562,
      "grad_norm": 0.75,
      "learning_rate": 0.00010208840884088408,
      "loss": 3.1076,
      "step": 489610
    },
    {
      "epoch": 0.2220782291162838,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010208640864086409,
      "loss": 3.2939,
      "step": 489620
    },
    {
      "epoch": 0.222096872352442,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010208440844084409,
      "loss": 3.2848,
      "step": 489630
    },
    {
      "epoch": 0.22211551558860018,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001020824082408241,
      "loss": 3.2435,
      "step": 489640
    },
    {
      "epoch": 0.22213415882475837,
      "grad_norm": 0.875,
      "learning_rate": 0.00010208040804080408,
      "loss": 3.2084,
      "step": 489650
    },
    {
      "epoch": 0.22215280206091653,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00010207840784078409,
      "loss": 3.2532,
      "step": 489660
    },
    {
      "epoch": 0.22217144529707472,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010207640764076407,
      "loss": 3.2852,
      "step": 489670
    },
    {
      "epoch": 0.2221900885332329,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010207440744074408,
      "loss": 3.3507,
      "step": 489680
    },
    {
      "epoch": 0.2222087317693911,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010207240724072409,
      "loss": 3.2749,
      "step": 489690
    },
    {
      "epoch": 0.2222273750055493,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010207040704070407,
      "loss": 3.2817,
      "step": 489700
    },
    {
      "epoch": 0.22224601824170745,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010206840684068408,
      "loss": 3.2839,
      "step": 489710
    },
    {
      "epoch": 0.22226466147786564,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010206640664066407,
      "loss": 3.2553,
      "step": 489720
    },
    {
      "epoch": 0.22228330471402383,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010206440644064408,
      "loss": 3.2549,
      "step": 489730
    },
    {
      "epoch": 0.22230194795018202,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010206240624062407,
      "loss": 3.2839,
      "step": 489740
    },
    {
      "epoch": 0.2223205911863402,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010206040604060408,
      "loss": 3.3501,
      "step": 489750
    },
    {
      "epoch": 0.22233923442249837,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010205840584058406,
      "loss": 3.2088,
      "step": 489760
    },
    {
      "epoch": 0.22235787765865656,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010205640564056407,
      "loss": 3.1324,
      "step": 489770
    },
    {
      "epoch": 0.22237652089481474,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010205440544054405,
      "loss": 3.1516,
      "step": 489780
    },
    {
      "epoch": 0.22239516413097293,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010205240524052406,
      "loss": 3.1385,
      "step": 489790
    },
    {
      "epoch": 0.22241380736713112,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010205040504050405,
      "loss": 3.2284,
      "step": 489800
    },
    {
      "epoch": 0.22243245060328928,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010204840484048406,
      "loss": 3.1226,
      "step": 489810
    },
    {
      "epoch": 0.22245109383944747,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010204640464046404,
      "loss": 3.2023,
      "step": 489820
    },
    {
      "epoch": 0.22246973707560566,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010204440444044405,
      "loss": 3.2402,
      "step": 489830
    },
    {
      "epoch": 0.22248838031176385,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010204240424042404,
      "loss": 3.1612,
      "step": 489840
    },
    {
      "epoch": 0.22250702354792204,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010204040404040404,
      "loss": 3.1207,
      "step": 489850
    },
    {
      "epoch": 0.2225256667840802,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010203840384038404,
      "loss": 3.1108,
      "step": 489860
    },
    {
      "epoch": 0.2225443100202384,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010203640364036404,
      "loss": 3.1638,
      "step": 489870
    },
    {
      "epoch": 0.22256295325639658,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010203440344034403,
      "loss": 3.1054,
      "step": 489880
    },
    {
      "epoch": 0.22258159649255477,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010203240324032404,
      "loss": 3.1893,
      "step": 489890
    },
    {
      "epoch": 0.22260023972871296,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010203040304030405,
      "loss": 3.104,
      "step": 489900
    },
    {
      "epoch": 0.22261888296487115,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010202840284028403,
      "loss": 3.2737,
      "step": 489910
    },
    {
      "epoch": 0.2226375262010293,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010202640264026404,
      "loss": 3.1109,
      "step": 489920
    },
    {
      "epoch": 0.2226561694371875,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010202440244024402,
      "loss": 3.1545,
      "step": 489930
    },
    {
      "epoch": 0.22267481267334568,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010202240224022403,
      "loss": 3.2341,
      "step": 489940
    },
    {
      "epoch": 0.22269345590950387,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010202040204020403,
      "loss": 3.2045,
      "step": 489950
    },
    {
      "epoch": 0.22271209914566206,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010201840184018402,
      "loss": 3.1604,
      "step": 489960
    },
    {
      "epoch": 0.22273074238182022,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010201640164016402,
      "loss": 3.1204,
      "step": 489970
    },
    {
      "epoch": 0.2227493856179784,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010201440144014403,
      "loss": 3.1832,
      "step": 489980
    },
    {
      "epoch": 0.2227680288541366,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010201240124012401,
      "loss": 3.2059,
      "step": 489990
    },
    {
      "epoch": 0.2227866720902948,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00010201040104010402,
      "loss": 3.1589,
      "step": 490000
    },
    {
      "epoch": 0.22280531532645298,
      "grad_norm": 0.859375,
      "learning_rate": 0.000102008400840084,
      "loss": 3.2308,
      "step": 490010
    },
    {
      "epoch": 0.22282395856261114,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010200640064006401,
      "loss": 3.2372,
      "step": 490020
    },
    {
      "epoch": 0.22284260179876933,
      "grad_norm": 0.7890625,
      "learning_rate": 0.000102004400440044,
      "loss": 3.354,
      "step": 490030
    },
    {
      "epoch": 0.22286124503492752,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010200240024002402,
      "loss": 3.2834,
      "step": 490040
    },
    {
      "epoch": 0.2228798882710857,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000102000400040004,
      "loss": 3.1224,
      "step": 490050
    },
    {
      "epoch": 0.2228985315072439,
      "grad_norm": 0.671875,
      "learning_rate": 0.000101998399839984,
      "loss": 3.1202,
      "step": 490060
    },
    {
      "epoch": 0.22291717474340206,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010199639963996399,
      "loss": 3.1635,
      "step": 490070
    },
    {
      "epoch": 0.22293581797956025,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000101994399439944,
      "loss": 3.1674,
      "step": 490080
    },
    {
      "epoch": 0.22295446121571844,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010199239923992401,
      "loss": 3.3386,
      "step": 490090
    },
    {
      "epoch": 0.22297310445187662,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010199039903990399,
      "loss": 3.2223,
      "step": 490100
    },
    {
      "epoch": 0.2229917476880348,
      "grad_norm": 0.609375,
      "learning_rate": 0.000101988398839884,
      "loss": 3.1162,
      "step": 490110
    },
    {
      "epoch": 0.223010390924193,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000101986398639864,
      "loss": 3.2853,
      "step": 490120
    },
    {
      "epoch": 0.22302903416035116,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000101984398439844,
      "loss": 3.396,
      "step": 490130
    },
    {
      "epoch": 0.22304767739650935,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010198239823982398,
      "loss": 3.1266,
      "step": 490140
    },
    {
      "epoch": 0.22306632063266754,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000101980398039804,
      "loss": 3.2777,
      "step": 490150
    },
    {
      "epoch": 0.22308496386882573,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010197839783978398,
      "loss": 3.0581,
      "step": 490160
    },
    {
      "epoch": 0.22310360710498392,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010197639763976399,
      "loss": 3.3415,
      "step": 490170
    },
    {
      "epoch": 0.22312225034114208,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010197439743974397,
      "loss": 3.2138,
      "step": 490180
    },
    {
      "epoch": 0.22314089357730027,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010197239723972398,
      "loss": 3.1457,
      "step": 490190
    },
    {
      "epoch": 0.22315953681345846,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010197039703970397,
      "loss": 3.2689,
      "step": 490200
    },
    {
      "epoch": 0.22317818004961665,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010196839683968398,
      "loss": 3.156,
      "step": 490210
    },
    {
      "epoch": 0.22319682328577484,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010196639663966396,
      "loss": 3.2196,
      "step": 490220
    },
    {
      "epoch": 0.223215466521933,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010196439643964397,
      "loss": 3.2157,
      "step": 490230
    },
    {
      "epoch": 0.2232341097580912,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010196239623962395,
      "loss": 3.1907,
      "step": 490240
    },
    {
      "epoch": 0.22325275299424938,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010196039603960396,
      "loss": 3.3356,
      "step": 490250
    },
    {
      "epoch": 0.22327139623040756,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010195839583958396,
      "loss": 3.1501,
      "step": 490260
    },
    {
      "epoch": 0.22329003946656575,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010195639563956397,
      "loss": 3.2041,
      "step": 490270
    },
    {
      "epoch": 0.22330868270272392,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010195439543954395,
      "loss": 3.2344,
      "step": 490280
    },
    {
      "epoch": 0.2233273259388821,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010195239523952396,
      "loss": 3.0855,
      "step": 490290
    },
    {
      "epoch": 0.2233459691750403,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010195039503950397,
      "loss": 3.2565,
      "step": 490300
    },
    {
      "epoch": 0.22336461241119848,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010194839483948395,
      "loss": 3.1542,
      "step": 490310
    },
    {
      "epoch": 0.22338325564735667,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010194639463946396,
      "loss": 3.2524,
      "step": 490320
    },
    {
      "epoch": 0.22340189888351486,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010194439443944394,
      "loss": 3.2423,
      "step": 490330
    },
    {
      "epoch": 0.22342054211967302,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010194239423942395,
      "loss": 3.2292,
      "step": 490340
    },
    {
      "epoch": 0.2234391853558312,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010194039403940395,
      "loss": 3.2084,
      "step": 490350
    },
    {
      "epoch": 0.2234578285919894,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010193839383938396,
      "loss": 3.2295,
      "step": 490360
    },
    {
      "epoch": 0.2234764718281476,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010193639363936394,
      "loss": 3.2581,
      "step": 490370
    },
    {
      "epoch": 0.22349511506430578,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010193439343934395,
      "loss": 3.1142,
      "step": 490380
    },
    {
      "epoch": 0.22351375830046394,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010193239323932393,
      "loss": 3.3288,
      "step": 490390
    },
    {
      "epoch": 0.22353240153662213,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010193039303930394,
      "loss": 3.2468,
      "step": 490400
    },
    {
      "epoch": 0.22355104477278032,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010192839283928392,
      "loss": 3.2283,
      "step": 490410
    },
    {
      "epoch": 0.2235696880089385,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010192639263926393,
      "loss": 3.1347,
      "step": 490420
    },
    {
      "epoch": 0.2235883312450967,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010192439243924392,
      "loss": 3.2651,
      "step": 490430
    },
    {
      "epoch": 0.22360697448125486,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010192239223922393,
      "loss": 3.2089,
      "step": 490440
    },
    {
      "epoch": 0.22362561771741304,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010192039203920392,
      "loss": 3.1585,
      "step": 490450
    },
    {
      "epoch": 0.22364426095357123,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010191839183918393,
      "loss": 3.1871,
      "step": 490460
    },
    {
      "epoch": 0.22366290418972942,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010191639163916391,
      "loss": 3.1786,
      "step": 490470
    },
    {
      "epoch": 0.2236815474258876,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010191439143914392,
      "loss": 3.1798,
      "step": 490480
    },
    {
      "epoch": 0.22370019066204577,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010191239123912393,
      "loss": 3.3485,
      "step": 490490
    },
    {
      "epoch": 0.22371883389820396,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010191039103910391,
      "loss": 3.1656,
      "step": 490500
    },
    {
      "epoch": 0.22373747713436215,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010190839083908392,
      "loss": 3.1193,
      "step": 490510
    },
    {
      "epoch": 0.22375612037052034,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010190639063906391,
      "loss": 3.2178,
      "step": 490520
    },
    {
      "epoch": 0.22377476360667853,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010190439043904392,
      "loss": 3.2575,
      "step": 490530
    },
    {
      "epoch": 0.22379340684283672,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001019023902390239,
      "loss": 3.2075,
      "step": 490540
    },
    {
      "epoch": 0.22381205007899488,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010190039003900391,
      "loss": 3.1933,
      "step": 490550
    },
    {
      "epoch": 0.22383069331515307,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001018983898389839,
      "loss": 3.2474,
      "step": 490560
    },
    {
      "epoch": 0.22384933655131126,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001018963896389639,
      "loss": 3.1334,
      "step": 490570
    },
    {
      "epoch": 0.22386797978746945,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001018943894389439,
      "loss": 3.1532,
      "step": 490580
    },
    {
      "epoch": 0.22388662302362763,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010189238923892391,
      "loss": 3.345,
      "step": 490590
    },
    {
      "epoch": 0.2239052662597858,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010189038903890389,
      "loss": 3.177,
      "step": 490600
    },
    {
      "epoch": 0.22392390949594398,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001018883888388839,
      "loss": 3.231,
      "step": 490610
    },
    {
      "epoch": 0.22394255273210217,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010188638863886388,
      "loss": 3.2772,
      "step": 490620
    },
    {
      "epoch": 0.22396119596826036,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010188438843884389,
      "loss": 3.2644,
      "step": 490630
    },
    {
      "epoch": 0.22397983920441855,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010188238823882387,
      "loss": 3.268,
      "step": 490640
    },
    {
      "epoch": 0.2239984824405767,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010188038803880388,
      "loss": 3.2084,
      "step": 490650
    },
    {
      "epoch": 0.2240171256767349,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010187838783878388,
      "loss": 3.2803,
      "step": 490660
    },
    {
      "epoch": 0.2240357689128931,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010187638763876389,
      "loss": 3.2837,
      "step": 490670
    },
    {
      "epoch": 0.22405441214905128,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001018743874387439,
      "loss": 3.232,
      "step": 490680
    },
    {
      "epoch": 0.22407305538520947,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010187238723872388,
      "loss": 3.3401,
      "step": 490690
    },
    {
      "epoch": 0.22409169862136763,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010187038703870389,
      "loss": 3.2311,
      "step": 490700
    },
    {
      "epoch": 0.22411034185752582,
      "grad_norm": 0.75,
      "learning_rate": 0.00010186838683868387,
      "loss": 3.0909,
      "step": 490710
    },
    {
      "epoch": 0.224128985093684,
      "grad_norm": 0.75,
      "learning_rate": 0.00010186638663866388,
      "loss": 3.17,
      "step": 490720
    },
    {
      "epoch": 0.2241476283298422,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010186438643864386,
      "loss": 3.2748,
      "step": 490730
    },
    {
      "epoch": 0.22416627156600039,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010186238623862387,
      "loss": 3.1914,
      "step": 490740
    },
    {
      "epoch": 0.22418491480215855,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010186038603860387,
      "loss": 3.0981,
      "step": 490750
    },
    {
      "epoch": 0.22420355803831674,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010185838583858387,
      "loss": 3.21,
      "step": 490760
    },
    {
      "epoch": 0.22422220127447492,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010185638563856386,
      "loss": 3.2654,
      "step": 490770
    },
    {
      "epoch": 0.2242408445106331,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010185438543854387,
      "loss": 3.3775,
      "step": 490780
    },
    {
      "epoch": 0.2242594877467913,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010185238523852385,
      "loss": 3.1415,
      "step": 490790
    },
    {
      "epoch": 0.2242781309829495,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010185038503850386,
      "loss": 3.2128,
      "step": 490800
    },
    {
      "epoch": 0.22429677421910765,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010184838483848385,
      "loss": 3.272,
      "step": 490810
    },
    {
      "epoch": 0.22431541745526584,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010184638463846385,
      "loss": 3.1944,
      "step": 490820
    },
    {
      "epoch": 0.22433406069142403,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010184438443844384,
      "loss": 3.2697,
      "step": 490830
    },
    {
      "epoch": 0.22435270392758222,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010184238423842385,
      "loss": 3.1117,
      "step": 490840
    },
    {
      "epoch": 0.2243713471637404,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010184038403840383,
      "loss": 3.1915,
      "step": 490850
    },
    {
      "epoch": 0.22438999039989857,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010183838383838384,
      "loss": 3.1458,
      "step": 490860
    },
    {
      "epoch": 0.22440863363605676,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010183638363836383,
      "loss": 3.2233,
      "step": 490870
    },
    {
      "epoch": 0.22442727687221495,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010183438343834383,
      "loss": 3.2427,
      "step": 490880
    },
    {
      "epoch": 0.22444592010837314,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010183238323832384,
      "loss": 3.1967,
      "step": 490890
    },
    {
      "epoch": 0.22446456334453133,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010183038303830384,
      "loss": 3.2105,
      "step": 490900
    },
    {
      "epoch": 0.2244832065806895,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010182838283828385,
      "loss": 3.1119,
      "step": 490910
    },
    {
      "epoch": 0.22450184981684768,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010182638263826383,
      "loss": 3.2231,
      "step": 490920
    },
    {
      "epoch": 0.22452049305300587,
      "grad_norm": 0.75,
      "learning_rate": 0.00010182438243824384,
      "loss": 3.197,
      "step": 490930
    },
    {
      "epoch": 0.22453913628916405,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010182238223822382,
      "loss": 3.2357,
      "step": 490940
    },
    {
      "epoch": 0.22455777952532224,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010182038203820383,
      "loss": 3.13,
      "step": 490950
    },
    {
      "epoch": 0.2245764227614804,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010181838183818381,
      "loss": 3.1789,
      "step": 490960
    },
    {
      "epoch": 0.2245950659976386,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010181638163816382,
      "loss": 3.1691,
      "step": 490970
    },
    {
      "epoch": 0.22461370923379678,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010181438143814382,
      "loss": 3.1729,
      "step": 490980
    },
    {
      "epoch": 0.22463235246995497,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010181238123812383,
      "loss": 3.3099,
      "step": 490990
    },
    {
      "epoch": 0.22465099570611316,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010181038103810381,
      "loss": 3.134,
      "step": 491000
    },
    {
      "epoch": 0.22466963894227135,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010180838083808382,
      "loss": 3.2383,
      "step": 491010
    },
    {
      "epoch": 0.2246882821784295,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001018063806380638,
      "loss": 3.2167,
      "step": 491020
    },
    {
      "epoch": 0.2247069254145877,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010180438043804381,
      "loss": 3.1195,
      "step": 491030
    },
    {
      "epoch": 0.2247255686507459,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010180238023802379,
      "loss": 3.0648,
      "step": 491040
    },
    {
      "epoch": 0.22474421188690408,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001018003800380038,
      "loss": 3.1843,
      "step": 491050
    },
    {
      "epoch": 0.22476285512306227,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001017983798379838,
      "loss": 3.2528,
      "step": 491060
    },
    {
      "epoch": 0.22478149835922043,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001017963796379638,
      "loss": 3.2515,
      "step": 491070
    },
    {
      "epoch": 0.22480014159537862,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010179437943794381,
      "loss": 3.2698,
      "step": 491080
    },
    {
      "epoch": 0.2248187848315368,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001017923792379238,
      "loss": 3.3807,
      "step": 491090
    },
    {
      "epoch": 0.224837428067695,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001017903790379038,
      "loss": 3.2206,
      "step": 491100
    },
    {
      "epoch": 0.22485607130385318,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010178837883788379,
      "loss": 3.1745,
      "step": 491110
    },
    {
      "epoch": 0.22487471454001134,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001017863786378638,
      "loss": 3.2763,
      "step": 491120
    },
    {
      "epoch": 0.22489335777616953,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010178437843784379,
      "loss": 3.2463,
      "step": 491130
    },
    {
      "epoch": 0.22491200101232772,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010178237823782379,
      "loss": 3.1995,
      "step": 491140
    },
    {
      "epoch": 0.2249306442484859,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010178037803780378,
      "loss": 3.1618,
      "step": 491150
    },
    {
      "epoch": 0.2249492874846441,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010177837783778379,
      "loss": 3.1457,
      "step": 491160
    },
    {
      "epoch": 0.22496793072080226,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010177637763776377,
      "loss": 3.2829,
      "step": 491170
    },
    {
      "epoch": 0.22498657395696045,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010177437743774378,
      "loss": 3.2055,
      "step": 491180
    },
    {
      "epoch": 0.22500521719311864,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010177237723772377,
      "loss": 3.1736,
      "step": 491190
    },
    {
      "epoch": 0.22502386042927683,
      "grad_norm": 0.75,
      "learning_rate": 0.00010177037703770378,
      "loss": 3.2834,
      "step": 491200
    },
    {
      "epoch": 0.22504250366543502,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010176837683768377,
      "loss": 3.1695,
      "step": 491210
    },
    {
      "epoch": 0.2250611469015932,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010176637663766378,
      "loss": 3.2021,
      "step": 491220
    },
    {
      "epoch": 0.22507979013775137,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010176437643764376,
      "loss": 3.0856,
      "step": 491230
    },
    {
      "epoch": 0.22509843337390956,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010176237623762377,
      "loss": 3.2078,
      "step": 491240
    },
    {
      "epoch": 0.22511707661006775,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010176037603760375,
      "loss": 3.2927,
      "step": 491250
    },
    {
      "epoch": 0.22513571984622593,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010175837583758376,
      "loss": 3.051,
      "step": 491260
    },
    {
      "epoch": 0.22515436308238412,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010175637563756377,
      "loss": 3.2363,
      "step": 491270
    },
    {
      "epoch": 0.22517300631854228,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010175437543754375,
      "loss": 3.2128,
      "step": 491280
    },
    {
      "epoch": 0.22519164955470047,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010175237523752376,
      "loss": 3.1871,
      "step": 491290
    },
    {
      "epoch": 0.22521029279085866,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010175037503750376,
      "loss": 3.2676,
      "step": 491300
    },
    {
      "epoch": 0.22522893602701685,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010174837483748377,
      "loss": 3.0504,
      "step": 491310
    },
    {
      "epoch": 0.22524757926317504,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010174637463746375,
      "loss": 3.2608,
      "step": 491320
    },
    {
      "epoch": 0.2252662224993332,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010174437443744376,
      "loss": 3.2865,
      "step": 491330
    },
    {
      "epoch": 0.2252848657354914,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010174237423742374,
      "loss": 3.1541,
      "step": 491340
    },
    {
      "epoch": 0.22530350897164958,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010174037403740375,
      "loss": 3.2512,
      "step": 491350
    },
    {
      "epoch": 0.22532215220780777,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010173837383738373,
      "loss": 3.2733,
      "step": 491360
    },
    {
      "epoch": 0.22534079544396596,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010173637363736374,
      "loss": 3.1972,
      "step": 491370
    },
    {
      "epoch": 0.22535943868012412,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010173437343734374,
      "loss": 3.1481,
      "step": 491380
    },
    {
      "epoch": 0.2253780819162823,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010173237323732375,
      "loss": 3.07,
      "step": 491390
    },
    {
      "epoch": 0.2253967251524405,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010173037303730373,
      "loss": 3.2395,
      "step": 491400
    },
    {
      "epoch": 0.22541536838859869,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010172837283728374,
      "loss": 3.2546,
      "step": 491410
    },
    {
      "epoch": 0.22543401162475687,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010172637263726372,
      "loss": 3.2333,
      "step": 491420
    },
    {
      "epoch": 0.22545265486091504,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010172437243724373,
      "loss": 3.1387,
      "step": 491430
    },
    {
      "epoch": 0.22547129809707323,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010172237223722372,
      "loss": 3.2164,
      "step": 491440
    },
    {
      "epoch": 0.22548994133323141,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010172037203720372,
      "loss": 3.263,
      "step": 491450
    },
    {
      "epoch": 0.2255085845693896,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010171837183718371,
      "loss": 3.2134,
      "step": 491460
    },
    {
      "epoch": 0.2255272278055478,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010171637163716372,
      "loss": 3.1435,
      "step": 491470
    },
    {
      "epoch": 0.22554587104170598,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010171437143714373,
      "loss": 3.0848,
      "step": 491480
    },
    {
      "epoch": 0.22556451427786414,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010171237123712372,
      "loss": 3.2815,
      "step": 491490
    },
    {
      "epoch": 0.22558315751402233,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010171037103710372,
      "loss": 3.1988,
      "step": 491500
    },
    {
      "epoch": 0.22560180075018052,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001017083708370837,
      "loss": 3.3331,
      "step": 491510
    },
    {
      "epoch": 0.2256204439863387,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010170637063706372,
      "loss": 3.0306,
      "step": 491520
    },
    {
      "epoch": 0.2256390872224969,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010170437043704371,
      "loss": 3.2062,
      "step": 491530
    },
    {
      "epoch": 0.22565773045865506,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00010170237023702372,
      "loss": 3.1808,
      "step": 491540
    },
    {
      "epoch": 0.22567637369481325,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001017003700370037,
      "loss": 3.2011,
      "step": 491550
    },
    {
      "epoch": 0.22569501693097144,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010169836983698371,
      "loss": 3.1837,
      "step": 491560
    },
    {
      "epoch": 0.22571366016712963,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001016963696369637,
      "loss": 3.0914,
      "step": 491570
    },
    {
      "epoch": 0.22573230340328782,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001016943694369437,
      "loss": 3.2071,
      "step": 491580
    },
    {
      "epoch": 0.22575094663944598,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010169236923692368,
      "loss": 3.2023,
      "step": 491590
    },
    {
      "epoch": 0.22576958987560417,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001016903690369037,
      "loss": 3.408,
      "step": 491600
    },
    {
      "epoch": 0.22578823311176235,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010168836883688369,
      "loss": 3.2743,
      "step": 491610
    },
    {
      "epoch": 0.22580687634792054,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001016863686368637,
      "loss": 3.0808,
      "step": 491620
    },
    {
      "epoch": 0.22582551958407873,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010168436843684368,
      "loss": 3.224,
      "step": 491630
    },
    {
      "epoch": 0.2258441628202369,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010168236823682369,
      "loss": 3.2028,
      "step": 491640
    },
    {
      "epoch": 0.22586280605639508,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010168036803680367,
      "loss": 3.2724,
      "step": 491650
    },
    {
      "epoch": 0.22588144929255327,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010167836783678368,
      "loss": 3.1844,
      "step": 491660
    },
    {
      "epoch": 0.22590009252871146,
      "grad_norm": 0.9375,
      "learning_rate": 0.00010167636763676369,
      "loss": 3.1412,
      "step": 491670
    },
    {
      "epoch": 0.22591873576486965,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010167436743674367,
      "loss": 3.1601,
      "step": 491680
    },
    {
      "epoch": 0.22593737900102784,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010167236723672368,
      "loss": 3.1209,
      "step": 491690
    },
    {
      "epoch": 0.225956022237186,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010167036703670368,
      "loss": 3.3341,
      "step": 491700
    },
    {
      "epoch": 0.2259746654733442,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010166836683668369,
      "loss": 3.282,
      "step": 491710
    },
    {
      "epoch": 0.22599330870950238,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010166636663666367,
      "loss": 3.2961,
      "step": 491720
    },
    {
      "epoch": 0.22601195194566057,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010166436643664368,
      "loss": 3.1979,
      "step": 491730
    },
    {
      "epoch": 0.22603059518181876,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010166236623662366,
      "loss": 3.2461,
      "step": 491740
    },
    {
      "epoch": 0.22604923841797692,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010166036603660367,
      "loss": 3.2305,
      "step": 491750
    },
    {
      "epoch": 0.2260678816541351,
      "grad_norm": 0.90625,
      "learning_rate": 0.00010165836583658366,
      "loss": 3.2616,
      "step": 491760
    },
    {
      "epoch": 0.2260865248902933,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010165636563656366,
      "loss": 3.2996,
      "step": 491770
    },
    {
      "epoch": 0.22610516812645148,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010165436543654366,
      "loss": 3.1706,
      "step": 491780
    },
    {
      "epoch": 0.22612381136260967,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010165236523652366,
      "loss": 3.2471,
      "step": 491790
    },
    {
      "epoch": 0.22614245459876783,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010165036503650365,
      "loss": 3.4398,
      "step": 491800
    },
    {
      "epoch": 0.22616109783492602,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010164836483648366,
      "loss": 3.1959,
      "step": 491810
    },
    {
      "epoch": 0.2261797410710842,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010164636463646364,
      "loss": 3.1041,
      "step": 491820
    },
    {
      "epoch": 0.2261983843072424,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010164436443644365,
      "loss": 3.145,
      "step": 491830
    },
    {
      "epoch": 0.2262170275434006,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010164236423642364,
      "loss": 3.2909,
      "step": 491840
    },
    {
      "epoch": 0.22623567077955875,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010164036403640365,
      "loss": 3.1281,
      "step": 491850
    },
    {
      "epoch": 0.22625431401571694,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010163836383638366,
      "loss": 3.3265,
      "step": 491860
    },
    {
      "epoch": 0.22627295725187513,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010163636363636364,
      "loss": 3.2094,
      "step": 491870
    },
    {
      "epoch": 0.22629160048803332,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010163436343634365,
      "loss": 3.1297,
      "step": 491880
    },
    {
      "epoch": 0.2263102437241915,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010163236323632363,
      "loss": 3.2833,
      "step": 491890
    },
    {
      "epoch": 0.2263288869603497,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010163036303630364,
      "loss": 3.177,
      "step": 491900
    },
    {
      "epoch": 0.22634753019650786,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010162836283628363,
      "loss": 3.1872,
      "step": 491910
    },
    {
      "epoch": 0.22636617343266605,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010162636263626363,
      "loss": 3.1493,
      "step": 491920
    },
    {
      "epoch": 0.22638481666882423,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00010162436243624363,
      "loss": 3.1351,
      "step": 491930
    },
    {
      "epoch": 0.22640345990498242,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010162236223622364,
      "loss": 3.2111,
      "step": 491940
    },
    {
      "epoch": 0.2264221031411406,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010162036203620362,
      "loss": 3.1906,
      "step": 491950
    },
    {
      "epoch": 0.22644074637729877,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010161836183618363,
      "loss": 3.118,
      "step": 491960
    },
    {
      "epoch": 0.22645938961345696,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010161636163616361,
      "loss": 3.289,
      "step": 491970
    },
    {
      "epoch": 0.22647803284961515,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010161436143614362,
      "loss": 3.2394,
      "step": 491980
    },
    {
      "epoch": 0.22649667608577334,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001016123612361236,
      "loss": 3.1654,
      "step": 491990
    },
    {
      "epoch": 0.22651531932193153,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010161036103610361,
      "loss": 3.2795,
      "step": 492000
    },
    {
      "epoch": 0.2265339625580897,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010160836083608361,
      "loss": 3.2712,
      "step": 492010
    },
    {
      "epoch": 0.22655260579424788,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00010160636063606362,
      "loss": 3.1357,
      "step": 492020
    },
    {
      "epoch": 0.22657124903040607,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001016043604360436,
      "loss": 3.187,
      "step": 492030
    },
    {
      "epoch": 0.22658989226656426,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010160236023602361,
      "loss": 3.0851,
      "step": 492040
    },
    {
      "epoch": 0.22660853550272245,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010160036003600359,
      "loss": 3.2709,
      "step": 492050
    },
    {
      "epoch": 0.2266271787388806,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001015983598359836,
      "loss": 3.2148,
      "step": 492060
    },
    {
      "epoch": 0.2266458219750388,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010159635963596361,
      "loss": 3.2774,
      "step": 492070
    },
    {
      "epoch": 0.22666446521119699,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001015943594359436,
      "loss": 3.1785,
      "step": 492080
    },
    {
      "epoch": 0.22668310844735517,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001015923592359236,
      "loss": 3.1945,
      "step": 492090
    },
    {
      "epoch": 0.22670175168351336,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001015903590359036,
      "loss": 3.2518,
      "step": 492100
    },
    {
      "epoch": 0.22672039491967153,
      "grad_norm": 0.75,
      "learning_rate": 0.0001015883588358836,
      "loss": 3.1653,
      "step": 492110
    },
    {
      "epoch": 0.22673903815582971,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010158635863586359,
      "loss": 3.1593,
      "step": 492120
    },
    {
      "epoch": 0.2267576813919879,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001015843584358436,
      "loss": 3.2287,
      "step": 492130
    },
    {
      "epoch": 0.2267763246281461,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010158235823582358,
      "loss": 3.3419,
      "step": 492140
    },
    {
      "epoch": 0.22679496786430428,
      "grad_norm": 0.625,
      "learning_rate": 0.00010158035803580359,
      "loss": 3.2002,
      "step": 492150
    },
    {
      "epoch": 0.22681361110046247,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010157835783578358,
      "loss": 3.2429,
      "step": 492160
    },
    {
      "epoch": 0.22683225433662063,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010157635763576359,
      "loss": 3.2962,
      "step": 492170
    },
    {
      "epoch": 0.22685089757277882,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010157435743574357,
      "loss": 3.2205,
      "step": 492180
    },
    {
      "epoch": 0.226869540808937,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00010157235723572358,
      "loss": 3.2055,
      "step": 492190
    },
    {
      "epoch": 0.2268881840450952,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010157035703570357,
      "loss": 3.2721,
      "step": 492200
    },
    {
      "epoch": 0.2269068272812534,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010156835683568357,
      "loss": 3.2889,
      "step": 492210
    },
    {
      "epoch": 0.22692547051741155,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010156635663566356,
      "loss": 3.1618,
      "step": 492220
    },
    {
      "epoch": 0.22694411375356974,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010156435643564357,
      "loss": 3.3319,
      "step": 492230
    },
    {
      "epoch": 0.22696275698972793,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010156235623562356,
      "loss": 3.3347,
      "step": 492240
    },
    {
      "epoch": 0.22698140022588612,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010156035603560357,
      "loss": 3.1962,
      "step": 492250
    },
    {
      "epoch": 0.2270000434620443,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010155835583558358,
      "loss": 3.1179,
      "step": 492260
    },
    {
      "epoch": 0.22701868669820247,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010155635563556356,
      "loss": 3.2022,
      "step": 492270
    },
    {
      "epoch": 0.22703732993436065,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010155435543554357,
      "loss": 3.2142,
      "step": 492280
    },
    {
      "epoch": 0.22705597317051884,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010155235523552355,
      "loss": 3.212,
      "step": 492290
    },
    {
      "epoch": 0.22707461640667703,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010155035503550356,
      "loss": 3.217,
      "step": 492300
    },
    {
      "epoch": 0.22709325964283522,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010154835483548354,
      "loss": 3.2119,
      "step": 492310
    },
    {
      "epoch": 0.22711190287899338,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010154635463546355,
      "loss": 3.2478,
      "step": 492320
    },
    {
      "epoch": 0.22713054611515157,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010154435443544355,
      "loss": 3.3843,
      "step": 492330
    },
    {
      "epoch": 0.22714918935130976,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010154235423542356,
      "loss": 3.1295,
      "step": 492340
    },
    {
      "epoch": 0.22716783258746795,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010154035403540354,
      "loss": 3.0995,
      "step": 492350
    },
    {
      "epoch": 0.22718647582362614,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010153835383538355,
      "loss": 3.1358,
      "step": 492360
    },
    {
      "epoch": 0.22720511905978433,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010153635363536353,
      "loss": 3.1836,
      "step": 492370
    },
    {
      "epoch": 0.2272237622959425,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010153435343534354,
      "loss": 3.2478,
      "step": 492380
    },
    {
      "epoch": 0.22724240553210068,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010153235323532354,
      "loss": 3.2612,
      "step": 492390
    },
    {
      "epoch": 0.22726104876825887,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010153035303530354,
      "loss": 3.1675,
      "step": 492400
    },
    {
      "epoch": 0.22727969200441706,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010152835283528353,
      "loss": 3.1958,
      "step": 492410
    },
    {
      "epoch": 0.22729833524057524,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010152635263526354,
      "loss": 3.1835,
      "step": 492420
    },
    {
      "epoch": 0.2273169784767334,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010152435243524352,
      "loss": 3.1601,
      "step": 492430
    },
    {
      "epoch": 0.2273356217128916,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010152235223522353,
      "loss": 3.2602,
      "step": 492440
    },
    {
      "epoch": 0.22735426494904978,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010152035203520351,
      "loss": 3.1179,
      "step": 492450
    },
    {
      "epoch": 0.22737290818520797,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010151835183518352,
      "loss": 3.138,
      "step": 492460
    },
    {
      "epoch": 0.22739155142136616,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010151635163516353,
      "loss": 3.2127,
      "step": 492470
    },
    {
      "epoch": 0.22741019465752432,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010151435143514352,
      "loss": 3.2391,
      "step": 492480
    },
    {
      "epoch": 0.2274288378936825,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010151235123512353,
      "loss": 3.277,
      "step": 492490
    },
    {
      "epoch": 0.2274474811298407,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010151035103510351,
      "loss": 3.1686,
      "step": 492500
    },
    {
      "epoch": 0.2274661243659989,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010150835083508352,
      "loss": 3.2237,
      "step": 492510
    },
    {
      "epoch": 0.22748476760215708,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001015063506350635,
      "loss": 3.1807,
      "step": 492520
    },
    {
      "epoch": 0.22750341083831524,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010150435043504351,
      "loss": 3.3289,
      "step": 492530
    },
    {
      "epoch": 0.22752205407447343,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001015023502350235,
      "loss": 3.2104,
      "step": 492540
    },
    {
      "epoch": 0.22754069731063162,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001015003500350035,
      "loss": 3.1718,
      "step": 492550
    },
    {
      "epoch": 0.2275593405467898,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001014983498349835,
      "loss": 3.1822,
      "step": 492560
    },
    {
      "epoch": 0.227577983782948,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010149634963496351,
      "loss": 3.0911,
      "step": 492570
    },
    {
      "epoch": 0.22759662701910618,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010149434943494349,
      "loss": 3.3161,
      "step": 492580
    },
    {
      "epoch": 0.22761527025526435,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001014923492349235,
      "loss": 3.2987,
      "step": 492590
    },
    {
      "epoch": 0.22763391349142253,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010149034903490348,
      "loss": 3.1908,
      "step": 492600
    },
    {
      "epoch": 0.22765255672758072,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010148834883488349,
      "loss": 3.2496,
      "step": 492610
    },
    {
      "epoch": 0.2276711999637389,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010148634863486349,
      "loss": 3.2246,
      "step": 492620
    },
    {
      "epoch": 0.2276898431998971,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010148434843484348,
      "loss": 3.2298,
      "step": 492630
    },
    {
      "epoch": 0.22770848643605526,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00010148234823482348,
      "loss": 3.173,
      "step": 492640
    },
    {
      "epoch": 0.22772712967221345,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010148034803480349,
      "loss": 3.1653,
      "step": 492650
    },
    {
      "epoch": 0.22774577290837164,
      "grad_norm": 0.75,
      "learning_rate": 0.0001014783478347835,
      "loss": 3.2217,
      "step": 492660
    },
    {
      "epoch": 0.22776441614452983,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010147634763476348,
      "loss": 3.1145,
      "step": 492670
    },
    {
      "epoch": 0.22778305938068802,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010147434743474349,
      "loss": 3.4133,
      "step": 492680
    },
    {
      "epoch": 0.22780170261684618,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010147234723472347,
      "loss": 3.126,
      "step": 492690
    },
    {
      "epoch": 0.22782034585300437,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010147034703470348,
      "loss": 3.2864,
      "step": 492700
    },
    {
      "epoch": 0.22783898908916256,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010146834683468348,
      "loss": 3.2603,
      "step": 492710
    },
    {
      "epoch": 0.22785763232532075,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010146634663466349,
      "loss": 3.2495,
      "step": 492720
    },
    {
      "epoch": 0.22787627556147894,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010146434643464347,
      "loss": 3.2558,
      "step": 492730
    },
    {
      "epoch": 0.2278949187976371,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010146234623462348,
      "loss": 3.1019,
      "step": 492740
    },
    {
      "epoch": 0.22791356203379529,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010146034603460346,
      "loss": 3.1605,
      "step": 492750
    },
    {
      "epoch": 0.22793220526995348,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010145834583458347,
      "loss": 3.2036,
      "step": 492760
    },
    {
      "epoch": 0.22795084850611166,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010145634563456345,
      "loss": 3.0781,
      "step": 492770
    },
    {
      "epoch": 0.22796949174226985,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010145434543454346,
      "loss": 3.2563,
      "step": 492780
    },
    {
      "epoch": 0.22798813497842801,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010145234523452345,
      "loss": 3.2848,
      "step": 492790
    },
    {
      "epoch": 0.2280067782145862,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010145034503450346,
      "loss": 3.2677,
      "step": 492800
    },
    {
      "epoch": 0.2280254214507444,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010144834483448345,
      "loss": 3.1766,
      "step": 492810
    },
    {
      "epoch": 0.22804406468690258,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010144634463446345,
      "loss": 3.4619,
      "step": 492820
    },
    {
      "epoch": 0.22806270792306077,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010144434443444344,
      "loss": 3.3158,
      "step": 492830
    },
    {
      "epoch": 0.22808135115921896,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010144234423442345,
      "loss": 3.1715,
      "step": 492840
    },
    {
      "epoch": 0.22809999439537712,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010144034403440345,
      "loss": 3.171,
      "step": 492850
    },
    {
      "epoch": 0.2281186376315353,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010143834383438344,
      "loss": 3.2359,
      "step": 492860
    },
    {
      "epoch": 0.2281372808676935,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010143634363436345,
      "loss": 3.0677,
      "step": 492870
    },
    {
      "epoch": 0.2281559241038517,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010143434343434344,
      "loss": 3.0984,
      "step": 492880
    },
    {
      "epoch": 0.22817456734000988,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010143234323432345,
      "loss": 3.1286,
      "step": 492890
    },
    {
      "epoch": 0.22819321057616804,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010143034303430343,
      "loss": 3.2126,
      "step": 492900
    },
    {
      "epoch": 0.22821185381232623,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010142834283428344,
      "loss": 3.1322,
      "step": 492910
    },
    {
      "epoch": 0.22823049704848442,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010142634263426342,
      "loss": 3.2416,
      "step": 492920
    },
    {
      "epoch": 0.2282491402846426,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010142434243424343,
      "loss": 3.1489,
      "step": 492930
    },
    {
      "epoch": 0.2282677835208008,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010142234223422342,
      "loss": 3.204,
      "step": 492940
    },
    {
      "epoch": 0.22828642675695895,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010142034203420342,
      "loss": 3.267,
      "step": 492950
    },
    {
      "epoch": 0.22830506999311714,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010141834183418342,
      "loss": 3.2882,
      "step": 492960
    },
    {
      "epoch": 0.22832371322927533,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010141634163416343,
      "loss": 3.2063,
      "step": 492970
    },
    {
      "epoch": 0.22834235646543352,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010141434143414341,
      "loss": 3.1892,
      "step": 492980
    },
    {
      "epoch": 0.2283609997015917,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010141234123412342,
      "loss": 3.2726,
      "step": 492990
    },
    {
      "epoch": 0.22837964293774987,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001014103410341034,
      "loss": 3.1353,
      "step": 493000
    },
    {
      "epoch": 0.22839828617390806,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010140834083408341,
      "loss": 3.1497,
      "step": 493010
    },
    {
      "epoch": 0.22841692941006625,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010140634063406341,
      "loss": 3.166,
      "step": 493020
    },
    {
      "epoch": 0.22843557264622444,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010140434043404342,
      "loss": 3.1143,
      "step": 493030
    },
    {
      "epoch": 0.22845421588238263,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001014023402340234,
      "loss": 3.2667,
      "step": 493040
    },
    {
      "epoch": 0.22847285911854082,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010140034003400341,
      "loss": 3.0987,
      "step": 493050
    },
    {
      "epoch": 0.22849150235469898,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010139833983398342,
      "loss": 3.1419,
      "step": 493060
    },
    {
      "epoch": 0.22851014559085717,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001013963396339634,
      "loss": 3.2029,
      "step": 493070
    },
    {
      "epoch": 0.22852878882701536,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010139433943394341,
      "loss": 3.311,
      "step": 493080
    },
    {
      "epoch": 0.22854743206317354,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010139233923392339,
      "loss": 3.0384,
      "step": 493090
    },
    {
      "epoch": 0.22856607529933173,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001013903390339034,
      "loss": 3.2301,
      "step": 493100
    },
    {
      "epoch": 0.2285847185354899,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001013883388338834,
      "loss": 3.2365,
      "step": 493110
    },
    {
      "epoch": 0.22860336177164808,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001013863386338634,
      "loss": 3.2653,
      "step": 493120
    },
    {
      "epoch": 0.22862200500780627,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010138433843384339,
      "loss": 3.1336,
      "step": 493130
    },
    {
      "epoch": 0.22864064824396446,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001013823382338234,
      "loss": 3.2914,
      "step": 493140
    },
    {
      "epoch": 0.22865929148012265,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010138033803380338,
      "loss": 3.2552,
      "step": 493150
    },
    {
      "epoch": 0.2286779347162808,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010137833783378339,
      "loss": 3.0755,
      "step": 493160
    },
    {
      "epoch": 0.228696577952439,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010137633763376337,
      "loss": 3.2342,
      "step": 493170
    },
    {
      "epoch": 0.2287152211885972,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010137433743374338,
      "loss": 3.0519,
      "step": 493180
    },
    {
      "epoch": 0.22873386442475538,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010137233723372337,
      "loss": 3.1556,
      "step": 493190
    },
    {
      "epoch": 0.22875250766091357,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010137033703370338,
      "loss": 3.2157,
      "step": 493200
    },
    {
      "epoch": 0.22877115089707173,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010136833683368336,
      "loss": 3.3129,
      "step": 493210
    },
    {
      "epoch": 0.22878979413322992,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010136633663366337,
      "loss": 3.191,
      "step": 493220
    },
    {
      "epoch": 0.2288084373693881,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010136433643364336,
      "loss": 3.0329,
      "step": 493230
    },
    {
      "epoch": 0.2288270806055463,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010136233623362336,
      "loss": 3.2424,
      "step": 493240
    },
    {
      "epoch": 0.22884572384170448,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010136033603360337,
      "loss": 3.1052,
      "step": 493250
    },
    {
      "epoch": 0.22886436707786267,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010135833583358336,
      "loss": 3.1446,
      "step": 493260
    },
    {
      "epoch": 0.22888301031402083,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010135633563356336,
      "loss": 3.1156,
      "step": 493270
    },
    {
      "epoch": 0.22890165355017902,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010135433543354336,
      "loss": 3.2114,
      "step": 493280
    },
    {
      "epoch": 0.2289202967863372,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010135233523352337,
      "loss": 3.151,
      "step": 493290
    },
    {
      "epoch": 0.2289389400224954,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010135033503350335,
      "loss": 3.1269,
      "step": 493300
    },
    {
      "epoch": 0.2289575832586536,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010134833483348336,
      "loss": 3.1553,
      "step": 493310
    },
    {
      "epoch": 0.22897622649481175,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010134633463346334,
      "loss": 3.152,
      "step": 493320
    },
    {
      "epoch": 0.22899486973096994,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010134433443344335,
      "loss": 3.2228,
      "step": 493330
    },
    {
      "epoch": 0.22901351296712813,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010134233423342335,
      "loss": 3.1501,
      "step": 493340
    },
    {
      "epoch": 0.22903215620328632,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010134033403340336,
      "loss": 3.1745,
      "step": 493350
    },
    {
      "epoch": 0.2290507994394445,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010133833383338334,
      "loss": 3.1451,
      "step": 493360
    },
    {
      "epoch": 0.22906944267560267,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010133633363336335,
      "loss": 3.1595,
      "step": 493370
    },
    {
      "epoch": 0.22908808591176086,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010133433343334333,
      "loss": 3.2957,
      "step": 493380
    },
    {
      "epoch": 0.22910672914791905,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010133233323332334,
      "loss": 3.3008,
      "step": 493390
    },
    {
      "epoch": 0.22912537238407724,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010133033303330332,
      "loss": 3.2483,
      "step": 493400
    },
    {
      "epoch": 0.22914401562023543,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010132833283328333,
      "loss": 3.3752,
      "step": 493410
    },
    {
      "epoch": 0.2291626588563936,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010132633263326333,
      "loss": 3.2536,
      "step": 493420
    },
    {
      "epoch": 0.22918130209255178,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010132433243324333,
      "loss": 3.2621,
      "step": 493430
    },
    {
      "epoch": 0.22919994532870996,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010132233223322334,
      "loss": 3.1477,
      "step": 493440
    },
    {
      "epoch": 0.22921858856486815,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010132033203320333,
      "loss": 3.2033,
      "step": 493450
    },
    {
      "epoch": 0.22923723180102634,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010131833183318334,
      "loss": 3.1744,
      "step": 493460
    },
    {
      "epoch": 0.2292558750371845,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010131633163316332,
      "loss": 3.281,
      "step": 493470
    },
    {
      "epoch": 0.2292745182733427,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010131433143314333,
      "loss": 3.1447,
      "step": 493480
    },
    {
      "epoch": 0.22929316150950088,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010131233123312331,
      "loss": 3.1918,
      "step": 493490
    },
    {
      "epoch": 0.22931180474565907,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010131033103310332,
      "loss": 3.1096,
      "step": 493500
    },
    {
      "epoch": 0.22933044798181726,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010130833083308331,
      "loss": 3.2528,
      "step": 493510
    },
    {
      "epoch": 0.22934909121797545,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010130633063306332,
      "loss": 3.0347,
      "step": 493520
    },
    {
      "epoch": 0.2293677344541336,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001013043304330433,
      "loss": 3.1223,
      "step": 493530
    },
    {
      "epoch": 0.2293863776902918,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010130233023302331,
      "loss": 3.1131,
      "step": 493540
    },
    {
      "epoch": 0.22940502092645,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001013003300330033,
      "loss": 3.1755,
      "step": 493550
    },
    {
      "epoch": 0.22942366416260818,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001012983298329833,
      "loss": 3.2892,
      "step": 493560
    },
    {
      "epoch": 0.22944230739876637,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001012963296329633,
      "loss": 3.2542,
      "step": 493570
    },
    {
      "epoch": 0.22946095063492453,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001012943294329433,
      "loss": 3.2613,
      "step": 493580
    },
    {
      "epoch": 0.22947959387108272,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010129232923292329,
      "loss": 3.3337,
      "step": 493590
    },
    {
      "epoch": 0.2294982371072409,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001012903290329033,
      "loss": 3.1356,
      "step": 493600
    },
    {
      "epoch": 0.2295168803433991,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010128832883288328,
      "loss": 3.252,
      "step": 493610
    },
    {
      "epoch": 0.22953552357955728,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010128632863286329,
      "loss": 3.2284,
      "step": 493620
    },
    {
      "epoch": 0.22955416681571544,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010128432843284327,
      "loss": 3.1904,
      "step": 493630
    },
    {
      "epoch": 0.22957281005187363,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010128232823282328,
      "loss": 3.2043,
      "step": 493640
    },
    {
      "epoch": 0.22959145328803182,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010128032803280329,
      "loss": 3.2187,
      "step": 493650
    },
    {
      "epoch": 0.22961009652419,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010127832783278329,
      "loss": 3.3867,
      "step": 493660
    },
    {
      "epoch": 0.2296287397603482,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001012763276327633,
      "loss": 3.1224,
      "step": 493670
    },
    {
      "epoch": 0.22964738299650636,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010127432743274328,
      "loss": 3.2877,
      "step": 493680
    },
    {
      "epoch": 0.22966602623266455,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010127232723272329,
      "loss": 3.1809,
      "step": 493690
    },
    {
      "epoch": 0.22968466946882274,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010127032703270327,
      "loss": 3.1021,
      "step": 493700
    },
    {
      "epoch": 0.22970331270498093,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010126832683268328,
      "loss": 3.302,
      "step": 493710
    },
    {
      "epoch": 0.22972195594113912,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010126632663266326,
      "loss": 3.1663,
      "step": 493720
    },
    {
      "epoch": 0.2297405991772973,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010126432643264327,
      "loss": 3.1626,
      "step": 493730
    },
    {
      "epoch": 0.22975924241345547,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010126232623262327,
      "loss": 3.2005,
      "step": 493740
    },
    {
      "epoch": 0.22977788564961366,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010126032603260328,
      "loss": 3.2336,
      "step": 493750
    },
    {
      "epoch": 0.22979652888577184,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010125832583258326,
      "loss": 3.2652,
      "step": 493760
    },
    {
      "epoch": 0.22981517212193003,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010125632563256327,
      "loss": 3.1938,
      "step": 493770
    },
    {
      "epoch": 0.22983381535808822,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010125432543254325,
      "loss": 3.1436,
      "step": 493780
    },
    {
      "epoch": 0.22985245859424638,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010125232523252326,
      "loss": 3.1212,
      "step": 493790
    },
    {
      "epoch": 0.22987110183040457,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010125032503250324,
      "loss": 3.2639,
      "step": 493800
    },
    {
      "epoch": 0.22988974506656276,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010124832483248325,
      "loss": 3.1308,
      "step": 493810
    },
    {
      "epoch": 0.22990838830272095,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010124632463246324,
      "loss": 3.2154,
      "step": 493820
    },
    {
      "epoch": 0.22992703153887914,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010124432443244325,
      "loss": 3.2166,
      "step": 493830
    },
    {
      "epoch": 0.2299456747750373,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010124232423242326,
      "loss": 3.1977,
      "step": 493840
    },
    {
      "epoch": 0.2299643180111955,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010124032403240324,
      "loss": 3.2447,
      "step": 493850
    },
    {
      "epoch": 0.22998296124735368,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010123832383238325,
      "loss": 3.1985,
      "step": 493860
    },
    {
      "epoch": 0.23000160448351187,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010123632363236324,
      "loss": 3.1481,
      "step": 493870
    },
    {
      "epoch": 0.23002024771967006,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010123432343234325,
      "loss": 3.2931,
      "step": 493880
    },
    {
      "epoch": 0.23003889095582822,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010123232323232324,
      "loss": 3.3256,
      "step": 493890
    },
    {
      "epoch": 0.2300575341919864,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010123032303230324,
      "loss": 3.2146,
      "step": 493900
    },
    {
      "epoch": 0.2300761774281446,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010122832283228323,
      "loss": 3.1376,
      "step": 493910
    },
    {
      "epoch": 0.23009482066430278,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010122632263226324,
      "loss": 3.1289,
      "step": 493920
    },
    {
      "epoch": 0.23011346390046097,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010122432243224322,
      "loss": 3.2222,
      "step": 493930
    },
    {
      "epoch": 0.23013210713661916,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010122232223222323,
      "loss": 3.4545,
      "step": 493940
    },
    {
      "epoch": 0.23015075037277732,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010122032203220321,
      "loss": 3.2483,
      "step": 493950
    },
    {
      "epoch": 0.2301693936089355,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010121832183218322,
      "loss": 3.2277,
      "step": 493960
    },
    {
      "epoch": 0.2301880368450937,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010121632163216322,
      "loss": 3.2432,
      "step": 493970
    },
    {
      "epoch": 0.2302066800812519,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010121432143214323,
      "loss": 3.3336,
      "step": 493980
    },
    {
      "epoch": 0.23022532331741008,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010121232123212321,
      "loss": 3.1967,
      "step": 493990
    },
    {
      "epoch": 0.23024396655356824,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010121032103210322,
      "loss": 3.1917,
      "step": 494000
    },
    {
      "epoch": 0.23026260978972643,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001012083208320832,
      "loss": 3.3637,
      "step": 494010
    },
    {
      "epoch": 0.23028125302588462,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010120632063206321,
      "loss": 3.2903,
      "step": 494020
    },
    {
      "epoch": 0.2302998962620428,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010120432043204319,
      "loss": 3.2731,
      "step": 494030
    },
    {
      "epoch": 0.230318539498201,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001012023202320232,
      "loss": 3.2195,
      "step": 494040
    },
    {
      "epoch": 0.23033718273435916,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010120032003200321,
      "loss": 3.2567,
      "step": 494050
    },
    {
      "epoch": 0.23035582597051735,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001011983198319832,
      "loss": 3.2538,
      "step": 494060
    },
    {
      "epoch": 0.23037446920667554,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010119631963196322,
      "loss": 3.2034,
      "step": 494070
    },
    {
      "epoch": 0.23039311244283373,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001011943194319432,
      "loss": 3.1342,
      "step": 494080
    },
    {
      "epoch": 0.23041175567899191,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001011923192319232,
      "loss": 3.212,
      "step": 494090
    },
    {
      "epoch": 0.23043039891515008,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010119031903190319,
      "loss": 3.3958,
      "step": 494100
    },
    {
      "epoch": 0.23044904215130826,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001011883188318832,
      "loss": 3.1935,
      "step": 494110
    },
    {
      "epoch": 0.23046768538746645,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010118631863186318,
      "loss": 3.191,
      "step": 494120
    },
    {
      "epoch": 0.23048632862362464,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010118431843184319,
      "loss": 2.9696,
      "step": 494130
    },
    {
      "epoch": 0.23050497185978283,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010118231823182318,
      "loss": 3.174,
      "step": 494140
    },
    {
      "epoch": 0.230523615095941,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001011803180318032,
      "loss": 3.2643,
      "step": 494150
    },
    {
      "epoch": 0.23054225833209918,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010117831783178318,
      "loss": 3.0673,
      "step": 494160
    },
    {
      "epoch": 0.23056090156825737,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010117631763176319,
      "loss": 3.1795,
      "step": 494170
    },
    {
      "epoch": 0.23057954480441556,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010117431743174317,
      "loss": 3.3217,
      "step": 494180
    },
    {
      "epoch": 0.23059818804057375,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010117231723172318,
      "loss": 3.3183,
      "step": 494190
    },
    {
      "epoch": 0.23061683127673194,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010117031703170317,
      "loss": 3.4017,
      "step": 494200
    },
    {
      "epoch": 0.2306354745128901,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010116831683168318,
      "loss": 3.1327,
      "step": 494210
    },
    {
      "epoch": 0.2306541177490483,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010116631663166316,
      "loss": 3.2462,
      "step": 494220
    },
    {
      "epoch": 0.23067276098520648,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010116431643164317,
      "loss": 3.2883,
      "step": 494230
    },
    {
      "epoch": 0.23069140422136467,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010116231623162318,
      "loss": 2.9817,
      "step": 494240
    },
    {
      "epoch": 0.23071004745752285,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010116031603160316,
      "loss": 3.2036,
      "step": 494250
    },
    {
      "epoch": 0.23072869069368102,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010115831583158317,
      "loss": 3.2045,
      "step": 494260
    },
    {
      "epoch": 0.2307473339298392,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010115631563156315,
      "loss": 2.9355,
      "step": 494270
    },
    {
      "epoch": 0.2307659771659974,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010115431543154316,
      "loss": 3.1651,
      "step": 494280
    },
    {
      "epoch": 0.23078462040215558,
      "grad_norm": 0.75,
      "learning_rate": 0.00010115231523152316,
      "loss": 3.1052,
      "step": 494290
    },
    {
      "epoch": 0.23080326363831377,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010115031503150317,
      "loss": 3.305,
      "step": 494300
    },
    {
      "epoch": 0.23082190687447193,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010114831483148315,
      "loss": 3.2536,
      "step": 494310
    },
    {
      "epoch": 0.23084055011063012,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010114631463146316,
      "loss": 3.1428,
      "step": 494320
    },
    {
      "epoch": 0.2308591933467883,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010114431443144314,
      "loss": 3.2016,
      "step": 494330
    },
    {
      "epoch": 0.2308778365829465,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010114231423142315,
      "loss": 3.2788,
      "step": 494340
    },
    {
      "epoch": 0.2308964798191047,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010114031403140313,
      "loss": 3.1512,
      "step": 494350
    },
    {
      "epoch": 0.23091512305526285,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010113831383138314,
      "loss": 3.1996,
      "step": 494360
    },
    {
      "epoch": 0.23093376629142104,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010113631363136314,
      "loss": 3.3486,
      "step": 494370
    },
    {
      "epoch": 0.23095240952757923,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010113431343134315,
      "loss": 3.3373,
      "step": 494380
    },
    {
      "epoch": 0.23097105276373742,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010113231323132313,
      "loss": 3.3096,
      "step": 494390
    },
    {
      "epoch": 0.2309896959998956,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010113031303130314,
      "loss": 3.389,
      "step": 494400
    },
    {
      "epoch": 0.2310083392360538,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010112831283128312,
      "loss": 3.2706,
      "step": 494410
    },
    {
      "epoch": 0.23102698247221196,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010112631263126313,
      "loss": 3.2694,
      "step": 494420
    },
    {
      "epoch": 0.23104562570837014,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010112431243124314,
      "loss": 3.0653,
      "step": 494430
    },
    {
      "epoch": 0.23106426894452833,
      "grad_norm": 0.75,
      "learning_rate": 0.00010112231223122312,
      "loss": 3.2345,
      "step": 494440
    },
    {
      "epoch": 0.23108291218068652,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010112031203120313,
      "loss": 3.3186,
      "step": 494450
    },
    {
      "epoch": 0.2311015554168447,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010111831183118313,
      "loss": 3.277,
      "step": 494460
    },
    {
      "epoch": 0.23112019865300287,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010111631163116313,
      "loss": 3.2019,
      "step": 494470
    },
    {
      "epoch": 0.23113884188916106,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010111431143114312,
      "loss": 3.2307,
      "step": 494480
    },
    {
      "epoch": 0.23115748512531925,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010111231123112313,
      "loss": 3.2404,
      "step": 494490
    },
    {
      "epoch": 0.23117612836147744,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010111031103110311,
      "loss": 3.3312,
      "step": 494500
    },
    {
      "epoch": 0.23119477159763563,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010110831083108312,
      "loss": 3.2362,
      "step": 494510
    },
    {
      "epoch": 0.2312134148337938,
      "grad_norm": 0.75,
      "learning_rate": 0.00010110631063106311,
      "loss": 3.0913,
      "step": 494520
    },
    {
      "epoch": 0.23123205806995198,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010110431043104311,
      "loss": 3.1793,
      "step": 494530
    },
    {
      "epoch": 0.23125070130611017,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001011023102310231,
      "loss": 3.2745,
      "step": 494540
    },
    {
      "epoch": 0.23126934454226836,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010110031003100311,
      "loss": 3.2265,
      "step": 494550
    },
    {
      "epoch": 0.23128798777842655,
      "grad_norm": 0.75,
      "learning_rate": 0.0001010983098309831,
      "loss": 3.2532,
      "step": 494560
    },
    {
      "epoch": 0.2313066310145847,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001010963096309631,
      "loss": 3.11,
      "step": 494570
    },
    {
      "epoch": 0.2313252742507429,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010109430943094309,
      "loss": 3.2096,
      "step": 494580
    },
    {
      "epoch": 0.23134391748690109,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001010923092309231,
      "loss": 3.2558,
      "step": 494590
    },
    {
      "epoch": 0.23136256072305927,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010109030903090309,
      "loss": 3.1967,
      "step": 494600
    },
    {
      "epoch": 0.23138120395921746,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001010883088308831,
      "loss": 3.1021,
      "step": 494610
    },
    {
      "epoch": 0.23139984719537565,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010108630863086308,
      "loss": 3.1418,
      "step": 494620
    },
    {
      "epoch": 0.2314184904315338,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010108430843084309,
      "loss": 3.308,
      "step": 494630
    },
    {
      "epoch": 0.231437133667692,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001010823082308231,
      "loss": 3.1404,
      "step": 494640
    },
    {
      "epoch": 0.2314557769038502,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010108030803080308,
      "loss": 3.2047,
      "step": 494650
    },
    {
      "epoch": 0.23147442014000838,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010107830783078309,
      "loss": 3.0986,
      "step": 494660
    },
    {
      "epoch": 0.23149306337616657,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010107630763076307,
      "loss": 3.2306,
      "step": 494670
    },
    {
      "epoch": 0.23151170661232473,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010107430743074308,
      "loss": 3.2583,
      "step": 494680
    },
    {
      "epoch": 0.23153034984848292,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010107230723072308,
      "loss": 3.3152,
      "step": 494690
    },
    {
      "epoch": 0.2315489930846411,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010107030703070309,
      "loss": 3.2863,
      "step": 494700
    },
    {
      "epoch": 0.2315676363207993,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010106830683068307,
      "loss": 3.3378,
      "step": 494710
    },
    {
      "epoch": 0.2315862795569575,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010106630663066308,
      "loss": 3.214,
      "step": 494720
    },
    {
      "epoch": 0.23160492279311565,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010106430643064306,
      "loss": 3.2383,
      "step": 494730
    },
    {
      "epoch": 0.23162356602927384,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010106230623062307,
      "loss": 3.2064,
      "step": 494740
    },
    {
      "epoch": 0.23164220926543203,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010106030603060305,
      "loss": 3.1762,
      "step": 494750
    },
    {
      "epoch": 0.23166085250159021,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010105830583058306,
      "loss": 3.2041,
      "step": 494760
    },
    {
      "epoch": 0.2316794957377484,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010105630563056306,
      "loss": 3.1135,
      "step": 494770
    },
    {
      "epoch": 0.23169813897390656,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010105430543054307,
      "loss": 3.1894,
      "step": 494780
    },
    {
      "epoch": 0.23171678221006475,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010105230523052305,
      "loss": 3.292,
      "step": 494790
    },
    {
      "epoch": 0.23173542544622294,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010105030503050306,
      "loss": 3.1205,
      "step": 494800
    },
    {
      "epoch": 0.23175406868238113,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010104830483048304,
      "loss": 3.2501,
      "step": 494810
    },
    {
      "epoch": 0.23177271191853932,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010104630463046305,
      "loss": 3.1788,
      "step": 494820
    },
    {
      "epoch": 0.2317913551546975,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010104430443044306,
      "loss": 3.2675,
      "step": 494830
    },
    {
      "epoch": 0.23180999839085567,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010104230423042305,
      "loss": 3.2394,
      "step": 494840
    },
    {
      "epoch": 0.23182864162701386,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010104030403040305,
      "loss": 3.1715,
      "step": 494850
    },
    {
      "epoch": 0.23184728486317205,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010103830383038304,
      "loss": 3.1633,
      "step": 494860
    },
    {
      "epoch": 0.23186592809933024,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010103630363036305,
      "loss": 3.1727,
      "step": 494870
    },
    {
      "epoch": 0.23188457133548843,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010103430343034303,
      "loss": 3.1536,
      "step": 494880
    },
    {
      "epoch": 0.2319032145716466,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010103230323032304,
      "loss": 3.3877,
      "step": 494890
    },
    {
      "epoch": 0.23192185780780478,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010103030303030303,
      "loss": 3.2029,
      "step": 494900
    },
    {
      "epoch": 0.23194050104396297,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010102830283028304,
      "loss": 3.3832,
      "step": 494910
    },
    {
      "epoch": 0.23195914428012115,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010102630263026303,
      "loss": 3.3218,
      "step": 494920
    },
    {
      "epoch": 0.23197778751627934,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010102430243024304,
      "loss": 3.2988,
      "step": 494930
    },
    {
      "epoch": 0.2319964307524375,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010102230223022302,
      "loss": 3.3199,
      "step": 494940
    },
    {
      "epoch": 0.2320150739885957,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010102030203020303,
      "loss": 3.2298,
      "step": 494950
    },
    {
      "epoch": 0.23203371722475388,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010101830183018301,
      "loss": 3.1444,
      "step": 494960
    },
    {
      "epoch": 0.23205236046091207,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010101630163016302,
      "loss": 3.2931,
      "step": 494970
    },
    {
      "epoch": 0.23207100369707026,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000101014301430143,
      "loss": 3.2015,
      "step": 494980
    },
    {
      "epoch": 0.23208964693322842,
      "grad_norm": 0.75,
      "learning_rate": 0.00010101230123012301,
      "loss": 3.2402,
      "step": 494990
    },
    {
      "epoch": 0.2321082901693866,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010101030103010301,
      "loss": 3.2728,
      "step": 495000
    },
    {
      "epoch": 0.2321269334055448,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010100830083008302,
      "loss": 3.1863,
      "step": 495010
    },
    {
      "epoch": 0.232145576641703,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010100630063006303,
      "loss": 3.1058,
      "step": 495020
    },
    {
      "epoch": 0.23216421987786118,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010100430043004301,
      "loss": 3.1389,
      "step": 495030
    },
    {
      "epoch": 0.23218286311401934,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010100230023002302,
      "loss": 3.1394,
      "step": 495040
    },
    {
      "epoch": 0.23220150635017753,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000101000300030003,
      "loss": 3.1874,
      "step": 495050
    },
    {
      "epoch": 0.23222014958633572,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010099829982998301,
      "loss": 3.2112,
      "step": 495060
    },
    {
      "epoch": 0.2322387928224939,
      "grad_norm": 0.625,
      "learning_rate": 0.00010099629962996299,
      "loss": 3.2192,
      "step": 495070
    },
    {
      "epoch": 0.2322574360586521,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000100994299429943,
      "loss": 3.3791,
      "step": 495080
    },
    {
      "epoch": 0.23227607929481028,
      "grad_norm": 0.703125,
      "learning_rate": 0.000100992299229923,
      "loss": 3.181,
      "step": 495090
    },
    {
      "epoch": 0.23229472253096844,
      "grad_norm": 0.7421875,
      "learning_rate": 0.000100990299029903,
      "loss": 3.1691,
      "step": 495100
    },
    {
      "epoch": 0.23231336576712663,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010098829882988299,
      "loss": 3.2175,
      "step": 495110
    },
    {
      "epoch": 0.23233200900328482,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000100986298629863,
      "loss": 3.2445,
      "step": 495120
    },
    {
      "epoch": 0.232350652239443,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010098429842984298,
      "loss": 3.2124,
      "step": 495130
    },
    {
      "epoch": 0.2323692954756012,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010098229822982299,
      "loss": 3.133,
      "step": 495140
    },
    {
      "epoch": 0.23238793871175936,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010098029802980298,
      "loss": 3.1799,
      "step": 495150
    },
    {
      "epoch": 0.23240658194791755,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010097829782978299,
      "loss": 3.2291,
      "step": 495160
    },
    {
      "epoch": 0.23242522518407574,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010097629762976297,
      "loss": 3.2412,
      "step": 495170
    },
    {
      "epoch": 0.23244386842023393,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010097429742974298,
      "loss": 3.0526,
      "step": 495180
    },
    {
      "epoch": 0.23246251165639212,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010097229722972297,
      "loss": 3.2621,
      "step": 495190
    },
    {
      "epoch": 0.23248115489255028,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010097029702970298,
      "loss": 3.2081,
      "step": 495200
    },
    {
      "epoch": 0.23249979812870847,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010096829682968296,
      "loss": 3.2027,
      "step": 495210
    },
    {
      "epoch": 0.23251844136486666,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010096629662966297,
      "loss": 3.206,
      "step": 495220
    },
    {
      "epoch": 0.23253708460102485,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010096429642964298,
      "loss": 3.2135,
      "step": 495230
    },
    {
      "epoch": 0.23255572783718303,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010096229622962297,
      "loss": 3.1812,
      "step": 495240
    },
    {
      "epoch": 0.2325743710733412,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010096029602960298,
      "loss": 3.0756,
      "step": 495250
    },
    {
      "epoch": 0.23259301430949939,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00010095829582958296,
      "loss": 3.12,
      "step": 495260
    },
    {
      "epoch": 0.23261165754565757,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010095629562956297,
      "loss": 3.1783,
      "step": 495270
    },
    {
      "epoch": 0.23263030078181576,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010095429542954295,
      "loss": 3.185,
      "step": 495280
    },
    {
      "epoch": 0.23264894401797395,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010095229522952296,
      "loss": 3.129,
      "step": 495290
    },
    {
      "epoch": 0.23266758725413214,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010095029502950294,
      "loss": 3.271,
      "step": 495300
    },
    {
      "epoch": 0.2326862304902903,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010094829482948295,
      "loss": 3.3243,
      "step": 495310
    },
    {
      "epoch": 0.2327048737264485,
      "grad_norm": 0.75,
      "learning_rate": 0.00010094629462946295,
      "loss": 3.0033,
      "step": 495320
    },
    {
      "epoch": 0.23272351696260668,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010094429442944296,
      "loss": 3.2374,
      "step": 495330
    },
    {
      "epoch": 0.23274216019876487,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010094229422942294,
      "loss": 3.259,
      "step": 495340
    },
    {
      "epoch": 0.23276080343492306,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010094029402940295,
      "loss": 3.0676,
      "step": 495350
    },
    {
      "epoch": 0.23277944667108122,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010093829382938293,
      "loss": 3.2522,
      "step": 495360
    },
    {
      "epoch": 0.2327980899072394,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010093629362936294,
      "loss": 3.2135,
      "step": 495370
    },
    {
      "epoch": 0.2328167331433976,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010093429342934294,
      "loss": 3.1568,
      "step": 495380
    },
    {
      "epoch": 0.2328353763795558,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010093229322932293,
      "loss": 3.2698,
      "step": 495390
    },
    {
      "epoch": 0.23285401961571398,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010093029302930293,
      "loss": 3.336,
      "step": 495400
    },
    {
      "epoch": 0.23287266285187214,
      "grad_norm": 0.75,
      "learning_rate": 0.00010092829282928294,
      "loss": 3.1652,
      "step": 495410
    },
    {
      "epoch": 0.23289130608803033,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010092629262926295,
      "loss": 3.3105,
      "step": 495420
    },
    {
      "epoch": 0.23290994932418851,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010092429242924293,
      "loss": 3.1682,
      "step": 495430
    },
    {
      "epoch": 0.2329285925603467,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010092229222922294,
      "loss": 3.2584,
      "step": 495440
    },
    {
      "epoch": 0.2329472357965049,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00010092029202920292,
      "loss": 3.3054,
      "step": 495450
    },
    {
      "epoch": 0.23296587903266305,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010091829182918293,
      "loss": 3.2329,
      "step": 495460
    },
    {
      "epoch": 0.23298452226882124,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010091629162916292,
      "loss": 3.1432,
      "step": 495470
    },
    {
      "epoch": 0.23300316550497943,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010091429142914293,
      "loss": 3.0859,
      "step": 495480
    },
    {
      "epoch": 0.23302180874113762,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010091229122912292,
      "loss": 3.1488,
      "step": 495490
    },
    {
      "epoch": 0.2330404519772958,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010091029102910292,
      "loss": 3.2097,
      "step": 495500
    },
    {
      "epoch": 0.233059095213454,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001009082908290829,
      "loss": 3.2378,
      "step": 495510
    },
    {
      "epoch": 0.23307773844961216,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010090629062906292,
      "loss": 3.27,
      "step": 495520
    },
    {
      "epoch": 0.23309638168577035,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001009042904290429,
      "loss": 3.0913,
      "step": 495530
    },
    {
      "epoch": 0.23311502492192854,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001009022902290229,
      "loss": 3.2215,
      "step": 495540
    },
    {
      "epoch": 0.23313366815808673,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001009002900290029,
      "loss": 3.2357,
      "step": 495550
    },
    {
      "epoch": 0.23315231139424492,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010089828982898291,
      "loss": 3.298,
      "step": 495560
    },
    {
      "epoch": 0.23317095463040308,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001008962896289629,
      "loss": 3.122,
      "step": 495570
    },
    {
      "epoch": 0.23318959786656127,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001008942894289429,
      "loss": 3.2979,
      "step": 495580
    },
    {
      "epoch": 0.23320824110271945,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010089228922892288,
      "loss": 3.1024,
      "step": 495590
    },
    {
      "epoch": 0.23322688433887764,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001008902890289029,
      "loss": 3.1047,
      "step": 495600
    },
    {
      "epoch": 0.23324552757503583,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010088828882888288,
      "loss": 3.1871,
      "step": 495610
    },
    {
      "epoch": 0.233264170811194,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00010088628862886289,
      "loss": 3.0703,
      "step": 495620
    },
    {
      "epoch": 0.23328281404735218,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001008842884288429,
      "loss": 3.1733,
      "step": 495630
    },
    {
      "epoch": 0.23330145728351037,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010088228822882289,
      "loss": 3.2694,
      "step": 495640
    },
    {
      "epoch": 0.23332010051966856,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001008802880288029,
      "loss": 3.3569,
      "step": 495650
    },
    {
      "epoch": 0.23333874375582675,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010087828782878288,
      "loss": 3.1516,
      "step": 495660
    },
    {
      "epoch": 0.2333573869919849,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010087628762876289,
      "loss": 3.1327,
      "step": 495670
    },
    {
      "epoch": 0.2333760302281431,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010087428742874287,
      "loss": 3.2447,
      "step": 495680
    },
    {
      "epoch": 0.2333946734643013,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010087228722872288,
      "loss": 3.0326,
      "step": 495690
    },
    {
      "epoch": 0.23341331670045948,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010087028702870286,
      "loss": 3.1789,
      "step": 495700
    },
    {
      "epoch": 0.23343195993661767,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010086828682868287,
      "loss": 3.2722,
      "step": 495710
    },
    {
      "epoch": 0.23345060317277583,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010086628662866287,
      "loss": 3.1053,
      "step": 495720
    },
    {
      "epoch": 0.23346924640893402,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010086428642864288,
      "loss": 3.3208,
      "step": 495730
    },
    {
      "epoch": 0.2334878896450922,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010086228622862286,
      "loss": 3.2722,
      "step": 495740
    },
    {
      "epoch": 0.2335065328812504,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010086028602860287,
      "loss": 3.1801,
      "step": 495750
    },
    {
      "epoch": 0.23352517611740858,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010085828582858285,
      "loss": 3.327,
      "step": 495760
    },
    {
      "epoch": 0.23354381935356677,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010085628562856286,
      "loss": 3.23,
      "step": 495770
    },
    {
      "epoch": 0.23356246258972493,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010085428542854286,
      "loss": 3.2712,
      "step": 495780
    },
    {
      "epoch": 0.23358110582588312,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010085228522852286,
      "loss": 3.2294,
      "step": 495790
    },
    {
      "epoch": 0.2335997490620413,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010085028502850285,
      "loss": 3.0701,
      "step": 495800
    },
    {
      "epoch": 0.2336183922981995,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010084828482848286,
      "loss": 3.228,
      "step": 495810
    },
    {
      "epoch": 0.2336370355343577,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010084628462846286,
      "loss": 3.1859,
      "step": 495820
    },
    {
      "epoch": 0.23365567877051585,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010084428442844285,
      "loss": 2.9909,
      "step": 495830
    },
    {
      "epoch": 0.23367432200667404,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010084228422842286,
      "loss": 3.137,
      "step": 495840
    },
    {
      "epoch": 0.23369296524283223,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010084028402840284,
      "loss": 3.257,
      "step": 495850
    },
    {
      "epoch": 0.23371160847899042,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010083828382838285,
      "loss": 3.2585,
      "step": 495860
    },
    {
      "epoch": 0.2337302517151486,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010083628362836284,
      "loss": 3.3394,
      "step": 495870
    },
    {
      "epoch": 0.23374889495130677,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010083428342834285,
      "loss": 3.3512,
      "step": 495880
    },
    {
      "epoch": 0.23376753818746496,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010083228322832283,
      "loss": 3.2095,
      "step": 495890
    },
    {
      "epoch": 0.23378618142362315,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010083028302830284,
      "loss": 3.1561,
      "step": 495900
    },
    {
      "epoch": 0.23380482465978134,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010082828282828283,
      "loss": 3.2785,
      "step": 495910
    },
    {
      "epoch": 0.23382346789593952,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010082628262826283,
      "loss": 3.1982,
      "step": 495920
    },
    {
      "epoch": 0.23384211113209769,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010082428242824282,
      "loss": 3.4649,
      "step": 495930
    },
    {
      "epoch": 0.23386075436825587,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010082228222822283,
      "loss": 3.2302,
      "step": 495940
    },
    {
      "epoch": 0.23387939760441406,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010082028202820282,
      "loss": 3.276,
      "step": 495950
    },
    {
      "epoch": 0.23389804084057225,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010081828182818283,
      "loss": 3.1842,
      "step": 495960
    },
    {
      "epoch": 0.23391668407673044,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010081628162816281,
      "loss": 3.3227,
      "step": 495970
    },
    {
      "epoch": 0.23393532731288863,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010081428142814282,
      "loss": 3.3121,
      "step": 495980
    },
    {
      "epoch": 0.2339539705490468,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001008122812281228,
      "loss": 3.1862,
      "step": 495990
    },
    {
      "epoch": 0.23397261378520498,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010081028102810281,
      "loss": 3.1462,
      "step": 496000
    },
    {
      "epoch": 0.23399125702136317,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010080828082808282,
      "loss": 3.2437,
      "step": 496010
    },
    {
      "epoch": 0.23400990025752136,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001008062806280628,
      "loss": 3.2305,
      "step": 496020
    },
    {
      "epoch": 0.23402854349367955,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010080428042804281,
      "loss": 3.2281,
      "step": 496030
    },
    {
      "epoch": 0.2340471867298377,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010080228022802281,
      "loss": 3.3364,
      "step": 496040
    },
    {
      "epoch": 0.2340658299659959,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010080028002800282,
      "loss": 3.1386,
      "step": 496050
    },
    {
      "epoch": 0.2340844732021541,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001007982798279828,
      "loss": 3.2159,
      "step": 496060
    },
    {
      "epoch": 0.23410311643831228,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010079627962796281,
      "loss": 3.2588,
      "step": 496070
    },
    {
      "epoch": 0.23412175967447046,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010079427942794279,
      "loss": 3.1299,
      "step": 496080
    },
    {
      "epoch": 0.23414040291062863,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001007922792279228,
      "loss": 3.195,
      "step": 496090
    },
    {
      "epoch": 0.23415904614678681,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001007902790279028,
      "loss": 3.2803,
      "step": 496100
    },
    {
      "epoch": 0.234177689382945,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001007882788278828,
      "loss": 3.2894,
      "step": 496110
    },
    {
      "epoch": 0.2341963326191032,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010078627862786279,
      "loss": 3.1918,
      "step": 496120
    },
    {
      "epoch": 0.23421497585526138,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001007842784278428,
      "loss": 3.154,
      "step": 496130
    },
    {
      "epoch": 0.23423361909141954,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010078227822782278,
      "loss": 3.3724,
      "step": 496140
    },
    {
      "epoch": 0.23425226232757773,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010078027802780279,
      "loss": 3.2805,
      "step": 496150
    },
    {
      "epoch": 0.23427090556373592,
      "grad_norm": 0.625,
      "learning_rate": 0.00010077827782778277,
      "loss": 3.2008,
      "step": 496160
    },
    {
      "epoch": 0.2342895487998941,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010077627762776278,
      "loss": 3.2281,
      "step": 496170
    },
    {
      "epoch": 0.2343081920360523,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010077427742774277,
      "loss": 3.2301,
      "step": 496180
    },
    {
      "epoch": 0.2343268352722105,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010077227722772278,
      "loss": 3.2041,
      "step": 496190
    },
    {
      "epoch": 0.23434547850836865,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010077027702770277,
      "loss": 3.2784,
      "step": 496200
    },
    {
      "epoch": 0.23436412174452684,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010076827682768277,
      "loss": 3.1707,
      "step": 496210
    },
    {
      "epoch": 0.23438276498068503,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010076627662766278,
      "loss": 2.9362,
      "step": 496220
    },
    {
      "epoch": 0.23440140821684322,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010076427642764277,
      "loss": 3.2733,
      "step": 496230
    },
    {
      "epoch": 0.2344200514530014,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010076227622762277,
      "loss": 3.1842,
      "step": 496240
    },
    {
      "epoch": 0.23443869468915957,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010076027602760276,
      "loss": 3.208,
      "step": 496250
    },
    {
      "epoch": 0.23445733792531775,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010075827582758277,
      "loss": 3.2534,
      "step": 496260
    },
    {
      "epoch": 0.23447598116147594,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010075627562756276,
      "loss": 3.2182,
      "step": 496270
    },
    {
      "epoch": 0.23449462439763413,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010075427542754277,
      "loss": 3.2971,
      "step": 496280
    },
    {
      "epoch": 0.23451326763379232,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010075227522752275,
      "loss": 3.1855,
      "step": 496290
    },
    {
      "epoch": 0.23453191086995048,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010075027502750276,
      "loss": 3.1763,
      "step": 496300
    },
    {
      "epoch": 0.23455055410610867,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010074827482748274,
      "loss": 3.0723,
      "step": 496310
    },
    {
      "epoch": 0.23456919734226686,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010074627462746275,
      "loss": 3.3531,
      "step": 496320
    },
    {
      "epoch": 0.23458784057842505,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010074427442744275,
      "loss": 3.2266,
      "step": 496330
    },
    {
      "epoch": 0.23460648381458324,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010074227422742274,
      "loss": 3.238,
      "step": 496340
    },
    {
      "epoch": 0.2346251270507414,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010074027402740274,
      "loss": 3.253,
      "step": 496350
    },
    {
      "epoch": 0.2346437702868996,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010073827382738275,
      "loss": 3.1959,
      "step": 496360
    },
    {
      "epoch": 0.23466241352305778,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010073627362736273,
      "loss": 3.1628,
      "step": 496370
    },
    {
      "epoch": 0.23468105675921597,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010073427342734274,
      "loss": 3.2939,
      "step": 496380
    },
    {
      "epoch": 0.23469969999537416,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010073227322732272,
      "loss": 3.1344,
      "step": 496390
    },
    {
      "epoch": 0.23471834323153232,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010073027302730273,
      "loss": 3.1025,
      "step": 496400
    },
    {
      "epoch": 0.2347369864676905,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010072827282728274,
      "loss": 3.1897,
      "step": 496410
    },
    {
      "epoch": 0.2347556297038487,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010072627262726274,
      "loss": 3.2174,
      "step": 496420
    },
    {
      "epoch": 0.23477427294000688,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010072427242724274,
      "loss": 3.1109,
      "step": 496430
    },
    {
      "epoch": 0.23479291617616507,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010072227222722273,
      "loss": 3.2183,
      "step": 496440
    },
    {
      "epoch": 0.23481155941232326,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010072027202720274,
      "loss": 3.1414,
      "step": 496450
    },
    {
      "epoch": 0.23483020264848142,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010071827182718272,
      "loss": 3.2473,
      "step": 496460
    },
    {
      "epoch": 0.2348488458846396,
      "grad_norm": 1.25,
      "learning_rate": 0.00010071627162716273,
      "loss": 3.2145,
      "step": 496470
    },
    {
      "epoch": 0.2348674891207978,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010071427142714271,
      "loss": 3.2768,
      "step": 496480
    },
    {
      "epoch": 0.234886132356956,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010071227122712272,
      "loss": 3.1827,
      "step": 496490
    },
    {
      "epoch": 0.23490477559311418,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010071027102710271,
      "loss": 3.247,
      "step": 496500
    },
    {
      "epoch": 0.23492341882927234,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010070827082708272,
      "loss": 3.2842,
      "step": 496510
    },
    {
      "epoch": 0.23494206206543053,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001007062706270627,
      "loss": 3.0318,
      "step": 496520
    },
    {
      "epoch": 0.23496070530158872,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010070427042704271,
      "loss": 3.1251,
      "step": 496530
    },
    {
      "epoch": 0.2349793485377469,
      "grad_norm": 0.75,
      "learning_rate": 0.0001007022702270227,
      "loss": 3.0932,
      "step": 496540
    },
    {
      "epoch": 0.2349979917739051,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001007002700270027,
      "loss": 3.1996,
      "step": 496550
    },
    {
      "epoch": 0.23501663501006326,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010069826982698269,
      "loss": 3.3174,
      "step": 496560
    },
    {
      "epoch": 0.23503527824622145,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001006962696269627,
      "loss": 3.1239,
      "step": 496570
    },
    {
      "epoch": 0.23505392148237964,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010069426942694269,
      "loss": 3.1912,
      "step": 496580
    },
    {
      "epoch": 0.23507256471853782,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001006922692269227,
      "loss": 3.1788,
      "step": 496590
    },
    {
      "epoch": 0.235091207954696,
      "grad_norm": 0.9375,
      "learning_rate": 0.00010069026902690271,
      "loss": 3.1288,
      "step": 496600
    },
    {
      "epoch": 0.23510985119085417,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010068826882688269,
      "loss": 3.2826,
      "step": 496610
    },
    {
      "epoch": 0.23512849442701236,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001006862686268627,
      "loss": 3.1682,
      "step": 496620
    },
    {
      "epoch": 0.23514713766317055,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010068426842684268,
      "loss": 3.1729,
      "step": 496630
    },
    {
      "epoch": 0.23516578089932874,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010068226822682269,
      "loss": 3.1997,
      "step": 496640
    },
    {
      "epoch": 0.23518442413548693,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010068026802680269,
      "loss": 3.1315,
      "step": 496650
    },
    {
      "epoch": 0.23520306737164512,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010067826782678268,
      "loss": 3.2039,
      "step": 496660
    },
    {
      "epoch": 0.23522171060780328,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010067626762676268,
      "loss": 3.1482,
      "step": 496670
    },
    {
      "epoch": 0.23524035384396147,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010067426742674269,
      "loss": 3.1665,
      "step": 496680
    },
    {
      "epoch": 0.23525899708011966,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010067226722672267,
      "loss": 3.1632,
      "step": 496690
    },
    {
      "epoch": 0.23527764031627785,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010067026702670268,
      "loss": 3.1776,
      "step": 496700
    },
    {
      "epoch": 0.23529628355243604,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010066826682668266,
      "loss": 3.2067,
      "step": 496710
    },
    {
      "epoch": 0.2353149267885942,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010066626662666267,
      "loss": 3.2086,
      "step": 496720
    },
    {
      "epoch": 0.2353335700247524,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010066426642664267,
      "loss": 3.1209,
      "step": 496730
    },
    {
      "epoch": 0.23535221326091058,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010066226622662268,
      "loss": 3.2785,
      "step": 496740
    },
    {
      "epoch": 0.23537085649706876,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010066026602660266,
      "loss": 3.1251,
      "step": 496750
    },
    {
      "epoch": 0.23538949973322695,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010065826582658267,
      "loss": 3.2298,
      "step": 496760
    },
    {
      "epoch": 0.23540814296938511,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010065626562656265,
      "loss": 3.2096,
      "step": 496770
    },
    {
      "epoch": 0.2354267862055433,
      "grad_norm": 0.75,
      "learning_rate": 0.00010065426542654266,
      "loss": 3.2419,
      "step": 496780
    },
    {
      "epoch": 0.2354454294417015,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010065226522652264,
      "loss": 3.1498,
      "step": 496790
    },
    {
      "epoch": 0.23546407267785968,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010065026502650265,
      "loss": 3.2259,
      "step": 496800
    },
    {
      "epoch": 0.23548271591401787,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010064826482648266,
      "loss": 3.1745,
      "step": 496810
    },
    {
      "epoch": 0.23550135915017603,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010064626462646265,
      "loss": 3.2892,
      "step": 496820
    },
    {
      "epoch": 0.23552000238633422,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010064426442644266,
      "loss": 3.2001,
      "step": 496830
    },
    {
      "epoch": 0.2355386456224924,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010064226422642265,
      "loss": 3.2055,
      "step": 496840
    },
    {
      "epoch": 0.2355572888586506,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010064026402640265,
      "loss": 3.095,
      "step": 496850
    },
    {
      "epoch": 0.2355759320948088,
      "grad_norm": 0.75,
      "learning_rate": 0.00010063826382638264,
      "loss": 3.2649,
      "step": 496860
    },
    {
      "epoch": 0.23559457533096698,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010063626362636265,
      "loss": 3.1864,
      "step": 496870
    },
    {
      "epoch": 0.23561321856712514,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010063426342634263,
      "loss": 3.1642,
      "step": 496880
    },
    {
      "epoch": 0.23563186180328333,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010063226322632264,
      "loss": 3.1799,
      "step": 496890
    },
    {
      "epoch": 0.23565050503944152,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010063026302630263,
      "loss": 3.1314,
      "step": 496900
    },
    {
      "epoch": 0.2356691482755997,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010062826282628264,
      "loss": 3.2469,
      "step": 496910
    },
    {
      "epoch": 0.2356877915117579,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010062626262626262,
      "loss": 3.2125,
      "step": 496920
    },
    {
      "epoch": 0.23570643474791605,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010062426242624263,
      "loss": 3.2448,
      "step": 496930
    },
    {
      "epoch": 0.23572507798407424,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00010062226222622262,
      "loss": 3.2226,
      "step": 496940
    },
    {
      "epoch": 0.23574372122023243,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010062026202620262,
      "loss": 3.3067,
      "step": 496950
    },
    {
      "epoch": 0.23576236445639062,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010061826182618262,
      "loss": 3.1618,
      "step": 496960
    },
    {
      "epoch": 0.2357810076925488,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010061626162616263,
      "loss": 3.0524,
      "step": 496970
    },
    {
      "epoch": 0.23579965092870697,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010061426142614261,
      "loss": 3.2618,
      "step": 496980
    },
    {
      "epoch": 0.23581829416486516,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010061226122612262,
      "loss": 3.3213,
      "step": 496990
    },
    {
      "epoch": 0.23583693740102335,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010061026102610263,
      "loss": 3.0822,
      "step": 497000
    },
    {
      "epoch": 0.23585558063718154,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010060826082608261,
      "loss": 3.2446,
      "step": 497010
    },
    {
      "epoch": 0.23587422387333973,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00010060626062606262,
      "loss": 3.1474,
      "step": 497020
    },
    {
      "epoch": 0.2358928671094979,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001006042604260426,
      "loss": 3.3212,
      "step": 497030
    },
    {
      "epoch": 0.23591151034565608,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010060226022602261,
      "loss": 3.1981,
      "step": 497040
    },
    {
      "epoch": 0.23593015358181427,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010060026002600261,
      "loss": 3.3836,
      "step": 497050
    },
    {
      "epoch": 0.23594879681797246,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010059825982598262,
      "loss": 3.2441,
      "step": 497060
    },
    {
      "epoch": 0.23596744005413064,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001005962596259626,
      "loss": 3.1383,
      "step": 497070
    },
    {
      "epoch": 0.2359860832902888,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010059425942594261,
      "loss": 3.2287,
      "step": 497080
    },
    {
      "epoch": 0.236004726526447,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010059225922592259,
      "loss": 3.2548,
      "step": 497090
    },
    {
      "epoch": 0.23602336976260518,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001005902590259026,
      "loss": 3.1936,
      "step": 497100
    },
    {
      "epoch": 0.23604201299876337,
      "grad_norm": 0.75,
      "learning_rate": 0.00010058825882588258,
      "loss": 3.1712,
      "step": 497110
    },
    {
      "epoch": 0.23606065623492156,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010058625862586259,
      "loss": 3.2789,
      "step": 497120
    },
    {
      "epoch": 0.23607929947107975,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010058425842584259,
      "loss": 3.233,
      "step": 497130
    },
    {
      "epoch": 0.2360979427072379,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001005822582258226,
      "loss": 3.2099,
      "step": 497140
    },
    {
      "epoch": 0.2361165859433961,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010058025802580258,
      "loss": 3.1001,
      "step": 497150
    },
    {
      "epoch": 0.2361352291795543,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010057825782578259,
      "loss": 3.2064,
      "step": 497160
    },
    {
      "epoch": 0.23615387241571248,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010057625762576257,
      "loss": 3.1848,
      "step": 497170
    },
    {
      "epoch": 0.23617251565187067,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010057425742574258,
      "loss": 3.1546,
      "step": 497180
    },
    {
      "epoch": 0.23619115888802883,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010057225722572259,
      "loss": 3.2116,
      "step": 497190
    },
    {
      "epoch": 0.23620980212418702,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010057025702570257,
      "loss": 3.1438,
      "step": 497200
    },
    {
      "epoch": 0.2362284453603452,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010056825682568258,
      "loss": 3.2436,
      "step": 497210
    },
    {
      "epoch": 0.2362470885965034,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010056625662566257,
      "loss": 3.3136,
      "step": 497220
    },
    {
      "epoch": 0.23626573183266159,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010056425642564258,
      "loss": 3.2112,
      "step": 497230
    },
    {
      "epoch": 0.23628437506881975,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010056225622562256,
      "loss": 3.2824,
      "step": 497240
    },
    {
      "epoch": 0.23630301830497794,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010056025602560257,
      "loss": 3.2802,
      "step": 497250
    },
    {
      "epoch": 0.23632166154113612,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010055825582558256,
      "loss": 3.1823,
      "step": 497260
    },
    {
      "epoch": 0.2363403047772943,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010055625562556256,
      "loss": 3.2253,
      "step": 497270
    },
    {
      "epoch": 0.2363589480134525,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010055425542554256,
      "loss": 3.1993,
      "step": 497280
    },
    {
      "epoch": 0.23637759124961066,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010055225522552256,
      "loss": 3.1951,
      "step": 497290
    },
    {
      "epoch": 0.23639623448576885,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010055025502550255,
      "loss": 3.0941,
      "step": 497300
    },
    {
      "epoch": 0.23641487772192704,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010054825482548256,
      "loss": 3.2626,
      "step": 497310
    },
    {
      "epoch": 0.23643352095808523,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010054625462546254,
      "loss": 3.2406,
      "step": 497320
    },
    {
      "epoch": 0.23645216419424342,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00010054425442544255,
      "loss": 3.3432,
      "step": 497330
    },
    {
      "epoch": 0.2364708074304016,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010054225422542253,
      "loss": 3.3024,
      "step": 497340
    },
    {
      "epoch": 0.23648945066655977,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010054025402540254,
      "loss": 3.0643,
      "step": 497350
    },
    {
      "epoch": 0.23650809390271796,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010053825382538254,
      "loss": 3.2844,
      "step": 497360
    },
    {
      "epoch": 0.23652673713887615,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010053625362536255,
      "loss": 3.1694,
      "step": 497370
    },
    {
      "epoch": 0.23654538037503434,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010053425342534253,
      "loss": 3.2433,
      "step": 497380
    },
    {
      "epoch": 0.23656402361119253,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010053225322532254,
      "loss": 3.2213,
      "step": 497390
    },
    {
      "epoch": 0.2365826668473507,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010053025302530255,
      "loss": 3.1249,
      "step": 497400
    },
    {
      "epoch": 0.23660131008350888,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010052825282528253,
      "loss": 3.1324,
      "step": 497410
    },
    {
      "epoch": 0.23661995331966706,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010052625262526254,
      "loss": 3.2247,
      "step": 497420
    },
    {
      "epoch": 0.23663859655582525,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010052425242524252,
      "loss": 3.1484,
      "step": 497430
    },
    {
      "epoch": 0.23665723979198344,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010052225222522253,
      "loss": 3.2034,
      "step": 497440
    },
    {
      "epoch": 0.2366758830281416,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010052025202520253,
      "loss": 3.2058,
      "step": 497450
    },
    {
      "epoch": 0.2366945262642998,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010051825182518254,
      "loss": 3.2161,
      "step": 497460
    },
    {
      "epoch": 0.23671316950045798,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010051625162516252,
      "loss": 3.1465,
      "step": 497470
    },
    {
      "epoch": 0.23673181273661617,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010051425142514253,
      "loss": 3.2315,
      "step": 497480
    },
    {
      "epoch": 0.23675045597277436,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010051225122512251,
      "loss": 3.15,
      "step": 497490
    },
    {
      "epoch": 0.23676909920893252,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010051025102510252,
      "loss": 3.1076,
      "step": 497500
    },
    {
      "epoch": 0.2367877424450907,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001005082508250825,
      "loss": 3.1888,
      "step": 497510
    },
    {
      "epoch": 0.2368063856812489,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010050625062506251,
      "loss": 3.1167,
      "step": 497520
    },
    {
      "epoch": 0.2368250289174071,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001005042504250425,
      "loss": 3.1462,
      "step": 497530
    },
    {
      "epoch": 0.23684367215356528,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010050225022502251,
      "loss": 3.1423,
      "step": 497540
    },
    {
      "epoch": 0.23686231538972347,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001005002500250025,
      "loss": 3.2939,
      "step": 497550
    },
    {
      "epoch": 0.23688095862588163,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001004982498249825,
      "loss": 3.2435,
      "step": 497560
    },
    {
      "epoch": 0.23689960186203982,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010049624962496249,
      "loss": 3.2741,
      "step": 497570
    },
    {
      "epoch": 0.236918245098198,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001004942494249425,
      "loss": 3.2111,
      "step": 497580
    },
    {
      "epoch": 0.2369368883343562,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001004922492249225,
      "loss": 3.1917,
      "step": 497590
    },
    {
      "epoch": 0.23695553157051438,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001004902490249025,
      "loss": 3.1806,
      "step": 497600
    },
    {
      "epoch": 0.23697417480667254,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001004882488248825,
      "loss": 3.3158,
      "step": 497610
    },
    {
      "epoch": 0.23699281804283073,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010048624862486249,
      "loss": 3.1339,
      "step": 497620
    },
    {
      "epoch": 0.23701146127898892,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001004842484248425,
      "loss": 3.1055,
      "step": 497630
    },
    {
      "epoch": 0.2370301045151471,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010048224822482248,
      "loss": 3.1387,
      "step": 497640
    },
    {
      "epoch": 0.2370487477513053,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010048024802480249,
      "loss": 3.2062,
      "step": 497650
    },
    {
      "epoch": 0.23706739098746346,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010047824782478247,
      "loss": 3.3509,
      "step": 497660
    },
    {
      "epoch": 0.23708603422362165,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010047624762476248,
      "loss": 3.0576,
      "step": 497670
    },
    {
      "epoch": 0.23710467745977984,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010047424742474248,
      "loss": 3.2207,
      "step": 497680
    },
    {
      "epoch": 0.23712332069593803,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010047224722472249,
      "loss": 3.3249,
      "step": 497690
    },
    {
      "epoch": 0.23714196393209622,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010047024702470247,
      "loss": 3.0812,
      "step": 497700
    },
    {
      "epoch": 0.23716060716825438,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010046824682468248,
      "loss": 3.2962,
      "step": 497710
    },
    {
      "epoch": 0.23717925040441257,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010046624662466246,
      "loss": 3.2182,
      "step": 497720
    },
    {
      "epoch": 0.23719789364057076,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010046424642464247,
      "loss": 3.2551,
      "step": 497730
    },
    {
      "epoch": 0.23721653687672895,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010046224622462245,
      "loss": 3.128,
      "step": 497740
    },
    {
      "epoch": 0.23723518011288713,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010046024602460246,
      "loss": 3.136,
      "step": 497750
    },
    {
      "epoch": 0.2372538233490453,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010045824582458246,
      "loss": 3.2118,
      "step": 497760
    },
    {
      "epoch": 0.23727246658520348,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010045624562456247,
      "loss": 3.1752,
      "step": 497770
    },
    {
      "epoch": 0.23729110982136167,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010045424542454245,
      "loss": 3.1602,
      "step": 497780
    },
    {
      "epoch": 0.23730975305751986,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010045224522452246,
      "loss": 3.387,
      "step": 497790
    },
    {
      "epoch": 0.23732839629367805,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010045024502450247,
      "loss": 3.282,
      "step": 497800
    },
    {
      "epoch": 0.23734703952983624,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010044824482448245,
      "loss": 3.1554,
      "step": 497810
    },
    {
      "epoch": 0.2373656827659944,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010044624462446246,
      "loss": 3.0971,
      "step": 497820
    },
    {
      "epoch": 0.2373843260021526,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010044424442444244,
      "loss": 3.2471,
      "step": 497830
    },
    {
      "epoch": 0.23740296923831078,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010044224422442245,
      "loss": 3.2596,
      "step": 497840
    },
    {
      "epoch": 0.23742161247446897,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010044024402440244,
      "loss": 2.9121,
      "step": 497850
    },
    {
      "epoch": 0.23744025571062716,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010043824382438245,
      "loss": 3.2059,
      "step": 497860
    },
    {
      "epoch": 0.23745889894678532,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010043624362436244,
      "loss": 3.1374,
      "step": 497870
    },
    {
      "epoch": 0.2374775421829435,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010043424342434244,
      "loss": 3.1046,
      "step": 497880
    },
    {
      "epoch": 0.2374961854191017,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010043224322432243,
      "loss": 3.2114,
      "step": 497890
    },
    {
      "epoch": 0.23751482865525989,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010043024302430244,
      "loss": 3.2766,
      "step": 497900
    },
    {
      "epoch": 0.23753347189141807,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010042824282428243,
      "loss": 3.2027,
      "step": 497910
    },
    {
      "epoch": 0.23755211512757624,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010042624262426244,
      "loss": 3.1471,
      "step": 497920
    },
    {
      "epoch": 0.23757075836373442,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010042424242424242,
      "loss": 3.2965,
      "step": 497930
    },
    {
      "epoch": 0.2375894015998926,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010042224222422243,
      "loss": 3.1422,
      "step": 497940
    },
    {
      "epoch": 0.2376080448360508,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010042024202420241,
      "loss": 3.1497,
      "step": 497950
    },
    {
      "epoch": 0.237626688072209,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010041824182418242,
      "loss": 3.1966,
      "step": 497960
    },
    {
      "epoch": 0.23764533130836715,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001004162416241624,
      "loss": 3.1708,
      "step": 497970
    },
    {
      "epoch": 0.23766397454452534,
      "grad_norm": 0.890625,
      "learning_rate": 0.00010041424142414241,
      "loss": 3.1125,
      "step": 497980
    },
    {
      "epoch": 0.23768261778068353,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010041224122412242,
      "loss": 3.2827,
      "step": 497990
    },
    {
      "epoch": 0.23770126101684172,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010041024102410242,
      "loss": 3.301,
      "step": 498000
    },
    {
      "epoch": 0.2377199042529999,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010040824082408243,
      "loss": 3.2773,
      "step": 498010
    },
    {
      "epoch": 0.2377385474891581,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010040624062406241,
      "loss": 3.1987,
      "step": 498020
    },
    {
      "epoch": 0.23775719072531626,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010040424042404242,
      "loss": 3.1984,
      "step": 498030
    },
    {
      "epoch": 0.23777583396147445,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001004022402240224,
      "loss": 3.233,
      "step": 498040
    },
    {
      "epoch": 0.23779447719763264,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010040024002400241,
      "loss": 3.318,
      "step": 498050
    },
    {
      "epoch": 0.23781312043379083,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010039823982398239,
      "loss": 3.1866,
      "step": 498060
    },
    {
      "epoch": 0.23783176366994901,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001003962396239624,
      "loss": 3.2727,
      "step": 498070
    },
    {
      "epoch": 0.23785040690610718,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001003942394239424,
      "loss": 3.1998,
      "step": 498080
    },
    {
      "epoch": 0.23786905014226536,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001003922392239224,
      "loss": 3.2564,
      "step": 498090
    },
    {
      "epoch": 0.23788769337842355,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010039023902390239,
      "loss": 3.2802,
      "step": 498100
    },
    {
      "epoch": 0.23790633661458174,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001003882388238824,
      "loss": 3.169,
      "step": 498110
    },
    {
      "epoch": 0.23792497985073993,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010038623862386238,
      "loss": 3.2326,
      "step": 498120
    },
    {
      "epoch": 0.2379436230868981,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010038423842384239,
      "loss": 3.2509,
      "step": 498130
    },
    {
      "epoch": 0.23796226632305628,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010038223822382238,
      "loss": 3.3834,
      "step": 498140
    },
    {
      "epoch": 0.23798090955921447,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010038023802380238,
      "loss": 3.214,
      "step": 498150
    },
    {
      "epoch": 0.23799955279537266,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010037823782378238,
      "loss": 3.0794,
      "step": 498160
    },
    {
      "epoch": 0.23801819603153085,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010037623762376238,
      "loss": 3.2221,
      "step": 498170
    },
    {
      "epoch": 0.238036839267689,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001003742374237424,
      "loss": 3.2492,
      "step": 498180
    },
    {
      "epoch": 0.2380554825038472,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010037223722372238,
      "loss": 3.1623,
      "step": 498190
    },
    {
      "epoch": 0.2380741257400054,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010037023702370239,
      "loss": 3.2996,
      "step": 498200
    },
    {
      "epoch": 0.23809276897616358,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010036823682368237,
      "loss": 3.2756,
      "step": 498210
    },
    {
      "epoch": 0.23811141221232177,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010036623662366238,
      "loss": 3.1453,
      "step": 498220
    },
    {
      "epoch": 0.23813005544847995,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010036423642364237,
      "loss": 3.2702,
      "step": 498230
    },
    {
      "epoch": 0.23814869868463812,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010036223622362238,
      "loss": 3.1623,
      "step": 498240
    },
    {
      "epoch": 0.2381673419207963,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010036023602360236,
      "loss": 3.1562,
      "step": 498250
    },
    {
      "epoch": 0.2381859851569545,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010035823582358237,
      "loss": 3.1404,
      "step": 498260
    },
    {
      "epoch": 0.23820462839311268,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010035623562356235,
      "loss": 3.2944,
      "step": 498270
    },
    {
      "epoch": 0.23822327162927087,
      "grad_norm": 0.625,
      "learning_rate": 0.00010035423542354236,
      "loss": 3.1627,
      "step": 498280
    },
    {
      "epoch": 0.23824191486542903,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010035223522352235,
      "loss": 3.1894,
      "step": 498290
    },
    {
      "epoch": 0.23826055810158722,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010035023502350235,
      "loss": 3.1712,
      "step": 498300
    },
    {
      "epoch": 0.2382792013377454,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010034823482348235,
      "loss": 3.1652,
      "step": 498310
    },
    {
      "epoch": 0.2382978445739036,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010034623462346236,
      "loss": 3.1746,
      "step": 498320
    },
    {
      "epoch": 0.2383164878100618,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010034423442344234,
      "loss": 3.2077,
      "step": 498330
    },
    {
      "epoch": 0.23833513104621995,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010034223422342235,
      "loss": 3.1062,
      "step": 498340
    },
    {
      "epoch": 0.23835377428237814,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010034023402340233,
      "loss": 3.1806,
      "step": 498350
    },
    {
      "epoch": 0.23837241751853633,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010033823382338234,
      "loss": 3.1944,
      "step": 498360
    },
    {
      "epoch": 0.23839106075469452,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010033623362336232,
      "loss": 3.1881,
      "step": 498370
    },
    {
      "epoch": 0.2384097039908527,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010033423342334233,
      "loss": 3.21,
      "step": 498380
    },
    {
      "epoch": 0.23842834722701087,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010033223322332234,
      "loss": 3.1414,
      "step": 498390
    },
    {
      "epoch": 0.23844699046316906,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010033023302330234,
      "loss": 3.2229,
      "step": 498400
    },
    {
      "epoch": 0.23846563369932725,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010032823282328235,
      "loss": 3.1571,
      "step": 498410
    },
    {
      "epoch": 0.23848427693548543,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010032623262326233,
      "loss": 3.26,
      "step": 498420
    },
    {
      "epoch": 0.23850292017164362,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010032423242324234,
      "loss": 2.9131,
      "step": 498430
    },
    {
      "epoch": 0.23852156340780178,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010032223222322232,
      "loss": 3.2737,
      "step": 498440
    },
    {
      "epoch": 0.23854020664395997,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010032023202320233,
      "loss": 3.1102,
      "step": 498450
    },
    {
      "epoch": 0.23855884988011816,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010031823182318232,
      "loss": 3.1375,
      "step": 498460
    },
    {
      "epoch": 0.23857749311627635,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010031623162316232,
      "loss": 3.2521,
      "step": 498470
    },
    {
      "epoch": 0.23859613635243454,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010031423142314232,
      "loss": 3.2457,
      "step": 498480
    },
    {
      "epoch": 0.23861477958859273,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010031223122312233,
      "loss": 3.2091,
      "step": 498490
    },
    {
      "epoch": 0.2386334228247509,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010031023102310231,
      "loss": 3.1853,
      "step": 498500
    },
    {
      "epoch": 0.23865206606090908,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010030823082308232,
      "loss": 3.2768,
      "step": 498510
    },
    {
      "epoch": 0.23867070929706727,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001003062306230623,
      "loss": 3.2133,
      "step": 498520
    },
    {
      "epoch": 0.23868935253322546,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00010030423042304231,
      "loss": 3.2477,
      "step": 498530
    },
    {
      "epoch": 0.23870799576938365,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001003022302230223,
      "loss": 3.1222,
      "step": 498540
    },
    {
      "epoch": 0.2387266390055418,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010030023002300231,
      "loss": 3.1905,
      "step": 498550
    },
    {
      "epoch": 0.2387452822417,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001002982298229823,
      "loss": 3.2629,
      "step": 498560
    },
    {
      "epoch": 0.23876392547785819,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001002962296229623,
      "loss": 3.1263,
      "step": 498570
    },
    {
      "epoch": 0.23878256871401637,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010029422942294231,
      "loss": 3.2149,
      "step": 498580
    },
    {
      "epoch": 0.23880121195017456,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001002922292229223,
      "loss": 3.1967,
      "step": 498590
    },
    {
      "epoch": 0.23881985518633272,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001002902290229023,
      "loss": 3.3218,
      "step": 498600
    },
    {
      "epoch": 0.2388384984224909,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010028822882288229,
      "loss": 3.2133,
      "step": 498610
    },
    {
      "epoch": 0.2388571416586491,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001002862286228623,
      "loss": 3.2008,
      "step": 498620
    },
    {
      "epoch": 0.2388757848948073,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010028422842284229,
      "loss": 3.2105,
      "step": 498630
    },
    {
      "epoch": 0.23889442813096548,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001002822282228223,
      "loss": 3.1316,
      "step": 498640
    },
    {
      "epoch": 0.23891307136712364,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010028022802280228,
      "loss": 3.2357,
      "step": 498650
    },
    {
      "epoch": 0.23893171460328183,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010027822782278229,
      "loss": 3.2308,
      "step": 498660
    },
    {
      "epoch": 0.23895035783944002,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010027622762276227,
      "loss": 3.0595,
      "step": 498670
    },
    {
      "epoch": 0.2389690010755982,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010027422742274228,
      "loss": 3.4178,
      "step": 498680
    },
    {
      "epoch": 0.2389876443117564,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010027222722272226,
      "loss": 3.2374,
      "step": 498690
    },
    {
      "epoch": 0.2390062875479146,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010027022702270227,
      "loss": 3.2497,
      "step": 498700
    },
    {
      "epoch": 0.23902493078407275,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010026822682268227,
      "loss": 3.1143,
      "step": 498710
    },
    {
      "epoch": 0.23904357402023094,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010026622662266228,
      "loss": 3.198,
      "step": 498720
    },
    {
      "epoch": 0.23906221725638913,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010026422642264226,
      "loss": 3.1233,
      "step": 498730
    },
    {
      "epoch": 0.23908086049254731,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010026222622262227,
      "loss": 3.2078,
      "step": 498740
    },
    {
      "epoch": 0.2390995037287055,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00010026022602260225,
      "loss": 3.3048,
      "step": 498750
    },
    {
      "epoch": 0.23911814696486366,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010025822582258226,
      "loss": 3.189,
      "step": 498760
    },
    {
      "epoch": 0.23913679020102185,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010025622562256227,
      "loss": 3.1086,
      "step": 498770
    },
    {
      "epoch": 0.23915543343718004,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010025422542254225,
      "loss": 3.1597,
      "step": 498780
    },
    {
      "epoch": 0.23917407667333823,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010025222522252226,
      "loss": 3.1653,
      "step": 498790
    },
    {
      "epoch": 0.23919271990949642,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010025022502250226,
      "loss": 3.1542,
      "step": 498800
    },
    {
      "epoch": 0.23921136314565458,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010024822482248227,
      "loss": 3.1891,
      "step": 498810
    },
    {
      "epoch": 0.23923000638181277,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010024622462246225,
      "loss": 3.1961,
      "step": 498820
    },
    {
      "epoch": 0.23924864961797096,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010024422442244226,
      "loss": 3.1126,
      "step": 498830
    },
    {
      "epoch": 0.23926729285412915,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010024222422242224,
      "loss": 3.0307,
      "step": 498840
    },
    {
      "epoch": 0.23928593609028734,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010024022402240225,
      "loss": 3.3283,
      "step": 498850
    },
    {
      "epoch": 0.2393045793264455,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010023822382238224,
      "loss": 3.1225,
      "step": 498860
    },
    {
      "epoch": 0.2393232225626037,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010023622362236225,
      "loss": 3.0646,
      "step": 498870
    },
    {
      "epoch": 0.23934186579876188,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010023422342234223,
      "loss": 3.1507,
      "step": 498880
    },
    {
      "epoch": 0.23936050903492007,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010023222322232224,
      "loss": 3.1022,
      "step": 498890
    },
    {
      "epoch": 0.23937915227107825,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010023022302230223,
      "loss": 3.2556,
      "step": 498900
    },
    {
      "epoch": 0.23939779550723644,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010022822282228224,
      "loss": 3.3042,
      "step": 498910
    },
    {
      "epoch": 0.2394164387433946,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010022622262226222,
      "loss": 3.198,
      "step": 498920
    },
    {
      "epoch": 0.2394350819795528,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010022422242224223,
      "loss": 3.1517,
      "step": 498930
    },
    {
      "epoch": 0.23945372521571098,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010022222222222222,
      "loss": 3.1462,
      "step": 498940
    },
    {
      "epoch": 0.23947236845186917,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010022022202220223,
      "loss": 3.131,
      "step": 498950
    },
    {
      "epoch": 0.23949101168802736,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010021822182218221,
      "loss": 3.2462,
      "step": 498960
    },
    {
      "epoch": 0.23950965492418552,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00010021622162216222,
      "loss": 3.1469,
      "step": 498970
    },
    {
      "epoch": 0.2395282981603437,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010021422142214223,
      "loss": 3.0929,
      "step": 498980
    },
    {
      "epoch": 0.2395469413965019,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010021222122212221,
      "loss": 3.25,
      "step": 498990
    },
    {
      "epoch": 0.2395655846326601,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010021022102210222,
      "loss": 3.1427,
      "step": 499000
    },
    {
      "epoch": 0.23958422786881828,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001002082208220822,
      "loss": 3.2952,
      "step": 499010
    },
    {
      "epoch": 0.23960287110497644,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010020622062206221,
      "loss": 3.2784,
      "step": 499020
    },
    {
      "epoch": 0.23962151434113463,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010020422042204221,
      "loss": 3.1016,
      "step": 499030
    },
    {
      "epoch": 0.23964015757729282,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010020222022202222,
      "loss": 3.1767,
      "step": 499040
    },
    {
      "epoch": 0.239658800813451,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001002002200220022,
      "loss": 3.161,
      "step": 499050
    },
    {
      "epoch": 0.2396774440496092,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010019821982198221,
      "loss": 3.1929,
      "step": 499060
    },
    {
      "epoch": 0.23969608728576736,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010019621962196219,
      "loss": 3.2743,
      "step": 499070
    },
    {
      "epoch": 0.23971473052192555,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001001942194219422,
      "loss": 3.0661,
      "step": 499080
    },
    {
      "epoch": 0.23973337375808373,
      "grad_norm": 0.75,
      "learning_rate": 0.0001001922192219222,
      "loss": 3.2315,
      "step": 499090
    },
    {
      "epoch": 0.23975201699424192,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010019021902190219,
      "loss": 3.2057,
      "step": 499100
    },
    {
      "epoch": 0.2397706602304001,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010018821882188219,
      "loss": 3.2397,
      "step": 499110
    },
    {
      "epoch": 0.2397893034665583,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001001862186218622,
      "loss": 3.0502,
      "step": 499120
    },
    {
      "epoch": 0.23980794670271646,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010018421842184218,
      "loss": 3.2659,
      "step": 499130
    },
    {
      "epoch": 0.23982658993887465,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010018221822182219,
      "loss": 3.1619,
      "step": 499140
    },
    {
      "epoch": 0.23984523317503284,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010018021802180217,
      "loss": 3.2741,
      "step": 499150
    },
    {
      "epoch": 0.23986387641119103,
      "grad_norm": 0.75,
      "learning_rate": 0.00010017821782178218,
      "loss": 3.2482,
      "step": 499160
    },
    {
      "epoch": 0.23988251964734922,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010017621762176219,
      "loss": 3.1393,
      "step": 499170
    },
    {
      "epoch": 0.23990116288350738,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010017421742174218,
      "loss": 3.2185,
      "step": 499180
    },
    {
      "epoch": 0.23991980611966557,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010017221722172219,
      "loss": 3.1882,
      "step": 499190
    },
    {
      "epoch": 0.23993844935582376,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010017021702170218,
      "loss": 3.1495,
      "step": 499200
    },
    {
      "epoch": 0.23995709259198195,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010016821682168218,
      "loss": 3.261,
      "step": 499210
    },
    {
      "epoch": 0.23997573582814014,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010016621662166217,
      "loss": 3.3579,
      "step": 499220
    },
    {
      "epoch": 0.2399943790642983,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010016421642164218,
      "loss": 3.1955,
      "step": 499230
    },
    {
      "epoch": 0.24001302230045649,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010016221622162216,
      "loss": 3.1676,
      "step": 499240
    },
    {
      "epoch": 0.24003166553661467,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010016021602160217,
      "loss": 3.1465,
      "step": 499250
    },
    {
      "epoch": 0.24005030877277286,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010015821582158216,
      "loss": 3.3018,
      "step": 499260
    },
    {
      "epoch": 0.24006895200893105,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010015621562156217,
      "loss": 3.2062,
      "step": 499270
    },
    {
      "epoch": 0.2400875952450892,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010015421542154215,
      "loss": 3.1847,
      "step": 499280
    },
    {
      "epoch": 0.2401062384812474,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010015221522152216,
      "loss": 3.2112,
      "step": 499290
    },
    {
      "epoch": 0.2401248817174056,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010015021502150214,
      "loss": 3.2976,
      "step": 499300
    },
    {
      "epoch": 0.24014352495356378,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010014821482148215,
      "loss": 3.3123,
      "step": 499310
    },
    {
      "epoch": 0.24016216818972197,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010014621462146214,
      "loss": 3.1125,
      "step": 499320
    },
    {
      "epoch": 0.24018081142588013,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00010014421442144214,
      "loss": 3.1586,
      "step": 499330
    },
    {
      "epoch": 0.24019945466203832,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010014221422142214,
      "loss": 3.2184,
      "step": 499340
    },
    {
      "epoch": 0.2402180978981965,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010014021402140215,
      "loss": 3.22,
      "step": 499350
    },
    {
      "epoch": 0.2402367411343547,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010013821382138213,
      "loss": 3.308,
      "step": 499360
    },
    {
      "epoch": 0.2402553843705129,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010013621362136214,
      "loss": 3.2133,
      "step": 499370
    },
    {
      "epoch": 0.24027402760667108,
      "grad_norm": 0.75,
      "learning_rate": 0.00010013421342134215,
      "loss": 3.1179,
      "step": 499380
    },
    {
      "epoch": 0.24029267084282924,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010013221322132213,
      "loss": 3.1999,
      "step": 499390
    },
    {
      "epoch": 0.24031131407898743,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010013021302130214,
      "loss": 3.2449,
      "step": 499400
    },
    {
      "epoch": 0.24032995731514561,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010012821282128214,
      "loss": 3.2688,
      "step": 499410
    },
    {
      "epoch": 0.2403486005513038,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010012621262126213,
      "loss": 3.1636,
      "step": 499420
    },
    {
      "epoch": 0.240367243787462,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010012421242124213,
      "loss": 3.2198,
      "step": 499430
    },
    {
      "epoch": 0.24038588702362015,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010012221222122214,
      "loss": 3.1827,
      "step": 499440
    },
    {
      "epoch": 0.24040453025977834,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010012021202120212,
      "loss": 3.2846,
      "step": 499450
    },
    {
      "epoch": 0.24042317349593653,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010011821182118213,
      "loss": 3.2504,
      "step": 499460
    },
    {
      "epoch": 0.24044181673209472,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010011621162116211,
      "loss": 3.218,
      "step": 499470
    },
    {
      "epoch": 0.2404604599682529,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010011421142114212,
      "loss": 3.0691,
      "step": 499480
    },
    {
      "epoch": 0.24047910320441107,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010011221122112212,
      "loss": 3.2708,
      "step": 499490
    },
    {
      "epoch": 0.24049774644056926,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010011021102110212,
      "loss": 3.3395,
      "step": 499500
    },
    {
      "epoch": 0.24051638967672745,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001001082108210821,
      "loss": 3.1068,
      "step": 499510
    },
    {
      "epoch": 0.24053503291288564,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010010621062106212,
      "loss": 3.1078,
      "step": 499520
    },
    {
      "epoch": 0.24055367614904383,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001001042104210421,
      "loss": 3.2743,
      "step": 499530
    },
    {
      "epoch": 0.240572319385202,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001001022102210221,
      "loss": 3.2478,
      "step": 499540
    },
    {
      "epoch": 0.24059096262136018,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010010021002100209,
      "loss": 3.2177,
      "step": 499550
    },
    {
      "epoch": 0.24060960585751837,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001000982098209821,
      "loss": 3.2089,
      "step": 499560
    },
    {
      "epoch": 0.24062824909367656,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00010009620962096211,
      "loss": 3.1327,
      "step": 499570
    },
    {
      "epoch": 0.24064689232983474,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001000942094209421,
      "loss": 3.1014,
      "step": 499580
    },
    {
      "epoch": 0.24066553556599293,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010009220922092211,
      "loss": 3.028,
      "step": 499590
    },
    {
      "epoch": 0.2406841788021511,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001000902090209021,
      "loss": 3.0519,
      "step": 499600
    },
    {
      "epoch": 0.24070282203830928,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001000882088208821,
      "loss": 3.2296,
      "step": 499610
    },
    {
      "epoch": 0.24072146527446747,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010008620862086208,
      "loss": 3.1725,
      "step": 499620
    },
    {
      "epoch": 0.24074010851062566,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001000842084208421,
      "loss": 3.2513,
      "step": 499630
    },
    {
      "epoch": 0.24075875174678385,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00010008220822082208,
      "loss": 3.1565,
      "step": 499640
    },
    {
      "epoch": 0.240777394982942,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010008020802080209,
      "loss": 3.2042,
      "step": 499650
    },
    {
      "epoch": 0.2407960382191002,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010007820782078208,
      "loss": 3.2365,
      "step": 499660
    },
    {
      "epoch": 0.2408146814552584,
      "grad_norm": 0.875,
      "learning_rate": 0.00010007620762076209,
      "loss": 3.1969,
      "step": 499670
    },
    {
      "epoch": 0.24083332469141658,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00010007420742074207,
      "loss": 3.181,
      "step": 499680
    },
    {
      "epoch": 0.24085196792757477,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010007220722072208,
      "loss": 3.2971,
      "step": 499690
    },
    {
      "epoch": 0.24087061116373293,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010007020702070206,
      "loss": 3.2605,
      "step": 499700
    },
    {
      "epoch": 0.24088925439989112,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010006820682068207,
      "loss": 3.3901,
      "step": 499710
    },
    {
      "epoch": 0.2409078976360493,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010006620662066207,
      "loss": 3.2095,
      "step": 499720
    },
    {
      "epoch": 0.2409265408722075,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010006420642064208,
      "loss": 3.2222,
      "step": 499730
    },
    {
      "epoch": 0.24094518410836568,
      "grad_norm": 0.78125,
      "learning_rate": 0.00010006220622062206,
      "loss": 3.3464,
      "step": 499740
    },
    {
      "epoch": 0.24096382734452385,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010006020602060207,
      "loss": 3.2591,
      "step": 499750
    },
    {
      "epoch": 0.24098247058068203,
      "grad_norm": 0.84375,
      "learning_rate": 0.00010005820582058208,
      "loss": 3.1111,
      "step": 499760
    },
    {
      "epoch": 0.24100111381684022,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010005620562056206,
      "loss": 3.2832,
      "step": 499770
    },
    {
      "epoch": 0.2410197570529984,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010005420542054207,
      "loss": 3.1941,
      "step": 499780
    },
    {
      "epoch": 0.2410384002891566,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00010005220522052205,
      "loss": 3.3439,
      "step": 499790
    },
    {
      "epoch": 0.2410570435253148,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010005020502050206,
      "loss": 3.1771,
      "step": 499800
    },
    {
      "epoch": 0.24107568676147295,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010004820482048206,
      "loss": 3.1981,
      "step": 499810
    },
    {
      "epoch": 0.24109432999763114,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010004620462046206,
      "loss": 3.0773,
      "step": 499820
    },
    {
      "epoch": 0.24111297323378933,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00010004420442044205,
      "loss": 3.1427,
      "step": 499830
    },
    {
      "epoch": 0.24113161646994752,
      "grad_norm": 0.625,
      "learning_rate": 0.00010004220422042206,
      "loss": 3.1773,
      "step": 499840
    },
    {
      "epoch": 0.2411502597061057,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010004020402040204,
      "loss": 3.3177,
      "step": 499850
    },
    {
      "epoch": 0.24116890294226387,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010003820382038205,
      "loss": 3.2122,
      "step": 499860
    },
    {
      "epoch": 0.24118754617842206,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010003620362036203,
      "loss": 3.2424,
      "step": 499870
    },
    {
      "epoch": 0.24120618941458025,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00010003420342034204,
      "loss": 3.1882,
      "step": 499880
    },
    {
      "epoch": 0.24122483265073844,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010003220322032203,
      "loss": 3.4052,
      "step": 499890
    },
    {
      "epoch": 0.24124347588689662,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010003020302030204,
      "loss": 3.1239,
      "step": 499900
    },
    {
      "epoch": 0.24126211912305479,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00010002820282028202,
      "loss": 3.1903,
      "step": 499910
    },
    {
      "epoch": 0.24128076235921297,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010002620262026203,
      "loss": 3.2084,
      "step": 499920
    },
    {
      "epoch": 0.24129940559537116,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010002420242024202,
      "loss": 3.2433,
      "step": 499930
    },
    {
      "epoch": 0.24131804883152935,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010002220222022203,
      "loss": 3.1791,
      "step": 499940
    },
    {
      "epoch": 0.24133669206768754,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00010002020202020201,
      "loss": 3.2119,
      "step": 499950
    },
    {
      "epoch": 0.2413553353038457,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010001820182018202,
      "loss": 3.1365,
      "step": 499960
    },
    {
      "epoch": 0.2413739785400039,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010001620162016203,
      "loss": 3.2433,
      "step": 499970
    },
    {
      "epoch": 0.24139262177616208,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010001420142014202,
      "loss": 3.21,
      "step": 499980
    },
    {
      "epoch": 0.24141126501232027,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010001220122012203,
      "loss": 3.341,
      "step": 499990
    },
    {
      "epoch": 0.24142990824847846,
      "grad_norm": 0.6875,
      "learning_rate": 0.00010001020102010201,
      "loss": 3.1637,
      "step": 500000
    },
    {
      "epoch": 0.24144855148463662,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010000820082008202,
      "loss": 3.1991,
      "step": 500010
    },
    {
      "epoch": 0.2414671947207948,
      "grad_norm": 0.7890625,
      "learning_rate": 0.000100006200620062,
      "loss": 3.2598,
      "step": 500020
    },
    {
      "epoch": 0.241485837956953,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00010000420042004201,
      "loss": 3.232,
      "step": 500030
    },
    {
      "epoch": 0.2415044811931112,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010000220022002201,
      "loss": 3.2345,
      "step": 500040
    },
    {
      "epoch": 0.24152312442926938,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000100000200020002,
      "loss": 3.2401,
      "step": 500050
    },
    {
      "epoch": 0.24154176766542756,
      "grad_norm": 0.73046875,
      "learning_rate": 9.999819981998201e-05,
      "loss": 3.1491,
      "step": 500060
    },
    {
      "epoch": 0.24156041090158573,
      "grad_norm": 0.71875,
      "learning_rate": 9.999619961996201e-05,
      "loss": 3.2268,
      "step": 500070
    },
    {
      "epoch": 0.24157905413774391,
      "grad_norm": 0.71484375,
      "learning_rate": 9.9994199419942e-05,
      "loss": 3.2242,
      "step": 500080
    },
    {
      "epoch": 0.2415976973739021,
      "grad_norm": 0.7421875,
      "learning_rate": 9.9992199219922e-05,
      "loss": 3.1363,
      "step": 500090
    },
    {
      "epoch": 0.2416163406100603,
      "grad_norm": 0.8046875,
      "learning_rate": 9.9990199019902e-05,
      "loss": 3.2426,
      "step": 500100
    },
    {
      "epoch": 0.24163498384621848,
      "grad_norm": 0.8203125,
      "learning_rate": 9.998819881988199e-05,
      "loss": 3.2537,
      "step": 500110
    },
    {
      "epoch": 0.24165362708237664,
      "grad_norm": 0.671875,
      "learning_rate": 9.998619861986199e-05,
      "loss": 3.2015,
      "step": 500120
    },
    {
      "epoch": 0.24167227031853483,
      "grad_norm": 0.76171875,
      "learning_rate": 9.9984198419842e-05,
      "loss": 3.1481,
      "step": 500130
    },
    {
      "epoch": 0.24169091355469302,
      "grad_norm": 0.79296875,
      "learning_rate": 9.998219821982199e-05,
      "loss": 3.1833,
      "step": 500140
    },
    {
      "epoch": 0.2417095567908512,
      "grad_norm": 0.671875,
      "learning_rate": 9.998019801980199e-05,
      "loss": 3.1205,
      "step": 500150
    },
    {
      "epoch": 0.2417282000270094,
      "grad_norm": 0.7734375,
      "learning_rate": 9.997819781978198e-05,
      "loss": 3.3288,
      "step": 500160
    },
    {
      "epoch": 0.24174684326316756,
      "grad_norm": 0.87109375,
      "learning_rate": 9.997619761976198e-05,
      "loss": 3.1721,
      "step": 500170
    },
    {
      "epoch": 0.24176548649932575,
      "grad_norm": 0.83203125,
      "learning_rate": 9.997419741974197e-05,
      "loss": 3.1509,
      "step": 500180
    },
    {
      "epoch": 0.24178412973548394,
      "grad_norm": 0.58984375,
      "learning_rate": 9.997219721972197e-05,
      "loss": 3.2007,
      "step": 500190
    },
    {
      "epoch": 0.24180277297164213,
      "grad_norm": 0.671875,
      "learning_rate": 9.997019701970198e-05,
      "loss": 3.2501,
      "step": 500200
    },
    {
      "epoch": 0.24182141620780032,
      "grad_norm": 0.74609375,
      "learning_rate": 9.996819681968197e-05,
      "loss": 3.2813,
      "step": 500210
    },
    {
      "epoch": 0.24184005944395848,
      "grad_norm": 0.70703125,
      "learning_rate": 9.996619661966197e-05,
      "loss": 3.2697,
      "step": 500220
    },
    {
      "epoch": 0.24185870268011667,
      "grad_norm": 0.71875,
      "learning_rate": 9.996419641964197e-05,
      "loss": 3.0533,
      "step": 500230
    },
    {
      "epoch": 0.24187734591627486,
      "grad_norm": 0.796875,
      "learning_rate": 9.996219621962196e-05,
      "loss": 3.1009,
      "step": 500240
    },
    {
      "epoch": 0.24189598915243304,
      "grad_norm": 0.6796875,
      "learning_rate": 9.996019601960197e-05,
      "loss": 3.2152,
      "step": 500250
    },
    {
      "epoch": 0.24191463238859123,
      "grad_norm": 0.65234375,
      "learning_rate": 9.995819581958197e-05,
      "loss": 3.2153,
      "step": 500260
    },
    {
      "epoch": 0.24193327562474942,
      "grad_norm": 0.80078125,
      "learning_rate": 9.995619561956196e-05,
      "loss": 3.1101,
      "step": 500270
    },
    {
      "epoch": 0.24195191886090758,
      "grad_norm": 0.7109375,
      "learning_rate": 9.995419541954196e-05,
      "loss": 3.1855,
      "step": 500280
    },
    {
      "epoch": 0.24197056209706577,
      "grad_norm": 0.70703125,
      "learning_rate": 9.995219521952197e-05,
      "loss": 3.2098,
      "step": 500290
    },
    {
      "epoch": 0.24198920533322396,
      "grad_norm": 0.73046875,
      "learning_rate": 9.995019501950196e-05,
      "loss": 3.2089,
      "step": 500300
    },
    {
      "epoch": 0.24200784856938215,
      "grad_norm": 0.73046875,
      "learning_rate": 9.994819481948196e-05,
      "loss": 3.2634,
      "step": 500310
    },
    {
      "epoch": 0.24202649180554034,
      "grad_norm": 0.72265625,
      "learning_rate": 9.994619461946195e-05,
      "loss": 3.2562,
      "step": 500320
    },
    {
      "epoch": 0.2420451350416985,
      "grad_norm": 0.7265625,
      "learning_rate": 9.994419441944195e-05,
      "loss": 3.1485,
      "step": 500330
    },
    {
      "epoch": 0.2420637782778567,
      "grad_norm": 0.703125,
      "learning_rate": 9.994219421942194e-05,
      "loss": 3.2081,
      "step": 500340
    },
    {
      "epoch": 0.24208242151401488,
      "grad_norm": 0.8046875,
      "learning_rate": 9.994019401940194e-05,
      "loss": 3.0631,
      "step": 500350
    },
    {
      "epoch": 0.24210106475017307,
      "grad_norm": 0.7265625,
      "learning_rate": 9.993819381938195e-05,
      "loss": 3.0535,
      "step": 500360
    },
    {
      "epoch": 0.24211970798633126,
      "grad_norm": 0.67578125,
      "learning_rate": 9.993619361936194e-05,
      "loss": 3.2563,
      "step": 500370
    },
    {
      "epoch": 0.24213835122248942,
      "grad_norm": 0.71484375,
      "learning_rate": 9.993419341934194e-05,
      "loss": 3.2743,
      "step": 500380
    },
    {
      "epoch": 0.2421569944586476,
      "grad_norm": 0.7734375,
      "learning_rate": 9.993219321932194e-05,
      "loss": 3.221,
      "step": 500390
    },
    {
      "epoch": 0.2421756376948058,
      "grad_norm": 0.7734375,
      "learning_rate": 9.993019301930193e-05,
      "loss": 3.1649,
      "step": 500400
    },
    {
      "epoch": 0.24219428093096398,
      "grad_norm": 0.78515625,
      "learning_rate": 9.992819281928193e-05,
      "loss": 3.2019,
      "step": 500410
    },
    {
      "epoch": 0.24221292416712217,
      "grad_norm": 0.8046875,
      "learning_rate": 9.992619261926192e-05,
      "loss": 3.222,
      "step": 500420
    },
    {
      "epoch": 0.24223156740328033,
      "grad_norm": 1.1640625,
      "learning_rate": 9.992419241924192e-05,
      "loss": 3.1695,
      "step": 500430
    },
    {
      "epoch": 0.24225021063943852,
      "grad_norm": 0.74609375,
      "learning_rate": 9.992219221922193e-05,
      "loss": 3.3029,
      "step": 500440
    },
    {
      "epoch": 0.2422688538755967,
      "grad_norm": 0.70703125,
      "learning_rate": 9.992019201920194e-05,
      "loss": 3.1738,
      "step": 500450
    },
    {
      "epoch": 0.2422874971117549,
      "grad_norm": 0.7421875,
      "learning_rate": 9.991819181918193e-05,
      "loss": 3.1446,
      "step": 500460
    },
    {
      "epoch": 0.2423061403479131,
      "grad_norm": 0.8671875,
      "learning_rate": 9.991619161916193e-05,
      "loss": 3.2035,
      "step": 500470
    },
    {
      "epoch": 0.24232478358407128,
      "grad_norm": 0.73828125,
      "learning_rate": 9.991419141914192e-05,
      "loss": 3.1585,
      "step": 500480
    },
    {
      "epoch": 0.24234342682022944,
      "grad_norm": 0.6484375,
      "learning_rate": 9.991219121912192e-05,
      "loss": 3.3666,
      "step": 500490
    },
    {
      "epoch": 0.24236207005638763,
      "grad_norm": 0.77734375,
      "learning_rate": 9.991019101910191e-05,
      "loss": 3.2475,
      "step": 500500
    },
    {
      "epoch": 0.24238071329254582,
      "grad_norm": 0.8125,
      "learning_rate": 9.990819081908191e-05,
      "loss": 3.3008,
      "step": 500510
    },
    {
      "epoch": 0.242399356528704,
      "grad_norm": 0.83203125,
      "learning_rate": 9.990619061906192e-05,
      "loss": 3.1049,
      "step": 500520
    },
    {
      "epoch": 0.2424179997648622,
      "grad_norm": 0.75390625,
      "learning_rate": 9.990419041904191e-05,
      "loss": 3.2543,
      "step": 500530
    },
    {
      "epoch": 0.24243664300102036,
      "grad_norm": 0.7421875,
      "learning_rate": 9.990219021902191e-05,
      "loss": 3.1767,
      "step": 500540
    },
    {
      "epoch": 0.24245528623717855,
      "grad_norm": 0.78125,
      "learning_rate": 9.99001900190019e-05,
      "loss": 3.1775,
      "step": 500550
    },
    {
      "epoch": 0.24247392947333674,
      "grad_norm": 0.65234375,
      "learning_rate": 9.98981898189819e-05,
      "loss": 3.1005,
      "step": 500560
    },
    {
      "epoch": 0.24249257270949492,
      "grad_norm": 0.75390625,
      "learning_rate": 9.98961896189619e-05,
      "loss": 3.251,
      "step": 500570
    },
    {
      "epoch": 0.2425112159456531,
      "grad_norm": 0.703125,
      "learning_rate": 9.989418941894189e-05,
      "loss": 3.154,
      "step": 500580
    },
    {
      "epoch": 0.24252985918181127,
      "grad_norm": 0.76953125,
      "learning_rate": 9.989218921892189e-05,
      "loss": 3.29,
      "step": 500590
    },
    {
      "epoch": 0.24254850241796946,
      "grad_norm": 0.59375,
      "learning_rate": 9.98901890189019e-05,
      "loss": 3.3427,
      "step": 500600
    },
    {
      "epoch": 0.24256714565412765,
      "grad_norm": 0.69140625,
      "learning_rate": 9.988818881888189e-05,
      "loss": 3.1983,
      "step": 500610
    },
    {
      "epoch": 0.24258578889028584,
      "grad_norm": 0.69921875,
      "learning_rate": 9.988618861886189e-05,
      "loss": 3.2171,
      "step": 500620
    },
    {
      "epoch": 0.24260443212644403,
      "grad_norm": 0.73046875,
      "learning_rate": 9.988418841884188e-05,
      "loss": 3.2403,
      "step": 500630
    },
    {
      "epoch": 0.2426230753626022,
      "grad_norm": 0.765625,
      "learning_rate": 9.988218821882188e-05,
      "loss": 3.0746,
      "step": 500640
    },
    {
      "epoch": 0.24264171859876038,
      "grad_norm": 0.7734375,
      "learning_rate": 9.988018801880189e-05,
      "loss": 3.2613,
      "step": 500650
    },
    {
      "epoch": 0.24266036183491857,
      "grad_norm": 0.79296875,
      "learning_rate": 9.987818781878188e-05,
      "loss": 3.0741,
      "step": 500660
    },
    {
      "epoch": 0.24267900507107676,
      "grad_norm": 0.68359375,
      "learning_rate": 9.987618761876188e-05,
      "loss": 3.3553,
      "step": 500670
    },
    {
      "epoch": 0.24269764830723495,
      "grad_norm": 0.74609375,
      "learning_rate": 9.987418741874189e-05,
      "loss": 3.1403,
      "step": 500680
    },
    {
      "epoch": 0.2427162915433931,
      "grad_norm": 0.70703125,
      "learning_rate": 9.987218721872188e-05,
      "loss": 3.2288,
      "step": 500690
    },
    {
      "epoch": 0.2427349347795513,
      "grad_norm": 0.875,
      "learning_rate": 9.987018701870188e-05,
      "loss": 3.1971,
      "step": 500700
    },
    {
      "epoch": 0.2427535780157095,
      "grad_norm": 0.765625,
      "learning_rate": 9.986818681868188e-05,
      "loss": 3.3304,
      "step": 500710
    },
    {
      "epoch": 0.24277222125186768,
      "grad_norm": 0.78515625,
      "learning_rate": 9.986618661866187e-05,
      "loss": 3.1936,
      "step": 500720
    },
    {
      "epoch": 0.24279086448802586,
      "grad_norm": 0.7265625,
      "learning_rate": 9.986418641864187e-05,
      "loss": 3.2185,
      "step": 500730
    },
    {
      "epoch": 0.24280950772418405,
      "grad_norm": 0.734375,
      "learning_rate": 9.986218621862186e-05,
      "loss": 3.1057,
      "step": 500740
    },
    {
      "epoch": 0.24282815096034222,
      "grad_norm": 0.734375,
      "learning_rate": 9.986018601860186e-05,
      "loss": 3.2387,
      "step": 500750
    },
    {
      "epoch": 0.2428467941965004,
      "grad_norm": 0.765625,
      "learning_rate": 9.985818581858187e-05,
      "loss": 3.2108,
      "step": 500760
    },
    {
      "epoch": 0.2428654374326586,
      "grad_norm": 0.65625,
      "learning_rate": 9.985618561856186e-05,
      "loss": 3.2186,
      "step": 500770
    },
    {
      "epoch": 0.24288408066881678,
      "grad_norm": 0.75390625,
      "learning_rate": 9.985418541854186e-05,
      "loss": 3.3081,
      "step": 500780
    },
    {
      "epoch": 0.24290272390497497,
      "grad_norm": 0.78125,
      "learning_rate": 9.985218521852185e-05,
      "loss": 3.214,
      "step": 500790
    },
    {
      "epoch": 0.24292136714113313,
      "grad_norm": 0.828125,
      "learning_rate": 9.985018501850185e-05,
      "loss": 3.2302,
      "step": 500800
    },
    {
      "epoch": 0.24294001037729132,
      "grad_norm": 0.69921875,
      "learning_rate": 9.984818481848185e-05,
      "loss": 3.1059,
      "step": 500810
    },
    {
      "epoch": 0.2429586536134495,
      "grad_norm": 0.69921875,
      "learning_rate": 9.984618461846184e-05,
      "loss": 3.0844,
      "step": 500820
    },
    {
      "epoch": 0.2429772968496077,
      "grad_norm": 0.73828125,
      "learning_rate": 9.984418441844185e-05,
      "loss": 3.3077,
      "step": 500830
    },
    {
      "epoch": 0.2429959400857659,
      "grad_norm": 0.73046875,
      "learning_rate": 9.984218421842185e-05,
      "loss": 3.2132,
      "step": 500840
    },
    {
      "epoch": 0.24301458332192405,
      "grad_norm": 0.78515625,
      "learning_rate": 9.984018401840185e-05,
      "loss": 3.1236,
      "step": 500850
    },
    {
      "epoch": 0.24303322655808224,
      "grad_norm": 0.73046875,
      "learning_rate": 9.983818381838185e-05,
      "loss": 3.3369,
      "step": 500860
    },
    {
      "epoch": 0.24305186979424043,
      "grad_norm": 0.68359375,
      "learning_rate": 9.983618361836185e-05,
      "loss": 3.3546,
      "step": 500870
    },
    {
      "epoch": 0.24307051303039862,
      "grad_norm": 0.70703125,
      "learning_rate": 9.983418341834184e-05,
      "loss": 3.1765,
      "step": 500880
    },
    {
      "epoch": 0.2430891562665568,
      "grad_norm": 0.69140625,
      "learning_rate": 9.983218321832184e-05,
      "loss": 3.2504,
      "step": 500890
    },
    {
      "epoch": 0.24310779950271497,
      "grad_norm": 0.7734375,
      "learning_rate": 9.983018301830183e-05,
      "loss": 3.2493,
      "step": 500900
    },
    {
      "epoch": 0.24312644273887316,
      "grad_norm": 0.76953125,
      "learning_rate": 9.982818281828183e-05,
      "loss": 3.2269,
      "step": 500910
    },
    {
      "epoch": 0.24314508597503134,
      "grad_norm": 0.85546875,
      "learning_rate": 9.982618261826184e-05,
      "loss": 3.2724,
      "step": 500920
    },
    {
      "epoch": 0.24316372921118953,
      "grad_norm": 0.70703125,
      "learning_rate": 9.982418241824183e-05,
      "loss": 3.2296,
      "step": 500930
    },
    {
      "epoch": 0.24318237244734772,
      "grad_norm": 0.6953125,
      "learning_rate": 9.982218221822183e-05,
      "loss": 3.1722,
      "step": 500940
    },
    {
      "epoch": 0.2432010156835059,
      "grad_norm": 0.734375,
      "learning_rate": 9.982018201820182e-05,
      "loss": 3.1199,
      "step": 500950
    },
    {
      "epoch": 0.24321965891966407,
      "grad_norm": 0.7265625,
      "learning_rate": 9.981818181818182e-05,
      "loss": 3.3007,
      "step": 500960
    },
    {
      "epoch": 0.24323830215582226,
      "grad_norm": 0.68359375,
      "learning_rate": 9.981618161816182e-05,
      "loss": 3.2022,
      "step": 500970
    },
    {
      "epoch": 0.24325694539198045,
      "grad_norm": 0.70703125,
      "learning_rate": 9.981418141814181e-05,
      "loss": 3.2554,
      "step": 500980
    },
    {
      "epoch": 0.24327558862813864,
      "grad_norm": 0.69140625,
      "learning_rate": 9.981218121812182e-05,
      "loss": 3.1744,
      "step": 500990
    },
    {
      "epoch": 0.24329423186429683,
      "grad_norm": 0.6796875,
      "learning_rate": 9.981018101810182e-05,
      "loss": 3.221,
      "step": 501000
    },
    {
      "epoch": 0.243312875100455,
      "grad_norm": 0.76171875,
      "learning_rate": 9.980818081808181e-05,
      "loss": 3.1428,
      "step": 501010
    },
    {
      "epoch": 0.24333151833661318,
      "grad_norm": 0.69921875,
      "learning_rate": 9.980618061806181e-05,
      "loss": 3.0035,
      "step": 501020
    },
    {
      "epoch": 0.24335016157277137,
      "grad_norm": 0.703125,
      "learning_rate": 9.98041804180418e-05,
      "loss": 3.2922,
      "step": 501030
    },
    {
      "epoch": 0.24336880480892956,
      "grad_norm": 0.73828125,
      "learning_rate": 9.980218021802181e-05,
      "loss": 3.0626,
      "step": 501040
    },
    {
      "epoch": 0.24338744804508775,
      "grad_norm": 0.73828125,
      "learning_rate": 9.980018001800181e-05,
      "loss": 3.1862,
      "step": 501050
    },
    {
      "epoch": 0.2434060912812459,
      "grad_norm": 0.72265625,
      "learning_rate": 9.97981798179818e-05,
      "loss": 3.1634,
      "step": 501060
    },
    {
      "epoch": 0.2434247345174041,
      "grad_norm": 0.73046875,
      "learning_rate": 9.97961796179618e-05,
      "loss": 3.1178,
      "step": 501070
    },
    {
      "epoch": 0.24344337775356228,
      "grad_norm": 0.75,
      "learning_rate": 9.979417941794181e-05,
      "loss": 3.2898,
      "step": 501080
    },
    {
      "epoch": 0.24346202098972047,
      "grad_norm": 0.77734375,
      "learning_rate": 9.97921792179218e-05,
      "loss": 3.2134,
      "step": 501090
    },
    {
      "epoch": 0.24348066422587866,
      "grad_norm": 0.6875,
      "learning_rate": 9.97901790179018e-05,
      "loss": 3.2434,
      "step": 501100
    },
    {
      "epoch": 0.24349930746203682,
      "grad_norm": 0.640625,
      "learning_rate": 9.97881788178818e-05,
      "loss": 3.0575,
      "step": 501110
    },
    {
      "epoch": 0.243517950698195,
      "grad_norm": 0.78515625,
      "learning_rate": 9.978617861786179e-05,
      "loss": 3.2591,
      "step": 501120
    },
    {
      "epoch": 0.2435365939343532,
      "grad_norm": 0.91796875,
      "learning_rate": 9.978417841784179e-05,
      "loss": 3.1039,
      "step": 501130
    },
    {
      "epoch": 0.2435552371705114,
      "grad_norm": 0.66796875,
      "learning_rate": 9.978217821782178e-05,
      "loss": 3.1928,
      "step": 501140
    },
    {
      "epoch": 0.24357388040666958,
      "grad_norm": 0.76953125,
      "learning_rate": 9.978017801780179e-05,
      "loss": 3.2611,
      "step": 501150
    },
    {
      "epoch": 0.24359252364282777,
      "grad_norm": 0.734375,
      "learning_rate": 9.977817781778179e-05,
      "loss": 3.3163,
      "step": 501160
    },
    {
      "epoch": 0.24361116687898593,
      "grad_norm": 0.6875,
      "learning_rate": 9.977617761776178e-05,
      "loss": 3.172,
      "step": 501170
    },
    {
      "epoch": 0.24362981011514412,
      "grad_norm": 0.625,
      "learning_rate": 9.977417741774178e-05,
      "loss": 3.2733,
      "step": 501180
    },
    {
      "epoch": 0.2436484533513023,
      "grad_norm": 0.74609375,
      "learning_rate": 9.977217721772177e-05,
      "loss": 3.2373,
      "step": 501190
    },
    {
      "epoch": 0.2436670965874605,
      "grad_norm": 0.734375,
      "learning_rate": 9.977017701770177e-05,
      "loss": 3.3581,
      "step": 501200
    },
    {
      "epoch": 0.24368573982361869,
      "grad_norm": 0.796875,
      "learning_rate": 9.976817681768176e-05,
      "loss": 3.1661,
      "step": 501210
    },
    {
      "epoch": 0.24370438305977685,
      "grad_norm": 0.69140625,
      "learning_rate": 9.976617661766177e-05,
      "loss": 3.2399,
      "step": 501220
    },
    {
      "epoch": 0.24372302629593504,
      "grad_norm": 0.66796875,
      "learning_rate": 9.976417641764177e-05,
      "loss": 3.1997,
      "step": 501230
    },
    {
      "epoch": 0.24374166953209322,
      "grad_norm": 0.80078125,
      "learning_rate": 9.976217621762178e-05,
      "loss": 3.1438,
      "step": 501240
    },
    {
      "epoch": 0.2437603127682514,
      "grad_norm": 0.6875,
      "learning_rate": 9.976017601760177e-05,
      "loss": 3.1386,
      "step": 501250
    },
    {
      "epoch": 0.2437789560044096,
      "grad_norm": 0.71875,
      "learning_rate": 9.975817581758177e-05,
      "loss": 3.1401,
      "step": 501260
    },
    {
      "epoch": 0.24379759924056776,
      "grad_norm": 0.7265625,
      "learning_rate": 9.975617561756176e-05,
      "loss": 3.1237,
      "step": 501270
    },
    {
      "epoch": 0.24381624247672595,
      "grad_norm": 0.82421875,
      "learning_rate": 9.975417541754176e-05,
      "loss": 3.1931,
      "step": 501280
    },
    {
      "epoch": 0.24383488571288414,
      "grad_norm": 0.765625,
      "learning_rate": 9.975217521752176e-05,
      "loss": 3.1014,
      "step": 501290
    },
    {
      "epoch": 0.24385352894904233,
      "grad_norm": 0.81640625,
      "learning_rate": 9.975017501750175e-05,
      "loss": 3.2604,
      "step": 501300
    },
    {
      "epoch": 0.24387217218520052,
      "grad_norm": 0.78515625,
      "learning_rate": 9.974817481748176e-05,
      "loss": 3.2205,
      "step": 501310
    },
    {
      "epoch": 0.24389081542135868,
      "grad_norm": 0.73828125,
      "learning_rate": 9.974617461746176e-05,
      "loss": 3.1871,
      "step": 501320
    },
    {
      "epoch": 0.24390945865751687,
      "grad_norm": 0.71875,
      "learning_rate": 9.974417441744175e-05,
      "loss": 3.1019,
      "step": 501330
    },
    {
      "epoch": 0.24392810189367506,
      "grad_norm": 0.73828125,
      "learning_rate": 9.974217421742175e-05,
      "loss": 3.1522,
      "step": 501340
    },
    {
      "epoch": 0.24394674512983325,
      "grad_norm": 0.78125,
      "learning_rate": 9.974017401740174e-05,
      "loss": 3.1235,
      "step": 501350
    },
    {
      "epoch": 0.24396538836599144,
      "grad_norm": 0.63671875,
      "learning_rate": 9.973817381738174e-05,
      "loss": 3.2066,
      "step": 501360
    },
    {
      "epoch": 0.2439840316021496,
      "grad_norm": 0.75,
      "learning_rate": 9.973617361736173e-05,
      "loss": 3.1859,
      "step": 501370
    },
    {
      "epoch": 0.2440026748383078,
      "grad_norm": 0.62109375,
      "learning_rate": 9.973417341734173e-05,
      "loss": 3.2857,
      "step": 501380
    },
    {
      "epoch": 0.24402131807446598,
      "grad_norm": 0.88671875,
      "learning_rate": 9.973217321732174e-05,
      "loss": 3.1783,
      "step": 501390
    },
    {
      "epoch": 0.24403996131062416,
      "grad_norm": 0.6640625,
      "learning_rate": 9.973017301730173e-05,
      "loss": 3.2291,
      "step": 501400
    },
    {
      "epoch": 0.24405860454678235,
      "grad_norm": 0.75,
      "learning_rate": 9.972817281728173e-05,
      "loss": 3.2067,
      "step": 501410
    },
    {
      "epoch": 0.24407724778294054,
      "grad_norm": 0.77734375,
      "learning_rate": 9.972617261726173e-05,
      "loss": 3.2617,
      "step": 501420
    },
    {
      "epoch": 0.2440958910190987,
      "grad_norm": 0.7578125,
      "learning_rate": 9.972417241724172e-05,
      "loss": 3.2167,
      "step": 501430
    },
    {
      "epoch": 0.2441145342552569,
      "grad_norm": 0.81640625,
      "learning_rate": 9.972217221722173e-05,
      "loss": 3.2552,
      "step": 501440
    },
    {
      "epoch": 0.24413317749141508,
      "grad_norm": 0.71484375,
      "learning_rate": 9.972017201720173e-05,
      "loss": 3.1953,
      "step": 501450
    },
    {
      "epoch": 0.24415182072757327,
      "grad_norm": 0.75,
      "learning_rate": 9.971817181718172e-05,
      "loss": 3.18,
      "step": 501460
    },
    {
      "epoch": 0.24417046396373146,
      "grad_norm": 0.7890625,
      "learning_rate": 9.971617161716173e-05,
      "loss": 3.2581,
      "step": 501470
    },
    {
      "epoch": 0.24418910719988962,
      "grad_norm": 0.671875,
      "learning_rate": 9.971417141714173e-05,
      "loss": 3.1717,
      "step": 501480
    },
    {
      "epoch": 0.2442077504360478,
      "grad_norm": 0.71484375,
      "learning_rate": 9.971217121712172e-05,
      "loss": 3.1521,
      "step": 501490
    },
    {
      "epoch": 0.244226393672206,
      "grad_norm": 0.7734375,
      "learning_rate": 9.971017101710172e-05,
      "loss": 3.2464,
      "step": 501500
    },
    {
      "epoch": 0.2442450369083642,
      "grad_norm": 0.8203125,
      "learning_rate": 9.970817081708171e-05,
      "loss": 3.1981,
      "step": 501510
    },
    {
      "epoch": 0.24426368014452238,
      "grad_norm": 0.71875,
      "learning_rate": 9.970617061706171e-05,
      "loss": 3.2193,
      "step": 501520
    },
    {
      "epoch": 0.24428232338068054,
      "grad_norm": 0.65234375,
      "learning_rate": 9.97041704170417e-05,
      "loss": 3.1857,
      "step": 501530
    },
    {
      "epoch": 0.24430096661683873,
      "grad_norm": 0.78125,
      "learning_rate": 9.97021702170217e-05,
      "loss": 3.2736,
      "step": 501540
    },
    {
      "epoch": 0.24431960985299692,
      "grad_norm": 0.6328125,
      "learning_rate": 9.970017001700171e-05,
      "loss": 3.2601,
      "step": 501550
    },
    {
      "epoch": 0.2443382530891551,
      "grad_norm": 0.80859375,
      "learning_rate": 9.96981698169817e-05,
      "loss": 3.2855,
      "step": 501560
    },
    {
      "epoch": 0.2443568963253133,
      "grad_norm": 0.7109375,
      "learning_rate": 9.96961696169617e-05,
      "loss": 3.2702,
      "step": 501570
    },
    {
      "epoch": 0.24437553956147146,
      "grad_norm": 0.73828125,
      "learning_rate": 9.96941694169417e-05,
      "loss": 3.2294,
      "step": 501580
    },
    {
      "epoch": 0.24439418279762964,
      "grad_norm": 0.74609375,
      "learning_rate": 9.969216921692169e-05,
      "loss": 3.136,
      "step": 501590
    },
    {
      "epoch": 0.24441282603378783,
      "grad_norm": 0.90625,
      "learning_rate": 9.969016901690169e-05,
      "loss": 3.1475,
      "step": 501600
    },
    {
      "epoch": 0.24443146926994602,
      "grad_norm": 0.68359375,
      "learning_rate": 9.968816881688168e-05,
      "loss": 3.2659,
      "step": 501610
    },
    {
      "epoch": 0.2444501125061042,
      "grad_norm": 0.73046875,
      "learning_rate": 9.968616861686169e-05,
      "loss": 3.1851,
      "step": 501620
    },
    {
      "epoch": 0.2444687557422624,
      "grad_norm": 0.671875,
      "learning_rate": 9.96841684168417e-05,
      "loss": 3.1977,
      "step": 501630
    },
    {
      "epoch": 0.24448739897842056,
      "grad_norm": 0.8984375,
      "learning_rate": 9.96821682168217e-05,
      "loss": 3.2177,
      "step": 501640
    },
    {
      "epoch": 0.24450604221457875,
      "grad_norm": 0.75390625,
      "learning_rate": 9.968016801680169e-05,
      "loss": 3.1939,
      "step": 501650
    },
    {
      "epoch": 0.24452468545073694,
      "grad_norm": 0.75390625,
      "learning_rate": 9.967816781678169e-05,
      "loss": 3.2309,
      "step": 501660
    },
    {
      "epoch": 0.24454332868689513,
      "grad_norm": 0.82421875,
      "learning_rate": 9.967616761676168e-05,
      "loss": 3.1138,
      "step": 501670
    },
    {
      "epoch": 0.24456197192305332,
      "grad_norm": 0.7578125,
      "learning_rate": 9.967416741674168e-05,
      "loss": 3.3079,
      "step": 501680
    },
    {
      "epoch": 0.24458061515921148,
      "grad_norm": 0.6640625,
      "learning_rate": 9.967216721672167e-05,
      "loss": 3.2522,
      "step": 501690
    },
    {
      "epoch": 0.24459925839536967,
      "grad_norm": 0.7265625,
      "learning_rate": 9.967016701670167e-05,
      "loss": 3.2403,
      "step": 501700
    },
    {
      "epoch": 0.24461790163152786,
      "grad_norm": 0.63671875,
      "learning_rate": 9.966816681668168e-05,
      "loss": 3.15,
      "step": 501710
    },
    {
      "epoch": 0.24463654486768605,
      "grad_norm": 0.80078125,
      "learning_rate": 9.966616661666167e-05,
      "loss": 3.2954,
      "step": 501720
    },
    {
      "epoch": 0.24465518810384423,
      "grad_norm": 0.65625,
      "learning_rate": 9.966416641664167e-05,
      "loss": 3.3737,
      "step": 501730
    },
    {
      "epoch": 0.2446738313400024,
      "grad_norm": 0.78125,
      "learning_rate": 9.966216621662167e-05,
      "loss": 3.2704,
      "step": 501740
    },
    {
      "epoch": 0.24469247457616058,
      "grad_norm": 0.77734375,
      "learning_rate": 9.966016601660166e-05,
      "loss": 3.1737,
      "step": 501750
    },
    {
      "epoch": 0.24471111781231877,
      "grad_norm": 0.7109375,
      "learning_rate": 9.965816581658166e-05,
      "loss": 3.1041,
      "step": 501760
    },
    {
      "epoch": 0.24472976104847696,
      "grad_norm": 0.84765625,
      "learning_rate": 9.965616561656165e-05,
      "loss": 3.1685,
      "step": 501770
    },
    {
      "epoch": 0.24474840428463515,
      "grad_norm": 0.85546875,
      "learning_rate": 9.965416541654166e-05,
      "loss": 3.2114,
      "step": 501780
    },
    {
      "epoch": 0.2447670475207933,
      "grad_norm": 0.68359375,
      "learning_rate": 9.965216521652166e-05,
      "loss": 3.274,
      "step": 501790
    },
    {
      "epoch": 0.2447856907569515,
      "grad_norm": 0.8828125,
      "learning_rate": 9.965016501650165e-05,
      "loss": 3.2092,
      "step": 501800
    },
    {
      "epoch": 0.2448043339931097,
      "grad_norm": 0.73046875,
      "learning_rate": 9.964816481648165e-05,
      "loss": 3.2419,
      "step": 501810
    },
    {
      "epoch": 0.24482297722926788,
      "grad_norm": 0.87109375,
      "learning_rate": 9.964616461646164e-05,
      "loss": 3.2224,
      "step": 501820
    },
    {
      "epoch": 0.24484162046542607,
      "grad_norm": 0.828125,
      "learning_rate": 9.964416441644165e-05,
      "loss": 3.1867,
      "step": 501830
    },
    {
      "epoch": 0.24486026370158426,
      "grad_norm": 0.73828125,
      "learning_rate": 9.964216421642165e-05,
      "loss": 3.2618,
      "step": 501840
    },
    {
      "epoch": 0.24487890693774242,
      "grad_norm": 0.73828125,
      "learning_rate": 9.964016401640164e-05,
      "loss": 3.2107,
      "step": 501850
    },
    {
      "epoch": 0.2448975501739006,
      "grad_norm": 0.7109375,
      "learning_rate": 9.963816381638164e-05,
      "loss": 3.2334,
      "step": 501860
    },
    {
      "epoch": 0.2449161934100588,
      "grad_norm": 0.8125,
      "learning_rate": 9.963616361636165e-05,
      "loss": 3.246,
      "step": 501870
    },
    {
      "epoch": 0.24493483664621699,
      "grad_norm": 0.7890625,
      "learning_rate": 9.963416341634164e-05,
      "loss": 3.0984,
      "step": 501880
    },
    {
      "epoch": 0.24495347988237517,
      "grad_norm": 0.765625,
      "learning_rate": 9.963216321632164e-05,
      "loss": 3.2826,
      "step": 501890
    },
    {
      "epoch": 0.24497212311853334,
      "grad_norm": 0.86328125,
      "learning_rate": 9.963016301630164e-05,
      "loss": 3.1549,
      "step": 501900
    },
    {
      "epoch": 0.24499076635469152,
      "grad_norm": 0.75390625,
      "learning_rate": 9.962816281628163e-05,
      "loss": 3.1347,
      "step": 501910
    },
    {
      "epoch": 0.2450094095908497,
      "grad_norm": 0.7578125,
      "learning_rate": 9.962616261626163e-05,
      "loss": 3.1729,
      "step": 501920
    },
    {
      "epoch": 0.2450280528270079,
      "grad_norm": 0.75390625,
      "learning_rate": 9.962416241624162e-05,
      "loss": 3.3147,
      "step": 501930
    },
    {
      "epoch": 0.2450466960631661,
      "grad_norm": 0.72265625,
      "learning_rate": 9.962216221622163e-05,
      "loss": 3.2164,
      "step": 501940
    },
    {
      "epoch": 0.24506533929932425,
      "grad_norm": 0.62109375,
      "learning_rate": 9.962016201620163e-05,
      "loss": 3.108,
      "step": 501950
    },
    {
      "epoch": 0.24508398253548244,
      "grad_norm": 0.76953125,
      "learning_rate": 9.961816181618162e-05,
      "loss": 3.2066,
      "step": 501960
    },
    {
      "epoch": 0.24510262577164063,
      "grad_norm": 0.7734375,
      "learning_rate": 9.961616161616162e-05,
      "loss": 3.3162,
      "step": 501970
    },
    {
      "epoch": 0.24512126900779882,
      "grad_norm": 0.75,
      "learning_rate": 9.961416141614161e-05,
      "loss": 3.2711,
      "step": 501980
    },
    {
      "epoch": 0.245139912243957,
      "grad_norm": 0.7734375,
      "learning_rate": 9.961216121612161e-05,
      "loss": 3.1479,
      "step": 501990
    },
    {
      "epoch": 0.24515855548011517,
      "grad_norm": 0.70703125,
      "learning_rate": 9.96101610161016e-05,
      "loss": 3.2633,
      "step": 502000
    },
    {
      "epoch": 0.24517719871627336,
      "grad_norm": 0.7578125,
      "learning_rate": 9.960816081608161e-05,
      "loss": 3.1467,
      "step": 502010
    },
    {
      "epoch": 0.24519584195243155,
      "grad_norm": 0.73828125,
      "learning_rate": 9.960616061606161e-05,
      "loss": 3.2482,
      "step": 502020
    },
    {
      "epoch": 0.24521448518858974,
      "grad_norm": 0.66796875,
      "learning_rate": 9.960416041604162e-05,
      "loss": 3.2895,
      "step": 502030
    },
    {
      "epoch": 0.24523312842474793,
      "grad_norm": 0.68359375,
      "learning_rate": 9.960216021602162e-05,
      "loss": 3.1946,
      "step": 502040
    },
    {
      "epoch": 0.2452517716609061,
      "grad_norm": 0.70703125,
      "learning_rate": 9.960016001600161e-05,
      "loss": 3.307,
      "step": 502050
    },
    {
      "epoch": 0.24527041489706428,
      "grad_norm": 0.66796875,
      "learning_rate": 9.95981598159816e-05,
      "loss": 3.1441,
      "step": 502060
    },
    {
      "epoch": 0.24528905813322247,
      "grad_norm": 0.69921875,
      "learning_rate": 9.95961596159616e-05,
      "loss": 3.1777,
      "step": 502070
    },
    {
      "epoch": 0.24530770136938065,
      "grad_norm": 0.7265625,
      "learning_rate": 9.95941594159416e-05,
      "loss": 3.2455,
      "step": 502080
    },
    {
      "epoch": 0.24532634460553884,
      "grad_norm": 0.75,
      "learning_rate": 9.959215921592159e-05,
      "loss": 3.2346,
      "step": 502090
    },
    {
      "epoch": 0.24534498784169703,
      "grad_norm": 0.66796875,
      "learning_rate": 9.95901590159016e-05,
      "loss": 3.2627,
      "step": 502100
    },
    {
      "epoch": 0.2453636310778552,
      "grad_norm": 0.7265625,
      "learning_rate": 9.95881588158816e-05,
      "loss": 3.2471,
      "step": 502110
    },
    {
      "epoch": 0.24538227431401338,
      "grad_norm": 0.7734375,
      "learning_rate": 9.95861586158616e-05,
      "loss": 3.2305,
      "step": 502120
    },
    {
      "epoch": 0.24540091755017157,
      "grad_norm": 0.625,
      "learning_rate": 9.958415841584159e-05,
      "loss": 3.2201,
      "step": 502130
    },
    {
      "epoch": 0.24541956078632976,
      "grad_norm": 0.66015625,
      "learning_rate": 9.958215821582158e-05,
      "loss": 3.0984,
      "step": 502140
    },
    {
      "epoch": 0.24543820402248795,
      "grad_norm": 0.828125,
      "learning_rate": 9.958015801580158e-05,
      "loss": 3.1846,
      "step": 502150
    },
    {
      "epoch": 0.2454568472586461,
      "grad_norm": 0.74609375,
      "learning_rate": 9.957815781578158e-05,
      "loss": 3.2895,
      "step": 502160
    },
    {
      "epoch": 0.2454754904948043,
      "grad_norm": 0.7421875,
      "learning_rate": 9.957615761576158e-05,
      "loss": 3.0894,
      "step": 502170
    },
    {
      "epoch": 0.2454941337309625,
      "grad_norm": 0.74609375,
      "learning_rate": 9.957415741574158e-05,
      "loss": 3.2536,
      "step": 502180
    },
    {
      "epoch": 0.24551277696712068,
      "grad_norm": 0.6875,
      "learning_rate": 9.957215721572158e-05,
      "loss": 3.1924,
      "step": 502190
    },
    {
      "epoch": 0.24553142020327887,
      "grad_norm": 0.7421875,
      "learning_rate": 9.957015701570157e-05,
      "loss": 3.103,
      "step": 502200
    },
    {
      "epoch": 0.24555006343943703,
      "grad_norm": 0.6640625,
      "learning_rate": 9.956815681568157e-05,
      "loss": 3.2641,
      "step": 502210
    },
    {
      "epoch": 0.24556870667559522,
      "grad_norm": 0.7734375,
      "learning_rate": 9.956615661566156e-05,
      "loss": 3.2203,
      "step": 502220
    },
    {
      "epoch": 0.2455873499117534,
      "grad_norm": 0.703125,
      "learning_rate": 9.956415641564157e-05,
      "loss": 3.2659,
      "step": 502230
    },
    {
      "epoch": 0.2456059931479116,
      "grad_norm": 0.70703125,
      "learning_rate": 9.956215621562157e-05,
      "loss": 3.2336,
      "step": 502240
    },
    {
      "epoch": 0.24562463638406978,
      "grad_norm": 0.76171875,
      "learning_rate": 9.956015601560156e-05,
      "loss": 3.2152,
      "step": 502250
    },
    {
      "epoch": 0.24564327962022794,
      "grad_norm": 0.70703125,
      "learning_rate": 9.955815581558157e-05,
      "loss": 3.1644,
      "step": 502260
    },
    {
      "epoch": 0.24566192285638613,
      "grad_norm": 0.8125,
      "learning_rate": 9.955615561556157e-05,
      "loss": 3.2317,
      "step": 502270
    },
    {
      "epoch": 0.24568056609254432,
      "grad_norm": 0.68359375,
      "learning_rate": 9.955415541554156e-05,
      "loss": 3.2805,
      "step": 502280
    },
    {
      "epoch": 0.2456992093287025,
      "grad_norm": 0.76171875,
      "learning_rate": 9.955215521552156e-05,
      "loss": 3.2651,
      "step": 502290
    },
    {
      "epoch": 0.2457178525648607,
      "grad_norm": 0.828125,
      "learning_rate": 9.955015501550155e-05,
      "loss": 3.1555,
      "step": 502300
    },
    {
      "epoch": 0.2457364958010189,
      "grad_norm": 0.7421875,
      "learning_rate": 9.954815481548155e-05,
      "loss": 3.2641,
      "step": 502310
    },
    {
      "epoch": 0.24575513903717705,
      "grad_norm": 0.7265625,
      "learning_rate": 9.954615461546155e-05,
      "loss": 3.0906,
      "step": 502320
    },
    {
      "epoch": 0.24577378227333524,
      "grad_norm": 0.76953125,
      "learning_rate": 9.954415441544155e-05,
      "loss": 3.227,
      "step": 502330
    },
    {
      "epoch": 0.24579242550949343,
      "grad_norm": 0.9296875,
      "learning_rate": 9.954215421542155e-05,
      "loss": 3.1375,
      "step": 502340
    },
    {
      "epoch": 0.24581106874565162,
      "grad_norm": 0.703125,
      "learning_rate": 9.954015401540155e-05,
      "loss": 3.0337,
      "step": 502350
    },
    {
      "epoch": 0.2458297119818098,
      "grad_norm": 0.8125,
      "learning_rate": 9.953815381538154e-05,
      "loss": 3.1385,
      "step": 502360
    },
    {
      "epoch": 0.24584835521796797,
      "grad_norm": 0.76171875,
      "learning_rate": 9.953615361536154e-05,
      "loss": 3.2897,
      "step": 502370
    },
    {
      "epoch": 0.24586699845412616,
      "grad_norm": 0.76171875,
      "learning_rate": 9.953415341534153e-05,
      "loss": 3.1594,
      "step": 502380
    },
    {
      "epoch": 0.24588564169028435,
      "grad_norm": 0.83984375,
      "learning_rate": 9.953215321532153e-05,
      "loss": 3.301,
      "step": 502390
    },
    {
      "epoch": 0.24590428492644253,
      "grad_norm": 0.83203125,
      "learning_rate": 9.953015301530152e-05,
      "loss": 3.2288,
      "step": 502400
    },
    {
      "epoch": 0.24592292816260072,
      "grad_norm": 0.76953125,
      "learning_rate": 9.952815281528153e-05,
      "loss": 3.2314,
      "step": 502410
    },
    {
      "epoch": 0.24594157139875888,
      "grad_norm": 0.77734375,
      "learning_rate": 9.952615261526154e-05,
      "loss": 3.316,
      "step": 502420
    },
    {
      "epoch": 0.24596021463491707,
      "grad_norm": 0.7578125,
      "learning_rate": 9.952415241524154e-05,
      "loss": 3.2665,
      "step": 502430
    },
    {
      "epoch": 0.24597885787107526,
      "grad_norm": 0.75390625,
      "learning_rate": 9.952215221522153e-05,
      "loss": 3.1327,
      "step": 502440
    },
    {
      "epoch": 0.24599750110723345,
      "grad_norm": 0.6875,
      "learning_rate": 9.952015201520153e-05,
      "loss": 3.2052,
      "step": 502450
    },
    {
      "epoch": 0.24601614434339164,
      "grad_norm": 0.83984375,
      "learning_rate": 9.951815181518152e-05,
      "loss": 3.1823,
      "step": 502460
    },
    {
      "epoch": 0.2460347875795498,
      "grad_norm": 0.7578125,
      "learning_rate": 9.951615161516152e-05,
      "loss": 3.2551,
      "step": 502470
    },
    {
      "epoch": 0.246053430815708,
      "grad_norm": 0.68359375,
      "learning_rate": 9.951415141514152e-05,
      "loss": 3.2001,
      "step": 502480
    },
    {
      "epoch": 0.24607207405186618,
      "grad_norm": 0.921875,
      "learning_rate": 9.951215121512153e-05,
      "loss": 3.2886,
      "step": 502490
    },
    {
      "epoch": 0.24609071728802437,
      "grad_norm": 0.69921875,
      "learning_rate": 9.951015101510152e-05,
      "loss": 3.1736,
      "step": 502500
    },
    {
      "epoch": 0.24610936052418256,
      "grad_norm": 0.734375,
      "learning_rate": 9.950815081508152e-05,
      "loss": 3.1895,
      "step": 502510
    },
    {
      "epoch": 0.24612800376034075,
      "grad_norm": 0.78515625,
      "learning_rate": 9.950615061506151e-05,
      "loss": 3.1235,
      "step": 502520
    },
    {
      "epoch": 0.2461466469964989,
      "grad_norm": 0.6640625,
      "learning_rate": 9.950415041504151e-05,
      "loss": 3.2037,
      "step": 502530
    },
    {
      "epoch": 0.2461652902326571,
      "grad_norm": 0.78515625,
      "learning_rate": 9.95021502150215e-05,
      "loss": 3.2232,
      "step": 502540
    },
    {
      "epoch": 0.24618393346881529,
      "grad_norm": 0.7265625,
      "learning_rate": 9.95001500150015e-05,
      "loss": 3.2776,
      "step": 502550
    },
    {
      "epoch": 0.24620257670497347,
      "grad_norm": 0.66015625,
      "learning_rate": 9.94981498149815e-05,
      "loss": 3.2316,
      "step": 502560
    },
    {
      "epoch": 0.24622121994113166,
      "grad_norm": 0.69921875,
      "learning_rate": 9.94961496149615e-05,
      "loss": 3.228,
      "step": 502570
    },
    {
      "epoch": 0.24623986317728982,
      "grad_norm": 0.69921875,
      "learning_rate": 9.94941494149415e-05,
      "loss": 3.1532,
      "step": 502580
    },
    {
      "epoch": 0.24625850641344801,
      "grad_norm": 0.796875,
      "learning_rate": 9.94921492149215e-05,
      "loss": 3.2643,
      "step": 502590
    },
    {
      "epoch": 0.2462771496496062,
      "grad_norm": 0.83203125,
      "learning_rate": 9.949014901490149e-05,
      "loss": 3.2485,
      "step": 502600
    },
    {
      "epoch": 0.2462957928857644,
      "grad_norm": 0.76171875,
      "learning_rate": 9.948814881488149e-05,
      "loss": 3.2058,
      "step": 502610
    },
    {
      "epoch": 0.24631443612192258,
      "grad_norm": 0.71875,
      "learning_rate": 9.94861486148615e-05,
      "loss": 3.0823,
      "step": 502620
    },
    {
      "epoch": 0.24633307935808074,
      "grad_norm": 0.8046875,
      "learning_rate": 9.948414841484149e-05,
      "loss": 3.2918,
      "step": 502630
    },
    {
      "epoch": 0.24635172259423893,
      "grad_norm": 0.65625,
      "learning_rate": 9.948214821482149e-05,
      "loss": 3.1002,
      "step": 502640
    },
    {
      "epoch": 0.24637036583039712,
      "grad_norm": 0.7578125,
      "learning_rate": 9.94801480148015e-05,
      "loss": 3.094,
      "step": 502650
    },
    {
      "epoch": 0.2463890090665553,
      "grad_norm": 0.7109375,
      "learning_rate": 9.947814781478149e-05,
      "loss": 3.2546,
      "step": 502660
    },
    {
      "epoch": 0.2464076523027135,
      "grad_norm": 0.76171875,
      "learning_rate": 9.947614761476149e-05,
      "loss": 3.1951,
      "step": 502670
    },
    {
      "epoch": 0.24642629553887166,
      "grad_norm": 0.62890625,
      "learning_rate": 9.947414741474148e-05,
      "loss": 3.2085,
      "step": 502680
    },
    {
      "epoch": 0.24644493877502985,
      "grad_norm": 0.72265625,
      "learning_rate": 9.947214721472148e-05,
      "loss": 3.2088,
      "step": 502690
    },
    {
      "epoch": 0.24646358201118804,
      "grad_norm": 0.76953125,
      "learning_rate": 9.947014701470147e-05,
      "loss": 3.216,
      "step": 502700
    },
    {
      "epoch": 0.24648222524734623,
      "grad_norm": 0.72265625,
      "learning_rate": 9.946814681468147e-05,
      "loss": 3.2331,
      "step": 502710
    },
    {
      "epoch": 0.24650086848350442,
      "grad_norm": 0.81640625,
      "learning_rate": 9.946614661466146e-05,
      "loss": 3.1955,
      "step": 502720
    },
    {
      "epoch": 0.24651951171966258,
      "grad_norm": 0.90234375,
      "learning_rate": 9.946414641464147e-05,
      "loss": 3.1613,
      "step": 502730
    },
    {
      "epoch": 0.24653815495582077,
      "grad_norm": 0.8515625,
      "learning_rate": 9.946214621462147e-05,
      "loss": 3.2689,
      "step": 502740
    },
    {
      "epoch": 0.24655679819197895,
      "grad_norm": 0.66015625,
      "learning_rate": 9.946014601460146e-05,
      "loss": 3.1279,
      "step": 502750
    },
    {
      "epoch": 0.24657544142813714,
      "grad_norm": 0.765625,
      "learning_rate": 9.945814581458146e-05,
      "loss": 3.1488,
      "step": 502760
    },
    {
      "epoch": 0.24659408466429533,
      "grad_norm": 0.80078125,
      "learning_rate": 9.945614561456146e-05,
      "loss": 3.1695,
      "step": 502770
    },
    {
      "epoch": 0.24661272790045352,
      "grad_norm": 0.73046875,
      "learning_rate": 9.945414541454145e-05,
      "loss": 3.1518,
      "step": 502780
    },
    {
      "epoch": 0.24663137113661168,
      "grad_norm": 0.74609375,
      "learning_rate": 9.945214521452145e-05,
      "loss": 3.2992,
      "step": 502790
    },
    {
      "epoch": 0.24665001437276987,
      "grad_norm": 0.82421875,
      "learning_rate": 9.945014501450146e-05,
      "loss": 3.2325,
      "step": 502800
    },
    {
      "epoch": 0.24666865760892806,
      "grad_norm": 0.68359375,
      "learning_rate": 9.944814481448145e-05,
      "loss": 3.1602,
      "step": 502810
    },
    {
      "epoch": 0.24668730084508625,
      "grad_norm": 0.80859375,
      "learning_rate": 9.944614461446146e-05,
      "loss": 3.2312,
      "step": 502820
    },
    {
      "epoch": 0.24670594408124444,
      "grad_norm": 0.75390625,
      "learning_rate": 9.944414441444146e-05,
      "loss": 3.1653,
      "step": 502830
    },
    {
      "epoch": 0.2467245873174026,
      "grad_norm": 1.2109375,
      "learning_rate": 9.944214421442145e-05,
      "loss": 3.2469,
      "step": 502840
    },
    {
      "epoch": 0.2467432305535608,
      "grad_norm": 0.71875,
      "learning_rate": 9.944014401440145e-05,
      "loss": 3.1604,
      "step": 502850
    },
    {
      "epoch": 0.24676187378971898,
      "grad_norm": 0.7109375,
      "learning_rate": 9.943814381438144e-05,
      "loss": 3.2924,
      "step": 502860
    },
    {
      "epoch": 0.24678051702587717,
      "grad_norm": 0.8203125,
      "learning_rate": 9.943614361436144e-05,
      "loss": 3.2888,
      "step": 502870
    },
    {
      "epoch": 0.24679916026203536,
      "grad_norm": 0.74609375,
      "learning_rate": 9.943414341434143e-05,
      "loss": 3.1502,
      "step": 502880
    },
    {
      "epoch": 0.24681780349819352,
      "grad_norm": 0.75,
      "learning_rate": 9.943214321432144e-05,
      "loss": 3.2595,
      "step": 502890
    },
    {
      "epoch": 0.2468364467343517,
      "grad_norm": 0.828125,
      "learning_rate": 9.943014301430144e-05,
      "loss": 3.2046,
      "step": 502900
    },
    {
      "epoch": 0.2468550899705099,
      "grad_norm": 0.86328125,
      "learning_rate": 9.942814281428143e-05,
      "loss": 3.116,
      "step": 502910
    },
    {
      "epoch": 0.24687373320666808,
      "grad_norm": 0.7421875,
      "learning_rate": 9.942614261426143e-05,
      "loss": 3.3112,
      "step": 502920
    },
    {
      "epoch": 0.24689237644282627,
      "grad_norm": 0.76171875,
      "learning_rate": 9.942414241424143e-05,
      "loss": 3.2417,
      "step": 502930
    },
    {
      "epoch": 0.24691101967898443,
      "grad_norm": 0.7421875,
      "learning_rate": 9.942214221422142e-05,
      "loss": 3.1251,
      "step": 502940
    },
    {
      "epoch": 0.24692966291514262,
      "grad_norm": 0.6875,
      "learning_rate": 9.942014201420142e-05,
      "loss": 3.2243,
      "step": 502950
    },
    {
      "epoch": 0.2469483061513008,
      "grad_norm": 0.7578125,
      "learning_rate": 9.941814181418143e-05,
      "loss": 3.1873,
      "step": 502960
    },
    {
      "epoch": 0.246966949387459,
      "grad_norm": 0.77734375,
      "learning_rate": 9.941614161416142e-05,
      "loss": 3.2233,
      "step": 502970
    },
    {
      "epoch": 0.2469855926236172,
      "grad_norm": 0.73828125,
      "learning_rate": 9.941414141414142e-05,
      "loss": 3.2971,
      "step": 502980
    },
    {
      "epoch": 0.24700423585977538,
      "grad_norm": 0.6796875,
      "learning_rate": 9.941214121412141e-05,
      "loss": 3.2803,
      "step": 502990
    },
    {
      "epoch": 0.24702287909593354,
      "grad_norm": 0.6875,
      "learning_rate": 9.941014101410141e-05,
      "loss": 3.2834,
      "step": 503000
    },
    {
      "epoch": 0.24704152233209173,
      "grad_norm": 0.640625,
      "learning_rate": 9.94081408140814e-05,
      "loss": 3.2637,
      "step": 503010
    },
    {
      "epoch": 0.24706016556824992,
      "grad_norm": 0.69140625,
      "learning_rate": 9.940614061406141e-05,
      "loss": 3.1589,
      "step": 503020
    },
    {
      "epoch": 0.2470788088044081,
      "grad_norm": 0.75390625,
      "learning_rate": 9.940414041404141e-05,
      "loss": 3.1687,
      "step": 503030
    },
    {
      "epoch": 0.2470974520405663,
      "grad_norm": 0.7890625,
      "learning_rate": 9.94021402140214e-05,
      "loss": 3.2226,
      "step": 503040
    },
    {
      "epoch": 0.24711609527672446,
      "grad_norm": 0.8046875,
      "learning_rate": 9.940014001400141e-05,
      "loss": 3.136,
      "step": 503050
    },
    {
      "epoch": 0.24713473851288265,
      "grad_norm": 0.83984375,
      "learning_rate": 9.939813981398141e-05,
      "loss": 3.1554,
      "step": 503060
    },
    {
      "epoch": 0.24715338174904083,
      "grad_norm": 0.66796875,
      "learning_rate": 9.93961396139614e-05,
      "loss": 3.2043,
      "step": 503070
    },
    {
      "epoch": 0.24717202498519902,
      "grad_norm": 0.6953125,
      "learning_rate": 9.93941394139414e-05,
      "loss": 3.1825,
      "step": 503080
    },
    {
      "epoch": 0.2471906682213572,
      "grad_norm": 0.84765625,
      "learning_rate": 9.93921392139214e-05,
      "loss": 3.2638,
      "step": 503090
    },
    {
      "epoch": 0.24720931145751537,
      "grad_norm": 0.71875,
      "learning_rate": 9.939013901390139e-05,
      "loss": 3.1775,
      "step": 503100
    },
    {
      "epoch": 0.24722795469367356,
      "grad_norm": 0.6953125,
      "learning_rate": 9.938813881388139e-05,
      "loss": 3.1798,
      "step": 503110
    },
    {
      "epoch": 0.24724659792983175,
      "grad_norm": 0.6953125,
      "learning_rate": 9.93861386138614e-05,
      "loss": 3.3521,
      "step": 503120
    },
    {
      "epoch": 0.24726524116598994,
      "grad_norm": 0.71875,
      "learning_rate": 9.938413841384139e-05,
      "loss": 3.2514,
      "step": 503130
    },
    {
      "epoch": 0.24728388440214813,
      "grad_norm": 0.75,
      "learning_rate": 9.938213821382139e-05,
      "loss": 3.1513,
      "step": 503140
    },
    {
      "epoch": 0.2473025276383063,
      "grad_norm": 0.71875,
      "learning_rate": 9.938013801380138e-05,
      "loss": 3.1555,
      "step": 503150
    },
    {
      "epoch": 0.24732117087446448,
      "grad_norm": 0.74609375,
      "learning_rate": 9.937813781378138e-05,
      "loss": 3.1056,
      "step": 503160
    },
    {
      "epoch": 0.24733981411062267,
      "grad_norm": 0.6640625,
      "learning_rate": 9.937613761376137e-05,
      "loss": 3.2048,
      "step": 503170
    },
    {
      "epoch": 0.24735845734678086,
      "grad_norm": 0.7578125,
      "learning_rate": 9.937413741374137e-05,
      "loss": 3.2257,
      "step": 503180
    },
    {
      "epoch": 0.24737710058293905,
      "grad_norm": 0.7421875,
      "learning_rate": 9.937213721372137e-05,
      "loss": 3.1265,
      "step": 503190
    },
    {
      "epoch": 0.24739574381909724,
      "grad_norm": 0.63671875,
      "learning_rate": 9.937013701370137e-05,
      "loss": 3.2787,
      "step": 503200
    },
    {
      "epoch": 0.2474143870552554,
      "grad_norm": 0.72265625,
      "learning_rate": 9.936813681368138e-05,
      "loss": 3.3585,
      "step": 503210
    },
    {
      "epoch": 0.24743303029141359,
      "grad_norm": 0.75,
      "learning_rate": 9.936613661366138e-05,
      "loss": 3.17,
      "step": 503220
    },
    {
      "epoch": 0.24745167352757177,
      "grad_norm": 0.8046875,
      "learning_rate": 9.936413641364138e-05,
      "loss": 3.0841,
      "step": 503230
    },
    {
      "epoch": 0.24747031676372996,
      "grad_norm": 0.77734375,
      "learning_rate": 9.936213621362137e-05,
      "loss": 3.2746,
      "step": 503240
    },
    {
      "epoch": 0.24748895999988815,
      "grad_norm": 0.73828125,
      "learning_rate": 9.936013601360137e-05,
      "loss": 3.1436,
      "step": 503250
    },
    {
      "epoch": 0.24750760323604631,
      "grad_norm": 0.80859375,
      "learning_rate": 9.935813581358136e-05,
      "loss": 3.206,
      "step": 503260
    },
    {
      "epoch": 0.2475262464722045,
      "grad_norm": 0.84375,
      "learning_rate": 9.935613561356136e-05,
      "loss": 3.1051,
      "step": 503270
    },
    {
      "epoch": 0.2475448897083627,
      "grad_norm": 0.65625,
      "learning_rate": 9.935413541354137e-05,
      "loss": 3.1264,
      "step": 503280
    },
    {
      "epoch": 0.24756353294452088,
      "grad_norm": 0.7578125,
      "learning_rate": 9.935213521352136e-05,
      "loss": 3.3342,
      "step": 503290
    },
    {
      "epoch": 0.24758217618067907,
      "grad_norm": 0.51171875,
      "learning_rate": 9.935013501350136e-05,
      "loss": 3.234,
      "step": 503300
    },
    {
      "epoch": 0.24760081941683723,
      "grad_norm": 0.81640625,
      "learning_rate": 9.934813481348135e-05,
      "loss": 3.2162,
      "step": 503310
    },
    {
      "epoch": 0.24761946265299542,
      "grad_norm": 0.7578125,
      "learning_rate": 9.934613461346135e-05,
      "loss": 3.3417,
      "step": 503320
    },
    {
      "epoch": 0.2476381058891536,
      "grad_norm": 0.75,
      "learning_rate": 9.934413441344134e-05,
      "loss": 3.2463,
      "step": 503330
    },
    {
      "epoch": 0.2476567491253118,
      "grad_norm": 0.77734375,
      "learning_rate": 9.934213421342134e-05,
      "loss": 3.1464,
      "step": 503340
    },
    {
      "epoch": 0.24767539236147,
      "grad_norm": 0.80078125,
      "learning_rate": 9.934013401340134e-05,
      "loss": 3.1527,
      "step": 503350
    },
    {
      "epoch": 0.24769403559762815,
      "grad_norm": 0.75390625,
      "learning_rate": 9.933813381338135e-05,
      "loss": 3.1415,
      "step": 503360
    },
    {
      "epoch": 0.24771267883378634,
      "grad_norm": 0.8125,
      "learning_rate": 9.933613361336134e-05,
      "loss": 3.1709,
      "step": 503370
    },
    {
      "epoch": 0.24773132206994453,
      "grad_norm": 0.74609375,
      "learning_rate": 9.933413341334134e-05,
      "loss": 3.211,
      "step": 503380
    },
    {
      "epoch": 0.24774996530610272,
      "grad_norm": 0.71875,
      "learning_rate": 9.933213321332133e-05,
      "loss": 3.2371,
      "step": 503390
    },
    {
      "epoch": 0.2477686085422609,
      "grad_norm": 0.75,
      "learning_rate": 9.933013301330133e-05,
      "loss": 3.304,
      "step": 503400
    },
    {
      "epoch": 0.24778725177841907,
      "grad_norm": 0.76953125,
      "learning_rate": 9.932813281328134e-05,
      "loss": 3.1914,
      "step": 503410
    },
    {
      "epoch": 0.24780589501457725,
      "grad_norm": 0.78515625,
      "learning_rate": 9.932613261326133e-05,
      "loss": 3.2386,
      "step": 503420
    },
    {
      "epoch": 0.24782453825073544,
      "grad_norm": 0.84375,
      "learning_rate": 9.932413241324133e-05,
      "loss": 3.2781,
      "step": 503430
    },
    {
      "epoch": 0.24784318148689363,
      "grad_norm": 0.875,
      "learning_rate": 9.932213221322134e-05,
      "loss": 3.3096,
      "step": 503440
    },
    {
      "epoch": 0.24786182472305182,
      "grad_norm": 0.73828125,
      "learning_rate": 9.932013201320133e-05,
      "loss": 3.2259,
      "step": 503450
    },
    {
      "epoch": 0.24788046795921,
      "grad_norm": 0.765625,
      "learning_rate": 9.931813181318133e-05,
      "loss": 3.1859,
      "step": 503460
    },
    {
      "epoch": 0.24789911119536817,
      "grad_norm": 0.765625,
      "learning_rate": 9.931613161316132e-05,
      "loss": 3.2109,
      "step": 503470
    },
    {
      "epoch": 0.24791775443152636,
      "grad_norm": 0.8203125,
      "learning_rate": 9.931413141314132e-05,
      "loss": 3.2041,
      "step": 503480
    },
    {
      "epoch": 0.24793639766768455,
      "grad_norm": 0.765625,
      "learning_rate": 9.931213121312131e-05,
      "loss": 3.0913,
      "step": 503490
    },
    {
      "epoch": 0.24795504090384274,
      "grad_norm": 0.65234375,
      "learning_rate": 9.931013101310131e-05,
      "loss": 3.3024,
      "step": 503500
    },
    {
      "epoch": 0.24797368414000093,
      "grad_norm": 0.78515625,
      "learning_rate": 9.93081308130813e-05,
      "loss": 3.0957,
      "step": 503510
    },
    {
      "epoch": 0.2479923273761591,
      "grad_norm": 0.76171875,
      "learning_rate": 9.930613061306132e-05,
      "loss": 3.1722,
      "step": 503520
    },
    {
      "epoch": 0.24801097061231728,
      "grad_norm": 0.80078125,
      "learning_rate": 9.930413041304131e-05,
      "loss": 3.1706,
      "step": 503530
    },
    {
      "epoch": 0.24802961384847547,
      "grad_norm": 0.70703125,
      "learning_rate": 9.93021302130213e-05,
      "loss": 3.3614,
      "step": 503540
    },
    {
      "epoch": 0.24804825708463366,
      "grad_norm": 0.7734375,
      "learning_rate": 9.93001300130013e-05,
      "loss": 3.1005,
      "step": 503550
    },
    {
      "epoch": 0.24806690032079184,
      "grad_norm": 0.75390625,
      "learning_rate": 9.92981298129813e-05,
      "loss": 3.1609,
      "step": 503560
    },
    {
      "epoch": 0.24808554355695,
      "grad_norm": 0.7109375,
      "learning_rate": 9.92961296129613e-05,
      "loss": 3.1393,
      "step": 503570
    },
    {
      "epoch": 0.2481041867931082,
      "grad_norm": 0.76171875,
      "learning_rate": 9.929412941294129e-05,
      "loss": 3.2764,
      "step": 503580
    },
    {
      "epoch": 0.24812283002926638,
      "grad_norm": 0.6953125,
      "learning_rate": 9.92921292129213e-05,
      "loss": 3.3253,
      "step": 503590
    },
    {
      "epoch": 0.24814147326542457,
      "grad_norm": 0.73828125,
      "learning_rate": 9.92901290129013e-05,
      "loss": 3.241,
      "step": 503600
    },
    {
      "epoch": 0.24816011650158276,
      "grad_norm": 0.8359375,
      "learning_rate": 9.92881288128813e-05,
      "loss": 3.0915,
      "step": 503610
    },
    {
      "epoch": 0.24817875973774092,
      "grad_norm": 0.6875,
      "learning_rate": 9.92861286128613e-05,
      "loss": 3.2026,
      "step": 503620
    },
    {
      "epoch": 0.2481974029738991,
      "grad_norm": 0.68359375,
      "learning_rate": 9.92841284128413e-05,
      "loss": 3.1885,
      "step": 503630
    },
    {
      "epoch": 0.2482160462100573,
      "grad_norm": 0.71484375,
      "learning_rate": 9.928212821282129e-05,
      "loss": 3.2777,
      "step": 503640
    },
    {
      "epoch": 0.2482346894462155,
      "grad_norm": 0.7109375,
      "learning_rate": 9.928012801280129e-05,
      "loss": 3.1606,
      "step": 503650
    },
    {
      "epoch": 0.24825333268237368,
      "grad_norm": 0.90234375,
      "learning_rate": 9.927812781278128e-05,
      "loss": 3.2247,
      "step": 503660
    },
    {
      "epoch": 0.24827197591853187,
      "grad_norm": 0.66015625,
      "learning_rate": 9.927612761276128e-05,
      "loss": 3.1739,
      "step": 503670
    },
    {
      "epoch": 0.24829061915469003,
      "grad_norm": 0.6953125,
      "learning_rate": 9.927412741274129e-05,
      "loss": 3.2444,
      "step": 503680
    },
    {
      "epoch": 0.24830926239084822,
      "grad_norm": 0.79296875,
      "learning_rate": 9.927212721272128e-05,
      "loss": 3.1666,
      "step": 503690
    },
    {
      "epoch": 0.2483279056270064,
      "grad_norm": 0.6796875,
      "learning_rate": 9.927012701270128e-05,
      "loss": 3.1582,
      "step": 503700
    },
    {
      "epoch": 0.2483465488631646,
      "grad_norm": 0.640625,
      "learning_rate": 9.926812681268127e-05,
      "loss": 3.2283,
      "step": 503710
    },
    {
      "epoch": 0.24836519209932278,
      "grad_norm": 0.7421875,
      "learning_rate": 9.926612661266127e-05,
      "loss": 3.0662,
      "step": 503720
    },
    {
      "epoch": 0.24838383533548095,
      "grad_norm": 0.8203125,
      "learning_rate": 9.926412641264126e-05,
      "loss": 3.2098,
      "step": 503730
    },
    {
      "epoch": 0.24840247857163913,
      "grad_norm": 0.75,
      "learning_rate": 9.926212621262126e-05,
      "loss": 3.1742,
      "step": 503740
    },
    {
      "epoch": 0.24842112180779732,
      "grad_norm": 0.6796875,
      "learning_rate": 9.926012601260127e-05,
      "loss": 3.1241,
      "step": 503750
    },
    {
      "epoch": 0.2484397650439555,
      "grad_norm": 0.7734375,
      "learning_rate": 9.925812581258126e-05,
      "loss": 3.2737,
      "step": 503760
    },
    {
      "epoch": 0.2484584082801137,
      "grad_norm": 0.89453125,
      "learning_rate": 9.925612561256126e-05,
      "loss": 3.1299,
      "step": 503770
    },
    {
      "epoch": 0.24847705151627186,
      "grad_norm": 0.734375,
      "learning_rate": 9.925412541254125e-05,
      "loss": 3.2024,
      "step": 503780
    },
    {
      "epoch": 0.24849569475243005,
      "grad_norm": 0.73046875,
      "learning_rate": 9.925212521252125e-05,
      "loss": 3.1829,
      "step": 503790
    },
    {
      "epoch": 0.24851433798858824,
      "grad_norm": 0.6953125,
      "learning_rate": 9.925012501250125e-05,
      "loss": 3.1505,
      "step": 503800
    },
    {
      "epoch": 0.24853298122474643,
      "grad_norm": 0.81640625,
      "learning_rate": 9.924812481248126e-05,
      "loss": 3.3257,
      "step": 503810
    },
    {
      "epoch": 0.24855162446090462,
      "grad_norm": 0.73828125,
      "learning_rate": 9.924612461246125e-05,
      "loss": 3.1419,
      "step": 503820
    },
    {
      "epoch": 0.24857026769706278,
      "grad_norm": 0.84375,
      "learning_rate": 9.924412441244125e-05,
      "loss": 3.1818,
      "step": 503830
    },
    {
      "epoch": 0.24858891093322097,
      "grad_norm": 0.67578125,
      "learning_rate": 9.924212421242126e-05,
      "loss": 3.1441,
      "step": 503840
    },
    {
      "epoch": 0.24860755416937916,
      "grad_norm": 0.63671875,
      "learning_rate": 9.924012401240125e-05,
      "loss": 3.2352,
      "step": 503850
    },
    {
      "epoch": 0.24862619740553735,
      "grad_norm": 0.828125,
      "learning_rate": 9.923812381238125e-05,
      "loss": 3.1046,
      "step": 503860
    },
    {
      "epoch": 0.24864484064169554,
      "grad_norm": 0.66796875,
      "learning_rate": 9.923612361236124e-05,
      "loss": 3.1847,
      "step": 503870
    },
    {
      "epoch": 0.24866348387785372,
      "grad_norm": 0.75390625,
      "learning_rate": 9.923412341234124e-05,
      "loss": 3.2508,
      "step": 503880
    },
    {
      "epoch": 0.24868212711401189,
      "grad_norm": 0.73046875,
      "learning_rate": 9.923212321232123e-05,
      "loss": 3.2385,
      "step": 503890
    },
    {
      "epoch": 0.24870077035017008,
      "grad_norm": 0.73828125,
      "learning_rate": 9.923012301230123e-05,
      "loss": 3.2317,
      "step": 503900
    },
    {
      "epoch": 0.24871941358632826,
      "grad_norm": 0.62890625,
      "learning_rate": 9.922812281228124e-05,
      "loss": 3.1947,
      "step": 503910
    },
    {
      "epoch": 0.24873805682248645,
      "grad_norm": 0.62890625,
      "learning_rate": 9.922612261226123e-05,
      "loss": 3.1106,
      "step": 503920
    },
    {
      "epoch": 0.24875670005864464,
      "grad_norm": 0.69140625,
      "learning_rate": 9.922412241224123e-05,
      "loss": 3.041,
      "step": 503930
    },
    {
      "epoch": 0.2487753432948028,
      "grad_norm": 0.6953125,
      "learning_rate": 9.922212221222123e-05,
      "loss": 3.2001,
      "step": 503940
    },
    {
      "epoch": 0.248793986530961,
      "grad_norm": 0.66015625,
      "learning_rate": 9.922012201220122e-05,
      "loss": 3.2089,
      "step": 503950
    },
    {
      "epoch": 0.24881262976711918,
      "grad_norm": 0.6328125,
      "learning_rate": 9.921812181218122e-05,
      "loss": 3.2292,
      "step": 503960
    },
    {
      "epoch": 0.24883127300327737,
      "grad_norm": 0.6953125,
      "learning_rate": 9.921612161216121e-05,
      "loss": 3.1813,
      "step": 503970
    },
    {
      "epoch": 0.24884991623943556,
      "grad_norm": 0.734375,
      "learning_rate": 9.921412141214122e-05,
      "loss": 3.2243,
      "step": 503980
    },
    {
      "epoch": 0.24886855947559372,
      "grad_norm": 0.71875,
      "learning_rate": 9.921212121212122e-05,
      "loss": 3.1,
      "step": 503990
    },
    {
      "epoch": 0.2488872027117519,
      "grad_norm": 0.67578125,
      "learning_rate": 9.921012101210123e-05,
      "loss": 3.2711,
      "step": 504000
    },
    {
      "epoch": 0.2489058459479101,
      "grad_norm": 0.66015625,
      "learning_rate": 9.920812081208122e-05,
      "loss": 3.2104,
      "step": 504010
    },
    {
      "epoch": 0.2489244891840683,
      "grad_norm": 0.79296875,
      "learning_rate": 9.920612061206122e-05,
      "loss": 3.2515,
      "step": 504020
    },
    {
      "epoch": 0.24894313242022648,
      "grad_norm": 0.80859375,
      "learning_rate": 9.920412041204121e-05,
      "loss": 3.3704,
      "step": 504030
    },
    {
      "epoch": 0.24896177565638464,
      "grad_norm": 0.8046875,
      "learning_rate": 9.920212021202121e-05,
      "loss": 3.1704,
      "step": 504040
    },
    {
      "epoch": 0.24898041889254283,
      "grad_norm": 0.828125,
      "learning_rate": 9.92001200120012e-05,
      "loss": 3.1382,
      "step": 504050
    },
    {
      "epoch": 0.24899906212870102,
      "grad_norm": 0.80859375,
      "learning_rate": 9.91981198119812e-05,
      "loss": 3.084,
      "step": 504060
    },
    {
      "epoch": 0.2490177053648592,
      "grad_norm": 0.671875,
      "learning_rate": 9.919611961196121e-05,
      "loss": 3.1839,
      "step": 504070
    },
    {
      "epoch": 0.2490363486010174,
      "grad_norm": 0.69140625,
      "learning_rate": 9.91941194119412e-05,
      "loss": 3.2432,
      "step": 504080
    },
    {
      "epoch": 0.24905499183717558,
      "grad_norm": 0.73828125,
      "learning_rate": 9.91921192119212e-05,
      "loss": 3.2376,
      "step": 504090
    },
    {
      "epoch": 0.24907363507333374,
      "grad_norm": 0.76953125,
      "learning_rate": 9.91901190119012e-05,
      "loss": 3.1323,
      "step": 504100
    },
    {
      "epoch": 0.24909227830949193,
      "grad_norm": 0.55078125,
      "learning_rate": 9.918811881188119e-05,
      "loss": 3.1396,
      "step": 504110
    },
    {
      "epoch": 0.24911092154565012,
      "grad_norm": 0.68359375,
      "learning_rate": 9.918611861186119e-05,
      "loss": 3.4172,
      "step": 504120
    },
    {
      "epoch": 0.2491295647818083,
      "grad_norm": 0.82421875,
      "learning_rate": 9.918411841184118e-05,
      "loss": 3.1523,
      "step": 504130
    },
    {
      "epoch": 0.2491482080179665,
      "grad_norm": 0.67578125,
      "learning_rate": 9.918211821182119e-05,
      "loss": 3.214,
      "step": 504140
    },
    {
      "epoch": 0.24916685125412466,
      "grad_norm": 0.76171875,
      "learning_rate": 9.918011801180119e-05,
      "loss": 3.2087,
      "step": 504150
    },
    {
      "epoch": 0.24918549449028285,
      "grad_norm": 0.671875,
      "learning_rate": 9.917811781178118e-05,
      "loss": 3.1168,
      "step": 504160
    },
    {
      "epoch": 0.24920413772644104,
      "grad_norm": 0.78125,
      "learning_rate": 9.917611761176118e-05,
      "loss": 3.1383,
      "step": 504170
    },
    {
      "epoch": 0.24922278096259923,
      "grad_norm": 0.66015625,
      "learning_rate": 9.917411741174117e-05,
      "loss": 3.3119,
      "step": 504180
    },
    {
      "epoch": 0.24924142419875742,
      "grad_norm": 0.63671875,
      "learning_rate": 9.917211721172117e-05,
      "loss": 3.2756,
      "step": 504190
    },
    {
      "epoch": 0.24926006743491558,
      "grad_norm": 0.73828125,
      "learning_rate": 9.917011701170118e-05,
      "loss": 3.3339,
      "step": 504200
    },
    {
      "epoch": 0.24927871067107377,
      "grad_norm": 0.71484375,
      "learning_rate": 9.916811681168117e-05,
      "loss": 3.2315,
      "step": 504210
    },
    {
      "epoch": 0.24929735390723196,
      "grad_norm": 0.81640625,
      "learning_rate": 9.916611661166117e-05,
      "loss": 3.1346,
      "step": 504220
    },
    {
      "epoch": 0.24931599714339014,
      "grad_norm": 0.76953125,
      "learning_rate": 9.916411641164118e-05,
      "loss": 3.2738,
      "step": 504230
    },
    {
      "epoch": 0.24933464037954833,
      "grad_norm": 0.75,
      "learning_rate": 9.916211621162117e-05,
      "loss": 3.2546,
      "step": 504240
    },
    {
      "epoch": 0.2493532836157065,
      "grad_norm": 0.78515625,
      "learning_rate": 9.916011601160117e-05,
      "loss": 3.1645,
      "step": 504250
    },
    {
      "epoch": 0.24937192685186468,
      "grad_norm": 0.86328125,
      "learning_rate": 9.915811581158117e-05,
      "loss": 3.1102,
      "step": 504260
    },
    {
      "epoch": 0.24939057008802287,
      "grad_norm": 0.72265625,
      "learning_rate": 9.915611561156116e-05,
      "loss": 3.1621,
      "step": 504270
    },
    {
      "epoch": 0.24940921332418106,
      "grad_norm": 0.69921875,
      "learning_rate": 9.915411541154116e-05,
      "loss": 3.2022,
      "step": 504280
    },
    {
      "epoch": 0.24942785656033925,
      "grad_norm": 0.90234375,
      "learning_rate": 9.915211521152115e-05,
      "loss": 3.2318,
      "step": 504290
    },
    {
      "epoch": 0.2494464997964974,
      "grad_norm": 0.87109375,
      "learning_rate": 9.915011501150115e-05,
      "loss": 3.068,
      "step": 504300
    },
    {
      "epoch": 0.2494651430326556,
      "grad_norm": 0.671875,
      "learning_rate": 9.914811481148116e-05,
      "loss": 3.1717,
      "step": 504310
    },
    {
      "epoch": 0.2494837862688138,
      "grad_norm": 0.7109375,
      "learning_rate": 9.914611461146115e-05,
      "loss": 3.1651,
      "step": 504320
    },
    {
      "epoch": 0.24950242950497198,
      "grad_norm": 0.68359375,
      "learning_rate": 9.914411441144115e-05,
      "loss": 3.0686,
      "step": 504330
    },
    {
      "epoch": 0.24952107274113017,
      "grad_norm": 0.59375,
      "learning_rate": 9.914211421142114e-05,
      "loss": 3.248,
      "step": 504340
    },
    {
      "epoch": 0.24953971597728836,
      "grad_norm": 0.6875,
      "learning_rate": 9.914011401140114e-05,
      "loss": 3.1721,
      "step": 504350
    },
    {
      "epoch": 0.24955835921344652,
      "grad_norm": 0.72265625,
      "learning_rate": 9.913811381138113e-05,
      "loss": 3.1793,
      "step": 504360
    },
    {
      "epoch": 0.2495770024496047,
      "grad_norm": 0.71875,
      "learning_rate": 9.913611361136113e-05,
      "loss": 3.3089,
      "step": 504370
    },
    {
      "epoch": 0.2495956456857629,
      "grad_norm": 0.80859375,
      "learning_rate": 9.913411341134114e-05,
      "loss": 3.1708,
      "step": 504380
    },
    {
      "epoch": 0.24961428892192108,
      "grad_norm": 0.75390625,
      "learning_rate": 9.913211321132114e-05,
      "loss": 3.3155,
      "step": 504390
    },
    {
      "epoch": 0.24963293215807927,
      "grad_norm": 0.7734375,
      "learning_rate": 9.913011301130114e-05,
      "loss": 3.215,
      "step": 504400
    },
    {
      "epoch": 0.24965157539423743,
      "grad_norm": 0.72265625,
      "learning_rate": 9.912811281128114e-05,
      "loss": 3.1629,
      "step": 504410
    },
    {
      "epoch": 0.24967021863039562,
      "grad_norm": 0.68359375,
      "learning_rate": 9.912611261126114e-05,
      "loss": 3.2785,
      "step": 504420
    },
    {
      "epoch": 0.2496888618665538,
      "grad_norm": 0.70703125,
      "learning_rate": 9.912411241124113e-05,
      "loss": 3.3568,
      "step": 504430
    },
    {
      "epoch": 0.249707505102712,
      "grad_norm": 0.8046875,
      "learning_rate": 9.912211221122113e-05,
      "loss": 3.1757,
      "step": 504440
    },
    {
      "epoch": 0.2497261483388702,
      "grad_norm": 0.765625,
      "learning_rate": 9.912011201120112e-05,
      "loss": 3.2558,
      "step": 504450
    },
    {
      "epoch": 0.24974479157502835,
      "grad_norm": 0.75390625,
      "learning_rate": 9.911811181118112e-05,
      "loss": 3.3275,
      "step": 504460
    },
    {
      "epoch": 0.24976343481118654,
      "grad_norm": 0.72265625,
      "learning_rate": 9.911611161116113e-05,
      "loss": 3.1452,
      "step": 504470
    },
    {
      "epoch": 0.24978207804734473,
      "grad_norm": 0.8359375,
      "learning_rate": 9.911411141114112e-05,
      "loss": 3.2084,
      "step": 504480
    },
    {
      "epoch": 0.24980072128350292,
      "grad_norm": 0.6875,
      "learning_rate": 9.911211121112112e-05,
      "loss": 3.3574,
      "step": 504490
    },
    {
      "epoch": 0.2498193645196611,
      "grad_norm": 0.72265625,
      "learning_rate": 9.911011101110111e-05,
      "loss": 3.3381,
      "step": 504500
    },
    {
      "epoch": 0.24983800775581927,
      "grad_norm": 0.6875,
      "learning_rate": 9.910811081108111e-05,
      "loss": 3.1409,
      "step": 504510
    },
    {
      "epoch": 0.24985665099197746,
      "grad_norm": 0.796875,
      "learning_rate": 9.91061106110611e-05,
      "loss": 3.2706,
      "step": 504520
    },
    {
      "epoch": 0.24987529422813565,
      "grad_norm": 0.65625,
      "learning_rate": 9.91041104110411e-05,
      "loss": 3.179,
      "step": 504530
    },
    {
      "epoch": 0.24989393746429384,
      "grad_norm": 0.76953125,
      "learning_rate": 9.910211021102111e-05,
      "loss": 3.2146,
      "step": 504540
    },
    {
      "epoch": 0.24991258070045202,
      "grad_norm": 0.78515625,
      "learning_rate": 9.91001100110011e-05,
      "loss": 3.2275,
      "step": 504550
    },
    {
      "epoch": 0.24993122393661021,
      "grad_norm": 0.7890625,
      "learning_rate": 9.90981098109811e-05,
      "loss": 3.0507,
      "step": 504560
    },
    {
      "epoch": 0.24994986717276838,
      "grad_norm": 0.71484375,
      "learning_rate": 9.90961096109611e-05,
      "loss": 3.1169,
      "step": 504570
    },
    {
      "epoch": 0.24996851040892656,
      "grad_norm": 0.66015625,
      "learning_rate": 9.909410941094109e-05,
      "loss": 3.1398,
      "step": 504580
    },
    {
      "epoch": 0.24998715364508475,
      "grad_norm": 0.66796875,
      "learning_rate": 9.909210921092109e-05,
      "loss": 3.1284,
      "step": 504590
    },
    {
      "epoch": 0.2500057968812429,
      "grad_norm": 0.765625,
      "learning_rate": 9.90901090109011e-05,
      "loss": 3.0384,
      "step": 504600
    },
    {
      "epoch": 0.2500244401174011,
      "grad_norm": 0.7734375,
      "learning_rate": 9.908810881088109e-05,
      "loss": 3.2525,
      "step": 504610
    },
    {
      "epoch": 0.2500430833535593,
      "grad_norm": 0.77734375,
      "learning_rate": 9.908610861086109e-05,
      "loss": 3.2077,
      "step": 504620
    },
    {
      "epoch": 0.2500617265897175,
      "grad_norm": 0.76171875,
      "learning_rate": 9.90841084108411e-05,
      "loss": 3.0987,
      "step": 504630
    },
    {
      "epoch": 0.25008036982587567,
      "grad_norm": 0.78515625,
      "learning_rate": 9.908210821082109e-05,
      "loss": 3.1884,
      "step": 504640
    },
    {
      "epoch": 0.25009901306203386,
      "grad_norm": 0.7265625,
      "learning_rate": 9.908010801080109e-05,
      "loss": 3.228,
      "step": 504650
    },
    {
      "epoch": 0.25011765629819205,
      "grad_norm": 0.796875,
      "learning_rate": 9.907810781078108e-05,
      "loss": 3.2482,
      "step": 504660
    },
    {
      "epoch": 0.25013629953435024,
      "grad_norm": 0.70703125,
      "learning_rate": 9.907610761076108e-05,
      "loss": 3.0824,
      "step": 504670
    },
    {
      "epoch": 0.2501549427705084,
      "grad_norm": 0.71484375,
      "learning_rate": 9.907410741074108e-05,
      "loss": 3.2857,
      "step": 504680
    },
    {
      "epoch": 0.2501735860066666,
      "grad_norm": 0.6484375,
      "learning_rate": 9.907210721072107e-05,
      "loss": 3.1217,
      "step": 504690
    },
    {
      "epoch": 0.25019222924282475,
      "grad_norm": 0.66015625,
      "learning_rate": 9.907010701070108e-05,
      "loss": 3.2725,
      "step": 504700
    },
    {
      "epoch": 0.25021087247898294,
      "grad_norm": 0.73046875,
      "learning_rate": 9.906810681068108e-05,
      "loss": 3.2663,
      "step": 504710
    },
    {
      "epoch": 0.2502295157151411,
      "grad_norm": 0.82421875,
      "learning_rate": 9.906610661066107e-05,
      "loss": 3.2377,
      "step": 504720
    },
    {
      "epoch": 0.2502481589512993,
      "grad_norm": 0.73828125,
      "learning_rate": 9.906410641064107e-05,
      "loss": 3.0698,
      "step": 504730
    },
    {
      "epoch": 0.2502668021874575,
      "grad_norm": 0.83203125,
      "learning_rate": 9.906210621062106e-05,
      "loss": 3.1496,
      "step": 504740
    },
    {
      "epoch": 0.2502854454236157,
      "grad_norm": 0.76171875,
      "learning_rate": 9.906010601060106e-05,
      "loss": 3.2479,
      "step": 504750
    },
    {
      "epoch": 0.2503040886597739,
      "grad_norm": 0.75390625,
      "learning_rate": 9.905810581058105e-05,
      "loss": 3.2853,
      "step": 504760
    },
    {
      "epoch": 0.25032273189593207,
      "grad_norm": 0.87109375,
      "learning_rate": 9.905610561056106e-05,
      "loss": 3.0525,
      "step": 504770
    },
    {
      "epoch": 0.25034137513209026,
      "grad_norm": 0.77734375,
      "learning_rate": 9.905410541054106e-05,
      "loss": 3.2988,
      "step": 504780
    },
    {
      "epoch": 0.25036001836824845,
      "grad_norm": 0.76171875,
      "learning_rate": 9.905210521052107e-05,
      "loss": 3.2215,
      "step": 504790
    },
    {
      "epoch": 0.2503786616044066,
      "grad_norm": 0.75,
      "learning_rate": 9.905010501050106e-05,
      "loss": 3.1178,
      "step": 504800
    },
    {
      "epoch": 0.25039730484056477,
      "grad_norm": 0.734375,
      "learning_rate": 9.904810481048106e-05,
      "loss": 3.2504,
      "step": 504810
    },
    {
      "epoch": 0.25041594807672296,
      "grad_norm": 0.71484375,
      "learning_rate": 9.904610461046105e-05,
      "loss": 3.1418,
      "step": 504820
    },
    {
      "epoch": 0.25043459131288115,
      "grad_norm": 0.77734375,
      "learning_rate": 9.904410441044105e-05,
      "loss": 3.1685,
      "step": 504830
    },
    {
      "epoch": 0.25045323454903934,
      "grad_norm": 0.7109375,
      "learning_rate": 9.904210421042105e-05,
      "loss": 3.1651,
      "step": 504840
    },
    {
      "epoch": 0.2504718777851975,
      "grad_norm": 0.7421875,
      "learning_rate": 9.904010401040104e-05,
      "loss": 3.1459,
      "step": 504850
    },
    {
      "epoch": 0.2504905210213557,
      "grad_norm": 0.71484375,
      "learning_rate": 9.903810381038105e-05,
      "loss": 3.1976,
      "step": 504860
    },
    {
      "epoch": 0.2505091642575139,
      "grad_norm": 0.734375,
      "learning_rate": 9.903610361036105e-05,
      "loss": 3.1962,
      "step": 504870
    },
    {
      "epoch": 0.2505278074936721,
      "grad_norm": 0.75390625,
      "learning_rate": 9.903410341034104e-05,
      "loss": 3.2459,
      "step": 504880
    },
    {
      "epoch": 0.2505464507298303,
      "grad_norm": 0.82421875,
      "learning_rate": 9.903210321032104e-05,
      "loss": 3.0314,
      "step": 504890
    },
    {
      "epoch": 0.2505650939659885,
      "grad_norm": 0.7265625,
      "learning_rate": 9.903010301030103e-05,
      "loss": 3.2647,
      "step": 504900
    },
    {
      "epoch": 0.2505837372021466,
      "grad_norm": 0.66796875,
      "learning_rate": 9.902810281028103e-05,
      "loss": 3.2498,
      "step": 504910
    },
    {
      "epoch": 0.2506023804383048,
      "grad_norm": 0.71875,
      "learning_rate": 9.902610261026102e-05,
      "loss": 3.2727,
      "step": 504920
    },
    {
      "epoch": 0.250621023674463,
      "grad_norm": 0.75,
      "learning_rate": 9.902410241024103e-05,
      "loss": 3.1739,
      "step": 504930
    },
    {
      "epoch": 0.2506396669106212,
      "grad_norm": 0.765625,
      "learning_rate": 9.902210221022103e-05,
      "loss": 3.0801,
      "step": 504940
    },
    {
      "epoch": 0.25065831014677936,
      "grad_norm": 0.80078125,
      "learning_rate": 9.902010201020102e-05,
      "loss": 3.095,
      "step": 504950
    },
    {
      "epoch": 0.25067695338293755,
      "grad_norm": 0.8359375,
      "learning_rate": 9.901810181018102e-05,
      "loss": 3.1813,
      "step": 504960
    },
    {
      "epoch": 0.25069559661909574,
      "grad_norm": 0.8359375,
      "learning_rate": 9.901610161016102e-05,
      "loss": 3.3177,
      "step": 504970
    },
    {
      "epoch": 0.25071423985525393,
      "grad_norm": 0.73046875,
      "learning_rate": 9.901410141014101e-05,
      "loss": 3.2107,
      "step": 504980
    },
    {
      "epoch": 0.2507328830914121,
      "grad_norm": 0.75,
      "learning_rate": 9.901210121012102e-05,
      "loss": 3.1437,
      "step": 504990
    },
    {
      "epoch": 0.2507515263275703,
      "grad_norm": 0.76171875,
      "learning_rate": 9.901010101010102e-05,
      "loss": 3.2117,
      "step": 505000
    },
    {
      "epoch": 0.25077016956372844,
      "grad_norm": 0.75,
      "learning_rate": 9.900810081008101e-05,
      "loss": 3.1233,
      "step": 505010
    },
    {
      "epoch": 0.25078881279988663,
      "grad_norm": 0.7109375,
      "learning_rate": 9.900610061006102e-05,
      "loss": 3.1769,
      "step": 505020
    },
    {
      "epoch": 0.2508074560360448,
      "grad_norm": 0.76953125,
      "learning_rate": 9.900410041004102e-05,
      "loss": 3.2008,
      "step": 505030
    },
    {
      "epoch": 0.250826099272203,
      "grad_norm": 0.7265625,
      "learning_rate": 9.900210021002101e-05,
      "loss": 3.1698,
      "step": 505040
    },
    {
      "epoch": 0.2508447425083612,
      "grad_norm": 0.77734375,
      "learning_rate": 9.900010001000101e-05,
      "loss": 3.2036,
      "step": 505050
    },
    {
      "epoch": 0.2508633857445194,
      "grad_norm": 0.83984375,
      "learning_rate": 9.8998099809981e-05,
      "loss": 3.2542,
      "step": 505060
    },
    {
      "epoch": 0.2508820289806776,
      "grad_norm": 0.70703125,
      "learning_rate": 9.8996099609961e-05,
      "loss": 3.3721,
      "step": 505070
    },
    {
      "epoch": 0.25090067221683576,
      "grad_norm": 0.6484375,
      "learning_rate": 9.8994099409941e-05,
      "loss": 3.2292,
      "step": 505080
    },
    {
      "epoch": 0.25091931545299395,
      "grad_norm": 0.7890625,
      "learning_rate": 9.8992099209921e-05,
      "loss": 3.2022,
      "step": 505090
    },
    {
      "epoch": 0.25093795868915214,
      "grad_norm": 0.7578125,
      "learning_rate": 9.8990099009901e-05,
      "loss": 3.0835,
      "step": 505100
    },
    {
      "epoch": 0.25095660192531033,
      "grad_norm": 0.828125,
      "learning_rate": 9.8988098809881e-05,
      "loss": 3.1915,
      "step": 505110
    },
    {
      "epoch": 0.25097524516146846,
      "grad_norm": 0.7265625,
      "learning_rate": 9.898609860986099e-05,
      "loss": 3.287,
      "step": 505120
    },
    {
      "epoch": 0.25099388839762665,
      "grad_norm": 0.8359375,
      "learning_rate": 9.898409840984099e-05,
      "loss": 3.107,
      "step": 505130
    },
    {
      "epoch": 0.25101253163378484,
      "grad_norm": 0.81640625,
      "learning_rate": 9.898209820982098e-05,
      "loss": 3.1789,
      "step": 505140
    },
    {
      "epoch": 0.25103117486994303,
      "grad_norm": 0.6875,
      "learning_rate": 9.898009800980098e-05,
      "loss": 3.2353,
      "step": 505150
    },
    {
      "epoch": 0.2510498181061012,
      "grad_norm": 0.78515625,
      "learning_rate": 9.897809780978097e-05,
      "loss": 3.2416,
      "step": 505160
    },
    {
      "epoch": 0.2510684613422594,
      "grad_norm": 0.671875,
      "learning_rate": 9.897609760976098e-05,
      "loss": 3.1574,
      "step": 505170
    },
    {
      "epoch": 0.2510871045784176,
      "grad_norm": 0.73046875,
      "learning_rate": 9.897409740974098e-05,
      "loss": 3.3023,
      "step": 505180
    },
    {
      "epoch": 0.2511057478145758,
      "grad_norm": 0.72265625,
      "learning_rate": 9.897209720972099e-05,
      "loss": 3.1487,
      "step": 505190
    },
    {
      "epoch": 0.251124391050734,
      "grad_norm": 0.80078125,
      "learning_rate": 9.897009700970098e-05,
      "loss": 3.1157,
      "step": 505200
    },
    {
      "epoch": 0.25114303428689216,
      "grad_norm": 0.7421875,
      "learning_rate": 9.896809680968098e-05,
      "loss": 3.2375,
      "step": 505210
    },
    {
      "epoch": 0.2511616775230503,
      "grad_norm": 0.7890625,
      "learning_rate": 9.896609660966097e-05,
      "loss": 3.0471,
      "step": 505220
    },
    {
      "epoch": 0.2511803207592085,
      "grad_norm": 0.84765625,
      "learning_rate": 9.896409640964097e-05,
      "loss": 3.1266,
      "step": 505230
    },
    {
      "epoch": 0.2511989639953667,
      "grad_norm": 0.76171875,
      "learning_rate": 9.896209620962096e-05,
      "loss": 3.185,
      "step": 505240
    },
    {
      "epoch": 0.25121760723152486,
      "grad_norm": 0.8125,
      "learning_rate": 9.896009600960097e-05,
      "loss": 3.3833,
      "step": 505250
    },
    {
      "epoch": 0.25123625046768305,
      "grad_norm": 0.8125,
      "learning_rate": 9.895809580958097e-05,
      "loss": 3.1621,
      "step": 505260
    },
    {
      "epoch": 0.25125489370384124,
      "grad_norm": 0.68359375,
      "learning_rate": 9.895609560956096e-05,
      "loss": 3.1756,
      "step": 505270
    },
    {
      "epoch": 0.25127353693999943,
      "grad_norm": 0.77734375,
      "learning_rate": 9.895409540954096e-05,
      "loss": 3.2373,
      "step": 505280
    },
    {
      "epoch": 0.2512921801761576,
      "grad_norm": 0.7421875,
      "learning_rate": 9.895209520952096e-05,
      "loss": 3.1965,
      "step": 505290
    },
    {
      "epoch": 0.2513108234123158,
      "grad_norm": 0.890625,
      "learning_rate": 9.895009500950095e-05,
      "loss": 3.1272,
      "step": 505300
    },
    {
      "epoch": 0.251329466648474,
      "grad_norm": 0.703125,
      "learning_rate": 9.894809480948095e-05,
      "loss": 3.1356,
      "step": 505310
    },
    {
      "epoch": 0.25134810988463213,
      "grad_norm": 0.7890625,
      "learning_rate": 9.894609460946094e-05,
      "loss": 3.2219,
      "step": 505320
    },
    {
      "epoch": 0.2513667531207903,
      "grad_norm": 0.7734375,
      "learning_rate": 9.894409440944095e-05,
      "loss": 3.1928,
      "step": 505330
    },
    {
      "epoch": 0.2513853963569485,
      "grad_norm": 0.7265625,
      "learning_rate": 9.894209420942095e-05,
      "loss": 3.1625,
      "step": 505340
    },
    {
      "epoch": 0.2514040395931067,
      "grad_norm": 0.70703125,
      "learning_rate": 9.894009400940094e-05,
      "loss": 3.1036,
      "step": 505350
    },
    {
      "epoch": 0.2514226828292649,
      "grad_norm": 0.7734375,
      "learning_rate": 9.893809380938094e-05,
      "loss": 3.1358,
      "step": 505360
    },
    {
      "epoch": 0.2514413260654231,
      "grad_norm": 0.703125,
      "learning_rate": 9.893609360936093e-05,
      "loss": 3.1978,
      "step": 505370
    },
    {
      "epoch": 0.25145996930158127,
      "grad_norm": 0.6328125,
      "learning_rate": 9.893409340934093e-05,
      "loss": 3.2093,
      "step": 505380
    },
    {
      "epoch": 0.25147861253773945,
      "grad_norm": 0.82421875,
      "learning_rate": 9.893209320932094e-05,
      "loss": 3.1807,
      "step": 505390
    },
    {
      "epoch": 0.25149725577389764,
      "grad_norm": 0.72265625,
      "learning_rate": 9.893009300930093e-05,
      "loss": 3.2724,
      "step": 505400
    },
    {
      "epoch": 0.25151589901005583,
      "grad_norm": 0.6328125,
      "learning_rate": 9.892809280928094e-05,
      "loss": 3.3529,
      "step": 505410
    },
    {
      "epoch": 0.251534542246214,
      "grad_norm": 0.73828125,
      "learning_rate": 9.892609260926094e-05,
      "loss": 3.16,
      "step": 505420
    },
    {
      "epoch": 0.25155318548237215,
      "grad_norm": 0.734375,
      "learning_rate": 9.892409240924093e-05,
      "loss": 3.3509,
      "step": 505430
    },
    {
      "epoch": 0.25157182871853034,
      "grad_norm": 0.72265625,
      "learning_rate": 9.892209220922093e-05,
      "loss": 3.0925,
      "step": 505440
    },
    {
      "epoch": 0.25159047195468853,
      "grad_norm": 0.734375,
      "learning_rate": 9.892009200920093e-05,
      "loss": 3.0777,
      "step": 505450
    },
    {
      "epoch": 0.2516091151908467,
      "grad_norm": 0.72265625,
      "learning_rate": 9.891809180918092e-05,
      "loss": 3.0951,
      "step": 505460
    },
    {
      "epoch": 0.2516277584270049,
      "grad_norm": 0.7734375,
      "learning_rate": 9.891609160916092e-05,
      "loss": 3.1621,
      "step": 505470
    },
    {
      "epoch": 0.2516464016631631,
      "grad_norm": 0.76171875,
      "learning_rate": 9.891409140914091e-05,
      "loss": 3.152,
      "step": 505480
    },
    {
      "epoch": 0.2516650448993213,
      "grad_norm": 0.734375,
      "learning_rate": 9.891209120912092e-05,
      "loss": 3.1713,
      "step": 505490
    },
    {
      "epoch": 0.2516836881354795,
      "grad_norm": 0.7421875,
      "learning_rate": 9.891009100910092e-05,
      "loss": 3.1228,
      "step": 505500
    },
    {
      "epoch": 0.25170233137163767,
      "grad_norm": 0.765625,
      "learning_rate": 9.890809080908091e-05,
      "loss": 3.1868,
      "step": 505510
    },
    {
      "epoch": 0.25172097460779586,
      "grad_norm": 0.87890625,
      "learning_rate": 9.890609060906091e-05,
      "loss": 3.2576,
      "step": 505520
    },
    {
      "epoch": 0.251739617843954,
      "grad_norm": 0.68359375,
      "learning_rate": 9.89040904090409e-05,
      "loss": 3.098,
      "step": 505530
    },
    {
      "epoch": 0.2517582610801122,
      "grad_norm": 1.515625,
      "learning_rate": 9.89020902090209e-05,
      "loss": 3.2315,
      "step": 505540
    },
    {
      "epoch": 0.25177690431627037,
      "grad_norm": 0.75,
      "learning_rate": 9.89000900090009e-05,
      "loss": 3.0718,
      "step": 505550
    },
    {
      "epoch": 0.25179554755242856,
      "grad_norm": 0.8515625,
      "learning_rate": 9.88980898089809e-05,
      "loss": 3.0389,
      "step": 505560
    },
    {
      "epoch": 0.25181419078858674,
      "grad_norm": 0.79296875,
      "learning_rate": 9.88960896089609e-05,
      "loss": 3.2722,
      "step": 505570
    },
    {
      "epoch": 0.25183283402474493,
      "grad_norm": 0.734375,
      "learning_rate": 9.889408940894091e-05,
      "loss": 3.2065,
      "step": 505580
    },
    {
      "epoch": 0.2518514772609031,
      "grad_norm": 0.7578125,
      "learning_rate": 9.88920892089209e-05,
      "loss": 3.1972,
      "step": 505590
    },
    {
      "epoch": 0.2518701204970613,
      "grad_norm": 0.73828125,
      "learning_rate": 9.88900890089009e-05,
      "loss": 3.1657,
      "step": 505600
    },
    {
      "epoch": 0.2518887637332195,
      "grad_norm": 0.6953125,
      "learning_rate": 9.88880888088809e-05,
      "loss": 3.2089,
      "step": 505610
    },
    {
      "epoch": 0.2519074069693777,
      "grad_norm": 0.8203125,
      "learning_rate": 9.888608860886089e-05,
      "loss": 3.1921,
      "step": 505620
    },
    {
      "epoch": 0.2519260502055359,
      "grad_norm": 0.7734375,
      "learning_rate": 9.888408840884089e-05,
      "loss": 3.267,
      "step": 505630
    },
    {
      "epoch": 0.251944693441694,
      "grad_norm": 0.765625,
      "learning_rate": 9.888208820882088e-05,
      "loss": 3.298,
      "step": 505640
    },
    {
      "epoch": 0.2519633366778522,
      "grad_norm": 0.75,
      "learning_rate": 9.888008800880089e-05,
      "loss": 3.2262,
      "step": 505650
    },
    {
      "epoch": 0.2519819799140104,
      "grad_norm": 0.76953125,
      "learning_rate": 9.887808780878089e-05,
      "loss": 3.2218,
      "step": 505660
    },
    {
      "epoch": 0.2520006231501686,
      "grad_norm": 0.7421875,
      "learning_rate": 9.887608760876088e-05,
      "loss": 3.2361,
      "step": 505670
    },
    {
      "epoch": 0.25201926638632677,
      "grad_norm": 0.7265625,
      "learning_rate": 9.887408740874088e-05,
      "loss": 3.209,
      "step": 505680
    },
    {
      "epoch": 0.25203790962248496,
      "grad_norm": 0.734375,
      "learning_rate": 9.887208720872087e-05,
      "loss": 3.2361,
      "step": 505690
    },
    {
      "epoch": 0.25205655285864315,
      "grad_norm": 0.83203125,
      "learning_rate": 9.887008700870087e-05,
      "loss": 3.2366,
      "step": 505700
    },
    {
      "epoch": 0.25207519609480133,
      "grad_norm": 0.78125,
      "learning_rate": 9.886808680868087e-05,
      "loss": 3.1881,
      "step": 505710
    },
    {
      "epoch": 0.2520938393309595,
      "grad_norm": 0.671875,
      "learning_rate": 9.886608660866087e-05,
      "loss": 3.2456,
      "step": 505720
    },
    {
      "epoch": 0.2521124825671177,
      "grad_norm": 0.83203125,
      "learning_rate": 9.886408640864087e-05,
      "loss": 3.2204,
      "step": 505730
    },
    {
      "epoch": 0.25213112580327585,
      "grad_norm": 0.8125,
      "learning_rate": 9.886208620862087e-05,
      "loss": 3.2258,
      "step": 505740
    },
    {
      "epoch": 0.25214976903943404,
      "grad_norm": 0.63671875,
      "learning_rate": 9.886008600860086e-05,
      "loss": 3.2654,
      "step": 505750
    },
    {
      "epoch": 0.2521684122755922,
      "grad_norm": 0.75390625,
      "learning_rate": 9.885808580858086e-05,
      "loss": 3.1966,
      "step": 505760
    },
    {
      "epoch": 0.2521870555117504,
      "grad_norm": 0.76171875,
      "learning_rate": 9.885608560856085e-05,
      "loss": 3.2811,
      "step": 505770
    },
    {
      "epoch": 0.2522056987479086,
      "grad_norm": 0.78125,
      "learning_rate": 9.885408540854086e-05,
      "loss": 3.2229,
      "step": 505780
    },
    {
      "epoch": 0.2522243419840668,
      "grad_norm": 0.76171875,
      "learning_rate": 9.885208520852086e-05,
      "loss": 3.1938,
      "step": 505790
    },
    {
      "epoch": 0.252242985220225,
      "grad_norm": 0.68359375,
      "learning_rate": 9.885008500850085e-05,
      "loss": 3.2864,
      "step": 505800
    },
    {
      "epoch": 0.25226162845638317,
      "grad_norm": 0.71484375,
      "learning_rate": 9.884808480848086e-05,
      "loss": 3.2252,
      "step": 505810
    },
    {
      "epoch": 0.25228027169254136,
      "grad_norm": 0.8828125,
      "learning_rate": 9.884608460846086e-05,
      "loss": 3.219,
      "step": 505820
    },
    {
      "epoch": 0.25229891492869955,
      "grad_norm": 0.671875,
      "learning_rate": 9.884408440844085e-05,
      "loss": 3.2258,
      "step": 505830
    },
    {
      "epoch": 0.25231755816485774,
      "grad_norm": 0.74609375,
      "learning_rate": 9.884208420842085e-05,
      "loss": 3.2193,
      "step": 505840
    },
    {
      "epoch": 0.25233620140101587,
      "grad_norm": 0.72265625,
      "learning_rate": 9.884008400840084e-05,
      "loss": 3.3083,
      "step": 505850
    },
    {
      "epoch": 0.25235484463717406,
      "grad_norm": 0.77734375,
      "learning_rate": 9.883808380838084e-05,
      "loss": 3.2453,
      "step": 505860
    },
    {
      "epoch": 0.25237348787333225,
      "grad_norm": 0.7734375,
      "learning_rate": 9.883608360836084e-05,
      "loss": 3.2225,
      "step": 505870
    },
    {
      "epoch": 0.25239213110949044,
      "grad_norm": 0.83203125,
      "learning_rate": 9.883408340834084e-05,
      "loss": 3.1267,
      "step": 505880
    },
    {
      "epoch": 0.2524107743456486,
      "grad_norm": 0.83203125,
      "learning_rate": 9.883208320832084e-05,
      "loss": 3.1793,
      "step": 505890
    },
    {
      "epoch": 0.2524294175818068,
      "grad_norm": 0.7265625,
      "learning_rate": 9.883008300830084e-05,
      "loss": 3.2627,
      "step": 505900
    },
    {
      "epoch": 0.252448060817965,
      "grad_norm": 0.62109375,
      "learning_rate": 9.882808280828083e-05,
      "loss": 3.1601,
      "step": 505910
    },
    {
      "epoch": 0.2524667040541232,
      "grad_norm": 0.78125,
      "learning_rate": 9.882608260826083e-05,
      "loss": 3.2258,
      "step": 505920
    },
    {
      "epoch": 0.2524853472902814,
      "grad_norm": 0.75390625,
      "learning_rate": 9.882408240824082e-05,
      "loss": 3.3321,
      "step": 505930
    },
    {
      "epoch": 0.25250399052643957,
      "grad_norm": 0.75,
      "learning_rate": 9.882208220822082e-05,
      "loss": 3.1982,
      "step": 505940
    },
    {
      "epoch": 0.2525226337625977,
      "grad_norm": 0.82421875,
      "learning_rate": 9.882008200820081e-05,
      "loss": 3.2108,
      "step": 505950
    },
    {
      "epoch": 0.2525412769987559,
      "grad_norm": 0.72265625,
      "learning_rate": 9.881808180818082e-05,
      "loss": 3.235,
      "step": 505960
    },
    {
      "epoch": 0.2525599202349141,
      "grad_norm": 0.82421875,
      "learning_rate": 9.881608160816082e-05,
      "loss": 3.1207,
      "step": 505970
    },
    {
      "epoch": 0.25257856347107227,
      "grad_norm": 0.66796875,
      "learning_rate": 9.881408140814083e-05,
      "loss": 3.1969,
      "step": 505980
    },
    {
      "epoch": 0.25259720670723046,
      "grad_norm": 0.72265625,
      "learning_rate": 9.881208120812082e-05,
      "loss": 3.0948,
      "step": 505990
    },
    {
      "epoch": 0.25261584994338865,
      "grad_norm": 0.7265625,
      "learning_rate": 9.881008100810082e-05,
      "loss": 3.2431,
      "step": 506000
    },
    {
      "epoch": 0.25263449317954684,
      "grad_norm": 0.8828125,
      "learning_rate": 9.880808080808081e-05,
      "loss": 3.2126,
      "step": 506010
    },
    {
      "epoch": 0.252653136415705,
      "grad_norm": 0.7578125,
      "learning_rate": 9.880608060806081e-05,
      "loss": 3.2131,
      "step": 506020
    },
    {
      "epoch": 0.2526717796518632,
      "grad_norm": 0.84765625,
      "learning_rate": 9.88040804080408e-05,
      "loss": 3.2095,
      "step": 506030
    },
    {
      "epoch": 0.2526904228880214,
      "grad_norm": 0.6015625,
      "learning_rate": 9.880208020802081e-05,
      "loss": 3.2136,
      "step": 506040
    },
    {
      "epoch": 0.2527090661241796,
      "grad_norm": 0.765625,
      "learning_rate": 9.880008000800081e-05,
      "loss": 3.2926,
      "step": 506050
    },
    {
      "epoch": 0.2527277093603377,
      "grad_norm": 0.89453125,
      "learning_rate": 9.87980798079808e-05,
      "loss": 3.2347,
      "step": 506060
    },
    {
      "epoch": 0.2527463525964959,
      "grad_norm": 0.6796875,
      "learning_rate": 9.87960796079608e-05,
      "loss": 3.2976,
      "step": 506070
    },
    {
      "epoch": 0.2527649958326541,
      "grad_norm": 0.7265625,
      "learning_rate": 9.87940794079408e-05,
      "loss": 3.2554,
      "step": 506080
    },
    {
      "epoch": 0.2527836390688123,
      "grad_norm": 0.6953125,
      "learning_rate": 9.879207920792079e-05,
      "loss": 3.0943,
      "step": 506090
    },
    {
      "epoch": 0.2528022823049705,
      "grad_norm": 0.734375,
      "learning_rate": 9.879007900790079e-05,
      "loss": 3.1052,
      "step": 506100
    },
    {
      "epoch": 0.25282092554112867,
      "grad_norm": 0.8359375,
      "learning_rate": 9.878807880788078e-05,
      "loss": 3.2176,
      "step": 506110
    },
    {
      "epoch": 0.25283956877728686,
      "grad_norm": 0.7890625,
      "learning_rate": 9.878607860786079e-05,
      "loss": 3.1132,
      "step": 506120
    },
    {
      "epoch": 0.25285821201344505,
      "grad_norm": 0.6875,
      "learning_rate": 9.878407840784079e-05,
      "loss": 3.3211,
      "step": 506130
    },
    {
      "epoch": 0.25287685524960324,
      "grad_norm": 0.76953125,
      "learning_rate": 9.878207820782078e-05,
      "loss": 3.1383,
      "step": 506140
    },
    {
      "epoch": 0.2528954984857614,
      "grad_norm": 0.76953125,
      "learning_rate": 9.878007800780078e-05,
      "loss": 3.2742,
      "step": 506150
    },
    {
      "epoch": 0.25291414172191956,
      "grad_norm": 0.72265625,
      "learning_rate": 9.877807780778078e-05,
      "loss": 3.1515,
      "step": 506160
    },
    {
      "epoch": 0.25293278495807775,
      "grad_norm": 0.82421875,
      "learning_rate": 9.877607760776078e-05,
      "loss": 3.3315,
      "step": 506170
    },
    {
      "epoch": 0.25295142819423594,
      "grad_norm": 0.6640625,
      "learning_rate": 9.877407740774078e-05,
      "loss": 3.3029,
      "step": 506180
    },
    {
      "epoch": 0.25297007143039413,
      "grad_norm": 0.7109375,
      "learning_rate": 9.877207720772078e-05,
      "loss": 3.2699,
      "step": 506190
    },
    {
      "epoch": 0.2529887146665523,
      "grad_norm": 0.73828125,
      "learning_rate": 9.877007700770079e-05,
      "loss": 3.1207,
      "step": 506200
    },
    {
      "epoch": 0.2530073579027105,
      "grad_norm": 0.8359375,
      "learning_rate": 9.876807680768078e-05,
      "loss": 3.2332,
      "step": 506210
    },
    {
      "epoch": 0.2530260011388687,
      "grad_norm": 0.7578125,
      "learning_rate": 9.876607660766078e-05,
      "loss": 3.2464,
      "step": 506220
    },
    {
      "epoch": 0.2530446443750269,
      "grad_norm": 0.83984375,
      "learning_rate": 9.876407640764077e-05,
      "loss": 3.1402,
      "step": 506230
    },
    {
      "epoch": 0.2530632876111851,
      "grad_norm": 0.7734375,
      "learning_rate": 9.876207620762077e-05,
      "loss": 3.3515,
      "step": 506240
    },
    {
      "epoch": 0.25308193084734326,
      "grad_norm": 0.66796875,
      "learning_rate": 9.876007600760076e-05,
      "loss": 3.2782,
      "step": 506250
    },
    {
      "epoch": 0.25310057408350145,
      "grad_norm": 0.72265625,
      "learning_rate": 9.875807580758076e-05,
      "loss": 3.1432,
      "step": 506260
    },
    {
      "epoch": 0.2531192173196596,
      "grad_norm": 0.79296875,
      "learning_rate": 9.875607560756075e-05,
      "loss": 3.127,
      "step": 506270
    },
    {
      "epoch": 0.2531378605558178,
      "grad_norm": 0.703125,
      "learning_rate": 9.875407540754076e-05,
      "loss": 3.1377,
      "step": 506280
    },
    {
      "epoch": 0.25315650379197596,
      "grad_norm": 0.85546875,
      "learning_rate": 9.875207520752076e-05,
      "loss": 3.2573,
      "step": 506290
    },
    {
      "epoch": 0.25317514702813415,
      "grad_norm": 0.67578125,
      "learning_rate": 9.875007500750075e-05,
      "loss": 3.3221,
      "step": 506300
    },
    {
      "epoch": 0.25319379026429234,
      "grad_norm": 0.828125,
      "learning_rate": 9.874807480748075e-05,
      "loss": 3.2772,
      "step": 506310
    },
    {
      "epoch": 0.25321243350045053,
      "grad_norm": 0.74609375,
      "learning_rate": 9.874607460746075e-05,
      "loss": 3.3389,
      "step": 506320
    },
    {
      "epoch": 0.2532310767366087,
      "grad_norm": 0.6328125,
      "learning_rate": 9.874407440744074e-05,
      "loss": 3.1992,
      "step": 506330
    },
    {
      "epoch": 0.2532497199727669,
      "grad_norm": 0.79296875,
      "learning_rate": 9.874207420742074e-05,
      "loss": 3.1506,
      "step": 506340
    },
    {
      "epoch": 0.2532683632089251,
      "grad_norm": 0.7890625,
      "learning_rate": 9.874007400740075e-05,
      "loss": 3.2149,
      "step": 506350
    },
    {
      "epoch": 0.2532870064450833,
      "grad_norm": 0.72265625,
      "learning_rate": 9.873807380738074e-05,
      "loss": 3.2693,
      "step": 506360
    },
    {
      "epoch": 0.2533056496812414,
      "grad_norm": 0.703125,
      "learning_rate": 9.873607360736075e-05,
      "loss": 3.2693,
      "step": 506370
    },
    {
      "epoch": 0.2533242929173996,
      "grad_norm": 0.671875,
      "learning_rate": 9.873407340734075e-05,
      "loss": 3.3973,
      "step": 506380
    },
    {
      "epoch": 0.2533429361535578,
      "grad_norm": 0.8125,
      "learning_rate": 9.873207320732074e-05,
      "loss": 3.1846,
      "step": 506390
    },
    {
      "epoch": 0.253361579389716,
      "grad_norm": 0.75,
      "learning_rate": 9.873007300730074e-05,
      "loss": 3.1227,
      "step": 506400
    },
    {
      "epoch": 0.2533802226258742,
      "grad_norm": 0.7578125,
      "learning_rate": 9.872807280728073e-05,
      "loss": 3.2307,
      "step": 506410
    },
    {
      "epoch": 0.25339886586203236,
      "grad_norm": 0.98046875,
      "learning_rate": 9.872607260726073e-05,
      "loss": 3.3541,
      "step": 506420
    },
    {
      "epoch": 0.25341750909819055,
      "grad_norm": 0.6953125,
      "learning_rate": 9.872407240724072e-05,
      "loss": 3.3135,
      "step": 506430
    },
    {
      "epoch": 0.25343615233434874,
      "grad_norm": 0.734375,
      "learning_rate": 9.872207220722073e-05,
      "loss": 3.1574,
      "step": 506440
    },
    {
      "epoch": 0.25345479557050693,
      "grad_norm": 0.71875,
      "learning_rate": 9.872007200720073e-05,
      "loss": 3.2963,
      "step": 506450
    },
    {
      "epoch": 0.2534734388066651,
      "grad_norm": 0.71875,
      "learning_rate": 9.871807180718072e-05,
      "loss": 3.106,
      "step": 506460
    },
    {
      "epoch": 0.2534920820428233,
      "grad_norm": 0.73828125,
      "learning_rate": 9.871607160716072e-05,
      "loss": 3.2491,
      "step": 506470
    },
    {
      "epoch": 0.25351072527898144,
      "grad_norm": 0.765625,
      "learning_rate": 9.871407140714072e-05,
      "loss": 3.1152,
      "step": 506480
    },
    {
      "epoch": 0.25352936851513963,
      "grad_norm": 0.6875,
      "learning_rate": 9.871207120712071e-05,
      "loss": 3.3446,
      "step": 506490
    },
    {
      "epoch": 0.2535480117512978,
      "grad_norm": 0.796875,
      "learning_rate": 9.871007100710071e-05,
      "loss": 3.1133,
      "step": 506500
    },
    {
      "epoch": 0.253566654987456,
      "grad_norm": 0.6875,
      "learning_rate": 9.870807080708072e-05,
      "loss": 3.2299,
      "step": 506510
    },
    {
      "epoch": 0.2535852982236142,
      "grad_norm": 0.74609375,
      "learning_rate": 9.870607060706071e-05,
      "loss": 3.2278,
      "step": 506520
    },
    {
      "epoch": 0.2536039414597724,
      "grad_norm": 0.73046875,
      "learning_rate": 9.870407040704071e-05,
      "loss": 3.1993,
      "step": 506530
    },
    {
      "epoch": 0.2536225846959306,
      "grad_norm": 0.8515625,
      "learning_rate": 9.87020702070207e-05,
      "loss": 3.2518,
      "step": 506540
    },
    {
      "epoch": 0.25364122793208876,
      "grad_norm": 0.71875,
      "learning_rate": 9.87000700070007e-05,
      "loss": 3.1094,
      "step": 506550
    },
    {
      "epoch": 0.25365987116824695,
      "grad_norm": 0.70703125,
      "learning_rate": 9.86980698069807e-05,
      "loss": 3.1759,
      "step": 506560
    },
    {
      "epoch": 0.25367851440440514,
      "grad_norm": 0.65625,
      "learning_rate": 9.86960696069607e-05,
      "loss": 3.3086,
      "step": 506570
    },
    {
      "epoch": 0.2536971576405633,
      "grad_norm": 0.7265625,
      "learning_rate": 9.86940694069407e-05,
      "loss": 3.1648,
      "step": 506580
    },
    {
      "epoch": 0.25371580087672146,
      "grad_norm": 0.703125,
      "learning_rate": 9.86920692069207e-05,
      "loss": 3.1705,
      "step": 506590
    },
    {
      "epoch": 0.25373444411287965,
      "grad_norm": 0.6953125,
      "learning_rate": 9.86900690069007e-05,
      "loss": 3.2849,
      "step": 506600
    },
    {
      "epoch": 0.25375308734903784,
      "grad_norm": 0.6484375,
      "learning_rate": 9.86880688068807e-05,
      "loss": 3.3015,
      "step": 506610
    },
    {
      "epoch": 0.25377173058519603,
      "grad_norm": 0.75,
      "learning_rate": 9.86860686068607e-05,
      "loss": 3.2309,
      "step": 506620
    },
    {
      "epoch": 0.2537903738213542,
      "grad_norm": 0.8359375,
      "learning_rate": 9.868406840684069e-05,
      "loss": 3.1085,
      "step": 506630
    },
    {
      "epoch": 0.2538090170575124,
      "grad_norm": 0.875,
      "learning_rate": 9.868206820682069e-05,
      "loss": 3.1205,
      "step": 506640
    },
    {
      "epoch": 0.2538276602936706,
      "grad_norm": 0.8828125,
      "learning_rate": 9.868006800680068e-05,
      "loss": 3.1388,
      "step": 506650
    },
    {
      "epoch": 0.2538463035298288,
      "grad_norm": 0.73046875,
      "learning_rate": 9.867806780678068e-05,
      "loss": 3.2572,
      "step": 506660
    },
    {
      "epoch": 0.253864946765987,
      "grad_norm": 0.73828125,
      "learning_rate": 9.867606760676069e-05,
      "loss": 3.3728,
      "step": 506670
    },
    {
      "epoch": 0.2538835900021451,
      "grad_norm": 0.68359375,
      "learning_rate": 9.867406740674068e-05,
      "loss": 3.2728,
      "step": 506680
    },
    {
      "epoch": 0.2539022332383033,
      "grad_norm": 0.8046875,
      "learning_rate": 9.867206720672068e-05,
      "loss": 3.2234,
      "step": 506690
    },
    {
      "epoch": 0.2539208764744615,
      "grad_norm": 0.6796875,
      "learning_rate": 9.867006700670067e-05,
      "loss": 3.245,
      "step": 506700
    },
    {
      "epoch": 0.2539395197106197,
      "grad_norm": 0.78125,
      "learning_rate": 9.866806680668067e-05,
      "loss": 3.2774,
      "step": 506710
    },
    {
      "epoch": 0.25395816294677787,
      "grad_norm": 0.68359375,
      "learning_rate": 9.866606660666066e-05,
      "loss": 3.2156,
      "step": 506720
    },
    {
      "epoch": 0.25397680618293605,
      "grad_norm": 0.72265625,
      "learning_rate": 9.866406640664066e-05,
      "loss": 3.2954,
      "step": 506730
    },
    {
      "epoch": 0.25399544941909424,
      "grad_norm": 0.78515625,
      "learning_rate": 9.866206620662067e-05,
      "loss": 3.1941,
      "step": 506740
    },
    {
      "epoch": 0.25401409265525243,
      "grad_norm": 0.75390625,
      "learning_rate": 9.866006600660066e-05,
      "loss": 3.2279,
      "step": 506750
    },
    {
      "epoch": 0.2540327358914106,
      "grad_norm": 0.7265625,
      "learning_rate": 9.865806580658066e-05,
      "loss": 3.1327,
      "step": 506760
    },
    {
      "epoch": 0.2540513791275688,
      "grad_norm": 0.734375,
      "learning_rate": 9.865606560656067e-05,
      "loss": 3.3217,
      "step": 506770
    },
    {
      "epoch": 0.254070022363727,
      "grad_norm": 0.89453125,
      "learning_rate": 9.865406540654067e-05,
      "loss": 3.2675,
      "step": 506780
    },
    {
      "epoch": 0.25408866559988513,
      "grad_norm": 0.72265625,
      "learning_rate": 9.865206520652066e-05,
      "loss": 3.1498,
      "step": 506790
    },
    {
      "epoch": 0.2541073088360433,
      "grad_norm": 0.71875,
      "learning_rate": 9.865006500650066e-05,
      "loss": 3.2387,
      "step": 506800
    },
    {
      "epoch": 0.2541259520722015,
      "grad_norm": 0.75,
      "learning_rate": 9.864806480648065e-05,
      "loss": 3.1335,
      "step": 506810
    },
    {
      "epoch": 0.2541445953083597,
      "grad_norm": 0.6640625,
      "learning_rate": 9.864606460646065e-05,
      "loss": 3.1625,
      "step": 506820
    },
    {
      "epoch": 0.2541632385445179,
      "grad_norm": 0.81640625,
      "learning_rate": 9.864406440644066e-05,
      "loss": 3.3424,
      "step": 506830
    },
    {
      "epoch": 0.2541818817806761,
      "grad_norm": 0.74609375,
      "learning_rate": 9.864206420642065e-05,
      "loss": 3.236,
      "step": 506840
    },
    {
      "epoch": 0.25420052501683427,
      "grad_norm": 0.69921875,
      "learning_rate": 9.864006400640065e-05,
      "loss": 3.1711,
      "step": 506850
    },
    {
      "epoch": 0.25421916825299246,
      "grad_norm": 0.703125,
      "learning_rate": 9.863806380638064e-05,
      "loss": 3.1371,
      "step": 506860
    },
    {
      "epoch": 0.25423781148915064,
      "grad_norm": 0.6953125,
      "learning_rate": 9.863606360636064e-05,
      "loss": 3.1877,
      "step": 506870
    },
    {
      "epoch": 0.25425645472530883,
      "grad_norm": 0.859375,
      "learning_rate": 9.863406340634063e-05,
      "loss": 3.2722,
      "step": 506880
    },
    {
      "epoch": 0.25427509796146697,
      "grad_norm": 0.77734375,
      "learning_rate": 9.863206320632063e-05,
      "loss": 3.1526,
      "step": 506890
    },
    {
      "epoch": 0.25429374119762516,
      "grad_norm": 0.73046875,
      "learning_rate": 9.863006300630064e-05,
      "loss": 3.0969,
      "step": 506900
    },
    {
      "epoch": 0.25431238443378335,
      "grad_norm": 0.72265625,
      "learning_rate": 9.862806280628063e-05,
      "loss": 3.1431,
      "step": 506910
    },
    {
      "epoch": 0.25433102766994153,
      "grad_norm": 0.66015625,
      "learning_rate": 9.862606260626063e-05,
      "loss": 3.1623,
      "step": 506920
    },
    {
      "epoch": 0.2543496709060997,
      "grad_norm": 0.66796875,
      "learning_rate": 9.862406240624063e-05,
      "loss": 3.3065,
      "step": 506930
    },
    {
      "epoch": 0.2543683141422579,
      "grad_norm": 0.703125,
      "learning_rate": 9.862206220622062e-05,
      "loss": 3.1331,
      "step": 506940
    },
    {
      "epoch": 0.2543869573784161,
      "grad_norm": 0.69140625,
      "learning_rate": 9.862006200620062e-05,
      "loss": 3.1904,
      "step": 506950
    },
    {
      "epoch": 0.2544056006145743,
      "grad_norm": 0.890625,
      "learning_rate": 9.861806180618063e-05,
      "loss": 3.232,
      "step": 506960
    },
    {
      "epoch": 0.2544242438507325,
      "grad_norm": 0.70703125,
      "learning_rate": 9.861606160616062e-05,
      "loss": 3.169,
      "step": 506970
    },
    {
      "epoch": 0.25444288708689067,
      "grad_norm": 0.671875,
      "learning_rate": 9.861406140614062e-05,
      "loss": 3.2327,
      "step": 506980
    },
    {
      "epoch": 0.25446153032304886,
      "grad_norm": 0.80078125,
      "learning_rate": 9.861206120612063e-05,
      "loss": 3.0348,
      "step": 506990
    },
    {
      "epoch": 0.254480173559207,
      "grad_norm": 0.7265625,
      "learning_rate": 9.861006100610062e-05,
      "loss": 3.3684,
      "step": 507000
    },
    {
      "epoch": 0.2544988167953652,
      "grad_norm": 0.84765625,
      "learning_rate": 9.860806080608062e-05,
      "loss": 3.132,
      "step": 507010
    },
    {
      "epoch": 0.25451746003152337,
      "grad_norm": 0.66015625,
      "learning_rate": 9.860606060606061e-05,
      "loss": 3.228,
      "step": 507020
    },
    {
      "epoch": 0.25453610326768156,
      "grad_norm": 0.61328125,
      "learning_rate": 9.860406040604061e-05,
      "loss": 3.2538,
      "step": 507030
    },
    {
      "epoch": 0.25455474650383975,
      "grad_norm": 0.78125,
      "learning_rate": 9.86020602060206e-05,
      "loss": 3.1672,
      "step": 507040
    },
    {
      "epoch": 0.25457338973999794,
      "grad_norm": 0.7265625,
      "learning_rate": 9.86000600060006e-05,
      "loss": 3.2113,
      "step": 507050
    },
    {
      "epoch": 0.2545920329761561,
      "grad_norm": 0.56640625,
      "learning_rate": 9.85980598059806e-05,
      "loss": 3.277,
      "step": 507060
    },
    {
      "epoch": 0.2546106762123143,
      "grad_norm": 0.76953125,
      "learning_rate": 9.85960596059606e-05,
      "loss": 3.0955,
      "step": 507070
    },
    {
      "epoch": 0.2546293194484725,
      "grad_norm": 0.578125,
      "learning_rate": 9.85940594059406e-05,
      "loss": 3.1389,
      "step": 507080
    },
    {
      "epoch": 0.2546479626846307,
      "grad_norm": 0.70703125,
      "learning_rate": 9.85920592059206e-05,
      "loss": 3.2349,
      "step": 507090
    },
    {
      "epoch": 0.2546666059207888,
      "grad_norm": 0.78515625,
      "learning_rate": 9.859005900590059e-05,
      "loss": 3.1968,
      "step": 507100
    },
    {
      "epoch": 0.254685249156947,
      "grad_norm": 0.7265625,
      "learning_rate": 9.858805880588059e-05,
      "loss": 3.2427,
      "step": 507110
    },
    {
      "epoch": 0.2547038923931052,
      "grad_norm": 0.734375,
      "learning_rate": 9.858605860586058e-05,
      "loss": 3.1078,
      "step": 507120
    },
    {
      "epoch": 0.2547225356292634,
      "grad_norm": 0.79296875,
      "learning_rate": 9.858405840584058e-05,
      "loss": 3.2007,
      "step": 507130
    },
    {
      "epoch": 0.2547411788654216,
      "grad_norm": 0.7109375,
      "learning_rate": 9.858205820582059e-05,
      "loss": 3.1662,
      "step": 507140
    },
    {
      "epoch": 0.25475982210157977,
      "grad_norm": 0.73046875,
      "learning_rate": 9.858005800580058e-05,
      "loss": 3.1788,
      "step": 507150
    },
    {
      "epoch": 0.25477846533773796,
      "grad_norm": 0.6875,
      "learning_rate": 9.857805780578059e-05,
      "loss": 3.3193,
      "step": 507160
    },
    {
      "epoch": 0.25479710857389615,
      "grad_norm": 0.68359375,
      "learning_rate": 9.857605760576059e-05,
      "loss": 3.1332,
      "step": 507170
    },
    {
      "epoch": 0.25481575181005434,
      "grad_norm": 0.72265625,
      "learning_rate": 9.857405740574058e-05,
      "loss": 3.3159,
      "step": 507180
    },
    {
      "epoch": 0.2548343950462125,
      "grad_norm": 0.69921875,
      "learning_rate": 9.857205720572058e-05,
      "loss": 3.197,
      "step": 507190
    },
    {
      "epoch": 0.2548530382823707,
      "grad_norm": 0.7109375,
      "learning_rate": 9.857005700570057e-05,
      "loss": 3.2166,
      "step": 507200
    },
    {
      "epoch": 0.25487168151852885,
      "grad_norm": 0.81640625,
      "learning_rate": 9.856805680568057e-05,
      "loss": 3.1295,
      "step": 507210
    },
    {
      "epoch": 0.25489032475468704,
      "grad_norm": 0.6640625,
      "learning_rate": 9.856605660566057e-05,
      "loss": 3.3149,
      "step": 507220
    },
    {
      "epoch": 0.2549089679908452,
      "grad_norm": 0.78125,
      "learning_rate": 9.856405640564058e-05,
      "loss": 3.2707,
      "step": 507230
    },
    {
      "epoch": 0.2549276112270034,
      "grad_norm": 0.73828125,
      "learning_rate": 9.856205620562057e-05,
      "loss": 3.128,
      "step": 507240
    },
    {
      "epoch": 0.2549462544631616,
      "grad_norm": 0.734375,
      "learning_rate": 9.856005600560057e-05,
      "loss": 3.1774,
      "step": 507250
    },
    {
      "epoch": 0.2549648976993198,
      "grad_norm": 0.74609375,
      "learning_rate": 9.855805580558056e-05,
      "loss": 3.3091,
      "step": 507260
    },
    {
      "epoch": 0.254983540935478,
      "grad_norm": 0.69140625,
      "learning_rate": 9.855605560556056e-05,
      "loss": 3.2015,
      "step": 507270
    },
    {
      "epoch": 0.25500218417163617,
      "grad_norm": 0.76953125,
      "learning_rate": 9.855405540554055e-05,
      "loss": 3.1837,
      "step": 507280
    },
    {
      "epoch": 0.25502082740779436,
      "grad_norm": 0.76953125,
      "learning_rate": 9.855205520552055e-05,
      "loss": 3.1491,
      "step": 507290
    },
    {
      "epoch": 0.25503947064395255,
      "grad_norm": 0.65234375,
      "learning_rate": 9.855005500550056e-05,
      "loss": 3.2744,
      "step": 507300
    },
    {
      "epoch": 0.2550581138801107,
      "grad_norm": 0.73046875,
      "learning_rate": 9.854805480548055e-05,
      "loss": 3.2565,
      "step": 507310
    },
    {
      "epoch": 0.25507675711626887,
      "grad_norm": 0.6953125,
      "learning_rate": 9.854605460546055e-05,
      "loss": 3.1551,
      "step": 507320
    },
    {
      "epoch": 0.25509540035242706,
      "grad_norm": 0.62890625,
      "learning_rate": 9.854405440544054e-05,
      "loss": 3.2168,
      "step": 507330
    },
    {
      "epoch": 0.25511404358858525,
      "grad_norm": 0.8828125,
      "learning_rate": 9.854205420542054e-05,
      "loss": 3.2597,
      "step": 507340
    },
    {
      "epoch": 0.25513268682474344,
      "grad_norm": 0.7421875,
      "learning_rate": 9.854005400540054e-05,
      "loss": 3.2512,
      "step": 507350
    },
    {
      "epoch": 0.2551513300609016,
      "grad_norm": 0.78515625,
      "learning_rate": 9.853805380538055e-05,
      "loss": 3.0465,
      "step": 507360
    },
    {
      "epoch": 0.2551699732970598,
      "grad_norm": 0.8359375,
      "learning_rate": 9.853605360536054e-05,
      "loss": 3.2341,
      "step": 507370
    },
    {
      "epoch": 0.255188616533218,
      "grad_norm": 0.75,
      "learning_rate": 9.853405340534054e-05,
      "loss": 3.0882,
      "step": 507380
    },
    {
      "epoch": 0.2552072597693762,
      "grad_norm": 0.7578125,
      "learning_rate": 9.853205320532055e-05,
      "loss": 3.1774,
      "step": 507390
    },
    {
      "epoch": 0.2552259030055344,
      "grad_norm": 0.66015625,
      "learning_rate": 9.853005300530054e-05,
      "loss": 3.2636,
      "step": 507400
    },
    {
      "epoch": 0.25524454624169257,
      "grad_norm": 0.7578125,
      "learning_rate": 9.852805280528054e-05,
      "loss": 3.153,
      "step": 507410
    },
    {
      "epoch": 0.2552631894778507,
      "grad_norm": 0.79296875,
      "learning_rate": 9.852605260526053e-05,
      "loss": 3.0855,
      "step": 507420
    },
    {
      "epoch": 0.2552818327140089,
      "grad_norm": 0.66015625,
      "learning_rate": 9.852405240524053e-05,
      "loss": 3.2909,
      "step": 507430
    },
    {
      "epoch": 0.2553004759501671,
      "grad_norm": 0.8125,
      "learning_rate": 9.852205220522052e-05,
      "loss": 3.135,
      "step": 507440
    },
    {
      "epoch": 0.25531911918632527,
      "grad_norm": 0.8125,
      "learning_rate": 9.852005200520052e-05,
      "loss": 3.1934,
      "step": 507450
    },
    {
      "epoch": 0.25533776242248346,
      "grad_norm": 0.78125,
      "learning_rate": 9.851805180518053e-05,
      "loss": 3.2475,
      "step": 507460
    },
    {
      "epoch": 0.25535640565864165,
      "grad_norm": 0.78125,
      "learning_rate": 9.851605160516052e-05,
      "loss": 3.1033,
      "step": 507470
    },
    {
      "epoch": 0.25537504889479984,
      "grad_norm": 0.8203125,
      "learning_rate": 9.851405140514052e-05,
      "loss": 3.2229,
      "step": 507480
    },
    {
      "epoch": 0.255393692130958,
      "grad_norm": 0.7421875,
      "learning_rate": 9.851205120512051e-05,
      "loss": 3.1077,
      "step": 507490
    },
    {
      "epoch": 0.2554123353671162,
      "grad_norm": 0.71484375,
      "learning_rate": 9.851005100510051e-05,
      "loss": 3.1812,
      "step": 507500
    },
    {
      "epoch": 0.2554309786032744,
      "grad_norm": 0.72265625,
      "learning_rate": 9.85080508050805e-05,
      "loss": 3.1175,
      "step": 507510
    },
    {
      "epoch": 0.25544962183943254,
      "grad_norm": 0.625,
      "learning_rate": 9.85060506050605e-05,
      "loss": 3.2452,
      "step": 507520
    },
    {
      "epoch": 0.25546826507559073,
      "grad_norm": 0.7890625,
      "learning_rate": 9.850405040504051e-05,
      "loss": 3.0437,
      "step": 507530
    },
    {
      "epoch": 0.2554869083117489,
      "grad_norm": 0.65625,
      "learning_rate": 9.85020502050205e-05,
      "loss": 3.2613,
      "step": 507540
    },
    {
      "epoch": 0.2555055515479071,
      "grad_norm": 0.765625,
      "learning_rate": 9.85000500050005e-05,
      "loss": 3.195,
      "step": 507550
    },
    {
      "epoch": 0.2555241947840653,
      "grad_norm": 0.66796875,
      "learning_rate": 9.849804980498051e-05,
      "loss": 3.1419,
      "step": 507560
    },
    {
      "epoch": 0.2555428380202235,
      "grad_norm": 0.734375,
      "learning_rate": 9.84960496049605e-05,
      "loss": 3.2649,
      "step": 507570
    },
    {
      "epoch": 0.2555614812563817,
      "grad_norm": 0.7734375,
      "learning_rate": 9.84940494049405e-05,
      "loss": 3.2525,
      "step": 507580
    },
    {
      "epoch": 0.25558012449253986,
      "grad_norm": 0.80078125,
      "learning_rate": 9.84920492049205e-05,
      "loss": 3.2967,
      "step": 507590
    },
    {
      "epoch": 0.25559876772869805,
      "grad_norm": 0.7890625,
      "learning_rate": 9.84900490049005e-05,
      "loss": 3.1578,
      "step": 507600
    },
    {
      "epoch": 0.25561741096485624,
      "grad_norm": 0.80859375,
      "learning_rate": 9.848804880488049e-05,
      "loss": 3.1476,
      "step": 507610
    },
    {
      "epoch": 0.25563605420101443,
      "grad_norm": 0.6953125,
      "learning_rate": 9.84860486048605e-05,
      "loss": 3.2749,
      "step": 507620
    },
    {
      "epoch": 0.25565469743717256,
      "grad_norm": 0.69140625,
      "learning_rate": 9.84840484048405e-05,
      "loss": 3.1717,
      "step": 507630
    },
    {
      "epoch": 0.25567334067333075,
      "grad_norm": 0.75,
      "learning_rate": 9.848204820482049e-05,
      "loss": 3.2227,
      "step": 507640
    },
    {
      "epoch": 0.25569198390948894,
      "grad_norm": 0.8359375,
      "learning_rate": 9.848004800480049e-05,
      "loss": 3.1844,
      "step": 507650
    },
    {
      "epoch": 0.25571062714564713,
      "grad_norm": 0.74609375,
      "learning_rate": 9.847804780478048e-05,
      "loss": 3.2585,
      "step": 507660
    },
    {
      "epoch": 0.2557292703818053,
      "grad_norm": 0.67578125,
      "learning_rate": 9.847604760476048e-05,
      "loss": 3.3366,
      "step": 507670
    },
    {
      "epoch": 0.2557479136179635,
      "grad_norm": 0.75,
      "learning_rate": 9.847404740474047e-05,
      "loss": 3.129,
      "step": 507680
    },
    {
      "epoch": 0.2557665568541217,
      "grad_norm": 0.734375,
      "learning_rate": 9.847204720472048e-05,
      "loss": 3.1852,
      "step": 507690
    },
    {
      "epoch": 0.2557852000902799,
      "grad_norm": 0.70703125,
      "learning_rate": 9.847004700470048e-05,
      "loss": 3.2008,
      "step": 507700
    },
    {
      "epoch": 0.2558038433264381,
      "grad_norm": 0.7265625,
      "learning_rate": 9.846804680468047e-05,
      "loss": 3.2414,
      "step": 507710
    },
    {
      "epoch": 0.25582248656259626,
      "grad_norm": 0.71484375,
      "learning_rate": 9.846604660466047e-05,
      "loss": 3.2368,
      "step": 507720
    },
    {
      "epoch": 0.2558411297987544,
      "grad_norm": 0.76171875,
      "learning_rate": 9.846404640464046e-05,
      "loss": 3.1658,
      "step": 507730
    },
    {
      "epoch": 0.2558597730349126,
      "grad_norm": 0.7578125,
      "learning_rate": 9.846204620462046e-05,
      "loss": 3.126,
      "step": 507740
    },
    {
      "epoch": 0.2558784162710708,
      "grad_norm": 0.75390625,
      "learning_rate": 9.846004600460047e-05,
      "loss": 3.174,
      "step": 507750
    },
    {
      "epoch": 0.25589705950722896,
      "grad_norm": 0.72265625,
      "learning_rate": 9.845804580458046e-05,
      "loss": 3.1193,
      "step": 507760
    },
    {
      "epoch": 0.25591570274338715,
      "grad_norm": 0.7578125,
      "learning_rate": 9.845604560456046e-05,
      "loss": 3.1991,
      "step": 507770
    },
    {
      "epoch": 0.25593434597954534,
      "grad_norm": 0.74609375,
      "learning_rate": 9.845404540454047e-05,
      "loss": 3.3443,
      "step": 507780
    },
    {
      "epoch": 0.25595298921570353,
      "grad_norm": 0.6875,
      "learning_rate": 9.845204520452046e-05,
      "loss": 3.2285,
      "step": 507790
    },
    {
      "epoch": 0.2559716324518617,
      "grad_norm": 0.7421875,
      "learning_rate": 9.845004500450046e-05,
      "loss": 3.3343,
      "step": 507800
    },
    {
      "epoch": 0.2559902756880199,
      "grad_norm": 0.75,
      "learning_rate": 9.844804480448046e-05,
      "loss": 3.1921,
      "step": 507810
    },
    {
      "epoch": 0.2560089189241781,
      "grad_norm": 0.734375,
      "learning_rate": 9.844604460446045e-05,
      "loss": 3.1596,
      "step": 507820
    },
    {
      "epoch": 0.2560275621603363,
      "grad_norm": 0.76171875,
      "learning_rate": 9.844404440444045e-05,
      "loss": 3.1783,
      "step": 507830
    },
    {
      "epoch": 0.2560462053964944,
      "grad_norm": 0.734375,
      "learning_rate": 9.844204420442044e-05,
      "loss": 3.2288,
      "step": 507840
    },
    {
      "epoch": 0.2560648486326526,
      "grad_norm": 0.84375,
      "learning_rate": 9.844004400440045e-05,
      "loss": 3.2004,
      "step": 507850
    },
    {
      "epoch": 0.2560834918688108,
      "grad_norm": 0.703125,
      "learning_rate": 9.843804380438045e-05,
      "loss": 3.155,
      "step": 507860
    },
    {
      "epoch": 0.256102135104969,
      "grad_norm": 0.76171875,
      "learning_rate": 9.843604360436044e-05,
      "loss": 3.1854,
      "step": 507870
    },
    {
      "epoch": 0.2561207783411272,
      "grad_norm": 0.796875,
      "learning_rate": 9.843404340434044e-05,
      "loss": 3.2559,
      "step": 507880
    },
    {
      "epoch": 0.25613942157728536,
      "grad_norm": 0.74609375,
      "learning_rate": 9.843204320432043e-05,
      "loss": 3.2659,
      "step": 507890
    },
    {
      "epoch": 0.25615806481344355,
      "grad_norm": 0.75,
      "learning_rate": 9.843004300430043e-05,
      "loss": 3.2591,
      "step": 507900
    },
    {
      "epoch": 0.25617670804960174,
      "grad_norm": 0.76953125,
      "learning_rate": 9.842804280428042e-05,
      "loss": 3.262,
      "step": 507910
    },
    {
      "epoch": 0.25619535128575993,
      "grad_norm": 0.8828125,
      "learning_rate": 9.842604260426042e-05,
      "loss": 3.1982,
      "step": 507920
    },
    {
      "epoch": 0.2562139945219181,
      "grad_norm": 0.68359375,
      "learning_rate": 9.842404240424043e-05,
      "loss": 3.2991,
      "step": 507930
    },
    {
      "epoch": 0.25623263775807625,
      "grad_norm": 0.7578125,
      "learning_rate": 9.842204220422043e-05,
      "loss": 3.1942,
      "step": 507940
    },
    {
      "epoch": 0.25625128099423444,
      "grad_norm": 0.73828125,
      "learning_rate": 9.842004200420043e-05,
      "loss": 3.2756,
      "step": 507950
    },
    {
      "epoch": 0.25626992423039263,
      "grad_norm": 0.796875,
      "learning_rate": 9.841804180418043e-05,
      "loss": 3.2136,
      "step": 507960
    },
    {
      "epoch": 0.2562885674665508,
      "grad_norm": 0.609375,
      "learning_rate": 9.841604160416043e-05,
      "loss": 3.2076,
      "step": 507970
    },
    {
      "epoch": 0.256307210702709,
      "grad_norm": 0.65625,
      "learning_rate": 9.841404140414042e-05,
      "loss": 3.1278,
      "step": 507980
    },
    {
      "epoch": 0.2563258539388672,
      "grad_norm": 0.828125,
      "learning_rate": 9.841204120412042e-05,
      "loss": 3.1757,
      "step": 507990
    },
    {
      "epoch": 0.2563444971750254,
      "grad_norm": 0.77734375,
      "learning_rate": 9.841004100410041e-05,
      "loss": 3.2702,
      "step": 508000
    },
    {
      "epoch": 0.2563631404111836,
      "grad_norm": 0.66796875,
      "learning_rate": 9.840804080408042e-05,
      "loss": 3.2256,
      "step": 508010
    },
    {
      "epoch": 0.25638178364734177,
      "grad_norm": 0.90625,
      "learning_rate": 9.840604060406042e-05,
      "loss": 3.2102,
      "step": 508020
    },
    {
      "epoch": 0.25640042688349995,
      "grad_norm": 0.77734375,
      "learning_rate": 9.840404040404041e-05,
      "loss": 3.0794,
      "step": 508030
    },
    {
      "epoch": 0.2564190701196581,
      "grad_norm": 0.59765625,
      "learning_rate": 9.840204020402041e-05,
      "loss": 3.0666,
      "step": 508040
    },
    {
      "epoch": 0.2564377133558163,
      "grad_norm": 0.75,
      "learning_rate": 9.84000400040004e-05,
      "loss": 3.2764,
      "step": 508050
    },
    {
      "epoch": 0.25645635659197447,
      "grad_norm": 0.69921875,
      "learning_rate": 9.83980398039804e-05,
      "loss": 3.3128,
      "step": 508060
    },
    {
      "epoch": 0.25647499982813265,
      "grad_norm": 0.71484375,
      "learning_rate": 9.83960396039604e-05,
      "loss": 3.2945,
      "step": 508070
    },
    {
      "epoch": 0.25649364306429084,
      "grad_norm": 0.73046875,
      "learning_rate": 9.839403940394039e-05,
      "loss": 3.1306,
      "step": 508080
    },
    {
      "epoch": 0.25651228630044903,
      "grad_norm": 0.8125,
      "learning_rate": 9.83920392039204e-05,
      "loss": 3.1532,
      "step": 508090
    },
    {
      "epoch": 0.2565309295366072,
      "grad_norm": 0.67578125,
      "learning_rate": 9.83900390039004e-05,
      "loss": 3.1723,
      "step": 508100
    },
    {
      "epoch": 0.2565495727727654,
      "grad_norm": 0.81640625,
      "learning_rate": 9.838803880388039e-05,
      "loss": 3.0657,
      "step": 508110
    },
    {
      "epoch": 0.2565682160089236,
      "grad_norm": 0.82421875,
      "learning_rate": 9.838603860386039e-05,
      "loss": 3.145,
      "step": 508120
    },
    {
      "epoch": 0.2565868592450818,
      "grad_norm": 0.69921875,
      "learning_rate": 9.838403840384038e-05,
      "loss": 3.1353,
      "step": 508130
    },
    {
      "epoch": 0.25660550248124,
      "grad_norm": 0.6875,
      "learning_rate": 9.838203820382038e-05,
      "loss": 3.272,
      "step": 508140
    },
    {
      "epoch": 0.2566241457173981,
      "grad_norm": 0.77734375,
      "learning_rate": 9.838003800380039e-05,
      "loss": 3.2793,
      "step": 508150
    },
    {
      "epoch": 0.2566427889535563,
      "grad_norm": 0.75390625,
      "learning_rate": 9.837803780378038e-05,
      "loss": 3.1723,
      "step": 508160
    },
    {
      "epoch": 0.2566614321897145,
      "grad_norm": 0.70703125,
      "learning_rate": 9.837603760376039e-05,
      "loss": 3.2807,
      "step": 508170
    },
    {
      "epoch": 0.2566800754258727,
      "grad_norm": 0.70703125,
      "learning_rate": 9.837403740374039e-05,
      "loss": 3.1146,
      "step": 508180
    },
    {
      "epoch": 0.25669871866203087,
      "grad_norm": 0.8203125,
      "learning_rate": 9.837203720372038e-05,
      "loss": 3.28,
      "step": 508190
    },
    {
      "epoch": 0.25671736189818906,
      "grad_norm": 0.7265625,
      "learning_rate": 9.837003700370038e-05,
      "loss": 3.2222,
      "step": 508200
    },
    {
      "epoch": 0.25673600513434724,
      "grad_norm": 0.70703125,
      "learning_rate": 9.836803680368037e-05,
      "loss": 3.2167,
      "step": 508210
    },
    {
      "epoch": 0.25675464837050543,
      "grad_norm": 0.80859375,
      "learning_rate": 9.836603660366037e-05,
      "loss": 3.2116,
      "step": 508220
    },
    {
      "epoch": 0.2567732916066636,
      "grad_norm": 0.71484375,
      "learning_rate": 9.836403640364037e-05,
      "loss": 3.1294,
      "step": 508230
    },
    {
      "epoch": 0.2567919348428218,
      "grad_norm": 0.7890625,
      "learning_rate": 9.836203620362036e-05,
      "loss": 3.2303,
      "step": 508240
    },
    {
      "epoch": 0.25681057807897995,
      "grad_norm": 0.859375,
      "learning_rate": 9.836003600360037e-05,
      "loss": 3.2039,
      "step": 508250
    },
    {
      "epoch": 0.25682922131513813,
      "grad_norm": 0.66796875,
      "learning_rate": 9.835803580358037e-05,
      "loss": 3.2741,
      "step": 508260
    },
    {
      "epoch": 0.2568478645512963,
      "grad_norm": 0.73828125,
      "learning_rate": 9.835603560356036e-05,
      "loss": 3.226,
      "step": 508270
    },
    {
      "epoch": 0.2568665077874545,
      "grad_norm": 0.671875,
      "learning_rate": 9.835403540354036e-05,
      "loss": 3.1917,
      "step": 508280
    },
    {
      "epoch": 0.2568851510236127,
      "grad_norm": 0.73046875,
      "learning_rate": 9.835203520352035e-05,
      "loss": 3.1457,
      "step": 508290
    },
    {
      "epoch": 0.2569037942597709,
      "grad_norm": 0.6484375,
      "learning_rate": 9.835003500350035e-05,
      "loss": 3.225,
      "step": 508300
    },
    {
      "epoch": 0.2569224374959291,
      "grad_norm": 0.80078125,
      "learning_rate": 9.834803480348034e-05,
      "loss": 3.1908,
      "step": 508310
    },
    {
      "epoch": 0.25694108073208727,
      "grad_norm": 0.67578125,
      "learning_rate": 9.834603460346035e-05,
      "loss": 3.2376,
      "step": 508320
    },
    {
      "epoch": 0.25695972396824546,
      "grad_norm": 0.703125,
      "learning_rate": 9.834403440344035e-05,
      "loss": 3.2188,
      "step": 508330
    },
    {
      "epoch": 0.25697836720440365,
      "grad_norm": 0.73046875,
      "learning_rate": 9.834203420342034e-05,
      "loss": 3.2842,
      "step": 508340
    },
    {
      "epoch": 0.25699701044056183,
      "grad_norm": 0.640625,
      "learning_rate": 9.834003400340035e-05,
      "loss": 3.3051,
      "step": 508350
    },
    {
      "epoch": 0.25701565367671997,
      "grad_norm": 0.91015625,
      "learning_rate": 9.833803380338035e-05,
      "loss": 3.29,
      "step": 508360
    },
    {
      "epoch": 0.25703429691287816,
      "grad_norm": 0.80859375,
      "learning_rate": 9.833603360336034e-05,
      "loss": 3.2252,
      "step": 508370
    },
    {
      "epoch": 0.25705294014903635,
      "grad_norm": 0.75,
      "learning_rate": 9.833403340334034e-05,
      "loss": 3.2121,
      "step": 508380
    },
    {
      "epoch": 0.25707158338519454,
      "grad_norm": 0.81640625,
      "learning_rate": 9.833203320332034e-05,
      "loss": 3.2061,
      "step": 508390
    },
    {
      "epoch": 0.2570902266213527,
      "grad_norm": 0.6875,
      "learning_rate": 9.833003300330033e-05,
      "loss": 3.3489,
      "step": 508400
    },
    {
      "epoch": 0.2571088698575109,
      "grad_norm": 0.765625,
      "learning_rate": 9.832803280328034e-05,
      "loss": 3.2656,
      "step": 508410
    },
    {
      "epoch": 0.2571275130936691,
      "grad_norm": 0.7734375,
      "learning_rate": 9.832603260326034e-05,
      "loss": 3.0589,
      "step": 508420
    },
    {
      "epoch": 0.2571461563298273,
      "grad_norm": 0.75390625,
      "learning_rate": 9.832403240324033e-05,
      "loss": 3.2562,
      "step": 508430
    },
    {
      "epoch": 0.2571647995659855,
      "grad_norm": 0.671875,
      "learning_rate": 9.832203220322033e-05,
      "loss": 3.273,
      "step": 508440
    },
    {
      "epoch": 0.25718344280214367,
      "grad_norm": 0.765625,
      "learning_rate": 9.832003200320032e-05,
      "loss": 3.1762,
      "step": 508450
    },
    {
      "epoch": 0.2572020860383018,
      "grad_norm": 0.70703125,
      "learning_rate": 9.831803180318032e-05,
      "loss": 3.3612,
      "step": 508460
    },
    {
      "epoch": 0.25722072927446,
      "grad_norm": 0.83984375,
      "learning_rate": 9.831603160316031e-05,
      "loss": 3.1946,
      "step": 508470
    },
    {
      "epoch": 0.2572393725106182,
      "grad_norm": 0.69140625,
      "learning_rate": 9.831403140314032e-05,
      "loss": 3.2614,
      "step": 508480
    },
    {
      "epoch": 0.25725801574677637,
      "grad_norm": 0.75,
      "learning_rate": 9.831203120312032e-05,
      "loss": 3.1719,
      "step": 508490
    },
    {
      "epoch": 0.25727665898293456,
      "grad_norm": 0.7890625,
      "learning_rate": 9.831003100310031e-05,
      "loss": 3.2656,
      "step": 508500
    },
    {
      "epoch": 0.25729530221909275,
      "grad_norm": 0.6484375,
      "learning_rate": 9.830803080308031e-05,
      "loss": 3.191,
      "step": 508510
    },
    {
      "epoch": 0.25731394545525094,
      "grad_norm": 0.71484375,
      "learning_rate": 9.83060306030603e-05,
      "loss": 3.2039,
      "step": 508520
    },
    {
      "epoch": 0.2573325886914091,
      "grad_norm": 0.7421875,
      "learning_rate": 9.83040304030403e-05,
      "loss": 3.306,
      "step": 508530
    },
    {
      "epoch": 0.2573512319275673,
      "grad_norm": 0.69921875,
      "learning_rate": 9.830203020302031e-05,
      "loss": 3.2803,
      "step": 508540
    },
    {
      "epoch": 0.2573698751637255,
      "grad_norm": 0.75390625,
      "learning_rate": 9.83000300030003e-05,
      "loss": 3.2152,
      "step": 508550
    },
    {
      "epoch": 0.2573885183998837,
      "grad_norm": 0.6875,
      "learning_rate": 9.82980298029803e-05,
      "loss": 3.2451,
      "step": 508560
    },
    {
      "epoch": 0.2574071616360418,
      "grad_norm": 0.578125,
      "learning_rate": 9.829602960296031e-05,
      "loss": 3.2576,
      "step": 508570
    },
    {
      "epoch": 0.2574258048722,
      "grad_norm": 0.75,
      "learning_rate": 9.82940294029403e-05,
      "loss": 3.2299,
      "step": 508580
    },
    {
      "epoch": 0.2574444481083582,
      "grad_norm": 0.765625,
      "learning_rate": 9.82920292029203e-05,
      "loss": 3.119,
      "step": 508590
    },
    {
      "epoch": 0.2574630913445164,
      "grad_norm": 0.671875,
      "learning_rate": 9.82900290029003e-05,
      "loss": 3.2082,
      "step": 508600
    },
    {
      "epoch": 0.2574817345806746,
      "grad_norm": 0.73046875,
      "learning_rate": 9.828802880288029e-05,
      "loss": 3.2563,
      "step": 508610
    },
    {
      "epoch": 0.25750037781683277,
      "grad_norm": 0.7890625,
      "learning_rate": 9.828602860286029e-05,
      "loss": 3.1961,
      "step": 508620
    },
    {
      "epoch": 0.25751902105299096,
      "grad_norm": 0.58984375,
      "learning_rate": 9.828402840284028e-05,
      "loss": 3.2317,
      "step": 508630
    },
    {
      "epoch": 0.25753766428914915,
      "grad_norm": 0.79296875,
      "learning_rate": 9.828202820282029e-05,
      "loss": 3.2672,
      "step": 508640
    },
    {
      "epoch": 0.25755630752530734,
      "grad_norm": 0.765625,
      "learning_rate": 9.828002800280029e-05,
      "loss": 3.0996,
      "step": 508650
    },
    {
      "epoch": 0.2575749507614655,
      "grad_norm": 0.671875,
      "learning_rate": 9.827802780278028e-05,
      "loss": 3.1234,
      "step": 508660
    },
    {
      "epoch": 0.25759359399762366,
      "grad_norm": 0.73046875,
      "learning_rate": 9.827602760276028e-05,
      "loss": 3.2407,
      "step": 508670
    },
    {
      "epoch": 0.25761223723378185,
      "grad_norm": 0.6796875,
      "learning_rate": 9.827402740274028e-05,
      "loss": 3.2621,
      "step": 508680
    },
    {
      "epoch": 0.25763088046994004,
      "grad_norm": 0.7578125,
      "learning_rate": 9.827202720272027e-05,
      "loss": 3.124,
      "step": 508690
    },
    {
      "epoch": 0.2576495237060982,
      "grad_norm": 0.80078125,
      "learning_rate": 9.827002700270027e-05,
      "loss": 3.3043,
      "step": 508700
    },
    {
      "epoch": 0.2576681669422564,
      "grad_norm": 0.73828125,
      "learning_rate": 9.826802680268026e-05,
      "loss": 3.2283,
      "step": 508710
    },
    {
      "epoch": 0.2576868101784146,
      "grad_norm": 0.8515625,
      "learning_rate": 9.826602660266027e-05,
      "loss": 3.1594,
      "step": 508720
    },
    {
      "epoch": 0.2577054534145728,
      "grad_norm": 0.828125,
      "learning_rate": 9.826402640264027e-05,
      "loss": 3.1473,
      "step": 508730
    },
    {
      "epoch": 0.257724096650731,
      "grad_norm": 0.7421875,
      "learning_rate": 9.826202620262028e-05,
      "loss": 3.2153,
      "step": 508740
    },
    {
      "epoch": 0.25774273988688917,
      "grad_norm": 0.75390625,
      "learning_rate": 9.826002600260027e-05,
      "loss": 3.1955,
      "step": 508750
    },
    {
      "epoch": 0.25776138312304736,
      "grad_norm": 0.640625,
      "learning_rate": 9.825802580258027e-05,
      "loss": 3.1615,
      "step": 508760
    },
    {
      "epoch": 0.25778002635920555,
      "grad_norm": 0.69921875,
      "learning_rate": 9.825602560256026e-05,
      "loss": 3.2884,
      "step": 508770
    },
    {
      "epoch": 0.2577986695953637,
      "grad_norm": 0.76171875,
      "learning_rate": 9.825402540254026e-05,
      "loss": 3.2019,
      "step": 508780
    },
    {
      "epoch": 0.25781731283152187,
      "grad_norm": 0.69140625,
      "learning_rate": 9.825202520252025e-05,
      "loss": 3.278,
      "step": 508790
    },
    {
      "epoch": 0.25783595606768006,
      "grad_norm": 0.671875,
      "learning_rate": 9.825002500250026e-05,
      "loss": 3.1383,
      "step": 508800
    },
    {
      "epoch": 0.25785459930383825,
      "grad_norm": 0.81640625,
      "learning_rate": 9.824802480248026e-05,
      "loss": 3.3568,
      "step": 508810
    },
    {
      "epoch": 0.25787324253999644,
      "grad_norm": 0.73828125,
      "learning_rate": 9.824602460246025e-05,
      "loss": 3.1521,
      "step": 508820
    },
    {
      "epoch": 0.25789188577615463,
      "grad_norm": 0.81640625,
      "learning_rate": 9.824402440244025e-05,
      "loss": 3.1859,
      "step": 508830
    },
    {
      "epoch": 0.2579105290123128,
      "grad_norm": 0.71875,
      "learning_rate": 9.824202420242025e-05,
      "loss": 3.1174,
      "step": 508840
    },
    {
      "epoch": 0.257929172248471,
      "grad_norm": 0.7734375,
      "learning_rate": 9.824002400240024e-05,
      "loss": 3.1633,
      "step": 508850
    },
    {
      "epoch": 0.2579478154846292,
      "grad_norm": 0.76171875,
      "learning_rate": 9.823802380238024e-05,
      "loss": 3.2808,
      "step": 508860
    },
    {
      "epoch": 0.2579664587207874,
      "grad_norm": 0.625,
      "learning_rate": 9.823602360236023e-05,
      "loss": 3.215,
      "step": 508870
    },
    {
      "epoch": 0.2579851019569455,
      "grad_norm": 0.65234375,
      "learning_rate": 9.823402340234024e-05,
      "loss": 3.3916,
      "step": 508880
    },
    {
      "epoch": 0.2580037451931037,
      "grad_norm": 0.70703125,
      "learning_rate": 9.823202320232024e-05,
      "loss": 3.2214,
      "step": 508890
    },
    {
      "epoch": 0.2580223884292619,
      "grad_norm": 0.6640625,
      "learning_rate": 9.823002300230023e-05,
      "loss": 3.2806,
      "step": 508900
    },
    {
      "epoch": 0.2580410316654201,
      "grad_norm": 0.7421875,
      "learning_rate": 9.822802280228023e-05,
      "loss": 3.2523,
      "step": 508910
    },
    {
      "epoch": 0.2580596749015783,
      "grad_norm": 0.71484375,
      "learning_rate": 9.822602260226022e-05,
      "loss": 3.3055,
      "step": 508920
    },
    {
      "epoch": 0.25807831813773646,
      "grad_norm": 0.6796875,
      "learning_rate": 9.822402240224022e-05,
      "loss": 3.3556,
      "step": 508930
    },
    {
      "epoch": 0.25809696137389465,
      "grad_norm": 0.8046875,
      "learning_rate": 9.822202220222023e-05,
      "loss": 3.2999,
      "step": 508940
    },
    {
      "epoch": 0.25811560461005284,
      "grad_norm": 0.828125,
      "learning_rate": 9.822002200220022e-05,
      "loss": 3.1692,
      "step": 508950
    },
    {
      "epoch": 0.25813424784621103,
      "grad_norm": 0.7578125,
      "learning_rate": 9.821802180218023e-05,
      "loss": 3.1002,
      "step": 508960
    },
    {
      "epoch": 0.2581528910823692,
      "grad_norm": 0.78125,
      "learning_rate": 9.821602160216023e-05,
      "loss": 3.1384,
      "step": 508970
    },
    {
      "epoch": 0.2581715343185274,
      "grad_norm": 0.75390625,
      "learning_rate": 9.821402140214022e-05,
      "loss": 3.1539,
      "step": 508980
    },
    {
      "epoch": 0.25819017755468554,
      "grad_norm": 0.7734375,
      "learning_rate": 9.821202120212022e-05,
      "loss": 3.2856,
      "step": 508990
    },
    {
      "epoch": 0.25820882079084373,
      "grad_norm": 0.7890625,
      "learning_rate": 9.821002100210022e-05,
      "loss": 3.2096,
      "step": 509000
    },
    {
      "epoch": 0.2582274640270019,
      "grad_norm": 0.69140625,
      "learning_rate": 9.820802080208021e-05,
      "loss": 3.0509,
      "step": 509010
    },
    {
      "epoch": 0.2582461072631601,
      "grad_norm": 0.65234375,
      "learning_rate": 9.82060206020602e-05,
      "loss": 3.169,
      "step": 509020
    },
    {
      "epoch": 0.2582647504993183,
      "grad_norm": 0.76953125,
      "learning_rate": 9.82040204020402e-05,
      "loss": 3.2937,
      "step": 509030
    },
    {
      "epoch": 0.2582833937354765,
      "grad_norm": 0.78515625,
      "learning_rate": 9.820202020202021e-05,
      "loss": 3.1687,
      "step": 509040
    },
    {
      "epoch": 0.2583020369716347,
      "grad_norm": 0.75390625,
      "learning_rate": 9.820002000200021e-05,
      "loss": 3.3139,
      "step": 509050
    },
    {
      "epoch": 0.25832068020779286,
      "grad_norm": 0.74609375,
      "learning_rate": 9.81980198019802e-05,
      "loss": 3.1557,
      "step": 509060
    },
    {
      "epoch": 0.25833932344395105,
      "grad_norm": 0.78515625,
      "learning_rate": 9.81960196019602e-05,
      "loss": 3.3285,
      "step": 509070
    },
    {
      "epoch": 0.25835796668010924,
      "grad_norm": 0.73046875,
      "learning_rate": 9.81940194019402e-05,
      "loss": 3.1715,
      "step": 509080
    },
    {
      "epoch": 0.2583766099162674,
      "grad_norm": 0.671875,
      "learning_rate": 9.819201920192019e-05,
      "loss": 3.1121,
      "step": 509090
    },
    {
      "epoch": 0.25839525315242556,
      "grad_norm": 0.69921875,
      "learning_rate": 9.819001900190018e-05,
      "loss": 3.2399,
      "step": 509100
    },
    {
      "epoch": 0.25841389638858375,
      "grad_norm": 0.61328125,
      "learning_rate": 9.81880188018802e-05,
      "loss": 3.1889,
      "step": 509110
    },
    {
      "epoch": 0.25843253962474194,
      "grad_norm": 0.81640625,
      "learning_rate": 9.818601860186019e-05,
      "loss": 3.3339,
      "step": 509120
    },
    {
      "epoch": 0.25845118286090013,
      "grad_norm": 0.67578125,
      "learning_rate": 9.818401840184019e-05,
      "loss": 3.312,
      "step": 509130
    },
    {
      "epoch": 0.2584698260970583,
      "grad_norm": 0.765625,
      "learning_rate": 9.81820182018202e-05,
      "loss": 3.1276,
      "step": 509140
    },
    {
      "epoch": 0.2584884693332165,
      "grad_norm": 0.75390625,
      "learning_rate": 9.818001800180019e-05,
      "loss": 3.2623,
      "step": 509150
    },
    {
      "epoch": 0.2585071125693747,
      "grad_norm": 0.81640625,
      "learning_rate": 9.817801780178019e-05,
      "loss": 3.2507,
      "step": 509160
    },
    {
      "epoch": 0.2585257558055329,
      "grad_norm": 0.7734375,
      "learning_rate": 9.817601760176018e-05,
      "loss": 3.2295,
      "step": 509170
    },
    {
      "epoch": 0.2585443990416911,
      "grad_norm": 0.77734375,
      "learning_rate": 9.817401740174018e-05,
      "loss": 3.3532,
      "step": 509180
    },
    {
      "epoch": 0.25856304227784926,
      "grad_norm": 0.78125,
      "learning_rate": 9.817201720172017e-05,
      "loss": 3.1885,
      "step": 509190
    },
    {
      "epoch": 0.2585816855140074,
      "grad_norm": 0.6640625,
      "learning_rate": 9.817001700170018e-05,
      "loss": 3.1755,
      "step": 509200
    },
    {
      "epoch": 0.2586003287501656,
      "grad_norm": 0.6640625,
      "learning_rate": 9.816801680168018e-05,
      "loss": 3.1588,
      "step": 509210
    },
    {
      "epoch": 0.2586189719863238,
      "grad_norm": 0.79296875,
      "learning_rate": 9.816601660166017e-05,
      "loss": 3.3274,
      "step": 509220
    },
    {
      "epoch": 0.25863761522248196,
      "grad_norm": 0.8359375,
      "learning_rate": 9.816401640164017e-05,
      "loss": 3.1872,
      "step": 509230
    },
    {
      "epoch": 0.25865625845864015,
      "grad_norm": 0.71484375,
      "learning_rate": 9.816201620162016e-05,
      "loss": 3.1351,
      "step": 509240
    },
    {
      "epoch": 0.25867490169479834,
      "grad_norm": 0.75,
      "learning_rate": 9.816001600160016e-05,
      "loss": 3.1821,
      "step": 509250
    },
    {
      "epoch": 0.25869354493095653,
      "grad_norm": 0.765625,
      "learning_rate": 9.815801580158016e-05,
      "loss": 3.1219,
      "step": 509260
    },
    {
      "epoch": 0.2587121881671147,
      "grad_norm": 0.7109375,
      "learning_rate": 9.815601560156016e-05,
      "loss": 3.1483,
      "step": 509270
    },
    {
      "epoch": 0.2587308314032729,
      "grad_norm": 0.703125,
      "learning_rate": 9.815401540154016e-05,
      "loss": 3.1083,
      "step": 509280
    },
    {
      "epoch": 0.2587494746394311,
      "grad_norm": 0.8359375,
      "learning_rate": 9.815201520152016e-05,
      "loss": 3.2226,
      "step": 509290
    },
    {
      "epoch": 0.25876811787558923,
      "grad_norm": 0.65625,
      "learning_rate": 9.815001500150015e-05,
      "loss": 3.3306,
      "step": 509300
    },
    {
      "epoch": 0.2587867611117474,
      "grad_norm": 0.8359375,
      "learning_rate": 9.814801480148015e-05,
      "loss": 3.2002,
      "step": 509310
    },
    {
      "epoch": 0.2588054043479056,
      "grad_norm": 0.7578125,
      "learning_rate": 9.814601460146014e-05,
      "loss": 3.2862,
      "step": 509320
    },
    {
      "epoch": 0.2588240475840638,
      "grad_norm": 0.72265625,
      "learning_rate": 9.814401440144015e-05,
      "loss": 3.1325,
      "step": 509330
    },
    {
      "epoch": 0.258842690820222,
      "grad_norm": 0.8984375,
      "learning_rate": 9.814201420142015e-05,
      "loss": 3.3092,
      "step": 509340
    },
    {
      "epoch": 0.2588613340563802,
      "grad_norm": 0.76953125,
      "learning_rate": 9.814001400140014e-05,
      "loss": 3.1875,
      "step": 509350
    },
    {
      "epoch": 0.25887997729253837,
      "grad_norm": 0.7578125,
      "learning_rate": 9.813801380138015e-05,
      "loss": 3.2446,
      "step": 509360
    },
    {
      "epoch": 0.25889862052869655,
      "grad_norm": 0.7578125,
      "learning_rate": 9.813601360136015e-05,
      "loss": 3.1065,
      "step": 509370
    },
    {
      "epoch": 0.25891726376485474,
      "grad_norm": 0.75390625,
      "learning_rate": 9.813401340134014e-05,
      "loss": 3.1628,
      "step": 509380
    },
    {
      "epoch": 0.25893590700101293,
      "grad_norm": 0.671875,
      "learning_rate": 9.813201320132014e-05,
      "loss": 3.3064,
      "step": 509390
    },
    {
      "epoch": 0.2589545502371711,
      "grad_norm": 0.7109375,
      "learning_rate": 9.813001300130013e-05,
      "loss": 3.1611,
      "step": 509400
    },
    {
      "epoch": 0.25897319347332926,
      "grad_norm": 0.79296875,
      "learning_rate": 9.812801280128013e-05,
      "loss": 3.2811,
      "step": 509410
    },
    {
      "epoch": 0.25899183670948744,
      "grad_norm": 0.73046875,
      "learning_rate": 9.812601260126013e-05,
      "loss": 3.2522,
      "step": 509420
    },
    {
      "epoch": 0.25901047994564563,
      "grad_norm": 0.78125,
      "learning_rate": 9.812401240124013e-05,
      "loss": 3.1461,
      "step": 509430
    },
    {
      "epoch": 0.2590291231818038,
      "grad_norm": 0.765625,
      "learning_rate": 9.812201220122013e-05,
      "loss": 3.1926,
      "step": 509440
    },
    {
      "epoch": 0.259047766417962,
      "grad_norm": 0.796875,
      "learning_rate": 9.812001200120013e-05,
      "loss": 3.3184,
      "step": 509450
    },
    {
      "epoch": 0.2590664096541202,
      "grad_norm": 0.6953125,
      "learning_rate": 9.811801180118012e-05,
      "loss": 3.2973,
      "step": 509460
    },
    {
      "epoch": 0.2590850528902784,
      "grad_norm": 0.6875,
      "learning_rate": 9.811601160116012e-05,
      "loss": 3.1492,
      "step": 509470
    },
    {
      "epoch": 0.2591036961264366,
      "grad_norm": 0.6484375,
      "learning_rate": 9.811401140114011e-05,
      "loss": 3.1878,
      "step": 509480
    },
    {
      "epoch": 0.25912233936259477,
      "grad_norm": 0.7578125,
      "learning_rate": 9.811201120112011e-05,
      "loss": 3.1904,
      "step": 509490
    },
    {
      "epoch": 0.25914098259875296,
      "grad_norm": 0.73046875,
      "learning_rate": 9.811001100110012e-05,
      "loss": 3.181,
      "step": 509500
    },
    {
      "epoch": 0.2591596258349111,
      "grad_norm": 0.62890625,
      "learning_rate": 9.810801080108011e-05,
      "loss": 3.2433,
      "step": 509510
    },
    {
      "epoch": 0.2591782690710693,
      "grad_norm": 0.7421875,
      "learning_rate": 9.810601060106011e-05,
      "loss": 3.1473,
      "step": 509520
    },
    {
      "epoch": 0.25919691230722747,
      "grad_norm": 0.82421875,
      "learning_rate": 9.810401040104012e-05,
      "loss": 3.1677,
      "step": 509530
    },
    {
      "epoch": 0.25921555554338566,
      "grad_norm": 0.8125,
      "learning_rate": 9.810201020102011e-05,
      "loss": 3.1294,
      "step": 509540
    },
    {
      "epoch": 0.25923419877954385,
      "grad_norm": 0.81640625,
      "learning_rate": 9.810001000100011e-05,
      "loss": 3.1852,
      "step": 509550
    },
    {
      "epoch": 0.25925284201570203,
      "grad_norm": 0.72265625,
      "learning_rate": 9.80980098009801e-05,
      "loss": 3.2768,
      "step": 509560
    },
    {
      "epoch": 0.2592714852518602,
      "grad_norm": 0.81640625,
      "learning_rate": 9.80960096009601e-05,
      "loss": 3.2529,
      "step": 509570
    },
    {
      "epoch": 0.2592901284880184,
      "grad_norm": 0.71875,
      "learning_rate": 9.80940094009401e-05,
      "loss": 3.1941,
      "step": 509580
    },
    {
      "epoch": 0.2593087717241766,
      "grad_norm": 0.67578125,
      "learning_rate": 9.80920092009201e-05,
      "loss": 3.1469,
      "step": 509590
    },
    {
      "epoch": 0.2593274149603348,
      "grad_norm": 0.7578125,
      "learning_rate": 9.80900090009001e-05,
      "loss": 3.2875,
      "step": 509600
    },
    {
      "epoch": 0.2593460581964929,
      "grad_norm": 0.7578125,
      "learning_rate": 9.80880088008801e-05,
      "loss": 3.163,
      "step": 509610
    },
    {
      "epoch": 0.2593647014326511,
      "grad_norm": 0.67578125,
      "learning_rate": 9.808600860086009e-05,
      "loss": 3.2999,
      "step": 509620
    },
    {
      "epoch": 0.2593833446688093,
      "grad_norm": 0.76171875,
      "learning_rate": 9.808400840084009e-05,
      "loss": 3.1281,
      "step": 509630
    },
    {
      "epoch": 0.2594019879049675,
      "grad_norm": 0.81640625,
      "learning_rate": 9.808200820082008e-05,
      "loss": 3.2414,
      "step": 509640
    },
    {
      "epoch": 0.2594206311411257,
      "grad_norm": 0.80078125,
      "learning_rate": 9.808000800080008e-05,
      "loss": 3.2834,
      "step": 509650
    },
    {
      "epoch": 0.25943927437728387,
      "grad_norm": 0.8125,
      "learning_rate": 9.807800780078009e-05,
      "loss": 3.2479,
      "step": 509660
    },
    {
      "epoch": 0.25945791761344206,
      "grad_norm": 0.8828125,
      "learning_rate": 9.807600760076008e-05,
      "loss": 3.1751,
      "step": 509670
    },
    {
      "epoch": 0.25947656084960025,
      "grad_norm": 0.7109375,
      "learning_rate": 9.807400740074008e-05,
      "loss": 3.1895,
      "step": 509680
    },
    {
      "epoch": 0.25949520408575844,
      "grad_norm": 0.78515625,
      "learning_rate": 9.807200720072007e-05,
      "loss": 2.9367,
      "step": 509690
    },
    {
      "epoch": 0.2595138473219166,
      "grad_norm": 0.76171875,
      "learning_rate": 9.807000700070007e-05,
      "loss": 3.2902,
      "step": 509700
    },
    {
      "epoch": 0.2595324905580748,
      "grad_norm": 0.796875,
      "learning_rate": 9.806800680068007e-05,
      "loss": 3.2243,
      "step": 509710
    },
    {
      "epoch": 0.25955113379423295,
      "grad_norm": 0.79296875,
      "learning_rate": 9.806600660066006e-05,
      "loss": 3.1788,
      "step": 509720
    },
    {
      "epoch": 0.25956977703039114,
      "grad_norm": 0.7578125,
      "learning_rate": 9.806400640064007e-05,
      "loss": 3.262,
      "step": 509730
    },
    {
      "epoch": 0.2595884202665493,
      "grad_norm": 0.79296875,
      "learning_rate": 9.806200620062007e-05,
      "loss": 3.1685,
      "step": 509740
    },
    {
      "epoch": 0.2596070635027075,
      "grad_norm": 0.83984375,
      "learning_rate": 9.806000600060007e-05,
      "loss": 3.2114,
      "step": 509750
    },
    {
      "epoch": 0.2596257067388657,
      "grad_norm": 0.70703125,
      "learning_rate": 9.805800580058007e-05,
      "loss": 3.2567,
      "step": 509760
    },
    {
      "epoch": 0.2596443499750239,
      "grad_norm": 0.78515625,
      "learning_rate": 9.805600560056007e-05,
      "loss": 3.2951,
      "step": 509770
    },
    {
      "epoch": 0.2596629932111821,
      "grad_norm": 0.78125,
      "learning_rate": 9.805400540054006e-05,
      "loss": 3.1012,
      "step": 509780
    },
    {
      "epoch": 0.25968163644734027,
      "grad_norm": 0.7265625,
      "learning_rate": 9.805200520052006e-05,
      "loss": 3.2727,
      "step": 509790
    },
    {
      "epoch": 0.25970027968349846,
      "grad_norm": 0.64453125,
      "learning_rate": 9.805000500050005e-05,
      "loss": 3.252,
      "step": 509800
    },
    {
      "epoch": 0.25971892291965665,
      "grad_norm": 0.60546875,
      "learning_rate": 9.804800480048005e-05,
      "loss": 3.2478,
      "step": 509810
    },
    {
      "epoch": 0.2597375661558148,
      "grad_norm": 0.8125,
      "learning_rate": 9.804600460046006e-05,
      "loss": 3.1276,
      "step": 509820
    },
    {
      "epoch": 0.25975620939197297,
      "grad_norm": 0.72265625,
      "learning_rate": 9.804400440044005e-05,
      "loss": 3.1025,
      "step": 509830
    },
    {
      "epoch": 0.25977485262813116,
      "grad_norm": 0.71484375,
      "learning_rate": 9.804200420042005e-05,
      "loss": 3.1209,
      "step": 509840
    },
    {
      "epoch": 0.25979349586428935,
      "grad_norm": 0.765625,
      "learning_rate": 9.804000400040004e-05,
      "loss": 3.1342,
      "step": 509850
    },
    {
      "epoch": 0.25981213910044754,
      "grad_norm": 0.75,
      "learning_rate": 9.803800380038004e-05,
      "loss": 3.204,
      "step": 509860
    },
    {
      "epoch": 0.2598307823366057,
      "grad_norm": 0.72265625,
      "learning_rate": 9.803600360036004e-05,
      "loss": 3.3286,
      "step": 509870
    },
    {
      "epoch": 0.2598494255727639,
      "grad_norm": 0.71875,
      "learning_rate": 9.803400340034003e-05,
      "loss": 3.2807,
      "step": 509880
    },
    {
      "epoch": 0.2598680688089221,
      "grad_norm": 0.8515625,
      "learning_rate": 9.803200320032003e-05,
      "loss": 3.2434,
      "step": 509890
    },
    {
      "epoch": 0.2598867120450803,
      "grad_norm": 0.79296875,
      "learning_rate": 9.803000300030004e-05,
      "loss": 3.2793,
      "step": 509900
    },
    {
      "epoch": 0.2599053552812385,
      "grad_norm": 0.796875,
      "learning_rate": 9.802800280028003e-05,
      "loss": 3.2274,
      "step": 509910
    },
    {
      "epoch": 0.25992399851739667,
      "grad_norm": 0.72265625,
      "learning_rate": 9.802600260026004e-05,
      "loss": 3.1753,
      "step": 509920
    },
    {
      "epoch": 0.2599426417535548,
      "grad_norm": 0.8125,
      "learning_rate": 9.802400240024004e-05,
      "loss": 3.1633,
      "step": 509930
    },
    {
      "epoch": 0.259961284989713,
      "grad_norm": 0.71875,
      "learning_rate": 9.802200220022003e-05,
      "loss": 3.2493,
      "step": 509940
    },
    {
      "epoch": 0.2599799282258712,
      "grad_norm": 0.765625,
      "learning_rate": 9.802000200020003e-05,
      "loss": 3.1888,
      "step": 509950
    },
    {
      "epoch": 0.25999857146202937,
      "grad_norm": 0.8046875,
      "learning_rate": 9.801800180018002e-05,
      "loss": 3.2686,
      "step": 509960
    },
    {
      "epoch": 0.26001721469818756,
      "grad_norm": 0.70703125,
      "learning_rate": 9.801600160016002e-05,
      "loss": 3.1776,
      "step": 509970
    },
    {
      "epoch": 0.26003585793434575,
      "grad_norm": 0.7734375,
      "learning_rate": 9.801400140014001e-05,
      "loss": 3.1735,
      "step": 509980
    },
    {
      "epoch": 0.26005450117050394,
      "grad_norm": 0.79296875,
      "learning_rate": 9.801200120012002e-05,
      "loss": 3.2574,
      "step": 509990
    },
    {
      "epoch": 0.2600731444066621,
      "grad_norm": 0.796875,
      "learning_rate": 9.801000100010002e-05,
      "loss": 3.1655,
      "step": 510000
    },
    {
      "epoch": 0.2600917876428203,
      "grad_norm": 0.7109375,
      "learning_rate": 9.800800080008001e-05,
      "loss": 3.1987,
      "step": 510010
    },
    {
      "epoch": 0.2601104308789785,
      "grad_norm": 0.73828125,
      "learning_rate": 9.800600060006001e-05,
      "loss": 3.2743,
      "step": 510020
    },
    {
      "epoch": 0.26012907411513664,
      "grad_norm": 0.80078125,
      "learning_rate": 9.800400040004e-05,
      "loss": 3.1672,
      "step": 510030
    },
    {
      "epoch": 0.2601477173512948,
      "grad_norm": 0.70703125,
      "learning_rate": 9.800200020002e-05,
      "loss": 3.1864,
      "step": 510040
    },
    {
      "epoch": 0.260166360587453,
      "grad_norm": 0.80078125,
      "learning_rate": 9.8e-05,
      "loss": 3.2293,
      "step": 510050
    },
    {
      "epoch": 0.2601850038236112,
      "grad_norm": 0.75,
      "learning_rate": 9.799799979998e-05,
      "loss": 3.3044,
      "step": 510060
    },
    {
      "epoch": 0.2602036470597694,
      "grad_norm": 0.66796875,
      "learning_rate": 9.799599959996e-05,
      "loss": 3.2207,
      "step": 510070
    },
    {
      "epoch": 0.2602222902959276,
      "grad_norm": 0.84375,
      "learning_rate": 9.799399939994e-05,
      "loss": 3.1842,
      "step": 510080
    },
    {
      "epoch": 0.26024093353208577,
      "grad_norm": 0.7109375,
      "learning_rate": 9.799199919991999e-05,
      "loss": 3.2274,
      "step": 510090
    },
    {
      "epoch": 0.26025957676824396,
      "grad_norm": 0.75,
      "learning_rate": 9.798999899989999e-05,
      "loss": 3.1807,
      "step": 510100
    },
    {
      "epoch": 0.26027822000440215,
      "grad_norm": 0.64453125,
      "learning_rate": 9.798799879987998e-05,
      "loss": 3.3049,
      "step": 510110
    },
    {
      "epoch": 0.26029686324056034,
      "grad_norm": 0.74609375,
      "learning_rate": 9.798599859985999e-05,
      "loss": 3.0823,
      "step": 510120
    },
    {
      "epoch": 0.26031550647671853,
      "grad_norm": 0.64453125,
      "learning_rate": 9.798399839983999e-05,
      "loss": 3.2156,
      "step": 510130
    },
    {
      "epoch": 0.26033414971287666,
      "grad_norm": 0.81640625,
      "learning_rate": 9.798199819981998e-05,
      "loss": 3.3041,
      "step": 510140
    },
    {
      "epoch": 0.26035279294903485,
      "grad_norm": 0.77734375,
      "learning_rate": 9.79799979998e-05,
      "loss": 3.2164,
      "step": 510150
    },
    {
      "epoch": 0.26037143618519304,
      "grad_norm": 0.8359375,
      "learning_rate": 9.797799779977999e-05,
      "loss": 3.1992,
      "step": 510160
    },
    {
      "epoch": 0.26039007942135123,
      "grad_norm": 0.7734375,
      "learning_rate": 9.797599759975998e-05,
      "loss": 3.2816,
      "step": 510170
    },
    {
      "epoch": 0.2604087226575094,
      "grad_norm": 0.734375,
      "learning_rate": 9.797399739973998e-05,
      "loss": 3.1632,
      "step": 510180
    },
    {
      "epoch": 0.2604273658936676,
      "grad_norm": 0.71875,
      "learning_rate": 9.797199719971998e-05,
      "loss": 3.2021,
      "step": 510190
    },
    {
      "epoch": 0.2604460091298258,
      "grad_norm": 0.78125,
      "learning_rate": 9.796999699969997e-05,
      "loss": 3.2781,
      "step": 510200
    },
    {
      "epoch": 0.260464652365984,
      "grad_norm": 0.75390625,
      "learning_rate": 9.796799679967997e-05,
      "loss": 3.2021,
      "step": 510210
    },
    {
      "epoch": 0.2604832956021422,
      "grad_norm": 0.796875,
      "learning_rate": 9.796599659965998e-05,
      "loss": 3.2711,
      "step": 510220
    },
    {
      "epoch": 0.26050193883830036,
      "grad_norm": 0.75390625,
      "learning_rate": 9.796399639963997e-05,
      "loss": 3.1084,
      "step": 510230
    },
    {
      "epoch": 0.2605205820744585,
      "grad_norm": 0.66796875,
      "learning_rate": 9.796199619961997e-05,
      "loss": 3.149,
      "step": 510240
    },
    {
      "epoch": 0.2605392253106167,
      "grad_norm": 0.82421875,
      "learning_rate": 9.795999599959996e-05,
      "loss": 3.2137,
      "step": 510250
    },
    {
      "epoch": 0.2605578685467749,
      "grad_norm": 0.87109375,
      "learning_rate": 9.795799579957996e-05,
      "loss": 3.1882,
      "step": 510260
    },
    {
      "epoch": 0.26057651178293306,
      "grad_norm": 0.7890625,
      "learning_rate": 9.795599559955995e-05,
      "loss": 3.3194,
      "step": 510270
    },
    {
      "epoch": 0.26059515501909125,
      "grad_norm": 0.7265625,
      "learning_rate": 9.795399539953995e-05,
      "loss": 3.1513,
      "step": 510280
    },
    {
      "epoch": 0.26061379825524944,
      "grad_norm": 0.79296875,
      "learning_rate": 9.795199519951996e-05,
      "loss": 3.183,
      "step": 510290
    },
    {
      "epoch": 0.26063244149140763,
      "grad_norm": 0.74609375,
      "learning_rate": 9.794999499949995e-05,
      "loss": 3.3038,
      "step": 510300
    },
    {
      "epoch": 0.2606510847275658,
      "grad_norm": 0.83984375,
      "learning_rate": 9.794799479947995e-05,
      "loss": 3.2228,
      "step": 510310
    },
    {
      "epoch": 0.260669727963724,
      "grad_norm": 0.70703125,
      "learning_rate": 9.794599459945996e-05,
      "loss": 3.1334,
      "step": 510320
    },
    {
      "epoch": 0.2606883711998822,
      "grad_norm": 0.8203125,
      "learning_rate": 9.794399439943995e-05,
      "loss": 3.1351,
      "step": 510330
    },
    {
      "epoch": 0.2607070144360404,
      "grad_norm": 0.71875,
      "learning_rate": 9.794199419941995e-05,
      "loss": 3.1899,
      "step": 510340
    },
    {
      "epoch": 0.2607256576721985,
      "grad_norm": 0.76171875,
      "learning_rate": 9.793999399939995e-05,
      "loss": 3.2882,
      "step": 510350
    },
    {
      "epoch": 0.2607443009083567,
      "grad_norm": 0.64453125,
      "learning_rate": 9.793799379937994e-05,
      "loss": 3.2899,
      "step": 510360
    },
    {
      "epoch": 0.2607629441445149,
      "grad_norm": 0.69921875,
      "learning_rate": 9.793599359935994e-05,
      "loss": 3.3142,
      "step": 510370
    },
    {
      "epoch": 0.2607815873806731,
      "grad_norm": 0.7421875,
      "learning_rate": 9.793399339933995e-05,
      "loss": 3.173,
      "step": 510380
    },
    {
      "epoch": 0.2608002306168313,
      "grad_norm": 0.7109375,
      "learning_rate": 9.793199319931994e-05,
      "loss": 3.2052,
      "step": 510390
    },
    {
      "epoch": 0.26081887385298946,
      "grad_norm": 0.72265625,
      "learning_rate": 9.792999299929994e-05,
      "loss": 3.1132,
      "step": 510400
    },
    {
      "epoch": 0.26083751708914765,
      "grad_norm": 0.703125,
      "learning_rate": 9.792799279927993e-05,
      "loss": 3.2014,
      "step": 510410
    },
    {
      "epoch": 0.26085616032530584,
      "grad_norm": 0.828125,
      "learning_rate": 9.792599259925993e-05,
      "loss": 3.1128,
      "step": 510420
    },
    {
      "epoch": 0.26087480356146403,
      "grad_norm": 0.77734375,
      "learning_rate": 9.792399239923992e-05,
      "loss": 3.1444,
      "step": 510430
    },
    {
      "epoch": 0.2608934467976222,
      "grad_norm": 0.81640625,
      "learning_rate": 9.792199219921992e-05,
      "loss": 3.165,
      "step": 510440
    },
    {
      "epoch": 0.26091209003378035,
      "grad_norm": 0.76171875,
      "learning_rate": 9.791999199919993e-05,
      "loss": 3.2371,
      "step": 510450
    },
    {
      "epoch": 0.26093073326993854,
      "grad_norm": 0.78125,
      "learning_rate": 9.791799179917992e-05,
      "loss": 3.2084,
      "step": 510460
    },
    {
      "epoch": 0.26094937650609673,
      "grad_norm": 0.69921875,
      "learning_rate": 9.791599159915992e-05,
      "loss": 3.2514,
      "step": 510470
    },
    {
      "epoch": 0.2609680197422549,
      "grad_norm": 0.734375,
      "learning_rate": 9.791399139913992e-05,
      "loss": 3.2657,
      "step": 510480
    },
    {
      "epoch": 0.2609866629784131,
      "grad_norm": 0.80859375,
      "learning_rate": 9.791199119911991e-05,
      "loss": 3.179,
      "step": 510490
    },
    {
      "epoch": 0.2610053062145713,
      "grad_norm": 0.625,
      "learning_rate": 9.790999099909991e-05,
      "loss": 3.186,
      "step": 510500
    },
    {
      "epoch": 0.2610239494507295,
      "grad_norm": 0.78125,
      "learning_rate": 9.79079907990799e-05,
      "loss": 3.2066,
      "step": 510510
    },
    {
      "epoch": 0.2610425926868877,
      "grad_norm": 0.8046875,
      "learning_rate": 9.790599059905991e-05,
      "loss": 3.2131,
      "step": 510520
    },
    {
      "epoch": 0.26106123592304586,
      "grad_norm": 0.66796875,
      "learning_rate": 9.790399039903991e-05,
      "loss": 3.1626,
      "step": 510530
    },
    {
      "epoch": 0.26107987915920405,
      "grad_norm": 0.80078125,
      "learning_rate": 9.790199019901992e-05,
      "loss": 3.2033,
      "step": 510540
    },
    {
      "epoch": 0.26109852239536224,
      "grad_norm": 0.7578125,
      "learning_rate": 9.789998999899991e-05,
      "loss": 3.2357,
      "step": 510550
    },
    {
      "epoch": 0.2611171656315204,
      "grad_norm": 0.6640625,
      "learning_rate": 9.789798979897991e-05,
      "loss": 3.163,
      "step": 510560
    },
    {
      "epoch": 0.26113580886767856,
      "grad_norm": 0.75,
      "learning_rate": 9.78959895989599e-05,
      "loss": 2.9781,
      "step": 510570
    },
    {
      "epoch": 0.26115445210383675,
      "grad_norm": 0.80078125,
      "learning_rate": 9.78939893989399e-05,
      "loss": 3.1738,
      "step": 510580
    },
    {
      "epoch": 0.26117309533999494,
      "grad_norm": 0.79296875,
      "learning_rate": 9.78919891989199e-05,
      "loss": 3.1193,
      "step": 510590
    },
    {
      "epoch": 0.26119173857615313,
      "grad_norm": 0.67578125,
      "learning_rate": 9.788998899889989e-05,
      "loss": 3.2361,
      "step": 510600
    },
    {
      "epoch": 0.2612103818123113,
      "grad_norm": 0.80859375,
      "learning_rate": 9.78879887988799e-05,
      "loss": 3.2218,
      "step": 510610
    },
    {
      "epoch": 0.2612290250484695,
      "grad_norm": 0.828125,
      "learning_rate": 9.78859885988599e-05,
      "loss": 3.2193,
      "step": 510620
    },
    {
      "epoch": 0.2612476682846277,
      "grad_norm": 0.75390625,
      "learning_rate": 9.788398839883989e-05,
      "loss": 3.1358,
      "step": 510630
    },
    {
      "epoch": 0.2612663115207859,
      "grad_norm": 0.8046875,
      "learning_rate": 9.788198819881989e-05,
      "loss": 3.2539,
      "step": 510640
    },
    {
      "epoch": 0.2612849547569441,
      "grad_norm": 0.71484375,
      "learning_rate": 9.787998799879988e-05,
      "loss": 3.295,
      "step": 510650
    },
    {
      "epoch": 0.2613035979931022,
      "grad_norm": 0.73828125,
      "learning_rate": 9.787798779877988e-05,
      "loss": 3.1448,
      "step": 510660
    },
    {
      "epoch": 0.2613222412292604,
      "grad_norm": 0.8828125,
      "learning_rate": 9.787598759875987e-05,
      "loss": 3.2898,
      "step": 510670
    },
    {
      "epoch": 0.2613408844654186,
      "grad_norm": 0.69140625,
      "learning_rate": 9.787398739873987e-05,
      "loss": 3.2254,
      "step": 510680
    },
    {
      "epoch": 0.2613595277015768,
      "grad_norm": 0.7734375,
      "learning_rate": 9.787198719871988e-05,
      "loss": 3.2409,
      "step": 510690
    },
    {
      "epoch": 0.26137817093773497,
      "grad_norm": 0.69921875,
      "learning_rate": 9.786998699869987e-05,
      "loss": 3.1481,
      "step": 510700
    },
    {
      "epoch": 0.26139681417389315,
      "grad_norm": 0.77734375,
      "learning_rate": 9.786798679867988e-05,
      "loss": 3.2746,
      "step": 510710
    },
    {
      "epoch": 0.26141545741005134,
      "grad_norm": 0.7109375,
      "learning_rate": 9.786598659865988e-05,
      "loss": 3.1158,
      "step": 510720
    },
    {
      "epoch": 0.26143410064620953,
      "grad_norm": 0.64453125,
      "learning_rate": 9.786398639863987e-05,
      "loss": 3.1793,
      "step": 510730
    },
    {
      "epoch": 0.2614527438823677,
      "grad_norm": 0.7578125,
      "learning_rate": 9.786198619861987e-05,
      "loss": 3.17,
      "step": 510740
    },
    {
      "epoch": 0.2614713871185259,
      "grad_norm": 0.671875,
      "learning_rate": 9.785998599859986e-05,
      "loss": 3.2613,
      "step": 510750
    },
    {
      "epoch": 0.2614900303546841,
      "grad_norm": 0.69921875,
      "learning_rate": 9.785798579857986e-05,
      "loss": 3.3274,
      "step": 510760
    },
    {
      "epoch": 0.26150867359084223,
      "grad_norm": 0.73046875,
      "learning_rate": 9.785598559855987e-05,
      "loss": 3.3188,
      "step": 510770
    },
    {
      "epoch": 0.2615273168270004,
      "grad_norm": 0.66015625,
      "learning_rate": 9.785398539853986e-05,
      "loss": 3.339,
      "step": 510780
    },
    {
      "epoch": 0.2615459600631586,
      "grad_norm": 0.71484375,
      "learning_rate": 9.785198519851986e-05,
      "loss": 3.236,
      "step": 510790
    },
    {
      "epoch": 0.2615646032993168,
      "grad_norm": 0.76171875,
      "learning_rate": 9.784998499849986e-05,
      "loss": 3.2416,
      "step": 510800
    },
    {
      "epoch": 0.261583246535475,
      "grad_norm": 0.84375,
      "learning_rate": 9.784798479847985e-05,
      "loss": 3.1609,
      "step": 510810
    },
    {
      "epoch": 0.2616018897716332,
      "grad_norm": 0.7734375,
      "learning_rate": 9.784598459845985e-05,
      "loss": 3.1676,
      "step": 510820
    },
    {
      "epoch": 0.26162053300779137,
      "grad_norm": 0.734375,
      "learning_rate": 9.784398439843984e-05,
      "loss": 3.2795,
      "step": 510830
    },
    {
      "epoch": 0.26163917624394956,
      "grad_norm": 0.74609375,
      "learning_rate": 9.784198419841984e-05,
      "loss": 3.1771,
      "step": 510840
    },
    {
      "epoch": 0.26165781948010775,
      "grad_norm": 0.75,
      "learning_rate": 9.783998399839985e-05,
      "loss": 3.1935,
      "step": 510850
    },
    {
      "epoch": 0.26167646271626593,
      "grad_norm": 0.65234375,
      "learning_rate": 9.783798379837984e-05,
      "loss": 3.1244,
      "step": 510860
    },
    {
      "epoch": 0.26169510595242407,
      "grad_norm": 0.69140625,
      "learning_rate": 9.783598359835984e-05,
      "loss": 3.2479,
      "step": 510870
    },
    {
      "epoch": 0.26171374918858226,
      "grad_norm": 0.7890625,
      "learning_rate": 9.783398339833983e-05,
      "loss": 3.2841,
      "step": 510880
    },
    {
      "epoch": 0.26173239242474045,
      "grad_norm": 0.81640625,
      "learning_rate": 9.783198319831983e-05,
      "loss": 3.1641,
      "step": 510890
    },
    {
      "epoch": 0.26175103566089863,
      "grad_norm": 0.7265625,
      "learning_rate": 9.782998299829983e-05,
      "loss": 3.1776,
      "step": 510900
    },
    {
      "epoch": 0.2617696788970568,
      "grad_norm": 0.734375,
      "learning_rate": 9.782798279827983e-05,
      "loss": 3.2574,
      "step": 510910
    },
    {
      "epoch": 0.261788322133215,
      "grad_norm": 0.66796875,
      "learning_rate": 9.782598259825983e-05,
      "loss": 3.2639,
      "step": 510920
    },
    {
      "epoch": 0.2618069653693732,
      "grad_norm": 0.77734375,
      "learning_rate": 9.782398239823984e-05,
      "loss": 3.1459,
      "step": 510930
    },
    {
      "epoch": 0.2618256086055314,
      "grad_norm": 0.8515625,
      "learning_rate": 9.782198219821984e-05,
      "loss": 3.237,
      "step": 510940
    },
    {
      "epoch": 0.2618442518416896,
      "grad_norm": 0.78515625,
      "learning_rate": 9.781998199819983e-05,
      "loss": 3.2672,
      "step": 510950
    },
    {
      "epoch": 0.26186289507784777,
      "grad_norm": 0.7109375,
      "learning_rate": 9.781798179817983e-05,
      "loss": 3.3206,
      "step": 510960
    },
    {
      "epoch": 0.2618815383140059,
      "grad_norm": 0.875,
      "learning_rate": 9.781598159815982e-05,
      "loss": 3.1833,
      "step": 510970
    },
    {
      "epoch": 0.2619001815501641,
      "grad_norm": 0.69140625,
      "learning_rate": 9.781398139813982e-05,
      "loss": 3.2146,
      "step": 510980
    },
    {
      "epoch": 0.2619188247863223,
      "grad_norm": 0.73828125,
      "learning_rate": 9.781198119811981e-05,
      "loss": 3.1968,
      "step": 510990
    },
    {
      "epoch": 0.26193746802248047,
      "grad_norm": 0.78125,
      "learning_rate": 9.780998099809981e-05,
      "loss": 3.1215,
      "step": 511000
    },
    {
      "epoch": 0.26195611125863866,
      "grad_norm": 0.6796875,
      "learning_rate": 9.780798079807982e-05,
      "loss": 3.3153,
      "step": 511010
    },
    {
      "epoch": 0.26197475449479685,
      "grad_norm": 0.6953125,
      "learning_rate": 9.780598059805981e-05,
      "loss": 3.1924,
      "step": 511020
    },
    {
      "epoch": 0.26199339773095504,
      "grad_norm": 0.6640625,
      "learning_rate": 9.780398039803981e-05,
      "loss": 3.254,
      "step": 511030
    },
    {
      "epoch": 0.2620120409671132,
      "grad_norm": 0.7890625,
      "learning_rate": 9.78019801980198e-05,
      "loss": 3.2153,
      "step": 511040
    },
    {
      "epoch": 0.2620306842032714,
      "grad_norm": 0.7421875,
      "learning_rate": 9.77999799979998e-05,
      "loss": 3.1305,
      "step": 511050
    },
    {
      "epoch": 0.2620493274394296,
      "grad_norm": 0.84765625,
      "learning_rate": 9.77979797979798e-05,
      "loss": 3.1857,
      "step": 511060
    },
    {
      "epoch": 0.2620679706755878,
      "grad_norm": 0.7421875,
      "learning_rate": 9.779597959795979e-05,
      "loss": 3.1098,
      "step": 511070
    },
    {
      "epoch": 0.2620866139117459,
      "grad_norm": 0.734375,
      "learning_rate": 9.77939793979398e-05,
      "loss": 3.1028,
      "step": 511080
    },
    {
      "epoch": 0.2621052571479041,
      "grad_norm": 0.8046875,
      "learning_rate": 9.77919791979198e-05,
      "loss": 3.3321,
      "step": 511090
    },
    {
      "epoch": 0.2621239003840623,
      "grad_norm": 0.71875,
      "learning_rate": 9.778997899789979e-05,
      "loss": 3.2387,
      "step": 511100
    },
    {
      "epoch": 0.2621425436202205,
      "grad_norm": 0.80078125,
      "learning_rate": 9.77879787978798e-05,
      "loss": 3.2312,
      "step": 511110
    },
    {
      "epoch": 0.2621611868563787,
      "grad_norm": 0.75390625,
      "learning_rate": 9.77859785978598e-05,
      "loss": 3.263,
      "step": 511120
    },
    {
      "epoch": 0.26217983009253687,
      "grad_norm": 0.7421875,
      "learning_rate": 9.778397839783979e-05,
      "loss": 3.2614,
      "step": 511130
    },
    {
      "epoch": 0.26219847332869506,
      "grad_norm": 0.74609375,
      "learning_rate": 9.778197819781979e-05,
      "loss": 3.0043,
      "step": 511140
    },
    {
      "epoch": 0.26221711656485325,
      "grad_norm": 0.73828125,
      "learning_rate": 9.777997799779978e-05,
      "loss": 3.1791,
      "step": 511150
    },
    {
      "epoch": 0.26223575980101144,
      "grad_norm": 0.7265625,
      "learning_rate": 9.777797779777978e-05,
      "loss": 3.1255,
      "step": 511160
    },
    {
      "epoch": 0.2622544030371696,
      "grad_norm": 0.7890625,
      "learning_rate": 9.777597759775979e-05,
      "loss": 3.2004,
      "step": 511170
    },
    {
      "epoch": 0.26227304627332776,
      "grad_norm": 0.80859375,
      "learning_rate": 9.777397739773978e-05,
      "loss": 3.2017,
      "step": 511180
    },
    {
      "epoch": 0.26229168950948595,
      "grad_norm": 0.6875,
      "learning_rate": 9.777197719771978e-05,
      "loss": 3.1118,
      "step": 511190
    },
    {
      "epoch": 0.26231033274564414,
      "grad_norm": 0.79296875,
      "learning_rate": 9.776997699769977e-05,
      "loss": 3.1791,
      "step": 511200
    },
    {
      "epoch": 0.2623289759818023,
      "grad_norm": 0.78515625,
      "learning_rate": 9.776797679767977e-05,
      "loss": 3.2668,
      "step": 511210
    },
    {
      "epoch": 0.2623476192179605,
      "grad_norm": 0.83203125,
      "learning_rate": 9.776597659765977e-05,
      "loss": 3.1683,
      "step": 511220
    },
    {
      "epoch": 0.2623662624541187,
      "grad_norm": 0.8046875,
      "learning_rate": 9.776397639763976e-05,
      "loss": 3.2603,
      "step": 511230
    },
    {
      "epoch": 0.2623849056902769,
      "grad_norm": 0.6796875,
      "learning_rate": 9.776197619761977e-05,
      "loss": 3.3081,
      "step": 511240
    },
    {
      "epoch": 0.2624035489264351,
      "grad_norm": 0.7109375,
      "learning_rate": 9.775997599759977e-05,
      "loss": 3.1745,
      "step": 511250
    },
    {
      "epoch": 0.26242219216259327,
      "grad_norm": 0.83984375,
      "learning_rate": 9.775797579757976e-05,
      "loss": 3.137,
      "step": 511260
    },
    {
      "epoch": 0.26244083539875146,
      "grad_norm": 0.828125,
      "learning_rate": 9.775597559755976e-05,
      "loss": 3.4072,
      "step": 511270
    },
    {
      "epoch": 0.26245947863490965,
      "grad_norm": 0.80078125,
      "learning_rate": 9.775397539753975e-05,
      "loss": 3.1029,
      "step": 511280
    },
    {
      "epoch": 0.2624781218710678,
      "grad_norm": 0.7265625,
      "learning_rate": 9.775197519751975e-05,
      "loss": 3.1987,
      "step": 511290
    },
    {
      "epoch": 0.26249676510722597,
      "grad_norm": 0.765625,
      "learning_rate": 9.774997499749974e-05,
      "loss": 3.1145,
      "step": 511300
    },
    {
      "epoch": 0.26251540834338416,
      "grad_norm": 0.76171875,
      "learning_rate": 9.774797479747975e-05,
      "loss": 3.2166,
      "step": 511310
    },
    {
      "epoch": 0.26253405157954235,
      "grad_norm": 0.80078125,
      "learning_rate": 9.774597459745975e-05,
      "loss": 3.1873,
      "step": 511320
    },
    {
      "epoch": 0.26255269481570054,
      "grad_norm": 0.828125,
      "learning_rate": 9.774397439743976e-05,
      "loss": 3.0325,
      "step": 511330
    },
    {
      "epoch": 0.2625713380518587,
      "grad_norm": 0.71484375,
      "learning_rate": 9.774197419741975e-05,
      "loss": 3.1956,
      "step": 511340
    },
    {
      "epoch": 0.2625899812880169,
      "grad_norm": 0.70703125,
      "learning_rate": 9.773997399739975e-05,
      "loss": 3.1234,
      "step": 511350
    },
    {
      "epoch": 0.2626086245241751,
      "grad_norm": 0.68359375,
      "learning_rate": 9.773797379737974e-05,
      "loss": 3.2115,
      "step": 511360
    },
    {
      "epoch": 0.2626272677603333,
      "grad_norm": 0.70703125,
      "learning_rate": 9.773597359735974e-05,
      "loss": 3.2228,
      "step": 511370
    },
    {
      "epoch": 0.2626459109964915,
      "grad_norm": 0.80078125,
      "learning_rate": 9.773397339733974e-05,
      "loss": 3.24,
      "step": 511380
    },
    {
      "epoch": 0.2626645542326496,
      "grad_norm": 0.75390625,
      "learning_rate": 9.773197319731973e-05,
      "loss": 3.1162,
      "step": 511390
    },
    {
      "epoch": 0.2626831974688078,
      "grad_norm": 0.8203125,
      "learning_rate": 9.772997299729974e-05,
      "loss": 3.1793,
      "step": 511400
    },
    {
      "epoch": 0.262701840704966,
      "grad_norm": 0.6875,
      "learning_rate": 9.772797279727974e-05,
      "loss": 3.1973,
      "step": 511410
    },
    {
      "epoch": 0.2627204839411242,
      "grad_norm": 0.68359375,
      "learning_rate": 9.772597259725973e-05,
      "loss": 3.2205,
      "step": 511420
    },
    {
      "epoch": 0.26273912717728237,
      "grad_norm": 0.8515625,
      "learning_rate": 9.772397239723973e-05,
      "loss": 3.0074,
      "step": 511430
    },
    {
      "epoch": 0.26275777041344056,
      "grad_norm": 0.75,
      "learning_rate": 9.772197219721972e-05,
      "loss": 3.1592,
      "step": 511440
    },
    {
      "epoch": 0.26277641364959875,
      "grad_norm": 0.765625,
      "learning_rate": 9.771997199719972e-05,
      "loss": 3.2098,
      "step": 511450
    },
    {
      "epoch": 0.26279505688575694,
      "grad_norm": 0.703125,
      "learning_rate": 9.771797179717971e-05,
      "loss": 3.2261,
      "step": 511460
    },
    {
      "epoch": 0.26281370012191513,
      "grad_norm": 0.73828125,
      "learning_rate": 9.771597159715971e-05,
      "loss": 3.1663,
      "step": 511470
    },
    {
      "epoch": 0.2628323433580733,
      "grad_norm": 0.70703125,
      "learning_rate": 9.771397139713972e-05,
      "loss": 3.1885,
      "step": 511480
    },
    {
      "epoch": 0.2628509865942315,
      "grad_norm": 0.76171875,
      "learning_rate": 9.771197119711971e-05,
      "loss": 3.3436,
      "step": 511490
    },
    {
      "epoch": 0.26286962983038964,
      "grad_norm": 0.67578125,
      "learning_rate": 9.770997099709972e-05,
      "loss": 3.2628,
      "step": 511500
    },
    {
      "epoch": 0.26288827306654783,
      "grad_norm": 0.828125,
      "learning_rate": 9.770797079707972e-05,
      "loss": 3.2353,
      "step": 511510
    },
    {
      "epoch": 0.262906916302706,
      "grad_norm": 0.62890625,
      "learning_rate": 9.770597059705972e-05,
      "loss": 3.2201,
      "step": 511520
    },
    {
      "epoch": 0.2629255595388642,
      "grad_norm": 0.75390625,
      "learning_rate": 9.770397039703971e-05,
      "loss": 3.1869,
      "step": 511530
    },
    {
      "epoch": 0.2629442027750224,
      "grad_norm": 0.66796875,
      "learning_rate": 9.77019701970197e-05,
      "loss": 3.2649,
      "step": 511540
    },
    {
      "epoch": 0.2629628460111806,
      "grad_norm": 1.0,
      "learning_rate": 9.76999699969997e-05,
      "loss": 3.3559,
      "step": 511550
    },
    {
      "epoch": 0.2629814892473388,
      "grad_norm": 0.71875,
      "learning_rate": 9.769796979697971e-05,
      "loss": 3.0883,
      "step": 511560
    },
    {
      "epoch": 0.26300013248349696,
      "grad_norm": 0.796875,
      "learning_rate": 9.76959695969597e-05,
      "loss": 3.1829,
      "step": 511570
    },
    {
      "epoch": 0.26301877571965515,
      "grad_norm": 0.81640625,
      "learning_rate": 9.76939693969397e-05,
      "loss": 3.2228,
      "step": 511580
    },
    {
      "epoch": 0.26303741895581334,
      "grad_norm": 0.69140625,
      "learning_rate": 9.76919691969197e-05,
      "loss": 3.2768,
      "step": 511590
    },
    {
      "epoch": 0.2630560621919715,
      "grad_norm": 0.6328125,
      "learning_rate": 9.76899689968997e-05,
      "loss": 3.1439,
      "step": 511600
    },
    {
      "epoch": 0.26307470542812966,
      "grad_norm": 0.73046875,
      "learning_rate": 9.768796879687969e-05,
      "loss": 3.1867,
      "step": 511610
    },
    {
      "epoch": 0.26309334866428785,
      "grad_norm": 0.8125,
      "learning_rate": 9.768596859685968e-05,
      "loss": 3.1288,
      "step": 511620
    },
    {
      "epoch": 0.26311199190044604,
      "grad_norm": 0.8125,
      "learning_rate": 9.768396839683968e-05,
      "loss": 3.1775,
      "step": 511630
    },
    {
      "epoch": 0.26313063513660423,
      "grad_norm": 0.82421875,
      "learning_rate": 9.768196819681969e-05,
      "loss": 3.1521,
      "step": 511640
    },
    {
      "epoch": 0.2631492783727624,
      "grad_norm": 0.73828125,
      "learning_rate": 9.767996799679968e-05,
      "loss": 3.0643,
      "step": 511650
    },
    {
      "epoch": 0.2631679216089206,
      "grad_norm": 0.78125,
      "learning_rate": 9.767796779677968e-05,
      "loss": 3.2555,
      "step": 511660
    },
    {
      "epoch": 0.2631865648450788,
      "grad_norm": 0.6875,
      "learning_rate": 9.767596759675968e-05,
      "loss": 3.0872,
      "step": 511670
    },
    {
      "epoch": 0.263205208081237,
      "grad_norm": 0.71875,
      "learning_rate": 9.767396739673967e-05,
      "loss": 3.1392,
      "step": 511680
    },
    {
      "epoch": 0.2632238513173952,
      "grad_norm": 0.83984375,
      "learning_rate": 9.767196719671967e-05,
      "loss": 3.0574,
      "step": 511690
    },
    {
      "epoch": 0.26324249455355336,
      "grad_norm": 0.71484375,
      "learning_rate": 9.766996699669968e-05,
      "loss": 3.1588,
      "step": 511700
    },
    {
      "epoch": 0.2632611377897115,
      "grad_norm": 0.75,
      "learning_rate": 9.766796679667967e-05,
      "loss": 3.0912,
      "step": 511710
    },
    {
      "epoch": 0.2632797810258697,
      "grad_norm": 0.67578125,
      "learning_rate": 9.766596659665968e-05,
      "loss": 3.1742,
      "step": 511720
    },
    {
      "epoch": 0.2632984242620279,
      "grad_norm": 0.66796875,
      "learning_rate": 9.766396639663968e-05,
      "loss": 3.2759,
      "step": 511730
    },
    {
      "epoch": 0.26331706749818606,
      "grad_norm": 0.73828125,
      "learning_rate": 9.766196619661967e-05,
      "loss": 3.3217,
      "step": 511740
    },
    {
      "epoch": 0.26333571073434425,
      "grad_norm": 0.71875,
      "learning_rate": 9.765996599659967e-05,
      "loss": 3.1706,
      "step": 511750
    },
    {
      "epoch": 0.26335435397050244,
      "grad_norm": 0.7109375,
      "learning_rate": 9.765796579657966e-05,
      "loss": 3.2075,
      "step": 511760
    },
    {
      "epoch": 0.26337299720666063,
      "grad_norm": 0.7265625,
      "learning_rate": 9.765596559655966e-05,
      "loss": 3.2729,
      "step": 511770
    },
    {
      "epoch": 0.2633916404428188,
      "grad_norm": 0.76171875,
      "learning_rate": 9.765396539653965e-05,
      "loss": 3.1167,
      "step": 511780
    },
    {
      "epoch": 0.263410283678977,
      "grad_norm": 0.69921875,
      "learning_rate": 9.765196519651965e-05,
      "loss": 3.1597,
      "step": 511790
    },
    {
      "epoch": 0.2634289269151352,
      "grad_norm": 0.80859375,
      "learning_rate": 9.764996499649966e-05,
      "loss": 3.1598,
      "step": 511800
    },
    {
      "epoch": 0.26344757015129333,
      "grad_norm": 0.75,
      "learning_rate": 9.764796479647966e-05,
      "loss": 3.3061,
      "step": 511810
    },
    {
      "epoch": 0.2634662133874515,
      "grad_norm": 0.8203125,
      "learning_rate": 9.764596459645965e-05,
      "loss": 3.1325,
      "step": 511820
    },
    {
      "epoch": 0.2634848566236097,
      "grad_norm": 0.79296875,
      "learning_rate": 9.764396439643965e-05,
      "loss": 3.1997,
      "step": 511830
    },
    {
      "epoch": 0.2635034998597679,
      "grad_norm": 0.70703125,
      "learning_rate": 9.764196419641964e-05,
      "loss": 3.1456,
      "step": 511840
    },
    {
      "epoch": 0.2635221430959261,
      "grad_norm": 0.8359375,
      "learning_rate": 9.763996399639964e-05,
      "loss": 3.2364,
      "step": 511850
    },
    {
      "epoch": 0.2635407863320843,
      "grad_norm": 0.76953125,
      "learning_rate": 9.763796379637963e-05,
      "loss": 3.2643,
      "step": 511860
    },
    {
      "epoch": 0.26355942956824246,
      "grad_norm": 0.69921875,
      "learning_rate": 9.763596359635964e-05,
      "loss": 3.2159,
      "step": 511870
    },
    {
      "epoch": 0.26357807280440065,
      "grad_norm": 0.6796875,
      "learning_rate": 9.763396339633964e-05,
      "loss": 3.0981,
      "step": 511880
    },
    {
      "epoch": 0.26359671604055884,
      "grad_norm": 0.65234375,
      "learning_rate": 9.763196319631963e-05,
      "loss": 3.2418,
      "step": 511890
    },
    {
      "epoch": 0.26361535927671703,
      "grad_norm": 0.7890625,
      "learning_rate": 9.762996299629964e-05,
      "loss": 3.0987,
      "step": 511900
    },
    {
      "epoch": 0.2636340025128752,
      "grad_norm": 0.78515625,
      "learning_rate": 9.762796279627964e-05,
      "loss": 3.1886,
      "step": 511910
    },
    {
      "epoch": 0.26365264574903335,
      "grad_norm": 0.71875,
      "learning_rate": 9.762596259625963e-05,
      "loss": 3.1527,
      "step": 511920
    },
    {
      "epoch": 0.26367128898519154,
      "grad_norm": 0.78125,
      "learning_rate": 9.762396239623963e-05,
      "loss": 3.355,
      "step": 511930
    },
    {
      "epoch": 0.26368993222134973,
      "grad_norm": 0.71875,
      "learning_rate": 9.762196219621962e-05,
      "loss": 3.2325,
      "step": 511940
    },
    {
      "epoch": 0.2637085754575079,
      "grad_norm": 0.859375,
      "learning_rate": 9.761996199619962e-05,
      "loss": 3.1637,
      "step": 511950
    },
    {
      "epoch": 0.2637272186936661,
      "grad_norm": 0.75390625,
      "learning_rate": 9.761796179617963e-05,
      "loss": 3.2655,
      "step": 511960
    },
    {
      "epoch": 0.2637458619298243,
      "grad_norm": 0.64453125,
      "learning_rate": 9.761596159615963e-05,
      "loss": 3.2039,
      "step": 511970
    },
    {
      "epoch": 0.2637645051659825,
      "grad_norm": 0.70703125,
      "learning_rate": 9.761396139613962e-05,
      "loss": 3.2106,
      "step": 511980
    },
    {
      "epoch": 0.2637831484021407,
      "grad_norm": 0.69921875,
      "learning_rate": 9.761196119611962e-05,
      "loss": 3.2723,
      "step": 511990
    },
    {
      "epoch": 0.26380179163829887,
      "grad_norm": 0.67578125,
      "learning_rate": 9.760996099609961e-05,
      "loss": 3.2364,
      "step": 512000
    },
    {
      "epoch": 0.26382043487445705,
      "grad_norm": 0.77734375,
      "learning_rate": 9.760796079607961e-05,
      "loss": 3.2332,
      "step": 512010
    },
    {
      "epoch": 0.2638390781106152,
      "grad_norm": 0.73046875,
      "learning_rate": 9.76059605960596e-05,
      "loss": 3.1355,
      "step": 512020
    },
    {
      "epoch": 0.2638577213467734,
      "grad_norm": 0.7890625,
      "learning_rate": 9.760396039603961e-05,
      "loss": 3.2676,
      "step": 512030
    },
    {
      "epoch": 0.26387636458293157,
      "grad_norm": 0.734375,
      "learning_rate": 9.760196019601961e-05,
      "loss": 3.203,
      "step": 512040
    },
    {
      "epoch": 0.26389500781908976,
      "grad_norm": 0.6875,
      "learning_rate": 9.75999599959996e-05,
      "loss": 3.1937,
      "step": 512050
    },
    {
      "epoch": 0.26391365105524794,
      "grad_norm": 0.734375,
      "learning_rate": 9.75979597959796e-05,
      "loss": 3.1692,
      "step": 512060
    },
    {
      "epoch": 0.26393229429140613,
      "grad_norm": 0.78515625,
      "learning_rate": 9.75959595959596e-05,
      "loss": 3.1705,
      "step": 512070
    },
    {
      "epoch": 0.2639509375275643,
      "grad_norm": 0.7578125,
      "learning_rate": 9.759395939593959e-05,
      "loss": 3.2035,
      "step": 512080
    },
    {
      "epoch": 0.2639695807637225,
      "grad_norm": 0.7109375,
      "learning_rate": 9.759195919591959e-05,
      "loss": 3.191,
      "step": 512090
    },
    {
      "epoch": 0.2639882239998807,
      "grad_norm": 0.65234375,
      "learning_rate": 9.75899589958996e-05,
      "loss": 3.166,
      "step": 512100
    },
    {
      "epoch": 0.2640068672360389,
      "grad_norm": 0.6953125,
      "learning_rate": 9.758795879587959e-05,
      "loss": 3.2161,
      "step": 512110
    },
    {
      "epoch": 0.2640255104721971,
      "grad_norm": 0.87890625,
      "learning_rate": 9.75859585958596e-05,
      "loss": 3.1368,
      "step": 512120
    },
    {
      "epoch": 0.2640441537083552,
      "grad_norm": 0.6875,
      "learning_rate": 9.75839583958396e-05,
      "loss": 3.1831,
      "step": 512130
    },
    {
      "epoch": 0.2640627969445134,
      "grad_norm": 0.78125,
      "learning_rate": 9.758195819581959e-05,
      "loss": 3.1779,
      "step": 512140
    },
    {
      "epoch": 0.2640814401806716,
      "grad_norm": 0.70703125,
      "learning_rate": 9.757995799579959e-05,
      "loss": 3.1948,
      "step": 512150
    },
    {
      "epoch": 0.2641000834168298,
      "grad_norm": 0.73046875,
      "learning_rate": 9.757795779577958e-05,
      "loss": 3.1845,
      "step": 512160
    },
    {
      "epoch": 0.26411872665298797,
      "grad_norm": 0.74609375,
      "learning_rate": 9.757595759575958e-05,
      "loss": 3.069,
      "step": 512170
    },
    {
      "epoch": 0.26413736988914616,
      "grad_norm": 0.74609375,
      "learning_rate": 9.757395739573957e-05,
      "loss": 3.1978,
      "step": 512180
    },
    {
      "epoch": 0.26415601312530435,
      "grad_norm": 0.72265625,
      "learning_rate": 9.757195719571958e-05,
      "loss": 3.23,
      "step": 512190
    },
    {
      "epoch": 0.26417465636146253,
      "grad_norm": 0.68359375,
      "learning_rate": 9.756995699569958e-05,
      "loss": 3.3008,
      "step": 512200
    },
    {
      "epoch": 0.2641932995976207,
      "grad_norm": 0.72265625,
      "learning_rate": 9.756795679567957e-05,
      "loss": 3.2691,
      "step": 512210
    },
    {
      "epoch": 0.2642119428337789,
      "grad_norm": 0.83984375,
      "learning_rate": 9.756595659565957e-05,
      "loss": 3.1816,
      "step": 512220
    },
    {
      "epoch": 0.26423058606993705,
      "grad_norm": 0.765625,
      "learning_rate": 9.756395639563956e-05,
      "loss": 3.2267,
      "step": 512230
    },
    {
      "epoch": 0.26424922930609523,
      "grad_norm": 0.765625,
      "learning_rate": 9.756195619561956e-05,
      "loss": 3.1797,
      "step": 512240
    },
    {
      "epoch": 0.2642678725422534,
      "grad_norm": 0.796875,
      "learning_rate": 9.755995599559956e-05,
      "loss": 3.1443,
      "step": 512250
    },
    {
      "epoch": 0.2642865157784116,
      "grad_norm": 0.734375,
      "learning_rate": 9.755795579557957e-05,
      "loss": 3.2434,
      "step": 512260
    },
    {
      "epoch": 0.2643051590145698,
      "grad_norm": 0.69921875,
      "learning_rate": 9.755595559555956e-05,
      "loss": 3.1579,
      "step": 512270
    },
    {
      "epoch": 0.264323802250728,
      "grad_norm": 0.68359375,
      "learning_rate": 9.755395539553956e-05,
      "loss": 3.1865,
      "step": 512280
    },
    {
      "epoch": 0.2643424454868862,
      "grad_norm": 0.6640625,
      "learning_rate": 9.755195519551957e-05,
      "loss": 3.0638,
      "step": 512290
    },
    {
      "epoch": 0.26436108872304437,
      "grad_norm": 0.7421875,
      "learning_rate": 9.754995499549956e-05,
      "loss": 3.2406,
      "step": 512300
    },
    {
      "epoch": 0.26437973195920256,
      "grad_norm": 0.76953125,
      "learning_rate": 9.754795479547956e-05,
      "loss": 3.1554,
      "step": 512310
    },
    {
      "epoch": 0.26439837519536075,
      "grad_norm": 0.6953125,
      "learning_rate": 9.754595459545955e-05,
      "loss": 3.1482,
      "step": 512320
    },
    {
      "epoch": 0.2644170184315189,
      "grad_norm": 0.68359375,
      "learning_rate": 9.754395439543955e-05,
      "loss": 3.3367,
      "step": 512330
    },
    {
      "epoch": 0.26443566166767707,
      "grad_norm": 0.78515625,
      "learning_rate": 9.754195419541954e-05,
      "loss": 3.3039,
      "step": 512340
    },
    {
      "epoch": 0.26445430490383526,
      "grad_norm": 0.78125,
      "learning_rate": 9.753995399539955e-05,
      "loss": 3.2968,
      "step": 512350
    },
    {
      "epoch": 0.26447294813999345,
      "grad_norm": 0.6328125,
      "learning_rate": 9.753795379537955e-05,
      "loss": 3.23,
      "step": 512360
    },
    {
      "epoch": 0.26449159137615164,
      "grad_norm": 0.77734375,
      "learning_rate": 9.753595359535954e-05,
      "loss": 3.2927,
      "step": 512370
    },
    {
      "epoch": 0.2645102346123098,
      "grad_norm": 0.69921875,
      "learning_rate": 9.753395339533954e-05,
      "loss": 3.1806,
      "step": 512380
    },
    {
      "epoch": 0.264528877848468,
      "grad_norm": 0.6875,
      "learning_rate": 9.753195319531954e-05,
      "loss": 3.2166,
      "step": 512390
    },
    {
      "epoch": 0.2645475210846262,
      "grad_norm": 0.703125,
      "learning_rate": 9.752995299529953e-05,
      "loss": 3.2325,
      "step": 512400
    },
    {
      "epoch": 0.2645661643207844,
      "grad_norm": 0.79296875,
      "learning_rate": 9.752795279527953e-05,
      "loss": 3.0039,
      "step": 512410
    },
    {
      "epoch": 0.2645848075569426,
      "grad_norm": 0.734375,
      "learning_rate": 9.752595259525954e-05,
      "loss": 3.1944,
      "step": 512420
    },
    {
      "epoch": 0.26460345079310077,
      "grad_norm": 0.6484375,
      "learning_rate": 9.752395239523953e-05,
      "loss": 3.2004,
      "step": 512430
    },
    {
      "epoch": 0.2646220940292589,
      "grad_norm": 0.8203125,
      "learning_rate": 9.752195219521953e-05,
      "loss": 3.2572,
      "step": 512440
    },
    {
      "epoch": 0.2646407372654171,
      "grad_norm": 0.76171875,
      "learning_rate": 9.751995199519952e-05,
      "loss": 3.171,
      "step": 512450
    },
    {
      "epoch": 0.2646593805015753,
      "grad_norm": 0.828125,
      "learning_rate": 9.751795179517952e-05,
      "loss": 3.1966,
      "step": 512460
    },
    {
      "epoch": 0.26467802373773347,
      "grad_norm": 0.734375,
      "learning_rate": 9.751595159515951e-05,
      "loss": 3.2361,
      "step": 512470
    },
    {
      "epoch": 0.26469666697389166,
      "grad_norm": 0.734375,
      "learning_rate": 9.751395139513951e-05,
      "loss": 3.2509,
      "step": 512480
    },
    {
      "epoch": 0.26471531021004985,
      "grad_norm": 0.734375,
      "learning_rate": 9.751195119511952e-05,
      "loss": 3.2598,
      "step": 512490
    },
    {
      "epoch": 0.26473395344620804,
      "grad_norm": 0.6953125,
      "learning_rate": 9.750995099509951e-05,
      "loss": 3.2397,
      "step": 512500
    },
    {
      "epoch": 0.2647525966823662,
      "grad_norm": 0.7578125,
      "learning_rate": 9.750795079507952e-05,
      "loss": 3.1576,
      "step": 512510
    },
    {
      "epoch": 0.2647712399185244,
      "grad_norm": 0.74609375,
      "learning_rate": 9.750595059505952e-05,
      "loss": 3.155,
      "step": 512520
    },
    {
      "epoch": 0.2647898831546826,
      "grad_norm": 0.70703125,
      "learning_rate": 9.750395039503951e-05,
      "loss": 3.2882,
      "step": 512530
    },
    {
      "epoch": 0.26480852639084074,
      "grad_norm": 0.8046875,
      "learning_rate": 9.750195019501951e-05,
      "loss": 3.284,
      "step": 512540
    },
    {
      "epoch": 0.2648271696269989,
      "grad_norm": 0.75390625,
      "learning_rate": 9.74999499949995e-05,
      "loss": 3.1055,
      "step": 512550
    },
    {
      "epoch": 0.2648458128631571,
      "grad_norm": 0.625,
      "learning_rate": 9.74979497949795e-05,
      "loss": 3.2273,
      "step": 512560
    },
    {
      "epoch": 0.2648644560993153,
      "grad_norm": 0.80859375,
      "learning_rate": 9.74959495949595e-05,
      "loss": 3.2299,
      "step": 512570
    },
    {
      "epoch": 0.2648830993354735,
      "grad_norm": 0.6640625,
      "learning_rate": 9.74939493949395e-05,
      "loss": 3.2616,
      "step": 512580
    },
    {
      "epoch": 0.2649017425716317,
      "grad_norm": 0.6796875,
      "learning_rate": 9.74919491949195e-05,
      "loss": 3.2348,
      "step": 512590
    },
    {
      "epoch": 0.26492038580778987,
      "grad_norm": 0.77734375,
      "learning_rate": 9.74899489948995e-05,
      "loss": 3.2799,
      "step": 512600
    },
    {
      "epoch": 0.26493902904394806,
      "grad_norm": 0.84375,
      "learning_rate": 9.748794879487949e-05,
      "loss": 3.1835,
      "step": 512610
    },
    {
      "epoch": 0.26495767228010625,
      "grad_norm": 0.703125,
      "learning_rate": 9.748594859485949e-05,
      "loss": 3.3163,
      "step": 512620
    },
    {
      "epoch": 0.26497631551626444,
      "grad_norm": 0.7421875,
      "learning_rate": 9.748394839483948e-05,
      "loss": 3.1937,
      "step": 512630
    },
    {
      "epoch": 0.2649949587524226,
      "grad_norm": 0.75390625,
      "learning_rate": 9.748194819481948e-05,
      "loss": 3.1096,
      "step": 512640
    },
    {
      "epoch": 0.26501360198858076,
      "grad_norm": 0.76171875,
      "learning_rate": 9.747994799479947e-05,
      "loss": 3.2369,
      "step": 512650
    },
    {
      "epoch": 0.26503224522473895,
      "grad_norm": 0.6953125,
      "learning_rate": 9.747794779477948e-05,
      "loss": 3.2573,
      "step": 512660
    },
    {
      "epoch": 0.26505088846089714,
      "grad_norm": 0.72265625,
      "learning_rate": 9.747594759475948e-05,
      "loss": 3.2855,
      "step": 512670
    },
    {
      "epoch": 0.2650695316970553,
      "grad_norm": 0.76953125,
      "learning_rate": 9.747394739473948e-05,
      "loss": 3.1892,
      "step": 512680
    },
    {
      "epoch": 0.2650881749332135,
      "grad_norm": 0.75,
      "learning_rate": 9.747194719471948e-05,
      "loss": 3.218,
      "step": 512690
    },
    {
      "epoch": 0.2651068181693717,
      "grad_norm": 0.6875,
      "learning_rate": 9.746994699469948e-05,
      "loss": 3.2695,
      "step": 512700
    },
    {
      "epoch": 0.2651254614055299,
      "grad_norm": 0.73828125,
      "learning_rate": 9.746794679467948e-05,
      "loss": 3.2943,
      "step": 512710
    },
    {
      "epoch": 0.2651441046416881,
      "grad_norm": 0.75390625,
      "learning_rate": 9.746594659465947e-05,
      "loss": 3.3061,
      "step": 512720
    },
    {
      "epoch": 0.26516274787784627,
      "grad_norm": 0.71484375,
      "learning_rate": 9.746394639463947e-05,
      "loss": 3.2209,
      "step": 512730
    },
    {
      "epoch": 0.26518139111400446,
      "grad_norm": 0.77734375,
      "learning_rate": 9.746194619461946e-05,
      "loss": 3.1262,
      "step": 512740
    },
    {
      "epoch": 0.2652000343501626,
      "grad_norm": 0.75390625,
      "learning_rate": 9.745994599459947e-05,
      "loss": 3.2175,
      "step": 512750
    },
    {
      "epoch": 0.2652186775863208,
      "grad_norm": 0.75,
      "learning_rate": 9.745794579457947e-05,
      "loss": 3.2692,
      "step": 512760
    },
    {
      "epoch": 0.265237320822479,
      "grad_norm": 0.796875,
      "learning_rate": 9.745594559455946e-05,
      "loss": 3.2009,
      "step": 512770
    },
    {
      "epoch": 0.26525596405863716,
      "grad_norm": 0.61328125,
      "learning_rate": 9.745394539453946e-05,
      "loss": 3.2494,
      "step": 512780
    },
    {
      "epoch": 0.26527460729479535,
      "grad_norm": 0.74609375,
      "learning_rate": 9.745194519451945e-05,
      "loss": 3.2185,
      "step": 512790
    },
    {
      "epoch": 0.26529325053095354,
      "grad_norm": 0.734375,
      "learning_rate": 9.744994499449945e-05,
      "loss": 3.1883,
      "step": 512800
    },
    {
      "epoch": 0.26531189376711173,
      "grad_norm": 0.765625,
      "learning_rate": 9.744794479447944e-05,
      "loss": 3.1697,
      "step": 512810
    },
    {
      "epoch": 0.2653305370032699,
      "grad_norm": 0.7734375,
      "learning_rate": 9.744594459445945e-05,
      "loss": 3.1667,
      "step": 512820
    },
    {
      "epoch": 0.2653491802394281,
      "grad_norm": 0.7578125,
      "learning_rate": 9.744394439443945e-05,
      "loss": 3.1892,
      "step": 512830
    },
    {
      "epoch": 0.2653678234755863,
      "grad_norm": 0.828125,
      "learning_rate": 9.744194419441945e-05,
      "loss": 3.3762,
      "step": 512840
    },
    {
      "epoch": 0.2653864667117445,
      "grad_norm": 0.828125,
      "learning_rate": 9.743994399439944e-05,
      "loss": 3.183,
      "step": 512850
    },
    {
      "epoch": 0.2654051099479026,
      "grad_norm": 0.703125,
      "learning_rate": 9.743794379437944e-05,
      "loss": 3.2109,
      "step": 512860
    },
    {
      "epoch": 0.2654237531840608,
      "grad_norm": 0.734375,
      "learning_rate": 9.743594359435943e-05,
      "loss": 3.2836,
      "step": 512870
    },
    {
      "epoch": 0.265442396420219,
      "grad_norm": 0.828125,
      "learning_rate": 9.743394339433943e-05,
      "loss": 3.3224,
      "step": 512880
    },
    {
      "epoch": 0.2654610396563772,
      "grad_norm": 0.74609375,
      "learning_rate": 9.743194319431944e-05,
      "loss": 3.2722,
      "step": 512890
    },
    {
      "epoch": 0.2654796828925354,
      "grad_norm": 0.78515625,
      "learning_rate": 9.742994299429943e-05,
      "loss": 3.1185,
      "step": 512900
    },
    {
      "epoch": 0.26549832612869356,
      "grad_norm": 0.78515625,
      "learning_rate": 9.742794279427944e-05,
      "loss": 3.1711,
      "step": 512910
    },
    {
      "epoch": 0.26551696936485175,
      "grad_norm": 0.796875,
      "learning_rate": 9.742594259425944e-05,
      "loss": 3.3007,
      "step": 512920
    },
    {
      "epoch": 0.26553561260100994,
      "grad_norm": 0.734375,
      "learning_rate": 9.742394239423943e-05,
      "loss": 3.3174,
      "step": 512930
    },
    {
      "epoch": 0.26555425583716813,
      "grad_norm": 0.73828125,
      "learning_rate": 9.742194219421943e-05,
      "loss": 3.2093,
      "step": 512940
    },
    {
      "epoch": 0.2655728990733263,
      "grad_norm": 0.7578125,
      "learning_rate": 9.741994199419942e-05,
      "loss": 3.2873,
      "step": 512950
    },
    {
      "epoch": 0.26559154230948445,
      "grad_norm": 0.83984375,
      "learning_rate": 9.741794179417942e-05,
      "loss": 3.2296,
      "step": 512960
    },
    {
      "epoch": 0.26561018554564264,
      "grad_norm": 0.70703125,
      "learning_rate": 9.741594159415942e-05,
      "loss": 3.1962,
      "step": 512970
    },
    {
      "epoch": 0.26562882878180083,
      "grad_norm": 0.796875,
      "learning_rate": 9.741394139413942e-05,
      "loss": 3.2071,
      "step": 512980
    },
    {
      "epoch": 0.265647472017959,
      "grad_norm": 0.79296875,
      "learning_rate": 9.741194119411942e-05,
      "loss": 3.1917,
      "step": 512990
    },
    {
      "epoch": 0.2656661152541172,
      "grad_norm": 0.75,
      "learning_rate": 9.740994099409942e-05,
      "loss": 3.2436,
      "step": 513000
    },
    {
      "epoch": 0.2656847584902754,
      "grad_norm": 0.7265625,
      "learning_rate": 9.740794079407941e-05,
      "loss": 3.1754,
      "step": 513010
    },
    {
      "epoch": 0.2657034017264336,
      "grad_norm": 0.75390625,
      "learning_rate": 9.74059405940594e-05,
      "loss": 3.3098,
      "step": 513020
    },
    {
      "epoch": 0.2657220449625918,
      "grad_norm": 0.7109375,
      "learning_rate": 9.74039403940394e-05,
      "loss": 3.1939,
      "step": 513030
    },
    {
      "epoch": 0.26574068819874996,
      "grad_norm": 0.7421875,
      "learning_rate": 9.74019401940194e-05,
      "loss": 3.1467,
      "step": 513040
    },
    {
      "epoch": 0.26575933143490815,
      "grad_norm": 0.7109375,
      "learning_rate": 9.739993999399941e-05,
      "loss": 3.172,
      "step": 513050
    },
    {
      "epoch": 0.26577797467106634,
      "grad_norm": 0.6953125,
      "learning_rate": 9.73979397939794e-05,
      "loss": 3.2482,
      "step": 513060
    },
    {
      "epoch": 0.2657966179072245,
      "grad_norm": 0.8046875,
      "learning_rate": 9.73959395939594e-05,
      "loss": 3.358,
      "step": 513070
    },
    {
      "epoch": 0.26581526114338266,
      "grad_norm": 0.87109375,
      "learning_rate": 9.739393939393941e-05,
      "loss": 3.2023,
      "step": 513080
    },
    {
      "epoch": 0.26583390437954085,
      "grad_norm": 0.859375,
      "learning_rate": 9.73919391939194e-05,
      "loss": 3.1877,
      "step": 513090
    },
    {
      "epoch": 0.26585254761569904,
      "grad_norm": 0.76953125,
      "learning_rate": 9.73899389938994e-05,
      "loss": 3.0943,
      "step": 513100
    },
    {
      "epoch": 0.26587119085185723,
      "grad_norm": 0.7578125,
      "learning_rate": 9.73879387938794e-05,
      "loss": 3.1229,
      "step": 513110
    },
    {
      "epoch": 0.2658898340880154,
      "grad_norm": 0.625,
      "learning_rate": 9.738593859385939e-05,
      "loss": 3.2245,
      "step": 513120
    },
    {
      "epoch": 0.2659084773241736,
      "grad_norm": 0.703125,
      "learning_rate": 9.738393839383939e-05,
      "loss": 3.2282,
      "step": 513130
    },
    {
      "epoch": 0.2659271205603318,
      "grad_norm": 0.66015625,
      "learning_rate": 9.73819381938194e-05,
      "loss": 3.1252,
      "step": 513140
    },
    {
      "epoch": 0.26594576379649,
      "grad_norm": 0.7421875,
      "learning_rate": 9.737993799379939e-05,
      "loss": 3.1252,
      "step": 513150
    },
    {
      "epoch": 0.2659644070326482,
      "grad_norm": 0.76171875,
      "learning_rate": 9.737793779377939e-05,
      "loss": 3.2318,
      "step": 513160
    },
    {
      "epoch": 0.2659830502688063,
      "grad_norm": 0.71875,
      "learning_rate": 9.737593759375938e-05,
      "loss": 3.0499,
      "step": 513170
    },
    {
      "epoch": 0.2660016935049645,
      "grad_norm": 0.80078125,
      "learning_rate": 9.737393739373938e-05,
      "loss": 3.2003,
      "step": 513180
    },
    {
      "epoch": 0.2660203367411227,
      "grad_norm": 0.890625,
      "learning_rate": 9.737193719371937e-05,
      "loss": 3.0256,
      "step": 513190
    },
    {
      "epoch": 0.2660389799772809,
      "grad_norm": 0.70703125,
      "learning_rate": 9.736993699369937e-05,
      "loss": 3.2268,
      "step": 513200
    },
    {
      "epoch": 0.26605762321343907,
      "grad_norm": 0.8125,
      "learning_rate": 9.736793679367938e-05,
      "loss": 3.1439,
      "step": 513210
    },
    {
      "epoch": 0.26607626644959725,
      "grad_norm": 0.75,
      "learning_rate": 9.736593659365937e-05,
      "loss": 3.301,
      "step": 513220
    },
    {
      "epoch": 0.26609490968575544,
      "grad_norm": 0.71484375,
      "learning_rate": 9.736393639363937e-05,
      "loss": 3.0916,
      "step": 513230
    },
    {
      "epoch": 0.26611355292191363,
      "grad_norm": 0.8125,
      "learning_rate": 9.736193619361936e-05,
      "loss": 3.2342,
      "step": 513240
    },
    {
      "epoch": 0.2661321961580718,
      "grad_norm": 0.84765625,
      "learning_rate": 9.735993599359936e-05,
      "loss": 3.2666,
      "step": 513250
    },
    {
      "epoch": 0.26615083939423,
      "grad_norm": 0.71484375,
      "learning_rate": 9.735793579357936e-05,
      "loss": 3.1779,
      "step": 513260
    },
    {
      "epoch": 0.2661694826303882,
      "grad_norm": 0.734375,
      "learning_rate": 9.735593559355935e-05,
      "loss": 3.2517,
      "step": 513270
    },
    {
      "epoch": 0.26618812586654633,
      "grad_norm": 0.76171875,
      "learning_rate": 9.735393539353936e-05,
      "loss": 3.324,
      "step": 513280
    },
    {
      "epoch": 0.2662067691027045,
      "grad_norm": 0.6953125,
      "learning_rate": 9.735193519351936e-05,
      "loss": 3.1383,
      "step": 513290
    },
    {
      "epoch": 0.2662254123388627,
      "grad_norm": 0.6875,
      "learning_rate": 9.734993499349936e-05,
      "loss": 3.2508,
      "step": 513300
    },
    {
      "epoch": 0.2662440555750209,
      "grad_norm": 0.6484375,
      "learning_rate": 9.734793479347936e-05,
      "loss": 3.1685,
      "step": 513310
    },
    {
      "epoch": 0.2662626988111791,
      "grad_norm": 0.72265625,
      "learning_rate": 9.734593459345936e-05,
      "loss": 3.1583,
      "step": 513320
    },
    {
      "epoch": 0.2662813420473373,
      "grad_norm": 0.73046875,
      "learning_rate": 9.734393439343935e-05,
      "loss": 3.1842,
      "step": 513330
    },
    {
      "epoch": 0.26629998528349547,
      "grad_norm": 0.75390625,
      "learning_rate": 9.734193419341935e-05,
      "loss": 3.2122,
      "step": 513340
    },
    {
      "epoch": 0.26631862851965366,
      "grad_norm": 0.65625,
      "learning_rate": 9.733993399339934e-05,
      "loss": 3.2221,
      "step": 513350
    },
    {
      "epoch": 0.26633727175581184,
      "grad_norm": 0.765625,
      "learning_rate": 9.733793379337934e-05,
      "loss": 3.2956,
      "step": 513360
    },
    {
      "epoch": 0.26635591499197003,
      "grad_norm": 0.6953125,
      "learning_rate": 9.733593359335935e-05,
      "loss": 3.117,
      "step": 513370
    },
    {
      "epoch": 0.26637455822812817,
      "grad_norm": 0.8359375,
      "learning_rate": 9.733393339333934e-05,
      "loss": 3.248,
      "step": 513380
    },
    {
      "epoch": 0.26639320146428636,
      "grad_norm": 0.69140625,
      "learning_rate": 9.733193319331934e-05,
      "loss": 3.117,
      "step": 513390
    },
    {
      "epoch": 0.26641184470044454,
      "grad_norm": 0.79296875,
      "learning_rate": 9.732993299329933e-05,
      "loss": 3.1472,
      "step": 513400
    },
    {
      "epoch": 0.26643048793660273,
      "grad_norm": 0.79296875,
      "learning_rate": 9.732793279327933e-05,
      "loss": 3.0836,
      "step": 513410
    },
    {
      "epoch": 0.2664491311727609,
      "grad_norm": 0.69921875,
      "learning_rate": 9.732593259325933e-05,
      "loss": 3.2475,
      "step": 513420
    },
    {
      "epoch": 0.2664677744089191,
      "grad_norm": 0.73046875,
      "learning_rate": 9.732393239323932e-05,
      "loss": 3.1662,
      "step": 513430
    },
    {
      "epoch": 0.2664864176450773,
      "grad_norm": 0.72265625,
      "learning_rate": 9.732193219321932e-05,
      "loss": 3.3631,
      "step": 513440
    },
    {
      "epoch": 0.2665050608812355,
      "grad_norm": 0.75,
      "learning_rate": 9.731993199319933e-05,
      "loss": 3.1994,
      "step": 513450
    },
    {
      "epoch": 0.2665237041173937,
      "grad_norm": 0.71875,
      "learning_rate": 9.731793179317932e-05,
      "loss": 3.3292,
      "step": 513460
    },
    {
      "epoch": 0.26654234735355187,
      "grad_norm": 0.8125,
      "learning_rate": 9.731593159315932e-05,
      "loss": 3.3065,
      "step": 513470
    },
    {
      "epoch": 0.26656099058971006,
      "grad_norm": 0.7421875,
      "learning_rate": 9.731393139313933e-05,
      "loss": 3.1505,
      "step": 513480
    },
    {
      "epoch": 0.2665796338258682,
      "grad_norm": 0.7578125,
      "learning_rate": 9.731193119311932e-05,
      "loss": 3.2551,
      "step": 513490
    },
    {
      "epoch": 0.2665982770620264,
      "grad_norm": 0.734375,
      "learning_rate": 9.730993099309932e-05,
      "loss": 3.274,
      "step": 513500
    },
    {
      "epoch": 0.26661692029818457,
      "grad_norm": 0.69921875,
      "learning_rate": 9.730793079307931e-05,
      "loss": 3.1325,
      "step": 513510
    },
    {
      "epoch": 0.26663556353434276,
      "grad_norm": 0.6875,
      "learning_rate": 9.730593059305931e-05,
      "loss": 3.1782,
      "step": 513520
    },
    {
      "epoch": 0.26665420677050095,
      "grad_norm": 0.68359375,
      "learning_rate": 9.730393039303932e-05,
      "loss": 3.2625,
      "step": 513530
    },
    {
      "epoch": 0.26667285000665913,
      "grad_norm": 0.75,
      "learning_rate": 9.730193019301931e-05,
      "loss": 3.2234,
      "step": 513540
    },
    {
      "epoch": 0.2666914932428173,
      "grad_norm": 0.86328125,
      "learning_rate": 9.729992999299931e-05,
      "loss": 3.2026,
      "step": 513550
    },
    {
      "epoch": 0.2667101364789755,
      "grad_norm": 0.73828125,
      "learning_rate": 9.72979297929793e-05,
      "loss": 3.2295,
      "step": 513560
    },
    {
      "epoch": 0.2667287797151337,
      "grad_norm": 0.77734375,
      "learning_rate": 9.72959295929593e-05,
      "loss": 3.2275,
      "step": 513570
    },
    {
      "epoch": 0.2667474229512919,
      "grad_norm": 0.84375,
      "learning_rate": 9.72939293929393e-05,
      "loss": 3.3515,
      "step": 513580
    },
    {
      "epoch": 0.26676606618745,
      "grad_norm": 0.84765625,
      "learning_rate": 9.729192919291929e-05,
      "loss": 3.1878,
      "step": 513590
    },
    {
      "epoch": 0.2667847094236082,
      "grad_norm": 0.765625,
      "learning_rate": 9.728992899289929e-05,
      "loss": 3.1419,
      "step": 513600
    },
    {
      "epoch": 0.2668033526597664,
      "grad_norm": 0.83203125,
      "learning_rate": 9.72879287928793e-05,
      "loss": 3.166,
      "step": 513610
    },
    {
      "epoch": 0.2668219958959246,
      "grad_norm": 0.80859375,
      "learning_rate": 9.728592859285929e-05,
      "loss": 3.2225,
      "step": 513620
    },
    {
      "epoch": 0.2668406391320828,
      "grad_norm": 0.734375,
      "learning_rate": 9.728392839283929e-05,
      "loss": 3.2661,
      "step": 513630
    },
    {
      "epoch": 0.26685928236824097,
      "grad_norm": 0.69140625,
      "learning_rate": 9.728192819281928e-05,
      "loss": 3.2591,
      "step": 513640
    },
    {
      "epoch": 0.26687792560439916,
      "grad_norm": 0.79296875,
      "learning_rate": 9.727992799279928e-05,
      "loss": 3.1773,
      "step": 513650
    },
    {
      "epoch": 0.26689656884055735,
      "grad_norm": 0.8359375,
      "learning_rate": 9.727792779277927e-05,
      "loss": 3.2412,
      "step": 513660
    },
    {
      "epoch": 0.26691521207671554,
      "grad_norm": 0.74609375,
      "learning_rate": 9.727592759275928e-05,
      "loss": 3.0537,
      "step": 513670
    },
    {
      "epoch": 0.2669338553128737,
      "grad_norm": 0.76953125,
      "learning_rate": 9.727392739273928e-05,
      "loss": 3.1647,
      "step": 513680
    },
    {
      "epoch": 0.2669524985490319,
      "grad_norm": 0.7265625,
      "learning_rate": 9.727192719271929e-05,
      "loss": 3.1738,
      "step": 513690
    },
    {
      "epoch": 0.26697114178519005,
      "grad_norm": 0.84765625,
      "learning_rate": 9.726992699269928e-05,
      "loss": 3.0889,
      "step": 513700
    },
    {
      "epoch": 0.26698978502134824,
      "grad_norm": 0.8203125,
      "learning_rate": 9.726792679267928e-05,
      "loss": 3.0853,
      "step": 513710
    },
    {
      "epoch": 0.2670084282575064,
      "grad_norm": 0.69140625,
      "learning_rate": 9.726592659265927e-05,
      "loss": 3.2424,
      "step": 513720
    },
    {
      "epoch": 0.2670270714936646,
      "grad_norm": 0.6953125,
      "learning_rate": 9.726392639263927e-05,
      "loss": 3.1166,
      "step": 513730
    },
    {
      "epoch": 0.2670457147298228,
      "grad_norm": 0.734375,
      "learning_rate": 9.726192619261927e-05,
      "loss": 3.1564,
      "step": 513740
    },
    {
      "epoch": 0.267064357965981,
      "grad_norm": 0.76953125,
      "learning_rate": 9.725992599259926e-05,
      "loss": 3.2172,
      "step": 513750
    },
    {
      "epoch": 0.2670830012021392,
      "grad_norm": 0.72265625,
      "learning_rate": 9.725792579257926e-05,
      "loss": 3.2156,
      "step": 513760
    },
    {
      "epoch": 0.26710164443829737,
      "grad_norm": 0.73046875,
      "learning_rate": 9.725592559255927e-05,
      "loss": 3.1973,
      "step": 513770
    },
    {
      "epoch": 0.26712028767445556,
      "grad_norm": 0.74609375,
      "learning_rate": 9.725392539253926e-05,
      "loss": 3.2321,
      "step": 513780
    },
    {
      "epoch": 0.26713893091061375,
      "grad_norm": 0.65234375,
      "learning_rate": 9.725192519251926e-05,
      "loss": 3.259,
      "step": 513790
    },
    {
      "epoch": 0.2671575741467719,
      "grad_norm": 0.7890625,
      "learning_rate": 9.724992499249925e-05,
      "loss": 3.0335,
      "step": 513800
    },
    {
      "epoch": 0.26717621738293007,
      "grad_norm": 0.62890625,
      "learning_rate": 9.724792479247925e-05,
      "loss": 3.2649,
      "step": 513810
    },
    {
      "epoch": 0.26719486061908826,
      "grad_norm": 0.71875,
      "learning_rate": 9.724592459245924e-05,
      "loss": 3.1904,
      "step": 513820
    },
    {
      "epoch": 0.26721350385524645,
      "grad_norm": 0.69921875,
      "learning_rate": 9.724392439243924e-05,
      "loss": 3.1247,
      "step": 513830
    },
    {
      "epoch": 0.26723214709140464,
      "grad_norm": 0.7265625,
      "learning_rate": 9.724192419241925e-05,
      "loss": 3.0764,
      "step": 513840
    },
    {
      "epoch": 0.2672507903275628,
      "grad_norm": 0.73046875,
      "learning_rate": 9.723992399239924e-05,
      "loss": 3.2072,
      "step": 513850
    },
    {
      "epoch": 0.267269433563721,
      "grad_norm": 0.77734375,
      "learning_rate": 9.723792379237924e-05,
      "loss": 3.2149,
      "step": 513860
    },
    {
      "epoch": 0.2672880767998792,
      "grad_norm": 0.703125,
      "learning_rate": 9.723592359235925e-05,
      "loss": 3.2244,
      "step": 513870
    },
    {
      "epoch": 0.2673067200360374,
      "grad_norm": 0.6484375,
      "learning_rate": 9.723392339233924e-05,
      "loss": 3.1837,
      "step": 513880
    },
    {
      "epoch": 0.2673253632721956,
      "grad_norm": 0.73046875,
      "learning_rate": 9.723192319231924e-05,
      "loss": 3.2639,
      "step": 513890
    },
    {
      "epoch": 0.2673440065083537,
      "grad_norm": 0.75,
      "learning_rate": 9.722992299229924e-05,
      "loss": 3.3385,
      "step": 513900
    },
    {
      "epoch": 0.2673626497445119,
      "grad_norm": 0.66015625,
      "learning_rate": 9.722792279227923e-05,
      "loss": 3.2733,
      "step": 513910
    },
    {
      "epoch": 0.2673812929806701,
      "grad_norm": 0.6953125,
      "learning_rate": 9.722592259225923e-05,
      "loss": 3.2326,
      "step": 513920
    },
    {
      "epoch": 0.2673999362168283,
      "grad_norm": 0.65234375,
      "learning_rate": 9.722392239223924e-05,
      "loss": 3.2501,
      "step": 513930
    },
    {
      "epoch": 0.26741857945298647,
      "grad_norm": 0.76953125,
      "learning_rate": 9.722192219221923e-05,
      "loss": 3.2383,
      "step": 513940
    },
    {
      "epoch": 0.26743722268914466,
      "grad_norm": 0.7890625,
      "learning_rate": 9.721992199219923e-05,
      "loss": 3.2382,
      "step": 513950
    },
    {
      "epoch": 0.26745586592530285,
      "grad_norm": 0.78515625,
      "learning_rate": 9.721792179217922e-05,
      "loss": 3.0129,
      "step": 513960
    },
    {
      "epoch": 0.26747450916146104,
      "grad_norm": 0.6484375,
      "learning_rate": 9.721592159215922e-05,
      "loss": 3.216,
      "step": 513970
    },
    {
      "epoch": 0.2674931523976192,
      "grad_norm": 0.75,
      "learning_rate": 9.721392139213921e-05,
      "loss": 3.3135,
      "step": 513980
    },
    {
      "epoch": 0.2675117956337774,
      "grad_norm": 0.73046875,
      "learning_rate": 9.721192119211921e-05,
      "loss": 3.2499,
      "step": 513990
    },
    {
      "epoch": 0.2675304388699356,
      "grad_norm": 0.70703125,
      "learning_rate": 9.720992099209922e-05,
      "loss": 3.112,
      "step": 514000
    },
    {
      "epoch": 0.26754908210609374,
      "grad_norm": 0.73828125,
      "learning_rate": 9.720792079207921e-05,
      "loss": 3.2289,
      "step": 514010
    },
    {
      "epoch": 0.2675677253422519,
      "grad_norm": 0.71484375,
      "learning_rate": 9.720592059205921e-05,
      "loss": 3.073,
      "step": 514020
    },
    {
      "epoch": 0.2675863685784101,
      "grad_norm": 0.79296875,
      "learning_rate": 9.72039203920392e-05,
      "loss": 3.2814,
      "step": 514030
    },
    {
      "epoch": 0.2676050118145683,
      "grad_norm": 0.84375,
      "learning_rate": 9.72019201920192e-05,
      "loss": 3.2251,
      "step": 514040
    },
    {
      "epoch": 0.2676236550507265,
      "grad_norm": 0.7890625,
      "learning_rate": 9.71999199919992e-05,
      "loss": 3.1938,
      "step": 514050
    },
    {
      "epoch": 0.2676422982868847,
      "grad_norm": 0.75,
      "learning_rate": 9.719791979197919e-05,
      "loss": 3.0867,
      "step": 514060
    },
    {
      "epoch": 0.2676609415230429,
      "grad_norm": 0.71484375,
      "learning_rate": 9.71959195919592e-05,
      "loss": 3.2521,
      "step": 514070
    },
    {
      "epoch": 0.26767958475920106,
      "grad_norm": 0.73828125,
      "learning_rate": 9.71939193919392e-05,
      "loss": 3.1759,
      "step": 514080
    },
    {
      "epoch": 0.26769822799535925,
      "grad_norm": 0.734375,
      "learning_rate": 9.71919191919192e-05,
      "loss": 3.2396,
      "step": 514090
    },
    {
      "epoch": 0.26771687123151744,
      "grad_norm": 0.796875,
      "learning_rate": 9.71899189918992e-05,
      "loss": 3.2079,
      "step": 514100
    },
    {
      "epoch": 0.2677355144676756,
      "grad_norm": 0.6640625,
      "learning_rate": 9.71879187918792e-05,
      "loss": 3.3872,
      "step": 514110
    },
    {
      "epoch": 0.26775415770383376,
      "grad_norm": 0.71875,
      "learning_rate": 9.718591859185919e-05,
      "loss": 3.3233,
      "step": 514120
    },
    {
      "epoch": 0.26777280093999195,
      "grad_norm": 1.03125,
      "learning_rate": 9.718391839183919e-05,
      "loss": 3.224,
      "step": 514130
    },
    {
      "epoch": 0.26779144417615014,
      "grad_norm": 0.71875,
      "learning_rate": 9.718191819181918e-05,
      "loss": 3.1919,
      "step": 514140
    },
    {
      "epoch": 0.26781008741230833,
      "grad_norm": 1.0546875,
      "learning_rate": 9.717991799179918e-05,
      "loss": 3.2638,
      "step": 514150
    },
    {
      "epoch": 0.2678287306484665,
      "grad_norm": 0.76171875,
      "learning_rate": 9.717791779177919e-05,
      "loss": 3.2634,
      "step": 514160
    },
    {
      "epoch": 0.2678473738846247,
      "grad_norm": 0.72265625,
      "learning_rate": 9.717591759175918e-05,
      "loss": 3.2748,
      "step": 514170
    },
    {
      "epoch": 0.2678660171207829,
      "grad_norm": 0.74609375,
      "learning_rate": 9.717391739173918e-05,
      "loss": 3.2079,
      "step": 514180
    },
    {
      "epoch": 0.2678846603569411,
      "grad_norm": 0.73046875,
      "learning_rate": 9.717191719171918e-05,
      "loss": 3.1702,
      "step": 514190
    },
    {
      "epoch": 0.2679033035930993,
      "grad_norm": 0.73046875,
      "learning_rate": 9.716991699169917e-05,
      "loss": 3.1991,
      "step": 514200
    },
    {
      "epoch": 0.26792194682925746,
      "grad_norm": 0.75390625,
      "learning_rate": 9.716791679167917e-05,
      "loss": 3.2363,
      "step": 514210
    },
    {
      "epoch": 0.2679405900654156,
      "grad_norm": 0.78515625,
      "learning_rate": 9.716591659165916e-05,
      "loss": 3.2002,
      "step": 514220
    },
    {
      "epoch": 0.2679592333015738,
      "grad_norm": 0.72265625,
      "learning_rate": 9.716391639163916e-05,
      "loss": 3.2541,
      "step": 514230
    },
    {
      "epoch": 0.267977876537732,
      "grad_norm": 0.6640625,
      "learning_rate": 9.716191619161917e-05,
      "loss": 3.2015,
      "step": 514240
    },
    {
      "epoch": 0.26799651977389016,
      "grad_norm": 0.6796875,
      "learning_rate": 9.715991599159916e-05,
      "loss": 3.2165,
      "step": 514250
    },
    {
      "epoch": 0.26801516301004835,
      "grad_norm": 0.6796875,
      "learning_rate": 9.715791579157916e-05,
      "loss": 3.2385,
      "step": 514260
    },
    {
      "epoch": 0.26803380624620654,
      "grad_norm": 0.7578125,
      "learning_rate": 9.715591559155917e-05,
      "loss": 3.2765,
      "step": 514270
    },
    {
      "epoch": 0.26805244948236473,
      "grad_norm": 0.67578125,
      "learning_rate": 9.715391539153916e-05,
      "loss": 3.2818,
      "step": 514280
    },
    {
      "epoch": 0.2680710927185229,
      "grad_norm": 0.7109375,
      "learning_rate": 9.715191519151916e-05,
      "loss": 3.2434,
      "step": 514290
    },
    {
      "epoch": 0.2680897359546811,
      "grad_norm": 0.65234375,
      "learning_rate": 9.714991499149915e-05,
      "loss": 3.2331,
      "step": 514300
    },
    {
      "epoch": 0.2681083791908393,
      "grad_norm": 0.7734375,
      "learning_rate": 9.714791479147915e-05,
      "loss": 3.2211,
      "step": 514310
    },
    {
      "epoch": 0.26812702242699743,
      "grad_norm": 0.72265625,
      "learning_rate": 9.714591459145916e-05,
      "loss": 3.2565,
      "step": 514320
    },
    {
      "epoch": 0.2681456656631556,
      "grad_norm": 0.765625,
      "learning_rate": 9.714391439143915e-05,
      "loss": 3.1553,
      "step": 514330
    },
    {
      "epoch": 0.2681643088993138,
      "grad_norm": 0.73828125,
      "learning_rate": 9.714191419141915e-05,
      "loss": 3.1925,
      "step": 514340
    },
    {
      "epoch": 0.268182952135472,
      "grad_norm": 0.76171875,
      "learning_rate": 9.713991399139915e-05,
      "loss": 3.1566,
      "step": 514350
    },
    {
      "epoch": 0.2682015953716302,
      "grad_norm": 0.65625,
      "learning_rate": 9.713791379137914e-05,
      "loss": 3.2577,
      "step": 514360
    },
    {
      "epoch": 0.2682202386077884,
      "grad_norm": 0.78515625,
      "learning_rate": 9.713591359135914e-05,
      "loss": 3.1787,
      "step": 514370
    },
    {
      "epoch": 0.26823888184394656,
      "grad_norm": 0.7109375,
      "learning_rate": 9.713391339133913e-05,
      "loss": 3.4046,
      "step": 514380
    },
    {
      "epoch": 0.26825752508010475,
      "grad_norm": 0.671875,
      "learning_rate": 9.713191319131913e-05,
      "loss": 3.2024,
      "step": 514390
    },
    {
      "epoch": 0.26827616831626294,
      "grad_norm": 0.76171875,
      "learning_rate": 9.712991299129914e-05,
      "loss": 3.1719,
      "step": 514400
    },
    {
      "epoch": 0.26829481155242113,
      "grad_norm": 0.66796875,
      "learning_rate": 9.712791279127913e-05,
      "loss": 3.1937,
      "step": 514410
    },
    {
      "epoch": 0.2683134547885793,
      "grad_norm": 0.75,
      "learning_rate": 9.712591259125913e-05,
      "loss": 3.1427,
      "step": 514420
    },
    {
      "epoch": 0.26833209802473745,
      "grad_norm": 0.75390625,
      "learning_rate": 9.712391239123912e-05,
      "loss": 3.1654,
      "step": 514430
    },
    {
      "epoch": 0.26835074126089564,
      "grad_norm": 0.84375,
      "learning_rate": 9.712191219121912e-05,
      "loss": 3.1799,
      "step": 514440
    },
    {
      "epoch": 0.26836938449705383,
      "grad_norm": 0.66015625,
      "learning_rate": 9.711991199119912e-05,
      "loss": 3.1309,
      "step": 514450
    },
    {
      "epoch": 0.268388027733212,
      "grad_norm": 0.70703125,
      "learning_rate": 9.711791179117912e-05,
      "loss": 3.181,
      "step": 514460
    },
    {
      "epoch": 0.2684066709693702,
      "grad_norm": 0.7109375,
      "learning_rate": 9.711591159115912e-05,
      "loss": 3.1747,
      "step": 514470
    },
    {
      "epoch": 0.2684253142055284,
      "grad_norm": 0.7421875,
      "learning_rate": 9.711391139113913e-05,
      "loss": 3.2302,
      "step": 514480
    },
    {
      "epoch": 0.2684439574416866,
      "grad_norm": 0.703125,
      "learning_rate": 9.711191119111912e-05,
      "loss": 3.3223,
      "step": 514490
    },
    {
      "epoch": 0.2684626006778448,
      "grad_norm": 0.65234375,
      "learning_rate": 9.710991099109912e-05,
      "loss": 3.263,
      "step": 514500
    },
    {
      "epoch": 0.26848124391400296,
      "grad_norm": 0.68359375,
      "learning_rate": 9.710791079107912e-05,
      "loss": 3.3197,
      "step": 514510
    },
    {
      "epoch": 0.26849988715016115,
      "grad_norm": 0.64453125,
      "learning_rate": 9.710591059105911e-05,
      "loss": 3.2829,
      "step": 514520
    },
    {
      "epoch": 0.2685185303863193,
      "grad_norm": 0.6875,
      "learning_rate": 9.710391039103911e-05,
      "loss": 3.2664,
      "step": 514530
    },
    {
      "epoch": 0.2685371736224775,
      "grad_norm": 0.69140625,
      "learning_rate": 9.71019101910191e-05,
      "loss": 3.2359,
      "step": 514540
    },
    {
      "epoch": 0.26855581685863567,
      "grad_norm": 0.69921875,
      "learning_rate": 9.70999099909991e-05,
      "loss": 3.1288,
      "step": 514550
    },
    {
      "epoch": 0.26857446009479385,
      "grad_norm": 0.83203125,
      "learning_rate": 9.709790979097911e-05,
      "loss": 3.1363,
      "step": 514560
    },
    {
      "epoch": 0.26859310333095204,
      "grad_norm": 0.70703125,
      "learning_rate": 9.70959095909591e-05,
      "loss": 3.1182,
      "step": 514570
    },
    {
      "epoch": 0.26861174656711023,
      "grad_norm": 0.83984375,
      "learning_rate": 9.70939093909391e-05,
      "loss": 3.2576,
      "step": 514580
    },
    {
      "epoch": 0.2686303898032684,
      "grad_norm": 0.74609375,
      "learning_rate": 9.70919091909191e-05,
      "loss": 3.1992,
      "step": 514590
    },
    {
      "epoch": 0.2686490330394266,
      "grad_norm": 0.65234375,
      "learning_rate": 9.708990899089909e-05,
      "loss": 3.2348,
      "step": 514600
    },
    {
      "epoch": 0.2686676762755848,
      "grad_norm": 0.69921875,
      "learning_rate": 9.708790879087909e-05,
      "loss": 3.3026,
      "step": 514610
    },
    {
      "epoch": 0.268686319511743,
      "grad_norm": 0.67578125,
      "learning_rate": 9.708590859085908e-05,
      "loss": 3.2504,
      "step": 514620
    },
    {
      "epoch": 0.2687049627479012,
      "grad_norm": 0.71875,
      "learning_rate": 9.708390839083909e-05,
      "loss": 3.2167,
      "step": 514630
    },
    {
      "epoch": 0.2687236059840593,
      "grad_norm": 0.71875,
      "learning_rate": 9.708190819081909e-05,
      "loss": 3.133,
      "step": 514640
    },
    {
      "epoch": 0.2687422492202175,
      "grad_norm": 0.66015625,
      "learning_rate": 9.707990799079908e-05,
      "loss": 3.2425,
      "step": 514650
    },
    {
      "epoch": 0.2687608924563757,
      "grad_norm": 0.69921875,
      "learning_rate": 9.707790779077909e-05,
      "loss": 3.1239,
      "step": 514660
    },
    {
      "epoch": 0.2687795356925339,
      "grad_norm": 0.69921875,
      "learning_rate": 9.707590759075909e-05,
      "loss": 3.2387,
      "step": 514670
    },
    {
      "epoch": 0.26879817892869207,
      "grad_norm": 0.6875,
      "learning_rate": 9.707390739073908e-05,
      "loss": 3.2296,
      "step": 514680
    },
    {
      "epoch": 0.26881682216485026,
      "grad_norm": 0.6953125,
      "learning_rate": 9.707190719071908e-05,
      "loss": 3.2521,
      "step": 514690
    },
    {
      "epoch": 0.26883546540100844,
      "grad_norm": 0.83984375,
      "learning_rate": 9.706990699069907e-05,
      "loss": 3.132,
      "step": 514700
    },
    {
      "epoch": 0.26885410863716663,
      "grad_norm": 0.72265625,
      "learning_rate": 9.706790679067907e-05,
      "loss": 3.2418,
      "step": 514710
    },
    {
      "epoch": 0.2688727518733248,
      "grad_norm": 0.73046875,
      "learning_rate": 9.706590659065908e-05,
      "loss": 3.2393,
      "step": 514720
    },
    {
      "epoch": 0.268891395109483,
      "grad_norm": 0.77734375,
      "learning_rate": 9.706390639063907e-05,
      "loss": 3.1544,
      "step": 514730
    },
    {
      "epoch": 0.26891003834564114,
      "grad_norm": 0.80859375,
      "learning_rate": 9.706190619061907e-05,
      "loss": 3.2189,
      "step": 514740
    },
    {
      "epoch": 0.26892868158179933,
      "grad_norm": 0.6328125,
      "learning_rate": 9.705990599059906e-05,
      "loss": 3.1639,
      "step": 514750
    },
    {
      "epoch": 0.2689473248179575,
      "grad_norm": 0.8359375,
      "learning_rate": 9.705790579057906e-05,
      "loss": 3.1777,
      "step": 514760
    },
    {
      "epoch": 0.2689659680541157,
      "grad_norm": 0.7578125,
      "learning_rate": 9.705590559055906e-05,
      "loss": 3.1476,
      "step": 514770
    },
    {
      "epoch": 0.2689846112902739,
      "grad_norm": 0.7421875,
      "learning_rate": 9.705390539053905e-05,
      "loss": 3.2116,
      "step": 514780
    },
    {
      "epoch": 0.2690032545264321,
      "grad_norm": 0.796875,
      "learning_rate": 9.705190519051906e-05,
      "loss": 3.2578,
      "step": 514790
    },
    {
      "epoch": 0.2690218977625903,
      "grad_norm": 0.79296875,
      "learning_rate": 9.704990499049906e-05,
      "loss": 3.3234,
      "step": 514800
    },
    {
      "epoch": 0.26904054099874847,
      "grad_norm": 0.6953125,
      "learning_rate": 9.704790479047905e-05,
      "loss": 3.1591,
      "step": 514810
    },
    {
      "epoch": 0.26905918423490666,
      "grad_norm": 0.70703125,
      "learning_rate": 9.704590459045905e-05,
      "loss": 3.1653,
      "step": 514820
    },
    {
      "epoch": 0.26907782747106485,
      "grad_norm": 0.79296875,
      "learning_rate": 9.704390439043904e-05,
      "loss": 3.2013,
      "step": 514830
    },
    {
      "epoch": 0.26909647070722303,
      "grad_norm": 0.7265625,
      "learning_rate": 9.704190419041904e-05,
      "loss": 3.1872,
      "step": 514840
    },
    {
      "epoch": 0.26911511394338117,
      "grad_norm": 0.6953125,
      "learning_rate": 9.703990399039903e-05,
      "loss": 3.181,
      "step": 514850
    },
    {
      "epoch": 0.26913375717953936,
      "grad_norm": 0.85546875,
      "learning_rate": 9.703790379037904e-05,
      "loss": 3.2428,
      "step": 514860
    },
    {
      "epoch": 0.26915240041569755,
      "grad_norm": 0.796875,
      "learning_rate": 9.703590359035904e-05,
      "loss": 3.2402,
      "step": 514870
    },
    {
      "epoch": 0.26917104365185573,
      "grad_norm": 0.70703125,
      "learning_rate": 9.703390339033905e-05,
      "loss": 3.2657,
      "step": 514880
    },
    {
      "epoch": 0.2691896868880139,
      "grad_norm": 0.79296875,
      "learning_rate": 9.703190319031904e-05,
      "loss": 3.2526,
      "step": 514890
    },
    {
      "epoch": 0.2692083301241721,
      "grad_norm": 0.73828125,
      "learning_rate": 9.702990299029904e-05,
      "loss": 3.1897,
      "step": 514900
    },
    {
      "epoch": 0.2692269733603303,
      "grad_norm": 0.73828125,
      "learning_rate": 9.702790279027903e-05,
      "loss": 3.2491,
      "step": 514910
    },
    {
      "epoch": 0.2692456165964885,
      "grad_norm": 0.76953125,
      "learning_rate": 9.702590259025903e-05,
      "loss": 3.2171,
      "step": 514920
    },
    {
      "epoch": 0.2692642598326467,
      "grad_norm": 0.8671875,
      "learning_rate": 9.702390239023903e-05,
      "loss": 3.281,
      "step": 514930
    },
    {
      "epoch": 0.26928290306880487,
      "grad_norm": 0.70703125,
      "learning_rate": 9.702190219021902e-05,
      "loss": 3.2835,
      "step": 514940
    },
    {
      "epoch": 0.269301546304963,
      "grad_norm": 0.7734375,
      "learning_rate": 9.701990199019903e-05,
      "loss": 3.2705,
      "step": 514950
    },
    {
      "epoch": 0.2693201895411212,
      "grad_norm": 0.72265625,
      "learning_rate": 9.701790179017903e-05,
      "loss": 3.1789,
      "step": 514960
    },
    {
      "epoch": 0.2693388327772794,
      "grad_norm": 0.6640625,
      "learning_rate": 9.701590159015902e-05,
      "loss": 3.1378,
      "step": 514970
    },
    {
      "epoch": 0.26935747601343757,
      "grad_norm": 0.88671875,
      "learning_rate": 9.701390139013902e-05,
      "loss": 3.1453,
      "step": 514980
    },
    {
      "epoch": 0.26937611924959576,
      "grad_norm": 0.69921875,
      "learning_rate": 9.701190119011901e-05,
      "loss": 3.194,
      "step": 514990
    },
    {
      "epoch": 0.26939476248575395,
      "grad_norm": 0.890625,
      "learning_rate": 9.700990099009901e-05,
      "loss": 3.2645,
      "step": 515000
    },
    {
      "epoch": 0.26941340572191214,
      "grad_norm": 0.77734375,
      "learning_rate": 9.7007900790079e-05,
      "loss": 3.2731,
      "step": 515010
    },
    {
      "epoch": 0.2694320489580703,
      "grad_norm": 0.75,
      "learning_rate": 9.700590059005901e-05,
      "loss": 3.2335,
      "step": 515020
    },
    {
      "epoch": 0.2694506921942285,
      "grad_norm": 0.7421875,
      "learning_rate": 9.700390039003901e-05,
      "loss": 3.3185,
      "step": 515030
    },
    {
      "epoch": 0.2694693354303867,
      "grad_norm": 0.67578125,
      "learning_rate": 9.7001900190019e-05,
      "loss": 3.1378,
      "step": 515040
    },
    {
      "epoch": 0.2694879786665449,
      "grad_norm": 0.75,
      "learning_rate": 9.6999899989999e-05,
      "loss": 3.2874,
      "step": 515050
    },
    {
      "epoch": 0.269506621902703,
      "grad_norm": 0.75390625,
      "learning_rate": 9.699789978997901e-05,
      "loss": 3.1868,
      "step": 515060
    },
    {
      "epoch": 0.2695252651388612,
      "grad_norm": 0.82421875,
      "learning_rate": 9.6995899589959e-05,
      "loss": 3.2021,
      "step": 515070
    },
    {
      "epoch": 0.2695439083750194,
      "grad_norm": 0.6875,
      "learning_rate": 9.6993899389939e-05,
      "loss": 3.175,
      "step": 515080
    },
    {
      "epoch": 0.2695625516111776,
      "grad_norm": 0.67578125,
      "learning_rate": 9.6991899189919e-05,
      "loss": 3.2149,
      "step": 515090
    },
    {
      "epoch": 0.2695811948473358,
      "grad_norm": 0.66015625,
      "learning_rate": 9.698989898989899e-05,
      "loss": 3.1965,
      "step": 515100
    },
    {
      "epoch": 0.26959983808349397,
      "grad_norm": 0.83984375,
      "learning_rate": 9.6987898789879e-05,
      "loss": 3.303,
      "step": 515110
    },
    {
      "epoch": 0.26961848131965216,
      "grad_norm": 0.7890625,
      "learning_rate": 9.6985898589859e-05,
      "loss": 3.1721,
      "step": 515120
    },
    {
      "epoch": 0.26963712455581035,
      "grad_norm": 0.68359375,
      "learning_rate": 9.698389838983899e-05,
      "loss": 3.3057,
      "step": 515130
    },
    {
      "epoch": 0.26965576779196854,
      "grad_norm": 0.70703125,
      "learning_rate": 9.698189818981899e-05,
      "loss": 3.2035,
      "step": 515140
    },
    {
      "epoch": 0.2696744110281267,
      "grad_norm": 0.75,
      "learning_rate": 9.697989798979898e-05,
      "loss": 3.2859,
      "step": 515150
    },
    {
      "epoch": 0.26969305426428486,
      "grad_norm": 0.7890625,
      "learning_rate": 9.697789778977898e-05,
      "loss": 3.2402,
      "step": 515160
    },
    {
      "epoch": 0.26971169750044305,
      "grad_norm": 0.8046875,
      "learning_rate": 9.697589758975897e-05,
      "loss": 3.2313,
      "step": 515170
    },
    {
      "epoch": 0.26973034073660124,
      "grad_norm": 0.90234375,
      "learning_rate": 9.697389738973898e-05,
      "loss": 3.0617,
      "step": 515180
    },
    {
      "epoch": 0.2697489839727594,
      "grad_norm": 0.70703125,
      "learning_rate": 9.697189718971898e-05,
      "loss": 3.3029,
      "step": 515190
    },
    {
      "epoch": 0.2697676272089176,
      "grad_norm": 0.66015625,
      "learning_rate": 9.696989698969897e-05,
      "loss": 3.2237,
      "step": 515200
    },
    {
      "epoch": 0.2697862704450758,
      "grad_norm": 0.7109375,
      "learning_rate": 9.696789678967897e-05,
      "loss": 3.2535,
      "step": 515210
    },
    {
      "epoch": 0.269804913681234,
      "grad_norm": 0.74609375,
      "learning_rate": 9.696589658965897e-05,
      "loss": 3.094,
      "step": 515220
    },
    {
      "epoch": 0.2698235569173922,
      "grad_norm": 0.6328125,
      "learning_rate": 9.696389638963896e-05,
      "loss": 3.144,
      "step": 515230
    },
    {
      "epoch": 0.26984220015355037,
      "grad_norm": 0.7109375,
      "learning_rate": 9.696189618961896e-05,
      "loss": 3.2302,
      "step": 515240
    },
    {
      "epoch": 0.26986084338970856,
      "grad_norm": 0.75,
      "learning_rate": 9.695989598959897e-05,
      "loss": 3.188,
      "step": 515250
    },
    {
      "epoch": 0.2698794866258667,
      "grad_norm": 0.72265625,
      "learning_rate": 9.695789578957896e-05,
      "loss": 3.0419,
      "step": 515260
    },
    {
      "epoch": 0.2698981298620249,
      "grad_norm": 0.734375,
      "learning_rate": 9.695589558955897e-05,
      "loss": 3.188,
      "step": 515270
    },
    {
      "epoch": 0.26991677309818307,
      "grad_norm": 0.62890625,
      "learning_rate": 9.695389538953897e-05,
      "loss": 3.1952,
      "step": 515280
    },
    {
      "epoch": 0.26993541633434126,
      "grad_norm": 0.75,
      "learning_rate": 9.695189518951896e-05,
      "loss": 3.1865,
      "step": 515290
    },
    {
      "epoch": 0.26995405957049945,
      "grad_norm": 0.84375,
      "learning_rate": 9.694989498949896e-05,
      "loss": 3.1404,
      "step": 515300
    },
    {
      "epoch": 0.26997270280665764,
      "grad_norm": 0.703125,
      "learning_rate": 9.694789478947895e-05,
      "loss": 3.0704,
      "step": 515310
    },
    {
      "epoch": 0.2699913460428158,
      "grad_norm": 0.8125,
      "learning_rate": 9.694589458945895e-05,
      "loss": 3.1531,
      "step": 515320
    },
    {
      "epoch": 0.270009989278974,
      "grad_norm": 0.64453125,
      "learning_rate": 9.694389438943894e-05,
      "loss": 3.2734,
      "step": 515330
    },
    {
      "epoch": 0.2700286325151322,
      "grad_norm": 0.94140625,
      "learning_rate": 9.694189418941895e-05,
      "loss": 3.1397,
      "step": 515340
    },
    {
      "epoch": 0.2700472757512904,
      "grad_norm": 0.8359375,
      "learning_rate": 9.693989398939895e-05,
      "loss": 3.2694,
      "step": 515350
    },
    {
      "epoch": 0.2700659189874486,
      "grad_norm": 0.69921875,
      "learning_rate": 9.693789378937894e-05,
      "loss": 3.0683,
      "step": 515360
    },
    {
      "epoch": 0.2700845622236067,
      "grad_norm": 0.84765625,
      "learning_rate": 9.693589358935894e-05,
      "loss": 3.2443,
      "step": 515370
    },
    {
      "epoch": 0.2701032054597649,
      "grad_norm": 0.74609375,
      "learning_rate": 9.693389338933894e-05,
      "loss": 3.2001,
      "step": 515380
    },
    {
      "epoch": 0.2701218486959231,
      "grad_norm": 0.6484375,
      "learning_rate": 9.693189318931893e-05,
      "loss": 3.3259,
      "step": 515390
    },
    {
      "epoch": 0.2701404919320813,
      "grad_norm": 0.79296875,
      "learning_rate": 9.692989298929893e-05,
      "loss": 3.0177,
      "step": 515400
    },
    {
      "epoch": 0.2701591351682395,
      "grad_norm": 0.71875,
      "learning_rate": 9.692789278927892e-05,
      "loss": 3.1582,
      "step": 515410
    },
    {
      "epoch": 0.27017777840439766,
      "grad_norm": 0.69921875,
      "learning_rate": 9.692589258925893e-05,
      "loss": 3.1226,
      "step": 515420
    },
    {
      "epoch": 0.27019642164055585,
      "grad_norm": 0.87109375,
      "learning_rate": 9.692389238923893e-05,
      "loss": 3.1919,
      "step": 515430
    },
    {
      "epoch": 0.27021506487671404,
      "grad_norm": 0.84765625,
      "learning_rate": 9.692189218921892e-05,
      "loss": 3.23,
      "step": 515440
    },
    {
      "epoch": 0.27023370811287223,
      "grad_norm": 0.84765625,
      "learning_rate": 9.691989198919893e-05,
      "loss": 3.2113,
      "step": 515450
    },
    {
      "epoch": 0.2702523513490304,
      "grad_norm": 0.73828125,
      "learning_rate": 9.691789178917893e-05,
      "loss": 3.1826,
      "step": 515460
    },
    {
      "epoch": 0.27027099458518855,
      "grad_norm": 0.78125,
      "learning_rate": 9.691589158915892e-05,
      "loss": 3.1458,
      "step": 515470
    },
    {
      "epoch": 0.27028963782134674,
      "grad_norm": 0.703125,
      "learning_rate": 9.691389138913892e-05,
      "loss": 3.1578,
      "step": 515480
    },
    {
      "epoch": 0.27030828105750493,
      "grad_norm": 0.74609375,
      "learning_rate": 9.691189118911891e-05,
      "loss": 3.2595,
      "step": 515490
    },
    {
      "epoch": 0.2703269242936631,
      "grad_norm": 0.74609375,
      "learning_rate": 9.690989098909892e-05,
      "loss": 3.164,
      "step": 515500
    },
    {
      "epoch": 0.2703455675298213,
      "grad_norm": 0.76171875,
      "learning_rate": 9.690789078907892e-05,
      "loss": 3.1205,
      "step": 515510
    },
    {
      "epoch": 0.2703642107659795,
      "grad_norm": 0.703125,
      "learning_rate": 9.690589058905891e-05,
      "loss": 3.2613,
      "step": 515520
    },
    {
      "epoch": 0.2703828540021377,
      "grad_norm": 0.80859375,
      "learning_rate": 9.690389038903891e-05,
      "loss": 3.2979,
      "step": 515530
    },
    {
      "epoch": 0.2704014972382959,
      "grad_norm": 0.83203125,
      "learning_rate": 9.69018901890189e-05,
      "loss": 3.0828,
      "step": 515540
    },
    {
      "epoch": 0.27042014047445406,
      "grad_norm": 0.671875,
      "learning_rate": 9.68998899889989e-05,
      "loss": 3.1845,
      "step": 515550
    },
    {
      "epoch": 0.27043878371061225,
      "grad_norm": 0.70703125,
      "learning_rate": 9.68978897889789e-05,
      "loss": 3.2611,
      "step": 515560
    },
    {
      "epoch": 0.27045742694677044,
      "grad_norm": 0.78515625,
      "learning_rate": 9.689588958895889e-05,
      "loss": 3.2173,
      "step": 515570
    },
    {
      "epoch": 0.2704760701829286,
      "grad_norm": 0.81640625,
      "learning_rate": 9.68938893889389e-05,
      "loss": 3.178,
      "step": 515580
    },
    {
      "epoch": 0.27049471341908676,
      "grad_norm": 0.75390625,
      "learning_rate": 9.68918891889189e-05,
      "loss": 3.2126,
      "step": 515590
    },
    {
      "epoch": 0.27051335665524495,
      "grad_norm": 0.7578125,
      "learning_rate": 9.68898889888989e-05,
      "loss": 3.2028,
      "step": 515600
    },
    {
      "epoch": 0.27053199989140314,
      "grad_norm": 0.765625,
      "learning_rate": 9.688788878887889e-05,
      "loss": 3.182,
      "step": 515610
    },
    {
      "epoch": 0.27055064312756133,
      "grad_norm": 0.78515625,
      "learning_rate": 9.688588858885888e-05,
      "loss": 3.2211,
      "step": 515620
    },
    {
      "epoch": 0.2705692863637195,
      "grad_norm": 0.7890625,
      "learning_rate": 9.688388838883888e-05,
      "loss": 3.0977,
      "step": 515630
    },
    {
      "epoch": 0.2705879295998777,
      "grad_norm": 0.6953125,
      "learning_rate": 9.688188818881888e-05,
      "loss": 3.1774,
      "step": 515640
    },
    {
      "epoch": 0.2706065728360359,
      "grad_norm": 0.7265625,
      "learning_rate": 9.687988798879888e-05,
      "loss": 3.1859,
      "step": 515650
    },
    {
      "epoch": 0.2706252160721941,
      "grad_norm": 0.6171875,
      "learning_rate": 9.687788778877888e-05,
      "loss": 3.2426,
      "step": 515660
    },
    {
      "epoch": 0.2706438593083523,
      "grad_norm": 0.80078125,
      "learning_rate": 9.687588758875889e-05,
      "loss": 3.2186,
      "step": 515670
    },
    {
      "epoch": 0.2706625025445104,
      "grad_norm": 0.72265625,
      "learning_rate": 9.687388738873889e-05,
      "loss": 3.2245,
      "step": 515680
    },
    {
      "epoch": 0.2706811457806686,
      "grad_norm": 0.76953125,
      "learning_rate": 9.687188718871888e-05,
      "loss": 3.2705,
      "step": 515690
    },
    {
      "epoch": 0.2706997890168268,
      "grad_norm": 0.7109375,
      "learning_rate": 9.686988698869888e-05,
      "loss": 3.1082,
      "step": 515700
    },
    {
      "epoch": 0.270718432252985,
      "grad_norm": 0.734375,
      "learning_rate": 9.686788678867887e-05,
      "loss": 3.2511,
      "step": 515710
    },
    {
      "epoch": 0.27073707548914316,
      "grad_norm": 0.7578125,
      "learning_rate": 9.686588658865887e-05,
      "loss": 3.2002,
      "step": 515720
    },
    {
      "epoch": 0.27075571872530135,
      "grad_norm": 0.7890625,
      "learning_rate": 9.686388638863886e-05,
      "loss": 3.1637,
      "step": 515730
    },
    {
      "epoch": 0.27077436196145954,
      "grad_norm": 0.73828125,
      "learning_rate": 9.686188618861887e-05,
      "loss": 3.1781,
      "step": 515740
    },
    {
      "epoch": 0.27079300519761773,
      "grad_norm": 0.72265625,
      "learning_rate": 9.685988598859887e-05,
      "loss": 3.2922,
      "step": 515750
    },
    {
      "epoch": 0.2708116484337759,
      "grad_norm": 0.73046875,
      "learning_rate": 9.685788578857886e-05,
      "loss": 3.254,
      "step": 515760
    },
    {
      "epoch": 0.2708302916699341,
      "grad_norm": 0.74609375,
      "learning_rate": 9.685588558855886e-05,
      "loss": 3.2578,
      "step": 515770
    },
    {
      "epoch": 0.2708489349060923,
      "grad_norm": 0.76171875,
      "learning_rate": 9.685388538853885e-05,
      "loss": 3.1316,
      "step": 515780
    },
    {
      "epoch": 0.27086757814225043,
      "grad_norm": 0.75390625,
      "learning_rate": 9.685188518851885e-05,
      "loss": 3.1457,
      "step": 515790
    },
    {
      "epoch": 0.2708862213784086,
      "grad_norm": 0.75,
      "learning_rate": 9.684988498849885e-05,
      "loss": 3.1349,
      "step": 515800
    },
    {
      "epoch": 0.2709048646145668,
      "grad_norm": 0.73046875,
      "learning_rate": 9.684788478847885e-05,
      "loss": 3.2425,
      "step": 515810
    },
    {
      "epoch": 0.270923507850725,
      "grad_norm": 0.6953125,
      "learning_rate": 9.684588458845885e-05,
      "loss": 3.1638,
      "step": 515820
    },
    {
      "epoch": 0.2709421510868832,
      "grad_norm": 0.74609375,
      "learning_rate": 9.684388438843885e-05,
      "loss": 3.2197,
      "step": 515830
    },
    {
      "epoch": 0.2709607943230414,
      "grad_norm": 0.7734375,
      "learning_rate": 9.684188418841884e-05,
      "loss": 3.1497,
      "step": 515840
    },
    {
      "epoch": 0.27097943755919957,
      "grad_norm": 0.734375,
      "learning_rate": 9.683988398839885e-05,
      "loss": 3.1931,
      "step": 515850
    },
    {
      "epoch": 0.27099808079535775,
      "grad_norm": 0.91015625,
      "learning_rate": 9.683788378837885e-05,
      "loss": 3.2529,
      "step": 515860
    },
    {
      "epoch": 0.27101672403151594,
      "grad_norm": 0.671875,
      "learning_rate": 9.683588358835884e-05,
      "loss": 3.142,
      "step": 515870
    },
    {
      "epoch": 0.27103536726767413,
      "grad_norm": 0.73828125,
      "learning_rate": 9.683388338833884e-05,
      "loss": 3.3261,
      "step": 515880
    },
    {
      "epoch": 0.27105401050383227,
      "grad_norm": 0.72265625,
      "learning_rate": 9.683188318831883e-05,
      "loss": 3.1785,
      "step": 515890
    },
    {
      "epoch": 0.27107265373999045,
      "grad_norm": 0.70703125,
      "learning_rate": 9.682988298829884e-05,
      "loss": 3.3058,
      "step": 515900
    },
    {
      "epoch": 0.27109129697614864,
      "grad_norm": 0.76953125,
      "learning_rate": 9.682788278827884e-05,
      "loss": 3.111,
      "step": 515910
    },
    {
      "epoch": 0.27110994021230683,
      "grad_norm": 0.68359375,
      "learning_rate": 9.682588258825883e-05,
      "loss": 3.0903,
      "step": 515920
    },
    {
      "epoch": 0.271128583448465,
      "grad_norm": 0.73046875,
      "learning_rate": 9.682388238823883e-05,
      "loss": 3.2663,
      "step": 515930
    },
    {
      "epoch": 0.2711472266846232,
      "grad_norm": 0.75390625,
      "learning_rate": 9.682188218821882e-05,
      "loss": 3.2808,
      "step": 515940
    },
    {
      "epoch": 0.2711658699207814,
      "grad_norm": 0.75390625,
      "learning_rate": 9.681988198819882e-05,
      "loss": 3.0486,
      "step": 515950
    },
    {
      "epoch": 0.2711845131569396,
      "grad_norm": 0.7421875,
      "learning_rate": 9.681788178817882e-05,
      "loss": 3.1408,
      "step": 515960
    },
    {
      "epoch": 0.2712031563930978,
      "grad_norm": 0.8359375,
      "learning_rate": 9.681588158815883e-05,
      "loss": 3.2283,
      "step": 515970
    },
    {
      "epoch": 0.27122179962925597,
      "grad_norm": 0.77734375,
      "learning_rate": 9.681388138813882e-05,
      "loss": 3.2028,
      "step": 515980
    },
    {
      "epoch": 0.27124044286541416,
      "grad_norm": 0.7890625,
      "learning_rate": 9.681188118811882e-05,
      "loss": 3.3499,
      "step": 515990
    },
    {
      "epoch": 0.2712590861015723,
      "grad_norm": 0.66796875,
      "learning_rate": 9.680988098809881e-05,
      "loss": 3.1876,
      "step": 516000
    },
    {
      "epoch": 0.2712777293377305,
      "grad_norm": 0.83984375,
      "learning_rate": 9.680788078807881e-05,
      "loss": 3.2201,
      "step": 516010
    },
    {
      "epoch": 0.27129637257388867,
      "grad_norm": 0.734375,
      "learning_rate": 9.68058805880588e-05,
      "loss": 3.1758,
      "step": 516020
    },
    {
      "epoch": 0.27131501581004686,
      "grad_norm": 0.8125,
      "learning_rate": 9.68038803880388e-05,
      "loss": 3.2739,
      "step": 516030
    },
    {
      "epoch": 0.27133365904620504,
      "grad_norm": 0.72265625,
      "learning_rate": 9.680188018801881e-05,
      "loss": 3.1326,
      "step": 516040
    },
    {
      "epoch": 0.27135230228236323,
      "grad_norm": 0.734375,
      "learning_rate": 9.67998799879988e-05,
      "loss": 3.1756,
      "step": 516050
    },
    {
      "epoch": 0.2713709455185214,
      "grad_norm": 0.8046875,
      "learning_rate": 9.679787978797881e-05,
      "loss": 3.258,
      "step": 516060
    },
    {
      "epoch": 0.2713895887546796,
      "grad_norm": 0.75390625,
      "learning_rate": 9.679587958795881e-05,
      "loss": 3.1785,
      "step": 516070
    },
    {
      "epoch": 0.2714082319908378,
      "grad_norm": 0.79296875,
      "learning_rate": 9.67938793879388e-05,
      "loss": 3.2047,
      "step": 516080
    },
    {
      "epoch": 0.271426875226996,
      "grad_norm": 0.7578125,
      "learning_rate": 9.67918791879188e-05,
      "loss": 3.2595,
      "step": 516090
    },
    {
      "epoch": 0.2714455184631541,
      "grad_norm": 0.78125,
      "learning_rate": 9.67898789878988e-05,
      "loss": 3.1932,
      "step": 516100
    },
    {
      "epoch": 0.2714641616993123,
      "grad_norm": 0.71484375,
      "learning_rate": 9.678787878787879e-05,
      "loss": 3.2324,
      "step": 516110
    },
    {
      "epoch": 0.2714828049354705,
      "grad_norm": 0.76953125,
      "learning_rate": 9.678587858785879e-05,
      "loss": 3.2259,
      "step": 516120
    },
    {
      "epoch": 0.2715014481716287,
      "grad_norm": 0.765625,
      "learning_rate": 9.67838783878388e-05,
      "loss": 3.1064,
      "step": 516130
    },
    {
      "epoch": 0.2715200914077869,
      "grad_norm": 0.78515625,
      "learning_rate": 9.678187818781879e-05,
      "loss": 3.2708,
      "step": 516140
    },
    {
      "epoch": 0.27153873464394507,
      "grad_norm": 0.93359375,
      "learning_rate": 9.677987798779879e-05,
      "loss": 3.2365,
      "step": 516150
    },
    {
      "epoch": 0.27155737788010326,
      "grad_norm": 0.70703125,
      "learning_rate": 9.677787778777878e-05,
      "loss": 3.2063,
      "step": 516160
    },
    {
      "epoch": 0.27157602111626145,
      "grad_norm": 0.74609375,
      "learning_rate": 9.677587758775878e-05,
      "loss": 3.309,
      "step": 516170
    },
    {
      "epoch": 0.27159466435241963,
      "grad_norm": 0.73828125,
      "learning_rate": 9.677387738773877e-05,
      "loss": 3.2331,
      "step": 516180
    },
    {
      "epoch": 0.2716133075885778,
      "grad_norm": 0.8359375,
      "learning_rate": 9.677187718771877e-05,
      "loss": 3.3345,
      "step": 516190
    },
    {
      "epoch": 0.271631950824736,
      "grad_norm": 0.70703125,
      "learning_rate": 9.676987698769876e-05,
      "loss": 3.3451,
      "step": 516200
    },
    {
      "epoch": 0.27165059406089415,
      "grad_norm": 0.796875,
      "learning_rate": 9.676787678767877e-05,
      "loss": 3.1338,
      "step": 516210
    },
    {
      "epoch": 0.27166923729705234,
      "grad_norm": 0.7578125,
      "learning_rate": 9.676587658765877e-05,
      "loss": 3.2281,
      "step": 516220
    },
    {
      "epoch": 0.2716878805332105,
      "grad_norm": 0.8125,
      "learning_rate": 9.676387638763876e-05,
      "loss": 3.2477,
      "step": 516230
    },
    {
      "epoch": 0.2717065237693687,
      "grad_norm": 0.69921875,
      "learning_rate": 9.676187618761877e-05,
      "loss": 3.1206,
      "step": 516240
    },
    {
      "epoch": 0.2717251670055269,
      "grad_norm": 0.75390625,
      "learning_rate": 9.675987598759877e-05,
      "loss": 3.1765,
      "step": 516250
    },
    {
      "epoch": 0.2717438102416851,
      "grad_norm": 0.75,
      "learning_rate": 9.675787578757877e-05,
      "loss": 3.1597,
      "step": 516260
    },
    {
      "epoch": 0.2717624534778433,
      "grad_norm": 0.71875,
      "learning_rate": 9.675587558755876e-05,
      "loss": 3.1609,
      "step": 516270
    },
    {
      "epoch": 0.27178109671400147,
      "grad_norm": 0.734375,
      "learning_rate": 9.675387538753876e-05,
      "loss": 3.1181,
      "step": 516280
    },
    {
      "epoch": 0.27179973995015966,
      "grad_norm": 0.79296875,
      "learning_rate": 9.675187518751877e-05,
      "loss": 3.2663,
      "step": 516290
    },
    {
      "epoch": 0.27181838318631785,
      "grad_norm": 0.77734375,
      "learning_rate": 9.674987498749876e-05,
      "loss": 3.2655,
      "step": 516300
    },
    {
      "epoch": 0.271837026422476,
      "grad_norm": 0.79296875,
      "learning_rate": 9.674787478747876e-05,
      "loss": 3.161,
      "step": 516310
    },
    {
      "epoch": 0.27185566965863417,
      "grad_norm": 0.69921875,
      "learning_rate": 9.674587458745875e-05,
      "loss": 3.1623,
      "step": 516320
    },
    {
      "epoch": 0.27187431289479236,
      "grad_norm": 0.734375,
      "learning_rate": 9.674387438743875e-05,
      "loss": 3.1606,
      "step": 516330
    },
    {
      "epoch": 0.27189295613095055,
      "grad_norm": 0.76953125,
      "learning_rate": 9.674187418741874e-05,
      "loss": 3.1494,
      "step": 516340
    },
    {
      "epoch": 0.27191159936710874,
      "grad_norm": 0.78125,
      "learning_rate": 9.673987398739874e-05,
      "loss": 3.1967,
      "step": 516350
    },
    {
      "epoch": 0.2719302426032669,
      "grad_norm": 0.734375,
      "learning_rate": 9.673787378737873e-05,
      "loss": 3.2692,
      "step": 516360
    },
    {
      "epoch": 0.2719488858394251,
      "grad_norm": 0.71875,
      "learning_rate": 9.673587358735874e-05,
      "loss": 3.1168,
      "step": 516370
    },
    {
      "epoch": 0.2719675290755833,
      "grad_norm": 0.7421875,
      "learning_rate": 9.673387338733874e-05,
      "loss": 3.2843,
      "step": 516380
    },
    {
      "epoch": 0.2719861723117415,
      "grad_norm": 0.6875,
      "learning_rate": 9.673187318731873e-05,
      "loss": 3.1991,
      "step": 516390
    },
    {
      "epoch": 0.2720048155478997,
      "grad_norm": 0.69140625,
      "learning_rate": 9.672987298729873e-05,
      "loss": 3.1314,
      "step": 516400
    },
    {
      "epoch": 0.27202345878405787,
      "grad_norm": 0.84765625,
      "learning_rate": 9.672787278727873e-05,
      "loss": 3.3154,
      "step": 516410
    },
    {
      "epoch": 0.272042102020216,
      "grad_norm": 0.66015625,
      "learning_rate": 9.672587258725872e-05,
      "loss": 3.1521,
      "step": 516420
    },
    {
      "epoch": 0.2720607452563742,
      "grad_norm": 0.80078125,
      "learning_rate": 9.672387238723872e-05,
      "loss": 3.1229,
      "step": 516430
    },
    {
      "epoch": 0.2720793884925324,
      "grad_norm": 0.66015625,
      "learning_rate": 9.672187218721873e-05,
      "loss": 3.1973,
      "step": 516440
    },
    {
      "epoch": 0.27209803172869057,
      "grad_norm": 0.75,
      "learning_rate": 9.671987198719874e-05,
      "loss": 3.2172,
      "step": 516450
    },
    {
      "epoch": 0.27211667496484876,
      "grad_norm": 0.78125,
      "learning_rate": 9.671787178717873e-05,
      "loss": 3.1904,
      "step": 516460
    },
    {
      "epoch": 0.27213531820100695,
      "grad_norm": 0.8046875,
      "learning_rate": 9.671587158715873e-05,
      "loss": 3.2347,
      "step": 516470
    },
    {
      "epoch": 0.27215396143716514,
      "grad_norm": 0.77734375,
      "learning_rate": 9.671387138713872e-05,
      "loss": 3.058,
      "step": 516480
    },
    {
      "epoch": 0.2721726046733233,
      "grad_norm": 0.81640625,
      "learning_rate": 9.671187118711872e-05,
      "loss": 3.229,
      "step": 516490
    },
    {
      "epoch": 0.2721912479094815,
      "grad_norm": 0.75,
      "learning_rate": 9.670987098709871e-05,
      "loss": 3.1802,
      "step": 516500
    },
    {
      "epoch": 0.2722098911456397,
      "grad_norm": 0.69921875,
      "learning_rate": 9.670787078707871e-05,
      "loss": 3.1678,
      "step": 516510
    },
    {
      "epoch": 0.27222853438179784,
      "grad_norm": 0.62109375,
      "learning_rate": 9.67058705870587e-05,
      "loss": 3.2404,
      "step": 516520
    },
    {
      "epoch": 0.272247177617956,
      "grad_norm": 0.81640625,
      "learning_rate": 9.670387038703871e-05,
      "loss": 3.3134,
      "step": 516530
    },
    {
      "epoch": 0.2722658208541142,
      "grad_norm": 0.65234375,
      "learning_rate": 9.670187018701871e-05,
      "loss": 3.2515,
      "step": 516540
    },
    {
      "epoch": 0.2722844640902724,
      "grad_norm": 0.88671875,
      "learning_rate": 9.66998699869987e-05,
      "loss": 3.1676,
      "step": 516550
    },
    {
      "epoch": 0.2723031073264306,
      "grad_norm": 0.72265625,
      "learning_rate": 9.66978697869787e-05,
      "loss": 3.1843,
      "step": 516560
    },
    {
      "epoch": 0.2723217505625888,
      "grad_norm": 0.80078125,
      "learning_rate": 9.66958695869587e-05,
      "loss": 3.2021,
      "step": 516570
    },
    {
      "epoch": 0.27234039379874697,
      "grad_norm": 0.84765625,
      "learning_rate": 9.669386938693869e-05,
      "loss": 3.2301,
      "step": 516580
    },
    {
      "epoch": 0.27235903703490516,
      "grad_norm": 0.8671875,
      "learning_rate": 9.669186918691869e-05,
      "loss": 3.2097,
      "step": 516590
    },
    {
      "epoch": 0.27237768027106335,
      "grad_norm": 0.83203125,
      "learning_rate": 9.66898689868987e-05,
      "loss": 3.205,
      "step": 516600
    },
    {
      "epoch": 0.27239632350722154,
      "grad_norm": 0.71875,
      "learning_rate": 9.668786878687869e-05,
      "loss": 3.3907,
      "step": 516610
    },
    {
      "epoch": 0.27241496674337967,
      "grad_norm": 0.65625,
      "learning_rate": 9.668586858685869e-05,
      "loss": 3.1971,
      "step": 516620
    },
    {
      "epoch": 0.27243360997953786,
      "grad_norm": 0.71484375,
      "learning_rate": 9.668386838683868e-05,
      "loss": 3.2254,
      "step": 516630
    },
    {
      "epoch": 0.27245225321569605,
      "grad_norm": 0.78515625,
      "learning_rate": 9.668186818681869e-05,
      "loss": 3.2838,
      "step": 516640
    },
    {
      "epoch": 0.27247089645185424,
      "grad_norm": 0.703125,
      "learning_rate": 9.667986798679869e-05,
      "loss": 3.0518,
      "step": 516650
    },
    {
      "epoch": 0.2724895396880124,
      "grad_norm": 0.77734375,
      "learning_rate": 9.667786778677868e-05,
      "loss": 3.2119,
      "step": 516660
    },
    {
      "epoch": 0.2725081829241706,
      "grad_norm": 0.84765625,
      "learning_rate": 9.667586758675868e-05,
      "loss": 3.3369,
      "step": 516670
    },
    {
      "epoch": 0.2725268261603288,
      "grad_norm": 0.66015625,
      "learning_rate": 9.667386738673867e-05,
      "loss": 3.3161,
      "step": 516680
    },
    {
      "epoch": 0.272545469396487,
      "grad_norm": 0.8671875,
      "learning_rate": 9.667186718671868e-05,
      "loss": 3.2752,
      "step": 516690
    },
    {
      "epoch": 0.2725641126326452,
      "grad_norm": 0.7578125,
      "learning_rate": 9.666986698669868e-05,
      "loss": 3.1934,
      "step": 516700
    },
    {
      "epoch": 0.2725827558688034,
      "grad_norm": 0.72265625,
      "learning_rate": 9.666786678667868e-05,
      "loss": 3.1979,
      "step": 516710
    },
    {
      "epoch": 0.27260139910496156,
      "grad_norm": 0.703125,
      "learning_rate": 9.666586658665867e-05,
      "loss": 3.1662,
      "step": 516720
    },
    {
      "epoch": 0.2726200423411197,
      "grad_norm": 0.67578125,
      "learning_rate": 9.666386638663867e-05,
      "loss": 3.1978,
      "step": 516730
    },
    {
      "epoch": 0.2726386855772779,
      "grad_norm": 0.86328125,
      "learning_rate": 9.666186618661866e-05,
      "loss": 3.182,
      "step": 516740
    },
    {
      "epoch": 0.2726573288134361,
      "grad_norm": 0.65625,
      "learning_rate": 9.665986598659866e-05,
      "loss": 3.157,
      "step": 516750
    },
    {
      "epoch": 0.27267597204959426,
      "grad_norm": 0.7578125,
      "learning_rate": 9.665786578657867e-05,
      "loss": 3.2543,
      "step": 516760
    },
    {
      "epoch": 0.27269461528575245,
      "grad_norm": 0.72265625,
      "learning_rate": 9.665586558655866e-05,
      "loss": 3.1224,
      "step": 516770
    },
    {
      "epoch": 0.27271325852191064,
      "grad_norm": 0.76171875,
      "learning_rate": 9.665386538653866e-05,
      "loss": 3.2519,
      "step": 516780
    },
    {
      "epoch": 0.27273190175806883,
      "grad_norm": 0.828125,
      "learning_rate": 9.665186518651865e-05,
      "loss": 3.1316,
      "step": 516790
    },
    {
      "epoch": 0.272750544994227,
      "grad_norm": 0.81640625,
      "learning_rate": 9.664986498649865e-05,
      "loss": 3.0658,
      "step": 516800
    },
    {
      "epoch": 0.2727691882303852,
      "grad_norm": 0.78515625,
      "learning_rate": 9.664786478647864e-05,
      "loss": 3.1799,
      "step": 516810
    },
    {
      "epoch": 0.2727878314665434,
      "grad_norm": 0.74609375,
      "learning_rate": 9.664586458645864e-05,
      "loss": 3.2486,
      "step": 516820
    },
    {
      "epoch": 0.27280647470270153,
      "grad_norm": 0.7421875,
      "learning_rate": 9.664386438643865e-05,
      "loss": 3.2191,
      "step": 516830
    },
    {
      "epoch": 0.2728251179388597,
      "grad_norm": 0.68359375,
      "learning_rate": 9.664186418641865e-05,
      "loss": 3.2345,
      "step": 516840
    },
    {
      "epoch": 0.2728437611750179,
      "grad_norm": 0.8046875,
      "learning_rate": 9.663986398639865e-05,
      "loss": 3.2828,
      "step": 516850
    },
    {
      "epoch": 0.2728624044111761,
      "grad_norm": 0.921875,
      "learning_rate": 9.663786378637865e-05,
      "loss": 3.2606,
      "step": 516860
    },
    {
      "epoch": 0.2728810476473343,
      "grad_norm": 0.73046875,
      "learning_rate": 9.663586358635865e-05,
      "loss": 3.2175,
      "step": 516870
    },
    {
      "epoch": 0.2728996908834925,
      "grad_norm": 0.78125,
      "learning_rate": 9.663386338633864e-05,
      "loss": 3.2266,
      "step": 516880
    },
    {
      "epoch": 0.27291833411965066,
      "grad_norm": 0.703125,
      "learning_rate": 9.663186318631864e-05,
      "loss": 3.3017,
      "step": 516890
    },
    {
      "epoch": 0.27293697735580885,
      "grad_norm": 0.75,
      "learning_rate": 9.662986298629863e-05,
      "loss": 3.1738,
      "step": 516900
    },
    {
      "epoch": 0.27295562059196704,
      "grad_norm": 0.75390625,
      "learning_rate": 9.662786278627863e-05,
      "loss": 3.2832,
      "step": 516910
    },
    {
      "epoch": 0.27297426382812523,
      "grad_norm": 0.80078125,
      "learning_rate": 9.662586258625864e-05,
      "loss": 3.1652,
      "step": 516920
    },
    {
      "epoch": 0.2729929070642834,
      "grad_norm": 0.70703125,
      "learning_rate": 9.662386238623863e-05,
      "loss": 3.1731,
      "step": 516930
    },
    {
      "epoch": 0.27301155030044155,
      "grad_norm": 0.859375,
      "learning_rate": 9.662186218621863e-05,
      "loss": 3.1921,
      "step": 516940
    },
    {
      "epoch": 0.27303019353659974,
      "grad_norm": 0.71484375,
      "learning_rate": 9.661986198619862e-05,
      "loss": 3.2316,
      "step": 516950
    },
    {
      "epoch": 0.27304883677275793,
      "grad_norm": 0.65234375,
      "learning_rate": 9.661786178617862e-05,
      "loss": 3.215,
      "step": 516960
    },
    {
      "epoch": 0.2730674800089161,
      "grad_norm": 0.6953125,
      "learning_rate": 9.661586158615861e-05,
      "loss": 3.2134,
      "step": 516970
    },
    {
      "epoch": 0.2730861232450743,
      "grad_norm": 0.73828125,
      "learning_rate": 9.661386138613861e-05,
      "loss": 3.2751,
      "step": 516980
    },
    {
      "epoch": 0.2731047664812325,
      "grad_norm": 0.62109375,
      "learning_rate": 9.661186118611862e-05,
      "loss": 3.3622,
      "step": 516990
    },
    {
      "epoch": 0.2731234097173907,
      "grad_norm": 0.6796875,
      "learning_rate": 9.660986098609862e-05,
      "loss": 3.1702,
      "step": 517000
    },
    {
      "epoch": 0.2731420529535489,
      "grad_norm": 0.86328125,
      "learning_rate": 9.660786078607861e-05,
      "loss": 3.3043,
      "step": 517010
    },
    {
      "epoch": 0.27316069618970706,
      "grad_norm": 0.7421875,
      "learning_rate": 9.66058605860586e-05,
      "loss": 3.2563,
      "step": 517020
    },
    {
      "epoch": 0.27317933942586525,
      "grad_norm": 0.75,
      "learning_rate": 9.660386038603862e-05,
      "loss": 3.1249,
      "step": 517030
    },
    {
      "epoch": 0.2731979826620234,
      "grad_norm": 0.78125,
      "learning_rate": 9.660186018601861e-05,
      "loss": 3.1967,
      "step": 517040
    },
    {
      "epoch": 0.2732166258981816,
      "grad_norm": 0.828125,
      "learning_rate": 9.659985998599861e-05,
      "loss": 3.1291,
      "step": 517050
    },
    {
      "epoch": 0.27323526913433976,
      "grad_norm": 0.796875,
      "learning_rate": 9.65978597859786e-05,
      "loss": 3.2882,
      "step": 517060
    },
    {
      "epoch": 0.27325391237049795,
      "grad_norm": 0.75,
      "learning_rate": 9.65958595859586e-05,
      "loss": 3.1257,
      "step": 517070
    },
    {
      "epoch": 0.27327255560665614,
      "grad_norm": 0.64453125,
      "learning_rate": 9.659385938593861e-05,
      "loss": 3.2387,
      "step": 517080
    },
    {
      "epoch": 0.27329119884281433,
      "grad_norm": 0.828125,
      "learning_rate": 9.65918591859186e-05,
      "loss": 3.241,
      "step": 517090
    },
    {
      "epoch": 0.2733098420789725,
      "grad_norm": 0.60546875,
      "learning_rate": 9.65898589858986e-05,
      "loss": 3.1541,
      "step": 517100
    },
    {
      "epoch": 0.2733284853151307,
      "grad_norm": 0.73046875,
      "learning_rate": 9.65878587858786e-05,
      "loss": 3.2842,
      "step": 517110
    },
    {
      "epoch": 0.2733471285512889,
      "grad_norm": 0.85546875,
      "learning_rate": 9.658585858585859e-05,
      "loss": 3.1808,
      "step": 517120
    },
    {
      "epoch": 0.2733657717874471,
      "grad_norm": 0.63671875,
      "learning_rate": 9.658385838583859e-05,
      "loss": 3.2635,
      "step": 517130
    },
    {
      "epoch": 0.2733844150236053,
      "grad_norm": 0.65625,
      "learning_rate": 9.658185818581858e-05,
      "loss": 3.1803,
      "step": 517140
    },
    {
      "epoch": 0.2734030582597634,
      "grad_norm": 0.72265625,
      "learning_rate": 9.657985798579858e-05,
      "loss": 3.1685,
      "step": 517150
    },
    {
      "epoch": 0.2734217014959216,
      "grad_norm": 0.75,
      "learning_rate": 9.657785778577859e-05,
      "loss": 3.1908,
      "step": 517160
    },
    {
      "epoch": 0.2734403447320798,
      "grad_norm": 0.8125,
      "learning_rate": 9.657585758575858e-05,
      "loss": 3.3255,
      "step": 517170
    },
    {
      "epoch": 0.273458987968238,
      "grad_norm": 0.7421875,
      "learning_rate": 9.657385738573858e-05,
      "loss": 3.2851,
      "step": 517180
    },
    {
      "epoch": 0.27347763120439617,
      "grad_norm": 0.6875,
      "learning_rate": 9.657185718571857e-05,
      "loss": 3.2137,
      "step": 517190
    },
    {
      "epoch": 0.27349627444055435,
      "grad_norm": 0.79296875,
      "learning_rate": 9.656985698569857e-05,
      "loss": 3.3363,
      "step": 517200
    },
    {
      "epoch": 0.27351491767671254,
      "grad_norm": 0.82421875,
      "learning_rate": 9.656785678567856e-05,
      "loss": 3.1188,
      "step": 517210
    },
    {
      "epoch": 0.27353356091287073,
      "grad_norm": 0.7890625,
      "learning_rate": 9.656585658565856e-05,
      "loss": 3.2002,
      "step": 517220
    },
    {
      "epoch": 0.2735522041490289,
      "grad_norm": 0.73828125,
      "learning_rate": 9.656385638563857e-05,
      "loss": 3.1913,
      "step": 517230
    },
    {
      "epoch": 0.2735708473851871,
      "grad_norm": 0.69921875,
      "learning_rate": 9.656185618561858e-05,
      "loss": 3.3009,
      "step": 517240
    },
    {
      "epoch": 0.27358949062134524,
      "grad_norm": 0.470703125,
      "learning_rate": 9.655985598559857e-05,
      "loss": 3.1136,
      "step": 517250
    },
    {
      "epoch": 0.27360813385750343,
      "grad_norm": 0.765625,
      "learning_rate": 9.655785578557857e-05,
      "loss": 3.1254,
      "step": 517260
    },
    {
      "epoch": 0.2736267770936616,
      "grad_norm": 0.68359375,
      "learning_rate": 9.655585558555856e-05,
      "loss": 3.1178,
      "step": 517270
    },
    {
      "epoch": 0.2736454203298198,
      "grad_norm": 0.78125,
      "learning_rate": 9.655385538553856e-05,
      "loss": 3.1801,
      "step": 517280
    },
    {
      "epoch": 0.273664063565978,
      "grad_norm": 0.8125,
      "learning_rate": 9.655185518551856e-05,
      "loss": 3.1599,
      "step": 517290
    },
    {
      "epoch": 0.2736827068021362,
      "grad_norm": 0.75,
      "learning_rate": 9.654985498549855e-05,
      "loss": 3.1076,
      "step": 517300
    },
    {
      "epoch": 0.2737013500382944,
      "grad_norm": 0.671875,
      "learning_rate": 9.654785478547855e-05,
      "loss": 3.3953,
      "step": 517310
    },
    {
      "epoch": 0.27371999327445257,
      "grad_norm": 0.75,
      "learning_rate": 9.654585458545856e-05,
      "loss": 3.3364,
      "step": 517320
    },
    {
      "epoch": 0.27373863651061076,
      "grad_norm": 0.75390625,
      "learning_rate": 9.654385438543855e-05,
      "loss": 3.2127,
      "step": 517330
    },
    {
      "epoch": 0.27375727974676894,
      "grad_norm": 0.67578125,
      "learning_rate": 9.654185418541855e-05,
      "loss": 3.2984,
      "step": 517340
    },
    {
      "epoch": 0.27377592298292713,
      "grad_norm": 0.68359375,
      "learning_rate": 9.653985398539854e-05,
      "loss": 3.2674,
      "step": 517350
    },
    {
      "epoch": 0.27379456621908527,
      "grad_norm": 0.6796875,
      "learning_rate": 9.653785378537854e-05,
      "loss": 3.1634,
      "step": 517360
    },
    {
      "epoch": 0.27381320945524346,
      "grad_norm": 0.79296875,
      "learning_rate": 9.653585358535853e-05,
      "loss": 3.2494,
      "step": 517370
    },
    {
      "epoch": 0.27383185269140164,
      "grad_norm": 0.77734375,
      "learning_rate": 9.653385338533853e-05,
      "loss": 3.139,
      "step": 517380
    },
    {
      "epoch": 0.27385049592755983,
      "grad_norm": 0.6796875,
      "learning_rate": 9.653185318531854e-05,
      "loss": 3.1746,
      "step": 517390
    },
    {
      "epoch": 0.273869139163718,
      "grad_norm": 0.765625,
      "learning_rate": 9.652985298529853e-05,
      "loss": 3.2613,
      "step": 517400
    },
    {
      "epoch": 0.2738877823998762,
      "grad_norm": 0.6953125,
      "learning_rate": 9.652785278527853e-05,
      "loss": 3.1764,
      "step": 517410
    },
    {
      "epoch": 0.2739064256360344,
      "grad_norm": 0.70703125,
      "learning_rate": 9.652585258525854e-05,
      "loss": 3.2758,
      "step": 517420
    },
    {
      "epoch": 0.2739250688721926,
      "grad_norm": 0.71484375,
      "learning_rate": 9.652385238523853e-05,
      "loss": 3.3451,
      "step": 517430
    },
    {
      "epoch": 0.2739437121083508,
      "grad_norm": 0.84765625,
      "learning_rate": 9.652185218521853e-05,
      "loss": 3.2104,
      "step": 517440
    },
    {
      "epoch": 0.27396235534450897,
      "grad_norm": 0.75,
      "learning_rate": 9.651985198519853e-05,
      "loss": 3.2849,
      "step": 517450
    },
    {
      "epoch": 0.2739809985806671,
      "grad_norm": 0.62109375,
      "learning_rate": 9.651785178517852e-05,
      "loss": 3.349,
      "step": 517460
    },
    {
      "epoch": 0.2739996418168253,
      "grad_norm": 0.7578125,
      "learning_rate": 9.651585158515852e-05,
      "loss": 3.2801,
      "step": 517470
    },
    {
      "epoch": 0.2740182850529835,
      "grad_norm": 0.734375,
      "learning_rate": 9.651385138513853e-05,
      "loss": 3.2425,
      "step": 517480
    },
    {
      "epoch": 0.27403692828914167,
      "grad_norm": 0.63671875,
      "learning_rate": 9.651185118511852e-05,
      "loss": 3.2368,
      "step": 517490
    },
    {
      "epoch": 0.27405557152529986,
      "grad_norm": 0.58984375,
      "learning_rate": 9.650985098509852e-05,
      "loss": 3.0813,
      "step": 517500
    },
    {
      "epoch": 0.27407421476145805,
      "grad_norm": 0.8515625,
      "learning_rate": 9.650785078507851e-05,
      "loss": 3.2295,
      "step": 517510
    },
    {
      "epoch": 0.27409285799761623,
      "grad_norm": 0.75,
      "learning_rate": 9.650585058505851e-05,
      "loss": 3.2312,
      "step": 517520
    },
    {
      "epoch": 0.2741115012337744,
      "grad_norm": 0.78125,
      "learning_rate": 9.65038503850385e-05,
      "loss": 3.1374,
      "step": 517530
    },
    {
      "epoch": 0.2741301444699326,
      "grad_norm": 0.71484375,
      "learning_rate": 9.65018501850185e-05,
      "loss": 3.1619,
      "step": 517540
    },
    {
      "epoch": 0.2741487877060908,
      "grad_norm": 0.80078125,
      "learning_rate": 9.649984998499851e-05,
      "loss": 3.1732,
      "step": 517550
    },
    {
      "epoch": 0.274167430942249,
      "grad_norm": 0.66796875,
      "learning_rate": 9.64978497849785e-05,
      "loss": 3.2867,
      "step": 517560
    },
    {
      "epoch": 0.2741860741784071,
      "grad_norm": 0.765625,
      "learning_rate": 9.64958495849585e-05,
      "loss": 3.1736,
      "step": 517570
    },
    {
      "epoch": 0.2742047174145653,
      "grad_norm": 0.72265625,
      "learning_rate": 9.64938493849385e-05,
      "loss": 3.2766,
      "step": 517580
    },
    {
      "epoch": 0.2742233606507235,
      "grad_norm": 0.703125,
      "learning_rate": 9.649184918491849e-05,
      "loss": 3.2601,
      "step": 517590
    },
    {
      "epoch": 0.2742420038868817,
      "grad_norm": 0.7734375,
      "learning_rate": 9.648984898489849e-05,
      "loss": 3.1712,
      "step": 517600
    },
    {
      "epoch": 0.2742606471230399,
      "grad_norm": 0.6640625,
      "learning_rate": 9.648784878487848e-05,
      "loss": 3.3237,
      "step": 517610
    },
    {
      "epoch": 0.27427929035919807,
      "grad_norm": 0.76171875,
      "learning_rate": 9.648584858485849e-05,
      "loss": 3.114,
      "step": 517620
    },
    {
      "epoch": 0.27429793359535626,
      "grad_norm": 0.70703125,
      "learning_rate": 9.648384838483849e-05,
      "loss": 3.1908,
      "step": 517630
    },
    {
      "epoch": 0.27431657683151445,
      "grad_norm": 0.8046875,
      "learning_rate": 9.64818481848185e-05,
      "loss": 3.2146,
      "step": 517640
    },
    {
      "epoch": 0.27433522006767264,
      "grad_norm": 0.8125,
      "learning_rate": 9.647984798479849e-05,
      "loss": 3.1737,
      "step": 517650
    },
    {
      "epoch": 0.2743538633038308,
      "grad_norm": 0.7109375,
      "learning_rate": 9.647784778477849e-05,
      "loss": 3.3268,
      "step": 517660
    },
    {
      "epoch": 0.27437250653998896,
      "grad_norm": 0.6953125,
      "learning_rate": 9.647584758475848e-05,
      "loss": 3.2556,
      "step": 517670
    },
    {
      "epoch": 0.27439114977614715,
      "grad_norm": 0.8125,
      "learning_rate": 9.647384738473848e-05,
      "loss": 3.1157,
      "step": 517680
    },
    {
      "epoch": 0.27440979301230534,
      "grad_norm": 0.69921875,
      "learning_rate": 9.647184718471847e-05,
      "loss": 3.2721,
      "step": 517690
    },
    {
      "epoch": 0.2744284362484635,
      "grad_norm": 0.75,
      "learning_rate": 9.646984698469847e-05,
      "loss": 3.1358,
      "step": 517700
    },
    {
      "epoch": 0.2744470794846217,
      "grad_norm": 0.83203125,
      "learning_rate": 9.646784678467848e-05,
      "loss": 3.2919,
      "step": 517710
    },
    {
      "epoch": 0.2744657227207799,
      "grad_norm": 0.6328125,
      "learning_rate": 9.646584658465847e-05,
      "loss": 3.2351,
      "step": 517720
    },
    {
      "epoch": 0.2744843659569381,
      "grad_norm": 0.7421875,
      "learning_rate": 9.646384638463847e-05,
      "loss": 3.1144,
      "step": 517730
    },
    {
      "epoch": 0.2745030091930963,
      "grad_norm": 0.68359375,
      "learning_rate": 9.646184618461847e-05,
      "loss": 3.1746,
      "step": 517740
    },
    {
      "epoch": 0.27452165242925447,
      "grad_norm": 0.6796875,
      "learning_rate": 9.645984598459846e-05,
      "loss": 3.2459,
      "step": 517750
    },
    {
      "epoch": 0.27454029566541266,
      "grad_norm": 0.7421875,
      "learning_rate": 9.645784578457846e-05,
      "loss": 3.1855,
      "step": 517760
    },
    {
      "epoch": 0.27455893890157085,
      "grad_norm": 0.8984375,
      "learning_rate": 9.645584558455845e-05,
      "loss": 3.2472,
      "step": 517770
    },
    {
      "epoch": 0.274577582137729,
      "grad_norm": 0.66796875,
      "learning_rate": 9.645384538453846e-05,
      "loss": 3.2017,
      "step": 517780
    },
    {
      "epoch": 0.27459622537388717,
      "grad_norm": 0.6171875,
      "learning_rate": 9.645184518451846e-05,
      "loss": 3.2104,
      "step": 517790
    },
    {
      "epoch": 0.27461486861004536,
      "grad_norm": 0.671875,
      "learning_rate": 9.644984498449845e-05,
      "loss": 3.2742,
      "step": 517800
    },
    {
      "epoch": 0.27463351184620355,
      "grad_norm": 0.671875,
      "learning_rate": 9.644784478447845e-05,
      "loss": 3.0832,
      "step": 517810
    },
    {
      "epoch": 0.27465215508236174,
      "grad_norm": 0.671875,
      "learning_rate": 9.644584458445846e-05,
      "loss": 3.171,
      "step": 517820
    },
    {
      "epoch": 0.2746707983185199,
      "grad_norm": 0.80078125,
      "learning_rate": 9.644384438443845e-05,
      "loss": 3.2612,
      "step": 517830
    },
    {
      "epoch": 0.2746894415546781,
      "grad_norm": 0.62890625,
      "learning_rate": 9.644184418441845e-05,
      "loss": 3.1767,
      "step": 517840
    },
    {
      "epoch": 0.2747080847908363,
      "grad_norm": 0.81640625,
      "learning_rate": 9.643984398439844e-05,
      "loss": 3.2216,
      "step": 517850
    },
    {
      "epoch": 0.2747267280269945,
      "grad_norm": 0.6953125,
      "learning_rate": 9.643784378437844e-05,
      "loss": 3.2392,
      "step": 517860
    },
    {
      "epoch": 0.2747453712631527,
      "grad_norm": 0.69921875,
      "learning_rate": 9.643584358435845e-05,
      "loss": 3.1897,
      "step": 517870
    },
    {
      "epoch": 0.2747640144993108,
      "grad_norm": 0.8125,
      "learning_rate": 9.643384338433844e-05,
      "loss": 3.2221,
      "step": 517880
    },
    {
      "epoch": 0.274782657735469,
      "grad_norm": 0.6875,
      "learning_rate": 9.643184318431844e-05,
      "loss": 3.2987,
      "step": 517890
    },
    {
      "epoch": 0.2748013009716272,
      "grad_norm": 0.6875,
      "learning_rate": 9.642984298429844e-05,
      "loss": 3.273,
      "step": 517900
    },
    {
      "epoch": 0.2748199442077854,
      "grad_norm": 0.734375,
      "learning_rate": 9.642784278427843e-05,
      "loss": 3.2063,
      "step": 517910
    },
    {
      "epoch": 0.27483858744394357,
      "grad_norm": 0.87890625,
      "learning_rate": 9.642584258425843e-05,
      "loss": 3.1871,
      "step": 517920
    },
    {
      "epoch": 0.27485723068010176,
      "grad_norm": 0.76953125,
      "learning_rate": 9.642384238423842e-05,
      "loss": 3.2682,
      "step": 517930
    },
    {
      "epoch": 0.27487587391625995,
      "grad_norm": 0.8125,
      "learning_rate": 9.642184218421843e-05,
      "loss": 3.2276,
      "step": 517940
    },
    {
      "epoch": 0.27489451715241814,
      "grad_norm": 0.68359375,
      "learning_rate": 9.641984198419843e-05,
      "loss": 3.2777,
      "step": 517950
    },
    {
      "epoch": 0.2749131603885763,
      "grad_norm": 0.78515625,
      "learning_rate": 9.641784178417842e-05,
      "loss": 3.306,
      "step": 517960
    },
    {
      "epoch": 0.2749318036247345,
      "grad_norm": 0.71875,
      "learning_rate": 9.641584158415842e-05,
      "loss": 3.134,
      "step": 517970
    },
    {
      "epoch": 0.2749504468608927,
      "grad_norm": 0.67578125,
      "learning_rate": 9.641384138413841e-05,
      "loss": 3.2737,
      "step": 517980
    },
    {
      "epoch": 0.27496909009705084,
      "grad_norm": 0.8125,
      "learning_rate": 9.641184118411841e-05,
      "loss": 3.3026,
      "step": 517990
    },
    {
      "epoch": 0.27498773333320903,
      "grad_norm": 0.7265625,
      "learning_rate": 9.64098409840984e-05,
      "loss": 3.1689,
      "step": 518000
    },
    {
      "epoch": 0.2750063765693672,
      "grad_norm": 0.80859375,
      "learning_rate": 9.64078407840784e-05,
      "loss": 3.1997,
      "step": 518010
    },
    {
      "epoch": 0.2750250198055254,
      "grad_norm": 0.7109375,
      "learning_rate": 9.640584058405841e-05,
      "loss": 3.2003,
      "step": 518020
    },
    {
      "epoch": 0.2750436630416836,
      "grad_norm": 0.69140625,
      "learning_rate": 9.640384038403842e-05,
      "loss": 3.1593,
      "step": 518030
    },
    {
      "epoch": 0.2750623062778418,
      "grad_norm": 0.76171875,
      "learning_rate": 9.640184018401841e-05,
      "loss": 3.2039,
      "step": 518040
    },
    {
      "epoch": 0.275080949514,
      "grad_norm": 0.7421875,
      "learning_rate": 9.639983998399841e-05,
      "loss": 3.1929,
      "step": 518050
    },
    {
      "epoch": 0.27509959275015816,
      "grad_norm": 0.80859375,
      "learning_rate": 9.63978397839784e-05,
      "loss": 3.2946,
      "step": 518060
    },
    {
      "epoch": 0.27511823598631635,
      "grad_norm": 0.75,
      "learning_rate": 9.63958395839584e-05,
      "loss": 3.1876,
      "step": 518070
    },
    {
      "epoch": 0.27513687922247454,
      "grad_norm": 0.76953125,
      "learning_rate": 9.63938393839384e-05,
      "loss": 3.2502,
      "step": 518080
    },
    {
      "epoch": 0.2751555224586327,
      "grad_norm": 0.671875,
      "learning_rate": 9.639183918391839e-05,
      "loss": 3.0914,
      "step": 518090
    },
    {
      "epoch": 0.27517416569479086,
      "grad_norm": 0.65234375,
      "learning_rate": 9.63898389838984e-05,
      "loss": 3.2467,
      "step": 518100
    },
    {
      "epoch": 0.27519280893094905,
      "grad_norm": 0.60546875,
      "learning_rate": 9.63878387838784e-05,
      "loss": 3.3664,
      "step": 518110
    },
    {
      "epoch": 0.27521145216710724,
      "grad_norm": 1.15625,
      "learning_rate": 9.638583858385839e-05,
      "loss": 3.2412,
      "step": 518120
    },
    {
      "epoch": 0.27523009540326543,
      "grad_norm": 0.6953125,
      "learning_rate": 9.638383838383839e-05,
      "loss": 3.1466,
      "step": 518130
    },
    {
      "epoch": 0.2752487386394236,
      "grad_norm": 0.66796875,
      "learning_rate": 9.638183818381838e-05,
      "loss": 3.1474,
      "step": 518140
    },
    {
      "epoch": 0.2752673818755818,
      "grad_norm": 0.73828125,
      "learning_rate": 9.637983798379838e-05,
      "loss": 3.0733,
      "step": 518150
    },
    {
      "epoch": 0.27528602511174,
      "grad_norm": 0.68359375,
      "learning_rate": 9.637783778377838e-05,
      "loss": 3.1709,
      "step": 518160
    },
    {
      "epoch": 0.2753046683478982,
      "grad_norm": 0.7421875,
      "learning_rate": 9.637583758375837e-05,
      "loss": 3.1572,
      "step": 518170
    },
    {
      "epoch": 0.2753233115840564,
      "grad_norm": 0.65234375,
      "learning_rate": 9.637383738373838e-05,
      "loss": 3.327,
      "step": 518180
    },
    {
      "epoch": 0.2753419548202145,
      "grad_norm": 0.64453125,
      "learning_rate": 9.637183718371838e-05,
      "loss": 3.0957,
      "step": 518190
    },
    {
      "epoch": 0.2753605980563727,
      "grad_norm": 0.81640625,
      "learning_rate": 9.636983698369837e-05,
      "loss": 3.2126,
      "step": 518200
    },
    {
      "epoch": 0.2753792412925309,
      "grad_norm": 0.8671875,
      "learning_rate": 9.636783678367838e-05,
      "loss": 3.1267,
      "step": 518210
    },
    {
      "epoch": 0.2753978845286891,
      "grad_norm": 0.78515625,
      "learning_rate": 9.636583658365838e-05,
      "loss": 3.1544,
      "step": 518220
    },
    {
      "epoch": 0.27541652776484726,
      "grad_norm": 0.7578125,
      "learning_rate": 9.636383638363837e-05,
      "loss": 3.245,
      "step": 518230
    },
    {
      "epoch": 0.27543517100100545,
      "grad_norm": 0.7109375,
      "learning_rate": 9.636183618361837e-05,
      "loss": 3.154,
      "step": 518240
    },
    {
      "epoch": 0.27545381423716364,
      "grad_norm": 0.625,
      "learning_rate": 9.635983598359836e-05,
      "loss": 3.2193,
      "step": 518250
    },
    {
      "epoch": 0.27547245747332183,
      "grad_norm": 0.80078125,
      "learning_rate": 9.635783578357837e-05,
      "loss": 3.189,
      "step": 518260
    },
    {
      "epoch": 0.27549110070948,
      "grad_norm": 0.69921875,
      "learning_rate": 9.635583558355837e-05,
      "loss": 3.1923,
      "step": 518270
    },
    {
      "epoch": 0.2755097439456382,
      "grad_norm": 0.80078125,
      "learning_rate": 9.635383538353836e-05,
      "loss": 3.2149,
      "step": 518280
    },
    {
      "epoch": 0.2755283871817964,
      "grad_norm": 0.64453125,
      "learning_rate": 9.635183518351836e-05,
      "loss": 3.1362,
      "step": 518290
    },
    {
      "epoch": 0.27554703041795453,
      "grad_norm": 0.6171875,
      "learning_rate": 9.634983498349835e-05,
      "loss": 3.2219,
      "step": 518300
    },
    {
      "epoch": 0.2755656736541127,
      "grad_norm": 0.8125,
      "learning_rate": 9.634783478347835e-05,
      "loss": 3.1546,
      "step": 518310
    },
    {
      "epoch": 0.2755843168902709,
      "grad_norm": 0.6796875,
      "learning_rate": 9.634583458345835e-05,
      "loss": 3.2342,
      "step": 518320
    },
    {
      "epoch": 0.2756029601264291,
      "grad_norm": 0.70703125,
      "learning_rate": 9.634383438343834e-05,
      "loss": 3.1605,
      "step": 518330
    },
    {
      "epoch": 0.2756216033625873,
      "grad_norm": 0.7578125,
      "learning_rate": 9.634183418341835e-05,
      "loss": 3.0988,
      "step": 518340
    },
    {
      "epoch": 0.2756402465987455,
      "grad_norm": 0.7265625,
      "learning_rate": 9.633983398339835e-05,
      "loss": 3.2681,
      "step": 518350
    },
    {
      "epoch": 0.27565888983490366,
      "grad_norm": 0.6640625,
      "learning_rate": 9.633783378337834e-05,
      "loss": 3.1523,
      "step": 518360
    },
    {
      "epoch": 0.27567753307106185,
      "grad_norm": 0.6875,
      "learning_rate": 9.633583358335834e-05,
      "loss": 3.1921,
      "step": 518370
    },
    {
      "epoch": 0.27569617630722004,
      "grad_norm": 0.671875,
      "learning_rate": 9.633383338333833e-05,
      "loss": 3.1983,
      "step": 518380
    },
    {
      "epoch": 0.27571481954337823,
      "grad_norm": 0.93359375,
      "learning_rate": 9.633183318331833e-05,
      "loss": 3.283,
      "step": 518390
    },
    {
      "epoch": 0.27573346277953636,
      "grad_norm": 0.703125,
      "learning_rate": 9.632983298329832e-05,
      "loss": 3.1919,
      "step": 518400
    },
    {
      "epoch": 0.27575210601569455,
      "grad_norm": 0.71484375,
      "learning_rate": 9.632783278327833e-05,
      "loss": 3.239,
      "step": 518410
    },
    {
      "epoch": 0.27577074925185274,
      "grad_norm": 0.68359375,
      "learning_rate": 9.632583258325834e-05,
      "loss": 3.261,
      "step": 518420
    },
    {
      "epoch": 0.27578939248801093,
      "grad_norm": 0.7578125,
      "learning_rate": 9.632383238323834e-05,
      "loss": 3.1582,
      "step": 518430
    },
    {
      "epoch": 0.2758080357241691,
      "grad_norm": 0.74609375,
      "learning_rate": 9.632183218321833e-05,
      "loss": 3.2293,
      "step": 518440
    },
    {
      "epoch": 0.2758266789603273,
      "grad_norm": 0.75390625,
      "learning_rate": 9.631983198319833e-05,
      "loss": 3.1791,
      "step": 518450
    },
    {
      "epoch": 0.2758453221964855,
      "grad_norm": 0.70703125,
      "learning_rate": 9.631783178317832e-05,
      "loss": 3.2875,
      "step": 518460
    },
    {
      "epoch": 0.2758639654326437,
      "grad_norm": 0.76171875,
      "learning_rate": 9.631583158315832e-05,
      "loss": 3.2767,
      "step": 518470
    },
    {
      "epoch": 0.2758826086688019,
      "grad_norm": 0.6484375,
      "learning_rate": 9.631383138313832e-05,
      "loss": 3.1877,
      "step": 518480
    },
    {
      "epoch": 0.27590125190496007,
      "grad_norm": 0.83203125,
      "learning_rate": 9.631183118311831e-05,
      "loss": 3.1037,
      "step": 518490
    },
    {
      "epoch": 0.27591989514111825,
      "grad_norm": 0.71875,
      "learning_rate": 9.630983098309832e-05,
      "loss": 3.2186,
      "step": 518500
    },
    {
      "epoch": 0.2759385383772764,
      "grad_norm": 0.75,
      "learning_rate": 9.630783078307832e-05,
      "loss": 3.1891,
      "step": 518510
    },
    {
      "epoch": 0.2759571816134346,
      "grad_norm": 0.68359375,
      "learning_rate": 9.630583058305831e-05,
      "loss": 3.2353,
      "step": 518520
    },
    {
      "epoch": 0.27597582484959277,
      "grad_norm": 0.8359375,
      "learning_rate": 9.630383038303831e-05,
      "loss": 3.2646,
      "step": 518530
    },
    {
      "epoch": 0.27599446808575095,
      "grad_norm": 0.828125,
      "learning_rate": 9.63018301830183e-05,
      "loss": 3.1903,
      "step": 518540
    },
    {
      "epoch": 0.27601311132190914,
      "grad_norm": 0.79296875,
      "learning_rate": 9.62998299829983e-05,
      "loss": 3.2536,
      "step": 518550
    },
    {
      "epoch": 0.27603175455806733,
      "grad_norm": 0.78515625,
      "learning_rate": 9.62978297829783e-05,
      "loss": 3.2623,
      "step": 518560
    },
    {
      "epoch": 0.2760503977942255,
      "grad_norm": 0.7578125,
      "learning_rate": 9.62958295829583e-05,
      "loss": 3.1829,
      "step": 518570
    },
    {
      "epoch": 0.2760690410303837,
      "grad_norm": 0.69921875,
      "learning_rate": 9.62938293829383e-05,
      "loss": 3.2169,
      "step": 518580
    },
    {
      "epoch": 0.2760876842665419,
      "grad_norm": 0.76171875,
      "learning_rate": 9.62918291829183e-05,
      "loss": 3.1896,
      "step": 518590
    },
    {
      "epoch": 0.2761063275027001,
      "grad_norm": 0.74609375,
      "learning_rate": 9.628982898289829e-05,
      "loss": 3.2649,
      "step": 518600
    },
    {
      "epoch": 0.2761249707388582,
      "grad_norm": 0.75390625,
      "learning_rate": 9.62878287828783e-05,
      "loss": 3.1981,
      "step": 518610
    },
    {
      "epoch": 0.2761436139750164,
      "grad_norm": 0.6875,
      "learning_rate": 9.62858285828583e-05,
      "loss": 3.2471,
      "step": 518620
    },
    {
      "epoch": 0.2761622572111746,
      "grad_norm": 0.703125,
      "learning_rate": 9.628382838283829e-05,
      "loss": 3.2125,
      "step": 518630
    },
    {
      "epoch": 0.2761809004473328,
      "grad_norm": 0.78515625,
      "learning_rate": 9.628182818281829e-05,
      "loss": 3.2188,
      "step": 518640
    },
    {
      "epoch": 0.276199543683491,
      "grad_norm": 0.80859375,
      "learning_rate": 9.627982798279828e-05,
      "loss": 3.0859,
      "step": 518650
    },
    {
      "epoch": 0.27621818691964917,
      "grad_norm": 0.8203125,
      "learning_rate": 9.627782778277829e-05,
      "loss": 3.2185,
      "step": 518660
    },
    {
      "epoch": 0.27623683015580736,
      "grad_norm": 0.84375,
      "learning_rate": 9.627582758275829e-05,
      "loss": 3.1443,
      "step": 518670
    },
    {
      "epoch": 0.27625547339196554,
      "grad_norm": 0.8359375,
      "learning_rate": 9.627382738273828e-05,
      "loss": 3.2656,
      "step": 518680
    },
    {
      "epoch": 0.27627411662812373,
      "grad_norm": 0.71875,
      "learning_rate": 9.627182718271828e-05,
      "loss": 3.1941,
      "step": 518690
    },
    {
      "epoch": 0.2762927598642819,
      "grad_norm": 0.6953125,
      "learning_rate": 9.626982698269827e-05,
      "loss": 3.1414,
      "step": 518700
    },
    {
      "epoch": 0.2763114031004401,
      "grad_norm": 0.671875,
      "learning_rate": 9.626782678267827e-05,
      "loss": 3.1338,
      "step": 518710
    },
    {
      "epoch": 0.27633004633659825,
      "grad_norm": 0.6875,
      "learning_rate": 9.626582658265826e-05,
      "loss": 3.1999,
      "step": 518720
    },
    {
      "epoch": 0.27634868957275643,
      "grad_norm": 0.75,
      "learning_rate": 9.626382638263827e-05,
      "loss": 3.2529,
      "step": 518730
    },
    {
      "epoch": 0.2763673328089146,
      "grad_norm": 0.76953125,
      "learning_rate": 9.626182618261827e-05,
      "loss": 3.1719,
      "step": 518740
    },
    {
      "epoch": 0.2763859760450728,
      "grad_norm": 0.80078125,
      "learning_rate": 9.625982598259826e-05,
      "loss": 3.2828,
      "step": 518750
    },
    {
      "epoch": 0.276404619281231,
      "grad_norm": 0.69140625,
      "learning_rate": 9.625782578257826e-05,
      "loss": 3.2819,
      "step": 518760
    },
    {
      "epoch": 0.2764232625173892,
      "grad_norm": 0.6875,
      "learning_rate": 9.625582558255826e-05,
      "loss": 3.2426,
      "step": 518770
    },
    {
      "epoch": 0.2764419057535474,
      "grad_norm": 0.703125,
      "learning_rate": 9.625382538253825e-05,
      "loss": 3.1994,
      "step": 518780
    },
    {
      "epoch": 0.27646054898970557,
      "grad_norm": 0.92578125,
      "learning_rate": 9.625182518251825e-05,
      "loss": 3.2896,
      "step": 518790
    },
    {
      "epoch": 0.27647919222586376,
      "grad_norm": 0.7421875,
      "learning_rate": 9.624982498249824e-05,
      "loss": 3.2804,
      "step": 518800
    },
    {
      "epoch": 0.27649783546202195,
      "grad_norm": 0.765625,
      "learning_rate": 9.624782478247825e-05,
      "loss": 3.2551,
      "step": 518810
    },
    {
      "epoch": 0.2765164786981801,
      "grad_norm": 0.671875,
      "learning_rate": 9.624582458245826e-05,
      "loss": 3.1455,
      "step": 518820
    },
    {
      "epoch": 0.27653512193433827,
      "grad_norm": 0.65625,
      "learning_rate": 9.624382438243826e-05,
      "loss": 3.1578,
      "step": 518830
    },
    {
      "epoch": 0.27655376517049646,
      "grad_norm": 0.7421875,
      "learning_rate": 9.624182418241825e-05,
      "loss": 3.2036,
      "step": 518840
    },
    {
      "epoch": 0.27657240840665465,
      "grad_norm": 0.8046875,
      "learning_rate": 9.623982398239825e-05,
      "loss": 3.1248,
      "step": 518850
    },
    {
      "epoch": 0.27659105164281284,
      "grad_norm": 0.73828125,
      "learning_rate": 9.623782378237824e-05,
      "loss": 3.1861,
      "step": 518860
    },
    {
      "epoch": 0.276609694878971,
      "grad_norm": 0.69140625,
      "learning_rate": 9.623582358235824e-05,
      "loss": 3.2377,
      "step": 518870
    },
    {
      "epoch": 0.2766283381151292,
      "grad_norm": 0.734375,
      "learning_rate": 9.623382338233823e-05,
      "loss": 3.1744,
      "step": 518880
    },
    {
      "epoch": 0.2766469813512874,
      "grad_norm": 0.796875,
      "learning_rate": 9.623182318231824e-05,
      "loss": 3.2423,
      "step": 518890
    },
    {
      "epoch": 0.2766656245874456,
      "grad_norm": 0.80859375,
      "learning_rate": 9.622982298229824e-05,
      "loss": 3.2101,
      "step": 518900
    },
    {
      "epoch": 0.2766842678236038,
      "grad_norm": 0.8046875,
      "learning_rate": 9.622782278227823e-05,
      "loss": 2.9402,
      "step": 518910
    },
    {
      "epoch": 0.27670291105976197,
      "grad_norm": 0.71484375,
      "learning_rate": 9.622582258225823e-05,
      "loss": 3.2194,
      "step": 518920
    },
    {
      "epoch": 0.2767215542959201,
      "grad_norm": 0.81640625,
      "learning_rate": 9.622382238223823e-05,
      "loss": 3.2797,
      "step": 518930
    },
    {
      "epoch": 0.2767401975320783,
      "grad_norm": 0.67578125,
      "learning_rate": 9.622182218221822e-05,
      "loss": 3.189,
      "step": 518940
    },
    {
      "epoch": 0.2767588407682365,
      "grad_norm": 0.73046875,
      "learning_rate": 9.621982198219822e-05,
      "loss": 3.2684,
      "step": 518950
    },
    {
      "epoch": 0.27677748400439467,
      "grad_norm": 0.68359375,
      "learning_rate": 9.621782178217821e-05,
      "loss": 3.2599,
      "step": 518960
    },
    {
      "epoch": 0.27679612724055286,
      "grad_norm": 0.66796875,
      "learning_rate": 9.621582158215822e-05,
      "loss": 3.1428,
      "step": 518970
    },
    {
      "epoch": 0.27681477047671105,
      "grad_norm": 0.83984375,
      "learning_rate": 9.621382138213822e-05,
      "loss": 3.2365,
      "step": 518980
    },
    {
      "epoch": 0.27683341371286924,
      "grad_norm": 0.8125,
      "learning_rate": 9.621182118211821e-05,
      "loss": 3.0918,
      "step": 518990
    },
    {
      "epoch": 0.2768520569490274,
      "grad_norm": 0.69921875,
      "learning_rate": 9.620982098209822e-05,
      "loss": 3.1613,
      "step": 519000
    },
    {
      "epoch": 0.2768707001851856,
      "grad_norm": 0.80859375,
      "learning_rate": 9.620782078207822e-05,
      "loss": 3.2949,
      "step": 519010
    },
    {
      "epoch": 0.2768893434213438,
      "grad_norm": 0.68359375,
      "learning_rate": 9.620582058205821e-05,
      "loss": 3.2153,
      "step": 519020
    },
    {
      "epoch": 0.27690798665750194,
      "grad_norm": 0.7109375,
      "learning_rate": 9.620382038203821e-05,
      "loss": 3.3726,
      "step": 519030
    },
    {
      "epoch": 0.2769266298936601,
      "grad_norm": 0.6796875,
      "learning_rate": 9.62018201820182e-05,
      "loss": 3.1499,
      "step": 519040
    },
    {
      "epoch": 0.2769452731298183,
      "grad_norm": 0.7421875,
      "learning_rate": 9.619981998199821e-05,
      "loss": 3.1705,
      "step": 519050
    },
    {
      "epoch": 0.2769639163659765,
      "grad_norm": 0.70703125,
      "learning_rate": 9.619781978197821e-05,
      "loss": 3.2454,
      "step": 519060
    },
    {
      "epoch": 0.2769825596021347,
      "grad_norm": 0.6875,
      "learning_rate": 9.61958195819582e-05,
      "loss": 3.3217,
      "step": 519070
    },
    {
      "epoch": 0.2770012028382929,
      "grad_norm": 0.84375,
      "learning_rate": 9.61938193819382e-05,
      "loss": 3.2431,
      "step": 519080
    },
    {
      "epoch": 0.27701984607445107,
      "grad_norm": 0.6796875,
      "learning_rate": 9.61918191819182e-05,
      "loss": 3.2287,
      "step": 519090
    },
    {
      "epoch": 0.27703848931060926,
      "grad_norm": 0.7578125,
      "learning_rate": 9.618981898189819e-05,
      "loss": 3.1517,
      "step": 519100
    },
    {
      "epoch": 0.27705713254676745,
      "grad_norm": 0.66015625,
      "learning_rate": 9.618781878187819e-05,
      "loss": 3.1798,
      "step": 519110
    },
    {
      "epoch": 0.27707577578292564,
      "grad_norm": 0.75390625,
      "learning_rate": 9.618581858185818e-05,
      "loss": 3.1221,
      "step": 519120
    },
    {
      "epoch": 0.2770944190190838,
      "grad_norm": 0.7578125,
      "learning_rate": 9.618381838183819e-05,
      "loss": 3.1557,
      "step": 519130
    },
    {
      "epoch": 0.27711306225524196,
      "grad_norm": 0.74609375,
      "learning_rate": 9.618181818181819e-05,
      "loss": 3.2491,
      "step": 519140
    },
    {
      "epoch": 0.27713170549140015,
      "grad_norm": 0.64453125,
      "learning_rate": 9.617981798179818e-05,
      "loss": 3.2325,
      "step": 519150
    },
    {
      "epoch": 0.27715034872755834,
      "grad_norm": 0.7265625,
      "learning_rate": 9.617781778177818e-05,
      "loss": 3.1457,
      "step": 519160
    },
    {
      "epoch": 0.2771689919637165,
      "grad_norm": 0.67578125,
      "learning_rate": 9.617581758175817e-05,
      "loss": 3.1456,
      "step": 519170
    },
    {
      "epoch": 0.2771876351998747,
      "grad_norm": 0.7109375,
      "learning_rate": 9.617381738173817e-05,
      "loss": 3.2873,
      "step": 519180
    },
    {
      "epoch": 0.2772062784360329,
      "grad_norm": 0.78515625,
      "learning_rate": 9.617181718171817e-05,
      "loss": 3.0937,
      "step": 519190
    },
    {
      "epoch": 0.2772249216721911,
      "grad_norm": 0.6640625,
      "learning_rate": 9.616981698169817e-05,
      "loss": 3.292,
      "step": 519200
    },
    {
      "epoch": 0.2772435649083493,
      "grad_norm": 0.7578125,
      "learning_rate": 9.616781678167818e-05,
      "loss": 3.357,
      "step": 519210
    },
    {
      "epoch": 0.27726220814450747,
      "grad_norm": 0.78125,
      "learning_rate": 9.616581658165818e-05,
      "loss": 3.0966,
      "step": 519220
    },
    {
      "epoch": 0.27728085138066566,
      "grad_norm": 0.734375,
      "learning_rate": 9.616381638163817e-05,
      "loss": 3.1873,
      "step": 519230
    },
    {
      "epoch": 0.2772994946168238,
      "grad_norm": 0.75,
      "learning_rate": 9.616181618161817e-05,
      "loss": 3.2454,
      "step": 519240
    },
    {
      "epoch": 0.277318137852982,
      "grad_norm": 0.703125,
      "learning_rate": 9.615981598159817e-05,
      "loss": 3.2199,
      "step": 519250
    },
    {
      "epoch": 0.27733678108914017,
      "grad_norm": 0.73828125,
      "learning_rate": 9.615781578157816e-05,
      "loss": 3.1999,
      "step": 519260
    },
    {
      "epoch": 0.27735542432529836,
      "grad_norm": 0.66015625,
      "learning_rate": 9.615581558155816e-05,
      "loss": 3.0433,
      "step": 519270
    },
    {
      "epoch": 0.27737406756145655,
      "grad_norm": 0.82421875,
      "learning_rate": 9.615381538153815e-05,
      "loss": 3.2192,
      "step": 519280
    },
    {
      "epoch": 0.27739271079761474,
      "grad_norm": 0.6953125,
      "learning_rate": 9.615181518151816e-05,
      "loss": 3.2081,
      "step": 519290
    },
    {
      "epoch": 0.27741135403377293,
      "grad_norm": 0.73046875,
      "learning_rate": 9.614981498149816e-05,
      "loss": 3.254,
      "step": 519300
    },
    {
      "epoch": 0.2774299972699311,
      "grad_norm": 0.75390625,
      "learning_rate": 9.614781478147815e-05,
      "loss": 3.197,
      "step": 519310
    },
    {
      "epoch": 0.2774486405060893,
      "grad_norm": 0.76953125,
      "learning_rate": 9.614581458145815e-05,
      "loss": 3.1441,
      "step": 519320
    },
    {
      "epoch": 0.2774672837422475,
      "grad_norm": 0.6640625,
      "learning_rate": 9.614381438143814e-05,
      "loss": 3.2525,
      "step": 519330
    },
    {
      "epoch": 0.2774859269784057,
      "grad_norm": 0.859375,
      "learning_rate": 9.614181418141814e-05,
      "loss": 3.2217,
      "step": 519340
    },
    {
      "epoch": 0.2775045702145638,
      "grad_norm": 0.71484375,
      "learning_rate": 9.613981398139814e-05,
      "loss": 3.1487,
      "step": 519350
    },
    {
      "epoch": 0.277523213450722,
      "grad_norm": 0.6796875,
      "learning_rate": 9.613781378137814e-05,
      "loss": 3.1073,
      "step": 519360
    },
    {
      "epoch": 0.2775418566868802,
      "grad_norm": 0.73828125,
      "learning_rate": 9.613581358135814e-05,
      "loss": 3.1873,
      "step": 519370
    },
    {
      "epoch": 0.2775604999230384,
      "grad_norm": 0.59765625,
      "learning_rate": 9.613381338133814e-05,
      "loss": 3.1618,
      "step": 519380
    },
    {
      "epoch": 0.2775791431591966,
      "grad_norm": 0.796875,
      "learning_rate": 9.613181318131813e-05,
      "loss": 3.1537,
      "step": 519390
    },
    {
      "epoch": 0.27759778639535476,
      "grad_norm": 0.73046875,
      "learning_rate": 9.612981298129814e-05,
      "loss": 3.2641,
      "step": 519400
    },
    {
      "epoch": 0.27761642963151295,
      "grad_norm": 0.671875,
      "learning_rate": 9.612781278127814e-05,
      "loss": 3.045,
      "step": 519410
    },
    {
      "epoch": 0.27763507286767114,
      "grad_norm": 0.7109375,
      "learning_rate": 9.612581258125813e-05,
      "loss": 3.2123,
      "step": 519420
    },
    {
      "epoch": 0.27765371610382933,
      "grad_norm": 0.69140625,
      "learning_rate": 9.612381238123813e-05,
      "loss": 3.0743,
      "step": 519430
    },
    {
      "epoch": 0.2776723593399875,
      "grad_norm": 0.62109375,
      "learning_rate": 9.612181218121812e-05,
      "loss": 3.3188,
      "step": 519440
    },
    {
      "epoch": 0.27769100257614565,
      "grad_norm": 0.75390625,
      "learning_rate": 9.611981198119813e-05,
      "loss": 3.1484,
      "step": 519450
    },
    {
      "epoch": 0.27770964581230384,
      "grad_norm": 0.66015625,
      "learning_rate": 9.611781178117813e-05,
      "loss": 3.1543,
      "step": 519460
    },
    {
      "epoch": 0.27772828904846203,
      "grad_norm": 0.78515625,
      "learning_rate": 9.611581158115812e-05,
      "loss": 3.121,
      "step": 519470
    },
    {
      "epoch": 0.2777469322846202,
      "grad_norm": 0.7578125,
      "learning_rate": 9.611381138113812e-05,
      "loss": 3.2474,
      "step": 519480
    },
    {
      "epoch": 0.2777655755207784,
      "grad_norm": 0.7421875,
      "learning_rate": 9.611181118111811e-05,
      "loss": 3.2095,
      "step": 519490
    },
    {
      "epoch": 0.2777842187569366,
      "grad_norm": 0.6875,
      "learning_rate": 9.610981098109811e-05,
      "loss": 3.2503,
      "step": 519500
    },
    {
      "epoch": 0.2778028619930948,
      "grad_norm": 0.78125,
      "learning_rate": 9.61078107810781e-05,
      "loss": 3.2349,
      "step": 519510
    },
    {
      "epoch": 0.277821505229253,
      "grad_norm": 0.7578125,
      "learning_rate": 9.610581058105811e-05,
      "loss": 3.1426,
      "step": 519520
    },
    {
      "epoch": 0.27784014846541116,
      "grad_norm": 0.8515625,
      "learning_rate": 9.610381038103811e-05,
      "loss": 3.2541,
      "step": 519530
    },
    {
      "epoch": 0.27785879170156935,
      "grad_norm": 0.71875,
      "learning_rate": 9.61018101810181e-05,
      "loss": 3.1342,
      "step": 519540
    },
    {
      "epoch": 0.2778774349377275,
      "grad_norm": 0.65625,
      "learning_rate": 9.60998099809981e-05,
      "loss": 3.1993,
      "step": 519550
    },
    {
      "epoch": 0.2778960781738857,
      "grad_norm": 0.828125,
      "learning_rate": 9.60978097809781e-05,
      "loss": 3.1296,
      "step": 519560
    },
    {
      "epoch": 0.27791472141004386,
      "grad_norm": 0.796875,
      "learning_rate": 9.609580958095809e-05,
      "loss": 3.1683,
      "step": 519570
    },
    {
      "epoch": 0.27793336464620205,
      "grad_norm": 0.7421875,
      "learning_rate": 9.609380938093809e-05,
      "loss": 3.2609,
      "step": 519580
    },
    {
      "epoch": 0.27795200788236024,
      "grad_norm": 0.69921875,
      "learning_rate": 9.60918091809181e-05,
      "loss": 3.2053,
      "step": 519590
    },
    {
      "epoch": 0.27797065111851843,
      "grad_norm": 0.7265625,
      "learning_rate": 9.608980898089809e-05,
      "loss": 3.1503,
      "step": 519600
    },
    {
      "epoch": 0.2779892943546766,
      "grad_norm": 0.8671875,
      "learning_rate": 9.60878087808781e-05,
      "loss": 3.25,
      "step": 519610
    },
    {
      "epoch": 0.2780079375908348,
      "grad_norm": 0.6484375,
      "learning_rate": 9.60858085808581e-05,
      "loss": 3.2116,
      "step": 519620
    },
    {
      "epoch": 0.278026580826993,
      "grad_norm": 0.7265625,
      "learning_rate": 9.60838083808381e-05,
      "loss": 3.2747,
      "step": 519630
    },
    {
      "epoch": 0.2780452240631512,
      "grad_norm": 0.703125,
      "learning_rate": 9.608180818081809e-05,
      "loss": 3.2965,
      "step": 519640
    },
    {
      "epoch": 0.2780638672993094,
      "grad_norm": 0.8359375,
      "learning_rate": 9.607980798079808e-05,
      "loss": 3.2803,
      "step": 519650
    },
    {
      "epoch": 0.2780825105354675,
      "grad_norm": 0.73828125,
      "learning_rate": 9.607780778077808e-05,
      "loss": 3.2119,
      "step": 519660
    },
    {
      "epoch": 0.2781011537716257,
      "grad_norm": 0.6953125,
      "learning_rate": 9.607580758075808e-05,
      "loss": 3.2913,
      "step": 519670
    },
    {
      "epoch": 0.2781197970077839,
      "grad_norm": 0.6640625,
      "learning_rate": 9.607380738073809e-05,
      "loss": 3.2903,
      "step": 519680
    },
    {
      "epoch": 0.2781384402439421,
      "grad_norm": 0.69140625,
      "learning_rate": 9.607180718071808e-05,
      "loss": 3.1369,
      "step": 519690
    },
    {
      "epoch": 0.27815708348010026,
      "grad_norm": 0.70703125,
      "learning_rate": 9.606980698069808e-05,
      "loss": 3.2731,
      "step": 519700
    },
    {
      "epoch": 0.27817572671625845,
      "grad_norm": 0.7890625,
      "learning_rate": 9.606780678067807e-05,
      "loss": 3.2071,
      "step": 519710
    },
    {
      "epoch": 0.27819436995241664,
      "grad_norm": 0.6484375,
      "learning_rate": 9.606580658065807e-05,
      "loss": 3.1987,
      "step": 519720
    },
    {
      "epoch": 0.27821301318857483,
      "grad_norm": 0.66796875,
      "learning_rate": 9.606380638063806e-05,
      "loss": 3.0672,
      "step": 519730
    },
    {
      "epoch": 0.278231656424733,
      "grad_norm": 0.76953125,
      "learning_rate": 9.606180618061806e-05,
      "loss": 3.2701,
      "step": 519740
    },
    {
      "epoch": 0.2782502996608912,
      "grad_norm": 0.875,
      "learning_rate": 9.605980598059807e-05,
      "loss": 3.3105,
      "step": 519750
    },
    {
      "epoch": 0.27826894289704934,
      "grad_norm": 0.625,
      "learning_rate": 9.605780578057806e-05,
      "loss": 3.2183,
      "step": 519760
    },
    {
      "epoch": 0.27828758613320753,
      "grad_norm": 0.7265625,
      "learning_rate": 9.605580558055806e-05,
      "loss": 3.3225,
      "step": 519770
    },
    {
      "epoch": 0.2783062293693657,
      "grad_norm": 0.875,
      "learning_rate": 9.605380538053805e-05,
      "loss": 3.2889,
      "step": 519780
    },
    {
      "epoch": 0.2783248726055239,
      "grad_norm": 0.7578125,
      "learning_rate": 9.605180518051806e-05,
      "loss": 3.322,
      "step": 519790
    },
    {
      "epoch": 0.2783435158416821,
      "grad_norm": 0.7578125,
      "learning_rate": 9.604980498049806e-05,
      "loss": 3.2443,
      "step": 519800
    },
    {
      "epoch": 0.2783621590778403,
      "grad_norm": 0.7265625,
      "learning_rate": 9.604780478047805e-05,
      "loss": 3.2776,
      "step": 519810
    },
    {
      "epoch": 0.2783808023139985,
      "grad_norm": 0.60546875,
      "learning_rate": 9.604580458045805e-05,
      "loss": 3.1908,
      "step": 519820
    },
    {
      "epoch": 0.27839944555015667,
      "grad_norm": 0.69921875,
      "learning_rate": 9.604380438043805e-05,
      "loss": 3.3016,
      "step": 519830
    },
    {
      "epoch": 0.27841808878631485,
      "grad_norm": 0.67578125,
      "learning_rate": 9.604180418041806e-05,
      "loss": 3.1546,
      "step": 519840
    },
    {
      "epoch": 0.27843673202247304,
      "grad_norm": 0.7109375,
      "learning_rate": 9.603980398039805e-05,
      "loss": 3.2082,
      "step": 519850
    },
    {
      "epoch": 0.27845537525863123,
      "grad_norm": 0.7421875,
      "learning_rate": 9.603780378037805e-05,
      "loss": 3.2104,
      "step": 519860
    },
    {
      "epoch": 0.27847401849478937,
      "grad_norm": 0.75,
      "learning_rate": 9.603580358035804e-05,
      "loss": 3.296,
      "step": 519870
    },
    {
      "epoch": 0.27849266173094755,
      "grad_norm": 0.75390625,
      "learning_rate": 9.603380338033804e-05,
      "loss": 3.1164,
      "step": 519880
    },
    {
      "epoch": 0.27851130496710574,
      "grad_norm": 0.7109375,
      "learning_rate": 9.603180318031803e-05,
      "loss": 3.2769,
      "step": 519890
    },
    {
      "epoch": 0.27852994820326393,
      "grad_norm": 0.67578125,
      "learning_rate": 9.602980298029803e-05,
      "loss": 3.2375,
      "step": 519900
    },
    {
      "epoch": 0.2785485914394221,
      "grad_norm": 0.82421875,
      "learning_rate": 9.602780278027802e-05,
      "loss": 3.2375,
      "step": 519910
    },
    {
      "epoch": 0.2785672346755803,
      "grad_norm": 0.68359375,
      "learning_rate": 9.602580258025803e-05,
      "loss": 3.1292,
      "step": 519920
    },
    {
      "epoch": 0.2785858779117385,
      "grad_norm": 0.75,
      "learning_rate": 9.602380238023803e-05,
      "loss": 3.2068,
      "step": 519930
    },
    {
      "epoch": 0.2786045211478967,
      "grad_norm": 0.73046875,
      "learning_rate": 9.602180218021802e-05,
      "loss": 3.2278,
      "step": 519940
    },
    {
      "epoch": 0.2786231643840549,
      "grad_norm": 0.7421875,
      "learning_rate": 9.601980198019802e-05,
      "loss": 3.253,
      "step": 519950
    },
    {
      "epoch": 0.27864180762021307,
      "grad_norm": 0.66796875,
      "learning_rate": 9.601780178017802e-05,
      "loss": 3.2261,
      "step": 519960
    },
    {
      "epoch": 0.2786604508563712,
      "grad_norm": 0.70703125,
      "learning_rate": 9.601580158015801e-05,
      "loss": 3.1566,
      "step": 519970
    },
    {
      "epoch": 0.2786790940925294,
      "grad_norm": 0.67578125,
      "learning_rate": 9.601380138013801e-05,
      "loss": 3.0158,
      "step": 519980
    },
    {
      "epoch": 0.2786977373286876,
      "grad_norm": 0.7578125,
      "learning_rate": 9.601180118011802e-05,
      "loss": 3.1906,
      "step": 519990
    },
    {
      "epoch": 0.27871638056484577,
      "grad_norm": 0.7578125,
      "learning_rate": 9.600980098009803e-05,
      "loss": 3.1686,
      "step": 520000
    },
    {
      "epoch": 0.27873502380100396,
      "grad_norm": 0.75,
      "learning_rate": 9.600780078007802e-05,
      "loss": 3.2784,
      "step": 520010
    },
    {
      "epoch": 0.27875366703716215,
      "grad_norm": 0.734375,
      "learning_rate": 9.600580058005802e-05,
      "loss": 3.2123,
      "step": 520020
    },
    {
      "epoch": 0.27877231027332033,
      "grad_norm": 0.71875,
      "learning_rate": 9.600380038003801e-05,
      "loss": 3.2377,
      "step": 520030
    },
    {
      "epoch": 0.2787909535094785,
      "grad_norm": 0.7265625,
      "learning_rate": 9.600180018001801e-05,
      "loss": 3.1599,
      "step": 520040
    },
    {
      "epoch": 0.2788095967456367,
      "grad_norm": 0.63671875,
      "learning_rate": 9.5999799979998e-05,
      "loss": 3.1459,
      "step": 520050
    },
    {
      "epoch": 0.2788282399817949,
      "grad_norm": 0.7421875,
      "learning_rate": 9.5997799779978e-05,
      "loss": 3.1427,
      "step": 520060
    },
    {
      "epoch": 0.2788468832179531,
      "grad_norm": 0.671875,
      "learning_rate": 9.5995799579958e-05,
      "loss": 3.1988,
      "step": 520070
    },
    {
      "epoch": 0.2788655264541112,
      "grad_norm": 0.8203125,
      "learning_rate": 9.5993799379938e-05,
      "loss": 3.0793,
      "step": 520080
    },
    {
      "epoch": 0.2788841696902694,
      "grad_norm": 0.765625,
      "learning_rate": 9.5991799179918e-05,
      "loss": 3.265,
      "step": 520090
    },
    {
      "epoch": 0.2789028129264276,
      "grad_norm": 0.8671875,
      "learning_rate": 9.5989798979898e-05,
      "loss": 3.1301,
      "step": 520100
    },
    {
      "epoch": 0.2789214561625858,
      "grad_norm": 0.74609375,
      "learning_rate": 9.598779877987799e-05,
      "loss": 3.1795,
      "step": 520110
    },
    {
      "epoch": 0.278940099398744,
      "grad_norm": 0.5859375,
      "learning_rate": 9.598579857985799e-05,
      "loss": 3.1715,
      "step": 520120
    },
    {
      "epoch": 0.27895874263490217,
      "grad_norm": 0.61328125,
      "learning_rate": 9.598379837983798e-05,
      "loss": 3.2843,
      "step": 520130
    },
    {
      "epoch": 0.27897738587106036,
      "grad_norm": 0.7421875,
      "learning_rate": 9.598179817981798e-05,
      "loss": 3.1611,
      "step": 520140
    },
    {
      "epoch": 0.27899602910721855,
      "grad_norm": 0.7890625,
      "learning_rate": 9.597979797979799e-05,
      "loss": 3.1938,
      "step": 520150
    },
    {
      "epoch": 0.27901467234337674,
      "grad_norm": 0.75390625,
      "learning_rate": 9.597779777977798e-05,
      "loss": 3.2525,
      "step": 520160
    },
    {
      "epoch": 0.2790333155795349,
      "grad_norm": 0.70703125,
      "learning_rate": 9.597579757975798e-05,
      "loss": 3.1729,
      "step": 520170
    },
    {
      "epoch": 0.27905195881569306,
      "grad_norm": 0.65625,
      "learning_rate": 9.597379737973797e-05,
      "loss": 3.1989,
      "step": 520180
    },
    {
      "epoch": 0.27907060205185125,
      "grad_norm": 0.7265625,
      "learning_rate": 9.597179717971798e-05,
      "loss": 3.242,
      "step": 520190
    },
    {
      "epoch": 0.27908924528800944,
      "grad_norm": 0.84375,
      "learning_rate": 9.596979697969798e-05,
      "loss": 3.1068,
      "step": 520200
    },
    {
      "epoch": 0.2791078885241676,
      "grad_norm": 0.83984375,
      "learning_rate": 9.596779677967797e-05,
      "loss": 3.1681,
      "step": 520210
    },
    {
      "epoch": 0.2791265317603258,
      "grad_norm": 0.6875,
      "learning_rate": 9.596579657965797e-05,
      "loss": 3.3182,
      "step": 520220
    },
    {
      "epoch": 0.279145174996484,
      "grad_norm": 0.73046875,
      "learning_rate": 9.596379637963796e-05,
      "loss": 3.1124,
      "step": 520230
    },
    {
      "epoch": 0.2791638182326422,
      "grad_norm": 0.75,
      "learning_rate": 9.596179617961797e-05,
      "loss": 3.1487,
      "step": 520240
    },
    {
      "epoch": 0.2791824614688004,
      "grad_norm": 0.6796875,
      "learning_rate": 9.595979597959797e-05,
      "loss": 3.2652,
      "step": 520250
    },
    {
      "epoch": 0.27920110470495857,
      "grad_norm": 0.6953125,
      "learning_rate": 9.595779577957797e-05,
      "loss": 3.0753,
      "step": 520260
    },
    {
      "epoch": 0.27921974794111676,
      "grad_norm": 0.7265625,
      "learning_rate": 9.595579557955796e-05,
      "loss": 3.3293,
      "step": 520270
    },
    {
      "epoch": 0.27923839117727495,
      "grad_norm": 0.765625,
      "learning_rate": 9.595379537953796e-05,
      "loss": 3.0934,
      "step": 520280
    },
    {
      "epoch": 0.2792570344134331,
      "grad_norm": 0.66796875,
      "learning_rate": 9.595179517951795e-05,
      "loss": 3.1976,
      "step": 520290
    },
    {
      "epoch": 0.27927567764959127,
      "grad_norm": 0.7265625,
      "learning_rate": 9.594979497949795e-05,
      "loss": 3.2548,
      "step": 520300
    },
    {
      "epoch": 0.27929432088574946,
      "grad_norm": 0.6875,
      "learning_rate": 9.594779477947796e-05,
      "loss": 3.2605,
      "step": 520310
    },
    {
      "epoch": 0.27931296412190765,
      "grad_norm": 1.15625,
      "learning_rate": 9.594579457945795e-05,
      "loss": 3.1359,
      "step": 520320
    },
    {
      "epoch": 0.27933160735806584,
      "grad_norm": 0.71875,
      "learning_rate": 9.594379437943795e-05,
      "loss": 3.2029,
      "step": 520330
    },
    {
      "epoch": 0.279350250594224,
      "grad_norm": 0.7421875,
      "learning_rate": 9.594179417941794e-05,
      "loss": 3.097,
      "step": 520340
    },
    {
      "epoch": 0.2793688938303822,
      "grad_norm": 0.66796875,
      "learning_rate": 9.593979397939794e-05,
      "loss": 3.3212,
      "step": 520350
    },
    {
      "epoch": 0.2793875370665404,
      "grad_norm": 0.77734375,
      "learning_rate": 9.593779377937793e-05,
      "loss": 3.1572,
      "step": 520360
    },
    {
      "epoch": 0.2794061803026986,
      "grad_norm": 0.765625,
      "learning_rate": 9.593579357935793e-05,
      "loss": 3.1367,
      "step": 520370
    },
    {
      "epoch": 0.2794248235388568,
      "grad_norm": 0.67578125,
      "learning_rate": 9.593379337933794e-05,
      "loss": 3.2206,
      "step": 520380
    },
    {
      "epoch": 0.2794434667750149,
      "grad_norm": 0.75,
      "learning_rate": 9.593179317931793e-05,
      "loss": 3.1167,
      "step": 520390
    },
    {
      "epoch": 0.2794621100111731,
      "grad_norm": 0.6796875,
      "learning_rate": 9.592979297929794e-05,
      "loss": 3.2158,
      "step": 520400
    },
    {
      "epoch": 0.2794807532473313,
      "grad_norm": 0.7265625,
      "learning_rate": 9.592779277927794e-05,
      "loss": 3.218,
      "step": 520410
    },
    {
      "epoch": 0.2794993964834895,
      "grad_norm": 0.73828125,
      "learning_rate": 9.592579257925794e-05,
      "loss": 3.236,
      "step": 520420
    },
    {
      "epoch": 0.27951803971964767,
      "grad_norm": 0.69921875,
      "learning_rate": 9.592379237923793e-05,
      "loss": 3.1608,
      "step": 520430
    },
    {
      "epoch": 0.27953668295580586,
      "grad_norm": 0.765625,
      "learning_rate": 9.592179217921793e-05,
      "loss": 3.1669,
      "step": 520440
    },
    {
      "epoch": 0.27955532619196405,
      "grad_norm": 0.71875,
      "learning_rate": 9.591979197919792e-05,
      "loss": 3.1274,
      "step": 520450
    },
    {
      "epoch": 0.27957396942812224,
      "grad_norm": 0.73046875,
      "learning_rate": 9.591779177917792e-05,
      "loss": 3.2193,
      "step": 520460
    },
    {
      "epoch": 0.2795926126642804,
      "grad_norm": 0.78125,
      "learning_rate": 9.591579157915793e-05,
      "loss": 3.1693,
      "step": 520470
    },
    {
      "epoch": 0.2796112559004386,
      "grad_norm": 0.6953125,
      "learning_rate": 9.591379137913792e-05,
      "loss": 3.3429,
      "step": 520480
    },
    {
      "epoch": 0.2796298991365968,
      "grad_norm": 0.81640625,
      "learning_rate": 9.591179117911792e-05,
      "loss": 3.2923,
      "step": 520490
    },
    {
      "epoch": 0.27964854237275494,
      "grad_norm": 0.765625,
      "learning_rate": 9.590979097909791e-05,
      "loss": 3.1781,
      "step": 520500
    },
    {
      "epoch": 0.2796671856089131,
      "grad_norm": 0.58984375,
      "learning_rate": 9.590779077907791e-05,
      "loss": 3.2205,
      "step": 520510
    },
    {
      "epoch": 0.2796858288450713,
      "grad_norm": 0.734375,
      "learning_rate": 9.59057905790579e-05,
      "loss": 3.1008,
      "step": 520520
    },
    {
      "epoch": 0.2797044720812295,
      "grad_norm": 0.78125,
      "learning_rate": 9.59037903790379e-05,
      "loss": 3.258,
      "step": 520530
    },
    {
      "epoch": 0.2797231153173877,
      "grad_norm": 0.74609375,
      "learning_rate": 9.590179017901791e-05,
      "loss": 3.1744,
      "step": 520540
    },
    {
      "epoch": 0.2797417585535459,
      "grad_norm": 0.66796875,
      "learning_rate": 9.58997899789979e-05,
      "loss": 3.108,
      "step": 520550
    },
    {
      "epoch": 0.27976040178970407,
      "grad_norm": 0.87890625,
      "learning_rate": 9.58977897789779e-05,
      "loss": 3.0728,
      "step": 520560
    },
    {
      "epoch": 0.27977904502586226,
      "grad_norm": 0.65625,
      "learning_rate": 9.58957895789579e-05,
      "loss": 3.2293,
      "step": 520570
    },
    {
      "epoch": 0.27979768826202045,
      "grad_norm": 0.7265625,
      "learning_rate": 9.58937893789379e-05,
      "loss": 3.217,
      "step": 520580
    },
    {
      "epoch": 0.27981633149817864,
      "grad_norm": 0.703125,
      "learning_rate": 9.58917891789179e-05,
      "loss": 3.2925,
      "step": 520590
    },
    {
      "epoch": 0.27983497473433677,
      "grad_norm": 0.7578125,
      "learning_rate": 9.58897889788979e-05,
      "loss": 3.1849,
      "step": 520600
    },
    {
      "epoch": 0.27985361797049496,
      "grad_norm": 0.78515625,
      "learning_rate": 9.588778877887789e-05,
      "loss": 3.1286,
      "step": 520610
    },
    {
      "epoch": 0.27987226120665315,
      "grad_norm": 0.70703125,
      "learning_rate": 9.588578857885789e-05,
      "loss": 3.3333,
      "step": 520620
    },
    {
      "epoch": 0.27989090444281134,
      "grad_norm": 0.66015625,
      "learning_rate": 9.58837883788379e-05,
      "loss": 3.1079,
      "step": 520630
    },
    {
      "epoch": 0.27990954767896953,
      "grad_norm": 0.73828125,
      "learning_rate": 9.588178817881789e-05,
      "loss": 3.2197,
      "step": 520640
    },
    {
      "epoch": 0.2799281909151277,
      "grad_norm": 0.7109375,
      "learning_rate": 9.587978797879789e-05,
      "loss": 3.309,
      "step": 520650
    },
    {
      "epoch": 0.2799468341512859,
      "grad_norm": 0.7109375,
      "learning_rate": 9.587778777877788e-05,
      "loss": 3.2606,
      "step": 520660
    },
    {
      "epoch": 0.2799654773874441,
      "grad_norm": 0.73828125,
      "learning_rate": 9.587578757875788e-05,
      "loss": 3.2132,
      "step": 520670
    },
    {
      "epoch": 0.2799841206236023,
      "grad_norm": 0.87890625,
      "learning_rate": 9.587378737873787e-05,
      "loss": 3.0675,
      "step": 520680
    },
    {
      "epoch": 0.2800027638597605,
      "grad_norm": 0.63671875,
      "learning_rate": 9.587178717871787e-05,
      "loss": 3.1682,
      "step": 520690
    },
    {
      "epoch": 0.28002140709591866,
      "grad_norm": 0.7890625,
      "learning_rate": 9.586978697869788e-05,
      "loss": 3.2268,
      "step": 520700
    },
    {
      "epoch": 0.2800400503320768,
      "grad_norm": 0.77734375,
      "learning_rate": 9.586778677867788e-05,
      "loss": 3.2162,
      "step": 520710
    },
    {
      "epoch": 0.280058693568235,
      "grad_norm": 0.7109375,
      "learning_rate": 9.586578657865787e-05,
      "loss": 3.1521,
      "step": 520720
    },
    {
      "epoch": 0.2800773368043932,
      "grad_norm": 0.74609375,
      "learning_rate": 9.586378637863787e-05,
      "loss": 3.2869,
      "step": 520730
    },
    {
      "epoch": 0.28009598004055136,
      "grad_norm": 0.8046875,
      "learning_rate": 9.586178617861786e-05,
      "loss": 3.1367,
      "step": 520740
    },
    {
      "epoch": 0.28011462327670955,
      "grad_norm": 0.73828125,
      "learning_rate": 9.585978597859786e-05,
      "loss": 3.2284,
      "step": 520750
    },
    {
      "epoch": 0.28013326651286774,
      "grad_norm": 0.79296875,
      "learning_rate": 9.585778577857785e-05,
      "loss": 3.238,
      "step": 520760
    },
    {
      "epoch": 0.28015190974902593,
      "grad_norm": 0.765625,
      "learning_rate": 9.585578557855785e-05,
      "loss": 3.1641,
      "step": 520770
    },
    {
      "epoch": 0.2801705529851841,
      "grad_norm": 0.76953125,
      "learning_rate": 9.585378537853786e-05,
      "loss": 3.2493,
      "step": 520780
    },
    {
      "epoch": 0.2801891962213423,
      "grad_norm": 0.80078125,
      "learning_rate": 9.585178517851787e-05,
      "loss": 3.0936,
      "step": 520790
    },
    {
      "epoch": 0.2802078394575005,
      "grad_norm": 0.703125,
      "learning_rate": 9.584978497849786e-05,
      "loss": 3.1934,
      "step": 520800
    },
    {
      "epoch": 0.28022648269365863,
      "grad_norm": 0.73046875,
      "learning_rate": 9.584778477847786e-05,
      "loss": 3.3437,
      "step": 520810
    },
    {
      "epoch": 0.2802451259298168,
      "grad_norm": 0.6953125,
      "learning_rate": 9.584578457845785e-05,
      "loss": 3.2098,
      "step": 520820
    },
    {
      "epoch": 0.280263769165975,
      "grad_norm": 0.6953125,
      "learning_rate": 9.584378437843785e-05,
      "loss": 3.0735,
      "step": 520830
    },
    {
      "epoch": 0.2802824124021332,
      "grad_norm": 0.65625,
      "learning_rate": 9.584178417841785e-05,
      "loss": 3.1982,
      "step": 520840
    },
    {
      "epoch": 0.2803010556382914,
      "grad_norm": 0.6953125,
      "learning_rate": 9.583978397839784e-05,
      "loss": 3.1571,
      "step": 520850
    },
    {
      "epoch": 0.2803196988744496,
      "grad_norm": 0.84765625,
      "learning_rate": 9.583778377837785e-05,
      "loss": 3.2975,
      "step": 520860
    },
    {
      "epoch": 0.28033834211060776,
      "grad_norm": 0.65234375,
      "learning_rate": 9.583578357835785e-05,
      "loss": 3.2099,
      "step": 520870
    },
    {
      "epoch": 0.28035698534676595,
      "grad_norm": 0.80078125,
      "learning_rate": 9.583378337833784e-05,
      "loss": 3.2039,
      "step": 520880
    },
    {
      "epoch": 0.28037562858292414,
      "grad_norm": 0.68359375,
      "learning_rate": 9.583178317831784e-05,
      "loss": 3.1614,
      "step": 520890
    },
    {
      "epoch": 0.28039427181908233,
      "grad_norm": 0.6953125,
      "learning_rate": 9.582978297829783e-05,
      "loss": 3.1813,
      "step": 520900
    },
    {
      "epoch": 0.28041291505524046,
      "grad_norm": 0.890625,
      "learning_rate": 9.582778277827783e-05,
      "loss": 3.2956,
      "step": 520910
    },
    {
      "epoch": 0.28043155829139865,
      "grad_norm": 0.80859375,
      "learning_rate": 9.582578257825782e-05,
      "loss": 3.0255,
      "step": 520920
    },
    {
      "epoch": 0.28045020152755684,
      "grad_norm": 0.78125,
      "learning_rate": 9.582378237823782e-05,
      "loss": 3.3445,
      "step": 520930
    },
    {
      "epoch": 0.28046884476371503,
      "grad_norm": 0.7578125,
      "learning_rate": 9.582178217821783e-05,
      "loss": 3.167,
      "step": 520940
    },
    {
      "epoch": 0.2804874879998732,
      "grad_norm": 0.796875,
      "learning_rate": 9.581978197819782e-05,
      "loss": 3.1226,
      "step": 520950
    },
    {
      "epoch": 0.2805061312360314,
      "grad_norm": 0.6796875,
      "learning_rate": 9.581778177817782e-05,
      "loss": 3.3063,
      "step": 520960
    },
    {
      "epoch": 0.2805247744721896,
      "grad_norm": 0.76953125,
      "learning_rate": 9.581578157815781e-05,
      "loss": 3.212,
      "step": 520970
    },
    {
      "epoch": 0.2805434177083478,
      "grad_norm": 0.640625,
      "learning_rate": 9.581378137813782e-05,
      "loss": 3.1641,
      "step": 520980
    },
    {
      "epoch": 0.280562060944506,
      "grad_norm": 0.73046875,
      "learning_rate": 9.581178117811782e-05,
      "loss": 3.1783,
      "step": 520990
    },
    {
      "epoch": 0.28058070418066416,
      "grad_norm": 0.7421875,
      "learning_rate": 9.580978097809782e-05,
      "loss": 3.0914,
      "step": 521000
    },
    {
      "epoch": 0.28059934741682235,
      "grad_norm": 0.69921875,
      "learning_rate": 9.580778077807781e-05,
      "loss": 3.1336,
      "step": 521010
    },
    {
      "epoch": 0.2806179906529805,
      "grad_norm": 0.7421875,
      "learning_rate": 9.580578057805782e-05,
      "loss": 3.1461,
      "step": 521020
    },
    {
      "epoch": 0.2806366338891387,
      "grad_norm": 0.6875,
      "learning_rate": 9.580378037803782e-05,
      "loss": 3.2953,
      "step": 521030
    },
    {
      "epoch": 0.28065527712529686,
      "grad_norm": 0.74609375,
      "learning_rate": 9.580178017801781e-05,
      "loss": 3.2768,
      "step": 521040
    },
    {
      "epoch": 0.28067392036145505,
      "grad_norm": 0.765625,
      "learning_rate": 9.57997799779978e-05,
      "loss": 3.2633,
      "step": 521050
    },
    {
      "epoch": 0.28069256359761324,
      "grad_norm": 0.75390625,
      "learning_rate": 9.57977797779778e-05,
      "loss": 3.2319,
      "step": 521060
    },
    {
      "epoch": 0.28071120683377143,
      "grad_norm": 0.7109375,
      "learning_rate": 9.57957795779578e-05,
      "loss": 3.2455,
      "step": 521070
    },
    {
      "epoch": 0.2807298500699296,
      "grad_norm": 0.734375,
      "learning_rate": 9.57937793779378e-05,
      "loss": 3.0259,
      "step": 521080
    },
    {
      "epoch": 0.2807484933060878,
      "grad_norm": 0.7265625,
      "learning_rate": 9.579177917791779e-05,
      "loss": 3.21,
      "step": 521090
    },
    {
      "epoch": 0.280767136542246,
      "grad_norm": 0.7109375,
      "learning_rate": 9.57897789778978e-05,
      "loss": 3.1434,
      "step": 521100
    },
    {
      "epoch": 0.2807857797784042,
      "grad_norm": 0.67578125,
      "learning_rate": 9.57877787778778e-05,
      "loss": 3.1737,
      "step": 521110
    },
    {
      "epoch": 0.2808044230145623,
      "grad_norm": 0.6875,
      "learning_rate": 9.578577857785779e-05,
      "loss": 3.3081,
      "step": 521120
    },
    {
      "epoch": 0.2808230662507205,
      "grad_norm": 0.69921875,
      "learning_rate": 9.578377837783779e-05,
      "loss": 3.2258,
      "step": 521130
    },
    {
      "epoch": 0.2808417094868787,
      "grad_norm": 0.8046875,
      "learning_rate": 9.578177817781778e-05,
      "loss": 3.0776,
      "step": 521140
    },
    {
      "epoch": 0.2808603527230369,
      "grad_norm": 0.828125,
      "learning_rate": 9.577977797779778e-05,
      "loss": 3.2264,
      "step": 521150
    },
    {
      "epoch": 0.2808789959591951,
      "grad_norm": 0.7109375,
      "learning_rate": 9.577777777777777e-05,
      "loss": 3.3929,
      "step": 521160
    },
    {
      "epoch": 0.28089763919535327,
      "grad_norm": 0.79296875,
      "learning_rate": 9.577577757775778e-05,
      "loss": 3.241,
      "step": 521170
    },
    {
      "epoch": 0.28091628243151145,
      "grad_norm": 0.7734375,
      "learning_rate": 9.577377737773779e-05,
      "loss": 3.2363,
      "step": 521180
    },
    {
      "epoch": 0.28093492566766964,
      "grad_norm": 0.66796875,
      "learning_rate": 9.577177717771779e-05,
      "loss": 3.1856,
      "step": 521190
    },
    {
      "epoch": 0.28095356890382783,
      "grad_norm": 0.78515625,
      "learning_rate": 9.576977697769778e-05,
      "loss": 3.2531,
      "step": 521200
    },
    {
      "epoch": 0.280972212139986,
      "grad_norm": 0.85546875,
      "learning_rate": 9.576777677767778e-05,
      "loss": 3.2386,
      "step": 521210
    },
    {
      "epoch": 0.2809908553761442,
      "grad_norm": 0.734375,
      "learning_rate": 9.576577657765777e-05,
      "loss": 3.2034,
      "step": 521220
    },
    {
      "epoch": 0.28100949861230234,
      "grad_norm": 0.72265625,
      "learning_rate": 9.576377637763777e-05,
      "loss": 3.3201,
      "step": 521230
    },
    {
      "epoch": 0.28102814184846053,
      "grad_norm": 0.8515625,
      "learning_rate": 9.576177617761776e-05,
      "loss": 3.2519,
      "step": 521240
    },
    {
      "epoch": 0.2810467850846187,
      "grad_norm": 0.71484375,
      "learning_rate": 9.575977597759776e-05,
      "loss": 3.2221,
      "step": 521250
    },
    {
      "epoch": 0.2810654283207769,
      "grad_norm": 0.796875,
      "learning_rate": 9.575777577757777e-05,
      "loss": 3.1339,
      "step": 521260
    },
    {
      "epoch": 0.2810840715569351,
      "grad_norm": 0.78515625,
      "learning_rate": 9.575577557755776e-05,
      "loss": 3.1033,
      "step": 521270
    },
    {
      "epoch": 0.2811027147930933,
      "grad_norm": 0.78125,
      "learning_rate": 9.575377537753776e-05,
      "loss": 3.2083,
      "step": 521280
    },
    {
      "epoch": 0.2811213580292515,
      "grad_norm": 0.60546875,
      "learning_rate": 9.575177517751776e-05,
      "loss": 3.236,
      "step": 521290
    },
    {
      "epoch": 0.28114000126540967,
      "grad_norm": 0.7109375,
      "learning_rate": 9.574977497749775e-05,
      "loss": 3.1809,
      "step": 521300
    },
    {
      "epoch": 0.28115864450156786,
      "grad_norm": 0.8046875,
      "learning_rate": 9.574777477747775e-05,
      "loss": 3.0623,
      "step": 521310
    },
    {
      "epoch": 0.28117728773772604,
      "grad_norm": 0.703125,
      "learning_rate": 9.574577457745774e-05,
      "loss": 3.2245,
      "step": 521320
    },
    {
      "epoch": 0.2811959309738842,
      "grad_norm": 0.8125,
      "learning_rate": 9.574377437743775e-05,
      "loss": 3.368,
      "step": 521330
    },
    {
      "epoch": 0.28121457421004237,
      "grad_norm": 0.79296875,
      "learning_rate": 9.574177417741775e-05,
      "loss": 3.1847,
      "step": 521340
    },
    {
      "epoch": 0.28123321744620056,
      "grad_norm": 0.73046875,
      "learning_rate": 9.573977397739774e-05,
      "loss": 3.1252,
      "step": 521350
    },
    {
      "epoch": 0.28125186068235875,
      "grad_norm": 0.8671875,
      "learning_rate": 9.573777377737774e-05,
      "loss": 3.2084,
      "step": 521360
    },
    {
      "epoch": 0.28127050391851693,
      "grad_norm": 0.7578125,
      "learning_rate": 9.573577357735775e-05,
      "loss": 3.2233,
      "step": 521370
    },
    {
      "epoch": 0.2812891471546751,
      "grad_norm": 0.71484375,
      "learning_rate": 9.573377337733774e-05,
      "loss": 3.2265,
      "step": 521380
    },
    {
      "epoch": 0.2813077903908333,
      "grad_norm": 0.80078125,
      "learning_rate": 9.573177317731774e-05,
      "loss": 3.2667,
      "step": 521390
    },
    {
      "epoch": 0.2813264336269915,
      "grad_norm": 0.79296875,
      "learning_rate": 9.572977297729773e-05,
      "loss": 3.3802,
      "step": 521400
    },
    {
      "epoch": 0.2813450768631497,
      "grad_norm": 0.80078125,
      "learning_rate": 9.572777277727773e-05,
      "loss": 3.243,
      "step": 521410
    },
    {
      "epoch": 0.2813637200993079,
      "grad_norm": 0.7890625,
      "learning_rate": 9.572577257725774e-05,
      "loss": 3.2214,
      "step": 521420
    },
    {
      "epoch": 0.28138236333546607,
      "grad_norm": 0.78125,
      "learning_rate": 9.572377237723773e-05,
      "loss": 3.1362,
      "step": 521430
    },
    {
      "epoch": 0.2814010065716242,
      "grad_norm": 0.87109375,
      "learning_rate": 9.572177217721773e-05,
      "loss": 3.2224,
      "step": 521440
    },
    {
      "epoch": 0.2814196498077824,
      "grad_norm": 0.78515625,
      "learning_rate": 9.571977197719773e-05,
      "loss": 3.2074,
      "step": 521450
    },
    {
      "epoch": 0.2814382930439406,
      "grad_norm": 0.7578125,
      "learning_rate": 9.571777177717772e-05,
      "loss": 3.2667,
      "step": 521460
    },
    {
      "epoch": 0.28145693628009877,
      "grad_norm": 0.73828125,
      "learning_rate": 9.571577157715772e-05,
      "loss": 3.2489,
      "step": 521470
    },
    {
      "epoch": 0.28147557951625696,
      "grad_norm": 0.77734375,
      "learning_rate": 9.571377137713771e-05,
      "loss": 3.2408,
      "step": 521480
    },
    {
      "epoch": 0.28149422275241515,
      "grad_norm": 0.78515625,
      "learning_rate": 9.571177117711772e-05,
      "loss": 3.2652,
      "step": 521490
    },
    {
      "epoch": 0.28151286598857334,
      "grad_norm": 0.77734375,
      "learning_rate": 9.570977097709772e-05,
      "loss": 3.1793,
      "step": 521500
    },
    {
      "epoch": 0.2815315092247315,
      "grad_norm": 0.75,
      "learning_rate": 9.570777077707771e-05,
      "loss": 3.2336,
      "step": 521510
    },
    {
      "epoch": 0.2815501524608897,
      "grad_norm": 0.83203125,
      "learning_rate": 9.570577057705771e-05,
      "loss": 3.273,
      "step": 521520
    },
    {
      "epoch": 0.2815687956970479,
      "grad_norm": 0.76171875,
      "learning_rate": 9.57037703770377e-05,
      "loss": 3.2103,
      "step": 521530
    },
    {
      "epoch": 0.28158743893320604,
      "grad_norm": 0.69140625,
      "learning_rate": 9.57017701770177e-05,
      "loss": 3.1214,
      "step": 521540
    },
    {
      "epoch": 0.2816060821693642,
      "grad_norm": 0.796875,
      "learning_rate": 9.56997699769977e-05,
      "loss": 3.2817,
      "step": 521550
    },
    {
      "epoch": 0.2816247254055224,
      "grad_norm": 0.73828125,
      "learning_rate": 9.569776977697769e-05,
      "loss": 3.2603,
      "step": 521560
    },
    {
      "epoch": 0.2816433686416806,
      "grad_norm": 0.703125,
      "learning_rate": 9.56957695769577e-05,
      "loss": 3.265,
      "step": 521570
    },
    {
      "epoch": 0.2816620118778388,
      "grad_norm": 0.859375,
      "learning_rate": 9.569376937693771e-05,
      "loss": 3.2305,
      "step": 521580
    },
    {
      "epoch": 0.281680655113997,
      "grad_norm": 0.734375,
      "learning_rate": 9.56917691769177e-05,
      "loss": 3.1084,
      "step": 521590
    },
    {
      "epoch": 0.28169929835015517,
      "grad_norm": 0.703125,
      "learning_rate": 9.56897689768977e-05,
      "loss": 3.1057,
      "step": 521600
    },
    {
      "epoch": 0.28171794158631336,
      "grad_norm": 0.7421875,
      "learning_rate": 9.56877687768777e-05,
      "loss": 3.1509,
      "step": 521610
    },
    {
      "epoch": 0.28173658482247155,
      "grad_norm": 0.71875,
      "learning_rate": 9.568576857685769e-05,
      "loss": 3.1228,
      "step": 521620
    },
    {
      "epoch": 0.28175522805862974,
      "grad_norm": 0.78125,
      "learning_rate": 9.568376837683769e-05,
      "loss": 3.2095,
      "step": 521630
    },
    {
      "epoch": 0.2817738712947879,
      "grad_norm": 0.734375,
      "learning_rate": 9.568176817681768e-05,
      "loss": 3.1685,
      "step": 521640
    },
    {
      "epoch": 0.28179251453094606,
      "grad_norm": 0.7421875,
      "learning_rate": 9.567976797679769e-05,
      "loss": 3.1111,
      "step": 521650
    },
    {
      "epoch": 0.28181115776710425,
      "grad_norm": 0.7734375,
      "learning_rate": 9.567776777677769e-05,
      "loss": 3.2074,
      "step": 521660
    },
    {
      "epoch": 0.28182980100326244,
      "grad_norm": 0.74609375,
      "learning_rate": 9.567576757675768e-05,
      "loss": 3.1117,
      "step": 521670
    },
    {
      "epoch": 0.2818484442394206,
      "grad_norm": 0.66796875,
      "learning_rate": 9.567376737673768e-05,
      "loss": 3.2138,
      "step": 521680
    },
    {
      "epoch": 0.2818670874755788,
      "grad_norm": 0.76171875,
      "learning_rate": 9.567176717671767e-05,
      "loss": 3.1864,
      "step": 521690
    },
    {
      "epoch": 0.281885730711737,
      "grad_norm": 0.87890625,
      "learning_rate": 9.566976697669767e-05,
      "loss": 3.0432,
      "step": 521700
    },
    {
      "epoch": 0.2819043739478952,
      "grad_norm": 0.7890625,
      "learning_rate": 9.566776677667767e-05,
      "loss": 3.2721,
      "step": 521710
    },
    {
      "epoch": 0.2819230171840534,
      "grad_norm": 0.8125,
      "learning_rate": 9.566576657665766e-05,
      "loss": 3.2251,
      "step": 521720
    },
    {
      "epoch": 0.28194166042021157,
      "grad_norm": 0.69921875,
      "learning_rate": 9.566376637663767e-05,
      "loss": 3.2564,
      "step": 521730
    },
    {
      "epoch": 0.28196030365636976,
      "grad_norm": 0.6796875,
      "learning_rate": 9.566176617661767e-05,
      "loss": 3.2566,
      "step": 521740
    },
    {
      "epoch": 0.2819789468925279,
      "grad_norm": 0.7421875,
      "learning_rate": 9.565976597659766e-05,
      "loss": 3.2174,
      "step": 521750
    },
    {
      "epoch": 0.2819975901286861,
      "grad_norm": 0.734375,
      "learning_rate": 9.565776577657766e-05,
      "loss": 3.3801,
      "step": 521760
    },
    {
      "epoch": 0.28201623336484427,
      "grad_norm": 0.71875,
      "learning_rate": 9.565576557655767e-05,
      "loss": 3.1476,
      "step": 521770
    },
    {
      "epoch": 0.28203487660100246,
      "grad_norm": 0.6484375,
      "learning_rate": 9.565376537653766e-05,
      "loss": 3.1611,
      "step": 521780
    },
    {
      "epoch": 0.28205351983716065,
      "grad_norm": 0.71875,
      "learning_rate": 9.565176517651766e-05,
      "loss": 3.1344,
      "step": 521790
    },
    {
      "epoch": 0.28207216307331884,
      "grad_norm": 0.69921875,
      "learning_rate": 9.564976497649765e-05,
      "loss": 3.2629,
      "step": 521800
    },
    {
      "epoch": 0.282090806309477,
      "grad_norm": 0.7109375,
      "learning_rate": 9.564776477647766e-05,
      "loss": 3.2306,
      "step": 521810
    },
    {
      "epoch": 0.2821094495456352,
      "grad_norm": 0.7421875,
      "learning_rate": 9.564576457645766e-05,
      "loss": 3.0517,
      "step": 521820
    },
    {
      "epoch": 0.2821280927817934,
      "grad_norm": 0.6171875,
      "learning_rate": 9.564376437643765e-05,
      "loss": 3.1635,
      "step": 521830
    },
    {
      "epoch": 0.2821467360179516,
      "grad_norm": 0.7421875,
      "learning_rate": 9.564176417641765e-05,
      "loss": 3.1729,
      "step": 521840
    },
    {
      "epoch": 0.2821653792541098,
      "grad_norm": 0.7421875,
      "learning_rate": 9.563976397639764e-05,
      "loss": 3.1894,
      "step": 521850
    },
    {
      "epoch": 0.2821840224902679,
      "grad_norm": 0.75,
      "learning_rate": 9.563776377637764e-05,
      "loss": 3.2063,
      "step": 521860
    },
    {
      "epoch": 0.2822026657264261,
      "grad_norm": 0.71875,
      "learning_rate": 9.563576357635764e-05,
      "loss": 3.3178,
      "step": 521870
    },
    {
      "epoch": 0.2822213089625843,
      "grad_norm": 0.828125,
      "learning_rate": 9.563376337633763e-05,
      "loss": 3.1665,
      "step": 521880
    },
    {
      "epoch": 0.2822399521987425,
      "grad_norm": 0.7734375,
      "learning_rate": 9.563176317631764e-05,
      "loss": 3.2685,
      "step": 521890
    },
    {
      "epoch": 0.28225859543490067,
      "grad_norm": 0.71484375,
      "learning_rate": 9.562976297629764e-05,
      "loss": 3.1711,
      "step": 521900
    },
    {
      "epoch": 0.28227723867105886,
      "grad_norm": 0.8671875,
      "learning_rate": 9.562776277627763e-05,
      "loss": 3.1509,
      "step": 521910
    },
    {
      "epoch": 0.28229588190721705,
      "grad_norm": 0.828125,
      "learning_rate": 9.562576257625763e-05,
      "loss": 3.1569,
      "step": 521920
    },
    {
      "epoch": 0.28231452514337524,
      "grad_norm": 0.84375,
      "learning_rate": 9.562376237623762e-05,
      "loss": 2.9994,
      "step": 521930
    },
    {
      "epoch": 0.28233316837953343,
      "grad_norm": 0.73828125,
      "learning_rate": 9.562176217621762e-05,
      "loss": 3.2569,
      "step": 521940
    },
    {
      "epoch": 0.2823518116156916,
      "grad_norm": 0.78125,
      "learning_rate": 9.561976197619761e-05,
      "loss": 3.2222,
      "step": 521950
    },
    {
      "epoch": 0.28237045485184975,
      "grad_norm": 0.859375,
      "learning_rate": 9.561776177617762e-05,
      "loss": 3.3146,
      "step": 521960
    },
    {
      "epoch": 0.28238909808800794,
      "grad_norm": 0.67578125,
      "learning_rate": 9.561576157615763e-05,
      "loss": 3.2812,
      "step": 521970
    },
    {
      "epoch": 0.28240774132416613,
      "grad_norm": 0.69921875,
      "learning_rate": 9.561376137613763e-05,
      "loss": 3.1505,
      "step": 521980
    },
    {
      "epoch": 0.2824263845603243,
      "grad_norm": 0.62890625,
      "learning_rate": 9.561176117611762e-05,
      "loss": 3.1723,
      "step": 521990
    },
    {
      "epoch": 0.2824450277964825,
      "grad_norm": 0.70703125,
      "learning_rate": 9.560976097609762e-05,
      "loss": 3.1539,
      "step": 522000
    },
    {
      "epoch": 0.2824636710326407,
      "grad_norm": 0.703125,
      "learning_rate": 9.560776077607761e-05,
      "loss": 3.2414,
      "step": 522010
    },
    {
      "epoch": 0.2824823142687989,
      "grad_norm": 0.7421875,
      "learning_rate": 9.560576057605761e-05,
      "loss": 3.1118,
      "step": 522020
    },
    {
      "epoch": 0.2825009575049571,
      "grad_norm": 0.765625,
      "learning_rate": 9.56037603760376e-05,
      "loss": 3.1952,
      "step": 522030
    },
    {
      "epoch": 0.28251960074111526,
      "grad_norm": 0.703125,
      "learning_rate": 9.56017601760176e-05,
      "loss": 3.2542,
      "step": 522040
    },
    {
      "epoch": 0.28253824397727345,
      "grad_norm": 0.70703125,
      "learning_rate": 9.559975997599761e-05,
      "loss": 3.1015,
      "step": 522050
    },
    {
      "epoch": 0.28255688721343164,
      "grad_norm": 0.71484375,
      "learning_rate": 9.55977597759776e-05,
      "loss": 3.2739,
      "step": 522060
    },
    {
      "epoch": 0.2825755304495898,
      "grad_norm": 0.66015625,
      "learning_rate": 9.55957595759576e-05,
      "loss": 3.2443,
      "step": 522070
    },
    {
      "epoch": 0.28259417368574796,
      "grad_norm": 0.8125,
      "learning_rate": 9.55937593759376e-05,
      "loss": 3.1785,
      "step": 522080
    },
    {
      "epoch": 0.28261281692190615,
      "grad_norm": 0.71484375,
      "learning_rate": 9.559175917591759e-05,
      "loss": 3.2486,
      "step": 522090
    },
    {
      "epoch": 0.28263146015806434,
      "grad_norm": 0.7109375,
      "learning_rate": 9.558975897589759e-05,
      "loss": 3.3255,
      "step": 522100
    },
    {
      "epoch": 0.28265010339422253,
      "grad_norm": 0.67578125,
      "learning_rate": 9.558775877587758e-05,
      "loss": 3.2402,
      "step": 522110
    },
    {
      "epoch": 0.2826687466303807,
      "grad_norm": 0.73828125,
      "learning_rate": 9.558575857585759e-05,
      "loss": 3.3125,
      "step": 522120
    },
    {
      "epoch": 0.2826873898665389,
      "grad_norm": 0.6484375,
      "learning_rate": 9.558375837583759e-05,
      "loss": 3.2824,
      "step": 522130
    },
    {
      "epoch": 0.2827060331026971,
      "grad_norm": 0.703125,
      "learning_rate": 9.558175817581758e-05,
      "loss": 3.273,
      "step": 522140
    },
    {
      "epoch": 0.2827246763388553,
      "grad_norm": 0.765625,
      "learning_rate": 9.557975797579758e-05,
      "loss": 3.0878,
      "step": 522150
    },
    {
      "epoch": 0.2827433195750135,
      "grad_norm": 0.76953125,
      "learning_rate": 9.557775777577759e-05,
      "loss": 3.2122,
      "step": 522160
    },
    {
      "epoch": 0.2827619628111716,
      "grad_norm": 0.84375,
      "learning_rate": 9.557575757575758e-05,
      "loss": 3.2452,
      "step": 522170
    },
    {
      "epoch": 0.2827806060473298,
      "grad_norm": 0.82421875,
      "learning_rate": 9.557375737573758e-05,
      "loss": 3.2491,
      "step": 522180
    },
    {
      "epoch": 0.282799249283488,
      "grad_norm": 0.8515625,
      "learning_rate": 9.557175717571758e-05,
      "loss": 3.0183,
      "step": 522190
    },
    {
      "epoch": 0.2828178925196462,
      "grad_norm": 0.70703125,
      "learning_rate": 9.556975697569757e-05,
      "loss": 3.0008,
      "step": 522200
    },
    {
      "epoch": 0.28283653575580436,
      "grad_norm": 0.84765625,
      "learning_rate": 9.556775677567758e-05,
      "loss": 3.2424,
      "step": 522210
    },
    {
      "epoch": 0.28285517899196255,
      "grad_norm": 0.72265625,
      "learning_rate": 9.556575657565758e-05,
      "loss": 3.2244,
      "step": 522220
    },
    {
      "epoch": 0.28287382222812074,
      "grad_norm": 0.77734375,
      "learning_rate": 9.556375637563757e-05,
      "loss": 3.2497,
      "step": 522230
    },
    {
      "epoch": 0.28289246546427893,
      "grad_norm": 0.7578125,
      "learning_rate": 9.556175617561757e-05,
      "loss": 3.1344,
      "step": 522240
    },
    {
      "epoch": 0.2829111087004371,
      "grad_norm": 0.76953125,
      "learning_rate": 9.555975597559756e-05,
      "loss": 3.106,
      "step": 522250
    },
    {
      "epoch": 0.2829297519365953,
      "grad_norm": 0.703125,
      "learning_rate": 9.555775577557756e-05,
      "loss": 3.2276,
      "step": 522260
    },
    {
      "epoch": 0.2829483951727535,
      "grad_norm": 0.890625,
      "learning_rate": 9.555575557555755e-05,
      "loss": 3.3959,
      "step": 522270
    },
    {
      "epoch": 0.28296703840891163,
      "grad_norm": 0.7109375,
      "learning_rate": 9.555375537553756e-05,
      "loss": 3.2509,
      "step": 522280
    },
    {
      "epoch": 0.2829856816450698,
      "grad_norm": 0.890625,
      "learning_rate": 9.555175517551756e-05,
      "loss": 3.2481,
      "step": 522290
    },
    {
      "epoch": 0.283004324881228,
      "grad_norm": 0.79296875,
      "learning_rate": 9.554975497549755e-05,
      "loss": 3.1733,
      "step": 522300
    },
    {
      "epoch": 0.2830229681173862,
      "grad_norm": 0.703125,
      "learning_rate": 9.554775477547755e-05,
      "loss": 3.2668,
      "step": 522310
    },
    {
      "epoch": 0.2830416113535444,
      "grad_norm": 0.70703125,
      "learning_rate": 9.554575457545755e-05,
      "loss": 3.2677,
      "step": 522320
    },
    {
      "epoch": 0.2830602545897026,
      "grad_norm": 0.73046875,
      "learning_rate": 9.554375437543754e-05,
      "loss": 3.2388,
      "step": 522330
    },
    {
      "epoch": 0.28307889782586076,
      "grad_norm": 0.70703125,
      "learning_rate": 9.554175417541754e-05,
      "loss": 3.2075,
      "step": 522340
    },
    {
      "epoch": 0.28309754106201895,
      "grad_norm": 0.70703125,
      "learning_rate": 9.553975397539755e-05,
      "loss": 3.217,
      "step": 522350
    },
    {
      "epoch": 0.28311618429817714,
      "grad_norm": 0.6796875,
      "learning_rate": 9.553775377537754e-05,
      "loss": 3.2332,
      "step": 522360
    },
    {
      "epoch": 0.28313482753433533,
      "grad_norm": 0.77734375,
      "learning_rate": 9.553575357535755e-05,
      "loss": 3.248,
      "step": 522370
    },
    {
      "epoch": 0.28315347077049347,
      "grad_norm": 0.7109375,
      "learning_rate": 9.553375337533755e-05,
      "loss": 3.1766,
      "step": 522380
    },
    {
      "epoch": 0.28317211400665165,
      "grad_norm": 0.8203125,
      "learning_rate": 9.553175317531754e-05,
      "loss": 3.1984,
      "step": 522390
    },
    {
      "epoch": 0.28319075724280984,
      "grad_norm": 0.70703125,
      "learning_rate": 9.552975297529754e-05,
      "loss": 3.335,
      "step": 522400
    },
    {
      "epoch": 0.28320940047896803,
      "grad_norm": 0.8125,
      "learning_rate": 9.552775277527753e-05,
      "loss": 3.1592,
      "step": 522410
    },
    {
      "epoch": 0.2832280437151262,
      "grad_norm": 0.73828125,
      "learning_rate": 9.552575257525753e-05,
      "loss": 3.2546,
      "step": 522420
    },
    {
      "epoch": 0.2832466869512844,
      "grad_norm": 0.76953125,
      "learning_rate": 9.552375237523752e-05,
      "loss": 3.2267,
      "step": 522430
    },
    {
      "epoch": 0.2832653301874426,
      "grad_norm": 0.765625,
      "learning_rate": 9.552175217521753e-05,
      "loss": 3.2264,
      "step": 522440
    },
    {
      "epoch": 0.2832839734236008,
      "grad_norm": 0.78125,
      "learning_rate": 9.551975197519753e-05,
      "loss": 3.3084,
      "step": 522450
    },
    {
      "epoch": 0.283302616659759,
      "grad_norm": 0.703125,
      "learning_rate": 9.551775177517752e-05,
      "loss": 3.3252,
      "step": 522460
    },
    {
      "epoch": 0.28332125989591717,
      "grad_norm": 0.69140625,
      "learning_rate": 9.551575157515752e-05,
      "loss": 3.1997,
      "step": 522470
    },
    {
      "epoch": 0.2833399031320753,
      "grad_norm": 0.75,
      "learning_rate": 9.551375137513752e-05,
      "loss": 3.2613,
      "step": 522480
    },
    {
      "epoch": 0.2833585463682335,
      "grad_norm": 0.86328125,
      "learning_rate": 9.551175117511751e-05,
      "loss": 3.1424,
      "step": 522490
    },
    {
      "epoch": 0.2833771896043917,
      "grad_norm": 0.7421875,
      "learning_rate": 9.55097509750975e-05,
      "loss": 3.2081,
      "step": 522500
    },
    {
      "epoch": 0.28339583284054987,
      "grad_norm": 0.6875,
      "learning_rate": 9.550775077507752e-05,
      "loss": 3.1917,
      "step": 522510
    },
    {
      "epoch": 0.28341447607670806,
      "grad_norm": 0.6875,
      "learning_rate": 9.550575057505751e-05,
      "loss": 3.1322,
      "step": 522520
    },
    {
      "epoch": 0.28343311931286624,
      "grad_norm": 0.63671875,
      "learning_rate": 9.550375037503751e-05,
      "loss": 3.3367,
      "step": 522530
    },
    {
      "epoch": 0.28345176254902443,
      "grad_norm": 0.76953125,
      "learning_rate": 9.55017501750175e-05,
      "loss": 3.207,
      "step": 522540
    },
    {
      "epoch": 0.2834704057851826,
      "grad_norm": 0.73046875,
      "learning_rate": 9.54997499749975e-05,
      "loss": 3.125,
      "step": 522550
    },
    {
      "epoch": 0.2834890490213408,
      "grad_norm": 0.703125,
      "learning_rate": 9.549774977497751e-05,
      "loss": 3.1319,
      "step": 522560
    },
    {
      "epoch": 0.283507692257499,
      "grad_norm": 0.76171875,
      "learning_rate": 9.54957495749575e-05,
      "loss": 3.1854,
      "step": 522570
    },
    {
      "epoch": 0.2835263354936572,
      "grad_norm": 0.75390625,
      "learning_rate": 9.54937493749375e-05,
      "loss": 3.1499,
      "step": 522580
    },
    {
      "epoch": 0.2835449787298153,
      "grad_norm": 0.69140625,
      "learning_rate": 9.54917491749175e-05,
      "loss": 3.2442,
      "step": 522590
    },
    {
      "epoch": 0.2835636219659735,
      "grad_norm": 0.6953125,
      "learning_rate": 9.54897489748975e-05,
      "loss": 3.1403,
      "step": 522600
    },
    {
      "epoch": 0.2835822652021317,
      "grad_norm": 0.76171875,
      "learning_rate": 9.54877487748775e-05,
      "loss": 3.2244,
      "step": 522610
    },
    {
      "epoch": 0.2836009084382899,
      "grad_norm": 0.84375,
      "learning_rate": 9.54857485748575e-05,
      "loss": 3.1105,
      "step": 522620
    },
    {
      "epoch": 0.2836195516744481,
      "grad_norm": 0.71875,
      "learning_rate": 9.548374837483749e-05,
      "loss": 3.2347,
      "step": 522630
    },
    {
      "epoch": 0.28363819491060627,
      "grad_norm": 0.75390625,
      "learning_rate": 9.548174817481749e-05,
      "loss": 3.2854,
      "step": 522640
    },
    {
      "epoch": 0.28365683814676446,
      "grad_norm": 0.65234375,
      "learning_rate": 9.547974797479748e-05,
      "loss": 3.1955,
      "step": 522650
    },
    {
      "epoch": 0.28367548138292265,
      "grad_norm": 0.6953125,
      "learning_rate": 9.547774777477748e-05,
      "loss": 3.2159,
      "step": 522660
    },
    {
      "epoch": 0.28369412461908083,
      "grad_norm": 0.7265625,
      "learning_rate": 9.547574757475749e-05,
      "loss": 3.2726,
      "step": 522670
    },
    {
      "epoch": 0.283712767855239,
      "grad_norm": 0.66796875,
      "learning_rate": 9.547374737473748e-05,
      "loss": 3.1473,
      "step": 522680
    },
    {
      "epoch": 0.28373141109139716,
      "grad_norm": 0.75390625,
      "learning_rate": 9.547174717471748e-05,
      "loss": 3.2186,
      "step": 522690
    },
    {
      "epoch": 0.28375005432755535,
      "grad_norm": 0.71875,
      "learning_rate": 9.546974697469747e-05,
      "loss": 3.2225,
      "step": 522700
    },
    {
      "epoch": 0.28376869756371353,
      "grad_norm": 0.71484375,
      "learning_rate": 9.546774677467747e-05,
      "loss": 3.1998,
      "step": 522710
    },
    {
      "epoch": 0.2837873407998717,
      "grad_norm": 0.6953125,
      "learning_rate": 9.546574657465746e-05,
      "loss": 3.0885,
      "step": 522720
    },
    {
      "epoch": 0.2838059840360299,
      "grad_norm": 0.75390625,
      "learning_rate": 9.546374637463746e-05,
      "loss": 3.1832,
      "step": 522730
    },
    {
      "epoch": 0.2838246272721881,
      "grad_norm": 0.69921875,
      "learning_rate": 9.546174617461746e-05,
      "loss": 2.9697,
      "step": 522740
    },
    {
      "epoch": 0.2838432705083463,
      "grad_norm": 0.69140625,
      "learning_rate": 9.545974597459746e-05,
      "loss": 3.0876,
      "step": 522750
    },
    {
      "epoch": 0.2838619137445045,
      "grad_norm": 0.70703125,
      "learning_rate": 9.545774577457747e-05,
      "loss": 3.3743,
      "step": 522760
    },
    {
      "epoch": 0.28388055698066267,
      "grad_norm": 0.71875,
      "learning_rate": 9.545574557455747e-05,
      "loss": 3.1746,
      "step": 522770
    },
    {
      "epoch": 0.28389920021682086,
      "grad_norm": 0.89453125,
      "learning_rate": 9.545374537453746e-05,
      "loss": 3.2339,
      "step": 522780
    },
    {
      "epoch": 0.28391784345297905,
      "grad_norm": 0.70703125,
      "learning_rate": 9.545174517451746e-05,
      "loss": 3.0726,
      "step": 522790
    },
    {
      "epoch": 0.2839364866891372,
      "grad_norm": 0.76171875,
      "learning_rate": 9.544974497449746e-05,
      "loss": 3.1699,
      "step": 522800
    },
    {
      "epoch": 0.28395512992529537,
      "grad_norm": 0.7109375,
      "learning_rate": 9.544774477447745e-05,
      "loss": 3.169,
      "step": 522810
    },
    {
      "epoch": 0.28397377316145356,
      "grad_norm": 0.8046875,
      "learning_rate": 9.544574457445745e-05,
      "loss": 3.3423,
      "step": 522820
    },
    {
      "epoch": 0.28399241639761175,
      "grad_norm": 0.859375,
      "learning_rate": 9.544374437443744e-05,
      "loss": 3.2876,
      "step": 522830
    },
    {
      "epoch": 0.28401105963376994,
      "grad_norm": 0.75,
      "learning_rate": 9.544174417441745e-05,
      "loss": 3.1304,
      "step": 522840
    },
    {
      "epoch": 0.2840297028699281,
      "grad_norm": 0.6875,
      "learning_rate": 9.543974397439745e-05,
      "loss": 3.1755,
      "step": 522850
    },
    {
      "epoch": 0.2840483461060863,
      "grad_norm": 0.828125,
      "learning_rate": 9.543774377437744e-05,
      "loss": 3.2262,
      "step": 522860
    },
    {
      "epoch": 0.2840669893422445,
      "grad_norm": 0.7421875,
      "learning_rate": 9.543574357435744e-05,
      "loss": 3.1652,
      "step": 522870
    },
    {
      "epoch": 0.2840856325784027,
      "grad_norm": 0.828125,
      "learning_rate": 9.543374337433743e-05,
      "loss": 3.1066,
      "step": 522880
    },
    {
      "epoch": 0.2841042758145609,
      "grad_norm": 0.6484375,
      "learning_rate": 9.543174317431743e-05,
      "loss": 3.3,
      "step": 522890
    },
    {
      "epoch": 0.284122919050719,
      "grad_norm": 0.67578125,
      "learning_rate": 9.542974297429743e-05,
      "loss": 3.1692,
      "step": 522900
    },
    {
      "epoch": 0.2841415622868772,
      "grad_norm": 0.76953125,
      "learning_rate": 9.542774277427743e-05,
      "loss": 3.1476,
      "step": 522910
    },
    {
      "epoch": 0.2841602055230354,
      "grad_norm": 0.69140625,
      "learning_rate": 9.542574257425743e-05,
      "loss": 3.2324,
      "step": 522920
    },
    {
      "epoch": 0.2841788487591936,
      "grad_norm": 0.62109375,
      "learning_rate": 9.542374237423743e-05,
      "loss": 3.2643,
      "step": 522930
    },
    {
      "epoch": 0.28419749199535177,
      "grad_norm": 0.76953125,
      "learning_rate": 9.542174217421742e-05,
      "loss": 3.1039,
      "step": 522940
    },
    {
      "epoch": 0.28421613523150996,
      "grad_norm": 0.6953125,
      "learning_rate": 9.541974197419743e-05,
      "loss": 3.274,
      "step": 522950
    },
    {
      "epoch": 0.28423477846766815,
      "grad_norm": 0.81640625,
      "learning_rate": 9.541774177417743e-05,
      "loss": 3.1655,
      "step": 522960
    },
    {
      "epoch": 0.28425342170382634,
      "grad_norm": 0.6953125,
      "learning_rate": 9.541574157415742e-05,
      "loss": 3.1482,
      "step": 522970
    },
    {
      "epoch": 0.2842720649399845,
      "grad_norm": 0.66796875,
      "learning_rate": 9.541374137413742e-05,
      "loss": 3.166,
      "step": 522980
    },
    {
      "epoch": 0.2842907081761427,
      "grad_norm": 0.65625,
      "learning_rate": 9.541174117411741e-05,
      "loss": 3.1223,
      "step": 522990
    },
    {
      "epoch": 0.2843093514123009,
      "grad_norm": 0.6796875,
      "learning_rate": 9.540974097409742e-05,
      "loss": 3.2621,
      "step": 523000
    },
    {
      "epoch": 0.28432799464845904,
      "grad_norm": 0.67578125,
      "learning_rate": 9.540774077407742e-05,
      "loss": 3.1549,
      "step": 523010
    },
    {
      "epoch": 0.2843466378846172,
      "grad_norm": 0.6953125,
      "learning_rate": 9.540574057405741e-05,
      "loss": 3.1752,
      "step": 523020
    },
    {
      "epoch": 0.2843652811207754,
      "grad_norm": 0.80859375,
      "learning_rate": 9.540374037403741e-05,
      "loss": 3.1717,
      "step": 523030
    },
    {
      "epoch": 0.2843839243569336,
      "grad_norm": 0.68359375,
      "learning_rate": 9.54017401740174e-05,
      "loss": 3.2105,
      "step": 523040
    },
    {
      "epoch": 0.2844025675930918,
      "grad_norm": 0.625,
      "learning_rate": 9.53997399739974e-05,
      "loss": 3.2906,
      "step": 523050
    },
    {
      "epoch": 0.28442121082925,
      "grad_norm": 0.73828125,
      "learning_rate": 9.53977397739774e-05,
      "loss": 3.2605,
      "step": 523060
    },
    {
      "epoch": 0.28443985406540817,
      "grad_norm": 0.7578125,
      "learning_rate": 9.53957395739574e-05,
      "loss": 3.2143,
      "step": 523070
    },
    {
      "epoch": 0.28445849730156636,
      "grad_norm": 0.80078125,
      "learning_rate": 9.53937393739374e-05,
      "loss": 3.19,
      "step": 523080
    },
    {
      "epoch": 0.28447714053772455,
      "grad_norm": 0.734375,
      "learning_rate": 9.53917391739174e-05,
      "loss": 3.2387,
      "step": 523090
    },
    {
      "epoch": 0.28449578377388274,
      "grad_norm": 0.7109375,
      "learning_rate": 9.538973897389739e-05,
      "loss": 3.2509,
      "step": 523100
    },
    {
      "epoch": 0.28451442701004087,
      "grad_norm": 0.75,
      "learning_rate": 9.538773877387739e-05,
      "loss": 3.197,
      "step": 523110
    },
    {
      "epoch": 0.28453307024619906,
      "grad_norm": 0.68359375,
      "learning_rate": 9.538573857385738e-05,
      "loss": 3.3166,
      "step": 523120
    },
    {
      "epoch": 0.28455171348235725,
      "grad_norm": 0.74609375,
      "learning_rate": 9.538373837383738e-05,
      "loss": 3.1814,
      "step": 523130
    },
    {
      "epoch": 0.28457035671851544,
      "grad_norm": 0.75390625,
      "learning_rate": 9.538173817381739e-05,
      "loss": 3.1249,
      "step": 523140
    },
    {
      "epoch": 0.2845889999546736,
      "grad_norm": 0.70703125,
      "learning_rate": 9.537973797379738e-05,
      "loss": 3.2248,
      "step": 523150
    },
    {
      "epoch": 0.2846076431908318,
      "grad_norm": 0.8671875,
      "learning_rate": 9.537773777377739e-05,
      "loss": 3.1962,
      "step": 523160
    },
    {
      "epoch": 0.28462628642699,
      "grad_norm": 0.65234375,
      "learning_rate": 9.537573757375739e-05,
      "loss": 3.0816,
      "step": 523170
    },
    {
      "epoch": 0.2846449296631482,
      "grad_norm": 0.80078125,
      "learning_rate": 9.537373737373738e-05,
      "loss": 3.1805,
      "step": 523180
    },
    {
      "epoch": 0.2846635728993064,
      "grad_norm": 0.7734375,
      "learning_rate": 9.537173717371738e-05,
      "loss": 3.1218,
      "step": 523190
    },
    {
      "epoch": 0.28468221613546457,
      "grad_norm": 0.66796875,
      "learning_rate": 9.536973697369737e-05,
      "loss": 3.2569,
      "step": 523200
    },
    {
      "epoch": 0.28470085937162276,
      "grad_norm": 0.67578125,
      "learning_rate": 9.536773677367737e-05,
      "loss": 3.2382,
      "step": 523210
    },
    {
      "epoch": 0.2847195026077809,
      "grad_norm": 0.73828125,
      "learning_rate": 9.536573657365737e-05,
      "loss": 3.2045,
      "step": 523220
    },
    {
      "epoch": 0.2847381458439391,
      "grad_norm": 0.78125,
      "learning_rate": 9.536373637363737e-05,
      "loss": 3.114,
      "step": 523230
    },
    {
      "epoch": 0.2847567890800973,
      "grad_norm": 0.734375,
      "learning_rate": 9.536173617361737e-05,
      "loss": 3.3058,
      "step": 523240
    },
    {
      "epoch": 0.28477543231625546,
      "grad_norm": 0.7578125,
      "learning_rate": 9.535973597359737e-05,
      "loss": 3.2019,
      "step": 523250
    },
    {
      "epoch": 0.28479407555241365,
      "grad_norm": 0.75390625,
      "learning_rate": 9.535773577357736e-05,
      "loss": 3.0401,
      "step": 523260
    },
    {
      "epoch": 0.28481271878857184,
      "grad_norm": 0.75,
      "learning_rate": 9.535573557355736e-05,
      "loss": 3.1056,
      "step": 523270
    },
    {
      "epoch": 0.28483136202473003,
      "grad_norm": 0.8203125,
      "learning_rate": 9.535373537353735e-05,
      "loss": 3.2611,
      "step": 523280
    },
    {
      "epoch": 0.2848500052608882,
      "grad_norm": 0.8203125,
      "learning_rate": 9.535173517351735e-05,
      "loss": 3.1987,
      "step": 523290
    },
    {
      "epoch": 0.2848686484970464,
      "grad_norm": 0.69140625,
      "learning_rate": 9.534973497349736e-05,
      "loss": 3.2512,
      "step": 523300
    },
    {
      "epoch": 0.2848872917332046,
      "grad_norm": 0.63671875,
      "learning_rate": 9.534773477347735e-05,
      "loss": 3.2362,
      "step": 523310
    },
    {
      "epoch": 0.28490593496936273,
      "grad_norm": 0.80859375,
      "learning_rate": 9.534573457345735e-05,
      "loss": 3.18,
      "step": 523320
    },
    {
      "epoch": 0.2849245782055209,
      "grad_norm": 0.69921875,
      "learning_rate": 9.534373437343734e-05,
      "loss": 3.1122,
      "step": 523330
    },
    {
      "epoch": 0.2849432214416791,
      "grad_norm": 0.84765625,
      "learning_rate": 9.534173417341734e-05,
      "loss": 3.1206,
      "step": 523340
    },
    {
      "epoch": 0.2849618646778373,
      "grad_norm": 0.7421875,
      "learning_rate": 9.533973397339735e-05,
      "loss": 3.1766,
      "step": 523350
    },
    {
      "epoch": 0.2849805079139955,
      "grad_norm": 0.64453125,
      "learning_rate": 9.533773377337734e-05,
      "loss": 2.9987,
      "step": 523360
    },
    {
      "epoch": 0.2849991511501537,
      "grad_norm": 0.7265625,
      "learning_rate": 9.533573357335734e-05,
      "loss": 3.305,
      "step": 523370
    },
    {
      "epoch": 0.28501779438631186,
      "grad_norm": 0.75390625,
      "learning_rate": 9.533373337333734e-05,
      "loss": 3.1707,
      "step": 523380
    },
    {
      "epoch": 0.28503643762247005,
      "grad_norm": 0.7421875,
      "learning_rate": 9.533173317331734e-05,
      "loss": 3.1781,
      "step": 523390
    },
    {
      "epoch": 0.28505508085862824,
      "grad_norm": 0.84375,
      "learning_rate": 9.532973297329734e-05,
      "loss": 3.1526,
      "step": 523400
    },
    {
      "epoch": 0.28507372409478643,
      "grad_norm": 0.7109375,
      "learning_rate": 9.532773277327734e-05,
      "loss": 3.2,
      "step": 523410
    },
    {
      "epoch": 0.2850923673309446,
      "grad_norm": 0.80078125,
      "learning_rate": 9.532573257325733e-05,
      "loss": 3.211,
      "step": 523420
    },
    {
      "epoch": 0.28511101056710275,
      "grad_norm": 0.703125,
      "learning_rate": 9.532373237323733e-05,
      "loss": 3.1522,
      "step": 523430
    },
    {
      "epoch": 0.28512965380326094,
      "grad_norm": 0.7890625,
      "learning_rate": 9.532173217321732e-05,
      "loss": 3.167,
      "step": 523440
    },
    {
      "epoch": 0.28514829703941913,
      "grad_norm": 0.63671875,
      "learning_rate": 9.531973197319732e-05,
      "loss": 3.3558,
      "step": 523450
    },
    {
      "epoch": 0.2851669402755773,
      "grad_norm": 0.6953125,
      "learning_rate": 9.531773177317733e-05,
      "loss": 3.1843,
      "step": 523460
    },
    {
      "epoch": 0.2851855835117355,
      "grad_norm": 0.9453125,
      "learning_rate": 9.531573157315732e-05,
      "loss": 3.1609,
      "step": 523470
    },
    {
      "epoch": 0.2852042267478937,
      "grad_norm": 0.671875,
      "learning_rate": 9.531373137313732e-05,
      "loss": 3.1081,
      "step": 523480
    },
    {
      "epoch": 0.2852228699840519,
      "grad_norm": 0.71875,
      "learning_rate": 9.531173117311731e-05,
      "loss": 3.2186,
      "step": 523490
    },
    {
      "epoch": 0.2852415132202101,
      "grad_norm": 0.78125,
      "learning_rate": 9.530973097309731e-05,
      "loss": 3.0906,
      "step": 523500
    },
    {
      "epoch": 0.28526015645636826,
      "grad_norm": 0.65625,
      "learning_rate": 9.53077307730773e-05,
      "loss": 3.1053,
      "step": 523510
    },
    {
      "epoch": 0.28527879969252645,
      "grad_norm": 0.78125,
      "learning_rate": 9.53057305730573e-05,
      "loss": 3.1435,
      "step": 523520
    },
    {
      "epoch": 0.2852974429286846,
      "grad_norm": 0.6640625,
      "learning_rate": 9.53037303730373e-05,
      "loss": 3.2875,
      "step": 523530
    },
    {
      "epoch": 0.2853160861648428,
      "grad_norm": 0.796875,
      "learning_rate": 9.53017301730173e-05,
      "loss": 3.1733,
      "step": 523540
    },
    {
      "epoch": 0.28533472940100096,
      "grad_norm": 0.76953125,
      "learning_rate": 9.529972997299732e-05,
      "loss": 3.1596,
      "step": 523550
    },
    {
      "epoch": 0.28535337263715915,
      "grad_norm": 0.75390625,
      "learning_rate": 9.529772977297731e-05,
      "loss": 3.2822,
      "step": 523560
    },
    {
      "epoch": 0.28537201587331734,
      "grad_norm": 0.78125,
      "learning_rate": 9.52957295729573e-05,
      "loss": 3.2653,
      "step": 523570
    },
    {
      "epoch": 0.28539065910947553,
      "grad_norm": 0.78515625,
      "learning_rate": 9.52937293729373e-05,
      "loss": 3.1977,
      "step": 523580
    },
    {
      "epoch": 0.2854093023456337,
      "grad_norm": 0.74609375,
      "learning_rate": 9.52917291729173e-05,
      "loss": 3.1605,
      "step": 523590
    },
    {
      "epoch": 0.2854279455817919,
      "grad_norm": 0.79296875,
      "learning_rate": 9.528972897289729e-05,
      "loss": 3.2902,
      "step": 523600
    },
    {
      "epoch": 0.2854465888179501,
      "grad_norm": 0.7109375,
      "learning_rate": 9.528772877287729e-05,
      "loss": 3.2498,
      "step": 523610
    },
    {
      "epoch": 0.2854652320541083,
      "grad_norm": 0.7421875,
      "learning_rate": 9.52857285728573e-05,
      "loss": 3.1997,
      "step": 523620
    },
    {
      "epoch": 0.2854838752902665,
      "grad_norm": 0.6953125,
      "learning_rate": 9.52837283728373e-05,
      "loss": 3.2935,
      "step": 523630
    },
    {
      "epoch": 0.2855025185264246,
      "grad_norm": 0.7109375,
      "learning_rate": 9.528172817281729e-05,
      "loss": 3.1883,
      "step": 523640
    },
    {
      "epoch": 0.2855211617625828,
      "grad_norm": 0.71484375,
      "learning_rate": 9.527972797279728e-05,
      "loss": 3.1846,
      "step": 523650
    },
    {
      "epoch": 0.285539804998741,
      "grad_norm": 0.8203125,
      "learning_rate": 9.527772777277728e-05,
      "loss": 3.1532,
      "step": 523660
    },
    {
      "epoch": 0.2855584482348992,
      "grad_norm": 0.8828125,
      "learning_rate": 9.527572757275728e-05,
      "loss": 3.2539,
      "step": 523670
    },
    {
      "epoch": 0.28557709147105736,
      "grad_norm": 0.69140625,
      "learning_rate": 9.527372737273727e-05,
      "loss": 3.204,
      "step": 523680
    },
    {
      "epoch": 0.28559573470721555,
      "grad_norm": 0.6015625,
      "learning_rate": 9.527172717271727e-05,
      "loss": 3.1098,
      "step": 523690
    },
    {
      "epoch": 0.28561437794337374,
      "grad_norm": 0.75,
      "learning_rate": 9.526972697269728e-05,
      "loss": 3.2466,
      "step": 523700
    },
    {
      "epoch": 0.28563302117953193,
      "grad_norm": 0.65234375,
      "learning_rate": 9.526772677267727e-05,
      "loss": 3.1989,
      "step": 523710
    },
    {
      "epoch": 0.2856516644156901,
      "grad_norm": 0.7421875,
      "learning_rate": 9.526572657265727e-05,
      "loss": 3.2205,
      "step": 523720
    },
    {
      "epoch": 0.2856703076518483,
      "grad_norm": 0.765625,
      "learning_rate": 9.526372637263726e-05,
      "loss": 3.2362,
      "step": 523730
    },
    {
      "epoch": 0.28568895088800644,
      "grad_norm": 0.6171875,
      "learning_rate": 9.526172617261727e-05,
      "loss": 3.1178,
      "step": 523740
    },
    {
      "epoch": 0.28570759412416463,
      "grad_norm": 0.71484375,
      "learning_rate": 9.525972597259727e-05,
      "loss": 3.12,
      "step": 523750
    },
    {
      "epoch": 0.2857262373603228,
      "grad_norm": 0.69921875,
      "learning_rate": 9.525772577257726e-05,
      "loss": 3.2061,
      "step": 523760
    },
    {
      "epoch": 0.285744880596481,
      "grad_norm": 0.78515625,
      "learning_rate": 9.525572557255726e-05,
      "loss": 3.1742,
      "step": 523770
    },
    {
      "epoch": 0.2857635238326392,
      "grad_norm": 0.75,
      "learning_rate": 9.525372537253727e-05,
      "loss": 3.1632,
      "step": 523780
    },
    {
      "epoch": 0.2857821670687974,
      "grad_norm": 0.77734375,
      "learning_rate": 9.525172517251726e-05,
      "loss": 3.211,
      "step": 523790
    },
    {
      "epoch": 0.2858008103049556,
      "grad_norm": 0.66796875,
      "learning_rate": 9.524972497249726e-05,
      "loss": 3.3965,
      "step": 523800
    },
    {
      "epoch": 0.28581945354111377,
      "grad_norm": 0.73828125,
      "learning_rate": 9.524772477247725e-05,
      "loss": 3.3206,
      "step": 523810
    },
    {
      "epoch": 0.28583809677727195,
      "grad_norm": 0.7890625,
      "learning_rate": 9.524572457245725e-05,
      "loss": 3.1922,
      "step": 523820
    },
    {
      "epoch": 0.28585674001343014,
      "grad_norm": 0.85546875,
      "learning_rate": 9.524372437243725e-05,
      "loss": 3.2929,
      "step": 523830
    },
    {
      "epoch": 0.2858753832495883,
      "grad_norm": 0.6796875,
      "learning_rate": 9.524172417241724e-05,
      "loss": 3.2743,
      "step": 523840
    },
    {
      "epoch": 0.28589402648574647,
      "grad_norm": 0.74609375,
      "learning_rate": 9.523972397239724e-05,
      "loss": 3.1756,
      "step": 523850
    },
    {
      "epoch": 0.28591266972190466,
      "grad_norm": 0.76171875,
      "learning_rate": 9.523772377237725e-05,
      "loss": 3.2775,
      "step": 523860
    },
    {
      "epoch": 0.28593131295806284,
      "grad_norm": 0.79296875,
      "learning_rate": 9.523572357235724e-05,
      "loss": 3.2571,
      "step": 523870
    },
    {
      "epoch": 0.28594995619422103,
      "grad_norm": 0.7109375,
      "learning_rate": 9.523372337233724e-05,
      "loss": 3.2703,
      "step": 523880
    },
    {
      "epoch": 0.2859685994303792,
      "grad_norm": 0.76171875,
      "learning_rate": 9.523172317231723e-05,
      "loss": 3.1991,
      "step": 523890
    },
    {
      "epoch": 0.2859872426665374,
      "grad_norm": 0.78515625,
      "learning_rate": 9.522972297229723e-05,
      "loss": 3.1058,
      "step": 523900
    },
    {
      "epoch": 0.2860058859026956,
      "grad_norm": 0.84375,
      "learning_rate": 9.522772277227722e-05,
      "loss": 3.0968,
      "step": 523910
    },
    {
      "epoch": 0.2860245291388538,
      "grad_norm": 0.73046875,
      "learning_rate": 9.522572257225722e-05,
      "loss": 3.2529,
      "step": 523920
    },
    {
      "epoch": 0.286043172375012,
      "grad_norm": 0.625,
      "learning_rate": 9.522372237223723e-05,
      "loss": 3.3054,
      "step": 523930
    },
    {
      "epoch": 0.28606181561117017,
      "grad_norm": 0.83984375,
      "learning_rate": 9.522172217221724e-05,
      "loss": 3.1855,
      "step": 523940
    },
    {
      "epoch": 0.2860804588473283,
      "grad_norm": 0.68359375,
      "learning_rate": 9.521972197219723e-05,
      "loss": 3.3249,
      "step": 523950
    },
    {
      "epoch": 0.2860991020834865,
      "grad_norm": 0.7578125,
      "learning_rate": 9.521772177217723e-05,
      "loss": 3.1872,
      "step": 523960
    },
    {
      "epoch": 0.2861177453196447,
      "grad_norm": 0.73046875,
      "learning_rate": 9.521572157215722e-05,
      "loss": 3.1919,
      "step": 523970
    },
    {
      "epoch": 0.28613638855580287,
      "grad_norm": 0.8125,
      "learning_rate": 9.521372137213722e-05,
      "loss": 3.1726,
      "step": 523980
    },
    {
      "epoch": 0.28615503179196106,
      "grad_norm": 0.7890625,
      "learning_rate": 9.521172117211722e-05,
      "loss": 3.28,
      "step": 523990
    },
    {
      "epoch": 0.28617367502811925,
      "grad_norm": 0.78515625,
      "learning_rate": 9.520972097209721e-05,
      "loss": 3.2284,
      "step": 524000
    },
    {
      "epoch": 0.28619231826427743,
      "grad_norm": 0.65625,
      "learning_rate": 9.520772077207721e-05,
      "loss": 3.2382,
      "step": 524010
    },
    {
      "epoch": 0.2862109615004356,
      "grad_norm": 0.75390625,
      "learning_rate": 9.520572057205722e-05,
      "loss": 3.2194,
      "step": 524020
    },
    {
      "epoch": 0.2862296047365938,
      "grad_norm": 0.6875,
      "learning_rate": 9.520372037203721e-05,
      "loss": 3.2177,
      "step": 524030
    },
    {
      "epoch": 0.286248247972752,
      "grad_norm": 0.71875,
      "learning_rate": 9.520172017201721e-05,
      "loss": 3.2675,
      "step": 524040
    },
    {
      "epoch": 0.28626689120891013,
      "grad_norm": 0.75390625,
      "learning_rate": 9.51997199719972e-05,
      "loss": 3.2712,
      "step": 524050
    },
    {
      "epoch": 0.2862855344450683,
      "grad_norm": 0.7109375,
      "learning_rate": 9.51977197719772e-05,
      "loss": 3.1359,
      "step": 524060
    },
    {
      "epoch": 0.2863041776812265,
      "grad_norm": 0.77734375,
      "learning_rate": 9.51957195719572e-05,
      "loss": 3.2445,
      "step": 524070
    },
    {
      "epoch": 0.2863228209173847,
      "grad_norm": 0.75390625,
      "learning_rate": 9.519371937193719e-05,
      "loss": 3.2289,
      "step": 524080
    },
    {
      "epoch": 0.2863414641535429,
      "grad_norm": 0.7734375,
      "learning_rate": 9.51917191719172e-05,
      "loss": 3.2788,
      "step": 524090
    },
    {
      "epoch": 0.2863601073897011,
      "grad_norm": 0.84765625,
      "learning_rate": 9.51897189718972e-05,
      "loss": 3.1063,
      "step": 524100
    },
    {
      "epoch": 0.28637875062585927,
      "grad_norm": 0.69921875,
      "learning_rate": 9.518771877187719e-05,
      "loss": 3.22,
      "step": 524110
    },
    {
      "epoch": 0.28639739386201746,
      "grad_norm": 0.83203125,
      "learning_rate": 9.518571857185719e-05,
      "loss": 3.0915,
      "step": 524120
    },
    {
      "epoch": 0.28641603709817565,
      "grad_norm": 0.70703125,
      "learning_rate": 9.518371837183718e-05,
      "loss": 3.2642,
      "step": 524130
    },
    {
      "epoch": 0.28643468033433384,
      "grad_norm": 0.92578125,
      "learning_rate": 9.518171817181719e-05,
      "loss": 3.1925,
      "step": 524140
    },
    {
      "epoch": 0.286453323570492,
      "grad_norm": 0.78125,
      "learning_rate": 9.517971797179719e-05,
      "loss": 3.2031,
      "step": 524150
    },
    {
      "epoch": 0.28647196680665016,
      "grad_norm": 0.80859375,
      "learning_rate": 9.517771777177718e-05,
      "loss": 3.2695,
      "step": 524160
    },
    {
      "epoch": 0.28649061004280835,
      "grad_norm": 0.71484375,
      "learning_rate": 9.517571757175718e-05,
      "loss": 3.1454,
      "step": 524170
    },
    {
      "epoch": 0.28650925327896654,
      "grad_norm": 0.7421875,
      "learning_rate": 9.517371737173719e-05,
      "loss": 3.1779,
      "step": 524180
    },
    {
      "epoch": 0.2865278965151247,
      "grad_norm": 0.703125,
      "learning_rate": 9.517171717171718e-05,
      "loss": 3.2524,
      "step": 524190
    },
    {
      "epoch": 0.2865465397512829,
      "grad_norm": 0.78125,
      "learning_rate": 9.516971697169718e-05,
      "loss": 3.2589,
      "step": 524200
    },
    {
      "epoch": 0.2865651829874411,
      "grad_norm": 0.73828125,
      "learning_rate": 9.516771677167717e-05,
      "loss": 3.2192,
      "step": 524210
    },
    {
      "epoch": 0.2865838262235993,
      "grad_norm": 0.72265625,
      "learning_rate": 9.516571657165717e-05,
      "loss": 3.1981,
      "step": 524220
    },
    {
      "epoch": 0.2866024694597575,
      "grad_norm": 0.80859375,
      "learning_rate": 9.516371637163716e-05,
      "loss": 3.2217,
      "step": 524230
    },
    {
      "epoch": 0.28662111269591567,
      "grad_norm": 0.75,
      "learning_rate": 9.516171617161716e-05,
      "loss": 3.1207,
      "step": 524240
    },
    {
      "epoch": 0.28663975593207386,
      "grad_norm": 0.765625,
      "learning_rate": 9.515971597159717e-05,
      "loss": 3.2244,
      "step": 524250
    },
    {
      "epoch": 0.286658399168232,
      "grad_norm": 0.7890625,
      "learning_rate": 9.515771577157716e-05,
      "loss": 3.1804,
      "step": 524260
    },
    {
      "epoch": 0.2866770424043902,
      "grad_norm": 0.67578125,
      "learning_rate": 9.515571557155716e-05,
      "loss": 3.1442,
      "step": 524270
    },
    {
      "epoch": 0.28669568564054837,
      "grad_norm": 0.76171875,
      "learning_rate": 9.515371537153716e-05,
      "loss": 3.126,
      "step": 524280
    },
    {
      "epoch": 0.28671432887670656,
      "grad_norm": 0.71484375,
      "learning_rate": 9.515171517151715e-05,
      "loss": 3.2282,
      "step": 524290
    },
    {
      "epoch": 0.28673297211286475,
      "grad_norm": 0.8203125,
      "learning_rate": 9.514971497149715e-05,
      "loss": 3.1383,
      "step": 524300
    },
    {
      "epoch": 0.28675161534902294,
      "grad_norm": 0.73828125,
      "learning_rate": 9.514771477147714e-05,
      "loss": 3.267,
      "step": 524310
    },
    {
      "epoch": 0.2867702585851811,
      "grad_norm": 0.83203125,
      "learning_rate": 9.514571457145714e-05,
      "loss": 3.1795,
      "step": 524320
    },
    {
      "epoch": 0.2867889018213393,
      "grad_norm": 0.7734375,
      "learning_rate": 9.514371437143715e-05,
      "loss": 3.2381,
      "step": 524330
    },
    {
      "epoch": 0.2868075450574975,
      "grad_norm": 0.75390625,
      "learning_rate": 9.514171417141716e-05,
      "loss": 3.3121,
      "step": 524340
    },
    {
      "epoch": 0.2868261882936557,
      "grad_norm": 0.72265625,
      "learning_rate": 9.513971397139715e-05,
      "loss": 3.3189,
      "step": 524350
    },
    {
      "epoch": 0.2868448315298139,
      "grad_norm": 0.6796875,
      "learning_rate": 9.513771377137715e-05,
      "loss": 3.2056,
      "step": 524360
    },
    {
      "epoch": 0.286863474765972,
      "grad_norm": 0.7578125,
      "learning_rate": 9.513571357135714e-05,
      "loss": 3.2779,
      "step": 524370
    },
    {
      "epoch": 0.2868821180021302,
      "grad_norm": 0.828125,
      "learning_rate": 9.513371337133714e-05,
      "loss": 3.1959,
      "step": 524380
    },
    {
      "epoch": 0.2869007612382884,
      "grad_norm": 0.70703125,
      "learning_rate": 9.513171317131713e-05,
      "loss": 3.2328,
      "step": 524390
    },
    {
      "epoch": 0.2869194044744466,
      "grad_norm": 0.63671875,
      "learning_rate": 9.512971297129713e-05,
      "loss": 3.075,
      "step": 524400
    },
    {
      "epoch": 0.28693804771060477,
      "grad_norm": 0.765625,
      "learning_rate": 9.512771277127714e-05,
      "loss": 3.1265,
      "step": 524410
    },
    {
      "epoch": 0.28695669094676296,
      "grad_norm": 0.66015625,
      "learning_rate": 9.512571257125714e-05,
      "loss": 3.2515,
      "step": 524420
    },
    {
      "epoch": 0.28697533418292115,
      "grad_norm": 0.72265625,
      "learning_rate": 9.512371237123713e-05,
      "loss": 3.2295,
      "step": 524430
    },
    {
      "epoch": 0.28699397741907934,
      "grad_norm": 0.7421875,
      "learning_rate": 9.512171217121713e-05,
      "loss": 3.3788,
      "step": 524440
    },
    {
      "epoch": 0.2870126206552375,
      "grad_norm": 0.75,
      "learning_rate": 9.511971197119712e-05,
      "loss": 3.2036,
      "step": 524450
    },
    {
      "epoch": 0.2870312638913957,
      "grad_norm": 0.62109375,
      "learning_rate": 9.511771177117712e-05,
      "loss": 3.1862,
      "step": 524460
    },
    {
      "epoch": 0.28704990712755385,
      "grad_norm": 0.73046875,
      "learning_rate": 9.511571157115711e-05,
      "loss": 3.3324,
      "step": 524470
    },
    {
      "epoch": 0.28706855036371204,
      "grad_norm": 0.60546875,
      "learning_rate": 9.511371137113711e-05,
      "loss": 3.1949,
      "step": 524480
    },
    {
      "epoch": 0.2870871935998702,
      "grad_norm": 0.8515625,
      "learning_rate": 9.511171117111712e-05,
      "loss": 3.1537,
      "step": 524490
    },
    {
      "epoch": 0.2871058368360284,
      "grad_norm": 0.65625,
      "learning_rate": 9.510971097109711e-05,
      "loss": 3.1904,
      "step": 524500
    },
    {
      "epoch": 0.2871244800721866,
      "grad_norm": 0.83984375,
      "learning_rate": 9.510771077107711e-05,
      "loss": 3.2027,
      "step": 524510
    },
    {
      "epoch": 0.2871431233083448,
      "grad_norm": 0.90234375,
      "learning_rate": 9.51057105710571e-05,
      "loss": 3.3184,
      "step": 524520
    },
    {
      "epoch": 0.287161766544503,
      "grad_norm": 0.7734375,
      "learning_rate": 9.510371037103711e-05,
      "loss": 3.2852,
      "step": 524530
    },
    {
      "epoch": 0.28718040978066117,
      "grad_norm": 0.796875,
      "learning_rate": 9.510171017101711e-05,
      "loss": 3.1793,
      "step": 524540
    },
    {
      "epoch": 0.28719905301681936,
      "grad_norm": 0.765625,
      "learning_rate": 9.50997099709971e-05,
      "loss": 3.1607,
      "step": 524550
    },
    {
      "epoch": 0.28721769625297755,
      "grad_norm": 0.73046875,
      "learning_rate": 9.50977097709771e-05,
      "loss": 3.2372,
      "step": 524560
    },
    {
      "epoch": 0.28723633948913574,
      "grad_norm": 0.76953125,
      "learning_rate": 9.509570957095711e-05,
      "loss": 3.2479,
      "step": 524570
    },
    {
      "epoch": 0.2872549827252939,
      "grad_norm": 0.8828125,
      "learning_rate": 9.50937093709371e-05,
      "loss": 3.204,
      "step": 524580
    },
    {
      "epoch": 0.28727362596145206,
      "grad_norm": 0.80859375,
      "learning_rate": 9.50917091709171e-05,
      "loss": 3.2758,
      "step": 524590
    },
    {
      "epoch": 0.28729226919761025,
      "grad_norm": 0.75,
      "learning_rate": 9.50897089708971e-05,
      "loss": 3.1472,
      "step": 524600
    },
    {
      "epoch": 0.28731091243376844,
      "grad_norm": 0.78125,
      "learning_rate": 9.508770877087709e-05,
      "loss": 3.1242,
      "step": 524610
    },
    {
      "epoch": 0.28732955566992663,
      "grad_norm": 0.69140625,
      "learning_rate": 9.508570857085709e-05,
      "loss": 3.241,
      "step": 524620
    },
    {
      "epoch": 0.2873481989060848,
      "grad_norm": 0.72265625,
      "learning_rate": 9.508370837083708e-05,
      "loss": 3.2004,
      "step": 524630
    },
    {
      "epoch": 0.287366842142243,
      "grad_norm": 0.6875,
      "learning_rate": 9.508170817081708e-05,
      "loss": 3.177,
      "step": 524640
    },
    {
      "epoch": 0.2873854853784012,
      "grad_norm": 0.69140625,
      "learning_rate": 9.507970797079709e-05,
      "loss": 3.1067,
      "step": 524650
    },
    {
      "epoch": 0.2874041286145594,
      "grad_norm": 0.76953125,
      "learning_rate": 9.507770777077708e-05,
      "loss": 3.15,
      "step": 524660
    },
    {
      "epoch": 0.2874227718507176,
      "grad_norm": 0.703125,
      "learning_rate": 9.507570757075708e-05,
      "loss": 3.2107,
      "step": 524670
    },
    {
      "epoch": 0.2874414150868757,
      "grad_norm": 0.69921875,
      "learning_rate": 9.507370737073707e-05,
      "loss": 3.2545,
      "step": 524680
    },
    {
      "epoch": 0.2874600583230339,
      "grad_norm": 0.67578125,
      "learning_rate": 9.507170717071707e-05,
      "loss": 3.2057,
      "step": 524690
    },
    {
      "epoch": 0.2874787015591921,
      "grad_norm": 1.1796875,
      "learning_rate": 9.506970697069707e-05,
      "loss": 3.0307,
      "step": 524700
    },
    {
      "epoch": 0.2874973447953503,
      "grad_norm": 0.75,
      "learning_rate": 9.506770677067706e-05,
      "loss": 3.0926,
      "step": 524710
    },
    {
      "epoch": 0.28751598803150846,
      "grad_norm": 0.68359375,
      "learning_rate": 9.506570657065707e-05,
      "loss": 3.2336,
      "step": 524720
    },
    {
      "epoch": 0.28753463126766665,
      "grad_norm": 0.6875,
      "learning_rate": 9.506370637063708e-05,
      "loss": 3.2876,
      "step": 524730
    },
    {
      "epoch": 0.28755327450382484,
      "grad_norm": 0.73046875,
      "learning_rate": 9.506170617061708e-05,
      "loss": 3.2313,
      "step": 524740
    },
    {
      "epoch": 0.28757191773998303,
      "grad_norm": 0.7734375,
      "learning_rate": 9.505970597059707e-05,
      "loss": 3.0867,
      "step": 524750
    },
    {
      "epoch": 0.2875905609761412,
      "grad_norm": 0.75,
      "learning_rate": 9.505770577057707e-05,
      "loss": 3.114,
      "step": 524760
    },
    {
      "epoch": 0.2876092042122994,
      "grad_norm": 0.79296875,
      "learning_rate": 9.505570557055706e-05,
      "loss": 3.1491,
      "step": 524770
    },
    {
      "epoch": 0.2876278474484576,
      "grad_norm": 0.765625,
      "learning_rate": 9.505370537053706e-05,
      "loss": 3.1684,
      "step": 524780
    },
    {
      "epoch": 0.28764649068461573,
      "grad_norm": 0.7265625,
      "learning_rate": 9.505170517051705e-05,
      "loss": 3.165,
      "step": 524790
    },
    {
      "epoch": 0.2876651339207739,
      "grad_norm": 0.78515625,
      "learning_rate": 9.504970497049705e-05,
      "loss": 3.1782,
      "step": 524800
    },
    {
      "epoch": 0.2876837771569321,
      "grad_norm": 0.76171875,
      "learning_rate": 9.504770477047706e-05,
      "loss": 3.3089,
      "step": 524810
    },
    {
      "epoch": 0.2877024203930903,
      "grad_norm": 0.921875,
      "learning_rate": 9.504570457045705e-05,
      "loss": 3.2926,
      "step": 524820
    },
    {
      "epoch": 0.2877210636292485,
      "grad_norm": 0.71484375,
      "learning_rate": 9.504370437043705e-05,
      "loss": 3.1963,
      "step": 524830
    },
    {
      "epoch": 0.2877397068654067,
      "grad_norm": 0.6484375,
      "learning_rate": 9.504170417041704e-05,
      "loss": 3.2015,
      "step": 524840
    },
    {
      "epoch": 0.28775835010156486,
      "grad_norm": 0.796875,
      "learning_rate": 9.503970397039704e-05,
      "loss": 3.1595,
      "step": 524850
    },
    {
      "epoch": 0.28777699333772305,
      "grad_norm": 0.73046875,
      "learning_rate": 9.503770377037704e-05,
      "loss": 3.17,
      "step": 524860
    },
    {
      "epoch": 0.28779563657388124,
      "grad_norm": 0.640625,
      "learning_rate": 9.503570357035703e-05,
      "loss": 3.2779,
      "step": 524870
    },
    {
      "epoch": 0.28781427981003943,
      "grad_norm": 0.8359375,
      "learning_rate": 9.503370337033704e-05,
      "loss": 3.2327,
      "step": 524880
    },
    {
      "epoch": 0.28783292304619756,
      "grad_norm": 0.71875,
      "learning_rate": 9.503170317031704e-05,
      "loss": 3.1809,
      "step": 524890
    },
    {
      "epoch": 0.28785156628235575,
      "grad_norm": 0.734375,
      "learning_rate": 9.502970297029703e-05,
      "loss": 3.0196,
      "step": 524900
    },
    {
      "epoch": 0.28787020951851394,
      "grad_norm": 0.71875,
      "learning_rate": 9.502770277027703e-05,
      "loss": 3.1332,
      "step": 524910
    },
    {
      "epoch": 0.28788885275467213,
      "grad_norm": 0.65234375,
      "learning_rate": 9.502570257025702e-05,
      "loss": 3.1547,
      "step": 524920
    },
    {
      "epoch": 0.2879074959908303,
      "grad_norm": 0.7265625,
      "learning_rate": 9.502370237023703e-05,
      "loss": 3.1705,
      "step": 524930
    },
    {
      "epoch": 0.2879261392269885,
      "grad_norm": 0.796875,
      "learning_rate": 9.502170217021703e-05,
      "loss": 3.2206,
      "step": 524940
    },
    {
      "epoch": 0.2879447824631467,
      "grad_norm": 0.78125,
      "learning_rate": 9.501970197019702e-05,
      "loss": 3.2434,
      "step": 524950
    },
    {
      "epoch": 0.2879634256993049,
      "grad_norm": 0.66015625,
      "learning_rate": 9.501770177017702e-05,
      "loss": 3.0985,
      "step": 524960
    },
    {
      "epoch": 0.2879820689354631,
      "grad_norm": 0.70703125,
      "learning_rate": 9.501570157015703e-05,
      "loss": 3.2236,
      "step": 524970
    },
    {
      "epoch": 0.28800071217162126,
      "grad_norm": 0.703125,
      "learning_rate": 9.501370137013702e-05,
      "loss": 3.2519,
      "step": 524980
    },
    {
      "epoch": 0.28801935540777945,
      "grad_norm": 0.66796875,
      "learning_rate": 9.501170117011702e-05,
      "loss": 3.2898,
      "step": 524990
    },
    {
      "epoch": 0.2880379986439376,
      "grad_norm": 0.78125,
      "learning_rate": 9.500970097009702e-05,
      "loss": 3.2595,
      "step": 525000
    },
    {
      "epoch": 0.2880566418800958,
      "grad_norm": 0.80078125,
      "learning_rate": 9.500770077007701e-05,
      "loss": 3.1741,
      "step": 525010
    },
    {
      "epoch": 0.28807528511625397,
      "grad_norm": 0.890625,
      "learning_rate": 9.5005700570057e-05,
      "loss": 3.2133,
      "step": 525020
    },
    {
      "epoch": 0.28809392835241215,
      "grad_norm": 0.76171875,
      "learning_rate": 9.5003700370037e-05,
      "loss": 3.093,
      "step": 525030
    },
    {
      "epoch": 0.28811257158857034,
      "grad_norm": 0.80078125,
      "learning_rate": 9.500170017001701e-05,
      "loss": 3.2164,
      "step": 525040
    },
    {
      "epoch": 0.28813121482472853,
      "grad_norm": 0.8125,
      "learning_rate": 9.4999699969997e-05,
      "loss": 3.2546,
      "step": 525050
    },
    {
      "epoch": 0.2881498580608867,
      "grad_norm": 0.78125,
      "learning_rate": 9.4997699769977e-05,
      "loss": 3.1467,
      "step": 525060
    },
    {
      "epoch": 0.2881685012970449,
      "grad_norm": 0.84765625,
      "learning_rate": 9.4995699569957e-05,
      "loss": 3.2661,
      "step": 525070
    },
    {
      "epoch": 0.2881871445332031,
      "grad_norm": 0.7421875,
      "learning_rate": 9.4993699369937e-05,
      "loss": 3.2214,
      "step": 525080
    },
    {
      "epoch": 0.2882057877693613,
      "grad_norm": 0.953125,
      "learning_rate": 9.499169916991699e-05,
      "loss": 3.2045,
      "step": 525090
    },
    {
      "epoch": 0.2882244310055194,
      "grad_norm": 0.73046875,
      "learning_rate": 9.498969896989698e-05,
      "loss": 3.2124,
      "step": 525100
    },
    {
      "epoch": 0.2882430742416776,
      "grad_norm": 0.66015625,
      "learning_rate": 9.4987698769877e-05,
      "loss": 3.2959,
      "step": 525110
    },
    {
      "epoch": 0.2882617174778358,
      "grad_norm": 0.7578125,
      "learning_rate": 9.498569856985699e-05,
      "loss": 3.1587,
      "step": 525120
    },
    {
      "epoch": 0.288280360713994,
      "grad_norm": 0.6796875,
      "learning_rate": 9.4983698369837e-05,
      "loss": 3.2722,
      "step": 525130
    },
    {
      "epoch": 0.2882990039501522,
      "grad_norm": 0.78125,
      "learning_rate": 9.4981698169817e-05,
      "loss": 3.2497,
      "step": 525140
    },
    {
      "epoch": 0.28831764718631037,
      "grad_norm": 0.68359375,
      "learning_rate": 9.497969796979699e-05,
      "loss": 3.1567,
      "step": 525150
    },
    {
      "epoch": 0.28833629042246856,
      "grad_norm": 0.80078125,
      "learning_rate": 9.497769776977699e-05,
      "loss": 3.2077,
      "step": 525160
    },
    {
      "epoch": 0.28835493365862674,
      "grad_norm": 0.6953125,
      "learning_rate": 9.497569756975698e-05,
      "loss": 3.2521,
      "step": 525170
    },
    {
      "epoch": 0.28837357689478493,
      "grad_norm": 0.58984375,
      "learning_rate": 9.497369736973698e-05,
      "loss": 3.2207,
      "step": 525180
    },
    {
      "epoch": 0.2883922201309431,
      "grad_norm": 0.73828125,
      "learning_rate": 9.497169716971697e-05,
      "loss": 3.1959,
      "step": 525190
    },
    {
      "epoch": 0.28841086336710126,
      "grad_norm": 0.71484375,
      "learning_rate": 9.496969696969698e-05,
      "loss": 3.2215,
      "step": 525200
    },
    {
      "epoch": 0.28842950660325944,
      "grad_norm": 0.75,
      "learning_rate": 9.496769676967698e-05,
      "loss": 3.2009,
      "step": 525210
    },
    {
      "epoch": 0.28844814983941763,
      "grad_norm": 0.6796875,
      "learning_rate": 9.496569656965697e-05,
      "loss": 3.2405,
      "step": 525220
    },
    {
      "epoch": 0.2884667930755758,
      "grad_norm": 0.6640625,
      "learning_rate": 9.496369636963697e-05,
      "loss": 3.2822,
      "step": 525230
    },
    {
      "epoch": 0.288485436311734,
      "grad_norm": 0.81640625,
      "learning_rate": 9.496169616961696e-05,
      "loss": 3.1842,
      "step": 525240
    },
    {
      "epoch": 0.2885040795478922,
      "grad_norm": 0.765625,
      "learning_rate": 9.495969596959696e-05,
      "loss": 3.2375,
      "step": 525250
    },
    {
      "epoch": 0.2885227227840504,
      "grad_norm": 0.66796875,
      "learning_rate": 9.495769576957695e-05,
      "loss": 3.2902,
      "step": 525260
    },
    {
      "epoch": 0.2885413660202086,
      "grad_norm": 0.79296875,
      "learning_rate": 9.495569556955696e-05,
      "loss": 3.2805,
      "step": 525270
    },
    {
      "epoch": 0.28856000925636677,
      "grad_norm": 0.7890625,
      "learning_rate": 9.495369536953696e-05,
      "loss": 3.2415,
      "step": 525280
    },
    {
      "epoch": 0.28857865249252496,
      "grad_norm": 0.75390625,
      "learning_rate": 9.495169516951696e-05,
      "loss": 3.1909,
      "step": 525290
    },
    {
      "epoch": 0.28859729572868315,
      "grad_norm": 0.83203125,
      "learning_rate": 9.494969496949695e-05,
      "loss": 3.3162,
      "step": 525300
    },
    {
      "epoch": 0.2886159389648413,
      "grad_norm": 0.7890625,
      "learning_rate": 9.494769476947695e-05,
      "loss": 3.2296,
      "step": 525310
    },
    {
      "epoch": 0.28863458220099947,
      "grad_norm": 0.69140625,
      "learning_rate": 9.494569456945696e-05,
      "loss": 3.1276,
      "step": 525320
    },
    {
      "epoch": 0.28865322543715766,
      "grad_norm": 0.74609375,
      "learning_rate": 9.494369436943695e-05,
      "loss": 3.2754,
      "step": 525330
    },
    {
      "epoch": 0.28867186867331585,
      "grad_norm": 0.7109375,
      "learning_rate": 9.494169416941695e-05,
      "loss": 3.2589,
      "step": 525340
    },
    {
      "epoch": 0.28869051190947403,
      "grad_norm": 0.65234375,
      "learning_rate": 9.493969396939694e-05,
      "loss": 3.082,
      "step": 525350
    },
    {
      "epoch": 0.2887091551456322,
      "grad_norm": 0.69921875,
      "learning_rate": 9.493769376937695e-05,
      "loss": 3.2378,
      "step": 525360
    },
    {
      "epoch": 0.2887277983817904,
      "grad_norm": 0.70703125,
      "learning_rate": 9.493569356935695e-05,
      "loss": 3.1074,
      "step": 525370
    },
    {
      "epoch": 0.2887464416179486,
      "grad_norm": 0.73046875,
      "learning_rate": 9.493369336933694e-05,
      "loss": 3.2413,
      "step": 525380
    },
    {
      "epoch": 0.2887650848541068,
      "grad_norm": 0.75390625,
      "learning_rate": 9.493169316931694e-05,
      "loss": 3.0598,
      "step": 525390
    },
    {
      "epoch": 0.288783728090265,
      "grad_norm": 0.75,
      "learning_rate": 9.492969296929693e-05,
      "loss": 3.1426,
      "step": 525400
    },
    {
      "epoch": 0.2888023713264231,
      "grad_norm": 0.78515625,
      "learning_rate": 9.492769276927693e-05,
      "loss": 3.2345,
      "step": 525410
    },
    {
      "epoch": 0.2888210145625813,
      "grad_norm": 0.6953125,
      "learning_rate": 9.492569256925692e-05,
      "loss": 3.3657,
      "step": 525420
    },
    {
      "epoch": 0.2888396577987395,
      "grad_norm": 0.671875,
      "learning_rate": 9.492369236923693e-05,
      "loss": 3.3594,
      "step": 525430
    },
    {
      "epoch": 0.2888583010348977,
      "grad_norm": 0.69921875,
      "learning_rate": 9.492169216921693e-05,
      "loss": 3.1961,
      "step": 525440
    },
    {
      "epoch": 0.28887694427105587,
      "grad_norm": 0.86328125,
      "learning_rate": 9.491969196919693e-05,
      "loss": 3.0926,
      "step": 525450
    },
    {
      "epoch": 0.28889558750721406,
      "grad_norm": 0.76953125,
      "learning_rate": 9.491769176917692e-05,
      "loss": 3.2533,
      "step": 525460
    },
    {
      "epoch": 0.28891423074337225,
      "grad_norm": 0.76171875,
      "learning_rate": 9.491569156915692e-05,
      "loss": 3.1372,
      "step": 525470
    },
    {
      "epoch": 0.28893287397953044,
      "grad_norm": 0.66796875,
      "learning_rate": 9.491369136913691e-05,
      "loss": 3.24,
      "step": 525480
    },
    {
      "epoch": 0.2889515172156886,
      "grad_norm": 0.734375,
      "learning_rate": 9.491169116911691e-05,
      "loss": 3.2445,
      "step": 525490
    },
    {
      "epoch": 0.2889701604518468,
      "grad_norm": 0.73828125,
      "learning_rate": 9.49096909690969e-05,
      "loss": 3.1847,
      "step": 525500
    },
    {
      "epoch": 0.288988803688005,
      "grad_norm": 0.6953125,
      "learning_rate": 9.490769076907691e-05,
      "loss": 3.2399,
      "step": 525510
    },
    {
      "epoch": 0.28900744692416314,
      "grad_norm": 0.84765625,
      "learning_rate": 9.490569056905692e-05,
      "loss": 3.2822,
      "step": 525520
    },
    {
      "epoch": 0.2890260901603213,
      "grad_norm": 1.109375,
      "learning_rate": 9.490369036903692e-05,
      "loss": 3.2435,
      "step": 525530
    },
    {
      "epoch": 0.2890447333964795,
      "grad_norm": 0.734375,
      "learning_rate": 9.490169016901691e-05,
      "loss": 3.1615,
      "step": 525540
    },
    {
      "epoch": 0.2890633766326377,
      "grad_norm": 0.8125,
      "learning_rate": 9.489968996899691e-05,
      "loss": 3.266,
      "step": 525550
    },
    {
      "epoch": 0.2890820198687959,
      "grad_norm": 0.77734375,
      "learning_rate": 9.48976897689769e-05,
      "loss": 3.3178,
      "step": 525560
    },
    {
      "epoch": 0.2891006631049541,
      "grad_norm": 0.69921875,
      "learning_rate": 9.48956895689569e-05,
      "loss": 3.2213,
      "step": 525570
    },
    {
      "epoch": 0.28911930634111227,
      "grad_norm": 0.8046875,
      "learning_rate": 9.48936893689369e-05,
      "loss": 3.2039,
      "step": 525580
    },
    {
      "epoch": 0.28913794957727046,
      "grad_norm": 0.859375,
      "learning_rate": 9.489168916891689e-05,
      "loss": 3.2168,
      "step": 525590
    },
    {
      "epoch": 0.28915659281342865,
      "grad_norm": 0.8046875,
      "learning_rate": 9.48896889688969e-05,
      "loss": 3.2445,
      "step": 525600
    },
    {
      "epoch": 0.28917523604958684,
      "grad_norm": 0.734375,
      "learning_rate": 9.48876887688769e-05,
      "loss": 3.1812,
      "step": 525610
    },
    {
      "epoch": 0.28919387928574497,
      "grad_norm": 1.015625,
      "learning_rate": 9.488568856885689e-05,
      "loss": 3.2648,
      "step": 525620
    },
    {
      "epoch": 0.28921252252190316,
      "grad_norm": 0.734375,
      "learning_rate": 9.488368836883689e-05,
      "loss": 3.0611,
      "step": 525630
    },
    {
      "epoch": 0.28923116575806135,
      "grad_norm": 0.640625,
      "learning_rate": 9.488168816881688e-05,
      "loss": 3.1925,
      "step": 525640
    },
    {
      "epoch": 0.28924980899421954,
      "grad_norm": 0.67578125,
      "learning_rate": 9.487968796879688e-05,
      "loss": 3.2009,
      "step": 525650
    },
    {
      "epoch": 0.2892684522303777,
      "grad_norm": 0.76171875,
      "learning_rate": 9.487768776877687e-05,
      "loss": 3.1825,
      "step": 525660
    },
    {
      "epoch": 0.2892870954665359,
      "grad_norm": 0.796875,
      "learning_rate": 9.487568756875688e-05,
      "loss": 3.2444,
      "step": 525670
    },
    {
      "epoch": 0.2893057387026941,
      "grad_norm": 0.66796875,
      "learning_rate": 9.487368736873688e-05,
      "loss": 3.1572,
      "step": 525680
    },
    {
      "epoch": 0.2893243819388523,
      "grad_norm": 0.7421875,
      "learning_rate": 9.487168716871687e-05,
      "loss": 3.1098,
      "step": 525690
    },
    {
      "epoch": 0.2893430251750105,
      "grad_norm": 0.765625,
      "learning_rate": 9.486968696869687e-05,
      "loss": 3.239,
      "step": 525700
    },
    {
      "epoch": 0.28936166841116867,
      "grad_norm": 0.73828125,
      "learning_rate": 9.486768676867688e-05,
      "loss": 3.3576,
      "step": 525710
    },
    {
      "epoch": 0.28938031164732686,
      "grad_norm": 0.71484375,
      "learning_rate": 9.486568656865687e-05,
      "loss": 3.2414,
      "step": 525720
    },
    {
      "epoch": 0.289398954883485,
      "grad_norm": 0.78125,
      "learning_rate": 9.486368636863687e-05,
      "loss": 3.231,
      "step": 525730
    },
    {
      "epoch": 0.2894175981196432,
      "grad_norm": 0.73046875,
      "learning_rate": 9.486168616861687e-05,
      "loss": 3.3087,
      "step": 525740
    },
    {
      "epoch": 0.28943624135580137,
      "grad_norm": 0.6328125,
      "learning_rate": 9.485968596859686e-05,
      "loss": 3.2181,
      "step": 525750
    },
    {
      "epoch": 0.28945488459195956,
      "grad_norm": 0.7578125,
      "learning_rate": 9.485768576857687e-05,
      "loss": 3.1339,
      "step": 525760
    },
    {
      "epoch": 0.28947352782811775,
      "grad_norm": 0.69140625,
      "learning_rate": 9.485568556855687e-05,
      "loss": 3.265,
      "step": 525770
    },
    {
      "epoch": 0.28949217106427594,
      "grad_norm": 0.75,
      "learning_rate": 9.485368536853686e-05,
      "loss": 3.2131,
      "step": 525780
    },
    {
      "epoch": 0.2895108143004341,
      "grad_norm": 0.8203125,
      "learning_rate": 9.485168516851686e-05,
      "loss": 3.2857,
      "step": 525790
    },
    {
      "epoch": 0.2895294575365923,
      "grad_norm": 0.76953125,
      "learning_rate": 9.484968496849685e-05,
      "loss": 3.0702,
      "step": 525800
    },
    {
      "epoch": 0.2895481007727505,
      "grad_norm": 0.62890625,
      "learning_rate": 9.484768476847685e-05,
      "loss": 3.1042,
      "step": 525810
    },
    {
      "epoch": 0.2895667440089087,
      "grad_norm": 0.71875,
      "learning_rate": 9.484568456845684e-05,
      "loss": 3.1898,
      "step": 525820
    },
    {
      "epoch": 0.2895853872450668,
      "grad_norm": 0.734375,
      "learning_rate": 9.484368436843685e-05,
      "loss": 3.2211,
      "step": 525830
    },
    {
      "epoch": 0.289604030481225,
      "grad_norm": 0.703125,
      "learning_rate": 9.484168416841685e-05,
      "loss": 3.1808,
      "step": 525840
    },
    {
      "epoch": 0.2896226737173832,
      "grad_norm": 0.73828125,
      "learning_rate": 9.483968396839684e-05,
      "loss": 3.1225,
      "step": 525850
    },
    {
      "epoch": 0.2896413169535414,
      "grad_norm": 0.80859375,
      "learning_rate": 9.483768376837684e-05,
      "loss": 3.0694,
      "step": 525860
    },
    {
      "epoch": 0.2896599601896996,
      "grad_norm": 0.75390625,
      "learning_rate": 9.483568356835684e-05,
      "loss": 3.233,
      "step": 525870
    },
    {
      "epoch": 0.2896786034258578,
      "grad_norm": 0.77734375,
      "learning_rate": 9.483368336833683e-05,
      "loss": 3.2047,
      "step": 525880
    },
    {
      "epoch": 0.28969724666201596,
      "grad_norm": 0.6640625,
      "learning_rate": 9.483168316831683e-05,
      "loss": 3.0796,
      "step": 525890
    },
    {
      "epoch": 0.28971588989817415,
      "grad_norm": 0.6953125,
      "learning_rate": 9.482968296829684e-05,
      "loss": 3.0694,
      "step": 525900
    },
    {
      "epoch": 0.28973453313433234,
      "grad_norm": 0.71484375,
      "learning_rate": 9.482768276827683e-05,
      "loss": 3.2115,
      "step": 525910
    },
    {
      "epoch": 0.28975317637049053,
      "grad_norm": 0.7265625,
      "learning_rate": 9.482568256825684e-05,
      "loss": 3.2346,
      "step": 525920
    },
    {
      "epoch": 0.2897718196066487,
      "grad_norm": 0.7578125,
      "learning_rate": 9.482368236823684e-05,
      "loss": 3.1603,
      "step": 525930
    },
    {
      "epoch": 0.28979046284280685,
      "grad_norm": 0.8515625,
      "learning_rate": 9.482168216821683e-05,
      "loss": 3.2044,
      "step": 525940
    },
    {
      "epoch": 0.28980910607896504,
      "grad_norm": 0.72265625,
      "learning_rate": 9.481968196819683e-05,
      "loss": 3.1675,
      "step": 525950
    },
    {
      "epoch": 0.28982774931512323,
      "grad_norm": 0.68359375,
      "learning_rate": 9.481768176817682e-05,
      "loss": 3.2859,
      "step": 525960
    },
    {
      "epoch": 0.2898463925512814,
      "grad_norm": 0.69140625,
      "learning_rate": 9.481568156815682e-05,
      "loss": 3.1879,
      "step": 525970
    },
    {
      "epoch": 0.2898650357874396,
      "grad_norm": 0.71875,
      "learning_rate": 9.481368136813681e-05,
      "loss": 3.2619,
      "step": 525980
    },
    {
      "epoch": 0.2898836790235978,
      "grad_norm": 0.68359375,
      "learning_rate": 9.481168116811682e-05,
      "loss": 3.3148,
      "step": 525990
    },
    {
      "epoch": 0.289902322259756,
      "grad_norm": 0.640625,
      "learning_rate": 9.480968096809682e-05,
      "loss": 3.1623,
      "step": 526000
    },
    {
      "epoch": 0.2899209654959142,
      "grad_norm": 0.7265625,
      "learning_rate": 9.480768076807681e-05,
      "loss": 3.1202,
      "step": 526010
    },
    {
      "epoch": 0.28993960873207236,
      "grad_norm": 0.82421875,
      "learning_rate": 9.480568056805681e-05,
      "loss": 3.1766,
      "step": 526020
    },
    {
      "epoch": 0.28995825196823055,
      "grad_norm": 0.80859375,
      "learning_rate": 9.48036803680368e-05,
      "loss": 3.2275,
      "step": 526030
    },
    {
      "epoch": 0.2899768952043887,
      "grad_norm": 0.78515625,
      "learning_rate": 9.48016801680168e-05,
      "loss": 3.1643,
      "step": 526040
    },
    {
      "epoch": 0.2899955384405469,
      "grad_norm": 0.64453125,
      "learning_rate": 9.47996799679968e-05,
      "loss": 3.1936,
      "step": 526050
    },
    {
      "epoch": 0.29001418167670506,
      "grad_norm": 0.8515625,
      "learning_rate": 9.47976797679768e-05,
      "loss": 3.2868,
      "step": 526060
    },
    {
      "epoch": 0.29003282491286325,
      "grad_norm": 1.125,
      "learning_rate": 9.47956795679568e-05,
      "loss": 3.2148,
      "step": 526070
    },
    {
      "epoch": 0.29005146814902144,
      "grad_norm": 0.64453125,
      "learning_rate": 9.47936793679368e-05,
      "loss": 3.2537,
      "step": 526080
    },
    {
      "epoch": 0.29007011138517963,
      "grad_norm": 0.69921875,
      "learning_rate": 9.479167916791679e-05,
      "loss": 3.2427,
      "step": 526090
    },
    {
      "epoch": 0.2900887546213378,
      "grad_norm": 0.69140625,
      "learning_rate": 9.478967896789679e-05,
      "loss": 3.073,
      "step": 526100
    },
    {
      "epoch": 0.290107397857496,
      "grad_norm": 0.74609375,
      "learning_rate": 9.47876787678768e-05,
      "loss": 3.2389,
      "step": 526110
    },
    {
      "epoch": 0.2901260410936542,
      "grad_norm": 0.671875,
      "learning_rate": 9.478567856785679e-05,
      "loss": 3.1764,
      "step": 526120
    },
    {
      "epoch": 0.2901446843298124,
      "grad_norm": 0.7265625,
      "learning_rate": 9.478367836783679e-05,
      "loss": 3.2061,
      "step": 526130
    },
    {
      "epoch": 0.2901633275659706,
      "grad_norm": 0.828125,
      "learning_rate": 9.478167816781678e-05,
      "loss": 3.1477,
      "step": 526140
    },
    {
      "epoch": 0.2901819708021287,
      "grad_norm": 0.73828125,
      "learning_rate": 9.477967796779679e-05,
      "loss": 3.1813,
      "step": 526150
    },
    {
      "epoch": 0.2902006140382869,
      "grad_norm": 1.09375,
      "learning_rate": 9.477767776777679e-05,
      "loss": 3.358,
      "step": 526160
    },
    {
      "epoch": 0.2902192572744451,
      "grad_norm": 0.734375,
      "learning_rate": 9.477567756775678e-05,
      "loss": 3.017,
      "step": 526170
    },
    {
      "epoch": 0.2902379005106033,
      "grad_norm": 0.73046875,
      "learning_rate": 9.477367736773678e-05,
      "loss": 3.225,
      "step": 526180
    },
    {
      "epoch": 0.29025654374676146,
      "grad_norm": 0.83203125,
      "learning_rate": 9.477167716771678e-05,
      "loss": 3.2399,
      "step": 526190
    },
    {
      "epoch": 0.29027518698291965,
      "grad_norm": 0.6953125,
      "learning_rate": 9.476967696769677e-05,
      "loss": 3.2788,
      "step": 526200
    },
    {
      "epoch": 0.29029383021907784,
      "grad_norm": 0.69140625,
      "learning_rate": 9.476767676767677e-05,
      "loss": 3.143,
      "step": 526210
    },
    {
      "epoch": 0.29031247345523603,
      "grad_norm": 0.71484375,
      "learning_rate": 9.476567656765678e-05,
      "loss": 3.393,
      "step": 526220
    },
    {
      "epoch": 0.2903311166913942,
      "grad_norm": 0.66796875,
      "learning_rate": 9.476367636763677e-05,
      "loss": 3.1876,
      "step": 526230
    },
    {
      "epoch": 0.2903497599275524,
      "grad_norm": 0.73828125,
      "learning_rate": 9.476167616761677e-05,
      "loss": 3.1542,
      "step": 526240
    },
    {
      "epoch": 0.29036840316371054,
      "grad_norm": 0.72265625,
      "learning_rate": 9.475967596759676e-05,
      "loss": 3.1467,
      "step": 526250
    },
    {
      "epoch": 0.29038704639986873,
      "grad_norm": 0.6953125,
      "learning_rate": 9.475767576757676e-05,
      "loss": 3.237,
      "step": 526260
    },
    {
      "epoch": 0.2904056896360269,
      "grad_norm": 0.78515625,
      "learning_rate": 9.475567556755675e-05,
      "loss": 3.1909,
      "step": 526270
    },
    {
      "epoch": 0.2904243328721851,
      "grad_norm": 0.76171875,
      "learning_rate": 9.475367536753675e-05,
      "loss": 3.2662,
      "step": 526280
    },
    {
      "epoch": 0.2904429761083433,
      "grad_norm": 0.765625,
      "learning_rate": 9.475167516751674e-05,
      "loss": 3.1359,
      "step": 526290
    },
    {
      "epoch": 0.2904616193445015,
      "grad_norm": 0.77734375,
      "learning_rate": 9.474967496749675e-05,
      "loss": 3.1986,
      "step": 526300
    },
    {
      "epoch": 0.2904802625806597,
      "grad_norm": 0.73828125,
      "learning_rate": 9.474767476747676e-05,
      "loss": 3.2889,
      "step": 526310
    },
    {
      "epoch": 0.29049890581681787,
      "grad_norm": 0.84765625,
      "learning_rate": 9.474567456745676e-05,
      "loss": 3.2176,
      "step": 526320
    },
    {
      "epoch": 0.29051754905297605,
      "grad_norm": 0.67578125,
      "learning_rate": 9.474367436743675e-05,
      "loss": 3.2486,
      "step": 526330
    },
    {
      "epoch": 0.29053619228913424,
      "grad_norm": 0.81640625,
      "learning_rate": 9.474167416741675e-05,
      "loss": 3.1908,
      "step": 526340
    },
    {
      "epoch": 0.29055483552529243,
      "grad_norm": 0.87109375,
      "learning_rate": 9.473967396739675e-05,
      "loss": 3.2441,
      "step": 526350
    },
    {
      "epoch": 0.29057347876145057,
      "grad_norm": 0.75390625,
      "learning_rate": 9.473767376737674e-05,
      "loss": 3.1378,
      "step": 526360
    },
    {
      "epoch": 0.29059212199760875,
      "grad_norm": 0.67578125,
      "learning_rate": 9.473567356735674e-05,
      "loss": 3.1232,
      "step": 526370
    },
    {
      "epoch": 0.29061076523376694,
      "grad_norm": 0.76171875,
      "learning_rate": 9.473367336733675e-05,
      "loss": 3.1791,
      "step": 526380
    },
    {
      "epoch": 0.29062940846992513,
      "grad_norm": 0.72265625,
      "learning_rate": 9.473167316731674e-05,
      "loss": 3.1696,
      "step": 526390
    },
    {
      "epoch": 0.2906480517060833,
      "grad_norm": 0.6640625,
      "learning_rate": 9.472967296729674e-05,
      "loss": 3.1231,
      "step": 526400
    },
    {
      "epoch": 0.2906666949422415,
      "grad_norm": 0.76953125,
      "learning_rate": 9.472767276727673e-05,
      "loss": 3.2306,
      "step": 526410
    },
    {
      "epoch": 0.2906853381783997,
      "grad_norm": 0.6640625,
      "learning_rate": 9.472567256725673e-05,
      "loss": 3.1017,
      "step": 526420
    },
    {
      "epoch": 0.2907039814145579,
      "grad_norm": 0.796875,
      "learning_rate": 9.472367236723672e-05,
      "loss": 3.1462,
      "step": 526430
    },
    {
      "epoch": 0.2907226246507161,
      "grad_norm": 0.765625,
      "learning_rate": 9.472167216721672e-05,
      "loss": 3.3198,
      "step": 526440
    },
    {
      "epoch": 0.29074126788687427,
      "grad_norm": 0.7109375,
      "learning_rate": 9.471967196719672e-05,
      "loss": 3.2775,
      "step": 526450
    },
    {
      "epoch": 0.2907599111230324,
      "grad_norm": 0.71484375,
      "learning_rate": 9.471767176717672e-05,
      "loss": 3.2157,
      "step": 526460
    },
    {
      "epoch": 0.2907785543591906,
      "grad_norm": 0.70703125,
      "learning_rate": 9.471567156715672e-05,
      "loss": 3.1445,
      "step": 526470
    },
    {
      "epoch": 0.2907971975953488,
      "grad_norm": 0.7109375,
      "learning_rate": 9.471367136713672e-05,
      "loss": 3.3544,
      "step": 526480
    },
    {
      "epoch": 0.29081584083150697,
      "grad_norm": 0.69140625,
      "learning_rate": 9.471167116711671e-05,
      "loss": 3.291,
      "step": 526490
    },
    {
      "epoch": 0.29083448406766516,
      "grad_norm": 0.6640625,
      "learning_rate": 9.470967096709672e-05,
      "loss": 3.3091,
      "step": 526500
    },
    {
      "epoch": 0.29085312730382334,
      "grad_norm": 0.796875,
      "learning_rate": 9.470767076707672e-05,
      "loss": 3.1443,
      "step": 526510
    },
    {
      "epoch": 0.29087177053998153,
      "grad_norm": 0.76171875,
      "learning_rate": 9.470567056705671e-05,
      "loss": 3.1734,
      "step": 526520
    },
    {
      "epoch": 0.2908904137761397,
      "grad_norm": 0.78515625,
      "learning_rate": 9.470367036703671e-05,
      "loss": 3.2194,
      "step": 526530
    },
    {
      "epoch": 0.2909090570122979,
      "grad_norm": 0.80859375,
      "learning_rate": 9.470167016701672e-05,
      "loss": 3.1057,
      "step": 526540
    },
    {
      "epoch": 0.2909277002484561,
      "grad_norm": 0.8125,
      "learning_rate": 9.469966996699671e-05,
      "loss": 3.2161,
      "step": 526550
    },
    {
      "epoch": 0.29094634348461423,
      "grad_norm": 0.86328125,
      "learning_rate": 9.469766976697671e-05,
      "loss": 3.2158,
      "step": 526560
    },
    {
      "epoch": 0.2909649867207724,
      "grad_norm": 0.66015625,
      "learning_rate": 9.46956695669567e-05,
      "loss": 3.1325,
      "step": 526570
    },
    {
      "epoch": 0.2909836299569306,
      "grad_norm": 0.71875,
      "learning_rate": 9.46936693669367e-05,
      "loss": 3.262,
      "step": 526580
    },
    {
      "epoch": 0.2910022731930888,
      "grad_norm": 0.6171875,
      "learning_rate": 9.46916691669167e-05,
      "loss": 3.1891,
      "step": 526590
    },
    {
      "epoch": 0.291020916429247,
      "grad_norm": 0.6328125,
      "learning_rate": 9.468966896689669e-05,
      "loss": 3.2555,
      "step": 526600
    },
    {
      "epoch": 0.2910395596654052,
      "grad_norm": 0.796875,
      "learning_rate": 9.468766876687669e-05,
      "loss": 3.1111,
      "step": 526610
    },
    {
      "epoch": 0.29105820290156337,
      "grad_norm": 0.6875,
      "learning_rate": 9.46856685668567e-05,
      "loss": 3.2247,
      "step": 526620
    },
    {
      "epoch": 0.29107684613772156,
      "grad_norm": 0.66015625,
      "learning_rate": 9.468366836683669e-05,
      "loss": 3.2954,
      "step": 526630
    },
    {
      "epoch": 0.29109548937387975,
      "grad_norm": 0.8125,
      "learning_rate": 9.468166816681669e-05,
      "loss": 3.1964,
      "step": 526640
    },
    {
      "epoch": 0.29111413261003793,
      "grad_norm": 0.71484375,
      "learning_rate": 9.467966796679668e-05,
      "loss": 3.1335,
      "step": 526650
    },
    {
      "epoch": 0.2911327758461961,
      "grad_norm": 0.79296875,
      "learning_rate": 9.467766776677668e-05,
      "loss": 3.1427,
      "step": 526660
    },
    {
      "epoch": 0.29115141908235426,
      "grad_norm": 0.83203125,
      "learning_rate": 9.467566756675667e-05,
      "loss": 3.2689,
      "step": 526670
    },
    {
      "epoch": 0.29117006231851245,
      "grad_norm": 0.71875,
      "learning_rate": 9.467366736673667e-05,
      "loss": 3.1753,
      "step": 526680
    },
    {
      "epoch": 0.29118870555467063,
      "grad_norm": 0.80078125,
      "learning_rate": 9.467166716671668e-05,
      "loss": 3.2219,
      "step": 526690
    },
    {
      "epoch": 0.2912073487908288,
      "grad_norm": 0.73046875,
      "learning_rate": 9.466966696669669e-05,
      "loss": 3.2536,
      "step": 526700
    },
    {
      "epoch": 0.291225992026987,
      "grad_norm": 0.6640625,
      "learning_rate": 9.466766676667668e-05,
      "loss": 3.2146,
      "step": 526710
    },
    {
      "epoch": 0.2912446352631452,
      "grad_norm": 0.8515625,
      "learning_rate": 9.466566656665668e-05,
      "loss": 3.1288,
      "step": 526720
    },
    {
      "epoch": 0.2912632784993034,
      "grad_norm": 0.76171875,
      "learning_rate": 9.466366636663667e-05,
      "loss": 3.0856,
      "step": 526730
    },
    {
      "epoch": 0.2912819217354616,
      "grad_norm": 0.72265625,
      "learning_rate": 9.466166616661667e-05,
      "loss": 3.2398,
      "step": 526740
    },
    {
      "epoch": 0.29130056497161977,
      "grad_norm": 0.71484375,
      "learning_rate": 9.465966596659666e-05,
      "loss": 3.3651,
      "step": 526750
    },
    {
      "epoch": 0.29131920820777796,
      "grad_norm": 0.78125,
      "learning_rate": 9.465766576657666e-05,
      "loss": 3.0578,
      "step": 526760
    },
    {
      "epoch": 0.2913378514439361,
      "grad_norm": 0.62890625,
      "learning_rate": 9.465566556655666e-05,
      "loss": 3.2063,
      "step": 526770
    },
    {
      "epoch": 0.2913564946800943,
      "grad_norm": 0.6484375,
      "learning_rate": 9.465366536653666e-05,
      "loss": 3.2736,
      "step": 526780
    },
    {
      "epoch": 0.29137513791625247,
      "grad_norm": 0.84375,
      "learning_rate": 9.465166516651666e-05,
      "loss": 3.2435,
      "step": 526790
    },
    {
      "epoch": 0.29139378115241066,
      "grad_norm": 0.70703125,
      "learning_rate": 9.464966496649666e-05,
      "loss": 3.2791,
      "step": 526800
    },
    {
      "epoch": 0.29141242438856885,
      "grad_norm": 0.71484375,
      "learning_rate": 9.464766476647665e-05,
      "loss": 3.1955,
      "step": 526810
    },
    {
      "epoch": 0.29143106762472704,
      "grad_norm": 0.75390625,
      "learning_rate": 9.464566456645665e-05,
      "loss": 3.1063,
      "step": 526820
    },
    {
      "epoch": 0.2914497108608852,
      "grad_norm": 0.75390625,
      "learning_rate": 9.464366436643664e-05,
      "loss": 3.2464,
      "step": 526830
    },
    {
      "epoch": 0.2914683540970434,
      "grad_norm": 0.6875,
      "learning_rate": 9.464166416641664e-05,
      "loss": 3.2049,
      "step": 526840
    },
    {
      "epoch": 0.2914869973332016,
      "grad_norm": 0.75,
      "learning_rate": 9.463966396639665e-05,
      "loss": 3.1687,
      "step": 526850
    },
    {
      "epoch": 0.2915056405693598,
      "grad_norm": 0.72265625,
      "learning_rate": 9.463766376637664e-05,
      "loss": 3.0531,
      "step": 526860
    },
    {
      "epoch": 0.291524283805518,
      "grad_norm": 0.8046875,
      "learning_rate": 9.463566356635664e-05,
      "loss": 3.2811,
      "step": 526870
    },
    {
      "epoch": 0.2915429270416761,
      "grad_norm": 0.71875,
      "learning_rate": 9.463366336633663e-05,
      "loss": 3.1384,
      "step": 526880
    },
    {
      "epoch": 0.2915615702778343,
      "grad_norm": 0.77734375,
      "learning_rate": 9.463166316631663e-05,
      "loss": 3.2299,
      "step": 526890
    },
    {
      "epoch": 0.2915802135139925,
      "grad_norm": 0.703125,
      "learning_rate": 9.462966296629664e-05,
      "loss": 3.2978,
      "step": 526900
    },
    {
      "epoch": 0.2915988567501507,
      "grad_norm": 0.82421875,
      "learning_rate": 9.462766276627663e-05,
      "loss": 3.1481,
      "step": 526910
    },
    {
      "epoch": 0.29161749998630887,
      "grad_norm": 0.671875,
      "learning_rate": 9.462566256625663e-05,
      "loss": 3.2974,
      "step": 526920
    },
    {
      "epoch": 0.29163614322246706,
      "grad_norm": 0.7734375,
      "learning_rate": 9.462366236623663e-05,
      "loss": 3.2611,
      "step": 526930
    },
    {
      "epoch": 0.29165478645862525,
      "grad_norm": 0.7265625,
      "learning_rate": 9.462166216621663e-05,
      "loss": 3.1637,
      "step": 526940
    },
    {
      "epoch": 0.29167342969478344,
      "grad_norm": 0.640625,
      "learning_rate": 9.461966196619663e-05,
      "loss": 3.0535,
      "step": 526950
    },
    {
      "epoch": 0.2916920729309416,
      "grad_norm": 0.78125,
      "learning_rate": 9.461766176617663e-05,
      "loss": 3.1618,
      "step": 526960
    },
    {
      "epoch": 0.2917107161670998,
      "grad_norm": 0.6796875,
      "learning_rate": 9.461566156615662e-05,
      "loss": 3.1988,
      "step": 526970
    },
    {
      "epoch": 0.29172935940325795,
      "grad_norm": 0.67578125,
      "learning_rate": 9.461366136613662e-05,
      "loss": 3.2375,
      "step": 526980
    },
    {
      "epoch": 0.29174800263941614,
      "grad_norm": 0.734375,
      "learning_rate": 9.461166116611661e-05,
      "loss": 3.2306,
      "step": 526990
    },
    {
      "epoch": 0.2917666458755743,
      "grad_norm": 0.671875,
      "learning_rate": 9.460966096609661e-05,
      "loss": 3.2111,
      "step": 527000
    },
    {
      "epoch": 0.2917852891117325,
      "grad_norm": 0.71875,
      "learning_rate": 9.460766076607662e-05,
      "loss": 3.1541,
      "step": 527010
    },
    {
      "epoch": 0.2918039323478907,
      "grad_norm": 0.72265625,
      "learning_rate": 9.460566056605661e-05,
      "loss": 3.1577,
      "step": 527020
    },
    {
      "epoch": 0.2918225755840489,
      "grad_norm": 0.72265625,
      "learning_rate": 9.460366036603661e-05,
      "loss": 3.2029,
      "step": 527030
    },
    {
      "epoch": 0.2918412188202071,
      "grad_norm": 0.72265625,
      "learning_rate": 9.46016601660166e-05,
      "loss": 3.2402,
      "step": 527040
    },
    {
      "epoch": 0.29185986205636527,
      "grad_norm": 0.76953125,
      "learning_rate": 9.45996599659966e-05,
      "loss": 3.0678,
      "step": 527050
    },
    {
      "epoch": 0.29187850529252346,
      "grad_norm": 0.78125,
      "learning_rate": 9.45976597659766e-05,
      "loss": 3.3475,
      "step": 527060
    },
    {
      "epoch": 0.29189714852868165,
      "grad_norm": 0.71875,
      "learning_rate": 9.459565956595659e-05,
      "loss": 3.1639,
      "step": 527070
    },
    {
      "epoch": 0.29191579176483984,
      "grad_norm": 0.69921875,
      "learning_rate": 9.459365936593659e-05,
      "loss": 3.1609,
      "step": 527080
    },
    {
      "epoch": 0.29193443500099797,
      "grad_norm": 0.6953125,
      "learning_rate": 9.45916591659166e-05,
      "loss": 3.2394,
      "step": 527090
    },
    {
      "epoch": 0.29195307823715616,
      "grad_norm": 0.73046875,
      "learning_rate": 9.45896589658966e-05,
      "loss": 3.2351,
      "step": 527100
    },
    {
      "epoch": 0.29197172147331435,
      "grad_norm": 0.71484375,
      "learning_rate": 9.45876587658766e-05,
      "loss": 3.2212,
      "step": 527110
    },
    {
      "epoch": 0.29199036470947254,
      "grad_norm": 0.80078125,
      "learning_rate": 9.45856585658566e-05,
      "loss": 3.3134,
      "step": 527120
    },
    {
      "epoch": 0.2920090079456307,
      "grad_norm": 0.71875,
      "learning_rate": 9.458365836583659e-05,
      "loss": 3.1554,
      "step": 527130
    },
    {
      "epoch": 0.2920276511817889,
      "grad_norm": 0.71484375,
      "learning_rate": 9.458165816581659e-05,
      "loss": 3.3148,
      "step": 527140
    },
    {
      "epoch": 0.2920462944179471,
      "grad_norm": 0.80859375,
      "learning_rate": 9.457965796579658e-05,
      "loss": 3.1108,
      "step": 527150
    },
    {
      "epoch": 0.2920649376541053,
      "grad_norm": 0.68359375,
      "learning_rate": 9.457765776577658e-05,
      "loss": 3.1734,
      "step": 527160
    },
    {
      "epoch": 0.2920835808902635,
      "grad_norm": 0.72265625,
      "learning_rate": 9.457565756575659e-05,
      "loss": 3.3444,
      "step": 527170
    },
    {
      "epoch": 0.2921022241264217,
      "grad_norm": 0.76953125,
      "learning_rate": 9.457365736573658e-05,
      "loss": 3.1223,
      "step": 527180
    },
    {
      "epoch": 0.2921208673625798,
      "grad_norm": 0.79296875,
      "learning_rate": 9.457165716571658e-05,
      "loss": 3.1993,
      "step": 527190
    },
    {
      "epoch": 0.292139510598738,
      "grad_norm": 0.73046875,
      "learning_rate": 9.456965696569657e-05,
      "loss": 3.2962,
      "step": 527200
    },
    {
      "epoch": 0.2921581538348962,
      "grad_norm": 0.6484375,
      "learning_rate": 9.456765676567657e-05,
      "loss": 3.1582,
      "step": 527210
    },
    {
      "epoch": 0.2921767970710544,
      "grad_norm": 0.75390625,
      "learning_rate": 9.456565656565657e-05,
      "loss": 3.2843,
      "step": 527220
    },
    {
      "epoch": 0.29219544030721256,
      "grad_norm": 0.82421875,
      "learning_rate": 9.456365636563656e-05,
      "loss": 3.2363,
      "step": 527230
    },
    {
      "epoch": 0.29221408354337075,
      "grad_norm": 0.7578125,
      "learning_rate": 9.456165616561656e-05,
      "loss": 3.2025,
      "step": 527240
    },
    {
      "epoch": 0.29223272677952894,
      "grad_norm": 0.72265625,
      "learning_rate": 9.455965596559657e-05,
      "loss": 3.3245,
      "step": 527250
    },
    {
      "epoch": 0.29225137001568713,
      "grad_norm": 0.68359375,
      "learning_rate": 9.455765576557656e-05,
      "loss": 3.2206,
      "step": 527260
    },
    {
      "epoch": 0.2922700132518453,
      "grad_norm": 0.78515625,
      "learning_rate": 9.455565556555656e-05,
      "loss": 3.2047,
      "step": 527270
    },
    {
      "epoch": 0.2922886564880035,
      "grad_norm": 0.7109375,
      "learning_rate": 9.455365536553655e-05,
      "loss": 3.1675,
      "step": 527280
    },
    {
      "epoch": 0.2923072997241617,
      "grad_norm": 0.6953125,
      "learning_rate": 9.455165516551656e-05,
      "loss": 3.1702,
      "step": 527290
    },
    {
      "epoch": 0.29232594296031983,
      "grad_norm": 0.734375,
      "learning_rate": 9.454965496549656e-05,
      "loss": 3.287,
      "step": 527300
    },
    {
      "epoch": 0.292344586196478,
      "grad_norm": 0.69921875,
      "learning_rate": 9.454765476547655e-05,
      "loss": 3.166,
      "step": 527310
    },
    {
      "epoch": 0.2923632294326362,
      "grad_norm": 0.65625,
      "learning_rate": 9.454565456545655e-05,
      "loss": 3.2374,
      "step": 527320
    },
    {
      "epoch": 0.2923818726687944,
      "grad_norm": 0.6796875,
      "learning_rate": 9.454365436543656e-05,
      "loss": 3.1986,
      "step": 527330
    },
    {
      "epoch": 0.2924005159049526,
      "grad_norm": 0.6953125,
      "learning_rate": 9.454165416541655e-05,
      "loss": 3.1959,
      "step": 527340
    },
    {
      "epoch": 0.2924191591411108,
      "grad_norm": 0.6953125,
      "learning_rate": 9.453965396539655e-05,
      "loss": 3.1939,
      "step": 527350
    },
    {
      "epoch": 0.29243780237726896,
      "grad_norm": 0.671875,
      "learning_rate": 9.453765376537654e-05,
      "loss": 3.2725,
      "step": 527360
    },
    {
      "epoch": 0.29245644561342715,
      "grad_norm": 0.81640625,
      "learning_rate": 9.453565356535654e-05,
      "loss": 3.1358,
      "step": 527370
    },
    {
      "epoch": 0.29247508884958534,
      "grad_norm": 0.78125,
      "learning_rate": 9.453365336533654e-05,
      "loss": 3.1758,
      "step": 527380
    },
    {
      "epoch": 0.29249373208574353,
      "grad_norm": 0.71875,
      "learning_rate": 9.453165316531653e-05,
      "loss": 3.177,
      "step": 527390
    },
    {
      "epoch": 0.29251237532190166,
      "grad_norm": 0.80078125,
      "learning_rate": 9.452965296529653e-05,
      "loss": 3.2452,
      "step": 527400
    },
    {
      "epoch": 0.29253101855805985,
      "grad_norm": 0.67578125,
      "learning_rate": 9.452765276527654e-05,
      "loss": 3.1923,
      "step": 527410
    },
    {
      "epoch": 0.29254966179421804,
      "grad_norm": 0.6171875,
      "learning_rate": 9.452565256525653e-05,
      "loss": 3.2562,
      "step": 527420
    },
    {
      "epoch": 0.29256830503037623,
      "grad_norm": 0.6796875,
      "learning_rate": 9.452365236523653e-05,
      "loss": 3.1576,
      "step": 527430
    },
    {
      "epoch": 0.2925869482665344,
      "grad_norm": 0.73828125,
      "learning_rate": 9.452165216521652e-05,
      "loss": 3.1199,
      "step": 527440
    },
    {
      "epoch": 0.2926055915026926,
      "grad_norm": 0.625,
      "learning_rate": 9.451965196519652e-05,
      "loss": 3.2161,
      "step": 527450
    },
    {
      "epoch": 0.2926242347388508,
      "grad_norm": 0.72265625,
      "learning_rate": 9.451765176517651e-05,
      "loss": 3.2483,
      "step": 527460
    },
    {
      "epoch": 0.292642877975009,
      "grad_norm": 0.7421875,
      "learning_rate": 9.451565156515651e-05,
      "loss": 3.2211,
      "step": 527470
    },
    {
      "epoch": 0.2926615212111672,
      "grad_norm": 0.74609375,
      "learning_rate": 9.451365136513652e-05,
      "loss": 3.2473,
      "step": 527480
    },
    {
      "epoch": 0.29268016444732536,
      "grad_norm": 0.74609375,
      "learning_rate": 9.451165116511653e-05,
      "loss": 3.3022,
      "step": 527490
    },
    {
      "epoch": 0.29269880768348355,
      "grad_norm": 0.8515625,
      "learning_rate": 9.450965096509652e-05,
      "loss": 3.2269,
      "step": 527500
    },
    {
      "epoch": 0.2927174509196417,
      "grad_norm": 0.73046875,
      "learning_rate": 9.450765076507652e-05,
      "loss": 3.1853,
      "step": 527510
    },
    {
      "epoch": 0.2927360941557999,
      "grad_norm": 0.83984375,
      "learning_rate": 9.450565056505651e-05,
      "loss": 3.2517,
      "step": 527520
    },
    {
      "epoch": 0.29275473739195806,
      "grad_norm": 0.66796875,
      "learning_rate": 9.450365036503651e-05,
      "loss": 3.1607,
      "step": 527530
    },
    {
      "epoch": 0.29277338062811625,
      "grad_norm": 0.7890625,
      "learning_rate": 9.45016501650165e-05,
      "loss": 3.2526,
      "step": 527540
    },
    {
      "epoch": 0.29279202386427444,
      "grad_norm": 0.72265625,
      "learning_rate": 9.44996499649965e-05,
      "loss": 3.2691,
      "step": 527550
    },
    {
      "epoch": 0.29281066710043263,
      "grad_norm": 0.8515625,
      "learning_rate": 9.44976497649765e-05,
      "loss": 3.2812,
      "step": 527560
    },
    {
      "epoch": 0.2928293103365908,
      "grad_norm": 0.75,
      "learning_rate": 9.44956495649565e-05,
      "loss": 3.2233,
      "step": 527570
    },
    {
      "epoch": 0.292847953572749,
      "grad_norm": 0.7109375,
      "learning_rate": 9.44936493649365e-05,
      "loss": 3.1371,
      "step": 527580
    },
    {
      "epoch": 0.2928665968089072,
      "grad_norm": 0.7890625,
      "learning_rate": 9.44916491649165e-05,
      "loss": 3.2085,
      "step": 527590
    },
    {
      "epoch": 0.2928852400450654,
      "grad_norm": 0.671875,
      "learning_rate": 9.448964896489649e-05,
      "loss": 3.2568,
      "step": 527600
    },
    {
      "epoch": 0.2929038832812235,
      "grad_norm": 0.734375,
      "learning_rate": 9.448764876487649e-05,
      "loss": 3.3107,
      "step": 527610
    },
    {
      "epoch": 0.2929225265173817,
      "grad_norm": 0.78515625,
      "learning_rate": 9.448564856485648e-05,
      "loss": 3.1114,
      "step": 527620
    },
    {
      "epoch": 0.2929411697535399,
      "grad_norm": 0.7421875,
      "learning_rate": 9.448364836483648e-05,
      "loss": 3.2978,
      "step": 527630
    },
    {
      "epoch": 0.2929598129896981,
      "grad_norm": 0.64453125,
      "learning_rate": 9.448164816481649e-05,
      "loss": 3.1379,
      "step": 527640
    },
    {
      "epoch": 0.2929784562258563,
      "grad_norm": 0.73828125,
      "learning_rate": 9.447964796479648e-05,
      "loss": 3.2824,
      "step": 527650
    },
    {
      "epoch": 0.29299709946201447,
      "grad_norm": 0.6484375,
      "learning_rate": 9.447764776477648e-05,
      "loss": 3.3145,
      "step": 527660
    },
    {
      "epoch": 0.29301574269817265,
      "grad_norm": 0.6875,
      "learning_rate": 9.447564756475648e-05,
      "loss": 3.1579,
      "step": 527670
    },
    {
      "epoch": 0.29303438593433084,
      "grad_norm": 0.765625,
      "learning_rate": 9.447364736473647e-05,
      "loss": 3.2043,
      "step": 527680
    },
    {
      "epoch": 0.29305302917048903,
      "grad_norm": 0.70703125,
      "learning_rate": 9.447164716471648e-05,
      "loss": 3.3173,
      "step": 527690
    },
    {
      "epoch": 0.2930716724066472,
      "grad_norm": 0.765625,
      "learning_rate": 9.446964696469648e-05,
      "loss": 3.2681,
      "step": 527700
    },
    {
      "epoch": 0.2930903156428054,
      "grad_norm": 0.64453125,
      "learning_rate": 9.446764676467647e-05,
      "loss": 3.1705,
      "step": 527710
    },
    {
      "epoch": 0.29310895887896354,
      "grad_norm": 0.80859375,
      "learning_rate": 9.446564656465647e-05,
      "loss": 3.2024,
      "step": 527720
    },
    {
      "epoch": 0.29312760211512173,
      "grad_norm": 0.80859375,
      "learning_rate": 9.446364636463648e-05,
      "loss": 3.1937,
      "step": 527730
    },
    {
      "epoch": 0.2931462453512799,
      "grad_norm": 0.72265625,
      "learning_rate": 9.446164616461647e-05,
      "loss": 3.2219,
      "step": 527740
    },
    {
      "epoch": 0.2931648885874381,
      "grad_norm": 0.671875,
      "learning_rate": 9.445964596459647e-05,
      "loss": 3.1752,
      "step": 527750
    },
    {
      "epoch": 0.2931835318235963,
      "grad_norm": 0.78515625,
      "learning_rate": 9.445764576457646e-05,
      "loss": 3.1006,
      "step": 527760
    },
    {
      "epoch": 0.2932021750597545,
      "grad_norm": 0.73046875,
      "learning_rate": 9.445564556455646e-05,
      "loss": 2.979,
      "step": 527770
    },
    {
      "epoch": 0.2932208182959127,
      "grad_norm": 0.71875,
      "learning_rate": 9.445364536453645e-05,
      "loss": 3.2597,
      "step": 527780
    },
    {
      "epoch": 0.29323946153207087,
      "grad_norm": 0.74609375,
      "learning_rate": 9.445164516451645e-05,
      "loss": 3.1578,
      "step": 527790
    },
    {
      "epoch": 0.29325810476822906,
      "grad_norm": 0.65625,
      "learning_rate": 9.444964496449646e-05,
      "loss": 3.232,
      "step": 527800
    },
    {
      "epoch": 0.29327674800438724,
      "grad_norm": 0.67578125,
      "learning_rate": 9.444764476447645e-05,
      "loss": 3.2255,
      "step": 527810
    },
    {
      "epoch": 0.2932953912405454,
      "grad_norm": 0.734375,
      "learning_rate": 9.444564456445645e-05,
      "loss": 3.3614,
      "step": 527820
    },
    {
      "epoch": 0.29331403447670357,
      "grad_norm": 0.75,
      "learning_rate": 9.444364436443645e-05,
      "loss": 3.1175,
      "step": 527830
    },
    {
      "epoch": 0.29333267771286176,
      "grad_norm": 0.73828125,
      "learning_rate": 9.444164416441644e-05,
      "loss": 3.3195,
      "step": 527840
    },
    {
      "epoch": 0.29335132094901994,
      "grad_norm": 0.73046875,
      "learning_rate": 9.443964396439644e-05,
      "loss": 3.2559,
      "step": 527850
    },
    {
      "epoch": 0.29336996418517813,
      "grad_norm": 0.7265625,
      "learning_rate": 9.443764376437643e-05,
      "loss": 3.1254,
      "step": 527860
    },
    {
      "epoch": 0.2933886074213363,
      "grad_norm": 0.8125,
      "learning_rate": 9.443564356435644e-05,
      "loss": 3.1027,
      "step": 527870
    },
    {
      "epoch": 0.2934072506574945,
      "grad_norm": 0.640625,
      "learning_rate": 9.443364336433644e-05,
      "loss": 3.2777,
      "step": 527880
    },
    {
      "epoch": 0.2934258938936527,
      "grad_norm": 0.6484375,
      "learning_rate": 9.443164316431645e-05,
      "loss": 3.2258,
      "step": 527890
    },
    {
      "epoch": 0.2934445371298109,
      "grad_norm": 0.7265625,
      "learning_rate": 9.442964296429644e-05,
      "loss": 3.1766,
      "step": 527900
    },
    {
      "epoch": 0.2934631803659691,
      "grad_norm": 0.7734375,
      "learning_rate": 9.442764276427644e-05,
      "loss": 3.1592,
      "step": 527910
    },
    {
      "epoch": 0.29348182360212727,
      "grad_norm": 0.67578125,
      "learning_rate": 9.442564256425643e-05,
      "loss": 3.1971,
      "step": 527920
    },
    {
      "epoch": 0.2935004668382854,
      "grad_norm": 0.86328125,
      "learning_rate": 9.442364236423643e-05,
      "loss": 3.2721,
      "step": 527930
    },
    {
      "epoch": 0.2935191100744436,
      "grad_norm": 0.671875,
      "learning_rate": 9.442164216421642e-05,
      "loss": 3.0764,
      "step": 527940
    },
    {
      "epoch": 0.2935377533106018,
      "grad_norm": 0.671875,
      "learning_rate": 9.441964196419642e-05,
      "loss": 3.1578,
      "step": 527950
    },
    {
      "epoch": 0.29355639654675997,
      "grad_norm": 0.76953125,
      "learning_rate": 9.441764176417643e-05,
      "loss": 3.1885,
      "step": 527960
    },
    {
      "epoch": 0.29357503978291816,
      "grad_norm": 0.890625,
      "learning_rate": 9.441564156415642e-05,
      "loss": 3.3328,
      "step": 527970
    },
    {
      "epoch": 0.29359368301907635,
      "grad_norm": 0.74609375,
      "learning_rate": 9.441364136413642e-05,
      "loss": 3.184,
      "step": 527980
    },
    {
      "epoch": 0.29361232625523453,
      "grad_norm": 0.69140625,
      "learning_rate": 9.441164116411642e-05,
      "loss": 3.1901,
      "step": 527990
    },
    {
      "epoch": 0.2936309694913927,
      "grad_norm": 0.7734375,
      "learning_rate": 9.440964096409641e-05,
      "loss": 3.2024,
      "step": 528000
    },
    {
      "epoch": 0.2936496127275509,
      "grad_norm": 0.734375,
      "learning_rate": 9.440764076407641e-05,
      "loss": 3.2028,
      "step": 528010
    },
    {
      "epoch": 0.2936682559637091,
      "grad_norm": 0.65234375,
      "learning_rate": 9.44056405640564e-05,
      "loss": 3.2322,
      "step": 528020
    },
    {
      "epoch": 0.29368689919986724,
      "grad_norm": 0.7890625,
      "learning_rate": 9.440364036403641e-05,
      "loss": 3.1851,
      "step": 528030
    },
    {
      "epoch": 0.2937055424360254,
      "grad_norm": 0.68359375,
      "learning_rate": 9.440164016401641e-05,
      "loss": 3.1235,
      "step": 528040
    },
    {
      "epoch": 0.2937241856721836,
      "grad_norm": 0.796875,
      "learning_rate": 9.43996399639964e-05,
      "loss": 3.2208,
      "step": 528050
    },
    {
      "epoch": 0.2937428289083418,
      "grad_norm": 0.78515625,
      "learning_rate": 9.43976397639764e-05,
      "loss": 3.1797,
      "step": 528060
    },
    {
      "epoch": 0.2937614721445,
      "grad_norm": 0.76171875,
      "learning_rate": 9.43956395639564e-05,
      "loss": 3.2903,
      "step": 528070
    },
    {
      "epoch": 0.2937801153806582,
      "grad_norm": 0.8359375,
      "learning_rate": 9.43936393639364e-05,
      "loss": 3.1587,
      "step": 528080
    },
    {
      "epoch": 0.29379875861681637,
      "grad_norm": 0.71875,
      "learning_rate": 9.43916391639164e-05,
      "loss": 3.1193,
      "step": 528090
    },
    {
      "epoch": 0.29381740185297456,
      "grad_norm": 0.8203125,
      "learning_rate": 9.43896389638964e-05,
      "loss": 3.0513,
      "step": 528100
    },
    {
      "epoch": 0.29383604508913275,
      "grad_norm": 0.89453125,
      "learning_rate": 9.438763876387639e-05,
      "loss": 3.214,
      "step": 528110
    },
    {
      "epoch": 0.29385468832529094,
      "grad_norm": 0.78515625,
      "learning_rate": 9.43856385638564e-05,
      "loss": 3.1639,
      "step": 528120
    },
    {
      "epoch": 0.29387333156144907,
      "grad_norm": 0.77734375,
      "learning_rate": 9.43836383638364e-05,
      "loss": 3.1489,
      "step": 528130
    },
    {
      "epoch": 0.29389197479760726,
      "grad_norm": 0.72265625,
      "learning_rate": 9.438163816381639e-05,
      "loss": 3.2589,
      "step": 528140
    },
    {
      "epoch": 0.29391061803376545,
      "grad_norm": 0.77734375,
      "learning_rate": 9.437963796379639e-05,
      "loss": 3.2689,
      "step": 528150
    },
    {
      "epoch": 0.29392926126992364,
      "grad_norm": 0.7890625,
      "learning_rate": 9.437763776377638e-05,
      "loss": 3.2137,
      "step": 528160
    },
    {
      "epoch": 0.2939479045060818,
      "grad_norm": 0.62109375,
      "learning_rate": 9.437563756375638e-05,
      "loss": 3.1921,
      "step": 528170
    },
    {
      "epoch": 0.29396654774224,
      "grad_norm": 0.83203125,
      "learning_rate": 9.437363736373637e-05,
      "loss": 3.1197,
      "step": 528180
    },
    {
      "epoch": 0.2939851909783982,
      "grad_norm": 0.875,
      "learning_rate": 9.437163716371638e-05,
      "loss": 3.2458,
      "step": 528190
    },
    {
      "epoch": 0.2940038342145564,
      "grad_norm": 0.71484375,
      "learning_rate": 9.436963696369638e-05,
      "loss": 3.3193,
      "step": 528200
    },
    {
      "epoch": 0.2940224774507146,
      "grad_norm": 0.625,
      "learning_rate": 9.436763676367637e-05,
      "loss": 3.1736,
      "step": 528210
    },
    {
      "epoch": 0.29404112068687277,
      "grad_norm": 0.765625,
      "learning_rate": 9.436563656365637e-05,
      "loss": 3.1729,
      "step": 528220
    },
    {
      "epoch": 0.29405976392303096,
      "grad_norm": 0.859375,
      "learning_rate": 9.436363636363636e-05,
      "loss": 3.1991,
      "step": 528230
    },
    {
      "epoch": 0.2940784071591891,
      "grad_norm": 0.82421875,
      "learning_rate": 9.436163616361636e-05,
      "loss": 3.0899,
      "step": 528240
    },
    {
      "epoch": 0.2940970503953473,
      "grad_norm": 0.859375,
      "learning_rate": 9.435963596359636e-05,
      "loss": 3.2377,
      "step": 528250
    },
    {
      "epoch": 0.29411569363150547,
      "grad_norm": 0.734375,
      "learning_rate": 9.435763576357635e-05,
      "loss": 3.2177,
      "step": 528260
    },
    {
      "epoch": 0.29413433686766366,
      "grad_norm": 0.640625,
      "learning_rate": 9.435563556355636e-05,
      "loss": 3.2469,
      "step": 528270
    },
    {
      "epoch": 0.29415298010382185,
      "grad_norm": 0.83203125,
      "learning_rate": 9.435363536353637e-05,
      "loss": 3.0532,
      "step": 528280
    },
    {
      "epoch": 0.29417162333998004,
      "grad_norm": 0.7265625,
      "learning_rate": 9.435163516351637e-05,
      "loss": 3.2723,
      "step": 528290
    },
    {
      "epoch": 0.2941902665761382,
      "grad_norm": 0.77734375,
      "learning_rate": 9.434963496349636e-05,
      "loss": 3.1407,
      "step": 528300
    },
    {
      "epoch": 0.2942089098122964,
      "grad_norm": 0.6875,
      "learning_rate": 9.434763476347636e-05,
      "loss": 3.072,
      "step": 528310
    },
    {
      "epoch": 0.2942275530484546,
      "grad_norm": 0.6796875,
      "learning_rate": 9.434563456345635e-05,
      "loss": 3.2198,
      "step": 528320
    },
    {
      "epoch": 0.2942461962846128,
      "grad_norm": 0.76953125,
      "learning_rate": 9.434363436343635e-05,
      "loss": 3.3154,
      "step": 528330
    },
    {
      "epoch": 0.2942648395207709,
      "grad_norm": 0.66015625,
      "learning_rate": 9.434163416341634e-05,
      "loss": 3.3211,
      "step": 528340
    },
    {
      "epoch": 0.2942834827569291,
      "grad_norm": 0.80078125,
      "learning_rate": 9.433963396339635e-05,
      "loss": 3.3179,
      "step": 528350
    },
    {
      "epoch": 0.2943021259930873,
      "grad_norm": 0.6796875,
      "learning_rate": 9.433763376337635e-05,
      "loss": 3.2457,
      "step": 528360
    },
    {
      "epoch": 0.2943207692292455,
      "grad_norm": 0.671875,
      "learning_rate": 9.433563356335634e-05,
      "loss": 3.1412,
      "step": 528370
    },
    {
      "epoch": 0.2943394124654037,
      "grad_norm": 0.7578125,
      "learning_rate": 9.433363336333634e-05,
      "loss": 3.1465,
      "step": 528380
    },
    {
      "epoch": 0.29435805570156187,
      "grad_norm": 0.765625,
      "learning_rate": 9.433163316331633e-05,
      "loss": 3.2729,
      "step": 528390
    },
    {
      "epoch": 0.29437669893772006,
      "grad_norm": 0.703125,
      "learning_rate": 9.432963296329633e-05,
      "loss": 3.1964,
      "step": 528400
    },
    {
      "epoch": 0.29439534217387825,
      "grad_norm": 0.71484375,
      "learning_rate": 9.432763276327633e-05,
      "loss": 3.0726,
      "step": 528410
    },
    {
      "epoch": 0.29441398541003644,
      "grad_norm": 0.75390625,
      "learning_rate": 9.432563256325632e-05,
      "loss": 3.1377,
      "step": 528420
    },
    {
      "epoch": 0.2944326286461946,
      "grad_norm": 0.671875,
      "learning_rate": 9.432363236323633e-05,
      "loss": 3.2376,
      "step": 528430
    },
    {
      "epoch": 0.2944512718823528,
      "grad_norm": 0.734375,
      "learning_rate": 9.432163216321633e-05,
      "loss": 3.2734,
      "step": 528440
    },
    {
      "epoch": 0.29446991511851095,
      "grad_norm": 0.7421875,
      "learning_rate": 9.431963196319632e-05,
      "loss": 3.1879,
      "step": 528450
    },
    {
      "epoch": 0.29448855835466914,
      "grad_norm": 0.65625,
      "learning_rate": 9.431763176317632e-05,
      "loss": 3.1556,
      "step": 528460
    },
    {
      "epoch": 0.2945072015908273,
      "grad_norm": 0.74609375,
      "learning_rate": 9.431563156315631e-05,
      "loss": 3.2184,
      "step": 528470
    },
    {
      "epoch": 0.2945258448269855,
      "grad_norm": 0.734375,
      "learning_rate": 9.431363136313632e-05,
      "loss": 3.2277,
      "step": 528480
    },
    {
      "epoch": 0.2945444880631437,
      "grad_norm": 0.70703125,
      "learning_rate": 9.431163116311632e-05,
      "loss": 3.3323,
      "step": 528490
    },
    {
      "epoch": 0.2945631312993019,
      "grad_norm": 0.70703125,
      "learning_rate": 9.430963096309631e-05,
      "loss": 3.1829,
      "step": 528500
    },
    {
      "epoch": 0.2945817745354601,
      "grad_norm": 0.78125,
      "learning_rate": 9.430763076307631e-05,
      "loss": 3.1271,
      "step": 528510
    },
    {
      "epoch": 0.2946004177716183,
      "grad_norm": 0.77734375,
      "learning_rate": 9.430563056305632e-05,
      "loss": 3.2835,
      "step": 528520
    },
    {
      "epoch": 0.29461906100777646,
      "grad_norm": 0.70703125,
      "learning_rate": 9.430363036303631e-05,
      "loss": 3.2159,
      "step": 528530
    },
    {
      "epoch": 0.29463770424393465,
      "grad_norm": 0.73828125,
      "learning_rate": 9.430163016301631e-05,
      "loss": 3.2554,
      "step": 528540
    },
    {
      "epoch": 0.2946563474800928,
      "grad_norm": 0.75390625,
      "learning_rate": 9.42996299629963e-05,
      "loss": 3.0525,
      "step": 528550
    },
    {
      "epoch": 0.294674990716251,
      "grad_norm": 0.79296875,
      "learning_rate": 9.42976297629763e-05,
      "loss": 3.1276,
      "step": 528560
    },
    {
      "epoch": 0.29469363395240916,
      "grad_norm": 0.7890625,
      "learning_rate": 9.42956295629563e-05,
      "loss": 3.2465,
      "step": 528570
    },
    {
      "epoch": 0.29471227718856735,
      "grad_norm": 0.8203125,
      "learning_rate": 9.429362936293629e-05,
      "loss": 3.2629,
      "step": 528580
    },
    {
      "epoch": 0.29473092042472554,
      "grad_norm": 0.69921875,
      "learning_rate": 9.42916291629163e-05,
      "loss": 3.1154,
      "step": 528590
    },
    {
      "epoch": 0.29474956366088373,
      "grad_norm": 0.66015625,
      "learning_rate": 9.42896289628963e-05,
      "loss": 3.2378,
      "step": 528600
    },
    {
      "epoch": 0.2947682068970419,
      "grad_norm": 0.74609375,
      "learning_rate": 9.428762876287629e-05,
      "loss": 3.3433,
      "step": 528610
    },
    {
      "epoch": 0.2947868501332001,
      "grad_norm": 0.7265625,
      "learning_rate": 9.428562856285629e-05,
      "loss": 3.1456,
      "step": 528620
    },
    {
      "epoch": 0.2948054933693583,
      "grad_norm": 0.71484375,
      "learning_rate": 9.428362836283628e-05,
      "loss": 3.1913,
      "step": 528630
    },
    {
      "epoch": 0.2948241366055165,
      "grad_norm": 0.79296875,
      "learning_rate": 9.428162816281628e-05,
      "loss": 3.252,
      "step": 528640
    },
    {
      "epoch": 0.2948427798416747,
      "grad_norm": 0.5859375,
      "learning_rate": 9.427962796279627e-05,
      "loss": 3.0743,
      "step": 528650
    },
    {
      "epoch": 0.2948614230778328,
      "grad_norm": 0.609375,
      "learning_rate": 9.427762776277628e-05,
      "loss": 3.2801,
      "step": 528660
    },
    {
      "epoch": 0.294880066313991,
      "grad_norm": 0.6328125,
      "learning_rate": 9.427562756275628e-05,
      "loss": 3.1192,
      "step": 528670
    },
    {
      "epoch": 0.2948987095501492,
      "grad_norm": 0.6796875,
      "learning_rate": 9.427362736273629e-05,
      "loss": 3.2148,
      "step": 528680
    },
    {
      "epoch": 0.2949173527863074,
      "grad_norm": 0.73046875,
      "learning_rate": 9.427162716271628e-05,
      "loss": 3.0894,
      "step": 528690
    },
    {
      "epoch": 0.29493599602246556,
      "grad_norm": 0.7109375,
      "learning_rate": 9.426962696269628e-05,
      "loss": 3.1022,
      "step": 528700
    },
    {
      "epoch": 0.29495463925862375,
      "grad_norm": 0.64453125,
      "learning_rate": 9.426762676267627e-05,
      "loss": 3.1421,
      "step": 528710
    },
    {
      "epoch": 0.29497328249478194,
      "grad_norm": 0.765625,
      "learning_rate": 9.426562656265627e-05,
      "loss": 3.2181,
      "step": 528720
    },
    {
      "epoch": 0.29499192573094013,
      "grad_norm": 0.83984375,
      "learning_rate": 9.426362636263627e-05,
      "loss": 3.1501,
      "step": 528730
    },
    {
      "epoch": 0.2950105689670983,
      "grad_norm": 0.78125,
      "learning_rate": 9.426162616261626e-05,
      "loss": 3.2409,
      "step": 528740
    },
    {
      "epoch": 0.2950292122032565,
      "grad_norm": 0.78125,
      "learning_rate": 9.425962596259627e-05,
      "loss": 3.0658,
      "step": 528750
    },
    {
      "epoch": 0.29504785543941464,
      "grad_norm": 0.640625,
      "learning_rate": 9.425762576257627e-05,
      "loss": 3.1938,
      "step": 528760
    },
    {
      "epoch": 0.29506649867557283,
      "grad_norm": 0.67578125,
      "learning_rate": 9.425562556255626e-05,
      "loss": 3.255,
      "step": 528770
    },
    {
      "epoch": 0.295085141911731,
      "grad_norm": 0.7578125,
      "learning_rate": 9.425362536253626e-05,
      "loss": 3.1421,
      "step": 528780
    },
    {
      "epoch": 0.2951037851478892,
      "grad_norm": 0.7578125,
      "learning_rate": 9.425162516251625e-05,
      "loss": 3.3311,
      "step": 528790
    },
    {
      "epoch": 0.2951224283840474,
      "grad_norm": 0.671875,
      "learning_rate": 9.424962496249625e-05,
      "loss": 3.2399,
      "step": 528800
    },
    {
      "epoch": 0.2951410716202056,
      "grad_norm": 0.79296875,
      "learning_rate": 9.424762476247624e-05,
      "loss": 3.2515,
      "step": 528810
    },
    {
      "epoch": 0.2951597148563638,
      "grad_norm": 0.72265625,
      "learning_rate": 9.424562456245625e-05,
      "loss": 3.2068,
      "step": 528820
    },
    {
      "epoch": 0.29517835809252196,
      "grad_norm": 0.71875,
      "learning_rate": 9.424362436243625e-05,
      "loss": 3.1682,
      "step": 528830
    },
    {
      "epoch": 0.29519700132868015,
      "grad_norm": 0.86328125,
      "learning_rate": 9.424162416241624e-05,
      "loss": 3.1364,
      "step": 528840
    },
    {
      "epoch": 0.29521564456483834,
      "grad_norm": 0.7109375,
      "learning_rate": 9.423962396239624e-05,
      "loss": 3.1746,
      "step": 528850
    },
    {
      "epoch": 0.29523428780099653,
      "grad_norm": 0.70703125,
      "learning_rate": 9.423762376237624e-05,
      "loss": 3.2229,
      "step": 528860
    },
    {
      "epoch": 0.29525293103715466,
      "grad_norm": 0.7578125,
      "learning_rate": 9.423562356235625e-05,
      "loss": 3.2752,
      "step": 528870
    },
    {
      "epoch": 0.29527157427331285,
      "grad_norm": 0.7578125,
      "learning_rate": 9.423362336233624e-05,
      "loss": 3.2725,
      "step": 528880
    },
    {
      "epoch": 0.29529021750947104,
      "grad_norm": 0.71875,
      "learning_rate": 9.423162316231624e-05,
      "loss": 3.2065,
      "step": 528890
    },
    {
      "epoch": 0.29530886074562923,
      "grad_norm": 0.80078125,
      "learning_rate": 9.422962296229623e-05,
      "loss": 3.1892,
      "step": 528900
    },
    {
      "epoch": 0.2953275039817874,
      "grad_norm": 0.7734375,
      "learning_rate": 9.422762276227624e-05,
      "loss": 3.2422,
      "step": 528910
    },
    {
      "epoch": 0.2953461472179456,
      "grad_norm": 0.73828125,
      "learning_rate": 9.422562256225624e-05,
      "loss": 3.1917,
      "step": 528920
    },
    {
      "epoch": 0.2953647904541038,
      "grad_norm": 0.84375,
      "learning_rate": 9.422362236223623e-05,
      "loss": 3.22,
      "step": 528930
    },
    {
      "epoch": 0.295383433690262,
      "grad_norm": 0.609375,
      "learning_rate": 9.422162216221623e-05,
      "loss": 3.293,
      "step": 528940
    },
    {
      "epoch": 0.2954020769264202,
      "grad_norm": 0.72265625,
      "learning_rate": 9.421962196219622e-05,
      "loss": 3.2201,
      "step": 528950
    },
    {
      "epoch": 0.29542072016257837,
      "grad_norm": 0.7734375,
      "learning_rate": 9.421762176217622e-05,
      "loss": 3.1196,
      "step": 528960
    },
    {
      "epoch": 0.2954393633987365,
      "grad_norm": 0.765625,
      "learning_rate": 9.421562156215621e-05,
      "loss": 3.2735,
      "step": 528970
    },
    {
      "epoch": 0.2954580066348947,
      "grad_norm": 0.6875,
      "learning_rate": 9.421362136213622e-05,
      "loss": 3.2551,
      "step": 528980
    },
    {
      "epoch": 0.2954766498710529,
      "grad_norm": 0.79296875,
      "learning_rate": 9.421162116211622e-05,
      "loss": 3.4138,
      "step": 528990
    },
    {
      "epoch": 0.29549529310721107,
      "grad_norm": 0.72265625,
      "learning_rate": 9.420962096209621e-05,
      "loss": 3.2073,
      "step": 529000
    },
    {
      "epoch": 0.29551393634336925,
      "grad_norm": 0.75,
      "learning_rate": 9.420762076207621e-05,
      "loss": 3.2552,
      "step": 529010
    },
    {
      "epoch": 0.29553257957952744,
      "grad_norm": 0.69140625,
      "learning_rate": 9.42056205620562e-05,
      "loss": 3.2714,
      "step": 529020
    },
    {
      "epoch": 0.29555122281568563,
      "grad_norm": 0.7421875,
      "learning_rate": 9.42036203620362e-05,
      "loss": 3.3125,
      "step": 529030
    },
    {
      "epoch": 0.2955698660518438,
      "grad_norm": 0.70703125,
      "learning_rate": 9.42016201620162e-05,
      "loss": 3.2344,
      "step": 529040
    },
    {
      "epoch": 0.295588509288002,
      "grad_norm": 0.8125,
      "learning_rate": 9.419961996199619e-05,
      "loss": 3.2496,
      "step": 529050
    },
    {
      "epoch": 0.2956071525241602,
      "grad_norm": 0.734375,
      "learning_rate": 9.41976197619762e-05,
      "loss": 3.2713,
      "step": 529060
    },
    {
      "epoch": 0.2956257957603184,
      "grad_norm": 0.81640625,
      "learning_rate": 9.419561956195621e-05,
      "loss": 3.2262,
      "step": 529070
    },
    {
      "epoch": 0.2956444389964765,
      "grad_norm": 0.828125,
      "learning_rate": 9.419361936193621e-05,
      "loss": 3.1651,
      "step": 529080
    },
    {
      "epoch": 0.2956630822326347,
      "grad_norm": 0.6484375,
      "learning_rate": 9.41916191619162e-05,
      "loss": 3.233,
      "step": 529090
    },
    {
      "epoch": 0.2956817254687929,
      "grad_norm": 0.796875,
      "learning_rate": 9.41896189618962e-05,
      "loss": 3.2443,
      "step": 529100
    },
    {
      "epoch": 0.2957003687049511,
      "grad_norm": 0.734375,
      "learning_rate": 9.41876187618762e-05,
      "loss": 3.2358,
      "step": 529110
    },
    {
      "epoch": 0.2957190119411093,
      "grad_norm": 0.6796875,
      "learning_rate": 9.418561856185619e-05,
      "loss": 3.2464,
      "step": 529120
    },
    {
      "epoch": 0.29573765517726747,
      "grad_norm": 0.66015625,
      "learning_rate": 9.418361836183618e-05,
      "loss": 3.0528,
      "step": 529130
    },
    {
      "epoch": 0.29575629841342566,
      "grad_norm": 0.78515625,
      "learning_rate": 9.41816181618162e-05,
      "loss": 3.3155,
      "step": 529140
    },
    {
      "epoch": 0.29577494164958384,
      "grad_norm": 0.7109375,
      "learning_rate": 9.417961796179619e-05,
      "loss": 3.253,
      "step": 529150
    },
    {
      "epoch": 0.29579358488574203,
      "grad_norm": 0.859375,
      "learning_rate": 9.417761776177619e-05,
      "loss": 3.1928,
      "step": 529160
    },
    {
      "epoch": 0.2958122281219002,
      "grad_norm": 0.74609375,
      "learning_rate": 9.417561756175618e-05,
      "loss": 3.1883,
      "step": 529170
    },
    {
      "epoch": 0.29583087135805836,
      "grad_norm": 0.90234375,
      "learning_rate": 9.417361736173618e-05,
      "loss": 3.2784,
      "step": 529180
    },
    {
      "epoch": 0.29584951459421654,
      "grad_norm": 0.7109375,
      "learning_rate": 9.417161716171617e-05,
      "loss": 3.1754,
      "step": 529190
    },
    {
      "epoch": 0.29586815783037473,
      "grad_norm": 0.63671875,
      "learning_rate": 9.416961696169617e-05,
      "loss": 3.2588,
      "step": 529200
    },
    {
      "epoch": 0.2958868010665329,
      "grad_norm": 0.71484375,
      "learning_rate": 9.416761676167616e-05,
      "loss": 3.2716,
      "step": 529210
    },
    {
      "epoch": 0.2959054443026911,
      "grad_norm": 0.63671875,
      "learning_rate": 9.416561656165617e-05,
      "loss": 3.1865,
      "step": 529220
    },
    {
      "epoch": 0.2959240875388493,
      "grad_norm": 0.7890625,
      "learning_rate": 9.416361636163617e-05,
      "loss": 3.1997,
      "step": 529230
    },
    {
      "epoch": 0.2959427307750075,
      "grad_norm": 0.6875,
      "learning_rate": 9.416161616161616e-05,
      "loss": 3.0888,
      "step": 529240
    },
    {
      "epoch": 0.2959613740111657,
      "grad_norm": 0.71484375,
      "learning_rate": 9.415961596159616e-05,
      "loss": 3.1391,
      "step": 529250
    },
    {
      "epoch": 0.29598001724732387,
      "grad_norm": 0.7578125,
      "learning_rate": 9.415761576157615e-05,
      "loss": 3.1499,
      "step": 529260
    },
    {
      "epoch": 0.29599866048348206,
      "grad_norm": 0.7265625,
      "learning_rate": 9.415561556155616e-05,
      "loss": 3.2754,
      "step": 529270
    },
    {
      "epoch": 0.29601730371964025,
      "grad_norm": 0.69921875,
      "learning_rate": 9.415361536153616e-05,
      "loss": 3.0754,
      "step": 529280
    },
    {
      "epoch": 0.2960359469557984,
      "grad_norm": 0.76953125,
      "learning_rate": 9.415161516151615e-05,
      "loss": 3.1778,
      "step": 529290
    },
    {
      "epoch": 0.29605459019195657,
      "grad_norm": 0.8515625,
      "learning_rate": 9.414961496149616e-05,
      "loss": 3.2091,
      "step": 529300
    },
    {
      "epoch": 0.29607323342811476,
      "grad_norm": 0.7734375,
      "learning_rate": 9.414761476147616e-05,
      "loss": 3.1536,
      "step": 529310
    },
    {
      "epoch": 0.29609187666427295,
      "grad_norm": 0.6328125,
      "learning_rate": 9.414561456145616e-05,
      "loss": 3.2628,
      "step": 529320
    },
    {
      "epoch": 0.29611051990043114,
      "grad_norm": 0.75390625,
      "learning_rate": 9.414361436143615e-05,
      "loss": 3.1364,
      "step": 529330
    },
    {
      "epoch": 0.2961291631365893,
      "grad_norm": 0.73046875,
      "learning_rate": 9.414161416141615e-05,
      "loss": 3.1168,
      "step": 529340
    },
    {
      "epoch": 0.2961478063727475,
      "grad_norm": 0.76171875,
      "learning_rate": 9.413961396139614e-05,
      "loss": 3.0976,
      "step": 529350
    },
    {
      "epoch": 0.2961664496089057,
      "grad_norm": 0.796875,
      "learning_rate": 9.413761376137614e-05,
      "loss": 3.1593,
      "step": 529360
    },
    {
      "epoch": 0.2961850928450639,
      "grad_norm": 0.7265625,
      "learning_rate": 9.413561356135613e-05,
      "loss": 3.1934,
      "step": 529370
    },
    {
      "epoch": 0.2962037360812221,
      "grad_norm": 0.85546875,
      "learning_rate": 9.413361336133614e-05,
      "loss": 3.2292,
      "step": 529380
    },
    {
      "epoch": 0.2962223793173802,
      "grad_norm": 0.71875,
      "learning_rate": 9.413161316131614e-05,
      "loss": 3.3131,
      "step": 529390
    },
    {
      "epoch": 0.2962410225535384,
      "grad_norm": 0.86328125,
      "learning_rate": 9.412961296129613e-05,
      "loss": 3.2494,
      "step": 529400
    },
    {
      "epoch": 0.2962596657896966,
      "grad_norm": 0.83203125,
      "learning_rate": 9.412761276127613e-05,
      "loss": 3.2043,
      "step": 529410
    },
    {
      "epoch": 0.2962783090258548,
      "grad_norm": 0.796875,
      "learning_rate": 9.412561256125612e-05,
      "loss": 3.207,
      "step": 529420
    },
    {
      "epoch": 0.29629695226201297,
      "grad_norm": 0.74609375,
      "learning_rate": 9.412361236123612e-05,
      "loss": 3.3138,
      "step": 529430
    },
    {
      "epoch": 0.29631559549817116,
      "grad_norm": 0.78515625,
      "learning_rate": 9.412161216121612e-05,
      "loss": 3.2436,
      "step": 529440
    },
    {
      "epoch": 0.29633423873432935,
      "grad_norm": 0.72265625,
      "learning_rate": 9.411961196119613e-05,
      "loss": 3.1948,
      "step": 529450
    },
    {
      "epoch": 0.29635288197048754,
      "grad_norm": 0.828125,
      "learning_rate": 9.411761176117613e-05,
      "loss": 3.1603,
      "step": 529460
    },
    {
      "epoch": 0.2963715252066457,
      "grad_norm": 0.60546875,
      "learning_rate": 9.411561156115613e-05,
      "loss": 3.1978,
      "step": 529470
    },
    {
      "epoch": 0.2963901684428039,
      "grad_norm": 0.76171875,
      "learning_rate": 9.411361136113613e-05,
      "loss": 3.2607,
      "step": 529480
    },
    {
      "epoch": 0.29640881167896205,
      "grad_norm": 0.62109375,
      "learning_rate": 9.411161116111612e-05,
      "loss": 3.1692,
      "step": 529490
    },
    {
      "epoch": 0.29642745491512024,
      "grad_norm": 0.73046875,
      "learning_rate": 9.410961096109612e-05,
      "loss": 3.2636,
      "step": 529500
    },
    {
      "epoch": 0.2964460981512784,
      "grad_norm": 0.75390625,
      "learning_rate": 9.410761076107611e-05,
      "loss": 3.184,
      "step": 529510
    },
    {
      "epoch": 0.2964647413874366,
      "grad_norm": 0.76953125,
      "learning_rate": 9.410561056105611e-05,
      "loss": 3.106,
      "step": 529520
    },
    {
      "epoch": 0.2964833846235948,
      "grad_norm": 0.76171875,
      "learning_rate": 9.41036103610361e-05,
      "loss": 3.1738,
      "step": 529530
    },
    {
      "epoch": 0.296502027859753,
      "grad_norm": 0.7421875,
      "learning_rate": 9.410161016101611e-05,
      "loss": 3.2271,
      "step": 529540
    },
    {
      "epoch": 0.2965206710959112,
      "grad_norm": 0.78125,
      "learning_rate": 9.409960996099611e-05,
      "loss": 3.3444,
      "step": 529550
    },
    {
      "epoch": 0.29653931433206937,
      "grad_norm": 0.65625,
      "learning_rate": 9.40976097609761e-05,
      "loss": 3.2623,
      "step": 529560
    },
    {
      "epoch": 0.29655795756822756,
      "grad_norm": 0.6875,
      "learning_rate": 9.40956095609561e-05,
      "loss": 3.094,
      "step": 529570
    },
    {
      "epoch": 0.29657660080438575,
      "grad_norm": 0.7890625,
      "learning_rate": 9.40936093609361e-05,
      "loss": 3.1774,
      "step": 529580
    },
    {
      "epoch": 0.29659524404054394,
      "grad_norm": 0.7734375,
      "learning_rate": 9.409160916091609e-05,
      "loss": 3.2264,
      "step": 529590
    },
    {
      "epoch": 0.29661388727670207,
      "grad_norm": 0.69921875,
      "learning_rate": 9.408960896089609e-05,
      "loss": 3.2493,
      "step": 529600
    },
    {
      "epoch": 0.29663253051286026,
      "grad_norm": 0.84375,
      "learning_rate": 9.40876087608761e-05,
      "loss": 3.2248,
      "step": 529610
    },
    {
      "epoch": 0.29665117374901845,
      "grad_norm": 0.6640625,
      "learning_rate": 9.408560856085609e-05,
      "loss": 3.2526,
      "step": 529620
    },
    {
      "epoch": 0.29666981698517664,
      "grad_norm": 0.7734375,
      "learning_rate": 9.408360836083609e-05,
      "loss": 3.2184,
      "step": 529630
    },
    {
      "epoch": 0.2966884602213348,
      "grad_norm": 0.6640625,
      "learning_rate": 9.408160816081608e-05,
      "loss": 3.2565,
      "step": 529640
    },
    {
      "epoch": 0.296707103457493,
      "grad_norm": 0.796875,
      "learning_rate": 9.407960796079608e-05,
      "loss": 3.3398,
      "step": 529650
    },
    {
      "epoch": 0.2967257466936512,
      "grad_norm": 0.72265625,
      "learning_rate": 9.407760776077609e-05,
      "loss": 3.2291,
      "step": 529660
    },
    {
      "epoch": 0.2967443899298094,
      "grad_norm": 0.734375,
      "learning_rate": 9.407560756075608e-05,
      "loss": 3.2624,
      "step": 529670
    },
    {
      "epoch": 0.2967630331659676,
      "grad_norm": 0.76953125,
      "learning_rate": 9.407360736073608e-05,
      "loss": 3.1234,
      "step": 529680
    },
    {
      "epoch": 0.29678167640212577,
      "grad_norm": 0.8046875,
      "learning_rate": 9.407160716071607e-05,
      "loss": 3.1371,
      "step": 529690
    },
    {
      "epoch": 0.2968003196382839,
      "grad_norm": 0.796875,
      "learning_rate": 9.406960696069608e-05,
      "loss": 3.1634,
      "step": 529700
    },
    {
      "epoch": 0.2968189628744421,
      "grad_norm": 0.6875,
      "learning_rate": 9.406760676067608e-05,
      "loss": 3.2709,
      "step": 529710
    },
    {
      "epoch": 0.2968376061106003,
      "grad_norm": 0.78515625,
      "learning_rate": 9.406560656065607e-05,
      "loss": 3.1623,
      "step": 529720
    },
    {
      "epoch": 0.29685624934675847,
      "grad_norm": 0.73828125,
      "learning_rate": 9.406360636063607e-05,
      "loss": 3.1419,
      "step": 529730
    },
    {
      "epoch": 0.29687489258291666,
      "grad_norm": 0.71484375,
      "learning_rate": 9.406160616061607e-05,
      "loss": 3.2885,
      "step": 529740
    },
    {
      "epoch": 0.29689353581907485,
      "grad_norm": 0.79296875,
      "learning_rate": 9.405960596059606e-05,
      "loss": 3.2016,
      "step": 529750
    },
    {
      "epoch": 0.29691217905523304,
      "grad_norm": 0.62109375,
      "learning_rate": 9.405760576057606e-05,
      "loss": 3.1748,
      "step": 529760
    },
    {
      "epoch": 0.2969308222913912,
      "grad_norm": 0.76953125,
      "learning_rate": 9.405560556055607e-05,
      "loss": 3.1952,
      "step": 529770
    },
    {
      "epoch": 0.2969494655275494,
      "grad_norm": 0.703125,
      "learning_rate": 9.405360536053606e-05,
      "loss": 3.2338,
      "step": 529780
    },
    {
      "epoch": 0.2969681087637076,
      "grad_norm": 0.71875,
      "learning_rate": 9.405160516051606e-05,
      "loss": 3.2863,
      "step": 529790
    },
    {
      "epoch": 0.2969867519998658,
      "grad_norm": 0.7265625,
      "learning_rate": 9.404960496049605e-05,
      "loss": 3.1492,
      "step": 529800
    },
    {
      "epoch": 0.29700539523602393,
      "grad_norm": 0.9296875,
      "learning_rate": 9.404760476047605e-05,
      "loss": 3.256,
      "step": 529810
    },
    {
      "epoch": 0.2970240384721821,
      "grad_norm": 0.74609375,
      "learning_rate": 9.404560456045604e-05,
      "loss": 3.3113,
      "step": 529820
    },
    {
      "epoch": 0.2970426817083403,
      "grad_norm": 0.7734375,
      "learning_rate": 9.404360436043604e-05,
      "loss": 3.2347,
      "step": 529830
    },
    {
      "epoch": 0.2970613249444985,
      "grad_norm": 0.84765625,
      "learning_rate": 9.404160416041603e-05,
      "loss": 3.3529,
      "step": 529840
    },
    {
      "epoch": 0.2970799681806567,
      "grad_norm": 0.74609375,
      "learning_rate": 9.403960396039604e-05,
      "loss": 3.266,
      "step": 529850
    },
    {
      "epoch": 0.2970986114168149,
      "grad_norm": 0.71875,
      "learning_rate": 9.403760376037605e-05,
      "loss": 3.2579,
      "step": 529860
    },
    {
      "epoch": 0.29711725465297306,
      "grad_norm": 0.7109375,
      "learning_rate": 9.403560356035605e-05,
      "loss": 3.1566,
      "step": 529870
    },
    {
      "epoch": 0.29713589788913125,
      "grad_norm": 0.671875,
      "learning_rate": 9.403360336033604e-05,
      "loss": 3.2252,
      "step": 529880
    },
    {
      "epoch": 0.29715454112528944,
      "grad_norm": 0.6328125,
      "learning_rate": 9.403160316031604e-05,
      "loss": 3.2239,
      "step": 529890
    },
    {
      "epoch": 0.29717318436144763,
      "grad_norm": 0.71484375,
      "learning_rate": 9.402960296029604e-05,
      "loss": 3.1986,
      "step": 529900
    },
    {
      "epoch": 0.29719182759760576,
      "grad_norm": 0.75390625,
      "learning_rate": 9.402760276027603e-05,
      "loss": 3.1894,
      "step": 529910
    },
    {
      "epoch": 0.29721047083376395,
      "grad_norm": 0.77734375,
      "learning_rate": 9.402560256025603e-05,
      "loss": 3.2482,
      "step": 529920
    },
    {
      "epoch": 0.29722911406992214,
      "grad_norm": 0.703125,
      "learning_rate": 9.402360236023604e-05,
      "loss": 3.2603,
      "step": 529930
    },
    {
      "epoch": 0.29724775730608033,
      "grad_norm": 0.73046875,
      "learning_rate": 9.402160216021603e-05,
      "loss": 3.1878,
      "step": 529940
    },
    {
      "epoch": 0.2972664005422385,
      "grad_norm": 0.6875,
      "learning_rate": 9.401960196019603e-05,
      "loss": 3.1248,
      "step": 529950
    },
    {
      "epoch": 0.2972850437783967,
      "grad_norm": 0.80859375,
      "learning_rate": 9.401760176017602e-05,
      "loss": 3.1382,
      "step": 529960
    },
    {
      "epoch": 0.2973036870145549,
      "grad_norm": 0.78515625,
      "learning_rate": 9.401560156015602e-05,
      "loss": 3.2218,
      "step": 529970
    },
    {
      "epoch": 0.2973223302507131,
      "grad_norm": 0.703125,
      "learning_rate": 9.401360136013601e-05,
      "loss": 3.1312,
      "step": 529980
    },
    {
      "epoch": 0.2973409734868713,
      "grad_norm": 0.73046875,
      "learning_rate": 9.401160116011601e-05,
      "loss": 3.2026,
      "step": 529990
    },
    {
      "epoch": 0.29735961672302946,
      "grad_norm": 0.75390625,
      "learning_rate": 9.4009600960096e-05,
      "loss": 3.2727,
      "step": 530000
    },
    {
      "epoch": 0.29737825995918765,
      "grad_norm": 0.7890625,
      "learning_rate": 9.400760076007601e-05,
      "loss": 3.1652,
      "step": 530010
    },
    {
      "epoch": 0.2973969031953458,
      "grad_norm": 0.83984375,
      "learning_rate": 9.400560056005601e-05,
      "loss": 3.2337,
      "step": 530020
    },
    {
      "epoch": 0.297415546431504,
      "grad_norm": 0.75390625,
      "learning_rate": 9.4003600360036e-05,
      "loss": 3.1583,
      "step": 530030
    },
    {
      "epoch": 0.29743418966766216,
      "grad_norm": 0.7421875,
      "learning_rate": 9.4001600160016e-05,
      "loss": 3.165,
      "step": 530040
    },
    {
      "epoch": 0.29745283290382035,
      "grad_norm": 0.7109375,
      "learning_rate": 9.3999599959996e-05,
      "loss": 3.1032,
      "step": 530050
    },
    {
      "epoch": 0.29747147613997854,
      "grad_norm": 0.79296875,
      "learning_rate": 9.3997599759976e-05,
      "loss": 3.1089,
      "step": 530060
    },
    {
      "epoch": 0.29749011937613673,
      "grad_norm": 0.66796875,
      "learning_rate": 9.3995599559956e-05,
      "loss": 3.2406,
      "step": 530070
    },
    {
      "epoch": 0.2975087626122949,
      "grad_norm": 0.72265625,
      "learning_rate": 9.3993599359936e-05,
      "loss": 3.2078,
      "step": 530080
    },
    {
      "epoch": 0.2975274058484531,
      "grad_norm": 0.68359375,
      "learning_rate": 9.3991599159916e-05,
      "loss": 3.2648,
      "step": 530090
    },
    {
      "epoch": 0.2975460490846113,
      "grad_norm": 0.75,
      "learning_rate": 9.3989598959896e-05,
      "loss": 3.2968,
      "step": 530100
    },
    {
      "epoch": 0.2975646923207695,
      "grad_norm": 0.796875,
      "learning_rate": 9.3987598759876e-05,
      "loss": 3.2649,
      "step": 530110
    },
    {
      "epoch": 0.2975833355569276,
      "grad_norm": 0.73828125,
      "learning_rate": 9.398559855985599e-05,
      "loss": 3.2476,
      "step": 530120
    },
    {
      "epoch": 0.2976019787930858,
      "grad_norm": 0.7734375,
      "learning_rate": 9.398359835983599e-05,
      "loss": 3.1254,
      "step": 530130
    },
    {
      "epoch": 0.297620622029244,
      "grad_norm": 0.67578125,
      "learning_rate": 9.398159815981598e-05,
      "loss": 3.0977,
      "step": 530140
    },
    {
      "epoch": 0.2976392652654022,
      "grad_norm": 0.78125,
      "learning_rate": 9.397959795979598e-05,
      "loss": 3.1291,
      "step": 530150
    },
    {
      "epoch": 0.2976579085015604,
      "grad_norm": 0.796875,
      "learning_rate": 9.397759775977597e-05,
      "loss": 3.3017,
      "step": 530160
    },
    {
      "epoch": 0.29767655173771856,
      "grad_norm": 0.7421875,
      "learning_rate": 9.397559755975598e-05,
      "loss": 3.245,
      "step": 530170
    },
    {
      "epoch": 0.29769519497387675,
      "grad_norm": 0.7890625,
      "learning_rate": 9.397359735973598e-05,
      "loss": 3.174,
      "step": 530180
    },
    {
      "epoch": 0.29771383821003494,
      "grad_norm": 0.8125,
      "learning_rate": 9.397159715971598e-05,
      "loss": 3.2259,
      "step": 530190
    },
    {
      "epoch": 0.29773248144619313,
      "grad_norm": 0.81640625,
      "learning_rate": 9.396959695969597e-05,
      "loss": 3.2141,
      "step": 530200
    },
    {
      "epoch": 0.2977511246823513,
      "grad_norm": 0.67578125,
      "learning_rate": 9.396759675967597e-05,
      "loss": 3.1871,
      "step": 530210
    },
    {
      "epoch": 0.2977697679185095,
      "grad_norm": 0.7265625,
      "learning_rate": 9.396559655965596e-05,
      "loss": 3.1643,
      "step": 530220
    },
    {
      "epoch": 0.29778841115466764,
      "grad_norm": 0.734375,
      "learning_rate": 9.396359635963596e-05,
      "loss": 3.1776,
      "step": 530230
    },
    {
      "epoch": 0.29780705439082583,
      "grad_norm": 0.6875,
      "learning_rate": 9.396159615961597e-05,
      "loss": 3.2504,
      "step": 530240
    },
    {
      "epoch": 0.297825697626984,
      "grad_norm": 0.75390625,
      "learning_rate": 9.395959595959598e-05,
      "loss": 3.2836,
      "step": 530250
    },
    {
      "epoch": 0.2978443408631422,
      "grad_norm": 0.72265625,
      "learning_rate": 9.395759575957597e-05,
      "loss": 3.366,
      "step": 530260
    },
    {
      "epoch": 0.2978629840993004,
      "grad_norm": 0.6484375,
      "learning_rate": 9.395559555955597e-05,
      "loss": 3.2421,
      "step": 530270
    },
    {
      "epoch": 0.2978816273354586,
      "grad_norm": 0.7578125,
      "learning_rate": 9.395359535953596e-05,
      "loss": 3.2516,
      "step": 530280
    },
    {
      "epoch": 0.2979002705716168,
      "grad_norm": 0.73046875,
      "learning_rate": 9.395159515951596e-05,
      "loss": 3.2213,
      "step": 530290
    },
    {
      "epoch": 0.29791891380777497,
      "grad_norm": 0.83984375,
      "learning_rate": 9.394959495949595e-05,
      "loss": 3.3583,
      "step": 530300
    },
    {
      "epoch": 0.29793755704393315,
      "grad_norm": 0.82421875,
      "learning_rate": 9.394759475947595e-05,
      "loss": 3.2834,
      "step": 530310
    },
    {
      "epoch": 0.29795620028009134,
      "grad_norm": 0.7578125,
      "learning_rate": 9.394559455945595e-05,
      "loss": 3.2469,
      "step": 530320
    },
    {
      "epoch": 0.2979748435162495,
      "grad_norm": 0.609375,
      "learning_rate": 9.394359435943595e-05,
      "loss": 3.1767,
      "step": 530330
    },
    {
      "epoch": 0.29799348675240767,
      "grad_norm": 0.640625,
      "learning_rate": 9.394159415941595e-05,
      "loss": 3.3272,
      "step": 530340
    },
    {
      "epoch": 0.29801212998856585,
      "grad_norm": 0.69921875,
      "learning_rate": 9.393959395939595e-05,
      "loss": 3.1518,
      "step": 530350
    },
    {
      "epoch": 0.29803077322472404,
      "grad_norm": 0.66796875,
      "learning_rate": 9.393759375937594e-05,
      "loss": 3.2754,
      "step": 530360
    },
    {
      "epoch": 0.29804941646088223,
      "grad_norm": 0.68359375,
      "learning_rate": 9.393559355935594e-05,
      "loss": 3.3454,
      "step": 530370
    },
    {
      "epoch": 0.2980680596970404,
      "grad_norm": 0.91796875,
      "learning_rate": 9.393359335933593e-05,
      "loss": 3.0645,
      "step": 530380
    },
    {
      "epoch": 0.2980867029331986,
      "grad_norm": 0.78515625,
      "learning_rate": 9.393159315931593e-05,
      "loss": 3.3045,
      "step": 530390
    },
    {
      "epoch": 0.2981053461693568,
      "grad_norm": 0.66796875,
      "learning_rate": 9.392959295929594e-05,
      "loss": 3.3528,
      "step": 530400
    },
    {
      "epoch": 0.298123989405515,
      "grad_norm": 0.80078125,
      "learning_rate": 9.392759275927593e-05,
      "loss": 3.1241,
      "step": 530410
    },
    {
      "epoch": 0.2981426326416732,
      "grad_norm": 0.7265625,
      "learning_rate": 9.392559255925593e-05,
      "loss": 3.2751,
      "step": 530420
    },
    {
      "epoch": 0.29816127587783137,
      "grad_norm": 0.7109375,
      "learning_rate": 9.392359235923592e-05,
      "loss": 3.2029,
      "step": 530430
    },
    {
      "epoch": 0.2981799191139895,
      "grad_norm": 0.74609375,
      "learning_rate": 9.392159215921592e-05,
      "loss": 3.2198,
      "step": 530440
    },
    {
      "epoch": 0.2981985623501477,
      "grad_norm": 0.73828125,
      "learning_rate": 9.391959195919593e-05,
      "loss": 3.2107,
      "step": 530450
    },
    {
      "epoch": 0.2982172055863059,
      "grad_norm": 0.78125,
      "learning_rate": 9.391759175917592e-05,
      "loss": 3.3104,
      "step": 530460
    },
    {
      "epoch": 0.29823584882246407,
      "grad_norm": 0.62109375,
      "learning_rate": 9.391559155915592e-05,
      "loss": 3.2008,
      "step": 530470
    },
    {
      "epoch": 0.29825449205862226,
      "grad_norm": 0.765625,
      "learning_rate": 9.391359135913592e-05,
      "loss": 3.2845,
      "step": 530480
    },
    {
      "epoch": 0.29827313529478044,
      "grad_norm": 0.734375,
      "learning_rate": 9.391159115911592e-05,
      "loss": 3.3371,
      "step": 530490
    },
    {
      "epoch": 0.29829177853093863,
      "grad_norm": 0.73828125,
      "learning_rate": 9.390959095909592e-05,
      "loss": 3.1526,
      "step": 530500
    },
    {
      "epoch": 0.2983104217670968,
      "grad_norm": 0.80859375,
      "learning_rate": 9.390759075907592e-05,
      "loss": 3.2247,
      "step": 530510
    },
    {
      "epoch": 0.298329065003255,
      "grad_norm": 0.76953125,
      "learning_rate": 9.390559055905591e-05,
      "loss": 3.2532,
      "step": 530520
    },
    {
      "epoch": 0.2983477082394132,
      "grad_norm": 0.75390625,
      "learning_rate": 9.390359035903591e-05,
      "loss": 3.2652,
      "step": 530530
    },
    {
      "epoch": 0.29836635147557133,
      "grad_norm": 0.8359375,
      "learning_rate": 9.39015901590159e-05,
      "loss": 3.1586,
      "step": 530540
    },
    {
      "epoch": 0.2983849947117295,
      "grad_norm": 0.78515625,
      "learning_rate": 9.38995899589959e-05,
      "loss": 3.2375,
      "step": 530550
    },
    {
      "epoch": 0.2984036379478877,
      "grad_norm": 0.69140625,
      "learning_rate": 9.389758975897591e-05,
      "loss": 3.1408,
      "step": 530560
    },
    {
      "epoch": 0.2984222811840459,
      "grad_norm": 0.69921875,
      "learning_rate": 9.38955895589559e-05,
      "loss": 3.192,
      "step": 530570
    },
    {
      "epoch": 0.2984409244202041,
      "grad_norm": 0.796875,
      "learning_rate": 9.38935893589359e-05,
      "loss": 3.1782,
      "step": 530580
    },
    {
      "epoch": 0.2984595676563623,
      "grad_norm": 0.6484375,
      "learning_rate": 9.38915891589159e-05,
      "loss": 3.3964,
      "step": 530590
    },
    {
      "epoch": 0.29847821089252047,
      "grad_norm": 0.7890625,
      "learning_rate": 9.388958895889589e-05,
      "loss": 3.3672,
      "step": 530600
    },
    {
      "epoch": 0.29849685412867866,
      "grad_norm": 0.75,
      "learning_rate": 9.388758875887589e-05,
      "loss": 3.1951,
      "step": 530610
    },
    {
      "epoch": 0.29851549736483685,
      "grad_norm": 0.76171875,
      "learning_rate": 9.388558855885588e-05,
      "loss": 3.146,
      "step": 530620
    },
    {
      "epoch": 0.29853414060099503,
      "grad_norm": 0.828125,
      "learning_rate": 9.388358835883589e-05,
      "loss": 3.1429,
      "step": 530630
    },
    {
      "epoch": 0.2985527838371532,
      "grad_norm": 0.88671875,
      "learning_rate": 9.388158815881589e-05,
      "loss": 3.1236,
      "step": 530640
    },
    {
      "epoch": 0.29857142707331136,
      "grad_norm": 0.76953125,
      "learning_rate": 9.38795879587959e-05,
      "loss": 3.3733,
      "step": 530650
    },
    {
      "epoch": 0.29859007030946955,
      "grad_norm": 0.65625,
      "learning_rate": 9.387758775877589e-05,
      "loss": 3.2428,
      "step": 530660
    },
    {
      "epoch": 0.29860871354562774,
      "grad_norm": 0.75,
      "learning_rate": 9.387558755875589e-05,
      "loss": 3.2536,
      "step": 530670
    },
    {
      "epoch": 0.2986273567817859,
      "grad_norm": 0.765625,
      "learning_rate": 9.387358735873588e-05,
      "loss": 3.2759,
      "step": 530680
    },
    {
      "epoch": 0.2986460000179441,
      "grad_norm": 0.81640625,
      "learning_rate": 9.387158715871588e-05,
      "loss": 3.2703,
      "step": 530690
    },
    {
      "epoch": 0.2986646432541023,
      "grad_norm": 0.65234375,
      "learning_rate": 9.386958695869587e-05,
      "loss": 3.2433,
      "step": 530700
    },
    {
      "epoch": 0.2986832864902605,
      "grad_norm": 0.8671875,
      "learning_rate": 9.386758675867587e-05,
      "loss": 3.112,
      "step": 530710
    },
    {
      "epoch": 0.2987019297264187,
      "grad_norm": 0.734375,
      "learning_rate": 9.386558655865588e-05,
      "loss": 3.2262,
      "step": 530720
    },
    {
      "epoch": 0.29872057296257687,
      "grad_norm": 0.6015625,
      "learning_rate": 9.386358635863587e-05,
      "loss": 3.1297,
      "step": 530730
    },
    {
      "epoch": 0.29873921619873506,
      "grad_norm": 0.6875,
      "learning_rate": 9.386158615861587e-05,
      "loss": 3.244,
      "step": 530740
    },
    {
      "epoch": 0.2987578594348932,
      "grad_norm": 0.57421875,
      "learning_rate": 9.385958595859586e-05,
      "loss": 3.2881,
      "step": 530750
    },
    {
      "epoch": 0.2987765026710514,
      "grad_norm": 0.79296875,
      "learning_rate": 9.385758575857586e-05,
      "loss": 3.1946,
      "step": 530760
    },
    {
      "epoch": 0.29879514590720957,
      "grad_norm": 0.69140625,
      "learning_rate": 9.385558555855586e-05,
      "loss": 3.1475,
      "step": 530770
    },
    {
      "epoch": 0.29881378914336776,
      "grad_norm": 0.78125,
      "learning_rate": 9.385358535853585e-05,
      "loss": 3.1264,
      "step": 530780
    },
    {
      "epoch": 0.29883243237952595,
      "grad_norm": 0.73046875,
      "learning_rate": 9.385158515851586e-05,
      "loss": 3.2987,
      "step": 530790
    },
    {
      "epoch": 0.29885107561568414,
      "grad_norm": 0.765625,
      "learning_rate": 9.384958495849586e-05,
      "loss": 3.1821,
      "step": 530800
    },
    {
      "epoch": 0.2988697188518423,
      "grad_norm": 0.70703125,
      "learning_rate": 9.384758475847585e-05,
      "loss": 3.1577,
      "step": 530810
    },
    {
      "epoch": 0.2988883620880005,
      "grad_norm": 0.765625,
      "learning_rate": 9.384558455845585e-05,
      "loss": 3.1617,
      "step": 530820
    },
    {
      "epoch": 0.2989070053241587,
      "grad_norm": 0.72265625,
      "learning_rate": 9.384358435843584e-05,
      "loss": 3.1444,
      "step": 530830
    },
    {
      "epoch": 0.2989256485603169,
      "grad_norm": 0.60546875,
      "learning_rate": 9.384158415841584e-05,
      "loss": 3.3167,
      "step": 530840
    },
    {
      "epoch": 0.298944291796475,
      "grad_norm": 0.76171875,
      "learning_rate": 9.383958395839585e-05,
      "loss": 3.2549,
      "step": 530850
    },
    {
      "epoch": 0.2989629350326332,
      "grad_norm": 0.85546875,
      "learning_rate": 9.383758375837584e-05,
      "loss": 3.2565,
      "step": 530860
    },
    {
      "epoch": 0.2989815782687914,
      "grad_norm": 0.79296875,
      "learning_rate": 9.383558355835584e-05,
      "loss": 3.2604,
      "step": 530870
    },
    {
      "epoch": 0.2990002215049496,
      "grad_norm": 0.65234375,
      "learning_rate": 9.383358335833585e-05,
      "loss": 3.2696,
      "step": 530880
    },
    {
      "epoch": 0.2990188647411078,
      "grad_norm": 0.81640625,
      "learning_rate": 9.383158315831584e-05,
      "loss": 3.3216,
      "step": 530890
    },
    {
      "epoch": 0.29903750797726597,
      "grad_norm": 0.80078125,
      "learning_rate": 9.382958295829584e-05,
      "loss": 3.137,
      "step": 530900
    },
    {
      "epoch": 0.29905615121342416,
      "grad_norm": 0.72265625,
      "learning_rate": 9.382758275827583e-05,
      "loss": 3.1524,
      "step": 530910
    },
    {
      "epoch": 0.29907479444958235,
      "grad_norm": 0.734375,
      "learning_rate": 9.382558255825583e-05,
      "loss": 3.124,
      "step": 530920
    },
    {
      "epoch": 0.29909343768574054,
      "grad_norm": 0.73828125,
      "learning_rate": 9.382358235823583e-05,
      "loss": 3.2564,
      "step": 530930
    },
    {
      "epoch": 0.2991120809218987,
      "grad_norm": 0.68359375,
      "learning_rate": 9.382158215821582e-05,
      "loss": 3.2005,
      "step": 530940
    },
    {
      "epoch": 0.2991307241580569,
      "grad_norm": 0.76171875,
      "learning_rate": 9.381958195819583e-05,
      "loss": 3.249,
      "step": 530950
    },
    {
      "epoch": 0.29914936739421505,
      "grad_norm": 0.86328125,
      "learning_rate": 9.381758175817583e-05,
      "loss": 3.1642,
      "step": 530960
    },
    {
      "epoch": 0.29916801063037324,
      "grad_norm": 0.83984375,
      "learning_rate": 9.381558155815582e-05,
      "loss": 3.2115,
      "step": 530970
    },
    {
      "epoch": 0.2991866538665314,
      "grad_norm": 0.55859375,
      "learning_rate": 9.381358135813582e-05,
      "loss": 3.3028,
      "step": 530980
    },
    {
      "epoch": 0.2992052971026896,
      "grad_norm": 0.90625,
      "learning_rate": 9.381158115811581e-05,
      "loss": 3.0807,
      "step": 530990
    },
    {
      "epoch": 0.2992239403388478,
      "grad_norm": 0.7421875,
      "learning_rate": 9.380958095809581e-05,
      "loss": 3.2182,
      "step": 531000
    },
    {
      "epoch": 0.299242583575006,
      "grad_norm": 0.6953125,
      "learning_rate": 9.38075807580758e-05,
      "loss": 3.2283,
      "step": 531010
    },
    {
      "epoch": 0.2992612268111642,
      "grad_norm": 0.77734375,
      "learning_rate": 9.38055805580558e-05,
      "loss": 3.1158,
      "step": 531020
    },
    {
      "epoch": 0.29927987004732237,
      "grad_norm": 0.8515625,
      "learning_rate": 9.380358035803581e-05,
      "loss": 3.2379,
      "step": 531030
    },
    {
      "epoch": 0.29929851328348056,
      "grad_norm": 0.68359375,
      "learning_rate": 9.380158015801582e-05,
      "loss": 3.2388,
      "step": 531040
    },
    {
      "epoch": 0.29931715651963875,
      "grad_norm": 0.796875,
      "learning_rate": 9.379957995799581e-05,
      "loss": 3.2376,
      "step": 531050
    },
    {
      "epoch": 0.2993357997557969,
      "grad_norm": 0.7578125,
      "learning_rate": 9.379757975797581e-05,
      "loss": 3.2559,
      "step": 531060
    },
    {
      "epoch": 0.29935444299195507,
      "grad_norm": 0.8984375,
      "learning_rate": 9.37955795579558e-05,
      "loss": 3.0972,
      "step": 531070
    },
    {
      "epoch": 0.29937308622811326,
      "grad_norm": 0.7734375,
      "learning_rate": 9.37935793579358e-05,
      "loss": 3.1622,
      "step": 531080
    },
    {
      "epoch": 0.29939172946427145,
      "grad_norm": 0.75,
      "learning_rate": 9.37915791579158e-05,
      "loss": 3.1735,
      "step": 531090
    },
    {
      "epoch": 0.29941037270042964,
      "grad_norm": 0.80078125,
      "learning_rate": 9.378957895789579e-05,
      "loss": 3.2304,
      "step": 531100
    },
    {
      "epoch": 0.29942901593658783,
      "grad_norm": 0.63671875,
      "learning_rate": 9.37875787578758e-05,
      "loss": 3.3067,
      "step": 531110
    },
    {
      "epoch": 0.299447659172746,
      "grad_norm": 0.7734375,
      "learning_rate": 9.37855785578558e-05,
      "loss": 3.1588,
      "step": 531120
    },
    {
      "epoch": 0.2994663024089042,
      "grad_norm": 0.796875,
      "learning_rate": 9.378357835783579e-05,
      "loss": 3.1749,
      "step": 531130
    },
    {
      "epoch": 0.2994849456450624,
      "grad_norm": 0.7578125,
      "learning_rate": 9.378157815781579e-05,
      "loss": 3.2322,
      "step": 531140
    },
    {
      "epoch": 0.2995035888812206,
      "grad_norm": 0.8359375,
      "learning_rate": 9.377957795779578e-05,
      "loss": 3.224,
      "step": 531150
    },
    {
      "epoch": 0.2995222321173788,
      "grad_norm": 0.71875,
      "learning_rate": 9.377757775777578e-05,
      "loss": 3.1472,
      "step": 531160
    },
    {
      "epoch": 0.2995408753535369,
      "grad_norm": 0.8125,
      "learning_rate": 9.377557755775577e-05,
      "loss": 3.2429,
      "step": 531170
    },
    {
      "epoch": 0.2995595185896951,
      "grad_norm": 0.6875,
      "learning_rate": 9.377357735773577e-05,
      "loss": 3.3415,
      "step": 531180
    },
    {
      "epoch": 0.2995781618258533,
      "grad_norm": 0.765625,
      "learning_rate": 9.377157715771578e-05,
      "loss": 3.2544,
      "step": 531190
    },
    {
      "epoch": 0.2995968050620115,
      "grad_norm": 0.7421875,
      "learning_rate": 9.376957695769577e-05,
      "loss": 3.1742,
      "step": 531200
    },
    {
      "epoch": 0.29961544829816966,
      "grad_norm": 0.7734375,
      "learning_rate": 9.376757675767577e-05,
      "loss": 3.169,
      "step": 531210
    },
    {
      "epoch": 0.29963409153432785,
      "grad_norm": 0.79296875,
      "learning_rate": 9.376557655765577e-05,
      "loss": 3.2298,
      "step": 531220
    },
    {
      "epoch": 0.29965273477048604,
      "grad_norm": 0.82421875,
      "learning_rate": 9.376357635763576e-05,
      "loss": 3.2436,
      "step": 531230
    },
    {
      "epoch": 0.29967137800664423,
      "grad_norm": 0.69140625,
      "learning_rate": 9.376157615761577e-05,
      "loss": 3.2409,
      "step": 531240
    },
    {
      "epoch": 0.2996900212428024,
      "grad_norm": 0.671875,
      "learning_rate": 9.375957595759577e-05,
      "loss": 3.3349,
      "step": 531250
    },
    {
      "epoch": 0.2997086644789606,
      "grad_norm": 0.59765625,
      "learning_rate": 9.375757575757576e-05,
      "loss": 3.2796,
      "step": 531260
    },
    {
      "epoch": 0.29972730771511874,
      "grad_norm": 0.76953125,
      "learning_rate": 9.375557555755576e-05,
      "loss": 3.2571,
      "step": 531270
    },
    {
      "epoch": 0.29974595095127693,
      "grad_norm": 0.84765625,
      "learning_rate": 9.375357535753577e-05,
      "loss": 3.1288,
      "step": 531280
    },
    {
      "epoch": 0.2997645941874351,
      "grad_norm": 0.73046875,
      "learning_rate": 9.375157515751576e-05,
      "loss": 3.2675,
      "step": 531290
    },
    {
      "epoch": 0.2997832374235933,
      "grad_norm": 0.81640625,
      "learning_rate": 9.374957495749576e-05,
      "loss": 3.1885,
      "step": 531300
    },
    {
      "epoch": 0.2998018806597515,
      "grad_norm": 0.67578125,
      "learning_rate": 9.374757475747575e-05,
      "loss": 3.2961,
      "step": 531310
    },
    {
      "epoch": 0.2998205238959097,
      "grad_norm": 0.69921875,
      "learning_rate": 9.374557455745575e-05,
      "loss": 3.2131,
      "step": 531320
    },
    {
      "epoch": 0.2998391671320679,
      "grad_norm": 0.796875,
      "learning_rate": 9.374357435743574e-05,
      "loss": 3.1082,
      "step": 531330
    },
    {
      "epoch": 0.29985781036822606,
      "grad_norm": 0.80078125,
      "learning_rate": 9.374157415741574e-05,
      "loss": 3.2298,
      "step": 531340
    },
    {
      "epoch": 0.29987645360438425,
      "grad_norm": 0.703125,
      "learning_rate": 9.373957395739575e-05,
      "loss": 3.2068,
      "step": 531350
    },
    {
      "epoch": 0.29989509684054244,
      "grad_norm": 0.765625,
      "learning_rate": 9.373757375737574e-05,
      "loss": 3.2545,
      "step": 531360
    },
    {
      "epoch": 0.29991374007670063,
      "grad_norm": 0.61328125,
      "learning_rate": 9.373557355735574e-05,
      "loss": 3.1476,
      "step": 531370
    },
    {
      "epoch": 0.29993238331285876,
      "grad_norm": 0.71875,
      "learning_rate": 9.373357335733574e-05,
      "loss": 3.0994,
      "step": 531380
    },
    {
      "epoch": 0.29995102654901695,
      "grad_norm": 0.81640625,
      "learning_rate": 9.373157315731573e-05,
      "loss": 3.2189,
      "step": 531390
    },
    {
      "epoch": 0.29996966978517514,
      "grad_norm": 0.7265625,
      "learning_rate": 9.372957295729573e-05,
      "loss": 3.2247,
      "step": 531400
    },
    {
      "epoch": 0.29998831302133333,
      "grad_norm": 0.7734375,
      "learning_rate": 9.372757275727572e-05,
      "loss": 3.2051,
      "step": 531410
    },
    {
      "epoch": 0.3000069562574915,
      "grad_norm": 0.734375,
      "learning_rate": 9.372557255725573e-05,
      "loss": 3.405,
      "step": 531420
    },
    {
      "epoch": 0.3000255994936497,
      "grad_norm": 0.85546875,
      "learning_rate": 9.372357235723573e-05,
      "loss": 3.1861,
      "step": 531430
    },
    {
      "epoch": 0.3000442427298079,
      "grad_norm": 0.8125,
      "learning_rate": 9.372157215721574e-05,
      "loss": 3.1018,
      "step": 531440
    },
    {
      "epoch": 0.3000628859659661,
      "grad_norm": 0.84765625,
      "learning_rate": 9.371957195719573e-05,
      "loss": 3.2296,
      "step": 531450
    },
    {
      "epoch": 0.3000815292021243,
      "grad_norm": 0.76171875,
      "learning_rate": 9.371757175717573e-05,
      "loss": 3.0982,
      "step": 531460
    },
    {
      "epoch": 0.30010017243828246,
      "grad_norm": 0.77734375,
      "learning_rate": 9.371557155715572e-05,
      "loss": 3.239,
      "step": 531470
    },
    {
      "epoch": 0.3001188156744406,
      "grad_norm": 0.71875,
      "learning_rate": 9.371357135713572e-05,
      "loss": 3.0918,
      "step": 531480
    },
    {
      "epoch": 0.3001374589105988,
      "grad_norm": 0.6953125,
      "learning_rate": 9.371157115711571e-05,
      "loss": 3.3025,
      "step": 531490
    },
    {
      "epoch": 0.300156102146757,
      "grad_norm": 0.78125,
      "learning_rate": 9.370957095709571e-05,
      "loss": 3.2892,
      "step": 531500
    }
  ],
  "logging_steps": 10,
  "max_steps": 1000000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.7940120013044777e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}